{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 100.0,
  "eval_steps": 500,
  "global_step": 112900,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0044286979627989375,
      "grad_norm": 11.820367813110352,
      "learning_rate": 4.428697962798938e-08,
      "loss": 1.7004,
      "step": 5
    },
    {
      "epoch": 0.008857395925597875,
      "grad_norm": 12.1470308303833,
      "learning_rate": 8.857395925597876e-08,
      "loss": 1.6149,
      "step": 10
    },
    {
      "epoch": 0.013286093888396812,
      "grad_norm": 11.656866073608398,
      "learning_rate": 1.3286093888396814e-07,
      "loss": 1.6436,
      "step": 15
    },
    {
      "epoch": 0.01771479185119575,
      "grad_norm": 11.312893867492676,
      "learning_rate": 1.7714791851195753e-07,
      "loss": 1.6183,
      "step": 20
    },
    {
      "epoch": 0.022143489813994686,
      "grad_norm": 11.812414169311523,
      "learning_rate": 2.214348981399469e-07,
      "loss": 1.6503,
      "step": 25
    },
    {
      "epoch": 0.026572187776793623,
      "grad_norm": 11.052279472351074,
      "learning_rate": 2.657218777679363e-07,
      "loss": 1.655,
      "step": 30
    },
    {
      "epoch": 0.03100088573959256,
      "grad_norm": 12.188301086425781,
      "learning_rate": 3.1000885739592564e-07,
      "loss": 1.6897,
      "step": 35
    },
    {
      "epoch": 0.0354295837023915,
      "grad_norm": 12.003920555114746,
      "learning_rate": 3.5429583702391505e-07,
      "loss": 1.5922,
      "step": 40
    },
    {
      "epoch": 0.03985828166519043,
      "grad_norm": 11.673710823059082,
      "learning_rate": 3.9858281665190436e-07,
      "loss": 1.6776,
      "step": 45
    },
    {
      "epoch": 0.04428697962798937,
      "grad_norm": 10.381232261657715,
      "learning_rate": 4.428697962798938e-07,
      "loss": 1.64,
      "step": 50
    },
    {
      "epoch": 0.048715677590788306,
      "grad_norm": 10.352510452270508,
      "learning_rate": 4.871567759078831e-07,
      "loss": 1.6747,
      "step": 55
    },
    {
      "epoch": 0.053144375553587246,
      "grad_norm": 10.216808319091797,
      "learning_rate": 5.314437555358726e-07,
      "loss": 1.6381,
      "step": 60
    },
    {
      "epoch": 0.05757307351638618,
      "grad_norm": 8.932103157043457,
      "learning_rate": 5.757307351638619e-07,
      "loss": 1.597,
      "step": 65
    },
    {
      "epoch": 0.06200177147918512,
      "grad_norm": 10.895230293273926,
      "learning_rate": 6.200177147918513e-07,
      "loss": 1.6513,
      "step": 70
    },
    {
      "epoch": 0.06643046944198405,
      "grad_norm": 10.943780899047852,
      "learning_rate": 6.643046944198406e-07,
      "loss": 1.611,
      "step": 75
    },
    {
      "epoch": 0.070859167404783,
      "grad_norm": 10.078645706176758,
      "learning_rate": 7.085916740478301e-07,
      "loss": 1.6542,
      "step": 80
    },
    {
      "epoch": 0.07528786536758193,
      "grad_norm": 8.50240421295166,
      "learning_rate": 7.528786536758194e-07,
      "loss": 1.6215,
      "step": 85
    },
    {
      "epoch": 0.07971656333038087,
      "grad_norm": 7.825551986694336,
      "learning_rate": 7.971656333038087e-07,
      "loss": 1.5225,
      "step": 90
    },
    {
      "epoch": 0.0841452612931798,
      "grad_norm": 7.054741382598877,
      "learning_rate": 8.41452612931798e-07,
      "loss": 1.5448,
      "step": 95
    },
    {
      "epoch": 0.08857395925597875,
      "grad_norm": 6.6052422523498535,
      "learning_rate": 8.857395925597875e-07,
      "loss": 1.504,
      "step": 100
    },
    {
      "epoch": 0.09300265721877768,
      "grad_norm": 6.6351752281188965,
      "learning_rate": 9.300265721877769e-07,
      "loss": 1.4697,
      "step": 105
    },
    {
      "epoch": 0.09743135518157661,
      "grad_norm": 7.448113918304443,
      "learning_rate": 9.743135518157663e-07,
      "loss": 1.5385,
      "step": 110
    },
    {
      "epoch": 0.10186005314437556,
      "grad_norm": 6.1844162940979,
      "learning_rate": 1.0186005314437557e-06,
      "loss": 1.4528,
      "step": 115
    },
    {
      "epoch": 0.10628875110717449,
      "grad_norm": 5.5842604637146,
      "learning_rate": 1.062887511071745e-06,
      "loss": 1.4363,
      "step": 120
    },
    {
      "epoch": 0.11071744906997343,
      "grad_norm": 4.3081440925598145,
      "learning_rate": 1.1071744906997343e-06,
      "loss": 1.3547,
      "step": 125
    },
    {
      "epoch": 0.11514614703277236,
      "grad_norm": 4.156342506408691,
      "learning_rate": 1.1514614703277237e-06,
      "loss": 1.3596,
      "step": 130
    },
    {
      "epoch": 0.1195748449955713,
      "grad_norm": 3.787323236465454,
      "learning_rate": 1.1957484499557131e-06,
      "loss": 1.4111,
      "step": 135
    },
    {
      "epoch": 0.12400354295837024,
      "grad_norm": 3.9504499435424805,
      "learning_rate": 1.2400354295837025e-06,
      "loss": 1.3179,
      "step": 140
    },
    {
      "epoch": 0.12843224092116917,
      "grad_norm": 3.5985236167907715,
      "learning_rate": 1.2843224092116918e-06,
      "loss": 1.2544,
      "step": 145
    },
    {
      "epoch": 0.1328609388839681,
      "grad_norm": 3.0680859088897705,
      "learning_rate": 1.3286093888396812e-06,
      "loss": 1.2642,
      "step": 150
    },
    {
      "epoch": 0.13728963684676704,
      "grad_norm": 2.8051581382751465,
      "learning_rate": 1.3728963684676706e-06,
      "loss": 1.2545,
      "step": 155
    },
    {
      "epoch": 0.141718334809566,
      "grad_norm": 2.2688193321228027,
      "learning_rate": 1.4171833480956602e-06,
      "loss": 1.2377,
      "step": 160
    },
    {
      "epoch": 0.14614703277236493,
      "grad_norm": 1.8552714586257935,
      "learning_rate": 1.4614703277236494e-06,
      "loss": 1.2488,
      "step": 165
    },
    {
      "epoch": 0.15057573073516387,
      "grad_norm": 1.4935394525527954,
      "learning_rate": 1.5057573073516388e-06,
      "loss": 1.185,
      "step": 170
    },
    {
      "epoch": 0.1550044286979628,
      "grad_norm": 0.948371946811676,
      "learning_rate": 1.550044286979628e-06,
      "loss": 1.1916,
      "step": 175
    },
    {
      "epoch": 0.15943312666076173,
      "grad_norm": 0.8102244734764099,
      "learning_rate": 1.5943312666076174e-06,
      "loss": 1.1421,
      "step": 180
    },
    {
      "epoch": 0.16386182462356066,
      "grad_norm": 0.5309062004089355,
      "learning_rate": 1.6386182462356069e-06,
      "loss": 1.1567,
      "step": 185
    },
    {
      "epoch": 0.1682905225863596,
      "grad_norm": 0.5291673541069031,
      "learning_rate": 1.682905225863596e-06,
      "loss": 1.1144,
      "step": 190
    },
    {
      "epoch": 0.17271922054915856,
      "grad_norm": 0.39175063371658325,
      "learning_rate": 1.7271922054915857e-06,
      "loss": 1.1344,
      "step": 195
    },
    {
      "epoch": 0.1771479185119575,
      "grad_norm": 0.42602333426475525,
      "learning_rate": 1.771479185119575e-06,
      "loss": 1.124,
      "step": 200
    },
    {
      "epoch": 0.18157661647475642,
      "grad_norm": 0.27804046869277954,
      "learning_rate": 1.8157661647475643e-06,
      "loss": 1.1309,
      "step": 205
    },
    {
      "epoch": 0.18600531443755536,
      "grad_norm": 0.3582090735435486,
      "learning_rate": 1.8600531443755537e-06,
      "loss": 1.0846,
      "step": 210
    },
    {
      "epoch": 0.1904340124003543,
      "grad_norm": 0.33198463916778564,
      "learning_rate": 1.9043401240035431e-06,
      "loss": 1.1095,
      "step": 215
    },
    {
      "epoch": 0.19486271036315322,
      "grad_norm": 0.312926322221756,
      "learning_rate": 1.9486271036315325e-06,
      "loss": 1.0828,
      "step": 220
    },
    {
      "epoch": 0.19929140832595216,
      "grad_norm": 0.3295394480228424,
      "learning_rate": 1.9929140832595215e-06,
      "loss": 1.0841,
      "step": 225
    },
    {
      "epoch": 0.20372010628875112,
      "grad_norm": 0.3140348792076111,
      "learning_rate": 2.0372010628875114e-06,
      "loss": 1.1161,
      "step": 230
    },
    {
      "epoch": 0.20814880425155005,
      "grad_norm": 0.264995813369751,
      "learning_rate": 2.0814880425155008e-06,
      "loss": 1.1086,
      "step": 235
    },
    {
      "epoch": 0.21257750221434898,
      "grad_norm": 0.27205732464790344,
      "learning_rate": 2.12577502214349e-06,
      "loss": 1.1358,
      "step": 240
    },
    {
      "epoch": 0.21700620017714792,
      "grad_norm": 0.31799212098121643,
      "learning_rate": 2.170062001771479e-06,
      "loss": 1.0842,
      "step": 245
    },
    {
      "epoch": 0.22143489813994685,
      "grad_norm": 0.27930590510368347,
      "learning_rate": 2.2143489813994686e-06,
      "loss": 1.1427,
      "step": 250
    },
    {
      "epoch": 0.22586359610274578,
      "grad_norm": 0.43252986669540405,
      "learning_rate": 2.258635961027458e-06,
      "loss": 1.1273,
      "step": 255
    },
    {
      "epoch": 0.23029229406554472,
      "grad_norm": 0.538478672504425,
      "learning_rate": 2.3029229406554474e-06,
      "loss": 1.085,
      "step": 260
    },
    {
      "epoch": 0.23472099202834368,
      "grad_norm": 0.25883543491363525,
      "learning_rate": 2.347209920283437e-06,
      "loss": 1.0959,
      "step": 265
    },
    {
      "epoch": 0.2391496899911426,
      "grad_norm": 0.3655385971069336,
      "learning_rate": 2.3914968999114263e-06,
      "loss": 1.1034,
      "step": 270
    },
    {
      "epoch": 0.24357838795394154,
      "grad_norm": 0.28594473004341125,
      "learning_rate": 2.4357838795394157e-06,
      "loss": 1.0776,
      "step": 275
    },
    {
      "epoch": 0.24800708591674048,
      "grad_norm": 0.34155675768852234,
      "learning_rate": 2.480070859167405e-06,
      "loss": 1.0311,
      "step": 280
    },
    {
      "epoch": 0.2524357838795394,
      "grad_norm": 0.4211924970149994,
      "learning_rate": 2.5243578387953945e-06,
      "loss": 1.0631,
      "step": 285
    },
    {
      "epoch": 0.25686448184233834,
      "grad_norm": 0.36909550428390503,
      "learning_rate": 2.5686448184233835e-06,
      "loss": 1.0825,
      "step": 290
    },
    {
      "epoch": 0.2612931798051373,
      "grad_norm": 0.37971094250679016,
      "learning_rate": 2.612931798051373e-06,
      "loss": 1.147,
      "step": 295
    },
    {
      "epoch": 0.2657218777679362,
      "grad_norm": 0.46819737553596497,
      "learning_rate": 2.6572187776793623e-06,
      "loss": 1.0498,
      "step": 300
    },
    {
      "epoch": 0.27015057573073514,
      "grad_norm": 0.6770675182342529,
      "learning_rate": 2.7015057573073517e-06,
      "loss": 1.1552,
      "step": 305
    },
    {
      "epoch": 0.2745792736935341,
      "grad_norm": 0.37761443853378296,
      "learning_rate": 2.745792736935341e-06,
      "loss": 1.1078,
      "step": 310
    },
    {
      "epoch": 0.27900797165633306,
      "grad_norm": 0.40357354283332825,
      "learning_rate": 2.790079716563331e-06,
      "loss": 1.1332,
      "step": 315
    },
    {
      "epoch": 0.283436669619132,
      "grad_norm": 0.28160160779953003,
      "learning_rate": 2.8343666961913204e-06,
      "loss": 1.1283,
      "step": 320
    },
    {
      "epoch": 0.28786536758193093,
      "grad_norm": 0.37395164370536804,
      "learning_rate": 2.8786536758193094e-06,
      "loss": 1.1381,
      "step": 325
    },
    {
      "epoch": 0.29229406554472986,
      "grad_norm": 0.3414997458457947,
      "learning_rate": 2.922940655447299e-06,
      "loss": 1.0947,
      "step": 330
    },
    {
      "epoch": 0.2967227635075288,
      "grad_norm": 0.5143349766731262,
      "learning_rate": 2.9672276350752882e-06,
      "loss": 1.1016,
      "step": 335
    },
    {
      "epoch": 0.30115146147032773,
      "grad_norm": 0.37565624713897705,
      "learning_rate": 3.0115146147032776e-06,
      "loss": 1.0792,
      "step": 340
    },
    {
      "epoch": 0.30558015943312666,
      "grad_norm": 0.308960884809494,
      "learning_rate": 3.055801594331267e-06,
      "loss": 1.0969,
      "step": 345
    },
    {
      "epoch": 0.3100088573959256,
      "grad_norm": 0.4782509207725525,
      "learning_rate": 3.100088573959256e-06,
      "loss": 1.1341,
      "step": 350
    },
    {
      "epoch": 0.31443755535872453,
      "grad_norm": 0.3736422061920166,
      "learning_rate": 3.1443755535872455e-06,
      "loss": 1.1104,
      "step": 355
    },
    {
      "epoch": 0.31886625332152346,
      "grad_norm": 0.4654044508934021,
      "learning_rate": 3.188662533215235e-06,
      "loss": 1.0979,
      "step": 360
    },
    {
      "epoch": 0.3232949512843224,
      "grad_norm": 0.35784783959388733,
      "learning_rate": 3.2329495128432243e-06,
      "loss": 1.0766,
      "step": 365
    },
    {
      "epoch": 0.32772364924712133,
      "grad_norm": 0.39122867584228516,
      "learning_rate": 3.2772364924712137e-06,
      "loss": 1.1236,
      "step": 370
    },
    {
      "epoch": 0.33215234720992026,
      "grad_norm": 0.3576284945011139,
      "learning_rate": 3.3215234720992027e-06,
      "loss": 1.0313,
      "step": 375
    },
    {
      "epoch": 0.3365810451727192,
      "grad_norm": 0.4109048545360565,
      "learning_rate": 3.365810451727192e-06,
      "loss": 1.0666,
      "step": 380
    },
    {
      "epoch": 0.3410097431355182,
      "grad_norm": 0.5468763113021851,
      "learning_rate": 3.410097431355182e-06,
      "loss": 1.0576,
      "step": 385
    },
    {
      "epoch": 0.3454384410983171,
      "grad_norm": 0.39048174023628235,
      "learning_rate": 3.4543844109831714e-06,
      "loss": 1.052,
      "step": 390
    },
    {
      "epoch": 0.34986713906111605,
      "grad_norm": 0.41467055678367615,
      "learning_rate": 3.4986713906111608e-06,
      "loss": 1.092,
      "step": 395
    },
    {
      "epoch": 0.354295837023915,
      "grad_norm": 0.41701918840408325,
      "learning_rate": 3.54295837023915e-06,
      "loss": 1.0536,
      "step": 400
    },
    {
      "epoch": 0.3587245349867139,
      "grad_norm": 0.5761395692825317,
      "learning_rate": 3.5872453498671396e-06,
      "loss": 1.1105,
      "step": 405
    },
    {
      "epoch": 0.36315323294951285,
      "grad_norm": 0.4602561295032501,
      "learning_rate": 3.6315323294951286e-06,
      "loss": 1.1196,
      "step": 410
    },
    {
      "epoch": 0.3675819309123118,
      "grad_norm": 0.4647201597690582,
      "learning_rate": 3.675819309123118e-06,
      "loss": 1.1325,
      "step": 415
    },
    {
      "epoch": 0.3720106288751107,
      "grad_norm": 0.3582036793231964,
      "learning_rate": 3.7201062887511074e-06,
      "loss": 1.101,
      "step": 420
    },
    {
      "epoch": 0.37643932683790965,
      "grad_norm": 0.44388604164123535,
      "learning_rate": 3.764393268379097e-06,
      "loss": 1.0995,
      "step": 425
    },
    {
      "epoch": 0.3808680248007086,
      "grad_norm": 0.42416882514953613,
      "learning_rate": 3.8086802480070863e-06,
      "loss": 1.0735,
      "step": 430
    },
    {
      "epoch": 0.3852967227635075,
      "grad_norm": 0.3444736897945404,
      "learning_rate": 3.852967227635076e-06,
      "loss": 1.0877,
      "step": 435
    },
    {
      "epoch": 0.38972542072630645,
      "grad_norm": 0.3834967017173767,
      "learning_rate": 3.897254207263065e-06,
      "loss": 1.0893,
      "step": 440
    },
    {
      "epoch": 0.3941541186891054,
      "grad_norm": 0.6447907090187073,
      "learning_rate": 3.9415411868910545e-06,
      "loss": 1.072,
      "step": 445
    },
    {
      "epoch": 0.3985828166519043,
      "grad_norm": 0.4464726448059082,
      "learning_rate": 3.985828166519043e-06,
      "loss": 1.0568,
      "step": 450
    },
    {
      "epoch": 0.4030115146147033,
      "grad_norm": 0.4295834004878998,
      "learning_rate": 4.030115146147033e-06,
      "loss": 1.0923,
      "step": 455
    },
    {
      "epoch": 0.40744021257750224,
      "grad_norm": 0.4237326979637146,
      "learning_rate": 4.074402125775023e-06,
      "loss": 1.0702,
      "step": 460
    },
    {
      "epoch": 0.41186891054030117,
      "grad_norm": 0.4239921271800995,
      "learning_rate": 4.118689105403012e-06,
      "loss": 1.0865,
      "step": 465
    },
    {
      "epoch": 0.4162976085031001,
      "grad_norm": 0.4564349353313446,
      "learning_rate": 4.1629760850310016e-06,
      "loss": 1.1196,
      "step": 470
    },
    {
      "epoch": 0.42072630646589904,
      "grad_norm": 0.4897279143333435,
      "learning_rate": 4.207263064658991e-06,
      "loss": 1.0389,
      "step": 475
    },
    {
      "epoch": 0.42515500442869797,
      "grad_norm": 0.4564661681652069,
      "learning_rate": 4.25155004428698e-06,
      "loss": 1.0837,
      "step": 480
    },
    {
      "epoch": 0.4295837023914969,
      "grad_norm": 0.43336793780326843,
      "learning_rate": 4.295837023914969e-06,
      "loss": 1.0765,
      "step": 485
    },
    {
      "epoch": 0.43401240035429584,
      "grad_norm": 0.5645413994789124,
      "learning_rate": 4.340124003542958e-06,
      "loss": 1.0094,
      "step": 490
    },
    {
      "epoch": 0.43844109831709477,
      "grad_norm": 0.93393874168396,
      "learning_rate": 4.384410983170948e-06,
      "loss": 1.1016,
      "step": 495
    },
    {
      "epoch": 0.4428697962798937,
      "grad_norm": 0.7553842067718506,
      "learning_rate": 4.428697962798937e-06,
      "loss": 1.0389,
      "step": 500
    },
    {
      "epoch": 0.44729849424269263,
      "grad_norm": 0.41328611969947815,
      "learning_rate": 4.472984942426927e-06,
      "loss": 1.0544,
      "step": 505
    },
    {
      "epoch": 0.45172719220549157,
      "grad_norm": 0.4961669445037842,
      "learning_rate": 4.517271922054916e-06,
      "loss": 1.0901,
      "step": 510
    },
    {
      "epoch": 0.4561558901682905,
      "grad_norm": 0.37691730260849,
      "learning_rate": 4.5615589016829055e-06,
      "loss": 1.0793,
      "step": 515
    },
    {
      "epoch": 0.46058458813108943,
      "grad_norm": 0.4213143587112427,
      "learning_rate": 4.605845881310895e-06,
      "loss": 1.1046,
      "step": 520
    },
    {
      "epoch": 0.4650132860938884,
      "grad_norm": 0.4811985492706299,
      "learning_rate": 4.650132860938884e-06,
      "loss": 1.1424,
      "step": 525
    },
    {
      "epoch": 0.46944198405668736,
      "grad_norm": 0.4970588982105255,
      "learning_rate": 4.694419840566874e-06,
      "loss": 1.045,
      "step": 530
    },
    {
      "epoch": 0.4738706820194863,
      "grad_norm": 0.38179582357406616,
      "learning_rate": 4.738706820194863e-06,
      "loss": 1.0647,
      "step": 535
    },
    {
      "epoch": 0.4782993799822852,
      "grad_norm": 0.6697129011154175,
      "learning_rate": 4.7829937998228525e-06,
      "loss": 1.1035,
      "step": 540
    },
    {
      "epoch": 0.48272807794508416,
      "grad_norm": 0.3649693429470062,
      "learning_rate": 4.827280779450842e-06,
      "loss": 1.0344,
      "step": 545
    },
    {
      "epoch": 0.4871567759078831,
      "grad_norm": 0.38832440972328186,
      "learning_rate": 4.871567759078831e-06,
      "loss": 1.11,
      "step": 550
    },
    {
      "epoch": 0.491585473870682,
      "grad_norm": 0.5177776217460632,
      "learning_rate": 4.915854738706821e-06,
      "loss": 1.0588,
      "step": 555
    },
    {
      "epoch": 0.49601417183348095,
      "grad_norm": 0.3817882239818573,
      "learning_rate": 4.96014171833481e-06,
      "loss": 1.0765,
      "step": 560
    },
    {
      "epoch": 0.5004428697962799,
      "grad_norm": 0.6858991980552673,
      "learning_rate": 5.0044286979628e-06,
      "loss": 1.0933,
      "step": 565
    },
    {
      "epoch": 0.5048715677590788,
      "grad_norm": 0.4919489622116089,
      "learning_rate": 5.048715677590789e-06,
      "loss": 1.1109,
      "step": 570
    },
    {
      "epoch": 0.5093002657218778,
      "grad_norm": 0.5251255035400391,
      "learning_rate": 5.0930026572187784e-06,
      "loss": 1.1175,
      "step": 575
    },
    {
      "epoch": 0.5137289636846767,
      "grad_norm": 0.5085564851760864,
      "learning_rate": 5.137289636846767e-06,
      "loss": 1.086,
      "step": 580
    },
    {
      "epoch": 0.5181576616474757,
      "grad_norm": 0.49915391206741333,
      "learning_rate": 5.181576616474757e-06,
      "loss": 1.0986,
      "step": 585
    },
    {
      "epoch": 0.5225863596102746,
      "grad_norm": 0.6538212895393372,
      "learning_rate": 5.225863596102746e-06,
      "loss": 1.0522,
      "step": 590
    },
    {
      "epoch": 0.5270150575730735,
      "grad_norm": 0.58580482006073,
      "learning_rate": 5.270150575730736e-06,
      "loss": 1.0854,
      "step": 595
    },
    {
      "epoch": 0.5314437555358724,
      "grad_norm": 0.5502675771713257,
      "learning_rate": 5.314437555358725e-06,
      "loss": 1.0861,
      "step": 600
    },
    {
      "epoch": 0.5358724534986714,
      "grad_norm": 0.4223230183124542,
      "learning_rate": 5.358724534986715e-06,
      "loss": 1.1022,
      "step": 605
    },
    {
      "epoch": 0.5403011514614703,
      "grad_norm": 0.6294427514076233,
      "learning_rate": 5.4030115146147035e-06,
      "loss": 1.068,
      "step": 610
    },
    {
      "epoch": 0.5447298494242693,
      "grad_norm": 0.5249088406562805,
      "learning_rate": 5.447298494242693e-06,
      "loss": 1.087,
      "step": 615
    },
    {
      "epoch": 0.5491585473870682,
      "grad_norm": 0.7817137241363525,
      "learning_rate": 5.491585473870682e-06,
      "loss": 1.0777,
      "step": 620
    },
    {
      "epoch": 0.5535872453498671,
      "grad_norm": 0.5898805260658264,
      "learning_rate": 5.535872453498672e-06,
      "loss": 1.0282,
      "step": 625
    },
    {
      "epoch": 0.5580159433126661,
      "grad_norm": 0.7486714124679565,
      "learning_rate": 5.580159433126662e-06,
      "loss": 1.0189,
      "step": 630
    },
    {
      "epoch": 0.562444641275465,
      "grad_norm": 0.49449342489242554,
      "learning_rate": 5.6244464127546506e-06,
      "loss": 1.0781,
      "step": 635
    },
    {
      "epoch": 0.566873339238264,
      "grad_norm": 0.483915776014328,
      "learning_rate": 5.668733392382641e-06,
      "loss": 1.0406,
      "step": 640
    },
    {
      "epoch": 0.5713020372010629,
      "grad_norm": 0.8464071750640869,
      "learning_rate": 5.713020372010629e-06,
      "loss": 1.0296,
      "step": 645
    },
    {
      "epoch": 0.5757307351638619,
      "grad_norm": 0.657326877117157,
      "learning_rate": 5.757307351638619e-06,
      "loss": 1.0481,
      "step": 650
    },
    {
      "epoch": 0.5801594331266607,
      "grad_norm": 0.5357707738876343,
      "learning_rate": 5.801594331266608e-06,
      "loss": 1.0536,
      "step": 655
    },
    {
      "epoch": 0.5845881310894597,
      "grad_norm": 0.4976673722267151,
      "learning_rate": 5.845881310894598e-06,
      "loss": 1.0818,
      "step": 660
    },
    {
      "epoch": 0.5890168290522586,
      "grad_norm": 0.6040676236152649,
      "learning_rate": 5.890168290522586e-06,
      "loss": 1.0511,
      "step": 665
    },
    {
      "epoch": 0.5934455270150576,
      "grad_norm": 0.8693132996559143,
      "learning_rate": 5.9344552701505765e-06,
      "loss": 1.0654,
      "step": 670
    },
    {
      "epoch": 0.5978742249778565,
      "grad_norm": 0.7987152338027954,
      "learning_rate": 5.978742249778565e-06,
      "loss": 1.0634,
      "step": 675
    },
    {
      "epoch": 0.6023029229406555,
      "grad_norm": 0.8263706564903259,
      "learning_rate": 6.023029229406555e-06,
      "loss": 1.0539,
      "step": 680
    },
    {
      "epoch": 0.6067316209034543,
      "grad_norm": 0.6553357839584351,
      "learning_rate": 6.067316209034544e-06,
      "loss": 1.0636,
      "step": 685
    },
    {
      "epoch": 0.6111603188662533,
      "grad_norm": 0.680133044719696,
      "learning_rate": 6.111603188662534e-06,
      "loss": 1.0842,
      "step": 690
    },
    {
      "epoch": 0.6155890168290522,
      "grad_norm": 0.6346431970596313,
      "learning_rate": 6.155890168290523e-06,
      "loss": 1.0823,
      "step": 695
    },
    {
      "epoch": 0.6200177147918512,
      "grad_norm": 0.6364468932151794,
      "learning_rate": 6.200177147918512e-06,
      "loss": 1.0705,
      "step": 700
    },
    {
      "epoch": 0.6244464127546502,
      "grad_norm": 0.9277119040489197,
      "learning_rate": 6.244464127546502e-06,
      "loss": 1.0438,
      "step": 705
    },
    {
      "epoch": 0.6288751107174491,
      "grad_norm": 0.9401694536209106,
      "learning_rate": 6.288751107174491e-06,
      "loss": 1.0416,
      "step": 710
    },
    {
      "epoch": 0.633303808680248,
      "grad_norm": 0.534271776676178,
      "learning_rate": 6.333038086802481e-06,
      "loss": 1.0831,
      "step": 715
    },
    {
      "epoch": 0.6377325066430469,
      "grad_norm": 0.7278724908828735,
      "learning_rate": 6.37732506643047e-06,
      "loss": 1.0662,
      "step": 720
    },
    {
      "epoch": 0.6421612046058459,
      "grad_norm": 0.6743765473365784,
      "learning_rate": 6.42161204605846e-06,
      "loss": 1.1087,
      "step": 725
    },
    {
      "epoch": 0.6465899025686448,
      "grad_norm": 0.5241950750350952,
      "learning_rate": 6.465899025686449e-06,
      "loss": 1.0673,
      "step": 730
    },
    {
      "epoch": 0.6510186005314438,
      "grad_norm": 0.6579424738883972,
      "learning_rate": 6.510186005314438e-06,
      "loss": 1.0607,
      "step": 735
    },
    {
      "epoch": 0.6554472984942427,
      "grad_norm": 0.7520736455917358,
      "learning_rate": 6.554472984942427e-06,
      "loss": 1.0426,
      "step": 740
    },
    {
      "epoch": 0.6598759964570416,
      "grad_norm": 0.5593547224998474,
      "learning_rate": 6.598759964570417e-06,
      "loss": 1.0462,
      "step": 745
    },
    {
      "epoch": 0.6643046944198405,
      "grad_norm": 0.7848285436630249,
      "learning_rate": 6.643046944198405e-06,
      "loss": 1.0614,
      "step": 750
    },
    {
      "epoch": 0.6687333923826395,
      "grad_norm": 0.5062673091888428,
      "learning_rate": 6.687333923826396e-06,
      "loss": 1.0777,
      "step": 755
    },
    {
      "epoch": 0.6731620903454384,
      "grad_norm": 0.6015821695327759,
      "learning_rate": 6.731620903454384e-06,
      "loss": 1.0379,
      "step": 760
    },
    {
      "epoch": 0.6775907883082374,
      "grad_norm": 0.570980966091156,
      "learning_rate": 6.7759078830823745e-06,
      "loss": 1.0463,
      "step": 765
    },
    {
      "epoch": 0.6820194862710364,
      "grad_norm": 0.6550763845443726,
      "learning_rate": 6.820194862710364e-06,
      "loss": 0.9967,
      "step": 770
    },
    {
      "epoch": 0.6864481842338352,
      "grad_norm": 0.5866822600364685,
      "learning_rate": 6.864481842338353e-06,
      "loss": 1.0862,
      "step": 775
    },
    {
      "epoch": 0.6908768821966342,
      "grad_norm": 0.6404900550842285,
      "learning_rate": 6.908768821966343e-06,
      "loss": 1.0833,
      "step": 780
    },
    {
      "epoch": 0.6953055801594331,
      "grad_norm": 0.4883545935153961,
      "learning_rate": 6.953055801594331e-06,
      "loss": 1.0391,
      "step": 785
    },
    {
      "epoch": 0.6997342781222321,
      "grad_norm": 0.45520517230033875,
      "learning_rate": 6.9973427812223216e-06,
      "loss": 1.0488,
      "step": 790
    },
    {
      "epoch": 0.704162976085031,
      "grad_norm": 0.5145585536956787,
      "learning_rate": 7.04162976085031e-06,
      "loss": 1.0492,
      "step": 795
    },
    {
      "epoch": 0.70859167404783,
      "grad_norm": 0.4754512906074524,
      "learning_rate": 7.0859167404783e-06,
      "loss": 1.0581,
      "step": 800
    },
    {
      "epoch": 0.7130203720106288,
      "grad_norm": 0.43282949924468994,
      "learning_rate": 7.130203720106289e-06,
      "loss": 1.0167,
      "step": 805
    },
    {
      "epoch": 0.7174490699734278,
      "grad_norm": 0.42789632081985474,
      "learning_rate": 7.174490699734279e-06,
      "loss": 1.0912,
      "step": 810
    },
    {
      "epoch": 0.7218777679362267,
      "grad_norm": 0.7601043581962585,
      "learning_rate": 7.218777679362268e-06,
      "loss": 1.0695,
      "step": 815
    },
    {
      "epoch": 0.7263064658990257,
      "grad_norm": 0.5453900694847107,
      "learning_rate": 7.263064658990257e-06,
      "loss": 1.0743,
      "step": 820
    },
    {
      "epoch": 0.7307351638618246,
      "grad_norm": 0.4569368064403534,
      "learning_rate": 7.307351638618247e-06,
      "loss": 1.0835,
      "step": 825
    },
    {
      "epoch": 0.7351638618246236,
      "grad_norm": 0.5344635248184204,
      "learning_rate": 7.351638618246236e-06,
      "loss": 0.9881,
      "step": 830
    },
    {
      "epoch": 0.7395925597874224,
      "grad_norm": 0.5342046618461609,
      "learning_rate": 7.3959255978742254e-06,
      "loss": 1.0668,
      "step": 835
    },
    {
      "epoch": 0.7440212577502214,
      "grad_norm": 0.5426737070083618,
      "learning_rate": 7.440212577502215e-06,
      "loss": 0.985,
      "step": 840
    },
    {
      "epoch": 0.7484499557130204,
      "grad_norm": 0.39210185408592224,
      "learning_rate": 7.484499557130205e-06,
      "loss": 1.0626,
      "step": 845
    },
    {
      "epoch": 0.7528786536758193,
      "grad_norm": 0.6499355435371399,
      "learning_rate": 7.528786536758194e-06,
      "loss": 1.0395,
      "step": 850
    },
    {
      "epoch": 0.7573073516386183,
      "grad_norm": 0.6279856562614441,
      "learning_rate": 7.573073516386183e-06,
      "loss": 1.0534,
      "step": 855
    },
    {
      "epoch": 0.7617360496014172,
      "grad_norm": 0.6102858185768127,
      "learning_rate": 7.6173604960141725e-06,
      "loss": 1.0939,
      "step": 860
    },
    {
      "epoch": 0.7661647475642162,
      "grad_norm": 0.5285583138465881,
      "learning_rate": 7.661647475642163e-06,
      "loss": 1.0494,
      "step": 865
    },
    {
      "epoch": 0.770593445527015,
      "grad_norm": 0.6597489714622498,
      "learning_rate": 7.705934455270151e-06,
      "loss": 1.0923,
      "step": 870
    },
    {
      "epoch": 0.775022143489814,
      "grad_norm": 0.37603759765625,
      "learning_rate": 7.750221434898142e-06,
      "loss": 1.0834,
      "step": 875
    },
    {
      "epoch": 0.7794508414526129,
      "grad_norm": 0.4725598394870758,
      "learning_rate": 7.79450841452613e-06,
      "loss": 1.0845,
      "step": 880
    },
    {
      "epoch": 0.7838795394154119,
      "grad_norm": 0.4172370731830597,
      "learning_rate": 7.838795394154119e-06,
      "loss": 1.0816,
      "step": 885
    },
    {
      "epoch": 0.7883082373782108,
      "grad_norm": 0.49854663014411926,
      "learning_rate": 7.883082373782109e-06,
      "loss": 1.0549,
      "step": 890
    },
    {
      "epoch": 0.7927369353410098,
      "grad_norm": 0.36657387018203735,
      "learning_rate": 7.927369353410098e-06,
      "loss": 1.0752,
      "step": 895
    },
    {
      "epoch": 0.7971656333038086,
      "grad_norm": 0.5820833444595337,
      "learning_rate": 7.971656333038086e-06,
      "loss": 1.052,
      "step": 900
    },
    {
      "epoch": 0.8015943312666076,
      "grad_norm": 0.6577838659286499,
      "learning_rate": 8.015943312666076e-06,
      "loss": 1.0638,
      "step": 905
    },
    {
      "epoch": 0.8060230292294066,
      "grad_norm": 0.6086249947547913,
      "learning_rate": 8.060230292294067e-06,
      "loss": 1.0278,
      "step": 910
    },
    {
      "epoch": 0.8104517271922055,
      "grad_norm": 0.7938503623008728,
      "learning_rate": 8.104517271922055e-06,
      "loss": 1.0165,
      "step": 915
    },
    {
      "epoch": 0.8148804251550045,
      "grad_norm": 0.7336840629577637,
      "learning_rate": 8.148804251550045e-06,
      "loss": 1.049,
      "step": 920
    },
    {
      "epoch": 0.8193091231178034,
      "grad_norm": 0.42411020398139954,
      "learning_rate": 8.193091231178034e-06,
      "loss": 1.0462,
      "step": 925
    },
    {
      "epoch": 0.8237378210806023,
      "grad_norm": 0.4607183337211609,
      "learning_rate": 8.237378210806024e-06,
      "loss": 1.0412,
      "step": 930
    },
    {
      "epoch": 0.8281665190434012,
      "grad_norm": 0.4874175488948822,
      "learning_rate": 8.281665190434013e-06,
      "loss": 1.0541,
      "step": 935
    },
    {
      "epoch": 0.8325952170062002,
      "grad_norm": 0.4741254448890686,
      "learning_rate": 8.325952170062003e-06,
      "loss": 1.0319,
      "step": 940
    },
    {
      "epoch": 0.8370239149689991,
      "grad_norm": 0.5217910408973694,
      "learning_rate": 8.370239149689992e-06,
      "loss": 1.0361,
      "step": 945
    },
    {
      "epoch": 0.8414526129317981,
      "grad_norm": 0.5650365948677063,
      "learning_rate": 8.414526129317982e-06,
      "loss": 1.0448,
      "step": 950
    },
    {
      "epoch": 0.845881310894597,
      "grad_norm": 0.4486577808856964,
      "learning_rate": 8.45881310894597e-06,
      "loss": 1.0569,
      "step": 955
    },
    {
      "epoch": 0.8503100088573959,
      "grad_norm": 0.5307552218437195,
      "learning_rate": 8.50310008857396e-06,
      "loss": 1.0653,
      "step": 960
    },
    {
      "epoch": 0.8547387068201948,
      "grad_norm": 0.8470759987831116,
      "learning_rate": 8.54738706820195e-06,
      "loss": 1.0722,
      "step": 965
    },
    {
      "epoch": 0.8591674047829938,
      "grad_norm": 0.6499133706092834,
      "learning_rate": 8.591674047829938e-06,
      "loss": 1.0423,
      "step": 970
    },
    {
      "epoch": 0.8635961027457927,
      "grad_norm": 0.46664857864379883,
      "learning_rate": 8.635961027457928e-06,
      "loss": 1.0776,
      "step": 975
    },
    {
      "epoch": 0.8680248007085917,
      "grad_norm": 0.4962954521179199,
      "learning_rate": 8.680248007085917e-06,
      "loss": 1.0191,
      "step": 980
    },
    {
      "epoch": 0.8724534986713907,
      "grad_norm": 0.6851896643638611,
      "learning_rate": 8.724534986713907e-06,
      "loss": 1.05,
      "step": 985
    },
    {
      "epoch": 0.8768821966341895,
      "grad_norm": 0.6915547251701355,
      "learning_rate": 8.768821966341896e-06,
      "loss": 1.0195,
      "step": 990
    },
    {
      "epoch": 0.8813108945969885,
      "grad_norm": 0.5120362639427185,
      "learning_rate": 8.813108945969886e-06,
      "loss": 1.0953,
      "step": 995
    },
    {
      "epoch": 0.8857395925597874,
      "grad_norm": 0.4280599057674408,
      "learning_rate": 8.857395925597874e-06,
      "loss": 1.067,
      "step": 1000
    },
    {
      "epoch": 0.8901682905225864,
      "grad_norm": 0.6475976705551147,
      "learning_rate": 8.901682905225865e-06,
      "loss": 1.1131,
      "step": 1005
    },
    {
      "epoch": 0.8945969884853853,
      "grad_norm": 0.5279582142829895,
      "learning_rate": 8.945969884853853e-06,
      "loss": 1.0436,
      "step": 1010
    },
    {
      "epoch": 0.8990256864481843,
      "grad_norm": 0.48513951897621155,
      "learning_rate": 8.990256864481844e-06,
      "loss": 1.074,
      "step": 1015
    },
    {
      "epoch": 0.9034543844109831,
      "grad_norm": 0.5839154124259949,
      "learning_rate": 9.034543844109832e-06,
      "loss": 1.035,
      "step": 1020
    },
    {
      "epoch": 0.9078830823737821,
      "grad_norm": 0.7936385273933411,
      "learning_rate": 9.078830823737822e-06,
      "loss": 1.0731,
      "step": 1025
    },
    {
      "epoch": 0.912311780336581,
      "grad_norm": 0.8670493960380554,
      "learning_rate": 9.123117803365811e-06,
      "loss": 1.065,
      "step": 1030
    },
    {
      "epoch": 0.91674047829938,
      "grad_norm": 0.6253863573074341,
      "learning_rate": 9.167404782993801e-06,
      "loss": 1.0698,
      "step": 1035
    },
    {
      "epoch": 0.9211691762621789,
      "grad_norm": 0.6447590589523315,
      "learning_rate": 9.21169176262179e-06,
      "loss": 1.0249,
      "step": 1040
    },
    {
      "epoch": 0.9255978742249779,
      "grad_norm": 0.6368715763092041,
      "learning_rate": 9.25597874224978e-06,
      "loss": 1.0293,
      "step": 1045
    },
    {
      "epoch": 0.9300265721877768,
      "grad_norm": 0.7156099081039429,
      "learning_rate": 9.300265721877769e-06,
      "loss": 1.0461,
      "step": 1050
    },
    {
      "epoch": 0.9344552701505757,
      "grad_norm": 0.7243468165397644,
      "learning_rate": 9.344552701505759e-06,
      "loss": 1.0236,
      "step": 1055
    },
    {
      "epoch": 0.9388839681133747,
      "grad_norm": 0.5831469893455505,
      "learning_rate": 9.388839681133747e-06,
      "loss": 1.0654,
      "step": 1060
    },
    {
      "epoch": 0.9433126660761736,
      "grad_norm": 0.6030169129371643,
      "learning_rate": 9.433126660761736e-06,
      "loss": 1.0206,
      "step": 1065
    },
    {
      "epoch": 0.9477413640389726,
      "grad_norm": 0.5703352093696594,
      "learning_rate": 9.477413640389726e-06,
      "loss": 1.0819,
      "step": 1070
    },
    {
      "epoch": 0.9521700620017715,
      "grad_norm": 0.6313313245773315,
      "learning_rate": 9.521700620017715e-06,
      "loss": 1.0658,
      "step": 1075
    },
    {
      "epoch": 0.9565987599645704,
      "grad_norm": 0.5377680659294128,
      "learning_rate": 9.565987599645705e-06,
      "loss": 1.0728,
      "step": 1080
    },
    {
      "epoch": 0.9610274579273693,
      "grad_norm": 0.5323534607887268,
      "learning_rate": 9.610274579273694e-06,
      "loss": 1.0578,
      "step": 1085
    },
    {
      "epoch": 0.9654561558901683,
      "grad_norm": 0.52129065990448,
      "learning_rate": 9.654561558901684e-06,
      "loss": 1.0116,
      "step": 1090
    },
    {
      "epoch": 0.9698848538529672,
      "grad_norm": 0.5387641191482544,
      "learning_rate": 9.698848538529672e-06,
      "loss": 1.0145,
      "step": 1095
    },
    {
      "epoch": 0.9743135518157662,
      "grad_norm": 0.8255786299705505,
      "learning_rate": 9.743135518157663e-06,
      "loss": 1.0599,
      "step": 1100
    },
    {
      "epoch": 0.978742249778565,
      "grad_norm": 0.8005098700523376,
      "learning_rate": 9.787422497785651e-06,
      "loss": 1.0571,
      "step": 1105
    },
    {
      "epoch": 0.983170947741364,
      "grad_norm": 0.5431801080703735,
      "learning_rate": 9.831709477413642e-06,
      "loss": 1.024,
      "step": 1110
    },
    {
      "epoch": 0.9875996457041629,
      "grad_norm": 0.6002717018127441,
      "learning_rate": 9.87599645704163e-06,
      "loss": 1.0321,
      "step": 1115
    },
    {
      "epoch": 0.9920283436669619,
      "grad_norm": 0.435141921043396,
      "learning_rate": 9.92028343666962e-06,
      "loss": 1.0913,
      "step": 1120
    },
    {
      "epoch": 0.9964570416297609,
      "grad_norm": 0.503257691860199,
      "learning_rate": 9.964570416297609e-06,
      "loss": 1.0394,
      "step": 1125
    },
    {
      "epoch": 1.0008857395925599,
      "grad_norm": 0.4300505816936493,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 1130
    },
    {
      "epoch": 1.0053144375553587,
      "grad_norm": 0.500792920589447,
      "learning_rate": 1e-05,
      "loss": 1.0992,
      "step": 1135
    },
    {
      "epoch": 1.0097431355181576,
      "grad_norm": 0.6591557264328003,
      "learning_rate": 1e-05,
      "loss": 1.0903,
      "step": 1140
    },
    {
      "epoch": 1.0141718334809566,
      "grad_norm": 0.4528239369392395,
      "learning_rate": 1e-05,
      "loss": 1.0596,
      "step": 1145
    },
    {
      "epoch": 1.0186005314437556,
      "grad_norm": 0.5716541409492493,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 1150
    },
    {
      "epoch": 1.0230292294065544,
      "grad_norm": 0.5290917754173279,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 1155
    },
    {
      "epoch": 1.0274579273693534,
      "grad_norm": 0.6618760824203491,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 1160
    },
    {
      "epoch": 1.0318866253321524,
      "grad_norm": 0.591672956943512,
      "learning_rate": 1e-05,
      "loss": 1.0734,
      "step": 1165
    },
    {
      "epoch": 1.0363153232949514,
      "grad_norm": 0.4395208954811096,
      "learning_rate": 1e-05,
      "loss": 1.0354,
      "step": 1170
    },
    {
      "epoch": 1.0407440212577501,
      "grad_norm": 0.5045232176780701,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 1175
    },
    {
      "epoch": 1.045172719220549,
      "grad_norm": 0.48325854539871216,
      "learning_rate": 1e-05,
      "loss": 1.0253,
      "step": 1180
    },
    {
      "epoch": 1.049601417183348,
      "grad_norm": 0.5216041803359985,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 1185
    },
    {
      "epoch": 1.054030115146147,
      "grad_norm": 0.7825465202331543,
      "learning_rate": 1e-05,
      "loss": 1.0839,
      "step": 1190
    },
    {
      "epoch": 1.058458813108946,
      "grad_norm": 0.46927815675735474,
      "learning_rate": 1e-05,
      "loss": 1.022,
      "step": 1195
    },
    {
      "epoch": 1.0628875110717448,
      "grad_norm": 0.5872330069541931,
      "learning_rate": 1e-05,
      "loss": 1.0285,
      "step": 1200
    },
    {
      "epoch": 1.0673162090345438,
      "grad_norm": 0.5527483820915222,
      "learning_rate": 1e-05,
      "loss": 1.0337,
      "step": 1205
    },
    {
      "epoch": 1.0717449069973428,
      "grad_norm": 0.8837293386459351,
      "learning_rate": 1e-05,
      "loss": 1.0523,
      "step": 1210
    },
    {
      "epoch": 1.0761736049601418,
      "grad_norm": 0.6395681500434875,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 1215
    },
    {
      "epoch": 1.0806023029229406,
      "grad_norm": 0.5509293079376221,
      "learning_rate": 1e-05,
      "loss": 1.0468,
      "step": 1220
    },
    {
      "epoch": 1.0850310008857396,
      "grad_norm": 0.593695342540741,
      "learning_rate": 1e-05,
      "loss": 1.0534,
      "step": 1225
    },
    {
      "epoch": 1.0894596988485385,
      "grad_norm": 0.5877532958984375,
      "learning_rate": 1e-05,
      "loss": 1.0454,
      "step": 1230
    },
    {
      "epoch": 1.0938883968113375,
      "grad_norm": 0.7444388270378113,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 1235
    },
    {
      "epoch": 1.0983170947741363,
      "grad_norm": 0.5822707414627075,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 1240
    },
    {
      "epoch": 1.1027457927369353,
      "grad_norm": 0.5833282470703125,
      "learning_rate": 1e-05,
      "loss": 1.0657,
      "step": 1245
    },
    {
      "epoch": 1.1071744906997343,
      "grad_norm": 0.6106334328651428,
      "learning_rate": 1e-05,
      "loss": 1.0895,
      "step": 1250
    },
    {
      "epoch": 1.1116031886625333,
      "grad_norm": 0.5250518321990967,
      "learning_rate": 1e-05,
      "loss": 1.04,
      "step": 1255
    },
    {
      "epoch": 1.1160318866253323,
      "grad_norm": 0.5018734335899353,
      "learning_rate": 1e-05,
      "loss": 1.0247,
      "step": 1260
    },
    {
      "epoch": 1.120460584588131,
      "grad_norm": 0.5880675911903381,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 1265
    },
    {
      "epoch": 1.12488928255093,
      "grad_norm": 0.39481469988822937,
      "learning_rate": 1e-05,
      "loss": 1.0912,
      "step": 1270
    },
    {
      "epoch": 1.129317980513729,
      "grad_norm": 0.5716440677642822,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 1275
    },
    {
      "epoch": 1.133746678476528,
      "grad_norm": 0.6783057451248169,
      "learning_rate": 1e-05,
      "loss": 1.0606,
      "step": 1280
    },
    {
      "epoch": 1.1381753764393268,
      "grad_norm": 0.758237361907959,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 1285
    },
    {
      "epoch": 1.1426040744021257,
      "grad_norm": 0.6016380190849304,
      "learning_rate": 1e-05,
      "loss": 1.0515,
      "step": 1290
    },
    {
      "epoch": 1.1470327723649247,
      "grad_norm": 0.5406017303466797,
      "learning_rate": 1e-05,
      "loss": 1.0224,
      "step": 1295
    },
    {
      "epoch": 1.1514614703277237,
      "grad_norm": 0.6456558704376221,
      "learning_rate": 1e-05,
      "loss": 1.0803,
      "step": 1300
    },
    {
      "epoch": 1.1558901682905225,
      "grad_norm": 0.5820453763008118,
      "learning_rate": 1e-05,
      "loss": 1.0439,
      "step": 1305
    },
    {
      "epoch": 1.1603188662533215,
      "grad_norm": 0.6696756482124329,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 1310
    },
    {
      "epoch": 1.1647475642161205,
      "grad_norm": 0.648473858833313,
      "learning_rate": 1e-05,
      "loss": 1.0683,
      "step": 1315
    },
    {
      "epoch": 1.1691762621789195,
      "grad_norm": 0.6276561617851257,
      "learning_rate": 1e-05,
      "loss": 1.1075,
      "step": 1320
    },
    {
      "epoch": 1.1736049601417182,
      "grad_norm": 0.6598888635635376,
      "learning_rate": 1e-05,
      "loss": 1.0439,
      "step": 1325
    },
    {
      "epoch": 1.1780336581045172,
      "grad_norm": 0.9274353384971619,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 1330
    },
    {
      "epoch": 1.1824623560673162,
      "grad_norm": 0.5601986050605774,
      "learning_rate": 1e-05,
      "loss": 1.0467,
      "step": 1335
    },
    {
      "epoch": 1.1868910540301152,
      "grad_norm": 0.42352792620658875,
      "learning_rate": 1e-05,
      "loss": 1.0422,
      "step": 1340
    },
    {
      "epoch": 1.1913197519929142,
      "grad_norm": 0.6717144250869751,
      "learning_rate": 1e-05,
      "loss": 1.0522,
      "step": 1345
    },
    {
      "epoch": 1.195748449955713,
      "grad_norm": 0.6188071966171265,
      "learning_rate": 1e-05,
      "loss": 1.0732,
      "step": 1350
    },
    {
      "epoch": 1.200177147918512,
      "grad_norm": 0.5769972205162048,
      "learning_rate": 1e-05,
      "loss": 1.0529,
      "step": 1355
    },
    {
      "epoch": 1.204605845881311,
      "grad_norm": 0.5368040204048157,
      "learning_rate": 1e-05,
      "loss": 1.0637,
      "step": 1360
    },
    {
      "epoch": 1.20903454384411,
      "grad_norm": 0.4155511260032654,
      "learning_rate": 1e-05,
      "loss": 1.0876,
      "step": 1365
    },
    {
      "epoch": 1.2134632418069087,
      "grad_norm": 0.5935255885124207,
      "learning_rate": 1e-05,
      "loss": 1.0483,
      "step": 1370
    },
    {
      "epoch": 1.2178919397697077,
      "grad_norm": 0.4535612165927887,
      "learning_rate": 1e-05,
      "loss": 1.0846,
      "step": 1375
    },
    {
      "epoch": 1.2223206377325067,
      "grad_norm": 0.6879892945289612,
      "learning_rate": 1e-05,
      "loss": 1.0726,
      "step": 1380
    },
    {
      "epoch": 1.2267493356953056,
      "grad_norm": 0.380599707365036,
      "learning_rate": 1e-05,
      "loss": 1.0291,
      "step": 1385
    },
    {
      "epoch": 1.2311780336581046,
      "grad_norm": 0.4734128713607788,
      "learning_rate": 1e-05,
      "loss": 1.0602,
      "step": 1390
    },
    {
      "epoch": 1.2356067316209034,
      "grad_norm": 0.6638430953025818,
      "learning_rate": 1e-05,
      "loss": 1.0537,
      "step": 1395
    },
    {
      "epoch": 1.2400354295837024,
      "grad_norm": 0.5650917291641235,
      "learning_rate": 1e-05,
      "loss": 1.0764,
      "step": 1400
    },
    {
      "epoch": 1.2444641275465014,
      "grad_norm": 0.4994293749332428,
      "learning_rate": 1e-05,
      "loss": 1.0738,
      "step": 1405
    },
    {
      "epoch": 1.2488928255093001,
      "grad_norm": 0.7376642823219299,
      "learning_rate": 1e-05,
      "loss": 1.0753,
      "step": 1410
    },
    {
      "epoch": 1.2533215234720991,
      "grad_norm": 0.45495113730430603,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 1415
    },
    {
      "epoch": 1.2577502214348981,
      "grad_norm": 0.5402055978775024,
      "learning_rate": 1e-05,
      "loss": 1.0506,
      "step": 1420
    },
    {
      "epoch": 1.262178919397697,
      "grad_norm": 0.9967095255851746,
      "learning_rate": 1e-05,
      "loss": 1.054,
      "step": 1425
    },
    {
      "epoch": 1.266607617360496,
      "grad_norm": 0.732774555683136,
      "learning_rate": 1e-05,
      "loss": 1.0701,
      "step": 1430
    },
    {
      "epoch": 1.271036315323295,
      "grad_norm": 0.7219616174697876,
      "learning_rate": 1e-05,
      "loss": 1.0511,
      "step": 1435
    },
    {
      "epoch": 1.2754650132860939,
      "grad_norm": 0.6792771220207214,
      "learning_rate": 1e-05,
      "loss": 1.0757,
      "step": 1440
    },
    {
      "epoch": 1.2798937112488928,
      "grad_norm": 0.5615405440330505,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 1445
    },
    {
      "epoch": 1.2843224092116918,
      "grad_norm": 0.4365237355232239,
      "learning_rate": 1e-05,
      "loss": 1.035,
      "step": 1450
    },
    {
      "epoch": 1.2887511071744906,
      "grad_norm": 0.418634831905365,
      "learning_rate": 1e-05,
      "loss": 1.033,
      "step": 1455
    },
    {
      "epoch": 1.2931798051372896,
      "grad_norm": 0.4341377913951874,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 1460
    },
    {
      "epoch": 1.2976085031000886,
      "grad_norm": 0.4297645092010498,
      "learning_rate": 1e-05,
      "loss": 1.0219,
      "step": 1465
    },
    {
      "epoch": 1.3020372010628876,
      "grad_norm": 0.44168952107429504,
      "learning_rate": 1e-05,
      "loss": 1.0329,
      "step": 1470
    },
    {
      "epoch": 1.3064658990256866,
      "grad_norm": 0.565904438495636,
      "learning_rate": 1e-05,
      "loss": 1.0327,
      "step": 1475
    },
    {
      "epoch": 1.3108945969884853,
      "grad_norm": 0.566770076751709,
      "learning_rate": 1e-05,
      "loss": 1.0554,
      "step": 1480
    },
    {
      "epoch": 1.3153232949512843,
      "grad_norm": 0.4992661774158478,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 1485
    },
    {
      "epoch": 1.3197519929140833,
      "grad_norm": 0.550749659538269,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 1490
    },
    {
      "epoch": 1.324180690876882,
      "grad_norm": 0.5356817245483398,
      "learning_rate": 1e-05,
      "loss": 1.0366,
      "step": 1495
    },
    {
      "epoch": 1.328609388839681,
      "grad_norm": 0.565149188041687,
      "learning_rate": 1e-05,
      "loss": 1.061,
      "step": 1500
    },
    {
      "epoch": 1.33303808680248,
      "grad_norm": 0.46444451808929443,
      "learning_rate": 1e-05,
      "loss": 1.0806,
      "step": 1505
    },
    {
      "epoch": 1.337466784765279,
      "grad_norm": 0.4967142343521118,
      "learning_rate": 1e-05,
      "loss": 1.0231,
      "step": 1510
    },
    {
      "epoch": 1.341895482728078,
      "grad_norm": 0.5111200213432312,
      "learning_rate": 1e-05,
      "loss": 1.0646,
      "step": 1515
    },
    {
      "epoch": 1.346324180690877,
      "grad_norm": 0.41596952080726624,
      "learning_rate": 1e-05,
      "loss": 1.0477,
      "step": 1520
    },
    {
      "epoch": 1.3507528786536758,
      "grad_norm": 0.476852685213089,
      "learning_rate": 1e-05,
      "loss": 1.0781,
      "step": 1525
    },
    {
      "epoch": 1.3551815766164748,
      "grad_norm": 0.5790765285491943,
      "learning_rate": 1e-05,
      "loss": 1.0344,
      "step": 1530
    },
    {
      "epoch": 1.3596102745792737,
      "grad_norm": 0.5472508072853088,
      "learning_rate": 1e-05,
      "loss": 1.0824,
      "step": 1535
    },
    {
      "epoch": 1.3640389725420725,
      "grad_norm": 0.5851724743843079,
      "learning_rate": 1e-05,
      "loss": 1.0261,
      "step": 1540
    },
    {
      "epoch": 1.3684676705048715,
      "grad_norm": 0.49602991342544556,
      "learning_rate": 1e-05,
      "loss": 1.04,
      "step": 1545
    },
    {
      "epoch": 1.3728963684676705,
      "grad_norm": 0.48241379857063293,
      "learning_rate": 1e-05,
      "loss": 1.1002,
      "step": 1550
    },
    {
      "epoch": 1.3773250664304695,
      "grad_norm": 0.4322584867477417,
      "learning_rate": 1e-05,
      "loss": 1.0923,
      "step": 1555
    },
    {
      "epoch": 1.3817537643932685,
      "grad_norm": 0.3377690613269806,
      "learning_rate": 1e-05,
      "loss": 1.0696,
      "step": 1560
    },
    {
      "epoch": 1.3861824623560672,
      "grad_norm": 0.4481833875179291,
      "learning_rate": 1e-05,
      "loss": 1.0519,
      "step": 1565
    },
    {
      "epoch": 1.3906111603188662,
      "grad_norm": 0.6130571961402893,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 1570
    },
    {
      "epoch": 1.3950398582816652,
      "grad_norm": 0.6391801238059998,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 1575
    },
    {
      "epoch": 1.3994685562444642,
      "grad_norm": 0.8034498691558838,
      "learning_rate": 1e-05,
      "loss": 1.0651,
      "step": 1580
    },
    {
      "epoch": 1.403897254207263,
      "grad_norm": 0.5643225312232971,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 1585
    },
    {
      "epoch": 1.408325952170062,
      "grad_norm": 0.6384730339050293,
      "learning_rate": 1e-05,
      "loss": 1.0725,
      "step": 1590
    },
    {
      "epoch": 1.412754650132861,
      "grad_norm": 0.8570442795753479,
      "learning_rate": 1e-05,
      "loss": 1.0485,
      "step": 1595
    },
    {
      "epoch": 1.41718334809566,
      "grad_norm": 0.4651219844818115,
      "learning_rate": 1e-05,
      "loss": 1.1197,
      "step": 1600
    },
    {
      "epoch": 1.421612046058459,
      "grad_norm": 0.6274141669273376,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 1605
    },
    {
      "epoch": 1.4260407440212577,
      "grad_norm": 0.6916234493255615,
      "learning_rate": 1e-05,
      "loss": 1.0658,
      "step": 1610
    },
    {
      "epoch": 1.4304694419840567,
      "grad_norm": 0.650422215461731,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 1615
    },
    {
      "epoch": 1.4348981399468557,
      "grad_norm": 0.6984378695487976,
      "learning_rate": 1e-05,
      "loss": 1.095,
      "step": 1620
    },
    {
      "epoch": 1.4393268379096544,
      "grad_norm": 0.5732066035270691,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 1625
    },
    {
      "epoch": 1.4437555358724534,
      "grad_norm": 0.4563065469264984,
      "learning_rate": 1e-05,
      "loss": 1.0247,
      "step": 1630
    },
    {
      "epoch": 1.4481842338352524,
      "grad_norm": 0.4518648386001587,
      "learning_rate": 1e-05,
      "loss": 1.0484,
      "step": 1635
    },
    {
      "epoch": 1.4526129317980514,
      "grad_norm": 0.5408558249473572,
      "learning_rate": 1e-05,
      "loss": 1.0337,
      "step": 1640
    },
    {
      "epoch": 1.4570416297608504,
      "grad_norm": 0.4345446228981018,
      "learning_rate": 1e-05,
      "loss": 1.037,
      "step": 1645
    },
    {
      "epoch": 1.4614703277236494,
      "grad_norm": 0.6300520300865173,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 1650
    },
    {
      "epoch": 1.4658990256864481,
      "grad_norm": 0.6697047352790833,
      "learning_rate": 1e-05,
      "loss": 1.0331,
      "step": 1655
    },
    {
      "epoch": 1.4703277236492471,
      "grad_norm": 0.5353518724441528,
      "learning_rate": 1e-05,
      "loss": 1.0606,
      "step": 1660
    },
    {
      "epoch": 1.4747564216120461,
      "grad_norm": 0.5491823554039001,
      "learning_rate": 1e-05,
      "loss": 1.032,
      "step": 1665
    },
    {
      "epoch": 1.4791851195748449,
      "grad_norm": 0.5808018445968628,
      "learning_rate": 1e-05,
      "loss": 1.0763,
      "step": 1670
    },
    {
      "epoch": 1.4836138175376439,
      "grad_norm": 0.4444665312767029,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 1675
    },
    {
      "epoch": 1.4880425155004429,
      "grad_norm": 0.5143681168556213,
      "learning_rate": 1e-05,
      "loss": 1.0296,
      "step": 1680
    },
    {
      "epoch": 1.4924712134632419,
      "grad_norm": 0.45289140939712524,
      "learning_rate": 1e-05,
      "loss": 1.0721,
      "step": 1685
    },
    {
      "epoch": 1.4968999114260408,
      "grad_norm": 0.608600914478302,
      "learning_rate": 1e-05,
      "loss": 1.0351,
      "step": 1690
    },
    {
      "epoch": 1.5013286093888398,
      "grad_norm": 0.6080148816108704,
      "learning_rate": 1e-05,
      "loss": 1.0666,
      "step": 1695
    },
    {
      "epoch": 1.5057573073516386,
      "grad_norm": 0.39750945568084717,
      "learning_rate": 1e-05,
      "loss": 1.1251,
      "step": 1700
    },
    {
      "epoch": 1.5101860053144376,
      "grad_norm": 0.6348580121994019,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 1705
    },
    {
      "epoch": 1.5146147032772364,
      "grad_norm": 0.5678335428237915,
      "learning_rate": 1e-05,
      "loss": 1.0744,
      "step": 1710
    },
    {
      "epoch": 1.5190434012400353,
      "grad_norm": 0.5787584781646729,
      "learning_rate": 1e-05,
      "loss": 1.0653,
      "step": 1715
    },
    {
      "epoch": 1.5234720992028343,
      "grad_norm": 0.44927048683166504,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 1720
    },
    {
      "epoch": 1.5279007971656333,
      "grad_norm": 0.38046860694885254,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 1725
    },
    {
      "epoch": 1.5323294951284323,
      "grad_norm": 0.4957244396209717,
      "learning_rate": 1e-05,
      "loss": 1.0357,
      "step": 1730
    },
    {
      "epoch": 1.5367581930912313,
      "grad_norm": 0.6060448884963989,
      "learning_rate": 1e-05,
      "loss": 1.0363,
      "step": 1735
    },
    {
      "epoch": 1.54118689105403,
      "grad_norm": 0.6205506920814514,
      "learning_rate": 1e-05,
      "loss": 1.0141,
      "step": 1740
    },
    {
      "epoch": 1.545615589016829,
      "grad_norm": 0.4897048771381378,
      "learning_rate": 1e-05,
      "loss": 1.1262,
      "step": 1745
    },
    {
      "epoch": 1.550044286979628,
      "grad_norm": 0.5701556205749512,
      "learning_rate": 1e-05,
      "loss": 1.0459,
      "step": 1750
    },
    {
      "epoch": 1.5544729849424268,
      "grad_norm": 0.4535682797431946,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 1755
    },
    {
      "epoch": 1.5589016829052258,
      "grad_norm": 0.5063267350196838,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 1760
    },
    {
      "epoch": 1.5633303808680248,
      "grad_norm": 0.4346560537815094,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 1765
    },
    {
      "epoch": 1.5677590788308238,
      "grad_norm": 0.7285047173500061,
      "learning_rate": 1e-05,
      "loss": 1.018,
      "step": 1770
    },
    {
      "epoch": 1.5721877767936228,
      "grad_norm": 0.7841598987579346,
      "learning_rate": 1e-05,
      "loss": 1.0508,
      "step": 1775
    },
    {
      "epoch": 1.5766164747564217,
      "grad_norm": 0.5126988887786865,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 1780
    },
    {
      "epoch": 1.5810451727192205,
      "grad_norm": 0.5649213194847107,
      "learning_rate": 1e-05,
      "loss": 1.1065,
      "step": 1785
    },
    {
      "epoch": 1.5854738706820195,
      "grad_norm": 0.4846344292163849,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 1790
    },
    {
      "epoch": 1.5899025686448183,
      "grad_norm": 0.4923091232776642,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 1795
    },
    {
      "epoch": 1.5943312666076173,
      "grad_norm": 0.4576951563358307,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 1800
    },
    {
      "epoch": 1.5987599645704162,
      "grad_norm": 0.6318348050117493,
      "learning_rate": 1e-05,
      "loss": 1.0603,
      "step": 1805
    },
    {
      "epoch": 1.6031886625332152,
      "grad_norm": 0.4736900329589844,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 1810
    },
    {
      "epoch": 1.6076173604960142,
      "grad_norm": 0.7124386429786682,
      "learning_rate": 1e-05,
      "loss": 1.0254,
      "step": 1815
    },
    {
      "epoch": 1.6120460584588132,
      "grad_norm": 0.6950289607048035,
      "learning_rate": 1e-05,
      "loss": 1.0501,
      "step": 1820
    },
    {
      "epoch": 1.6164747564216122,
      "grad_norm": 0.6133359670639038,
      "learning_rate": 1e-05,
      "loss": 1.0468,
      "step": 1825
    },
    {
      "epoch": 1.620903454384411,
      "grad_norm": 0.6645231246948242,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 1830
    },
    {
      "epoch": 1.62533215234721,
      "grad_norm": 0.41025203466415405,
      "learning_rate": 1e-05,
      "loss": 1.0513,
      "step": 1835
    },
    {
      "epoch": 1.6297608503100087,
      "grad_norm": 0.3575768768787384,
      "learning_rate": 1e-05,
      "loss": 1.0384,
      "step": 1840
    },
    {
      "epoch": 1.6341895482728077,
      "grad_norm": 0.7201238870620728,
      "learning_rate": 1e-05,
      "loss": 1.0273,
      "step": 1845
    },
    {
      "epoch": 1.6386182462356067,
      "grad_norm": 0.4535777270793915,
      "learning_rate": 1e-05,
      "loss": 1.0226,
      "step": 1850
    },
    {
      "epoch": 1.6430469441984057,
      "grad_norm": 0.7136516571044922,
      "learning_rate": 1e-05,
      "loss": 1.0704,
      "step": 1855
    },
    {
      "epoch": 1.6474756421612047,
      "grad_norm": 0.4246818423271179,
      "learning_rate": 1e-05,
      "loss": 1.0558,
      "step": 1860
    },
    {
      "epoch": 1.6519043401240037,
      "grad_norm": 0.4250449240207672,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 1865
    },
    {
      "epoch": 1.6563330380868024,
      "grad_norm": 0.5725380778312683,
      "learning_rate": 1e-05,
      "loss": 1.086,
      "step": 1870
    },
    {
      "epoch": 1.6607617360496014,
      "grad_norm": 0.5512462258338928,
      "learning_rate": 1e-05,
      "loss": 1.0546,
      "step": 1875
    },
    {
      "epoch": 1.6651904340124002,
      "grad_norm": 0.4659876525402069,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 1880
    },
    {
      "epoch": 1.6696191319751992,
      "grad_norm": 0.47894182801246643,
      "learning_rate": 1e-05,
      "loss": 1.0323,
      "step": 1885
    },
    {
      "epoch": 1.6740478299379982,
      "grad_norm": 0.6409828066825867,
      "learning_rate": 1e-05,
      "loss": 1.064,
      "step": 1890
    },
    {
      "epoch": 1.6784765279007972,
      "grad_norm": 0.5657796859741211,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 1895
    },
    {
      "epoch": 1.6829052258635961,
      "grad_norm": 0.5008763074874878,
      "learning_rate": 1e-05,
      "loss": 1.0149,
      "step": 1900
    },
    {
      "epoch": 1.6873339238263951,
      "grad_norm": 0.47699323296546936,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 1905
    },
    {
      "epoch": 1.6917626217891941,
      "grad_norm": 0.5299811959266663,
      "learning_rate": 1e-05,
      "loss": 1.0474,
      "step": 1910
    },
    {
      "epoch": 1.6961913197519929,
      "grad_norm": 0.47808408737182617,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 1915
    },
    {
      "epoch": 1.7006200177147919,
      "grad_norm": 0.5127821564674377,
      "learning_rate": 1e-05,
      "loss": 1.0615,
      "step": 1920
    },
    {
      "epoch": 1.7050487156775906,
      "grad_norm": 0.42068690061569214,
      "learning_rate": 1e-05,
      "loss": 1.066,
      "step": 1925
    },
    {
      "epoch": 1.7094774136403896,
      "grad_norm": 0.5648940801620483,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 1930
    },
    {
      "epoch": 1.7139061116031886,
      "grad_norm": 0.4861409664154053,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 1935
    },
    {
      "epoch": 1.7183348095659876,
      "grad_norm": 0.5075104832649231,
      "learning_rate": 1e-05,
      "loss": 1.1242,
      "step": 1940
    },
    {
      "epoch": 1.7227635075287866,
      "grad_norm": 0.5100075602531433,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 1945
    },
    {
      "epoch": 1.7271922054915856,
      "grad_norm": 0.4531674385070801,
      "learning_rate": 1e-05,
      "loss": 1.0312,
      "step": 1950
    },
    {
      "epoch": 1.7316209034543846,
      "grad_norm": 0.447305828332901,
      "learning_rate": 1e-05,
      "loss": 1.0537,
      "step": 1955
    },
    {
      "epoch": 1.7360496014171833,
      "grad_norm": 0.45211970806121826,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 1960
    },
    {
      "epoch": 1.7404782993799823,
      "grad_norm": 0.4799877405166626,
      "learning_rate": 1e-05,
      "loss": 1.0458,
      "step": 1965
    },
    {
      "epoch": 1.744906997342781,
      "grad_norm": 0.4379311501979828,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 1970
    },
    {
      "epoch": 1.74933569530558,
      "grad_norm": 0.45466145873069763,
      "learning_rate": 1e-05,
      "loss": 1.0432,
      "step": 1975
    },
    {
      "epoch": 1.753764393268379,
      "grad_norm": 0.368390828371048,
      "learning_rate": 1e-05,
      "loss": 1.0219,
      "step": 1980
    },
    {
      "epoch": 1.758193091231178,
      "grad_norm": 0.6489937901496887,
      "learning_rate": 1e-05,
      "loss": 1.0748,
      "step": 1985
    },
    {
      "epoch": 1.762621789193977,
      "grad_norm": 0.41698235273361206,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 1990
    },
    {
      "epoch": 1.767050487156776,
      "grad_norm": 0.4092991054058075,
      "learning_rate": 1e-05,
      "loss": 1.05,
      "step": 1995
    },
    {
      "epoch": 1.7714791851195748,
      "grad_norm": 0.360099732875824,
      "learning_rate": 1e-05,
      "loss": 1.0842,
      "step": 2000
    },
    {
      "epoch": 1.7759078830823738,
      "grad_norm": 0.34736168384552,
      "learning_rate": 1e-05,
      "loss": 1.0539,
      "step": 2005
    },
    {
      "epoch": 1.7803365810451726,
      "grad_norm": 0.3399425745010376,
      "learning_rate": 1e-05,
      "loss": 1.034,
      "step": 2010
    },
    {
      "epoch": 1.7847652790079716,
      "grad_norm": 0.5124260783195496,
      "learning_rate": 1e-05,
      "loss": 1.0546,
      "step": 2015
    },
    {
      "epoch": 1.7891939769707705,
      "grad_norm": 0.6677430868148804,
      "learning_rate": 1e-05,
      "loss": 1.0402,
      "step": 2020
    },
    {
      "epoch": 1.7936226749335695,
      "grad_norm": 0.5623106956481934,
      "learning_rate": 1e-05,
      "loss": 1.0666,
      "step": 2025
    },
    {
      "epoch": 1.7980513728963685,
      "grad_norm": 0.3347500264644623,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 2030
    },
    {
      "epoch": 1.8024800708591675,
      "grad_norm": 0.35451000928878784,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 2035
    },
    {
      "epoch": 1.8069087688219665,
      "grad_norm": 0.4026930332183838,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 2040
    },
    {
      "epoch": 1.8113374667847653,
      "grad_norm": 0.6083166003227234,
      "learning_rate": 1e-05,
      "loss": 1.0409,
      "step": 2045
    },
    {
      "epoch": 1.8157661647475642,
      "grad_norm": 0.3873279094696045,
      "learning_rate": 1e-05,
      "loss": 1.0847,
      "step": 2050
    },
    {
      "epoch": 1.820194862710363,
      "grad_norm": 0.7231238484382629,
      "learning_rate": 1e-05,
      "loss": 1.0247,
      "step": 2055
    },
    {
      "epoch": 1.824623560673162,
      "grad_norm": 0.4188913106918335,
      "learning_rate": 1e-05,
      "loss": 1.051,
      "step": 2060
    },
    {
      "epoch": 1.829052258635961,
      "grad_norm": 0.4547122120857239,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 2065
    },
    {
      "epoch": 1.83348095659876,
      "grad_norm": 0.47021517157554626,
      "learning_rate": 1e-05,
      "loss": 1.0363,
      "step": 2070
    },
    {
      "epoch": 1.837909654561559,
      "grad_norm": 0.5474349856376648,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 2075
    },
    {
      "epoch": 1.842338352524358,
      "grad_norm": 0.47033992409706116,
      "learning_rate": 1e-05,
      "loss": 1.0477,
      "step": 2080
    },
    {
      "epoch": 1.8467670504871567,
      "grad_norm": 0.6260009407997131,
      "learning_rate": 1e-05,
      "loss": 1.0587,
      "step": 2085
    },
    {
      "epoch": 1.8511957484499557,
      "grad_norm": 0.44886916875839233,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 2090
    },
    {
      "epoch": 1.8556244464127547,
      "grad_norm": 0.31309664249420166,
      "learning_rate": 1e-05,
      "loss": 1.0449,
      "step": 2095
    },
    {
      "epoch": 1.8600531443755535,
      "grad_norm": 0.4695584774017334,
      "learning_rate": 1e-05,
      "loss": 1.0642,
      "step": 2100
    },
    {
      "epoch": 1.8644818423383525,
      "grad_norm": 0.5193604826927185,
      "learning_rate": 1e-05,
      "loss": 1.0631,
      "step": 2105
    },
    {
      "epoch": 1.8689105403011514,
      "grad_norm": 0.5674470663070679,
      "learning_rate": 1e-05,
      "loss": 1.0733,
      "step": 2110
    },
    {
      "epoch": 1.8733392382639504,
      "grad_norm": 0.5005362033843994,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 2115
    },
    {
      "epoch": 1.8777679362267494,
      "grad_norm": 0.4786299467086792,
      "learning_rate": 1e-05,
      "loss": 1.0587,
      "step": 2120
    },
    {
      "epoch": 1.8821966341895484,
      "grad_norm": 0.31839287281036377,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 2125
    },
    {
      "epoch": 1.8866253321523472,
      "grad_norm": 0.6383420825004578,
      "learning_rate": 1e-05,
      "loss": 1.0354,
      "step": 2130
    },
    {
      "epoch": 1.8910540301151462,
      "grad_norm": 0.4370393753051758,
      "learning_rate": 1e-05,
      "loss": 1.0526,
      "step": 2135
    },
    {
      "epoch": 1.895482728077945,
      "grad_norm": 0.4346504211425781,
      "learning_rate": 1e-05,
      "loss": 1.0233,
      "step": 2140
    },
    {
      "epoch": 1.899911426040744,
      "grad_norm": 0.4525904059410095,
      "learning_rate": 1e-05,
      "loss": 1.0617,
      "step": 2145
    },
    {
      "epoch": 1.904340124003543,
      "grad_norm": 0.4848250448703766,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 2150
    },
    {
      "epoch": 1.908768821966342,
      "grad_norm": 0.4482121765613556,
      "learning_rate": 1e-05,
      "loss": 1.0218,
      "step": 2155
    },
    {
      "epoch": 1.9131975199291409,
      "grad_norm": 0.5617077946662903,
      "learning_rate": 1e-05,
      "loss": 1.0317,
      "step": 2160
    },
    {
      "epoch": 1.9176262178919399,
      "grad_norm": 0.4350849390029907,
      "learning_rate": 1e-05,
      "loss": 1.0902,
      "step": 2165
    },
    {
      "epoch": 1.9220549158547389,
      "grad_norm": 0.3970003128051758,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 2170
    },
    {
      "epoch": 1.9264836138175376,
      "grad_norm": 0.4444156587123871,
      "learning_rate": 1e-05,
      "loss": 1.0388,
      "step": 2175
    },
    {
      "epoch": 1.9309123117803366,
      "grad_norm": 0.6810417175292969,
      "learning_rate": 1e-05,
      "loss": 1.0351,
      "step": 2180
    },
    {
      "epoch": 1.9353410097431354,
      "grad_norm": 0.5398831367492676,
      "learning_rate": 1e-05,
      "loss": 1.0344,
      "step": 2185
    },
    {
      "epoch": 1.9397697077059344,
      "grad_norm": 0.39851683378219604,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 2190
    },
    {
      "epoch": 1.9441984056687334,
      "grad_norm": 0.47903212904930115,
      "learning_rate": 1e-05,
      "loss": 1.0439,
      "step": 2195
    },
    {
      "epoch": 1.9486271036315324,
      "grad_norm": 0.468469500541687,
      "learning_rate": 1e-05,
      "loss": 1.0225,
      "step": 2200
    },
    {
      "epoch": 1.9530558015943313,
      "grad_norm": 0.46060481667518616,
      "learning_rate": 1e-05,
      "loss": 1.0477,
      "step": 2205
    },
    {
      "epoch": 1.9574844995571303,
      "grad_norm": 0.45541658997535706,
      "learning_rate": 1e-05,
      "loss": 1.0601,
      "step": 2210
    },
    {
      "epoch": 1.961913197519929,
      "grad_norm": 0.42831093072891235,
      "learning_rate": 1e-05,
      "loss": 1.0615,
      "step": 2215
    },
    {
      "epoch": 1.966341895482728,
      "grad_norm": 0.283743679523468,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 2220
    },
    {
      "epoch": 1.9707705934455269,
      "grad_norm": 0.44202861189842224,
      "learning_rate": 1e-05,
      "loss": 1.0832,
      "step": 2225
    },
    {
      "epoch": 1.9751992914083258,
      "grad_norm": 0.4106939733028412,
      "learning_rate": 1e-05,
      "loss": 1.0506,
      "step": 2230
    },
    {
      "epoch": 1.9796279893711248,
      "grad_norm": 0.34922152757644653,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 2235
    },
    {
      "epoch": 1.9840566873339238,
      "grad_norm": 0.5265505909919739,
      "learning_rate": 1e-05,
      "loss": 1.0309,
      "step": 2240
    },
    {
      "epoch": 1.9884853852967228,
      "grad_norm": 0.5609035491943359,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 2245
    },
    {
      "epoch": 1.9929140832595218,
      "grad_norm": 0.5041390657424927,
      "learning_rate": 1e-05,
      "loss": 1.03,
      "step": 2250
    },
    {
      "epoch": 1.9973427812223208,
      "grad_norm": 0.6953744292259216,
      "learning_rate": 1e-05,
      "loss": 1.0603,
      "step": 2255
    },
    {
      "epoch": 2.0017714791851198,
      "grad_norm": 0.3584790527820587,
      "learning_rate": 1e-05,
      "loss": 1.0633,
      "step": 2260
    },
    {
      "epoch": 2.0062001771479183,
      "grad_norm": 0.4668038785457611,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 2265
    },
    {
      "epoch": 2.0106288751107173,
      "grad_norm": 0.44779878854751587,
      "learning_rate": 1e-05,
      "loss": 1.0322,
      "step": 2270
    },
    {
      "epoch": 2.0150575730735163,
      "grad_norm": 0.5731096267700195,
      "learning_rate": 1e-05,
      "loss": 1.0594,
      "step": 2275
    },
    {
      "epoch": 2.0194862710363153,
      "grad_norm": 0.5145208239555359,
      "learning_rate": 1e-05,
      "loss": 1.0314,
      "step": 2280
    },
    {
      "epoch": 2.0239149689991143,
      "grad_norm": 0.6408817768096924,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 2285
    },
    {
      "epoch": 2.0283436669619133,
      "grad_norm": 0.6031640768051147,
      "learning_rate": 1e-05,
      "loss": 1.0694,
      "step": 2290
    },
    {
      "epoch": 2.0327723649247122,
      "grad_norm": 0.6322607398033142,
      "learning_rate": 1e-05,
      "loss": 1.032,
      "step": 2295
    },
    {
      "epoch": 2.0372010628875112,
      "grad_norm": 0.6165448427200317,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 2300
    },
    {
      "epoch": 2.0416297608503102,
      "grad_norm": 0.34421566128730774,
      "learning_rate": 1e-05,
      "loss": 1.0773,
      "step": 2305
    },
    {
      "epoch": 2.0460584588131088,
      "grad_norm": 0.5919576287269592,
      "learning_rate": 1e-05,
      "loss": 1.0579,
      "step": 2310
    },
    {
      "epoch": 2.0504871567759078,
      "grad_norm": 0.773813009262085,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 2315
    },
    {
      "epoch": 2.0549158547387067,
      "grad_norm": 0.43329134583473206,
      "learning_rate": 1e-05,
      "loss": 1.057,
      "step": 2320
    },
    {
      "epoch": 2.0593445527015057,
      "grad_norm": 0.5484357476234436,
      "learning_rate": 1e-05,
      "loss": 1.0441,
      "step": 2325
    },
    {
      "epoch": 2.0637732506643047,
      "grad_norm": 0.4514033794403076,
      "learning_rate": 1e-05,
      "loss": 1.0415,
      "step": 2330
    },
    {
      "epoch": 2.0682019486271037,
      "grad_norm": 0.5863833427429199,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 2335
    },
    {
      "epoch": 2.0726306465899027,
      "grad_norm": 0.4827769696712494,
      "learning_rate": 1e-05,
      "loss": 1.0384,
      "step": 2340
    },
    {
      "epoch": 2.0770593445527017,
      "grad_norm": 0.4218099117279053,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 2345
    },
    {
      "epoch": 2.0814880425155002,
      "grad_norm": 0.49901875853538513,
      "learning_rate": 1e-05,
      "loss": 1.0729,
      "step": 2350
    },
    {
      "epoch": 2.0859167404782992,
      "grad_norm": 0.4048095643520355,
      "learning_rate": 1e-05,
      "loss": 1.0413,
      "step": 2355
    },
    {
      "epoch": 2.090345438441098,
      "grad_norm": 0.5221948027610779,
      "learning_rate": 1e-05,
      "loss": 1.0444,
      "step": 2360
    },
    {
      "epoch": 2.094774136403897,
      "grad_norm": 0.4813966155052185,
      "learning_rate": 1e-05,
      "loss": 1.0545,
      "step": 2365
    },
    {
      "epoch": 2.099202834366696,
      "grad_norm": 0.4535312056541443,
      "learning_rate": 1e-05,
      "loss": 1.0314,
      "step": 2370
    },
    {
      "epoch": 2.103631532329495,
      "grad_norm": 0.5918130874633789,
      "learning_rate": 1e-05,
      "loss": 1.0561,
      "step": 2375
    },
    {
      "epoch": 2.108060230292294,
      "grad_norm": 0.4560534656047821,
      "learning_rate": 1e-05,
      "loss": 1.0732,
      "step": 2380
    },
    {
      "epoch": 2.112488928255093,
      "grad_norm": 0.5319624543190002,
      "learning_rate": 1e-05,
      "loss": 1.0379,
      "step": 2385
    },
    {
      "epoch": 2.116917626217892,
      "grad_norm": 0.6658979654312134,
      "learning_rate": 1e-05,
      "loss": 1.018,
      "step": 2390
    },
    {
      "epoch": 2.1213463241806907,
      "grad_norm": 0.4870125651359558,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 2395
    },
    {
      "epoch": 2.1257750221434897,
      "grad_norm": 0.3799990117549896,
      "learning_rate": 1e-05,
      "loss": 1.0476,
      "step": 2400
    },
    {
      "epoch": 2.1302037201062887,
      "grad_norm": 0.42643675208091736,
      "learning_rate": 1e-05,
      "loss": 1.0433,
      "step": 2405
    },
    {
      "epoch": 2.1346324180690877,
      "grad_norm": 0.5843912959098816,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 2410
    },
    {
      "epoch": 2.1390611160318866,
      "grad_norm": 0.40163424611091614,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 2415
    },
    {
      "epoch": 2.1434898139946856,
      "grad_norm": 0.6528959274291992,
      "learning_rate": 1e-05,
      "loss": 1.038,
      "step": 2420
    },
    {
      "epoch": 2.1479185119574846,
      "grad_norm": 0.48753559589385986,
      "learning_rate": 1e-05,
      "loss": 1.0305,
      "step": 2425
    },
    {
      "epoch": 2.1523472099202836,
      "grad_norm": 0.4759777784347534,
      "learning_rate": 1e-05,
      "loss": 1.0776,
      "step": 2430
    },
    {
      "epoch": 2.156775907883082,
      "grad_norm": 0.4306032061576843,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 2435
    },
    {
      "epoch": 2.161204605845881,
      "grad_norm": 0.4647599458694458,
      "learning_rate": 1e-05,
      "loss": 1.0478,
      "step": 2440
    },
    {
      "epoch": 2.16563330380868,
      "grad_norm": 0.4583887755870819,
      "learning_rate": 1e-05,
      "loss": 1.0759,
      "step": 2445
    },
    {
      "epoch": 2.170062001771479,
      "grad_norm": 0.39155516028404236,
      "learning_rate": 1e-05,
      "loss": 1.0184,
      "step": 2450
    },
    {
      "epoch": 2.174490699734278,
      "grad_norm": 0.5356603264808655,
      "learning_rate": 1e-05,
      "loss": 1.0384,
      "step": 2455
    },
    {
      "epoch": 2.178919397697077,
      "grad_norm": 0.48529770970344543,
      "learning_rate": 1e-05,
      "loss": 1.0464,
      "step": 2460
    },
    {
      "epoch": 2.183348095659876,
      "grad_norm": 0.4828105568885803,
      "learning_rate": 1e-05,
      "loss": 1.0219,
      "step": 2465
    },
    {
      "epoch": 2.187776793622675,
      "grad_norm": 0.46481409668922424,
      "learning_rate": 1e-05,
      "loss": 1.0551,
      "step": 2470
    },
    {
      "epoch": 2.192205491585474,
      "grad_norm": 0.5605440139770508,
      "learning_rate": 1e-05,
      "loss": 1.0225,
      "step": 2475
    },
    {
      "epoch": 2.1966341895482726,
      "grad_norm": 0.5987666249275208,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 2480
    },
    {
      "epoch": 2.2010628875110716,
      "grad_norm": 0.33373814821243286,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 2485
    },
    {
      "epoch": 2.2054915854738706,
      "grad_norm": 0.5175365209579468,
      "learning_rate": 1e-05,
      "loss": 1.062,
      "step": 2490
    },
    {
      "epoch": 2.2099202834366696,
      "grad_norm": 0.45836082100868225,
      "learning_rate": 1e-05,
      "loss": 1.0729,
      "step": 2495
    },
    {
      "epoch": 2.2143489813994686,
      "grad_norm": 0.38386160135269165,
      "learning_rate": 1e-05,
      "loss": 1.0725,
      "step": 2500
    },
    {
      "epoch": 2.2187776793622676,
      "grad_norm": 0.44808462262153625,
      "learning_rate": 1e-05,
      "loss": 1.0641,
      "step": 2505
    },
    {
      "epoch": 2.2232063773250665,
      "grad_norm": 0.3938499391078949,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 2510
    },
    {
      "epoch": 2.2276350752878655,
      "grad_norm": 0.4187353253364563,
      "learning_rate": 1e-05,
      "loss": 1.093,
      "step": 2515
    },
    {
      "epoch": 2.2320637732506645,
      "grad_norm": 0.4338715672492981,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 2520
    },
    {
      "epoch": 2.236492471213463,
      "grad_norm": 0.8376438617706299,
      "learning_rate": 1e-05,
      "loss": 1.0319,
      "step": 2525
    },
    {
      "epoch": 2.240921169176262,
      "grad_norm": 0.5062357187271118,
      "learning_rate": 1e-05,
      "loss": 1.0377,
      "step": 2530
    },
    {
      "epoch": 2.245349867139061,
      "grad_norm": 0.4610828757286072,
      "learning_rate": 1e-05,
      "loss": 1.0726,
      "step": 2535
    },
    {
      "epoch": 2.24977856510186,
      "grad_norm": 0.47126075625419617,
      "learning_rate": 1e-05,
      "loss": 1.0618,
      "step": 2540
    },
    {
      "epoch": 2.254207263064659,
      "grad_norm": 0.4205315411090851,
      "learning_rate": 1e-05,
      "loss": 1.025,
      "step": 2545
    },
    {
      "epoch": 2.258635961027458,
      "grad_norm": 0.40285390615463257,
      "learning_rate": 1e-05,
      "loss": 1.0446,
      "step": 2550
    },
    {
      "epoch": 2.263064658990257,
      "grad_norm": 0.37645837664604187,
      "learning_rate": 1e-05,
      "loss": 1.0637,
      "step": 2555
    },
    {
      "epoch": 2.267493356953056,
      "grad_norm": 0.7137925624847412,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 2560
    },
    {
      "epoch": 2.271922054915855,
      "grad_norm": 0.558600664138794,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 2565
    },
    {
      "epoch": 2.2763507528786535,
      "grad_norm": 0.7317633032798767,
      "learning_rate": 1e-05,
      "loss": 1.0795,
      "step": 2570
    },
    {
      "epoch": 2.2807794508414525,
      "grad_norm": 0.4917280077934265,
      "learning_rate": 1e-05,
      "loss": 1.0471,
      "step": 2575
    },
    {
      "epoch": 2.2852081488042515,
      "grad_norm": 0.40624135732650757,
      "learning_rate": 1e-05,
      "loss": 1.0626,
      "step": 2580
    },
    {
      "epoch": 2.2896368467670505,
      "grad_norm": 0.5770871639251709,
      "learning_rate": 1e-05,
      "loss": 1.0705,
      "step": 2585
    },
    {
      "epoch": 2.2940655447298495,
      "grad_norm": 0.3286740779876709,
      "learning_rate": 1e-05,
      "loss": 1.0578,
      "step": 2590
    },
    {
      "epoch": 2.2984942426926485,
      "grad_norm": 0.6179378628730774,
      "learning_rate": 1e-05,
      "loss": 1.0379,
      "step": 2595
    },
    {
      "epoch": 2.3029229406554474,
      "grad_norm": 0.3655185103416443,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 2600
    },
    {
      "epoch": 2.307351638618246,
      "grad_norm": 0.45200085639953613,
      "learning_rate": 1e-05,
      "loss": 1.059,
      "step": 2605
    },
    {
      "epoch": 2.311780336581045,
      "grad_norm": 0.3856170177459717,
      "learning_rate": 1e-05,
      "loss": 1.0273,
      "step": 2610
    },
    {
      "epoch": 2.316209034543844,
      "grad_norm": 0.3704911470413208,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 2615
    },
    {
      "epoch": 2.320637732506643,
      "grad_norm": 0.5131767988204956,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 2620
    },
    {
      "epoch": 2.325066430469442,
      "grad_norm": 0.4233784079551697,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 2625
    },
    {
      "epoch": 2.329495128432241,
      "grad_norm": 0.524312436580658,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 2630
    },
    {
      "epoch": 2.33392382639504,
      "grad_norm": 0.5103597044944763,
      "learning_rate": 1e-05,
      "loss": 1.093,
      "step": 2635
    },
    {
      "epoch": 2.338352524357839,
      "grad_norm": 0.5735695958137512,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 2640
    },
    {
      "epoch": 2.342781222320638,
      "grad_norm": 0.5327978134155273,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 2645
    },
    {
      "epoch": 2.3472099202834364,
      "grad_norm": 0.4777471721172333,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 2650
    },
    {
      "epoch": 2.3516386182462354,
      "grad_norm": 0.3995833694934845,
      "learning_rate": 1e-05,
      "loss": 1.031,
      "step": 2655
    },
    {
      "epoch": 2.3560673162090344,
      "grad_norm": 0.5510890483856201,
      "learning_rate": 1e-05,
      "loss": 1.0196,
      "step": 2660
    },
    {
      "epoch": 2.3604960141718334,
      "grad_norm": 0.4626731872558594,
      "learning_rate": 1e-05,
      "loss": 1.0977,
      "step": 2665
    },
    {
      "epoch": 2.3649247121346324,
      "grad_norm": 0.36934515833854675,
      "learning_rate": 1e-05,
      "loss": 1.0728,
      "step": 2670
    },
    {
      "epoch": 2.3693534100974314,
      "grad_norm": 0.5003888607025146,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 2675
    },
    {
      "epoch": 2.3737821080602304,
      "grad_norm": 0.4252766966819763,
      "learning_rate": 1e-05,
      "loss": 1.0376,
      "step": 2680
    },
    {
      "epoch": 2.3782108060230294,
      "grad_norm": 0.376730352640152,
      "learning_rate": 1e-05,
      "loss": 1.0314,
      "step": 2685
    },
    {
      "epoch": 2.3826395039858284,
      "grad_norm": 0.43287017941474915,
      "learning_rate": 1e-05,
      "loss": 1.0538,
      "step": 2690
    },
    {
      "epoch": 2.387068201948627,
      "grad_norm": 0.4191151261329651,
      "learning_rate": 1e-05,
      "loss": 1.073,
      "step": 2695
    },
    {
      "epoch": 2.391496899911426,
      "grad_norm": 0.3899643123149872,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 2700
    },
    {
      "epoch": 2.395925597874225,
      "grad_norm": 0.49577733874320984,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 2705
    },
    {
      "epoch": 2.400354295837024,
      "grad_norm": 0.5975940227508545,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 2710
    },
    {
      "epoch": 2.404782993799823,
      "grad_norm": 0.4224880635738373,
      "learning_rate": 1e-05,
      "loss": 1.0275,
      "step": 2715
    },
    {
      "epoch": 2.409211691762622,
      "grad_norm": 0.5402734279632568,
      "learning_rate": 1e-05,
      "loss": 1.0528,
      "step": 2720
    },
    {
      "epoch": 2.413640389725421,
      "grad_norm": 0.31992432475090027,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 2725
    },
    {
      "epoch": 2.41806908768822,
      "grad_norm": 0.32328641414642334,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 2730
    },
    {
      "epoch": 2.422497785651019,
      "grad_norm": 0.3336014151573181,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 2735
    },
    {
      "epoch": 2.4269264836138174,
      "grad_norm": 0.5074469447135925,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 2740
    },
    {
      "epoch": 2.4313551815766163,
      "grad_norm": 0.5504379272460938,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 2745
    },
    {
      "epoch": 2.4357838795394153,
      "grad_norm": 0.4262321889400482,
      "learning_rate": 1e-05,
      "loss": 1.0344,
      "step": 2750
    },
    {
      "epoch": 2.4402125775022143,
      "grad_norm": 0.5119176506996155,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 2755
    },
    {
      "epoch": 2.4446412754650133,
      "grad_norm": 0.3485449552536011,
      "learning_rate": 1e-05,
      "loss": 1.0961,
      "step": 2760
    },
    {
      "epoch": 2.4490699734278123,
      "grad_norm": 0.29425960779190063,
      "learning_rate": 1e-05,
      "loss": 1.0491,
      "step": 2765
    },
    {
      "epoch": 2.4534986713906113,
      "grad_norm": 0.3820127844810486,
      "learning_rate": 1e-05,
      "loss": 1.0248,
      "step": 2770
    },
    {
      "epoch": 2.4579273693534103,
      "grad_norm": 0.3655147850513458,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 2775
    },
    {
      "epoch": 2.4623560673162093,
      "grad_norm": 0.4104725420475006,
      "learning_rate": 1e-05,
      "loss": 1.0458,
      "step": 2780
    },
    {
      "epoch": 2.466784765279008,
      "grad_norm": 0.595163106918335,
      "learning_rate": 1e-05,
      "loss": 1.0521,
      "step": 2785
    },
    {
      "epoch": 2.471213463241807,
      "grad_norm": 0.6888088583946228,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 2790
    },
    {
      "epoch": 2.475642161204606,
      "grad_norm": 0.46057769656181335,
      "learning_rate": 1e-05,
      "loss": 1.0485,
      "step": 2795
    },
    {
      "epoch": 2.4800708591674048,
      "grad_norm": 0.5151911973953247,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 2800
    },
    {
      "epoch": 2.4844995571302038,
      "grad_norm": 0.5031762719154358,
      "learning_rate": 1e-05,
      "loss": 1.0411,
      "step": 2805
    },
    {
      "epoch": 2.4889282550930028,
      "grad_norm": 0.4458197355270386,
      "learning_rate": 1e-05,
      "loss": 1.0447,
      "step": 2810
    },
    {
      "epoch": 2.4933569530558017,
      "grad_norm": 0.4074757397174835,
      "learning_rate": 1e-05,
      "loss": 1.0566,
      "step": 2815
    },
    {
      "epoch": 2.4977856510186003,
      "grad_norm": 0.5228142738342285,
      "learning_rate": 1e-05,
      "loss": 1.0889,
      "step": 2820
    },
    {
      "epoch": 2.5022143489813997,
      "grad_norm": 0.47482267022132874,
      "learning_rate": 1e-05,
      "loss": 1.0413,
      "step": 2825
    },
    {
      "epoch": 2.5066430469441983,
      "grad_norm": 0.6530681252479553,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 2830
    },
    {
      "epoch": 2.5110717449069972,
      "grad_norm": 0.5544772148132324,
      "learning_rate": 1e-05,
      "loss": 1.0779,
      "step": 2835
    },
    {
      "epoch": 2.5155004428697962,
      "grad_norm": 0.5468206405639648,
      "learning_rate": 1e-05,
      "loss": 1.0152,
      "step": 2840
    },
    {
      "epoch": 2.5199291408325952,
      "grad_norm": 0.42827850580215454,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 2845
    },
    {
      "epoch": 2.524357838795394,
      "grad_norm": 0.4947317838668823,
      "learning_rate": 1e-05,
      "loss": 1.0598,
      "step": 2850
    },
    {
      "epoch": 2.528786536758193,
      "grad_norm": 0.4925984740257263,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 2855
    },
    {
      "epoch": 2.533215234720992,
      "grad_norm": 0.3933313190937042,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 2860
    },
    {
      "epoch": 2.5376439326837907,
      "grad_norm": 0.4389501214027405,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 2865
    },
    {
      "epoch": 2.54207263064659,
      "grad_norm": 0.40963491797447205,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 2870
    },
    {
      "epoch": 2.5465013286093887,
      "grad_norm": 0.37005141377449036,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 2875
    },
    {
      "epoch": 2.5509300265721877,
      "grad_norm": 0.5001286864280701,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 2880
    },
    {
      "epoch": 2.5553587245349867,
      "grad_norm": 0.38077735900878906,
      "learning_rate": 1e-05,
      "loss": 1.0412,
      "step": 2885
    },
    {
      "epoch": 2.5597874224977857,
      "grad_norm": 0.5739526748657227,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 2890
    },
    {
      "epoch": 2.5642161204605847,
      "grad_norm": 0.5770562291145325,
      "learning_rate": 1e-05,
      "loss": 1.0938,
      "step": 2895
    },
    {
      "epoch": 2.5686448184233837,
      "grad_norm": 0.40662881731987,
      "learning_rate": 1e-05,
      "loss": 1.0354,
      "step": 2900
    },
    {
      "epoch": 2.5730735163861826,
      "grad_norm": 0.39924120903015137,
      "learning_rate": 1e-05,
      "loss": 1.0447,
      "step": 2905
    },
    {
      "epoch": 2.577502214348981,
      "grad_norm": 0.4533263146877289,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 2910
    },
    {
      "epoch": 2.58193091231178,
      "grad_norm": 0.5112301111221313,
      "learning_rate": 1e-05,
      "loss": 1.0464,
      "step": 2915
    },
    {
      "epoch": 2.586359610274579,
      "grad_norm": 0.4359137713909149,
      "learning_rate": 1e-05,
      "loss": 1.0375,
      "step": 2920
    },
    {
      "epoch": 2.590788308237378,
      "grad_norm": 0.4898315370082855,
      "learning_rate": 1e-05,
      "loss": 1.0335,
      "step": 2925
    },
    {
      "epoch": 2.595217006200177,
      "grad_norm": 0.43125033378601074,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 2930
    },
    {
      "epoch": 2.599645704162976,
      "grad_norm": 0.5227671265602112,
      "learning_rate": 1e-05,
      "loss": 1.0986,
      "step": 2935
    },
    {
      "epoch": 2.604074402125775,
      "grad_norm": 0.4317053556442261,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 2940
    },
    {
      "epoch": 2.608503100088574,
      "grad_norm": 0.425248384475708,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 2945
    },
    {
      "epoch": 2.612931798051373,
      "grad_norm": 0.5229920744895935,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 2950
    },
    {
      "epoch": 2.6173604960141716,
      "grad_norm": 0.3360884189605713,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 2955
    },
    {
      "epoch": 2.6217891939769706,
      "grad_norm": 0.3374822437763214,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 2960
    },
    {
      "epoch": 2.6262178919397696,
      "grad_norm": 0.4743426740169525,
      "learning_rate": 1e-05,
      "loss": 1.0885,
      "step": 2965
    },
    {
      "epoch": 2.6306465899025686,
      "grad_norm": 0.43363791704177856,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 2970
    },
    {
      "epoch": 2.6350752878653676,
      "grad_norm": 0.3984227776527405,
      "learning_rate": 1e-05,
      "loss": 1.039,
      "step": 2975
    },
    {
      "epoch": 2.6395039858281666,
      "grad_norm": 0.5370942950248718,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 2980
    },
    {
      "epoch": 2.6439326837909656,
      "grad_norm": 0.39962244033813477,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 2985
    },
    {
      "epoch": 2.648361381753764,
      "grad_norm": 0.40735554695129395,
      "learning_rate": 1e-05,
      "loss": 1.0281,
      "step": 2990
    },
    {
      "epoch": 2.6527900797165636,
      "grad_norm": 0.46149295568466187,
      "learning_rate": 1e-05,
      "loss": 1.036,
      "step": 2995
    },
    {
      "epoch": 2.657218777679362,
      "grad_norm": 0.4877465069293976,
      "learning_rate": 1e-05,
      "loss": 1.0401,
      "step": 3000
    },
    {
      "epoch": 2.661647475642161,
      "grad_norm": 0.5125572085380554,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 3005
    },
    {
      "epoch": 2.66607617360496,
      "grad_norm": 0.5010902881622314,
      "learning_rate": 1e-05,
      "loss": 1.0656,
      "step": 3010
    },
    {
      "epoch": 2.670504871567759,
      "grad_norm": 0.507574200630188,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 3015
    },
    {
      "epoch": 2.674933569530558,
      "grad_norm": 0.44138211011886597,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 3020
    },
    {
      "epoch": 2.679362267493357,
      "grad_norm": 0.3149421215057373,
      "learning_rate": 1e-05,
      "loss": 1.1063,
      "step": 3025
    },
    {
      "epoch": 2.683790965456156,
      "grad_norm": 0.4658683240413666,
      "learning_rate": 1e-05,
      "loss": 1.0262,
      "step": 3030
    },
    {
      "epoch": 2.6882196634189546,
      "grad_norm": 0.373843252658844,
      "learning_rate": 1e-05,
      "loss": 1.0337,
      "step": 3035
    },
    {
      "epoch": 2.692648361381754,
      "grad_norm": 0.3963046371936798,
      "learning_rate": 1e-05,
      "loss": 1.025,
      "step": 3040
    },
    {
      "epoch": 2.6970770593445526,
      "grad_norm": 0.46280255913734436,
      "learning_rate": 1e-05,
      "loss": 1.0499,
      "step": 3045
    },
    {
      "epoch": 2.7015057573073515,
      "grad_norm": 0.412231981754303,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 3050
    },
    {
      "epoch": 2.7059344552701505,
      "grad_norm": 0.3762901723384857,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 3055
    },
    {
      "epoch": 2.7103631532329495,
      "grad_norm": 0.4995385408401489,
      "learning_rate": 1e-05,
      "loss": 1.0288,
      "step": 3060
    },
    {
      "epoch": 2.7147918511957485,
      "grad_norm": 0.38599368929862976,
      "learning_rate": 1e-05,
      "loss": 1.018,
      "step": 3065
    },
    {
      "epoch": 2.7192205491585475,
      "grad_norm": 0.31549742817878723,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 3070
    },
    {
      "epoch": 2.7236492471213465,
      "grad_norm": 0.3889061510562897,
      "learning_rate": 1e-05,
      "loss": 1.0463,
      "step": 3075
    },
    {
      "epoch": 2.728077945084145,
      "grad_norm": 0.6376797556877136,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 3080
    },
    {
      "epoch": 2.7325066430469445,
      "grad_norm": 0.549766480922699,
      "learning_rate": 1e-05,
      "loss": 1.0545,
      "step": 3085
    },
    {
      "epoch": 2.736935341009743,
      "grad_norm": 0.3776531517505646,
      "learning_rate": 1e-05,
      "loss": 1.0798,
      "step": 3090
    },
    {
      "epoch": 2.741364038972542,
      "grad_norm": 0.40356865525245667,
      "learning_rate": 1e-05,
      "loss": 1.034,
      "step": 3095
    },
    {
      "epoch": 2.745792736935341,
      "grad_norm": 0.41116103529930115,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 3100
    },
    {
      "epoch": 2.75022143489814,
      "grad_norm": 0.41147708892822266,
      "learning_rate": 1e-05,
      "loss": 1.0362,
      "step": 3105
    },
    {
      "epoch": 2.754650132860939,
      "grad_norm": 0.43723002076148987,
      "learning_rate": 1e-05,
      "loss": 1.0685,
      "step": 3110
    },
    {
      "epoch": 2.759078830823738,
      "grad_norm": 0.5626050233840942,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 3115
    },
    {
      "epoch": 2.763507528786537,
      "grad_norm": 0.43236517906188965,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 3120
    },
    {
      "epoch": 2.7679362267493355,
      "grad_norm": 0.3476856052875519,
      "learning_rate": 1e-05,
      "loss": 1.0311,
      "step": 3125
    },
    {
      "epoch": 2.7723649247121345,
      "grad_norm": 0.47323328256607056,
      "learning_rate": 1e-05,
      "loss": 1.0661,
      "step": 3130
    },
    {
      "epoch": 2.7767936226749335,
      "grad_norm": 0.6006180047988892,
      "learning_rate": 1e-05,
      "loss": 1.0616,
      "step": 3135
    },
    {
      "epoch": 2.7812223206377324,
      "grad_norm": 0.6216714382171631,
      "learning_rate": 1e-05,
      "loss": 0.9522,
      "step": 3140
    },
    {
      "epoch": 2.7856510186005314,
      "grad_norm": 0.31843629479408264,
      "learning_rate": 1e-05,
      "loss": 1.0526,
      "step": 3145
    },
    {
      "epoch": 2.7900797165633304,
      "grad_norm": 0.32713088393211365,
      "learning_rate": 1e-05,
      "loss": 1.0475,
      "step": 3150
    },
    {
      "epoch": 2.7945084145261294,
      "grad_norm": 0.49055880308151245,
      "learning_rate": 1e-05,
      "loss": 1.069,
      "step": 3155
    },
    {
      "epoch": 2.7989371124889284,
      "grad_norm": 0.33268678188323975,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 3160
    },
    {
      "epoch": 2.8033658104517274,
      "grad_norm": 0.34377628564834595,
      "learning_rate": 1e-05,
      "loss": 1.0493,
      "step": 3165
    },
    {
      "epoch": 2.807794508414526,
      "grad_norm": 0.43863555788993835,
      "learning_rate": 1e-05,
      "loss": 1.0266,
      "step": 3170
    },
    {
      "epoch": 2.812223206377325,
      "grad_norm": 0.4929737448692322,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 3175
    },
    {
      "epoch": 2.816651904340124,
      "grad_norm": 0.3163629174232483,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 3180
    },
    {
      "epoch": 2.821080602302923,
      "grad_norm": 0.4430500864982605,
      "learning_rate": 1e-05,
      "loss": 1.0753,
      "step": 3185
    },
    {
      "epoch": 2.825509300265722,
      "grad_norm": 0.41517603397369385,
      "learning_rate": 1e-05,
      "loss": 1.0346,
      "step": 3190
    },
    {
      "epoch": 2.829937998228521,
      "grad_norm": 0.39873167872428894,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 3195
    },
    {
      "epoch": 2.83436669619132,
      "grad_norm": 0.35759586095809937,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 3200
    },
    {
      "epoch": 2.8387953941541184,
      "grad_norm": 0.48786240816116333,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 3205
    },
    {
      "epoch": 2.843224092116918,
      "grad_norm": 0.4037776291370392,
      "learning_rate": 1e-05,
      "loss": 1.067,
      "step": 3210
    },
    {
      "epoch": 2.8476527900797164,
      "grad_norm": 0.4887538254261017,
      "learning_rate": 1e-05,
      "loss": 1.083,
      "step": 3215
    },
    {
      "epoch": 2.8520814880425154,
      "grad_norm": 0.39434826374053955,
      "learning_rate": 1e-05,
      "loss": 1.0528,
      "step": 3220
    },
    {
      "epoch": 2.8565101860053144,
      "grad_norm": 0.5967410206794739,
      "learning_rate": 1e-05,
      "loss": 1.0284,
      "step": 3225
    },
    {
      "epoch": 2.8609388839681134,
      "grad_norm": 0.4126602113246918,
      "learning_rate": 1e-05,
      "loss": 1.0614,
      "step": 3230
    },
    {
      "epoch": 2.8653675819309123,
      "grad_norm": 0.48035362362861633,
      "learning_rate": 1e-05,
      "loss": 1.0772,
      "step": 3235
    },
    {
      "epoch": 2.8697962798937113,
      "grad_norm": 0.5496764183044434,
      "learning_rate": 1e-05,
      "loss": 1.0454,
      "step": 3240
    },
    {
      "epoch": 2.8742249778565103,
      "grad_norm": 0.3879026770591736,
      "learning_rate": 1e-05,
      "loss": 1.0418,
      "step": 3245
    },
    {
      "epoch": 2.878653675819309,
      "grad_norm": 0.37894827127456665,
      "learning_rate": 1e-05,
      "loss": 1.0413,
      "step": 3250
    },
    {
      "epoch": 2.8830823737821083,
      "grad_norm": 0.3943926990032196,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 3255
    },
    {
      "epoch": 2.887511071744907,
      "grad_norm": 0.5473713874816895,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 3260
    },
    {
      "epoch": 2.891939769707706,
      "grad_norm": 0.3977048397064209,
      "learning_rate": 1e-05,
      "loss": 1.0898,
      "step": 3265
    },
    {
      "epoch": 2.896368467670505,
      "grad_norm": 0.5678221583366394,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 3270
    },
    {
      "epoch": 2.900797165633304,
      "grad_norm": 0.4889892041683197,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 3275
    },
    {
      "epoch": 2.905225863596103,
      "grad_norm": 0.5965889096260071,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 3280
    },
    {
      "epoch": 2.909654561558902,
      "grad_norm": 0.5212200880050659,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 3285
    },
    {
      "epoch": 2.9140832595217008,
      "grad_norm": 0.5184657573699951,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 3290
    },
    {
      "epoch": 2.9185119574844993,
      "grad_norm": 0.4270380139350891,
      "learning_rate": 1e-05,
      "loss": 1.0375,
      "step": 3295
    },
    {
      "epoch": 2.9229406554472988,
      "grad_norm": 0.4573812782764435,
      "learning_rate": 1e-05,
      "loss": 1.0676,
      "step": 3300
    },
    {
      "epoch": 2.9273693534100973,
      "grad_norm": 0.5198473930358887,
      "learning_rate": 1e-05,
      "loss": 1.0479,
      "step": 3305
    },
    {
      "epoch": 2.9317980513728963,
      "grad_norm": 0.41153794527053833,
      "learning_rate": 1e-05,
      "loss": 1.0194,
      "step": 3310
    },
    {
      "epoch": 2.9362267493356953,
      "grad_norm": 0.4575025737285614,
      "learning_rate": 1e-05,
      "loss": 1.0378,
      "step": 3315
    },
    {
      "epoch": 2.9406554472984943,
      "grad_norm": 0.6119422316551208,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 3320
    },
    {
      "epoch": 2.9450841452612933,
      "grad_norm": 0.31815919280052185,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 3325
    },
    {
      "epoch": 2.9495128432240922,
      "grad_norm": 0.6472691297531128,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 3330
    },
    {
      "epoch": 2.9539415411868912,
      "grad_norm": 0.39716836810112,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 3335
    },
    {
      "epoch": 2.9583702391496898,
      "grad_norm": 0.4567423462867737,
      "learning_rate": 1e-05,
      "loss": 1.0525,
      "step": 3340
    },
    {
      "epoch": 2.9627989371124888,
      "grad_norm": 0.4389401376247406,
      "learning_rate": 1e-05,
      "loss": 1.0581,
      "step": 3345
    },
    {
      "epoch": 2.9672276350752878,
      "grad_norm": 0.42704617977142334,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 3350
    },
    {
      "epoch": 2.9716563330380867,
      "grad_norm": 0.5419511795043945,
      "learning_rate": 1e-05,
      "loss": 1.0305,
      "step": 3355
    },
    {
      "epoch": 2.9760850310008857,
      "grad_norm": 0.33577871322631836,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 3360
    },
    {
      "epoch": 2.9805137289636847,
      "grad_norm": 0.45892855525016785,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 3365
    },
    {
      "epoch": 2.9849424269264837,
      "grad_norm": 0.391928106546402,
      "learning_rate": 1e-05,
      "loss": 1.0537,
      "step": 3370
    },
    {
      "epoch": 2.9893711248892827,
      "grad_norm": 0.5902290940284729,
      "learning_rate": 1e-05,
      "loss": 1.0863,
      "step": 3375
    },
    {
      "epoch": 2.9937998228520817,
      "grad_norm": 0.4113743305206299,
      "learning_rate": 1e-05,
      "loss": 1.0401,
      "step": 3380
    },
    {
      "epoch": 2.9982285208148802,
      "grad_norm": 0.3413982391357422,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 3385
    },
    {
      "epoch": 3.002657218777679,
      "grad_norm": 0.41479942202568054,
      "learning_rate": 1e-05,
      "loss": 1.046,
      "step": 3390
    },
    {
      "epoch": 3.007085916740478,
      "grad_norm": 0.5093465447425842,
      "learning_rate": 1e-05,
      "loss": 1.0409,
      "step": 3395
    },
    {
      "epoch": 3.011514614703277,
      "grad_norm": 0.42168518900871277,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 3400
    },
    {
      "epoch": 3.015943312666076,
      "grad_norm": 0.38336870074272156,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 3405
    },
    {
      "epoch": 3.020372010628875,
      "grad_norm": 0.3866322636604309,
      "learning_rate": 1e-05,
      "loss": 1.0428,
      "step": 3410
    },
    {
      "epoch": 3.024800708591674,
      "grad_norm": 0.5771451592445374,
      "learning_rate": 1e-05,
      "loss": 1.053,
      "step": 3415
    },
    {
      "epoch": 3.029229406554473,
      "grad_norm": 0.4466056525707245,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 3420
    },
    {
      "epoch": 3.033658104517272,
      "grad_norm": 0.44519755244255066,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 3425
    },
    {
      "epoch": 3.0380868024800707,
      "grad_norm": 0.3985057771205902,
      "learning_rate": 1e-05,
      "loss": 1.0415,
      "step": 3430
    },
    {
      "epoch": 3.0425155004428697,
      "grad_norm": 0.3949190378189087,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 3435
    },
    {
      "epoch": 3.0469441984056687,
      "grad_norm": 0.39209190011024475,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 3440
    },
    {
      "epoch": 3.0513728963684676,
      "grad_norm": 0.3770267367362976,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 3445
    },
    {
      "epoch": 3.0558015943312666,
      "grad_norm": 0.47111639380455017,
      "learning_rate": 1e-05,
      "loss": 1.0541,
      "step": 3450
    },
    {
      "epoch": 3.0602302922940656,
      "grad_norm": 0.4482271373271942,
      "learning_rate": 1e-05,
      "loss": 1.0278,
      "step": 3455
    },
    {
      "epoch": 3.0646589902568646,
      "grad_norm": 0.377704918384552,
      "learning_rate": 1e-05,
      "loss": 1.0235,
      "step": 3460
    },
    {
      "epoch": 3.0690876882196636,
      "grad_norm": 0.4829293191432953,
      "learning_rate": 1e-05,
      "loss": 1.0646,
      "step": 3465
    },
    {
      "epoch": 3.073516386182462,
      "grad_norm": 0.30433109402656555,
      "learning_rate": 1e-05,
      "loss": 1.037,
      "step": 3470
    },
    {
      "epoch": 3.077945084145261,
      "grad_norm": 0.4092828631401062,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 3475
    },
    {
      "epoch": 3.08237378210806,
      "grad_norm": 0.37381264567375183,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 3480
    },
    {
      "epoch": 3.086802480070859,
      "grad_norm": 0.5895156860351562,
      "learning_rate": 1e-05,
      "loss": 1.038,
      "step": 3485
    },
    {
      "epoch": 3.091231178033658,
      "grad_norm": 0.4740341305732727,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 3490
    },
    {
      "epoch": 3.095659875996457,
      "grad_norm": 0.4457544982433319,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 3495
    },
    {
      "epoch": 3.100088573959256,
      "grad_norm": 0.42571043968200684,
      "learning_rate": 1e-05,
      "loss": 1.1006,
      "step": 3500
    },
    {
      "epoch": 3.104517271922055,
      "grad_norm": 0.4492760896682739,
      "learning_rate": 1e-05,
      "loss": 1.0413,
      "step": 3505
    },
    {
      "epoch": 3.108945969884854,
      "grad_norm": 0.4131598174571991,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 3510
    },
    {
      "epoch": 3.1133746678476526,
      "grad_norm": 0.44474515318870544,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 3515
    },
    {
      "epoch": 3.1178033658104516,
      "grad_norm": 0.4319981634616852,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 3520
    },
    {
      "epoch": 3.1222320637732506,
      "grad_norm": 0.3335142433643341,
      "learning_rate": 1e-05,
      "loss": 1.0304,
      "step": 3525
    },
    {
      "epoch": 3.1266607617360496,
      "grad_norm": 0.3271971642971039,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 3530
    },
    {
      "epoch": 3.1310894596988486,
      "grad_norm": 0.30540257692337036,
      "learning_rate": 1e-05,
      "loss": 1.0638,
      "step": 3535
    },
    {
      "epoch": 3.1355181576616475,
      "grad_norm": 0.4695254862308502,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 3540
    },
    {
      "epoch": 3.1399468556244465,
      "grad_norm": 0.42064210772514343,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 3545
    },
    {
      "epoch": 3.1443755535872455,
      "grad_norm": 0.5295586585998535,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 3550
    },
    {
      "epoch": 3.148804251550044,
      "grad_norm": 0.4814288318157196,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 3555
    },
    {
      "epoch": 3.153232949512843,
      "grad_norm": 0.4444942772388458,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 3560
    },
    {
      "epoch": 3.157661647475642,
      "grad_norm": 0.3699358403682709,
      "learning_rate": 1e-05,
      "loss": 1.0632,
      "step": 3565
    },
    {
      "epoch": 3.162090345438441,
      "grad_norm": 0.5086784958839417,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 3570
    },
    {
      "epoch": 3.16651904340124,
      "grad_norm": 0.3851536810398102,
      "learning_rate": 1e-05,
      "loss": 1.0256,
      "step": 3575
    },
    {
      "epoch": 3.170947741364039,
      "grad_norm": 0.37175634503364563,
      "learning_rate": 1e-05,
      "loss": 1.0707,
      "step": 3580
    },
    {
      "epoch": 3.175376439326838,
      "grad_norm": 0.4465455710887909,
      "learning_rate": 1e-05,
      "loss": 1.0456,
      "step": 3585
    },
    {
      "epoch": 3.179805137289637,
      "grad_norm": 0.37006422877311707,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 3590
    },
    {
      "epoch": 3.184233835252436,
      "grad_norm": 0.4222365617752075,
      "learning_rate": 1e-05,
      "loss": 1.0334,
      "step": 3595
    },
    {
      "epoch": 3.1886625332152345,
      "grad_norm": 0.3667023181915283,
      "learning_rate": 1e-05,
      "loss": 1.0511,
      "step": 3600
    },
    {
      "epoch": 3.1930912311780335,
      "grad_norm": 0.4231782555580139,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 3605
    },
    {
      "epoch": 3.1975199291408325,
      "grad_norm": 0.3751493990421295,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 3610
    },
    {
      "epoch": 3.2019486271036315,
      "grad_norm": 0.47144243121147156,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 3615
    },
    {
      "epoch": 3.2063773250664305,
      "grad_norm": 0.4281963109970093,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 3620
    },
    {
      "epoch": 3.2108060230292295,
      "grad_norm": 0.6021721363067627,
      "learning_rate": 1e-05,
      "loss": 1.0396,
      "step": 3625
    },
    {
      "epoch": 3.2152347209920284,
      "grad_norm": 0.39457863569259644,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 3630
    },
    {
      "epoch": 3.2196634189548274,
      "grad_norm": 0.3613373935222626,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 3635
    },
    {
      "epoch": 3.2240921169176264,
      "grad_norm": 0.4260545074939728,
      "learning_rate": 1e-05,
      "loss": 1.0288,
      "step": 3640
    },
    {
      "epoch": 3.228520814880425,
      "grad_norm": 0.5399686694145203,
      "learning_rate": 1e-05,
      "loss": 1.0817,
      "step": 3645
    },
    {
      "epoch": 3.232949512843224,
      "grad_norm": 0.42582985758781433,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 3650
    },
    {
      "epoch": 3.237378210806023,
      "grad_norm": 0.313127338886261,
      "learning_rate": 1e-05,
      "loss": 1.0512,
      "step": 3655
    },
    {
      "epoch": 3.241806908768822,
      "grad_norm": 0.4694111943244934,
      "learning_rate": 1e-05,
      "loss": 1.0592,
      "step": 3660
    },
    {
      "epoch": 3.246235606731621,
      "grad_norm": 0.6192038655281067,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 3665
    },
    {
      "epoch": 3.25066430469442,
      "grad_norm": 0.47336092591285706,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 3670
    },
    {
      "epoch": 3.255093002657219,
      "grad_norm": 0.34903597831726074,
      "learning_rate": 1e-05,
      "loss": 1.0314,
      "step": 3675
    },
    {
      "epoch": 3.259521700620018,
      "grad_norm": 0.4449535310268402,
      "learning_rate": 1e-05,
      "loss": 1.0467,
      "step": 3680
    },
    {
      "epoch": 3.263950398582817,
      "grad_norm": 0.5163513422012329,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 3685
    },
    {
      "epoch": 3.2683790965456154,
      "grad_norm": 0.36432549357414246,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 3690
    },
    {
      "epoch": 3.2728077945084144,
      "grad_norm": 0.4025719165802002,
      "learning_rate": 1e-05,
      "loss": 1.0195,
      "step": 3695
    },
    {
      "epoch": 3.2772364924712134,
      "grad_norm": 0.6147280335426331,
      "learning_rate": 1e-05,
      "loss": 1.014,
      "step": 3700
    },
    {
      "epoch": 3.2816651904340124,
      "grad_norm": 0.4857144355773926,
      "learning_rate": 1e-05,
      "loss": 1.0309,
      "step": 3705
    },
    {
      "epoch": 3.2860938883968114,
      "grad_norm": 0.5235254168510437,
      "learning_rate": 1e-05,
      "loss": 1.0957,
      "step": 3710
    },
    {
      "epoch": 3.2905225863596104,
      "grad_norm": 0.5222679376602173,
      "learning_rate": 1e-05,
      "loss": 1.0711,
      "step": 3715
    },
    {
      "epoch": 3.2949512843224094,
      "grad_norm": 0.2747630476951599,
      "learning_rate": 1e-05,
      "loss": 1.0445,
      "step": 3720
    },
    {
      "epoch": 3.299379982285208,
      "grad_norm": 0.36898380517959595,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 3725
    },
    {
      "epoch": 3.3038086802480073,
      "grad_norm": 0.4518287777900696,
      "learning_rate": 1e-05,
      "loss": 1.0301,
      "step": 3730
    },
    {
      "epoch": 3.308237378210806,
      "grad_norm": 0.37403711676597595,
      "learning_rate": 1e-05,
      "loss": 1.0613,
      "step": 3735
    },
    {
      "epoch": 3.312666076173605,
      "grad_norm": 0.3976152241230011,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 3740
    },
    {
      "epoch": 3.317094774136404,
      "grad_norm": 0.3707811236381531,
      "learning_rate": 1e-05,
      "loss": 1.0511,
      "step": 3745
    },
    {
      "epoch": 3.321523472099203,
      "grad_norm": 0.490507036447525,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 3750
    },
    {
      "epoch": 3.325952170062002,
      "grad_norm": 0.3149888515472412,
      "learning_rate": 1e-05,
      "loss": 1.0207,
      "step": 3755
    },
    {
      "epoch": 3.330380868024801,
      "grad_norm": 0.4916192591190338,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 3760
    },
    {
      "epoch": 3.3348095659876,
      "grad_norm": 0.4619700312614441,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 3765
    },
    {
      "epoch": 3.3392382639503984,
      "grad_norm": 0.4629688858985901,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 3770
    },
    {
      "epoch": 3.3436669619131973,
      "grad_norm": 0.4954208731651306,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 3775
    },
    {
      "epoch": 3.3480956598759963,
      "grad_norm": 0.5487666726112366,
      "learning_rate": 1e-05,
      "loss": 1.0343,
      "step": 3780
    },
    {
      "epoch": 3.3525243578387953,
      "grad_norm": 0.36240583658218384,
      "learning_rate": 1e-05,
      "loss": 1.0301,
      "step": 3785
    },
    {
      "epoch": 3.3569530558015943,
      "grad_norm": 0.38365602493286133,
      "learning_rate": 1e-05,
      "loss": 1.0415,
      "step": 3790
    },
    {
      "epoch": 3.3613817537643933,
      "grad_norm": 0.32367461919784546,
      "learning_rate": 1e-05,
      "loss": 1.0428,
      "step": 3795
    },
    {
      "epoch": 3.3658104517271923,
      "grad_norm": 0.38669154047966003,
      "learning_rate": 1e-05,
      "loss": 1.0663,
      "step": 3800
    },
    {
      "epoch": 3.3702391496899913,
      "grad_norm": 0.38724860548973083,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 3805
    },
    {
      "epoch": 3.3746678476527903,
      "grad_norm": 0.37929293513298035,
      "learning_rate": 1e-05,
      "loss": 1.0438,
      "step": 3810
    },
    {
      "epoch": 3.379096545615589,
      "grad_norm": 0.40706324577331543,
      "learning_rate": 1e-05,
      "loss": 1.0703,
      "step": 3815
    },
    {
      "epoch": 3.383525243578388,
      "grad_norm": 0.47311824560165405,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 3820
    },
    {
      "epoch": 3.387953941541187,
      "grad_norm": 0.3395290672779083,
      "learning_rate": 1e-05,
      "loss": 1.0565,
      "step": 3825
    },
    {
      "epoch": 3.3923826395039858,
      "grad_norm": 0.4534793198108673,
      "learning_rate": 1e-05,
      "loss": 1.0226,
      "step": 3830
    },
    {
      "epoch": 3.3968113374667848,
      "grad_norm": 0.33159688115119934,
      "learning_rate": 1e-05,
      "loss": 1.0085,
      "step": 3835
    },
    {
      "epoch": 3.4012400354295838,
      "grad_norm": 0.4188603460788727,
      "learning_rate": 1e-05,
      "loss": 1.0387,
      "step": 3840
    },
    {
      "epoch": 3.4056687333923827,
      "grad_norm": 0.5187346935272217,
      "learning_rate": 1e-05,
      "loss": 1.0389,
      "step": 3845
    },
    {
      "epoch": 3.4100974313551817,
      "grad_norm": 0.4985080063343048,
      "learning_rate": 1e-05,
      "loss": 1.0507,
      "step": 3850
    },
    {
      "epoch": 3.4145261293179807,
      "grad_norm": 0.38136884570121765,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 3855
    },
    {
      "epoch": 3.4189548272807793,
      "grad_norm": 0.31632256507873535,
      "learning_rate": 1e-05,
      "loss": 1.0318,
      "step": 3860
    },
    {
      "epoch": 3.4233835252435783,
      "grad_norm": 0.386005163192749,
      "learning_rate": 1e-05,
      "loss": 1.0435,
      "step": 3865
    },
    {
      "epoch": 3.4278122232063772,
      "grad_norm": 0.36566078662872314,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 3870
    },
    {
      "epoch": 3.4322409211691762,
      "grad_norm": 0.389634370803833,
      "learning_rate": 1e-05,
      "loss": 1.0668,
      "step": 3875
    },
    {
      "epoch": 3.436669619131975,
      "grad_norm": 0.31114742159843445,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 3880
    },
    {
      "epoch": 3.441098317094774,
      "grad_norm": 0.4838503301143646,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 3885
    },
    {
      "epoch": 3.445527015057573,
      "grad_norm": 0.3525892198085785,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 3890
    },
    {
      "epoch": 3.449955713020372,
      "grad_norm": 0.4195665717124939,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 3895
    },
    {
      "epoch": 3.454384410983171,
      "grad_norm": 0.5166270732879639,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 3900
    },
    {
      "epoch": 3.4588131089459697,
      "grad_norm": 0.3707009255886078,
      "learning_rate": 1e-05,
      "loss": 1.0629,
      "step": 3905
    },
    {
      "epoch": 3.4632418069087687,
      "grad_norm": 0.5166239738464355,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 3910
    },
    {
      "epoch": 3.4676705048715677,
      "grad_norm": 0.38136765360832214,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 3915
    },
    {
      "epoch": 3.4720992028343667,
      "grad_norm": 0.38072213530540466,
      "learning_rate": 1e-05,
      "loss": 1.0893,
      "step": 3920
    },
    {
      "epoch": 3.4765279007971657,
      "grad_norm": 0.3407192528247833,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 3925
    },
    {
      "epoch": 3.4809565987599647,
      "grad_norm": 0.4799145758152008,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 3930
    },
    {
      "epoch": 3.4853852967227636,
      "grad_norm": 0.520026445388794,
      "learning_rate": 1e-05,
      "loss": 1.0846,
      "step": 3935
    },
    {
      "epoch": 3.4898139946855626,
      "grad_norm": 0.5156550407409668,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 3940
    },
    {
      "epoch": 3.4942426926483616,
      "grad_norm": 0.5291915535926819,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 3945
    },
    {
      "epoch": 3.49867139061116,
      "grad_norm": 0.3861340880393982,
      "learning_rate": 1e-05,
      "loss": 1.0674,
      "step": 3950
    },
    {
      "epoch": 3.503100088573959,
      "grad_norm": 0.44092243909835815,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 3955
    },
    {
      "epoch": 3.507528786536758,
      "grad_norm": 0.4885048270225525,
      "learning_rate": 1e-05,
      "loss": 1.03,
      "step": 3960
    },
    {
      "epoch": 3.511957484499557,
      "grad_norm": 0.4576728940010071,
      "learning_rate": 1e-05,
      "loss": 1.0867,
      "step": 3965
    },
    {
      "epoch": 3.516386182462356,
      "grad_norm": 0.380595326423645,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 3970
    },
    {
      "epoch": 3.520814880425155,
      "grad_norm": 0.38781315088272095,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 3975
    },
    {
      "epoch": 3.525243578387954,
      "grad_norm": 0.33252057433128357,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 3980
    },
    {
      "epoch": 3.5296722763507526,
      "grad_norm": 0.3965751528739929,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 3985
    },
    {
      "epoch": 3.534100974313552,
      "grad_norm": 0.324137806892395,
      "learning_rate": 1e-05,
      "loss": 1.0325,
      "step": 3990
    },
    {
      "epoch": 3.5385296722763506,
      "grad_norm": 0.35043829679489136,
      "learning_rate": 1e-05,
      "loss": 1.0312,
      "step": 3995
    },
    {
      "epoch": 3.5429583702391496,
      "grad_norm": 0.2821948230266571,
      "learning_rate": 1e-05,
      "loss": 1.0524,
      "step": 4000
    },
    {
      "epoch": 3.5473870682019486,
      "grad_norm": 0.3412671685218811,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 4005
    },
    {
      "epoch": 3.5518157661647476,
      "grad_norm": 0.47750186920166016,
      "learning_rate": 1e-05,
      "loss": 1.0512,
      "step": 4010
    },
    {
      "epoch": 3.5562444641275466,
      "grad_norm": 0.4440419375896454,
      "learning_rate": 1e-05,
      "loss": 1.043,
      "step": 4015
    },
    {
      "epoch": 3.5606731620903456,
      "grad_norm": 0.3451400101184845,
      "learning_rate": 1e-05,
      "loss": 1.0281,
      "step": 4020
    },
    {
      "epoch": 3.5651018600531446,
      "grad_norm": 0.4672628939151764,
      "learning_rate": 1e-05,
      "loss": 1.0341,
      "step": 4025
    },
    {
      "epoch": 3.569530558015943,
      "grad_norm": 0.4284445643424988,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 4030
    },
    {
      "epoch": 3.573959255978742,
      "grad_norm": 0.45521488785743713,
      "learning_rate": 1e-05,
      "loss": 1.0337,
      "step": 4035
    },
    {
      "epoch": 3.578387953941541,
      "grad_norm": 0.35136884450912476,
      "learning_rate": 1e-05,
      "loss": 1.0349,
      "step": 4040
    },
    {
      "epoch": 3.58281665190434,
      "grad_norm": 0.37805497646331787,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 4045
    },
    {
      "epoch": 3.587245349867139,
      "grad_norm": 0.5786856412887573,
      "learning_rate": 1e-05,
      "loss": 1.0367,
      "step": 4050
    },
    {
      "epoch": 3.591674047829938,
      "grad_norm": 0.38704222440719604,
      "learning_rate": 1e-05,
      "loss": 1.0292,
      "step": 4055
    },
    {
      "epoch": 3.596102745792737,
      "grad_norm": 0.41924527287483215,
      "learning_rate": 1e-05,
      "loss": 1.0571,
      "step": 4060
    },
    {
      "epoch": 3.600531443755536,
      "grad_norm": 0.4547056257724762,
      "learning_rate": 1e-05,
      "loss": 1.0481,
      "step": 4065
    },
    {
      "epoch": 3.604960141718335,
      "grad_norm": 0.3528039753437042,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 4070
    },
    {
      "epoch": 3.6093888396811336,
      "grad_norm": 0.3567149043083191,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 4075
    },
    {
      "epoch": 3.6138175376439325,
      "grad_norm": 0.3547581434249878,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 4080
    },
    {
      "epoch": 3.6182462356067315,
      "grad_norm": 0.40108251571655273,
      "learning_rate": 1e-05,
      "loss": 1.0377,
      "step": 4085
    },
    {
      "epoch": 3.6226749335695305,
      "grad_norm": 0.3745425045490265,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 4090
    },
    {
      "epoch": 3.6271036315323295,
      "grad_norm": 0.38116466999053955,
      "learning_rate": 1e-05,
      "loss": 1.0318,
      "step": 4095
    },
    {
      "epoch": 3.6315323294951285,
      "grad_norm": 0.34331002831459045,
      "learning_rate": 1e-05,
      "loss": 1.0311,
      "step": 4100
    },
    {
      "epoch": 3.6359610274579275,
      "grad_norm": 0.3361397087574005,
      "learning_rate": 1e-05,
      "loss": 1.1007,
      "step": 4105
    },
    {
      "epoch": 3.640389725420726,
      "grad_norm": 0.3507891297340393,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 4110
    },
    {
      "epoch": 3.6448184233835255,
      "grad_norm": 0.32863402366638184,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 4115
    },
    {
      "epoch": 3.649247121346324,
      "grad_norm": 0.34813687205314636,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 4120
    },
    {
      "epoch": 3.653675819309123,
      "grad_norm": 0.28008124232292175,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 4125
    },
    {
      "epoch": 3.658104517271922,
      "grad_norm": 0.33971190452575684,
      "learning_rate": 1e-05,
      "loss": 1.0368,
      "step": 4130
    },
    {
      "epoch": 3.662533215234721,
      "grad_norm": 0.32608115673065186,
      "learning_rate": 1e-05,
      "loss": 1.0921,
      "step": 4135
    },
    {
      "epoch": 3.66696191319752,
      "grad_norm": 0.34819838404655457,
      "learning_rate": 1e-05,
      "loss": 1.0333,
      "step": 4140
    },
    {
      "epoch": 3.671390611160319,
      "grad_norm": 0.46355897188186646,
      "learning_rate": 1e-05,
      "loss": 1.0235,
      "step": 4145
    },
    {
      "epoch": 3.675819309123118,
      "grad_norm": 0.5182165503501892,
      "learning_rate": 1e-05,
      "loss": 1.0276,
      "step": 4150
    },
    {
      "epoch": 3.6802480070859165,
      "grad_norm": 0.34073352813720703,
      "learning_rate": 1e-05,
      "loss": 1.0493,
      "step": 4155
    },
    {
      "epoch": 3.684676705048716,
      "grad_norm": 0.460151344537735,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 4160
    },
    {
      "epoch": 3.6891054030115145,
      "grad_norm": 0.45734596252441406,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 4165
    },
    {
      "epoch": 3.6935341009743134,
      "grad_norm": 0.3882424831390381,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 4170
    },
    {
      "epoch": 3.6979627989371124,
      "grad_norm": 0.42943891882896423,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 4175
    },
    {
      "epoch": 3.7023914968999114,
      "grad_norm": 0.390083372592926,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 4180
    },
    {
      "epoch": 3.7068201948627104,
      "grad_norm": 0.42090412974357605,
      "learning_rate": 1e-05,
      "loss": 1.037,
      "step": 4185
    },
    {
      "epoch": 3.7112488928255094,
      "grad_norm": 0.4169060289859772,
      "learning_rate": 1e-05,
      "loss": 1.0417,
      "step": 4190
    },
    {
      "epoch": 3.7156775907883084,
      "grad_norm": 0.32734933495521545,
      "learning_rate": 1e-05,
      "loss": 1.0452,
      "step": 4195
    },
    {
      "epoch": 3.720106288751107,
      "grad_norm": 0.29035845398902893,
      "learning_rate": 1e-05,
      "loss": 1.0485,
      "step": 4200
    },
    {
      "epoch": 3.7245349867139064,
      "grad_norm": 0.3693220317363739,
      "learning_rate": 1e-05,
      "loss": 1.0334,
      "step": 4205
    },
    {
      "epoch": 3.728963684676705,
      "grad_norm": 0.39305785298347473,
      "learning_rate": 1e-05,
      "loss": 1.0245,
      "step": 4210
    },
    {
      "epoch": 3.733392382639504,
      "grad_norm": 0.335788369178772,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 4215
    },
    {
      "epoch": 3.737821080602303,
      "grad_norm": 0.41085320711135864,
      "learning_rate": 1e-05,
      "loss": 1.0559,
      "step": 4220
    },
    {
      "epoch": 3.742249778565102,
      "grad_norm": 0.3124822974205017,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 4225
    },
    {
      "epoch": 3.746678476527901,
      "grad_norm": 0.5323129296302795,
      "learning_rate": 1e-05,
      "loss": 1.0152,
      "step": 4230
    },
    {
      "epoch": 3.7511071744907,
      "grad_norm": 0.40696221590042114,
      "learning_rate": 1e-05,
      "loss": 1.0174,
      "step": 4235
    },
    {
      "epoch": 3.755535872453499,
      "grad_norm": 0.4836999475955963,
      "learning_rate": 1e-05,
      "loss": 1.04,
      "step": 4240
    },
    {
      "epoch": 3.7599645704162974,
      "grad_norm": 0.33299151062965393,
      "learning_rate": 1e-05,
      "loss": 1.0555,
      "step": 4245
    },
    {
      "epoch": 3.7643932683790964,
      "grad_norm": 0.2661699950695038,
      "learning_rate": 1e-05,
      "loss": 1.0672,
      "step": 4250
    },
    {
      "epoch": 3.7688219663418954,
      "grad_norm": 0.43583810329437256,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 4255
    },
    {
      "epoch": 3.7732506643046944,
      "grad_norm": 0.5268107056617737,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 4260
    },
    {
      "epoch": 3.7776793622674933,
      "grad_norm": 0.4620998799800873,
      "learning_rate": 1e-05,
      "loss": 1.0392,
      "step": 4265
    },
    {
      "epoch": 3.7821080602302923,
      "grad_norm": 0.5220122337341309,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 4270
    },
    {
      "epoch": 3.7865367581930913,
      "grad_norm": 0.3385135233402252,
      "learning_rate": 1e-05,
      "loss": 1.0208,
      "step": 4275
    },
    {
      "epoch": 3.7909654561558903,
      "grad_norm": 0.39374107122421265,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 4280
    },
    {
      "epoch": 3.7953941541186893,
      "grad_norm": 0.29298868775367737,
      "learning_rate": 1e-05,
      "loss": 1.0903,
      "step": 4285
    },
    {
      "epoch": 3.799822852081488,
      "grad_norm": 0.3600406348705292,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 4290
    },
    {
      "epoch": 3.804251550044287,
      "grad_norm": 0.374218225479126,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 4295
    },
    {
      "epoch": 3.808680248007086,
      "grad_norm": 0.4733864665031433,
      "learning_rate": 1e-05,
      "loss": 1.0359,
      "step": 4300
    },
    {
      "epoch": 3.813108945969885,
      "grad_norm": 0.36350077390670776,
      "learning_rate": 1e-05,
      "loss": 1.0391,
      "step": 4305
    },
    {
      "epoch": 3.817537643932684,
      "grad_norm": 0.5348228812217712,
      "learning_rate": 1e-05,
      "loss": 1.0807,
      "step": 4310
    },
    {
      "epoch": 3.821966341895483,
      "grad_norm": 0.40654081106185913,
      "learning_rate": 1e-05,
      "loss": 1.0395,
      "step": 4315
    },
    {
      "epoch": 3.8263950398582818,
      "grad_norm": 0.4531782567501068,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 4320
    },
    {
      "epoch": 3.8308237378210808,
      "grad_norm": 0.3837543725967407,
      "learning_rate": 1e-05,
      "loss": 1.0258,
      "step": 4325
    },
    {
      "epoch": 3.8352524357838798,
      "grad_norm": 0.43151649832725525,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 4330
    },
    {
      "epoch": 3.8396811337466783,
      "grad_norm": 0.2781372368335724,
      "learning_rate": 1e-05,
      "loss": 1.0303,
      "step": 4335
    },
    {
      "epoch": 3.8441098317094773,
      "grad_norm": 0.3800680637359619,
      "learning_rate": 1e-05,
      "loss": 1.0343,
      "step": 4340
    },
    {
      "epoch": 3.8485385296722763,
      "grad_norm": 0.38863804936408997,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 4345
    },
    {
      "epoch": 3.8529672276350753,
      "grad_norm": 0.42367759346961975,
      "learning_rate": 1e-05,
      "loss": 1.0513,
      "step": 4350
    },
    {
      "epoch": 3.8573959255978743,
      "grad_norm": 0.31877681612968445,
      "learning_rate": 1e-05,
      "loss": 1.0538,
      "step": 4355
    },
    {
      "epoch": 3.8618246235606732,
      "grad_norm": 0.2576321065425873,
      "learning_rate": 1e-05,
      "loss": 1.0323,
      "step": 4360
    },
    {
      "epoch": 3.8662533215234722,
      "grad_norm": 0.39924493432044983,
      "learning_rate": 1e-05,
      "loss": 1.041,
      "step": 4365
    },
    {
      "epoch": 3.8706820194862708,
      "grad_norm": 0.31105536222457886,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 4370
    },
    {
      "epoch": 3.87511071744907,
      "grad_norm": 0.37262478470802307,
      "learning_rate": 1e-05,
      "loss": 1.0558,
      "step": 4375
    },
    {
      "epoch": 3.8795394154118688,
      "grad_norm": 0.5016164183616638,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 4380
    },
    {
      "epoch": 3.8839681133746677,
      "grad_norm": 0.43220239877700806,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 4385
    },
    {
      "epoch": 3.8883968113374667,
      "grad_norm": 0.3780975639820099,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 4390
    },
    {
      "epoch": 3.8928255093002657,
      "grad_norm": 0.3637257516384125,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 4395
    },
    {
      "epoch": 3.8972542072630647,
      "grad_norm": 0.49972036480903625,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 4400
    },
    {
      "epoch": 3.9016829052258637,
      "grad_norm": 0.4010217487812042,
      "learning_rate": 1e-05,
      "loss": 1.0325,
      "step": 4405
    },
    {
      "epoch": 3.9061116031886627,
      "grad_norm": 0.44208645820617676,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 4410
    },
    {
      "epoch": 3.9105403011514612,
      "grad_norm": 0.41336748003959656,
      "learning_rate": 1e-05,
      "loss": 1.0216,
      "step": 4415
    },
    {
      "epoch": 3.9149689991142607,
      "grad_norm": 0.3971027135848999,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 4420
    },
    {
      "epoch": 3.919397697077059,
      "grad_norm": 0.406230628490448,
      "learning_rate": 1e-05,
      "loss": 1.0288,
      "step": 4425
    },
    {
      "epoch": 3.923826395039858,
      "grad_norm": 0.3609033524990082,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 4430
    },
    {
      "epoch": 3.928255093002657,
      "grad_norm": 0.4624992907047272,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 4435
    },
    {
      "epoch": 3.932683790965456,
      "grad_norm": 0.4618331491947174,
      "learning_rate": 1e-05,
      "loss": 1.0447,
      "step": 4440
    },
    {
      "epoch": 3.937112488928255,
      "grad_norm": 0.3795495629310608,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 4445
    },
    {
      "epoch": 3.941541186891054,
      "grad_norm": 0.3128076195716858,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 4450
    },
    {
      "epoch": 3.945969884853853,
      "grad_norm": 0.3996869623661041,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 4455
    },
    {
      "epoch": 3.9503985828166517,
      "grad_norm": 0.39952194690704346,
      "learning_rate": 1e-05,
      "loss": 1.035,
      "step": 4460
    },
    {
      "epoch": 3.954827280779451,
      "grad_norm": 0.3561505973339081,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 4465
    },
    {
      "epoch": 3.9592559787422497,
      "grad_norm": 0.32471564412117004,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 4470
    },
    {
      "epoch": 3.9636846767050486,
      "grad_norm": 0.38149505853652954,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 4475
    },
    {
      "epoch": 3.9681133746678476,
      "grad_norm": 0.3936023414134979,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 4480
    },
    {
      "epoch": 3.9725420726306466,
      "grad_norm": 0.48117342591285706,
      "learning_rate": 1e-05,
      "loss": 1.0255,
      "step": 4485
    },
    {
      "epoch": 3.9769707705934456,
      "grad_norm": 0.4666324257850647,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 4490
    },
    {
      "epoch": 3.9813994685562446,
      "grad_norm": 0.39509865641593933,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 4495
    },
    {
      "epoch": 3.9858281665190436,
      "grad_norm": 0.4648655354976654,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 4500
    },
    {
      "epoch": 3.990256864481842,
      "grad_norm": 0.3578872084617615,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 4505
    },
    {
      "epoch": 3.994685562444641,
      "grad_norm": 0.30674901604652405,
      "learning_rate": 1e-05,
      "loss": 1.0559,
      "step": 4510
    },
    {
      "epoch": 3.99911426040744,
      "grad_norm": 0.3139508068561554,
      "learning_rate": 1e-05,
      "loss": 1.0363,
      "step": 4515
    },
    {
      "epoch": 4.0035429583702395,
      "grad_norm": 0.3991316854953766,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 4520
    },
    {
      "epoch": 4.007971656333038,
      "grad_norm": 0.4668462574481964,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 4525
    },
    {
      "epoch": 4.012400354295837,
      "grad_norm": 0.32502973079681396,
      "learning_rate": 1e-05,
      "loss": 1.0315,
      "step": 4530
    },
    {
      "epoch": 4.016829052258636,
      "grad_norm": 0.40989378094673157,
      "learning_rate": 1e-05,
      "loss": 1.0713,
      "step": 4535
    },
    {
      "epoch": 4.021257750221435,
      "grad_norm": 0.3096681535243988,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 4540
    },
    {
      "epoch": 4.025686448184234,
      "grad_norm": 0.32582685351371765,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 4545
    },
    {
      "epoch": 4.030115146147033,
      "grad_norm": 0.40166813135147095,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 4550
    },
    {
      "epoch": 4.034543844109832,
      "grad_norm": 0.4353068470954895,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 4555
    },
    {
      "epoch": 4.038972542072631,
      "grad_norm": 0.32297283411026,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 4560
    },
    {
      "epoch": 4.04340124003543,
      "grad_norm": 0.4393858313560486,
      "learning_rate": 1e-05,
      "loss": 1.0391,
      "step": 4565
    },
    {
      "epoch": 4.0478299379982285,
      "grad_norm": 0.319776713848114,
      "learning_rate": 1e-05,
      "loss": 1.018,
      "step": 4570
    },
    {
      "epoch": 4.052258635961027,
      "grad_norm": 0.3933486044406891,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 4575
    },
    {
      "epoch": 4.0566873339238265,
      "grad_norm": 0.3859753906726837,
      "learning_rate": 1e-05,
      "loss": 1.0596,
      "step": 4580
    },
    {
      "epoch": 4.061116031886625,
      "grad_norm": 0.30829575657844543,
      "learning_rate": 1e-05,
      "loss": 1.0623,
      "step": 4585
    },
    {
      "epoch": 4.0655447298494245,
      "grad_norm": 0.42734357714653015,
      "learning_rate": 1e-05,
      "loss": 1.0327,
      "step": 4590
    },
    {
      "epoch": 4.069973427812223,
      "grad_norm": 0.3717409372329712,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 4595
    },
    {
      "epoch": 4.0744021257750225,
      "grad_norm": 0.3354444205760956,
      "learning_rate": 1e-05,
      "loss": 1.0411,
      "step": 4600
    },
    {
      "epoch": 4.078830823737821,
      "grad_norm": 0.39263391494750977,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 4605
    },
    {
      "epoch": 4.0832595217006205,
      "grad_norm": 0.3263624608516693,
      "learning_rate": 1e-05,
      "loss": 1.039,
      "step": 4610
    },
    {
      "epoch": 4.087688219663419,
      "grad_norm": 0.499742716550827,
      "learning_rate": 1e-05,
      "loss": 1.0247,
      "step": 4615
    },
    {
      "epoch": 4.0921169176262175,
      "grad_norm": 0.3455238342285156,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 4620
    },
    {
      "epoch": 4.096545615589017,
      "grad_norm": 0.4965772032737732,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 4625
    },
    {
      "epoch": 4.1009743135518155,
      "grad_norm": 0.5790495276451111,
      "learning_rate": 1e-05,
      "loss": 1.0334,
      "step": 4630
    },
    {
      "epoch": 4.105403011514615,
      "grad_norm": 0.44276583194732666,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 4635
    },
    {
      "epoch": 4.1098317094774135,
      "grad_norm": 0.5250453948974609,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 4640
    },
    {
      "epoch": 4.114260407440213,
      "grad_norm": 0.42392274737358093,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 4645
    },
    {
      "epoch": 4.1186891054030115,
      "grad_norm": 0.39138004183769226,
      "learning_rate": 1e-05,
      "loss": 1.0554,
      "step": 4650
    },
    {
      "epoch": 4.12311780336581,
      "grad_norm": 0.4503367841243744,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 4655
    },
    {
      "epoch": 4.1275465013286095,
      "grad_norm": 0.38771429657936096,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 4660
    },
    {
      "epoch": 4.131975199291408,
      "grad_norm": 0.44300052523612976,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 4665
    },
    {
      "epoch": 4.136403897254207,
      "grad_norm": 0.4752773940563202,
      "learning_rate": 1e-05,
      "loss": 1.0295,
      "step": 4670
    },
    {
      "epoch": 4.140832595217006,
      "grad_norm": 0.36452868580818176,
      "learning_rate": 1e-05,
      "loss": 1.0217,
      "step": 4675
    },
    {
      "epoch": 4.145261293179805,
      "grad_norm": 0.4705621600151062,
      "learning_rate": 1e-05,
      "loss": 1.0451,
      "step": 4680
    },
    {
      "epoch": 4.149689991142604,
      "grad_norm": 0.5243046283721924,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 4685
    },
    {
      "epoch": 4.154118689105403,
      "grad_norm": 0.44808846712112427,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 4690
    },
    {
      "epoch": 4.158547387068202,
      "grad_norm": 0.503298819065094,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 4695
    },
    {
      "epoch": 4.1629760850310005,
      "grad_norm": 0.3674895763397217,
      "learning_rate": 1e-05,
      "loss": 1.0545,
      "step": 4700
    },
    {
      "epoch": 4.1674047829938,
      "grad_norm": 0.47434374690055847,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 4705
    },
    {
      "epoch": 4.1718334809565985,
      "grad_norm": 0.36587774753570557,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 4710
    },
    {
      "epoch": 4.176262178919398,
      "grad_norm": 0.30373701453208923,
      "learning_rate": 1e-05,
      "loss": 1.0342,
      "step": 4715
    },
    {
      "epoch": 4.180690876882196,
      "grad_norm": 0.3133320212364197,
      "learning_rate": 1e-05,
      "loss": 1.0308,
      "step": 4720
    },
    {
      "epoch": 4.185119574844996,
      "grad_norm": 0.412952184677124,
      "learning_rate": 1e-05,
      "loss": 1.0523,
      "step": 4725
    },
    {
      "epoch": 4.189548272807794,
      "grad_norm": 0.40483686327934265,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 4730
    },
    {
      "epoch": 4.193976970770594,
      "grad_norm": 0.3221893608570099,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 4735
    },
    {
      "epoch": 4.198405668733392,
      "grad_norm": 0.40332549810409546,
      "learning_rate": 1e-05,
      "loss": 1.0535,
      "step": 4740
    },
    {
      "epoch": 4.202834366696191,
      "grad_norm": 0.41692882776260376,
      "learning_rate": 1e-05,
      "loss": 1.0522,
      "step": 4745
    },
    {
      "epoch": 4.20726306465899,
      "grad_norm": 0.3916093409061432,
      "learning_rate": 1e-05,
      "loss": 1.0682,
      "step": 4750
    },
    {
      "epoch": 4.211691762621789,
      "grad_norm": 0.48341992497444153,
      "learning_rate": 1e-05,
      "loss": 1.0653,
      "step": 4755
    },
    {
      "epoch": 4.216120460584588,
      "grad_norm": 0.4701288044452667,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 4760
    },
    {
      "epoch": 4.220549158547387,
      "grad_norm": 0.3812442123889923,
      "learning_rate": 1e-05,
      "loss": 1.0362,
      "step": 4765
    },
    {
      "epoch": 4.224977856510186,
      "grad_norm": 0.4342661499977112,
      "learning_rate": 1e-05,
      "loss": 1.0458,
      "step": 4770
    },
    {
      "epoch": 4.229406554472985,
      "grad_norm": 0.516766369342804,
      "learning_rate": 1e-05,
      "loss": 1.0587,
      "step": 4775
    },
    {
      "epoch": 4.233835252435784,
      "grad_norm": 0.33460134267807007,
      "learning_rate": 1e-05,
      "loss": 1.0267,
      "step": 4780
    },
    {
      "epoch": 4.238263950398583,
      "grad_norm": 0.36863911151885986,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 4785
    },
    {
      "epoch": 4.242692648361381,
      "grad_norm": 0.4102412760257721,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 4790
    },
    {
      "epoch": 4.247121346324181,
      "grad_norm": 0.3994167447090149,
      "learning_rate": 1e-05,
      "loss": 1.0491,
      "step": 4795
    },
    {
      "epoch": 4.251550044286979,
      "grad_norm": 0.26637136936187744,
      "learning_rate": 1e-05,
      "loss": 1.0732,
      "step": 4800
    },
    {
      "epoch": 4.255978742249779,
      "grad_norm": 0.27351829409599304,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 4805
    },
    {
      "epoch": 4.260407440212577,
      "grad_norm": 0.3481573164463043,
      "learning_rate": 1e-05,
      "loss": 1.0641,
      "step": 4810
    },
    {
      "epoch": 4.264836138175377,
      "grad_norm": 0.27660074830055237,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 4815
    },
    {
      "epoch": 4.269264836138175,
      "grad_norm": 0.3420484662055969,
      "learning_rate": 1e-05,
      "loss": 1.0894,
      "step": 4820
    },
    {
      "epoch": 4.273693534100975,
      "grad_norm": 0.4297480583190918,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 4825
    },
    {
      "epoch": 4.278122232063773,
      "grad_norm": 0.3223203420639038,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 4830
    },
    {
      "epoch": 4.282550930026572,
      "grad_norm": 0.45501500368118286,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 4835
    },
    {
      "epoch": 4.286979627989371,
      "grad_norm": 0.37511059641838074,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 4840
    },
    {
      "epoch": 4.29140832595217,
      "grad_norm": 0.29621511697769165,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 4845
    },
    {
      "epoch": 4.295837023914969,
      "grad_norm": 0.36821597814559937,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 4850
    },
    {
      "epoch": 4.300265721877768,
      "grad_norm": 0.29799211025238037,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 4855
    },
    {
      "epoch": 4.304694419840567,
      "grad_norm": 0.31813478469848633,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 4860
    },
    {
      "epoch": 4.309123117803366,
      "grad_norm": 0.3424695134162903,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 4865
    },
    {
      "epoch": 4.313551815766164,
      "grad_norm": 0.35493436455726624,
      "learning_rate": 1e-05,
      "loss": 1.014,
      "step": 4870
    },
    {
      "epoch": 4.317980513728964,
      "grad_norm": 0.3901638984680176,
      "learning_rate": 1e-05,
      "loss": 1.0704,
      "step": 4875
    },
    {
      "epoch": 4.322409211691762,
      "grad_norm": 0.3743135631084442,
      "learning_rate": 1e-05,
      "loss": 1.078,
      "step": 4880
    },
    {
      "epoch": 4.326837909654562,
      "grad_norm": 0.4754677414894104,
      "learning_rate": 1e-05,
      "loss": 1.0855,
      "step": 4885
    },
    {
      "epoch": 4.33126660761736,
      "grad_norm": 0.38147851824760437,
      "learning_rate": 1e-05,
      "loss": 1.0449,
      "step": 4890
    },
    {
      "epoch": 4.33569530558016,
      "grad_norm": 0.4686789810657501,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 4895
    },
    {
      "epoch": 4.340124003542958,
      "grad_norm": 0.4210522770881653,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 4900
    },
    {
      "epoch": 4.344552701505758,
      "grad_norm": 0.4124254286289215,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 4905
    },
    {
      "epoch": 4.348981399468556,
      "grad_norm": 0.39260315895080566,
      "learning_rate": 1e-05,
      "loss": 1.037,
      "step": 4910
    },
    {
      "epoch": 4.353410097431356,
      "grad_norm": 0.3679765462875366,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 4915
    },
    {
      "epoch": 4.357838795394154,
      "grad_norm": 0.3751234710216522,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 4920
    },
    {
      "epoch": 4.362267493356953,
      "grad_norm": 0.3851834237575531,
      "learning_rate": 1e-05,
      "loss": 1.0381,
      "step": 4925
    },
    {
      "epoch": 4.366696191319752,
      "grad_norm": 0.33739200234413147,
      "learning_rate": 1e-05,
      "loss": 1.0989,
      "step": 4930
    },
    {
      "epoch": 4.371124889282551,
      "grad_norm": 0.30182185769081116,
      "learning_rate": 1e-05,
      "loss": 1.0277,
      "step": 4935
    },
    {
      "epoch": 4.37555358724535,
      "grad_norm": 0.26282113790512085,
      "learning_rate": 1e-05,
      "loss": 1.0479,
      "step": 4940
    },
    {
      "epoch": 4.379982285208149,
      "grad_norm": 0.37779316306114197,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 4945
    },
    {
      "epoch": 4.384410983170948,
      "grad_norm": 0.3736492693424225,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 4950
    },
    {
      "epoch": 4.388839681133747,
      "grad_norm": 0.4341072142124176,
      "learning_rate": 1e-05,
      "loss": 1.0483,
      "step": 4955
    },
    {
      "epoch": 4.393268379096545,
      "grad_norm": 0.34500136971473694,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 4960
    },
    {
      "epoch": 4.397697077059345,
      "grad_norm": 0.29406043887138367,
      "learning_rate": 1e-05,
      "loss": 1.0256,
      "step": 4965
    },
    {
      "epoch": 4.402125775022143,
      "grad_norm": 0.3825207054615021,
      "learning_rate": 1e-05,
      "loss": 1.05,
      "step": 4970
    },
    {
      "epoch": 4.406554472984943,
      "grad_norm": 0.34601831436157227,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 4975
    },
    {
      "epoch": 4.410983170947741,
      "grad_norm": 0.30903878808021545,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 4980
    },
    {
      "epoch": 4.415411868910541,
      "grad_norm": 0.35714828968048096,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 4985
    },
    {
      "epoch": 4.419840566873339,
      "grad_norm": 0.47219887375831604,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 4990
    },
    {
      "epoch": 4.424269264836139,
      "grad_norm": 0.37371787428855896,
      "learning_rate": 1e-05,
      "loss": 1.0335,
      "step": 4995
    },
    {
      "epoch": 4.428697962798937,
      "grad_norm": 0.3369366228580475,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 5000
    },
    {
      "epoch": 4.433126660761736,
      "grad_norm": 0.2944571375846863,
      "learning_rate": 1e-05,
      "loss": 1.0388,
      "step": 5005
    },
    {
      "epoch": 4.437555358724535,
      "grad_norm": 0.3116125762462616,
      "learning_rate": 1e-05,
      "loss": 1.0226,
      "step": 5010
    },
    {
      "epoch": 4.441984056687334,
      "grad_norm": 0.31112319231033325,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 5015
    },
    {
      "epoch": 4.446412754650133,
      "grad_norm": 0.3763453960418701,
      "learning_rate": 1e-05,
      "loss": 1.0616,
      "step": 5020
    },
    {
      "epoch": 4.450841452612932,
      "grad_norm": 0.3956683278083801,
      "learning_rate": 1e-05,
      "loss": 1.0688,
      "step": 5025
    },
    {
      "epoch": 4.455270150575731,
      "grad_norm": 0.47690436244010925,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 5030
    },
    {
      "epoch": 4.45969884853853,
      "grad_norm": 0.41445139050483704,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 5035
    },
    {
      "epoch": 4.464127546501329,
      "grad_norm": 0.45534929633140564,
      "learning_rate": 1e-05,
      "loss": 1.0363,
      "step": 5040
    },
    {
      "epoch": 4.468556244464128,
      "grad_norm": 0.3589898347854614,
      "learning_rate": 1e-05,
      "loss": 1.0277,
      "step": 5045
    },
    {
      "epoch": 4.472984942426926,
      "grad_norm": 0.38962066173553467,
      "learning_rate": 1e-05,
      "loss": 1.0148,
      "step": 5050
    },
    {
      "epoch": 4.477413640389726,
      "grad_norm": 0.5035517811775208,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 5055
    },
    {
      "epoch": 4.481842338352524,
      "grad_norm": 0.4337952733039856,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 5060
    },
    {
      "epoch": 4.4862710363153235,
      "grad_norm": 0.4255777597427368,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 5065
    },
    {
      "epoch": 4.490699734278122,
      "grad_norm": 0.3382212519645691,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 5070
    },
    {
      "epoch": 4.4951284322409215,
      "grad_norm": 0.3560464382171631,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 5075
    },
    {
      "epoch": 4.49955713020372,
      "grad_norm": 0.3801257610321045,
      "learning_rate": 1e-05,
      "loss": 1.0553,
      "step": 5080
    },
    {
      "epoch": 4.503985828166519,
      "grad_norm": 0.4087178707122803,
      "learning_rate": 1e-05,
      "loss": 1.0257,
      "step": 5085
    },
    {
      "epoch": 4.508414526129318,
      "grad_norm": 0.32567787170410156,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 5090
    },
    {
      "epoch": 4.512843224092117,
      "grad_norm": 0.2973158657550812,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 5095
    },
    {
      "epoch": 4.517271922054916,
      "grad_norm": 0.40281176567077637,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 5100
    },
    {
      "epoch": 4.5217006200177146,
      "grad_norm": 0.3604399263858795,
      "learning_rate": 1e-05,
      "loss": 1.1224,
      "step": 5105
    },
    {
      "epoch": 4.526129317980514,
      "grad_norm": 0.3796055018901825,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 5110
    },
    {
      "epoch": 4.5305580159433125,
      "grad_norm": 0.35533571243286133,
      "learning_rate": 1e-05,
      "loss": 1.0238,
      "step": 5115
    },
    {
      "epoch": 4.534986713906112,
      "grad_norm": 0.40700632333755493,
      "learning_rate": 1e-05,
      "loss": 1.0199,
      "step": 5120
    },
    {
      "epoch": 4.5394154118689105,
      "grad_norm": 0.39343124628067017,
      "learning_rate": 1e-05,
      "loss": 1.0286,
      "step": 5125
    },
    {
      "epoch": 4.54384410983171,
      "grad_norm": 0.40103879570961,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 5130
    },
    {
      "epoch": 4.5482728077945085,
      "grad_norm": 0.4494595527648926,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 5135
    },
    {
      "epoch": 4.552701505757307,
      "grad_norm": 0.35697704553604126,
      "learning_rate": 1e-05,
      "loss": 1.0345,
      "step": 5140
    },
    {
      "epoch": 4.5571302037201065,
      "grad_norm": 0.3446771800518036,
      "learning_rate": 1e-05,
      "loss": 1.0439,
      "step": 5145
    },
    {
      "epoch": 4.561558901682905,
      "grad_norm": 0.3937239646911621,
      "learning_rate": 1e-05,
      "loss": 1.0465,
      "step": 5150
    },
    {
      "epoch": 4.565987599645704,
      "grad_norm": 0.3969931900501251,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 5155
    },
    {
      "epoch": 4.570416297608503,
      "grad_norm": 0.31436094641685486,
      "learning_rate": 1e-05,
      "loss": 1.0385,
      "step": 5160
    },
    {
      "epoch": 4.574844995571302,
      "grad_norm": 0.29186370968818665,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 5165
    },
    {
      "epoch": 4.579273693534101,
      "grad_norm": 0.37925904989242554,
      "learning_rate": 1e-05,
      "loss": 1.0557,
      "step": 5170
    },
    {
      "epoch": 4.5837023914968995,
      "grad_norm": 0.342354953289032,
      "learning_rate": 1e-05,
      "loss": 1.0477,
      "step": 5175
    },
    {
      "epoch": 4.588131089459699,
      "grad_norm": 0.3777170777320862,
      "learning_rate": 1e-05,
      "loss": 1.0643,
      "step": 5180
    },
    {
      "epoch": 4.5925597874224975,
      "grad_norm": 0.28083372116088867,
      "learning_rate": 1e-05,
      "loss": 1.0472,
      "step": 5185
    },
    {
      "epoch": 4.596988485385297,
      "grad_norm": 0.4039880931377411,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 5190
    },
    {
      "epoch": 4.6014171833480955,
      "grad_norm": 0.41336092352867126,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 5195
    },
    {
      "epoch": 4.605845881310895,
      "grad_norm": 0.30506569147109985,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 5200
    },
    {
      "epoch": 4.610274579273693,
      "grad_norm": 0.33881711959838867,
      "learning_rate": 1e-05,
      "loss": 1.0301,
      "step": 5205
    },
    {
      "epoch": 4.614703277236492,
      "grad_norm": 0.30085551738739014,
      "learning_rate": 1e-05,
      "loss": 1.0335,
      "step": 5210
    },
    {
      "epoch": 4.619131975199291,
      "grad_norm": 0.28912144899368286,
      "learning_rate": 1e-05,
      "loss": 1.0556,
      "step": 5215
    },
    {
      "epoch": 4.62356067316209,
      "grad_norm": 0.3428463935852051,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 5220
    },
    {
      "epoch": 4.627989371124889,
      "grad_norm": 0.3984996974468231,
      "learning_rate": 1e-05,
      "loss": 1.0655,
      "step": 5225
    },
    {
      "epoch": 4.632418069087688,
      "grad_norm": 0.39433929324150085,
      "learning_rate": 1e-05,
      "loss": 1.0554,
      "step": 5230
    },
    {
      "epoch": 4.636846767050487,
      "grad_norm": 0.3705815374851227,
      "learning_rate": 1e-05,
      "loss": 1.0477,
      "step": 5235
    },
    {
      "epoch": 4.641275465013286,
      "grad_norm": 0.366614431142807,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 5240
    },
    {
      "epoch": 4.645704162976085,
      "grad_norm": 0.45391011238098145,
      "learning_rate": 1e-05,
      "loss": 1.0658,
      "step": 5245
    },
    {
      "epoch": 4.650132860938884,
      "grad_norm": 0.32938307523727417,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 5250
    },
    {
      "epoch": 4.654561558901683,
      "grad_norm": 0.36539316177368164,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 5255
    },
    {
      "epoch": 4.658990256864482,
      "grad_norm": 0.3959774076938629,
      "learning_rate": 1e-05,
      "loss": 1.0238,
      "step": 5260
    },
    {
      "epoch": 4.66341895482728,
      "grad_norm": 0.3841852843761444,
      "learning_rate": 1e-05,
      "loss": 1.0748,
      "step": 5265
    },
    {
      "epoch": 4.66784765279008,
      "grad_norm": 0.3290482461452484,
      "learning_rate": 1e-05,
      "loss": 1.0626,
      "step": 5270
    },
    {
      "epoch": 4.672276350752878,
      "grad_norm": 0.47915369272232056,
      "learning_rate": 1e-05,
      "loss": 1.0348,
      "step": 5275
    },
    {
      "epoch": 4.676705048715678,
      "grad_norm": 0.565383791923523,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 5280
    },
    {
      "epoch": 4.681133746678476,
      "grad_norm": 0.3323745131492615,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 5285
    },
    {
      "epoch": 4.685562444641276,
      "grad_norm": 0.37112122774124146,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 5290
    },
    {
      "epoch": 4.689991142604074,
      "grad_norm": 0.38679757714271545,
      "learning_rate": 1e-05,
      "loss": 1.0413,
      "step": 5295
    },
    {
      "epoch": 4.694419840566873,
      "grad_norm": 0.3508121371269226,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 5300
    },
    {
      "epoch": 4.698848538529672,
      "grad_norm": 0.4351249039173126,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 5305
    },
    {
      "epoch": 4.703277236492471,
      "grad_norm": 0.42216530442237854,
      "learning_rate": 1e-05,
      "loss": 1.0333,
      "step": 5310
    },
    {
      "epoch": 4.70770593445527,
      "grad_norm": 0.3658766448497772,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 5315
    },
    {
      "epoch": 4.712134632418069,
      "grad_norm": 0.3895967900753021,
      "learning_rate": 1e-05,
      "loss": 1.0463,
      "step": 5320
    },
    {
      "epoch": 4.716563330380868,
      "grad_norm": 0.2990342080593109,
      "learning_rate": 1e-05,
      "loss": 1.0478,
      "step": 5325
    },
    {
      "epoch": 4.720992028343667,
      "grad_norm": 0.3200947940349579,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 5330
    },
    {
      "epoch": 4.725420726306466,
      "grad_norm": 0.3705001175403595,
      "learning_rate": 1e-05,
      "loss": 1.0254,
      "step": 5335
    },
    {
      "epoch": 4.729849424269265,
      "grad_norm": 0.29840797185897827,
      "learning_rate": 1e-05,
      "loss": 1.0335,
      "step": 5340
    },
    {
      "epoch": 4.734278122232064,
      "grad_norm": 0.2868799865245819,
      "learning_rate": 1e-05,
      "loss": 1.0111,
      "step": 5345
    },
    {
      "epoch": 4.738706820194863,
      "grad_norm": 0.3441306948661804,
      "learning_rate": 1e-05,
      "loss": 1.0187,
      "step": 5350
    },
    {
      "epoch": 4.743135518157661,
      "grad_norm": 0.3503575026988983,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 5355
    },
    {
      "epoch": 4.747564216120461,
      "grad_norm": 0.36174270510673523,
      "learning_rate": 1e-05,
      "loss": 1.0814,
      "step": 5360
    },
    {
      "epoch": 4.751992914083259,
      "grad_norm": 0.37903475761413574,
      "learning_rate": 1e-05,
      "loss": 1.0349,
      "step": 5365
    },
    {
      "epoch": 4.756421612046059,
      "grad_norm": 0.300153523683548,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 5370
    },
    {
      "epoch": 4.760850310008857,
      "grad_norm": 0.45964720845222473,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 5375
    },
    {
      "epoch": 4.765279007971657,
      "grad_norm": 0.33609259128570557,
      "learning_rate": 1e-05,
      "loss": 1.0273,
      "step": 5380
    },
    {
      "epoch": 4.769707705934455,
      "grad_norm": 0.389454185962677,
      "learning_rate": 1e-05,
      "loss": 1.0324,
      "step": 5385
    },
    {
      "epoch": 4.774136403897254,
      "grad_norm": 0.3468781113624573,
      "learning_rate": 1e-05,
      "loss": 1.049,
      "step": 5390
    },
    {
      "epoch": 4.778565101860053,
      "grad_norm": 0.3363134562969208,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 5395
    },
    {
      "epoch": 4.782993799822852,
      "grad_norm": 0.43030187487602234,
      "learning_rate": 1e-05,
      "loss": 1.0199,
      "step": 5400
    },
    {
      "epoch": 4.787422497785651,
      "grad_norm": 0.30518239736557007,
      "learning_rate": 1e-05,
      "loss": 1.0708,
      "step": 5405
    },
    {
      "epoch": 4.79185119574845,
      "grad_norm": 0.3768819570541382,
      "learning_rate": 1e-05,
      "loss": 1.0709,
      "step": 5410
    },
    {
      "epoch": 4.796279893711249,
      "grad_norm": 0.26274043321609497,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 5415
    },
    {
      "epoch": 4.800708591674048,
      "grad_norm": 0.28756001591682434,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 5420
    },
    {
      "epoch": 4.805137289636846,
      "grad_norm": 0.3681223392486572,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 5425
    },
    {
      "epoch": 4.809565987599646,
      "grad_norm": 0.3625973165035248,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 5430
    },
    {
      "epoch": 4.813994685562444,
      "grad_norm": 0.34455597400665283,
      "learning_rate": 1e-05,
      "loss": 1.0193,
      "step": 5435
    },
    {
      "epoch": 4.818423383525244,
      "grad_norm": 0.4125065207481384,
      "learning_rate": 1e-05,
      "loss": 1.0597,
      "step": 5440
    },
    {
      "epoch": 4.822852081488042,
      "grad_norm": 0.5526455640792847,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 5445
    },
    {
      "epoch": 4.827280779450842,
      "grad_norm": 0.35034605860710144,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 5450
    },
    {
      "epoch": 4.83170947741364,
      "grad_norm": 0.3632095754146576,
      "learning_rate": 1e-05,
      "loss": 1.031,
      "step": 5455
    },
    {
      "epoch": 4.83613817537644,
      "grad_norm": 0.4786195158958435,
      "learning_rate": 1e-05,
      "loss": 1.0573,
      "step": 5460
    },
    {
      "epoch": 4.840566873339238,
      "grad_norm": 0.328865647315979,
      "learning_rate": 1e-05,
      "loss": 1.0393,
      "step": 5465
    },
    {
      "epoch": 4.844995571302038,
      "grad_norm": 0.3569698929786682,
      "learning_rate": 1e-05,
      "loss": 1.052,
      "step": 5470
    },
    {
      "epoch": 4.849424269264836,
      "grad_norm": 0.3329259753227234,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 5475
    },
    {
      "epoch": 4.853852967227635,
      "grad_norm": 0.4242902994155884,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 5480
    },
    {
      "epoch": 4.858281665190434,
      "grad_norm": 0.3652363419532776,
      "learning_rate": 1e-05,
      "loss": 1.0326,
      "step": 5485
    },
    {
      "epoch": 4.862710363153233,
      "grad_norm": 0.35778918862342834,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 5490
    },
    {
      "epoch": 4.867139061116032,
      "grad_norm": 0.3289540410041809,
      "learning_rate": 1e-05,
      "loss": 1.0494,
      "step": 5495
    },
    {
      "epoch": 4.871567759078831,
      "grad_norm": 0.34093955159187317,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 5500
    },
    {
      "epoch": 4.87599645704163,
      "grad_norm": 0.46902990341186523,
      "learning_rate": 1e-05,
      "loss": 1.0458,
      "step": 5505
    },
    {
      "epoch": 4.880425155004429,
      "grad_norm": 0.3542724847793579,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 5510
    },
    {
      "epoch": 4.884853852967227,
      "grad_norm": 0.2924833297729492,
      "learning_rate": 1e-05,
      "loss": 1.0801,
      "step": 5515
    },
    {
      "epoch": 4.889282550930027,
      "grad_norm": 0.28937458992004395,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 5520
    },
    {
      "epoch": 4.893711248892825,
      "grad_norm": 0.3082599937915802,
      "learning_rate": 1e-05,
      "loss": 1.0456,
      "step": 5525
    },
    {
      "epoch": 4.898139946855625,
      "grad_norm": 0.34604334831237793,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 5530
    },
    {
      "epoch": 4.902568644818423,
      "grad_norm": 0.3090795576572418,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 5535
    },
    {
      "epoch": 4.906997342781223,
      "grad_norm": 0.41077151894569397,
      "learning_rate": 1e-05,
      "loss": 1.0241,
      "step": 5540
    },
    {
      "epoch": 4.911426040744021,
      "grad_norm": 0.4031723141670227,
      "learning_rate": 1e-05,
      "loss": 1.0385,
      "step": 5545
    },
    {
      "epoch": 4.9158547387068205,
      "grad_norm": 0.4754953980445862,
      "learning_rate": 1e-05,
      "loss": 0.9347,
      "step": 5550
    },
    {
      "epoch": 4.920283436669619,
      "grad_norm": 0.43536168336868286,
      "learning_rate": 1e-05,
      "loss": 1.0214,
      "step": 5555
    },
    {
      "epoch": 4.9247121346324185,
      "grad_norm": 0.49753400683403015,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 5560
    },
    {
      "epoch": 4.929140832595217,
      "grad_norm": 0.43414339423179626,
      "learning_rate": 1e-05,
      "loss": 1.0569,
      "step": 5565
    },
    {
      "epoch": 4.933569530558016,
      "grad_norm": 0.4094363749027252,
      "learning_rate": 1e-05,
      "loss": 1.0604,
      "step": 5570
    },
    {
      "epoch": 4.937998228520815,
      "grad_norm": 0.41692155599594116,
      "learning_rate": 1e-05,
      "loss": 1.018,
      "step": 5575
    },
    {
      "epoch": 4.942426926483614,
      "grad_norm": 0.38908472657203674,
      "learning_rate": 1e-05,
      "loss": 1.0606,
      "step": 5580
    },
    {
      "epoch": 4.946855624446413,
      "grad_norm": 0.3818272650241852,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 5585
    },
    {
      "epoch": 4.951284322409212,
      "grad_norm": 0.3907962441444397,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 5590
    },
    {
      "epoch": 4.955713020372011,
      "grad_norm": 0.33932164311408997,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 5595
    },
    {
      "epoch": 4.9601417183348095,
      "grad_norm": 0.31713759899139404,
      "learning_rate": 1e-05,
      "loss": 1.0512,
      "step": 5600
    },
    {
      "epoch": 4.964570416297608,
      "grad_norm": 0.31604087352752686,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 5605
    },
    {
      "epoch": 4.9689991142604075,
      "grad_norm": 0.38807857036590576,
      "learning_rate": 1e-05,
      "loss": 1.0524,
      "step": 5610
    },
    {
      "epoch": 4.973427812223206,
      "grad_norm": 0.3664707839488983,
      "learning_rate": 1e-05,
      "loss": 1.0314,
      "step": 5615
    },
    {
      "epoch": 4.9778565101860055,
      "grad_norm": 0.40883156657218933,
      "learning_rate": 1e-05,
      "loss": 1.0402,
      "step": 5620
    },
    {
      "epoch": 4.982285208148804,
      "grad_norm": 0.40304985642433167,
      "learning_rate": 1e-05,
      "loss": 1.014,
      "step": 5625
    },
    {
      "epoch": 4.9867139061116035,
      "grad_norm": 0.37292009592056274,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 5630
    },
    {
      "epoch": 4.991142604074402,
      "grad_norm": 0.46004778146743774,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 5635
    },
    {
      "epoch": 4.995571302037201,
      "grad_norm": 0.3352963328361511,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 5640
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.28555047512054443,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 5645
    },
    {
      "epoch": 5.0044286979627985,
      "grad_norm": 0.35492247343063354,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 5650
    },
    {
      "epoch": 5.008857395925598,
      "grad_norm": 0.41121748089790344,
      "learning_rate": 1e-05,
      "loss": 1.0374,
      "step": 5655
    },
    {
      "epoch": 5.0132860938883965,
      "grad_norm": 0.3643803298473358,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 5660
    },
    {
      "epoch": 5.017714791851196,
      "grad_norm": 0.373393714427948,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 5665
    },
    {
      "epoch": 5.0221434898139945,
      "grad_norm": 0.44527116417884827,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 5670
    },
    {
      "epoch": 5.026572187776794,
      "grad_norm": 0.3466438353061676,
      "learning_rate": 1e-05,
      "loss": 1.043,
      "step": 5675
    },
    {
      "epoch": 5.0310008857395925,
      "grad_norm": 0.36244356632232666,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 5680
    },
    {
      "epoch": 5.035429583702392,
      "grad_norm": 0.39741718769073486,
      "learning_rate": 1e-05,
      "loss": 1.036,
      "step": 5685
    },
    {
      "epoch": 5.0398582816651905,
      "grad_norm": 0.28051066398620605,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 5690
    },
    {
      "epoch": 5.044286979627989,
      "grad_norm": 0.3079621195793152,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 5695
    },
    {
      "epoch": 5.048715677590788,
      "grad_norm": 0.3123464584350586,
      "learning_rate": 1e-05,
      "loss": 1.0388,
      "step": 5700
    },
    {
      "epoch": 5.053144375553587,
      "grad_norm": 0.40218472480773926,
      "learning_rate": 1e-05,
      "loss": 1.0483,
      "step": 5705
    },
    {
      "epoch": 5.057573073516386,
      "grad_norm": 0.3287288248538971,
      "learning_rate": 1e-05,
      "loss": 1.0642,
      "step": 5710
    },
    {
      "epoch": 5.062001771479185,
      "grad_norm": 0.30074939131736755,
      "learning_rate": 1e-05,
      "loss": 1.0579,
      "step": 5715
    },
    {
      "epoch": 5.066430469441984,
      "grad_norm": 0.37240633368492126,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 5720
    },
    {
      "epoch": 5.070859167404783,
      "grad_norm": 0.3434273600578308,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 5725
    },
    {
      "epoch": 5.075287865367582,
      "grad_norm": 0.44496390223503113,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 5730
    },
    {
      "epoch": 5.079716563330381,
      "grad_norm": 0.3801027834415436,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 5735
    },
    {
      "epoch": 5.0841452612931795,
      "grad_norm": 0.3687537908554077,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 5740
    },
    {
      "epoch": 5.088573959255979,
      "grad_norm": 0.3070183992385864,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 5745
    },
    {
      "epoch": 5.093002657218777,
      "grad_norm": 0.2721776068210602,
      "learning_rate": 1e-05,
      "loss": 1.0342,
      "step": 5750
    },
    {
      "epoch": 5.097431355181577,
      "grad_norm": 0.43468162417411804,
      "learning_rate": 1e-05,
      "loss": 1.0439,
      "step": 5755
    },
    {
      "epoch": 5.101860053144375,
      "grad_norm": 0.4132360816001892,
      "learning_rate": 1e-05,
      "loss": 1.0348,
      "step": 5760
    },
    {
      "epoch": 5.106288751107175,
      "grad_norm": 0.34460538625717163,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 5765
    },
    {
      "epoch": 5.110717449069973,
      "grad_norm": 0.31522879004478455,
      "learning_rate": 1e-05,
      "loss": 1.0915,
      "step": 5770
    },
    {
      "epoch": 5.115146147032772,
      "grad_norm": 0.40316328406333923,
      "learning_rate": 1e-05,
      "loss": 1.0537,
      "step": 5775
    },
    {
      "epoch": 5.119574844995571,
      "grad_norm": 0.3921210765838623,
      "learning_rate": 1e-05,
      "loss": 1.0297,
      "step": 5780
    },
    {
      "epoch": 5.12400354295837,
      "grad_norm": 0.3546719253063202,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 5785
    },
    {
      "epoch": 5.128432240921169,
      "grad_norm": 0.29872897267341614,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 5790
    },
    {
      "epoch": 5.132860938883968,
      "grad_norm": 0.316112220287323,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 5795
    },
    {
      "epoch": 5.137289636846767,
      "grad_norm": 0.4852432310581207,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 5800
    },
    {
      "epoch": 5.141718334809566,
      "grad_norm": 0.46665048599243164,
      "learning_rate": 1e-05,
      "loss": 1.023,
      "step": 5805
    },
    {
      "epoch": 5.146147032772365,
      "grad_norm": 0.30212077498435974,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 5810
    },
    {
      "epoch": 5.150575730735164,
      "grad_norm": 0.37133854627609253,
      "learning_rate": 1e-05,
      "loss": 1.0369,
      "step": 5815
    },
    {
      "epoch": 5.155004428697962,
      "grad_norm": 0.3628295958042145,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 5820
    },
    {
      "epoch": 5.159433126660762,
      "grad_norm": 0.5432576537132263,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 5825
    },
    {
      "epoch": 5.16386182462356,
      "grad_norm": 0.36904704570770264,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 5830
    },
    {
      "epoch": 5.16829052258636,
      "grad_norm": 0.2848658859729767,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 5835
    },
    {
      "epoch": 5.172719220549158,
      "grad_norm": 0.3406495451927185,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 5840
    },
    {
      "epoch": 5.177147918511958,
      "grad_norm": 0.32348519563674927,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 5845
    },
    {
      "epoch": 5.181576616474756,
      "grad_norm": 0.3358650803565979,
      "learning_rate": 1e-05,
      "loss": 1.0514,
      "step": 5850
    },
    {
      "epoch": 5.186005314437556,
      "grad_norm": 0.32775822281837463,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 5855
    },
    {
      "epoch": 5.190434012400354,
      "grad_norm": 0.42163965106010437,
      "learning_rate": 1e-05,
      "loss": 1.0257,
      "step": 5860
    },
    {
      "epoch": 5.194862710363153,
      "grad_norm": 0.30723923444747925,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 5865
    },
    {
      "epoch": 5.199291408325952,
      "grad_norm": 0.44126787781715393,
      "learning_rate": 1e-05,
      "loss": 1.0324,
      "step": 5870
    },
    {
      "epoch": 5.203720106288751,
      "grad_norm": 0.4131248891353607,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 5875
    },
    {
      "epoch": 5.20814880425155,
      "grad_norm": 0.29048019647598267,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 5880
    },
    {
      "epoch": 5.212577502214349,
      "grad_norm": 0.3277253806591034,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 5885
    },
    {
      "epoch": 5.217006200177148,
      "grad_norm": 0.2832876145839691,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 5890
    },
    {
      "epoch": 5.221434898139947,
      "grad_norm": 0.4724528193473816,
      "learning_rate": 1e-05,
      "loss": 1.0942,
      "step": 5895
    },
    {
      "epoch": 5.225863596102746,
      "grad_norm": 0.4407491087913513,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 5900
    },
    {
      "epoch": 5.230292294065545,
      "grad_norm": 0.3977293074131012,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 5905
    },
    {
      "epoch": 5.234720992028343,
      "grad_norm": 0.35610103607177734,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 5910
    },
    {
      "epoch": 5.239149689991143,
      "grad_norm": 0.3498757779598236,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 5915
    },
    {
      "epoch": 5.243578387953941,
      "grad_norm": 0.35207265615463257,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 5920
    },
    {
      "epoch": 5.248007085916741,
      "grad_norm": 0.287803590297699,
      "learning_rate": 1e-05,
      "loss": 1.0526,
      "step": 5925
    },
    {
      "epoch": 5.252435783879539,
      "grad_norm": 0.30702874064445496,
      "learning_rate": 1e-05,
      "loss": 1.0434,
      "step": 5930
    },
    {
      "epoch": 5.256864481842339,
      "grad_norm": 0.3165454566478729,
      "learning_rate": 1e-05,
      "loss": 1.0441,
      "step": 5935
    },
    {
      "epoch": 5.261293179805137,
      "grad_norm": 0.3398440182209015,
      "learning_rate": 1e-05,
      "loss": 1.0405,
      "step": 5940
    },
    {
      "epoch": 5.265721877767936,
      "grad_norm": 0.29266223311424255,
      "learning_rate": 1e-05,
      "loss": 1.0724,
      "step": 5945
    },
    {
      "epoch": 5.270150575730735,
      "grad_norm": 0.46261587738990784,
      "learning_rate": 1e-05,
      "loss": 1.0193,
      "step": 5950
    },
    {
      "epoch": 5.274579273693534,
      "grad_norm": 0.3286849856376648,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 5955
    },
    {
      "epoch": 5.279007971656333,
      "grad_norm": 0.3285291790962219,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 5960
    },
    {
      "epoch": 5.283436669619132,
      "grad_norm": 0.34139150381088257,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 5965
    },
    {
      "epoch": 5.287865367581931,
      "grad_norm": 0.41966500878334045,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 5970
    },
    {
      "epoch": 5.29229406554473,
      "grad_norm": 0.47533154487609863,
      "learning_rate": 1e-05,
      "loss": 1.0204,
      "step": 5975
    },
    {
      "epoch": 5.296722763507529,
      "grad_norm": 0.34932243824005127,
      "learning_rate": 1e-05,
      "loss": 1.0951,
      "step": 5980
    },
    {
      "epoch": 5.301151461470328,
      "grad_norm": 0.3282047510147095,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 5985
    },
    {
      "epoch": 5.305580159433127,
      "grad_norm": 0.382445752620697,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 5990
    },
    {
      "epoch": 5.310008857395926,
      "grad_norm": 0.39635875821113586,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 5995
    },
    {
      "epoch": 5.314437555358724,
      "grad_norm": 0.5262877941131592,
      "learning_rate": 1e-05,
      "loss": 1.0692,
      "step": 6000
    },
    {
      "epoch": 5.318866253321524,
      "grad_norm": 0.4436338245868683,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 6005
    },
    {
      "epoch": 5.323294951284322,
      "grad_norm": 0.26665282249450684,
      "learning_rate": 1e-05,
      "loss": 1.0315,
      "step": 6010
    },
    {
      "epoch": 5.327723649247122,
      "grad_norm": 0.27709686756134033,
      "learning_rate": 1e-05,
      "loss": 1.0408,
      "step": 6015
    },
    {
      "epoch": 5.33215234720992,
      "grad_norm": 0.25978901982307434,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 6020
    },
    {
      "epoch": 5.33658104517272,
      "grad_norm": 0.4198380708694458,
      "learning_rate": 1e-05,
      "loss": 1.0478,
      "step": 6025
    },
    {
      "epoch": 5.341009743135518,
      "grad_norm": 0.35256171226501465,
      "learning_rate": 1e-05,
      "loss": 1.0658,
      "step": 6030
    },
    {
      "epoch": 5.345438441098317,
      "grad_norm": 0.3989357054233551,
      "learning_rate": 1e-05,
      "loss": 1.0334,
      "step": 6035
    },
    {
      "epoch": 5.349867139061116,
      "grad_norm": 0.3216528296470642,
      "learning_rate": 1e-05,
      "loss": 1.0454,
      "step": 6040
    },
    {
      "epoch": 5.354295837023915,
      "grad_norm": 0.3307344913482666,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 6045
    },
    {
      "epoch": 5.358724534986714,
      "grad_norm": 0.3836325407028198,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 6050
    },
    {
      "epoch": 5.363153232949513,
      "grad_norm": 0.29982152581214905,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 6055
    },
    {
      "epoch": 5.367581930912312,
      "grad_norm": 0.3325580060482025,
      "learning_rate": 1e-05,
      "loss": 1.0861,
      "step": 6060
    },
    {
      "epoch": 5.372010628875111,
      "grad_norm": 0.3494728207588196,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 6065
    },
    {
      "epoch": 5.37643932683791,
      "grad_norm": 0.3120420575141907,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 6070
    },
    {
      "epoch": 5.380868024800709,
      "grad_norm": 0.2987075448036194,
      "learning_rate": 1e-05,
      "loss": 1.0438,
      "step": 6075
    },
    {
      "epoch": 5.385296722763507,
      "grad_norm": 0.28469255566596985,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 6080
    },
    {
      "epoch": 5.389725420726307,
      "grad_norm": 0.333445280790329,
      "learning_rate": 1e-05,
      "loss": 1.0366,
      "step": 6085
    },
    {
      "epoch": 5.394154118689105,
      "grad_norm": 0.3890826404094696,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 6090
    },
    {
      "epoch": 5.3985828166519045,
      "grad_norm": 0.26968681812286377,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 6095
    },
    {
      "epoch": 5.403011514614703,
      "grad_norm": 0.30162420868873596,
      "learning_rate": 1e-05,
      "loss": 1.0351,
      "step": 6100
    },
    {
      "epoch": 5.4074402125775025,
      "grad_norm": 0.3432179391384125,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 6105
    },
    {
      "epoch": 5.411868910540301,
      "grad_norm": 0.3125027120113373,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 6110
    },
    {
      "epoch": 5.4162976085031005,
      "grad_norm": 0.3560992479324341,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 6115
    },
    {
      "epoch": 5.420726306465899,
      "grad_norm": 0.33001136779785156,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 6120
    },
    {
      "epoch": 5.425155004428698,
      "grad_norm": 0.32153934240341187,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 6125
    },
    {
      "epoch": 5.429583702391497,
      "grad_norm": 0.3118373453617096,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 6130
    },
    {
      "epoch": 5.434012400354296,
      "grad_norm": 0.2876591980457306,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 6135
    },
    {
      "epoch": 5.438441098317095,
      "grad_norm": 0.29356780648231506,
      "learning_rate": 1e-05,
      "loss": 1.0502,
      "step": 6140
    },
    {
      "epoch": 5.4428697962798935,
      "grad_norm": 0.350515753030777,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 6145
    },
    {
      "epoch": 5.447298494242693,
      "grad_norm": 0.388534814119339,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 6150
    },
    {
      "epoch": 5.4517271922054915,
      "grad_norm": 0.40516066551208496,
      "learning_rate": 1e-05,
      "loss": 1.0262,
      "step": 6155
    },
    {
      "epoch": 5.45615589016829,
      "grad_norm": 0.4120570421218872,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 6160
    },
    {
      "epoch": 5.4605845881310895,
      "grad_norm": 0.5171772241592407,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 6165
    },
    {
      "epoch": 5.465013286093888,
      "grad_norm": 0.28847429156303406,
      "learning_rate": 1e-05,
      "loss": 1.0448,
      "step": 6170
    },
    {
      "epoch": 5.4694419840566875,
      "grad_norm": 0.43600407242774963,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 6175
    },
    {
      "epoch": 5.473870682019486,
      "grad_norm": 0.5186824202537537,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 6180
    },
    {
      "epoch": 5.4782993799822854,
      "grad_norm": 0.3558749854564667,
      "learning_rate": 1e-05,
      "loss": 1.0511,
      "step": 6185
    },
    {
      "epoch": 5.482728077945084,
      "grad_norm": 0.34864541888237,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 6190
    },
    {
      "epoch": 5.487156775907883,
      "grad_norm": 0.3080834448337555,
      "learning_rate": 1e-05,
      "loss": 1.0381,
      "step": 6195
    },
    {
      "epoch": 5.491585473870682,
      "grad_norm": 0.3084205985069275,
      "learning_rate": 1e-05,
      "loss": 1.0515,
      "step": 6200
    },
    {
      "epoch": 5.496014171833481,
      "grad_norm": 0.33310824632644653,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 6205
    },
    {
      "epoch": 5.50044286979628,
      "grad_norm": 0.28664156794548035,
      "learning_rate": 1e-05,
      "loss": 1.0429,
      "step": 6210
    },
    {
      "epoch": 5.5048715677590785,
      "grad_norm": 0.43311434984207153,
      "learning_rate": 1e-05,
      "loss": 1.0374,
      "step": 6215
    },
    {
      "epoch": 5.509300265721878,
      "grad_norm": 0.3207617998123169,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 6220
    },
    {
      "epoch": 5.5137289636846765,
      "grad_norm": 0.3104935884475708,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 6225
    },
    {
      "epoch": 5.518157661647476,
      "grad_norm": 0.3671448528766632,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 6230
    },
    {
      "epoch": 5.522586359610274,
      "grad_norm": 0.2939125895500183,
      "learning_rate": 1e-05,
      "loss": 1.0523,
      "step": 6235
    },
    {
      "epoch": 5.527015057573074,
      "grad_norm": 0.2815743088722229,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 6240
    },
    {
      "epoch": 5.531443755535872,
      "grad_norm": 0.34176525473594666,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 6245
    },
    {
      "epoch": 5.535872453498671,
      "grad_norm": 0.3273089528083801,
      "learning_rate": 1e-05,
      "loss": 1.0284,
      "step": 6250
    },
    {
      "epoch": 5.54030115146147,
      "grad_norm": 0.2730557918548584,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 6255
    },
    {
      "epoch": 5.544729849424269,
      "grad_norm": 0.3903951048851013,
      "learning_rate": 1e-05,
      "loss": 1.0465,
      "step": 6260
    },
    {
      "epoch": 5.549158547387068,
      "grad_norm": 0.3390137553215027,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 6265
    },
    {
      "epoch": 5.553587245349867,
      "grad_norm": 0.46740761399269104,
      "learning_rate": 1e-05,
      "loss": 1.0443,
      "step": 6270
    },
    {
      "epoch": 5.558015943312666,
      "grad_norm": 0.4413903057575226,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 6275
    },
    {
      "epoch": 5.562444641275465,
      "grad_norm": 0.3875576853752136,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 6280
    },
    {
      "epoch": 5.566873339238264,
      "grad_norm": 0.28578558564186096,
      "learning_rate": 1e-05,
      "loss": 1.0295,
      "step": 6285
    },
    {
      "epoch": 5.571302037201063,
      "grad_norm": 0.3622230291366577,
      "learning_rate": 1e-05,
      "loss": 1.0238,
      "step": 6290
    },
    {
      "epoch": 5.575730735163862,
      "grad_norm": 0.33592015504837036,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 6295
    },
    {
      "epoch": 5.580159433126661,
      "grad_norm": 0.3847084045410156,
      "learning_rate": 1e-05,
      "loss": 1.0386,
      "step": 6300
    },
    {
      "epoch": 5.584588131089459,
      "grad_norm": 0.4086059629917145,
      "learning_rate": 1e-05,
      "loss": 1.0343,
      "step": 6305
    },
    {
      "epoch": 5.589016829052259,
      "grad_norm": 0.3374808430671692,
      "learning_rate": 1e-05,
      "loss": 1.0388,
      "step": 6310
    },
    {
      "epoch": 5.593445527015057,
      "grad_norm": 0.30118468403816223,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 6315
    },
    {
      "epoch": 5.597874224977857,
      "grad_norm": 0.37272441387176514,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 6320
    },
    {
      "epoch": 5.602302922940655,
      "grad_norm": 0.44374728202819824,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 6325
    },
    {
      "epoch": 5.606731620903455,
      "grad_norm": 0.3207448422908783,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 6330
    },
    {
      "epoch": 5.611160318866253,
      "grad_norm": 0.34067943692207336,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 6335
    },
    {
      "epoch": 5.615589016829052,
      "grad_norm": 0.3046121299266815,
      "learning_rate": 1e-05,
      "loss": 1.0542,
      "step": 6340
    },
    {
      "epoch": 5.620017714791851,
      "grad_norm": 0.3301253318786621,
      "learning_rate": 1e-05,
      "loss": 1.0189,
      "step": 6345
    },
    {
      "epoch": 5.62444641275465,
      "grad_norm": 0.3467637002468109,
      "learning_rate": 1e-05,
      "loss": 1.0373,
      "step": 6350
    },
    {
      "epoch": 5.628875110717449,
      "grad_norm": 0.35957252979278564,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 6355
    },
    {
      "epoch": 5.633303808680248,
      "grad_norm": 0.3335168957710266,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 6360
    },
    {
      "epoch": 5.637732506643047,
      "grad_norm": 0.3452107012271881,
      "learning_rate": 1e-05,
      "loss": 1.0225,
      "step": 6365
    },
    {
      "epoch": 5.642161204605846,
      "grad_norm": 0.3340684771537781,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 6370
    },
    {
      "epoch": 5.646589902568644,
      "grad_norm": 0.3786199688911438,
      "learning_rate": 1e-05,
      "loss": 1.0824,
      "step": 6375
    },
    {
      "epoch": 5.651018600531444,
      "grad_norm": 0.36782315373420715,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 6380
    },
    {
      "epoch": 5.655447298494242,
      "grad_norm": 0.378689706325531,
      "learning_rate": 1e-05,
      "loss": 1.0468,
      "step": 6385
    },
    {
      "epoch": 5.659875996457042,
      "grad_norm": 0.320144385099411,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 6390
    },
    {
      "epoch": 5.66430469441984,
      "grad_norm": 0.3400636911392212,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 6395
    },
    {
      "epoch": 5.66873339238264,
      "grad_norm": 0.3092327117919922,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 6400
    },
    {
      "epoch": 5.673162090345438,
      "grad_norm": 0.30968233942985535,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 6405
    },
    {
      "epoch": 5.677590788308238,
      "grad_norm": 0.3212548792362213,
      "learning_rate": 1e-05,
      "loss": 1.0531,
      "step": 6410
    },
    {
      "epoch": 5.682019486271036,
      "grad_norm": 0.4057113826274872,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 6415
    },
    {
      "epoch": 5.686448184233836,
      "grad_norm": 0.3626675009727478,
      "learning_rate": 1e-05,
      "loss": 1.0395,
      "step": 6420
    },
    {
      "epoch": 5.690876882196634,
      "grad_norm": 0.3725353479385376,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 6425
    },
    {
      "epoch": 5.695305580159433,
      "grad_norm": 0.3042195737361908,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 6430
    },
    {
      "epoch": 5.699734278122232,
      "grad_norm": 0.321402370929718,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 6435
    },
    {
      "epoch": 5.704162976085031,
      "grad_norm": 0.4093253016471863,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 6440
    },
    {
      "epoch": 5.70859167404783,
      "grad_norm": 0.29677829146385193,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 6445
    },
    {
      "epoch": 5.713020372010629,
      "grad_norm": 0.25508174300193787,
      "learning_rate": 1e-05,
      "loss": 1.0238,
      "step": 6450
    },
    {
      "epoch": 5.717449069973428,
      "grad_norm": 0.350716233253479,
      "learning_rate": 1e-05,
      "loss": 1.045,
      "step": 6455
    },
    {
      "epoch": 5.721877767936227,
      "grad_norm": 0.4051620662212372,
      "learning_rate": 1e-05,
      "loss": 1.0576,
      "step": 6460
    },
    {
      "epoch": 5.726306465899025,
      "grad_norm": 0.3167639970779419,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 6465
    },
    {
      "epoch": 5.730735163861825,
      "grad_norm": 0.39503225684165955,
      "learning_rate": 1e-05,
      "loss": 1.0222,
      "step": 6470
    },
    {
      "epoch": 5.735163861824623,
      "grad_norm": 0.4134061336517334,
      "learning_rate": 1e-05,
      "loss": 1.0518,
      "step": 6475
    },
    {
      "epoch": 5.739592559787423,
      "grad_norm": 0.3831997811794281,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 6480
    },
    {
      "epoch": 5.744021257750221,
      "grad_norm": 0.4127909541130066,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 6485
    },
    {
      "epoch": 5.748449955713021,
      "grad_norm": 0.37857189774513245,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 6490
    },
    {
      "epoch": 5.752878653675819,
      "grad_norm": 0.47841256856918335,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 6495
    },
    {
      "epoch": 5.757307351638619,
      "grad_norm": 0.4187529981136322,
      "learning_rate": 1e-05,
      "loss": 1.028,
      "step": 6500
    },
    {
      "epoch": 5.761736049601417,
      "grad_norm": 0.36477819085121155,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 6505
    },
    {
      "epoch": 5.766164747564217,
      "grad_norm": 0.3008206784725189,
      "learning_rate": 1e-05,
      "loss": 1.0461,
      "step": 6510
    },
    {
      "epoch": 5.770593445527015,
      "grad_norm": 0.2751103937625885,
      "learning_rate": 1e-05,
      "loss": 1.0708,
      "step": 6515
    },
    {
      "epoch": 5.775022143489814,
      "grad_norm": 0.28461137413978577,
      "learning_rate": 1e-05,
      "loss": 1.0415,
      "step": 6520
    },
    {
      "epoch": 5.779450841452613,
      "grad_norm": 0.4289863109588623,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 6525
    },
    {
      "epoch": 5.783879539415412,
      "grad_norm": 0.3430834412574768,
      "learning_rate": 1e-05,
      "loss": 1.0371,
      "step": 6530
    },
    {
      "epoch": 5.788308237378211,
      "grad_norm": 0.30186137557029724,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 6535
    },
    {
      "epoch": 5.79273693534101,
      "grad_norm": 0.29980048537254333,
      "learning_rate": 1e-05,
      "loss": 1.0322,
      "step": 6540
    },
    {
      "epoch": 5.797165633303809,
      "grad_norm": 0.3881379961967468,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 6545
    },
    {
      "epoch": 5.801594331266608,
      "grad_norm": 0.3307119905948639,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 6550
    },
    {
      "epoch": 5.806023029229406,
      "grad_norm": 0.34576040506362915,
      "learning_rate": 1e-05,
      "loss": 1.0481,
      "step": 6555
    },
    {
      "epoch": 5.810451727192206,
      "grad_norm": 0.34578219056129456,
      "learning_rate": 1e-05,
      "loss": 1.0477,
      "step": 6560
    },
    {
      "epoch": 5.814880425155004,
      "grad_norm": 0.3267001807689667,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 6565
    },
    {
      "epoch": 5.819309123117804,
      "grad_norm": 0.4689924418926239,
      "learning_rate": 1e-05,
      "loss": 1.0475,
      "step": 6570
    },
    {
      "epoch": 5.823737821080602,
      "grad_norm": 0.3635270595550537,
      "learning_rate": 1e-05,
      "loss": 1.091,
      "step": 6575
    },
    {
      "epoch": 5.8281665190434015,
      "grad_norm": 0.35070130228996277,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 6580
    },
    {
      "epoch": 5.8325952170062,
      "grad_norm": 0.3956962525844574,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 6585
    },
    {
      "epoch": 5.837023914968999,
      "grad_norm": 0.3035469353199005,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 6590
    },
    {
      "epoch": 5.841452612931798,
      "grad_norm": 0.33993586897850037,
      "learning_rate": 1e-05,
      "loss": 1.0385,
      "step": 6595
    },
    {
      "epoch": 5.845881310894597,
      "grad_norm": 0.38742855191230774,
      "learning_rate": 1e-05,
      "loss": 1.0265,
      "step": 6600
    },
    {
      "epoch": 5.850310008857396,
      "grad_norm": 0.23851977288722992,
      "learning_rate": 1e-05,
      "loss": 1.0263,
      "step": 6605
    },
    {
      "epoch": 5.854738706820195,
      "grad_norm": 0.3214780390262604,
      "learning_rate": 1e-05,
      "loss": 1.0386,
      "step": 6610
    },
    {
      "epoch": 5.859167404782994,
      "grad_norm": 0.3350125253200531,
      "learning_rate": 1e-05,
      "loss": 1.0447,
      "step": 6615
    },
    {
      "epoch": 5.863596102745793,
      "grad_norm": 0.2939947545528412,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 6620
    },
    {
      "epoch": 5.868024800708592,
      "grad_norm": 0.31134822964668274,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 6625
    },
    {
      "epoch": 5.8724534986713905,
      "grad_norm": 0.4234241843223572,
      "learning_rate": 1e-05,
      "loss": 1.0297,
      "step": 6630
    },
    {
      "epoch": 5.87688219663419,
      "grad_norm": 0.37806475162506104,
      "learning_rate": 1e-05,
      "loss": 1.0333,
      "step": 6635
    },
    {
      "epoch": 5.8813108945969885,
      "grad_norm": 0.2995615005493164,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 6640
    },
    {
      "epoch": 5.885739592559787,
      "grad_norm": 0.3165876865386963,
      "learning_rate": 1e-05,
      "loss": 1.0297,
      "step": 6645
    },
    {
      "epoch": 5.8901682905225865,
      "grad_norm": 0.32325923442840576,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 6650
    },
    {
      "epoch": 5.894596988485385,
      "grad_norm": 0.45908138155937195,
      "learning_rate": 1e-05,
      "loss": 1.0219,
      "step": 6655
    },
    {
      "epoch": 5.8990256864481845,
      "grad_norm": 0.35252439975738525,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 6660
    },
    {
      "epoch": 5.903454384410983,
      "grad_norm": 0.3436559736728668,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 6665
    },
    {
      "epoch": 5.9078830823737825,
      "grad_norm": 0.31436410546302795,
      "learning_rate": 1e-05,
      "loss": 1.0649,
      "step": 6670
    },
    {
      "epoch": 5.912311780336581,
      "grad_norm": 0.4750252962112427,
      "learning_rate": 1e-05,
      "loss": 1.0326,
      "step": 6675
    },
    {
      "epoch": 5.9167404782993795,
      "grad_norm": 0.34484434127807617,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 6680
    },
    {
      "epoch": 5.921169176262179,
      "grad_norm": 0.385506272315979,
      "learning_rate": 1e-05,
      "loss": 1.0489,
      "step": 6685
    },
    {
      "epoch": 5.9255978742249775,
      "grad_norm": 0.39610734581947327,
      "learning_rate": 1e-05,
      "loss": 1.0353,
      "step": 6690
    },
    {
      "epoch": 5.930026572187777,
      "grad_norm": 0.4324401021003723,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 6695
    },
    {
      "epoch": 5.9344552701505755,
      "grad_norm": 0.3518627882003784,
      "learning_rate": 1e-05,
      "loss": 1.0505,
      "step": 6700
    },
    {
      "epoch": 5.938883968113375,
      "grad_norm": 0.32452812790870667,
      "learning_rate": 1e-05,
      "loss": 1.0417,
      "step": 6705
    },
    {
      "epoch": 5.9433126660761735,
      "grad_norm": 0.4187712073326111,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 6710
    },
    {
      "epoch": 5.947741364038973,
      "grad_norm": 0.3556433320045471,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 6715
    },
    {
      "epoch": 5.9521700620017715,
      "grad_norm": 0.4036000669002533,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 6720
    },
    {
      "epoch": 5.956598759964571,
      "grad_norm": 0.2611381709575653,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 6725
    },
    {
      "epoch": 5.961027457927369,
      "grad_norm": 0.3407951593399048,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 6730
    },
    {
      "epoch": 5.965456155890168,
      "grad_norm": 0.28380531072616577,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 6735
    },
    {
      "epoch": 5.969884853852967,
      "grad_norm": 0.3297921121120453,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 6740
    },
    {
      "epoch": 5.974313551815766,
      "grad_norm": 0.3118314743041992,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 6745
    },
    {
      "epoch": 5.978742249778565,
      "grad_norm": 0.30400076508522034,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 6750
    },
    {
      "epoch": 5.983170947741364,
      "grad_norm": 0.4714670181274414,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 6755
    },
    {
      "epoch": 5.987599645704163,
      "grad_norm": 0.40930747985839844,
      "learning_rate": 1e-05,
      "loss": 1.0222,
      "step": 6760
    },
    {
      "epoch": 5.992028343666962,
      "grad_norm": 0.29554399847984314,
      "learning_rate": 1e-05,
      "loss": 1.0344,
      "step": 6765
    },
    {
      "epoch": 5.9964570416297605,
      "grad_norm": 0.4346447288990021,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 6770
    },
    {
      "epoch": 6.00088573959256,
      "grad_norm": 0.4221264719963074,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 6775
    },
    {
      "epoch": 6.005314437555358,
      "grad_norm": 0.3572789430618286,
      "learning_rate": 1e-05,
      "loss": 1.0364,
      "step": 6780
    },
    {
      "epoch": 6.009743135518158,
      "grad_norm": 0.3365592360496521,
      "learning_rate": 1e-05,
      "loss": 1.0474,
      "step": 6785
    },
    {
      "epoch": 6.014171833480956,
      "grad_norm": 0.29700925946235657,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 6790
    },
    {
      "epoch": 6.018600531443756,
      "grad_norm": 0.34870168566703796,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 6795
    },
    {
      "epoch": 6.023029229406554,
      "grad_norm": 0.3146915137767792,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 6800
    },
    {
      "epoch": 6.027457927369354,
      "grad_norm": 0.36624160408973694,
      "learning_rate": 1e-05,
      "loss": 1.0184,
      "step": 6805
    },
    {
      "epoch": 6.031886625332152,
      "grad_norm": 0.3642502725124359,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 6810
    },
    {
      "epoch": 6.036315323294951,
      "grad_norm": 0.28912538290023804,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 6815
    },
    {
      "epoch": 6.04074402125775,
      "grad_norm": 0.3548716902732849,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 6820
    },
    {
      "epoch": 6.045172719220549,
      "grad_norm": 0.3012695908546448,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 6825
    },
    {
      "epoch": 6.049601417183348,
      "grad_norm": 0.33697596192359924,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 6830
    },
    {
      "epoch": 6.054030115146147,
      "grad_norm": 0.35951241850852966,
      "learning_rate": 1e-05,
      "loss": 1.0524,
      "step": 6835
    },
    {
      "epoch": 6.058458813108946,
      "grad_norm": 0.3536849021911621,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 6840
    },
    {
      "epoch": 6.062887511071745,
      "grad_norm": 0.3755175769329071,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 6845
    },
    {
      "epoch": 6.067316209034544,
      "grad_norm": 0.4255145490169525,
      "learning_rate": 1e-05,
      "loss": 1.0248,
      "step": 6850
    },
    {
      "epoch": 6.071744906997343,
      "grad_norm": 0.39630162715911865,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 6855
    },
    {
      "epoch": 6.076173604960141,
      "grad_norm": 0.36435675621032715,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 6860
    },
    {
      "epoch": 6.080602302922941,
      "grad_norm": 0.2913612425327301,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 6865
    },
    {
      "epoch": 6.085031000885739,
      "grad_norm": 0.35636329650878906,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 6870
    },
    {
      "epoch": 6.089459698848539,
      "grad_norm": 0.3127959370613098,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 6875
    },
    {
      "epoch": 6.093888396811337,
      "grad_norm": 0.5212419629096985,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 6880
    },
    {
      "epoch": 6.098317094774137,
      "grad_norm": 0.3824059069156647,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 6885
    },
    {
      "epoch": 6.102745792736935,
      "grad_norm": 0.33459386229515076,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 6890
    },
    {
      "epoch": 6.107174490699734,
      "grad_norm": 0.34141600131988525,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 6895
    },
    {
      "epoch": 6.111603188662533,
      "grad_norm": 0.3059738576412201,
      "learning_rate": 1e-05,
      "loss": 1.0584,
      "step": 6900
    },
    {
      "epoch": 6.116031886625332,
      "grad_norm": 0.3926936388015747,
      "learning_rate": 1e-05,
      "loss": 1.0286,
      "step": 6905
    },
    {
      "epoch": 6.120460584588131,
      "grad_norm": 0.3398391604423523,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 6910
    },
    {
      "epoch": 6.12488928255093,
      "grad_norm": 0.359720379114151,
      "learning_rate": 1e-05,
      "loss": 1.0305,
      "step": 6915
    },
    {
      "epoch": 6.129317980513729,
      "grad_norm": 0.2975776493549347,
      "learning_rate": 1e-05,
      "loss": 1.0252,
      "step": 6920
    },
    {
      "epoch": 6.133746678476528,
      "grad_norm": 0.34713348746299744,
      "learning_rate": 1e-05,
      "loss": 1.027,
      "step": 6925
    },
    {
      "epoch": 6.138175376439327,
      "grad_norm": 0.3530220687389374,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 6930
    },
    {
      "epoch": 6.142604074402126,
      "grad_norm": 0.3547761142253876,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 6935
    },
    {
      "epoch": 6.147032772364924,
      "grad_norm": 0.3353806138038635,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 6940
    },
    {
      "epoch": 6.151461470327724,
      "grad_norm": 0.3081471920013428,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 6945
    },
    {
      "epoch": 6.155890168290522,
      "grad_norm": 0.364685595035553,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 6950
    },
    {
      "epoch": 6.160318866253322,
      "grad_norm": 0.35186582803726196,
      "learning_rate": 1e-05,
      "loss": 1.0361,
      "step": 6955
    },
    {
      "epoch": 6.16474756421612,
      "grad_norm": 0.35470500588417053,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 6960
    },
    {
      "epoch": 6.16917626217892,
      "grad_norm": 0.35173505544662476,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 6965
    },
    {
      "epoch": 6.173604960141718,
      "grad_norm": 0.32691577076911926,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 6970
    },
    {
      "epoch": 6.178033658104518,
      "grad_norm": 0.28472381830215454,
      "learning_rate": 1e-05,
      "loss": 1.0519,
      "step": 6975
    },
    {
      "epoch": 6.182462356067316,
      "grad_norm": 0.46015751361846924,
      "learning_rate": 1e-05,
      "loss": 1.0406,
      "step": 6980
    },
    {
      "epoch": 6.186891054030115,
      "grad_norm": 0.2897954285144806,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 6985
    },
    {
      "epoch": 6.191319751992914,
      "grad_norm": 0.2893534302711487,
      "learning_rate": 1e-05,
      "loss": 1.0625,
      "step": 6990
    },
    {
      "epoch": 6.195748449955713,
      "grad_norm": 0.3138471841812134,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 6995
    },
    {
      "epoch": 6.200177147918512,
      "grad_norm": 0.4907361567020416,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 7000
    },
    {
      "epoch": 6.204605845881311,
      "grad_norm": 0.31292858719825745,
      "learning_rate": 1e-05,
      "loss": 1.028,
      "step": 7005
    },
    {
      "epoch": 6.20903454384411,
      "grad_norm": 0.298232764005661,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 7010
    },
    {
      "epoch": 6.213463241806909,
      "grad_norm": 0.47002145648002625,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 7015
    },
    {
      "epoch": 6.217891939769708,
      "grad_norm": 0.4330032169818878,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 7020
    },
    {
      "epoch": 6.222320637732507,
      "grad_norm": 0.4534287750720978,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 7025
    },
    {
      "epoch": 6.226749335695305,
      "grad_norm": 0.3723717927932739,
      "learning_rate": 1e-05,
      "loss": 1.0756,
      "step": 7030
    },
    {
      "epoch": 6.231178033658105,
      "grad_norm": 0.3415564298629761,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 7035
    },
    {
      "epoch": 6.235606731620903,
      "grad_norm": 0.3624410331249237,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 7040
    },
    {
      "epoch": 6.240035429583703,
      "grad_norm": 0.3186006247997284,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 7045
    },
    {
      "epoch": 6.244464127546501,
      "grad_norm": 0.5132346749305725,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 7050
    },
    {
      "epoch": 6.248892825509301,
      "grad_norm": 0.3773728907108307,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 7055
    },
    {
      "epoch": 6.253321523472099,
      "grad_norm": 0.33810749650001526,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 7060
    },
    {
      "epoch": 6.257750221434899,
      "grad_norm": 0.24548912048339844,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 7065
    },
    {
      "epoch": 6.262178919397697,
      "grad_norm": 0.32872912287712097,
      "learning_rate": 1e-05,
      "loss": 1.0397,
      "step": 7070
    },
    {
      "epoch": 6.266607617360496,
      "grad_norm": 0.36943259835243225,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 7075
    },
    {
      "epoch": 6.271036315323295,
      "grad_norm": 0.4258362352848053,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 7080
    },
    {
      "epoch": 6.275465013286094,
      "grad_norm": 0.28767505288124084,
      "learning_rate": 1e-05,
      "loss": 1.05,
      "step": 7085
    },
    {
      "epoch": 6.279893711248893,
      "grad_norm": 0.3052264451980591,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 7090
    },
    {
      "epoch": 6.284322409211692,
      "grad_norm": 0.3374520540237427,
      "learning_rate": 1e-05,
      "loss": 1.0456,
      "step": 7095
    },
    {
      "epoch": 6.288751107174491,
      "grad_norm": 0.3040774166584015,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 7100
    },
    {
      "epoch": 6.29317980513729,
      "grad_norm": 0.2610459625720978,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 7105
    },
    {
      "epoch": 6.297608503100088,
      "grad_norm": 0.31213927268981934,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 7110
    },
    {
      "epoch": 6.302037201062888,
      "grad_norm": 0.2727365791797638,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 7115
    },
    {
      "epoch": 6.306465899025686,
      "grad_norm": 0.266261488199234,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 7120
    },
    {
      "epoch": 6.3108945969884855,
      "grad_norm": 0.27193114161491394,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 7125
    },
    {
      "epoch": 6.315323294951284,
      "grad_norm": 0.29136723279953003,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 7130
    },
    {
      "epoch": 6.3197519929140835,
      "grad_norm": 0.2739744484424591,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 7135
    },
    {
      "epoch": 6.324180690876882,
      "grad_norm": 0.3441196084022522,
      "learning_rate": 1e-05,
      "loss": 1.0437,
      "step": 7140
    },
    {
      "epoch": 6.3286093888396815,
      "grad_norm": 0.30992305278778076,
      "learning_rate": 1e-05,
      "loss": 1.054,
      "step": 7145
    },
    {
      "epoch": 6.33303808680248,
      "grad_norm": 0.3201008141040802,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 7150
    },
    {
      "epoch": 6.3374667847652795,
      "grad_norm": 0.3894982933998108,
      "learning_rate": 1e-05,
      "loss": 1.0232,
      "step": 7155
    },
    {
      "epoch": 6.341895482728078,
      "grad_norm": 0.2990393042564392,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 7160
    },
    {
      "epoch": 6.346324180690877,
      "grad_norm": 0.3568171262741089,
      "learning_rate": 1e-05,
      "loss": 1.0426,
      "step": 7165
    },
    {
      "epoch": 6.350752878653676,
      "grad_norm": 0.3259207010269165,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 7170
    },
    {
      "epoch": 6.3551815766164745,
      "grad_norm": 0.2781645357608795,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 7175
    },
    {
      "epoch": 6.359610274579274,
      "grad_norm": 0.3118613660335541,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 7180
    },
    {
      "epoch": 6.3640389725420725,
      "grad_norm": 0.3418436050415039,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 7185
    },
    {
      "epoch": 6.368467670504872,
      "grad_norm": 0.31584247946739197,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 7190
    },
    {
      "epoch": 6.3728963684676705,
      "grad_norm": 0.3011067807674408,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 7195
    },
    {
      "epoch": 6.377325066430469,
      "grad_norm": 0.3170275092124939,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 7200
    },
    {
      "epoch": 6.3817537643932685,
      "grad_norm": 0.31234514713287354,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 7205
    },
    {
      "epoch": 6.386182462356067,
      "grad_norm": 0.34424668550491333,
      "learning_rate": 1e-05,
      "loss": 1.0356,
      "step": 7210
    },
    {
      "epoch": 6.3906111603188664,
      "grad_norm": 0.31194567680358887,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 7215
    },
    {
      "epoch": 6.395039858281665,
      "grad_norm": 0.3398292660713196,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 7220
    },
    {
      "epoch": 6.399468556244464,
      "grad_norm": 0.3615325391292572,
      "learning_rate": 1e-05,
      "loss": 1.0593,
      "step": 7225
    },
    {
      "epoch": 6.403897254207263,
      "grad_norm": 0.3728867471218109,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 7230
    },
    {
      "epoch": 6.408325952170062,
      "grad_norm": 0.28887295722961426,
      "learning_rate": 1e-05,
      "loss": 1.0258,
      "step": 7235
    },
    {
      "epoch": 6.412754650132861,
      "grad_norm": 0.31499141454696655,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 7240
    },
    {
      "epoch": 6.4171833480956595,
      "grad_norm": 0.29244956374168396,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 7245
    },
    {
      "epoch": 6.421612046058459,
      "grad_norm": 0.3293982148170471,
      "learning_rate": 1e-05,
      "loss": 1.0491,
      "step": 7250
    },
    {
      "epoch": 6.4260407440212575,
      "grad_norm": 0.3518960773944855,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 7255
    },
    {
      "epoch": 6.430469441984057,
      "grad_norm": 0.3628379702568054,
      "learning_rate": 1e-05,
      "loss": 1.0357,
      "step": 7260
    },
    {
      "epoch": 6.4348981399468554,
      "grad_norm": 0.36089372634887695,
      "learning_rate": 1e-05,
      "loss": 1.0492,
      "step": 7265
    },
    {
      "epoch": 6.439326837909655,
      "grad_norm": 0.4071352779865265,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 7270
    },
    {
      "epoch": 6.443755535872453,
      "grad_norm": 0.32147347927093506,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 7275
    },
    {
      "epoch": 6.448184233835253,
      "grad_norm": 0.316927045583725,
      "learning_rate": 1e-05,
      "loss": 1.0383,
      "step": 7280
    },
    {
      "epoch": 6.452612931798051,
      "grad_norm": 0.412502259016037,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 7285
    },
    {
      "epoch": 6.45704162976085,
      "grad_norm": 0.3707854449748993,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 7290
    },
    {
      "epoch": 6.461470327723649,
      "grad_norm": 0.38297298550605774,
      "learning_rate": 1e-05,
      "loss": 1.0723,
      "step": 7295
    },
    {
      "epoch": 6.465899025686448,
      "grad_norm": 0.47720569372177124,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 7300
    },
    {
      "epoch": 6.470327723649247,
      "grad_norm": 0.3851393163204193,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 7305
    },
    {
      "epoch": 6.474756421612046,
      "grad_norm": 0.31918996572494507,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 7310
    },
    {
      "epoch": 6.479185119574845,
      "grad_norm": 0.3066047728061676,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 7315
    },
    {
      "epoch": 6.483613817537644,
      "grad_norm": 0.300331711769104,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 7320
    },
    {
      "epoch": 6.488042515500442,
      "grad_norm": 0.34394770860671997,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 7325
    },
    {
      "epoch": 6.492471213463242,
      "grad_norm": 0.284976601600647,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 7330
    },
    {
      "epoch": 6.49689991142604,
      "grad_norm": 0.39693188667297363,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 7335
    },
    {
      "epoch": 6.50132860938884,
      "grad_norm": 0.29884403944015503,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 7340
    },
    {
      "epoch": 6.505757307351638,
      "grad_norm": 0.3038507401943207,
      "learning_rate": 1e-05,
      "loss": 1.032,
      "step": 7345
    },
    {
      "epoch": 6.510186005314438,
      "grad_norm": 0.39658111333847046,
      "learning_rate": 1e-05,
      "loss": 1.0629,
      "step": 7350
    },
    {
      "epoch": 6.514614703277236,
      "grad_norm": 0.3359249234199524,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 7355
    },
    {
      "epoch": 6.519043401240036,
      "grad_norm": 0.42337730526924133,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 7360
    },
    {
      "epoch": 6.523472099202834,
      "grad_norm": 0.41874006390571594,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 7365
    },
    {
      "epoch": 6.527900797165634,
      "grad_norm": 0.3224058449268341,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 7370
    },
    {
      "epoch": 6.532329495128432,
      "grad_norm": 0.4191104471683502,
      "learning_rate": 1e-05,
      "loss": 1.0749,
      "step": 7375
    },
    {
      "epoch": 6.536758193091231,
      "grad_norm": 0.4798573851585388,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 7380
    },
    {
      "epoch": 6.54118689105403,
      "grad_norm": 0.41390278935432434,
      "learning_rate": 1e-05,
      "loss": 1.0226,
      "step": 7385
    },
    {
      "epoch": 6.545615589016829,
      "grad_norm": 0.41620805859565735,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 7390
    },
    {
      "epoch": 6.550044286979628,
      "grad_norm": 0.37604033946990967,
      "learning_rate": 1e-05,
      "loss": 1.0195,
      "step": 7395
    },
    {
      "epoch": 6.554472984942427,
      "grad_norm": 0.28521087765693665,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 7400
    },
    {
      "epoch": 6.558901682905226,
      "grad_norm": 0.38479727506637573,
      "learning_rate": 1e-05,
      "loss": 1.0183,
      "step": 7405
    },
    {
      "epoch": 6.563330380868025,
      "grad_norm": 0.3014092743396759,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 7410
    },
    {
      "epoch": 6.567759078830823,
      "grad_norm": 0.35304468870162964,
      "learning_rate": 1e-05,
      "loss": 1.0267,
      "step": 7415
    },
    {
      "epoch": 6.572187776793623,
      "grad_norm": 0.2886923551559448,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 7420
    },
    {
      "epoch": 6.576616474756421,
      "grad_norm": 0.3872258961200714,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 7425
    },
    {
      "epoch": 6.581045172719221,
      "grad_norm": 0.33828261494636536,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 7430
    },
    {
      "epoch": 6.585473870682019,
      "grad_norm": 0.2929351031780243,
      "learning_rate": 1e-05,
      "loss": 1.0349,
      "step": 7435
    },
    {
      "epoch": 6.589902568644819,
      "grad_norm": 0.2872735559940338,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 7440
    },
    {
      "epoch": 6.594331266607617,
      "grad_norm": 0.30917489528656006,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 7445
    },
    {
      "epoch": 6.598759964570416,
      "grad_norm": 0.3415820598602295,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 7450
    },
    {
      "epoch": 6.603188662533215,
      "grad_norm": 0.29959750175476074,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 7455
    },
    {
      "epoch": 6.607617360496015,
      "grad_norm": 0.47754573822021484,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 7460
    },
    {
      "epoch": 6.612046058458813,
      "grad_norm": 0.2885790765285492,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 7465
    },
    {
      "epoch": 6.616474756421612,
      "grad_norm": 0.30503877997398376,
      "learning_rate": 1e-05,
      "loss": 1.0532,
      "step": 7470
    },
    {
      "epoch": 6.620903454384411,
      "grad_norm": 0.35308367013931274,
      "learning_rate": 1e-05,
      "loss": 1.0488,
      "step": 7475
    },
    {
      "epoch": 6.62533215234721,
      "grad_norm": 0.3534207344055176,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 7480
    },
    {
      "epoch": 6.629760850310009,
      "grad_norm": 0.379452645778656,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 7485
    },
    {
      "epoch": 6.634189548272808,
      "grad_norm": 0.31310799717903137,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 7490
    },
    {
      "epoch": 6.638618246235607,
      "grad_norm": 0.3678398132324219,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 7495
    },
    {
      "epoch": 6.643046944198406,
      "grad_norm": 0.4244448244571686,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 7500
    },
    {
      "epoch": 6.647475642161204,
      "grad_norm": 0.3662121891975403,
      "learning_rate": 1e-05,
      "loss": 1.0648,
      "step": 7505
    },
    {
      "epoch": 6.651904340124004,
      "grad_norm": 0.3143360912799835,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 7510
    },
    {
      "epoch": 6.656333038086802,
      "grad_norm": 0.3744223713874817,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 7515
    },
    {
      "epoch": 6.660761736049602,
      "grad_norm": 0.3605046272277832,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 7520
    },
    {
      "epoch": 6.6651904340124,
      "grad_norm": 0.28498315811157227,
      "learning_rate": 1e-05,
      "loss": 1.0855,
      "step": 7525
    },
    {
      "epoch": 6.6696191319752,
      "grad_norm": 0.3095240592956543,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 7530
    },
    {
      "epoch": 6.674047829937998,
      "grad_norm": 0.3136258125305176,
      "learning_rate": 1e-05,
      "loss": 1.0287,
      "step": 7535
    },
    {
      "epoch": 6.678476527900797,
      "grad_norm": 0.317767471075058,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 7540
    },
    {
      "epoch": 6.682905225863596,
      "grad_norm": 0.39734598994255066,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 7545
    },
    {
      "epoch": 6.687333923826395,
      "grad_norm": 0.36481449007987976,
      "learning_rate": 1e-05,
      "loss": 1.0662,
      "step": 7550
    },
    {
      "epoch": 6.691762621789194,
      "grad_norm": 0.261600524187088,
      "learning_rate": 1e-05,
      "loss": 1.0522,
      "step": 7555
    },
    {
      "epoch": 6.696191319751993,
      "grad_norm": 0.27716726064682007,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 7560
    },
    {
      "epoch": 6.700620017714792,
      "grad_norm": 0.25577282905578613,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 7565
    },
    {
      "epoch": 6.705048715677591,
      "grad_norm": 0.34284737706184387,
      "learning_rate": 1e-05,
      "loss": 1.0181,
      "step": 7570
    },
    {
      "epoch": 6.70947741364039,
      "grad_norm": 0.41684550046920776,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 7575
    },
    {
      "epoch": 6.713906111603189,
      "grad_norm": 0.257420152425766,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 7580
    },
    {
      "epoch": 6.718334809565988,
      "grad_norm": 0.32291096448898315,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 7585
    },
    {
      "epoch": 6.722763507528787,
      "grad_norm": 0.3305424451828003,
      "learning_rate": 1e-05,
      "loss": 1.0186,
      "step": 7590
    },
    {
      "epoch": 6.727192205491585,
      "grad_norm": 0.3456735908985138,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 7595
    },
    {
      "epoch": 6.731620903454385,
      "grad_norm": 0.33171603083610535,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 7600
    },
    {
      "epoch": 6.736049601417183,
      "grad_norm": 0.313313364982605,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 7605
    },
    {
      "epoch": 6.7404782993799826,
      "grad_norm": 0.3528946340084076,
      "learning_rate": 1e-05,
      "loss": 1.0641,
      "step": 7610
    },
    {
      "epoch": 6.744906997342781,
      "grad_norm": 0.36428534984588623,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 7615
    },
    {
      "epoch": 6.7493356953055805,
      "grad_norm": 0.36096686124801636,
      "learning_rate": 1e-05,
      "loss": 1.0435,
      "step": 7620
    },
    {
      "epoch": 6.753764393268379,
      "grad_norm": 0.37182632088661194,
      "learning_rate": 1e-05,
      "loss": 1.0101,
      "step": 7625
    },
    {
      "epoch": 6.758193091231178,
      "grad_norm": 0.3743254244327545,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 7630
    },
    {
      "epoch": 6.762621789193977,
      "grad_norm": 0.3137091398239136,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 7635
    },
    {
      "epoch": 6.767050487156776,
      "grad_norm": 0.3580295741558075,
      "learning_rate": 1e-05,
      "loss": 1.061,
      "step": 7640
    },
    {
      "epoch": 6.771479185119575,
      "grad_norm": 0.27975258231163025,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 7645
    },
    {
      "epoch": 6.775907883082374,
      "grad_norm": 0.29722923040390015,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 7650
    },
    {
      "epoch": 6.780336581045173,
      "grad_norm": 0.3239089846611023,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 7655
    },
    {
      "epoch": 6.7847652790079716,
      "grad_norm": 0.3350318670272827,
      "learning_rate": 1e-05,
      "loss": 1.0444,
      "step": 7660
    },
    {
      "epoch": 6.78919397697077,
      "grad_norm": 0.31723690032958984,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 7665
    },
    {
      "epoch": 6.7936226749335695,
      "grad_norm": 0.333515465259552,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 7670
    },
    {
      "epoch": 6.798051372896369,
      "grad_norm": 0.33454298973083496,
      "learning_rate": 1e-05,
      "loss": 1.0431,
      "step": 7675
    },
    {
      "epoch": 6.8024800708591675,
      "grad_norm": 0.338902086019516,
      "learning_rate": 1e-05,
      "loss": 1.0608,
      "step": 7680
    },
    {
      "epoch": 6.806908768821966,
      "grad_norm": 0.28785985708236694,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 7685
    },
    {
      "epoch": 6.8113374667847655,
      "grad_norm": 0.31911033391952515,
      "learning_rate": 1e-05,
      "loss": 1.0382,
      "step": 7690
    },
    {
      "epoch": 6.815766164747564,
      "grad_norm": 0.31498414278030396,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 7695
    },
    {
      "epoch": 6.8201948627103635,
      "grad_norm": 0.34170210361480713,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 7700
    },
    {
      "epoch": 6.824623560673162,
      "grad_norm": 0.3100164830684662,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 7705
    },
    {
      "epoch": 6.829052258635961,
      "grad_norm": 0.32382506132125854,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 7710
    },
    {
      "epoch": 6.83348095659876,
      "grad_norm": 0.3402826488018036,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 7715
    },
    {
      "epoch": 6.8379096545615585,
      "grad_norm": 0.3127281665802002,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 7720
    },
    {
      "epoch": 6.842338352524358,
      "grad_norm": 0.2640603482723236,
      "learning_rate": 1e-05,
      "loss": 1.0534,
      "step": 7725
    },
    {
      "epoch": 6.8467670504871565,
      "grad_norm": 0.2885826528072357,
      "learning_rate": 1e-05,
      "loss": 1.0262,
      "step": 7730
    },
    {
      "epoch": 6.851195748449956,
      "grad_norm": 0.3311317563056946,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 7735
    },
    {
      "epoch": 6.8556244464127545,
      "grad_norm": 0.3045313358306885,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 7740
    },
    {
      "epoch": 6.860053144375554,
      "grad_norm": 0.28137636184692383,
      "learning_rate": 1e-05,
      "loss": 1.0326,
      "step": 7745
    },
    {
      "epoch": 6.8644818423383525,
      "grad_norm": 0.30333489179611206,
      "learning_rate": 1e-05,
      "loss": 1.0267,
      "step": 7750
    },
    {
      "epoch": 6.868910540301151,
      "grad_norm": 0.2819533050060272,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 7755
    },
    {
      "epoch": 6.87333923826395,
      "grad_norm": 0.3045574426651001,
      "learning_rate": 1e-05,
      "loss": 1.0149,
      "step": 7760
    },
    {
      "epoch": 6.877767936226749,
      "grad_norm": 0.3403189480304718,
      "learning_rate": 1e-05,
      "loss": 1.0472,
      "step": 7765
    },
    {
      "epoch": 6.882196634189548,
      "grad_norm": 0.3321974277496338,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 7770
    },
    {
      "epoch": 6.886625332152347,
      "grad_norm": 0.3050408661365509,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 7775
    },
    {
      "epoch": 6.891054030115146,
      "grad_norm": 0.2941606044769287,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 7780
    },
    {
      "epoch": 6.895482728077945,
      "grad_norm": 0.27922290563583374,
      "learning_rate": 1e-05,
      "loss": 1.048,
      "step": 7785
    },
    {
      "epoch": 6.899911426040744,
      "grad_norm": 0.3234655559062958,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 7790
    },
    {
      "epoch": 6.904340124003543,
      "grad_norm": 0.3068198263645172,
      "learning_rate": 1e-05,
      "loss": 0.9392,
      "step": 7795
    },
    {
      "epoch": 6.908768821966342,
      "grad_norm": 0.4504777789115906,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 7800
    },
    {
      "epoch": 6.913197519929141,
      "grad_norm": 0.3687359094619751,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 7805
    },
    {
      "epoch": 6.917626217891939,
      "grad_norm": 0.34978917241096497,
      "learning_rate": 1e-05,
      "loss": 1.1029,
      "step": 7810
    },
    {
      "epoch": 6.922054915854739,
      "grad_norm": 0.3262089490890503,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 7815
    },
    {
      "epoch": 6.926483613817537,
      "grad_norm": 0.32655099034309387,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 7820
    },
    {
      "epoch": 6.930912311780337,
      "grad_norm": 0.3247694969177246,
      "learning_rate": 1e-05,
      "loss": 1.0111,
      "step": 7825
    },
    {
      "epoch": 6.935341009743135,
      "grad_norm": 0.2828722894191742,
      "learning_rate": 1e-05,
      "loss": 1.0642,
      "step": 7830
    },
    {
      "epoch": 6.939769707705935,
      "grad_norm": 0.30532127618789673,
      "learning_rate": 1e-05,
      "loss": 1.024,
      "step": 7835
    },
    {
      "epoch": 6.944198405668733,
      "grad_norm": 0.36493581533432007,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 7840
    },
    {
      "epoch": 6.948627103631532,
      "grad_norm": 0.272352397441864,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 7845
    },
    {
      "epoch": 6.953055801594331,
      "grad_norm": 0.39458003640174866,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 7850
    },
    {
      "epoch": 6.95748449955713,
      "grad_norm": 0.3530968129634857,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 7855
    },
    {
      "epoch": 6.961913197519929,
      "grad_norm": 0.3298160135746002,
      "learning_rate": 1e-05,
      "loss": 1.0299,
      "step": 7860
    },
    {
      "epoch": 6.966341895482728,
      "grad_norm": 0.2778833210468292,
      "learning_rate": 1e-05,
      "loss": 1.0361,
      "step": 7865
    },
    {
      "epoch": 6.970770593445527,
      "grad_norm": 0.26152563095092773,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 7870
    },
    {
      "epoch": 6.975199291408326,
      "grad_norm": 0.31589120626449585,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 7875
    },
    {
      "epoch": 6.979627989371125,
      "grad_norm": 0.30574697256088257,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 7880
    },
    {
      "epoch": 6.984056687333924,
      "grad_norm": 0.32277002930641174,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 7885
    },
    {
      "epoch": 6.988485385296723,
      "grad_norm": 0.28040793538093567,
      "learning_rate": 1e-05,
      "loss": 1.0278,
      "step": 7890
    },
    {
      "epoch": 6.992914083259522,
      "grad_norm": 0.3733680546283722,
      "learning_rate": 1e-05,
      "loss": 1.034,
      "step": 7895
    },
    {
      "epoch": 6.99734278122232,
      "grad_norm": 0.2930971086025238,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 7900
    },
    {
      "epoch": 7.00177147918512,
      "grad_norm": 0.31791961193084717,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 7905
    },
    {
      "epoch": 7.006200177147918,
      "grad_norm": 0.3523207902908325,
      "learning_rate": 1e-05,
      "loss": 1.0546,
      "step": 7910
    },
    {
      "epoch": 7.010628875110718,
      "grad_norm": 0.38635241985321045,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 7915
    },
    {
      "epoch": 7.015057573073516,
      "grad_norm": 0.31179079413414,
      "learning_rate": 1e-05,
      "loss": 1.041,
      "step": 7920
    },
    {
      "epoch": 7.019486271036316,
      "grad_norm": 0.3547854423522949,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 7925
    },
    {
      "epoch": 7.023914968999114,
      "grad_norm": 0.32669755816459656,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 7930
    },
    {
      "epoch": 7.028343666961913,
      "grad_norm": 0.45759037137031555,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 7935
    },
    {
      "epoch": 7.032772364924712,
      "grad_norm": 0.329446405172348,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 7940
    },
    {
      "epoch": 7.037201062887511,
      "grad_norm": 0.38219141960144043,
      "learning_rate": 1e-05,
      "loss": 1.0455,
      "step": 7945
    },
    {
      "epoch": 7.04162976085031,
      "grad_norm": 0.3277440369129181,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 7950
    },
    {
      "epoch": 7.046058458813109,
      "grad_norm": 0.44572392106056213,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 7955
    },
    {
      "epoch": 7.050487156775908,
      "grad_norm": 0.31423547863960266,
      "learning_rate": 1e-05,
      "loss": 1.0217,
      "step": 7960
    },
    {
      "epoch": 7.054915854738707,
      "grad_norm": 0.40082406997680664,
      "learning_rate": 1e-05,
      "loss": 1.0375,
      "step": 7965
    },
    {
      "epoch": 7.059344552701506,
      "grad_norm": 0.34407326579093933,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 7970
    },
    {
      "epoch": 7.063773250664305,
      "grad_norm": 0.33923476934432983,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 7975
    },
    {
      "epoch": 7.068201948627103,
      "grad_norm": 0.27645084261894226,
      "learning_rate": 1e-05,
      "loss": 1.0362,
      "step": 7980
    },
    {
      "epoch": 7.072630646589903,
      "grad_norm": 0.3475607931613922,
      "learning_rate": 1e-05,
      "loss": 1.0956,
      "step": 7985
    },
    {
      "epoch": 7.077059344552701,
      "grad_norm": 0.3091600835323334,
      "learning_rate": 1e-05,
      "loss": 1.0392,
      "step": 7990
    },
    {
      "epoch": 7.081488042515501,
      "grad_norm": 0.3222597539424896,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 7995
    },
    {
      "epoch": 7.085916740478299,
      "grad_norm": 0.30701637268066406,
      "learning_rate": 1e-05,
      "loss": 1.0434,
      "step": 8000
    },
    {
      "epoch": 7.090345438441099,
      "grad_norm": 0.3054087162017822,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 8005
    },
    {
      "epoch": 7.094774136403897,
      "grad_norm": 0.3620450794696808,
      "learning_rate": 1e-05,
      "loss": 1.0468,
      "step": 8010
    },
    {
      "epoch": 7.099202834366697,
      "grad_norm": 0.34009912610054016,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 8015
    },
    {
      "epoch": 7.103631532329495,
      "grad_norm": 0.2626882493495941,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 8020
    },
    {
      "epoch": 7.108060230292294,
      "grad_norm": 0.2886578440666199,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 8025
    },
    {
      "epoch": 7.112488928255093,
      "grad_norm": 0.3738825023174286,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 8030
    },
    {
      "epoch": 7.116917626217892,
      "grad_norm": 0.31492140889167786,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 8035
    },
    {
      "epoch": 7.121346324180691,
      "grad_norm": 0.32313334941864014,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 8040
    },
    {
      "epoch": 7.12577502214349,
      "grad_norm": 0.2910026013851166,
      "learning_rate": 1e-05,
      "loss": 1.04,
      "step": 8045
    },
    {
      "epoch": 7.130203720106289,
      "grad_norm": 0.3253958225250244,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 8050
    },
    {
      "epoch": 7.134632418069088,
      "grad_norm": 0.3761049211025238,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 8055
    },
    {
      "epoch": 7.139061116031886,
      "grad_norm": 0.3223036825656891,
      "learning_rate": 1e-05,
      "loss": 1.0268,
      "step": 8060
    },
    {
      "epoch": 7.143489813994686,
      "grad_norm": 0.3698316514492035,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 8065
    },
    {
      "epoch": 7.147918511957484,
      "grad_norm": 0.28210216760635376,
      "learning_rate": 1e-05,
      "loss": 1.108,
      "step": 8070
    },
    {
      "epoch": 7.152347209920284,
      "grad_norm": 0.2705000340938568,
      "learning_rate": 1e-05,
      "loss": 1.0323,
      "step": 8075
    },
    {
      "epoch": 7.156775907883082,
      "grad_norm": 0.31756705045700073,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 8080
    },
    {
      "epoch": 7.161204605845882,
      "grad_norm": 0.34153518080711365,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 8085
    },
    {
      "epoch": 7.16563330380868,
      "grad_norm": 0.2473386526107788,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 8090
    },
    {
      "epoch": 7.17006200177148,
      "grad_norm": 0.35500189661979675,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 8095
    },
    {
      "epoch": 7.174490699734278,
      "grad_norm": 0.33213868737220764,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 8100
    },
    {
      "epoch": 7.178919397697077,
      "grad_norm": 0.34023064374923706,
      "learning_rate": 1e-05,
      "loss": 1.0471,
      "step": 8105
    },
    {
      "epoch": 7.183348095659876,
      "grad_norm": 0.3616093695163727,
      "learning_rate": 1e-05,
      "loss": 1.0574,
      "step": 8110
    },
    {
      "epoch": 7.187776793622675,
      "grad_norm": 0.32585567235946655,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 8115
    },
    {
      "epoch": 7.192205491585474,
      "grad_norm": 0.3519092798233032,
      "learning_rate": 1e-05,
      "loss": 1.0766,
      "step": 8120
    },
    {
      "epoch": 7.196634189548273,
      "grad_norm": 0.4222314655780792,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 8125
    },
    {
      "epoch": 7.201062887511072,
      "grad_norm": 0.3706689774990082,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 8130
    },
    {
      "epoch": 7.205491585473871,
      "grad_norm": 0.31903958320617676,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 8135
    },
    {
      "epoch": 7.20992028343667,
      "grad_norm": 0.30987489223480225,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 8140
    },
    {
      "epoch": 7.214348981399469,
      "grad_norm": 0.32475340366363525,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 8145
    },
    {
      "epoch": 7.218777679362267,
      "grad_norm": 0.3950023651123047,
      "learning_rate": 1e-05,
      "loss": 1.0435,
      "step": 8150
    },
    {
      "epoch": 7.2232063773250665,
      "grad_norm": 0.31994789838790894,
      "learning_rate": 1e-05,
      "loss": 1.0335,
      "step": 8155
    },
    {
      "epoch": 7.227635075287865,
      "grad_norm": 0.35178259015083313,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 8160
    },
    {
      "epoch": 7.2320637732506645,
      "grad_norm": 0.2776545584201813,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 8165
    },
    {
      "epoch": 7.236492471213463,
      "grad_norm": 0.35874345898628235,
      "learning_rate": 1e-05,
      "loss": 1.0581,
      "step": 8170
    },
    {
      "epoch": 7.2409211691762625,
      "grad_norm": 0.40755918622016907,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 8175
    },
    {
      "epoch": 7.245349867139061,
      "grad_norm": 0.30432066321372986,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 8180
    },
    {
      "epoch": 7.2497785651018605,
      "grad_norm": 0.2718200385570526,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 8185
    },
    {
      "epoch": 7.254207263064659,
      "grad_norm": 0.35999277234077454,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 8190
    },
    {
      "epoch": 7.258635961027458,
      "grad_norm": 0.33571064472198486,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 8195
    },
    {
      "epoch": 7.263064658990257,
      "grad_norm": 0.41914239525794983,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 8200
    },
    {
      "epoch": 7.2674933569530555,
      "grad_norm": 0.32580021023750305,
      "learning_rate": 1e-05,
      "loss": 1.0225,
      "step": 8205
    },
    {
      "epoch": 7.271922054915855,
      "grad_norm": 0.3584657311439514,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 8210
    },
    {
      "epoch": 7.2763507528786535,
      "grad_norm": 0.3185144364833832,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 8215
    },
    {
      "epoch": 7.280779450841453,
      "grad_norm": 0.322122722864151,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 8220
    },
    {
      "epoch": 7.2852081488042515,
      "grad_norm": 0.37027859687805176,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 8225
    },
    {
      "epoch": 7.289636846767051,
      "grad_norm": 0.3315444886684418,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 8230
    },
    {
      "epoch": 7.2940655447298495,
      "grad_norm": 0.312578946352005,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 8235
    },
    {
      "epoch": 7.298494242692648,
      "grad_norm": 0.318161278963089,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 8240
    },
    {
      "epoch": 7.3029229406554474,
      "grad_norm": 0.2673146426677704,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 8245
    },
    {
      "epoch": 7.307351638618246,
      "grad_norm": 0.3477378785610199,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 8250
    },
    {
      "epoch": 7.311780336581045,
      "grad_norm": 0.2590877115726471,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 8255
    },
    {
      "epoch": 7.316209034543844,
      "grad_norm": 0.34417563676834106,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 8260
    },
    {
      "epoch": 7.320637732506643,
      "grad_norm": 0.2503780126571655,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 8265
    },
    {
      "epoch": 7.325066430469442,
      "grad_norm": 0.2567213177680969,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 8270
    },
    {
      "epoch": 7.3294951284322405,
      "grad_norm": 0.3780187666416168,
      "learning_rate": 1e-05,
      "loss": 1.0241,
      "step": 8275
    },
    {
      "epoch": 7.33392382639504,
      "grad_norm": 0.34917977452278137,
      "learning_rate": 1e-05,
      "loss": 1.0263,
      "step": 8280
    },
    {
      "epoch": 7.3383525243578385,
      "grad_norm": 0.33574095368385315,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 8285
    },
    {
      "epoch": 7.342781222320638,
      "grad_norm": 0.3511805236339569,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 8290
    },
    {
      "epoch": 7.3472099202834364,
      "grad_norm": 0.3084450364112854,
      "learning_rate": 1e-05,
      "loss": 1.0196,
      "step": 8295
    },
    {
      "epoch": 7.351638618246236,
      "grad_norm": 0.38242924213409424,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 8300
    },
    {
      "epoch": 7.356067316209034,
      "grad_norm": 0.37005117535591125,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 8305
    },
    {
      "epoch": 7.360496014171834,
      "grad_norm": 0.34039926528930664,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 8310
    },
    {
      "epoch": 7.364924712134632,
      "grad_norm": 0.37982606887817383,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 8315
    },
    {
      "epoch": 7.369353410097431,
      "grad_norm": 0.37362316250801086,
      "learning_rate": 1e-05,
      "loss": 1.0159,
      "step": 8320
    },
    {
      "epoch": 7.37378210806023,
      "grad_norm": 0.28669190406799316,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 8325
    },
    {
      "epoch": 7.378210806023029,
      "grad_norm": 0.2529672384262085,
      "learning_rate": 1e-05,
      "loss": 1.0548,
      "step": 8330
    },
    {
      "epoch": 7.382639503985828,
      "grad_norm": 0.28348371386528015,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 8335
    },
    {
      "epoch": 7.387068201948627,
      "grad_norm": 0.34466010332107544,
      "learning_rate": 1e-05,
      "loss": 1.0318,
      "step": 8340
    },
    {
      "epoch": 7.391496899911426,
      "grad_norm": 0.28888288140296936,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 8345
    },
    {
      "epoch": 7.395925597874225,
      "grad_norm": 0.3542207181453705,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 8350
    },
    {
      "epoch": 7.400354295837024,
      "grad_norm": 0.2847772538661957,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 8355
    },
    {
      "epoch": 7.404782993799823,
      "grad_norm": 0.32181715965270996,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 8360
    },
    {
      "epoch": 7.409211691762621,
      "grad_norm": 0.34742075204849243,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 8365
    },
    {
      "epoch": 7.413640389725421,
      "grad_norm": 0.3222198784351349,
      "learning_rate": 1e-05,
      "loss": 1.0181,
      "step": 8370
    },
    {
      "epoch": 7.418069087688219,
      "grad_norm": 0.38159647583961487,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 8375
    },
    {
      "epoch": 7.422497785651019,
      "grad_norm": 0.25137943029403687,
      "learning_rate": 1e-05,
      "loss": 1.0101,
      "step": 8380
    },
    {
      "epoch": 7.426926483613817,
      "grad_norm": 0.24861066043376923,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 8385
    },
    {
      "epoch": 7.431355181576617,
      "grad_norm": 0.31143897771835327,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 8390
    },
    {
      "epoch": 7.435783879539415,
      "grad_norm": 0.43090397119522095,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 8395
    },
    {
      "epoch": 7.440212577502215,
      "grad_norm": 0.3416902422904968,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 8400
    },
    {
      "epoch": 7.444641275465013,
      "grad_norm": 0.3006885051727295,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 8405
    },
    {
      "epoch": 7.449069973427812,
      "grad_norm": 0.304620623588562,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 8410
    },
    {
      "epoch": 7.453498671390611,
      "grad_norm": 0.38202860951423645,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 8415
    },
    {
      "epoch": 7.45792736935341,
      "grad_norm": 0.3281434178352356,
      "learning_rate": 1e-05,
      "loss": 1.0343,
      "step": 8420
    },
    {
      "epoch": 7.462356067316209,
      "grad_norm": 0.31972700357437134,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 8425
    },
    {
      "epoch": 7.466784765279008,
      "grad_norm": 0.37000319361686707,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 8430
    },
    {
      "epoch": 7.471213463241807,
      "grad_norm": 0.27409106492996216,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 8435
    },
    {
      "epoch": 7.475642161204606,
      "grad_norm": 0.2912721335887909,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 8440
    },
    {
      "epoch": 7.480070859167405,
      "grad_norm": 0.3768261969089508,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 8445
    },
    {
      "epoch": 7.484499557130204,
      "grad_norm": 0.33346015214920044,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 8450
    },
    {
      "epoch": 7.488928255093002,
      "grad_norm": 0.34781017899513245,
      "learning_rate": 1e-05,
      "loss": 1.0218,
      "step": 8455
    },
    {
      "epoch": 7.493356953055802,
      "grad_norm": 0.27549079060554504,
      "learning_rate": 1e-05,
      "loss": 1.0436,
      "step": 8460
    },
    {
      "epoch": 7.4977856510186,
      "grad_norm": 0.33016541600227356,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 8465
    },
    {
      "epoch": 7.5022143489814,
      "grad_norm": 0.36840030550956726,
      "learning_rate": 1e-05,
      "loss": 1.0258,
      "step": 8470
    },
    {
      "epoch": 7.506643046944198,
      "grad_norm": 0.33794450759887695,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 8475
    },
    {
      "epoch": 7.511071744906998,
      "grad_norm": 0.34260109066963196,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 8480
    },
    {
      "epoch": 7.515500442869796,
      "grad_norm": 0.3097268044948578,
      "learning_rate": 1e-05,
      "loss": 1.0851,
      "step": 8485
    },
    {
      "epoch": 7.519929140832595,
      "grad_norm": 0.3689298629760742,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 8490
    },
    {
      "epoch": 7.524357838795394,
      "grad_norm": 0.30302244424819946,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 8495
    },
    {
      "epoch": 7.528786536758193,
      "grad_norm": 0.40244540572166443,
      "learning_rate": 1e-05,
      "loss": 1.0305,
      "step": 8500
    },
    {
      "epoch": 7.533215234720992,
      "grad_norm": 0.3475583791732788,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 8505
    },
    {
      "epoch": 7.537643932683791,
      "grad_norm": 0.3201814889907837,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 8510
    },
    {
      "epoch": 7.54207263064659,
      "grad_norm": 0.3686660826206207,
      "learning_rate": 1e-05,
      "loss": 1.0371,
      "step": 8515
    },
    {
      "epoch": 7.546501328609389,
      "grad_norm": 0.3371261954307556,
      "learning_rate": 1e-05,
      "loss": 1.0725,
      "step": 8520
    },
    {
      "epoch": 7.550930026572188,
      "grad_norm": 0.29111093282699585,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 8525
    },
    {
      "epoch": 7.555358724534987,
      "grad_norm": 0.2628686726093292,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 8530
    },
    {
      "epoch": 7.559787422497786,
      "grad_norm": 0.28007084131240845,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 8535
    },
    {
      "epoch": 7.564216120460585,
      "grad_norm": 0.27675244212150574,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 8540
    },
    {
      "epoch": 7.568644818423383,
      "grad_norm": 0.31111079454421997,
      "learning_rate": 1e-05,
      "loss": 1.0445,
      "step": 8545
    },
    {
      "epoch": 7.573073516386183,
      "grad_norm": 0.3076365888118744,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 8550
    },
    {
      "epoch": 7.577502214348981,
      "grad_norm": 0.2831222116947174,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 8555
    },
    {
      "epoch": 7.581930912311781,
      "grad_norm": 0.2999606132507324,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 8560
    },
    {
      "epoch": 7.586359610274579,
      "grad_norm": 0.356234073638916,
      "learning_rate": 1e-05,
      "loss": 1.034,
      "step": 8565
    },
    {
      "epoch": 7.590788308237379,
      "grad_norm": 0.4144793748855591,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 8570
    },
    {
      "epoch": 7.595217006200177,
      "grad_norm": 0.3226306140422821,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 8575
    },
    {
      "epoch": 7.599645704162976,
      "grad_norm": 0.2927561104297638,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 8580
    },
    {
      "epoch": 7.604074402125775,
      "grad_norm": 0.358500599861145,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 8585
    },
    {
      "epoch": 7.608503100088574,
      "grad_norm": 0.3640202283859253,
      "learning_rate": 1e-05,
      "loss": 1.0368,
      "step": 8590
    },
    {
      "epoch": 7.612931798051373,
      "grad_norm": 0.3261796534061432,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 8595
    },
    {
      "epoch": 7.617360496014172,
      "grad_norm": 0.3256380558013916,
      "learning_rate": 1e-05,
      "loss": 1.0681,
      "step": 8600
    },
    {
      "epoch": 7.621789193976971,
      "grad_norm": 0.329542875289917,
      "learning_rate": 1e-05,
      "loss": 1.0504,
      "step": 8605
    },
    {
      "epoch": 7.62621789193977,
      "grad_norm": 0.31438469886779785,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 8610
    },
    {
      "epoch": 7.630646589902568,
      "grad_norm": 0.24086038768291473,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 8615
    },
    {
      "epoch": 7.635075287865368,
      "grad_norm": 0.32759779691696167,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 8620
    },
    {
      "epoch": 7.639503985828166,
      "grad_norm": 0.3202802538871765,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 8625
    },
    {
      "epoch": 7.643932683790966,
      "grad_norm": 0.2695578336715698,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 8630
    },
    {
      "epoch": 7.648361381753764,
      "grad_norm": 0.25730758905410767,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 8635
    },
    {
      "epoch": 7.6527900797165636,
      "grad_norm": 0.32701051235198975,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 8640
    },
    {
      "epoch": 7.657218777679362,
      "grad_norm": 0.24104008078575134,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 8645
    },
    {
      "epoch": 7.6616474756421615,
      "grad_norm": 0.35857489705085754,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 8650
    },
    {
      "epoch": 7.66607617360496,
      "grad_norm": 0.3006949722766876,
      "learning_rate": 1e-05,
      "loss": 1.0431,
      "step": 8655
    },
    {
      "epoch": 7.6705048715677595,
      "grad_norm": 0.3171907663345337,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 8660
    },
    {
      "epoch": 7.674933569530558,
      "grad_norm": 0.3297688663005829,
      "learning_rate": 1e-05,
      "loss": 1.0297,
      "step": 8665
    },
    {
      "epoch": 7.679362267493357,
      "grad_norm": 0.3140852451324463,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 8670
    },
    {
      "epoch": 7.683790965456156,
      "grad_norm": 0.3075326979160309,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 8675
    },
    {
      "epoch": 7.688219663418955,
      "grad_norm": 0.39814651012420654,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 8680
    },
    {
      "epoch": 7.692648361381754,
      "grad_norm": 0.3181552588939667,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 8685
    },
    {
      "epoch": 7.6970770593445526,
      "grad_norm": 0.3182726800441742,
      "learning_rate": 1e-05,
      "loss": 1.0288,
      "step": 8690
    },
    {
      "epoch": 7.701505757307352,
      "grad_norm": 0.3198147714138031,
      "learning_rate": 1e-05,
      "loss": 1.0354,
      "step": 8695
    },
    {
      "epoch": 7.7059344552701505,
      "grad_norm": 0.29476943612098694,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 8700
    },
    {
      "epoch": 7.710363153232949,
      "grad_norm": 0.2936825752258301,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 8705
    },
    {
      "epoch": 7.7147918511957485,
      "grad_norm": 0.29142528772354126,
      "learning_rate": 1e-05,
      "loss": 1.0325,
      "step": 8710
    },
    {
      "epoch": 7.719220549158547,
      "grad_norm": 0.30274105072021484,
      "learning_rate": 1e-05,
      "loss": 1.0506,
      "step": 8715
    },
    {
      "epoch": 7.7236492471213465,
      "grad_norm": 0.2681039869785309,
      "learning_rate": 1e-05,
      "loss": 1.0536,
      "step": 8720
    },
    {
      "epoch": 7.728077945084145,
      "grad_norm": 0.3139670491218567,
      "learning_rate": 1e-05,
      "loss": 1.0238,
      "step": 8725
    },
    {
      "epoch": 7.7325066430469445,
      "grad_norm": 0.28641027212142944,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 8730
    },
    {
      "epoch": 7.736935341009743,
      "grad_norm": 0.31871822476387024,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 8735
    },
    {
      "epoch": 7.741364038972542,
      "grad_norm": 0.3115043342113495,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 8740
    },
    {
      "epoch": 7.745792736935341,
      "grad_norm": 0.2972046136856079,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 8745
    },
    {
      "epoch": 7.75022143489814,
      "grad_norm": 0.2684633135795593,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 8750
    },
    {
      "epoch": 7.754650132860939,
      "grad_norm": 0.363823264837265,
      "learning_rate": 1e-05,
      "loss": 1.0181,
      "step": 8755
    },
    {
      "epoch": 7.7590788308237375,
      "grad_norm": 0.387590229511261,
      "learning_rate": 1e-05,
      "loss": 1.0107,
      "step": 8760
    },
    {
      "epoch": 7.763507528786537,
      "grad_norm": 0.3175017535686493,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 8765
    },
    {
      "epoch": 7.7679362267493355,
      "grad_norm": 0.28583285212516785,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 8770
    },
    {
      "epoch": 7.772364924712135,
      "grad_norm": 0.29801133275032043,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 8775
    },
    {
      "epoch": 7.7767936226749335,
      "grad_norm": 0.34351205825805664,
      "learning_rate": 1e-05,
      "loss": 1.0459,
      "step": 8780
    },
    {
      "epoch": 7.781222320637733,
      "grad_norm": 0.3609464168548584,
      "learning_rate": 1e-05,
      "loss": 1.0085,
      "step": 8785
    },
    {
      "epoch": 7.785651018600531,
      "grad_norm": 0.33855128288269043,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 8790
    },
    {
      "epoch": 7.79007971656333,
      "grad_norm": 0.2705983519554138,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 8795
    },
    {
      "epoch": 7.794508414526129,
      "grad_norm": 0.2934018075466156,
      "learning_rate": 1e-05,
      "loss": 1.0837,
      "step": 8800
    },
    {
      "epoch": 7.798937112488928,
      "grad_norm": 0.3145749866962433,
      "learning_rate": 1e-05,
      "loss": 1.0841,
      "step": 8805
    },
    {
      "epoch": 7.803365810451727,
      "grad_norm": 0.3640284836292267,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 8810
    },
    {
      "epoch": 7.807794508414526,
      "grad_norm": 0.3094344437122345,
      "learning_rate": 1e-05,
      "loss": 1.0228,
      "step": 8815
    },
    {
      "epoch": 7.812223206377325,
      "grad_norm": 0.39733627438545227,
      "learning_rate": 1e-05,
      "loss": 1.0396,
      "step": 8820
    },
    {
      "epoch": 7.816651904340124,
      "grad_norm": 0.3264067471027374,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 8825
    },
    {
      "epoch": 7.8210806023029225,
      "grad_norm": 0.40689459443092346,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 8830
    },
    {
      "epoch": 7.825509300265722,
      "grad_norm": 0.29561591148376465,
      "learning_rate": 1e-05,
      "loss": 1.0402,
      "step": 8835
    },
    {
      "epoch": 7.82993799822852,
      "grad_norm": 0.3005363941192627,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 8840
    },
    {
      "epoch": 7.83436669619132,
      "grad_norm": 0.3895195424556732,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 8845
    },
    {
      "epoch": 7.838795394154118,
      "grad_norm": 0.28007766604423523,
      "learning_rate": 1e-05,
      "loss": 1.0308,
      "step": 8850
    },
    {
      "epoch": 7.843224092116918,
      "grad_norm": 0.32203447818756104,
      "learning_rate": 1e-05,
      "loss": 1.0309,
      "step": 8855
    },
    {
      "epoch": 7.847652790079716,
      "grad_norm": 0.3152931034564972,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 8860
    },
    {
      "epoch": 7.852081488042516,
      "grad_norm": 0.41865599155426025,
      "learning_rate": 1e-05,
      "loss": 1.0489,
      "step": 8865
    },
    {
      "epoch": 7.856510186005314,
      "grad_norm": 0.30333995819091797,
      "learning_rate": 1e-05,
      "loss": 1.0491,
      "step": 8870
    },
    {
      "epoch": 7.860938883968114,
      "grad_norm": 0.2809825539588928,
      "learning_rate": 1e-05,
      "loss": 1.0654,
      "step": 8875
    },
    {
      "epoch": 7.865367581930912,
      "grad_norm": 0.2574847340583801,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 8880
    },
    {
      "epoch": 7.869796279893711,
      "grad_norm": 0.2967134416103363,
      "learning_rate": 1e-05,
      "loss": 1.0506,
      "step": 8885
    },
    {
      "epoch": 7.87422497785651,
      "grad_norm": 0.303195059299469,
      "learning_rate": 1e-05,
      "loss": 1.0594,
      "step": 8890
    },
    {
      "epoch": 7.878653675819309,
      "grad_norm": 0.30553334951400757,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 8895
    },
    {
      "epoch": 7.883082373782108,
      "grad_norm": 0.46372443437576294,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 8900
    },
    {
      "epoch": 7.887511071744907,
      "grad_norm": 0.34033286571502686,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 8905
    },
    {
      "epoch": 7.891939769707706,
      "grad_norm": 0.32112789154052734,
      "learning_rate": 1e-05,
      "loss": 1.0511,
      "step": 8910
    },
    {
      "epoch": 7.896368467670505,
      "grad_norm": 0.2972681522369385,
      "learning_rate": 1e-05,
      "loss": 1.0184,
      "step": 8915
    },
    {
      "epoch": 7.900797165633303,
      "grad_norm": 0.3520476520061493,
      "learning_rate": 1e-05,
      "loss": 1.0683,
      "step": 8920
    },
    {
      "epoch": 7.905225863596103,
      "grad_norm": 0.2974855601787567,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 8925
    },
    {
      "epoch": 7.909654561558901,
      "grad_norm": 0.29914185404777527,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 8930
    },
    {
      "epoch": 7.914083259521701,
      "grad_norm": 0.3232276439666748,
      "learning_rate": 1e-05,
      "loss": 1.0336,
      "step": 8935
    },
    {
      "epoch": 7.918511957484499,
      "grad_norm": 0.3576143682003021,
      "learning_rate": 1e-05,
      "loss": 1.0301,
      "step": 8940
    },
    {
      "epoch": 7.922940655447299,
      "grad_norm": 0.34765493869781494,
      "learning_rate": 1e-05,
      "loss": 1.0382,
      "step": 8945
    },
    {
      "epoch": 7.927369353410097,
      "grad_norm": 0.3478412926197052,
      "learning_rate": 1e-05,
      "loss": 1.0287,
      "step": 8950
    },
    {
      "epoch": 7.931798051372897,
      "grad_norm": 0.33409714698791504,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 8955
    },
    {
      "epoch": 7.936226749335695,
      "grad_norm": 0.282105416059494,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 8960
    },
    {
      "epoch": 7.940655447298495,
      "grad_norm": 0.35349369049072266,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 8965
    },
    {
      "epoch": 7.945084145261293,
      "grad_norm": 0.27753743529319763,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 8970
    },
    {
      "epoch": 7.949512843224092,
      "grad_norm": 0.34601280093193054,
      "learning_rate": 1e-05,
      "loss": 1.054,
      "step": 8975
    },
    {
      "epoch": 7.953941541186891,
      "grad_norm": 0.35734254121780396,
      "learning_rate": 1e-05,
      "loss": 1.0565,
      "step": 8980
    },
    {
      "epoch": 7.95837023914969,
      "grad_norm": 0.3318418860435486,
      "learning_rate": 1e-05,
      "loss": 1.025,
      "step": 8985
    },
    {
      "epoch": 7.962798937112489,
      "grad_norm": 0.33603084087371826,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 8990
    },
    {
      "epoch": 7.967227635075288,
      "grad_norm": 0.2874302864074707,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 8995
    },
    {
      "epoch": 7.971656333038087,
      "grad_norm": 0.3076196312904358,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 9000
    },
    {
      "epoch": 7.976085031000886,
      "grad_norm": 0.3234718441963196,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 9005
    },
    {
      "epoch": 7.980513728963684,
      "grad_norm": 0.25036486983299255,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 9010
    },
    {
      "epoch": 7.984942426926484,
      "grad_norm": 0.36438146233558655,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 9015
    },
    {
      "epoch": 7.989371124889282,
      "grad_norm": 0.2823215425014496,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 9020
    },
    {
      "epoch": 7.993799822852082,
      "grad_norm": 0.32932335138320923,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 9025
    },
    {
      "epoch": 7.99822852081488,
      "grad_norm": 0.30196577310562134,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 9030
    },
    {
      "epoch": 8.00265721877768,
      "grad_norm": 0.3501168489456177,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 9035
    },
    {
      "epoch": 8.007085916740479,
      "grad_norm": 0.2672116458415985,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 9040
    },
    {
      "epoch": 8.011514614703277,
      "grad_norm": 0.3172587454319,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 9045
    },
    {
      "epoch": 8.015943312666076,
      "grad_norm": 0.33237701654434204,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 9050
    },
    {
      "epoch": 8.020372010628876,
      "grad_norm": 0.3432263731956482,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 9055
    },
    {
      "epoch": 8.024800708591673,
      "grad_norm": 0.36441463232040405,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 9060
    },
    {
      "epoch": 8.029229406554473,
      "grad_norm": 0.4518466889858246,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 9065
    },
    {
      "epoch": 8.033658104517272,
      "grad_norm": 0.40251004695892334,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 9070
    },
    {
      "epoch": 8.038086802480072,
      "grad_norm": 0.3918655514717102,
      "learning_rate": 1e-05,
      "loss": 1.0409,
      "step": 9075
    },
    {
      "epoch": 8.04251550044287,
      "grad_norm": 0.36182475090026855,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 9080
    },
    {
      "epoch": 8.046944198405669,
      "grad_norm": 0.3470155596733093,
      "learning_rate": 1e-05,
      "loss": 1.0252,
      "step": 9085
    },
    {
      "epoch": 8.051372896368468,
      "grad_norm": 0.4093534052371979,
      "learning_rate": 1e-05,
      "loss": 1.0261,
      "step": 9090
    },
    {
      "epoch": 8.055801594331266,
      "grad_norm": 0.32986247539520264,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 9095
    },
    {
      "epoch": 8.060230292294065,
      "grad_norm": 0.32325616478919983,
      "learning_rate": 1e-05,
      "loss": 1.0691,
      "step": 9100
    },
    {
      "epoch": 8.064658990256865,
      "grad_norm": 0.30846190452575684,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 9105
    },
    {
      "epoch": 8.069087688219664,
      "grad_norm": 0.3124985098838806,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 9110
    },
    {
      "epoch": 8.073516386182462,
      "grad_norm": 0.29013264179229736,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 9115
    },
    {
      "epoch": 8.077945084145261,
      "grad_norm": 0.3013417422771454,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 9120
    },
    {
      "epoch": 8.08237378210806,
      "grad_norm": 0.2700832486152649,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 9125
    },
    {
      "epoch": 8.08680248007086,
      "grad_norm": 0.30386435985565186,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 9130
    },
    {
      "epoch": 8.091231178033658,
      "grad_norm": 0.4277620315551758,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 9135
    },
    {
      "epoch": 8.095659875996457,
      "grad_norm": 0.32337477803230286,
      "learning_rate": 1e-05,
      "loss": 1.0262,
      "step": 9140
    },
    {
      "epoch": 8.100088573959257,
      "grad_norm": 0.2845984995365143,
      "learning_rate": 1e-05,
      "loss": 1.0359,
      "step": 9145
    },
    {
      "epoch": 8.104517271922054,
      "grad_norm": 0.347700834274292,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 9150
    },
    {
      "epoch": 8.108945969884854,
      "grad_norm": 0.3128695785999298,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 9155
    },
    {
      "epoch": 8.113374667847653,
      "grad_norm": 0.3797442317008972,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 9160
    },
    {
      "epoch": 8.117803365810452,
      "grad_norm": 0.35018256306648254,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 9165
    },
    {
      "epoch": 8.12223206377325,
      "grad_norm": 0.3066740334033966,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 9170
    },
    {
      "epoch": 8.12666076173605,
      "grad_norm": 0.35448938608169556,
      "learning_rate": 1e-05,
      "loss": 1.0447,
      "step": 9175
    },
    {
      "epoch": 8.131089459698849,
      "grad_norm": 0.27987968921661377,
      "learning_rate": 1e-05,
      "loss": 1.0344,
      "step": 9180
    },
    {
      "epoch": 8.135518157661647,
      "grad_norm": 0.3672999441623688,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 9185
    },
    {
      "epoch": 8.139946855624446,
      "grad_norm": 0.2840280830860138,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 9190
    },
    {
      "epoch": 8.144375553587246,
      "grad_norm": 0.3593059182167053,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 9195
    },
    {
      "epoch": 8.148804251550045,
      "grad_norm": 0.3591805398464203,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 9200
    },
    {
      "epoch": 8.153232949512843,
      "grad_norm": 0.36199432611465454,
      "learning_rate": 1e-05,
      "loss": 1.0571,
      "step": 9205
    },
    {
      "epoch": 8.157661647475642,
      "grad_norm": 0.31503650546073914,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 9210
    },
    {
      "epoch": 8.162090345438441,
      "grad_norm": 0.4618072509765625,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 9215
    },
    {
      "epoch": 8.166519043401241,
      "grad_norm": 0.3053239583969116,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 9220
    },
    {
      "epoch": 8.170947741364039,
      "grad_norm": 0.3260848820209503,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 9225
    },
    {
      "epoch": 8.175376439326838,
      "grad_norm": 0.2825438976287842,
      "learning_rate": 1e-05,
      "loss": 1.0181,
      "step": 9230
    },
    {
      "epoch": 8.179805137289637,
      "grad_norm": 0.3291209936141968,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 9235
    },
    {
      "epoch": 8.184233835252435,
      "grad_norm": 0.39499205350875854,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 9240
    },
    {
      "epoch": 8.188662533215235,
      "grad_norm": 0.34457486867904663,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 9245
    },
    {
      "epoch": 8.193091231178034,
      "grad_norm": 0.3637145161628723,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 9250
    },
    {
      "epoch": 8.197519929140833,
      "grad_norm": 0.34986281394958496,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 9255
    },
    {
      "epoch": 8.201948627103631,
      "grad_norm": 0.3071766793727875,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 9260
    },
    {
      "epoch": 8.20637732506643,
      "grad_norm": 0.3395848274230957,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 9265
    },
    {
      "epoch": 8.21080602302923,
      "grad_norm": 0.30788081884384155,
      "learning_rate": 1e-05,
      "loss": 1.0404,
      "step": 9270
    },
    {
      "epoch": 8.215234720992028,
      "grad_norm": 0.3265116512775421,
      "learning_rate": 1e-05,
      "loss": 1.0473,
      "step": 9275
    },
    {
      "epoch": 8.219663418954827,
      "grad_norm": 0.3096961975097656,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 9280
    },
    {
      "epoch": 8.224092116917626,
      "grad_norm": 0.2829352915287018,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 9285
    },
    {
      "epoch": 8.228520814880426,
      "grad_norm": 0.35683074593544006,
      "learning_rate": 1e-05,
      "loss": 1.0399,
      "step": 9290
    },
    {
      "epoch": 8.232949512843224,
      "grad_norm": 0.32789289951324463,
      "learning_rate": 1e-05,
      "loss": 1.0258,
      "step": 9295
    },
    {
      "epoch": 8.237378210806023,
      "grad_norm": 0.2792492210865021,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 9300
    },
    {
      "epoch": 8.241806908768822,
      "grad_norm": 0.26763540506362915,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 9305
    },
    {
      "epoch": 8.24623560673162,
      "grad_norm": 0.3517155349254608,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 9310
    },
    {
      "epoch": 8.25066430469442,
      "grad_norm": 0.3247127830982208,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 9315
    },
    {
      "epoch": 8.255093002657219,
      "grad_norm": 0.3012772798538208,
      "learning_rate": 1e-05,
      "loss": 1.0463,
      "step": 9320
    },
    {
      "epoch": 8.259521700620018,
      "grad_norm": 0.43280452489852905,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 9325
    },
    {
      "epoch": 8.263950398582816,
      "grad_norm": 0.4521568715572357,
      "learning_rate": 1e-05,
      "loss": 1.0111,
      "step": 9330
    },
    {
      "epoch": 8.268379096545615,
      "grad_norm": 0.2858237624168396,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 9335
    },
    {
      "epoch": 8.272807794508415,
      "grad_norm": 0.2798160910606384,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 9340
    },
    {
      "epoch": 8.277236492471214,
      "grad_norm": 0.29831013083457947,
      "learning_rate": 1e-05,
      "loss": 1.0276,
      "step": 9345
    },
    {
      "epoch": 8.281665190434012,
      "grad_norm": 0.2587404251098633,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 9350
    },
    {
      "epoch": 8.286093888396811,
      "grad_norm": 0.3005838096141815,
      "learning_rate": 1e-05,
      "loss": 1.014,
      "step": 9355
    },
    {
      "epoch": 8.29052258635961,
      "grad_norm": 0.3595026135444641,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 9360
    },
    {
      "epoch": 8.294951284322408,
      "grad_norm": 0.2960871756076813,
      "learning_rate": 1e-05,
      "loss": 1.0517,
      "step": 9365
    },
    {
      "epoch": 8.299379982285208,
      "grad_norm": 0.41662898659706116,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 9370
    },
    {
      "epoch": 8.303808680248007,
      "grad_norm": 0.29070064425468445,
      "learning_rate": 1e-05,
      "loss": 1.0183,
      "step": 9375
    },
    {
      "epoch": 8.308237378210807,
      "grad_norm": 0.24964648485183716,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 9380
    },
    {
      "epoch": 8.312666076173604,
      "grad_norm": 0.3184340298175812,
      "learning_rate": 1e-05,
      "loss": 1.0482,
      "step": 9385
    },
    {
      "epoch": 8.317094774136404,
      "grad_norm": 0.32899683713912964,
      "learning_rate": 1e-05,
      "loss": 1.0359,
      "step": 9390
    },
    {
      "epoch": 8.321523472099203,
      "grad_norm": 0.3268736004829407,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 9395
    },
    {
      "epoch": 8.325952170062001,
      "grad_norm": 0.3246442377567291,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 9400
    },
    {
      "epoch": 8.3303808680248,
      "grad_norm": 0.28695979714393616,
      "learning_rate": 1e-05,
      "loss": 1.0446,
      "step": 9405
    },
    {
      "epoch": 8.3348095659876,
      "grad_norm": 0.31747546792030334,
      "learning_rate": 1e-05,
      "loss": 1.0422,
      "step": 9410
    },
    {
      "epoch": 8.3392382639504,
      "grad_norm": 0.2943037152290344,
      "learning_rate": 1e-05,
      "loss": 1.0362,
      "step": 9415
    },
    {
      "epoch": 8.343666961913197,
      "grad_norm": 0.27293357253074646,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 9420
    },
    {
      "epoch": 8.348095659875996,
      "grad_norm": 0.320142537355423,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 9425
    },
    {
      "epoch": 8.352524357838796,
      "grad_norm": 0.3383481204509735,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 9430
    },
    {
      "epoch": 8.356953055801593,
      "grad_norm": 0.2932541072368622,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 9435
    },
    {
      "epoch": 8.361381753764393,
      "grad_norm": 0.3102038502693176,
      "learning_rate": 1e-05,
      "loss": 1.0203,
      "step": 9440
    },
    {
      "epoch": 8.365810451727192,
      "grad_norm": 0.2597452998161316,
      "learning_rate": 1e-05,
      "loss": 1.0226,
      "step": 9445
    },
    {
      "epoch": 8.370239149689992,
      "grad_norm": 0.2563137412071228,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 9450
    },
    {
      "epoch": 8.37466784765279,
      "grad_norm": 0.3111991584300995,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 9455
    },
    {
      "epoch": 8.379096545615589,
      "grad_norm": 0.3485407531261444,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 9460
    },
    {
      "epoch": 8.383525243578388,
      "grad_norm": 0.3654576539993286,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 9465
    },
    {
      "epoch": 8.387953941541188,
      "grad_norm": 0.28132757544517517,
      "learning_rate": 1e-05,
      "loss": 1.0276,
      "step": 9470
    },
    {
      "epoch": 8.392382639503985,
      "grad_norm": 0.34759706258773804,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 9475
    },
    {
      "epoch": 8.396811337466785,
      "grad_norm": 0.2717755138874054,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 9480
    },
    {
      "epoch": 8.401240035429584,
      "grad_norm": 0.3453042507171631,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 9485
    },
    {
      "epoch": 8.405668733392382,
      "grad_norm": 0.3692055642604828,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 9490
    },
    {
      "epoch": 8.410097431355181,
      "grad_norm": 0.3445853590965271,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 9495
    },
    {
      "epoch": 8.41452612931798,
      "grad_norm": 0.25680211186408997,
      "learning_rate": 1e-05,
      "loss": 1.0668,
      "step": 9500
    },
    {
      "epoch": 8.41895482728078,
      "grad_norm": 0.31134283542633057,
      "learning_rate": 1e-05,
      "loss": 1.0269,
      "step": 9505
    },
    {
      "epoch": 8.423383525243578,
      "grad_norm": 0.32206547260284424,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 9510
    },
    {
      "epoch": 8.427812223206377,
      "grad_norm": 0.3236282467842102,
      "learning_rate": 1e-05,
      "loss": 1.0338,
      "step": 9515
    },
    {
      "epoch": 8.432240921169177,
      "grad_norm": 0.3217984139919281,
      "learning_rate": 1e-05,
      "loss": 1.0422,
      "step": 9520
    },
    {
      "epoch": 8.436669619131976,
      "grad_norm": 0.2910170555114746,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 9525
    },
    {
      "epoch": 8.441098317094774,
      "grad_norm": 0.29590415954589844,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 9530
    },
    {
      "epoch": 8.445527015057573,
      "grad_norm": 0.37308189272880554,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 9535
    },
    {
      "epoch": 8.449955713020373,
      "grad_norm": 0.4636489152908325,
      "learning_rate": 1e-05,
      "loss": 1.0459,
      "step": 9540
    },
    {
      "epoch": 8.45438441098317,
      "grad_norm": 0.316521555185318,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 9545
    },
    {
      "epoch": 8.45881310894597,
      "grad_norm": 0.3285221457481384,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 9550
    },
    {
      "epoch": 8.46324180690877,
      "grad_norm": 0.34427571296691895,
      "learning_rate": 1e-05,
      "loss": 1.0855,
      "step": 9555
    },
    {
      "epoch": 8.467670504871569,
      "grad_norm": 0.34588485956192017,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 9560
    },
    {
      "epoch": 8.472099202834366,
      "grad_norm": 0.26303595304489136,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 9565
    },
    {
      "epoch": 8.476527900797166,
      "grad_norm": 0.27641695737838745,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 9570
    },
    {
      "epoch": 8.480956598759965,
      "grad_norm": 0.3545122444629669,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 9575
    },
    {
      "epoch": 8.485385296722763,
      "grad_norm": 0.24362750351428986,
      "learning_rate": 1e-05,
      "loss": 1.0584,
      "step": 9580
    },
    {
      "epoch": 8.489813994685562,
      "grad_norm": 0.36961522698402405,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 9585
    },
    {
      "epoch": 8.494242692648362,
      "grad_norm": 0.2861690819263458,
      "learning_rate": 1e-05,
      "loss": 1.0232,
      "step": 9590
    },
    {
      "epoch": 8.498671390611161,
      "grad_norm": 0.28869161009788513,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 9595
    },
    {
      "epoch": 8.503100088573959,
      "grad_norm": 0.3293612003326416,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 9600
    },
    {
      "epoch": 8.507528786536758,
      "grad_norm": 0.280514121055603,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 9605
    },
    {
      "epoch": 8.511957484499558,
      "grad_norm": 0.26847410202026367,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 9610
    },
    {
      "epoch": 8.516386182462355,
      "grad_norm": 0.2584664225578308,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 9615
    },
    {
      "epoch": 8.520814880425155,
      "grad_norm": 0.24842916429042816,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 9620
    },
    {
      "epoch": 8.525243578387954,
      "grad_norm": 0.45260944962501526,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 9625
    },
    {
      "epoch": 8.529672276350754,
      "grad_norm": 0.3274429142475128,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 9630
    },
    {
      "epoch": 8.534100974313551,
      "grad_norm": 0.2761222720146179,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 9635
    },
    {
      "epoch": 8.53852967227635,
      "grad_norm": 0.29830145835876465,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 9640
    },
    {
      "epoch": 8.54295837023915,
      "grad_norm": 0.3462134003639221,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 9645
    },
    {
      "epoch": 8.54738706820195,
      "grad_norm": 0.24739807844161987,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 9650
    },
    {
      "epoch": 8.551815766164747,
      "grad_norm": 0.2875482439994812,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 9655
    },
    {
      "epoch": 8.556244464127547,
      "grad_norm": 0.27581486105918884,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 9660
    },
    {
      "epoch": 8.560673162090346,
      "grad_norm": 0.2932671308517456,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 9665
    },
    {
      "epoch": 8.565101860053144,
      "grad_norm": 0.27838656306266785,
      "learning_rate": 1e-05,
      "loss": 1.0598,
      "step": 9670
    },
    {
      "epoch": 8.569530558015943,
      "grad_norm": 0.2761351764202118,
      "learning_rate": 1e-05,
      "loss": 1.0256,
      "step": 9675
    },
    {
      "epoch": 8.573959255978743,
      "grad_norm": 0.30724138021469116,
      "learning_rate": 1e-05,
      "loss": 1.0187,
      "step": 9680
    },
    {
      "epoch": 8.578387953941542,
      "grad_norm": 0.3189345598220825,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 9685
    },
    {
      "epoch": 8.58281665190434,
      "grad_norm": 0.3159189224243164,
      "learning_rate": 1e-05,
      "loss": 1.0225,
      "step": 9690
    },
    {
      "epoch": 8.587245349867139,
      "grad_norm": 0.3966004252433777,
      "learning_rate": 1e-05,
      "loss": 1.0457,
      "step": 9695
    },
    {
      "epoch": 8.591674047829938,
      "grad_norm": 0.34525537490844727,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 9700
    },
    {
      "epoch": 8.596102745792736,
      "grad_norm": 0.363925963640213,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 9705
    },
    {
      "epoch": 8.600531443755536,
      "grad_norm": 0.33375084400177,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 9710
    },
    {
      "epoch": 8.604960141718335,
      "grad_norm": 0.2654789090156555,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 9715
    },
    {
      "epoch": 8.609388839681134,
      "grad_norm": 0.299304336309433,
      "learning_rate": 1e-05,
      "loss": 1.0422,
      "step": 9720
    },
    {
      "epoch": 8.613817537643932,
      "grad_norm": 0.35462525486946106,
      "learning_rate": 1e-05,
      "loss": 1.0494,
      "step": 9725
    },
    {
      "epoch": 8.618246235606732,
      "grad_norm": 0.31569480895996094,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 9730
    },
    {
      "epoch": 8.622674933569531,
      "grad_norm": 0.32788023352622986,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 9735
    },
    {
      "epoch": 8.627103631532329,
      "grad_norm": 0.27471575140953064,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 9740
    },
    {
      "epoch": 8.631532329495128,
      "grad_norm": 0.3337244391441345,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 9745
    },
    {
      "epoch": 8.635961027457927,
      "grad_norm": 0.2886463403701782,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 9750
    },
    {
      "epoch": 8.640389725420727,
      "grad_norm": 0.2794244885444641,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 9755
    },
    {
      "epoch": 8.644818423383525,
      "grad_norm": 0.2633725702762604,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 9760
    },
    {
      "epoch": 8.649247121346324,
      "grad_norm": 0.34546712040901184,
      "learning_rate": 1e-05,
      "loss": 1.0547,
      "step": 9765
    },
    {
      "epoch": 8.653675819309123,
      "grad_norm": 0.3061356842517853,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 9770
    },
    {
      "epoch": 8.658104517271923,
      "grad_norm": 0.25276973843574524,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 9775
    },
    {
      "epoch": 8.66253321523472,
      "grad_norm": 0.3296339809894562,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 9780
    },
    {
      "epoch": 8.66696191319752,
      "grad_norm": 0.4279148578643799,
      "learning_rate": 1e-05,
      "loss": 1.0461,
      "step": 9785
    },
    {
      "epoch": 8.67139061116032,
      "grad_norm": 0.4023708701133728,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 9790
    },
    {
      "epoch": 8.675819309123117,
      "grad_norm": 0.30620500445365906,
      "learning_rate": 1e-05,
      "loss": 1.0208,
      "step": 9795
    },
    {
      "epoch": 8.680248007085916,
      "grad_norm": 0.28190991282463074,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 9800
    },
    {
      "epoch": 8.684676705048716,
      "grad_norm": 0.3588258922100067,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 9805
    },
    {
      "epoch": 8.689105403011515,
      "grad_norm": 0.2799982726573944,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 9810
    },
    {
      "epoch": 8.693534100974313,
      "grad_norm": 0.3799862265586853,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 9815
    },
    {
      "epoch": 8.697962798937112,
      "grad_norm": 0.3779396712779999,
      "learning_rate": 1e-05,
      "loss": 1.0214,
      "step": 9820
    },
    {
      "epoch": 8.702391496899912,
      "grad_norm": 0.2611728608608246,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 9825
    },
    {
      "epoch": 8.706820194862711,
      "grad_norm": 0.30146512389183044,
      "learning_rate": 1e-05,
      "loss": 1.0539,
      "step": 9830
    },
    {
      "epoch": 8.711248892825509,
      "grad_norm": 0.3535897433757782,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 9835
    },
    {
      "epoch": 8.715677590788308,
      "grad_norm": 0.32646411657333374,
      "learning_rate": 1e-05,
      "loss": 1.0389,
      "step": 9840
    },
    {
      "epoch": 8.720106288751108,
      "grad_norm": 0.2533220052719116,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 9845
    },
    {
      "epoch": 8.724534986713905,
      "grad_norm": 0.29469966888427734,
      "learning_rate": 1e-05,
      "loss": 1.0357,
      "step": 9850
    },
    {
      "epoch": 8.728963684676705,
      "grad_norm": 0.34019947052001953,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 9855
    },
    {
      "epoch": 8.733392382639504,
      "grad_norm": 0.3179537057876587,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 9860
    },
    {
      "epoch": 8.737821080602302,
      "grad_norm": 0.270094633102417,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 9865
    },
    {
      "epoch": 8.742249778565101,
      "grad_norm": 0.3085406720638275,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 9870
    },
    {
      "epoch": 8.7466784765279,
      "grad_norm": 0.35700610280036926,
      "learning_rate": 1e-05,
      "loss": 0.9198,
      "step": 9875
    },
    {
      "epoch": 8.7511071744907,
      "grad_norm": 0.29298990964889526,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 9880
    },
    {
      "epoch": 8.755535872453498,
      "grad_norm": 0.28406137228012085,
      "learning_rate": 1e-05,
      "loss": 1.0337,
      "step": 9885
    },
    {
      "epoch": 8.759964570416297,
      "grad_norm": 0.3093281686306,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 9890
    },
    {
      "epoch": 8.764393268379097,
      "grad_norm": 0.3061217963695526,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 9895
    },
    {
      "epoch": 8.768821966341896,
      "grad_norm": 0.34029045701026917,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 9900
    },
    {
      "epoch": 8.773250664304694,
      "grad_norm": 0.3023642897605896,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 9905
    },
    {
      "epoch": 8.777679362267493,
      "grad_norm": 0.2699114680290222,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 9910
    },
    {
      "epoch": 8.782108060230293,
      "grad_norm": 0.34366509318351746,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 9915
    },
    {
      "epoch": 8.78653675819309,
      "grad_norm": 0.2885497212409973,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 9920
    },
    {
      "epoch": 8.79096545615589,
      "grad_norm": 0.3078930675983429,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 9925
    },
    {
      "epoch": 8.79539415411869,
      "grad_norm": 0.2534559369087219,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 9930
    },
    {
      "epoch": 8.799822852081489,
      "grad_norm": 0.20130226016044617,
      "learning_rate": 1e-05,
      "loss": 1.0258,
      "step": 9935
    },
    {
      "epoch": 8.804251550044286,
      "grad_norm": 0.3134182095527649,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 9940
    },
    {
      "epoch": 8.808680248007086,
      "grad_norm": 0.31002235412597656,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 9945
    },
    {
      "epoch": 8.813108945969885,
      "grad_norm": 0.27708548307418823,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 9950
    },
    {
      "epoch": 8.817537643932685,
      "grad_norm": 0.30106914043426514,
      "learning_rate": 1e-05,
      "loss": 1.0626,
      "step": 9955
    },
    {
      "epoch": 8.821966341895482,
      "grad_norm": 0.26456499099731445,
      "learning_rate": 1e-05,
      "loss": 1.0455,
      "step": 9960
    },
    {
      "epoch": 8.826395039858282,
      "grad_norm": 0.2438918799161911,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 9965
    },
    {
      "epoch": 8.830823737821081,
      "grad_norm": 0.26591259241104126,
      "learning_rate": 1e-05,
      "loss": 1.0171,
      "step": 9970
    },
    {
      "epoch": 8.835252435783879,
      "grad_norm": 0.32115495204925537,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 9975
    },
    {
      "epoch": 8.839681133746678,
      "grad_norm": 0.2797943949699402,
      "learning_rate": 1e-05,
      "loss": 1.058,
      "step": 9980
    },
    {
      "epoch": 8.844109831709478,
      "grad_norm": 0.2658204138278961,
      "learning_rate": 1e-05,
      "loss": 1.0326,
      "step": 9985
    },
    {
      "epoch": 8.848538529672277,
      "grad_norm": 0.32358720898628235,
      "learning_rate": 1e-05,
      "loss": 1.0605,
      "step": 9990
    },
    {
      "epoch": 8.852967227635075,
      "grad_norm": 0.25266775488853455,
      "learning_rate": 1e-05,
      "loss": 1.0497,
      "step": 9995
    },
    {
      "epoch": 8.857395925597874,
      "grad_norm": 0.2671074867248535,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 10000
    },
    {
      "epoch": 8.861824623560674,
      "grad_norm": 0.26102587580680847,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 10005
    },
    {
      "epoch": 8.866253321523471,
      "grad_norm": 0.26402416825294495,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 10010
    },
    {
      "epoch": 8.87068201948627,
      "grad_norm": 0.28006434440612793,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 10015
    },
    {
      "epoch": 8.87511071744907,
      "grad_norm": 0.3361985981464386,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 10020
    },
    {
      "epoch": 8.87953941541187,
      "grad_norm": 0.26976171135902405,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 10025
    },
    {
      "epoch": 8.883968113374667,
      "grad_norm": 0.36532947421073914,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 10030
    },
    {
      "epoch": 8.888396811337467,
      "grad_norm": 0.3031025528907776,
      "learning_rate": 1e-05,
      "loss": 1.0172,
      "step": 10035
    },
    {
      "epoch": 8.892825509300266,
      "grad_norm": 0.27416276931762695,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 10040
    },
    {
      "epoch": 8.897254207263064,
      "grad_norm": 0.3802085518836975,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 10045
    },
    {
      "epoch": 8.901682905225863,
      "grad_norm": 0.34512272477149963,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 10050
    },
    {
      "epoch": 8.906111603188663,
      "grad_norm": 0.27620893716812134,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 10055
    },
    {
      "epoch": 8.910540301151462,
      "grad_norm": 0.3376884162425995,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 10060
    },
    {
      "epoch": 8.91496899911426,
      "grad_norm": 0.2946058213710785,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 10065
    },
    {
      "epoch": 8.91939769707706,
      "grad_norm": 0.3219498097896576,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 10070
    },
    {
      "epoch": 8.923826395039859,
      "grad_norm": 0.3931785225868225,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 10075
    },
    {
      "epoch": 8.928255093002658,
      "grad_norm": 0.29029181599617004,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 10080
    },
    {
      "epoch": 8.932683790965456,
      "grad_norm": 0.3026534914970398,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 10085
    },
    {
      "epoch": 8.937112488928255,
      "grad_norm": 0.25995296239852905,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 10090
    },
    {
      "epoch": 8.941541186891055,
      "grad_norm": 0.2809275686740875,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 10095
    },
    {
      "epoch": 8.945969884853852,
      "grad_norm": 0.2803073823451996,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 10100
    },
    {
      "epoch": 8.950398582816652,
      "grad_norm": 0.2728479206562042,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 10105
    },
    {
      "epoch": 8.954827280779451,
      "grad_norm": 0.3069097101688385,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 10110
    },
    {
      "epoch": 8.95925597874225,
      "grad_norm": 0.36019957065582275,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 10115
    },
    {
      "epoch": 8.963684676705048,
      "grad_norm": 0.3736335039138794,
      "learning_rate": 1e-05,
      "loss": 1.0473,
      "step": 10120
    },
    {
      "epoch": 8.968113374667848,
      "grad_norm": 0.3105757534503937,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 10125
    },
    {
      "epoch": 8.972542072630647,
      "grad_norm": 0.422336608171463,
      "learning_rate": 1e-05,
      "loss": 1.1144,
      "step": 10130
    },
    {
      "epoch": 8.976970770593445,
      "grad_norm": 0.3580183982849121,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 10135
    },
    {
      "epoch": 8.981399468556244,
      "grad_norm": 0.3884885609149933,
      "learning_rate": 1e-05,
      "loss": 1.0335,
      "step": 10140
    },
    {
      "epoch": 8.985828166519044,
      "grad_norm": 0.35174691677093506,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 10145
    },
    {
      "epoch": 8.990256864481843,
      "grad_norm": 0.3623945116996765,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 10150
    },
    {
      "epoch": 8.99468556244464,
      "grad_norm": 0.3279879689216614,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 10155
    },
    {
      "epoch": 8.99911426040744,
      "grad_norm": 0.33078688383102417,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 10160
    },
    {
      "epoch": 9.00354295837024,
      "grad_norm": 0.31664299964904785,
      "learning_rate": 1e-05,
      "loss": 1.0159,
      "step": 10165
    },
    {
      "epoch": 9.007971656333037,
      "grad_norm": 0.35084065794944763,
      "learning_rate": 1e-05,
      "loss": 1.052,
      "step": 10170
    },
    {
      "epoch": 9.012400354295837,
      "grad_norm": 0.28195109963417053,
      "learning_rate": 1e-05,
      "loss": 1.0429,
      "step": 10175
    },
    {
      "epoch": 9.016829052258636,
      "grad_norm": 0.3798409700393677,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 10180
    },
    {
      "epoch": 9.021257750221436,
      "grad_norm": 0.2715992331504822,
      "learning_rate": 1e-05,
      "loss": 1.0245,
      "step": 10185
    },
    {
      "epoch": 9.025686448184233,
      "grad_norm": 0.2833086848258972,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 10190
    },
    {
      "epoch": 9.030115146147033,
      "grad_norm": 0.4003225266933441,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 10195
    },
    {
      "epoch": 9.034543844109832,
      "grad_norm": 0.37667742371559143,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 10200
    },
    {
      "epoch": 9.038972542072631,
      "grad_norm": 0.3836769461631775,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 10205
    },
    {
      "epoch": 9.043401240035429,
      "grad_norm": 0.36238908767700195,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 10210
    },
    {
      "epoch": 9.047829937998229,
      "grad_norm": 0.3144984841346741,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 10215
    },
    {
      "epoch": 9.052258635961028,
      "grad_norm": 0.30484485626220703,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 10220
    },
    {
      "epoch": 9.056687333923826,
      "grad_norm": 0.3397437334060669,
      "learning_rate": 1e-05,
      "loss": 1.022,
      "step": 10225
    },
    {
      "epoch": 9.061116031886625,
      "grad_norm": 0.28322675824165344,
      "learning_rate": 1e-05,
      "loss": 1.0097,
      "step": 10230
    },
    {
      "epoch": 9.065544729849424,
      "grad_norm": 0.3227088153362274,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 10235
    },
    {
      "epoch": 9.069973427812224,
      "grad_norm": 0.28987735509872437,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 10240
    },
    {
      "epoch": 9.074402125775022,
      "grad_norm": 0.33533304929733276,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 10245
    },
    {
      "epoch": 9.078830823737821,
      "grad_norm": 0.2863968014717102,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 10250
    },
    {
      "epoch": 9.08325952170062,
      "grad_norm": 0.320400595664978,
      "learning_rate": 1e-05,
      "loss": 1.0281,
      "step": 10255
    },
    {
      "epoch": 9.087688219663418,
      "grad_norm": 0.43020936846733093,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 10260
    },
    {
      "epoch": 9.092116917626218,
      "grad_norm": 0.350770503282547,
      "learning_rate": 1e-05,
      "loss": 1.0572,
      "step": 10265
    },
    {
      "epoch": 9.096545615589017,
      "grad_norm": 0.31009382009506226,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 10270
    },
    {
      "epoch": 9.100974313551816,
      "grad_norm": 0.3370135724544525,
      "learning_rate": 1e-05,
      "loss": 1.0231,
      "step": 10275
    },
    {
      "epoch": 9.105403011514614,
      "grad_norm": 0.31504905223846436,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 10280
    },
    {
      "epoch": 9.109831709477413,
      "grad_norm": 0.3323787748813629,
      "learning_rate": 1e-05,
      "loss": 1.022,
      "step": 10285
    },
    {
      "epoch": 9.114260407440213,
      "grad_norm": 0.23321537673473358,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 10290
    },
    {
      "epoch": 9.118689105403012,
      "grad_norm": 0.29984867572784424,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 10295
    },
    {
      "epoch": 9.12311780336581,
      "grad_norm": 0.28411856293678284,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 10300
    },
    {
      "epoch": 9.12754650132861,
      "grad_norm": 0.2836872935295105,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 10305
    },
    {
      "epoch": 9.131975199291409,
      "grad_norm": 0.2888641953468323,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 10310
    },
    {
      "epoch": 9.136403897254207,
      "grad_norm": 0.2679941952228546,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 10315
    },
    {
      "epoch": 9.140832595217006,
      "grad_norm": 0.35586997866630554,
      "learning_rate": 1e-05,
      "loss": 1.0633,
      "step": 10320
    },
    {
      "epoch": 9.145261293179805,
      "grad_norm": 0.2879399359226227,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 10325
    },
    {
      "epoch": 9.149689991142605,
      "grad_norm": 0.35500773787498474,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 10330
    },
    {
      "epoch": 9.154118689105402,
      "grad_norm": 0.27988946437835693,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 10335
    },
    {
      "epoch": 9.158547387068202,
      "grad_norm": 0.33403074741363525,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 10340
    },
    {
      "epoch": 9.162976085031001,
      "grad_norm": 0.3000040650367737,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 10345
    },
    {
      "epoch": 9.167404782993799,
      "grad_norm": 0.35046878457069397,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 10350
    },
    {
      "epoch": 9.171833480956598,
      "grad_norm": 0.35079461336135864,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 10355
    },
    {
      "epoch": 9.176262178919398,
      "grad_norm": 0.3019295036792755,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 10360
    },
    {
      "epoch": 9.180690876882197,
      "grad_norm": 0.3077620565891266,
      "learning_rate": 1e-05,
      "loss": 1.0257,
      "step": 10365
    },
    {
      "epoch": 9.185119574844995,
      "grad_norm": 0.2527737319469452,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 10370
    },
    {
      "epoch": 9.189548272807794,
      "grad_norm": 0.32669419050216675,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 10375
    },
    {
      "epoch": 9.193976970770594,
      "grad_norm": 0.31431785225868225,
      "learning_rate": 1e-05,
      "loss": 1.0353,
      "step": 10380
    },
    {
      "epoch": 9.198405668733393,
      "grad_norm": 0.30633842945098877,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 10385
    },
    {
      "epoch": 9.202834366696191,
      "grad_norm": 0.34725508093833923,
      "learning_rate": 1e-05,
      "loss": 1.0085,
      "step": 10390
    },
    {
      "epoch": 9.20726306465899,
      "grad_norm": 0.26888903975486755,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 10395
    },
    {
      "epoch": 9.21169176262179,
      "grad_norm": 0.34096434712409973,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 10400
    },
    {
      "epoch": 9.216120460584587,
      "grad_norm": 0.3352886736392975,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 10405
    },
    {
      "epoch": 9.220549158547387,
      "grad_norm": 0.35172680020332336,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 10410
    },
    {
      "epoch": 9.224977856510186,
      "grad_norm": 0.24038206040859222,
      "learning_rate": 1e-05,
      "loss": 1.034,
      "step": 10415
    },
    {
      "epoch": 9.229406554472986,
      "grad_norm": 0.32749372720718384,
      "learning_rate": 1e-05,
      "loss": 1.0401,
      "step": 10420
    },
    {
      "epoch": 9.233835252435783,
      "grad_norm": 0.3877907395362854,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 10425
    },
    {
      "epoch": 9.238263950398583,
      "grad_norm": 0.29924723505973816,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 10430
    },
    {
      "epoch": 9.242692648361382,
      "grad_norm": 0.3780488073825836,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 10435
    },
    {
      "epoch": 9.24712134632418,
      "grad_norm": 0.3642600476741791,
      "learning_rate": 1e-05,
      "loss": 0.9071,
      "step": 10440
    },
    {
      "epoch": 9.25155004428698,
      "grad_norm": 0.30567747354507446,
      "learning_rate": 1e-05,
      "loss": 1.063,
      "step": 10445
    },
    {
      "epoch": 9.255978742249779,
      "grad_norm": 0.2744263708591461,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 10450
    },
    {
      "epoch": 9.260407440212578,
      "grad_norm": 0.3133305013179779,
      "learning_rate": 1e-05,
      "loss": 1.0339,
      "step": 10455
    },
    {
      "epoch": 9.264836138175376,
      "grad_norm": 0.30763304233551025,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 10460
    },
    {
      "epoch": 9.269264836138175,
      "grad_norm": 0.3267073333263397,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 10465
    },
    {
      "epoch": 9.273693534100975,
      "grad_norm": 0.3251609206199646,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 10470
    },
    {
      "epoch": 9.278122232063772,
      "grad_norm": 0.29866859316825867,
      "learning_rate": 1e-05,
      "loss": 1.0516,
      "step": 10475
    },
    {
      "epoch": 9.282550930026572,
      "grad_norm": 0.4649989902973175,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 10480
    },
    {
      "epoch": 9.286979627989371,
      "grad_norm": 0.27347126603126526,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 10485
    },
    {
      "epoch": 9.29140832595217,
      "grad_norm": 0.2536153793334961,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 10490
    },
    {
      "epoch": 9.295837023914968,
      "grad_norm": 0.23823118209838867,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 10495
    },
    {
      "epoch": 9.300265721877768,
      "grad_norm": 0.31557103991508484,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 10500
    },
    {
      "epoch": 9.304694419840567,
      "grad_norm": 0.2430633157491684,
      "learning_rate": 1e-05,
      "loss": 1.0457,
      "step": 10505
    },
    {
      "epoch": 9.309123117803367,
      "grad_norm": 0.23746606707572937,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 10510
    },
    {
      "epoch": 9.313551815766164,
      "grad_norm": 0.26026999950408936,
      "learning_rate": 1e-05,
      "loss": 1.0742,
      "step": 10515
    },
    {
      "epoch": 9.317980513728964,
      "grad_norm": 0.23958712816238403,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 10520
    },
    {
      "epoch": 9.322409211691763,
      "grad_norm": 0.25504070520401,
      "learning_rate": 1e-05,
      "loss": 1.0252,
      "step": 10525
    },
    {
      "epoch": 9.32683790965456,
      "grad_norm": 0.40648138523101807,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 10530
    },
    {
      "epoch": 9.33126660761736,
      "grad_norm": 0.2843705415725708,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 10535
    },
    {
      "epoch": 9.33569530558016,
      "grad_norm": 0.3267786502838135,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 10540
    },
    {
      "epoch": 9.34012400354296,
      "grad_norm": 0.28439798951148987,
      "learning_rate": 1e-05,
      "loss": 1.0967,
      "step": 10545
    },
    {
      "epoch": 9.344552701505757,
      "grad_norm": 0.3294345736503601,
      "learning_rate": 1e-05,
      "loss": 1.0258,
      "step": 10550
    },
    {
      "epoch": 9.348981399468556,
      "grad_norm": 0.3477773070335388,
      "learning_rate": 1e-05,
      "loss": 1.0435,
      "step": 10555
    },
    {
      "epoch": 9.353410097431356,
      "grad_norm": 0.40877801179885864,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 10560
    },
    {
      "epoch": 9.357838795394153,
      "grad_norm": 0.3087904751300812,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 10565
    },
    {
      "epoch": 9.362267493356953,
      "grad_norm": 0.28086230158805847,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 10570
    },
    {
      "epoch": 9.366696191319752,
      "grad_norm": 0.23268984258174896,
      "learning_rate": 1e-05,
      "loss": 1.0567,
      "step": 10575
    },
    {
      "epoch": 9.371124889282552,
      "grad_norm": 0.2668893337249756,
      "learning_rate": 1e-05,
      "loss": 1.0388,
      "step": 10580
    },
    {
      "epoch": 9.37555358724535,
      "grad_norm": 0.23762261867523193,
      "learning_rate": 1e-05,
      "loss": 1.0222,
      "step": 10585
    },
    {
      "epoch": 9.379982285208149,
      "grad_norm": 0.31048086285591125,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 10590
    },
    {
      "epoch": 9.384410983170948,
      "grad_norm": 0.2556580603122711,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 10595
    },
    {
      "epoch": 9.388839681133746,
      "grad_norm": 0.35179075598716736,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 10600
    },
    {
      "epoch": 9.393268379096545,
      "grad_norm": 0.3239005208015442,
      "learning_rate": 1e-05,
      "loss": 1.0904,
      "step": 10605
    },
    {
      "epoch": 9.397697077059345,
      "grad_norm": 0.3315274119377136,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 10610
    },
    {
      "epoch": 9.402125775022144,
      "grad_norm": 0.3327401578426361,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 10615
    },
    {
      "epoch": 9.406554472984942,
      "grad_norm": 0.39855074882507324,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 10620
    },
    {
      "epoch": 9.410983170947741,
      "grad_norm": 0.386879563331604,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 10625
    },
    {
      "epoch": 9.41541186891054,
      "grad_norm": 0.3346370756626129,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 10630
    },
    {
      "epoch": 9.41984056687334,
      "grad_norm": 0.37306782603263855,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 10635
    },
    {
      "epoch": 9.424269264836138,
      "grad_norm": 0.29044109582901,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 10640
    },
    {
      "epoch": 9.428697962798937,
      "grad_norm": 0.3142159879207611,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 10645
    },
    {
      "epoch": 9.433126660761737,
      "grad_norm": 0.32744020223617554,
      "learning_rate": 1e-05,
      "loss": 1.0255,
      "step": 10650
    },
    {
      "epoch": 9.437555358724534,
      "grad_norm": 0.2751801013946533,
      "learning_rate": 1e-05,
      "loss": 1.0497,
      "step": 10655
    },
    {
      "epoch": 9.441984056687334,
      "grad_norm": 0.21770940721035004,
      "learning_rate": 1e-05,
      "loss": 1.0506,
      "step": 10660
    },
    {
      "epoch": 9.446412754650133,
      "grad_norm": 0.261288046836853,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 10665
    },
    {
      "epoch": 9.450841452612933,
      "grad_norm": 0.3091477155685425,
      "learning_rate": 1e-05,
      "loss": 1.0214,
      "step": 10670
    },
    {
      "epoch": 9.45527015057573,
      "grad_norm": 0.3383648991584778,
      "learning_rate": 1e-05,
      "loss": 1.0377,
      "step": 10675
    },
    {
      "epoch": 9.45969884853853,
      "grad_norm": 0.30078986287117004,
      "learning_rate": 1e-05,
      "loss": 1.0248,
      "step": 10680
    },
    {
      "epoch": 9.464127546501329,
      "grad_norm": 0.3739725947380066,
      "learning_rate": 1e-05,
      "loss": 1.032,
      "step": 10685
    },
    {
      "epoch": 9.468556244464127,
      "grad_norm": 0.3867415487766266,
      "learning_rate": 1e-05,
      "loss": 1.041,
      "step": 10690
    },
    {
      "epoch": 9.472984942426926,
      "grad_norm": 0.28349366784095764,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 10695
    },
    {
      "epoch": 9.477413640389726,
      "grad_norm": 0.3332308828830719,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 10700
    },
    {
      "epoch": 9.481842338352525,
      "grad_norm": 0.304450124502182,
      "learning_rate": 1e-05,
      "loss": 1.0218,
      "step": 10705
    },
    {
      "epoch": 9.486271036315323,
      "grad_norm": 0.3437693119049072,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 10710
    },
    {
      "epoch": 9.490699734278122,
      "grad_norm": 0.2844833731651306,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 10715
    },
    {
      "epoch": 9.495128432240922,
      "grad_norm": 0.30860450863838196,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 10720
    },
    {
      "epoch": 9.499557130203721,
      "grad_norm": 0.23085078597068787,
      "learning_rate": 1e-05,
      "loss": 1.0431,
      "step": 10725
    },
    {
      "epoch": 9.503985828166519,
      "grad_norm": 0.3047216832637787,
      "learning_rate": 1e-05,
      "loss": 1.0284,
      "step": 10730
    },
    {
      "epoch": 9.508414526129318,
      "grad_norm": 0.2583774924278259,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 10735
    },
    {
      "epoch": 9.512843224092117,
      "grad_norm": 0.18981026113033295,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 10740
    },
    {
      "epoch": 9.517271922054915,
      "grad_norm": 0.2867439091205597,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 10745
    },
    {
      "epoch": 9.521700620017715,
      "grad_norm": 0.32015806436538696,
      "learning_rate": 1e-05,
      "loss": 1.0314,
      "step": 10750
    },
    {
      "epoch": 9.526129317980514,
      "grad_norm": 0.31143102049827576,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 10755
    },
    {
      "epoch": 9.530558015943313,
      "grad_norm": 0.32476815581321716,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 10760
    },
    {
      "epoch": 9.534986713906111,
      "grad_norm": 0.34490182995796204,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 10765
    },
    {
      "epoch": 9.53941541186891,
      "grad_norm": 0.26716142892837524,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 10770
    },
    {
      "epoch": 9.54384410983171,
      "grad_norm": 0.2967308461666107,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 10775
    },
    {
      "epoch": 9.548272807794508,
      "grad_norm": 0.28547367453575134,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 10780
    },
    {
      "epoch": 9.552701505757307,
      "grad_norm": 0.29418671131134033,
      "learning_rate": 1e-05,
      "loss": 1.0324,
      "step": 10785
    },
    {
      "epoch": 9.557130203720106,
      "grad_norm": 0.30048003792762756,
      "learning_rate": 1e-05,
      "loss": 1.0245,
      "step": 10790
    },
    {
      "epoch": 9.561558901682906,
      "grad_norm": 0.28666961193084717,
      "learning_rate": 1e-05,
      "loss": 1.0528,
      "step": 10795
    },
    {
      "epoch": 9.565987599645704,
      "grad_norm": 0.3216293454170227,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 10800
    },
    {
      "epoch": 9.570416297608503,
      "grad_norm": 0.2466462254524231,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 10805
    },
    {
      "epoch": 9.574844995571302,
      "grad_norm": 0.24364490807056427,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 10810
    },
    {
      "epoch": 9.579273693534102,
      "grad_norm": 0.2657265067100525,
      "learning_rate": 1e-05,
      "loss": 1.0224,
      "step": 10815
    },
    {
      "epoch": 9.5837023914969,
      "grad_norm": 0.28962242603302,
      "learning_rate": 1e-05,
      "loss": 1.0327,
      "step": 10820
    },
    {
      "epoch": 9.588131089459699,
      "grad_norm": 0.34054914116859436,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 10825
    },
    {
      "epoch": 9.592559787422498,
      "grad_norm": 0.290404736995697,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 10830
    },
    {
      "epoch": 9.596988485385296,
      "grad_norm": 0.34479761123657227,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 10835
    },
    {
      "epoch": 9.601417183348095,
      "grad_norm": 0.3987703323364258,
      "learning_rate": 1e-05,
      "loss": 1.069,
      "step": 10840
    },
    {
      "epoch": 9.605845881310895,
      "grad_norm": 0.2594773471355438,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 10845
    },
    {
      "epoch": 9.610274579273694,
      "grad_norm": 0.30426791310310364,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 10850
    },
    {
      "epoch": 9.614703277236492,
      "grad_norm": 0.31898564100265503,
      "learning_rate": 1e-05,
      "loss": 1.0365,
      "step": 10855
    },
    {
      "epoch": 9.619131975199291,
      "grad_norm": 0.35877910256385803,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 10860
    },
    {
      "epoch": 9.62356067316209,
      "grad_norm": 0.2700461149215698,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 10865
    },
    {
      "epoch": 9.627989371124889,
      "grad_norm": 0.36071059107780457,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 10870
    },
    {
      "epoch": 9.632418069087688,
      "grad_norm": 0.25075316429138184,
      "learning_rate": 1e-05,
      "loss": 1.0199,
      "step": 10875
    },
    {
      "epoch": 9.636846767050487,
      "grad_norm": 0.2619266211986542,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 10880
    },
    {
      "epoch": 9.641275465013287,
      "grad_norm": 0.23985588550567627,
      "learning_rate": 1e-05,
      "loss": 1.0571,
      "step": 10885
    },
    {
      "epoch": 9.645704162976084,
      "grad_norm": 0.3363223671913147,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 10890
    },
    {
      "epoch": 9.650132860938884,
      "grad_norm": 0.3055192828178406,
      "learning_rate": 1e-05,
      "loss": 1.0496,
      "step": 10895
    },
    {
      "epoch": 9.654561558901683,
      "grad_norm": 0.3000132739543915,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 10900
    },
    {
      "epoch": 9.658990256864481,
      "grad_norm": 0.282427579164505,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 10905
    },
    {
      "epoch": 9.66341895482728,
      "grad_norm": 0.3415239453315735,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 10910
    },
    {
      "epoch": 9.66784765279008,
      "grad_norm": 0.2961592972278595,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 10915
    },
    {
      "epoch": 9.67227635075288,
      "grad_norm": 0.2643917202949524,
      "learning_rate": 1e-05,
      "loss": 1.0346,
      "step": 10920
    },
    {
      "epoch": 9.676705048715677,
      "grad_norm": 0.41411229968070984,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 10925
    },
    {
      "epoch": 9.681133746678476,
      "grad_norm": 0.33263343572616577,
      "learning_rate": 1e-05,
      "loss": 1.0492,
      "step": 10930
    },
    {
      "epoch": 9.685562444641276,
      "grad_norm": 0.26576805114746094,
      "learning_rate": 1e-05,
      "loss": 1.0486,
      "step": 10935
    },
    {
      "epoch": 9.689991142604075,
      "grad_norm": 0.3351614773273468,
      "learning_rate": 1e-05,
      "loss": 1.0281,
      "step": 10940
    },
    {
      "epoch": 9.694419840566873,
      "grad_norm": 0.296962171792984,
      "learning_rate": 1e-05,
      "loss": 1.0511,
      "step": 10945
    },
    {
      "epoch": 9.698848538529672,
      "grad_norm": 0.247733473777771,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 10950
    },
    {
      "epoch": 9.703277236492472,
      "grad_norm": 0.23276546597480774,
      "learning_rate": 1e-05,
      "loss": 1.0407,
      "step": 10955
    },
    {
      "epoch": 9.70770593445527,
      "grad_norm": 0.2448033094406128,
      "learning_rate": 1e-05,
      "loss": 1.0233,
      "step": 10960
    },
    {
      "epoch": 9.712134632418069,
      "grad_norm": 0.31032833456993103,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 10965
    },
    {
      "epoch": 9.716563330380868,
      "grad_norm": 0.2827463746070862,
      "learning_rate": 1e-05,
      "loss": 1.0408,
      "step": 10970
    },
    {
      "epoch": 9.720992028343668,
      "grad_norm": 0.2876776456832886,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 10975
    },
    {
      "epoch": 9.725420726306465,
      "grad_norm": 0.28047484159469604,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 10980
    },
    {
      "epoch": 9.729849424269265,
      "grad_norm": 0.31798845529556274,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 10985
    },
    {
      "epoch": 9.734278122232064,
      "grad_norm": 0.2628239393234253,
      "learning_rate": 1e-05,
      "loss": 0.942,
      "step": 10990
    },
    {
      "epoch": 9.738706820194862,
      "grad_norm": 0.3181191384792328,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 10995
    },
    {
      "epoch": 9.743135518157661,
      "grad_norm": 0.26345333456993103,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 11000
    },
    {
      "epoch": 9.74756421612046,
      "grad_norm": 0.30158525705337524,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 11005
    },
    {
      "epoch": 9.75199291408326,
      "grad_norm": 0.34285974502563477,
      "learning_rate": 1e-05,
      "loss": 1.0368,
      "step": 11010
    },
    {
      "epoch": 9.756421612046058,
      "grad_norm": 0.2689751386642456,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 11015
    },
    {
      "epoch": 9.760850310008857,
      "grad_norm": 0.2519364356994629,
      "learning_rate": 1e-05,
      "loss": 1.0259,
      "step": 11020
    },
    {
      "epoch": 9.765279007971657,
      "grad_norm": 0.2979722321033478,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 11025
    },
    {
      "epoch": 9.769707705934454,
      "grad_norm": 0.316421240568161,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 11030
    },
    {
      "epoch": 9.774136403897254,
      "grad_norm": 0.31883957982063293,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 11035
    },
    {
      "epoch": 9.778565101860053,
      "grad_norm": 0.29930931329727173,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 11040
    },
    {
      "epoch": 9.782993799822853,
      "grad_norm": 0.31545814871788025,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 11045
    },
    {
      "epoch": 9.78742249778565,
      "grad_norm": 0.31619441509246826,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 11050
    },
    {
      "epoch": 9.79185119574845,
      "grad_norm": 0.3094756305217743,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 11055
    },
    {
      "epoch": 9.79627989371125,
      "grad_norm": 0.31359949707984924,
      "learning_rate": 1e-05,
      "loss": 1.0584,
      "step": 11060
    },
    {
      "epoch": 9.800708591674049,
      "grad_norm": 0.3016626238822937,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 11065
    },
    {
      "epoch": 9.805137289636846,
      "grad_norm": 0.3067277669906616,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 11070
    },
    {
      "epoch": 9.809565987599646,
      "grad_norm": 0.3074761927127838,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 11075
    },
    {
      "epoch": 9.813994685562445,
      "grad_norm": 0.23297877609729767,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 11080
    },
    {
      "epoch": 9.818423383525243,
      "grad_norm": 0.26705634593963623,
      "learning_rate": 1e-05,
      "loss": 1.0453,
      "step": 11085
    },
    {
      "epoch": 9.822852081488042,
      "grad_norm": 0.45414841175079346,
      "learning_rate": 1e-05,
      "loss": 1.0549,
      "step": 11090
    },
    {
      "epoch": 9.827280779450842,
      "grad_norm": 0.26370587944984436,
      "learning_rate": 1e-05,
      "loss": 1.0193,
      "step": 11095
    },
    {
      "epoch": 9.831709477413641,
      "grad_norm": 0.33747348189353943,
      "learning_rate": 1e-05,
      "loss": 1.0473,
      "step": 11100
    },
    {
      "epoch": 9.836138175376439,
      "grad_norm": 0.2905241549015045,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 11105
    },
    {
      "epoch": 9.840566873339238,
      "grad_norm": 0.2959686517715454,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 11110
    },
    {
      "epoch": 9.844995571302038,
      "grad_norm": 0.24476315081119537,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 11115
    },
    {
      "epoch": 9.849424269264837,
      "grad_norm": 0.2643963396549225,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 11120
    },
    {
      "epoch": 9.853852967227635,
      "grad_norm": 0.2926819622516632,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 11125
    },
    {
      "epoch": 9.858281665190434,
      "grad_norm": 0.2401208132505417,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 11130
    },
    {
      "epoch": 9.862710363153234,
      "grad_norm": 0.3018626272678375,
      "learning_rate": 1e-05,
      "loss": 1.0404,
      "step": 11135
    },
    {
      "epoch": 9.867139061116031,
      "grad_norm": 0.298589289188385,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 11140
    },
    {
      "epoch": 9.87156775907883,
      "grad_norm": 0.30023714900016785,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 11145
    },
    {
      "epoch": 9.87599645704163,
      "grad_norm": 0.2657444477081299,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 11150
    },
    {
      "epoch": 9.88042515500443,
      "grad_norm": 0.29500603675842285,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 11155
    },
    {
      "epoch": 9.884853852967227,
      "grad_norm": 0.29076966643333435,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 11160
    },
    {
      "epoch": 9.889282550930027,
      "grad_norm": 0.3475627899169922,
      "learning_rate": 1e-05,
      "loss": 0.9348,
      "step": 11165
    },
    {
      "epoch": 9.893711248892826,
      "grad_norm": 0.26014071702957153,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 11170
    },
    {
      "epoch": 9.898139946855624,
      "grad_norm": 0.2554040253162384,
      "learning_rate": 1e-05,
      "loss": 1.0628,
      "step": 11175
    },
    {
      "epoch": 9.902568644818423,
      "grad_norm": 0.25429725646972656,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 11180
    },
    {
      "epoch": 9.906997342781223,
      "grad_norm": 0.3491935729980469,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 11185
    },
    {
      "epoch": 9.911426040744022,
      "grad_norm": 0.40521156787872314,
      "learning_rate": 1e-05,
      "loss": 1.0203,
      "step": 11190
    },
    {
      "epoch": 9.91585473870682,
      "grad_norm": 0.37809327244758606,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 11195
    },
    {
      "epoch": 9.920283436669619,
      "grad_norm": 0.32339635491371155,
      "learning_rate": 1e-05,
      "loss": 1.0253,
      "step": 11200
    },
    {
      "epoch": 9.924712134632419,
      "grad_norm": 0.3003422021865845,
      "learning_rate": 1e-05,
      "loss": 1.038,
      "step": 11205
    },
    {
      "epoch": 9.929140832595216,
      "grad_norm": 0.23051877319812775,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 11210
    },
    {
      "epoch": 9.933569530558016,
      "grad_norm": 0.30091825127601624,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 11215
    },
    {
      "epoch": 9.937998228520815,
      "grad_norm": 0.3318042755126953,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 11220
    },
    {
      "epoch": 9.942426926483614,
      "grad_norm": 0.24081188440322876,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 11225
    },
    {
      "epoch": 9.946855624446412,
      "grad_norm": 0.26931074261665344,
      "learning_rate": 1e-05,
      "loss": 1.0551,
      "step": 11230
    },
    {
      "epoch": 9.951284322409212,
      "grad_norm": 0.2333841174840927,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 11235
    },
    {
      "epoch": 9.955713020372011,
      "grad_norm": 0.29254716634750366,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 11240
    },
    {
      "epoch": 9.96014171833481,
      "grad_norm": 0.2870270609855652,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 11245
    },
    {
      "epoch": 9.964570416297608,
      "grad_norm": 0.22885331511497498,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 11250
    },
    {
      "epoch": 9.968999114260408,
      "grad_norm": 0.3018409311771393,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 11255
    },
    {
      "epoch": 9.973427812223207,
      "grad_norm": 0.2703056335449219,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 11260
    },
    {
      "epoch": 9.977856510186005,
      "grad_norm": 0.23221834003925323,
      "learning_rate": 1e-05,
      "loss": 0.9202,
      "step": 11265
    },
    {
      "epoch": 9.982285208148804,
      "grad_norm": 0.2505277395248413,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 11270
    },
    {
      "epoch": 9.986713906111603,
      "grad_norm": 0.32141977548599243,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 11275
    },
    {
      "epoch": 9.991142604074403,
      "grad_norm": 0.285922110080719,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 11280
    },
    {
      "epoch": 9.9955713020372,
      "grad_norm": 0.3130255341529846,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 11285
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.29459288716316223,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 11290
    },
    {
      "epoch": 10.0044286979628,
      "grad_norm": 0.3114648461341858,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 11295
    },
    {
      "epoch": 10.008857395925597,
      "grad_norm": 0.34349703788757324,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 11300
    },
    {
      "epoch": 10.013286093888397,
      "grad_norm": 0.3219708800315857,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 11305
    },
    {
      "epoch": 10.017714791851196,
      "grad_norm": 0.34660136699676514,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 11310
    },
    {
      "epoch": 10.022143489813995,
      "grad_norm": 0.3349578380584717,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 11315
    },
    {
      "epoch": 10.026572187776793,
      "grad_norm": 0.3159612715244293,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 11320
    },
    {
      "epoch": 10.031000885739592,
      "grad_norm": 0.28448784351348877,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 11325
    },
    {
      "epoch": 10.035429583702392,
      "grad_norm": 0.2708514928817749,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 11330
    },
    {
      "epoch": 10.03985828166519,
      "grad_norm": 0.22899767756462097,
      "learning_rate": 1e-05,
      "loss": 1.0341,
      "step": 11335
    },
    {
      "epoch": 10.044286979627989,
      "grad_norm": 0.2686099410057068,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 11340
    },
    {
      "epoch": 10.048715677590788,
      "grad_norm": 0.3064838647842407,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 11345
    },
    {
      "epoch": 10.053144375553588,
      "grad_norm": 0.3376131057739258,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 11350
    },
    {
      "epoch": 10.057573073516386,
      "grad_norm": 0.26854780316352844,
      "learning_rate": 1e-05,
      "loss": 1.0448,
      "step": 11355
    },
    {
      "epoch": 10.062001771479185,
      "grad_norm": 0.27277448773384094,
      "learning_rate": 1e-05,
      "loss": 1.0456,
      "step": 11360
    },
    {
      "epoch": 10.066430469441984,
      "grad_norm": 0.29424813389778137,
      "learning_rate": 1e-05,
      "loss": 1.0828,
      "step": 11365
    },
    {
      "epoch": 10.070859167404784,
      "grad_norm": 0.4012397229671478,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 11370
    },
    {
      "epoch": 10.075287865367581,
      "grad_norm": 0.34967800974845886,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 11375
    },
    {
      "epoch": 10.079716563330381,
      "grad_norm": 0.3148069381713867,
      "learning_rate": 1e-05,
      "loss": 1.0482,
      "step": 11380
    },
    {
      "epoch": 10.08414526129318,
      "grad_norm": 0.36094632744789124,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 11385
    },
    {
      "epoch": 10.088573959255978,
      "grad_norm": 0.31681621074676514,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 11390
    },
    {
      "epoch": 10.093002657218777,
      "grad_norm": 0.264861136674881,
      "learning_rate": 1e-05,
      "loss": 1.0488,
      "step": 11395
    },
    {
      "epoch": 10.097431355181577,
      "grad_norm": 0.3462587594985962,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 11400
    },
    {
      "epoch": 10.101860053144376,
      "grad_norm": 0.2840697467327118,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 11405
    },
    {
      "epoch": 10.106288751107174,
      "grad_norm": 0.2997097373008728,
      "learning_rate": 1e-05,
      "loss": 1.031,
      "step": 11410
    },
    {
      "epoch": 10.110717449069973,
      "grad_norm": 0.2907783091068268,
      "learning_rate": 1e-05,
      "loss": 1.0462,
      "step": 11415
    },
    {
      "epoch": 10.115146147032773,
      "grad_norm": 0.2680007517337799,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 11420
    },
    {
      "epoch": 10.11957484499557,
      "grad_norm": 0.29785585403442383,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 11425
    },
    {
      "epoch": 10.12400354295837,
      "grad_norm": 0.2777506709098816,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 11430
    },
    {
      "epoch": 10.12843224092117,
      "grad_norm": 0.30228719115257263,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 11435
    },
    {
      "epoch": 10.132860938883969,
      "grad_norm": 0.2686280608177185,
      "learning_rate": 1e-05,
      "loss": 1.0511,
      "step": 11440
    },
    {
      "epoch": 10.137289636846766,
      "grad_norm": 0.29569387435913086,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 11445
    },
    {
      "epoch": 10.141718334809566,
      "grad_norm": 0.3754440248012543,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 11450
    },
    {
      "epoch": 10.146147032772365,
      "grad_norm": 0.30589839816093445,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 11455
    },
    {
      "epoch": 10.150575730735165,
      "grad_norm": 0.2895824611186981,
      "learning_rate": 1e-05,
      "loss": 1.0415,
      "step": 11460
    },
    {
      "epoch": 10.155004428697962,
      "grad_norm": 0.23917438089847565,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 11465
    },
    {
      "epoch": 10.159433126660762,
      "grad_norm": 0.3071250021457672,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 11470
    },
    {
      "epoch": 10.163861824623561,
      "grad_norm": 0.25300514698028564,
      "learning_rate": 1e-05,
      "loss": 1.0335,
      "step": 11475
    },
    {
      "epoch": 10.168290522586359,
      "grad_norm": 0.29810330271720886,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 11480
    },
    {
      "epoch": 10.172719220549158,
      "grad_norm": 0.26134008169174194,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 11485
    },
    {
      "epoch": 10.177147918511958,
      "grad_norm": 0.28486084938049316,
      "learning_rate": 1e-05,
      "loss": 1.0226,
      "step": 11490
    },
    {
      "epoch": 10.181576616474757,
      "grad_norm": 0.2644582688808441,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 11495
    },
    {
      "epoch": 10.186005314437555,
      "grad_norm": 0.25823184847831726,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 11500
    },
    {
      "epoch": 10.190434012400354,
      "grad_norm": 0.2837287485599518,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 11505
    },
    {
      "epoch": 10.194862710363154,
      "grad_norm": 0.31140053272247314,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 11510
    },
    {
      "epoch": 10.199291408325951,
      "grad_norm": 0.2545853555202484,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 11515
    },
    {
      "epoch": 10.20372010628875,
      "grad_norm": 0.26009130477905273,
      "learning_rate": 1e-05,
      "loss": 1.0256,
      "step": 11520
    },
    {
      "epoch": 10.20814880425155,
      "grad_norm": 0.3061967194080353,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 11525
    },
    {
      "epoch": 10.21257750221435,
      "grad_norm": 0.28326642513275146,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 11530
    },
    {
      "epoch": 10.217006200177147,
      "grad_norm": 0.27370598912239075,
      "learning_rate": 1e-05,
      "loss": 1.0474,
      "step": 11535
    },
    {
      "epoch": 10.221434898139947,
      "grad_norm": 0.3138013482093811,
      "learning_rate": 1e-05,
      "loss": 1.023,
      "step": 11540
    },
    {
      "epoch": 10.225863596102746,
      "grad_norm": 0.3052850067615509,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 11545
    },
    {
      "epoch": 10.230292294065544,
      "grad_norm": 0.30269312858581543,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 11550
    },
    {
      "epoch": 10.234720992028343,
      "grad_norm": 0.3532915711402893,
      "learning_rate": 1e-05,
      "loss": 1.0414,
      "step": 11555
    },
    {
      "epoch": 10.239149689991143,
      "grad_norm": 0.30972394347190857,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 11560
    },
    {
      "epoch": 10.243578387953942,
      "grad_norm": 0.27440938353538513,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 11565
    },
    {
      "epoch": 10.24800708591674,
      "grad_norm": 0.28840890526771545,
      "learning_rate": 1e-05,
      "loss": 1.0256,
      "step": 11570
    },
    {
      "epoch": 10.25243578387954,
      "grad_norm": 0.2969173192977905,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 11575
    },
    {
      "epoch": 10.256864481842339,
      "grad_norm": 0.3268223702907562,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 11580
    },
    {
      "epoch": 10.261293179805138,
      "grad_norm": 0.2664327025413513,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 11585
    },
    {
      "epoch": 10.265721877767936,
      "grad_norm": 0.28675204515457153,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 11590
    },
    {
      "epoch": 10.270150575730735,
      "grad_norm": 0.34164151549339294,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 11595
    },
    {
      "epoch": 10.274579273693535,
      "grad_norm": 0.3371868431568146,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 11600
    },
    {
      "epoch": 10.279007971656332,
      "grad_norm": 0.3159741461277008,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 11605
    },
    {
      "epoch": 10.283436669619132,
      "grad_norm": 0.2591995298862457,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 11610
    },
    {
      "epoch": 10.287865367581931,
      "grad_norm": 0.26970839500427246,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 11615
    },
    {
      "epoch": 10.29229406554473,
      "grad_norm": 0.29870346188545227,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 11620
    },
    {
      "epoch": 10.296722763507528,
      "grad_norm": 0.30782654881477356,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 11625
    },
    {
      "epoch": 10.301151461470328,
      "grad_norm": 0.35799652338027954,
      "learning_rate": 1e-05,
      "loss": 1.0563,
      "step": 11630
    },
    {
      "epoch": 10.305580159433127,
      "grad_norm": 0.28341639041900635,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 11635
    },
    {
      "epoch": 10.310008857395925,
      "grad_norm": 0.31674161553382874,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 11640
    },
    {
      "epoch": 10.314437555358724,
      "grad_norm": 0.29714614152908325,
      "learning_rate": 1e-05,
      "loss": 1.0111,
      "step": 11645
    },
    {
      "epoch": 10.318866253321524,
      "grad_norm": 0.28980761766433716,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 11650
    },
    {
      "epoch": 10.323294951284323,
      "grad_norm": 0.25496095418930054,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 11655
    },
    {
      "epoch": 10.32772364924712,
      "grad_norm": 0.2615630030632019,
      "learning_rate": 1e-05,
      "loss": 1.0467,
      "step": 11660
    },
    {
      "epoch": 10.33215234720992,
      "grad_norm": 0.24543699622154236,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 11665
    },
    {
      "epoch": 10.33658104517272,
      "grad_norm": 0.3174920082092285,
      "learning_rate": 1e-05,
      "loss": 1.0152,
      "step": 11670
    },
    {
      "epoch": 10.341009743135519,
      "grad_norm": 0.26769909262657166,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 11675
    },
    {
      "epoch": 10.345438441098317,
      "grad_norm": 0.29964545369148254,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 11680
    },
    {
      "epoch": 10.349867139061116,
      "grad_norm": 0.28694555163383484,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 11685
    },
    {
      "epoch": 10.354295837023916,
      "grad_norm": 0.303021103143692,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 11690
    },
    {
      "epoch": 10.358724534986713,
      "grad_norm": 0.32018807530403137,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 11695
    },
    {
      "epoch": 10.363153232949513,
      "grad_norm": 0.252888023853302,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 11700
    },
    {
      "epoch": 10.367581930912312,
      "grad_norm": 0.2665899991989136,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 11705
    },
    {
      "epoch": 10.372010628875111,
      "grad_norm": 0.3029266893863678,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 11710
    },
    {
      "epoch": 10.37643932683791,
      "grad_norm": 0.32538360357284546,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 11715
    },
    {
      "epoch": 10.380868024800709,
      "grad_norm": 0.3245810568332672,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 11720
    },
    {
      "epoch": 10.385296722763508,
      "grad_norm": 0.2661267817020416,
      "learning_rate": 1e-05,
      "loss": 1.0415,
      "step": 11725
    },
    {
      "epoch": 10.389725420726306,
      "grad_norm": 0.26853328943252563,
      "learning_rate": 1e-05,
      "loss": 1.0365,
      "step": 11730
    },
    {
      "epoch": 10.394154118689105,
      "grad_norm": 0.27677589654922485,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 11735
    },
    {
      "epoch": 10.398582816651905,
      "grad_norm": 0.3096354305744171,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 11740
    },
    {
      "epoch": 10.403011514614704,
      "grad_norm": 0.2818625867366791,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 11745
    },
    {
      "epoch": 10.407440212577502,
      "grad_norm": 0.3005274534225464,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 11750
    },
    {
      "epoch": 10.411868910540301,
      "grad_norm": 0.2543414235115051,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 11755
    },
    {
      "epoch": 10.4162976085031,
      "grad_norm": 0.2781553864479065,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 11760
    },
    {
      "epoch": 10.420726306465898,
      "grad_norm": 0.3184894919395447,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 11765
    },
    {
      "epoch": 10.425155004428698,
      "grad_norm": 0.27732446789741516,
      "learning_rate": 1e-05,
      "loss": 0.91,
      "step": 11770
    },
    {
      "epoch": 10.429583702391497,
      "grad_norm": 0.28109610080718994,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 11775
    },
    {
      "epoch": 10.434012400354296,
      "grad_norm": 0.3391228914260864,
      "learning_rate": 1e-05,
      "loss": 1.051,
      "step": 11780
    },
    {
      "epoch": 10.438441098317094,
      "grad_norm": 0.3156225383281708,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 11785
    },
    {
      "epoch": 10.442869796279894,
      "grad_norm": 0.3333657681941986,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 11790
    },
    {
      "epoch": 10.447298494242693,
      "grad_norm": 0.29760661721229553,
      "learning_rate": 1e-05,
      "loss": 0.9493,
      "step": 11795
    },
    {
      "epoch": 10.451727192205492,
      "grad_norm": 0.3054805099964142,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 11800
    },
    {
      "epoch": 10.45615589016829,
      "grad_norm": 0.31078213453292847,
      "learning_rate": 1e-05,
      "loss": 1.0386,
      "step": 11805
    },
    {
      "epoch": 10.46058458813109,
      "grad_norm": 0.3612339198589325,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 11810
    },
    {
      "epoch": 10.465013286093889,
      "grad_norm": 0.23194549977779388,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 11815
    },
    {
      "epoch": 10.469441984056687,
      "grad_norm": 0.32136034965515137,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 11820
    },
    {
      "epoch": 10.473870682019486,
      "grad_norm": 0.2960103452205658,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 11825
    },
    {
      "epoch": 10.478299379982285,
      "grad_norm": 0.27054598927497864,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 11830
    },
    {
      "epoch": 10.482728077945085,
      "grad_norm": 0.29483795166015625,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 11835
    },
    {
      "epoch": 10.487156775907883,
      "grad_norm": 0.37229597568511963,
      "learning_rate": 1e-05,
      "loss": 1.024,
      "step": 11840
    },
    {
      "epoch": 10.491585473870682,
      "grad_norm": 0.28565678000450134,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 11845
    },
    {
      "epoch": 10.496014171833481,
      "grad_norm": 0.4019598662853241,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 11850
    },
    {
      "epoch": 10.50044286979628,
      "grad_norm": 0.36128920316696167,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 11855
    },
    {
      "epoch": 10.504871567759078,
      "grad_norm": 0.3604027032852173,
      "learning_rate": 1e-05,
      "loss": 1.065,
      "step": 11860
    },
    {
      "epoch": 10.509300265721878,
      "grad_norm": 0.34436139464378357,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 11865
    },
    {
      "epoch": 10.513728963684677,
      "grad_norm": 0.2565373182296753,
      "learning_rate": 1e-05,
      "loss": 1.0445,
      "step": 11870
    },
    {
      "epoch": 10.518157661647475,
      "grad_norm": 0.2698712646961212,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 11875
    },
    {
      "epoch": 10.522586359610274,
      "grad_norm": 0.32578006386756897,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 11880
    },
    {
      "epoch": 10.527015057573074,
      "grad_norm": 0.2972579896450043,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 11885
    },
    {
      "epoch": 10.531443755535872,
      "grad_norm": 0.3600703477859497,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 11890
    },
    {
      "epoch": 10.535872453498671,
      "grad_norm": 0.2374672293663025,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 11895
    },
    {
      "epoch": 10.54030115146147,
      "grad_norm": 0.250801146030426,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 11900
    },
    {
      "epoch": 10.54472984942427,
      "grad_norm": 0.3013153672218323,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 11905
    },
    {
      "epoch": 10.549158547387067,
      "grad_norm": 0.3373948037624359,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 11910
    },
    {
      "epoch": 10.553587245349867,
      "grad_norm": 0.30375102162361145,
      "learning_rate": 1e-05,
      "loss": 1.0259,
      "step": 11915
    },
    {
      "epoch": 10.558015943312666,
      "grad_norm": 0.2554892599582672,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 11920
    },
    {
      "epoch": 10.562444641275466,
      "grad_norm": 0.29160356521606445,
      "learning_rate": 1e-05,
      "loss": 1.0085,
      "step": 11925
    },
    {
      "epoch": 10.566873339238263,
      "grad_norm": 0.27245867252349854,
      "learning_rate": 1e-05,
      "loss": 1.0546,
      "step": 11930
    },
    {
      "epoch": 10.571302037201063,
      "grad_norm": 0.2780960500240326,
      "learning_rate": 1e-05,
      "loss": 1.0347,
      "step": 11935
    },
    {
      "epoch": 10.575730735163862,
      "grad_norm": 0.31066030263900757,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 11940
    },
    {
      "epoch": 10.58015943312666,
      "grad_norm": 0.2587489187717438,
      "learning_rate": 1e-05,
      "loss": 1.0267,
      "step": 11945
    },
    {
      "epoch": 10.58458813108946,
      "grad_norm": 0.44822588562965393,
      "learning_rate": 1e-05,
      "loss": 0.9048,
      "step": 11950
    },
    {
      "epoch": 10.589016829052259,
      "grad_norm": 0.352801650762558,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 11955
    },
    {
      "epoch": 10.593445527015058,
      "grad_norm": 0.3258765935897827,
      "learning_rate": 1e-05,
      "loss": 1.049,
      "step": 11960
    },
    {
      "epoch": 10.597874224977856,
      "grad_norm": 0.31580856442451477,
      "learning_rate": 1e-05,
      "loss": 1.0204,
      "step": 11965
    },
    {
      "epoch": 10.602302922940655,
      "grad_norm": 0.2915458381175995,
      "learning_rate": 1e-05,
      "loss": 1.0549,
      "step": 11970
    },
    {
      "epoch": 10.606731620903455,
      "grad_norm": 0.31409668922424316,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 11975
    },
    {
      "epoch": 10.611160318866254,
      "grad_norm": 0.28547248244285583,
      "learning_rate": 1e-05,
      "loss": 1.0299,
      "step": 11980
    },
    {
      "epoch": 10.615589016829052,
      "grad_norm": 0.28606781363487244,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 11985
    },
    {
      "epoch": 10.620017714791851,
      "grad_norm": 0.3044525980949402,
      "learning_rate": 1e-05,
      "loss": 1.0183,
      "step": 11990
    },
    {
      "epoch": 10.62444641275465,
      "grad_norm": 0.31772077083587646,
      "learning_rate": 1e-05,
      "loss": 1.0424,
      "step": 11995
    },
    {
      "epoch": 10.628875110717448,
      "grad_norm": 0.24412225186824799,
      "learning_rate": 1e-05,
      "loss": 1.0245,
      "step": 12000
    },
    {
      "epoch": 10.633303808680248,
      "grad_norm": 0.3045209050178528,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 12005
    },
    {
      "epoch": 10.637732506643047,
      "grad_norm": 0.2466321438550949,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 12010
    },
    {
      "epoch": 10.642161204605847,
      "grad_norm": 0.30908432602882385,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 12015
    },
    {
      "epoch": 10.646589902568644,
      "grad_norm": 0.27930381894111633,
      "learning_rate": 1e-05,
      "loss": 1.023,
      "step": 12020
    },
    {
      "epoch": 10.651018600531444,
      "grad_norm": 0.26829177141189575,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 12025
    },
    {
      "epoch": 10.655447298494243,
      "grad_norm": 0.28117358684539795,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 12030
    },
    {
      "epoch": 10.65987599645704,
      "grad_norm": 0.30929335951805115,
      "learning_rate": 1e-05,
      "loss": 1.0337,
      "step": 12035
    },
    {
      "epoch": 10.66430469441984,
      "grad_norm": 0.327304869890213,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 12040
    },
    {
      "epoch": 10.66873339238264,
      "grad_norm": 0.25552797317504883,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 12045
    },
    {
      "epoch": 10.67316209034544,
      "grad_norm": 0.3285534083843231,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 12050
    },
    {
      "epoch": 10.677590788308237,
      "grad_norm": 0.2624794542789459,
      "learning_rate": 1e-05,
      "loss": 1.0296,
      "step": 12055
    },
    {
      "epoch": 10.682019486271036,
      "grad_norm": 0.2602158784866333,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 12060
    },
    {
      "epoch": 10.686448184233836,
      "grad_norm": 0.2626660466194153,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 12065
    },
    {
      "epoch": 10.690876882196633,
      "grad_norm": 0.33077991008758545,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 12070
    },
    {
      "epoch": 10.695305580159433,
      "grad_norm": 0.31237027049064636,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 12075
    },
    {
      "epoch": 10.699734278122232,
      "grad_norm": 0.25750496983528137,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 12080
    },
    {
      "epoch": 10.704162976085032,
      "grad_norm": 0.3051992654800415,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 12085
    },
    {
      "epoch": 10.70859167404783,
      "grad_norm": 0.2875952422618866,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 12090
    },
    {
      "epoch": 10.713020372010629,
      "grad_norm": 0.3118644952774048,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 12095
    },
    {
      "epoch": 10.717449069973428,
      "grad_norm": 0.19404920935630798,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 12100
    },
    {
      "epoch": 10.721877767936228,
      "grad_norm": 0.25898414850234985,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 12105
    },
    {
      "epoch": 10.726306465899025,
      "grad_norm": 0.3232988715171814,
      "learning_rate": 1e-05,
      "loss": 1.043,
      "step": 12110
    },
    {
      "epoch": 10.730735163861825,
      "grad_norm": 0.35013917088508606,
      "learning_rate": 1e-05,
      "loss": 1.0174,
      "step": 12115
    },
    {
      "epoch": 10.735163861824624,
      "grad_norm": 0.3086259663105011,
      "learning_rate": 1e-05,
      "loss": 1.0359,
      "step": 12120
    },
    {
      "epoch": 10.739592559787422,
      "grad_norm": 0.2938600480556488,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 12125
    },
    {
      "epoch": 10.744021257750221,
      "grad_norm": 0.32055458426475525,
      "learning_rate": 1e-05,
      "loss": 1.0428,
      "step": 12130
    },
    {
      "epoch": 10.74844995571302,
      "grad_norm": 0.3220217823982239,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 12135
    },
    {
      "epoch": 10.75287865367582,
      "grad_norm": 0.3461529612541199,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 12140
    },
    {
      "epoch": 10.757307351638618,
      "grad_norm": 0.29279544949531555,
      "learning_rate": 1e-05,
      "loss": 1.0688,
      "step": 12145
    },
    {
      "epoch": 10.761736049601417,
      "grad_norm": 0.2685081660747528,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 12150
    },
    {
      "epoch": 10.766164747564217,
      "grad_norm": 0.26211896538734436,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 12155
    },
    {
      "epoch": 10.770593445527014,
      "grad_norm": 0.30870041251182556,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 12160
    },
    {
      "epoch": 10.775022143489814,
      "grad_norm": 0.27631425857543945,
      "learning_rate": 1e-05,
      "loss": 1.0581,
      "step": 12165
    },
    {
      "epoch": 10.779450841452613,
      "grad_norm": 0.2699381113052368,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 12170
    },
    {
      "epoch": 10.783879539415413,
      "grad_norm": 0.27515122294425964,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 12175
    },
    {
      "epoch": 10.78830823737821,
      "grad_norm": 0.2865341305732727,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 12180
    },
    {
      "epoch": 10.79273693534101,
      "grad_norm": 0.4199225902557373,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 12185
    },
    {
      "epoch": 10.797165633303809,
      "grad_norm": 0.2637953758239746,
      "learning_rate": 1e-05,
      "loss": 1.0183,
      "step": 12190
    },
    {
      "epoch": 10.801594331266607,
      "grad_norm": 0.32217082381248474,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 12195
    },
    {
      "epoch": 10.806023029229406,
      "grad_norm": 0.3124724328517914,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 12200
    },
    {
      "epoch": 10.810451727192206,
      "grad_norm": 0.2634015679359436,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 12205
    },
    {
      "epoch": 10.814880425155005,
      "grad_norm": 0.3478938341140747,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 12210
    },
    {
      "epoch": 10.819309123117803,
      "grad_norm": 0.25399765372276306,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 12215
    },
    {
      "epoch": 10.823737821080602,
      "grad_norm": 0.22675226628780365,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 12220
    },
    {
      "epoch": 10.828166519043402,
      "grad_norm": 0.28305405378341675,
      "learning_rate": 1e-05,
      "loss": 1.0386,
      "step": 12225
    },
    {
      "epoch": 10.832595217006201,
      "grad_norm": 0.3156740069389343,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 12230
    },
    {
      "epoch": 10.837023914968999,
      "grad_norm": 0.29016852378845215,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 12235
    },
    {
      "epoch": 10.841452612931798,
      "grad_norm": 0.3123618960380554,
      "learning_rate": 1e-05,
      "loss": 1.0325,
      "step": 12240
    },
    {
      "epoch": 10.845881310894598,
      "grad_norm": 0.3538264334201813,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 12245
    },
    {
      "epoch": 10.850310008857395,
      "grad_norm": 0.31346404552459717,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 12250
    },
    {
      "epoch": 10.854738706820195,
      "grad_norm": 0.22529560327529907,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 12255
    },
    {
      "epoch": 10.859167404782994,
      "grad_norm": 0.3076067268848419,
      "learning_rate": 1e-05,
      "loss": 1.0149,
      "step": 12260
    },
    {
      "epoch": 10.863596102745793,
      "grad_norm": 0.255454421043396,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 12265
    },
    {
      "epoch": 10.868024800708591,
      "grad_norm": 0.2832607328891754,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 12270
    },
    {
      "epoch": 10.87245349867139,
      "grad_norm": 0.2525108754634857,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 12275
    },
    {
      "epoch": 10.87688219663419,
      "grad_norm": 0.2743304669857025,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 12280
    },
    {
      "epoch": 10.88131089459699,
      "grad_norm": 0.26145845651626587,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 12285
    },
    {
      "epoch": 10.885739592559787,
      "grad_norm": 0.2718985676765442,
      "learning_rate": 1e-05,
      "loss": 1.017,
      "step": 12290
    },
    {
      "epoch": 10.890168290522587,
      "grad_norm": 0.26267197728157043,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 12295
    },
    {
      "epoch": 10.894596988485386,
      "grad_norm": 0.28427135944366455,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 12300
    },
    {
      "epoch": 10.899025686448184,
      "grad_norm": 0.3193788528442383,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 12305
    },
    {
      "epoch": 10.903454384410983,
      "grad_norm": 0.27316877245903015,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 12310
    },
    {
      "epoch": 10.907883082373782,
      "grad_norm": 0.36816081404685974,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 12315
    },
    {
      "epoch": 10.91231178033658,
      "grad_norm": 0.35779786109924316,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 12320
    },
    {
      "epoch": 10.91674047829938,
      "grad_norm": 0.3075735867023468,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 12325
    },
    {
      "epoch": 10.921169176262179,
      "grad_norm": 0.287017285823822,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 12330
    },
    {
      "epoch": 10.925597874224978,
      "grad_norm": 0.36687302589416504,
      "learning_rate": 1e-05,
      "loss": 1.0233,
      "step": 12335
    },
    {
      "epoch": 10.930026572187776,
      "grad_norm": 0.3728812336921692,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 12340
    },
    {
      "epoch": 10.934455270150576,
      "grad_norm": 0.34260761737823486,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 12345
    },
    {
      "epoch": 10.938883968113375,
      "grad_norm": 0.46748384833335876,
      "learning_rate": 1e-05,
      "loss": 1.0141,
      "step": 12350
    },
    {
      "epoch": 10.943312666076174,
      "grad_norm": 0.3414219617843628,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 12355
    },
    {
      "epoch": 10.947741364038972,
      "grad_norm": 0.35716310143470764,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 12360
    },
    {
      "epoch": 10.952170062001771,
      "grad_norm": 0.35552752017974854,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 12365
    },
    {
      "epoch": 10.956598759964571,
      "grad_norm": 0.29145801067352295,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 12370
    },
    {
      "epoch": 10.961027457927369,
      "grad_norm": 0.26375558972358704,
      "learning_rate": 1e-05,
      "loss": 1.0521,
      "step": 12375
    },
    {
      "epoch": 10.965456155890168,
      "grad_norm": 0.2923826575279236,
      "learning_rate": 1e-05,
      "loss": 1.0387,
      "step": 12380
    },
    {
      "epoch": 10.969884853852967,
      "grad_norm": 0.2746538519859314,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 12385
    },
    {
      "epoch": 10.974313551815767,
      "grad_norm": 0.298706591129303,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 12390
    },
    {
      "epoch": 10.978742249778564,
      "grad_norm": 0.29562684893608093,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 12395
    },
    {
      "epoch": 10.983170947741364,
      "grad_norm": 0.2692628502845764,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 12400
    },
    {
      "epoch": 10.987599645704163,
      "grad_norm": 0.3113064467906952,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 12405
    },
    {
      "epoch": 10.992028343666963,
      "grad_norm": 0.2874666154384613,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 12410
    },
    {
      "epoch": 10.99645704162976,
      "grad_norm": 0.33159539103507996,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 12415
    },
    {
      "epoch": 11.00088573959256,
      "grad_norm": 0.29299935698509216,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 12420
    },
    {
      "epoch": 11.00531443755536,
      "grad_norm": 0.3103204369544983,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 12425
    },
    {
      "epoch": 11.009743135518157,
      "grad_norm": 0.27742722630500793,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 12430
    },
    {
      "epoch": 11.014171833480956,
      "grad_norm": 0.26999253034591675,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 12435
    },
    {
      "epoch": 11.018600531443756,
      "grad_norm": 0.314456969499588,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 12440
    },
    {
      "epoch": 11.023029229406555,
      "grad_norm": 0.26246508955955505,
      "learning_rate": 1e-05,
      "loss": 1.0273,
      "step": 12445
    },
    {
      "epoch": 11.027457927369353,
      "grad_norm": 0.3206881582736969,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 12450
    },
    {
      "epoch": 11.031886625332152,
      "grad_norm": 0.2791249752044678,
      "learning_rate": 1e-05,
      "loss": 1.0301,
      "step": 12455
    },
    {
      "epoch": 11.036315323294952,
      "grad_norm": 0.3195202350616455,
      "learning_rate": 1e-05,
      "loss": 1.0487,
      "step": 12460
    },
    {
      "epoch": 11.04074402125775,
      "grad_norm": 0.3175537586212158,
      "learning_rate": 1e-05,
      "loss": 1.0673,
      "step": 12465
    },
    {
      "epoch": 11.045172719220549,
      "grad_norm": 0.29901549220085144,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 12470
    },
    {
      "epoch": 11.049601417183348,
      "grad_norm": 0.3121190369129181,
      "learning_rate": 1e-05,
      "loss": 1.0398,
      "step": 12475
    },
    {
      "epoch": 11.054030115146148,
      "grad_norm": 0.3043995797634125,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 12480
    },
    {
      "epoch": 11.058458813108945,
      "grad_norm": 0.2613113224506378,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 12485
    },
    {
      "epoch": 11.062887511071745,
      "grad_norm": 0.28122082352638245,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 12490
    },
    {
      "epoch": 11.067316209034544,
      "grad_norm": 0.2508530020713806,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 12495
    },
    {
      "epoch": 11.071744906997342,
      "grad_norm": 0.31216961145401,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 12500
    },
    {
      "epoch": 11.076173604960141,
      "grad_norm": 0.23662249743938446,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 12505
    },
    {
      "epoch": 11.08060230292294,
      "grad_norm": 0.2580465078353882,
      "learning_rate": 1e-05,
      "loss": 1.0339,
      "step": 12510
    },
    {
      "epoch": 11.08503100088574,
      "grad_norm": 0.30637606978416443,
      "learning_rate": 1e-05,
      "loss": 1.0275,
      "step": 12515
    },
    {
      "epoch": 11.089459698848538,
      "grad_norm": 0.28177255392074585,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 12520
    },
    {
      "epoch": 11.093888396811337,
      "grad_norm": 0.23359251022338867,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 12525
    },
    {
      "epoch": 11.098317094774137,
      "grad_norm": 0.28313735127449036,
      "learning_rate": 1e-05,
      "loss": 1.0054,
      "step": 12530
    },
    {
      "epoch": 11.102745792736936,
      "grad_norm": 0.25916263461112976,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 12535
    },
    {
      "epoch": 11.107174490699734,
      "grad_norm": 0.223174050450325,
      "learning_rate": 1e-05,
      "loss": 1.0284,
      "step": 12540
    },
    {
      "epoch": 11.111603188662533,
      "grad_norm": 0.270033597946167,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 12545
    },
    {
      "epoch": 11.116031886625333,
      "grad_norm": 0.294217586517334,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 12550
    },
    {
      "epoch": 11.12046058458813,
      "grad_norm": 0.254237562417984,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 12555
    },
    {
      "epoch": 11.12488928255093,
      "grad_norm": 0.2888711392879486,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 12560
    },
    {
      "epoch": 11.12931798051373,
      "grad_norm": 0.3375939428806305,
      "learning_rate": 1e-05,
      "loss": 1.0444,
      "step": 12565
    },
    {
      "epoch": 11.133746678476529,
      "grad_norm": 0.33700940012931824,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 12570
    },
    {
      "epoch": 11.138175376439326,
      "grad_norm": 0.25875553488731384,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 12575
    },
    {
      "epoch": 11.142604074402126,
      "grad_norm": 0.29349377751350403,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 12580
    },
    {
      "epoch": 11.147032772364925,
      "grad_norm": 0.27773380279541016,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 12585
    },
    {
      "epoch": 11.151461470327723,
      "grad_norm": 0.26748961210250854,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 12590
    },
    {
      "epoch": 11.155890168290522,
      "grad_norm": 0.2450910359621048,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 12595
    },
    {
      "epoch": 11.160318866253322,
      "grad_norm": 0.2941138744354248,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 12600
    },
    {
      "epoch": 11.164747564216121,
      "grad_norm": 0.25659123063087463,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 12605
    },
    {
      "epoch": 11.169176262178919,
      "grad_norm": 0.29851406812667847,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 12610
    },
    {
      "epoch": 11.173604960141718,
      "grad_norm": 0.26001042127609253,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 12615
    },
    {
      "epoch": 11.178033658104518,
      "grad_norm": 0.27933594584465027,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 12620
    },
    {
      "epoch": 11.182462356067315,
      "grad_norm": 0.3293021023273468,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 12625
    },
    {
      "epoch": 11.186891054030115,
      "grad_norm": 0.3214893937110901,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 12630
    },
    {
      "epoch": 11.191319751992914,
      "grad_norm": 0.29869943857192993,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 12635
    },
    {
      "epoch": 11.195748449955714,
      "grad_norm": 0.27732065320014954,
      "learning_rate": 1e-05,
      "loss": 0.9427,
      "step": 12640
    },
    {
      "epoch": 11.200177147918511,
      "grad_norm": 0.3572227954864502,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 12645
    },
    {
      "epoch": 11.20460584588131,
      "grad_norm": 0.3276127576828003,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 12650
    },
    {
      "epoch": 11.20903454384411,
      "grad_norm": 0.31231093406677246,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 12655
    },
    {
      "epoch": 11.21346324180691,
      "grad_norm": 0.27797722816467285,
      "learning_rate": 1e-05,
      "loss": 1.0233,
      "step": 12660
    },
    {
      "epoch": 11.217891939769707,
      "grad_norm": 0.23629514873027802,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 12665
    },
    {
      "epoch": 11.222320637732507,
      "grad_norm": 0.2862871587276459,
      "learning_rate": 1e-05,
      "loss": 1.033,
      "step": 12670
    },
    {
      "epoch": 11.226749335695306,
      "grad_norm": 0.2610045075416565,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 12675
    },
    {
      "epoch": 11.231178033658104,
      "grad_norm": 0.28060027956962585,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 12680
    },
    {
      "epoch": 11.235606731620903,
      "grad_norm": 0.2740086019039154,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 12685
    },
    {
      "epoch": 11.240035429583703,
      "grad_norm": 0.27540123462677,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 12690
    },
    {
      "epoch": 11.244464127546502,
      "grad_norm": 0.3479073643684387,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 12695
    },
    {
      "epoch": 11.2488928255093,
      "grad_norm": 0.26659345626831055,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 12700
    },
    {
      "epoch": 11.2533215234721,
      "grad_norm": 0.2560499906539917,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 12705
    },
    {
      "epoch": 11.257750221434899,
      "grad_norm": 0.2852429747581482,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 12710
    },
    {
      "epoch": 11.262178919397696,
      "grad_norm": 0.30607637763023376,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 12715
    },
    {
      "epoch": 11.266607617360496,
      "grad_norm": 0.2971040904521942,
      "learning_rate": 1e-05,
      "loss": 1.0265,
      "step": 12720
    },
    {
      "epoch": 11.271036315323295,
      "grad_norm": 0.2594987154006958,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 12725
    },
    {
      "epoch": 11.275465013286095,
      "grad_norm": 0.2745828628540039,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 12730
    },
    {
      "epoch": 11.279893711248892,
      "grad_norm": 0.22700214385986328,
      "learning_rate": 1e-05,
      "loss": 1.0297,
      "step": 12735
    },
    {
      "epoch": 11.284322409211692,
      "grad_norm": 0.2602710425853729,
      "learning_rate": 1e-05,
      "loss": 1.0453,
      "step": 12740
    },
    {
      "epoch": 11.288751107174491,
      "grad_norm": 0.27829915285110474,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 12745
    },
    {
      "epoch": 11.29317980513729,
      "grad_norm": 0.26040542125701904,
      "learning_rate": 1e-05,
      "loss": 1.0311,
      "step": 12750
    },
    {
      "epoch": 11.297608503100088,
      "grad_norm": 0.32715657353401184,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 12755
    },
    {
      "epoch": 11.302037201062888,
      "grad_norm": 0.3321079611778259,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 12760
    },
    {
      "epoch": 11.306465899025687,
      "grad_norm": 0.3088231384754181,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 12765
    },
    {
      "epoch": 11.310894596988485,
      "grad_norm": 0.2821297347545624,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 12770
    },
    {
      "epoch": 11.315323294951284,
      "grad_norm": 0.30211564898490906,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 12775
    },
    {
      "epoch": 11.319751992914084,
      "grad_norm": 0.2941362261772156,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 12780
    },
    {
      "epoch": 11.324180690876883,
      "grad_norm": 0.26926636695861816,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 12785
    },
    {
      "epoch": 11.32860938883968,
      "grad_norm": 0.26685798168182373,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 12790
    },
    {
      "epoch": 11.33303808680248,
      "grad_norm": 0.262614369392395,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 12795
    },
    {
      "epoch": 11.33746678476528,
      "grad_norm": 0.2847999036312103,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 12800
    },
    {
      "epoch": 11.341895482728077,
      "grad_norm": 0.2977809011936188,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 12805
    },
    {
      "epoch": 11.346324180690877,
      "grad_norm": 0.250999391078949,
      "learning_rate": 1e-05,
      "loss": 1.0754,
      "step": 12810
    },
    {
      "epoch": 11.350752878653676,
      "grad_norm": 0.268209844827652,
      "learning_rate": 1e-05,
      "loss": 1.0267,
      "step": 12815
    },
    {
      "epoch": 11.355181576616475,
      "grad_norm": 0.2630920708179474,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 12820
    },
    {
      "epoch": 11.359610274579273,
      "grad_norm": 0.24793335795402527,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 12825
    },
    {
      "epoch": 11.364038972542073,
      "grad_norm": 0.26168251037597656,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 12830
    },
    {
      "epoch": 11.368467670504872,
      "grad_norm": 0.27676111459732056,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 12835
    },
    {
      "epoch": 11.372896368467671,
      "grad_norm": 0.2518727779388428,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 12840
    },
    {
      "epoch": 11.377325066430469,
      "grad_norm": 0.3044617176055908,
      "learning_rate": 1e-05,
      "loss": 1.0361,
      "step": 12845
    },
    {
      "epoch": 11.381753764393268,
      "grad_norm": 0.2791688144207001,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 12850
    },
    {
      "epoch": 11.386182462356068,
      "grad_norm": 0.2488403022289276,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 12855
    },
    {
      "epoch": 11.390611160318866,
      "grad_norm": 0.26252153515815735,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 12860
    },
    {
      "epoch": 11.395039858281665,
      "grad_norm": 0.283425509929657,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 12865
    },
    {
      "epoch": 11.399468556244464,
      "grad_norm": 0.2645430266857147,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 12870
    },
    {
      "epoch": 11.403897254207264,
      "grad_norm": 0.27936238050460815,
      "learning_rate": 1e-05,
      "loss": 1.0344,
      "step": 12875
    },
    {
      "epoch": 11.408325952170062,
      "grad_norm": 0.2836114168167114,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 12880
    },
    {
      "epoch": 11.412754650132861,
      "grad_norm": 0.28614330291748047,
      "learning_rate": 1e-05,
      "loss": 1.0234,
      "step": 12885
    },
    {
      "epoch": 11.41718334809566,
      "grad_norm": 0.2818671464920044,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 12890
    },
    {
      "epoch": 11.421612046058458,
      "grad_norm": 0.24952994287014008,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 12895
    },
    {
      "epoch": 11.426040744021257,
      "grad_norm": 0.3071129024028778,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 12900
    },
    {
      "epoch": 11.430469441984057,
      "grad_norm": 0.31698107719421387,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 12905
    },
    {
      "epoch": 11.434898139946856,
      "grad_norm": 0.2559114396572113,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 12910
    },
    {
      "epoch": 11.439326837909654,
      "grad_norm": 0.23474229872226715,
      "learning_rate": 1e-05,
      "loss": 1.0444,
      "step": 12915
    },
    {
      "epoch": 11.443755535872453,
      "grad_norm": 0.25353118777275085,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 12920
    },
    {
      "epoch": 11.448184233835253,
      "grad_norm": 0.3092925250530243,
      "learning_rate": 1e-05,
      "loss": 1.0526,
      "step": 12925
    },
    {
      "epoch": 11.45261293179805,
      "grad_norm": 0.31358668208122253,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 12930
    },
    {
      "epoch": 11.45704162976085,
      "grad_norm": 0.3259412348270416,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 12935
    },
    {
      "epoch": 11.46147032772365,
      "grad_norm": 0.3386422395706177,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 12940
    },
    {
      "epoch": 11.465899025686449,
      "grad_norm": 0.28312399983406067,
      "learning_rate": 1e-05,
      "loss": 1.0378,
      "step": 12945
    },
    {
      "epoch": 11.470327723649246,
      "grad_norm": 0.25567251443862915,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 12950
    },
    {
      "epoch": 11.474756421612046,
      "grad_norm": 0.24410630762577057,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 12955
    },
    {
      "epoch": 11.479185119574845,
      "grad_norm": 0.28404220938682556,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 12960
    },
    {
      "epoch": 11.483613817537645,
      "grad_norm": 0.2507125437259674,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 12965
    },
    {
      "epoch": 11.488042515500442,
      "grad_norm": 0.286196768283844,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 12970
    },
    {
      "epoch": 11.492471213463242,
      "grad_norm": 0.2457255721092224,
      "learning_rate": 1e-05,
      "loss": 1.0308,
      "step": 12975
    },
    {
      "epoch": 11.496899911426041,
      "grad_norm": 0.2932203710079193,
      "learning_rate": 1e-05,
      "loss": 1.0113,
      "step": 12980
    },
    {
      "epoch": 11.501328609388839,
      "grad_norm": 0.2764868140220642,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 12985
    },
    {
      "epoch": 11.505757307351638,
      "grad_norm": 0.2516365349292755,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 12990
    },
    {
      "epoch": 11.510186005314438,
      "grad_norm": 0.2501620650291443,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 12995
    },
    {
      "epoch": 11.514614703277237,
      "grad_norm": 0.23135757446289062,
      "learning_rate": 1e-05,
      "loss": 1.0501,
      "step": 13000
    },
    {
      "epoch": 11.519043401240035,
      "grad_norm": 0.2551914155483246,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 13005
    },
    {
      "epoch": 11.523472099202834,
      "grad_norm": 0.30834686756134033,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 13010
    },
    {
      "epoch": 11.527900797165634,
      "grad_norm": 0.2901008427143097,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 13015
    },
    {
      "epoch": 11.532329495128431,
      "grad_norm": 0.3713749945163727,
      "learning_rate": 1e-05,
      "loss": 1.0256,
      "step": 13020
    },
    {
      "epoch": 11.53675819309123,
      "grad_norm": 0.25785550475120544,
      "learning_rate": 1e-05,
      "loss": 1.0338,
      "step": 13025
    },
    {
      "epoch": 11.54118689105403,
      "grad_norm": 0.28028014302253723,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 13030
    },
    {
      "epoch": 11.54561558901683,
      "grad_norm": 0.27772173285484314,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 13035
    },
    {
      "epoch": 11.550044286979627,
      "grad_norm": 0.2639353573322296,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 13040
    },
    {
      "epoch": 11.554472984942427,
      "grad_norm": 0.26178157329559326,
      "learning_rate": 1e-05,
      "loss": 1.0429,
      "step": 13045
    },
    {
      "epoch": 11.558901682905226,
      "grad_norm": 0.29230260848999023,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 13050
    },
    {
      "epoch": 11.563330380868024,
      "grad_norm": 0.27721700072288513,
      "learning_rate": 1e-05,
      "loss": 1.0588,
      "step": 13055
    },
    {
      "epoch": 11.567759078830823,
      "grad_norm": 0.30772387981414795,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 13060
    },
    {
      "epoch": 11.572187776793623,
      "grad_norm": 0.21189692616462708,
      "learning_rate": 1e-05,
      "loss": 1.0594,
      "step": 13065
    },
    {
      "epoch": 11.576616474756422,
      "grad_norm": 0.3092624545097351,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 13070
    },
    {
      "epoch": 11.58104517271922,
      "grad_norm": 0.24715109169483185,
      "learning_rate": 1e-05,
      "loss": 1.0549,
      "step": 13075
    },
    {
      "epoch": 11.58547387068202,
      "grad_norm": 0.25297677516937256,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 13080
    },
    {
      "epoch": 11.589902568644819,
      "grad_norm": 0.23887625336647034,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 13085
    },
    {
      "epoch": 11.594331266607618,
      "grad_norm": 0.31239053606987,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 13090
    },
    {
      "epoch": 11.598759964570416,
      "grad_norm": 0.2441648691892624,
      "learning_rate": 1e-05,
      "loss": 1.0524,
      "step": 13095
    },
    {
      "epoch": 11.603188662533215,
      "grad_norm": 0.22316597402095795,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 13100
    },
    {
      "epoch": 11.607617360496015,
      "grad_norm": 0.25960877537727356,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 13105
    },
    {
      "epoch": 11.612046058458812,
      "grad_norm": 0.2770078182220459,
      "learning_rate": 1e-05,
      "loss": 1.028,
      "step": 13110
    },
    {
      "epoch": 11.616474756421612,
      "grad_norm": 0.24256984889507294,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 13115
    },
    {
      "epoch": 11.620903454384411,
      "grad_norm": 0.2775229811668396,
      "learning_rate": 1e-05,
      "loss": 1.0283,
      "step": 13120
    },
    {
      "epoch": 11.62533215234721,
      "grad_norm": 0.21990017592906952,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 13125
    },
    {
      "epoch": 11.629760850310008,
      "grad_norm": 0.25390660762786865,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 13130
    },
    {
      "epoch": 11.634189548272808,
      "grad_norm": 0.27939173579216003,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 13135
    },
    {
      "epoch": 11.638618246235607,
      "grad_norm": 0.31400325894355774,
      "learning_rate": 1e-05,
      "loss": 1.0566,
      "step": 13140
    },
    {
      "epoch": 11.643046944198407,
      "grad_norm": 0.2791631519794464,
      "learning_rate": 1e-05,
      "loss": 1.0582,
      "step": 13145
    },
    {
      "epoch": 11.647475642161204,
      "grad_norm": 0.3174920976161957,
      "learning_rate": 1e-05,
      "loss": 1.0546,
      "step": 13150
    },
    {
      "epoch": 11.651904340124004,
      "grad_norm": 0.2516086995601654,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 13155
    },
    {
      "epoch": 11.656333038086803,
      "grad_norm": 0.33027520775794983,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 13160
    },
    {
      "epoch": 11.6607617360496,
      "grad_norm": 0.2980208992958069,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 13165
    },
    {
      "epoch": 11.6651904340124,
      "grad_norm": 0.25837254524230957,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 13170
    },
    {
      "epoch": 11.6696191319752,
      "grad_norm": 0.3473252058029175,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 13175
    },
    {
      "epoch": 11.674047829937999,
      "grad_norm": 0.2550885081291199,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 13180
    },
    {
      "epoch": 11.678476527900797,
      "grad_norm": 0.27462732791900635,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 13185
    },
    {
      "epoch": 11.682905225863596,
      "grad_norm": 0.336978018283844,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 13190
    },
    {
      "epoch": 11.687333923826396,
      "grad_norm": 0.2903445065021515,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 13195
    },
    {
      "epoch": 11.691762621789193,
      "grad_norm": 0.3348015248775482,
      "learning_rate": 1e-05,
      "loss": 1.0301,
      "step": 13200
    },
    {
      "epoch": 11.696191319751993,
      "grad_norm": 0.25575220584869385,
      "learning_rate": 1e-05,
      "loss": 1.0248,
      "step": 13205
    },
    {
      "epoch": 11.700620017714792,
      "grad_norm": 0.24518495798110962,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 13210
    },
    {
      "epoch": 11.705048715677592,
      "grad_norm": 0.3174123167991638,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 13215
    },
    {
      "epoch": 11.70947741364039,
      "grad_norm": 0.33312612771987915,
      "learning_rate": 1e-05,
      "loss": 0.9234,
      "step": 13220
    },
    {
      "epoch": 11.713906111603189,
      "grad_norm": 0.27455177903175354,
      "learning_rate": 1e-05,
      "loss": 1.0444,
      "step": 13225
    },
    {
      "epoch": 11.718334809565988,
      "grad_norm": 0.2920316159725189,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 13230
    },
    {
      "epoch": 11.722763507528786,
      "grad_norm": 0.28126898407936096,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 13235
    },
    {
      "epoch": 11.727192205491585,
      "grad_norm": 0.26298972964286804,
      "learning_rate": 1e-05,
      "loss": 1.046,
      "step": 13240
    },
    {
      "epoch": 11.731620903454385,
      "grad_norm": 0.27447426319122314,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 13245
    },
    {
      "epoch": 11.736049601417184,
      "grad_norm": 0.27569296956062317,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 13250
    },
    {
      "epoch": 11.740478299379982,
      "grad_norm": 0.28101348876953125,
      "learning_rate": 1e-05,
      "loss": 1.0523,
      "step": 13255
    },
    {
      "epoch": 11.744906997342781,
      "grad_norm": 0.28040438890457153,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 13260
    },
    {
      "epoch": 11.74933569530558,
      "grad_norm": 0.3497065007686615,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 13265
    },
    {
      "epoch": 11.75376439326838,
      "grad_norm": 0.26495182514190674,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 13270
    },
    {
      "epoch": 11.758193091231178,
      "grad_norm": 0.3278653621673584,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 13275
    },
    {
      "epoch": 11.762621789193977,
      "grad_norm": 0.26182252168655396,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 13280
    },
    {
      "epoch": 11.767050487156776,
      "grad_norm": 0.2797078490257263,
      "learning_rate": 1e-05,
      "loss": 1.022,
      "step": 13285
    },
    {
      "epoch": 11.771479185119574,
      "grad_norm": 0.2660171091556549,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 13290
    },
    {
      "epoch": 11.775907883082374,
      "grad_norm": 0.26958000659942627,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 13295
    },
    {
      "epoch": 11.780336581045173,
      "grad_norm": 0.3416246771812439,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 13300
    },
    {
      "epoch": 11.784765279007972,
      "grad_norm": 0.314456045627594,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 13305
    },
    {
      "epoch": 11.78919397697077,
      "grad_norm": 0.31488505005836487,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 13310
    },
    {
      "epoch": 11.79362267493357,
      "grad_norm": 0.25869429111480713,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 13315
    },
    {
      "epoch": 11.798051372896369,
      "grad_norm": 0.28862735629081726,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 13320
    },
    {
      "epoch": 11.802480070859167,
      "grad_norm": 0.2620314657688141,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 13325
    },
    {
      "epoch": 11.806908768821966,
      "grad_norm": 0.21097129583358765,
      "learning_rate": 1e-05,
      "loss": 1.0504,
      "step": 13330
    },
    {
      "epoch": 11.811337466784765,
      "grad_norm": 0.26400476694107056,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 13335
    },
    {
      "epoch": 11.815766164747565,
      "grad_norm": 0.2999187111854553,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 13340
    },
    {
      "epoch": 11.820194862710363,
      "grad_norm": 0.31398919224739075,
      "learning_rate": 1e-05,
      "loss": 1.0232,
      "step": 13345
    },
    {
      "epoch": 11.824623560673162,
      "grad_norm": 0.3374306559562683,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 13350
    },
    {
      "epoch": 11.829052258635961,
      "grad_norm": 0.2914050221443176,
      "learning_rate": 1e-05,
      "loss": 1.0418,
      "step": 13355
    },
    {
      "epoch": 11.833480956598759,
      "grad_norm": 0.3007832169532776,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 13360
    },
    {
      "epoch": 11.837909654561559,
      "grad_norm": 0.33798086643218994,
      "learning_rate": 1e-05,
      "loss": 1.0284,
      "step": 13365
    },
    {
      "epoch": 11.842338352524358,
      "grad_norm": 0.26915472745895386,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 13370
    },
    {
      "epoch": 11.846767050487157,
      "grad_norm": 0.2517496347427368,
      "learning_rate": 1e-05,
      "loss": 1.0505,
      "step": 13375
    },
    {
      "epoch": 11.851195748449955,
      "grad_norm": 0.2744987905025482,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 13380
    },
    {
      "epoch": 11.855624446412754,
      "grad_norm": 0.2502387762069702,
      "learning_rate": 1e-05,
      "loss": 1.0161,
      "step": 13385
    },
    {
      "epoch": 11.860053144375554,
      "grad_norm": 0.3029354512691498,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 13390
    },
    {
      "epoch": 11.864481842338353,
      "grad_norm": 0.2605747878551483,
      "learning_rate": 1e-05,
      "loss": 1.0526,
      "step": 13395
    },
    {
      "epoch": 11.868910540301151,
      "grad_norm": 0.29193899035453796,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 13400
    },
    {
      "epoch": 11.87333923826395,
      "grad_norm": 0.27779293060302734,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 13405
    },
    {
      "epoch": 11.87776793622675,
      "grad_norm": 0.3530624210834503,
      "learning_rate": 1e-05,
      "loss": 1.0221,
      "step": 13410
    },
    {
      "epoch": 11.882196634189548,
      "grad_norm": 0.31932714581489563,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 13415
    },
    {
      "epoch": 11.886625332152347,
      "grad_norm": 0.297229528427124,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 13420
    },
    {
      "epoch": 11.891054030115146,
      "grad_norm": 0.2655245363712311,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 13425
    },
    {
      "epoch": 11.895482728077946,
      "grad_norm": 0.2570720613002777,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 13430
    },
    {
      "epoch": 11.899911426040743,
      "grad_norm": 0.23089171946048737,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 13435
    },
    {
      "epoch": 11.904340124003543,
      "grad_norm": 0.23117727041244507,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 13440
    },
    {
      "epoch": 11.908768821966342,
      "grad_norm": 0.41261041164398193,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 13445
    },
    {
      "epoch": 11.91319751992914,
      "grad_norm": 0.3185042142868042,
      "learning_rate": 1e-05,
      "loss": 1.0346,
      "step": 13450
    },
    {
      "epoch": 11.91762621789194,
      "grad_norm": 0.26604071259498596,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 13455
    },
    {
      "epoch": 11.922054915854739,
      "grad_norm": 0.24834921956062317,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 13460
    },
    {
      "epoch": 11.926483613817538,
      "grad_norm": 0.3106094300746918,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 13465
    },
    {
      "epoch": 11.930912311780336,
      "grad_norm": 0.35537686944007874,
      "learning_rate": 1e-05,
      "loss": 1.0281,
      "step": 13470
    },
    {
      "epoch": 11.935341009743135,
      "grad_norm": 0.2999175488948822,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 13475
    },
    {
      "epoch": 11.939769707705935,
      "grad_norm": 0.2737196385860443,
      "learning_rate": 1e-05,
      "loss": 1.0263,
      "step": 13480
    },
    {
      "epoch": 11.944198405668732,
      "grad_norm": 0.2924957573413849,
      "learning_rate": 1e-05,
      "loss": 1.0422,
      "step": 13485
    },
    {
      "epoch": 11.948627103631532,
      "grad_norm": 0.2330625206232071,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 13490
    },
    {
      "epoch": 11.953055801594331,
      "grad_norm": 0.2882536053657532,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 13495
    },
    {
      "epoch": 11.95748449955713,
      "grad_norm": 0.2570204436779022,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 13500
    },
    {
      "epoch": 11.961913197519928,
      "grad_norm": 0.2935815155506134,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 13505
    },
    {
      "epoch": 11.966341895482728,
      "grad_norm": 0.277477890253067,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 13510
    },
    {
      "epoch": 11.970770593445527,
      "grad_norm": 0.34977397322654724,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 13515
    },
    {
      "epoch": 11.975199291408327,
      "grad_norm": 0.31164172291755676,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 13520
    },
    {
      "epoch": 11.979627989371124,
      "grad_norm": 0.24536538124084473,
      "learning_rate": 1e-05,
      "loss": 1.0473,
      "step": 13525
    },
    {
      "epoch": 11.984056687333924,
      "grad_norm": 0.2435460239648819,
      "learning_rate": 1e-05,
      "loss": 1.0224,
      "step": 13530
    },
    {
      "epoch": 11.988485385296723,
      "grad_norm": 0.2834871709346771,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 13535
    },
    {
      "epoch": 11.992914083259521,
      "grad_norm": 0.3505094349384308,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 13540
    },
    {
      "epoch": 11.99734278122232,
      "grad_norm": 0.24276615679264069,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 13545
    },
    {
      "epoch": 12.00177147918512,
      "grad_norm": 0.2683594524860382,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 13550
    },
    {
      "epoch": 12.00620017714792,
      "grad_norm": 0.2599976658821106,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 13555
    },
    {
      "epoch": 12.010628875110717,
      "grad_norm": 0.3061163127422333,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 13560
    },
    {
      "epoch": 12.015057573073516,
      "grad_norm": 0.26979687809944153,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 13565
    },
    {
      "epoch": 12.019486271036316,
      "grad_norm": 0.34937670826911926,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 13570
    },
    {
      "epoch": 12.023914968999113,
      "grad_norm": 0.2727797031402588,
      "learning_rate": 1e-05,
      "loss": 1.0185,
      "step": 13575
    },
    {
      "epoch": 12.028343666961913,
      "grad_norm": 0.27641206979751587,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 13580
    },
    {
      "epoch": 12.032772364924712,
      "grad_norm": 0.25684574246406555,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 13585
    },
    {
      "epoch": 12.037201062887512,
      "grad_norm": 0.26939576864242554,
      "learning_rate": 1e-05,
      "loss": 1.068,
      "step": 13590
    },
    {
      "epoch": 12.04162976085031,
      "grad_norm": 0.2836732268333435,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 13595
    },
    {
      "epoch": 12.046058458813109,
      "grad_norm": 0.3030366897583008,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 13600
    },
    {
      "epoch": 12.050487156775908,
      "grad_norm": 0.25969383120536804,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 13605
    },
    {
      "epoch": 12.054915854738708,
      "grad_norm": 0.25452321767807007,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 13610
    },
    {
      "epoch": 12.059344552701505,
      "grad_norm": 0.26574021577835083,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 13615
    },
    {
      "epoch": 12.063773250664305,
      "grad_norm": 0.2719283699989319,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 13620
    },
    {
      "epoch": 12.068201948627104,
      "grad_norm": 0.284067839384079,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 13625
    },
    {
      "epoch": 12.072630646589902,
      "grad_norm": 0.25207099318504333,
      "learning_rate": 1e-05,
      "loss": 1.0247,
      "step": 13630
    },
    {
      "epoch": 12.077059344552701,
      "grad_norm": 0.2667444944381714,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 13635
    },
    {
      "epoch": 12.0814880425155,
      "grad_norm": 0.2897314727306366,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 13640
    },
    {
      "epoch": 12.0859167404783,
      "grad_norm": 0.278303325176239,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 13645
    },
    {
      "epoch": 12.090345438441098,
      "grad_norm": 0.3381325602531433,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 13650
    },
    {
      "epoch": 12.094774136403897,
      "grad_norm": 0.3587614893913269,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 13655
    },
    {
      "epoch": 12.099202834366697,
      "grad_norm": 0.28937163949012756,
      "learning_rate": 1e-05,
      "loss": 1.0398,
      "step": 13660
    },
    {
      "epoch": 12.103631532329494,
      "grad_norm": 0.22299537062644958,
      "learning_rate": 1e-05,
      "loss": 1.0414,
      "step": 13665
    },
    {
      "epoch": 12.108060230292294,
      "grad_norm": 0.24265766143798828,
      "learning_rate": 1e-05,
      "loss": 1.0269,
      "step": 13670
    },
    {
      "epoch": 12.112488928255093,
      "grad_norm": 0.3386152386665344,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 13675
    },
    {
      "epoch": 12.116917626217893,
      "grad_norm": 0.258344441652298,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 13680
    },
    {
      "epoch": 12.12134632418069,
      "grad_norm": 0.23197130858898163,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 13685
    },
    {
      "epoch": 12.12577502214349,
      "grad_norm": 0.31310248374938965,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 13690
    },
    {
      "epoch": 12.130203720106289,
      "grad_norm": 0.2991909086704254,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 13695
    },
    {
      "epoch": 12.134632418069089,
      "grad_norm": 0.29839247465133667,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 13700
    },
    {
      "epoch": 12.139061116031886,
      "grad_norm": 0.27661699056625366,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 13705
    },
    {
      "epoch": 12.143489813994686,
      "grad_norm": 0.26128870248794556,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 13710
    },
    {
      "epoch": 12.147918511957485,
      "grad_norm": 0.2674477994441986,
      "learning_rate": 1e-05,
      "loss": 1.0581,
      "step": 13715
    },
    {
      "epoch": 12.152347209920283,
      "grad_norm": 0.3240382969379425,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 13720
    },
    {
      "epoch": 12.156775907883082,
      "grad_norm": 0.34261512756347656,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 13725
    },
    {
      "epoch": 12.161204605845882,
      "grad_norm": 0.3624170422554016,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 13730
    },
    {
      "epoch": 12.165633303808681,
      "grad_norm": 0.25599101185798645,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 13735
    },
    {
      "epoch": 12.170062001771479,
      "grad_norm": 0.31879791617393494,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 13740
    },
    {
      "epoch": 12.174490699734278,
      "grad_norm": 0.24751757085323334,
      "learning_rate": 1e-05,
      "loss": 1.0343,
      "step": 13745
    },
    {
      "epoch": 12.178919397697078,
      "grad_norm": 0.29482147097587585,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 13750
    },
    {
      "epoch": 12.183348095659875,
      "grad_norm": 0.2824556231498718,
      "learning_rate": 1e-05,
      "loss": 1.0231,
      "step": 13755
    },
    {
      "epoch": 12.187776793622675,
      "grad_norm": 0.25332191586494446,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 13760
    },
    {
      "epoch": 12.192205491585474,
      "grad_norm": 0.28341910243034363,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 13765
    },
    {
      "epoch": 12.196634189548273,
      "grad_norm": 0.25882938504219055,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 13770
    },
    {
      "epoch": 12.201062887511071,
      "grad_norm": 0.29963967204093933,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 13775
    },
    {
      "epoch": 12.20549158547387,
      "grad_norm": 0.35134243965148926,
      "learning_rate": 1e-05,
      "loss": 1.0217,
      "step": 13780
    },
    {
      "epoch": 12.20992028343667,
      "grad_norm": 0.26065540313720703,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 13785
    },
    {
      "epoch": 12.214348981399468,
      "grad_norm": 0.26940080523490906,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 13790
    },
    {
      "epoch": 12.218777679362267,
      "grad_norm": 0.3097437024116516,
      "learning_rate": 1e-05,
      "loss": 1.0479,
      "step": 13795
    },
    {
      "epoch": 12.223206377325067,
      "grad_norm": 0.2877049446105957,
      "learning_rate": 1e-05,
      "loss": 1.0317,
      "step": 13800
    },
    {
      "epoch": 12.227635075287866,
      "grad_norm": 0.2793222963809967,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 13805
    },
    {
      "epoch": 12.232063773250664,
      "grad_norm": 0.3042561411857605,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 13810
    },
    {
      "epoch": 12.236492471213463,
      "grad_norm": 0.36875292658805847,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 13815
    },
    {
      "epoch": 12.240921169176262,
      "grad_norm": 0.27395907044410706,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 13820
    },
    {
      "epoch": 12.245349867139062,
      "grad_norm": 0.23993048071861267,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 13825
    },
    {
      "epoch": 12.24977856510186,
      "grad_norm": 0.3032304644584656,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 13830
    },
    {
      "epoch": 12.254207263064659,
      "grad_norm": 0.29428017139434814,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 13835
    },
    {
      "epoch": 12.258635961027458,
      "grad_norm": 0.32754456996917725,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 13840
    },
    {
      "epoch": 12.263064658990256,
      "grad_norm": 0.29243528842926025,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 13845
    },
    {
      "epoch": 12.267493356953056,
      "grad_norm": 0.2800409197807312,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 13850
    },
    {
      "epoch": 12.271922054915855,
      "grad_norm": 0.22223135828971863,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 13855
    },
    {
      "epoch": 12.276350752878654,
      "grad_norm": 0.2942938506603241,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 13860
    },
    {
      "epoch": 12.280779450841452,
      "grad_norm": 0.27925005555152893,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 13865
    },
    {
      "epoch": 12.285208148804251,
      "grad_norm": 0.2256144881248474,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 13870
    },
    {
      "epoch": 12.289636846767051,
      "grad_norm": 0.26106294989585876,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 13875
    },
    {
      "epoch": 12.294065544729849,
      "grad_norm": 0.23196075856685638,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 13880
    },
    {
      "epoch": 12.298494242692648,
      "grad_norm": 0.24064002931118011,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 13885
    },
    {
      "epoch": 12.302922940655447,
      "grad_norm": 0.2666199207305908,
      "learning_rate": 1e-05,
      "loss": 1.0571,
      "step": 13890
    },
    {
      "epoch": 12.307351638618247,
      "grad_norm": 0.2753749489784241,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 13895
    },
    {
      "epoch": 12.311780336581045,
      "grad_norm": 0.2881872355937958,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 13900
    },
    {
      "epoch": 12.316209034543844,
      "grad_norm": 0.2434229552745819,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 13905
    },
    {
      "epoch": 12.320637732506643,
      "grad_norm": 0.2382996827363968,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 13910
    },
    {
      "epoch": 12.325066430469443,
      "grad_norm": 0.27602484822273254,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 13915
    },
    {
      "epoch": 12.32949512843224,
      "grad_norm": 0.28263893723487854,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 13920
    },
    {
      "epoch": 12.33392382639504,
      "grad_norm": 0.3976885974407196,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 13925
    },
    {
      "epoch": 12.33835252435784,
      "grad_norm": 0.2859059274196625,
      "learning_rate": 1e-05,
      "loss": 1.0181,
      "step": 13930
    },
    {
      "epoch": 12.342781222320637,
      "grad_norm": 0.35167238116264343,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 13935
    },
    {
      "epoch": 12.347209920283436,
      "grad_norm": 0.28873613476753235,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 13940
    },
    {
      "epoch": 12.351638618246236,
      "grad_norm": 0.30664119124412537,
      "learning_rate": 1e-05,
      "loss": 1.0172,
      "step": 13945
    },
    {
      "epoch": 12.356067316209035,
      "grad_norm": 0.3501007854938507,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 13950
    },
    {
      "epoch": 12.360496014171833,
      "grad_norm": 0.31964924931526184,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 13955
    },
    {
      "epoch": 12.364924712134632,
      "grad_norm": 0.39151525497436523,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 13960
    },
    {
      "epoch": 12.369353410097432,
      "grad_norm": 0.36976611614227295,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 13965
    },
    {
      "epoch": 12.37378210806023,
      "grad_norm": 0.3437179923057556,
      "learning_rate": 1e-05,
      "loss": 1.0485,
      "step": 13970
    },
    {
      "epoch": 12.378210806023029,
      "grad_norm": 0.2536727786064148,
      "learning_rate": 1e-05,
      "loss": 1.0469,
      "step": 13975
    },
    {
      "epoch": 12.382639503985828,
      "grad_norm": 0.2405978888273239,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 13980
    },
    {
      "epoch": 12.387068201948628,
      "grad_norm": 0.31488922238349915,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 13985
    },
    {
      "epoch": 12.391496899911425,
      "grad_norm": 0.25583669543266296,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 13990
    },
    {
      "epoch": 12.395925597874225,
      "grad_norm": 0.35331764817237854,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 13995
    },
    {
      "epoch": 12.400354295837024,
      "grad_norm": 0.29949167370796204,
      "learning_rate": 1e-05,
      "loss": 1.0232,
      "step": 14000
    },
    {
      "epoch": 12.404782993799824,
      "grad_norm": 0.3051189184188843,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 14005
    },
    {
      "epoch": 12.409211691762621,
      "grad_norm": 0.29661527276039124,
      "learning_rate": 1e-05,
      "loss": 1.025,
      "step": 14010
    },
    {
      "epoch": 12.41364038972542,
      "grad_norm": 0.3046285808086395,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 14015
    },
    {
      "epoch": 12.41806908768822,
      "grad_norm": 0.2921496331691742,
      "learning_rate": 1e-05,
      "loss": 1.0285,
      "step": 14020
    },
    {
      "epoch": 12.422497785651018,
      "grad_norm": 0.28083136677742004,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 14025
    },
    {
      "epoch": 12.426926483613817,
      "grad_norm": 0.27116382122039795,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 14030
    },
    {
      "epoch": 12.431355181576617,
      "grad_norm": 0.32082560658454895,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 14035
    },
    {
      "epoch": 12.435783879539416,
      "grad_norm": 0.30781230330467224,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 14040
    },
    {
      "epoch": 12.440212577502214,
      "grad_norm": 0.26844415068626404,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 14045
    },
    {
      "epoch": 12.444641275465013,
      "grad_norm": 0.2612963020801544,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 14050
    },
    {
      "epoch": 12.449069973427813,
      "grad_norm": 0.2543257772922516,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 14055
    },
    {
      "epoch": 12.45349867139061,
      "grad_norm": 0.2535167336463928,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 14060
    },
    {
      "epoch": 12.45792736935341,
      "grad_norm": 0.30882173776626587,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 14065
    },
    {
      "epoch": 12.46235606731621,
      "grad_norm": 0.32208821177482605,
      "learning_rate": 1e-05,
      "loss": 1.034,
      "step": 14070
    },
    {
      "epoch": 12.466784765279009,
      "grad_norm": 0.35279417037963867,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 14075
    },
    {
      "epoch": 12.471213463241806,
      "grad_norm": 0.37035006284713745,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 14080
    },
    {
      "epoch": 12.475642161204606,
      "grad_norm": 0.33337658643722534,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 14085
    },
    {
      "epoch": 12.480070859167405,
      "grad_norm": 0.28878286480903625,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 14090
    },
    {
      "epoch": 12.484499557130203,
      "grad_norm": 0.2391795814037323,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 14095
    },
    {
      "epoch": 12.488928255093002,
      "grad_norm": 0.3022184669971466,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 14100
    },
    {
      "epoch": 12.493356953055802,
      "grad_norm": 0.27281835675239563,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 14105
    },
    {
      "epoch": 12.497785651018601,
      "grad_norm": 0.26640427112579346,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 14110
    },
    {
      "epoch": 12.502214348981399,
      "grad_norm": 0.26127880811691284,
      "learning_rate": 1e-05,
      "loss": 1.0375,
      "step": 14115
    },
    {
      "epoch": 12.506643046944198,
      "grad_norm": 0.2945282459259033,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 14120
    },
    {
      "epoch": 12.511071744906998,
      "grad_norm": 0.2432110607624054,
      "learning_rate": 1e-05,
      "loss": 1.0182,
      "step": 14125
    },
    {
      "epoch": 12.515500442869797,
      "grad_norm": 0.24952492117881775,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 14130
    },
    {
      "epoch": 12.519929140832595,
      "grad_norm": 0.26578807830810547,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 14135
    },
    {
      "epoch": 12.524357838795394,
      "grad_norm": 0.2699507772922516,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 14140
    },
    {
      "epoch": 12.528786536758194,
      "grad_norm": 0.30454644560813904,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 14145
    },
    {
      "epoch": 12.533215234720991,
      "grad_norm": 0.38640955090522766,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 14150
    },
    {
      "epoch": 12.53764393268379,
      "grad_norm": 0.33499661087989807,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 14155
    },
    {
      "epoch": 12.54207263064659,
      "grad_norm": 0.32563361525535583,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 14160
    },
    {
      "epoch": 12.54650132860939,
      "grad_norm": 0.30377498269081116,
      "learning_rate": 1e-05,
      "loss": 1.0553,
      "step": 14165
    },
    {
      "epoch": 12.550930026572187,
      "grad_norm": 0.29143550992012024,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 14170
    },
    {
      "epoch": 12.555358724534987,
      "grad_norm": 0.24909621477127075,
      "learning_rate": 1e-05,
      "loss": 1.024,
      "step": 14175
    },
    {
      "epoch": 12.559787422497786,
      "grad_norm": 0.2271912693977356,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 14180
    },
    {
      "epoch": 12.564216120460584,
      "grad_norm": 0.2555599510669708,
      "learning_rate": 1e-05,
      "loss": 1.0473,
      "step": 14185
    },
    {
      "epoch": 12.568644818423383,
      "grad_norm": 0.23209033906459808,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 14190
    },
    {
      "epoch": 12.573073516386183,
      "grad_norm": 0.2560659348964691,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 14195
    },
    {
      "epoch": 12.577502214348982,
      "grad_norm": 0.3476274907588959,
      "learning_rate": 1e-05,
      "loss": 1.0273,
      "step": 14200
    },
    {
      "epoch": 12.58193091231178,
      "grad_norm": 0.2799437940120697,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 14205
    },
    {
      "epoch": 12.58635961027458,
      "grad_norm": 0.30695584416389465,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 14210
    },
    {
      "epoch": 12.590788308237379,
      "grad_norm": 0.2842298746109009,
      "learning_rate": 1e-05,
      "loss": 1.0303,
      "step": 14215
    },
    {
      "epoch": 12.595217006200176,
      "grad_norm": 0.26325708627700806,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 14220
    },
    {
      "epoch": 12.599645704162976,
      "grad_norm": 0.36153414845466614,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 14225
    },
    {
      "epoch": 12.604074402125775,
      "grad_norm": 0.2362048476934433,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 14230
    },
    {
      "epoch": 12.608503100088575,
      "grad_norm": 0.22038434445858002,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 14235
    },
    {
      "epoch": 12.612931798051372,
      "grad_norm": 0.29843786358833313,
      "learning_rate": 1e-05,
      "loss": 1.0414,
      "step": 14240
    },
    {
      "epoch": 12.617360496014172,
      "grad_norm": 0.3027302026748657,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 14245
    },
    {
      "epoch": 12.621789193976971,
      "grad_norm": 0.301037460565567,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 14250
    },
    {
      "epoch": 12.62621789193977,
      "grad_norm": 0.2412690669298172,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 14255
    },
    {
      "epoch": 12.630646589902568,
      "grad_norm": 0.2833200693130493,
      "learning_rate": 1e-05,
      "loss": 1.0354,
      "step": 14260
    },
    {
      "epoch": 12.635075287865368,
      "grad_norm": 0.31504499912261963,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 14265
    },
    {
      "epoch": 12.639503985828167,
      "grad_norm": 0.32337433099746704,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 14270
    },
    {
      "epoch": 12.643932683790965,
      "grad_norm": 0.2757129967212677,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 14275
    },
    {
      "epoch": 12.648361381753764,
      "grad_norm": 0.21794089674949646,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 14280
    },
    {
      "epoch": 12.652790079716564,
      "grad_norm": 0.31632256507873535,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 14285
    },
    {
      "epoch": 12.657218777679363,
      "grad_norm": 0.28485068678855896,
      "learning_rate": 1e-05,
      "loss": 1.0257,
      "step": 14290
    },
    {
      "epoch": 12.66164747564216,
      "grad_norm": 0.28050705790519714,
      "learning_rate": 1e-05,
      "loss": 1.0317,
      "step": 14295
    },
    {
      "epoch": 12.66607617360496,
      "grad_norm": 0.31393176317214966,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 14300
    },
    {
      "epoch": 12.67050487156776,
      "grad_norm": 0.3161180317401886,
      "learning_rate": 1e-05,
      "loss": 1.0295,
      "step": 14305
    },
    {
      "epoch": 12.674933569530559,
      "grad_norm": 0.3357236683368683,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 14310
    },
    {
      "epoch": 12.679362267493357,
      "grad_norm": 0.27490562200546265,
      "learning_rate": 1e-05,
      "loss": 1.0383,
      "step": 14315
    },
    {
      "epoch": 12.683790965456156,
      "grad_norm": 0.29009804129600525,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 14320
    },
    {
      "epoch": 12.688219663418955,
      "grad_norm": 0.34791311621665955,
      "learning_rate": 1e-05,
      "loss": 1.0633,
      "step": 14325
    },
    {
      "epoch": 12.692648361381753,
      "grad_norm": 0.27698037028312683,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 14330
    },
    {
      "epoch": 12.697077059344553,
      "grad_norm": 0.23974473774433136,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 14335
    },
    {
      "epoch": 12.701505757307352,
      "grad_norm": 0.25287118554115295,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 14340
    },
    {
      "epoch": 12.70593445527015,
      "grad_norm": 0.2530422806739807,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 14345
    },
    {
      "epoch": 12.710363153232949,
      "grad_norm": 0.26218244433403015,
      "learning_rate": 1e-05,
      "loss": 1.0155,
      "step": 14350
    },
    {
      "epoch": 12.714791851195749,
      "grad_norm": 0.26334822177886963,
      "learning_rate": 1e-05,
      "loss": 1.0397,
      "step": 14355
    },
    {
      "epoch": 12.719220549158548,
      "grad_norm": 0.24559754133224487,
      "learning_rate": 1e-05,
      "loss": 1.0172,
      "step": 14360
    },
    {
      "epoch": 12.723649247121346,
      "grad_norm": 0.24789783358573914,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 14365
    },
    {
      "epoch": 12.728077945084145,
      "grad_norm": 0.25581347942352295,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 14370
    },
    {
      "epoch": 12.732506643046944,
      "grad_norm": 0.33454272150993347,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 14375
    },
    {
      "epoch": 12.736935341009744,
      "grad_norm": 0.2632225751876831,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 14380
    },
    {
      "epoch": 12.741364038972542,
      "grad_norm": 0.26623839139938354,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 14385
    },
    {
      "epoch": 12.745792736935341,
      "grad_norm": 0.2904810905456543,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 14390
    },
    {
      "epoch": 12.75022143489814,
      "grad_norm": 0.3001648485660553,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 14395
    },
    {
      "epoch": 12.754650132860938,
      "grad_norm": 0.27608737349510193,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 14400
    },
    {
      "epoch": 12.759078830823738,
      "grad_norm": 0.2190631926059723,
      "learning_rate": 1e-05,
      "loss": 1.0343,
      "step": 14405
    },
    {
      "epoch": 12.763507528786537,
      "grad_norm": 0.23140758275985718,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 14410
    },
    {
      "epoch": 12.767936226749336,
      "grad_norm": 0.24402092397212982,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 14415
    },
    {
      "epoch": 12.772364924712134,
      "grad_norm": 0.26170435547828674,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 14420
    },
    {
      "epoch": 12.776793622674933,
      "grad_norm": 0.2686377763748169,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 14425
    },
    {
      "epoch": 12.781222320637733,
      "grad_norm": 0.36198344826698303,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 14430
    },
    {
      "epoch": 12.785651018600532,
      "grad_norm": 0.3081618547439575,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 14435
    },
    {
      "epoch": 12.79007971656333,
      "grad_norm": 0.2812959849834442,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 14440
    },
    {
      "epoch": 12.79450841452613,
      "grad_norm": 0.42400217056274414,
      "learning_rate": 1e-05,
      "loss": 1.0262,
      "step": 14445
    },
    {
      "epoch": 12.798937112488929,
      "grad_norm": 0.30481165647506714,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 14450
    },
    {
      "epoch": 12.803365810451727,
      "grad_norm": 0.28901466727256775,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 14455
    },
    {
      "epoch": 12.807794508414526,
      "grad_norm": 0.2570507228374481,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 14460
    },
    {
      "epoch": 12.812223206377325,
      "grad_norm": 0.2827906012535095,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 14465
    },
    {
      "epoch": 12.816651904340125,
      "grad_norm": 0.2440817803144455,
      "learning_rate": 1e-05,
      "loss": 1.031,
      "step": 14470
    },
    {
      "epoch": 12.821080602302922,
      "grad_norm": 0.3164633810520172,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 14475
    },
    {
      "epoch": 12.825509300265722,
      "grad_norm": 0.25699707865715027,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 14480
    },
    {
      "epoch": 12.829937998228521,
      "grad_norm": 0.2710092067718506,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 14485
    },
    {
      "epoch": 12.834366696191319,
      "grad_norm": 0.2854827344417572,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 14490
    },
    {
      "epoch": 12.838795394154118,
      "grad_norm": 0.24846535921096802,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 14495
    },
    {
      "epoch": 12.843224092116918,
      "grad_norm": 0.29199254512786865,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 14500
    },
    {
      "epoch": 12.847652790079717,
      "grad_norm": 0.24273204803466797,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 14505
    },
    {
      "epoch": 12.852081488042515,
      "grad_norm": 0.30488133430480957,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 14510
    },
    {
      "epoch": 12.856510186005314,
      "grad_norm": 0.27750492095947266,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 14515
    },
    {
      "epoch": 12.860938883968114,
      "grad_norm": 0.2598741054534912,
      "learning_rate": 1e-05,
      "loss": 1.0818,
      "step": 14520
    },
    {
      "epoch": 12.865367581930911,
      "grad_norm": 0.26312053203582764,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 14525
    },
    {
      "epoch": 12.869796279893711,
      "grad_norm": 0.3174665868282318,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 14530
    },
    {
      "epoch": 12.87422497785651,
      "grad_norm": 0.2978462278842926,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 14535
    },
    {
      "epoch": 12.87865367581931,
      "grad_norm": 0.26777568459510803,
      "learning_rate": 1e-05,
      "loss": 1.0483,
      "step": 14540
    },
    {
      "epoch": 12.883082373782107,
      "grad_norm": 0.32283392548561096,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 14545
    },
    {
      "epoch": 12.887511071744907,
      "grad_norm": 0.2876666784286499,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 14550
    },
    {
      "epoch": 12.891939769707706,
      "grad_norm": 0.2791752219200134,
      "learning_rate": 1e-05,
      "loss": 1.0175,
      "step": 14555
    },
    {
      "epoch": 12.896368467670506,
      "grad_norm": 0.26223334670066833,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 14560
    },
    {
      "epoch": 12.900797165633303,
      "grad_norm": 0.29784226417541504,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 14565
    },
    {
      "epoch": 12.905225863596103,
      "grad_norm": 0.2935771644115448,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 14570
    },
    {
      "epoch": 12.909654561558902,
      "grad_norm": 0.3291126787662506,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 14575
    },
    {
      "epoch": 12.9140832595217,
      "grad_norm": 0.2663058042526245,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 14580
    },
    {
      "epoch": 12.9185119574845,
      "grad_norm": 0.321157842874527,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 14585
    },
    {
      "epoch": 12.922940655447299,
      "grad_norm": 0.28205373883247375,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 14590
    },
    {
      "epoch": 12.927369353410098,
      "grad_norm": 0.3134406805038452,
      "learning_rate": 1e-05,
      "loss": 1.0195,
      "step": 14595
    },
    {
      "epoch": 12.931798051372896,
      "grad_norm": 0.3107638359069824,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 14600
    },
    {
      "epoch": 12.936226749335695,
      "grad_norm": 0.3422958552837372,
      "learning_rate": 1e-05,
      "loss": 0.9251,
      "step": 14605
    },
    {
      "epoch": 12.940655447298495,
      "grad_norm": 0.2904197573661804,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 14610
    },
    {
      "epoch": 12.945084145261292,
      "grad_norm": 0.25404492020606995,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 14615
    },
    {
      "epoch": 12.949512843224092,
      "grad_norm": 0.2975197732448578,
      "learning_rate": 1e-05,
      "loss": 1.0371,
      "step": 14620
    },
    {
      "epoch": 12.953941541186891,
      "grad_norm": 0.28089869022369385,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 14625
    },
    {
      "epoch": 12.95837023914969,
      "grad_norm": 0.2694743871688843,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 14630
    },
    {
      "epoch": 12.962798937112488,
      "grad_norm": 0.24171431362628937,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 14635
    },
    {
      "epoch": 12.967227635075288,
      "grad_norm": 0.3107065260410309,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 14640
    },
    {
      "epoch": 12.971656333038087,
      "grad_norm": 0.24199900031089783,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 14645
    },
    {
      "epoch": 12.976085031000885,
      "grad_norm": 0.28428131341934204,
      "learning_rate": 1e-05,
      "loss": 0.9283,
      "step": 14650
    },
    {
      "epoch": 12.980513728963684,
      "grad_norm": 0.3267878293991089,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 14655
    },
    {
      "epoch": 12.984942426926484,
      "grad_norm": 0.27152273058891296,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 14660
    },
    {
      "epoch": 12.989371124889283,
      "grad_norm": 0.26075512170791626,
      "learning_rate": 1e-05,
      "loss": 1.038,
      "step": 14665
    },
    {
      "epoch": 12.99379982285208,
      "grad_norm": 0.25247660279273987,
      "learning_rate": 1e-05,
      "loss": 1.0845,
      "step": 14670
    },
    {
      "epoch": 12.99822852081488,
      "grad_norm": 0.2509274482727051,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 14675
    },
    {
      "epoch": 13.00265721877768,
      "grad_norm": 0.2458716183900833,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 14680
    },
    {
      "epoch": 13.007085916740479,
      "grad_norm": 0.29686760902404785,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 14685
    },
    {
      "epoch": 13.011514614703277,
      "grad_norm": 0.23911021649837494,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 14690
    },
    {
      "epoch": 13.015943312666076,
      "grad_norm": 0.32030215859413147,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 14695
    },
    {
      "epoch": 13.020372010628876,
      "grad_norm": 0.3203235864639282,
      "learning_rate": 1e-05,
      "loss": 1.0365,
      "step": 14700
    },
    {
      "epoch": 13.024800708591673,
      "grad_norm": 0.2907693386077881,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 14705
    },
    {
      "epoch": 13.029229406554473,
      "grad_norm": 0.26345759630203247,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 14710
    },
    {
      "epoch": 13.033658104517272,
      "grad_norm": 0.2642922103404999,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 14715
    },
    {
      "epoch": 13.038086802480072,
      "grad_norm": 0.25867316126823425,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 14720
    },
    {
      "epoch": 13.04251550044287,
      "grad_norm": 0.28513777256011963,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 14725
    },
    {
      "epoch": 13.046944198405669,
      "grad_norm": 0.24561457335948944,
      "learning_rate": 1e-05,
      "loss": 1.0285,
      "step": 14730
    },
    {
      "epoch": 13.051372896368468,
      "grad_norm": 0.27027419209480286,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 14735
    },
    {
      "epoch": 13.055801594331266,
      "grad_norm": 0.28715813159942627,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 14740
    },
    {
      "epoch": 13.060230292294065,
      "grad_norm": 0.2848362624645233,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 14745
    },
    {
      "epoch": 13.064658990256865,
      "grad_norm": 0.2359524369239807,
      "learning_rate": 1e-05,
      "loss": 1.0312,
      "step": 14750
    },
    {
      "epoch": 13.069087688219664,
      "grad_norm": 0.2665725648403168,
      "learning_rate": 1e-05,
      "loss": 1.0897,
      "step": 14755
    },
    {
      "epoch": 13.073516386182462,
      "grad_norm": 0.2580292224884033,
      "learning_rate": 1e-05,
      "loss": 1.0357,
      "step": 14760
    },
    {
      "epoch": 13.077945084145261,
      "grad_norm": 0.2622484564781189,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 14765
    },
    {
      "epoch": 13.08237378210806,
      "grad_norm": 0.2700832784175873,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 14770
    },
    {
      "epoch": 13.08680248007086,
      "grad_norm": 0.285881906747818,
      "learning_rate": 1e-05,
      "loss": 1.057,
      "step": 14775
    },
    {
      "epoch": 13.091231178033658,
      "grad_norm": 0.33594414591789246,
      "learning_rate": 1e-05,
      "loss": 0.9121,
      "step": 14780
    },
    {
      "epoch": 13.095659875996457,
      "grad_norm": 0.23105424642562866,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 14785
    },
    {
      "epoch": 13.100088573959257,
      "grad_norm": 0.2254118025302887,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 14790
    },
    {
      "epoch": 13.104517271922054,
      "grad_norm": 0.29767879843711853,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 14795
    },
    {
      "epoch": 13.108945969884854,
      "grad_norm": 0.2523473799228668,
      "learning_rate": 1e-05,
      "loss": 1.0359,
      "step": 14800
    },
    {
      "epoch": 13.113374667847653,
      "grad_norm": 0.2674669623374939,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 14805
    },
    {
      "epoch": 13.117803365810452,
      "grad_norm": 0.2697570025920868,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 14810
    },
    {
      "epoch": 13.12223206377325,
      "grad_norm": 0.2735148072242737,
      "learning_rate": 1e-05,
      "loss": 1.039,
      "step": 14815
    },
    {
      "epoch": 13.12666076173605,
      "grad_norm": 0.28674471378326416,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 14820
    },
    {
      "epoch": 13.131089459698849,
      "grad_norm": 0.290833055973053,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 14825
    },
    {
      "epoch": 13.135518157661647,
      "grad_norm": 0.29840123653411865,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 14830
    },
    {
      "epoch": 13.139946855624446,
      "grad_norm": 0.2405647486448288,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 14835
    },
    {
      "epoch": 13.144375553587246,
      "grad_norm": 0.3389081060886383,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 14840
    },
    {
      "epoch": 13.148804251550045,
      "grad_norm": 0.30209752917289734,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 14845
    },
    {
      "epoch": 13.153232949512843,
      "grad_norm": 0.25667551159858704,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 14850
    },
    {
      "epoch": 13.157661647475642,
      "grad_norm": 0.2599428594112396,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 14855
    },
    {
      "epoch": 13.162090345438441,
      "grad_norm": 0.2966095507144928,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 14860
    },
    {
      "epoch": 13.166519043401241,
      "grad_norm": 0.2184838205575943,
      "learning_rate": 1e-05,
      "loss": 1.0457,
      "step": 14865
    },
    {
      "epoch": 13.170947741364039,
      "grad_norm": 0.2505369782447815,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 14870
    },
    {
      "epoch": 13.175376439326838,
      "grad_norm": 0.2525484561920166,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 14875
    },
    {
      "epoch": 13.179805137289637,
      "grad_norm": 0.2752380967140198,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 14880
    },
    {
      "epoch": 13.184233835252435,
      "grad_norm": 0.2934929430484772,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 14885
    },
    {
      "epoch": 13.188662533215235,
      "grad_norm": 0.29155921936035156,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 14890
    },
    {
      "epoch": 13.193091231178034,
      "grad_norm": 0.278621643781662,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 14895
    },
    {
      "epoch": 13.197519929140833,
      "grad_norm": 0.2534531354904175,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 14900
    },
    {
      "epoch": 13.201948627103631,
      "grad_norm": 0.2333713173866272,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 14905
    },
    {
      "epoch": 13.20637732506643,
      "grad_norm": 0.2530066967010498,
      "learning_rate": 1e-05,
      "loss": 1.0505,
      "step": 14910
    },
    {
      "epoch": 13.21080602302923,
      "grad_norm": 0.24350881576538086,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 14915
    },
    {
      "epoch": 13.215234720992028,
      "grad_norm": 0.2508932054042816,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 14920
    },
    {
      "epoch": 13.219663418954827,
      "grad_norm": 0.25571197271347046,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 14925
    },
    {
      "epoch": 13.224092116917626,
      "grad_norm": 0.2576509416103363,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 14930
    },
    {
      "epoch": 13.228520814880426,
      "grad_norm": 0.2673387825489044,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 14935
    },
    {
      "epoch": 13.232949512843224,
      "grad_norm": 0.29158371686935425,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 14940
    },
    {
      "epoch": 13.237378210806023,
      "grad_norm": 0.26073262095451355,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 14945
    },
    {
      "epoch": 13.241806908768822,
      "grad_norm": 0.28596410155296326,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 14950
    },
    {
      "epoch": 13.24623560673162,
      "grad_norm": 0.28583839535713196,
      "learning_rate": 1e-05,
      "loss": 1.0221,
      "step": 14955
    },
    {
      "epoch": 13.25066430469442,
      "grad_norm": 0.3165747821331024,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 14960
    },
    {
      "epoch": 13.255093002657219,
      "grad_norm": 0.3266614079475403,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 14965
    },
    {
      "epoch": 13.259521700620018,
      "grad_norm": 0.25403618812561035,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 14970
    },
    {
      "epoch": 13.263950398582816,
      "grad_norm": 0.3080897331237793,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 14975
    },
    {
      "epoch": 13.268379096545615,
      "grad_norm": 0.253658264875412,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 14980
    },
    {
      "epoch": 13.272807794508415,
      "grad_norm": 0.26744991540908813,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 14985
    },
    {
      "epoch": 13.277236492471214,
      "grad_norm": 0.3480512201786041,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 14990
    },
    {
      "epoch": 13.281665190434012,
      "grad_norm": 0.4289158880710602,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 14995
    },
    {
      "epoch": 13.286093888396811,
      "grad_norm": 0.28872525691986084,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 15000
    },
    {
      "epoch": 13.29052258635961,
      "grad_norm": 0.3190418481826782,
      "learning_rate": 1e-05,
      "loss": 1.0054,
      "step": 15005
    },
    {
      "epoch": 13.294951284322408,
      "grad_norm": 0.3040004074573517,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 15010
    },
    {
      "epoch": 13.299379982285208,
      "grad_norm": 0.32927224040031433,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 15015
    },
    {
      "epoch": 13.303808680248007,
      "grad_norm": 0.26025819778442383,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 15020
    },
    {
      "epoch": 13.308237378210807,
      "grad_norm": 0.24589228630065918,
      "learning_rate": 1e-05,
      "loss": 1.0233,
      "step": 15025
    },
    {
      "epoch": 13.312666076173604,
      "grad_norm": 0.24863485991954803,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 15030
    },
    {
      "epoch": 13.317094774136404,
      "grad_norm": 0.2284671813249588,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 15035
    },
    {
      "epoch": 13.321523472099203,
      "grad_norm": 0.32172033190727234,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 15040
    },
    {
      "epoch": 13.325952170062001,
      "grad_norm": 0.2671842575073242,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 15045
    },
    {
      "epoch": 13.3303808680248,
      "grad_norm": 0.2727336585521698,
      "learning_rate": 1e-05,
      "loss": 1.0107,
      "step": 15050
    },
    {
      "epoch": 13.3348095659876,
      "grad_norm": 0.2501024901866913,
      "learning_rate": 1e-05,
      "loss": 1.0398,
      "step": 15055
    },
    {
      "epoch": 13.3392382639504,
      "grad_norm": 0.29526761174201965,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 15060
    },
    {
      "epoch": 13.343666961913197,
      "grad_norm": 0.27258872985839844,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 15065
    },
    {
      "epoch": 13.348095659875996,
      "grad_norm": 0.2876748740673065,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 15070
    },
    {
      "epoch": 13.352524357838796,
      "grad_norm": 0.29867053031921387,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 15075
    },
    {
      "epoch": 13.356953055801593,
      "grad_norm": 0.33151668310165405,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 15080
    },
    {
      "epoch": 13.361381753764393,
      "grad_norm": 0.2648833692073822,
      "learning_rate": 1e-05,
      "loss": 1.0728,
      "step": 15085
    },
    {
      "epoch": 13.365810451727192,
      "grad_norm": 0.2855778932571411,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 15090
    },
    {
      "epoch": 13.370239149689992,
      "grad_norm": 0.268180787563324,
      "learning_rate": 1e-05,
      "loss": 1.0241,
      "step": 15095
    },
    {
      "epoch": 13.37466784765279,
      "grad_norm": 0.23653945326805115,
      "learning_rate": 1e-05,
      "loss": 1.0189,
      "step": 15100
    },
    {
      "epoch": 13.379096545615589,
      "grad_norm": 0.2820050120353699,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 15105
    },
    {
      "epoch": 13.383525243578388,
      "grad_norm": 0.2409396767616272,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 15110
    },
    {
      "epoch": 13.387953941541188,
      "grad_norm": 0.3344702422618866,
      "learning_rate": 1e-05,
      "loss": 1.0461,
      "step": 15115
    },
    {
      "epoch": 13.392382639503985,
      "grad_norm": 0.2607822120189667,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 15120
    },
    {
      "epoch": 13.396811337466785,
      "grad_norm": 0.24322141706943512,
      "learning_rate": 1e-05,
      "loss": 1.0225,
      "step": 15125
    },
    {
      "epoch": 13.401240035429584,
      "grad_norm": 0.27175086736679077,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 15130
    },
    {
      "epoch": 13.405668733392382,
      "grad_norm": 0.3235015869140625,
      "learning_rate": 1e-05,
      "loss": 1.0584,
      "step": 15135
    },
    {
      "epoch": 13.410097431355181,
      "grad_norm": 0.2846199572086334,
      "learning_rate": 1e-05,
      "loss": 1.0427,
      "step": 15140
    },
    {
      "epoch": 13.41452612931798,
      "grad_norm": 0.2673349976539612,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 15145
    },
    {
      "epoch": 13.41895482728078,
      "grad_norm": 0.22861461341381073,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 15150
    },
    {
      "epoch": 13.423383525243578,
      "grad_norm": 0.22525490820407867,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 15155
    },
    {
      "epoch": 13.427812223206377,
      "grad_norm": 0.2689531147480011,
      "learning_rate": 1e-05,
      "loss": 1.0252,
      "step": 15160
    },
    {
      "epoch": 13.432240921169177,
      "grad_norm": 0.2856645882129669,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 15165
    },
    {
      "epoch": 13.436669619131976,
      "grad_norm": 0.23056024312973022,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 15170
    },
    {
      "epoch": 13.441098317094774,
      "grad_norm": 0.24246810376644135,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 15175
    },
    {
      "epoch": 13.445527015057573,
      "grad_norm": 0.23817594349384308,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 15180
    },
    {
      "epoch": 13.449955713020373,
      "grad_norm": 0.23851200938224792,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 15185
    },
    {
      "epoch": 13.45438441098317,
      "grad_norm": 0.2606596350669861,
      "learning_rate": 1e-05,
      "loss": 1.0776,
      "step": 15190
    },
    {
      "epoch": 13.45881310894597,
      "grad_norm": 0.2664880156517029,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 15195
    },
    {
      "epoch": 13.46324180690877,
      "grad_norm": 0.32686474919319153,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 15200
    },
    {
      "epoch": 13.467670504871569,
      "grad_norm": 0.2740435004234314,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 15205
    },
    {
      "epoch": 13.472099202834366,
      "grad_norm": 0.2651437819004059,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 15210
    },
    {
      "epoch": 13.476527900797166,
      "grad_norm": 0.2624920904636383,
      "learning_rate": 1e-05,
      "loss": 1.0268,
      "step": 15215
    },
    {
      "epoch": 13.480956598759965,
      "grad_norm": 0.29028382897377014,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 15220
    },
    {
      "epoch": 13.485385296722763,
      "grad_norm": 0.2924567461013794,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 15225
    },
    {
      "epoch": 13.489813994685562,
      "grad_norm": 0.29711809754371643,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 15230
    },
    {
      "epoch": 13.494242692648362,
      "grad_norm": 0.2705939710140228,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 15235
    },
    {
      "epoch": 13.498671390611161,
      "grad_norm": 0.240801140666008,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 15240
    },
    {
      "epoch": 13.503100088573959,
      "grad_norm": 0.30542322993278503,
      "learning_rate": 1e-05,
      "loss": 1.0561,
      "step": 15245
    },
    {
      "epoch": 13.507528786536758,
      "grad_norm": 0.33220988512039185,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 15250
    },
    {
      "epoch": 13.511957484499558,
      "grad_norm": 0.2919374406337738,
      "learning_rate": 1e-05,
      "loss": 1.0353,
      "step": 15255
    },
    {
      "epoch": 13.516386182462355,
      "grad_norm": 0.30203157663345337,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 15260
    },
    {
      "epoch": 13.520814880425155,
      "grad_norm": 0.2783576250076294,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 15265
    },
    {
      "epoch": 13.525243578387954,
      "grad_norm": 0.26149386167526245,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 15270
    },
    {
      "epoch": 13.529672276350754,
      "grad_norm": 0.3352258503437042,
      "learning_rate": 1e-05,
      "loss": 1.0392,
      "step": 15275
    },
    {
      "epoch": 13.534100974313551,
      "grad_norm": 0.25567755103111267,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 15280
    },
    {
      "epoch": 13.53852967227635,
      "grad_norm": 0.24535495042800903,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 15285
    },
    {
      "epoch": 13.54295837023915,
      "grad_norm": 0.3483872413635254,
      "learning_rate": 1e-05,
      "loss": 1.0367,
      "step": 15290
    },
    {
      "epoch": 13.54738706820195,
      "grad_norm": 0.27215635776519775,
      "learning_rate": 1e-05,
      "loss": 1.0862,
      "step": 15295
    },
    {
      "epoch": 13.551815766164747,
      "grad_norm": 0.24170774221420288,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 15300
    },
    {
      "epoch": 13.556244464127547,
      "grad_norm": 0.22384804487228394,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 15305
    },
    {
      "epoch": 13.560673162090346,
      "grad_norm": 0.3383476734161377,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 15310
    },
    {
      "epoch": 13.565101860053144,
      "grad_norm": 0.2526433765888214,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 15315
    },
    {
      "epoch": 13.569530558015943,
      "grad_norm": 0.26605209708213806,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 15320
    },
    {
      "epoch": 13.573959255978743,
      "grad_norm": 0.2773497998714447,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 15325
    },
    {
      "epoch": 13.578387953941542,
      "grad_norm": 0.28566789627075195,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 15330
    },
    {
      "epoch": 13.58281665190434,
      "grad_norm": 0.25376608967781067,
      "learning_rate": 1e-05,
      "loss": 1.0461,
      "step": 15335
    },
    {
      "epoch": 13.587245349867139,
      "grad_norm": 0.3573301136493683,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 15340
    },
    {
      "epoch": 13.591674047829938,
      "grad_norm": 0.2915681004524231,
      "learning_rate": 1e-05,
      "loss": 1.0186,
      "step": 15345
    },
    {
      "epoch": 13.596102745792736,
      "grad_norm": 0.30945998430252075,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 15350
    },
    {
      "epoch": 13.600531443755536,
      "grad_norm": 0.35102200508117676,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 15355
    },
    {
      "epoch": 13.604960141718335,
      "grad_norm": 0.25039467215538025,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 15360
    },
    {
      "epoch": 13.609388839681134,
      "grad_norm": 0.23250414431095123,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 15365
    },
    {
      "epoch": 13.613817537643932,
      "grad_norm": 0.29039010405540466,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 15370
    },
    {
      "epoch": 13.618246235606732,
      "grad_norm": 0.28464218974113464,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 15375
    },
    {
      "epoch": 13.622674933569531,
      "grad_norm": 0.2606031894683838,
      "learning_rate": 1e-05,
      "loss": 0.9272,
      "step": 15380
    },
    {
      "epoch": 13.627103631532329,
      "grad_norm": 0.27778005599975586,
      "learning_rate": 1e-05,
      "loss": 1.0419,
      "step": 15385
    },
    {
      "epoch": 13.631532329495128,
      "grad_norm": 0.2646855115890503,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 15390
    },
    {
      "epoch": 13.635961027457927,
      "grad_norm": 0.25512513518333435,
      "learning_rate": 1e-05,
      "loss": 0.9166,
      "step": 15395
    },
    {
      "epoch": 13.640389725420727,
      "grad_norm": 0.2692898213863373,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 15400
    },
    {
      "epoch": 13.644818423383525,
      "grad_norm": 0.21172656118869781,
      "learning_rate": 1e-05,
      "loss": 1.0305,
      "step": 15405
    },
    {
      "epoch": 13.649247121346324,
      "grad_norm": 0.23341864347457886,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 15410
    },
    {
      "epoch": 13.653675819309123,
      "grad_norm": 0.2582659125328064,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 15415
    },
    {
      "epoch": 13.658104517271923,
      "grad_norm": 0.2700680196285248,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 15420
    },
    {
      "epoch": 13.66253321523472,
      "grad_norm": 0.22885730862617493,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 15425
    },
    {
      "epoch": 13.66696191319752,
      "grad_norm": 0.36526116728782654,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 15430
    },
    {
      "epoch": 13.67139061116032,
      "grad_norm": 0.2604011595249176,
      "learning_rate": 1e-05,
      "loss": 1.0284,
      "step": 15435
    },
    {
      "epoch": 13.675819309123117,
      "grad_norm": 0.29298102855682373,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 15440
    },
    {
      "epoch": 13.680248007085916,
      "grad_norm": 0.23797187209129333,
      "learning_rate": 1e-05,
      "loss": 1.0333,
      "step": 15445
    },
    {
      "epoch": 13.684676705048716,
      "grad_norm": 0.25061294436454773,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 15450
    },
    {
      "epoch": 13.689105403011515,
      "grad_norm": 0.3172397017478943,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 15455
    },
    {
      "epoch": 13.693534100974313,
      "grad_norm": 0.2732192575931549,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 15460
    },
    {
      "epoch": 13.697962798937112,
      "grad_norm": 0.3379781246185303,
      "learning_rate": 1e-05,
      "loss": 1.0113,
      "step": 15465
    },
    {
      "epoch": 13.702391496899912,
      "grad_norm": 0.29626280069351196,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 15470
    },
    {
      "epoch": 13.706820194862711,
      "grad_norm": 0.25145697593688965,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 15475
    },
    {
      "epoch": 13.711248892825509,
      "grad_norm": 0.28558701276779175,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 15480
    },
    {
      "epoch": 13.715677590788308,
      "grad_norm": 0.2868648171424866,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 15485
    },
    {
      "epoch": 13.720106288751108,
      "grad_norm": 0.22914022207260132,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 15490
    },
    {
      "epoch": 13.724534986713905,
      "grad_norm": 0.2646102011203766,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 15495
    },
    {
      "epoch": 13.728963684676705,
      "grad_norm": 0.29670560359954834,
      "learning_rate": 1e-05,
      "loss": 1.0009,
      "step": 15500
    },
    {
      "epoch": 13.733392382639504,
      "grad_norm": 0.2894361913204193,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 15505
    },
    {
      "epoch": 13.737821080602302,
      "grad_norm": 0.32102397084236145,
      "learning_rate": 1e-05,
      "loss": 1.0388,
      "step": 15510
    },
    {
      "epoch": 13.742249778565101,
      "grad_norm": 0.3547017574310303,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 15515
    },
    {
      "epoch": 13.7466784765279,
      "grad_norm": 0.2745082974433899,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 15520
    },
    {
      "epoch": 13.7511071744907,
      "grad_norm": 0.2461821287870407,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 15525
    },
    {
      "epoch": 13.755535872453498,
      "grad_norm": 0.30014970898628235,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 15530
    },
    {
      "epoch": 13.759964570416297,
      "grad_norm": 0.28270408511161804,
      "learning_rate": 1e-05,
      "loss": 1.0204,
      "step": 15535
    },
    {
      "epoch": 13.764393268379097,
      "grad_norm": 0.2666551470756531,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 15540
    },
    {
      "epoch": 13.768821966341896,
      "grad_norm": 0.2758060693740845,
      "learning_rate": 1e-05,
      "loss": 1.0317,
      "step": 15545
    },
    {
      "epoch": 13.773250664304694,
      "grad_norm": 0.28551164269447327,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 15550
    },
    {
      "epoch": 13.777679362267493,
      "grad_norm": 0.28669363260269165,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 15555
    },
    {
      "epoch": 13.782108060230293,
      "grad_norm": 0.27511075139045715,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 15560
    },
    {
      "epoch": 13.78653675819309,
      "grad_norm": 0.28586411476135254,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 15565
    },
    {
      "epoch": 13.79096545615589,
      "grad_norm": 0.2807426452636719,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 15570
    },
    {
      "epoch": 13.79539415411869,
      "grad_norm": 0.25462839007377625,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 15575
    },
    {
      "epoch": 13.799822852081489,
      "grad_norm": 0.2450139969587326,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 15580
    },
    {
      "epoch": 13.804251550044286,
      "grad_norm": 0.25302863121032715,
      "learning_rate": 1e-05,
      "loss": 1.0281,
      "step": 15585
    },
    {
      "epoch": 13.808680248007086,
      "grad_norm": 0.2476828545331955,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 15590
    },
    {
      "epoch": 13.813108945969885,
      "grad_norm": 0.23024851083755493,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 15595
    },
    {
      "epoch": 13.817537643932685,
      "grad_norm": 0.24292601644992828,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 15600
    },
    {
      "epoch": 13.821966341895482,
      "grad_norm": 0.2939973473548889,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 15605
    },
    {
      "epoch": 13.826395039858282,
      "grad_norm": 0.25731831789016724,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 15610
    },
    {
      "epoch": 13.830823737821081,
      "grad_norm": 0.26878997683525085,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 15615
    },
    {
      "epoch": 13.835252435783879,
      "grad_norm": 0.3293323218822479,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 15620
    },
    {
      "epoch": 13.839681133746678,
      "grad_norm": 0.25292903184890747,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 15625
    },
    {
      "epoch": 13.844109831709478,
      "grad_norm": 0.24890176951885223,
      "learning_rate": 1e-05,
      "loss": 1.0377,
      "step": 15630
    },
    {
      "epoch": 13.848538529672277,
      "grad_norm": 0.28238019347190857,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 15635
    },
    {
      "epoch": 13.852967227635075,
      "grad_norm": 0.28742435574531555,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 15640
    },
    {
      "epoch": 13.857395925597874,
      "grad_norm": 0.27353227138519287,
      "learning_rate": 1e-05,
      "loss": 1.0555,
      "step": 15645
    },
    {
      "epoch": 13.861824623560674,
      "grad_norm": 0.24027560651302338,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 15650
    },
    {
      "epoch": 13.866253321523471,
      "grad_norm": 0.2460683435201645,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 15655
    },
    {
      "epoch": 13.87068201948627,
      "grad_norm": 0.29546400904655457,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 15660
    },
    {
      "epoch": 13.87511071744907,
      "grad_norm": 0.23563817143440247,
      "learning_rate": 1e-05,
      "loss": 1.0281,
      "step": 15665
    },
    {
      "epoch": 13.87953941541187,
      "grad_norm": 0.2567593455314636,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 15670
    },
    {
      "epoch": 13.883968113374667,
      "grad_norm": 0.2727123498916626,
      "learning_rate": 1e-05,
      "loss": 1.0415,
      "step": 15675
    },
    {
      "epoch": 13.888396811337467,
      "grad_norm": 0.31490030884742737,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 15680
    },
    {
      "epoch": 13.892825509300266,
      "grad_norm": 0.25928354263305664,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 15685
    },
    {
      "epoch": 13.897254207263064,
      "grad_norm": 0.24376000463962555,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 15690
    },
    {
      "epoch": 13.901682905225863,
      "grad_norm": 0.266998291015625,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 15695
    },
    {
      "epoch": 13.906111603188663,
      "grad_norm": 0.23325850069522858,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 15700
    },
    {
      "epoch": 13.910540301151462,
      "grad_norm": 0.2657238841056824,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 15705
    },
    {
      "epoch": 13.91496899911426,
      "grad_norm": 0.28277379274368286,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 15710
    },
    {
      "epoch": 13.91939769707706,
      "grad_norm": 0.2881607413291931,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 15715
    },
    {
      "epoch": 13.923826395039859,
      "grad_norm": 0.30713793635368347,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 15720
    },
    {
      "epoch": 13.928255093002658,
      "grad_norm": 0.2747843265533447,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 15725
    },
    {
      "epoch": 13.932683790965456,
      "grad_norm": 0.3049401044845581,
      "learning_rate": 1e-05,
      "loss": 1.0305,
      "step": 15730
    },
    {
      "epoch": 13.937112488928255,
      "grad_norm": 0.2870447635650635,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 15735
    },
    {
      "epoch": 13.941541186891055,
      "grad_norm": 0.23038822412490845,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 15740
    },
    {
      "epoch": 13.945969884853852,
      "grad_norm": 0.22686836123466492,
      "learning_rate": 1e-05,
      "loss": 1.0305,
      "step": 15745
    },
    {
      "epoch": 13.950398582816652,
      "grad_norm": 0.2563081681728363,
      "learning_rate": 1e-05,
      "loss": 1.0248,
      "step": 15750
    },
    {
      "epoch": 13.954827280779451,
      "grad_norm": 0.2691231966018677,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 15755
    },
    {
      "epoch": 13.95925597874225,
      "grad_norm": 0.2593192756175995,
      "learning_rate": 1e-05,
      "loss": 1.0345,
      "step": 15760
    },
    {
      "epoch": 13.963684676705048,
      "grad_norm": 0.2814178466796875,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 15765
    },
    {
      "epoch": 13.968113374667848,
      "grad_norm": 0.3045186400413513,
      "learning_rate": 1e-05,
      "loss": 1.0113,
      "step": 15770
    },
    {
      "epoch": 13.972542072630647,
      "grad_norm": 0.2753313481807709,
      "learning_rate": 1e-05,
      "loss": 1.0216,
      "step": 15775
    },
    {
      "epoch": 13.976970770593445,
      "grad_norm": 0.2835383415222168,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 15780
    },
    {
      "epoch": 13.981399468556244,
      "grad_norm": 0.2729502022266388,
      "learning_rate": 1e-05,
      "loss": 1.0389,
      "step": 15785
    },
    {
      "epoch": 13.985828166519044,
      "grad_norm": 0.325484037399292,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 15790
    },
    {
      "epoch": 13.990256864481843,
      "grad_norm": 0.28955528140068054,
      "learning_rate": 1e-05,
      "loss": 1.0341,
      "step": 15795
    },
    {
      "epoch": 13.99468556244464,
      "grad_norm": 0.22102151811122894,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 15800
    },
    {
      "epoch": 13.99911426040744,
      "grad_norm": 0.23810943961143494,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 15805
    },
    {
      "epoch": 14.00354295837024,
      "grad_norm": 0.2610328495502472,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 15810
    },
    {
      "epoch": 14.007971656333037,
      "grad_norm": 0.2607215940952301,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 15815
    },
    {
      "epoch": 14.012400354295837,
      "grad_norm": 0.28001588582992554,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 15820
    },
    {
      "epoch": 14.016829052258636,
      "grad_norm": 0.32271939516067505,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 15825
    },
    {
      "epoch": 14.021257750221436,
      "grad_norm": 0.26282185316085815,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 15830
    },
    {
      "epoch": 14.025686448184233,
      "grad_norm": 0.2943328619003296,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 15835
    },
    {
      "epoch": 14.030115146147033,
      "grad_norm": 0.31595247983932495,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 15840
    },
    {
      "epoch": 14.034543844109832,
      "grad_norm": 0.2925513982772827,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 15845
    },
    {
      "epoch": 14.038972542072631,
      "grad_norm": 0.2770253121852875,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 15850
    },
    {
      "epoch": 14.043401240035429,
      "grad_norm": 0.34785792231559753,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 15855
    },
    {
      "epoch": 14.047829937998229,
      "grad_norm": 0.25001445412635803,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 15860
    },
    {
      "epoch": 14.052258635961028,
      "grad_norm": 0.28387728333473206,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 15865
    },
    {
      "epoch": 14.056687333923826,
      "grad_norm": 0.2730706036090851,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 15870
    },
    {
      "epoch": 14.061116031886625,
      "grad_norm": 0.2684321105480194,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 15875
    },
    {
      "epoch": 14.065544729849424,
      "grad_norm": 0.25972771644592285,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 15880
    },
    {
      "epoch": 14.069973427812224,
      "grad_norm": 0.26421937346458435,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 15885
    },
    {
      "epoch": 14.074402125775022,
      "grad_norm": 0.31100571155548096,
      "learning_rate": 1e-05,
      "loss": 1.0172,
      "step": 15890
    },
    {
      "epoch": 14.078830823737821,
      "grad_norm": 0.26712560653686523,
      "learning_rate": 1e-05,
      "loss": 1.0526,
      "step": 15895
    },
    {
      "epoch": 14.08325952170062,
      "grad_norm": 0.27068427205085754,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 15900
    },
    {
      "epoch": 14.087688219663418,
      "grad_norm": 0.3587118983268738,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 15905
    },
    {
      "epoch": 14.092116917626218,
      "grad_norm": 0.38153502345085144,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 15910
    },
    {
      "epoch": 14.096545615589017,
      "grad_norm": 0.25527530908584595,
      "learning_rate": 1e-05,
      "loss": 1.0246,
      "step": 15915
    },
    {
      "epoch": 14.100974313551816,
      "grad_norm": 0.2993118166923523,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 15920
    },
    {
      "epoch": 14.105403011514614,
      "grad_norm": 0.2507144510746002,
      "learning_rate": 1e-05,
      "loss": 1.035,
      "step": 15925
    },
    {
      "epoch": 14.109831709477413,
      "grad_norm": 0.3291673958301544,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 15930
    },
    {
      "epoch": 14.114260407440213,
      "grad_norm": 0.22528529167175293,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 15935
    },
    {
      "epoch": 14.118689105403012,
      "grad_norm": 0.24888576567173004,
      "learning_rate": 1e-05,
      "loss": 1.0312,
      "step": 15940
    },
    {
      "epoch": 14.12311780336581,
      "grad_norm": 0.24165961146354675,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 15945
    },
    {
      "epoch": 14.12754650132861,
      "grad_norm": 0.28978678584098816,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 15950
    },
    {
      "epoch": 14.131975199291409,
      "grad_norm": 0.2681102752685547,
      "learning_rate": 1e-05,
      "loss": 1.0319,
      "step": 15955
    },
    {
      "epoch": 14.136403897254207,
      "grad_norm": 0.24786819517612457,
      "learning_rate": 1e-05,
      "loss": 1.0259,
      "step": 15960
    },
    {
      "epoch": 14.140832595217006,
      "grad_norm": 0.2648155987262726,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 15965
    },
    {
      "epoch": 14.145261293179805,
      "grad_norm": 0.25509393215179443,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 15970
    },
    {
      "epoch": 14.149689991142605,
      "grad_norm": 0.33331871032714844,
      "learning_rate": 1e-05,
      "loss": 1.0318,
      "step": 15975
    },
    {
      "epoch": 14.154118689105402,
      "grad_norm": 0.2945652902126312,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 15980
    },
    {
      "epoch": 14.158547387068202,
      "grad_norm": 0.41670629382133484,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 15985
    },
    {
      "epoch": 14.162976085031001,
      "grad_norm": 0.2906208336353302,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 15990
    },
    {
      "epoch": 14.167404782993799,
      "grad_norm": 0.2625887095928192,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 15995
    },
    {
      "epoch": 14.171833480956598,
      "grad_norm": 0.2908119559288025,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 16000
    },
    {
      "epoch": 14.176262178919398,
      "grad_norm": 0.2747705578804016,
      "learning_rate": 1e-05,
      "loss": 1.0255,
      "step": 16005
    },
    {
      "epoch": 14.180690876882197,
      "grad_norm": 0.3522818386554718,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 16010
    },
    {
      "epoch": 14.185119574844995,
      "grad_norm": 0.27590030431747437,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 16015
    },
    {
      "epoch": 14.189548272807794,
      "grad_norm": 0.24589639902114868,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 16020
    },
    {
      "epoch": 14.193976970770594,
      "grad_norm": 0.28107768297195435,
      "learning_rate": 1e-05,
      "loss": 0.9196,
      "step": 16025
    },
    {
      "epoch": 14.198405668733393,
      "grad_norm": 0.26836106181144714,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 16030
    },
    {
      "epoch": 14.202834366696191,
      "grad_norm": 0.2444487065076828,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 16035
    },
    {
      "epoch": 14.20726306465899,
      "grad_norm": 0.26155325770378113,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 16040
    },
    {
      "epoch": 14.21169176262179,
      "grad_norm": 0.2920330762863159,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 16045
    },
    {
      "epoch": 14.216120460584587,
      "grad_norm": 0.3188605010509491,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 16050
    },
    {
      "epoch": 14.220549158547387,
      "grad_norm": 0.27870675921440125,
      "learning_rate": 1e-05,
      "loss": 1.0632,
      "step": 16055
    },
    {
      "epoch": 14.224977856510186,
      "grad_norm": 0.24375474452972412,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 16060
    },
    {
      "epoch": 14.229406554472986,
      "grad_norm": 0.28569549322128296,
      "learning_rate": 1e-05,
      "loss": 1.0486,
      "step": 16065
    },
    {
      "epoch": 14.233835252435783,
      "grad_norm": 0.3168725371360779,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 16070
    },
    {
      "epoch": 14.238263950398583,
      "grad_norm": 0.2294890433549881,
      "learning_rate": 1e-05,
      "loss": 1.0174,
      "step": 16075
    },
    {
      "epoch": 14.242692648361382,
      "grad_norm": 0.30843108892440796,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 16080
    },
    {
      "epoch": 14.24712134632418,
      "grad_norm": 0.23705920577049255,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 16085
    },
    {
      "epoch": 14.25155004428698,
      "grad_norm": 0.22615958750247955,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 16090
    },
    {
      "epoch": 14.255978742249779,
      "grad_norm": 0.27387434244155884,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 16095
    },
    {
      "epoch": 14.260407440212578,
      "grad_norm": 0.31500357389450073,
      "learning_rate": 1e-05,
      "loss": 1.0383,
      "step": 16100
    },
    {
      "epoch": 14.264836138175376,
      "grad_norm": 0.27954399585723877,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 16105
    },
    {
      "epoch": 14.269264836138175,
      "grad_norm": 0.26311737298965454,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 16110
    },
    {
      "epoch": 14.273693534100975,
      "grad_norm": 0.2789212763309479,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 16115
    },
    {
      "epoch": 14.278122232063772,
      "grad_norm": 0.2928401529788971,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 16120
    },
    {
      "epoch": 14.282550930026572,
      "grad_norm": 0.2525635361671448,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 16125
    },
    {
      "epoch": 14.286979627989371,
      "grad_norm": 0.2728481888771057,
      "learning_rate": 1e-05,
      "loss": 1.0276,
      "step": 16130
    },
    {
      "epoch": 14.29140832595217,
      "grad_norm": 0.2740583121776581,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 16135
    },
    {
      "epoch": 14.295837023914968,
      "grad_norm": 0.24192453920841217,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 16140
    },
    {
      "epoch": 14.300265721877768,
      "grad_norm": 0.36275580525398254,
      "learning_rate": 1e-05,
      "loss": 1.0312,
      "step": 16145
    },
    {
      "epoch": 14.304694419840567,
      "grad_norm": 0.2272033393383026,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 16150
    },
    {
      "epoch": 14.309123117803367,
      "grad_norm": 0.22436775267124176,
      "learning_rate": 1e-05,
      "loss": 1.0248,
      "step": 16155
    },
    {
      "epoch": 14.313551815766164,
      "grad_norm": 0.246480792760849,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 16160
    },
    {
      "epoch": 14.317980513728964,
      "grad_norm": 0.21877822279930115,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 16165
    },
    {
      "epoch": 14.322409211691763,
      "grad_norm": 0.23440659046173096,
      "learning_rate": 1e-05,
      "loss": 1.0475,
      "step": 16170
    },
    {
      "epoch": 14.32683790965456,
      "grad_norm": 0.2668089270591736,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 16175
    },
    {
      "epoch": 14.33126660761736,
      "grad_norm": 0.2701428532600403,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 16180
    },
    {
      "epoch": 14.33569530558016,
      "grad_norm": 0.32622238993644714,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 16185
    },
    {
      "epoch": 14.34012400354296,
      "grad_norm": 0.23200897872447968,
      "learning_rate": 1e-05,
      "loss": 1.0193,
      "step": 16190
    },
    {
      "epoch": 14.344552701505757,
      "grad_norm": 0.3101765215396881,
      "learning_rate": 1e-05,
      "loss": 0.8924,
      "step": 16195
    },
    {
      "epoch": 14.348981399468556,
      "grad_norm": 0.26896390318870544,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 16200
    },
    {
      "epoch": 14.353410097431356,
      "grad_norm": 0.277150958776474,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 16205
    },
    {
      "epoch": 14.357838795394153,
      "grad_norm": 0.23689205944538116,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 16210
    },
    {
      "epoch": 14.362267493356953,
      "grad_norm": 0.25146445631980896,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 16215
    },
    {
      "epoch": 14.366696191319752,
      "grad_norm": 0.2717793583869934,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 16220
    },
    {
      "epoch": 14.371124889282552,
      "grad_norm": 0.2160385698080063,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 16225
    },
    {
      "epoch": 14.37555358724535,
      "grad_norm": 0.226392924785614,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 16230
    },
    {
      "epoch": 14.379982285208149,
      "grad_norm": 0.21924401819705963,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 16235
    },
    {
      "epoch": 14.384410983170948,
      "grad_norm": 0.23507100343704224,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 16240
    },
    {
      "epoch": 14.388839681133746,
      "grad_norm": 0.2684524655342102,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 16245
    },
    {
      "epoch": 14.393268379096545,
      "grad_norm": 0.2606136202812195,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 16250
    },
    {
      "epoch": 14.397697077059345,
      "grad_norm": 0.27165958285331726,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 16255
    },
    {
      "epoch": 14.402125775022144,
      "grad_norm": 0.27855613827705383,
      "learning_rate": 1e-05,
      "loss": 1.0446,
      "step": 16260
    },
    {
      "epoch": 14.406554472984942,
      "grad_norm": 0.28144940733909607,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 16265
    },
    {
      "epoch": 14.410983170947741,
      "grad_norm": 0.285361111164093,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 16270
    },
    {
      "epoch": 14.41541186891054,
      "grad_norm": 0.2869277596473694,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 16275
    },
    {
      "epoch": 14.41984056687334,
      "grad_norm": 0.28857702016830444,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 16280
    },
    {
      "epoch": 14.424269264836138,
      "grad_norm": 0.25510743260383606,
      "learning_rate": 1e-05,
      "loss": 0.9083,
      "step": 16285
    },
    {
      "epoch": 14.428697962798937,
      "grad_norm": 0.2594030201435089,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 16290
    },
    {
      "epoch": 14.433126660761737,
      "grad_norm": 0.26197192072868347,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 16295
    },
    {
      "epoch": 14.437555358724534,
      "grad_norm": 0.25105172395706177,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 16300
    },
    {
      "epoch": 14.441984056687334,
      "grad_norm": 0.26715198159217834,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 16305
    },
    {
      "epoch": 14.446412754650133,
      "grad_norm": 0.2774592638015747,
      "learning_rate": 1e-05,
      "loss": 1.0382,
      "step": 16310
    },
    {
      "epoch": 14.450841452612933,
      "grad_norm": 0.2662948668003082,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 16315
    },
    {
      "epoch": 14.45527015057573,
      "grad_norm": 0.35703012347221375,
      "learning_rate": 1e-05,
      "loss": 1.0323,
      "step": 16320
    },
    {
      "epoch": 14.45969884853853,
      "grad_norm": 0.29315707087516785,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 16325
    },
    {
      "epoch": 14.464127546501329,
      "grad_norm": 0.2548542022705078,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 16330
    },
    {
      "epoch": 14.468556244464127,
      "grad_norm": 0.2775784432888031,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 16335
    },
    {
      "epoch": 14.472984942426926,
      "grad_norm": 0.24884721636772156,
      "learning_rate": 1e-05,
      "loss": 1.0493,
      "step": 16340
    },
    {
      "epoch": 14.477413640389726,
      "grad_norm": 0.2941446602344513,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 16345
    },
    {
      "epoch": 14.481842338352525,
      "grad_norm": 0.2529769837856293,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 16350
    },
    {
      "epoch": 14.486271036315323,
      "grad_norm": 0.2483808547258377,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 16355
    },
    {
      "epoch": 14.490699734278122,
      "grad_norm": 0.29735067486763,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 16360
    },
    {
      "epoch": 14.495128432240922,
      "grad_norm": 0.2199345976114273,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 16365
    },
    {
      "epoch": 14.499557130203721,
      "grad_norm": 0.25083208084106445,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 16370
    },
    {
      "epoch": 14.503985828166519,
      "grad_norm": 0.2711193263530731,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 16375
    },
    {
      "epoch": 14.508414526129318,
      "grad_norm": 0.2512422800064087,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 16380
    },
    {
      "epoch": 14.512843224092117,
      "grad_norm": 0.277103453874588,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 16385
    },
    {
      "epoch": 14.517271922054915,
      "grad_norm": 0.27260157465934753,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 16390
    },
    {
      "epoch": 14.521700620017715,
      "grad_norm": 0.2731979191303253,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 16395
    },
    {
      "epoch": 14.526129317980514,
      "grad_norm": 0.27902352809906006,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 16400
    },
    {
      "epoch": 14.530558015943313,
      "grad_norm": 0.28178659081459045,
      "learning_rate": 1e-05,
      "loss": 1.0216,
      "step": 16405
    },
    {
      "epoch": 14.534986713906111,
      "grad_norm": 0.2443675696849823,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 16410
    },
    {
      "epoch": 14.53941541186891,
      "grad_norm": 0.2959047555923462,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 16415
    },
    {
      "epoch": 14.54384410983171,
      "grad_norm": 0.3143279254436493,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 16420
    },
    {
      "epoch": 14.548272807794508,
      "grad_norm": 0.281217098236084,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 16425
    },
    {
      "epoch": 14.552701505757307,
      "grad_norm": 0.31636083126068115,
      "learning_rate": 1e-05,
      "loss": 1.0368,
      "step": 16430
    },
    {
      "epoch": 14.557130203720106,
      "grad_norm": 0.2782772183418274,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 16435
    },
    {
      "epoch": 14.561558901682906,
      "grad_norm": 0.28139472007751465,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 16440
    },
    {
      "epoch": 14.565987599645704,
      "grad_norm": 0.2928841710090637,
      "learning_rate": 1e-05,
      "loss": 1.0301,
      "step": 16445
    },
    {
      "epoch": 14.570416297608503,
      "grad_norm": 0.26062625646591187,
      "learning_rate": 1e-05,
      "loss": 1.0515,
      "step": 16450
    },
    {
      "epoch": 14.574844995571302,
      "grad_norm": 0.29303669929504395,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 16455
    },
    {
      "epoch": 14.579273693534102,
      "grad_norm": 0.2500944435596466,
      "learning_rate": 1e-05,
      "loss": 1.0297,
      "step": 16460
    },
    {
      "epoch": 14.5837023914969,
      "grad_norm": 0.26715782284736633,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 16465
    },
    {
      "epoch": 14.588131089459699,
      "grad_norm": 0.2846585512161255,
      "learning_rate": 1e-05,
      "loss": 1.0171,
      "step": 16470
    },
    {
      "epoch": 14.592559787422498,
      "grad_norm": 0.26248589158058167,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 16475
    },
    {
      "epoch": 14.596988485385296,
      "grad_norm": 0.3085353374481201,
      "learning_rate": 1e-05,
      "loss": 1.0399,
      "step": 16480
    },
    {
      "epoch": 14.601417183348095,
      "grad_norm": 0.37500736117362976,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 16485
    },
    {
      "epoch": 14.605845881310895,
      "grad_norm": 0.27004265785217285,
      "learning_rate": 1e-05,
      "loss": 1.0709,
      "step": 16490
    },
    {
      "epoch": 14.610274579273694,
      "grad_norm": 0.30227288603782654,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 16495
    },
    {
      "epoch": 14.614703277236492,
      "grad_norm": 0.2669404447078705,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 16500
    },
    {
      "epoch": 14.619131975199291,
      "grad_norm": 0.2740556001663208,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 16505
    },
    {
      "epoch": 14.62356067316209,
      "grad_norm": 0.2449372559785843,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 16510
    },
    {
      "epoch": 14.627989371124889,
      "grad_norm": 0.30259495973587036,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 16515
    },
    {
      "epoch": 14.632418069087688,
      "grad_norm": 0.2543676197528839,
      "learning_rate": 1e-05,
      "loss": 1.0349,
      "step": 16520
    },
    {
      "epoch": 14.636846767050487,
      "grad_norm": 0.2868063747882843,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 16525
    },
    {
      "epoch": 14.641275465013287,
      "grad_norm": 0.24118371307849884,
      "learning_rate": 1e-05,
      "loss": 0.9197,
      "step": 16530
    },
    {
      "epoch": 14.645704162976084,
      "grad_norm": 0.2835771143436432,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 16535
    },
    {
      "epoch": 14.650132860938884,
      "grad_norm": 0.2593700587749481,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 16540
    },
    {
      "epoch": 14.654561558901683,
      "grad_norm": 0.21953213214874268,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 16545
    },
    {
      "epoch": 14.658990256864481,
      "grad_norm": 0.26301050186157227,
      "learning_rate": 1e-05,
      "loss": 1.0542,
      "step": 16550
    },
    {
      "epoch": 14.66341895482728,
      "grad_norm": 0.27383026480674744,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 16555
    },
    {
      "epoch": 14.66784765279008,
      "grad_norm": 0.2769801914691925,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 16560
    },
    {
      "epoch": 14.67227635075288,
      "grad_norm": 0.29033446311950684,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 16565
    },
    {
      "epoch": 14.676705048715677,
      "grad_norm": 0.2810085415840149,
      "learning_rate": 1e-05,
      "loss": 1.0159,
      "step": 16570
    },
    {
      "epoch": 14.681133746678476,
      "grad_norm": 0.30164459347724915,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 16575
    },
    {
      "epoch": 14.685562444641276,
      "grad_norm": 0.2507741451263428,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 16580
    },
    {
      "epoch": 14.689991142604075,
      "grad_norm": 0.296539843082428,
      "learning_rate": 1e-05,
      "loss": 1.0318,
      "step": 16585
    },
    {
      "epoch": 14.694419840566873,
      "grad_norm": 0.28560638427734375,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 16590
    },
    {
      "epoch": 14.698848538529672,
      "grad_norm": 0.2627999782562256,
      "learning_rate": 1e-05,
      "loss": 1.0426,
      "step": 16595
    },
    {
      "epoch": 14.703277236492472,
      "grad_norm": 0.26439911127090454,
      "learning_rate": 1e-05,
      "loss": 1.079,
      "step": 16600
    },
    {
      "epoch": 14.70770593445527,
      "grad_norm": 0.25219106674194336,
      "learning_rate": 1e-05,
      "loss": 1.0303,
      "step": 16605
    },
    {
      "epoch": 14.712134632418069,
      "grad_norm": 0.23781970143318176,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 16610
    },
    {
      "epoch": 14.716563330380868,
      "grad_norm": 0.27697667479515076,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 16615
    },
    {
      "epoch": 14.720992028343668,
      "grad_norm": 0.26851117610931396,
      "learning_rate": 1e-05,
      "loss": 1.0184,
      "step": 16620
    },
    {
      "epoch": 14.725420726306465,
      "grad_norm": 0.2835143804550171,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 16625
    },
    {
      "epoch": 14.729849424269265,
      "grad_norm": 0.27118510007858276,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 16630
    },
    {
      "epoch": 14.734278122232064,
      "grad_norm": 0.28629037737846375,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 16635
    },
    {
      "epoch": 14.738706820194862,
      "grad_norm": 0.3202347159385681,
      "learning_rate": 1e-05,
      "loss": 1.0425,
      "step": 16640
    },
    {
      "epoch": 14.743135518157661,
      "grad_norm": 0.24710731208324432,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 16645
    },
    {
      "epoch": 14.74756421612046,
      "grad_norm": 0.25110748410224915,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 16650
    },
    {
      "epoch": 14.75199291408326,
      "grad_norm": 0.271636039018631,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 16655
    },
    {
      "epoch": 14.756421612046058,
      "grad_norm": 0.2625683546066284,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 16660
    },
    {
      "epoch": 14.760850310008857,
      "grad_norm": 0.26619279384613037,
      "learning_rate": 1e-05,
      "loss": 1.0247,
      "step": 16665
    },
    {
      "epoch": 14.765279007971657,
      "grad_norm": 0.3262878656387329,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 16670
    },
    {
      "epoch": 14.769707705934454,
      "grad_norm": 0.2604900896549225,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 16675
    },
    {
      "epoch": 14.774136403897254,
      "grad_norm": 0.26651960611343384,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 16680
    },
    {
      "epoch": 14.778565101860053,
      "grad_norm": 0.2893853485584259,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 16685
    },
    {
      "epoch": 14.782993799822853,
      "grad_norm": 0.24873779714107513,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 16690
    },
    {
      "epoch": 14.78742249778565,
      "grad_norm": 0.2951345443725586,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 16695
    },
    {
      "epoch": 14.79185119574845,
      "grad_norm": 0.25161123275756836,
      "learning_rate": 1e-05,
      "loss": 1.0316,
      "step": 16700
    },
    {
      "epoch": 14.79627989371125,
      "grad_norm": 0.27895843982696533,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 16705
    },
    {
      "epoch": 14.800708591674049,
      "grad_norm": 0.2510391175746918,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 16710
    },
    {
      "epoch": 14.805137289636846,
      "grad_norm": 0.29707637429237366,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 16715
    },
    {
      "epoch": 14.809565987599646,
      "grad_norm": 0.27752935886383057,
      "learning_rate": 1e-05,
      "loss": 1.025,
      "step": 16720
    },
    {
      "epoch": 14.813994685562445,
      "grad_norm": 0.24980412423610687,
      "learning_rate": 1e-05,
      "loss": 1.0286,
      "step": 16725
    },
    {
      "epoch": 14.818423383525243,
      "grad_norm": 0.24784047901630402,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 16730
    },
    {
      "epoch": 14.822852081488042,
      "grad_norm": 0.27620649337768555,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 16735
    },
    {
      "epoch": 14.827280779450842,
      "grad_norm": 0.25357839465141296,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 16740
    },
    {
      "epoch": 14.831709477413641,
      "grad_norm": 0.2991798520088196,
      "learning_rate": 1e-05,
      "loss": 1.0519,
      "step": 16745
    },
    {
      "epoch": 14.836138175376439,
      "grad_norm": 0.25299325585365295,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 16750
    },
    {
      "epoch": 14.840566873339238,
      "grad_norm": 0.24106444418430328,
      "learning_rate": 1e-05,
      "loss": 1.0152,
      "step": 16755
    },
    {
      "epoch": 14.844995571302038,
      "grad_norm": 0.29380327463150024,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 16760
    },
    {
      "epoch": 14.849424269264837,
      "grad_norm": 0.23624031245708466,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 16765
    },
    {
      "epoch": 14.853852967227635,
      "grad_norm": 0.26048368215560913,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 16770
    },
    {
      "epoch": 14.858281665190434,
      "grad_norm": 0.22700917720794678,
      "learning_rate": 1e-05,
      "loss": 1.061,
      "step": 16775
    },
    {
      "epoch": 14.862710363153234,
      "grad_norm": 0.28412002325057983,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 16780
    },
    {
      "epoch": 14.867139061116031,
      "grad_norm": 0.2695987820625305,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 16785
    },
    {
      "epoch": 14.87156775907883,
      "grad_norm": 0.23892877995967865,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 16790
    },
    {
      "epoch": 14.87599645704163,
      "grad_norm": 0.21462759375572205,
      "learning_rate": 1e-05,
      "loss": 1.0327,
      "step": 16795
    },
    {
      "epoch": 14.88042515500443,
      "grad_norm": 0.3129950761795044,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 16800
    },
    {
      "epoch": 14.884853852967227,
      "grad_norm": 0.26655280590057373,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 16805
    },
    {
      "epoch": 14.889282550930027,
      "grad_norm": 0.2924897074699402,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 16810
    },
    {
      "epoch": 14.893711248892826,
      "grad_norm": 0.2594379186630249,
      "learning_rate": 1e-05,
      "loss": 1.0318,
      "step": 16815
    },
    {
      "epoch": 14.898139946855624,
      "grad_norm": 0.3770524263381958,
      "learning_rate": 1e-05,
      "loss": 1.023,
      "step": 16820
    },
    {
      "epoch": 14.902568644818423,
      "grad_norm": 0.32294076681137085,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 16825
    },
    {
      "epoch": 14.906997342781223,
      "grad_norm": 0.35033518075942993,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 16830
    },
    {
      "epoch": 14.911426040744022,
      "grad_norm": 0.2783731520175934,
      "learning_rate": 1e-05,
      "loss": 1.0506,
      "step": 16835
    },
    {
      "epoch": 14.91585473870682,
      "grad_norm": 0.30879294872283936,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 16840
    },
    {
      "epoch": 14.920283436669619,
      "grad_norm": 0.262994647026062,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 16845
    },
    {
      "epoch": 14.924712134632419,
      "grad_norm": 0.30566802620887756,
      "learning_rate": 1e-05,
      "loss": 1.0238,
      "step": 16850
    },
    {
      "epoch": 14.929140832595216,
      "grad_norm": 0.2475215494632721,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 16855
    },
    {
      "epoch": 14.933569530558016,
      "grad_norm": 0.2576778829097748,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 16860
    },
    {
      "epoch": 14.937998228520815,
      "grad_norm": 0.22945937514305115,
      "learning_rate": 1e-05,
      "loss": 1.0148,
      "step": 16865
    },
    {
      "epoch": 14.942426926483614,
      "grad_norm": 0.2262829840183258,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 16870
    },
    {
      "epoch": 14.946855624446412,
      "grad_norm": 0.2827739119529724,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 16875
    },
    {
      "epoch": 14.951284322409212,
      "grad_norm": 0.20690865814685822,
      "learning_rate": 1e-05,
      "loss": 1.0391,
      "step": 16880
    },
    {
      "epoch": 14.955713020372011,
      "grad_norm": 0.2357817441225052,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 16885
    },
    {
      "epoch": 14.96014171833481,
      "grad_norm": 0.2403680831193924,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 16890
    },
    {
      "epoch": 14.964570416297608,
      "grad_norm": 0.24647489190101624,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 16895
    },
    {
      "epoch": 14.968999114260408,
      "grad_norm": 0.26419657468795776,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 16900
    },
    {
      "epoch": 14.973427812223207,
      "grad_norm": 0.25355908274650574,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 16905
    },
    {
      "epoch": 14.977856510186005,
      "grad_norm": 0.2276563048362732,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 16910
    },
    {
      "epoch": 14.982285208148804,
      "grad_norm": 0.253778338432312,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 16915
    },
    {
      "epoch": 14.986713906111603,
      "grad_norm": 0.2397511899471283,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 16920
    },
    {
      "epoch": 14.991142604074403,
      "grad_norm": 0.22408337891101837,
      "learning_rate": 1e-05,
      "loss": 1.0343,
      "step": 16925
    },
    {
      "epoch": 14.9955713020372,
      "grad_norm": 0.25532180070877075,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 16930
    },
    {
      "epoch": 15.0,
      "grad_norm": 0.32064101099967957,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 16935
    },
    {
      "epoch": 15.0044286979628,
      "grad_norm": 0.25387218594551086,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 16940
    },
    {
      "epoch": 15.008857395925597,
      "grad_norm": 0.29483988881111145,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 16945
    },
    {
      "epoch": 15.013286093888397,
      "grad_norm": 0.28784963488578796,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 16950
    },
    {
      "epoch": 15.017714791851196,
      "grad_norm": 0.2789418697357178,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 16955
    },
    {
      "epoch": 15.022143489813995,
      "grad_norm": 0.2883960008621216,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 16960
    },
    {
      "epoch": 15.026572187776793,
      "grad_norm": 0.27577394247055054,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 16965
    },
    {
      "epoch": 15.031000885739592,
      "grad_norm": 0.23802579939365387,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 16970
    },
    {
      "epoch": 15.035429583702392,
      "grad_norm": 0.3305317163467407,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 16975
    },
    {
      "epoch": 15.03985828166519,
      "grad_norm": 0.23631387948989868,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 16980
    },
    {
      "epoch": 15.044286979627989,
      "grad_norm": 0.31919947266578674,
      "learning_rate": 1e-05,
      "loss": 1.0233,
      "step": 16985
    },
    {
      "epoch": 15.048715677590788,
      "grad_norm": 0.2521390914916992,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 16990
    },
    {
      "epoch": 15.053144375553588,
      "grad_norm": 0.34347400069236755,
      "learning_rate": 1e-05,
      "loss": 1.0387,
      "step": 16995
    },
    {
      "epoch": 15.057573073516386,
      "grad_norm": 0.21043822169303894,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 17000
    },
    {
      "epoch": 15.062001771479185,
      "grad_norm": 0.22088706493377686,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 17005
    },
    {
      "epoch": 15.066430469441984,
      "grad_norm": 0.2586420476436615,
      "learning_rate": 1e-05,
      "loss": 1.0483,
      "step": 17010
    },
    {
      "epoch": 15.070859167404784,
      "grad_norm": 0.27061590552330017,
      "learning_rate": 1e-05,
      "loss": 0.915,
      "step": 17015
    },
    {
      "epoch": 15.075287865367581,
      "grad_norm": 0.2675482928752899,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 17020
    },
    {
      "epoch": 15.079716563330381,
      "grad_norm": 0.21281589567661285,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 17025
    },
    {
      "epoch": 15.08414526129318,
      "grad_norm": 0.3287443220615387,
      "learning_rate": 1e-05,
      "loss": 1.0194,
      "step": 17030
    },
    {
      "epoch": 15.088573959255978,
      "grad_norm": 0.2601924538612366,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 17035
    },
    {
      "epoch": 15.093002657218777,
      "grad_norm": 0.27621152997016907,
      "learning_rate": 1e-05,
      "loss": 1.0465,
      "step": 17040
    },
    {
      "epoch": 15.097431355181577,
      "grad_norm": 0.24964499473571777,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 17045
    },
    {
      "epoch": 15.101860053144376,
      "grad_norm": 0.31866148114204407,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 17050
    },
    {
      "epoch": 15.106288751107174,
      "grad_norm": 0.26239389181137085,
      "learning_rate": 1e-05,
      "loss": 1.0817,
      "step": 17055
    },
    {
      "epoch": 15.110717449069973,
      "grad_norm": 0.2661646604537964,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 17060
    },
    {
      "epoch": 15.115146147032773,
      "grad_norm": 0.2909359037876129,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 17065
    },
    {
      "epoch": 15.11957484499557,
      "grad_norm": 0.25504469871520996,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 17070
    },
    {
      "epoch": 15.12400354295837,
      "grad_norm": 0.27992212772369385,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 17075
    },
    {
      "epoch": 15.12843224092117,
      "grad_norm": 0.2818259596824646,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 17080
    },
    {
      "epoch": 15.132860938883969,
      "grad_norm": 0.31795716285705566,
      "learning_rate": 1e-05,
      "loss": 1.071,
      "step": 17085
    },
    {
      "epoch": 15.137289636846766,
      "grad_norm": 0.2748221457004547,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 17090
    },
    {
      "epoch": 15.141718334809566,
      "grad_norm": 0.3239392936229706,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 17095
    },
    {
      "epoch": 15.146147032772365,
      "grad_norm": 0.2508675754070282,
      "learning_rate": 1e-05,
      "loss": 1.0222,
      "step": 17100
    },
    {
      "epoch": 15.150575730735165,
      "grad_norm": 0.2542046308517456,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 17105
    },
    {
      "epoch": 15.155004428697962,
      "grad_norm": 0.2578985393047333,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 17110
    },
    {
      "epoch": 15.159433126660762,
      "grad_norm": 0.3133172392845154,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 17115
    },
    {
      "epoch": 15.163861824623561,
      "grad_norm": 0.26482972502708435,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 17120
    },
    {
      "epoch": 15.168290522586359,
      "grad_norm": 0.304808109998703,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 17125
    },
    {
      "epoch": 15.172719220549158,
      "grad_norm": 0.28081241250038147,
      "learning_rate": 1e-05,
      "loss": 1.0218,
      "step": 17130
    },
    {
      "epoch": 15.177147918511958,
      "grad_norm": 0.25393274426460266,
      "learning_rate": 1e-05,
      "loss": 1.0319,
      "step": 17135
    },
    {
      "epoch": 15.181576616474757,
      "grad_norm": 0.25129199028015137,
      "learning_rate": 1e-05,
      "loss": 1.0346,
      "step": 17140
    },
    {
      "epoch": 15.186005314437555,
      "grad_norm": 0.26994654536247253,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 17145
    },
    {
      "epoch": 15.190434012400354,
      "grad_norm": 0.3445732891559601,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 17150
    },
    {
      "epoch": 15.194862710363154,
      "grad_norm": 0.2910650372505188,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 17155
    },
    {
      "epoch": 15.199291408325951,
      "grad_norm": 0.2666941285133362,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 17160
    },
    {
      "epoch": 15.20372010628875,
      "grad_norm": 0.3099244236946106,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 17165
    },
    {
      "epoch": 15.20814880425155,
      "grad_norm": 0.2541021406650543,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 17170
    },
    {
      "epoch": 15.21257750221435,
      "grad_norm": 0.2885095179080963,
      "learning_rate": 1e-05,
      "loss": 1.0234,
      "step": 17175
    },
    {
      "epoch": 15.217006200177147,
      "grad_norm": 0.2297290563583374,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 17180
    },
    {
      "epoch": 15.221434898139947,
      "grad_norm": 0.26323774456977844,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 17185
    },
    {
      "epoch": 15.225863596102746,
      "grad_norm": 0.23744550347328186,
      "learning_rate": 1e-05,
      "loss": 1.039,
      "step": 17190
    },
    {
      "epoch": 15.230292294065544,
      "grad_norm": 0.22292733192443848,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 17195
    },
    {
      "epoch": 15.234720992028343,
      "grad_norm": 0.23988264799118042,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 17200
    },
    {
      "epoch": 15.239149689991143,
      "grad_norm": 0.22324340045452118,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 17205
    },
    {
      "epoch": 15.243578387953942,
      "grad_norm": 0.2578936815261841,
      "learning_rate": 1e-05,
      "loss": 1.0426,
      "step": 17210
    },
    {
      "epoch": 15.24800708591674,
      "grad_norm": 0.24282030761241913,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 17215
    },
    {
      "epoch": 15.25243578387954,
      "grad_norm": 0.25213000178337097,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 17220
    },
    {
      "epoch": 15.256864481842339,
      "grad_norm": 0.23575039207935333,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 17225
    },
    {
      "epoch": 15.261293179805138,
      "grad_norm": 0.38254740834236145,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 17230
    },
    {
      "epoch": 15.265721877767936,
      "grad_norm": 0.2365749627351761,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 17235
    },
    {
      "epoch": 15.270150575730735,
      "grad_norm": 0.26967090368270874,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 17240
    },
    {
      "epoch": 15.274579273693535,
      "grad_norm": 0.26564309000968933,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 17245
    },
    {
      "epoch": 15.279007971656332,
      "grad_norm": 0.31534746289253235,
      "learning_rate": 1e-05,
      "loss": 1.0384,
      "step": 17250
    },
    {
      "epoch": 15.283436669619132,
      "grad_norm": 0.294215589761734,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 17255
    },
    {
      "epoch": 15.287865367581931,
      "grad_norm": 0.26330846548080444,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 17260
    },
    {
      "epoch": 15.29229406554473,
      "grad_norm": 0.24788245558738708,
      "learning_rate": 1e-05,
      "loss": 1.0338,
      "step": 17265
    },
    {
      "epoch": 15.296722763507528,
      "grad_norm": 0.3743017911911011,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 17270
    },
    {
      "epoch": 15.301151461470328,
      "grad_norm": 0.27872133255004883,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 17275
    },
    {
      "epoch": 15.305580159433127,
      "grad_norm": 0.2927137017250061,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 17280
    },
    {
      "epoch": 15.310008857395925,
      "grad_norm": 0.2726747989654541,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 17285
    },
    {
      "epoch": 15.314437555358724,
      "grad_norm": 0.31740278005599976,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 17290
    },
    {
      "epoch": 15.318866253321524,
      "grad_norm": 0.2880260646343231,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 17295
    },
    {
      "epoch": 15.323294951284323,
      "grad_norm": 0.26660725474357605,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 17300
    },
    {
      "epoch": 15.32772364924712,
      "grad_norm": 0.26563549041748047,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 17305
    },
    {
      "epoch": 15.33215234720992,
      "grad_norm": 0.2138364464044571,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 17310
    },
    {
      "epoch": 15.33658104517272,
      "grad_norm": 0.288858562707901,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 17315
    },
    {
      "epoch": 15.341009743135519,
      "grad_norm": 0.3685668408870697,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 17320
    },
    {
      "epoch": 15.345438441098317,
      "grad_norm": 0.31687313318252563,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 17325
    },
    {
      "epoch": 15.349867139061116,
      "grad_norm": 0.2954452633857727,
      "learning_rate": 1e-05,
      "loss": 1.0448,
      "step": 17330
    },
    {
      "epoch": 15.354295837023916,
      "grad_norm": 0.2854871451854706,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 17335
    },
    {
      "epoch": 15.358724534986713,
      "grad_norm": 0.3266577422618866,
      "learning_rate": 1e-05,
      "loss": 1.017,
      "step": 17340
    },
    {
      "epoch": 15.363153232949513,
      "grad_norm": 0.25883886218070984,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 17345
    },
    {
      "epoch": 15.367581930912312,
      "grad_norm": 0.21190109848976135,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 17350
    },
    {
      "epoch": 15.372010628875111,
      "grad_norm": 0.26922836899757385,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 17355
    },
    {
      "epoch": 15.37643932683791,
      "grad_norm": 0.254291832447052,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 17360
    },
    {
      "epoch": 15.380868024800709,
      "grad_norm": 0.28723376989364624,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 17365
    },
    {
      "epoch": 15.385296722763508,
      "grad_norm": 0.2999916970729828,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 17370
    },
    {
      "epoch": 15.389725420726306,
      "grad_norm": 0.23715093731880188,
      "learning_rate": 1e-05,
      "loss": 1.0356,
      "step": 17375
    },
    {
      "epoch": 15.394154118689105,
      "grad_norm": 0.2447013556957245,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 17380
    },
    {
      "epoch": 15.398582816651905,
      "grad_norm": 0.24983754754066467,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 17385
    },
    {
      "epoch": 15.403011514614704,
      "grad_norm": 0.27488842606544495,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 17390
    },
    {
      "epoch": 15.407440212577502,
      "grad_norm": 0.3296554386615753,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 17395
    },
    {
      "epoch": 15.411868910540301,
      "grad_norm": 0.2354331761598587,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 17400
    },
    {
      "epoch": 15.4162976085031,
      "grad_norm": 0.27824997901916504,
      "learning_rate": 1e-05,
      "loss": 0.9305,
      "step": 17405
    },
    {
      "epoch": 15.420726306465898,
      "grad_norm": 0.2649856507778168,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 17410
    },
    {
      "epoch": 15.425155004428698,
      "grad_norm": 0.2622207701206207,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 17415
    },
    {
      "epoch": 15.429583702391497,
      "grad_norm": 0.24964140355587006,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 17420
    },
    {
      "epoch": 15.434012400354296,
      "grad_norm": 0.22974726557731628,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 17425
    },
    {
      "epoch": 15.438441098317094,
      "grad_norm": 0.2717423737049103,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 17430
    },
    {
      "epoch": 15.442869796279894,
      "grad_norm": 0.2755749821662903,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 17435
    },
    {
      "epoch": 15.447298494242693,
      "grad_norm": 0.22714024782180786,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 17440
    },
    {
      "epoch": 15.451727192205492,
      "grad_norm": 0.322359174489975,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 17445
    },
    {
      "epoch": 15.45615589016829,
      "grad_norm": 0.250095933675766,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 17450
    },
    {
      "epoch": 15.46058458813109,
      "grad_norm": 0.28568190336227417,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 17455
    },
    {
      "epoch": 15.465013286093889,
      "grad_norm": 0.2286706417798996,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 17460
    },
    {
      "epoch": 15.469441984056687,
      "grad_norm": 0.22113493084907532,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 17465
    },
    {
      "epoch": 15.473870682019486,
      "grad_norm": 0.2511354982852936,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 17470
    },
    {
      "epoch": 15.478299379982285,
      "grad_norm": 0.26192641258239746,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 17475
    },
    {
      "epoch": 15.482728077945085,
      "grad_norm": 0.31223440170288086,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 17480
    },
    {
      "epoch": 15.487156775907883,
      "grad_norm": 0.34236565232276917,
      "learning_rate": 1e-05,
      "loss": 1.0297,
      "step": 17485
    },
    {
      "epoch": 15.491585473870682,
      "grad_norm": 0.2933385372161865,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 17490
    },
    {
      "epoch": 15.496014171833481,
      "grad_norm": 0.24082890152931213,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 17495
    },
    {
      "epoch": 15.50044286979628,
      "grad_norm": 0.2509610056877136,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 17500
    },
    {
      "epoch": 15.504871567759078,
      "grad_norm": 0.34095117449760437,
      "learning_rate": 1e-05,
      "loss": 1.0432,
      "step": 17505
    },
    {
      "epoch": 15.509300265721878,
      "grad_norm": 0.26015010476112366,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 17510
    },
    {
      "epoch": 15.513728963684677,
      "grad_norm": 0.2837755084037781,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 17515
    },
    {
      "epoch": 15.518157661647475,
      "grad_norm": 0.2888769805431366,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 17520
    },
    {
      "epoch": 15.522586359610274,
      "grad_norm": 0.2791120111942291,
      "learning_rate": 1e-05,
      "loss": 1.0497,
      "step": 17525
    },
    {
      "epoch": 15.527015057573074,
      "grad_norm": 0.2765829563140869,
      "learning_rate": 1e-05,
      "loss": 1.0621,
      "step": 17530
    },
    {
      "epoch": 15.531443755535872,
      "grad_norm": 0.22506879270076752,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 17535
    },
    {
      "epoch": 15.535872453498671,
      "grad_norm": 0.288679301738739,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 17540
    },
    {
      "epoch": 15.54030115146147,
      "grad_norm": 0.27534568309783936,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 17545
    },
    {
      "epoch": 15.54472984942427,
      "grad_norm": 0.2309505194425583,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 17550
    },
    {
      "epoch": 15.549158547387067,
      "grad_norm": 0.2653706967830658,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 17555
    },
    {
      "epoch": 15.553587245349867,
      "grad_norm": 0.2603389322757721,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 17560
    },
    {
      "epoch": 15.558015943312666,
      "grad_norm": 0.24747580289840698,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 17565
    },
    {
      "epoch": 15.562444641275466,
      "grad_norm": 0.299694299697876,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 17570
    },
    {
      "epoch": 15.566873339238263,
      "grad_norm": 0.2646327018737793,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 17575
    },
    {
      "epoch": 15.571302037201063,
      "grad_norm": 0.297442227602005,
      "learning_rate": 1e-05,
      "loss": 0.9215,
      "step": 17580
    },
    {
      "epoch": 15.575730735163862,
      "grad_norm": 0.24240759015083313,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 17585
    },
    {
      "epoch": 15.58015943312666,
      "grad_norm": 0.25861620903015137,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 17590
    },
    {
      "epoch": 15.58458813108946,
      "grad_norm": 0.36966672539711,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 17595
    },
    {
      "epoch": 15.589016829052259,
      "grad_norm": 0.2738770544528961,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 17600
    },
    {
      "epoch": 15.593445527015058,
      "grad_norm": 0.2926274240016937,
      "learning_rate": 1e-05,
      "loss": 1.0309,
      "step": 17605
    },
    {
      "epoch": 15.597874224977856,
      "grad_norm": 0.40111541748046875,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 17610
    },
    {
      "epoch": 15.602302922940655,
      "grad_norm": 0.3129487633705139,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 17615
    },
    {
      "epoch": 15.606731620903455,
      "grad_norm": 0.3202768564224243,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 17620
    },
    {
      "epoch": 15.611160318866254,
      "grad_norm": 0.27919039130210876,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 17625
    },
    {
      "epoch": 15.615589016829052,
      "grad_norm": 0.25771623849868774,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 17630
    },
    {
      "epoch": 15.620017714791851,
      "grad_norm": 0.22565802931785583,
      "learning_rate": 1e-05,
      "loss": 1.0371,
      "step": 17635
    },
    {
      "epoch": 15.62444641275465,
      "grad_norm": 0.3081500232219696,
      "learning_rate": 1e-05,
      "loss": 1.0682,
      "step": 17640
    },
    {
      "epoch": 15.628875110717448,
      "grad_norm": 0.22506597638130188,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 17645
    },
    {
      "epoch": 15.633303808680248,
      "grad_norm": 0.24768738448619843,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 17650
    },
    {
      "epoch": 15.637732506643047,
      "grad_norm": 0.22748365998268127,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 17655
    },
    {
      "epoch": 15.642161204605847,
      "grad_norm": 0.2742569148540497,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 17660
    },
    {
      "epoch": 15.646589902568644,
      "grad_norm": 0.2463819533586502,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 17665
    },
    {
      "epoch": 15.651018600531444,
      "grad_norm": 0.25562024116516113,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 17670
    },
    {
      "epoch": 15.655447298494243,
      "grad_norm": 0.240044966340065,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 17675
    },
    {
      "epoch": 15.65987599645704,
      "grad_norm": 0.29254019260406494,
      "learning_rate": 1e-05,
      "loss": 0.9259,
      "step": 17680
    },
    {
      "epoch": 15.66430469441984,
      "grad_norm": 0.3491929769515991,
      "learning_rate": 1e-05,
      "loss": 1.0331,
      "step": 17685
    },
    {
      "epoch": 15.66873339238264,
      "grad_norm": 0.22034764289855957,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 17690
    },
    {
      "epoch": 15.67316209034544,
      "grad_norm": 0.21789421141147614,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 17695
    },
    {
      "epoch": 15.677590788308237,
      "grad_norm": 0.2813005745410919,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 17700
    },
    {
      "epoch": 15.682019486271036,
      "grad_norm": 0.2813015282154083,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 17705
    },
    {
      "epoch": 15.686448184233836,
      "grad_norm": 0.22260402143001556,
      "learning_rate": 1e-05,
      "loss": 1.028,
      "step": 17710
    },
    {
      "epoch": 15.690876882196633,
      "grad_norm": 0.24874848127365112,
      "learning_rate": 1e-05,
      "loss": 1.0218,
      "step": 17715
    },
    {
      "epoch": 15.695305580159433,
      "grad_norm": 0.24613982439041138,
      "learning_rate": 1e-05,
      "loss": 1.0398,
      "step": 17720
    },
    {
      "epoch": 15.699734278122232,
      "grad_norm": 0.3024621605873108,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 17725
    },
    {
      "epoch": 15.704162976085032,
      "grad_norm": 0.33095231652259827,
      "learning_rate": 1e-05,
      "loss": 0.9253,
      "step": 17730
    },
    {
      "epoch": 15.70859167404783,
      "grad_norm": 0.2548953890800476,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 17735
    },
    {
      "epoch": 15.713020372010629,
      "grad_norm": 0.24033677577972412,
      "learning_rate": 1e-05,
      "loss": 1.0228,
      "step": 17740
    },
    {
      "epoch": 15.717449069973428,
      "grad_norm": 0.27227887511253357,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 17745
    },
    {
      "epoch": 15.721877767936228,
      "grad_norm": 0.2598018944263458,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 17750
    },
    {
      "epoch": 15.726306465899025,
      "grad_norm": 0.24490581452846527,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 17755
    },
    {
      "epoch": 15.730735163861825,
      "grad_norm": 0.2886039614677429,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 17760
    },
    {
      "epoch": 15.735163861824624,
      "grad_norm": 0.26948562264442444,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 17765
    },
    {
      "epoch": 15.739592559787422,
      "grad_norm": 0.24950982630252838,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 17770
    },
    {
      "epoch": 15.744021257750221,
      "grad_norm": 0.3059339225292206,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 17775
    },
    {
      "epoch": 15.74844995571302,
      "grad_norm": 0.22925111651420593,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 17780
    },
    {
      "epoch": 15.75287865367582,
      "grad_norm": 0.3430158495903015,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 17785
    },
    {
      "epoch": 15.757307351638618,
      "grad_norm": 0.28006812930107117,
      "learning_rate": 1e-05,
      "loss": 1.0362,
      "step": 17790
    },
    {
      "epoch": 15.761736049601417,
      "grad_norm": 0.2532021105289459,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 17795
    },
    {
      "epoch": 15.766164747564217,
      "grad_norm": 0.2364649623632431,
      "learning_rate": 1e-05,
      "loss": 1.0297,
      "step": 17800
    },
    {
      "epoch": 15.770593445527014,
      "grad_norm": 0.290894091129303,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 17805
    },
    {
      "epoch": 15.775022143489814,
      "grad_norm": 0.29338327050209045,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 17810
    },
    {
      "epoch": 15.779450841452613,
      "grad_norm": 0.2634062170982361,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 17815
    },
    {
      "epoch": 15.783879539415413,
      "grad_norm": 0.25008612871170044,
      "learning_rate": 1e-05,
      "loss": 1.0432,
      "step": 17820
    },
    {
      "epoch": 15.78830823737821,
      "grad_norm": 0.21823376417160034,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 17825
    },
    {
      "epoch": 15.79273693534101,
      "grad_norm": 0.24414171278476715,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 17830
    },
    {
      "epoch": 15.797165633303809,
      "grad_norm": 0.2700551152229309,
      "learning_rate": 1e-05,
      "loss": 1.0561,
      "step": 17835
    },
    {
      "epoch": 15.801594331266607,
      "grad_norm": 0.28298768401145935,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 17840
    },
    {
      "epoch": 15.806023029229406,
      "grad_norm": 0.31701624393463135,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 17845
    },
    {
      "epoch": 15.810451727192206,
      "grad_norm": 0.3518561124801636,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 17850
    },
    {
      "epoch": 15.814880425155005,
      "grad_norm": 0.27521559596061707,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 17855
    },
    {
      "epoch": 15.819309123117803,
      "grad_norm": 0.24579580128192902,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 17860
    },
    {
      "epoch": 15.823737821080602,
      "grad_norm": 0.2322177141904831,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 17865
    },
    {
      "epoch": 15.828166519043402,
      "grad_norm": 0.2954336106777191,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 17870
    },
    {
      "epoch": 15.832595217006201,
      "grad_norm": 0.24369779229164124,
      "learning_rate": 1e-05,
      "loss": 1.0155,
      "step": 17875
    },
    {
      "epoch": 15.837023914968999,
      "grad_norm": 0.26766467094421387,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 17880
    },
    {
      "epoch": 15.841452612931798,
      "grad_norm": 0.3284904360771179,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 17885
    },
    {
      "epoch": 15.845881310894598,
      "grad_norm": 0.3282492160797119,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 17890
    },
    {
      "epoch": 15.850310008857395,
      "grad_norm": 0.31012794375419617,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 17895
    },
    {
      "epoch": 15.854738706820195,
      "grad_norm": 0.23211978375911713,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 17900
    },
    {
      "epoch": 15.859167404782994,
      "grad_norm": 0.30380022525787354,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 17905
    },
    {
      "epoch": 15.863596102745793,
      "grad_norm": 0.25331732630729675,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 17910
    },
    {
      "epoch": 15.868024800708591,
      "grad_norm": 0.2511480152606964,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 17915
    },
    {
      "epoch": 15.87245349867139,
      "grad_norm": 0.2963160276412964,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 17920
    },
    {
      "epoch": 15.87688219663419,
      "grad_norm": 0.254862517118454,
      "learning_rate": 1e-05,
      "loss": 1.0235,
      "step": 17925
    },
    {
      "epoch": 15.88131089459699,
      "grad_norm": 0.2827042043209076,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 17930
    },
    {
      "epoch": 15.885739592559787,
      "grad_norm": 0.2421710193157196,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 17935
    },
    {
      "epoch": 15.890168290522587,
      "grad_norm": 0.2819615602493286,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 17940
    },
    {
      "epoch": 15.894596988485386,
      "grad_norm": 0.3103281855583191,
      "learning_rate": 1e-05,
      "loss": 1.0219,
      "step": 17945
    },
    {
      "epoch": 15.899025686448184,
      "grad_norm": 0.25584855675697327,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 17950
    },
    {
      "epoch": 15.903454384410983,
      "grad_norm": 0.24295596778392792,
      "learning_rate": 1e-05,
      "loss": 1.0319,
      "step": 17955
    },
    {
      "epoch": 15.907883082373782,
      "grad_norm": 0.2826685309410095,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 17960
    },
    {
      "epoch": 15.91231178033658,
      "grad_norm": 0.29204365611076355,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 17965
    },
    {
      "epoch": 15.91674047829938,
      "grad_norm": 0.2577894628047943,
      "learning_rate": 1e-05,
      "loss": 1.0475,
      "step": 17970
    },
    {
      "epoch": 15.921169176262179,
      "grad_norm": 0.24750283360481262,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 17975
    },
    {
      "epoch": 15.925597874224978,
      "grad_norm": 0.2631934881210327,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 17980
    },
    {
      "epoch": 15.930026572187776,
      "grad_norm": 0.3149290680885315,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 17985
    },
    {
      "epoch": 15.934455270150576,
      "grad_norm": 0.3718342185020447,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 17990
    },
    {
      "epoch": 15.938883968113375,
      "grad_norm": 0.2407686859369278,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 17995
    },
    {
      "epoch": 15.943312666076174,
      "grad_norm": 0.34004491567611694,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 18000
    },
    {
      "epoch": 15.947741364038972,
      "grad_norm": 0.25165820121765137,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 18005
    },
    {
      "epoch": 15.952170062001771,
      "grad_norm": 0.25692176818847656,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 18010
    },
    {
      "epoch": 15.956598759964571,
      "grad_norm": 0.26404044032096863,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 18015
    },
    {
      "epoch": 15.961027457927369,
      "grad_norm": 0.2877795398235321,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 18020
    },
    {
      "epoch": 15.965456155890168,
      "grad_norm": 0.2511865198612213,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 18025
    },
    {
      "epoch": 15.969884853852967,
      "grad_norm": 0.24652396142482758,
      "learning_rate": 1e-05,
      "loss": 1.0372,
      "step": 18030
    },
    {
      "epoch": 15.974313551815767,
      "grad_norm": 0.2523021697998047,
      "learning_rate": 1e-05,
      "loss": 1.0665,
      "step": 18035
    },
    {
      "epoch": 15.978742249778564,
      "grad_norm": 0.23684607446193695,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 18040
    },
    {
      "epoch": 15.983170947741364,
      "grad_norm": 0.25142917037010193,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 18045
    },
    {
      "epoch": 15.987599645704163,
      "grad_norm": 0.2796723246574402,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 18050
    },
    {
      "epoch": 15.992028343666963,
      "grad_norm": 0.26551374793052673,
      "learning_rate": 1e-05,
      "loss": 1.0363,
      "step": 18055
    },
    {
      "epoch": 15.99645704162976,
      "grad_norm": 0.2703036963939667,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 18060
    },
    {
      "epoch": 16.000885739592558,
      "grad_norm": 0.24998635053634644,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 18065
    },
    {
      "epoch": 16.00531443755536,
      "grad_norm": 0.24102988839149475,
      "learning_rate": 1e-05,
      "loss": 1.0682,
      "step": 18070
    },
    {
      "epoch": 16.009743135518157,
      "grad_norm": 0.23604339361190796,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 18075
    },
    {
      "epoch": 16.014171833480958,
      "grad_norm": 0.23942531645298004,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 18080
    },
    {
      "epoch": 16.018600531443756,
      "grad_norm": 0.2519570291042328,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 18085
    },
    {
      "epoch": 16.023029229406553,
      "grad_norm": 0.30462661385536194,
      "learning_rate": 1e-05,
      "loss": 1.0506,
      "step": 18090
    },
    {
      "epoch": 16.027457927369355,
      "grad_norm": 0.24723510444164276,
      "learning_rate": 1e-05,
      "loss": 1.0329,
      "step": 18095
    },
    {
      "epoch": 16.031886625332152,
      "grad_norm": 0.23702853918075562,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 18100
    },
    {
      "epoch": 16.03631532329495,
      "grad_norm": 0.24404092133045197,
      "learning_rate": 1e-05,
      "loss": 1.0329,
      "step": 18105
    },
    {
      "epoch": 16.04074402125775,
      "grad_norm": 0.2478272020816803,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 18110
    },
    {
      "epoch": 16.04517271922055,
      "grad_norm": 0.32819488644599915,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 18115
    },
    {
      "epoch": 16.049601417183347,
      "grad_norm": 0.31817978620529175,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 18120
    },
    {
      "epoch": 16.054030115146148,
      "grad_norm": 0.31144028902053833,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 18125
    },
    {
      "epoch": 16.058458813108945,
      "grad_norm": 0.23166677355766296,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 18130
    },
    {
      "epoch": 16.062887511071747,
      "grad_norm": 0.26017239689826965,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 18135
    },
    {
      "epoch": 16.067316209034544,
      "grad_norm": 0.2920660078525543,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 18140
    },
    {
      "epoch": 16.071744906997342,
      "grad_norm": 0.2695963382720947,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 18145
    },
    {
      "epoch": 16.076173604960143,
      "grad_norm": 0.27189207077026367,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 18150
    },
    {
      "epoch": 16.08060230292294,
      "grad_norm": 0.257854700088501,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 18155
    },
    {
      "epoch": 16.08503100088574,
      "grad_norm": 0.31605926156044006,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 18160
    },
    {
      "epoch": 16.08945969884854,
      "grad_norm": 0.2787530720233917,
      "learning_rate": 1e-05,
      "loss": 1.0505,
      "step": 18165
    },
    {
      "epoch": 16.093888396811337,
      "grad_norm": 0.33719149231910706,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 18170
    },
    {
      "epoch": 16.098317094774135,
      "grad_norm": 0.32978686690330505,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 18175
    },
    {
      "epoch": 16.102745792736936,
      "grad_norm": 0.3089555501937866,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 18180
    },
    {
      "epoch": 16.107174490699734,
      "grad_norm": 0.23977087438106537,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 18185
    },
    {
      "epoch": 16.11160318866253,
      "grad_norm": 0.265188068151474,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 18190
    },
    {
      "epoch": 16.116031886625333,
      "grad_norm": 0.2618181109428406,
      "learning_rate": 1e-05,
      "loss": 1.0382,
      "step": 18195
    },
    {
      "epoch": 16.12046058458813,
      "grad_norm": 0.21465323865413666,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 18200
    },
    {
      "epoch": 16.12488928255093,
      "grad_norm": 0.2547129690647125,
      "learning_rate": 1e-05,
      "loss": 1.0187,
      "step": 18205
    },
    {
      "epoch": 16.12931798051373,
      "grad_norm": 0.2909153997898102,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 18210
    },
    {
      "epoch": 16.133746678476527,
      "grad_norm": 0.2824973464012146,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 18215
    },
    {
      "epoch": 16.138175376439328,
      "grad_norm": 0.2601968050003052,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 18220
    },
    {
      "epoch": 16.142604074402126,
      "grad_norm": 0.30577826499938965,
      "learning_rate": 1e-05,
      "loss": 1.0494,
      "step": 18225
    },
    {
      "epoch": 16.147032772364923,
      "grad_norm": 0.25143304467201233,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 18230
    },
    {
      "epoch": 16.151461470327725,
      "grad_norm": 0.2998024523258209,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 18235
    },
    {
      "epoch": 16.155890168290522,
      "grad_norm": 0.2602280080318451,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 18240
    },
    {
      "epoch": 16.16031886625332,
      "grad_norm": 0.31770411133766174,
      "learning_rate": 1e-05,
      "loss": 1.0739,
      "step": 18245
    },
    {
      "epoch": 16.16474756421612,
      "grad_norm": 0.22414326667785645,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 18250
    },
    {
      "epoch": 16.16917626217892,
      "grad_norm": 0.24530647695064545,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 18255
    },
    {
      "epoch": 16.17360496014172,
      "grad_norm": 0.2370666265487671,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 18260
    },
    {
      "epoch": 16.178033658104518,
      "grad_norm": 0.27960169315338135,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 18265
    },
    {
      "epoch": 16.182462356067315,
      "grad_norm": 0.25410163402557373,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 18270
    },
    {
      "epoch": 16.186891054030117,
      "grad_norm": 0.2481650412082672,
      "learning_rate": 1e-05,
      "loss": 0.9151,
      "step": 18275
    },
    {
      "epoch": 16.191319751992914,
      "grad_norm": 0.25486287474632263,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 18280
    },
    {
      "epoch": 16.195748449955712,
      "grad_norm": 0.2724369764328003,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 18285
    },
    {
      "epoch": 16.200177147918513,
      "grad_norm": 0.34363308548927307,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 18290
    },
    {
      "epoch": 16.20460584588131,
      "grad_norm": 0.3404398262500763,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 18295
    },
    {
      "epoch": 16.20903454384411,
      "grad_norm": 0.2944444715976715,
      "learning_rate": 1e-05,
      "loss": 1.042,
      "step": 18300
    },
    {
      "epoch": 16.21346324180691,
      "grad_norm": 0.23392118513584137,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 18305
    },
    {
      "epoch": 16.217891939769707,
      "grad_norm": 0.2643710970878601,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 18310
    },
    {
      "epoch": 16.22232063773251,
      "grad_norm": 0.3132951855659485,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 18315
    },
    {
      "epoch": 16.226749335695306,
      "grad_norm": 0.2576334476470947,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 18320
    },
    {
      "epoch": 16.231178033658104,
      "grad_norm": 0.2385687530040741,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 18325
    },
    {
      "epoch": 16.235606731620905,
      "grad_norm": 0.2157694697380066,
      "learning_rate": 1e-05,
      "loss": 1.0194,
      "step": 18330
    },
    {
      "epoch": 16.240035429583703,
      "grad_norm": 0.28696709871292114,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 18335
    },
    {
      "epoch": 16.2444641275465,
      "grad_norm": 0.25031882524490356,
      "learning_rate": 1e-05,
      "loss": 1.052,
      "step": 18340
    },
    {
      "epoch": 16.2488928255093,
      "grad_norm": 0.2438434660434723,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 18345
    },
    {
      "epoch": 16.2533215234721,
      "grad_norm": 0.2602662146091461,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 18350
    },
    {
      "epoch": 16.257750221434897,
      "grad_norm": 0.24205753207206726,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 18355
    },
    {
      "epoch": 16.262178919397698,
      "grad_norm": 0.30614742636680603,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 18360
    },
    {
      "epoch": 16.266607617360496,
      "grad_norm": 0.2930641770362854,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 18365
    },
    {
      "epoch": 16.271036315323293,
      "grad_norm": 0.2607276439666748,
      "learning_rate": 1e-05,
      "loss": 0.918,
      "step": 18370
    },
    {
      "epoch": 16.275465013286095,
      "grad_norm": 0.32311829924583435,
      "learning_rate": 1e-05,
      "loss": 1.031,
      "step": 18375
    },
    {
      "epoch": 16.279893711248892,
      "grad_norm": 0.2691332697868347,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 18380
    },
    {
      "epoch": 16.284322409211693,
      "grad_norm": 0.2581843435764313,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 18385
    },
    {
      "epoch": 16.28875110717449,
      "grad_norm": 0.23530881106853485,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 18390
    },
    {
      "epoch": 16.29317980513729,
      "grad_norm": 0.2343352735042572,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 18395
    },
    {
      "epoch": 16.29760850310009,
      "grad_norm": 0.236292764544487,
      "learning_rate": 1e-05,
      "loss": 1.0618,
      "step": 18400
    },
    {
      "epoch": 16.302037201062888,
      "grad_norm": 0.2484918087720871,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 18405
    },
    {
      "epoch": 16.306465899025685,
      "grad_norm": 0.2452331930398941,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 18410
    },
    {
      "epoch": 16.310894596988486,
      "grad_norm": 0.23057523369789124,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 18415
    },
    {
      "epoch": 16.315323294951284,
      "grad_norm": 0.27240073680877686,
      "learning_rate": 1e-05,
      "loss": 1.0333,
      "step": 18420
    },
    {
      "epoch": 16.31975199291408,
      "grad_norm": 0.24729584157466888,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 18425
    },
    {
      "epoch": 16.324180690876883,
      "grad_norm": 0.2590065598487854,
      "learning_rate": 1e-05,
      "loss": 1.0326,
      "step": 18430
    },
    {
      "epoch": 16.32860938883968,
      "grad_norm": 0.21345050632953644,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 18435
    },
    {
      "epoch": 16.333038086802482,
      "grad_norm": 0.23559211194515228,
      "learning_rate": 1e-05,
      "loss": 1.0345,
      "step": 18440
    },
    {
      "epoch": 16.33746678476528,
      "grad_norm": 0.23431992530822754,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 18445
    },
    {
      "epoch": 16.341895482728077,
      "grad_norm": 0.22062960267066956,
      "learning_rate": 1e-05,
      "loss": 1.082,
      "step": 18450
    },
    {
      "epoch": 16.34632418069088,
      "grad_norm": 0.2790276110172272,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 18455
    },
    {
      "epoch": 16.350752878653676,
      "grad_norm": 0.2819046974182129,
      "learning_rate": 1e-05,
      "loss": 1.0389,
      "step": 18460
    },
    {
      "epoch": 16.355181576616474,
      "grad_norm": 0.25946441292762756,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 18465
    },
    {
      "epoch": 16.359610274579275,
      "grad_norm": 0.3128257691860199,
      "learning_rate": 1e-05,
      "loss": 1.0208,
      "step": 18470
    },
    {
      "epoch": 16.364038972542073,
      "grad_norm": 0.28413891792297363,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 18475
    },
    {
      "epoch": 16.36846767050487,
      "grad_norm": 0.27322763204574585,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 18480
    },
    {
      "epoch": 16.37289636846767,
      "grad_norm": 0.28998783230781555,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 18485
    },
    {
      "epoch": 16.37732506643047,
      "grad_norm": 0.2913134694099426,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 18490
    },
    {
      "epoch": 16.381753764393267,
      "grad_norm": 0.226438969373703,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 18495
    },
    {
      "epoch": 16.386182462356068,
      "grad_norm": 0.2568718194961548,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 18500
    },
    {
      "epoch": 16.390611160318866,
      "grad_norm": 0.24007783830165863,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 18505
    },
    {
      "epoch": 16.395039858281667,
      "grad_norm": 0.2564709782600403,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 18510
    },
    {
      "epoch": 16.399468556244464,
      "grad_norm": 0.23964142799377441,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 18515
    },
    {
      "epoch": 16.403897254207262,
      "grad_norm": 0.2868022322654724,
      "learning_rate": 1e-05,
      "loss": 1.0464,
      "step": 18520
    },
    {
      "epoch": 16.408325952170063,
      "grad_norm": 0.2683990001678467,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 18525
    },
    {
      "epoch": 16.41275465013286,
      "grad_norm": 0.27346548438072205,
      "learning_rate": 1e-05,
      "loss": 1.0497,
      "step": 18530
    },
    {
      "epoch": 16.41718334809566,
      "grad_norm": 0.2398541420698166,
      "learning_rate": 1e-05,
      "loss": 1.0379,
      "step": 18535
    },
    {
      "epoch": 16.42161204605846,
      "grad_norm": 0.2636696696281433,
      "learning_rate": 1e-05,
      "loss": 1.0247,
      "step": 18540
    },
    {
      "epoch": 16.426040744021257,
      "grad_norm": 0.2778176963329315,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 18545
    },
    {
      "epoch": 16.430469441984055,
      "grad_norm": 0.24303869903087616,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 18550
    },
    {
      "epoch": 16.434898139946856,
      "grad_norm": 0.21459044516086578,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 18555
    },
    {
      "epoch": 16.439326837909654,
      "grad_norm": 0.2780698835849762,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 18560
    },
    {
      "epoch": 16.443755535872455,
      "grad_norm": 0.211567685008049,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 18565
    },
    {
      "epoch": 16.448184233835253,
      "grad_norm": 0.30449023842811584,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 18570
    },
    {
      "epoch": 16.45261293179805,
      "grad_norm": 0.2632780373096466,
      "learning_rate": 1e-05,
      "loss": 1.0269,
      "step": 18575
    },
    {
      "epoch": 16.45704162976085,
      "grad_norm": 0.28514331579208374,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 18580
    },
    {
      "epoch": 16.46147032772365,
      "grad_norm": 0.2687274217605591,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 18585
    },
    {
      "epoch": 16.465899025686447,
      "grad_norm": 0.304848313331604,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 18590
    },
    {
      "epoch": 16.47032772364925,
      "grad_norm": 0.25122809410095215,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 18595
    },
    {
      "epoch": 16.474756421612046,
      "grad_norm": 0.22518500685691833,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 18600
    },
    {
      "epoch": 16.479185119574844,
      "grad_norm": 0.20703738927841187,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 18605
    },
    {
      "epoch": 16.483613817537645,
      "grad_norm": 0.2702414095401764,
      "learning_rate": 1e-05,
      "loss": 1.0283,
      "step": 18610
    },
    {
      "epoch": 16.488042515500442,
      "grad_norm": 0.2425641268491745,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 18615
    },
    {
      "epoch": 16.49247121346324,
      "grad_norm": 0.2771373689174652,
      "learning_rate": 1e-05,
      "loss": 1.022,
      "step": 18620
    },
    {
      "epoch": 16.49689991142604,
      "grad_norm": 0.25770580768585205,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 18625
    },
    {
      "epoch": 16.50132860938884,
      "grad_norm": 0.28298768401145935,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 18630
    },
    {
      "epoch": 16.50575730735164,
      "grad_norm": 0.27143004536628723,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 18635
    },
    {
      "epoch": 16.510186005314438,
      "grad_norm": 0.25454238057136536,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 18640
    },
    {
      "epoch": 16.514614703277235,
      "grad_norm": 0.2803010046482086,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 18645
    },
    {
      "epoch": 16.519043401240037,
      "grad_norm": 0.27099743485450745,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 18650
    },
    {
      "epoch": 16.523472099202834,
      "grad_norm": 0.3108830749988556,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 18655
    },
    {
      "epoch": 16.527900797165632,
      "grad_norm": 0.24949656426906586,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 18660
    },
    {
      "epoch": 16.532329495128433,
      "grad_norm": 0.28336501121520996,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 18665
    },
    {
      "epoch": 16.53675819309123,
      "grad_norm": 0.23413774371147156,
      "learning_rate": 1e-05,
      "loss": 1.0506,
      "step": 18670
    },
    {
      "epoch": 16.54118689105403,
      "grad_norm": 0.2716546952724457,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 18675
    },
    {
      "epoch": 16.54561558901683,
      "grad_norm": 0.28943341970443726,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 18680
    },
    {
      "epoch": 16.550044286979627,
      "grad_norm": 0.2586379945278168,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 18685
    },
    {
      "epoch": 16.55447298494243,
      "grad_norm": 0.29518845677375793,
      "learning_rate": 1e-05,
      "loss": 1.0241,
      "step": 18690
    },
    {
      "epoch": 16.558901682905226,
      "grad_norm": 0.2944999039173126,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 18695
    },
    {
      "epoch": 16.563330380868024,
      "grad_norm": 0.27595025300979614,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 18700
    },
    {
      "epoch": 16.567759078830825,
      "grad_norm": 0.3008978068828583,
      "learning_rate": 1e-05,
      "loss": 1.0224,
      "step": 18705
    },
    {
      "epoch": 16.572187776793623,
      "grad_norm": 0.3106701672077179,
      "learning_rate": 1e-05,
      "loss": 1.0561,
      "step": 18710
    },
    {
      "epoch": 16.57661647475642,
      "grad_norm": 0.2804649770259857,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 18715
    },
    {
      "epoch": 16.58104517271922,
      "grad_norm": 0.24391058087348938,
      "learning_rate": 1e-05,
      "loss": 1.0379,
      "step": 18720
    },
    {
      "epoch": 16.58547387068202,
      "grad_norm": 0.2532961368560791,
      "learning_rate": 1e-05,
      "loss": 1.0526,
      "step": 18725
    },
    {
      "epoch": 16.589902568644817,
      "grad_norm": 0.23720023036003113,
      "learning_rate": 1e-05,
      "loss": 0.9325,
      "step": 18730
    },
    {
      "epoch": 16.594331266607618,
      "grad_norm": 0.23242931067943573,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 18735
    },
    {
      "epoch": 16.598759964570416,
      "grad_norm": 0.27134084701538086,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 18740
    },
    {
      "epoch": 16.603188662533213,
      "grad_norm": 0.26074808835983276,
      "learning_rate": 1e-05,
      "loss": 0.9432,
      "step": 18745
    },
    {
      "epoch": 16.607617360496015,
      "grad_norm": 0.23537185788154602,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 18750
    },
    {
      "epoch": 16.612046058458812,
      "grad_norm": 0.27737313508987427,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 18755
    },
    {
      "epoch": 16.616474756421614,
      "grad_norm": 0.2559964656829834,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 18760
    },
    {
      "epoch": 16.62090345438441,
      "grad_norm": 0.2800896167755127,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 18765
    },
    {
      "epoch": 16.62533215234721,
      "grad_norm": 0.26467326283454895,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 18770
    },
    {
      "epoch": 16.62976085031001,
      "grad_norm": 0.21582509577274323,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 18775
    },
    {
      "epoch": 16.634189548272808,
      "grad_norm": 0.2690695822238922,
      "learning_rate": 1e-05,
      "loss": 1.0407,
      "step": 18780
    },
    {
      "epoch": 16.638618246235605,
      "grad_norm": 0.296164870262146,
      "learning_rate": 1e-05,
      "loss": 0.9292,
      "step": 18785
    },
    {
      "epoch": 16.643046944198407,
      "grad_norm": 0.2603996992111206,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 18790
    },
    {
      "epoch": 16.647475642161204,
      "grad_norm": 0.2523553669452667,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 18795
    },
    {
      "epoch": 16.651904340124002,
      "grad_norm": 0.221808522939682,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 18800
    },
    {
      "epoch": 16.656333038086803,
      "grad_norm": 0.29286250472068787,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 18805
    },
    {
      "epoch": 16.6607617360496,
      "grad_norm": 0.3606424033641815,
      "learning_rate": 1e-05,
      "loss": 1.0284,
      "step": 18810
    },
    {
      "epoch": 16.665190434012402,
      "grad_norm": 0.25356325507164,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 18815
    },
    {
      "epoch": 16.6696191319752,
      "grad_norm": 0.2800666093826294,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 18820
    },
    {
      "epoch": 16.674047829937997,
      "grad_norm": 0.23566178977489471,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 18825
    },
    {
      "epoch": 16.6784765279008,
      "grad_norm": 0.24341733753681183,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 18830
    },
    {
      "epoch": 16.682905225863596,
      "grad_norm": 0.29197293519973755,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 18835
    },
    {
      "epoch": 16.687333923826394,
      "grad_norm": 0.29102814197540283,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 18840
    },
    {
      "epoch": 16.691762621789195,
      "grad_norm": 0.2886674404144287,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 18845
    },
    {
      "epoch": 16.696191319751993,
      "grad_norm": 0.2126307338476181,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 18850
    },
    {
      "epoch": 16.70062001771479,
      "grad_norm": 0.23335541784763336,
      "learning_rate": 1e-05,
      "loss": 1.0548,
      "step": 18855
    },
    {
      "epoch": 16.70504871567759,
      "grad_norm": 0.29025331139564514,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 18860
    },
    {
      "epoch": 16.70947741364039,
      "grad_norm": 0.3079996407032013,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 18865
    },
    {
      "epoch": 16.713906111603187,
      "grad_norm": 0.26167216897010803,
      "learning_rate": 1e-05,
      "loss": 1.0281,
      "step": 18870
    },
    {
      "epoch": 16.718334809565988,
      "grad_norm": 0.22602461278438568,
      "learning_rate": 1e-05,
      "loss": 1.0407,
      "step": 18875
    },
    {
      "epoch": 16.722763507528786,
      "grad_norm": 0.2575356364250183,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 18880
    },
    {
      "epoch": 16.727192205491587,
      "grad_norm": 0.2534136474132538,
      "learning_rate": 1e-05,
      "loss": 1.0265,
      "step": 18885
    },
    {
      "epoch": 16.731620903454385,
      "grad_norm": 0.263756662607193,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 18890
    },
    {
      "epoch": 16.736049601417182,
      "grad_norm": 0.2415543496608734,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 18895
    },
    {
      "epoch": 16.740478299379983,
      "grad_norm": 0.29941806197166443,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 18900
    },
    {
      "epoch": 16.74490699734278,
      "grad_norm": 0.3705863356590271,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 18905
    },
    {
      "epoch": 16.74933569530558,
      "grad_norm": 0.28699252009391785,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 18910
    },
    {
      "epoch": 16.75376439326838,
      "grad_norm": 0.27103716135025024,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 18915
    },
    {
      "epoch": 16.758193091231178,
      "grad_norm": 0.29779547452926636,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 18920
    },
    {
      "epoch": 16.762621789193975,
      "grad_norm": 0.2174983024597168,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 18925
    },
    {
      "epoch": 16.767050487156776,
      "grad_norm": 0.24685895442962646,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 18930
    },
    {
      "epoch": 16.771479185119574,
      "grad_norm": 0.25331911444664,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 18935
    },
    {
      "epoch": 16.775907883082375,
      "grad_norm": 0.23541639745235443,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 18940
    },
    {
      "epoch": 16.780336581045173,
      "grad_norm": 0.2841220796108246,
      "learning_rate": 1e-05,
      "loss": 1.0241,
      "step": 18945
    },
    {
      "epoch": 16.78476527900797,
      "grad_norm": 0.32827258110046387,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 18950
    },
    {
      "epoch": 16.789193976970772,
      "grad_norm": 0.24203045666217804,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 18955
    },
    {
      "epoch": 16.79362267493357,
      "grad_norm": 0.2676389217376709,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 18960
    },
    {
      "epoch": 16.798051372896367,
      "grad_norm": 0.24541980028152466,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 18965
    },
    {
      "epoch": 16.80248007085917,
      "grad_norm": 0.24969732761383057,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 18970
    },
    {
      "epoch": 16.806908768821966,
      "grad_norm": 0.2652243375778198,
      "learning_rate": 1e-05,
      "loss": 0.9153,
      "step": 18975
    },
    {
      "epoch": 16.811337466784764,
      "grad_norm": 0.2739236056804657,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 18980
    },
    {
      "epoch": 16.815766164747565,
      "grad_norm": 0.3239922523498535,
      "learning_rate": 1e-05,
      "loss": 1.0515,
      "step": 18985
    },
    {
      "epoch": 16.820194862710363,
      "grad_norm": 0.24756105244159698,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 18990
    },
    {
      "epoch": 16.824623560673164,
      "grad_norm": 0.2689042091369629,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 18995
    },
    {
      "epoch": 16.82905225863596,
      "grad_norm": 0.2473742961883545,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 19000
    },
    {
      "epoch": 16.83348095659876,
      "grad_norm": 0.2622603178024292,
      "learning_rate": 1e-05,
      "loss": 0.9211,
      "step": 19005
    },
    {
      "epoch": 16.83790965456156,
      "grad_norm": 0.3478242754936218,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 19010
    },
    {
      "epoch": 16.842338352524358,
      "grad_norm": 0.2884719669818878,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 19015
    },
    {
      "epoch": 16.846767050487156,
      "grad_norm": 0.23757636547088623,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 19020
    },
    {
      "epoch": 16.851195748449957,
      "grad_norm": 0.25557246804237366,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 19025
    },
    {
      "epoch": 16.855624446412754,
      "grad_norm": 0.24929285049438477,
      "learning_rate": 1e-05,
      "loss": 1.0502,
      "step": 19030
    },
    {
      "epoch": 16.860053144375552,
      "grad_norm": 0.2467043548822403,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 19035
    },
    {
      "epoch": 16.864481842338353,
      "grad_norm": 0.24875454604625702,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 19040
    },
    {
      "epoch": 16.86891054030115,
      "grad_norm": 0.24771851301193237,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 19045
    },
    {
      "epoch": 16.873339238263952,
      "grad_norm": 0.2596139907836914,
      "learning_rate": 1e-05,
      "loss": 1.036,
      "step": 19050
    },
    {
      "epoch": 16.87776793622675,
      "grad_norm": 0.2807680070400238,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 19055
    },
    {
      "epoch": 16.882196634189548,
      "grad_norm": 0.25339943170547485,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 19060
    },
    {
      "epoch": 16.88662533215235,
      "grad_norm": 0.29721444845199585,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 19065
    },
    {
      "epoch": 16.891054030115146,
      "grad_norm": 0.2381032556295395,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 19070
    },
    {
      "epoch": 16.895482728077944,
      "grad_norm": 0.2421633005142212,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 19075
    },
    {
      "epoch": 16.899911426040745,
      "grad_norm": 0.3250364065170288,
      "learning_rate": 1e-05,
      "loss": 1.0199,
      "step": 19080
    },
    {
      "epoch": 16.904340124003543,
      "grad_norm": 0.2659304141998291,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 19085
    },
    {
      "epoch": 16.90876882196634,
      "grad_norm": 0.3063162565231323,
      "learning_rate": 1e-05,
      "loss": 1.0085,
      "step": 19090
    },
    {
      "epoch": 16.913197519929142,
      "grad_norm": 0.38583600521087646,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 19095
    },
    {
      "epoch": 16.91762621789194,
      "grad_norm": 0.2514832615852356,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 19100
    },
    {
      "epoch": 16.922054915854737,
      "grad_norm": 0.2967970371246338,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 19105
    },
    {
      "epoch": 16.92648361381754,
      "grad_norm": 0.2786117494106293,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 19110
    },
    {
      "epoch": 16.930912311780336,
      "grad_norm": 0.2897469103336334,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 19115
    },
    {
      "epoch": 16.935341009743137,
      "grad_norm": 0.25544604659080505,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 19120
    },
    {
      "epoch": 16.939769707705935,
      "grad_norm": 0.2646866738796234,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 19125
    },
    {
      "epoch": 16.944198405668732,
      "grad_norm": 0.28695422410964966,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 19130
    },
    {
      "epoch": 16.948627103631534,
      "grad_norm": 0.22553794085979462,
      "learning_rate": 1e-05,
      "loss": 1.0261,
      "step": 19135
    },
    {
      "epoch": 16.95305580159433,
      "grad_norm": 0.2556069493293762,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 19140
    },
    {
      "epoch": 16.95748449955713,
      "grad_norm": 0.24725207686424255,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 19145
    },
    {
      "epoch": 16.96191319751993,
      "grad_norm": 0.2623295485973358,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 19150
    },
    {
      "epoch": 16.966341895482728,
      "grad_norm": 0.2691655158996582,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 19155
    },
    {
      "epoch": 16.970770593445526,
      "grad_norm": 0.25209763646125793,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 19160
    },
    {
      "epoch": 16.975199291408327,
      "grad_norm": 0.23078106343746185,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 19165
    },
    {
      "epoch": 16.979627989371124,
      "grad_norm": 0.2285294383764267,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 19170
    },
    {
      "epoch": 16.984056687333926,
      "grad_norm": 0.2527830898761749,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 19175
    },
    {
      "epoch": 16.988485385296723,
      "grad_norm": 0.26528918743133545,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 19180
    },
    {
      "epoch": 16.99291408325952,
      "grad_norm": 0.27150407433509827,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 19185
    },
    {
      "epoch": 16.997342781222322,
      "grad_norm": 0.2638486325740814,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 19190
    },
    {
      "epoch": 17.00177147918512,
      "grad_norm": 0.27760282158851624,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 19195
    },
    {
      "epoch": 17.006200177147917,
      "grad_norm": 0.28149768710136414,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 19200
    },
    {
      "epoch": 17.01062887511072,
      "grad_norm": 0.23505602777004242,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 19205
    },
    {
      "epoch": 17.015057573073516,
      "grad_norm": 0.2850729823112488,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 19210
    },
    {
      "epoch": 17.019486271036314,
      "grad_norm": 0.24557830393314362,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 19215
    },
    {
      "epoch": 17.023914968999115,
      "grad_norm": 0.29252949357032776,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 19220
    },
    {
      "epoch": 17.028343666961913,
      "grad_norm": 0.2909540832042694,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 19225
    },
    {
      "epoch": 17.03277236492471,
      "grad_norm": 0.25772130489349365,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 19230
    },
    {
      "epoch": 17.03720106288751,
      "grad_norm": 0.2648557424545288,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 19235
    },
    {
      "epoch": 17.04162976085031,
      "grad_norm": 0.23941372334957123,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 19240
    },
    {
      "epoch": 17.04605845881311,
      "grad_norm": 0.24207670986652374,
      "learning_rate": 1e-05,
      "loss": 0.9522,
      "step": 19245
    },
    {
      "epoch": 17.050487156775908,
      "grad_norm": 0.2905253469944,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 19250
    },
    {
      "epoch": 17.054915854738706,
      "grad_norm": 0.257564514875412,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 19255
    },
    {
      "epoch": 17.059344552701507,
      "grad_norm": 0.25863656401634216,
      "learning_rate": 1e-05,
      "loss": 1.0196,
      "step": 19260
    },
    {
      "epoch": 17.063773250664305,
      "grad_norm": 0.2568065822124481,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 19265
    },
    {
      "epoch": 17.068201948627102,
      "grad_norm": 0.23939284682273865,
      "learning_rate": 1e-05,
      "loss": 0.9126,
      "step": 19270
    },
    {
      "epoch": 17.072630646589904,
      "grad_norm": 0.2636503577232361,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 19275
    },
    {
      "epoch": 17.0770593445527,
      "grad_norm": 0.2507743239402771,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 19280
    },
    {
      "epoch": 17.0814880425155,
      "grad_norm": 0.2822876572608948,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 19285
    },
    {
      "epoch": 17.0859167404783,
      "grad_norm": 0.22202524542808533,
      "learning_rate": 1e-05,
      "loss": 1.0189,
      "step": 19290
    },
    {
      "epoch": 17.090345438441098,
      "grad_norm": 0.29415300488471985,
      "learning_rate": 1e-05,
      "loss": 1.0393,
      "step": 19295
    },
    {
      "epoch": 17.0947741364039,
      "grad_norm": 0.2532123923301697,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 19300
    },
    {
      "epoch": 17.099202834366697,
      "grad_norm": 0.26051363348960876,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 19305
    },
    {
      "epoch": 17.103631532329494,
      "grad_norm": 0.22066263854503632,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 19310
    },
    {
      "epoch": 17.108060230292296,
      "grad_norm": 0.2342502772808075,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 19315
    },
    {
      "epoch": 17.112488928255093,
      "grad_norm": 0.24816593527793884,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 19320
    },
    {
      "epoch": 17.11691762621789,
      "grad_norm": 0.24889181554317474,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 19325
    },
    {
      "epoch": 17.121346324180692,
      "grad_norm": 0.2589760720729828,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 19330
    },
    {
      "epoch": 17.12577502214349,
      "grad_norm": 0.27571389079093933,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 19335
    },
    {
      "epoch": 17.130203720106287,
      "grad_norm": 0.28763172030448914,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 19340
    },
    {
      "epoch": 17.13463241806909,
      "grad_norm": 0.2309671938419342,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 19345
    },
    {
      "epoch": 17.139061116031886,
      "grad_norm": 0.27404001355171204,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 19350
    },
    {
      "epoch": 17.143489813994684,
      "grad_norm": 0.2963801324367523,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 19355
    },
    {
      "epoch": 17.147918511957485,
      "grad_norm": 0.264787882566452,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 19360
    },
    {
      "epoch": 17.152347209920283,
      "grad_norm": 0.28650131821632385,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 19365
    },
    {
      "epoch": 17.156775907883084,
      "grad_norm": 0.23257899284362793,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 19370
    },
    {
      "epoch": 17.16120460584588,
      "grad_norm": 0.26733118295669556,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 19375
    },
    {
      "epoch": 17.16563330380868,
      "grad_norm": 0.22564467787742615,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 19380
    },
    {
      "epoch": 17.17006200177148,
      "grad_norm": 0.22003979980945587,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 19385
    },
    {
      "epoch": 17.174490699734278,
      "grad_norm": 0.26003363728523254,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 19390
    },
    {
      "epoch": 17.178919397697076,
      "grad_norm": 0.31560972332954407,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 19395
    },
    {
      "epoch": 17.183348095659877,
      "grad_norm": 0.24204380810260773,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 19400
    },
    {
      "epoch": 17.187776793622675,
      "grad_norm": 0.282644659280777,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 19405
    },
    {
      "epoch": 17.192205491585472,
      "grad_norm": 0.22840814292430878,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 19410
    },
    {
      "epoch": 17.196634189548273,
      "grad_norm": 0.30209341645240784,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 19415
    },
    {
      "epoch": 17.20106288751107,
      "grad_norm": 0.27143368124961853,
      "learning_rate": 1e-05,
      "loss": 1.045,
      "step": 19420
    },
    {
      "epoch": 17.205491585473872,
      "grad_norm": 0.2402728646993637,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 19425
    },
    {
      "epoch": 17.20992028343667,
      "grad_norm": 0.27113693952560425,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 19430
    },
    {
      "epoch": 17.214348981399468,
      "grad_norm": 0.23963776230812073,
      "learning_rate": 1e-05,
      "loss": 1.0443,
      "step": 19435
    },
    {
      "epoch": 17.21877767936227,
      "grad_norm": 0.2239329218864441,
      "learning_rate": 1e-05,
      "loss": 1.0159,
      "step": 19440
    },
    {
      "epoch": 17.223206377325067,
      "grad_norm": 0.2643023133277893,
      "learning_rate": 1e-05,
      "loss": 1.0324,
      "step": 19445
    },
    {
      "epoch": 17.227635075287864,
      "grad_norm": 0.2785678505897522,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 19450
    },
    {
      "epoch": 17.232063773250665,
      "grad_norm": 0.3162379860877991,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 19455
    },
    {
      "epoch": 17.236492471213463,
      "grad_norm": 0.29354482889175415,
      "learning_rate": 1e-05,
      "loss": 1.0262,
      "step": 19460
    },
    {
      "epoch": 17.24092116917626,
      "grad_norm": 0.27754074335098267,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 19465
    },
    {
      "epoch": 17.245349867139062,
      "grad_norm": 0.238135427236557,
      "learning_rate": 1e-05,
      "loss": 1.0372,
      "step": 19470
    },
    {
      "epoch": 17.24977856510186,
      "grad_norm": 0.23985199630260468,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 19475
    },
    {
      "epoch": 17.254207263064657,
      "grad_norm": 0.30725282430648804,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 19480
    },
    {
      "epoch": 17.25863596102746,
      "grad_norm": 0.2880932092666626,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 19485
    },
    {
      "epoch": 17.263064658990256,
      "grad_norm": 0.27494096755981445,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 19490
    },
    {
      "epoch": 17.267493356953057,
      "grad_norm": 0.26676058769226074,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 19495
    },
    {
      "epoch": 17.271922054915855,
      "grad_norm": 0.2333136647939682,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 19500
    },
    {
      "epoch": 17.276350752878653,
      "grad_norm": 0.23762977123260498,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 19505
    },
    {
      "epoch": 17.280779450841454,
      "grad_norm": 0.2730015218257904,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 19510
    },
    {
      "epoch": 17.28520814880425,
      "grad_norm": 0.26295778155326843,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 19515
    },
    {
      "epoch": 17.28963684676705,
      "grad_norm": 0.29106032848358154,
      "learning_rate": 1e-05,
      "loss": 1.0345,
      "step": 19520
    },
    {
      "epoch": 17.29406554472985,
      "grad_norm": 0.288153737783432,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 19525
    },
    {
      "epoch": 17.298494242692648,
      "grad_norm": 0.250797837972641,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 19530
    },
    {
      "epoch": 17.302922940655446,
      "grad_norm": 0.22978240251541138,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 19535
    },
    {
      "epoch": 17.307351638618247,
      "grad_norm": 0.26314058899879456,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 19540
    },
    {
      "epoch": 17.311780336581045,
      "grad_norm": 0.24025462567806244,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 19545
    },
    {
      "epoch": 17.316209034543846,
      "grad_norm": 0.25958991050720215,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 19550
    },
    {
      "epoch": 17.320637732506643,
      "grad_norm": 0.25435057282447815,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 19555
    },
    {
      "epoch": 17.32506643046944,
      "grad_norm": 0.22832071781158447,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 19560
    },
    {
      "epoch": 17.329495128432242,
      "grad_norm": 0.285186767578125,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 19565
    },
    {
      "epoch": 17.33392382639504,
      "grad_norm": 0.27580833435058594,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 19570
    },
    {
      "epoch": 17.338352524357838,
      "grad_norm": 0.22905929386615753,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 19575
    },
    {
      "epoch": 17.34278122232064,
      "grad_norm": 0.28417590260505676,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 19580
    },
    {
      "epoch": 17.347209920283436,
      "grad_norm": 0.21974067389965057,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 19585
    },
    {
      "epoch": 17.351638618246234,
      "grad_norm": 0.2521921396255493,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 19590
    },
    {
      "epoch": 17.356067316209035,
      "grad_norm": 0.2822516858577728,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 19595
    },
    {
      "epoch": 17.360496014171833,
      "grad_norm": 0.2621997892856598,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 19600
    },
    {
      "epoch": 17.36492471213463,
      "grad_norm": 0.2842501401901245,
      "learning_rate": 1e-05,
      "loss": 1.0336,
      "step": 19605
    },
    {
      "epoch": 17.369353410097432,
      "grad_norm": 0.2468596249818802,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 19610
    },
    {
      "epoch": 17.37378210806023,
      "grad_norm": 0.24540621042251587,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 19615
    },
    {
      "epoch": 17.37821080602303,
      "grad_norm": 0.2523883581161499,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 19620
    },
    {
      "epoch": 17.38263950398583,
      "grad_norm": 0.23667879402637482,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 19625
    },
    {
      "epoch": 17.387068201948626,
      "grad_norm": 0.3174433410167694,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 19630
    },
    {
      "epoch": 17.391496899911427,
      "grad_norm": 0.292094886302948,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 19635
    },
    {
      "epoch": 17.395925597874225,
      "grad_norm": 0.3052208721637726,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 19640
    },
    {
      "epoch": 17.400354295837023,
      "grad_norm": 0.3293698728084564,
      "learning_rate": 1e-05,
      "loss": 0.9289,
      "step": 19645
    },
    {
      "epoch": 17.404782993799824,
      "grad_norm": 0.28416889905929565,
      "learning_rate": 1e-05,
      "loss": 1.0463,
      "step": 19650
    },
    {
      "epoch": 17.40921169176262,
      "grad_norm": 0.264961302280426,
      "learning_rate": 1e-05,
      "loss": 1.0473,
      "step": 19655
    },
    {
      "epoch": 17.41364038972542,
      "grad_norm": 0.23802722990512848,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 19660
    },
    {
      "epoch": 17.41806908768822,
      "grad_norm": 0.3156811594963074,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 19665
    },
    {
      "epoch": 17.422497785651018,
      "grad_norm": 0.3185417056083679,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 19670
    },
    {
      "epoch": 17.42692648361382,
      "grad_norm": 0.25376373529434204,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 19675
    },
    {
      "epoch": 17.431355181576617,
      "grad_norm": 0.28718098998069763,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 19680
    },
    {
      "epoch": 17.435783879539414,
      "grad_norm": 0.23882503807544708,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 19685
    },
    {
      "epoch": 17.440212577502216,
      "grad_norm": 0.2596502900123596,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 19690
    },
    {
      "epoch": 17.444641275465013,
      "grad_norm": 0.25458356738090515,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 19695
    },
    {
      "epoch": 17.44906997342781,
      "grad_norm": 0.24817614257335663,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 19700
    },
    {
      "epoch": 17.453498671390612,
      "grad_norm": 0.23650948703289032,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 19705
    },
    {
      "epoch": 17.45792736935341,
      "grad_norm": 0.24727743864059448,
      "learning_rate": 1e-05,
      "loss": 1.03,
      "step": 19710
    },
    {
      "epoch": 17.462356067316207,
      "grad_norm": 0.26096096634864807,
      "learning_rate": 1e-05,
      "loss": 1.0307,
      "step": 19715
    },
    {
      "epoch": 17.46678476527901,
      "grad_norm": 0.23782479763031006,
      "learning_rate": 1e-05,
      "loss": 1.0255,
      "step": 19720
    },
    {
      "epoch": 17.471213463241806,
      "grad_norm": 0.26561567187309265,
      "learning_rate": 1e-05,
      "loss": 1.0538,
      "step": 19725
    },
    {
      "epoch": 17.475642161204608,
      "grad_norm": 0.2778032720088959,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 19730
    },
    {
      "epoch": 17.480070859167405,
      "grad_norm": 0.262604683637619,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 19735
    },
    {
      "epoch": 17.484499557130203,
      "grad_norm": 0.2597878575325012,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 19740
    },
    {
      "epoch": 17.488928255093004,
      "grad_norm": 0.35033324360847473,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 19745
    },
    {
      "epoch": 17.4933569530558,
      "grad_norm": 0.2768417000770569,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 19750
    },
    {
      "epoch": 17.4977856510186,
      "grad_norm": 0.25707241892814636,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 19755
    },
    {
      "epoch": 17.5022143489814,
      "grad_norm": 0.2521880269050598,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 19760
    },
    {
      "epoch": 17.5066430469442,
      "grad_norm": 0.3279540240764618,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 19765
    },
    {
      "epoch": 17.511071744906996,
      "grad_norm": 0.25486913323402405,
      "learning_rate": 1e-05,
      "loss": 1.0453,
      "step": 19770
    },
    {
      "epoch": 17.515500442869797,
      "grad_norm": 0.26148685812950134,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 19775
    },
    {
      "epoch": 17.519929140832595,
      "grad_norm": 0.3092002868652344,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 19780
    },
    {
      "epoch": 17.524357838795392,
      "grad_norm": 0.21546992659568787,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 19785
    },
    {
      "epoch": 17.528786536758194,
      "grad_norm": 0.258698970079422,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 19790
    },
    {
      "epoch": 17.53321523472099,
      "grad_norm": 0.26470106840133667,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 19795
    },
    {
      "epoch": 17.537643932683793,
      "grad_norm": 0.29198598861694336,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 19800
    },
    {
      "epoch": 17.54207263064659,
      "grad_norm": 0.3538688123226166,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 19805
    },
    {
      "epoch": 17.546501328609388,
      "grad_norm": 0.3227647542953491,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 19810
    },
    {
      "epoch": 17.55093002657219,
      "grad_norm": 0.2500825822353363,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 19815
    },
    {
      "epoch": 17.555358724534987,
      "grad_norm": 0.25851163268089294,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 19820
    },
    {
      "epoch": 17.559787422497784,
      "grad_norm": 0.3146398663520813,
      "learning_rate": 1e-05,
      "loss": 1.0418,
      "step": 19825
    },
    {
      "epoch": 17.564216120460586,
      "grad_norm": 0.3041551113128662,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 19830
    },
    {
      "epoch": 17.568644818423383,
      "grad_norm": 0.2577322721481323,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 19835
    },
    {
      "epoch": 17.57307351638618,
      "grad_norm": 0.2891632616519928,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 19840
    },
    {
      "epoch": 17.577502214348982,
      "grad_norm": 0.3622133433818817,
      "learning_rate": 1e-05,
      "loss": 1.0599,
      "step": 19845
    },
    {
      "epoch": 17.58193091231178,
      "grad_norm": 0.3105239272117615,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 19850
    },
    {
      "epoch": 17.58635961027458,
      "grad_norm": 0.31793999671936035,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 19855
    },
    {
      "epoch": 17.59078830823738,
      "grad_norm": 0.2550145983695984,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 19860
    },
    {
      "epoch": 17.595217006200176,
      "grad_norm": 0.2385762482881546,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 19865
    },
    {
      "epoch": 17.599645704162977,
      "grad_norm": 0.25345996022224426,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 19870
    },
    {
      "epoch": 17.604074402125775,
      "grad_norm": 0.24802087247371674,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 19875
    },
    {
      "epoch": 17.608503100088573,
      "grad_norm": 0.22456267476081848,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 19880
    },
    {
      "epoch": 17.612931798051374,
      "grad_norm": 0.250970721244812,
      "learning_rate": 1e-05,
      "loss": 1.0148,
      "step": 19885
    },
    {
      "epoch": 17.61736049601417,
      "grad_norm": 0.24302758276462555,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 19890
    },
    {
      "epoch": 17.62178919397697,
      "grad_norm": 0.22629830241203308,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 19895
    },
    {
      "epoch": 17.62621789193977,
      "grad_norm": 0.27670133113861084,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 19900
    },
    {
      "epoch": 17.630646589902568,
      "grad_norm": 0.22957493364810944,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 19905
    },
    {
      "epoch": 17.63507528786537,
      "grad_norm": 0.25864338874816895,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 19910
    },
    {
      "epoch": 17.639503985828167,
      "grad_norm": 0.2682626247406006,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 19915
    },
    {
      "epoch": 17.643932683790965,
      "grad_norm": 0.30448803305625916,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 19920
    },
    {
      "epoch": 17.648361381753766,
      "grad_norm": 0.2132716029882431,
      "learning_rate": 1e-05,
      "loss": 1.0369,
      "step": 19925
    },
    {
      "epoch": 17.652790079716564,
      "grad_norm": 0.27838215231895447,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 19930
    },
    {
      "epoch": 17.65721877767936,
      "grad_norm": 0.2461836338043213,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 19935
    },
    {
      "epoch": 17.661647475642162,
      "grad_norm": 0.24796424806118011,
      "learning_rate": 1e-05,
      "loss": 0.9079,
      "step": 19940
    },
    {
      "epoch": 17.66607617360496,
      "grad_norm": 0.2612871825695038,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 19945
    },
    {
      "epoch": 17.670504871567758,
      "grad_norm": 0.22839590907096863,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 19950
    },
    {
      "epoch": 17.67493356953056,
      "grad_norm": 0.28730660676956177,
      "learning_rate": 1e-05,
      "loss": 1.051,
      "step": 19955
    },
    {
      "epoch": 17.679362267493357,
      "grad_norm": 0.2569023370742798,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 19960
    },
    {
      "epoch": 17.683790965456154,
      "grad_norm": 0.2763356566429138,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 19965
    },
    {
      "epoch": 17.688219663418955,
      "grad_norm": 0.29813385009765625,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 19970
    },
    {
      "epoch": 17.692648361381753,
      "grad_norm": 0.2158433049917221,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 19975
    },
    {
      "epoch": 17.697077059344554,
      "grad_norm": 0.25374796986579895,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 19980
    },
    {
      "epoch": 17.701505757307352,
      "grad_norm": 0.2582029104232788,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 19985
    },
    {
      "epoch": 17.70593445527015,
      "grad_norm": 0.27137789130210876,
      "learning_rate": 1e-05,
      "loss": 1.0536,
      "step": 19990
    },
    {
      "epoch": 17.71036315323295,
      "grad_norm": 0.2295052856206894,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 19995
    },
    {
      "epoch": 17.71479185119575,
      "grad_norm": 0.24351628124713898,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 20000
    },
    {
      "epoch": 17.719220549158546,
      "grad_norm": 0.23730763792991638,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 20005
    },
    {
      "epoch": 17.723649247121347,
      "grad_norm": 0.29798734188079834,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 20010
    },
    {
      "epoch": 17.728077945084145,
      "grad_norm": 0.2506239712238312,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 20015
    },
    {
      "epoch": 17.732506643046943,
      "grad_norm": 0.2482958883047104,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 20020
    },
    {
      "epoch": 17.736935341009744,
      "grad_norm": 0.2259867936372757,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 20025
    },
    {
      "epoch": 17.74136403897254,
      "grad_norm": 0.256012886762619,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 20030
    },
    {
      "epoch": 17.745792736935343,
      "grad_norm": 0.2893749177455902,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 20035
    },
    {
      "epoch": 17.75022143489814,
      "grad_norm": 0.27459439635276794,
      "learning_rate": 1e-05,
      "loss": 1.0438,
      "step": 20040
    },
    {
      "epoch": 17.754650132860938,
      "grad_norm": 0.26578623056411743,
      "learning_rate": 1e-05,
      "loss": 1.0214,
      "step": 20045
    },
    {
      "epoch": 17.75907883082374,
      "grad_norm": 0.2591007947921753,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 20050
    },
    {
      "epoch": 17.763507528786537,
      "grad_norm": 0.3230084478855133,
      "learning_rate": 1e-05,
      "loss": 0.9188,
      "step": 20055
    },
    {
      "epoch": 17.767936226749335,
      "grad_norm": 0.22423215210437775,
      "learning_rate": 1e-05,
      "loss": 1.0346,
      "step": 20060
    },
    {
      "epoch": 17.772364924712136,
      "grad_norm": 0.2446160465478897,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 20065
    },
    {
      "epoch": 17.776793622674933,
      "grad_norm": 0.245269313454628,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 20070
    },
    {
      "epoch": 17.78122232063773,
      "grad_norm": 0.2621501684188843,
      "learning_rate": 1e-05,
      "loss": 1.014,
      "step": 20075
    },
    {
      "epoch": 17.785651018600532,
      "grad_norm": 0.2698981463909149,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 20080
    },
    {
      "epoch": 17.79007971656333,
      "grad_norm": 0.25634124875068665,
      "learning_rate": 1e-05,
      "loss": 1.0256,
      "step": 20085
    },
    {
      "epoch": 17.794508414526128,
      "grad_norm": 0.23192934691905975,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 20090
    },
    {
      "epoch": 17.79893711248893,
      "grad_norm": 0.27244633436203003,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 20095
    },
    {
      "epoch": 17.803365810451727,
      "grad_norm": 0.29565033316612244,
      "learning_rate": 1e-05,
      "loss": 0.8974,
      "step": 20100
    },
    {
      "epoch": 17.807794508414528,
      "grad_norm": 0.209321528673172,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 20105
    },
    {
      "epoch": 17.812223206377325,
      "grad_norm": 0.2633292078971863,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 20110
    },
    {
      "epoch": 17.816651904340123,
      "grad_norm": 0.22794212400913239,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 20115
    },
    {
      "epoch": 17.821080602302924,
      "grad_norm": 0.29816770553588867,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 20120
    },
    {
      "epoch": 17.825509300265722,
      "grad_norm": 0.21468093991279602,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 20125
    },
    {
      "epoch": 17.82993799822852,
      "grad_norm": 0.24957433342933655,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 20130
    },
    {
      "epoch": 17.83436669619132,
      "grad_norm": 0.2944706380367279,
      "learning_rate": 1e-05,
      "loss": 1.0226,
      "step": 20135
    },
    {
      "epoch": 17.83879539415412,
      "grad_norm": 0.3037720024585724,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 20140
    },
    {
      "epoch": 17.843224092116916,
      "grad_norm": 0.2549017071723938,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 20145
    },
    {
      "epoch": 17.847652790079717,
      "grad_norm": 0.24500052630901337,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 20150
    },
    {
      "epoch": 17.852081488042515,
      "grad_norm": 0.2671549320220947,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 20155
    },
    {
      "epoch": 17.856510186005316,
      "grad_norm": 0.2728942930698395,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 20160
    },
    {
      "epoch": 17.860938883968114,
      "grad_norm": 0.24169975519180298,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 20165
    },
    {
      "epoch": 17.86536758193091,
      "grad_norm": 0.24461784958839417,
      "learning_rate": 1e-05,
      "loss": 1.0303,
      "step": 20170
    },
    {
      "epoch": 17.869796279893713,
      "grad_norm": 0.32894495129585266,
      "learning_rate": 1e-05,
      "loss": 1.0428,
      "step": 20175
    },
    {
      "epoch": 17.87422497785651,
      "grad_norm": 0.2735280692577362,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 20180
    },
    {
      "epoch": 17.878653675819308,
      "grad_norm": 0.2700766921043396,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 20185
    },
    {
      "epoch": 17.88308237378211,
      "grad_norm": 0.31885093450546265,
      "learning_rate": 1e-05,
      "loss": 1.0873,
      "step": 20190
    },
    {
      "epoch": 17.887511071744907,
      "grad_norm": 0.2657311260700226,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 20195
    },
    {
      "epoch": 17.891939769707704,
      "grad_norm": 0.25965866446495056,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 20200
    },
    {
      "epoch": 17.896368467670506,
      "grad_norm": 0.28984150290489197,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 20205
    },
    {
      "epoch": 17.900797165633303,
      "grad_norm": 0.2735573351383209,
      "learning_rate": 1e-05,
      "loss": 1.0354,
      "step": 20210
    },
    {
      "epoch": 17.9052258635961,
      "grad_norm": 0.31076914072036743,
      "learning_rate": 1e-05,
      "loss": 1.0263,
      "step": 20215
    },
    {
      "epoch": 17.909654561558902,
      "grad_norm": 0.26216214895248413,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 20220
    },
    {
      "epoch": 17.9140832595217,
      "grad_norm": 0.25458693504333496,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 20225
    },
    {
      "epoch": 17.9185119574845,
      "grad_norm": 0.26108530163764954,
      "learning_rate": 1e-05,
      "loss": 1.0766,
      "step": 20230
    },
    {
      "epoch": 17.9229406554473,
      "grad_norm": 0.2990536689758301,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 20235
    },
    {
      "epoch": 17.927369353410096,
      "grad_norm": 0.28272974491119385,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 20240
    },
    {
      "epoch": 17.931798051372898,
      "grad_norm": 0.22939245402812958,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 20245
    },
    {
      "epoch": 17.936226749335695,
      "grad_norm": 0.2612392008304596,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 20250
    },
    {
      "epoch": 17.940655447298493,
      "grad_norm": 0.2584625482559204,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 20255
    },
    {
      "epoch": 17.945084145261294,
      "grad_norm": 0.2809886634349823,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 20260
    },
    {
      "epoch": 17.949512843224092,
      "grad_norm": 0.31402015686035156,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 20265
    },
    {
      "epoch": 17.95394154118689,
      "grad_norm": 0.3004207909107208,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 20270
    },
    {
      "epoch": 17.95837023914969,
      "grad_norm": 0.32444968819618225,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 20275
    },
    {
      "epoch": 17.96279893711249,
      "grad_norm": 0.26725834608078003,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 20280
    },
    {
      "epoch": 17.96722763507529,
      "grad_norm": 0.234386146068573,
      "learning_rate": 1e-05,
      "loss": 1.0239,
      "step": 20285
    },
    {
      "epoch": 17.971656333038087,
      "grad_norm": 0.2573343813419342,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 20290
    },
    {
      "epoch": 17.976085031000885,
      "grad_norm": 0.2270103543996811,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 20295
    },
    {
      "epoch": 17.980513728963686,
      "grad_norm": 0.23110328614711761,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 20300
    },
    {
      "epoch": 17.984942426926484,
      "grad_norm": 0.22924678027629852,
      "learning_rate": 1e-05,
      "loss": 1.067,
      "step": 20305
    },
    {
      "epoch": 17.98937112488928,
      "grad_norm": 0.29164668917655945,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 20310
    },
    {
      "epoch": 17.993799822852083,
      "grad_norm": 0.2458817958831787,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 20315
    },
    {
      "epoch": 17.99822852081488,
      "grad_norm": 0.23047512769699097,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 20320
    },
    {
      "epoch": 18.002657218777678,
      "grad_norm": 0.24460440874099731,
      "learning_rate": 1e-05,
      "loss": 1.044,
      "step": 20325
    },
    {
      "epoch": 18.00708591674048,
      "grad_norm": 0.22777819633483887,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 20330
    },
    {
      "epoch": 18.011514614703277,
      "grad_norm": 0.23051854968070984,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 20335
    },
    {
      "epoch": 18.015943312666074,
      "grad_norm": 0.2609602212905884,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 20340
    },
    {
      "epoch": 18.020372010628876,
      "grad_norm": 0.24906164407730103,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 20345
    },
    {
      "epoch": 18.024800708591673,
      "grad_norm": 0.25315409898757935,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 20350
    },
    {
      "epoch": 18.029229406554474,
      "grad_norm": 0.2424316555261612,
      "learning_rate": 1e-05,
      "loss": 1.0213,
      "step": 20355
    },
    {
      "epoch": 18.033658104517272,
      "grad_norm": 0.2625334858894348,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 20360
    },
    {
      "epoch": 18.03808680248007,
      "grad_norm": 0.26623883843421936,
      "learning_rate": 1e-05,
      "loss": 1.0951,
      "step": 20365
    },
    {
      "epoch": 18.04251550044287,
      "grad_norm": 0.25353771448135376,
      "learning_rate": 1e-05,
      "loss": 1.0217,
      "step": 20370
    },
    {
      "epoch": 18.04694419840567,
      "grad_norm": 0.23834754526615143,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 20375
    },
    {
      "epoch": 18.051372896368466,
      "grad_norm": 0.2945196330547333,
      "learning_rate": 1e-05,
      "loss": 1.0186,
      "step": 20380
    },
    {
      "epoch": 18.055801594331268,
      "grad_norm": 0.2569020092487335,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 20385
    },
    {
      "epoch": 18.060230292294065,
      "grad_norm": 0.26425111293792725,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 20390
    },
    {
      "epoch": 18.064658990256863,
      "grad_norm": 0.22954672574996948,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 20395
    },
    {
      "epoch": 18.069087688219664,
      "grad_norm": 0.22191312909126282,
      "learning_rate": 1e-05,
      "loss": 1.0175,
      "step": 20400
    },
    {
      "epoch": 18.07351638618246,
      "grad_norm": 0.27791184186935425,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 20405
    },
    {
      "epoch": 18.077945084145263,
      "grad_norm": 0.264342337846756,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 20410
    },
    {
      "epoch": 18.08237378210806,
      "grad_norm": 0.23839698731899261,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 20415
    },
    {
      "epoch": 18.086802480070858,
      "grad_norm": 0.2891393005847931,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 20420
    },
    {
      "epoch": 18.09123117803366,
      "grad_norm": 0.27449479699134827,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 20425
    },
    {
      "epoch": 18.095659875996457,
      "grad_norm": 0.28650403022766113,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 20430
    },
    {
      "epoch": 18.100088573959255,
      "grad_norm": 0.2671445608139038,
      "learning_rate": 1e-05,
      "loss": 1.0286,
      "step": 20435
    },
    {
      "epoch": 18.104517271922056,
      "grad_norm": 0.2786725163459778,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 20440
    },
    {
      "epoch": 18.108945969884854,
      "grad_norm": 0.23936715722084045,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 20445
    },
    {
      "epoch": 18.11337466784765,
      "grad_norm": 0.24532751739025116,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 20450
    },
    {
      "epoch": 18.117803365810452,
      "grad_norm": 0.26958322525024414,
      "learning_rate": 1e-05,
      "loss": 1.0512,
      "step": 20455
    },
    {
      "epoch": 18.12223206377325,
      "grad_norm": 0.2908604145050049,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 20460
    },
    {
      "epoch": 18.12666076173605,
      "grad_norm": 0.2929830849170685,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 20465
    },
    {
      "epoch": 18.13108945969885,
      "grad_norm": 0.23224352300167084,
      "learning_rate": 1e-05,
      "loss": 1.0149,
      "step": 20470
    },
    {
      "epoch": 18.135518157661647,
      "grad_norm": 0.2708369791507721,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 20475
    },
    {
      "epoch": 18.139946855624448,
      "grad_norm": 0.2332603931427002,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 20480
    },
    {
      "epoch": 18.144375553587246,
      "grad_norm": 0.2812580168247223,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 20485
    },
    {
      "epoch": 18.148804251550043,
      "grad_norm": 0.2496076226234436,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 20490
    },
    {
      "epoch": 18.153232949512844,
      "grad_norm": 0.2712249755859375,
      "learning_rate": 1e-05,
      "loss": 1.0427,
      "step": 20495
    },
    {
      "epoch": 18.157661647475642,
      "grad_norm": 0.24284881353378296,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 20500
    },
    {
      "epoch": 18.16209034543844,
      "grad_norm": 0.2146773487329483,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 20505
    },
    {
      "epoch": 18.16651904340124,
      "grad_norm": 0.23569054901599884,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 20510
    },
    {
      "epoch": 18.17094774136404,
      "grad_norm": 0.24455448985099792,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 20515
    },
    {
      "epoch": 18.175376439326836,
      "grad_norm": 0.2086784839630127,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 20520
    },
    {
      "epoch": 18.179805137289637,
      "grad_norm": 0.23628121614456177,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 20525
    },
    {
      "epoch": 18.184233835252435,
      "grad_norm": 0.24341154098510742,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 20530
    },
    {
      "epoch": 18.188662533215236,
      "grad_norm": 0.2880188822746277,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 20535
    },
    {
      "epoch": 18.193091231178034,
      "grad_norm": 0.28739896416664124,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 20540
    },
    {
      "epoch": 18.19751992914083,
      "grad_norm": 0.22721165418624878,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 20545
    },
    {
      "epoch": 18.201948627103633,
      "grad_norm": 0.240403413772583,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 20550
    },
    {
      "epoch": 18.20637732506643,
      "grad_norm": 0.27363017201423645,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 20555
    },
    {
      "epoch": 18.210806023029228,
      "grad_norm": 0.23993827402591705,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 20560
    },
    {
      "epoch": 18.21523472099203,
      "grad_norm": 0.288872629404068,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 20565
    },
    {
      "epoch": 18.219663418954827,
      "grad_norm": 0.24095289409160614,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 20570
    },
    {
      "epoch": 18.224092116917625,
      "grad_norm": 0.2589903473854065,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 20575
    },
    {
      "epoch": 18.228520814880426,
      "grad_norm": 0.24059975147247314,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 20580
    },
    {
      "epoch": 18.232949512843224,
      "grad_norm": 0.21484382450580597,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 20585
    },
    {
      "epoch": 18.237378210806025,
      "grad_norm": 0.22517722845077515,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 20590
    },
    {
      "epoch": 18.241806908768822,
      "grad_norm": 0.22644075751304626,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 20595
    },
    {
      "epoch": 18.24623560673162,
      "grad_norm": 0.25431379675865173,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 20600
    },
    {
      "epoch": 18.25066430469442,
      "grad_norm": 0.24803446233272552,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 20605
    },
    {
      "epoch": 18.25509300265722,
      "grad_norm": 0.23967397212982178,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 20610
    },
    {
      "epoch": 18.259521700620017,
      "grad_norm": 0.23101189732551575,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 20615
    },
    {
      "epoch": 18.263950398582818,
      "grad_norm": 0.23987622559070587,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 20620
    },
    {
      "epoch": 18.268379096545615,
      "grad_norm": 0.2931157648563385,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 20625
    },
    {
      "epoch": 18.272807794508413,
      "grad_norm": 0.21868695318698883,
      "learning_rate": 1e-05,
      "loss": 1.0238,
      "step": 20630
    },
    {
      "epoch": 18.277236492471214,
      "grad_norm": 0.2845988869667053,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 20635
    },
    {
      "epoch": 18.281665190434012,
      "grad_norm": 0.2677435278892517,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 20640
    },
    {
      "epoch": 18.28609388839681,
      "grad_norm": 0.2774239182472229,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 20645
    },
    {
      "epoch": 18.29052258635961,
      "grad_norm": 0.27630990743637085,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 20650
    },
    {
      "epoch": 18.29495128432241,
      "grad_norm": 0.22741125524044037,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 20655
    },
    {
      "epoch": 18.29937998228521,
      "grad_norm": 0.2839937210083008,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 20660
    },
    {
      "epoch": 18.303808680248007,
      "grad_norm": 0.27133598923683167,
      "learning_rate": 1e-05,
      "loss": 1.0355,
      "step": 20665
    },
    {
      "epoch": 18.308237378210805,
      "grad_norm": 0.2232791632413864,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 20670
    },
    {
      "epoch": 18.312666076173606,
      "grad_norm": 0.255623459815979,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 20675
    },
    {
      "epoch": 18.317094774136404,
      "grad_norm": 0.28772732615470886,
      "learning_rate": 1e-05,
      "loss": 1.0381,
      "step": 20680
    },
    {
      "epoch": 18.3215234720992,
      "grad_norm": 0.27251097559928894,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 20685
    },
    {
      "epoch": 18.325952170062003,
      "grad_norm": 0.22250190377235413,
      "learning_rate": 1e-05,
      "loss": 1.0544,
      "step": 20690
    },
    {
      "epoch": 18.3303808680248,
      "grad_norm": 0.28801122307777405,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 20695
    },
    {
      "epoch": 18.334809565987598,
      "grad_norm": 0.34806159138679504,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 20700
    },
    {
      "epoch": 18.3392382639504,
      "grad_norm": 0.30037423968315125,
      "learning_rate": 1e-05,
      "loss": 1.0235,
      "step": 20705
    },
    {
      "epoch": 18.343666961913197,
      "grad_norm": 0.2328796535730362,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 20710
    },
    {
      "epoch": 18.348095659875998,
      "grad_norm": 0.3051951825618744,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 20715
    },
    {
      "epoch": 18.352524357838796,
      "grad_norm": 0.2697443664073944,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 20720
    },
    {
      "epoch": 18.356953055801593,
      "grad_norm": 0.26864537596702576,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 20725
    },
    {
      "epoch": 18.361381753764395,
      "grad_norm": 0.2774394750595093,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 20730
    },
    {
      "epoch": 18.365810451727192,
      "grad_norm": 0.2447633594274521,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 20735
    },
    {
      "epoch": 18.37023914968999,
      "grad_norm": 0.24478772282600403,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 20740
    },
    {
      "epoch": 18.37466784765279,
      "grad_norm": 0.2232452780008316,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 20745
    },
    {
      "epoch": 18.37909654561559,
      "grad_norm": 0.26278409361839294,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 20750
    },
    {
      "epoch": 18.383525243578386,
      "grad_norm": 0.2082904428243637,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 20755
    },
    {
      "epoch": 18.387953941541188,
      "grad_norm": 0.23361815512180328,
      "learning_rate": 1e-05,
      "loss": 1.034,
      "step": 20760
    },
    {
      "epoch": 18.392382639503985,
      "grad_norm": 0.2113049328327179,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 20765
    },
    {
      "epoch": 18.396811337466787,
      "grad_norm": 0.2505301237106323,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 20770
    },
    {
      "epoch": 18.401240035429584,
      "grad_norm": 0.28443095088005066,
      "learning_rate": 1e-05,
      "loss": 1.0261,
      "step": 20775
    },
    {
      "epoch": 18.405668733392382,
      "grad_norm": 0.22276189923286438,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 20780
    },
    {
      "epoch": 18.410097431355183,
      "grad_norm": 0.21942079067230225,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 20785
    },
    {
      "epoch": 18.41452612931798,
      "grad_norm": 0.23163661360740662,
      "learning_rate": 1e-05,
      "loss": 1.0567,
      "step": 20790
    },
    {
      "epoch": 18.41895482728078,
      "grad_norm": 0.26401811838150024,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 20795
    },
    {
      "epoch": 18.42338352524358,
      "grad_norm": 0.23706109821796417,
      "learning_rate": 1e-05,
      "loss": 0.9147,
      "step": 20800
    },
    {
      "epoch": 18.427812223206377,
      "grad_norm": 0.24434785544872284,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 20805
    },
    {
      "epoch": 18.432240921169175,
      "grad_norm": 0.21827857196331024,
      "learning_rate": 1e-05,
      "loss": 1.0303,
      "step": 20810
    },
    {
      "epoch": 18.436669619131976,
      "grad_norm": 0.2557186186313629,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 20815
    },
    {
      "epoch": 18.441098317094774,
      "grad_norm": 0.23795369267463684,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 20820
    },
    {
      "epoch": 18.44552701505757,
      "grad_norm": 0.2577052414417267,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 20825
    },
    {
      "epoch": 18.449955713020373,
      "grad_norm": 0.19321399927139282,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 20830
    },
    {
      "epoch": 18.45438441098317,
      "grad_norm": 0.27815720438957214,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 20835
    },
    {
      "epoch": 18.45881310894597,
      "grad_norm": 0.24520039558410645,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 20840
    },
    {
      "epoch": 18.46324180690877,
      "grad_norm": 0.3463394343852997,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 20845
    },
    {
      "epoch": 18.467670504871567,
      "grad_norm": 0.2585110068321228,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 20850
    },
    {
      "epoch": 18.472099202834368,
      "grad_norm": 0.2485339641571045,
      "learning_rate": 1e-05,
      "loss": 1.0613,
      "step": 20855
    },
    {
      "epoch": 18.476527900797166,
      "grad_norm": 0.20855893194675446,
      "learning_rate": 1e-05,
      "loss": 1.0257,
      "step": 20860
    },
    {
      "epoch": 18.480956598759963,
      "grad_norm": 0.28965285420417786,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 20865
    },
    {
      "epoch": 18.485385296722765,
      "grad_norm": 0.2055974006652832,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 20870
    },
    {
      "epoch": 18.489813994685562,
      "grad_norm": 0.4266437888145447,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 20875
    },
    {
      "epoch": 18.49424269264836,
      "grad_norm": 0.30046847462654114,
      "learning_rate": 1e-05,
      "loss": 1.0794,
      "step": 20880
    },
    {
      "epoch": 18.49867139061116,
      "grad_norm": 0.25240740180015564,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 20885
    },
    {
      "epoch": 18.50310008857396,
      "grad_norm": 0.29113587737083435,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 20890
    },
    {
      "epoch": 18.50752878653676,
      "grad_norm": 0.2869458496570587,
      "learning_rate": 1e-05,
      "loss": 1.031,
      "step": 20895
    },
    {
      "epoch": 18.511957484499558,
      "grad_norm": 0.2689855992794037,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 20900
    },
    {
      "epoch": 18.516386182462355,
      "grad_norm": 0.2189704328775406,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 20905
    },
    {
      "epoch": 18.520814880425156,
      "grad_norm": 0.2508397698402405,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 20910
    },
    {
      "epoch": 18.525243578387954,
      "grad_norm": 0.4034369885921478,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 20915
    },
    {
      "epoch": 18.52967227635075,
      "grad_norm": 0.22806298732757568,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 20920
    },
    {
      "epoch": 18.534100974313553,
      "grad_norm": 0.2971166968345642,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 20925
    },
    {
      "epoch": 18.53852967227635,
      "grad_norm": 0.2362562119960785,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 20930
    },
    {
      "epoch": 18.54295837023915,
      "grad_norm": 0.2426292896270752,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 20935
    },
    {
      "epoch": 18.54738706820195,
      "grad_norm": 0.2474449723958969,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 20940
    },
    {
      "epoch": 18.551815766164747,
      "grad_norm": 0.2669737935066223,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 20945
    },
    {
      "epoch": 18.556244464127545,
      "grad_norm": 0.26444774866104126,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 20950
    },
    {
      "epoch": 18.560673162090346,
      "grad_norm": 0.3404650092124939,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 20955
    },
    {
      "epoch": 18.565101860053144,
      "grad_norm": 0.222050279378891,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 20960
    },
    {
      "epoch": 18.569530558015945,
      "grad_norm": 0.25178003311157227,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 20965
    },
    {
      "epoch": 18.573959255978743,
      "grad_norm": 0.3069479167461395,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 20970
    },
    {
      "epoch": 18.57838795394154,
      "grad_norm": 0.2649993598461151,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 20975
    },
    {
      "epoch": 18.58281665190434,
      "grad_norm": 0.25494804978370667,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 20980
    },
    {
      "epoch": 18.58724534986714,
      "grad_norm": 0.23358748853206635,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 20985
    },
    {
      "epoch": 18.591674047829937,
      "grad_norm": 0.258725106716156,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 20990
    },
    {
      "epoch": 18.596102745792738,
      "grad_norm": 0.2568466067314148,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 20995
    },
    {
      "epoch": 18.600531443755536,
      "grad_norm": 0.2945703864097595,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 21000
    },
    {
      "epoch": 18.604960141718333,
      "grad_norm": 0.3263799846172333,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 21005
    },
    {
      "epoch": 18.609388839681134,
      "grad_norm": 0.25300148129463196,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 21010
    },
    {
      "epoch": 18.613817537643932,
      "grad_norm": 0.23674839735031128,
      "learning_rate": 1e-05,
      "loss": 1.0204,
      "step": 21015
    },
    {
      "epoch": 18.618246235606733,
      "grad_norm": 0.2520981431007385,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 21020
    },
    {
      "epoch": 18.62267493356953,
      "grad_norm": 0.240373894572258,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 21025
    },
    {
      "epoch": 18.62710363153233,
      "grad_norm": 0.26391932368278503,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 21030
    },
    {
      "epoch": 18.63153232949513,
      "grad_norm": 0.2686443030834198,
      "learning_rate": 1e-05,
      "loss": 1.0182,
      "step": 21035
    },
    {
      "epoch": 18.635961027457927,
      "grad_norm": 0.2981419861316681,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 21040
    },
    {
      "epoch": 18.640389725420725,
      "grad_norm": 0.24534614384174347,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 21045
    },
    {
      "epoch": 18.644818423383526,
      "grad_norm": 0.29010310769081116,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 21050
    },
    {
      "epoch": 18.649247121346324,
      "grad_norm": 0.2278757393360138,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 21055
    },
    {
      "epoch": 18.65367581930912,
      "grad_norm": 0.2694181203842163,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 21060
    },
    {
      "epoch": 18.658104517271923,
      "grad_norm": 0.2664870023727417,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 21065
    },
    {
      "epoch": 18.66253321523472,
      "grad_norm": 0.23377925157546997,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 21070
    },
    {
      "epoch": 18.666961913197518,
      "grad_norm": 0.2550694942474365,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 21075
    },
    {
      "epoch": 18.67139061116032,
      "grad_norm": 0.2993178069591522,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 21080
    },
    {
      "epoch": 18.675819309123117,
      "grad_norm": 0.25599873065948486,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 21085
    },
    {
      "epoch": 18.68024800708592,
      "grad_norm": 0.2198406308889389,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 21090
    },
    {
      "epoch": 18.684676705048716,
      "grad_norm": 0.3192952275276184,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 21095
    },
    {
      "epoch": 18.689105403011514,
      "grad_norm": 0.2867943048477173,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 21100
    },
    {
      "epoch": 18.693534100974315,
      "grad_norm": 0.23722036182880402,
      "learning_rate": 1e-05,
      "loss": 1.0405,
      "step": 21105
    },
    {
      "epoch": 18.697962798937112,
      "grad_norm": 0.2737114429473877,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 21110
    },
    {
      "epoch": 18.70239149689991,
      "grad_norm": 0.28285425901412964,
      "learning_rate": 1e-05,
      "loss": 0.9279,
      "step": 21115
    },
    {
      "epoch": 18.70682019486271,
      "grad_norm": 0.2990366518497467,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 21120
    },
    {
      "epoch": 18.71124889282551,
      "grad_norm": 0.302739679813385,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 21125
    },
    {
      "epoch": 18.715677590788307,
      "grad_norm": 0.2583636939525604,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 21130
    },
    {
      "epoch": 18.720106288751108,
      "grad_norm": 0.20303533971309662,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 21135
    },
    {
      "epoch": 18.724534986713905,
      "grad_norm": 0.28236815333366394,
      "learning_rate": 1e-05,
      "loss": 1.0276,
      "step": 21140
    },
    {
      "epoch": 18.728963684676707,
      "grad_norm": 0.33288443088531494,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 21145
    },
    {
      "epoch": 18.733392382639504,
      "grad_norm": 0.24907001852989197,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 21150
    },
    {
      "epoch": 18.737821080602302,
      "grad_norm": 0.28821179270744324,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 21155
    },
    {
      "epoch": 18.742249778565103,
      "grad_norm": 0.3428443968296051,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 21160
    },
    {
      "epoch": 18.7466784765279,
      "grad_norm": 0.2658557891845703,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 21165
    },
    {
      "epoch": 18.7511071744907,
      "grad_norm": 0.30375638604164124,
      "learning_rate": 1e-05,
      "loss": 1.0382,
      "step": 21170
    },
    {
      "epoch": 18.7555358724535,
      "grad_norm": 0.2688475549221039,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 21175
    },
    {
      "epoch": 18.759964570416297,
      "grad_norm": 0.2684788107872009,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 21180
    },
    {
      "epoch": 18.764393268379095,
      "grad_norm": 0.2773300111293793,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 21185
    },
    {
      "epoch": 18.768821966341896,
      "grad_norm": 0.26814454793930054,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 21190
    },
    {
      "epoch": 18.773250664304694,
      "grad_norm": 0.263419508934021,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 21195
    },
    {
      "epoch": 18.77767936226749,
      "grad_norm": 0.22362379729747772,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 21200
    },
    {
      "epoch": 18.782108060230293,
      "grad_norm": 0.21374699473381042,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 21205
    },
    {
      "epoch": 18.78653675819309,
      "grad_norm": 0.3004043996334076,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 21210
    },
    {
      "epoch": 18.79096545615589,
      "grad_norm": 0.2347671240568161,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 21215
    },
    {
      "epoch": 18.79539415411869,
      "grad_norm": 0.22683128714561462,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 21220
    },
    {
      "epoch": 18.799822852081487,
      "grad_norm": 0.22115366160869598,
      "learning_rate": 1e-05,
      "loss": 0.9206,
      "step": 21225
    },
    {
      "epoch": 18.804251550044288,
      "grad_norm": 0.33044692873954773,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 21230
    },
    {
      "epoch": 18.808680248007086,
      "grad_norm": 0.2138751894235611,
      "learning_rate": 1e-05,
      "loss": 1.0315,
      "step": 21235
    },
    {
      "epoch": 18.813108945969883,
      "grad_norm": 0.21777422726154327,
      "learning_rate": 1e-05,
      "loss": 1.0757,
      "step": 21240
    },
    {
      "epoch": 18.817537643932685,
      "grad_norm": 0.24124111235141754,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 21245
    },
    {
      "epoch": 18.821966341895482,
      "grad_norm": 0.2945777475833893,
      "learning_rate": 1e-05,
      "loss": 1.0232,
      "step": 21250
    },
    {
      "epoch": 18.82639503985828,
      "grad_norm": 0.24781376123428345,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 21255
    },
    {
      "epoch": 18.83082373782108,
      "grad_norm": 0.26145169138908386,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 21260
    },
    {
      "epoch": 18.83525243578388,
      "grad_norm": 0.25660327076911926,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 21265
    },
    {
      "epoch": 18.83968113374668,
      "grad_norm": 0.23725764453411102,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 21270
    },
    {
      "epoch": 18.844109831709478,
      "grad_norm": 0.23738716542720795,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 21275
    },
    {
      "epoch": 18.848538529672275,
      "grad_norm": 0.30419716238975525,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 21280
    },
    {
      "epoch": 18.852967227635077,
      "grad_norm": 0.28879550099372864,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 21285
    },
    {
      "epoch": 18.857395925597874,
      "grad_norm": 0.25562360882759094,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 21290
    },
    {
      "epoch": 18.861824623560672,
      "grad_norm": 0.2460346519947052,
      "learning_rate": 1e-05,
      "loss": 0.9408,
      "step": 21295
    },
    {
      "epoch": 18.866253321523473,
      "grad_norm": 0.23227177560329437,
      "learning_rate": 1e-05,
      "loss": 1.0332,
      "step": 21300
    },
    {
      "epoch": 18.87068201948627,
      "grad_norm": 0.23724380135536194,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 21305
    },
    {
      "epoch": 18.87511071744907,
      "grad_norm": 0.2601776123046875,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 21310
    },
    {
      "epoch": 18.87953941541187,
      "grad_norm": 0.2933485209941864,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 21315
    },
    {
      "epoch": 18.883968113374667,
      "grad_norm": 0.28493037819862366,
      "learning_rate": 1e-05,
      "loss": 1.0288,
      "step": 21320
    },
    {
      "epoch": 18.888396811337465,
      "grad_norm": 0.2708951532840729,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 21325
    },
    {
      "epoch": 18.892825509300266,
      "grad_norm": 0.2811450958251953,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 21330
    },
    {
      "epoch": 18.897254207263064,
      "grad_norm": 0.2341175228357315,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 21335
    },
    {
      "epoch": 18.901682905225865,
      "grad_norm": 0.22528398036956787,
      "learning_rate": 1e-05,
      "loss": 1.0494,
      "step": 21340
    },
    {
      "epoch": 18.906111603188663,
      "grad_norm": 0.26082026958465576,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 21345
    },
    {
      "epoch": 18.91054030115146,
      "grad_norm": 0.27095597982406616,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 21350
    },
    {
      "epoch": 18.91496899911426,
      "grad_norm": 0.27901455760002136,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 21355
    },
    {
      "epoch": 18.91939769707706,
      "grad_norm": 0.2891344130039215,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 21360
    },
    {
      "epoch": 18.923826395039857,
      "grad_norm": 0.2786041796207428,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 21365
    },
    {
      "epoch": 18.928255093002658,
      "grad_norm": 0.22194671630859375,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 21370
    },
    {
      "epoch": 18.932683790965456,
      "grad_norm": 0.2665034830570221,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 21375
    },
    {
      "epoch": 18.937112488928253,
      "grad_norm": 0.23111851513385773,
      "learning_rate": 1e-05,
      "loss": 0.9297,
      "step": 21380
    },
    {
      "epoch": 18.941541186891055,
      "grad_norm": 0.22685515880584717,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 21385
    },
    {
      "epoch": 18.945969884853852,
      "grad_norm": 0.24442416429519653,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 21390
    },
    {
      "epoch": 18.950398582816653,
      "grad_norm": 0.2431734800338745,
      "learning_rate": 1e-05,
      "loss": 1.0489,
      "step": 21395
    },
    {
      "epoch": 18.95482728077945,
      "grad_norm": 0.20329251885414124,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 21400
    },
    {
      "epoch": 18.95925597874225,
      "grad_norm": 0.22948205471038818,
      "learning_rate": 1e-05,
      "loss": 0.9195,
      "step": 21405
    },
    {
      "epoch": 18.96368467670505,
      "grad_norm": 0.2835366427898407,
      "learning_rate": 1e-05,
      "loss": 1.0221,
      "step": 21410
    },
    {
      "epoch": 18.968113374667848,
      "grad_norm": 0.2749370038509369,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 21415
    },
    {
      "epoch": 18.972542072630645,
      "grad_norm": 0.2701703608036041,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 21420
    },
    {
      "epoch": 18.976970770593447,
      "grad_norm": 0.25853419303894043,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 21425
    },
    {
      "epoch": 18.981399468556244,
      "grad_norm": 0.304542601108551,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 21430
    },
    {
      "epoch": 18.985828166519042,
      "grad_norm": 0.2370123565196991,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 21435
    },
    {
      "epoch": 18.990256864481843,
      "grad_norm": 0.3070536255836487,
      "learning_rate": 1e-05,
      "loss": 0.9001,
      "step": 21440
    },
    {
      "epoch": 18.99468556244464,
      "grad_norm": 0.2961980998516083,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 21445
    },
    {
      "epoch": 18.999114260407442,
      "grad_norm": 0.27768170833587646,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 21450
    },
    {
      "epoch": 19.00354295837024,
      "grad_norm": 0.2616350054740906,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 21455
    },
    {
      "epoch": 19.007971656333037,
      "grad_norm": 0.2388584166765213,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 21460
    },
    {
      "epoch": 19.01240035429584,
      "grad_norm": 0.2257193773984909,
      "learning_rate": 1e-05,
      "loss": 1.0207,
      "step": 21465
    },
    {
      "epoch": 19.016829052258636,
      "grad_norm": 0.2784733176231384,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 21470
    },
    {
      "epoch": 19.021257750221434,
      "grad_norm": 0.27685242891311646,
      "learning_rate": 1e-05,
      "loss": 0.9253,
      "step": 21475
    },
    {
      "epoch": 19.025686448184235,
      "grad_norm": 0.23150534927845,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 21480
    },
    {
      "epoch": 19.030115146147033,
      "grad_norm": 0.34025201201438904,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 21485
    },
    {
      "epoch": 19.03454384410983,
      "grad_norm": 0.26765719056129456,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 21490
    },
    {
      "epoch": 19.03897254207263,
      "grad_norm": 0.23880422115325928,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 21495
    },
    {
      "epoch": 19.04340124003543,
      "grad_norm": 0.38661840558052063,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 21500
    },
    {
      "epoch": 19.047829937998227,
      "grad_norm": 0.3080753982067108,
      "learning_rate": 1e-05,
      "loss": 1.0451,
      "step": 21505
    },
    {
      "epoch": 19.052258635961028,
      "grad_norm": 0.30306461453437805,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 21510
    },
    {
      "epoch": 19.056687333923826,
      "grad_norm": 0.297817200422287,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 21515
    },
    {
      "epoch": 19.061116031886627,
      "grad_norm": 0.3119041621685028,
      "learning_rate": 1e-05,
      "loss": 1.0438,
      "step": 21520
    },
    {
      "epoch": 19.065544729849424,
      "grad_norm": 0.2956949770450592,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 21525
    },
    {
      "epoch": 19.069973427812222,
      "grad_norm": 0.27365535497665405,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 21530
    },
    {
      "epoch": 19.074402125775023,
      "grad_norm": 0.2655046582221985,
      "learning_rate": 1e-05,
      "loss": 1.0085,
      "step": 21535
    },
    {
      "epoch": 19.07883082373782,
      "grad_norm": 0.23782466351985931,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 21540
    },
    {
      "epoch": 19.08325952170062,
      "grad_norm": 0.2659986615180969,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 21545
    },
    {
      "epoch": 19.08768821966342,
      "grad_norm": 0.2969227731227875,
      "learning_rate": 1e-05,
      "loss": 1.0613,
      "step": 21550
    },
    {
      "epoch": 19.092116917626218,
      "grad_norm": 0.3280172049999237,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 21555
    },
    {
      "epoch": 19.096545615589015,
      "grad_norm": 0.2885426878929138,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 21560
    },
    {
      "epoch": 19.100974313551816,
      "grad_norm": 0.22005954384803772,
      "learning_rate": 1e-05,
      "loss": 1.0181,
      "step": 21565
    },
    {
      "epoch": 19.105403011514614,
      "grad_norm": 0.20952272415161133,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 21570
    },
    {
      "epoch": 19.109831709477415,
      "grad_norm": 0.23301991820335388,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 21575
    },
    {
      "epoch": 19.114260407440213,
      "grad_norm": 0.2801075875759125,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 21580
    },
    {
      "epoch": 19.11868910540301,
      "grad_norm": 0.23552002012729645,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 21585
    },
    {
      "epoch": 19.123117803365812,
      "grad_norm": 0.24808785319328308,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 21590
    },
    {
      "epoch": 19.12754650132861,
      "grad_norm": 0.2408321350812912,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 21595
    },
    {
      "epoch": 19.131975199291407,
      "grad_norm": 0.2559404671192169,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 21600
    },
    {
      "epoch": 19.13640389725421,
      "grad_norm": 0.26444193720817566,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 21605
    },
    {
      "epoch": 19.140832595217006,
      "grad_norm": 0.2920604944229126,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 21610
    },
    {
      "epoch": 19.145261293179804,
      "grad_norm": 0.263216495513916,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 21615
    },
    {
      "epoch": 19.149689991142605,
      "grad_norm": 0.28683018684387207,
      "learning_rate": 1e-05,
      "loss": 0.9163,
      "step": 21620
    },
    {
      "epoch": 19.154118689105402,
      "grad_norm": 0.24363479018211365,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 21625
    },
    {
      "epoch": 19.158547387068204,
      "grad_norm": 0.2464655339717865,
      "learning_rate": 1e-05,
      "loss": 0.9212,
      "step": 21630
    },
    {
      "epoch": 19.162976085031,
      "grad_norm": 0.2400534600019455,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 21635
    },
    {
      "epoch": 19.1674047829938,
      "grad_norm": 0.24684584140777588,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 21640
    },
    {
      "epoch": 19.1718334809566,
      "grad_norm": 0.2557712495326996,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 21645
    },
    {
      "epoch": 19.176262178919398,
      "grad_norm": 0.25647491216659546,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 21650
    },
    {
      "epoch": 19.180690876882196,
      "grad_norm": 0.3155398666858673,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 21655
    },
    {
      "epoch": 19.185119574844997,
      "grad_norm": 0.2577785849571228,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 21660
    },
    {
      "epoch": 19.189548272807794,
      "grad_norm": 0.21081209182739258,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 21665
    },
    {
      "epoch": 19.193976970770592,
      "grad_norm": 0.21020257472991943,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 21670
    },
    {
      "epoch": 19.198405668733393,
      "grad_norm": 0.2684951722621918,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 21675
    },
    {
      "epoch": 19.20283436669619,
      "grad_norm": 0.22918616235256195,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 21680
    },
    {
      "epoch": 19.20726306465899,
      "grad_norm": 0.23693285882472992,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 21685
    },
    {
      "epoch": 19.21169176262179,
      "grad_norm": 0.259158194065094,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 21690
    },
    {
      "epoch": 19.216120460584587,
      "grad_norm": 0.35630491375923157,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 21695
    },
    {
      "epoch": 19.22054915854739,
      "grad_norm": 0.2685184180736542,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 21700
    },
    {
      "epoch": 19.224977856510186,
      "grad_norm": 0.2263508439064026,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 21705
    },
    {
      "epoch": 19.229406554472984,
      "grad_norm": 0.24197499454021454,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 21710
    },
    {
      "epoch": 19.233835252435785,
      "grad_norm": 0.2383718341588974,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 21715
    },
    {
      "epoch": 19.238263950398583,
      "grad_norm": 0.2473609894514084,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 21720
    },
    {
      "epoch": 19.24269264836138,
      "grad_norm": 0.2580435276031494,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 21725
    },
    {
      "epoch": 19.24712134632418,
      "grad_norm": 0.2924043834209442,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 21730
    },
    {
      "epoch": 19.25155004428698,
      "grad_norm": 0.23957420885562897,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 21735
    },
    {
      "epoch": 19.255978742249777,
      "grad_norm": 0.2752174437046051,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 21740
    },
    {
      "epoch": 19.260407440212578,
      "grad_norm": 0.3000291585922241,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 21745
    },
    {
      "epoch": 19.264836138175376,
      "grad_norm": 0.26889264583587646,
      "learning_rate": 1e-05,
      "loss": 1.0141,
      "step": 21750
    },
    {
      "epoch": 19.269264836138177,
      "grad_norm": 0.3213759660720825,
      "learning_rate": 1e-05,
      "loss": 1.0338,
      "step": 21755
    },
    {
      "epoch": 19.273693534100975,
      "grad_norm": 0.31049710512161255,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 21760
    },
    {
      "epoch": 19.278122232063772,
      "grad_norm": 0.286528080701828,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 21765
    },
    {
      "epoch": 19.282550930026574,
      "grad_norm": 0.22837868332862854,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 21770
    },
    {
      "epoch": 19.28697962798937,
      "grad_norm": 0.27868813276290894,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 21775
    },
    {
      "epoch": 19.29140832595217,
      "grad_norm": 0.22765976190567017,
      "learning_rate": 1e-05,
      "loss": 0.9319,
      "step": 21780
    },
    {
      "epoch": 19.29583702391497,
      "grad_norm": 0.2459191083908081,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 21785
    },
    {
      "epoch": 19.300265721877768,
      "grad_norm": 0.2618340849876404,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 21790
    },
    {
      "epoch": 19.304694419840565,
      "grad_norm": 0.24597235023975372,
      "learning_rate": 1e-05,
      "loss": 1.052,
      "step": 21795
    },
    {
      "epoch": 19.309123117803367,
      "grad_norm": 0.25850048661231995,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 21800
    },
    {
      "epoch": 19.313551815766164,
      "grad_norm": 0.2148985117673874,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 21805
    },
    {
      "epoch": 19.317980513728962,
      "grad_norm": 0.2388210892677307,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 21810
    },
    {
      "epoch": 19.322409211691763,
      "grad_norm": 0.22469930350780487,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 21815
    },
    {
      "epoch": 19.32683790965456,
      "grad_norm": 0.29969337582588196,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 21820
    },
    {
      "epoch": 19.331266607617362,
      "grad_norm": 0.2328653335571289,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 21825
    },
    {
      "epoch": 19.33569530558016,
      "grad_norm": 0.2838694155216217,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 21830
    },
    {
      "epoch": 19.340124003542957,
      "grad_norm": 0.2548776865005493,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 21835
    },
    {
      "epoch": 19.34455270150576,
      "grad_norm": 0.2791413366794586,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 21840
    },
    {
      "epoch": 19.348981399468556,
      "grad_norm": 0.21865776181221008,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 21845
    },
    {
      "epoch": 19.353410097431354,
      "grad_norm": 0.31672561168670654,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 21850
    },
    {
      "epoch": 19.357838795394155,
      "grad_norm": 0.27901971340179443,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 21855
    },
    {
      "epoch": 19.362267493356953,
      "grad_norm": 0.2634013593196869,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 21860
    },
    {
      "epoch": 19.36669619131975,
      "grad_norm": 0.28205999732017517,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 21865
    },
    {
      "epoch": 19.37112488928255,
      "grad_norm": 0.2697298526763916,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 21870
    },
    {
      "epoch": 19.37555358724535,
      "grad_norm": 0.25014305114746094,
      "learning_rate": 1e-05,
      "loss": 1.0438,
      "step": 21875
    },
    {
      "epoch": 19.37998228520815,
      "grad_norm": 0.23310589790344238,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 21880
    },
    {
      "epoch": 19.384410983170948,
      "grad_norm": 0.27912992238998413,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 21885
    },
    {
      "epoch": 19.388839681133746,
      "grad_norm": 0.3555620610713959,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 21890
    },
    {
      "epoch": 19.393268379096547,
      "grad_norm": 0.260863333940506,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 21895
    },
    {
      "epoch": 19.397697077059345,
      "grad_norm": 0.2534933388233185,
      "learning_rate": 1e-05,
      "loss": 1.027,
      "step": 21900
    },
    {
      "epoch": 19.402125775022142,
      "grad_norm": 0.2706056237220764,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 21905
    },
    {
      "epoch": 19.406554472984944,
      "grad_norm": 0.2713649272918701,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 21910
    },
    {
      "epoch": 19.41098317094774,
      "grad_norm": 0.3194676339626312,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 21915
    },
    {
      "epoch": 19.41541186891054,
      "grad_norm": 0.2466733455657959,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 21920
    },
    {
      "epoch": 19.41984056687334,
      "grad_norm": 0.20290613174438477,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 21925
    },
    {
      "epoch": 19.424269264836138,
      "grad_norm": 0.202281191945076,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 21930
    },
    {
      "epoch": 19.428697962798935,
      "grad_norm": 0.242069274187088,
      "learning_rate": 1e-05,
      "loss": 1.0148,
      "step": 21935
    },
    {
      "epoch": 19.433126660761737,
      "grad_norm": 0.3062588572502136,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 21940
    },
    {
      "epoch": 19.437555358724534,
      "grad_norm": 0.2583722472190857,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 21945
    },
    {
      "epoch": 19.441984056687335,
      "grad_norm": 0.2504075765609741,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 21950
    },
    {
      "epoch": 19.446412754650133,
      "grad_norm": 0.2103910595178604,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 21955
    },
    {
      "epoch": 19.45084145261293,
      "grad_norm": 0.2799288034439087,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 21960
    },
    {
      "epoch": 19.455270150575732,
      "grad_norm": 0.2419409602880478,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 21965
    },
    {
      "epoch": 19.45969884853853,
      "grad_norm": 0.26130715012550354,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 21970
    },
    {
      "epoch": 19.464127546501327,
      "grad_norm": 0.23237186670303345,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 21975
    },
    {
      "epoch": 19.46855624446413,
      "grad_norm": 0.26077741384506226,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 21980
    },
    {
      "epoch": 19.472984942426926,
      "grad_norm": 0.2457740753889084,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 21985
    },
    {
      "epoch": 19.477413640389724,
      "grad_norm": 0.28531643748283386,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 21990
    },
    {
      "epoch": 19.481842338352525,
      "grad_norm": 0.24584434926509857,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 21995
    },
    {
      "epoch": 19.486271036315323,
      "grad_norm": 0.2677319347858429,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 22000
    },
    {
      "epoch": 19.490699734278124,
      "grad_norm": 0.25762227177619934,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 22005
    },
    {
      "epoch": 19.49512843224092,
      "grad_norm": 0.2576358914375305,
      "learning_rate": 1e-05,
      "loss": 1.0578,
      "step": 22010
    },
    {
      "epoch": 19.49955713020372,
      "grad_norm": 0.2575262784957886,
      "learning_rate": 1e-05,
      "loss": 1.0425,
      "step": 22015
    },
    {
      "epoch": 19.50398582816652,
      "grad_norm": 0.2733260691165924,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 22020
    },
    {
      "epoch": 19.508414526129318,
      "grad_norm": 0.24955831468105316,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 22025
    },
    {
      "epoch": 19.512843224092116,
      "grad_norm": 0.23731659352779388,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 22030
    },
    {
      "epoch": 19.517271922054917,
      "grad_norm": 0.27653467655181885,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 22035
    },
    {
      "epoch": 19.521700620017715,
      "grad_norm": 0.2617150545120239,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 22040
    },
    {
      "epoch": 19.526129317980512,
      "grad_norm": 0.26434868574142456,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 22045
    },
    {
      "epoch": 19.530558015943313,
      "grad_norm": 0.26113566756248474,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 22050
    },
    {
      "epoch": 19.53498671390611,
      "grad_norm": 0.2489299476146698,
      "learning_rate": 1e-05,
      "loss": 1.0568,
      "step": 22055
    },
    {
      "epoch": 19.53941541186891,
      "grad_norm": 0.23893140256404877,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 22060
    },
    {
      "epoch": 19.54384410983171,
      "grad_norm": 0.2802172601222992,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 22065
    },
    {
      "epoch": 19.548272807794508,
      "grad_norm": 0.28759679198265076,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 22070
    },
    {
      "epoch": 19.55270150575731,
      "grad_norm": 0.2095140516757965,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 22075
    },
    {
      "epoch": 19.557130203720106,
      "grad_norm": 0.2768591642379761,
      "learning_rate": 1e-05,
      "loss": 0.9045,
      "step": 22080
    },
    {
      "epoch": 19.561558901682904,
      "grad_norm": 0.21753664314746857,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 22085
    },
    {
      "epoch": 19.565987599645705,
      "grad_norm": 0.22399666905403137,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 22090
    },
    {
      "epoch": 19.570416297608503,
      "grad_norm": 0.25039008259773254,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 22095
    },
    {
      "epoch": 19.5748449955713,
      "grad_norm": 0.2238444983959198,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 22100
    },
    {
      "epoch": 19.579273693534102,
      "grad_norm": 0.27676069736480713,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 22105
    },
    {
      "epoch": 19.5837023914969,
      "grad_norm": 0.26960572600364685,
      "learning_rate": 1e-05,
      "loss": 1.0351,
      "step": 22110
    },
    {
      "epoch": 19.588131089459697,
      "grad_norm": 0.2727920711040497,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 22115
    },
    {
      "epoch": 19.5925597874225,
      "grad_norm": 0.2702040374279022,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 22120
    },
    {
      "epoch": 19.596988485385296,
      "grad_norm": 0.31197258830070496,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 22125
    },
    {
      "epoch": 19.601417183348097,
      "grad_norm": 0.2902476489543915,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 22130
    },
    {
      "epoch": 19.605845881310895,
      "grad_norm": 0.2764741778373718,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 22135
    },
    {
      "epoch": 19.610274579273693,
      "grad_norm": 0.2726614773273468,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 22140
    },
    {
      "epoch": 19.614703277236494,
      "grad_norm": 0.24448367953300476,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 22145
    },
    {
      "epoch": 19.61913197519929,
      "grad_norm": 0.24003101885318756,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 22150
    },
    {
      "epoch": 19.62356067316209,
      "grad_norm": 0.21956351399421692,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 22155
    },
    {
      "epoch": 19.62798937112489,
      "grad_norm": 0.2665596902370453,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 22160
    },
    {
      "epoch": 19.632418069087688,
      "grad_norm": 0.2401365339756012,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 22165
    },
    {
      "epoch": 19.636846767050486,
      "grad_norm": 0.23235240578651428,
      "learning_rate": 1e-05,
      "loss": 1.0517,
      "step": 22170
    },
    {
      "epoch": 19.641275465013287,
      "grad_norm": 0.2293304055929184,
      "learning_rate": 1e-05,
      "loss": 1.0054,
      "step": 22175
    },
    {
      "epoch": 19.645704162976084,
      "grad_norm": 0.2629548907279968,
      "learning_rate": 1e-05,
      "loss": 1.0645,
      "step": 22180
    },
    {
      "epoch": 19.650132860938886,
      "grad_norm": 0.27476292848587036,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 22185
    },
    {
      "epoch": 19.654561558901683,
      "grad_norm": 0.23981112241744995,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 22190
    },
    {
      "epoch": 19.65899025686448,
      "grad_norm": 0.2123728096485138,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 22195
    },
    {
      "epoch": 19.663418954827282,
      "grad_norm": 0.25020381808280945,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 22200
    },
    {
      "epoch": 19.66784765279008,
      "grad_norm": 0.25911882519721985,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 22205
    },
    {
      "epoch": 19.672276350752878,
      "grad_norm": 0.2560748755931854,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 22210
    },
    {
      "epoch": 19.67670504871568,
      "grad_norm": 0.2542824447154999,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 22215
    },
    {
      "epoch": 19.681133746678476,
      "grad_norm": 0.24735695123672485,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 22220
    },
    {
      "epoch": 19.685562444641274,
      "grad_norm": 0.2450762242078781,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 22225
    },
    {
      "epoch": 19.689991142604075,
      "grad_norm": 0.2648891806602478,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 22230
    },
    {
      "epoch": 19.694419840566873,
      "grad_norm": 0.26009857654571533,
      "learning_rate": 1e-05,
      "loss": 1.0316,
      "step": 22235
    },
    {
      "epoch": 19.698848538529674,
      "grad_norm": 0.2615038752555847,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 22240
    },
    {
      "epoch": 19.70327723649247,
      "grad_norm": 0.2833743393421173,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 22245
    },
    {
      "epoch": 19.70770593445527,
      "grad_norm": 0.2354872077703476,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 22250
    },
    {
      "epoch": 19.71213463241807,
      "grad_norm": 0.24126538634300232,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 22255
    },
    {
      "epoch": 19.71656333038087,
      "grad_norm": 0.22607719898223877,
      "learning_rate": 1e-05,
      "loss": 1.0269,
      "step": 22260
    },
    {
      "epoch": 19.720992028343666,
      "grad_norm": 0.27492403984069824,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 22265
    },
    {
      "epoch": 19.725420726306467,
      "grad_norm": 0.2301681786775589,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 22270
    },
    {
      "epoch": 19.729849424269265,
      "grad_norm": 0.26673516631126404,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 22275
    },
    {
      "epoch": 19.734278122232062,
      "grad_norm": 0.2584981620311737,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 22280
    },
    {
      "epoch": 19.738706820194864,
      "grad_norm": 0.27978476881980896,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 22285
    },
    {
      "epoch": 19.74313551815766,
      "grad_norm": 0.2619239091873169,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 22290
    },
    {
      "epoch": 19.74756421612046,
      "grad_norm": 0.2218603789806366,
      "learning_rate": 1e-05,
      "loss": 1.0468,
      "step": 22295
    },
    {
      "epoch": 19.75199291408326,
      "grad_norm": 0.27625197172164917,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 22300
    },
    {
      "epoch": 19.756421612046058,
      "grad_norm": 0.2624746561050415,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 22305
    },
    {
      "epoch": 19.76085031000886,
      "grad_norm": 0.24015900492668152,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 22310
    },
    {
      "epoch": 19.765279007971657,
      "grad_norm": 0.2397843301296234,
      "learning_rate": 1e-05,
      "loss": 1.0216,
      "step": 22315
    },
    {
      "epoch": 19.769707705934454,
      "grad_norm": 0.26035407185554504,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 22320
    },
    {
      "epoch": 19.774136403897256,
      "grad_norm": 0.22754158079624176,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 22325
    },
    {
      "epoch": 19.778565101860053,
      "grad_norm": 0.23876163363456726,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 22330
    },
    {
      "epoch": 19.78299379982285,
      "grad_norm": 0.23701980710029602,
      "learning_rate": 1e-05,
      "loss": 1.0421,
      "step": 22335
    },
    {
      "epoch": 19.787422497785652,
      "grad_norm": 0.24347496032714844,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 22340
    },
    {
      "epoch": 19.79185119574845,
      "grad_norm": 0.2831503748893738,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 22345
    },
    {
      "epoch": 19.796279893711247,
      "grad_norm": 0.2691107988357544,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 22350
    },
    {
      "epoch": 19.80070859167405,
      "grad_norm": 0.274152934551239,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 22355
    },
    {
      "epoch": 19.805137289636846,
      "grad_norm": 0.2526584267616272,
      "learning_rate": 1e-05,
      "loss": 0.9277,
      "step": 22360
    },
    {
      "epoch": 19.809565987599647,
      "grad_norm": 0.2391812801361084,
      "learning_rate": 1e-05,
      "loss": 1.0368,
      "step": 22365
    },
    {
      "epoch": 19.813994685562445,
      "grad_norm": 0.22304296493530273,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 22370
    },
    {
      "epoch": 19.818423383525243,
      "grad_norm": 0.2387648969888687,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 22375
    },
    {
      "epoch": 19.822852081488044,
      "grad_norm": 0.29134365916252136,
      "learning_rate": 1e-05,
      "loss": 0.9295,
      "step": 22380
    },
    {
      "epoch": 19.82728077945084,
      "grad_norm": 0.25216689705848694,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 22385
    },
    {
      "epoch": 19.83170947741364,
      "grad_norm": 0.2788964807987213,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 22390
    },
    {
      "epoch": 19.83613817537644,
      "grad_norm": 0.24531854689121246,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 22395
    },
    {
      "epoch": 19.840566873339238,
      "grad_norm": 0.23592737317085266,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 22400
    },
    {
      "epoch": 19.844995571302036,
      "grad_norm": 0.26679787039756775,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 22405
    },
    {
      "epoch": 19.849424269264837,
      "grad_norm": 0.24392372369766235,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 22410
    },
    {
      "epoch": 19.853852967227635,
      "grad_norm": 0.34096843004226685,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 22415
    },
    {
      "epoch": 19.858281665190432,
      "grad_norm": 0.3159085214138031,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 22420
    },
    {
      "epoch": 19.862710363153234,
      "grad_norm": 0.3020704984664917,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 22425
    },
    {
      "epoch": 19.86713906111603,
      "grad_norm": 0.23647166788578033,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 22430
    },
    {
      "epoch": 19.871567759078832,
      "grad_norm": 0.2795269191265106,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 22435
    },
    {
      "epoch": 19.87599645704163,
      "grad_norm": 0.22462648153305054,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 22440
    },
    {
      "epoch": 19.880425155004428,
      "grad_norm": 0.2533538341522217,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 22445
    },
    {
      "epoch": 19.88485385296723,
      "grad_norm": 0.24216006696224213,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 22450
    },
    {
      "epoch": 19.889282550930027,
      "grad_norm": 0.2674252986907959,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 22455
    },
    {
      "epoch": 19.893711248892824,
      "grad_norm": 0.2583083212375641,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 22460
    },
    {
      "epoch": 19.898139946855625,
      "grad_norm": 0.21754281222820282,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 22465
    },
    {
      "epoch": 19.902568644818423,
      "grad_norm": 0.22411461174488068,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 22470
    },
    {
      "epoch": 19.90699734278122,
      "grad_norm": 0.28816187381744385,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 22475
    },
    {
      "epoch": 19.911426040744022,
      "grad_norm": 0.2748092710971832,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 22480
    },
    {
      "epoch": 19.91585473870682,
      "grad_norm": 0.3295396566390991,
      "learning_rate": 1e-05,
      "loss": 1.0708,
      "step": 22485
    },
    {
      "epoch": 19.92028343666962,
      "grad_norm": 0.36053869128227234,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 22490
    },
    {
      "epoch": 19.92471213463242,
      "grad_norm": 0.2501281499862671,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 22495
    },
    {
      "epoch": 19.929140832595216,
      "grad_norm": 0.2649358808994293,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 22500
    },
    {
      "epoch": 19.933569530558017,
      "grad_norm": 0.2116309404373169,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 22505
    },
    {
      "epoch": 19.937998228520815,
      "grad_norm": 0.23288613557815552,
      "learning_rate": 1e-05,
      "loss": 0.9364,
      "step": 22510
    },
    {
      "epoch": 19.942426926483613,
      "grad_norm": 0.20370088517665863,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 22515
    },
    {
      "epoch": 19.946855624446414,
      "grad_norm": 0.24225036799907684,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 22520
    },
    {
      "epoch": 19.95128432240921,
      "grad_norm": 0.2097589075565338,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 22525
    },
    {
      "epoch": 19.95571302037201,
      "grad_norm": 0.22259247303009033,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 22530
    },
    {
      "epoch": 19.96014171833481,
      "grad_norm": 0.27597904205322266,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 22535
    },
    {
      "epoch": 19.964570416297608,
      "grad_norm": 0.23908229172229767,
      "learning_rate": 1e-05,
      "loss": 1.0241,
      "step": 22540
    },
    {
      "epoch": 19.968999114260406,
      "grad_norm": 0.2578098177909851,
      "learning_rate": 1e-05,
      "loss": 1.034,
      "step": 22545
    },
    {
      "epoch": 19.973427812223207,
      "grad_norm": 0.2551516592502594,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 22550
    },
    {
      "epoch": 19.977856510186005,
      "grad_norm": 0.2383735626935959,
      "learning_rate": 1e-05,
      "loss": 0.9271,
      "step": 22555
    },
    {
      "epoch": 19.982285208148806,
      "grad_norm": 0.2537039816379547,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 22560
    },
    {
      "epoch": 19.986713906111603,
      "grad_norm": 0.25406530499458313,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 22565
    },
    {
      "epoch": 19.9911426040744,
      "grad_norm": 0.2307620644569397,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 22570
    },
    {
      "epoch": 19.995571302037202,
      "grad_norm": 0.24770940840244293,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 22575
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.27704715728759766,
      "learning_rate": 1e-05,
      "loss": 1.0308,
      "step": 22580
    },
    {
      "epoch": 20.004428697962798,
      "grad_norm": 0.30626973509788513,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 22585
    },
    {
      "epoch": 20.0088573959256,
      "grad_norm": 0.2711142897605896,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 22590
    },
    {
      "epoch": 20.013286093888397,
      "grad_norm": 0.23982557654380798,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 22595
    },
    {
      "epoch": 20.017714791851194,
      "grad_norm": 0.2538910508155823,
      "learning_rate": 1e-05,
      "loss": 1.0194,
      "step": 22600
    },
    {
      "epoch": 20.022143489813995,
      "grad_norm": 0.23520027101039886,
      "learning_rate": 1e-05,
      "loss": 0.922,
      "step": 22605
    },
    {
      "epoch": 20.026572187776793,
      "grad_norm": 0.24669069051742554,
      "learning_rate": 1e-05,
      "loss": 1.041,
      "step": 22610
    },
    {
      "epoch": 20.031000885739594,
      "grad_norm": 0.28411203622817993,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 22615
    },
    {
      "epoch": 20.035429583702392,
      "grad_norm": 0.22540497779846191,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 22620
    },
    {
      "epoch": 20.03985828166519,
      "grad_norm": 0.2427888959646225,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 22625
    },
    {
      "epoch": 20.04428697962799,
      "grad_norm": 0.2336624711751938,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 22630
    },
    {
      "epoch": 20.04871567759079,
      "grad_norm": 0.2394520789384842,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 22635
    },
    {
      "epoch": 20.053144375553586,
      "grad_norm": 0.28735679388046265,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 22640
    },
    {
      "epoch": 20.057573073516387,
      "grad_norm": 0.24385744333267212,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 22645
    },
    {
      "epoch": 20.062001771479185,
      "grad_norm": 0.22770138084888458,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 22650
    },
    {
      "epoch": 20.066430469441983,
      "grad_norm": 0.2785815894603729,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 22655
    },
    {
      "epoch": 20.070859167404784,
      "grad_norm": 0.3103032410144806,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 22660
    },
    {
      "epoch": 20.07528786536758,
      "grad_norm": 0.24815575778484344,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 22665
    },
    {
      "epoch": 20.07971656333038,
      "grad_norm": 0.24150699377059937,
      "learning_rate": 1e-05,
      "loss": 1.0175,
      "step": 22670
    },
    {
      "epoch": 20.08414526129318,
      "grad_norm": 0.29132309556007385,
      "learning_rate": 1e-05,
      "loss": 1.0414,
      "step": 22675
    },
    {
      "epoch": 20.088573959255978,
      "grad_norm": 0.300889253616333,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 22680
    },
    {
      "epoch": 20.09300265721878,
      "grad_norm": 0.27988967299461365,
      "learning_rate": 1e-05,
      "loss": 1.0309,
      "step": 22685
    },
    {
      "epoch": 20.097431355181577,
      "grad_norm": 0.27934205532073975,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 22690
    },
    {
      "epoch": 20.101860053144375,
      "grad_norm": 0.25654634833335876,
      "learning_rate": 1e-05,
      "loss": 1.0359,
      "step": 22695
    },
    {
      "epoch": 20.106288751107176,
      "grad_norm": 0.24366258084774017,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 22700
    },
    {
      "epoch": 20.110717449069973,
      "grad_norm": 0.2615424692630768,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 22705
    },
    {
      "epoch": 20.11514614703277,
      "grad_norm": 0.2076205313205719,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 22710
    },
    {
      "epoch": 20.119574844995572,
      "grad_norm": 0.2148674577474594,
      "learning_rate": 1e-05,
      "loss": 1.0405,
      "step": 22715
    },
    {
      "epoch": 20.12400354295837,
      "grad_norm": 0.2351323664188385,
      "learning_rate": 1e-05,
      "loss": 1.0314,
      "step": 22720
    },
    {
      "epoch": 20.128432240921168,
      "grad_norm": 0.21035417914390564,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 22725
    },
    {
      "epoch": 20.13286093888397,
      "grad_norm": 0.33949536085128784,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 22730
    },
    {
      "epoch": 20.137289636846766,
      "grad_norm": 0.3018169701099396,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 22735
    },
    {
      "epoch": 20.141718334809568,
      "grad_norm": 0.3865547180175781,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 22740
    },
    {
      "epoch": 20.146147032772365,
      "grad_norm": 0.32810071110725403,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 22745
    },
    {
      "epoch": 20.150575730735163,
      "grad_norm": 0.24213823676109314,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 22750
    },
    {
      "epoch": 20.155004428697964,
      "grad_norm": 0.2582855820655823,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 22755
    },
    {
      "epoch": 20.159433126660762,
      "grad_norm": 0.26272791624069214,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 22760
    },
    {
      "epoch": 20.16386182462356,
      "grad_norm": 0.25247475504875183,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 22765
    },
    {
      "epoch": 20.16829052258636,
      "grad_norm": 0.3538379669189453,
      "learning_rate": 1e-05,
      "loss": 1.0009,
      "step": 22770
    },
    {
      "epoch": 20.17271922054916,
      "grad_norm": 0.2905731201171875,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 22775
    },
    {
      "epoch": 20.177147918511956,
      "grad_norm": 0.29343536496162415,
      "learning_rate": 1e-05,
      "loss": 1.0399,
      "step": 22780
    },
    {
      "epoch": 20.181576616474757,
      "grad_norm": 0.2544710636138916,
      "learning_rate": 1e-05,
      "loss": 1.0339,
      "step": 22785
    },
    {
      "epoch": 20.186005314437555,
      "grad_norm": 0.22626975178718567,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 22790
    },
    {
      "epoch": 20.190434012400353,
      "grad_norm": 0.25439634919166565,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 22795
    },
    {
      "epoch": 20.194862710363154,
      "grad_norm": 0.2761237919330597,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 22800
    },
    {
      "epoch": 20.19929140832595,
      "grad_norm": 0.24387550354003906,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 22805
    },
    {
      "epoch": 20.203720106288753,
      "grad_norm": 0.23981715738773346,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 22810
    },
    {
      "epoch": 20.20814880425155,
      "grad_norm": 0.24644440412521362,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 22815
    },
    {
      "epoch": 20.212577502214348,
      "grad_norm": 0.249264657497406,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 22820
    },
    {
      "epoch": 20.21700620017715,
      "grad_norm": 0.2768544852733612,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 22825
    },
    {
      "epoch": 20.221434898139947,
      "grad_norm": 0.29739975929260254,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 22830
    },
    {
      "epoch": 20.225863596102744,
      "grad_norm": 0.2584186792373657,
      "learning_rate": 1e-05,
      "loss": 1.0362,
      "step": 22835
    },
    {
      "epoch": 20.230292294065546,
      "grad_norm": 0.362277626991272,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 22840
    },
    {
      "epoch": 20.234720992028343,
      "grad_norm": 0.2594701051712036,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 22845
    },
    {
      "epoch": 20.23914968999114,
      "grad_norm": 0.2504386305809021,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 22850
    },
    {
      "epoch": 20.243578387953942,
      "grad_norm": 0.21802587807178497,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 22855
    },
    {
      "epoch": 20.24800708591674,
      "grad_norm": 0.20525391399860382,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 22860
    },
    {
      "epoch": 20.25243578387954,
      "grad_norm": 0.2622871994972229,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 22865
    },
    {
      "epoch": 20.25686448184234,
      "grad_norm": 0.2646791636943817,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 22870
    },
    {
      "epoch": 20.261293179805136,
      "grad_norm": 0.26397275924682617,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 22875
    },
    {
      "epoch": 20.265721877767938,
      "grad_norm": 0.3050857186317444,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 22880
    },
    {
      "epoch": 20.270150575730735,
      "grad_norm": 0.3151940405368805,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 22885
    },
    {
      "epoch": 20.274579273693533,
      "grad_norm": 0.24490241706371307,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 22890
    },
    {
      "epoch": 20.279007971656334,
      "grad_norm": 0.28745564818382263,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 22895
    },
    {
      "epoch": 20.28343666961913,
      "grad_norm": 0.2804032862186432,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 22900
    },
    {
      "epoch": 20.28786536758193,
      "grad_norm": 0.2028869390487671,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 22905
    },
    {
      "epoch": 20.29229406554473,
      "grad_norm": 0.21999183297157288,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 22910
    },
    {
      "epoch": 20.29672276350753,
      "grad_norm": 0.29831916093826294,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 22915
    },
    {
      "epoch": 20.30115146147033,
      "grad_norm": 0.3363560438156128,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 22920
    },
    {
      "epoch": 20.305580159433127,
      "grad_norm": 0.24571125209331512,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 22925
    },
    {
      "epoch": 20.310008857395925,
      "grad_norm": 0.2533973455429077,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 22930
    },
    {
      "epoch": 20.314437555358726,
      "grad_norm": 0.2748803198337555,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 22935
    },
    {
      "epoch": 20.318866253321524,
      "grad_norm": 0.31570225954055786,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 22940
    },
    {
      "epoch": 20.32329495128432,
      "grad_norm": 0.2510981559753418,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 22945
    },
    {
      "epoch": 20.327723649247122,
      "grad_norm": 0.28418129682540894,
      "learning_rate": 1e-05,
      "loss": 1.0448,
      "step": 22950
    },
    {
      "epoch": 20.33215234720992,
      "grad_norm": 0.26578933000564575,
      "learning_rate": 1e-05,
      "loss": 1.0242,
      "step": 22955
    },
    {
      "epoch": 20.336581045172718,
      "grad_norm": 0.2867553234100342,
      "learning_rate": 1e-05,
      "loss": 1.0326,
      "step": 22960
    },
    {
      "epoch": 20.34100974313552,
      "grad_norm": 0.27747517824172974,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 22965
    },
    {
      "epoch": 20.345438441098317,
      "grad_norm": 0.2678050398826599,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 22970
    },
    {
      "epoch": 20.349867139061114,
      "grad_norm": 0.263844758272171,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 22975
    },
    {
      "epoch": 20.354295837023916,
      "grad_norm": 0.2709397077560425,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 22980
    },
    {
      "epoch": 20.358724534986713,
      "grad_norm": 0.26302099227905273,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 22985
    },
    {
      "epoch": 20.363153232949514,
      "grad_norm": 0.22915199398994446,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 22990
    },
    {
      "epoch": 20.367581930912312,
      "grad_norm": 0.2734391987323761,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 22995
    },
    {
      "epoch": 20.37201062887511,
      "grad_norm": 0.23513121902942657,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 23000
    },
    {
      "epoch": 20.37643932683791,
      "grad_norm": 0.2585228979587555,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 23005
    },
    {
      "epoch": 20.38086802480071,
      "grad_norm": 0.3046770691871643,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 23010
    },
    {
      "epoch": 20.385296722763506,
      "grad_norm": 0.31108105182647705,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 23015
    },
    {
      "epoch": 20.389725420726307,
      "grad_norm": 0.2511794865131378,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 23020
    },
    {
      "epoch": 20.394154118689105,
      "grad_norm": 0.27853289246559143,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 23025
    },
    {
      "epoch": 20.398582816651903,
      "grad_norm": 0.25214579701423645,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 23030
    },
    {
      "epoch": 20.403011514614704,
      "grad_norm": 0.25975358486175537,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 23035
    },
    {
      "epoch": 20.4074402125775,
      "grad_norm": 0.26760604977607727,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 23040
    },
    {
      "epoch": 20.411868910540303,
      "grad_norm": 0.2410479485988617,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 23045
    },
    {
      "epoch": 20.4162976085031,
      "grad_norm": 0.26017051935195923,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 23050
    },
    {
      "epoch": 20.420726306465898,
      "grad_norm": 0.25673961639404297,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 23055
    },
    {
      "epoch": 20.4251550044287,
      "grad_norm": 0.22346404194831848,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 23060
    },
    {
      "epoch": 20.429583702391497,
      "grad_norm": 0.23063363134860992,
      "learning_rate": 1e-05,
      "loss": 0.9253,
      "step": 23065
    },
    {
      "epoch": 20.434012400354295,
      "grad_norm": 0.2615554630756378,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 23070
    },
    {
      "epoch": 20.438441098317096,
      "grad_norm": 0.2427557408809662,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 23075
    },
    {
      "epoch": 20.442869796279894,
      "grad_norm": 0.23984217643737793,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 23080
    },
    {
      "epoch": 20.44729849424269,
      "grad_norm": 0.24669641256332397,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 23085
    },
    {
      "epoch": 20.451727192205492,
      "grad_norm": 0.2829103469848633,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 23090
    },
    {
      "epoch": 20.45615589016829,
      "grad_norm": 0.2261546105146408,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 23095
    },
    {
      "epoch": 20.460584588131088,
      "grad_norm": 0.26145729422569275,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 23100
    },
    {
      "epoch": 20.46501328609389,
      "grad_norm": 0.2522001266479492,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 23105
    },
    {
      "epoch": 20.469441984056687,
      "grad_norm": 0.26703497767448425,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 23110
    },
    {
      "epoch": 20.473870682019488,
      "grad_norm": 0.2748298645019531,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 23115
    },
    {
      "epoch": 20.478299379982285,
      "grad_norm": 0.281241238117218,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 23120
    },
    {
      "epoch": 20.482728077945083,
      "grad_norm": 0.2569335997104645,
      "learning_rate": 1e-05,
      "loss": 1.0488,
      "step": 23125
    },
    {
      "epoch": 20.487156775907884,
      "grad_norm": 0.26251575350761414,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 23130
    },
    {
      "epoch": 20.491585473870682,
      "grad_norm": 0.27247336506843567,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 23135
    },
    {
      "epoch": 20.49601417183348,
      "grad_norm": 0.23398526012897491,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 23140
    },
    {
      "epoch": 20.50044286979628,
      "grad_norm": 0.258674293756485,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 23145
    },
    {
      "epoch": 20.50487156775908,
      "grad_norm": 0.284990131855011,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 23150
    },
    {
      "epoch": 20.509300265721876,
      "grad_norm": 0.3295852541923523,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 23155
    },
    {
      "epoch": 20.513728963684677,
      "grad_norm": 0.2717971205711365,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 23160
    },
    {
      "epoch": 20.518157661647475,
      "grad_norm": 0.25128209590911865,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 23165
    },
    {
      "epoch": 20.522586359610276,
      "grad_norm": 0.2315579354763031,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 23170
    },
    {
      "epoch": 20.527015057573074,
      "grad_norm": 0.24004024267196655,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 23175
    },
    {
      "epoch": 20.53144375553587,
      "grad_norm": 0.25360557436943054,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 23180
    },
    {
      "epoch": 20.535872453498673,
      "grad_norm": 0.2642561197280884,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 23185
    },
    {
      "epoch": 20.54030115146147,
      "grad_norm": 0.32052522897720337,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 23190
    },
    {
      "epoch": 20.544729849424268,
      "grad_norm": 0.2145920842885971,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 23195
    },
    {
      "epoch": 20.54915854738707,
      "grad_norm": 0.288447767496109,
      "learning_rate": 1e-05,
      "loss": 1.0422,
      "step": 23200
    },
    {
      "epoch": 20.553587245349867,
      "grad_norm": 0.26760920882225037,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 23205
    },
    {
      "epoch": 20.558015943312665,
      "grad_norm": 0.25050586462020874,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 23210
    },
    {
      "epoch": 20.562444641275466,
      "grad_norm": 0.293376624584198,
      "learning_rate": 1e-05,
      "loss": 1.0505,
      "step": 23215
    },
    {
      "epoch": 20.566873339238263,
      "grad_norm": 0.2831237018108368,
      "learning_rate": 1e-05,
      "loss": 0.9408,
      "step": 23220
    },
    {
      "epoch": 20.571302037201065,
      "grad_norm": 0.2854495048522949,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 23225
    },
    {
      "epoch": 20.575730735163862,
      "grad_norm": 0.23675929009914398,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 23230
    },
    {
      "epoch": 20.58015943312666,
      "grad_norm": 0.2307220697402954,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 23235
    },
    {
      "epoch": 20.58458813108946,
      "grad_norm": 0.258803129196167,
      "learning_rate": 1e-05,
      "loss": 1.0341,
      "step": 23240
    },
    {
      "epoch": 20.58901682905226,
      "grad_norm": 0.35433143377304077,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 23245
    },
    {
      "epoch": 20.593445527015056,
      "grad_norm": 0.2867363691329956,
      "learning_rate": 1e-05,
      "loss": 1.0272,
      "step": 23250
    },
    {
      "epoch": 20.597874224977858,
      "grad_norm": 0.31190216541290283,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 23255
    },
    {
      "epoch": 20.602302922940655,
      "grad_norm": 0.25837382674217224,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 23260
    },
    {
      "epoch": 20.606731620903453,
      "grad_norm": 0.29861852526664734,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 23265
    },
    {
      "epoch": 20.611160318866254,
      "grad_norm": 0.2336566150188446,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 23270
    },
    {
      "epoch": 20.615589016829052,
      "grad_norm": 0.319244384765625,
      "learning_rate": 1e-05,
      "loss": 0.9125,
      "step": 23275
    },
    {
      "epoch": 20.62001771479185,
      "grad_norm": 0.2418716847896576,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 23280
    },
    {
      "epoch": 20.62444641275465,
      "grad_norm": 0.2546244263648987,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 23285
    },
    {
      "epoch": 20.62887511071745,
      "grad_norm": 0.22904939949512482,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 23290
    },
    {
      "epoch": 20.63330380868025,
      "grad_norm": 0.28012776374816895,
      "learning_rate": 1e-05,
      "loss": 1.0341,
      "step": 23295
    },
    {
      "epoch": 20.637732506643047,
      "grad_norm": 0.30803826451301575,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 23300
    },
    {
      "epoch": 20.642161204605845,
      "grad_norm": 0.3214375078678131,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 23305
    },
    {
      "epoch": 20.646589902568646,
      "grad_norm": 0.2672041356563568,
      "learning_rate": 1e-05,
      "loss": 1.017,
      "step": 23310
    },
    {
      "epoch": 20.651018600531444,
      "grad_norm": 0.23594099283218384,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 23315
    },
    {
      "epoch": 20.65544729849424,
      "grad_norm": 0.22315728664398193,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 23320
    },
    {
      "epoch": 20.659875996457043,
      "grad_norm": 0.25356242060661316,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 23325
    },
    {
      "epoch": 20.66430469441984,
      "grad_norm": 0.3012852966785431,
      "learning_rate": 1e-05,
      "loss": 1.0357,
      "step": 23330
    },
    {
      "epoch": 20.668733392382638,
      "grad_norm": 0.21883249282836914,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 23335
    },
    {
      "epoch": 20.67316209034544,
      "grad_norm": 0.21413317322731018,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 23340
    },
    {
      "epoch": 20.677590788308237,
      "grad_norm": 0.2563398778438568,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 23345
    },
    {
      "epoch": 20.682019486271038,
      "grad_norm": 0.2521705627441406,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 23350
    },
    {
      "epoch": 20.686448184233836,
      "grad_norm": 0.2602318227291107,
      "learning_rate": 1e-05,
      "loss": 1.0174,
      "step": 23355
    },
    {
      "epoch": 20.690876882196633,
      "grad_norm": 0.21690435707569122,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 23360
    },
    {
      "epoch": 20.695305580159435,
      "grad_norm": 0.2532104253768921,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 23365
    },
    {
      "epoch": 20.699734278122232,
      "grad_norm": 0.23877738416194916,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 23370
    },
    {
      "epoch": 20.70416297608503,
      "grad_norm": 0.28318607807159424,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 23375
    },
    {
      "epoch": 20.70859167404783,
      "grad_norm": 0.2561405301094055,
      "learning_rate": 1e-05,
      "loss": 1.0329,
      "step": 23380
    },
    {
      "epoch": 20.71302037201063,
      "grad_norm": 0.24713988602161407,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 23385
    },
    {
      "epoch": 20.717449069973426,
      "grad_norm": 0.25251656770706177,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 23390
    },
    {
      "epoch": 20.721877767936228,
      "grad_norm": 0.24469034373760223,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 23395
    },
    {
      "epoch": 20.726306465899025,
      "grad_norm": 0.2381730079650879,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 23400
    },
    {
      "epoch": 20.730735163861823,
      "grad_norm": 0.24882963299751282,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 23405
    },
    {
      "epoch": 20.735163861824624,
      "grad_norm": 0.2904219329357147,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 23410
    },
    {
      "epoch": 20.739592559787422,
      "grad_norm": 0.2203953117132187,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 23415
    },
    {
      "epoch": 20.744021257750223,
      "grad_norm": 0.2897529602050781,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 23420
    },
    {
      "epoch": 20.74844995571302,
      "grad_norm": 0.25622791051864624,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 23425
    },
    {
      "epoch": 20.75287865367582,
      "grad_norm": 0.3021980822086334,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 23430
    },
    {
      "epoch": 20.75730735163862,
      "grad_norm": 0.27540022134780884,
      "learning_rate": 1e-05,
      "loss": 1.0412,
      "step": 23435
    },
    {
      "epoch": 20.761736049601417,
      "grad_norm": 0.26078498363494873,
      "learning_rate": 1e-05,
      "loss": 0.9296,
      "step": 23440
    },
    {
      "epoch": 20.766164747564215,
      "grad_norm": 0.2253457009792328,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 23445
    },
    {
      "epoch": 20.770593445527016,
      "grad_norm": 0.24515530467033386,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 23450
    },
    {
      "epoch": 20.775022143489814,
      "grad_norm": 0.28406015038490295,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 23455
    },
    {
      "epoch": 20.77945084145261,
      "grad_norm": 0.2737155854701996,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 23460
    },
    {
      "epoch": 20.783879539415413,
      "grad_norm": 0.2011604607105255,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 23465
    },
    {
      "epoch": 20.78830823737821,
      "grad_norm": 0.22934001684188843,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 23470
    },
    {
      "epoch": 20.79273693534101,
      "grad_norm": 0.26330819725990295,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 23475
    },
    {
      "epoch": 20.79716563330381,
      "grad_norm": 0.29069122672080994,
      "learning_rate": 1e-05,
      "loss": 1.0148,
      "step": 23480
    },
    {
      "epoch": 20.801594331266607,
      "grad_norm": 0.2525894045829773,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 23485
    },
    {
      "epoch": 20.806023029229408,
      "grad_norm": 0.23904164135456085,
      "learning_rate": 1e-05,
      "loss": 1.0183,
      "step": 23490
    },
    {
      "epoch": 20.810451727192206,
      "grad_norm": 0.2577880024909973,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 23495
    },
    {
      "epoch": 20.814880425155003,
      "grad_norm": 0.2428380846977234,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 23500
    },
    {
      "epoch": 20.819309123117804,
      "grad_norm": 0.240816131234169,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 23505
    },
    {
      "epoch": 20.823737821080602,
      "grad_norm": 0.21072085201740265,
      "learning_rate": 1e-05,
      "loss": 1.0543,
      "step": 23510
    },
    {
      "epoch": 20.8281665190434,
      "grad_norm": 0.23351332545280457,
      "learning_rate": 1e-05,
      "loss": 1.0621,
      "step": 23515
    },
    {
      "epoch": 20.8325952170062,
      "grad_norm": 0.27252712845802307,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 23520
    },
    {
      "epoch": 20.837023914969,
      "grad_norm": 0.2045300453901291,
      "learning_rate": 1e-05,
      "loss": 1.0194,
      "step": 23525
    },
    {
      "epoch": 20.841452612931796,
      "grad_norm": 0.2848270833492279,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 23530
    },
    {
      "epoch": 20.845881310894598,
      "grad_norm": 0.2638493478298187,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 23535
    },
    {
      "epoch": 20.850310008857395,
      "grad_norm": 0.22406628727912903,
      "learning_rate": 1e-05,
      "loss": 1.0184,
      "step": 23540
    },
    {
      "epoch": 20.854738706820196,
      "grad_norm": 0.29318952560424805,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 23545
    },
    {
      "epoch": 20.859167404782994,
      "grad_norm": 0.27850982546806335,
      "learning_rate": 1e-05,
      "loss": 1.0382,
      "step": 23550
    },
    {
      "epoch": 20.86359610274579,
      "grad_norm": 0.2463536560535431,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 23555
    },
    {
      "epoch": 20.868024800708593,
      "grad_norm": 0.2756263017654419,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 23560
    },
    {
      "epoch": 20.87245349867139,
      "grad_norm": 0.24165046215057373,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 23565
    },
    {
      "epoch": 20.876882196634188,
      "grad_norm": 0.22937564551830292,
      "learning_rate": 1e-05,
      "loss": 1.0296,
      "step": 23570
    },
    {
      "epoch": 20.88131089459699,
      "grad_norm": 0.23448596894741058,
      "learning_rate": 1e-05,
      "loss": 1.0403,
      "step": 23575
    },
    {
      "epoch": 20.885739592559787,
      "grad_norm": 0.24200119078159332,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 23580
    },
    {
      "epoch": 20.890168290522585,
      "grad_norm": 0.25208789110183716,
      "learning_rate": 1e-05,
      "loss": 1.0246,
      "step": 23585
    },
    {
      "epoch": 20.894596988485386,
      "grad_norm": 0.2373015284538269,
      "learning_rate": 1e-05,
      "loss": 1.0111,
      "step": 23590
    },
    {
      "epoch": 20.899025686448184,
      "grad_norm": 0.29320967197418213,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 23595
    },
    {
      "epoch": 20.903454384410985,
      "grad_norm": 0.2730688750743866,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 23600
    },
    {
      "epoch": 20.907883082373782,
      "grad_norm": 0.28743427991867065,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 23605
    },
    {
      "epoch": 20.91231178033658,
      "grad_norm": 0.29893437027931213,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 23610
    },
    {
      "epoch": 20.91674047829938,
      "grad_norm": 0.23643238842487335,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 23615
    },
    {
      "epoch": 20.92116917626218,
      "grad_norm": 0.24388255178928375,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 23620
    },
    {
      "epoch": 20.925597874224977,
      "grad_norm": 0.2582964599132538,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 23625
    },
    {
      "epoch": 20.930026572187778,
      "grad_norm": 0.27074819803237915,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 23630
    },
    {
      "epoch": 20.934455270150576,
      "grad_norm": 0.25435471534729004,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 23635
    },
    {
      "epoch": 20.938883968113373,
      "grad_norm": 0.22766220569610596,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 23640
    },
    {
      "epoch": 20.943312666076174,
      "grad_norm": 0.2583090662956238,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 23645
    },
    {
      "epoch": 20.947741364038972,
      "grad_norm": 0.2744172513484955,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 23650
    },
    {
      "epoch": 20.95217006200177,
      "grad_norm": 0.22137551009655,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 23655
    },
    {
      "epoch": 20.95659875996457,
      "grad_norm": 0.2544724941253662,
      "learning_rate": 1e-05,
      "loss": 1.031,
      "step": 23660
    },
    {
      "epoch": 20.96102745792737,
      "grad_norm": 0.2587026059627533,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 23665
    },
    {
      "epoch": 20.96545615589017,
      "grad_norm": 0.24419590830802917,
      "learning_rate": 1e-05,
      "loss": 1.0277,
      "step": 23670
    },
    {
      "epoch": 20.969884853852967,
      "grad_norm": 0.2535645663738251,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 23675
    },
    {
      "epoch": 20.974313551815765,
      "grad_norm": 0.26606446504592896,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 23680
    },
    {
      "epoch": 20.978742249778566,
      "grad_norm": 0.2687552571296692,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 23685
    },
    {
      "epoch": 20.983170947741364,
      "grad_norm": 0.288351833820343,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 23690
    },
    {
      "epoch": 20.98759964570416,
      "grad_norm": 0.2936451733112335,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 23695
    },
    {
      "epoch": 20.992028343666963,
      "grad_norm": 0.2874999940395355,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 23700
    },
    {
      "epoch": 20.99645704162976,
      "grad_norm": 0.2770979702472687,
      "learning_rate": 1e-05,
      "loss": 1.0412,
      "step": 23705
    },
    {
      "epoch": 21.000885739592558,
      "grad_norm": 0.25790080428123474,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 23710
    },
    {
      "epoch": 21.00531443755536,
      "grad_norm": 0.1928500235080719,
      "learning_rate": 1e-05,
      "loss": 1.0185,
      "step": 23715
    },
    {
      "epoch": 21.009743135518157,
      "grad_norm": 0.2296430766582489,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 23720
    },
    {
      "epoch": 21.014171833480958,
      "grad_norm": 0.2535626292228699,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 23725
    },
    {
      "epoch": 21.018600531443756,
      "grad_norm": 0.22882448136806488,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 23730
    },
    {
      "epoch": 21.023029229406553,
      "grad_norm": 0.24649257957935333,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 23735
    },
    {
      "epoch": 21.027457927369355,
      "grad_norm": 0.21738092601299286,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 23740
    },
    {
      "epoch": 21.031886625332152,
      "grad_norm": 0.25209927558898926,
      "learning_rate": 1e-05,
      "loss": 1.0515,
      "step": 23745
    },
    {
      "epoch": 21.03631532329495,
      "grad_norm": 0.20074862241744995,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 23750
    },
    {
      "epoch": 21.04074402125775,
      "grad_norm": 0.21273909509181976,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 23755
    },
    {
      "epoch": 21.04517271922055,
      "grad_norm": 0.22330664098262787,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 23760
    },
    {
      "epoch": 21.049601417183347,
      "grad_norm": 0.28758561611175537,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 23765
    },
    {
      "epoch": 21.054030115146148,
      "grad_norm": 0.22504737973213196,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 23770
    },
    {
      "epoch": 21.058458813108945,
      "grad_norm": 0.23204205930233002,
      "learning_rate": 1e-05,
      "loss": 1.0172,
      "step": 23775
    },
    {
      "epoch": 21.062887511071747,
      "grad_norm": 0.26115328073501587,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 23780
    },
    {
      "epoch": 21.067316209034544,
      "grad_norm": 0.2827441990375519,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 23785
    },
    {
      "epoch": 21.071744906997342,
      "grad_norm": 0.2512715756893158,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 23790
    },
    {
      "epoch": 21.076173604960143,
      "grad_norm": 0.32470351457595825,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 23795
    },
    {
      "epoch": 21.08060230292294,
      "grad_norm": 0.2564871311187744,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 23800
    },
    {
      "epoch": 21.08503100088574,
      "grad_norm": 0.3475436866283417,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 23805
    },
    {
      "epoch": 21.08945969884854,
      "grad_norm": 0.24595697224140167,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 23810
    },
    {
      "epoch": 21.093888396811337,
      "grad_norm": 0.20949521660804749,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 23815
    },
    {
      "epoch": 21.098317094774135,
      "grad_norm": 0.22952190041542053,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 23820
    },
    {
      "epoch": 21.102745792736936,
      "grad_norm": 0.2702259421348572,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 23825
    },
    {
      "epoch": 21.107174490699734,
      "grad_norm": 0.23519152402877808,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 23830
    },
    {
      "epoch": 21.11160318866253,
      "grad_norm": 0.2529354989528656,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 23835
    },
    {
      "epoch": 21.116031886625333,
      "grad_norm": 0.2953972816467285,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 23840
    },
    {
      "epoch": 21.12046058458813,
      "grad_norm": 0.21525080502033234,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 23845
    },
    {
      "epoch": 21.12488928255093,
      "grad_norm": 0.23865020275115967,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 23850
    },
    {
      "epoch": 21.12931798051373,
      "grad_norm": 0.23286272585391998,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 23855
    },
    {
      "epoch": 21.133746678476527,
      "grad_norm": 0.29467570781707764,
      "learning_rate": 1e-05,
      "loss": 1.0257,
      "step": 23860
    },
    {
      "epoch": 21.138175376439328,
      "grad_norm": 0.2763665020465851,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 23865
    },
    {
      "epoch": 21.142604074402126,
      "grad_norm": 0.22636842727661133,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 23870
    },
    {
      "epoch": 21.147032772364923,
      "grad_norm": 0.3360186219215393,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 23875
    },
    {
      "epoch": 21.151461470327725,
      "grad_norm": 0.2123258411884308,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 23880
    },
    {
      "epoch": 21.155890168290522,
      "grad_norm": 0.26806992292404175,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 23885
    },
    {
      "epoch": 21.16031886625332,
      "grad_norm": 0.24717672169208527,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 23890
    },
    {
      "epoch": 21.16474756421612,
      "grad_norm": 0.2531307637691498,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 23895
    },
    {
      "epoch": 21.16917626217892,
      "grad_norm": 0.21396209299564362,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 23900
    },
    {
      "epoch": 21.17360496014172,
      "grad_norm": 0.26138243079185486,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 23905
    },
    {
      "epoch": 21.178033658104518,
      "grad_norm": 0.317848801612854,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 23910
    },
    {
      "epoch": 21.182462356067315,
      "grad_norm": 0.25748392939567566,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 23915
    },
    {
      "epoch": 21.186891054030117,
      "grad_norm": 0.23696640133857727,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 23920
    },
    {
      "epoch": 21.191319751992914,
      "grad_norm": 0.2723008692264557,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 23925
    },
    {
      "epoch": 21.195748449955712,
      "grad_norm": 0.245529443025589,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 23930
    },
    {
      "epoch": 21.200177147918513,
      "grad_norm": 0.3217242360115051,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 23935
    },
    {
      "epoch": 21.20460584588131,
      "grad_norm": 0.24378260970115662,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 23940
    },
    {
      "epoch": 21.20903454384411,
      "grad_norm": 0.2365870177745819,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 23945
    },
    {
      "epoch": 21.21346324180691,
      "grad_norm": 0.2485848218202591,
      "learning_rate": 1e-05,
      "loss": 1.034,
      "step": 23950
    },
    {
      "epoch": 21.217891939769707,
      "grad_norm": 0.331729918718338,
      "learning_rate": 1e-05,
      "loss": 1.0576,
      "step": 23955
    },
    {
      "epoch": 21.22232063773251,
      "grad_norm": 0.3377779424190521,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 23960
    },
    {
      "epoch": 21.226749335695306,
      "grad_norm": 0.22519612312316895,
      "learning_rate": 1e-05,
      "loss": 1.0476,
      "step": 23965
    },
    {
      "epoch": 21.231178033658104,
      "grad_norm": 0.2346481829881668,
      "learning_rate": 1e-05,
      "loss": 0.9299,
      "step": 23970
    },
    {
      "epoch": 21.235606731620905,
      "grad_norm": 0.20597229897975922,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 23975
    },
    {
      "epoch": 21.240035429583703,
      "grad_norm": 0.24908660352230072,
      "learning_rate": 1e-05,
      "loss": 1.0221,
      "step": 23980
    },
    {
      "epoch": 21.2444641275465,
      "grad_norm": 0.23830406367778778,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 23985
    },
    {
      "epoch": 21.2488928255093,
      "grad_norm": 0.32512572407722473,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 23990
    },
    {
      "epoch": 21.2533215234721,
      "grad_norm": 0.26207512617111206,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 23995
    },
    {
      "epoch": 21.257750221434897,
      "grad_norm": 0.2201402187347412,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 24000
    },
    {
      "epoch": 21.262178919397698,
      "grad_norm": 0.23515959084033966,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 24005
    },
    {
      "epoch": 21.266607617360496,
      "grad_norm": 0.25755035877227783,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 24010
    },
    {
      "epoch": 21.271036315323293,
      "grad_norm": 0.23354493081569672,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 24015
    },
    {
      "epoch": 21.275465013286095,
      "grad_norm": 0.2569540739059448,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 24020
    },
    {
      "epoch": 21.279893711248892,
      "grad_norm": 0.24763429164886475,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 24025
    },
    {
      "epoch": 21.284322409211693,
      "grad_norm": 0.23891708254814148,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 24030
    },
    {
      "epoch": 21.28875110717449,
      "grad_norm": 0.24977341294288635,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 24035
    },
    {
      "epoch": 21.29317980513729,
      "grad_norm": 0.25577160716056824,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 24040
    },
    {
      "epoch": 21.29760850310009,
      "grad_norm": 0.2855280935764313,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 24045
    },
    {
      "epoch": 21.302037201062888,
      "grad_norm": 0.27523383498191833,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 24050
    },
    {
      "epoch": 21.306465899025685,
      "grad_norm": 0.23196375370025635,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 24055
    },
    {
      "epoch": 21.310894596988486,
      "grad_norm": 0.2422071099281311,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 24060
    },
    {
      "epoch": 21.315323294951284,
      "grad_norm": 0.23991167545318604,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 24065
    },
    {
      "epoch": 21.31975199291408,
      "grad_norm": 0.23510125279426575,
      "learning_rate": 1e-05,
      "loss": 1.0097,
      "step": 24070
    },
    {
      "epoch": 21.324180690876883,
      "grad_norm": 0.212108314037323,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 24075
    },
    {
      "epoch": 21.32860938883968,
      "grad_norm": 0.23386695981025696,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 24080
    },
    {
      "epoch": 21.333038086802482,
      "grad_norm": 0.25324028730392456,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 24085
    },
    {
      "epoch": 21.33746678476528,
      "grad_norm": 0.2725676894187927,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 24090
    },
    {
      "epoch": 21.341895482728077,
      "grad_norm": 0.2590475380420685,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 24095
    },
    {
      "epoch": 21.34632418069088,
      "grad_norm": 0.26223626732826233,
      "learning_rate": 1e-05,
      "loss": 1.0532,
      "step": 24100
    },
    {
      "epoch": 21.350752878653676,
      "grad_norm": 0.23049882054328918,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 24105
    },
    {
      "epoch": 21.355181576616474,
      "grad_norm": 0.264771431684494,
      "learning_rate": 1e-05,
      "loss": 1.0395,
      "step": 24110
    },
    {
      "epoch": 21.359610274579275,
      "grad_norm": 0.22449934482574463,
      "learning_rate": 1e-05,
      "loss": 1.0097,
      "step": 24115
    },
    {
      "epoch": 21.364038972542073,
      "grad_norm": 0.24789178371429443,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 24120
    },
    {
      "epoch": 21.36846767050487,
      "grad_norm": 0.2665213346481323,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 24125
    },
    {
      "epoch": 21.37289636846767,
      "grad_norm": 0.2734752297401428,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 24130
    },
    {
      "epoch": 21.37732506643047,
      "grad_norm": 0.2670280933380127,
      "learning_rate": 1e-05,
      "loss": 1.0409,
      "step": 24135
    },
    {
      "epoch": 21.381753764393267,
      "grad_norm": 0.23711976408958435,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 24140
    },
    {
      "epoch": 21.386182462356068,
      "grad_norm": 0.23111297190189362,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 24145
    },
    {
      "epoch": 21.390611160318866,
      "grad_norm": 0.2747046947479248,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 24150
    },
    {
      "epoch": 21.395039858281667,
      "grad_norm": 0.24125155806541443,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 24155
    },
    {
      "epoch": 21.399468556244464,
      "grad_norm": 0.25119295716285706,
      "learning_rate": 1e-05,
      "loss": 0.9261,
      "step": 24160
    },
    {
      "epoch": 21.403897254207262,
      "grad_norm": 0.23565930128097534,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 24165
    },
    {
      "epoch": 21.408325952170063,
      "grad_norm": 0.23585757613182068,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 24170
    },
    {
      "epoch": 21.41275465013286,
      "grad_norm": 0.23910439014434814,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 24175
    },
    {
      "epoch": 21.41718334809566,
      "grad_norm": 0.233554869890213,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 24180
    },
    {
      "epoch": 21.42161204605846,
      "grad_norm": 0.26669591665267944,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 24185
    },
    {
      "epoch": 21.426040744021257,
      "grad_norm": 0.2674793601036072,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 24190
    },
    {
      "epoch": 21.430469441984055,
      "grad_norm": 0.239483043551445,
      "learning_rate": 1e-05,
      "loss": 1.0398,
      "step": 24195
    },
    {
      "epoch": 21.434898139946856,
      "grad_norm": 0.2558232545852661,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 24200
    },
    {
      "epoch": 21.439326837909654,
      "grad_norm": 0.262599378824234,
      "learning_rate": 1e-05,
      "loss": 1.0268,
      "step": 24205
    },
    {
      "epoch": 21.443755535872455,
      "grad_norm": 0.2378520518541336,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 24210
    },
    {
      "epoch": 21.448184233835253,
      "grad_norm": 0.35304948687553406,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 24215
    },
    {
      "epoch": 21.45261293179805,
      "grad_norm": 0.25574415922164917,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 24220
    },
    {
      "epoch": 21.45704162976085,
      "grad_norm": 0.3049648404121399,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 24225
    },
    {
      "epoch": 21.46147032772365,
      "grad_norm": 0.220179945230484,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 24230
    },
    {
      "epoch": 21.465899025686447,
      "grad_norm": 0.22716927528381348,
      "learning_rate": 1e-05,
      "loss": 0.9408,
      "step": 24235
    },
    {
      "epoch": 21.47032772364925,
      "grad_norm": 0.25995850563049316,
      "learning_rate": 1e-05,
      "loss": 1.0297,
      "step": 24240
    },
    {
      "epoch": 21.474756421612046,
      "grad_norm": 0.22999300062656403,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 24245
    },
    {
      "epoch": 21.479185119574844,
      "grad_norm": 0.21122734248638153,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 24250
    },
    {
      "epoch": 21.483613817537645,
      "grad_norm": 0.197481170296669,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 24255
    },
    {
      "epoch": 21.488042515500442,
      "grad_norm": 0.22244811058044434,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 24260
    },
    {
      "epoch": 21.49247121346324,
      "grad_norm": 0.22211964428424835,
      "learning_rate": 1e-05,
      "loss": 1.0269,
      "step": 24265
    },
    {
      "epoch": 21.49689991142604,
      "grad_norm": 0.28239572048187256,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 24270
    },
    {
      "epoch": 21.50132860938884,
      "grad_norm": 0.21171702444553375,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 24275
    },
    {
      "epoch": 21.50575730735164,
      "grad_norm": 0.2408788949251175,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 24280
    },
    {
      "epoch": 21.510186005314438,
      "grad_norm": 0.23125869035720825,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 24285
    },
    {
      "epoch": 21.514614703277235,
      "grad_norm": 0.2452387511730194,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 24290
    },
    {
      "epoch": 21.519043401240037,
      "grad_norm": 0.224269837141037,
      "learning_rate": 1e-05,
      "loss": 1.0303,
      "step": 24295
    },
    {
      "epoch": 21.523472099202834,
      "grad_norm": 0.28405871987342834,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 24300
    },
    {
      "epoch": 21.527900797165632,
      "grad_norm": 0.2887643575668335,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 24305
    },
    {
      "epoch": 21.532329495128433,
      "grad_norm": 0.29802432656288147,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 24310
    },
    {
      "epoch": 21.53675819309123,
      "grad_norm": 0.23205694556236267,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 24315
    },
    {
      "epoch": 21.54118689105403,
      "grad_norm": 0.28083962202072144,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 24320
    },
    {
      "epoch": 21.54561558901683,
      "grad_norm": 0.3045998811721802,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 24325
    },
    {
      "epoch": 21.550044286979627,
      "grad_norm": 0.23261308670043945,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 24330
    },
    {
      "epoch": 21.55447298494243,
      "grad_norm": 0.27049073576927185,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 24335
    },
    {
      "epoch": 21.558901682905226,
      "grad_norm": 0.2971896231174469,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 24340
    },
    {
      "epoch": 21.563330380868024,
      "grad_norm": 0.2367928922176361,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 24345
    },
    {
      "epoch": 21.567759078830825,
      "grad_norm": 0.28476718068122864,
      "learning_rate": 1e-05,
      "loss": 1.0602,
      "step": 24350
    },
    {
      "epoch": 21.572187776793623,
      "grad_norm": 0.23783431947231293,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 24355
    },
    {
      "epoch": 21.57661647475642,
      "grad_norm": 0.3048815131187439,
      "learning_rate": 1e-05,
      "loss": 1.0457,
      "step": 24360
    },
    {
      "epoch": 21.58104517271922,
      "grad_norm": 0.2317344695329666,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 24365
    },
    {
      "epoch": 21.58547387068202,
      "grad_norm": 0.25966066122055054,
      "learning_rate": 1e-05,
      "loss": 1.0365,
      "step": 24370
    },
    {
      "epoch": 21.589902568644817,
      "grad_norm": 0.23673969507217407,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 24375
    },
    {
      "epoch": 21.594331266607618,
      "grad_norm": 0.26256829500198364,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 24380
    },
    {
      "epoch": 21.598759964570416,
      "grad_norm": 0.23464412987232208,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 24385
    },
    {
      "epoch": 21.603188662533213,
      "grad_norm": 0.26200950145721436,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 24390
    },
    {
      "epoch": 21.607617360496015,
      "grad_norm": 0.24826548993587494,
      "learning_rate": 1e-05,
      "loss": 1.0343,
      "step": 24395
    },
    {
      "epoch": 21.612046058458812,
      "grad_norm": 0.26936453580856323,
      "learning_rate": 1e-05,
      "loss": 0.9522,
      "step": 24400
    },
    {
      "epoch": 21.616474756421614,
      "grad_norm": 0.3161003887653351,
      "learning_rate": 1e-05,
      "loss": 1.0235,
      "step": 24405
    },
    {
      "epoch": 21.62090345438441,
      "grad_norm": 0.25600945949554443,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 24410
    },
    {
      "epoch": 21.62533215234721,
      "grad_norm": 0.24475471675395966,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 24415
    },
    {
      "epoch": 21.62976085031001,
      "grad_norm": 0.24165140092372894,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 24420
    },
    {
      "epoch": 21.634189548272808,
      "grad_norm": 0.22416894137859344,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 24425
    },
    {
      "epoch": 21.638618246235605,
      "grad_norm": 0.24341599643230438,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 24430
    },
    {
      "epoch": 21.643046944198407,
      "grad_norm": 0.21842268109321594,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 24435
    },
    {
      "epoch": 21.647475642161204,
      "grad_norm": 0.28387102484703064,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 24440
    },
    {
      "epoch": 21.651904340124002,
      "grad_norm": 0.2577364444732666,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 24445
    },
    {
      "epoch": 21.656333038086803,
      "grad_norm": 0.30951401591300964,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 24450
    },
    {
      "epoch": 21.6607617360496,
      "grad_norm": 0.418165385723114,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 24455
    },
    {
      "epoch": 21.665190434012402,
      "grad_norm": 0.2846079468727112,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 24460
    },
    {
      "epoch": 21.6696191319752,
      "grad_norm": 0.28966230154037476,
      "learning_rate": 1e-05,
      "loss": 1.0712,
      "step": 24465
    },
    {
      "epoch": 21.674047829937997,
      "grad_norm": 0.2394290566444397,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 24470
    },
    {
      "epoch": 21.6784765279008,
      "grad_norm": 0.2498738318681717,
      "learning_rate": 1e-05,
      "loss": 1.0283,
      "step": 24475
    },
    {
      "epoch": 21.682905225863596,
      "grad_norm": 0.27644601464271545,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 24480
    },
    {
      "epoch": 21.687333923826394,
      "grad_norm": 0.2564210295677185,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 24485
    },
    {
      "epoch": 21.691762621789195,
      "grad_norm": 0.21085625886917114,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 24490
    },
    {
      "epoch": 21.696191319751993,
      "grad_norm": 0.20350492000579834,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 24495
    },
    {
      "epoch": 21.70062001771479,
      "grad_norm": 0.2390928566455841,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 24500
    },
    {
      "epoch": 21.70504871567759,
      "grad_norm": 0.26607394218444824,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 24505
    },
    {
      "epoch": 21.70947741364039,
      "grad_norm": 0.23229064047336578,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 24510
    },
    {
      "epoch": 21.713906111603187,
      "grad_norm": 0.21496142446994781,
      "learning_rate": 1e-05,
      "loss": 1.0238,
      "step": 24515
    },
    {
      "epoch": 21.718334809565988,
      "grad_norm": 0.24185022711753845,
      "learning_rate": 1e-05,
      "loss": 1.0588,
      "step": 24520
    },
    {
      "epoch": 21.722763507528786,
      "grad_norm": 0.2623401880264282,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 24525
    },
    {
      "epoch": 21.727192205491587,
      "grad_norm": 0.27695274353027344,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 24530
    },
    {
      "epoch": 21.731620903454385,
      "grad_norm": 0.20083735883235931,
      "learning_rate": 1e-05,
      "loss": 1.037,
      "step": 24535
    },
    {
      "epoch": 21.736049601417182,
      "grad_norm": 0.2114323377609253,
      "learning_rate": 1e-05,
      "loss": 1.0218,
      "step": 24540
    },
    {
      "epoch": 21.740478299379983,
      "grad_norm": 0.2885844111442566,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 24545
    },
    {
      "epoch": 21.74490699734278,
      "grad_norm": 0.2764827311038971,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 24550
    },
    {
      "epoch": 21.74933569530558,
      "grad_norm": 0.26501521468162537,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 24555
    },
    {
      "epoch": 21.75376439326838,
      "grad_norm": 0.25946244597435,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 24560
    },
    {
      "epoch": 21.758193091231178,
      "grad_norm": 0.2269723266363144,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 24565
    },
    {
      "epoch": 21.762621789193975,
      "grad_norm": 0.2374456822872162,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 24570
    },
    {
      "epoch": 21.767050487156776,
      "grad_norm": 0.2643706202507019,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 24575
    },
    {
      "epoch": 21.771479185119574,
      "grad_norm": 0.2324141561985016,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 24580
    },
    {
      "epoch": 21.775907883082375,
      "grad_norm": 0.23101358115673065,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 24585
    },
    {
      "epoch": 21.780336581045173,
      "grad_norm": 0.2550812065601349,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 24590
    },
    {
      "epoch": 21.78476527900797,
      "grad_norm": 0.3111308217048645,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 24595
    },
    {
      "epoch": 21.789193976970772,
      "grad_norm": 0.21285420656204224,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 24600
    },
    {
      "epoch": 21.79362267493357,
      "grad_norm": 0.22681789100170135,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 24605
    },
    {
      "epoch": 21.798051372896367,
      "grad_norm": 0.2354830652475357,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 24610
    },
    {
      "epoch": 21.80248007085917,
      "grad_norm": 0.23641620576381683,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 24615
    },
    {
      "epoch": 21.806908768821966,
      "grad_norm": 0.2156049758195877,
      "learning_rate": 1e-05,
      "loss": 1.035,
      "step": 24620
    },
    {
      "epoch": 21.811337466784764,
      "grad_norm": 0.27553847432136536,
      "learning_rate": 1e-05,
      "loss": 0.9116,
      "step": 24625
    },
    {
      "epoch": 21.815766164747565,
      "grad_norm": 0.23027409613132477,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 24630
    },
    {
      "epoch": 21.820194862710363,
      "grad_norm": 0.2639000713825226,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 24635
    },
    {
      "epoch": 21.824623560673164,
      "grad_norm": 0.2749571204185486,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 24640
    },
    {
      "epoch": 21.82905225863596,
      "grad_norm": 0.23388560116291046,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 24645
    },
    {
      "epoch": 21.83348095659876,
      "grad_norm": 0.266191691160202,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 24650
    },
    {
      "epoch": 21.83790965456156,
      "grad_norm": 0.23108968138694763,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 24655
    },
    {
      "epoch": 21.842338352524358,
      "grad_norm": 0.23208607733249664,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 24660
    },
    {
      "epoch": 21.846767050487156,
      "grad_norm": 0.2253040373325348,
      "learning_rate": 1e-05,
      "loss": 0.9312,
      "step": 24665
    },
    {
      "epoch": 21.851195748449957,
      "grad_norm": 0.21282123029232025,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 24670
    },
    {
      "epoch": 21.855624446412754,
      "grad_norm": 0.26727738976478577,
      "learning_rate": 1e-05,
      "loss": 0.9381,
      "step": 24675
    },
    {
      "epoch": 21.860053144375552,
      "grad_norm": 0.2842693030834198,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 24680
    },
    {
      "epoch": 21.864481842338353,
      "grad_norm": 0.19972069561481476,
      "learning_rate": 1e-05,
      "loss": 1.0242,
      "step": 24685
    },
    {
      "epoch": 21.86891054030115,
      "grad_norm": 0.24891071021556854,
      "learning_rate": 1e-05,
      "loss": 1.0439,
      "step": 24690
    },
    {
      "epoch": 21.873339238263952,
      "grad_norm": 0.23052257299423218,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 24695
    },
    {
      "epoch": 21.87776793622675,
      "grad_norm": 0.23111027479171753,
      "learning_rate": 1e-05,
      "loss": 1.0394,
      "step": 24700
    },
    {
      "epoch": 21.882196634189548,
      "grad_norm": 0.2436663806438446,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 24705
    },
    {
      "epoch": 21.88662533215235,
      "grad_norm": 0.27078261971473694,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 24710
    },
    {
      "epoch": 21.891054030115146,
      "grad_norm": 0.26738667488098145,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 24715
    },
    {
      "epoch": 21.895482728077944,
      "grad_norm": 0.2642519176006317,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 24720
    },
    {
      "epoch": 21.899911426040745,
      "grad_norm": 0.2486964464187622,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 24725
    },
    {
      "epoch": 21.904340124003543,
      "grad_norm": 0.2267775982618332,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 24730
    },
    {
      "epoch": 21.90876882196634,
      "grad_norm": 0.30306345224380493,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 24735
    },
    {
      "epoch": 21.913197519929142,
      "grad_norm": 0.2734224796295166,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 24740
    },
    {
      "epoch": 21.91762621789194,
      "grad_norm": 0.22469951212406158,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 24745
    },
    {
      "epoch": 21.922054915854737,
      "grad_norm": 0.24321064352989197,
      "learning_rate": 1e-05,
      "loss": 1.0267,
      "step": 24750
    },
    {
      "epoch": 21.92648361381754,
      "grad_norm": 0.23115235567092896,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 24755
    },
    {
      "epoch": 21.930912311780336,
      "grad_norm": 0.2903413474559784,
      "learning_rate": 1e-05,
      "loss": 1.0276,
      "step": 24760
    },
    {
      "epoch": 21.935341009743137,
      "grad_norm": 0.21366137266159058,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 24765
    },
    {
      "epoch": 21.939769707705935,
      "grad_norm": 0.23998533189296722,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 24770
    },
    {
      "epoch": 21.944198405668732,
      "grad_norm": 0.2669883072376251,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 24775
    },
    {
      "epoch": 21.948627103631534,
      "grad_norm": 0.2695714235305786,
      "learning_rate": 1e-05,
      "loss": 1.0447,
      "step": 24780
    },
    {
      "epoch": 21.95305580159433,
      "grad_norm": 0.24322670698165894,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 24785
    },
    {
      "epoch": 21.95748449955713,
      "grad_norm": 0.21124529838562012,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 24790
    },
    {
      "epoch": 21.96191319751993,
      "grad_norm": 0.21688516438007355,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 24795
    },
    {
      "epoch": 21.966341895482728,
      "grad_norm": 0.23600834608078003,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 24800
    },
    {
      "epoch": 21.970770593445526,
      "grad_norm": 0.28674063086509705,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 24805
    },
    {
      "epoch": 21.975199291408327,
      "grad_norm": 0.26195546984672546,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 24810
    },
    {
      "epoch": 21.979627989371124,
      "grad_norm": 0.25015777349472046,
      "learning_rate": 1e-05,
      "loss": 0.9087,
      "step": 24815
    },
    {
      "epoch": 21.984056687333926,
      "grad_norm": 0.25432541966438293,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 24820
    },
    {
      "epoch": 21.988485385296723,
      "grad_norm": 0.27454882860183716,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 24825
    },
    {
      "epoch": 21.99291408325952,
      "grad_norm": 0.28187116980552673,
      "learning_rate": 1e-05,
      "loss": 1.0441,
      "step": 24830
    },
    {
      "epoch": 21.997342781222322,
      "grad_norm": 0.29521968960762024,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 24835
    },
    {
      "epoch": 22.00177147918512,
      "grad_norm": 0.3057933449745178,
      "learning_rate": 1e-05,
      "loss": 1.0242,
      "step": 24840
    },
    {
      "epoch": 22.006200177147917,
      "grad_norm": 0.2525843679904938,
      "learning_rate": 1e-05,
      "loss": 1.0189,
      "step": 24845
    },
    {
      "epoch": 22.01062887511072,
      "grad_norm": 0.28382664918899536,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 24850
    },
    {
      "epoch": 22.015057573073516,
      "grad_norm": 0.23374539613723755,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 24855
    },
    {
      "epoch": 22.019486271036314,
      "grad_norm": 0.2440481185913086,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 24860
    },
    {
      "epoch": 22.023914968999115,
      "grad_norm": 0.24902604520320892,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 24865
    },
    {
      "epoch": 22.028343666961913,
      "grad_norm": 0.2957521378993988,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 24870
    },
    {
      "epoch": 22.03277236492471,
      "grad_norm": 0.22571849822998047,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 24875
    },
    {
      "epoch": 22.03720106288751,
      "grad_norm": 0.24343352019786835,
      "learning_rate": 1e-05,
      "loss": 1.0588,
      "step": 24880
    },
    {
      "epoch": 22.04162976085031,
      "grad_norm": 0.23574069142341614,
      "learning_rate": 1e-05,
      "loss": 1.0383,
      "step": 24885
    },
    {
      "epoch": 22.04605845881311,
      "grad_norm": 0.24992144107818604,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 24890
    },
    {
      "epoch": 22.050487156775908,
      "grad_norm": 0.28429991006851196,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 24895
    },
    {
      "epoch": 22.054915854738706,
      "grad_norm": 0.2270321100950241,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 24900
    },
    {
      "epoch": 22.059344552701507,
      "grad_norm": 0.22781485319137573,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 24905
    },
    {
      "epoch": 22.063773250664305,
      "grad_norm": 0.25927671790122986,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 24910
    },
    {
      "epoch": 22.068201948627102,
      "grad_norm": 0.22343440353870392,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 24915
    },
    {
      "epoch": 22.072630646589904,
      "grad_norm": 0.25268563628196716,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 24920
    },
    {
      "epoch": 22.0770593445527,
      "grad_norm": 0.22193363308906555,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 24925
    },
    {
      "epoch": 22.0814880425155,
      "grad_norm": 0.30172044038772583,
      "learning_rate": 1e-05,
      "loss": 1.0355,
      "step": 24930
    },
    {
      "epoch": 22.0859167404783,
      "grad_norm": 0.21960721909999847,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 24935
    },
    {
      "epoch": 22.090345438441098,
      "grad_norm": 0.23669998347759247,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 24940
    },
    {
      "epoch": 22.0947741364039,
      "grad_norm": 0.31388846039772034,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 24945
    },
    {
      "epoch": 22.099202834366697,
      "grad_norm": 0.30553311109542847,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 24950
    },
    {
      "epoch": 22.103631532329494,
      "grad_norm": 0.29542359709739685,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 24955
    },
    {
      "epoch": 22.108060230292296,
      "grad_norm": 0.21039451658725739,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 24960
    },
    {
      "epoch": 22.112488928255093,
      "grad_norm": 0.23354622721672058,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 24965
    },
    {
      "epoch": 22.11691762621789,
      "grad_norm": 0.257804811000824,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 24970
    },
    {
      "epoch": 22.121346324180692,
      "grad_norm": 0.2120206207036972,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 24975
    },
    {
      "epoch": 22.12577502214349,
      "grad_norm": 0.26745209097862244,
      "learning_rate": 1e-05,
      "loss": 0.9196,
      "step": 24980
    },
    {
      "epoch": 22.130203720106287,
      "grad_norm": 0.23815174400806427,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 24985
    },
    {
      "epoch": 22.13463241806909,
      "grad_norm": 0.2787434756755829,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 24990
    },
    {
      "epoch": 22.139061116031886,
      "grad_norm": 0.3281756639480591,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 24995
    },
    {
      "epoch": 22.143489813994684,
      "grad_norm": 0.24894240498542786,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 25000
    },
    {
      "epoch": 22.147918511957485,
      "grad_norm": 0.26528269052505493,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 25005
    },
    {
      "epoch": 22.152347209920283,
      "grad_norm": 0.243265300989151,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 25010
    },
    {
      "epoch": 22.156775907883084,
      "grad_norm": 0.2229057103395462,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 25015
    },
    {
      "epoch": 22.16120460584588,
      "grad_norm": 0.269471675157547,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 25020
    },
    {
      "epoch": 22.16563330380868,
      "grad_norm": 0.24262364208698273,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 25025
    },
    {
      "epoch": 22.17006200177148,
      "grad_norm": 0.21221621334552765,
      "learning_rate": 1e-05,
      "loss": 1.0349,
      "step": 25030
    },
    {
      "epoch": 22.174490699734278,
      "grad_norm": 0.24311906099319458,
      "learning_rate": 1e-05,
      "loss": 1.0526,
      "step": 25035
    },
    {
      "epoch": 22.178919397697076,
      "grad_norm": 0.28096672892570496,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 25040
    },
    {
      "epoch": 22.183348095659877,
      "grad_norm": 0.25694939494132996,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 25045
    },
    {
      "epoch": 22.187776793622675,
      "grad_norm": 0.27143892645835876,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 25050
    },
    {
      "epoch": 22.192205491585472,
      "grad_norm": 0.30177608132362366,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 25055
    },
    {
      "epoch": 22.196634189548273,
      "grad_norm": 0.25788095593452454,
      "learning_rate": 1e-05,
      "loss": 1.0314,
      "step": 25060
    },
    {
      "epoch": 22.20106288751107,
      "grad_norm": 0.2544099986553192,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 25065
    },
    {
      "epoch": 22.205491585473872,
      "grad_norm": 0.19766612350940704,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 25070
    },
    {
      "epoch": 22.20992028343667,
      "grad_norm": 0.2524663507938385,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 25075
    },
    {
      "epoch": 22.214348981399468,
      "grad_norm": 0.2551048696041107,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 25080
    },
    {
      "epoch": 22.21877767936227,
      "grad_norm": 0.28066185116767883,
      "learning_rate": 1e-05,
      "loss": 1.0417,
      "step": 25085
    },
    {
      "epoch": 22.223206377325067,
      "grad_norm": 0.2449391633272171,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 25090
    },
    {
      "epoch": 22.227635075287864,
      "grad_norm": 0.26418206095695496,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 25095
    },
    {
      "epoch": 22.232063773250665,
      "grad_norm": 0.31985703110694885,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 25100
    },
    {
      "epoch": 22.236492471213463,
      "grad_norm": 0.2525765895843506,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 25105
    },
    {
      "epoch": 22.24092116917626,
      "grad_norm": 0.23749807476997375,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 25110
    },
    {
      "epoch": 22.245349867139062,
      "grad_norm": 0.23759925365447998,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 25115
    },
    {
      "epoch": 22.24977856510186,
      "grad_norm": 0.24245911836624146,
      "learning_rate": 1e-05,
      "loss": 1.0402,
      "step": 25120
    },
    {
      "epoch": 22.254207263064657,
      "grad_norm": 0.2141829878091812,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 25125
    },
    {
      "epoch": 22.25863596102746,
      "grad_norm": 0.33661559224128723,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 25130
    },
    {
      "epoch": 22.263064658990256,
      "grad_norm": 0.2650635242462158,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 25135
    },
    {
      "epoch": 22.267493356953057,
      "grad_norm": 0.29339638352394104,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 25140
    },
    {
      "epoch": 22.271922054915855,
      "grad_norm": 0.24675299227237701,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 25145
    },
    {
      "epoch": 22.276350752878653,
      "grad_norm": 0.26497048139572144,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 25150
    },
    {
      "epoch": 22.280779450841454,
      "grad_norm": 0.29946672916412354,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 25155
    },
    {
      "epoch": 22.28520814880425,
      "grad_norm": 0.2353670448064804,
      "learning_rate": 1e-05,
      "loss": 1.0286,
      "step": 25160
    },
    {
      "epoch": 22.28963684676705,
      "grad_norm": 0.24496451020240784,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 25165
    },
    {
      "epoch": 22.29406554472985,
      "grad_norm": 0.22993889451026917,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 25170
    },
    {
      "epoch": 22.298494242692648,
      "grad_norm": 0.23675395548343658,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 25175
    },
    {
      "epoch": 22.302922940655446,
      "grad_norm": 0.23097476363182068,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 25180
    },
    {
      "epoch": 22.307351638618247,
      "grad_norm": 0.2836301326751709,
      "learning_rate": 1e-05,
      "loss": 1.0221,
      "step": 25185
    },
    {
      "epoch": 22.311780336581045,
      "grad_norm": 0.2484685629606247,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 25190
    },
    {
      "epoch": 22.316209034543846,
      "grad_norm": 0.25793904066085815,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 25195
    },
    {
      "epoch": 22.320637732506643,
      "grad_norm": 0.21760345995426178,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 25200
    },
    {
      "epoch": 22.32506643046944,
      "grad_norm": 0.1999516636133194,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 25205
    },
    {
      "epoch": 22.329495128432242,
      "grad_norm": 0.2596040666103363,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 25210
    },
    {
      "epoch": 22.33392382639504,
      "grad_norm": 0.24976402521133423,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 25215
    },
    {
      "epoch": 22.338352524357838,
      "grad_norm": 0.2822668254375458,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 25220
    },
    {
      "epoch": 22.34278122232064,
      "grad_norm": 0.22268618643283844,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 25225
    },
    {
      "epoch": 22.347209920283436,
      "grad_norm": 0.24922682344913483,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 25230
    },
    {
      "epoch": 22.351638618246234,
      "grad_norm": 0.2675280272960663,
      "learning_rate": 1e-05,
      "loss": 1.0375,
      "step": 25235
    },
    {
      "epoch": 22.356067316209035,
      "grad_norm": 0.3110574185848236,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 25240
    },
    {
      "epoch": 22.360496014171833,
      "grad_norm": 0.30851760506629944,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 25245
    },
    {
      "epoch": 22.36492471213463,
      "grad_norm": 0.27786239981651306,
      "learning_rate": 1e-05,
      "loss": 1.024,
      "step": 25250
    },
    {
      "epoch": 22.369353410097432,
      "grad_norm": 0.2358911782503128,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 25255
    },
    {
      "epoch": 22.37378210806023,
      "grad_norm": 0.23460493981838226,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 25260
    },
    {
      "epoch": 22.37821080602303,
      "grad_norm": 0.2363087385892868,
      "learning_rate": 1e-05,
      "loss": 1.0448,
      "step": 25265
    },
    {
      "epoch": 22.38263950398583,
      "grad_norm": 0.2724899351596832,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 25270
    },
    {
      "epoch": 22.387068201948626,
      "grad_norm": 0.2441389113664627,
      "learning_rate": 1e-05,
      "loss": 1.0316,
      "step": 25275
    },
    {
      "epoch": 22.391496899911427,
      "grad_norm": 0.2583214044570923,
      "learning_rate": 1e-05,
      "loss": 1.0316,
      "step": 25280
    },
    {
      "epoch": 22.395925597874225,
      "grad_norm": 0.2398071140050888,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 25285
    },
    {
      "epoch": 22.400354295837023,
      "grad_norm": 0.23686298727989197,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 25290
    },
    {
      "epoch": 22.404782993799824,
      "grad_norm": 0.25865161418914795,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 25295
    },
    {
      "epoch": 22.40921169176262,
      "grad_norm": 0.30793774127960205,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 25300
    },
    {
      "epoch": 22.41364038972542,
      "grad_norm": 0.2522560954093933,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 25305
    },
    {
      "epoch": 22.41806908768822,
      "grad_norm": 0.23933781683444977,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 25310
    },
    {
      "epoch": 22.422497785651018,
      "grad_norm": 0.23642078042030334,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 25315
    },
    {
      "epoch": 22.42692648361382,
      "grad_norm": 0.23104454576969147,
      "learning_rate": 1e-05,
      "loss": 1.014,
      "step": 25320
    },
    {
      "epoch": 22.431355181576617,
      "grad_norm": 0.20739449560642242,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 25325
    },
    {
      "epoch": 22.435783879539414,
      "grad_norm": 0.2163451761007309,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 25330
    },
    {
      "epoch": 22.440212577502216,
      "grad_norm": 0.25509095191955566,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 25335
    },
    {
      "epoch": 22.444641275465013,
      "grad_norm": 0.24274949729442596,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 25340
    },
    {
      "epoch": 22.44906997342781,
      "grad_norm": 0.2529536485671997,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 25345
    },
    {
      "epoch": 22.453498671390612,
      "grad_norm": 0.2632298767566681,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 25350
    },
    {
      "epoch": 22.45792736935341,
      "grad_norm": 0.24525117874145508,
      "learning_rate": 1e-05,
      "loss": 1.0269,
      "step": 25355
    },
    {
      "epoch": 22.462356067316207,
      "grad_norm": 0.2170303463935852,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 25360
    },
    {
      "epoch": 22.46678476527901,
      "grad_norm": 0.26130905747413635,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 25365
    },
    {
      "epoch": 22.471213463241806,
      "grad_norm": 0.23054029047489166,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 25370
    },
    {
      "epoch": 22.475642161204608,
      "grad_norm": 0.2737879157066345,
      "learning_rate": 1e-05,
      "loss": 1.0221,
      "step": 25375
    },
    {
      "epoch": 22.480070859167405,
      "grad_norm": 0.21854202449321747,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 25380
    },
    {
      "epoch": 22.484499557130203,
      "grad_norm": 0.2287951409816742,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 25385
    },
    {
      "epoch": 22.488928255093004,
      "grad_norm": 0.3347814679145813,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 25390
    },
    {
      "epoch": 22.4933569530558,
      "grad_norm": 0.22403250634670258,
      "learning_rate": 1e-05,
      "loss": 1.0374,
      "step": 25395
    },
    {
      "epoch": 22.4977856510186,
      "grad_norm": 0.21907730400562286,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 25400
    },
    {
      "epoch": 22.5022143489814,
      "grad_norm": 0.21052448451519012,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 25405
    },
    {
      "epoch": 22.5066430469442,
      "grad_norm": 0.24547868967056274,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 25410
    },
    {
      "epoch": 22.511071744906996,
      "grad_norm": 0.2388623058795929,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 25415
    },
    {
      "epoch": 22.515500442869797,
      "grad_norm": 0.2339182198047638,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 25420
    },
    {
      "epoch": 22.519929140832595,
      "grad_norm": 0.28296196460723877,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 25425
    },
    {
      "epoch": 22.524357838795392,
      "grad_norm": 0.2755812406539917,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 25430
    },
    {
      "epoch": 22.528786536758194,
      "grad_norm": 0.20562784373760223,
      "learning_rate": 1e-05,
      "loss": 1.0522,
      "step": 25435
    },
    {
      "epoch": 22.53321523472099,
      "grad_norm": 0.24974970519542694,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 25440
    },
    {
      "epoch": 22.537643932683793,
      "grad_norm": 0.2414158582687378,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 25445
    },
    {
      "epoch": 22.54207263064659,
      "grad_norm": 0.2662864923477173,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 25450
    },
    {
      "epoch": 22.546501328609388,
      "grad_norm": 0.21896205842494965,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 25455
    },
    {
      "epoch": 22.55093002657219,
      "grad_norm": 0.24726009368896484,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 25460
    },
    {
      "epoch": 22.555358724534987,
      "grad_norm": 0.24781477451324463,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 25465
    },
    {
      "epoch": 22.559787422497784,
      "grad_norm": 0.26446297764778137,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 25470
    },
    {
      "epoch": 22.564216120460586,
      "grad_norm": 0.23708654940128326,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 25475
    },
    {
      "epoch": 22.568644818423383,
      "grad_norm": 0.2327011376619339,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 25480
    },
    {
      "epoch": 22.57307351638618,
      "grad_norm": 0.24683986604213715,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 25485
    },
    {
      "epoch": 22.577502214348982,
      "grad_norm": 0.29164671897888184,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 25490
    },
    {
      "epoch": 22.58193091231178,
      "grad_norm": 0.2717437148094177,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 25495
    },
    {
      "epoch": 22.58635961027458,
      "grad_norm": 0.26747655868530273,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 25500
    },
    {
      "epoch": 22.59078830823738,
      "grad_norm": 0.3003387153148651,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 25505
    },
    {
      "epoch": 22.595217006200176,
      "grad_norm": 0.2480490356683731,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 25510
    },
    {
      "epoch": 22.599645704162977,
      "grad_norm": 0.2751440107822418,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 25515
    },
    {
      "epoch": 22.604074402125775,
      "grad_norm": 0.25806310772895813,
      "learning_rate": 1e-05,
      "loss": 0.9287,
      "step": 25520
    },
    {
      "epoch": 22.608503100088573,
      "grad_norm": 0.23098978400230408,
      "learning_rate": 1e-05,
      "loss": 1.0187,
      "step": 25525
    },
    {
      "epoch": 22.612931798051374,
      "grad_norm": 0.25656262040138245,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 25530
    },
    {
      "epoch": 22.61736049601417,
      "grad_norm": 0.26678669452667236,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 25535
    },
    {
      "epoch": 22.62178919397697,
      "grad_norm": 0.23283767700195312,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 25540
    },
    {
      "epoch": 22.62621789193977,
      "grad_norm": 0.3079436719417572,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 25545
    },
    {
      "epoch": 22.630646589902568,
      "grad_norm": 0.25933048129081726,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 25550
    },
    {
      "epoch": 22.63507528786537,
      "grad_norm": 0.27540329098701477,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 25555
    },
    {
      "epoch": 22.639503985828167,
      "grad_norm": 0.25537315011024475,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 25560
    },
    {
      "epoch": 22.643932683790965,
      "grad_norm": 0.28954339027404785,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 25565
    },
    {
      "epoch": 22.648361381753766,
      "grad_norm": 0.24918121099472046,
      "learning_rate": 1e-05,
      "loss": 1.0299,
      "step": 25570
    },
    {
      "epoch": 22.652790079716564,
      "grad_norm": 0.31572872400283813,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 25575
    },
    {
      "epoch": 22.65721877767936,
      "grad_norm": 0.23251767456531525,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 25580
    },
    {
      "epoch": 22.661647475642162,
      "grad_norm": 0.23751093447208405,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 25585
    },
    {
      "epoch": 22.66607617360496,
      "grad_norm": 0.24201689660549164,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 25590
    },
    {
      "epoch": 22.670504871567758,
      "grad_norm": 0.28515568375587463,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 25595
    },
    {
      "epoch": 22.67493356953056,
      "grad_norm": 0.2828501760959625,
      "learning_rate": 1e-05,
      "loss": 1.0325,
      "step": 25600
    },
    {
      "epoch": 22.679362267493357,
      "grad_norm": 0.2390662580728531,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 25605
    },
    {
      "epoch": 22.683790965456154,
      "grad_norm": 0.22238297760486603,
      "learning_rate": 1e-05,
      "loss": 0.9522,
      "step": 25610
    },
    {
      "epoch": 22.688219663418955,
      "grad_norm": 0.24652238190174103,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 25615
    },
    {
      "epoch": 22.692648361381753,
      "grad_norm": 0.20941878855228424,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 25620
    },
    {
      "epoch": 22.697077059344554,
      "grad_norm": 0.24829435348510742,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 25625
    },
    {
      "epoch": 22.701505757307352,
      "grad_norm": 0.2920697331428528,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 25630
    },
    {
      "epoch": 22.70593445527015,
      "grad_norm": 0.2333025187253952,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 25635
    },
    {
      "epoch": 22.71036315323295,
      "grad_norm": 0.27829641103744507,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 25640
    },
    {
      "epoch": 22.71479185119575,
      "grad_norm": 0.28581103682518005,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 25645
    },
    {
      "epoch": 22.719220549158546,
      "grad_norm": 0.21458880603313446,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 25650
    },
    {
      "epoch": 22.723649247121347,
      "grad_norm": 0.26263871788978577,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 25655
    },
    {
      "epoch": 22.728077945084145,
      "grad_norm": 0.2184339314699173,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 25660
    },
    {
      "epoch": 22.732506643046943,
      "grad_norm": 0.23649661242961884,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 25665
    },
    {
      "epoch": 22.736935341009744,
      "grad_norm": 0.22391146421432495,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 25670
    },
    {
      "epoch": 22.74136403897254,
      "grad_norm": 0.2300713062286377,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 25675
    },
    {
      "epoch": 22.745792736935343,
      "grad_norm": 0.25490954518318176,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 25680
    },
    {
      "epoch": 22.75022143489814,
      "grad_norm": 0.23086613416671753,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 25685
    },
    {
      "epoch": 22.754650132860938,
      "grad_norm": 0.2372780740261078,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 25690
    },
    {
      "epoch": 22.75907883082374,
      "grad_norm": 0.19394199550151825,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 25695
    },
    {
      "epoch": 22.763507528786537,
      "grad_norm": 0.18885891139507294,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 25700
    },
    {
      "epoch": 22.767936226749335,
      "grad_norm": 0.32187581062316895,
      "learning_rate": 1e-05,
      "loss": 1.0287,
      "step": 25705
    },
    {
      "epoch": 22.772364924712136,
      "grad_norm": 0.27318906784057617,
      "learning_rate": 1e-05,
      "loss": 1.0339,
      "step": 25710
    },
    {
      "epoch": 22.776793622674933,
      "grad_norm": 0.2578780949115753,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 25715
    },
    {
      "epoch": 22.78122232063773,
      "grad_norm": 0.25151413679122925,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 25720
    },
    {
      "epoch": 22.785651018600532,
      "grad_norm": 0.27815499901771545,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 25725
    },
    {
      "epoch": 22.79007971656333,
      "grad_norm": 0.22998100519180298,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 25730
    },
    {
      "epoch": 22.794508414526128,
      "grad_norm": 0.2738983631134033,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 25735
    },
    {
      "epoch": 22.79893711248893,
      "grad_norm": 0.24925392866134644,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 25740
    },
    {
      "epoch": 22.803365810451727,
      "grad_norm": 0.26162779331207275,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 25745
    },
    {
      "epoch": 22.807794508414528,
      "grad_norm": 0.24740613996982574,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 25750
    },
    {
      "epoch": 22.812223206377325,
      "grad_norm": 0.25766319036483765,
      "learning_rate": 1e-05,
      "loss": 1.0604,
      "step": 25755
    },
    {
      "epoch": 22.816651904340123,
      "grad_norm": 0.270840585231781,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 25760
    },
    {
      "epoch": 22.821080602302924,
      "grad_norm": 0.26112550497055054,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 25765
    },
    {
      "epoch": 22.825509300265722,
      "grad_norm": 0.2253311723470688,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 25770
    },
    {
      "epoch": 22.82993799822852,
      "grad_norm": 0.25054895877838135,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 25775
    },
    {
      "epoch": 22.83436669619132,
      "grad_norm": 0.2464187890291214,
      "learning_rate": 1e-05,
      "loss": 1.0316,
      "step": 25780
    },
    {
      "epoch": 22.83879539415412,
      "grad_norm": 0.2638521194458008,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 25785
    },
    {
      "epoch": 22.843224092116916,
      "grad_norm": 0.27301034331321716,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 25790
    },
    {
      "epoch": 22.847652790079717,
      "grad_norm": 0.22551095485687256,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 25795
    },
    {
      "epoch": 22.852081488042515,
      "grad_norm": 0.28099676966667175,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 25800
    },
    {
      "epoch": 22.856510186005316,
      "grad_norm": 0.2262679636478424,
      "learning_rate": 1e-05,
      "loss": 1.0458,
      "step": 25805
    },
    {
      "epoch": 22.860938883968114,
      "grad_norm": 0.2315719574689865,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 25810
    },
    {
      "epoch": 22.86536758193091,
      "grad_norm": 0.19639046490192413,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 25815
    },
    {
      "epoch": 22.869796279893713,
      "grad_norm": 0.2535834014415741,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 25820
    },
    {
      "epoch": 22.87422497785651,
      "grad_norm": 0.25161483883857727,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 25825
    },
    {
      "epoch": 22.878653675819308,
      "grad_norm": 0.2458488792181015,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 25830
    },
    {
      "epoch": 22.88308237378211,
      "grad_norm": 0.27616190910339355,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 25835
    },
    {
      "epoch": 22.887511071744907,
      "grad_norm": 0.22954954206943512,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 25840
    },
    {
      "epoch": 22.891939769707704,
      "grad_norm": 0.2157122641801834,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 25845
    },
    {
      "epoch": 22.896368467670506,
      "grad_norm": 0.3065982162952423,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 25850
    },
    {
      "epoch": 22.900797165633303,
      "grad_norm": 0.28051337599754333,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 25855
    },
    {
      "epoch": 22.9052258635961,
      "grad_norm": 0.3243139684200287,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 25860
    },
    {
      "epoch": 22.909654561558902,
      "grad_norm": 0.25547072291374207,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 25865
    },
    {
      "epoch": 22.9140832595217,
      "grad_norm": 0.28228655457496643,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 25870
    },
    {
      "epoch": 22.9185119574845,
      "grad_norm": 0.3086131811141968,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 25875
    },
    {
      "epoch": 22.9229406554473,
      "grad_norm": 0.25552958250045776,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 25880
    },
    {
      "epoch": 22.927369353410096,
      "grad_norm": 0.2505502998828888,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 25885
    },
    {
      "epoch": 22.931798051372898,
      "grad_norm": 0.24610909819602966,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 25890
    },
    {
      "epoch": 22.936226749335695,
      "grad_norm": 0.240942120552063,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 25895
    },
    {
      "epoch": 22.940655447298493,
      "grad_norm": 0.2842324674129486,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 25900
    },
    {
      "epoch": 22.945084145261294,
      "grad_norm": 0.252953439950943,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 25905
    },
    {
      "epoch": 22.949512843224092,
      "grad_norm": 0.23336157202720642,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 25910
    },
    {
      "epoch": 22.95394154118689,
      "grad_norm": 0.28841614723205566,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 25915
    },
    {
      "epoch": 22.95837023914969,
      "grad_norm": 0.2475043088197708,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 25920
    },
    {
      "epoch": 22.96279893711249,
      "grad_norm": 0.20819233357906342,
      "learning_rate": 1e-05,
      "loss": 1.0152,
      "step": 25925
    },
    {
      "epoch": 22.96722763507529,
      "grad_norm": 0.27295488119125366,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 25930
    },
    {
      "epoch": 22.971656333038087,
      "grad_norm": 0.2852492928504944,
      "learning_rate": 1e-05,
      "loss": 1.0578,
      "step": 25935
    },
    {
      "epoch": 22.976085031000885,
      "grad_norm": 0.23579590022563934,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 25940
    },
    {
      "epoch": 22.980513728963686,
      "grad_norm": 0.28827714920043945,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 25945
    },
    {
      "epoch": 22.984942426926484,
      "grad_norm": 0.29914212226867676,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 25950
    },
    {
      "epoch": 22.98937112488928,
      "grad_norm": 0.22872090339660645,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 25955
    },
    {
      "epoch": 22.993799822852083,
      "grad_norm": 0.22265173494815826,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 25960
    },
    {
      "epoch": 22.99822852081488,
      "grad_norm": 0.3121262192726135,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 25965
    },
    {
      "epoch": 23.002657218777678,
      "grad_norm": 0.26447778940200806,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 25970
    },
    {
      "epoch": 23.00708591674048,
      "grad_norm": 0.2370622158050537,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 25975
    },
    {
      "epoch": 23.011514614703277,
      "grad_norm": 0.2635786831378937,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 25980
    },
    {
      "epoch": 23.015943312666074,
      "grad_norm": 0.28557124733924866,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 25985
    },
    {
      "epoch": 23.020372010628876,
      "grad_norm": 0.237621009349823,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 25990
    },
    {
      "epoch": 23.024800708591673,
      "grad_norm": 0.22556810081005096,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 25995
    },
    {
      "epoch": 23.029229406554474,
      "grad_norm": 0.2727847099304199,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 26000
    },
    {
      "epoch": 23.033658104517272,
      "grad_norm": 0.22376036643981934,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 26005
    },
    {
      "epoch": 23.03808680248007,
      "grad_norm": 0.2665121853351593,
      "learning_rate": 1e-05,
      "loss": 1.0376,
      "step": 26010
    },
    {
      "epoch": 23.04251550044287,
      "grad_norm": 0.2564091980457306,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 26015
    },
    {
      "epoch": 23.04694419840567,
      "grad_norm": 0.21694253385066986,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 26020
    },
    {
      "epoch": 23.051372896368466,
      "grad_norm": 0.25690361857414246,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 26025
    },
    {
      "epoch": 23.055801594331268,
      "grad_norm": 0.2262437343597412,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 26030
    },
    {
      "epoch": 23.060230292294065,
      "grad_norm": 0.28153860569000244,
      "learning_rate": 1e-05,
      "loss": 1.0009,
      "step": 26035
    },
    {
      "epoch": 23.064658990256863,
      "grad_norm": 0.22980055212974548,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 26040
    },
    {
      "epoch": 23.069087688219664,
      "grad_norm": 0.22773821651935577,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 26045
    },
    {
      "epoch": 23.07351638618246,
      "grad_norm": 0.25532132387161255,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 26050
    },
    {
      "epoch": 23.077945084145263,
      "grad_norm": 0.2534859776496887,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 26055
    },
    {
      "epoch": 23.08237378210806,
      "grad_norm": 0.23883554339408875,
      "learning_rate": 1e-05,
      "loss": 1.0309,
      "step": 26060
    },
    {
      "epoch": 23.086802480070858,
      "grad_norm": 0.26781368255615234,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 26065
    },
    {
      "epoch": 23.09123117803366,
      "grad_norm": 0.26020872592926025,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 26070
    },
    {
      "epoch": 23.095659875996457,
      "grad_norm": 0.2362697422504425,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 26075
    },
    {
      "epoch": 23.100088573959255,
      "grad_norm": 0.21506093442440033,
      "learning_rate": 1e-05,
      "loss": 1.0409,
      "step": 26080
    },
    {
      "epoch": 23.104517271922056,
      "grad_norm": 0.2723117470741272,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 26085
    },
    {
      "epoch": 23.108945969884854,
      "grad_norm": 0.2622233033180237,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 26090
    },
    {
      "epoch": 23.11337466784765,
      "grad_norm": 0.21874207258224487,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 26095
    },
    {
      "epoch": 23.117803365810452,
      "grad_norm": 0.2519303262233734,
      "learning_rate": 1e-05,
      "loss": 1.0597,
      "step": 26100
    },
    {
      "epoch": 23.12223206377325,
      "grad_norm": 0.2639300227165222,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 26105
    },
    {
      "epoch": 23.12666076173605,
      "grad_norm": 0.21876296401023865,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 26110
    },
    {
      "epoch": 23.13108945969885,
      "grad_norm": 0.29018235206604004,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 26115
    },
    {
      "epoch": 23.135518157661647,
      "grad_norm": 0.27658510208129883,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 26120
    },
    {
      "epoch": 23.139946855624448,
      "grad_norm": 0.26128217577934265,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 26125
    },
    {
      "epoch": 23.144375553587246,
      "grad_norm": 0.21736447513103485,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 26130
    },
    {
      "epoch": 23.148804251550043,
      "grad_norm": 0.28176945447921753,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 26135
    },
    {
      "epoch": 23.153232949512844,
      "grad_norm": 0.2477501779794693,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 26140
    },
    {
      "epoch": 23.157661647475642,
      "grad_norm": 0.26699912548065186,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 26145
    },
    {
      "epoch": 23.16209034543844,
      "grad_norm": 0.2769632041454315,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 26150
    },
    {
      "epoch": 23.16651904340124,
      "grad_norm": 0.24970702826976776,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 26155
    },
    {
      "epoch": 23.17094774136404,
      "grad_norm": 0.19275449216365814,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 26160
    },
    {
      "epoch": 23.175376439326836,
      "grad_norm": 0.25007882714271545,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 26165
    },
    {
      "epoch": 23.179805137289637,
      "grad_norm": 0.25314250588417053,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 26170
    },
    {
      "epoch": 23.184233835252435,
      "grad_norm": 0.2311410754919052,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 26175
    },
    {
      "epoch": 23.188662533215236,
      "grad_norm": 0.2643571197986603,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 26180
    },
    {
      "epoch": 23.193091231178034,
      "grad_norm": 0.240512877702713,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 26185
    },
    {
      "epoch": 23.19751992914083,
      "grad_norm": 0.273680180311203,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 26190
    },
    {
      "epoch": 23.201948627103633,
      "grad_norm": 0.25345948338508606,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 26195
    },
    {
      "epoch": 23.20637732506643,
      "grad_norm": 0.27781733870506287,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 26200
    },
    {
      "epoch": 23.210806023029228,
      "grad_norm": 0.26070523262023926,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 26205
    },
    {
      "epoch": 23.21523472099203,
      "grad_norm": 0.23446990549564362,
      "learning_rate": 1e-05,
      "loss": 1.0359,
      "step": 26210
    },
    {
      "epoch": 23.219663418954827,
      "grad_norm": 0.23746508359909058,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 26215
    },
    {
      "epoch": 23.224092116917625,
      "grad_norm": 0.2397100180387497,
      "learning_rate": 1e-05,
      "loss": 1.0564,
      "step": 26220
    },
    {
      "epoch": 23.228520814880426,
      "grad_norm": 0.23432020843029022,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 26225
    },
    {
      "epoch": 23.232949512843224,
      "grad_norm": 0.26136159896850586,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 26230
    },
    {
      "epoch": 23.237378210806025,
      "grad_norm": 0.2443007081747055,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 26235
    },
    {
      "epoch": 23.241806908768822,
      "grad_norm": 0.2665751278400421,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 26240
    },
    {
      "epoch": 23.24623560673162,
      "grad_norm": 0.2399630844593048,
      "learning_rate": 1e-05,
      "loss": 0.9312,
      "step": 26245
    },
    {
      "epoch": 23.25066430469442,
      "grad_norm": 0.26037630438804626,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 26250
    },
    {
      "epoch": 23.25509300265722,
      "grad_norm": 0.2658340036869049,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 26255
    },
    {
      "epoch": 23.259521700620017,
      "grad_norm": 0.23740796744823456,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 26260
    },
    {
      "epoch": 23.263950398582818,
      "grad_norm": 0.28052619099617004,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 26265
    },
    {
      "epoch": 23.268379096545615,
      "grad_norm": 0.2324618250131607,
      "learning_rate": 1e-05,
      "loss": 1.0464,
      "step": 26270
    },
    {
      "epoch": 23.272807794508413,
      "grad_norm": 0.23283697664737701,
      "learning_rate": 1e-05,
      "loss": 0.904,
      "step": 26275
    },
    {
      "epoch": 23.277236492471214,
      "grad_norm": 0.25780996680259705,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 26280
    },
    {
      "epoch": 23.281665190434012,
      "grad_norm": 0.2611989378929138,
      "learning_rate": 1e-05,
      "loss": 1.0288,
      "step": 26285
    },
    {
      "epoch": 23.28609388839681,
      "grad_norm": 0.2390451729297638,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 26290
    },
    {
      "epoch": 23.29052258635961,
      "grad_norm": 0.28081047534942627,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 26295
    },
    {
      "epoch": 23.29495128432241,
      "grad_norm": 0.2402516007423401,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 26300
    },
    {
      "epoch": 23.29937998228521,
      "grad_norm": 0.30495649576187134,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 26305
    },
    {
      "epoch": 23.303808680248007,
      "grad_norm": 0.24615336954593658,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 26310
    },
    {
      "epoch": 23.308237378210805,
      "grad_norm": 0.23095014691352844,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 26315
    },
    {
      "epoch": 23.312666076173606,
      "grad_norm": 0.23776906728744507,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 26320
    },
    {
      "epoch": 23.317094774136404,
      "grad_norm": 0.24809081852436066,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 26325
    },
    {
      "epoch": 23.3215234720992,
      "grad_norm": 0.2409420609474182,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 26330
    },
    {
      "epoch": 23.325952170062003,
      "grad_norm": 0.2269292175769806,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 26335
    },
    {
      "epoch": 23.3303808680248,
      "grad_norm": 0.22666223347187042,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 26340
    },
    {
      "epoch": 23.334809565987598,
      "grad_norm": 0.25741687417030334,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 26345
    },
    {
      "epoch": 23.3392382639504,
      "grad_norm": 0.23735107481479645,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 26350
    },
    {
      "epoch": 23.343666961913197,
      "grad_norm": 0.24654647707939148,
      "learning_rate": 1e-05,
      "loss": 0.9333,
      "step": 26355
    },
    {
      "epoch": 23.348095659875998,
      "grad_norm": 0.2872961759567261,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 26360
    },
    {
      "epoch": 23.352524357838796,
      "grad_norm": 0.204074427485466,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 26365
    },
    {
      "epoch": 23.356953055801593,
      "grad_norm": 0.2723001539707184,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 26370
    },
    {
      "epoch": 23.361381753764395,
      "grad_norm": 0.22920246422290802,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 26375
    },
    {
      "epoch": 23.365810451727192,
      "grad_norm": 0.2627025842666626,
      "learning_rate": 1e-05,
      "loss": 1.055,
      "step": 26380
    },
    {
      "epoch": 23.37023914968999,
      "grad_norm": 0.222743421792984,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 26385
    },
    {
      "epoch": 23.37466784765279,
      "grad_norm": 0.24090582132339478,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 26390
    },
    {
      "epoch": 23.37909654561559,
      "grad_norm": 0.24897602200508118,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 26395
    },
    {
      "epoch": 23.383525243578386,
      "grad_norm": 0.20534689724445343,
      "learning_rate": 1e-05,
      "loss": 1.0292,
      "step": 26400
    },
    {
      "epoch": 23.387953941541188,
      "grad_norm": 0.2780756652355194,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 26405
    },
    {
      "epoch": 23.392382639503985,
      "grad_norm": 0.22269535064697266,
      "learning_rate": 1e-05,
      "loss": 0.9124,
      "step": 26410
    },
    {
      "epoch": 23.396811337466787,
      "grad_norm": 0.24354100227355957,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 26415
    },
    {
      "epoch": 23.401240035429584,
      "grad_norm": 0.2916207015514374,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 26420
    },
    {
      "epoch": 23.405668733392382,
      "grad_norm": 0.24010638892650604,
      "learning_rate": 1e-05,
      "loss": 0.8943,
      "step": 26425
    },
    {
      "epoch": 23.410097431355183,
      "grad_norm": 0.2418605238199234,
      "learning_rate": 1e-05,
      "loss": 0.9242,
      "step": 26430
    },
    {
      "epoch": 23.41452612931798,
      "grad_norm": 0.2563546299934387,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 26435
    },
    {
      "epoch": 23.41895482728078,
      "grad_norm": 0.2244047373533249,
      "learning_rate": 1e-05,
      "loss": 1.0009,
      "step": 26440
    },
    {
      "epoch": 23.42338352524358,
      "grad_norm": 0.27586227655410767,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 26445
    },
    {
      "epoch": 23.427812223206377,
      "grad_norm": 0.3002455234527588,
      "learning_rate": 1e-05,
      "loss": 1.0401,
      "step": 26450
    },
    {
      "epoch": 23.432240921169175,
      "grad_norm": 0.26960045099258423,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 26455
    },
    {
      "epoch": 23.436669619131976,
      "grad_norm": 0.22955931723117828,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 26460
    },
    {
      "epoch": 23.441098317094774,
      "grad_norm": 0.2313987910747528,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 26465
    },
    {
      "epoch": 23.44552701505757,
      "grad_norm": 0.26987579464912415,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 26470
    },
    {
      "epoch": 23.449955713020373,
      "grad_norm": 0.23977801203727722,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 26475
    },
    {
      "epoch": 23.45438441098317,
      "grad_norm": 0.26626765727996826,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 26480
    },
    {
      "epoch": 23.45881310894597,
      "grad_norm": 0.22593915462493896,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 26485
    },
    {
      "epoch": 23.46324180690877,
      "grad_norm": 0.28595519065856934,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 26490
    },
    {
      "epoch": 23.467670504871567,
      "grad_norm": 0.3095269203186035,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 26495
    },
    {
      "epoch": 23.472099202834368,
      "grad_norm": 0.3001960217952728,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 26500
    },
    {
      "epoch": 23.476527900797166,
      "grad_norm": 0.24239255487918854,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 26505
    },
    {
      "epoch": 23.480956598759963,
      "grad_norm": 0.2582003176212311,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 26510
    },
    {
      "epoch": 23.485385296722765,
      "grad_norm": 0.2729235291481018,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 26515
    },
    {
      "epoch": 23.489813994685562,
      "grad_norm": 0.3105107545852661,
      "learning_rate": 1e-05,
      "loss": 1.0541,
      "step": 26520
    },
    {
      "epoch": 23.49424269264836,
      "grad_norm": 0.29212746024131775,
      "learning_rate": 1e-05,
      "loss": 1.0159,
      "step": 26525
    },
    {
      "epoch": 23.49867139061116,
      "grad_norm": 0.21326103806495667,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 26530
    },
    {
      "epoch": 23.50310008857396,
      "grad_norm": 0.2569035589694977,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 26535
    },
    {
      "epoch": 23.50752878653676,
      "grad_norm": 0.23314517736434937,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 26540
    },
    {
      "epoch": 23.511957484499558,
      "grad_norm": 0.24351991713047028,
      "learning_rate": 1e-05,
      "loss": 1.0429,
      "step": 26545
    },
    {
      "epoch": 23.516386182462355,
      "grad_norm": 0.23723074793815613,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 26550
    },
    {
      "epoch": 23.520814880425156,
      "grad_norm": 0.2797541320323944,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 26555
    },
    {
      "epoch": 23.525243578387954,
      "grad_norm": 0.2910453677177429,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 26560
    },
    {
      "epoch": 23.52967227635075,
      "grad_norm": 0.23377171158790588,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 26565
    },
    {
      "epoch": 23.534100974313553,
      "grad_norm": 0.24713104963302612,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 26570
    },
    {
      "epoch": 23.53852967227635,
      "grad_norm": 0.24373966455459595,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 26575
    },
    {
      "epoch": 23.54295837023915,
      "grad_norm": 0.3163232207298279,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 26580
    },
    {
      "epoch": 23.54738706820195,
      "grad_norm": 0.2568933963775635,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 26585
    },
    {
      "epoch": 23.551815766164747,
      "grad_norm": 0.2469266951084137,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 26590
    },
    {
      "epoch": 23.556244464127545,
      "grad_norm": 0.2674744129180908,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 26595
    },
    {
      "epoch": 23.560673162090346,
      "grad_norm": 0.2934143543243408,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 26600
    },
    {
      "epoch": 23.565101860053144,
      "grad_norm": 0.22752077877521515,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 26605
    },
    {
      "epoch": 23.569530558015945,
      "grad_norm": 0.21492210030555725,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 26610
    },
    {
      "epoch": 23.573959255978743,
      "grad_norm": 0.23824001848697662,
      "learning_rate": 1e-05,
      "loss": 1.036,
      "step": 26615
    },
    {
      "epoch": 23.57838795394154,
      "grad_norm": 0.25493600964546204,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 26620
    },
    {
      "epoch": 23.58281665190434,
      "grad_norm": 0.2807728052139282,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 26625
    },
    {
      "epoch": 23.58724534986714,
      "grad_norm": 0.2720225155353546,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 26630
    },
    {
      "epoch": 23.591674047829937,
      "grad_norm": 0.3188835382461548,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 26635
    },
    {
      "epoch": 23.596102745792738,
      "grad_norm": 0.22400449216365814,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 26640
    },
    {
      "epoch": 23.600531443755536,
      "grad_norm": 0.30098628997802734,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 26645
    },
    {
      "epoch": 23.604960141718333,
      "grad_norm": 0.2980384826660156,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 26650
    },
    {
      "epoch": 23.609388839681134,
      "grad_norm": 0.2819817364215851,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 26655
    },
    {
      "epoch": 23.613817537643932,
      "grad_norm": 0.2706122398376465,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 26660
    },
    {
      "epoch": 23.618246235606733,
      "grad_norm": 0.22448629140853882,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 26665
    },
    {
      "epoch": 23.62267493356953,
      "grad_norm": 0.2219371348619461,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 26670
    },
    {
      "epoch": 23.62710363153233,
      "grad_norm": 0.2392638772726059,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 26675
    },
    {
      "epoch": 23.63153232949513,
      "grad_norm": 0.25862807035446167,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 26680
    },
    {
      "epoch": 23.635961027457927,
      "grad_norm": 0.25160324573516846,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 26685
    },
    {
      "epoch": 23.640389725420725,
      "grad_norm": 0.22142229974269867,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 26690
    },
    {
      "epoch": 23.644818423383526,
      "grad_norm": 0.24858084321022034,
      "learning_rate": 1e-05,
      "loss": 1.0403,
      "step": 26695
    },
    {
      "epoch": 23.649247121346324,
      "grad_norm": 0.2293468862771988,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 26700
    },
    {
      "epoch": 23.65367581930912,
      "grad_norm": 0.2500147521495819,
      "learning_rate": 1e-05,
      "loss": 1.0336,
      "step": 26705
    },
    {
      "epoch": 23.658104517271923,
      "grad_norm": 0.25640490651130676,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 26710
    },
    {
      "epoch": 23.66253321523472,
      "grad_norm": 0.25516122579574585,
      "learning_rate": 1e-05,
      "loss": 0.9103,
      "step": 26715
    },
    {
      "epoch": 23.666961913197518,
      "grad_norm": 0.23220662772655487,
      "learning_rate": 1e-05,
      "loss": 1.0207,
      "step": 26720
    },
    {
      "epoch": 23.67139061116032,
      "grad_norm": 0.2558310925960541,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 26725
    },
    {
      "epoch": 23.675819309123117,
      "grad_norm": 0.2426174432039261,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 26730
    },
    {
      "epoch": 23.68024800708592,
      "grad_norm": 0.20205755531787872,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 26735
    },
    {
      "epoch": 23.684676705048716,
      "grad_norm": 0.2598371207714081,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 26740
    },
    {
      "epoch": 23.689105403011514,
      "grad_norm": 0.2728954553604126,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 26745
    },
    {
      "epoch": 23.693534100974315,
      "grad_norm": 0.2757314145565033,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 26750
    },
    {
      "epoch": 23.697962798937112,
      "grad_norm": 0.2727743089199066,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 26755
    },
    {
      "epoch": 23.70239149689991,
      "grad_norm": 0.2958875000476837,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 26760
    },
    {
      "epoch": 23.70682019486271,
      "grad_norm": 0.25355657935142517,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 26765
    },
    {
      "epoch": 23.71124889282551,
      "grad_norm": 0.24307024478912354,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 26770
    },
    {
      "epoch": 23.715677590788307,
      "grad_norm": 0.23818360269069672,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 26775
    },
    {
      "epoch": 23.720106288751108,
      "grad_norm": 0.18384967744350433,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 26780
    },
    {
      "epoch": 23.724534986713905,
      "grad_norm": 0.253822386264801,
      "learning_rate": 1e-05,
      "loss": 1.039,
      "step": 26785
    },
    {
      "epoch": 23.728963684676707,
      "grad_norm": 0.26108798384666443,
      "learning_rate": 1e-05,
      "loss": 1.0323,
      "step": 26790
    },
    {
      "epoch": 23.733392382639504,
      "grad_norm": 0.2243223637342453,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 26795
    },
    {
      "epoch": 23.737821080602302,
      "grad_norm": 0.2187839150428772,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 26800
    },
    {
      "epoch": 23.742249778565103,
      "grad_norm": 0.28664642572402954,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 26805
    },
    {
      "epoch": 23.7466784765279,
      "grad_norm": 0.31537994742393494,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 26810
    },
    {
      "epoch": 23.7511071744907,
      "grad_norm": 0.28046277165412903,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 26815
    },
    {
      "epoch": 23.7555358724535,
      "grad_norm": 0.23828428983688354,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 26820
    },
    {
      "epoch": 23.759964570416297,
      "grad_norm": 0.2972710132598877,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 26825
    },
    {
      "epoch": 23.764393268379095,
      "grad_norm": 0.266639769077301,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 26830
    },
    {
      "epoch": 23.768821966341896,
      "grad_norm": 0.21086819469928741,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 26835
    },
    {
      "epoch": 23.773250664304694,
      "grad_norm": 0.303099125623703,
      "learning_rate": 1e-05,
      "loss": 1.0193,
      "step": 26840
    },
    {
      "epoch": 23.77767936226749,
      "grad_norm": 0.2552616596221924,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 26845
    },
    {
      "epoch": 23.782108060230293,
      "grad_norm": 0.21964603662490845,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 26850
    },
    {
      "epoch": 23.78653675819309,
      "grad_norm": 0.23729275166988373,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 26855
    },
    {
      "epoch": 23.79096545615589,
      "grad_norm": 0.26136288046836853,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 26860
    },
    {
      "epoch": 23.79539415411869,
      "grad_norm": 0.2399042397737503,
      "learning_rate": 1e-05,
      "loss": 1.0261,
      "step": 26865
    },
    {
      "epoch": 23.799822852081487,
      "grad_norm": 0.25391727685928345,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 26870
    },
    {
      "epoch": 23.804251550044288,
      "grad_norm": 0.280653178691864,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 26875
    },
    {
      "epoch": 23.808680248007086,
      "grad_norm": 0.27748042345046997,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 26880
    },
    {
      "epoch": 23.813108945969883,
      "grad_norm": 0.2171100676059723,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 26885
    },
    {
      "epoch": 23.817537643932685,
      "grad_norm": 0.26859402656555176,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 26890
    },
    {
      "epoch": 23.821966341895482,
      "grad_norm": 0.2783423066139221,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 26895
    },
    {
      "epoch": 23.82639503985828,
      "grad_norm": 0.3145160377025604,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 26900
    },
    {
      "epoch": 23.83082373782108,
      "grad_norm": 0.26615825295448303,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 26905
    },
    {
      "epoch": 23.83525243578388,
      "grad_norm": 0.26348012685775757,
      "learning_rate": 1e-05,
      "loss": 0.9234,
      "step": 26910
    },
    {
      "epoch": 23.83968113374668,
      "grad_norm": 0.25390365719795227,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 26915
    },
    {
      "epoch": 23.844109831709478,
      "grad_norm": 0.22636613249778748,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 26920
    },
    {
      "epoch": 23.848538529672275,
      "grad_norm": 0.24785874783992767,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 26925
    },
    {
      "epoch": 23.852967227635077,
      "grad_norm": 0.23465396463871002,
      "learning_rate": 1e-05,
      "loss": 1.0359,
      "step": 26930
    },
    {
      "epoch": 23.857395925597874,
      "grad_norm": 0.2235262095928192,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 26935
    },
    {
      "epoch": 23.861824623560672,
      "grad_norm": 0.24740758538246155,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 26940
    },
    {
      "epoch": 23.866253321523473,
      "grad_norm": 0.26221954822540283,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 26945
    },
    {
      "epoch": 23.87068201948627,
      "grad_norm": 0.23343157768249512,
      "learning_rate": 1e-05,
      "loss": 0.9031,
      "step": 26950
    },
    {
      "epoch": 23.87511071744907,
      "grad_norm": 0.24470555782318115,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 26955
    },
    {
      "epoch": 23.87953941541187,
      "grad_norm": 0.24176916480064392,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 26960
    },
    {
      "epoch": 23.883968113374667,
      "grad_norm": 0.2797878086566925,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 26965
    },
    {
      "epoch": 23.888396811337465,
      "grad_norm": 0.2574193775653839,
      "learning_rate": 1e-05,
      "loss": 1.0255,
      "step": 26970
    },
    {
      "epoch": 23.892825509300266,
      "grad_norm": 0.2504587769508362,
      "learning_rate": 1e-05,
      "loss": 0.9221,
      "step": 26975
    },
    {
      "epoch": 23.897254207263064,
      "grad_norm": 0.20924057066440582,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 26980
    },
    {
      "epoch": 23.901682905225865,
      "grad_norm": 0.2794031500816345,
      "learning_rate": 1e-05,
      "loss": 1.048,
      "step": 26985
    },
    {
      "epoch": 23.906111603188663,
      "grad_norm": 0.28351205587387085,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 26990
    },
    {
      "epoch": 23.91054030115146,
      "grad_norm": 0.26499977707862854,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 26995
    },
    {
      "epoch": 23.91496899911426,
      "grad_norm": 0.2845896780490875,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 27000
    },
    {
      "epoch": 23.91939769707706,
      "grad_norm": 0.23948568105697632,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 27005
    },
    {
      "epoch": 23.923826395039857,
      "grad_norm": 0.22813214361667633,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 27010
    },
    {
      "epoch": 23.928255093002658,
      "grad_norm": 0.2147350162267685,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 27015
    },
    {
      "epoch": 23.932683790965456,
      "grad_norm": 0.2525482177734375,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 27020
    },
    {
      "epoch": 23.937112488928253,
      "grad_norm": 0.226675882935524,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 27025
    },
    {
      "epoch": 23.941541186891055,
      "grad_norm": 0.25734806060791016,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 27030
    },
    {
      "epoch": 23.945969884853852,
      "grad_norm": 0.27355778217315674,
      "learning_rate": 1e-05,
      "loss": 1.0494,
      "step": 27035
    },
    {
      "epoch": 23.950398582816653,
      "grad_norm": 0.24465897679328918,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 27040
    },
    {
      "epoch": 23.95482728077945,
      "grad_norm": 0.24087879061698914,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 27045
    },
    {
      "epoch": 23.95925597874225,
      "grad_norm": 0.23881228268146515,
      "learning_rate": 1e-05,
      "loss": 1.0382,
      "step": 27050
    },
    {
      "epoch": 23.96368467670505,
      "grad_norm": 0.2686094343662262,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 27055
    },
    {
      "epoch": 23.968113374667848,
      "grad_norm": 0.28021782636642456,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 27060
    },
    {
      "epoch": 23.972542072630645,
      "grad_norm": 0.29325753450393677,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 27065
    },
    {
      "epoch": 23.976970770593447,
      "grad_norm": 0.2697674632072449,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 27070
    },
    {
      "epoch": 23.981399468556244,
      "grad_norm": 0.30482274293899536,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 27075
    },
    {
      "epoch": 23.985828166519042,
      "grad_norm": 0.26910868287086487,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 27080
    },
    {
      "epoch": 23.990256864481843,
      "grad_norm": 0.3090648055076599,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 27085
    },
    {
      "epoch": 23.99468556244464,
      "grad_norm": 0.2467624545097351,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 27090
    },
    {
      "epoch": 23.999114260407442,
      "grad_norm": 0.2364131510257721,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 27095
    },
    {
      "epoch": 24.00354295837024,
      "grad_norm": 0.2566116750240326,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 27100
    },
    {
      "epoch": 24.007971656333037,
      "grad_norm": 0.22930893301963806,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 27105
    },
    {
      "epoch": 24.01240035429584,
      "grad_norm": 0.2578718364238739,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 27110
    },
    {
      "epoch": 24.016829052258636,
      "grad_norm": 0.26853352785110474,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 27115
    },
    {
      "epoch": 24.021257750221434,
      "grad_norm": 0.20939262211322784,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 27120
    },
    {
      "epoch": 24.025686448184235,
      "grad_norm": 0.2888830304145813,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 27125
    },
    {
      "epoch": 24.030115146147033,
      "grad_norm": 0.29009345173835754,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 27130
    },
    {
      "epoch": 24.03454384410983,
      "grad_norm": 0.274711936712265,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 27135
    },
    {
      "epoch": 24.03897254207263,
      "grad_norm": 0.26842740178108215,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 27140
    },
    {
      "epoch": 24.04340124003543,
      "grad_norm": 0.279020756483078,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 27145
    },
    {
      "epoch": 24.047829937998227,
      "grad_norm": 0.2553081214427948,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 27150
    },
    {
      "epoch": 24.052258635961028,
      "grad_norm": 0.30284035205841064,
      "learning_rate": 1e-05,
      "loss": 1.0402,
      "step": 27155
    },
    {
      "epoch": 24.056687333923826,
      "grad_norm": 0.2558462917804718,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 27160
    },
    {
      "epoch": 24.061116031886627,
      "grad_norm": 0.2524622678756714,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 27165
    },
    {
      "epoch": 24.065544729849424,
      "grad_norm": 0.24786490201950073,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 27170
    },
    {
      "epoch": 24.069973427812222,
      "grad_norm": 0.3078889548778534,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 27175
    },
    {
      "epoch": 24.074402125775023,
      "grad_norm": 0.24278715252876282,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 27180
    },
    {
      "epoch": 24.07883082373782,
      "grad_norm": 0.22713172435760498,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 27185
    },
    {
      "epoch": 24.08325952170062,
      "grad_norm": 0.2256680727005005,
      "learning_rate": 1e-05,
      "loss": 1.0314,
      "step": 27190
    },
    {
      "epoch": 24.08768821966342,
      "grad_norm": 0.24600300192832947,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 27195
    },
    {
      "epoch": 24.092116917626218,
      "grad_norm": 0.26029497385025024,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 27200
    },
    {
      "epoch": 24.096545615589015,
      "grad_norm": 0.28288689255714417,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 27205
    },
    {
      "epoch": 24.100974313551816,
      "grad_norm": 0.26191774010658264,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 27210
    },
    {
      "epoch": 24.105403011514614,
      "grad_norm": 0.23438803851604462,
      "learning_rate": 1e-05,
      "loss": 0.8922,
      "step": 27215
    },
    {
      "epoch": 24.109831709477415,
      "grad_norm": 0.23090486228466034,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 27220
    },
    {
      "epoch": 24.114260407440213,
      "grad_norm": 0.26005062460899353,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 27225
    },
    {
      "epoch": 24.11868910540301,
      "grad_norm": 0.2271093875169754,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 27230
    },
    {
      "epoch": 24.123117803365812,
      "grad_norm": 0.2458283007144928,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 27235
    },
    {
      "epoch": 24.12754650132861,
      "grad_norm": 0.2611621022224426,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 27240
    },
    {
      "epoch": 24.131975199291407,
      "grad_norm": 0.22730910778045654,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 27245
    },
    {
      "epoch": 24.13640389725421,
      "grad_norm": 0.23347064852714539,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 27250
    },
    {
      "epoch": 24.140832595217006,
      "grad_norm": 0.2405199259519577,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 27255
    },
    {
      "epoch": 24.145261293179804,
      "grad_norm": 0.29748761653900146,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 27260
    },
    {
      "epoch": 24.149689991142605,
      "grad_norm": 0.29782795906066895,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 27265
    },
    {
      "epoch": 24.154118689105402,
      "grad_norm": 0.2286754995584488,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 27270
    },
    {
      "epoch": 24.158547387068204,
      "grad_norm": 0.26479247212409973,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 27275
    },
    {
      "epoch": 24.162976085031,
      "grad_norm": 0.2673901319503784,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 27280
    },
    {
      "epoch": 24.1674047829938,
      "grad_norm": 0.267720103263855,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 27285
    },
    {
      "epoch": 24.1718334809566,
      "grad_norm": 0.23425528407096863,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 27290
    },
    {
      "epoch": 24.176262178919398,
      "grad_norm": 0.27703917026519775,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 27295
    },
    {
      "epoch": 24.180690876882196,
      "grad_norm": 0.25991371273994446,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 27300
    },
    {
      "epoch": 24.185119574844997,
      "grad_norm": 0.27498453855514526,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 27305
    },
    {
      "epoch": 24.189548272807794,
      "grad_norm": 0.25150948762893677,
      "learning_rate": 1e-05,
      "loss": 1.023,
      "step": 27310
    },
    {
      "epoch": 24.193976970770592,
      "grad_norm": 0.22732007503509521,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 27315
    },
    {
      "epoch": 24.198405668733393,
      "grad_norm": 0.21285602450370789,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 27320
    },
    {
      "epoch": 24.20283436669619,
      "grad_norm": 0.2449236959218979,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 27325
    },
    {
      "epoch": 24.20726306465899,
      "grad_norm": 0.24172784388065338,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 27330
    },
    {
      "epoch": 24.21169176262179,
      "grad_norm": 0.3550632894039154,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 27335
    },
    {
      "epoch": 24.216120460584587,
      "grad_norm": 0.30615052580833435,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 27340
    },
    {
      "epoch": 24.22054915854739,
      "grad_norm": 0.2539338767528534,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 27345
    },
    {
      "epoch": 24.224977856510186,
      "grad_norm": 0.28723493218421936,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 27350
    },
    {
      "epoch": 24.229406554472984,
      "grad_norm": 0.2576487958431244,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 27355
    },
    {
      "epoch": 24.233835252435785,
      "grad_norm": 0.2898792624473572,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 27360
    },
    {
      "epoch": 24.238263950398583,
      "grad_norm": 0.2693891227245331,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 27365
    },
    {
      "epoch": 24.24269264836138,
      "grad_norm": 0.24380403757095337,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 27370
    },
    {
      "epoch": 24.24712134632418,
      "grad_norm": 0.26417067646980286,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 27375
    },
    {
      "epoch": 24.25155004428698,
      "grad_norm": 0.2408972829580307,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 27380
    },
    {
      "epoch": 24.255978742249777,
      "grad_norm": 0.24388913810253143,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 27385
    },
    {
      "epoch": 24.260407440212578,
      "grad_norm": 0.26220259070396423,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 27390
    },
    {
      "epoch": 24.264836138175376,
      "grad_norm": 0.2537146806716919,
      "learning_rate": 1e-05,
      "loss": 1.017,
      "step": 27395
    },
    {
      "epoch": 24.269264836138177,
      "grad_norm": 0.2604083716869354,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 27400
    },
    {
      "epoch": 24.273693534100975,
      "grad_norm": 0.21949341893196106,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 27405
    },
    {
      "epoch": 24.278122232063772,
      "grad_norm": 0.22510309517383575,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 27410
    },
    {
      "epoch": 24.282550930026574,
      "grad_norm": 0.2591549754142761,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 27415
    },
    {
      "epoch": 24.28697962798937,
      "grad_norm": 0.2637147605419159,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 27420
    },
    {
      "epoch": 24.29140832595217,
      "grad_norm": 0.22815559804439545,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 27425
    },
    {
      "epoch": 24.29583702391497,
      "grad_norm": 0.23618504405021667,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 27430
    },
    {
      "epoch": 24.300265721877768,
      "grad_norm": 0.25723832845687866,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 27435
    },
    {
      "epoch": 24.304694419840565,
      "grad_norm": 0.2191542685031891,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 27440
    },
    {
      "epoch": 24.309123117803367,
      "grad_norm": 0.2241484522819519,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 27445
    },
    {
      "epoch": 24.313551815766164,
      "grad_norm": 0.2691168487071991,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 27450
    },
    {
      "epoch": 24.317980513728962,
      "grad_norm": 0.21560446918010712,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 27455
    },
    {
      "epoch": 24.322409211691763,
      "grad_norm": 0.23158149421215057,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 27460
    },
    {
      "epoch": 24.32683790965456,
      "grad_norm": 0.24868138134479523,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 27465
    },
    {
      "epoch": 24.331266607617362,
      "grad_norm": 0.219271719455719,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 27470
    },
    {
      "epoch": 24.33569530558016,
      "grad_norm": 0.25831377506256104,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 27475
    },
    {
      "epoch": 24.340124003542957,
      "grad_norm": 0.24598956108093262,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 27480
    },
    {
      "epoch": 24.34455270150576,
      "grad_norm": 0.25745877623558044,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 27485
    },
    {
      "epoch": 24.348981399468556,
      "grad_norm": 0.23913982510566711,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 27490
    },
    {
      "epoch": 24.353410097431354,
      "grad_norm": 0.28351959586143494,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 27495
    },
    {
      "epoch": 24.357838795394155,
      "grad_norm": 0.24848684668540955,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 27500
    },
    {
      "epoch": 24.362267493356953,
      "grad_norm": 0.20931607484817505,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 27505
    },
    {
      "epoch": 24.36669619131975,
      "grad_norm": 0.23935669660568237,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 27510
    },
    {
      "epoch": 24.37112488928255,
      "grad_norm": 0.2865438163280487,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 27515
    },
    {
      "epoch": 24.37555358724535,
      "grad_norm": 0.25080606341362,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 27520
    },
    {
      "epoch": 24.37998228520815,
      "grad_norm": 0.22788484394550323,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 27525
    },
    {
      "epoch": 24.384410983170948,
      "grad_norm": 0.2690378725528717,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 27530
    },
    {
      "epoch": 24.388839681133746,
      "grad_norm": 0.2662249803543091,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 27535
    },
    {
      "epoch": 24.393268379096547,
      "grad_norm": 0.25115519762039185,
      "learning_rate": 1e-05,
      "loss": 0.9295,
      "step": 27540
    },
    {
      "epoch": 24.397697077059345,
      "grad_norm": 0.27152177691459656,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 27545
    },
    {
      "epoch": 24.402125775022142,
      "grad_norm": 0.2977381646633148,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 27550
    },
    {
      "epoch": 24.406554472984944,
      "grad_norm": 0.2588966488838196,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 27555
    },
    {
      "epoch": 24.41098317094774,
      "grad_norm": 0.22470641136169434,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 27560
    },
    {
      "epoch": 24.41541186891054,
      "grad_norm": 0.27989575266838074,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 27565
    },
    {
      "epoch": 24.41984056687334,
      "grad_norm": 0.2517945468425751,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 27570
    },
    {
      "epoch": 24.424269264836138,
      "grad_norm": 0.23654067516326904,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 27575
    },
    {
      "epoch": 24.428697962798935,
      "grad_norm": 0.23164743185043335,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 27580
    },
    {
      "epoch": 24.433126660761737,
      "grad_norm": 0.24988412857055664,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 27585
    },
    {
      "epoch": 24.437555358724534,
      "grad_norm": 0.27342689037323,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 27590
    },
    {
      "epoch": 24.441984056687335,
      "grad_norm": 0.22409027814865112,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 27595
    },
    {
      "epoch": 24.446412754650133,
      "grad_norm": 0.21091631054878235,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 27600
    },
    {
      "epoch": 24.45084145261293,
      "grad_norm": 0.20956432819366455,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 27605
    },
    {
      "epoch": 24.455270150575732,
      "grad_norm": 0.257333368062973,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 27610
    },
    {
      "epoch": 24.45969884853853,
      "grad_norm": 0.2603904604911804,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 27615
    },
    {
      "epoch": 24.464127546501327,
      "grad_norm": 0.24283568561077118,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 27620
    },
    {
      "epoch": 24.46855624446413,
      "grad_norm": 0.26459357142448425,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 27625
    },
    {
      "epoch": 24.472984942426926,
      "grad_norm": 0.23536963760852814,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 27630
    },
    {
      "epoch": 24.477413640389724,
      "grad_norm": 0.26077204942703247,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 27635
    },
    {
      "epoch": 24.481842338352525,
      "grad_norm": 0.19926729798316956,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 27640
    },
    {
      "epoch": 24.486271036315323,
      "grad_norm": 0.24146565794944763,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 27645
    },
    {
      "epoch": 24.490699734278124,
      "grad_norm": 0.24525809288024902,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 27650
    },
    {
      "epoch": 24.49512843224092,
      "grad_norm": 0.275239497423172,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 27655
    },
    {
      "epoch": 24.49955713020372,
      "grad_norm": 0.39674660563468933,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 27660
    },
    {
      "epoch": 24.50398582816652,
      "grad_norm": 0.25674673914909363,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 27665
    },
    {
      "epoch": 24.508414526129318,
      "grad_norm": 0.27644819021224976,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 27670
    },
    {
      "epoch": 24.512843224092116,
      "grad_norm": 0.22909076511859894,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 27675
    },
    {
      "epoch": 24.517271922054917,
      "grad_norm": 0.2632433772087097,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 27680
    },
    {
      "epoch": 24.521700620017715,
      "grad_norm": 0.24753408133983612,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 27685
    },
    {
      "epoch": 24.526129317980512,
      "grad_norm": 0.21823808550834656,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 27690
    },
    {
      "epoch": 24.530558015943313,
      "grad_norm": 0.25986164808273315,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 27695
    },
    {
      "epoch": 24.53498671390611,
      "grad_norm": 0.2600781321525574,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 27700
    },
    {
      "epoch": 24.53941541186891,
      "grad_norm": 0.2880846858024597,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 27705
    },
    {
      "epoch": 24.54384410983171,
      "grad_norm": 0.27464109659194946,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 27710
    },
    {
      "epoch": 24.548272807794508,
      "grad_norm": 0.26930370926856995,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 27715
    },
    {
      "epoch": 24.55270150575731,
      "grad_norm": 0.23947246372699738,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 27720
    },
    {
      "epoch": 24.557130203720106,
      "grad_norm": 0.23891033232212067,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 27725
    },
    {
      "epoch": 24.561558901682904,
      "grad_norm": 0.21519169211387634,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 27730
    },
    {
      "epoch": 24.565987599645705,
      "grad_norm": 0.22921505570411682,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 27735
    },
    {
      "epoch": 24.570416297608503,
      "grad_norm": 0.2791348099708557,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 27740
    },
    {
      "epoch": 24.5748449955713,
      "grad_norm": 0.293082594871521,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 27745
    },
    {
      "epoch": 24.579273693534102,
      "grad_norm": 0.2287125140428543,
      "learning_rate": 1e-05,
      "loss": 1.0203,
      "step": 27750
    },
    {
      "epoch": 24.5837023914969,
      "grad_norm": 0.23439842462539673,
      "learning_rate": 1e-05,
      "loss": 0.9283,
      "step": 27755
    },
    {
      "epoch": 24.588131089459697,
      "grad_norm": 0.2438921481370926,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 27760
    },
    {
      "epoch": 24.5925597874225,
      "grad_norm": 0.27476295828819275,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 27765
    },
    {
      "epoch": 24.596988485385296,
      "grad_norm": 0.23507197201251984,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 27770
    },
    {
      "epoch": 24.601417183348097,
      "grad_norm": 0.28731539845466614,
      "learning_rate": 1e-05,
      "loss": 1.033,
      "step": 27775
    },
    {
      "epoch": 24.605845881310895,
      "grad_norm": 0.24496515095233917,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 27780
    },
    {
      "epoch": 24.610274579273693,
      "grad_norm": 0.2469015121459961,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 27785
    },
    {
      "epoch": 24.614703277236494,
      "grad_norm": 0.23915302753448486,
      "learning_rate": 1e-05,
      "loss": 0.917,
      "step": 27790
    },
    {
      "epoch": 24.61913197519929,
      "grad_norm": 0.22897915542125702,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 27795
    },
    {
      "epoch": 24.62356067316209,
      "grad_norm": 0.2170686423778534,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 27800
    },
    {
      "epoch": 24.62798937112489,
      "grad_norm": 0.26316043734550476,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 27805
    },
    {
      "epoch": 24.632418069087688,
      "grad_norm": 0.25438329577445984,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 27810
    },
    {
      "epoch": 24.636846767050486,
      "grad_norm": 0.25038447976112366,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 27815
    },
    {
      "epoch": 24.641275465013287,
      "grad_norm": 0.20531101524829865,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 27820
    },
    {
      "epoch": 24.645704162976084,
      "grad_norm": 0.24842555820941925,
      "learning_rate": 1e-05,
      "loss": 1.0317,
      "step": 27825
    },
    {
      "epoch": 24.650132860938886,
      "grad_norm": 0.25941750407218933,
      "learning_rate": 1e-05,
      "loss": 1.042,
      "step": 27830
    },
    {
      "epoch": 24.654561558901683,
      "grad_norm": 0.24607211351394653,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 27835
    },
    {
      "epoch": 24.65899025686448,
      "grad_norm": 0.22874706983566284,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 27840
    },
    {
      "epoch": 24.663418954827282,
      "grad_norm": 0.2419596016407013,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 27845
    },
    {
      "epoch": 24.66784765279008,
      "grad_norm": 0.273303359746933,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 27850
    },
    {
      "epoch": 24.672276350752878,
      "grad_norm": 0.2364151030778885,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 27855
    },
    {
      "epoch": 24.67670504871568,
      "grad_norm": 0.2614619731903076,
      "learning_rate": 1e-05,
      "loss": 1.0484,
      "step": 27860
    },
    {
      "epoch": 24.681133746678476,
      "grad_norm": 0.2701669931411743,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 27865
    },
    {
      "epoch": 24.685562444641274,
      "grad_norm": 0.2146516740322113,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 27870
    },
    {
      "epoch": 24.689991142604075,
      "grad_norm": 0.23337946832180023,
      "learning_rate": 1e-05,
      "loss": 1.05,
      "step": 27875
    },
    {
      "epoch": 24.694419840566873,
      "grad_norm": 0.20593266189098358,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 27880
    },
    {
      "epoch": 24.698848538529674,
      "grad_norm": 0.2650032639503479,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 27885
    },
    {
      "epoch": 24.70327723649247,
      "grad_norm": 0.26032283902168274,
      "learning_rate": 1e-05,
      "loss": 1.0688,
      "step": 27890
    },
    {
      "epoch": 24.70770593445527,
      "grad_norm": 0.24456565082073212,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 27895
    },
    {
      "epoch": 24.71213463241807,
      "grad_norm": 0.23210467398166656,
      "learning_rate": 1e-05,
      "loss": 0.9242,
      "step": 27900
    },
    {
      "epoch": 24.71656333038087,
      "grad_norm": 0.2659805715084076,
      "learning_rate": 1e-05,
      "loss": 1.0241,
      "step": 27905
    },
    {
      "epoch": 24.720992028343666,
      "grad_norm": 0.2703510522842407,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 27910
    },
    {
      "epoch": 24.725420726306467,
      "grad_norm": 0.2797582447528839,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 27915
    },
    {
      "epoch": 24.729849424269265,
      "grad_norm": 0.22575101256370544,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 27920
    },
    {
      "epoch": 24.734278122232062,
      "grad_norm": 0.22218960523605347,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 27925
    },
    {
      "epoch": 24.738706820194864,
      "grad_norm": 0.2900409400463104,
      "learning_rate": 1e-05,
      "loss": 1.0614,
      "step": 27930
    },
    {
      "epoch": 24.74313551815766,
      "grad_norm": 0.28105300664901733,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 27935
    },
    {
      "epoch": 24.74756421612046,
      "grad_norm": 0.22506463527679443,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 27940
    },
    {
      "epoch": 24.75199291408326,
      "grad_norm": 0.2369133085012436,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 27945
    },
    {
      "epoch": 24.756421612046058,
      "grad_norm": 0.24252861738204956,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 27950
    },
    {
      "epoch": 24.76085031000886,
      "grad_norm": 0.2130277007818222,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 27955
    },
    {
      "epoch": 24.765279007971657,
      "grad_norm": 0.2696848511695862,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 27960
    },
    {
      "epoch": 24.769707705934454,
      "grad_norm": 0.28770482540130615,
      "learning_rate": 1e-05,
      "loss": 1.0187,
      "step": 27965
    },
    {
      "epoch": 24.774136403897256,
      "grad_norm": 0.2645743489265442,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 27970
    },
    {
      "epoch": 24.778565101860053,
      "grad_norm": 0.31868138909339905,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 27975
    },
    {
      "epoch": 24.78299379982285,
      "grad_norm": 0.2536707818508148,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 27980
    },
    {
      "epoch": 24.787422497785652,
      "grad_norm": 0.2427499145269394,
      "learning_rate": 1e-05,
      "loss": 0.9003,
      "step": 27985
    },
    {
      "epoch": 24.79185119574845,
      "grad_norm": 0.2844630479812622,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 27990
    },
    {
      "epoch": 24.796279893711247,
      "grad_norm": 0.34206822514533997,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 27995
    },
    {
      "epoch": 24.80070859167405,
      "grad_norm": 0.2651706337928772,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 28000
    },
    {
      "epoch": 24.805137289636846,
      "grad_norm": 0.243279829621315,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 28005
    },
    {
      "epoch": 24.809565987599647,
      "grad_norm": 0.25350362062454224,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 28010
    },
    {
      "epoch": 24.813994685562445,
      "grad_norm": 0.219048872590065,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 28015
    },
    {
      "epoch": 24.818423383525243,
      "grad_norm": 0.24372893571853638,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 28020
    },
    {
      "epoch": 24.822852081488044,
      "grad_norm": 0.2852746844291687,
      "learning_rate": 1e-05,
      "loss": 1.0375,
      "step": 28025
    },
    {
      "epoch": 24.82728077945084,
      "grad_norm": 0.22398953139781952,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 28030
    },
    {
      "epoch": 24.83170947741364,
      "grad_norm": 0.2317638099193573,
      "learning_rate": 1e-05,
      "loss": 1.0511,
      "step": 28035
    },
    {
      "epoch": 24.83613817537644,
      "grad_norm": 0.258124977350235,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 28040
    },
    {
      "epoch": 24.840566873339238,
      "grad_norm": 0.25679975748062134,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 28045
    },
    {
      "epoch": 24.844995571302036,
      "grad_norm": 0.22252784669399261,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 28050
    },
    {
      "epoch": 24.849424269264837,
      "grad_norm": 0.24934500455856323,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 28055
    },
    {
      "epoch": 24.853852967227635,
      "grad_norm": 0.2803652584552765,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 28060
    },
    {
      "epoch": 24.858281665190432,
      "grad_norm": 0.21808862686157227,
      "learning_rate": 1e-05,
      "loss": 1.0203,
      "step": 28065
    },
    {
      "epoch": 24.862710363153234,
      "grad_norm": 0.22926940023899078,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 28070
    },
    {
      "epoch": 24.86713906111603,
      "grad_norm": 0.2412165403366089,
      "learning_rate": 1e-05,
      "loss": 1.0588,
      "step": 28075
    },
    {
      "epoch": 24.871567759078832,
      "grad_norm": 0.24139633774757385,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 28080
    },
    {
      "epoch": 24.87599645704163,
      "grad_norm": 0.24621202051639557,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 28085
    },
    {
      "epoch": 24.880425155004428,
      "grad_norm": 0.24362148344516754,
      "learning_rate": 1e-05,
      "loss": 1.0373,
      "step": 28090
    },
    {
      "epoch": 24.88485385296723,
      "grad_norm": 0.24921147525310516,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 28095
    },
    {
      "epoch": 24.889282550930027,
      "grad_norm": 0.2551196217536926,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 28100
    },
    {
      "epoch": 24.893711248892824,
      "grad_norm": 0.23885582387447357,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 28105
    },
    {
      "epoch": 24.898139946855625,
      "grad_norm": 0.21594491600990295,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 28110
    },
    {
      "epoch": 24.902568644818423,
      "grad_norm": 0.2223508208990097,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 28115
    },
    {
      "epoch": 24.90699734278122,
      "grad_norm": 0.2156670242547989,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 28120
    },
    {
      "epoch": 24.911426040744022,
      "grad_norm": 0.2858951687812805,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 28125
    },
    {
      "epoch": 24.91585473870682,
      "grad_norm": 0.2882167398929596,
      "learning_rate": 1e-05,
      "loss": 1.0159,
      "step": 28130
    },
    {
      "epoch": 24.92028343666962,
      "grad_norm": 0.2706875801086426,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 28135
    },
    {
      "epoch": 24.92471213463242,
      "grad_norm": 0.23093226552009583,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 28140
    },
    {
      "epoch": 24.929140832595216,
      "grad_norm": 0.2316305786371231,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 28145
    },
    {
      "epoch": 24.933569530558017,
      "grad_norm": 0.21752788126468658,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 28150
    },
    {
      "epoch": 24.937998228520815,
      "grad_norm": 0.19479702413082123,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 28155
    },
    {
      "epoch": 24.942426926483613,
      "grad_norm": 0.2344137579202652,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 28160
    },
    {
      "epoch": 24.946855624446414,
      "grad_norm": 0.23480965197086334,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 28165
    },
    {
      "epoch": 24.95128432240921,
      "grad_norm": 0.25966697931289673,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 28170
    },
    {
      "epoch": 24.95571302037201,
      "grad_norm": 0.2760307490825653,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 28175
    },
    {
      "epoch": 24.96014171833481,
      "grad_norm": 0.25913602113723755,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 28180
    },
    {
      "epoch": 24.964570416297608,
      "grad_norm": 0.24504348635673523,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 28185
    },
    {
      "epoch": 24.968999114260406,
      "grad_norm": 0.2237614244222641,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 28190
    },
    {
      "epoch": 24.973427812223207,
      "grad_norm": 0.31712502241134644,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 28195
    },
    {
      "epoch": 24.977856510186005,
      "grad_norm": 0.23510074615478516,
      "learning_rate": 1e-05,
      "loss": 1.0628,
      "step": 28200
    },
    {
      "epoch": 24.982285208148806,
      "grad_norm": 0.3259398341178894,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 28205
    },
    {
      "epoch": 24.986713906111603,
      "grad_norm": 0.24236249923706055,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 28210
    },
    {
      "epoch": 24.9911426040744,
      "grad_norm": 0.23763632774353027,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 28215
    },
    {
      "epoch": 24.995571302037202,
      "grad_norm": 0.22181124985218048,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 28220
    },
    {
      "epoch": 25.0,
      "grad_norm": 0.26796409487724304,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 28225
    },
    {
      "epoch": 25.004428697962798,
      "grad_norm": 0.3231574296951294,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 28230
    },
    {
      "epoch": 25.0088573959256,
      "grad_norm": 0.2569660246372223,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 28235
    },
    {
      "epoch": 25.013286093888397,
      "grad_norm": 0.26533979177474976,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 28240
    },
    {
      "epoch": 25.017714791851194,
      "grad_norm": 0.22557643055915833,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 28245
    },
    {
      "epoch": 25.022143489813995,
      "grad_norm": 0.30640771985054016,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 28250
    },
    {
      "epoch": 25.026572187776793,
      "grad_norm": 0.29978352785110474,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 28255
    },
    {
      "epoch": 25.031000885739594,
      "grad_norm": 0.27134472131729126,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 28260
    },
    {
      "epoch": 25.035429583702392,
      "grad_norm": 0.2424149513244629,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 28265
    },
    {
      "epoch": 25.03985828166519,
      "grad_norm": 0.22151607275009155,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 28270
    },
    {
      "epoch": 25.04428697962799,
      "grad_norm": 0.27944135665893555,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 28275
    },
    {
      "epoch": 25.04871567759079,
      "grad_norm": 0.2413647174835205,
      "learning_rate": 1e-05,
      "loss": 0.9319,
      "step": 28280
    },
    {
      "epoch": 25.053144375553586,
      "grad_norm": 0.3381389081478119,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 28285
    },
    {
      "epoch": 25.057573073516387,
      "grad_norm": 0.2824645936489105,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 28290
    },
    {
      "epoch": 25.062001771479185,
      "grad_norm": 0.26023802161216736,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 28295
    },
    {
      "epoch": 25.066430469441983,
      "grad_norm": 0.2313794046640396,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 28300
    },
    {
      "epoch": 25.070859167404784,
      "grad_norm": 0.24783073365688324,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 28305
    },
    {
      "epoch": 25.07528786536758,
      "grad_norm": 0.2511592209339142,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 28310
    },
    {
      "epoch": 25.07971656333038,
      "grad_norm": 0.22317899763584137,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 28315
    },
    {
      "epoch": 25.08414526129318,
      "grad_norm": 0.3170972764492035,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 28320
    },
    {
      "epoch": 25.088573959255978,
      "grad_norm": 0.2434535026550293,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 28325
    },
    {
      "epoch": 25.09300265721878,
      "grad_norm": 0.2370983362197876,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 28330
    },
    {
      "epoch": 25.097431355181577,
      "grad_norm": 0.23729123175144196,
      "learning_rate": 1e-05,
      "loss": 0.9091,
      "step": 28335
    },
    {
      "epoch": 25.101860053144375,
      "grad_norm": 0.2768557369709015,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 28340
    },
    {
      "epoch": 25.106288751107176,
      "grad_norm": 0.22914046049118042,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 28345
    },
    {
      "epoch": 25.110717449069973,
      "grad_norm": 0.2848317623138428,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 28350
    },
    {
      "epoch": 25.11514614703277,
      "grad_norm": 0.24439536035060883,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 28355
    },
    {
      "epoch": 25.119574844995572,
      "grad_norm": 0.24731579422950745,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 28360
    },
    {
      "epoch": 25.12400354295837,
      "grad_norm": 0.2253962755203247,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 28365
    },
    {
      "epoch": 25.128432240921168,
      "grad_norm": 0.23305702209472656,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 28370
    },
    {
      "epoch": 25.13286093888397,
      "grad_norm": 0.2980017364025116,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 28375
    },
    {
      "epoch": 25.137289636846766,
      "grad_norm": 0.23258838057518005,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 28380
    },
    {
      "epoch": 25.141718334809568,
      "grad_norm": 0.2909734845161438,
      "learning_rate": 1e-05,
      "loss": 1.0273,
      "step": 28385
    },
    {
      "epoch": 25.146147032772365,
      "grad_norm": 0.2500873804092407,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 28390
    },
    {
      "epoch": 25.150575730735163,
      "grad_norm": 0.22985301911830902,
      "learning_rate": 1e-05,
      "loss": 1.0276,
      "step": 28395
    },
    {
      "epoch": 25.155004428697964,
      "grad_norm": 0.2257264405488968,
      "learning_rate": 1e-05,
      "loss": 1.0225,
      "step": 28400
    },
    {
      "epoch": 25.159433126660762,
      "grad_norm": 0.23549792170524597,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 28405
    },
    {
      "epoch": 25.16386182462356,
      "grad_norm": 0.2427898645401001,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 28410
    },
    {
      "epoch": 25.16829052258636,
      "grad_norm": 0.2809029221534729,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 28415
    },
    {
      "epoch": 25.17271922054916,
      "grad_norm": 0.21593166887760162,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 28420
    },
    {
      "epoch": 25.177147918511956,
      "grad_norm": 0.27121323347091675,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 28425
    },
    {
      "epoch": 25.181576616474757,
      "grad_norm": 0.24392178654670715,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 28430
    },
    {
      "epoch": 25.186005314437555,
      "grad_norm": 0.21362219750881195,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 28435
    },
    {
      "epoch": 25.190434012400353,
      "grad_norm": 0.23220038414001465,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 28440
    },
    {
      "epoch": 25.194862710363154,
      "grad_norm": 0.20844097435474396,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 28445
    },
    {
      "epoch": 25.19929140832595,
      "grad_norm": 0.22274212539196014,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 28450
    },
    {
      "epoch": 25.203720106288753,
      "grad_norm": 0.213849276304245,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 28455
    },
    {
      "epoch": 25.20814880425155,
      "grad_norm": 0.22715076804161072,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 28460
    },
    {
      "epoch": 25.212577502214348,
      "grad_norm": 0.25142112374305725,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 28465
    },
    {
      "epoch": 25.21700620017715,
      "grad_norm": 0.23326852917671204,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 28470
    },
    {
      "epoch": 25.221434898139947,
      "grad_norm": 0.26229164004325867,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 28475
    },
    {
      "epoch": 25.225863596102744,
      "grad_norm": 0.2494177520275116,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 28480
    },
    {
      "epoch": 25.230292294065546,
      "grad_norm": 0.27019643783569336,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 28485
    },
    {
      "epoch": 25.234720992028343,
      "grad_norm": 0.2603653073310852,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 28490
    },
    {
      "epoch": 25.23914968999114,
      "grad_norm": 0.2675544023513794,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 28495
    },
    {
      "epoch": 25.243578387953942,
      "grad_norm": 0.24533584713935852,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 28500
    },
    {
      "epoch": 25.24800708591674,
      "grad_norm": 0.22727340459823608,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 28505
    },
    {
      "epoch": 25.25243578387954,
      "grad_norm": 0.22361747920513153,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 28510
    },
    {
      "epoch": 25.25686448184234,
      "grad_norm": 0.24781891703605652,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 28515
    },
    {
      "epoch": 25.261293179805136,
      "grad_norm": 0.2413349747657776,
      "learning_rate": 1e-05,
      "loss": 1.0345,
      "step": 28520
    },
    {
      "epoch": 25.265721877767938,
      "grad_norm": 0.25148409605026245,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 28525
    },
    {
      "epoch": 25.270150575730735,
      "grad_norm": 0.2741221785545349,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 28530
    },
    {
      "epoch": 25.274579273693533,
      "grad_norm": 0.21583959460258484,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 28535
    },
    {
      "epoch": 25.279007971656334,
      "grad_norm": 0.22502276301383972,
      "learning_rate": 1e-05,
      "loss": 1.0111,
      "step": 28540
    },
    {
      "epoch": 25.28343666961913,
      "grad_norm": 0.23255762457847595,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 28545
    },
    {
      "epoch": 25.28786536758193,
      "grad_norm": 0.257028728723526,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 28550
    },
    {
      "epoch": 25.29229406554473,
      "grad_norm": 0.22574196755886078,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 28555
    },
    {
      "epoch": 25.29672276350753,
      "grad_norm": 0.24432362616062164,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 28560
    },
    {
      "epoch": 25.30115146147033,
      "grad_norm": 0.2594378888607025,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 28565
    },
    {
      "epoch": 25.305580159433127,
      "grad_norm": 0.23691865801811218,
      "learning_rate": 1e-05,
      "loss": 1.0283,
      "step": 28570
    },
    {
      "epoch": 25.310008857395925,
      "grad_norm": 0.28134483098983765,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 28575
    },
    {
      "epoch": 25.314437555358726,
      "grad_norm": 0.2507847845554352,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 28580
    },
    {
      "epoch": 25.318866253321524,
      "grad_norm": 0.23688143491744995,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 28585
    },
    {
      "epoch": 25.32329495128432,
      "grad_norm": 0.26104605197906494,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 28590
    },
    {
      "epoch": 25.327723649247122,
      "grad_norm": 0.23105032742023468,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 28595
    },
    {
      "epoch": 25.33215234720992,
      "grad_norm": 0.2216823548078537,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 28600
    },
    {
      "epoch": 25.336581045172718,
      "grad_norm": 0.27124977111816406,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 28605
    },
    {
      "epoch": 25.34100974313552,
      "grad_norm": 0.2161342352628708,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 28610
    },
    {
      "epoch": 25.345438441098317,
      "grad_norm": 0.26108431816101074,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 28615
    },
    {
      "epoch": 25.349867139061114,
      "grad_norm": 0.25465962290763855,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 28620
    },
    {
      "epoch": 25.354295837023916,
      "grad_norm": 0.28791970014572144,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 28625
    },
    {
      "epoch": 25.358724534986713,
      "grad_norm": 0.26569464802742004,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 28630
    },
    {
      "epoch": 25.363153232949514,
      "grad_norm": 0.24875789880752563,
      "learning_rate": 1e-05,
      "loss": 1.0475,
      "step": 28635
    },
    {
      "epoch": 25.367581930912312,
      "grad_norm": 0.20327898859977722,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 28640
    },
    {
      "epoch": 25.37201062887511,
      "grad_norm": 0.2711411416530609,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 28645
    },
    {
      "epoch": 25.37643932683791,
      "grad_norm": 0.20519310235977173,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 28650
    },
    {
      "epoch": 25.38086802480071,
      "grad_norm": 0.2470201849937439,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 28655
    },
    {
      "epoch": 25.385296722763506,
      "grad_norm": 0.2653522789478302,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 28660
    },
    {
      "epoch": 25.389725420726307,
      "grad_norm": 0.2669956386089325,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 28665
    },
    {
      "epoch": 25.394154118689105,
      "grad_norm": 0.2532452344894409,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 28670
    },
    {
      "epoch": 25.398582816651903,
      "grad_norm": 0.2404826581478119,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 28675
    },
    {
      "epoch": 25.403011514614704,
      "grad_norm": 0.21702100336551666,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 28680
    },
    {
      "epoch": 25.4074402125775,
      "grad_norm": 0.24474535882472992,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 28685
    },
    {
      "epoch": 25.411868910540303,
      "grad_norm": 0.28809377551078796,
      "learning_rate": 1e-05,
      "loss": 1.0199,
      "step": 28690
    },
    {
      "epoch": 25.4162976085031,
      "grad_norm": 0.27852576971054077,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 28695
    },
    {
      "epoch": 25.420726306465898,
      "grad_norm": 0.20958438515663147,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 28700
    },
    {
      "epoch": 25.4251550044287,
      "grad_norm": 0.20069308578968048,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 28705
    },
    {
      "epoch": 25.429583702391497,
      "grad_norm": 0.2399550825357437,
      "learning_rate": 1e-05,
      "loss": 0.9273,
      "step": 28710
    },
    {
      "epoch": 25.434012400354295,
      "grad_norm": 0.2795233428478241,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 28715
    },
    {
      "epoch": 25.438441098317096,
      "grad_norm": 0.22888731956481934,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 28720
    },
    {
      "epoch": 25.442869796279894,
      "grad_norm": 0.23873721063137054,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 28725
    },
    {
      "epoch": 25.44729849424269,
      "grad_norm": 0.25387635827064514,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 28730
    },
    {
      "epoch": 25.451727192205492,
      "grad_norm": 0.30345356464385986,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 28735
    },
    {
      "epoch": 25.45615589016829,
      "grad_norm": 0.29723215103149414,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 28740
    },
    {
      "epoch": 25.460584588131088,
      "grad_norm": 0.24982304871082306,
      "learning_rate": 1e-05,
      "loss": 1.0278,
      "step": 28745
    },
    {
      "epoch": 25.46501328609389,
      "grad_norm": 0.3130345344543457,
      "learning_rate": 1e-05,
      "loss": 1.0473,
      "step": 28750
    },
    {
      "epoch": 25.469441984056687,
      "grad_norm": 0.20515911281108856,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 28755
    },
    {
      "epoch": 25.473870682019488,
      "grad_norm": 0.23972462117671967,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 28760
    },
    {
      "epoch": 25.478299379982285,
      "grad_norm": 0.2404281198978424,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 28765
    },
    {
      "epoch": 25.482728077945083,
      "grad_norm": 0.23902878165245056,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 28770
    },
    {
      "epoch": 25.487156775907884,
      "grad_norm": 0.25248533487319946,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 28775
    },
    {
      "epoch": 25.491585473870682,
      "grad_norm": 0.23704902827739716,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 28780
    },
    {
      "epoch": 25.49601417183348,
      "grad_norm": 0.2655435800552368,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 28785
    },
    {
      "epoch": 25.50044286979628,
      "grad_norm": 0.23425692319869995,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 28790
    },
    {
      "epoch": 25.50487156775908,
      "grad_norm": 0.2840118110179901,
      "learning_rate": 1e-05,
      "loss": 1.0216,
      "step": 28795
    },
    {
      "epoch": 25.509300265721876,
      "grad_norm": 0.31624239683151245,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 28800
    },
    {
      "epoch": 25.513728963684677,
      "grad_norm": 0.24195078015327454,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 28805
    },
    {
      "epoch": 25.518157661647475,
      "grad_norm": 0.24833817780017853,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 28810
    },
    {
      "epoch": 25.522586359610276,
      "grad_norm": 0.22840385138988495,
      "learning_rate": 1e-05,
      "loss": 1.014,
      "step": 28815
    },
    {
      "epoch": 25.527015057573074,
      "grad_norm": 0.2724335193634033,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 28820
    },
    {
      "epoch": 25.53144375553587,
      "grad_norm": 0.31222543120384216,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 28825
    },
    {
      "epoch": 25.535872453498673,
      "grad_norm": 0.2051672637462616,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 28830
    },
    {
      "epoch": 25.54030115146147,
      "grad_norm": 0.2828862965106964,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 28835
    },
    {
      "epoch": 25.544729849424268,
      "grad_norm": 0.22247949242591858,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 28840
    },
    {
      "epoch": 25.54915854738707,
      "grad_norm": 0.2899076044559479,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 28845
    },
    {
      "epoch": 25.553587245349867,
      "grad_norm": 0.2460116147994995,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 28850
    },
    {
      "epoch": 25.558015943312665,
      "grad_norm": 0.25456270575523376,
      "learning_rate": 1e-05,
      "loss": 0.9158,
      "step": 28855
    },
    {
      "epoch": 25.562444641275466,
      "grad_norm": 0.23259401321411133,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 28860
    },
    {
      "epoch": 25.566873339238263,
      "grad_norm": 0.22429171204566956,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 28865
    },
    {
      "epoch": 25.571302037201065,
      "grad_norm": 0.21937568485736847,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 28870
    },
    {
      "epoch": 25.575730735163862,
      "grad_norm": 0.20632602274417877,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 28875
    },
    {
      "epoch": 25.58015943312666,
      "grad_norm": 0.25110140442848206,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 28880
    },
    {
      "epoch": 25.58458813108946,
      "grad_norm": 0.2793775200843811,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 28885
    },
    {
      "epoch": 25.58901682905226,
      "grad_norm": 0.28905123472213745,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 28890
    },
    {
      "epoch": 25.593445527015056,
      "grad_norm": 0.25348004698753357,
      "learning_rate": 1e-05,
      "loss": 1.034,
      "step": 28895
    },
    {
      "epoch": 25.597874224977858,
      "grad_norm": 0.3241104483604431,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 28900
    },
    {
      "epoch": 25.602302922940655,
      "grad_norm": 0.2509087324142456,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 28905
    },
    {
      "epoch": 25.606731620903453,
      "grad_norm": 0.2558797001838684,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 28910
    },
    {
      "epoch": 25.611160318866254,
      "grad_norm": 0.25341325998306274,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 28915
    },
    {
      "epoch": 25.615589016829052,
      "grad_norm": 0.24359464645385742,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 28920
    },
    {
      "epoch": 25.62001771479185,
      "grad_norm": 0.23044449090957642,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 28925
    },
    {
      "epoch": 25.62444641275465,
      "grad_norm": 0.23174871504306793,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 28930
    },
    {
      "epoch": 25.62887511071745,
      "grad_norm": 0.25438809394836426,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 28935
    },
    {
      "epoch": 25.63330380868025,
      "grad_norm": 0.2380950003862381,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 28940
    },
    {
      "epoch": 25.637732506643047,
      "grad_norm": 0.26295799016952515,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 28945
    },
    {
      "epoch": 25.642161204605845,
      "grad_norm": 0.2500235140323639,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 28950
    },
    {
      "epoch": 25.646589902568646,
      "grad_norm": 0.25703057646751404,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 28955
    },
    {
      "epoch": 25.651018600531444,
      "grad_norm": 0.23264075815677643,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 28960
    },
    {
      "epoch": 25.65544729849424,
      "grad_norm": 0.25241145491600037,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 28965
    },
    {
      "epoch": 25.659875996457043,
      "grad_norm": 0.2852046489715576,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 28970
    },
    {
      "epoch": 25.66430469441984,
      "grad_norm": 0.3080235421657562,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 28975
    },
    {
      "epoch": 25.668733392382638,
      "grad_norm": 0.22542570531368256,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 28980
    },
    {
      "epoch": 25.67316209034544,
      "grad_norm": 0.22936002910137177,
      "learning_rate": 1e-05,
      "loss": 0.9273,
      "step": 28985
    },
    {
      "epoch": 25.677590788308237,
      "grad_norm": 0.23263134062290192,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 28990
    },
    {
      "epoch": 25.682019486271038,
      "grad_norm": 0.21266630291938782,
      "learning_rate": 1e-05,
      "loss": 0.9286,
      "step": 28995
    },
    {
      "epoch": 25.686448184233836,
      "grad_norm": 0.23370856046676636,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 29000
    },
    {
      "epoch": 25.690876882196633,
      "grad_norm": 0.21539267897605896,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 29005
    },
    {
      "epoch": 25.695305580159435,
      "grad_norm": 0.2574823796749115,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 29010
    },
    {
      "epoch": 25.699734278122232,
      "grad_norm": 0.24346426129341125,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 29015
    },
    {
      "epoch": 25.70416297608503,
      "grad_norm": 0.2353721708059311,
      "learning_rate": 1e-05,
      "loss": 1.0255,
      "step": 29020
    },
    {
      "epoch": 25.70859167404783,
      "grad_norm": 0.24475936591625214,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 29025
    },
    {
      "epoch": 25.71302037201063,
      "grad_norm": 0.24502651393413544,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 29030
    },
    {
      "epoch": 25.717449069973426,
      "grad_norm": 0.24546490609645844,
      "learning_rate": 1e-05,
      "loss": 1.0171,
      "step": 29035
    },
    {
      "epoch": 25.721877767936228,
      "grad_norm": 0.23093603551387787,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 29040
    },
    {
      "epoch": 25.726306465899025,
      "grad_norm": 0.22885775566101074,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 29045
    },
    {
      "epoch": 25.730735163861823,
      "grad_norm": 0.2644462585449219,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 29050
    },
    {
      "epoch": 25.735163861824624,
      "grad_norm": 0.2482924908399582,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 29055
    },
    {
      "epoch": 25.739592559787422,
      "grad_norm": 0.27061980962753296,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 29060
    },
    {
      "epoch": 25.744021257750223,
      "grad_norm": 0.2516295909881592,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 29065
    },
    {
      "epoch": 25.74844995571302,
      "grad_norm": 0.2653324007987976,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 29070
    },
    {
      "epoch": 25.75287865367582,
      "grad_norm": 0.2714124619960785,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 29075
    },
    {
      "epoch": 25.75730735163862,
      "grad_norm": 0.28832945227622986,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 29080
    },
    {
      "epoch": 25.761736049601417,
      "grad_norm": 0.3036993443965912,
      "learning_rate": 1e-05,
      "loss": 1.0387,
      "step": 29085
    },
    {
      "epoch": 25.766164747564215,
      "grad_norm": 0.269141286611557,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 29090
    },
    {
      "epoch": 25.770593445527016,
      "grad_norm": 0.27736595273017883,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 29095
    },
    {
      "epoch": 25.775022143489814,
      "grad_norm": 0.2391963005065918,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 29100
    },
    {
      "epoch": 25.77945084145261,
      "grad_norm": 0.25906017422676086,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 29105
    },
    {
      "epoch": 25.783879539415413,
      "grad_norm": 0.24674907326698303,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 29110
    },
    {
      "epoch": 25.78830823737821,
      "grad_norm": 0.24169249832630157,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 29115
    },
    {
      "epoch": 25.79273693534101,
      "grad_norm": 0.22263750433921814,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 29120
    },
    {
      "epoch": 25.79716563330381,
      "grad_norm": 0.24689054489135742,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 29125
    },
    {
      "epoch": 25.801594331266607,
      "grad_norm": 0.2252499759197235,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 29130
    },
    {
      "epoch": 25.806023029229408,
      "grad_norm": 0.21843624114990234,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 29135
    },
    {
      "epoch": 25.810451727192206,
      "grad_norm": 0.23872670531272888,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 29140
    },
    {
      "epoch": 25.814880425155003,
      "grad_norm": 0.23328737914562225,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 29145
    },
    {
      "epoch": 25.819309123117804,
      "grad_norm": 0.2513371706008911,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 29150
    },
    {
      "epoch": 25.823737821080602,
      "grad_norm": 0.24121609330177307,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 29155
    },
    {
      "epoch": 25.8281665190434,
      "grad_norm": 0.2561293840408325,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 29160
    },
    {
      "epoch": 25.8325952170062,
      "grad_norm": 0.2826625108718872,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 29165
    },
    {
      "epoch": 25.837023914969,
      "grad_norm": 0.22753024101257324,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 29170
    },
    {
      "epoch": 25.841452612931796,
      "grad_norm": 0.28123176097869873,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 29175
    },
    {
      "epoch": 25.845881310894598,
      "grad_norm": 0.25102218985557556,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 29180
    },
    {
      "epoch": 25.850310008857395,
      "grad_norm": 0.24924074113368988,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 29185
    },
    {
      "epoch": 25.854738706820196,
      "grad_norm": 0.2137407809495926,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 29190
    },
    {
      "epoch": 25.859167404782994,
      "grad_norm": 0.2659122347831726,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 29195
    },
    {
      "epoch": 25.86359610274579,
      "grad_norm": 0.235785573720932,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 29200
    },
    {
      "epoch": 25.868024800708593,
      "grad_norm": 0.20222915709018707,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 29205
    },
    {
      "epoch": 25.87245349867139,
      "grad_norm": 0.22771041095256805,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 29210
    },
    {
      "epoch": 25.876882196634188,
      "grad_norm": 0.2526921331882477,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 29215
    },
    {
      "epoch": 25.88131089459699,
      "grad_norm": 0.24338418245315552,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 29220
    },
    {
      "epoch": 25.885739592559787,
      "grad_norm": 0.25345486402511597,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 29225
    },
    {
      "epoch": 25.890168290522585,
      "grad_norm": 0.21984684467315674,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 29230
    },
    {
      "epoch": 25.894596988485386,
      "grad_norm": 0.23430711030960083,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 29235
    },
    {
      "epoch": 25.899025686448184,
      "grad_norm": 0.24225330352783203,
      "learning_rate": 1e-05,
      "loss": 1.035,
      "step": 29240
    },
    {
      "epoch": 25.903454384410985,
      "grad_norm": 0.2422517091035843,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 29245
    },
    {
      "epoch": 25.907883082373782,
      "grad_norm": 0.23553062975406647,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 29250
    },
    {
      "epoch": 25.91231178033658,
      "grad_norm": 0.32093924283981323,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 29255
    },
    {
      "epoch": 25.91674047829938,
      "grad_norm": 0.23774869740009308,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 29260
    },
    {
      "epoch": 25.92116917626218,
      "grad_norm": 0.216340571641922,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 29265
    },
    {
      "epoch": 25.925597874224977,
      "grad_norm": 0.20214806497097015,
      "learning_rate": 1e-05,
      "loss": 0.9212,
      "step": 29270
    },
    {
      "epoch": 25.930026572187778,
      "grad_norm": 0.21292157471179962,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 29275
    },
    {
      "epoch": 25.934455270150576,
      "grad_norm": 0.22030073404312134,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 29280
    },
    {
      "epoch": 25.938883968113373,
      "grad_norm": 0.2435506135225296,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 29285
    },
    {
      "epoch": 25.943312666076174,
      "grad_norm": 0.2213594764471054,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 29290
    },
    {
      "epoch": 25.947741364038972,
      "grad_norm": 0.2633857727050781,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 29295
    },
    {
      "epoch": 25.95217006200177,
      "grad_norm": 0.23637594282627106,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 29300
    },
    {
      "epoch": 25.95659875996457,
      "grad_norm": 0.2753964960575104,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 29305
    },
    {
      "epoch": 25.96102745792737,
      "grad_norm": 0.22327548265457153,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 29310
    },
    {
      "epoch": 25.96545615589017,
      "grad_norm": 0.2317304164171219,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 29315
    },
    {
      "epoch": 25.969884853852967,
      "grad_norm": 0.24318106472492218,
      "learning_rate": 1e-05,
      "loss": 1.033,
      "step": 29320
    },
    {
      "epoch": 25.974313551815765,
      "grad_norm": 0.2720038592815399,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 29325
    },
    {
      "epoch": 25.978742249778566,
      "grad_norm": 0.26555389165878296,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 29330
    },
    {
      "epoch": 25.983170947741364,
      "grad_norm": 0.2475748062133789,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 29335
    },
    {
      "epoch": 25.98759964570416,
      "grad_norm": 0.25087156891822815,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 29340
    },
    {
      "epoch": 25.992028343666963,
      "grad_norm": 0.24291250109672546,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 29345
    },
    {
      "epoch": 25.99645704162976,
      "grad_norm": 0.25244009494781494,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 29350
    },
    {
      "epoch": 26.000885739592558,
      "grad_norm": 0.19791699945926666,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 29355
    },
    {
      "epoch": 26.00531443755536,
      "grad_norm": 0.244821235537529,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 29360
    },
    {
      "epoch": 26.009743135518157,
      "grad_norm": 0.28714078664779663,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 29365
    },
    {
      "epoch": 26.014171833480958,
      "grad_norm": 0.2341163009405136,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 29370
    },
    {
      "epoch": 26.018600531443756,
      "grad_norm": 0.2755536735057831,
      "learning_rate": 1e-05,
      "loss": 1.0148,
      "step": 29375
    },
    {
      "epoch": 26.023029229406553,
      "grad_norm": 0.24109947681427002,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 29380
    },
    {
      "epoch": 26.027457927369355,
      "grad_norm": 0.23363451659679413,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 29385
    },
    {
      "epoch": 26.031886625332152,
      "grad_norm": 0.24639639258384705,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 29390
    },
    {
      "epoch": 26.03631532329495,
      "grad_norm": 0.27804136276245117,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 29395
    },
    {
      "epoch": 26.04074402125775,
      "grad_norm": 0.27713873982429504,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 29400
    },
    {
      "epoch": 26.04517271922055,
      "grad_norm": 0.27685269713401794,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 29405
    },
    {
      "epoch": 26.049601417183347,
      "grad_norm": 0.2663515508174896,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 29410
    },
    {
      "epoch": 26.054030115146148,
      "grad_norm": 0.24985182285308838,
      "learning_rate": 1e-05,
      "loss": 1.0405,
      "step": 29415
    },
    {
      "epoch": 26.058458813108945,
      "grad_norm": 0.2278592586517334,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 29420
    },
    {
      "epoch": 26.062887511071747,
      "grad_norm": 0.21478793025016785,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 29425
    },
    {
      "epoch": 26.067316209034544,
      "grad_norm": 0.24094019830226898,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 29430
    },
    {
      "epoch": 26.071744906997342,
      "grad_norm": 0.3006095588207245,
      "learning_rate": 1e-05,
      "loss": 1.0507,
      "step": 29435
    },
    {
      "epoch": 26.076173604960143,
      "grad_norm": 0.23204229772090912,
      "learning_rate": 1e-05,
      "loss": 1.0561,
      "step": 29440
    },
    {
      "epoch": 26.08060230292294,
      "grad_norm": 0.24658440053462982,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 29445
    },
    {
      "epoch": 26.08503100088574,
      "grad_norm": 0.26472967863082886,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 29450
    },
    {
      "epoch": 26.08945969884854,
      "grad_norm": 0.22890281677246094,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 29455
    },
    {
      "epoch": 26.093888396811337,
      "grad_norm": 0.2702520787715912,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 29460
    },
    {
      "epoch": 26.098317094774135,
      "grad_norm": 0.2833285629749298,
      "learning_rate": 1e-05,
      "loss": 1.0355,
      "step": 29465
    },
    {
      "epoch": 26.102745792736936,
      "grad_norm": 0.29363709688186646,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 29470
    },
    {
      "epoch": 26.107174490699734,
      "grad_norm": 0.19844835996627808,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 29475
    },
    {
      "epoch": 26.11160318866253,
      "grad_norm": 0.20662382245063782,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 29480
    },
    {
      "epoch": 26.116031886625333,
      "grad_norm": 0.2563677728176117,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 29485
    },
    {
      "epoch": 26.12046058458813,
      "grad_norm": 0.2552192807197571,
      "learning_rate": 1e-05,
      "loss": 1.033,
      "step": 29490
    },
    {
      "epoch": 26.12488928255093,
      "grad_norm": 0.24195873737335205,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 29495
    },
    {
      "epoch": 26.12931798051373,
      "grad_norm": 0.2592959702014923,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 29500
    },
    {
      "epoch": 26.133746678476527,
      "grad_norm": 0.2350163459777832,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 29505
    },
    {
      "epoch": 26.138175376439328,
      "grad_norm": 0.25672343373298645,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 29510
    },
    {
      "epoch": 26.142604074402126,
      "grad_norm": 0.20711758732795715,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 29515
    },
    {
      "epoch": 26.147032772364923,
      "grad_norm": 0.27545568346977234,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 29520
    },
    {
      "epoch": 26.151461470327725,
      "grad_norm": 0.2389090359210968,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 29525
    },
    {
      "epoch": 26.155890168290522,
      "grad_norm": 0.21873405575752258,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 29530
    },
    {
      "epoch": 26.16031886625332,
      "grad_norm": 0.25126391649246216,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 29535
    },
    {
      "epoch": 26.16474756421612,
      "grad_norm": 0.2871840000152588,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 29540
    },
    {
      "epoch": 26.16917626217892,
      "grad_norm": 0.22746191918849945,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 29545
    },
    {
      "epoch": 26.17360496014172,
      "grad_norm": 0.23948286473751068,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 29550
    },
    {
      "epoch": 26.178033658104518,
      "grad_norm": 0.23553849756717682,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 29555
    },
    {
      "epoch": 26.182462356067315,
      "grad_norm": 0.2790740132331848,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 29560
    },
    {
      "epoch": 26.186891054030117,
      "grad_norm": 0.24037010967731476,
      "learning_rate": 1e-05,
      "loss": 0.9086,
      "step": 29565
    },
    {
      "epoch": 26.191319751992914,
      "grad_norm": 0.22348038852214813,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 29570
    },
    {
      "epoch": 26.195748449955712,
      "grad_norm": 0.25426185131073,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 29575
    },
    {
      "epoch": 26.200177147918513,
      "grad_norm": 0.31084033846855164,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 29580
    },
    {
      "epoch": 26.20460584588131,
      "grad_norm": 0.2768554389476776,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 29585
    },
    {
      "epoch": 26.20903454384411,
      "grad_norm": 0.22014443576335907,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 29590
    },
    {
      "epoch": 26.21346324180691,
      "grad_norm": 0.2207522690296173,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 29595
    },
    {
      "epoch": 26.217891939769707,
      "grad_norm": 0.2294379025697708,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 29600
    },
    {
      "epoch": 26.22232063773251,
      "grad_norm": 0.26175636053085327,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 29605
    },
    {
      "epoch": 26.226749335695306,
      "grad_norm": 0.22617003321647644,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 29610
    },
    {
      "epoch": 26.231178033658104,
      "grad_norm": 0.2585447132587433,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 29615
    },
    {
      "epoch": 26.235606731620905,
      "grad_norm": 0.18052157759666443,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 29620
    },
    {
      "epoch": 26.240035429583703,
      "grad_norm": 0.22431935369968414,
      "learning_rate": 1e-05,
      "loss": 0.9271,
      "step": 29625
    },
    {
      "epoch": 26.2444641275465,
      "grad_norm": 0.22433078289031982,
      "learning_rate": 1e-05,
      "loss": 1.0149,
      "step": 29630
    },
    {
      "epoch": 26.2488928255093,
      "grad_norm": 0.285879522562027,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 29635
    },
    {
      "epoch": 26.2533215234721,
      "grad_norm": 0.2589271366596222,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 29640
    },
    {
      "epoch": 26.257750221434897,
      "grad_norm": 0.23681409657001495,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 29645
    },
    {
      "epoch": 26.262178919397698,
      "grad_norm": 0.2588694095611572,
      "learning_rate": 1e-05,
      "loss": 1.052,
      "step": 29650
    },
    {
      "epoch": 26.266607617360496,
      "grad_norm": 0.2883615791797638,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 29655
    },
    {
      "epoch": 26.271036315323293,
      "grad_norm": 0.25754493474960327,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 29660
    },
    {
      "epoch": 26.275465013286095,
      "grad_norm": 0.25957587361335754,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 29665
    },
    {
      "epoch": 26.279893711248892,
      "grad_norm": 0.32041653990745544,
      "learning_rate": 1e-05,
      "loss": 1.0325,
      "step": 29670
    },
    {
      "epoch": 26.284322409211693,
      "grad_norm": 0.23580865561962128,
      "learning_rate": 1e-05,
      "loss": 1.0189,
      "step": 29675
    },
    {
      "epoch": 26.28875110717449,
      "grad_norm": 0.25689220428466797,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 29680
    },
    {
      "epoch": 26.29317980513729,
      "grad_norm": 0.2997405230998993,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 29685
    },
    {
      "epoch": 26.29760850310009,
      "grad_norm": 0.2702265679836273,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 29690
    },
    {
      "epoch": 26.302037201062888,
      "grad_norm": 0.27100005745887756,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 29695
    },
    {
      "epoch": 26.306465899025685,
      "grad_norm": 0.2557739019393921,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 29700
    },
    {
      "epoch": 26.310894596988486,
      "grad_norm": 0.23851528763771057,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 29705
    },
    {
      "epoch": 26.315323294951284,
      "grad_norm": 0.25407180190086365,
      "learning_rate": 1e-05,
      "loss": 1.0598,
      "step": 29710
    },
    {
      "epoch": 26.31975199291408,
      "grad_norm": 0.2696511447429657,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 29715
    },
    {
      "epoch": 26.324180690876883,
      "grad_norm": 0.23409490287303925,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 29720
    },
    {
      "epoch": 26.32860938883968,
      "grad_norm": 0.20391660928726196,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 29725
    },
    {
      "epoch": 26.333038086802482,
      "grad_norm": 0.23955675959587097,
      "learning_rate": 1e-05,
      "loss": 0.9295,
      "step": 29730
    },
    {
      "epoch": 26.33746678476528,
      "grad_norm": 0.25519853830337524,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 29735
    },
    {
      "epoch": 26.341895482728077,
      "grad_norm": 0.3160719871520996,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 29740
    },
    {
      "epoch": 26.34632418069088,
      "grad_norm": 0.2571757435798645,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 29745
    },
    {
      "epoch": 26.350752878653676,
      "grad_norm": 0.25098538398742676,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 29750
    },
    {
      "epoch": 26.355181576616474,
      "grad_norm": 0.2709645628929138,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 29755
    },
    {
      "epoch": 26.359610274579275,
      "grad_norm": 0.3633810579776764,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 29760
    },
    {
      "epoch": 26.364038972542073,
      "grad_norm": 0.273794949054718,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 29765
    },
    {
      "epoch": 26.36846767050487,
      "grad_norm": 0.21977777779102325,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 29770
    },
    {
      "epoch": 26.37289636846767,
      "grad_norm": 0.24285535514354706,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 29775
    },
    {
      "epoch": 26.37732506643047,
      "grad_norm": 0.22680318355560303,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 29780
    },
    {
      "epoch": 26.381753764393267,
      "grad_norm": 0.21320123970508575,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 29785
    },
    {
      "epoch": 26.386182462356068,
      "grad_norm": 0.3039967715740204,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 29790
    },
    {
      "epoch": 26.390611160318866,
      "grad_norm": 0.3038237392902374,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 29795
    },
    {
      "epoch": 26.395039858281667,
      "grad_norm": 0.23035113513469696,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 29800
    },
    {
      "epoch": 26.399468556244464,
      "grad_norm": 0.24465003609657288,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 29805
    },
    {
      "epoch": 26.403897254207262,
      "grad_norm": 0.2605549395084381,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 29810
    },
    {
      "epoch": 26.408325952170063,
      "grad_norm": 0.23271866142749786,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 29815
    },
    {
      "epoch": 26.41275465013286,
      "grad_norm": 0.2813819944858551,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 29820
    },
    {
      "epoch": 26.41718334809566,
      "grad_norm": 0.23463965952396393,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 29825
    },
    {
      "epoch": 26.42161204605846,
      "grad_norm": 0.2190408855676651,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 29830
    },
    {
      "epoch": 26.426040744021257,
      "grad_norm": 0.2265297770500183,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 29835
    },
    {
      "epoch": 26.430469441984055,
      "grad_norm": 0.280929833650589,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 29840
    },
    {
      "epoch": 26.434898139946856,
      "grad_norm": 0.2773053050041199,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 29845
    },
    {
      "epoch": 26.439326837909654,
      "grad_norm": 0.2736125588417053,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 29850
    },
    {
      "epoch": 26.443755535872455,
      "grad_norm": 0.21759001910686493,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 29855
    },
    {
      "epoch": 26.448184233835253,
      "grad_norm": 0.2628740966320038,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 29860
    },
    {
      "epoch": 26.45261293179805,
      "grad_norm": 0.23844057321548462,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 29865
    },
    {
      "epoch": 26.45704162976085,
      "grad_norm": 0.23006317019462585,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 29870
    },
    {
      "epoch": 26.46147032772365,
      "grad_norm": 0.24007225036621094,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 29875
    },
    {
      "epoch": 26.465899025686447,
      "grad_norm": 0.29693323373794556,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 29880
    },
    {
      "epoch": 26.47032772364925,
      "grad_norm": 0.23353172838687897,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 29885
    },
    {
      "epoch": 26.474756421612046,
      "grad_norm": 0.21376121044158936,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 29890
    },
    {
      "epoch": 26.479185119574844,
      "grad_norm": 0.2312644124031067,
      "learning_rate": 1e-05,
      "loss": 1.0286,
      "step": 29895
    },
    {
      "epoch": 26.483613817537645,
      "grad_norm": 0.2439718395471573,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 29900
    },
    {
      "epoch": 26.488042515500442,
      "grad_norm": 0.28455981612205505,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 29905
    },
    {
      "epoch": 26.49247121346324,
      "grad_norm": 0.2266928255558014,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 29910
    },
    {
      "epoch": 26.49689991142604,
      "grad_norm": 0.22934497892856598,
      "learning_rate": 1e-05,
      "loss": 1.0183,
      "step": 29915
    },
    {
      "epoch": 26.50132860938884,
      "grad_norm": 0.23625163733959198,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 29920
    },
    {
      "epoch": 26.50575730735164,
      "grad_norm": 0.21642480790615082,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 29925
    },
    {
      "epoch": 26.510186005314438,
      "grad_norm": 0.21227529644966125,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 29930
    },
    {
      "epoch": 26.514614703277235,
      "grad_norm": 0.22259730100631714,
      "learning_rate": 1e-05,
      "loss": 1.0482,
      "step": 29935
    },
    {
      "epoch": 26.519043401240037,
      "grad_norm": 0.22608643770217896,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 29940
    },
    {
      "epoch": 26.523472099202834,
      "grad_norm": 0.22166387736797333,
      "learning_rate": 1e-05,
      "loss": 1.0214,
      "step": 29945
    },
    {
      "epoch": 26.527900797165632,
      "grad_norm": 0.27043792605400085,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 29950
    },
    {
      "epoch": 26.532329495128433,
      "grad_norm": 0.24039876461029053,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 29955
    },
    {
      "epoch": 26.53675819309123,
      "grad_norm": 0.23946158587932587,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 29960
    },
    {
      "epoch": 26.54118689105403,
      "grad_norm": 0.25826382637023926,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 29965
    },
    {
      "epoch": 26.54561558901683,
      "grad_norm": 0.2664659023284912,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 29970
    },
    {
      "epoch": 26.550044286979627,
      "grad_norm": 0.22836853563785553,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 29975
    },
    {
      "epoch": 26.55447298494243,
      "grad_norm": 0.2150292694568634,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 29980
    },
    {
      "epoch": 26.558901682905226,
      "grad_norm": 0.28751879930496216,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 29985
    },
    {
      "epoch": 26.563330380868024,
      "grad_norm": 0.26578569412231445,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 29990
    },
    {
      "epoch": 26.567759078830825,
      "grad_norm": 0.2752525806427002,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 29995
    },
    {
      "epoch": 26.572187776793623,
      "grad_norm": 0.20902705192565918,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 30000
    },
    {
      "epoch": 26.57661647475642,
      "grad_norm": 0.247431218624115,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 30005
    },
    {
      "epoch": 26.58104517271922,
      "grad_norm": 0.2535596787929535,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 30010
    },
    {
      "epoch": 26.58547387068202,
      "grad_norm": 0.30459001660346985,
      "learning_rate": 1e-05,
      "loss": 0.9239,
      "step": 30015
    },
    {
      "epoch": 26.589902568644817,
      "grad_norm": 0.24338839948177338,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 30020
    },
    {
      "epoch": 26.594331266607618,
      "grad_norm": 0.21212758123874664,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 30025
    },
    {
      "epoch": 26.598759964570416,
      "grad_norm": 0.27010592818260193,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 30030
    },
    {
      "epoch": 26.603188662533213,
      "grad_norm": 0.226255863904953,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 30035
    },
    {
      "epoch": 26.607617360496015,
      "grad_norm": 0.22579582035541534,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 30040
    },
    {
      "epoch": 26.612046058458812,
      "grad_norm": 0.23832368850708008,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 30045
    },
    {
      "epoch": 26.616474756421614,
      "grad_norm": 0.24104498326778412,
      "learning_rate": 1e-05,
      "loss": 1.0171,
      "step": 30050
    },
    {
      "epoch": 26.62090345438441,
      "grad_norm": 0.25107672810554504,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 30055
    },
    {
      "epoch": 26.62533215234721,
      "grad_norm": 0.24267186224460602,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 30060
    },
    {
      "epoch": 26.62976085031001,
      "grad_norm": 0.24306286871433258,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 30065
    },
    {
      "epoch": 26.634189548272808,
      "grad_norm": 0.22851808369159698,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 30070
    },
    {
      "epoch": 26.638618246235605,
      "grad_norm": 0.2540264427661896,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 30075
    },
    {
      "epoch": 26.643046944198407,
      "grad_norm": 0.23075030744075775,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 30080
    },
    {
      "epoch": 26.647475642161204,
      "grad_norm": 0.25420352816581726,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 30085
    },
    {
      "epoch": 26.651904340124002,
      "grad_norm": 0.23552055656909943,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 30090
    },
    {
      "epoch": 26.656333038086803,
      "grad_norm": 0.27110713720321655,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 30095
    },
    {
      "epoch": 26.6607617360496,
      "grad_norm": 0.3211550712585449,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 30100
    },
    {
      "epoch": 26.665190434012402,
      "grad_norm": 0.28638020157814026,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 30105
    },
    {
      "epoch": 26.6696191319752,
      "grad_norm": 0.2437484860420227,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 30110
    },
    {
      "epoch": 26.674047829937997,
      "grad_norm": 0.23589135706424713,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 30115
    },
    {
      "epoch": 26.6784765279008,
      "grad_norm": 0.2294664829969406,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 30120
    },
    {
      "epoch": 26.682905225863596,
      "grad_norm": 0.2592249810695648,
      "learning_rate": 1e-05,
      "loss": 1.0193,
      "step": 30125
    },
    {
      "epoch": 26.687333923826394,
      "grad_norm": 0.2469903528690338,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 30130
    },
    {
      "epoch": 26.691762621789195,
      "grad_norm": 0.23921608924865723,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 30135
    },
    {
      "epoch": 26.696191319751993,
      "grad_norm": 0.24063198268413544,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 30140
    },
    {
      "epoch": 26.70062001771479,
      "grad_norm": 0.2603273093700409,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 30145
    },
    {
      "epoch": 26.70504871567759,
      "grad_norm": 0.2889840304851532,
      "learning_rate": 1e-05,
      "loss": 1.0324,
      "step": 30150
    },
    {
      "epoch": 26.70947741364039,
      "grad_norm": 0.2850040793418884,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 30155
    },
    {
      "epoch": 26.713906111603187,
      "grad_norm": 0.2411225140094757,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 30160
    },
    {
      "epoch": 26.718334809565988,
      "grad_norm": 0.2227754145860672,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 30165
    },
    {
      "epoch": 26.722763507528786,
      "grad_norm": 0.2847329080104828,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 30170
    },
    {
      "epoch": 26.727192205491587,
      "grad_norm": 0.2909183204174042,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 30175
    },
    {
      "epoch": 26.731620903454385,
      "grad_norm": 0.23366843163967133,
      "learning_rate": 1e-05,
      "loss": 1.0485,
      "step": 30180
    },
    {
      "epoch": 26.736049601417182,
      "grad_norm": 0.25701847672462463,
      "learning_rate": 1e-05,
      "loss": 1.0325,
      "step": 30185
    },
    {
      "epoch": 26.740478299379983,
      "grad_norm": 0.241773322224617,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 30190
    },
    {
      "epoch": 26.74490699734278,
      "grad_norm": 0.26165956258773804,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 30195
    },
    {
      "epoch": 26.74933569530558,
      "grad_norm": 0.2630947232246399,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 30200
    },
    {
      "epoch": 26.75376439326838,
      "grad_norm": 0.30287283658981323,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 30205
    },
    {
      "epoch": 26.758193091231178,
      "grad_norm": 0.2831924855709076,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 30210
    },
    {
      "epoch": 26.762621789193975,
      "grad_norm": 0.2565245032310486,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 30215
    },
    {
      "epoch": 26.767050487156776,
      "grad_norm": 0.22565393149852753,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 30220
    },
    {
      "epoch": 26.771479185119574,
      "grad_norm": 0.29895102977752686,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 30225
    },
    {
      "epoch": 26.775907883082375,
      "grad_norm": 0.2604537904262543,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 30230
    },
    {
      "epoch": 26.780336581045173,
      "grad_norm": 0.30843308568000793,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 30235
    },
    {
      "epoch": 26.78476527900797,
      "grad_norm": 0.3403940200805664,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 30240
    },
    {
      "epoch": 26.789193976970772,
      "grad_norm": 0.21627341210842133,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 30245
    },
    {
      "epoch": 26.79362267493357,
      "grad_norm": 0.2071480005979538,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 30250
    },
    {
      "epoch": 26.798051372896367,
      "grad_norm": 0.24159276485443115,
      "learning_rate": 1e-05,
      "loss": 1.0304,
      "step": 30255
    },
    {
      "epoch": 26.80248007085917,
      "grad_norm": 0.1980116367340088,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 30260
    },
    {
      "epoch": 26.806908768821966,
      "grad_norm": 0.21964040398597717,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 30265
    },
    {
      "epoch": 26.811337466784764,
      "grad_norm": 0.2412816286087036,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 30270
    },
    {
      "epoch": 26.815766164747565,
      "grad_norm": 0.261136531829834,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 30275
    },
    {
      "epoch": 26.820194862710363,
      "grad_norm": 0.23315303027629852,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 30280
    },
    {
      "epoch": 26.824623560673164,
      "grad_norm": 0.21400538086891174,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 30285
    },
    {
      "epoch": 26.82905225863596,
      "grad_norm": 0.2653231620788574,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 30290
    },
    {
      "epoch": 26.83348095659876,
      "grad_norm": 0.23129649460315704,
      "learning_rate": 1e-05,
      "loss": 1.0217,
      "step": 30295
    },
    {
      "epoch": 26.83790965456156,
      "grad_norm": 0.39349043369293213,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 30300
    },
    {
      "epoch": 26.842338352524358,
      "grad_norm": 0.21276351809501648,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 30305
    },
    {
      "epoch": 26.846767050487156,
      "grad_norm": 0.2156832367181778,
      "learning_rate": 1e-05,
      "loss": 1.0458,
      "step": 30310
    },
    {
      "epoch": 26.851195748449957,
      "grad_norm": 0.2102944254875183,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 30315
    },
    {
      "epoch": 26.855624446412754,
      "grad_norm": 0.30699586868286133,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 30320
    },
    {
      "epoch": 26.860053144375552,
      "grad_norm": 0.24376645684242249,
      "learning_rate": 1e-05,
      "loss": 1.053,
      "step": 30325
    },
    {
      "epoch": 26.864481842338353,
      "grad_norm": 0.23732689023017883,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 30330
    },
    {
      "epoch": 26.86891054030115,
      "grad_norm": 0.24358725547790527,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 30335
    },
    {
      "epoch": 26.873339238263952,
      "grad_norm": 0.22225724160671234,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 30340
    },
    {
      "epoch": 26.87776793622675,
      "grad_norm": 0.3010594844818115,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 30345
    },
    {
      "epoch": 26.882196634189548,
      "grad_norm": 0.21877768635749817,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 30350
    },
    {
      "epoch": 26.88662533215235,
      "grad_norm": 0.29026171565055847,
      "learning_rate": 1e-05,
      "loss": 1.0364,
      "step": 30355
    },
    {
      "epoch": 26.891054030115146,
      "grad_norm": 0.25234198570251465,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 30360
    },
    {
      "epoch": 26.895482728077944,
      "grad_norm": 0.22945880889892578,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 30365
    },
    {
      "epoch": 26.899911426040745,
      "grad_norm": 0.25953271985054016,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 30370
    },
    {
      "epoch": 26.904340124003543,
      "grad_norm": 0.27650877833366394,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 30375
    },
    {
      "epoch": 26.90876882196634,
      "grad_norm": 0.2854640483856201,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 30380
    },
    {
      "epoch": 26.913197519929142,
      "grad_norm": 0.3169996440410614,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 30385
    },
    {
      "epoch": 26.91762621789194,
      "grad_norm": 0.20262137055397034,
      "learning_rate": 1e-05,
      "loss": 1.0222,
      "step": 30390
    },
    {
      "epoch": 26.922054915854737,
      "grad_norm": 0.2926405370235443,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 30395
    },
    {
      "epoch": 26.92648361381754,
      "grad_norm": 0.24764838814735413,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 30400
    },
    {
      "epoch": 26.930912311780336,
      "grad_norm": 0.2266751080751419,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 30405
    },
    {
      "epoch": 26.935341009743137,
      "grad_norm": 0.2836935818195343,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 30410
    },
    {
      "epoch": 26.939769707705935,
      "grad_norm": 0.26348793506622314,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 30415
    },
    {
      "epoch": 26.944198405668732,
      "grad_norm": 0.25096118450164795,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 30420
    },
    {
      "epoch": 26.948627103631534,
      "grad_norm": 0.24840302765369415,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 30425
    },
    {
      "epoch": 26.95305580159433,
      "grad_norm": 0.25247102975845337,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 30430
    },
    {
      "epoch": 26.95748449955713,
      "grad_norm": 0.23725663125514984,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 30435
    },
    {
      "epoch": 26.96191319751993,
      "grad_norm": 0.2660263478755951,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 30440
    },
    {
      "epoch": 26.966341895482728,
      "grad_norm": 0.24798409640789032,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 30445
    },
    {
      "epoch": 26.970770593445526,
      "grad_norm": 0.2632289528846741,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 30450
    },
    {
      "epoch": 26.975199291408327,
      "grad_norm": 0.2203081101179123,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 30455
    },
    {
      "epoch": 26.979627989371124,
      "grad_norm": 0.25215771794319153,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 30460
    },
    {
      "epoch": 26.984056687333926,
      "grad_norm": 0.25819385051727295,
      "learning_rate": 1e-05,
      "loss": 1.0196,
      "step": 30465
    },
    {
      "epoch": 26.988485385296723,
      "grad_norm": 0.24728737771511078,
      "learning_rate": 1e-05,
      "loss": 1.0101,
      "step": 30470
    },
    {
      "epoch": 26.99291408325952,
      "grad_norm": 0.21859033405780792,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 30475
    },
    {
      "epoch": 26.997342781222322,
      "grad_norm": 0.2844151556491852,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 30480
    },
    {
      "epoch": 27.00177147918512,
      "grad_norm": 0.3071330487728119,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 30485
    },
    {
      "epoch": 27.006200177147917,
      "grad_norm": 0.23135030269622803,
      "learning_rate": 1e-05,
      "loss": 1.0283,
      "step": 30490
    },
    {
      "epoch": 27.01062887511072,
      "grad_norm": 0.22536525130271912,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 30495
    },
    {
      "epoch": 27.015057573073516,
      "grad_norm": 0.24716205894947052,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 30500
    },
    {
      "epoch": 27.019486271036314,
      "grad_norm": 0.2711803615093231,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 30505
    },
    {
      "epoch": 27.023914968999115,
      "grad_norm": 0.21956685185432434,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 30510
    },
    {
      "epoch": 27.028343666961913,
      "grad_norm": 0.24644488096237183,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 30515
    },
    {
      "epoch": 27.03277236492471,
      "grad_norm": 0.21150024235248566,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 30520
    },
    {
      "epoch": 27.03720106288751,
      "grad_norm": 0.288254052400589,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 30525
    },
    {
      "epoch": 27.04162976085031,
      "grad_norm": 0.22987912595272064,
      "learning_rate": 1e-05,
      "loss": 1.0402,
      "step": 30530
    },
    {
      "epoch": 27.04605845881311,
      "grad_norm": 0.22410458326339722,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 30535
    },
    {
      "epoch": 27.050487156775908,
      "grad_norm": 0.2752321660518646,
      "learning_rate": 1e-05,
      "loss": 1.0426,
      "step": 30540
    },
    {
      "epoch": 27.054915854738706,
      "grad_norm": 0.25184276700019836,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 30545
    },
    {
      "epoch": 27.059344552701507,
      "grad_norm": 0.26193737983703613,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 30550
    },
    {
      "epoch": 27.063773250664305,
      "grad_norm": 0.2575017809867859,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 30555
    },
    {
      "epoch": 27.068201948627102,
      "grad_norm": 0.24047674238681793,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 30560
    },
    {
      "epoch": 27.072630646589904,
      "grad_norm": 0.3317027688026428,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 30565
    },
    {
      "epoch": 27.0770593445527,
      "grad_norm": 0.2667787969112396,
      "learning_rate": 1e-05,
      "loss": 1.0454,
      "step": 30570
    },
    {
      "epoch": 27.0814880425155,
      "grad_norm": 0.24323886632919312,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 30575
    },
    {
      "epoch": 27.0859167404783,
      "grad_norm": 0.20749755203723907,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 30580
    },
    {
      "epoch": 27.090345438441098,
      "grad_norm": 0.2156824767589569,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 30585
    },
    {
      "epoch": 27.0947741364039,
      "grad_norm": 0.27501180768013,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 30590
    },
    {
      "epoch": 27.099202834366697,
      "grad_norm": 0.23036827147006989,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 30595
    },
    {
      "epoch": 27.103631532329494,
      "grad_norm": 0.23696354031562805,
      "learning_rate": 1e-05,
      "loss": 1.0656,
      "step": 30600
    },
    {
      "epoch": 27.108060230292296,
      "grad_norm": 0.23960939049720764,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 30605
    },
    {
      "epoch": 27.112488928255093,
      "grad_norm": 0.24240179359912872,
      "learning_rate": 1e-05,
      "loss": 1.0534,
      "step": 30610
    },
    {
      "epoch": 27.11691762621789,
      "grad_norm": 0.21522845327854156,
      "learning_rate": 1e-05,
      "loss": 1.048,
      "step": 30615
    },
    {
      "epoch": 27.121346324180692,
      "grad_norm": 0.24755887687206268,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 30620
    },
    {
      "epoch": 27.12577502214349,
      "grad_norm": 0.2615501880645752,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 30625
    },
    {
      "epoch": 27.130203720106287,
      "grad_norm": 0.2601321339607239,
      "learning_rate": 1e-05,
      "loss": 1.0272,
      "step": 30630
    },
    {
      "epoch": 27.13463241806909,
      "grad_norm": 0.27466320991516113,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 30635
    },
    {
      "epoch": 27.139061116031886,
      "grad_norm": 0.23908357322216034,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 30640
    },
    {
      "epoch": 27.143489813994684,
      "grad_norm": 0.28776228427886963,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 30645
    },
    {
      "epoch": 27.147918511957485,
      "grad_norm": 0.20808722078800201,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 30650
    },
    {
      "epoch": 27.152347209920283,
      "grad_norm": 0.24358826875686646,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 30655
    },
    {
      "epoch": 27.156775907883084,
      "grad_norm": 0.18493781983852386,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 30660
    },
    {
      "epoch": 27.16120460584588,
      "grad_norm": 0.23946912586688995,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 30665
    },
    {
      "epoch": 27.16563330380868,
      "grad_norm": 0.25911280512809753,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 30670
    },
    {
      "epoch": 27.17006200177148,
      "grad_norm": 0.21719790995121002,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 30675
    },
    {
      "epoch": 27.174490699734278,
      "grad_norm": 0.22320930659770966,
      "learning_rate": 1e-05,
      "loss": 1.0539,
      "step": 30680
    },
    {
      "epoch": 27.178919397697076,
      "grad_norm": 0.22360873222351074,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 30685
    },
    {
      "epoch": 27.183348095659877,
      "grad_norm": 0.24945010244846344,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 30690
    },
    {
      "epoch": 27.187776793622675,
      "grad_norm": 0.2389744222164154,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 30695
    },
    {
      "epoch": 27.192205491585472,
      "grad_norm": 0.22741013765335083,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 30700
    },
    {
      "epoch": 27.196634189548273,
      "grad_norm": 0.23818716406822205,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 30705
    },
    {
      "epoch": 27.20106288751107,
      "grad_norm": 0.23088152706623077,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 30710
    },
    {
      "epoch": 27.205491585473872,
      "grad_norm": 0.20602001249790192,
      "learning_rate": 1e-05,
      "loss": 1.0543,
      "step": 30715
    },
    {
      "epoch": 27.20992028343667,
      "grad_norm": 0.2460908591747284,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 30720
    },
    {
      "epoch": 27.214348981399468,
      "grad_norm": 0.22483669221401215,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 30725
    },
    {
      "epoch": 27.21877767936227,
      "grad_norm": 0.2405012995004654,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 30730
    },
    {
      "epoch": 27.223206377325067,
      "grad_norm": 0.2231002002954483,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 30735
    },
    {
      "epoch": 27.227635075287864,
      "grad_norm": 0.2702237367630005,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 30740
    },
    {
      "epoch": 27.232063773250665,
      "grad_norm": 0.27780887484550476,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 30745
    },
    {
      "epoch": 27.236492471213463,
      "grad_norm": 0.2398044317960739,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 30750
    },
    {
      "epoch": 27.24092116917626,
      "grad_norm": 0.21106013655662537,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 30755
    },
    {
      "epoch": 27.245349867139062,
      "grad_norm": 0.2806750237941742,
      "learning_rate": 1e-05,
      "loss": 1.0571,
      "step": 30760
    },
    {
      "epoch": 27.24977856510186,
      "grad_norm": 0.3151966333389282,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 30765
    },
    {
      "epoch": 27.254207263064657,
      "grad_norm": 0.23356717824935913,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 30770
    },
    {
      "epoch": 27.25863596102746,
      "grad_norm": 0.26284605264663696,
      "learning_rate": 1e-05,
      "loss": 1.0577,
      "step": 30775
    },
    {
      "epoch": 27.263064658990256,
      "grad_norm": 0.2558484673500061,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 30780
    },
    {
      "epoch": 27.267493356953057,
      "grad_norm": 0.239601731300354,
      "learning_rate": 1e-05,
      "loss": 1.0183,
      "step": 30785
    },
    {
      "epoch": 27.271922054915855,
      "grad_norm": 0.2502806484699249,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 30790
    },
    {
      "epoch": 27.276350752878653,
      "grad_norm": 0.22724245488643646,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 30795
    },
    {
      "epoch": 27.280779450841454,
      "grad_norm": 0.22719928622245789,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 30800
    },
    {
      "epoch": 27.28520814880425,
      "grad_norm": 0.2547604739665985,
      "learning_rate": 1e-05,
      "loss": 1.0107,
      "step": 30805
    },
    {
      "epoch": 27.28963684676705,
      "grad_norm": 0.23706574738025665,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 30810
    },
    {
      "epoch": 27.29406554472985,
      "grad_norm": 0.2269642949104309,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 30815
    },
    {
      "epoch": 27.298494242692648,
      "grad_norm": 0.21294556558132172,
      "learning_rate": 1e-05,
      "loss": 1.0224,
      "step": 30820
    },
    {
      "epoch": 27.302922940655446,
      "grad_norm": 0.23332804441452026,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 30825
    },
    {
      "epoch": 27.307351638618247,
      "grad_norm": 0.2975403964519501,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 30830
    },
    {
      "epoch": 27.311780336581045,
      "grad_norm": 0.22429126501083374,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 30835
    },
    {
      "epoch": 27.316209034543846,
      "grad_norm": 0.2342357337474823,
      "learning_rate": 1e-05,
      "loss": 0.9299,
      "step": 30840
    },
    {
      "epoch": 27.320637732506643,
      "grad_norm": 0.2459591180086136,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 30845
    },
    {
      "epoch": 27.32506643046944,
      "grad_norm": 0.21968969702720642,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 30850
    },
    {
      "epoch": 27.329495128432242,
      "grad_norm": 0.25242751836776733,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 30855
    },
    {
      "epoch": 27.33392382639504,
      "grad_norm": 0.29921382665634155,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 30860
    },
    {
      "epoch": 27.338352524357838,
      "grad_norm": 0.26967331767082214,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 30865
    },
    {
      "epoch": 27.34278122232064,
      "grad_norm": 0.29027464985847473,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 30870
    },
    {
      "epoch": 27.347209920283436,
      "grad_norm": 0.2600877285003662,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 30875
    },
    {
      "epoch": 27.351638618246234,
      "grad_norm": 0.24870745837688446,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 30880
    },
    {
      "epoch": 27.356067316209035,
      "grad_norm": 0.2901571989059448,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 30885
    },
    {
      "epoch": 27.360496014171833,
      "grad_norm": 0.2893698513507843,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 30890
    },
    {
      "epoch": 27.36492471213463,
      "grad_norm": 0.2636055052280426,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 30895
    },
    {
      "epoch": 27.369353410097432,
      "grad_norm": 0.2832021117210388,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 30900
    },
    {
      "epoch": 27.37378210806023,
      "grad_norm": 0.23784591257572174,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 30905
    },
    {
      "epoch": 27.37821080602303,
      "grad_norm": 0.23933926224708557,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 30910
    },
    {
      "epoch": 27.38263950398583,
      "grad_norm": 0.20947566628456116,
      "learning_rate": 1e-05,
      "loss": 1.0974,
      "step": 30915
    },
    {
      "epoch": 27.387068201948626,
      "grad_norm": 0.23381993174552917,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 30920
    },
    {
      "epoch": 27.391496899911427,
      "grad_norm": 0.24552865326404572,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 30925
    },
    {
      "epoch": 27.395925597874225,
      "grad_norm": 0.24369806051254272,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 30930
    },
    {
      "epoch": 27.400354295837023,
      "grad_norm": 0.2227746844291687,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 30935
    },
    {
      "epoch": 27.404782993799824,
      "grad_norm": 0.2737066149711609,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 30940
    },
    {
      "epoch": 27.40921169176262,
      "grad_norm": 0.2747707962989807,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 30945
    },
    {
      "epoch": 27.41364038972542,
      "grad_norm": 0.1957477331161499,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 30950
    },
    {
      "epoch": 27.41806908768822,
      "grad_norm": 0.2482810616493225,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 30955
    },
    {
      "epoch": 27.422497785651018,
      "grad_norm": 0.24321091175079346,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 30960
    },
    {
      "epoch": 27.42692648361382,
      "grad_norm": 0.2579958438873291,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 30965
    },
    {
      "epoch": 27.431355181576617,
      "grad_norm": 0.20409387350082397,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 30970
    },
    {
      "epoch": 27.435783879539414,
      "grad_norm": 0.23095521330833435,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 30975
    },
    {
      "epoch": 27.440212577502216,
      "grad_norm": 0.22307327389717102,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 30980
    },
    {
      "epoch": 27.444641275465013,
      "grad_norm": 0.23792791366577148,
      "learning_rate": 1e-05,
      "loss": 0.9097,
      "step": 30985
    },
    {
      "epoch": 27.44906997342781,
      "grad_norm": 0.23176367580890656,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 30990
    },
    {
      "epoch": 27.453498671390612,
      "grad_norm": 0.2149556428194046,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 30995
    },
    {
      "epoch": 27.45792736935341,
      "grad_norm": 0.21995078027248383,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 31000
    },
    {
      "epoch": 27.462356067316207,
      "grad_norm": 0.2358456552028656,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 31005
    },
    {
      "epoch": 27.46678476527901,
      "grad_norm": 0.25158458948135376,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 31010
    },
    {
      "epoch": 27.471213463241806,
      "grad_norm": 0.23530571162700653,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 31015
    },
    {
      "epoch": 27.475642161204608,
      "grad_norm": 0.24893873929977417,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 31020
    },
    {
      "epoch": 27.480070859167405,
      "grad_norm": 0.23119038343429565,
      "learning_rate": 1e-05,
      "loss": 0.9312,
      "step": 31025
    },
    {
      "epoch": 27.484499557130203,
      "grad_norm": 0.21887554228305817,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 31030
    },
    {
      "epoch": 27.488928255093004,
      "grad_norm": 0.258976012468338,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 31035
    },
    {
      "epoch": 27.4933569530558,
      "grad_norm": 0.2321832776069641,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 31040
    },
    {
      "epoch": 27.4977856510186,
      "grad_norm": 0.22107261419296265,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 31045
    },
    {
      "epoch": 27.5022143489814,
      "grad_norm": 0.2509700357913971,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 31050
    },
    {
      "epoch": 27.5066430469442,
      "grad_norm": 0.31972596049308777,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 31055
    },
    {
      "epoch": 27.511071744906996,
      "grad_norm": 0.2478359490633011,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 31060
    },
    {
      "epoch": 27.515500442869797,
      "grad_norm": 0.24427036941051483,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 31065
    },
    {
      "epoch": 27.519929140832595,
      "grad_norm": 0.2811807096004486,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 31070
    },
    {
      "epoch": 27.524357838795392,
      "grad_norm": 0.23756143450737,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 31075
    },
    {
      "epoch": 27.528786536758194,
      "grad_norm": 0.2270578145980835,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 31080
    },
    {
      "epoch": 27.53321523472099,
      "grad_norm": 0.24863387644290924,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 31085
    },
    {
      "epoch": 27.537643932683793,
      "grad_norm": 0.23438431322574615,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 31090
    },
    {
      "epoch": 27.54207263064659,
      "grad_norm": 0.29311737418174744,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 31095
    },
    {
      "epoch": 27.546501328609388,
      "grad_norm": 0.23446708917617798,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 31100
    },
    {
      "epoch": 27.55093002657219,
      "grad_norm": 0.23869706690311432,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 31105
    },
    {
      "epoch": 27.555358724534987,
      "grad_norm": 0.2336687445640564,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 31110
    },
    {
      "epoch": 27.559787422497784,
      "grad_norm": 0.28089210391044617,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 31115
    },
    {
      "epoch": 27.564216120460586,
      "grad_norm": 0.23779523372650146,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 31120
    },
    {
      "epoch": 27.568644818423383,
      "grad_norm": 0.22791455686092377,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 31125
    },
    {
      "epoch": 27.57307351638618,
      "grad_norm": 0.26404720544815063,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 31130
    },
    {
      "epoch": 27.577502214348982,
      "grad_norm": 0.30810046195983887,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 31135
    },
    {
      "epoch": 27.58193091231178,
      "grad_norm": 0.2704094648361206,
      "learning_rate": 1e-05,
      "loss": 0.9306,
      "step": 31140
    },
    {
      "epoch": 27.58635961027458,
      "grad_norm": 0.2914482355117798,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 31145
    },
    {
      "epoch": 27.59078830823738,
      "grad_norm": 0.23683074116706848,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 31150
    },
    {
      "epoch": 27.595217006200176,
      "grad_norm": 0.22593069076538086,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 31155
    },
    {
      "epoch": 27.599645704162977,
      "grad_norm": 0.23497292399406433,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 31160
    },
    {
      "epoch": 27.604074402125775,
      "grad_norm": 0.22512322664260864,
      "learning_rate": 1e-05,
      "loss": 1.0225,
      "step": 31165
    },
    {
      "epoch": 27.608503100088573,
      "grad_norm": 0.22328370809555054,
      "learning_rate": 1e-05,
      "loss": 1.0325,
      "step": 31170
    },
    {
      "epoch": 27.612931798051374,
      "grad_norm": 0.2554346024990082,
      "learning_rate": 1e-05,
      "loss": 1.0485,
      "step": 31175
    },
    {
      "epoch": 27.61736049601417,
      "grad_norm": 0.23263664543628693,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 31180
    },
    {
      "epoch": 27.62178919397697,
      "grad_norm": 0.19805167615413666,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 31185
    },
    {
      "epoch": 27.62621789193977,
      "grad_norm": 0.2554255723953247,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 31190
    },
    {
      "epoch": 27.630646589902568,
      "grad_norm": 0.24750465154647827,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 31195
    },
    {
      "epoch": 27.63507528786537,
      "grad_norm": 0.23358836770057678,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 31200
    },
    {
      "epoch": 27.639503985828167,
      "grad_norm": 0.23879177868366241,
      "learning_rate": 1e-05,
      "loss": 0.9303,
      "step": 31205
    },
    {
      "epoch": 27.643932683790965,
      "grad_norm": 0.26125022768974304,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 31210
    },
    {
      "epoch": 27.648361381753766,
      "grad_norm": 0.2203727662563324,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 31215
    },
    {
      "epoch": 27.652790079716564,
      "grad_norm": 0.2464284747838974,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 31220
    },
    {
      "epoch": 27.65721877767936,
      "grad_norm": 0.22900137305259705,
      "learning_rate": 1e-05,
      "loss": 1.0085,
      "step": 31225
    },
    {
      "epoch": 27.661647475642162,
      "grad_norm": 0.2325325459241867,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 31230
    },
    {
      "epoch": 27.66607617360496,
      "grad_norm": 0.2320452481508255,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 31235
    },
    {
      "epoch": 27.670504871567758,
      "grad_norm": 0.23795297741889954,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 31240
    },
    {
      "epoch": 27.67493356953056,
      "grad_norm": 0.2289896011352539,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 31245
    },
    {
      "epoch": 27.679362267493357,
      "grad_norm": 0.20696084201335907,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 31250
    },
    {
      "epoch": 27.683790965456154,
      "grad_norm": 0.2510433495044708,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 31255
    },
    {
      "epoch": 27.688219663418955,
      "grad_norm": 0.2481033354997635,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 31260
    },
    {
      "epoch": 27.692648361381753,
      "grad_norm": 0.2543386220932007,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 31265
    },
    {
      "epoch": 27.697077059344554,
      "grad_norm": 0.25980255007743835,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 31270
    },
    {
      "epoch": 27.701505757307352,
      "grad_norm": 0.2612070143222809,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 31275
    },
    {
      "epoch": 27.70593445527015,
      "grad_norm": 0.23598235845565796,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 31280
    },
    {
      "epoch": 27.71036315323295,
      "grad_norm": 0.2432212233543396,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 31285
    },
    {
      "epoch": 27.71479185119575,
      "grad_norm": 0.37720274925231934,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 31290
    },
    {
      "epoch": 27.719220549158546,
      "grad_norm": 0.276174396276474,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 31295
    },
    {
      "epoch": 27.723649247121347,
      "grad_norm": 0.2447313517332077,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 31300
    },
    {
      "epoch": 27.728077945084145,
      "grad_norm": 0.2556127607822418,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 31305
    },
    {
      "epoch": 27.732506643046943,
      "grad_norm": 0.28501713275909424,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 31310
    },
    {
      "epoch": 27.736935341009744,
      "grad_norm": 0.25870856642723083,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 31315
    },
    {
      "epoch": 27.74136403897254,
      "grad_norm": 0.21241873502731323,
      "learning_rate": 1e-05,
      "loss": 0.9219,
      "step": 31320
    },
    {
      "epoch": 27.745792736935343,
      "grad_norm": 0.25511202216148376,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 31325
    },
    {
      "epoch": 27.75022143489814,
      "grad_norm": 0.29273536801338196,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 31330
    },
    {
      "epoch": 27.754650132860938,
      "grad_norm": 0.23387330770492554,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 31335
    },
    {
      "epoch": 27.75907883082374,
      "grad_norm": 0.21892881393432617,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 31340
    },
    {
      "epoch": 27.763507528786537,
      "grad_norm": 0.2555164694786072,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 31345
    },
    {
      "epoch": 27.767936226749335,
      "grad_norm": 0.23252488672733307,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 31350
    },
    {
      "epoch": 27.772364924712136,
      "grad_norm": 0.24156518280506134,
      "learning_rate": 1e-05,
      "loss": 1.0252,
      "step": 31355
    },
    {
      "epoch": 27.776793622674933,
      "grad_norm": 0.20946422219276428,
      "learning_rate": 1e-05,
      "loss": 1.0392,
      "step": 31360
    },
    {
      "epoch": 27.78122232063773,
      "grad_norm": 0.24066127836704254,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 31365
    },
    {
      "epoch": 27.785651018600532,
      "grad_norm": 0.2086721807718277,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 31370
    },
    {
      "epoch": 27.79007971656333,
      "grad_norm": 0.2736165225505829,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 31375
    },
    {
      "epoch": 27.794508414526128,
      "grad_norm": 0.24205589294433594,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 31380
    },
    {
      "epoch": 27.79893711248893,
      "grad_norm": 0.23163186013698578,
      "learning_rate": 1e-05,
      "loss": 1.017,
      "step": 31385
    },
    {
      "epoch": 27.803365810451727,
      "grad_norm": 0.24665170907974243,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 31390
    },
    {
      "epoch": 27.807794508414528,
      "grad_norm": 0.22288623452186584,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 31395
    },
    {
      "epoch": 27.812223206377325,
      "grad_norm": 0.2558608949184418,
      "learning_rate": 1e-05,
      "loss": 1.0295,
      "step": 31400
    },
    {
      "epoch": 27.816651904340123,
      "grad_norm": 0.22187495231628418,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 31405
    },
    {
      "epoch": 27.821080602302924,
      "grad_norm": 0.2203754037618637,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 31410
    },
    {
      "epoch": 27.825509300265722,
      "grad_norm": 0.22384601831436157,
      "learning_rate": 1e-05,
      "loss": 1.0548,
      "step": 31415
    },
    {
      "epoch": 27.82993799822852,
      "grad_norm": 0.26528772711753845,
      "learning_rate": 1e-05,
      "loss": 1.0441,
      "step": 31420
    },
    {
      "epoch": 27.83436669619132,
      "grad_norm": 0.2644975185394287,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 31425
    },
    {
      "epoch": 27.83879539415412,
      "grad_norm": 0.3055729866027832,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 31430
    },
    {
      "epoch": 27.843224092116916,
      "grad_norm": 0.2652888894081116,
      "learning_rate": 1e-05,
      "loss": 1.0467,
      "step": 31435
    },
    {
      "epoch": 27.847652790079717,
      "grad_norm": 0.21909134089946747,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 31440
    },
    {
      "epoch": 27.852081488042515,
      "grad_norm": 0.29096031188964844,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 31445
    },
    {
      "epoch": 27.856510186005316,
      "grad_norm": 0.25493115186691284,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 31450
    },
    {
      "epoch": 27.860938883968114,
      "grad_norm": 0.2630530893802643,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 31455
    },
    {
      "epoch": 27.86536758193091,
      "grad_norm": 0.250218003988266,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 31460
    },
    {
      "epoch": 27.869796279893713,
      "grad_norm": 0.2662123441696167,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 31465
    },
    {
      "epoch": 27.87422497785651,
      "grad_norm": 0.21547740697860718,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 31470
    },
    {
      "epoch": 27.878653675819308,
      "grad_norm": 0.2514686584472656,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 31475
    },
    {
      "epoch": 27.88308237378211,
      "grad_norm": 0.2314177006483078,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 31480
    },
    {
      "epoch": 27.887511071744907,
      "grad_norm": 0.23453478515148163,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 31485
    },
    {
      "epoch": 27.891939769707704,
      "grad_norm": 0.2426118701696396,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 31490
    },
    {
      "epoch": 27.896368467670506,
      "grad_norm": 0.3054236173629761,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 31495
    },
    {
      "epoch": 27.900797165633303,
      "grad_norm": 0.25926703214645386,
      "learning_rate": 1e-05,
      "loss": 0.9242,
      "step": 31500
    },
    {
      "epoch": 27.9052258635961,
      "grad_norm": 0.26455721259117126,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 31505
    },
    {
      "epoch": 27.909654561558902,
      "grad_norm": 0.27374276518821716,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 31510
    },
    {
      "epoch": 27.9140832595217,
      "grad_norm": 0.26286399364471436,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 31515
    },
    {
      "epoch": 27.9185119574845,
      "grad_norm": 0.25271520018577576,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 31520
    },
    {
      "epoch": 27.9229406554473,
      "grad_norm": 0.2456914186477661,
      "learning_rate": 1e-05,
      "loss": 1.0598,
      "step": 31525
    },
    {
      "epoch": 27.927369353410096,
      "grad_norm": 0.2313370555639267,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 31530
    },
    {
      "epoch": 27.931798051372898,
      "grad_norm": 0.23602987825870514,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 31535
    },
    {
      "epoch": 27.936226749335695,
      "grad_norm": 0.2285691648721695,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 31540
    },
    {
      "epoch": 27.940655447298493,
      "grad_norm": 0.2799905240535736,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 31545
    },
    {
      "epoch": 27.945084145261294,
      "grad_norm": 0.20500244200229645,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 31550
    },
    {
      "epoch": 27.949512843224092,
      "grad_norm": 0.24910742044448853,
      "learning_rate": 1e-05,
      "loss": 1.0392,
      "step": 31555
    },
    {
      "epoch": 27.95394154118689,
      "grad_norm": 0.2464452087879181,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 31560
    },
    {
      "epoch": 27.95837023914969,
      "grad_norm": 0.25134047865867615,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 31565
    },
    {
      "epoch": 27.96279893711249,
      "grad_norm": 0.2531101107597351,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 31570
    },
    {
      "epoch": 27.96722763507529,
      "grad_norm": 0.23719653487205505,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 31575
    },
    {
      "epoch": 27.971656333038087,
      "grad_norm": 0.20806440711021423,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 31580
    },
    {
      "epoch": 27.976085031000885,
      "grad_norm": 0.28420737385749817,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 31585
    },
    {
      "epoch": 27.980513728963686,
      "grad_norm": 0.22390660643577576,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 31590
    },
    {
      "epoch": 27.984942426926484,
      "grad_norm": 0.25261765718460083,
      "learning_rate": 1e-05,
      "loss": 0.9028,
      "step": 31595
    },
    {
      "epoch": 27.98937112488928,
      "grad_norm": 0.23809413611888885,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 31600
    },
    {
      "epoch": 27.993799822852083,
      "grad_norm": 0.22561772167682648,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 31605
    },
    {
      "epoch": 27.99822852081488,
      "grad_norm": 0.25934693217277527,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 31610
    },
    {
      "epoch": 28.002657218777678,
      "grad_norm": 0.23673613369464874,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 31615
    },
    {
      "epoch": 28.00708591674048,
      "grad_norm": 0.18960891664028168,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 31620
    },
    {
      "epoch": 28.011514614703277,
      "grad_norm": 0.23437142372131348,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 31625
    },
    {
      "epoch": 28.015943312666074,
      "grad_norm": 0.22854219377040863,
      "learning_rate": 1e-05,
      "loss": 1.0332,
      "step": 31630
    },
    {
      "epoch": 28.020372010628876,
      "grad_norm": 0.2848300337791443,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 31635
    },
    {
      "epoch": 28.024800708591673,
      "grad_norm": 0.280914843082428,
      "learning_rate": 1e-05,
      "loss": 1.0204,
      "step": 31640
    },
    {
      "epoch": 28.029229406554474,
      "grad_norm": 0.2586367130279541,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 31645
    },
    {
      "epoch": 28.033658104517272,
      "grad_norm": 0.26630550622940063,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 31650
    },
    {
      "epoch": 28.03808680248007,
      "grad_norm": 0.2387808859348297,
      "learning_rate": 1e-05,
      "loss": 1.0523,
      "step": 31655
    },
    {
      "epoch": 28.04251550044287,
      "grad_norm": 0.2943151295185089,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 31660
    },
    {
      "epoch": 28.04694419840567,
      "grad_norm": 0.28493833541870117,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 31665
    },
    {
      "epoch": 28.051372896368466,
      "grad_norm": 0.2756597399711609,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 31670
    },
    {
      "epoch": 28.055801594331268,
      "grad_norm": 0.2798602879047394,
      "learning_rate": 1e-05,
      "loss": 1.0266,
      "step": 31675
    },
    {
      "epoch": 28.060230292294065,
      "grad_norm": 0.2521456182003021,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 31680
    },
    {
      "epoch": 28.064658990256863,
      "grad_norm": 0.20207074284553528,
      "learning_rate": 1e-05,
      "loss": 1.0216,
      "step": 31685
    },
    {
      "epoch": 28.069087688219664,
      "grad_norm": 0.2529992461204529,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 31690
    },
    {
      "epoch": 28.07351638618246,
      "grad_norm": 0.3253951370716095,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 31695
    },
    {
      "epoch": 28.077945084145263,
      "grad_norm": 0.300656795501709,
      "learning_rate": 1e-05,
      "loss": 1.0453,
      "step": 31700
    },
    {
      "epoch": 28.08237378210806,
      "grad_norm": 0.25386694073677063,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 31705
    },
    {
      "epoch": 28.086802480070858,
      "grad_norm": 0.25687623023986816,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 31710
    },
    {
      "epoch": 28.09123117803366,
      "grad_norm": 0.2788044214248657,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 31715
    },
    {
      "epoch": 28.095659875996457,
      "grad_norm": 0.2612397372722626,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 31720
    },
    {
      "epoch": 28.100088573959255,
      "grad_norm": 0.22717978060245514,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 31725
    },
    {
      "epoch": 28.104517271922056,
      "grad_norm": 0.27017271518707275,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 31730
    },
    {
      "epoch": 28.108945969884854,
      "grad_norm": 0.24155868589878082,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 31735
    },
    {
      "epoch": 28.11337466784765,
      "grad_norm": 0.2399553507566452,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 31740
    },
    {
      "epoch": 28.117803365810452,
      "grad_norm": 0.24093075096607208,
      "learning_rate": 1e-05,
      "loss": 1.0196,
      "step": 31745
    },
    {
      "epoch": 28.12223206377325,
      "grad_norm": 0.21842342615127563,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 31750
    },
    {
      "epoch": 28.12666076173605,
      "grad_norm": 0.26092880964279175,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 31755
    },
    {
      "epoch": 28.13108945969885,
      "grad_norm": 0.22773610055446625,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 31760
    },
    {
      "epoch": 28.135518157661647,
      "grad_norm": 0.305690735578537,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 31765
    },
    {
      "epoch": 28.139946855624448,
      "grad_norm": 0.2465287297964096,
      "learning_rate": 1e-05,
      "loss": 1.0374,
      "step": 31770
    },
    {
      "epoch": 28.144375553587246,
      "grad_norm": 0.2904472053050995,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 31775
    },
    {
      "epoch": 28.148804251550043,
      "grad_norm": 0.28554725646972656,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 31780
    },
    {
      "epoch": 28.153232949512844,
      "grad_norm": 0.2604232728481293,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 31785
    },
    {
      "epoch": 28.157661647475642,
      "grad_norm": 0.2199588119983673,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 31790
    },
    {
      "epoch": 28.16209034543844,
      "grad_norm": 0.21487219631671906,
      "learning_rate": 1e-05,
      "loss": 1.0226,
      "step": 31795
    },
    {
      "epoch": 28.16651904340124,
      "grad_norm": 0.23685072362422943,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 31800
    },
    {
      "epoch": 28.17094774136404,
      "grad_norm": 0.21897755563259125,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 31805
    },
    {
      "epoch": 28.175376439326836,
      "grad_norm": 0.20956380665302277,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 31810
    },
    {
      "epoch": 28.179805137289637,
      "grad_norm": 0.25374090671539307,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 31815
    },
    {
      "epoch": 28.184233835252435,
      "grad_norm": 0.23682576417922974,
      "learning_rate": 1e-05,
      "loss": 1.0491,
      "step": 31820
    },
    {
      "epoch": 28.188662533215236,
      "grad_norm": 0.2301643043756485,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 31825
    },
    {
      "epoch": 28.193091231178034,
      "grad_norm": 0.2204231172800064,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 31830
    },
    {
      "epoch": 28.19751992914083,
      "grad_norm": 0.23439417779445648,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 31835
    },
    {
      "epoch": 28.201948627103633,
      "grad_norm": 0.24330754578113556,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 31840
    },
    {
      "epoch": 28.20637732506643,
      "grad_norm": 0.23869043588638306,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 31845
    },
    {
      "epoch": 28.210806023029228,
      "grad_norm": 0.2518260180950165,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 31850
    },
    {
      "epoch": 28.21523472099203,
      "grad_norm": 0.2115086168050766,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 31855
    },
    {
      "epoch": 28.219663418954827,
      "grad_norm": 0.23373179137706757,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 31860
    },
    {
      "epoch": 28.224092116917625,
      "grad_norm": 0.23405955731868744,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 31865
    },
    {
      "epoch": 28.228520814880426,
      "grad_norm": 0.21524518728256226,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 31870
    },
    {
      "epoch": 28.232949512843224,
      "grad_norm": 0.23748517036437988,
      "learning_rate": 1e-05,
      "loss": 1.0461,
      "step": 31875
    },
    {
      "epoch": 28.237378210806025,
      "grad_norm": 0.2518925070762634,
      "learning_rate": 1e-05,
      "loss": 1.0291,
      "step": 31880
    },
    {
      "epoch": 28.241806908768822,
      "grad_norm": 0.21329110860824585,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 31885
    },
    {
      "epoch": 28.24623560673162,
      "grad_norm": 0.2637166380882263,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 31890
    },
    {
      "epoch": 28.25066430469442,
      "grad_norm": 0.252742737531662,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 31895
    },
    {
      "epoch": 28.25509300265722,
      "grad_norm": 0.2738778293132782,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 31900
    },
    {
      "epoch": 28.259521700620017,
      "grad_norm": 0.21903753280639648,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 31905
    },
    {
      "epoch": 28.263950398582818,
      "grad_norm": 0.2650688588619232,
      "learning_rate": 1e-05,
      "loss": 1.0396,
      "step": 31910
    },
    {
      "epoch": 28.268379096545615,
      "grad_norm": 0.25602075457572937,
      "learning_rate": 1e-05,
      "loss": 1.0443,
      "step": 31915
    },
    {
      "epoch": 28.272807794508413,
      "grad_norm": 0.24006669223308563,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 31920
    },
    {
      "epoch": 28.277236492471214,
      "grad_norm": 0.21383509039878845,
      "learning_rate": 1e-05,
      "loss": 1.0054,
      "step": 31925
    },
    {
      "epoch": 28.281665190434012,
      "grad_norm": 0.20268988609313965,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 31930
    },
    {
      "epoch": 28.28609388839681,
      "grad_norm": 0.23327352106571198,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 31935
    },
    {
      "epoch": 28.29052258635961,
      "grad_norm": 0.23685763776302338,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 31940
    },
    {
      "epoch": 28.29495128432241,
      "grad_norm": 0.2382057011127472,
      "learning_rate": 1e-05,
      "loss": 1.0354,
      "step": 31945
    },
    {
      "epoch": 28.29937998228521,
      "grad_norm": 0.21725286543369293,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 31950
    },
    {
      "epoch": 28.303808680248007,
      "grad_norm": 0.2097073495388031,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 31955
    },
    {
      "epoch": 28.308237378210805,
      "grad_norm": 0.21233265101909637,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 31960
    },
    {
      "epoch": 28.312666076173606,
      "grad_norm": 0.2257704734802246,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 31965
    },
    {
      "epoch": 28.317094774136404,
      "grad_norm": 0.23280641436576843,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 31970
    },
    {
      "epoch": 28.3215234720992,
      "grad_norm": 0.22488297522068024,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 31975
    },
    {
      "epoch": 28.325952170062003,
      "grad_norm": 0.27357837557792664,
      "learning_rate": 1e-05,
      "loss": 1.03,
      "step": 31980
    },
    {
      "epoch": 28.3303808680248,
      "grad_norm": 0.2501538097858429,
      "learning_rate": 1e-05,
      "loss": 0.9273,
      "step": 31985
    },
    {
      "epoch": 28.334809565987598,
      "grad_norm": 0.2617991864681244,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 31990
    },
    {
      "epoch": 28.3392382639504,
      "grad_norm": 0.23400233685970306,
      "learning_rate": 1e-05,
      "loss": 1.0437,
      "step": 31995
    },
    {
      "epoch": 28.343666961913197,
      "grad_norm": 0.2104942947626114,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 32000
    },
    {
      "epoch": 28.348095659875998,
      "grad_norm": 0.25284701585769653,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 32005
    },
    {
      "epoch": 28.352524357838796,
      "grad_norm": 0.20697462558746338,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 32010
    },
    {
      "epoch": 28.356953055801593,
      "grad_norm": 0.23990274965763092,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 32015
    },
    {
      "epoch": 28.361381753764395,
      "grad_norm": 0.26316896080970764,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 32020
    },
    {
      "epoch": 28.365810451727192,
      "grad_norm": 0.26972419023513794,
      "learning_rate": 1e-05,
      "loss": 1.0189,
      "step": 32025
    },
    {
      "epoch": 28.37023914968999,
      "grad_norm": 0.22244566679000854,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 32030
    },
    {
      "epoch": 28.37466784765279,
      "grad_norm": 0.2899254262447357,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 32035
    },
    {
      "epoch": 28.37909654561559,
      "grad_norm": 0.2493455857038498,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 32040
    },
    {
      "epoch": 28.383525243578386,
      "grad_norm": 0.2560606896877289,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 32045
    },
    {
      "epoch": 28.387953941541188,
      "grad_norm": 0.2786065340042114,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 32050
    },
    {
      "epoch": 28.392382639503985,
      "grad_norm": 0.2601239085197449,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 32055
    },
    {
      "epoch": 28.396811337466787,
      "grad_norm": 0.2901366353034973,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 32060
    },
    {
      "epoch": 28.401240035429584,
      "grad_norm": 0.23854628205299377,
      "learning_rate": 1e-05,
      "loss": 0.8997,
      "step": 32065
    },
    {
      "epoch": 28.405668733392382,
      "grad_norm": 0.24251247942447662,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 32070
    },
    {
      "epoch": 28.410097431355183,
      "grad_norm": 0.25075048208236694,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 32075
    },
    {
      "epoch": 28.41452612931798,
      "grad_norm": 0.22861126065254211,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 32080
    },
    {
      "epoch": 28.41895482728078,
      "grad_norm": 0.2756919264793396,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 32085
    },
    {
      "epoch": 28.42338352524358,
      "grad_norm": 0.21984876692295074,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 32090
    },
    {
      "epoch": 28.427812223206377,
      "grad_norm": 0.2257862538099289,
      "learning_rate": 1e-05,
      "loss": 1.0318,
      "step": 32095
    },
    {
      "epoch": 28.432240921169175,
      "grad_norm": 0.24166934192180634,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 32100
    },
    {
      "epoch": 28.436669619131976,
      "grad_norm": 0.22282323241233826,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 32105
    },
    {
      "epoch": 28.441098317094774,
      "grad_norm": 0.21808739006519318,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 32110
    },
    {
      "epoch": 28.44552701505757,
      "grad_norm": 0.2878672480583191,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 32115
    },
    {
      "epoch": 28.449955713020373,
      "grad_norm": 0.22229643166065216,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 32120
    },
    {
      "epoch": 28.45438441098317,
      "grad_norm": 0.23301953077316284,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 32125
    },
    {
      "epoch": 28.45881310894597,
      "grad_norm": 0.23343119025230408,
      "learning_rate": 1e-05,
      "loss": 0.9413,
      "step": 32130
    },
    {
      "epoch": 28.46324180690877,
      "grad_norm": 0.2900716960430145,
      "learning_rate": 1e-05,
      "loss": 1.037,
      "step": 32135
    },
    {
      "epoch": 28.467670504871567,
      "grad_norm": 0.23313802480697632,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 32140
    },
    {
      "epoch": 28.472099202834368,
      "grad_norm": 0.23826976120471954,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 32145
    },
    {
      "epoch": 28.476527900797166,
      "grad_norm": 0.19099058210849762,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 32150
    },
    {
      "epoch": 28.480956598759963,
      "grad_norm": 0.25383460521698,
      "learning_rate": 1e-05,
      "loss": 1.0155,
      "step": 32155
    },
    {
      "epoch": 28.485385296722765,
      "grad_norm": 0.23315663635730743,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 32160
    },
    {
      "epoch": 28.489813994685562,
      "grad_norm": 0.36980384588241577,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 32165
    },
    {
      "epoch": 28.49424269264836,
      "grad_norm": 0.2621196210384369,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 32170
    },
    {
      "epoch": 28.49867139061116,
      "grad_norm": 0.23126965761184692,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 32175
    },
    {
      "epoch": 28.50310008857396,
      "grad_norm": 0.2714141309261322,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 32180
    },
    {
      "epoch": 28.50752878653676,
      "grad_norm": 0.21692310273647308,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 32185
    },
    {
      "epoch": 28.511957484499558,
      "grad_norm": 0.2549686133861542,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 32190
    },
    {
      "epoch": 28.516386182462355,
      "grad_norm": 0.27794280648231506,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 32195
    },
    {
      "epoch": 28.520814880425156,
      "grad_norm": 0.26158109307289124,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 32200
    },
    {
      "epoch": 28.525243578387954,
      "grad_norm": 0.25001096725463867,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 32205
    },
    {
      "epoch": 28.52967227635075,
      "grad_norm": 0.2353513091802597,
      "learning_rate": 1e-05,
      "loss": 1.0318,
      "step": 32210
    },
    {
      "epoch": 28.534100974313553,
      "grad_norm": 0.23650024831295013,
      "learning_rate": 1e-05,
      "loss": 1.0183,
      "step": 32215
    },
    {
      "epoch": 28.53852967227635,
      "grad_norm": 0.23523296415805817,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 32220
    },
    {
      "epoch": 28.54295837023915,
      "grad_norm": 0.30064430832862854,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 32225
    },
    {
      "epoch": 28.54738706820195,
      "grad_norm": 0.2481122612953186,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 32230
    },
    {
      "epoch": 28.551815766164747,
      "grad_norm": 0.19871559739112854,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 32235
    },
    {
      "epoch": 28.556244464127545,
      "grad_norm": 0.2222663015127182,
      "learning_rate": 1e-05,
      "loss": 1.0247,
      "step": 32240
    },
    {
      "epoch": 28.560673162090346,
      "grad_norm": 0.24340662360191345,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 32245
    },
    {
      "epoch": 28.565101860053144,
      "grad_norm": 0.2281303107738495,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 32250
    },
    {
      "epoch": 28.569530558015945,
      "grad_norm": 0.272045761346817,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 32255
    },
    {
      "epoch": 28.573959255978743,
      "grad_norm": 0.24185380339622498,
      "learning_rate": 1e-05,
      "loss": 1.0574,
      "step": 32260
    },
    {
      "epoch": 28.57838795394154,
      "grad_norm": 0.2618290185928345,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 32265
    },
    {
      "epoch": 28.58281665190434,
      "grad_norm": 0.24950416386127472,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 32270
    },
    {
      "epoch": 28.58724534986714,
      "grad_norm": 0.2625531256198883,
      "learning_rate": 1e-05,
      "loss": 1.0152,
      "step": 32275
    },
    {
      "epoch": 28.591674047829937,
      "grad_norm": 0.2395915538072586,
      "learning_rate": 1e-05,
      "loss": 1.0183,
      "step": 32280
    },
    {
      "epoch": 28.596102745792738,
      "grad_norm": 0.2766381502151489,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 32285
    },
    {
      "epoch": 28.600531443755536,
      "grad_norm": 0.2338201403617859,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 32290
    },
    {
      "epoch": 28.604960141718333,
      "grad_norm": 0.35288164019584656,
      "learning_rate": 1e-05,
      "loss": 0.9252,
      "step": 32295
    },
    {
      "epoch": 28.609388839681134,
      "grad_norm": 0.23449599742889404,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 32300
    },
    {
      "epoch": 28.613817537643932,
      "grad_norm": 0.22771088778972626,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 32305
    },
    {
      "epoch": 28.618246235606733,
      "grad_norm": 0.25646376609802246,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 32310
    },
    {
      "epoch": 28.62267493356953,
      "grad_norm": 0.21815960109233856,
      "learning_rate": 1e-05,
      "loss": 1.0406,
      "step": 32315
    },
    {
      "epoch": 28.62710363153233,
      "grad_norm": 0.24311071634292603,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 32320
    },
    {
      "epoch": 28.63153232949513,
      "grad_norm": 0.23015135526657104,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 32325
    },
    {
      "epoch": 28.635961027457927,
      "grad_norm": 0.22268445789813995,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 32330
    },
    {
      "epoch": 28.640389725420725,
      "grad_norm": 0.2108345627784729,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 32335
    },
    {
      "epoch": 28.644818423383526,
      "grad_norm": 0.22398991882801056,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 32340
    },
    {
      "epoch": 28.649247121346324,
      "grad_norm": 0.22614456713199615,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 32345
    },
    {
      "epoch": 28.65367581930912,
      "grad_norm": 0.22226692736148834,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 32350
    },
    {
      "epoch": 28.658104517271923,
      "grad_norm": 0.21315021812915802,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 32355
    },
    {
      "epoch": 28.66253321523472,
      "grad_norm": 0.23941238224506378,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 32360
    },
    {
      "epoch": 28.666961913197518,
      "grad_norm": 0.2649783790111542,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 32365
    },
    {
      "epoch": 28.67139061116032,
      "grad_norm": 0.22079895436763763,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 32370
    },
    {
      "epoch": 28.675819309123117,
      "grad_norm": 0.29518985748291016,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 32375
    },
    {
      "epoch": 28.68024800708592,
      "grad_norm": 0.23155109584331512,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 32380
    },
    {
      "epoch": 28.684676705048716,
      "grad_norm": 0.2357022762298584,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 32385
    },
    {
      "epoch": 28.689105403011514,
      "grad_norm": 0.2670823335647583,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 32390
    },
    {
      "epoch": 28.693534100974315,
      "grad_norm": 0.2107042372226715,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 32395
    },
    {
      "epoch": 28.697962798937112,
      "grad_norm": 0.24749968945980072,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 32400
    },
    {
      "epoch": 28.70239149689991,
      "grad_norm": 0.23842880129814148,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 32405
    },
    {
      "epoch": 28.70682019486271,
      "grad_norm": 0.2762989401817322,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 32410
    },
    {
      "epoch": 28.71124889282551,
      "grad_norm": 0.23752766847610474,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 32415
    },
    {
      "epoch": 28.715677590788307,
      "grad_norm": 0.26040777564048767,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 32420
    },
    {
      "epoch": 28.720106288751108,
      "grad_norm": 0.2159717082977295,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 32425
    },
    {
      "epoch": 28.724534986713905,
      "grad_norm": 0.222582146525383,
      "learning_rate": 1e-05,
      "loss": 1.0459,
      "step": 32430
    },
    {
      "epoch": 28.728963684676707,
      "grad_norm": 0.28095534443855286,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 32435
    },
    {
      "epoch": 28.733392382639504,
      "grad_norm": 0.2394607663154602,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 32440
    },
    {
      "epoch": 28.737821080602302,
      "grad_norm": 0.237933948636055,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 32445
    },
    {
      "epoch": 28.742249778565103,
      "grad_norm": 0.3280743956565857,
      "learning_rate": 1e-05,
      "loss": 0.9273,
      "step": 32450
    },
    {
      "epoch": 28.7466784765279,
      "grad_norm": 0.2793219983577728,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 32455
    },
    {
      "epoch": 28.7511071744907,
      "grad_norm": 0.27851080894470215,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 32460
    },
    {
      "epoch": 28.7555358724535,
      "grad_norm": 0.24829672276973724,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 32465
    },
    {
      "epoch": 28.759964570416297,
      "grad_norm": 0.2521245777606964,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 32470
    },
    {
      "epoch": 28.764393268379095,
      "grad_norm": 0.22625643014907837,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 32475
    },
    {
      "epoch": 28.768821966341896,
      "grad_norm": 0.2462923675775528,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 32480
    },
    {
      "epoch": 28.773250664304694,
      "grad_norm": 0.23189212381839752,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 32485
    },
    {
      "epoch": 28.77767936226749,
      "grad_norm": 0.2547517716884613,
      "learning_rate": 1e-05,
      "loss": 1.0383,
      "step": 32490
    },
    {
      "epoch": 28.782108060230293,
      "grad_norm": 0.24846181273460388,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 32495
    },
    {
      "epoch": 28.78653675819309,
      "grad_norm": 0.2744988799095154,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 32500
    },
    {
      "epoch": 28.79096545615589,
      "grad_norm": 0.2284986972808838,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 32505
    },
    {
      "epoch": 28.79539415411869,
      "grad_norm": 0.25157272815704346,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 32510
    },
    {
      "epoch": 28.799822852081487,
      "grad_norm": 0.2612571716308594,
      "learning_rate": 1e-05,
      "loss": 1.0314,
      "step": 32515
    },
    {
      "epoch": 28.804251550044288,
      "grad_norm": 0.25162070989608765,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 32520
    },
    {
      "epoch": 28.808680248007086,
      "grad_norm": 0.23993653059005737,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 32525
    },
    {
      "epoch": 28.813108945969883,
      "grad_norm": 0.22653470933437347,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 32530
    },
    {
      "epoch": 28.817537643932685,
      "grad_norm": 0.2827704846858978,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 32535
    },
    {
      "epoch": 28.821966341895482,
      "grad_norm": 0.2577563226222992,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 32540
    },
    {
      "epoch": 28.82639503985828,
      "grad_norm": 0.28066474199295044,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 32545
    },
    {
      "epoch": 28.83082373782108,
      "grad_norm": 0.2725309133529663,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 32550
    },
    {
      "epoch": 28.83525243578388,
      "grad_norm": 0.23435166478157043,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 32555
    },
    {
      "epoch": 28.83968113374668,
      "grad_norm": 1.3431416749954224,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 32560
    },
    {
      "epoch": 28.844109831709478,
      "grad_norm": 0.21664059162139893,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 32565
    },
    {
      "epoch": 28.848538529672275,
      "grad_norm": 0.256254106760025,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 32570
    },
    {
      "epoch": 28.852967227635077,
      "grad_norm": 0.25111788511276245,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 32575
    },
    {
      "epoch": 28.857395925597874,
      "grad_norm": 0.24342425167560577,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 32580
    },
    {
      "epoch": 28.861824623560672,
      "grad_norm": 0.23971812427043915,
      "learning_rate": 1e-05,
      "loss": 1.0253,
      "step": 32585
    },
    {
      "epoch": 28.866253321523473,
      "grad_norm": 0.23585836589336395,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 32590
    },
    {
      "epoch": 28.87068201948627,
      "grad_norm": 0.2613334357738495,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 32595
    },
    {
      "epoch": 28.87511071744907,
      "grad_norm": 0.25037524104118347,
      "learning_rate": 1e-05,
      "loss": 1.0331,
      "step": 32600
    },
    {
      "epoch": 28.87953941541187,
      "grad_norm": 0.25890594720840454,
      "learning_rate": 1e-05,
      "loss": 1.0745,
      "step": 32605
    },
    {
      "epoch": 28.883968113374667,
      "grad_norm": 0.25616317987442017,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 32610
    },
    {
      "epoch": 28.888396811337465,
      "grad_norm": 0.2339998185634613,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 32615
    },
    {
      "epoch": 28.892825509300266,
      "grad_norm": 0.2537119686603546,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 32620
    },
    {
      "epoch": 28.897254207263064,
      "grad_norm": 0.21938014030456543,
      "learning_rate": 1e-05,
      "loss": 1.0399,
      "step": 32625
    },
    {
      "epoch": 28.901682905225865,
      "grad_norm": 0.2520517408847809,
      "learning_rate": 1e-05,
      "loss": 1.0335,
      "step": 32630
    },
    {
      "epoch": 28.906111603188663,
      "grad_norm": 0.24985922873020172,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 32635
    },
    {
      "epoch": 28.91054030115146,
      "grad_norm": 0.26078352332115173,
      "learning_rate": 1e-05,
      "loss": 1.0374,
      "step": 32640
    },
    {
      "epoch": 28.91496899911426,
      "grad_norm": 0.22888430953025818,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 32645
    },
    {
      "epoch": 28.91939769707706,
      "grad_norm": 0.23833692073822021,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 32650
    },
    {
      "epoch": 28.923826395039857,
      "grad_norm": 0.27575263381004333,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 32655
    },
    {
      "epoch": 28.928255093002658,
      "grad_norm": 0.19048675894737244,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 32660
    },
    {
      "epoch": 28.932683790965456,
      "grad_norm": 0.24705664813518524,
      "learning_rate": 1e-05,
      "loss": 0.9239,
      "step": 32665
    },
    {
      "epoch": 28.937112488928253,
      "grad_norm": 0.23557865619659424,
      "learning_rate": 1e-05,
      "loss": 0.9522,
      "step": 32670
    },
    {
      "epoch": 28.941541186891055,
      "grad_norm": 0.23979802429676056,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 32675
    },
    {
      "epoch": 28.945969884853852,
      "grad_norm": 0.2484261840581894,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 32680
    },
    {
      "epoch": 28.950398582816653,
      "grad_norm": 0.2429129183292389,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 32685
    },
    {
      "epoch": 28.95482728077945,
      "grad_norm": 0.23566299676895142,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 32690
    },
    {
      "epoch": 28.95925597874225,
      "grad_norm": 0.2131844460964203,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 32695
    },
    {
      "epoch": 28.96368467670505,
      "grad_norm": 0.22451364994049072,
      "learning_rate": 1e-05,
      "loss": 1.018,
      "step": 32700
    },
    {
      "epoch": 28.968113374667848,
      "grad_norm": 0.3154303729534149,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 32705
    },
    {
      "epoch": 28.972542072630645,
      "grad_norm": 0.2687758505344391,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 32710
    },
    {
      "epoch": 28.976970770593447,
      "grad_norm": 0.2667892277240753,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 32715
    },
    {
      "epoch": 28.981399468556244,
      "grad_norm": 0.2537265717983246,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 32720
    },
    {
      "epoch": 28.985828166519042,
      "grad_norm": 0.25515344738960266,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 32725
    },
    {
      "epoch": 28.990256864481843,
      "grad_norm": 0.24838703870773315,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 32730
    },
    {
      "epoch": 28.99468556244464,
      "grad_norm": 0.23681841790676117,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 32735
    },
    {
      "epoch": 28.999114260407442,
      "grad_norm": 0.21567222476005554,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 32740
    },
    {
      "epoch": 29.00354295837024,
      "grad_norm": 0.3280039429664612,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 32745
    },
    {
      "epoch": 29.007971656333037,
      "grad_norm": 0.21983349323272705,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 32750
    },
    {
      "epoch": 29.01240035429584,
      "grad_norm": 0.22912150621414185,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 32755
    },
    {
      "epoch": 29.016829052258636,
      "grad_norm": 0.2517050802707672,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 32760
    },
    {
      "epoch": 29.021257750221434,
      "grad_norm": 0.22547712922096252,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 32765
    },
    {
      "epoch": 29.025686448184235,
      "grad_norm": 0.2503494620323181,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 32770
    },
    {
      "epoch": 29.030115146147033,
      "grad_norm": 0.2700424790382385,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 32775
    },
    {
      "epoch": 29.03454384410983,
      "grad_norm": 0.2730877101421356,
      "learning_rate": 1e-05,
      "loss": 1.0097,
      "step": 32780
    },
    {
      "epoch": 29.03897254207263,
      "grad_norm": 0.23071296513080597,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 32785
    },
    {
      "epoch": 29.04340124003543,
      "grad_norm": 0.2713901996612549,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 32790
    },
    {
      "epoch": 29.047829937998227,
      "grad_norm": 0.291415810585022,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 32795
    },
    {
      "epoch": 29.052258635961028,
      "grad_norm": 0.25327184796333313,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 32800
    },
    {
      "epoch": 29.056687333923826,
      "grad_norm": 0.25974413752555847,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 32805
    },
    {
      "epoch": 29.061116031886627,
      "grad_norm": 0.2495075911283493,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 32810
    },
    {
      "epoch": 29.065544729849424,
      "grad_norm": 0.27440500259399414,
      "learning_rate": 1e-05,
      "loss": 1.0301,
      "step": 32815
    },
    {
      "epoch": 29.069973427812222,
      "grad_norm": 0.2577050030231476,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 32820
    },
    {
      "epoch": 29.074402125775023,
      "grad_norm": 0.24887725710868835,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 32825
    },
    {
      "epoch": 29.07883082373782,
      "grad_norm": 0.22878865897655487,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 32830
    },
    {
      "epoch": 29.08325952170062,
      "grad_norm": 0.22497308254241943,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 32835
    },
    {
      "epoch": 29.08768821966342,
      "grad_norm": 0.22363094985485077,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 32840
    },
    {
      "epoch": 29.092116917626218,
      "grad_norm": 0.26189014315605164,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 32845
    },
    {
      "epoch": 29.096545615589015,
      "grad_norm": 0.22668828070163727,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 32850
    },
    {
      "epoch": 29.100974313551816,
      "grad_norm": 0.23449285328388214,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 32855
    },
    {
      "epoch": 29.105403011514614,
      "grad_norm": 0.24742555618286133,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 32860
    },
    {
      "epoch": 29.109831709477415,
      "grad_norm": 0.22500079870224,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 32865
    },
    {
      "epoch": 29.114260407440213,
      "grad_norm": 0.22786012291908264,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 32870
    },
    {
      "epoch": 29.11868910540301,
      "grad_norm": 0.2354092299938202,
      "learning_rate": 1e-05,
      "loss": 1.0527,
      "step": 32875
    },
    {
      "epoch": 29.123117803365812,
      "grad_norm": 0.22803618013858795,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 32880
    },
    {
      "epoch": 29.12754650132861,
      "grad_norm": 0.2696690857410431,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 32885
    },
    {
      "epoch": 29.131975199291407,
      "grad_norm": 0.2248559445142746,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 32890
    },
    {
      "epoch": 29.13640389725421,
      "grad_norm": 0.23205527663230896,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 32895
    },
    {
      "epoch": 29.140832595217006,
      "grad_norm": 0.27893131971359253,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 32900
    },
    {
      "epoch": 29.145261293179804,
      "grad_norm": 0.24877573549747467,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 32905
    },
    {
      "epoch": 29.149689991142605,
      "grad_norm": 0.2536066472530365,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 32910
    },
    {
      "epoch": 29.154118689105402,
      "grad_norm": 0.2403651773929596,
      "learning_rate": 1e-05,
      "loss": 1.063,
      "step": 32915
    },
    {
      "epoch": 29.158547387068204,
      "grad_norm": 0.3604414463043213,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 32920
    },
    {
      "epoch": 29.162976085031,
      "grad_norm": 0.2457941621541977,
      "learning_rate": 1e-05,
      "loss": 1.0217,
      "step": 32925
    },
    {
      "epoch": 29.1674047829938,
      "grad_norm": 0.22010116279125214,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 32930
    },
    {
      "epoch": 29.1718334809566,
      "grad_norm": 0.24960453808307648,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 32935
    },
    {
      "epoch": 29.176262178919398,
      "grad_norm": 0.23997916281223297,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 32940
    },
    {
      "epoch": 29.180690876882196,
      "grad_norm": 0.27191200852394104,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 32945
    },
    {
      "epoch": 29.185119574844997,
      "grad_norm": 0.22481368482112885,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 32950
    },
    {
      "epoch": 29.189548272807794,
      "grad_norm": 0.2542947232723236,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 32955
    },
    {
      "epoch": 29.193976970770592,
      "grad_norm": 0.2834567129611969,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 32960
    },
    {
      "epoch": 29.198405668733393,
      "grad_norm": 0.2695099711418152,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 32965
    },
    {
      "epoch": 29.20283436669619,
      "grad_norm": 0.2747369408607483,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 32970
    },
    {
      "epoch": 29.20726306465899,
      "grad_norm": 0.2420107126235962,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 32975
    },
    {
      "epoch": 29.21169176262179,
      "grad_norm": 0.2598002552986145,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 32980
    },
    {
      "epoch": 29.216120460584587,
      "grad_norm": 0.2828283905982971,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 32985
    },
    {
      "epoch": 29.22054915854739,
      "grad_norm": 0.2351001352071762,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 32990
    },
    {
      "epoch": 29.224977856510186,
      "grad_norm": 0.2394755631685257,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 32995
    },
    {
      "epoch": 29.229406554472984,
      "grad_norm": 0.31276336312294006,
      "learning_rate": 1e-05,
      "loss": 1.017,
      "step": 33000
    },
    {
      "epoch": 29.233835252435785,
      "grad_norm": 0.2706705629825592,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 33005
    },
    {
      "epoch": 29.238263950398583,
      "grad_norm": 0.24142980575561523,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 33010
    },
    {
      "epoch": 29.24269264836138,
      "grad_norm": 0.23878198862075806,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 33015
    },
    {
      "epoch": 29.24712134632418,
      "grad_norm": 0.22935231029987335,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 33020
    },
    {
      "epoch": 29.25155004428698,
      "grad_norm": 0.2256104201078415,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 33025
    },
    {
      "epoch": 29.255978742249777,
      "grad_norm": 0.21620772778987885,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 33030
    },
    {
      "epoch": 29.260407440212578,
      "grad_norm": 0.2574934661388397,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 33035
    },
    {
      "epoch": 29.264836138175376,
      "grad_norm": 0.2542046904563904,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 33040
    },
    {
      "epoch": 29.269264836138177,
      "grad_norm": 0.26493820548057556,
      "learning_rate": 1e-05,
      "loss": 0.9396,
      "step": 33045
    },
    {
      "epoch": 29.273693534100975,
      "grad_norm": 0.25389036536216736,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 33050
    },
    {
      "epoch": 29.278122232063772,
      "grad_norm": 0.2502876818180084,
      "learning_rate": 1e-05,
      "loss": 1.0333,
      "step": 33055
    },
    {
      "epoch": 29.282550930026574,
      "grad_norm": 0.2571655809879303,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 33060
    },
    {
      "epoch": 29.28697962798937,
      "grad_norm": 0.24559326469898224,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 33065
    },
    {
      "epoch": 29.29140832595217,
      "grad_norm": 0.23028038442134857,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 33070
    },
    {
      "epoch": 29.29583702391497,
      "grad_norm": 0.22738775610923767,
      "learning_rate": 1e-05,
      "loss": 1.0204,
      "step": 33075
    },
    {
      "epoch": 29.300265721877768,
      "grad_norm": 0.2694743275642395,
      "learning_rate": 1e-05,
      "loss": 1.0262,
      "step": 33080
    },
    {
      "epoch": 29.304694419840565,
      "grad_norm": 0.2586454749107361,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 33085
    },
    {
      "epoch": 29.309123117803367,
      "grad_norm": 0.23461739718914032,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 33090
    },
    {
      "epoch": 29.313551815766164,
      "grad_norm": 0.22517363727092743,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 33095
    },
    {
      "epoch": 29.317980513728962,
      "grad_norm": 0.2183738797903061,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 33100
    },
    {
      "epoch": 29.322409211691763,
      "grad_norm": 0.2568979263305664,
      "learning_rate": 1e-05,
      "loss": 1.0218,
      "step": 33105
    },
    {
      "epoch": 29.32683790965456,
      "grad_norm": 0.2764614522457123,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 33110
    },
    {
      "epoch": 29.331266607617362,
      "grad_norm": 0.2150292694568634,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 33115
    },
    {
      "epoch": 29.33569530558016,
      "grad_norm": 0.29166650772094727,
      "learning_rate": 1e-05,
      "loss": 0.9275,
      "step": 33120
    },
    {
      "epoch": 29.340124003542957,
      "grad_norm": 0.22223339974880219,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 33125
    },
    {
      "epoch": 29.34455270150576,
      "grad_norm": 0.32859939336776733,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 33130
    },
    {
      "epoch": 29.348981399468556,
      "grad_norm": 0.2441742718219757,
      "learning_rate": 1e-05,
      "loss": 1.0515,
      "step": 33135
    },
    {
      "epoch": 29.353410097431354,
      "grad_norm": 0.2407771348953247,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 33140
    },
    {
      "epoch": 29.357838795394155,
      "grad_norm": 0.24372783303260803,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 33145
    },
    {
      "epoch": 29.362267493356953,
      "grad_norm": 0.20713920891284943,
      "learning_rate": 1e-05,
      "loss": 1.0254,
      "step": 33150
    },
    {
      "epoch": 29.36669619131975,
      "grad_norm": 0.24818897247314453,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 33155
    },
    {
      "epoch": 29.37112488928255,
      "grad_norm": 0.25761672854423523,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 33160
    },
    {
      "epoch": 29.37555358724535,
      "grad_norm": 0.23868688941001892,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 33165
    },
    {
      "epoch": 29.37998228520815,
      "grad_norm": 0.1938958764076233,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 33170
    },
    {
      "epoch": 29.384410983170948,
      "grad_norm": 0.2537584900856018,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 33175
    },
    {
      "epoch": 29.388839681133746,
      "grad_norm": 0.2777320444583893,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 33180
    },
    {
      "epoch": 29.393268379096547,
      "grad_norm": 0.23738200962543488,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 33185
    },
    {
      "epoch": 29.397697077059345,
      "grad_norm": 0.27272143959999084,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 33190
    },
    {
      "epoch": 29.402125775022142,
      "grad_norm": 0.25351670384407043,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 33195
    },
    {
      "epoch": 29.406554472984944,
      "grad_norm": 0.2402755618095398,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 33200
    },
    {
      "epoch": 29.41098317094774,
      "grad_norm": 0.22165927290916443,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 33205
    },
    {
      "epoch": 29.41541186891054,
      "grad_norm": 0.22902365028858185,
      "learning_rate": 1e-05,
      "loss": 1.0204,
      "step": 33210
    },
    {
      "epoch": 29.41984056687334,
      "grad_norm": 0.25669386982917786,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 33215
    },
    {
      "epoch": 29.424269264836138,
      "grad_norm": 0.25663214921951294,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 33220
    },
    {
      "epoch": 29.428697962798935,
      "grad_norm": 0.22321319580078125,
      "learning_rate": 1e-05,
      "loss": 1.0196,
      "step": 33225
    },
    {
      "epoch": 29.433126660761737,
      "grad_norm": 0.26744386553764343,
      "learning_rate": 1e-05,
      "loss": 0.9042,
      "step": 33230
    },
    {
      "epoch": 29.437555358724534,
      "grad_norm": 0.21041367948055267,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 33235
    },
    {
      "epoch": 29.441984056687335,
      "grad_norm": 0.2211267203092575,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 33240
    },
    {
      "epoch": 29.446412754650133,
      "grad_norm": 0.22458960115909576,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 33245
    },
    {
      "epoch": 29.45084145261293,
      "grad_norm": 0.25537192821502686,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 33250
    },
    {
      "epoch": 29.455270150575732,
      "grad_norm": 0.25810956954956055,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 33255
    },
    {
      "epoch": 29.45969884853853,
      "grad_norm": 0.20794254541397095,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 33260
    },
    {
      "epoch": 29.464127546501327,
      "grad_norm": 0.24391856789588928,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 33265
    },
    {
      "epoch": 29.46855624446413,
      "grad_norm": 0.2551543712615967,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 33270
    },
    {
      "epoch": 29.472984942426926,
      "grad_norm": 0.24688716232776642,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 33275
    },
    {
      "epoch": 29.477413640389724,
      "grad_norm": 0.2589344084262848,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 33280
    },
    {
      "epoch": 29.481842338352525,
      "grad_norm": 0.20920489728450775,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 33285
    },
    {
      "epoch": 29.486271036315323,
      "grad_norm": 0.26096221804618835,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 33290
    },
    {
      "epoch": 29.490699734278124,
      "grad_norm": 0.23136945068836212,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 33295
    },
    {
      "epoch": 29.49512843224092,
      "grad_norm": 0.24354800581932068,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 33300
    },
    {
      "epoch": 29.49955713020372,
      "grad_norm": 0.26048028469085693,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 33305
    },
    {
      "epoch": 29.50398582816652,
      "grad_norm": 0.26856541633605957,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 33310
    },
    {
      "epoch": 29.508414526129318,
      "grad_norm": 0.2925977110862732,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 33315
    },
    {
      "epoch": 29.512843224092116,
      "grad_norm": 0.2360851913690567,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 33320
    },
    {
      "epoch": 29.517271922054917,
      "grad_norm": 0.25888973474502563,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 33325
    },
    {
      "epoch": 29.521700620017715,
      "grad_norm": 0.27774688601493835,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 33330
    },
    {
      "epoch": 29.526129317980512,
      "grad_norm": 0.29836344718933105,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 33335
    },
    {
      "epoch": 29.530558015943313,
      "grad_norm": 0.23448826372623444,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 33340
    },
    {
      "epoch": 29.53498671390611,
      "grad_norm": 0.24063892662525177,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 33345
    },
    {
      "epoch": 29.53941541186891,
      "grad_norm": 0.28010237216949463,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 33350
    },
    {
      "epoch": 29.54384410983171,
      "grad_norm": 0.23950083553791046,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 33355
    },
    {
      "epoch": 29.548272807794508,
      "grad_norm": 0.2847791016101837,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 33360
    },
    {
      "epoch": 29.55270150575731,
      "grad_norm": 0.21922768652439117,
      "learning_rate": 1e-05,
      "loss": 0.9258,
      "step": 33365
    },
    {
      "epoch": 29.557130203720106,
      "grad_norm": 0.23477822542190552,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 33370
    },
    {
      "epoch": 29.561558901682904,
      "grad_norm": 0.22649945318698883,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 33375
    },
    {
      "epoch": 29.565987599645705,
      "grad_norm": 0.25750523805618286,
      "learning_rate": 1e-05,
      "loss": 1.0392,
      "step": 33380
    },
    {
      "epoch": 29.570416297608503,
      "grad_norm": 0.29351806640625,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 33385
    },
    {
      "epoch": 29.5748449955713,
      "grad_norm": 0.2347741574048996,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 33390
    },
    {
      "epoch": 29.579273693534102,
      "grad_norm": 0.27449727058410645,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 33395
    },
    {
      "epoch": 29.5837023914969,
      "grad_norm": 0.24438513815402985,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 33400
    },
    {
      "epoch": 29.588131089459697,
      "grad_norm": 0.2961277961730957,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 33405
    },
    {
      "epoch": 29.5925597874225,
      "grad_norm": 0.25197264552116394,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 33410
    },
    {
      "epoch": 29.596988485385296,
      "grad_norm": 0.26049402356147766,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 33415
    },
    {
      "epoch": 29.601417183348097,
      "grad_norm": 0.3087747097015381,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 33420
    },
    {
      "epoch": 29.605845881310895,
      "grad_norm": 0.24062579870224,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 33425
    },
    {
      "epoch": 29.610274579273693,
      "grad_norm": 0.2511786222457886,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 33430
    },
    {
      "epoch": 29.614703277236494,
      "grad_norm": 0.22619308531284332,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 33435
    },
    {
      "epoch": 29.61913197519929,
      "grad_norm": 0.2527395486831665,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 33440
    },
    {
      "epoch": 29.62356067316209,
      "grad_norm": 0.19290590286254883,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 33445
    },
    {
      "epoch": 29.62798937112489,
      "grad_norm": 0.3469868302345276,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 33450
    },
    {
      "epoch": 29.632418069087688,
      "grad_norm": 0.24371160566806793,
      "learning_rate": 1e-05,
      "loss": 0.8963,
      "step": 33455
    },
    {
      "epoch": 29.636846767050486,
      "grad_norm": 0.24923472106456757,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 33460
    },
    {
      "epoch": 29.641275465013287,
      "grad_norm": 0.23934711515903473,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 33465
    },
    {
      "epoch": 29.645704162976084,
      "grad_norm": 0.2713635563850403,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 33470
    },
    {
      "epoch": 29.650132860938886,
      "grad_norm": 0.24598468840122223,
      "learning_rate": 1e-05,
      "loss": 1.0265,
      "step": 33475
    },
    {
      "epoch": 29.654561558901683,
      "grad_norm": 0.24211840331554413,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 33480
    },
    {
      "epoch": 29.65899025686448,
      "grad_norm": 0.22946064174175262,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 33485
    },
    {
      "epoch": 29.663418954827282,
      "grad_norm": 0.26737797260284424,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 33490
    },
    {
      "epoch": 29.66784765279008,
      "grad_norm": 0.2769491374492645,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 33495
    },
    {
      "epoch": 29.672276350752878,
      "grad_norm": 0.2214026153087616,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 33500
    },
    {
      "epoch": 29.67670504871568,
      "grad_norm": 0.231183260679245,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 33505
    },
    {
      "epoch": 29.681133746678476,
      "grad_norm": 0.25053277611732483,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 33510
    },
    {
      "epoch": 29.685562444641274,
      "grad_norm": 0.22274485230445862,
      "learning_rate": 1e-05,
      "loss": 1.0318,
      "step": 33515
    },
    {
      "epoch": 29.689991142604075,
      "grad_norm": 0.2466314435005188,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 33520
    },
    {
      "epoch": 29.694419840566873,
      "grad_norm": 0.2397489696741104,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 33525
    },
    {
      "epoch": 29.698848538529674,
      "grad_norm": 0.22980856895446777,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 33530
    },
    {
      "epoch": 29.70327723649247,
      "grad_norm": 0.2572290897369385,
      "learning_rate": 1e-05,
      "loss": 0.9145,
      "step": 33535
    },
    {
      "epoch": 29.70770593445527,
      "grad_norm": 0.24892254173755646,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 33540
    },
    {
      "epoch": 29.71213463241807,
      "grad_norm": 0.2390914112329483,
      "learning_rate": 1e-05,
      "loss": 0.9229,
      "step": 33545
    },
    {
      "epoch": 29.71656333038087,
      "grad_norm": 0.21791300177574158,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 33550
    },
    {
      "epoch": 29.720992028343666,
      "grad_norm": 0.24028660356998444,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 33555
    },
    {
      "epoch": 29.725420726306467,
      "grad_norm": 0.2410203218460083,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 33560
    },
    {
      "epoch": 29.729849424269265,
      "grad_norm": 0.24294373393058777,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 33565
    },
    {
      "epoch": 29.734278122232062,
      "grad_norm": 0.20225472748279572,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 33570
    },
    {
      "epoch": 29.738706820194864,
      "grad_norm": 0.19303831458091736,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 33575
    },
    {
      "epoch": 29.74313551815766,
      "grad_norm": 0.20625481009483337,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 33580
    },
    {
      "epoch": 29.74756421612046,
      "grad_norm": 0.24322716891765594,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 33585
    },
    {
      "epoch": 29.75199291408326,
      "grad_norm": 0.26344043016433716,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 33590
    },
    {
      "epoch": 29.756421612046058,
      "grad_norm": 0.28257396817207336,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 33595
    },
    {
      "epoch": 29.76085031000886,
      "grad_norm": 0.27443960309028625,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 33600
    },
    {
      "epoch": 29.765279007971657,
      "grad_norm": 0.24726985394954681,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 33605
    },
    {
      "epoch": 29.769707705934454,
      "grad_norm": 0.22249121963977814,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 33610
    },
    {
      "epoch": 29.774136403897256,
      "grad_norm": 0.25300875306129456,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 33615
    },
    {
      "epoch": 29.778565101860053,
      "grad_norm": 0.2905903458595276,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 33620
    },
    {
      "epoch": 29.78299379982285,
      "grad_norm": 0.2594776153564453,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 33625
    },
    {
      "epoch": 29.787422497785652,
      "grad_norm": 0.4132271707057953,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 33630
    },
    {
      "epoch": 29.79185119574845,
      "grad_norm": 0.23372139036655426,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 33635
    },
    {
      "epoch": 29.796279893711247,
      "grad_norm": 0.25129473209381104,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 33640
    },
    {
      "epoch": 29.80070859167405,
      "grad_norm": 0.2372855246067047,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 33645
    },
    {
      "epoch": 29.805137289636846,
      "grad_norm": 0.25713202357292175,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 33650
    },
    {
      "epoch": 29.809565987599647,
      "grad_norm": 0.25309890508651733,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 33655
    },
    {
      "epoch": 29.813994685562445,
      "grad_norm": 0.23268768191337585,
      "learning_rate": 1e-05,
      "loss": 1.0311,
      "step": 33660
    },
    {
      "epoch": 29.818423383525243,
      "grad_norm": 0.22995048761367798,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 33665
    },
    {
      "epoch": 29.822852081488044,
      "grad_norm": 0.3111424744129181,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 33670
    },
    {
      "epoch": 29.82728077945084,
      "grad_norm": 0.2710413336753845,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 33675
    },
    {
      "epoch": 29.83170947741364,
      "grad_norm": 0.3212091624736786,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 33680
    },
    {
      "epoch": 29.83613817537644,
      "grad_norm": 0.3030320107936859,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 33685
    },
    {
      "epoch": 29.840566873339238,
      "grad_norm": 0.24937406182289124,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 33690
    },
    {
      "epoch": 29.844995571302036,
      "grad_norm": 0.24058672785758972,
      "learning_rate": 1e-05,
      "loss": 1.0373,
      "step": 33695
    },
    {
      "epoch": 29.849424269264837,
      "grad_norm": 0.23151683807373047,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 33700
    },
    {
      "epoch": 29.853852967227635,
      "grad_norm": 0.29559141397476196,
      "learning_rate": 1e-05,
      "loss": 1.0257,
      "step": 33705
    },
    {
      "epoch": 29.858281665190432,
      "grad_norm": 0.2709830403327942,
      "learning_rate": 1e-05,
      "loss": 1.0242,
      "step": 33710
    },
    {
      "epoch": 29.862710363153234,
      "grad_norm": 0.30582910776138306,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 33715
    },
    {
      "epoch": 29.86713906111603,
      "grad_norm": 0.23479345440864563,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 33720
    },
    {
      "epoch": 29.871567759078832,
      "grad_norm": 0.1971449851989746,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 33725
    },
    {
      "epoch": 29.87599645704163,
      "grad_norm": 0.276248037815094,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 33730
    },
    {
      "epoch": 29.880425155004428,
      "grad_norm": 0.2343430519104004,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 33735
    },
    {
      "epoch": 29.88485385296723,
      "grad_norm": 0.2501111328601837,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 33740
    },
    {
      "epoch": 29.889282550930027,
      "grad_norm": 0.2660665810108185,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 33745
    },
    {
      "epoch": 29.893711248892824,
      "grad_norm": 0.2440873384475708,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 33750
    },
    {
      "epoch": 29.898139946855625,
      "grad_norm": 0.2634141445159912,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 33755
    },
    {
      "epoch": 29.902568644818423,
      "grad_norm": 0.2766169309616089,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 33760
    },
    {
      "epoch": 29.90699734278122,
      "grad_norm": 0.25497835874557495,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 33765
    },
    {
      "epoch": 29.911426040744022,
      "grad_norm": 0.23452550172805786,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 33770
    },
    {
      "epoch": 29.91585473870682,
      "grad_norm": 0.27484825253486633,
      "learning_rate": 1e-05,
      "loss": 1.0107,
      "step": 33775
    },
    {
      "epoch": 29.92028343666962,
      "grad_norm": 0.24014128744602203,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 33780
    },
    {
      "epoch": 29.92471213463242,
      "grad_norm": 0.24675621092319489,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 33785
    },
    {
      "epoch": 29.929140832595216,
      "grad_norm": 0.2273753583431244,
      "learning_rate": 1e-05,
      "loss": 1.0349,
      "step": 33790
    },
    {
      "epoch": 29.933569530558017,
      "grad_norm": 0.23943696916103363,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 33795
    },
    {
      "epoch": 29.937998228520815,
      "grad_norm": 0.21891751885414124,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 33800
    },
    {
      "epoch": 29.942426926483613,
      "grad_norm": 0.23428450524806976,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 33805
    },
    {
      "epoch": 29.946855624446414,
      "grad_norm": 0.21084064245224,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 33810
    },
    {
      "epoch": 29.95128432240921,
      "grad_norm": 0.22127531468868256,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 33815
    },
    {
      "epoch": 29.95571302037201,
      "grad_norm": 0.21720542013645172,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 33820
    },
    {
      "epoch": 29.96014171833481,
      "grad_norm": 0.21275512874126434,
      "learning_rate": 1e-05,
      "loss": 0.922,
      "step": 33825
    },
    {
      "epoch": 29.964570416297608,
      "grad_norm": 0.21496634185314178,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 33830
    },
    {
      "epoch": 29.968999114260406,
      "grad_norm": 0.29388946294784546,
      "learning_rate": 1e-05,
      "loss": 1.0267,
      "step": 33835
    },
    {
      "epoch": 29.973427812223207,
      "grad_norm": 0.2560620903968811,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 33840
    },
    {
      "epoch": 29.977856510186005,
      "grad_norm": 0.2432190179824829,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 33845
    },
    {
      "epoch": 29.982285208148806,
      "grad_norm": 0.2676409184932709,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 33850
    },
    {
      "epoch": 29.986713906111603,
      "grad_norm": 0.20951853692531586,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 33855
    },
    {
      "epoch": 29.9911426040744,
      "grad_norm": 0.2243216186761856,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 33860
    },
    {
      "epoch": 29.995571302037202,
      "grad_norm": 0.23263275623321533,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 33865
    },
    {
      "epoch": 30.0,
      "grad_norm": 0.22279272973537445,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 33870
    },
    {
      "epoch": 30.004428697962798,
      "grad_norm": 0.2128625214099884,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 33875
    },
    {
      "epoch": 30.0088573959256,
      "grad_norm": 0.24543286859989166,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 33880
    },
    {
      "epoch": 30.013286093888397,
      "grad_norm": 0.21484240889549255,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 33885
    },
    {
      "epoch": 30.017714791851194,
      "grad_norm": 0.2522362172603607,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 33890
    },
    {
      "epoch": 30.022143489813995,
      "grad_norm": 0.20335379242897034,
      "learning_rate": 1e-05,
      "loss": 0.913,
      "step": 33895
    },
    {
      "epoch": 30.026572187776793,
      "grad_norm": 0.27319905161857605,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 33900
    },
    {
      "epoch": 30.031000885739594,
      "grad_norm": 0.24649815261363983,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 33905
    },
    {
      "epoch": 30.035429583702392,
      "grad_norm": 0.2384646236896515,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 33910
    },
    {
      "epoch": 30.03985828166519,
      "grad_norm": 0.24109268188476562,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 33915
    },
    {
      "epoch": 30.04428697962799,
      "grad_norm": 0.22765232622623444,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 33920
    },
    {
      "epoch": 30.04871567759079,
      "grad_norm": 0.23042981326580048,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 33925
    },
    {
      "epoch": 30.053144375553586,
      "grad_norm": 0.22180402278900146,
      "learning_rate": 1e-05,
      "loss": 1.0232,
      "step": 33930
    },
    {
      "epoch": 30.057573073516387,
      "grad_norm": 0.2460329681634903,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 33935
    },
    {
      "epoch": 30.062001771479185,
      "grad_norm": 0.2586100399494171,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 33940
    },
    {
      "epoch": 30.066430469441983,
      "grad_norm": 0.2281045913696289,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 33945
    },
    {
      "epoch": 30.070859167404784,
      "grad_norm": 0.31259095668792725,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 33950
    },
    {
      "epoch": 30.07528786536758,
      "grad_norm": 0.29469239711761475,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 33955
    },
    {
      "epoch": 30.07971656333038,
      "grad_norm": 0.2334946095943451,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 33960
    },
    {
      "epoch": 30.08414526129318,
      "grad_norm": 0.2896636426448822,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 33965
    },
    {
      "epoch": 30.088573959255978,
      "grad_norm": 0.24355122447013855,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 33970
    },
    {
      "epoch": 30.09300265721878,
      "grad_norm": 0.26359230279922485,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 33975
    },
    {
      "epoch": 30.097431355181577,
      "grad_norm": 0.27373090386390686,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 33980
    },
    {
      "epoch": 30.101860053144375,
      "grad_norm": 0.21663948893547058,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 33985
    },
    {
      "epoch": 30.106288751107176,
      "grad_norm": 0.22245050966739655,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 33990
    },
    {
      "epoch": 30.110717449069973,
      "grad_norm": 0.22601357102394104,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 33995
    },
    {
      "epoch": 30.11514614703277,
      "grad_norm": 0.20984213054180145,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 34000
    },
    {
      "epoch": 30.119574844995572,
      "grad_norm": 0.21778535842895508,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 34005
    },
    {
      "epoch": 30.12400354295837,
      "grad_norm": 0.2527742385864258,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 34010
    },
    {
      "epoch": 30.128432240921168,
      "grad_norm": 0.24183496832847595,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 34015
    },
    {
      "epoch": 30.13286093888397,
      "grad_norm": 0.2694372236728668,
      "learning_rate": 1e-05,
      "loss": 1.0414,
      "step": 34020
    },
    {
      "epoch": 30.137289636846766,
      "grad_norm": 0.2790459990501404,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 34025
    },
    {
      "epoch": 30.141718334809568,
      "grad_norm": 0.3224250376224518,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 34030
    },
    {
      "epoch": 30.146147032772365,
      "grad_norm": 0.28082728385925293,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 34035
    },
    {
      "epoch": 30.150575730735163,
      "grad_norm": 0.25160282850265503,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 34040
    },
    {
      "epoch": 30.155004428697964,
      "grad_norm": 0.2314085215330124,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 34045
    },
    {
      "epoch": 30.159433126660762,
      "grad_norm": 0.21914249658584595,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 34050
    },
    {
      "epoch": 30.16386182462356,
      "grad_norm": 0.25499778985977173,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 34055
    },
    {
      "epoch": 30.16829052258636,
      "grad_norm": 0.2513638138771057,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 34060
    },
    {
      "epoch": 30.17271922054916,
      "grad_norm": 0.23948869109153748,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 34065
    },
    {
      "epoch": 30.177147918511956,
      "grad_norm": 0.24730654060840607,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 34070
    },
    {
      "epoch": 30.181576616474757,
      "grad_norm": 0.22750549018383026,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 34075
    },
    {
      "epoch": 30.186005314437555,
      "grad_norm": 0.24042083323001862,
      "learning_rate": 1e-05,
      "loss": 1.0371,
      "step": 34080
    },
    {
      "epoch": 30.190434012400353,
      "grad_norm": 0.26706311106681824,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 34085
    },
    {
      "epoch": 30.194862710363154,
      "grad_norm": 0.2427464723587036,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 34090
    },
    {
      "epoch": 30.19929140832595,
      "grad_norm": 0.21277616918087006,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 34095
    },
    {
      "epoch": 30.203720106288753,
      "grad_norm": 0.26772087812423706,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 34100
    },
    {
      "epoch": 30.20814880425155,
      "grad_norm": 0.3057091534137726,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 34105
    },
    {
      "epoch": 30.212577502214348,
      "grad_norm": 0.21423472464084625,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 34110
    },
    {
      "epoch": 30.21700620017715,
      "grad_norm": 0.23690512776374817,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 34115
    },
    {
      "epoch": 30.221434898139947,
      "grad_norm": 0.22269457578659058,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 34120
    },
    {
      "epoch": 30.225863596102744,
      "grad_norm": 0.2575618326663971,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 34125
    },
    {
      "epoch": 30.230292294065546,
      "grad_norm": 0.29142439365386963,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 34130
    },
    {
      "epoch": 30.234720992028343,
      "grad_norm": 0.24798934161663055,
      "learning_rate": 1e-05,
      "loss": 1.0161,
      "step": 34135
    },
    {
      "epoch": 30.23914968999114,
      "grad_norm": 0.2384113222360611,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 34140
    },
    {
      "epoch": 30.243578387953942,
      "grad_norm": 0.24455608427524567,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 34145
    },
    {
      "epoch": 30.24800708591674,
      "grad_norm": 0.23404699563980103,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 34150
    },
    {
      "epoch": 30.25243578387954,
      "grad_norm": 0.23720529675483704,
      "learning_rate": 1e-05,
      "loss": 0.9301,
      "step": 34155
    },
    {
      "epoch": 30.25686448184234,
      "grad_norm": 0.2367081642150879,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 34160
    },
    {
      "epoch": 30.261293179805136,
      "grad_norm": 0.2246650755405426,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 34165
    },
    {
      "epoch": 30.265721877767938,
      "grad_norm": 0.24377083778381348,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 34170
    },
    {
      "epoch": 30.270150575730735,
      "grad_norm": 0.2659006714820862,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 34175
    },
    {
      "epoch": 30.274579273693533,
      "grad_norm": 0.23865897953510284,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 34180
    },
    {
      "epoch": 30.279007971656334,
      "grad_norm": 0.32606515288352966,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 34185
    },
    {
      "epoch": 30.28343666961913,
      "grad_norm": 0.2819402813911438,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 34190
    },
    {
      "epoch": 30.28786536758193,
      "grad_norm": 0.23415009677410126,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 34195
    },
    {
      "epoch": 30.29229406554473,
      "grad_norm": 0.22377543151378632,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 34200
    },
    {
      "epoch": 30.29672276350753,
      "grad_norm": 0.2716352045536041,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 34205
    },
    {
      "epoch": 30.30115146147033,
      "grad_norm": 0.2506796717643738,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 34210
    },
    {
      "epoch": 30.305580159433127,
      "grad_norm": 0.245045006275177,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 34215
    },
    {
      "epoch": 30.310008857395925,
      "grad_norm": 0.22532270848751068,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 34220
    },
    {
      "epoch": 30.314437555358726,
      "grad_norm": 0.2955477833747864,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 34225
    },
    {
      "epoch": 30.318866253321524,
      "grad_norm": 0.3092818856239319,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 34230
    },
    {
      "epoch": 30.32329495128432,
      "grad_norm": 0.22338555753231049,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 34235
    },
    {
      "epoch": 30.327723649247122,
      "grad_norm": 0.22364799678325653,
      "learning_rate": 1e-05,
      "loss": 0.9232,
      "step": 34240
    },
    {
      "epoch": 30.33215234720992,
      "grad_norm": 0.23154060542583466,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 34245
    },
    {
      "epoch": 30.336581045172718,
      "grad_norm": 0.240864560008049,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 34250
    },
    {
      "epoch": 30.34100974313552,
      "grad_norm": 0.24423836171627045,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 34255
    },
    {
      "epoch": 30.345438441098317,
      "grad_norm": 0.24207143485546112,
      "learning_rate": 1e-05,
      "loss": 0.9293,
      "step": 34260
    },
    {
      "epoch": 30.349867139061114,
      "grad_norm": 0.29829666018486023,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 34265
    },
    {
      "epoch": 30.354295837023916,
      "grad_norm": 0.2518916726112366,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 34270
    },
    {
      "epoch": 30.358724534986713,
      "grad_norm": 0.2573820650577545,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 34275
    },
    {
      "epoch": 30.363153232949514,
      "grad_norm": 0.26762500405311584,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 34280
    },
    {
      "epoch": 30.367581930912312,
      "grad_norm": 0.2401001900434494,
      "learning_rate": 1e-05,
      "loss": 1.0604,
      "step": 34285
    },
    {
      "epoch": 30.37201062887511,
      "grad_norm": 0.2129121869802475,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 34290
    },
    {
      "epoch": 30.37643932683791,
      "grad_norm": 0.2617007791996002,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 34295
    },
    {
      "epoch": 30.38086802480071,
      "grad_norm": 0.282577782869339,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 34300
    },
    {
      "epoch": 30.385296722763506,
      "grad_norm": 0.28779372572898865,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 34305
    },
    {
      "epoch": 30.389725420726307,
      "grad_norm": 0.2623580992221832,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 34310
    },
    {
      "epoch": 30.394154118689105,
      "grad_norm": 0.32232269644737244,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 34315
    },
    {
      "epoch": 30.398582816651903,
      "grad_norm": 0.2766963839530945,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 34320
    },
    {
      "epoch": 30.403011514614704,
      "grad_norm": 0.20745114982128143,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 34325
    },
    {
      "epoch": 30.4074402125775,
      "grad_norm": 0.20107071101665497,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 34330
    },
    {
      "epoch": 30.411868910540303,
      "grad_norm": 0.2113850861787796,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 34335
    },
    {
      "epoch": 30.4162976085031,
      "grad_norm": 0.2446461021900177,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 34340
    },
    {
      "epoch": 30.420726306465898,
      "grad_norm": 0.24066166579723358,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 34345
    },
    {
      "epoch": 30.4251550044287,
      "grad_norm": 0.23494704067707062,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 34350
    },
    {
      "epoch": 30.429583702391497,
      "grad_norm": 0.23799870908260345,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 34355
    },
    {
      "epoch": 30.434012400354295,
      "grad_norm": 0.28301307559013367,
      "learning_rate": 1e-05,
      "loss": 1.0245,
      "step": 34360
    },
    {
      "epoch": 30.438441098317096,
      "grad_norm": 0.2635035514831543,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 34365
    },
    {
      "epoch": 30.442869796279894,
      "grad_norm": 0.24691958725452423,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 34370
    },
    {
      "epoch": 30.44729849424269,
      "grad_norm": 0.22498434782028198,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 34375
    },
    {
      "epoch": 30.451727192205492,
      "grad_norm": 0.2796193063259125,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 34380
    },
    {
      "epoch": 30.45615589016829,
      "grad_norm": 0.2543242275714874,
      "learning_rate": 1e-05,
      "loss": 1.0287,
      "step": 34385
    },
    {
      "epoch": 30.460584588131088,
      "grad_norm": 0.26206764578819275,
      "learning_rate": 1e-05,
      "loss": 1.014,
      "step": 34390
    },
    {
      "epoch": 30.46501328609389,
      "grad_norm": 0.24417613446712494,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 34395
    },
    {
      "epoch": 30.469441984056687,
      "grad_norm": 0.22031310200691223,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 34400
    },
    {
      "epoch": 30.473870682019488,
      "grad_norm": 0.22819258272647858,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 34405
    },
    {
      "epoch": 30.478299379982285,
      "grad_norm": 0.28085991740226746,
      "learning_rate": 1e-05,
      "loss": 0.9287,
      "step": 34410
    },
    {
      "epoch": 30.482728077945083,
      "grad_norm": 0.2402373105287552,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 34415
    },
    {
      "epoch": 30.487156775907884,
      "grad_norm": 0.2271738499403,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 34420
    },
    {
      "epoch": 30.491585473870682,
      "grad_norm": 0.2838076651096344,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 34425
    },
    {
      "epoch": 30.49601417183348,
      "grad_norm": 0.21655474603176117,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 34430
    },
    {
      "epoch": 30.50044286979628,
      "grad_norm": 0.22896727919578552,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 34435
    },
    {
      "epoch": 30.50487156775908,
      "grad_norm": 0.28526267409324646,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 34440
    },
    {
      "epoch": 30.509300265721876,
      "grad_norm": 0.26764383912086487,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 34445
    },
    {
      "epoch": 30.513728963684677,
      "grad_norm": 0.24766159057617188,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 34450
    },
    {
      "epoch": 30.518157661647475,
      "grad_norm": 0.2382461130619049,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 34455
    },
    {
      "epoch": 30.522586359610276,
      "grad_norm": 0.2269027978181839,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 34460
    },
    {
      "epoch": 30.527015057573074,
      "grad_norm": 0.2292771190404892,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 34465
    },
    {
      "epoch": 30.53144375553587,
      "grad_norm": 0.2514874041080475,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 34470
    },
    {
      "epoch": 30.535872453498673,
      "grad_norm": 0.2683906555175781,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 34475
    },
    {
      "epoch": 30.54030115146147,
      "grad_norm": 0.2551718056201935,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 34480
    },
    {
      "epoch": 30.544729849424268,
      "grad_norm": 0.2350943386554718,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 34485
    },
    {
      "epoch": 30.54915854738707,
      "grad_norm": 0.2871323823928833,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 34490
    },
    {
      "epoch": 30.553587245349867,
      "grad_norm": 0.23525409400463104,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 34495
    },
    {
      "epoch": 30.558015943312665,
      "grad_norm": 0.21111206710338593,
      "learning_rate": 1e-05,
      "loss": 0.8939,
      "step": 34500
    },
    {
      "epoch": 30.562444641275466,
      "grad_norm": 0.2725718319416046,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 34505
    },
    {
      "epoch": 30.566873339238263,
      "grad_norm": 0.21595382690429688,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 34510
    },
    {
      "epoch": 30.571302037201065,
      "grad_norm": 0.3127965033054352,
      "learning_rate": 1e-05,
      "loss": 0.8932,
      "step": 34515
    },
    {
      "epoch": 30.575730735163862,
      "grad_norm": 0.2232382446527481,
      "learning_rate": 1e-05,
      "loss": 1.0382,
      "step": 34520
    },
    {
      "epoch": 30.58015943312666,
      "grad_norm": 0.25699564814567566,
      "learning_rate": 1e-05,
      "loss": 1.0486,
      "step": 34525
    },
    {
      "epoch": 30.58458813108946,
      "grad_norm": 0.32374072074890137,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 34530
    },
    {
      "epoch": 30.58901682905226,
      "grad_norm": 0.31563764810562134,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 34535
    },
    {
      "epoch": 30.593445527015056,
      "grad_norm": 0.2529101073741913,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 34540
    },
    {
      "epoch": 30.597874224977858,
      "grad_norm": 0.3023061752319336,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 34545
    },
    {
      "epoch": 30.602302922940655,
      "grad_norm": 0.25037938356399536,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 34550
    },
    {
      "epoch": 30.606731620903453,
      "grad_norm": 0.2639210522174835,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 34555
    },
    {
      "epoch": 30.611160318866254,
      "grad_norm": 0.22708819806575775,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 34560
    },
    {
      "epoch": 30.615589016829052,
      "grad_norm": 0.26096218824386597,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 34565
    },
    {
      "epoch": 30.62001771479185,
      "grad_norm": 0.2173214554786682,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 34570
    },
    {
      "epoch": 30.62444641275465,
      "grad_norm": 0.2637869715690613,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 34575
    },
    {
      "epoch": 30.62887511071745,
      "grad_norm": 0.2391861230134964,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 34580
    },
    {
      "epoch": 30.63330380868025,
      "grad_norm": 0.22723832726478577,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 34585
    },
    {
      "epoch": 30.637732506643047,
      "grad_norm": 0.2361384779214859,
      "learning_rate": 1e-05,
      "loss": 1.0161,
      "step": 34590
    },
    {
      "epoch": 30.642161204605845,
      "grad_norm": 0.2518586218357086,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 34595
    },
    {
      "epoch": 30.646589902568646,
      "grad_norm": 0.2275327891111374,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 34600
    },
    {
      "epoch": 30.651018600531444,
      "grad_norm": 0.2624276876449585,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 34605
    },
    {
      "epoch": 30.65544729849424,
      "grad_norm": 0.2679789662361145,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 34610
    },
    {
      "epoch": 30.659875996457043,
      "grad_norm": 0.2824432849884033,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 34615
    },
    {
      "epoch": 30.66430469441984,
      "grad_norm": 0.2547994554042816,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 34620
    },
    {
      "epoch": 30.668733392382638,
      "grad_norm": 0.20962588489055634,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 34625
    },
    {
      "epoch": 30.67316209034544,
      "grad_norm": 0.22719013690948486,
      "learning_rate": 1e-05,
      "loss": 1.0357,
      "step": 34630
    },
    {
      "epoch": 30.677590788308237,
      "grad_norm": 0.2688770592212677,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 34635
    },
    {
      "epoch": 30.682019486271038,
      "grad_norm": 0.22420749068260193,
      "learning_rate": 1e-05,
      "loss": 1.0203,
      "step": 34640
    },
    {
      "epoch": 30.686448184233836,
      "grad_norm": 0.2644655704498291,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 34645
    },
    {
      "epoch": 30.690876882196633,
      "grad_norm": 0.23710519075393677,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 34650
    },
    {
      "epoch": 30.695305580159435,
      "grad_norm": 0.2559422552585602,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 34655
    },
    {
      "epoch": 30.699734278122232,
      "grad_norm": 0.3059367835521698,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 34660
    },
    {
      "epoch": 30.70416297608503,
      "grad_norm": 0.2646207809448242,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 34665
    },
    {
      "epoch": 30.70859167404783,
      "grad_norm": 0.23796968162059784,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 34670
    },
    {
      "epoch": 30.71302037201063,
      "grad_norm": 0.29492005705833435,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 34675
    },
    {
      "epoch": 30.717449069973426,
      "grad_norm": 0.24444738030433655,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 34680
    },
    {
      "epoch": 30.721877767936228,
      "grad_norm": 0.2072303593158722,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 34685
    },
    {
      "epoch": 30.726306465899025,
      "grad_norm": 0.25287508964538574,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 34690
    },
    {
      "epoch": 30.730735163861823,
      "grad_norm": 0.28292641043663025,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 34695
    },
    {
      "epoch": 30.735163861824624,
      "grad_norm": 0.26195114850997925,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 34700
    },
    {
      "epoch": 30.739592559787422,
      "grad_norm": 0.2798844575881958,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 34705
    },
    {
      "epoch": 30.744021257750223,
      "grad_norm": 0.25873300433158875,
      "learning_rate": 1e-05,
      "loss": 0.9261,
      "step": 34710
    },
    {
      "epoch": 30.74844995571302,
      "grad_norm": 0.21659459173679352,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 34715
    },
    {
      "epoch": 30.75287865367582,
      "grad_norm": 0.25124579668045044,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 34720
    },
    {
      "epoch": 30.75730735163862,
      "grad_norm": 0.2484748363494873,
      "learning_rate": 1e-05,
      "loss": 1.0391,
      "step": 34725
    },
    {
      "epoch": 30.761736049601417,
      "grad_norm": 0.2469712346792221,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 34730
    },
    {
      "epoch": 30.766164747564215,
      "grad_norm": 0.23426826298236847,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 34735
    },
    {
      "epoch": 30.770593445527016,
      "grad_norm": 0.25094476342201233,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 34740
    },
    {
      "epoch": 30.775022143489814,
      "grad_norm": 0.23613260686397552,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 34745
    },
    {
      "epoch": 30.77945084145261,
      "grad_norm": 0.2904433608055115,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 34750
    },
    {
      "epoch": 30.783879539415413,
      "grad_norm": 0.2220756858587265,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 34755
    },
    {
      "epoch": 30.78830823737821,
      "grad_norm": 0.2451700121164322,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 34760
    },
    {
      "epoch": 30.79273693534101,
      "grad_norm": 0.25129827857017517,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 34765
    },
    {
      "epoch": 30.79716563330381,
      "grad_norm": 0.2640751004219055,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 34770
    },
    {
      "epoch": 30.801594331266607,
      "grad_norm": 0.22449372708797455,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 34775
    },
    {
      "epoch": 30.806023029229408,
      "grad_norm": 0.19808551669120789,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 34780
    },
    {
      "epoch": 30.810451727192206,
      "grad_norm": 0.24920284748077393,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 34785
    },
    {
      "epoch": 30.814880425155003,
      "grad_norm": 0.26047414541244507,
      "learning_rate": 1e-05,
      "loss": 1.0399,
      "step": 34790
    },
    {
      "epoch": 30.819309123117804,
      "grad_norm": 0.23394909501075745,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 34795
    },
    {
      "epoch": 30.823737821080602,
      "grad_norm": 0.2258889526128769,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 34800
    },
    {
      "epoch": 30.8281665190434,
      "grad_norm": 0.2383006066083908,
      "learning_rate": 1e-05,
      "loss": 1.0085,
      "step": 34805
    },
    {
      "epoch": 30.8325952170062,
      "grad_norm": 0.2643567621707916,
      "learning_rate": 1e-05,
      "loss": 0.9203,
      "step": 34810
    },
    {
      "epoch": 30.837023914969,
      "grad_norm": 0.22734931111335754,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 34815
    },
    {
      "epoch": 30.841452612931796,
      "grad_norm": 0.25397050380706787,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 34820
    },
    {
      "epoch": 30.845881310894598,
      "grad_norm": 0.27157706022262573,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 34825
    },
    {
      "epoch": 30.850310008857395,
      "grad_norm": 0.2450697124004364,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 34830
    },
    {
      "epoch": 30.854738706820196,
      "grad_norm": 0.21117836236953735,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 34835
    },
    {
      "epoch": 30.859167404782994,
      "grad_norm": 0.2756665349006653,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 34840
    },
    {
      "epoch": 30.86359610274579,
      "grad_norm": 0.24095040559768677,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 34845
    },
    {
      "epoch": 30.868024800708593,
      "grad_norm": 0.2671302258968353,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 34850
    },
    {
      "epoch": 30.87245349867139,
      "grad_norm": 0.22376534342765808,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 34855
    },
    {
      "epoch": 30.876882196634188,
      "grad_norm": 0.27429595589637756,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 34860
    },
    {
      "epoch": 30.88131089459699,
      "grad_norm": 0.24483215808868408,
      "learning_rate": 1e-05,
      "loss": 0.924,
      "step": 34865
    },
    {
      "epoch": 30.885739592559787,
      "grad_norm": 0.2594766318798065,
      "learning_rate": 1e-05,
      "loss": 1.0141,
      "step": 34870
    },
    {
      "epoch": 30.890168290522585,
      "grad_norm": 0.22799572348594666,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 34875
    },
    {
      "epoch": 30.894596988485386,
      "grad_norm": 0.22435466945171356,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 34880
    },
    {
      "epoch": 30.899025686448184,
      "grad_norm": 0.2150404304265976,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 34885
    },
    {
      "epoch": 30.903454384410985,
      "grad_norm": 0.27419155836105347,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 34890
    },
    {
      "epoch": 30.907883082373782,
      "grad_norm": 0.2596563398838043,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 34895
    },
    {
      "epoch": 30.91231178033658,
      "grad_norm": 0.28533440828323364,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 34900
    },
    {
      "epoch": 30.91674047829938,
      "grad_norm": 0.2412666231393814,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 34905
    },
    {
      "epoch": 30.92116917626218,
      "grad_norm": 0.22542257606983185,
      "learning_rate": 1e-05,
      "loss": 1.0391,
      "step": 34910
    },
    {
      "epoch": 30.925597874224977,
      "grad_norm": 0.22069807350635529,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 34915
    },
    {
      "epoch": 30.930026572187778,
      "grad_norm": 0.24135540425777435,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 34920
    },
    {
      "epoch": 30.934455270150576,
      "grad_norm": 0.2341495156288147,
      "learning_rate": 1e-05,
      "loss": 1.0268,
      "step": 34925
    },
    {
      "epoch": 30.938883968113373,
      "grad_norm": 0.22325782477855682,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 34930
    },
    {
      "epoch": 30.943312666076174,
      "grad_norm": 0.25892776250839233,
      "learning_rate": 1e-05,
      "loss": 0.9367,
      "step": 34935
    },
    {
      "epoch": 30.947741364038972,
      "grad_norm": 0.2588663101196289,
      "learning_rate": 1e-05,
      "loss": 1.0342,
      "step": 34940
    },
    {
      "epoch": 30.95217006200177,
      "grad_norm": 0.2352781891822815,
      "learning_rate": 1e-05,
      "loss": 1.0213,
      "step": 34945
    },
    {
      "epoch": 30.95659875996457,
      "grad_norm": 0.23250240087509155,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 34950
    },
    {
      "epoch": 30.96102745792737,
      "grad_norm": 0.2730807662010193,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 34955
    },
    {
      "epoch": 30.96545615589017,
      "grad_norm": 0.23375511169433594,
      "learning_rate": 1e-05,
      "loss": 1.0445,
      "step": 34960
    },
    {
      "epoch": 30.969884853852967,
      "grad_norm": 0.2920178174972534,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 34965
    },
    {
      "epoch": 30.974313551815765,
      "grad_norm": 0.2189139574766159,
      "learning_rate": 1e-05,
      "loss": 1.0159,
      "step": 34970
    },
    {
      "epoch": 30.978742249778566,
      "grad_norm": 0.27846527099609375,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 34975
    },
    {
      "epoch": 30.983170947741364,
      "grad_norm": 0.20329032838344574,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 34980
    },
    {
      "epoch": 30.98759964570416,
      "grad_norm": 0.2962951958179474,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 34985
    },
    {
      "epoch": 30.992028343666963,
      "grad_norm": 0.27440345287323,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 34990
    },
    {
      "epoch": 30.99645704162976,
      "grad_norm": 0.28769195079803467,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 34995
    },
    {
      "epoch": 31.000885739592558,
      "grad_norm": 0.2570668160915375,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 35000
    },
    {
      "epoch": 31.00531443755536,
      "grad_norm": 0.24807776510715485,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 35005
    },
    {
      "epoch": 31.009743135518157,
      "grad_norm": 0.2410086691379547,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 35010
    },
    {
      "epoch": 31.014171833480958,
      "grad_norm": 0.23655954003334045,
      "learning_rate": 1e-05,
      "loss": 0.9211,
      "step": 35015
    },
    {
      "epoch": 31.018600531443756,
      "grad_norm": 0.27551335096359253,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 35020
    },
    {
      "epoch": 31.023029229406553,
      "grad_norm": 0.2707156538963318,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 35025
    },
    {
      "epoch": 31.027457927369355,
      "grad_norm": 0.22851836681365967,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 35030
    },
    {
      "epoch": 31.031886625332152,
      "grad_norm": 0.31851282715797424,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 35035
    },
    {
      "epoch": 31.03631532329495,
      "grad_norm": 0.2293817549943924,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 35040
    },
    {
      "epoch": 31.04074402125775,
      "grad_norm": 0.24714027345180511,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 35045
    },
    {
      "epoch": 31.04517271922055,
      "grad_norm": 0.2505913972854614,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 35050
    },
    {
      "epoch": 31.049601417183347,
      "grad_norm": 0.275619238615036,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 35055
    },
    {
      "epoch": 31.054030115146148,
      "grad_norm": 0.24901467561721802,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 35060
    },
    {
      "epoch": 31.058458813108945,
      "grad_norm": 0.2166486233472824,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 35065
    },
    {
      "epoch": 31.062887511071747,
      "grad_norm": 0.20286552608013153,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 35070
    },
    {
      "epoch": 31.067316209034544,
      "grad_norm": 0.2698226571083069,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 35075
    },
    {
      "epoch": 31.071744906997342,
      "grad_norm": 0.2966356873512268,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 35080
    },
    {
      "epoch": 31.076173604960143,
      "grad_norm": 0.21147166192531586,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 35085
    },
    {
      "epoch": 31.08060230292294,
      "grad_norm": 0.24268680810928345,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 35090
    },
    {
      "epoch": 31.08503100088574,
      "grad_norm": 0.2531626522541046,
      "learning_rate": 1e-05,
      "loss": 1.0475,
      "step": 35095
    },
    {
      "epoch": 31.08945969884854,
      "grad_norm": 0.25439777970314026,
      "learning_rate": 1e-05,
      "loss": 1.0354,
      "step": 35100
    },
    {
      "epoch": 31.093888396811337,
      "grad_norm": 0.4166276752948761,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 35105
    },
    {
      "epoch": 31.098317094774135,
      "grad_norm": 0.2791493237018585,
      "learning_rate": 1e-05,
      "loss": 1.0185,
      "step": 35110
    },
    {
      "epoch": 31.102745792736936,
      "grad_norm": 0.2618753910064697,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 35115
    },
    {
      "epoch": 31.107174490699734,
      "grad_norm": 0.236869677901268,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 35120
    },
    {
      "epoch": 31.11160318866253,
      "grad_norm": 0.24910107254981995,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 35125
    },
    {
      "epoch": 31.116031886625333,
      "grad_norm": 0.2720644474029541,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 35130
    },
    {
      "epoch": 31.12046058458813,
      "grad_norm": 0.2633616626262665,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 35135
    },
    {
      "epoch": 31.12488928255093,
      "grad_norm": 0.288844496011734,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 35140
    },
    {
      "epoch": 31.12931798051373,
      "grad_norm": 0.2554263770580292,
      "learning_rate": 1e-05,
      "loss": 0.9311,
      "step": 35145
    },
    {
      "epoch": 31.133746678476527,
      "grad_norm": 0.28100958466529846,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 35150
    },
    {
      "epoch": 31.138175376439328,
      "grad_norm": 0.22852961719036102,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 35155
    },
    {
      "epoch": 31.142604074402126,
      "grad_norm": 0.21116362512111664,
      "learning_rate": 1e-05,
      "loss": 0.9068,
      "step": 35160
    },
    {
      "epoch": 31.147032772364923,
      "grad_norm": 0.2336074411869049,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 35165
    },
    {
      "epoch": 31.151461470327725,
      "grad_norm": 0.2704808712005615,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 35170
    },
    {
      "epoch": 31.155890168290522,
      "grad_norm": 0.26952216029167175,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 35175
    },
    {
      "epoch": 31.16031886625332,
      "grad_norm": 0.26170507073402405,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 35180
    },
    {
      "epoch": 31.16474756421612,
      "grad_norm": 0.282736212015152,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 35185
    },
    {
      "epoch": 31.16917626217892,
      "grad_norm": 0.23463398218154907,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 35190
    },
    {
      "epoch": 31.17360496014172,
      "grad_norm": 0.24890385568141937,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 35195
    },
    {
      "epoch": 31.178033658104518,
      "grad_norm": 0.23202618956565857,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 35200
    },
    {
      "epoch": 31.182462356067315,
      "grad_norm": 0.236739844083786,
      "learning_rate": 1e-05,
      "loss": 1.0408,
      "step": 35205
    },
    {
      "epoch": 31.186891054030117,
      "grad_norm": 0.29719746112823486,
      "learning_rate": 1e-05,
      "loss": 1.0371,
      "step": 35210
    },
    {
      "epoch": 31.191319751992914,
      "grad_norm": 0.26827743649482727,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 35215
    },
    {
      "epoch": 31.195748449955712,
      "grad_norm": 0.2626824975013733,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 35220
    },
    {
      "epoch": 31.200177147918513,
      "grad_norm": 0.2704109251499176,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 35225
    },
    {
      "epoch": 31.20460584588131,
      "grad_norm": 0.27876201272010803,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 35230
    },
    {
      "epoch": 31.20903454384411,
      "grad_norm": 0.2840498387813568,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 35235
    },
    {
      "epoch": 31.21346324180691,
      "grad_norm": 0.214911550283432,
      "learning_rate": 1e-05,
      "loss": 0.9174,
      "step": 35240
    },
    {
      "epoch": 31.217891939769707,
      "grad_norm": 0.25282469391822815,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 35245
    },
    {
      "epoch": 31.22232063773251,
      "grad_norm": 0.2781233787536621,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 35250
    },
    {
      "epoch": 31.226749335695306,
      "grad_norm": 0.21406224370002747,
      "learning_rate": 1e-05,
      "loss": 1.0258,
      "step": 35255
    },
    {
      "epoch": 31.231178033658104,
      "grad_norm": 0.22580429911613464,
      "learning_rate": 1e-05,
      "loss": 0.9236,
      "step": 35260
    },
    {
      "epoch": 31.235606731620905,
      "grad_norm": 0.21106700599193573,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 35265
    },
    {
      "epoch": 31.240035429583703,
      "grad_norm": 0.23357057571411133,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 35270
    },
    {
      "epoch": 31.2444641275465,
      "grad_norm": 0.24178718030452728,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 35275
    },
    {
      "epoch": 31.2488928255093,
      "grad_norm": 0.2801498770713806,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 35280
    },
    {
      "epoch": 31.2533215234721,
      "grad_norm": 0.25350314378738403,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 35285
    },
    {
      "epoch": 31.257750221434897,
      "grad_norm": 0.225112184882164,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 35290
    },
    {
      "epoch": 31.262178919397698,
      "grad_norm": 0.2511879801750183,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 35295
    },
    {
      "epoch": 31.266607617360496,
      "grad_norm": 0.28343436121940613,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 35300
    },
    {
      "epoch": 31.271036315323293,
      "grad_norm": 0.2766794264316559,
      "learning_rate": 1e-05,
      "loss": 1.0161,
      "step": 35305
    },
    {
      "epoch": 31.275465013286095,
      "grad_norm": 0.2446972131729126,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 35310
    },
    {
      "epoch": 31.279893711248892,
      "grad_norm": 0.25362056493759155,
      "learning_rate": 1e-05,
      "loss": 1.03,
      "step": 35315
    },
    {
      "epoch": 31.284322409211693,
      "grad_norm": 0.21366441249847412,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 35320
    },
    {
      "epoch": 31.28875110717449,
      "grad_norm": 0.22904399037361145,
      "learning_rate": 1e-05,
      "loss": 1.0376,
      "step": 35325
    },
    {
      "epoch": 31.29317980513729,
      "grad_norm": 0.2796422243118286,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 35330
    },
    {
      "epoch": 31.29760850310009,
      "grad_norm": 0.2913896143436432,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 35335
    },
    {
      "epoch": 31.302037201062888,
      "grad_norm": 0.23101086914539337,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 35340
    },
    {
      "epoch": 31.306465899025685,
      "grad_norm": 0.2066575437784195,
      "learning_rate": 1e-05,
      "loss": 0.9313,
      "step": 35345
    },
    {
      "epoch": 31.310894596988486,
      "grad_norm": 0.21769800782203674,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 35350
    },
    {
      "epoch": 31.315323294951284,
      "grad_norm": 0.2609650790691376,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 35355
    },
    {
      "epoch": 31.31975199291408,
      "grad_norm": 0.22938460111618042,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 35360
    },
    {
      "epoch": 31.324180690876883,
      "grad_norm": 0.24664834141731262,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 35365
    },
    {
      "epoch": 31.32860938883968,
      "grad_norm": 0.18712028861045837,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 35370
    },
    {
      "epoch": 31.333038086802482,
      "grad_norm": 0.2431710660457611,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 35375
    },
    {
      "epoch": 31.33746678476528,
      "grad_norm": 0.23196014761924744,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 35380
    },
    {
      "epoch": 31.341895482728077,
      "grad_norm": 0.2867108881473541,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 35385
    },
    {
      "epoch": 31.34632418069088,
      "grad_norm": 0.24104748666286469,
      "learning_rate": 1e-05,
      "loss": 1.0371,
      "step": 35390
    },
    {
      "epoch": 31.350752878653676,
      "grad_norm": 0.23968102037906647,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 35395
    },
    {
      "epoch": 31.355181576616474,
      "grad_norm": 0.23314544558525085,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 35400
    },
    {
      "epoch": 31.359610274579275,
      "grad_norm": 0.2784644067287445,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 35405
    },
    {
      "epoch": 31.364038972542073,
      "grad_norm": 0.21619026362895966,
      "learning_rate": 1e-05,
      "loss": 0.9192,
      "step": 35410
    },
    {
      "epoch": 31.36846767050487,
      "grad_norm": 0.23371988534927368,
      "learning_rate": 1e-05,
      "loss": 1.0272,
      "step": 35415
    },
    {
      "epoch": 31.37289636846767,
      "grad_norm": 0.23752902448177338,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 35420
    },
    {
      "epoch": 31.37732506643047,
      "grad_norm": 0.24812424182891846,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 35425
    },
    {
      "epoch": 31.381753764393267,
      "grad_norm": 0.22310978174209595,
      "learning_rate": 1e-05,
      "loss": 1.0111,
      "step": 35430
    },
    {
      "epoch": 31.386182462356068,
      "grad_norm": 0.24477465450763702,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 35435
    },
    {
      "epoch": 31.390611160318866,
      "grad_norm": 0.20999886095523834,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 35440
    },
    {
      "epoch": 31.395039858281667,
      "grad_norm": 0.22738344967365265,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 35445
    },
    {
      "epoch": 31.399468556244464,
      "grad_norm": 0.2496766597032547,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 35450
    },
    {
      "epoch": 31.403897254207262,
      "grad_norm": 0.2513379752635956,
      "learning_rate": 1e-05,
      "loss": 0.9245,
      "step": 35455
    },
    {
      "epoch": 31.408325952170063,
      "grad_norm": 0.22780577838420868,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 35460
    },
    {
      "epoch": 31.41275465013286,
      "grad_norm": 0.23735111951828003,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 35465
    },
    {
      "epoch": 31.41718334809566,
      "grad_norm": 0.2360881119966507,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 35470
    },
    {
      "epoch": 31.42161204605846,
      "grad_norm": 0.23891125619411469,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 35475
    },
    {
      "epoch": 31.426040744021257,
      "grad_norm": 0.24041160941123962,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 35480
    },
    {
      "epoch": 31.430469441984055,
      "grad_norm": 0.24010957777500153,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 35485
    },
    {
      "epoch": 31.434898139946856,
      "grad_norm": 0.22992154955863953,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 35490
    },
    {
      "epoch": 31.439326837909654,
      "grad_norm": 0.3005712330341339,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 35495
    },
    {
      "epoch": 31.443755535872455,
      "grad_norm": 0.2488047182559967,
      "learning_rate": 1e-05,
      "loss": 1.0496,
      "step": 35500
    },
    {
      "epoch": 31.448184233835253,
      "grad_norm": 0.2128056436777115,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 35505
    },
    {
      "epoch": 31.45261293179805,
      "grad_norm": 0.22438634932041168,
      "learning_rate": 1e-05,
      "loss": 1.0507,
      "step": 35510
    },
    {
      "epoch": 31.45704162976085,
      "grad_norm": 0.2282673865556717,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 35515
    },
    {
      "epoch": 31.46147032772365,
      "grad_norm": 0.235970601439476,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 35520
    },
    {
      "epoch": 31.465899025686447,
      "grad_norm": 0.2444128692150116,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 35525
    },
    {
      "epoch": 31.47032772364925,
      "grad_norm": 0.25187110900878906,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 35530
    },
    {
      "epoch": 31.474756421612046,
      "grad_norm": 0.23714660108089447,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 35535
    },
    {
      "epoch": 31.479185119574844,
      "grad_norm": 0.27497971057891846,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 35540
    },
    {
      "epoch": 31.483613817537645,
      "grad_norm": 0.2124537229537964,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 35545
    },
    {
      "epoch": 31.488042515500442,
      "grad_norm": 0.22905023396015167,
      "learning_rate": 1e-05,
      "loss": 1.0189,
      "step": 35550
    },
    {
      "epoch": 31.49247121346324,
      "grad_norm": 0.3222437798976898,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 35555
    },
    {
      "epoch": 31.49689991142604,
      "grad_norm": 0.2093300074338913,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 35560
    },
    {
      "epoch": 31.50132860938884,
      "grad_norm": 0.237563818693161,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 35565
    },
    {
      "epoch": 31.50575730735164,
      "grad_norm": 0.2434566169977188,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 35570
    },
    {
      "epoch": 31.510186005314438,
      "grad_norm": 0.2535463571548462,
      "learning_rate": 1e-05,
      "loss": 1.0228,
      "step": 35575
    },
    {
      "epoch": 31.514614703277235,
      "grad_norm": 0.2376113086938858,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 35580
    },
    {
      "epoch": 31.519043401240037,
      "grad_norm": 0.20796199142932892,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 35585
    },
    {
      "epoch": 31.523472099202834,
      "grad_norm": 0.23710638284683228,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 35590
    },
    {
      "epoch": 31.527900797165632,
      "grad_norm": 0.2609461545944214,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 35595
    },
    {
      "epoch": 31.532329495128433,
      "grad_norm": 0.2826554775238037,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 35600
    },
    {
      "epoch": 31.53675819309123,
      "grad_norm": 0.25112223625183105,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 35605
    },
    {
      "epoch": 31.54118689105403,
      "grad_norm": 0.2330673485994339,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 35610
    },
    {
      "epoch": 31.54561558901683,
      "grad_norm": 0.2788885235786438,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 35615
    },
    {
      "epoch": 31.550044286979627,
      "grad_norm": 0.23976612091064453,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 35620
    },
    {
      "epoch": 31.55447298494243,
      "grad_norm": 0.2367788553237915,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 35625
    },
    {
      "epoch": 31.558901682905226,
      "grad_norm": 0.27491727471351624,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 35630
    },
    {
      "epoch": 31.563330380868024,
      "grad_norm": 0.263215035200119,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 35635
    },
    {
      "epoch": 31.567759078830825,
      "grad_norm": 0.2804422974586487,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 35640
    },
    {
      "epoch": 31.572187776793623,
      "grad_norm": 0.2390197515487671,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 35645
    },
    {
      "epoch": 31.57661647475642,
      "grad_norm": 0.26763594150543213,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 35650
    },
    {
      "epoch": 31.58104517271922,
      "grad_norm": 0.22215326130390167,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 35655
    },
    {
      "epoch": 31.58547387068202,
      "grad_norm": 0.2609909772872925,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 35660
    },
    {
      "epoch": 31.589902568644817,
      "grad_norm": 0.20740638673305511,
      "learning_rate": 1e-05,
      "loss": 1.0779,
      "step": 35665
    },
    {
      "epoch": 31.594331266607618,
      "grad_norm": 0.2855990529060364,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 35670
    },
    {
      "epoch": 31.598759964570416,
      "grad_norm": 0.25453928112983704,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 35675
    },
    {
      "epoch": 31.603188662533213,
      "grad_norm": 0.2798418402671814,
      "learning_rate": 1e-05,
      "loss": 1.0516,
      "step": 35680
    },
    {
      "epoch": 31.607617360496015,
      "grad_norm": 0.27569082379341125,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 35685
    },
    {
      "epoch": 31.612046058458812,
      "grad_norm": 0.2389264851808548,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 35690
    },
    {
      "epoch": 31.616474756421614,
      "grad_norm": 0.2625390291213989,
      "learning_rate": 1e-05,
      "loss": 1.0054,
      "step": 35695
    },
    {
      "epoch": 31.62090345438441,
      "grad_norm": 0.2644560635089874,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 35700
    },
    {
      "epoch": 31.62533215234721,
      "grad_norm": 0.24798107147216797,
      "learning_rate": 1e-05,
      "loss": 0.9174,
      "step": 35705
    },
    {
      "epoch": 31.62976085031001,
      "grad_norm": 0.2319185584783554,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 35710
    },
    {
      "epoch": 31.634189548272808,
      "grad_norm": 0.2435673177242279,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 35715
    },
    {
      "epoch": 31.638618246235605,
      "grad_norm": 0.24127261340618134,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 35720
    },
    {
      "epoch": 31.643046944198407,
      "grad_norm": 0.2230539619922638,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 35725
    },
    {
      "epoch": 31.647475642161204,
      "grad_norm": 0.2332291603088379,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 35730
    },
    {
      "epoch": 31.651904340124002,
      "grad_norm": 0.2614711821079254,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 35735
    },
    {
      "epoch": 31.656333038086803,
      "grad_norm": 0.23311543464660645,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 35740
    },
    {
      "epoch": 31.6607617360496,
      "grad_norm": 0.3181777596473694,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 35745
    },
    {
      "epoch": 31.665190434012402,
      "grad_norm": 0.2673760652542114,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 35750
    },
    {
      "epoch": 31.6696191319752,
      "grad_norm": 0.2845006287097931,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 35755
    },
    {
      "epoch": 31.674047829937997,
      "grad_norm": 0.2288849800825119,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 35760
    },
    {
      "epoch": 31.6784765279008,
      "grad_norm": 0.2399853765964508,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 35765
    },
    {
      "epoch": 31.682905225863596,
      "grad_norm": 0.27550503611564636,
      "learning_rate": 1e-05,
      "loss": 0.9214,
      "step": 35770
    },
    {
      "epoch": 31.687333923826394,
      "grad_norm": 0.2351623773574829,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 35775
    },
    {
      "epoch": 31.691762621789195,
      "grad_norm": 0.3011205494403839,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 35780
    },
    {
      "epoch": 31.696191319751993,
      "grad_norm": 0.24457640945911407,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 35785
    },
    {
      "epoch": 31.70062001771479,
      "grad_norm": 0.2897714078426361,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 35790
    },
    {
      "epoch": 31.70504871567759,
      "grad_norm": 0.27010974287986755,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 35795
    },
    {
      "epoch": 31.70947741364039,
      "grad_norm": 0.24359668791294098,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 35800
    },
    {
      "epoch": 31.713906111603187,
      "grad_norm": 0.2919084131717682,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 35805
    },
    {
      "epoch": 31.718334809565988,
      "grad_norm": 0.2341143786907196,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 35810
    },
    {
      "epoch": 31.722763507528786,
      "grad_norm": 0.24202905595302582,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 35815
    },
    {
      "epoch": 31.727192205491587,
      "grad_norm": 0.23509721457958221,
      "learning_rate": 1e-05,
      "loss": 0.9101,
      "step": 35820
    },
    {
      "epoch": 31.731620903454385,
      "grad_norm": 0.2174912691116333,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 35825
    },
    {
      "epoch": 31.736049601417182,
      "grad_norm": 0.24426499009132385,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 35830
    },
    {
      "epoch": 31.740478299379983,
      "grad_norm": 0.238041490316391,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 35835
    },
    {
      "epoch": 31.74490699734278,
      "grad_norm": 0.26409682631492615,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 35840
    },
    {
      "epoch": 31.74933569530558,
      "grad_norm": 0.23438508808612823,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 35845
    },
    {
      "epoch": 31.75376439326838,
      "grad_norm": 0.24072594940662384,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 35850
    },
    {
      "epoch": 31.758193091231178,
      "grad_norm": 0.2742479145526886,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 35855
    },
    {
      "epoch": 31.762621789193975,
      "grad_norm": 0.24791787564754486,
      "learning_rate": 1e-05,
      "loss": 0.9306,
      "step": 35860
    },
    {
      "epoch": 31.767050487156776,
      "grad_norm": 0.24108822643756866,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 35865
    },
    {
      "epoch": 31.771479185119574,
      "grad_norm": 0.2175181359052658,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 35870
    },
    {
      "epoch": 31.775907883082375,
      "grad_norm": 0.2114318460226059,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 35875
    },
    {
      "epoch": 31.780336581045173,
      "grad_norm": 0.2672019600868225,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 35880
    },
    {
      "epoch": 31.78476527900797,
      "grad_norm": 0.3060592710971832,
      "learning_rate": 1e-05,
      "loss": 0.9342,
      "step": 35885
    },
    {
      "epoch": 31.789193976970772,
      "grad_norm": 0.22689026594161987,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 35890
    },
    {
      "epoch": 31.79362267493357,
      "grad_norm": 0.23403868079185486,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 35895
    },
    {
      "epoch": 31.798051372896367,
      "grad_norm": 0.2175912857055664,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 35900
    },
    {
      "epoch": 31.80248007085917,
      "grad_norm": 0.21413464844226837,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 35905
    },
    {
      "epoch": 31.806908768821966,
      "grad_norm": 0.27171841263771057,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 35910
    },
    {
      "epoch": 31.811337466784764,
      "grad_norm": 0.24016597867012024,
      "learning_rate": 1e-05,
      "loss": 1.0174,
      "step": 35915
    },
    {
      "epoch": 31.815766164747565,
      "grad_norm": 0.24857477843761444,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 35920
    },
    {
      "epoch": 31.820194862710363,
      "grad_norm": 0.22688615322113037,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 35925
    },
    {
      "epoch": 31.824623560673164,
      "grad_norm": 0.2265274077653885,
      "learning_rate": 1e-05,
      "loss": 0.9139,
      "step": 35930
    },
    {
      "epoch": 31.82905225863596,
      "grad_norm": 0.2208435982465744,
      "learning_rate": 1e-05,
      "loss": 1.0199,
      "step": 35935
    },
    {
      "epoch": 31.83348095659876,
      "grad_norm": 0.24390463531017303,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 35940
    },
    {
      "epoch": 31.83790965456156,
      "grad_norm": 0.26262837648391724,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 35945
    },
    {
      "epoch": 31.842338352524358,
      "grad_norm": 0.23161430656909943,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 35950
    },
    {
      "epoch": 31.846767050487156,
      "grad_norm": 0.22293663024902344,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 35955
    },
    {
      "epoch": 31.851195748449957,
      "grad_norm": 0.26411643624305725,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 35960
    },
    {
      "epoch": 31.855624446412754,
      "grad_norm": 0.2719457149505615,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 35965
    },
    {
      "epoch": 31.860053144375552,
      "grad_norm": 0.2952626049518585,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 35970
    },
    {
      "epoch": 31.864481842338353,
      "grad_norm": 0.23821762204170227,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 35975
    },
    {
      "epoch": 31.86891054030115,
      "grad_norm": 0.21682235598564148,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 35980
    },
    {
      "epoch": 31.873339238263952,
      "grad_norm": 0.24717587232589722,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 35985
    },
    {
      "epoch": 31.87776793622675,
      "grad_norm": 0.24953578412532806,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 35990
    },
    {
      "epoch": 31.882196634189548,
      "grad_norm": 0.26740214228630066,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 35995
    },
    {
      "epoch": 31.88662533215235,
      "grad_norm": 0.22675389051437378,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 36000
    },
    {
      "epoch": 31.891054030115146,
      "grad_norm": 0.2416391372680664,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 36005
    },
    {
      "epoch": 31.895482728077944,
      "grad_norm": 0.2632206082344055,
      "learning_rate": 1e-05,
      "loss": 0.93,
      "step": 36010
    },
    {
      "epoch": 31.899911426040745,
      "grad_norm": 0.254170686006546,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 36015
    },
    {
      "epoch": 31.904340124003543,
      "grad_norm": 0.27379873394966125,
      "learning_rate": 1e-05,
      "loss": 1.023,
      "step": 36020
    },
    {
      "epoch": 31.90876882196634,
      "grad_norm": 0.29201215505599976,
      "learning_rate": 1e-05,
      "loss": 1.0441,
      "step": 36025
    },
    {
      "epoch": 31.913197519929142,
      "grad_norm": 0.2925633192062378,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 36030
    },
    {
      "epoch": 31.91762621789194,
      "grad_norm": 0.22624832391738892,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 36035
    },
    {
      "epoch": 31.922054915854737,
      "grad_norm": 0.24902020394802094,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 36040
    },
    {
      "epoch": 31.92648361381754,
      "grad_norm": 0.22631345689296722,
      "learning_rate": 1e-05,
      "loss": 0.9106,
      "step": 36045
    },
    {
      "epoch": 31.930912311780336,
      "grad_norm": 0.2157122641801834,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 36050
    },
    {
      "epoch": 31.935341009743137,
      "grad_norm": 0.2781873643398285,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 36055
    },
    {
      "epoch": 31.939769707705935,
      "grad_norm": 0.22202855348587036,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 36060
    },
    {
      "epoch": 31.944198405668732,
      "grad_norm": 0.26727306842803955,
      "learning_rate": 1e-05,
      "loss": 1.0339,
      "step": 36065
    },
    {
      "epoch": 31.948627103631534,
      "grad_norm": 0.23619753122329712,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 36070
    },
    {
      "epoch": 31.95305580159433,
      "grad_norm": 0.24048994481563568,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 36075
    },
    {
      "epoch": 31.95748449955713,
      "grad_norm": 0.30554598569869995,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 36080
    },
    {
      "epoch": 31.96191319751993,
      "grad_norm": 0.23777183890342712,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 36085
    },
    {
      "epoch": 31.966341895482728,
      "grad_norm": 0.23298799991607666,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 36090
    },
    {
      "epoch": 31.970770593445526,
      "grad_norm": 0.22514529526233673,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 36095
    },
    {
      "epoch": 31.975199291408327,
      "grad_norm": 0.22329914569854736,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 36100
    },
    {
      "epoch": 31.979627989371124,
      "grad_norm": 0.23244720697402954,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 36105
    },
    {
      "epoch": 31.984056687333926,
      "grad_norm": 0.2390136420726776,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 36110
    },
    {
      "epoch": 31.988485385296723,
      "grad_norm": 0.24254535138607025,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 36115
    },
    {
      "epoch": 31.99291408325952,
      "grad_norm": 0.21646682918071747,
      "learning_rate": 1e-05,
      "loss": 1.0331,
      "step": 36120
    },
    {
      "epoch": 31.997342781222322,
      "grad_norm": 0.24873514473438263,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 36125
    },
    {
      "epoch": 32.001771479185116,
      "grad_norm": 0.270109087228775,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 36130
    },
    {
      "epoch": 32.00620017714792,
      "grad_norm": 0.23483222723007202,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 36135
    },
    {
      "epoch": 32.01062887511072,
      "grad_norm": 0.21630969643592834,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 36140
    },
    {
      "epoch": 32.01505757307352,
      "grad_norm": 0.22074012458324432,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 36145
    },
    {
      "epoch": 32.019486271036314,
      "grad_norm": 0.19626078009605408,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 36150
    },
    {
      "epoch": 32.023914968999115,
      "grad_norm": 0.25371024012565613,
      "learning_rate": 1e-05,
      "loss": 1.0171,
      "step": 36155
    },
    {
      "epoch": 32.028343666961916,
      "grad_norm": 0.3242085874080658,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 36160
    },
    {
      "epoch": 32.03277236492471,
      "grad_norm": 0.3156336545944214,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 36165
    },
    {
      "epoch": 32.03720106288751,
      "grad_norm": 0.24700365960597992,
      "learning_rate": 1e-05,
      "loss": 0.9152,
      "step": 36170
    },
    {
      "epoch": 32.04162976085031,
      "grad_norm": 0.21226954460144043,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 36175
    },
    {
      "epoch": 32.04605845881311,
      "grad_norm": 0.24382300674915314,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 36180
    },
    {
      "epoch": 32.05048715677591,
      "grad_norm": 0.23433832824230194,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 36185
    },
    {
      "epoch": 32.05491585473871,
      "grad_norm": 0.22456440329551697,
      "learning_rate": 1e-05,
      "loss": 1.0184,
      "step": 36190
    },
    {
      "epoch": 32.0593445527015,
      "grad_norm": 0.2309112548828125,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 36195
    },
    {
      "epoch": 32.063773250664305,
      "grad_norm": 0.3553697466850281,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 36200
    },
    {
      "epoch": 32.068201948627106,
      "grad_norm": 0.21518997848033905,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 36205
    },
    {
      "epoch": 32.0726306465899,
      "grad_norm": 0.26817819476127625,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 36210
    },
    {
      "epoch": 32.0770593445527,
      "grad_norm": 0.24733425676822662,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 36215
    },
    {
      "epoch": 32.0814880425155,
      "grad_norm": 0.26431870460510254,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 36220
    },
    {
      "epoch": 32.0859167404783,
      "grad_norm": 0.2452416718006134,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 36225
    },
    {
      "epoch": 32.0903454384411,
      "grad_norm": 0.24736879765987396,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 36230
    },
    {
      "epoch": 32.0947741364039,
      "grad_norm": 0.24133014678955078,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 36235
    },
    {
      "epoch": 32.09920283436669,
      "grad_norm": 0.24304157495498657,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 36240
    },
    {
      "epoch": 32.103631532329494,
      "grad_norm": 0.227181077003479,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 36245
    },
    {
      "epoch": 32.108060230292296,
      "grad_norm": 0.20828892290592194,
      "learning_rate": 1e-05,
      "loss": 1.054,
      "step": 36250
    },
    {
      "epoch": 32.11248892825509,
      "grad_norm": 0.21861423552036285,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 36255
    },
    {
      "epoch": 32.11691762621789,
      "grad_norm": 0.2248683124780655,
      "learning_rate": 1e-05,
      "loss": 1.0217,
      "step": 36260
    },
    {
      "epoch": 32.12134632418069,
      "grad_norm": 0.21037541329860687,
      "learning_rate": 1e-05,
      "loss": 0.9381,
      "step": 36265
    },
    {
      "epoch": 32.12577502214349,
      "grad_norm": 0.26288631558418274,
      "learning_rate": 1e-05,
      "loss": 1.0111,
      "step": 36270
    },
    {
      "epoch": 32.13020372010629,
      "grad_norm": 0.24942642450332642,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 36275
    },
    {
      "epoch": 32.13463241806909,
      "grad_norm": 0.26169219613075256,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 36280
    },
    {
      "epoch": 32.13906111603189,
      "grad_norm": 0.2686249613761902,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 36285
    },
    {
      "epoch": 32.143489813994684,
      "grad_norm": 0.28162044286727905,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 36290
    },
    {
      "epoch": 32.147918511957485,
      "grad_norm": 0.28954973816871643,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 36295
    },
    {
      "epoch": 32.152347209920286,
      "grad_norm": 0.25665611028671265,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 36300
    },
    {
      "epoch": 32.15677590788308,
      "grad_norm": 0.31064364314079285,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 36305
    },
    {
      "epoch": 32.16120460584588,
      "grad_norm": 0.2692579925060272,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 36310
    },
    {
      "epoch": 32.16563330380868,
      "grad_norm": 0.22714948654174805,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 36315
    },
    {
      "epoch": 32.17006200177148,
      "grad_norm": 0.3256995975971222,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 36320
    },
    {
      "epoch": 32.17449069973428,
      "grad_norm": 0.24564258754253387,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 36325
    },
    {
      "epoch": 32.17891939769708,
      "grad_norm": 0.23182697594165802,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 36330
    },
    {
      "epoch": 32.18334809565987,
      "grad_norm": 0.2725316286087036,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 36335
    },
    {
      "epoch": 32.187776793622675,
      "grad_norm": 0.23019357025623322,
      "learning_rate": 1e-05,
      "loss": 0.9239,
      "step": 36340
    },
    {
      "epoch": 32.192205491585476,
      "grad_norm": 0.2802714705467224,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 36345
    },
    {
      "epoch": 32.19663418954827,
      "grad_norm": 0.238783061504364,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 36350
    },
    {
      "epoch": 32.20106288751107,
      "grad_norm": 0.23554763197898865,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 36355
    },
    {
      "epoch": 32.20549158547387,
      "grad_norm": 0.23464348912239075,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 36360
    },
    {
      "epoch": 32.20992028343667,
      "grad_norm": 0.2215508371591568,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 36365
    },
    {
      "epoch": 32.21434898139947,
      "grad_norm": 0.23354178667068481,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 36370
    },
    {
      "epoch": 32.21877767936227,
      "grad_norm": 0.2414330393075943,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 36375
    },
    {
      "epoch": 32.22320637732506,
      "grad_norm": 0.2330123782157898,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 36380
    },
    {
      "epoch": 32.227635075287864,
      "grad_norm": 0.23892280459403992,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 36385
    },
    {
      "epoch": 32.232063773250665,
      "grad_norm": 0.2523578405380249,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 36390
    },
    {
      "epoch": 32.23649247121347,
      "grad_norm": 0.23887327313423157,
      "learning_rate": 1e-05,
      "loss": 1.0389,
      "step": 36395
    },
    {
      "epoch": 32.24092116917626,
      "grad_norm": 0.31854650378227234,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 36400
    },
    {
      "epoch": 32.24534986713906,
      "grad_norm": 0.24610836803913116,
      "learning_rate": 1e-05,
      "loss": 1.0341,
      "step": 36405
    },
    {
      "epoch": 32.24977856510186,
      "grad_norm": 0.2544173002243042,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 36410
    },
    {
      "epoch": 32.25420726306466,
      "grad_norm": 0.22621837258338928,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 36415
    },
    {
      "epoch": 32.25863596102746,
      "grad_norm": 0.3374495208263397,
      "learning_rate": 1e-05,
      "loss": 1.0304,
      "step": 36420
    },
    {
      "epoch": 32.26306465899026,
      "grad_norm": 0.25838255882263184,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 36425
    },
    {
      "epoch": 32.267493356953054,
      "grad_norm": 0.2937672436237335,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 36430
    },
    {
      "epoch": 32.271922054915855,
      "grad_norm": 0.27002236247062683,
      "learning_rate": 1e-05,
      "loss": 1.0426,
      "step": 36435
    },
    {
      "epoch": 32.276350752878656,
      "grad_norm": 0.3041854500770569,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 36440
    },
    {
      "epoch": 32.28077945084145,
      "grad_norm": 0.24303793907165527,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 36445
    },
    {
      "epoch": 32.28520814880425,
      "grad_norm": 0.2140151411294937,
      "learning_rate": 1e-05,
      "loss": 1.024,
      "step": 36450
    },
    {
      "epoch": 32.28963684676705,
      "grad_norm": 0.22937098145484924,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 36455
    },
    {
      "epoch": 32.29406554472985,
      "grad_norm": 0.23301666975021362,
      "learning_rate": 1e-05,
      "loss": 1.0324,
      "step": 36460
    },
    {
      "epoch": 32.29849424269265,
      "grad_norm": 0.2180456519126892,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 36465
    },
    {
      "epoch": 32.30292294065545,
      "grad_norm": 0.26369139552116394,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 36470
    },
    {
      "epoch": 32.30735163861824,
      "grad_norm": 0.22635501623153687,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 36475
    },
    {
      "epoch": 32.311780336581045,
      "grad_norm": 0.21336348354816437,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 36480
    },
    {
      "epoch": 32.316209034543846,
      "grad_norm": 0.24370889365673065,
      "learning_rate": 1e-05,
      "loss": 1.0561,
      "step": 36485
    },
    {
      "epoch": 32.32063773250664,
      "grad_norm": 0.20133469998836517,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 36490
    },
    {
      "epoch": 32.32506643046944,
      "grad_norm": 0.20760925114154816,
      "learning_rate": 1e-05,
      "loss": 1.0519,
      "step": 36495
    },
    {
      "epoch": 32.32949512843224,
      "grad_norm": 0.2572067677974701,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 36500
    },
    {
      "epoch": 32.333923826395036,
      "grad_norm": 0.2773112654685974,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 36505
    },
    {
      "epoch": 32.33835252435784,
      "grad_norm": 0.2422502040863037,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 36510
    },
    {
      "epoch": 32.34278122232064,
      "grad_norm": 0.22528409957885742,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 36515
    },
    {
      "epoch": 32.34720992028344,
      "grad_norm": 0.2012006640434265,
      "learning_rate": 1e-05,
      "loss": 1.0331,
      "step": 36520
    },
    {
      "epoch": 32.351638618246234,
      "grad_norm": 0.26770204305648804,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 36525
    },
    {
      "epoch": 32.356067316209035,
      "grad_norm": 0.31134894490242004,
      "learning_rate": 1e-05,
      "loss": 0.9116,
      "step": 36530
    },
    {
      "epoch": 32.36049601417184,
      "grad_norm": 0.27741578221321106,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 36535
    },
    {
      "epoch": 32.36492471213463,
      "grad_norm": 0.2827765941619873,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 36540
    },
    {
      "epoch": 32.36935341009743,
      "grad_norm": 0.2371225357055664,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 36545
    },
    {
      "epoch": 32.37378210806023,
      "grad_norm": 0.24451832473278046,
      "learning_rate": 1e-05,
      "loss": 1.0431,
      "step": 36550
    },
    {
      "epoch": 32.37821080602303,
      "grad_norm": 0.2703625559806824,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 36555
    },
    {
      "epoch": 32.38263950398583,
      "grad_norm": 0.27809298038482666,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 36560
    },
    {
      "epoch": 32.38706820194863,
      "grad_norm": 0.24478842318058014,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 36565
    },
    {
      "epoch": 32.391496899911424,
      "grad_norm": 0.23801636695861816,
      "learning_rate": 1e-05,
      "loss": 1.0367,
      "step": 36570
    },
    {
      "epoch": 32.395925597874225,
      "grad_norm": 0.2981101870536804,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 36575
    },
    {
      "epoch": 32.400354295837026,
      "grad_norm": 0.2488527148962021,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 36580
    },
    {
      "epoch": 32.40478299379982,
      "grad_norm": 0.26797574758529663,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 36585
    },
    {
      "epoch": 32.40921169176262,
      "grad_norm": 0.25990450382232666,
      "learning_rate": 1e-05,
      "loss": 1.0207,
      "step": 36590
    },
    {
      "epoch": 32.41364038972542,
      "grad_norm": 0.2383684664964676,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 36595
    },
    {
      "epoch": 32.41806908768822,
      "grad_norm": 0.24780191481113434,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 36600
    },
    {
      "epoch": 32.42249778565102,
      "grad_norm": 0.21007679402828217,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 36605
    },
    {
      "epoch": 32.42692648361382,
      "grad_norm": 0.22435930371284485,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 36610
    },
    {
      "epoch": 32.43135518157661,
      "grad_norm": 0.2192358672618866,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 36615
    },
    {
      "epoch": 32.435783879539414,
      "grad_norm": 0.23997455835342407,
      "learning_rate": 1e-05,
      "loss": 1.0358,
      "step": 36620
    },
    {
      "epoch": 32.440212577502216,
      "grad_norm": 0.2107658088207245,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 36625
    },
    {
      "epoch": 32.44464127546502,
      "grad_norm": 0.19654399156570435,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 36630
    },
    {
      "epoch": 32.44906997342781,
      "grad_norm": 0.2635065019130707,
      "learning_rate": 1e-05,
      "loss": 1.0292,
      "step": 36635
    },
    {
      "epoch": 32.45349867139061,
      "grad_norm": 0.19647496938705444,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 36640
    },
    {
      "epoch": 32.45792736935341,
      "grad_norm": 0.2359771728515625,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 36645
    },
    {
      "epoch": 32.46235606731621,
      "grad_norm": 0.23199866712093353,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 36650
    },
    {
      "epoch": 32.46678476527901,
      "grad_norm": 0.24042458832263947,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 36655
    },
    {
      "epoch": 32.47121346324181,
      "grad_norm": 0.20143255591392517,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 36660
    },
    {
      "epoch": 32.475642161204604,
      "grad_norm": 0.22234395146369934,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 36665
    },
    {
      "epoch": 32.480070859167405,
      "grad_norm": 0.25506138801574707,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 36670
    },
    {
      "epoch": 32.484499557130206,
      "grad_norm": 0.25806450843811035,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 36675
    },
    {
      "epoch": 32.488928255093,
      "grad_norm": 0.26445791125297546,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 36680
    },
    {
      "epoch": 32.4933569530558,
      "grad_norm": 0.21333833038806915,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 36685
    },
    {
      "epoch": 32.4977856510186,
      "grad_norm": 0.23295442759990692,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 36690
    },
    {
      "epoch": 32.5022143489814,
      "grad_norm": 0.2343805730342865,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 36695
    },
    {
      "epoch": 32.5066430469442,
      "grad_norm": 0.24746939539909363,
      "learning_rate": 1e-05,
      "loss": 1.0295,
      "step": 36700
    },
    {
      "epoch": 32.511071744907,
      "grad_norm": 0.24456191062927246,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 36705
    },
    {
      "epoch": 32.515500442869794,
      "grad_norm": 0.2585187256336212,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 36710
    },
    {
      "epoch": 32.519929140832595,
      "grad_norm": 0.23885998129844666,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 36715
    },
    {
      "epoch": 32.524357838795396,
      "grad_norm": 0.27841898798942566,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 36720
    },
    {
      "epoch": 32.52878653675819,
      "grad_norm": 0.2269294261932373,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 36725
    },
    {
      "epoch": 32.53321523472099,
      "grad_norm": 0.2419809103012085,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 36730
    },
    {
      "epoch": 32.53764393268379,
      "grad_norm": 0.2230791598558426,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 36735
    },
    {
      "epoch": 32.54207263064659,
      "grad_norm": 0.25819098949432373,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 36740
    },
    {
      "epoch": 32.54650132860939,
      "grad_norm": 0.2441270500421524,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 36745
    },
    {
      "epoch": 32.55093002657219,
      "grad_norm": 0.27701112627983093,
      "learning_rate": 1e-05,
      "loss": 1.0354,
      "step": 36750
    },
    {
      "epoch": 32.55535872453498,
      "grad_norm": 0.24796810746192932,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 36755
    },
    {
      "epoch": 32.559787422497784,
      "grad_norm": 0.2414970099925995,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 36760
    },
    {
      "epoch": 32.564216120460586,
      "grad_norm": 0.2906976640224457,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 36765
    },
    {
      "epoch": 32.56864481842339,
      "grad_norm": 0.2600421905517578,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 36770
    },
    {
      "epoch": 32.57307351638618,
      "grad_norm": 0.31107842922210693,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 36775
    },
    {
      "epoch": 32.57750221434898,
      "grad_norm": 0.2882780134677887,
      "learning_rate": 1e-05,
      "loss": 0.9277,
      "step": 36780
    },
    {
      "epoch": 32.58193091231178,
      "grad_norm": 0.27277156710624695,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 36785
    },
    {
      "epoch": 32.58635961027458,
      "grad_norm": 0.26888182759284973,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 36790
    },
    {
      "epoch": 32.59078830823738,
      "grad_norm": 0.2638552486896515,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 36795
    },
    {
      "epoch": 32.59521700620018,
      "grad_norm": 0.2250799536705017,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 36800
    },
    {
      "epoch": 32.599645704162974,
      "grad_norm": 0.2779911160469055,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 36805
    },
    {
      "epoch": 32.604074402125775,
      "grad_norm": 0.2575404644012451,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 36810
    },
    {
      "epoch": 32.608503100088576,
      "grad_norm": 0.23828734457492828,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 36815
    },
    {
      "epoch": 32.61293179805137,
      "grad_norm": 0.22199828922748566,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 36820
    },
    {
      "epoch": 32.61736049601417,
      "grad_norm": 0.2699551582336426,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 36825
    },
    {
      "epoch": 32.62178919397697,
      "grad_norm": 0.24250787496566772,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 36830
    },
    {
      "epoch": 32.62621789193977,
      "grad_norm": 0.22502340376377106,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 36835
    },
    {
      "epoch": 32.63064658990257,
      "grad_norm": 0.22491468489170074,
      "learning_rate": 1e-05,
      "loss": 1.0339,
      "step": 36840
    },
    {
      "epoch": 32.63507528786537,
      "grad_norm": 0.21636025607585907,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 36845
    },
    {
      "epoch": 32.63950398582816,
      "grad_norm": 0.20897270739078522,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 36850
    },
    {
      "epoch": 32.643932683790965,
      "grad_norm": 0.28615567088127136,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 36855
    },
    {
      "epoch": 32.648361381753766,
      "grad_norm": 0.224213644862175,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 36860
    },
    {
      "epoch": 32.65279007971656,
      "grad_norm": 0.2577323019504547,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 36865
    },
    {
      "epoch": 32.65721877767936,
      "grad_norm": 0.2592145800590515,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 36870
    },
    {
      "epoch": 32.66164747564216,
      "grad_norm": 0.19868357479572296,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 36875
    },
    {
      "epoch": 32.666076173604964,
      "grad_norm": 0.29028773307800293,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 36880
    },
    {
      "epoch": 32.67050487156776,
      "grad_norm": 0.29770392179489136,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 36885
    },
    {
      "epoch": 32.67493356953056,
      "grad_norm": 0.23048870265483856,
      "learning_rate": 1e-05,
      "loss": 0.9109,
      "step": 36890
    },
    {
      "epoch": 32.67936226749336,
      "grad_norm": 0.23651660978794098,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 36895
    },
    {
      "epoch": 32.683790965456154,
      "grad_norm": 0.21591228246688843,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 36900
    },
    {
      "epoch": 32.688219663418955,
      "grad_norm": 0.25752878189086914,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 36905
    },
    {
      "epoch": 32.69264836138176,
      "grad_norm": 0.2440830022096634,
      "learning_rate": 1e-05,
      "loss": 0.936,
      "step": 36910
    },
    {
      "epoch": 32.69707705934455,
      "grad_norm": 0.28019312024116516,
      "learning_rate": 1e-05,
      "loss": 0.9347,
      "step": 36915
    },
    {
      "epoch": 32.70150575730735,
      "grad_norm": 0.30171093344688416,
      "learning_rate": 1e-05,
      "loss": 1.0466,
      "step": 36920
    },
    {
      "epoch": 32.70593445527015,
      "grad_norm": 0.3084123432636261,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 36925
    },
    {
      "epoch": 32.71036315323295,
      "grad_norm": 0.2576330900192261,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 36930
    },
    {
      "epoch": 32.71479185119575,
      "grad_norm": 0.22833141684532166,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 36935
    },
    {
      "epoch": 32.71922054915855,
      "grad_norm": 0.22222119569778442,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 36940
    },
    {
      "epoch": 32.723649247121344,
      "grad_norm": 0.24523021280765533,
      "learning_rate": 1e-05,
      "loss": 1.0316,
      "step": 36945
    },
    {
      "epoch": 32.728077945084145,
      "grad_norm": 0.2447596937417984,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 36950
    },
    {
      "epoch": 32.732506643046946,
      "grad_norm": 0.28538066148757935,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 36955
    },
    {
      "epoch": 32.73693534100974,
      "grad_norm": 0.23064063489437103,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 36960
    },
    {
      "epoch": 32.74136403897254,
      "grad_norm": 0.24713575839996338,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 36965
    },
    {
      "epoch": 32.74579273693534,
      "grad_norm": 0.27793022990226746,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 36970
    },
    {
      "epoch": 32.75022143489814,
      "grad_norm": 0.28085562586784363,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 36975
    },
    {
      "epoch": 32.75465013286094,
      "grad_norm": 0.19376224279403687,
      "learning_rate": 1e-05,
      "loss": 0.9318,
      "step": 36980
    },
    {
      "epoch": 32.75907883082374,
      "grad_norm": 0.25680866837501526,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 36985
    },
    {
      "epoch": 32.76350752878653,
      "grad_norm": 0.2485509216785431,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 36990
    },
    {
      "epoch": 32.767936226749335,
      "grad_norm": 0.23558840155601501,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 36995
    },
    {
      "epoch": 32.772364924712136,
      "grad_norm": 0.2355247437953949,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 37000
    },
    {
      "epoch": 32.77679362267494,
      "grad_norm": 0.24214035272598267,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 37005
    },
    {
      "epoch": 32.78122232063773,
      "grad_norm": 0.23325000703334808,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 37010
    },
    {
      "epoch": 32.78565101860053,
      "grad_norm": 0.2599441111087799,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 37015
    },
    {
      "epoch": 32.79007971656333,
      "grad_norm": 0.26771649718284607,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 37020
    },
    {
      "epoch": 32.79450841452613,
      "grad_norm": 0.24848684668540955,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 37025
    },
    {
      "epoch": 32.79893711248893,
      "grad_norm": 0.2554144561290741,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 37030
    },
    {
      "epoch": 32.80336581045173,
      "grad_norm": 0.3027076721191406,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 37035
    },
    {
      "epoch": 32.807794508414524,
      "grad_norm": 0.24313239753246307,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 37040
    },
    {
      "epoch": 32.812223206377325,
      "grad_norm": 0.3010101914405823,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 37045
    },
    {
      "epoch": 32.81665190434013,
      "grad_norm": 0.24170802533626556,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 37050
    },
    {
      "epoch": 32.82108060230292,
      "grad_norm": 0.2526654899120331,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 37055
    },
    {
      "epoch": 32.82550930026572,
      "grad_norm": 0.2299080342054367,
      "learning_rate": 1e-05,
      "loss": 1.0322,
      "step": 37060
    },
    {
      "epoch": 32.82993799822852,
      "grad_norm": 0.35731419920921326,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 37065
    },
    {
      "epoch": 32.83436669619132,
      "grad_norm": 0.2991877794265747,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 37070
    },
    {
      "epoch": 32.83879539415412,
      "grad_norm": 0.2344764918088913,
      "learning_rate": 1e-05,
      "loss": 1.0598,
      "step": 37075
    },
    {
      "epoch": 32.84322409211692,
      "grad_norm": 0.2575565576553345,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 37080
    },
    {
      "epoch": 32.847652790079714,
      "grad_norm": 0.26056885719299316,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 37085
    },
    {
      "epoch": 32.852081488042515,
      "grad_norm": 0.26875463128089905,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 37090
    },
    {
      "epoch": 32.856510186005316,
      "grad_norm": 0.2692602872848511,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 37095
    },
    {
      "epoch": 32.86093888396811,
      "grad_norm": 0.27811992168426514,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 37100
    },
    {
      "epoch": 32.86536758193091,
      "grad_norm": 0.22851362824440002,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 37105
    },
    {
      "epoch": 32.86979627989371,
      "grad_norm": 0.2500940263271332,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 37110
    },
    {
      "epoch": 32.87422497785651,
      "grad_norm": 0.22791844606399536,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 37115
    },
    {
      "epoch": 32.87865367581931,
      "grad_norm": 0.22995440661907196,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 37120
    },
    {
      "epoch": 32.88308237378211,
      "grad_norm": 0.27069565653800964,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 37125
    },
    {
      "epoch": 32.88751107174491,
      "grad_norm": 0.24803012609481812,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 37130
    },
    {
      "epoch": 32.891939769707704,
      "grad_norm": 0.2297472208738327,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 37135
    },
    {
      "epoch": 32.896368467670506,
      "grad_norm": 0.2674930691719055,
      "learning_rate": 1e-05,
      "loss": 1.0097,
      "step": 37140
    },
    {
      "epoch": 32.90079716563331,
      "grad_norm": 0.25241905450820923,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 37145
    },
    {
      "epoch": 32.9052258635961,
      "grad_norm": 0.24673676490783691,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 37150
    },
    {
      "epoch": 32.9096545615589,
      "grad_norm": 0.2551254630088806,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 37155
    },
    {
      "epoch": 32.9140832595217,
      "grad_norm": 0.23645739257335663,
      "learning_rate": 1e-05,
      "loss": 1.0155,
      "step": 37160
    },
    {
      "epoch": 32.9185119574845,
      "grad_norm": 0.3785167634487152,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 37165
    },
    {
      "epoch": 32.9229406554473,
      "grad_norm": 0.2699519693851471,
      "learning_rate": 1e-05,
      "loss": 1.06,
      "step": 37170
    },
    {
      "epoch": 32.9273693534101,
      "grad_norm": 0.2335544228553772,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 37175
    },
    {
      "epoch": 32.931798051372894,
      "grad_norm": 0.24266362190246582,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 37180
    },
    {
      "epoch": 32.936226749335695,
      "grad_norm": 0.22529281675815582,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 37185
    },
    {
      "epoch": 32.9406554472985,
      "grad_norm": 0.2407943457365036,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 37190
    },
    {
      "epoch": 32.94508414526129,
      "grad_norm": 0.2171785831451416,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 37195
    },
    {
      "epoch": 32.94951284322409,
      "grad_norm": 0.2314230352640152,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 37200
    },
    {
      "epoch": 32.95394154118689,
      "grad_norm": 0.3619070053100586,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 37205
    },
    {
      "epoch": 32.95837023914969,
      "grad_norm": 0.21950994431972504,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 37210
    },
    {
      "epoch": 32.96279893711249,
      "grad_norm": 0.20312044024467468,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 37215
    },
    {
      "epoch": 32.96722763507529,
      "grad_norm": 0.2392464131116867,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 37220
    },
    {
      "epoch": 32.971656333038084,
      "grad_norm": 0.2703521251678467,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 37225
    },
    {
      "epoch": 32.976085031000885,
      "grad_norm": 0.27658864855766296,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 37230
    },
    {
      "epoch": 32.980513728963686,
      "grad_norm": 0.2587783634662628,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 37235
    },
    {
      "epoch": 32.98494242692648,
      "grad_norm": 0.2905009984970093,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 37240
    },
    {
      "epoch": 32.98937112488928,
      "grad_norm": 0.2012159526348114,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 37245
    },
    {
      "epoch": 32.99379982285208,
      "grad_norm": 0.2734716236591339,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 37250
    },
    {
      "epoch": 32.998228520814884,
      "grad_norm": 0.22511662542819977,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 37255
    },
    {
      "epoch": 33.00265721877768,
      "grad_norm": 0.2462189793586731,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 37260
    },
    {
      "epoch": 33.00708591674048,
      "grad_norm": 0.22202102839946747,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 37265
    },
    {
      "epoch": 33.01151461470328,
      "grad_norm": 0.26523357629776,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 37270
    },
    {
      "epoch": 33.015943312666074,
      "grad_norm": 0.267270565032959,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 37275
    },
    {
      "epoch": 33.020372010628876,
      "grad_norm": 0.23890122771263123,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 37280
    },
    {
      "epoch": 33.02480070859168,
      "grad_norm": 0.29531559348106384,
      "learning_rate": 1e-05,
      "loss": 1.0377,
      "step": 37285
    },
    {
      "epoch": 33.02922940655447,
      "grad_norm": 0.2053052932024002,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 37290
    },
    {
      "epoch": 33.03365810451727,
      "grad_norm": 0.24436959624290466,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 37295
    },
    {
      "epoch": 33.03808680248007,
      "grad_norm": 0.33839404582977295,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 37300
    },
    {
      "epoch": 33.04251550044287,
      "grad_norm": 0.2618570923805237,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 37305
    },
    {
      "epoch": 33.04694419840567,
      "grad_norm": 0.24721820652484894,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 37310
    },
    {
      "epoch": 33.05137289636847,
      "grad_norm": 0.23903165757656097,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 37315
    },
    {
      "epoch": 33.055801594331264,
      "grad_norm": 0.2669444978237152,
      "learning_rate": 1e-05,
      "loss": 1.0607,
      "step": 37320
    },
    {
      "epoch": 33.060230292294065,
      "grad_norm": 0.274209588766098,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 37325
    },
    {
      "epoch": 33.064658990256866,
      "grad_norm": 0.19675327837467194,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 37330
    },
    {
      "epoch": 33.06908768821966,
      "grad_norm": 0.26304563879966736,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 37335
    },
    {
      "epoch": 33.07351638618246,
      "grad_norm": 0.25058713555336,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 37340
    },
    {
      "epoch": 33.07794508414526,
      "grad_norm": 0.27034229040145874,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 37345
    },
    {
      "epoch": 33.08237378210806,
      "grad_norm": 0.21662580966949463,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 37350
    },
    {
      "epoch": 33.08680248007086,
      "grad_norm": 0.2343723028898239,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 37355
    },
    {
      "epoch": 33.09123117803366,
      "grad_norm": 0.2491210550069809,
      "learning_rate": 1e-05,
      "loss": 1.0268,
      "step": 37360
    },
    {
      "epoch": 33.09565987599645,
      "grad_norm": 0.22822631895542145,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 37365
    },
    {
      "epoch": 33.100088573959255,
      "grad_norm": 0.2211805135011673,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 37370
    },
    {
      "epoch": 33.104517271922056,
      "grad_norm": 0.2651304602622986,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 37375
    },
    {
      "epoch": 33.10894596988486,
      "grad_norm": 0.28861936926841736,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 37380
    },
    {
      "epoch": 33.11337466784765,
      "grad_norm": 0.22887682914733887,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 37385
    },
    {
      "epoch": 33.11780336581045,
      "grad_norm": 0.383199006319046,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 37390
    },
    {
      "epoch": 33.122232063773254,
      "grad_norm": 0.23534031212329865,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 37395
    },
    {
      "epoch": 33.12666076173605,
      "grad_norm": 0.27983418107032776,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 37400
    },
    {
      "epoch": 33.13108945969885,
      "grad_norm": 0.24169385433197021,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 37405
    },
    {
      "epoch": 33.13551815766165,
      "grad_norm": 0.2552761435508728,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 37410
    },
    {
      "epoch": 33.139946855624444,
      "grad_norm": 0.24624589085578918,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 37415
    },
    {
      "epoch": 33.144375553587246,
      "grad_norm": 0.2403360903263092,
      "learning_rate": 1e-05,
      "loss": 1.0364,
      "step": 37420
    },
    {
      "epoch": 33.14880425155005,
      "grad_norm": 0.2598419487476349,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 37425
    },
    {
      "epoch": 33.15323294951284,
      "grad_norm": 0.28563302755355835,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 37430
    },
    {
      "epoch": 33.15766164747564,
      "grad_norm": 0.24167422950267792,
      "learning_rate": 1e-05,
      "loss": 0.9206,
      "step": 37435
    },
    {
      "epoch": 33.16209034543844,
      "grad_norm": 0.24953114986419678,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 37440
    },
    {
      "epoch": 33.16651904340124,
      "grad_norm": 0.22629134356975555,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 37445
    },
    {
      "epoch": 33.17094774136404,
      "grad_norm": 0.2352973371744156,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 37450
    },
    {
      "epoch": 33.17537643932684,
      "grad_norm": 0.219156414270401,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 37455
    },
    {
      "epoch": 33.179805137289634,
      "grad_norm": 0.25489112734794617,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 37460
    },
    {
      "epoch": 33.184233835252435,
      "grad_norm": 0.25097987055778503,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 37465
    },
    {
      "epoch": 33.188662533215236,
      "grad_norm": 0.23594914376735687,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 37470
    },
    {
      "epoch": 33.19309123117803,
      "grad_norm": 0.23238977789878845,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 37475
    },
    {
      "epoch": 33.19751992914083,
      "grad_norm": 0.2475544810295105,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 37480
    },
    {
      "epoch": 33.20194862710363,
      "grad_norm": 0.25583094358444214,
      "learning_rate": 1e-05,
      "loss": 1.0343,
      "step": 37485
    },
    {
      "epoch": 33.20637732506643,
      "grad_norm": 0.2261056900024414,
      "learning_rate": 1e-05,
      "loss": 1.0421,
      "step": 37490
    },
    {
      "epoch": 33.21080602302923,
      "grad_norm": 0.25020086765289307,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 37495
    },
    {
      "epoch": 33.21523472099203,
      "grad_norm": 0.2253435105085373,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 37500
    },
    {
      "epoch": 33.21966341895483,
      "grad_norm": 0.22151602804660797,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 37505
    },
    {
      "epoch": 33.224092116917625,
      "grad_norm": 0.2499372810125351,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 37510
    },
    {
      "epoch": 33.228520814880426,
      "grad_norm": 0.2618054747581482,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 37515
    },
    {
      "epoch": 33.23294951284323,
      "grad_norm": 0.22210164368152618,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 37520
    },
    {
      "epoch": 33.23737821080602,
      "grad_norm": 0.20142731070518494,
      "learning_rate": 1e-05,
      "loss": 0.9176,
      "step": 37525
    },
    {
      "epoch": 33.24180690876882,
      "grad_norm": 0.26337745785713196,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 37530
    },
    {
      "epoch": 33.246235606731624,
      "grad_norm": 0.2420770823955536,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 37535
    },
    {
      "epoch": 33.25066430469442,
      "grad_norm": 0.28763192892074585,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 37540
    },
    {
      "epoch": 33.25509300265722,
      "grad_norm": 0.2696307301521301,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 37545
    },
    {
      "epoch": 33.25952170062002,
      "grad_norm": 0.25449714064598083,
      "learning_rate": 1e-05,
      "loss": 1.018,
      "step": 37550
    },
    {
      "epoch": 33.263950398582814,
      "grad_norm": 0.2824353575706482,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 37555
    },
    {
      "epoch": 33.268379096545615,
      "grad_norm": 0.25014811754226685,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 37560
    },
    {
      "epoch": 33.27280779450842,
      "grad_norm": 0.2584537863731384,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 37565
    },
    {
      "epoch": 33.27723649247121,
      "grad_norm": 0.24283677339553833,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 37570
    },
    {
      "epoch": 33.28166519043401,
      "grad_norm": 0.23383092880249023,
      "learning_rate": 1e-05,
      "loss": 0.9153,
      "step": 37575
    },
    {
      "epoch": 33.28609388839681,
      "grad_norm": 0.24581915140151978,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 37580
    },
    {
      "epoch": 33.29052258635961,
      "grad_norm": 0.2930060029029846,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 37585
    },
    {
      "epoch": 33.29495128432241,
      "grad_norm": 0.20185275375843048,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 37590
    },
    {
      "epoch": 33.29937998228521,
      "grad_norm": 0.29508915543556213,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 37595
    },
    {
      "epoch": 33.303808680248004,
      "grad_norm": 0.2665797770023346,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 37600
    },
    {
      "epoch": 33.308237378210805,
      "grad_norm": 0.219760924577713,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 37605
    },
    {
      "epoch": 33.312666076173606,
      "grad_norm": 0.243989497423172,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 37610
    },
    {
      "epoch": 33.31709477413641,
      "grad_norm": 0.18913330137729645,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 37615
    },
    {
      "epoch": 33.3215234720992,
      "grad_norm": 0.23927466571331024,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 37620
    },
    {
      "epoch": 33.325952170062,
      "grad_norm": 0.2826748490333557,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 37625
    },
    {
      "epoch": 33.330380868024804,
      "grad_norm": 0.2496422678232193,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 37630
    },
    {
      "epoch": 33.3348095659876,
      "grad_norm": 0.2906489074230194,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 37635
    },
    {
      "epoch": 33.3392382639504,
      "grad_norm": 0.23769623041152954,
      "learning_rate": 1e-05,
      "loss": 1.058,
      "step": 37640
    },
    {
      "epoch": 33.3436669619132,
      "grad_norm": 0.2164970487356186,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 37645
    },
    {
      "epoch": 33.348095659875995,
      "grad_norm": 0.2195953130722046,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 37650
    },
    {
      "epoch": 33.352524357838796,
      "grad_norm": 0.2180243581533432,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 37655
    },
    {
      "epoch": 33.3569530558016,
      "grad_norm": 0.2616342008113861,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 37660
    },
    {
      "epoch": 33.36138175376439,
      "grad_norm": 0.26874032616615295,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 37665
    },
    {
      "epoch": 33.36581045172719,
      "grad_norm": 0.27089524269104004,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 37670
    },
    {
      "epoch": 33.37023914968999,
      "grad_norm": 0.23958326876163483,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 37675
    },
    {
      "epoch": 33.37466784765279,
      "grad_norm": 0.2638654112815857,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 37680
    },
    {
      "epoch": 33.37909654561559,
      "grad_norm": 0.2748025059700012,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 37685
    },
    {
      "epoch": 33.38352524357839,
      "grad_norm": 0.20542241632938385,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 37690
    },
    {
      "epoch": 33.387953941541184,
      "grad_norm": 0.253123015165329,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 37695
    },
    {
      "epoch": 33.392382639503985,
      "grad_norm": 0.22315892577171326,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 37700
    },
    {
      "epoch": 33.39681133746679,
      "grad_norm": 0.2307894229888916,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 37705
    },
    {
      "epoch": 33.40124003542958,
      "grad_norm": 0.24775582551956177,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 37710
    },
    {
      "epoch": 33.40566873339238,
      "grad_norm": 0.23357945680618286,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 37715
    },
    {
      "epoch": 33.41009743135518,
      "grad_norm": 0.22723504900932312,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 37720
    },
    {
      "epoch": 33.41452612931798,
      "grad_norm": 0.212600976228714,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 37725
    },
    {
      "epoch": 33.41895482728078,
      "grad_norm": 0.22390806674957275,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 37730
    },
    {
      "epoch": 33.42338352524358,
      "grad_norm": 0.25025299191474915,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 37735
    },
    {
      "epoch": 33.42781222320638,
      "grad_norm": 0.25924888253211975,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 37740
    },
    {
      "epoch": 33.432240921169175,
      "grad_norm": 0.23782186210155487,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 37745
    },
    {
      "epoch": 33.436669619131976,
      "grad_norm": 0.2332630753517151,
      "learning_rate": 1e-05,
      "loss": 0.912,
      "step": 37750
    },
    {
      "epoch": 33.44109831709478,
      "grad_norm": 0.22997522354125977,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 37755
    },
    {
      "epoch": 33.44552701505757,
      "grad_norm": 0.2595265209674835,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 37760
    },
    {
      "epoch": 33.44995571302037,
      "grad_norm": 0.23804420232772827,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 37765
    },
    {
      "epoch": 33.454384410983174,
      "grad_norm": 0.2505871653556824,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 37770
    },
    {
      "epoch": 33.45881310894597,
      "grad_norm": 0.2468378245830536,
      "learning_rate": 1e-05,
      "loss": 1.0161,
      "step": 37775
    },
    {
      "epoch": 33.46324180690877,
      "grad_norm": 0.24924004077911377,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 37780
    },
    {
      "epoch": 33.46767050487157,
      "grad_norm": 0.25373706221580505,
      "learning_rate": 1e-05,
      "loss": 1.0195,
      "step": 37785
    },
    {
      "epoch": 33.472099202834364,
      "grad_norm": 0.1956234872341156,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 37790
    },
    {
      "epoch": 33.476527900797166,
      "grad_norm": 0.2079487442970276,
      "learning_rate": 1e-05,
      "loss": 0.9211,
      "step": 37795
    },
    {
      "epoch": 33.48095659875997,
      "grad_norm": 0.2684883773326874,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 37800
    },
    {
      "epoch": 33.48538529672276,
      "grad_norm": 0.23346030712127686,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 37805
    },
    {
      "epoch": 33.48981399468556,
      "grad_norm": 0.3997478783130646,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 37810
    },
    {
      "epoch": 33.49424269264836,
      "grad_norm": 0.24162228405475616,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 37815
    },
    {
      "epoch": 33.49867139061116,
      "grad_norm": 0.21404288709163666,
      "learning_rate": 1e-05,
      "loss": 0.9267,
      "step": 37820
    },
    {
      "epoch": 33.50310008857396,
      "grad_norm": 0.27143868803977966,
      "learning_rate": 1e-05,
      "loss": 1.062,
      "step": 37825
    },
    {
      "epoch": 33.50752878653676,
      "grad_norm": 0.19717994332313538,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 37830
    },
    {
      "epoch": 33.511957484499554,
      "grad_norm": 0.20435220003128052,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 37835
    },
    {
      "epoch": 33.516386182462355,
      "grad_norm": 0.24851441383361816,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 37840
    },
    {
      "epoch": 33.520814880425156,
      "grad_norm": 0.23568382859230042,
      "learning_rate": 1e-05,
      "loss": 1.0269,
      "step": 37845
    },
    {
      "epoch": 33.52524357838795,
      "grad_norm": 0.25395458936691284,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 37850
    },
    {
      "epoch": 33.52967227635075,
      "grad_norm": 0.2722824215888977,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 37855
    },
    {
      "epoch": 33.53410097431355,
      "grad_norm": 0.26487404108047485,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 37860
    },
    {
      "epoch": 33.538529672276354,
      "grad_norm": 0.23948870599269867,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 37865
    },
    {
      "epoch": 33.54295837023915,
      "grad_norm": 0.22190731763839722,
      "learning_rate": 1e-05,
      "loss": 1.0318,
      "step": 37870
    },
    {
      "epoch": 33.54738706820195,
      "grad_norm": 0.26469072699546814,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 37875
    },
    {
      "epoch": 33.55181576616475,
      "grad_norm": 0.22756771743297577,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 37880
    },
    {
      "epoch": 33.556244464127545,
      "grad_norm": 0.23696641623973846,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 37885
    },
    {
      "epoch": 33.560673162090346,
      "grad_norm": 0.25579628348350525,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 37890
    },
    {
      "epoch": 33.56510186005315,
      "grad_norm": 0.2269541174173355,
      "learning_rate": 1e-05,
      "loss": 1.0376,
      "step": 37895
    },
    {
      "epoch": 33.56953055801594,
      "grad_norm": 0.2187008112668991,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 37900
    },
    {
      "epoch": 33.57395925597874,
      "grad_norm": 0.2544189691543579,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 37905
    },
    {
      "epoch": 33.578387953941544,
      "grad_norm": 0.21207940578460693,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 37910
    },
    {
      "epoch": 33.58281665190434,
      "grad_norm": 0.24516642093658447,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 37915
    },
    {
      "epoch": 33.58724534986714,
      "grad_norm": 0.2136494368314743,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 37920
    },
    {
      "epoch": 33.59167404782994,
      "grad_norm": 0.25357040762901306,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 37925
    },
    {
      "epoch": 33.596102745792734,
      "grad_norm": 0.24380625784397125,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 37930
    },
    {
      "epoch": 33.600531443755536,
      "grad_norm": 0.2423871010541916,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 37935
    },
    {
      "epoch": 33.60496014171834,
      "grad_norm": 0.3245455026626587,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 37940
    },
    {
      "epoch": 33.60938883968113,
      "grad_norm": 0.2679072320461273,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 37945
    },
    {
      "epoch": 33.61381753764393,
      "grad_norm": 0.24811644852161407,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 37950
    },
    {
      "epoch": 33.61824623560673,
      "grad_norm": 0.27384641766548157,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 37955
    },
    {
      "epoch": 33.62267493356953,
      "grad_norm": 0.21501348912715912,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 37960
    },
    {
      "epoch": 33.62710363153233,
      "grad_norm": 0.23064468801021576,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 37965
    },
    {
      "epoch": 33.63153232949513,
      "grad_norm": 0.22785690426826477,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 37970
    },
    {
      "epoch": 33.635961027457924,
      "grad_norm": 0.24837546050548553,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 37975
    },
    {
      "epoch": 33.640389725420725,
      "grad_norm": 0.25972801446914673,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 37980
    },
    {
      "epoch": 33.644818423383526,
      "grad_norm": 0.2836465537548065,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 37985
    },
    {
      "epoch": 33.64924712134633,
      "grad_norm": 0.2290176898241043,
      "learning_rate": 1e-05,
      "loss": 1.0113,
      "step": 37990
    },
    {
      "epoch": 33.65367581930912,
      "grad_norm": 0.3135979473590851,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 37995
    },
    {
      "epoch": 33.65810451727192,
      "grad_norm": 0.2480962723493576,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 38000
    },
    {
      "epoch": 33.662533215234724,
      "grad_norm": 0.21964405477046967,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 38005
    },
    {
      "epoch": 33.66696191319752,
      "grad_norm": 0.2652662396430969,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 38010
    },
    {
      "epoch": 33.67139061116032,
      "grad_norm": 0.26028746366500854,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 38015
    },
    {
      "epoch": 33.67581930912312,
      "grad_norm": 0.24155287444591522,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 38020
    },
    {
      "epoch": 33.680248007085915,
      "grad_norm": 0.2609650492668152,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 38025
    },
    {
      "epoch": 33.684676705048716,
      "grad_norm": 0.26211851835250854,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 38030
    },
    {
      "epoch": 33.68910540301152,
      "grad_norm": 0.21739651262760162,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 38035
    },
    {
      "epoch": 33.69353410097431,
      "grad_norm": 0.22474536299705505,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 38040
    },
    {
      "epoch": 33.69796279893711,
      "grad_norm": 0.22200100123882294,
      "learning_rate": 1e-05,
      "loss": 1.0222,
      "step": 38045
    },
    {
      "epoch": 33.702391496899914,
      "grad_norm": 0.23596346378326416,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 38050
    },
    {
      "epoch": 33.70682019486271,
      "grad_norm": 0.2362101525068283,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 38055
    },
    {
      "epoch": 33.71124889282551,
      "grad_norm": 0.2920089066028595,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 38060
    },
    {
      "epoch": 33.71567759078831,
      "grad_norm": 0.23172269761562347,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 38065
    },
    {
      "epoch": 33.720106288751104,
      "grad_norm": 0.2064635157585144,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 38070
    },
    {
      "epoch": 33.724534986713905,
      "grad_norm": 0.2808133661746979,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 38075
    },
    {
      "epoch": 33.72896368467671,
      "grad_norm": 0.27570268511772156,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 38080
    },
    {
      "epoch": 33.7333923826395,
      "grad_norm": 0.2796964943408966,
      "learning_rate": 1e-05,
      "loss": 1.0364,
      "step": 38085
    },
    {
      "epoch": 33.7378210806023,
      "grad_norm": 0.23116172850131989,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 38090
    },
    {
      "epoch": 33.7422497785651,
      "grad_norm": 0.3091551661491394,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 38095
    },
    {
      "epoch": 33.7466784765279,
      "grad_norm": 0.26836469769477844,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 38100
    },
    {
      "epoch": 33.7511071744907,
      "grad_norm": 0.2579007148742676,
      "learning_rate": 1e-05,
      "loss": 0.9338,
      "step": 38105
    },
    {
      "epoch": 33.7555358724535,
      "grad_norm": 0.21425236761569977,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 38110
    },
    {
      "epoch": 33.7599645704163,
      "grad_norm": 0.2666289210319519,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 38115
    },
    {
      "epoch": 33.764393268379095,
      "grad_norm": 0.2515363395214081,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 38120
    },
    {
      "epoch": 33.768821966341896,
      "grad_norm": 0.2566606104373932,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 38125
    },
    {
      "epoch": 33.7732506643047,
      "grad_norm": 0.29022499918937683,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 38130
    },
    {
      "epoch": 33.77767936226749,
      "grad_norm": 0.2834495007991791,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 38135
    },
    {
      "epoch": 33.78210806023029,
      "grad_norm": 0.24446694552898407,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 38140
    },
    {
      "epoch": 33.786536758193094,
      "grad_norm": 0.27261218428611755,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 38145
    },
    {
      "epoch": 33.79096545615589,
      "grad_norm": 0.23751747608184814,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 38150
    },
    {
      "epoch": 33.79539415411869,
      "grad_norm": 0.20280642807483673,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 38155
    },
    {
      "epoch": 33.79982285208149,
      "grad_norm": 0.25098899006843567,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 38160
    },
    {
      "epoch": 33.804251550044285,
      "grad_norm": 0.23242275416851044,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 38165
    },
    {
      "epoch": 33.808680248007086,
      "grad_norm": 0.24211683869361877,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 38170
    },
    {
      "epoch": 33.81310894596989,
      "grad_norm": 0.2172280102968216,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 38175
    },
    {
      "epoch": 33.81753764393268,
      "grad_norm": 0.2778877019882202,
      "learning_rate": 1e-05,
      "loss": 1.0226,
      "step": 38180
    },
    {
      "epoch": 33.82196634189548,
      "grad_norm": 0.2905314862728119,
      "learning_rate": 1e-05,
      "loss": 1.0409,
      "step": 38185
    },
    {
      "epoch": 33.826395039858284,
      "grad_norm": 0.2853711247444153,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 38190
    },
    {
      "epoch": 33.83082373782108,
      "grad_norm": 0.2654324173927307,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 38195
    },
    {
      "epoch": 33.83525243578388,
      "grad_norm": 0.2581745386123657,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 38200
    },
    {
      "epoch": 33.83968113374668,
      "grad_norm": 0.25141727924346924,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 38205
    },
    {
      "epoch": 33.844109831709474,
      "grad_norm": 0.2173549085855484,
      "learning_rate": 1e-05,
      "loss": 0.9193,
      "step": 38210
    },
    {
      "epoch": 33.848538529672275,
      "grad_norm": 0.2668944001197815,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 38215
    },
    {
      "epoch": 33.85296722763508,
      "grad_norm": 0.21575461328029633,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 38220
    },
    {
      "epoch": 33.85739592559787,
      "grad_norm": 0.2107408344745636,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 38225
    },
    {
      "epoch": 33.86182462356067,
      "grad_norm": 0.21856476366519928,
      "learning_rate": 1e-05,
      "loss": 1.0577,
      "step": 38230
    },
    {
      "epoch": 33.86625332152347,
      "grad_norm": 0.225790336728096,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 38235
    },
    {
      "epoch": 33.870682019486274,
      "grad_norm": 0.21448101103305817,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 38240
    },
    {
      "epoch": 33.87511071744907,
      "grad_norm": 0.2115173041820526,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 38245
    },
    {
      "epoch": 33.87953941541187,
      "grad_norm": 0.25188231468200684,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 38250
    },
    {
      "epoch": 33.88396811337467,
      "grad_norm": 0.2773911952972412,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 38255
    },
    {
      "epoch": 33.888396811337465,
      "grad_norm": 0.1982952356338501,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 38260
    },
    {
      "epoch": 33.892825509300266,
      "grad_norm": 0.27818411588668823,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 38265
    },
    {
      "epoch": 33.89725420726307,
      "grad_norm": 0.26349830627441406,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 38270
    },
    {
      "epoch": 33.90168290522586,
      "grad_norm": 0.24820630252361298,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 38275
    },
    {
      "epoch": 33.90611160318866,
      "grad_norm": 0.20903825759887695,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 38280
    },
    {
      "epoch": 33.910540301151464,
      "grad_norm": 0.24729932844638824,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 38285
    },
    {
      "epoch": 33.91496899911426,
      "grad_norm": 0.21811094880104065,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 38290
    },
    {
      "epoch": 33.91939769707706,
      "grad_norm": 0.23259656131267548,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 38295
    },
    {
      "epoch": 33.92382639503986,
      "grad_norm": 0.257358193397522,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 38300
    },
    {
      "epoch": 33.928255093002655,
      "grad_norm": 0.1952032893896103,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 38305
    },
    {
      "epoch": 33.932683790965456,
      "grad_norm": 0.2545221149921417,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 38310
    },
    {
      "epoch": 33.93711248892826,
      "grad_norm": 0.27006372809410095,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 38315
    },
    {
      "epoch": 33.94154118689105,
      "grad_norm": 0.21550485491752625,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 38320
    },
    {
      "epoch": 33.94596988485385,
      "grad_norm": 0.225944384932518,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 38325
    },
    {
      "epoch": 33.95039858281665,
      "grad_norm": 0.2626259922981262,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 38330
    },
    {
      "epoch": 33.95482728077945,
      "grad_norm": 0.22060105204582214,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 38335
    },
    {
      "epoch": 33.95925597874225,
      "grad_norm": 0.23608282208442688,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 38340
    },
    {
      "epoch": 33.96368467670505,
      "grad_norm": 0.21818242967128754,
      "learning_rate": 1e-05,
      "loss": 1.0262,
      "step": 38345
    },
    {
      "epoch": 33.96811337466785,
      "grad_norm": 0.2523571848869324,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 38350
    },
    {
      "epoch": 33.972542072630645,
      "grad_norm": 0.24640929698944092,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 38355
    },
    {
      "epoch": 33.97697077059345,
      "grad_norm": 0.24513690173625946,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 38360
    },
    {
      "epoch": 33.98139946855625,
      "grad_norm": 0.23903600871562958,
      "learning_rate": 1e-05,
      "loss": 1.0311,
      "step": 38365
    },
    {
      "epoch": 33.98582816651904,
      "grad_norm": 0.2276596575975418,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 38370
    },
    {
      "epoch": 33.99025686448184,
      "grad_norm": 0.30598318576812744,
      "learning_rate": 1e-05,
      "loss": 1.0492,
      "step": 38375
    },
    {
      "epoch": 33.994685562444644,
      "grad_norm": 0.20703360438346863,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 38380
    },
    {
      "epoch": 33.99911426040744,
      "grad_norm": 0.2632818818092346,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 38385
    },
    {
      "epoch": 34.00354295837024,
      "grad_norm": 0.25833189487457275,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 38390
    },
    {
      "epoch": 34.00797165633304,
      "grad_norm": 0.2433941662311554,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 38395
    },
    {
      "epoch": 34.012400354295835,
      "grad_norm": 0.22201590240001678,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 38400
    },
    {
      "epoch": 34.016829052258636,
      "grad_norm": 0.25580641627311707,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 38405
    },
    {
      "epoch": 34.02125775022144,
      "grad_norm": 0.22945557534694672,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 38410
    },
    {
      "epoch": 34.02568644818423,
      "grad_norm": 0.2832637131214142,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 38415
    },
    {
      "epoch": 34.03011514614703,
      "grad_norm": 0.25894853472709656,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 38420
    },
    {
      "epoch": 34.034543844109834,
      "grad_norm": 0.2717205584049225,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 38425
    },
    {
      "epoch": 34.03897254207263,
      "grad_norm": 0.22346657514572144,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 38430
    },
    {
      "epoch": 34.04340124003543,
      "grad_norm": 0.25589868426322937,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 38435
    },
    {
      "epoch": 34.04782993799823,
      "grad_norm": 0.24962647259235382,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 38440
    },
    {
      "epoch": 34.052258635961024,
      "grad_norm": 0.2575848698616028,
      "learning_rate": 1e-05,
      "loss": 1.0672,
      "step": 38445
    },
    {
      "epoch": 34.056687333923826,
      "grad_norm": 0.24088113009929657,
      "learning_rate": 1e-05,
      "loss": 1.0532,
      "step": 38450
    },
    {
      "epoch": 34.06111603188663,
      "grad_norm": 0.21632416546344757,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 38455
    },
    {
      "epoch": 34.06554472984942,
      "grad_norm": 0.2102256864309311,
      "learning_rate": 1e-05,
      "loss": 1.0161,
      "step": 38460
    },
    {
      "epoch": 34.06997342781222,
      "grad_norm": 0.22546735405921936,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 38465
    },
    {
      "epoch": 34.07440212577502,
      "grad_norm": 0.24194324016571045,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 38470
    },
    {
      "epoch": 34.078830823737825,
      "grad_norm": 0.21429917216300964,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 38475
    },
    {
      "epoch": 34.08325952170062,
      "grad_norm": 0.21943578124046326,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 38480
    },
    {
      "epoch": 34.08768821966342,
      "grad_norm": 0.2599654197692871,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 38485
    },
    {
      "epoch": 34.09211691762622,
      "grad_norm": 0.27270013093948364,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 38490
    },
    {
      "epoch": 34.096545615589015,
      "grad_norm": 0.22257526218891144,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 38495
    },
    {
      "epoch": 34.100974313551816,
      "grad_norm": 0.24975185096263885,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 38500
    },
    {
      "epoch": 34.10540301151462,
      "grad_norm": 0.22256126999855042,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 38505
    },
    {
      "epoch": 34.10983170947741,
      "grad_norm": 0.2118120789527893,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 38510
    },
    {
      "epoch": 34.11426040744021,
      "grad_norm": 0.20445938408374786,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 38515
    },
    {
      "epoch": 34.118689105403014,
      "grad_norm": 0.2057626098394394,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 38520
    },
    {
      "epoch": 34.12311780336581,
      "grad_norm": 0.2255978137254715,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 38525
    },
    {
      "epoch": 34.12754650132861,
      "grad_norm": 0.24340316653251648,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 38530
    },
    {
      "epoch": 34.13197519929141,
      "grad_norm": 0.2793462872505188,
      "learning_rate": 1e-05,
      "loss": 1.0341,
      "step": 38535
    },
    {
      "epoch": 34.136403897254205,
      "grad_norm": 0.27937790751457214,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 38540
    },
    {
      "epoch": 34.140832595217006,
      "grad_norm": 0.2662443220615387,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 38545
    },
    {
      "epoch": 34.14526129317981,
      "grad_norm": 0.2826865315437317,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 38550
    },
    {
      "epoch": 34.1496899911426,
      "grad_norm": 0.2715863585472107,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 38555
    },
    {
      "epoch": 34.1541186891054,
      "grad_norm": 0.23822639882564545,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 38560
    },
    {
      "epoch": 34.158547387068204,
      "grad_norm": 0.30323848128318787,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 38565
    },
    {
      "epoch": 34.162976085031,
      "grad_norm": 0.2628025412559509,
      "learning_rate": 1e-05,
      "loss": 1.0421,
      "step": 38570
    },
    {
      "epoch": 34.1674047829938,
      "grad_norm": 0.21528016030788422,
      "learning_rate": 1e-05,
      "loss": 0.9277,
      "step": 38575
    },
    {
      "epoch": 34.1718334809566,
      "grad_norm": 0.2175915688276291,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 38580
    },
    {
      "epoch": 34.176262178919394,
      "grad_norm": 0.25207310914993286,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 38585
    },
    {
      "epoch": 34.180690876882196,
      "grad_norm": 0.2646348178386688,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 38590
    },
    {
      "epoch": 34.185119574845,
      "grad_norm": 0.22858981788158417,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 38595
    },
    {
      "epoch": 34.1895482728078,
      "grad_norm": 0.24562247097492218,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 38600
    },
    {
      "epoch": 34.19397697077059,
      "grad_norm": 0.24626575410366058,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 38605
    },
    {
      "epoch": 34.19840566873339,
      "grad_norm": 0.2909865379333496,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 38610
    },
    {
      "epoch": 34.202834366696194,
      "grad_norm": 0.2305697798728943,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 38615
    },
    {
      "epoch": 34.20726306465899,
      "grad_norm": 0.27132418751716614,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 38620
    },
    {
      "epoch": 34.21169176262179,
      "grad_norm": 0.25381582975387573,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 38625
    },
    {
      "epoch": 34.21612046058459,
      "grad_norm": 0.26009371876716614,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 38630
    },
    {
      "epoch": 34.220549158547385,
      "grad_norm": 0.23428818583488464,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 38635
    },
    {
      "epoch": 34.224977856510186,
      "grad_norm": 0.2927929759025574,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 38640
    },
    {
      "epoch": 34.22940655447299,
      "grad_norm": 0.24169239401817322,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 38645
    },
    {
      "epoch": 34.23383525243578,
      "grad_norm": 0.23842723667621613,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 38650
    },
    {
      "epoch": 34.23826395039858,
      "grad_norm": 0.19830359518527985,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 38655
    },
    {
      "epoch": 34.242692648361384,
      "grad_norm": 0.24296438694000244,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 38660
    },
    {
      "epoch": 34.24712134632418,
      "grad_norm": 0.2316388040781021,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 38665
    },
    {
      "epoch": 34.25155004428698,
      "grad_norm": 0.22557847201824188,
      "learning_rate": 1e-05,
      "loss": 0.917,
      "step": 38670
    },
    {
      "epoch": 34.25597874224978,
      "grad_norm": 0.2527371048927307,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 38675
    },
    {
      "epoch": 34.260407440212575,
      "grad_norm": 0.26969581842422485,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 38680
    },
    {
      "epoch": 34.264836138175376,
      "grad_norm": 0.24290160834789276,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 38685
    },
    {
      "epoch": 34.26926483613818,
      "grad_norm": 0.25625211000442505,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 38690
    },
    {
      "epoch": 34.27369353410097,
      "grad_norm": 0.2537311315536499,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 38695
    },
    {
      "epoch": 34.27812223206377,
      "grad_norm": 0.2425544410943985,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 38700
    },
    {
      "epoch": 34.282550930026574,
      "grad_norm": 0.2746416926383972,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 38705
    },
    {
      "epoch": 34.28697962798937,
      "grad_norm": 0.2213379591703415,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 38710
    },
    {
      "epoch": 34.29140832595217,
      "grad_norm": 0.23981286585330963,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 38715
    },
    {
      "epoch": 34.29583702391497,
      "grad_norm": 0.2633059620857239,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 38720
    },
    {
      "epoch": 34.30026572187777,
      "grad_norm": 0.2280246466398239,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 38725
    },
    {
      "epoch": 34.304694419840565,
      "grad_norm": 0.22413896024227142,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 38730
    },
    {
      "epoch": 34.30912311780337,
      "grad_norm": 0.2260398417711258,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 38735
    },
    {
      "epoch": 34.31355181576617,
      "grad_norm": 0.2194809466600418,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 38740
    },
    {
      "epoch": 34.31798051372896,
      "grad_norm": 0.26567405462265015,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 38745
    },
    {
      "epoch": 34.32240921169176,
      "grad_norm": 0.2985020875930786,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 38750
    },
    {
      "epoch": 34.326837909654564,
      "grad_norm": 0.27523016929626465,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 38755
    },
    {
      "epoch": 34.33126660761736,
      "grad_norm": 0.257125586271286,
      "learning_rate": 1e-05,
      "loss": 1.0009,
      "step": 38760
    },
    {
      "epoch": 34.33569530558016,
      "grad_norm": 0.24871249496936798,
      "learning_rate": 1e-05,
      "loss": 1.0432,
      "step": 38765
    },
    {
      "epoch": 34.34012400354296,
      "grad_norm": 0.2568099796772003,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 38770
    },
    {
      "epoch": 34.344552701505755,
      "grad_norm": 0.3367270231246948,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 38775
    },
    {
      "epoch": 34.348981399468556,
      "grad_norm": 0.2605668008327484,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 38780
    },
    {
      "epoch": 34.35341009743136,
      "grad_norm": 0.24907691776752472,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 38785
    },
    {
      "epoch": 34.35783879539415,
      "grad_norm": 0.2581411302089691,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 38790
    },
    {
      "epoch": 34.36226749335695,
      "grad_norm": 0.31799376010894775,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 38795
    },
    {
      "epoch": 34.366696191319754,
      "grad_norm": 0.249354749917984,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 38800
    },
    {
      "epoch": 34.37112488928255,
      "grad_norm": 0.27642133831977844,
      "learning_rate": 1e-05,
      "loss": 0.9264,
      "step": 38805
    },
    {
      "epoch": 34.37555358724535,
      "grad_norm": 0.22749485075473785,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 38810
    },
    {
      "epoch": 34.37998228520815,
      "grad_norm": 0.23874270915985107,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 38815
    },
    {
      "epoch": 34.384410983170945,
      "grad_norm": 0.2383110225200653,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 38820
    },
    {
      "epoch": 34.388839681133746,
      "grad_norm": 0.2674820125102997,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 38825
    },
    {
      "epoch": 34.39326837909655,
      "grad_norm": 0.3682923913002014,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 38830
    },
    {
      "epoch": 34.39769707705934,
      "grad_norm": 0.31268978118896484,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 38835
    },
    {
      "epoch": 34.40212577502214,
      "grad_norm": 0.26730576157569885,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 38840
    },
    {
      "epoch": 34.40655447298494,
      "grad_norm": 0.234216570854187,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 38845
    },
    {
      "epoch": 34.410983170947745,
      "grad_norm": 0.2032884657382965,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 38850
    },
    {
      "epoch": 34.41541186891054,
      "grad_norm": 0.24037154018878937,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 38855
    },
    {
      "epoch": 34.41984056687334,
      "grad_norm": 0.2099924236536026,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 38860
    },
    {
      "epoch": 34.42426926483614,
      "grad_norm": 0.21875235438346863,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 38865
    },
    {
      "epoch": 34.428697962798935,
      "grad_norm": 0.2246522158384323,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 38870
    },
    {
      "epoch": 34.43312666076174,
      "grad_norm": 0.2580392062664032,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 38875
    },
    {
      "epoch": 34.43755535872454,
      "grad_norm": 0.21624650061130524,
      "learning_rate": 1e-05,
      "loss": 0.9263,
      "step": 38880
    },
    {
      "epoch": 34.44198405668733,
      "grad_norm": 0.19981111586093903,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 38885
    },
    {
      "epoch": 34.44641275465013,
      "grad_norm": 0.2141062617301941,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 38890
    },
    {
      "epoch": 34.450841452612934,
      "grad_norm": 0.21368104219436646,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 38895
    },
    {
      "epoch": 34.45527015057573,
      "grad_norm": 0.27000999450683594,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 38900
    },
    {
      "epoch": 34.45969884853853,
      "grad_norm": 0.2587941288948059,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 38905
    },
    {
      "epoch": 34.46412754650133,
      "grad_norm": 0.2249888777732849,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 38910
    },
    {
      "epoch": 34.468556244464125,
      "grad_norm": 0.26439368724823,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 38915
    },
    {
      "epoch": 34.472984942426926,
      "grad_norm": 0.23644816875457764,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 38920
    },
    {
      "epoch": 34.47741364038973,
      "grad_norm": 0.22862541675567627,
      "learning_rate": 1e-05,
      "loss": 0.921,
      "step": 38925
    },
    {
      "epoch": 34.48184233835252,
      "grad_norm": 0.22303332388401031,
      "learning_rate": 1e-05,
      "loss": 0.9209,
      "step": 38930
    },
    {
      "epoch": 34.48627103631532,
      "grad_norm": 0.23572546243667603,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 38935
    },
    {
      "epoch": 34.490699734278124,
      "grad_norm": 0.19901789724826813,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 38940
    },
    {
      "epoch": 34.49512843224092,
      "grad_norm": 0.25095057487487793,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 38945
    },
    {
      "epoch": 34.49955713020372,
      "grad_norm": 0.2433033287525177,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 38950
    },
    {
      "epoch": 34.50398582816652,
      "grad_norm": 0.2725948691368103,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 38955
    },
    {
      "epoch": 34.508414526129314,
      "grad_norm": 0.275831013917923,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 38960
    },
    {
      "epoch": 34.512843224092116,
      "grad_norm": 0.26582422852516174,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 38965
    },
    {
      "epoch": 34.51727192205492,
      "grad_norm": 0.22523094713687897,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 38970
    },
    {
      "epoch": 34.52170062001772,
      "grad_norm": 0.2237347811460495,
      "learning_rate": 1e-05,
      "loss": 1.0344,
      "step": 38975
    },
    {
      "epoch": 34.52612931798051,
      "grad_norm": 0.23026876151561737,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 38980
    },
    {
      "epoch": 34.53055801594331,
      "grad_norm": 0.2516075670719147,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 38985
    },
    {
      "epoch": 34.534986713906115,
      "grad_norm": 0.2620795667171478,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 38990
    },
    {
      "epoch": 34.53941541186891,
      "grad_norm": 0.2485264390707016,
      "learning_rate": 1e-05,
      "loss": 0.9364,
      "step": 38995
    },
    {
      "epoch": 34.54384410983171,
      "grad_norm": 0.26360583305358887,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 39000
    },
    {
      "epoch": 34.54827280779451,
      "grad_norm": 0.2605079412460327,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 39005
    },
    {
      "epoch": 34.552701505757305,
      "grad_norm": 0.20501109957695007,
      "learning_rate": 1e-05,
      "loss": 1.0172,
      "step": 39010
    },
    {
      "epoch": 34.55713020372011,
      "grad_norm": 0.23418430984020233,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 39015
    },
    {
      "epoch": 34.56155890168291,
      "grad_norm": 0.25498753786087036,
      "learning_rate": 1e-05,
      "loss": 0.9381,
      "step": 39020
    },
    {
      "epoch": 34.5659875996457,
      "grad_norm": 0.2587830722332001,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 39025
    },
    {
      "epoch": 34.5704162976085,
      "grad_norm": 0.2637541890144348,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 39030
    },
    {
      "epoch": 34.574844995571304,
      "grad_norm": 0.25061196088790894,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 39035
    },
    {
      "epoch": 34.5792736935341,
      "grad_norm": 0.26215267181396484,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 39040
    },
    {
      "epoch": 34.5837023914969,
      "grad_norm": 0.223074808716774,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 39045
    },
    {
      "epoch": 34.5881310894597,
      "grad_norm": 0.2833113372325897,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 39050
    },
    {
      "epoch": 34.592559787422495,
      "grad_norm": 0.2382892668247223,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 39055
    },
    {
      "epoch": 34.596988485385296,
      "grad_norm": 0.29259565472602844,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 39060
    },
    {
      "epoch": 34.6014171833481,
      "grad_norm": 0.3025352954864502,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 39065
    },
    {
      "epoch": 34.60584588131089,
      "grad_norm": 0.263430655002594,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 39070
    },
    {
      "epoch": 34.61027457927369,
      "grad_norm": 0.2834121584892273,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 39075
    },
    {
      "epoch": 34.614703277236494,
      "grad_norm": 0.21686379611492157,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 39080
    },
    {
      "epoch": 34.619131975199295,
      "grad_norm": 0.2334703654050827,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 39085
    },
    {
      "epoch": 34.62356067316209,
      "grad_norm": 0.23110033571720123,
      "learning_rate": 1e-05,
      "loss": 1.0545,
      "step": 39090
    },
    {
      "epoch": 34.62798937112489,
      "grad_norm": 0.2550921142101288,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 39095
    },
    {
      "epoch": 34.63241806908769,
      "grad_norm": 0.20780588686466217,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 39100
    },
    {
      "epoch": 34.636846767050486,
      "grad_norm": 0.2538759410381317,
      "learning_rate": 1e-05,
      "loss": 1.0107,
      "step": 39105
    },
    {
      "epoch": 34.64127546501329,
      "grad_norm": 0.2352839708328247,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 39110
    },
    {
      "epoch": 34.64570416297609,
      "grad_norm": 0.2324318140745163,
      "learning_rate": 1e-05,
      "loss": 1.0232,
      "step": 39115
    },
    {
      "epoch": 34.65013286093888,
      "grad_norm": 0.2471347451210022,
      "learning_rate": 1e-05,
      "loss": 0.9299,
      "step": 39120
    },
    {
      "epoch": 34.65456155890168,
      "grad_norm": 0.22780823707580566,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 39125
    },
    {
      "epoch": 34.658990256864485,
      "grad_norm": 0.2593483626842499,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 39130
    },
    {
      "epoch": 34.66341895482728,
      "grad_norm": 0.24731208384037018,
      "learning_rate": 1e-05,
      "loss": 1.0208,
      "step": 39135
    },
    {
      "epoch": 34.66784765279008,
      "grad_norm": 0.2653765380382538,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 39140
    },
    {
      "epoch": 34.67227635075288,
      "grad_norm": 0.22021587193012238,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 39145
    },
    {
      "epoch": 34.676705048715675,
      "grad_norm": 0.22607065737247467,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 39150
    },
    {
      "epoch": 34.681133746678476,
      "grad_norm": 0.28130224347114563,
      "learning_rate": 1e-05,
      "loss": 1.024,
      "step": 39155
    },
    {
      "epoch": 34.68556244464128,
      "grad_norm": 0.22432592511177063,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 39160
    },
    {
      "epoch": 34.68999114260407,
      "grad_norm": 0.268875390291214,
      "learning_rate": 1e-05,
      "loss": 1.0473,
      "step": 39165
    },
    {
      "epoch": 34.69441984056687,
      "grad_norm": 0.24163834750652313,
      "learning_rate": 1e-05,
      "loss": 0.9306,
      "step": 39170
    },
    {
      "epoch": 34.698848538529674,
      "grad_norm": 0.22692064940929413,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 39175
    },
    {
      "epoch": 34.70327723649247,
      "grad_norm": 0.25467365980148315,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 39180
    },
    {
      "epoch": 34.70770593445527,
      "grad_norm": 0.25165635347366333,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 39185
    },
    {
      "epoch": 34.71213463241807,
      "grad_norm": 0.2267739325761795,
      "learning_rate": 1e-05,
      "loss": 0.9296,
      "step": 39190
    },
    {
      "epoch": 34.716563330380865,
      "grad_norm": 0.2208961546421051,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 39195
    },
    {
      "epoch": 34.720992028343666,
      "grad_norm": 0.21921059489250183,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 39200
    },
    {
      "epoch": 34.72542072630647,
      "grad_norm": 0.25159573554992676,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 39205
    },
    {
      "epoch": 34.72984942426926,
      "grad_norm": 0.2510949373245239,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 39210
    },
    {
      "epoch": 34.73427812223206,
      "grad_norm": 0.2238406538963318,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 39215
    },
    {
      "epoch": 34.738706820194864,
      "grad_norm": 0.23065055906772614,
      "learning_rate": 1e-05,
      "loss": 1.0101,
      "step": 39220
    },
    {
      "epoch": 34.743135518157665,
      "grad_norm": 0.2058543562889099,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 39225
    },
    {
      "epoch": 34.74756421612046,
      "grad_norm": 0.21793067455291748,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 39230
    },
    {
      "epoch": 34.75199291408326,
      "grad_norm": 0.2603197693824768,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 39235
    },
    {
      "epoch": 34.75642161204606,
      "grad_norm": 0.23578760027885437,
      "learning_rate": 1e-05,
      "loss": 0.915,
      "step": 39240
    },
    {
      "epoch": 34.760850310008856,
      "grad_norm": 0.218476340174675,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 39245
    },
    {
      "epoch": 34.76527900797166,
      "grad_norm": 0.26012128591537476,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 39250
    },
    {
      "epoch": 34.76970770593446,
      "grad_norm": 0.21792545914649963,
      "learning_rate": 1e-05,
      "loss": 1.0365,
      "step": 39255
    },
    {
      "epoch": 34.77413640389725,
      "grad_norm": 0.2458389401435852,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 39260
    },
    {
      "epoch": 34.77856510186005,
      "grad_norm": 0.19058483839035034,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 39265
    },
    {
      "epoch": 34.782993799822854,
      "grad_norm": 0.2451803833246231,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 39270
    },
    {
      "epoch": 34.78742249778565,
      "grad_norm": 0.3127826750278473,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 39275
    },
    {
      "epoch": 34.79185119574845,
      "grad_norm": 0.2494623363018036,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 39280
    },
    {
      "epoch": 34.79627989371125,
      "grad_norm": 0.2247854471206665,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 39285
    },
    {
      "epoch": 34.800708591674045,
      "grad_norm": 0.23555737733840942,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 39290
    },
    {
      "epoch": 34.805137289636846,
      "grad_norm": 0.20528742671012878,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 39295
    },
    {
      "epoch": 34.80956598759965,
      "grad_norm": 0.243865966796875,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 39300
    },
    {
      "epoch": 34.81399468556244,
      "grad_norm": 0.2291424721479416,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 39305
    },
    {
      "epoch": 34.81842338352524,
      "grad_norm": 0.3096823990345001,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 39310
    },
    {
      "epoch": 34.822852081488044,
      "grad_norm": 0.3650055527687073,
      "learning_rate": 1e-05,
      "loss": 1.0347,
      "step": 39315
    },
    {
      "epoch": 34.82728077945084,
      "grad_norm": 0.22102035582065582,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 39320
    },
    {
      "epoch": 34.83170947741364,
      "grad_norm": 0.20257841050624847,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 39325
    },
    {
      "epoch": 34.83613817537644,
      "grad_norm": 0.2753404974937439,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 39330
    },
    {
      "epoch": 34.84056687333924,
      "grad_norm": 0.226443350315094,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 39335
    },
    {
      "epoch": 34.844995571302036,
      "grad_norm": 0.23265786468982697,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 39340
    },
    {
      "epoch": 34.84942426926484,
      "grad_norm": 0.2845379114151001,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 39345
    },
    {
      "epoch": 34.85385296722764,
      "grad_norm": 0.21679973602294922,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 39350
    },
    {
      "epoch": 34.85828166519043,
      "grad_norm": 0.21578486263751984,
      "learning_rate": 1e-05,
      "loss": 0.9172,
      "step": 39355
    },
    {
      "epoch": 34.862710363153234,
      "grad_norm": 0.2823736071586609,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 39360
    },
    {
      "epoch": 34.867139061116035,
      "grad_norm": 0.23898911476135254,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 39365
    },
    {
      "epoch": 34.87156775907883,
      "grad_norm": 0.21913652122020721,
      "learning_rate": 1e-05,
      "loss": 1.0596,
      "step": 39370
    },
    {
      "epoch": 34.87599645704163,
      "grad_norm": 0.25310397148132324,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 39375
    },
    {
      "epoch": 34.88042515500443,
      "grad_norm": 0.2798282504081726,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 39380
    },
    {
      "epoch": 34.884853852967225,
      "grad_norm": 0.23490358889102936,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 39385
    },
    {
      "epoch": 34.88928255093003,
      "grad_norm": 0.2537546753883362,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 39390
    },
    {
      "epoch": 34.89371124889283,
      "grad_norm": 0.2565337121486664,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 39395
    },
    {
      "epoch": 34.89813994685562,
      "grad_norm": 0.20885735750198364,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 39400
    },
    {
      "epoch": 34.90256864481842,
      "grad_norm": 0.1987423151731491,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 39405
    },
    {
      "epoch": 34.906997342781224,
      "grad_norm": 0.22835476696491241,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 39410
    },
    {
      "epoch": 34.91142604074402,
      "grad_norm": 0.27420395612716675,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 39415
    },
    {
      "epoch": 34.91585473870682,
      "grad_norm": 0.2783346474170685,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 39420
    },
    {
      "epoch": 34.92028343666962,
      "grad_norm": 0.2649555802345276,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 39425
    },
    {
      "epoch": 34.924712134632415,
      "grad_norm": 0.2542673647403717,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 39430
    },
    {
      "epoch": 34.929140832595216,
      "grad_norm": 0.21622420847415924,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 39435
    },
    {
      "epoch": 34.93356953055802,
      "grad_norm": 0.2599319517612457,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 39440
    },
    {
      "epoch": 34.93799822852081,
      "grad_norm": 0.18314099311828613,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 39445
    },
    {
      "epoch": 34.94242692648361,
      "grad_norm": 0.3421282470226288,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 39450
    },
    {
      "epoch": 34.946855624446414,
      "grad_norm": 0.23871494829654694,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 39455
    },
    {
      "epoch": 34.951284322409215,
      "grad_norm": 0.23737779259681702,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 39460
    },
    {
      "epoch": 34.95571302037201,
      "grad_norm": 0.2778080403804779,
      "learning_rate": 1e-05,
      "loss": 0.9155,
      "step": 39465
    },
    {
      "epoch": 34.96014171833481,
      "grad_norm": 0.23091882467269897,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 39470
    },
    {
      "epoch": 34.96457041629761,
      "grad_norm": 0.20534102618694305,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 39475
    },
    {
      "epoch": 34.968999114260406,
      "grad_norm": 0.26630571484565735,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 39480
    },
    {
      "epoch": 34.97342781222321,
      "grad_norm": 0.26347237825393677,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 39485
    },
    {
      "epoch": 34.97785651018601,
      "grad_norm": 0.2595798671245575,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 39490
    },
    {
      "epoch": 34.9822852081488,
      "grad_norm": 0.22914078831672668,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 39495
    },
    {
      "epoch": 34.9867139061116,
      "grad_norm": 0.20719504356384277,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 39500
    },
    {
      "epoch": 34.991142604074405,
      "grad_norm": 0.22324219346046448,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 39505
    },
    {
      "epoch": 34.9955713020372,
      "grad_norm": 0.22987961769104004,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 39510
    },
    {
      "epoch": 35.0,
      "grad_norm": 0.2068951427936554,
      "learning_rate": 1e-05,
      "loss": 1.0492,
      "step": 39515
    },
    {
      "epoch": 35.0044286979628,
      "grad_norm": 0.23593661189079285,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 39520
    },
    {
      "epoch": 35.008857395925595,
      "grad_norm": 0.315938264131546,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 39525
    },
    {
      "epoch": 35.0132860938884,
      "grad_norm": 0.2850911617279053,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 39530
    },
    {
      "epoch": 35.0177147918512,
      "grad_norm": 0.21571305394172668,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 39535
    },
    {
      "epoch": 35.02214348981399,
      "grad_norm": 0.23679733276367188,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 39540
    },
    {
      "epoch": 35.02657218777679,
      "grad_norm": 0.22444473206996918,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 39545
    },
    {
      "epoch": 35.031000885739594,
      "grad_norm": 0.22798216342926025,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 39550
    },
    {
      "epoch": 35.03542958370239,
      "grad_norm": 0.23759490251541138,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 39555
    },
    {
      "epoch": 35.03985828166519,
      "grad_norm": 0.25344133377075195,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 39560
    },
    {
      "epoch": 35.04428697962799,
      "grad_norm": 0.2339189499616623,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 39565
    },
    {
      "epoch": 35.048715677590785,
      "grad_norm": 0.24252775311470032,
      "learning_rate": 1e-05,
      "loss": 0.9104,
      "step": 39570
    },
    {
      "epoch": 35.053144375553586,
      "grad_norm": 0.24591848254203796,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 39575
    },
    {
      "epoch": 35.05757307351639,
      "grad_norm": 0.2737513780593872,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 39580
    },
    {
      "epoch": 35.06200177147919,
      "grad_norm": 0.24790601432323456,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 39585
    },
    {
      "epoch": 35.06643046944198,
      "grad_norm": 0.21346385776996613,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 39590
    },
    {
      "epoch": 35.070859167404784,
      "grad_norm": 0.2736019492149353,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 39595
    },
    {
      "epoch": 35.075287865367585,
      "grad_norm": 0.2315564900636673,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 39600
    },
    {
      "epoch": 35.07971656333038,
      "grad_norm": 0.23989319801330566,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 39605
    },
    {
      "epoch": 35.08414526129318,
      "grad_norm": 0.3100568652153015,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 39610
    },
    {
      "epoch": 35.08857395925598,
      "grad_norm": 0.26376354694366455,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 39615
    },
    {
      "epoch": 35.093002657218776,
      "grad_norm": 0.2705543637275696,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 39620
    },
    {
      "epoch": 35.09743135518158,
      "grad_norm": 0.21611632406711578,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 39625
    },
    {
      "epoch": 35.10186005314438,
      "grad_norm": 0.2686470150947571,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 39630
    },
    {
      "epoch": 35.10628875110717,
      "grad_norm": 0.21684277057647705,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 39635
    },
    {
      "epoch": 35.11071744906997,
      "grad_norm": 0.23775018751621246,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 39640
    },
    {
      "epoch": 35.115146147032775,
      "grad_norm": 0.251823753118515,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 39645
    },
    {
      "epoch": 35.11957484499557,
      "grad_norm": 0.27192607522010803,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 39650
    },
    {
      "epoch": 35.12400354295837,
      "grad_norm": 0.21933510899543762,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 39655
    },
    {
      "epoch": 35.12843224092117,
      "grad_norm": 0.2719583511352539,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 39660
    },
    {
      "epoch": 35.132860938883965,
      "grad_norm": 0.26673954725265503,
      "learning_rate": 1e-05,
      "loss": 0.8995,
      "step": 39665
    },
    {
      "epoch": 35.137289636846766,
      "grad_norm": 0.26488256454467773,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 39670
    },
    {
      "epoch": 35.14171833480957,
      "grad_norm": 0.2852939963340759,
      "learning_rate": 1e-05,
      "loss": 0.9168,
      "step": 39675
    },
    {
      "epoch": 35.14614703277236,
      "grad_norm": 0.2433881014585495,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 39680
    },
    {
      "epoch": 35.15057573073516,
      "grad_norm": 0.2733299136161804,
      "learning_rate": 1e-05,
      "loss": 1.0203,
      "step": 39685
    },
    {
      "epoch": 35.155004428697964,
      "grad_norm": 0.2582400143146515,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 39690
    },
    {
      "epoch": 35.15943312666076,
      "grad_norm": 0.2586967349052429,
      "learning_rate": 1e-05,
      "loss": 1.0354,
      "step": 39695
    },
    {
      "epoch": 35.16386182462356,
      "grad_norm": 0.24145683646202087,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 39700
    },
    {
      "epoch": 35.16829052258636,
      "grad_norm": 0.22906555235385895,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 39705
    },
    {
      "epoch": 35.17271922054916,
      "grad_norm": 0.22689169645309448,
      "learning_rate": 1e-05,
      "loss": 0.9319,
      "step": 39710
    },
    {
      "epoch": 35.177147918511956,
      "grad_norm": 0.2927563488483429,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 39715
    },
    {
      "epoch": 35.18157661647476,
      "grad_norm": 0.24924597144126892,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 39720
    },
    {
      "epoch": 35.18600531443756,
      "grad_norm": 0.2501330077648163,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 39725
    },
    {
      "epoch": 35.19043401240035,
      "grad_norm": 0.3229110836982727,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 39730
    },
    {
      "epoch": 35.194862710363154,
      "grad_norm": 0.25542911887168884,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 39735
    },
    {
      "epoch": 35.199291408325955,
      "grad_norm": 0.22127750515937805,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 39740
    },
    {
      "epoch": 35.20372010628875,
      "grad_norm": 0.22302144765853882,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 39745
    },
    {
      "epoch": 35.20814880425155,
      "grad_norm": 0.24056854844093323,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 39750
    },
    {
      "epoch": 35.21257750221435,
      "grad_norm": 0.23887591063976288,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 39755
    },
    {
      "epoch": 35.217006200177146,
      "grad_norm": 0.29898229241371155,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 39760
    },
    {
      "epoch": 35.22143489813995,
      "grad_norm": 0.24393925070762634,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 39765
    },
    {
      "epoch": 35.22586359610275,
      "grad_norm": 0.2737123668193817,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 39770
    },
    {
      "epoch": 35.23029229406554,
      "grad_norm": 0.22036582231521606,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 39775
    },
    {
      "epoch": 35.23472099202834,
      "grad_norm": 0.28304049372673035,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 39780
    },
    {
      "epoch": 35.239149689991144,
      "grad_norm": 0.25792399048805237,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 39785
    },
    {
      "epoch": 35.24357838795394,
      "grad_norm": 0.2297327071428299,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 39790
    },
    {
      "epoch": 35.24800708591674,
      "grad_norm": 0.24497422575950623,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 39795
    },
    {
      "epoch": 35.25243578387954,
      "grad_norm": 0.24111978709697723,
      "learning_rate": 1e-05,
      "loss": 1.0242,
      "step": 39800
    },
    {
      "epoch": 35.256864481842335,
      "grad_norm": 0.21529093384742737,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 39805
    },
    {
      "epoch": 35.261293179805136,
      "grad_norm": 0.19911101460456848,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 39810
    },
    {
      "epoch": 35.26572187776794,
      "grad_norm": 0.22599324584007263,
      "learning_rate": 1e-05,
      "loss": 0.9252,
      "step": 39815
    },
    {
      "epoch": 35.27015057573073,
      "grad_norm": 0.3066536784172058,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 39820
    },
    {
      "epoch": 35.27457927369353,
      "grad_norm": 0.22253525257110596,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 39825
    },
    {
      "epoch": 35.279007971656334,
      "grad_norm": 0.24628353118896484,
      "learning_rate": 1e-05,
      "loss": 1.0054,
      "step": 39830
    },
    {
      "epoch": 35.283436669619135,
      "grad_norm": 0.28410613536834717,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 39835
    },
    {
      "epoch": 35.28786536758193,
      "grad_norm": 0.2228560894727707,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 39840
    },
    {
      "epoch": 35.29229406554473,
      "grad_norm": 0.2113189399242401,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 39845
    },
    {
      "epoch": 35.29672276350753,
      "grad_norm": 0.2591398060321808,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 39850
    },
    {
      "epoch": 35.301151461470326,
      "grad_norm": 0.2836805582046509,
      "learning_rate": 1e-05,
      "loss": 1.0385,
      "step": 39855
    },
    {
      "epoch": 35.30558015943313,
      "grad_norm": 0.23124606907367706,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 39860
    },
    {
      "epoch": 35.31000885739593,
      "grad_norm": 0.25152158737182617,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 39865
    },
    {
      "epoch": 35.31443755535872,
      "grad_norm": 0.273102730512619,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 39870
    },
    {
      "epoch": 35.318866253321524,
      "grad_norm": 0.2898319959640503,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 39875
    },
    {
      "epoch": 35.323294951284325,
      "grad_norm": 0.23062893748283386,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 39880
    },
    {
      "epoch": 35.32772364924712,
      "grad_norm": 0.2247716337442398,
      "learning_rate": 1e-05,
      "loss": 1.0273,
      "step": 39885
    },
    {
      "epoch": 35.33215234720992,
      "grad_norm": 0.2621605694293976,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 39890
    },
    {
      "epoch": 35.33658104517272,
      "grad_norm": 0.24151019752025604,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 39895
    },
    {
      "epoch": 35.341009743135515,
      "grad_norm": 0.21506355702877045,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 39900
    },
    {
      "epoch": 35.34543844109832,
      "grad_norm": 0.23143921792507172,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 39905
    },
    {
      "epoch": 35.34986713906112,
      "grad_norm": 0.22458681464195251,
      "learning_rate": 1e-05,
      "loss": 1.0175,
      "step": 39910
    },
    {
      "epoch": 35.35429583702391,
      "grad_norm": 0.25023940205574036,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 39915
    },
    {
      "epoch": 35.35872453498671,
      "grad_norm": 0.24931608140468597,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 39920
    },
    {
      "epoch": 35.363153232949514,
      "grad_norm": 0.24820096790790558,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 39925
    },
    {
      "epoch": 35.36758193091231,
      "grad_norm": 0.2608039081096649,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 39930
    },
    {
      "epoch": 35.37201062887511,
      "grad_norm": 0.26962557435035706,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 39935
    },
    {
      "epoch": 35.37643932683791,
      "grad_norm": 0.25606101751327515,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 39940
    },
    {
      "epoch": 35.380868024800705,
      "grad_norm": 0.2733430862426758,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 39945
    },
    {
      "epoch": 35.385296722763506,
      "grad_norm": 0.2497810423374176,
      "learning_rate": 1e-05,
      "loss": 0.9209,
      "step": 39950
    },
    {
      "epoch": 35.38972542072631,
      "grad_norm": 0.22486664354801178,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 39955
    },
    {
      "epoch": 35.39415411868911,
      "grad_norm": 0.24633093178272247,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 39960
    },
    {
      "epoch": 35.3985828166519,
      "grad_norm": 0.232732892036438,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 39965
    },
    {
      "epoch": 35.403011514614704,
      "grad_norm": 0.2210260033607483,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 39970
    },
    {
      "epoch": 35.407440212577505,
      "grad_norm": 0.23340089619159698,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 39975
    },
    {
      "epoch": 35.4118689105403,
      "grad_norm": 0.23357854783535004,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 39980
    },
    {
      "epoch": 35.4162976085031,
      "grad_norm": 0.2467680126428604,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 39985
    },
    {
      "epoch": 35.4207263064659,
      "grad_norm": 0.24557289481163025,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 39990
    },
    {
      "epoch": 35.425155004428696,
      "grad_norm": 0.22820210456848145,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 39995
    },
    {
      "epoch": 35.4295837023915,
      "grad_norm": 0.27644363045692444,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 40000
    },
    {
      "epoch": 35.4340124003543,
      "grad_norm": 0.3036370277404785,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 40005
    },
    {
      "epoch": 35.43844109831709,
      "grad_norm": 0.2707492709159851,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 40010
    },
    {
      "epoch": 35.44286979627989,
      "grad_norm": 0.22467005252838135,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 40015
    },
    {
      "epoch": 35.447298494242695,
      "grad_norm": 0.2091584950685501,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 40020
    },
    {
      "epoch": 35.45172719220549,
      "grad_norm": 0.2864615321159363,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 40025
    },
    {
      "epoch": 35.45615589016829,
      "grad_norm": 0.2828342616558075,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 40030
    },
    {
      "epoch": 35.46058458813109,
      "grad_norm": 0.23655110597610474,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 40035
    },
    {
      "epoch": 35.465013286093885,
      "grad_norm": 0.2519761621952057,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 40040
    },
    {
      "epoch": 35.46944198405669,
      "grad_norm": 0.2154015600681305,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 40045
    },
    {
      "epoch": 35.47387068201949,
      "grad_norm": 0.2232111096382141,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 40050
    },
    {
      "epoch": 35.47829937998228,
      "grad_norm": 0.22750414907932281,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 40055
    },
    {
      "epoch": 35.48272807794508,
      "grad_norm": 0.22685526311397552,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 40060
    },
    {
      "epoch": 35.487156775907884,
      "grad_norm": 0.218906432390213,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 40065
    },
    {
      "epoch": 35.491585473870686,
      "grad_norm": 0.271619588136673,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 40070
    },
    {
      "epoch": 35.49601417183348,
      "grad_norm": 0.23013056814670563,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 40075
    },
    {
      "epoch": 35.50044286979628,
      "grad_norm": 0.25824683904647827,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 40080
    },
    {
      "epoch": 35.50487156775908,
      "grad_norm": 0.26319095492362976,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 40085
    },
    {
      "epoch": 35.509300265721876,
      "grad_norm": 0.279031902551651,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 40090
    },
    {
      "epoch": 35.51372896368468,
      "grad_norm": 0.2286226451396942,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 40095
    },
    {
      "epoch": 35.51815766164748,
      "grad_norm": 0.28177499771118164,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 40100
    },
    {
      "epoch": 35.52258635961027,
      "grad_norm": 0.2695818543434143,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 40105
    },
    {
      "epoch": 35.527015057573074,
      "grad_norm": 0.22232311964035034,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 40110
    },
    {
      "epoch": 35.531443755535875,
      "grad_norm": 0.2320026308298111,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 40115
    },
    {
      "epoch": 35.53587245349867,
      "grad_norm": 0.21246756613254547,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 40120
    },
    {
      "epoch": 35.54030115146147,
      "grad_norm": 0.2931070625782013,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 40125
    },
    {
      "epoch": 35.54472984942427,
      "grad_norm": 0.21984563767910004,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 40130
    },
    {
      "epoch": 35.549158547387066,
      "grad_norm": 0.2637583911418915,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 40135
    },
    {
      "epoch": 35.55358724534987,
      "grad_norm": 0.2627270519733429,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 40140
    },
    {
      "epoch": 35.55801594331267,
      "grad_norm": 0.23151454329490662,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 40145
    },
    {
      "epoch": 35.56244464127546,
      "grad_norm": 0.2779940962791443,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 40150
    },
    {
      "epoch": 35.56687333923826,
      "grad_norm": 0.22090676426887512,
      "learning_rate": 1e-05,
      "loss": 1.0245,
      "step": 40155
    },
    {
      "epoch": 35.571302037201065,
      "grad_norm": 0.2810547649860382,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 40160
    },
    {
      "epoch": 35.57573073516386,
      "grad_norm": 0.23355476558208466,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 40165
    },
    {
      "epoch": 35.58015943312666,
      "grad_norm": 0.23681817948818207,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 40170
    },
    {
      "epoch": 35.58458813108946,
      "grad_norm": 0.2516898810863495,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 40175
    },
    {
      "epoch": 35.589016829052255,
      "grad_norm": 0.2581563889980316,
      "learning_rate": 1e-05,
      "loss": 1.0385,
      "step": 40180
    },
    {
      "epoch": 35.59344552701506,
      "grad_norm": 0.2318749874830246,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 40185
    },
    {
      "epoch": 35.59787422497786,
      "grad_norm": 0.3203091323375702,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 40190
    },
    {
      "epoch": 35.60230292294066,
      "grad_norm": 0.2570744752883911,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 40195
    },
    {
      "epoch": 35.60673162090345,
      "grad_norm": 0.28338080644607544,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 40200
    },
    {
      "epoch": 35.611160318866254,
      "grad_norm": 0.23535174131393433,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 40205
    },
    {
      "epoch": 35.615589016829055,
      "grad_norm": 0.2677377760410309,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 40210
    },
    {
      "epoch": 35.62001771479185,
      "grad_norm": 0.26248013973236084,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 40215
    },
    {
      "epoch": 35.62444641275465,
      "grad_norm": 0.28670191764831543,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 40220
    },
    {
      "epoch": 35.62887511071745,
      "grad_norm": 0.2198255956172943,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 40225
    },
    {
      "epoch": 35.633303808680246,
      "grad_norm": 0.24472220242023468,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 40230
    },
    {
      "epoch": 35.63773250664305,
      "grad_norm": 0.19831225275993347,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 40235
    },
    {
      "epoch": 35.64216120460585,
      "grad_norm": 0.2829912304878235,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 40240
    },
    {
      "epoch": 35.64658990256864,
      "grad_norm": 0.24533922970294952,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 40245
    },
    {
      "epoch": 35.651018600531444,
      "grad_norm": 0.2577371895313263,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 40250
    },
    {
      "epoch": 35.655447298494245,
      "grad_norm": 0.2771191895008087,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 40255
    },
    {
      "epoch": 35.65987599645704,
      "grad_norm": 0.2723691761493683,
      "learning_rate": 1e-05,
      "loss": 1.0382,
      "step": 40260
    },
    {
      "epoch": 35.66430469441984,
      "grad_norm": 0.23274526000022888,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 40265
    },
    {
      "epoch": 35.66873339238264,
      "grad_norm": 0.2028035968542099,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 40270
    },
    {
      "epoch": 35.673162090345436,
      "grad_norm": 0.24089133739471436,
      "learning_rate": 1e-05,
      "loss": 0.9342,
      "step": 40275
    },
    {
      "epoch": 35.67759078830824,
      "grad_norm": 0.2692558765411377,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 40280
    },
    {
      "epoch": 35.68201948627104,
      "grad_norm": 0.23566405475139618,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 40285
    },
    {
      "epoch": 35.68644818423383,
      "grad_norm": 0.22322937846183777,
      "learning_rate": 1e-05,
      "loss": 1.0599,
      "step": 40290
    },
    {
      "epoch": 35.69087688219663,
      "grad_norm": 0.2418806105852127,
      "learning_rate": 1e-05,
      "loss": 1.0301,
      "step": 40295
    },
    {
      "epoch": 35.695305580159435,
      "grad_norm": 0.2599877417087555,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 40300
    },
    {
      "epoch": 35.69973427812223,
      "grad_norm": 0.27923688292503357,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 40305
    },
    {
      "epoch": 35.70416297608503,
      "grad_norm": 0.28744441270828247,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 40310
    },
    {
      "epoch": 35.70859167404783,
      "grad_norm": 0.25891977548599243,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 40315
    },
    {
      "epoch": 35.71302037201063,
      "grad_norm": 0.30485567450523376,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 40320
    },
    {
      "epoch": 35.717449069973426,
      "grad_norm": 0.22354458272457123,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 40325
    },
    {
      "epoch": 35.72187776793623,
      "grad_norm": 0.26116660237312317,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 40330
    },
    {
      "epoch": 35.72630646589903,
      "grad_norm": 0.24663841724395752,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 40335
    },
    {
      "epoch": 35.73073516386182,
      "grad_norm": 0.27957314252853394,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 40340
    },
    {
      "epoch": 35.735163861824624,
      "grad_norm": 0.27709126472473145,
      "learning_rate": 1e-05,
      "loss": 1.0266,
      "step": 40345
    },
    {
      "epoch": 35.739592559787425,
      "grad_norm": 0.2336152046918869,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 40350
    },
    {
      "epoch": 35.74402125775022,
      "grad_norm": 0.24564549326896667,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 40355
    },
    {
      "epoch": 35.74844995571302,
      "grad_norm": 0.2178659737110138,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 40360
    },
    {
      "epoch": 35.75287865367582,
      "grad_norm": 0.2633017897605896,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 40365
    },
    {
      "epoch": 35.757307351638616,
      "grad_norm": 0.23075000941753387,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 40370
    },
    {
      "epoch": 35.76173604960142,
      "grad_norm": 0.24644388258457184,
      "learning_rate": 1e-05,
      "loss": 0.9283,
      "step": 40375
    },
    {
      "epoch": 35.76616474756422,
      "grad_norm": 0.1923307478427887,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 40380
    },
    {
      "epoch": 35.77059344552701,
      "grad_norm": 0.2611455023288727,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 40385
    },
    {
      "epoch": 35.775022143489814,
      "grad_norm": 0.25513362884521484,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 40390
    },
    {
      "epoch": 35.779450841452615,
      "grad_norm": 0.2518458366394043,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 40395
    },
    {
      "epoch": 35.78387953941541,
      "grad_norm": 0.299014151096344,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 40400
    },
    {
      "epoch": 35.78830823737821,
      "grad_norm": 0.2497718781232834,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 40405
    },
    {
      "epoch": 35.79273693534101,
      "grad_norm": 0.23372980952262878,
      "learning_rate": 1e-05,
      "loss": 1.0559,
      "step": 40410
    },
    {
      "epoch": 35.797165633303806,
      "grad_norm": 0.2446320503950119,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 40415
    },
    {
      "epoch": 35.80159433126661,
      "grad_norm": 0.2172650396823883,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 40420
    },
    {
      "epoch": 35.80602302922941,
      "grad_norm": 0.22092875838279724,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 40425
    },
    {
      "epoch": 35.8104517271922,
      "grad_norm": 0.2555411756038666,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 40430
    },
    {
      "epoch": 35.814880425155,
      "grad_norm": 0.2262481302022934,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 40435
    },
    {
      "epoch": 35.819309123117804,
      "grad_norm": 0.24759456515312195,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 40440
    },
    {
      "epoch": 35.823737821080606,
      "grad_norm": 0.2330486923456192,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 40445
    },
    {
      "epoch": 35.8281665190434,
      "grad_norm": 0.2492523342370987,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 40450
    },
    {
      "epoch": 35.8325952170062,
      "grad_norm": 0.2578888237476349,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 40455
    },
    {
      "epoch": 35.837023914969,
      "grad_norm": 0.23786818981170654,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 40460
    },
    {
      "epoch": 35.841452612931796,
      "grad_norm": 0.2568472921848297,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 40465
    },
    {
      "epoch": 35.8458813108946,
      "grad_norm": 0.2453257292509079,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 40470
    },
    {
      "epoch": 35.8503100088574,
      "grad_norm": 0.2947475016117096,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 40475
    },
    {
      "epoch": 35.85473870682019,
      "grad_norm": 0.223874032497406,
      "learning_rate": 1e-05,
      "loss": 0.9081,
      "step": 40480
    },
    {
      "epoch": 35.859167404782994,
      "grad_norm": 0.23459452390670776,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 40485
    },
    {
      "epoch": 35.863596102745795,
      "grad_norm": 0.26651930809020996,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 40490
    },
    {
      "epoch": 35.86802480070859,
      "grad_norm": 0.2747059166431427,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 40495
    },
    {
      "epoch": 35.87245349867139,
      "grad_norm": 0.22639040648937225,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 40500
    },
    {
      "epoch": 35.87688219663419,
      "grad_norm": 0.2218010574579239,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 40505
    },
    {
      "epoch": 35.881310894596986,
      "grad_norm": 0.22835911810398102,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 40510
    },
    {
      "epoch": 35.88573959255979,
      "grad_norm": 0.24543921649456024,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 40515
    },
    {
      "epoch": 35.89016829052259,
      "grad_norm": 0.26899459958076477,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 40520
    },
    {
      "epoch": 35.89459698848538,
      "grad_norm": 0.22196191549301147,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 40525
    },
    {
      "epoch": 35.899025686448184,
      "grad_norm": 0.23421040177345276,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 40530
    },
    {
      "epoch": 35.903454384410985,
      "grad_norm": 0.2716995179653168,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 40535
    },
    {
      "epoch": 35.90788308237378,
      "grad_norm": 0.20850670337677002,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 40540
    },
    {
      "epoch": 35.91231178033658,
      "grad_norm": 0.23377592861652374,
      "learning_rate": 1e-05,
      "loss": 1.0152,
      "step": 40545
    },
    {
      "epoch": 35.91674047829938,
      "grad_norm": 0.21797043085098267,
      "learning_rate": 1e-05,
      "loss": 1.0193,
      "step": 40550
    },
    {
      "epoch": 35.921169176262175,
      "grad_norm": 0.2463165670633316,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 40555
    },
    {
      "epoch": 35.92559787422498,
      "grad_norm": 0.22919940948486328,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 40560
    },
    {
      "epoch": 35.93002657218778,
      "grad_norm": 0.2925037145614624,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 40565
    },
    {
      "epoch": 35.93445527015058,
      "grad_norm": 0.22773735225200653,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 40570
    },
    {
      "epoch": 35.93888396811337,
      "grad_norm": 0.22901175916194916,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 40575
    },
    {
      "epoch": 35.943312666076174,
      "grad_norm": 0.3081228733062744,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 40580
    },
    {
      "epoch": 35.947741364038976,
      "grad_norm": 0.3218163251876831,
      "learning_rate": 1e-05,
      "loss": 0.9114,
      "step": 40585
    },
    {
      "epoch": 35.95217006200177,
      "grad_norm": 0.22988595068454742,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 40590
    },
    {
      "epoch": 35.95659875996457,
      "grad_norm": 0.23910291492938995,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 40595
    },
    {
      "epoch": 35.96102745792737,
      "grad_norm": 0.3272089660167694,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 40600
    },
    {
      "epoch": 35.965456155890166,
      "grad_norm": 0.23385964334011078,
      "learning_rate": 1e-05,
      "loss": 1.0337,
      "step": 40605
    },
    {
      "epoch": 35.96988485385297,
      "grad_norm": 0.24690861999988556,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 40610
    },
    {
      "epoch": 35.97431355181577,
      "grad_norm": 0.2402074784040451,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 40615
    },
    {
      "epoch": 35.97874224977856,
      "grad_norm": 0.2547287344932556,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 40620
    },
    {
      "epoch": 35.983170947741364,
      "grad_norm": 0.25014543533325195,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 40625
    },
    {
      "epoch": 35.987599645704165,
      "grad_norm": 0.2261633425951004,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 40630
    },
    {
      "epoch": 35.99202834366696,
      "grad_norm": 0.22299930453300476,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 40635
    },
    {
      "epoch": 35.99645704162976,
      "grad_norm": 0.23826146125793457,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 40640
    },
    {
      "epoch": 36.00088573959256,
      "grad_norm": 0.22658748924732208,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 40645
    },
    {
      "epoch": 36.005314437555356,
      "grad_norm": 0.20685550570487976,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 40650
    },
    {
      "epoch": 36.00974313551816,
      "grad_norm": 0.24671468138694763,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 40655
    },
    {
      "epoch": 36.01417183348096,
      "grad_norm": 0.24889251589775085,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 40660
    },
    {
      "epoch": 36.01860053144375,
      "grad_norm": 0.2482403814792633,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 40665
    },
    {
      "epoch": 36.02302922940655,
      "grad_norm": 0.22407257556915283,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 40670
    },
    {
      "epoch": 36.027457927369355,
      "grad_norm": 0.22811315953731537,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 40675
    },
    {
      "epoch": 36.03188662533215,
      "grad_norm": 0.24326738715171814,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 40680
    },
    {
      "epoch": 36.03631532329495,
      "grad_norm": 0.22853440046310425,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 40685
    },
    {
      "epoch": 36.04074402125775,
      "grad_norm": 0.24981658160686493,
      "learning_rate": 1e-05,
      "loss": 0.9271,
      "step": 40690
    },
    {
      "epoch": 36.04517271922055,
      "grad_norm": 0.22410358488559723,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 40695
    },
    {
      "epoch": 36.04960141718335,
      "grad_norm": 0.2527514398097992,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 40700
    },
    {
      "epoch": 36.05403011514615,
      "grad_norm": 0.22499558329582214,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 40705
    },
    {
      "epoch": 36.05845881310895,
      "grad_norm": 0.2345588058233261,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 40710
    },
    {
      "epoch": 36.06288751107174,
      "grad_norm": 0.23372536897659302,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 40715
    },
    {
      "epoch": 36.067316209034544,
      "grad_norm": 0.24152472615242004,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 40720
    },
    {
      "epoch": 36.071744906997345,
      "grad_norm": 0.23476995527744293,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 40725
    },
    {
      "epoch": 36.07617360496014,
      "grad_norm": 0.332564115524292,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 40730
    },
    {
      "epoch": 36.08060230292294,
      "grad_norm": 0.23601870238780975,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 40735
    },
    {
      "epoch": 36.08503100088574,
      "grad_norm": 0.24262824654579163,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 40740
    },
    {
      "epoch": 36.089459698848536,
      "grad_norm": 0.2730354070663452,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 40745
    },
    {
      "epoch": 36.09388839681134,
      "grad_norm": 0.28003156185150146,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 40750
    },
    {
      "epoch": 36.09831709477414,
      "grad_norm": 0.26177123188972473,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 40755
    },
    {
      "epoch": 36.10274579273693,
      "grad_norm": 0.2058447003364563,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 40760
    },
    {
      "epoch": 36.107174490699734,
      "grad_norm": 0.21758095920085907,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 40765
    },
    {
      "epoch": 36.111603188662535,
      "grad_norm": 0.21229171752929688,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 40770
    },
    {
      "epoch": 36.11603188662533,
      "grad_norm": 0.2983495593070984,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 40775
    },
    {
      "epoch": 36.12046058458813,
      "grad_norm": 0.23964758217334747,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 40780
    },
    {
      "epoch": 36.12488928255093,
      "grad_norm": 0.2386459857225418,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 40785
    },
    {
      "epoch": 36.129317980513726,
      "grad_norm": 0.2813543975353241,
      "learning_rate": 1e-05,
      "loss": 0.918,
      "step": 40790
    },
    {
      "epoch": 36.13374667847653,
      "grad_norm": 0.26468729972839355,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 40795
    },
    {
      "epoch": 36.13817537643933,
      "grad_norm": 0.25739598274230957,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 40800
    },
    {
      "epoch": 36.14260407440213,
      "grad_norm": 0.23630237579345703,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 40805
    },
    {
      "epoch": 36.14703277236492,
      "grad_norm": 0.2426365166902542,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 40810
    },
    {
      "epoch": 36.151461470327725,
      "grad_norm": 0.24553999304771423,
      "learning_rate": 1e-05,
      "loss": 1.0207,
      "step": 40815
    },
    {
      "epoch": 36.155890168290526,
      "grad_norm": 0.23183932900428772,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 40820
    },
    {
      "epoch": 36.16031886625332,
      "grad_norm": 0.2705097496509552,
      "learning_rate": 1e-05,
      "loss": 1.0494,
      "step": 40825
    },
    {
      "epoch": 36.16474756421612,
      "grad_norm": 0.20186571776866913,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 40830
    },
    {
      "epoch": 36.16917626217892,
      "grad_norm": 0.21930205821990967,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 40835
    },
    {
      "epoch": 36.173604960141716,
      "grad_norm": 0.27030783891677856,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 40840
    },
    {
      "epoch": 36.17803365810452,
      "grad_norm": 0.23785150051116943,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 40845
    },
    {
      "epoch": 36.18246235606732,
      "grad_norm": 0.23985576629638672,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 40850
    },
    {
      "epoch": 36.18689105403011,
      "grad_norm": 0.24489888548851013,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 40855
    },
    {
      "epoch": 36.191319751992914,
      "grad_norm": 0.22954946756362915,
      "learning_rate": 1e-05,
      "loss": 1.0347,
      "step": 40860
    },
    {
      "epoch": 36.195748449955715,
      "grad_norm": 0.23638089001178741,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 40865
    },
    {
      "epoch": 36.20017714791851,
      "grad_norm": 0.23923087120056152,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 40870
    },
    {
      "epoch": 36.20460584588131,
      "grad_norm": 0.2198793590068817,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 40875
    },
    {
      "epoch": 36.20903454384411,
      "grad_norm": 0.2216985672712326,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 40880
    },
    {
      "epoch": 36.213463241806906,
      "grad_norm": 0.25956448912620544,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 40885
    },
    {
      "epoch": 36.21789193976971,
      "grad_norm": 0.21048910915851593,
      "learning_rate": 1e-05,
      "loss": 1.0148,
      "step": 40890
    },
    {
      "epoch": 36.22232063773251,
      "grad_norm": 0.277170330286026,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 40895
    },
    {
      "epoch": 36.2267493356953,
      "grad_norm": 0.23861654102802277,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 40900
    },
    {
      "epoch": 36.231178033658104,
      "grad_norm": 0.2342303991317749,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 40905
    },
    {
      "epoch": 36.235606731620905,
      "grad_norm": 0.27772581577301025,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 40910
    },
    {
      "epoch": 36.2400354295837,
      "grad_norm": 0.25007620453834534,
      "learning_rate": 1e-05,
      "loss": 0.9301,
      "step": 40915
    },
    {
      "epoch": 36.2444641275465,
      "grad_norm": 0.234497532248497,
      "learning_rate": 1e-05,
      "loss": 1.0619,
      "step": 40920
    },
    {
      "epoch": 36.2488928255093,
      "grad_norm": 0.2606408894062042,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 40925
    },
    {
      "epoch": 36.2533215234721,
      "grad_norm": 0.2422298938035965,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 40930
    },
    {
      "epoch": 36.2577502214349,
      "grad_norm": 0.2119947373867035,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 40935
    },
    {
      "epoch": 36.2621789193977,
      "grad_norm": 0.25420472025871277,
      "learning_rate": 1e-05,
      "loss": 1.0521,
      "step": 40940
    },
    {
      "epoch": 36.2666076173605,
      "grad_norm": 0.25499996542930603,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 40945
    },
    {
      "epoch": 36.27103631532329,
      "grad_norm": 0.2024070769548416,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 40950
    },
    {
      "epoch": 36.275465013286095,
      "grad_norm": 0.21779686212539673,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 40955
    },
    {
      "epoch": 36.279893711248896,
      "grad_norm": 0.24275332689285278,
      "learning_rate": 1e-05,
      "loss": 1.0219,
      "step": 40960
    },
    {
      "epoch": 36.28432240921169,
      "grad_norm": 0.23373295366764069,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 40965
    },
    {
      "epoch": 36.28875110717449,
      "grad_norm": 0.24112732708454132,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 40970
    },
    {
      "epoch": 36.29317980513729,
      "grad_norm": 0.22080321609973907,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 40975
    },
    {
      "epoch": 36.297608503100086,
      "grad_norm": 0.21972325444221497,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 40980
    },
    {
      "epoch": 36.30203720106289,
      "grad_norm": 0.23970209062099457,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 40985
    },
    {
      "epoch": 36.30646589902569,
      "grad_norm": 0.24055175483226776,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 40990
    },
    {
      "epoch": 36.31089459698848,
      "grad_norm": 0.24951638281345367,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 40995
    },
    {
      "epoch": 36.315323294951284,
      "grad_norm": 0.23876352608203888,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 41000
    },
    {
      "epoch": 36.319751992914085,
      "grad_norm": 0.29305002093315125,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 41005
    },
    {
      "epoch": 36.32418069087688,
      "grad_norm": 0.2657853364944458,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 41010
    },
    {
      "epoch": 36.32860938883968,
      "grad_norm": 0.25105082988739014,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 41015
    },
    {
      "epoch": 36.33303808680248,
      "grad_norm": 0.2244410514831543,
      "learning_rate": 1e-05,
      "loss": 1.0627,
      "step": 41020
    },
    {
      "epoch": 36.337466784765276,
      "grad_norm": 0.23671582341194153,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 41025
    },
    {
      "epoch": 36.34189548272808,
      "grad_norm": 0.2655124366283417,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 41030
    },
    {
      "epoch": 36.34632418069088,
      "grad_norm": 0.2666459083557129,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 41035
    },
    {
      "epoch": 36.35075287865367,
      "grad_norm": 0.25170886516571045,
      "learning_rate": 1e-05,
      "loss": 1.0348,
      "step": 41040
    },
    {
      "epoch": 36.355181576616474,
      "grad_norm": 0.2830251157283783,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 41045
    },
    {
      "epoch": 36.359610274579275,
      "grad_norm": 0.27102768421173096,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 41050
    },
    {
      "epoch": 36.364038972542076,
      "grad_norm": 0.24004510045051575,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 41055
    },
    {
      "epoch": 36.36846767050487,
      "grad_norm": 0.2342505156993866,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 41060
    },
    {
      "epoch": 36.37289636846767,
      "grad_norm": 0.22039338946342468,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 41065
    },
    {
      "epoch": 36.37732506643047,
      "grad_norm": 0.2308778464794159,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 41070
    },
    {
      "epoch": 36.38175376439327,
      "grad_norm": 0.2716803252696991,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 41075
    },
    {
      "epoch": 36.38618246235607,
      "grad_norm": 0.2282794564962387,
      "learning_rate": 1e-05,
      "loss": 1.0362,
      "step": 41080
    },
    {
      "epoch": 36.39061116031887,
      "grad_norm": 0.21496593952178955,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 41085
    },
    {
      "epoch": 36.39503985828166,
      "grad_norm": 0.2691848874092102,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 41090
    },
    {
      "epoch": 36.399468556244464,
      "grad_norm": 0.25886887311935425,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 41095
    },
    {
      "epoch": 36.403897254207266,
      "grad_norm": 0.23968742787837982,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 41100
    },
    {
      "epoch": 36.40832595217006,
      "grad_norm": 0.22284342348575592,
      "learning_rate": 1e-05,
      "loss": 1.0186,
      "step": 41105
    },
    {
      "epoch": 36.41275465013286,
      "grad_norm": 0.27596932649612427,
      "learning_rate": 1e-05,
      "loss": 1.0498,
      "step": 41110
    },
    {
      "epoch": 36.41718334809566,
      "grad_norm": 0.24692389369010925,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 41115
    },
    {
      "epoch": 36.421612046058456,
      "grad_norm": 0.274220734834671,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 41120
    },
    {
      "epoch": 36.42604074402126,
      "grad_norm": 0.222972109913826,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 41125
    },
    {
      "epoch": 36.43046944198406,
      "grad_norm": 0.20551925897598267,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 41130
    },
    {
      "epoch": 36.43489813994685,
      "grad_norm": 0.266122043132782,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 41135
    },
    {
      "epoch": 36.439326837909654,
      "grad_norm": 0.2282188981771469,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 41140
    },
    {
      "epoch": 36.443755535872455,
      "grad_norm": 0.23219631612300873,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 41145
    },
    {
      "epoch": 36.44818423383525,
      "grad_norm": 0.22225667536258698,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 41150
    },
    {
      "epoch": 36.45261293179805,
      "grad_norm": 0.2564220130443573,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 41155
    },
    {
      "epoch": 36.45704162976085,
      "grad_norm": 0.22233328223228455,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 41160
    },
    {
      "epoch": 36.461470327723646,
      "grad_norm": 0.24494224786758423,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 41165
    },
    {
      "epoch": 36.46589902568645,
      "grad_norm": 0.2292802631855011,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 41170
    },
    {
      "epoch": 36.47032772364925,
      "grad_norm": 0.2072295844554901,
      "learning_rate": 1e-05,
      "loss": 0.9134,
      "step": 41175
    },
    {
      "epoch": 36.47475642161205,
      "grad_norm": 0.2157590687274933,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 41180
    },
    {
      "epoch": 36.479185119574844,
      "grad_norm": 0.2594149112701416,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 41185
    },
    {
      "epoch": 36.483613817537645,
      "grad_norm": 0.23352308571338654,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 41190
    },
    {
      "epoch": 36.488042515500446,
      "grad_norm": 0.2546723186969757,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 41195
    },
    {
      "epoch": 36.49247121346324,
      "grad_norm": 0.24370384216308594,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 41200
    },
    {
      "epoch": 36.49689991142604,
      "grad_norm": 0.23751384019851685,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 41205
    },
    {
      "epoch": 36.50132860938884,
      "grad_norm": 0.20981431007385254,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 41210
    },
    {
      "epoch": 36.50575730735164,
      "grad_norm": 0.24499230086803436,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 41215
    },
    {
      "epoch": 36.51018600531444,
      "grad_norm": 0.19730959832668304,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 41220
    },
    {
      "epoch": 36.51461470327724,
      "grad_norm": 0.22785040736198425,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 41225
    },
    {
      "epoch": 36.51904340124003,
      "grad_norm": 0.22181697189807892,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 41230
    },
    {
      "epoch": 36.523472099202834,
      "grad_norm": 0.23306769132614136,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 41235
    },
    {
      "epoch": 36.527900797165636,
      "grad_norm": 0.2977795898914337,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 41240
    },
    {
      "epoch": 36.53232949512843,
      "grad_norm": 0.24598073959350586,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 41245
    },
    {
      "epoch": 36.53675819309123,
      "grad_norm": 0.26347070932388306,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 41250
    },
    {
      "epoch": 36.54118689105403,
      "grad_norm": 0.24939309060573578,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 41255
    },
    {
      "epoch": 36.545615589016826,
      "grad_norm": 0.3335600793361664,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 41260
    },
    {
      "epoch": 36.55004428697963,
      "grad_norm": 0.22563804686069489,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 41265
    },
    {
      "epoch": 36.55447298494243,
      "grad_norm": 0.22194381058216095,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 41270
    },
    {
      "epoch": 36.55890168290522,
      "grad_norm": 0.2519119083881378,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 41275
    },
    {
      "epoch": 36.563330380868024,
      "grad_norm": 0.24392183125019073,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 41280
    },
    {
      "epoch": 36.567759078830825,
      "grad_norm": 0.3094474971294403,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 41285
    },
    {
      "epoch": 36.57218777679362,
      "grad_norm": 0.21822114288806915,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 41290
    },
    {
      "epoch": 36.57661647475642,
      "grad_norm": 0.23856975138187408,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 41295
    },
    {
      "epoch": 36.58104517271922,
      "grad_norm": 0.2759832441806793,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 41300
    },
    {
      "epoch": 36.58547387068202,
      "grad_norm": 0.28036031126976013,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 41305
    },
    {
      "epoch": 36.58990256864482,
      "grad_norm": 0.21053095161914825,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 41310
    },
    {
      "epoch": 36.59433126660762,
      "grad_norm": 0.2949060797691345,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 41315
    },
    {
      "epoch": 36.59875996457042,
      "grad_norm": 0.2263978272676468,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 41320
    },
    {
      "epoch": 36.60318866253321,
      "grad_norm": 0.2795294225215912,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 41325
    },
    {
      "epoch": 36.607617360496015,
      "grad_norm": 0.2060815691947937,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 41330
    },
    {
      "epoch": 36.612046058458816,
      "grad_norm": 0.2050504982471466,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 41335
    },
    {
      "epoch": 36.61647475642161,
      "grad_norm": 0.2511064410209656,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 41340
    },
    {
      "epoch": 36.62090345438441,
      "grad_norm": 0.26368412375450134,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 41345
    },
    {
      "epoch": 36.62533215234721,
      "grad_norm": 0.21881631016731262,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 41350
    },
    {
      "epoch": 36.62976085031001,
      "grad_norm": 0.24021776020526886,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 41355
    },
    {
      "epoch": 36.63418954827281,
      "grad_norm": 0.26761144399642944,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 41360
    },
    {
      "epoch": 36.63861824623561,
      "grad_norm": 0.3264281451702118,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 41365
    },
    {
      "epoch": 36.6430469441984,
      "grad_norm": 0.21894054114818573,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 41370
    },
    {
      "epoch": 36.647475642161204,
      "grad_norm": 0.24946807324886322,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 41375
    },
    {
      "epoch": 36.651904340124005,
      "grad_norm": 0.22466206550598145,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 41380
    },
    {
      "epoch": 36.6563330380868,
      "grad_norm": 0.2501908242702484,
      "learning_rate": 1e-05,
      "loss": 1.0323,
      "step": 41385
    },
    {
      "epoch": 36.6607617360496,
      "grad_norm": 0.2907739281654358,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 41390
    },
    {
      "epoch": 36.6651904340124,
      "grad_norm": 0.2530531883239746,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 41395
    },
    {
      "epoch": 36.669619131975196,
      "grad_norm": 0.27774450182914734,
      "learning_rate": 1e-05,
      "loss": 1.0385,
      "step": 41400
    },
    {
      "epoch": 36.674047829938,
      "grad_norm": 0.2448064535856247,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 41405
    },
    {
      "epoch": 36.6784765279008,
      "grad_norm": 0.2538725733757019,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 41410
    },
    {
      "epoch": 36.68290522586359,
      "grad_norm": 0.2417149692773819,
      "learning_rate": 1e-05,
      "loss": 0.9131,
      "step": 41415
    },
    {
      "epoch": 36.687333923826394,
      "grad_norm": 0.22603629529476166,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 41420
    },
    {
      "epoch": 36.691762621789195,
      "grad_norm": 0.2294352948665619,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 41425
    },
    {
      "epoch": 36.696191319751996,
      "grad_norm": 0.24379414319992065,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 41430
    },
    {
      "epoch": 36.70062001771479,
      "grad_norm": 0.24497553706169128,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 41435
    },
    {
      "epoch": 36.70504871567759,
      "grad_norm": 0.2463923841714859,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 41440
    },
    {
      "epoch": 36.70947741364039,
      "grad_norm": 0.2909233868122101,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 41445
    },
    {
      "epoch": 36.71390611160319,
      "grad_norm": 0.24570921063423157,
      "learning_rate": 1e-05,
      "loss": 1.0233,
      "step": 41450
    },
    {
      "epoch": 36.71833480956599,
      "grad_norm": 0.23515303432941437,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 41455
    },
    {
      "epoch": 36.72276350752879,
      "grad_norm": 0.28930291533470154,
      "learning_rate": 1e-05,
      "loss": 1.0255,
      "step": 41460
    },
    {
      "epoch": 36.72719220549158,
      "grad_norm": 0.2584001421928406,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 41465
    },
    {
      "epoch": 36.731620903454385,
      "grad_norm": 0.22280341386795044,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 41470
    },
    {
      "epoch": 36.736049601417186,
      "grad_norm": 0.250087708234787,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 41475
    },
    {
      "epoch": 36.74047829937998,
      "grad_norm": 0.255700945854187,
      "learning_rate": 1e-05,
      "loss": 1.0305,
      "step": 41480
    },
    {
      "epoch": 36.74490699734278,
      "grad_norm": 0.23839564621448517,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 41485
    },
    {
      "epoch": 36.74933569530558,
      "grad_norm": 0.26836100220680237,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 41490
    },
    {
      "epoch": 36.753764393268376,
      "grad_norm": 0.21705758571624756,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 41495
    },
    {
      "epoch": 36.75819309123118,
      "grad_norm": 0.204350084066391,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 41500
    },
    {
      "epoch": 36.76262178919398,
      "grad_norm": 0.2528984546661377,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 41505
    },
    {
      "epoch": 36.76705048715677,
      "grad_norm": 0.19084975123405457,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 41510
    },
    {
      "epoch": 36.771479185119574,
      "grad_norm": 0.2052205353975296,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 41515
    },
    {
      "epoch": 36.775907883082375,
      "grad_norm": 0.24439504742622375,
      "learning_rate": 1e-05,
      "loss": 1.039,
      "step": 41520
    },
    {
      "epoch": 36.78033658104517,
      "grad_norm": 0.25444209575653076,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 41525
    },
    {
      "epoch": 36.78476527900797,
      "grad_norm": 0.23856742680072784,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 41530
    },
    {
      "epoch": 36.78919397697077,
      "grad_norm": 0.2234344482421875,
      "learning_rate": 1e-05,
      "loss": 1.027,
      "step": 41535
    },
    {
      "epoch": 36.79362267493357,
      "grad_norm": 0.270016074180603,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 41540
    },
    {
      "epoch": 36.79805137289637,
      "grad_norm": 0.2673543393611908,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 41545
    },
    {
      "epoch": 36.80248007085917,
      "grad_norm": 0.24772733449935913,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 41550
    },
    {
      "epoch": 36.80690876882197,
      "grad_norm": 0.23683005571365356,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 41555
    },
    {
      "epoch": 36.811337466784764,
      "grad_norm": 0.23282134532928467,
      "learning_rate": 1e-05,
      "loss": 0.9265,
      "step": 41560
    },
    {
      "epoch": 36.815766164747565,
      "grad_norm": 0.21743695437908173,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 41565
    },
    {
      "epoch": 36.820194862710366,
      "grad_norm": 0.2388668954372406,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 41570
    },
    {
      "epoch": 36.82462356067316,
      "grad_norm": 0.24171118438243866,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 41575
    },
    {
      "epoch": 36.82905225863596,
      "grad_norm": 0.20600350201129913,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 41580
    },
    {
      "epoch": 36.83348095659876,
      "grad_norm": 0.22473964095115662,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 41585
    },
    {
      "epoch": 36.83790965456156,
      "grad_norm": 0.2945292592048645,
      "learning_rate": 1e-05,
      "loss": 1.0587,
      "step": 41590
    },
    {
      "epoch": 36.84233835252436,
      "grad_norm": 0.21868018805980682,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 41595
    },
    {
      "epoch": 36.84676705048716,
      "grad_norm": 0.21495646238327026,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 41600
    },
    {
      "epoch": 36.85119574844995,
      "grad_norm": 0.235142320394516,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 41605
    },
    {
      "epoch": 36.855624446412754,
      "grad_norm": 0.2598854601383209,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 41610
    },
    {
      "epoch": 36.860053144375556,
      "grad_norm": 0.2535991370677948,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 41615
    },
    {
      "epoch": 36.86448184233835,
      "grad_norm": 0.22093145549297333,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 41620
    },
    {
      "epoch": 36.86891054030115,
      "grad_norm": 0.22329245507717133,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 41625
    },
    {
      "epoch": 36.87333923826395,
      "grad_norm": 0.24171333014965057,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 41630
    },
    {
      "epoch": 36.877767936226746,
      "grad_norm": 0.23357275128364563,
      "learning_rate": 1e-05,
      "loss": 1.0445,
      "step": 41635
    },
    {
      "epoch": 36.88219663418955,
      "grad_norm": 0.2335280030965805,
      "learning_rate": 1e-05,
      "loss": 0.9161,
      "step": 41640
    },
    {
      "epoch": 36.88662533215235,
      "grad_norm": 0.2635389566421509,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 41645
    },
    {
      "epoch": 36.89105403011514,
      "grad_norm": 0.21385855972766876,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 41650
    },
    {
      "epoch": 36.895482728077944,
      "grad_norm": 0.22912786900997162,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 41655
    },
    {
      "epoch": 36.899911426040745,
      "grad_norm": 0.25736311078071594,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 41660
    },
    {
      "epoch": 36.90434012400354,
      "grad_norm": 0.21302156150341034,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 41665
    },
    {
      "epoch": 36.90876882196634,
      "grad_norm": 0.2374745011329651,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 41670
    },
    {
      "epoch": 36.91319751992914,
      "grad_norm": 0.23203137516975403,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 41675
    },
    {
      "epoch": 36.91762621789194,
      "grad_norm": 0.21209439635276794,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 41680
    },
    {
      "epoch": 36.92205491585474,
      "grad_norm": 0.23146556317806244,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 41685
    },
    {
      "epoch": 36.92648361381754,
      "grad_norm": 0.2478378564119339,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 41690
    },
    {
      "epoch": 36.93091231178034,
      "grad_norm": 0.20333725214004517,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 41695
    },
    {
      "epoch": 36.935341009743134,
      "grad_norm": 0.2504282295703888,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 41700
    },
    {
      "epoch": 36.939769707705935,
      "grad_norm": 0.20910699665546417,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 41705
    },
    {
      "epoch": 36.944198405668736,
      "grad_norm": 0.23719508945941925,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 41710
    },
    {
      "epoch": 36.94862710363153,
      "grad_norm": 0.26588308811187744,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 41715
    },
    {
      "epoch": 36.95305580159433,
      "grad_norm": 0.24046368896961212,
      "learning_rate": 1e-05,
      "loss": 1.0432,
      "step": 41720
    },
    {
      "epoch": 36.95748449955713,
      "grad_norm": 0.2910871207714081,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 41725
    },
    {
      "epoch": 36.96191319751993,
      "grad_norm": 0.2554914355278015,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 41730
    },
    {
      "epoch": 36.96634189548273,
      "grad_norm": 0.22151389718055725,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 41735
    },
    {
      "epoch": 36.97077059344553,
      "grad_norm": 0.23031800985336304,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 41740
    },
    {
      "epoch": 36.97519929140832,
      "grad_norm": 0.24828629195690155,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 41745
    },
    {
      "epoch": 36.979627989371124,
      "grad_norm": 0.20596984028816223,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 41750
    },
    {
      "epoch": 36.984056687333926,
      "grad_norm": 0.23722344636917114,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 41755
    },
    {
      "epoch": 36.98848538529672,
      "grad_norm": 0.21555255353450775,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 41760
    },
    {
      "epoch": 36.99291408325952,
      "grad_norm": 0.2219674289226532,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 41765
    },
    {
      "epoch": 36.99734278122232,
      "grad_norm": 0.28827813267707825,
      "learning_rate": 1e-05,
      "loss": 1.0603,
      "step": 41770
    },
    {
      "epoch": 37.001771479185116,
      "grad_norm": 0.2842288911342621,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 41775
    },
    {
      "epoch": 37.00620017714792,
      "grad_norm": 0.29359903931617737,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 41780
    },
    {
      "epoch": 37.01062887511072,
      "grad_norm": 0.23733164370059967,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 41785
    },
    {
      "epoch": 37.01505757307352,
      "grad_norm": 0.23545072972774506,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 41790
    },
    {
      "epoch": 37.019486271036314,
      "grad_norm": 0.2726118564605713,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 41795
    },
    {
      "epoch": 37.023914968999115,
      "grad_norm": 0.207491934299469,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 41800
    },
    {
      "epoch": 37.028343666961916,
      "grad_norm": 0.3502281606197357,
      "learning_rate": 1e-05,
      "loss": 0.936,
      "step": 41805
    },
    {
      "epoch": 37.03277236492471,
      "grad_norm": 0.21178008615970612,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 41810
    },
    {
      "epoch": 37.03720106288751,
      "grad_norm": 0.2580762505531311,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 41815
    },
    {
      "epoch": 37.04162976085031,
      "grad_norm": 0.23341518640518188,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 41820
    },
    {
      "epoch": 37.04605845881311,
      "grad_norm": 0.2185945212841034,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 41825
    },
    {
      "epoch": 37.05048715677591,
      "grad_norm": 0.25499385595321655,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 41830
    },
    {
      "epoch": 37.05491585473871,
      "grad_norm": 0.23552343249320984,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 41835
    },
    {
      "epoch": 37.0593445527015,
      "grad_norm": 0.22409363090991974,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 41840
    },
    {
      "epoch": 37.063773250664305,
      "grad_norm": 0.27179014682769775,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 41845
    },
    {
      "epoch": 37.068201948627106,
      "grad_norm": 0.23248104751110077,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 41850
    },
    {
      "epoch": 37.0726306465899,
      "grad_norm": 0.227968230843544,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 41855
    },
    {
      "epoch": 37.0770593445527,
      "grad_norm": 0.2531631886959076,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 41860
    },
    {
      "epoch": 37.0814880425155,
      "grad_norm": 0.26105639338493347,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 41865
    },
    {
      "epoch": 37.0859167404783,
      "grad_norm": 0.22814108431339264,
      "learning_rate": 1e-05,
      "loss": 1.0374,
      "step": 41870
    },
    {
      "epoch": 37.0903454384411,
      "grad_norm": 0.26840445399284363,
      "learning_rate": 1e-05,
      "loss": 0.923,
      "step": 41875
    },
    {
      "epoch": 37.0947741364039,
      "grad_norm": 0.2333166003227234,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 41880
    },
    {
      "epoch": 37.09920283436669,
      "grad_norm": 0.28321367502212524,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 41885
    },
    {
      "epoch": 37.103631532329494,
      "grad_norm": 0.23226363956928253,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 41890
    },
    {
      "epoch": 37.108060230292296,
      "grad_norm": 0.2403482049703598,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 41895
    },
    {
      "epoch": 37.11248892825509,
      "grad_norm": 0.26225462555885315,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 41900
    },
    {
      "epoch": 37.11691762621789,
      "grad_norm": 0.245097354054451,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 41905
    },
    {
      "epoch": 37.12134632418069,
      "grad_norm": 0.23149815201759338,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 41910
    },
    {
      "epoch": 37.12577502214349,
      "grad_norm": 0.25361597537994385,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 41915
    },
    {
      "epoch": 37.13020372010629,
      "grad_norm": 0.27507415413856506,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 41920
    },
    {
      "epoch": 37.13463241806909,
      "grad_norm": 0.23268146812915802,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 41925
    },
    {
      "epoch": 37.13906111603189,
      "grad_norm": 0.22741036117076874,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 41930
    },
    {
      "epoch": 37.143489813994684,
      "grad_norm": 0.2483917474746704,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 41935
    },
    {
      "epoch": 37.147918511957485,
      "grad_norm": 0.2293664962053299,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 41940
    },
    {
      "epoch": 37.152347209920286,
      "grad_norm": 0.24596326053142548,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 41945
    },
    {
      "epoch": 37.15677590788308,
      "grad_norm": 0.25122880935668945,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 41950
    },
    {
      "epoch": 37.16120460584588,
      "grad_norm": 0.25325894355773926,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 41955
    },
    {
      "epoch": 37.16563330380868,
      "grad_norm": 0.22633004188537598,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 41960
    },
    {
      "epoch": 37.17006200177148,
      "grad_norm": 0.25204646587371826,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 41965
    },
    {
      "epoch": 37.17449069973428,
      "grad_norm": 0.20441824197769165,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 41970
    },
    {
      "epoch": 37.17891939769708,
      "grad_norm": 0.2294832319021225,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 41975
    },
    {
      "epoch": 37.18334809565987,
      "grad_norm": 0.231958270072937,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 41980
    },
    {
      "epoch": 37.187776793622675,
      "grad_norm": 0.22556857764720917,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 41985
    },
    {
      "epoch": 37.192205491585476,
      "grad_norm": 0.25151780247688293,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 41990
    },
    {
      "epoch": 37.19663418954827,
      "grad_norm": 0.2676210403442383,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 41995
    },
    {
      "epoch": 37.20106288751107,
      "grad_norm": 0.23232288658618927,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 42000
    },
    {
      "epoch": 37.20549158547387,
      "grad_norm": 0.2164548635482788,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 42005
    },
    {
      "epoch": 37.20992028343667,
      "grad_norm": 0.2006070911884308,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 42010
    },
    {
      "epoch": 37.21434898139947,
      "grad_norm": 0.22555896639823914,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 42015
    },
    {
      "epoch": 37.21877767936227,
      "grad_norm": 0.24526165425777435,
      "learning_rate": 1e-05,
      "loss": 1.0351,
      "step": 42020
    },
    {
      "epoch": 37.22320637732506,
      "grad_norm": 0.22633777558803558,
      "learning_rate": 1e-05,
      "loss": 1.0161,
      "step": 42025
    },
    {
      "epoch": 37.227635075287864,
      "grad_norm": 0.2592720091342926,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 42030
    },
    {
      "epoch": 37.232063773250665,
      "grad_norm": 0.24568088352680206,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 42035
    },
    {
      "epoch": 37.23649247121347,
      "grad_norm": 0.23025763034820557,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 42040
    },
    {
      "epoch": 37.24092116917626,
      "grad_norm": 0.224689781665802,
      "learning_rate": 1e-05,
      "loss": 0.9293,
      "step": 42045
    },
    {
      "epoch": 37.24534986713906,
      "grad_norm": 0.21344514191150665,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 42050
    },
    {
      "epoch": 37.24977856510186,
      "grad_norm": 0.23611518740653992,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 42055
    },
    {
      "epoch": 37.25420726306466,
      "grad_norm": 0.22356264293193817,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 42060
    },
    {
      "epoch": 37.25863596102746,
      "grad_norm": 0.26769402623176575,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 42065
    },
    {
      "epoch": 37.26306465899026,
      "grad_norm": 0.26295900344848633,
      "learning_rate": 1e-05,
      "loss": 1.0254,
      "step": 42070
    },
    {
      "epoch": 37.267493356953054,
      "grad_norm": 0.25025105476379395,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 42075
    },
    {
      "epoch": 37.271922054915855,
      "grad_norm": 0.2221536487340927,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 42080
    },
    {
      "epoch": 37.276350752878656,
      "grad_norm": 0.25521183013916016,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 42085
    },
    {
      "epoch": 37.28077945084145,
      "grad_norm": 0.24157032370567322,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 42090
    },
    {
      "epoch": 37.28520814880425,
      "grad_norm": 0.25145530700683594,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 42095
    },
    {
      "epoch": 37.28963684676705,
      "grad_norm": 0.23742009699344635,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 42100
    },
    {
      "epoch": 37.29406554472985,
      "grad_norm": 0.2109355926513672,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 42105
    },
    {
      "epoch": 37.29849424269265,
      "grad_norm": 0.22865231335163116,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 42110
    },
    {
      "epoch": 37.30292294065545,
      "grad_norm": 0.23039765655994415,
      "learning_rate": 1e-05,
      "loss": 1.017,
      "step": 42115
    },
    {
      "epoch": 37.30735163861824,
      "grad_norm": 0.23206283152103424,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 42120
    },
    {
      "epoch": 37.311780336581045,
      "grad_norm": 0.20329523086547852,
      "learning_rate": 1e-05,
      "loss": 1.0216,
      "step": 42125
    },
    {
      "epoch": 37.316209034543846,
      "grad_norm": 0.2435583621263504,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 42130
    },
    {
      "epoch": 37.32063773250664,
      "grad_norm": 0.21034710109233856,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 42135
    },
    {
      "epoch": 37.32506643046944,
      "grad_norm": 0.22206032276153564,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 42140
    },
    {
      "epoch": 37.32949512843224,
      "grad_norm": 0.24671857059001923,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 42145
    },
    {
      "epoch": 37.333923826395036,
      "grad_norm": 0.2695288062095642,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 42150
    },
    {
      "epoch": 37.33835252435784,
      "grad_norm": 0.2319755107164383,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 42155
    },
    {
      "epoch": 37.34278122232064,
      "grad_norm": 0.27511054277420044,
      "learning_rate": 1e-05,
      "loss": 1.1001,
      "step": 42160
    },
    {
      "epoch": 37.34720992028344,
      "grad_norm": 0.214742511510849,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 42165
    },
    {
      "epoch": 37.351638618246234,
      "grad_norm": 0.24615897238254547,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 42170
    },
    {
      "epoch": 37.356067316209035,
      "grad_norm": 0.31881681084632874,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 42175
    },
    {
      "epoch": 37.36049601417184,
      "grad_norm": 0.2792544364929199,
      "learning_rate": 1e-05,
      "loss": 1.0591,
      "step": 42180
    },
    {
      "epoch": 37.36492471213463,
      "grad_norm": 0.2725650370121002,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 42185
    },
    {
      "epoch": 37.36935341009743,
      "grad_norm": 0.2541310489177704,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 42190
    },
    {
      "epoch": 37.37378210806023,
      "grad_norm": 0.2675561010837555,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 42195
    },
    {
      "epoch": 37.37821080602303,
      "grad_norm": 0.21635383367538452,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 42200
    },
    {
      "epoch": 37.38263950398583,
      "grad_norm": 0.2331739217042923,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 42205
    },
    {
      "epoch": 37.38706820194863,
      "grad_norm": 0.302857369184494,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 42210
    },
    {
      "epoch": 37.391496899911424,
      "grad_norm": 0.2117435783147812,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 42215
    },
    {
      "epoch": 37.395925597874225,
      "grad_norm": 0.23086801171302795,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 42220
    },
    {
      "epoch": 37.400354295837026,
      "grad_norm": 0.23856469988822937,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 42225
    },
    {
      "epoch": 37.40478299379982,
      "grad_norm": 0.2832426428794861,
      "learning_rate": 1e-05,
      "loss": 1.0441,
      "step": 42230
    },
    {
      "epoch": 37.40921169176262,
      "grad_norm": 0.22969260811805725,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 42235
    },
    {
      "epoch": 37.41364038972542,
      "grad_norm": 0.20280721783638,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 42240
    },
    {
      "epoch": 37.41806908768822,
      "grad_norm": 0.23133990168571472,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 42245
    },
    {
      "epoch": 37.42249778565102,
      "grad_norm": 0.2690482437610626,
      "learning_rate": 1e-05,
      "loss": 1.0172,
      "step": 42250
    },
    {
      "epoch": 37.42692648361382,
      "grad_norm": 0.20983082056045532,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 42255
    },
    {
      "epoch": 37.43135518157661,
      "grad_norm": 0.2177984118461609,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 42260
    },
    {
      "epoch": 37.435783879539414,
      "grad_norm": 0.22971859574317932,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 42265
    },
    {
      "epoch": 37.440212577502216,
      "grad_norm": 0.22158464789390564,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 42270
    },
    {
      "epoch": 37.44464127546502,
      "grad_norm": 0.24342280626296997,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 42275
    },
    {
      "epoch": 37.44906997342781,
      "grad_norm": 0.23846440017223358,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 42280
    },
    {
      "epoch": 37.45349867139061,
      "grad_norm": 0.219167560338974,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 42285
    },
    {
      "epoch": 37.45792736935341,
      "grad_norm": 0.23654691874980927,
      "learning_rate": 1e-05,
      "loss": 0.9252,
      "step": 42290
    },
    {
      "epoch": 37.46235606731621,
      "grad_norm": 0.2326965034008026,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 42295
    },
    {
      "epoch": 37.46678476527901,
      "grad_norm": 0.25994324684143066,
      "learning_rate": 1e-05,
      "loss": 1.0317,
      "step": 42300
    },
    {
      "epoch": 37.47121346324181,
      "grad_norm": 0.21999827027320862,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 42305
    },
    {
      "epoch": 37.475642161204604,
      "grad_norm": 0.2213774174451828,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 42310
    },
    {
      "epoch": 37.480070859167405,
      "grad_norm": 0.24380311369895935,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 42315
    },
    {
      "epoch": 37.484499557130206,
      "grad_norm": 0.24933016300201416,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 42320
    },
    {
      "epoch": 37.488928255093,
      "grad_norm": 0.28443336486816406,
      "learning_rate": 1e-05,
      "loss": 1.0276,
      "step": 42325
    },
    {
      "epoch": 37.4933569530558,
      "grad_norm": 0.2330223023891449,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 42330
    },
    {
      "epoch": 37.4977856510186,
      "grad_norm": 0.22605866193771362,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 42335
    },
    {
      "epoch": 37.5022143489814,
      "grad_norm": 0.23135940730571747,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 42340
    },
    {
      "epoch": 37.5066430469442,
      "grad_norm": 0.2751252353191376,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 42345
    },
    {
      "epoch": 37.511071744907,
      "grad_norm": 0.2597109079360962,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 42350
    },
    {
      "epoch": 37.515500442869794,
      "grad_norm": 0.23957784473896027,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 42355
    },
    {
      "epoch": 37.519929140832595,
      "grad_norm": 0.22048811614513397,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 42360
    },
    {
      "epoch": 37.524357838795396,
      "grad_norm": 0.22523728013038635,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 42365
    },
    {
      "epoch": 37.52878653675819,
      "grad_norm": 0.24047182500362396,
      "learning_rate": 1e-05,
      "loss": 1.0356,
      "step": 42370
    },
    {
      "epoch": 37.53321523472099,
      "grad_norm": 0.23936474323272705,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 42375
    },
    {
      "epoch": 37.53764393268379,
      "grad_norm": 0.23928560316562653,
      "learning_rate": 1e-05,
      "loss": 1.0217,
      "step": 42380
    },
    {
      "epoch": 37.54207263064659,
      "grad_norm": 0.3341608941555023,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 42385
    },
    {
      "epoch": 37.54650132860939,
      "grad_norm": 0.24284934997558594,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 42390
    },
    {
      "epoch": 37.55093002657219,
      "grad_norm": 0.26976075768470764,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 42395
    },
    {
      "epoch": 37.55535872453498,
      "grad_norm": 0.2603054642677307,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 42400
    },
    {
      "epoch": 37.559787422497784,
      "grad_norm": 0.2568533718585968,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 42405
    },
    {
      "epoch": 37.564216120460586,
      "grad_norm": 0.26029160618782043,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 42410
    },
    {
      "epoch": 37.56864481842339,
      "grad_norm": 0.24650785326957703,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 42415
    },
    {
      "epoch": 37.57307351638618,
      "grad_norm": 0.2329334318637848,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 42420
    },
    {
      "epoch": 37.57750221434898,
      "grad_norm": 0.26726096868515015,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 42425
    },
    {
      "epoch": 37.58193091231178,
      "grad_norm": 0.24087293446063995,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 42430
    },
    {
      "epoch": 37.58635961027458,
      "grad_norm": 0.30094900727272034,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 42435
    },
    {
      "epoch": 37.59078830823738,
      "grad_norm": 0.2575588524341583,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 42440
    },
    {
      "epoch": 37.59521700620018,
      "grad_norm": 0.20232011377811432,
      "learning_rate": 1e-05,
      "loss": 0.9031,
      "step": 42445
    },
    {
      "epoch": 37.599645704162974,
      "grad_norm": 0.21574756503105164,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 42450
    },
    {
      "epoch": 37.604074402125775,
      "grad_norm": 0.2227298468351364,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 42455
    },
    {
      "epoch": 37.608503100088576,
      "grad_norm": 0.2319817841053009,
      "learning_rate": 1e-05,
      "loss": 1.0514,
      "step": 42460
    },
    {
      "epoch": 37.61293179805137,
      "grad_norm": 0.22359859943389893,
      "learning_rate": 1e-05,
      "loss": 1.0336,
      "step": 42465
    },
    {
      "epoch": 37.61736049601417,
      "grad_norm": 0.24118474125862122,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 42470
    },
    {
      "epoch": 37.62178919397697,
      "grad_norm": 0.2634837329387665,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 42475
    },
    {
      "epoch": 37.62621789193977,
      "grad_norm": 0.23753149807453156,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 42480
    },
    {
      "epoch": 37.63064658990257,
      "grad_norm": 0.22305628657341003,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 42485
    },
    {
      "epoch": 37.63507528786537,
      "grad_norm": 0.21795621514320374,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 42490
    },
    {
      "epoch": 37.63950398582816,
      "grad_norm": 0.22745104134082794,
      "learning_rate": 1e-05,
      "loss": 1.0455,
      "step": 42495
    },
    {
      "epoch": 37.643932683790965,
      "grad_norm": 0.2625316083431244,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 42500
    },
    {
      "epoch": 37.648361381753766,
      "grad_norm": 0.26233750581741333,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 42505
    },
    {
      "epoch": 37.65279007971656,
      "grad_norm": 0.2161959856748581,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 42510
    },
    {
      "epoch": 37.65721877767936,
      "grad_norm": 0.19854971766471863,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 42515
    },
    {
      "epoch": 37.66164747564216,
      "grad_norm": 0.21185769140720367,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 42520
    },
    {
      "epoch": 37.666076173604964,
      "grad_norm": 0.23087051510810852,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 42525
    },
    {
      "epoch": 37.67050487156776,
      "grad_norm": 0.223099946975708,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 42530
    },
    {
      "epoch": 37.67493356953056,
      "grad_norm": 0.23845601081848145,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 42535
    },
    {
      "epoch": 37.67936226749336,
      "grad_norm": 0.23099152743816376,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 42540
    },
    {
      "epoch": 37.683790965456154,
      "grad_norm": 0.23859800398349762,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 42545
    },
    {
      "epoch": 37.688219663418955,
      "grad_norm": 0.23860591650009155,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 42550
    },
    {
      "epoch": 37.69264836138176,
      "grad_norm": 0.23042136430740356,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 42555
    },
    {
      "epoch": 37.69707705934455,
      "grad_norm": 0.23309612274169922,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 42560
    },
    {
      "epoch": 37.70150575730735,
      "grad_norm": 0.2884846031665802,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 42565
    },
    {
      "epoch": 37.70593445527015,
      "grad_norm": 0.2549743354320526,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 42570
    },
    {
      "epoch": 37.71036315323295,
      "grad_norm": 0.26618364453315735,
      "learning_rate": 1e-05,
      "loss": 1.0189,
      "step": 42575
    },
    {
      "epoch": 37.71479185119575,
      "grad_norm": 0.2884793281555176,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 42580
    },
    {
      "epoch": 37.71922054915855,
      "grad_norm": 0.2261142134666443,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 42585
    },
    {
      "epoch": 37.723649247121344,
      "grad_norm": 0.2919895350933075,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 42590
    },
    {
      "epoch": 37.728077945084145,
      "grad_norm": 0.24840529263019562,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 42595
    },
    {
      "epoch": 37.732506643046946,
      "grad_norm": 0.25027358531951904,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 42600
    },
    {
      "epoch": 37.73693534100974,
      "grad_norm": 0.21553629636764526,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 42605
    },
    {
      "epoch": 37.74136403897254,
      "grad_norm": 0.2082803100347519,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 42610
    },
    {
      "epoch": 37.74579273693534,
      "grad_norm": 0.23363971710205078,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 42615
    },
    {
      "epoch": 37.75022143489814,
      "grad_norm": 0.23775963485240936,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 42620
    },
    {
      "epoch": 37.75465013286094,
      "grad_norm": 0.23447728157043457,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 42625
    },
    {
      "epoch": 37.75907883082374,
      "grad_norm": 0.2571112811565399,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 42630
    },
    {
      "epoch": 37.76350752878653,
      "grad_norm": 0.26900970935821533,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 42635
    },
    {
      "epoch": 37.767936226749335,
      "grad_norm": 0.2583164870738983,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 42640
    },
    {
      "epoch": 37.772364924712136,
      "grad_norm": 0.22950932383537292,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 42645
    },
    {
      "epoch": 37.77679362267494,
      "grad_norm": 0.29403239488601685,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 42650
    },
    {
      "epoch": 37.78122232063773,
      "grad_norm": 0.2755625247955322,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 42655
    },
    {
      "epoch": 37.78565101860053,
      "grad_norm": 0.2673417329788208,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 42660
    },
    {
      "epoch": 37.79007971656333,
      "grad_norm": 0.23710231482982635,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 42665
    },
    {
      "epoch": 37.79450841452613,
      "grad_norm": 0.2526950538158417,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 42670
    },
    {
      "epoch": 37.79893711248893,
      "grad_norm": 0.23061133921146393,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 42675
    },
    {
      "epoch": 37.80336581045173,
      "grad_norm": 0.25248298048973083,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 42680
    },
    {
      "epoch": 37.807794508414524,
      "grad_norm": 0.2378055304288864,
      "learning_rate": 1e-05,
      "loss": 1.0148,
      "step": 42685
    },
    {
      "epoch": 37.812223206377325,
      "grad_norm": 0.23923656344413757,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 42690
    },
    {
      "epoch": 37.81665190434013,
      "grad_norm": 0.22410640120506287,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 42695
    },
    {
      "epoch": 37.82108060230292,
      "grad_norm": 0.2792249917984009,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 42700
    },
    {
      "epoch": 37.82550930026572,
      "grad_norm": 0.2228323072195053,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 42705
    },
    {
      "epoch": 37.82993799822852,
      "grad_norm": 0.24203184247016907,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 42710
    },
    {
      "epoch": 37.83436669619132,
      "grad_norm": 0.25683319568634033,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 42715
    },
    {
      "epoch": 37.83879539415412,
      "grad_norm": 0.257548451423645,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 42720
    },
    {
      "epoch": 37.84322409211692,
      "grad_norm": 0.2615880072116852,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 42725
    },
    {
      "epoch": 37.847652790079714,
      "grad_norm": 0.2446560561656952,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 42730
    },
    {
      "epoch": 37.852081488042515,
      "grad_norm": 0.28907597064971924,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 42735
    },
    {
      "epoch": 37.856510186005316,
      "grad_norm": 0.25594547390937805,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 42740
    },
    {
      "epoch": 37.86093888396811,
      "grad_norm": 0.23299019038677216,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 42745
    },
    {
      "epoch": 37.86536758193091,
      "grad_norm": 0.2175140678882599,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 42750
    },
    {
      "epoch": 37.86979627989371,
      "grad_norm": 0.2387598156929016,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 42755
    },
    {
      "epoch": 37.87422497785651,
      "grad_norm": 0.21592430770397186,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 42760
    },
    {
      "epoch": 37.87865367581931,
      "grad_norm": 0.21279841661453247,
      "learning_rate": 1e-05,
      "loss": 1.0434,
      "step": 42765
    },
    {
      "epoch": 37.88308237378211,
      "grad_norm": 0.29665660858154297,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 42770
    },
    {
      "epoch": 37.88751107174491,
      "grad_norm": 0.2409961223602295,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 42775
    },
    {
      "epoch": 37.891939769707704,
      "grad_norm": 0.23460136353969574,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 42780
    },
    {
      "epoch": 37.896368467670506,
      "grad_norm": 0.25908422470092773,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 42785
    },
    {
      "epoch": 37.90079716563331,
      "grad_norm": 0.26455163955688477,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 42790
    },
    {
      "epoch": 37.9052258635961,
      "grad_norm": 0.26139652729034424,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 42795
    },
    {
      "epoch": 37.9096545615589,
      "grad_norm": 0.26581835746765137,
      "learning_rate": 1e-05,
      "loss": 1.0417,
      "step": 42800
    },
    {
      "epoch": 37.9140832595217,
      "grad_norm": 0.29815199971199036,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 42805
    },
    {
      "epoch": 37.9185119574845,
      "grad_norm": 0.27939558029174805,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 42810
    },
    {
      "epoch": 37.9229406554473,
      "grad_norm": 0.21969373524188995,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 42815
    },
    {
      "epoch": 37.9273693534101,
      "grad_norm": 0.25108328461647034,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 42820
    },
    {
      "epoch": 37.931798051372894,
      "grad_norm": 0.23512962460517883,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 42825
    },
    {
      "epoch": 37.936226749335695,
      "grad_norm": 0.24433231353759766,
      "learning_rate": 1e-05,
      "loss": 0.9271,
      "step": 42830
    },
    {
      "epoch": 37.9406554472985,
      "grad_norm": 0.22840668261051178,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 42835
    },
    {
      "epoch": 37.94508414526129,
      "grad_norm": 0.2117973417043686,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 42840
    },
    {
      "epoch": 37.94951284322409,
      "grad_norm": 0.2078549861907959,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 42845
    },
    {
      "epoch": 37.95394154118689,
      "grad_norm": 0.2514343559741974,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 42850
    },
    {
      "epoch": 37.95837023914969,
      "grad_norm": 0.262573778629303,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 42855
    },
    {
      "epoch": 37.96279893711249,
      "grad_norm": 0.22597801685333252,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 42860
    },
    {
      "epoch": 37.96722763507529,
      "grad_norm": 0.23276439309120178,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 42865
    },
    {
      "epoch": 37.971656333038084,
      "grad_norm": 0.25960466265678406,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 42870
    },
    {
      "epoch": 37.976085031000885,
      "grad_norm": 0.25331589579582214,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 42875
    },
    {
      "epoch": 37.980513728963686,
      "grad_norm": 0.2831728160381317,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 42880
    },
    {
      "epoch": 37.98494242692648,
      "grad_norm": 0.24022911489009857,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 42885
    },
    {
      "epoch": 37.98937112488928,
      "grad_norm": 0.2533038854598999,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 42890
    },
    {
      "epoch": 37.99379982285208,
      "grad_norm": 0.20172680914402008,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 42895
    },
    {
      "epoch": 37.998228520814884,
      "grad_norm": 0.2641580104827881,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 42900
    },
    {
      "epoch": 38.00265721877768,
      "grad_norm": 0.2513423562049866,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 42905
    },
    {
      "epoch": 38.00708591674048,
      "grad_norm": 0.24848389625549316,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 42910
    },
    {
      "epoch": 38.01151461470328,
      "grad_norm": 0.2690280079841614,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 42915
    },
    {
      "epoch": 38.015943312666074,
      "grad_norm": 0.2773149609565735,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 42920
    },
    {
      "epoch": 38.020372010628876,
      "grad_norm": 0.2997073531150818,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 42925
    },
    {
      "epoch": 38.02480070859168,
      "grad_norm": 0.259006142616272,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 42930
    },
    {
      "epoch": 38.02922940655447,
      "grad_norm": 0.27275604009628296,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 42935
    },
    {
      "epoch": 38.03365810451727,
      "grad_norm": 0.27408191561698914,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 42940
    },
    {
      "epoch": 38.03808680248007,
      "grad_norm": 0.26808083057403564,
      "learning_rate": 1e-05,
      "loss": 1.0501,
      "step": 42945
    },
    {
      "epoch": 38.04251550044287,
      "grad_norm": 0.28611332178115845,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 42950
    },
    {
      "epoch": 38.04694419840567,
      "grad_norm": 0.25341981649398804,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 42955
    },
    {
      "epoch": 38.05137289636847,
      "grad_norm": 0.26704496145248413,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 42960
    },
    {
      "epoch": 38.055801594331264,
      "grad_norm": 0.25065797567367554,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 42965
    },
    {
      "epoch": 38.060230292294065,
      "grad_norm": 0.2529211938381195,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 42970
    },
    {
      "epoch": 38.064658990256866,
      "grad_norm": 0.22175614535808563,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 42975
    },
    {
      "epoch": 38.06908768821966,
      "grad_norm": 0.23338641226291656,
      "learning_rate": 1e-05,
      "loss": 1.0403,
      "step": 42980
    },
    {
      "epoch": 38.07351638618246,
      "grad_norm": 0.24338971078395844,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 42985
    },
    {
      "epoch": 38.07794508414526,
      "grad_norm": 0.2220805585384369,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 42990
    },
    {
      "epoch": 38.08237378210806,
      "grad_norm": 0.24641622602939606,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 42995
    },
    {
      "epoch": 38.08680248007086,
      "grad_norm": 0.26084789633750916,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 43000
    },
    {
      "epoch": 38.09123117803366,
      "grad_norm": 0.25485512614250183,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 43005
    },
    {
      "epoch": 38.09565987599645,
      "grad_norm": 0.254149854183197,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 43010
    },
    {
      "epoch": 38.100088573959255,
      "grad_norm": 0.2314128428697586,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 43015
    },
    {
      "epoch": 38.104517271922056,
      "grad_norm": 0.25796201825141907,
      "learning_rate": 1e-05,
      "loss": 0.9191,
      "step": 43020
    },
    {
      "epoch": 38.10894596988486,
      "grad_norm": 0.22887766361236572,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 43025
    },
    {
      "epoch": 38.11337466784765,
      "grad_norm": 0.2184365689754486,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 43030
    },
    {
      "epoch": 38.11780336581045,
      "grad_norm": 0.23559419810771942,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 43035
    },
    {
      "epoch": 38.122232063773254,
      "grad_norm": 0.22974500060081482,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 43040
    },
    {
      "epoch": 38.12666076173605,
      "grad_norm": 0.2795167863368988,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 43045
    },
    {
      "epoch": 38.13108945969885,
      "grad_norm": 0.28331977128982544,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 43050
    },
    {
      "epoch": 38.13551815766165,
      "grad_norm": 0.2635152339935303,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 43055
    },
    {
      "epoch": 38.139946855624444,
      "grad_norm": 0.23196662962436676,
      "learning_rate": 1e-05,
      "loss": 1.0604,
      "step": 43060
    },
    {
      "epoch": 38.144375553587246,
      "grad_norm": 0.24418827891349792,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 43065
    },
    {
      "epoch": 38.14880425155005,
      "grad_norm": 0.24671323597431183,
      "learning_rate": 1e-05,
      "loss": 0.9188,
      "step": 43070
    },
    {
      "epoch": 38.15323294951284,
      "grad_norm": 0.23762334883213043,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 43075
    },
    {
      "epoch": 38.15766164747564,
      "grad_norm": 0.224981427192688,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 43080
    },
    {
      "epoch": 38.16209034543844,
      "grad_norm": 0.21803024411201477,
      "learning_rate": 1e-05,
      "loss": 0.9205,
      "step": 43085
    },
    {
      "epoch": 38.16651904340124,
      "grad_norm": 0.22328072786331177,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 43090
    },
    {
      "epoch": 38.17094774136404,
      "grad_norm": 0.25803542137145996,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 43095
    },
    {
      "epoch": 38.17537643932684,
      "grad_norm": 0.25329846143722534,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 43100
    },
    {
      "epoch": 38.179805137289634,
      "grad_norm": 0.25826695561408997,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 43105
    },
    {
      "epoch": 38.184233835252435,
      "grad_norm": 0.2522648572921753,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 43110
    },
    {
      "epoch": 38.188662533215236,
      "grad_norm": 0.26407769322395325,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 43115
    },
    {
      "epoch": 38.19309123117803,
      "grad_norm": 0.2603117525577545,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 43120
    },
    {
      "epoch": 38.19751992914083,
      "grad_norm": 0.21637728810310364,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 43125
    },
    {
      "epoch": 38.20194862710363,
      "grad_norm": 0.26354262232780457,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 43130
    },
    {
      "epoch": 38.20637732506643,
      "grad_norm": 0.24848845601081848,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 43135
    },
    {
      "epoch": 38.21080602302923,
      "grad_norm": 0.24224057793617249,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 43140
    },
    {
      "epoch": 38.21523472099203,
      "grad_norm": 0.22280791401863098,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 43145
    },
    {
      "epoch": 38.21966341895483,
      "grad_norm": 0.21780434250831604,
      "learning_rate": 1e-05,
      "loss": 0.941,
      "step": 43150
    },
    {
      "epoch": 38.224092116917625,
      "grad_norm": 0.2702253460884094,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 43155
    },
    {
      "epoch": 38.228520814880426,
      "grad_norm": 0.2269081175327301,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 43160
    },
    {
      "epoch": 38.23294951284323,
      "grad_norm": 0.183292418718338,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 43165
    },
    {
      "epoch": 38.23737821080602,
      "grad_norm": 0.22489820420742035,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 43170
    },
    {
      "epoch": 38.24180690876882,
      "grad_norm": 0.29860106110572815,
      "learning_rate": 1e-05,
      "loss": 0.915,
      "step": 43175
    },
    {
      "epoch": 38.246235606731624,
      "grad_norm": 0.2841985523700714,
      "learning_rate": 1e-05,
      "loss": 1.0521,
      "step": 43180
    },
    {
      "epoch": 38.25066430469442,
      "grad_norm": 0.2412446141242981,
      "learning_rate": 1e-05,
      "loss": 1.047,
      "step": 43185
    },
    {
      "epoch": 38.25509300265722,
      "grad_norm": 0.23004049062728882,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 43190
    },
    {
      "epoch": 38.25952170062002,
      "grad_norm": 0.19633902609348297,
      "learning_rate": 1e-05,
      "loss": 1.0324,
      "step": 43195
    },
    {
      "epoch": 38.263950398582814,
      "grad_norm": 0.2862204909324646,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 43200
    },
    {
      "epoch": 38.268379096545615,
      "grad_norm": 0.23032046854496002,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 43205
    },
    {
      "epoch": 38.27280779450842,
      "grad_norm": 0.28612473607063293,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 43210
    },
    {
      "epoch": 38.27723649247121,
      "grad_norm": 0.2367434948682785,
      "learning_rate": 1e-05,
      "loss": 0.9346,
      "step": 43215
    },
    {
      "epoch": 38.28166519043401,
      "grad_norm": 0.24974460899829865,
      "learning_rate": 1e-05,
      "loss": 1.0288,
      "step": 43220
    },
    {
      "epoch": 38.28609388839681,
      "grad_norm": 0.22161123156547546,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 43225
    },
    {
      "epoch": 38.29052258635961,
      "grad_norm": 0.2765234410762787,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 43230
    },
    {
      "epoch": 38.29495128432241,
      "grad_norm": 0.23298610746860504,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 43235
    },
    {
      "epoch": 38.29937998228521,
      "grad_norm": 0.20875632762908936,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 43240
    },
    {
      "epoch": 38.303808680248004,
      "grad_norm": 0.24329780042171478,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 43245
    },
    {
      "epoch": 38.308237378210805,
      "grad_norm": 0.21711863577365875,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 43250
    },
    {
      "epoch": 38.312666076173606,
      "grad_norm": 0.24762588739395142,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 43255
    },
    {
      "epoch": 38.31709477413641,
      "grad_norm": 0.23348502814769745,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 43260
    },
    {
      "epoch": 38.3215234720992,
      "grad_norm": 0.2355189323425293,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 43265
    },
    {
      "epoch": 38.325952170062,
      "grad_norm": 0.20766477286815643,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 43270
    },
    {
      "epoch": 38.330380868024804,
      "grad_norm": 0.24386738240718842,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 43275
    },
    {
      "epoch": 38.3348095659876,
      "grad_norm": 0.23066936433315277,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 43280
    },
    {
      "epoch": 38.3392382639504,
      "grad_norm": 0.22878050804138184,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 43285
    },
    {
      "epoch": 38.3436669619132,
      "grad_norm": 0.21676266193389893,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 43290
    },
    {
      "epoch": 38.348095659875995,
      "grad_norm": 0.22341535985469818,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 43295
    },
    {
      "epoch": 38.352524357838796,
      "grad_norm": 0.2633419930934906,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 43300
    },
    {
      "epoch": 38.3569530558016,
      "grad_norm": 0.27283814549446106,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 43305
    },
    {
      "epoch": 38.36138175376439,
      "grad_norm": 0.2556975483894348,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 43310
    },
    {
      "epoch": 38.36581045172719,
      "grad_norm": 0.21136456727981567,
      "learning_rate": 1e-05,
      "loss": 1.0326,
      "step": 43315
    },
    {
      "epoch": 38.37023914968999,
      "grad_norm": 0.2684818208217621,
      "learning_rate": 1e-05,
      "loss": 0.8982,
      "step": 43320
    },
    {
      "epoch": 38.37466784765279,
      "grad_norm": 0.23107388615608215,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 43325
    },
    {
      "epoch": 38.37909654561559,
      "grad_norm": 0.2445419728755951,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 43330
    },
    {
      "epoch": 38.38352524357839,
      "grad_norm": 0.2546462416648865,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 43335
    },
    {
      "epoch": 38.387953941541184,
      "grad_norm": 0.23402030766010284,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 43340
    },
    {
      "epoch": 38.392382639503985,
      "grad_norm": 0.22267848253250122,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 43345
    },
    {
      "epoch": 38.39681133746679,
      "grad_norm": 0.27227842807769775,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 43350
    },
    {
      "epoch": 38.40124003542958,
      "grad_norm": 0.24881525337696075,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 43355
    },
    {
      "epoch": 38.40566873339238,
      "grad_norm": 0.21689139306545258,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 43360
    },
    {
      "epoch": 38.41009743135518,
      "grad_norm": 0.2058870643377304,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 43365
    },
    {
      "epoch": 38.41452612931798,
      "grad_norm": 0.20513705909252167,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 43370
    },
    {
      "epoch": 38.41895482728078,
      "grad_norm": 0.20016486942768097,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 43375
    },
    {
      "epoch": 38.42338352524358,
      "grad_norm": 0.22196191549301147,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 43380
    },
    {
      "epoch": 38.42781222320638,
      "grad_norm": 0.2295418083667755,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 43385
    },
    {
      "epoch": 38.432240921169175,
      "grad_norm": 0.2352699637413025,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 43390
    },
    {
      "epoch": 38.436669619131976,
      "grad_norm": 0.31505686044692993,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 43395
    },
    {
      "epoch": 38.44109831709478,
      "grad_norm": 0.22873611748218536,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 43400
    },
    {
      "epoch": 38.44552701505757,
      "grad_norm": 0.21728169918060303,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 43405
    },
    {
      "epoch": 38.44995571302037,
      "grad_norm": 0.2645941972732544,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 43410
    },
    {
      "epoch": 38.454384410983174,
      "grad_norm": 0.22846366465091705,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 43415
    },
    {
      "epoch": 38.45881310894597,
      "grad_norm": 0.22175458073616028,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 43420
    },
    {
      "epoch": 38.46324180690877,
      "grad_norm": 0.26965638995170593,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 43425
    },
    {
      "epoch": 38.46767050487157,
      "grad_norm": 0.22983549535274506,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 43430
    },
    {
      "epoch": 38.472099202834364,
      "grad_norm": 0.24273781478405,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 43435
    },
    {
      "epoch": 38.476527900797166,
      "grad_norm": 0.20609554648399353,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 43440
    },
    {
      "epoch": 38.48095659875997,
      "grad_norm": 0.22920441627502441,
      "learning_rate": 1e-05,
      "loss": 0.9086,
      "step": 43445
    },
    {
      "epoch": 38.48538529672276,
      "grad_norm": 0.24241000413894653,
      "learning_rate": 1e-05,
      "loss": 0.9157,
      "step": 43450
    },
    {
      "epoch": 38.48981399468556,
      "grad_norm": 0.29323849081993103,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 43455
    },
    {
      "epoch": 38.49424269264836,
      "grad_norm": 0.20561303198337555,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 43460
    },
    {
      "epoch": 38.49867139061116,
      "grad_norm": 0.22029079496860504,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 43465
    },
    {
      "epoch": 38.50310008857396,
      "grad_norm": 0.26725682616233826,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 43470
    },
    {
      "epoch": 38.50752878653676,
      "grad_norm": 0.2451079785823822,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 43475
    },
    {
      "epoch": 38.511957484499554,
      "grad_norm": 0.20564872026443481,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 43480
    },
    {
      "epoch": 38.516386182462355,
      "grad_norm": 0.2674044072628021,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 43485
    },
    {
      "epoch": 38.520814880425156,
      "grad_norm": 0.26122304797172546,
      "learning_rate": 1e-05,
      "loss": 0.9169,
      "step": 43490
    },
    {
      "epoch": 38.52524357838795,
      "grad_norm": 0.2647591233253479,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 43495
    },
    {
      "epoch": 38.52967227635075,
      "grad_norm": 0.2281855046749115,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 43500
    },
    {
      "epoch": 38.53410097431355,
      "grad_norm": 0.2355351597070694,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 43505
    },
    {
      "epoch": 38.538529672276354,
      "grad_norm": 0.25589779019355774,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 43510
    },
    {
      "epoch": 38.54295837023915,
      "grad_norm": 0.211058109998703,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 43515
    },
    {
      "epoch": 38.54738706820195,
      "grad_norm": 0.2609139382839203,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 43520
    },
    {
      "epoch": 38.55181576616475,
      "grad_norm": 0.2622933089733124,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 43525
    },
    {
      "epoch": 38.556244464127545,
      "grad_norm": 0.20722365379333496,
      "learning_rate": 1e-05,
      "loss": 1.0252,
      "step": 43530
    },
    {
      "epoch": 38.560673162090346,
      "grad_norm": 0.21178418397903442,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 43535
    },
    {
      "epoch": 38.56510186005315,
      "grad_norm": 0.21725904941558838,
      "learning_rate": 1e-05,
      "loss": 0.9318,
      "step": 43540
    },
    {
      "epoch": 38.56953055801594,
      "grad_norm": 0.2667466700077057,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 43545
    },
    {
      "epoch": 38.57395925597874,
      "grad_norm": 0.21910931169986725,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 43550
    },
    {
      "epoch": 38.578387953941544,
      "grad_norm": 0.2225453406572342,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 43555
    },
    {
      "epoch": 38.58281665190434,
      "grad_norm": 0.23588016629219055,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 43560
    },
    {
      "epoch": 38.58724534986714,
      "grad_norm": 0.23776131868362427,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 43565
    },
    {
      "epoch": 38.59167404782994,
      "grad_norm": 0.26252466440200806,
      "learning_rate": 1e-05,
      "loss": 1.0281,
      "step": 43570
    },
    {
      "epoch": 38.596102745792734,
      "grad_norm": 0.24215373396873474,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 43575
    },
    {
      "epoch": 38.600531443755536,
      "grad_norm": 0.23254767060279846,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 43580
    },
    {
      "epoch": 38.60496014171834,
      "grad_norm": 0.2724493443965912,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 43585
    },
    {
      "epoch": 38.60938883968113,
      "grad_norm": 0.27124494314193726,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 43590
    },
    {
      "epoch": 38.61381753764393,
      "grad_norm": 0.23203571140766144,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 43595
    },
    {
      "epoch": 38.61824623560673,
      "grad_norm": 0.21647340059280396,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 43600
    },
    {
      "epoch": 38.62267493356953,
      "grad_norm": 0.24302645027637482,
      "learning_rate": 1e-05,
      "loss": 1.0629,
      "step": 43605
    },
    {
      "epoch": 38.62710363153233,
      "grad_norm": 0.2427358478307724,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 43610
    },
    {
      "epoch": 38.63153232949513,
      "grad_norm": 0.2331506907939911,
      "learning_rate": 1e-05,
      "loss": 1.0363,
      "step": 43615
    },
    {
      "epoch": 38.635961027457924,
      "grad_norm": 0.2289654016494751,
      "learning_rate": 1e-05,
      "loss": 0.9121,
      "step": 43620
    },
    {
      "epoch": 38.640389725420725,
      "grad_norm": 0.21937672793865204,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 43625
    },
    {
      "epoch": 38.644818423383526,
      "grad_norm": 0.226113960146904,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 43630
    },
    {
      "epoch": 38.64924712134633,
      "grad_norm": 0.239945188164711,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 43635
    },
    {
      "epoch": 38.65367581930912,
      "grad_norm": 0.21665291488170624,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 43640
    },
    {
      "epoch": 38.65810451727192,
      "grad_norm": 0.2411862015724182,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 43645
    },
    {
      "epoch": 38.662533215234724,
      "grad_norm": 0.2256542295217514,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 43650
    },
    {
      "epoch": 38.66696191319752,
      "grad_norm": 0.24472227692604065,
      "learning_rate": 1e-05,
      "loss": 0.9129,
      "step": 43655
    },
    {
      "epoch": 38.67139061116032,
      "grad_norm": 0.24234147369861603,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 43660
    },
    {
      "epoch": 38.67581930912312,
      "grad_norm": 0.2653104364871979,
      "learning_rate": 1e-05,
      "loss": 1.025,
      "step": 43665
    },
    {
      "epoch": 38.680248007085915,
      "grad_norm": 0.18867559731006622,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 43670
    },
    {
      "epoch": 38.684676705048716,
      "grad_norm": 0.24823516607284546,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 43675
    },
    {
      "epoch": 38.68910540301152,
      "grad_norm": 0.23131290078163147,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 43680
    },
    {
      "epoch": 38.69353410097431,
      "grad_norm": 0.21727074682712555,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 43685
    },
    {
      "epoch": 38.69796279893711,
      "grad_norm": 0.26791298389434814,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 43690
    },
    {
      "epoch": 38.702391496899914,
      "grad_norm": 0.23814387619495392,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 43695
    },
    {
      "epoch": 38.70682019486271,
      "grad_norm": 0.23660992085933685,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 43700
    },
    {
      "epoch": 38.71124889282551,
      "grad_norm": 0.24554641544818878,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 43705
    },
    {
      "epoch": 38.71567759078831,
      "grad_norm": 0.246957466006279,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 43710
    },
    {
      "epoch": 38.720106288751104,
      "grad_norm": 0.22561015188694,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 43715
    },
    {
      "epoch": 38.724534986713905,
      "grad_norm": 0.25264501571655273,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 43720
    },
    {
      "epoch": 38.72896368467671,
      "grad_norm": 0.21695616841316223,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 43725
    },
    {
      "epoch": 38.7333923826395,
      "grad_norm": 0.24012257158756256,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 43730
    },
    {
      "epoch": 38.7378210806023,
      "grad_norm": 0.23363974690437317,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 43735
    },
    {
      "epoch": 38.7422497785651,
      "grad_norm": 0.2782191038131714,
      "learning_rate": 1e-05,
      "loss": 0.8982,
      "step": 43740
    },
    {
      "epoch": 38.7466784765279,
      "grad_norm": 0.2390768826007843,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 43745
    },
    {
      "epoch": 38.7511071744907,
      "grad_norm": 0.2396044135093689,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 43750
    },
    {
      "epoch": 38.7555358724535,
      "grad_norm": 0.23891164362430573,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 43755
    },
    {
      "epoch": 38.7599645704163,
      "grad_norm": 0.21045605838298798,
      "learning_rate": 1e-05,
      "loss": 0.8854,
      "step": 43760
    },
    {
      "epoch": 38.764393268379095,
      "grad_norm": 0.24695606529712677,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 43765
    },
    {
      "epoch": 38.768821966341896,
      "grad_norm": 0.2641974985599518,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 43770
    },
    {
      "epoch": 38.7732506643047,
      "grad_norm": 0.25685933232307434,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 43775
    },
    {
      "epoch": 38.77767936226749,
      "grad_norm": 0.2249232828617096,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 43780
    },
    {
      "epoch": 38.78210806023029,
      "grad_norm": 0.21370990574359894,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 43785
    },
    {
      "epoch": 38.786536758193094,
      "grad_norm": 0.24395334720611572,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 43790
    },
    {
      "epoch": 38.79096545615589,
      "grad_norm": 0.23896770179271698,
      "learning_rate": 1e-05,
      "loss": 0.9148,
      "step": 43795
    },
    {
      "epoch": 38.79539415411869,
      "grad_norm": 0.2118944376707077,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 43800
    },
    {
      "epoch": 38.79982285208149,
      "grad_norm": 0.21172665059566498,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 43805
    },
    {
      "epoch": 38.804251550044285,
      "grad_norm": 0.2833765149116516,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 43810
    },
    {
      "epoch": 38.808680248007086,
      "grad_norm": 0.23825910687446594,
      "learning_rate": 1e-05,
      "loss": 1.0199,
      "step": 43815
    },
    {
      "epoch": 38.81310894596989,
      "grad_norm": 0.1943647712469101,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 43820
    },
    {
      "epoch": 38.81753764393268,
      "grad_norm": 0.27112287282943726,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 43825
    },
    {
      "epoch": 38.82196634189548,
      "grad_norm": 0.3868490755558014,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 43830
    },
    {
      "epoch": 38.826395039858284,
      "grad_norm": 0.31806594133377075,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 43835
    },
    {
      "epoch": 38.83082373782108,
      "grad_norm": 0.25791141390800476,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 43840
    },
    {
      "epoch": 38.83525243578388,
      "grad_norm": 0.3068712055683136,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 43845
    },
    {
      "epoch": 38.83968113374668,
      "grad_norm": 0.26004689931869507,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 43850
    },
    {
      "epoch": 38.844109831709474,
      "grad_norm": 0.2442476749420166,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 43855
    },
    {
      "epoch": 38.848538529672275,
      "grad_norm": 0.2509099841117859,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 43860
    },
    {
      "epoch": 38.85296722763508,
      "grad_norm": 0.22360014915466309,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 43865
    },
    {
      "epoch": 38.85739592559787,
      "grad_norm": 0.25066977739334106,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 43870
    },
    {
      "epoch": 38.86182462356067,
      "grad_norm": 0.2108364701271057,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 43875
    },
    {
      "epoch": 38.86625332152347,
      "grad_norm": 0.23675523698329926,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 43880
    },
    {
      "epoch": 38.870682019486274,
      "grad_norm": 0.22983263432979584,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 43885
    },
    {
      "epoch": 38.87511071744907,
      "grad_norm": 0.21928900480270386,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 43890
    },
    {
      "epoch": 38.87953941541187,
      "grad_norm": 0.23863732814788818,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 43895
    },
    {
      "epoch": 38.88396811337467,
      "grad_norm": 0.22324906289577484,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 43900
    },
    {
      "epoch": 38.888396811337465,
      "grad_norm": 0.26141077280044556,
      "learning_rate": 1e-05,
      "loss": 1.0267,
      "step": 43905
    },
    {
      "epoch": 38.892825509300266,
      "grad_norm": 0.2276572585105896,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 43910
    },
    {
      "epoch": 38.89725420726307,
      "grad_norm": 0.21929779648780823,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 43915
    },
    {
      "epoch": 38.90168290522586,
      "grad_norm": 0.25231146812438965,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 43920
    },
    {
      "epoch": 38.90611160318866,
      "grad_norm": 0.28124314546585083,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 43925
    },
    {
      "epoch": 38.910540301151464,
      "grad_norm": 0.2180287390947342,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 43930
    },
    {
      "epoch": 38.91496899911426,
      "grad_norm": 0.22384630143642426,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 43935
    },
    {
      "epoch": 38.91939769707706,
      "grad_norm": 0.20836429297924042,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 43940
    },
    {
      "epoch": 38.92382639503986,
      "grad_norm": 0.21534837782382965,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 43945
    },
    {
      "epoch": 38.928255093002655,
      "grad_norm": 0.22883909940719604,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 43950
    },
    {
      "epoch": 38.932683790965456,
      "grad_norm": 0.2744563817977905,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 43955
    },
    {
      "epoch": 38.93711248892826,
      "grad_norm": 0.26108402013778687,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 43960
    },
    {
      "epoch": 38.94154118689105,
      "grad_norm": 0.20949621498584747,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 43965
    },
    {
      "epoch": 38.94596988485385,
      "grad_norm": 0.22525851428508759,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 43970
    },
    {
      "epoch": 38.95039858281665,
      "grad_norm": 0.288987398147583,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 43975
    },
    {
      "epoch": 38.95482728077945,
      "grad_norm": 0.2601258158683777,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 43980
    },
    {
      "epoch": 38.95925597874225,
      "grad_norm": 0.2004460245370865,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 43985
    },
    {
      "epoch": 38.96368467670505,
      "grad_norm": 0.26141634583473206,
      "learning_rate": 1e-05,
      "loss": 1.0159,
      "step": 43990
    },
    {
      "epoch": 38.96811337466785,
      "grad_norm": 0.26430144906044006,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 43995
    },
    {
      "epoch": 38.972542072630645,
      "grad_norm": 0.29508286714553833,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 44000
    },
    {
      "epoch": 38.97697077059345,
      "grad_norm": 0.23690460622310638,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 44005
    },
    {
      "epoch": 38.98139946855625,
      "grad_norm": 0.2896316647529602,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 44010
    },
    {
      "epoch": 38.98582816651904,
      "grad_norm": 0.2283962517976761,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 44015
    },
    {
      "epoch": 38.99025686448184,
      "grad_norm": 0.24760879576206207,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 44020
    },
    {
      "epoch": 38.994685562444644,
      "grad_norm": 0.2084726095199585,
      "learning_rate": 1e-05,
      "loss": 0.8882,
      "step": 44025
    },
    {
      "epoch": 38.99911426040744,
      "grad_norm": 0.21469901502132416,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 44030
    },
    {
      "epoch": 39.00354295837024,
      "grad_norm": 0.207458958029747,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 44035
    },
    {
      "epoch": 39.00797165633304,
      "grad_norm": 0.23130010068416595,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 44040
    },
    {
      "epoch": 39.012400354295835,
      "grad_norm": 0.2177809774875641,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 44045
    },
    {
      "epoch": 39.016829052258636,
      "grad_norm": 0.25430431962013245,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 44050
    },
    {
      "epoch": 39.02125775022144,
      "grad_norm": 0.2396925538778305,
      "learning_rate": 1e-05,
      "loss": 1.0184,
      "step": 44055
    },
    {
      "epoch": 39.02568644818423,
      "grad_norm": 0.2683342397212982,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 44060
    },
    {
      "epoch": 39.03011514614703,
      "grad_norm": 0.24209195375442505,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 44065
    },
    {
      "epoch": 39.034543844109834,
      "grad_norm": 0.26828837394714355,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 44070
    },
    {
      "epoch": 39.03897254207263,
      "grad_norm": 0.23773469030857086,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 44075
    },
    {
      "epoch": 39.04340124003543,
      "grad_norm": 0.26834332942962646,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 44080
    },
    {
      "epoch": 39.04782993799823,
      "grad_norm": 0.2255697250366211,
      "learning_rate": 1e-05,
      "loss": 0.9245,
      "step": 44085
    },
    {
      "epoch": 39.052258635961024,
      "grad_norm": 0.2594926357269287,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 44090
    },
    {
      "epoch": 39.056687333923826,
      "grad_norm": 0.29831230640411377,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 44095
    },
    {
      "epoch": 39.06111603188663,
      "grad_norm": 0.22961190342903137,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 44100
    },
    {
      "epoch": 39.06554472984942,
      "grad_norm": 0.25696319341659546,
      "learning_rate": 1e-05,
      "loss": 1.023,
      "step": 44105
    },
    {
      "epoch": 39.06997342781222,
      "grad_norm": 0.2482203096151352,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 44110
    },
    {
      "epoch": 39.07440212577502,
      "grad_norm": 0.2834431231021881,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 44115
    },
    {
      "epoch": 39.078830823737825,
      "grad_norm": 0.24590784311294556,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 44120
    },
    {
      "epoch": 39.08325952170062,
      "grad_norm": 0.2518172264099121,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 44125
    },
    {
      "epoch": 39.08768821966342,
      "grad_norm": 0.25441253185272217,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 44130
    },
    {
      "epoch": 39.09211691762622,
      "grad_norm": 0.35825350880622864,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 44135
    },
    {
      "epoch": 39.096545615589015,
      "grad_norm": 0.2582775950431824,
      "learning_rate": 1e-05,
      "loss": 1.0231,
      "step": 44140
    },
    {
      "epoch": 39.100974313551816,
      "grad_norm": 0.29992473125457764,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 44145
    },
    {
      "epoch": 39.10540301151462,
      "grad_norm": 0.2526478171348572,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 44150
    },
    {
      "epoch": 39.10983170947741,
      "grad_norm": 0.2766793966293335,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 44155
    },
    {
      "epoch": 39.11426040744021,
      "grad_norm": 0.2710573971271515,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 44160
    },
    {
      "epoch": 39.118689105403014,
      "grad_norm": 0.2608216106891632,
      "learning_rate": 1e-05,
      "loss": 1.0402,
      "step": 44165
    },
    {
      "epoch": 39.12311780336581,
      "grad_norm": 0.23495997488498688,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 44170
    },
    {
      "epoch": 39.12754650132861,
      "grad_norm": 0.27490705251693726,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 44175
    },
    {
      "epoch": 39.13197519929141,
      "grad_norm": 0.2554977536201477,
      "learning_rate": 1e-05,
      "loss": 0.8788,
      "step": 44180
    },
    {
      "epoch": 39.136403897254205,
      "grad_norm": 0.26549553871154785,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 44185
    },
    {
      "epoch": 39.140832595217006,
      "grad_norm": 0.24583964049816132,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 44190
    },
    {
      "epoch": 39.14526129317981,
      "grad_norm": 0.20966410636901855,
      "learning_rate": 1e-05,
      "loss": 1.0439,
      "step": 44195
    },
    {
      "epoch": 39.1496899911426,
      "grad_norm": 0.23834633827209473,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 44200
    },
    {
      "epoch": 39.1541186891054,
      "grad_norm": 0.19447845220565796,
      "learning_rate": 1e-05,
      "loss": 0.9367,
      "step": 44205
    },
    {
      "epoch": 39.158547387068204,
      "grad_norm": 0.3450884222984314,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 44210
    },
    {
      "epoch": 39.162976085031,
      "grad_norm": 0.22826288640499115,
      "learning_rate": 1e-05,
      "loss": 1.028,
      "step": 44215
    },
    {
      "epoch": 39.1674047829938,
      "grad_norm": 0.2646857798099518,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 44220
    },
    {
      "epoch": 39.1718334809566,
      "grad_norm": 0.2580079436302185,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 44225
    },
    {
      "epoch": 39.176262178919394,
      "grad_norm": 0.26738640666007996,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 44230
    },
    {
      "epoch": 39.180690876882196,
      "grad_norm": 0.26614123582839966,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 44235
    },
    {
      "epoch": 39.185119574845,
      "grad_norm": 0.21896515786647797,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 44240
    },
    {
      "epoch": 39.1895482728078,
      "grad_norm": 0.20820488035678864,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 44245
    },
    {
      "epoch": 39.19397697077059,
      "grad_norm": 0.23873235285282135,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 44250
    },
    {
      "epoch": 39.19840566873339,
      "grad_norm": 0.22646278142929077,
      "learning_rate": 1e-05,
      "loss": 0.9205,
      "step": 44255
    },
    {
      "epoch": 39.202834366696194,
      "grad_norm": 0.23012956976890564,
      "learning_rate": 1e-05,
      "loss": 1.0508,
      "step": 44260
    },
    {
      "epoch": 39.20726306465899,
      "grad_norm": 0.23358553647994995,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 44265
    },
    {
      "epoch": 39.21169176262179,
      "grad_norm": 0.22419176995754242,
      "learning_rate": 1e-05,
      "loss": 1.0257,
      "step": 44270
    },
    {
      "epoch": 39.21612046058459,
      "grad_norm": 0.29865843057632446,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 44275
    },
    {
      "epoch": 39.220549158547385,
      "grad_norm": 0.23881004750728607,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 44280
    },
    {
      "epoch": 39.224977856510186,
      "grad_norm": 0.2720155417919159,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 44285
    },
    {
      "epoch": 39.22940655447299,
      "grad_norm": 0.2511255741119385,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 44290
    },
    {
      "epoch": 39.23383525243578,
      "grad_norm": 0.24253632128238678,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 44295
    },
    {
      "epoch": 39.23826395039858,
      "grad_norm": 0.2567937970161438,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 44300
    },
    {
      "epoch": 39.242692648361384,
      "grad_norm": 0.2816894054412842,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 44305
    },
    {
      "epoch": 39.24712134632418,
      "grad_norm": 0.2455597072839737,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 44310
    },
    {
      "epoch": 39.25155004428698,
      "grad_norm": 0.22847037017345428,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 44315
    },
    {
      "epoch": 39.25597874224978,
      "grad_norm": 0.24603794515132904,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 44320
    },
    {
      "epoch": 39.260407440212575,
      "grad_norm": 0.25504645705223083,
      "learning_rate": 1e-05,
      "loss": 0.9277,
      "step": 44325
    },
    {
      "epoch": 39.264836138175376,
      "grad_norm": 0.24797974526882172,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 44330
    },
    {
      "epoch": 39.26926483613818,
      "grad_norm": 0.2609892785549164,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 44335
    },
    {
      "epoch": 39.27369353410097,
      "grad_norm": 0.24699504673480988,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 44340
    },
    {
      "epoch": 39.27812223206377,
      "grad_norm": 0.17874492704868317,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 44345
    },
    {
      "epoch": 39.282550930026574,
      "grad_norm": 0.25838321447372437,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 44350
    },
    {
      "epoch": 39.28697962798937,
      "grad_norm": 0.2813611328601837,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 44355
    },
    {
      "epoch": 39.29140832595217,
      "grad_norm": 0.24154804646968842,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 44360
    },
    {
      "epoch": 39.29583702391497,
      "grad_norm": 0.2507898211479187,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 44365
    },
    {
      "epoch": 39.30026572187777,
      "grad_norm": 0.24096839129924774,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 44370
    },
    {
      "epoch": 39.304694419840565,
      "grad_norm": 0.18299080431461334,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 44375
    },
    {
      "epoch": 39.30912311780337,
      "grad_norm": 0.2190800905227661,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 44380
    },
    {
      "epoch": 39.31355181576617,
      "grad_norm": 0.23172251880168915,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 44385
    },
    {
      "epoch": 39.31798051372896,
      "grad_norm": 0.21666519343852997,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 44390
    },
    {
      "epoch": 39.32240921169176,
      "grad_norm": 0.22334641218185425,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 44395
    },
    {
      "epoch": 39.326837909654564,
      "grad_norm": 0.2638227045536041,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 44400
    },
    {
      "epoch": 39.33126660761736,
      "grad_norm": 0.23349282145500183,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 44405
    },
    {
      "epoch": 39.33569530558016,
      "grad_norm": 0.24016672372817993,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 44410
    },
    {
      "epoch": 39.34012400354296,
      "grad_norm": 0.23537923395633698,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 44415
    },
    {
      "epoch": 39.344552701505755,
      "grad_norm": 0.3087572157382965,
      "learning_rate": 1e-05,
      "loss": 1.0216,
      "step": 44420
    },
    {
      "epoch": 39.348981399468556,
      "grad_norm": 0.28580746054649353,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 44425
    },
    {
      "epoch": 39.35341009743136,
      "grad_norm": 0.22272680699825287,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 44430
    },
    {
      "epoch": 39.35783879539415,
      "grad_norm": 0.2559676170349121,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 44435
    },
    {
      "epoch": 39.36226749335695,
      "grad_norm": 0.23016200959682465,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 44440
    },
    {
      "epoch": 39.366696191319754,
      "grad_norm": 0.25372010469436646,
      "learning_rate": 1e-05,
      "loss": 1.0639,
      "step": 44445
    },
    {
      "epoch": 39.37112488928255,
      "grad_norm": 0.25866615772247314,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 44450
    },
    {
      "epoch": 39.37555358724535,
      "grad_norm": 0.21047712862491608,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 44455
    },
    {
      "epoch": 39.37998228520815,
      "grad_norm": 0.2168559730052948,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 44460
    },
    {
      "epoch": 39.384410983170945,
      "grad_norm": 0.21960103511810303,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 44465
    },
    {
      "epoch": 39.388839681133746,
      "grad_norm": 0.2648416757583618,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 44470
    },
    {
      "epoch": 39.39326837909655,
      "grad_norm": 0.24154886603355408,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 44475
    },
    {
      "epoch": 39.39769707705934,
      "grad_norm": 0.22506415843963623,
      "learning_rate": 1e-05,
      "loss": 0.9077,
      "step": 44480
    },
    {
      "epoch": 39.40212577502214,
      "grad_norm": 0.21510253846645355,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 44485
    },
    {
      "epoch": 39.40655447298494,
      "grad_norm": 0.24056263267993927,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 44490
    },
    {
      "epoch": 39.410983170947745,
      "grad_norm": 0.2314099222421646,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 44495
    },
    {
      "epoch": 39.41541186891054,
      "grad_norm": 0.2339179664850235,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 44500
    },
    {
      "epoch": 39.41984056687334,
      "grad_norm": 0.21904712915420532,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 44505
    },
    {
      "epoch": 39.42426926483614,
      "grad_norm": 0.24298132956027985,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 44510
    },
    {
      "epoch": 39.428697962798935,
      "grad_norm": 0.2730855643749237,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 44515
    },
    {
      "epoch": 39.43312666076174,
      "grad_norm": 0.276470422744751,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 44520
    },
    {
      "epoch": 39.43755535872454,
      "grad_norm": 0.2324453443288803,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 44525
    },
    {
      "epoch": 39.44198405668733,
      "grad_norm": 0.27155014872550964,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 44530
    },
    {
      "epoch": 39.44641275465013,
      "grad_norm": 0.20579677820205688,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 44535
    },
    {
      "epoch": 39.450841452612934,
      "grad_norm": 0.27589836716651917,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 44540
    },
    {
      "epoch": 39.45527015057573,
      "grad_norm": 0.25818344950675964,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 44545
    },
    {
      "epoch": 39.45969884853853,
      "grad_norm": 0.2616889774799347,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 44550
    },
    {
      "epoch": 39.46412754650133,
      "grad_norm": 0.261217325925827,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 44555
    },
    {
      "epoch": 39.468556244464125,
      "grad_norm": 0.2635051906108856,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 44560
    },
    {
      "epoch": 39.472984942426926,
      "grad_norm": 0.24376420676708221,
      "learning_rate": 1e-05,
      "loss": 0.9493,
      "step": 44565
    },
    {
      "epoch": 39.47741364038973,
      "grad_norm": 0.26818716526031494,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 44570
    },
    {
      "epoch": 39.48184233835252,
      "grad_norm": 0.23902402818202972,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 44575
    },
    {
      "epoch": 39.48627103631532,
      "grad_norm": 0.26013222336769104,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 44580
    },
    {
      "epoch": 39.490699734278124,
      "grad_norm": 0.2219470739364624,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 44585
    },
    {
      "epoch": 39.49512843224092,
      "grad_norm": 0.22296161949634552,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 44590
    },
    {
      "epoch": 39.49955713020372,
      "grad_norm": 0.28529879450798035,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 44595
    },
    {
      "epoch": 39.50398582816652,
      "grad_norm": 0.314669132232666,
      "learning_rate": 1e-05,
      "loss": 1.0405,
      "step": 44600
    },
    {
      "epoch": 39.508414526129314,
      "grad_norm": 0.2913729250431061,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 44605
    },
    {
      "epoch": 39.512843224092116,
      "grad_norm": 0.27496740221977234,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 44610
    },
    {
      "epoch": 39.51727192205492,
      "grad_norm": 0.269618421792984,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 44615
    },
    {
      "epoch": 39.52170062001772,
      "grad_norm": 0.2604779601097107,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 44620
    },
    {
      "epoch": 39.52612931798051,
      "grad_norm": 0.23224522173404694,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 44625
    },
    {
      "epoch": 39.53055801594331,
      "grad_norm": 0.2347531020641327,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 44630
    },
    {
      "epoch": 39.534986713906115,
      "grad_norm": 0.21391569077968597,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 44635
    },
    {
      "epoch": 39.53941541186891,
      "grad_norm": 0.26430365443229675,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 44640
    },
    {
      "epoch": 39.54384410983171,
      "grad_norm": 0.250061571598053,
      "learning_rate": 1e-05,
      "loss": 1.0185,
      "step": 44645
    },
    {
      "epoch": 39.54827280779451,
      "grad_norm": 0.24326933920383453,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 44650
    },
    {
      "epoch": 39.552701505757305,
      "grad_norm": 0.23102015256881714,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 44655
    },
    {
      "epoch": 39.55713020372011,
      "grad_norm": 0.24936458468437195,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 44660
    },
    {
      "epoch": 39.56155890168291,
      "grad_norm": 0.22950631380081177,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 44665
    },
    {
      "epoch": 39.5659875996457,
      "grad_norm": 0.25710630416870117,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 44670
    },
    {
      "epoch": 39.5704162976085,
      "grad_norm": 0.28747037053108215,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 44675
    },
    {
      "epoch": 39.574844995571304,
      "grad_norm": 0.2632983922958374,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 44680
    },
    {
      "epoch": 39.5792736935341,
      "grad_norm": 0.2290157526731491,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 44685
    },
    {
      "epoch": 39.5837023914969,
      "grad_norm": 0.24512092769145966,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 44690
    },
    {
      "epoch": 39.5881310894597,
      "grad_norm": 0.2741369903087616,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 44695
    },
    {
      "epoch": 39.592559787422495,
      "grad_norm": 0.24561484158039093,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 44700
    },
    {
      "epoch": 39.596988485385296,
      "grad_norm": 0.24511860311031342,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 44705
    },
    {
      "epoch": 39.6014171833481,
      "grad_norm": 0.2898678183555603,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 44710
    },
    {
      "epoch": 39.60584588131089,
      "grad_norm": 0.22866323590278625,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 44715
    },
    {
      "epoch": 39.61027457927369,
      "grad_norm": 0.2523864209651947,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 44720
    },
    {
      "epoch": 39.614703277236494,
      "grad_norm": 0.25530171394348145,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 44725
    },
    {
      "epoch": 39.619131975199295,
      "grad_norm": 0.2613741159439087,
      "learning_rate": 1e-05,
      "loss": 0.942,
      "step": 44730
    },
    {
      "epoch": 39.62356067316209,
      "grad_norm": 0.23648029565811157,
      "learning_rate": 1e-05,
      "loss": 0.9269,
      "step": 44735
    },
    {
      "epoch": 39.62798937112489,
      "grad_norm": 0.22205957770347595,
      "learning_rate": 1e-05,
      "loss": 0.903,
      "step": 44740
    },
    {
      "epoch": 39.63241806908769,
      "grad_norm": 0.228758305311203,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 44745
    },
    {
      "epoch": 39.636846767050486,
      "grad_norm": 0.22042939066886902,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 44750
    },
    {
      "epoch": 39.64127546501329,
      "grad_norm": 0.2286497950553894,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 44755
    },
    {
      "epoch": 39.64570416297609,
      "grad_norm": 0.24070866405963898,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 44760
    },
    {
      "epoch": 39.65013286093888,
      "grad_norm": 0.26237747073173523,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 44765
    },
    {
      "epoch": 39.65456155890168,
      "grad_norm": 0.22717924416065216,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 44770
    },
    {
      "epoch": 39.658990256864485,
      "grad_norm": 0.2244872748851776,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 44775
    },
    {
      "epoch": 39.66341895482728,
      "grad_norm": 0.2297087013721466,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 44780
    },
    {
      "epoch": 39.66784765279008,
      "grad_norm": 0.2541778087615967,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 44785
    },
    {
      "epoch": 39.67227635075288,
      "grad_norm": 0.22266943752765656,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 44790
    },
    {
      "epoch": 39.676705048715675,
      "grad_norm": 0.1932983547449112,
      "learning_rate": 1e-05,
      "loss": 0.9221,
      "step": 44795
    },
    {
      "epoch": 39.681133746678476,
      "grad_norm": 0.2179950326681137,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 44800
    },
    {
      "epoch": 39.68556244464128,
      "grad_norm": 0.24509498476982117,
      "learning_rate": 1e-05,
      "loss": 0.9295,
      "step": 44805
    },
    {
      "epoch": 39.68999114260407,
      "grad_norm": 0.24425208568572998,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 44810
    },
    {
      "epoch": 39.69441984056687,
      "grad_norm": 0.21673190593719482,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 44815
    },
    {
      "epoch": 39.698848538529674,
      "grad_norm": 0.2017115205526352,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 44820
    },
    {
      "epoch": 39.70327723649247,
      "grad_norm": 0.28510695695877075,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 44825
    },
    {
      "epoch": 39.70770593445527,
      "grad_norm": 0.2330552488565445,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 44830
    },
    {
      "epoch": 39.71213463241807,
      "grad_norm": 0.31918102502822876,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 44835
    },
    {
      "epoch": 39.716563330380865,
      "grad_norm": 0.27928364276885986,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 44840
    },
    {
      "epoch": 39.720992028343666,
      "grad_norm": 0.20045199990272522,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 44845
    },
    {
      "epoch": 39.72542072630647,
      "grad_norm": 0.28553086519241333,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 44850
    },
    {
      "epoch": 39.72984942426926,
      "grad_norm": 0.26095589995384216,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 44855
    },
    {
      "epoch": 39.73427812223206,
      "grad_norm": 0.19937722384929657,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 44860
    },
    {
      "epoch": 39.738706820194864,
      "grad_norm": 0.2984931766986847,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 44865
    },
    {
      "epoch": 39.743135518157665,
      "grad_norm": 0.22275038063526154,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 44870
    },
    {
      "epoch": 39.74756421612046,
      "grad_norm": 0.237240731716156,
      "learning_rate": 1e-05,
      "loss": 1.0505,
      "step": 44875
    },
    {
      "epoch": 39.75199291408326,
      "grad_norm": 0.25869351625442505,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 44880
    },
    {
      "epoch": 39.75642161204606,
      "grad_norm": 0.2375112920999527,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 44885
    },
    {
      "epoch": 39.760850310008856,
      "grad_norm": 0.28122302889823914,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 44890
    },
    {
      "epoch": 39.76527900797166,
      "grad_norm": 0.2714742422103882,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 44895
    },
    {
      "epoch": 39.76970770593446,
      "grad_norm": 0.25567108392715454,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 44900
    },
    {
      "epoch": 39.77413640389725,
      "grad_norm": 0.2884957790374756,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 44905
    },
    {
      "epoch": 39.77856510186005,
      "grad_norm": 0.20873017609119415,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 44910
    },
    {
      "epoch": 39.782993799822854,
      "grad_norm": 0.27229395508766174,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 44915
    },
    {
      "epoch": 39.78742249778565,
      "grad_norm": 0.23368670046329498,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 44920
    },
    {
      "epoch": 39.79185119574845,
      "grad_norm": 0.28879836201667786,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 44925
    },
    {
      "epoch": 39.79627989371125,
      "grad_norm": 0.25292420387268066,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 44930
    },
    {
      "epoch": 39.800708591674045,
      "grad_norm": 0.22302865982055664,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 44935
    },
    {
      "epoch": 39.805137289636846,
      "grad_norm": 0.22722569108009338,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 44940
    },
    {
      "epoch": 39.80956598759965,
      "grad_norm": 0.23947083950042725,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 44945
    },
    {
      "epoch": 39.81399468556244,
      "grad_norm": 0.20041674375534058,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 44950
    },
    {
      "epoch": 39.81842338352524,
      "grad_norm": 0.23563252389431,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 44955
    },
    {
      "epoch": 39.822852081488044,
      "grad_norm": 0.2579123377799988,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 44960
    },
    {
      "epoch": 39.82728077945084,
      "grad_norm": 0.22976310551166534,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 44965
    },
    {
      "epoch": 39.83170947741364,
      "grad_norm": 0.24820353090763092,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 44970
    },
    {
      "epoch": 39.83613817537644,
      "grad_norm": 0.25754719972610474,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 44975
    },
    {
      "epoch": 39.84056687333924,
      "grad_norm": 0.2591226100921631,
      "learning_rate": 1e-05,
      "loss": 1.0233,
      "step": 44980
    },
    {
      "epoch": 39.844995571302036,
      "grad_norm": 0.2403859794139862,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 44985
    },
    {
      "epoch": 39.84942426926484,
      "grad_norm": 0.23662184178829193,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 44990
    },
    {
      "epoch": 39.85385296722764,
      "grad_norm": 0.24221083521842957,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 44995
    },
    {
      "epoch": 39.85828166519043,
      "grad_norm": 0.2532975375652313,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 45000
    },
    {
      "epoch": 39.862710363153234,
      "grad_norm": 0.27036911249160767,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 45005
    },
    {
      "epoch": 39.867139061116035,
      "grad_norm": 0.2022770345211029,
      "learning_rate": 1e-05,
      "loss": 0.9151,
      "step": 45010
    },
    {
      "epoch": 39.87156775907883,
      "grad_norm": 0.21985706686973572,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 45015
    },
    {
      "epoch": 39.87599645704163,
      "grad_norm": 0.22863242030143738,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 45020
    },
    {
      "epoch": 39.88042515500443,
      "grad_norm": 0.2397555112838745,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 45025
    },
    {
      "epoch": 39.884853852967225,
      "grad_norm": 0.21645426750183105,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 45030
    },
    {
      "epoch": 39.88928255093003,
      "grad_norm": 0.2565898597240448,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 45035
    },
    {
      "epoch": 39.89371124889283,
      "grad_norm": 0.21214188635349274,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 45040
    },
    {
      "epoch": 39.89813994685562,
      "grad_norm": 0.2289351522922516,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 45045
    },
    {
      "epoch": 39.90256864481842,
      "grad_norm": 0.23046784102916718,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 45050
    },
    {
      "epoch": 39.906997342781224,
      "grad_norm": 0.20969471335411072,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 45055
    },
    {
      "epoch": 39.91142604074402,
      "grad_norm": 0.21135160326957703,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 45060
    },
    {
      "epoch": 39.91585473870682,
      "grad_norm": 0.23426280915737152,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 45065
    },
    {
      "epoch": 39.92028343666962,
      "grad_norm": 0.24642334878444672,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 45070
    },
    {
      "epoch": 39.924712134632415,
      "grad_norm": 0.23854975402355194,
      "learning_rate": 1e-05,
      "loss": 1.036,
      "step": 45075
    },
    {
      "epoch": 39.929140832595216,
      "grad_norm": 0.24300029873847961,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 45080
    },
    {
      "epoch": 39.93356953055802,
      "grad_norm": 0.23616068065166473,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 45085
    },
    {
      "epoch": 39.93799822852081,
      "grad_norm": 0.2391427606344223,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 45090
    },
    {
      "epoch": 39.94242692648361,
      "grad_norm": 0.23161187767982483,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 45095
    },
    {
      "epoch": 39.946855624446414,
      "grad_norm": 0.23924826085567474,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 45100
    },
    {
      "epoch": 39.951284322409215,
      "grad_norm": 0.25415587425231934,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 45105
    },
    {
      "epoch": 39.95571302037201,
      "grad_norm": 0.28035393357276917,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 45110
    },
    {
      "epoch": 39.96014171833481,
      "grad_norm": 0.22854293882846832,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 45115
    },
    {
      "epoch": 39.96457041629761,
      "grad_norm": 0.20898598432540894,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 45120
    },
    {
      "epoch": 39.968999114260406,
      "grad_norm": 0.2618158757686615,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 45125
    },
    {
      "epoch": 39.97342781222321,
      "grad_norm": 0.26269733905792236,
      "learning_rate": 1e-05,
      "loss": 1.0246,
      "step": 45130
    },
    {
      "epoch": 39.97785651018601,
      "grad_norm": 0.269710898399353,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 45135
    },
    {
      "epoch": 39.9822852081488,
      "grad_norm": 0.2545463740825653,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 45140
    },
    {
      "epoch": 39.9867139061116,
      "grad_norm": 0.23376058042049408,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 45145
    },
    {
      "epoch": 39.991142604074405,
      "grad_norm": 0.2609884738922119,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 45150
    },
    {
      "epoch": 39.9955713020372,
      "grad_norm": 0.22352980077266693,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 45155
    },
    {
      "epoch": 40.0,
      "grad_norm": 0.25171589851379395,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 45160
    },
    {
      "epoch": 40.0044286979628,
      "grad_norm": 0.27114593982696533,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 45165
    },
    {
      "epoch": 40.008857395925595,
      "grad_norm": 0.27931538224220276,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 45170
    },
    {
      "epoch": 40.0132860938884,
      "grad_norm": 0.2411375492811203,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 45175
    },
    {
      "epoch": 40.0177147918512,
      "grad_norm": 0.2151385247707367,
      "learning_rate": 1e-05,
      "loss": 0.9305,
      "step": 45180
    },
    {
      "epoch": 40.02214348981399,
      "grad_norm": 0.23496542870998383,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 45185
    },
    {
      "epoch": 40.02657218777679,
      "grad_norm": 0.22750380635261536,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 45190
    },
    {
      "epoch": 40.031000885739594,
      "grad_norm": 0.2113644927740097,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 45195
    },
    {
      "epoch": 40.03542958370239,
      "grad_norm": 0.25214388966560364,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 45200
    },
    {
      "epoch": 40.03985828166519,
      "grad_norm": 0.22801996767520905,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 45205
    },
    {
      "epoch": 40.04428697962799,
      "grad_norm": 0.20237503945827484,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 45210
    },
    {
      "epoch": 40.048715677590785,
      "grad_norm": 0.24248719215393066,
      "learning_rate": 1e-05,
      "loss": 1.0204,
      "step": 45215
    },
    {
      "epoch": 40.053144375553586,
      "grad_norm": 0.21527644991874695,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 45220
    },
    {
      "epoch": 40.05757307351639,
      "grad_norm": 0.22583015263080597,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 45225
    },
    {
      "epoch": 40.06200177147919,
      "grad_norm": 0.24841605126857758,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 45230
    },
    {
      "epoch": 40.06643046944198,
      "grad_norm": 0.23951411247253418,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 45235
    },
    {
      "epoch": 40.070859167404784,
      "grad_norm": 0.27444639801979065,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 45240
    },
    {
      "epoch": 40.075287865367585,
      "grad_norm": 0.2693447768688202,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 45245
    },
    {
      "epoch": 40.07971656333038,
      "grad_norm": 0.226759672164917,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 45250
    },
    {
      "epoch": 40.08414526129318,
      "grad_norm": 0.2614608108997345,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 45255
    },
    {
      "epoch": 40.08857395925598,
      "grad_norm": 0.27981653809547424,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 45260
    },
    {
      "epoch": 40.093002657218776,
      "grad_norm": 0.24565313756465912,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 45265
    },
    {
      "epoch": 40.09743135518158,
      "grad_norm": 0.24127835035324097,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 45270
    },
    {
      "epoch": 40.10186005314438,
      "grad_norm": 0.25702059268951416,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 45275
    },
    {
      "epoch": 40.10628875110717,
      "grad_norm": 0.21388256549835205,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 45280
    },
    {
      "epoch": 40.11071744906997,
      "grad_norm": 0.26808103919029236,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 45285
    },
    {
      "epoch": 40.115146147032775,
      "grad_norm": 0.2567979395389557,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 45290
    },
    {
      "epoch": 40.11957484499557,
      "grad_norm": 0.25704577565193176,
      "learning_rate": 1e-05,
      "loss": 1.0309,
      "step": 45295
    },
    {
      "epoch": 40.12400354295837,
      "grad_norm": 0.25858715176582336,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 45300
    },
    {
      "epoch": 40.12843224092117,
      "grad_norm": 0.245374396443367,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 45305
    },
    {
      "epoch": 40.132860938883965,
      "grad_norm": 0.27113077044487,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 45310
    },
    {
      "epoch": 40.137289636846766,
      "grad_norm": 0.299653023481369,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 45315
    },
    {
      "epoch": 40.14171833480957,
      "grad_norm": 0.2712441086769104,
      "learning_rate": 1e-05,
      "loss": 1.0347,
      "step": 45320
    },
    {
      "epoch": 40.14614703277236,
      "grad_norm": 0.2394365668296814,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 45325
    },
    {
      "epoch": 40.15057573073516,
      "grad_norm": 0.24231483042240143,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 45330
    },
    {
      "epoch": 40.155004428697964,
      "grad_norm": 0.2390504628419876,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 45335
    },
    {
      "epoch": 40.15943312666076,
      "grad_norm": 0.22679764032363892,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 45340
    },
    {
      "epoch": 40.16386182462356,
      "grad_norm": 0.230784609913826,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 45345
    },
    {
      "epoch": 40.16829052258636,
      "grad_norm": 0.2817385792732239,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 45350
    },
    {
      "epoch": 40.17271922054916,
      "grad_norm": 0.24811577796936035,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 45355
    },
    {
      "epoch": 40.177147918511956,
      "grad_norm": 0.2963844835758209,
      "learning_rate": 1e-05,
      "loss": 0.9246,
      "step": 45360
    },
    {
      "epoch": 40.18157661647476,
      "grad_norm": 0.2743324935436249,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 45365
    },
    {
      "epoch": 40.18600531443756,
      "grad_norm": 0.2275199294090271,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 45370
    },
    {
      "epoch": 40.19043401240035,
      "grad_norm": 0.3600218892097473,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 45375
    },
    {
      "epoch": 40.194862710363154,
      "grad_norm": 0.2674665153026581,
      "learning_rate": 1e-05,
      "loss": 0.9072,
      "step": 45380
    },
    {
      "epoch": 40.199291408325955,
      "grad_norm": 0.2581815719604492,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 45385
    },
    {
      "epoch": 40.20372010628875,
      "grad_norm": 0.24808089435100555,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 45390
    },
    {
      "epoch": 40.20814880425155,
      "grad_norm": 0.2820883095264435,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 45395
    },
    {
      "epoch": 40.21257750221435,
      "grad_norm": 0.25388291478157043,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 45400
    },
    {
      "epoch": 40.217006200177146,
      "grad_norm": 0.2618962228298187,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 45405
    },
    {
      "epoch": 40.22143489813995,
      "grad_norm": 0.25588858127593994,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 45410
    },
    {
      "epoch": 40.22586359610275,
      "grad_norm": 0.23774223029613495,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 45415
    },
    {
      "epoch": 40.23029229406554,
      "grad_norm": 0.29644522070884705,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 45420
    },
    {
      "epoch": 40.23472099202834,
      "grad_norm": 0.24150510132312775,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 45425
    },
    {
      "epoch": 40.239149689991144,
      "grad_norm": 0.23090475797653198,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 45430
    },
    {
      "epoch": 40.24357838795394,
      "grad_norm": 0.22343404591083527,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 45435
    },
    {
      "epoch": 40.24800708591674,
      "grad_norm": 0.21493154764175415,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 45440
    },
    {
      "epoch": 40.25243578387954,
      "grad_norm": 0.26069170236587524,
      "learning_rate": 1e-05,
      "loss": 0.9279,
      "step": 45445
    },
    {
      "epoch": 40.256864481842335,
      "grad_norm": 0.24710218608379364,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 45450
    },
    {
      "epoch": 40.261293179805136,
      "grad_norm": 0.25690266489982605,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 45455
    },
    {
      "epoch": 40.26572187776794,
      "grad_norm": 0.23312783241271973,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 45460
    },
    {
      "epoch": 40.27015057573073,
      "grad_norm": 0.27539923787117004,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 45465
    },
    {
      "epoch": 40.27457927369353,
      "grad_norm": 0.20210836827754974,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 45470
    },
    {
      "epoch": 40.279007971656334,
      "grad_norm": 0.26728883385658264,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 45475
    },
    {
      "epoch": 40.283436669619135,
      "grad_norm": 0.2526356279850006,
      "learning_rate": 1e-05,
      "loss": 1.0231,
      "step": 45480
    },
    {
      "epoch": 40.28786536758193,
      "grad_norm": 0.23037558794021606,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 45485
    },
    {
      "epoch": 40.29229406554473,
      "grad_norm": 0.2466672956943512,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 45490
    },
    {
      "epoch": 40.29672276350753,
      "grad_norm": 0.22266316413879395,
      "learning_rate": 1e-05,
      "loss": 1.039,
      "step": 45495
    },
    {
      "epoch": 40.301151461470326,
      "grad_norm": 0.24533483386039734,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 45500
    },
    {
      "epoch": 40.30558015943313,
      "grad_norm": 0.2344052493572235,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 45505
    },
    {
      "epoch": 40.31000885739593,
      "grad_norm": 0.24981491267681122,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 45510
    },
    {
      "epoch": 40.31443755535872,
      "grad_norm": 0.23269736766815186,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 45515
    },
    {
      "epoch": 40.318866253321524,
      "grad_norm": 0.22463656961917877,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 45520
    },
    {
      "epoch": 40.323294951284325,
      "grad_norm": 0.2597326636314392,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 45525
    },
    {
      "epoch": 40.32772364924712,
      "grad_norm": 0.23905695974826813,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 45530
    },
    {
      "epoch": 40.33215234720992,
      "grad_norm": 0.22376959025859833,
      "learning_rate": 1e-05,
      "loss": 1.0595,
      "step": 45535
    },
    {
      "epoch": 40.33658104517272,
      "grad_norm": 0.28511202335357666,
      "learning_rate": 1e-05,
      "loss": 0.9117,
      "step": 45540
    },
    {
      "epoch": 40.341009743135515,
      "grad_norm": 0.2111230343580246,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 45545
    },
    {
      "epoch": 40.34543844109832,
      "grad_norm": 0.23622570931911469,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 45550
    },
    {
      "epoch": 40.34986713906112,
      "grad_norm": 0.257079154253006,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 45555
    },
    {
      "epoch": 40.35429583702391,
      "grad_norm": 0.25046560168266296,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 45560
    },
    {
      "epoch": 40.35872453498671,
      "grad_norm": 0.24342671036720276,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 45565
    },
    {
      "epoch": 40.363153232949514,
      "grad_norm": 0.24034452438354492,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 45570
    },
    {
      "epoch": 40.36758193091231,
      "grad_norm": 0.22056357562541962,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 45575
    },
    {
      "epoch": 40.37201062887511,
      "grad_norm": 0.26125818490982056,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 45580
    },
    {
      "epoch": 40.37643932683791,
      "grad_norm": 0.23894613981246948,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 45585
    },
    {
      "epoch": 40.380868024800705,
      "grad_norm": 0.3081366717815399,
      "learning_rate": 1e-05,
      "loss": 1.0554,
      "step": 45590
    },
    {
      "epoch": 40.385296722763506,
      "grad_norm": 0.299436092376709,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 45595
    },
    {
      "epoch": 40.38972542072631,
      "grad_norm": 0.23816809058189392,
      "learning_rate": 1e-05,
      "loss": 0.9162,
      "step": 45600
    },
    {
      "epoch": 40.39415411868911,
      "grad_norm": 0.262315034866333,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 45605
    },
    {
      "epoch": 40.3985828166519,
      "grad_norm": 0.24232378602027893,
      "learning_rate": 1e-05,
      "loss": 1.0427,
      "step": 45610
    },
    {
      "epoch": 40.403011514614704,
      "grad_norm": 0.21910181641578674,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 45615
    },
    {
      "epoch": 40.407440212577505,
      "grad_norm": 0.2129480093717575,
      "learning_rate": 1e-05,
      "loss": 0.8943,
      "step": 45620
    },
    {
      "epoch": 40.4118689105403,
      "grad_norm": 0.23968534171581268,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 45625
    },
    {
      "epoch": 40.4162976085031,
      "grad_norm": 0.24089445173740387,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 45630
    },
    {
      "epoch": 40.4207263064659,
      "grad_norm": 0.2468075007200241,
      "learning_rate": 1e-05,
      "loss": 0.9348,
      "step": 45635
    },
    {
      "epoch": 40.425155004428696,
      "grad_norm": 0.2679606080055237,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 45640
    },
    {
      "epoch": 40.4295837023915,
      "grad_norm": 0.23917730152606964,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 45645
    },
    {
      "epoch": 40.4340124003543,
      "grad_norm": 0.24306005239486694,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 45650
    },
    {
      "epoch": 40.43844109831709,
      "grad_norm": 0.20310689508914948,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 45655
    },
    {
      "epoch": 40.44286979627989,
      "grad_norm": 0.230911985039711,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 45660
    },
    {
      "epoch": 40.447298494242695,
      "grad_norm": 0.2517818510532379,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 45665
    },
    {
      "epoch": 40.45172719220549,
      "grad_norm": 0.236632838845253,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 45670
    },
    {
      "epoch": 40.45615589016829,
      "grad_norm": 0.24472634494304657,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 45675
    },
    {
      "epoch": 40.46058458813109,
      "grad_norm": 0.26011422276496887,
      "learning_rate": 1e-05,
      "loss": 0.924,
      "step": 45680
    },
    {
      "epoch": 40.465013286093885,
      "grad_norm": 0.22335882484912872,
      "learning_rate": 1e-05,
      "loss": 1.0818,
      "step": 45685
    },
    {
      "epoch": 40.46944198405669,
      "grad_norm": 0.22788962721824646,
      "learning_rate": 1e-05,
      "loss": 1.0175,
      "step": 45690
    },
    {
      "epoch": 40.47387068201949,
      "grad_norm": 0.211948961019516,
      "learning_rate": 1e-05,
      "loss": 1.0448,
      "step": 45695
    },
    {
      "epoch": 40.47829937998228,
      "grad_norm": 0.23885637521743774,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 45700
    },
    {
      "epoch": 40.48272807794508,
      "grad_norm": 0.20586352050304413,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 45705
    },
    {
      "epoch": 40.487156775907884,
      "grad_norm": 0.2192770391702652,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 45710
    },
    {
      "epoch": 40.491585473870686,
      "grad_norm": 0.22180071473121643,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 45715
    },
    {
      "epoch": 40.49601417183348,
      "grad_norm": 0.2147900015115738,
      "learning_rate": 1e-05,
      "loss": 0.9232,
      "step": 45720
    },
    {
      "epoch": 40.50044286979628,
      "grad_norm": 0.2357667237520218,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 45725
    },
    {
      "epoch": 40.50487156775908,
      "grad_norm": 0.27332115173339844,
      "learning_rate": 1e-05,
      "loss": 1.0309,
      "step": 45730
    },
    {
      "epoch": 40.509300265721876,
      "grad_norm": 0.23799879848957062,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 45735
    },
    {
      "epoch": 40.51372896368468,
      "grad_norm": 0.24116341769695282,
      "learning_rate": 1e-05,
      "loss": 1.0344,
      "step": 45740
    },
    {
      "epoch": 40.51815766164748,
      "grad_norm": 0.2594386041164398,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 45745
    },
    {
      "epoch": 40.52258635961027,
      "grad_norm": 0.22958551347255707,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 45750
    },
    {
      "epoch": 40.527015057573074,
      "grad_norm": 0.24454988539218903,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 45755
    },
    {
      "epoch": 40.531443755535875,
      "grad_norm": 0.2570692002773285,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 45760
    },
    {
      "epoch": 40.53587245349867,
      "grad_norm": 0.21165458858013153,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 45765
    },
    {
      "epoch": 40.54030115146147,
      "grad_norm": 0.27606502175331116,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 45770
    },
    {
      "epoch": 40.54472984942427,
      "grad_norm": 0.19953720271587372,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 45775
    },
    {
      "epoch": 40.549158547387066,
      "grad_norm": 0.23468749225139618,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 45780
    },
    {
      "epoch": 40.55358724534987,
      "grad_norm": 0.22528351843357086,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 45785
    },
    {
      "epoch": 40.55801594331267,
      "grad_norm": 0.20409393310546875,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 45790
    },
    {
      "epoch": 40.56244464127546,
      "grad_norm": 0.3014966547489166,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 45795
    },
    {
      "epoch": 40.56687333923826,
      "grad_norm": 0.26219120621681213,
      "learning_rate": 1e-05,
      "loss": 1.0427,
      "step": 45800
    },
    {
      "epoch": 40.571302037201065,
      "grad_norm": 0.3385133147239685,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 45805
    },
    {
      "epoch": 40.57573073516386,
      "grad_norm": 0.22322909533977509,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 45810
    },
    {
      "epoch": 40.58015943312666,
      "grad_norm": 3.132784605026245,
      "learning_rate": 1e-05,
      "loss": 1.0196,
      "step": 45815
    },
    {
      "epoch": 40.58458813108946,
      "grad_norm": 0.26024380326271057,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 45820
    },
    {
      "epoch": 40.589016829052255,
      "grad_norm": 0.24028509855270386,
      "learning_rate": 1e-05,
      "loss": 0.9329,
      "step": 45825
    },
    {
      "epoch": 40.59344552701506,
      "grad_norm": 0.2672039270401001,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 45830
    },
    {
      "epoch": 40.59787422497786,
      "grad_norm": 0.30118849873542786,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 45835
    },
    {
      "epoch": 40.60230292294066,
      "grad_norm": 0.24499301612377167,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 45840
    },
    {
      "epoch": 40.60673162090345,
      "grad_norm": 0.2375156283378601,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 45845
    },
    {
      "epoch": 40.611160318866254,
      "grad_norm": 0.20132434368133545,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 45850
    },
    {
      "epoch": 40.615589016829055,
      "grad_norm": 0.2787694036960602,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 45855
    },
    {
      "epoch": 40.62001771479185,
      "grad_norm": 0.2578490972518921,
      "learning_rate": 1e-05,
      "loss": 0.9218,
      "step": 45860
    },
    {
      "epoch": 40.62444641275465,
      "grad_norm": 0.26339927315711975,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 45865
    },
    {
      "epoch": 40.62887511071745,
      "grad_norm": 0.23578201234340668,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 45870
    },
    {
      "epoch": 40.633303808680246,
      "grad_norm": 0.25542744994163513,
      "learning_rate": 1e-05,
      "loss": 0.9144,
      "step": 45875
    },
    {
      "epoch": 40.63773250664305,
      "grad_norm": 0.2786359488964081,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 45880
    },
    {
      "epoch": 40.64216120460585,
      "grad_norm": 0.2705237865447998,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 45885
    },
    {
      "epoch": 40.64658990256864,
      "grad_norm": 0.2309940904378891,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 45890
    },
    {
      "epoch": 40.651018600531444,
      "grad_norm": 0.2568571865558624,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 45895
    },
    {
      "epoch": 40.655447298494245,
      "grad_norm": 0.20694012939929962,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 45900
    },
    {
      "epoch": 40.65987599645704,
      "grad_norm": 0.259566992521286,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 45905
    },
    {
      "epoch": 40.66430469441984,
      "grad_norm": 0.27784380316734314,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 45910
    },
    {
      "epoch": 40.66873339238264,
      "grad_norm": 0.2501930892467499,
      "learning_rate": 1e-05,
      "loss": 0.927,
      "step": 45915
    },
    {
      "epoch": 40.673162090345436,
      "grad_norm": 0.22544093430042267,
      "learning_rate": 1e-05,
      "loss": 1.0296,
      "step": 45920
    },
    {
      "epoch": 40.67759078830824,
      "grad_norm": 0.3154388666152954,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 45925
    },
    {
      "epoch": 40.68201948627104,
      "grad_norm": 0.20174498856067657,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 45930
    },
    {
      "epoch": 40.68644818423383,
      "grad_norm": 0.24484170973300934,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 45935
    },
    {
      "epoch": 40.69087688219663,
      "grad_norm": 0.22526822984218597,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 45940
    },
    {
      "epoch": 40.695305580159435,
      "grad_norm": 0.2262207567691803,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 45945
    },
    {
      "epoch": 40.69973427812223,
      "grad_norm": 0.2243526428937912,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 45950
    },
    {
      "epoch": 40.70416297608503,
      "grad_norm": 0.2703973650932312,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 45955
    },
    {
      "epoch": 40.70859167404783,
      "grad_norm": 0.2283085584640503,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 45960
    },
    {
      "epoch": 40.71302037201063,
      "grad_norm": 0.2553483247756958,
      "learning_rate": 1e-05,
      "loss": 1.0253,
      "step": 45965
    },
    {
      "epoch": 40.717449069973426,
      "grad_norm": 0.24605821073055267,
      "learning_rate": 1e-05,
      "loss": 0.9065,
      "step": 45970
    },
    {
      "epoch": 40.72187776793623,
      "grad_norm": 0.21523445844650269,
      "learning_rate": 1e-05,
      "loss": 0.926,
      "step": 45975
    },
    {
      "epoch": 40.72630646589903,
      "grad_norm": 0.2937183678150177,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 45980
    },
    {
      "epoch": 40.73073516386182,
      "grad_norm": 0.2621420621871948,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 45985
    },
    {
      "epoch": 40.735163861824624,
      "grad_norm": 0.23658692836761475,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 45990
    },
    {
      "epoch": 40.739592559787425,
      "grad_norm": 0.2258015275001526,
      "learning_rate": 1e-05,
      "loss": 1.0693,
      "step": 45995
    },
    {
      "epoch": 40.74402125775022,
      "grad_norm": 0.24768316745758057,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 46000
    },
    {
      "epoch": 40.74844995571302,
      "grad_norm": 0.19924908876419067,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 46005
    },
    {
      "epoch": 40.75287865367582,
      "grad_norm": 0.28765299916267395,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 46010
    },
    {
      "epoch": 40.757307351638616,
      "grad_norm": 0.26688098907470703,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 46015
    },
    {
      "epoch": 40.76173604960142,
      "grad_norm": 0.2504536807537079,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 46020
    },
    {
      "epoch": 40.76616474756422,
      "grad_norm": 0.25557008385658264,
      "learning_rate": 1e-05,
      "loss": 1.0221,
      "step": 46025
    },
    {
      "epoch": 40.77059344552701,
      "grad_norm": 0.25966283679008484,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 46030
    },
    {
      "epoch": 40.775022143489814,
      "grad_norm": 0.2781597077846527,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 46035
    },
    {
      "epoch": 40.779450841452615,
      "grad_norm": 0.28727951645851135,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 46040
    },
    {
      "epoch": 40.78387953941541,
      "grad_norm": 0.22551856935024261,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 46045
    },
    {
      "epoch": 40.78830823737821,
      "grad_norm": 0.19982650876045227,
      "learning_rate": 1e-05,
      "loss": 0.9267,
      "step": 46050
    },
    {
      "epoch": 40.79273693534101,
      "grad_norm": 0.2332746833562851,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 46055
    },
    {
      "epoch": 40.797165633303806,
      "grad_norm": 0.2241213172674179,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 46060
    },
    {
      "epoch": 40.80159433126661,
      "grad_norm": 0.24330177903175354,
      "learning_rate": 1e-05,
      "loss": 0.9156,
      "step": 46065
    },
    {
      "epoch": 40.80602302922941,
      "grad_norm": 0.2048812061548233,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 46070
    },
    {
      "epoch": 40.8104517271922,
      "grad_norm": 0.23944677412509918,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 46075
    },
    {
      "epoch": 40.814880425155,
      "grad_norm": 0.2650429308414459,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 46080
    },
    {
      "epoch": 40.819309123117804,
      "grad_norm": 0.2534618079662323,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 46085
    },
    {
      "epoch": 40.823737821080606,
      "grad_norm": 0.2590779960155487,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 46090
    },
    {
      "epoch": 40.8281665190434,
      "grad_norm": 0.241688534617424,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 46095
    },
    {
      "epoch": 40.8325952170062,
      "grad_norm": 0.2734193503856659,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 46100
    },
    {
      "epoch": 40.837023914969,
      "grad_norm": 0.21250741183757782,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 46105
    },
    {
      "epoch": 40.841452612931796,
      "grad_norm": 0.2548217475414276,
      "learning_rate": 1e-05,
      "loss": 0.9118,
      "step": 46110
    },
    {
      "epoch": 40.8458813108946,
      "grad_norm": 0.2470664530992508,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 46115
    },
    {
      "epoch": 40.8503100088574,
      "grad_norm": 0.22192998230457306,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 46120
    },
    {
      "epoch": 40.85473870682019,
      "grad_norm": 0.24488016963005066,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 46125
    },
    {
      "epoch": 40.859167404782994,
      "grad_norm": 0.24498888850212097,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 46130
    },
    {
      "epoch": 40.863596102745795,
      "grad_norm": 0.25724685192108154,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 46135
    },
    {
      "epoch": 40.86802480070859,
      "grad_norm": 0.25136297941207886,
      "learning_rate": 1e-05,
      "loss": 0.9346,
      "step": 46140
    },
    {
      "epoch": 40.87245349867139,
      "grad_norm": 0.21111196279525757,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 46145
    },
    {
      "epoch": 40.87688219663419,
      "grad_norm": 0.25202807784080505,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 46150
    },
    {
      "epoch": 40.881310894596986,
      "grad_norm": 0.2228946089744568,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 46155
    },
    {
      "epoch": 40.88573959255979,
      "grad_norm": 0.2557080090045929,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 46160
    },
    {
      "epoch": 40.89016829052259,
      "grad_norm": 0.24860051274299622,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 46165
    },
    {
      "epoch": 40.89459698848538,
      "grad_norm": 0.23795831203460693,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 46170
    },
    {
      "epoch": 40.899025686448184,
      "grad_norm": 0.27710476517677307,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 46175
    },
    {
      "epoch": 40.903454384410985,
      "grad_norm": 0.22504942119121552,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 46180
    },
    {
      "epoch": 40.90788308237378,
      "grad_norm": 0.2174450010061264,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 46185
    },
    {
      "epoch": 40.91231178033658,
      "grad_norm": 0.21438786387443542,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 46190
    },
    {
      "epoch": 40.91674047829938,
      "grad_norm": 0.23867042362689972,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 46195
    },
    {
      "epoch": 40.921169176262175,
      "grad_norm": 0.21805495023727417,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 46200
    },
    {
      "epoch": 40.92559787422498,
      "grad_norm": 0.18959949910640717,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 46205
    },
    {
      "epoch": 40.93002657218778,
      "grad_norm": 0.2630828619003296,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 46210
    },
    {
      "epoch": 40.93445527015058,
      "grad_norm": 0.255587100982666,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 46215
    },
    {
      "epoch": 40.93888396811337,
      "grad_norm": 0.2215876579284668,
      "learning_rate": 1e-05,
      "loss": 0.912,
      "step": 46220
    },
    {
      "epoch": 40.943312666076174,
      "grad_norm": 0.28492793440818787,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 46225
    },
    {
      "epoch": 40.947741364038976,
      "grad_norm": 0.2775213420391083,
      "learning_rate": 1e-05,
      "loss": 1.0361,
      "step": 46230
    },
    {
      "epoch": 40.95217006200177,
      "grad_norm": 0.19882459938526154,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 46235
    },
    {
      "epoch": 40.95659875996457,
      "grad_norm": 0.21408990025520325,
      "learning_rate": 1e-05,
      "loss": 1.0204,
      "step": 46240
    },
    {
      "epoch": 40.96102745792737,
      "grad_norm": 0.22075635194778442,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 46245
    },
    {
      "epoch": 40.965456155890166,
      "grad_norm": 0.26780733466148376,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 46250
    },
    {
      "epoch": 40.96988485385297,
      "grad_norm": 0.19951346516609192,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 46255
    },
    {
      "epoch": 40.97431355181577,
      "grad_norm": 0.23445382714271545,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 46260
    },
    {
      "epoch": 40.97874224977856,
      "grad_norm": 0.24156714975833893,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 46265
    },
    {
      "epoch": 40.983170947741364,
      "grad_norm": 0.22810764610767365,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 46270
    },
    {
      "epoch": 40.987599645704165,
      "grad_norm": 0.25293776392936707,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 46275
    },
    {
      "epoch": 40.99202834366696,
      "grad_norm": 0.2452917844057083,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 46280
    },
    {
      "epoch": 40.99645704162976,
      "grad_norm": 0.25986042618751526,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 46285
    },
    {
      "epoch": 41.00088573959256,
      "grad_norm": 0.255265474319458,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 46290
    },
    {
      "epoch": 41.005314437555356,
      "grad_norm": 0.22851619124412537,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 46295
    },
    {
      "epoch": 41.00974313551816,
      "grad_norm": 0.2514009475708008,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 46300
    },
    {
      "epoch": 41.01417183348096,
      "grad_norm": 0.2622789442539215,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 46305
    },
    {
      "epoch": 41.01860053144375,
      "grad_norm": 0.23823486268520355,
      "learning_rate": 1e-05,
      "loss": 1.0336,
      "step": 46310
    },
    {
      "epoch": 41.02302922940655,
      "grad_norm": 0.23992542922496796,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 46315
    },
    {
      "epoch": 41.027457927369355,
      "grad_norm": 0.2484361082315445,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 46320
    },
    {
      "epoch": 41.03188662533215,
      "grad_norm": 0.27358919382095337,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 46325
    },
    {
      "epoch": 41.03631532329495,
      "grad_norm": 0.25942134857177734,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 46330
    },
    {
      "epoch": 41.04074402125775,
      "grad_norm": 0.2241154909133911,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 46335
    },
    {
      "epoch": 41.04517271922055,
      "grad_norm": 0.22613605856895447,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 46340
    },
    {
      "epoch": 41.04960141718335,
      "grad_norm": 0.29719963669776917,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 46345
    },
    {
      "epoch": 41.05403011514615,
      "grad_norm": 0.3122774064540863,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 46350
    },
    {
      "epoch": 41.05845881310895,
      "grad_norm": 0.2574634850025177,
      "learning_rate": 1e-05,
      "loss": 0.9348,
      "step": 46355
    },
    {
      "epoch": 41.06288751107174,
      "grad_norm": 0.2685568928718567,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 46360
    },
    {
      "epoch": 41.067316209034544,
      "grad_norm": 0.2701598107814789,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 46365
    },
    {
      "epoch": 41.071744906997345,
      "grad_norm": 0.2244083136320114,
      "learning_rate": 1e-05,
      "loss": 0.9149,
      "step": 46370
    },
    {
      "epoch": 41.07617360496014,
      "grad_norm": 0.2717266380786896,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 46375
    },
    {
      "epoch": 41.08060230292294,
      "grad_norm": 0.23584267497062683,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 46380
    },
    {
      "epoch": 41.08503100088574,
      "grad_norm": 0.27556976675987244,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 46385
    },
    {
      "epoch": 41.089459698848536,
      "grad_norm": 0.23604941368103027,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 46390
    },
    {
      "epoch": 41.09388839681134,
      "grad_norm": 0.2693665623664856,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 46395
    },
    {
      "epoch": 41.09831709477414,
      "grad_norm": 0.2334788739681244,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 46400
    },
    {
      "epoch": 41.10274579273693,
      "grad_norm": 0.2547294795513153,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 46405
    },
    {
      "epoch": 41.107174490699734,
      "grad_norm": 0.23960816860198975,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 46410
    },
    {
      "epoch": 41.111603188662535,
      "grad_norm": 0.1880325824022293,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 46415
    },
    {
      "epoch": 41.11603188662533,
      "grad_norm": 0.21492323279380798,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 46420
    },
    {
      "epoch": 41.12046058458813,
      "grad_norm": 0.23511657118797302,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 46425
    },
    {
      "epoch": 41.12488928255093,
      "grad_norm": 0.2740682363510132,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 46430
    },
    {
      "epoch": 41.129317980513726,
      "grad_norm": 0.22582025825977325,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 46435
    },
    {
      "epoch": 41.13374667847653,
      "grad_norm": 0.2642294466495514,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 46440
    },
    {
      "epoch": 41.13817537643933,
      "grad_norm": 0.27966028451919556,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 46445
    },
    {
      "epoch": 41.14260407440213,
      "grad_norm": 0.22969523072242737,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 46450
    },
    {
      "epoch": 41.14703277236492,
      "grad_norm": 0.24469727277755737,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 46455
    },
    {
      "epoch": 41.151461470327725,
      "grad_norm": 0.20409634709358215,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 46460
    },
    {
      "epoch": 41.155890168290526,
      "grad_norm": 0.24594646692276,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 46465
    },
    {
      "epoch": 41.16031886625332,
      "grad_norm": 0.2107960283756256,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 46470
    },
    {
      "epoch": 41.16474756421612,
      "grad_norm": 0.23458872735500336,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 46475
    },
    {
      "epoch": 41.16917626217892,
      "grad_norm": 0.22560983896255493,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 46480
    },
    {
      "epoch": 41.173604960141716,
      "grad_norm": 0.2523638904094696,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 46485
    },
    {
      "epoch": 41.17803365810452,
      "grad_norm": 0.23163332045078278,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 46490
    },
    {
      "epoch": 41.18246235606732,
      "grad_norm": 0.2533573508262634,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 46495
    },
    {
      "epoch": 41.18689105403011,
      "grad_norm": 0.20747590065002441,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 46500
    },
    {
      "epoch": 41.191319751992914,
      "grad_norm": 0.2377549111843109,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 46505
    },
    {
      "epoch": 41.195748449955715,
      "grad_norm": 0.24137932062149048,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 46510
    },
    {
      "epoch": 41.20017714791851,
      "grad_norm": 0.2392067313194275,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 46515
    },
    {
      "epoch": 41.20460584588131,
      "grad_norm": 0.24320088326931,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 46520
    },
    {
      "epoch": 41.20903454384411,
      "grad_norm": 0.2416033297777176,
      "learning_rate": 1e-05,
      "loss": 1.033,
      "step": 46525
    },
    {
      "epoch": 41.213463241806906,
      "grad_norm": 0.2187662422657013,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 46530
    },
    {
      "epoch": 41.21789193976971,
      "grad_norm": 0.21655720472335815,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 46535
    },
    {
      "epoch": 41.22232063773251,
      "grad_norm": 0.2829791009426117,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 46540
    },
    {
      "epoch": 41.2267493356953,
      "grad_norm": 0.2016926407814026,
      "learning_rate": 1e-05,
      "loss": 1.0187,
      "step": 46545
    },
    {
      "epoch": 41.231178033658104,
      "grad_norm": 0.24520066380500793,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 46550
    },
    {
      "epoch": 41.235606731620905,
      "grad_norm": 0.2519649267196655,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 46555
    },
    {
      "epoch": 41.2400354295837,
      "grad_norm": 0.2173595130443573,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 46560
    },
    {
      "epoch": 41.2444641275465,
      "grad_norm": 0.2718411386013031,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 46565
    },
    {
      "epoch": 41.2488928255093,
      "grad_norm": 0.29615840315818787,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 46570
    },
    {
      "epoch": 41.2533215234721,
      "grad_norm": 0.21959950029850006,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 46575
    },
    {
      "epoch": 41.2577502214349,
      "grad_norm": 0.2201385349035263,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 46580
    },
    {
      "epoch": 41.2621789193977,
      "grad_norm": 0.2673913240432739,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 46585
    },
    {
      "epoch": 41.2666076173605,
      "grad_norm": 0.31370651721954346,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 46590
    },
    {
      "epoch": 41.27103631532329,
      "grad_norm": 0.22648300230503082,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 46595
    },
    {
      "epoch": 41.275465013286095,
      "grad_norm": 0.2525435984134674,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 46600
    },
    {
      "epoch": 41.279893711248896,
      "grad_norm": 0.2291669100522995,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 46605
    },
    {
      "epoch": 41.28432240921169,
      "grad_norm": 0.23526166379451752,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 46610
    },
    {
      "epoch": 41.28875110717449,
      "grad_norm": 0.2228817492723465,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 46615
    },
    {
      "epoch": 41.29317980513729,
      "grad_norm": 0.23752111196517944,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 46620
    },
    {
      "epoch": 41.297608503100086,
      "grad_norm": 0.2538985311985016,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 46625
    },
    {
      "epoch": 41.30203720106289,
      "grad_norm": 0.20848622918128967,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 46630
    },
    {
      "epoch": 41.30646589902569,
      "grad_norm": 0.20615798234939575,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 46635
    },
    {
      "epoch": 41.31089459698848,
      "grad_norm": 0.30227023363113403,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 46640
    },
    {
      "epoch": 41.315323294951284,
      "grad_norm": 0.2978127896785736,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 46645
    },
    {
      "epoch": 41.319751992914085,
      "grad_norm": 0.25392451882362366,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 46650
    },
    {
      "epoch": 41.32418069087688,
      "grad_norm": 0.23451435565948486,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 46655
    },
    {
      "epoch": 41.32860938883968,
      "grad_norm": 0.22238238155841827,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 46660
    },
    {
      "epoch": 41.33303808680248,
      "grad_norm": 0.2178657203912735,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 46665
    },
    {
      "epoch": 41.337466784765276,
      "grad_norm": 0.19798831641674042,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 46670
    },
    {
      "epoch": 41.34189548272808,
      "grad_norm": 0.26017433404922485,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 46675
    },
    {
      "epoch": 41.34632418069088,
      "grad_norm": 0.2545863389968872,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 46680
    },
    {
      "epoch": 41.35075287865367,
      "grad_norm": 0.2588142156600952,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 46685
    },
    {
      "epoch": 41.355181576616474,
      "grad_norm": 0.2616218328475952,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 46690
    },
    {
      "epoch": 41.359610274579275,
      "grad_norm": 0.3297959864139557,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 46695
    },
    {
      "epoch": 41.364038972542076,
      "grad_norm": 0.24753610789775848,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 46700
    },
    {
      "epoch": 41.36846767050487,
      "grad_norm": 0.28048592805862427,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 46705
    },
    {
      "epoch": 41.37289636846767,
      "grad_norm": 0.22273565828800201,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 46710
    },
    {
      "epoch": 41.37732506643047,
      "grad_norm": 0.24823904037475586,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 46715
    },
    {
      "epoch": 41.38175376439327,
      "grad_norm": 0.24420718848705292,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 46720
    },
    {
      "epoch": 41.38618246235607,
      "grad_norm": 0.2724437415599823,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 46725
    },
    {
      "epoch": 41.39061116031887,
      "grad_norm": 0.250664621591568,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 46730
    },
    {
      "epoch": 41.39503985828166,
      "grad_norm": 0.20065896213054657,
      "learning_rate": 1e-05,
      "loss": 1.0323,
      "step": 46735
    },
    {
      "epoch": 41.399468556244464,
      "grad_norm": 0.26644062995910645,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 46740
    },
    {
      "epoch": 41.403897254207266,
      "grad_norm": 0.20965181291103363,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 46745
    },
    {
      "epoch": 41.40832595217006,
      "grad_norm": 0.22881768643856049,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 46750
    },
    {
      "epoch": 41.41275465013286,
      "grad_norm": 0.25017255544662476,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 46755
    },
    {
      "epoch": 41.41718334809566,
      "grad_norm": 0.222843736410141,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 46760
    },
    {
      "epoch": 41.421612046058456,
      "grad_norm": 0.24611032009124756,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 46765
    },
    {
      "epoch": 41.42604074402126,
      "grad_norm": 0.23981942236423492,
      "learning_rate": 1e-05,
      "loss": 0.9138,
      "step": 46770
    },
    {
      "epoch": 41.43046944198406,
      "grad_norm": 0.24829858541488647,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 46775
    },
    {
      "epoch": 41.43489813994685,
      "grad_norm": 0.2527877390384674,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 46780
    },
    {
      "epoch": 41.439326837909654,
      "grad_norm": 0.21177981793880463,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 46785
    },
    {
      "epoch": 41.443755535872455,
      "grad_norm": 0.21250608563423157,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 46790
    },
    {
      "epoch": 41.44818423383525,
      "grad_norm": 0.21410933136940002,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 46795
    },
    {
      "epoch": 41.45261293179805,
      "grad_norm": 0.2529192864894867,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 46800
    },
    {
      "epoch": 41.45704162976085,
      "grad_norm": 0.23823131620883942,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 46805
    },
    {
      "epoch": 41.461470327723646,
      "grad_norm": 0.24617500603199005,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 46810
    },
    {
      "epoch": 41.46589902568645,
      "grad_norm": 0.2661856710910797,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 46815
    },
    {
      "epoch": 41.47032772364925,
      "grad_norm": 0.2571755647659302,
      "learning_rate": 1e-05,
      "loss": 1.0337,
      "step": 46820
    },
    {
      "epoch": 41.47475642161205,
      "grad_norm": 0.20305150747299194,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 46825
    },
    {
      "epoch": 41.479185119574844,
      "grad_norm": 0.24163159728050232,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 46830
    },
    {
      "epoch": 41.483613817537645,
      "grad_norm": 0.2743411660194397,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 46835
    },
    {
      "epoch": 41.488042515500446,
      "grad_norm": 0.24505265057086945,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 46840
    },
    {
      "epoch": 41.49247121346324,
      "grad_norm": 0.24050122499465942,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 46845
    },
    {
      "epoch": 41.49689991142604,
      "grad_norm": 0.2746000289916992,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 46850
    },
    {
      "epoch": 41.50132860938884,
      "grad_norm": 0.22860535979270935,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 46855
    },
    {
      "epoch": 41.50575730735164,
      "grad_norm": 0.21988466382026672,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 46860
    },
    {
      "epoch": 41.51018600531444,
      "grad_norm": 0.24702921509742737,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 46865
    },
    {
      "epoch": 41.51461470327724,
      "grad_norm": 0.24761199951171875,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 46870
    },
    {
      "epoch": 41.51904340124003,
      "grad_norm": 0.2534991204738617,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 46875
    },
    {
      "epoch": 41.523472099202834,
      "grad_norm": 0.23372985422611237,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 46880
    },
    {
      "epoch": 41.527900797165636,
      "grad_norm": 0.2518102526664734,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 46885
    },
    {
      "epoch": 41.53232949512843,
      "grad_norm": 0.24182449281215668,
      "learning_rate": 1e-05,
      "loss": 0.9305,
      "step": 46890
    },
    {
      "epoch": 41.53675819309123,
      "grad_norm": 0.23864220082759857,
      "learning_rate": 1e-05,
      "loss": 1.018,
      "step": 46895
    },
    {
      "epoch": 41.54118689105403,
      "grad_norm": 0.23089124262332916,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 46900
    },
    {
      "epoch": 41.545615589016826,
      "grad_norm": 0.26446375250816345,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 46905
    },
    {
      "epoch": 41.55004428697963,
      "grad_norm": 0.23237252235412598,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 46910
    },
    {
      "epoch": 41.55447298494243,
      "grad_norm": 0.22160682082176208,
      "learning_rate": 1e-05,
      "loss": 1.03,
      "step": 46915
    },
    {
      "epoch": 41.55890168290522,
      "grad_norm": 0.22599604725837708,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 46920
    },
    {
      "epoch": 41.563330380868024,
      "grad_norm": 0.2181016355752945,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 46925
    },
    {
      "epoch": 41.567759078830825,
      "grad_norm": 0.28839850425720215,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 46930
    },
    {
      "epoch": 41.57218777679362,
      "grad_norm": 0.21832482516765594,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 46935
    },
    {
      "epoch": 41.57661647475642,
      "grad_norm": 0.23229339718818665,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 46940
    },
    {
      "epoch": 41.58104517271922,
      "grad_norm": 0.22854207456111908,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 46945
    },
    {
      "epoch": 41.58547387068202,
      "grad_norm": 0.2785719633102417,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 46950
    },
    {
      "epoch": 41.58990256864482,
      "grad_norm": 0.26977190375328064,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 46955
    },
    {
      "epoch": 41.59433126660762,
      "grad_norm": 0.23604361712932587,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 46960
    },
    {
      "epoch": 41.59875996457042,
      "grad_norm": 0.19699259102344513,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 46965
    },
    {
      "epoch": 41.60318866253321,
      "grad_norm": 0.2620716691017151,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 46970
    },
    {
      "epoch": 41.607617360496015,
      "grad_norm": 0.2337796688079834,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 46975
    },
    {
      "epoch": 41.612046058458816,
      "grad_norm": 0.22643381357192993,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 46980
    },
    {
      "epoch": 41.61647475642161,
      "grad_norm": 0.2471427023410797,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 46985
    },
    {
      "epoch": 41.62090345438441,
      "grad_norm": 0.27139613032341003,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 46990
    },
    {
      "epoch": 41.62533215234721,
      "grad_norm": 0.26607316732406616,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 46995
    },
    {
      "epoch": 41.62976085031001,
      "grad_norm": 0.24246305227279663,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 47000
    },
    {
      "epoch": 41.63418954827281,
      "grad_norm": 0.2571573853492737,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 47005
    },
    {
      "epoch": 41.63861824623561,
      "grad_norm": 0.20895442366600037,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 47010
    },
    {
      "epoch": 41.6430469441984,
      "grad_norm": 0.2300744354724884,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 47015
    },
    {
      "epoch": 41.647475642161204,
      "grad_norm": 0.23912738263607025,
      "learning_rate": 1e-05,
      "loss": 0.9273,
      "step": 47020
    },
    {
      "epoch": 41.651904340124005,
      "grad_norm": 0.20212675631046295,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 47025
    },
    {
      "epoch": 41.6563330380868,
      "grad_norm": 0.24596139788627625,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 47030
    },
    {
      "epoch": 41.6607617360496,
      "grad_norm": 0.27843809127807617,
      "learning_rate": 1e-05,
      "loss": 1.0358,
      "step": 47035
    },
    {
      "epoch": 41.6651904340124,
      "grad_norm": 0.2495543658733368,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 47040
    },
    {
      "epoch": 41.669619131975196,
      "grad_norm": 0.24879902601242065,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 47045
    },
    {
      "epoch": 41.674047829938,
      "grad_norm": 0.20944470167160034,
      "learning_rate": 1e-05,
      "loss": 1.0376,
      "step": 47050
    },
    {
      "epoch": 41.6784765279008,
      "grad_norm": 0.26936009526252747,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 47055
    },
    {
      "epoch": 41.68290522586359,
      "grad_norm": 0.28627660870552063,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 47060
    },
    {
      "epoch": 41.687333923826394,
      "grad_norm": 0.2799301743507385,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 47065
    },
    {
      "epoch": 41.691762621789195,
      "grad_norm": 0.2388470321893692,
      "learning_rate": 1e-05,
      "loss": 1.0303,
      "step": 47070
    },
    {
      "epoch": 41.696191319751996,
      "grad_norm": 0.19623346626758575,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 47075
    },
    {
      "epoch": 41.70062001771479,
      "grad_norm": 0.24366050958633423,
      "learning_rate": 1e-05,
      "loss": 0.9087,
      "step": 47080
    },
    {
      "epoch": 41.70504871567759,
      "grad_norm": 0.2991557717323303,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 47085
    },
    {
      "epoch": 41.70947741364039,
      "grad_norm": 0.21608667075634003,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 47090
    },
    {
      "epoch": 41.71390611160319,
      "grad_norm": 0.23013699054718018,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 47095
    },
    {
      "epoch": 41.71833480956599,
      "grad_norm": 0.21956031024456024,
      "learning_rate": 1e-05,
      "loss": 1.0843,
      "step": 47100
    },
    {
      "epoch": 41.72276350752879,
      "grad_norm": 0.2651800215244293,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 47105
    },
    {
      "epoch": 41.72719220549158,
      "grad_norm": 0.22979427874088287,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 47110
    },
    {
      "epoch": 41.731620903454385,
      "grad_norm": 0.21967464685440063,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 47115
    },
    {
      "epoch": 41.736049601417186,
      "grad_norm": 0.24914321303367615,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 47120
    },
    {
      "epoch": 41.74047829937998,
      "grad_norm": 0.2248576134443283,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 47125
    },
    {
      "epoch": 41.74490699734278,
      "grad_norm": 0.2356075793504715,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 47130
    },
    {
      "epoch": 41.74933569530558,
      "grad_norm": 0.2672087550163269,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 47135
    },
    {
      "epoch": 41.753764393268376,
      "grad_norm": 0.290438711643219,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 47140
    },
    {
      "epoch": 41.75819309123118,
      "grad_norm": 0.25266802310943604,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 47145
    },
    {
      "epoch": 41.76262178919398,
      "grad_norm": 0.23994064331054688,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 47150
    },
    {
      "epoch": 41.76705048715677,
      "grad_norm": 0.24676162004470825,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 47155
    },
    {
      "epoch": 41.771479185119574,
      "grad_norm": 0.23655670881271362,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 47160
    },
    {
      "epoch": 41.775907883082375,
      "grad_norm": 0.21515819430351257,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 47165
    },
    {
      "epoch": 41.78033658104517,
      "grad_norm": 0.2658790051937103,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 47170
    },
    {
      "epoch": 41.78476527900797,
      "grad_norm": 0.26729583740234375,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 47175
    },
    {
      "epoch": 41.78919397697077,
      "grad_norm": 0.24590493738651276,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 47180
    },
    {
      "epoch": 41.79362267493357,
      "grad_norm": 0.23929621279239655,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 47185
    },
    {
      "epoch": 41.79805137289637,
      "grad_norm": 0.22001193463802338,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 47190
    },
    {
      "epoch": 41.80248007085917,
      "grad_norm": 0.24116064608097076,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 47195
    },
    {
      "epoch": 41.80690876882197,
      "grad_norm": 0.20740506052970886,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 47200
    },
    {
      "epoch": 41.811337466784764,
      "grad_norm": 0.23944616317749023,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 47205
    },
    {
      "epoch": 41.815766164747565,
      "grad_norm": 0.24343454837799072,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 47210
    },
    {
      "epoch": 41.820194862710366,
      "grad_norm": 0.2211264818906784,
      "learning_rate": 1e-05,
      "loss": 1.0187,
      "step": 47215
    },
    {
      "epoch": 41.82462356067316,
      "grad_norm": 0.2685020864009857,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 47220
    },
    {
      "epoch": 41.82905225863596,
      "grad_norm": 0.24459992349147797,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 47225
    },
    {
      "epoch": 41.83348095659876,
      "grad_norm": 0.21922965347766876,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 47230
    },
    {
      "epoch": 41.83790965456156,
      "grad_norm": 0.238324373960495,
      "learning_rate": 1e-05,
      "loss": 0.9061,
      "step": 47235
    },
    {
      "epoch": 41.84233835252436,
      "grad_norm": 0.22449564933776855,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 47240
    },
    {
      "epoch": 41.84676705048716,
      "grad_norm": 0.20336972177028656,
      "learning_rate": 1e-05,
      "loss": 1.0357,
      "step": 47245
    },
    {
      "epoch": 41.85119574844995,
      "grad_norm": 0.2726646065711975,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 47250
    },
    {
      "epoch": 41.855624446412754,
      "grad_norm": 0.23348622024059296,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 47255
    },
    {
      "epoch": 41.860053144375556,
      "grad_norm": 0.24239620566368103,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 47260
    },
    {
      "epoch": 41.86448184233835,
      "grad_norm": 0.21317002177238464,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 47265
    },
    {
      "epoch": 41.86891054030115,
      "grad_norm": 0.2872420847415924,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 47270
    },
    {
      "epoch": 41.87333923826395,
      "grad_norm": 0.31914740800857544,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 47275
    },
    {
      "epoch": 41.877767936226746,
      "grad_norm": 0.2558329105377197,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 47280
    },
    {
      "epoch": 41.88219663418955,
      "grad_norm": 0.21335332095623016,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 47285
    },
    {
      "epoch": 41.88662533215235,
      "grad_norm": 0.24892573058605194,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 47290
    },
    {
      "epoch": 41.89105403011514,
      "grad_norm": 0.24181511998176575,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 47295
    },
    {
      "epoch": 41.895482728077944,
      "grad_norm": 0.2369491308927536,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 47300
    },
    {
      "epoch": 41.899911426040745,
      "grad_norm": 0.23254704475402832,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 47305
    },
    {
      "epoch": 41.90434012400354,
      "grad_norm": 0.22457970678806305,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 47310
    },
    {
      "epoch": 41.90876882196634,
      "grad_norm": 0.2847231924533844,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 47315
    },
    {
      "epoch": 41.91319751992914,
      "grad_norm": 0.2645552456378937,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 47320
    },
    {
      "epoch": 41.91762621789194,
      "grad_norm": 0.23353981971740723,
      "learning_rate": 1e-05,
      "loss": 1.0181,
      "step": 47325
    },
    {
      "epoch": 41.92205491585474,
      "grad_norm": 0.25501686334609985,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 47330
    },
    {
      "epoch": 41.92648361381754,
      "grad_norm": 0.2782838046550751,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 47335
    },
    {
      "epoch": 41.93091231178034,
      "grad_norm": 0.24164332449436188,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 47340
    },
    {
      "epoch": 41.935341009743134,
      "grad_norm": 0.22004728019237518,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 47345
    },
    {
      "epoch": 41.939769707705935,
      "grad_norm": 0.2279643565416336,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 47350
    },
    {
      "epoch": 41.944198405668736,
      "grad_norm": 0.26370400190353394,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 47355
    },
    {
      "epoch": 41.94862710363153,
      "grad_norm": 0.22065910696983337,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 47360
    },
    {
      "epoch": 41.95305580159433,
      "grad_norm": 0.22977013885974884,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 47365
    },
    {
      "epoch": 41.95748449955713,
      "grad_norm": 0.2535136938095093,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 47370
    },
    {
      "epoch": 41.96191319751993,
      "grad_norm": 0.2568632960319519,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 47375
    },
    {
      "epoch": 41.96634189548273,
      "grad_norm": 0.23541945219039917,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 47380
    },
    {
      "epoch": 41.97077059344553,
      "grad_norm": 0.28623607754707336,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 47385
    },
    {
      "epoch": 41.97519929140832,
      "grad_norm": 0.2416672557592392,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 47390
    },
    {
      "epoch": 41.979627989371124,
      "grad_norm": 0.24725818634033203,
      "learning_rate": 1e-05,
      "loss": 1.0389,
      "step": 47395
    },
    {
      "epoch": 41.984056687333926,
      "grad_norm": 0.24128949642181396,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 47400
    },
    {
      "epoch": 41.98848538529672,
      "grad_norm": 0.23148638010025024,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 47405
    },
    {
      "epoch": 41.99291408325952,
      "grad_norm": 0.2444465309381485,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 47410
    },
    {
      "epoch": 41.99734278122232,
      "grad_norm": 0.2789562940597534,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 47415
    },
    {
      "epoch": 42.001771479185116,
      "grad_norm": 0.23833613097667694,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 47420
    },
    {
      "epoch": 42.00620017714792,
      "grad_norm": 0.26139411330223083,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 47425
    },
    {
      "epoch": 42.01062887511072,
      "grad_norm": 0.2773147523403168,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 47430
    },
    {
      "epoch": 42.01505757307352,
      "grad_norm": 0.2189108431339264,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 47435
    },
    {
      "epoch": 42.019486271036314,
      "grad_norm": 0.20204530656337738,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 47440
    },
    {
      "epoch": 42.023914968999115,
      "grad_norm": 0.2207849770784378,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 47445
    },
    {
      "epoch": 42.028343666961916,
      "grad_norm": 0.28971290588378906,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 47450
    },
    {
      "epoch": 42.03277236492471,
      "grad_norm": 0.2455006092786789,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 47455
    },
    {
      "epoch": 42.03720106288751,
      "grad_norm": 0.2976686656475067,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 47460
    },
    {
      "epoch": 42.04162976085031,
      "grad_norm": 0.2381303608417511,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 47465
    },
    {
      "epoch": 42.04605845881311,
      "grad_norm": 0.26214805245399475,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 47470
    },
    {
      "epoch": 42.05048715677591,
      "grad_norm": 0.23552539944648743,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 47475
    },
    {
      "epoch": 42.05491585473871,
      "grad_norm": 0.23731185495853424,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 47480
    },
    {
      "epoch": 42.0593445527015,
      "grad_norm": 0.2604026198387146,
      "learning_rate": 1e-05,
      "loss": 0.927,
      "step": 47485
    },
    {
      "epoch": 42.063773250664305,
      "grad_norm": 0.29012829065322876,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 47490
    },
    {
      "epoch": 42.068201948627106,
      "grad_norm": 0.24425168335437775,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 47495
    },
    {
      "epoch": 42.0726306465899,
      "grad_norm": 0.26090261340141296,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 47500
    },
    {
      "epoch": 42.0770593445527,
      "grad_norm": 0.23037965595722198,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 47505
    },
    {
      "epoch": 42.0814880425155,
      "grad_norm": 0.2707884609699249,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 47510
    },
    {
      "epoch": 42.0859167404783,
      "grad_norm": 0.2334928959608078,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 47515
    },
    {
      "epoch": 42.0903454384411,
      "grad_norm": 0.22577998042106628,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 47520
    },
    {
      "epoch": 42.0947741364039,
      "grad_norm": 0.24617010354995728,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 47525
    },
    {
      "epoch": 42.09920283436669,
      "grad_norm": 0.27806687355041504,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 47530
    },
    {
      "epoch": 42.103631532329494,
      "grad_norm": 0.2106403261423111,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 47535
    },
    {
      "epoch": 42.108060230292296,
      "grad_norm": 0.2449704259634018,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 47540
    },
    {
      "epoch": 42.11248892825509,
      "grad_norm": 0.21838924288749695,
      "learning_rate": 1e-05,
      "loss": 1.0085,
      "step": 47545
    },
    {
      "epoch": 42.11691762621789,
      "grad_norm": 0.23402725160121918,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 47550
    },
    {
      "epoch": 42.12134632418069,
      "grad_norm": 0.2137393206357956,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 47555
    },
    {
      "epoch": 42.12577502214349,
      "grad_norm": 0.2771582007408142,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 47560
    },
    {
      "epoch": 42.13020372010629,
      "grad_norm": 0.22381120920181274,
      "learning_rate": 1e-05,
      "loss": 0.9126,
      "step": 47565
    },
    {
      "epoch": 42.13463241806909,
      "grad_norm": 0.24643754959106445,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 47570
    },
    {
      "epoch": 42.13906111603189,
      "grad_norm": 0.22045674920082092,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 47575
    },
    {
      "epoch": 42.143489813994684,
      "grad_norm": 0.2493283599615097,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 47580
    },
    {
      "epoch": 42.147918511957485,
      "grad_norm": 0.3196389377117157,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 47585
    },
    {
      "epoch": 42.152347209920286,
      "grad_norm": 0.3015509843826294,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 47590
    },
    {
      "epoch": 42.15677590788308,
      "grad_norm": 0.25217247009277344,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 47595
    },
    {
      "epoch": 42.16120460584588,
      "grad_norm": 0.24276064336299896,
      "learning_rate": 1e-05,
      "loss": 0.9258,
      "step": 47600
    },
    {
      "epoch": 42.16563330380868,
      "grad_norm": 0.21076928079128265,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 47605
    },
    {
      "epoch": 42.17006200177148,
      "grad_norm": 0.21807684004306793,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 47610
    },
    {
      "epoch": 42.17449069973428,
      "grad_norm": 0.25889700651168823,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 47615
    },
    {
      "epoch": 42.17891939769708,
      "grad_norm": 0.2158796638250351,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 47620
    },
    {
      "epoch": 42.18334809565987,
      "grad_norm": 0.24467463791370392,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 47625
    },
    {
      "epoch": 42.187776793622675,
      "grad_norm": 0.22895513474941254,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 47630
    },
    {
      "epoch": 42.192205491585476,
      "grad_norm": 0.2406184822320938,
      "learning_rate": 1e-05,
      "loss": 0.863,
      "step": 47635
    },
    {
      "epoch": 42.19663418954827,
      "grad_norm": 0.22785741090774536,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 47640
    },
    {
      "epoch": 42.20106288751107,
      "grad_norm": 0.20653823018074036,
      "learning_rate": 1e-05,
      "loss": 1.0427,
      "step": 47645
    },
    {
      "epoch": 42.20549158547387,
      "grad_norm": 0.22851921617984772,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 47650
    },
    {
      "epoch": 42.20992028343667,
      "grad_norm": 0.22391921281814575,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 47655
    },
    {
      "epoch": 42.21434898139947,
      "grad_norm": 0.2410077452659607,
      "learning_rate": 1e-05,
      "loss": 0.9159,
      "step": 47660
    },
    {
      "epoch": 42.21877767936227,
      "grad_norm": 0.2572147250175476,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 47665
    },
    {
      "epoch": 42.22320637732506,
      "grad_norm": 0.23231694102287292,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 47670
    },
    {
      "epoch": 42.227635075287864,
      "grad_norm": 0.2209526002407074,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 47675
    },
    {
      "epoch": 42.232063773250665,
      "grad_norm": 0.23461313545703888,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 47680
    },
    {
      "epoch": 42.23649247121347,
      "grad_norm": 0.2990756034851074,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 47685
    },
    {
      "epoch": 42.24092116917626,
      "grad_norm": 0.2768782079219818,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 47690
    },
    {
      "epoch": 42.24534986713906,
      "grad_norm": 0.22221550345420837,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 47695
    },
    {
      "epoch": 42.24977856510186,
      "grad_norm": 0.2784070074558258,
      "learning_rate": 1e-05,
      "loss": 1.0797,
      "step": 47700
    },
    {
      "epoch": 42.25420726306466,
      "grad_norm": 0.22987860441207886,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 47705
    },
    {
      "epoch": 42.25863596102746,
      "grad_norm": 0.30230504274368286,
      "learning_rate": 1e-05,
      "loss": 1.0465,
      "step": 47710
    },
    {
      "epoch": 42.26306465899026,
      "grad_norm": 0.22940903902053833,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 47715
    },
    {
      "epoch": 42.267493356953054,
      "grad_norm": 0.24579615890979767,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 47720
    },
    {
      "epoch": 42.271922054915855,
      "grad_norm": 0.2738560140132904,
      "learning_rate": 1e-05,
      "loss": 1.0234,
      "step": 47725
    },
    {
      "epoch": 42.276350752878656,
      "grad_norm": 0.2987234592437744,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 47730
    },
    {
      "epoch": 42.28077945084145,
      "grad_norm": 0.21220698952674866,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 47735
    },
    {
      "epoch": 42.28520814880425,
      "grad_norm": 0.22221818566322327,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 47740
    },
    {
      "epoch": 42.28963684676705,
      "grad_norm": 0.22247502207756042,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 47745
    },
    {
      "epoch": 42.29406554472985,
      "grad_norm": 0.2361268401145935,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 47750
    },
    {
      "epoch": 42.29849424269265,
      "grad_norm": 0.2383842170238495,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 47755
    },
    {
      "epoch": 42.30292294065545,
      "grad_norm": 0.221732035279274,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 47760
    },
    {
      "epoch": 42.30735163861824,
      "grad_norm": 0.25873833894729614,
      "learning_rate": 1e-05,
      "loss": 1.0272,
      "step": 47765
    },
    {
      "epoch": 42.311780336581045,
      "grad_norm": 0.21419979631900787,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 47770
    },
    {
      "epoch": 42.316209034543846,
      "grad_norm": 0.2709173262119293,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 47775
    },
    {
      "epoch": 42.32063773250664,
      "grad_norm": 0.228935107588768,
      "learning_rate": 1e-05,
      "loss": 1.0217,
      "step": 47780
    },
    {
      "epoch": 42.32506643046944,
      "grad_norm": 0.2365936040878296,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 47785
    },
    {
      "epoch": 42.32949512843224,
      "grad_norm": 0.22589780390262604,
      "learning_rate": 1e-05,
      "loss": 0.9215,
      "step": 47790
    },
    {
      "epoch": 42.333923826395036,
      "grad_norm": 0.2352641224861145,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 47795
    },
    {
      "epoch": 42.33835252435784,
      "grad_norm": 0.2637138366699219,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 47800
    },
    {
      "epoch": 42.34278122232064,
      "grad_norm": 0.25251033902168274,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 47805
    },
    {
      "epoch": 42.34720992028344,
      "grad_norm": 0.25087177753448486,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 47810
    },
    {
      "epoch": 42.351638618246234,
      "grad_norm": 0.26238250732421875,
      "learning_rate": 1e-05,
      "loss": 0.9051,
      "step": 47815
    },
    {
      "epoch": 42.356067316209035,
      "grad_norm": 0.2517799437046051,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 47820
    },
    {
      "epoch": 42.36049601417184,
      "grad_norm": 0.27053138613700867,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 47825
    },
    {
      "epoch": 42.36492471213463,
      "grad_norm": 0.25898489356040955,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 47830
    },
    {
      "epoch": 42.36935341009743,
      "grad_norm": 0.22970426082611084,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 47835
    },
    {
      "epoch": 42.37378210806023,
      "grad_norm": 0.22901815176010132,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 47840
    },
    {
      "epoch": 42.37821080602303,
      "grad_norm": 0.2113620787858963,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 47845
    },
    {
      "epoch": 42.38263950398583,
      "grad_norm": 0.24157191812992096,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 47850
    },
    {
      "epoch": 42.38706820194863,
      "grad_norm": 0.25221458077430725,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 47855
    },
    {
      "epoch": 42.391496899911424,
      "grad_norm": 0.22772446274757385,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 47860
    },
    {
      "epoch": 42.395925597874225,
      "grad_norm": 0.21750780940055847,
      "learning_rate": 1e-05,
      "loss": 0.9432,
      "step": 47865
    },
    {
      "epoch": 42.400354295837026,
      "grad_norm": 0.2313288152217865,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 47870
    },
    {
      "epoch": 42.40478299379982,
      "grad_norm": 0.26093485951423645,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 47875
    },
    {
      "epoch": 42.40921169176262,
      "grad_norm": 0.23907415568828583,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 47880
    },
    {
      "epoch": 42.41364038972542,
      "grad_norm": 0.2607579827308655,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 47885
    },
    {
      "epoch": 42.41806908768822,
      "grad_norm": 0.22821852564811707,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 47890
    },
    {
      "epoch": 42.42249778565102,
      "grad_norm": 0.209013432264328,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 47895
    },
    {
      "epoch": 42.42692648361382,
      "grad_norm": 0.22845275700092316,
      "learning_rate": 1e-05,
      "loss": 0.9192,
      "step": 47900
    },
    {
      "epoch": 42.43135518157661,
      "grad_norm": 0.21673406660556793,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 47905
    },
    {
      "epoch": 42.435783879539414,
      "grad_norm": 0.22812657058238983,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 47910
    },
    {
      "epoch": 42.440212577502216,
      "grad_norm": 0.239319309592247,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 47915
    },
    {
      "epoch": 42.44464127546502,
      "grad_norm": 0.21700850129127502,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 47920
    },
    {
      "epoch": 42.44906997342781,
      "grad_norm": 0.26806920766830444,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 47925
    },
    {
      "epoch": 42.45349867139061,
      "grad_norm": 0.27256736159324646,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 47930
    },
    {
      "epoch": 42.45792736935341,
      "grad_norm": 0.23432551324367523,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 47935
    },
    {
      "epoch": 42.46235606731621,
      "grad_norm": 0.23915445804595947,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 47940
    },
    {
      "epoch": 42.46678476527901,
      "grad_norm": 0.28664112091064453,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 47945
    },
    {
      "epoch": 42.47121346324181,
      "grad_norm": 0.22265088558197021,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 47950
    },
    {
      "epoch": 42.475642161204604,
      "grad_norm": 0.23771323263645172,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 47955
    },
    {
      "epoch": 42.480070859167405,
      "grad_norm": 0.25379106402397156,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 47960
    },
    {
      "epoch": 42.484499557130206,
      "grad_norm": 0.2564697861671448,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 47965
    },
    {
      "epoch": 42.488928255093,
      "grad_norm": 0.2633035182952881,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 47970
    },
    {
      "epoch": 42.4933569530558,
      "grad_norm": 0.28511014580726624,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 47975
    },
    {
      "epoch": 42.4977856510186,
      "grad_norm": 0.30362117290496826,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 47980
    },
    {
      "epoch": 42.5022143489814,
      "grad_norm": 0.2917020618915558,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 47985
    },
    {
      "epoch": 42.5066430469442,
      "grad_norm": 0.2303413599729538,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 47990
    },
    {
      "epoch": 42.511071744907,
      "grad_norm": 0.2529556453227997,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 47995
    },
    {
      "epoch": 42.515500442869794,
      "grad_norm": 0.23653215169906616,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 48000
    },
    {
      "epoch": 42.519929140832595,
      "grad_norm": 0.2300071120262146,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 48005
    },
    {
      "epoch": 42.524357838795396,
      "grad_norm": 0.24020180106163025,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 48010
    },
    {
      "epoch": 42.52878653675819,
      "grad_norm": 0.26055529713630676,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 48015
    },
    {
      "epoch": 42.53321523472099,
      "grad_norm": 0.23629535734653473,
      "learning_rate": 1e-05,
      "loss": 0.9079,
      "step": 48020
    },
    {
      "epoch": 42.53764393268379,
      "grad_norm": 0.22570590674877167,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 48025
    },
    {
      "epoch": 42.54207263064659,
      "grad_norm": 0.21297363936901093,
      "learning_rate": 1e-05,
      "loss": 1.0286,
      "step": 48030
    },
    {
      "epoch": 42.54650132860939,
      "grad_norm": 0.24435722827911377,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 48035
    },
    {
      "epoch": 42.55093002657219,
      "grad_norm": 0.21717384457588196,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 48040
    },
    {
      "epoch": 42.55535872453498,
      "grad_norm": 0.2731130123138428,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 48045
    },
    {
      "epoch": 42.559787422497784,
      "grad_norm": 0.27486127614974976,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 48050
    },
    {
      "epoch": 42.564216120460586,
      "grad_norm": 0.24913103878498077,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 48055
    },
    {
      "epoch": 42.56864481842339,
      "grad_norm": 0.2720363140106201,
      "learning_rate": 1e-05,
      "loss": 1.0523,
      "step": 48060
    },
    {
      "epoch": 42.57307351638618,
      "grad_norm": 0.2424292266368866,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 48065
    },
    {
      "epoch": 42.57750221434898,
      "grad_norm": 0.26393935084342957,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 48070
    },
    {
      "epoch": 42.58193091231178,
      "grad_norm": 0.2699756920337677,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 48075
    },
    {
      "epoch": 42.58635961027458,
      "grad_norm": 0.2601114809513092,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 48080
    },
    {
      "epoch": 42.59078830823738,
      "grad_norm": 0.2980334460735321,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 48085
    },
    {
      "epoch": 42.59521700620018,
      "grad_norm": 0.2408342808485031,
      "learning_rate": 1e-05,
      "loss": 1.0461,
      "step": 48090
    },
    {
      "epoch": 42.599645704162974,
      "grad_norm": 0.2292022556066513,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 48095
    },
    {
      "epoch": 42.604074402125775,
      "grad_norm": 0.2296779453754425,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 48100
    },
    {
      "epoch": 42.608503100088576,
      "grad_norm": 0.21083290874958038,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 48105
    },
    {
      "epoch": 42.61293179805137,
      "grad_norm": 0.2390083372592926,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 48110
    },
    {
      "epoch": 42.61736049601417,
      "grad_norm": 0.23495185375213623,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 48115
    },
    {
      "epoch": 42.62178919397697,
      "grad_norm": 0.2481905072927475,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 48120
    },
    {
      "epoch": 42.62621789193977,
      "grad_norm": 0.254972904920578,
      "learning_rate": 1e-05,
      "loss": 1.0497,
      "step": 48125
    },
    {
      "epoch": 42.63064658990257,
      "grad_norm": 0.2681357264518738,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 48130
    },
    {
      "epoch": 42.63507528786537,
      "grad_norm": 0.23341068625450134,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 48135
    },
    {
      "epoch": 42.63950398582816,
      "grad_norm": 0.2620530128479004,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 48140
    },
    {
      "epoch": 42.643932683790965,
      "grad_norm": 0.22171328961849213,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 48145
    },
    {
      "epoch": 42.648361381753766,
      "grad_norm": 0.22144654393196106,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 48150
    },
    {
      "epoch": 42.65279007971656,
      "grad_norm": 0.3005954921245575,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 48155
    },
    {
      "epoch": 42.65721877767936,
      "grad_norm": 0.2432592660188675,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 48160
    },
    {
      "epoch": 42.66164747564216,
      "grad_norm": 0.2259034365415573,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 48165
    },
    {
      "epoch": 42.666076173604964,
      "grad_norm": 0.25215601921081543,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 48170
    },
    {
      "epoch": 42.67050487156776,
      "grad_norm": 0.2721492350101471,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 48175
    },
    {
      "epoch": 42.67493356953056,
      "grad_norm": 0.21123404800891876,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 48180
    },
    {
      "epoch": 42.67936226749336,
      "grad_norm": 0.2151031792163849,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 48185
    },
    {
      "epoch": 42.683790965456154,
      "grad_norm": 0.22641023993492126,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 48190
    },
    {
      "epoch": 42.688219663418955,
      "grad_norm": 0.3041313588619232,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 48195
    },
    {
      "epoch": 42.69264836138176,
      "grad_norm": 0.24665305018424988,
      "learning_rate": 1e-05,
      "loss": 0.9251,
      "step": 48200
    },
    {
      "epoch": 42.69707705934455,
      "grad_norm": 0.27113863825798035,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 48205
    },
    {
      "epoch": 42.70150575730735,
      "grad_norm": 0.24964125454425812,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 48210
    },
    {
      "epoch": 42.70593445527015,
      "grad_norm": 0.24823129177093506,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 48215
    },
    {
      "epoch": 42.71036315323295,
      "grad_norm": 0.24530290067195892,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 48220
    },
    {
      "epoch": 42.71479185119575,
      "grad_norm": 0.28934958577156067,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 48225
    },
    {
      "epoch": 42.71922054915855,
      "grad_norm": 0.24014613032341003,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 48230
    },
    {
      "epoch": 42.723649247121344,
      "grad_norm": 0.27102765440940857,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 48235
    },
    {
      "epoch": 42.728077945084145,
      "grad_norm": 0.2744861841201782,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 48240
    },
    {
      "epoch": 42.732506643046946,
      "grad_norm": 0.23346906900405884,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 48245
    },
    {
      "epoch": 42.73693534100974,
      "grad_norm": 0.2375788688659668,
      "learning_rate": 1e-05,
      "loss": 0.8964,
      "step": 48250
    },
    {
      "epoch": 42.74136403897254,
      "grad_norm": 0.25419747829437256,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 48255
    },
    {
      "epoch": 42.74579273693534,
      "grad_norm": 0.339944988489151,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 48260
    },
    {
      "epoch": 42.75022143489814,
      "grad_norm": 0.26090502738952637,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 48265
    },
    {
      "epoch": 42.75465013286094,
      "grad_norm": 0.2445857673883438,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 48270
    },
    {
      "epoch": 42.75907883082374,
      "grad_norm": 0.25257250666618347,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 48275
    },
    {
      "epoch": 42.76350752878653,
      "grad_norm": 0.26221832633018494,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 48280
    },
    {
      "epoch": 42.767936226749335,
      "grad_norm": 0.2826215624809265,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 48285
    },
    {
      "epoch": 42.772364924712136,
      "grad_norm": 0.24633489549160004,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 48290
    },
    {
      "epoch": 42.77679362267494,
      "grad_norm": 0.2523399889469147,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 48295
    },
    {
      "epoch": 42.78122232063773,
      "grad_norm": 0.23993588984012604,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 48300
    },
    {
      "epoch": 42.78565101860053,
      "grad_norm": 0.24645495414733887,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 48305
    },
    {
      "epoch": 42.79007971656333,
      "grad_norm": 0.2708395719528198,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 48310
    },
    {
      "epoch": 42.79450841452613,
      "grad_norm": 0.24099871516227722,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 48315
    },
    {
      "epoch": 42.79893711248893,
      "grad_norm": 0.25115475058555603,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 48320
    },
    {
      "epoch": 42.80336581045173,
      "grad_norm": 0.25144869089126587,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 48325
    },
    {
      "epoch": 42.807794508414524,
      "grad_norm": 0.21381622552871704,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 48330
    },
    {
      "epoch": 42.812223206377325,
      "grad_norm": 0.23303286731243134,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 48335
    },
    {
      "epoch": 42.81665190434013,
      "grad_norm": 0.19767539203166962,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 48340
    },
    {
      "epoch": 42.82108060230292,
      "grad_norm": 0.29785776138305664,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 48345
    },
    {
      "epoch": 42.82550930026572,
      "grad_norm": 0.23110856115818024,
      "learning_rate": 1e-05,
      "loss": 0.883,
      "step": 48350
    },
    {
      "epoch": 42.82993799822852,
      "grad_norm": 0.27621594071388245,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 48355
    },
    {
      "epoch": 42.83436669619132,
      "grad_norm": 0.2695445120334625,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 48360
    },
    {
      "epoch": 42.83879539415412,
      "grad_norm": 0.25764527916908264,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 48365
    },
    {
      "epoch": 42.84322409211692,
      "grad_norm": 0.2536025941371918,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 48370
    },
    {
      "epoch": 42.847652790079714,
      "grad_norm": 0.28541821241378784,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 48375
    },
    {
      "epoch": 42.852081488042515,
      "grad_norm": 0.24982883036136627,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 48380
    },
    {
      "epoch": 42.856510186005316,
      "grad_norm": 0.23558664321899414,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 48385
    },
    {
      "epoch": 42.86093888396811,
      "grad_norm": 0.24839410185813904,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 48390
    },
    {
      "epoch": 42.86536758193091,
      "grad_norm": 0.2402290403842926,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 48395
    },
    {
      "epoch": 42.86979627989371,
      "grad_norm": 0.21183155477046967,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 48400
    },
    {
      "epoch": 42.87422497785651,
      "grad_norm": 0.21802331507205963,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 48405
    },
    {
      "epoch": 42.87865367581931,
      "grad_norm": 0.19899477064609528,
      "learning_rate": 1e-05,
      "loss": 1.0564,
      "step": 48410
    },
    {
      "epoch": 42.88308237378211,
      "grad_norm": 0.25269243121147156,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 48415
    },
    {
      "epoch": 42.88751107174491,
      "grad_norm": 0.25537973642349243,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 48420
    },
    {
      "epoch": 42.891939769707704,
      "grad_norm": 0.21263942122459412,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 48425
    },
    {
      "epoch": 42.896368467670506,
      "grad_norm": 0.22235079109668732,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 48430
    },
    {
      "epoch": 42.90079716563331,
      "grad_norm": 0.22799000144004822,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 48435
    },
    {
      "epoch": 42.9052258635961,
      "grad_norm": 0.21052277088165283,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 48440
    },
    {
      "epoch": 42.9096545615589,
      "grad_norm": 0.31826117634773254,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 48445
    },
    {
      "epoch": 42.9140832595217,
      "grad_norm": 0.2638929486274719,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 48450
    },
    {
      "epoch": 42.9185119574845,
      "grad_norm": 0.2635389268398285,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 48455
    },
    {
      "epoch": 42.9229406554473,
      "grad_norm": 0.22469651699066162,
      "learning_rate": 1e-05,
      "loss": 1.0448,
      "step": 48460
    },
    {
      "epoch": 42.9273693534101,
      "grad_norm": 0.2392352670431137,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 48465
    },
    {
      "epoch": 42.931798051372894,
      "grad_norm": 0.2513946294784546,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 48470
    },
    {
      "epoch": 42.936226749335695,
      "grad_norm": 0.2198619693517685,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 48475
    },
    {
      "epoch": 42.9406554472985,
      "grad_norm": 0.3263920843601227,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 48480
    },
    {
      "epoch": 42.94508414526129,
      "grad_norm": 0.26297903060913086,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 48485
    },
    {
      "epoch": 42.94951284322409,
      "grad_norm": 0.23898695409297943,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 48490
    },
    {
      "epoch": 42.95394154118689,
      "grad_norm": 0.2855616807937622,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 48495
    },
    {
      "epoch": 42.95837023914969,
      "grad_norm": 0.26662975549697876,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 48500
    },
    {
      "epoch": 42.96279893711249,
      "grad_norm": 0.22201858460903168,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 48505
    },
    {
      "epoch": 42.96722763507529,
      "grad_norm": 0.318624347448349,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 48510
    },
    {
      "epoch": 42.971656333038084,
      "grad_norm": 0.254317969083786,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 48515
    },
    {
      "epoch": 42.976085031000885,
      "grad_norm": 0.2682293653488159,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 48520
    },
    {
      "epoch": 42.980513728963686,
      "grad_norm": 0.22450757026672363,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 48525
    },
    {
      "epoch": 42.98494242692648,
      "grad_norm": 0.23072472214698792,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 48530
    },
    {
      "epoch": 42.98937112488928,
      "grad_norm": 0.2410205453634262,
      "learning_rate": 1e-05,
      "loss": 0.894,
      "step": 48535
    },
    {
      "epoch": 42.99379982285208,
      "grad_norm": 0.21451111137866974,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 48540
    },
    {
      "epoch": 42.998228520814884,
      "grad_norm": 0.2471478134393692,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 48545
    },
    {
      "epoch": 43.00265721877768,
      "grad_norm": 0.24021108448505402,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 48550
    },
    {
      "epoch": 43.00708591674048,
      "grad_norm": 0.2044842541217804,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 48555
    },
    {
      "epoch": 43.01151461470328,
      "grad_norm": 0.2626218795776367,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 48560
    },
    {
      "epoch": 43.015943312666074,
      "grad_norm": 0.19783316552639008,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 48565
    },
    {
      "epoch": 43.020372010628876,
      "grad_norm": 0.23617658019065857,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 48570
    },
    {
      "epoch": 43.02480070859168,
      "grad_norm": 0.22546809911727905,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 48575
    },
    {
      "epoch": 43.02922940655447,
      "grad_norm": 0.23770242929458618,
      "learning_rate": 1e-05,
      "loss": 1.0435,
      "step": 48580
    },
    {
      "epoch": 43.03365810451727,
      "grad_norm": 0.2602790594100952,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 48585
    },
    {
      "epoch": 43.03808680248007,
      "grad_norm": 0.22907041013240814,
      "learning_rate": 1e-05,
      "loss": 1.0221,
      "step": 48590
    },
    {
      "epoch": 43.04251550044287,
      "grad_norm": 0.25333258509635925,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 48595
    },
    {
      "epoch": 43.04694419840567,
      "grad_norm": 0.2384660691022873,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 48600
    },
    {
      "epoch": 43.05137289636847,
      "grad_norm": 0.2633801996707916,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 48605
    },
    {
      "epoch": 43.055801594331264,
      "grad_norm": 0.2746524214744568,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 48610
    },
    {
      "epoch": 43.060230292294065,
      "grad_norm": 0.30926743149757385,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 48615
    },
    {
      "epoch": 43.064658990256866,
      "grad_norm": 0.24673625826835632,
      "learning_rate": 1e-05,
      "loss": 0.9522,
      "step": 48620
    },
    {
      "epoch": 43.06908768821966,
      "grad_norm": 0.24810610711574554,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 48625
    },
    {
      "epoch": 43.07351638618246,
      "grad_norm": 0.2859072685241699,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 48630
    },
    {
      "epoch": 43.07794508414526,
      "grad_norm": 0.2518060505390167,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 48635
    },
    {
      "epoch": 43.08237378210806,
      "grad_norm": 0.26977333426475525,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 48640
    },
    {
      "epoch": 43.08680248007086,
      "grad_norm": 0.26873618364334106,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 48645
    },
    {
      "epoch": 43.09123117803366,
      "grad_norm": 0.23334825038909912,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 48650
    },
    {
      "epoch": 43.09565987599645,
      "grad_norm": 0.24947552382946014,
      "learning_rate": 1e-05,
      "loss": 0.904,
      "step": 48655
    },
    {
      "epoch": 43.100088573959255,
      "grad_norm": 0.2441912442445755,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 48660
    },
    {
      "epoch": 43.104517271922056,
      "grad_norm": 0.26713937520980835,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 48665
    },
    {
      "epoch": 43.10894596988486,
      "grad_norm": 0.23476269841194153,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 48670
    },
    {
      "epoch": 43.11337466784765,
      "grad_norm": 0.22453244030475616,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 48675
    },
    {
      "epoch": 43.11780336581045,
      "grad_norm": 0.2722662687301636,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 48680
    },
    {
      "epoch": 43.122232063773254,
      "grad_norm": 0.24403497576713562,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 48685
    },
    {
      "epoch": 43.12666076173605,
      "grad_norm": 0.23619729280471802,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 48690
    },
    {
      "epoch": 43.13108945969885,
      "grad_norm": 0.2588210105895996,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 48695
    },
    {
      "epoch": 43.13551815766165,
      "grad_norm": 0.2949660122394562,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 48700
    },
    {
      "epoch": 43.139946855624444,
      "grad_norm": 0.2258627861738205,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 48705
    },
    {
      "epoch": 43.144375553587246,
      "grad_norm": 0.2544408142566681,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 48710
    },
    {
      "epoch": 43.14880425155005,
      "grad_norm": 0.24393408000469208,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 48715
    },
    {
      "epoch": 43.15323294951284,
      "grad_norm": 0.2392231673002243,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 48720
    },
    {
      "epoch": 43.15766164747564,
      "grad_norm": 0.23040099442005157,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 48725
    },
    {
      "epoch": 43.16209034543844,
      "grad_norm": 0.2195800244808197,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 48730
    },
    {
      "epoch": 43.16651904340124,
      "grad_norm": 0.22495730221271515,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 48735
    },
    {
      "epoch": 43.17094774136404,
      "grad_norm": 0.2142936736345291,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 48740
    },
    {
      "epoch": 43.17537643932684,
      "grad_norm": 0.21201835572719574,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 48745
    },
    {
      "epoch": 43.179805137289634,
      "grad_norm": 0.2537386119365692,
      "learning_rate": 1e-05,
      "loss": 1.0184,
      "step": 48750
    },
    {
      "epoch": 43.184233835252435,
      "grad_norm": 0.24516579508781433,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 48755
    },
    {
      "epoch": 43.188662533215236,
      "grad_norm": 0.22724126279354095,
      "learning_rate": 1e-05,
      "loss": 0.9212,
      "step": 48760
    },
    {
      "epoch": 43.19309123117803,
      "grad_norm": 0.2207074910402298,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 48765
    },
    {
      "epoch": 43.19751992914083,
      "grad_norm": 0.23460142314434052,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 48770
    },
    {
      "epoch": 43.20194862710363,
      "grad_norm": 0.27136340737342834,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 48775
    },
    {
      "epoch": 43.20637732506643,
      "grad_norm": 0.241058811545372,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 48780
    },
    {
      "epoch": 43.21080602302923,
      "grad_norm": 0.2589196562767029,
      "learning_rate": 1e-05,
      "loss": 1.0185,
      "step": 48785
    },
    {
      "epoch": 43.21523472099203,
      "grad_norm": 0.221303790807724,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 48790
    },
    {
      "epoch": 43.21966341895483,
      "grad_norm": 0.21532781422138214,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 48795
    },
    {
      "epoch": 43.224092116917625,
      "grad_norm": 0.2571173906326294,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 48800
    },
    {
      "epoch": 43.228520814880426,
      "grad_norm": 0.24057568609714508,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 48805
    },
    {
      "epoch": 43.23294951284323,
      "grad_norm": 0.25805070996284485,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 48810
    },
    {
      "epoch": 43.23737821080602,
      "grad_norm": 0.2347421497106552,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 48815
    },
    {
      "epoch": 43.24180690876882,
      "grad_norm": 0.2759992182254791,
      "learning_rate": 1e-05,
      "loss": 1.0187,
      "step": 48820
    },
    {
      "epoch": 43.246235606731624,
      "grad_norm": 0.23638631403446198,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 48825
    },
    {
      "epoch": 43.25066430469442,
      "grad_norm": 0.29003486037254333,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 48830
    },
    {
      "epoch": 43.25509300265722,
      "grad_norm": 0.2998713552951813,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 48835
    },
    {
      "epoch": 43.25952170062002,
      "grad_norm": 0.23364496231079102,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 48840
    },
    {
      "epoch": 43.263950398582814,
      "grad_norm": 0.3096825182437897,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 48845
    },
    {
      "epoch": 43.268379096545615,
      "grad_norm": 0.20321637392044067,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 48850
    },
    {
      "epoch": 43.27280779450842,
      "grad_norm": 0.2535122334957123,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 48855
    },
    {
      "epoch": 43.27723649247121,
      "grad_norm": 0.29232820868492126,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 48860
    },
    {
      "epoch": 43.28166519043401,
      "grad_norm": 0.2545989453792572,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 48865
    },
    {
      "epoch": 43.28609388839681,
      "grad_norm": 0.2291771024465561,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 48870
    },
    {
      "epoch": 43.29052258635961,
      "grad_norm": 0.23977535963058472,
      "learning_rate": 1e-05,
      "loss": 1.0148,
      "step": 48875
    },
    {
      "epoch": 43.29495128432241,
      "grad_norm": 0.22587409615516663,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 48880
    },
    {
      "epoch": 43.29937998228521,
      "grad_norm": 0.25296398997306824,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 48885
    },
    {
      "epoch": 43.303808680248004,
      "grad_norm": 0.23332521319389343,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 48890
    },
    {
      "epoch": 43.308237378210805,
      "grad_norm": 0.2421172559261322,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 48895
    },
    {
      "epoch": 43.312666076173606,
      "grad_norm": 0.2455790489912033,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 48900
    },
    {
      "epoch": 43.31709477413641,
      "grad_norm": 0.25551503896713257,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 48905
    },
    {
      "epoch": 43.3215234720992,
      "grad_norm": 0.23959125578403473,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 48910
    },
    {
      "epoch": 43.325952170062,
      "grad_norm": 0.25829899311065674,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 48915
    },
    {
      "epoch": 43.330380868024804,
      "grad_norm": 0.25805699825286865,
      "learning_rate": 1e-05,
      "loss": 0.9113,
      "step": 48920
    },
    {
      "epoch": 43.3348095659876,
      "grad_norm": 0.28306129574775696,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 48925
    },
    {
      "epoch": 43.3392382639504,
      "grad_norm": 0.2557009160518646,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 48930
    },
    {
      "epoch": 43.3436669619132,
      "grad_norm": 0.21177564561367035,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 48935
    },
    {
      "epoch": 43.348095659875995,
      "grad_norm": 0.27018865942955017,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 48940
    },
    {
      "epoch": 43.352524357838796,
      "grad_norm": 0.2154812514781952,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 48945
    },
    {
      "epoch": 43.3569530558016,
      "grad_norm": 0.27615541219711304,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 48950
    },
    {
      "epoch": 43.36138175376439,
      "grad_norm": 0.2018997073173523,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 48955
    },
    {
      "epoch": 43.36581045172719,
      "grad_norm": 0.22869576513767242,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 48960
    },
    {
      "epoch": 43.37023914968999,
      "grad_norm": 0.24255912005901337,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 48965
    },
    {
      "epoch": 43.37466784765279,
      "grad_norm": 0.2782955765724182,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 48970
    },
    {
      "epoch": 43.37909654561559,
      "grad_norm": 0.24312376976013184,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 48975
    },
    {
      "epoch": 43.38352524357839,
      "grad_norm": 0.213237002491951,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 48980
    },
    {
      "epoch": 43.387953941541184,
      "grad_norm": 0.2938636839389801,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 48985
    },
    {
      "epoch": 43.392382639503985,
      "grad_norm": 0.21172958612442017,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 48990
    },
    {
      "epoch": 43.39681133746679,
      "grad_norm": 0.27214258909225464,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 48995
    },
    {
      "epoch": 43.40124003542958,
      "grad_norm": 0.26839783787727356,
      "learning_rate": 1e-05,
      "loss": 1.0364,
      "step": 49000
    },
    {
      "epoch": 43.40566873339238,
      "grad_norm": 0.2545689046382904,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 49005
    },
    {
      "epoch": 43.41009743135518,
      "grad_norm": 0.2299758344888687,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 49010
    },
    {
      "epoch": 43.41452612931798,
      "grad_norm": 0.25038138031959534,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 49015
    },
    {
      "epoch": 43.41895482728078,
      "grad_norm": 0.230449840426445,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 49020
    },
    {
      "epoch": 43.42338352524358,
      "grad_norm": 0.23532941937446594,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 49025
    },
    {
      "epoch": 43.42781222320638,
      "grad_norm": 0.21966420114040375,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 49030
    },
    {
      "epoch": 43.432240921169175,
      "grad_norm": 0.2668024003505707,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 49035
    },
    {
      "epoch": 43.436669619131976,
      "grad_norm": 0.253147691488266,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 49040
    },
    {
      "epoch": 43.44109831709478,
      "grad_norm": 0.22405873239040375,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 49045
    },
    {
      "epoch": 43.44552701505757,
      "grad_norm": 0.2223701924085617,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 49050
    },
    {
      "epoch": 43.44995571302037,
      "grad_norm": 0.24943919479846954,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 49055
    },
    {
      "epoch": 43.454384410983174,
      "grad_norm": 0.20813694596290588,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 49060
    },
    {
      "epoch": 43.45881310894597,
      "grad_norm": 0.2050868570804596,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 49065
    },
    {
      "epoch": 43.46324180690877,
      "grad_norm": 0.2796988785266876,
      "learning_rate": 1e-05,
      "loss": 0.9156,
      "step": 49070
    },
    {
      "epoch": 43.46767050487157,
      "grad_norm": 0.24679169058799744,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 49075
    },
    {
      "epoch": 43.472099202834364,
      "grad_norm": 0.27094992995262146,
      "learning_rate": 1e-05,
      "loss": 0.8916,
      "step": 49080
    },
    {
      "epoch": 43.476527900797166,
      "grad_norm": 0.23632866144180298,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 49085
    },
    {
      "epoch": 43.48095659875997,
      "grad_norm": 0.2382800132036209,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 49090
    },
    {
      "epoch": 43.48538529672276,
      "grad_norm": 0.2205362170934677,
      "learning_rate": 1e-05,
      "loss": 0.9245,
      "step": 49095
    },
    {
      "epoch": 43.48981399468556,
      "grad_norm": 0.3077715039253235,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 49100
    },
    {
      "epoch": 43.49424269264836,
      "grad_norm": 0.24289454519748688,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 49105
    },
    {
      "epoch": 43.49867139061116,
      "grad_norm": 0.27689915895462036,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 49110
    },
    {
      "epoch": 43.50310008857396,
      "grad_norm": 0.2686055302619934,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 49115
    },
    {
      "epoch": 43.50752878653676,
      "grad_norm": 0.2293749451637268,
      "learning_rate": 1e-05,
      "loss": 0.923,
      "step": 49120
    },
    {
      "epoch": 43.511957484499554,
      "grad_norm": 0.21473269164562225,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 49125
    },
    {
      "epoch": 43.516386182462355,
      "grad_norm": 0.21567752957344055,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 49130
    },
    {
      "epoch": 43.520814880425156,
      "grad_norm": 0.23887348175048828,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 49135
    },
    {
      "epoch": 43.52524357838795,
      "grad_norm": 0.2501644790172577,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 49140
    },
    {
      "epoch": 43.52967227635075,
      "grad_norm": 0.2574361264705658,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 49145
    },
    {
      "epoch": 43.53410097431355,
      "grad_norm": 0.2448074221611023,
      "learning_rate": 1e-05,
      "loss": 1.0149,
      "step": 49150
    },
    {
      "epoch": 43.538529672276354,
      "grad_norm": 0.26575735211372375,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 49155
    },
    {
      "epoch": 43.54295837023915,
      "grad_norm": 0.23696841299533844,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 49160
    },
    {
      "epoch": 43.54738706820195,
      "grad_norm": 0.2419581562280655,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 49165
    },
    {
      "epoch": 43.55181576616475,
      "grad_norm": 0.2198808342218399,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 49170
    },
    {
      "epoch": 43.556244464127545,
      "grad_norm": 0.24984802305698395,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 49175
    },
    {
      "epoch": 43.560673162090346,
      "grad_norm": 0.23892787098884583,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 49180
    },
    {
      "epoch": 43.56510186005315,
      "grad_norm": 0.2153821438550949,
      "learning_rate": 1e-05,
      "loss": 1.0395,
      "step": 49185
    },
    {
      "epoch": 43.56953055801594,
      "grad_norm": 0.2657289206981659,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 49190
    },
    {
      "epoch": 43.57395925597874,
      "grad_norm": 0.2194409966468811,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 49195
    },
    {
      "epoch": 43.578387953941544,
      "grad_norm": 0.2293374240398407,
      "learning_rate": 1e-05,
      "loss": 0.9462,
      "step": 49200
    },
    {
      "epoch": 43.58281665190434,
      "grad_norm": 0.24256011843681335,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 49205
    },
    {
      "epoch": 43.58724534986714,
      "grad_norm": 0.27824878692626953,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 49210
    },
    {
      "epoch": 43.59167404782994,
      "grad_norm": 0.2267044335603714,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 49215
    },
    {
      "epoch": 43.596102745792734,
      "grad_norm": 0.2308742105960846,
      "learning_rate": 1e-05,
      "loss": 1.0285,
      "step": 49220
    },
    {
      "epoch": 43.600531443755536,
      "grad_norm": 0.23583976924419403,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 49225
    },
    {
      "epoch": 43.60496014171834,
      "grad_norm": 0.266474187374115,
      "learning_rate": 1e-05,
      "loss": 0.9204,
      "step": 49230
    },
    {
      "epoch": 43.60938883968113,
      "grad_norm": 0.2643164396286011,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 49235
    },
    {
      "epoch": 43.61381753764393,
      "grad_norm": 0.23752117156982422,
      "learning_rate": 1e-05,
      "loss": 1.0172,
      "step": 49240
    },
    {
      "epoch": 43.61824623560673,
      "grad_norm": 0.27046123147010803,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 49245
    },
    {
      "epoch": 43.62267493356953,
      "grad_norm": 0.21744512021541595,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 49250
    },
    {
      "epoch": 43.62710363153233,
      "grad_norm": 0.2618338465690613,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 49255
    },
    {
      "epoch": 43.63153232949513,
      "grad_norm": 0.28749337792396545,
      "learning_rate": 1e-05,
      "loss": 0.9178,
      "step": 49260
    },
    {
      "epoch": 43.635961027457924,
      "grad_norm": 0.22716765105724335,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 49265
    },
    {
      "epoch": 43.640389725420725,
      "grad_norm": 0.20710992813110352,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 49270
    },
    {
      "epoch": 43.644818423383526,
      "grad_norm": 0.24519795179367065,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 49275
    },
    {
      "epoch": 43.64924712134633,
      "grad_norm": 0.23651088774204254,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 49280
    },
    {
      "epoch": 43.65367581930912,
      "grad_norm": 0.23251208662986755,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 49285
    },
    {
      "epoch": 43.65810451727192,
      "grad_norm": 0.2389114499092102,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 49290
    },
    {
      "epoch": 43.662533215234724,
      "grad_norm": 0.25229859352111816,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 49295
    },
    {
      "epoch": 43.66696191319752,
      "grad_norm": 0.2183249592781067,
      "learning_rate": 1e-05,
      "loss": 1.0296,
      "step": 49300
    },
    {
      "epoch": 43.67139061116032,
      "grad_norm": 0.24305486679077148,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 49305
    },
    {
      "epoch": 43.67581930912312,
      "grad_norm": 0.26113682985305786,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 49310
    },
    {
      "epoch": 43.680248007085915,
      "grad_norm": 0.21981877088546753,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 49315
    },
    {
      "epoch": 43.684676705048716,
      "grad_norm": 0.23341485857963562,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 49320
    },
    {
      "epoch": 43.68910540301152,
      "grad_norm": 0.2075640857219696,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 49325
    },
    {
      "epoch": 43.69353410097431,
      "grad_norm": 0.18894685804843903,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 49330
    },
    {
      "epoch": 43.69796279893711,
      "grad_norm": 0.2324332296848297,
      "learning_rate": 1e-05,
      "loss": 0.9095,
      "step": 49335
    },
    {
      "epoch": 43.702391496899914,
      "grad_norm": 0.22342362999916077,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 49340
    },
    {
      "epoch": 43.70682019486271,
      "grad_norm": 0.2466936856508255,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 49345
    },
    {
      "epoch": 43.71124889282551,
      "grad_norm": 0.27215689420700073,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 49350
    },
    {
      "epoch": 43.71567759078831,
      "grad_norm": 0.30186519026756287,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 49355
    },
    {
      "epoch": 43.720106288751104,
      "grad_norm": 0.24005964398384094,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 49360
    },
    {
      "epoch": 43.724534986713905,
      "grad_norm": 0.279868483543396,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 49365
    },
    {
      "epoch": 43.72896368467671,
      "grad_norm": 0.21704097092151642,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 49370
    },
    {
      "epoch": 43.7333923826395,
      "grad_norm": 0.23223647475242615,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 49375
    },
    {
      "epoch": 43.7378210806023,
      "grad_norm": 0.2283603549003601,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 49380
    },
    {
      "epoch": 43.7422497785651,
      "grad_norm": 0.25047197937965393,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 49385
    },
    {
      "epoch": 43.7466784765279,
      "grad_norm": 0.23389460146427155,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 49390
    },
    {
      "epoch": 43.7511071744907,
      "grad_norm": 0.2856042683124542,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 49395
    },
    {
      "epoch": 43.7555358724535,
      "grad_norm": 0.21818546950817108,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 49400
    },
    {
      "epoch": 43.7599645704163,
      "grad_norm": 0.22605416178703308,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 49405
    },
    {
      "epoch": 43.764393268379095,
      "grad_norm": 0.24196910858154297,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 49410
    },
    {
      "epoch": 43.768821966341896,
      "grad_norm": 0.27956026792526245,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 49415
    },
    {
      "epoch": 43.7732506643047,
      "grad_norm": 0.23798374831676483,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 49420
    },
    {
      "epoch": 43.77767936226749,
      "grad_norm": 0.24072261154651642,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 49425
    },
    {
      "epoch": 43.78210806023029,
      "grad_norm": 0.19725191593170166,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 49430
    },
    {
      "epoch": 43.786536758193094,
      "grad_norm": 0.24050062894821167,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 49435
    },
    {
      "epoch": 43.79096545615589,
      "grad_norm": 0.2125917375087738,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 49440
    },
    {
      "epoch": 43.79539415411869,
      "grad_norm": 0.25900572538375854,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 49445
    },
    {
      "epoch": 43.79982285208149,
      "grad_norm": 0.21378600597381592,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 49450
    },
    {
      "epoch": 43.804251550044285,
      "grad_norm": 0.2513464689254761,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 49455
    },
    {
      "epoch": 43.808680248007086,
      "grad_norm": 0.2141907960176468,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 49460
    },
    {
      "epoch": 43.81310894596989,
      "grad_norm": 0.26176124811172485,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 49465
    },
    {
      "epoch": 43.81753764393268,
      "grad_norm": 0.30373674631118774,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 49470
    },
    {
      "epoch": 43.82196634189548,
      "grad_norm": 0.26641765236854553,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 49475
    },
    {
      "epoch": 43.826395039858284,
      "grad_norm": 0.25141939520835876,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 49480
    },
    {
      "epoch": 43.83082373782108,
      "grad_norm": 0.23124532401561737,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 49485
    },
    {
      "epoch": 43.83525243578388,
      "grad_norm": 0.23579838871955872,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 49490
    },
    {
      "epoch": 43.83968113374668,
      "grad_norm": 0.23973248898983002,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 49495
    },
    {
      "epoch": 43.844109831709474,
      "grad_norm": 0.22211840748786926,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 49500
    },
    {
      "epoch": 43.848538529672275,
      "grad_norm": 0.24297572672367096,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 49505
    },
    {
      "epoch": 43.85296722763508,
      "grad_norm": 0.22632776200771332,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 49510
    },
    {
      "epoch": 43.85739592559787,
      "grad_norm": 0.24866943061351776,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 49515
    },
    {
      "epoch": 43.86182462356067,
      "grad_norm": 0.2901684641838074,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 49520
    },
    {
      "epoch": 43.86625332152347,
      "grad_norm": 0.2154412716627121,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 49525
    },
    {
      "epoch": 43.870682019486274,
      "grad_norm": 0.22443675994873047,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 49530
    },
    {
      "epoch": 43.87511071744907,
      "grad_norm": 0.23283013701438904,
      "learning_rate": 1e-05,
      "loss": 1.0224,
      "step": 49535
    },
    {
      "epoch": 43.87953941541187,
      "grad_norm": 0.26056212186813354,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 49540
    },
    {
      "epoch": 43.88396811337467,
      "grad_norm": 0.24146683514118195,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 49545
    },
    {
      "epoch": 43.888396811337465,
      "grad_norm": 0.2114381194114685,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 49550
    },
    {
      "epoch": 43.892825509300266,
      "grad_norm": 0.2068805694580078,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 49555
    },
    {
      "epoch": 43.89725420726307,
      "grad_norm": 0.2212674766778946,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 49560
    },
    {
      "epoch": 43.90168290522586,
      "grad_norm": 0.2440878301858902,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 49565
    },
    {
      "epoch": 43.90611160318866,
      "grad_norm": 0.2579110562801361,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 49570
    },
    {
      "epoch": 43.910540301151464,
      "grad_norm": 0.2541361153125763,
      "learning_rate": 1e-05,
      "loss": 1.0305,
      "step": 49575
    },
    {
      "epoch": 43.91496899911426,
      "grad_norm": 0.24004533886909485,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 49580
    },
    {
      "epoch": 43.91939769707706,
      "grad_norm": 0.23401138186454773,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 49585
    },
    {
      "epoch": 43.92382639503986,
      "grad_norm": 0.28666403889656067,
      "learning_rate": 1e-05,
      "loss": 0.9152,
      "step": 49590
    },
    {
      "epoch": 43.928255093002655,
      "grad_norm": 0.2494373917579651,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 49595
    },
    {
      "epoch": 43.932683790965456,
      "grad_norm": 0.2411811649799347,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 49600
    },
    {
      "epoch": 43.93711248892826,
      "grad_norm": 0.22462542355060577,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 49605
    },
    {
      "epoch": 43.94154118689105,
      "grad_norm": 0.2938650846481323,
      "learning_rate": 1e-05,
      "loss": 0.9032,
      "step": 49610
    },
    {
      "epoch": 43.94596988485385,
      "grad_norm": 0.24739117920398712,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 49615
    },
    {
      "epoch": 43.95039858281665,
      "grad_norm": 0.22504116594791412,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 49620
    },
    {
      "epoch": 43.95482728077945,
      "grad_norm": 0.2561790943145752,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 49625
    },
    {
      "epoch": 43.95925597874225,
      "grad_norm": 0.22255125641822815,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 49630
    },
    {
      "epoch": 43.96368467670505,
      "grad_norm": 0.21574559807777405,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 49635
    },
    {
      "epoch": 43.96811337466785,
      "grad_norm": 0.2735796272754669,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 49640
    },
    {
      "epoch": 43.972542072630645,
      "grad_norm": 0.24575546383857727,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 49645
    },
    {
      "epoch": 43.97697077059345,
      "grad_norm": 0.2672165632247925,
      "learning_rate": 1e-05,
      "loss": 0.9237,
      "step": 49650
    },
    {
      "epoch": 43.98139946855625,
      "grad_norm": 0.3126530051231384,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 49655
    },
    {
      "epoch": 43.98582816651904,
      "grad_norm": 0.2688734829425812,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 49660
    },
    {
      "epoch": 43.99025686448184,
      "grad_norm": 0.2695014774799347,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 49665
    },
    {
      "epoch": 43.994685562444644,
      "grad_norm": 0.23811259865760803,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 49670
    },
    {
      "epoch": 43.99911426040744,
      "grad_norm": 0.2175956815481186,
      "learning_rate": 1e-05,
      "loss": 1.0484,
      "step": 49675
    },
    {
      "epoch": 44.00354295837024,
      "grad_norm": 0.2453879415988922,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 49680
    },
    {
      "epoch": 44.00797165633304,
      "grad_norm": 0.2216404676437378,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 49685
    },
    {
      "epoch": 44.012400354295835,
      "grad_norm": 0.25337862968444824,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 49690
    },
    {
      "epoch": 44.016829052258636,
      "grad_norm": 0.2043168991804123,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 49695
    },
    {
      "epoch": 44.02125775022144,
      "grad_norm": 0.24075400829315186,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 49700
    },
    {
      "epoch": 44.02568644818423,
      "grad_norm": 0.2637993097305298,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 49705
    },
    {
      "epoch": 44.03011514614703,
      "grad_norm": 0.2577388882637024,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 49710
    },
    {
      "epoch": 44.034543844109834,
      "grad_norm": 0.2601379454135895,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 49715
    },
    {
      "epoch": 44.03897254207263,
      "grad_norm": 0.3299559950828552,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 49720
    },
    {
      "epoch": 44.04340124003543,
      "grad_norm": 0.29679402709007263,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 49725
    },
    {
      "epoch": 44.04782993799823,
      "grad_norm": 0.2444685995578766,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 49730
    },
    {
      "epoch": 44.052258635961024,
      "grad_norm": 0.2680133283138275,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 49735
    },
    {
      "epoch": 44.056687333923826,
      "grad_norm": 0.23599475622177124,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 49740
    },
    {
      "epoch": 44.06111603188663,
      "grad_norm": 0.2521182894706726,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 49745
    },
    {
      "epoch": 44.06554472984942,
      "grad_norm": 0.2508750855922699,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 49750
    },
    {
      "epoch": 44.06997342781222,
      "grad_norm": 0.2555854916572571,
      "learning_rate": 1e-05,
      "loss": 0.9269,
      "step": 49755
    },
    {
      "epoch": 44.07440212577502,
      "grad_norm": 0.22437188029289246,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 49760
    },
    {
      "epoch": 44.078830823737825,
      "grad_norm": 0.22609266638755798,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 49765
    },
    {
      "epoch": 44.08325952170062,
      "grad_norm": 0.21780064702033997,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 49770
    },
    {
      "epoch": 44.08768821966342,
      "grad_norm": 0.23511624336242676,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 49775
    },
    {
      "epoch": 44.09211691762622,
      "grad_norm": 0.355492502450943,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 49780
    },
    {
      "epoch": 44.096545615589015,
      "grad_norm": 0.3028511703014374,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 49785
    },
    {
      "epoch": 44.100974313551816,
      "grad_norm": 0.35851171612739563,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 49790
    },
    {
      "epoch": 44.10540301151462,
      "grad_norm": 0.25587978959083557,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 49795
    },
    {
      "epoch": 44.10983170947741,
      "grad_norm": 0.23267780244350433,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 49800
    },
    {
      "epoch": 44.11426040744021,
      "grad_norm": 0.2884264886379242,
      "learning_rate": 1e-05,
      "loss": 1.0532,
      "step": 49805
    },
    {
      "epoch": 44.118689105403014,
      "grad_norm": 0.23949430882930756,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 49810
    },
    {
      "epoch": 44.12311780336581,
      "grad_norm": 0.23606997728347778,
      "learning_rate": 1e-05,
      "loss": 1.0379,
      "step": 49815
    },
    {
      "epoch": 44.12754650132861,
      "grad_norm": 0.2520465552806854,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 49820
    },
    {
      "epoch": 44.13197519929141,
      "grad_norm": 0.2147519588470459,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 49825
    },
    {
      "epoch": 44.136403897254205,
      "grad_norm": 0.2593078911304474,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 49830
    },
    {
      "epoch": 44.140832595217006,
      "grad_norm": 0.28337571024894714,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 49835
    },
    {
      "epoch": 44.14526129317981,
      "grad_norm": 0.2923653721809387,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 49840
    },
    {
      "epoch": 44.1496899911426,
      "grad_norm": 0.2859342396259308,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 49845
    },
    {
      "epoch": 44.1541186891054,
      "grad_norm": 0.23269562423229218,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 49850
    },
    {
      "epoch": 44.158547387068204,
      "grad_norm": 0.3974922299385071,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 49855
    },
    {
      "epoch": 44.162976085031,
      "grad_norm": 0.2668763995170593,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 49860
    },
    {
      "epoch": 44.1674047829938,
      "grad_norm": 0.23206990957260132,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 49865
    },
    {
      "epoch": 44.1718334809566,
      "grad_norm": 0.2509148418903351,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 49870
    },
    {
      "epoch": 44.176262178919394,
      "grad_norm": 0.2504391670227051,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 49875
    },
    {
      "epoch": 44.180690876882196,
      "grad_norm": 0.2371387928724289,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 49880
    },
    {
      "epoch": 44.185119574845,
      "grad_norm": 0.2349540740251541,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 49885
    },
    {
      "epoch": 44.1895482728078,
      "grad_norm": 0.2520585060119629,
      "learning_rate": 1e-05,
      "loss": 0.9253,
      "step": 49890
    },
    {
      "epoch": 44.19397697077059,
      "grad_norm": 0.290661096572876,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 49895
    },
    {
      "epoch": 44.19840566873339,
      "grad_norm": 0.26891621947288513,
      "learning_rate": 1e-05,
      "loss": 0.9396,
      "step": 49900
    },
    {
      "epoch": 44.202834366696194,
      "grad_norm": 0.23947502672672272,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 49905
    },
    {
      "epoch": 44.20726306465899,
      "grad_norm": 0.24221175909042358,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 49910
    },
    {
      "epoch": 44.21169176262179,
      "grad_norm": 0.28259408473968506,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 49915
    },
    {
      "epoch": 44.21612046058459,
      "grad_norm": 0.25452733039855957,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 49920
    },
    {
      "epoch": 44.220549158547385,
      "grad_norm": 0.22958946228027344,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 49925
    },
    {
      "epoch": 44.224977856510186,
      "grad_norm": 0.23510897159576416,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 49930
    },
    {
      "epoch": 44.22940655447299,
      "grad_norm": 0.250924289226532,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 49935
    },
    {
      "epoch": 44.23383525243578,
      "grad_norm": 0.2388603538274765,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 49940
    },
    {
      "epoch": 44.23826395039858,
      "grad_norm": 0.2535986602306366,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 49945
    },
    {
      "epoch": 44.242692648361384,
      "grad_norm": 0.2537343502044678,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 49950
    },
    {
      "epoch": 44.24712134632418,
      "grad_norm": 0.22179286181926727,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 49955
    },
    {
      "epoch": 44.25155004428698,
      "grad_norm": 0.23771311342716217,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 49960
    },
    {
      "epoch": 44.25597874224978,
      "grad_norm": 0.2930850088596344,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 49965
    },
    {
      "epoch": 44.260407440212575,
      "grad_norm": 0.22962325811386108,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 49970
    },
    {
      "epoch": 44.264836138175376,
      "grad_norm": 0.2151719182729721,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 49975
    },
    {
      "epoch": 44.26926483613818,
      "grad_norm": 0.2562272846698761,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 49980
    },
    {
      "epoch": 44.27369353410097,
      "grad_norm": 0.2469555139541626,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 49985
    },
    {
      "epoch": 44.27812223206377,
      "grad_norm": 0.23190981149673462,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 49990
    },
    {
      "epoch": 44.282550930026574,
      "grad_norm": 0.2526366710662842,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 49995
    },
    {
      "epoch": 44.28697962798937,
      "grad_norm": 0.2519548535346985,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 50000
    },
    {
      "epoch": 44.29140832595217,
      "grad_norm": 0.2310408651828766,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 50005
    },
    {
      "epoch": 44.29583702391497,
      "grad_norm": 0.24374184012413025,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 50010
    },
    {
      "epoch": 44.30026572187777,
      "grad_norm": 0.20592452585697174,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 50015
    },
    {
      "epoch": 44.304694419840565,
      "grad_norm": 0.23045875132083893,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 50020
    },
    {
      "epoch": 44.30912311780337,
      "grad_norm": 0.23315216600894928,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 50025
    },
    {
      "epoch": 44.31355181576617,
      "grad_norm": 0.23353172838687897,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 50030
    },
    {
      "epoch": 44.31798051372896,
      "grad_norm": 0.2239086925983429,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 50035
    },
    {
      "epoch": 44.32240921169176,
      "grad_norm": 0.23434504866600037,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 50040
    },
    {
      "epoch": 44.326837909654564,
      "grad_norm": 0.25160253047943115,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 50045
    },
    {
      "epoch": 44.33126660761736,
      "grad_norm": 0.22875922918319702,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 50050
    },
    {
      "epoch": 44.33569530558016,
      "grad_norm": 0.2536490857601166,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 50055
    },
    {
      "epoch": 44.34012400354296,
      "grad_norm": 0.25222399830818176,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 50060
    },
    {
      "epoch": 44.344552701505755,
      "grad_norm": 0.24515140056610107,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 50065
    },
    {
      "epoch": 44.348981399468556,
      "grad_norm": 0.20229072868824005,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 50070
    },
    {
      "epoch": 44.35341009743136,
      "grad_norm": 0.22831720113754272,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 50075
    },
    {
      "epoch": 44.35783879539415,
      "grad_norm": 0.25126734375953674,
      "learning_rate": 1e-05,
      "loss": 1.0495,
      "step": 50080
    },
    {
      "epoch": 44.36226749335695,
      "grad_norm": 0.24788789451122284,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 50085
    },
    {
      "epoch": 44.366696191319754,
      "grad_norm": 0.23605138063430786,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 50090
    },
    {
      "epoch": 44.37112488928255,
      "grad_norm": 0.31426650285720825,
      "learning_rate": 1e-05,
      "loss": 0.9221,
      "step": 50095
    },
    {
      "epoch": 44.37555358724535,
      "grad_norm": 0.22368678450584412,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 50100
    },
    {
      "epoch": 44.37998228520815,
      "grad_norm": 0.23569771647453308,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 50105
    },
    {
      "epoch": 44.384410983170945,
      "grad_norm": 0.3110570013523102,
      "learning_rate": 1e-05,
      "loss": 1.0301,
      "step": 50110
    },
    {
      "epoch": 44.388839681133746,
      "grad_norm": 0.25935137271881104,
      "learning_rate": 1e-05,
      "loss": 1.0228,
      "step": 50115
    },
    {
      "epoch": 44.39326837909655,
      "grad_norm": 0.28817296028137207,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 50120
    },
    {
      "epoch": 44.39769707705934,
      "grad_norm": 0.28363922238349915,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 50125
    },
    {
      "epoch": 44.40212577502214,
      "grad_norm": 0.2399965524673462,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 50130
    },
    {
      "epoch": 44.40655447298494,
      "grad_norm": 0.21465909481048584,
      "learning_rate": 1e-05,
      "loss": 0.9294,
      "step": 50135
    },
    {
      "epoch": 44.410983170947745,
      "grad_norm": 0.22371327877044678,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 50140
    },
    {
      "epoch": 44.41541186891054,
      "grad_norm": 0.22318057715892792,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 50145
    },
    {
      "epoch": 44.41984056687334,
      "grad_norm": 0.22202415764331818,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 50150
    },
    {
      "epoch": 44.42426926483614,
      "grad_norm": 0.24073345959186554,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 50155
    },
    {
      "epoch": 44.428697962798935,
      "grad_norm": 0.2259293496608734,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 50160
    },
    {
      "epoch": 44.43312666076174,
      "grad_norm": 0.21760140359401703,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 50165
    },
    {
      "epoch": 44.43755535872454,
      "grad_norm": 0.22334861755371094,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 50170
    },
    {
      "epoch": 44.44198405668733,
      "grad_norm": 0.26698681712150574,
      "learning_rate": 1e-05,
      "loss": 0.8922,
      "step": 50175
    },
    {
      "epoch": 44.44641275465013,
      "grad_norm": 0.24635346233844757,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 50180
    },
    {
      "epoch": 44.450841452612934,
      "grad_norm": 0.251383513212204,
      "learning_rate": 1e-05,
      "loss": 0.92,
      "step": 50185
    },
    {
      "epoch": 44.45527015057573,
      "grad_norm": 0.28668734431266785,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 50190
    },
    {
      "epoch": 44.45969884853853,
      "grad_norm": 0.22254236042499542,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 50195
    },
    {
      "epoch": 44.46412754650133,
      "grad_norm": 0.21074004471302032,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 50200
    },
    {
      "epoch": 44.468556244464125,
      "grad_norm": 0.26638713479042053,
      "learning_rate": 1e-05,
      "loss": 0.9357,
      "step": 50205
    },
    {
      "epoch": 44.472984942426926,
      "grad_norm": 0.2354864925146103,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 50210
    },
    {
      "epoch": 44.47741364038973,
      "grad_norm": 0.22966063022613525,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 50215
    },
    {
      "epoch": 44.48184233835252,
      "grad_norm": 0.27157625555992126,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 50220
    },
    {
      "epoch": 44.48627103631532,
      "grad_norm": 0.23083417117595673,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 50225
    },
    {
      "epoch": 44.490699734278124,
      "grad_norm": 0.22872625291347504,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 50230
    },
    {
      "epoch": 44.49512843224092,
      "grad_norm": 0.23940850794315338,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 50235
    },
    {
      "epoch": 44.49955713020372,
      "grad_norm": 0.2548903226852417,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 50240
    },
    {
      "epoch": 44.50398582816652,
      "grad_norm": 0.268465131521225,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 50245
    },
    {
      "epoch": 44.508414526129314,
      "grad_norm": 0.25344371795654297,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 50250
    },
    {
      "epoch": 44.512843224092116,
      "grad_norm": 0.20049132406711578,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 50255
    },
    {
      "epoch": 44.51727192205492,
      "grad_norm": 0.2764725685119629,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 50260
    },
    {
      "epoch": 44.52170062001772,
      "grad_norm": 0.23784217238426208,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 50265
    },
    {
      "epoch": 44.52612931798051,
      "grad_norm": 0.29805687069892883,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 50270
    },
    {
      "epoch": 44.53055801594331,
      "grad_norm": 0.3250724673271179,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 50275
    },
    {
      "epoch": 44.534986713906115,
      "grad_norm": 0.23270177841186523,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 50280
    },
    {
      "epoch": 44.53941541186891,
      "grad_norm": 0.2521251142024994,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 50285
    },
    {
      "epoch": 44.54384410983171,
      "grad_norm": 0.25402703881263733,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 50290
    },
    {
      "epoch": 44.54827280779451,
      "grad_norm": 0.26695531606674194,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 50295
    },
    {
      "epoch": 44.552701505757305,
      "grad_norm": 0.2517922520637512,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 50300
    },
    {
      "epoch": 44.55713020372011,
      "grad_norm": 0.21802237629890442,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 50305
    },
    {
      "epoch": 44.56155890168291,
      "grad_norm": 0.20943692326545715,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 50310
    },
    {
      "epoch": 44.5659875996457,
      "grad_norm": 0.22353410720825195,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 50315
    },
    {
      "epoch": 44.5704162976085,
      "grad_norm": 0.248456671833992,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 50320
    },
    {
      "epoch": 44.574844995571304,
      "grad_norm": 0.2801806330680847,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 50325
    },
    {
      "epoch": 44.5792736935341,
      "grad_norm": 0.28935131430625916,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 50330
    },
    {
      "epoch": 44.5837023914969,
      "grad_norm": 0.20283353328704834,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 50335
    },
    {
      "epoch": 44.5881310894597,
      "grad_norm": 0.22326818108558655,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 50340
    },
    {
      "epoch": 44.592559787422495,
      "grad_norm": 0.2251996397972107,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 50345
    },
    {
      "epoch": 44.596988485385296,
      "grad_norm": 0.23880022764205933,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 50350
    },
    {
      "epoch": 44.6014171833481,
      "grad_norm": 0.29937225580215454,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 50355
    },
    {
      "epoch": 44.60584588131089,
      "grad_norm": 0.2504315972328186,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 50360
    },
    {
      "epoch": 44.61027457927369,
      "grad_norm": 0.2714006006717682,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 50365
    },
    {
      "epoch": 44.614703277236494,
      "grad_norm": 0.2577667534351349,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 50370
    },
    {
      "epoch": 44.619131975199295,
      "grad_norm": 0.2723548114299774,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 50375
    },
    {
      "epoch": 44.62356067316209,
      "grad_norm": 0.22927972674369812,
      "learning_rate": 1e-05,
      "loss": 1.0186,
      "step": 50380
    },
    {
      "epoch": 44.62798937112489,
      "grad_norm": 0.2656670808792114,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 50385
    },
    {
      "epoch": 44.63241806908769,
      "grad_norm": 0.24565164744853973,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 50390
    },
    {
      "epoch": 44.636846767050486,
      "grad_norm": 0.23352958261966705,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 50395
    },
    {
      "epoch": 44.64127546501329,
      "grad_norm": 0.24695728719234467,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 50400
    },
    {
      "epoch": 44.64570416297609,
      "grad_norm": 0.24550925195217133,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 50405
    },
    {
      "epoch": 44.65013286093888,
      "grad_norm": 0.24056026339530945,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 50410
    },
    {
      "epoch": 44.65456155890168,
      "grad_norm": 0.23579323291778564,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 50415
    },
    {
      "epoch": 44.658990256864485,
      "grad_norm": 0.2570098638534546,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 50420
    },
    {
      "epoch": 44.66341895482728,
      "grad_norm": 0.22641190886497498,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 50425
    },
    {
      "epoch": 44.66784765279008,
      "grad_norm": 0.23759019374847412,
      "learning_rate": 1e-05,
      "loss": 0.882,
      "step": 50430
    },
    {
      "epoch": 44.67227635075288,
      "grad_norm": 0.2774142622947693,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 50435
    },
    {
      "epoch": 44.676705048715675,
      "grad_norm": 0.23752395808696747,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 50440
    },
    {
      "epoch": 44.681133746678476,
      "grad_norm": 0.23774580657482147,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 50445
    },
    {
      "epoch": 44.68556244464128,
      "grad_norm": 0.2331254482269287,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 50450
    },
    {
      "epoch": 44.68999114260407,
      "grad_norm": 0.23788121342658997,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 50455
    },
    {
      "epoch": 44.69441984056687,
      "grad_norm": 0.2610919177532196,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 50460
    },
    {
      "epoch": 44.698848538529674,
      "grad_norm": 0.24243460595607758,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 50465
    },
    {
      "epoch": 44.70327723649247,
      "grad_norm": 0.25183090567588806,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 50470
    },
    {
      "epoch": 44.70770593445527,
      "grad_norm": 0.2365809977054596,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 50475
    },
    {
      "epoch": 44.71213463241807,
      "grad_norm": 0.3140511214733124,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 50480
    },
    {
      "epoch": 44.716563330380865,
      "grad_norm": 0.19660690426826477,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 50485
    },
    {
      "epoch": 44.720992028343666,
      "grad_norm": 0.21523156762123108,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 50490
    },
    {
      "epoch": 44.72542072630647,
      "grad_norm": 0.23465435206890106,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 50495
    },
    {
      "epoch": 44.72984942426926,
      "grad_norm": 0.2012229710817337,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 50500
    },
    {
      "epoch": 44.73427812223206,
      "grad_norm": 0.2181909829378128,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 50505
    },
    {
      "epoch": 44.738706820194864,
      "grad_norm": 0.2760373651981354,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 50510
    },
    {
      "epoch": 44.743135518157665,
      "grad_norm": 0.2988514006137848,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 50515
    },
    {
      "epoch": 44.74756421612046,
      "grad_norm": 0.23796944320201874,
      "learning_rate": 1e-05,
      "loss": 0.9073,
      "step": 50520
    },
    {
      "epoch": 44.75199291408326,
      "grad_norm": 0.23518982529640198,
      "learning_rate": 1e-05,
      "loss": 1.0174,
      "step": 50525
    },
    {
      "epoch": 44.75642161204606,
      "grad_norm": 0.2623854875564575,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 50530
    },
    {
      "epoch": 44.760850310008856,
      "grad_norm": 0.2675706148147583,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 50535
    },
    {
      "epoch": 44.76527900797166,
      "grad_norm": 0.27577558159828186,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 50540
    },
    {
      "epoch": 44.76970770593446,
      "grad_norm": 0.26294589042663574,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 50545
    },
    {
      "epoch": 44.77413640389725,
      "grad_norm": 0.2526409924030304,
      "learning_rate": 1e-05,
      "loss": 0.9121,
      "step": 50550
    },
    {
      "epoch": 44.77856510186005,
      "grad_norm": 0.30389928817749023,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 50555
    },
    {
      "epoch": 44.782993799822854,
      "grad_norm": 0.2704399824142456,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 50560
    },
    {
      "epoch": 44.78742249778565,
      "grad_norm": 0.2622881233692169,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 50565
    },
    {
      "epoch": 44.79185119574845,
      "grad_norm": 0.27411016821861267,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 50570
    },
    {
      "epoch": 44.79627989371125,
      "grad_norm": 0.24862916767597198,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 50575
    },
    {
      "epoch": 44.800708591674045,
      "grad_norm": 0.2665644586086273,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 50580
    },
    {
      "epoch": 44.805137289636846,
      "grad_norm": 0.24571362137794495,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 50585
    },
    {
      "epoch": 44.80956598759965,
      "grad_norm": 0.23265193402767181,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 50590
    },
    {
      "epoch": 44.81399468556244,
      "grad_norm": 0.2324245721101761,
      "learning_rate": 1e-05,
      "loss": 0.929,
      "step": 50595
    },
    {
      "epoch": 44.81842338352524,
      "grad_norm": 0.22979331016540527,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 50600
    },
    {
      "epoch": 44.822852081488044,
      "grad_norm": 0.21760888397693634,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 50605
    },
    {
      "epoch": 44.82728077945084,
      "grad_norm": 0.23312786221504211,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 50610
    },
    {
      "epoch": 44.83170947741364,
      "grad_norm": 0.2248609960079193,
      "learning_rate": 1e-05,
      "loss": 1.0563,
      "step": 50615
    },
    {
      "epoch": 44.83613817537644,
      "grad_norm": 0.25403907895088196,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 50620
    },
    {
      "epoch": 44.84056687333924,
      "grad_norm": 0.18929818272590637,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 50625
    },
    {
      "epoch": 44.844995571302036,
      "grad_norm": 0.22148524224758148,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 50630
    },
    {
      "epoch": 44.84942426926484,
      "grad_norm": 0.2323078215122223,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 50635
    },
    {
      "epoch": 44.85385296722764,
      "grad_norm": 0.250608891248703,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 50640
    },
    {
      "epoch": 44.85828166519043,
      "grad_norm": 0.2476261556148529,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 50645
    },
    {
      "epoch": 44.862710363153234,
      "grad_norm": 0.27119264006614685,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 50650
    },
    {
      "epoch": 44.867139061116035,
      "grad_norm": 0.24649271368980408,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 50655
    },
    {
      "epoch": 44.87156775907883,
      "grad_norm": 0.26250141859054565,
      "learning_rate": 1e-05,
      "loss": 1.023,
      "step": 50660
    },
    {
      "epoch": 44.87599645704163,
      "grad_norm": 0.2345227748155594,
      "learning_rate": 1e-05,
      "loss": 0.915,
      "step": 50665
    },
    {
      "epoch": 44.88042515500443,
      "grad_norm": 0.2239430844783783,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 50670
    },
    {
      "epoch": 44.884853852967225,
      "grad_norm": 0.22486598789691925,
      "learning_rate": 1e-05,
      "loss": 1.0295,
      "step": 50675
    },
    {
      "epoch": 44.88928255093003,
      "grad_norm": 0.2772603929042816,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 50680
    },
    {
      "epoch": 44.89371124889283,
      "grad_norm": 0.2388048619031906,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 50685
    },
    {
      "epoch": 44.89813994685562,
      "grad_norm": 0.2416672706604004,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 50690
    },
    {
      "epoch": 44.90256864481842,
      "grad_norm": 0.21063125133514404,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 50695
    },
    {
      "epoch": 44.906997342781224,
      "grad_norm": 0.24858325719833374,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 50700
    },
    {
      "epoch": 44.91142604074402,
      "grad_norm": 0.2745394706726074,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 50705
    },
    {
      "epoch": 44.91585473870682,
      "grad_norm": 0.2880264222621918,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 50710
    },
    {
      "epoch": 44.92028343666962,
      "grad_norm": 0.22836659848690033,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 50715
    },
    {
      "epoch": 44.924712134632415,
      "grad_norm": 0.22982192039489746,
      "learning_rate": 1e-05,
      "loss": 0.8998,
      "step": 50720
    },
    {
      "epoch": 44.929140832595216,
      "grad_norm": 0.2595456540584564,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 50725
    },
    {
      "epoch": 44.93356953055802,
      "grad_norm": 0.2588038444519043,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 50730
    },
    {
      "epoch": 44.93799822852081,
      "grad_norm": 0.2293577641248703,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 50735
    },
    {
      "epoch": 44.94242692648361,
      "grad_norm": 0.2277379035949707,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 50740
    },
    {
      "epoch": 44.946855624446414,
      "grad_norm": 0.22213824093341827,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 50745
    },
    {
      "epoch": 44.951284322409215,
      "grad_norm": 0.24956080317497253,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 50750
    },
    {
      "epoch": 44.95571302037201,
      "grad_norm": 0.22188463807106018,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 50755
    },
    {
      "epoch": 44.96014171833481,
      "grad_norm": 0.22646038234233856,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 50760
    },
    {
      "epoch": 44.96457041629761,
      "grad_norm": 0.20522108674049377,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 50765
    },
    {
      "epoch": 44.968999114260406,
      "grad_norm": 0.2669883668422699,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 50770
    },
    {
      "epoch": 44.97342781222321,
      "grad_norm": 0.24617964029312134,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 50775
    },
    {
      "epoch": 44.97785651018601,
      "grad_norm": 0.20243538916110992,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 50780
    },
    {
      "epoch": 44.9822852081488,
      "grad_norm": 0.22546884417533875,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 50785
    },
    {
      "epoch": 44.9867139061116,
      "grad_norm": 0.21208958327770233,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 50790
    },
    {
      "epoch": 44.991142604074405,
      "grad_norm": 0.20937855541706085,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 50795
    },
    {
      "epoch": 44.9955713020372,
      "grad_norm": 0.1977820247411728,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 50800
    },
    {
      "epoch": 45.0,
      "grad_norm": 0.22402580082416534,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 50805
    },
    {
      "epoch": 45.0044286979628,
      "grad_norm": 0.2275746762752533,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 50810
    },
    {
      "epoch": 45.008857395925595,
      "grad_norm": 0.23878690600395203,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 50815
    },
    {
      "epoch": 45.0132860938884,
      "grad_norm": 0.257638543844223,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 50820
    },
    {
      "epoch": 45.0177147918512,
      "grad_norm": 0.26441490650177,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 50825
    },
    {
      "epoch": 45.02214348981399,
      "grad_norm": 0.24332650005817413,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 50830
    },
    {
      "epoch": 45.02657218777679,
      "grad_norm": 0.2957375645637512,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 50835
    },
    {
      "epoch": 45.031000885739594,
      "grad_norm": 0.239308699965477,
      "learning_rate": 1e-05,
      "loss": 0.9462,
      "step": 50840
    },
    {
      "epoch": 45.03542958370239,
      "grad_norm": 0.22293758392333984,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 50845
    },
    {
      "epoch": 45.03985828166519,
      "grad_norm": 0.23287378251552582,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 50850
    },
    {
      "epoch": 45.04428697962799,
      "grad_norm": 0.23699893057346344,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 50855
    },
    {
      "epoch": 45.048715677590785,
      "grad_norm": 0.28027772903442383,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 50860
    },
    {
      "epoch": 45.053144375553586,
      "grad_norm": 0.2623668313026428,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 50865
    },
    {
      "epoch": 45.05757307351639,
      "grad_norm": 0.20903250575065613,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 50870
    },
    {
      "epoch": 45.06200177147919,
      "grad_norm": 0.2196718454360962,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 50875
    },
    {
      "epoch": 45.06643046944198,
      "grad_norm": 0.19057652354240417,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 50880
    },
    {
      "epoch": 45.070859167404784,
      "grad_norm": 0.21576182544231415,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 50885
    },
    {
      "epoch": 45.075287865367585,
      "grad_norm": 0.23814213275909424,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 50890
    },
    {
      "epoch": 45.07971656333038,
      "grad_norm": 0.26735758781433105,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 50895
    },
    {
      "epoch": 45.08414526129318,
      "grad_norm": 0.24511125683784485,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 50900
    },
    {
      "epoch": 45.08857395925598,
      "grad_norm": 0.22583621740341187,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 50905
    },
    {
      "epoch": 45.093002657218776,
      "grad_norm": 0.2847575843334198,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 50910
    },
    {
      "epoch": 45.09743135518158,
      "grad_norm": 0.24746952950954437,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 50915
    },
    {
      "epoch": 45.10186005314438,
      "grad_norm": 0.22392427921295166,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 50920
    },
    {
      "epoch": 45.10628875110717,
      "grad_norm": 0.20248247683048248,
      "learning_rate": 1e-05,
      "loss": 1.0252,
      "step": 50925
    },
    {
      "epoch": 45.11071744906997,
      "grad_norm": 0.24867719411849976,
      "learning_rate": 1e-05,
      "loss": 0.9273,
      "step": 50930
    },
    {
      "epoch": 45.115146147032775,
      "grad_norm": 0.2462959736585617,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 50935
    },
    {
      "epoch": 45.11957484499557,
      "grad_norm": 0.24415653944015503,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 50940
    },
    {
      "epoch": 45.12400354295837,
      "grad_norm": 0.24125155806541443,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 50945
    },
    {
      "epoch": 45.12843224092117,
      "grad_norm": 0.264469176530838,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 50950
    },
    {
      "epoch": 45.132860938883965,
      "grad_norm": 0.2606436610221863,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 50955
    },
    {
      "epoch": 45.137289636846766,
      "grad_norm": 0.20980465412139893,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 50960
    },
    {
      "epoch": 45.14171833480957,
      "grad_norm": 0.2637585699558258,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 50965
    },
    {
      "epoch": 45.14614703277236,
      "grad_norm": 0.2547243535518646,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 50970
    },
    {
      "epoch": 45.15057573073516,
      "grad_norm": 0.2801729738712311,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 50975
    },
    {
      "epoch": 45.155004428697964,
      "grad_norm": 0.2803928256034851,
      "learning_rate": 1e-05,
      "loss": 0.9305,
      "step": 50980
    },
    {
      "epoch": 45.15943312666076,
      "grad_norm": 0.27185243368148804,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 50985
    },
    {
      "epoch": 45.16386182462356,
      "grad_norm": 0.23501235246658325,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 50990
    },
    {
      "epoch": 45.16829052258636,
      "grad_norm": 0.24999774992465973,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 50995
    },
    {
      "epoch": 45.17271922054916,
      "grad_norm": 0.2192055881023407,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 51000
    },
    {
      "epoch": 45.177147918511956,
      "grad_norm": 0.26313719153404236,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 51005
    },
    {
      "epoch": 45.18157661647476,
      "grad_norm": 0.2927393317222595,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 51010
    },
    {
      "epoch": 45.18600531443756,
      "grad_norm": 0.2240595519542694,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 51015
    },
    {
      "epoch": 45.19043401240035,
      "grad_norm": 0.32358798384666443,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 51020
    },
    {
      "epoch": 45.194862710363154,
      "grad_norm": 0.26124095916748047,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 51025
    },
    {
      "epoch": 45.199291408325955,
      "grad_norm": 0.22017693519592285,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 51030
    },
    {
      "epoch": 45.20372010628875,
      "grad_norm": 0.21079875528812408,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 51035
    },
    {
      "epoch": 45.20814880425155,
      "grad_norm": 0.2244163453578949,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 51040
    },
    {
      "epoch": 45.21257750221435,
      "grad_norm": 0.2547610104084015,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 51045
    },
    {
      "epoch": 45.217006200177146,
      "grad_norm": 0.24453681707382202,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 51050
    },
    {
      "epoch": 45.22143489813995,
      "grad_norm": 0.25053828954696655,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 51055
    },
    {
      "epoch": 45.22586359610275,
      "grad_norm": 0.24711820483207703,
      "learning_rate": 1e-05,
      "loss": 0.9074,
      "step": 51060
    },
    {
      "epoch": 45.23029229406554,
      "grad_norm": 0.32900500297546387,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 51065
    },
    {
      "epoch": 45.23472099202834,
      "grad_norm": 0.23388823866844177,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 51070
    },
    {
      "epoch": 45.239149689991144,
      "grad_norm": 0.21588705480098724,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 51075
    },
    {
      "epoch": 45.24357838795394,
      "grad_norm": 0.2930537462234497,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 51080
    },
    {
      "epoch": 45.24800708591674,
      "grad_norm": 0.2271912395954132,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 51085
    },
    {
      "epoch": 45.25243578387954,
      "grad_norm": 0.241968035697937,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 51090
    },
    {
      "epoch": 45.256864481842335,
      "grad_norm": 0.23127003014087677,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 51095
    },
    {
      "epoch": 45.261293179805136,
      "grad_norm": 0.2244143784046173,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 51100
    },
    {
      "epoch": 45.26572187776794,
      "grad_norm": 0.2516489624977112,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 51105
    },
    {
      "epoch": 45.27015057573073,
      "grad_norm": 0.2522209882736206,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 51110
    },
    {
      "epoch": 45.27457927369353,
      "grad_norm": 0.2186204046010971,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 51115
    },
    {
      "epoch": 45.279007971656334,
      "grad_norm": 0.24312123656272888,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 51120
    },
    {
      "epoch": 45.283436669619135,
      "grad_norm": 0.21799670159816742,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 51125
    },
    {
      "epoch": 45.28786536758193,
      "grad_norm": 0.22940537333488464,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 51130
    },
    {
      "epoch": 45.29229406554473,
      "grad_norm": 0.22109197080135345,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 51135
    },
    {
      "epoch": 45.29672276350753,
      "grad_norm": 0.27144771814346313,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 51140
    },
    {
      "epoch": 45.301151461470326,
      "grad_norm": 0.23288264870643616,
      "learning_rate": 1e-05,
      "loss": 0.9191,
      "step": 51145
    },
    {
      "epoch": 45.30558015943313,
      "grad_norm": 0.2264794111251831,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 51150
    },
    {
      "epoch": 45.31000885739593,
      "grad_norm": 0.23706716299057007,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 51155
    },
    {
      "epoch": 45.31443755535872,
      "grad_norm": 0.2442762702703476,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 51160
    },
    {
      "epoch": 45.318866253321524,
      "grad_norm": 0.24715778231620789,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 51165
    },
    {
      "epoch": 45.323294951284325,
      "grad_norm": 0.31723394989967346,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 51170
    },
    {
      "epoch": 45.32772364924712,
      "grad_norm": 0.2172892540693283,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 51175
    },
    {
      "epoch": 45.33215234720992,
      "grad_norm": 0.24963992834091187,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 51180
    },
    {
      "epoch": 45.33658104517272,
      "grad_norm": 0.24094165861606598,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 51185
    },
    {
      "epoch": 45.341009743135515,
      "grad_norm": 0.20671811699867249,
      "learning_rate": 1e-05,
      "loss": 0.9333,
      "step": 51190
    },
    {
      "epoch": 45.34543844109832,
      "grad_norm": 0.25364503264427185,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 51195
    },
    {
      "epoch": 45.34986713906112,
      "grad_norm": 0.2803034484386444,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 51200
    },
    {
      "epoch": 45.35429583702391,
      "grad_norm": 0.24685046076774597,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 51205
    },
    {
      "epoch": 45.35872453498671,
      "grad_norm": 0.23616907000541687,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 51210
    },
    {
      "epoch": 45.363153232949514,
      "grad_norm": 0.2507687211036682,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 51215
    },
    {
      "epoch": 45.36758193091231,
      "grad_norm": 0.2694598436355591,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 51220
    },
    {
      "epoch": 45.37201062887511,
      "grad_norm": 0.2728506624698639,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 51225
    },
    {
      "epoch": 45.37643932683791,
      "grad_norm": 0.21888592839241028,
      "learning_rate": 1e-05,
      "loss": 0.9342,
      "step": 51230
    },
    {
      "epoch": 45.380868024800705,
      "grad_norm": 0.25766709446907043,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 51235
    },
    {
      "epoch": 45.385296722763506,
      "grad_norm": 0.28223270177841187,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 51240
    },
    {
      "epoch": 45.38972542072631,
      "grad_norm": 0.251563161611557,
      "learning_rate": 1e-05,
      "loss": 1.0292,
      "step": 51245
    },
    {
      "epoch": 45.39415411868911,
      "grad_norm": 0.24172428250312805,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 51250
    },
    {
      "epoch": 45.3985828166519,
      "grad_norm": 0.25851836800575256,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 51255
    },
    {
      "epoch": 45.403011514614704,
      "grad_norm": 0.21488241851329803,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 51260
    },
    {
      "epoch": 45.407440212577505,
      "grad_norm": 0.24836045503616333,
      "learning_rate": 1e-05,
      "loss": 0.9084,
      "step": 51265
    },
    {
      "epoch": 45.4118689105403,
      "grad_norm": 0.2217872142791748,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 51270
    },
    {
      "epoch": 45.4162976085031,
      "grad_norm": 0.2358526587486267,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 51275
    },
    {
      "epoch": 45.4207263064659,
      "grad_norm": 0.2298174649477005,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 51280
    },
    {
      "epoch": 45.425155004428696,
      "grad_norm": 0.22027640044689178,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 51285
    },
    {
      "epoch": 45.4295837023915,
      "grad_norm": 0.22802740335464478,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 51290
    },
    {
      "epoch": 45.4340124003543,
      "grad_norm": 0.25210607051849365,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 51295
    },
    {
      "epoch": 45.43844109831709,
      "grad_norm": 0.22045212984085083,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 51300
    },
    {
      "epoch": 45.44286979627989,
      "grad_norm": 0.21925128996372223,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 51305
    },
    {
      "epoch": 45.447298494242695,
      "grad_norm": 0.23857742547988892,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 51310
    },
    {
      "epoch": 45.45172719220549,
      "grad_norm": 0.2602781653404236,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 51315
    },
    {
      "epoch": 45.45615589016829,
      "grad_norm": 0.28282612562179565,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 51320
    },
    {
      "epoch": 45.46058458813109,
      "grad_norm": 0.2592891752719879,
      "learning_rate": 1e-05,
      "loss": 1.0222,
      "step": 51325
    },
    {
      "epoch": 45.465013286093885,
      "grad_norm": 0.2841559946537018,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 51330
    },
    {
      "epoch": 45.46944198405669,
      "grad_norm": 0.20972949266433716,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 51335
    },
    {
      "epoch": 45.47387068201949,
      "grad_norm": 0.20510976016521454,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 51340
    },
    {
      "epoch": 45.47829937998228,
      "grad_norm": 0.24025958776474,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 51345
    },
    {
      "epoch": 45.48272807794508,
      "grad_norm": 0.25290006399154663,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 51350
    },
    {
      "epoch": 45.487156775907884,
      "grad_norm": 0.2119912952184677,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 51355
    },
    {
      "epoch": 45.491585473870686,
      "grad_norm": 0.24001964926719666,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 51360
    },
    {
      "epoch": 45.49601417183348,
      "grad_norm": 0.22937263548374176,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 51365
    },
    {
      "epoch": 45.50044286979628,
      "grad_norm": 0.20668862760066986,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 51370
    },
    {
      "epoch": 45.50487156775908,
      "grad_norm": 0.26289650797843933,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 51375
    },
    {
      "epoch": 45.509300265721876,
      "grad_norm": 0.27244558930397034,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 51380
    },
    {
      "epoch": 45.51372896368468,
      "grad_norm": 0.2312142699956894,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 51385
    },
    {
      "epoch": 45.51815766164748,
      "grad_norm": 0.27039605379104614,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 51390
    },
    {
      "epoch": 45.52258635961027,
      "grad_norm": 0.27057579159736633,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 51395
    },
    {
      "epoch": 45.527015057573074,
      "grad_norm": 0.21195568144321442,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 51400
    },
    {
      "epoch": 45.531443755535875,
      "grad_norm": 0.22906166315078735,
      "learning_rate": 1e-05,
      "loss": 0.942,
      "step": 51405
    },
    {
      "epoch": 45.53587245349867,
      "grad_norm": 0.25505486130714417,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 51410
    },
    {
      "epoch": 45.54030115146147,
      "grad_norm": 0.2492789626121521,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 51415
    },
    {
      "epoch": 45.54472984942427,
      "grad_norm": 0.25532904267311096,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 51420
    },
    {
      "epoch": 45.549158547387066,
      "grad_norm": 0.23320272564888,
      "learning_rate": 1e-05,
      "loss": 0.9029,
      "step": 51425
    },
    {
      "epoch": 45.55358724534987,
      "grad_norm": 0.279902845621109,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 51430
    },
    {
      "epoch": 45.55801594331267,
      "grad_norm": 0.21254844963550568,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 51435
    },
    {
      "epoch": 45.56244464127546,
      "grad_norm": 0.2398146092891693,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 51440
    },
    {
      "epoch": 45.56687333923826,
      "grad_norm": 0.2217506319284439,
      "learning_rate": 1e-05,
      "loss": 1.0097,
      "step": 51445
    },
    {
      "epoch": 45.571302037201065,
      "grad_norm": 0.26074621081352234,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 51450
    },
    {
      "epoch": 45.57573073516386,
      "grad_norm": 0.2479376345872879,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 51455
    },
    {
      "epoch": 45.58015943312666,
      "grad_norm": 0.2737751603126526,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 51460
    },
    {
      "epoch": 45.58458813108946,
      "grad_norm": 0.27061790227890015,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 51465
    },
    {
      "epoch": 45.589016829052255,
      "grad_norm": 0.2958587408065796,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 51470
    },
    {
      "epoch": 45.59344552701506,
      "grad_norm": 0.3015517592430115,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 51475
    },
    {
      "epoch": 45.59787422497786,
      "grad_norm": 0.2944192588329315,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 51480
    },
    {
      "epoch": 45.60230292294066,
      "grad_norm": 0.26193124055862427,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 51485
    },
    {
      "epoch": 45.60673162090345,
      "grad_norm": 0.2511567175388336,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 51490
    },
    {
      "epoch": 45.611160318866254,
      "grad_norm": 0.20771130919456482,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 51495
    },
    {
      "epoch": 45.615589016829055,
      "grad_norm": 0.26946723461151123,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 51500
    },
    {
      "epoch": 45.62001771479185,
      "grad_norm": 0.23836848139762878,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 51505
    },
    {
      "epoch": 45.62444641275465,
      "grad_norm": 0.21833090484142303,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 51510
    },
    {
      "epoch": 45.62887511071745,
      "grad_norm": 0.24224352836608887,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 51515
    },
    {
      "epoch": 45.633303808680246,
      "grad_norm": 0.22550438344478607,
      "learning_rate": 1e-05,
      "loss": 1.0255,
      "step": 51520
    },
    {
      "epoch": 45.63773250664305,
      "grad_norm": 0.21138441562652588,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 51525
    },
    {
      "epoch": 45.64216120460585,
      "grad_norm": 0.2626248598098755,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 51530
    },
    {
      "epoch": 45.64658990256864,
      "grad_norm": 0.3008551001548767,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 51535
    },
    {
      "epoch": 45.651018600531444,
      "grad_norm": 0.2048415243625641,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 51540
    },
    {
      "epoch": 45.655447298494245,
      "grad_norm": 0.23698744177818298,
      "learning_rate": 1e-05,
      "loss": 0.9038,
      "step": 51545
    },
    {
      "epoch": 45.65987599645704,
      "grad_norm": 0.2327071577310562,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 51550
    },
    {
      "epoch": 45.66430469441984,
      "grad_norm": 0.3224489092826843,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 51555
    },
    {
      "epoch": 45.66873339238264,
      "grad_norm": 0.2258215844631195,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 51560
    },
    {
      "epoch": 45.673162090345436,
      "grad_norm": 0.2476758062839508,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 51565
    },
    {
      "epoch": 45.67759078830824,
      "grad_norm": 0.2695644795894623,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 51570
    },
    {
      "epoch": 45.68201948627104,
      "grad_norm": 0.2607501447200775,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 51575
    },
    {
      "epoch": 45.68644818423383,
      "grad_norm": 0.21129955351352692,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 51580
    },
    {
      "epoch": 45.69087688219663,
      "grad_norm": 0.2089204490184784,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 51585
    },
    {
      "epoch": 45.695305580159435,
      "grad_norm": 0.234258770942688,
      "learning_rate": 1e-05,
      "loss": 0.9215,
      "step": 51590
    },
    {
      "epoch": 45.69973427812223,
      "grad_norm": 0.24929247796535492,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 51595
    },
    {
      "epoch": 45.70416297608503,
      "grad_norm": 0.2664491832256317,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 51600
    },
    {
      "epoch": 45.70859167404783,
      "grad_norm": 0.23142677545547485,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 51605
    },
    {
      "epoch": 45.71302037201063,
      "grad_norm": 0.23572951555252075,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 51610
    },
    {
      "epoch": 45.717449069973426,
      "grad_norm": 0.2164895087480545,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 51615
    },
    {
      "epoch": 45.72187776793623,
      "grad_norm": 0.23070400953292847,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 51620
    },
    {
      "epoch": 45.72630646589903,
      "grad_norm": 0.2819378674030304,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 51625
    },
    {
      "epoch": 45.73073516386182,
      "grad_norm": 0.22256404161453247,
      "learning_rate": 1e-05,
      "loss": 0.9002,
      "step": 51630
    },
    {
      "epoch": 45.735163861824624,
      "grad_norm": 0.2503831386566162,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 51635
    },
    {
      "epoch": 45.739592559787425,
      "grad_norm": 0.2594408392906189,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 51640
    },
    {
      "epoch": 45.74402125775022,
      "grad_norm": 0.2545607089996338,
      "learning_rate": 1e-05,
      "loss": 0.8817,
      "step": 51645
    },
    {
      "epoch": 45.74844995571302,
      "grad_norm": 0.21343974769115448,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 51650
    },
    {
      "epoch": 45.75287865367582,
      "grad_norm": 0.24160389602184296,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 51655
    },
    {
      "epoch": 45.757307351638616,
      "grad_norm": 0.2479962408542633,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 51660
    },
    {
      "epoch": 45.76173604960142,
      "grad_norm": 0.2532936930656433,
      "learning_rate": 1e-05,
      "loss": 0.9254,
      "step": 51665
    },
    {
      "epoch": 45.76616474756422,
      "grad_norm": 0.22991423308849335,
      "learning_rate": 1e-05,
      "loss": 0.921,
      "step": 51670
    },
    {
      "epoch": 45.77059344552701,
      "grad_norm": 0.23111528158187866,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 51675
    },
    {
      "epoch": 45.775022143489814,
      "grad_norm": 0.23853762447834015,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 51680
    },
    {
      "epoch": 45.779450841452615,
      "grad_norm": 0.28785794973373413,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 51685
    },
    {
      "epoch": 45.78387953941541,
      "grad_norm": 0.20950739085674286,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 51690
    },
    {
      "epoch": 45.78830823737821,
      "grad_norm": 0.2349524199962616,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 51695
    },
    {
      "epoch": 45.79273693534101,
      "grad_norm": 0.22990046441555023,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 51700
    },
    {
      "epoch": 45.797165633303806,
      "grad_norm": 0.23833337426185608,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 51705
    },
    {
      "epoch": 45.80159433126661,
      "grad_norm": 0.2167847752571106,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 51710
    },
    {
      "epoch": 45.80602302922941,
      "grad_norm": 0.24349065124988556,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 51715
    },
    {
      "epoch": 45.8104517271922,
      "grad_norm": 0.24023036658763885,
      "learning_rate": 1e-05,
      "loss": 1.0219,
      "step": 51720
    },
    {
      "epoch": 45.814880425155,
      "grad_norm": 0.21085967123508453,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 51725
    },
    {
      "epoch": 45.819309123117804,
      "grad_norm": 0.20929795503616333,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 51730
    },
    {
      "epoch": 45.823737821080606,
      "grad_norm": 0.22110675275325775,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 51735
    },
    {
      "epoch": 45.8281665190434,
      "grad_norm": 0.2524730861186981,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 51740
    },
    {
      "epoch": 45.8325952170062,
      "grad_norm": 0.24815669655799866,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 51745
    },
    {
      "epoch": 45.837023914969,
      "grad_norm": 0.2149936705827713,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 51750
    },
    {
      "epoch": 45.841452612931796,
      "grad_norm": 0.2428337037563324,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 51755
    },
    {
      "epoch": 45.8458813108946,
      "grad_norm": 0.23148475587368011,
      "learning_rate": 1e-05,
      "loss": 0.9338,
      "step": 51760
    },
    {
      "epoch": 45.8503100088574,
      "grad_norm": 0.23069745302200317,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 51765
    },
    {
      "epoch": 45.85473870682019,
      "grad_norm": 0.24695904552936554,
      "learning_rate": 1e-05,
      "loss": 0.93,
      "step": 51770
    },
    {
      "epoch": 45.859167404782994,
      "grad_norm": 0.25191372632980347,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 51775
    },
    {
      "epoch": 45.863596102745795,
      "grad_norm": 0.23737689852714539,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 51780
    },
    {
      "epoch": 45.86802480070859,
      "grad_norm": 0.27893298864364624,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 51785
    },
    {
      "epoch": 45.87245349867139,
      "grad_norm": 0.2078242301940918,
      "learning_rate": 1e-05,
      "loss": 0.9174,
      "step": 51790
    },
    {
      "epoch": 45.87688219663419,
      "grad_norm": 0.2302054464817047,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 51795
    },
    {
      "epoch": 45.881310894596986,
      "grad_norm": 0.2452826052904129,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 51800
    },
    {
      "epoch": 45.88573959255979,
      "grad_norm": 0.2006453573703766,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 51805
    },
    {
      "epoch": 45.89016829052259,
      "grad_norm": 0.25416144728660583,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 51810
    },
    {
      "epoch": 45.89459698848538,
      "grad_norm": 0.24128666520118713,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 51815
    },
    {
      "epoch": 45.899025686448184,
      "grad_norm": 0.24687378108501434,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 51820
    },
    {
      "epoch": 45.903454384410985,
      "grad_norm": 0.22308023273944855,
      "learning_rate": 1e-05,
      "loss": 0.9159,
      "step": 51825
    },
    {
      "epoch": 45.90788308237378,
      "grad_norm": 0.22135116159915924,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 51830
    },
    {
      "epoch": 45.91231178033658,
      "grad_norm": 0.21836702525615692,
      "learning_rate": 1e-05,
      "loss": 0.8916,
      "step": 51835
    },
    {
      "epoch": 45.91674047829938,
      "grad_norm": 0.2132343053817749,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 51840
    },
    {
      "epoch": 45.921169176262175,
      "grad_norm": 0.22857122123241425,
      "learning_rate": 1e-05,
      "loss": 1.0141,
      "step": 51845
    },
    {
      "epoch": 45.92559787422498,
      "grad_norm": 0.22367356717586517,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 51850
    },
    {
      "epoch": 45.93002657218778,
      "grad_norm": 0.23527488112449646,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 51855
    },
    {
      "epoch": 45.93445527015058,
      "grad_norm": 0.2949398458003998,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 51860
    },
    {
      "epoch": 45.93888396811337,
      "grad_norm": 0.21424800157546997,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 51865
    },
    {
      "epoch": 45.943312666076174,
      "grad_norm": 0.2660035490989685,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 51870
    },
    {
      "epoch": 45.947741364038976,
      "grad_norm": 0.2730546295642853,
      "learning_rate": 1e-05,
      "loss": 1.0172,
      "step": 51875
    },
    {
      "epoch": 45.95217006200177,
      "grad_norm": 0.23108869791030884,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 51880
    },
    {
      "epoch": 45.95659875996457,
      "grad_norm": 0.22088433802127838,
      "learning_rate": 1e-05,
      "loss": 0.9066,
      "step": 51885
    },
    {
      "epoch": 45.96102745792737,
      "grad_norm": 0.20338571071624756,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 51890
    },
    {
      "epoch": 45.965456155890166,
      "grad_norm": 0.22179971635341644,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 51895
    },
    {
      "epoch": 45.96988485385297,
      "grad_norm": 0.22934699058532715,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 51900
    },
    {
      "epoch": 45.97431355181577,
      "grad_norm": 0.25358590483665466,
      "learning_rate": 1e-05,
      "loss": 1.039,
      "step": 51905
    },
    {
      "epoch": 45.97874224977856,
      "grad_norm": 0.31548011302948,
      "learning_rate": 1e-05,
      "loss": 0.9132,
      "step": 51910
    },
    {
      "epoch": 45.983170947741364,
      "grad_norm": 0.2892420291900635,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 51915
    },
    {
      "epoch": 45.987599645704165,
      "grad_norm": 0.24548064172267914,
      "learning_rate": 1e-05,
      "loss": 0.9275,
      "step": 51920
    },
    {
      "epoch": 45.99202834366696,
      "grad_norm": 0.2229636162519455,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 51925
    },
    {
      "epoch": 45.99645704162976,
      "grad_norm": 0.24473243951797485,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 51930
    },
    {
      "epoch": 46.00088573959256,
      "grad_norm": 0.22059190273284912,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 51935
    },
    {
      "epoch": 46.005314437555356,
      "grad_norm": 0.23424828052520752,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 51940
    },
    {
      "epoch": 46.00974313551816,
      "grad_norm": 0.24413150548934937,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 51945
    },
    {
      "epoch": 46.01417183348096,
      "grad_norm": 0.29263216257095337,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 51950
    },
    {
      "epoch": 46.01860053144375,
      "grad_norm": 0.2872030735015869,
      "learning_rate": 1e-05,
      "loss": 1.0185,
      "step": 51955
    },
    {
      "epoch": 46.02302922940655,
      "grad_norm": 0.25302475690841675,
      "learning_rate": 1e-05,
      "loss": 0.9227,
      "step": 51960
    },
    {
      "epoch": 46.027457927369355,
      "grad_norm": 0.27512097358703613,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 51965
    },
    {
      "epoch": 46.03188662533215,
      "grad_norm": 0.28898295760154724,
      "learning_rate": 1e-05,
      "loss": 0.942,
      "step": 51970
    },
    {
      "epoch": 46.03631532329495,
      "grad_norm": 0.23756752908229828,
      "learning_rate": 1e-05,
      "loss": 0.9105,
      "step": 51975
    },
    {
      "epoch": 46.04074402125775,
      "grad_norm": 0.24522282183170319,
      "learning_rate": 1e-05,
      "loss": 0.926,
      "step": 51980
    },
    {
      "epoch": 46.04517271922055,
      "grad_norm": 0.2577900290489197,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 51985
    },
    {
      "epoch": 46.04960141718335,
      "grad_norm": 0.23627310991287231,
      "learning_rate": 1e-05,
      "loss": 1.0009,
      "step": 51990
    },
    {
      "epoch": 46.05403011514615,
      "grad_norm": 0.22613753378391266,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 51995
    },
    {
      "epoch": 46.05845881310895,
      "grad_norm": 0.23439499735832214,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 52000
    },
    {
      "epoch": 46.06288751107174,
      "grad_norm": 0.2566295266151428,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 52005
    },
    {
      "epoch": 46.067316209034544,
      "grad_norm": 0.2557554841041565,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 52010
    },
    {
      "epoch": 46.071744906997345,
      "grad_norm": 0.2788117229938507,
      "learning_rate": 1e-05,
      "loss": 1.0405,
      "step": 52015
    },
    {
      "epoch": 46.07617360496014,
      "grad_norm": 0.2325088381767273,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 52020
    },
    {
      "epoch": 46.08060230292294,
      "grad_norm": 0.25615447759628296,
      "learning_rate": 1e-05,
      "loss": 1.0299,
      "step": 52025
    },
    {
      "epoch": 46.08503100088574,
      "grad_norm": 0.23676101863384247,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 52030
    },
    {
      "epoch": 46.089459698848536,
      "grad_norm": 0.2522531747817993,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 52035
    },
    {
      "epoch": 46.09388839681134,
      "grad_norm": 0.2111278623342514,
      "learning_rate": 1e-05,
      "loss": 1.0284,
      "step": 52040
    },
    {
      "epoch": 46.09831709477414,
      "grad_norm": 0.2221577763557434,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 52045
    },
    {
      "epoch": 46.10274579273693,
      "grad_norm": 0.22191591560840607,
      "learning_rate": 1e-05,
      "loss": 0.9161,
      "step": 52050
    },
    {
      "epoch": 46.107174490699734,
      "grad_norm": 0.2434382140636444,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 52055
    },
    {
      "epoch": 46.111603188662535,
      "grad_norm": 0.23474010825157166,
      "learning_rate": 1e-05,
      "loss": 1.0408,
      "step": 52060
    },
    {
      "epoch": 46.11603188662533,
      "grad_norm": 0.2551935017108917,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 52065
    },
    {
      "epoch": 46.12046058458813,
      "grad_norm": 0.220155268907547,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 52070
    },
    {
      "epoch": 46.12488928255093,
      "grad_norm": 0.2255057990550995,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 52075
    },
    {
      "epoch": 46.129317980513726,
      "grad_norm": 0.26529330015182495,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 52080
    },
    {
      "epoch": 46.13374667847653,
      "grad_norm": 0.23907260596752167,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 52085
    },
    {
      "epoch": 46.13817537643933,
      "grad_norm": 0.27552422881126404,
      "learning_rate": 1e-05,
      "loss": 1.028,
      "step": 52090
    },
    {
      "epoch": 46.14260407440213,
      "grad_norm": 0.23915618658065796,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 52095
    },
    {
      "epoch": 46.14703277236492,
      "grad_norm": 0.24072015285491943,
      "learning_rate": 1e-05,
      "loss": 0.9137,
      "step": 52100
    },
    {
      "epoch": 46.151461470327725,
      "grad_norm": 0.22663548588752747,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 52105
    },
    {
      "epoch": 46.155890168290526,
      "grad_norm": 0.2206011712551117,
      "learning_rate": 1e-05,
      "loss": 0.9214,
      "step": 52110
    },
    {
      "epoch": 46.16031886625332,
      "grad_norm": 0.23620931804180145,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 52115
    },
    {
      "epoch": 46.16474756421612,
      "grad_norm": 0.2478964924812317,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 52120
    },
    {
      "epoch": 46.16917626217892,
      "grad_norm": 0.24203436076641083,
      "learning_rate": 1e-05,
      "loss": 1.0389,
      "step": 52125
    },
    {
      "epoch": 46.173604960141716,
      "grad_norm": 0.20357997715473175,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 52130
    },
    {
      "epoch": 46.17803365810452,
      "grad_norm": 0.24347221851348877,
      "learning_rate": 1e-05,
      "loss": 0.9113,
      "step": 52135
    },
    {
      "epoch": 46.18246235606732,
      "grad_norm": 0.24830567836761475,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 52140
    },
    {
      "epoch": 46.18689105403011,
      "grad_norm": 0.25571784377098083,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 52145
    },
    {
      "epoch": 46.191319751992914,
      "grad_norm": 0.2986680567264557,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 52150
    },
    {
      "epoch": 46.195748449955715,
      "grad_norm": 0.2642230987548828,
      "learning_rate": 1e-05,
      "loss": 1.0307,
      "step": 52155
    },
    {
      "epoch": 46.20017714791851,
      "grad_norm": 0.22545696794986725,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 52160
    },
    {
      "epoch": 46.20460584588131,
      "grad_norm": 0.21802228689193726,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 52165
    },
    {
      "epoch": 46.20903454384411,
      "grad_norm": 0.22780759632587433,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 52170
    },
    {
      "epoch": 46.213463241806906,
      "grad_norm": 0.21695683896541595,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 52175
    },
    {
      "epoch": 46.21789193976971,
      "grad_norm": 0.24940991401672363,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 52180
    },
    {
      "epoch": 46.22232063773251,
      "grad_norm": 0.2487255185842514,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 52185
    },
    {
      "epoch": 46.2267493356953,
      "grad_norm": 0.20448024570941925,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 52190
    },
    {
      "epoch": 46.231178033658104,
      "grad_norm": 0.2528068423271179,
      "learning_rate": 1e-05,
      "loss": 1.0509,
      "step": 52195
    },
    {
      "epoch": 46.235606731620905,
      "grad_norm": 0.22307176887989044,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 52200
    },
    {
      "epoch": 46.2400354295837,
      "grad_norm": 0.2304147630929947,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 52205
    },
    {
      "epoch": 46.2444641275465,
      "grad_norm": 0.2592250108718872,
      "learning_rate": 1e-05,
      "loss": 1.0273,
      "step": 52210
    },
    {
      "epoch": 46.2488928255093,
      "grad_norm": 0.25643765926361084,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 52215
    },
    {
      "epoch": 46.2533215234721,
      "grad_norm": 0.265780508518219,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 52220
    },
    {
      "epoch": 46.2577502214349,
      "grad_norm": 0.24850358068943024,
      "learning_rate": 1e-05,
      "loss": 0.9181,
      "step": 52225
    },
    {
      "epoch": 46.2621789193977,
      "grad_norm": 0.24540919065475464,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 52230
    },
    {
      "epoch": 46.2666076173605,
      "grad_norm": 0.2508479952812195,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 52235
    },
    {
      "epoch": 46.27103631532329,
      "grad_norm": 0.22535806894302368,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 52240
    },
    {
      "epoch": 46.275465013286095,
      "grad_norm": 0.2499355971813202,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 52245
    },
    {
      "epoch": 46.279893711248896,
      "grad_norm": 0.25124189257621765,
      "learning_rate": 1e-05,
      "loss": 0.9427,
      "step": 52250
    },
    {
      "epoch": 46.28432240921169,
      "grad_norm": 0.2670814096927643,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 52255
    },
    {
      "epoch": 46.28875110717449,
      "grad_norm": 0.25615066289901733,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 52260
    },
    {
      "epoch": 46.29317980513729,
      "grad_norm": 0.23560094833374023,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 52265
    },
    {
      "epoch": 46.297608503100086,
      "grad_norm": 0.25396451354026794,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 52270
    },
    {
      "epoch": 46.30203720106289,
      "grad_norm": 0.22137683629989624,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 52275
    },
    {
      "epoch": 46.30646589902569,
      "grad_norm": 0.23736625909805298,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 52280
    },
    {
      "epoch": 46.31089459698848,
      "grad_norm": 0.2688720226287842,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 52285
    },
    {
      "epoch": 46.315323294951284,
      "grad_norm": 0.2810586392879486,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 52290
    },
    {
      "epoch": 46.319751992914085,
      "grad_norm": 0.27852848172187805,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 52295
    },
    {
      "epoch": 46.32418069087688,
      "grad_norm": 0.28398391604423523,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 52300
    },
    {
      "epoch": 46.32860938883968,
      "grad_norm": 0.26669609546661377,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 52305
    },
    {
      "epoch": 46.33303808680248,
      "grad_norm": 0.27653294801712036,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 52310
    },
    {
      "epoch": 46.337466784765276,
      "grad_norm": 0.25108420848846436,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 52315
    },
    {
      "epoch": 46.34189548272808,
      "grad_norm": 0.267616868019104,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 52320
    },
    {
      "epoch": 46.34632418069088,
      "grad_norm": 0.22993402183055878,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 52325
    },
    {
      "epoch": 46.35075287865367,
      "grad_norm": 0.25327572226524353,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 52330
    },
    {
      "epoch": 46.355181576616474,
      "grad_norm": 0.25147002935409546,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 52335
    },
    {
      "epoch": 46.359610274579275,
      "grad_norm": 0.3183804750442505,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 52340
    },
    {
      "epoch": 46.364038972542076,
      "grad_norm": 0.27869829535484314,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 52345
    },
    {
      "epoch": 46.36846767050487,
      "grad_norm": 0.2506049573421478,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 52350
    },
    {
      "epoch": 46.37289636846767,
      "grad_norm": 0.23751741647720337,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 52355
    },
    {
      "epoch": 46.37732506643047,
      "grad_norm": 0.33607232570648193,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 52360
    },
    {
      "epoch": 46.38175376439327,
      "grad_norm": 0.23826178908348083,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 52365
    },
    {
      "epoch": 46.38618246235607,
      "grad_norm": 0.26693665981292725,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 52370
    },
    {
      "epoch": 46.39061116031887,
      "grad_norm": 0.24751687049865723,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 52375
    },
    {
      "epoch": 46.39503985828166,
      "grad_norm": 0.2192055881023407,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 52380
    },
    {
      "epoch": 46.399468556244464,
      "grad_norm": 0.2583507001399994,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 52385
    },
    {
      "epoch": 46.403897254207266,
      "grad_norm": 0.23104475438594818,
      "learning_rate": 1e-05,
      "loss": 0.9115,
      "step": 52390
    },
    {
      "epoch": 46.40832595217006,
      "grad_norm": 0.2703949511051178,
      "learning_rate": 1e-05,
      "loss": 1.0194,
      "step": 52395
    },
    {
      "epoch": 46.41275465013286,
      "grad_norm": 0.24414384365081787,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 52400
    },
    {
      "epoch": 46.41718334809566,
      "grad_norm": 0.22799710929393768,
      "learning_rate": 1e-05,
      "loss": 1.023,
      "step": 52405
    },
    {
      "epoch": 46.421612046058456,
      "grad_norm": 0.23787589371204376,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 52410
    },
    {
      "epoch": 46.42604074402126,
      "grad_norm": 0.2353285402059555,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 52415
    },
    {
      "epoch": 46.43046944198406,
      "grad_norm": 0.3217039704322815,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 52420
    },
    {
      "epoch": 46.43489813994685,
      "grad_norm": 0.25284624099731445,
      "learning_rate": 1e-05,
      "loss": 0.9211,
      "step": 52425
    },
    {
      "epoch": 46.439326837909654,
      "grad_norm": 0.2597035765647888,
      "learning_rate": 1e-05,
      "loss": 1.0327,
      "step": 52430
    },
    {
      "epoch": 46.443755535872455,
      "grad_norm": 0.2135528326034546,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 52435
    },
    {
      "epoch": 46.44818423383525,
      "grad_norm": 0.281564861536026,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 52440
    },
    {
      "epoch": 46.45261293179805,
      "grad_norm": 0.21969282627105713,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 52445
    },
    {
      "epoch": 46.45704162976085,
      "grad_norm": 0.259161114692688,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 52450
    },
    {
      "epoch": 46.461470327723646,
      "grad_norm": 0.22296705842018127,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 52455
    },
    {
      "epoch": 46.46589902568645,
      "grad_norm": 0.22984325885772705,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 52460
    },
    {
      "epoch": 46.47032772364925,
      "grad_norm": 0.2613532841205597,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 52465
    },
    {
      "epoch": 46.47475642161205,
      "grad_norm": 0.22698812186717987,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 52470
    },
    {
      "epoch": 46.479185119574844,
      "grad_norm": 0.20282553136348724,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 52475
    },
    {
      "epoch": 46.483613817537645,
      "grad_norm": 0.22666481137275696,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 52480
    },
    {
      "epoch": 46.488042515500446,
      "grad_norm": 0.2294263243675232,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 52485
    },
    {
      "epoch": 46.49247121346324,
      "grad_norm": 0.21758776903152466,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 52490
    },
    {
      "epoch": 46.49689991142604,
      "grad_norm": 0.23589198291301727,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 52495
    },
    {
      "epoch": 46.50132860938884,
      "grad_norm": 0.21198731660842896,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 52500
    },
    {
      "epoch": 46.50575730735164,
      "grad_norm": 0.21669331192970276,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 52505
    },
    {
      "epoch": 46.51018600531444,
      "grad_norm": 0.18793557584285736,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 52510
    },
    {
      "epoch": 46.51461470327724,
      "grad_norm": 0.2524021863937378,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 52515
    },
    {
      "epoch": 46.51904340124003,
      "grad_norm": 0.25322335958480835,
      "learning_rate": 1e-05,
      "loss": 1.0375,
      "step": 52520
    },
    {
      "epoch": 46.523472099202834,
      "grad_norm": 0.2473878711462021,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 52525
    },
    {
      "epoch": 46.527900797165636,
      "grad_norm": 0.2516082227230072,
      "learning_rate": 1e-05,
      "loss": 0.9211,
      "step": 52530
    },
    {
      "epoch": 46.53232949512843,
      "grad_norm": 0.2698797881603241,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 52535
    },
    {
      "epoch": 46.53675819309123,
      "grad_norm": 0.22685733437538147,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 52540
    },
    {
      "epoch": 46.54118689105403,
      "grad_norm": 0.23823480308055878,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 52545
    },
    {
      "epoch": 46.545615589016826,
      "grad_norm": 0.20379240810871124,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 52550
    },
    {
      "epoch": 46.55004428697963,
      "grad_norm": 0.2341829538345337,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 52555
    },
    {
      "epoch": 46.55447298494243,
      "grad_norm": 0.23614032566547394,
      "learning_rate": 1e-05,
      "loss": 1.0101,
      "step": 52560
    },
    {
      "epoch": 46.55890168290522,
      "grad_norm": 0.2548588812351227,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 52565
    },
    {
      "epoch": 46.563330380868024,
      "grad_norm": 0.26242223381996155,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 52570
    },
    {
      "epoch": 46.567759078830825,
      "grad_norm": 0.30709484219551086,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 52575
    },
    {
      "epoch": 46.57218777679362,
      "grad_norm": 0.2339414805173874,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 52580
    },
    {
      "epoch": 46.57661647475642,
      "grad_norm": 0.25077104568481445,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 52585
    },
    {
      "epoch": 46.58104517271922,
      "grad_norm": 0.23809531331062317,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 52590
    },
    {
      "epoch": 46.58547387068202,
      "grad_norm": 0.26254934072494507,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 52595
    },
    {
      "epoch": 46.58990256864482,
      "grad_norm": 0.24272571504116058,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 52600
    },
    {
      "epoch": 46.59433126660762,
      "grad_norm": 0.25622281432151794,
      "learning_rate": 1e-05,
      "loss": 1.0159,
      "step": 52605
    },
    {
      "epoch": 46.59875996457042,
      "grad_norm": 0.2809261083602905,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 52610
    },
    {
      "epoch": 46.60318866253321,
      "grad_norm": 0.26929810643196106,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 52615
    },
    {
      "epoch": 46.607617360496015,
      "grad_norm": 0.209640234708786,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 52620
    },
    {
      "epoch": 46.612046058458816,
      "grad_norm": 0.24667634069919586,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 52625
    },
    {
      "epoch": 46.61647475642161,
      "grad_norm": 0.24101629853248596,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 52630
    },
    {
      "epoch": 46.62090345438441,
      "grad_norm": 0.2497033178806305,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 52635
    },
    {
      "epoch": 46.62533215234721,
      "grad_norm": 0.25059953331947327,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 52640
    },
    {
      "epoch": 46.62976085031001,
      "grad_norm": 0.25847601890563965,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 52645
    },
    {
      "epoch": 46.63418954827281,
      "grad_norm": 0.2661818861961365,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 52650
    },
    {
      "epoch": 46.63861824623561,
      "grad_norm": 0.2577047049999237,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 52655
    },
    {
      "epoch": 46.6430469441984,
      "grad_norm": 0.25426873564720154,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 52660
    },
    {
      "epoch": 46.647475642161204,
      "grad_norm": 0.22114379703998566,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 52665
    },
    {
      "epoch": 46.651904340124005,
      "grad_norm": 0.27035418152809143,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 52670
    },
    {
      "epoch": 46.6563330380868,
      "grad_norm": 0.256034255027771,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 52675
    },
    {
      "epoch": 46.6607617360496,
      "grad_norm": 0.29731854796409607,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 52680
    },
    {
      "epoch": 46.6651904340124,
      "grad_norm": 0.23128163814544678,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 52685
    },
    {
      "epoch": 46.669619131975196,
      "grad_norm": 0.22359953820705414,
      "learning_rate": 1e-05,
      "loss": 0.909,
      "step": 52690
    },
    {
      "epoch": 46.674047829938,
      "grad_norm": 0.21049128472805023,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 52695
    },
    {
      "epoch": 46.6784765279008,
      "grad_norm": 0.26686444878578186,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 52700
    },
    {
      "epoch": 46.68290522586359,
      "grad_norm": 0.29327306151390076,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 52705
    },
    {
      "epoch": 46.687333923826394,
      "grad_norm": 0.264400839805603,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 52710
    },
    {
      "epoch": 46.691762621789195,
      "grad_norm": 0.28055617213249207,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 52715
    },
    {
      "epoch": 46.696191319751996,
      "grad_norm": 0.24878482520580292,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 52720
    },
    {
      "epoch": 46.70062001771479,
      "grad_norm": 0.26178857684135437,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 52725
    },
    {
      "epoch": 46.70504871567759,
      "grad_norm": 0.25145775079727173,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 52730
    },
    {
      "epoch": 46.70947741364039,
      "grad_norm": 0.21283026039600372,
      "learning_rate": 1e-05,
      "loss": 1.0149,
      "step": 52735
    },
    {
      "epoch": 46.71390611160319,
      "grad_norm": 0.2254999727010727,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 52740
    },
    {
      "epoch": 46.71833480956599,
      "grad_norm": 0.1945682168006897,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 52745
    },
    {
      "epoch": 46.72276350752879,
      "grad_norm": 0.2572639286518097,
      "learning_rate": 1e-05,
      "loss": 1.0452,
      "step": 52750
    },
    {
      "epoch": 46.72719220549158,
      "grad_norm": 0.2676490545272827,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 52755
    },
    {
      "epoch": 46.731620903454385,
      "grad_norm": 0.25357845425605774,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 52760
    },
    {
      "epoch": 46.736049601417186,
      "grad_norm": 0.2400568425655365,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 52765
    },
    {
      "epoch": 46.74047829937998,
      "grad_norm": 0.26042479276657104,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 52770
    },
    {
      "epoch": 46.74490699734278,
      "grad_norm": 0.24639937281608582,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 52775
    },
    {
      "epoch": 46.74933569530558,
      "grad_norm": 0.2331128567457199,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 52780
    },
    {
      "epoch": 46.753764393268376,
      "grad_norm": 0.2004622220993042,
      "learning_rate": 1e-05,
      "loss": 0.9028,
      "step": 52785
    },
    {
      "epoch": 46.75819309123118,
      "grad_norm": 0.23124127089977264,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 52790
    },
    {
      "epoch": 46.76262178919398,
      "grad_norm": 0.22730982303619385,
      "learning_rate": 1e-05,
      "loss": 1.022,
      "step": 52795
    },
    {
      "epoch": 46.76705048715677,
      "grad_norm": 0.2400243580341339,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 52800
    },
    {
      "epoch": 46.771479185119574,
      "grad_norm": 0.23408199846744537,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 52805
    },
    {
      "epoch": 46.775907883082375,
      "grad_norm": 0.2206108272075653,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 52810
    },
    {
      "epoch": 46.78033658104517,
      "grad_norm": 0.2546866238117218,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 52815
    },
    {
      "epoch": 46.78476527900797,
      "grad_norm": 0.24876323342323303,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 52820
    },
    {
      "epoch": 46.78919397697077,
      "grad_norm": 0.21203528344631195,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 52825
    },
    {
      "epoch": 46.79362267493357,
      "grad_norm": 0.2326049953699112,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 52830
    },
    {
      "epoch": 46.79805137289637,
      "grad_norm": 0.21757188439369202,
      "learning_rate": 1e-05,
      "loss": 1.0436,
      "step": 52835
    },
    {
      "epoch": 46.80248007085917,
      "grad_norm": 0.2163865715265274,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 52840
    },
    {
      "epoch": 46.80690876882197,
      "grad_norm": 0.2674662470817566,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 52845
    },
    {
      "epoch": 46.811337466784764,
      "grad_norm": 0.26493677496910095,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 52850
    },
    {
      "epoch": 46.815766164747565,
      "grad_norm": 0.27579277753829956,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 52855
    },
    {
      "epoch": 46.820194862710366,
      "grad_norm": 0.24195124208927155,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 52860
    },
    {
      "epoch": 46.82462356067316,
      "grad_norm": 0.2612171769142151,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 52865
    },
    {
      "epoch": 46.82905225863596,
      "grad_norm": 0.2395055890083313,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 52870
    },
    {
      "epoch": 46.83348095659876,
      "grad_norm": 0.23697564005851746,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 52875
    },
    {
      "epoch": 46.83790965456156,
      "grad_norm": 0.2651281952857971,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 52880
    },
    {
      "epoch": 46.84233835252436,
      "grad_norm": 0.23796038329601288,
      "learning_rate": 1e-05,
      "loss": 1.0297,
      "step": 52885
    },
    {
      "epoch": 46.84676705048716,
      "grad_norm": 0.23892737925052643,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 52890
    },
    {
      "epoch": 46.85119574844995,
      "grad_norm": 0.28136616945266724,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 52895
    },
    {
      "epoch": 46.855624446412754,
      "grad_norm": 0.24340462684631348,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 52900
    },
    {
      "epoch": 46.860053144375556,
      "grad_norm": 0.2332012802362442,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 52905
    },
    {
      "epoch": 46.86448184233835,
      "grad_norm": 0.21644599735736847,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 52910
    },
    {
      "epoch": 46.86891054030115,
      "grad_norm": 0.2537345588207245,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 52915
    },
    {
      "epoch": 46.87333923826395,
      "grad_norm": 0.2520318627357483,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 52920
    },
    {
      "epoch": 46.877767936226746,
      "grad_norm": 0.241068035364151,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 52925
    },
    {
      "epoch": 46.88219663418955,
      "grad_norm": 0.21747778356075287,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 52930
    },
    {
      "epoch": 46.88662533215235,
      "grad_norm": 0.22705544531345367,
      "learning_rate": 1e-05,
      "loss": 0.9215,
      "step": 52935
    },
    {
      "epoch": 46.89105403011514,
      "grad_norm": 0.24327921867370605,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 52940
    },
    {
      "epoch": 46.895482728077944,
      "grad_norm": 0.21224263310432434,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 52945
    },
    {
      "epoch": 46.899911426040745,
      "grad_norm": 0.28321564197540283,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 52950
    },
    {
      "epoch": 46.90434012400354,
      "grad_norm": 0.24056753516197205,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 52955
    },
    {
      "epoch": 46.90876882196634,
      "grad_norm": 0.3190236985683441,
      "learning_rate": 1e-05,
      "loss": 0.9294,
      "step": 52960
    },
    {
      "epoch": 46.91319751992914,
      "grad_norm": 0.2720867991447449,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 52965
    },
    {
      "epoch": 46.91762621789194,
      "grad_norm": 0.22937411069869995,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 52970
    },
    {
      "epoch": 46.92205491585474,
      "grad_norm": 0.2518390715122223,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 52975
    },
    {
      "epoch": 46.92648361381754,
      "grad_norm": 0.19199016690254211,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 52980
    },
    {
      "epoch": 46.93091231178034,
      "grad_norm": 0.1845492422580719,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 52985
    },
    {
      "epoch": 46.935341009743134,
      "grad_norm": 0.2563817799091339,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 52990
    },
    {
      "epoch": 46.939769707705935,
      "grad_norm": 0.235687255859375,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 52995
    },
    {
      "epoch": 46.944198405668736,
      "grad_norm": 0.25768518447875977,
      "learning_rate": 1e-05,
      "loss": 0.908,
      "step": 53000
    },
    {
      "epoch": 46.94862710363153,
      "grad_norm": 0.25883081555366516,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 53005
    },
    {
      "epoch": 46.95305580159433,
      "grad_norm": 0.23598439991474152,
      "learning_rate": 1e-05,
      "loss": 1.0461,
      "step": 53010
    },
    {
      "epoch": 46.95748449955713,
      "grad_norm": 0.21958620846271515,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 53015
    },
    {
      "epoch": 46.96191319751993,
      "grad_norm": 0.22603271901607513,
      "learning_rate": 1e-05,
      "loss": 0.8979,
      "step": 53020
    },
    {
      "epoch": 46.96634189548273,
      "grad_norm": 0.3098353445529938,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 53025
    },
    {
      "epoch": 46.97077059344553,
      "grad_norm": 0.25202155113220215,
      "learning_rate": 1e-05,
      "loss": 1.0113,
      "step": 53030
    },
    {
      "epoch": 46.97519929140832,
      "grad_norm": 0.24985453486442566,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 53035
    },
    {
      "epoch": 46.979627989371124,
      "grad_norm": 0.23234573006629944,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 53040
    },
    {
      "epoch": 46.984056687333926,
      "grad_norm": 0.22871297597885132,
      "learning_rate": 1e-05,
      "loss": 0.9235,
      "step": 53045
    },
    {
      "epoch": 46.98848538529672,
      "grad_norm": 0.25631675124168396,
      "learning_rate": 1e-05,
      "loss": 0.9206,
      "step": 53050
    },
    {
      "epoch": 46.99291408325952,
      "grad_norm": 0.25236403942108154,
      "learning_rate": 1e-05,
      "loss": 0.9237,
      "step": 53055
    },
    {
      "epoch": 46.99734278122232,
      "grad_norm": 0.26691293716430664,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 53060
    },
    {
      "epoch": 47.001771479185116,
      "grad_norm": 0.26703035831451416,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 53065
    },
    {
      "epoch": 47.00620017714792,
      "grad_norm": 0.24113349616527557,
      "learning_rate": 1e-05,
      "loss": 1.0245,
      "step": 53070
    },
    {
      "epoch": 47.01062887511072,
      "grad_norm": 0.22206377983093262,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 53075
    },
    {
      "epoch": 47.01505757307352,
      "grad_norm": 0.24047952890396118,
      "learning_rate": 1e-05,
      "loss": 0.9251,
      "step": 53080
    },
    {
      "epoch": 47.019486271036314,
      "grad_norm": 0.26321879029273987,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 53085
    },
    {
      "epoch": 47.023914968999115,
      "grad_norm": 0.2279481440782547,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 53090
    },
    {
      "epoch": 47.028343666961916,
      "grad_norm": 0.3014968931674957,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 53095
    },
    {
      "epoch": 47.03277236492471,
      "grad_norm": 0.23308083415031433,
      "learning_rate": 1e-05,
      "loss": 1.0541,
      "step": 53100
    },
    {
      "epoch": 47.03720106288751,
      "grad_norm": 0.2552223205566406,
      "learning_rate": 1e-05,
      "loss": 1.0253,
      "step": 53105
    },
    {
      "epoch": 47.04162976085031,
      "grad_norm": 0.2541239559650421,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 53110
    },
    {
      "epoch": 47.04605845881311,
      "grad_norm": 0.2731362283229828,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 53115
    },
    {
      "epoch": 47.05048715677591,
      "grad_norm": 0.2793257236480713,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 53120
    },
    {
      "epoch": 47.05491585473871,
      "grad_norm": 0.2322128713130951,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 53125
    },
    {
      "epoch": 47.0593445527015,
      "grad_norm": 0.2558023929595947,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 53130
    },
    {
      "epoch": 47.063773250664305,
      "grad_norm": 0.3224157989025116,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 53135
    },
    {
      "epoch": 47.068201948627106,
      "grad_norm": 0.22113892436027527,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 53140
    },
    {
      "epoch": 47.0726306465899,
      "grad_norm": 0.21309953927993774,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 53145
    },
    {
      "epoch": 47.0770593445527,
      "grad_norm": 0.24870753288269043,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 53150
    },
    {
      "epoch": 47.0814880425155,
      "grad_norm": 0.2497696727514267,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 53155
    },
    {
      "epoch": 47.0859167404783,
      "grad_norm": 0.2304968684911728,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 53160
    },
    {
      "epoch": 47.0903454384411,
      "grad_norm": 0.2342444509267807,
      "learning_rate": 1e-05,
      "loss": 0.9149,
      "step": 53165
    },
    {
      "epoch": 47.0947741364039,
      "grad_norm": 0.28979218006134033,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 53170
    },
    {
      "epoch": 47.09920283436669,
      "grad_norm": 0.23976381123065948,
      "learning_rate": 1e-05,
      "loss": 0.9199,
      "step": 53175
    },
    {
      "epoch": 47.103631532329494,
      "grad_norm": 0.2094898670911789,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 53180
    },
    {
      "epoch": 47.108060230292296,
      "grad_norm": 0.21998946368694305,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 53185
    },
    {
      "epoch": 47.11248892825509,
      "grad_norm": 0.22302959859371185,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 53190
    },
    {
      "epoch": 47.11691762621789,
      "grad_norm": 0.2214338630437851,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 53195
    },
    {
      "epoch": 47.12134632418069,
      "grad_norm": 0.23871473968029022,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 53200
    },
    {
      "epoch": 47.12577502214349,
      "grad_norm": 0.252048522233963,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 53205
    },
    {
      "epoch": 47.13020372010629,
      "grad_norm": 0.3042607605457306,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 53210
    },
    {
      "epoch": 47.13463241806909,
      "grad_norm": 0.2738129794597626,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 53215
    },
    {
      "epoch": 47.13906111603189,
      "grad_norm": 0.24043193459510803,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 53220
    },
    {
      "epoch": 47.143489813994684,
      "grad_norm": 0.22027045488357544,
      "learning_rate": 1e-05,
      "loss": 0.9043,
      "step": 53225
    },
    {
      "epoch": 47.147918511957485,
      "grad_norm": 0.26409274339675903,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 53230
    },
    {
      "epoch": 47.152347209920286,
      "grad_norm": 0.21213717758655548,
      "learning_rate": 1e-05,
      "loss": 1.0403,
      "step": 53235
    },
    {
      "epoch": 47.15677590788308,
      "grad_norm": 0.207491934299469,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 53240
    },
    {
      "epoch": 47.16120460584588,
      "grad_norm": 0.265442818403244,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 53245
    },
    {
      "epoch": 47.16563330380868,
      "grad_norm": 0.21938614547252655,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 53250
    },
    {
      "epoch": 47.17006200177148,
      "grad_norm": 0.25768953561782837,
      "learning_rate": 1e-05,
      "loss": 0.9311,
      "step": 53255
    },
    {
      "epoch": 47.17449069973428,
      "grad_norm": 0.22516220808029175,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 53260
    },
    {
      "epoch": 47.17891939769708,
      "grad_norm": 0.25387775897979736,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 53265
    },
    {
      "epoch": 47.18334809565987,
      "grad_norm": 0.26574185490608215,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 53270
    },
    {
      "epoch": 47.187776793622675,
      "grad_norm": 0.2780439853668213,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 53275
    },
    {
      "epoch": 47.192205491585476,
      "grad_norm": 0.2373197376728058,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 53280
    },
    {
      "epoch": 47.19663418954827,
      "grad_norm": 0.22846949100494385,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 53285
    },
    {
      "epoch": 47.20106288751107,
      "grad_norm": 0.20439369976520538,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 53290
    },
    {
      "epoch": 47.20549158547387,
      "grad_norm": 0.22882992029190063,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 53295
    },
    {
      "epoch": 47.20992028343667,
      "grad_norm": 0.21933390200138092,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 53300
    },
    {
      "epoch": 47.21434898139947,
      "grad_norm": 0.23676227033138275,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 53305
    },
    {
      "epoch": 47.21877767936227,
      "grad_norm": 0.2338477224111557,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 53310
    },
    {
      "epoch": 47.22320637732506,
      "grad_norm": 0.21973244845867157,
      "learning_rate": 1e-05,
      "loss": 0.9244,
      "step": 53315
    },
    {
      "epoch": 47.227635075287864,
      "grad_norm": 0.2668595314025879,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 53320
    },
    {
      "epoch": 47.232063773250665,
      "grad_norm": 0.2607423663139343,
      "learning_rate": 1e-05,
      "loss": 1.0101,
      "step": 53325
    },
    {
      "epoch": 47.23649247121347,
      "grad_norm": 0.22921982407569885,
      "learning_rate": 1e-05,
      "loss": 1.0264,
      "step": 53330
    },
    {
      "epoch": 47.24092116917626,
      "grad_norm": 0.2475089430809021,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 53335
    },
    {
      "epoch": 47.24534986713906,
      "grad_norm": 0.24275843799114227,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 53340
    },
    {
      "epoch": 47.24977856510186,
      "grad_norm": 0.24358443915843964,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 53345
    },
    {
      "epoch": 47.25420726306466,
      "grad_norm": 0.20708854496479034,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 53350
    },
    {
      "epoch": 47.25863596102746,
      "grad_norm": 0.2137162834405899,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 53355
    },
    {
      "epoch": 47.26306465899026,
      "grad_norm": 0.2265847772359848,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 53360
    },
    {
      "epoch": 47.267493356953054,
      "grad_norm": 0.20596414804458618,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 53365
    },
    {
      "epoch": 47.271922054915855,
      "grad_norm": 0.23469996452331543,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 53370
    },
    {
      "epoch": 47.276350752878656,
      "grad_norm": 0.2332441210746765,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 53375
    },
    {
      "epoch": 47.28077945084145,
      "grad_norm": 0.2466612458229065,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 53380
    },
    {
      "epoch": 47.28520814880425,
      "grad_norm": 0.24347557127475739,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 53385
    },
    {
      "epoch": 47.28963684676705,
      "grad_norm": 0.24266250431537628,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 53390
    },
    {
      "epoch": 47.29406554472985,
      "grad_norm": 0.2543930411338806,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 53395
    },
    {
      "epoch": 47.29849424269265,
      "grad_norm": 0.24881596863269806,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 53400
    },
    {
      "epoch": 47.30292294065545,
      "grad_norm": 0.264466792345047,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 53405
    },
    {
      "epoch": 47.30735163861824,
      "grad_norm": 0.2405591905117035,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 53410
    },
    {
      "epoch": 47.311780336581045,
      "grad_norm": 0.2447965145111084,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 53415
    },
    {
      "epoch": 47.316209034543846,
      "grad_norm": 0.21730738878250122,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 53420
    },
    {
      "epoch": 47.32063773250664,
      "grad_norm": 0.22900474071502686,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 53425
    },
    {
      "epoch": 47.32506643046944,
      "grad_norm": 0.18706859648227692,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 53430
    },
    {
      "epoch": 47.32949512843224,
      "grad_norm": 0.23171640932559967,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 53435
    },
    {
      "epoch": 47.333923826395036,
      "grad_norm": 0.21889935433864594,
      "learning_rate": 1e-05,
      "loss": 0.9093,
      "step": 53440
    },
    {
      "epoch": 47.33835252435784,
      "grad_norm": 0.199715718626976,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 53445
    },
    {
      "epoch": 47.34278122232064,
      "grad_norm": 0.23362258076667786,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 53450
    },
    {
      "epoch": 47.34720992028344,
      "grad_norm": 0.25792279839515686,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 53455
    },
    {
      "epoch": 47.351638618246234,
      "grad_norm": 0.261173814535141,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 53460
    },
    {
      "epoch": 47.356067316209035,
      "grad_norm": 0.21843306720256805,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 53465
    },
    {
      "epoch": 47.36049601417184,
      "grad_norm": 0.2818938195705414,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 53470
    },
    {
      "epoch": 47.36492471213463,
      "grad_norm": 0.27318310737609863,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 53475
    },
    {
      "epoch": 47.36935341009743,
      "grad_norm": 0.24385161697864532,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 53480
    },
    {
      "epoch": 47.37378210806023,
      "grad_norm": 0.22357363998889923,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 53485
    },
    {
      "epoch": 47.37821080602303,
      "grad_norm": 0.23831617832183838,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 53490
    },
    {
      "epoch": 47.38263950398583,
      "grad_norm": 0.2219279557466507,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 53495
    },
    {
      "epoch": 47.38706820194863,
      "grad_norm": 0.2879375219345093,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 53500
    },
    {
      "epoch": 47.391496899911424,
      "grad_norm": 0.23290784657001495,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 53505
    },
    {
      "epoch": 47.395925597874225,
      "grad_norm": 0.2344203144311905,
      "learning_rate": 1e-05,
      "loss": 1.0241,
      "step": 53510
    },
    {
      "epoch": 47.400354295837026,
      "grad_norm": 0.2480880469083786,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 53515
    },
    {
      "epoch": 47.40478299379982,
      "grad_norm": 0.2302074283361435,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 53520
    },
    {
      "epoch": 47.40921169176262,
      "grad_norm": 0.2596975862979889,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 53525
    },
    {
      "epoch": 47.41364038972542,
      "grad_norm": 0.20955277979373932,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 53530
    },
    {
      "epoch": 47.41806908768822,
      "grad_norm": 0.23876962065696716,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 53535
    },
    {
      "epoch": 47.42249778565102,
      "grad_norm": 0.23004160821437836,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 53540
    },
    {
      "epoch": 47.42692648361382,
      "grad_norm": 0.2267003059387207,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 53545
    },
    {
      "epoch": 47.43135518157661,
      "grad_norm": 0.2454613447189331,
      "learning_rate": 1e-05,
      "loss": 0.8968,
      "step": 53550
    },
    {
      "epoch": 47.435783879539414,
      "grad_norm": 0.2609652876853943,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 53555
    },
    {
      "epoch": 47.440212577502216,
      "grad_norm": 0.22250857949256897,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 53560
    },
    {
      "epoch": 47.44464127546502,
      "grad_norm": 0.24546082317829132,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 53565
    },
    {
      "epoch": 47.44906997342781,
      "grad_norm": 0.25648465752601624,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 53570
    },
    {
      "epoch": 47.45349867139061,
      "grad_norm": 0.22747467458248138,
      "learning_rate": 1e-05,
      "loss": 0.8994,
      "step": 53575
    },
    {
      "epoch": 47.45792736935341,
      "grad_norm": 0.2028781920671463,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 53580
    },
    {
      "epoch": 47.46235606731621,
      "grad_norm": 0.21915559470653534,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 53585
    },
    {
      "epoch": 47.46678476527901,
      "grad_norm": 0.28994935750961304,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 53590
    },
    {
      "epoch": 47.47121346324181,
      "grad_norm": 0.2160787731409073,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 53595
    },
    {
      "epoch": 47.475642161204604,
      "grad_norm": 0.2394951432943344,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 53600
    },
    {
      "epoch": 47.480070859167405,
      "grad_norm": 0.22000344097614288,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 53605
    },
    {
      "epoch": 47.484499557130206,
      "grad_norm": 0.21622705459594727,
      "learning_rate": 1e-05,
      "loss": 0.941,
      "step": 53610
    },
    {
      "epoch": 47.488928255093,
      "grad_norm": 0.2536621391773224,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 53615
    },
    {
      "epoch": 47.4933569530558,
      "grad_norm": 0.26189449429512024,
      "learning_rate": 1e-05,
      "loss": 1.0113,
      "step": 53620
    },
    {
      "epoch": 47.4977856510186,
      "grad_norm": 0.25883767008781433,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 53625
    },
    {
      "epoch": 47.5022143489814,
      "grad_norm": 0.2206197828054428,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 53630
    },
    {
      "epoch": 47.5066430469442,
      "grad_norm": 0.28032901883125305,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 53635
    },
    {
      "epoch": 47.511071744907,
      "grad_norm": 0.2708328664302826,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 53640
    },
    {
      "epoch": 47.515500442869794,
      "grad_norm": 0.24346905946731567,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 53645
    },
    {
      "epoch": 47.519929140832595,
      "grad_norm": 0.23383086919784546,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 53650
    },
    {
      "epoch": 47.524357838795396,
      "grad_norm": 0.2327410876750946,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 53655
    },
    {
      "epoch": 47.52878653675819,
      "grad_norm": 0.24351760745048523,
      "learning_rate": 1e-05,
      "loss": 0.9258,
      "step": 53660
    },
    {
      "epoch": 47.53321523472099,
      "grad_norm": 0.2381892055273056,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 53665
    },
    {
      "epoch": 47.53764393268379,
      "grad_norm": 0.2196206897497177,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 53670
    },
    {
      "epoch": 47.54207263064659,
      "grad_norm": 0.2274877429008484,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 53675
    },
    {
      "epoch": 47.54650132860939,
      "grad_norm": 0.24286487698554993,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 53680
    },
    {
      "epoch": 47.55093002657219,
      "grad_norm": 0.2319401651620865,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 53685
    },
    {
      "epoch": 47.55535872453498,
      "grad_norm": 0.24155199527740479,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 53690
    },
    {
      "epoch": 47.559787422497784,
      "grad_norm": 0.20747703313827515,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 53695
    },
    {
      "epoch": 47.564216120460586,
      "grad_norm": 0.24954165518283844,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 53700
    },
    {
      "epoch": 47.56864481842339,
      "grad_norm": 0.2240035980939865,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 53705
    },
    {
      "epoch": 47.57307351638618,
      "grad_norm": 0.24623899161815643,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 53710
    },
    {
      "epoch": 47.57750221434898,
      "grad_norm": 0.2567157447338104,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 53715
    },
    {
      "epoch": 47.58193091231178,
      "grad_norm": 0.2522636353969574,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 53720
    },
    {
      "epoch": 47.58635961027458,
      "grad_norm": 0.27009207010269165,
      "learning_rate": 1e-05,
      "loss": 0.9299,
      "step": 53725
    },
    {
      "epoch": 47.59078830823738,
      "grad_norm": 0.2906951904296875,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 53730
    },
    {
      "epoch": 47.59521700620018,
      "grad_norm": 0.22079971432685852,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 53735
    },
    {
      "epoch": 47.599645704162974,
      "grad_norm": 0.21642599999904633,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 53740
    },
    {
      "epoch": 47.604074402125775,
      "grad_norm": 0.27416425943374634,
      "learning_rate": 1e-05,
      "loss": 1.051,
      "step": 53745
    },
    {
      "epoch": 47.608503100088576,
      "grad_norm": 0.24693958461284637,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 53750
    },
    {
      "epoch": 47.61293179805137,
      "grad_norm": 0.2254188358783722,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 53755
    },
    {
      "epoch": 47.61736049601417,
      "grad_norm": 0.24221350252628326,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 53760
    },
    {
      "epoch": 47.62178919397697,
      "grad_norm": 0.23769111931324005,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 53765
    },
    {
      "epoch": 47.62621789193977,
      "grad_norm": 0.20381732285022736,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 53770
    },
    {
      "epoch": 47.63064658990257,
      "grad_norm": 0.23721915483474731,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 53775
    },
    {
      "epoch": 47.63507528786537,
      "grad_norm": 0.23032695055007935,
      "learning_rate": 1e-05,
      "loss": 0.9208,
      "step": 53780
    },
    {
      "epoch": 47.63950398582816,
      "grad_norm": 0.19805924594402313,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 53785
    },
    {
      "epoch": 47.643932683790965,
      "grad_norm": 0.29126277565956116,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 53790
    },
    {
      "epoch": 47.648361381753766,
      "grad_norm": 0.23259305953979492,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 53795
    },
    {
      "epoch": 47.65279007971656,
      "grad_norm": 0.2581517994403839,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 53800
    },
    {
      "epoch": 47.65721877767936,
      "grad_norm": 0.25572654604911804,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 53805
    },
    {
      "epoch": 47.66164747564216,
      "grad_norm": 0.21386483311653137,
      "learning_rate": 1e-05,
      "loss": 0.9297,
      "step": 53810
    },
    {
      "epoch": 47.666076173604964,
      "grad_norm": 0.25590816140174866,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 53815
    },
    {
      "epoch": 47.67050487156776,
      "grad_norm": 0.31813332438468933,
      "learning_rate": 1e-05,
      "loss": 0.9261,
      "step": 53820
    },
    {
      "epoch": 47.67493356953056,
      "grad_norm": 0.24896399676799774,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 53825
    },
    {
      "epoch": 47.67936226749336,
      "grad_norm": 0.22610758244991302,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 53830
    },
    {
      "epoch": 47.683790965456154,
      "grad_norm": 0.24754638969898224,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 53835
    },
    {
      "epoch": 47.688219663418955,
      "grad_norm": 0.2399856001138687,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 53840
    },
    {
      "epoch": 47.69264836138176,
      "grad_norm": 0.21230283379554749,
      "learning_rate": 1e-05,
      "loss": 1.0101,
      "step": 53845
    },
    {
      "epoch": 47.69707705934455,
      "grad_norm": 0.2711069583892822,
      "learning_rate": 1e-05,
      "loss": 0.9166,
      "step": 53850
    },
    {
      "epoch": 47.70150575730735,
      "grad_norm": 0.2500757873058319,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 53855
    },
    {
      "epoch": 47.70593445527015,
      "grad_norm": 0.21669919788837433,
      "learning_rate": 1e-05,
      "loss": 1.0311,
      "step": 53860
    },
    {
      "epoch": 47.71036315323295,
      "grad_norm": 0.207510307431221,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 53865
    },
    {
      "epoch": 47.71479185119575,
      "grad_norm": 0.2725810408592224,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 53870
    },
    {
      "epoch": 47.71922054915855,
      "grad_norm": 0.2077665477991104,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 53875
    },
    {
      "epoch": 47.723649247121344,
      "grad_norm": 0.2801382541656494,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 53880
    },
    {
      "epoch": 47.728077945084145,
      "grad_norm": 0.2574128806591034,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 53885
    },
    {
      "epoch": 47.732506643046946,
      "grad_norm": 0.28305867314338684,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 53890
    },
    {
      "epoch": 47.73693534100974,
      "grad_norm": 0.27080875635147095,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 53895
    },
    {
      "epoch": 47.74136403897254,
      "grad_norm": 0.24162890017032623,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 53900
    },
    {
      "epoch": 47.74579273693534,
      "grad_norm": 0.2586243152618408,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 53905
    },
    {
      "epoch": 47.75022143489814,
      "grad_norm": 0.26357874274253845,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 53910
    },
    {
      "epoch": 47.75465013286094,
      "grad_norm": 0.2479870319366455,
      "learning_rate": 1e-05,
      "loss": 1.0255,
      "step": 53915
    },
    {
      "epoch": 47.75907883082374,
      "grad_norm": 0.28403475880622864,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 53920
    },
    {
      "epoch": 47.76350752878653,
      "grad_norm": 0.2357282191514969,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 53925
    },
    {
      "epoch": 47.767936226749335,
      "grad_norm": 0.27125847339630127,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 53930
    },
    {
      "epoch": 47.772364924712136,
      "grad_norm": 0.22664161026477814,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 53935
    },
    {
      "epoch": 47.77679362267494,
      "grad_norm": 0.22646038234233856,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 53940
    },
    {
      "epoch": 47.78122232063773,
      "grad_norm": 0.22992365062236786,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 53945
    },
    {
      "epoch": 47.78565101860053,
      "grad_norm": 0.22834976017475128,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 53950
    },
    {
      "epoch": 47.79007971656333,
      "grad_norm": 0.24656164646148682,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 53955
    },
    {
      "epoch": 47.79450841452613,
      "grad_norm": 0.25544074177742004,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 53960
    },
    {
      "epoch": 47.79893711248893,
      "grad_norm": 0.2300511598587036,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 53965
    },
    {
      "epoch": 47.80336581045173,
      "grad_norm": 0.21994829177856445,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 53970
    },
    {
      "epoch": 47.807794508414524,
      "grad_norm": 0.21942482888698578,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 53975
    },
    {
      "epoch": 47.812223206377325,
      "grad_norm": 0.24433690309524536,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 53980
    },
    {
      "epoch": 47.81665190434013,
      "grad_norm": 0.21427181363105774,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 53985
    },
    {
      "epoch": 47.82108060230292,
      "grad_norm": 0.24657323956489563,
      "learning_rate": 1e-05,
      "loss": 1.0113,
      "step": 53990
    },
    {
      "epoch": 47.82550930026572,
      "grad_norm": 0.24297063052654266,
      "learning_rate": 1e-05,
      "loss": 0.9196,
      "step": 53995
    },
    {
      "epoch": 47.82993799822852,
      "grad_norm": 0.25518327951431274,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 54000
    },
    {
      "epoch": 47.83436669619132,
      "grad_norm": 0.2253987193107605,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 54005
    },
    {
      "epoch": 47.83879539415412,
      "grad_norm": 0.2907208204269409,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 54010
    },
    {
      "epoch": 47.84322409211692,
      "grad_norm": 0.2752867639064789,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 54015
    },
    {
      "epoch": 47.847652790079714,
      "grad_norm": 0.24904656410217285,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 54020
    },
    {
      "epoch": 47.852081488042515,
      "grad_norm": 0.24027155339717865,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 54025
    },
    {
      "epoch": 47.856510186005316,
      "grad_norm": 0.2629442512989044,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 54030
    },
    {
      "epoch": 47.86093888396811,
      "grad_norm": 0.2401168793439865,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 54035
    },
    {
      "epoch": 47.86536758193091,
      "grad_norm": 0.3038504123687744,
      "learning_rate": 1e-05,
      "loss": 0.9161,
      "step": 54040
    },
    {
      "epoch": 47.86979627989371,
      "grad_norm": 0.2369646579027176,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 54045
    },
    {
      "epoch": 47.87422497785651,
      "grad_norm": 0.2360883504152298,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 54050
    },
    {
      "epoch": 47.87865367581931,
      "grad_norm": 0.23813392221927643,
      "learning_rate": 1e-05,
      "loss": 1.058,
      "step": 54055
    },
    {
      "epoch": 47.88308237378211,
      "grad_norm": 0.24803613126277924,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 54060
    },
    {
      "epoch": 47.88751107174491,
      "grad_norm": 0.28278231620788574,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 54065
    },
    {
      "epoch": 47.891939769707704,
      "grad_norm": 0.2390805184841156,
      "learning_rate": 1e-05,
      "loss": 1.0258,
      "step": 54070
    },
    {
      "epoch": 47.896368467670506,
      "grad_norm": 0.2573857009410858,
      "learning_rate": 1e-05,
      "loss": 0.9186,
      "step": 54075
    },
    {
      "epoch": 47.90079716563331,
      "grad_norm": 0.2704625725746155,
      "learning_rate": 1e-05,
      "loss": 0.9066,
      "step": 54080
    },
    {
      "epoch": 47.9052258635961,
      "grad_norm": 0.2473248392343521,
      "learning_rate": 1e-05,
      "loss": 1.03,
      "step": 54085
    },
    {
      "epoch": 47.9096545615589,
      "grad_norm": 0.2661992907524109,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 54090
    },
    {
      "epoch": 47.9140832595217,
      "grad_norm": 0.2230280339717865,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 54095
    },
    {
      "epoch": 47.9185119574845,
      "grad_norm": 0.24677376449108124,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 54100
    },
    {
      "epoch": 47.9229406554473,
      "grad_norm": 0.25990813970565796,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 54105
    },
    {
      "epoch": 47.9273693534101,
      "grad_norm": 0.24985814094543457,
      "learning_rate": 1e-05,
      "loss": 0.942,
      "step": 54110
    },
    {
      "epoch": 47.931798051372894,
      "grad_norm": 0.24179169535636902,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 54115
    },
    {
      "epoch": 47.936226749335695,
      "grad_norm": 0.2248622179031372,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 54120
    },
    {
      "epoch": 47.9406554472985,
      "grad_norm": 0.2241470366716385,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 54125
    },
    {
      "epoch": 47.94508414526129,
      "grad_norm": 0.23480424284934998,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 54130
    },
    {
      "epoch": 47.94951284322409,
      "grad_norm": 0.2681584060192108,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 54135
    },
    {
      "epoch": 47.95394154118689,
      "grad_norm": 0.24585366249084473,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 54140
    },
    {
      "epoch": 47.95837023914969,
      "grad_norm": 0.2310543656349182,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 54145
    },
    {
      "epoch": 47.96279893711249,
      "grad_norm": 0.24443264305591583,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 54150
    },
    {
      "epoch": 47.96722763507529,
      "grad_norm": 0.20922762155532837,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 54155
    },
    {
      "epoch": 47.971656333038084,
      "grad_norm": 0.25041335821151733,
      "learning_rate": 1e-05,
      "loss": 0.9999,
      "step": 54160
    },
    {
      "epoch": 47.976085031000885,
      "grad_norm": 0.22820109128952026,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 54165
    },
    {
      "epoch": 47.980513728963686,
      "grad_norm": 0.21311110258102417,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 54170
    },
    {
      "epoch": 47.98494242692648,
      "grad_norm": 0.2056189477443695,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 54175
    },
    {
      "epoch": 47.98937112488928,
      "grad_norm": 0.21524974703788757,
      "learning_rate": 1e-05,
      "loss": 1.0085,
      "step": 54180
    },
    {
      "epoch": 47.99379982285208,
      "grad_norm": 0.2759053111076355,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 54185
    },
    {
      "epoch": 47.998228520814884,
      "grad_norm": 0.26053863763809204,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 54190
    },
    {
      "epoch": 48.00265721877768,
      "grad_norm": 0.27095502614974976,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 54195
    },
    {
      "epoch": 48.00708591674048,
      "grad_norm": 0.21030853688716888,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 54200
    },
    {
      "epoch": 48.01151461470328,
      "grad_norm": 0.21274219453334808,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 54205
    },
    {
      "epoch": 48.015943312666074,
      "grad_norm": 0.22843638062477112,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 54210
    },
    {
      "epoch": 48.020372010628876,
      "grad_norm": 0.21671180427074432,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 54215
    },
    {
      "epoch": 48.02480070859168,
      "grad_norm": 0.20990025997161865,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 54220
    },
    {
      "epoch": 48.02922940655447,
      "grad_norm": 0.24164745211601257,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 54225
    },
    {
      "epoch": 48.03365810451727,
      "grad_norm": 0.22842633724212646,
      "learning_rate": 1e-05,
      "loss": 1.0373,
      "step": 54230
    },
    {
      "epoch": 48.03808680248007,
      "grad_norm": 0.2710988521575928,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 54235
    },
    {
      "epoch": 48.04251550044287,
      "grad_norm": 0.23926052451133728,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 54240
    },
    {
      "epoch": 48.04694419840567,
      "grad_norm": 0.25839963555336,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 54245
    },
    {
      "epoch": 48.05137289636847,
      "grad_norm": 0.23707906901836395,
      "learning_rate": 1e-05,
      "loss": 1.0424,
      "step": 54250
    },
    {
      "epoch": 48.055801594331264,
      "grad_norm": 0.26977527141571045,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 54255
    },
    {
      "epoch": 48.060230292294065,
      "grad_norm": 0.2278953343629837,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 54260
    },
    {
      "epoch": 48.064658990256866,
      "grad_norm": 0.24550196528434753,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 54265
    },
    {
      "epoch": 48.06908768821966,
      "grad_norm": 0.23541414737701416,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 54270
    },
    {
      "epoch": 48.07351638618246,
      "grad_norm": 0.2648276388645172,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 54275
    },
    {
      "epoch": 48.07794508414526,
      "grad_norm": 0.22903665900230408,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 54280
    },
    {
      "epoch": 48.08237378210806,
      "grad_norm": 0.26113361120224,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 54285
    },
    {
      "epoch": 48.08680248007086,
      "grad_norm": 0.24845163524150848,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 54290
    },
    {
      "epoch": 48.09123117803366,
      "grad_norm": 0.2625034749507904,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 54295
    },
    {
      "epoch": 48.09565987599645,
      "grad_norm": 0.2668391466140747,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 54300
    },
    {
      "epoch": 48.100088573959255,
      "grad_norm": 0.2723771333694458,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 54305
    },
    {
      "epoch": 48.104517271922056,
      "grad_norm": 0.2708938717842102,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 54310
    },
    {
      "epoch": 48.10894596988486,
      "grad_norm": 0.2522117495536804,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 54315
    },
    {
      "epoch": 48.11337466784765,
      "grad_norm": 0.2573252320289612,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 54320
    },
    {
      "epoch": 48.11780336581045,
      "grad_norm": 0.2446250021457672,
      "learning_rate": 1e-05,
      "loss": 1.014,
      "step": 54325
    },
    {
      "epoch": 48.122232063773254,
      "grad_norm": 0.23871248960494995,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 54330
    },
    {
      "epoch": 48.12666076173605,
      "grad_norm": 0.23324066400527954,
      "learning_rate": 1e-05,
      "loss": 0.9253,
      "step": 54335
    },
    {
      "epoch": 48.13108945969885,
      "grad_norm": 0.220278799533844,
      "learning_rate": 1e-05,
      "loss": 0.9251,
      "step": 54340
    },
    {
      "epoch": 48.13551815766165,
      "grad_norm": 0.2946358025074005,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 54345
    },
    {
      "epoch": 48.139946855624444,
      "grad_norm": 0.22677218914031982,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 54350
    },
    {
      "epoch": 48.144375553587246,
      "grad_norm": 0.24953903257846832,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 54355
    },
    {
      "epoch": 48.14880425155005,
      "grad_norm": 0.2694138288497925,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 54360
    },
    {
      "epoch": 48.15323294951284,
      "grad_norm": 0.26679837703704834,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 54365
    },
    {
      "epoch": 48.15766164747564,
      "grad_norm": 0.2267667055130005,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 54370
    },
    {
      "epoch": 48.16209034543844,
      "grad_norm": 0.2701794505119324,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 54375
    },
    {
      "epoch": 48.16651904340124,
      "grad_norm": 0.23114758729934692,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 54380
    },
    {
      "epoch": 48.17094774136404,
      "grad_norm": 0.2564833164215088,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 54385
    },
    {
      "epoch": 48.17537643932684,
      "grad_norm": 0.2633117437362671,
      "learning_rate": 1e-05,
      "loss": 1.0632,
      "step": 54390
    },
    {
      "epoch": 48.179805137289634,
      "grad_norm": 0.24890990555286407,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 54395
    },
    {
      "epoch": 48.184233835252435,
      "grad_norm": 0.2447938770055771,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 54400
    },
    {
      "epoch": 48.188662533215236,
      "grad_norm": 0.2054462432861328,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 54405
    },
    {
      "epoch": 48.19309123117803,
      "grad_norm": 0.23189909756183624,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 54410
    },
    {
      "epoch": 48.19751992914083,
      "grad_norm": 0.22702151536941528,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 54415
    },
    {
      "epoch": 48.20194862710363,
      "grad_norm": 0.22041070461273193,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 54420
    },
    {
      "epoch": 48.20637732506643,
      "grad_norm": 0.22508110105991364,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 54425
    },
    {
      "epoch": 48.21080602302923,
      "grad_norm": 0.24688751995563507,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 54430
    },
    {
      "epoch": 48.21523472099203,
      "grad_norm": 0.2732565999031067,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 54435
    },
    {
      "epoch": 48.21966341895483,
      "grad_norm": 0.20759116113185883,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 54440
    },
    {
      "epoch": 48.224092116917625,
      "grad_norm": 0.26913273334503174,
      "learning_rate": 1e-05,
      "loss": 0.8892,
      "step": 54445
    },
    {
      "epoch": 48.228520814880426,
      "grad_norm": 0.2323606163263321,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 54450
    },
    {
      "epoch": 48.23294951284323,
      "grad_norm": 0.2318078726530075,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 54455
    },
    {
      "epoch": 48.23737821080602,
      "grad_norm": 0.2297491878271103,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 54460
    },
    {
      "epoch": 48.24180690876882,
      "grad_norm": 0.24085623025894165,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 54465
    },
    {
      "epoch": 48.246235606731624,
      "grad_norm": 0.23873399198055267,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 54470
    },
    {
      "epoch": 48.25066430469442,
      "grad_norm": 0.2656705975532532,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 54475
    },
    {
      "epoch": 48.25509300265722,
      "grad_norm": 0.24265353381633759,
      "learning_rate": 1e-05,
      "loss": 0.9222,
      "step": 54480
    },
    {
      "epoch": 48.25952170062002,
      "grad_norm": 0.2673168182373047,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 54485
    },
    {
      "epoch": 48.263950398582814,
      "grad_norm": 0.28807422518730164,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 54490
    },
    {
      "epoch": 48.268379096545615,
      "grad_norm": 0.24280890822410583,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 54495
    },
    {
      "epoch": 48.27280779450842,
      "grad_norm": 0.2847305238246918,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 54500
    },
    {
      "epoch": 48.27723649247121,
      "grad_norm": 0.23908749222755432,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 54505
    },
    {
      "epoch": 48.28166519043401,
      "grad_norm": 0.20606811344623566,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 54510
    },
    {
      "epoch": 48.28609388839681,
      "grad_norm": 0.30251064896583557,
      "learning_rate": 1e-05,
      "loss": 1.0054,
      "step": 54515
    },
    {
      "epoch": 48.29052258635961,
      "grad_norm": 0.27221035957336426,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 54520
    },
    {
      "epoch": 48.29495128432241,
      "grad_norm": 0.22023601830005646,
      "learning_rate": 1e-05,
      "loss": 1.0239,
      "step": 54525
    },
    {
      "epoch": 48.29937998228521,
      "grad_norm": 0.2757318317890167,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 54530
    },
    {
      "epoch": 48.303808680248004,
      "grad_norm": 0.2227504700422287,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 54535
    },
    {
      "epoch": 48.308237378210805,
      "grad_norm": 0.19213366508483887,
      "learning_rate": 1e-05,
      "loss": 0.9191,
      "step": 54540
    },
    {
      "epoch": 48.312666076173606,
      "grad_norm": 0.21350793540477753,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 54545
    },
    {
      "epoch": 48.31709477413641,
      "grad_norm": 0.26234567165374756,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 54550
    },
    {
      "epoch": 48.3215234720992,
      "grad_norm": 0.2475329041481018,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 54555
    },
    {
      "epoch": 48.325952170062,
      "grad_norm": 0.2819952070713043,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 54560
    },
    {
      "epoch": 48.330380868024804,
      "grad_norm": 0.23078052699565887,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 54565
    },
    {
      "epoch": 48.3348095659876,
      "grad_norm": 0.2415207177400589,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 54570
    },
    {
      "epoch": 48.3392382639504,
      "grad_norm": 0.21792201697826385,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 54575
    },
    {
      "epoch": 48.3436669619132,
      "grad_norm": 0.21483545005321503,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 54580
    },
    {
      "epoch": 48.348095659875995,
      "grad_norm": 0.24992138147354126,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 54585
    },
    {
      "epoch": 48.352524357838796,
      "grad_norm": 0.21953345835208893,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 54590
    },
    {
      "epoch": 48.3569530558016,
      "grad_norm": 0.2584550678730011,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 54595
    },
    {
      "epoch": 48.36138175376439,
      "grad_norm": 0.24125786125659943,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 54600
    },
    {
      "epoch": 48.36581045172719,
      "grad_norm": 0.2452802062034607,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 54605
    },
    {
      "epoch": 48.37023914968999,
      "grad_norm": 0.22685956954956055,
      "learning_rate": 1e-05,
      "loss": 0.9205,
      "step": 54610
    },
    {
      "epoch": 48.37466784765279,
      "grad_norm": 0.22611494362354279,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 54615
    },
    {
      "epoch": 48.37909654561559,
      "grad_norm": 0.2419653683900833,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 54620
    },
    {
      "epoch": 48.38352524357839,
      "grad_norm": 0.22503463923931122,
      "learning_rate": 1e-05,
      "loss": 0.9233,
      "step": 54625
    },
    {
      "epoch": 48.387953941541184,
      "grad_norm": 0.22026920318603516,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 54630
    },
    {
      "epoch": 48.392382639503985,
      "grad_norm": 0.22886820137500763,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 54635
    },
    {
      "epoch": 48.39681133746679,
      "grad_norm": 0.26335409283638,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 54640
    },
    {
      "epoch": 48.40124003542958,
      "grad_norm": 0.24266499280929565,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 54645
    },
    {
      "epoch": 48.40566873339238,
      "grad_norm": 0.22106784582138062,
      "learning_rate": 1e-05,
      "loss": 0.9298,
      "step": 54650
    },
    {
      "epoch": 48.41009743135518,
      "grad_norm": 0.23873266577720642,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 54655
    },
    {
      "epoch": 48.41452612931798,
      "grad_norm": 0.263793408870697,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 54660
    },
    {
      "epoch": 48.41895482728078,
      "grad_norm": 0.20200185477733612,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 54665
    },
    {
      "epoch": 48.42338352524358,
      "grad_norm": 0.21077078580856323,
      "learning_rate": 1e-05,
      "loss": 0.9099,
      "step": 54670
    },
    {
      "epoch": 48.42781222320638,
      "grad_norm": 0.22018828988075256,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 54675
    },
    {
      "epoch": 48.432240921169175,
      "grad_norm": 0.2709357440471649,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 54680
    },
    {
      "epoch": 48.436669619131976,
      "grad_norm": 0.24960030615329742,
      "learning_rate": 1e-05,
      "loss": 1.0213,
      "step": 54685
    },
    {
      "epoch": 48.44109831709478,
      "grad_norm": 0.30026060342788696,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 54690
    },
    {
      "epoch": 48.44552701505757,
      "grad_norm": 0.2196469008922577,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 54695
    },
    {
      "epoch": 48.44995571302037,
      "grad_norm": 0.2054162323474884,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 54700
    },
    {
      "epoch": 48.454384410983174,
      "grad_norm": 0.264117032289505,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 54705
    },
    {
      "epoch": 48.45881310894597,
      "grad_norm": 0.2796826958656311,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 54710
    },
    {
      "epoch": 48.46324180690877,
      "grad_norm": 0.26269227266311646,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 54715
    },
    {
      "epoch": 48.46767050487157,
      "grad_norm": 0.2235548496246338,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 54720
    },
    {
      "epoch": 48.472099202834364,
      "grad_norm": 0.23668207228183746,
      "learning_rate": 1e-05,
      "loss": 1.0468,
      "step": 54725
    },
    {
      "epoch": 48.476527900797166,
      "grad_norm": 0.24984893202781677,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 54730
    },
    {
      "epoch": 48.48095659875997,
      "grad_norm": 0.2638162672519684,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 54735
    },
    {
      "epoch": 48.48538529672276,
      "grad_norm": 0.22550000250339508,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 54740
    },
    {
      "epoch": 48.48981399468556,
      "grad_norm": 0.2607899308204651,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 54745
    },
    {
      "epoch": 48.49424269264836,
      "grad_norm": 0.25486019253730774,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 54750
    },
    {
      "epoch": 48.49867139061116,
      "grad_norm": 0.22319072484970093,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 54755
    },
    {
      "epoch": 48.50310008857396,
      "grad_norm": 0.25386983156204224,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 54760
    },
    {
      "epoch": 48.50752878653676,
      "grad_norm": 0.20265452563762665,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 54765
    },
    {
      "epoch": 48.511957484499554,
      "grad_norm": 0.2146308869123459,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 54770
    },
    {
      "epoch": 48.516386182462355,
      "grad_norm": 0.23876622319221497,
      "learning_rate": 1e-05,
      "loss": 1.0354,
      "step": 54775
    },
    {
      "epoch": 48.520814880425156,
      "grad_norm": 0.2343628853559494,
      "learning_rate": 1e-05,
      "loss": 1.0444,
      "step": 54780
    },
    {
      "epoch": 48.52524357838795,
      "grad_norm": 0.25492337346076965,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 54785
    },
    {
      "epoch": 48.52967227635075,
      "grad_norm": 0.221262127161026,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 54790
    },
    {
      "epoch": 48.53410097431355,
      "grad_norm": 0.25594913959503174,
      "learning_rate": 1e-05,
      "loss": 1.0174,
      "step": 54795
    },
    {
      "epoch": 48.538529672276354,
      "grad_norm": 0.24185062944889069,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 54800
    },
    {
      "epoch": 48.54295837023915,
      "grad_norm": 0.23660625517368317,
      "learning_rate": 1e-05,
      "loss": 1.0054,
      "step": 54805
    },
    {
      "epoch": 48.54738706820195,
      "grad_norm": 0.2552359402179718,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 54810
    },
    {
      "epoch": 48.55181576616475,
      "grad_norm": 0.21541759371757507,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 54815
    },
    {
      "epoch": 48.556244464127545,
      "grad_norm": 0.26602432131767273,
      "learning_rate": 1e-05,
      "loss": 0.9204,
      "step": 54820
    },
    {
      "epoch": 48.560673162090346,
      "grad_norm": 0.23079116642475128,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 54825
    },
    {
      "epoch": 48.56510186005315,
      "grad_norm": 0.23874498903751373,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 54830
    },
    {
      "epoch": 48.56953055801594,
      "grad_norm": 0.26916539669036865,
      "learning_rate": 1e-05,
      "loss": 0.9427,
      "step": 54835
    },
    {
      "epoch": 48.57395925597874,
      "grad_norm": 0.22758601605892181,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 54840
    },
    {
      "epoch": 48.578387953941544,
      "grad_norm": 0.2415654957294464,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 54845
    },
    {
      "epoch": 48.58281665190434,
      "grad_norm": 0.27855557203292847,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 54850
    },
    {
      "epoch": 48.58724534986714,
      "grad_norm": 0.2430223971605301,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 54855
    },
    {
      "epoch": 48.59167404782994,
      "grad_norm": 0.2339971363544464,
      "learning_rate": 1e-05,
      "loss": 1.0437,
      "step": 54860
    },
    {
      "epoch": 48.596102745792734,
      "grad_norm": 0.2321372628211975,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 54865
    },
    {
      "epoch": 48.600531443755536,
      "grad_norm": 0.23226068913936615,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 54870
    },
    {
      "epoch": 48.60496014171834,
      "grad_norm": 0.19789043068885803,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 54875
    },
    {
      "epoch": 48.60938883968113,
      "grad_norm": 0.2241741269826889,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 54880
    },
    {
      "epoch": 48.61381753764393,
      "grad_norm": 0.22698169946670532,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 54885
    },
    {
      "epoch": 48.61824623560673,
      "grad_norm": 0.2848401963710785,
      "learning_rate": 1e-05,
      "loss": 1.0345,
      "step": 54890
    },
    {
      "epoch": 48.62267493356953,
      "grad_norm": 0.2099856585264206,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 54895
    },
    {
      "epoch": 48.62710363153233,
      "grad_norm": 0.24514034390449524,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 54900
    },
    {
      "epoch": 48.63153232949513,
      "grad_norm": 0.2556520998477936,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 54905
    },
    {
      "epoch": 48.635961027457924,
      "grad_norm": 0.22490465641021729,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 54910
    },
    {
      "epoch": 48.640389725420725,
      "grad_norm": 0.2267404943704605,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 54915
    },
    {
      "epoch": 48.644818423383526,
      "grad_norm": 0.25889432430267334,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 54920
    },
    {
      "epoch": 48.64924712134633,
      "grad_norm": 0.27019771933555603,
      "learning_rate": 1e-05,
      "loss": 1.0759,
      "step": 54925
    },
    {
      "epoch": 48.65367581930912,
      "grad_norm": 0.22395308315753937,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 54930
    },
    {
      "epoch": 48.65810451727192,
      "grad_norm": 0.22619661688804626,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 54935
    },
    {
      "epoch": 48.662533215234724,
      "grad_norm": 0.2858215868473053,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 54940
    },
    {
      "epoch": 48.66696191319752,
      "grad_norm": 0.2700244188308716,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 54945
    },
    {
      "epoch": 48.67139061116032,
      "grad_norm": 0.2147485613822937,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 54950
    },
    {
      "epoch": 48.67581930912312,
      "grad_norm": 0.23243673145771027,
      "learning_rate": 1e-05,
      "loss": 1.0111,
      "step": 54955
    },
    {
      "epoch": 48.680248007085915,
      "grad_norm": 0.2546096444129944,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 54960
    },
    {
      "epoch": 48.684676705048716,
      "grad_norm": 0.2619113028049469,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 54965
    },
    {
      "epoch": 48.68910540301152,
      "grad_norm": 0.2887438237667084,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 54970
    },
    {
      "epoch": 48.69353410097431,
      "grad_norm": 0.23999159038066864,
      "learning_rate": 1e-05,
      "loss": 1.0149,
      "step": 54975
    },
    {
      "epoch": 48.69796279893711,
      "grad_norm": 0.2825166583061218,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 54980
    },
    {
      "epoch": 48.702391496899914,
      "grad_norm": 0.2746540606021881,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 54985
    },
    {
      "epoch": 48.70682019486271,
      "grad_norm": 0.21760587394237518,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 54990
    },
    {
      "epoch": 48.71124889282551,
      "grad_norm": 0.25527656078338623,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 54995
    },
    {
      "epoch": 48.71567759078831,
      "grad_norm": 0.21348458528518677,
      "learning_rate": 1e-05,
      "loss": 1.0256,
      "step": 55000
    },
    {
      "epoch": 48.720106288751104,
      "grad_norm": 0.2200413942337036,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 55005
    },
    {
      "epoch": 48.724534986713905,
      "grad_norm": 0.26661837100982666,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 55010
    },
    {
      "epoch": 48.72896368467671,
      "grad_norm": 0.2207280397415161,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 55015
    },
    {
      "epoch": 48.7333923826395,
      "grad_norm": 0.21707503497600555,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 55020
    },
    {
      "epoch": 48.7378210806023,
      "grad_norm": 0.21548376977443695,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 55025
    },
    {
      "epoch": 48.7422497785651,
      "grad_norm": 0.3349223732948303,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 55030
    },
    {
      "epoch": 48.7466784765279,
      "grad_norm": 0.2791121006011963,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 55035
    },
    {
      "epoch": 48.7511071744907,
      "grad_norm": 0.24870958924293518,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 55040
    },
    {
      "epoch": 48.7555358724535,
      "grad_norm": 0.2350766658782959,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 55045
    },
    {
      "epoch": 48.7599645704163,
      "grad_norm": 0.26081809401512146,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 55050
    },
    {
      "epoch": 48.764393268379095,
      "grad_norm": 0.24910062551498413,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 55055
    },
    {
      "epoch": 48.768821966341896,
      "grad_norm": 0.29383233189582825,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 55060
    },
    {
      "epoch": 48.7732506643047,
      "grad_norm": 0.24867188930511475,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 55065
    },
    {
      "epoch": 48.77767936226749,
      "grad_norm": 0.23956997692584991,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 55070
    },
    {
      "epoch": 48.78210806023029,
      "grad_norm": 0.25531238317489624,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 55075
    },
    {
      "epoch": 48.786536758193094,
      "grad_norm": 0.23537547886371613,
      "learning_rate": 1e-05,
      "loss": 1.0054,
      "step": 55080
    },
    {
      "epoch": 48.79096545615589,
      "grad_norm": 0.26684388518333435,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 55085
    },
    {
      "epoch": 48.79539415411869,
      "grad_norm": 0.2253996580839157,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 55090
    },
    {
      "epoch": 48.79982285208149,
      "grad_norm": 0.2288932353258133,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 55095
    },
    {
      "epoch": 48.804251550044285,
      "grad_norm": 0.25369077920913696,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 55100
    },
    {
      "epoch": 48.808680248007086,
      "grad_norm": 0.2032015323638916,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 55105
    },
    {
      "epoch": 48.81310894596989,
      "grad_norm": 0.2597110867500305,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 55110
    },
    {
      "epoch": 48.81753764393268,
      "grad_norm": 0.2512738108634949,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 55115
    },
    {
      "epoch": 48.82196634189548,
      "grad_norm": 0.35800260305404663,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 55120
    },
    {
      "epoch": 48.826395039858284,
      "grad_norm": 0.2719745934009552,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 55125
    },
    {
      "epoch": 48.83082373782108,
      "grad_norm": 0.21742922067642212,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 55130
    },
    {
      "epoch": 48.83525243578388,
      "grad_norm": 0.2501205503940582,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 55135
    },
    {
      "epoch": 48.83968113374668,
      "grad_norm": 0.26601022481918335,
      "learning_rate": 1e-05,
      "loss": 0.9126,
      "step": 55140
    },
    {
      "epoch": 48.844109831709474,
      "grad_norm": 0.24707823991775513,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 55145
    },
    {
      "epoch": 48.848538529672275,
      "grad_norm": 0.2571483850479126,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 55150
    },
    {
      "epoch": 48.85296722763508,
      "grad_norm": 0.262888103723526,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 55155
    },
    {
      "epoch": 48.85739592559787,
      "grad_norm": 0.21559688448905945,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 55160
    },
    {
      "epoch": 48.86182462356067,
      "grad_norm": 0.24982097744941711,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 55165
    },
    {
      "epoch": 48.86625332152347,
      "grad_norm": 0.24011340737342834,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 55170
    },
    {
      "epoch": 48.870682019486274,
      "grad_norm": 0.23663382232189178,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 55175
    },
    {
      "epoch": 48.87511071744907,
      "grad_norm": 0.22923614084720612,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 55180
    },
    {
      "epoch": 48.87953941541187,
      "grad_norm": 0.22929036617279053,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 55185
    },
    {
      "epoch": 48.88396811337467,
      "grad_norm": 0.2659350335597992,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 55190
    },
    {
      "epoch": 48.888396811337465,
      "grad_norm": 0.26742982864379883,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 55195
    },
    {
      "epoch": 48.892825509300266,
      "grad_norm": 0.26196664571762085,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 55200
    },
    {
      "epoch": 48.89725420726307,
      "grad_norm": 0.1893482208251953,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 55205
    },
    {
      "epoch": 48.90168290522586,
      "grad_norm": 0.24410848319530487,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 55210
    },
    {
      "epoch": 48.90611160318866,
      "grad_norm": 0.2510742247104645,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 55215
    },
    {
      "epoch": 48.910540301151464,
      "grad_norm": 0.2423769235610962,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 55220
    },
    {
      "epoch": 48.91496899911426,
      "grad_norm": 0.24973231554031372,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 55225
    },
    {
      "epoch": 48.91939769707706,
      "grad_norm": 0.2149386703968048,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 55230
    },
    {
      "epoch": 48.92382639503986,
      "grad_norm": 0.2188480794429779,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 55235
    },
    {
      "epoch": 48.928255093002655,
      "grad_norm": 0.2532457709312439,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 55240
    },
    {
      "epoch": 48.932683790965456,
      "grad_norm": 0.28841453790664673,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 55245
    },
    {
      "epoch": 48.93711248892826,
      "grad_norm": 0.23716680705547333,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 55250
    },
    {
      "epoch": 48.94154118689105,
      "grad_norm": 0.239442378282547,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 55255
    },
    {
      "epoch": 48.94596988485385,
      "grad_norm": 0.2541291415691376,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 55260
    },
    {
      "epoch": 48.95039858281665,
      "grad_norm": 0.27392786741256714,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 55265
    },
    {
      "epoch": 48.95482728077945,
      "grad_norm": 0.24729777872562408,
      "learning_rate": 1e-05,
      "loss": 1.0285,
      "step": 55270
    },
    {
      "epoch": 48.95925597874225,
      "grad_norm": 0.2830755114555359,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 55275
    },
    {
      "epoch": 48.96368467670505,
      "grad_norm": 0.2020728439092636,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 55280
    },
    {
      "epoch": 48.96811337466785,
      "grad_norm": 0.2305692881345749,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 55285
    },
    {
      "epoch": 48.972542072630645,
      "grad_norm": 0.22978775203227997,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 55290
    },
    {
      "epoch": 48.97697077059345,
      "grad_norm": 0.28941699862480164,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 55295
    },
    {
      "epoch": 48.98139946855625,
      "grad_norm": 0.2947419285774231,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 55300
    },
    {
      "epoch": 48.98582816651904,
      "grad_norm": 0.21778425574302673,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 55305
    },
    {
      "epoch": 48.99025686448184,
      "grad_norm": 0.3094514310359955,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 55310
    },
    {
      "epoch": 48.994685562444644,
      "grad_norm": 0.22837582230567932,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 55315
    },
    {
      "epoch": 48.99911426040744,
      "grad_norm": 0.20818109810352325,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 55320
    },
    {
      "epoch": 49.00354295837024,
      "grad_norm": 0.28077957034111023,
      "learning_rate": 1e-05,
      "loss": 0.9192,
      "step": 55325
    },
    {
      "epoch": 49.00797165633304,
      "grad_norm": 0.2179676741361618,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 55330
    },
    {
      "epoch": 49.012400354295835,
      "grad_norm": 0.2138155847787857,
      "learning_rate": 1e-05,
      "loss": 0.8731,
      "step": 55335
    },
    {
      "epoch": 49.016829052258636,
      "grad_norm": 0.31002724170684814,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 55340
    },
    {
      "epoch": 49.02125775022144,
      "grad_norm": 0.23113851249217987,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 55345
    },
    {
      "epoch": 49.02568644818423,
      "grad_norm": 0.2922385632991791,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 55350
    },
    {
      "epoch": 49.03011514614703,
      "grad_norm": 0.28313514590263367,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 55355
    },
    {
      "epoch": 49.034543844109834,
      "grad_norm": 0.2304021269083023,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 55360
    },
    {
      "epoch": 49.03897254207263,
      "grad_norm": 0.23780737817287445,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 55365
    },
    {
      "epoch": 49.04340124003543,
      "grad_norm": 0.2541655898094177,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 55370
    },
    {
      "epoch": 49.04782993799823,
      "grad_norm": 0.23372630774974823,
      "learning_rate": 1e-05,
      "loss": 0.9154,
      "step": 55375
    },
    {
      "epoch": 49.052258635961024,
      "grad_norm": 0.2583569288253784,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 55380
    },
    {
      "epoch": 49.056687333923826,
      "grad_norm": 0.28064998984336853,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 55385
    },
    {
      "epoch": 49.06111603188663,
      "grad_norm": 0.23718559741973877,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 55390
    },
    {
      "epoch": 49.06554472984942,
      "grad_norm": 0.24140843749046326,
      "learning_rate": 1e-05,
      "loss": 0.9324,
      "step": 55395
    },
    {
      "epoch": 49.06997342781222,
      "grad_norm": 0.250434935092926,
      "learning_rate": 1e-05,
      "loss": 1.0298,
      "step": 55400
    },
    {
      "epoch": 49.07440212577502,
      "grad_norm": 0.2934398651123047,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 55405
    },
    {
      "epoch": 49.078830823737825,
      "grad_norm": 0.21975839138031006,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 55410
    },
    {
      "epoch": 49.08325952170062,
      "grad_norm": 0.24368907511234283,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 55415
    },
    {
      "epoch": 49.08768821966342,
      "grad_norm": 0.2522338032722473,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 55420
    },
    {
      "epoch": 49.09211691762622,
      "grad_norm": 0.30634424090385437,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 55425
    },
    {
      "epoch": 49.096545615589015,
      "grad_norm": 0.27043506503105164,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 55430
    },
    {
      "epoch": 49.100974313551816,
      "grad_norm": 0.24883215129375458,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 55435
    },
    {
      "epoch": 49.10540301151462,
      "grad_norm": 0.2334386110305786,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 55440
    },
    {
      "epoch": 49.10983170947741,
      "grad_norm": 0.2280300408601761,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 55445
    },
    {
      "epoch": 49.11426040744021,
      "grad_norm": 0.22359693050384521,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 55450
    },
    {
      "epoch": 49.118689105403014,
      "grad_norm": 0.22574204206466675,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 55455
    },
    {
      "epoch": 49.12311780336581,
      "grad_norm": 0.26959648728370667,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 55460
    },
    {
      "epoch": 49.12754650132861,
      "grad_norm": 0.2523326873779297,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 55465
    },
    {
      "epoch": 49.13197519929141,
      "grad_norm": 0.26202499866485596,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 55470
    },
    {
      "epoch": 49.136403897254205,
      "grad_norm": 0.2383251190185547,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 55475
    },
    {
      "epoch": 49.140832595217006,
      "grad_norm": 0.22497080266475677,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 55480
    },
    {
      "epoch": 49.14526129317981,
      "grad_norm": 0.363065242767334,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 55485
    },
    {
      "epoch": 49.1496899911426,
      "grad_norm": 0.3056315779685974,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 55490
    },
    {
      "epoch": 49.1541186891054,
      "grad_norm": 0.2373078316450119,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 55495
    },
    {
      "epoch": 49.158547387068204,
      "grad_norm": 0.3665105104446411,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 55500
    },
    {
      "epoch": 49.162976085031,
      "grad_norm": 0.2530321776866913,
      "learning_rate": 1e-05,
      "loss": 1.0281,
      "step": 55505
    },
    {
      "epoch": 49.1674047829938,
      "grad_norm": 0.2833568751811981,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 55510
    },
    {
      "epoch": 49.1718334809566,
      "grad_norm": 0.24073496460914612,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 55515
    },
    {
      "epoch": 49.176262178919394,
      "grad_norm": 0.23924696445465088,
      "learning_rate": 1e-05,
      "loss": 1.0373,
      "step": 55520
    },
    {
      "epoch": 49.180690876882196,
      "grad_norm": 0.24713070690631866,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 55525
    },
    {
      "epoch": 49.185119574845,
      "grad_norm": 0.2487911432981491,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 55530
    },
    {
      "epoch": 49.1895482728078,
      "grad_norm": 0.2104041874408722,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 55535
    },
    {
      "epoch": 49.19397697077059,
      "grad_norm": 0.2619360387325287,
      "learning_rate": 1e-05,
      "loss": 0.9249,
      "step": 55540
    },
    {
      "epoch": 49.19840566873339,
      "grad_norm": 0.2509959638118744,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 55545
    },
    {
      "epoch": 49.202834366696194,
      "grad_norm": 0.20325206220149994,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 55550
    },
    {
      "epoch": 49.20726306465899,
      "grad_norm": 0.2482072114944458,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 55555
    },
    {
      "epoch": 49.21169176262179,
      "grad_norm": 0.24647383391857147,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 55560
    },
    {
      "epoch": 49.21612046058459,
      "grad_norm": 0.3046664595603943,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 55565
    },
    {
      "epoch": 49.220549158547385,
      "grad_norm": 0.25379493832588196,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 55570
    },
    {
      "epoch": 49.224977856510186,
      "grad_norm": 0.24232938885688782,
      "learning_rate": 1e-05,
      "loss": 0.9187,
      "step": 55575
    },
    {
      "epoch": 49.22940655447299,
      "grad_norm": 0.27989089488983154,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 55580
    },
    {
      "epoch": 49.23383525243578,
      "grad_norm": 0.24518831074237823,
      "learning_rate": 1e-05,
      "loss": 0.9185,
      "step": 55585
    },
    {
      "epoch": 49.23826395039858,
      "grad_norm": 0.2454182207584381,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 55590
    },
    {
      "epoch": 49.242692648361384,
      "grad_norm": 0.21499726176261902,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 55595
    },
    {
      "epoch": 49.24712134632418,
      "grad_norm": 0.2654757797718048,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 55600
    },
    {
      "epoch": 49.25155004428698,
      "grad_norm": 0.20775367319583893,
      "learning_rate": 1e-05,
      "loss": 1.0113,
      "step": 55605
    },
    {
      "epoch": 49.25597874224978,
      "grad_norm": 0.22467748820781708,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 55610
    },
    {
      "epoch": 49.260407440212575,
      "grad_norm": 0.27343037724494934,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 55615
    },
    {
      "epoch": 49.264836138175376,
      "grad_norm": 0.21698635816574097,
      "learning_rate": 1e-05,
      "loss": 1.0323,
      "step": 55620
    },
    {
      "epoch": 49.26926483613818,
      "grad_norm": 0.220206618309021,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 55625
    },
    {
      "epoch": 49.27369353410097,
      "grad_norm": 0.23211440443992615,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 55630
    },
    {
      "epoch": 49.27812223206377,
      "grad_norm": 0.20855194330215454,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 55635
    },
    {
      "epoch": 49.282550930026574,
      "grad_norm": 0.2288268357515335,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 55640
    },
    {
      "epoch": 49.28697962798937,
      "grad_norm": 0.2961913049221039,
      "learning_rate": 1e-05,
      "loss": 1.0407,
      "step": 55645
    },
    {
      "epoch": 49.29140832595217,
      "grad_norm": 0.2488223910331726,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 55650
    },
    {
      "epoch": 49.29583702391497,
      "grad_norm": 0.2510570287704468,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 55655
    },
    {
      "epoch": 49.30026572187777,
      "grad_norm": 0.2550215423107147,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 55660
    },
    {
      "epoch": 49.304694419840565,
      "grad_norm": 0.2813078761100769,
      "learning_rate": 1e-05,
      "loss": 1.0561,
      "step": 55665
    },
    {
      "epoch": 49.30912311780337,
      "grad_norm": 0.2419459968805313,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 55670
    },
    {
      "epoch": 49.31355181576617,
      "grad_norm": 0.22254474461078644,
      "learning_rate": 1e-05,
      "loss": 0.9188,
      "step": 55675
    },
    {
      "epoch": 49.31798051372896,
      "grad_norm": 0.20271334052085876,
      "learning_rate": 1e-05,
      "loss": 0.9141,
      "step": 55680
    },
    {
      "epoch": 49.32240921169176,
      "grad_norm": 0.22568435966968536,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 55685
    },
    {
      "epoch": 49.326837909654564,
      "grad_norm": 0.2265196591615677,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 55690
    },
    {
      "epoch": 49.33126660761736,
      "grad_norm": 0.20363369584083557,
      "learning_rate": 1e-05,
      "loss": 0.9264,
      "step": 55695
    },
    {
      "epoch": 49.33569530558016,
      "grad_norm": 0.2713947892189026,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 55700
    },
    {
      "epoch": 49.34012400354296,
      "grad_norm": 0.21646949648857117,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 55705
    },
    {
      "epoch": 49.344552701505755,
      "grad_norm": 0.24461214244365692,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 55710
    },
    {
      "epoch": 49.348981399468556,
      "grad_norm": 0.21809403598308563,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 55715
    },
    {
      "epoch": 49.35341009743136,
      "grad_norm": 0.23036445677280426,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 55720
    },
    {
      "epoch": 49.35783879539415,
      "grad_norm": 0.24903655052185059,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 55725
    },
    {
      "epoch": 49.36226749335695,
      "grad_norm": 0.22437074780464172,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 55730
    },
    {
      "epoch": 49.366696191319754,
      "grad_norm": 0.2485104650259018,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 55735
    },
    {
      "epoch": 49.37112488928255,
      "grad_norm": 0.27319252490997314,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 55740
    },
    {
      "epoch": 49.37555358724535,
      "grad_norm": 0.24037888646125793,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 55745
    },
    {
      "epoch": 49.37998228520815,
      "grad_norm": 0.2011907398700714,
      "learning_rate": 1e-05,
      "loss": 1.0278,
      "step": 55750
    },
    {
      "epoch": 49.384410983170945,
      "grad_norm": 0.2516098916530609,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 55755
    },
    {
      "epoch": 49.388839681133746,
      "grad_norm": 0.21974478662014008,
      "learning_rate": 1e-05,
      "loss": 1.0609,
      "step": 55760
    },
    {
      "epoch": 49.39326837909655,
      "grad_norm": 0.22345365583896637,
      "learning_rate": 1e-05,
      "loss": 0.9019,
      "step": 55765
    },
    {
      "epoch": 49.39769707705934,
      "grad_norm": 0.28094354271888733,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 55770
    },
    {
      "epoch": 49.40212577502214,
      "grad_norm": 0.2135746031999588,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 55775
    },
    {
      "epoch": 49.40655447298494,
      "grad_norm": 0.22173616290092468,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 55780
    },
    {
      "epoch": 49.410983170947745,
      "grad_norm": 0.19864509999752045,
      "learning_rate": 1e-05,
      "loss": 0.9164,
      "step": 55785
    },
    {
      "epoch": 49.41541186891054,
      "grad_norm": 0.23265066742897034,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 55790
    },
    {
      "epoch": 49.41984056687334,
      "grad_norm": 0.21120624244213104,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 55795
    },
    {
      "epoch": 49.42426926483614,
      "grad_norm": 0.22201330959796906,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 55800
    },
    {
      "epoch": 49.428697962798935,
      "grad_norm": 0.21634584665298462,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 55805
    },
    {
      "epoch": 49.43312666076174,
      "grad_norm": 0.23441413044929504,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 55810
    },
    {
      "epoch": 49.43755535872454,
      "grad_norm": 0.23186996579170227,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 55815
    },
    {
      "epoch": 49.44198405668733,
      "grad_norm": 0.22654154896736145,
      "learning_rate": 1e-05,
      "loss": 0.9099,
      "step": 55820
    },
    {
      "epoch": 49.44641275465013,
      "grad_norm": 0.2407923936843872,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 55825
    },
    {
      "epoch": 49.450841452612934,
      "grad_norm": 0.23158203065395355,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 55830
    },
    {
      "epoch": 49.45527015057573,
      "grad_norm": 0.2513697147369385,
      "learning_rate": 1e-05,
      "loss": 1.0309,
      "step": 55835
    },
    {
      "epoch": 49.45969884853853,
      "grad_norm": 0.23609429597854614,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 55840
    },
    {
      "epoch": 49.46412754650133,
      "grad_norm": 0.27875301241874695,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 55845
    },
    {
      "epoch": 49.468556244464125,
      "grad_norm": 0.2642456889152527,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 55850
    },
    {
      "epoch": 49.472984942426926,
      "grad_norm": 0.24122361838817596,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 55855
    },
    {
      "epoch": 49.47741364038973,
      "grad_norm": 0.22178712487220764,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 55860
    },
    {
      "epoch": 49.48184233835252,
      "grad_norm": 0.23801103234291077,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 55865
    },
    {
      "epoch": 49.48627103631532,
      "grad_norm": 0.2524906396865845,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 55870
    },
    {
      "epoch": 49.490699734278124,
      "grad_norm": 0.22201824188232422,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 55875
    },
    {
      "epoch": 49.49512843224092,
      "grad_norm": 0.2502390444278717,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 55880
    },
    {
      "epoch": 49.49955713020372,
      "grad_norm": 0.23540426790714264,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 55885
    },
    {
      "epoch": 49.50398582816652,
      "grad_norm": 0.23848573863506317,
      "learning_rate": 1e-05,
      "loss": 1.0155,
      "step": 55890
    },
    {
      "epoch": 49.508414526129314,
      "grad_norm": 0.22835133969783783,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 55895
    },
    {
      "epoch": 49.512843224092116,
      "grad_norm": 0.22900861501693726,
      "learning_rate": 1e-05,
      "loss": 1.0224,
      "step": 55900
    },
    {
      "epoch": 49.51727192205492,
      "grad_norm": 0.27086400985717773,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 55905
    },
    {
      "epoch": 49.52170062001772,
      "grad_norm": 0.2634629011154175,
      "learning_rate": 1e-05,
      "loss": 1.0256,
      "step": 55910
    },
    {
      "epoch": 49.52612931798051,
      "grad_norm": 0.22731907665729523,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 55915
    },
    {
      "epoch": 49.53055801594331,
      "grad_norm": 0.26295971870422363,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 55920
    },
    {
      "epoch": 49.534986713906115,
      "grad_norm": 0.2209186851978302,
      "learning_rate": 1e-05,
      "loss": 0.9199,
      "step": 55925
    },
    {
      "epoch": 49.53941541186891,
      "grad_norm": 0.274909108877182,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 55930
    },
    {
      "epoch": 49.54384410983171,
      "grad_norm": 0.23730503022670746,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 55935
    },
    {
      "epoch": 49.54827280779451,
      "grad_norm": 0.25347045063972473,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 55940
    },
    {
      "epoch": 49.552701505757305,
      "grad_norm": 0.27422139048576355,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 55945
    },
    {
      "epoch": 49.55713020372011,
      "grad_norm": 0.22927168011665344,
      "learning_rate": 1e-05,
      "loss": 1.0513,
      "step": 55950
    },
    {
      "epoch": 49.56155890168291,
      "grad_norm": 0.2302422821521759,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 55955
    },
    {
      "epoch": 49.5659875996457,
      "grad_norm": 0.2383836954832077,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 55960
    },
    {
      "epoch": 49.5704162976085,
      "grad_norm": 0.36718258261680603,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 55965
    },
    {
      "epoch": 49.574844995571304,
      "grad_norm": 0.23085416853427887,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 55970
    },
    {
      "epoch": 49.5792736935341,
      "grad_norm": 0.23889276385307312,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 55975
    },
    {
      "epoch": 49.5837023914969,
      "grad_norm": 0.19973666965961456,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 55980
    },
    {
      "epoch": 49.5881310894597,
      "grad_norm": 0.2542799115180969,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 55985
    },
    {
      "epoch": 49.592559787422495,
      "grad_norm": 0.22943666577339172,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 55990
    },
    {
      "epoch": 49.596988485385296,
      "grad_norm": 0.25890737771987915,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 55995
    },
    {
      "epoch": 49.6014171833481,
      "grad_norm": 0.2690388560295105,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 56000
    },
    {
      "epoch": 49.60584588131089,
      "grad_norm": 0.23407095670700073,
      "learning_rate": 1e-05,
      "loss": 0.9413,
      "step": 56005
    },
    {
      "epoch": 49.61027457927369,
      "grad_norm": 0.25506293773651123,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 56010
    },
    {
      "epoch": 49.614703277236494,
      "grad_norm": 0.2488018125295639,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 56015
    },
    {
      "epoch": 49.619131975199295,
      "grad_norm": 0.2653346061706543,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 56020
    },
    {
      "epoch": 49.62356067316209,
      "grad_norm": 0.2499607354402542,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 56025
    },
    {
      "epoch": 49.62798937112489,
      "grad_norm": 0.32554203271865845,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 56030
    },
    {
      "epoch": 49.63241806908769,
      "grad_norm": 0.23772965371608734,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 56035
    },
    {
      "epoch": 49.636846767050486,
      "grad_norm": 0.2505536377429962,
      "learning_rate": 1e-05,
      "loss": 1.0324,
      "step": 56040
    },
    {
      "epoch": 49.64127546501329,
      "grad_norm": 0.2918280363082886,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 56045
    },
    {
      "epoch": 49.64570416297609,
      "grad_norm": 0.22287245094776154,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 56050
    },
    {
      "epoch": 49.65013286093888,
      "grad_norm": 0.22217696905136108,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 56055
    },
    {
      "epoch": 49.65456155890168,
      "grad_norm": 0.24594427645206451,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 56060
    },
    {
      "epoch": 49.658990256864485,
      "grad_norm": 0.21668019890785217,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 56065
    },
    {
      "epoch": 49.66341895482728,
      "grad_norm": 0.23349064588546753,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 56070
    },
    {
      "epoch": 49.66784765279008,
      "grad_norm": 0.2514871656894684,
      "learning_rate": 1e-05,
      "loss": 1.0396,
      "step": 56075
    },
    {
      "epoch": 49.67227635075288,
      "grad_norm": 0.24177692830562592,
      "learning_rate": 1e-05,
      "loss": 0.9138,
      "step": 56080
    },
    {
      "epoch": 49.676705048715675,
      "grad_norm": 0.2102849930524826,
      "learning_rate": 1e-05,
      "loss": 0.925,
      "step": 56085
    },
    {
      "epoch": 49.681133746678476,
      "grad_norm": 0.21542243659496307,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 56090
    },
    {
      "epoch": 49.68556244464128,
      "grad_norm": 0.2515002191066742,
      "learning_rate": 1e-05,
      "loss": 1.0193,
      "step": 56095
    },
    {
      "epoch": 49.68999114260407,
      "grad_norm": 0.24471041560173035,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 56100
    },
    {
      "epoch": 49.69441984056687,
      "grad_norm": 0.23970235884189606,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 56105
    },
    {
      "epoch": 49.698848538529674,
      "grad_norm": 0.2658633887767792,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 56110
    },
    {
      "epoch": 49.70327723649247,
      "grad_norm": 0.2718665599822998,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 56115
    },
    {
      "epoch": 49.70770593445527,
      "grad_norm": 0.23668673634529114,
      "learning_rate": 1e-05,
      "loss": 0.9261,
      "step": 56120
    },
    {
      "epoch": 49.71213463241807,
      "grad_norm": 0.2891448438167572,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 56125
    },
    {
      "epoch": 49.716563330380865,
      "grad_norm": 0.19356776773929596,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 56130
    },
    {
      "epoch": 49.720992028343666,
      "grad_norm": 0.23055894672870636,
      "learning_rate": 1e-05,
      "loss": 0.9231,
      "step": 56135
    },
    {
      "epoch": 49.72542072630647,
      "grad_norm": 0.21958231925964355,
      "learning_rate": 1e-05,
      "loss": 0.9155,
      "step": 56140
    },
    {
      "epoch": 49.72984942426926,
      "grad_norm": 0.21882109344005585,
      "learning_rate": 1e-05,
      "loss": 0.9233,
      "step": 56145
    },
    {
      "epoch": 49.73427812223206,
      "grad_norm": 0.21798716485500336,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 56150
    },
    {
      "epoch": 49.738706820194864,
      "grad_norm": 0.23968777060508728,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 56155
    },
    {
      "epoch": 49.743135518157665,
      "grad_norm": 0.2466309666633606,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 56160
    },
    {
      "epoch": 49.74756421612046,
      "grad_norm": 0.2300512045621872,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 56165
    },
    {
      "epoch": 49.75199291408326,
      "grad_norm": 0.2710813283920288,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 56170
    },
    {
      "epoch": 49.75642161204606,
      "grad_norm": 0.258838027715683,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 56175
    },
    {
      "epoch": 49.760850310008856,
      "grad_norm": 0.20486044883728027,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 56180
    },
    {
      "epoch": 49.76527900797166,
      "grad_norm": 0.28276970982551575,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 56185
    },
    {
      "epoch": 49.76970770593446,
      "grad_norm": 0.23578336834907532,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 56190
    },
    {
      "epoch": 49.77413640389725,
      "grad_norm": 0.2509406805038452,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 56195
    },
    {
      "epoch": 49.77856510186005,
      "grad_norm": 0.24274130165576935,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 56200
    },
    {
      "epoch": 49.782993799822854,
      "grad_norm": 0.24467487633228302,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 56205
    },
    {
      "epoch": 49.78742249778565,
      "grad_norm": 0.2643640339374542,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 56210
    },
    {
      "epoch": 49.79185119574845,
      "grad_norm": 0.25002598762512207,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 56215
    },
    {
      "epoch": 49.79627989371125,
      "grad_norm": 0.2515376806259155,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 56220
    },
    {
      "epoch": 49.800708591674045,
      "grad_norm": 0.23220328986644745,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 56225
    },
    {
      "epoch": 49.805137289636846,
      "grad_norm": 0.22976285219192505,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 56230
    },
    {
      "epoch": 49.80956598759965,
      "grad_norm": 0.2217559665441513,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 56235
    },
    {
      "epoch": 49.81399468556244,
      "grad_norm": 0.249554842710495,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 56240
    },
    {
      "epoch": 49.81842338352524,
      "grad_norm": 0.2501259446144104,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 56245
    },
    {
      "epoch": 49.822852081488044,
      "grad_norm": 0.22955256700515747,
      "learning_rate": 1e-05,
      "loss": 0.9188,
      "step": 56250
    },
    {
      "epoch": 49.82728077945084,
      "grad_norm": 0.24748598039150238,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 56255
    },
    {
      "epoch": 49.83170947741364,
      "grad_norm": 0.23254449665546417,
      "learning_rate": 1e-05,
      "loss": 1.0387,
      "step": 56260
    },
    {
      "epoch": 49.83613817537644,
      "grad_norm": 0.2125118225812912,
      "learning_rate": 1e-05,
      "loss": 0.9247,
      "step": 56265
    },
    {
      "epoch": 49.84056687333924,
      "grad_norm": 0.2508005201816559,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 56270
    },
    {
      "epoch": 49.844995571302036,
      "grad_norm": 0.24274130165576935,
      "learning_rate": 1e-05,
      "loss": 0.9213,
      "step": 56275
    },
    {
      "epoch": 49.84942426926484,
      "grad_norm": 0.28861716389656067,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 56280
    },
    {
      "epoch": 49.85385296722764,
      "grad_norm": 0.2913484275341034,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 56285
    },
    {
      "epoch": 49.85828166519043,
      "grad_norm": 0.2529178559780121,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 56290
    },
    {
      "epoch": 49.862710363153234,
      "grad_norm": 0.29265162348747253,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 56295
    },
    {
      "epoch": 49.867139061116035,
      "grad_norm": 0.22390113770961761,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 56300
    },
    {
      "epoch": 49.87156775907883,
      "grad_norm": 0.2138739824295044,
      "learning_rate": 1e-05,
      "loss": 0.9061,
      "step": 56305
    },
    {
      "epoch": 49.87599645704163,
      "grad_norm": 0.22618485987186432,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 56310
    },
    {
      "epoch": 49.88042515500443,
      "grad_norm": 0.22366419434547424,
      "learning_rate": 1e-05,
      "loss": 0.9338,
      "step": 56315
    },
    {
      "epoch": 49.884853852967225,
      "grad_norm": 0.25958943367004395,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 56320
    },
    {
      "epoch": 49.88928255093003,
      "grad_norm": 0.24752451479434967,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 56325
    },
    {
      "epoch": 49.89371124889283,
      "grad_norm": 0.24582603573799133,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 56330
    },
    {
      "epoch": 49.89813994685562,
      "grad_norm": 0.2090013325214386,
      "learning_rate": 1e-05,
      "loss": 0.9462,
      "step": 56335
    },
    {
      "epoch": 49.90256864481842,
      "grad_norm": 0.216317281126976,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 56340
    },
    {
      "epoch": 49.906997342781224,
      "grad_norm": 0.20565621554851532,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 56345
    },
    {
      "epoch": 49.91142604074402,
      "grad_norm": 0.22992201149463654,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 56350
    },
    {
      "epoch": 49.91585473870682,
      "grad_norm": 0.24856363236904144,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 56355
    },
    {
      "epoch": 49.92028343666962,
      "grad_norm": 0.24793480336666107,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 56360
    },
    {
      "epoch": 49.924712134632415,
      "grad_norm": 0.2484944462776184,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 56365
    },
    {
      "epoch": 49.929140832595216,
      "grad_norm": 0.24044661223888397,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 56370
    },
    {
      "epoch": 49.93356953055802,
      "grad_norm": 0.2586592733860016,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 56375
    },
    {
      "epoch": 49.93799822852081,
      "grad_norm": 0.22751368582248688,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 56380
    },
    {
      "epoch": 49.94242692648361,
      "grad_norm": 0.2624330222606659,
      "learning_rate": 1e-05,
      "loss": 1.0404,
      "step": 56385
    },
    {
      "epoch": 49.946855624446414,
      "grad_norm": 0.23961280286312103,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 56390
    },
    {
      "epoch": 49.951284322409215,
      "grad_norm": 0.28060203790664673,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 56395
    },
    {
      "epoch": 49.95571302037201,
      "grad_norm": 0.2486191689968109,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 56400
    },
    {
      "epoch": 49.96014171833481,
      "grad_norm": 0.22898992896080017,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 56405
    },
    {
      "epoch": 49.96457041629761,
      "grad_norm": 0.20768101513385773,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 56410
    },
    {
      "epoch": 49.968999114260406,
      "grad_norm": 0.25775259733200073,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 56415
    },
    {
      "epoch": 49.97342781222321,
      "grad_norm": 0.2138027846813202,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 56420
    },
    {
      "epoch": 49.97785651018601,
      "grad_norm": 0.24015085399150848,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 56425
    },
    {
      "epoch": 49.9822852081488,
      "grad_norm": 0.24268838763237,
      "learning_rate": 1e-05,
      "loss": 0.9274,
      "step": 56430
    },
    {
      "epoch": 49.9867139061116,
      "grad_norm": 0.23679347336292267,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 56435
    },
    {
      "epoch": 49.991142604074405,
      "grad_norm": 0.2042813003063202,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 56440
    },
    {
      "epoch": 49.9955713020372,
      "grad_norm": 0.2307308465242386,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 56445
    },
    {
      "epoch": 50.0,
      "grad_norm": 0.20903560519218445,
      "learning_rate": 1e-05,
      "loss": 0.9109,
      "step": 56450
    },
    {
      "epoch": 50.0044286979628,
      "grad_norm": 0.256826251745224,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 56455
    },
    {
      "epoch": 50.008857395925595,
      "grad_norm": 0.28415969014167786,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 56460
    },
    {
      "epoch": 50.0132860938884,
      "grad_norm": 0.24987296760082245,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 56465
    },
    {
      "epoch": 50.0177147918512,
      "grad_norm": 0.2883746027946472,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 56470
    },
    {
      "epoch": 50.02214348981399,
      "grad_norm": 0.2436843365430832,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 56475
    },
    {
      "epoch": 50.02657218777679,
      "grad_norm": 0.25531429052352905,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 56480
    },
    {
      "epoch": 50.031000885739594,
      "grad_norm": 0.23130466043949127,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 56485
    },
    {
      "epoch": 50.03542958370239,
      "grad_norm": 0.28787776827812195,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 56490
    },
    {
      "epoch": 50.03985828166519,
      "grad_norm": 0.22041937708854675,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 56495
    },
    {
      "epoch": 50.04428697962799,
      "grad_norm": 0.22946397960186005,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 56500
    },
    {
      "epoch": 50.048715677590785,
      "grad_norm": 0.2536393702030182,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 56505
    },
    {
      "epoch": 50.053144375553586,
      "grad_norm": 0.24152427911758423,
      "learning_rate": 1e-05,
      "loss": 0.9273,
      "step": 56510
    },
    {
      "epoch": 50.05757307351639,
      "grad_norm": 0.2372935265302658,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 56515
    },
    {
      "epoch": 50.06200177147919,
      "grad_norm": 0.2500895857810974,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 56520
    },
    {
      "epoch": 50.06643046944198,
      "grad_norm": 0.258232980966568,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 56525
    },
    {
      "epoch": 50.070859167404784,
      "grad_norm": 0.27169135212898254,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 56530
    },
    {
      "epoch": 50.075287865367585,
      "grad_norm": 0.22759337723255157,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 56535
    },
    {
      "epoch": 50.07971656333038,
      "grad_norm": 0.2306303232908249,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 56540
    },
    {
      "epoch": 50.08414526129318,
      "grad_norm": 0.3019965887069702,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 56545
    },
    {
      "epoch": 50.08857395925598,
      "grad_norm": 0.2953367531299591,
      "learning_rate": 1e-05,
      "loss": 0.9194,
      "step": 56550
    },
    {
      "epoch": 50.093002657218776,
      "grad_norm": 0.22084151208400726,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 56555
    },
    {
      "epoch": 50.09743135518158,
      "grad_norm": 0.2320505976676941,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 56560
    },
    {
      "epoch": 50.10186005314438,
      "grad_norm": 0.23217327892780304,
      "learning_rate": 1e-05,
      "loss": 1.0213,
      "step": 56565
    },
    {
      "epoch": 50.10628875110717,
      "grad_norm": 0.21947097778320312,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 56570
    },
    {
      "epoch": 50.11071744906997,
      "grad_norm": 0.219364732503891,
      "learning_rate": 1e-05,
      "loss": 0.9179,
      "step": 56575
    },
    {
      "epoch": 50.115146147032775,
      "grad_norm": 0.23172561824321747,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 56580
    },
    {
      "epoch": 50.11957484499557,
      "grad_norm": 0.24873532354831696,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 56585
    },
    {
      "epoch": 50.12400354295837,
      "grad_norm": 0.2821251451969147,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 56590
    },
    {
      "epoch": 50.12843224092117,
      "grad_norm": 0.22603997588157654,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 56595
    },
    {
      "epoch": 50.132860938883965,
      "grad_norm": 0.26354584097862244,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 56600
    },
    {
      "epoch": 50.137289636846766,
      "grad_norm": 0.22468827664852142,
      "learning_rate": 1e-05,
      "loss": 1.0216,
      "step": 56605
    },
    {
      "epoch": 50.14171833480957,
      "grad_norm": 0.2797124981880188,
      "learning_rate": 1e-05,
      "loss": 0.9207,
      "step": 56610
    },
    {
      "epoch": 50.14614703277236,
      "grad_norm": 0.2565435767173767,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 56615
    },
    {
      "epoch": 50.15057573073516,
      "grad_norm": 0.25486549735069275,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 56620
    },
    {
      "epoch": 50.155004428697964,
      "grad_norm": 0.22764211893081665,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 56625
    },
    {
      "epoch": 50.15943312666076,
      "grad_norm": 0.2739883065223694,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 56630
    },
    {
      "epoch": 50.16386182462356,
      "grad_norm": 0.23255641758441925,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 56635
    },
    {
      "epoch": 50.16829052258636,
      "grad_norm": 0.28168371319770813,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 56640
    },
    {
      "epoch": 50.17271922054916,
      "grad_norm": 0.2573017179965973,
      "learning_rate": 1e-05,
      "loss": 0.9133,
      "step": 56645
    },
    {
      "epoch": 50.177147918511956,
      "grad_norm": 0.24071170389652252,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 56650
    },
    {
      "epoch": 50.18157661647476,
      "grad_norm": 0.27621400356292725,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 56655
    },
    {
      "epoch": 50.18600531443756,
      "grad_norm": 0.21441973745822906,
      "learning_rate": 1e-05,
      "loss": 0.9239,
      "step": 56660
    },
    {
      "epoch": 50.19043401240035,
      "grad_norm": 0.2881956100463867,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 56665
    },
    {
      "epoch": 50.194862710363154,
      "grad_norm": 0.24569925665855408,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 56670
    },
    {
      "epoch": 50.199291408325955,
      "grad_norm": 0.21228636801242828,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 56675
    },
    {
      "epoch": 50.20372010628875,
      "grad_norm": 0.23929989337921143,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 56680
    },
    {
      "epoch": 50.20814880425155,
      "grad_norm": 0.24594345688819885,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 56685
    },
    {
      "epoch": 50.21257750221435,
      "grad_norm": 0.23644809424877167,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 56690
    },
    {
      "epoch": 50.217006200177146,
      "grad_norm": 0.23983396589756012,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 56695
    },
    {
      "epoch": 50.22143489813995,
      "grad_norm": 0.2325279265642166,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 56700
    },
    {
      "epoch": 50.22586359610275,
      "grad_norm": 0.24303388595581055,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 56705
    },
    {
      "epoch": 50.23029229406554,
      "grad_norm": 0.22000473737716675,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 56710
    },
    {
      "epoch": 50.23472099202834,
      "grad_norm": 0.29801854491233826,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 56715
    },
    {
      "epoch": 50.239149689991144,
      "grad_norm": 0.1950961798429489,
      "learning_rate": 1e-05,
      "loss": 0.9095,
      "step": 56720
    },
    {
      "epoch": 50.24357838795394,
      "grad_norm": 0.2332957237958908,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 56725
    },
    {
      "epoch": 50.24800708591674,
      "grad_norm": 0.24296385049819946,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 56730
    },
    {
      "epoch": 50.25243578387954,
      "grad_norm": 0.21270343661308289,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 56735
    },
    {
      "epoch": 50.256864481842335,
      "grad_norm": 0.2131119966506958,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 56740
    },
    {
      "epoch": 50.261293179805136,
      "grad_norm": 0.244968980550766,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 56745
    },
    {
      "epoch": 50.26572187776794,
      "grad_norm": 0.23098643124103546,
      "learning_rate": 1e-05,
      "loss": 1.0405,
      "step": 56750
    },
    {
      "epoch": 50.27015057573073,
      "grad_norm": 0.28778502345085144,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 56755
    },
    {
      "epoch": 50.27457927369353,
      "grad_norm": 0.2344450056552887,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 56760
    },
    {
      "epoch": 50.279007971656334,
      "grad_norm": 0.2548731565475464,
      "learning_rate": 1e-05,
      "loss": 0.9111,
      "step": 56765
    },
    {
      "epoch": 50.283436669619135,
      "grad_norm": 0.26159363985061646,
      "learning_rate": 1e-05,
      "loss": 1.0261,
      "step": 56770
    },
    {
      "epoch": 50.28786536758193,
      "grad_norm": 0.2169611006975174,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 56775
    },
    {
      "epoch": 50.29229406554473,
      "grad_norm": 0.2346806526184082,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 56780
    },
    {
      "epoch": 50.29672276350753,
      "grad_norm": 0.22332145273685455,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 56785
    },
    {
      "epoch": 50.301151461470326,
      "grad_norm": 0.24709761142730713,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 56790
    },
    {
      "epoch": 50.30558015943313,
      "grad_norm": 0.25908616185188293,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 56795
    },
    {
      "epoch": 50.31000885739593,
      "grad_norm": 0.2909523546695709,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 56800
    },
    {
      "epoch": 50.31443755535872,
      "grad_norm": 0.2912805378437042,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 56805
    },
    {
      "epoch": 50.318866253321524,
      "grad_norm": 0.26910462975502014,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 56810
    },
    {
      "epoch": 50.323294951284325,
      "grad_norm": 0.3363611102104187,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 56815
    },
    {
      "epoch": 50.32772364924712,
      "grad_norm": 0.24463380873203278,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 56820
    },
    {
      "epoch": 50.33215234720992,
      "grad_norm": 0.19113120436668396,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 56825
    },
    {
      "epoch": 50.33658104517272,
      "grad_norm": 0.2420957237482071,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 56830
    },
    {
      "epoch": 50.341009743135515,
      "grad_norm": 0.2199445515871048,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 56835
    },
    {
      "epoch": 50.34543844109832,
      "grad_norm": 0.2994040250778198,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 56840
    },
    {
      "epoch": 50.34986713906112,
      "grad_norm": 0.22258982062339783,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 56845
    },
    {
      "epoch": 50.35429583702391,
      "grad_norm": 0.2610372006893158,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 56850
    },
    {
      "epoch": 50.35872453498671,
      "grad_norm": 0.28715774416923523,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 56855
    },
    {
      "epoch": 50.363153232949514,
      "grad_norm": 0.25883904099464417,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 56860
    },
    {
      "epoch": 50.36758193091231,
      "grad_norm": 0.24693803489208221,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 56865
    },
    {
      "epoch": 50.37201062887511,
      "grad_norm": 0.21784591674804688,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 56870
    },
    {
      "epoch": 50.37643932683791,
      "grad_norm": 0.22341829538345337,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 56875
    },
    {
      "epoch": 50.380868024800705,
      "grad_norm": 0.23047322034835815,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 56880
    },
    {
      "epoch": 50.385296722763506,
      "grad_norm": 0.20648445188999176,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 56885
    },
    {
      "epoch": 50.38972542072631,
      "grad_norm": 0.21252861618995667,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 56890
    },
    {
      "epoch": 50.39415411868911,
      "grad_norm": 0.22594702243804932,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 56895
    },
    {
      "epoch": 50.3985828166519,
      "grad_norm": 0.23270262777805328,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 56900
    },
    {
      "epoch": 50.403011514614704,
      "grad_norm": 0.2160016894340515,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 56905
    },
    {
      "epoch": 50.407440212577505,
      "grad_norm": 0.24679569900035858,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 56910
    },
    {
      "epoch": 50.4118689105403,
      "grad_norm": 0.2292606085538864,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 56915
    },
    {
      "epoch": 50.4162976085031,
      "grad_norm": 0.2807747423648834,
      "learning_rate": 1e-05,
      "loss": 0.9195,
      "step": 56920
    },
    {
      "epoch": 50.4207263064659,
      "grad_norm": 0.23891091346740723,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 56925
    },
    {
      "epoch": 50.425155004428696,
      "grad_norm": 0.24489007890224457,
      "learning_rate": 1e-05,
      "loss": 0.9184,
      "step": 56930
    },
    {
      "epoch": 50.4295837023915,
      "grad_norm": 0.2609945237636566,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 56935
    },
    {
      "epoch": 50.4340124003543,
      "grad_norm": 0.2422347366809845,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 56940
    },
    {
      "epoch": 50.43844109831709,
      "grad_norm": 0.2711182236671448,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 56945
    },
    {
      "epoch": 50.44286979627989,
      "grad_norm": 0.23269061744213104,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 56950
    },
    {
      "epoch": 50.447298494242695,
      "grad_norm": 0.20936839282512665,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 56955
    },
    {
      "epoch": 50.45172719220549,
      "grad_norm": 0.2907756567001343,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 56960
    },
    {
      "epoch": 50.45615589016829,
      "grad_norm": 0.2308727502822876,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 56965
    },
    {
      "epoch": 50.46058458813109,
      "grad_norm": 0.28089532256126404,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 56970
    },
    {
      "epoch": 50.465013286093885,
      "grad_norm": 0.25694093108177185,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 56975
    },
    {
      "epoch": 50.46944198405669,
      "grad_norm": 0.21947243809700012,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 56980
    },
    {
      "epoch": 50.47387068201949,
      "grad_norm": 0.21457254886627197,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 56985
    },
    {
      "epoch": 50.47829937998228,
      "grad_norm": 0.2480718046426773,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 56990
    },
    {
      "epoch": 50.48272807794508,
      "grad_norm": 0.2229345291852951,
      "learning_rate": 1e-05,
      "loss": 0.8955,
      "step": 56995
    },
    {
      "epoch": 50.487156775907884,
      "grad_norm": 0.2067663073539734,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 57000
    },
    {
      "epoch": 50.491585473870686,
      "grad_norm": 0.2179500013589859,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 57005
    },
    {
      "epoch": 50.49601417183348,
      "grad_norm": 0.21371547877788544,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 57010
    },
    {
      "epoch": 50.50044286979628,
      "grad_norm": 0.2740658223628998,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 57015
    },
    {
      "epoch": 50.50487156775908,
      "grad_norm": 0.2660622298717499,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 57020
    },
    {
      "epoch": 50.509300265721876,
      "grad_norm": 0.2441960722208023,
      "learning_rate": 1e-05,
      "loss": 0.9165,
      "step": 57025
    },
    {
      "epoch": 50.51372896368468,
      "grad_norm": 0.24387258291244507,
      "learning_rate": 1e-05,
      "loss": 1.0255,
      "step": 57030
    },
    {
      "epoch": 50.51815766164748,
      "grad_norm": 0.24928133189678192,
      "learning_rate": 1e-05,
      "loss": 0.924,
      "step": 57035
    },
    {
      "epoch": 50.52258635961027,
      "grad_norm": 0.2652866840362549,
      "learning_rate": 1e-05,
      "loss": 0.9144,
      "step": 57040
    },
    {
      "epoch": 50.527015057573074,
      "grad_norm": 0.19332289695739746,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 57045
    },
    {
      "epoch": 50.531443755535875,
      "grad_norm": 0.20250988006591797,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 57050
    },
    {
      "epoch": 50.53587245349867,
      "grad_norm": 0.23639582097530365,
      "learning_rate": 1e-05,
      "loss": 1.0186,
      "step": 57055
    },
    {
      "epoch": 50.54030115146147,
      "grad_norm": 0.26621049642562866,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 57060
    },
    {
      "epoch": 50.54472984942427,
      "grad_norm": 0.2117260992527008,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 57065
    },
    {
      "epoch": 50.549158547387066,
      "grad_norm": 0.1955445408821106,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 57070
    },
    {
      "epoch": 50.55358724534987,
      "grad_norm": 0.2266274094581604,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 57075
    },
    {
      "epoch": 50.55801594331267,
      "grad_norm": 0.23665519058704376,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 57080
    },
    {
      "epoch": 50.56244464127546,
      "grad_norm": 0.260951966047287,
      "learning_rate": 1e-05,
      "loss": 1.0097,
      "step": 57085
    },
    {
      "epoch": 50.56687333923826,
      "grad_norm": 0.2310333549976349,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 57090
    },
    {
      "epoch": 50.571302037201065,
      "grad_norm": 0.26228269934654236,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 57095
    },
    {
      "epoch": 50.57573073516386,
      "grad_norm": 0.22783657908439636,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 57100
    },
    {
      "epoch": 50.58015943312666,
      "grad_norm": 0.21104858815670013,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 57105
    },
    {
      "epoch": 50.58458813108946,
      "grad_norm": 0.21797597408294678,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 57110
    },
    {
      "epoch": 50.589016829052255,
      "grad_norm": 0.29294660687446594,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 57115
    },
    {
      "epoch": 50.59344552701506,
      "grad_norm": 0.2762058973312378,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 57120
    },
    {
      "epoch": 50.59787422497786,
      "grad_norm": 0.3349522054195404,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 57125
    },
    {
      "epoch": 50.60230292294066,
      "grad_norm": 0.25857076048851013,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 57130
    },
    {
      "epoch": 50.60673162090345,
      "grad_norm": 0.3034328520298004,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 57135
    },
    {
      "epoch": 50.611160318866254,
      "grad_norm": 0.21597638726234436,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 57140
    },
    {
      "epoch": 50.615589016829055,
      "grad_norm": 0.2535270154476166,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 57145
    },
    {
      "epoch": 50.62001771479185,
      "grad_norm": 0.26540327072143555,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 57150
    },
    {
      "epoch": 50.62444641275465,
      "grad_norm": 0.2667843997478485,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 57155
    },
    {
      "epoch": 50.62887511071745,
      "grad_norm": 0.24924424290657043,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 57160
    },
    {
      "epoch": 50.633303808680246,
      "grad_norm": 0.22557270526885986,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 57165
    },
    {
      "epoch": 50.63773250664305,
      "grad_norm": 0.2317129224538803,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 57170
    },
    {
      "epoch": 50.64216120460585,
      "grad_norm": 0.22792044281959534,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 57175
    },
    {
      "epoch": 50.64658990256864,
      "grad_norm": 0.2907612919807434,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 57180
    },
    {
      "epoch": 50.651018600531444,
      "grad_norm": 0.25826042890548706,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 57185
    },
    {
      "epoch": 50.655447298494245,
      "grad_norm": 0.23000043630599976,
      "learning_rate": 1e-05,
      "loss": 0.9054,
      "step": 57190
    },
    {
      "epoch": 50.65987599645704,
      "grad_norm": 0.21878407895565033,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 57195
    },
    {
      "epoch": 50.66430469441984,
      "grad_norm": 0.2538566589355469,
      "learning_rate": 1e-05,
      "loss": 0.9317,
      "step": 57200
    },
    {
      "epoch": 50.66873339238264,
      "grad_norm": 0.2425651103258133,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 57205
    },
    {
      "epoch": 50.673162090345436,
      "grad_norm": 0.23855812847614288,
      "learning_rate": 1e-05,
      "loss": 0.9235,
      "step": 57210
    },
    {
      "epoch": 50.67759078830824,
      "grad_norm": 0.2407747507095337,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 57215
    },
    {
      "epoch": 50.68201948627104,
      "grad_norm": 0.23114387691020966,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 57220
    },
    {
      "epoch": 50.68644818423383,
      "grad_norm": 0.24291351437568665,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 57225
    },
    {
      "epoch": 50.69087688219663,
      "grad_norm": 0.21679618954658508,
      "learning_rate": 1e-05,
      "loss": 0.9197,
      "step": 57230
    },
    {
      "epoch": 50.695305580159435,
      "grad_norm": 0.2194240540266037,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 57235
    },
    {
      "epoch": 50.69973427812223,
      "grad_norm": 0.2436395138502121,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 57240
    },
    {
      "epoch": 50.70416297608503,
      "grad_norm": 0.26533743739128113,
      "learning_rate": 1e-05,
      "loss": 1.034,
      "step": 57245
    },
    {
      "epoch": 50.70859167404783,
      "grad_norm": 0.24759019911289215,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 57250
    },
    {
      "epoch": 50.71302037201063,
      "grad_norm": 0.2354304939508438,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 57255
    },
    {
      "epoch": 50.717449069973426,
      "grad_norm": 0.23703746497631073,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 57260
    },
    {
      "epoch": 50.72187776793623,
      "grad_norm": 0.23415827751159668,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 57265
    },
    {
      "epoch": 50.72630646589903,
      "grad_norm": 0.22479312121868134,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 57270
    },
    {
      "epoch": 50.73073516386182,
      "grad_norm": 0.2348138391971588,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 57275
    },
    {
      "epoch": 50.735163861824624,
      "grad_norm": 0.2475343644618988,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 57280
    },
    {
      "epoch": 50.739592559787425,
      "grad_norm": 0.222883939743042,
      "learning_rate": 1e-05,
      "loss": 1.0705,
      "step": 57285
    },
    {
      "epoch": 50.74402125775022,
      "grad_norm": 0.25917381048202515,
      "learning_rate": 1e-05,
      "loss": 0.8941,
      "step": 57290
    },
    {
      "epoch": 50.74844995571302,
      "grad_norm": 0.21381698548793793,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 57295
    },
    {
      "epoch": 50.75287865367582,
      "grad_norm": 0.3045254349708557,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 57300
    },
    {
      "epoch": 50.757307351638616,
      "grad_norm": 0.2285705804824829,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 57305
    },
    {
      "epoch": 50.76173604960142,
      "grad_norm": 0.28514525294303894,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 57310
    },
    {
      "epoch": 50.76616474756422,
      "grad_norm": 0.2468743771314621,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 57315
    },
    {
      "epoch": 50.77059344552701,
      "grad_norm": 0.26387229561805725,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 57320
    },
    {
      "epoch": 50.775022143489814,
      "grad_norm": 0.29074862599372864,
      "learning_rate": 1e-05,
      "loss": 0.9259,
      "step": 57325
    },
    {
      "epoch": 50.779450841452615,
      "grad_norm": 0.2334967404603958,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 57330
    },
    {
      "epoch": 50.78387953941541,
      "grad_norm": 0.2385290414094925,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 57335
    },
    {
      "epoch": 50.78830823737821,
      "grad_norm": 0.24056480824947357,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 57340
    },
    {
      "epoch": 50.79273693534101,
      "grad_norm": 0.25320765376091003,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 57345
    },
    {
      "epoch": 50.797165633303806,
      "grad_norm": 0.2529081404209137,
      "learning_rate": 1e-05,
      "loss": 0.9243,
      "step": 57350
    },
    {
      "epoch": 50.80159433126661,
      "grad_norm": 0.25356802344322205,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 57355
    },
    {
      "epoch": 50.80602302922941,
      "grad_norm": 0.23200450837612152,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 57360
    },
    {
      "epoch": 50.8104517271922,
      "grad_norm": 0.2691291868686676,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 57365
    },
    {
      "epoch": 50.814880425155,
      "grad_norm": 0.22183488309383392,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 57370
    },
    {
      "epoch": 50.819309123117804,
      "grad_norm": 0.2938096523284912,
      "learning_rate": 1e-05,
      "loss": 1.0272,
      "step": 57375
    },
    {
      "epoch": 50.823737821080606,
      "grad_norm": 0.20272104442119598,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 57380
    },
    {
      "epoch": 50.8281665190434,
      "grad_norm": 0.24421222507953644,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 57385
    },
    {
      "epoch": 50.8325952170062,
      "grad_norm": 0.25468650460243225,
      "learning_rate": 1e-05,
      "loss": 1.0416,
      "step": 57390
    },
    {
      "epoch": 50.837023914969,
      "grad_norm": 0.21876370906829834,
      "learning_rate": 1e-05,
      "loss": 0.9244,
      "step": 57395
    },
    {
      "epoch": 50.841452612931796,
      "grad_norm": 0.27346208691596985,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 57400
    },
    {
      "epoch": 50.8458813108946,
      "grad_norm": 0.26185932755470276,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 57405
    },
    {
      "epoch": 50.8503100088574,
      "grad_norm": 0.28568413853645325,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 57410
    },
    {
      "epoch": 50.85473870682019,
      "grad_norm": 0.22601120173931122,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 57415
    },
    {
      "epoch": 50.859167404782994,
      "grad_norm": 0.28095149993896484,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 57420
    },
    {
      "epoch": 50.863596102745795,
      "grad_norm": 0.26968300342559814,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 57425
    },
    {
      "epoch": 50.86802480070859,
      "grad_norm": 0.27069154381752014,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 57430
    },
    {
      "epoch": 50.87245349867139,
      "grad_norm": 0.21626657247543335,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 57435
    },
    {
      "epoch": 50.87688219663419,
      "grad_norm": 0.2365463227033615,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 57440
    },
    {
      "epoch": 50.881310894596986,
      "grad_norm": 0.2627531886100769,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 57445
    },
    {
      "epoch": 50.88573959255979,
      "grad_norm": 0.24699030816555023,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 57450
    },
    {
      "epoch": 50.89016829052259,
      "grad_norm": 0.24027204513549805,
      "learning_rate": 1e-05,
      "loss": 0.9018,
      "step": 57455
    },
    {
      "epoch": 50.89459698848538,
      "grad_norm": 0.22851277887821198,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 57460
    },
    {
      "epoch": 50.899025686448184,
      "grad_norm": 0.25088736414909363,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 57465
    },
    {
      "epoch": 50.903454384410985,
      "grad_norm": 0.22899356484413147,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 57470
    },
    {
      "epoch": 50.90788308237378,
      "grad_norm": 0.22089873254299164,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 57475
    },
    {
      "epoch": 50.91231178033658,
      "grad_norm": 0.2423372119665146,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 57480
    },
    {
      "epoch": 50.91674047829938,
      "grad_norm": 0.2189023792743683,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 57485
    },
    {
      "epoch": 50.921169176262175,
      "grad_norm": 0.2416541427373886,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 57490
    },
    {
      "epoch": 50.92559787422498,
      "grad_norm": 0.2738267481327057,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 57495
    },
    {
      "epoch": 50.93002657218778,
      "grad_norm": 0.233888640999794,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 57500
    },
    {
      "epoch": 50.93445527015058,
      "grad_norm": 0.22839553654193878,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 57505
    },
    {
      "epoch": 50.93888396811337,
      "grad_norm": 0.2798800468444824,
      "learning_rate": 1e-05,
      "loss": 0.9462,
      "step": 57510
    },
    {
      "epoch": 50.943312666076174,
      "grad_norm": 0.20363402366638184,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 57515
    },
    {
      "epoch": 50.947741364038976,
      "grad_norm": 0.27506139874458313,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 57520
    },
    {
      "epoch": 50.95217006200177,
      "grad_norm": 0.2358459234237671,
      "learning_rate": 1e-05,
      "loss": 1.0208,
      "step": 57525
    },
    {
      "epoch": 50.95659875996457,
      "grad_norm": 0.22368596494197845,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 57530
    },
    {
      "epoch": 50.96102745792737,
      "grad_norm": 0.26471343636512756,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 57535
    },
    {
      "epoch": 50.965456155890166,
      "grad_norm": 0.24759797751903534,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 57540
    },
    {
      "epoch": 50.96988485385297,
      "grad_norm": 0.21278832852840424,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 57545
    },
    {
      "epoch": 50.97431355181577,
      "grad_norm": 0.22913242876529694,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 57550
    },
    {
      "epoch": 50.97874224977856,
      "grad_norm": 0.28342562913894653,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 57555
    },
    {
      "epoch": 50.983170947741364,
      "grad_norm": 0.2534969747066498,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 57560
    },
    {
      "epoch": 50.987599645704165,
      "grad_norm": 0.3334161639213562,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 57565
    },
    {
      "epoch": 50.99202834366696,
      "grad_norm": 0.24183982610702515,
      "learning_rate": 1e-05,
      "loss": 1.0285,
      "step": 57570
    },
    {
      "epoch": 50.99645704162976,
      "grad_norm": 0.23501867055892944,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 57575
    },
    {
      "epoch": 51.00088573959256,
      "grad_norm": 0.2851235866546631,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 57580
    },
    {
      "epoch": 51.005314437555356,
      "grad_norm": 0.21883146464824677,
      "learning_rate": 1e-05,
      "loss": 1.0316,
      "step": 57585
    },
    {
      "epoch": 51.00974313551816,
      "grad_norm": 0.2711237370967865,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 57590
    },
    {
      "epoch": 51.01417183348096,
      "grad_norm": 0.22978302836418152,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 57595
    },
    {
      "epoch": 51.01860053144375,
      "grad_norm": 0.21813155710697174,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 57600
    },
    {
      "epoch": 51.02302922940655,
      "grad_norm": 0.2518755793571472,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 57605
    },
    {
      "epoch": 51.027457927369355,
      "grad_norm": 0.22469745576381683,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 57610
    },
    {
      "epoch": 51.03188662533215,
      "grad_norm": 0.27151429653167725,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 57615
    },
    {
      "epoch": 51.03631532329495,
      "grad_norm": 0.2737997770309448,
      "learning_rate": 1e-05,
      "loss": 1.0624,
      "step": 57620
    },
    {
      "epoch": 51.04074402125775,
      "grad_norm": 0.21020102500915527,
      "learning_rate": 1e-05,
      "loss": 0.9175,
      "step": 57625
    },
    {
      "epoch": 51.04517271922055,
      "grad_norm": 0.2312561571598053,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 57630
    },
    {
      "epoch": 51.04960141718335,
      "grad_norm": 0.2422817200422287,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 57635
    },
    {
      "epoch": 51.05403011514615,
      "grad_norm": 0.21907754242420197,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 57640
    },
    {
      "epoch": 51.05845881310895,
      "grad_norm": 0.2058318704366684,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 57645
    },
    {
      "epoch": 51.06288751107174,
      "grad_norm": 0.24165132641792297,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 57650
    },
    {
      "epoch": 51.067316209034544,
      "grad_norm": 0.2245476394891739,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 57655
    },
    {
      "epoch": 51.071744906997345,
      "grad_norm": 0.21088151633739471,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 57660
    },
    {
      "epoch": 51.07617360496014,
      "grad_norm": 0.2458590865135193,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 57665
    },
    {
      "epoch": 51.08060230292294,
      "grad_norm": 0.2767459750175476,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 57670
    },
    {
      "epoch": 51.08503100088574,
      "grad_norm": 0.22455215454101562,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 57675
    },
    {
      "epoch": 51.089459698848536,
      "grad_norm": 0.2461453229188919,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 57680
    },
    {
      "epoch": 51.09388839681134,
      "grad_norm": 0.20260339975357056,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 57685
    },
    {
      "epoch": 51.09831709477414,
      "grad_norm": 0.2502760589122772,
      "learning_rate": 1e-05,
      "loss": 0.9046,
      "step": 57690
    },
    {
      "epoch": 51.10274579273693,
      "grad_norm": 0.2781901955604553,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 57695
    },
    {
      "epoch": 51.107174490699734,
      "grad_norm": 0.22900964319705963,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 57700
    },
    {
      "epoch": 51.111603188662535,
      "grad_norm": 0.22861196100711823,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 57705
    },
    {
      "epoch": 51.11603188662533,
      "grad_norm": 0.30818867683410645,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 57710
    },
    {
      "epoch": 51.12046058458813,
      "grad_norm": 0.2083790898323059,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 57715
    },
    {
      "epoch": 51.12488928255093,
      "grad_norm": 0.2377614825963974,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 57720
    },
    {
      "epoch": 51.129317980513726,
      "grad_norm": 0.2846793830394745,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 57725
    },
    {
      "epoch": 51.13374667847653,
      "grad_norm": 0.284907728433609,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 57730
    },
    {
      "epoch": 51.13817537643933,
      "grad_norm": 0.2716125547885895,
      "learning_rate": 1e-05,
      "loss": 0.9296,
      "step": 57735
    },
    {
      "epoch": 51.14260407440213,
      "grad_norm": 0.25886595249176025,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 57740
    },
    {
      "epoch": 51.14703277236492,
      "grad_norm": 0.29313427209854126,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 57745
    },
    {
      "epoch": 51.151461470327725,
      "grad_norm": 0.2650250196456909,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 57750
    },
    {
      "epoch": 51.155890168290526,
      "grad_norm": 0.25280410051345825,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 57755
    },
    {
      "epoch": 51.16031886625332,
      "grad_norm": 0.2823229730129242,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 57760
    },
    {
      "epoch": 51.16474756421612,
      "grad_norm": 0.2749771773815155,
      "learning_rate": 1e-05,
      "loss": 1.0363,
      "step": 57765
    },
    {
      "epoch": 51.16917626217892,
      "grad_norm": 0.26156389713287354,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 57770
    },
    {
      "epoch": 51.173604960141716,
      "grad_norm": 0.23654894530773163,
      "learning_rate": 1e-05,
      "loss": 1.0199,
      "step": 57775
    },
    {
      "epoch": 51.17803365810452,
      "grad_norm": 0.2652944028377533,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 57780
    },
    {
      "epoch": 51.18246235606732,
      "grad_norm": 0.2423141747713089,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 57785
    },
    {
      "epoch": 51.18689105403011,
      "grad_norm": 0.25852829217910767,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 57790
    },
    {
      "epoch": 51.191319751992914,
      "grad_norm": 0.22827711701393127,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 57795
    },
    {
      "epoch": 51.195748449955715,
      "grad_norm": 0.2187909483909607,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 57800
    },
    {
      "epoch": 51.20017714791851,
      "grad_norm": 0.2305816113948822,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 57805
    },
    {
      "epoch": 51.20460584588131,
      "grad_norm": 0.24282309412956238,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 57810
    },
    {
      "epoch": 51.20903454384411,
      "grad_norm": 0.21139930188655853,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 57815
    },
    {
      "epoch": 51.213463241806906,
      "grad_norm": 0.2523021101951599,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 57820
    },
    {
      "epoch": 51.21789193976971,
      "grad_norm": 0.24135370552539825,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 57825
    },
    {
      "epoch": 51.22232063773251,
      "grad_norm": 0.27971288561820984,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 57830
    },
    {
      "epoch": 51.2267493356953,
      "grad_norm": 0.222804456949234,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 57835
    },
    {
      "epoch": 51.231178033658104,
      "grad_norm": 0.20857460796833038,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 57840
    },
    {
      "epoch": 51.235606731620905,
      "grad_norm": 0.228929802775383,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 57845
    },
    {
      "epoch": 51.2400354295837,
      "grad_norm": 0.2285294383764267,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 57850
    },
    {
      "epoch": 51.2444641275465,
      "grad_norm": 0.2479756772518158,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 57855
    },
    {
      "epoch": 51.2488928255093,
      "grad_norm": 0.24195349216461182,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 57860
    },
    {
      "epoch": 51.2533215234721,
      "grad_norm": 0.24460451304912567,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 57865
    },
    {
      "epoch": 51.2577502214349,
      "grad_norm": 0.23649516701698303,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 57870
    },
    {
      "epoch": 51.2621789193977,
      "grad_norm": 0.2504752278327942,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 57875
    },
    {
      "epoch": 51.2666076173605,
      "grad_norm": 0.30173248052597046,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 57880
    },
    {
      "epoch": 51.27103631532329,
      "grad_norm": 0.24462582170963287,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 57885
    },
    {
      "epoch": 51.275465013286095,
      "grad_norm": 0.2838738262653351,
      "learning_rate": 1e-05,
      "loss": 0.9163,
      "step": 57890
    },
    {
      "epoch": 51.279893711248896,
      "grad_norm": 0.2863309383392334,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 57895
    },
    {
      "epoch": 51.28432240921169,
      "grad_norm": 0.24349147081375122,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 57900
    },
    {
      "epoch": 51.28875110717449,
      "grad_norm": 0.2216518521308899,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 57905
    },
    {
      "epoch": 51.29317980513729,
      "grad_norm": 0.20430532097816467,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 57910
    },
    {
      "epoch": 51.297608503100086,
      "grad_norm": 0.2482028603553772,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 57915
    },
    {
      "epoch": 51.30203720106289,
      "grad_norm": 0.23034650087356567,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 57920
    },
    {
      "epoch": 51.30646589902569,
      "grad_norm": 0.25499558448791504,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 57925
    },
    {
      "epoch": 51.31089459698848,
      "grad_norm": 0.21269191801548004,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 57930
    },
    {
      "epoch": 51.315323294951284,
      "grad_norm": 0.26415735483169556,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 57935
    },
    {
      "epoch": 51.319751992914085,
      "grad_norm": 0.23134873807430267,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 57940
    },
    {
      "epoch": 51.32418069087688,
      "grad_norm": 0.2758760154247284,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 57945
    },
    {
      "epoch": 51.32860938883968,
      "grad_norm": 0.2471761554479599,
      "learning_rate": 1e-05,
      "loss": 0.9056,
      "step": 57950
    },
    {
      "epoch": 51.33303808680248,
      "grad_norm": 0.2567176818847656,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 57955
    },
    {
      "epoch": 51.337466784765276,
      "grad_norm": 0.2770969271659851,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 57960
    },
    {
      "epoch": 51.34189548272808,
      "grad_norm": 0.2107272446155548,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 57965
    },
    {
      "epoch": 51.34632418069088,
      "grad_norm": 0.24066130816936493,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 57970
    },
    {
      "epoch": 51.35075287865367,
      "grad_norm": 0.20916897058486938,
      "learning_rate": 1e-05,
      "loss": 1.0464,
      "step": 57975
    },
    {
      "epoch": 51.355181576616474,
      "grad_norm": 0.23597535490989685,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 57980
    },
    {
      "epoch": 51.359610274579275,
      "grad_norm": 0.2893282473087311,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 57985
    },
    {
      "epoch": 51.364038972542076,
      "grad_norm": 0.20720933377742767,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 57990
    },
    {
      "epoch": 51.36846767050487,
      "grad_norm": 0.21616671979427338,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 57995
    },
    {
      "epoch": 51.37289636846767,
      "grad_norm": 0.22247853875160217,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 58000
    },
    {
      "epoch": 51.37732506643047,
      "grad_norm": 0.1928655356168747,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 58005
    },
    {
      "epoch": 51.38175376439327,
      "grad_norm": 0.2537802755832672,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 58010
    },
    {
      "epoch": 51.38618246235607,
      "grad_norm": 0.24991494417190552,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 58015
    },
    {
      "epoch": 51.39061116031887,
      "grad_norm": 0.22834308445453644,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 58020
    },
    {
      "epoch": 51.39503985828166,
      "grad_norm": 0.23812471330165863,
      "learning_rate": 1e-05,
      "loss": 0.9237,
      "step": 58025
    },
    {
      "epoch": 51.399468556244464,
      "grad_norm": 0.21515800058841705,
      "learning_rate": 1e-05,
      "loss": 0.9292,
      "step": 58030
    },
    {
      "epoch": 51.403897254207266,
      "grad_norm": 0.242784321308136,
      "learning_rate": 1e-05,
      "loss": 1.0316,
      "step": 58035
    },
    {
      "epoch": 51.40832595217006,
      "grad_norm": 0.23172993957996368,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 58040
    },
    {
      "epoch": 51.41275465013286,
      "grad_norm": 0.2517952024936676,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 58045
    },
    {
      "epoch": 51.41718334809566,
      "grad_norm": 0.23771822452545166,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 58050
    },
    {
      "epoch": 51.421612046058456,
      "grad_norm": 0.2543613314628601,
      "learning_rate": 1e-05,
      "loss": 1.0384,
      "step": 58055
    },
    {
      "epoch": 51.42604074402126,
      "grad_norm": 0.25572437047958374,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 58060
    },
    {
      "epoch": 51.43046944198406,
      "grad_norm": 0.2784786820411682,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 58065
    },
    {
      "epoch": 51.43489813994685,
      "grad_norm": 0.22722876071929932,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 58070
    },
    {
      "epoch": 51.439326837909654,
      "grad_norm": 0.23458446562290192,
      "learning_rate": 1e-05,
      "loss": 0.9258,
      "step": 58075
    },
    {
      "epoch": 51.443755535872455,
      "grad_norm": 0.2506619691848755,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 58080
    },
    {
      "epoch": 51.44818423383525,
      "grad_norm": 0.22852088510990143,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 58085
    },
    {
      "epoch": 51.45261293179805,
      "grad_norm": 0.23912429809570312,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 58090
    },
    {
      "epoch": 51.45704162976085,
      "grad_norm": 0.22203591465950012,
      "learning_rate": 1e-05,
      "loss": 0.9186,
      "step": 58095
    },
    {
      "epoch": 51.461470327723646,
      "grad_norm": 0.28027209639549255,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 58100
    },
    {
      "epoch": 51.46589902568645,
      "grad_norm": 0.2391355186700821,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 58105
    },
    {
      "epoch": 51.47032772364925,
      "grad_norm": 0.23409736156463623,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 58110
    },
    {
      "epoch": 51.47475642161205,
      "grad_norm": 0.23817473649978638,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 58115
    },
    {
      "epoch": 51.479185119574844,
      "grad_norm": 0.21340115368366241,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 58120
    },
    {
      "epoch": 51.483613817537645,
      "grad_norm": 0.21516743302345276,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 58125
    },
    {
      "epoch": 51.488042515500446,
      "grad_norm": 0.2183932363986969,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 58130
    },
    {
      "epoch": 51.49247121346324,
      "grad_norm": 0.2642355263233185,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 58135
    },
    {
      "epoch": 51.49689991142604,
      "grad_norm": 0.243107870221138,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 58140
    },
    {
      "epoch": 51.50132860938884,
      "grad_norm": 0.2183106541633606,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 58145
    },
    {
      "epoch": 51.50575730735164,
      "grad_norm": 0.22349101305007935,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 58150
    },
    {
      "epoch": 51.51018600531444,
      "grad_norm": 0.22907379269599915,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 58155
    },
    {
      "epoch": 51.51461470327724,
      "grad_norm": 0.24334098398685455,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 58160
    },
    {
      "epoch": 51.51904340124003,
      "grad_norm": 0.24807412922382355,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 58165
    },
    {
      "epoch": 51.523472099202834,
      "grad_norm": 0.21496319770812988,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 58170
    },
    {
      "epoch": 51.527900797165636,
      "grad_norm": 0.27487337589263916,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 58175
    },
    {
      "epoch": 51.53232949512843,
      "grad_norm": 0.24121910333633423,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 58180
    },
    {
      "epoch": 51.53675819309123,
      "grad_norm": 0.19430440664291382,
      "learning_rate": 1e-05,
      "loss": 1.0438,
      "step": 58185
    },
    {
      "epoch": 51.54118689105403,
      "grad_norm": 0.22432756423950195,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 58190
    },
    {
      "epoch": 51.545615589016826,
      "grad_norm": 0.25665029883384705,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 58195
    },
    {
      "epoch": 51.55004428697963,
      "grad_norm": 0.24494412541389465,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 58200
    },
    {
      "epoch": 51.55447298494243,
      "grad_norm": 0.20735104382038116,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 58205
    },
    {
      "epoch": 51.55890168290522,
      "grad_norm": 0.23223921656608582,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 58210
    },
    {
      "epoch": 51.563330380868024,
      "grad_norm": 0.2448100596666336,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 58215
    },
    {
      "epoch": 51.567759078830825,
      "grad_norm": 0.28599679470062256,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 58220
    },
    {
      "epoch": 51.57218777679362,
      "grad_norm": 0.2101174145936966,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 58225
    },
    {
      "epoch": 51.57661647475642,
      "grad_norm": 0.23095417022705078,
      "learning_rate": 1e-05,
      "loss": 0.9208,
      "step": 58230
    },
    {
      "epoch": 51.58104517271922,
      "grad_norm": 0.22340860962867737,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 58235
    },
    {
      "epoch": 51.58547387068202,
      "grad_norm": 0.22469645738601685,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 58240
    },
    {
      "epoch": 51.58990256864482,
      "grad_norm": 0.21839606761932373,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 58245
    },
    {
      "epoch": 51.59433126660762,
      "grad_norm": 0.3325742483139038,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 58250
    },
    {
      "epoch": 51.59875996457042,
      "grad_norm": 0.2494598925113678,
      "learning_rate": 1e-05,
      "loss": 1.0228,
      "step": 58255
    },
    {
      "epoch": 51.60318866253321,
      "grad_norm": 0.2741064429283142,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 58260
    },
    {
      "epoch": 51.607617360496015,
      "grad_norm": 0.2602717876434326,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 58265
    },
    {
      "epoch": 51.612046058458816,
      "grad_norm": 0.3043157756328583,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 58270
    },
    {
      "epoch": 51.61647475642161,
      "grad_norm": 0.30509912967681885,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 58275
    },
    {
      "epoch": 51.62090345438441,
      "grad_norm": 0.28390470147132874,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 58280
    },
    {
      "epoch": 51.62533215234721,
      "grad_norm": 0.22268792986869812,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 58285
    },
    {
      "epoch": 51.62976085031001,
      "grad_norm": 0.23918232321739197,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 58290
    },
    {
      "epoch": 51.63418954827281,
      "grad_norm": 0.22927765548229218,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 58295
    },
    {
      "epoch": 51.63861824623561,
      "grad_norm": 0.2979111671447754,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 58300
    },
    {
      "epoch": 51.6430469441984,
      "grad_norm": 0.2556305527687073,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 58305
    },
    {
      "epoch": 51.647475642161204,
      "grad_norm": 0.23678626120090485,
      "learning_rate": 1e-05,
      "loss": 1.0268,
      "step": 58310
    },
    {
      "epoch": 51.651904340124005,
      "grad_norm": 0.2691119909286499,
      "learning_rate": 1e-05,
      "loss": 0.9077,
      "step": 58315
    },
    {
      "epoch": 51.6563330380868,
      "grad_norm": 0.2591462731361389,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 58320
    },
    {
      "epoch": 51.6607617360496,
      "grad_norm": 0.2804064154624939,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 58325
    },
    {
      "epoch": 51.6651904340124,
      "grad_norm": 0.24454502761363983,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 58330
    },
    {
      "epoch": 51.669619131975196,
      "grad_norm": 0.24861130118370056,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 58335
    },
    {
      "epoch": 51.674047829938,
      "grad_norm": 0.26016995310783386,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 58340
    },
    {
      "epoch": 51.6784765279008,
      "grad_norm": 0.2355981022119522,
      "learning_rate": 1e-05,
      "loss": 0.8913,
      "step": 58345
    },
    {
      "epoch": 51.68290522586359,
      "grad_norm": 0.24048180878162384,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 58350
    },
    {
      "epoch": 51.687333923826394,
      "grad_norm": 0.22731292247772217,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 58355
    },
    {
      "epoch": 51.691762621789195,
      "grad_norm": 0.2594451606273651,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 58360
    },
    {
      "epoch": 51.696191319751996,
      "grad_norm": 0.21449124813079834,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 58365
    },
    {
      "epoch": 51.70062001771479,
      "grad_norm": 0.24885571002960205,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 58370
    },
    {
      "epoch": 51.70504871567759,
      "grad_norm": 0.24724626541137695,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 58375
    },
    {
      "epoch": 51.70947741364039,
      "grad_norm": 0.2644142210483551,
      "learning_rate": 1e-05,
      "loss": 1.0189,
      "step": 58380
    },
    {
      "epoch": 51.71390611160319,
      "grad_norm": 0.21985891461372375,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 58385
    },
    {
      "epoch": 51.71833480956599,
      "grad_norm": 0.2015070915222168,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 58390
    },
    {
      "epoch": 51.72276350752879,
      "grad_norm": 0.2373541295528412,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 58395
    },
    {
      "epoch": 51.72719220549158,
      "grad_norm": 0.20183350145816803,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 58400
    },
    {
      "epoch": 51.731620903454385,
      "grad_norm": 0.21155519783496857,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 58405
    },
    {
      "epoch": 51.736049601417186,
      "grad_norm": 0.26506441831588745,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 58410
    },
    {
      "epoch": 51.74047829937998,
      "grad_norm": 0.24553550779819489,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 58415
    },
    {
      "epoch": 51.74490699734278,
      "grad_norm": 0.20289985835552216,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 58420
    },
    {
      "epoch": 51.74933569530558,
      "grad_norm": 0.19446690380573273,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 58425
    },
    {
      "epoch": 51.753764393268376,
      "grad_norm": 0.26361045241355896,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 58430
    },
    {
      "epoch": 51.75819309123118,
      "grad_norm": 0.28387776017189026,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 58435
    },
    {
      "epoch": 51.76262178919398,
      "grad_norm": 0.2082238644361496,
      "learning_rate": 1e-05,
      "loss": 0.9242,
      "step": 58440
    },
    {
      "epoch": 51.76705048715677,
      "grad_norm": 0.22151020169258118,
      "learning_rate": 1e-05,
      "loss": 0.9247,
      "step": 58445
    },
    {
      "epoch": 51.771479185119574,
      "grad_norm": 0.21319808065891266,
      "learning_rate": 1e-05,
      "loss": 0.9265,
      "step": 58450
    },
    {
      "epoch": 51.775907883082375,
      "grad_norm": 0.1932530701160431,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 58455
    },
    {
      "epoch": 51.78033658104517,
      "grad_norm": 0.22282566130161285,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 58460
    },
    {
      "epoch": 51.78476527900797,
      "grad_norm": 0.2406737357378006,
      "learning_rate": 1e-05,
      "loss": 0.9211,
      "step": 58465
    },
    {
      "epoch": 51.78919397697077,
      "grad_norm": 0.255719393491745,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 58470
    },
    {
      "epoch": 51.79362267493357,
      "grad_norm": 0.2293304055929184,
      "learning_rate": 1e-05,
      "loss": 1.0174,
      "step": 58475
    },
    {
      "epoch": 51.79805137289637,
      "grad_norm": 0.2705524265766144,
      "learning_rate": 1e-05,
      "loss": 1.0306,
      "step": 58480
    },
    {
      "epoch": 51.80248007085917,
      "grad_norm": 0.20759350061416626,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 58485
    },
    {
      "epoch": 51.80690876882197,
      "grad_norm": 0.22513075172901154,
      "learning_rate": 1e-05,
      "loss": 1.0285,
      "step": 58490
    },
    {
      "epoch": 51.811337466784764,
      "grad_norm": 0.2505941390991211,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 58495
    },
    {
      "epoch": 51.815766164747565,
      "grad_norm": 0.2880514860153198,
      "learning_rate": 1e-05,
      "loss": 0.9217,
      "step": 58500
    },
    {
      "epoch": 51.820194862710366,
      "grad_norm": 0.2535289525985718,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 58505
    },
    {
      "epoch": 51.82462356067316,
      "grad_norm": 0.2720993757247925,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 58510
    },
    {
      "epoch": 51.82905225863596,
      "grad_norm": 0.23074521124362946,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 58515
    },
    {
      "epoch": 51.83348095659876,
      "grad_norm": 0.24251681566238403,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 58520
    },
    {
      "epoch": 51.83790965456156,
      "grad_norm": 0.24032023549079895,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 58525
    },
    {
      "epoch": 51.84233835252436,
      "grad_norm": 0.22962750494480133,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 58530
    },
    {
      "epoch": 51.84676705048716,
      "grad_norm": 0.2727820575237274,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 58535
    },
    {
      "epoch": 51.85119574844995,
      "grad_norm": 0.27132317423820496,
      "learning_rate": 1e-05,
      "loss": 0.8938,
      "step": 58540
    },
    {
      "epoch": 51.855624446412754,
      "grad_norm": 0.27526429295539856,
      "learning_rate": 1e-05,
      "loss": 0.9191,
      "step": 58545
    },
    {
      "epoch": 51.860053144375556,
      "grad_norm": 0.27985382080078125,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 58550
    },
    {
      "epoch": 51.86448184233835,
      "grad_norm": 0.2495328187942505,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 58555
    },
    {
      "epoch": 51.86891054030115,
      "grad_norm": 0.24981878697872162,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 58560
    },
    {
      "epoch": 51.87333923826395,
      "grad_norm": 0.2278611958026886,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 58565
    },
    {
      "epoch": 51.877767936226746,
      "grad_norm": 0.2569010853767395,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 58570
    },
    {
      "epoch": 51.88219663418955,
      "grad_norm": 0.22449199855327606,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 58575
    },
    {
      "epoch": 51.88662533215235,
      "grad_norm": 0.27370843291282654,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 58580
    },
    {
      "epoch": 51.89105403011514,
      "grad_norm": 0.24513772130012512,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 58585
    },
    {
      "epoch": 51.895482728077944,
      "grad_norm": 0.2431306540966034,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 58590
    },
    {
      "epoch": 51.899911426040745,
      "grad_norm": 0.20609644055366516,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 58595
    },
    {
      "epoch": 51.90434012400354,
      "grad_norm": 0.24189017713069916,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 58600
    },
    {
      "epoch": 51.90876882196634,
      "grad_norm": 0.29227396845817566,
      "learning_rate": 1e-05,
      "loss": 0.888,
      "step": 58605
    },
    {
      "epoch": 51.91319751992914,
      "grad_norm": 0.31686240434646606,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 58610
    },
    {
      "epoch": 51.91762621789194,
      "grad_norm": 0.228986456990242,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 58615
    },
    {
      "epoch": 51.92205491585474,
      "grad_norm": 0.24290288984775543,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 58620
    },
    {
      "epoch": 51.92648361381754,
      "grad_norm": 0.20453818142414093,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 58625
    },
    {
      "epoch": 51.93091231178034,
      "grad_norm": 0.20300108194351196,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 58630
    },
    {
      "epoch": 51.935341009743134,
      "grad_norm": 0.2920350134372711,
      "learning_rate": 1e-05,
      "loss": 1.042,
      "step": 58635
    },
    {
      "epoch": 51.939769707705935,
      "grad_norm": 0.2604334354400635,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 58640
    },
    {
      "epoch": 51.944198405668736,
      "grad_norm": 0.20409448444843292,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 58645
    },
    {
      "epoch": 51.94862710363153,
      "grad_norm": 0.24128533899784088,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 58650
    },
    {
      "epoch": 51.95305580159433,
      "grad_norm": 0.2413007766008377,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 58655
    },
    {
      "epoch": 51.95748449955713,
      "grad_norm": 0.25519412755966187,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 58660
    },
    {
      "epoch": 51.96191319751993,
      "grad_norm": 0.2560592293739319,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 58665
    },
    {
      "epoch": 51.96634189548273,
      "grad_norm": 0.2350081354379654,
      "learning_rate": 1e-05,
      "loss": 0.9251,
      "step": 58670
    },
    {
      "epoch": 51.97077059344553,
      "grad_norm": 0.2191975861787796,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 58675
    },
    {
      "epoch": 51.97519929140832,
      "grad_norm": 0.260464608669281,
      "learning_rate": 1e-05,
      "loss": 0.9143,
      "step": 58680
    },
    {
      "epoch": 51.979627989371124,
      "grad_norm": 0.24714504182338715,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 58685
    },
    {
      "epoch": 51.984056687333926,
      "grad_norm": 0.26472267508506775,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 58690
    },
    {
      "epoch": 51.98848538529672,
      "grad_norm": 0.21740540862083435,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 58695
    },
    {
      "epoch": 51.99291408325952,
      "grad_norm": 0.23431916534900665,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 58700
    },
    {
      "epoch": 51.99734278122232,
      "grad_norm": 0.26568934321403503,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 58705
    },
    {
      "epoch": 52.001771479185116,
      "grad_norm": 0.24277284741401672,
      "learning_rate": 1e-05,
      "loss": 1.0266,
      "step": 58710
    },
    {
      "epoch": 52.00620017714792,
      "grad_norm": 0.19721931219100952,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 58715
    },
    {
      "epoch": 52.01062887511072,
      "grad_norm": 0.2184523493051529,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 58720
    },
    {
      "epoch": 52.01505757307352,
      "grad_norm": 0.2964482009410858,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 58725
    },
    {
      "epoch": 52.019486271036314,
      "grad_norm": 0.23082707822322845,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 58730
    },
    {
      "epoch": 52.023914968999115,
      "grad_norm": 0.20773455500602722,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 58735
    },
    {
      "epoch": 52.028343666961916,
      "grad_norm": 0.20699819922447205,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 58740
    },
    {
      "epoch": 52.03277236492471,
      "grad_norm": 0.23711200058460236,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 58745
    },
    {
      "epoch": 52.03720106288751,
      "grad_norm": 0.2855316698551178,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 58750
    },
    {
      "epoch": 52.04162976085031,
      "grad_norm": 0.2499338984489441,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 58755
    },
    {
      "epoch": 52.04605845881311,
      "grad_norm": 0.23982737958431244,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 58760
    },
    {
      "epoch": 52.05048715677591,
      "grad_norm": 0.2482318878173828,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 58765
    },
    {
      "epoch": 52.05491585473871,
      "grad_norm": 0.2540883719921112,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 58770
    },
    {
      "epoch": 52.0593445527015,
      "grad_norm": 0.26081958413124084,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 58775
    },
    {
      "epoch": 52.063773250664305,
      "grad_norm": 0.2717559337615967,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 58780
    },
    {
      "epoch": 52.068201948627106,
      "grad_norm": 0.2518419027328491,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 58785
    },
    {
      "epoch": 52.0726306465899,
      "grad_norm": 0.2804437577724457,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 58790
    },
    {
      "epoch": 52.0770593445527,
      "grad_norm": 0.2536376416683197,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 58795
    },
    {
      "epoch": 52.0814880425155,
      "grad_norm": 0.25808805227279663,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 58800
    },
    {
      "epoch": 52.0859167404783,
      "grad_norm": 0.24577517807483673,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 58805
    },
    {
      "epoch": 52.0903454384411,
      "grad_norm": 0.2653447091579437,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 58810
    },
    {
      "epoch": 52.0947741364039,
      "grad_norm": 0.2296014428138733,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 58815
    },
    {
      "epoch": 52.09920283436669,
      "grad_norm": 0.24378883838653564,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 58820
    },
    {
      "epoch": 52.103631532329494,
      "grad_norm": 0.23796670138835907,
      "learning_rate": 1e-05,
      "loss": 0.9207,
      "step": 58825
    },
    {
      "epoch": 52.108060230292296,
      "grad_norm": 0.24171461164951324,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 58830
    },
    {
      "epoch": 52.11248892825509,
      "grad_norm": 0.24397942423820496,
      "learning_rate": 1e-05,
      "loss": 0.9313,
      "step": 58835
    },
    {
      "epoch": 52.11691762621789,
      "grad_norm": 0.2165588140487671,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 58840
    },
    {
      "epoch": 52.12134632418069,
      "grad_norm": 0.2648729979991913,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 58845
    },
    {
      "epoch": 52.12577502214349,
      "grad_norm": 0.2547149658203125,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 58850
    },
    {
      "epoch": 52.13020372010629,
      "grad_norm": 0.22421130537986755,
      "learning_rate": 1e-05,
      "loss": 0.9137,
      "step": 58855
    },
    {
      "epoch": 52.13463241806909,
      "grad_norm": 0.2462160885334015,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 58860
    },
    {
      "epoch": 52.13906111603189,
      "grad_norm": 0.2373134046792984,
      "learning_rate": 1e-05,
      "loss": 0.9427,
      "step": 58865
    },
    {
      "epoch": 52.143489813994684,
      "grad_norm": 0.22970210015773773,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 58870
    },
    {
      "epoch": 52.147918511957485,
      "grad_norm": 0.2565675377845764,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 58875
    },
    {
      "epoch": 52.152347209920286,
      "grad_norm": 0.2633257806301117,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 58880
    },
    {
      "epoch": 52.15677590788308,
      "grad_norm": 0.22295983135700226,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 58885
    },
    {
      "epoch": 52.16120460584588,
      "grad_norm": 0.22809961438179016,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 58890
    },
    {
      "epoch": 52.16563330380868,
      "grad_norm": 0.20048491656780243,
      "learning_rate": 1e-05,
      "loss": 0.9307,
      "step": 58895
    },
    {
      "epoch": 52.17006200177148,
      "grad_norm": 0.23810194432735443,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 58900
    },
    {
      "epoch": 52.17449069973428,
      "grad_norm": 0.2021421492099762,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 58905
    },
    {
      "epoch": 52.17891939769708,
      "grad_norm": 0.2306343913078308,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 58910
    },
    {
      "epoch": 52.18334809565987,
      "grad_norm": 0.21147963404655457,
      "learning_rate": 1e-05,
      "loss": 1.0458,
      "step": 58915
    },
    {
      "epoch": 52.187776793622675,
      "grad_norm": 0.2717270851135254,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 58920
    },
    {
      "epoch": 52.192205491585476,
      "grad_norm": 0.27564477920532227,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 58925
    },
    {
      "epoch": 52.19663418954827,
      "grad_norm": 0.2741462290287018,
      "learning_rate": 1e-05,
      "loss": 1.0386,
      "step": 58930
    },
    {
      "epoch": 52.20106288751107,
      "grad_norm": 0.23094050586223602,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 58935
    },
    {
      "epoch": 52.20549158547387,
      "grad_norm": 0.21067003905773163,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 58940
    },
    {
      "epoch": 52.20992028343667,
      "grad_norm": 0.2271854728460312,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 58945
    },
    {
      "epoch": 52.21434898139947,
      "grad_norm": 0.23402239382266998,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 58950
    },
    {
      "epoch": 52.21877767936227,
      "grad_norm": 0.2510242760181427,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 58955
    },
    {
      "epoch": 52.22320637732506,
      "grad_norm": 0.2374059110879898,
      "learning_rate": 1e-05,
      "loss": 0.9227,
      "step": 58960
    },
    {
      "epoch": 52.227635075287864,
      "grad_norm": 0.2022409588098526,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 58965
    },
    {
      "epoch": 52.232063773250665,
      "grad_norm": 0.2773072421550751,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 58970
    },
    {
      "epoch": 52.23649247121347,
      "grad_norm": 0.27207493782043457,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 58975
    },
    {
      "epoch": 52.24092116917626,
      "grad_norm": 0.21166494488716125,
      "learning_rate": 1e-05,
      "loss": 0.924,
      "step": 58980
    },
    {
      "epoch": 52.24534986713906,
      "grad_norm": 0.24492958188056946,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 58985
    },
    {
      "epoch": 52.24977856510186,
      "grad_norm": 0.22873732447624207,
      "learning_rate": 1e-05,
      "loss": 0.9229,
      "step": 58990
    },
    {
      "epoch": 52.25420726306466,
      "grad_norm": 0.21485452353954315,
      "learning_rate": 1e-05,
      "loss": 0.9249,
      "step": 58995
    },
    {
      "epoch": 52.25863596102746,
      "grad_norm": 0.2711961269378662,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 59000
    },
    {
      "epoch": 52.26306465899026,
      "grad_norm": 0.21365612745285034,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 59005
    },
    {
      "epoch": 52.267493356953054,
      "grad_norm": 0.23931793868541718,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 59010
    },
    {
      "epoch": 52.271922054915855,
      "grad_norm": 0.24862323701381683,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 59015
    },
    {
      "epoch": 52.276350752878656,
      "grad_norm": 0.261353999376297,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 59020
    },
    {
      "epoch": 52.28077945084145,
      "grad_norm": 0.2403019219636917,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 59025
    },
    {
      "epoch": 52.28520814880425,
      "grad_norm": 0.26119378209114075,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 59030
    },
    {
      "epoch": 52.28963684676705,
      "grad_norm": 0.2288169264793396,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 59035
    },
    {
      "epoch": 52.29406554472985,
      "grad_norm": 0.25187814235687256,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 59040
    },
    {
      "epoch": 52.29849424269265,
      "grad_norm": 0.21173524856567383,
      "learning_rate": 1e-05,
      "loss": 0.8987,
      "step": 59045
    },
    {
      "epoch": 52.30292294065545,
      "grad_norm": 0.22896890342235565,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 59050
    },
    {
      "epoch": 52.30735163861824,
      "grad_norm": 0.2732599079608917,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 59055
    },
    {
      "epoch": 52.311780336581045,
      "grad_norm": 0.21000850200653076,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 59060
    },
    {
      "epoch": 52.316209034543846,
      "grad_norm": 0.2537667751312256,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 59065
    },
    {
      "epoch": 52.32063773250664,
      "grad_norm": 0.22175449132919312,
      "learning_rate": 1e-05,
      "loss": 1.0291,
      "step": 59070
    },
    {
      "epoch": 52.32506643046944,
      "grad_norm": 0.22910772264003754,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 59075
    },
    {
      "epoch": 52.32949512843224,
      "grad_norm": 0.24023257195949554,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 59080
    },
    {
      "epoch": 52.333923826395036,
      "grad_norm": 0.23657000064849854,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 59085
    },
    {
      "epoch": 52.33835252435784,
      "grad_norm": 0.27525779604911804,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 59090
    },
    {
      "epoch": 52.34278122232064,
      "grad_norm": 0.24265964329242706,
      "learning_rate": 1e-05,
      "loss": 1.0475,
      "step": 59095
    },
    {
      "epoch": 52.34720992028344,
      "grad_norm": 0.2675337493419647,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 59100
    },
    {
      "epoch": 52.351638618246234,
      "grad_norm": 0.2790255546569824,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 59105
    },
    {
      "epoch": 52.356067316209035,
      "grad_norm": 0.3175674080848694,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 59110
    },
    {
      "epoch": 52.36049601417184,
      "grad_norm": 0.25975850224494934,
      "learning_rate": 1e-05,
      "loss": 0.9039,
      "step": 59115
    },
    {
      "epoch": 52.36492471213463,
      "grad_norm": 0.26649001240730286,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 59120
    },
    {
      "epoch": 52.36935341009743,
      "grad_norm": 0.2806892991065979,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 59125
    },
    {
      "epoch": 52.37378210806023,
      "grad_norm": 0.26368820667266846,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 59130
    },
    {
      "epoch": 52.37821080602303,
      "grad_norm": 0.23274797201156616,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 59135
    },
    {
      "epoch": 52.38263950398583,
      "grad_norm": 0.22270150482654572,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 59140
    },
    {
      "epoch": 52.38706820194863,
      "grad_norm": 0.28753140568733215,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 59145
    },
    {
      "epoch": 52.391496899911424,
      "grad_norm": 0.2770533561706543,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 59150
    },
    {
      "epoch": 52.395925597874225,
      "grad_norm": 0.26124048233032227,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 59155
    },
    {
      "epoch": 52.400354295837026,
      "grad_norm": 0.23832140862941742,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 59160
    },
    {
      "epoch": 52.40478299379982,
      "grad_norm": 0.2152429223060608,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 59165
    },
    {
      "epoch": 52.40921169176262,
      "grad_norm": 0.2796470820903778,
      "learning_rate": 1e-05,
      "loss": 0.93,
      "step": 59170
    },
    {
      "epoch": 52.41364038972542,
      "grad_norm": 0.22918136417865753,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 59175
    },
    {
      "epoch": 52.41806908768822,
      "grad_norm": 0.2409706562757492,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 59180
    },
    {
      "epoch": 52.42249778565102,
      "grad_norm": 0.21738368272781372,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 59185
    },
    {
      "epoch": 52.42692648361382,
      "grad_norm": 0.239488422870636,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 59190
    },
    {
      "epoch": 52.43135518157661,
      "grad_norm": 0.2247210144996643,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 59195
    },
    {
      "epoch": 52.435783879539414,
      "grad_norm": 0.22405464947223663,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 59200
    },
    {
      "epoch": 52.440212577502216,
      "grad_norm": 0.2425481379032135,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 59205
    },
    {
      "epoch": 52.44464127546502,
      "grad_norm": 0.23909392952919006,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 59210
    },
    {
      "epoch": 52.44906997342781,
      "grad_norm": 0.23658297955989838,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 59215
    },
    {
      "epoch": 52.45349867139061,
      "grad_norm": 0.23252840340137482,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 59220
    },
    {
      "epoch": 52.45792736935341,
      "grad_norm": 0.2774755358695984,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 59225
    },
    {
      "epoch": 52.46235606731621,
      "grad_norm": 0.2380921095609665,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 59230
    },
    {
      "epoch": 52.46678476527901,
      "grad_norm": 0.25685325264930725,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 59235
    },
    {
      "epoch": 52.47121346324181,
      "grad_norm": 0.2130991518497467,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 59240
    },
    {
      "epoch": 52.475642161204604,
      "grad_norm": 0.24722576141357422,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 59245
    },
    {
      "epoch": 52.480070859167405,
      "grad_norm": 0.21808280050754547,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 59250
    },
    {
      "epoch": 52.484499557130206,
      "grad_norm": 0.21170203387737274,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 59255
    },
    {
      "epoch": 52.488928255093,
      "grad_norm": 0.32557445764541626,
      "learning_rate": 1e-05,
      "loss": 1.0337,
      "step": 59260
    },
    {
      "epoch": 52.4933569530558,
      "grad_norm": 0.24916352331638336,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 59265
    },
    {
      "epoch": 52.4977856510186,
      "grad_norm": 0.2145821452140808,
      "learning_rate": 1e-05,
      "loss": 0.9056,
      "step": 59270
    },
    {
      "epoch": 52.5022143489814,
      "grad_norm": 0.22541280090808868,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 59275
    },
    {
      "epoch": 52.5066430469442,
      "grad_norm": 0.2691725492477417,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 59280
    },
    {
      "epoch": 52.511071744907,
      "grad_norm": 0.2309570461511612,
      "learning_rate": 1e-05,
      "loss": 0.9231,
      "step": 59285
    },
    {
      "epoch": 52.515500442869794,
      "grad_norm": 0.2116842120885849,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 59290
    },
    {
      "epoch": 52.519929140832595,
      "grad_norm": 0.2270590364933014,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 59295
    },
    {
      "epoch": 52.524357838795396,
      "grad_norm": 0.22602643072605133,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 59300
    },
    {
      "epoch": 52.52878653675819,
      "grad_norm": 0.23234984278678894,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 59305
    },
    {
      "epoch": 52.53321523472099,
      "grad_norm": 0.27178335189819336,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 59310
    },
    {
      "epoch": 52.53764393268379,
      "grad_norm": 0.20679497718811035,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 59315
    },
    {
      "epoch": 52.54207263064659,
      "grad_norm": 0.21883054077625275,
      "learning_rate": 1e-05,
      "loss": 0.9254,
      "step": 59320
    },
    {
      "epoch": 52.54650132860939,
      "grad_norm": 0.25714194774627686,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 59325
    },
    {
      "epoch": 52.55093002657219,
      "grad_norm": 0.25423967838287354,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 59330
    },
    {
      "epoch": 52.55535872453498,
      "grad_norm": 0.25764816999435425,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 59335
    },
    {
      "epoch": 52.559787422497784,
      "grad_norm": 0.28911277651786804,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 59340
    },
    {
      "epoch": 52.564216120460586,
      "grad_norm": 0.3020361661911011,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 59345
    },
    {
      "epoch": 52.56864481842339,
      "grad_norm": 0.19778360426425934,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 59350
    },
    {
      "epoch": 52.57307351638618,
      "grad_norm": 0.21709156036376953,
      "learning_rate": 1e-05,
      "loss": 1.024,
      "step": 59355
    },
    {
      "epoch": 52.57750221434898,
      "grad_norm": 0.25256097316741943,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 59360
    },
    {
      "epoch": 52.58193091231178,
      "grad_norm": 0.24452269077301025,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 59365
    },
    {
      "epoch": 52.58635961027458,
      "grad_norm": 0.3096843361854553,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 59370
    },
    {
      "epoch": 52.59078830823738,
      "grad_norm": 0.2754860520362854,
      "learning_rate": 1e-05,
      "loss": 0.9325,
      "step": 59375
    },
    {
      "epoch": 52.59521700620018,
      "grad_norm": 0.22189363837242126,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 59380
    },
    {
      "epoch": 52.599645704162974,
      "grad_norm": 0.2509881854057312,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 59385
    },
    {
      "epoch": 52.604074402125775,
      "grad_norm": 0.2636805474758148,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 59390
    },
    {
      "epoch": 52.608503100088576,
      "grad_norm": 0.2145776003599167,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 59395
    },
    {
      "epoch": 52.61293179805137,
      "grad_norm": 0.25512391328811646,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 59400
    },
    {
      "epoch": 52.61736049601417,
      "grad_norm": 0.24466027319431305,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 59405
    },
    {
      "epoch": 52.62178919397697,
      "grad_norm": 0.2131464183330536,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 59410
    },
    {
      "epoch": 52.62621789193977,
      "grad_norm": 0.2435818314552307,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 59415
    },
    {
      "epoch": 52.63064658990257,
      "grad_norm": 0.22133409976959229,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 59420
    },
    {
      "epoch": 52.63507528786537,
      "grad_norm": 0.2613498568534851,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 59425
    },
    {
      "epoch": 52.63950398582816,
      "grad_norm": 0.2328488826751709,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 59430
    },
    {
      "epoch": 52.643932683790965,
      "grad_norm": 0.26889708638191223,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 59435
    },
    {
      "epoch": 52.648361381753766,
      "grad_norm": 0.2561200261116028,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 59440
    },
    {
      "epoch": 52.65279007971656,
      "grad_norm": 0.23625174164772034,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 59445
    },
    {
      "epoch": 52.65721877767936,
      "grad_norm": 0.24090686440467834,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 59450
    },
    {
      "epoch": 52.66164747564216,
      "grad_norm": 0.23700059950351715,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 59455
    },
    {
      "epoch": 52.666076173604964,
      "grad_norm": 0.24252142012119293,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 59460
    },
    {
      "epoch": 52.67050487156776,
      "grad_norm": 0.246984601020813,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 59465
    },
    {
      "epoch": 52.67493356953056,
      "grad_norm": 0.21925924718379974,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 59470
    },
    {
      "epoch": 52.67936226749336,
      "grad_norm": 0.25783872604370117,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 59475
    },
    {
      "epoch": 52.683790965456154,
      "grad_norm": 0.2226601541042328,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 59480
    },
    {
      "epoch": 52.688219663418955,
      "grad_norm": 0.27122756838798523,
      "learning_rate": 1e-05,
      "loss": 0.917,
      "step": 59485
    },
    {
      "epoch": 52.69264836138176,
      "grad_norm": 0.2551897168159485,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 59490
    },
    {
      "epoch": 52.69707705934455,
      "grad_norm": 0.23085863888263702,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 59495
    },
    {
      "epoch": 52.70150575730735,
      "grad_norm": 0.26000431180000305,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 59500
    },
    {
      "epoch": 52.70593445527015,
      "grad_norm": 0.2330954223871231,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 59505
    },
    {
      "epoch": 52.71036315323295,
      "grad_norm": 0.2411959320306778,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 59510
    },
    {
      "epoch": 52.71479185119575,
      "grad_norm": 0.24666187167167664,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 59515
    },
    {
      "epoch": 52.71922054915855,
      "grad_norm": 0.25113222002983093,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 59520
    },
    {
      "epoch": 52.723649247121344,
      "grad_norm": 0.2843528091907501,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 59525
    },
    {
      "epoch": 52.728077945084145,
      "grad_norm": 0.24040670692920685,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 59530
    },
    {
      "epoch": 52.732506643046946,
      "grad_norm": 0.241177499294281,
      "learning_rate": 1e-05,
      "loss": 0.9264,
      "step": 59535
    },
    {
      "epoch": 52.73693534100974,
      "grad_norm": 0.23298807442188263,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 59540
    },
    {
      "epoch": 52.74136403897254,
      "grad_norm": 0.23927631974220276,
      "learning_rate": 1e-05,
      "loss": 1.0273,
      "step": 59545
    },
    {
      "epoch": 52.74579273693534,
      "grad_norm": 0.3062893748283386,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 59550
    },
    {
      "epoch": 52.75022143489814,
      "grad_norm": 0.26091793179512024,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 59555
    },
    {
      "epoch": 52.75465013286094,
      "grad_norm": 0.25685369968414307,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 59560
    },
    {
      "epoch": 52.75907883082374,
      "grad_norm": 0.23045146465301514,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 59565
    },
    {
      "epoch": 52.76350752878653,
      "grad_norm": 0.2705872058868408,
      "learning_rate": 1e-05,
      "loss": 0.9149,
      "step": 59570
    },
    {
      "epoch": 52.767936226749335,
      "grad_norm": 0.2371685951948166,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 59575
    },
    {
      "epoch": 52.772364924712136,
      "grad_norm": 0.2210664302110672,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 59580
    },
    {
      "epoch": 52.77679362267494,
      "grad_norm": 0.21319898962974548,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 59585
    },
    {
      "epoch": 52.78122232063773,
      "grad_norm": 0.23021219670772552,
      "learning_rate": 1e-05,
      "loss": 0.918,
      "step": 59590
    },
    {
      "epoch": 52.78565101860053,
      "grad_norm": 0.2274142950773239,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 59595
    },
    {
      "epoch": 52.79007971656333,
      "grad_norm": 0.23048655688762665,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 59600
    },
    {
      "epoch": 52.79450841452613,
      "grad_norm": 0.23316003382205963,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 59605
    },
    {
      "epoch": 52.79893711248893,
      "grad_norm": 0.24283450841903687,
      "learning_rate": 1e-05,
      "loss": 0.9085,
      "step": 59610
    },
    {
      "epoch": 52.80336581045173,
      "grad_norm": 0.2414061725139618,
      "learning_rate": 1e-05,
      "loss": 0.8966,
      "step": 59615
    },
    {
      "epoch": 52.807794508414524,
      "grad_norm": 0.2697695791721344,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 59620
    },
    {
      "epoch": 52.812223206377325,
      "grad_norm": 0.29279714822769165,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 59625
    },
    {
      "epoch": 52.81665190434013,
      "grad_norm": 0.2392735630273819,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 59630
    },
    {
      "epoch": 52.82108060230292,
      "grad_norm": 0.28301066160202026,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 59635
    },
    {
      "epoch": 52.82550930026572,
      "grad_norm": 0.25010064244270325,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 59640
    },
    {
      "epoch": 52.82993799822852,
      "grad_norm": 0.23396393656730652,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 59645
    },
    {
      "epoch": 52.83436669619132,
      "grad_norm": 0.277478963136673,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 59650
    },
    {
      "epoch": 52.83879539415412,
      "grad_norm": 0.2712346613407135,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 59655
    },
    {
      "epoch": 52.84322409211692,
      "grad_norm": 0.21306876838207245,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 59660
    },
    {
      "epoch": 52.847652790079714,
      "grad_norm": 0.23523080348968506,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 59665
    },
    {
      "epoch": 52.852081488042515,
      "grad_norm": 0.22288550436496735,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 59670
    },
    {
      "epoch": 52.856510186005316,
      "grad_norm": 0.22876958549022675,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 59675
    },
    {
      "epoch": 52.86093888396811,
      "grad_norm": 0.24014483392238617,
      "learning_rate": 1e-05,
      "loss": 0.9236,
      "step": 59680
    },
    {
      "epoch": 52.86536758193091,
      "grad_norm": 0.2395474910736084,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 59685
    },
    {
      "epoch": 52.86979627989371,
      "grad_norm": 0.23993068933486938,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 59690
    },
    {
      "epoch": 52.87422497785651,
      "grad_norm": 0.20460699498653412,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 59695
    },
    {
      "epoch": 52.87865367581931,
      "grad_norm": 0.22421720623970032,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 59700
    },
    {
      "epoch": 52.88308237378211,
      "grad_norm": 0.21821144223213196,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 59705
    },
    {
      "epoch": 52.88751107174491,
      "grad_norm": 0.21350087225437164,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 59710
    },
    {
      "epoch": 52.891939769707704,
      "grad_norm": 0.22925147414207458,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 59715
    },
    {
      "epoch": 52.896368467670506,
      "grad_norm": 0.2710317373275757,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 59720
    },
    {
      "epoch": 52.90079716563331,
      "grad_norm": 0.23127196729183197,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 59725
    },
    {
      "epoch": 52.9052258635961,
      "grad_norm": 0.24701432883739471,
      "learning_rate": 1e-05,
      "loss": 0.9311,
      "step": 59730
    },
    {
      "epoch": 52.9096545615589,
      "grad_norm": 0.24243749678134918,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 59735
    },
    {
      "epoch": 52.9140832595217,
      "grad_norm": 0.2752993106842041,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 59740
    },
    {
      "epoch": 52.9185119574845,
      "grad_norm": 0.269914448261261,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 59745
    },
    {
      "epoch": 52.9229406554473,
      "grad_norm": 0.22686247527599335,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 59750
    },
    {
      "epoch": 52.9273693534101,
      "grad_norm": 0.2274935394525528,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 59755
    },
    {
      "epoch": 52.931798051372894,
      "grad_norm": 0.2664961516857147,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 59760
    },
    {
      "epoch": 52.936226749335695,
      "grad_norm": 0.24603243172168732,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 59765
    },
    {
      "epoch": 52.9406554472985,
      "grad_norm": 0.2533240020275116,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 59770
    },
    {
      "epoch": 52.94508414526129,
      "grad_norm": 0.2096920609474182,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 59775
    },
    {
      "epoch": 52.94951284322409,
      "grad_norm": 0.21883811056613922,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 59780
    },
    {
      "epoch": 52.95394154118689,
      "grad_norm": 0.23622551560401917,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 59785
    },
    {
      "epoch": 52.95837023914969,
      "grad_norm": 0.2807080149650574,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 59790
    },
    {
      "epoch": 52.96279893711249,
      "grad_norm": 0.22725415229797363,
      "learning_rate": 1e-05,
      "loss": 0.9206,
      "step": 59795
    },
    {
      "epoch": 52.96722763507529,
      "grad_norm": 0.2504546046257019,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 59800
    },
    {
      "epoch": 52.971656333038084,
      "grad_norm": 0.2076190710067749,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 59805
    },
    {
      "epoch": 52.976085031000885,
      "grad_norm": 0.24882617592811584,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 59810
    },
    {
      "epoch": 52.980513728963686,
      "grad_norm": 0.24668893218040466,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 59815
    },
    {
      "epoch": 52.98494242692648,
      "grad_norm": 0.23608694970607758,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 59820
    },
    {
      "epoch": 52.98937112488928,
      "grad_norm": 0.23497037589550018,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 59825
    },
    {
      "epoch": 52.99379982285208,
      "grad_norm": 0.2623255252838135,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 59830
    },
    {
      "epoch": 52.998228520814884,
      "grad_norm": 0.2677062749862671,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 59835
    },
    {
      "epoch": 53.00265721877768,
      "grad_norm": 0.2218473106622696,
      "learning_rate": 1e-05,
      "loss": 0.9252,
      "step": 59840
    },
    {
      "epoch": 53.00708591674048,
      "grad_norm": 0.22999630868434906,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 59845
    },
    {
      "epoch": 53.01151461470328,
      "grad_norm": 0.22826425731182098,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 59850
    },
    {
      "epoch": 53.015943312666074,
      "grad_norm": 0.2073640376329422,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 59855
    },
    {
      "epoch": 53.020372010628876,
      "grad_norm": 0.25264009833335876,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 59860
    },
    {
      "epoch": 53.02480070859168,
      "grad_norm": 0.2786754369735718,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 59865
    },
    {
      "epoch": 53.02922940655447,
      "grad_norm": 0.22319111227989197,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 59870
    },
    {
      "epoch": 53.03365810451727,
      "grad_norm": 0.2614092528820038,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 59875
    },
    {
      "epoch": 53.03808680248007,
      "grad_norm": 0.25744086503982544,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 59880
    },
    {
      "epoch": 53.04251550044287,
      "grad_norm": 0.23858335614204407,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 59885
    },
    {
      "epoch": 53.04694419840567,
      "grad_norm": 0.22827693819999695,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 59890
    },
    {
      "epoch": 53.05137289636847,
      "grad_norm": 0.24994118511676788,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 59895
    },
    {
      "epoch": 53.055801594331264,
      "grad_norm": 0.21131977438926697,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 59900
    },
    {
      "epoch": 53.060230292294065,
      "grad_norm": 0.24249157309532166,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 59905
    },
    {
      "epoch": 53.064658990256866,
      "grad_norm": 0.24080723524093628,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 59910
    },
    {
      "epoch": 53.06908768821966,
      "grad_norm": 0.21769703924655914,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 59915
    },
    {
      "epoch": 53.07351638618246,
      "grad_norm": 0.26333436369895935,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 59920
    },
    {
      "epoch": 53.07794508414526,
      "grad_norm": 0.229929119348526,
      "learning_rate": 1e-05,
      "loss": 0.9253,
      "step": 59925
    },
    {
      "epoch": 53.08237378210806,
      "grad_norm": 0.2134549915790558,
      "learning_rate": 1e-05,
      "loss": 1.0387,
      "step": 59930
    },
    {
      "epoch": 53.08680248007086,
      "grad_norm": 0.22480005025863647,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 59935
    },
    {
      "epoch": 53.09123117803366,
      "grad_norm": 0.19069701433181763,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 59940
    },
    {
      "epoch": 53.09565987599645,
      "grad_norm": 0.23737424612045288,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 59945
    },
    {
      "epoch": 53.100088573959255,
      "grad_norm": 0.2166704684495926,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 59950
    },
    {
      "epoch": 53.104517271922056,
      "grad_norm": 0.23745734989643097,
      "learning_rate": 1e-05,
      "loss": 1.0161,
      "step": 59955
    },
    {
      "epoch": 53.10894596988486,
      "grad_norm": 0.23875974118709564,
      "learning_rate": 1e-05,
      "loss": 0.9137,
      "step": 59960
    },
    {
      "epoch": 53.11337466784765,
      "grad_norm": 0.22952696681022644,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 59965
    },
    {
      "epoch": 53.11780336581045,
      "grad_norm": 0.2571762800216675,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 59970
    },
    {
      "epoch": 53.122232063773254,
      "grad_norm": 0.23149101436138153,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 59975
    },
    {
      "epoch": 53.12666076173605,
      "grad_norm": 0.2507995069026947,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 59980
    },
    {
      "epoch": 53.13108945969885,
      "grad_norm": 0.22844278812408447,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 59985
    },
    {
      "epoch": 53.13551815766165,
      "grad_norm": 0.27623775601387024,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 59990
    },
    {
      "epoch": 53.139946855624444,
      "grad_norm": 0.234405055642128,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 59995
    },
    {
      "epoch": 53.144375553587246,
      "grad_norm": 0.23527589440345764,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 60000
    },
    {
      "epoch": 53.14880425155005,
      "grad_norm": 0.22647306323051453,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 60005
    },
    {
      "epoch": 53.15323294951284,
      "grad_norm": 0.25990360975265503,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 60010
    },
    {
      "epoch": 53.15766164747564,
      "grad_norm": 0.2278909683227539,
      "learning_rate": 1e-05,
      "loss": 0.9301,
      "step": 60015
    },
    {
      "epoch": 53.16209034543844,
      "grad_norm": 0.22557573020458221,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 60020
    },
    {
      "epoch": 53.16651904340124,
      "grad_norm": 0.21799437701702118,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 60025
    },
    {
      "epoch": 53.17094774136404,
      "grad_norm": 0.22027651965618134,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 60030
    },
    {
      "epoch": 53.17537643932684,
      "grad_norm": 0.2320096790790558,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 60035
    },
    {
      "epoch": 53.179805137289634,
      "grad_norm": 0.23430219292640686,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 60040
    },
    {
      "epoch": 53.184233835252435,
      "grad_norm": 0.2303348183631897,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 60045
    },
    {
      "epoch": 53.188662533215236,
      "grad_norm": 0.20613956451416016,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 60050
    },
    {
      "epoch": 53.19309123117803,
      "grad_norm": 0.26425671577453613,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 60055
    },
    {
      "epoch": 53.19751992914083,
      "grad_norm": 0.23431171476840973,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 60060
    },
    {
      "epoch": 53.20194862710363,
      "grad_norm": 0.2612230181694031,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 60065
    },
    {
      "epoch": 53.20637732506643,
      "grad_norm": 0.2589799463748932,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 60070
    },
    {
      "epoch": 53.21080602302923,
      "grad_norm": 0.2895714342594147,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 60075
    },
    {
      "epoch": 53.21523472099203,
      "grad_norm": 0.26066553592681885,
      "learning_rate": 1e-05,
      "loss": 0.9299,
      "step": 60080
    },
    {
      "epoch": 53.21966341895483,
      "grad_norm": 0.25754573941230774,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 60085
    },
    {
      "epoch": 53.224092116917625,
      "grad_norm": 0.27685636281967163,
      "learning_rate": 1e-05,
      "loss": 1.0261,
      "step": 60090
    },
    {
      "epoch": 53.228520814880426,
      "grad_norm": 0.23839853703975677,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 60095
    },
    {
      "epoch": 53.23294951284323,
      "grad_norm": 0.22421537339687347,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 60100
    },
    {
      "epoch": 53.23737821080602,
      "grad_norm": 0.21918919682502747,
      "learning_rate": 1e-05,
      "loss": 0.9167,
      "step": 60105
    },
    {
      "epoch": 53.24180690876882,
      "grad_norm": 0.2267196774482727,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 60110
    },
    {
      "epoch": 53.246235606731624,
      "grad_norm": 0.21427926421165466,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 60115
    },
    {
      "epoch": 53.25066430469442,
      "grad_norm": 0.2027479112148285,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 60120
    },
    {
      "epoch": 53.25509300265722,
      "grad_norm": 0.22369693219661713,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 60125
    },
    {
      "epoch": 53.25952170062002,
      "grad_norm": 0.22161203622817993,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 60130
    },
    {
      "epoch": 53.263950398582814,
      "grad_norm": 0.30367374420166016,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 60135
    },
    {
      "epoch": 53.268379096545615,
      "grad_norm": 0.22306816279888153,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 60140
    },
    {
      "epoch": 53.27280779450842,
      "grad_norm": 0.23401185870170593,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 60145
    },
    {
      "epoch": 53.27723649247121,
      "grad_norm": 0.2387910783290863,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 60150
    },
    {
      "epoch": 53.28166519043401,
      "grad_norm": 0.2351236492395401,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 60155
    },
    {
      "epoch": 53.28609388839681,
      "grad_norm": 0.23330087959766388,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 60160
    },
    {
      "epoch": 53.29052258635961,
      "grad_norm": 0.2882341146469116,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 60165
    },
    {
      "epoch": 53.29495128432241,
      "grad_norm": 0.23369009792804718,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 60170
    },
    {
      "epoch": 53.29937998228521,
      "grad_norm": 0.24204441905021667,
      "learning_rate": 1e-05,
      "loss": 0.9194,
      "step": 60175
    },
    {
      "epoch": 53.303808680248004,
      "grad_norm": 0.2050991952419281,
      "learning_rate": 1e-05,
      "loss": 1.0372,
      "step": 60180
    },
    {
      "epoch": 53.308237378210805,
      "grad_norm": 0.2358250766992569,
      "learning_rate": 1e-05,
      "loss": 1.0303,
      "step": 60185
    },
    {
      "epoch": 53.312666076173606,
      "grad_norm": 0.27120256423950195,
      "learning_rate": 1e-05,
      "loss": 0.9289,
      "step": 60190
    },
    {
      "epoch": 53.31709477413641,
      "grad_norm": 0.2972188889980316,
      "learning_rate": 1e-05,
      "loss": 1.0248,
      "step": 60195
    },
    {
      "epoch": 53.3215234720992,
      "grad_norm": 0.24034792184829712,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 60200
    },
    {
      "epoch": 53.325952170062,
      "grad_norm": 0.19845075905323029,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 60205
    },
    {
      "epoch": 53.330380868024804,
      "grad_norm": 0.23408180475234985,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 60210
    },
    {
      "epoch": 53.3348095659876,
      "grad_norm": 0.20962047576904297,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 60215
    },
    {
      "epoch": 53.3392382639504,
      "grad_norm": 0.25279760360717773,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 60220
    },
    {
      "epoch": 53.3436669619132,
      "grad_norm": 0.26563024520874023,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 60225
    },
    {
      "epoch": 53.348095659875995,
      "grad_norm": 0.27955740690231323,
      "learning_rate": 1e-05,
      "loss": 0.9142,
      "step": 60230
    },
    {
      "epoch": 53.352524357838796,
      "grad_norm": 0.26782798767089844,
      "learning_rate": 1e-05,
      "loss": 0.9064,
      "step": 60235
    },
    {
      "epoch": 53.3569530558016,
      "grad_norm": 0.2246531993150711,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 60240
    },
    {
      "epoch": 53.36138175376439,
      "grad_norm": 0.2505984902381897,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 60245
    },
    {
      "epoch": 53.36581045172719,
      "grad_norm": 0.21696841716766357,
      "learning_rate": 1e-05,
      "loss": 0.8989,
      "step": 60250
    },
    {
      "epoch": 53.37023914968999,
      "grad_norm": 0.2331947237253189,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 60255
    },
    {
      "epoch": 53.37466784765279,
      "grad_norm": 0.24772530794143677,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 60260
    },
    {
      "epoch": 53.37909654561559,
      "grad_norm": 0.23740357160568237,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 60265
    },
    {
      "epoch": 53.38352524357839,
      "grad_norm": 0.21484477818012238,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 60270
    },
    {
      "epoch": 53.387953941541184,
      "grad_norm": 0.26013872027397156,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 60275
    },
    {
      "epoch": 53.392382639503985,
      "grad_norm": 0.23608742654323578,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 60280
    },
    {
      "epoch": 53.39681133746679,
      "grad_norm": 0.24695591628551483,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 60285
    },
    {
      "epoch": 53.40124003542958,
      "grad_norm": 0.22766916453838348,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 60290
    },
    {
      "epoch": 53.40566873339238,
      "grad_norm": 0.26797938346862793,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 60295
    },
    {
      "epoch": 53.41009743135518,
      "grad_norm": 0.21601247787475586,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 60300
    },
    {
      "epoch": 53.41452612931798,
      "grad_norm": 0.2364301234483719,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 60305
    },
    {
      "epoch": 53.41895482728078,
      "grad_norm": 0.22183459997177124,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 60310
    },
    {
      "epoch": 53.42338352524358,
      "grad_norm": 0.2335205227136612,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 60315
    },
    {
      "epoch": 53.42781222320638,
      "grad_norm": 0.2170056849718094,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 60320
    },
    {
      "epoch": 53.432240921169175,
      "grad_norm": 0.23669470846652985,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 60325
    },
    {
      "epoch": 53.436669619131976,
      "grad_norm": 0.27642059326171875,
      "learning_rate": 1e-05,
      "loss": 0.9249,
      "step": 60330
    },
    {
      "epoch": 53.44109831709478,
      "grad_norm": 0.21286150813102722,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 60335
    },
    {
      "epoch": 53.44552701505757,
      "grad_norm": 0.22230182588100433,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 60340
    },
    {
      "epoch": 53.44995571302037,
      "grad_norm": 0.24709220230579376,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 60345
    },
    {
      "epoch": 53.454384410983174,
      "grad_norm": 0.22660088539123535,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 60350
    },
    {
      "epoch": 53.45881310894597,
      "grad_norm": 0.24670866131782532,
      "learning_rate": 1e-05,
      "loss": 0.9234,
      "step": 60355
    },
    {
      "epoch": 53.46324180690877,
      "grad_norm": 0.239283949136734,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 60360
    },
    {
      "epoch": 53.46767050487157,
      "grad_norm": 0.243422731757164,
      "learning_rate": 1e-05,
      "loss": 1.0216,
      "step": 60365
    },
    {
      "epoch": 53.472099202834364,
      "grad_norm": 0.2267739623785019,
      "learning_rate": 1e-05,
      "loss": 0.9235,
      "step": 60370
    },
    {
      "epoch": 53.476527900797166,
      "grad_norm": 0.23952731490135193,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 60375
    },
    {
      "epoch": 53.48095659875997,
      "grad_norm": 0.24978460371494293,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 60380
    },
    {
      "epoch": 53.48538529672276,
      "grad_norm": 0.20844578742980957,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 60385
    },
    {
      "epoch": 53.48981399468556,
      "grad_norm": 0.26046037673950195,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 60390
    },
    {
      "epoch": 53.49424269264836,
      "grad_norm": 0.2372545599937439,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 60395
    },
    {
      "epoch": 53.49867139061116,
      "grad_norm": 0.21985235810279846,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 60400
    },
    {
      "epoch": 53.50310008857396,
      "grad_norm": 0.24295611679553986,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 60405
    },
    {
      "epoch": 53.50752878653676,
      "grad_norm": 0.23884452879428864,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 60410
    },
    {
      "epoch": 53.511957484499554,
      "grad_norm": 0.21319353580474854,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 60415
    },
    {
      "epoch": 53.516386182462355,
      "grad_norm": 0.2133975625038147,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 60420
    },
    {
      "epoch": 53.520814880425156,
      "grad_norm": 0.2694464325904846,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 60425
    },
    {
      "epoch": 53.52524357838795,
      "grad_norm": 0.24523159861564636,
      "learning_rate": 1e-05,
      "loss": 0.9173,
      "step": 60430
    },
    {
      "epoch": 53.52967227635075,
      "grad_norm": 0.24195018410682678,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 60435
    },
    {
      "epoch": 53.53410097431355,
      "grad_norm": 0.22272612154483795,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 60440
    },
    {
      "epoch": 53.538529672276354,
      "grad_norm": 0.22450736165046692,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 60445
    },
    {
      "epoch": 53.54295837023915,
      "grad_norm": 0.26652538776397705,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 60450
    },
    {
      "epoch": 53.54738706820195,
      "grad_norm": 0.2680169939994812,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 60455
    },
    {
      "epoch": 53.55181576616475,
      "grad_norm": 0.21304230391979218,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 60460
    },
    {
      "epoch": 53.556244464127545,
      "grad_norm": 0.2085220366716385,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 60465
    },
    {
      "epoch": 53.560673162090346,
      "grad_norm": 0.21584953367710114,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 60470
    },
    {
      "epoch": 53.56510186005315,
      "grad_norm": 0.2310914248228073,
      "learning_rate": 1e-05,
      "loss": 1.0239,
      "step": 60475
    },
    {
      "epoch": 53.56953055801594,
      "grad_norm": 0.21490155160427094,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 60480
    },
    {
      "epoch": 53.57395925597874,
      "grad_norm": 0.2255762219429016,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 60485
    },
    {
      "epoch": 53.578387953941544,
      "grad_norm": 0.21962149441242218,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 60490
    },
    {
      "epoch": 53.58281665190434,
      "grad_norm": 0.2350894659757614,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 60495
    },
    {
      "epoch": 53.58724534986714,
      "grad_norm": 0.2455650269985199,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 60500
    },
    {
      "epoch": 53.59167404782994,
      "grad_norm": 0.29344671964645386,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 60505
    },
    {
      "epoch": 53.596102745792734,
      "grad_norm": 0.2824232280254364,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 60510
    },
    {
      "epoch": 53.600531443755536,
      "grad_norm": 0.20357902348041534,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 60515
    },
    {
      "epoch": 53.60496014171834,
      "grad_norm": 0.23408834636211395,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 60520
    },
    {
      "epoch": 53.60938883968113,
      "grad_norm": 0.23880112171173096,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 60525
    },
    {
      "epoch": 53.61381753764393,
      "grad_norm": 0.24110502004623413,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 60530
    },
    {
      "epoch": 53.61824623560673,
      "grad_norm": 0.2096918523311615,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 60535
    },
    {
      "epoch": 53.62267493356953,
      "grad_norm": 0.2138698846101761,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 60540
    },
    {
      "epoch": 53.62710363153233,
      "grad_norm": 0.23307427763938904,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 60545
    },
    {
      "epoch": 53.63153232949513,
      "grad_norm": 0.26484254002571106,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 60550
    },
    {
      "epoch": 53.635961027457924,
      "grad_norm": 0.23737332224845886,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 60555
    },
    {
      "epoch": 53.640389725420725,
      "grad_norm": 0.20891714096069336,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 60560
    },
    {
      "epoch": 53.644818423383526,
      "grad_norm": 0.23625357449054718,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 60565
    },
    {
      "epoch": 53.64924712134633,
      "grad_norm": 0.22194480895996094,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 60570
    },
    {
      "epoch": 53.65367581930912,
      "grad_norm": 0.23048807680606842,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 60575
    },
    {
      "epoch": 53.65810451727192,
      "grad_norm": 0.23959128558635712,
      "learning_rate": 1e-05,
      "loss": 0.929,
      "step": 60580
    },
    {
      "epoch": 53.662533215234724,
      "grad_norm": 0.23040126264095306,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 60585
    },
    {
      "epoch": 53.66696191319752,
      "grad_norm": 0.2538955509662628,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 60590
    },
    {
      "epoch": 53.67139061116032,
      "grad_norm": 0.2324790209531784,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 60595
    },
    {
      "epoch": 53.67581930912312,
      "grad_norm": 0.19082011282444,
      "learning_rate": 1e-05,
      "loss": 1.0353,
      "step": 60600
    },
    {
      "epoch": 53.680248007085915,
      "grad_norm": 0.22211992740631104,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 60605
    },
    {
      "epoch": 53.684676705048716,
      "grad_norm": 0.22224803268909454,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 60610
    },
    {
      "epoch": 53.68910540301152,
      "grad_norm": 0.2658023536205292,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 60615
    },
    {
      "epoch": 53.69353410097431,
      "grad_norm": 0.22065530717372894,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 60620
    },
    {
      "epoch": 53.69796279893711,
      "grad_norm": 0.2552250027656555,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 60625
    },
    {
      "epoch": 53.702391496899914,
      "grad_norm": 0.23224309086799622,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 60630
    },
    {
      "epoch": 53.70682019486271,
      "grad_norm": 0.2216968834400177,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 60635
    },
    {
      "epoch": 53.71124889282551,
      "grad_norm": 0.25914743542671204,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 60640
    },
    {
      "epoch": 53.71567759078831,
      "grad_norm": 0.27898186445236206,
      "learning_rate": 1e-05,
      "loss": 1.0263,
      "step": 60645
    },
    {
      "epoch": 53.720106288751104,
      "grad_norm": 0.2800960838794708,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 60650
    },
    {
      "epoch": 53.724534986713905,
      "grad_norm": 0.230106383562088,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 60655
    },
    {
      "epoch": 53.72896368467671,
      "grad_norm": 0.23093128204345703,
      "learning_rate": 1e-05,
      "loss": 1.0294,
      "step": 60660
    },
    {
      "epoch": 53.7333923826395,
      "grad_norm": 0.2400493025779724,
      "learning_rate": 1e-05,
      "loss": 0.9271,
      "step": 60665
    },
    {
      "epoch": 53.7378210806023,
      "grad_norm": 0.2102871537208557,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 60670
    },
    {
      "epoch": 53.7422497785651,
      "grad_norm": 0.2573719620704651,
      "learning_rate": 1e-05,
      "loss": 0.9199,
      "step": 60675
    },
    {
      "epoch": 53.7466784765279,
      "grad_norm": 0.24573856592178345,
      "learning_rate": 1e-05,
      "loss": 0.9081,
      "step": 60680
    },
    {
      "epoch": 53.7511071744907,
      "grad_norm": 0.24206885695457458,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 60685
    },
    {
      "epoch": 53.7555358724535,
      "grad_norm": 0.2640511691570282,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 60690
    },
    {
      "epoch": 53.7599645704163,
      "grad_norm": 0.2531708776950836,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 60695
    },
    {
      "epoch": 53.764393268379095,
      "grad_norm": 0.24144858121871948,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 60700
    },
    {
      "epoch": 53.768821966341896,
      "grad_norm": 0.2880392372608185,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 60705
    },
    {
      "epoch": 53.7732506643047,
      "grad_norm": 0.2411138117313385,
      "learning_rate": 1e-05,
      "loss": 0.9212,
      "step": 60710
    },
    {
      "epoch": 53.77767936226749,
      "grad_norm": 0.3010050356388092,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 60715
    },
    {
      "epoch": 53.78210806023029,
      "grad_norm": 0.21168889105319977,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 60720
    },
    {
      "epoch": 53.786536758193094,
      "grad_norm": 0.24816155433654785,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 60725
    },
    {
      "epoch": 53.79096545615589,
      "grad_norm": 0.30651214718818665,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 60730
    },
    {
      "epoch": 53.79539415411869,
      "grad_norm": 0.27157482504844666,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 60735
    },
    {
      "epoch": 53.79982285208149,
      "grad_norm": 0.2219117134809494,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 60740
    },
    {
      "epoch": 53.804251550044285,
      "grad_norm": 0.26084062457084656,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 60745
    },
    {
      "epoch": 53.808680248007086,
      "grad_norm": 0.2546119689941406,
      "learning_rate": 1e-05,
      "loss": 1.0296,
      "step": 60750
    },
    {
      "epoch": 53.81310894596989,
      "grad_norm": 0.23620547354221344,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 60755
    },
    {
      "epoch": 53.81753764393268,
      "grad_norm": 0.277325838804245,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 60760
    },
    {
      "epoch": 53.82196634189548,
      "grad_norm": 0.25658735632896423,
      "learning_rate": 1e-05,
      "loss": 0.9188,
      "step": 60765
    },
    {
      "epoch": 53.826395039858284,
      "grad_norm": 0.24100135266780853,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 60770
    },
    {
      "epoch": 53.83082373782108,
      "grad_norm": 0.2314073145389557,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 60775
    },
    {
      "epoch": 53.83525243578388,
      "grad_norm": 0.25841328501701355,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 60780
    },
    {
      "epoch": 53.83968113374668,
      "grad_norm": 0.24181579053401947,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 60785
    },
    {
      "epoch": 53.844109831709474,
      "grad_norm": 0.21728689968585968,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 60790
    },
    {
      "epoch": 53.848538529672275,
      "grad_norm": 0.22766830027103424,
      "learning_rate": 1e-05,
      "loss": 0.9249,
      "step": 60795
    },
    {
      "epoch": 53.85296722763508,
      "grad_norm": 0.22813519835472107,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 60800
    },
    {
      "epoch": 53.85739592559787,
      "grad_norm": 0.2287793606519699,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 60805
    },
    {
      "epoch": 53.86182462356067,
      "grad_norm": 0.2280627191066742,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 60810
    },
    {
      "epoch": 53.86625332152347,
      "grad_norm": 0.23534254729747772,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 60815
    },
    {
      "epoch": 53.870682019486274,
      "grad_norm": 0.20020708441734314,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 60820
    },
    {
      "epoch": 53.87511071744907,
      "grad_norm": 0.2288442701101303,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 60825
    },
    {
      "epoch": 53.87953941541187,
      "grad_norm": 0.2380603700876236,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 60830
    },
    {
      "epoch": 53.88396811337467,
      "grad_norm": 0.2640037536621094,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 60835
    },
    {
      "epoch": 53.888396811337465,
      "grad_norm": 0.2480287402868271,
      "learning_rate": 1e-05,
      "loss": 1.0181,
      "step": 60840
    },
    {
      "epoch": 53.892825509300266,
      "grad_norm": 0.21241746842861176,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 60845
    },
    {
      "epoch": 53.89725420726307,
      "grad_norm": 0.22319309413433075,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 60850
    },
    {
      "epoch": 53.90168290522586,
      "grad_norm": 0.21784153580665588,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 60855
    },
    {
      "epoch": 53.90611160318866,
      "grad_norm": 0.23020143806934357,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 60860
    },
    {
      "epoch": 53.910540301151464,
      "grad_norm": 0.24832095205783844,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 60865
    },
    {
      "epoch": 53.91496899911426,
      "grad_norm": 0.19770771265029907,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 60870
    },
    {
      "epoch": 53.91939769707706,
      "grad_norm": 0.25580593943595886,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 60875
    },
    {
      "epoch": 53.92382639503986,
      "grad_norm": 0.2854255735874176,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 60880
    },
    {
      "epoch": 53.928255093002655,
      "grad_norm": 0.24428914487361908,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 60885
    },
    {
      "epoch": 53.932683790965456,
      "grad_norm": 0.2942395508289337,
      "learning_rate": 1e-05,
      "loss": 1.0376,
      "step": 60890
    },
    {
      "epoch": 53.93711248892826,
      "grad_norm": 0.28321489691734314,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 60895
    },
    {
      "epoch": 53.94154118689105,
      "grad_norm": 0.24364061653614044,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 60900
    },
    {
      "epoch": 53.94596988485385,
      "grad_norm": 0.22843430936336517,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 60905
    },
    {
      "epoch": 53.95039858281665,
      "grad_norm": 0.2668422758579254,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 60910
    },
    {
      "epoch": 53.95482728077945,
      "grad_norm": 0.2423567920923233,
      "learning_rate": 1e-05,
      "loss": 0.9082,
      "step": 60915
    },
    {
      "epoch": 53.95925597874225,
      "grad_norm": 0.21341903507709503,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 60920
    },
    {
      "epoch": 53.96368467670505,
      "grad_norm": 0.20688697695732117,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 60925
    },
    {
      "epoch": 53.96811337466785,
      "grad_norm": 0.2544124722480774,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 60930
    },
    {
      "epoch": 53.972542072630645,
      "grad_norm": 0.2535913586616516,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 60935
    },
    {
      "epoch": 53.97697077059345,
      "grad_norm": 0.2406681627035141,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 60940
    },
    {
      "epoch": 53.98139946855625,
      "grad_norm": 0.18434345722198486,
      "learning_rate": 1e-05,
      "loss": 0.9084,
      "step": 60945
    },
    {
      "epoch": 53.98582816651904,
      "grad_norm": 0.23859114944934845,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 60950
    },
    {
      "epoch": 53.99025686448184,
      "grad_norm": 0.26423484086990356,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 60955
    },
    {
      "epoch": 53.994685562444644,
      "grad_norm": 0.2631922662258148,
      "learning_rate": 1e-05,
      "loss": 0.9301,
      "step": 60960
    },
    {
      "epoch": 53.99911426040744,
      "grad_norm": 0.24404844641685486,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 60965
    },
    {
      "epoch": 54.00354295837024,
      "grad_norm": 0.22955921292304993,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 60970
    },
    {
      "epoch": 54.00797165633304,
      "grad_norm": 0.22842659056186676,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 60975
    },
    {
      "epoch": 54.012400354295835,
      "grad_norm": 0.25447097420692444,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 60980
    },
    {
      "epoch": 54.016829052258636,
      "grad_norm": 0.28639480471611023,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 60985
    },
    {
      "epoch": 54.02125775022144,
      "grad_norm": 0.2449595332145691,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 60990
    },
    {
      "epoch": 54.02568644818423,
      "grad_norm": 0.29549911618232727,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 60995
    },
    {
      "epoch": 54.03011514614703,
      "grad_norm": 0.26542067527770996,
      "learning_rate": 1e-05,
      "loss": 1.0195,
      "step": 61000
    },
    {
      "epoch": 54.034543844109834,
      "grad_norm": 0.25493472814559937,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 61005
    },
    {
      "epoch": 54.03897254207263,
      "grad_norm": 0.24698977172374725,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 61010
    },
    {
      "epoch": 54.04340124003543,
      "grad_norm": 0.2290639728307724,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 61015
    },
    {
      "epoch": 54.04782993799823,
      "grad_norm": 0.2510537803173065,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 61020
    },
    {
      "epoch": 54.052258635961024,
      "grad_norm": 0.24928176403045654,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 61025
    },
    {
      "epoch": 54.056687333923826,
      "grad_norm": 0.2595762312412262,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 61030
    },
    {
      "epoch": 54.06111603188663,
      "grad_norm": 0.22515444457530975,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 61035
    },
    {
      "epoch": 54.06554472984942,
      "grad_norm": 0.23509569466114044,
      "learning_rate": 1e-05,
      "loss": 0.9121,
      "step": 61040
    },
    {
      "epoch": 54.06997342781222,
      "grad_norm": 0.2688516676425934,
      "learning_rate": 1e-05,
      "loss": 0.9292,
      "step": 61045
    },
    {
      "epoch": 54.07440212577502,
      "grad_norm": 0.21259772777557373,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 61050
    },
    {
      "epoch": 54.078830823737825,
      "grad_norm": 0.22343476116657257,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 61055
    },
    {
      "epoch": 54.08325952170062,
      "grad_norm": 0.210588738322258,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 61060
    },
    {
      "epoch": 54.08768821966342,
      "grad_norm": 0.2256932258605957,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 61065
    },
    {
      "epoch": 54.09211691762622,
      "grad_norm": 0.29084935784339905,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 61070
    },
    {
      "epoch": 54.096545615589015,
      "grad_norm": 0.28121477365493774,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 61075
    },
    {
      "epoch": 54.100974313551816,
      "grad_norm": 0.3363478481769562,
      "learning_rate": 1e-05,
      "loss": 1.0254,
      "step": 61080
    },
    {
      "epoch": 54.10540301151462,
      "grad_norm": 0.2818223237991333,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 61085
    },
    {
      "epoch": 54.10983170947741,
      "grad_norm": 0.21597203612327576,
      "learning_rate": 1e-05,
      "loss": 0.9076,
      "step": 61090
    },
    {
      "epoch": 54.11426040744021,
      "grad_norm": 0.2366877943277359,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 61095
    },
    {
      "epoch": 54.118689105403014,
      "grad_norm": 0.22500531375408173,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 61100
    },
    {
      "epoch": 54.12311780336581,
      "grad_norm": 0.2209455519914627,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 61105
    },
    {
      "epoch": 54.12754650132861,
      "grad_norm": 0.23634248971939087,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 61110
    },
    {
      "epoch": 54.13197519929141,
      "grad_norm": 0.25187307596206665,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 61115
    },
    {
      "epoch": 54.136403897254205,
      "grad_norm": 0.273495614528656,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 61120
    },
    {
      "epoch": 54.140832595217006,
      "grad_norm": 0.3204086124897003,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 61125
    },
    {
      "epoch": 54.14526129317981,
      "grad_norm": 0.3518579304218292,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 61130
    },
    {
      "epoch": 54.1496899911426,
      "grad_norm": 0.3865751028060913,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 61135
    },
    {
      "epoch": 54.1541186891054,
      "grad_norm": 0.22252608835697174,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 61140
    },
    {
      "epoch": 54.158547387068204,
      "grad_norm": 0.2905064523220062,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 61145
    },
    {
      "epoch": 54.162976085031,
      "grad_norm": 0.22203530371189117,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 61150
    },
    {
      "epoch": 54.1674047829938,
      "grad_norm": 0.2612822353839874,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 61155
    },
    {
      "epoch": 54.1718334809566,
      "grad_norm": 0.23783627152442932,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 61160
    },
    {
      "epoch": 54.176262178919394,
      "grad_norm": 0.2373826652765274,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 61165
    },
    {
      "epoch": 54.180690876882196,
      "grad_norm": 0.2468205988407135,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 61170
    },
    {
      "epoch": 54.185119574845,
      "grad_norm": 0.2272808998823166,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 61175
    },
    {
      "epoch": 54.1895482728078,
      "grad_norm": 0.27046507596969604,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 61180
    },
    {
      "epoch": 54.19397697077059,
      "grad_norm": 0.23064535856246948,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 61185
    },
    {
      "epoch": 54.19840566873339,
      "grad_norm": 0.22172294557094574,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 61190
    },
    {
      "epoch": 54.202834366696194,
      "grad_norm": 0.226271390914917,
      "learning_rate": 1e-05,
      "loss": 0.9065,
      "step": 61195
    },
    {
      "epoch": 54.20726306465899,
      "grad_norm": 0.22542870044708252,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 61200
    },
    {
      "epoch": 54.21169176262179,
      "grad_norm": 0.22811514139175415,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 61205
    },
    {
      "epoch": 54.21612046058459,
      "grad_norm": 0.26387643814086914,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 61210
    },
    {
      "epoch": 54.220549158547385,
      "grad_norm": 0.23597018420696259,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 61215
    },
    {
      "epoch": 54.224977856510186,
      "grad_norm": 0.2678360939025879,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 61220
    },
    {
      "epoch": 54.22940655447299,
      "grad_norm": 0.2404780387878418,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 61225
    },
    {
      "epoch": 54.23383525243578,
      "grad_norm": 0.2540590167045593,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 61230
    },
    {
      "epoch": 54.23826395039858,
      "grad_norm": 0.24822445213794708,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 61235
    },
    {
      "epoch": 54.242692648361384,
      "grad_norm": 0.23118668794631958,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 61240
    },
    {
      "epoch": 54.24712134632418,
      "grad_norm": 0.24964898824691772,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 61245
    },
    {
      "epoch": 54.25155004428698,
      "grad_norm": 0.22792859375476837,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 61250
    },
    {
      "epoch": 54.25597874224978,
      "grad_norm": 0.22979103028774261,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 61255
    },
    {
      "epoch": 54.260407440212575,
      "grad_norm": 0.2890489399433136,
      "learning_rate": 1e-05,
      "loss": 0.9211,
      "step": 61260
    },
    {
      "epoch": 54.264836138175376,
      "grad_norm": 0.2525692582130432,
      "learning_rate": 1e-05,
      "loss": 1.0305,
      "step": 61265
    },
    {
      "epoch": 54.26926483613818,
      "grad_norm": 0.24704942107200623,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 61270
    },
    {
      "epoch": 54.27369353410097,
      "grad_norm": 0.2595321834087372,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 61275
    },
    {
      "epoch": 54.27812223206377,
      "grad_norm": 0.22018831968307495,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 61280
    },
    {
      "epoch": 54.282550930026574,
      "grad_norm": 0.2448827475309372,
      "learning_rate": 1e-05,
      "loss": 0.9174,
      "step": 61285
    },
    {
      "epoch": 54.28697962798937,
      "grad_norm": 0.2333334982395172,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 61290
    },
    {
      "epoch": 54.29140832595217,
      "grad_norm": 0.22767828404903412,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 61295
    },
    {
      "epoch": 54.29583702391497,
      "grad_norm": 0.23467841744422913,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 61300
    },
    {
      "epoch": 54.30026572187777,
      "grad_norm": 0.2241162359714508,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 61305
    },
    {
      "epoch": 54.304694419840565,
      "grad_norm": 0.233202263712883,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 61310
    },
    {
      "epoch": 54.30912311780337,
      "grad_norm": 0.21783636510372162,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 61315
    },
    {
      "epoch": 54.31355181576617,
      "grad_norm": 0.2490084171295166,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 61320
    },
    {
      "epoch": 54.31798051372896,
      "grad_norm": 0.21604113280773163,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 61325
    },
    {
      "epoch": 54.32240921169176,
      "grad_norm": 0.21787811815738678,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 61330
    },
    {
      "epoch": 54.326837909654564,
      "grad_norm": 0.2443985939025879,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 61335
    },
    {
      "epoch": 54.33126660761736,
      "grad_norm": 0.23070906102657318,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 61340
    },
    {
      "epoch": 54.33569530558016,
      "grad_norm": 0.23045994341373444,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 61345
    },
    {
      "epoch": 54.34012400354296,
      "grad_norm": 0.27131208777427673,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 61350
    },
    {
      "epoch": 54.344552701505755,
      "grad_norm": 0.2660194933414459,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 61355
    },
    {
      "epoch": 54.348981399468556,
      "grad_norm": 0.23644393682479858,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 61360
    },
    {
      "epoch": 54.35341009743136,
      "grad_norm": 0.26887020468711853,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 61365
    },
    {
      "epoch": 54.35783879539415,
      "grad_norm": 0.2516176104545593,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 61370
    },
    {
      "epoch": 54.36226749335695,
      "grad_norm": 0.21114002168178558,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 61375
    },
    {
      "epoch": 54.366696191319754,
      "grad_norm": 0.25534120202064514,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 61380
    },
    {
      "epoch": 54.37112488928255,
      "grad_norm": 0.21948450803756714,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 61385
    },
    {
      "epoch": 54.37555358724535,
      "grad_norm": 0.21734066307544708,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 61390
    },
    {
      "epoch": 54.37998228520815,
      "grad_norm": 0.25333961844444275,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 61395
    },
    {
      "epoch": 54.384410983170945,
      "grad_norm": 0.21795569360256195,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 61400
    },
    {
      "epoch": 54.388839681133746,
      "grad_norm": 0.26207274198532104,
      "learning_rate": 1e-05,
      "loss": 1.0238,
      "step": 61405
    },
    {
      "epoch": 54.39326837909655,
      "grad_norm": 0.2518703043460846,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 61410
    },
    {
      "epoch": 54.39769707705934,
      "grad_norm": 0.2144148051738739,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 61415
    },
    {
      "epoch": 54.40212577502214,
      "grad_norm": 0.231561079621315,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 61420
    },
    {
      "epoch": 54.40655447298494,
      "grad_norm": 0.268319696187973,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 61425
    },
    {
      "epoch": 54.410983170947745,
      "grad_norm": 0.2864091992378235,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 61430
    },
    {
      "epoch": 54.41541186891054,
      "grad_norm": 0.25611644983291626,
      "learning_rate": 1e-05,
      "loss": 0.9168,
      "step": 61435
    },
    {
      "epoch": 54.41984056687334,
      "grad_norm": 0.2154947817325592,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 61440
    },
    {
      "epoch": 54.42426926483614,
      "grad_norm": 0.21742206811904907,
      "learning_rate": 1e-05,
      "loss": 0.9231,
      "step": 61445
    },
    {
      "epoch": 54.428697962798935,
      "grad_norm": 0.24300965666770935,
      "learning_rate": 1e-05,
      "loss": 1.0471,
      "step": 61450
    },
    {
      "epoch": 54.43312666076174,
      "grad_norm": 0.22295871376991272,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 61455
    },
    {
      "epoch": 54.43755535872454,
      "grad_norm": 0.24372051656246185,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 61460
    },
    {
      "epoch": 54.44198405668733,
      "grad_norm": 0.21082350611686707,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 61465
    },
    {
      "epoch": 54.44641275465013,
      "grad_norm": 0.2071627825498581,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 61470
    },
    {
      "epoch": 54.450841452612934,
      "grad_norm": 0.26812344789505005,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 61475
    },
    {
      "epoch": 54.45527015057573,
      "grad_norm": 0.24720941483974457,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 61480
    },
    {
      "epoch": 54.45969884853853,
      "grad_norm": 0.22771954536437988,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 61485
    },
    {
      "epoch": 54.46412754650133,
      "grad_norm": 0.2672770321369171,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 61490
    },
    {
      "epoch": 54.468556244464125,
      "grad_norm": 0.25041714310646057,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 61495
    },
    {
      "epoch": 54.472984942426926,
      "grad_norm": 0.263938844203949,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 61500
    },
    {
      "epoch": 54.47741364038973,
      "grad_norm": 0.2439061999320984,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 61505
    },
    {
      "epoch": 54.48184233835252,
      "grad_norm": 0.23846527934074402,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 61510
    },
    {
      "epoch": 54.48627103631532,
      "grad_norm": 0.23370777070522308,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 61515
    },
    {
      "epoch": 54.490699734278124,
      "grad_norm": 0.22891657054424286,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 61520
    },
    {
      "epoch": 54.49512843224092,
      "grad_norm": 0.2401563972234726,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 61525
    },
    {
      "epoch": 54.49955713020372,
      "grad_norm": 0.23497898876667023,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 61530
    },
    {
      "epoch": 54.50398582816652,
      "grad_norm": 0.2689017355442047,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 61535
    },
    {
      "epoch": 54.508414526129314,
      "grad_norm": 0.2731553912162781,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 61540
    },
    {
      "epoch": 54.512843224092116,
      "grad_norm": 0.2824132442474365,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 61545
    },
    {
      "epoch": 54.51727192205492,
      "grad_norm": 0.2665017247200012,
      "learning_rate": 1e-05,
      "loss": 0.9206,
      "step": 61550
    },
    {
      "epoch": 54.52170062001772,
      "grad_norm": 0.24270696938037872,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 61555
    },
    {
      "epoch": 54.52612931798051,
      "grad_norm": 0.23882102966308594,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 61560
    },
    {
      "epoch": 54.53055801594331,
      "grad_norm": 0.25425294041633606,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 61565
    },
    {
      "epoch": 54.534986713906115,
      "grad_norm": 0.238195538520813,
      "learning_rate": 1e-05,
      "loss": 1.0277,
      "step": 61570
    },
    {
      "epoch": 54.53941541186891,
      "grad_norm": 0.26347872614860535,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 61575
    },
    {
      "epoch": 54.54384410983171,
      "grad_norm": 0.2912469804286957,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 61580
    },
    {
      "epoch": 54.54827280779451,
      "grad_norm": 0.2978285849094391,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 61585
    },
    {
      "epoch": 54.552701505757305,
      "grad_norm": 0.252096027135849,
      "learning_rate": 1e-05,
      "loss": 1.0593,
      "step": 61590
    },
    {
      "epoch": 54.55713020372011,
      "grad_norm": 0.21461506187915802,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 61595
    },
    {
      "epoch": 54.56155890168291,
      "grad_norm": 0.2181003987789154,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 61600
    },
    {
      "epoch": 54.5659875996457,
      "grad_norm": 0.2642095386981964,
      "learning_rate": 1e-05,
      "loss": 1.017,
      "step": 61605
    },
    {
      "epoch": 54.5704162976085,
      "grad_norm": 0.276835173368454,
      "learning_rate": 1e-05,
      "loss": 1.0413,
      "step": 61610
    },
    {
      "epoch": 54.574844995571304,
      "grad_norm": 0.261502742767334,
      "learning_rate": 1e-05,
      "loss": 0.9119,
      "step": 61615
    },
    {
      "epoch": 54.5792736935341,
      "grad_norm": 0.23839789628982544,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 61620
    },
    {
      "epoch": 54.5837023914969,
      "grad_norm": 0.243050679564476,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 61625
    },
    {
      "epoch": 54.5881310894597,
      "grad_norm": 0.249448761343956,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 61630
    },
    {
      "epoch": 54.592559787422495,
      "grad_norm": 0.2467772364616394,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 61635
    },
    {
      "epoch": 54.596988485385296,
      "grad_norm": 0.2678309381008148,
      "learning_rate": 1e-05,
      "loss": 0.8949,
      "step": 61640
    },
    {
      "epoch": 54.6014171833481,
      "grad_norm": 0.2426210343837738,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 61645
    },
    {
      "epoch": 54.60584588131089,
      "grad_norm": 0.20903442800045013,
      "learning_rate": 1e-05,
      "loss": 0.9008,
      "step": 61650
    },
    {
      "epoch": 54.61027457927369,
      "grad_norm": 0.24812667071819305,
      "learning_rate": 1e-05,
      "loss": 1.0231,
      "step": 61655
    },
    {
      "epoch": 54.614703277236494,
      "grad_norm": 0.2490062117576599,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 61660
    },
    {
      "epoch": 54.619131975199295,
      "grad_norm": 0.22590519487857819,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 61665
    },
    {
      "epoch": 54.62356067316209,
      "grad_norm": 0.2515568137168884,
      "learning_rate": 1e-05,
      "loss": 1.0303,
      "step": 61670
    },
    {
      "epoch": 54.62798937112489,
      "grad_norm": 0.25598007440567017,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 61675
    },
    {
      "epoch": 54.63241806908769,
      "grad_norm": 0.2397070676088333,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 61680
    },
    {
      "epoch": 54.636846767050486,
      "grad_norm": 0.20996831357479095,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 61685
    },
    {
      "epoch": 54.64127546501329,
      "grad_norm": 0.2177884876728058,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 61690
    },
    {
      "epoch": 54.64570416297609,
      "grad_norm": 0.26674431562423706,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 61695
    },
    {
      "epoch": 54.65013286093888,
      "grad_norm": 0.24462474882602692,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 61700
    },
    {
      "epoch": 54.65456155890168,
      "grad_norm": 0.24660229682922363,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 61705
    },
    {
      "epoch": 54.658990256864485,
      "grad_norm": 0.21257789433002472,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 61710
    },
    {
      "epoch": 54.66341895482728,
      "grad_norm": 0.23424005508422852,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 61715
    },
    {
      "epoch": 54.66784765279008,
      "grad_norm": 0.280011922121048,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 61720
    },
    {
      "epoch": 54.67227635075288,
      "grad_norm": 0.2793463468551636,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 61725
    },
    {
      "epoch": 54.676705048715675,
      "grad_norm": 0.23466026782989502,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 61730
    },
    {
      "epoch": 54.681133746678476,
      "grad_norm": 0.20742113888263702,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 61735
    },
    {
      "epoch": 54.68556244464128,
      "grad_norm": 0.24423709511756897,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 61740
    },
    {
      "epoch": 54.68999114260407,
      "grad_norm": 0.2402326911687851,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 61745
    },
    {
      "epoch": 54.69441984056687,
      "grad_norm": 0.22081421315670013,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 61750
    },
    {
      "epoch": 54.698848538529674,
      "grad_norm": 0.24656403064727783,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 61755
    },
    {
      "epoch": 54.70327723649247,
      "grad_norm": 0.284909188747406,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 61760
    },
    {
      "epoch": 54.70770593445527,
      "grad_norm": 0.23226577043533325,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 61765
    },
    {
      "epoch": 54.71213463241807,
      "grad_norm": 0.21470679342746735,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 61770
    },
    {
      "epoch": 54.716563330380865,
      "grad_norm": 0.2293403595685959,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 61775
    },
    {
      "epoch": 54.720992028343666,
      "grad_norm": 0.2368682473897934,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 61780
    },
    {
      "epoch": 54.72542072630647,
      "grad_norm": 0.22414761781692505,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 61785
    },
    {
      "epoch": 54.72984942426926,
      "grad_norm": 0.23385803401470184,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 61790
    },
    {
      "epoch": 54.73427812223206,
      "grad_norm": 0.21191293001174927,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 61795
    },
    {
      "epoch": 54.738706820194864,
      "grad_norm": 0.24699129164218903,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 61800
    },
    {
      "epoch": 54.743135518157665,
      "grad_norm": 0.22854012250900269,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 61805
    },
    {
      "epoch": 54.74756421612046,
      "grad_norm": 0.2828068137168884,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 61810
    },
    {
      "epoch": 54.75199291408326,
      "grad_norm": 0.2359173595905304,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 61815
    },
    {
      "epoch": 54.75642161204606,
      "grad_norm": 0.23383831977844238,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 61820
    },
    {
      "epoch": 54.760850310008856,
      "grad_norm": 0.26490044593811035,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 61825
    },
    {
      "epoch": 54.76527900797166,
      "grad_norm": 0.24586886167526245,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 61830
    },
    {
      "epoch": 54.76970770593446,
      "grad_norm": 0.25765252113342285,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 61835
    },
    {
      "epoch": 54.77413640389725,
      "grad_norm": 0.27040475606918335,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 61840
    },
    {
      "epoch": 54.77856510186005,
      "grad_norm": 0.2563571035861969,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 61845
    },
    {
      "epoch": 54.782993799822854,
      "grad_norm": 0.25582483410835266,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 61850
    },
    {
      "epoch": 54.78742249778565,
      "grad_norm": 0.22667090594768524,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 61855
    },
    {
      "epoch": 54.79185119574845,
      "grad_norm": 0.2700563669204712,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 61860
    },
    {
      "epoch": 54.79627989371125,
      "grad_norm": 0.23663084208965302,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 61865
    },
    {
      "epoch": 54.800708591674045,
      "grad_norm": 0.22395679354667664,
      "learning_rate": 1e-05,
      "loss": 1.0679,
      "step": 61870
    },
    {
      "epoch": 54.805137289636846,
      "grad_norm": 0.21704262495040894,
      "learning_rate": 1e-05,
      "loss": 0.9209,
      "step": 61875
    },
    {
      "epoch": 54.80956598759965,
      "grad_norm": 0.21953852474689484,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 61880
    },
    {
      "epoch": 54.81399468556244,
      "grad_norm": 0.22336454689502716,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 61885
    },
    {
      "epoch": 54.81842338352524,
      "grad_norm": 0.24294808506965637,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 61890
    },
    {
      "epoch": 54.822852081488044,
      "grad_norm": 0.25395455956459045,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 61895
    },
    {
      "epoch": 54.82728077945084,
      "grad_norm": 0.21166537702083588,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 61900
    },
    {
      "epoch": 54.83170947741364,
      "grad_norm": 0.2385307401418686,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 61905
    },
    {
      "epoch": 54.83613817537644,
      "grad_norm": 0.24236778914928436,
      "learning_rate": 1e-05,
      "loss": 0.915,
      "step": 61910
    },
    {
      "epoch": 54.84056687333924,
      "grad_norm": 0.2466571033000946,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 61915
    },
    {
      "epoch": 54.844995571302036,
      "grad_norm": 0.26242566108703613,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 61920
    },
    {
      "epoch": 54.84942426926484,
      "grad_norm": 0.2473292201757431,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 61925
    },
    {
      "epoch": 54.85385296722764,
      "grad_norm": 0.24475917220115662,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 61930
    },
    {
      "epoch": 54.85828166519043,
      "grad_norm": 0.23263022303581238,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 61935
    },
    {
      "epoch": 54.862710363153234,
      "grad_norm": 0.2234267145395279,
      "learning_rate": 1e-05,
      "loss": 1.0441,
      "step": 61940
    },
    {
      "epoch": 54.867139061116035,
      "grad_norm": 0.2610851228237152,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 61945
    },
    {
      "epoch": 54.87156775907883,
      "grad_norm": 0.24277083575725555,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 61950
    },
    {
      "epoch": 54.87599645704163,
      "grad_norm": 0.24271883070468903,
      "learning_rate": 1e-05,
      "loss": 1.0161,
      "step": 61955
    },
    {
      "epoch": 54.88042515500443,
      "grad_norm": 0.24794341623783112,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 61960
    },
    {
      "epoch": 54.884853852967225,
      "grad_norm": 0.2069029062986374,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 61965
    },
    {
      "epoch": 54.88928255093003,
      "grad_norm": 0.21977069973945618,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 61970
    },
    {
      "epoch": 54.89371124889283,
      "grad_norm": 0.23933996260166168,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 61975
    },
    {
      "epoch": 54.89813994685562,
      "grad_norm": 0.22514726221561432,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 61980
    },
    {
      "epoch": 54.90256864481842,
      "grad_norm": 0.22039583325386047,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 61985
    },
    {
      "epoch": 54.906997342781224,
      "grad_norm": 0.23116083443164825,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 61990
    },
    {
      "epoch": 54.91142604074402,
      "grad_norm": 0.19800740480422974,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 61995
    },
    {
      "epoch": 54.91585473870682,
      "grad_norm": 0.2333899438381195,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 62000
    },
    {
      "epoch": 54.92028343666962,
      "grad_norm": 0.22694028913974762,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 62005
    },
    {
      "epoch": 54.924712134632415,
      "grad_norm": 0.27045968174934387,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 62010
    },
    {
      "epoch": 54.929140832595216,
      "grad_norm": 0.2575031518936157,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 62015
    },
    {
      "epoch": 54.93356953055802,
      "grad_norm": 0.24480172991752625,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 62020
    },
    {
      "epoch": 54.93799822852081,
      "grad_norm": 0.2962493300437927,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 62025
    },
    {
      "epoch": 54.94242692648361,
      "grad_norm": 0.27831003069877625,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 62030
    },
    {
      "epoch": 54.946855624446414,
      "grad_norm": 0.2670118510723114,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 62035
    },
    {
      "epoch": 54.951284322409215,
      "grad_norm": 0.26412615180015564,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 62040
    },
    {
      "epoch": 54.95571302037201,
      "grad_norm": 0.2617094814777374,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 62045
    },
    {
      "epoch": 54.96014171833481,
      "grad_norm": 0.22700825333595276,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 62050
    },
    {
      "epoch": 54.96457041629761,
      "grad_norm": 0.18606488406658173,
      "learning_rate": 1e-05,
      "loss": 0.9342,
      "step": 62055
    },
    {
      "epoch": 54.968999114260406,
      "grad_norm": 0.2505088448524475,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 62060
    },
    {
      "epoch": 54.97342781222321,
      "grad_norm": 0.2616252601146698,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 62065
    },
    {
      "epoch": 54.97785651018601,
      "grad_norm": 0.24996070563793182,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 62070
    },
    {
      "epoch": 54.9822852081488,
      "grad_norm": 0.22112376987934113,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 62075
    },
    {
      "epoch": 54.9867139061116,
      "grad_norm": 0.21114081144332886,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 62080
    },
    {
      "epoch": 54.991142604074405,
      "grad_norm": 0.2259313315153122,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 62085
    },
    {
      "epoch": 54.9955713020372,
      "grad_norm": 0.21121808886528015,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 62090
    },
    {
      "epoch": 55.0,
      "grad_norm": 0.25856801867485046,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 62095
    },
    {
      "epoch": 55.0044286979628,
      "grad_norm": 0.2713790535926819,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 62100
    },
    {
      "epoch": 55.008857395925595,
      "grad_norm": 0.2832414209842682,
      "learning_rate": 1e-05,
      "loss": 0.9167,
      "step": 62105
    },
    {
      "epoch": 55.0132860938884,
      "grad_norm": 0.23071420192718506,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 62110
    },
    {
      "epoch": 55.0177147918512,
      "grad_norm": 0.22246162593364716,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 62115
    },
    {
      "epoch": 55.02214348981399,
      "grad_norm": 0.2495613545179367,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 62120
    },
    {
      "epoch": 55.02657218777679,
      "grad_norm": 0.270028293132782,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 62125
    },
    {
      "epoch": 55.031000885739594,
      "grad_norm": 0.2383924275636673,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 62130
    },
    {
      "epoch": 55.03542958370239,
      "grad_norm": 0.21703945100307465,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 62135
    },
    {
      "epoch": 55.03985828166519,
      "grad_norm": 0.24551594257354736,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 62140
    },
    {
      "epoch": 55.04428697962799,
      "grad_norm": 0.2273833304643631,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 62145
    },
    {
      "epoch": 55.048715677590785,
      "grad_norm": 0.25210875272750854,
      "learning_rate": 1e-05,
      "loss": 0.9249,
      "step": 62150
    },
    {
      "epoch": 55.053144375553586,
      "grad_norm": 0.26989150047302246,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 62155
    },
    {
      "epoch": 55.05757307351639,
      "grad_norm": 0.256803423166275,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 62160
    },
    {
      "epoch": 55.06200177147919,
      "grad_norm": 0.22256067395210266,
      "learning_rate": 1e-05,
      "loss": 0.8808,
      "step": 62165
    },
    {
      "epoch": 55.06643046944198,
      "grad_norm": 0.2533958852291107,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 62170
    },
    {
      "epoch": 55.070859167404784,
      "grad_norm": 0.2591835856437683,
      "learning_rate": 1e-05,
      "loss": 0.9193,
      "step": 62175
    },
    {
      "epoch": 55.075287865367585,
      "grad_norm": 0.22120441496372223,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 62180
    },
    {
      "epoch": 55.07971656333038,
      "grad_norm": 0.20032048225402832,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 62185
    },
    {
      "epoch": 55.08414526129318,
      "grad_norm": 0.27565130591392517,
      "learning_rate": 1e-05,
      "loss": 1.0251,
      "step": 62190
    },
    {
      "epoch": 55.08857395925598,
      "grad_norm": 0.20962965488433838,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 62195
    },
    {
      "epoch": 55.093002657218776,
      "grad_norm": 0.2532493472099304,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 62200
    },
    {
      "epoch": 55.09743135518158,
      "grad_norm": 0.22159592807292938,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 62205
    },
    {
      "epoch": 55.10186005314438,
      "grad_norm": 0.23838955163955688,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 62210
    },
    {
      "epoch": 55.10628875110717,
      "grad_norm": 0.230186328291893,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 62215
    },
    {
      "epoch": 55.11071744906997,
      "grad_norm": 0.23161542415618896,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 62220
    },
    {
      "epoch": 55.115146147032775,
      "grad_norm": 0.22316773235797882,
      "learning_rate": 1e-05,
      "loss": 0.9277,
      "step": 62225
    },
    {
      "epoch": 55.11957484499557,
      "grad_norm": 0.2775658965110779,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 62230
    },
    {
      "epoch": 55.12400354295837,
      "grad_norm": 0.21759754419326782,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 62235
    },
    {
      "epoch": 55.12843224092117,
      "grad_norm": 0.210435688495636,
      "learning_rate": 1e-05,
      "loss": 0.9191,
      "step": 62240
    },
    {
      "epoch": 55.132860938883965,
      "grad_norm": 0.2566128373146057,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 62245
    },
    {
      "epoch": 55.137289636846766,
      "grad_norm": 0.22963926196098328,
      "learning_rate": 1e-05,
      "loss": 1.0394,
      "step": 62250
    },
    {
      "epoch": 55.14171833480957,
      "grad_norm": 0.31815585494041443,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 62255
    },
    {
      "epoch": 55.14614703277236,
      "grad_norm": 0.26324036717414856,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 62260
    },
    {
      "epoch": 55.15057573073516,
      "grad_norm": 0.2819535434246063,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 62265
    },
    {
      "epoch": 55.155004428697964,
      "grad_norm": 0.263812392950058,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 62270
    },
    {
      "epoch": 55.15943312666076,
      "grad_norm": 0.22461117804050446,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 62275
    },
    {
      "epoch": 55.16386182462356,
      "grad_norm": 0.21655142307281494,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 62280
    },
    {
      "epoch": 55.16829052258636,
      "grad_norm": 0.28236937522888184,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 62285
    },
    {
      "epoch": 55.17271922054916,
      "grad_norm": 0.243770569562912,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 62290
    },
    {
      "epoch": 55.177147918511956,
      "grad_norm": 0.24992871284484863,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 62295
    },
    {
      "epoch": 55.18157661647476,
      "grad_norm": 0.24871692061424255,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 62300
    },
    {
      "epoch": 55.18600531443756,
      "grad_norm": 0.22770540416240692,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 62305
    },
    {
      "epoch": 55.19043401240035,
      "grad_norm": 0.3503894507884979,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 62310
    },
    {
      "epoch": 55.194862710363154,
      "grad_norm": 0.24219578504562378,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 62315
    },
    {
      "epoch": 55.199291408325955,
      "grad_norm": 0.21635305881500244,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 62320
    },
    {
      "epoch": 55.20372010628875,
      "grad_norm": 0.27642425894737244,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 62325
    },
    {
      "epoch": 55.20814880425155,
      "grad_norm": 0.2553144097328186,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 62330
    },
    {
      "epoch": 55.21257750221435,
      "grad_norm": 0.25659167766571045,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 62335
    },
    {
      "epoch": 55.217006200177146,
      "grad_norm": 0.24272117018699646,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 62340
    },
    {
      "epoch": 55.22143489813995,
      "grad_norm": 0.21669627726078033,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 62345
    },
    {
      "epoch": 55.22586359610275,
      "grad_norm": 0.26535534858703613,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 62350
    },
    {
      "epoch": 55.23029229406554,
      "grad_norm": 0.2685007154941559,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 62355
    },
    {
      "epoch": 55.23472099202834,
      "grad_norm": 0.24896568059921265,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 62360
    },
    {
      "epoch": 55.239149689991144,
      "grad_norm": 0.21610885858535767,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 62365
    },
    {
      "epoch": 55.24357838795394,
      "grad_norm": 0.21651025116443634,
      "learning_rate": 1e-05,
      "loss": 1.0341,
      "step": 62370
    },
    {
      "epoch": 55.24800708591674,
      "grad_norm": 0.19530245661735535,
      "learning_rate": 1e-05,
      "loss": 1.0292,
      "step": 62375
    },
    {
      "epoch": 55.25243578387954,
      "grad_norm": 0.2430613785982132,
      "learning_rate": 1e-05,
      "loss": 1.0261,
      "step": 62380
    },
    {
      "epoch": 55.256864481842335,
      "grad_norm": 0.23671753704547882,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 62385
    },
    {
      "epoch": 55.261293179805136,
      "grad_norm": 0.22328652441501617,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 62390
    },
    {
      "epoch": 55.26572187776794,
      "grad_norm": 0.24272359907627106,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 62395
    },
    {
      "epoch": 55.27015057573073,
      "grad_norm": 0.2391354888677597,
      "learning_rate": 1e-05,
      "loss": 1.0434,
      "step": 62400
    },
    {
      "epoch": 55.27457927369353,
      "grad_norm": 0.23660340905189514,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 62405
    },
    {
      "epoch": 55.279007971656334,
      "grad_norm": 0.28350868821144104,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 62410
    },
    {
      "epoch": 55.283436669619135,
      "grad_norm": 0.24310103058815002,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 62415
    },
    {
      "epoch": 55.28786536758193,
      "grad_norm": 0.2373107224702835,
      "learning_rate": 1e-05,
      "loss": 0.9263,
      "step": 62420
    },
    {
      "epoch": 55.29229406554473,
      "grad_norm": 0.2172693908214569,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 62425
    },
    {
      "epoch": 55.29672276350753,
      "grad_norm": 0.2619110345840454,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 62430
    },
    {
      "epoch": 55.301151461470326,
      "grad_norm": 0.26987820863723755,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 62435
    },
    {
      "epoch": 55.30558015943313,
      "grad_norm": 0.2333763986825943,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 62440
    },
    {
      "epoch": 55.31000885739593,
      "grad_norm": 0.21843530237674713,
      "learning_rate": 1e-05,
      "loss": 0.9136,
      "step": 62445
    },
    {
      "epoch": 55.31443755535872,
      "grad_norm": 0.2594922184944153,
      "learning_rate": 1e-05,
      "loss": 0.9364,
      "step": 62450
    },
    {
      "epoch": 55.318866253321524,
      "grad_norm": 0.28343212604522705,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 62455
    },
    {
      "epoch": 55.323294951284325,
      "grad_norm": 0.2748631238937378,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 62460
    },
    {
      "epoch": 55.32772364924712,
      "grad_norm": 0.26620927453041077,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 62465
    },
    {
      "epoch": 55.33215234720992,
      "grad_norm": 0.2769966721534729,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 62470
    },
    {
      "epoch": 55.33658104517272,
      "grad_norm": 0.26412954926490784,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 62475
    },
    {
      "epoch": 55.341009743135515,
      "grad_norm": 0.29838427901268005,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 62480
    },
    {
      "epoch": 55.34543844109832,
      "grad_norm": 0.24841853976249695,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 62485
    },
    {
      "epoch": 55.34986713906112,
      "grad_norm": 0.25717467069625854,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 62490
    },
    {
      "epoch": 55.35429583702391,
      "grad_norm": 0.22899077832698822,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 62495
    },
    {
      "epoch": 55.35872453498671,
      "grad_norm": 0.25663626194000244,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 62500
    },
    {
      "epoch": 55.363153232949514,
      "grad_norm": 0.2554994523525238,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 62505
    },
    {
      "epoch": 55.36758193091231,
      "grad_norm": 0.33785203099250793,
      "learning_rate": 1e-05,
      "loss": 1.0199,
      "step": 62510
    },
    {
      "epoch": 55.37201062887511,
      "grad_norm": 0.2501871585845947,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 62515
    },
    {
      "epoch": 55.37643932683791,
      "grad_norm": 0.24996094405651093,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 62520
    },
    {
      "epoch": 55.380868024800705,
      "grad_norm": 0.227278470993042,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 62525
    },
    {
      "epoch": 55.385296722763506,
      "grad_norm": 0.26868489384651184,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 62530
    },
    {
      "epoch": 55.38972542072631,
      "grad_norm": 0.27864688634872437,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 62535
    },
    {
      "epoch": 55.39415411868911,
      "grad_norm": 0.2396826595067978,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 62540
    },
    {
      "epoch": 55.3985828166519,
      "grad_norm": 0.2545613944530487,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 62545
    },
    {
      "epoch": 55.403011514614704,
      "grad_norm": 0.22027942538261414,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 62550
    },
    {
      "epoch": 55.407440212577505,
      "grad_norm": 0.265720397233963,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 62555
    },
    {
      "epoch": 55.4118689105403,
      "grad_norm": 0.2706334888935089,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 62560
    },
    {
      "epoch": 55.4162976085031,
      "grad_norm": 0.23240944743156433,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 62565
    },
    {
      "epoch": 55.4207263064659,
      "grad_norm": 0.2491762936115265,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 62570
    },
    {
      "epoch": 55.425155004428696,
      "grad_norm": 0.20693622529506683,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 62575
    },
    {
      "epoch": 55.4295837023915,
      "grad_norm": 0.2690315544605255,
      "learning_rate": 1e-05,
      "loss": 1.0097,
      "step": 62580
    },
    {
      "epoch": 55.4340124003543,
      "grad_norm": 0.22171130776405334,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 62585
    },
    {
      "epoch": 55.43844109831709,
      "grad_norm": 0.27453330159187317,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 62590
    },
    {
      "epoch": 55.44286979627989,
      "grad_norm": 0.2499241679906845,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 62595
    },
    {
      "epoch": 55.447298494242695,
      "grad_norm": 0.28359460830688477,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 62600
    },
    {
      "epoch": 55.45172719220549,
      "grad_norm": 0.22056515514850616,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 62605
    },
    {
      "epoch": 55.45615589016829,
      "grad_norm": 0.26162782311439514,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 62610
    },
    {
      "epoch": 55.46058458813109,
      "grad_norm": 0.24669480323791504,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 62615
    },
    {
      "epoch": 55.465013286093885,
      "grad_norm": 0.2601143419742584,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 62620
    },
    {
      "epoch": 55.46944198405669,
      "grad_norm": 0.2427825629711151,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 62625
    },
    {
      "epoch": 55.47387068201949,
      "grad_norm": 0.19705727696418762,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 62630
    },
    {
      "epoch": 55.47829937998228,
      "grad_norm": 0.2382509559392929,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 62635
    },
    {
      "epoch": 55.48272807794508,
      "grad_norm": 0.21953454613685608,
      "learning_rate": 1e-05,
      "loss": 0.9006,
      "step": 62640
    },
    {
      "epoch": 55.487156775907884,
      "grad_norm": 0.2171293944120407,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 62645
    },
    {
      "epoch": 55.491585473870686,
      "grad_norm": 0.24557551741600037,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 62650
    },
    {
      "epoch": 55.49601417183348,
      "grad_norm": 0.20853084325790405,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 62655
    },
    {
      "epoch": 55.50044286979628,
      "grad_norm": 0.25548630952835083,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 62660
    },
    {
      "epoch": 55.50487156775908,
      "grad_norm": 0.23453408479690552,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 62665
    },
    {
      "epoch": 55.509300265721876,
      "grad_norm": 0.23328697681427002,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 62670
    },
    {
      "epoch": 55.51372896368468,
      "grad_norm": 0.2442711889743805,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 62675
    },
    {
      "epoch": 55.51815766164748,
      "grad_norm": 0.25138574838638306,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 62680
    },
    {
      "epoch": 55.52258635961027,
      "grad_norm": 0.22825178503990173,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 62685
    },
    {
      "epoch": 55.527015057573074,
      "grad_norm": 0.25139036774635315,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 62690
    },
    {
      "epoch": 55.531443755535875,
      "grad_norm": 0.2730069160461426,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 62695
    },
    {
      "epoch": 55.53587245349867,
      "grad_norm": 0.23859600722789764,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 62700
    },
    {
      "epoch": 55.54030115146147,
      "grad_norm": 0.2873833477497101,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 62705
    },
    {
      "epoch": 55.54472984942427,
      "grad_norm": 0.2638431191444397,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 62710
    },
    {
      "epoch": 55.549158547387066,
      "grad_norm": 0.2742599844932556,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 62715
    },
    {
      "epoch": 55.55358724534987,
      "grad_norm": 0.24800468981266022,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 62720
    },
    {
      "epoch": 55.55801594331267,
      "grad_norm": 0.22815638780593872,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 62725
    },
    {
      "epoch": 55.56244464127546,
      "grad_norm": 0.31424880027770996,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 62730
    },
    {
      "epoch": 55.56687333923826,
      "grad_norm": 0.20046547055244446,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 62735
    },
    {
      "epoch": 55.571302037201065,
      "grad_norm": 0.24561159312725067,
      "learning_rate": 1e-05,
      "loss": 1.0186,
      "step": 62740
    },
    {
      "epoch": 55.57573073516386,
      "grad_norm": 0.21277067065238953,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 62745
    },
    {
      "epoch": 55.58015943312666,
      "grad_norm": 0.2338026463985443,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 62750
    },
    {
      "epoch": 55.58458813108946,
      "grad_norm": 0.22151383757591248,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 62755
    },
    {
      "epoch": 55.589016829052255,
      "grad_norm": 0.21774446964263916,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 62760
    },
    {
      "epoch": 55.59344552701506,
      "grad_norm": 0.23514288663864136,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 62765
    },
    {
      "epoch": 55.59787422497786,
      "grad_norm": 0.26878175139427185,
      "learning_rate": 1e-05,
      "loss": 0.941,
      "step": 62770
    },
    {
      "epoch": 55.60230292294066,
      "grad_norm": 0.22452412545681,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 62775
    },
    {
      "epoch": 55.60673162090345,
      "grad_norm": 0.26823243498802185,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 62780
    },
    {
      "epoch": 55.611160318866254,
      "grad_norm": 0.22869546711444855,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 62785
    },
    {
      "epoch": 55.615589016829055,
      "grad_norm": 0.23047584295272827,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 62790
    },
    {
      "epoch": 55.62001771479185,
      "grad_norm": 0.21435987949371338,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 62795
    },
    {
      "epoch": 55.62444641275465,
      "grad_norm": 0.2282906174659729,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 62800
    },
    {
      "epoch": 55.62887511071745,
      "grad_norm": 0.22043752670288086,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 62805
    },
    {
      "epoch": 55.633303808680246,
      "grad_norm": 0.23046617209911346,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 62810
    },
    {
      "epoch": 55.63773250664305,
      "grad_norm": 0.24408136308193207,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 62815
    },
    {
      "epoch": 55.64216120460585,
      "grad_norm": 0.24182654917240143,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 62820
    },
    {
      "epoch": 55.64658990256864,
      "grad_norm": 0.2297814041376114,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 62825
    },
    {
      "epoch": 55.651018600531444,
      "grad_norm": 0.2758806645870209,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 62830
    },
    {
      "epoch": 55.655447298494245,
      "grad_norm": 0.2332402616739273,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 62835
    },
    {
      "epoch": 55.65987599645704,
      "grad_norm": 0.23059028387069702,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 62840
    },
    {
      "epoch": 55.66430469441984,
      "grad_norm": 0.2054479569196701,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 62845
    },
    {
      "epoch": 55.66873339238264,
      "grad_norm": 0.227760449051857,
      "learning_rate": 1e-05,
      "loss": 0.8962,
      "step": 62850
    },
    {
      "epoch": 55.673162090345436,
      "grad_norm": 0.22935718297958374,
      "learning_rate": 1e-05,
      "loss": 0.9297,
      "step": 62855
    },
    {
      "epoch": 55.67759078830824,
      "grad_norm": 0.21174666285514832,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 62860
    },
    {
      "epoch": 55.68201948627104,
      "grad_norm": 0.2854756712913513,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 62865
    },
    {
      "epoch": 55.68644818423383,
      "grad_norm": 0.2622082233428955,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 62870
    },
    {
      "epoch": 55.69087688219663,
      "grad_norm": 0.24696001410484314,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 62875
    },
    {
      "epoch": 55.695305580159435,
      "grad_norm": 0.22844815254211426,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 62880
    },
    {
      "epoch": 55.69973427812223,
      "grad_norm": 0.2736019492149353,
      "learning_rate": 1e-05,
      "loss": 0.9396,
      "step": 62885
    },
    {
      "epoch": 55.70416297608503,
      "grad_norm": 0.2845125198364258,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 62890
    },
    {
      "epoch": 55.70859167404783,
      "grad_norm": 0.25794246792793274,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 62895
    },
    {
      "epoch": 55.71302037201063,
      "grad_norm": 0.2491748183965683,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 62900
    },
    {
      "epoch": 55.717449069973426,
      "grad_norm": 0.22434252500534058,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 62905
    },
    {
      "epoch": 55.72187776793623,
      "grad_norm": 0.257744163274765,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 62910
    },
    {
      "epoch": 55.72630646589903,
      "grad_norm": 0.28021812438964844,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 62915
    },
    {
      "epoch": 55.73073516386182,
      "grad_norm": 0.2581353783607483,
      "learning_rate": 1e-05,
      "loss": 1.0141,
      "step": 62920
    },
    {
      "epoch": 55.735163861824624,
      "grad_norm": 0.2833695709705353,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 62925
    },
    {
      "epoch": 55.739592559787425,
      "grad_norm": 0.23834869265556335,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 62930
    },
    {
      "epoch": 55.74402125775022,
      "grad_norm": 0.24291785061359406,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 62935
    },
    {
      "epoch": 55.74844995571302,
      "grad_norm": 0.2367095947265625,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 62940
    },
    {
      "epoch": 55.75287865367582,
      "grad_norm": 0.22828327119350433,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 62945
    },
    {
      "epoch": 55.757307351638616,
      "grad_norm": 0.1983243227005005,
      "learning_rate": 1e-05,
      "loss": 0.9299,
      "step": 62950
    },
    {
      "epoch": 55.76173604960142,
      "grad_norm": 0.21977515518665314,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 62955
    },
    {
      "epoch": 55.76616474756422,
      "grad_norm": 0.2196301370859146,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 62960
    },
    {
      "epoch": 55.77059344552701,
      "grad_norm": 0.2925870418548584,
      "learning_rate": 1e-05,
      "loss": 1.0181,
      "step": 62965
    },
    {
      "epoch": 55.775022143489814,
      "grad_norm": 0.2700136601924896,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 62970
    },
    {
      "epoch": 55.779450841452615,
      "grad_norm": 0.24620164930820465,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 62975
    },
    {
      "epoch": 55.78387953941541,
      "grad_norm": 0.2316984087228775,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 62980
    },
    {
      "epoch": 55.78830823737821,
      "grad_norm": 0.2617681920528412,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 62985
    },
    {
      "epoch": 55.79273693534101,
      "grad_norm": 0.2237551361322403,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 62990
    },
    {
      "epoch": 55.797165633303806,
      "grad_norm": 0.24159231781959534,
      "learning_rate": 1e-05,
      "loss": 0.9324,
      "step": 62995
    },
    {
      "epoch": 55.80159433126661,
      "grad_norm": 0.24352505803108215,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 63000
    },
    {
      "epoch": 55.80602302922941,
      "grad_norm": 0.23825091123580933,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 63005
    },
    {
      "epoch": 55.8104517271922,
      "grad_norm": 0.24412472546100616,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 63010
    },
    {
      "epoch": 55.814880425155,
      "grad_norm": 0.21329577267169952,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 63015
    },
    {
      "epoch": 55.819309123117804,
      "grad_norm": 0.2143968641757965,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 63020
    },
    {
      "epoch": 55.823737821080606,
      "grad_norm": 0.2275923490524292,
      "learning_rate": 1e-05,
      "loss": 1.0318,
      "step": 63025
    },
    {
      "epoch": 55.8281665190434,
      "grad_norm": 0.22037973999977112,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 63030
    },
    {
      "epoch": 55.8325952170062,
      "grad_norm": 0.23195819556713104,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 63035
    },
    {
      "epoch": 55.837023914969,
      "grad_norm": 0.23441609740257263,
      "learning_rate": 1e-05,
      "loss": 1.0513,
      "step": 63040
    },
    {
      "epoch": 55.841452612931796,
      "grad_norm": 0.26434823870658875,
      "learning_rate": 1e-05,
      "loss": 1.0424,
      "step": 63045
    },
    {
      "epoch": 55.8458813108946,
      "grad_norm": 0.2524208724498749,
      "learning_rate": 1e-05,
      "loss": 1.0403,
      "step": 63050
    },
    {
      "epoch": 55.8503100088574,
      "grad_norm": 0.278835266828537,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 63055
    },
    {
      "epoch": 55.85473870682019,
      "grad_norm": 0.20644482970237732,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 63060
    },
    {
      "epoch": 55.859167404782994,
      "grad_norm": 0.21144837141036987,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 63065
    },
    {
      "epoch": 55.863596102745795,
      "grad_norm": 0.22869908809661865,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 63070
    },
    {
      "epoch": 55.86802480070859,
      "grad_norm": 0.24267901480197906,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 63075
    },
    {
      "epoch": 55.87245349867139,
      "grad_norm": 0.2329319566488266,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 63080
    },
    {
      "epoch": 55.87688219663419,
      "grad_norm": 0.2512611746788025,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 63085
    },
    {
      "epoch": 55.881310894596986,
      "grad_norm": 0.25006699562072754,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 63090
    },
    {
      "epoch": 55.88573959255979,
      "grad_norm": 0.2464587241411209,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 63095
    },
    {
      "epoch": 55.89016829052259,
      "grad_norm": 0.24423974752426147,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 63100
    },
    {
      "epoch": 55.89459698848538,
      "grad_norm": 0.22483476996421814,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 63105
    },
    {
      "epoch": 55.899025686448184,
      "grad_norm": 0.27327942848205566,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 63110
    },
    {
      "epoch": 55.903454384410985,
      "grad_norm": 0.22841021418571472,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 63115
    },
    {
      "epoch": 55.90788308237378,
      "grad_norm": 0.21562814712524414,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 63120
    },
    {
      "epoch": 55.91231178033658,
      "grad_norm": 0.2504241466522217,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 63125
    },
    {
      "epoch": 55.91674047829938,
      "grad_norm": 0.23530380427837372,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 63130
    },
    {
      "epoch": 55.921169176262175,
      "grad_norm": 0.2534446120262146,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 63135
    },
    {
      "epoch": 55.92559787422498,
      "grad_norm": 0.2118387073278427,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 63140
    },
    {
      "epoch": 55.93002657218778,
      "grad_norm": 0.23165062069892883,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 63145
    },
    {
      "epoch": 55.93445527015058,
      "grad_norm": 0.23897498846054077,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 63150
    },
    {
      "epoch": 55.93888396811337,
      "grad_norm": 0.18769510090351105,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 63155
    },
    {
      "epoch": 55.943312666076174,
      "grad_norm": 0.21453003585338593,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 63160
    },
    {
      "epoch": 55.947741364038976,
      "grad_norm": 0.25377440452575684,
      "learning_rate": 1e-05,
      "loss": 0.909,
      "step": 63165
    },
    {
      "epoch": 55.95217006200177,
      "grad_norm": 0.25588148832321167,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 63170
    },
    {
      "epoch": 55.95659875996457,
      "grad_norm": 0.2416241466999054,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 63175
    },
    {
      "epoch": 55.96102745792737,
      "grad_norm": 0.28508681058883667,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 63180
    },
    {
      "epoch": 55.965456155890166,
      "grad_norm": 0.25988131761550903,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 63185
    },
    {
      "epoch": 55.96988485385297,
      "grad_norm": 0.2363196611404419,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 63190
    },
    {
      "epoch": 55.97431355181577,
      "grad_norm": 0.21184957027435303,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 63195
    },
    {
      "epoch": 55.97874224977856,
      "grad_norm": 0.25400838255882263,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 63200
    },
    {
      "epoch": 55.983170947741364,
      "grad_norm": 0.2314334362745285,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 63205
    },
    {
      "epoch": 55.987599645704165,
      "grad_norm": 0.23777985572814941,
      "learning_rate": 1e-05,
      "loss": 1.0291,
      "step": 63210
    },
    {
      "epoch": 55.99202834366696,
      "grad_norm": 0.27431249618530273,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 63215
    },
    {
      "epoch": 55.99645704162976,
      "grad_norm": 0.24531115591526031,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 63220
    },
    {
      "epoch": 56.00088573959256,
      "grad_norm": 0.23937658965587616,
      "learning_rate": 1e-05,
      "loss": 1.0392,
      "step": 63225
    },
    {
      "epoch": 56.005314437555356,
      "grad_norm": 0.25790509581565857,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 63230
    },
    {
      "epoch": 56.00974313551816,
      "grad_norm": 0.22111348807811737,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 63235
    },
    {
      "epoch": 56.01417183348096,
      "grad_norm": 0.24599124491214752,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 63240
    },
    {
      "epoch": 56.01860053144375,
      "grad_norm": 0.24326390027999878,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 63245
    },
    {
      "epoch": 56.02302922940655,
      "grad_norm": 0.26520904898643494,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 63250
    },
    {
      "epoch": 56.027457927369355,
      "grad_norm": 0.2518685460090637,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 63255
    },
    {
      "epoch": 56.03188662533215,
      "grad_norm": 0.22588272392749786,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 63260
    },
    {
      "epoch": 56.03631532329495,
      "grad_norm": 0.2146756649017334,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 63265
    },
    {
      "epoch": 56.04074402125775,
      "grad_norm": 0.2459445744752884,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 63270
    },
    {
      "epoch": 56.04517271922055,
      "grad_norm": 0.2339220941066742,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 63275
    },
    {
      "epoch": 56.04960141718335,
      "grad_norm": 0.2387182116508484,
      "learning_rate": 1e-05,
      "loss": 0.908,
      "step": 63280
    },
    {
      "epoch": 56.05403011514615,
      "grad_norm": 0.20047685503959656,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 63285
    },
    {
      "epoch": 56.05845881310895,
      "grad_norm": 0.18126839399337769,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 63290
    },
    {
      "epoch": 56.06288751107174,
      "grad_norm": 0.2056189626455307,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 63295
    },
    {
      "epoch": 56.067316209034544,
      "grad_norm": 0.20262739062309265,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 63300
    },
    {
      "epoch": 56.071744906997345,
      "grad_norm": 0.2546745538711548,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 63305
    },
    {
      "epoch": 56.07617360496014,
      "grad_norm": 0.24659445881843567,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 63310
    },
    {
      "epoch": 56.08060230292294,
      "grad_norm": 0.2330092042684555,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 63315
    },
    {
      "epoch": 56.08503100088574,
      "grad_norm": 0.23480361700057983,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 63320
    },
    {
      "epoch": 56.089459698848536,
      "grad_norm": 0.2409174144268036,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 63325
    },
    {
      "epoch": 56.09388839681134,
      "grad_norm": 0.26574721932411194,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 63330
    },
    {
      "epoch": 56.09831709477414,
      "grad_norm": 0.2549513280391693,
      "learning_rate": 1e-05,
      "loss": 0.9338,
      "step": 63335
    },
    {
      "epoch": 56.10274579273693,
      "grad_norm": 0.25018665194511414,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 63340
    },
    {
      "epoch": 56.107174490699734,
      "grad_norm": 0.23832666873931885,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 63345
    },
    {
      "epoch": 56.111603188662535,
      "grad_norm": 0.22562791407108307,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 63350
    },
    {
      "epoch": 56.11603188662533,
      "grad_norm": 0.251179963350296,
      "learning_rate": 1e-05,
      "loss": 0.9181,
      "step": 63355
    },
    {
      "epoch": 56.12046058458813,
      "grad_norm": 0.2523738741874695,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 63360
    },
    {
      "epoch": 56.12488928255093,
      "grad_norm": 0.2886253893375397,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 63365
    },
    {
      "epoch": 56.129317980513726,
      "grad_norm": 0.27339133620262146,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 63370
    },
    {
      "epoch": 56.13374667847653,
      "grad_norm": 0.22999246418476105,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 63375
    },
    {
      "epoch": 56.13817537643933,
      "grad_norm": 0.25712668895721436,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 63380
    },
    {
      "epoch": 56.14260407440213,
      "grad_norm": 0.2335035502910614,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 63385
    },
    {
      "epoch": 56.14703277236492,
      "grad_norm": 0.2485598772764206,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 63390
    },
    {
      "epoch": 56.151461470327725,
      "grad_norm": 0.20362693071365356,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 63395
    },
    {
      "epoch": 56.155890168290526,
      "grad_norm": 0.22485344111919403,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 63400
    },
    {
      "epoch": 56.16031886625332,
      "grad_norm": 0.25063422322273254,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 63405
    },
    {
      "epoch": 56.16474756421612,
      "grad_norm": 0.26546475291252136,
      "learning_rate": 1e-05,
      "loss": 0.9155,
      "step": 63410
    },
    {
      "epoch": 56.16917626217892,
      "grad_norm": 0.2285129725933075,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 63415
    },
    {
      "epoch": 56.173604960141716,
      "grad_norm": 0.2408449649810791,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 63420
    },
    {
      "epoch": 56.17803365810452,
      "grad_norm": 0.2592220604419708,
      "learning_rate": 1e-05,
      "loss": 0.924,
      "step": 63425
    },
    {
      "epoch": 56.18246235606732,
      "grad_norm": 0.2781519591808319,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 63430
    },
    {
      "epoch": 56.18689105403011,
      "grad_norm": 0.3001905679702759,
      "learning_rate": 1e-05,
      "loss": 0.9205,
      "step": 63435
    },
    {
      "epoch": 56.191319751992914,
      "grad_norm": 0.24128659069538116,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 63440
    },
    {
      "epoch": 56.195748449955715,
      "grad_norm": 0.30806219577789307,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 63445
    },
    {
      "epoch": 56.20017714791851,
      "grad_norm": 0.28717541694641113,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 63450
    },
    {
      "epoch": 56.20460584588131,
      "grad_norm": 0.22641092538833618,
      "learning_rate": 1e-05,
      "loss": 1.0252,
      "step": 63455
    },
    {
      "epoch": 56.20903454384411,
      "grad_norm": 0.31907275319099426,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 63460
    },
    {
      "epoch": 56.213463241806906,
      "grad_norm": 0.23403482139110565,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 63465
    },
    {
      "epoch": 56.21789193976971,
      "grad_norm": 0.2310042679309845,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 63470
    },
    {
      "epoch": 56.22232063773251,
      "grad_norm": 0.2366589456796646,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 63475
    },
    {
      "epoch": 56.2267493356953,
      "grad_norm": 0.2359618991613388,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 63480
    },
    {
      "epoch": 56.231178033658104,
      "grad_norm": 0.26032111048698425,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 63485
    },
    {
      "epoch": 56.235606731620905,
      "grad_norm": 0.22414495050907135,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 63490
    },
    {
      "epoch": 56.2400354295837,
      "grad_norm": 0.22042685747146606,
      "learning_rate": 1e-05,
      "loss": 0.9175,
      "step": 63495
    },
    {
      "epoch": 56.2444641275465,
      "grad_norm": 0.24749435484409332,
      "learning_rate": 1e-05,
      "loss": 0.8955,
      "step": 63500
    },
    {
      "epoch": 56.2488928255093,
      "grad_norm": 0.2552974224090576,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 63505
    },
    {
      "epoch": 56.2533215234721,
      "grad_norm": 0.24924516677856445,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 63510
    },
    {
      "epoch": 56.2577502214349,
      "grad_norm": 0.22684572637081146,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 63515
    },
    {
      "epoch": 56.2621789193977,
      "grad_norm": 0.24484111368656158,
      "learning_rate": 1e-05,
      "loss": 0.9167,
      "step": 63520
    },
    {
      "epoch": 56.2666076173605,
      "grad_norm": 0.28961053490638733,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 63525
    },
    {
      "epoch": 56.27103631532329,
      "grad_norm": 0.2719418406486511,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 63530
    },
    {
      "epoch": 56.275465013286095,
      "grad_norm": 0.19193609058856964,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 63535
    },
    {
      "epoch": 56.279893711248896,
      "grad_norm": 0.2647283375263214,
      "learning_rate": 1e-05,
      "loss": 1.0258,
      "step": 63540
    },
    {
      "epoch": 56.28432240921169,
      "grad_norm": 0.26923179626464844,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 63545
    },
    {
      "epoch": 56.28875110717449,
      "grad_norm": 0.22998864948749542,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 63550
    },
    {
      "epoch": 56.29317980513729,
      "grad_norm": 0.257528692483902,
      "learning_rate": 1e-05,
      "loss": 0.9266,
      "step": 63555
    },
    {
      "epoch": 56.297608503100086,
      "grad_norm": 0.2713509500026703,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 63560
    },
    {
      "epoch": 56.30203720106289,
      "grad_norm": 0.2289247363805771,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 63565
    },
    {
      "epoch": 56.30646589902569,
      "grad_norm": 0.22357754409313202,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 63570
    },
    {
      "epoch": 56.31089459698848,
      "grad_norm": 0.26244646310806274,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 63575
    },
    {
      "epoch": 56.315323294951284,
      "grad_norm": 0.27205440402030945,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 63580
    },
    {
      "epoch": 56.319751992914085,
      "grad_norm": 0.24197699129581451,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 63585
    },
    {
      "epoch": 56.32418069087688,
      "grad_norm": 0.24268017709255219,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 63590
    },
    {
      "epoch": 56.32860938883968,
      "grad_norm": 0.25126227736473083,
      "learning_rate": 1e-05,
      "loss": 0.9133,
      "step": 63595
    },
    {
      "epoch": 56.33303808680248,
      "grad_norm": 0.2567299008369446,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 63600
    },
    {
      "epoch": 56.337466784765276,
      "grad_norm": 0.22538305819034576,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 63605
    },
    {
      "epoch": 56.34189548272808,
      "grad_norm": 0.2222796082496643,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 63610
    },
    {
      "epoch": 56.34632418069088,
      "grad_norm": 0.24867665767669678,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 63615
    },
    {
      "epoch": 56.35075287865367,
      "grad_norm": 0.23514172434806824,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 63620
    },
    {
      "epoch": 56.355181576616474,
      "grad_norm": 0.2335110753774643,
      "learning_rate": 1e-05,
      "loss": 1.0379,
      "step": 63625
    },
    {
      "epoch": 56.359610274579275,
      "grad_norm": 0.2373688966035843,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 63630
    },
    {
      "epoch": 56.364038972542076,
      "grad_norm": 0.24348297715187073,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 63635
    },
    {
      "epoch": 56.36846767050487,
      "grad_norm": 0.2737266719341278,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 63640
    },
    {
      "epoch": 56.37289636846767,
      "grad_norm": 0.278772234916687,
      "learning_rate": 1e-05,
      "loss": 1.0141,
      "step": 63645
    },
    {
      "epoch": 56.37732506643047,
      "grad_norm": 0.29180070757865906,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 63650
    },
    {
      "epoch": 56.38175376439327,
      "grad_norm": 0.2496812492609024,
      "learning_rate": 1e-05,
      "loss": 0.9247,
      "step": 63655
    },
    {
      "epoch": 56.38618246235607,
      "grad_norm": 0.28697994351387024,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 63660
    },
    {
      "epoch": 56.39061116031887,
      "grad_norm": 0.23187129199504852,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 63665
    },
    {
      "epoch": 56.39503985828166,
      "grad_norm": 0.21356260776519775,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 63670
    },
    {
      "epoch": 56.399468556244464,
      "grad_norm": 0.22953714430332184,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 63675
    },
    {
      "epoch": 56.403897254207266,
      "grad_norm": 0.25784069299697876,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 63680
    },
    {
      "epoch": 56.40832595217006,
      "grad_norm": 0.24349458515644073,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 63685
    },
    {
      "epoch": 56.41275465013286,
      "grad_norm": 0.21518701314926147,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 63690
    },
    {
      "epoch": 56.41718334809566,
      "grad_norm": 0.2251671850681305,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 63695
    },
    {
      "epoch": 56.421612046058456,
      "grad_norm": 0.22276322543621063,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 63700
    },
    {
      "epoch": 56.42604074402126,
      "grad_norm": 0.21394184231758118,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 63705
    },
    {
      "epoch": 56.43046944198406,
      "grad_norm": 0.2657020688056946,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 63710
    },
    {
      "epoch": 56.43489813994685,
      "grad_norm": 0.24964112043380737,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 63715
    },
    {
      "epoch": 56.439326837909654,
      "grad_norm": 0.23910826444625854,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 63720
    },
    {
      "epoch": 56.443755535872455,
      "grad_norm": 0.27830642461776733,
      "learning_rate": 1e-05,
      "loss": 1.0319,
      "step": 63725
    },
    {
      "epoch": 56.44818423383525,
      "grad_norm": 0.2774246335029602,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 63730
    },
    {
      "epoch": 56.45261293179805,
      "grad_norm": 0.22147469222545624,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 63735
    },
    {
      "epoch": 56.45704162976085,
      "grad_norm": 0.23240873217582703,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 63740
    },
    {
      "epoch": 56.461470327723646,
      "grad_norm": 0.26253893971443176,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 63745
    },
    {
      "epoch": 56.46589902568645,
      "grad_norm": 0.29025065898895264,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 63750
    },
    {
      "epoch": 56.47032772364925,
      "grad_norm": 0.24803084135055542,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 63755
    },
    {
      "epoch": 56.47475642161205,
      "grad_norm": 0.20899616181850433,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 63760
    },
    {
      "epoch": 56.479185119574844,
      "grad_norm": 0.2042752355337143,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 63765
    },
    {
      "epoch": 56.483613817537645,
      "grad_norm": 0.21896463632583618,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 63770
    },
    {
      "epoch": 56.488042515500446,
      "grad_norm": 0.23999616503715515,
      "learning_rate": 1e-05,
      "loss": 0.9318,
      "step": 63775
    },
    {
      "epoch": 56.49247121346324,
      "grad_norm": 0.22065283358097076,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 63780
    },
    {
      "epoch": 56.49689991142604,
      "grad_norm": 0.22317902743816376,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 63785
    },
    {
      "epoch": 56.50132860938884,
      "grad_norm": 0.20797133445739746,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 63790
    },
    {
      "epoch": 56.50575730735164,
      "grad_norm": 0.22109512984752655,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 63795
    },
    {
      "epoch": 56.51018600531444,
      "grad_norm": 0.24690577387809753,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 63800
    },
    {
      "epoch": 56.51461470327724,
      "grad_norm": 0.26962345838546753,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 63805
    },
    {
      "epoch": 56.51904340124003,
      "grad_norm": 0.25095251202583313,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 63810
    },
    {
      "epoch": 56.523472099202834,
      "grad_norm": 0.24364377558231354,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 63815
    },
    {
      "epoch": 56.527900797165636,
      "grad_norm": 0.25177696347236633,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 63820
    },
    {
      "epoch": 56.53232949512843,
      "grad_norm": 0.26411545276641846,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 63825
    },
    {
      "epoch": 56.53675819309123,
      "grad_norm": 0.22823084890842438,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 63830
    },
    {
      "epoch": 56.54118689105403,
      "grad_norm": 0.22628812491893768,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 63835
    },
    {
      "epoch": 56.545615589016826,
      "grad_norm": 0.25449636578559875,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 63840
    },
    {
      "epoch": 56.55004428697963,
      "grad_norm": 0.21194985508918762,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 63845
    },
    {
      "epoch": 56.55447298494243,
      "grad_norm": 0.25784966349601746,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 63850
    },
    {
      "epoch": 56.55890168290522,
      "grad_norm": 0.2687908113002777,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 63855
    },
    {
      "epoch": 56.563330380868024,
      "grad_norm": 0.23463495075702667,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 63860
    },
    {
      "epoch": 56.567759078830825,
      "grad_norm": 0.3146330416202545,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 63865
    },
    {
      "epoch": 56.57218777679362,
      "grad_norm": 0.24020206928253174,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 63870
    },
    {
      "epoch": 56.57661647475642,
      "grad_norm": 0.25961068272590637,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 63875
    },
    {
      "epoch": 56.58104517271922,
      "grad_norm": 0.2569449245929718,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 63880
    },
    {
      "epoch": 56.58547387068202,
      "grad_norm": 0.2551226019859314,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 63885
    },
    {
      "epoch": 56.58990256864482,
      "grad_norm": 0.2499077320098877,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 63890
    },
    {
      "epoch": 56.59433126660762,
      "grad_norm": 0.2654375731945038,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 63895
    },
    {
      "epoch": 56.59875996457042,
      "grad_norm": 0.21383053064346313,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 63900
    },
    {
      "epoch": 56.60318866253321,
      "grad_norm": 0.2844485938549042,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 63905
    },
    {
      "epoch": 56.607617360496015,
      "grad_norm": 0.20691964030265808,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 63910
    },
    {
      "epoch": 56.612046058458816,
      "grad_norm": 0.2167731076478958,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 63915
    },
    {
      "epoch": 56.61647475642161,
      "grad_norm": 0.23787663877010345,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 63920
    },
    {
      "epoch": 56.62090345438441,
      "grad_norm": 0.3365321159362793,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 63925
    },
    {
      "epoch": 56.62533215234721,
      "grad_norm": 0.20767658948898315,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 63930
    },
    {
      "epoch": 56.62976085031001,
      "grad_norm": 0.21107585728168488,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 63935
    },
    {
      "epoch": 56.63418954827281,
      "grad_norm": 0.20760977268218994,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 63940
    },
    {
      "epoch": 56.63861824623561,
      "grad_norm": 0.21589045226573944,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 63945
    },
    {
      "epoch": 56.6430469441984,
      "grad_norm": 0.23618140816688538,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 63950
    },
    {
      "epoch": 56.647475642161204,
      "grad_norm": 0.23637351393699646,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 63955
    },
    {
      "epoch": 56.651904340124005,
      "grad_norm": 0.23837397992610931,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 63960
    },
    {
      "epoch": 56.6563330380868,
      "grad_norm": 0.2561452388763428,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 63965
    },
    {
      "epoch": 56.6607617360496,
      "grad_norm": 0.3384799361228943,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 63970
    },
    {
      "epoch": 56.6651904340124,
      "grad_norm": 0.23272429406642914,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 63975
    },
    {
      "epoch": 56.669619131975196,
      "grad_norm": 0.23863771557807922,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 63980
    },
    {
      "epoch": 56.674047829938,
      "grad_norm": 0.23360808193683624,
      "learning_rate": 1e-05,
      "loss": 0.9018,
      "step": 63985
    },
    {
      "epoch": 56.6784765279008,
      "grad_norm": 0.24241039156913757,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 63990
    },
    {
      "epoch": 56.68290522586359,
      "grad_norm": 0.31063148379325867,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 63995
    },
    {
      "epoch": 56.687333923826394,
      "grad_norm": 0.23626072704792023,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 64000
    },
    {
      "epoch": 56.691762621789195,
      "grad_norm": 0.23614118993282318,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 64005
    },
    {
      "epoch": 56.696191319751996,
      "grad_norm": 0.20562206208705902,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 64010
    },
    {
      "epoch": 56.70062001771479,
      "grad_norm": 0.19414405524730682,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 64015
    },
    {
      "epoch": 56.70504871567759,
      "grad_norm": 0.2645584046840668,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 64020
    },
    {
      "epoch": 56.70947741364039,
      "grad_norm": 0.19743388891220093,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 64025
    },
    {
      "epoch": 56.71390611160319,
      "grad_norm": 0.22869670391082764,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 64030
    },
    {
      "epoch": 56.71833480956599,
      "grad_norm": 0.2251833975315094,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 64035
    },
    {
      "epoch": 56.72276350752879,
      "grad_norm": 0.22205980122089386,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 64040
    },
    {
      "epoch": 56.72719220549158,
      "grad_norm": 0.2329990565776825,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 64045
    },
    {
      "epoch": 56.731620903454385,
      "grad_norm": 0.22200244665145874,
      "learning_rate": 1e-05,
      "loss": 0.8899,
      "step": 64050
    },
    {
      "epoch": 56.736049601417186,
      "grad_norm": 0.25576964020729065,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 64055
    },
    {
      "epoch": 56.74047829937998,
      "grad_norm": 0.24480198323726654,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 64060
    },
    {
      "epoch": 56.74490699734278,
      "grad_norm": 0.2794451117515564,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 64065
    },
    {
      "epoch": 56.74933569530558,
      "grad_norm": 0.23102684319019318,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 64070
    },
    {
      "epoch": 56.753764393268376,
      "grad_norm": 0.25073865056037903,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 64075
    },
    {
      "epoch": 56.75819309123118,
      "grad_norm": 0.22945536673069,
      "learning_rate": 1e-05,
      "loss": 0.9229,
      "step": 64080
    },
    {
      "epoch": 56.76262178919398,
      "grad_norm": 0.2440461367368698,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 64085
    },
    {
      "epoch": 56.76705048715677,
      "grad_norm": 0.2708947956562042,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 64090
    },
    {
      "epoch": 56.771479185119574,
      "grad_norm": 0.23065300285816193,
      "learning_rate": 1e-05,
      "loss": 0.9248,
      "step": 64095
    },
    {
      "epoch": 56.775907883082375,
      "grad_norm": 0.24007974565029144,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 64100
    },
    {
      "epoch": 56.78033658104517,
      "grad_norm": 0.22570888698101044,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 64105
    },
    {
      "epoch": 56.78476527900797,
      "grad_norm": 0.24484147131443024,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 64110
    },
    {
      "epoch": 56.78919397697077,
      "grad_norm": 0.2136494517326355,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 64115
    },
    {
      "epoch": 56.79362267493357,
      "grad_norm": 0.25291118025779724,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 64120
    },
    {
      "epoch": 56.79805137289637,
      "grad_norm": 0.2680213451385498,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 64125
    },
    {
      "epoch": 56.80248007085917,
      "grad_norm": 0.24497048556804657,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 64130
    },
    {
      "epoch": 56.80690876882197,
      "grad_norm": 0.20482584834098816,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 64135
    },
    {
      "epoch": 56.811337466784764,
      "grad_norm": 0.2691558599472046,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 64140
    },
    {
      "epoch": 56.815766164747565,
      "grad_norm": 0.29624661803245544,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 64145
    },
    {
      "epoch": 56.820194862710366,
      "grad_norm": 0.26249465346336365,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 64150
    },
    {
      "epoch": 56.82462356067316,
      "grad_norm": 0.2631811499595642,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 64155
    },
    {
      "epoch": 56.82905225863596,
      "grad_norm": 0.20773299038410187,
      "learning_rate": 1e-05,
      "loss": 0.9157,
      "step": 64160
    },
    {
      "epoch": 56.83348095659876,
      "grad_norm": 0.24310153722763062,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 64165
    },
    {
      "epoch": 56.83790965456156,
      "grad_norm": 0.26155924797058105,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 64170
    },
    {
      "epoch": 56.84233835252436,
      "grad_norm": 0.2059398740530014,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 64175
    },
    {
      "epoch": 56.84676705048716,
      "grad_norm": 0.19910509884357452,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 64180
    },
    {
      "epoch": 56.85119574844995,
      "grad_norm": 0.24665945768356323,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 64185
    },
    {
      "epoch": 56.855624446412754,
      "grad_norm": 0.23373277485370636,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 64190
    },
    {
      "epoch": 56.860053144375556,
      "grad_norm": 0.18853700160980225,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 64195
    },
    {
      "epoch": 56.86448184233835,
      "grad_norm": 0.29547590017318726,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 64200
    },
    {
      "epoch": 56.86891054030115,
      "grad_norm": 0.2515179216861725,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 64205
    },
    {
      "epoch": 56.87333923826395,
      "grad_norm": 0.23519451916217804,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 64210
    },
    {
      "epoch": 56.877767936226746,
      "grad_norm": 0.2409517765045166,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 64215
    },
    {
      "epoch": 56.88219663418955,
      "grad_norm": 0.26572102308273315,
      "learning_rate": 1e-05,
      "loss": 0.9427,
      "step": 64220
    },
    {
      "epoch": 56.88662533215235,
      "grad_norm": 0.29910725355148315,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 64225
    },
    {
      "epoch": 56.89105403011514,
      "grad_norm": 0.2584010362625122,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 64230
    },
    {
      "epoch": 56.895482728077944,
      "grad_norm": 0.20503230392932892,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 64235
    },
    {
      "epoch": 56.899911426040745,
      "grad_norm": 0.22223049402236938,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 64240
    },
    {
      "epoch": 56.90434012400354,
      "grad_norm": 0.23342706263065338,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 64245
    },
    {
      "epoch": 56.90876882196634,
      "grad_norm": 0.2384214848279953,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 64250
    },
    {
      "epoch": 56.91319751992914,
      "grad_norm": 0.27266746759414673,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 64255
    },
    {
      "epoch": 56.91762621789194,
      "grad_norm": 0.2111235111951828,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 64260
    },
    {
      "epoch": 56.92205491585474,
      "grad_norm": 0.23333798348903656,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 64265
    },
    {
      "epoch": 56.92648361381754,
      "grad_norm": 0.2777155935764313,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 64270
    },
    {
      "epoch": 56.93091231178034,
      "grad_norm": 0.22184447944164276,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 64275
    },
    {
      "epoch": 56.935341009743134,
      "grad_norm": 0.23562589287757874,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 64280
    },
    {
      "epoch": 56.939769707705935,
      "grad_norm": 0.23291298747062683,
      "learning_rate": 1e-05,
      "loss": 0.8889,
      "step": 64285
    },
    {
      "epoch": 56.944198405668736,
      "grad_norm": 0.226912721991539,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 64290
    },
    {
      "epoch": 56.94862710363153,
      "grad_norm": 0.2113107591867447,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 64295
    },
    {
      "epoch": 56.95305580159433,
      "grad_norm": 0.2539554536342621,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 64300
    },
    {
      "epoch": 56.95748449955713,
      "grad_norm": 0.2438468486070633,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 64305
    },
    {
      "epoch": 56.96191319751993,
      "grad_norm": 0.25723347067832947,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 64310
    },
    {
      "epoch": 56.96634189548273,
      "grad_norm": 0.28310999274253845,
      "learning_rate": 1e-05,
      "loss": 0.9325,
      "step": 64315
    },
    {
      "epoch": 56.97077059344553,
      "grad_norm": 0.2382184863090515,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 64320
    },
    {
      "epoch": 56.97519929140832,
      "grad_norm": 0.2212591916322708,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 64325
    },
    {
      "epoch": 56.979627989371124,
      "grad_norm": 0.23368185758590698,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 64330
    },
    {
      "epoch": 56.984056687333926,
      "grad_norm": 0.22670021653175354,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 64335
    },
    {
      "epoch": 56.98848538529672,
      "grad_norm": 0.2316238433122635,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 64340
    },
    {
      "epoch": 56.99291408325952,
      "grad_norm": 0.22660046815872192,
      "learning_rate": 1e-05,
      "loss": 1.028,
      "step": 64345
    },
    {
      "epoch": 56.99734278122232,
      "grad_norm": 0.2520727515220642,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 64350
    },
    {
      "epoch": 57.001771479185116,
      "grad_norm": 0.2616822421550751,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 64355
    },
    {
      "epoch": 57.00620017714792,
      "grad_norm": 0.22308708727359772,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 64360
    },
    {
      "epoch": 57.01062887511072,
      "grad_norm": 0.2275661826133728,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 64365
    },
    {
      "epoch": 57.01505757307352,
      "grad_norm": 0.2338424175977707,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 64370
    },
    {
      "epoch": 57.019486271036314,
      "grad_norm": 0.22162552177906036,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 64375
    },
    {
      "epoch": 57.023914968999115,
      "grad_norm": 0.23149742186069489,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 64380
    },
    {
      "epoch": 57.028343666961916,
      "grad_norm": 0.2888084650039673,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 64385
    },
    {
      "epoch": 57.03277236492471,
      "grad_norm": 0.21528057754039764,
      "learning_rate": 1e-05,
      "loss": 0.9103,
      "step": 64390
    },
    {
      "epoch": 57.03720106288751,
      "grad_norm": 0.27679142355918884,
      "learning_rate": 1e-05,
      "loss": 1.0257,
      "step": 64395
    },
    {
      "epoch": 57.04162976085031,
      "grad_norm": 0.2184085249900818,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 64400
    },
    {
      "epoch": 57.04605845881311,
      "grad_norm": 0.20083005726337433,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 64405
    },
    {
      "epoch": 57.05048715677591,
      "grad_norm": 0.23817558586597443,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 64410
    },
    {
      "epoch": 57.05491585473871,
      "grad_norm": 0.2634621262550354,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 64415
    },
    {
      "epoch": 57.0593445527015,
      "grad_norm": 0.24302643537521362,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 64420
    },
    {
      "epoch": 57.063773250664305,
      "grad_norm": 0.2861745357513428,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 64425
    },
    {
      "epoch": 57.068201948627106,
      "grad_norm": 0.2536584734916687,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 64430
    },
    {
      "epoch": 57.0726306465899,
      "grad_norm": 0.2796446681022644,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 64435
    },
    {
      "epoch": 57.0770593445527,
      "grad_norm": 0.2579536736011505,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 64440
    },
    {
      "epoch": 57.0814880425155,
      "grad_norm": 0.24723190069198608,
      "learning_rate": 1e-05,
      "loss": 0.9007,
      "step": 64445
    },
    {
      "epoch": 57.0859167404783,
      "grad_norm": 0.22301775217056274,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 64450
    },
    {
      "epoch": 57.0903454384411,
      "grad_norm": 0.2747363746166229,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 64455
    },
    {
      "epoch": 57.0947741364039,
      "grad_norm": 0.23562118411064148,
      "learning_rate": 1e-05,
      "loss": 0.9201,
      "step": 64460
    },
    {
      "epoch": 57.09920283436669,
      "grad_norm": 0.24335841834545135,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 64465
    },
    {
      "epoch": 57.103631532329494,
      "grad_norm": 0.23442089557647705,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 64470
    },
    {
      "epoch": 57.108060230292296,
      "grad_norm": 0.22402559220790863,
      "learning_rate": 1e-05,
      "loss": 0.936,
      "step": 64475
    },
    {
      "epoch": 57.11248892825509,
      "grad_norm": 0.25184866786003113,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 64480
    },
    {
      "epoch": 57.11691762621789,
      "grad_norm": 0.23988264799118042,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 64485
    },
    {
      "epoch": 57.12134632418069,
      "grad_norm": 0.2549890875816345,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 64490
    },
    {
      "epoch": 57.12577502214349,
      "grad_norm": 0.23983615636825562,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 64495
    },
    {
      "epoch": 57.13020372010629,
      "grad_norm": 0.2442067563533783,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 64500
    },
    {
      "epoch": 57.13463241806909,
      "grad_norm": 0.28639140725135803,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 64505
    },
    {
      "epoch": 57.13906111603189,
      "grad_norm": 0.2667704224586487,
      "learning_rate": 1e-05,
      "loss": 1.0331,
      "step": 64510
    },
    {
      "epoch": 57.143489813994684,
      "grad_norm": 0.2432333528995514,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 64515
    },
    {
      "epoch": 57.147918511957485,
      "grad_norm": 0.2359638810157776,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 64520
    },
    {
      "epoch": 57.152347209920286,
      "grad_norm": 0.20179174840450287,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 64525
    },
    {
      "epoch": 57.15677590788308,
      "grad_norm": 0.23003005981445312,
      "learning_rate": 1e-05,
      "loss": 0.915,
      "step": 64530
    },
    {
      "epoch": 57.16120460584588,
      "grad_norm": 0.2588357925415039,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 64535
    },
    {
      "epoch": 57.16563330380868,
      "grad_norm": 0.2229629009962082,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 64540
    },
    {
      "epoch": 57.17006200177148,
      "grad_norm": 0.2676277160644531,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 64545
    },
    {
      "epoch": 57.17449069973428,
      "grad_norm": 0.23335818946361542,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 64550
    },
    {
      "epoch": 57.17891939769708,
      "grad_norm": 0.25431424379348755,
      "learning_rate": 1e-05,
      "loss": 0.9223,
      "step": 64555
    },
    {
      "epoch": 57.18334809565987,
      "grad_norm": 0.2280757576227188,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 64560
    },
    {
      "epoch": 57.187776793622675,
      "grad_norm": 0.2610466778278351,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 64565
    },
    {
      "epoch": 57.192205491585476,
      "grad_norm": 0.22489522397518158,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 64570
    },
    {
      "epoch": 57.19663418954827,
      "grad_norm": 0.22267022728919983,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 64575
    },
    {
      "epoch": 57.20106288751107,
      "grad_norm": 0.25446072220802307,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 64580
    },
    {
      "epoch": 57.20549158547387,
      "grad_norm": 0.2352152019739151,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 64585
    },
    {
      "epoch": 57.20992028343667,
      "grad_norm": 0.25363969802856445,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 64590
    },
    {
      "epoch": 57.21434898139947,
      "grad_norm": 0.26549476385116577,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 64595
    },
    {
      "epoch": 57.21877767936227,
      "grad_norm": 0.2486758977174759,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 64600
    },
    {
      "epoch": 57.22320637732506,
      "grad_norm": 0.26905712485313416,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 64605
    },
    {
      "epoch": 57.227635075287864,
      "grad_norm": 0.2641041576862335,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 64610
    },
    {
      "epoch": 57.232063773250665,
      "grad_norm": 0.2319459766149521,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 64615
    },
    {
      "epoch": 57.23649247121347,
      "grad_norm": 0.24442654848098755,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 64620
    },
    {
      "epoch": 57.24092116917626,
      "grad_norm": 0.21868108212947845,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 64625
    },
    {
      "epoch": 57.24534986713906,
      "grad_norm": 0.22921469807624817,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 64630
    },
    {
      "epoch": 57.24977856510186,
      "grad_norm": 0.27202725410461426,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 64635
    },
    {
      "epoch": 57.25420726306466,
      "grad_norm": 0.22790783643722534,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 64640
    },
    {
      "epoch": 57.25863596102746,
      "grad_norm": 0.22276154160499573,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 64645
    },
    {
      "epoch": 57.26306465899026,
      "grad_norm": 0.24896281957626343,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 64650
    },
    {
      "epoch": 57.267493356953054,
      "grad_norm": 0.2236364632844925,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 64655
    },
    {
      "epoch": 57.271922054915855,
      "grad_norm": 0.22617529332637787,
      "learning_rate": 1e-05,
      "loss": 0.9242,
      "step": 64660
    },
    {
      "epoch": 57.276350752878656,
      "grad_norm": 0.24415220320224762,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 64665
    },
    {
      "epoch": 57.28077945084145,
      "grad_norm": 0.23086804151535034,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 64670
    },
    {
      "epoch": 57.28520814880425,
      "grad_norm": 0.27890124917030334,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 64675
    },
    {
      "epoch": 57.28963684676705,
      "grad_norm": 0.26853156089782715,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 64680
    },
    {
      "epoch": 57.29406554472985,
      "grad_norm": 0.23927192389965057,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 64685
    },
    {
      "epoch": 57.29849424269265,
      "grad_norm": 0.21735401451587677,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 64690
    },
    {
      "epoch": 57.30292294065545,
      "grad_norm": 0.25924548506736755,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 64695
    },
    {
      "epoch": 57.30735163861824,
      "grad_norm": 0.24101030826568604,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 64700
    },
    {
      "epoch": 57.311780336581045,
      "grad_norm": 0.2158128321170807,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 64705
    },
    {
      "epoch": 57.316209034543846,
      "grad_norm": 0.2520535886287689,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 64710
    },
    {
      "epoch": 57.32063773250664,
      "grad_norm": 0.23017117381095886,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 64715
    },
    {
      "epoch": 57.32506643046944,
      "grad_norm": 0.2008381485939026,
      "learning_rate": 1e-05,
      "loss": 1.0187,
      "step": 64720
    },
    {
      "epoch": 57.32949512843224,
      "grad_norm": 0.22195418179035187,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 64725
    },
    {
      "epoch": 57.333923826395036,
      "grad_norm": 0.25837013125419617,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 64730
    },
    {
      "epoch": 57.33835252435784,
      "grad_norm": 0.20500344038009644,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 64735
    },
    {
      "epoch": 57.34278122232064,
      "grad_norm": 0.2097352147102356,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 64740
    },
    {
      "epoch": 57.34720992028344,
      "grad_norm": 0.20659856498241425,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 64745
    },
    {
      "epoch": 57.351638618246234,
      "grad_norm": 0.23098255693912506,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 64750
    },
    {
      "epoch": 57.356067316209035,
      "grad_norm": 0.2713516652584076,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 64755
    },
    {
      "epoch": 57.36049601417184,
      "grad_norm": 0.2024286836385727,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 64760
    },
    {
      "epoch": 57.36492471213463,
      "grad_norm": 0.24792177975177765,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 64765
    },
    {
      "epoch": 57.36935341009743,
      "grad_norm": 0.21796000003814697,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 64770
    },
    {
      "epoch": 57.37378210806023,
      "grad_norm": 0.2694271504878998,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 64775
    },
    {
      "epoch": 57.37821080602303,
      "grad_norm": 0.22968241572380066,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 64780
    },
    {
      "epoch": 57.38263950398583,
      "grad_norm": 0.19997096061706543,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 64785
    },
    {
      "epoch": 57.38706820194863,
      "grad_norm": 0.23471786081790924,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 64790
    },
    {
      "epoch": 57.391496899911424,
      "grad_norm": 0.20355014503002167,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 64795
    },
    {
      "epoch": 57.395925597874225,
      "grad_norm": 0.22634902596473694,
      "learning_rate": 1e-05,
      "loss": 0.9263,
      "step": 64800
    },
    {
      "epoch": 57.400354295837026,
      "grad_norm": 0.20699720084667206,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 64805
    },
    {
      "epoch": 57.40478299379982,
      "grad_norm": 0.21015790104866028,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 64810
    },
    {
      "epoch": 57.40921169176262,
      "grad_norm": 0.23895959556102753,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 64815
    },
    {
      "epoch": 57.41364038972542,
      "grad_norm": 0.2263335883617401,
      "learning_rate": 1e-05,
      "loss": 0.9251,
      "step": 64820
    },
    {
      "epoch": 57.41806908768822,
      "grad_norm": 0.23941682279109955,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 64825
    },
    {
      "epoch": 57.42249778565102,
      "grad_norm": 0.21777811646461487,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 64830
    },
    {
      "epoch": 57.42692648361382,
      "grad_norm": 0.205412358045578,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 64835
    },
    {
      "epoch": 57.43135518157661,
      "grad_norm": 0.24491752684116364,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 64840
    },
    {
      "epoch": 57.435783879539414,
      "grad_norm": 0.24434705078601837,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 64845
    },
    {
      "epoch": 57.440212577502216,
      "grad_norm": 0.20678195357322693,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 64850
    },
    {
      "epoch": 57.44464127546502,
      "grad_norm": 0.22470834851264954,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 64855
    },
    {
      "epoch": 57.44906997342781,
      "grad_norm": 0.22402292490005493,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 64860
    },
    {
      "epoch": 57.45349867139061,
      "grad_norm": 0.25211504101753235,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 64865
    },
    {
      "epoch": 57.45792736935341,
      "grad_norm": 0.22374984622001648,
      "learning_rate": 1e-05,
      "loss": 0.9236,
      "step": 64870
    },
    {
      "epoch": 57.46235606731621,
      "grad_norm": 0.2464490383863449,
      "learning_rate": 1e-05,
      "loss": 1.0107,
      "step": 64875
    },
    {
      "epoch": 57.46678476527901,
      "grad_norm": 0.23377221822738647,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 64880
    },
    {
      "epoch": 57.47121346324181,
      "grad_norm": 0.21537736058235168,
      "learning_rate": 1e-05,
      "loss": 1.0498,
      "step": 64885
    },
    {
      "epoch": 57.475642161204604,
      "grad_norm": 0.23741796612739563,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 64890
    },
    {
      "epoch": 57.480070859167405,
      "grad_norm": 0.21205613017082214,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 64895
    },
    {
      "epoch": 57.484499557130206,
      "grad_norm": 0.2406456619501114,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 64900
    },
    {
      "epoch": 57.488928255093,
      "grad_norm": 0.24922002851963043,
      "learning_rate": 1e-05,
      "loss": 1.0193,
      "step": 64905
    },
    {
      "epoch": 57.4933569530558,
      "grad_norm": 0.22549712657928467,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 64910
    },
    {
      "epoch": 57.4977856510186,
      "grad_norm": 0.24909362196922302,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 64915
    },
    {
      "epoch": 57.5022143489814,
      "grad_norm": 0.3197036385536194,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 64920
    },
    {
      "epoch": 57.5066430469442,
      "grad_norm": 0.2692038118839264,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 64925
    },
    {
      "epoch": 57.511071744907,
      "grad_norm": 0.25321099162101746,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 64930
    },
    {
      "epoch": 57.515500442869794,
      "grad_norm": 0.22069984674453735,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 64935
    },
    {
      "epoch": 57.519929140832595,
      "grad_norm": 0.23525096476078033,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 64940
    },
    {
      "epoch": 57.524357838795396,
      "grad_norm": 0.2202586978673935,
      "learning_rate": 1e-05,
      "loss": 0.9147,
      "step": 64945
    },
    {
      "epoch": 57.52878653675819,
      "grad_norm": 0.2157680094242096,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 64950
    },
    {
      "epoch": 57.53321523472099,
      "grad_norm": 0.3081792891025543,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 64955
    },
    {
      "epoch": 57.53764393268379,
      "grad_norm": 0.2438458502292633,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 64960
    },
    {
      "epoch": 57.54207263064659,
      "grad_norm": 0.23848648369312286,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 64965
    },
    {
      "epoch": 57.54650132860939,
      "grad_norm": 0.24289342761039734,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 64970
    },
    {
      "epoch": 57.55093002657219,
      "grad_norm": 0.21775168180465698,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 64975
    },
    {
      "epoch": 57.55535872453498,
      "grad_norm": 0.25212976336479187,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 64980
    },
    {
      "epoch": 57.559787422497784,
      "grad_norm": 0.23616497218608856,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 64985
    },
    {
      "epoch": 57.564216120460586,
      "grad_norm": 0.24934092164039612,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 64990
    },
    {
      "epoch": 57.56864481842339,
      "grad_norm": 0.23486179113388062,
      "learning_rate": 1e-05,
      "loss": 1.0593,
      "step": 64995
    },
    {
      "epoch": 57.57307351638618,
      "grad_norm": 0.21842266619205475,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 65000
    },
    {
      "epoch": 57.57750221434898,
      "grad_norm": 0.2404446005821228,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 65005
    },
    {
      "epoch": 57.58193091231178,
      "grad_norm": 0.2785521149635315,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 65010
    },
    {
      "epoch": 57.58635961027458,
      "grad_norm": 0.270435631275177,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 65015
    },
    {
      "epoch": 57.59078830823738,
      "grad_norm": 0.29943397641181946,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 65020
    },
    {
      "epoch": 57.59521700620018,
      "grad_norm": 0.23091958463191986,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 65025
    },
    {
      "epoch": 57.599645704162974,
      "grad_norm": 0.2477135956287384,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 65030
    },
    {
      "epoch": 57.604074402125775,
      "grad_norm": 0.2208290547132492,
      "learning_rate": 1e-05,
      "loss": 1.0232,
      "step": 65035
    },
    {
      "epoch": 57.608503100088576,
      "grad_norm": 0.2288370430469513,
      "learning_rate": 1e-05,
      "loss": 0.9299,
      "step": 65040
    },
    {
      "epoch": 57.61293179805137,
      "grad_norm": 0.22634239494800568,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 65045
    },
    {
      "epoch": 57.61736049601417,
      "grad_norm": 0.25695568323135376,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 65050
    },
    {
      "epoch": 57.62178919397697,
      "grad_norm": 0.1895153820514679,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 65055
    },
    {
      "epoch": 57.62621789193977,
      "grad_norm": 0.24530749022960663,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 65060
    },
    {
      "epoch": 57.63064658990257,
      "grad_norm": 0.2582763731479645,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 65065
    },
    {
      "epoch": 57.63507528786537,
      "grad_norm": 0.2380317747592926,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 65070
    },
    {
      "epoch": 57.63950398582816,
      "grad_norm": 0.19777148962020874,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 65075
    },
    {
      "epoch": 57.643932683790965,
      "grad_norm": 0.31268760561943054,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 65080
    },
    {
      "epoch": 57.648361381753766,
      "grad_norm": 0.2274935096502304,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 65085
    },
    {
      "epoch": 57.65279007971656,
      "grad_norm": 0.22767941653728485,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 65090
    },
    {
      "epoch": 57.65721877767936,
      "grad_norm": 0.2537647485733032,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 65095
    },
    {
      "epoch": 57.66164747564216,
      "grad_norm": 0.27609172463417053,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 65100
    },
    {
      "epoch": 57.666076173604964,
      "grad_norm": 0.2544642388820648,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 65105
    },
    {
      "epoch": 57.67050487156776,
      "grad_norm": 0.2952786386013031,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 65110
    },
    {
      "epoch": 57.67493356953056,
      "grad_norm": 0.22622506320476532,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 65115
    },
    {
      "epoch": 57.67936226749336,
      "grad_norm": 0.22703221440315247,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 65120
    },
    {
      "epoch": 57.683790965456154,
      "grad_norm": 0.2238233983516693,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 65125
    },
    {
      "epoch": 57.688219663418955,
      "grad_norm": 0.24899868667125702,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 65130
    },
    {
      "epoch": 57.69264836138176,
      "grad_norm": 0.25768086314201355,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 65135
    },
    {
      "epoch": 57.69707705934455,
      "grad_norm": 0.2340216338634491,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 65140
    },
    {
      "epoch": 57.70150575730735,
      "grad_norm": 0.26033055782318115,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 65145
    },
    {
      "epoch": 57.70593445527015,
      "grad_norm": 0.23662853240966797,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 65150
    },
    {
      "epoch": 57.71036315323295,
      "grad_norm": 0.27634745836257935,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 65155
    },
    {
      "epoch": 57.71479185119575,
      "grad_norm": 0.24156300723552704,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 65160
    },
    {
      "epoch": 57.71922054915855,
      "grad_norm": 0.20061972737312317,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 65165
    },
    {
      "epoch": 57.723649247121344,
      "grad_norm": 0.24070684611797333,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 65170
    },
    {
      "epoch": 57.728077945084145,
      "grad_norm": 0.2417374700307846,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 65175
    },
    {
      "epoch": 57.732506643046946,
      "grad_norm": 0.2408277839422226,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 65180
    },
    {
      "epoch": 57.73693534100974,
      "grad_norm": 0.2178143411874771,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 65185
    },
    {
      "epoch": 57.74136403897254,
      "grad_norm": 0.2403598129749298,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 65190
    },
    {
      "epoch": 57.74579273693534,
      "grad_norm": 0.26669999957084656,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 65195
    },
    {
      "epoch": 57.75022143489814,
      "grad_norm": 0.2620389461517334,
      "learning_rate": 1e-05,
      "loss": 0.9188,
      "step": 65200
    },
    {
      "epoch": 57.75465013286094,
      "grad_norm": 0.23346661031246185,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 65205
    },
    {
      "epoch": 57.75907883082374,
      "grad_norm": 0.21242472529411316,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 65210
    },
    {
      "epoch": 57.76350752878653,
      "grad_norm": 0.25294920802116394,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 65215
    },
    {
      "epoch": 57.767936226749335,
      "grad_norm": 0.23865795135498047,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 65220
    },
    {
      "epoch": 57.772364924712136,
      "grad_norm": 0.2392657846212387,
      "learning_rate": 1e-05,
      "loss": 0.9264,
      "step": 65225
    },
    {
      "epoch": 57.77679362267494,
      "grad_norm": 0.2450830340385437,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 65230
    },
    {
      "epoch": 57.78122232063773,
      "grad_norm": 0.2542381286621094,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 65235
    },
    {
      "epoch": 57.78565101860053,
      "grad_norm": 0.21462008357048035,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 65240
    },
    {
      "epoch": 57.79007971656333,
      "grad_norm": 0.24379025399684906,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 65245
    },
    {
      "epoch": 57.79450841452613,
      "grad_norm": 0.25722989439964294,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 65250
    },
    {
      "epoch": 57.79893711248893,
      "grad_norm": 0.2545693516731262,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 65255
    },
    {
      "epoch": 57.80336581045173,
      "grad_norm": 0.22995683550834656,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 65260
    },
    {
      "epoch": 57.807794508414524,
      "grad_norm": 0.29748353362083435,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 65265
    },
    {
      "epoch": 57.812223206377325,
      "grad_norm": 0.24034743010997772,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 65270
    },
    {
      "epoch": 57.81665190434013,
      "grad_norm": 0.23386414349079132,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 65275
    },
    {
      "epoch": 57.82108060230292,
      "grad_norm": 0.2299608737230301,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 65280
    },
    {
      "epoch": 57.82550930026572,
      "grad_norm": 0.23577432334423065,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 65285
    },
    {
      "epoch": 57.82993799822852,
      "grad_norm": 0.2799549996852875,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 65290
    },
    {
      "epoch": 57.83436669619132,
      "grad_norm": 0.22159375250339508,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 65295
    },
    {
      "epoch": 57.83879539415412,
      "grad_norm": 0.2742268443107605,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 65300
    },
    {
      "epoch": 57.84322409211692,
      "grad_norm": 0.2504993975162506,
      "learning_rate": 1e-05,
      "loss": 0.9155,
      "step": 65305
    },
    {
      "epoch": 57.847652790079714,
      "grad_norm": 0.2482503354549408,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 65310
    },
    {
      "epoch": 57.852081488042515,
      "grad_norm": 0.21859294176101685,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 65315
    },
    {
      "epoch": 57.856510186005316,
      "grad_norm": 0.23677146434783936,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 65320
    },
    {
      "epoch": 57.86093888396811,
      "grad_norm": 0.25044289231300354,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 65325
    },
    {
      "epoch": 57.86536758193091,
      "grad_norm": 0.1946915239095688,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 65330
    },
    {
      "epoch": 57.86979627989371,
      "grad_norm": 0.20747967064380646,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 65335
    },
    {
      "epoch": 57.87422497785651,
      "grad_norm": 0.25214090943336487,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 65340
    },
    {
      "epoch": 57.87865367581931,
      "grad_norm": 0.23532888293266296,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 65345
    },
    {
      "epoch": 57.88308237378211,
      "grad_norm": 0.25461849570274353,
      "learning_rate": 1e-05,
      "loss": 1.0009,
      "step": 65350
    },
    {
      "epoch": 57.88751107174491,
      "grad_norm": 0.25920936465263367,
      "learning_rate": 1e-05,
      "loss": 1.0195,
      "step": 65355
    },
    {
      "epoch": 57.891939769707704,
      "grad_norm": 0.21980534493923187,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 65360
    },
    {
      "epoch": 57.896368467670506,
      "grad_norm": 0.286761999130249,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 65365
    },
    {
      "epoch": 57.90079716563331,
      "grad_norm": 0.2770629823207855,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 65370
    },
    {
      "epoch": 57.9052258635961,
      "grad_norm": 0.2169332653284073,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 65375
    },
    {
      "epoch": 57.9096545615589,
      "grad_norm": 0.2575649321079254,
      "learning_rate": 1e-05,
      "loss": 1.0107,
      "step": 65380
    },
    {
      "epoch": 57.9140832595217,
      "grad_norm": 0.2296285629272461,
      "learning_rate": 1e-05,
      "loss": 0.9255,
      "step": 65385
    },
    {
      "epoch": 57.9185119574845,
      "grad_norm": 0.2089402675628662,
      "learning_rate": 1e-05,
      "loss": 1.0392,
      "step": 65390
    },
    {
      "epoch": 57.9229406554473,
      "grad_norm": 0.2604351043701172,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 65395
    },
    {
      "epoch": 57.9273693534101,
      "grad_norm": 0.26109015941619873,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 65400
    },
    {
      "epoch": 57.931798051372894,
      "grad_norm": 0.268934041261673,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 65405
    },
    {
      "epoch": 57.936226749335695,
      "grad_norm": 0.22601456940174103,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 65410
    },
    {
      "epoch": 57.9406554472985,
      "grad_norm": 0.2878456115722656,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 65415
    },
    {
      "epoch": 57.94508414526129,
      "grad_norm": 0.22090114653110504,
      "learning_rate": 1e-05,
      "loss": 0.9312,
      "step": 65420
    },
    {
      "epoch": 57.94951284322409,
      "grad_norm": 0.2888115346431732,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 65425
    },
    {
      "epoch": 57.95394154118689,
      "grad_norm": 0.39820554852485657,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 65430
    },
    {
      "epoch": 57.95837023914969,
      "grad_norm": 0.24735014140605927,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 65435
    },
    {
      "epoch": 57.96279893711249,
      "grad_norm": 0.21726997196674347,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 65440
    },
    {
      "epoch": 57.96722763507529,
      "grad_norm": 0.23674222826957703,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 65445
    },
    {
      "epoch": 57.971656333038084,
      "grad_norm": 0.23199085891246796,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 65450
    },
    {
      "epoch": 57.976085031000885,
      "grad_norm": 0.23483982682228088,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 65455
    },
    {
      "epoch": 57.980513728963686,
      "grad_norm": 0.23024198412895203,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 65460
    },
    {
      "epoch": 57.98494242692648,
      "grad_norm": 0.22566211223602295,
      "learning_rate": 1e-05,
      "loss": 0.9237,
      "step": 65465
    },
    {
      "epoch": 57.98937112488928,
      "grad_norm": 0.2305525541305542,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 65470
    },
    {
      "epoch": 57.99379982285208,
      "grad_norm": 0.2151440531015396,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 65475
    },
    {
      "epoch": 57.998228520814884,
      "grad_norm": 0.21990051865577698,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 65480
    },
    {
      "epoch": 58.00265721877768,
      "grad_norm": 0.3026951551437378,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 65485
    },
    {
      "epoch": 58.00708591674048,
      "grad_norm": 0.1917659491300583,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 65490
    },
    {
      "epoch": 58.01151461470328,
      "grad_norm": 0.23054540157318115,
      "learning_rate": 1e-05,
      "loss": 1.0412,
      "step": 65495
    },
    {
      "epoch": 58.015943312666074,
      "grad_norm": 0.20408916473388672,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 65500
    },
    {
      "epoch": 58.020372010628876,
      "grad_norm": 0.23431792855262756,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 65505
    },
    {
      "epoch": 58.02480070859168,
      "grad_norm": 0.2742765247821808,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 65510
    },
    {
      "epoch": 58.02922940655447,
      "grad_norm": 0.22317524254322052,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 65515
    },
    {
      "epoch": 58.03365810451727,
      "grad_norm": 0.2639908492565155,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 65520
    },
    {
      "epoch": 58.03808680248007,
      "grad_norm": 0.22771699726581573,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 65525
    },
    {
      "epoch": 58.04251550044287,
      "grad_norm": 0.2904604375362396,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 65530
    },
    {
      "epoch": 58.04694419840567,
      "grad_norm": 0.25606799125671387,
      "learning_rate": 1e-05,
      "loss": 0.9212,
      "step": 65535
    },
    {
      "epoch": 58.05137289636847,
      "grad_norm": 0.22992970049381256,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 65540
    },
    {
      "epoch": 58.055801594331264,
      "grad_norm": 0.26750385761260986,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 65545
    },
    {
      "epoch": 58.060230292294065,
      "grad_norm": 0.23747022449970245,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 65550
    },
    {
      "epoch": 58.064658990256866,
      "grad_norm": 0.24407124519348145,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 65555
    },
    {
      "epoch": 58.06908768821966,
      "grad_norm": 0.26146432757377625,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 65560
    },
    {
      "epoch": 58.07351638618246,
      "grad_norm": 0.2496596872806549,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 65565
    },
    {
      "epoch": 58.07794508414526,
      "grad_norm": 0.22963643074035645,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 65570
    },
    {
      "epoch": 58.08237378210806,
      "grad_norm": 0.24085082113742828,
      "learning_rate": 1e-05,
      "loss": 0.9493,
      "step": 65575
    },
    {
      "epoch": 58.08680248007086,
      "grad_norm": 0.2758622169494629,
      "learning_rate": 1e-05,
      "loss": 0.9232,
      "step": 65580
    },
    {
      "epoch": 58.09123117803366,
      "grad_norm": 0.24488963186740875,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 65585
    },
    {
      "epoch": 58.09565987599645,
      "grad_norm": 0.242008775472641,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 65590
    },
    {
      "epoch": 58.100088573959255,
      "grad_norm": 0.23270395398139954,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 65595
    },
    {
      "epoch": 58.104517271922056,
      "grad_norm": 0.27526354789733887,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 65600
    },
    {
      "epoch": 58.10894596988486,
      "grad_norm": 0.20777063071727753,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 65605
    },
    {
      "epoch": 58.11337466784765,
      "grad_norm": 0.2342921942472458,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 65610
    },
    {
      "epoch": 58.11780336581045,
      "grad_norm": 0.27047473192214966,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 65615
    },
    {
      "epoch": 58.122232063773254,
      "grad_norm": 0.25943392515182495,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 65620
    },
    {
      "epoch": 58.12666076173605,
      "grad_norm": 0.25543007254600525,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 65625
    },
    {
      "epoch": 58.13108945969885,
      "grad_norm": 0.2618057131767273,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 65630
    },
    {
      "epoch": 58.13551815766165,
      "grad_norm": 0.21959581971168518,
      "learning_rate": 1e-05,
      "loss": 0.9313,
      "step": 65635
    },
    {
      "epoch": 58.139946855624444,
      "grad_norm": 0.21703411638736725,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 65640
    },
    {
      "epoch": 58.144375553587246,
      "grad_norm": 0.2822389602661133,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 65645
    },
    {
      "epoch": 58.14880425155005,
      "grad_norm": 0.23886580765247345,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 65650
    },
    {
      "epoch": 58.15323294951284,
      "grad_norm": 0.28538328409194946,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 65655
    },
    {
      "epoch": 58.15766164747564,
      "grad_norm": 0.22715330123901367,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 65660
    },
    {
      "epoch": 58.16209034543844,
      "grad_norm": 0.22549784183502197,
      "learning_rate": 1e-05,
      "loss": 0.9381,
      "step": 65665
    },
    {
      "epoch": 58.16651904340124,
      "grad_norm": 0.22157911956310272,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 65670
    },
    {
      "epoch": 58.17094774136404,
      "grad_norm": 0.21394436061382294,
      "learning_rate": 1e-05,
      "loss": 0.9335,
      "step": 65675
    },
    {
      "epoch": 58.17537643932684,
      "grad_norm": 0.2460552453994751,
      "learning_rate": 1e-05,
      "loss": 0.8801,
      "step": 65680
    },
    {
      "epoch": 58.179805137289634,
      "grad_norm": 0.2419407218694687,
      "learning_rate": 1e-05,
      "loss": 0.921,
      "step": 65685
    },
    {
      "epoch": 58.184233835252435,
      "grad_norm": 0.2834322154521942,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 65690
    },
    {
      "epoch": 58.188662533215236,
      "grad_norm": 0.23645217716693878,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 65695
    },
    {
      "epoch": 58.19309123117803,
      "grad_norm": 0.22626784443855286,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 65700
    },
    {
      "epoch": 58.19751992914083,
      "grad_norm": 0.2656615376472473,
      "learning_rate": 1e-05,
      "loss": 1.0159,
      "step": 65705
    },
    {
      "epoch": 58.20194862710363,
      "grad_norm": 0.2266595959663391,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 65710
    },
    {
      "epoch": 58.20637732506643,
      "grad_norm": 0.24668626487255096,
      "learning_rate": 1e-05,
      "loss": 0.933,
      "step": 65715
    },
    {
      "epoch": 58.21080602302923,
      "grad_norm": 0.2371007353067398,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 65720
    },
    {
      "epoch": 58.21523472099203,
      "grad_norm": 0.2515372931957245,
      "learning_rate": 1e-05,
      "loss": 0.8912,
      "step": 65725
    },
    {
      "epoch": 58.21966341895483,
      "grad_norm": 0.21236766874790192,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 65730
    },
    {
      "epoch": 58.224092116917625,
      "grad_norm": 0.23803332448005676,
      "learning_rate": 1e-05,
      "loss": 1.0373,
      "step": 65735
    },
    {
      "epoch": 58.228520814880426,
      "grad_norm": 0.2148313969373703,
      "learning_rate": 1e-05,
      "loss": 0.941,
      "step": 65740
    },
    {
      "epoch": 58.23294951284323,
      "grad_norm": 0.2393161654472351,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 65745
    },
    {
      "epoch": 58.23737821080602,
      "grad_norm": 0.24256375432014465,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 65750
    },
    {
      "epoch": 58.24180690876882,
      "grad_norm": 0.21639619767665863,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 65755
    },
    {
      "epoch": 58.246235606731624,
      "grad_norm": 0.2605295479297638,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 65760
    },
    {
      "epoch": 58.25066430469442,
      "grad_norm": 0.2850554287433624,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 65765
    },
    {
      "epoch": 58.25509300265722,
      "grad_norm": 0.2696142792701721,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 65770
    },
    {
      "epoch": 58.25952170062002,
      "grad_norm": 0.21933311223983765,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 65775
    },
    {
      "epoch": 58.263950398582814,
      "grad_norm": 0.2362578958272934,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 65780
    },
    {
      "epoch": 58.268379096545615,
      "grad_norm": 0.2514587342739105,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 65785
    },
    {
      "epoch": 58.27280779450842,
      "grad_norm": 0.25394684076309204,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 65790
    },
    {
      "epoch": 58.27723649247121,
      "grad_norm": 0.26241201162338257,
      "learning_rate": 1e-05,
      "loss": 1.045,
      "step": 65795
    },
    {
      "epoch": 58.28166519043401,
      "grad_norm": 0.2563392221927643,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 65800
    },
    {
      "epoch": 58.28609388839681,
      "grad_norm": 0.26506853103637695,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 65805
    },
    {
      "epoch": 58.29052258635961,
      "grad_norm": 0.27237415313720703,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 65810
    },
    {
      "epoch": 58.29495128432241,
      "grad_norm": 0.27189236879348755,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 65815
    },
    {
      "epoch": 58.29937998228521,
      "grad_norm": 0.25804075598716736,
      "learning_rate": 1e-05,
      "loss": 0.9493,
      "step": 65820
    },
    {
      "epoch": 58.303808680248004,
      "grad_norm": 0.23183204233646393,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 65825
    },
    {
      "epoch": 58.308237378210805,
      "grad_norm": 0.20005829632282257,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 65830
    },
    {
      "epoch": 58.312666076173606,
      "grad_norm": 0.24902881681919098,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 65835
    },
    {
      "epoch": 58.31709477413641,
      "grad_norm": 0.2612319588661194,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 65840
    },
    {
      "epoch": 58.3215234720992,
      "grad_norm": 0.252181738615036,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 65845
    },
    {
      "epoch": 58.325952170062,
      "grad_norm": 0.23472100496292114,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 65850
    },
    {
      "epoch": 58.330380868024804,
      "grad_norm": 0.2129027545452118,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 65855
    },
    {
      "epoch": 58.3348095659876,
      "grad_norm": 0.2990647554397583,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 65860
    },
    {
      "epoch": 58.3392382639504,
      "grad_norm": 0.24078285694122314,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 65865
    },
    {
      "epoch": 58.3436669619132,
      "grad_norm": 0.23519694805145264,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 65870
    },
    {
      "epoch": 58.348095659875995,
      "grad_norm": 0.22987228631973267,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 65875
    },
    {
      "epoch": 58.352524357838796,
      "grad_norm": 0.22575676441192627,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 65880
    },
    {
      "epoch": 58.3569530558016,
      "grad_norm": 0.23451492190361023,
      "learning_rate": 1e-05,
      "loss": 0.8912,
      "step": 65885
    },
    {
      "epoch": 58.36138175376439,
      "grad_norm": 0.2505142390727997,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 65890
    },
    {
      "epoch": 58.36581045172719,
      "grad_norm": 0.19898705184459686,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 65895
    },
    {
      "epoch": 58.37023914968999,
      "grad_norm": 0.21096228063106537,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 65900
    },
    {
      "epoch": 58.37466784765279,
      "grad_norm": 0.21735480427742004,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 65905
    },
    {
      "epoch": 58.37909654561559,
      "grad_norm": 0.23830744624137878,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 65910
    },
    {
      "epoch": 58.38352524357839,
      "grad_norm": 0.25660133361816406,
      "learning_rate": 1e-05,
      "loss": 1.0189,
      "step": 65915
    },
    {
      "epoch": 58.387953941541184,
      "grad_norm": 0.24429598450660706,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 65920
    },
    {
      "epoch": 58.392382639503985,
      "grad_norm": 0.2170325517654419,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 65925
    },
    {
      "epoch": 58.39681133746679,
      "grad_norm": 0.23029619455337524,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 65930
    },
    {
      "epoch": 58.40124003542958,
      "grad_norm": 0.24783751368522644,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 65935
    },
    {
      "epoch": 58.40566873339238,
      "grad_norm": 0.30578845739364624,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 65940
    },
    {
      "epoch": 58.41009743135518,
      "grad_norm": 0.2600264549255371,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 65945
    },
    {
      "epoch": 58.41452612931798,
      "grad_norm": 0.21792787313461304,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 65950
    },
    {
      "epoch": 58.41895482728078,
      "grad_norm": 0.23134055733680725,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 65955
    },
    {
      "epoch": 58.42338352524358,
      "grad_norm": 0.20161613821983337,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 65960
    },
    {
      "epoch": 58.42781222320638,
      "grad_norm": 0.21276476979255676,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 65965
    },
    {
      "epoch": 58.432240921169175,
      "grad_norm": 0.23096302151679993,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 65970
    },
    {
      "epoch": 58.436669619131976,
      "grad_norm": 0.2491220086812973,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 65975
    },
    {
      "epoch": 58.44109831709478,
      "grad_norm": 0.23723995685577393,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 65980
    },
    {
      "epoch": 58.44552701505757,
      "grad_norm": 0.24288012087345123,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 65985
    },
    {
      "epoch": 58.44995571302037,
      "grad_norm": 0.2346174716949463,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 65990
    },
    {
      "epoch": 58.454384410983174,
      "grad_norm": 0.21272841095924377,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 65995
    },
    {
      "epoch": 58.45881310894597,
      "grad_norm": 0.2549314498901367,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 66000
    },
    {
      "epoch": 58.46324180690877,
      "grad_norm": 0.3597918152809143,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 66005
    },
    {
      "epoch": 58.46767050487157,
      "grad_norm": 0.22955277562141418,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 66010
    },
    {
      "epoch": 58.472099202834364,
      "grad_norm": 0.2038521021604538,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 66015
    },
    {
      "epoch": 58.476527900797166,
      "grad_norm": 0.20900289714336395,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 66020
    },
    {
      "epoch": 58.48095659875997,
      "grad_norm": 0.19945380091667175,
      "learning_rate": 1e-05,
      "loss": 0.9146,
      "step": 66025
    },
    {
      "epoch": 58.48538529672276,
      "grad_norm": 0.21381662786006927,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 66030
    },
    {
      "epoch": 58.48981399468556,
      "grad_norm": 0.24177663028240204,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 66035
    },
    {
      "epoch": 58.49424269264836,
      "grad_norm": 0.2254743129014969,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 66040
    },
    {
      "epoch": 58.49867139061116,
      "grad_norm": 0.24336585402488708,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 66045
    },
    {
      "epoch": 58.50310008857396,
      "grad_norm": 0.2734587788581848,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 66050
    },
    {
      "epoch": 58.50752878653676,
      "grad_norm": 0.21274012327194214,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 66055
    },
    {
      "epoch": 58.511957484499554,
      "grad_norm": 0.21428635716438293,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 66060
    },
    {
      "epoch": 58.516386182462355,
      "grad_norm": 0.2442777454853058,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 66065
    },
    {
      "epoch": 58.520814880425156,
      "grad_norm": 0.22740158438682556,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 66070
    },
    {
      "epoch": 58.52524357838795,
      "grad_norm": 0.24372412264347076,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 66075
    },
    {
      "epoch": 58.52967227635075,
      "grad_norm": 0.20975163578987122,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 66080
    },
    {
      "epoch": 58.53410097431355,
      "grad_norm": 0.24740298092365265,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 66085
    },
    {
      "epoch": 58.538529672276354,
      "grad_norm": 0.3319302201271057,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 66090
    },
    {
      "epoch": 58.54295837023915,
      "grad_norm": 0.25016993284225464,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 66095
    },
    {
      "epoch": 58.54738706820195,
      "grad_norm": 0.23626646399497986,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 66100
    },
    {
      "epoch": 58.55181576616475,
      "grad_norm": 0.1829553097486496,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 66105
    },
    {
      "epoch": 58.556244464127545,
      "grad_norm": 0.2579843997955322,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 66110
    },
    {
      "epoch": 58.560673162090346,
      "grad_norm": 0.21693141758441925,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 66115
    },
    {
      "epoch": 58.56510186005315,
      "grad_norm": 0.2358841449022293,
      "learning_rate": 1e-05,
      "loss": 1.0554,
      "step": 66120
    },
    {
      "epoch": 58.56953055801594,
      "grad_norm": 0.23074689507484436,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 66125
    },
    {
      "epoch": 58.57395925597874,
      "grad_norm": 0.2559589445590973,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 66130
    },
    {
      "epoch": 58.578387953941544,
      "grad_norm": 0.2241109311580658,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 66135
    },
    {
      "epoch": 58.58281665190434,
      "grad_norm": 0.2654598653316498,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 66140
    },
    {
      "epoch": 58.58724534986714,
      "grad_norm": 0.22279340028762817,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 66145
    },
    {
      "epoch": 58.59167404782994,
      "grad_norm": 0.22340507805347443,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 66150
    },
    {
      "epoch": 58.596102745792734,
      "grad_norm": 0.23884937167167664,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 66155
    },
    {
      "epoch": 58.600531443755536,
      "grad_norm": 0.2491408735513687,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 66160
    },
    {
      "epoch": 58.60496014171834,
      "grad_norm": 0.2564755082130432,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 66165
    },
    {
      "epoch": 58.60938883968113,
      "grad_norm": 0.22645698487758636,
      "learning_rate": 1e-05,
      "loss": 1.0501,
      "step": 66170
    },
    {
      "epoch": 58.61381753764393,
      "grad_norm": 0.2449694126844406,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 66175
    },
    {
      "epoch": 58.61824623560673,
      "grad_norm": 0.22285647690296173,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 66180
    },
    {
      "epoch": 58.62267493356953,
      "grad_norm": 0.2291710525751114,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 66185
    },
    {
      "epoch": 58.62710363153233,
      "grad_norm": 0.2962253987789154,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 66190
    },
    {
      "epoch": 58.63153232949513,
      "grad_norm": 0.2888045310974121,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 66195
    },
    {
      "epoch": 58.635961027457924,
      "grad_norm": 0.24166929721832275,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 66200
    },
    {
      "epoch": 58.640389725420725,
      "grad_norm": 0.23009037971496582,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 66205
    },
    {
      "epoch": 58.644818423383526,
      "grad_norm": 0.25208649039268494,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 66210
    },
    {
      "epoch": 58.64924712134633,
      "grad_norm": 0.22103185951709747,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 66215
    },
    {
      "epoch": 58.65367581930912,
      "grad_norm": 0.24001199007034302,
      "learning_rate": 1e-05,
      "loss": 0.936,
      "step": 66220
    },
    {
      "epoch": 58.65810451727192,
      "grad_norm": 0.2976042628288269,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 66225
    },
    {
      "epoch": 58.662533215234724,
      "grad_norm": 0.20681537687778473,
      "learning_rate": 1e-05,
      "loss": 0.9094,
      "step": 66230
    },
    {
      "epoch": 58.66696191319752,
      "grad_norm": 0.20726196467876434,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 66235
    },
    {
      "epoch": 58.67139061116032,
      "grad_norm": 0.21387752890586853,
      "learning_rate": 1e-05,
      "loss": 1.0204,
      "step": 66240
    },
    {
      "epoch": 58.67581930912312,
      "grad_norm": 0.2361944019794464,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 66245
    },
    {
      "epoch": 58.680248007085915,
      "grad_norm": 0.19607706367969513,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 66250
    },
    {
      "epoch": 58.684676705048716,
      "grad_norm": 0.24780048429965973,
      "learning_rate": 1e-05,
      "loss": 0.8867,
      "step": 66255
    },
    {
      "epoch": 58.68910540301152,
      "grad_norm": 0.22258004546165466,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 66260
    },
    {
      "epoch": 58.69353410097431,
      "grad_norm": 0.21379536390304565,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 66265
    },
    {
      "epoch": 58.69796279893711,
      "grad_norm": 0.23343965411186218,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 66270
    },
    {
      "epoch": 58.702391496899914,
      "grad_norm": 0.24378928542137146,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 66275
    },
    {
      "epoch": 58.70682019486271,
      "grad_norm": 0.23671655356884003,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 66280
    },
    {
      "epoch": 58.71124889282551,
      "grad_norm": 0.2634745240211487,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 66285
    },
    {
      "epoch": 58.71567759078831,
      "grad_norm": 0.206903338432312,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 66290
    },
    {
      "epoch": 58.720106288751104,
      "grad_norm": 0.24206727743148804,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 66295
    },
    {
      "epoch": 58.724534986713905,
      "grad_norm": 0.22427155077457428,
      "learning_rate": 1e-05,
      "loss": 1.0347,
      "step": 66300
    },
    {
      "epoch": 58.72896368467671,
      "grad_norm": 0.21122358739376068,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 66305
    },
    {
      "epoch": 58.7333923826395,
      "grad_norm": 0.21206024289131165,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 66310
    },
    {
      "epoch": 58.7378210806023,
      "grad_norm": 0.23835526406764984,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 66315
    },
    {
      "epoch": 58.7422497785651,
      "grad_norm": 0.274087518453598,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 66320
    },
    {
      "epoch": 58.7466784765279,
      "grad_norm": 0.314848393201828,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 66325
    },
    {
      "epoch": 58.7511071744907,
      "grad_norm": 0.23780398070812225,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 66330
    },
    {
      "epoch": 58.7555358724535,
      "grad_norm": 0.24787330627441406,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 66335
    },
    {
      "epoch": 58.7599645704163,
      "grad_norm": 0.23480159044265747,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 66340
    },
    {
      "epoch": 58.764393268379095,
      "grad_norm": 0.2621453106403351,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 66345
    },
    {
      "epoch": 58.768821966341896,
      "grad_norm": 0.27325886487960815,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 66350
    },
    {
      "epoch": 58.7732506643047,
      "grad_norm": 0.21999648213386536,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 66355
    },
    {
      "epoch": 58.77767936226749,
      "grad_norm": 0.23174482583999634,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 66360
    },
    {
      "epoch": 58.78210806023029,
      "grad_norm": 0.23891036212444305,
      "learning_rate": 1e-05,
      "loss": 0.903,
      "step": 66365
    },
    {
      "epoch": 58.786536758193094,
      "grad_norm": 0.23595242202281952,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 66370
    },
    {
      "epoch": 58.79096545615589,
      "grad_norm": 0.2569817006587982,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 66375
    },
    {
      "epoch": 58.79539415411869,
      "grad_norm": 0.21331319212913513,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 66380
    },
    {
      "epoch": 58.79982285208149,
      "grad_norm": 0.21353302896022797,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 66385
    },
    {
      "epoch": 58.804251550044285,
      "grad_norm": 0.2393171638250351,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 66390
    },
    {
      "epoch": 58.808680248007086,
      "grad_norm": 0.26260244846343994,
      "learning_rate": 1e-05,
      "loss": 0.9271,
      "step": 66395
    },
    {
      "epoch": 58.81310894596989,
      "grad_norm": 0.2421475052833557,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 66400
    },
    {
      "epoch": 58.81753764393268,
      "grad_norm": 0.24958375096321106,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 66405
    },
    {
      "epoch": 58.82196634189548,
      "grad_norm": 0.3130318522453308,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 66410
    },
    {
      "epoch": 58.826395039858284,
      "grad_norm": 0.25453537702560425,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 66415
    },
    {
      "epoch": 58.83082373782108,
      "grad_norm": 0.22835572063922882,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 66420
    },
    {
      "epoch": 58.83525243578388,
      "grad_norm": 0.3002355098724365,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 66425
    },
    {
      "epoch": 58.83968113374668,
      "grad_norm": 0.2848992943763733,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 66430
    },
    {
      "epoch": 58.844109831709474,
      "grad_norm": 0.2286827117204666,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 66435
    },
    {
      "epoch": 58.848538529672275,
      "grad_norm": 0.23456798493862152,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 66440
    },
    {
      "epoch": 58.85296722763508,
      "grad_norm": 0.2134729027748108,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 66445
    },
    {
      "epoch": 58.85739592559787,
      "grad_norm": 0.2205529361963272,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 66450
    },
    {
      "epoch": 58.86182462356067,
      "grad_norm": 0.25917816162109375,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 66455
    },
    {
      "epoch": 58.86625332152347,
      "grad_norm": 0.25884154438972473,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 66460
    },
    {
      "epoch": 58.870682019486274,
      "grad_norm": 0.2629377543926239,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 66465
    },
    {
      "epoch": 58.87511071744907,
      "grad_norm": 0.2892385721206665,
      "learning_rate": 1e-05,
      "loss": 0.9213,
      "step": 66470
    },
    {
      "epoch": 58.87953941541187,
      "grad_norm": 0.21622851490974426,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 66475
    },
    {
      "epoch": 58.88396811337467,
      "grad_norm": 0.2472354918718338,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 66480
    },
    {
      "epoch": 58.888396811337465,
      "grad_norm": 0.244821235537529,
      "learning_rate": 1e-05,
      "loss": 0.9259,
      "step": 66485
    },
    {
      "epoch": 58.892825509300266,
      "grad_norm": 0.2422996610403061,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 66490
    },
    {
      "epoch": 58.89725420726307,
      "grad_norm": 0.2427123486995697,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 66495
    },
    {
      "epoch": 58.90168290522586,
      "grad_norm": 0.2249944508075714,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 66500
    },
    {
      "epoch": 58.90611160318866,
      "grad_norm": 0.2916962504386902,
      "learning_rate": 1e-05,
      "loss": 0.8985,
      "step": 66505
    },
    {
      "epoch": 58.910540301151464,
      "grad_norm": 0.2462603598833084,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 66510
    },
    {
      "epoch": 58.91496899911426,
      "grad_norm": 0.2573831379413605,
      "learning_rate": 1e-05,
      "loss": 1.0366,
      "step": 66515
    },
    {
      "epoch": 58.91939769707706,
      "grad_norm": 0.22972248494625092,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 66520
    },
    {
      "epoch": 58.92382639503986,
      "grad_norm": 0.24964314699172974,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 66525
    },
    {
      "epoch": 58.928255093002655,
      "grad_norm": 0.27196383476257324,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 66530
    },
    {
      "epoch": 58.932683790965456,
      "grad_norm": 0.2552023231983185,
      "learning_rate": 1e-05,
      "loss": 1.0175,
      "step": 66535
    },
    {
      "epoch": 58.93711248892826,
      "grad_norm": 0.23833294212818146,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 66540
    },
    {
      "epoch": 58.94154118689105,
      "grad_norm": 0.25160494446754456,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 66545
    },
    {
      "epoch": 58.94596988485385,
      "grad_norm": 0.21554109454154968,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 66550
    },
    {
      "epoch": 58.95039858281665,
      "grad_norm": 0.2272166609764099,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 66555
    },
    {
      "epoch": 58.95482728077945,
      "grad_norm": 0.263322114944458,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 66560
    },
    {
      "epoch": 58.95925597874225,
      "grad_norm": 0.2646535336971283,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 66565
    },
    {
      "epoch": 58.96368467670505,
      "grad_norm": 0.23884084820747375,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 66570
    },
    {
      "epoch": 58.96811337466785,
      "grad_norm": 0.24935728311538696,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 66575
    },
    {
      "epoch": 58.972542072630645,
      "grad_norm": 0.24558894336223602,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 66580
    },
    {
      "epoch": 58.97697077059345,
      "grad_norm": 0.2120255082845688,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 66585
    },
    {
      "epoch": 58.98139946855625,
      "grad_norm": 0.23919865489006042,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 66590
    },
    {
      "epoch": 58.98582816651904,
      "grad_norm": 0.24285271763801575,
      "learning_rate": 1e-05,
      "loss": 0.8847,
      "step": 66595
    },
    {
      "epoch": 58.99025686448184,
      "grad_norm": 0.26579606533050537,
      "learning_rate": 1e-05,
      "loss": 1.0232,
      "step": 66600
    },
    {
      "epoch": 58.994685562444644,
      "grad_norm": 0.24005885422229767,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 66605
    },
    {
      "epoch": 58.99911426040744,
      "grad_norm": 0.22951431572437286,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 66610
    },
    {
      "epoch": 59.00354295837024,
      "grad_norm": 0.2932699918746948,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 66615
    },
    {
      "epoch": 59.00797165633304,
      "grad_norm": 0.22845040261745453,
      "learning_rate": 1e-05,
      "loss": 0.9347,
      "step": 66620
    },
    {
      "epoch": 59.012400354295835,
      "grad_norm": 0.21848931908607483,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 66625
    },
    {
      "epoch": 59.016829052258636,
      "grad_norm": 0.24698922038078308,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 66630
    },
    {
      "epoch": 59.02125775022144,
      "grad_norm": 0.2556541860103607,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 66635
    },
    {
      "epoch": 59.02568644818423,
      "grad_norm": 0.23641115427017212,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 66640
    },
    {
      "epoch": 59.03011514614703,
      "grad_norm": 0.2486625611782074,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 66645
    },
    {
      "epoch": 59.034543844109834,
      "grad_norm": 0.23156021535396576,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 66650
    },
    {
      "epoch": 59.03897254207263,
      "grad_norm": 0.2528839707374573,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 66655
    },
    {
      "epoch": 59.04340124003543,
      "grad_norm": 0.23054809868335724,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 66660
    },
    {
      "epoch": 59.04782993799823,
      "grad_norm": 0.2053862065076828,
      "learning_rate": 1e-05,
      "loss": 0.9246,
      "step": 66665
    },
    {
      "epoch": 59.052258635961024,
      "grad_norm": 0.22063937783241272,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 66670
    },
    {
      "epoch": 59.056687333923826,
      "grad_norm": 0.22324319183826447,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 66675
    },
    {
      "epoch": 59.06111603188663,
      "grad_norm": 0.23338168859481812,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 66680
    },
    {
      "epoch": 59.06554472984942,
      "grad_norm": 0.2395501434803009,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 66685
    },
    {
      "epoch": 59.06997342781222,
      "grad_norm": 0.2651427984237671,
      "learning_rate": 1e-05,
      "loss": 1.0267,
      "step": 66690
    },
    {
      "epoch": 59.07440212577502,
      "grad_norm": 0.24108341336250305,
      "learning_rate": 1e-05,
      "loss": 0.922,
      "step": 66695
    },
    {
      "epoch": 59.078830823737825,
      "grad_norm": 0.24153071641921997,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 66700
    },
    {
      "epoch": 59.08325952170062,
      "grad_norm": 0.2270612120628357,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 66705
    },
    {
      "epoch": 59.08768821966342,
      "grad_norm": 0.23814406991004944,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 66710
    },
    {
      "epoch": 59.09211691762622,
      "grad_norm": 0.24804134666919708,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 66715
    },
    {
      "epoch": 59.096545615589015,
      "grad_norm": 0.27677488327026367,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 66720
    },
    {
      "epoch": 59.100974313551816,
      "grad_norm": 0.2481439858675003,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 66725
    },
    {
      "epoch": 59.10540301151462,
      "grad_norm": 0.21477828919887543,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 66730
    },
    {
      "epoch": 59.10983170947741,
      "grad_norm": 0.23443078994750977,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 66735
    },
    {
      "epoch": 59.11426040744021,
      "grad_norm": 0.19984757900238037,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 66740
    },
    {
      "epoch": 59.118689105403014,
      "grad_norm": 0.22607550024986267,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 66745
    },
    {
      "epoch": 59.12311780336581,
      "grad_norm": 0.2566843330860138,
      "learning_rate": 1e-05,
      "loss": 1.0295,
      "step": 66750
    },
    {
      "epoch": 59.12754650132861,
      "grad_norm": 0.24119414389133453,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 66755
    },
    {
      "epoch": 59.13197519929141,
      "grad_norm": 0.24013695120811462,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 66760
    },
    {
      "epoch": 59.136403897254205,
      "grad_norm": 0.2711068093776703,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 66765
    },
    {
      "epoch": 59.140832595217006,
      "grad_norm": 0.25309956073760986,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 66770
    },
    {
      "epoch": 59.14526129317981,
      "grad_norm": 0.26761960983276367,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 66775
    },
    {
      "epoch": 59.1496899911426,
      "grad_norm": 0.27214285731315613,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 66780
    },
    {
      "epoch": 59.1541186891054,
      "grad_norm": 0.22049346566200256,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 66785
    },
    {
      "epoch": 59.158547387068204,
      "grad_norm": 0.3557475209236145,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 66790
    },
    {
      "epoch": 59.162976085031,
      "grad_norm": 0.2651252746582031,
      "learning_rate": 1e-05,
      "loss": 0.9149,
      "step": 66795
    },
    {
      "epoch": 59.1674047829938,
      "grad_norm": 0.29058703780174255,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 66800
    },
    {
      "epoch": 59.1718334809566,
      "grad_norm": 0.27897506952285767,
      "learning_rate": 1e-05,
      "loss": 0.9239,
      "step": 66805
    },
    {
      "epoch": 59.176262178919394,
      "grad_norm": 0.22232943773269653,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 66810
    },
    {
      "epoch": 59.180690876882196,
      "grad_norm": 0.24779817461967468,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 66815
    },
    {
      "epoch": 59.185119574845,
      "grad_norm": 0.22302231192588806,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 66820
    },
    {
      "epoch": 59.1895482728078,
      "grad_norm": 0.24547143280506134,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 66825
    },
    {
      "epoch": 59.19397697077059,
      "grad_norm": 0.21537873148918152,
      "learning_rate": 1e-05,
      "loss": 0.9271,
      "step": 66830
    },
    {
      "epoch": 59.19840566873339,
      "grad_norm": 0.3073902130126953,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 66835
    },
    {
      "epoch": 59.202834366696194,
      "grad_norm": 0.23865914344787598,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 66840
    },
    {
      "epoch": 59.20726306465899,
      "grad_norm": 0.2591618299484253,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 66845
    },
    {
      "epoch": 59.21169176262179,
      "grad_norm": 0.23636005818843842,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 66850
    },
    {
      "epoch": 59.21612046058459,
      "grad_norm": 0.2438323199748993,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 66855
    },
    {
      "epoch": 59.220549158547385,
      "grad_norm": 0.24456475675106049,
      "learning_rate": 1e-05,
      "loss": 1.0107,
      "step": 66860
    },
    {
      "epoch": 59.224977856510186,
      "grad_norm": 0.23916451632976532,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 66865
    },
    {
      "epoch": 59.22940655447299,
      "grad_norm": 0.2286665141582489,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 66870
    },
    {
      "epoch": 59.23383525243578,
      "grad_norm": 0.22660048305988312,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 66875
    },
    {
      "epoch": 59.23826395039858,
      "grad_norm": 0.2248922437429428,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 66880
    },
    {
      "epoch": 59.242692648361384,
      "grad_norm": 0.24523353576660156,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 66885
    },
    {
      "epoch": 59.24712134632418,
      "grad_norm": 0.2399088591337204,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 66890
    },
    {
      "epoch": 59.25155004428698,
      "grad_norm": 0.23760074377059937,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 66895
    },
    {
      "epoch": 59.25597874224978,
      "grad_norm": 0.2238783836364746,
      "learning_rate": 1e-05,
      "loss": 0.9297,
      "step": 66900
    },
    {
      "epoch": 59.260407440212575,
      "grad_norm": 0.287847101688385,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 66905
    },
    {
      "epoch": 59.264836138175376,
      "grad_norm": 0.24397455155849457,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 66910
    },
    {
      "epoch": 59.26926483613818,
      "grad_norm": 0.24414679408073425,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 66915
    },
    {
      "epoch": 59.27369353410097,
      "grad_norm": 0.23794770240783691,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 66920
    },
    {
      "epoch": 59.27812223206377,
      "grad_norm": 0.26321831345558167,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 66925
    },
    {
      "epoch": 59.282550930026574,
      "grad_norm": 0.2558186650276184,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 66930
    },
    {
      "epoch": 59.28697962798937,
      "grad_norm": 0.25666743516921997,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 66935
    },
    {
      "epoch": 59.29140832595217,
      "grad_norm": 0.2599847912788391,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 66940
    },
    {
      "epoch": 59.29583702391497,
      "grad_norm": 0.24281898140907288,
      "learning_rate": 1e-05,
      "loss": 0.9211,
      "step": 66945
    },
    {
      "epoch": 59.30026572187777,
      "grad_norm": 0.2657301723957062,
      "learning_rate": 1e-05,
      "loss": 0.9251,
      "step": 66950
    },
    {
      "epoch": 59.304694419840565,
      "grad_norm": 0.2535754442214966,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 66955
    },
    {
      "epoch": 59.30912311780337,
      "grad_norm": 0.2630328834056854,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 66960
    },
    {
      "epoch": 59.31355181576617,
      "grad_norm": 0.20457661151885986,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 66965
    },
    {
      "epoch": 59.31798051372896,
      "grad_norm": 0.23641666769981384,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 66970
    },
    {
      "epoch": 59.32240921169176,
      "grad_norm": 0.2554995119571686,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 66975
    },
    {
      "epoch": 59.326837909654564,
      "grad_norm": 0.2306179702281952,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 66980
    },
    {
      "epoch": 59.33126660761736,
      "grad_norm": 0.25697603821754456,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 66985
    },
    {
      "epoch": 59.33569530558016,
      "grad_norm": 0.25965002179145813,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 66990
    },
    {
      "epoch": 59.34012400354296,
      "grad_norm": 0.24917984008789062,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 66995
    },
    {
      "epoch": 59.344552701505755,
      "grad_norm": 0.26654982566833496,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 67000
    },
    {
      "epoch": 59.348981399468556,
      "grad_norm": 0.2426074892282486,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 67005
    },
    {
      "epoch": 59.35341009743136,
      "grad_norm": 0.26889070868492126,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 67010
    },
    {
      "epoch": 59.35783879539415,
      "grad_norm": 0.20574183762073517,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 67015
    },
    {
      "epoch": 59.36226749335695,
      "grad_norm": 0.23412829637527466,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 67020
    },
    {
      "epoch": 59.366696191319754,
      "grad_norm": 0.24843290448188782,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 67025
    },
    {
      "epoch": 59.37112488928255,
      "grad_norm": 0.2229468673467636,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 67030
    },
    {
      "epoch": 59.37555358724535,
      "grad_norm": 0.23482932150363922,
      "learning_rate": 1e-05,
      "loss": 0.9305,
      "step": 67035
    },
    {
      "epoch": 59.37998228520815,
      "grad_norm": 0.23547935485839844,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 67040
    },
    {
      "epoch": 59.384410983170945,
      "grad_norm": 0.24420316517353058,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 67045
    },
    {
      "epoch": 59.388839681133746,
      "grad_norm": 0.2632458209991455,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 67050
    },
    {
      "epoch": 59.39326837909655,
      "grad_norm": 0.24076366424560547,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 67055
    },
    {
      "epoch": 59.39769707705934,
      "grad_norm": 0.27422770857810974,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 67060
    },
    {
      "epoch": 59.40212577502214,
      "grad_norm": 0.2265530526638031,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 67065
    },
    {
      "epoch": 59.40655447298494,
      "grad_norm": 0.22102276980876923,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 67070
    },
    {
      "epoch": 59.410983170947745,
      "grad_norm": 0.21235457062721252,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 67075
    },
    {
      "epoch": 59.41541186891054,
      "grad_norm": 0.24659648537635803,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 67080
    },
    {
      "epoch": 59.41984056687334,
      "grad_norm": 0.2166571319103241,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 67085
    },
    {
      "epoch": 59.42426926483614,
      "grad_norm": 0.22712436318397522,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 67090
    },
    {
      "epoch": 59.428697962798935,
      "grad_norm": 0.21531012654304504,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 67095
    },
    {
      "epoch": 59.43312666076174,
      "grad_norm": 0.21641729772090912,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 67100
    },
    {
      "epoch": 59.43755535872454,
      "grad_norm": 0.20961648225784302,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 67105
    },
    {
      "epoch": 59.44198405668733,
      "grad_norm": 0.19643104076385498,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 67110
    },
    {
      "epoch": 59.44641275465013,
      "grad_norm": 0.21212148666381836,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 67115
    },
    {
      "epoch": 59.450841452612934,
      "grad_norm": 0.21542568504810333,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 67120
    },
    {
      "epoch": 59.45527015057573,
      "grad_norm": 0.26023104786872864,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 67125
    },
    {
      "epoch": 59.45969884853853,
      "grad_norm": 0.23674340546131134,
      "learning_rate": 1e-05,
      "loss": 0.9092,
      "step": 67130
    },
    {
      "epoch": 59.46412754650133,
      "grad_norm": 0.23538298904895782,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 67135
    },
    {
      "epoch": 59.468556244464125,
      "grad_norm": 0.2596363425254822,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 67140
    },
    {
      "epoch": 59.472984942426926,
      "grad_norm": 0.20869964361190796,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 67145
    },
    {
      "epoch": 59.47741364038973,
      "grad_norm": 0.2394796460866928,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 67150
    },
    {
      "epoch": 59.48184233835252,
      "grad_norm": 0.22037747502326965,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 67155
    },
    {
      "epoch": 59.48627103631532,
      "grad_norm": 0.20692504942417145,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 67160
    },
    {
      "epoch": 59.490699734278124,
      "grad_norm": 0.22223250567913055,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 67165
    },
    {
      "epoch": 59.49512843224092,
      "grad_norm": 0.2629796862602234,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 67170
    },
    {
      "epoch": 59.49955713020372,
      "grad_norm": 0.2817767858505249,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 67175
    },
    {
      "epoch": 59.50398582816652,
      "grad_norm": 0.2689191997051239,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 67180
    },
    {
      "epoch": 59.508414526129314,
      "grad_norm": 0.27091455459594727,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 67185
    },
    {
      "epoch": 59.512843224092116,
      "grad_norm": 0.26691892743110657,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 67190
    },
    {
      "epoch": 59.51727192205492,
      "grad_norm": 0.26606935262680054,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 67195
    },
    {
      "epoch": 59.52170062001772,
      "grad_norm": 0.22358602285385132,
      "learning_rate": 1e-05,
      "loss": 0.921,
      "step": 67200
    },
    {
      "epoch": 59.52612931798051,
      "grad_norm": 0.22314418852329254,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 67205
    },
    {
      "epoch": 59.53055801594331,
      "grad_norm": 0.22729524970054626,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 67210
    },
    {
      "epoch": 59.534986713906115,
      "grad_norm": 0.2256029099225998,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 67215
    },
    {
      "epoch": 59.53941541186891,
      "grad_norm": 0.27966466546058655,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 67220
    },
    {
      "epoch": 59.54384410983171,
      "grad_norm": 0.2753465473651886,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 67225
    },
    {
      "epoch": 59.54827280779451,
      "grad_norm": 0.24972563982009888,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 67230
    },
    {
      "epoch": 59.552701505757305,
      "grad_norm": 0.28063878417015076,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 67235
    },
    {
      "epoch": 59.55713020372011,
      "grad_norm": 0.26607462763786316,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 67240
    },
    {
      "epoch": 59.56155890168291,
      "grad_norm": 0.2708020806312561,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 67245
    },
    {
      "epoch": 59.5659875996457,
      "grad_norm": 0.2864118814468384,
      "learning_rate": 1e-05,
      "loss": 1.0101,
      "step": 67250
    },
    {
      "epoch": 59.5704162976085,
      "grad_norm": 0.2689034938812256,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 67255
    },
    {
      "epoch": 59.574844995571304,
      "grad_norm": 0.24226604402065277,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 67260
    },
    {
      "epoch": 59.5792736935341,
      "grad_norm": 0.23298649489879608,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 67265
    },
    {
      "epoch": 59.5837023914969,
      "grad_norm": 0.23314152657985687,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 67270
    },
    {
      "epoch": 59.5881310894597,
      "grad_norm": 0.24087823927402496,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 67275
    },
    {
      "epoch": 59.592559787422495,
      "grad_norm": 0.2168496549129486,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 67280
    },
    {
      "epoch": 59.596988485385296,
      "grad_norm": 0.27107295393943787,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 67285
    },
    {
      "epoch": 59.6014171833481,
      "grad_norm": 0.25947487354278564,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 67290
    },
    {
      "epoch": 59.60584588131089,
      "grad_norm": 0.23305225372314453,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 67295
    },
    {
      "epoch": 59.61027457927369,
      "grad_norm": 0.2442581206560135,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 67300
    },
    {
      "epoch": 59.614703277236494,
      "grad_norm": 0.26204654574394226,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 67305
    },
    {
      "epoch": 59.619131975199295,
      "grad_norm": 0.2580435574054718,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 67310
    },
    {
      "epoch": 59.62356067316209,
      "grad_norm": 0.23890714347362518,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 67315
    },
    {
      "epoch": 59.62798937112489,
      "grad_norm": 0.249770388007164,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 67320
    },
    {
      "epoch": 59.63241806908769,
      "grad_norm": 0.22697529196739197,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 67325
    },
    {
      "epoch": 59.636846767050486,
      "grad_norm": 0.23145148158073425,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 67330
    },
    {
      "epoch": 59.64127546501329,
      "grad_norm": 0.23793938755989075,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 67335
    },
    {
      "epoch": 59.64570416297609,
      "grad_norm": 0.2357412874698639,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 67340
    },
    {
      "epoch": 59.65013286093888,
      "grad_norm": 0.2740411162376404,
      "learning_rate": 1e-05,
      "loss": 0.9135,
      "step": 67345
    },
    {
      "epoch": 59.65456155890168,
      "grad_norm": 0.2274979203939438,
      "learning_rate": 1e-05,
      "loss": 0.903,
      "step": 67350
    },
    {
      "epoch": 59.658990256864485,
      "grad_norm": 0.2634359300136566,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 67355
    },
    {
      "epoch": 59.66341895482728,
      "grad_norm": 0.2276473194360733,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 67360
    },
    {
      "epoch": 59.66784765279008,
      "grad_norm": 0.2399526834487915,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 67365
    },
    {
      "epoch": 59.67227635075288,
      "grad_norm": 0.23105016350746155,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 67370
    },
    {
      "epoch": 59.676705048715675,
      "grad_norm": 0.24216358363628387,
      "learning_rate": 1e-05,
      "loss": 1.0355,
      "step": 67375
    },
    {
      "epoch": 59.681133746678476,
      "grad_norm": 0.21329949796199799,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 67380
    },
    {
      "epoch": 59.68556244464128,
      "grad_norm": 0.2208309918642044,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 67385
    },
    {
      "epoch": 59.68999114260407,
      "grad_norm": 0.24061959981918335,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 67390
    },
    {
      "epoch": 59.69441984056687,
      "grad_norm": 0.2035655379295349,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 67395
    },
    {
      "epoch": 59.698848538529674,
      "grad_norm": 0.21024549007415771,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 67400
    },
    {
      "epoch": 59.70327723649247,
      "grad_norm": 0.24159760773181915,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 67405
    },
    {
      "epoch": 59.70770593445527,
      "grad_norm": 0.2286030650138855,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 67410
    },
    {
      "epoch": 59.71213463241807,
      "grad_norm": 0.28946295380592346,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 67415
    },
    {
      "epoch": 59.716563330380865,
      "grad_norm": 0.19123387336730957,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 67420
    },
    {
      "epoch": 59.720992028343666,
      "grad_norm": 0.2614642083644867,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 67425
    },
    {
      "epoch": 59.72542072630647,
      "grad_norm": 0.2773922383785248,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 67430
    },
    {
      "epoch": 59.72984942426926,
      "grad_norm": 0.2428750842809677,
      "learning_rate": 1e-05,
      "loss": 0.9184,
      "step": 67435
    },
    {
      "epoch": 59.73427812223206,
      "grad_norm": 0.22674207389354706,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 67440
    },
    {
      "epoch": 59.738706820194864,
      "grad_norm": 0.24473629891872406,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 67445
    },
    {
      "epoch": 59.743135518157665,
      "grad_norm": 0.24417509138584137,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 67450
    },
    {
      "epoch": 59.74756421612046,
      "grad_norm": 0.23501455783843994,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 67455
    },
    {
      "epoch": 59.75199291408326,
      "grad_norm": 0.217117041349411,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 67460
    },
    {
      "epoch": 59.75642161204606,
      "grad_norm": 0.2634599506855011,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 67465
    },
    {
      "epoch": 59.760850310008856,
      "grad_norm": 0.251139760017395,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 67470
    },
    {
      "epoch": 59.76527900797166,
      "grad_norm": 0.248135045170784,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 67475
    },
    {
      "epoch": 59.76970770593446,
      "grad_norm": 0.24345804750919342,
      "learning_rate": 1e-05,
      "loss": 1.0309,
      "step": 67480
    },
    {
      "epoch": 59.77413640389725,
      "grad_norm": 0.2599170506000519,
      "learning_rate": 1e-05,
      "loss": 0.9493,
      "step": 67485
    },
    {
      "epoch": 59.77856510186005,
      "grad_norm": 0.2228216677904129,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 67490
    },
    {
      "epoch": 59.782993799822854,
      "grad_norm": 0.21924491226673126,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 67495
    },
    {
      "epoch": 59.78742249778565,
      "grad_norm": 0.2735077440738678,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 67500
    },
    {
      "epoch": 59.79185119574845,
      "grad_norm": 0.24382410943508148,
      "learning_rate": 1e-05,
      "loss": 1.0471,
      "step": 67505
    },
    {
      "epoch": 59.79627989371125,
      "grad_norm": 0.25825035572052,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 67510
    },
    {
      "epoch": 59.800708591674045,
      "grad_norm": 0.23420469462871552,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 67515
    },
    {
      "epoch": 59.805137289636846,
      "grad_norm": 0.24119849503040314,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 67520
    },
    {
      "epoch": 59.80956598759965,
      "grad_norm": 0.251857191324234,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 67525
    },
    {
      "epoch": 59.81399468556244,
      "grad_norm": 0.25429603457450867,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 67530
    },
    {
      "epoch": 59.81842338352524,
      "grad_norm": 0.3338758945465088,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 67535
    },
    {
      "epoch": 59.822852081488044,
      "grad_norm": 0.2282615602016449,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 67540
    },
    {
      "epoch": 59.82728077945084,
      "grad_norm": 0.23681332170963287,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 67545
    },
    {
      "epoch": 59.83170947741364,
      "grad_norm": 0.21596649289131165,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 67550
    },
    {
      "epoch": 59.83613817537644,
      "grad_norm": 0.25154420733451843,
      "learning_rate": 1e-05,
      "loss": 0.9151,
      "step": 67555
    },
    {
      "epoch": 59.84056687333924,
      "grad_norm": 0.2358541339635849,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 67560
    },
    {
      "epoch": 59.844995571302036,
      "grad_norm": 0.23139184713363647,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 67565
    },
    {
      "epoch": 59.84942426926484,
      "grad_norm": 0.24705787003040314,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 67570
    },
    {
      "epoch": 59.85385296722764,
      "grad_norm": 0.2336421012878418,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 67575
    },
    {
      "epoch": 59.85828166519043,
      "grad_norm": 0.22496020793914795,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 67580
    },
    {
      "epoch": 59.862710363153234,
      "grad_norm": 0.23790419101715088,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 67585
    },
    {
      "epoch": 59.867139061116035,
      "grad_norm": 0.23613722622394562,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 67590
    },
    {
      "epoch": 59.87156775907883,
      "grad_norm": 0.23906050622463226,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 67595
    },
    {
      "epoch": 59.87599645704163,
      "grad_norm": 0.2269524186849594,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 67600
    },
    {
      "epoch": 59.88042515500443,
      "grad_norm": 0.22984082996845245,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 67605
    },
    {
      "epoch": 59.884853852967225,
      "grad_norm": 0.27834585309028625,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 67610
    },
    {
      "epoch": 59.88928255093003,
      "grad_norm": 0.2387055605649948,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 67615
    },
    {
      "epoch": 59.89371124889283,
      "grad_norm": 0.23092499375343323,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 67620
    },
    {
      "epoch": 59.89813994685562,
      "grad_norm": 0.2322911024093628,
      "learning_rate": 1e-05,
      "loss": 1.0418,
      "step": 67625
    },
    {
      "epoch": 59.90256864481842,
      "grad_norm": 0.24539130926132202,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 67630
    },
    {
      "epoch": 59.906997342781224,
      "grad_norm": 0.27204397320747375,
      "learning_rate": 1e-05,
      "loss": 0.9427,
      "step": 67635
    },
    {
      "epoch": 59.91142604074402,
      "grad_norm": 0.2431749850511551,
      "learning_rate": 1e-05,
      "loss": 1.0219,
      "step": 67640
    },
    {
      "epoch": 59.91585473870682,
      "grad_norm": 0.2722782790660858,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 67645
    },
    {
      "epoch": 59.92028343666962,
      "grad_norm": 0.271264910697937,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 67650
    },
    {
      "epoch": 59.924712134632415,
      "grad_norm": 0.24249565601348877,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 67655
    },
    {
      "epoch": 59.929140832595216,
      "grad_norm": 0.23625680804252625,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 67660
    },
    {
      "epoch": 59.93356953055802,
      "grad_norm": 0.19955313205718994,
      "learning_rate": 1e-05,
      "loss": 0.9292,
      "step": 67665
    },
    {
      "epoch": 59.93799822852081,
      "grad_norm": 0.27409690618515015,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 67670
    },
    {
      "epoch": 59.94242692648361,
      "grad_norm": 0.2384069412946701,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 67675
    },
    {
      "epoch": 59.946855624446414,
      "grad_norm": 0.20589803159236908,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 67680
    },
    {
      "epoch": 59.951284322409215,
      "grad_norm": 0.22315742075443268,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 67685
    },
    {
      "epoch": 59.95571302037201,
      "grad_norm": 0.2215559333562851,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 67690
    },
    {
      "epoch": 59.96014171833481,
      "grad_norm": 0.21825172007083893,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 67695
    },
    {
      "epoch": 59.96457041629761,
      "grad_norm": 0.20135079324245453,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 67700
    },
    {
      "epoch": 59.968999114260406,
      "grad_norm": 0.22235675156116486,
      "learning_rate": 1e-05,
      "loss": 1.0107,
      "step": 67705
    },
    {
      "epoch": 59.97342781222321,
      "grad_norm": 0.27094313502311707,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 67710
    },
    {
      "epoch": 59.97785651018601,
      "grad_norm": 0.2268366515636444,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 67715
    },
    {
      "epoch": 59.9822852081488,
      "grad_norm": 0.23001061379909515,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 67720
    },
    {
      "epoch": 59.9867139061116,
      "grad_norm": 0.24745343625545502,
      "learning_rate": 1e-05,
      "loss": 1.0217,
      "step": 67725
    },
    {
      "epoch": 59.991142604074405,
      "grad_norm": 0.24070978164672852,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 67730
    },
    {
      "epoch": 59.9955713020372,
      "grad_norm": 0.24640832841396332,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 67735
    },
    {
      "epoch": 60.0,
      "grad_norm": 0.25133851170539856,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 67740
    },
    {
      "epoch": 60.0044286979628,
      "grad_norm": 0.22156170010566711,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 67745
    },
    {
      "epoch": 60.008857395925595,
      "grad_norm": 0.2612433433532715,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 67750
    },
    {
      "epoch": 60.0132860938884,
      "grad_norm": 0.22379903495311737,
      "learning_rate": 1e-05,
      "loss": 1.022,
      "step": 67755
    },
    {
      "epoch": 60.0177147918512,
      "grad_norm": 0.2242935597896576,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 67760
    },
    {
      "epoch": 60.02214348981399,
      "grad_norm": 0.23966464400291443,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 67765
    },
    {
      "epoch": 60.02657218777679,
      "grad_norm": 0.21759504079818726,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 67770
    },
    {
      "epoch": 60.031000885739594,
      "grad_norm": 0.20940038561820984,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 67775
    },
    {
      "epoch": 60.03542958370239,
      "grad_norm": 0.24632173776626587,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 67780
    },
    {
      "epoch": 60.03985828166519,
      "grad_norm": 0.26936087012290955,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 67785
    },
    {
      "epoch": 60.04428697962799,
      "grad_norm": 0.23884139955043793,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 67790
    },
    {
      "epoch": 60.048715677590785,
      "grad_norm": 0.2195722460746765,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 67795
    },
    {
      "epoch": 60.053144375553586,
      "grad_norm": 0.23998527228832245,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 67800
    },
    {
      "epoch": 60.05757307351639,
      "grad_norm": 0.24039466679096222,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 67805
    },
    {
      "epoch": 60.06200177147919,
      "grad_norm": 0.23541930317878723,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 67810
    },
    {
      "epoch": 60.06643046944198,
      "grad_norm": 0.23188871145248413,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 67815
    },
    {
      "epoch": 60.070859167404784,
      "grad_norm": 0.2790294289588928,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 67820
    },
    {
      "epoch": 60.075287865367585,
      "grad_norm": 0.21344634890556335,
      "learning_rate": 1e-05,
      "loss": 1.0105,
      "step": 67825
    },
    {
      "epoch": 60.07971656333038,
      "grad_norm": 0.21285206079483032,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 67830
    },
    {
      "epoch": 60.08414526129318,
      "grad_norm": 0.2517503798007965,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 67835
    },
    {
      "epoch": 60.08857395925598,
      "grad_norm": 0.2339160442352295,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 67840
    },
    {
      "epoch": 60.093002657218776,
      "grad_norm": 0.31511059403419495,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 67845
    },
    {
      "epoch": 60.09743135518158,
      "grad_norm": 0.24189969897270203,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 67850
    },
    {
      "epoch": 60.10186005314438,
      "grad_norm": 0.27861279249191284,
      "learning_rate": 1e-05,
      "loss": 0.9087,
      "step": 67855
    },
    {
      "epoch": 60.10628875110717,
      "grad_norm": 0.22302816808223724,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 67860
    },
    {
      "epoch": 60.11071744906997,
      "grad_norm": 0.23074880242347717,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 67865
    },
    {
      "epoch": 60.115146147032775,
      "grad_norm": 0.24772585928440094,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 67870
    },
    {
      "epoch": 60.11957484499557,
      "grad_norm": 0.23156459629535675,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 67875
    },
    {
      "epoch": 60.12400354295837,
      "grad_norm": 0.3078056871891022,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 67880
    },
    {
      "epoch": 60.12843224092117,
      "grad_norm": 0.21533367037773132,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 67885
    },
    {
      "epoch": 60.132860938883965,
      "grad_norm": 0.2413732260465622,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 67890
    },
    {
      "epoch": 60.137289636846766,
      "grad_norm": 0.2541654109954834,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 67895
    },
    {
      "epoch": 60.14171833480957,
      "grad_norm": 0.30525773763656616,
      "learning_rate": 1e-05,
      "loss": 0.9198,
      "step": 67900
    },
    {
      "epoch": 60.14614703277236,
      "grad_norm": 0.276557981967926,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 67905
    },
    {
      "epoch": 60.15057573073516,
      "grad_norm": 0.2996844947338104,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 67910
    },
    {
      "epoch": 60.155004428697964,
      "grad_norm": 0.2943823039531708,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 67915
    },
    {
      "epoch": 60.15943312666076,
      "grad_norm": 0.2414475530385971,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 67920
    },
    {
      "epoch": 60.16386182462356,
      "grad_norm": 0.26345542073249817,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 67925
    },
    {
      "epoch": 60.16829052258636,
      "grad_norm": 0.2540135681629181,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 67930
    },
    {
      "epoch": 60.17271922054916,
      "grad_norm": 0.2111683338880539,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 67935
    },
    {
      "epoch": 60.177147918511956,
      "grad_norm": 0.25722554326057434,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 67940
    },
    {
      "epoch": 60.18157661647476,
      "grad_norm": 0.27004504203796387,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 67945
    },
    {
      "epoch": 60.18600531443756,
      "grad_norm": 0.21575520932674408,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 67950
    },
    {
      "epoch": 60.19043401240035,
      "grad_norm": 0.26626771688461304,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 67955
    },
    {
      "epoch": 60.194862710363154,
      "grad_norm": 0.22579853236675262,
      "learning_rate": 1e-05,
      "loss": 0.9028,
      "step": 67960
    },
    {
      "epoch": 60.199291408325955,
      "grad_norm": 0.22904513776302338,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 67965
    },
    {
      "epoch": 60.20372010628875,
      "grad_norm": 0.2567936182022095,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 67970
    },
    {
      "epoch": 60.20814880425155,
      "grad_norm": 0.2563650608062744,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 67975
    },
    {
      "epoch": 60.21257750221435,
      "grad_norm": 0.2461683303117752,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 67980
    },
    {
      "epoch": 60.217006200177146,
      "grad_norm": 0.22083435952663422,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 67985
    },
    {
      "epoch": 60.22143489813995,
      "grad_norm": 0.23430413007736206,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 67990
    },
    {
      "epoch": 60.22586359610275,
      "grad_norm": 0.2729833126068115,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 67995
    },
    {
      "epoch": 60.23029229406554,
      "grad_norm": 0.237448051571846,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 68000
    },
    {
      "epoch": 60.23472099202834,
      "grad_norm": 0.2860821783542633,
      "learning_rate": 1e-05,
      "loss": 1.0156,
      "step": 68005
    },
    {
      "epoch": 60.239149689991144,
      "grad_norm": 0.22187405824661255,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 68010
    },
    {
      "epoch": 60.24357838795394,
      "grad_norm": 0.21306779980659485,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 68015
    },
    {
      "epoch": 60.24800708591674,
      "grad_norm": 0.22761847078800201,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 68020
    },
    {
      "epoch": 60.25243578387954,
      "grad_norm": 0.2301272302865982,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 68025
    },
    {
      "epoch": 60.256864481842335,
      "grad_norm": 0.24498267471790314,
      "learning_rate": 1e-05,
      "loss": 0.9325,
      "step": 68030
    },
    {
      "epoch": 60.261293179805136,
      "grad_norm": 0.24395066499710083,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 68035
    },
    {
      "epoch": 60.26572187776794,
      "grad_norm": 0.27467772364616394,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 68040
    },
    {
      "epoch": 60.27015057573073,
      "grad_norm": 0.3105911612510681,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 68045
    },
    {
      "epoch": 60.27457927369353,
      "grad_norm": 0.24254311621189117,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 68050
    },
    {
      "epoch": 60.279007971656334,
      "grad_norm": 0.2539561688899994,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 68055
    },
    {
      "epoch": 60.283436669619135,
      "grad_norm": 0.2549138069152832,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 68060
    },
    {
      "epoch": 60.28786536758193,
      "grad_norm": 0.24409809708595276,
      "learning_rate": 1e-05,
      "loss": 1.0497,
      "step": 68065
    },
    {
      "epoch": 60.29229406554473,
      "grad_norm": 0.22572097182273865,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 68070
    },
    {
      "epoch": 60.29672276350753,
      "grad_norm": 0.24944044649600983,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 68075
    },
    {
      "epoch": 60.301151461470326,
      "grad_norm": 0.25153833627700806,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 68080
    },
    {
      "epoch": 60.30558015943313,
      "grad_norm": 0.24118006229400635,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 68085
    },
    {
      "epoch": 60.31000885739593,
      "grad_norm": 0.2610335946083069,
      "learning_rate": 1e-05,
      "loss": 1.0456,
      "step": 68090
    },
    {
      "epoch": 60.31443755535872,
      "grad_norm": 0.24294546246528625,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 68095
    },
    {
      "epoch": 60.318866253321524,
      "grad_norm": 0.2512407898902893,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 68100
    },
    {
      "epoch": 60.323294951284325,
      "grad_norm": 0.24164848029613495,
      "learning_rate": 1e-05,
      "loss": 0.92,
      "step": 68105
    },
    {
      "epoch": 60.32772364924712,
      "grad_norm": 0.2508971095085144,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 68110
    },
    {
      "epoch": 60.33215234720992,
      "grad_norm": 0.21086767315864563,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 68115
    },
    {
      "epoch": 60.33658104517272,
      "grad_norm": 0.2512386739253998,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 68120
    },
    {
      "epoch": 60.341009743135515,
      "grad_norm": 0.20561937987804413,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 68125
    },
    {
      "epoch": 60.34543844109832,
      "grad_norm": 0.23540417850017548,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 68130
    },
    {
      "epoch": 60.34986713906112,
      "grad_norm": 0.2360813468694687,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 68135
    },
    {
      "epoch": 60.35429583702391,
      "grad_norm": 0.230765238404274,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 68140
    },
    {
      "epoch": 60.35872453498671,
      "grad_norm": 0.23727911710739136,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 68145
    },
    {
      "epoch": 60.363153232949514,
      "grad_norm": 0.21590246260166168,
      "learning_rate": 1e-05,
      "loss": 0.916,
      "step": 68150
    },
    {
      "epoch": 60.36758193091231,
      "grad_norm": 0.2584565281867981,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 68155
    },
    {
      "epoch": 60.37201062887511,
      "grad_norm": 0.24334752559661865,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 68160
    },
    {
      "epoch": 60.37643932683791,
      "grad_norm": 0.2383350282907486,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 68165
    },
    {
      "epoch": 60.380868024800705,
      "grad_norm": 0.2528057098388672,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 68170
    },
    {
      "epoch": 60.385296722763506,
      "grad_norm": 0.2439594715833664,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 68175
    },
    {
      "epoch": 60.38972542072631,
      "grad_norm": 0.2513636648654938,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 68180
    },
    {
      "epoch": 60.39415411868911,
      "grad_norm": 0.22704513370990753,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 68185
    },
    {
      "epoch": 60.3985828166519,
      "grad_norm": 0.22684215009212494,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 68190
    },
    {
      "epoch": 60.403011514614704,
      "grad_norm": 0.22785650193691254,
      "learning_rate": 1e-05,
      "loss": 1.0427,
      "step": 68195
    },
    {
      "epoch": 60.407440212577505,
      "grad_norm": 0.22285433113574982,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 68200
    },
    {
      "epoch": 60.4118689105403,
      "grad_norm": 0.21892952919006348,
      "learning_rate": 1e-05,
      "loss": 0.9112,
      "step": 68205
    },
    {
      "epoch": 60.4162976085031,
      "grad_norm": 0.2665686011314392,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 68210
    },
    {
      "epoch": 60.4207263064659,
      "grad_norm": 0.22768713533878326,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 68215
    },
    {
      "epoch": 60.425155004428696,
      "grad_norm": 0.24704597890377045,
      "learning_rate": 1e-05,
      "loss": 0.8956,
      "step": 68220
    },
    {
      "epoch": 60.4295837023915,
      "grad_norm": 0.26176685094833374,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 68225
    },
    {
      "epoch": 60.4340124003543,
      "grad_norm": 0.23113496601581573,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 68230
    },
    {
      "epoch": 60.43844109831709,
      "grad_norm": 0.22440707683563232,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 68235
    },
    {
      "epoch": 60.44286979627989,
      "grad_norm": 0.23483550548553467,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 68240
    },
    {
      "epoch": 60.447298494242695,
      "grad_norm": 0.24866607785224915,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 68245
    },
    {
      "epoch": 60.45172719220549,
      "grad_norm": 0.2702188789844513,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 68250
    },
    {
      "epoch": 60.45615589016829,
      "grad_norm": 0.28178319334983826,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 68255
    },
    {
      "epoch": 60.46058458813109,
      "grad_norm": 0.2174639254808426,
      "learning_rate": 1e-05,
      "loss": 0.9462,
      "step": 68260
    },
    {
      "epoch": 60.465013286093885,
      "grad_norm": 0.2783980071544647,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 68265
    },
    {
      "epoch": 60.46944198405669,
      "grad_norm": 0.2589445412158966,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 68270
    },
    {
      "epoch": 60.47387068201949,
      "grad_norm": 0.2268449068069458,
      "learning_rate": 1e-05,
      "loss": 0.929,
      "step": 68275
    },
    {
      "epoch": 60.47829937998228,
      "grad_norm": 0.2240566462278366,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 68280
    },
    {
      "epoch": 60.48272807794508,
      "grad_norm": 0.207584947347641,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 68285
    },
    {
      "epoch": 60.487156775907884,
      "grad_norm": 0.23902861773967743,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 68290
    },
    {
      "epoch": 60.491585473870686,
      "grad_norm": 0.23124824464321136,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 68295
    },
    {
      "epoch": 60.49601417183348,
      "grad_norm": 0.21634815633296967,
      "learning_rate": 1e-05,
      "loss": 1.0253,
      "step": 68300
    },
    {
      "epoch": 60.50044286979628,
      "grad_norm": 0.27311766147613525,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 68305
    },
    {
      "epoch": 60.50487156775908,
      "grad_norm": 0.23520460724830627,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 68310
    },
    {
      "epoch": 60.509300265721876,
      "grad_norm": 0.23590585589408875,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 68315
    },
    {
      "epoch": 60.51372896368468,
      "grad_norm": 0.25773492455482483,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 68320
    },
    {
      "epoch": 60.51815766164748,
      "grad_norm": 0.2539404034614563,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 68325
    },
    {
      "epoch": 60.52258635961027,
      "grad_norm": 0.24535350501537323,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 68330
    },
    {
      "epoch": 60.527015057573074,
      "grad_norm": 0.22788722813129425,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 68335
    },
    {
      "epoch": 60.531443755535875,
      "grad_norm": 0.227084681391716,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 68340
    },
    {
      "epoch": 60.53587245349867,
      "grad_norm": 0.21022918820381165,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 68345
    },
    {
      "epoch": 60.54030115146147,
      "grad_norm": 0.27771732211112976,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 68350
    },
    {
      "epoch": 60.54472984942427,
      "grad_norm": 0.2801872789859772,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 68355
    },
    {
      "epoch": 60.549158547387066,
      "grad_norm": 0.27351677417755127,
      "learning_rate": 1e-05,
      "loss": 1.0152,
      "step": 68360
    },
    {
      "epoch": 60.55358724534987,
      "grad_norm": 0.2551421821117401,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 68365
    },
    {
      "epoch": 60.55801594331267,
      "grad_norm": 0.26367899775505066,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 68370
    },
    {
      "epoch": 60.56244464127546,
      "grad_norm": 0.2702999711036682,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 68375
    },
    {
      "epoch": 60.56687333923826,
      "grad_norm": 0.21274510025978088,
      "learning_rate": 1e-05,
      "loss": 1.0235,
      "step": 68380
    },
    {
      "epoch": 60.571302037201065,
      "grad_norm": 0.23331323266029358,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 68385
    },
    {
      "epoch": 60.57573073516386,
      "grad_norm": 0.257802277803421,
      "learning_rate": 1e-05,
      "loss": 0.9432,
      "step": 68390
    },
    {
      "epoch": 60.58015943312666,
      "grad_norm": 0.22249586880207062,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 68395
    },
    {
      "epoch": 60.58458813108946,
      "grad_norm": 0.24841439723968506,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 68400
    },
    {
      "epoch": 60.589016829052255,
      "grad_norm": 0.2400336116552353,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 68405
    },
    {
      "epoch": 60.59344552701506,
      "grad_norm": 0.22133579850196838,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 68410
    },
    {
      "epoch": 60.59787422497786,
      "grad_norm": 0.2465466558933258,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 68415
    },
    {
      "epoch": 60.60230292294066,
      "grad_norm": 0.25301554799079895,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 68420
    },
    {
      "epoch": 60.60673162090345,
      "grad_norm": 0.26860591769218445,
      "learning_rate": 1e-05,
      "loss": 0.9186,
      "step": 68425
    },
    {
      "epoch": 60.611160318866254,
      "grad_norm": 0.20792792737483978,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 68430
    },
    {
      "epoch": 60.615589016829055,
      "grad_norm": 0.22394777834415436,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 68435
    },
    {
      "epoch": 60.62001771479185,
      "grad_norm": 0.24083735048770905,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 68440
    },
    {
      "epoch": 60.62444641275465,
      "grad_norm": 0.2446780651807785,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 68445
    },
    {
      "epoch": 60.62887511071745,
      "grad_norm": 0.22034183144569397,
      "learning_rate": 1e-05,
      "loss": 1.0409,
      "step": 68450
    },
    {
      "epoch": 60.633303808680246,
      "grad_norm": 0.24650482833385468,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 68455
    },
    {
      "epoch": 60.63773250664305,
      "grad_norm": 0.23790277540683746,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 68460
    },
    {
      "epoch": 60.64216120460585,
      "grad_norm": 0.32248207926750183,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 68465
    },
    {
      "epoch": 60.64658990256864,
      "grad_norm": 0.22248025238513947,
      "learning_rate": 1e-05,
      "loss": 0.9273,
      "step": 68470
    },
    {
      "epoch": 60.651018600531444,
      "grad_norm": 0.23740394413471222,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 68475
    },
    {
      "epoch": 60.655447298494245,
      "grad_norm": 0.250934362411499,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 68480
    },
    {
      "epoch": 60.65987599645704,
      "grad_norm": 0.24006050825119019,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 68485
    },
    {
      "epoch": 60.66430469441984,
      "grad_norm": 0.23684562742710114,
      "learning_rate": 1e-05,
      "loss": 0.9033,
      "step": 68490
    },
    {
      "epoch": 60.66873339238264,
      "grad_norm": 0.2845231294631958,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 68495
    },
    {
      "epoch": 60.673162090345436,
      "grad_norm": 0.25366225838661194,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 68500
    },
    {
      "epoch": 60.67759078830824,
      "grad_norm": 0.2804090678691864,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 68505
    },
    {
      "epoch": 60.68201948627104,
      "grad_norm": 0.21341632306575775,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 68510
    },
    {
      "epoch": 60.68644818423383,
      "grad_norm": 0.23643432557582855,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 68515
    },
    {
      "epoch": 60.69087688219663,
      "grad_norm": 0.30427131056785583,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 68520
    },
    {
      "epoch": 60.695305580159435,
      "grad_norm": 0.2572616934776306,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 68525
    },
    {
      "epoch": 60.69973427812223,
      "grad_norm": 0.2615016996860504,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 68530
    },
    {
      "epoch": 60.70416297608503,
      "grad_norm": 0.2761181592941284,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 68535
    },
    {
      "epoch": 60.70859167404783,
      "grad_norm": 0.21619199216365814,
      "learning_rate": 1e-05,
      "loss": 1.0539,
      "step": 68540
    },
    {
      "epoch": 60.71302037201063,
      "grad_norm": 0.2630160450935364,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 68545
    },
    {
      "epoch": 60.717449069973426,
      "grad_norm": 0.21130770444869995,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 68550
    },
    {
      "epoch": 60.72187776793623,
      "grad_norm": 0.22268280386924744,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 68555
    },
    {
      "epoch": 60.72630646589903,
      "grad_norm": 0.32420459389686584,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 68560
    },
    {
      "epoch": 60.73073516386182,
      "grad_norm": 0.24895937740802765,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 68565
    },
    {
      "epoch": 60.735163861824624,
      "grad_norm": 0.2566019594669342,
      "learning_rate": 1e-05,
      "loss": 0.9228,
      "step": 68570
    },
    {
      "epoch": 60.739592559787425,
      "grad_norm": 0.22634539008140564,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 68575
    },
    {
      "epoch": 60.74402125775022,
      "grad_norm": 0.26682814955711365,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 68580
    },
    {
      "epoch": 60.74844995571302,
      "grad_norm": 0.275595098733902,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 68585
    },
    {
      "epoch": 60.75287865367582,
      "grad_norm": 0.23859696090221405,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 68590
    },
    {
      "epoch": 60.757307351638616,
      "grad_norm": 0.2459527850151062,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 68595
    },
    {
      "epoch": 60.76173604960142,
      "grad_norm": 0.24977359175682068,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 68600
    },
    {
      "epoch": 60.76616474756422,
      "grad_norm": 0.2629532217979431,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 68605
    },
    {
      "epoch": 60.77059344552701,
      "grad_norm": 0.28251326084136963,
      "learning_rate": 1e-05,
      "loss": 0.9251,
      "step": 68610
    },
    {
      "epoch": 60.775022143489814,
      "grad_norm": 0.26810160279273987,
      "learning_rate": 1e-05,
      "loss": 0.9301,
      "step": 68615
    },
    {
      "epoch": 60.779450841452615,
      "grad_norm": 0.28561556339263916,
      "learning_rate": 1e-05,
      "loss": 0.9182,
      "step": 68620
    },
    {
      "epoch": 60.78387953941541,
      "grad_norm": 0.2667543590068817,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 68625
    },
    {
      "epoch": 60.78830823737821,
      "grad_norm": 0.22049254179000854,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 68630
    },
    {
      "epoch": 60.79273693534101,
      "grad_norm": 0.23839691281318665,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 68635
    },
    {
      "epoch": 60.797165633303806,
      "grad_norm": 0.2234589159488678,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 68640
    },
    {
      "epoch": 60.80159433126661,
      "grad_norm": 0.21641317009925842,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 68645
    },
    {
      "epoch": 60.80602302922941,
      "grad_norm": 0.21696758270263672,
      "learning_rate": 1e-05,
      "loss": 0.9275,
      "step": 68650
    },
    {
      "epoch": 60.8104517271922,
      "grad_norm": 0.2315734177827835,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 68655
    },
    {
      "epoch": 60.814880425155,
      "grad_norm": 0.2406282126903534,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 68660
    },
    {
      "epoch": 60.819309123117804,
      "grad_norm": 0.22853884100914001,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 68665
    },
    {
      "epoch": 60.823737821080606,
      "grad_norm": 0.2348943054676056,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 68670
    },
    {
      "epoch": 60.8281665190434,
      "grad_norm": 0.25485584139823914,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 68675
    },
    {
      "epoch": 60.8325952170062,
      "grad_norm": 0.2539708614349365,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 68680
    },
    {
      "epoch": 60.837023914969,
      "grad_norm": 0.26831015944480896,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 68685
    },
    {
      "epoch": 60.841452612931796,
      "grad_norm": 0.23185327649116516,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 68690
    },
    {
      "epoch": 60.8458813108946,
      "grad_norm": 0.22609320282936096,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 68695
    },
    {
      "epoch": 60.8503100088574,
      "grad_norm": 0.21681274473667145,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 68700
    },
    {
      "epoch": 60.85473870682019,
      "grad_norm": 0.2144000083208084,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 68705
    },
    {
      "epoch": 60.859167404782994,
      "grad_norm": 0.24931775033473969,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 68710
    },
    {
      "epoch": 60.863596102745795,
      "grad_norm": 0.28115612268447876,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 68715
    },
    {
      "epoch": 60.86802480070859,
      "grad_norm": 0.22621887922286987,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 68720
    },
    {
      "epoch": 60.87245349867139,
      "grad_norm": 0.22621527314186096,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 68725
    },
    {
      "epoch": 60.87688219663419,
      "grad_norm": 0.26475536823272705,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 68730
    },
    {
      "epoch": 60.881310894596986,
      "grad_norm": 0.25169700384140015,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 68735
    },
    {
      "epoch": 60.88573959255979,
      "grad_norm": 0.22355714440345764,
      "learning_rate": 1e-05,
      "loss": 1.0279,
      "step": 68740
    },
    {
      "epoch": 60.89016829052259,
      "grad_norm": 0.23615559935569763,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 68745
    },
    {
      "epoch": 60.89459698848538,
      "grad_norm": 0.23883506655693054,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 68750
    },
    {
      "epoch": 60.899025686448184,
      "grad_norm": 0.23506960272789001,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 68755
    },
    {
      "epoch": 60.903454384410985,
      "grad_norm": 0.27814409136772156,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 68760
    },
    {
      "epoch": 60.90788308237378,
      "grad_norm": 0.2074914276599884,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 68765
    },
    {
      "epoch": 60.91231178033658,
      "grad_norm": 0.2358940690755844,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 68770
    },
    {
      "epoch": 60.91674047829938,
      "grad_norm": 0.21459928154945374,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 68775
    },
    {
      "epoch": 60.921169176262175,
      "grad_norm": 0.2349170446395874,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 68780
    },
    {
      "epoch": 60.92559787422498,
      "grad_norm": 0.22285522520542145,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 68785
    },
    {
      "epoch": 60.93002657218778,
      "grad_norm": 0.23253588378429413,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 68790
    },
    {
      "epoch": 60.93445527015058,
      "grad_norm": 0.22562578320503235,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 68795
    },
    {
      "epoch": 60.93888396811337,
      "grad_norm": 0.24191094934940338,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 68800
    },
    {
      "epoch": 60.943312666076174,
      "grad_norm": 0.22991205751895905,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 68805
    },
    {
      "epoch": 60.947741364038976,
      "grad_norm": 0.22426697611808777,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 68810
    },
    {
      "epoch": 60.95217006200177,
      "grad_norm": 0.25657710433006287,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 68815
    },
    {
      "epoch": 60.95659875996457,
      "grad_norm": 0.23975779116153717,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 68820
    },
    {
      "epoch": 60.96102745792737,
      "grad_norm": 0.22793243825435638,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 68825
    },
    {
      "epoch": 60.965456155890166,
      "grad_norm": 0.24359433352947235,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 68830
    },
    {
      "epoch": 60.96988485385297,
      "grad_norm": 0.24737828969955444,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 68835
    },
    {
      "epoch": 60.97431355181577,
      "grad_norm": 0.2817375659942627,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 68840
    },
    {
      "epoch": 60.97874224977856,
      "grad_norm": 0.24856781959533691,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 68845
    },
    {
      "epoch": 60.983170947741364,
      "grad_norm": 0.253393292427063,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 68850
    },
    {
      "epoch": 60.987599645704165,
      "grad_norm": 0.2796756327152252,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 68855
    },
    {
      "epoch": 60.99202834366696,
      "grad_norm": 0.2869615852832794,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 68860
    },
    {
      "epoch": 60.99645704162976,
      "grad_norm": 0.24789300560951233,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 68865
    },
    {
      "epoch": 61.00088573959256,
      "grad_norm": 0.2543843388557434,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 68870
    },
    {
      "epoch": 61.005314437555356,
      "grad_norm": 0.24900545179843903,
      "learning_rate": 1e-05,
      "loss": 0.91,
      "step": 68875
    },
    {
      "epoch": 61.00974313551816,
      "grad_norm": 0.2447466254234314,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 68880
    },
    {
      "epoch": 61.01417183348096,
      "grad_norm": 0.21506695449352264,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 68885
    },
    {
      "epoch": 61.01860053144375,
      "grad_norm": 0.22232094407081604,
      "learning_rate": 1e-05,
      "loss": 0.8969,
      "step": 68890
    },
    {
      "epoch": 61.02302922940655,
      "grad_norm": 0.2680467367172241,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 68895
    },
    {
      "epoch": 61.027457927369355,
      "grad_norm": 0.2284451574087143,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 68900
    },
    {
      "epoch": 61.03188662533215,
      "grad_norm": 0.26245033740997314,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 68905
    },
    {
      "epoch": 61.03631532329495,
      "grad_norm": 0.2640277147293091,
      "learning_rate": 1e-05,
      "loss": 0.9396,
      "step": 68910
    },
    {
      "epoch": 61.04074402125775,
      "grad_norm": 0.18897074460983276,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 68915
    },
    {
      "epoch": 61.04517271922055,
      "grad_norm": 0.2382601499557495,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 68920
    },
    {
      "epoch": 61.04960141718335,
      "grad_norm": 0.24305549263954163,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 68925
    },
    {
      "epoch": 61.05403011514615,
      "grad_norm": 0.2528451085090637,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 68930
    },
    {
      "epoch": 61.05845881310895,
      "grad_norm": 0.22123828530311584,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 68935
    },
    {
      "epoch": 61.06288751107174,
      "grad_norm": 0.20078840851783752,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 68940
    },
    {
      "epoch": 61.067316209034544,
      "grad_norm": 0.27237752079963684,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 68945
    },
    {
      "epoch": 61.071744906997345,
      "grad_norm": 0.24376441538333893,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 68950
    },
    {
      "epoch": 61.07617360496014,
      "grad_norm": 0.256112664937973,
      "learning_rate": 1e-05,
      "loss": 0.9346,
      "step": 68955
    },
    {
      "epoch": 61.08060230292294,
      "grad_norm": 0.21611861884593964,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 68960
    },
    {
      "epoch": 61.08503100088574,
      "grad_norm": 0.26536816358566284,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 68965
    },
    {
      "epoch": 61.089459698848536,
      "grad_norm": 0.27056974172592163,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 68970
    },
    {
      "epoch": 61.09388839681134,
      "grad_norm": 0.24422673881053925,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 68975
    },
    {
      "epoch": 61.09831709477414,
      "grad_norm": 0.23578929901123047,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 68980
    },
    {
      "epoch": 61.10274579273693,
      "grad_norm": 0.22743748128414154,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 68985
    },
    {
      "epoch": 61.107174490699734,
      "grad_norm": 0.23487642407417297,
      "learning_rate": 1e-05,
      "loss": 0.9241,
      "step": 68990
    },
    {
      "epoch": 61.111603188662535,
      "grad_norm": 0.21708784997463226,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 68995
    },
    {
      "epoch": 61.11603188662533,
      "grad_norm": 0.24450981616973877,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 69000
    },
    {
      "epoch": 61.12046058458813,
      "grad_norm": 0.21709676086902618,
      "learning_rate": 1e-05,
      "loss": 0.9364,
      "step": 69005
    },
    {
      "epoch": 61.12488928255093,
      "grad_norm": 0.2173592448234558,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 69010
    },
    {
      "epoch": 61.129317980513726,
      "grad_norm": 0.2793889045715332,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 69015
    },
    {
      "epoch": 61.13374667847653,
      "grad_norm": 0.22945021092891693,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 69020
    },
    {
      "epoch": 61.13817537643933,
      "grad_norm": 0.21406635642051697,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 69025
    },
    {
      "epoch": 61.14260407440213,
      "grad_norm": 0.21664361655712128,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 69030
    },
    {
      "epoch": 61.14703277236492,
      "grad_norm": 0.2671864926815033,
      "learning_rate": 1e-05,
      "loss": 0.9462,
      "step": 69035
    },
    {
      "epoch": 61.151461470327725,
      "grad_norm": 0.2216176688671112,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 69040
    },
    {
      "epoch": 61.155890168290526,
      "grad_norm": 0.2204846739768982,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 69045
    },
    {
      "epoch": 61.16031886625332,
      "grad_norm": 0.2371831089258194,
      "learning_rate": 1e-05,
      "loss": 0.9219,
      "step": 69050
    },
    {
      "epoch": 61.16474756421612,
      "grad_norm": 0.22620365023612976,
      "learning_rate": 1e-05,
      "loss": 1.0386,
      "step": 69055
    },
    {
      "epoch": 61.16917626217892,
      "grad_norm": 0.22587302327156067,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 69060
    },
    {
      "epoch": 61.173604960141716,
      "grad_norm": 0.2310447245836258,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 69065
    },
    {
      "epoch": 61.17803365810452,
      "grad_norm": 0.26117467880249023,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 69070
    },
    {
      "epoch": 61.18246235606732,
      "grad_norm": 0.28922730684280396,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 69075
    },
    {
      "epoch": 61.18689105403011,
      "grad_norm": 0.267564594745636,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 69080
    },
    {
      "epoch": 61.191319751992914,
      "grad_norm": 0.24295002222061157,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 69085
    },
    {
      "epoch": 61.195748449955715,
      "grad_norm": 0.24249619245529175,
      "learning_rate": 1e-05,
      "loss": 0.9218,
      "step": 69090
    },
    {
      "epoch": 61.20017714791851,
      "grad_norm": 0.29141199588775635,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 69095
    },
    {
      "epoch": 61.20460584588131,
      "grad_norm": 0.2403004914522171,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 69100
    },
    {
      "epoch": 61.20903454384411,
      "grad_norm": 0.24814929068088531,
      "learning_rate": 1e-05,
      "loss": 0.9364,
      "step": 69105
    },
    {
      "epoch": 61.213463241806906,
      "grad_norm": 0.21212254464626312,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 69110
    },
    {
      "epoch": 61.21789193976971,
      "grad_norm": 0.23725003004074097,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 69115
    },
    {
      "epoch": 61.22232063773251,
      "grad_norm": 0.2563864290714264,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 69120
    },
    {
      "epoch": 61.2267493356953,
      "grad_norm": 0.2553084194660187,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 69125
    },
    {
      "epoch": 61.231178033658104,
      "grad_norm": 0.23428797721862793,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 69130
    },
    {
      "epoch": 61.235606731620905,
      "grad_norm": 0.23866522312164307,
      "learning_rate": 1e-05,
      "loss": 1.0095,
      "step": 69135
    },
    {
      "epoch": 61.2400354295837,
      "grad_norm": 0.23985204100608826,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 69140
    },
    {
      "epoch": 61.2444641275465,
      "grad_norm": 0.25277647376060486,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 69145
    },
    {
      "epoch": 61.2488928255093,
      "grad_norm": 0.2376217246055603,
      "learning_rate": 1e-05,
      "loss": 0.9142,
      "step": 69150
    },
    {
      "epoch": 61.2533215234721,
      "grad_norm": 0.24289919435977936,
      "learning_rate": 1e-05,
      "loss": 0.9153,
      "step": 69155
    },
    {
      "epoch": 61.2577502214349,
      "grad_norm": 0.22350744903087616,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 69160
    },
    {
      "epoch": 61.2621789193977,
      "grad_norm": 0.23198938369750977,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 69165
    },
    {
      "epoch": 61.2666076173605,
      "grad_norm": 0.23566776514053345,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 69170
    },
    {
      "epoch": 61.27103631532329,
      "grad_norm": 0.2958187460899353,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 69175
    },
    {
      "epoch": 61.275465013286095,
      "grad_norm": 0.22491714358329773,
      "learning_rate": 1e-05,
      "loss": 0.9145,
      "step": 69180
    },
    {
      "epoch": 61.279893711248896,
      "grad_norm": 0.24291083216667175,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 69185
    },
    {
      "epoch": 61.28432240921169,
      "grad_norm": 0.22924542427062988,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 69190
    },
    {
      "epoch": 61.28875110717449,
      "grad_norm": 0.25897637009620667,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 69195
    },
    {
      "epoch": 61.29317980513729,
      "grad_norm": 0.25369951128959656,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 69200
    },
    {
      "epoch": 61.297608503100086,
      "grad_norm": 0.22002443671226501,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 69205
    },
    {
      "epoch": 61.30203720106289,
      "grad_norm": 0.22557134926319122,
      "learning_rate": 1e-05,
      "loss": 0.9022,
      "step": 69210
    },
    {
      "epoch": 61.30646589902569,
      "grad_norm": 0.2160862535238266,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 69215
    },
    {
      "epoch": 61.31089459698848,
      "grad_norm": 0.21808575093746185,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 69220
    },
    {
      "epoch": 61.315323294951284,
      "grad_norm": 0.25484880805015564,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 69225
    },
    {
      "epoch": 61.319751992914085,
      "grad_norm": 0.2583205997943878,
      "learning_rate": 1e-05,
      "loss": 0.8996,
      "step": 69230
    },
    {
      "epoch": 61.32418069087688,
      "grad_norm": 0.23051223158836365,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 69235
    },
    {
      "epoch": 61.32860938883968,
      "grad_norm": 0.24472104012966156,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 69240
    },
    {
      "epoch": 61.33303808680248,
      "grad_norm": 0.21966078877449036,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 69245
    },
    {
      "epoch": 61.337466784765276,
      "grad_norm": 0.24408718943595886,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 69250
    },
    {
      "epoch": 61.34189548272808,
      "grad_norm": 0.24638210237026215,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 69255
    },
    {
      "epoch": 61.34632418069088,
      "grad_norm": 0.22625134885311127,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 69260
    },
    {
      "epoch": 61.35075287865367,
      "grad_norm": 0.264849454164505,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 69265
    },
    {
      "epoch": 61.355181576616474,
      "grad_norm": 0.23788122832775116,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 69270
    },
    {
      "epoch": 61.359610274579275,
      "grad_norm": 0.28192004561424255,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 69275
    },
    {
      "epoch": 61.364038972542076,
      "grad_norm": 0.226658433675766,
      "learning_rate": 1e-05,
      "loss": 0.9119,
      "step": 69280
    },
    {
      "epoch": 61.36846767050487,
      "grad_norm": 0.22849087417125702,
      "learning_rate": 1e-05,
      "loss": 0.9338,
      "step": 69285
    },
    {
      "epoch": 61.37289636846767,
      "grad_norm": 0.2611321210861206,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 69290
    },
    {
      "epoch": 61.37732506643047,
      "grad_norm": 0.2517118752002716,
      "learning_rate": 1e-05,
      "loss": 0.9206,
      "step": 69295
    },
    {
      "epoch": 61.38175376439327,
      "grad_norm": 0.2303445190191269,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 69300
    },
    {
      "epoch": 61.38618246235607,
      "grad_norm": 0.23463520407676697,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 69305
    },
    {
      "epoch": 61.39061116031887,
      "grad_norm": 0.23300568759441376,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 69310
    },
    {
      "epoch": 61.39503985828166,
      "grad_norm": 0.24406839907169342,
      "learning_rate": 1e-05,
      "loss": 0.8969,
      "step": 69315
    },
    {
      "epoch": 61.399468556244464,
      "grad_norm": 0.23840808868408203,
      "learning_rate": 1e-05,
      "loss": 0.921,
      "step": 69320
    },
    {
      "epoch": 61.403897254207266,
      "grad_norm": 0.26248326897621155,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 69325
    },
    {
      "epoch": 61.40832595217006,
      "grad_norm": 0.2521049380302429,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 69330
    },
    {
      "epoch": 61.41275465013286,
      "grad_norm": 0.24987314641475677,
      "learning_rate": 1e-05,
      "loss": 0.9227,
      "step": 69335
    },
    {
      "epoch": 61.41718334809566,
      "grad_norm": 0.2493465393781662,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 69340
    },
    {
      "epoch": 61.421612046058456,
      "grad_norm": 0.26827895641326904,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 69345
    },
    {
      "epoch": 61.42604074402126,
      "grad_norm": 0.22761604189872742,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 69350
    },
    {
      "epoch": 61.43046944198406,
      "grad_norm": 0.2533360719680786,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 69355
    },
    {
      "epoch": 61.43489813994685,
      "grad_norm": 0.2639541029930115,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 69360
    },
    {
      "epoch": 61.439326837909654,
      "grad_norm": 0.24893233180046082,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 69365
    },
    {
      "epoch": 61.443755535872455,
      "grad_norm": 0.2455807626247406,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 69370
    },
    {
      "epoch": 61.44818423383525,
      "grad_norm": 0.3321240544319153,
      "learning_rate": 1e-05,
      "loss": 0.9097,
      "step": 69375
    },
    {
      "epoch": 61.45261293179805,
      "grad_norm": 0.2909684181213379,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 69380
    },
    {
      "epoch": 61.45704162976085,
      "grad_norm": 0.2340477854013443,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 69385
    },
    {
      "epoch": 61.461470327723646,
      "grad_norm": 0.25586262345314026,
      "learning_rate": 1e-05,
      "loss": 0.9177,
      "step": 69390
    },
    {
      "epoch": 61.46589902568645,
      "grad_norm": 0.24378307163715363,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 69395
    },
    {
      "epoch": 61.47032772364925,
      "grad_norm": 0.25312042236328125,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 69400
    },
    {
      "epoch": 61.47475642161205,
      "grad_norm": 0.22455553710460663,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 69405
    },
    {
      "epoch": 61.479185119574844,
      "grad_norm": 0.2228417992591858,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 69410
    },
    {
      "epoch": 61.483613817537645,
      "grad_norm": 0.23340564966201782,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 69415
    },
    {
      "epoch": 61.488042515500446,
      "grad_norm": 0.2407195121049881,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 69420
    },
    {
      "epoch": 61.49247121346324,
      "grad_norm": 0.23649799823760986,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 69425
    },
    {
      "epoch": 61.49689991142604,
      "grad_norm": 0.2822832763195038,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 69430
    },
    {
      "epoch": 61.50132860938884,
      "grad_norm": 0.21479719877243042,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 69435
    },
    {
      "epoch": 61.50575730735164,
      "grad_norm": 0.24879147112369537,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 69440
    },
    {
      "epoch": 61.51018600531444,
      "grad_norm": 0.2494964450597763,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 69445
    },
    {
      "epoch": 61.51461470327724,
      "grad_norm": 0.26404666900634766,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 69450
    },
    {
      "epoch": 61.51904340124003,
      "grad_norm": 0.2151639759540558,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 69455
    },
    {
      "epoch": 61.523472099202834,
      "grad_norm": 0.21767747402191162,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 69460
    },
    {
      "epoch": 61.527900797165636,
      "grad_norm": 0.26045215129852295,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 69465
    },
    {
      "epoch": 61.53232949512843,
      "grad_norm": 0.23753869533538818,
      "learning_rate": 1e-05,
      "loss": 0.9204,
      "step": 69470
    },
    {
      "epoch": 61.53675819309123,
      "grad_norm": 0.21348710358142853,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 69475
    },
    {
      "epoch": 61.54118689105403,
      "grad_norm": 0.22218547761440277,
      "learning_rate": 1e-05,
      "loss": 0.941,
      "step": 69480
    },
    {
      "epoch": 61.545615589016826,
      "grad_norm": 0.23249605298042297,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 69485
    },
    {
      "epoch": 61.55004428697963,
      "grad_norm": 0.25759604573249817,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 69490
    },
    {
      "epoch": 61.55447298494243,
      "grad_norm": 0.2337723970413208,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 69495
    },
    {
      "epoch": 61.55890168290522,
      "grad_norm": 0.20633439719676971,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 69500
    },
    {
      "epoch": 61.563330380868024,
      "grad_norm": 0.24570763111114502,
      "learning_rate": 1e-05,
      "loss": 1.0241,
      "step": 69505
    },
    {
      "epoch": 61.567759078830825,
      "grad_norm": 0.25317081809043884,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 69510
    },
    {
      "epoch": 61.57218777679362,
      "grad_norm": 0.2093810886144638,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 69515
    },
    {
      "epoch": 61.57661647475642,
      "grad_norm": 0.2280571311712265,
      "learning_rate": 1e-05,
      "loss": 1.0272,
      "step": 69520
    },
    {
      "epoch": 61.58104517271922,
      "grad_norm": 0.23558080196380615,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 69525
    },
    {
      "epoch": 61.58547387068202,
      "grad_norm": 0.24911385774612427,
      "learning_rate": 1e-05,
      "loss": 0.912,
      "step": 69530
    },
    {
      "epoch": 61.58990256864482,
      "grad_norm": 0.24621254205703735,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 69535
    },
    {
      "epoch": 61.59433126660762,
      "grad_norm": 0.24742159247398376,
      "learning_rate": 1e-05,
      "loss": 0.9204,
      "step": 69540
    },
    {
      "epoch": 61.59875996457042,
      "grad_norm": 0.2405882477760315,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 69545
    },
    {
      "epoch": 61.60318866253321,
      "grad_norm": 0.24292337894439697,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 69550
    },
    {
      "epoch": 61.607617360496015,
      "grad_norm": 0.23934440314769745,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 69555
    },
    {
      "epoch": 61.612046058458816,
      "grad_norm": 0.23483149707317352,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 69560
    },
    {
      "epoch": 61.61647475642161,
      "grad_norm": 0.23129723966121674,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 69565
    },
    {
      "epoch": 61.62090345438441,
      "grad_norm": 0.22783353924751282,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 69570
    },
    {
      "epoch": 61.62533215234721,
      "grad_norm": 0.2411622554063797,
      "learning_rate": 1e-05,
      "loss": 0.9297,
      "step": 69575
    },
    {
      "epoch": 61.62976085031001,
      "grad_norm": 0.23716354370117188,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 69580
    },
    {
      "epoch": 61.63418954827281,
      "grad_norm": 0.22356528043746948,
      "learning_rate": 1e-05,
      "loss": 1.0344,
      "step": 69585
    },
    {
      "epoch": 61.63861824623561,
      "grad_norm": 0.25518184900283813,
      "learning_rate": 1e-05,
      "loss": 1.0172,
      "step": 69590
    },
    {
      "epoch": 61.6430469441984,
      "grad_norm": 0.20435483753681183,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 69595
    },
    {
      "epoch": 61.647475642161204,
      "grad_norm": 0.2521303594112396,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 69600
    },
    {
      "epoch": 61.651904340124005,
      "grad_norm": 0.255447655916214,
      "learning_rate": 1e-05,
      "loss": 0.9267,
      "step": 69605
    },
    {
      "epoch": 61.6563330380868,
      "grad_norm": 0.23234157264232635,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 69610
    },
    {
      "epoch": 61.6607617360496,
      "grad_norm": 0.30756673216819763,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 69615
    },
    {
      "epoch": 61.6651904340124,
      "grad_norm": 0.21867334842681885,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 69620
    },
    {
      "epoch": 61.669619131975196,
      "grad_norm": 0.22728191316127777,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 69625
    },
    {
      "epoch": 61.674047829938,
      "grad_norm": 0.27508410811424255,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 69630
    },
    {
      "epoch": 61.6784765279008,
      "grad_norm": 0.23752889037132263,
      "learning_rate": 1e-05,
      "loss": 0.9043,
      "step": 69635
    },
    {
      "epoch": 61.68290522586359,
      "grad_norm": 0.2650087773799896,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 69640
    },
    {
      "epoch": 61.687333923826394,
      "grad_norm": 0.2092810422182083,
      "learning_rate": 1e-05,
      "loss": 0.9258,
      "step": 69645
    },
    {
      "epoch": 61.691762621789195,
      "grad_norm": 0.24645790457725525,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 69650
    },
    {
      "epoch": 61.696191319751996,
      "grad_norm": 0.2182760089635849,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 69655
    },
    {
      "epoch": 61.70062001771479,
      "grad_norm": 0.2168237566947937,
      "learning_rate": 1e-05,
      "loss": 1.017,
      "step": 69660
    },
    {
      "epoch": 61.70504871567759,
      "grad_norm": 0.24498319625854492,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 69665
    },
    {
      "epoch": 61.70947741364039,
      "grad_norm": 0.23338940739631653,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 69670
    },
    {
      "epoch": 61.71390611160319,
      "grad_norm": 0.2907017469406128,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 69675
    },
    {
      "epoch": 61.71833480956599,
      "grad_norm": 0.23821742832660675,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 69680
    },
    {
      "epoch": 61.72276350752879,
      "grad_norm": 0.2678869366645813,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 69685
    },
    {
      "epoch": 61.72719220549158,
      "grad_norm": 0.22680503129959106,
      "learning_rate": 1e-05,
      "loss": 0.942,
      "step": 69690
    },
    {
      "epoch": 61.731620903454385,
      "grad_norm": 0.18340849876403809,
      "learning_rate": 1e-05,
      "loss": 0.9277,
      "step": 69695
    },
    {
      "epoch": 61.736049601417186,
      "grad_norm": 0.2458120435476303,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 69700
    },
    {
      "epoch": 61.74047829937998,
      "grad_norm": 0.2705990672111511,
      "learning_rate": 1e-05,
      "loss": 1.0319,
      "step": 69705
    },
    {
      "epoch": 61.74490699734278,
      "grad_norm": 0.24120868742465973,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 69710
    },
    {
      "epoch": 61.74933569530558,
      "grad_norm": 0.20870280265808105,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 69715
    },
    {
      "epoch": 61.753764393268376,
      "grad_norm": 0.21546325087547302,
      "learning_rate": 1e-05,
      "loss": 1.0359,
      "step": 69720
    },
    {
      "epoch": 61.75819309123118,
      "grad_norm": 0.2226325422525406,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 69725
    },
    {
      "epoch": 61.76262178919398,
      "grad_norm": 0.250235915184021,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 69730
    },
    {
      "epoch": 61.76705048715677,
      "grad_norm": 0.24084600806236267,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 69735
    },
    {
      "epoch": 61.771479185119574,
      "grad_norm": 0.24350860714912415,
      "learning_rate": 1e-05,
      "loss": 0.9333,
      "step": 69740
    },
    {
      "epoch": 61.775907883082375,
      "grad_norm": 0.24917399883270264,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 69745
    },
    {
      "epoch": 61.78033658104517,
      "grad_norm": 0.23366564512252808,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 69750
    },
    {
      "epoch": 61.78476527900797,
      "grad_norm": 0.24146410822868347,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 69755
    },
    {
      "epoch": 61.78919397697077,
      "grad_norm": 0.22105148434638977,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 69760
    },
    {
      "epoch": 61.79362267493357,
      "grad_norm": 0.21910695731639862,
      "learning_rate": 1e-05,
      "loss": 0.8971,
      "step": 69765
    },
    {
      "epoch": 61.79805137289637,
      "grad_norm": 0.22292330861091614,
      "learning_rate": 1e-05,
      "loss": 1.0111,
      "step": 69770
    },
    {
      "epoch": 61.80248007085917,
      "grad_norm": 0.22276896238327026,
      "learning_rate": 1e-05,
      "loss": 0.9269,
      "step": 69775
    },
    {
      "epoch": 61.80690876882197,
      "grad_norm": 0.2110166847705841,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 69780
    },
    {
      "epoch": 61.811337466784764,
      "grad_norm": 0.22309884428977966,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 69785
    },
    {
      "epoch": 61.815766164747565,
      "grad_norm": 0.2487981766462326,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 69790
    },
    {
      "epoch": 61.820194862710366,
      "grad_norm": 0.27073320746421814,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 69795
    },
    {
      "epoch": 61.82462356067316,
      "grad_norm": 0.22163337469100952,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 69800
    },
    {
      "epoch": 61.82905225863596,
      "grad_norm": 0.23547673225402832,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 69805
    },
    {
      "epoch": 61.83348095659876,
      "grad_norm": 0.24921423196792603,
      "learning_rate": 1e-05,
      "loss": 0.9126,
      "step": 69810
    },
    {
      "epoch": 61.83790965456156,
      "grad_norm": 0.24800798296928406,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 69815
    },
    {
      "epoch": 61.84233835252436,
      "grad_norm": 0.24171341955661774,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 69820
    },
    {
      "epoch": 61.84676705048716,
      "grad_norm": 0.20354153215885162,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 69825
    },
    {
      "epoch": 61.85119574844995,
      "grad_norm": 0.2333570122718811,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 69830
    },
    {
      "epoch": 61.855624446412754,
      "grad_norm": 0.2476746290922165,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 69835
    },
    {
      "epoch": 61.860053144375556,
      "grad_norm": 0.22960172593593597,
      "learning_rate": 1e-05,
      "loss": 1.0265,
      "step": 69840
    },
    {
      "epoch": 61.86448184233835,
      "grad_norm": 0.22044894099235535,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 69845
    },
    {
      "epoch": 61.86891054030115,
      "grad_norm": 0.20115643739700317,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 69850
    },
    {
      "epoch": 61.87333923826395,
      "grad_norm": 0.24040508270263672,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 69855
    },
    {
      "epoch": 61.877767936226746,
      "grad_norm": 0.255266934633255,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 69860
    },
    {
      "epoch": 61.88219663418955,
      "grad_norm": 0.20590266585350037,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 69865
    },
    {
      "epoch": 61.88662533215235,
      "grad_norm": 0.24051718413829803,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 69870
    },
    {
      "epoch": 61.89105403011514,
      "grad_norm": 0.20873750746250153,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 69875
    },
    {
      "epoch": 61.895482728077944,
      "grad_norm": 0.25857263803482056,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 69880
    },
    {
      "epoch": 61.899911426040745,
      "grad_norm": 0.2931264340877533,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 69885
    },
    {
      "epoch": 61.90434012400354,
      "grad_norm": 0.2375732809305191,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 69890
    },
    {
      "epoch": 61.90876882196634,
      "grad_norm": 0.24140547215938568,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 69895
    },
    {
      "epoch": 61.91319751992914,
      "grad_norm": 0.23719781637191772,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 69900
    },
    {
      "epoch": 61.91762621789194,
      "grad_norm": 0.22623473405838013,
      "learning_rate": 1e-05,
      "loss": 1.0349,
      "step": 69905
    },
    {
      "epoch": 61.92205491585474,
      "grad_norm": 0.2743055522441864,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 69910
    },
    {
      "epoch": 61.92648361381754,
      "grad_norm": 0.23880910873413086,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 69915
    },
    {
      "epoch": 61.93091231178034,
      "grad_norm": 0.2551307678222656,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 69920
    },
    {
      "epoch": 61.935341009743134,
      "grad_norm": 0.2756955325603485,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 69925
    },
    {
      "epoch": 61.939769707705935,
      "grad_norm": 0.24531473219394684,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 69930
    },
    {
      "epoch": 61.944198405668736,
      "grad_norm": 0.2602996230125427,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 69935
    },
    {
      "epoch": 61.94862710363153,
      "grad_norm": 0.2426404505968094,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 69940
    },
    {
      "epoch": 61.95305580159433,
      "grad_norm": 0.2642937898635864,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 69945
    },
    {
      "epoch": 61.95748449955713,
      "grad_norm": 0.2578623294830322,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 69950
    },
    {
      "epoch": 61.96191319751993,
      "grad_norm": 0.25468024611473083,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 69955
    },
    {
      "epoch": 61.96634189548273,
      "grad_norm": 0.23923556506633759,
      "learning_rate": 1e-05,
      "loss": 1.0009,
      "step": 69960
    },
    {
      "epoch": 61.97077059344553,
      "grad_norm": 0.250541090965271,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 69965
    },
    {
      "epoch": 61.97519929140832,
      "grad_norm": 0.24246199429035187,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 69970
    },
    {
      "epoch": 61.979627989371124,
      "grad_norm": 0.23483256995677948,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 69975
    },
    {
      "epoch": 61.984056687333926,
      "grad_norm": 0.2540358603000641,
      "learning_rate": 1e-05,
      "loss": 1.0195,
      "step": 69980
    },
    {
      "epoch": 61.98848538529672,
      "grad_norm": 0.22133643925189972,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 69985
    },
    {
      "epoch": 61.99291408325952,
      "grad_norm": 0.23597507178783417,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 69990
    },
    {
      "epoch": 61.99734278122232,
      "grad_norm": 0.2267870455980301,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 69995
    },
    {
      "epoch": 62.001771479185116,
      "grad_norm": 0.25215357542037964,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 70000
    },
    {
      "epoch": 62.00620017714792,
      "grad_norm": 0.21305692195892334,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 70005
    },
    {
      "epoch": 62.01062887511072,
      "grad_norm": 0.2484603077173233,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 70010
    },
    {
      "epoch": 62.01505757307352,
      "grad_norm": 0.2464698851108551,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 70015
    },
    {
      "epoch": 62.019486271036314,
      "grad_norm": 0.21229656040668488,
      "learning_rate": 1e-05,
      "loss": 0.8915,
      "step": 70020
    },
    {
      "epoch": 62.023914968999115,
      "grad_norm": 0.250098317861557,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 70025
    },
    {
      "epoch": 62.028343666961916,
      "grad_norm": 0.23205868899822235,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 70030
    },
    {
      "epoch": 62.03277236492471,
      "grad_norm": 0.21758829057216644,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 70035
    },
    {
      "epoch": 62.03720106288751,
      "grad_norm": 0.2773418426513672,
      "learning_rate": 1e-05,
      "loss": 0.8937,
      "step": 70040
    },
    {
      "epoch": 62.04162976085031,
      "grad_norm": 0.2416914701461792,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 70045
    },
    {
      "epoch": 62.04605845881311,
      "grad_norm": 0.23452645540237427,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 70050
    },
    {
      "epoch": 62.05048715677591,
      "grad_norm": 0.25741443037986755,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 70055
    },
    {
      "epoch": 62.05491585473871,
      "grad_norm": 0.23296645283699036,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 70060
    },
    {
      "epoch": 62.0593445527015,
      "grad_norm": 0.2542116940021515,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 70065
    },
    {
      "epoch": 62.063773250664305,
      "grad_norm": 0.23072609305381775,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 70070
    },
    {
      "epoch": 62.068201948627106,
      "grad_norm": 0.2435091882944107,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 70075
    },
    {
      "epoch": 62.0726306465899,
      "grad_norm": 0.25461915135383606,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 70080
    },
    {
      "epoch": 62.0770593445527,
      "grad_norm": 0.24691841006278992,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 70085
    },
    {
      "epoch": 62.0814880425155,
      "grad_norm": 0.23524101078510284,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 70090
    },
    {
      "epoch": 62.0859167404783,
      "grad_norm": 0.21469426155090332,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 70095
    },
    {
      "epoch": 62.0903454384411,
      "grad_norm": 0.2569428086280823,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 70100
    },
    {
      "epoch": 62.0947741364039,
      "grad_norm": 0.25578826665878296,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 70105
    },
    {
      "epoch": 62.09920283436669,
      "grad_norm": 0.2667850852012634,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 70110
    },
    {
      "epoch": 62.103631532329494,
      "grad_norm": 0.2515510320663452,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 70115
    },
    {
      "epoch": 62.108060230292296,
      "grad_norm": 0.21684180200099945,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 70120
    },
    {
      "epoch": 62.11248892825509,
      "grad_norm": 0.25035208463668823,
      "learning_rate": 1e-05,
      "loss": 1.0175,
      "step": 70125
    },
    {
      "epoch": 62.11691762621789,
      "grad_norm": 0.23104338347911835,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 70130
    },
    {
      "epoch": 62.12134632418069,
      "grad_norm": 0.22898367047309875,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 70135
    },
    {
      "epoch": 62.12577502214349,
      "grad_norm": 0.2066776305437088,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 70140
    },
    {
      "epoch": 62.13020372010629,
      "grad_norm": 0.2225179374217987,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 70145
    },
    {
      "epoch": 62.13463241806909,
      "grad_norm": 0.24942712485790253,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 70150
    },
    {
      "epoch": 62.13906111603189,
      "grad_norm": 0.19825562834739685,
      "learning_rate": 1e-05,
      "loss": 0.9187,
      "step": 70155
    },
    {
      "epoch": 62.143489813994684,
      "grad_norm": 0.2645453214645386,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 70160
    },
    {
      "epoch": 62.147918511957485,
      "grad_norm": 0.24226155877113342,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 70165
    },
    {
      "epoch": 62.152347209920286,
      "grad_norm": 0.2127036154270172,
      "learning_rate": 1e-05,
      "loss": 0.9177,
      "step": 70170
    },
    {
      "epoch": 62.15677590788308,
      "grad_norm": 0.22022129595279694,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 70175
    },
    {
      "epoch": 62.16120460584588,
      "grad_norm": 0.2422579973936081,
      "learning_rate": 1e-05,
      "loss": 0.9427,
      "step": 70180
    },
    {
      "epoch": 62.16563330380868,
      "grad_norm": 0.19733184576034546,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 70185
    },
    {
      "epoch": 62.17006200177148,
      "grad_norm": 0.2555609941482544,
      "learning_rate": 1e-05,
      "loss": 1.0455,
      "step": 70190
    },
    {
      "epoch": 62.17449069973428,
      "grad_norm": 0.2240137755870819,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 70195
    },
    {
      "epoch": 62.17891939769708,
      "grad_norm": 0.26627662777900696,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 70200
    },
    {
      "epoch": 62.18334809565987,
      "grad_norm": 0.31554529070854187,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 70205
    },
    {
      "epoch": 62.187776793622675,
      "grad_norm": 0.2500566244125366,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 70210
    },
    {
      "epoch": 62.192205491585476,
      "grad_norm": 0.24197325110435486,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 70215
    },
    {
      "epoch": 62.19663418954827,
      "grad_norm": 0.26426810026168823,
      "learning_rate": 1e-05,
      "loss": 1.027,
      "step": 70220
    },
    {
      "epoch": 62.20106288751107,
      "grad_norm": 0.2679857611656189,
      "learning_rate": 1e-05,
      "loss": 0.9058,
      "step": 70225
    },
    {
      "epoch": 62.20549158547387,
      "grad_norm": 0.20104993879795074,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 70230
    },
    {
      "epoch": 62.20992028343667,
      "grad_norm": 0.24380265176296234,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 70235
    },
    {
      "epoch": 62.21434898139947,
      "grad_norm": 0.22323939204216003,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 70240
    },
    {
      "epoch": 62.21877767936227,
      "grad_norm": 0.25179633498191833,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 70245
    },
    {
      "epoch": 62.22320637732506,
      "grad_norm": 0.23339669406414032,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 70250
    },
    {
      "epoch": 62.227635075287864,
      "grad_norm": 0.26794078946113586,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 70255
    },
    {
      "epoch": 62.232063773250665,
      "grad_norm": 0.24298353493213654,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 70260
    },
    {
      "epoch": 62.23649247121347,
      "grad_norm": 0.22416622936725616,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 70265
    },
    {
      "epoch": 62.24092116917626,
      "grad_norm": 0.24579066038131714,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 70270
    },
    {
      "epoch": 62.24534986713906,
      "grad_norm": 0.2506297826766968,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 70275
    },
    {
      "epoch": 62.24977856510186,
      "grad_norm": 0.24719461798667908,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 70280
    },
    {
      "epoch": 62.25420726306466,
      "grad_norm": 0.19893841445446014,
      "learning_rate": 1e-05,
      "loss": 0.9147,
      "step": 70285
    },
    {
      "epoch": 62.25863596102746,
      "grad_norm": 0.2632090449333191,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 70290
    },
    {
      "epoch": 62.26306465899026,
      "grad_norm": 0.27455762028694153,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 70295
    },
    {
      "epoch": 62.267493356953054,
      "grad_norm": 0.3039446175098419,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 70300
    },
    {
      "epoch": 62.271922054915855,
      "grad_norm": 0.2396402359008789,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 70305
    },
    {
      "epoch": 62.276350752878656,
      "grad_norm": 0.2342238873243332,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 70310
    },
    {
      "epoch": 62.28077945084145,
      "grad_norm": 0.20139279961585999,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 70315
    },
    {
      "epoch": 62.28520814880425,
      "grad_norm": 0.2541227638721466,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 70320
    },
    {
      "epoch": 62.28963684676705,
      "grad_norm": 0.26092657446861267,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 70325
    },
    {
      "epoch": 62.29406554472985,
      "grad_norm": 0.2117588073015213,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 70330
    },
    {
      "epoch": 62.29849424269265,
      "grad_norm": 0.22186078131198883,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 70335
    },
    {
      "epoch": 62.30292294065545,
      "grad_norm": 0.2326059639453888,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 70340
    },
    {
      "epoch": 62.30735163861824,
      "grad_norm": 0.251787006855011,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 70345
    },
    {
      "epoch": 62.311780336581045,
      "grad_norm": 0.28133392333984375,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 70350
    },
    {
      "epoch": 62.316209034543846,
      "grad_norm": 0.2618042528629303,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 70355
    },
    {
      "epoch": 62.32063773250664,
      "grad_norm": 0.21033257246017456,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 70360
    },
    {
      "epoch": 62.32506643046944,
      "grad_norm": 0.2005743533372879,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 70365
    },
    {
      "epoch": 62.32949512843224,
      "grad_norm": 0.21412776410579681,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 70370
    },
    {
      "epoch": 62.333923826395036,
      "grad_norm": 0.24447116255760193,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 70375
    },
    {
      "epoch": 62.33835252435784,
      "grad_norm": 0.20145437121391296,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 70380
    },
    {
      "epoch": 62.34278122232064,
      "grad_norm": 0.2918694317340851,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 70385
    },
    {
      "epoch": 62.34720992028344,
      "grad_norm": 0.25959402322769165,
      "learning_rate": 1e-05,
      "loss": 1.0275,
      "step": 70390
    },
    {
      "epoch": 62.351638618246234,
      "grad_norm": 0.29668527841567993,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 70395
    },
    {
      "epoch": 62.356067316209035,
      "grad_norm": 0.2272024303674698,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 70400
    },
    {
      "epoch": 62.36049601417184,
      "grad_norm": 0.23765182495117188,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 70405
    },
    {
      "epoch": 62.36492471213463,
      "grad_norm": 0.2663049101829529,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 70410
    },
    {
      "epoch": 62.36935341009743,
      "grad_norm": 0.23255185782909393,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 70415
    },
    {
      "epoch": 62.37378210806023,
      "grad_norm": 0.27086231112480164,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 70420
    },
    {
      "epoch": 62.37821080602303,
      "grad_norm": 0.21752439439296722,
      "learning_rate": 1e-05,
      "loss": 1.046,
      "step": 70425
    },
    {
      "epoch": 62.38263950398583,
      "grad_norm": 0.2265576422214508,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 70430
    },
    {
      "epoch": 62.38706820194863,
      "grad_norm": 0.2743570804595947,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 70435
    },
    {
      "epoch": 62.391496899911424,
      "grad_norm": 0.26250413060188293,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 70440
    },
    {
      "epoch": 62.395925597874225,
      "grad_norm": 0.24101440608501434,
      "learning_rate": 1e-05,
      "loss": 0.9165,
      "step": 70445
    },
    {
      "epoch": 62.400354295837026,
      "grad_norm": 0.22953088581562042,
      "learning_rate": 1e-05,
      "loss": 0.9021,
      "step": 70450
    },
    {
      "epoch": 62.40478299379982,
      "grad_norm": 0.32851457595825195,
      "learning_rate": 1e-05,
      "loss": 0.9319,
      "step": 70455
    },
    {
      "epoch": 62.40921169176262,
      "grad_norm": 0.2697694003582001,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 70460
    },
    {
      "epoch": 62.41364038972542,
      "grad_norm": 0.21063600480556488,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 70465
    },
    {
      "epoch": 62.41806908768822,
      "grad_norm": 0.21948233246803284,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 70470
    },
    {
      "epoch": 62.42249778565102,
      "grad_norm": 0.23780766129493713,
      "learning_rate": 1e-05,
      "loss": 0.9095,
      "step": 70475
    },
    {
      "epoch": 62.42692648361382,
      "grad_norm": 0.22337086498737335,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 70480
    },
    {
      "epoch": 62.43135518157661,
      "grad_norm": 0.24864935874938965,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 70485
    },
    {
      "epoch": 62.435783879539414,
      "grad_norm": 0.25221261382102966,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 70490
    },
    {
      "epoch": 62.440212577502216,
      "grad_norm": 0.22788521647453308,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 70495
    },
    {
      "epoch": 62.44464127546502,
      "grad_norm": 0.23247942328453064,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 70500
    },
    {
      "epoch": 62.44906997342781,
      "grad_norm": 0.24569058418273926,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 70505
    },
    {
      "epoch": 62.45349867139061,
      "grad_norm": 0.3016890585422516,
      "learning_rate": 1e-05,
      "loss": 1.0195,
      "step": 70510
    },
    {
      "epoch": 62.45792736935341,
      "grad_norm": 0.24460209906101227,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 70515
    },
    {
      "epoch": 62.46235606731621,
      "grad_norm": 0.22484734654426575,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 70520
    },
    {
      "epoch": 62.46678476527901,
      "grad_norm": 0.22628504037857056,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 70525
    },
    {
      "epoch": 62.47121346324181,
      "grad_norm": 0.2309778332710266,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 70530
    },
    {
      "epoch": 62.475642161204604,
      "grad_norm": 0.23494674265384674,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 70535
    },
    {
      "epoch": 62.480070859167405,
      "grad_norm": 0.22516237199306488,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 70540
    },
    {
      "epoch": 62.484499557130206,
      "grad_norm": 0.22986817359924316,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 70545
    },
    {
      "epoch": 62.488928255093,
      "grad_norm": 0.22091831266880035,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 70550
    },
    {
      "epoch": 62.4933569530558,
      "grad_norm": 0.21616840362548828,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 70555
    },
    {
      "epoch": 62.4977856510186,
      "grad_norm": 0.25040966272354126,
      "learning_rate": 1e-05,
      "loss": 1.0382,
      "step": 70560
    },
    {
      "epoch": 62.5022143489814,
      "grad_norm": 0.2095143347978592,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 70565
    },
    {
      "epoch": 62.5066430469442,
      "grad_norm": 0.21836015582084656,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 70570
    },
    {
      "epoch": 62.511071744907,
      "grad_norm": 0.22668179869651794,
      "learning_rate": 1e-05,
      "loss": 0.9029,
      "step": 70575
    },
    {
      "epoch": 62.515500442869794,
      "grad_norm": 0.23770293593406677,
      "learning_rate": 1e-05,
      "loss": 0.9286,
      "step": 70580
    },
    {
      "epoch": 62.519929140832595,
      "grad_norm": 0.26848098635673523,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 70585
    },
    {
      "epoch": 62.524357838795396,
      "grad_norm": 0.2318633645772934,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 70590
    },
    {
      "epoch": 62.52878653675819,
      "grad_norm": 0.2602775990962982,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 70595
    },
    {
      "epoch": 62.53321523472099,
      "grad_norm": 0.2255481779575348,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 70600
    },
    {
      "epoch": 62.53764393268379,
      "grad_norm": 0.23722165822982788,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 70605
    },
    {
      "epoch": 62.54207263064659,
      "grad_norm": 0.23507237434387207,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 70610
    },
    {
      "epoch": 62.54650132860939,
      "grad_norm": 0.24398557841777802,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 70615
    },
    {
      "epoch": 62.55093002657219,
      "grad_norm": 0.22806020081043243,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 70620
    },
    {
      "epoch": 62.55535872453498,
      "grad_norm": 0.24399439990520477,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 70625
    },
    {
      "epoch": 62.559787422497784,
      "grad_norm": 0.21834306418895721,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 70630
    },
    {
      "epoch": 62.564216120460586,
      "grad_norm": 0.2814664840698242,
      "learning_rate": 1e-05,
      "loss": 0.9165,
      "step": 70635
    },
    {
      "epoch": 62.56864481842339,
      "grad_norm": 0.22061903774738312,
      "learning_rate": 1e-05,
      "loss": 0.9158,
      "step": 70640
    },
    {
      "epoch": 62.57307351638618,
      "grad_norm": 0.22100220620632172,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 70645
    },
    {
      "epoch": 62.57750221434898,
      "grad_norm": 0.27740535140037537,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 70650
    },
    {
      "epoch": 62.58193091231178,
      "grad_norm": 0.23855076730251312,
      "learning_rate": 1e-05,
      "loss": 0.9287,
      "step": 70655
    },
    {
      "epoch": 62.58635961027458,
      "grad_norm": 0.2657449245452881,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 70660
    },
    {
      "epoch": 62.59078830823738,
      "grad_norm": 0.28157928586006165,
      "learning_rate": 1e-05,
      "loss": 1.0015,
      "step": 70665
    },
    {
      "epoch": 62.59521700620018,
      "grad_norm": 0.24486863613128662,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 70670
    },
    {
      "epoch": 62.599645704162974,
      "grad_norm": 0.2499570995569229,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 70675
    },
    {
      "epoch": 62.604074402125775,
      "grad_norm": 0.25258275866508484,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 70680
    },
    {
      "epoch": 62.608503100088576,
      "grad_norm": 0.24013161659240723,
      "learning_rate": 1e-05,
      "loss": 0.9239,
      "step": 70685
    },
    {
      "epoch": 62.61293179805137,
      "grad_norm": 0.2438352257013321,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 70690
    },
    {
      "epoch": 62.61736049601417,
      "grad_norm": 0.25572359561920166,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 70695
    },
    {
      "epoch": 62.62178919397697,
      "grad_norm": 0.22293159365653992,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 70700
    },
    {
      "epoch": 62.62621789193977,
      "grad_norm": 0.2221287488937378,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 70705
    },
    {
      "epoch": 62.63064658990257,
      "grad_norm": 0.2349444031715393,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 70710
    },
    {
      "epoch": 62.63507528786537,
      "grad_norm": 0.21984359622001648,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 70715
    },
    {
      "epoch": 62.63950398582816,
      "grad_norm": 0.22304648160934448,
      "learning_rate": 1e-05,
      "loss": 1.0195,
      "step": 70720
    },
    {
      "epoch": 62.643932683790965,
      "grad_norm": 0.2281741350889206,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 70725
    },
    {
      "epoch": 62.648361381753766,
      "grad_norm": 0.22168071568012238,
      "learning_rate": 1e-05,
      "loss": 1.0511,
      "step": 70730
    },
    {
      "epoch": 62.65279007971656,
      "grad_norm": 0.20521099865436554,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 70735
    },
    {
      "epoch": 62.65721877767936,
      "grad_norm": 0.2321094423532486,
      "learning_rate": 1e-05,
      "loss": 1.0273,
      "step": 70740
    },
    {
      "epoch": 62.66164747564216,
      "grad_norm": 0.20615778863430023,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 70745
    },
    {
      "epoch": 62.666076173604964,
      "grad_norm": 0.23900240659713745,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 70750
    },
    {
      "epoch": 62.67050487156776,
      "grad_norm": 0.23599305748939514,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 70755
    },
    {
      "epoch": 62.67493356953056,
      "grad_norm": 0.26711955666542053,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 70760
    },
    {
      "epoch": 62.67936226749336,
      "grad_norm": 0.22688394784927368,
      "learning_rate": 1e-05,
      "loss": 1.0397,
      "step": 70765
    },
    {
      "epoch": 62.683790965456154,
      "grad_norm": 0.2360047549009323,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 70770
    },
    {
      "epoch": 62.688219663418955,
      "grad_norm": 0.286018043756485,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 70775
    },
    {
      "epoch": 62.69264836138176,
      "grad_norm": 0.21145670115947723,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 70780
    },
    {
      "epoch": 62.69707705934455,
      "grad_norm": 0.2701103389263153,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 70785
    },
    {
      "epoch": 62.70150575730735,
      "grad_norm": 0.2744518518447876,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 70790
    },
    {
      "epoch": 62.70593445527015,
      "grad_norm": 0.25096407532691956,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 70795
    },
    {
      "epoch": 62.71036315323295,
      "grad_norm": 0.24646686017513275,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 70800
    },
    {
      "epoch": 62.71479185119575,
      "grad_norm": 0.2550307810306549,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 70805
    },
    {
      "epoch": 62.71922054915855,
      "grad_norm": 0.240295872092247,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 70810
    },
    {
      "epoch": 62.723649247121344,
      "grad_norm": 0.2812483608722687,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 70815
    },
    {
      "epoch": 62.728077945084145,
      "grad_norm": 0.231979638338089,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 70820
    },
    {
      "epoch": 62.732506643046946,
      "grad_norm": 0.24539978802204132,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 70825
    },
    {
      "epoch": 62.73693534100974,
      "grad_norm": 0.27017080783843994,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 70830
    },
    {
      "epoch": 62.74136403897254,
      "grad_norm": 0.2220301777124405,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 70835
    },
    {
      "epoch": 62.74579273693534,
      "grad_norm": 0.2858640253543854,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 70840
    },
    {
      "epoch": 62.75022143489814,
      "grad_norm": 0.25268128514289856,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 70845
    },
    {
      "epoch": 62.75465013286094,
      "grad_norm": 0.21712400019168854,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 70850
    },
    {
      "epoch": 62.75907883082374,
      "grad_norm": 0.22115670144557953,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 70855
    },
    {
      "epoch": 62.76350752878653,
      "grad_norm": 0.20956386625766754,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 70860
    },
    {
      "epoch": 62.767936226749335,
      "grad_norm": 0.22501781582832336,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 70865
    },
    {
      "epoch": 62.772364924712136,
      "grad_norm": 0.24337220191955566,
      "learning_rate": 1e-05,
      "loss": 0.9043,
      "step": 70870
    },
    {
      "epoch": 62.77679362267494,
      "grad_norm": 0.22969840466976166,
      "learning_rate": 1e-05,
      "loss": 0.9206,
      "step": 70875
    },
    {
      "epoch": 62.78122232063773,
      "grad_norm": 0.22875837981700897,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 70880
    },
    {
      "epoch": 62.78565101860053,
      "grad_norm": 0.19865119457244873,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 70885
    },
    {
      "epoch": 62.79007971656333,
      "grad_norm": 0.23736608028411865,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 70890
    },
    {
      "epoch": 62.79450841452613,
      "grad_norm": 0.2708814740180969,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 70895
    },
    {
      "epoch": 62.79893711248893,
      "grad_norm": 0.22967125475406647,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 70900
    },
    {
      "epoch": 62.80336581045173,
      "grad_norm": 0.22558431327342987,
      "learning_rate": 1e-05,
      "loss": 0.9026,
      "step": 70905
    },
    {
      "epoch": 62.807794508414524,
      "grad_norm": 0.22295479476451874,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 70910
    },
    {
      "epoch": 62.812223206377325,
      "grad_norm": 0.2839282751083374,
      "learning_rate": 1e-05,
      "loss": 0.9266,
      "step": 70915
    },
    {
      "epoch": 62.81665190434013,
      "grad_norm": 0.23600752651691437,
      "learning_rate": 1e-05,
      "loss": 0.9052,
      "step": 70920
    },
    {
      "epoch": 62.82108060230292,
      "grad_norm": 0.30957773327827454,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 70925
    },
    {
      "epoch": 62.82550930026572,
      "grad_norm": 0.21585112810134888,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 70930
    },
    {
      "epoch": 62.82993799822852,
      "grad_norm": 0.2283477783203125,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 70935
    },
    {
      "epoch": 62.83436669619132,
      "grad_norm": 0.21538548171520233,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 70940
    },
    {
      "epoch": 62.83879539415412,
      "grad_norm": 0.2529657483100891,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 70945
    },
    {
      "epoch": 62.84322409211692,
      "grad_norm": 0.22416697442531586,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 70950
    },
    {
      "epoch": 62.847652790079714,
      "grad_norm": 0.2417793720960617,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 70955
    },
    {
      "epoch": 62.852081488042515,
      "grad_norm": 0.2744355797767639,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 70960
    },
    {
      "epoch": 62.856510186005316,
      "grad_norm": 0.2282475084066391,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 70965
    },
    {
      "epoch": 62.86093888396811,
      "grad_norm": 0.23291133344173431,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 70970
    },
    {
      "epoch": 62.86536758193091,
      "grad_norm": 0.2341146320104599,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 70975
    },
    {
      "epoch": 62.86979627989371,
      "grad_norm": 0.2175760120153427,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 70980
    },
    {
      "epoch": 62.87422497785651,
      "grad_norm": 0.21026121079921722,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 70985
    },
    {
      "epoch": 62.87865367581931,
      "grad_norm": 0.22480472922325134,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 70990
    },
    {
      "epoch": 62.88308237378211,
      "grad_norm": 0.22405695915222168,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 70995
    },
    {
      "epoch": 62.88751107174491,
      "grad_norm": 0.27734410762786865,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 71000
    },
    {
      "epoch": 62.891939769707704,
      "grad_norm": 0.25987330079078674,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 71005
    },
    {
      "epoch": 62.896368467670506,
      "grad_norm": 0.2726013660430908,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 71010
    },
    {
      "epoch": 62.90079716563331,
      "grad_norm": 0.2747352719306946,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 71015
    },
    {
      "epoch": 62.9052258635961,
      "grad_norm": 0.2267664074897766,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 71020
    },
    {
      "epoch": 62.9096545615589,
      "grad_norm": 0.24789299070835114,
      "learning_rate": 1e-05,
      "loss": 0.9231,
      "step": 71025
    },
    {
      "epoch": 62.9140832595217,
      "grad_norm": 0.25955846905708313,
      "learning_rate": 1e-05,
      "loss": 0.9246,
      "step": 71030
    },
    {
      "epoch": 62.9185119574845,
      "grad_norm": 0.24115349352359772,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 71035
    },
    {
      "epoch": 62.9229406554473,
      "grad_norm": 0.2540762424468994,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 71040
    },
    {
      "epoch": 62.9273693534101,
      "grad_norm": 0.30991625785827637,
      "learning_rate": 1e-05,
      "loss": 0.9059,
      "step": 71045
    },
    {
      "epoch": 62.931798051372894,
      "grad_norm": 0.30098727345466614,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 71050
    },
    {
      "epoch": 62.936226749335695,
      "grad_norm": 0.2261722981929779,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 71055
    },
    {
      "epoch": 62.9406554472985,
      "grad_norm": 0.2749830484390259,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 71060
    },
    {
      "epoch": 62.94508414526129,
      "grad_norm": 0.23942245543003082,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 71065
    },
    {
      "epoch": 62.94951284322409,
      "grad_norm": 0.2497217208147049,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 71070
    },
    {
      "epoch": 62.95394154118689,
      "grad_norm": 0.2423870861530304,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 71075
    },
    {
      "epoch": 62.95837023914969,
      "grad_norm": 0.23776225745677948,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 71080
    },
    {
      "epoch": 62.96279893711249,
      "grad_norm": 0.22410358488559723,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 71085
    },
    {
      "epoch": 62.96722763507529,
      "grad_norm": 0.2592860758304596,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 71090
    },
    {
      "epoch": 62.971656333038084,
      "grad_norm": 0.2347031831741333,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 71095
    },
    {
      "epoch": 62.976085031000885,
      "grad_norm": 0.256392240524292,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 71100
    },
    {
      "epoch": 62.980513728963686,
      "grad_norm": 0.31371161341667175,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 71105
    },
    {
      "epoch": 62.98494242692648,
      "grad_norm": 0.25879257917404175,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 71110
    },
    {
      "epoch": 62.98937112488928,
      "grad_norm": 0.26602286100387573,
      "learning_rate": 1e-05,
      "loss": 0.9154,
      "step": 71115
    },
    {
      "epoch": 62.99379982285208,
      "grad_norm": 0.28012028336524963,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 71120
    },
    {
      "epoch": 62.998228520814884,
      "grad_norm": 0.2472427934408188,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 71125
    },
    {
      "epoch": 63.00265721877768,
      "grad_norm": 0.21624986827373505,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 71130
    },
    {
      "epoch": 63.00708591674048,
      "grad_norm": 0.2152177393436432,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 71135
    },
    {
      "epoch": 63.01151461470328,
      "grad_norm": 0.2457621693611145,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 71140
    },
    {
      "epoch": 63.015943312666074,
      "grad_norm": 0.21310003101825714,
      "learning_rate": 1e-05,
      "loss": 0.923,
      "step": 71145
    },
    {
      "epoch": 63.020372010628876,
      "grad_norm": 0.2093529999256134,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 71150
    },
    {
      "epoch": 63.02480070859168,
      "grad_norm": 0.2449687123298645,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 71155
    },
    {
      "epoch": 63.02922940655447,
      "grad_norm": 0.2772822976112366,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 71160
    },
    {
      "epoch": 63.03365810451727,
      "grad_norm": 0.23291809856891632,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 71165
    },
    {
      "epoch": 63.03808680248007,
      "grad_norm": 0.23847796022891998,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 71170
    },
    {
      "epoch": 63.04251550044287,
      "grad_norm": 0.2273930460214615,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 71175
    },
    {
      "epoch": 63.04694419840567,
      "grad_norm": 0.22644150257110596,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 71180
    },
    {
      "epoch": 63.05137289636847,
      "grad_norm": 0.25376737117767334,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 71185
    },
    {
      "epoch": 63.055801594331264,
      "grad_norm": 0.2361001819372177,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 71190
    },
    {
      "epoch": 63.060230292294065,
      "grad_norm": 0.2948321998119354,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 71195
    },
    {
      "epoch": 63.064658990256866,
      "grad_norm": 0.23853449523448944,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 71200
    },
    {
      "epoch": 63.06908768821966,
      "grad_norm": 0.23073089122772217,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 71205
    },
    {
      "epoch": 63.07351638618246,
      "grad_norm": 0.26085132360458374,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 71210
    },
    {
      "epoch": 63.07794508414526,
      "grad_norm": 0.24262838065624237,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 71215
    },
    {
      "epoch": 63.08237378210806,
      "grad_norm": 0.22725196182727814,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 71220
    },
    {
      "epoch": 63.08680248007086,
      "grad_norm": 0.25621891021728516,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 71225
    },
    {
      "epoch": 63.09123117803366,
      "grad_norm": 0.20483575761318207,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 71230
    },
    {
      "epoch": 63.09565987599645,
      "grad_norm": 0.23137125372886658,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 71235
    },
    {
      "epoch": 63.100088573959255,
      "grad_norm": 0.22602321207523346,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 71240
    },
    {
      "epoch": 63.104517271922056,
      "grad_norm": 0.2657042443752289,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 71245
    },
    {
      "epoch": 63.10894596988486,
      "grad_norm": 0.23578548431396484,
      "learning_rate": 1e-05,
      "loss": 0.9019,
      "step": 71250
    },
    {
      "epoch": 63.11337466784765,
      "grad_norm": 0.22353316843509674,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 71255
    },
    {
      "epoch": 63.11780336581045,
      "grad_norm": 0.2512418329715729,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 71260
    },
    {
      "epoch": 63.122232063773254,
      "grad_norm": 0.22224655747413635,
      "learning_rate": 1e-05,
      "loss": 1.0213,
      "step": 71265
    },
    {
      "epoch": 63.12666076173605,
      "grad_norm": 0.2689298093318939,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 71270
    },
    {
      "epoch": 63.13108945969885,
      "grad_norm": 0.2617366313934326,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 71275
    },
    {
      "epoch": 63.13551815766165,
      "grad_norm": 0.3360458016395569,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 71280
    },
    {
      "epoch": 63.139946855624444,
      "grad_norm": 0.2393290251493454,
      "learning_rate": 1e-05,
      "loss": 0.8593,
      "step": 71285
    },
    {
      "epoch": 63.144375553587246,
      "grad_norm": 0.24505038559436798,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 71290
    },
    {
      "epoch": 63.14880425155005,
      "grad_norm": 0.27710598707199097,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 71295
    },
    {
      "epoch": 63.15323294951284,
      "grad_norm": 0.2641274929046631,
      "learning_rate": 1e-05,
      "loss": 0.8996,
      "step": 71300
    },
    {
      "epoch": 63.15766164747564,
      "grad_norm": 0.2361448109149933,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 71305
    },
    {
      "epoch": 63.16209034543844,
      "grad_norm": 0.2303447425365448,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 71310
    },
    {
      "epoch": 63.16651904340124,
      "grad_norm": 0.238552525639534,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 71315
    },
    {
      "epoch": 63.17094774136404,
      "grad_norm": 0.23807229101657867,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 71320
    },
    {
      "epoch": 63.17537643932684,
      "grad_norm": 0.2130032479763031,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 71325
    },
    {
      "epoch": 63.179805137289634,
      "grad_norm": 0.19494573771953583,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 71330
    },
    {
      "epoch": 63.184233835252435,
      "grad_norm": 0.24512140452861786,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 71335
    },
    {
      "epoch": 63.188662533215236,
      "grad_norm": 0.26029345393180847,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 71340
    },
    {
      "epoch": 63.19309123117803,
      "grad_norm": 0.2286328673362732,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 71345
    },
    {
      "epoch": 63.19751992914083,
      "grad_norm": 0.23320408165454865,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 71350
    },
    {
      "epoch": 63.20194862710363,
      "grad_norm": 0.2648443579673767,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 71355
    },
    {
      "epoch": 63.20637732506643,
      "grad_norm": 0.25996845960617065,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 71360
    },
    {
      "epoch": 63.21080602302923,
      "grad_norm": 0.27111363410949707,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 71365
    },
    {
      "epoch": 63.21523472099203,
      "grad_norm": 0.2943268418312073,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 71370
    },
    {
      "epoch": 63.21966341895483,
      "grad_norm": 0.25065112113952637,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 71375
    },
    {
      "epoch": 63.224092116917625,
      "grad_norm": 0.26021096110343933,
      "learning_rate": 1e-05,
      "loss": 0.9223,
      "step": 71380
    },
    {
      "epoch": 63.228520814880426,
      "grad_norm": 0.2268586903810501,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 71385
    },
    {
      "epoch": 63.23294951284323,
      "grad_norm": 0.23720504343509674,
      "learning_rate": 1e-05,
      "loss": 0.9289,
      "step": 71390
    },
    {
      "epoch": 63.23737821080602,
      "grad_norm": 0.24163025617599487,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 71395
    },
    {
      "epoch": 63.24180690876882,
      "grad_norm": 0.2378177046775818,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 71400
    },
    {
      "epoch": 63.246235606731624,
      "grad_norm": 0.24011115729808807,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 71405
    },
    {
      "epoch": 63.25066430469442,
      "grad_norm": 0.2730252146720886,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 71410
    },
    {
      "epoch": 63.25509300265722,
      "grad_norm": 0.23889029026031494,
      "learning_rate": 1e-05,
      "loss": 0.9217,
      "step": 71415
    },
    {
      "epoch": 63.25952170062002,
      "grad_norm": 0.7274526953697205,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 71420
    },
    {
      "epoch": 63.263950398582814,
      "grad_norm": 0.27280551195144653,
      "learning_rate": 1e-05,
      "loss": 0.9186,
      "step": 71425
    },
    {
      "epoch": 63.268379096545615,
      "grad_norm": 0.27427947521209717,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 71430
    },
    {
      "epoch": 63.27280779450842,
      "grad_norm": 0.26814350485801697,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 71435
    },
    {
      "epoch": 63.27723649247121,
      "grad_norm": 0.2442561835050583,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 71440
    },
    {
      "epoch": 63.28166519043401,
      "grad_norm": 0.22382545471191406,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 71445
    },
    {
      "epoch": 63.28609388839681,
      "grad_norm": 0.2901572287082672,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 71450
    },
    {
      "epoch": 63.29052258635961,
      "grad_norm": 0.23307915031909943,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 71455
    },
    {
      "epoch": 63.29495128432241,
      "grad_norm": 0.24821032583713531,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 71460
    },
    {
      "epoch": 63.29937998228521,
      "grad_norm": 0.25503966212272644,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 71465
    },
    {
      "epoch": 63.303808680248004,
      "grad_norm": 0.27275925874710083,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 71470
    },
    {
      "epoch": 63.308237378210805,
      "grad_norm": 0.2198983132839203,
      "learning_rate": 1e-05,
      "loss": 0.9047,
      "step": 71475
    },
    {
      "epoch": 63.312666076173606,
      "grad_norm": 0.22909466922283173,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 71480
    },
    {
      "epoch": 63.31709477413641,
      "grad_norm": 0.3188401460647583,
      "learning_rate": 1e-05,
      "loss": 0.8985,
      "step": 71485
    },
    {
      "epoch": 63.3215234720992,
      "grad_norm": 0.2388790398836136,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 71490
    },
    {
      "epoch": 63.325952170062,
      "grad_norm": 0.21546371281147003,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 71495
    },
    {
      "epoch": 63.330380868024804,
      "grad_norm": 0.2413410097360611,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 71500
    },
    {
      "epoch": 63.3348095659876,
      "grad_norm": 0.33520326018333435,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 71505
    },
    {
      "epoch": 63.3392382639504,
      "grad_norm": 0.24245035648345947,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 71510
    },
    {
      "epoch": 63.3436669619132,
      "grad_norm": 0.24469655752182007,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 71515
    },
    {
      "epoch": 63.348095659875995,
      "grad_norm": 0.2601113021373749,
      "learning_rate": 1e-05,
      "loss": 1.0263,
      "step": 71520
    },
    {
      "epoch": 63.352524357838796,
      "grad_norm": 0.2387658804655075,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 71525
    },
    {
      "epoch": 63.3569530558016,
      "grad_norm": 0.23423263430595398,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 71530
    },
    {
      "epoch": 63.36138175376439,
      "grad_norm": 0.24021795392036438,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 71535
    },
    {
      "epoch": 63.36581045172719,
      "grad_norm": 0.22292114794254303,
      "learning_rate": 1e-05,
      "loss": 0.9427,
      "step": 71540
    },
    {
      "epoch": 63.37023914968999,
      "grad_norm": 0.21965794265270233,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 71545
    },
    {
      "epoch": 63.37466784765279,
      "grad_norm": 0.2741168737411499,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 71550
    },
    {
      "epoch": 63.37909654561559,
      "grad_norm": 0.24786996841430664,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 71555
    },
    {
      "epoch": 63.38352524357839,
      "grad_norm": 0.26409363746643066,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 71560
    },
    {
      "epoch": 63.387953941541184,
      "grad_norm": 0.26504209637641907,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 71565
    },
    {
      "epoch": 63.392382639503985,
      "grad_norm": 0.2318962812423706,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 71570
    },
    {
      "epoch": 63.39681133746679,
      "grad_norm": 0.2678954601287842,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 71575
    },
    {
      "epoch": 63.40124003542958,
      "grad_norm": 0.23237016797065735,
      "learning_rate": 1e-05,
      "loss": 0.9244,
      "step": 71580
    },
    {
      "epoch": 63.40566873339238,
      "grad_norm": 0.2803463935852051,
      "learning_rate": 1e-05,
      "loss": 0.9263,
      "step": 71585
    },
    {
      "epoch": 63.41009743135518,
      "grad_norm": 0.22890187799930573,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 71590
    },
    {
      "epoch": 63.41452612931798,
      "grad_norm": 0.23410005867481232,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 71595
    },
    {
      "epoch": 63.41895482728078,
      "grad_norm": 0.20497246086597443,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 71600
    },
    {
      "epoch": 63.42338352524358,
      "grad_norm": 0.23170189559459686,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 71605
    },
    {
      "epoch": 63.42781222320638,
      "grad_norm": 0.2319885641336441,
      "learning_rate": 1e-05,
      "loss": 0.9031,
      "step": 71610
    },
    {
      "epoch": 63.432240921169175,
      "grad_norm": 0.24319502711296082,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 71615
    },
    {
      "epoch": 63.436669619131976,
      "grad_norm": 0.2243577241897583,
      "learning_rate": 1e-05,
      "loss": 0.9292,
      "step": 71620
    },
    {
      "epoch": 63.44109831709478,
      "grad_norm": 0.22944214940071106,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 71625
    },
    {
      "epoch": 63.44552701505757,
      "grad_norm": 0.26419004797935486,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 71630
    },
    {
      "epoch": 63.44995571302037,
      "grad_norm": 0.2424696683883667,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 71635
    },
    {
      "epoch": 63.454384410983174,
      "grad_norm": 0.29082199931144714,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 71640
    },
    {
      "epoch": 63.45881310894597,
      "grad_norm": 0.22004655003547668,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 71645
    },
    {
      "epoch": 63.46324180690877,
      "grad_norm": 0.24860699474811554,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 71650
    },
    {
      "epoch": 63.46767050487157,
      "grad_norm": 0.24170424044132233,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 71655
    },
    {
      "epoch": 63.472099202834364,
      "grad_norm": 0.2371632158756256,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 71660
    },
    {
      "epoch": 63.476527900797166,
      "grad_norm": 0.22572371363639832,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 71665
    },
    {
      "epoch": 63.48095659875997,
      "grad_norm": 0.23342150449752808,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 71670
    },
    {
      "epoch": 63.48538529672276,
      "grad_norm": 0.21047848463058472,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 71675
    },
    {
      "epoch": 63.48981399468556,
      "grad_norm": 0.37906166911125183,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 71680
    },
    {
      "epoch": 63.49424269264836,
      "grad_norm": 0.20648927986621857,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 71685
    },
    {
      "epoch": 63.49867139061116,
      "grad_norm": 0.2352408915758133,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 71690
    },
    {
      "epoch": 63.50310008857396,
      "grad_norm": 0.2716488540172577,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 71695
    },
    {
      "epoch": 63.50752878653676,
      "grad_norm": 0.2358202040195465,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 71700
    },
    {
      "epoch": 63.511957484499554,
      "grad_norm": 0.24556708335876465,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 71705
    },
    {
      "epoch": 63.516386182462355,
      "grad_norm": 0.23424102365970612,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 71710
    },
    {
      "epoch": 63.520814880425156,
      "grad_norm": 0.2296905219554901,
      "learning_rate": 1e-05,
      "loss": 0.9081,
      "step": 71715
    },
    {
      "epoch": 63.52524357838795,
      "grad_norm": 0.2563684284687042,
      "learning_rate": 1e-05,
      "loss": 0.9178,
      "step": 71720
    },
    {
      "epoch": 63.52967227635075,
      "grad_norm": 0.26498961448669434,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 71725
    },
    {
      "epoch": 63.53410097431355,
      "grad_norm": 0.2527771592140198,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 71730
    },
    {
      "epoch": 63.538529672276354,
      "grad_norm": 0.30983543395996094,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 71735
    },
    {
      "epoch": 63.54295837023915,
      "grad_norm": 0.2953641712665558,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 71740
    },
    {
      "epoch": 63.54738706820195,
      "grad_norm": 0.2481699287891388,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 71745
    },
    {
      "epoch": 63.55181576616475,
      "grad_norm": 0.23895703256130219,
      "learning_rate": 1e-05,
      "loss": 0.9357,
      "step": 71750
    },
    {
      "epoch": 63.556244464127545,
      "grad_norm": 0.24418756365776062,
      "learning_rate": 1e-05,
      "loss": 0.9091,
      "step": 71755
    },
    {
      "epoch": 63.560673162090346,
      "grad_norm": 0.24183018505573273,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 71760
    },
    {
      "epoch": 63.56510186005315,
      "grad_norm": 0.23627054691314697,
      "learning_rate": 1e-05,
      "loss": 1.0355,
      "step": 71765
    },
    {
      "epoch": 63.56953055801594,
      "grad_norm": 0.2245316207408905,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 71770
    },
    {
      "epoch": 63.57395925597874,
      "grad_norm": 0.23440031707286835,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 71775
    },
    {
      "epoch": 63.578387953941544,
      "grad_norm": 0.2233452945947647,
      "learning_rate": 1e-05,
      "loss": 0.9233,
      "step": 71780
    },
    {
      "epoch": 63.58281665190434,
      "grad_norm": 0.28246229887008667,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 71785
    },
    {
      "epoch": 63.58724534986714,
      "grad_norm": 0.25369954109191895,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 71790
    },
    {
      "epoch": 63.59167404782994,
      "grad_norm": 0.2215196043252945,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 71795
    },
    {
      "epoch": 63.596102745792734,
      "grad_norm": 0.22083483636379242,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 71800
    },
    {
      "epoch": 63.600531443755536,
      "grad_norm": 0.2275639921426773,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 71805
    },
    {
      "epoch": 63.60496014171834,
      "grad_norm": 0.23594975471496582,
      "learning_rate": 1e-05,
      "loss": 1.0485,
      "step": 71810
    },
    {
      "epoch": 63.60938883968113,
      "grad_norm": 0.23736220598220825,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 71815
    },
    {
      "epoch": 63.61381753764393,
      "grad_norm": 0.23253846168518066,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 71820
    },
    {
      "epoch": 63.61824623560673,
      "grad_norm": 0.23370932042598724,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 71825
    },
    {
      "epoch": 63.62267493356953,
      "grad_norm": 0.2638593912124634,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 71830
    },
    {
      "epoch": 63.62710363153233,
      "grad_norm": 0.2509211301803589,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 71835
    },
    {
      "epoch": 63.63153232949513,
      "grad_norm": 0.23171059787273407,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 71840
    },
    {
      "epoch": 63.635961027457924,
      "grad_norm": 0.2203523814678192,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 71845
    },
    {
      "epoch": 63.640389725420725,
      "grad_norm": 0.2221454530954361,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 71850
    },
    {
      "epoch": 63.644818423383526,
      "grad_norm": 0.26503869891166687,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 71855
    },
    {
      "epoch": 63.64924712134633,
      "grad_norm": 0.2604830265045166,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 71860
    },
    {
      "epoch": 63.65367581930912,
      "grad_norm": 0.257392555475235,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 71865
    },
    {
      "epoch": 63.65810451727192,
      "grad_norm": 0.257062166929245,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 71870
    },
    {
      "epoch": 63.662533215234724,
      "grad_norm": 0.23192249238491058,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 71875
    },
    {
      "epoch": 63.66696191319752,
      "grad_norm": 0.2728661000728607,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 71880
    },
    {
      "epoch": 63.67139061116032,
      "grad_norm": 0.2383316457271576,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 71885
    },
    {
      "epoch": 63.67581930912312,
      "grad_norm": 0.24647721648216248,
      "learning_rate": 1e-05,
      "loss": 1.0405,
      "step": 71890
    },
    {
      "epoch": 63.680248007085915,
      "grad_norm": 0.18873251974582672,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 71895
    },
    {
      "epoch": 63.684676705048716,
      "grad_norm": 0.20158159732818604,
      "learning_rate": 1e-05,
      "loss": 0.888,
      "step": 71900
    },
    {
      "epoch": 63.68910540301152,
      "grad_norm": 0.222474604845047,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 71905
    },
    {
      "epoch": 63.69353410097431,
      "grad_norm": 0.21104910969734192,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 71910
    },
    {
      "epoch": 63.69796279893711,
      "grad_norm": 0.23700547218322754,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 71915
    },
    {
      "epoch": 63.702391496899914,
      "grad_norm": 0.21648821234703064,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 71920
    },
    {
      "epoch": 63.70682019486271,
      "grad_norm": 0.23354972898960114,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 71925
    },
    {
      "epoch": 63.71124889282551,
      "grad_norm": 0.25457149744033813,
      "learning_rate": 1e-05,
      "loss": 0.925,
      "step": 71930
    },
    {
      "epoch": 63.71567759078831,
      "grad_norm": 0.25546082854270935,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 71935
    },
    {
      "epoch": 63.720106288751104,
      "grad_norm": 0.24389851093292236,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 71940
    },
    {
      "epoch": 63.724534986713905,
      "grad_norm": 0.27593424916267395,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 71945
    },
    {
      "epoch": 63.72896368467671,
      "grad_norm": 0.23904871940612793,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 71950
    },
    {
      "epoch": 63.7333923826395,
      "grad_norm": 0.2235032021999359,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 71955
    },
    {
      "epoch": 63.7378210806023,
      "grad_norm": 0.2393193393945694,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 71960
    },
    {
      "epoch": 63.7422497785651,
      "grad_norm": 0.2622387409210205,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 71965
    },
    {
      "epoch": 63.7466784765279,
      "grad_norm": 0.23345951735973358,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 71970
    },
    {
      "epoch": 63.7511071744907,
      "grad_norm": 0.28554511070251465,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 71975
    },
    {
      "epoch": 63.7555358724535,
      "grad_norm": 0.33774247765541077,
      "learning_rate": 1e-05,
      "loss": 0.9396,
      "step": 71980
    },
    {
      "epoch": 63.7599645704163,
      "grad_norm": 0.23593991994857788,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 71985
    },
    {
      "epoch": 63.764393268379095,
      "grad_norm": 0.26405423879623413,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 71990
    },
    {
      "epoch": 63.768821966341896,
      "grad_norm": 0.26620954275131226,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 71995
    },
    {
      "epoch": 63.7732506643047,
      "grad_norm": 0.25992855429649353,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 72000
    },
    {
      "epoch": 63.77767936226749,
      "grad_norm": 0.27696675062179565,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 72005
    },
    {
      "epoch": 63.78210806023029,
      "grad_norm": 0.263491153717041,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 72010
    },
    {
      "epoch": 63.786536758193094,
      "grad_norm": 0.26416218280792236,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 72015
    },
    {
      "epoch": 63.79096545615589,
      "grad_norm": 0.23178862035274506,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 72020
    },
    {
      "epoch": 63.79539415411869,
      "grad_norm": 0.23109832406044006,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 72025
    },
    {
      "epoch": 63.79982285208149,
      "grad_norm": 0.2070980817079544,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 72030
    },
    {
      "epoch": 63.804251550044285,
      "grad_norm": 0.2491668462753296,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 72035
    },
    {
      "epoch": 63.808680248007086,
      "grad_norm": 0.24766898155212402,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 72040
    },
    {
      "epoch": 63.81310894596989,
      "grad_norm": 0.21834276616573334,
      "learning_rate": 1e-05,
      "loss": 0.9083,
      "step": 72045
    },
    {
      "epoch": 63.81753764393268,
      "grad_norm": 0.24240577220916748,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 72050
    },
    {
      "epoch": 63.82196634189548,
      "grad_norm": 0.28192341327667236,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 72055
    },
    {
      "epoch": 63.826395039858284,
      "grad_norm": 0.2615720331668854,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 72060
    },
    {
      "epoch": 63.83082373782108,
      "grad_norm": 0.28793010115623474,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 72065
    },
    {
      "epoch": 63.83525243578388,
      "grad_norm": 0.27413439750671387,
      "learning_rate": 1e-05,
      "loss": 0.9249,
      "step": 72070
    },
    {
      "epoch": 63.83968113374668,
      "grad_norm": 0.25113892555236816,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 72075
    },
    {
      "epoch": 63.844109831709474,
      "grad_norm": 0.21922902762889862,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 72080
    },
    {
      "epoch": 63.848538529672275,
      "grad_norm": 0.23861823976039886,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 72085
    },
    {
      "epoch": 63.85296722763508,
      "grad_norm": 0.2628633677959442,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 72090
    },
    {
      "epoch": 63.85739592559787,
      "grad_norm": 0.20280368626117706,
      "learning_rate": 1e-05,
      "loss": 0.9009,
      "step": 72095
    },
    {
      "epoch": 63.86182462356067,
      "grad_norm": 0.22723224759101868,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 72100
    },
    {
      "epoch": 63.86625332152347,
      "grad_norm": 0.2559661269187927,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 72105
    },
    {
      "epoch": 63.870682019486274,
      "grad_norm": 0.22751013934612274,
      "learning_rate": 1e-05,
      "loss": 0.9013,
      "step": 72110
    },
    {
      "epoch": 63.87511071744907,
      "grad_norm": 0.2760406732559204,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 72115
    },
    {
      "epoch": 63.87953941541187,
      "grad_norm": 0.23910555243492126,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 72120
    },
    {
      "epoch": 63.88396811337467,
      "grad_norm": 0.2394097000360489,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 72125
    },
    {
      "epoch": 63.888396811337465,
      "grad_norm": 0.2322576940059662,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 72130
    },
    {
      "epoch": 63.892825509300266,
      "grad_norm": 0.20410165190696716,
      "learning_rate": 1e-05,
      "loss": 0.9003,
      "step": 72135
    },
    {
      "epoch": 63.89725420726307,
      "grad_norm": 0.2347601056098938,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 72140
    },
    {
      "epoch": 63.90168290522586,
      "grad_norm": 0.256761759519577,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 72145
    },
    {
      "epoch": 63.90611160318866,
      "grad_norm": 0.22375032305717468,
      "learning_rate": 1e-05,
      "loss": 0.9116,
      "step": 72150
    },
    {
      "epoch": 63.910540301151464,
      "grad_norm": 0.23194104433059692,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 72155
    },
    {
      "epoch": 63.91496899911426,
      "grad_norm": 0.21748337149620056,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 72160
    },
    {
      "epoch": 63.91939769707706,
      "grad_norm": 0.22062312066555023,
      "learning_rate": 1e-05,
      "loss": 1.0071,
      "step": 72165
    },
    {
      "epoch": 63.92382639503986,
      "grad_norm": 0.30419921875,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 72170
    },
    {
      "epoch": 63.928255093002655,
      "grad_norm": 0.24396952986717224,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 72175
    },
    {
      "epoch": 63.932683790965456,
      "grad_norm": 0.2516191601753235,
      "learning_rate": 1e-05,
      "loss": 1.0284,
      "step": 72180
    },
    {
      "epoch": 63.93711248892826,
      "grad_norm": 0.23053854703903198,
      "learning_rate": 1e-05,
      "loss": 0.9303,
      "step": 72185
    },
    {
      "epoch": 63.94154118689105,
      "grad_norm": 0.2863028049468994,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 72190
    },
    {
      "epoch": 63.94596988485385,
      "grad_norm": 0.20314890146255493,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 72195
    },
    {
      "epoch": 63.95039858281665,
      "grad_norm": 0.20461708307266235,
      "learning_rate": 1e-05,
      "loss": 0.942,
      "step": 72200
    },
    {
      "epoch": 63.95482728077945,
      "grad_norm": 0.26836422085762024,
      "learning_rate": 1e-05,
      "loss": 0.9273,
      "step": 72205
    },
    {
      "epoch": 63.95925597874225,
      "grad_norm": 0.28519341349601746,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 72210
    },
    {
      "epoch": 63.96368467670505,
      "grad_norm": 0.25519445538520813,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 72215
    },
    {
      "epoch": 63.96811337466785,
      "grad_norm": 0.315518856048584,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 72220
    },
    {
      "epoch": 63.972542072630645,
      "grad_norm": 0.2512063980102539,
      "learning_rate": 1e-05,
      "loss": 1.0419,
      "step": 72225
    },
    {
      "epoch": 63.97697077059345,
      "grad_norm": 0.2222267985343933,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 72230
    },
    {
      "epoch": 63.98139946855625,
      "grad_norm": 0.2543501555919647,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 72235
    },
    {
      "epoch": 63.98582816651904,
      "grad_norm": 0.2504511773586273,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 72240
    },
    {
      "epoch": 63.99025686448184,
      "grad_norm": 0.2905332148075104,
      "learning_rate": 1e-05,
      "loss": 0.9329,
      "step": 72245
    },
    {
      "epoch": 63.994685562444644,
      "grad_norm": 0.2392033487558365,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 72250
    },
    {
      "epoch": 63.99911426040744,
      "grad_norm": 0.26713433861732483,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 72255
    },
    {
      "epoch": 64.00354295837023,
      "grad_norm": 0.2496492713689804,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 72260
    },
    {
      "epoch": 64.00797165633304,
      "grad_norm": 0.2492613047361374,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 72265
    },
    {
      "epoch": 64.01240035429583,
      "grad_norm": 0.230791836977005,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 72270
    },
    {
      "epoch": 64.01682905225863,
      "grad_norm": 0.21724149584770203,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 72275
    },
    {
      "epoch": 64.02125775022144,
      "grad_norm": 0.21066518127918243,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 72280
    },
    {
      "epoch": 64.02568644818423,
      "grad_norm": 0.2353580743074417,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 72285
    },
    {
      "epoch": 64.03011514614704,
      "grad_norm": 0.2535284757614136,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 72290
    },
    {
      "epoch": 64.03454384410983,
      "grad_norm": 0.2701921761035919,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 72295
    },
    {
      "epoch": 64.03897254207263,
      "grad_norm": 0.22650481760501862,
      "learning_rate": 1e-05,
      "loss": 0.9108,
      "step": 72300
    },
    {
      "epoch": 64.04340124003544,
      "grad_norm": 0.26875796914100647,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 72305
    },
    {
      "epoch": 64.04782993799823,
      "grad_norm": 0.27226752042770386,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 72310
    },
    {
      "epoch": 64.05225863596102,
      "grad_norm": 0.26001182198524475,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 72315
    },
    {
      "epoch": 64.05668733392383,
      "grad_norm": 0.2839227318763733,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 72320
    },
    {
      "epoch": 64.06111603188663,
      "grad_norm": 0.2377711981534958,
      "learning_rate": 1e-05,
      "loss": 1.0246,
      "step": 72325
    },
    {
      "epoch": 64.06554472984942,
      "grad_norm": 0.2790060341358185,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 72330
    },
    {
      "epoch": 64.06997342781223,
      "grad_norm": 0.2965708076953888,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 72335
    },
    {
      "epoch": 64.07440212577502,
      "grad_norm": 0.2182554006576538,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 72340
    },
    {
      "epoch": 64.07883082373782,
      "grad_norm": 0.2384357750415802,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 72345
    },
    {
      "epoch": 64.08325952170063,
      "grad_norm": 0.2502974569797516,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 72350
    },
    {
      "epoch": 64.08768821966342,
      "grad_norm": 0.24464303255081177,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 72355
    },
    {
      "epoch": 64.09211691762621,
      "grad_norm": 0.3505728244781494,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 72360
    },
    {
      "epoch": 64.09654561558902,
      "grad_norm": 0.2568131685256958,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 72365
    },
    {
      "epoch": 64.10097431355182,
      "grad_norm": 0.26593470573425293,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 72370
    },
    {
      "epoch": 64.10540301151461,
      "grad_norm": 0.24346861243247986,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 72375
    },
    {
      "epoch": 64.10983170947742,
      "grad_norm": 0.22596806287765503,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 72380
    },
    {
      "epoch": 64.11426040744021,
      "grad_norm": 0.2355467975139618,
      "learning_rate": 1e-05,
      "loss": 1.0159,
      "step": 72385
    },
    {
      "epoch": 64.118689105403,
      "grad_norm": 0.26663997769355774,
      "learning_rate": 1e-05,
      "loss": 0.8946,
      "step": 72390
    },
    {
      "epoch": 64.12311780336582,
      "grad_norm": 0.2563101053237915,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 72395
    },
    {
      "epoch": 64.12754650132861,
      "grad_norm": 0.23266133666038513,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 72400
    },
    {
      "epoch": 64.1319751992914,
      "grad_norm": 0.25268301367759705,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 72405
    },
    {
      "epoch": 64.13640389725421,
      "grad_norm": 0.27930742502212524,
      "learning_rate": 1e-05,
      "loss": 0.8997,
      "step": 72410
    },
    {
      "epoch": 64.140832595217,
      "grad_norm": 0.2327995002269745,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 72415
    },
    {
      "epoch": 64.1452612931798,
      "grad_norm": 0.24939550459384918,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 72420
    },
    {
      "epoch": 64.14968999114261,
      "grad_norm": 0.3613887131214142,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 72425
    },
    {
      "epoch": 64.1541186891054,
      "grad_norm": 0.25087034702301025,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 72430
    },
    {
      "epoch": 64.1585473870682,
      "grad_norm": 0.2715052366256714,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 72435
    },
    {
      "epoch": 64.162976085031,
      "grad_norm": 0.2986374795436859,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 72440
    },
    {
      "epoch": 64.1674047829938,
      "grad_norm": 0.3153024911880493,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 72445
    },
    {
      "epoch": 64.1718334809566,
      "grad_norm": 0.2618449032306671,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 72450
    },
    {
      "epoch": 64.1762621789194,
      "grad_norm": 0.24955107271671295,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 72455
    },
    {
      "epoch": 64.1806908768822,
      "grad_norm": 0.267992228269577,
      "learning_rate": 1e-05,
      "loss": 0.9084,
      "step": 72460
    },
    {
      "epoch": 64.18511957484499,
      "grad_norm": 0.23598214983940125,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 72465
    },
    {
      "epoch": 64.1895482728078,
      "grad_norm": 0.262320876121521,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 72470
    },
    {
      "epoch": 64.19397697077059,
      "grad_norm": 0.20598241686820984,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 72475
    },
    {
      "epoch": 64.19840566873339,
      "grad_norm": 0.28596749901771545,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 72480
    },
    {
      "epoch": 64.2028343666962,
      "grad_norm": 0.2384020835161209,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 72485
    },
    {
      "epoch": 64.20726306465899,
      "grad_norm": 0.27076131105422974,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 72490
    },
    {
      "epoch": 64.21169176262178,
      "grad_norm": 0.22116747498512268,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 72495
    },
    {
      "epoch": 64.21612046058459,
      "grad_norm": 0.3124556243419647,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 72500
    },
    {
      "epoch": 64.22054915854739,
      "grad_norm": 0.2269141525030136,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 72505
    },
    {
      "epoch": 64.22497785651018,
      "grad_norm": 0.22330693900585175,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 72510
    },
    {
      "epoch": 64.22940655447299,
      "grad_norm": 0.2259770631790161,
      "learning_rate": 1e-05,
      "loss": 1.0324,
      "step": 72515
    },
    {
      "epoch": 64.23383525243578,
      "grad_norm": 0.2648800015449524,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 72520
    },
    {
      "epoch": 64.23826395039858,
      "grad_norm": 0.22514502704143524,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 72525
    },
    {
      "epoch": 64.24269264836138,
      "grad_norm": 0.22584038972854614,
      "learning_rate": 1e-05,
      "loss": 0.8964,
      "step": 72530
    },
    {
      "epoch": 64.24712134632418,
      "grad_norm": 0.21360959112644196,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 72535
    },
    {
      "epoch": 64.25155004428699,
      "grad_norm": 0.23636780679225922,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 72540
    },
    {
      "epoch": 64.25597874224978,
      "grad_norm": 0.2663382291793823,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 72545
    },
    {
      "epoch": 64.26040744021257,
      "grad_norm": 0.3076880872249603,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 72550
    },
    {
      "epoch": 64.26483613817538,
      "grad_norm": 0.2816512882709503,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 72555
    },
    {
      "epoch": 64.26926483613818,
      "grad_norm": 0.3166913688182831,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 72560
    },
    {
      "epoch": 64.27369353410097,
      "grad_norm": 0.26130399107933044,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 72565
    },
    {
      "epoch": 64.27812223206378,
      "grad_norm": 0.18850895762443542,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 72570
    },
    {
      "epoch": 64.28255093002657,
      "grad_norm": 0.24359814822673798,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 72575
    },
    {
      "epoch": 64.28697962798937,
      "grad_norm": 0.21340949833393097,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 72580
    },
    {
      "epoch": 64.29140832595218,
      "grad_norm": 0.22033889591693878,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 72585
    },
    {
      "epoch": 64.29583702391497,
      "grad_norm": 0.24071216583251953,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 72590
    },
    {
      "epoch": 64.30026572187776,
      "grad_norm": 0.21995486319065094,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 72595
    },
    {
      "epoch": 64.30469441984057,
      "grad_norm": 0.23339150846004486,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 72600
    },
    {
      "epoch": 64.30912311780337,
      "grad_norm": 0.24846330285072327,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 72605
    },
    {
      "epoch": 64.31355181576616,
      "grad_norm": 0.21609875559806824,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 72610
    },
    {
      "epoch": 64.31798051372897,
      "grad_norm": 0.22578540444374084,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 72615
    },
    {
      "epoch": 64.32240921169176,
      "grad_norm": 0.2506599426269531,
      "learning_rate": 1e-05,
      "loss": 0.9192,
      "step": 72620
    },
    {
      "epoch": 64.32683790965456,
      "grad_norm": 0.23522557318210602,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 72625
    },
    {
      "epoch": 64.33126660761737,
      "grad_norm": 0.2518937587738037,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 72630
    },
    {
      "epoch": 64.33569530558016,
      "grad_norm": 0.2291484773159027,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 72635
    },
    {
      "epoch": 64.34012400354295,
      "grad_norm": 0.2440386414527893,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 72640
    },
    {
      "epoch": 64.34455270150576,
      "grad_norm": 0.24349361658096313,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 72645
    },
    {
      "epoch": 64.34898139946856,
      "grad_norm": 0.240842804312706,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 72650
    },
    {
      "epoch": 64.35341009743135,
      "grad_norm": 0.235815167427063,
      "learning_rate": 1e-05,
      "loss": 0.9364,
      "step": 72655
    },
    {
      "epoch": 64.35783879539416,
      "grad_norm": 0.27491965889930725,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 72660
    },
    {
      "epoch": 64.36226749335695,
      "grad_norm": 0.22957608103752136,
      "learning_rate": 1e-05,
      "loss": 1.0316,
      "step": 72665
    },
    {
      "epoch": 64.36669619131975,
      "grad_norm": 0.2438986599445343,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 72670
    },
    {
      "epoch": 64.37112488928256,
      "grad_norm": 0.2598194479942322,
      "learning_rate": 1e-05,
      "loss": 1.0219,
      "step": 72675
    },
    {
      "epoch": 64.37555358724535,
      "grad_norm": 0.2896093428134918,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 72680
    },
    {
      "epoch": 64.37998228520814,
      "grad_norm": 0.2230769246816635,
      "learning_rate": 1e-05,
      "loss": 0.9096,
      "step": 72685
    },
    {
      "epoch": 64.38441098317095,
      "grad_norm": 0.24404607713222504,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 72690
    },
    {
      "epoch": 64.38883968113375,
      "grad_norm": 0.2649652659893036,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 72695
    },
    {
      "epoch": 64.39326837909654,
      "grad_norm": 0.25047820806503296,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 72700
    },
    {
      "epoch": 64.39769707705935,
      "grad_norm": 0.2452995777130127,
      "learning_rate": 1e-05,
      "loss": 0.897,
      "step": 72705
    },
    {
      "epoch": 64.40212577502214,
      "grad_norm": 0.22855930030345917,
      "learning_rate": 1e-05,
      "loss": 1.0535,
      "step": 72710
    },
    {
      "epoch": 64.40655447298494,
      "grad_norm": 0.21578523516654968,
      "learning_rate": 1e-05,
      "loss": 0.9146,
      "step": 72715
    },
    {
      "epoch": 64.41098317094774,
      "grad_norm": 0.20851679146289825,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 72720
    },
    {
      "epoch": 64.41541186891054,
      "grad_norm": 0.22085556387901306,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 72725
    },
    {
      "epoch": 64.41984056687333,
      "grad_norm": 0.24598106741905212,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 72730
    },
    {
      "epoch": 64.42426926483614,
      "grad_norm": 0.2659984230995178,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 72735
    },
    {
      "epoch": 64.42869796279894,
      "grad_norm": 0.20612941682338715,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 72740
    },
    {
      "epoch": 64.43312666076173,
      "grad_norm": 0.21529501676559448,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 72745
    },
    {
      "epoch": 64.43755535872454,
      "grad_norm": 0.263980507850647,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 72750
    },
    {
      "epoch": 64.44198405668733,
      "grad_norm": 0.22312942147254944,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 72755
    },
    {
      "epoch": 64.44641275465013,
      "grad_norm": 0.24249732494354248,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 72760
    },
    {
      "epoch": 64.45084145261293,
      "grad_norm": 0.21512573957443237,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 72765
    },
    {
      "epoch": 64.45527015057573,
      "grad_norm": 0.23423995077610016,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 72770
    },
    {
      "epoch": 64.45969884853854,
      "grad_norm": 0.26808351278305054,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 72775
    },
    {
      "epoch": 64.46412754650133,
      "grad_norm": 0.2010992467403412,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 72780
    },
    {
      "epoch": 64.46855624446412,
      "grad_norm": 0.26097404956817627,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 72785
    },
    {
      "epoch": 64.47298494242693,
      "grad_norm": 0.23877692222595215,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 72790
    },
    {
      "epoch": 64.47741364038973,
      "grad_norm": 0.19634488224983215,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 72795
    },
    {
      "epoch": 64.48184233835252,
      "grad_norm": 0.21636557579040527,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 72800
    },
    {
      "epoch": 64.48627103631533,
      "grad_norm": 0.2246290147304535,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 72805
    },
    {
      "epoch": 64.49069973427812,
      "grad_norm": 0.22305701673030853,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 72810
    },
    {
      "epoch": 64.49512843224092,
      "grad_norm": 0.20937050879001617,
      "learning_rate": 1e-05,
      "loss": 0.9275,
      "step": 72815
    },
    {
      "epoch": 64.49955713020373,
      "grad_norm": 0.22782689332962036,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 72820
    },
    {
      "epoch": 64.50398582816652,
      "grad_norm": 0.24238668382167816,
      "learning_rate": 1e-05,
      "loss": 0.9293,
      "step": 72825
    },
    {
      "epoch": 64.50841452612931,
      "grad_norm": 0.2320430874824524,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 72830
    },
    {
      "epoch": 64.51284322409212,
      "grad_norm": 0.31743624806404114,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 72835
    },
    {
      "epoch": 64.51727192205492,
      "grad_norm": 0.216452956199646,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 72840
    },
    {
      "epoch": 64.52170062001771,
      "grad_norm": 0.23494507372379303,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 72845
    },
    {
      "epoch": 64.52612931798052,
      "grad_norm": 0.22299674153327942,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 72850
    },
    {
      "epoch": 64.53055801594331,
      "grad_norm": 0.22777986526489258,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 72855
    },
    {
      "epoch": 64.53498671390611,
      "grad_norm": 0.22655801475048065,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 72860
    },
    {
      "epoch": 64.53941541186892,
      "grad_norm": 0.26625117659568787,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 72865
    },
    {
      "epoch": 64.54384410983171,
      "grad_norm": 0.22187185287475586,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 72870
    },
    {
      "epoch": 64.5482728077945,
      "grad_norm": 0.2565276324748993,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 72875
    },
    {
      "epoch": 64.55270150575731,
      "grad_norm": 0.31812727451324463,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 72880
    },
    {
      "epoch": 64.5571302037201,
      "grad_norm": 0.30237773060798645,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 72885
    },
    {
      "epoch": 64.5615589016829,
      "grad_norm": 0.22413788735866547,
      "learning_rate": 1e-05,
      "loss": 0.9292,
      "step": 72890
    },
    {
      "epoch": 64.56598759964571,
      "grad_norm": 0.26362931728363037,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 72895
    },
    {
      "epoch": 64.5704162976085,
      "grad_norm": 0.22944939136505127,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 72900
    },
    {
      "epoch": 64.5748449955713,
      "grad_norm": 0.24276171624660492,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 72905
    },
    {
      "epoch": 64.5792736935341,
      "grad_norm": 0.2938569486141205,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 72910
    },
    {
      "epoch": 64.5837023914969,
      "grad_norm": 0.2229621559381485,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 72915
    },
    {
      "epoch": 64.5881310894597,
      "grad_norm": 0.22491954267024994,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 72920
    },
    {
      "epoch": 64.5925597874225,
      "grad_norm": 0.2031039297580719,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 72925
    },
    {
      "epoch": 64.5969884853853,
      "grad_norm": 0.23491178452968597,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 72930
    },
    {
      "epoch": 64.60141718334809,
      "grad_norm": 0.278828501701355,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 72935
    },
    {
      "epoch": 64.6058458813109,
      "grad_norm": 0.2112453579902649,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 72940
    },
    {
      "epoch": 64.61027457927369,
      "grad_norm": 0.3014611601829529,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 72945
    },
    {
      "epoch": 64.61470327723649,
      "grad_norm": 0.22272783517837524,
      "learning_rate": 1e-05,
      "loss": 0.9255,
      "step": 72950
    },
    {
      "epoch": 64.6191319751993,
      "grad_norm": 0.2961048185825348,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 72955
    },
    {
      "epoch": 64.62356067316209,
      "grad_norm": 0.22155489027500153,
      "learning_rate": 1e-05,
      "loss": 0.8963,
      "step": 72960
    },
    {
      "epoch": 64.62798937112488,
      "grad_norm": 0.22372622787952423,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 72965
    },
    {
      "epoch": 64.63241806908769,
      "grad_norm": 0.22023390233516693,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 72970
    },
    {
      "epoch": 64.63684676705049,
      "grad_norm": 0.23097185790538788,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 72975
    },
    {
      "epoch": 64.64127546501328,
      "grad_norm": 0.2689822316169739,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 72980
    },
    {
      "epoch": 64.64570416297609,
      "grad_norm": 0.24470609426498413,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 72985
    },
    {
      "epoch": 64.65013286093888,
      "grad_norm": 0.23677454888820648,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 72990
    },
    {
      "epoch": 64.65456155890168,
      "grad_norm": 0.19466763734817505,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 72995
    },
    {
      "epoch": 64.65899025686448,
      "grad_norm": 0.22303299605846405,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 73000
    },
    {
      "epoch": 64.66341895482728,
      "grad_norm": 0.2066963165998459,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 73005
    },
    {
      "epoch": 64.66784765279007,
      "grad_norm": 0.23120933771133423,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 73010
    },
    {
      "epoch": 64.67227635075288,
      "grad_norm": 0.2797692120075226,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 73015
    },
    {
      "epoch": 64.67670504871568,
      "grad_norm": 0.2526929974555969,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 73020
    },
    {
      "epoch": 64.68113374667848,
      "grad_norm": 0.2414504438638687,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 73025
    },
    {
      "epoch": 64.68556244464128,
      "grad_norm": 0.23376959562301636,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 73030
    },
    {
      "epoch": 64.68999114260407,
      "grad_norm": 0.2564552426338196,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 73035
    },
    {
      "epoch": 64.69441984056688,
      "grad_norm": 0.2301609367132187,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 73040
    },
    {
      "epoch": 64.69884853852967,
      "grad_norm": 0.22173574566841125,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 73045
    },
    {
      "epoch": 64.70327723649247,
      "grad_norm": 0.25225400924682617,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 73050
    },
    {
      "epoch": 64.70770593445528,
      "grad_norm": 0.21192489564418793,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 73055
    },
    {
      "epoch": 64.71213463241807,
      "grad_norm": 0.26313936710357666,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 73060
    },
    {
      "epoch": 64.71656333038086,
      "grad_norm": 0.22775939106941223,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 73065
    },
    {
      "epoch": 64.72099202834367,
      "grad_norm": 0.22733718156814575,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 73070
    },
    {
      "epoch": 64.72542072630647,
      "grad_norm": 0.21024346351623535,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 73075
    },
    {
      "epoch": 64.72984942426926,
      "grad_norm": 0.24041904509067535,
      "learning_rate": 1e-05,
      "loss": 0.9312,
      "step": 73080
    },
    {
      "epoch": 64.73427812223207,
      "grad_norm": 0.20395861566066742,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 73085
    },
    {
      "epoch": 64.73870682019486,
      "grad_norm": 0.22118213772773743,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 73090
    },
    {
      "epoch": 64.74313551815766,
      "grad_norm": 0.21690507233142853,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 73095
    },
    {
      "epoch": 64.74756421612047,
      "grad_norm": 0.2162257432937622,
      "learning_rate": 1e-05,
      "loss": 1.0369,
      "step": 73100
    },
    {
      "epoch": 64.75199291408326,
      "grad_norm": 0.20733825862407684,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 73105
    },
    {
      "epoch": 64.75642161204605,
      "grad_norm": 0.24995101988315582,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 73110
    },
    {
      "epoch": 64.76085031000886,
      "grad_norm": 0.23254866898059845,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 73115
    },
    {
      "epoch": 64.76527900797166,
      "grad_norm": 0.266143262386322,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 73120
    },
    {
      "epoch": 64.76970770593445,
      "grad_norm": 0.2318686693906784,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 73125
    },
    {
      "epoch": 64.77413640389726,
      "grad_norm": 0.20897050201892853,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 73130
    },
    {
      "epoch": 64.77856510186005,
      "grad_norm": 0.3021281957626343,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 73135
    },
    {
      "epoch": 64.78299379982285,
      "grad_norm": 0.24280782043933868,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 73140
    },
    {
      "epoch": 64.78742249778566,
      "grad_norm": 0.2503491938114166,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 73145
    },
    {
      "epoch": 64.79185119574845,
      "grad_norm": 0.21739760041236877,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 73150
    },
    {
      "epoch": 64.79627989371124,
      "grad_norm": 0.2679872214794159,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 73155
    },
    {
      "epoch": 64.80070859167405,
      "grad_norm": 0.2514238953590393,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 73160
    },
    {
      "epoch": 64.80513728963685,
      "grad_norm": 0.24457912147045135,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 73165
    },
    {
      "epoch": 64.80956598759964,
      "grad_norm": 0.23919068276882172,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 73170
    },
    {
      "epoch": 64.81399468556245,
      "grad_norm": 0.24494221806526184,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 73175
    },
    {
      "epoch": 64.81842338352524,
      "grad_norm": 0.258117139339447,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 73180
    },
    {
      "epoch": 64.82285208148804,
      "grad_norm": 0.23422940075397491,
      "learning_rate": 1e-05,
      "loss": 0.9259,
      "step": 73185
    },
    {
      "epoch": 64.82728077945085,
      "grad_norm": 0.2692304253578186,
      "learning_rate": 1e-05,
      "loss": 1.0308,
      "step": 73190
    },
    {
      "epoch": 64.83170947741364,
      "grad_norm": 0.23676486313343048,
      "learning_rate": 1e-05,
      "loss": 0.9493,
      "step": 73195
    },
    {
      "epoch": 64.83613817537643,
      "grad_norm": 0.2411409467458725,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 73200
    },
    {
      "epoch": 64.84056687333924,
      "grad_norm": 0.24172060191631317,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 73205
    },
    {
      "epoch": 64.84499557130204,
      "grad_norm": 0.24388645589351654,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 73210
    },
    {
      "epoch": 64.84942426926483,
      "grad_norm": 0.27366477251052856,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 73215
    },
    {
      "epoch": 64.85385296722764,
      "grad_norm": 0.3080722391605377,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 73220
    },
    {
      "epoch": 64.85828166519043,
      "grad_norm": 0.2512790858745575,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 73225
    },
    {
      "epoch": 64.86271036315323,
      "grad_norm": 0.259541392326355,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 73230
    },
    {
      "epoch": 64.86713906111603,
      "grad_norm": 0.3149668574333191,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 73235
    },
    {
      "epoch": 64.87156775907883,
      "grad_norm": 0.24256470799446106,
      "learning_rate": 1e-05,
      "loss": 0.9163,
      "step": 73240
    },
    {
      "epoch": 64.87599645704162,
      "grad_norm": 0.2324226349592209,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 73245
    },
    {
      "epoch": 64.88042515500443,
      "grad_norm": 0.2165728211402893,
      "learning_rate": 1e-05,
      "loss": 0.9296,
      "step": 73250
    },
    {
      "epoch": 64.88485385296723,
      "grad_norm": 0.21871446073055267,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 73255
    },
    {
      "epoch": 64.88928255093003,
      "grad_norm": 0.21416890621185303,
      "learning_rate": 1e-05,
      "loss": 1.0059,
      "step": 73260
    },
    {
      "epoch": 64.89371124889283,
      "grad_norm": 0.2610345780849457,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 73265
    },
    {
      "epoch": 64.89813994685562,
      "grad_norm": 0.23710957169532776,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 73270
    },
    {
      "epoch": 64.90256864481843,
      "grad_norm": 0.2321448177099228,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 73275
    },
    {
      "epoch": 64.90699734278122,
      "grad_norm": 0.23969289660453796,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 73280
    },
    {
      "epoch": 64.91142604074402,
      "grad_norm": 0.23932261765003204,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 73285
    },
    {
      "epoch": 64.91585473870683,
      "grad_norm": 0.24695177376270294,
      "learning_rate": 1e-05,
      "loss": 0.9254,
      "step": 73290
    },
    {
      "epoch": 64.92028343666962,
      "grad_norm": 0.21476314961910248,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 73295
    },
    {
      "epoch": 64.92471213463241,
      "grad_norm": 0.24135638773441315,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 73300
    },
    {
      "epoch": 64.92914083259522,
      "grad_norm": 0.2539912164211273,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 73305
    },
    {
      "epoch": 64.93356953055802,
      "grad_norm": 0.2453600913286209,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 73310
    },
    {
      "epoch": 64.93799822852081,
      "grad_norm": 0.2225649058818817,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 73315
    },
    {
      "epoch": 64.94242692648362,
      "grad_norm": 0.21839207410812378,
      "learning_rate": 1e-05,
      "loss": 0.9125,
      "step": 73320
    },
    {
      "epoch": 64.94685562444641,
      "grad_norm": 0.2366826832294464,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 73325
    },
    {
      "epoch": 64.95128432240921,
      "grad_norm": 0.22882409393787384,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 73330
    },
    {
      "epoch": 64.95571302037202,
      "grad_norm": 0.21799173951148987,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 73335
    },
    {
      "epoch": 64.96014171833481,
      "grad_norm": 0.23177845776081085,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 73340
    },
    {
      "epoch": 64.9645704162976,
      "grad_norm": 0.20961230993270874,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 73345
    },
    {
      "epoch": 64.96899911426041,
      "grad_norm": 0.27216798067092896,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 73350
    },
    {
      "epoch": 64.9734278122232,
      "grad_norm": 0.20005322992801666,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 73355
    },
    {
      "epoch": 64.977856510186,
      "grad_norm": 0.22230158746242523,
      "learning_rate": 1e-05,
      "loss": 1.048,
      "step": 73360
    },
    {
      "epoch": 64.98228520814881,
      "grad_norm": 0.2673405706882477,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 73365
    },
    {
      "epoch": 64.9867139061116,
      "grad_norm": 0.2241121232509613,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 73370
    },
    {
      "epoch": 64.9911426040744,
      "grad_norm": 0.2412911057472229,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 73375
    },
    {
      "epoch": 64.9955713020372,
      "grad_norm": 0.2260071337223053,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 73380
    },
    {
      "epoch": 65.0,
      "grad_norm": 0.2273583561182022,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 73385
    },
    {
      "epoch": 65.0044286979628,
      "grad_norm": 0.2455538660287857,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 73390
    },
    {
      "epoch": 65.0088573959256,
      "grad_norm": 0.30271026492118835,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 73395
    },
    {
      "epoch": 65.0132860938884,
      "grad_norm": 0.2328118085861206,
      "learning_rate": 1e-05,
      "loss": 0.9367,
      "step": 73400
    },
    {
      "epoch": 65.01771479185119,
      "grad_norm": 0.245832160115242,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 73405
    },
    {
      "epoch": 65.022143489814,
      "grad_norm": 0.26279518008232117,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 73410
    },
    {
      "epoch": 65.0265721877768,
      "grad_norm": 0.2553558051586151,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 73415
    },
    {
      "epoch": 65.03100088573959,
      "grad_norm": 0.2141643613576889,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 73420
    },
    {
      "epoch": 65.0354295837024,
      "grad_norm": 0.23492883145809174,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 73425
    },
    {
      "epoch": 65.03985828166519,
      "grad_norm": 0.2577956020832062,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 73430
    },
    {
      "epoch": 65.04428697962798,
      "grad_norm": 0.21198050677776337,
      "learning_rate": 1e-05,
      "loss": 0.912,
      "step": 73435
    },
    {
      "epoch": 65.04871567759079,
      "grad_norm": 0.24318081140518188,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 73440
    },
    {
      "epoch": 65.05314437555359,
      "grad_norm": 0.25270235538482666,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 73445
    },
    {
      "epoch": 65.05757307351638,
      "grad_norm": 0.23072810471057892,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 73450
    },
    {
      "epoch": 65.06200177147919,
      "grad_norm": 0.24660111963748932,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 73455
    },
    {
      "epoch": 65.06643046944198,
      "grad_norm": 0.2252207100391388,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 73460
    },
    {
      "epoch": 65.07085916740478,
      "grad_norm": 0.25394079089164734,
      "learning_rate": 1e-05,
      "loss": 0.9381,
      "step": 73465
    },
    {
      "epoch": 65.07528786536759,
      "grad_norm": 0.25519198179244995,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 73470
    },
    {
      "epoch": 65.07971656333038,
      "grad_norm": 0.23888011276721954,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 73475
    },
    {
      "epoch": 65.08414526129317,
      "grad_norm": 0.2619520425796509,
      "learning_rate": 1e-05,
      "loss": 0.9342,
      "step": 73480
    },
    {
      "epoch": 65.08857395925598,
      "grad_norm": 0.25942203402519226,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 73485
    },
    {
      "epoch": 65.09300265721878,
      "grad_norm": 0.24151575565338135,
      "learning_rate": 1e-05,
      "loss": 0.9224,
      "step": 73490
    },
    {
      "epoch": 65.09743135518157,
      "grad_norm": 0.23074884712696075,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 73495
    },
    {
      "epoch": 65.10186005314438,
      "grad_norm": 0.28484755754470825,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 73500
    },
    {
      "epoch": 65.10628875110717,
      "grad_norm": 0.24937203526496887,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 73505
    },
    {
      "epoch": 65.11071744906998,
      "grad_norm": 0.22056695818901062,
      "learning_rate": 1e-05,
      "loss": 0.9183,
      "step": 73510
    },
    {
      "epoch": 65.11514614703277,
      "grad_norm": 0.2412230223417282,
      "learning_rate": 1e-05,
      "loss": 0.9408,
      "step": 73515
    },
    {
      "epoch": 65.11957484499557,
      "grad_norm": 0.2385357767343521,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 73520
    },
    {
      "epoch": 65.12400354295838,
      "grad_norm": 0.226994127035141,
      "learning_rate": 1e-05,
      "loss": 0.9266,
      "step": 73525
    },
    {
      "epoch": 65.12843224092117,
      "grad_norm": 0.23409655690193176,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 73530
    },
    {
      "epoch": 65.13286093888397,
      "grad_norm": 0.2716817557811737,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 73535
    },
    {
      "epoch": 65.13728963684677,
      "grad_norm": 0.22187738120555878,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 73540
    },
    {
      "epoch": 65.14171833480957,
      "grad_norm": 0.28643107414245605,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 73545
    },
    {
      "epoch": 65.14614703277236,
      "grad_norm": 0.23184771835803986,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 73550
    },
    {
      "epoch": 65.15057573073517,
      "grad_norm": 0.2321070432662964,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 73555
    },
    {
      "epoch": 65.15500442869796,
      "grad_norm": 0.23224906623363495,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 73560
    },
    {
      "epoch": 65.15943312666076,
      "grad_norm": 0.22908441722393036,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 73565
    },
    {
      "epoch": 65.16386182462357,
      "grad_norm": 0.22927556931972504,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 73570
    },
    {
      "epoch": 65.16829052258636,
      "grad_norm": 0.26013466715812683,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 73575
    },
    {
      "epoch": 65.17271922054915,
      "grad_norm": 0.2474551647901535,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 73580
    },
    {
      "epoch": 65.17714791851196,
      "grad_norm": 0.23870502412319183,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 73585
    },
    {
      "epoch": 65.18157661647476,
      "grad_norm": 0.24945245683193207,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 73590
    },
    {
      "epoch": 65.18600531443755,
      "grad_norm": 0.21831509470939636,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 73595
    },
    {
      "epoch": 65.19043401240036,
      "grad_norm": 0.26080167293548584,
      "learning_rate": 1e-05,
      "loss": 0.9408,
      "step": 73600
    },
    {
      "epoch": 65.19486271036315,
      "grad_norm": 0.22626294195652008,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 73605
    },
    {
      "epoch": 65.19929140832595,
      "grad_norm": 0.22210374474525452,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 73610
    },
    {
      "epoch": 65.20372010628876,
      "grad_norm": 0.2490384578704834,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 73615
    },
    {
      "epoch": 65.20814880425155,
      "grad_norm": 0.2724076509475708,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 73620
    },
    {
      "epoch": 65.21257750221434,
      "grad_norm": 0.26132825016975403,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 73625
    },
    {
      "epoch": 65.21700620017715,
      "grad_norm": 0.2411799132823944,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 73630
    },
    {
      "epoch": 65.22143489813995,
      "grad_norm": 0.23282384872436523,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 73635
    },
    {
      "epoch": 65.22586359610274,
      "grad_norm": 0.2657560110092163,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 73640
    },
    {
      "epoch": 65.23029229406555,
      "grad_norm": 0.23267632722854614,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 73645
    },
    {
      "epoch": 65.23472099202834,
      "grad_norm": 0.2692844569683075,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 73650
    },
    {
      "epoch": 65.23914968999114,
      "grad_norm": 0.2364145815372467,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 73655
    },
    {
      "epoch": 65.24357838795395,
      "grad_norm": 0.24558281898498535,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 73660
    },
    {
      "epoch": 65.24800708591674,
      "grad_norm": 0.23422978818416595,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 73665
    },
    {
      "epoch": 65.25243578387953,
      "grad_norm": 0.2597558796405792,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 73670
    },
    {
      "epoch": 65.25686448184234,
      "grad_norm": 0.22502420842647552,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 73675
    },
    {
      "epoch": 65.26129317980514,
      "grad_norm": 0.30387818813323975,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 73680
    },
    {
      "epoch": 65.26572187776793,
      "grad_norm": 0.2563590705394745,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 73685
    },
    {
      "epoch": 65.27015057573074,
      "grad_norm": 0.22517463564872742,
      "learning_rate": 1e-05,
      "loss": 0.9267,
      "step": 73690
    },
    {
      "epoch": 65.27457927369353,
      "grad_norm": 0.210673525929451,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 73695
    },
    {
      "epoch": 65.27900797165633,
      "grad_norm": 0.24069519340991974,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 73700
    },
    {
      "epoch": 65.28343666961914,
      "grad_norm": 0.2506633996963501,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 73705
    },
    {
      "epoch": 65.28786536758193,
      "grad_norm": 0.22733008861541748,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 73710
    },
    {
      "epoch": 65.29229406554472,
      "grad_norm": 0.21477723121643066,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 73715
    },
    {
      "epoch": 65.29672276350753,
      "grad_norm": 0.23473884165287018,
      "learning_rate": 1e-05,
      "loss": 0.9283,
      "step": 73720
    },
    {
      "epoch": 65.30115146147033,
      "grad_norm": 0.23256225883960724,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 73725
    },
    {
      "epoch": 65.30558015943312,
      "grad_norm": 0.2348213940858841,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 73730
    },
    {
      "epoch": 65.31000885739593,
      "grad_norm": 0.2542742192745209,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 73735
    },
    {
      "epoch": 65.31443755535872,
      "grad_norm": 0.24554532766342163,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 73740
    },
    {
      "epoch": 65.31886625332152,
      "grad_norm": 0.24468572437763214,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 73745
    },
    {
      "epoch": 65.32329495128432,
      "grad_norm": 0.24310672283172607,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 73750
    },
    {
      "epoch": 65.32772364924712,
      "grad_norm": 0.24722982943058014,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 73755
    },
    {
      "epoch": 65.33215234720993,
      "grad_norm": 0.23034626245498657,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 73760
    },
    {
      "epoch": 65.33658104517272,
      "grad_norm": 0.23216694593429565,
      "learning_rate": 1e-05,
      "loss": 0.9195,
      "step": 73765
    },
    {
      "epoch": 65.34100974313552,
      "grad_norm": 0.22310815751552582,
      "learning_rate": 1e-05,
      "loss": 0.9149,
      "step": 73770
    },
    {
      "epoch": 65.34543844109832,
      "grad_norm": 0.246931791305542,
      "learning_rate": 1e-05,
      "loss": 0.9287,
      "step": 73775
    },
    {
      "epoch": 65.34986713906112,
      "grad_norm": 0.23285101354122162,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 73780
    },
    {
      "epoch": 65.35429583702391,
      "grad_norm": 0.23956657946109772,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 73785
    },
    {
      "epoch": 65.35872453498672,
      "grad_norm": 0.2419046014547348,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 73790
    },
    {
      "epoch": 65.36315323294951,
      "grad_norm": 0.23861533403396606,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 73795
    },
    {
      "epoch": 65.36758193091231,
      "grad_norm": 0.25837117433547974,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 73800
    },
    {
      "epoch": 65.37201062887512,
      "grad_norm": 0.2363186776638031,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 73805
    },
    {
      "epoch": 65.37643932683791,
      "grad_norm": 0.23103666305541992,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 73810
    },
    {
      "epoch": 65.3808680248007,
      "grad_norm": 0.24365819990634918,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 73815
    },
    {
      "epoch": 65.38529672276351,
      "grad_norm": 0.2451895922422409,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 73820
    },
    {
      "epoch": 65.38972542072631,
      "grad_norm": 0.24540302157402039,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 73825
    },
    {
      "epoch": 65.3941541186891,
      "grad_norm": 0.2697092890739441,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 73830
    },
    {
      "epoch": 65.39858281665191,
      "grad_norm": 0.2869071364402771,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 73835
    },
    {
      "epoch": 65.4030115146147,
      "grad_norm": 0.2370927929878235,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 73840
    },
    {
      "epoch": 65.4074402125775,
      "grad_norm": 0.2766813635826111,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 73845
    },
    {
      "epoch": 65.4118689105403,
      "grad_norm": 0.25103282928466797,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 73850
    },
    {
      "epoch": 65.4162976085031,
      "grad_norm": 0.2520640194416046,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 73855
    },
    {
      "epoch": 65.4207263064659,
      "grad_norm": 0.215667262673378,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 73860
    },
    {
      "epoch": 65.4251550044287,
      "grad_norm": 0.22285738587379456,
      "learning_rate": 1e-05,
      "loss": 0.9207,
      "step": 73865
    },
    {
      "epoch": 65.4295837023915,
      "grad_norm": 0.2278720736503601,
      "learning_rate": 1e-05,
      "loss": 0.9381,
      "step": 73870
    },
    {
      "epoch": 65.43401240035429,
      "grad_norm": 0.24399523437023163,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 73875
    },
    {
      "epoch": 65.4384410983171,
      "grad_norm": 0.23723246157169342,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 73880
    },
    {
      "epoch": 65.4428697962799,
      "grad_norm": 0.2579454481601715,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 73885
    },
    {
      "epoch": 65.44729849424269,
      "grad_norm": 0.27759817242622375,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 73890
    },
    {
      "epoch": 65.4517271922055,
      "grad_norm": 0.24446061253547668,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 73895
    },
    {
      "epoch": 65.45615589016829,
      "grad_norm": 0.26221561431884766,
      "learning_rate": 1e-05,
      "loss": 0.9069,
      "step": 73900
    },
    {
      "epoch": 65.46058458813108,
      "grad_norm": 0.2655339241027832,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 73905
    },
    {
      "epoch": 65.46501328609389,
      "grad_norm": 0.25286513566970825,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 73910
    },
    {
      "epoch": 65.46944198405669,
      "grad_norm": 0.22625558078289032,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 73915
    },
    {
      "epoch": 65.47387068201948,
      "grad_norm": 0.24174721539020538,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 73920
    },
    {
      "epoch": 65.47829937998229,
      "grad_norm": 0.2104085385799408,
      "learning_rate": 1e-05,
      "loss": 0.9286,
      "step": 73925
    },
    {
      "epoch": 65.48272807794508,
      "grad_norm": 0.21638138592243195,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 73930
    },
    {
      "epoch": 65.48715677590788,
      "grad_norm": 0.23180367052555084,
      "learning_rate": 1e-05,
      "loss": 0.9265,
      "step": 73935
    },
    {
      "epoch": 65.49158547387069,
      "grad_norm": 0.29226547479629517,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 73940
    },
    {
      "epoch": 65.49601417183348,
      "grad_norm": 0.2184530347585678,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 73945
    },
    {
      "epoch": 65.50044286979627,
      "grad_norm": 0.24443991482257843,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 73950
    },
    {
      "epoch": 65.50487156775908,
      "grad_norm": 0.31451648473739624,
      "learning_rate": 1e-05,
      "loss": 1.0302,
      "step": 73955
    },
    {
      "epoch": 65.50930026572188,
      "grad_norm": 0.2482128143310547,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 73960
    },
    {
      "epoch": 65.51372896368467,
      "grad_norm": 0.22831083834171295,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 73965
    },
    {
      "epoch": 65.51815766164748,
      "grad_norm": 0.27633482217788696,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 73970
    },
    {
      "epoch": 65.52258635961027,
      "grad_norm": 0.250397652387619,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 73975
    },
    {
      "epoch": 65.52701505757307,
      "grad_norm": 0.2273891717195511,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 73980
    },
    {
      "epoch": 65.53144375553588,
      "grad_norm": 0.24531710147857666,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 73985
    },
    {
      "epoch": 65.53587245349867,
      "grad_norm": 0.22834379971027374,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 73990
    },
    {
      "epoch": 65.54030115146146,
      "grad_norm": 0.28772053122520447,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 73995
    },
    {
      "epoch": 65.54472984942427,
      "grad_norm": 0.24808190762996674,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 74000
    },
    {
      "epoch": 65.54915854738707,
      "grad_norm": 0.24969829618930817,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 74005
    },
    {
      "epoch": 65.55358724534987,
      "grad_norm": 0.2515394985675812,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 74010
    },
    {
      "epoch": 65.55801594331267,
      "grad_norm": 0.22127506136894226,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 74015
    },
    {
      "epoch": 65.56244464127546,
      "grad_norm": 0.29235970973968506,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 74020
    },
    {
      "epoch": 65.56687333923827,
      "grad_norm": 0.21909882128238678,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 74025
    },
    {
      "epoch": 65.57130203720106,
      "grad_norm": 0.2424968034029007,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 74030
    },
    {
      "epoch": 65.57573073516386,
      "grad_norm": 0.2307959496974945,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 74035
    },
    {
      "epoch": 65.58015943312667,
      "grad_norm": 0.2197333723306656,
      "learning_rate": 1e-05,
      "loss": 0.9278,
      "step": 74040
    },
    {
      "epoch": 65.58458813108946,
      "grad_norm": 0.2280236929655075,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 74045
    },
    {
      "epoch": 65.58901682905226,
      "grad_norm": 0.3449203670024872,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 74050
    },
    {
      "epoch": 65.59344552701506,
      "grad_norm": 0.25828278064727783,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 74055
    },
    {
      "epoch": 65.59787422497786,
      "grad_norm": 0.29056501388549805,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 74060
    },
    {
      "epoch": 65.60230292294065,
      "grad_norm": 0.2404964417219162,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 74065
    },
    {
      "epoch": 65.60673162090346,
      "grad_norm": 0.23660553991794586,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 74070
    },
    {
      "epoch": 65.61116031886625,
      "grad_norm": 0.26986849308013916,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 74075
    },
    {
      "epoch": 65.61558901682905,
      "grad_norm": 0.27582237124443054,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 74080
    },
    {
      "epoch": 65.62001771479186,
      "grad_norm": 0.2464091181755066,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 74085
    },
    {
      "epoch": 65.62444641275465,
      "grad_norm": 0.21993358433246613,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 74090
    },
    {
      "epoch": 65.62887511071744,
      "grad_norm": 0.23220138251781464,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 74095
    },
    {
      "epoch": 65.63330380868025,
      "grad_norm": 0.1981947273015976,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 74100
    },
    {
      "epoch": 65.63773250664305,
      "grad_norm": 0.28225255012512207,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 74105
    },
    {
      "epoch": 65.64216120460584,
      "grad_norm": 0.2839025557041168,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 74110
    },
    {
      "epoch": 65.64658990256865,
      "grad_norm": 0.2281114161014557,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 74115
    },
    {
      "epoch": 65.65101860053144,
      "grad_norm": 0.2327207624912262,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 74120
    },
    {
      "epoch": 65.65544729849424,
      "grad_norm": 0.20485705137252808,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 74125
    },
    {
      "epoch": 65.65987599645705,
      "grad_norm": 0.21389590203762054,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 74130
    },
    {
      "epoch": 65.66430469441984,
      "grad_norm": 0.25099921226501465,
      "learning_rate": 1e-05,
      "loss": 0.9269,
      "step": 74135
    },
    {
      "epoch": 65.66873339238263,
      "grad_norm": 0.27750277519226074,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 74140
    },
    {
      "epoch": 65.67316209034544,
      "grad_norm": 0.25508829951286316,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 74145
    },
    {
      "epoch": 65.67759078830824,
      "grad_norm": 0.24631275236606598,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 74150
    },
    {
      "epoch": 65.68201948627103,
      "grad_norm": 0.23914340138435364,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 74155
    },
    {
      "epoch": 65.68644818423384,
      "grad_norm": 0.24078847467899323,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 74160
    },
    {
      "epoch": 65.69087688219663,
      "grad_norm": 0.22129502892494202,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 74165
    },
    {
      "epoch": 65.69530558015943,
      "grad_norm": 0.2491893470287323,
      "learning_rate": 1e-05,
      "loss": 0.9033,
      "step": 74170
    },
    {
      "epoch": 65.69973427812224,
      "grad_norm": 0.24785545468330383,
      "learning_rate": 1e-05,
      "loss": 0.9313,
      "step": 74175
    },
    {
      "epoch": 65.70416297608503,
      "grad_norm": 0.2541751265525818,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 74180
    },
    {
      "epoch": 65.70859167404782,
      "grad_norm": 0.22128938138484955,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 74185
    },
    {
      "epoch": 65.71302037201063,
      "grad_norm": 0.21123871207237244,
      "learning_rate": 1e-05,
      "loss": 0.9217,
      "step": 74190
    },
    {
      "epoch": 65.71744906997343,
      "grad_norm": 0.2514908015727997,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 74195
    },
    {
      "epoch": 65.72187776793622,
      "grad_norm": 0.26354873180389404,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 74200
    },
    {
      "epoch": 65.72630646589903,
      "grad_norm": 0.2506631016731262,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 74205
    },
    {
      "epoch": 65.73073516386182,
      "grad_norm": 0.2458077222108841,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 74210
    },
    {
      "epoch": 65.73516386182462,
      "grad_norm": 0.22851774096488953,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 74215
    },
    {
      "epoch": 65.73959255978743,
      "grad_norm": 0.25505608320236206,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 74220
    },
    {
      "epoch": 65.74402125775022,
      "grad_norm": 0.263139933347702,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 74225
    },
    {
      "epoch": 65.74844995571301,
      "grad_norm": 0.20865462720394135,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 74230
    },
    {
      "epoch": 65.75287865367582,
      "grad_norm": 0.21958093345165253,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 74235
    },
    {
      "epoch": 65.75730735163862,
      "grad_norm": 0.24629531800746918,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 74240
    },
    {
      "epoch": 65.76173604960141,
      "grad_norm": 0.25311487913131714,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 74245
    },
    {
      "epoch": 65.76616474756422,
      "grad_norm": 0.26450374722480774,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 74250
    },
    {
      "epoch": 65.77059344552701,
      "grad_norm": 0.273154616355896,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 74255
    },
    {
      "epoch": 65.77502214348982,
      "grad_norm": 0.2843685448169708,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 74260
    },
    {
      "epoch": 65.77945084145261,
      "grad_norm": 0.24004797637462616,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 74265
    },
    {
      "epoch": 65.78387953941541,
      "grad_norm": 0.21119549870491028,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 74270
    },
    {
      "epoch": 65.78830823737822,
      "grad_norm": 0.24425192177295685,
      "learning_rate": 1e-05,
      "loss": 1.0246,
      "step": 74275
    },
    {
      "epoch": 65.79273693534101,
      "grad_norm": 0.2254219800233841,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 74280
    },
    {
      "epoch": 65.7971656333038,
      "grad_norm": 0.22379523515701294,
      "learning_rate": 1e-05,
      "loss": 0.9324,
      "step": 74285
    },
    {
      "epoch": 65.80159433126661,
      "grad_norm": 0.23435257375240326,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 74290
    },
    {
      "epoch": 65.80602302922941,
      "grad_norm": 0.2412867397069931,
      "learning_rate": 1e-05,
      "loss": 0.8612,
      "step": 74295
    },
    {
      "epoch": 65.8104517271922,
      "grad_norm": 0.25451773405075073,
      "learning_rate": 1e-05,
      "loss": 1.0113,
      "step": 74300
    },
    {
      "epoch": 65.81488042515501,
      "grad_norm": 0.24115265905857086,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 74305
    },
    {
      "epoch": 65.8193091231178,
      "grad_norm": 0.21350601315498352,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 74310
    },
    {
      "epoch": 65.8237378210806,
      "grad_norm": 0.22150246798992157,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 74315
    },
    {
      "epoch": 65.8281665190434,
      "grad_norm": 0.22074013948440552,
      "learning_rate": 1e-05,
      "loss": 0.9157,
      "step": 74320
    },
    {
      "epoch": 65.8325952170062,
      "grad_norm": 0.30737680196762085,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 74325
    },
    {
      "epoch": 65.837023914969,
      "grad_norm": 0.23562325537204742,
      "learning_rate": 1e-05,
      "loss": 0.929,
      "step": 74330
    },
    {
      "epoch": 65.8414526129318,
      "grad_norm": 0.1977197527885437,
      "learning_rate": 1e-05,
      "loss": 1.0242,
      "step": 74335
    },
    {
      "epoch": 65.8458813108946,
      "grad_norm": 0.23536312580108643,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 74340
    },
    {
      "epoch": 65.85031000885739,
      "grad_norm": 0.2220110446214676,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 74345
    },
    {
      "epoch": 65.8547387068202,
      "grad_norm": 0.23775538802146912,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 74350
    },
    {
      "epoch": 65.859167404783,
      "grad_norm": 0.24388448894023895,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 74355
    },
    {
      "epoch": 65.86359610274579,
      "grad_norm": 0.23526248335838318,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 74360
    },
    {
      "epoch": 65.8680248007086,
      "grad_norm": 0.23477354645729065,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 74365
    },
    {
      "epoch": 65.87245349867139,
      "grad_norm": 0.26307356357574463,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 74370
    },
    {
      "epoch": 65.87688219663418,
      "grad_norm": 0.27177801728248596,
      "learning_rate": 1e-05,
      "loss": 0.9175,
      "step": 74375
    },
    {
      "epoch": 65.881310894597,
      "grad_norm": 0.23924562335014343,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 74380
    },
    {
      "epoch": 65.88573959255979,
      "grad_norm": 0.25265198945999146,
      "learning_rate": 1e-05,
      "loss": 0.9224,
      "step": 74385
    },
    {
      "epoch": 65.89016829052258,
      "grad_norm": 0.273100346326828,
      "learning_rate": 1e-05,
      "loss": 0.9261,
      "step": 74390
    },
    {
      "epoch": 65.89459698848539,
      "grad_norm": 0.2339247465133667,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 74395
    },
    {
      "epoch": 65.89902568644818,
      "grad_norm": 0.22921106219291687,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 74400
    },
    {
      "epoch": 65.90345438441098,
      "grad_norm": 0.22646678984165192,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 74405
    },
    {
      "epoch": 65.90788308237379,
      "grad_norm": 0.25305068492889404,
      "learning_rate": 1e-05,
      "loss": 0.9381,
      "step": 74410
    },
    {
      "epoch": 65.91231178033658,
      "grad_norm": 0.2483273148536682,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 74415
    },
    {
      "epoch": 65.91674047829937,
      "grad_norm": 0.22778117656707764,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 74420
    },
    {
      "epoch": 65.92116917626218,
      "grad_norm": 0.2253522127866745,
      "learning_rate": 1e-05,
      "loss": 1.0101,
      "step": 74425
    },
    {
      "epoch": 65.92559787422498,
      "grad_norm": 0.1965600550174713,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 74430
    },
    {
      "epoch": 65.93002657218777,
      "grad_norm": 0.2007874697446823,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 74435
    },
    {
      "epoch": 65.93445527015058,
      "grad_norm": 0.22544962167739868,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 74440
    },
    {
      "epoch": 65.93888396811337,
      "grad_norm": 0.2848958373069763,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 74445
    },
    {
      "epoch": 65.94331266607617,
      "grad_norm": 0.2699938714504242,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 74450
    },
    {
      "epoch": 65.94774136403898,
      "grad_norm": 0.2770007252693176,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 74455
    },
    {
      "epoch": 65.95217006200177,
      "grad_norm": 0.2443130612373352,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 74460
    },
    {
      "epoch": 65.95659875996456,
      "grad_norm": 0.24658843874931335,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 74465
    },
    {
      "epoch": 65.96102745792737,
      "grad_norm": 0.22709037363529205,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 74470
    },
    {
      "epoch": 65.96545615589017,
      "grad_norm": 0.2640805244445801,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 74475
    },
    {
      "epoch": 65.96988485385296,
      "grad_norm": 0.21873131394386292,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 74480
    },
    {
      "epoch": 65.97431355181577,
      "grad_norm": 0.21051909029483795,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 74485
    },
    {
      "epoch": 65.97874224977856,
      "grad_norm": 0.28755322098731995,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 74490
    },
    {
      "epoch": 65.98317094774137,
      "grad_norm": 0.2537829279899597,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 74495
    },
    {
      "epoch": 65.98759964570417,
      "grad_norm": 0.25761479139328003,
      "learning_rate": 1e-05,
      "loss": 0.9239,
      "step": 74500
    },
    {
      "epoch": 65.99202834366696,
      "grad_norm": 0.23359157145023346,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 74505
    },
    {
      "epoch": 65.99645704162977,
      "grad_norm": 0.2516773045063019,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 74510
    },
    {
      "epoch": 66.00088573959256,
      "grad_norm": 0.28383520245552063,
      "learning_rate": 1e-05,
      "loss": 0.9178,
      "step": 74515
    },
    {
      "epoch": 66.00531443755536,
      "grad_norm": 0.20141509175300598,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 74520
    },
    {
      "epoch": 66.00974313551816,
      "grad_norm": 0.23093950748443604,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 74525
    },
    {
      "epoch": 66.01417183348096,
      "grad_norm": 0.21200443804264069,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 74530
    },
    {
      "epoch": 66.01860053144375,
      "grad_norm": 0.2544075846672058,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 74535
    },
    {
      "epoch": 66.02302922940656,
      "grad_norm": 0.26948198676109314,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 74540
    },
    {
      "epoch": 66.02745792736935,
      "grad_norm": 0.2253427356481552,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 74545
    },
    {
      "epoch": 66.03188662533215,
      "grad_norm": 0.2516725957393646,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 74550
    },
    {
      "epoch": 66.03631532329496,
      "grad_norm": 0.25191354751586914,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 74555
    },
    {
      "epoch": 66.04074402125775,
      "grad_norm": 0.21623797714710236,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 74560
    },
    {
      "epoch": 66.04517271922055,
      "grad_norm": 0.28452759981155396,
      "learning_rate": 1e-05,
      "loss": 0.9098,
      "step": 74565
    },
    {
      "epoch": 66.04960141718335,
      "grad_norm": 0.23730280995368958,
      "learning_rate": 1e-05,
      "loss": 0.9247,
      "step": 74570
    },
    {
      "epoch": 66.05403011514615,
      "grad_norm": 0.2643267810344696,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 74575
    },
    {
      "epoch": 66.05845881310894,
      "grad_norm": 0.2191995084285736,
      "learning_rate": 1e-05,
      "loss": 0.9153,
      "step": 74580
    },
    {
      "epoch": 66.06288751107175,
      "grad_norm": 0.24467483162879944,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 74585
    },
    {
      "epoch": 66.06731620903454,
      "grad_norm": 0.25363361835479736,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 74590
    },
    {
      "epoch": 66.07174490699734,
      "grad_norm": 0.29141151905059814,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 74595
    },
    {
      "epoch": 66.07617360496015,
      "grad_norm": 0.2630162537097931,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 74600
    },
    {
      "epoch": 66.08060230292294,
      "grad_norm": 0.2733655869960785,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 74605
    },
    {
      "epoch": 66.08503100088573,
      "grad_norm": 0.2453538179397583,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 74610
    },
    {
      "epoch": 66.08945969884854,
      "grad_norm": 0.2916133999824524,
      "learning_rate": 1e-05,
      "loss": 0.9228,
      "step": 74615
    },
    {
      "epoch": 66.09388839681134,
      "grad_norm": 0.23237910866737366,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 74620
    },
    {
      "epoch": 66.09831709477413,
      "grad_norm": 0.2668355703353882,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 74625
    },
    {
      "epoch": 66.10274579273694,
      "grad_norm": 0.23668146133422852,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 74630
    },
    {
      "epoch": 66.10717449069973,
      "grad_norm": 0.24643099308013916,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 74635
    },
    {
      "epoch": 66.11160318866253,
      "grad_norm": 0.22631680965423584,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 74640
    },
    {
      "epoch": 66.11603188662534,
      "grad_norm": 0.23542432487010956,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 74645
    },
    {
      "epoch": 66.12046058458813,
      "grad_norm": 0.28409436345100403,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 74650
    },
    {
      "epoch": 66.12488928255092,
      "grad_norm": 0.23772531747817993,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 74655
    },
    {
      "epoch": 66.12931798051373,
      "grad_norm": 0.2425490915775299,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 74660
    },
    {
      "epoch": 66.13374667847653,
      "grad_norm": 0.26845383644104004,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 74665
    },
    {
      "epoch": 66.13817537643932,
      "grad_norm": 0.22238360345363617,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 74670
    },
    {
      "epoch": 66.14260407440213,
      "grad_norm": 0.2718292474746704,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 74675
    },
    {
      "epoch": 66.14703277236492,
      "grad_norm": 0.25698161125183105,
      "learning_rate": 1e-05,
      "loss": 0.9493,
      "step": 74680
    },
    {
      "epoch": 66.15146147032772,
      "grad_norm": 0.2645103931427002,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 74685
    },
    {
      "epoch": 66.15589016829053,
      "grad_norm": 0.2501746714115143,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 74690
    },
    {
      "epoch": 66.16031886625332,
      "grad_norm": 0.23271210491657257,
      "learning_rate": 1e-05,
      "loss": 0.9254,
      "step": 74695
    },
    {
      "epoch": 66.16474756421611,
      "grad_norm": 0.22491124272346497,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 74700
    },
    {
      "epoch": 66.16917626217892,
      "grad_norm": 0.2692507803440094,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 74705
    },
    {
      "epoch": 66.17360496014172,
      "grad_norm": 0.22798876464366913,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 74710
    },
    {
      "epoch": 66.17803365810451,
      "grad_norm": 0.2707555294036865,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 74715
    },
    {
      "epoch": 66.18246235606732,
      "grad_norm": 0.2852475345134735,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 74720
    },
    {
      "epoch": 66.18689105403011,
      "grad_norm": 0.24982333183288574,
      "learning_rate": 1e-05,
      "loss": 0.9301,
      "step": 74725
    },
    {
      "epoch": 66.1913197519929,
      "grad_norm": 0.23818860948085785,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 74730
    },
    {
      "epoch": 66.19574844995572,
      "grad_norm": 0.24792850017547607,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 74735
    },
    {
      "epoch": 66.20017714791851,
      "grad_norm": 0.26514294743537903,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 74740
    },
    {
      "epoch": 66.20460584588132,
      "grad_norm": 0.28952962160110474,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 74745
    },
    {
      "epoch": 66.20903454384411,
      "grad_norm": 0.2227969616651535,
      "learning_rate": 1e-05,
      "loss": 0.9413,
      "step": 74750
    },
    {
      "epoch": 66.2134632418069,
      "grad_norm": 0.21819466352462769,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 74755
    },
    {
      "epoch": 66.21789193976971,
      "grad_norm": 0.2691785991191864,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 74760
    },
    {
      "epoch": 66.22232063773251,
      "grad_norm": 0.25031018257141113,
      "learning_rate": 1e-05,
      "loss": 0.9207,
      "step": 74765
    },
    {
      "epoch": 66.2267493356953,
      "grad_norm": 0.21965476870536804,
      "learning_rate": 1e-05,
      "loss": 0.9139,
      "step": 74770
    },
    {
      "epoch": 66.23117803365811,
      "grad_norm": 0.22157379984855652,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 74775
    },
    {
      "epoch": 66.2356067316209,
      "grad_norm": 0.21819068491458893,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 74780
    },
    {
      "epoch": 66.2400354295837,
      "grad_norm": 0.24777469038963318,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 74785
    },
    {
      "epoch": 66.24446412754651,
      "grad_norm": 0.28822726011276245,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 74790
    },
    {
      "epoch": 66.2488928255093,
      "grad_norm": 0.2769969701766968,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 74795
    },
    {
      "epoch": 66.2533215234721,
      "grad_norm": 0.22472482919692993,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 74800
    },
    {
      "epoch": 66.2577502214349,
      "grad_norm": 0.2421802580356598,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 74805
    },
    {
      "epoch": 66.2621789193977,
      "grad_norm": 0.2716982364654541,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 74810
    },
    {
      "epoch": 66.26660761736049,
      "grad_norm": 0.2902645170688629,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 74815
    },
    {
      "epoch": 66.2710363153233,
      "grad_norm": 0.23928608000278473,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 74820
    },
    {
      "epoch": 66.2754650132861,
      "grad_norm": 0.24107012152671814,
      "learning_rate": 1e-05,
      "loss": 0.936,
      "step": 74825
    },
    {
      "epoch": 66.27989371124889,
      "grad_norm": 0.23326340317726135,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 74830
    },
    {
      "epoch": 66.2843224092117,
      "grad_norm": 0.22210094332695007,
      "learning_rate": 1e-05,
      "loss": 0.9364,
      "step": 74835
    },
    {
      "epoch": 66.28875110717449,
      "grad_norm": 0.2259248048067093,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 74840
    },
    {
      "epoch": 66.29317980513729,
      "grad_norm": 0.2693815529346466,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 74845
    },
    {
      "epoch": 66.2976085031001,
      "grad_norm": 0.2788546085357666,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 74850
    },
    {
      "epoch": 66.30203720106289,
      "grad_norm": 0.22876811027526855,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 74855
    },
    {
      "epoch": 66.30646589902568,
      "grad_norm": 0.24928165972232819,
      "learning_rate": 1e-05,
      "loss": 1.0002,
      "step": 74860
    },
    {
      "epoch": 66.31089459698849,
      "grad_norm": 0.2944789528846741,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 74865
    },
    {
      "epoch": 66.31532329495128,
      "grad_norm": 0.2194368839263916,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 74870
    },
    {
      "epoch": 66.31975199291408,
      "grad_norm": 0.2220890074968338,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 74875
    },
    {
      "epoch": 66.32418069087689,
      "grad_norm": 0.2729151248931885,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 74880
    },
    {
      "epoch": 66.32860938883968,
      "grad_norm": 0.22160911560058594,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 74885
    },
    {
      "epoch": 66.33303808680247,
      "grad_norm": 0.22314119338989258,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 74890
    },
    {
      "epoch": 66.33746678476528,
      "grad_norm": 0.22863273322582245,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 74895
    },
    {
      "epoch": 66.34189548272808,
      "grad_norm": 0.24421074986457825,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 74900
    },
    {
      "epoch": 66.34632418069087,
      "grad_norm": 0.22514158487319946,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 74905
    },
    {
      "epoch": 66.35075287865368,
      "grad_norm": 0.2802341878414154,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 74910
    },
    {
      "epoch": 66.35518157661647,
      "grad_norm": 0.24391338229179382,
      "learning_rate": 1e-05,
      "loss": 0.9087,
      "step": 74915
    },
    {
      "epoch": 66.35961027457927,
      "grad_norm": 0.2581331133842468,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 74920
    },
    {
      "epoch": 66.36403897254208,
      "grad_norm": 0.2371310293674469,
      "learning_rate": 1e-05,
      "loss": 0.9125,
      "step": 74925
    },
    {
      "epoch": 66.36846767050487,
      "grad_norm": 0.27201002836227417,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 74930
    },
    {
      "epoch": 66.37289636846766,
      "grad_norm": 0.2739558517932892,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 74935
    },
    {
      "epoch": 66.37732506643047,
      "grad_norm": 0.24453838169574738,
      "learning_rate": 1e-05,
      "loss": 1.0419,
      "step": 74940
    },
    {
      "epoch": 66.38175376439327,
      "grad_norm": 0.2607939839363098,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 74945
    },
    {
      "epoch": 66.38618246235606,
      "grad_norm": 0.22902780771255493,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 74950
    },
    {
      "epoch": 66.39061116031887,
      "grad_norm": 0.22381523251533508,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 74955
    },
    {
      "epoch": 66.39503985828166,
      "grad_norm": 0.2658435106277466,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 74960
    },
    {
      "epoch": 66.39946855624446,
      "grad_norm": 0.22766417264938354,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 74965
    },
    {
      "epoch": 66.40389725420727,
      "grad_norm": 0.21936286985874176,
      "learning_rate": 1e-05,
      "loss": 0.9357,
      "step": 74970
    },
    {
      "epoch": 66.40832595217006,
      "grad_norm": 0.24941982328891754,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 74975
    },
    {
      "epoch": 66.41275465013285,
      "grad_norm": 0.25433406233787537,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 74980
    },
    {
      "epoch": 66.41718334809566,
      "grad_norm": 0.23422107100486755,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 74985
    },
    {
      "epoch": 66.42161204605846,
      "grad_norm": 0.2508363425731659,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 74990
    },
    {
      "epoch": 66.42604074402126,
      "grad_norm": 0.2616310119628906,
      "learning_rate": 1e-05,
      "loss": 0.9079,
      "step": 74995
    },
    {
      "epoch": 66.43046944198406,
      "grad_norm": 0.2855851650238037,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 75000
    },
    {
      "epoch": 66.43489813994685,
      "grad_norm": 0.2324899435043335,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 75005
    },
    {
      "epoch": 66.43932683790966,
      "grad_norm": 0.2708640396595001,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 75010
    },
    {
      "epoch": 66.44375553587246,
      "grad_norm": 0.24579563736915588,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 75015
    },
    {
      "epoch": 66.44818423383525,
      "grad_norm": 0.23982743918895721,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 75020
    },
    {
      "epoch": 66.45261293179806,
      "grad_norm": 0.2698413133621216,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 75025
    },
    {
      "epoch": 66.45704162976085,
      "grad_norm": 0.2634207308292389,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 75030
    },
    {
      "epoch": 66.46147032772365,
      "grad_norm": 0.2523522973060608,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 75035
    },
    {
      "epoch": 66.46589902568645,
      "grad_norm": 0.20769092440605164,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 75040
    },
    {
      "epoch": 66.47032772364925,
      "grad_norm": 0.23730620741844177,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 75045
    },
    {
      "epoch": 66.47475642161204,
      "grad_norm": 0.2567179501056671,
      "learning_rate": 1e-05,
      "loss": 0.9293,
      "step": 75050
    },
    {
      "epoch": 66.47918511957485,
      "grad_norm": 0.23126822710037231,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 75055
    },
    {
      "epoch": 66.48361381753764,
      "grad_norm": 0.23821425437927246,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 75060
    },
    {
      "epoch": 66.48804251550044,
      "grad_norm": 0.25249359011650085,
      "learning_rate": 1e-05,
      "loss": 0.9027,
      "step": 75065
    },
    {
      "epoch": 66.49247121346325,
      "grad_norm": 0.28909602761268616,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 75070
    },
    {
      "epoch": 66.49689991142604,
      "grad_norm": 0.2202627956867218,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 75075
    },
    {
      "epoch": 66.50132860938884,
      "grad_norm": 0.24920712411403656,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 75080
    },
    {
      "epoch": 66.50575730735164,
      "grad_norm": 0.22108668088912964,
      "learning_rate": 1e-05,
      "loss": 1.0182,
      "step": 75085
    },
    {
      "epoch": 66.51018600531444,
      "grad_norm": 0.26375913619995117,
      "learning_rate": 1e-05,
      "loss": 0.9043,
      "step": 75090
    },
    {
      "epoch": 66.51461470327723,
      "grad_norm": 0.266618013381958,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 75095
    },
    {
      "epoch": 66.51904340124004,
      "grad_norm": 0.24958953261375427,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 75100
    },
    {
      "epoch": 66.52347209920283,
      "grad_norm": 0.21478323638439178,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 75105
    },
    {
      "epoch": 66.52790079716563,
      "grad_norm": 0.2612699568271637,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 75110
    },
    {
      "epoch": 66.53232949512844,
      "grad_norm": 0.20067405700683594,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 75115
    },
    {
      "epoch": 66.53675819309123,
      "grad_norm": 0.20420116186141968,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 75120
    },
    {
      "epoch": 66.54118689105402,
      "grad_norm": 0.2514948844909668,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 75125
    },
    {
      "epoch": 66.54561558901683,
      "grad_norm": 0.24784861505031586,
      "learning_rate": 1e-05,
      "loss": 0.9242,
      "step": 75130
    },
    {
      "epoch": 66.55004428697963,
      "grad_norm": 0.1898975521326065,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 75135
    },
    {
      "epoch": 66.55447298494242,
      "grad_norm": 0.2076498419046402,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 75140
    },
    {
      "epoch": 66.55890168290523,
      "grad_norm": 0.2726189196109772,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 75145
    },
    {
      "epoch": 66.56333038086802,
      "grad_norm": 0.2076188027858734,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 75150
    },
    {
      "epoch": 66.56775907883082,
      "grad_norm": 0.21906183660030365,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 75155
    },
    {
      "epoch": 66.57218777679363,
      "grad_norm": 0.20580622553825378,
      "learning_rate": 1e-05,
      "loss": 0.9095,
      "step": 75160
    },
    {
      "epoch": 66.57661647475642,
      "grad_norm": 0.23688721656799316,
      "learning_rate": 1e-05,
      "loss": 0.9392,
      "step": 75165
    },
    {
      "epoch": 66.58104517271921,
      "grad_norm": 0.24313907325267792,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 75170
    },
    {
      "epoch": 66.58547387068202,
      "grad_norm": 0.23236240446567535,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 75175
    },
    {
      "epoch": 66.58990256864482,
      "grad_norm": 0.26118573546409607,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 75180
    },
    {
      "epoch": 66.59433126660761,
      "grad_norm": 0.2630217373371124,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 75185
    },
    {
      "epoch": 66.59875996457042,
      "grad_norm": 0.2817946970462799,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 75190
    },
    {
      "epoch": 66.60318866253321,
      "grad_norm": 0.2811495363712311,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 75195
    },
    {
      "epoch": 66.60761736049601,
      "grad_norm": 0.319832444190979,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 75200
    },
    {
      "epoch": 66.61204605845882,
      "grad_norm": 0.23089855909347534,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 75205
    },
    {
      "epoch": 66.61647475642161,
      "grad_norm": 0.22274315357208252,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 75210
    },
    {
      "epoch": 66.6209034543844,
      "grad_norm": 0.24172724783420563,
      "learning_rate": 1e-05,
      "loss": 0.9241,
      "step": 75215
    },
    {
      "epoch": 66.62533215234721,
      "grad_norm": 0.22229234874248505,
      "learning_rate": 1e-05,
      "loss": 0.9148,
      "step": 75220
    },
    {
      "epoch": 66.62976085031,
      "grad_norm": 0.23254527151584625,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 75225
    },
    {
      "epoch": 66.63418954827281,
      "grad_norm": 0.22574812173843384,
      "learning_rate": 1e-05,
      "loss": 0.918,
      "step": 75230
    },
    {
      "epoch": 66.63861824623561,
      "grad_norm": 0.2288178652524948,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 75235
    },
    {
      "epoch": 66.6430469441984,
      "grad_norm": 0.241951584815979,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 75240
    },
    {
      "epoch": 66.64747564216121,
      "grad_norm": 0.2685646712779999,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 75245
    },
    {
      "epoch": 66.651904340124,
      "grad_norm": 0.2204943150281906,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 75250
    },
    {
      "epoch": 66.6563330380868,
      "grad_norm": 0.23781593143939972,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 75255
    },
    {
      "epoch": 66.66076173604961,
      "grad_norm": 0.2906077802181244,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 75260
    },
    {
      "epoch": 66.6651904340124,
      "grad_norm": 0.22208988666534424,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 75265
    },
    {
      "epoch": 66.6696191319752,
      "grad_norm": 0.27872180938720703,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 75270
    },
    {
      "epoch": 66.674047829938,
      "grad_norm": 0.20481392741203308,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 75275
    },
    {
      "epoch": 66.6784765279008,
      "grad_norm": 0.25637343525886536,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 75280
    },
    {
      "epoch": 66.68290522586359,
      "grad_norm": 0.25928816199302673,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 75285
    },
    {
      "epoch": 66.6873339238264,
      "grad_norm": 0.19369734823703766,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 75290
    },
    {
      "epoch": 66.6917626217892,
      "grad_norm": 0.22767478227615356,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 75295
    },
    {
      "epoch": 66.69619131975199,
      "grad_norm": 0.21035458147525787,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 75300
    },
    {
      "epoch": 66.7006200177148,
      "grad_norm": 0.22688359022140503,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 75305
    },
    {
      "epoch": 66.70504871567759,
      "grad_norm": 0.26498523354530334,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 75310
    },
    {
      "epoch": 66.70947741364039,
      "grad_norm": 0.23060598969459534,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 75315
    },
    {
      "epoch": 66.7139061116032,
      "grad_norm": 0.22982072830200195,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 75320
    },
    {
      "epoch": 66.71833480956599,
      "grad_norm": 0.2353755235671997,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 75325
    },
    {
      "epoch": 66.72276350752878,
      "grad_norm": 0.25840964913368225,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 75330
    },
    {
      "epoch": 66.72719220549159,
      "grad_norm": 0.235965758562088,
      "learning_rate": 1e-05,
      "loss": 0.9068,
      "step": 75335
    },
    {
      "epoch": 66.73162090345438,
      "grad_norm": 0.21959485113620758,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 75340
    },
    {
      "epoch": 66.73604960141718,
      "grad_norm": 0.2299426794052124,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 75345
    },
    {
      "epoch": 66.74047829937999,
      "grad_norm": 0.24419432878494263,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 75350
    },
    {
      "epoch": 66.74490699734278,
      "grad_norm": 0.24157685041427612,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 75355
    },
    {
      "epoch": 66.74933569530558,
      "grad_norm": 0.2289917767047882,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 75360
    },
    {
      "epoch": 66.75376439326838,
      "grad_norm": 0.2568418085575104,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 75365
    },
    {
      "epoch": 66.75819309123118,
      "grad_norm": 0.23512817919254303,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 75370
    },
    {
      "epoch": 66.76262178919397,
      "grad_norm": 0.24344559013843536,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 75375
    },
    {
      "epoch": 66.76705048715678,
      "grad_norm": 0.22590403258800507,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 75380
    },
    {
      "epoch": 66.77147918511957,
      "grad_norm": 0.2334146499633789,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 75385
    },
    {
      "epoch": 66.77590788308237,
      "grad_norm": 0.25481337308883667,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 75390
    },
    {
      "epoch": 66.78033658104518,
      "grad_norm": 0.2354278862476349,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 75395
    },
    {
      "epoch": 66.78476527900797,
      "grad_norm": 0.2461446076631546,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 75400
    },
    {
      "epoch": 66.78919397697076,
      "grad_norm": 0.25934159755706787,
      "learning_rate": 1e-05,
      "loss": 0.9347,
      "step": 75405
    },
    {
      "epoch": 66.79362267493357,
      "grad_norm": 0.2537505030632019,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 75410
    },
    {
      "epoch": 66.79805137289637,
      "grad_norm": 0.27161723375320435,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 75415
    },
    {
      "epoch": 66.80248007085916,
      "grad_norm": 0.21545828878879547,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 75420
    },
    {
      "epoch": 66.80690876882197,
      "grad_norm": 0.23820339143276215,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 75425
    },
    {
      "epoch": 66.81133746678476,
      "grad_norm": 0.3309570252895355,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 75430
    },
    {
      "epoch": 66.81576616474756,
      "grad_norm": 0.2800322473049164,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 75435
    },
    {
      "epoch": 66.82019486271037,
      "grad_norm": 0.30508455634117126,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 75440
    },
    {
      "epoch": 66.82462356067316,
      "grad_norm": 0.27522146701812744,
      "learning_rate": 1e-05,
      "loss": 1.0242,
      "step": 75445
    },
    {
      "epoch": 66.82905225863595,
      "grad_norm": 0.22513563930988312,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 75450
    },
    {
      "epoch": 66.83348095659876,
      "grad_norm": 0.22156572341918945,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 75455
    },
    {
      "epoch": 66.83790965456156,
      "grad_norm": 0.2650841772556305,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 75460
    },
    {
      "epoch": 66.84233835252435,
      "grad_norm": 0.24405457079410553,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 75465
    },
    {
      "epoch": 66.84676705048716,
      "grad_norm": 0.23092027008533478,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 75470
    },
    {
      "epoch": 66.85119574844995,
      "grad_norm": 0.23895317316055298,
      "learning_rate": 1e-05,
      "loss": 0.9254,
      "step": 75475
    },
    {
      "epoch": 66.85562444641276,
      "grad_norm": 0.24698390066623688,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 75480
    },
    {
      "epoch": 66.86005314437556,
      "grad_norm": 0.25817641615867615,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 75485
    },
    {
      "epoch": 66.86448184233835,
      "grad_norm": 0.2691112160682678,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 75490
    },
    {
      "epoch": 66.86891054030116,
      "grad_norm": 0.21688330173492432,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 75495
    },
    {
      "epoch": 66.87333923826395,
      "grad_norm": 0.2268580198287964,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 75500
    },
    {
      "epoch": 66.87776793622675,
      "grad_norm": 0.24221055209636688,
      "learning_rate": 1e-05,
      "loss": 0.926,
      "step": 75505
    },
    {
      "epoch": 66.88219663418955,
      "grad_norm": 0.23131303489208221,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 75510
    },
    {
      "epoch": 66.88662533215235,
      "grad_norm": 0.2452724277973175,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 75515
    },
    {
      "epoch": 66.89105403011514,
      "grad_norm": 0.23758156597614288,
      "learning_rate": 1e-05,
      "loss": 0.9077,
      "step": 75520
    },
    {
      "epoch": 66.89548272807795,
      "grad_norm": 0.21675114333629608,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 75525
    },
    {
      "epoch": 66.89991142604075,
      "grad_norm": 0.23800943791866302,
      "learning_rate": 1e-05,
      "loss": 0.9462,
      "step": 75530
    },
    {
      "epoch": 66.90434012400354,
      "grad_norm": 0.23256763815879822,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 75535
    },
    {
      "epoch": 66.90876882196635,
      "grad_norm": 0.26479703187942505,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 75540
    },
    {
      "epoch": 66.91319751992914,
      "grad_norm": 0.23046721518039703,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 75545
    },
    {
      "epoch": 66.91762621789194,
      "grad_norm": 0.22760146856307983,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 75550
    },
    {
      "epoch": 66.92205491585474,
      "grad_norm": 0.21838697791099548,
      "learning_rate": 1e-05,
      "loss": 0.8722,
      "step": 75555
    },
    {
      "epoch": 66.92648361381754,
      "grad_norm": 0.19916194677352905,
      "learning_rate": 1e-05,
      "loss": 0.8918,
      "step": 75560
    },
    {
      "epoch": 66.93091231178033,
      "grad_norm": 0.21423226594924927,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 75565
    },
    {
      "epoch": 66.93534100974314,
      "grad_norm": 0.21948851644992828,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 75570
    },
    {
      "epoch": 66.93976970770593,
      "grad_norm": 0.22058556973934174,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 75575
    },
    {
      "epoch": 66.94419840566873,
      "grad_norm": 0.3051481544971466,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 75580
    },
    {
      "epoch": 66.94862710363154,
      "grad_norm": 0.2431422770023346,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 75585
    },
    {
      "epoch": 66.95305580159433,
      "grad_norm": 0.24674725532531738,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 75590
    },
    {
      "epoch": 66.95748449955713,
      "grad_norm": 0.2501239776611328,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 75595
    },
    {
      "epoch": 66.96191319751993,
      "grad_norm": 0.21950183808803558,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 75600
    },
    {
      "epoch": 66.96634189548273,
      "grad_norm": 0.27658769488334656,
      "learning_rate": 1e-05,
      "loss": 0.9139,
      "step": 75605
    },
    {
      "epoch": 66.97077059344552,
      "grad_norm": 0.19335128366947174,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 75610
    },
    {
      "epoch": 66.97519929140833,
      "grad_norm": 0.23977304995059967,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 75615
    },
    {
      "epoch": 66.97962798937112,
      "grad_norm": 0.20509642362594604,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 75620
    },
    {
      "epoch": 66.98405668733392,
      "grad_norm": 0.2605193853378296,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 75625
    },
    {
      "epoch": 66.98848538529673,
      "grad_norm": 0.2361188679933548,
      "learning_rate": 1e-05,
      "loss": 0.9237,
      "step": 75630
    },
    {
      "epoch": 66.99291408325952,
      "grad_norm": 0.22738058865070343,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 75635
    },
    {
      "epoch": 66.99734278122232,
      "grad_norm": 0.2395820915699005,
      "learning_rate": 1e-05,
      "loss": 1.0295,
      "step": 75640
    },
    {
      "epoch": 67.00177147918512,
      "grad_norm": 0.22118353843688965,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 75645
    },
    {
      "epoch": 67.00620017714792,
      "grad_norm": 0.2724268138408661,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 75650
    },
    {
      "epoch": 67.01062887511071,
      "grad_norm": 0.2503476142883301,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 75655
    },
    {
      "epoch": 67.01505757307352,
      "grad_norm": 0.26182955503463745,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 75660
    },
    {
      "epoch": 67.01948627103631,
      "grad_norm": 0.22933301329612732,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 75665
    },
    {
      "epoch": 67.02391496899911,
      "grad_norm": 0.23889955878257751,
      "learning_rate": 1e-05,
      "loss": 0.9183,
      "step": 75670
    },
    {
      "epoch": 67.02834366696192,
      "grad_norm": 0.26819315552711487,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 75675
    },
    {
      "epoch": 67.03277236492471,
      "grad_norm": 0.26547348499298096,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 75680
    },
    {
      "epoch": 67.0372010628875,
      "grad_norm": 0.26852715015411377,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 75685
    },
    {
      "epoch": 67.04162976085031,
      "grad_norm": 0.24702468514442444,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 75690
    },
    {
      "epoch": 67.0460584588131,
      "grad_norm": 0.22735893726348877,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 75695
    },
    {
      "epoch": 67.0504871567759,
      "grad_norm": 0.2753344178199768,
      "learning_rate": 1e-05,
      "loss": 1.0473,
      "step": 75700
    },
    {
      "epoch": 67.05491585473871,
      "grad_norm": 0.25895124673843384,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 75705
    },
    {
      "epoch": 67.0593445527015,
      "grad_norm": 0.23908720910549164,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 75710
    },
    {
      "epoch": 67.0637732506643,
      "grad_norm": 0.2247406542301178,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 75715
    },
    {
      "epoch": 67.0682019486271,
      "grad_norm": 0.22783435881137848,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 75720
    },
    {
      "epoch": 67.0726306465899,
      "grad_norm": 0.26499298214912415,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 75725
    },
    {
      "epoch": 67.07705934455271,
      "grad_norm": 0.23504024744033813,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 75730
    },
    {
      "epoch": 67.0814880425155,
      "grad_norm": 0.25943422317504883,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 75735
    },
    {
      "epoch": 67.0859167404783,
      "grad_norm": 0.21212315559387207,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 75740
    },
    {
      "epoch": 67.0903454384411,
      "grad_norm": 0.243398517370224,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 75745
    },
    {
      "epoch": 67.0947741364039,
      "grad_norm": 0.22973883152008057,
      "learning_rate": 1e-05,
      "loss": 1.0122,
      "step": 75750
    },
    {
      "epoch": 67.0992028343667,
      "grad_norm": 0.25598523020744324,
      "learning_rate": 1e-05,
      "loss": 0.9221,
      "step": 75755
    },
    {
      "epoch": 67.1036315323295,
      "grad_norm": 0.21338117122650146,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 75760
    },
    {
      "epoch": 67.1080602302923,
      "grad_norm": 0.2169879674911499,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 75765
    },
    {
      "epoch": 67.11248892825509,
      "grad_norm": 0.26075443625450134,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 75770
    },
    {
      "epoch": 67.1169176262179,
      "grad_norm": 0.2342262864112854,
      "learning_rate": 1e-05,
      "loss": 0.9307,
      "step": 75775
    },
    {
      "epoch": 67.12134632418069,
      "grad_norm": 0.2636463940143585,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 75780
    },
    {
      "epoch": 67.12577502214349,
      "grad_norm": 0.24479275941848755,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 75785
    },
    {
      "epoch": 67.1302037201063,
      "grad_norm": 0.2589849829673767,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 75790
    },
    {
      "epoch": 67.13463241806909,
      "grad_norm": 0.22806964814662933,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 75795
    },
    {
      "epoch": 67.13906111603188,
      "grad_norm": 0.2187642753124237,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 75800
    },
    {
      "epoch": 67.14348981399469,
      "grad_norm": 0.31209951639175415,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 75805
    },
    {
      "epoch": 67.14791851195749,
      "grad_norm": 0.33480408787727356,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 75810
    },
    {
      "epoch": 67.15234720992028,
      "grad_norm": 0.28045713901519775,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 75815
    },
    {
      "epoch": 67.15677590788309,
      "grad_norm": 0.1930042952299118,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 75820
    },
    {
      "epoch": 67.16120460584588,
      "grad_norm": 0.22959069907665253,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 75825
    },
    {
      "epoch": 67.16563330380868,
      "grad_norm": 0.23556780815124512,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 75830
    },
    {
      "epoch": 67.17006200177148,
      "grad_norm": 0.232491135597229,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 75835
    },
    {
      "epoch": 67.17449069973428,
      "grad_norm": 0.2298961579799652,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 75840
    },
    {
      "epoch": 67.17891939769707,
      "grad_norm": 0.2644525170326233,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 75845
    },
    {
      "epoch": 67.18334809565988,
      "grad_norm": 0.2615306079387665,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 75850
    },
    {
      "epoch": 67.18777679362267,
      "grad_norm": 0.26570606231689453,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 75855
    },
    {
      "epoch": 67.19220549158547,
      "grad_norm": 0.25981390476226807,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 75860
    },
    {
      "epoch": 67.19663418954828,
      "grad_norm": 0.2342883199453354,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 75865
    },
    {
      "epoch": 67.20106288751107,
      "grad_norm": 0.25465264916419983,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 75870
    },
    {
      "epoch": 67.20549158547387,
      "grad_norm": 0.23989440500736237,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 75875
    },
    {
      "epoch": 67.20992028343667,
      "grad_norm": 0.2354646474123001,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 75880
    },
    {
      "epoch": 67.21434898139947,
      "grad_norm": 0.21341370046138763,
      "learning_rate": 1e-05,
      "loss": 1.023,
      "step": 75885
    },
    {
      "epoch": 67.21877767936226,
      "grad_norm": 0.27982768416404724,
      "learning_rate": 1e-05,
      "loss": 1.0196,
      "step": 75890
    },
    {
      "epoch": 67.22320637732507,
      "grad_norm": 0.2206200212240219,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 75895
    },
    {
      "epoch": 67.22763507528786,
      "grad_norm": 0.2396899163722992,
      "learning_rate": 1e-05,
      "loss": 1.0184,
      "step": 75900
    },
    {
      "epoch": 67.23206377325066,
      "grad_norm": 0.23123286664485931,
      "learning_rate": 1e-05,
      "loss": 0.93,
      "step": 75905
    },
    {
      "epoch": 67.23649247121347,
      "grad_norm": 0.288381963968277,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 75910
    },
    {
      "epoch": 67.24092116917626,
      "grad_norm": 0.29143276810646057,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 75915
    },
    {
      "epoch": 67.24534986713905,
      "grad_norm": 0.23502743244171143,
      "learning_rate": 1e-05,
      "loss": 1.025,
      "step": 75920
    },
    {
      "epoch": 67.24977856510186,
      "grad_norm": 0.24762378633022308,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 75925
    },
    {
      "epoch": 67.25420726306466,
      "grad_norm": 0.23010413348674774,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 75930
    },
    {
      "epoch": 67.25863596102745,
      "grad_norm": 0.281925767660141,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 75935
    },
    {
      "epoch": 67.26306465899026,
      "grad_norm": 0.26682430505752563,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 75940
    },
    {
      "epoch": 67.26749335695305,
      "grad_norm": 0.2477923333644867,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 75945
    },
    {
      "epoch": 67.27192205491585,
      "grad_norm": 0.21757370233535767,
      "learning_rate": 1e-05,
      "loss": 0.8989,
      "step": 75950
    },
    {
      "epoch": 67.27635075287866,
      "grad_norm": 0.26036861538887024,
      "learning_rate": 1e-05,
      "loss": 1.0361,
      "step": 75955
    },
    {
      "epoch": 67.28077945084145,
      "grad_norm": 0.22255021333694458,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 75960
    },
    {
      "epoch": 67.28520814880426,
      "grad_norm": 0.2614385187625885,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 75965
    },
    {
      "epoch": 67.28963684676705,
      "grad_norm": 0.2175007164478302,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 75970
    },
    {
      "epoch": 67.29406554472985,
      "grad_norm": 0.2614530324935913,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 75975
    },
    {
      "epoch": 67.29849424269266,
      "grad_norm": 0.23791691660881042,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 75980
    },
    {
      "epoch": 67.30292294065545,
      "grad_norm": 0.2422555536031723,
      "learning_rate": 1e-05,
      "loss": 1.0253,
      "step": 75985
    },
    {
      "epoch": 67.30735163861824,
      "grad_norm": 0.22131690382957458,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 75990
    },
    {
      "epoch": 67.31178033658105,
      "grad_norm": 0.24223895370960236,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 75995
    },
    {
      "epoch": 67.31620903454385,
      "grad_norm": 0.2355361133813858,
      "learning_rate": 1e-05,
      "loss": 0.9029,
      "step": 76000
    },
    {
      "epoch": 67.32063773250664,
      "grad_norm": 0.18825343251228333,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 76005
    },
    {
      "epoch": 67.32506643046945,
      "grad_norm": 0.21182124316692352,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 76010
    },
    {
      "epoch": 67.32949512843224,
      "grad_norm": 0.24794258177280426,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 76015
    },
    {
      "epoch": 67.33392382639504,
      "grad_norm": 0.24796223640441895,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 76020
    },
    {
      "epoch": 67.33835252435784,
      "grad_norm": 0.23214022815227509,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 76025
    },
    {
      "epoch": 67.34278122232064,
      "grad_norm": 0.22341832518577576,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 76030
    },
    {
      "epoch": 67.34720992028343,
      "grad_norm": 0.2317054718732834,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 76035
    },
    {
      "epoch": 67.35163861824624,
      "grad_norm": 0.27360010147094727,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 76040
    },
    {
      "epoch": 67.35606731620904,
      "grad_norm": 0.2764146029949188,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 76045
    },
    {
      "epoch": 67.36049601417183,
      "grad_norm": 0.2772946357727051,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 76050
    },
    {
      "epoch": 67.36492471213464,
      "grad_norm": 0.2732570469379425,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 76055
    },
    {
      "epoch": 67.36935341009743,
      "grad_norm": 0.2558162212371826,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 76060
    },
    {
      "epoch": 67.37378210806023,
      "grad_norm": 0.2680899202823639,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 76065
    },
    {
      "epoch": 67.37821080602303,
      "grad_norm": 0.2657979428768158,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 76070
    },
    {
      "epoch": 67.38263950398583,
      "grad_norm": 0.24710823595523834,
      "learning_rate": 1e-05,
      "loss": 0.9208,
      "step": 76075
    },
    {
      "epoch": 67.38706820194862,
      "grad_norm": 0.24437837302684784,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 76080
    },
    {
      "epoch": 67.39149689991143,
      "grad_norm": 0.26440921425819397,
      "learning_rate": 1e-05,
      "loss": 1.0412,
      "step": 76085
    },
    {
      "epoch": 67.39592559787422,
      "grad_norm": 0.2714998424053192,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 76090
    },
    {
      "epoch": 67.40035429583702,
      "grad_norm": 0.22913290560245514,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 76095
    },
    {
      "epoch": 67.40478299379983,
      "grad_norm": 0.20261049270629883,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 76100
    },
    {
      "epoch": 67.40921169176262,
      "grad_norm": 0.27024707198143005,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 76105
    },
    {
      "epoch": 67.41364038972542,
      "grad_norm": 0.2339015007019043,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 76110
    },
    {
      "epoch": 67.41806908768822,
      "grad_norm": 0.22890833020210266,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 76115
    },
    {
      "epoch": 67.42249778565102,
      "grad_norm": 0.2372104525566101,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 76120
    },
    {
      "epoch": 67.42692648361381,
      "grad_norm": 0.21062377095222473,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 76125
    },
    {
      "epoch": 67.43135518157662,
      "grad_norm": 0.22386609017848969,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 76130
    },
    {
      "epoch": 67.43578387953941,
      "grad_norm": 0.21555323898792267,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 76135
    },
    {
      "epoch": 67.44021257750221,
      "grad_norm": 0.23043711483478546,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 76140
    },
    {
      "epoch": 67.44464127546502,
      "grad_norm": 0.23496712744235992,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 76145
    },
    {
      "epoch": 67.44906997342781,
      "grad_norm": 0.21874217689037323,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 76150
    },
    {
      "epoch": 67.4534986713906,
      "grad_norm": 0.242909237742424,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 76155
    },
    {
      "epoch": 67.45792736935341,
      "grad_norm": 0.25705698132514954,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 76160
    },
    {
      "epoch": 67.46235606731621,
      "grad_norm": 0.24262379109859467,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 76165
    },
    {
      "epoch": 67.466784765279,
      "grad_norm": 0.2792976200580597,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 76170
    },
    {
      "epoch": 67.47121346324181,
      "grad_norm": 0.2542523741722107,
      "learning_rate": 1e-05,
      "loss": 0.9019,
      "step": 76175
    },
    {
      "epoch": 67.4756421612046,
      "grad_norm": 0.2473069280385971,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 76180
    },
    {
      "epoch": 67.4800708591674,
      "grad_norm": 0.2294023185968399,
      "learning_rate": 1e-05,
      "loss": 1.0494,
      "step": 76185
    },
    {
      "epoch": 67.4844995571302,
      "grad_norm": 0.2472751885652542,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 76190
    },
    {
      "epoch": 67.488928255093,
      "grad_norm": 0.24871501326560974,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 76195
    },
    {
      "epoch": 67.4933569530558,
      "grad_norm": 0.29277753829956055,
      "learning_rate": 1e-05,
      "loss": 1.0161,
      "step": 76200
    },
    {
      "epoch": 67.4977856510186,
      "grad_norm": 0.219540074467659,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 76205
    },
    {
      "epoch": 67.5022143489814,
      "grad_norm": 0.2593120336532593,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 76210
    },
    {
      "epoch": 67.5066430469442,
      "grad_norm": 0.25804784893989563,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 76215
    },
    {
      "epoch": 67.511071744907,
      "grad_norm": 0.26734939217567444,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 76220
    },
    {
      "epoch": 67.5155004428698,
      "grad_norm": 0.22027775645256042,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 76225
    },
    {
      "epoch": 67.5199291408326,
      "grad_norm": 0.21133993566036224,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 76230
    },
    {
      "epoch": 67.5243578387954,
      "grad_norm": 0.19820855557918549,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 76235
    },
    {
      "epoch": 67.52878653675819,
      "grad_norm": 0.2640177011489868,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 76240
    },
    {
      "epoch": 67.533215234721,
      "grad_norm": 0.2408740520477295,
      "learning_rate": 1e-05,
      "loss": 0.9157,
      "step": 76245
    },
    {
      "epoch": 67.53764393268379,
      "grad_norm": 0.25717005133628845,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 76250
    },
    {
      "epoch": 67.54207263064659,
      "grad_norm": 0.22637727856636047,
      "learning_rate": 1e-05,
      "loss": 0.926,
      "step": 76255
    },
    {
      "epoch": 67.5465013286094,
      "grad_norm": 0.24869249761104584,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 76260
    },
    {
      "epoch": 67.55093002657219,
      "grad_norm": 0.23160229623317719,
      "learning_rate": 1e-05,
      "loss": 0.9204,
      "step": 76265
    },
    {
      "epoch": 67.55535872453498,
      "grad_norm": 0.2552292048931122,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 76270
    },
    {
      "epoch": 67.55978742249779,
      "grad_norm": 0.286464661359787,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 76275
    },
    {
      "epoch": 67.56421612046059,
      "grad_norm": 0.3222089111804962,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 76280
    },
    {
      "epoch": 67.56864481842338,
      "grad_norm": 0.23707778751850128,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 76285
    },
    {
      "epoch": 67.57307351638619,
      "grad_norm": 0.25252413749694824,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 76290
    },
    {
      "epoch": 67.57750221434898,
      "grad_norm": 0.21557208895683289,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 76295
    },
    {
      "epoch": 67.58193091231178,
      "grad_norm": 0.26272785663604736,
      "learning_rate": 1e-05,
      "loss": 0.9066,
      "step": 76300
    },
    {
      "epoch": 67.58635961027458,
      "grad_norm": 0.2887038290500641,
      "learning_rate": 1e-05,
      "loss": 0.9003,
      "step": 76305
    },
    {
      "epoch": 67.59078830823738,
      "grad_norm": 0.24808894097805023,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 76310
    },
    {
      "epoch": 67.59521700620017,
      "grad_norm": 0.22453254461288452,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 76315
    },
    {
      "epoch": 67.59964570416298,
      "grad_norm": 0.2560519278049469,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 76320
    },
    {
      "epoch": 67.60407440212578,
      "grad_norm": 0.2581895887851715,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 76325
    },
    {
      "epoch": 67.60850310008857,
      "grad_norm": 0.25403323769569397,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 76330
    },
    {
      "epoch": 67.61293179805138,
      "grad_norm": 0.2543439567089081,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 76335
    },
    {
      "epoch": 67.61736049601417,
      "grad_norm": 0.2541045844554901,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 76340
    },
    {
      "epoch": 67.62178919397697,
      "grad_norm": 0.2141459584236145,
      "learning_rate": 1e-05,
      "loss": 0.9271,
      "step": 76345
    },
    {
      "epoch": 67.62621789193977,
      "grad_norm": 0.21260610222816467,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 76350
    },
    {
      "epoch": 67.63064658990257,
      "grad_norm": 0.24872179329395294,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 76355
    },
    {
      "epoch": 67.63507528786536,
      "grad_norm": 0.23809215426445007,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 76360
    },
    {
      "epoch": 67.63950398582817,
      "grad_norm": 0.2212904691696167,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 76365
    },
    {
      "epoch": 67.64393268379096,
      "grad_norm": 0.22988073527812958,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 76370
    },
    {
      "epoch": 67.64836138175376,
      "grad_norm": 0.2459574192762375,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 76375
    },
    {
      "epoch": 67.65279007971657,
      "grad_norm": 0.24165329337120056,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 76380
    },
    {
      "epoch": 67.65721877767936,
      "grad_norm": 0.25526028871536255,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 76385
    },
    {
      "epoch": 67.66164747564216,
      "grad_norm": 0.23187051713466644,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 76390
    },
    {
      "epoch": 67.66607617360496,
      "grad_norm": 0.25210532546043396,
      "learning_rate": 1e-05,
      "loss": 0.9335,
      "step": 76395
    },
    {
      "epoch": 67.67050487156776,
      "grad_norm": 0.22754041850566864,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 76400
    },
    {
      "epoch": 67.67493356953055,
      "grad_norm": 0.2311403751373291,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 76405
    },
    {
      "epoch": 67.67936226749336,
      "grad_norm": 0.21909789741039276,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 76410
    },
    {
      "epoch": 67.68379096545615,
      "grad_norm": 0.21831664443016052,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 76415
    },
    {
      "epoch": 67.68821966341895,
      "grad_norm": 0.2344667911529541,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 76420
    },
    {
      "epoch": 67.69264836138176,
      "grad_norm": 0.2096155881881714,
      "learning_rate": 1e-05,
      "loss": 0.9158,
      "step": 76425
    },
    {
      "epoch": 67.69707705934455,
      "grad_norm": 0.27816441655158997,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 76430
    },
    {
      "epoch": 67.70150575730734,
      "grad_norm": 0.2850865423679352,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 76435
    },
    {
      "epoch": 67.70593445527015,
      "grad_norm": 0.2764822840690613,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 76440
    },
    {
      "epoch": 67.71036315323295,
      "grad_norm": 0.22478429973125458,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 76445
    },
    {
      "epoch": 67.71479185119574,
      "grad_norm": 0.24560759961605072,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 76450
    },
    {
      "epoch": 67.71922054915855,
      "grad_norm": 0.24513061344623566,
      "learning_rate": 1e-05,
      "loss": 0.9381,
      "step": 76455
    },
    {
      "epoch": 67.72364924712134,
      "grad_norm": 0.2477315217256546,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 76460
    },
    {
      "epoch": 67.72807794508415,
      "grad_norm": 0.2127884030342102,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 76465
    },
    {
      "epoch": 67.73250664304695,
      "grad_norm": 0.21078099310398102,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 76470
    },
    {
      "epoch": 67.73693534100974,
      "grad_norm": 0.2255050539970398,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 76475
    },
    {
      "epoch": 67.74136403897255,
      "grad_norm": 0.2513887882232666,
      "learning_rate": 1e-05,
      "loss": 0.9186,
      "step": 76480
    },
    {
      "epoch": 67.74579273693534,
      "grad_norm": 0.24259239435195923,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 76485
    },
    {
      "epoch": 67.75022143489814,
      "grad_norm": 0.2723296284675598,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 76490
    },
    {
      "epoch": 67.75465013286095,
      "grad_norm": 0.2045954316854477,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 76495
    },
    {
      "epoch": 67.75907883082374,
      "grad_norm": 0.21716976165771484,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 76500
    },
    {
      "epoch": 67.76350752878653,
      "grad_norm": 0.24256424605846405,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 76505
    },
    {
      "epoch": 67.76793622674934,
      "grad_norm": 0.2862918972969055,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 76510
    },
    {
      "epoch": 67.77236492471214,
      "grad_norm": 0.2170039415359497,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 76515
    },
    {
      "epoch": 67.77679362267493,
      "grad_norm": 0.22306427359580994,
      "learning_rate": 1e-05,
      "loss": 0.9119,
      "step": 76520
    },
    {
      "epoch": 67.78122232063774,
      "grad_norm": 0.2219875603914261,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 76525
    },
    {
      "epoch": 67.78565101860053,
      "grad_norm": 0.23259595036506653,
      "learning_rate": 1e-05,
      "loss": 0.9213,
      "step": 76530
    },
    {
      "epoch": 67.79007971656333,
      "grad_norm": 0.21992826461791992,
      "learning_rate": 1e-05,
      "loss": 1.0191,
      "step": 76535
    },
    {
      "epoch": 67.79450841452613,
      "grad_norm": 0.3031964898109436,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 76540
    },
    {
      "epoch": 67.79893711248893,
      "grad_norm": 0.22539545595645905,
      "learning_rate": 1e-05,
      "loss": 0.9305,
      "step": 76545
    },
    {
      "epoch": 67.80336581045172,
      "grad_norm": 0.2568967044353485,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 76550
    },
    {
      "epoch": 67.80779450841453,
      "grad_norm": 0.21805106103420258,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 76555
    },
    {
      "epoch": 67.81222320637733,
      "grad_norm": 0.2425696849822998,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 76560
    },
    {
      "epoch": 67.81665190434012,
      "grad_norm": 0.233871191740036,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 76565
    },
    {
      "epoch": 67.82108060230293,
      "grad_norm": 0.2168416976928711,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 76570
    },
    {
      "epoch": 67.82550930026572,
      "grad_norm": 0.2798667550086975,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 76575
    },
    {
      "epoch": 67.82993799822852,
      "grad_norm": 0.2331400215625763,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 76580
    },
    {
      "epoch": 67.83436669619132,
      "grad_norm": 0.21549779176712036,
      "learning_rate": 1e-05,
      "loss": 0.9136,
      "step": 76585
    },
    {
      "epoch": 67.83879539415412,
      "grad_norm": 0.2231917679309845,
      "learning_rate": 1e-05,
      "loss": 1.0246,
      "step": 76590
    },
    {
      "epoch": 67.84322409211691,
      "grad_norm": 0.3170589506626129,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 76595
    },
    {
      "epoch": 67.84765279007972,
      "grad_norm": 0.2921904921531677,
      "learning_rate": 1e-05,
      "loss": 0.8998,
      "step": 76600
    },
    {
      "epoch": 67.85208148804251,
      "grad_norm": 0.2759345471858978,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 76605
    },
    {
      "epoch": 67.85651018600531,
      "grad_norm": 0.24915920197963715,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 76610
    },
    {
      "epoch": 67.86093888396812,
      "grad_norm": 0.2416379153728485,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 76615
    },
    {
      "epoch": 67.86536758193091,
      "grad_norm": 0.24495647847652435,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 76620
    },
    {
      "epoch": 67.8697962798937,
      "grad_norm": 0.2333097606897354,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 76625
    },
    {
      "epoch": 67.87422497785651,
      "grad_norm": 0.22763358056545258,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 76630
    },
    {
      "epoch": 67.87865367581931,
      "grad_norm": 0.209817036986351,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 76635
    },
    {
      "epoch": 67.8830823737821,
      "grad_norm": 0.24745959043502808,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 76640
    },
    {
      "epoch": 67.88751107174491,
      "grad_norm": 0.24498023092746735,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 76645
    },
    {
      "epoch": 67.8919397697077,
      "grad_norm": 0.23769810795783997,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 76650
    },
    {
      "epoch": 67.8963684676705,
      "grad_norm": 0.2656491696834564,
      "learning_rate": 1e-05,
      "loss": 0.9292,
      "step": 76655
    },
    {
      "epoch": 67.9007971656333,
      "grad_norm": 0.30382663011550903,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 76660
    },
    {
      "epoch": 67.9052258635961,
      "grad_norm": 0.263300359249115,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 76665
    },
    {
      "epoch": 67.9096545615589,
      "grad_norm": 0.25266414880752563,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 76670
    },
    {
      "epoch": 67.9140832595217,
      "grad_norm": 0.27038756012916565,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 76675
    },
    {
      "epoch": 67.9185119574845,
      "grad_norm": 0.219991996884346,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 76680
    },
    {
      "epoch": 67.92294065544729,
      "grad_norm": 0.2520654499530792,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 76685
    },
    {
      "epoch": 67.9273693534101,
      "grad_norm": 0.25472912192344666,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 76690
    },
    {
      "epoch": 67.9317980513729,
      "grad_norm": 0.21687844395637512,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 76695
    },
    {
      "epoch": 67.9362267493357,
      "grad_norm": 0.24386608600616455,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 76700
    },
    {
      "epoch": 67.9406554472985,
      "grad_norm": 0.29538071155548096,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 76705
    },
    {
      "epoch": 67.94508414526129,
      "grad_norm": 0.27116674184799194,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 76710
    },
    {
      "epoch": 67.9495128432241,
      "grad_norm": 0.24953193962574005,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 76715
    },
    {
      "epoch": 67.9539415411869,
      "grad_norm": 0.27773234248161316,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 76720
    },
    {
      "epoch": 67.95837023914969,
      "grad_norm": 0.23349931836128235,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 76725
    },
    {
      "epoch": 67.9627989371125,
      "grad_norm": 0.2621541917324066,
      "learning_rate": 1e-05,
      "loss": 0.9153,
      "step": 76730
    },
    {
      "epoch": 67.96722763507529,
      "grad_norm": 0.2179231345653534,
      "learning_rate": 1e-05,
      "loss": 0.9342,
      "step": 76735
    },
    {
      "epoch": 67.97165633303808,
      "grad_norm": 0.24042630195617676,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 76740
    },
    {
      "epoch": 67.97608503100089,
      "grad_norm": 0.22015798091888428,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 76745
    },
    {
      "epoch": 67.98051372896369,
      "grad_norm": 0.24946820735931396,
      "learning_rate": 1e-05,
      "loss": 0.9247,
      "step": 76750
    },
    {
      "epoch": 67.98494242692648,
      "grad_norm": 0.22434455156326294,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 76755
    },
    {
      "epoch": 67.98937112488929,
      "grad_norm": 0.2252250462770462,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 76760
    },
    {
      "epoch": 67.99379982285208,
      "grad_norm": 0.21331030130386353,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 76765
    },
    {
      "epoch": 67.99822852081488,
      "grad_norm": 0.25047460198402405,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 76770
    },
    {
      "epoch": 68.00265721877768,
      "grad_norm": 0.25584807991981506,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 76775
    },
    {
      "epoch": 68.00708591674048,
      "grad_norm": 0.23728016018867493,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 76780
    },
    {
      "epoch": 68.01151461470327,
      "grad_norm": 0.22622765600681305,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 76785
    },
    {
      "epoch": 68.01594331266608,
      "grad_norm": 0.2121357023715973,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 76790
    },
    {
      "epoch": 68.02037201062888,
      "grad_norm": 0.25021448731422424,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 76795
    },
    {
      "epoch": 68.02480070859167,
      "grad_norm": 0.22793389856815338,
      "learning_rate": 1e-05,
      "loss": 0.8953,
      "step": 76800
    },
    {
      "epoch": 68.02922940655448,
      "grad_norm": 0.23099902272224426,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 76805
    },
    {
      "epoch": 68.03365810451727,
      "grad_norm": 0.27883380651474,
      "learning_rate": 1e-05,
      "loss": 0.9198,
      "step": 76810
    },
    {
      "epoch": 68.03808680248007,
      "grad_norm": 0.2255604863166809,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 76815
    },
    {
      "epoch": 68.04251550044287,
      "grad_norm": 0.22914519906044006,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 76820
    },
    {
      "epoch": 68.04694419840567,
      "grad_norm": 0.21722693741321564,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 76825
    },
    {
      "epoch": 68.05137289636846,
      "grad_norm": 0.22584277391433716,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 76830
    },
    {
      "epoch": 68.05580159433127,
      "grad_norm": 0.2503102123737335,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 76835
    },
    {
      "epoch": 68.06023029229407,
      "grad_norm": 0.30725279450416565,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 76840
    },
    {
      "epoch": 68.06465899025686,
      "grad_norm": 0.24035684764385223,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 76845
    },
    {
      "epoch": 68.06908768821967,
      "grad_norm": 0.3344380557537079,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 76850
    },
    {
      "epoch": 68.07351638618246,
      "grad_norm": 0.2741321325302124,
      "learning_rate": 1e-05,
      "loss": 0.8868,
      "step": 76855
    },
    {
      "epoch": 68.07794508414526,
      "grad_norm": 0.2542209327220917,
      "learning_rate": 1e-05,
      "loss": 0.8985,
      "step": 76860
    },
    {
      "epoch": 68.08237378210806,
      "grad_norm": 0.21474997699260712,
      "learning_rate": 1e-05,
      "loss": 1.025,
      "step": 76865
    },
    {
      "epoch": 68.08680248007086,
      "grad_norm": 0.2442864626646042,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 76870
    },
    {
      "epoch": 68.09123117803365,
      "grad_norm": 0.23459120094776154,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 76875
    },
    {
      "epoch": 68.09565987599646,
      "grad_norm": 0.24858005344867706,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 76880
    },
    {
      "epoch": 68.10008857395925,
      "grad_norm": 0.2764456570148468,
      "learning_rate": 1e-05,
      "loss": 0.8921,
      "step": 76885
    },
    {
      "epoch": 68.10451727192205,
      "grad_norm": 0.2397177517414093,
      "learning_rate": 1e-05,
      "loss": 0.9011,
      "step": 76890
    },
    {
      "epoch": 68.10894596988486,
      "grad_norm": 0.19807660579681396,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 76895
    },
    {
      "epoch": 68.11337466784765,
      "grad_norm": 0.2176160216331482,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 76900
    },
    {
      "epoch": 68.11780336581045,
      "grad_norm": 0.19930031895637512,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 76905
    },
    {
      "epoch": 68.12223206377325,
      "grad_norm": 0.24108199775218964,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 76910
    },
    {
      "epoch": 68.12666076173605,
      "grad_norm": 0.26400989294052124,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 76915
    },
    {
      "epoch": 68.13108945969884,
      "grad_norm": 0.2549172341823578,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 76920
    },
    {
      "epoch": 68.13551815766165,
      "grad_norm": 0.2952554523944855,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 76925
    },
    {
      "epoch": 68.13994685562444,
      "grad_norm": 0.2596129775047302,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 76930
    },
    {
      "epoch": 68.14437555358724,
      "grad_norm": 0.23718851804733276,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 76935
    },
    {
      "epoch": 68.14880425155005,
      "grad_norm": 0.25222161412239075,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 76940
    },
    {
      "epoch": 68.15323294951284,
      "grad_norm": 0.24107126891613007,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 76945
    },
    {
      "epoch": 68.15766164747565,
      "grad_norm": 0.23306317627429962,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 76950
    },
    {
      "epoch": 68.16209034543844,
      "grad_norm": 0.21922940015792847,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 76955
    },
    {
      "epoch": 68.16651904340124,
      "grad_norm": 0.2458428144454956,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 76960
    },
    {
      "epoch": 68.17094774136405,
      "grad_norm": 0.23581838607788086,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 76965
    },
    {
      "epoch": 68.17537643932684,
      "grad_norm": 0.24623973667621613,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 76970
    },
    {
      "epoch": 68.17980513728963,
      "grad_norm": 0.2370588779449463,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 76975
    },
    {
      "epoch": 68.18423383525244,
      "grad_norm": 0.2924964725971222,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 76980
    },
    {
      "epoch": 68.18866253321524,
      "grad_norm": 0.23916597664356232,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 76985
    },
    {
      "epoch": 68.19309123117803,
      "grad_norm": 0.27028536796569824,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 76990
    },
    {
      "epoch": 68.19751992914084,
      "grad_norm": 0.27892959117889404,
      "learning_rate": 1e-05,
      "loss": 0.9153,
      "step": 76995
    },
    {
      "epoch": 68.20194862710363,
      "grad_norm": 0.2842993140220642,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 77000
    },
    {
      "epoch": 68.20637732506643,
      "grad_norm": 0.31390640139579773,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 77005
    },
    {
      "epoch": 68.21080602302924,
      "grad_norm": 0.2571030855178833,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 77010
    },
    {
      "epoch": 68.21523472099203,
      "grad_norm": 0.22763369977474213,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 77015
    },
    {
      "epoch": 68.21966341895482,
      "grad_norm": 0.20157192647457123,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 77020
    },
    {
      "epoch": 68.22409211691763,
      "grad_norm": 0.23055174946784973,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 77025
    },
    {
      "epoch": 68.22852081488043,
      "grad_norm": 0.22462643682956696,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 77030
    },
    {
      "epoch": 68.23294951284322,
      "grad_norm": 0.2218571901321411,
      "learning_rate": 1e-05,
      "loss": 0.909,
      "step": 77035
    },
    {
      "epoch": 68.23737821080603,
      "grad_norm": 0.23939426243305206,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 77040
    },
    {
      "epoch": 68.24180690876882,
      "grad_norm": 0.2975097596645355,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 77045
    },
    {
      "epoch": 68.24623560673162,
      "grad_norm": 0.2307479828596115,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 77050
    },
    {
      "epoch": 68.25066430469442,
      "grad_norm": 0.25362545251846313,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 77055
    },
    {
      "epoch": 68.25509300265722,
      "grad_norm": 0.2312287986278534,
      "learning_rate": 1e-05,
      "loss": 0.9144,
      "step": 77060
    },
    {
      "epoch": 68.25952170062001,
      "grad_norm": 0.21733015775680542,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 77065
    },
    {
      "epoch": 68.26395039858282,
      "grad_norm": 0.2657482326030731,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 77070
    },
    {
      "epoch": 68.26837909654562,
      "grad_norm": 0.24333474040031433,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 77075
    },
    {
      "epoch": 68.27280779450841,
      "grad_norm": 0.2510969042778015,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 77080
    },
    {
      "epoch": 68.27723649247122,
      "grad_norm": 0.2663164436817169,
      "learning_rate": 1e-05,
      "loss": 0.9038,
      "step": 77085
    },
    {
      "epoch": 68.28166519043401,
      "grad_norm": 0.21536394953727722,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 77090
    },
    {
      "epoch": 68.2860938883968,
      "grad_norm": 0.2670002281665802,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 77095
    },
    {
      "epoch": 68.29052258635961,
      "grad_norm": 0.23812291026115417,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 77100
    },
    {
      "epoch": 68.29495128432241,
      "grad_norm": 0.2553909420967102,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 77105
    },
    {
      "epoch": 68.2993799822852,
      "grad_norm": 0.26331189274787903,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 77110
    },
    {
      "epoch": 68.30380868024801,
      "grad_norm": 0.21576452255249023,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 77115
    },
    {
      "epoch": 68.3082373782108,
      "grad_norm": 0.20230774581432343,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 77120
    },
    {
      "epoch": 68.3126660761736,
      "grad_norm": 0.23358643054962158,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 77125
    },
    {
      "epoch": 68.31709477413641,
      "grad_norm": 0.2323332577943802,
      "learning_rate": 1e-05,
      "loss": 0.9272,
      "step": 77130
    },
    {
      "epoch": 68.3215234720992,
      "grad_norm": 0.22187022864818573,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 77135
    },
    {
      "epoch": 68.325952170062,
      "grad_norm": 0.2275269478559494,
      "learning_rate": 1e-05,
      "loss": 1.0549,
      "step": 77140
    },
    {
      "epoch": 68.3303808680248,
      "grad_norm": 0.24728339910507202,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 77145
    },
    {
      "epoch": 68.3348095659876,
      "grad_norm": 0.24441349506378174,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 77150
    },
    {
      "epoch": 68.33923826395039,
      "grad_norm": 0.26552215218544006,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 77155
    },
    {
      "epoch": 68.3436669619132,
      "grad_norm": 0.20027394592761993,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 77160
    },
    {
      "epoch": 68.348095659876,
      "grad_norm": 0.2559523582458496,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 77165
    },
    {
      "epoch": 68.35252435783879,
      "grad_norm": 0.2647932171821594,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 77170
    },
    {
      "epoch": 68.3569530558016,
      "grad_norm": 0.30525845289230347,
      "learning_rate": 1e-05,
      "loss": 0.9191,
      "step": 77175
    },
    {
      "epoch": 68.36138175376439,
      "grad_norm": 0.22653600573539734,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 77180
    },
    {
      "epoch": 68.36581045172719,
      "grad_norm": 0.22548870742321014,
      "learning_rate": 1e-05,
      "loss": 1.0263,
      "step": 77185
    },
    {
      "epoch": 68.37023914969,
      "grad_norm": 0.21463324129581451,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 77190
    },
    {
      "epoch": 68.37466784765279,
      "grad_norm": 0.21944944560527802,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 77195
    },
    {
      "epoch": 68.3790965456156,
      "grad_norm": 0.23399657011032104,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 77200
    },
    {
      "epoch": 68.38352524357839,
      "grad_norm": 0.2551707327365875,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 77205
    },
    {
      "epoch": 68.38795394154118,
      "grad_norm": 0.3327534794807434,
      "learning_rate": 1e-05,
      "loss": 0.925,
      "step": 77210
    },
    {
      "epoch": 68.39238263950399,
      "grad_norm": 0.26631003618240356,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 77215
    },
    {
      "epoch": 68.39681133746679,
      "grad_norm": 0.23770877718925476,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 77220
    },
    {
      "epoch": 68.40124003542958,
      "grad_norm": 0.21755321323871613,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 77225
    },
    {
      "epoch": 68.40566873339239,
      "grad_norm": 0.2293049693107605,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 77230
    },
    {
      "epoch": 68.41009743135518,
      "grad_norm": 0.23316511511802673,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 77235
    },
    {
      "epoch": 68.41452612931798,
      "grad_norm": 0.25545260310173035,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 77240
    },
    {
      "epoch": 68.41895482728079,
      "grad_norm": 0.25524574518203735,
      "learning_rate": 1e-05,
      "loss": 0.9493,
      "step": 77245
    },
    {
      "epoch": 68.42338352524358,
      "grad_norm": 0.22837354242801666,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 77250
    },
    {
      "epoch": 68.42781222320637,
      "grad_norm": 0.23328398168087006,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 77255
    },
    {
      "epoch": 68.43224092116918,
      "grad_norm": 0.26440852880477905,
      "learning_rate": 1e-05,
      "loss": 0.9214,
      "step": 77260
    },
    {
      "epoch": 68.43666961913198,
      "grad_norm": 0.23230767250061035,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 77265
    },
    {
      "epoch": 68.44109831709477,
      "grad_norm": 0.21214210987091064,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 77270
    },
    {
      "epoch": 68.44552701505758,
      "grad_norm": 0.2702535092830658,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 77275
    },
    {
      "epoch": 68.44995571302037,
      "grad_norm": 0.1878858357667923,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 77280
    },
    {
      "epoch": 68.45438441098317,
      "grad_norm": 0.23052692413330078,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 77285
    },
    {
      "epoch": 68.45881310894598,
      "grad_norm": 0.22295252978801727,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 77290
    },
    {
      "epoch": 68.46324180690877,
      "grad_norm": 0.27759110927581787,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 77295
    },
    {
      "epoch": 68.46767050487156,
      "grad_norm": 0.2125776708126068,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 77300
    },
    {
      "epoch": 68.47209920283437,
      "grad_norm": 0.21196389198303223,
      "learning_rate": 1e-05,
      "loss": 0.9342,
      "step": 77305
    },
    {
      "epoch": 68.47652790079717,
      "grad_norm": 0.20559704303741455,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 77310
    },
    {
      "epoch": 68.48095659875996,
      "grad_norm": 0.21400666236877441,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 77315
    },
    {
      "epoch": 68.48538529672277,
      "grad_norm": 0.20403701066970825,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 77320
    },
    {
      "epoch": 68.48981399468556,
      "grad_norm": 0.2335355132818222,
      "learning_rate": 1e-05,
      "loss": 1.0186,
      "step": 77325
    },
    {
      "epoch": 68.49424269264836,
      "grad_norm": 0.2715393304824829,
      "learning_rate": 1e-05,
      "loss": 0.9235,
      "step": 77330
    },
    {
      "epoch": 68.49867139061116,
      "grad_norm": 0.22402842342853546,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 77335
    },
    {
      "epoch": 68.50310008857396,
      "grad_norm": 0.25089943408966064,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 77340
    },
    {
      "epoch": 68.50752878653675,
      "grad_norm": 0.2760465145111084,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 77345
    },
    {
      "epoch": 68.51195748449956,
      "grad_norm": 0.24571135640144348,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 77350
    },
    {
      "epoch": 68.51638618246236,
      "grad_norm": 0.24833685159683228,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 77355
    },
    {
      "epoch": 68.52081488042515,
      "grad_norm": 0.27041807770729065,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 77360
    },
    {
      "epoch": 68.52524357838796,
      "grad_norm": 0.243954598903656,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 77365
    },
    {
      "epoch": 68.52967227635075,
      "grad_norm": 0.21405239403247833,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 77370
    },
    {
      "epoch": 68.53410097431355,
      "grad_norm": 0.30342036485671997,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 77375
    },
    {
      "epoch": 68.53852967227635,
      "grad_norm": 0.35024258494377136,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 77380
    },
    {
      "epoch": 68.54295837023915,
      "grad_norm": 0.21902072429656982,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 77385
    },
    {
      "epoch": 68.54738706820194,
      "grad_norm": 0.2056579887866974,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 77390
    },
    {
      "epoch": 68.55181576616475,
      "grad_norm": 0.2437870055437088,
      "learning_rate": 1e-05,
      "loss": 0.9493,
      "step": 77395
    },
    {
      "epoch": 68.55624446412754,
      "grad_norm": 0.2283148467540741,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 77400
    },
    {
      "epoch": 68.56067316209034,
      "grad_norm": 0.2600709795951843,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 77405
    },
    {
      "epoch": 68.56510186005315,
      "grad_norm": 0.24737876653671265,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 77410
    },
    {
      "epoch": 68.56953055801594,
      "grad_norm": 0.2758330702781677,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 77415
    },
    {
      "epoch": 68.57395925597874,
      "grad_norm": 0.210517019033432,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 77420
    },
    {
      "epoch": 68.57838795394154,
      "grad_norm": 0.23354770243167877,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 77425
    },
    {
      "epoch": 68.58281665190434,
      "grad_norm": 0.27804309129714966,
      "learning_rate": 1e-05,
      "loss": 0.9231,
      "step": 77430
    },
    {
      "epoch": 68.58724534986715,
      "grad_norm": 0.28320494294166565,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 77435
    },
    {
      "epoch": 68.59167404782994,
      "grad_norm": 0.3446701467037201,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 77440
    },
    {
      "epoch": 68.59610274579273,
      "grad_norm": 0.20666784048080444,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 77445
    },
    {
      "epoch": 68.60053144375554,
      "grad_norm": 0.23360778391361237,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 77450
    },
    {
      "epoch": 68.60496014171834,
      "grad_norm": 0.2181546837091446,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 77455
    },
    {
      "epoch": 68.60938883968113,
      "grad_norm": 0.24975289404392242,
      "learning_rate": 1e-05,
      "loss": 1.0397,
      "step": 77460
    },
    {
      "epoch": 68.61381753764394,
      "grad_norm": 0.2663228511810303,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 77465
    },
    {
      "epoch": 68.61824623560673,
      "grad_norm": 0.22524394094944,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 77470
    },
    {
      "epoch": 68.62267493356953,
      "grad_norm": 0.22815193235874176,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 77475
    },
    {
      "epoch": 68.62710363153234,
      "grad_norm": 0.22721263766288757,
      "learning_rate": 1e-05,
      "loss": 1.0396,
      "step": 77480
    },
    {
      "epoch": 68.63153232949513,
      "grad_norm": 0.23396463692188263,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 77485
    },
    {
      "epoch": 68.63596102745792,
      "grad_norm": 0.2337217479944229,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 77490
    },
    {
      "epoch": 68.64038972542073,
      "grad_norm": 0.20511044561862946,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 77495
    },
    {
      "epoch": 68.64481842338353,
      "grad_norm": 0.21474257111549377,
      "learning_rate": 1e-05,
      "loss": 0.9221,
      "step": 77500
    },
    {
      "epoch": 68.64924712134632,
      "grad_norm": 0.23007886111736298,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 77505
    },
    {
      "epoch": 68.65367581930913,
      "grad_norm": 0.24416375160217285,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 77510
    },
    {
      "epoch": 68.65810451727192,
      "grad_norm": 0.2489532232284546,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 77515
    },
    {
      "epoch": 68.66253321523472,
      "grad_norm": 0.23852939903736115,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 77520
    },
    {
      "epoch": 68.66696191319753,
      "grad_norm": 0.25823014974594116,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 77525
    },
    {
      "epoch": 68.67139061116032,
      "grad_norm": 0.21028785407543182,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 77530
    },
    {
      "epoch": 68.67581930912311,
      "grad_norm": 0.2341497391462326,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 77535
    },
    {
      "epoch": 68.68024800708592,
      "grad_norm": 0.19045452773571014,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 77540
    },
    {
      "epoch": 68.68467670504872,
      "grad_norm": 0.25875353813171387,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 77545
    },
    {
      "epoch": 68.68910540301151,
      "grad_norm": 0.19904853403568268,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 77550
    },
    {
      "epoch": 68.69353410097432,
      "grad_norm": 0.22108907997608185,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 77555
    },
    {
      "epoch": 68.69796279893711,
      "grad_norm": 0.2110719531774521,
      "learning_rate": 1e-05,
      "loss": 0.9161,
      "step": 77560
    },
    {
      "epoch": 68.7023914968999,
      "grad_norm": 0.2077806442975998,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 77565
    },
    {
      "epoch": 68.70682019486271,
      "grad_norm": 0.23411232233047485,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 77570
    },
    {
      "epoch": 68.71124889282551,
      "grad_norm": 0.23621974885463715,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 77575
    },
    {
      "epoch": 68.7156775907883,
      "grad_norm": 0.26220467686653137,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 77580
    },
    {
      "epoch": 68.72010628875111,
      "grad_norm": 0.19367679953575134,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 77585
    },
    {
      "epoch": 68.7245349867139,
      "grad_norm": 0.2192482054233551,
      "learning_rate": 1e-05,
      "loss": 0.9102,
      "step": 77590
    },
    {
      "epoch": 68.7289636846767,
      "grad_norm": 0.21327932178974152,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 77595
    },
    {
      "epoch": 68.73339238263951,
      "grad_norm": 0.2203483283519745,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 77600
    },
    {
      "epoch": 68.7378210806023,
      "grad_norm": 0.23164600133895874,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 77605
    },
    {
      "epoch": 68.7422497785651,
      "grad_norm": 0.2410828173160553,
      "learning_rate": 1e-05,
      "loss": 0.9095,
      "step": 77610
    },
    {
      "epoch": 68.7466784765279,
      "grad_norm": 0.22069929540157318,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 77615
    },
    {
      "epoch": 68.7511071744907,
      "grad_norm": 0.1951211541891098,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 77620
    },
    {
      "epoch": 68.75553587245349,
      "grad_norm": 0.22460252046585083,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 77625
    },
    {
      "epoch": 68.7599645704163,
      "grad_norm": 0.30915582180023193,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 77630
    },
    {
      "epoch": 68.7643932683791,
      "grad_norm": 0.2821124494075775,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 77635
    },
    {
      "epoch": 68.76882196634189,
      "grad_norm": 0.31025904417037964,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 77640
    },
    {
      "epoch": 68.7732506643047,
      "grad_norm": 0.25534045696258545,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 77645
    },
    {
      "epoch": 68.77767936226749,
      "grad_norm": 0.301859974861145,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 77650
    },
    {
      "epoch": 68.78210806023029,
      "grad_norm": 0.23446762561798096,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 77655
    },
    {
      "epoch": 68.7865367581931,
      "grad_norm": 0.2862578332424164,
      "learning_rate": 1e-05,
      "loss": 0.941,
      "step": 77660
    },
    {
      "epoch": 68.79096545615589,
      "grad_norm": 0.22713255882263184,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 77665
    },
    {
      "epoch": 68.79539415411868,
      "grad_norm": 0.24764148890972137,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 77670
    },
    {
      "epoch": 68.79982285208149,
      "grad_norm": 0.2324870377779007,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 77675
    },
    {
      "epoch": 68.80425155004428,
      "grad_norm": 0.24814823269844055,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 77680
    },
    {
      "epoch": 68.8086802480071,
      "grad_norm": 0.22148555517196655,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 77685
    },
    {
      "epoch": 68.81310894596989,
      "grad_norm": 0.23959319293498993,
      "learning_rate": 1e-05,
      "loss": 0.893,
      "step": 77690
    },
    {
      "epoch": 68.81753764393268,
      "grad_norm": 0.2729448676109314,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 77695
    },
    {
      "epoch": 68.82196634189549,
      "grad_norm": 0.24853262305259705,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 77700
    },
    {
      "epoch": 68.82639503985828,
      "grad_norm": 0.2562715709209442,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 77705
    },
    {
      "epoch": 68.83082373782108,
      "grad_norm": 0.25085675716400146,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 77710
    },
    {
      "epoch": 68.83525243578389,
      "grad_norm": 0.25476396083831787,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 77715
    },
    {
      "epoch": 68.83968113374668,
      "grad_norm": 0.2540120780467987,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 77720
    },
    {
      "epoch": 68.84410983170947,
      "grad_norm": 0.217283695936203,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 77725
    },
    {
      "epoch": 68.84853852967228,
      "grad_norm": 0.2352837473154068,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 77730
    },
    {
      "epoch": 68.85296722763508,
      "grad_norm": 0.22519177198410034,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 77735
    },
    {
      "epoch": 68.85739592559787,
      "grad_norm": 0.219848170876503,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 77740
    },
    {
      "epoch": 68.86182462356068,
      "grad_norm": 0.24652816355228424,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 77745
    },
    {
      "epoch": 68.86625332152347,
      "grad_norm": 0.2178501933813095,
      "learning_rate": 1e-05,
      "loss": 1.0261,
      "step": 77750
    },
    {
      "epoch": 68.87068201948627,
      "grad_norm": 0.22229357063770294,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 77755
    },
    {
      "epoch": 68.87511071744908,
      "grad_norm": 0.2508009076118469,
      "learning_rate": 1e-05,
      "loss": 1.0261,
      "step": 77760
    },
    {
      "epoch": 68.87953941541187,
      "grad_norm": 0.2688422203063965,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 77765
    },
    {
      "epoch": 68.88396811337466,
      "grad_norm": 0.27963051199913025,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 77770
    },
    {
      "epoch": 68.88839681133747,
      "grad_norm": 0.2947990298271179,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 77775
    },
    {
      "epoch": 68.89282550930027,
      "grad_norm": 0.24646307528018951,
      "learning_rate": 1e-05,
      "loss": 0.9289,
      "step": 77780
    },
    {
      "epoch": 68.89725420726306,
      "grad_norm": 0.2942536473274231,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 77785
    },
    {
      "epoch": 68.90168290522587,
      "grad_norm": 0.2099767029285431,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 77790
    },
    {
      "epoch": 68.90611160318866,
      "grad_norm": 0.29026347398757935,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 77795
    },
    {
      "epoch": 68.91054030115146,
      "grad_norm": 0.21217495203018188,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 77800
    },
    {
      "epoch": 68.91496899911427,
      "grad_norm": 0.21227198839187622,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 77805
    },
    {
      "epoch": 68.91939769707706,
      "grad_norm": 0.2930909991264343,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 77810
    },
    {
      "epoch": 68.92382639503985,
      "grad_norm": 0.23876406252384186,
      "learning_rate": 1e-05,
      "loss": 0.9081,
      "step": 77815
    },
    {
      "epoch": 68.92825509300266,
      "grad_norm": 0.21298159658908844,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 77820
    },
    {
      "epoch": 68.93268379096546,
      "grad_norm": 0.22780898213386536,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 77825
    },
    {
      "epoch": 68.93711248892825,
      "grad_norm": 0.22752320766448975,
      "learning_rate": 1e-05,
      "loss": 1.0245,
      "step": 77830
    },
    {
      "epoch": 68.94154118689106,
      "grad_norm": 0.2558114230632782,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 77835
    },
    {
      "epoch": 68.94596988485385,
      "grad_norm": 0.20823952555656433,
      "learning_rate": 1e-05,
      "loss": 0.9347,
      "step": 77840
    },
    {
      "epoch": 68.95039858281665,
      "grad_norm": 0.2360299527645111,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 77845
    },
    {
      "epoch": 68.95482728077945,
      "grad_norm": 0.21293459832668304,
      "learning_rate": 1e-05,
      "loss": 0.9427,
      "step": 77850
    },
    {
      "epoch": 68.95925597874225,
      "grad_norm": 0.2501783072948456,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 77855
    },
    {
      "epoch": 68.96368467670504,
      "grad_norm": 0.24438054859638214,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 77860
    },
    {
      "epoch": 68.96811337466785,
      "grad_norm": 0.21473801136016846,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 77865
    },
    {
      "epoch": 68.97254207263065,
      "grad_norm": 0.25880977511405945,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 77870
    },
    {
      "epoch": 68.97697077059344,
      "grad_norm": 0.2433009147644043,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 77875
    },
    {
      "epoch": 68.98139946855625,
      "grad_norm": 0.21708384156227112,
      "learning_rate": 1e-05,
      "loss": 0.9261,
      "step": 77880
    },
    {
      "epoch": 68.98582816651904,
      "grad_norm": 0.2144363969564438,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 77885
    },
    {
      "epoch": 68.99025686448184,
      "grad_norm": 0.2765505015850067,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 77890
    },
    {
      "epoch": 68.99468556244464,
      "grad_norm": 0.213337242603302,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 77895
    },
    {
      "epoch": 68.99911426040744,
      "grad_norm": 0.26184558868408203,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 77900
    },
    {
      "epoch": 69.00354295837023,
      "grad_norm": 0.24421700835227966,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 77905
    },
    {
      "epoch": 69.00797165633304,
      "grad_norm": 0.22132588922977448,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 77910
    },
    {
      "epoch": 69.01240035429583,
      "grad_norm": 0.2508220374584198,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 77915
    },
    {
      "epoch": 69.01682905225863,
      "grad_norm": 0.23341313004493713,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 77920
    },
    {
      "epoch": 69.02125775022144,
      "grad_norm": 0.2203708291053772,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 77925
    },
    {
      "epoch": 69.02568644818423,
      "grad_norm": 0.23375320434570312,
      "learning_rate": 1e-05,
      "loss": 0.9038,
      "step": 77930
    },
    {
      "epoch": 69.03011514614704,
      "grad_norm": 0.2515154182910919,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 77935
    },
    {
      "epoch": 69.03454384410983,
      "grad_norm": 0.27470308542251587,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 77940
    },
    {
      "epoch": 69.03897254207263,
      "grad_norm": 0.240151047706604,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 77945
    },
    {
      "epoch": 69.04340124003544,
      "grad_norm": 0.2680210769176483,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 77950
    },
    {
      "epoch": 69.04782993799823,
      "grad_norm": 0.21505847573280334,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 77955
    },
    {
      "epoch": 69.05225863596102,
      "grad_norm": 0.21608306467533112,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 77960
    },
    {
      "epoch": 69.05668733392383,
      "grad_norm": 0.2254691869020462,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 77965
    },
    {
      "epoch": 69.06111603188663,
      "grad_norm": 0.2361876368522644,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 77970
    },
    {
      "epoch": 69.06554472984942,
      "grad_norm": 0.22258225083351135,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 77975
    },
    {
      "epoch": 69.06997342781223,
      "grad_norm": 0.31070664525032043,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 77980
    },
    {
      "epoch": 69.07440212577502,
      "grad_norm": 0.28711432218551636,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 77985
    },
    {
      "epoch": 69.07883082373782,
      "grad_norm": 0.23573873937129974,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 77990
    },
    {
      "epoch": 69.08325952170063,
      "grad_norm": 0.25527769327163696,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 77995
    },
    {
      "epoch": 69.08768821966342,
      "grad_norm": 0.27001112699508667,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 78000
    },
    {
      "epoch": 69.09211691762621,
      "grad_norm": 0.28292331099510193,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 78005
    },
    {
      "epoch": 69.09654561558902,
      "grad_norm": 0.2472730278968811,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 78010
    },
    {
      "epoch": 69.10097431355182,
      "grad_norm": 0.2121076136827469,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 78015
    },
    {
      "epoch": 69.10540301151461,
      "grad_norm": 0.21238595247268677,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 78020
    },
    {
      "epoch": 69.10983170947742,
      "grad_norm": 0.20333240926265717,
      "learning_rate": 1e-05,
      "loss": 0.9294,
      "step": 78025
    },
    {
      "epoch": 69.11426040744021,
      "grad_norm": 0.23517295718193054,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 78030
    },
    {
      "epoch": 69.118689105403,
      "grad_norm": 0.21922169625759125,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 78035
    },
    {
      "epoch": 69.12311780336582,
      "grad_norm": 0.2950936257839203,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 78040
    },
    {
      "epoch": 69.12754650132861,
      "grad_norm": 0.210884690284729,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 78045
    },
    {
      "epoch": 69.1319751992914,
      "grad_norm": 0.2227676659822464,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 78050
    },
    {
      "epoch": 69.13640389725421,
      "grad_norm": 0.25897300243377686,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 78055
    },
    {
      "epoch": 69.140832595217,
      "grad_norm": 0.23228009045124054,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 78060
    },
    {
      "epoch": 69.1452612931798,
      "grad_norm": 0.23589274287223816,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 78065
    },
    {
      "epoch": 69.14968999114261,
      "grad_norm": 0.2592189610004425,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 78070
    },
    {
      "epoch": 69.1541186891054,
      "grad_norm": 0.21341092884540558,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 78075
    },
    {
      "epoch": 69.1585473870682,
      "grad_norm": 0.26577284932136536,
      "learning_rate": 1e-05,
      "loss": 1.036,
      "step": 78080
    },
    {
      "epoch": 69.162976085031,
      "grad_norm": 0.24455757439136505,
      "learning_rate": 1e-05,
      "loss": 0.9165,
      "step": 78085
    },
    {
      "epoch": 69.1674047829938,
      "grad_norm": 0.28228557109832764,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 78090
    },
    {
      "epoch": 69.1718334809566,
      "grad_norm": 0.253810316324234,
      "learning_rate": 1e-05,
      "loss": 0.9157,
      "step": 78095
    },
    {
      "epoch": 69.1762621789194,
      "grad_norm": 0.26682087779045105,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 78100
    },
    {
      "epoch": 69.1806908768822,
      "grad_norm": 0.23522455990314484,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 78105
    },
    {
      "epoch": 69.18511957484499,
      "grad_norm": 0.22824148833751678,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 78110
    },
    {
      "epoch": 69.1895482728078,
      "grad_norm": 0.25177592039108276,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 78115
    },
    {
      "epoch": 69.19397697077059,
      "grad_norm": 0.2501276135444641,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 78120
    },
    {
      "epoch": 69.19840566873339,
      "grad_norm": 0.24373245239257812,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 78125
    },
    {
      "epoch": 69.2028343666962,
      "grad_norm": 0.2491762489080429,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 78130
    },
    {
      "epoch": 69.20726306465899,
      "grad_norm": 0.2440991997718811,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 78135
    },
    {
      "epoch": 69.21169176262178,
      "grad_norm": 0.21512573957443237,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 78140
    },
    {
      "epoch": 69.21612046058459,
      "grad_norm": 0.22907035052776337,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 78145
    },
    {
      "epoch": 69.22054915854739,
      "grad_norm": 0.22105741500854492,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 78150
    },
    {
      "epoch": 69.22497785651018,
      "grad_norm": 0.22191956639289856,
      "learning_rate": 1e-05,
      "loss": 0.9175,
      "step": 78155
    },
    {
      "epoch": 69.22940655447299,
      "grad_norm": 0.23449578881263733,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 78160
    },
    {
      "epoch": 69.23383525243578,
      "grad_norm": 0.27362146973609924,
      "learning_rate": 1e-05,
      "loss": 0.9063,
      "step": 78165
    },
    {
      "epoch": 69.23826395039858,
      "grad_norm": 0.23117782175540924,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 78170
    },
    {
      "epoch": 69.24269264836138,
      "grad_norm": 0.2439649999141693,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 78175
    },
    {
      "epoch": 69.24712134632418,
      "grad_norm": 0.23415379226207733,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 78180
    },
    {
      "epoch": 69.25155004428699,
      "grad_norm": 0.22621458768844604,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 78185
    },
    {
      "epoch": 69.25597874224978,
      "grad_norm": 0.2816198170185089,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 78190
    },
    {
      "epoch": 69.26040744021257,
      "grad_norm": 0.24278493225574493,
      "learning_rate": 1e-05,
      "loss": 0.9493,
      "step": 78195
    },
    {
      "epoch": 69.26483613817538,
      "grad_norm": 0.27565550804138184,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 78200
    },
    {
      "epoch": 69.26926483613818,
      "grad_norm": 0.21678772568702698,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 78205
    },
    {
      "epoch": 69.27369353410097,
      "grad_norm": 0.26439160108566284,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 78210
    },
    {
      "epoch": 69.27812223206378,
      "grad_norm": 0.21772649884223938,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 78215
    },
    {
      "epoch": 69.28255093002657,
      "grad_norm": 0.2593878209590912,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 78220
    },
    {
      "epoch": 69.28697962798937,
      "grad_norm": 0.24496276676654816,
      "learning_rate": 1e-05,
      "loss": 0.9133,
      "step": 78225
    },
    {
      "epoch": 69.29140832595218,
      "grad_norm": 0.2970627248287201,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 78230
    },
    {
      "epoch": 69.29583702391497,
      "grad_norm": 0.24245810508728027,
      "learning_rate": 1e-05,
      "loss": 1.0205,
      "step": 78235
    },
    {
      "epoch": 69.30026572187776,
      "grad_norm": 0.23077483475208282,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 78240
    },
    {
      "epoch": 69.30469441984057,
      "grad_norm": 0.27574291825294495,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 78245
    },
    {
      "epoch": 69.30912311780337,
      "grad_norm": 0.2636722922325134,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 78250
    },
    {
      "epoch": 69.31355181576616,
      "grad_norm": 0.21882973611354828,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 78255
    },
    {
      "epoch": 69.31798051372897,
      "grad_norm": 0.27143245935440063,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 78260
    },
    {
      "epoch": 69.32240921169176,
      "grad_norm": 0.2611597776412964,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 78265
    },
    {
      "epoch": 69.32683790965456,
      "grad_norm": 0.264215886592865,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 78270
    },
    {
      "epoch": 69.33126660761737,
      "grad_norm": 0.2768572270870209,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 78275
    },
    {
      "epoch": 69.33569530558016,
      "grad_norm": 0.3088644742965698,
      "learning_rate": 1e-05,
      "loss": 0.9075,
      "step": 78280
    },
    {
      "epoch": 69.34012400354295,
      "grad_norm": 0.2964226305484772,
      "learning_rate": 1e-05,
      "loss": 0.9307,
      "step": 78285
    },
    {
      "epoch": 69.34455270150576,
      "grad_norm": 0.30500662326812744,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 78290
    },
    {
      "epoch": 69.34898139946856,
      "grad_norm": 0.21796412765979767,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 78295
    },
    {
      "epoch": 69.35341009743135,
      "grad_norm": 0.2305801808834076,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 78300
    },
    {
      "epoch": 69.35783879539416,
      "grad_norm": 0.22625824809074402,
      "learning_rate": 1e-05,
      "loss": 0.8954,
      "step": 78305
    },
    {
      "epoch": 69.36226749335695,
      "grad_norm": 0.24406488239765167,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 78310
    },
    {
      "epoch": 69.36669619131975,
      "grad_norm": 0.2312549501657486,
      "learning_rate": 1e-05,
      "loss": 0.9041,
      "step": 78315
    },
    {
      "epoch": 69.37112488928256,
      "grad_norm": 0.2332194447517395,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 78320
    },
    {
      "epoch": 69.37555358724535,
      "grad_norm": 0.22481530904769897,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 78325
    },
    {
      "epoch": 69.37998228520814,
      "grad_norm": 0.2588733732700348,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 78330
    },
    {
      "epoch": 69.38441098317095,
      "grad_norm": 0.23142071068286896,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 78335
    },
    {
      "epoch": 69.38883968113375,
      "grad_norm": 0.22105400264263153,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 78340
    },
    {
      "epoch": 69.39326837909654,
      "grad_norm": 0.22179068624973297,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 78345
    },
    {
      "epoch": 69.39769707705935,
      "grad_norm": 0.24478228390216827,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 78350
    },
    {
      "epoch": 69.40212577502214,
      "grad_norm": 0.22888080775737762,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 78355
    },
    {
      "epoch": 69.40655447298494,
      "grad_norm": 0.2257438600063324,
      "learning_rate": 1e-05,
      "loss": 0.9244,
      "step": 78360
    },
    {
      "epoch": 69.41098317094774,
      "grad_norm": 0.21681204438209534,
      "learning_rate": 1e-05,
      "loss": 0.9298,
      "step": 78365
    },
    {
      "epoch": 69.41541186891054,
      "grad_norm": 0.21038585901260376,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 78370
    },
    {
      "epoch": 69.41984056687333,
      "grad_norm": 0.21871738135814667,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 78375
    },
    {
      "epoch": 69.42426926483614,
      "grad_norm": 0.18269741535186768,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 78380
    },
    {
      "epoch": 69.42869796279894,
      "grad_norm": 0.26974815130233765,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 78385
    },
    {
      "epoch": 69.43312666076173,
      "grad_norm": 0.24466097354888916,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 78390
    },
    {
      "epoch": 69.43755535872454,
      "grad_norm": 0.257714182138443,
      "learning_rate": 1e-05,
      "loss": 0.9347,
      "step": 78395
    },
    {
      "epoch": 69.44198405668733,
      "grad_norm": 0.22286882996559143,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 78400
    },
    {
      "epoch": 69.44641275465013,
      "grad_norm": 0.24475909769535065,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 78405
    },
    {
      "epoch": 69.45084145261293,
      "grad_norm": 0.2519271969795227,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 78410
    },
    {
      "epoch": 69.45527015057573,
      "grad_norm": 0.25500184297561646,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 78415
    },
    {
      "epoch": 69.45969884853854,
      "grad_norm": 0.2197030782699585,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 78420
    },
    {
      "epoch": 69.46412754650133,
      "grad_norm": 0.2599322199821472,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 78425
    },
    {
      "epoch": 69.46855624446412,
      "grad_norm": 0.2638357877731323,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 78430
    },
    {
      "epoch": 69.47298494242693,
      "grad_norm": 0.2599092125892639,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 78435
    },
    {
      "epoch": 69.47741364038973,
      "grad_norm": 0.231001615524292,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 78440
    },
    {
      "epoch": 69.48184233835252,
      "grad_norm": 0.24882809817790985,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 78445
    },
    {
      "epoch": 69.48627103631533,
      "grad_norm": 0.2364180088043213,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 78450
    },
    {
      "epoch": 69.49069973427812,
      "grad_norm": 0.26200833916664124,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 78455
    },
    {
      "epoch": 69.49512843224092,
      "grad_norm": 0.25467485189437866,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 78460
    },
    {
      "epoch": 69.49955713020373,
      "grad_norm": 0.2419613152742386,
      "learning_rate": 1e-05,
      "loss": 1.0285,
      "step": 78465
    },
    {
      "epoch": 69.50398582816652,
      "grad_norm": 0.2418401837348938,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 78470
    },
    {
      "epoch": 69.50841452612931,
      "grad_norm": 0.27183210849761963,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 78475
    },
    {
      "epoch": 69.51284322409212,
      "grad_norm": 0.23429618775844574,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 78480
    },
    {
      "epoch": 69.51727192205492,
      "grad_norm": 0.2655896246433258,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 78485
    },
    {
      "epoch": 69.52170062001771,
      "grad_norm": 0.2690069079399109,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 78490
    },
    {
      "epoch": 69.52612931798052,
      "grad_norm": 0.24193212389945984,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 78495
    },
    {
      "epoch": 69.53055801594331,
      "grad_norm": 0.23649504780769348,
      "learning_rate": 1e-05,
      "loss": 0.9085,
      "step": 78500
    },
    {
      "epoch": 69.53498671390611,
      "grad_norm": 0.22794711589813232,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 78505
    },
    {
      "epoch": 69.53941541186892,
      "grad_norm": 0.31951290369033813,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 78510
    },
    {
      "epoch": 69.54384410983171,
      "grad_norm": 0.24057066440582275,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 78515
    },
    {
      "epoch": 69.5482728077945,
      "grad_norm": 0.24476291239261627,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 78520
    },
    {
      "epoch": 69.55270150575731,
      "grad_norm": 0.3005260229110718,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 78525
    },
    {
      "epoch": 69.5571302037201,
      "grad_norm": 0.2852076292037964,
      "learning_rate": 1e-05,
      "loss": 1.0075,
      "step": 78530
    },
    {
      "epoch": 69.5615589016829,
      "grad_norm": 0.2528403401374817,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 78535
    },
    {
      "epoch": 69.56598759964571,
      "grad_norm": 0.29602670669555664,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 78540
    },
    {
      "epoch": 69.5704162976085,
      "grad_norm": 0.2505547106266022,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 78545
    },
    {
      "epoch": 69.5748449955713,
      "grad_norm": 0.22935326397418976,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 78550
    },
    {
      "epoch": 69.5792736935341,
      "grad_norm": 0.2139900177717209,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 78555
    },
    {
      "epoch": 69.5837023914969,
      "grad_norm": 0.23865588009357452,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 78560
    },
    {
      "epoch": 69.5881310894597,
      "grad_norm": 0.22912459075450897,
      "learning_rate": 1e-05,
      "loss": 0.9168,
      "step": 78565
    },
    {
      "epoch": 69.5925597874225,
      "grad_norm": 0.259901762008667,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 78570
    },
    {
      "epoch": 69.5969884853853,
      "grad_norm": 0.24522702395915985,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 78575
    },
    {
      "epoch": 69.60141718334809,
      "grad_norm": 0.25122779607772827,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 78580
    },
    {
      "epoch": 69.6058458813109,
      "grad_norm": 0.25533708930015564,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 78585
    },
    {
      "epoch": 69.61027457927369,
      "grad_norm": 0.2665473520755768,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 78590
    },
    {
      "epoch": 69.61470327723649,
      "grad_norm": 0.2831224799156189,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 78595
    },
    {
      "epoch": 69.6191319751993,
      "grad_norm": 0.23435364663600922,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 78600
    },
    {
      "epoch": 69.62356067316209,
      "grad_norm": 0.23897412419319153,
      "learning_rate": 1e-05,
      "loss": 0.9102,
      "step": 78605
    },
    {
      "epoch": 69.62798937112488,
      "grad_norm": 0.2745687663555145,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 78610
    },
    {
      "epoch": 69.63241806908769,
      "grad_norm": 0.21054910123348236,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 78615
    },
    {
      "epoch": 69.63684676705049,
      "grad_norm": 0.23244932293891907,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 78620
    },
    {
      "epoch": 69.64127546501328,
      "grad_norm": 0.2235032021999359,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 78625
    },
    {
      "epoch": 69.64570416297609,
      "grad_norm": 0.25063931941986084,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 78630
    },
    {
      "epoch": 69.65013286093888,
      "grad_norm": 0.24528315663337708,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 78635
    },
    {
      "epoch": 69.65456155890168,
      "grad_norm": 0.21745511889457703,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 78640
    },
    {
      "epoch": 69.65899025686448,
      "grad_norm": 0.24475501477718353,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 78645
    },
    {
      "epoch": 69.66341895482728,
      "grad_norm": 0.22695426642894745,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 78650
    },
    {
      "epoch": 69.66784765279007,
      "grad_norm": 0.25105684995651245,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 78655
    },
    {
      "epoch": 69.67227635075288,
      "grad_norm": 0.23489674925804138,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 78660
    },
    {
      "epoch": 69.67670504871568,
      "grad_norm": 0.212635338306427,
      "learning_rate": 1e-05,
      "loss": 0.9212,
      "step": 78665
    },
    {
      "epoch": 69.68113374667848,
      "grad_norm": 0.20544585585594177,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 78670
    },
    {
      "epoch": 69.68556244464128,
      "grad_norm": 0.26581746339797974,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 78675
    },
    {
      "epoch": 69.68999114260407,
      "grad_norm": 0.29279255867004395,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 78680
    },
    {
      "epoch": 69.69441984056688,
      "grad_norm": 0.23660996556282043,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 78685
    },
    {
      "epoch": 69.69884853852967,
      "grad_norm": 0.3415919542312622,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 78690
    },
    {
      "epoch": 69.70327723649247,
      "grad_norm": 0.2696986794471741,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 78695
    },
    {
      "epoch": 69.70770593445528,
      "grad_norm": 0.2356952726840973,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 78700
    },
    {
      "epoch": 69.71213463241807,
      "grad_norm": 0.22103829681873322,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 78705
    },
    {
      "epoch": 69.71656333038086,
      "grad_norm": 0.2711811363697052,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 78710
    },
    {
      "epoch": 69.72099202834367,
      "grad_norm": 0.21430125832557678,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 78715
    },
    {
      "epoch": 69.72542072630647,
      "grad_norm": 0.29422467947006226,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 78720
    },
    {
      "epoch": 69.72984942426926,
      "grad_norm": 0.22700706124305725,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 78725
    },
    {
      "epoch": 69.73427812223207,
      "grad_norm": 0.21754463016986847,
      "learning_rate": 1e-05,
      "loss": 0.9271,
      "step": 78730
    },
    {
      "epoch": 69.73870682019486,
      "grad_norm": 0.22344212234020233,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 78735
    },
    {
      "epoch": 69.74313551815766,
      "grad_norm": 0.26332640647888184,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 78740
    },
    {
      "epoch": 69.74756421612047,
      "grad_norm": 0.22788086533546448,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 78745
    },
    {
      "epoch": 69.75199291408326,
      "grad_norm": 0.26933398842811584,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 78750
    },
    {
      "epoch": 69.75642161204605,
      "grad_norm": 0.2200969010591507,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 78755
    },
    {
      "epoch": 69.76085031000886,
      "grad_norm": 0.21304844319820404,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 78760
    },
    {
      "epoch": 69.76527900797166,
      "grad_norm": 0.2831621766090393,
      "learning_rate": 1e-05,
      "loss": 0.8625,
      "step": 78765
    },
    {
      "epoch": 69.76970770593445,
      "grad_norm": 0.25681835412979126,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 78770
    },
    {
      "epoch": 69.77413640389726,
      "grad_norm": 0.2893185615539551,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 78775
    },
    {
      "epoch": 69.77856510186005,
      "grad_norm": 0.2517160475254059,
      "learning_rate": 1e-05,
      "loss": 1.0208,
      "step": 78780
    },
    {
      "epoch": 69.78299379982285,
      "grad_norm": 0.2583816349506378,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 78785
    },
    {
      "epoch": 69.78742249778566,
      "grad_norm": 0.24481059610843658,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 78790
    },
    {
      "epoch": 69.79185119574845,
      "grad_norm": 0.2583755850791931,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 78795
    },
    {
      "epoch": 69.79627989371124,
      "grad_norm": 0.2394489049911499,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 78800
    },
    {
      "epoch": 69.80070859167405,
      "grad_norm": 0.24175888299942017,
      "learning_rate": 1e-05,
      "loss": 0.9258,
      "step": 78805
    },
    {
      "epoch": 69.80513728963685,
      "grad_norm": 0.22404834628105164,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 78810
    },
    {
      "epoch": 69.80956598759964,
      "grad_norm": 0.2633533477783203,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 78815
    },
    {
      "epoch": 69.81399468556245,
      "grad_norm": 0.22801053524017334,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 78820
    },
    {
      "epoch": 69.81842338352524,
      "grad_norm": 0.2548644244670868,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 78825
    },
    {
      "epoch": 69.82285208148804,
      "grad_norm": 0.21931657195091248,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 78830
    },
    {
      "epoch": 69.82728077945085,
      "grad_norm": 0.22544324398040771,
      "learning_rate": 1e-05,
      "loss": 1.0259,
      "step": 78835
    },
    {
      "epoch": 69.83170947741364,
      "grad_norm": 0.206543430685997,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 78840
    },
    {
      "epoch": 69.83613817537643,
      "grad_norm": 0.26624545454978943,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 78845
    },
    {
      "epoch": 69.84056687333924,
      "grad_norm": 0.23478882014751434,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 78850
    },
    {
      "epoch": 69.84499557130204,
      "grad_norm": 0.26856914162635803,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 78855
    },
    {
      "epoch": 69.84942426926483,
      "grad_norm": 0.25222253799438477,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 78860
    },
    {
      "epoch": 69.85385296722764,
      "grad_norm": 0.23517948389053345,
      "learning_rate": 1e-05,
      "loss": 0.9104,
      "step": 78865
    },
    {
      "epoch": 69.85828166519043,
      "grad_norm": 0.2656784653663635,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 78870
    },
    {
      "epoch": 69.86271036315323,
      "grad_norm": 0.21019315719604492,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 78875
    },
    {
      "epoch": 69.86713906111603,
      "grad_norm": 0.2421507090330124,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 78880
    },
    {
      "epoch": 69.87156775907883,
      "grad_norm": 0.24652041494846344,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 78885
    },
    {
      "epoch": 69.87599645704162,
      "grad_norm": 0.20786546170711517,
      "learning_rate": 1e-05,
      "loss": 0.9141,
      "step": 78890
    },
    {
      "epoch": 69.88042515500443,
      "grad_norm": 0.22649672627449036,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 78895
    },
    {
      "epoch": 69.88485385296723,
      "grad_norm": 0.2741950452327728,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 78900
    },
    {
      "epoch": 69.88928255093003,
      "grad_norm": 0.1955946832895279,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 78905
    },
    {
      "epoch": 69.89371124889283,
      "grad_norm": 0.23754297196865082,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 78910
    },
    {
      "epoch": 69.89813994685562,
      "grad_norm": 0.21939845383167267,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 78915
    },
    {
      "epoch": 69.90256864481843,
      "grad_norm": 0.19615289568901062,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 78920
    },
    {
      "epoch": 69.90699734278122,
      "grad_norm": 0.2401057779788971,
      "learning_rate": 1e-05,
      "loss": 0.9325,
      "step": 78925
    },
    {
      "epoch": 69.91142604074402,
      "grad_norm": 0.2540321946144104,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 78930
    },
    {
      "epoch": 69.91585473870683,
      "grad_norm": 0.24981608986854553,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 78935
    },
    {
      "epoch": 69.92028343666962,
      "grad_norm": 0.22702178359031677,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 78940
    },
    {
      "epoch": 69.92471213463241,
      "grad_norm": 0.21843881905078888,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 78945
    },
    {
      "epoch": 69.92914083259522,
      "grad_norm": 0.2285066395998001,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 78950
    },
    {
      "epoch": 69.93356953055802,
      "grad_norm": 0.23991787433624268,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 78955
    },
    {
      "epoch": 69.93799822852081,
      "grad_norm": 0.24507230520248413,
      "learning_rate": 1e-05,
      "loss": 0.9278,
      "step": 78960
    },
    {
      "epoch": 69.94242692648362,
      "grad_norm": 0.2080422341823578,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 78965
    },
    {
      "epoch": 69.94685562444641,
      "grad_norm": 0.2088281810283661,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 78970
    },
    {
      "epoch": 69.95128432240921,
      "grad_norm": 0.22768926620483398,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 78975
    },
    {
      "epoch": 69.95571302037202,
      "grad_norm": 0.24241498112678528,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 78980
    },
    {
      "epoch": 69.96014171833481,
      "grad_norm": 0.2541194260120392,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 78985
    },
    {
      "epoch": 69.9645704162976,
      "grad_norm": 0.2567010223865509,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 78990
    },
    {
      "epoch": 69.96899911426041,
      "grad_norm": 0.29952529072761536,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 78995
    },
    {
      "epoch": 69.9734278122232,
      "grad_norm": 0.31167447566986084,
      "learning_rate": 1e-05,
      "loss": 0.9215,
      "step": 79000
    },
    {
      "epoch": 69.977856510186,
      "grad_norm": 0.24813435971736908,
      "learning_rate": 1e-05,
      "loss": 0.9329,
      "step": 79005
    },
    {
      "epoch": 69.98228520814881,
      "grad_norm": 0.2542593479156494,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 79010
    },
    {
      "epoch": 69.9867139061116,
      "grad_norm": 0.30488407611846924,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 79015
    },
    {
      "epoch": 69.9911426040744,
      "grad_norm": 0.235349640250206,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 79020
    },
    {
      "epoch": 69.9955713020372,
      "grad_norm": 0.23907087743282318,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 79025
    },
    {
      "epoch": 70.0,
      "grad_norm": 0.19082893431186676,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 79030
    },
    {
      "epoch": 70.0044286979628,
      "grad_norm": 0.2625238299369812,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 79035
    },
    {
      "epoch": 70.0088573959256,
      "grad_norm": 0.24825170636177063,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 79040
    },
    {
      "epoch": 70.0132860938884,
      "grad_norm": 0.2068416178226471,
      "learning_rate": 1e-05,
      "loss": 0.9178,
      "step": 79045
    },
    {
      "epoch": 70.01771479185119,
      "grad_norm": 0.23527489602565765,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 79050
    },
    {
      "epoch": 70.022143489814,
      "grad_norm": 0.22223863005638123,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 79055
    },
    {
      "epoch": 70.0265721877768,
      "grad_norm": 0.23251789808273315,
      "learning_rate": 1e-05,
      "loss": 0.9357,
      "step": 79060
    },
    {
      "epoch": 70.03100088573959,
      "grad_norm": 0.1982160061597824,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 79065
    },
    {
      "epoch": 70.0354295837024,
      "grad_norm": 0.22085952758789062,
      "learning_rate": 1e-05,
      "loss": 0.9306,
      "step": 79070
    },
    {
      "epoch": 70.03985828166519,
      "grad_norm": 0.23387576639652252,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 79075
    },
    {
      "epoch": 70.04428697962798,
      "grad_norm": 0.22022143006324768,
      "learning_rate": 1e-05,
      "loss": 0.922,
      "step": 79080
    },
    {
      "epoch": 70.04871567759079,
      "grad_norm": 0.2559187710285187,
      "learning_rate": 1e-05,
      "loss": 0.9254,
      "step": 79085
    },
    {
      "epoch": 70.05314437555359,
      "grad_norm": 0.2740534245967865,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 79090
    },
    {
      "epoch": 70.05757307351638,
      "grad_norm": 0.22527728974819183,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 79095
    },
    {
      "epoch": 70.06200177147919,
      "grad_norm": 0.22429199516773224,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 79100
    },
    {
      "epoch": 70.06643046944198,
      "grad_norm": 0.24807880818843842,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 79105
    },
    {
      "epoch": 70.07085916740478,
      "grad_norm": 0.28592634201049805,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 79110
    },
    {
      "epoch": 70.07528786536759,
      "grad_norm": 0.224885493516922,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 79115
    },
    {
      "epoch": 70.07971656333038,
      "grad_norm": 0.2100324183702469,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 79120
    },
    {
      "epoch": 70.08414526129317,
      "grad_norm": 0.2536398470401764,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 79125
    },
    {
      "epoch": 70.08857395925598,
      "grad_norm": 0.20576679706573486,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 79130
    },
    {
      "epoch": 70.09300265721878,
      "grad_norm": 0.23106379806995392,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 79135
    },
    {
      "epoch": 70.09743135518157,
      "grad_norm": 0.2338072806596756,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 79140
    },
    {
      "epoch": 70.10186005314438,
      "grad_norm": 0.27735385298728943,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 79145
    },
    {
      "epoch": 70.10628875110717,
      "grad_norm": 0.21151185035705566,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 79150
    },
    {
      "epoch": 70.11071744906998,
      "grad_norm": 0.2400214970111847,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 79155
    },
    {
      "epoch": 70.11514614703277,
      "grad_norm": 0.2658084034919739,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 79160
    },
    {
      "epoch": 70.11957484499557,
      "grad_norm": 0.22883708775043488,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 79165
    },
    {
      "epoch": 70.12400354295838,
      "grad_norm": 0.2803514301776886,
      "learning_rate": 1e-05,
      "loss": 0.912,
      "step": 79170
    },
    {
      "epoch": 70.12843224092117,
      "grad_norm": 0.2579149007797241,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 79175
    },
    {
      "epoch": 70.13286093888397,
      "grad_norm": 0.24845117330551147,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 79180
    },
    {
      "epoch": 70.13728963684677,
      "grad_norm": 0.2834876775741577,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 79185
    },
    {
      "epoch": 70.14171833480957,
      "grad_norm": 0.2607656717300415,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 79190
    },
    {
      "epoch": 70.14614703277236,
      "grad_norm": 0.2543274462223053,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 79195
    },
    {
      "epoch": 70.15057573073517,
      "grad_norm": 0.27856314182281494,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 79200
    },
    {
      "epoch": 70.15500442869796,
      "grad_norm": 0.2586901783943176,
      "learning_rate": 1e-05,
      "loss": 1.0232,
      "step": 79205
    },
    {
      "epoch": 70.15943312666076,
      "grad_norm": 0.24000760912895203,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 79210
    },
    {
      "epoch": 70.16386182462357,
      "grad_norm": 0.26503807306289673,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 79215
    },
    {
      "epoch": 70.16829052258636,
      "grad_norm": 0.23649385571479797,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 79220
    },
    {
      "epoch": 70.17271922054915,
      "grad_norm": 0.25321412086486816,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 79225
    },
    {
      "epoch": 70.17714791851196,
      "grad_norm": 0.2565304636955261,
      "learning_rate": 1e-05,
      "loss": 0.9139,
      "step": 79230
    },
    {
      "epoch": 70.18157661647476,
      "grad_norm": 0.23995286226272583,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 79235
    },
    {
      "epoch": 70.18600531443755,
      "grad_norm": 0.24533501267433167,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 79240
    },
    {
      "epoch": 70.19043401240036,
      "grad_norm": 0.23964214324951172,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 79245
    },
    {
      "epoch": 70.19486271036315,
      "grad_norm": 0.2198006808757782,
      "learning_rate": 1e-05,
      "loss": 0.9289,
      "step": 79250
    },
    {
      "epoch": 70.19929140832595,
      "grad_norm": 0.26828646659851074,
      "learning_rate": 1e-05,
      "loss": 1.014,
      "step": 79255
    },
    {
      "epoch": 70.20372010628876,
      "grad_norm": 0.24599570035934448,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 79260
    },
    {
      "epoch": 70.20814880425155,
      "grad_norm": 0.27050408720970154,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 79265
    },
    {
      "epoch": 70.21257750221434,
      "grad_norm": 0.23714813590049744,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 79270
    },
    {
      "epoch": 70.21700620017715,
      "grad_norm": 0.21290239691734314,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 79275
    },
    {
      "epoch": 70.22143489813995,
      "grad_norm": 0.2118254005908966,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 79280
    },
    {
      "epoch": 70.22586359610274,
      "grad_norm": 0.26099902391433716,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 79285
    },
    {
      "epoch": 70.23029229406555,
      "grad_norm": 0.23516890406608582,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 79290
    },
    {
      "epoch": 70.23472099202834,
      "grad_norm": 0.27695444226264954,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 79295
    },
    {
      "epoch": 70.23914968999114,
      "grad_norm": 0.22662830352783203,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 79300
    },
    {
      "epoch": 70.24357838795395,
      "grad_norm": 0.2804628014564514,
      "learning_rate": 1e-05,
      "loss": 0.9252,
      "step": 79305
    },
    {
      "epoch": 70.24800708591674,
      "grad_norm": 0.27573657035827637,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 79310
    },
    {
      "epoch": 70.25243578387953,
      "grad_norm": 0.2663476765155792,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 79315
    },
    {
      "epoch": 70.25686448184234,
      "grad_norm": 0.22174453735351562,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 79320
    },
    {
      "epoch": 70.26129317980514,
      "grad_norm": 0.2670125961303711,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 79325
    },
    {
      "epoch": 70.26572187776793,
      "grad_norm": 0.2670302093029022,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 79330
    },
    {
      "epoch": 70.27015057573074,
      "grad_norm": 0.27516040205955505,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 79335
    },
    {
      "epoch": 70.27457927369353,
      "grad_norm": 0.24188938736915588,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 79340
    },
    {
      "epoch": 70.27900797165633,
      "grad_norm": 0.2529669404029846,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 79345
    },
    {
      "epoch": 70.28343666961914,
      "grad_norm": 0.2787066400051117,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 79350
    },
    {
      "epoch": 70.28786536758193,
      "grad_norm": 0.2558532655239105,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 79355
    },
    {
      "epoch": 70.29229406554472,
      "grad_norm": 0.24675633013248444,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 79360
    },
    {
      "epoch": 70.29672276350753,
      "grad_norm": 0.23144817352294922,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 79365
    },
    {
      "epoch": 70.30115146147033,
      "grad_norm": 0.25335338711738586,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 79370
    },
    {
      "epoch": 70.30558015943312,
      "grad_norm": 0.24836066365242004,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 79375
    },
    {
      "epoch": 70.31000885739593,
      "grad_norm": 0.24997423589229584,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 79380
    },
    {
      "epoch": 70.31443755535872,
      "grad_norm": 0.24430014193058014,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 79385
    },
    {
      "epoch": 70.31886625332152,
      "grad_norm": 0.2486829161643982,
      "learning_rate": 1e-05,
      "loss": 0.8792,
      "step": 79390
    },
    {
      "epoch": 70.32329495128432,
      "grad_norm": 0.2654065489768982,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 79395
    },
    {
      "epoch": 70.32772364924712,
      "grad_norm": 0.19608627259731293,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 79400
    },
    {
      "epoch": 70.33215234720993,
      "grad_norm": 0.20926392078399658,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 79405
    },
    {
      "epoch": 70.33658104517272,
      "grad_norm": 0.24228957295417786,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 79410
    },
    {
      "epoch": 70.34100974313552,
      "grad_norm": 0.23176786303520203,
      "learning_rate": 1e-05,
      "loss": 0.9264,
      "step": 79415
    },
    {
      "epoch": 70.34543844109832,
      "grad_norm": 0.26702576875686646,
      "learning_rate": 1e-05,
      "loss": 1.0175,
      "step": 79420
    },
    {
      "epoch": 70.34986713906112,
      "grad_norm": 0.2655392289161682,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 79425
    },
    {
      "epoch": 70.35429583702391,
      "grad_norm": 0.22481006383895874,
      "learning_rate": 1e-05,
      "loss": 0.923,
      "step": 79430
    },
    {
      "epoch": 70.35872453498672,
      "grad_norm": 0.23333734273910522,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 79435
    },
    {
      "epoch": 70.36315323294951,
      "grad_norm": 0.23280106484889984,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 79440
    },
    {
      "epoch": 70.36758193091231,
      "grad_norm": 0.223131000995636,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 79445
    },
    {
      "epoch": 70.37201062887512,
      "grad_norm": 0.25419679284095764,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 79450
    },
    {
      "epoch": 70.37643932683791,
      "grad_norm": 0.2271285355091095,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 79455
    },
    {
      "epoch": 70.3808680248007,
      "grad_norm": 0.27452829480171204,
      "learning_rate": 1e-05,
      "loss": 1.0252,
      "step": 79460
    },
    {
      "epoch": 70.38529672276351,
      "grad_norm": 0.22552655637264252,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 79465
    },
    {
      "epoch": 70.38972542072631,
      "grad_norm": 0.28763824701309204,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 79470
    },
    {
      "epoch": 70.3941541186891,
      "grad_norm": 0.2525554597377777,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 79475
    },
    {
      "epoch": 70.39858281665191,
      "grad_norm": 0.2524948716163635,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 79480
    },
    {
      "epoch": 70.4030115146147,
      "grad_norm": 0.21762143075466156,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 79485
    },
    {
      "epoch": 70.4074402125775,
      "grad_norm": 0.24756409227848053,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 79490
    },
    {
      "epoch": 70.4118689105403,
      "grad_norm": 0.22603274881839752,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 79495
    },
    {
      "epoch": 70.4162976085031,
      "grad_norm": 0.2580925226211548,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 79500
    },
    {
      "epoch": 70.4207263064659,
      "grad_norm": 0.26012122631073,
      "learning_rate": 1e-05,
      "loss": 0.9299,
      "step": 79505
    },
    {
      "epoch": 70.4251550044287,
      "grad_norm": 0.21958008408546448,
      "learning_rate": 1e-05,
      "loss": 0.896,
      "step": 79510
    },
    {
      "epoch": 70.4295837023915,
      "grad_norm": 0.23766876757144928,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 79515
    },
    {
      "epoch": 70.43401240035429,
      "grad_norm": 0.24751250445842743,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 79520
    },
    {
      "epoch": 70.4384410983171,
      "grad_norm": 0.2513173520565033,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 79525
    },
    {
      "epoch": 70.4428697962799,
      "grad_norm": 0.20737197995185852,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 79530
    },
    {
      "epoch": 70.44729849424269,
      "grad_norm": 0.2462579905986786,
      "learning_rate": 1e-05,
      "loss": 1.0375,
      "step": 79535
    },
    {
      "epoch": 70.4517271922055,
      "grad_norm": 0.2730797231197357,
      "learning_rate": 1e-05,
      "loss": 0.9333,
      "step": 79540
    },
    {
      "epoch": 70.45615589016829,
      "grad_norm": 0.20168955624103546,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 79545
    },
    {
      "epoch": 70.46058458813108,
      "grad_norm": 0.22944588959217072,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 79550
    },
    {
      "epoch": 70.46501328609389,
      "grad_norm": 0.24466437101364136,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 79555
    },
    {
      "epoch": 70.46944198405669,
      "grad_norm": 0.22957712411880493,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 79560
    },
    {
      "epoch": 70.47387068201948,
      "grad_norm": 0.2648143470287323,
      "learning_rate": 1e-05,
      "loss": 1.0275,
      "step": 79565
    },
    {
      "epoch": 70.47829937998229,
      "grad_norm": 0.2452268898487091,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 79570
    },
    {
      "epoch": 70.48272807794508,
      "grad_norm": 0.230984166264534,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 79575
    },
    {
      "epoch": 70.48715677590788,
      "grad_norm": 0.2508372962474823,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 79580
    },
    {
      "epoch": 70.49158547387069,
      "grad_norm": 0.2418755739927292,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 79585
    },
    {
      "epoch": 70.49601417183348,
      "grad_norm": 0.20538198947906494,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 79590
    },
    {
      "epoch": 70.50044286979627,
      "grad_norm": 0.22137463092803955,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 79595
    },
    {
      "epoch": 70.50487156775908,
      "grad_norm": 0.22290237247943878,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 79600
    },
    {
      "epoch": 70.50930026572188,
      "grad_norm": 0.23100435733795166,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 79605
    },
    {
      "epoch": 70.51372896368467,
      "grad_norm": 0.26087355613708496,
      "learning_rate": 1e-05,
      "loss": 0.9215,
      "step": 79610
    },
    {
      "epoch": 70.51815766164748,
      "grad_norm": 0.22224442660808563,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 79615
    },
    {
      "epoch": 70.52258635961027,
      "grad_norm": 0.2380107194185257,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 79620
    },
    {
      "epoch": 70.52701505757307,
      "grad_norm": 0.24505631625652313,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 79625
    },
    {
      "epoch": 70.53144375553588,
      "grad_norm": 0.22856011986732483,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 79630
    },
    {
      "epoch": 70.53587245349867,
      "grad_norm": 0.24574953317642212,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 79635
    },
    {
      "epoch": 70.54030115146146,
      "grad_norm": 0.22716130316257477,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 79640
    },
    {
      "epoch": 70.54472984942427,
      "grad_norm": 0.20567893981933594,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 79645
    },
    {
      "epoch": 70.54915854738707,
      "grad_norm": 0.23963038623332977,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 79650
    },
    {
      "epoch": 70.55358724534987,
      "grad_norm": 0.23295466601848602,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 79655
    },
    {
      "epoch": 70.55801594331267,
      "grad_norm": 0.26455336809158325,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 79660
    },
    {
      "epoch": 70.56244464127546,
      "grad_norm": 0.2628478705883026,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 79665
    },
    {
      "epoch": 70.56687333923827,
      "grad_norm": 0.22359447181224823,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 79670
    },
    {
      "epoch": 70.57130203720106,
      "grad_norm": 0.239747554063797,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 79675
    },
    {
      "epoch": 70.57573073516386,
      "grad_norm": 0.19292421638965607,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 79680
    },
    {
      "epoch": 70.58015943312667,
      "grad_norm": 0.24212707579135895,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 79685
    },
    {
      "epoch": 70.58458813108946,
      "grad_norm": 0.23006929457187653,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 79690
    },
    {
      "epoch": 70.58901682905226,
      "grad_norm": 0.32561323046684265,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 79695
    },
    {
      "epoch": 70.59344552701506,
      "grad_norm": 0.22944635152816772,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 79700
    },
    {
      "epoch": 70.59787422497786,
      "grad_norm": 0.2449757158756256,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 79705
    },
    {
      "epoch": 70.60230292294065,
      "grad_norm": 0.21088825166225433,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 79710
    },
    {
      "epoch": 70.60673162090346,
      "grad_norm": 0.2369798719882965,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 79715
    },
    {
      "epoch": 70.61116031886625,
      "grad_norm": 0.24787916243076324,
      "learning_rate": 1e-05,
      "loss": 0.9234,
      "step": 79720
    },
    {
      "epoch": 70.61558901682905,
      "grad_norm": 0.26056164503097534,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 79725
    },
    {
      "epoch": 70.62001771479186,
      "grad_norm": 0.20133395493030548,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 79730
    },
    {
      "epoch": 70.62444641275465,
      "grad_norm": 0.24059037864208221,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 79735
    },
    {
      "epoch": 70.62887511071744,
      "grad_norm": 0.2369968444108963,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 79740
    },
    {
      "epoch": 70.63330380868025,
      "grad_norm": 0.21638371050357819,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 79745
    },
    {
      "epoch": 70.63773250664305,
      "grad_norm": 0.19638800621032715,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 79750
    },
    {
      "epoch": 70.64216120460584,
      "grad_norm": 0.2876845896244049,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 79755
    },
    {
      "epoch": 70.64658990256865,
      "grad_norm": 0.25584158301353455,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 79760
    },
    {
      "epoch": 70.65101860053144,
      "grad_norm": 0.24557648599147797,
      "learning_rate": 1e-05,
      "loss": 1.0226,
      "step": 79765
    },
    {
      "epoch": 70.65544729849424,
      "grad_norm": 0.2286657691001892,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 79770
    },
    {
      "epoch": 70.65987599645705,
      "grad_norm": 0.23045624792575836,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 79775
    },
    {
      "epoch": 70.66430469441984,
      "grad_norm": 0.2590855360031128,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 79780
    },
    {
      "epoch": 70.66873339238263,
      "grad_norm": 0.22924913465976715,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 79785
    },
    {
      "epoch": 70.67316209034544,
      "grad_norm": 0.26343294978141785,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 79790
    },
    {
      "epoch": 70.67759078830824,
      "grad_norm": 0.23316098749637604,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 79795
    },
    {
      "epoch": 70.68201948627103,
      "grad_norm": 0.2377476692199707,
      "learning_rate": 1e-05,
      "loss": 0.9381,
      "step": 79800
    },
    {
      "epoch": 70.68644818423384,
      "grad_norm": 0.2573939859867096,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 79805
    },
    {
      "epoch": 70.69087688219663,
      "grad_norm": 0.243405282497406,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 79810
    },
    {
      "epoch": 70.69530558015943,
      "grad_norm": 0.2553041875362396,
      "learning_rate": 1e-05,
      "loss": 0.8978,
      "step": 79815
    },
    {
      "epoch": 70.69973427812224,
      "grad_norm": 0.24558883905410767,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 79820
    },
    {
      "epoch": 70.70416297608503,
      "grad_norm": 0.21875348687171936,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 79825
    },
    {
      "epoch": 70.70859167404782,
      "grad_norm": 0.23658457398414612,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 79830
    },
    {
      "epoch": 70.71302037201063,
      "grad_norm": 0.22629718482494354,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 79835
    },
    {
      "epoch": 70.71744906997343,
      "grad_norm": 0.25328344106674194,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 79840
    },
    {
      "epoch": 70.72187776793622,
      "grad_norm": 0.22597786784172058,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 79845
    },
    {
      "epoch": 70.72630646589903,
      "grad_norm": 0.23131953179836273,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 79850
    },
    {
      "epoch": 70.73073516386182,
      "grad_norm": 0.2347148358821869,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 79855
    },
    {
      "epoch": 70.73516386182462,
      "grad_norm": 0.23361410200595856,
      "learning_rate": 1e-05,
      "loss": 0.9001,
      "step": 79860
    },
    {
      "epoch": 70.73959255978743,
      "grad_norm": 0.2147773951292038,
      "learning_rate": 1e-05,
      "loss": 0.9139,
      "step": 79865
    },
    {
      "epoch": 70.74402125775022,
      "grad_norm": 0.22081850469112396,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 79870
    },
    {
      "epoch": 70.74844995571301,
      "grad_norm": 0.21614603698253632,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 79875
    },
    {
      "epoch": 70.75287865367582,
      "grad_norm": 0.22959750890731812,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 79880
    },
    {
      "epoch": 70.75730735163862,
      "grad_norm": 0.23026128113269806,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 79885
    },
    {
      "epoch": 70.76173604960141,
      "grad_norm": 0.24133040010929108,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 79890
    },
    {
      "epoch": 70.76616474756422,
      "grad_norm": 0.20410820841789246,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 79895
    },
    {
      "epoch": 70.77059344552701,
      "grad_norm": 0.25879955291748047,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 79900
    },
    {
      "epoch": 70.77502214348982,
      "grad_norm": 0.23236370086669922,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 79905
    },
    {
      "epoch": 70.77945084145261,
      "grad_norm": 0.269452840089798,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 79910
    },
    {
      "epoch": 70.78387953941541,
      "grad_norm": 0.22299674153327942,
      "learning_rate": 1e-05,
      "loss": 0.9214,
      "step": 79915
    },
    {
      "epoch": 70.78830823737822,
      "grad_norm": 0.24492210149765015,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 79920
    },
    {
      "epoch": 70.79273693534101,
      "grad_norm": 0.2115638256072998,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 79925
    },
    {
      "epoch": 70.7971656333038,
      "grad_norm": 0.25674957036972046,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 79930
    },
    {
      "epoch": 70.80159433126661,
      "grad_norm": 0.23233728110790253,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 79935
    },
    {
      "epoch": 70.80602302922941,
      "grad_norm": 0.22099140286445618,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 79940
    },
    {
      "epoch": 70.8104517271922,
      "grad_norm": 0.23403653502464294,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 79945
    },
    {
      "epoch": 70.81488042515501,
      "grad_norm": 0.22350962460041046,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 79950
    },
    {
      "epoch": 70.8193091231178,
      "grad_norm": 0.24749670922756195,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 79955
    },
    {
      "epoch": 70.8237378210806,
      "grad_norm": 0.2343074530363083,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 79960
    },
    {
      "epoch": 70.8281665190434,
      "grad_norm": 0.2226838916540146,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 79965
    },
    {
      "epoch": 70.8325952170062,
      "grad_norm": 0.34486261010169983,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 79970
    },
    {
      "epoch": 70.837023914969,
      "grad_norm": 0.25514116883277893,
      "learning_rate": 1e-05,
      "loss": 0.9294,
      "step": 79975
    },
    {
      "epoch": 70.8414526129318,
      "grad_norm": 0.2079382687807083,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 79980
    },
    {
      "epoch": 70.8458813108946,
      "grad_norm": 0.24633552134037018,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 79985
    },
    {
      "epoch": 70.85031000885739,
      "grad_norm": 0.21729297935962677,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 79990
    },
    {
      "epoch": 70.8547387068202,
      "grad_norm": 0.21760140359401703,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 79995
    },
    {
      "epoch": 70.859167404783,
      "grad_norm": 0.2608988285064697,
      "learning_rate": 1e-05,
      "loss": 1.0259,
      "step": 80000
    },
    {
      "epoch": 70.86359610274579,
      "grad_norm": 0.2253909558057785,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 80005
    },
    {
      "epoch": 70.8680248007086,
      "grad_norm": 0.2454213798046112,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 80010
    },
    {
      "epoch": 70.87245349867139,
      "grad_norm": 0.235086590051651,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 80015
    },
    {
      "epoch": 70.87688219663418,
      "grad_norm": 0.23786352574825287,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 80020
    },
    {
      "epoch": 70.881310894597,
      "grad_norm": 0.21125487983226776,
      "learning_rate": 1e-05,
      "loss": 0.8755,
      "step": 80025
    },
    {
      "epoch": 70.88573959255979,
      "grad_norm": 0.23355448246002197,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 80030
    },
    {
      "epoch": 70.89016829052258,
      "grad_norm": 0.24758510291576385,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 80035
    },
    {
      "epoch": 70.89459698848539,
      "grad_norm": 0.2557387351989746,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 80040
    },
    {
      "epoch": 70.89902568644818,
      "grad_norm": 0.24200406670570374,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 80045
    },
    {
      "epoch": 70.90345438441098,
      "grad_norm": 0.2286711484193802,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 80050
    },
    {
      "epoch": 70.90788308237379,
      "grad_norm": 0.2260790765285492,
      "learning_rate": 1e-05,
      "loss": 0.9189,
      "step": 80055
    },
    {
      "epoch": 70.91231178033658,
      "grad_norm": 0.23960740864276886,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 80060
    },
    {
      "epoch": 70.91674047829937,
      "grad_norm": 0.2657438814640045,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 80065
    },
    {
      "epoch": 70.92116917626218,
      "grad_norm": 0.2587682604789734,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 80070
    },
    {
      "epoch": 70.92559787422498,
      "grad_norm": 0.22517229616641998,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 80075
    },
    {
      "epoch": 70.93002657218777,
      "grad_norm": 0.29738885164260864,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 80080
    },
    {
      "epoch": 70.93445527015058,
      "grad_norm": 0.2340104728937149,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 80085
    },
    {
      "epoch": 70.93888396811337,
      "grad_norm": 0.2501141428947449,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 80090
    },
    {
      "epoch": 70.94331266607617,
      "grad_norm": 0.22814619541168213,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 80095
    },
    {
      "epoch": 70.94774136403898,
      "grad_norm": 0.27238234877586365,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 80100
    },
    {
      "epoch": 70.95217006200177,
      "grad_norm": 0.2445191890001297,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 80105
    },
    {
      "epoch": 70.95659875996456,
      "grad_norm": 0.23224687576293945,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 80110
    },
    {
      "epoch": 70.96102745792737,
      "grad_norm": 0.23445631563663483,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 80115
    },
    {
      "epoch": 70.96545615589017,
      "grad_norm": 0.26628297567367554,
      "learning_rate": 1e-05,
      "loss": 1.028,
      "step": 80120
    },
    {
      "epoch": 70.96988485385296,
      "grad_norm": 0.3078608512878418,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 80125
    },
    {
      "epoch": 70.97431355181577,
      "grad_norm": 0.2374720424413681,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 80130
    },
    {
      "epoch": 70.97874224977856,
      "grad_norm": 0.31722986698150635,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 80135
    },
    {
      "epoch": 70.98317094774137,
      "grad_norm": 0.2344239056110382,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 80140
    },
    {
      "epoch": 70.98759964570417,
      "grad_norm": 0.2642884850502014,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 80145
    },
    {
      "epoch": 70.99202834366696,
      "grad_norm": 0.2550107538700104,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 80150
    },
    {
      "epoch": 70.99645704162977,
      "grad_norm": 0.2752811014652252,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 80155
    },
    {
      "epoch": 71.00088573959256,
      "grad_norm": 0.2818542420864105,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 80160
    },
    {
      "epoch": 71.00531443755536,
      "grad_norm": 0.2628529369831085,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 80165
    },
    {
      "epoch": 71.00974313551816,
      "grad_norm": 0.2648138999938965,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 80170
    },
    {
      "epoch": 71.01417183348096,
      "grad_norm": 0.21259281039237976,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 80175
    },
    {
      "epoch": 71.01860053144375,
      "grad_norm": 0.239787757396698,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 80180
    },
    {
      "epoch": 71.02302922940656,
      "grad_norm": 0.24108575284481049,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 80185
    },
    {
      "epoch": 71.02745792736935,
      "grad_norm": 0.2325608879327774,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 80190
    },
    {
      "epoch": 71.03188662533215,
      "grad_norm": 0.26507678627967834,
      "learning_rate": 1e-05,
      "loss": 0.9217,
      "step": 80195
    },
    {
      "epoch": 71.03631532329496,
      "grad_norm": 0.23155640065670013,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 80200
    },
    {
      "epoch": 71.04074402125775,
      "grad_norm": 0.221502885222435,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 80205
    },
    {
      "epoch": 71.04517271922055,
      "grad_norm": 0.2519630193710327,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 80210
    },
    {
      "epoch": 71.04960141718335,
      "grad_norm": 0.22802503407001495,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 80215
    },
    {
      "epoch": 71.05403011514615,
      "grad_norm": 0.21995733678340912,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 80220
    },
    {
      "epoch": 71.05845881310894,
      "grad_norm": 0.23330344259738922,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 80225
    },
    {
      "epoch": 71.06288751107175,
      "grad_norm": 0.20638588070869446,
      "learning_rate": 1e-05,
      "loss": 0.925,
      "step": 80230
    },
    {
      "epoch": 71.06731620903454,
      "grad_norm": 0.2572697401046753,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 80235
    },
    {
      "epoch": 71.07174490699734,
      "grad_norm": 0.2319895625114441,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 80240
    },
    {
      "epoch": 71.07617360496015,
      "grad_norm": 0.27966466546058655,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 80245
    },
    {
      "epoch": 71.08060230292294,
      "grad_norm": 0.2795937955379486,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 80250
    },
    {
      "epoch": 71.08503100088573,
      "grad_norm": 0.27428826689720154,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 80255
    },
    {
      "epoch": 71.08945969884854,
      "grad_norm": 0.2572523355484009,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 80260
    },
    {
      "epoch": 71.09388839681134,
      "grad_norm": 0.2241087406873703,
      "learning_rate": 1e-05,
      "loss": 1.0161,
      "step": 80265
    },
    {
      "epoch": 71.09831709477413,
      "grad_norm": 0.2676275670528412,
      "learning_rate": 1e-05,
      "loss": 1.0268,
      "step": 80270
    },
    {
      "epoch": 71.10274579273694,
      "grad_norm": 0.275318443775177,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 80275
    },
    {
      "epoch": 71.10717449069973,
      "grad_norm": 0.2725452184677124,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 80280
    },
    {
      "epoch": 71.11160318866253,
      "grad_norm": 0.23706059157848358,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 80285
    },
    {
      "epoch": 71.11603188662534,
      "grad_norm": 0.261922150850296,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 80290
    },
    {
      "epoch": 71.12046058458813,
      "grad_norm": 0.26080769300460815,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 80295
    },
    {
      "epoch": 71.12488928255092,
      "grad_norm": 0.25540903210639954,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 80300
    },
    {
      "epoch": 71.12931798051373,
      "grad_norm": 0.25440841913223267,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 80305
    },
    {
      "epoch": 71.13374667847653,
      "grad_norm": 0.22302967309951782,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 80310
    },
    {
      "epoch": 71.13817537643932,
      "grad_norm": 0.26216715574264526,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 80315
    },
    {
      "epoch": 71.14260407440213,
      "grad_norm": 0.2351176142692566,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 80320
    },
    {
      "epoch": 71.14703277236492,
      "grad_norm": 0.27764636278152466,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 80325
    },
    {
      "epoch": 71.15146147032772,
      "grad_norm": 0.26274585723876953,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 80330
    },
    {
      "epoch": 71.15589016829053,
      "grad_norm": 0.2509876489639282,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 80335
    },
    {
      "epoch": 71.16031886625332,
      "grad_norm": 0.23072203993797302,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 80340
    },
    {
      "epoch": 71.16474756421611,
      "grad_norm": 0.21615734696388245,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 80345
    },
    {
      "epoch": 71.16917626217892,
      "grad_norm": 0.25847092270851135,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 80350
    },
    {
      "epoch": 71.17360496014172,
      "grad_norm": 0.24947138130664825,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 80355
    },
    {
      "epoch": 71.17803365810451,
      "grad_norm": 0.2748432755470276,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 80360
    },
    {
      "epoch": 71.18246235606732,
      "grad_norm": 0.24430978298187256,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 80365
    },
    {
      "epoch": 71.18689105403011,
      "grad_norm": 0.22269472479820251,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 80370
    },
    {
      "epoch": 71.1913197519929,
      "grad_norm": 0.24195894598960876,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 80375
    },
    {
      "epoch": 71.19574844995572,
      "grad_norm": 0.23681263625621796,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 80380
    },
    {
      "epoch": 71.20017714791851,
      "grad_norm": 0.2955668270587921,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 80385
    },
    {
      "epoch": 71.20460584588132,
      "grad_norm": 0.25044575333595276,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 80390
    },
    {
      "epoch": 71.20903454384411,
      "grad_norm": 0.24081918597221375,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 80395
    },
    {
      "epoch": 71.2134632418069,
      "grad_norm": 0.20423902571201324,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 80400
    },
    {
      "epoch": 71.21789193976971,
      "grad_norm": 0.23293322324752808,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 80405
    },
    {
      "epoch": 71.22232063773251,
      "grad_norm": 0.23395177721977234,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 80410
    },
    {
      "epoch": 71.2267493356953,
      "grad_norm": 0.26237043738365173,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 80415
    },
    {
      "epoch": 71.23117803365811,
      "grad_norm": 0.24904310703277588,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 80420
    },
    {
      "epoch": 71.2356067316209,
      "grad_norm": 0.21965089440345764,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 80425
    },
    {
      "epoch": 71.2400354295837,
      "grad_norm": 0.22282753884792328,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 80430
    },
    {
      "epoch": 71.24446412754651,
      "grad_norm": 0.26240798830986023,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 80435
    },
    {
      "epoch": 71.2488928255093,
      "grad_norm": 0.2755425274372101,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 80440
    },
    {
      "epoch": 71.2533215234721,
      "grad_norm": 0.22789554297924042,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 80445
    },
    {
      "epoch": 71.2577502214349,
      "grad_norm": 0.24028095602989197,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 80450
    },
    {
      "epoch": 71.2621789193977,
      "grad_norm": 0.24441474676132202,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 80455
    },
    {
      "epoch": 71.26660761736049,
      "grad_norm": 0.20573052763938904,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 80460
    },
    {
      "epoch": 71.2710363153233,
      "grad_norm": 0.24030780792236328,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 80465
    },
    {
      "epoch": 71.2754650132861,
      "grad_norm": 0.23251719772815704,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 80470
    },
    {
      "epoch": 71.27989371124889,
      "grad_norm": 0.23238717019557953,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 80475
    },
    {
      "epoch": 71.2843224092117,
      "grad_norm": 0.21438805758953094,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 80480
    },
    {
      "epoch": 71.28875110717449,
      "grad_norm": 0.21263061463832855,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 80485
    },
    {
      "epoch": 71.29317980513729,
      "grad_norm": 0.24810579419136047,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 80490
    },
    {
      "epoch": 71.2976085031001,
      "grad_norm": 0.2711457908153534,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 80495
    },
    {
      "epoch": 71.30203720106289,
      "grad_norm": 0.24710458517074585,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 80500
    },
    {
      "epoch": 71.30646589902568,
      "grad_norm": 0.23652152717113495,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 80505
    },
    {
      "epoch": 71.31089459698849,
      "grad_norm": 0.23710598051548004,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 80510
    },
    {
      "epoch": 71.31532329495128,
      "grad_norm": 0.22188624739646912,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 80515
    },
    {
      "epoch": 71.31975199291408,
      "grad_norm": 0.26686355471611023,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 80520
    },
    {
      "epoch": 71.32418069087689,
      "grad_norm": 0.251725435256958,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 80525
    },
    {
      "epoch": 71.32860938883968,
      "grad_norm": 0.2258744090795517,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 80530
    },
    {
      "epoch": 71.33303808680247,
      "grad_norm": 0.2634340524673462,
      "learning_rate": 1e-05,
      "loss": 0.93,
      "step": 80535
    },
    {
      "epoch": 71.33746678476528,
      "grad_norm": 0.26706716418266296,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 80540
    },
    {
      "epoch": 71.34189548272808,
      "grad_norm": 0.2580114006996155,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 80545
    },
    {
      "epoch": 71.34632418069087,
      "grad_norm": 0.23479372262954712,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 80550
    },
    {
      "epoch": 71.35075287865368,
      "grad_norm": 0.2756200432777405,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 80555
    },
    {
      "epoch": 71.35518157661647,
      "grad_norm": 0.27503982186317444,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 80560
    },
    {
      "epoch": 71.35961027457927,
      "grad_norm": 0.23908555507659912,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 80565
    },
    {
      "epoch": 71.36403897254208,
      "grad_norm": 0.2406509816646576,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 80570
    },
    {
      "epoch": 71.36846767050487,
      "grad_norm": 0.32873573899269104,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 80575
    },
    {
      "epoch": 71.37289636846766,
      "grad_norm": 0.20651979744434357,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 80580
    },
    {
      "epoch": 71.37732506643047,
      "grad_norm": 0.23700939118862152,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 80585
    },
    {
      "epoch": 71.38175376439327,
      "grad_norm": 0.258241206407547,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 80590
    },
    {
      "epoch": 71.38618246235606,
      "grad_norm": 0.24222588539123535,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 80595
    },
    {
      "epoch": 71.39061116031887,
      "grad_norm": 0.2494470477104187,
      "learning_rate": 1e-05,
      "loss": 0.8667,
      "step": 80600
    },
    {
      "epoch": 71.39503985828166,
      "grad_norm": 0.23111043870449066,
      "learning_rate": 1e-05,
      "loss": 0.9052,
      "step": 80605
    },
    {
      "epoch": 71.39946855624446,
      "grad_norm": 0.27360033988952637,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 80610
    },
    {
      "epoch": 71.40389725420727,
      "grad_norm": 0.23422251641750336,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 80615
    },
    {
      "epoch": 71.40832595217006,
      "grad_norm": 0.24004213511943817,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 80620
    },
    {
      "epoch": 71.41275465013285,
      "grad_norm": 0.34923917055130005,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 80625
    },
    {
      "epoch": 71.41718334809566,
      "grad_norm": 0.23606614768505096,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 80630
    },
    {
      "epoch": 71.42161204605846,
      "grad_norm": 0.22275462746620178,
      "learning_rate": 1e-05,
      "loss": 0.9227,
      "step": 80635
    },
    {
      "epoch": 71.42604074402126,
      "grad_norm": 0.23076629638671875,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 80640
    },
    {
      "epoch": 71.43046944198406,
      "grad_norm": 0.2563914954662323,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 80645
    },
    {
      "epoch": 71.43489813994685,
      "grad_norm": 0.2539900839328766,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 80650
    },
    {
      "epoch": 71.43932683790966,
      "grad_norm": 0.20931419730186462,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 80655
    },
    {
      "epoch": 71.44375553587246,
      "grad_norm": 0.23672591149806976,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 80660
    },
    {
      "epoch": 71.44818423383525,
      "grad_norm": 0.2351336032152176,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 80665
    },
    {
      "epoch": 71.45261293179806,
      "grad_norm": 0.22419917583465576,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 80670
    },
    {
      "epoch": 71.45704162976085,
      "grad_norm": 0.2815806269645691,
      "learning_rate": 1e-05,
      "loss": 0.9432,
      "step": 80675
    },
    {
      "epoch": 71.46147032772365,
      "grad_norm": 0.25801798701286316,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 80680
    },
    {
      "epoch": 71.46589902568645,
      "grad_norm": 0.24311938881874084,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 80685
    },
    {
      "epoch": 71.47032772364925,
      "grad_norm": 0.2610177993774414,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 80690
    },
    {
      "epoch": 71.47475642161204,
      "grad_norm": 0.23146659135818481,
      "learning_rate": 1e-05,
      "loss": 1.0054,
      "step": 80695
    },
    {
      "epoch": 71.47918511957485,
      "grad_norm": 0.2188626527786255,
      "learning_rate": 1e-05,
      "loss": 0.9289,
      "step": 80700
    },
    {
      "epoch": 71.48361381753764,
      "grad_norm": 0.23460035026073456,
      "learning_rate": 1e-05,
      "loss": 0.9153,
      "step": 80705
    },
    {
      "epoch": 71.48804251550044,
      "grad_norm": 0.23641106486320496,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 80710
    },
    {
      "epoch": 71.49247121346325,
      "grad_norm": 0.24951107800006866,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 80715
    },
    {
      "epoch": 71.49689991142604,
      "grad_norm": 0.2044644057750702,
      "learning_rate": 1e-05,
      "loss": 0.9201,
      "step": 80720
    },
    {
      "epoch": 71.50132860938884,
      "grad_norm": 0.2317964881658554,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 80725
    },
    {
      "epoch": 71.50575730735164,
      "grad_norm": 0.21704602241516113,
      "learning_rate": 1e-05,
      "loss": 0.933,
      "step": 80730
    },
    {
      "epoch": 71.51018600531444,
      "grad_norm": 0.22970236837863922,
      "learning_rate": 1e-05,
      "loss": 1.0101,
      "step": 80735
    },
    {
      "epoch": 71.51461470327723,
      "grad_norm": 0.21163766086101532,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 80740
    },
    {
      "epoch": 71.51904340124004,
      "grad_norm": 0.21668152511119843,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 80745
    },
    {
      "epoch": 71.52347209920283,
      "grad_norm": 0.2421610802412033,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 80750
    },
    {
      "epoch": 71.52790079716563,
      "grad_norm": 0.20976890623569489,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 80755
    },
    {
      "epoch": 71.53232949512844,
      "grad_norm": 0.2770538032054901,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 80760
    },
    {
      "epoch": 71.53675819309123,
      "grad_norm": 0.22088707983493805,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 80765
    },
    {
      "epoch": 71.54118689105402,
      "grad_norm": 0.22720152139663696,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 80770
    },
    {
      "epoch": 71.54561558901683,
      "grad_norm": 0.21053992211818695,
      "learning_rate": 1e-05,
      "loss": 0.9165,
      "step": 80775
    },
    {
      "epoch": 71.55004428697963,
      "grad_norm": 0.2627178430557251,
      "learning_rate": 1e-05,
      "loss": 1.0406,
      "step": 80780
    },
    {
      "epoch": 71.55447298494242,
      "grad_norm": 0.20366695523262024,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 80785
    },
    {
      "epoch": 71.55890168290523,
      "grad_norm": 0.26274093985557556,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 80790
    },
    {
      "epoch": 71.56333038086802,
      "grad_norm": 0.24148856103420258,
      "learning_rate": 1e-05,
      "loss": 1.0237,
      "step": 80795
    },
    {
      "epoch": 71.56775907883082,
      "grad_norm": 0.21823592483997345,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 80800
    },
    {
      "epoch": 71.57218777679363,
      "grad_norm": 0.2478497177362442,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 80805
    },
    {
      "epoch": 71.57661647475642,
      "grad_norm": 0.2173428237438202,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 80810
    },
    {
      "epoch": 71.58104517271921,
      "grad_norm": 0.22447757422924042,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 80815
    },
    {
      "epoch": 71.58547387068202,
      "grad_norm": 0.23631472885608673,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 80820
    },
    {
      "epoch": 71.58990256864482,
      "grad_norm": 0.25925299525260925,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 80825
    },
    {
      "epoch": 71.59433126660761,
      "grad_norm": 0.25300338864326477,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 80830
    },
    {
      "epoch": 71.59875996457042,
      "grad_norm": 0.30748093128204346,
      "learning_rate": 1e-05,
      "loss": 0.9242,
      "step": 80835
    },
    {
      "epoch": 71.60318866253321,
      "grad_norm": 0.27988728880882263,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 80840
    },
    {
      "epoch": 71.60761736049601,
      "grad_norm": 0.2745327949523926,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 80845
    },
    {
      "epoch": 71.61204605845882,
      "grad_norm": 0.22556182742118835,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 80850
    },
    {
      "epoch": 71.61647475642161,
      "grad_norm": 0.2211562991142273,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 80855
    },
    {
      "epoch": 71.6209034543844,
      "grad_norm": 0.28095027804374695,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 80860
    },
    {
      "epoch": 71.62533215234721,
      "grad_norm": 0.21826890110969543,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 80865
    },
    {
      "epoch": 71.62976085031,
      "grad_norm": 0.20751270651817322,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 80870
    },
    {
      "epoch": 71.63418954827281,
      "grad_norm": 0.22506152093410492,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 80875
    },
    {
      "epoch": 71.63861824623561,
      "grad_norm": 0.2570043206214905,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 80880
    },
    {
      "epoch": 71.6430469441984,
      "grad_norm": 0.26006948947906494,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 80885
    },
    {
      "epoch": 71.64747564216121,
      "grad_norm": 0.2555810809135437,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 80890
    },
    {
      "epoch": 71.651904340124,
      "grad_norm": 0.21721920371055603,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 80895
    },
    {
      "epoch": 71.6563330380868,
      "grad_norm": 0.2691342234611511,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 80900
    },
    {
      "epoch": 71.66076173604961,
      "grad_norm": 0.28396010398864746,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 80905
    },
    {
      "epoch": 71.6651904340124,
      "grad_norm": 0.27304667234420776,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 80910
    },
    {
      "epoch": 71.6696191319752,
      "grad_norm": 0.2724715769290924,
      "learning_rate": 1e-05,
      "loss": 0.9192,
      "step": 80915
    },
    {
      "epoch": 71.674047829938,
      "grad_norm": 0.27306482195854187,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 80920
    },
    {
      "epoch": 71.6784765279008,
      "grad_norm": 0.2507292926311493,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 80925
    },
    {
      "epoch": 71.68290522586359,
      "grad_norm": 0.23401039838790894,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 80930
    },
    {
      "epoch": 71.6873339238264,
      "grad_norm": 0.26273906230926514,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 80935
    },
    {
      "epoch": 71.6917626217892,
      "grad_norm": 0.3119945824146271,
      "learning_rate": 1e-05,
      "loss": 1.022,
      "step": 80940
    },
    {
      "epoch": 71.69619131975199,
      "grad_norm": 0.23534446954727173,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 80945
    },
    {
      "epoch": 71.7006200177148,
      "grad_norm": 0.25379493832588196,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 80950
    },
    {
      "epoch": 71.70504871567759,
      "grad_norm": 0.2314881533384323,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 80955
    },
    {
      "epoch": 71.70947741364039,
      "grad_norm": 0.2233014851808548,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 80960
    },
    {
      "epoch": 71.7139061116032,
      "grad_norm": 0.21803206205368042,
      "learning_rate": 1e-05,
      "loss": 0.8981,
      "step": 80965
    },
    {
      "epoch": 71.71833480956599,
      "grad_norm": 0.23618222773075104,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 80970
    },
    {
      "epoch": 71.72276350752878,
      "grad_norm": 0.26837074756622314,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 80975
    },
    {
      "epoch": 71.72719220549159,
      "grad_norm": 0.2029809057712555,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 80980
    },
    {
      "epoch": 71.73162090345438,
      "grad_norm": 0.21906058490276337,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 80985
    },
    {
      "epoch": 71.73604960141718,
      "grad_norm": 0.24671000242233276,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 80990
    },
    {
      "epoch": 71.74047829937999,
      "grad_norm": 0.24907661974430084,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 80995
    },
    {
      "epoch": 71.74490699734278,
      "grad_norm": 0.21554157137870789,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 81000
    },
    {
      "epoch": 71.74933569530558,
      "grad_norm": 0.22353962063789368,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 81005
    },
    {
      "epoch": 71.75376439326838,
      "grad_norm": 0.22581346333026886,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 81010
    },
    {
      "epoch": 71.75819309123118,
      "grad_norm": 0.2615215480327606,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 81015
    },
    {
      "epoch": 71.76262178919397,
      "grad_norm": 0.24465018510818481,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 81020
    },
    {
      "epoch": 71.76705048715678,
      "grad_norm": 0.2122029960155487,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 81025
    },
    {
      "epoch": 71.77147918511957,
      "grad_norm": 0.23267334699630737,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 81030
    },
    {
      "epoch": 71.77590788308237,
      "grad_norm": 0.19814740121364594,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 81035
    },
    {
      "epoch": 71.78033658104518,
      "grad_norm": 0.27602100372314453,
      "learning_rate": 1e-05,
      "loss": 0.9187,
      "step": 81040
    },
    {
      "epoch": 71.78476527900797,
      "grad_norm": 0.2953493893146515,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 81045
    },
    {
      "epoch": 71.78919397697076,
      "grad_norm": 0.24956394731998444,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 81050
    },
    {
      "epoch": 71.79362267493357,
      "grad_norm": 0.2468077540397644,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 81055
    },
    {
      "epoch": 71.79805137289637,
      "grad_norm": 0.2389238029718399,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 81060
    },
    {
      "epoch": 71.80248007085916,
      "grad_norm": 0.2530921697616577,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 81065
    },
    {
      "epoch": 71.80690876882197,
      "grad_norm": 0.22603349387645721,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 81070
    },
    {
      "epoch": 71.81133746678476,
      "grad_norm": 0.230739027261734,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 81075
    },
    {
      "epoch": 71.81576616474756,
      "grad_norm": 0.2293829619884491,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 81080
    },
    {
      "epoch": 71.82019486271037,
      "grad_norm": 0.2462838739156723,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 81085
    },
    {
      "epoch": 71.82462356067316,
      "grad_norm": 0.2447291910648346,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 81090
    },
    {
      "epoch": 71.82905225863595,
      "grad_norm": 0.22890999913215637,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 81095
    },
    {
      "epoch": 71.83348095659876,
      "grad_norm": 0.24218162894248962,
      "learning_rate": 1e-05,
      "loss": 0.924,
      "step": 81100
    },
    {
      "epoch": 71.83790965456156,
      "grad_norm": 0.23259367048740387,
      "learning_rate": 1e-05,
      "loss": 0.9103,
      "step": 81105
    },
    {
      "epoch": 71.84233835252435,
      "grad_norm": 0.21914659440517426,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 81110
    },
    {
      "epoch": 71.84676705048716,
      "grad_norm": 0.2505854368209839,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 81115
    },
    {
      "epoch": 71.85119574844995,
      "grad_norm": 0.2469121366739273,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 81120
    },
    {
      "epoch": 71.85562444641276,
      "grad_norm": 0.2482776790857315,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 81125
    },
    {
      "epoch": 71.86005314437556,
      "grad_norm": 0.2741515338420868,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 81130
    },
    {
      "epoch": 71.86448184233835,
      "grad_norm": 0.22157226502895355,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 81135
    },
    {
      "epoch": 71.86891054030116,
      "grad_norm": 0.22523397207260132,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 81140
    },
    {
      "epoch": 71.87333923826395,
      "grad_norm": 0.2368723601102829,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 81145
    },
    {
      "epoch": 71.87776793622675,
      "grad_norm": 0.25687870383262634,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 81150
    },
    {
      "epoch": 71.88219663418955,
      "grad_norm": 0.22510018944740295,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 81155
    },
    {
      "epoch": 71.88662533215235,
      "grad_norm": 0.2791641354560852,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 81160
    },
    {
      "epoch": 71.89105403011514,
      "grad_norm": 0.27973106503486633,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 81165
    },
    {
      "epoch": 71.89548272807795,
      "grad_norm": 0.2402273565530777,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 81170
    },
    {
      "epoch": 71.89991142604075,
      "grad_norm": 0.2389194816350937,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 81175
    },
    {
      "epoch": 71.90434012400354,
      "grad_norm": 0.21920032799243927,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 81180
    },
    {
      "epoch": 71.90876882196635,
      "grad_norm": 0.2521321773529053,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 81185
    },
    {
      "epoch": 71.91319751992914,
      "grad_norm": 0.26153382658958435,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 81190
    },
    {
      "epoch": 71.91762621789194,
      "grad_norm": 0.22406961023807526,
      "learning_rate": 1e-05,
      "loss": 1.0196,
      "step": 81195
    },
    {
      "epoch": 71.92205491585474,
      "grad_norm": 0.22841869294643402,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 81200
    },
    {
      "epoch": 71.92648361381754,
      "grad_norm": 0.2409314215183258,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 81205
    },
    {
      "epoch": 71.93091231178033,
      "grad_norm": 0.22601507604122162,
      "learning_rate": 1e-05,
      "loss": 1.0239,
      "step": 81210
    },
    {
      "epoch": 71.93534100974314,
      "grad_norm": 0.24212512373924255,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 81215
    },
    {
      "epoch": 71.93976970770593,
      "grad_norm": 0.25856292247772217,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 81220
    },
    {
      "epoch": 71.94419840566873,
      "grad_norm": 0.24418391287326813,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 81225
    },
    {
      "epoch": 71.94862710363154,
      "grad_norm": 0.2174445539712906,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 81230
    },
    {
      "epoch": 71.95305580159433,
      "grad_norm": 0.23913152515888214,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 81235
    },
    {
      "epoch": 71.95748449955713,
      "grad_norm": 0.20793089270591736,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 81240
    },
    {
      "epoch": 71.96191319751993,
      "grad_norm": 0.21642142534255981,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 81245
    },
    {
      "epoch": 71.96634189548273,
      "grad_norm": 0.25521984696388245,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 81250
    },
    {
      "epoch": 71.97077059344552,
      "grad_norm": 0.24558059871196747,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 81255
    },
    {
      "epoch": 71.97519929140833,
      "grad_norm": 0.2338753193616867,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 81260
    },
    {
      "epoch": 71.97962798937112,
      "grad_norm": 0.24042771756649017,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 81265
    },
    {
      "epoch": 71.98405668733392,
      "grad_norm": 0.24682267010211945,
      "learning_rate": 1e-05,
      "loss": 0.9124,
      "step": 81270
    },
    {
      "epoch": 71.98848538529673,
      "grad_norm": 0.23481950163841248,
      "learning_rate": 1e-05,
      "loss": 0.9051,
      "step": 81275
    },
    {
      "epoch": 71.99291408325952,
      "grad_norm": 0.2177044153213501,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 81280
    },
    {
      "epoch": 71.99734278122232,
      "grad_norm": 0.2158825844526291,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 81285
    },
    {
      "epoch": 72.00177147918512,
      "grad_norm": 0.23622381687164307,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 81290
    },
    {
      "epoch": 72.00620017714792,
      "grad_norm": 0.21890859305858612,
      "learning_rate": 1e-05,
      "loss": 0.9202,
      "step": 81295
    },
    {
      "epoch": 72.01062887511071,
      "grad_norm": 0.242213174700737,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 81300
    },
    {
      "epoch": 72.01505757307352,
      "grad_norm": 0.27944111824035645,
      "learning_rate": 1e-05,
      "loss": 0.9156,
      "step": 81305
    },
    {
      "epoch": 72.01948627103631,
      "grad_norm": 0.23651334643363953,
      "learning_rate": 1e-05,
      "loss": 0.9052,
      "step": 81310
    },
    {
      "epoch": 72.02391496899911,
      "grad_norm": 0.22058548033237457,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 81315
    },
    {
      "epoch": 72.02834366696192,
      "grad_norm": 0.25605371594429016,
      "learning_rate": 1e-05,
      "loss": 0.9252,
      "step": 81320
    },
    {
      "epoch": 72.03277236492471,
      "grad_norm": 0.22399884462356567,
      "learning_rate": 1e-05,
      "loss": 0.9335,
      "step": 81325
    },
    {
      "epoch": 72.0372010628875,
      "grad_norm": 0.2504664361476898,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 81330
    },
    {
      "epoch": 72.04162976085031,
      "grad_norm": 0.23969127237796783,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 81335
    },
    {
      "epoch": 72.0460584588131,
      "grad_norm": 0.24141551554203033,
      "learning_rate": 1e-05,
      "loss": 0.9293,
      "step": 81340
    },
    {
      "epoch": 72.0504871567759,
      "grad_norm": 0.24725572764873505,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 81345
    },
    {
      "epoch": 72.05491585473871,
      "grad_norm": 0.20956510305404663,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 81350
    },
    {
      "epoch": 72.0593445527015,
      "grad_norm": 0.22496171295642853,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 81355
    },
    {
      "epoch": 72.0637732506643,
      "grad_norm": 0.236468106508255,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 81360
    },
    {
      "epoch": 72.0682019486271,
      "grad_norm": 0.24994243681430817,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 81365
    },
    {
      "epoch": 72.0726306465899,
      "grad_norm": 0.23941124975681305,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 81370
    },
    {
      "epoch": 72.07705934455271,
      "grad_norm": 0.25477737188339233,
      "learning_rate": 1e-05,
      "loss": 1.0119,
      "step": 81375
    },
    {
      "epoch": 72.0814880425155,
      "grad_norm": 0.23598423600196838,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 81380
    },
    {
      "epoch": 72.0859167404783,
      "grad_norm": 0.21149305999279022,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 81385
    },
    {
      "epoch": 72.0903454384411,
      "grad_norm": 0.2616944909095764,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 81390
    },
    {
      "epoch": 72.0947741364039,
      "grad_norm": 0.27005478739738464,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 81395
    },
    {
      "epoch": 72.0992028343667,
      "grad_norm": 0.27633461356163025,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 81400
    },
    {
      "epoch": 72.1036315323295,
      "grad_norm": 0.23954515159130096,
      "learning_rate": 1e-05,
      "loss": 0.9232,
      "step": 81405
    },
    {
      "epoch": 72.1080602302923,
      "grad_norm": 0.2105601578950882,
      "learning_rate": 1e-05,
      "loss": 0.929,
      "step": 81410
    },
    {
      "epoch": 72.11248892825509,
      "grad_norm": 0.2877689301967621,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 81415
    },
    {
      "epoch": 72.1169176262179,
      "grad_norm": 0.2449050396680832,
      "learning_rate": 1e-05,
      "loss": 0.9102,
      "step": 81420
    },
    {
      "epoch": 72.12134632418069,
      "grad_norm": 0.24011529982089996,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 81425
    },
    {
      "epoch": 72.12577502214349,
      "grad_norm": 0.26916563510894775,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 81430
    },
    {
      "epoch": 72.1302037201063,
      "grad_norm": 0.1996055245399475,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 81435
    },
    {
      "epoch": 72.13463241806909,
      "grad_norm": 0.24843572080135345,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 81440
    },
    {
      "epoch": 72.13906111603188,
      "grad_norm": 0.21508702635765076,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 81445
    },
    {
      "epoch": 72.14348981399469,
      "grad_norm": 0.23762810230255127,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 81450
    },
    {
      "epoch": 72.14791851195749,
      "grad_norm": 0.28272709250450134,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 81455
    },
    {
      "epoch": 72.15234720992028,
      "grad_norm": 0.26446881890296936,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 81460
    },
    {
      "epoch": 72.15677590788309,
      "grad_norm": 0.24066747725009918,
      "learning_rate": 1e-05,
      "loss": 0.9166,
      "step": 81465
    },
    {
      "epoch": 72.16120460584588,
      "grad_norm": 0.2595427334308624,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 81470
    },
    {
      "epoch": 72.16563330380868,
      "grad_norm": 0.20953522622585297,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 81475
    },
    {
      "epoch": 72.17006200177148,
      "grad_norm": 0.255551815032959,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 81480
    },
    {
      "epoch": 72.17449069973428,
      "grad_norm": 0.21852417290210724,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 81485
    },
    {
      "epoch": 72.17891939769707,
      "grad_norm": 0.22914937138557434,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 81490
    },
    {
      "epoch": 72.18334809565988,
      "grad_norm": 0.2236819714307785,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 81495
    },
    {
      "epoch": 72.18777679362267,
      "grad_norm": 0.2010485827922821,
      "learning_rate": 1e-05,
      "loss": 0.9432,
      "step": 81500
    },
    {
      "epoch": 72.19220549158547,
      "grad_norm": 0.22047671675682068,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 81505
    },
    {
      "epoch": 72.19663418954828,
      "grad_norm": 0.2272142916917801,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 81510
    },
    {
      "epoch": 72.20106288751107,
      "grad_norm": 0.2484421581029892,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 81515
    },
    {
      "epoch": 72.20549158547387,
      "grad_norm": 0.2548709809780121,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 81520
    },
    {
      "epoch": 72.20992028343667,
      "grad_norm": 0.212657630443573,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 81525
    },
    {
      "epoch": 72.21434898139947,
      "grad_norm": 0.22249644994735718,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 81530
    },
    {
      "epoch": 72.21877767936226,
      "grad_norm": 0.2488417774438858,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 81535
    },
    {
      "epoch": 72.22320637732507,
      "grad_norm": 0.25050610303878784,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 81540
    },
    {
      "epoch": 72.22763507528786,
      "grad_norm": 0.22389383614063263,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 81545
    },
    {
      "epoch": 72.23206377325066,
      "grad_norm": 0.2913344204425812,
      "learning_rate": 1e-05,
      "loss": 0.9207,
      "step": 81550
    },
    {
      "epoch": 72.23649247121347,
      "grad_norm": 0.26752591133117676,
      "learning_rate": 1e-05,
      "loss": 1.0218,
      "step": 81555
    },
    {
      "epoch": 72.24092116917626,
      "grad_norm": 0.29146572947502136,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 81560
    },
    {
      "epoch": 72.24534986713905,
      "grad_norm": 0.2561086416244507,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 81565
    },
    {
      "epoch": 72.24977856510186,
      "grad_norm": 0.24353192746639252,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 81570
    },
    {
      "epoch": 72.25420726306466,
      "grad_norm": 0.20363597571849823,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 81575
    },
    {
      "epoch": 72.25863596102745,
      "grad_norm": 0.28131386637687683,
      "learning_rate": 1e-05,
      "loss": 0.9347,
      "step": 81580
    },
    {
      "epoch": 72.26306465899026,
      "grad_norm": 0.23872306942939758,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 81585
    },
    {
      "epoch": 72.26749335695305,
      "grad_norm": 0.24349729716777802,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 81590
    },
    {
      "epoch": 72.27192205491585,
      "grad_norm": 0.22526611387729645,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 81595
    },
    {
      "epoch": 72.27635075287866,
      "grad_norm": 0.2616110146045685,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 81600
    },
    {
      "epoch": 72.28077945084145,
      "grad_norm": 0.26459765434265137,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 81605
    },
    {
      "epoch": 72.28520814880426,
      "grad_norm": 0.2639113664627075,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 81610
    },
    {
      "epoch": 72.28963684676705,
      "grad_norm": 0.23226481676101685,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 81615
    },
    {
      "epoch": 72.29406554472985,
      "grad_norm": 0.27576157450675964,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 81620
    },
    {
      "epoch": 72.29849424269266,
      "grad_norm": 0.2584199011325836,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 81625
    },
    {
      "epoch": 72.30292294065545,
      "grad_norm": 0.27295801043510437,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 81630
    },
    {
      "epoch": 72.30735163861824,
      "grad_norm": 0.23922015726566315,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 81635
    },
    {
      "epoch": 72.31178033658105,
      "grad_norm": 0.2583041489124298,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 81640
    },
    {
      "epoch": 72.31620903454385,
      "grad_norm": 0.27850082516670227,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 81645
    },
    {
      "epoch": 72.32063773250664,
      "grad_norm": 0.20707598328590393,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 81650
    },
    {
      "epoch": 72.32506643046945,
      "grad_norm": 0.23906508088111877,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 81655
    },
    {
      "epoch": 72.32949512843224,
      "grad_norm": 0.23260806500911713,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 81660
    },
    {
      "epoch": 72.33392382639504,
      "grad_norm": 0.21275664865970612,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 81665
    },
    {
      "epoch": 72.33835252435784,
      "grad_norm": 0.2309500277042389,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 81670
    },
    {
      "epoch": 72.34278122232064,
      "grad_norm": 0.2329455018043518,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 81675
    },
    {
      "epoch": 72.34720992028343,
      "grad_norm": 0.22734816372394562,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 81680
    },
    {
      "epoch": 72.35163861824624,
      "grad_norm": 0.22864748537540436,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 81685
    },
    {
      "epoch": 72.35606731620904,
      "grad_norm": 0.2648402750492096,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 81690
    },
    {
      "epoch": 72.36049601417183,
      "grad_norm": 0.25199875235557556,
      "learning_rate": 1e-05,
      "loss": 0.941,
      "step": 81695
    },
    {
      "epoch": 72.36492471213464,
      "grad_norm": 0.28546321392059326,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 81700
    },
    {
      "epoch": 72.36935341009743,
      "grad_norm": 0.23969654738903046,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 81705
    },
    {
      "epoch": 72.37378210806023,
      "grad_norm": 0.2365540713071823,
      "learning_rate": 1e-05,
      "loss": 1.0338,
      "step": 81710
    },
    {
      "epoch": 72.37821080602303,
      "grad_norm": 0.22369222342967987,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 81715
    },
    {
      "epoch": 72.38263950398583,
      "grad_norm": 0.2490292489528656,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 81720
    },
    {
      "epoch": 72.38706820194862,
      "grad_norm": 0.27210327982902527,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 81725
    },
    {
      "epoch": 72.39149689991143,
      "grad_norm": 0.19943967461585999,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 81730
    },
    {
      "epoch": 72.39592559787422,
      "grad_norm": 0.2582855522632599,
      "learning_rate": 1e-05,
      "loss": 0.8777,
      "step": 81735
    },
    {
      "epoch": 72.40035429583702,
      "grad_norm": 0.22129642963409424,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 81740
    },
    {
      "epoch": 72.40478299379983,
      "grad_norm": 0.23241719603538513,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 81745
    },
    {
      "epoch": 72.40921169176262,
      "grad_norm": 0.28924551606178284,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 81750
    },
    {
      "epoch": 72.41364038972542,
      "grad_norm": 0.254770427942276,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 81755
    },
    {
      "epoch": 72.41806908768822,
      "grad_norm": 0.2669125199317932,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 81760
    },
    {
      "epoch": 72.42249778565102,
      "grad_norm": 0.29147234559059143,
      "learning_rate": 1e-05,
      "loss": 0.9364,
      "step": 81765
    },
    {
      "epoch": 72.42692648361381,
      "grad_norm": 0.21068868041038513,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 81770
    },
    {
      "epoch": 72.43135518157662,
      "grad_norm": 0.23538866639137268,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 81775
    },
    {
      "epoch": 72.43578387953941,
      "grad_norm": 0.2441510260105133,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 81780
    },
    {
      "epoch": 72.44021257750221,
      "grad_norm": 0.22194457054138184,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 81785
    },
    {
      "epoch": 72.44464127546502,
      "grad_norm": 0.2323472499847412,
      "learning_rate": 1e-05,
      "loss": 0.9094,
      "step": 81790
    },
    {
      "epoch": 72.44906997342781,
      "grad_norm": 0.2350277602672577,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 81795
    },
    {
      "epoch": 72.4534986713906,
      "grad_norm": 0.22526490688323975,
      "learning_rate": 1e-05,
      "loss": 0.9195,
      "step": 81800
    },
    {
      "epoch": 72.45792736935341,
      "grad_norm": 0.20925948023796082,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 81805
    },
    {
      "epoch": 72.46235606731621,
      "grad_norm": 0.21776196360588074,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 81810
    },
    {
      "epoch": 72.466784765279,
      "grad_norm": 0.2173004001379013,
      "learning_rate": 1e-05,
      "loss": 1.0448,
      "step": 81815
    },
    {
      "epoch": 72.47121346324181,
      "grad_norm": 0.2210615575313568,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 81820
    },
    {
      "epoch": 72.4756421612046,
      "grad_norm": 0.22636619210243225,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 81825
    },
    {
      "epoch": 72.4800708591674,
      "grad_norm": 0.21062739193439484,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 81830
    },
    {
      "epoch": 72.4844995571302,
      "grad_norm": 0.22763392329216003,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 81835
    },
    {
      "epoch": 72.488928255093,
      "grad_norm": 0.29344770312309265,
      "learning_rate": 1e-05,
      "loss": 0.9134,
      "step": 81840
    },
    {
      "epoch": 72.4933569530558,
      "grad_norm": 0.24963222444057465,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 81845
    },
    {
      "epoch": 72.4977856510186,
      "grad_norm": 0.22271636128425598,
      "learning_rate": 1e-05,
      "loss": 0.8833,
      "step": 81850
    },
    {
      "epoch": 72.5022143489814,
      "grad_norm": 0.19670172035694122,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 81855
    },
    {
      "epoch": 72.5066430469442,
      "grad_norm": 0.2319105714559555,
      "learning_rate": 1e-05,
      "loss": 1.0384,
      "step": 81860
    },
    {
      "epoch": 72.511071744907,
      "grad_norm": 0.24714207649230957,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 81865
    },
    {
      "epoch": 72.5155004428698,
      "grad_norm": 0.2168656885623932,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 81870
    },
    {
      "epoch": 72.5199291408326,
      "grad_norm": 0.24972131848335266,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 81875
    },
    {
      "epoch": 72.5243578387954,
      "grad_norm": 0.23894548416137695,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 81880
    },
    {
      "epoch": 72.52878653675819,
      "grad_norm": 0.21484695374965668,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 81885
    },
    {
      "epoch": 72.533215234721,
      "grad_norm": 0.22791941463947296,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 81890
    },
    {
      "epoch": 72.53764393268379,
      "grad_norm": 0.21178391575813293,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 81895
    },
    {
      "epoch": 72.54207263064659,
      "grad_norm": 0.2729080617427826,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 81900
    },
    {
      "epoch": 72.5465013286094,
      "grad_norm": 0.24734045565128326,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 81905
    },
    {
      "epoch": 72.55093002657219,
      "grad_norm": 0.2369898110628128,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 81910
    },
    {
      "epoch": 72.55535872453498,
      "grad_norm": 0.29310154914855957,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 81915
    },
    {
      "epoch": 72.55978742249779,
      "grad_norm": 0.2385549396276474,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 81920
    },
    {
      "epoch": 72.56421612046059,
      "grad_norm": 0.25564342737197876,
      "learning_rate": 1e-05,
      "loss": 0.917,
      "step": 81925
    },
    {
      "epoch": 72.56864481842338,
      "grad_norm": 0.21443584561347961,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 81930
    },
    {
      "epoch": 72.57307351638619,
      "grad_norm": 0.2723100781440735,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 81935
    },
    {
      "epoch": 72.57750221434898,
      "grad_norm": 0.26437267661094666,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 81940
    },
    {
      "epoch": 72.58193091231178,
      "grad_norm": 0.24043592810630798,
      "learning_rate": 1e-05,
      "loss": 1.0126,
      "step": 81945
    },
    {
      "epoch": 72.58635961027458,
      "grad_norm": 0.25841444730758667,
      "learning_rate": 1e-05,
      "loss": 1.0247,
      "step": 81950
    },
    {
      "epoch": 72.59078830823738,
      "grad_norm": 0.2862551510334015,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 81955
    },
    {
      "epoch": 72.59521700620017,
      "grad_norm": 0.2475762963294983,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 81960
    },
    {
      "epoch": 72.59964570416298,
      "grad_norm": 0.2246282547712326,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 81965
    },
    {
      "epoch": 72.60407440212578,
      "grad_norm": 0.28102266788482666,
      "learning_rate": 1e-05,
      "loss": 0.9325,
      "step": 81970
    },
    {
      "epoch": 72.60850310008857,
      "grad_norm": 0.22349387407302856,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 81975
    },
    {
      "epoch": 72.61293179805138,
      "grad_norm": 0.3062332570552826,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 81980
    },
    {
      "epoch": 72.61736049601417,
      "grad_norm": 0.2933221459388733,
      "learning_rate": 1e-05,
      "loss": 1.0086,
      "step": 81985
    },
    {
      "epoch": 72.62178919397697,
      "grad_norm": 0.23921425640583038,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 81990
    },
    {
      "epoch": 72.62621789193977,
      "grad_norm": 0.23220431804656982,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 81995
    },
    {
      "epoch": 72.63064658990257,
      "grad_norm": 0.24322591722011566,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 82000
    },
    {
      "epoch": 72.63507528786536,
      "grad_norm": 0.22642558813095093,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 82005
    },
    {
      "epoch": 72.63950398582817,
      "grad_norm": 0.22948335111141205,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 82010
    },
    {
      "epoch": 72.64393268379096,
      "grad_norm": 0.29430514574050903,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 82015
    },
    {
      "epoch": 72.64836138175376,
      "grad_norm": 0.26079872250556946,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 82020
    },
    {
      "epoch": 72.65279007971657,
      "grad_norm": 0.24384747445583344,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 82025
    },
    {
      "epoch": 72.65721877767936,
      "grad_norm": 0.2269025295972824,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 82030
    },
    {
      "epoch": 72.66164747564216,
      "grad_norm": 0.1892281323671341,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 82035
    },
    {
      "epoch": 72.66607617360496,
      "grad_norm": 0.23388154804706573,
      "learning_rate": 1e-05,
      "loss": 0.9178,
      "step": 82040
    },
    {
      "epoch": 72.67050487156776,
      "grad_norm": 0.25561922788619995,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 82045
    },
    {
      "epoch": 72.67493356953055,
      "grad_norm": 0.23890981078147888,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 82050
    },
    {
      "epoch": 72.67936226749336,
      "grad_norm": 0.2638990581035614,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 82055
    },
    {
      "epoch": 72.68379096545615,
      "grad_norm": 0.260124146938324,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 82060
    },
    {
      "epoch": 72.68821966341895,
      "grad_norm": 0.2417614609003067,
      "learning_rate": 1e-05,
      "loss": 1.0233,
      "step": 82065
    },
    {
      "epoch": 72.69264836138176,
      "grad_norm": 0.2613557279109955,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 82070
    },
    {
      "epoch": 72.69707705934455,
      "grad_norm": 0.27670252323150635,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 82075
    },
    {
      "epoch": 72.70150575730734,
      "grad_norm": 0.22210197150707245,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 82080
    },
    {
      "epoch": 72.70593445527015,
      "grad_norm": 0.2397746443748474,
      "learning_rate": 1e-05,
      "loss": 0.9006,
      "step": 82085
    },
    {
      "epoch": 72.71036315323295,
      "grad_norm": 0.25654351711273193,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 82090
    },
    {
      "epoch": 72.71479185119574,
      "grad_norm": 0.24264849722385406,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 82095
    },
    {
      "epoch": 72.71922054915855,
      "grad_norm": 0.207445427775383,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 82100
    },
    {
      "epoch": 72.72364924712134,
      "grad_norm": 0.2603868544101715,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 82105
    },
    {
      "epoch": 72.72807794508415,
      "grad_norm": 0.2347344607114792,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 82110
    },
    {
      "epoch": 72.73250664304695,
      "grad_norm": 0.20686018466949463,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 82115
    },
    {
      "epoch": 72.73693534100974,
      "grad_norm": 0.23141111433506012,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 82120
    },
    {
      "epoch": 72.74136403897255,
      "grad_norm": 0.23409943282604218,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 82125
    },
    {
      "epoch": 72.74579273693534,
      "grad_norm": 0.22090205550193787,
      "learning_rate": 1e-05,
      "loss": 0.9224,
      "step": 82130
    },
    {
      "epoch": 72.75022143489814,
      "grad_norm": 0.23999835550785065,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 82135
    },
    {
      "epoch": 72.75465013286095,
      "grad_norm": 0.22561438381671906,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 82140
    },
    {
      "epoch": 72.75907883082374,
      "grad_norm": 0.2848077714443207,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 82145
    },
    {
      "epoch": 72.76350752878653,
      "grad_norm": 0.2460223138332367,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 82150
    },
    {
      "epoch": 72.76793622674934,
      "grad_norm": 0.2385382503271103,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 82155
    },
    {
      "epoch": 72.77236492471214,
      "grad_norm": 0.21186718344688416,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 82160
    },
    {
      "epoch": 72.77679362267493,
      "grad_norm": 0.24680039286613464,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 82165
    },
    {
      "epoch": 72.78122232063774,
      "grad_norm": 0.237496480345726,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 82170
    },
    {
      "epoch": 72.78565101860053,
      "grad_norm": 0.21692587435245514,
      "learning_rate": 1e-05,
      "loss": 1.018,
      "step": 82175
    },
    {
      "epoch": 72.79007971656333,
      "grad_norm": 0.2576132118701935,
      "learning_rate": 1e-05,
      "loss": 0.9229,
      "step": 82180
    },
    {
      "epoch": 72.79450841452613,
      "grad_norm": 0.2511577606201172,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 82185
    },
    {
      "epoch": 72.79893711248893,
      "grad_norm": 0.2365998476743698,
      "learning_rate": 1e-05,
      "loss": 1.0149,
      "step": 82190
    },
    {
      "epoch": 72.80336581045172,
      "grad_norm": 0.3116912841796875,
      "learning_rate": 1e-05,
      "loss": 0.9088,
      "step": 82195
    },
    {
      "epoch": 72.80779450841453,
      "grad_norm": 0.20943093299865723,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 82200
    },
    {
      "epoch": 72.81222320637733,
      "grad_norm": 0.24530185759067535,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 82205
    },
    {
      "epoch": 72.81665190434012,
      "grad_norm": 0.21564076840877533,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 82210
    },
    {
      "epoch": 72.82108060230293,
      "grad_norm": 0.27512311935424805,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 82215
    },
    {
      "epoch": 72.82550930026572,
      "grad_norm": 0.2699989676475525,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 82220
    },
    {
      "epoch": 72.82993799822852,
      "grad_norm": 0.2616477310657501,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 82225
    },
    {
      "epoch": 72.83436669619132,
      "grad_norm": 0.200178861618042,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 82230
    },
    {
      "epoch": 72.83879539415412,
      "grad_norm": 0.26071739196777344,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 82235
    },
    {
      "epoch": 72.84322409211691,
      "grad_norm": 0.2706190347671509,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 82240
    },
    {
      "epoch": 72.84765279007972,
      "grad_norm": 0.23201747238636017,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 82245
    },
    {
      "epoch": 72.85208148804251,
      "grad_norm": 0.2525644898414612,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 82250
    },
    {
      "epoch": 72.85651018600531,
      "grad_norm": 0.28115618228912354,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 82255
    },
    {
      "epoch": 72.86093888396812,
      "grad_norm": 0.264055997133255,
      "learning_rate": 1e-05,
      "loss": 1.0422,
      "step": 82260
    },
    {
      "epoch": 72.86536758193091,
      "grad_norm": 0.2329600304365158,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 82265
    },
    {
      "epoch": 72.8697962798937,
      "grad_norm": 0.237477108836174,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 82270
    },
    {
      "epoch": 72.87422497785651,
      "grad_norm": 0.22517579793930054,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 82275
    },
    {
      "epoch": 72.87865367581931,
      "grad_norm": 0.2691676616668701,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 82280
    },
    {
      "epoch": 72.8830823737821,
      "grad_norm": 0.2528765797615051,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 82285
    },
    {
      "epoch": 72.88751107174491,
      "grad_norm": 0.2579416036605835,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 82290
    },
    {
      "epoch": 72.8919397697077,
      "grad_norm": 0.25088420510292053,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 82295
    },
    {
      "epoch": 72.8963684676705,
      "grad_norm": 0.24960538744926453,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 82300
    },
    {
      "epoch": 72.9007971656333,
      "grad_norm": 0.24997778236865997,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 82305
    },
    {
      "epoch": 72.9052258635961,
      "grad_norm": 0.2649942636489868,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 82310
    },
    {
      "epoch": 72.9096545615589,
      "grad_norm": 0.22673450410366058,
      "learning_rate": 1e-05,
      "loss": 0.9303,
      "step": 82315
    },
    {
      "epoch": 72.9140832595217,
      "grad_norm": 0.29669609665870667,
      "learning_rate": 1e-05,
      "loss": 0.9215,
      "step": 82320
    },
    {
      "epoch": 72.9185119574845,
      "grad_norm": 0.22993719577789307,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 82325
    },
    {
      "epoch": 72.92294065544729,
      "grad_norm": 0.2600945234298706,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 82330
    },
    {
      "epoch": 72.9273693534101,
      "grad_norm": 0.26497912406921387,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 82335
    },
    {
      "epoch": 72.9317980513729,
      "grad_norm": 0.23068685829639435,
      "learning_rate": 1e-05,
      "loss": 1.0239,
      "step": 82340
    },
    {
      "epoch": 72.9362267493357,
      "grad_norm": 0.24579927325248718,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 82345
    },
    {
      "epoch": 72.9406554472985,
      "grad_norm": 0.24136674404144287,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 82350
    },
    {
      "epoch": 72.94508414526129,
      "grad_norm": 0.24702665209770203,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 82355
    },
    {
      "epoch": 72.9495128432241,
      "grad_norm": 0.2417319118976593,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 82360
    },
    {
      "epoch": 72.9539415411869,
      "grad_norm": 0.2343817800283432,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 82365
    },
    {
      "epoch": 72.95837023914969,
      "grad_norm": 0.2281305342912674,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 82370
    },
    {
      "epoch": 72.9627989371125,
      "grad_norm": 0.24599644541740417,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 82375
    },
    {
      "epoch": 72.96722763507529,
      "grad_norm": 0.2539338767528534,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 82380
    },
    {
      "epoch": 72.97165633303808,
      "grad_norm": 0.2511104643344879,
      "learning_rate": 1e-05,
      "loss": 0.9213,
      "step": 82385
    },
    {
      "epoch": 72.97608503100089,
      "grad_norm": 0.2469913363456726,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 82390
    },
    {
      "epoch": 72.98051372896369,
      "grad_norm": 0.21827562153339386,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 82395
    },
    {
      "epoch": 72.98494242692648,
      "grad_norm": 0.21459631621837616,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 82400
    },
    {
      "epoch": 72.98937112488929,
      "grad_norm": 0.23577465116977692,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 82405
    },
    {
      "epoch": 72.99379982285208,
      "grad_norm": 0.21619725227355957,
      "learning_rate": 1e-05,
      "loss": 0.9392,
      "step": 82410
    },
    {
      "epoch": 72.99822852081488,
      "grad_norm": 0.23966482281684875,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 82415
    },
    {
      "epoch": 73.00265721877768,
      "grad_norm": 0.2208172082901001,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 82420
    },
    {
      "epoch": 73.00708591674048,
      "grad_norm": 0.21438747644424438,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 82425
    },
    {
      "epoch": 73.01151461470327,
      "grad_norm": 0.22542330622673035,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 82430
    },
    {
      "epoch": 73.01594331266608,
      "grad_norm": 0.2583635747432709,
      "learning_rate": 1e-05,
      "loss": 0.8995,
      "step": 82435
    },
    {
      "epoch": 73.02037201062888,
      "grad_norm": 0.23522783815860748,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 82440
    },
    {
      "epoch": 73.02480070859167,
      "grad_norm": 0.20607879757881165,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 82445
    },
    {
      "epoch": 73.02922940655448,
      "grad_norm": 0.22864672541618347,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 82450
    },
    {
      "epoch": 73.03365810451727,
      "grad_norm": 0.23980331420898438,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 82455
    },
    {
      "epoch": 73.03808680248007,
      "grad_norm": 0.227765753865242,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 82460
    },
    {
      "epoch": 73.04251550044287,
      "grad_norm": 0.25453898310661316,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 82465
    },
    {
      "epoch": 73.04694419840567,
      "grad_norm": 0.22375932335853577,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 82470
    },
    {
      "epoch": 73.05137289636846,
      "grad_norm": 0.2136206030845642,
      "learning_rate": 1e-05,
      "loss": 0.887,
      "step": 82475
    },
    {
      "epoch": 73.05580159433127,
      "grad_norm": 0.28091931343078613,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 82480
    },
    {
      "epoch": 73.06023029229407,
      "grad_norm": 0.26697659492492676,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 82485
    },
    {
      "epoch": 73.06465899025686,
      "grad_norm": 0.2722189724445343,
      "learning_rate": 1e-05,
      "loss": 1.0574,
      "step": 82490
    },
    {
      "epoch": 73.06908768821967,
      "grad_norm": 0.23614715039730072,
      "learning_rate": 1e-05,
      "loss": 1.0005,
      "step": 82495
    },
    {
      "epoch": 73.07351638618246,
      "grad_norm": 0.2785855829715729,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 82500
    },
    {
      "epoch": 73.07794508414526,
      "grad_norm": 0.2589624226093292,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 82505
    },
    {
      "epoch": 73.08237378210806,
      "grad_norm": 0.23542962968349457,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 82510
    },
    {
      "epoch": 73.08680248007086,
      "grad_norm": 0.228175550699234,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 82515
    },
    {
      "epoch": 73.09123117803365,
      "grad_norm": 0.2737840712070465,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 82520
    },
    {
      "epoch": 73.09565987599646,
      "grad_norm": 0.26945221424102783,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 82525
    },
    {
      "epoch": 73.10008857395925,
      "grad_norm": 0.24570070207118988,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 82530
    },
    {
      "epoch": 73.10451727192205,
      "grad_norm": 0.24858075380325317,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 82535
    },
    {
      "epoch": 73.10894596988486,
      "grad_norm": 0.21923251450061798,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 82540
    },
    {
      "epoch": 73.11337466784765,
      "grad_norm": 0.2247978150844574,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 82545
    },
    {
      "epoch": 73.11780336581045,
      "grad_norm": 0.22685687243938446,
      "learning_rate": 1e-05,
      "loss": 0.9198,
      "step": 82550
    },
    {
      "epoch": 73.12223206377325,
      "grad_norm": 0.25057026743888855,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 82555
    },
    {
      "epoch": 73.12666076173605,
      "grad_norm": 0.28565722703933716,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 82560
    },
    {
      "epoch": 73.13108945969884,
      "grad_norm": 0.22555232048034668,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 82565
    },
    {
      "epoch": 73.13551815766165,
      "grad_norm": 0.24804599583148956,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 82570
    },
    {
      "epoch": 73.13994685562444,
      "grad_norm": 0.25632360577583313,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 82575
    },
    {
      "epoch": 73.14437555358724,
      "grad_norm": 0.22083309292793274,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 82580
    },
    {
      "epoch": 73.14880425155005,
      "grad_norm": 0.2413714975118637,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 82585
    },
    {
      "epoch": 73.15323294951284,
      "grad_norm": 0.22615182399749756,
      "learning_rate": 1e-05,
      "loss": 1.0245,
      "step": 82590
    },
    {
      "epoch": 73.15766164747565,
      "grad_norm": 0.275217205286026,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 82595
    },
    {
      "epoch": 73.16209034543844,
      "grad_norm": 0.2280871719121933,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 82600
    },
    {
      "epoch": 73.16651904340124,
      "grad_norm": 0.24297204613685608,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 82605
    },
    {
      "epoch": 73.17094774136405,
      "grad_norm": 0.21975819766521454,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 82610
    },
    {
      "epoch": 73.17537643932684,
      "grad_norm": 0.23662470281124115,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 82615
    },
    {
      "epoch": 73.17980513728963,
      "grad_norm": 0.2460552304983139,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 82620
    },
    {
      "epoch": 73.18423383525244,
      "grad_norm": 0.21850033104419708,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 82625
    },
    {
      "epoch": 73.18866253321524,
      "grad_norm": 0.24351060390472412,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 82630
    },
    {
      "epoch": 73.19309123117803,
      "grad_norm": 0.21935692429542542,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 82635
    },
    {
      "epoch": 73.19751992914084,
      "grad_norm": 0.28177767992019653,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 82640
    },
    {
      "epoch": 73.20194862710363,
      "grad_norm": 0.21997953951358795,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 82645
    },
    {
      "epoch": 73.20637732506643,
      "grad_norm": 0.23230719566345215,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 82650
    },
    {
      "epoch": 73.21080602302924,
      "grad_norm": 0.2655057907104492,
      "learning_rate": 1e-05,
      "loss": 0.8886,
      "step": 82655
    },
    {
      "epoch": 73.21523472099203,
      "grad_norm": 0.22441129386425018,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 82660
    },
    {
      "epoch": 73.21966341895482,
      "grad_norm": 0.23323912918567657,
      "learning_rate": 1e-05,
      "loss": 0.9286,
      "step": 82665
    },
    {
      "epoch": 73.22409211691763,
      "grad_norm": 0.23481370508670807,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 82670
    },
    {
      "epoch": 73.22852081488043,
      "grad_norm": 0.2173687219619751,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 82675
    },
    {
      "epoch": 73.23294951284322,
      "grad_norm": 0.25321415066719055,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 82680
    },
    {
      "epoch": 73.23737821080603,
      "grad_norm": 0.21834953129291534,
      "learning_rate": 1e-05,
      "loss": 0.9243,
      "step": 82685
    },
    {
      "epoch": 73.24180690876882,
      "grad_norm": 0.297186940908432,
      "learning_rate": 1e-05,
      "loss": 0.916,
      "step": 82690
    },
    {
      "epoch": 73.24623560673162,
      "grad_norm": 0.2575085759162903,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 82695
    },
    {
      "epoch": 73.25066430469442,
      "grad_norm": 0.2264639139175415,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 82700
    },
    {
      "epoch": 73.25509300265722,
      "grad_norm": 0.22753046452999115,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 82705
    },
    {
      "epoch": 73.25952170062001,
      "grad_norm": 0.22967657446861267,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 82710
    },
    {
      "epoch": 73.26395039858282,
      "grad_norm": 0.2646719515323639,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 82715
    },
    {
      "epoch": 73.26837909654562,
      "grad_norm": 0.2503224015235901,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 82720
    },
    {
      "epoch": 73.27280779450841,
      "grad_norm": 0.2612695097923279,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 82725
    },
    {
      "epoch": 73.27723649247122,
      "grad_norm": 0.2478746771812439,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 82730
    },
    {
      "epoch": 73.28166519043401,
      "grad_norm": 0.28299659490585327,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 82735
    },
    {
      "epoch": 73.2860938883968,
      "grad_norm": 0.26097890734672546,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 82740
    },
    {
      "epoch": 73.29052258635961,
      "grad_norm": 0.24240058660507202,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 82745
    },
    {
      "epoch": 73.29495128432241,
      "grad_norm": 0.2340339571237564,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 82750
    },
    {
      "epoch": 73.2993799822852,
      "grad_norm": 0.2833745777606964,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 82755
    },
    {
      "epoch": 73.30380868024801,
      "grad_norm": 0.22378607094287872,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 82760
    },
    {
      "epoch": 73.3082373782108,
      "grad_norm": 0.2732107937335968,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 82765
    },
    {
      "epoch": 73.3126660761736,
      "grad_norm": 0.2284357100725174,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 82770
    },
    {
      "epoch": 73.31709477413641,
      "grad_norm": 0.24475449323654175,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 82775
    },
    {
      "epoch": 73.3215234720992,
      "grad_norm": 0.22579967975616455,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 82780
    },
    {
      "epoch": 73.325952170062,
      "grad_norm": 0.2380644530057907,
      "learning_rate": 1e-05,
      "loss": 1.0477,
      "step": 82785
    },
    {
      "epoch": 73.3303808680248,
      "grad_norm": 0.24849334359169006,
      "learning_rate": 1e-05,
      "loss": 0.902,
      "step": 82790
    },
    {
      "epoch": 73.3348095659876,
      "grad_norm": 0.2405947595834732,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 82795
    },
    {
      "epoch": 73.33923826395039,
      "grad_norm": 0.2670816481113434,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 82800
    },
    {
      "epoch": 73.3436669619132,
      "grad_norm": 0.23080633580684662,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 82805
    },
    {
      "epoch": 73.348095659876,
      "grad_norm": 0.23334714770317078,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 82810
    },
    {
      "epoch": 73.35252435783879,
      "grad_norm": 0.25967931747436523,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 82815
    },
    {
      "epoch": 73.3569530558016,
      "grad_norm": 0.24247030913829803,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 82820
    },
    {
      "epoch": 73.36138175376439,
      "grad_norm": 0.23570400476455688,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 82825
    },
    {
      "epoch": 73.36581045172719,
      "grad_norm": 0.23097412288188934,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 82830
    },
    {
      "epoch": 73.37023914969,
      "grad_norm": 0.22044362127780914,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 82835
    },
    {
      "epoch": 73.37466784765279,
      "grad_norm": 0.24457845091819763,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 82840
    },
    {
      "epoch": 73.3790965456156,
      "grad_norm": 0.25608763098716736,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 82845
    },
    {
      "epoch": 73.38352524357839,
      "grad_norm": 0.268093466758728,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 82850
    },
    {
      "epoch": 73.38795394154118,
      "grad_norm": 0.2697315216064453,
      "learning_rate": 1e-05,
      "loss": 0.9079,
      "step": 82855
    },
    {
      "epoch": 73.39238263950399,
      "grad_norm": 0.2225591391324997,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 82860
    },
    {
      "epoch": 73.39681133746679,
      "grad_norm": 0.2316623479127884,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 82865
    },
    {
      "epoch": 73.40124003542958,
      "grad_norm": 0.25394290685653687,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 82870
    },
    {
      "epoch": 73.40566873339239,
      "grad_norm": 0.25854936242103577,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 82875
    },
    {
      "epoch": 73.41009743135518,
      "grad_norm": 0.2392587661743164,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 82880
    },
    {
      "epoch": 73.41452612931798,
      "grad_norm": 0.23477937281131744,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 82885
    },
    {
      "epoch": 73.41895482728079,
      "grad_norm": 0.23929959535598755,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 82890
    },
    {
      "epoch": 73.42338352524358,
      "grad_norm": 0.2164512723684311,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 82895
    },
    {
      "epoch": 73.42781222320637,
      "grad_norm": 0.21970677375793457,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 82900
    },
    {
      "epoch": 73.43224092116918,
      "grad_norm": 0.24705880880355835,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 82905
    },
    {
      "epoch": 73.43666961913198,
      "grad_norm": 0.23214955627918243,
      "learning_rate": 1e-05,
      "loss": 0.9289,
      "step": 82910
    },
    {
      "epoch": 73.44109831709477,
      "grad_norm": 0.27697354555130005,
      "learning_rate": 1e-05,
      "loss": 0.9234,
      "step": 82915
    },
    {
      "epoch": 73.44552701505758,
      "grad_norm": 0.2728331685066223,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 82920
    },
    {
      "epoch": 73.44995571302037,
      "grad_norm": 0.24829663336277008,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 82925
    },
    {
      "epoch": 73.45438441098317,
      "grad_norm": 0.3199656307697296,
      "learning_rate": 1e-05,
      "loss": 0.9293,
      "step": 82930
    },
    {
      "epoch": 73.45881310894598,
      "grad_norm": 0.2506788671016693,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 82935
    },
    {
      "epoch": 73.46324180690877,
      "grad_norm": 0.2483828067779541,
      "learning_rate": 1e-05,
      "loss": 1.0273,
      "step": 82940
    },
    {
      "epoch": 73.46767050487156,
      "grad_norm": 0.22795487940311432,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 82945
    },
    {
      "epoch": 73.47209920283437,
      "grad_norm": 0.23457051813602448,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 82950
    },
    {
      "epoch": 73.47652790079717,
      "grad_norm": 0.26390478014945984,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 82955
    },
    {
      "epoch": 73.48095659875996,
      "grad_norm": 0.22247827053070068,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 82960
    },
    {
      "epoch": 73.48538529672277,
      "grad_norm": 0.22308751940727234,
      "learning_rate": 1e-05,
      "loss": 0.9229,
      "step": 82965
    },
    {
      "epoch": 73.48981399468556,
      "grad_norm": 0.27113136649131775,
      "learning_rate": 1e-05,
      "loss": 0.941,
      "step": 82970
    },
    {
      "epoch": 73.49424269264836,
      "grad_norm": 0.2309257984161377,
      "learning_rate": 1e-05,
      "loss": 0.9202,
      "step": 82975
    },
    {
      "epoch": 73.49867139061116,
      "grad_norm": 0.20855991542339325,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 82980
    },
    {
      "epoch": 73.50310008857396,
      "grad_norm": 0.2625739872455597,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 82985
    },
    {
      "epoch": 73.50752878653675,
      "grad_norm": 0.20825833082199097,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 82990
    },
    {
      "epoch": 73.51195748449956,
      "grad_norm": 0.2604326009750366,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 82995
    },
    {
      "epoch": 73.51638618246236,
      "grad_norm": 0.25394269824028015,
      "learning_rate": 1e-05,
      "loss": 0.9189,
      "step": 83000
    },
    {
      "epoch": 73.52081488042515,
      "grad_norm": 0.22562089562416077,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 83005
    },
    {
      "epoch": 73.52524357838796,
      "grad_norm": 0.22459916770458221,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 83010
    },
    {
      "epoch": 73.52967227635075,
      "grad_norm": 0.23401667177677155,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 83015
    },
    {
      "epoch": 73.53410097431355,
      "grad_norm": 0.25175946950912476,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 83020
    },
    {
      "epoch": 73.53852967227635,
      "grad_norm": 0.20914360880851746,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 83025
    },
    {
      "epoch": 73.54295837023915,
      "grad_norm": 0.38618069887161255,
      "learning_rate": 1e-05,
      "loss": 0.9007,
      "step": 83030
    },
    {
      "epoch": 73.54738706820194,
      "grad_norm": 0.24266676604747772,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 83035
    },
    {
      "epoch": 73.55181576616475,
      "grad_norm": 0.24347826838493347,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 83040
    },
    {
      "epoch": 73.55624446412754,
      "grad_norm": 0.21379315853118896,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 83045
    },
    {
      "epoch": 73.56067316209034,
      "grad_norm": 0.23964732885360718,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 83050
    },
    {
      "epoch": 73.56510186005315,
      "grad_norm": 0.19100530445575714,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 83055
    },
    {
      "epoch": 73.56953055801594,
      "grad_norm": 0.237108513712883,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 83060
    },
    {
      "epoch": 73.57395925597874,
      "grad_norm": 0.25132131576538086,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 83065
    },
    {
      "epoch": 73.57838795394154,
      "grad_norm": 0.25894325971603394,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 83070
    },
    {
      "epoch": 73.58281665190434,
      "grad_norm": 0.2741222381591797,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 83075
    },
    {
      "epoch": 73.58724534986715,
      "grad_norm": 0.2366926670074463,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 83080
    },
    {
      "epoch": 73.59167404782994,
      "grad_norm": 0.2619396150112152,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 83085
    },
    {
      "epoch": 73.59610274579273,
      "grad_norm": 0.22304563224315643,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 83090
    },
    {
      "epoch": 73.60053144375554,
      "grad_norm": 0.20240060985088348,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 83095
    },
    {
      "epoch": 73.60496014171834,
      "grad_norm": 0.2572607100009918,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 83100
    },
    {
      "epoch": 73.60938883968113,
      "grad_norm": 0.2057189792394638,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 83105
    },
    {
      "epoch": 73.61381753764394,
      "grad_norm": 0.20827379822731018,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 83110
    },
    {
      "epoch": 73.61824623560673,
      "grad_norm": 0.204784095287323,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 83115
    },
    {
      "epoch": 73.62267493356953,
      "grad_norm": 0.25376030802726746,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 83120
    },
    {
      "epoch": 73.62710363153234,
      "grad_norm": 0.26508691906929016,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 83125
    },
    {
      "epoch": 73.63153232949513,
      "grad_norm": 0.2635312080383301,
      "learning_rate": 1e-05,
      "loss": 0.9165,
      "step": 83130
    },
    {
      "epoch": 73.63596102745792,
      "grad_norm": 0.22795486450195312,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 83135
    },
    {
      "epoch": 73.64038972542073,
      "grad_norm": 0.2265150398015976,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 83140
    },
    {
      "epoch": 73.64481842338353,
      "grad_norm": 0.21949546039104462,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 83145
    },
    {
      "epoch": 73.64924712134632,
      "grad_norm": 0.23996548354625702,
      "learning_rate": 1e-05,
      "loss": 0.9265,
      "step": 83150
    },
    {
      "epoch": 73.65367581930913,
      "grad_norm": 0.24817617237567902,
      "learning_rate": 1e-05,
      "loss": 0.9252,
      "step": 83155
    },
    {
      "epoch": 73.65810451727192,
      "grad_norm": 0.24546799063682556,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 83160
    },
    {
      "epoch": 73.66253321523472,
      "grad_norm": 0.2327755242586136,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 83165
    },
    {
      "epoch": 73.66696191319753,
      "grad_norm": 0.2609008550643921,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 83170
    },
    {
      "epoch": 73.67139061116032,
      "grad_norm": 0.2279651015996933,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 83175
    },
    {
      "epoch": 73.67581930912311,
      "grad_norm": 0.2589198648929596,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 83180
    },
    {
      "epoch": 73.68024800708592,
      "grad_norm": 0.2010425478219986,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 83185
    },
    {
      "epoch": 73.68467670504872,
      "grad_norm": 0.20207096636295319,
      "learning_rate": 1e-05,
      "loss": 0.9112,
      "step": 83190
    },
    {
      "epoch": 73.68910540301151,
      "grad_norm": 0.24023544788360596,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 83195
    },
    {
      "epoch": 73.69353410097432,
      "grad_norm": 0.21951812505722046,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 83200
    },
    {
      "epoch": 73.69796279893711,
      "grad_norm": 0.24895143508911133,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 83205
    },
    {
      "epoch": 73.7023914968999,
      "grad_norm": 0.22392037510871887,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 83210
    },
    {
      "epoch": 73.70682019486271,
      "grad_norm": 0.2230488806962967,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 83215
    },
    {
      "epoch": 73.71124889282551,
      "grad_norm": 0.25851505994796753,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 83220
    },
    {
      "epoch": 73.7156775907883,
      "grad_norm": 0.2613077461719513,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 83225
    },
    {
      "epoch": 73.72010628875111,
      "grad_norm": 0.25208404660224915,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 83230
    },
    {
      "epoch": 73.7245349867139,
      "grad_norm": 0.2495100200176239,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 83235
    },
    {
      "epoch": 73.7289636846767,
      "grad_norm": 0.2525523602962494,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 83240
    },
    {
      "epoch": 73.73339238263951,
      "grad_norm": 0.21497578918933868,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 83245
    },
    {
      "epoch": 73.7378210806023,
      "grad_norm": 0.2412101775407791,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 83250
    },
    {
      "epoch": 73.7422497785651,
      "grad_norm": 0.27067261934280396,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 83255
    },
    {
      "epoch": 73.7466784765279,
      "grad_norm": 0.27735018730163574,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 83260
    },
    {
      "epoch": 73.7511071744907,
      "grad_norm": 0.22332754731178284,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 83265
    },
    {
      "epoch": 73.75553587245349,
      "grad_norm": 0.21955981850624084,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 83270
    },
    {
      "epoch": 73.7599645704163,
      "grad_norm": 0.24124151468276978,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 83275
    },
    {
      "epoch": 73.7643932683791,
      "grad_norm": 0.2841683626174927,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 83280
    },
    {
      "epoch": 73.76882196634189,
      "grad_norm": 0.2517390251159668,
      "learning_rate": 1e-05,
      "loss": 1.0646,
      "step": 83285
    },
    {
      "epoch": 73.7732506643047,
      "grad_norm": 0.25931620597839355,
      "learning_rate": 1e-05,
      "loss": 0.8927,
      "step": 83290
    },
    {
      "epoch": 73.77767936226749,
      "grad_norm": 0.23221397399902344,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 83295
    },
    {
      "epoch": 73.78210806023029,
      "grad_norm": 0.2714071571826935,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 83300
    },
    {
      "epoch": 73.7865367581931,
      "grad_norm": 0.25025326013565063,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 83305
    },
    {
      "epoch": 73.79096545615589,
      "grad_norm": 0.23740044236183167,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 83310
    },
    {
      "epoch": 73.79539415411868,
      "grad_norm": 0.2720984220504761,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 83315
    },
    {
      "epoch": 73.79982285208149,
      "grad_norm": 0.22173546254634857,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 83320
    },
    {
      "epoch": 73.80425155004428,
      "grad_norm": 0.26164770126342773,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 83325
    },
    {
      "epoch": 73.8086802480071,
      "grad_norm": 0.23208299279212952,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 83330
    },
    {
      "epoch": 73.81310894596989,
      "grad_norm": 0.2349410206079483,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 83335
    },
    {
      "epoch": 73.81753764393268,
      "grad_norm": 0.26026564836502075,
      "learning_rate": 1e-05,
      "loss": 0.9221,
      "step": 83340
    },
    {
      "epoch": 73.82196634189549,
      "grad_norm": 0.24884694814682007,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 83345
    },
    {
      "epoch": 73.82639503985828,
      "grad_norm": 0.2892662584781647,
      "learning_rate": 1e-05,
      "loss": 0.9267,
      "step": 83350
    },
    {
      "epoch": 73.83082373782108,
      "grad_norm": 0.26073071360588074,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 83355
    },
    {
      "epoch": 73.83525243578389,
      "grad_norm": 0.27598854899406433,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 83360
    },
    {
      "epoch": 73.83968113374668,
      "grad_norm": 0.22965501248836517,
      "learning_rate": 1e-05,
      "loss": 0.8721,
      "step": 83365
    },
    {
      "epoch": 73.84410983170947,
      "grad_norm": 0.21010957658290863,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 83370
    },
    {
      "epoch": 73.84853852967228,
      "grad_norm": 0.22666208446025848,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 83375
    },
    {
      "epoch": 73.85296722763508,
      "grad_norm": 0.2133139967918396,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 83380
    },
    {
      "epoch": 73.85739592559787,
      "grad_norm": 0.21636372804641724,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 83385
    },
    {
      "epoch": 73.86182462356068,
      "grad_norm": 0.23691928386688232,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 83390
    },
    {
      "epoch": 73.86625332152347,
      "grad_norm": 0.2837972939014435,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 83395
    },
    {
      "epoch": 73.87068201948627,
      "grad_norm": 0.24087807536125183,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 83400
    },
    {
      "epoch": 73.87511071744908,
      "grad_norm": 0.2588447630405426,
      "learning_rate": 1e-05,
      "loss": 0.9179,
      "step": 83405
    },
    {
      "epoch": 73.87953941541187,
      "grad_norm": 0.2112160474061966,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 83410
    },
    {
      "epoch": 73.88396811337466,
      "grad_norm": 0.2502126395702362,
      "learning_rate": 1e-05,
      "loss": 0.9367,
      "step": 83415
    },
    {
      "epoch": 73.88839681133747,
      "grad_norm": 0.250557005405426,
      "learning_rate": 1e-05,
      "loss": 0.8922,
      "step": 83420
    },
    {
      "epoch": 73.89282550930027,
      "grad_norm": 0.23844382166862488,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 83425
    },
    {
      "epoch": 73.89725420726306,
      "grad_norm": 0.24517607688903809,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 83430
    },
    {
      "epoch": 73.90168290522587,
      "grad_norm": 0.2454157918691635,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 83435
    },
    {
      "epoch": 73.90611160318866,
      "grad_norm": 0.20270434021949768,
      "learning_rate": 1e-05,
      "loss": 0.9296,
      "step": 83440
    },
    {
      "epoch": 73.91054030115146,
      "grad_norm": 0.23969697952270508,
      "learning_rate": 1e-05,
      "loss": 0.93,
      "step": 83445
    },
    {
      "epoch": 73.91496899911427,
      "grad_norm": 0.2249893993139267,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 83450
    },
    {
      "epoch": 73.91939769707706,
      "grad_norm": 0.24709507822990417,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 83455
    },
    {
      "epoch": 73.92382639503985,
      "grad_norm": 0.22869141399860382,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 83460
    },
    {
      "epoch": 73.92825509300266,
      "grad_norm": 0.2149641215801239,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 83465
    },
    {
      "epoch": 73.93268379096546,
      "grad_norm": 0.2631613314151764,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 83470
    },
    {
      "epoch": 73.93711248892825,
      "grad_norm": 0.21642833948135376,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 83475
    },
    {
      "epoch": 73.94154118689106,
      "grad_norm": 0.21103224158287048,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 83480
    },
    {
      "epoch": 73.94596988485385,
      "grad_norm": 0.20202122628688812,
      "learning_rate": 1e-05,
      "loss": 1.0347,
      "step": 83485
    },
    {
      "epoch": 73.95039858281665,
      "grad_norm": 0.23157206177711487,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 83490
    },
    {
      "epoch": 73.95482728077945,
      "grad_norm": 0.23936142027378082,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 83495
    },
    {
      "epoch": 73.95925597874225,
      "grad_norm": 0.19143910706043243,
      "learning_rate": 1e-05,
      "loss": 0.8995,
      "step": 83500
    },
    {
      "epoch": 73.96368467670504,
      "grad_norm": 0.297951877117157,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 83505
    },
    {
      "epoch": 73.96811337466785,
      "grad_norm": 0.26755356788635254,
      "learning_rate": 1e-05,
      "loss": 1.0491,
      "step": 83510
    },
    {
      "epoch": 73.97254207263065,
      "grad_norm": 0.24796029925346375,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 83515
    },
    {
      "epoch": 73.97697077059344,
      "grad_norm": 0.28519660234451294,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 83520
    },
    {
      "epoch": 73.98139946855625,
      "grad_norm": 0.24823503196239471,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 83525
    },
    {
      "epoch": 73.98582816651904,
      "grad_norm": 0.24633878469467163,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 83530
    },
    {
      "epoch": 73.99025686448184,
      "grad_norm": 0.22910018265247345,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 83535
    },
    {
      "epoch": 73.99468556244464,
      "grad_norm": 0.26459747552871704,
      "learning_rate": 1e-05,
      "loss": 0.9209,
      "step": 83540
    },
    {
      "epoch": 73.99911426040744,
      "grad_norm": 0.25838255882263184,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 83545
    },
    {
      "epoch": 74.00354295837023,
      "grad_norm": 0.21516035497188568,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 83550
    },
    {
      "epoch": 74.00797165633304,
      "grad_norm": 0.2380024641752243,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 83555
    },
    {
      "epoch": 74.01240035429583,
      "grad_norm": 0.25423529744148254,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 83560
    },
    {
      "epoch": 74.01682905225863,
      "grad_norm": 0.18841908872127533,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 83565
    },
    {
      "epoch": 74.02125775022144,
      "grad_norm": 0.22169649600982666,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 83570
    },
    {
      "epoch": 74.02568644818423,
      "grad_norm": 0.24355438351631165,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 83575
    },
    {
      "epoch": 74.03011514614704,
      "grad_norm": 0.2513810694217682,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 83580
    },
    {
      "epoch": 74.03454384410983,
      "grad_norm": 0.2404751479625702,
      "learning_rate": 1e-05,
      "loss": 0.919,
      "step": 83585
    },
    {
      "epoch": 74.03897254207263,
      "grad_norm": 0.21027123928070068,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 83590
    },
    {
      "epoch": 74.04340124003544,
      "grad_norm": 0.23689305782318115,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 83595
    },
    {
      "epoch": 74.04782993799823,
      "grad_norm": 0.2254077047109604,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 83600
    },
    {
      "epoch": 74.05225863596102,
      "grad_norm": 0.22020323574543,
      "learning_rate": 1e-05,
      "loss": 1.0195,
      "step": 83605
    },
    {
      "epoch": 74.05668733392383,
      "grad_norm": 0.24068717658519745,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 83610
    },
    {
      "epoch": 74.06111603188663,
      "grad_norm": 0.25655418634414673,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 83615
    },
    {
      "epoch": 74.06554472984942,
      "grad_norm": 0.23864029347896576,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 83620
    },
    {
      "epoch": 74.06997342781223,
      "grad_norm": 0.2524789869785309,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 83625
    },
    {
      "epoch": 74.07440212577502,
      "grad_norm": 0.244278222322464,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 83630
    },
    {
      "epoch": 74.07883082373782,
      "grad_norm": 0.2437145859003067,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 83635
    },
    {
      "epoch": 74.08325952170063,
      "grad_norm": 0.2703615427017212,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 83640
    },
    {
      "epoch": 74.08768821966342,
      "grad_norm": 0.22829675674438477,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 83645
    },
    {
      "epoch": 74.09211691762621,
      "grad_norm": 0.276937872171402,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 83650
    },
    {
      "epoch": 74.09654561558902,
      "grad_norm": 0.2446051687002182,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 83655
    },
    {
      "epoch": 74.10097431355182,
      "grad_norm": 0.26338323950767517,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 83660
    },
    {
      "epoch": 74.10540301151461,
      "grad_norm": 0.24262087047100067,
      "learning_rate": 1e-05,
      "loss": 0.9346,
      "step": 83665
    },
    {
      "epoch": 74.10983170947742,
      "grad_norm": 0.2427598237991333,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 83670
    },
    {
      "epoch": 74.11426040744021,
      "grad_norm": 0.19425921142101288,
      "learning_rate": 1e-05,
      "loss": 1.017,
      "step": 83675
    },
    {
      "epoch": 74.118689105403,
      "grad_norm": 0.23510712385177612,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 83680
    },
    {
      "epoch": 74.12311780336582,
      "grad_norm": 0.27288198471069336,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 83685
    },
    {
      "epoch": 74.12754650132861,
      "grad_norm": 0.2708515524864197,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 83690
    },
    {
      "epoch": 74.1319751992914,
      "grad_norm": 0.22248634696006775,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 83695
    },
    {
      "epoch": 74.13640389725421,
      "grad_norm": 0.26697608828544617,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 83700
    },
    {
      "epoch": 74.140832595217,
      "grad_norm": 0.2455085813999176,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 83705
    },
    {
      "epoch": 74.1452612931798,
      "grad_norm": 0.23500816524028778,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 83710
    },
    {
      "epoch": 74.14968999114261,
      "grad_norm": 0.2902398109436035,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 83715
    },
    {
      "epoch": 74.1541186891054,
      "grad_norm": 0.23216289281845093,
      "learning_rate": 1e-05,
      "loss": 0.9244,
      "step": 83720
    },
    {
      "epoch": 74.1585473870682,
      "grad_norm": 0.24378244578838348,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 83725
    },
    {
      "epoch": 74.162976085031,
      "grad_norm": 0.2358853667974472,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 83730
    },
    {
      "epoch": 74.1674047829938,
      "grad_norm": 0.2271830439567566,
      "learning_rate": 1e-05,
      "loss": 0.9116,
      "step": 83735
    },
    {
      "epoch": 74.1718334809566,
      "grad_norm": 0.21153266727924347,
      "learning_rate": 1e-05,
      "loss": 0.9009,
      "step": 83740
    },
    {
      "epoch": 74.1762621789194,
      "grad_norm": 0.24842631816864014,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 83745
    },
    {
      "epoch": 74.1806908768822,
      "grad_norm": 0.2902752757072449,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 83750
    },
    {
      "epoch": 74.18511957484499,
      "grad_norm": 0.23537321388721466,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 83755
    },
    {
      "epoch": 74.1895482728078,
      "grad_norm": 0.25719574093818665,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 83760
    },
    {
      "epoch": 74.19397697077059,
      "grad_norm": 0.229341059923172,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 83765
    },
    {
      "epoch": 74.19840566873339,
      "grad_norm": 0.253781795501709,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 83770
    },
    {
      "epoch": 74.2028343666962,
      "grad_norm": 0.22993256151676178,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 83775
    },
    {
      "epoch": 74.20726306465899,
      "grad_norm": 0.22871562838554382,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 83780
    },
    {
      "epoch": 74.21169176262178,
      "grad_norm": 0.24034012854099274,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 83785
    },
    {
      "epoch": 74.21612046058459,
      "grad_norm": 0.23713849484920502,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 83790
    },
    {
      "epoch": 74.22054915854739,
      "grad_norm": 0.24475929141044617,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 83795
    },
    {
      "epoch": 74.22497785651018,
      "grad_norm": 0.2614419758319855,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 83800
    },
    {
      "epoch": 74.22940655447299,
      "grad_norm": 0.2253301590681076,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 83805
    },
    {
      "epoch": 74.23383525243578,
      "grad_norm": 0.2708972990512848,
      "learning_rate": 1e-05,
      "loss": 0.9138,
      "step": 83810
    },
    {
      "epoch": 74.23826395039858,
      "grad_norm": 0.24148797988891602,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 83815
    },
    {
      "epoch": 74.24269264836138,
      "grad_norm": 0.2569144070148468,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 83820
    },
    {
      "epoch": 74.24712134632418,
      "grad_norm": 0.2280445098876953,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 83825
    },
    {
      "epoch": 74.25155004428699,
      "grad_norm": 0.2669070363044739,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 83830
    },
    {
      "epoch": 74.25597874224978,
      "grad_norm": 0.21201454102993011,
      "learning_rate": 1e-05,
      "loss": 0.9236,
      "step": 83835
    },
    {
      "epoch": 74.26040744021257,
      "grad_norm": 0.27060744166374207,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 83840
    },
    {
      "epoch": 74.26483613817538,
      "grad_norm": 0.2669195234775543,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 83845
    },
    {
      "epoch": 74.26926483613818,
      "grad_norm": 0.30005455017089844,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 83850
    },
    {
      "epoch": 74.27369353410097,
      "grad_norm": 0.2485140711069107,
      "learning_rate": 1e-05,
      "loss": 0.9233,
      "step": 83855
    },
    {
      "epoch": 74.27812223206378,
      "grad_norm": 0.23749056458473206,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 83860
    },
    {
      "epoch": 74.28255093002657,
      "grad_norm": 0.29755693674087524,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 83865
    },
    {
      "epoch": 74.28697962798937,
      "grad_norm": 0.2690080404281616,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 83870
    },
    {
      "epoch": 74.29140832595218,
      "grad_norm": 0.26526302099227905,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 83875
    },
    {
      "epoch": 74.29583702391497,
      "grad_norm": 0.22737567126750946,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 83880
    },
    {
      "epoch": 74.30026572187776,
      "grad_norm": 0.20818877220153809,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 83885
    },
    {
      "epoch": 74.30469441984057,
      "grad_norm": 0.23308666050434113,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 83890
    },
    {
      "epoch": 74.30912311780337,
      "grad_norm": 0.2413572072982788,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 83895
    },
    {
      "epoch": 74.31355181576616,
      "grad_norm": 0.23688402771949768,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 83900
    },
    {
      "epoch": 74.31798051372897,
      "grad_norm": 0.24354438483715057,
      "learning_rate": 1e-05,
      "loss": 0.8995,
      "step": 83905
    },
    {
      "epoch": 74.32240921169176,
      "grad_norm": 0.22352060675621033,
      "learning_rate": 1e-05,
      "loss": 0.9157,
      "step": 83910
    },
    {
      "epoch": 74.32683790965456,
      "grad_norm": 0.247068390250206,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 83915
    },
    {
      "epoch": 74.33126660761737,
      "grad_norm": 0.23777520656585693,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 83920
    },
    {
      "epoch": 74.33569530558016,
      "grad_norm": 0.2843903601169586,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 83925
    },
    {
      "epoch": 74.34012400354295,
      "grad_norm": 0.2221992164850235,
      "learning_rate": 1e-05,
      "loss": 0.9246,
      "step": 83930
    },
    {
      "epoch": 74.34455270150576,
      "grad_norm": 0.22756318747997284,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 83935
    },
    {
      "epoch": 74.34898139946856,
      "grad_norm": 0.2188238501548767,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 83940
    },
    {
      "epoch": 74.35341009743135,
      "grad_norm": 0.23725594580173492,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 83945
    },
    {
      "epoch": 74.35783879539416,
      "grad_norm": 0.23624736070632935,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 83950
    },
    {
      "epoch": 74.36226749335695,
      "grad_norm": 0.22528259456157684,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 83955
    },
    {
      "epoch": 74.36669619131975,
      "grad_norm": 0.21942612528800964,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 83960
    },
    {
      "epoch": 74.37112488928256,
      "grad_norm": 0.24189907312393188,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 83965
    },
    {
      "epoch": 74.37555358724535,
      "grad_norm": 0.2378605455160141,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 83970
    },
    {
      "epoch": 74.37998228520814,
      "grad_norm": 0.23476238548755646,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 83975
    },
    {
      "epoch": 74.38441098317095,
      "grad_norm": 0.25473737716674805,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 83980
    },
    {
      "epoch": 74.38883968113375,
      "grad_norm": 0.27539709210395813,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 83985
    },
    {
      "epoch": 74.39326837909654,
      "grad_norm": 0.21888090670108795,
      "learning_rate": 1e-05,
      "loss": 0.9162,
      "step": 83990
    },
    {
      "epoch": 74.39769707705935,
      "grad_norm": 0.28328707814216614,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 83995
    },
    {
      "epoch": 74.40212577502214,
      "grad_norm": 0.2491467148065567,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 84000
    },
    {
      "epoch": 74.40655447298494,
      "grad_norm": 0.24903646111488342,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 84005
    },
    {
      "epoch": 74.41098317094774,
      "grad_norm": 0.2287069708108902,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 84010
    },
    {
      "epoch": 74.41541186891054,
      "grad_norm": 0.24306733906269073,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 84015
    },
    {
      "epoch": 74.41984056687333,
      "grad_norm": 0.22020022571086884,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 84020
    },
    {
      "epoch": 74.42426926483614,
      "grad_norm": 0.2536507546901703,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 84025
    },
    {
      "epoch": 74.42869796279894,
      "grad_norm": 0.20949707925319672,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 84030
    },
    {
      "epoch": 74.43312666076173,
      "grad_norm": 0.2516788840293884,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 84035
    },
    {
      "epoch": 74.43755535872454,
      "grad_norm": 0.25324663519859314,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 84040
    },
    {
      "epoch": 74.44198405668733,
      "grad_norm": 0.22406317293643951,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 84045
    },
    {
      "epoch": 74.44641275465013,
      "grad_norm": 0.25456079840660095,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 84050
    },
    {
      "epoch": 74.45084145261293,
      "grad_norm": 0.23706777393817902,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 84055
    },
    {
      "epoch": 74.45527015057573,
      "grad_norm": 0.2422616183757782,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 84060
    },
    {
      "epoch": 74.45969884853854,
      "grad_norm": 0.2437332719564438,
      "learning_rate": 1e-05,
      "loss": 0.9004,
      "step": 84065
    },
    {
      "epoch": 74.46412754650133,
      "grad_norm": 0.25202903151512146,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 84070
    },
    {
      "epoch": 74.46855624446412,
      "grad_norm": 0.25213393568992615,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 84075
    },
    {
      "epoch": 74.47298494242693,
      "grad_norm": 0.2552889585494995,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 84080
    },
    {
      "epoch": 74.47741364038973,
      "grad_norm": 0.23778022825717926,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 84085
    },
    {
      "epoch": 74.48184233835252,
      "grad_norm": 0.22344079613685608,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 84090
    },
    {
      "epoch": 74.48627103631533,
      "grad_norm": 0.25441935658454895,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 84095
    },
    {
      "epoch": 74.49069973427812,
      "grad_norm": 0.2612452208995819,
      "learning_rate": 1e-05,
      "loss": 0.9318,
      "step": 84100
    },
    {
      "epoch": 74.49512843224092,
      "grad_norm": 0.2253248542547226,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 84105
    },
    {
      "epoch": 74.49955713020373,
      "grad_norm": 0.24539245665073395,
      "learning_rate": 1e-05,
      "loss": 1.0394,
      "step": 84110
    },
    {
      "epoch": 74.50398582816652,
      "grad_norm": 0.23280608654022217,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 84115
    },
    {
      "epoch": 74.50841452612931,
      "grad_norm": 0.22873243689537048,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 84120
    },
    {
      "epoch": 74.51284322409212,
      "grad_norm": 0.2524687647819519,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 84125
    },
    {
      "epoch": 74.51727192205492,
      "grad_norm": 0.22405165433883667,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 84130
    },
    {
      "epoch": 74.52170062001771,
      "grad_norm": 0.24014557898044586,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 84135
    },
    {
      "epoch": 74.52612931798052,
      "grad_norm": 0.2226085066795349,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 84140
    },
    {
      "epoch": 74.53055801594331,
      "grad_norm": 0.25360554456710815,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 84145
    },
    {
      "epoch": 74.53498671390611,
      "grad_norm": 0.2232673466205597,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 84150
    },
    {
      "epoch": 74.53941541186892,
      "grad_norm": 0.28131598234176636,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 84155
    },
    {
      "epoch": 74.54384410983171,
      "grad_norm": 0.23310501873493195,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 84160
    },
    {
      "epoch": 74.5482728077945,
      "grad_norm": 0.21536749601364136,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 84165
    },
    {
      "epoch": 74.55270150575731,
      "grad_norm": 0.2521098554134369,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 84170
    },
    {
      "epoch": 74.5571302037201,
      "grad_norm": 0.26201021671295166,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 84175
    },
    {
      "epoch": 74.5615589016829,
      "grad_norm": 0.23942938446998596,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 84180
    },
    {
      "epoch": 74.56598759964571,
      "grad_norm": 0.23541148006916046,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 84185
    },
    {
      "epoch": 74.5704162976085,
      "grad_norm": 0.240321084856987,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 84190
    },
    {
      "epoch": 74.5748449955713,
      "grad_norm": 0.2605343759059906,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 84195
    },
    {
      "epoch": 74.5792736935341,
      "grad_norm": 0.28967800736427307,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 84200
    },
    {
      "epoch": 74.5837023914969,
      "grad_norm": 0.25075262784957886,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 84205
    },
    {
      "epoch": 74.5881310894597,
      "grad_norm": 0.2145904004573822,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 84210
    },
    {
      "epoch": 74.5925597874225,
      "grad_norm": 0.221253901720047,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 84215
    },
    {
      "epoch": 74.5969884853853,
      "grad_norm": 0.23714208602905273,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 84220
    },
    {
      "epoch": 74.60141718334809,
      "grad_norm": 0.24737609922885895,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 84225
    },
    {
      "epoch": 74.6058458813109,
      "grad_norm": 0.1893225461244583,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 84230
    },
    {
      "epoch": 74.61027457927369,
      "grad_norm": 0.27101677656173706,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 84235
    },
    {
      "epoch": 74.61470327723649,
      "grad_norm": 0.26572537422180176,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 84240
    },
    {
      "epoch": 74.6191319751993,
      "grad_norm": 0.2822604179382324,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 84245
    },
    {
      "epoch": 74.62356067316209,
      "grad_norm": 0.2608496844768524,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 84250
    },
    {
      "epoch": 74.62798937112488,
      "grad_norm": 0.24638453125953674,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 84255
    },
    {
      "epoch": 74.63241806908769,
      "grad_norm": 0.2655778229236603,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 84260
    },
    {
      "epoch": 74.63684676705049,
      "grad_norm": 0.2132103443145752,
      "learning_rate": 1e-05,
      "loss": 0.9175,
      "step": 84265
    },
    {
      "epoch": 74.64127546501328,
      "grad_norm": 0.27513816952705383,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 84270
    },
    {
      "epoch": 74.64570416297609,
      "grad_norm": 0.23372705280780792,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 84275
    },
    {
      "epoch": 74.65013286093888,
      "grad_norm": 0.2767297029495239,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 84280
    },
    {
      "epoch": 74.65456155890168,
      "grad_norm": 0.2111191302537918,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 84285
    },
    {
      "epoch": 74.65899025686448,
      "grad_norm": 0.21248102188110352,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 84290
    },
    {
      "epoch": 74.66341895482728,
      "grad_norm": 0.21242599189281464,
      "learning_rate": 1e-05,
      "loss": 0.9413,
      "step": 84295
    },
    {
      "epoch": 74.66784765279007,
      "grad_norm": 0.28285810351371765,
      "learning_rate": 1e-05,
      "loss": 0.9275,
      "step": 84300
    },
    {
      "epoch": 74.67227635075288,
      "grad_norm": 0.22369693219661713,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 84305
    },
    {
      "epoch": 74.67670504871568,
      "grad_norm": 0.218536838889122,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 84310
    },
    {
      "epoch": 74.68113374667848,
      "grad_norm": 0.23357917368412018,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 84315
    },
    {
      "epoch": 74.68556244464128,
      "grad_norm": 0.22577446699142456,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 84320
    },
    {
      "epoch": 74.68999114260407,
      "grad_norm": 0.3132842481136322,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 84325
    },
    {
      "epoch": 74.69441984056688,
      "grad_norm": 0.23717401921749115,
      "learning_rate": 1e-05,
      "loss": 0.9253,
      "step": 84330
    },
    {
      "epoch": 74.69884853852967,
      "grad_norm": 0.26239141821861267,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 84335
    },
    {
      "epoch": 74.70327723649247,
      "grad_norm": 0.2510971128940582,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 84340
    },
    {
      "epoch": 74.70770593445528,
      "grad_norm": 0.24655738472938538,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 84345
    },
    {
      "epoch": 74.71213463241807,
      "grad_norm": 0.2461847960948944,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 84350
    },
    {
      "epoch": 74.71656333038086,
      "grad_norm": 0.22440913319587708,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 84355
    },
    {
      "epoch": 74.72099202834367,
      "grad_norm": 0.22830170392990112,
      "learning_rate": 1e-05,
      "loss": 1.0138,
      "step": 84360
    },
    {
      "epoch": 74.72542072630647,
      "grad_norm": 0.2553552985191345,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 84365
    },
    {
      "epoch": 74.72984942426926,
      "grad_norm": 0.21517328917980194,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 84370
    },
    {
      "epoch": 74.73427812223207,
      "grad_norm": 0.238071009516716,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 84375
    },
    {
      "epoch": 74.73870682019486,
      "grad_norm": 0.21605949103832245,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 84380
    },
    {
      "epoch": 74.74313551815766,
      "grad_norm": 0.22177082300186157,
      "learning_rate": 1e-05,
      "loss": 1.0396,
      "step": 84385
    },
    {
      "epoch": 74.74756421612047,
      "grad_norm": 0.22120670974254608,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 84390
    },
    {
      "epoch": 74.75199291408326,
      "grad_norm": 0.23397816717624664,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 84395
    },
    {
      "epoch": 74.75642161204605,
      "grad_norm": 0.24542438983917236,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 84400
    },
    {
      "epoch": 74.76085031000886,
      "grad_norm": 0.21681912243366241,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 84405
    },
    {
      "epoch": 74.76527900797166,
      "grad_norm": 0.24025574326515198,
      "learning_rate": 1e-05,
      "loss": 0.921,
      "step": 84410
    },
    {
      "epoch": 74.76970770593445,
      "grad_norm": 0.23549969494342804,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 84415
    },
    {
      "epoch": 74.77413640389726,
      "grad_norm": 0.26955610513687134,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 84420
    },
    {
      "epoch": 74.77856510186005,
      "grad_norm": 0.2821662724018097,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 84425
    },
    {
      "epoch": 74.78299379982285,
      "grad_norm": 0.2508194148540497,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 84430
    },
    {
      "epoch": 74.78742249778566,
      "grad_norm": 0.2517653703689575,
      "learning_rate": 1e-05,
      "loss": 0.8962,
      "step": 84435
    },
    {
      "epoch": 74.79185119574845,
      "grad_norm": 0.23977205157279968,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 84440
    },
    {
      "epoch": 74.79627989371124,
      "grad_norm": 0.2645267844200134,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 84445
    },
    {
      "epoch": 74.80070859167405,
      "grad_norm": 0.21609662473201752,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 84450
    },
    {
      "epoch": 74.80513728963685,
      "grad_norm": 0.25307223200798035,
      "learning_rate": 1e-05,
      "loss": 0.8657,
      "step": 84455
    },
    {
      "epoch": 74.80956598759964,
      "grad_norm": 0.27134793996810913,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 84460
    },
    {
      "epoch": 74.81399468556245,
      "grad_norm": 0.24788208305835724,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 84465
    },
    {
      "epoch": 74.81842338352524,
      "grad_norm": 0.22885899245738983,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 84470
    },
    {
      "epoch": 74.82285208148804,
      "grad_norm": 0.23829950392246246,
      "learning_rate": 1e-05,
      "loss": 1.0304,
      "step": 84475
    },
    {
      "epoch": 74.82728077945085,
      "grad_norm": 0.2025105059146881,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 84480
    },
    {
      "epoch": 74.83170947741364,
      "grad_norm": 0.25875964760780334,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 84485
    },
    {
      "epoch": 74.83613817537643,
      "grad_norm": 0.23047824203968048,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 84490
    },
    {
      "epoch": 74.84056687333924,
      "grad_norm": 0.1988583207130432,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 84495
    },
    {
      "epoch": 74.84499557130204,
      "grad_norm": 0.21686246991157532,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 84500
    },
    {
      "epoch": 74.84942426926483,
      "grad_norm": 0.24168746173381805,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 84505
    },
    {
      "epoch": 74.85385296722764,
      "grad_norm": 0.21927835047245026,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 84510
    },
    {
      "epoch": 74.85828166519043,
      "grad_norm": 0.25283053517341614,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 84515
    },
    {
      "epoch": 74.86271036315323,
      "grad_norm": 0.2474817931652069,
      "learning_rate": 1e-05,
      "loss": 0.921,
      "step": 84520
    },
    {
      "epoch": 74.86713906111603,
      "grad_norm": 0.2392907589673996,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 84525
    },
    {
      "epoch": 74.87156775907883,
      "grad_norm": 0.2124769389629364,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 84530
    },
    {
      "epoch": 74.87599645704162,
      "grad_norm": 0.24759037792682648,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 84535
    },
    {
      "epoch": 74.88042515500443,
      "grad_norm": 0.22522376477718353,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 84540
    },
    {
      "epoch": 74.88485385296723,
      "grad_norm": 0.21356217563152313,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 84545
    },
    {
      "epoch": 74.88928255093003,
      "grad_norm": 0.2679639756679535,
      "learning_rate": 1e-05,
      "loss": 0.9564,
      "step": 84550
    },
    {
      "epoch": 74.89371124889283,
      "grad_norm": 0.25876304507255554,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 84555
    },
    {
      "epoch": 74.89813994685562,
      "grad_norm": 0.22637389600276947,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 84560
    },
    {
      "epoch": 74.90256864481843,
      "grad_norm": 0.21849526464939117,
      "learning_rate": 1e-05,
      "loss": 0.9307,
      "step": 84565
    },
    {
      "epoch": 74.90699734278122,
      "grad_norm": 0.25481897592544556,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 84570
    },
    {
      "epoch": 74.91142604074402,
      "grad_norm": 0.23954732716083527,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 84575
    },
    {
      "epoch": 74.91585473870683,
      "grad_norm": 0.2385462075471878,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 84580
    },
    {
      "epoch": 74.92028343666962,
      "grad_norm": 0.24303081631660461,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 84585
    },
    {
      "epoch": 74.92471213463241,
      "grad_norm": 0.24699783325195312,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 84590
    },
    {
      "epoch": 74.92914083259522,
      "grad_norm": 0.21922394633293152,
      "learning_rate": 1e-05,
      "loss": 0.9075,
      "step": 84595
    },
    {
      "epoch": 74.93356953055802,
      "grad_norm": 0.22815263271331787,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 84600
    },
    {
      "epoch": 74.93799822852081,
      "grad_norm": 0.23953737318515778,
      "learning_rate": 1e-05,
      "loss": 0.9235,
      "step": 84605
    },
    {
      "epoch": 74.94242692648362,
      "grad_norm": 0.23508131504058838,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 84610
    },
    {
      "epoch": 74.94685562444641,
      "grad_norm": 0.288730263710022,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 84615
    },
    {
      "epoch": 74.95128432240921,
      "grad_norm": 0.2662127912044525,
      "learning_rate": 1e-05,
      "loss": 0.9333,
      "step": 84620
    },
    {
      "epoch": 74.95571302037202,
      "grad_norm": 0.3024565577507019,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 84625
    },
    {
      "epoch": 74.96014171833481,
      "grad_norm": 0.21293513476848602,
      "learning_rate": 1e-05,
      "loss": 0.9043,
      "step": 84630
    },
    {
      "epoch": 74.9645704162976,
      "grad_norm": 0.1992969512939453,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 84635
    },
    {
      "epoch": 74.96899911426041,
      "grad_norm": 0.24272489547729492,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 84640
    },
    {
      "epoch": 74.9734278122232,
      "grad_norm": 0.26822975277900696,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 84645
    },
    {
      "epoch": 74.977856510186,
      "grad_norm": 0.23995016515254974,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 84650
    },
    {
      "epoch": 74.98228520814881,
      "grad_norm": 0.2250189483165741,
      "learning_rate": 1e-05,
      "loss": 0.9169,
      "step": 84655
    },
    {
      "epoch": 74.9867139061116,
      "grad_norm": 0.22450236976146698,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 84660
    },
    {
      "epoch": 74.9911426040744,
      "grad_norm": 0.2503156065940857,
      "learning_rate": 1e-05,
      "loss": 0.9183,
      "step": 84665
    },
    {
      "epoch": 74.9955713020372,
      "grad_norm": 0.2240157276391983,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 84670
    },
    {
      "epoch": 75.0,
      "grad_norm": 0.22003693878650665,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 84675
    },
    {
      "epoch": 75.0044286979628,
      "grad_norm": 0.21870656311511993,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 84680
    },
    {
      "epoch": 75.0088573959256,
      "grad_norm": 0.22483786940574646,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 84685
    },
    {
      "epoch": 75.0132860938884,
      "grad_norm": 0.22884005308151245,
      "learning_rate": 1e-05,
      "loss": 0.9191,
      "step": 84690
    },
    {
      "epoch": 75.01771479185119,
      "grad_norm": 0.28699758648872375,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 84695
    },
    {
      "epoch": 75.022143489814,
      "grad_norm": 0.25010424852371216,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 84700
    },
    {
      "epoch": 75.0265721877768,
      "grad_norm": 0.22076232731342316,
      "learning_rate": 1e-05,
      "loss": 0.8877,
      "step": 84705
    },
    {
      "epoch": 75.03100088573959,
      "grad_norm": 0.22312040627002716,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 84710
    },
    {
      "epoch": 75.0354295837024,
      "grad_norm": 0.2565702497959137,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 84715
    },
    {
      "epoch": 75.03985828166519,
      "grad_norm": 0.21622341871261597,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 84720
    },
    {
      "epoch": 75.04428697962798,
      "grad_norm": 0.23020702600479126,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 84725
    },
    {
      "epoch": 75.04871567759079,
      "grad_norm": 0.26327991485595703,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 84730
    },
    {
      "epoch": 75.05314437555359,
      "grad_norm": 0.22826343774795532,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 84735
    },
    {
      "epoch": 75.05757307351638,
      "grad_norm": 0.20753420889377594,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 84740
    },
    {
      "epoch": 75.06200177147919,
      "grad_norm": 0.2340545803308487,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 84745
    },
    {
      "epoch": 75.06643046944198,
      "grad_norm": 0.2826903462409973,
      "learning_rate": 1e-05,
      "loss": 0.9152,
      "step": 84750
    },
    {
      "epoch": 75.07085916740478,
      "grad_norm": 0.25694406032562256,
      "learning_rate": 1e-05,
      "loss": 1.0487,
      "step": 84755
    },
    {
      "epoch": 75.07528786536759,
      "grad_norm": 0.23377755284309387,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 84760
    },
    {
      "epoch": 75.07971656333038,
      "grad_norm": 0.22043955326080322,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 84765
    },
    {
      "epoch": 75.08414526129317,
      "grad_norm": 0.23454740643501282,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 84770
    },
    {
      "epoch": 75.08857395925598,
      "grad_norm": 0.27510154247283936,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 84775
    },
    {
      "epoch": 75.09300265721878,
      "grad_norm": 0.23960064351558685,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 84780
    },
    {
      "epoch": 75.09743135518157,
      "grad_norm": 0.24262239038944244,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 84785
    },
    {
      "epoch": 75.10186005314438,
      "grad_norm": 0.26969027519226074,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 84790
    },
    {
      "epoch": 75.10628875110717,
      "grad_norm": 0.24327705800533295,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 84795
    },
    {
      "epoch": 75.11071744906998,
      "grad_norm": 0.22460506856441498,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 84800
    },
    {
      "epoch": 75.11514614703277,
      "grad_norm": 0.2183789163827896,
      "learning_rate": 1e-05,
      "loss": 0.9408,
      "step": 84805
    },
    {
      "epoch": 75.11957484499557,
      "grad_norm": 0.25380414724349976,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 84810
    },
    {
      "epoch": 75.12400354295838,
      "grad_norm": 0.2654682993888855,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 84815
    },
    {
      "epoch": 75.12843224092117,
      "grad_norm": 0.1973486691713333,
      "learning_rate": 1e-05,
      "loss": 0.9097,
      "step": 84820
    },
    {
      "epoch": 75.13286093888397,
      "grad_norm": 0.2441629022359848,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 84825
    },
    {
      "epoch": 75.13728963684677,
      "grad_norm": 0.24071498215198517,
      "learning_rate": 1e-05,
      "loss": 0.9181,
      "step": 84830
    },
    {
      "epoch": 75.14171833480957,
      "grad_norm": 0.31346943974494934,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 84835
    },
    {
      "epoch": 75.14614703277236,
      "grad_norm": 0.308391273021698,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 84840
    },
    {
      "epoch": 75.15057573073517,
      "grad_norm": 0.2587350606918335,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 84845
    },
    {
      "epoch": 75.15500442869796,
      "grad_norm": 0.20671920478343964,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 84850
    },
    {
      "epoch": 75.15943312666076,
      "grad_norm": 0.25550347566604614,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 84855
    },
    {
      "epoch": 75.16386182462357,
      "grad_norm": 0.25200262665748596,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 84860
    },
    {
      "epoch": 75.16829052258636,
      "grad_norm": 0.29642894864082336,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 84865
    },
    {
      "epoch": 75.17271922054915,
      "grad_norm": 0.2914987802505493,
      "learning_rate": 1e-05,
      "loss": 1.0135,
      "step": 84870
    },
    {
      "epoch": 75.17714791851196,
      "grad_norm": 0.2569211423397064,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 84875
    },
    {
      "epoch": 75.18157661647476,
      "grad_norm": 0.2678290605545044,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 84880
    },
    {
      "epoch": 75.18600531443755,
      "grad_norm": 0.24707165360450745,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 84885
    },
    {
      "epoch": 75.19043401240036,
      "grad_norm": 0.28388920426368713,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 84890
    },
    {
      "epoch": 75.19486271036315,
      "grad_norm": 0.2352336347103119,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 84895
    },
    {
      "epoch": 75.19929140832595,
      "grad_norm": 0.2437903881072998,
      "learning_rate": 1e-05,
      "loss": 0.9346,
      "step": 84900
    },
    {
      "epoch": 75.20372010628876,
      "grad_norm": 0.20318947732448578,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 84905
    },
    {
      "epoch": 75.20814880425155,
      "grad_norm": 0.275124192237854,
      "learning_rate": 1e-05,
      "loss": 0.9317,
      "step": 84910
    },
    {
      "epoch": 75.21257750221434,
      "grad_norm": 0.2471960484981537,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 84915
    },
    {
      "epoch": 75.21700620017715,
      "grad_norm": 0.2619740664958954,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 84920
    },
    {
      "epoch": 75.22143489813995,
      "grad_norm": 0.2774341106414795,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 84925
    },
    {
      "epoch": 75.22586359610274,
      "grad_norm": 0.24442876875400543,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 84930
    },
    {
      "epoch": 75.23029229406555,
      "grad_norm": 0.25724250078201294,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 84935
    },
    {
      "epoch": 75.23472099202834,
      "grad_norm": 0.24393513798713684,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 84940
    },
    {
      "epoch": 75.23914968999114,
      "grad_norm": 0.23192188143730164,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 84945
    },
    {
      "epoch": 75.24357838795395,
      "grad_norm": 0.22205850481987,
      "learning_rate": 1e-05,
      "loss": 1.0053,
      "step": 84950
    },
    {
      "epoch": 75.24800708591674,
      "grad_norm": 0.2539752125740051,
      "learning_rate": 1e-05,
      "loss": 1.0369,
      "step": 84955
    },
    {
      "epoch": 75.25243578387953,
      "grad_norm": 0.2708742022514343,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 84960
    },
    {
      "epoch": 75.25686448184234,
      "grad_norm": 0.229655921459198,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 84965
    },
    {
      "epoch": 75.26129317980514,
      "grad_norm": 0.2129310667514801,
      "learning_rate": 1e-05,
      "loss": 1.036,
      "step": 84970
    },
    {
      "epoch": 75.26572187776793,
      "grad_norm": 0.24959397315979004,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 84975
    },
    {
      "epoch": 75.27015057573074,
      "grad_norm": 0.258658230304718,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 84980
    },
    {
      "epoch": 75.27457927369353,
      "grad_norm": 0.22723299264907837,
      "learning_rate": 1e-05,
      "loss": 0.933,
      "step": 84985
    },
    {
      "epoch": 75.27900797165633,
      "grad_norm": 0.23444893956184387,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 84990
    },
    {
      "epoch": 75.28343666961914,
      "grad_norm": 0.26359301805496216,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 84995
    },
    {
      "epoch": 75.28786536758193,
      "grad_norm": 0.21894216537475586,
      "learning_rate": 1e-05,
      "loss": 0.9108,
      "step": 85000
    },
    {
      "epoch": 75.29229406554472,
      "grad_norm": 0.2418021559715271,
      "learning_rate": 1e-05,
      "loss": 0.8826,
      "step": 85005
    },
    {
      "epoch": 75.29672276350753,
      "grad_norm": 0.2179296761751175,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 85010
    },
    {
      "epoch": 75.30115146147033,
      "grad_norm": 0.28593719005584717,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 85015
    },
    {
      "epoch": 75.30558015943312,
      "grad_norm": 0.2666097581386566,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 85020
    },
    {
      "epoch": 75.31000885739593,
      "grad_norm": 0.2627635598182678,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 85025
    },
    {
      "epoch": 75.31443755535872,
      "grad_norm": 0.2633863389492035,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 85030
    },
    {
      "epoch": 75.31886625332152,
      "grad_norm": 0.2474062740802765,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 85035
    },
    {
      "epoch": 75.32329495128432,
      "grad_norm": 0.24775506556034088,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 85040
    },
    {
      "epoch": 75.32772364924712,
      "grad_norm": 0.23711426556110382,
      "learning_rate": 1e-05,
      "loss": 1.029,
      "step": 85045
    },
    {
      "epoch": 75.33215234720993,
      "grad_norm": 0.24943478405475616,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 85050
    },
    {
      "epoch": 75.33658104517272,
      "grad_norm": 0.230828657746315,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 85055
    },
    {
      "epoch": 75.34100974313552,
      "grad_norm": 0.2685759365558624,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 85060
    },
    {
      "epoch": 75.34543844109832,
      "grad_norm": 0.23378632962703705,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 85065
    },
    {
      "epoch": 75.34986713906112,
      "grad_norm": 0.2377382218837738,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 85070
    },
    {
      "epoch": 75.35429583702391,
      "grad_norm": 0.2443789392709732,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 85075
    },
    {
      "epoch": 75.35872453498672,
      "grad_norm": 0.2492707520723343,
      "learning_rate": 1e-05,
      "loss": 0.9097,
      "step": 85080
    },
    {
      "epoch": 75.36315323294951,
      "grad_norm": 0.23865725100040436,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 85085
    },
    {
      "epoch": 75.36758193091231,
      "grad_norm": 0.21588142216205597,
      "learning_rate": 1e-05,
      "loss": 0.9123,
      "step": 85090
    },
    {
      "epoch": 75.37201062887512,
      "grad_norm": 0.24121707677841187,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 85095
    },
    {
      "epoch": 75.37643932683791,
      "grad_norm": 0.2339557260274887,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 85100
    },
    {
      "epoch": 75.3808680248007,
      "grad_norm": 0.22046083211898804,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 85105
    },
    {
      "epoch": 75.38529672276351,
      "grad_norm": 0.31256505846977234,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 85110
    },
    {
      "epoch": 75.38972542072631,
      "grad_norm": 0.2579330801963806,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 85115
    },
    {
      "epoch": 75.3941541186891,
      "grad_norm": 0.27714377641677856,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 85120
    },
    {
      "epoch": 75.39858281665191,
      "grad_norm": 0.29912069439888,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 85125
    },
    {
      "epoch": 75.4030115146147,
      "grad_norm": 0.23450332880020142,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 85130
    },
    {
      "epoch": 75.4074402125775,
      "grad_norm": 0.27591490745544434,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 85135
    },
    {
      "epoch": 75.4118689105403,
      "grad_norm": 0.23395629227161407,
      "learning_rate": 1e-05,
      "loss": 0.9227,
      "step": 85140
    },
    {
      "epoch": 75.4162976085031,
      "grad_norm": 0.29644298553466797,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 85145
    },
    {
      "epoch": 75.4207263064659,
      "grad_norm": 0.23182235658168793,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 85150
    },
    {
      "epoch": 75.4251550044287,
      "grad_norm": 0.21023684740066528,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 85155
    },
    {
      "epoch": 75.4295837023915,
      "grad_norm": 0.22046741843223572,
      "learning_rate": 1e-05,
      "loss": 1.0186,
      "step": 85160
    },
    {
      "epoch": 75.43401240035429,
      "grad_norm": 0.2222185879945755,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 85165
    },
    {
      "epoch": 75.4384410983171,
      "grad_norm": 0.2118159830570221,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 85170
    },
    {
      "epoch": 75.4428697962799,
      "grad_norm": 0.22569133341312408,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 85175
    },
    {
      "epoch": 75.44729849424269,
      "grad_norm": 0.22361229360103607,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 85180
    },
    {
      "epoch": 75.4517271922055,
      "grad_norm": 0.2534103989601135,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 85185
    },
    {
      "epoch": 75.45615589016829,
      "grad_norm": 0.2374987155199051,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 85190
    },
    {
      "epoch": 75.46058458813108,
      "grad_norm": 0.27647802233695984,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 85195
    },
    {
      "epoch": 75.46501328609389,
      "grad_norm": 0.23736269772052765,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 85200
    },
    {
      "epoch": 75.46944198405669,
      "grad_norm": 0.27589935064315796,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 85205
    },
    {
      "epoch": 75.47387068201948,
      "grad_norm": 0.24836444854736328,
      "learning_rate": 1e-05,
      "loss": 1.025,
      "step": 85210
    },
    {
      "epoch": 75.47829937998229,
      "grad_norm": 0.270071417093277,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 85215
    },
    {
      "epoch": 75.48272807794508,
      "grad_norm": 0.23580753803253174,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 85220
    },
    {
      "epoch": 75.48715677590788,
      "grad_norm": 0.23322385549545288,
      "learning_rate": 1e-05,
      "loss": 0.9293,
      "step": 85225
    },
    {
      "epoch": 75.49158547387069,
      "grad_norm": 0.2386915385723114,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 85230
    },
    {
      "epoch": 75.49601417183348,
      "grad_norm": 0.2402641326189041,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 85235
    },
    {
      "epoch": 75.50044286979627,
      "grad_norm": 0.25126728415489197,
      "learning_rate": 1e-05,
      "loss": 1.0234,
      "step": 85240
    },
    {
      "epoch": 75.50487156775908,
      "grad_norm": 0.3170195519924164,
      "learning_rate": 1e-05,
      "loss": 0.9272,
      "step": 85245
    },
    {
      "epoch": 75.50930026572188,
      "grad_norm": 0.22746683657169342,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 85250
    },
    {
      "epoch": 75.51372896368467,
      "grad_norm": 0.24262942373752594,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 85255
    },
    {
      "epoch": 75.51815766164748,
      "grad_norm": 0.2949182093143463,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 85260
    },
    {
      "epoch": 75.52258635961027,
      "grad_norm": 0.2304728776216507,
      "learning_rate": 1e-05,
      "loss": 1.0231,
      "step": 85265
    },
    {
      "epoch": 75.52701505757307,
      "grad_norm": 0.2478756606578827,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 85270
    },
    {
      "epoch": 75.53144375553588,
      "grad_norm": 0.2361639440059662,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 85275
    },
    {
      "epoch": 75.53587245349867,
      "grad_norm": 0.2413511872291565,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 85280
    },
    {
      "epoch": 75.54030115146146,
      "grad_norm": 0.2701733708381653,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 85285
    },
    {
      "epoch": 75.54472984942427,
      "grad_norm": 0.21128009259700775,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 85290
    },
    {
      "epoch": 75.54915854738707,
      "grad_norm": 0.22484073042869568,
      "learning_rate": 1e-05,
      "loss": 0.8959,
      "step": 85295
    },
    {
      "epoch": 75.55358724534987,
      "grad_norm": 0.2168583869934082,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 85300
    },
    {
      "epoch": 75.55801594331267,
      "grad_norm": 0.22390127182006836,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 85305
    },
    {
      "epoch": 75.56244464127546,
      "grad_norm": 0.258327454328537,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 85310
    },
    {
      "epoch": 75.56687333923827,
      "grad_norm": 0.23413020372390747,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 85315
    },
    {
      "epoch": 75.57130203720106,
      "grad_norm": 0.24476855993270874,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 85320
    },
    {
      "epoch": 75.57573073516386,
      "grad_norm": 0.23538783192634583,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 85325
    },
    {
      "epoch": 75.58015943312667,
      "grad_norm": 0.26011019945144653,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 85330
    },
    {
      "epoch": 75.58458813108946,
      "grad_norm": 0.2031121551990509,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 85335
    },
    {
      "epoch": 75.58901682905226,
      "grad_norm": 0.23786626756191254,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 85340
    },
    {
      "epoch": 75.59344552701506,
      "grad_norm": 0.24605628848075867,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 85345
    },
    {
      "epoch": 75.59787422497786,
      "grad_norm": 0.2458154261112213,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 85350
    },
    {
      "epoch": 75.60230292294065,
      "grad_norm": 0.22241619229316711,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 85355
    },
    {
      "epoch": 75.60673162090346,
      "grad_norm": 0.22901132702827454,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 85360
    },
    {
      "epoch": 75.61116031886625,
      "grad_norm": 0.23304331302642822,
      "learning_rate": 1e-05,
      "loss": 0.9136,
      "step": 85365
    },
    {
      "epoch": 75.61558901682905,
      "grad_norm": 0.22018621861934662,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 85370
    },
    {
      "epoch": 75.62001771479186,
      "grad_norm": 0.25559377670288086,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 85375
    },
    {
      "epoch": 75.62444641275465,
      "grad_norm": 0.25102365016937256,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 85380
    },
    {
      "epoch": 75.62887511071744,
      "grad_norm": 0.208938866853714,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 85385
    },
    {
      "epoch": 75.63330380868025,
      "grad_norm": 0.21545442938804626,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 85390
    },
    {
      "epoch": 75.63773250664305,
      "grad_norm": 0.1989172399044037,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 85395
    },
    {
      "epoch": 75.64216120460584,
      "grad_norm": 0.23315006494522095,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 85400
    },
    {
      "epoch": 75.64658990256865,
      "grad_norm": 0.21885494887828827,
      "learning_rate": 1e-05,
      "loss": 0.925,
      "step": 85405
    },
    {
      "epoch": 75.65101860053144,
      "grad_norm": 0.23954246938228607,
      "learning_rate": 1e-05,
      "loss": 0.9221,
      "step": 85410
    },
    {
      "epoch": 75.65544729849424,
      "grad_norm": 0.2486918568611145,
      "learning_rate": 1e-05,
      "loss": 0.9319,
      "step": 85415
    },
    {
      "epoch": 75.65987599645705,
      "grad_norm": 0.2685340940952301,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 85420
    },
    {
      "epoch": 75.66430469441984,
      "grad_norm": 0.259987473487854,
      "learning_rate": 1e-05,
      "loss": 0.9195,
      "step": 85425
    },
    {
      "epoch": 75.66873339238263,
      "grad_norm": 0.22894006967544556,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 85430
    },
    {
      "epoch": 75.67316209034544,
      "grad_norm": 0.2500392198562622,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 85435
    },
    {
      "epoch": 75.67759078830824,
      "grad_norm": 0.24679076671600342,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 85440
    },
    {
      "epoch": 75.68201948627103,
      "grad_norm": 0.23897510766983032,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 85445
    },
    {
      "epoch": 75.68644818423384,
      "grad_norm": 0.24865782260894775,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 85450
    },
    {
      "epoch": 75.69087688219663,
      "grad_norm": 0.21969492733478546,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 85455
    },
    {
      "epoch": 75.69530558015943,
      "grad_norm": 0.22037707269191742,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 85460
    },
    {
      "epoch": 75.69973427812224,
      "grad_norm": 0.2533146142959595,
      "learning_rate": 1e-05,
      "loss": 0.8999,
      "step": 85465
    },
    {
      "epoch": 75.70416297608503,
      "grad_norm": 0.2391863316297531,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 85470
    },
    {
      "epoch": 75.70859167404782,
      "grad_norm": 0.23583179712295532,
      "learning_rate": 1e-05,
      "loss": 1.0309,
      "step": 85475
    },
    {
      "epoch": 75.71302037201063,
      "grad_norm": 0.3049645721912384,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 85480
    },
    {
      "epoch": 75.71744906997343,
      "grad_norm": 0.2273280769586563,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 85485
    },
    {
      "epoch": 75.72187776793622,
      "grad_norm": 0.2354247123003006,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 85490
    },
    {
      "epoch": 75.72630646589903,
      "grad_norm": 0.21188464760780334,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 85495
    },
    {
      "epoch": 75.73073516386182,
      "grad_norm": 0.307801216840744,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 85500
    },
    {
      "epoch": 75.73516386182462,
      "grad_norm": 0.27640068531036377,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 85505
    },
    {
      "epoch": 75.73959255978743,
      "grad_norm": 0.24601681530475616,
      "learning_rate": 1e-05,
      "loss": 1.0698,
      "step": 85510
    },
    {
      "epoch": 75.74402125775022,
      "grad_norm": 0.20356449484825134,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 85515
    },
    {
      "epoch": 75.74844995571301,
      "grad_norm": 0.23101134598255157,
      "learning_rate": 1e-05,
      "loss": 0.9026,
      "step": 85520
    },
    {
      "epoch": 75.75287865367582,
      "grad_norm": 0.2281644493341446,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 85525
    },
    {
      "epoch": 75.75730735163862,
      "grad_norm": 0.24147437512874603,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 85530
    },
    {
      "epoch": 75.76173604960141,
      "grad_norm": 0.24739526212215424,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 85535
    },
    {
      "epoch": 75.76616474756422,
      "grad_norm": 0.22871942818164825,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 85540
    },
    {
      "epoch": 75.77059344552701,
      "grad_norm": 0.2706684172153473,
      "learning_rate": 1e-05,
      "loss": 0.9191,
      "step": 85545
    },
    {
      "epoch": 75.77502214348982,
      "grad_norm": 0.24718569219112396,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 85550
    },
    {
      "epoch": 75.77945084145261,
      "grad_norm": 0.2716258466243744,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 85555
    },
    {
      "epoch": 75.78387953941541,
      "grad_norm": 0.25657913088798523,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 85560
    },
    {
      "epoch": 75.78830823737822,
      "grad_norm": 0.24244873225688934,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 85565
    },
    {
      "epoch": 75.79273693534101,
      "grad_norm": 0.24041496217250824,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 85570
    },
    {
      "epoch": 75.7971656333038,
      "grad_norm": 0.2565837502479553,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 85575
    },
    {
      "epoch": 75.80159433126661,
      "grad_norm": 0.2787538468837738,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 85580
    },
    {
      "epoch": 75.80602302922941,
      "grad_norm": 0.2501988112926483,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 85585
    },
    {
      "epoch": 75.8104517271922,
      "grad_norm": 0.2141208052635193,
      "learning_rate": 1e-05,
      "loss": 0.93,
      "step": 85590
    },
    {
      "epoch": 75.81488042515501,
      "grad_norm": 0.25114116072654724,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 85595
    },
    {
      "epoch": 75.8193091231178,
      "grad_norm": 0.24835629761219025,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 85600
    },
    {
      "epoch": 75.8237378210806,
      "grad_norm": 0.21733909845352173,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 85605
    },
    {
      "epoch": 75.8281665190434,
      "grad_norm": 0.20196415483951569,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 85610
    },
    {
      "epoch": 75.8325952170062,
      "grad_norm": 0.30430835485458374,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 85615
    },
    {
      "epoch": 75.837023914969,
      "grad_norm": 0.24052922427654266,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 85620
    },
    {
      "epoch": 75.8414526129318,
      "grad_norm": 0.24825434386730194,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 85625
    },
    {
      "epoch": 75.8458813108946,
      "grad_norm": 0.2518892288208008,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 85630
    },
    {
      "epoch": 75.85031000885739,
      "grad_norm": 0.20691558718681335,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 85635
    },
    {
      "epoch": 75.8547387068202,
      "grad_norm": 0.2416483461856842,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 85640
    },
    {
      "epoch": 75.859167404783,
      "grad_norm": 0.24811483919620514,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 85645
    },
    {
      "epoch": 75.86359610274579,
      "grad_norm": 0.23103493452072144,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 85650
    },
    {
      "epoch": 75.8680248007086,
      "grad_norm": 0.2679225206375122,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 85655
    },
    {
      "epoch": 75.87245349867139,
      "grad_norm": 0.2682875692844391,
      "learning_rate": 1e-05,
      "loss": 0.9165,
      "step": 85660
    },
    {
      "epoch": 75.87688219663418,
      "grad_norm": 0.20635049045085907,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 85665
    },
    {
      "epoch": 75.881310894597,
      "grad_norm": 0.19470936059951782,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 85670
    },
    {
      "epoch": 75.88573959255979,
      "grad_norm": 0.24073436856269836,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 85675
    },
    {
      "epoch": 75.89016829052258,
      "grad_norm": 0.2124311476945877,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 85680
    },
    {
      "epoch": 75.89459698848539,
      "grad_norm": 0.2729822099208832,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 85685
    },
    {
      "epoch": 75.89902568644818,
      "grad_norm": 0.22725269198417664,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 85690
    },
    {
      "epoch": 75.90345438441098,
      "grad_norm": 0.27036288380622864,
      "learning_rate": 1e-05,
      "loss": 0.9211,
      "step": 85695
    },
    {
      "epoch": 75.90788308237379,
      "grad_norm": 0.22100743651390076,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 85700
    },
    {
      "epoch": 75.91231178033658,
      "grad_norm": 0.24741415679454803,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 85705
    },
    {
      "epoch": 75.91674047829937,
      "grad_norm": 0.20851606130599976,
      "learning_rate": 1e-05,
      "loss": 0.9183,
      "step": 85710
    },
    {
      "epoch": 75.92116917626218,
      "grad_norm": 0.24909481406211853,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 85715
    },
    {
      "epoch": 75.92559787422498,
      "grad_norm": 0.2866688668727875,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 85720
    },
    {
      "epoch": 75.93002657218777,
      "grad_norm": 0.22481217980384827,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 85725
    },
    {
      "epoch": 75.93445527015058,
      "grad_norm": 0.24417854845523834,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 85730
    },
    {
      "epoch": 75.93888396811337,
      "grad_norm": 0.203885018825531,
      "learning_rate": 1e-05,
      "loss": 0.8979,
      "step": 85735
    },
    {
      "epoch": 75.94331266607617,
      "grad_norm": 0.22813384234905243,
      "learning_rate": 1e-05,
      "loss": 0.9185,
      "step": 85740
    },
    {
      "epoch": 75.94774136403898,
      "grad_norm": 0.26505497097969055,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 85745
    },
    {
      "epoch": 75.95217006200177,
      "grad_norm": 0.24845972657203674,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 85750
    },
    {
      "epoch": 75.95659875996456,
      "grad_norm": 0.23623836040496826,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 85755
    },
    {
      "epoch": 75.96102745792737,
      "grad_norm": 0.24243469536304474,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 85760
    },
    {
      "epoch": 75.96545615589017,
      "grad_norm": 0.33199119567871094,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 85765
    },
    {
      "epoch": 75.96988485385296,
      "grad_norm": 0.21568551659584045,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 85770
    },
    {
      "epoch": 75.97431355181577,
      "grad_norm": 0.23183706402778625,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 85775
    },
    {
      "epoch": 75.97874224977856,
      "grad_norm": 0.23481948673725128,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 85780
    },
    {
      "epoch": 75.98317094774137,
      "grad_norm": 0.26898589730262756,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 85785
    },
    {
      "epoch": 75.98759964570417,
      "grad_norm": 0.2503397762775421,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 85790
    },
    {
      "epoch": 75.99202834366696,
      "grad_norm": 0.24289657175540924,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 85795
    },
    {
      "epoch": 75.99645704162977,
      "grad_norm": 0.2242513746023178,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 85800
    },
    {
      "epoch": 76.00088573959256,
      "grad_norm": 0.2130116969347,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 85805
    },
    {
      "epoch": 76.00531443755536,
      "grad_norm": 0.22323259711265564,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 85810
    },
    {
      "epoch": 76.00974313551816,
      "grad_norm": 0.23010407388210297,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 85815
    },
    {
      "epoch": 76.01417183348096,
      "grad_norm": 0.222050741314888,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 85820
    },
    {
      "epoch": 76.01860053144375,
      "grad_norm": 0.26141923666000366,
      "learning_rate": 1e-05,
      "loss": 0.9193,
      "step": 85825
    },
    {
      "epoch": 76.02302922940656,
      "grad_norm": 0.29551437497138977,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 85830
    },
    {
      "epoch": 76.02745792736935,
      "grad_norm": 0.2618325650691986,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 85835
    },
    {
      "epoch": 76.03188662533215,
      "grad_norm": 0.24414639174938202,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 85840
    },
    {
      "epoch": 76.03631532329496,
      "grad_norm": 0.2505650520324707,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 85845
    },
    {
      "epoch": 76.04074402125775,
      "grad_norm": 0.23258769512176514,
      "learning_rate": 1e-05,
      "loss": 1.0183,
      "step": 85850
    },
    {
      "epoch": 76.04517271922055,
      "grad_norm": 0.24757912755012512,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 85855
    },
    {
      "epoch": 76.04960141718335,
      "grad_norm": 0.25555816292762756,
      "learning_rate": 1e-05,
      "loss": 1.0208,
      "step": 85860
    },
    {
      "epoch": 76.05403011514615,
      "grad_norm": 0.2425612360239029,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 85865
    },
    {
      "epoch": 76.05845881310894,
      "grad_norm": 0.20050077140331268,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 85870
    },
    {
      "epoch": 76.06288751107175,
      "grad_norm": 0.23308445513248444,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 85875
    },
    {
      "epoch": 76.06731620903454,
      "grad_norm": 0.22100616991519928,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 85880
    },
    {
      "epoch": 76.07174490699734,
      "grad_norm": 0.2164248824119568,
      "learning_rate": 1e-05,
      "loss": 0.914,
      "step": 85885
    },
    {
      "epoch": 76.07617360496015,
      "grad_norm": 0.24151375889778137,
      "learning_rate": 1e-05,
      "loss": 0.9019,
      "step": 85890
    },
    {
      "epoch": 76.08060230292294,
      "grad_norm": 0.2733105421066284,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 85895
    },
    {
      "epoch": 76.08503100088573,
      "grad_norm": 0.25998198986053467,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 85900
    },
    {
      "epoch": 76.08945969884854,
      "grad_norm": 0.28050321340560913,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 85905
    },
    {
      "epoch": 76.09388839681134,
      "grad_norm": 0.22001048922538757,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 85910
    },
    {
      "epoch": 76.09831709477413,
      "grad_norm": 0.2429816722869873,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 85915
    },
    {
      "epoch": 76.10274579273694,
      "grad_norm": 0.2649989426136017,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 85920
    },
    {
      "epoch": 76.10717449069973,
      "grad_norm": 0.27369949221611023,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 85925
    },
    {
      "epoch": 76.11160318866253,
      "grad_norm": 0.2501014471054077,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 85930
    },
    {
      "epoch": 76.11603188662534,
      "grad_norm": 0.25790613889694214,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 85935
    },
    {
      "epoch": 76.12046058458813,
      "grad_norm": 0.2612212002277374,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 85940
    },
    {
      "epoch": 76.12488928255092,
      "grad_norm": 0.2193792462348938,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 85945
    },
    {
      "epoch": 76.12931798051373,
      "grad_norm": 0.24109342694282532,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 85950
    },
    {
      "epoch": 76.13374667847653,
      "grad_norm": 0.2330726534128189,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 85955
    },
    {
      "epoch": 76.13817537643932,
      "grad_norm": 0.2948412597179413,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 85960
    },
    {
      "epoch": 76.14260407440213,
      "grad_norm": 0.24219980835914612,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 85965
    },
    {
      "epoch": 76.14703277236492,
      "grad_norm": 0.26734235882759094,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 85970
    },
    {
      "epoch": 76.15146147032772,
      "grad_norm": 0.2455536276102066,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 85975
    },
    {
      "epoch": 76.15589016829053,
      "grad_norm": 0.30622678995132446,
      "learning_rate": 1e-05,
      "loss": 1.0361,
      "step": 85980
    },
    {
      "epoch": 76.16031886625332,
      "grad_norm": 0.2327612340450287,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 85985
    },
    {
      "epoch": 76.16474756421611,
      "grad_norm": 0.23492680490016937,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 85990
    },
    {
      "epoch": 76.16917626217892,
      "grad_norm": 0.220939502120018,
      "learning_rate": 1e-05,
      "loss": 0.936,
      "step": 85995
    },
    {
      "epoch": 76.17360496014172,
      "grad_norm": 0.20230375230312347,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 86000
    },
    {
      "epoch": 76.17803365810451,
      "grad_norm": 0.25440168380737305,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 86005
    },
    {
      "epoch": 76.18246235606732,
      "grad_norm": 0.27093505859375,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 86010
    },
    {
      "epoch": 76.18689105403011,
      "grad_norm": 0.23706993460655212,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 86015
    },
    {
      "epoch": 76.1913197519929,
      "grad_norm": 0.23958848416805267,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 86020
    },
    {
      "epoch": 76.19574844995572,
      "grad_norm": 0.2547640800476074,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 86025
    },
    {
      "epoch": 76.20017714791851,
      "grad_norm": 0.2764376699924469,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 86030
    },
    {
      "epoch": 76.20460584588132,
      "grad_norm": 0.22230862081050873,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 86035
    },
    {
      "epoch": 76.20903454384411,
      "grad_norm": 0.24868865311145782,
      "learning_rate": 1e-05,
      "loss": 1.0189,
      "step": 86040
    },
    {
      "epoch": 76.2134632418069,
      "grad_norm": 0.2329602837562561,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 86045
    },
    {
      "epoch": 76.21789193976971,
      "grad_norm": 0.22833050787448883,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 86050
    },
    {
      "epoch": 76.22232063773251,
      "grad_norm": 0.24775633215904236,
      "learning_rate": 1e-05,
      "loss": 1.031,
      "step": 86055
    },
    {
      "epoch": 76.2267493356953,
      "grad_norm": 0.2267179787158966,
      "learning_rate": 1e-05,
      "loss": 0.9118,
      "step": 86060
    },
    {
      "epoch": 76.23117803365811,
      "grad_norm": 0.22376282513141632,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 86065
    },
    {
      "epoch": 76.2356067316209,
      "grad_norm": 0.24551080167293549,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 86070
    },
    {
      "epoch": 76.2400354295837,
      "grad_norm": 0.2261231392621994,
      "learning_rate": 1e-05,
      "loss": 1.0263,
      "step": 86075
    },
    {
      "epoch": 76.24446412754651,
      "grad_norm": 0.23337920010089874,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 86080
    },
    {
      "epoch": 76.2488928255093,
      "grad_norm": 0.3091772198677063,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 86085
    },
    {
      "epoch": 76.2533215234721,
      "grad_norm": 0.2600855231285095,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 86090
    },
    {
      "epoch": 76.2577502214349,
      "grad_norm": 0.28874391317367554,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 86095
    },
    {
      "epoch": 76.2621789193977,
      "grad_norm": 0.26457393169403076,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 86100
    },
    {
      "epoch": 76.26660761736049,
      "grad_norm": 0.22813531756401062,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 86105
    },
    {
      "epoch": 76.2710363153233,
      "grad_norm": 0.22116704285144806,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 86110
    },
    {
      "epoch": 76.2754650132861,
      "grad_norm": 0.2731364369392395,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 86115
    },
    {
      "epoch": 76.27989371124889,
      "grad_norm": 0.21815571188926697,
      "learning_rate": 1e-05,
      "loss": 0.9038,
      "step": 86120
    },
    {
      "epoch": 76.2843224092117,
      "grad_norm": 0.23345430195331573,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 86125
    },
    {
      "epoch": 76.28875110717449,
      "grad_norm": 0.29991385340690613,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 86130
    },
    {
      "epoch": 76.29317980513729,
      "grad_norm": 0.24409854412078857,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 86135
    },
    {
      "epoch": 76.2976085031001,
      "grad_norm": 0.25381743907928467,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 86140
    },
    {
      "epoch": 76.30203720106289,
      "grad_norm": 0.23244820535182953,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 86145
    },
    {
      "epoch": 76.30646589902568,
      "grad_norm": 0.29370757937431335,
      "learning_rate": 1e-05,
      "loss": 0.9335,
      "step": 86150
    },
    {
      "epoch": 76.31089459698849,
      "grad_norm": 0.22407659888267517,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 86155
    },
    {
      "epoch": 76.31532329495128,
      "grad_norm": 0.23379559814929962,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 86160
    },
    {
      "epoch": 76.31975199291408,
      "grad_norm": 0.21955782175064087,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 86165
    },
    {
      "epoch": 76.32418069087689,
      "grad_norm": 0.25068891048431396,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 86170
    },
    {
      "epoch": 76.32860938883968,
      "grad_norm": 0.23807179927825928,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 86175
    },
    {
      "epoch": 76.33303808680247,
      "grad_norm": 0.23320156335830688,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 86180
    },
    {
      "epoch": 76.33746678476528,
      "grad_norm": 0.21294766664505005,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 86185
    },
    {
      "epoch": 76.34189548272808,
      "grad_norm": 0.2287573665380478,
      "learning_rate": 1e-05,
      "loss": 0.9099,
      "step": 86190
    },
    {
      "epoch": 76.34632418069087,
      "grad_norm": 0.21637177467346191,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 86195
    },
    {
      "epoch": 76.35075287865368,
      "grad_norm": 0.25280267000198364,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 86200
    },
    {
      "epoch": 76.35518157661647,
      "grad_norm": 0.21149703860282898,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 86205
    },
    {
      "epoch": 76.35961027457927,
      "grad_norm": 0.22107967734336853,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 86210
    },
    {
      "epoch": 76.36403897254208,
      "grad_norm": 0.23441214859485626,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 86215
    },
    {
      "epoch": 76.36846767050487,
      "grad_norm": 0.24445748329162598,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 86220
    },
    {
      "epoch": 76.37289636846766,
      "grad_norm": 0.22310857474803925,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 86225
    },
    {
      "epoch": 76.37732506643047,
      "grad_norm": 0.24026520550251007,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 86230
    },
    {
      "epoch": 76.38175376439327,
      "grad_norm": 0.2497514933347702,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 86235
    },
    {
      "epoch": 76.38618246235606,
      "grad_norm": 0.3094234764575958,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 86240
    },
    {
      "epoch": 76.39061116031887,
      "grad_norm": 0.28352686762809753,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 86245
    },
    {
      "epoch": 76.39503985828166,
      "grad_norm": 0.22026361525058746,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 86250
    },
    {
      "epoch": 76.39946855624446,
      "grad_norm": 0.22804172337055206,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 86255
    },
    {
      "epoch": 76.40389725420727,
      "grad_norm": 0.21392962336540222,
      "learning_rate": 1e-05,
      "loss": 1.0445,
      "step": 86260
    },
    {
      "epoch": 76.40832595217006,
      "grad_norm": 0.24531374871730804,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 86265
    },
    {
      "epoch": 76.41275465013285,
      "grad_norm": 0.23857592046260834,
      "learning_rate": 1e-05,
      "loss": 0.9241,
      "step": 86270
    },
    {
      "epoch": 76.41718334809566,
      "grad_norm": 0.20373912155628204,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 86275
    },
    {
      "epoch": 76.42161204605846,
      "grad_norm": 0.24145077168941498,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 86280
    },
    {
      "epoch": 76.42604074402126,
      "grad_norm": 0.23858612775802612,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 86285
    },
    {
      "epoch": 76.43046944198406,
      "grad_norm": 0.247470885515213,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 86290
    },
    {
      "epoch": 76.43489813994685,
      "grad_norm": 0.23213699460029602,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 86295
    },
    {
      "epoch": 76.43932683790966,
      "grad_norm": 0.2299901843070984,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 86300
    },
    {
      "epoch": 76.44375553587246,
      "grad_norm": 0.2666286826133728,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 86305
    },
    {
      "epoch": 76.44818423383525,
      "grad_norm": 0.22047480940818787,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 86310
    },
    {
      "epoch": 76.45261293179806,
      "grad_norm": 0.24304699897766113,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 86315
    },
    {
      "epoch": 76.45704162976085,
      "grad_norm": 0.2529726028442383,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 86320
    },
    {
      "epoch": 76.46147032772365,
      "grad_norm": 0.22365185618400574,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 86325
    },
    {
      "epoch": 76.46589902568645,
      "grad_norm": 0.23253794014453888,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 86330
    },
    {
      "epoch": 76.47032772364925,
      "grad_norm": 0.2619085907936096,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 86335
    },
    {
      "epoch": 76.47475642161204,
      "grad_norm": 0.22694091498851776,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 86340
    },
    {
      "epoch": 76.47918511957485,
      "grad_norm": 0.2369011640548706,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 86345
    },
    {
      "epoch": 76.48361381753764,
      "grad_norm": 0.22451049089431763,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 86350
    },
    {
      "epoch": 76.48804251550044,
      "grad_norm": 0.2325989156961441,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 86355
    },
    {
      "epoch": 76.49247121346325,
      "grad_norm": 0.227561354637146,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 86360
    },
    {
      "epoch": 76.49689991142604,
      "grad_norm": 0.22985471785068512,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 86365
    },
    {
      "epoch": 76.50132860938884,
      "grad_norm": 0.19532828032970428,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 86370
    },
    {
      "epoch": 76.50575730735164,
      "grad_norm": 0.2506723999977112,
      "learning_rate": 1e-05,
      "loss": 0.9207,
      "step": 86375
    },
    {
      "epoch": 76.51018600531444,
      "grad_norm": 0.2511369287967682,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 86380
    },
    {
      "epoch": 76.51461470327723,
      "grad_norm": 0.3116391897201538,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 86385
    },
    {
      "epoch": 76.51904340124004,
      "grad_norm": 0.2753237187862396,
      "learning_rate": 1e-05,
      "loss": 0.904,
      "step": 86390
    },
    {
      "epoch": 76.52347209920283,
      "grad_norm": 0.27732256054878235,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 86395
    },
    {
      "epoch": 76.52790079716563,
      "grad_norm": 0.25070759654045105,
      "learning_rate": 1e-05,
      "loss": 0.9279,
      "step": 86400
    },
    {
      "epoch": 76.53232949512844,
      "grad_norm": 0.2630537748336792,
      "learning_rate": 1e-05,
      "loss": 0.9046,
      "step": 86405
    },
    {
      "epoch": 76.53675819309123,
      "grad_norm": 0.24246330559253693,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 86410
    },
    {
      "epoch": 76.54118689105402,
      "grad_norm": 0.20930306613445282,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 86415
    },
    {
      "epoch": 76.54561558901683,
      "grad_norm": 0.2310047447681427,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 86420
    },
    {
      "epoch": 76.55004428697963,
      "grad_norm": 0.26790744066238403,
      "learning_rate": 1e-05,
      "loss": 1.024,
      "step": 86425
    },
    {
      "epoch": 76.55447298494242,
      "grad_norm": 0.2147175818681717,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 86430
    },
    {
      "epoch": 76.55890168290523,
      "grad_norm": 0.298657089471817,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 86435
    },
    {
      "epoch": 76.56333038086802,
      "grad_norm": 0.2431868463754654,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 86440
    },
    {
      "epoch": 76.56775907883082,
      "grad_norm": 0.33144888281822205,
      "learning_rate": 1e-05,
      "loss": 0.917,
      "step": 86445
    },
    {
      "epoch": 76.57218777679363,
      "grad_norm": 0.24151641130447388,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 86450
    },
    {
      "epoch": 76.57661647475642,
      "grad_norm": 0.23077145218849182,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 86455
    },
    {
      "epoch": 76.58104517271921,
      "grad_norm": 0.3118334710597992,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 86460
    },
    {
      "epoch": 76.58547387068202,
      "grad_norm": 0.24082234501838684,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 86465
    },
    {
      "epoch": 76.58990256864482,
      "grad_norm": 0.25131234526634216,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 86470
    },
    {
      "epoch": 76.59433126660761,
      "grad_norm": 0.25931814312934875,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 86475
    },
    {
      "epoch": 76.59875996457042,
      "grad_norm": 0.2307201772928238,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 86480
    },
    {
      "epoch": 76.60318866253321,
      "grad_norm": 0.28370875120162964,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 86485
    },
    {
      "epoch": 76.60761736049601,
      "grad_norm": 0.25544503331184387,
      "learning_rate": 1e-05,
      "loss": 0.9266,
      "step": 86490
    },
    {
      "epoch": 76.61204605845882,
      "grad_norm": 0.1927243173122406,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 86495
    },
    {
      "epoch": 76.61647475642161,
      "grad_norm": 0.2231311947107315,
      "learning_rate": 1e-05,
      "loss": 0.9214,
      "step": 86500
    },
    {
      "epoch": 76.6209034543844,
      "grad_norm": 0.21392396092414856,
      "learning_rate": 1e-05,
      "loss": 0.9227,
      "step": 86505
    },
    {
      "epoch": 76.62533215234721,
      "grad_norm": 0.20430269837379456,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 86510
    },
    {
      "epoch": 76.62976085031,
      "grad_norm": 0.25092628598213196,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 86515
    },
    {
      "epoch": 76.63418954827281,
      "grad_norm": 0.23163533210754395,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 86520
    },
    {
      "epoch": 76.63861824623561,
      "grad_norm": 0.23117592930793762,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 86525
    },
    {
      "epoch": 76.6430469441984,
      "grad_norm": 0.20848457515239716,
      "learning_rate": 1e-05,
      "loss": 0.9187,
      "step": 86530
    },
    {
      "epoch": 76.64747564216121,
      "grad_norm": 0.24528324604034424,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 86535
    },
    {
      "epoch": 76.651904340124,
      "grad_norm": 0.23941963911056519,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 86540
    },
    {
      "epoch": 76.6563330380868,
      "grad_norm": 0.22736983001232147,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 86545
    },
    {
      "epoch": 76.66076173604961,
      "grad_norm": 0.25687968730926514,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 86550
    },
    {
      "epoch": 76.6651904340124,
      "grad_norm": 0.21230284869670868,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 86555
    },
    {
      "epoch": 76.6696191319752,
      "grad_norm": 0.24746738374233246,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 86560
    },
    {
      "epoch": 76.674047829938,
      "grad_norm": 0.2330058366060257,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 86565
    },
    {
      "epoch": 76.6784765279008,
      "grad_norm": 0.2323242723941803,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 86570
    },
    {
      "epoch": 76.68290522586359,
      "grad_norm": 0.23065844178199768,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 86575
    },
    {
      "epoch": 76.6873339238264,
      "grad_norm": 0.2441716492176056,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 86580
    },
    {
      "epoch": 76.6917626217892,
      "grad_norm": 0.25631096959114075,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 86585
    },
    {
      "epoch": 76.69619131975199,
      "grad_norm": 0.2615143954753876,
      "learning_rate": 1e-05,
      "loss": 0.9235,
      "step": 86590
    },
    {
      "epoch": 76.7006200177148,
      "grad_norm": 0.2249242663383484,
      "learning_rate": 1e-05,
      "loss": 0.9134,
      "step": 86595
    },
    {
      "epoch": 76.70504871567759,
      "grad_norm": 0.24702699482440948,
      "learning_rate": 1e-05,
      "loss": 0.9318,
      "step": 86600
    },
    {
      "epoch": 76.70947741364039,
      "grad_norm": 0.23183965682983398,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 86605
    },
    {
      "epoch": 76.7139061116032,
      "grad_norm": 0.2163151055574417,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 86610
    },
    {
      "epoch": 76.71833480956599,
      "grad_norm": 0.20894266664981842,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 86615
    },
    {
      "epoch": 76.72276350752878,
      "grad_norm": 0.2231171727180481,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 86620
    },
    {
      "epoch": 76.72719220549159,
      "grad_norm": 0.24931246042251587,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 86625
    },
    {
      "epoch": 76.73162090345438,
      "grad_norm": 0.22917351126670837,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 86630
    },
    {
      "epoch": 76.73604960141718,
      "grad_norm": 0.2686276137828827,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 86635
    },
    {
      "epoch": 76.74047829937999,
      "grad_norm": 0.225505530834198,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 86640
    },
    {
      "epoch": 76.74490699734278,
      "grad_norm": 0.2655288577079773,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 86645
    },
    {
      "epoch": 76.74933569530558,
      "grad_norm": 0.2716740071773529,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 86650
    },
    {
      "epoch": 76.75376439326838,
      "grad_norm": 0.23587769269943237,
      "learning_rate": 1e-05,
      "loss": 0.9234,
      "step": 86655
    },
    {
      "epoch": 76.75819309123118,
      "grad_norm": 0.2702116072177887,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 86660
    },
    {
      "epoch": 76.76262178919397,
      "grad_norm": 0.26375889778137207,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 86665
    },
    {
      "epoch": 76.76705048715678,
      "grad_norm": 0.2620989680290222,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 86670
    },
    {
      "epoch": 76.77147918511957,
      "grad_norm": 0.2296069711446762,
      "learning_rate": 1e-05,
      "loss": 0.9324,
      "step": 86675
    },
    {
      "epoch": 76.77590788308237,
      "grad_norm": 0.2209530770778656,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 86680
    },
    {
      "epoch": 76.78033658104518,
      "grad_norm": 0.236846461892128,
      "learning_rate": 1e-05,
      "loss": 0.9243,
      "step": 86685
    },
    {
      "epoch": 76.78476527900797,
      "grad_norm": 0.23009677231311798,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 86690
    },
    {
      "epoch": 76.78919397697076,
      "grad_norm": 0.2381734699010849,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 86695
    },
    {
      "epoch": 76.79362267493357,
      "grad_norm": 0.24124813079833984,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 86700
    },
    {
      "epoch": 76.79805137289637,
      "grad_norm": 0.22239917516708374,
      "learning_rate": 1e-05,
      "loss": 0.9072,
      "step": 86705
    },
    {
      "epoch": 76.80248007085916,
      "grad_norm": 0.22841840982437134,
      "learning_rate": 1e-05,
      "loss": 0.9181,
      "step": 86710
    },
    {
      "epoch": 76.80690876882197,
      "grad_norm": 0.22420372068881989,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 86715
    },
    {
      "epoch": 76.81133746678476,
      "grad_norm": 0.22873982787132263,
      "learning_rate": 1e-05,
      "loss": 0.936,
      "step": 86720
    },
    {
      "epoch": 76.81576616474756,
      "grad_norm": 0.2477078139781952,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 86725
    },
    {
      "epoch": 76.82019486271037,
      "grad_norm": 0.2596331238746643,
      "learning_rate": 1e-05,
      "loss": 0.9221,
      "step": 86730
    },
    {
      "epoch": 76.82462356067316,
      "grad_norm": 0.25076568126678467,
      "learning_rate": 1e-05,
      "loss": 0.903,
      "step": 86735
    },
    {
      "epoch": 76.82905225863595,
      "grad_norm": 0.2529015839099884,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 86740
    },
    {
      "epoch": 76.83348095659876,
      "grad_norm": 0.27693265676498413,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 86745
    },
    {
      "epoch": 76.83790965456156,
      "grad_norm": 0.22784750163555145,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 86750
    },
    {
      "epoch": 76.84233835252435,
      "grad_norm": 0.2062016874551773,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 86755
    },
    {
      "epoch": 76.84676705048716,
      "grad_norm": 0.2384243607521057,
      "learning_rate": 1e-05,
      "loss": 0.9413,
      "step": 86760
    },
    {
      "epoch": 76.85119574844995,
      "grad_norm": 0.23561438918113708,
      "learning_rate": 1e-05,
      "loss": 0.9223,
      "step": 86765
    },
    {
      "epoch": 76.85562444641276,
      "grad_norm": 0.23912149667739868,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 86770
    },
    {
      "epoch": 76.86005314437556,
      "grad_norm": 0.25323787331581116,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 86775
    },
    {
      "epoch": 76.86448184233835,
      "grad_norm": 0.23842278122901917,
      "learning_rate": 1e-05,
      "loss": 0.9182,
      "step": 86780
    },
    {
      "epoch": 76.86891054030116,
      "grad_norm": 0.2314378321170807,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 86785
    },
    {
      "epoch": 76.87333923826395,
      "grad_norm": 0.2654232680797577,
      "learning_rate": 1e-05,
      "loss": 0.9179,
      "step": 86790
    },
    {
      "epoch": 76.87776793622675,
      "grad_norm": 0.27247875928878784,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 86795
    },
    {
      "epoch": 76.88219663418955,
      "grad_norm": 0.220348060131073,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 86800
    },
    {
      "epoch": 76.88662533215235,
      "grad_norm": 0.225404292345047,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 86805
    },
    {
      "epoch": 76.89105403011514,
      "grad_norm": 0.2505453824996948,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 86810
    },
    {
      "epoch": 76.89548272807795,
      "grad_norm": 0.2077699899673462,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 86815
    },
    {
      "epoch": 76.89991142604075,
      "grad_norm": 0.2339121550321579,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 86820
    },
    {
      "epoch": 76.90434012400354,
      "grad_norm": 0.2175145298242569,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 86825
    },
    {
      "epoch": 76.90876882196635,
      "grad_norm": 0.23567888140678406,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 86830
    },
    {
      "epoch": 76.91319751992914,
      "grad_norm": 0.23579953610897064,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 86835
    },
    {
      "epoch": 76.91762621789194,
      "grad_norm": 0.20956461131572723,
      "learning_rate": 1e-05,
      "loss": 1.0361,
      "step": 86840
    },
    {
      "epoch": 76.92205491585474,
      "grad_norm": 0.24520070850849152,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 86845
    },
    {
      "epoch": 76.92648361381754,
      "grad_norm": 0.23382751643657684,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 86850
    },
    {
      "epoch": 76.93091231178033,
      "grad_norm": 0.27794188261032104,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 86855
    },
    {
      "epoch": 76.93534100974314,
      "grad_norm": 0.22913743555545807,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 86860
    },
    {
      "epoch": 76.93976970770593,
      "grad_norm": 0.2015327662229538,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 86865
    },
    {
      "epoch": 76.94419840566873,
      "grad_norm": 0.2482973039150238,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 86870
    },
    {
      "epoch": 76.94862710363154,
      "grad_norm": 0.23543602228164673,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 86875
    },
    {
      "epoch": 76.95305580159433,
      "grad_norm": 0.26965153217315674,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 86880
    },
    {
      "epoch": 76.95748449955713,
      "grad_norm": 0.23308025300502777,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 86885
    },
    {
      "epoch": 76.96191319751993,
      "grad_norm": 0.2172469049692154,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 86890
    },
    {
      "epoch": 76.96634189548273,
      "grad_norm": 0.2006201595067978,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 86895
    },
    {
      "epoch": 76.97077059344552,
      "grad_norm": 0.23705631494522095,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 86900
    },
    {
      "epoch": 76.97519929140833,
      "grad_norm": 0.2509170174598694,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 86905
    },
    {
      "epoch": 76.97962798937112,
      "grad_norm": 0.31459057331085205,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 86910
    },
    {
      "epoch": 76.98405668733392,
      "grad_norm": 0.24824321269989014,
      "learning_rate": 1e-05,
      "loss": 0.8998,
      "step": 86915
    },
    {
      "epoch": 76.98848538529673,
      "grad_norm": 0.2622279226779938,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 86920
    },
    {
      "epoch": 76.99291408325952,
      "grad_norm": 0.2236255705356598,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 86925
    },
    {
      "epoch": 76.99734278122232,
      "grad_norm": 0.24968622624874115,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 86930
    },
    {
      "epoch": 77.00177147918512,
      "grad_norm": 0.25333523750305176,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 86935
    },
    {
      "epoch": 77.00620017714792,
      "grad_norm": 0.2364991158246994,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 86940
    },
    {
      "epoch": 77.01062887511071,
      "grad_norm": 0.2518218755722046,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 86945
    },
    {
      "epoch": 77.01505757307352,
      "grad_norm": 0.21034954488277435,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 86950
    },
    {
      "epoch": 77.01948627103631,
      "grad_norm": 0.26711803674697876,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 86955
    },
    {
      "epoch": 77.02391496899911,
      "grad_norm": 0.28568345308303833,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 86960
    },
    {
      "epoch": 77.02834366696192,
      "grad_norm": 0.22148174047470093,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 86965
    },
    {
      "epoch": 77.03277236492471,
      "grad_norm": 0.1994667500257492,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 86970
    },
    {
      "epoch": 77.0372010628875,
      "grad_norm": 0.22593165934085846,
      "learning_rate": 1e-05,
      "loss": 1.0304,
      "step": 86975
    },
    {
      "epoch": 77.04162976085031,
      "grad_norm": 0.2594856023788452,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 86980
    },
    {
      "epoch": 77.0460584588131,
      "grad_norm": 0.2465016096830368,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 86985
    },
    {
      "epoch": 77.0504871567759,
      "grad_norm": 0.21591578423976898,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 86990
    },
    {
      "epoch": 77.05491585473871,
      "grad_norm": 0.28775516152381897,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 86995
    },
    {
      "epoch": 77.0593445527015,
      "grad_norm": 0.2348496913909912,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 87000
    },
    {
      "epoch": 77.0637732506643,
      "grad_norm": 0.23258076608181,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 87005
    },
    {
      "epoch": 77.0682019486271,
      "grad_norm": 0.21153271198272705,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 87010
    },
    {
      "epoch": 77.0726306465899,
      "grad_norm": 0.2139168083667755,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 87015
    },
    {
      "epoch": 77.07705934455271,
      "grad_norm": 0.2559416890144348,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 87020
    },
    {
      "epoch": 77.0814880425155,
      "grad_norm": 0.26610124111175537,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 87025
    },
    {
      "epoch": 77.0859167404783,
      "grad_norm": 0.2436743825674057,
      "learning_rate": 1e-05,
      "loss": 0.9255,
      "step": 87030
    },
    {
      "epoch": 77.0903454384411,
      "grad_norm": 0.2805301547050476,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 87035
    },
    {
      "epoch": 77.0947741364039,
      "grad_norm": 0.2683490812778473,
      "learning_rate": 1e-05,
      "loss": 0.9278,
      "step": 87040
    },
    {
      "epoch": 77.0992028343667,
      "grad_norm": 0.21891073882579803,
      "learning_rate": 1e-05,
      "loss": 1.0377,
      "step": 87045
    },
    {
      "epoch": 77.1036315323295,
      "grad_norm": 0.25783663988113403,
      "learning_rate": 1e-05,
      "loss": 0.8835,
      "step": 87050
    },
    {
      "epoch": 77.1080602302923,
      "grad_norm": 0.23595960438251495,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 87055
    },
    {
      "epoch": 77.11248892825509,
      "grad_norm": 0.2305869609117508,
      "learning_rate": 1e-05,
      "loss": 0.9234,
      "step": 87060
    },
    {
      "epoch": 77.1169176262179,
      "grad_norm": 0.23492711782455444,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 87065
    },
    {
      "epoch": 77.12134632418069,
      "grad_norm": 0.233218714594841,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 87070
    },
    {
      "epoch": 77.12577502214349,
      "grad_norm": 0.2314988374710083,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 87075
    },
    {
      "epoch": 77.1302037201063,
      "grad_norm": 0.2374245524406433,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 87080
    },
    {
      "epoch": 77.13463241806909,
      "grad_norm": 0.25992587208747864,
      "learning_rate": 1e-05,
      "loss": 0.9325,
      "step": 87085
    },
    {
      "epoch": 77.13906111603188,
      "grad_norm": 0.233268603682518,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 87090
    },
    {
      "epoch": 77.14348981399469,
      "grad_norm": 0.2239455282688141,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 87095
    },
    {
      "epoch": 77.14791851195749,
      "grad_norm": 0.24732601642608643,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 87100
    },
    {
      "epoch": 77.15234720992028,
      "grad_norm": 0.26330986618995667,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 87105
    },
    {
      "epoch": 77.15677590788309,
      "grad_norm": 0.3085660934448242,
      "learning_rate": 1e-05,
      "loss": 0.923,
      "step": 87110
    },
    {
      "epoch": 77.16120460584588,
      "grad_norm": 0.2828408181667328,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 87115
    },
    {
      "epoch": 77.16563330380868,
      "grad_norm": 0.23471574485301971,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 87120
    },
    {
      "epoch": 77.17006200177148,
      "grad_norm": 0.22224895656108856,
      "learning_rate": 1e-05,
      "loss": 0.8937,
      "step": 87125
    },
    {
      "epoch": 77.17449069973428,
      "grad_norm": 0.24026210606098175,
      "learning_rate": 1e-05,
      "loss": 0.8974,
      "step": 87130
    },
    {
      "epoch": 77.17891939769707,
      "grad_norm": 0.2387019544839859,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 87135
    },
    {
      "epoch": 77.18334809565988,
      "grad_norm": 0.248616024851799,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 87140
    },
    {
      "epoch": 77.18777679362267,
      "grad_norm": 0.25219297409057617,
      "learning_rate": 1e-05,
      "loss": 0.9292,
      "step": 87145
    },
    {
      "epoch": 77.19220549158547,
      "grad_norm": 0.2501124143600464,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 87150
    },
    {
      "epoch": 77.19663418954828,
      "grad_norm": 0.2432943433523178,
      "learning_rate": 1e-05,
      "loss": 0.9096,
      "step": 87155
    },
    {
      "epoch": 77.20106288751107,
      "grad_norm": 0.26558417081832886,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 87160
    },
    {
      "epoch": 77.20549158547387,
      "grad_norm": 0.2285250723361969,
      "learning_rate": 1e-05,
      "loss": 0.8989,
      "step": 87165
    },
    {
      "epoch": 77.20992028343667,
      "grad_norm": 0.2780630588531494,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 87170
    },
    {
      "epoch": 77.21434898139947,
      "grad_norm": 0.2030693143606186,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 87175
    },
    {
      "epoch": 77.21877767936226,
      "grad_norm": 0.26875707507133484,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 87180
    },
    {
      "epoch": 77.22320637732507,
      "grad_norm": 0.2510926127433777,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 87185
    },
    {
      "epoch": 77.22763507528786,
      "grad_norm": 0.24003753066062927,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 87190
    },
    {
      "epoch": 77.23206377325066,
      "grad_norm": 0.24508704245090485,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 87195
    },
    {
      "epoch": 77.23649247121347,
      "grad_norm": 0.2086341381072998,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 87200
    },
    {
      "epoch": 77.24092116917626,
      "grad_norm": 0.26420485973358154,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 87205
    },
    {
      "epoch": 77.24534986713905,
      "grad_norm": 0.25184863805770874,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 87210
    },
    {
      "epoch": 77.24977856510186,
      "grad_norm": 0.23711846768856049,
      "learning_rate": 1e-05,
      "loss": 1.0242,
      "step": 87215
    },
    {
      "epoch": 77.25420726306466,
      "grad_norm": 0.2293325662612915,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 87220
    },
    {
      "epoch": 77.25863596102745,
      "grad_norm": 0.2318987250328064,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 87225
    },
    {
      "epoch": 77.26306465899026,
      "grad_norm": 0.26365166902542114,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 87230
    },
    {
      "epoch": 77.26749335695305,
      "grad_norm": 0.22320283949375153,
      "learning_rate": 1e-05,
      "loss": 1.0335,
      "step": 87235
    },
    {
      "epoch": 77.27192205491585,
      "grad_norm": 0.26045191287994385,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 87240
    },
    {
      "epoch": 77.27635075287866,
      "grad_norm": 0.2685241401195526,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 87245
    },
    {
      "epoch": 77.28077945084145,
      "grad_norm": 0.2308633178472519,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 87250
    },
    {
      "epoch": 77.28520814880426,
      "grad_norm": 0.23745161294937134,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 87255
    },
    {
      "epoch": 77.28963684676705,
      "grad_norm": 0.21297231316566467,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 87260
    },
    {
      "epoch": 77.29406554472985,
      "grad_norm": 0.29279035329818726,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 87265
    },
    {
      "epoch": 77.29849424269266,
      "grad_norm": 0.29635563492774963,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 87270
    },
    {
      "epoch": 77.30292294065545,
      "grad_norm": 0.2522418797016144,
      "learning_rate": 1e-05,
      "loss": 0.9194,
      "step": 87275
    },
    {
      "epoch": 77.30735163861824,
      "grad_norm": 0.23736627399921417,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 87280
    },
    {
      "epoch": 77.31178033658105,
      "grad_norm": 0.21685829758644104,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 87285
    },
    {
      "epoch": 77.31620903454385,
      "grad_norm": 0.24377422034740448,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 87290
    },
    {
      "epoch": 77.32063773250664,
      "grad_norm": 0.23714256286621094,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 87295
    },
    {
      "epoch": 77.32506643046945,
      "grad_norm": 0.20305867493152618,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 87300
    },
    {
      "epoch": 77.32949512843224,
      "grad_norm": 0.22839532792568207,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 87305
    },
    {
      "epoch": 77.33392382639504,
      "grad_norm": 0.2554134726524353,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 87310
    },
    {
      "epoch": 77.33835252435784,
      "grad_norm": 0.20826709270477295,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 87315
    },
    {
      "epoch": 77.34278122232064,
      "grad_norm": 0.25881198048591614,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 87320
    },
    {
      "epoch": 77.34720992028343,
      "grad_norm": 0.21859923005104065,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 87325
    },
    {
      "epoch": 77.35163861824624,
      "grad_norm": 0.22646629810333252,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 87330
    },
    {
      "epoch": 77.35606731620904,
      "grad_norm": 0.22934813797473907,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 87335
    },
    {
      "epoch": 77.36049601417183,
      "grad_norm": 0.2469106912612915,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 87340
    },
    {
      "epoch": 77.36492471213464,
      "grad_norm": 0.23186412453651428,
      "learning_rate": 1e-05,
      "loss": 0.8968,
      "step": 87345
    },
    {
      "epoch": 77.36935341009743,
      "grad_norm": 0.25305822491645813,
      "learning_rate": 1e-05,
      "loss": 0.9195,
      "step": 87350
    },
    {
      "epoch": 77.37378210806023,
      "grad_norm": 0.2502666711807251,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 87355
    },
    {
      "epoch": 77.37821080602303,
      "grad_norm": 0.22845736145973206,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 87360
    },
    {
      "epoch": 77.38263950398583,
      "grad_norm": 0.1960151642560959,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 87365
    },
    {
      "epoch": 77.38706820194862,
      "grad_norm": 0.22671368718147278,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 87370
    },
    {
      "epoch": 77.39149689991143,
      "grad_norm": 0.2935572564601898,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 87375
    },
    {
      "epoch": 77.39592559787422,
      "grad_norm": 0.2797524034976959,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 87380
    },
    {
      "epoch": 77.40035429583702,
      "grad_norm": 0.24903050065040588,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 87385
    },
    {
      "epoch": 77.40478299379983,
      "grad_norm": 0.26196572184562683,
      "learning_rate": 1e-05,
      "loss": 0.9091,
      "step": 87390
    },
    {
      "epoch": 77.40921169176262,
      "grad_norm": 0.2857091724872589,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 87395
    },
    {
      "epoch": 77.41364038972542,
      "grad_norm": 0.2554527223110199,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 87400
    },
    {
      "epoch": 77.41806908768822,
      "grad_norm": 0.214559406042099,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 87405
    },
    {
      "epoch": 77.42249778565102,
      "grad_norm": 0.20117905735969543,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 87410
    },
    {
      "epoch": 77.42692648361381,
      "grad_norm": 0.22684235870838165,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 87415
    },
    {
      "epoch": 77.43135518157662,
      "grad_norm": 0.22124417126178741,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 87420
    },
    {
      "epoch": 77.43578387953941,
      "grad_norm": 0.22114130854606628,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 87425
    },
    {
      "epoch": 77.44021257750221,
      "grad_norm": 0.22874243557453156,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 87430
    },
    {
      "epoch": 77.44464127546502,
      "grad_norm": 0.2213953584432602,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 87435
    },
    {
      "epoch": 77.44906997342781,
      "grad_norm": 0.2078544646501541,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 87440
    },
    {
      "epoch": 77.4534986713906,
      "grad_norm": 0.280277281999588,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 87445
    },
    {
      "epoch": 77.45792736935341,
      "grad_norm": 0.20978453755378723,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 87450
    },
    {
      "epoch": 77.46235606731621,
      "grad_norm": 0.2221294343471527,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 87455
    },
    {
      "epoch": 77.466784765279,
      "grad_norm": 0.2494630664587021,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 87460
    },
    {
      "epoch": 77.47121346324181,
      "grad_norm": 0.2390281707048416,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 87465
    },
    {
      "epoch": 77.4756421612046,
      "grad_norm": 0.2283698320388794,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 87470
    },
    {
      "epoch": 77.4800708591674,
      "grad_norm": 0.21567218005657196,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 87475
    },
    {
      "epoch": 77.4844995571302,
      "grad_norm": 0.24495169520378113,
      "learning_rate": 1e-05,
      "loss": 1.0283,
      "step": 87480
    },
    {
      "epoch": 77.488928255093,
      "grad_norm": 0.30073457956314087,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 87485
    },
    {
      "epoch": 77.4933569530558,
      "grad_norm": 0.2280602902173996,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 87490
    },
    {
      "epoch": 77.4977856510186,
      "grad_norm": 0.24839480221271515,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 87495
    },
    {
      "epoch": 77.5022143489814,
      "grad_norm": 0.26367878913879395,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 87500
    },
    {
      "epoch": 77.5066430469442,
      "grad_norm": 0.22025717794895172,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 87505
    },
    {
      "epoch": 77.511071744907,
      "grad_norm": 0.21963772177696228,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 87510
    },
    {
      "epoch": 77.5155004428698,
      "grad_norm": 0.22366157174110413,
      "learning_rate": 1e-05,
      "loss": 1.0051,
      "step": 87515
    },
    {
      "epoch": 77.5199291408326,
      "grad_norm": 0.22449831664562225,
      "learning_rate": 1e-05,
      "loss": 0.9335,
      "step": 87520
    },
    {
      "epoch": 77.5243578387954,
      "grad_norm": 0.23920750617980957,
      "learning_rate": 1e-05,
      "loss": 0.9303,
      "step": 87525
    },
    {
      "epoch": 77.52878653675819,
      "grad_norm": 0.21678584814071655,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 87530
    },
    {
      "epoch": 77.533215234721,
      "grad_norm": 0.2494162917137146,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 87535
    },
    {
      "epoch": 77.53764393268379,
      "grad_norm": 0.2327771931886673,
      "learning_rate": 1e-05,
      "loss": 0.9253,
      "step": 87540
    },
    {
      "epoch": 77.54207263064659,
      "grad_norm": 0.22130730748176575,
      "learning_rate": 1e-05,
      "loss": 0.9209,
      "step": 87545
    },
    {
      "epoch": 77.5465013286094,
      "grad_norm": 0.24990203976631165,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 87550
    },
    {
      "epoch": 77.55093002657219,
      "grad_norm": 0.23619632422924042,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 87555
    },
    {
      "epoch": 77.55535872453498,
      "grad_norm": 0.22352677583694458,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 87560
    },
    {
      "epoch": 77.55978742249779,
      "grad_norm": 0.24528370797634125,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 87565
    },
    {
      "epoch": 77.56421612046059,
      "grad_norm": 0.2725163996219635,
      "learning_rate": 1e-05,
      "loss": 1.0187,
      "step": 87570
    },
    {
      "epoch": 77.56864481842338,
      "grad_norm": 0.208133265376091,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 87575
    },
    {
      "epoch": 77.57307351638619,
      "grad_norm": 0.23830865323543549,
      "learning_rate": 1e-05,
      "loss": 0.9169,
      "step": 87580
    },
    {
      "epoch": 77.57750221434898,
      "grad_norm": 0.23724903166294098,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 87585
    },
    {
      "epoch": 77.58193091231178,
      "grad_norm": 0.2735835909843445,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 87590
    },
    {
      "epoch": 77.58635961027458,
      "grad_norm": 0.23597557842731476,
      "learning_rate": 1e-05,
      "loss": 1.0213,
      "step": 87595
    },
    {
      "epoch": 77.59078830823738,
      "grad_norm": 0.267072856426239,
      "learning_rate": 1e-05,
      "loss": 1.0178,
      "step": 87600
    },
    {
      "epoch": 77.59521700620017,
      "grad_norm": 0.2662096917629242,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 87605
    },
    {
      "epoch": 77.59964570416298,
      "grad_norm": 0.2577826976776123,
      "learning_rate": 1e-05,
      "loss": 1.0185,
      "step": 87610
    },
    {
      "epoch": 77.60407440212578,
      "grad_norm": 0.23772192001342773,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 87615
    },
    {
      "epoch": 77.60850310008857,
      "grad_norm": 0.2308945655822754,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 87620
    },
    {
      "epoch": 77.61293179805138,
      "grad_norm": 0.2857731878757477,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 87625
    },
    {
      "epoch": 77.61736049601417,
      "grad_norm": 0.2752941846847534,
      "learning_rate": 1e-05,
      "loss": 0.9246,
      "step": 87630
    },
    {
      "epoch": 77.62178919397697,
      "grad_norm": 0.22603636980056763,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 87635
    },
    {
      "epoch": 77.62621789193977,
      "grad_norm": 0.2126711905002594,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 87640
    },
    {
      "epoch": 77.63064658990257,
      "grad_norm": 0.2467328906059265,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 87645
    },
    {
      "epoch": 77.63507528786536,
      "grad_norm": 0.2128833383321762,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 87650
    },
    {
      "epoch": 77.63950398582817,
      "grad_norm": 0.22866934537887573,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 87655
    },
    {
      "epoch": 77.64393268379096,
      "grad_norm": 0.25474923849105835,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 87660
    },
    {
      "epoch": 77.64836138175376,
      "grad_norm": 0.28724879026412964,
      "learning_rate": 1e-05,
      "loss": 1.0425,
      "step": 87665
    },
    {
      "epoch": 77.65279007971657,
      "grad_norm": 0.28412649035453796,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 87670
    },
    {
      "epoch": 77.65721877767936,
      "grad_norm": 0.24339643120765686,
      "learning_rate": 1e-05,
      "loss": 0.8986,
      "step": 87675
    },
    {
      "epoch": 77.66164747564216,
      "grad_norm": 0.26179274916648865,
      "learning_rate": 1e-05,
      "loss": 0.9152,
      "step": 87680
    },
    {
      "epoch": 77.66607617360496,
      "grad_norm": 0.20889487862586975,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 87685
    },
    {
      "epoch": 77.67050487156776,
      "grad_norm": 0.22540608048439026,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 87690
    },
    {
      "epoch": 77.67493356953055,
      "grad_norm": 0.23712977766990662,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 87695
    },
    {
      "epoch": 77.67936226749336,
      "grad_norm": 0.23204006254673004,
      "learning_rate": 1e-05,
      "loss": 0.9313,
      "step": 87700
    },
    {
      "epoch": 77.68379096545615,
      "grad_norm": 0.22438377141952515,
      "learning_rate": 1e-05,
      "loss": 0.8877,
      "step": 87705
    },
    {
      "epoch": 77.68821966341895,
      "grad_norm": 0.24378086626529694,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 87710
    },
    {
      "epoch": 77.69264836138176,
      "grad_norm": 0.23039083182811737,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 87715
    },
    {
      "epoch": 77.69707705934455,
      "grad_norm": 0.21411454677581787,
      "learning_rate": 1e-05,
      "loss": 0.9227,
      "step": 87720
    },
    {
      "epoch": 77.70150575730734,
      "grad_norm": 0.23939155042171478,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 87725
    },
    {
      "epoch": 77.70593445527015,
      "grad_norm": 0.24701708555221558,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 87730
    },
    {
      "epoch": 77.71036315323295,
      "grad_norm": 0.26469677686691284,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 87735
    },
    {
      "epoch": 77.71479185119574,
      "grad_norm": 0.1970253437757492,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 87740
    },
    {
      "epoch": 77.71922054915855,
      "grad_norm": 0.23921014368534088,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 87745
    },
    {
      "epoch": 77.72364924712134,
      "grad_norm": 0.2604734003543854,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 87750
    },
    {
      "epoch": 77.72807794508415,
      "grad_norm": 0.2523438632488251,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 87755
    },
    {
      "epoch": 77.73250664304695,
      "grad_norm": 0.23461085557937622,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 87760
    },
    {
      "epoch": 77.73693534100974,
      "grad_norm": 0.2170139104127884,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 87765
    },
    {
      "epoch": 77.74136403897255,
      "grad_norm": 0.21661920845508575,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 87770
    },
    {
      "epoch": 77.74579273693534,
      "grad_norm": 0.20358291268348694,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 87775
    },
    {
      "epoch": 77.75022143489814,
      "grad_norm": 0.23301099240779877,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 87780
    },
    {
      "epoch": 77.75465013286095,
      "grad_norm": 0.2260369211435318,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 87785
    },
    {
      "epoch": 77.75907883082374,
      "grad_norm": 0.22996878623962402,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 87790
    },
    {
      "epoch": 77.76350752878653,
      "grad_norm": 0.25925150513648987,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 87795
    },
    {
      "epoch": 77.76793622674934,
      "grad_norm": 0.23372302949428558,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 87800
    },
    {
      "epoch": 77.77236492471214,
      "grad_norm": 0.2515539526939392,
      "learning_rate": 1e-05,
      "loss": 0.9307,
      "step": 87805
    },
    {
      "epoch": 77.77679362267493,
      "grad_norm": 0.22752253711223602,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 87810
    },
    {
      "epoch": 77.78122232063774,
      "grad_norm": 0.21126540005207062,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 87815
    },
    {
      "epoch": 77.78565101860053,
      "grad_norm": 0.2386217713356018,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 87820
    },
    {
      "epoch": 77.79007971656333,
      "grad_norm": 0.2570544183254242,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 87825
    },
    {
      "epoch": 77.79450841452613,
      "grad_norm": 0.20882780849933624,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 87830
    },
    {
      "epoch": 77.79893711248893,
      "grad_norm": 0.20831246674060822,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 87835
    },
    {
      "epoch": 77.80336581045172,
      "grad_norm": 0.2543601393699646,
      "learning_rate": 1e-05,
      "loss": 0.9296,
      "step": 87840
    },
    {
      "epoch": 77.80779450841453,
      "grad_norm": 0.23801063001155853,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 87845
    },
    {
      "epoch": 77.81222320637733,
      "grad_norm": 0.26623669266700745,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 87850
    },
    {
      "epoch": 77.81665190434012,
      "grad_norm": 0.26402807235717773,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 87855
    },
    {
      "epoch": 77.82108060230293,
      "grad_norm": 0.26201727986335754,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 87860
    },
    {
      "epoch": 77.82550930026572,
      "grad_norm": 0.26105183362960815,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 87865
    },
    {
      "epoch": 77.82993799822852,
      "grad_norm": 0.2734915316104889,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 87870
    },
    {
      "epoch": 77.83436669619132,
      "grad_norm": 0.22708305716514587,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 87875
    },
    {
      "epoch": 77.83879539415412,
      "grad_norm": 0.24792613089084625,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 87880
    },
    {
      "epoch": 77.84322409211691,
      "grad_norm": 0.2871337831020355,
      "learning_rate": 1e-05,
      "loss": 0.9233,
      "step": 87885
    },
    {
      "epoch": 77.84765279007972,
      "grad_norm": 0.26960790157318115,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 87890
    },
    {
      "epoch": 77.85208148804251,
      "grad_norm": 0.2633669376373291,
      "learning_rate": 1e-05,
      "loss": 1.0288,
      "step": 87895
    },
    {
      "epoch": 77.85651018600531,
      "grad_norm": 0.22795478999614716,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 87900
    },
    {
      "epoch": 77.86093888396812,
      "grad_norm": 0.22482727468013763,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 87905
    },
    {
      "epoch": 77.86536758193091,
      "grad_norm": 0.21699407696723938,
      "learning_rate": 1e-05,
      "loss": 0.9231,
      "step": 87910
    },
    {
      "epoch": 77.8697962798937,
      "grad_norm": 0.22380073368549347,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 87915
    },
    {
      "epoch": 77.87422497785651,
      "grad_norm": 0.2231159210205078,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 87920
    },
    {
      "epoch": 77.87865367581931,
      "grad_norm": 0.22316156327724457,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 87925
    },
    {
      "epoch": 77.8830823737821,
      "grad_norm": 0.24263930320739746,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 87930
    },
    {
      "epoch": 77.88751107174491,
      "grad_norm": 0.22376778721809387,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 87935
    },
    {
      "epoch": 77.8919397697077,
      "grad_norm": 0.23587343096733093,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 87940
    },
    {
      "epoch": 77.8963684676705,
      "grad_norm": 0.23348736763000488,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 87945
    },
    {
      "epoch": 77.9007971656333,
      "grad_norm": 0.2668285071849823,
      "learning_rate": 1e-05,
      "loss": 1.0384,
      "step": 87950
    },
    {
      "epoch": 77.9052258635961,
      "grad_norm": 0.24204878509044647,
      "learning_rate": 1e-05,
      "loss": 0.9275,
      "step": 87955
    },
    {
      "epoch": 77.9096545615589,
      "grad_norm": 0.24683523178100586,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 87960
    },
    {
      "epoch": 77.9140832595217,
      "grad_norm": 0.27237313985824585,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 87965
    },
    {
      "epoch": 77.9185119574845,
      "grad_norm": 0.220925971865654,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 87970
    },
    {
      "epoch": 77.92294065544729,
      "grad_norm": 0.2781432271003723,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 87975
    },
    {
      "epoch": 77.9273693534101,
      "grad_norm": 0.22386308014392853,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 87980
    },
    {
      "epoch": 77.9317980513729,
      "grad_norm": 0.2146822065114975,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 87985
    },
    {
      "epoch": 77.9362267493357,
      "grad_norm": 0.21760590374469757,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 87990
    },
    {
      "epoch": 77.9406554472985,
      "grad_norm": 0.23479552567005157,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 87995
    },
    {
      "epoch": 77.94508414526129,
      "grad_norm": 0.2356594353914261,
      "learning_rate": 1e-05,
      "loss": 0.9182,
      "step": 88000
    },
    {
      "epoch": 77.9495128432241,
      "grad_norm": 0.261869877576828,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 88005
    },
    {
      "epoch": 77.9539415411869,
      "grad_norm": 0.2639724314212799,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 88010
    },
    {
      "epoch": 77.95837023914969,
      "grad_norm": 0.2590380012989044,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 88015
    },
    {
      "epoch": 77.9627989371125,
      "grad_norm": 0.2274603396654129,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 88020
    },
    {
      "epoch": 77.96722763507529,
      "grad_norm": 0.2468695491552353,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 88025
    },
    {
      "epoch": 77.97165633303808,
      "grad_norm": 0.2372131049633026,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 88030
    },
    {
      "epoch": 77.97608503100089,
      "grad_norm": 0.22166943550109863,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 88035
    },
    {
      "epoch": 77.98051372896369,
      "grad_norm": 0.23885931074619293,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 88040
    },
    {
      "epoch": 77.98494242692648,
      "grad_norm": 0.23855599761009216,
      "learning_rate": 1e-05,
      "loss": 1.0206,
      "step": 88045
    },
    {
      "epoch": 77.98937112488929,
      "grad_norm": 0.22656738758087158,
      "learning_rate": 1e-05,
      "loss": 0.9248,
      "step": 88050
    },
    {
      "epoch": 77.99379982285208,
      "grad_norm": 0.2272365689277649,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 88055
    },
    {
      "epoch": 77.99822852081488,
      "grad_norm": 0.24719080328941345,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 88060
    },
    {
      "epoch": 78.00265721877768,
      "grad_norm": 0.21056294441223145,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 88065
    },
    {
      "epoch": 78.00708591674048,
      "grad_norm": 0.21819378435611725,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 88070
    },
    {
      "epoch": 78.01151461470327,
      "grad_norm": 0.23172403872013092,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 88075
    },
    {
      "epoch": 78.01594331266608,
      "grad_norm": 0.2348961979150772,
      "learning_rate": 1e-05,
      "loss": 0.9324,
      "step": 88080
    },
    {
      "epoch": 78.02037201062888,
      "grad_norm": 0.25585222244262695,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 88085
    },
    {
      "epoch": 78.02480070859167,
      "grad_norm": 0.2880314290523529,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 88090
    },
    {
      "epoch": 78.02922940655448,
      "grad_norm": 0.3272036612033844,
      "learning_rate": 1e-05,
      "loss": 0.924,
      "step": 88095
    },
    {
      "epoch": 78.03365810451727,
      "grad_norm": 0.2883589565753937,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 88100
    },
    {
      "epoch": 78.03808680248007,
      "grad_norm": 0.2714685797691345,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 88105
    },
    {
      "epoch": 78.04251550044287,
      "grad_norm": 0.24066340923309326,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 88110
    },
    {
      "epoch": 78.04694419840567,
      "grad_norm": 0.23197868466377258,
      "learning_rate": 1e-05,
      "loss": 0.9124,
      "step": 88115
    },
    {
      "epoch": 78.05137289636846,
      "grad_norm": 0.21554987132549286,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 88120
    },
    {
      "epoch": 78.05580159433127,
      "grad_norm": 0.2372073233127594,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 88125
    },
    {
      "epoch": 78.06023029229407,
      "grad_norm": 0.24696719646453857,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 88130
    },
    {
      "epoch": 78.06465899025686,
      "grad_norm": 0.26418161392211914,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 88135
    },
    {
      "epoch": 78.06908768821967,
      "grad_norm": 0.2248615324497223,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 88140
    },
    {
      "epoch": 78.07351638618246,
      "grad_norm": 0.2294880449771881,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 88145
    },
    {
      "epoch": 78.07794508414526,
      "grad_norm": 0.2494615912437439,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 88150
    },
    {
      "epoch": 78.08237378210806,
      "grad_norm": 0.25067639350891113,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 88155
    },
    {
      "epoch": 78.08680248007086,
      "grad_norm": 0.27057117223739624,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 88160
    },
    {
      "epoch": 78.09123117803365,
      "grad_norm": 0.2219180017709732,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 88165
    },
    {
      "epoch": 78.09565987599646,
      "grad_norm": 0.2449643760919571,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 88170
    },
    {
      "epoch": 78.10008857395925,
      "grad_norm": 0.24394257366657257,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 88175
    },
    {
      "epoch": 78.10451727192205,
      "grad_norm": 0.28916069865226746,
      "learning_rate": 1e-05,
      "loss": 0.9278,
      "step": 88180
    },
    {
      "epoch": 78.10894596988486,
      "grad_norm": 0.21044951677322388,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 88185
    },
    {
      "epoch": 78.11337466784765,
      "grad_norm": 0.21388471126556396,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 88190
    },
    {
      "epoch": 78.11780336581045,
      "grad_norm": 0.26471683382987976,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 88195
    },
    {
      "epoch": 78.12223206377325,
      "grad_norm": 0.23909151554107666,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 88200
    },
    {
      "epoch": 78.12666076173605,
      "grad_norm": 0.2227104902267456,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 88205
    },
    {
      "epoch": 78.13108945969884,
      "grad_norm": 0.2222926765680313,
      "learning_rate": 1e-05,
      "loss": 0.9017,
      "step": 88210
    },
    {
      "epoch": 78.13551815766165,
      "grad_norm": 0.27274656295776367,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 88215
    },
    {
      "epoch": 78.13994685562444,
      "grad_norm": 0.2666155695915222,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 88220
    },
    {
      "epoch": 78.14437555358724,
      "grad_norm": 0.2284383624792099,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 88225
    },
    {
      "epoch": 78.14880425155005,
      "grad_norm": 0.22449897229671478,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 88230
    },
    {
      "epoch": 78.15323294951284,
      "grad_norm": 0.24238385260105133,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 88235
    },
    {
      "epoch": 78.15766164747565,
      "grad_norm": 0.23849505186080933,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 88240
    },
    {
      "epoch": 78.16209034543844,
      "grad_norm": 0.21940478682518005,
      "learning_rate": 1e-05,
      "loss": 1.0036,
      "step": 88245
    },
    {
      "epoch": 78.16651904340124,
      "grad_norm": 0.2441585212945938,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 88250
    },
    {
      "epoch": 78.17094774136405,
      "grad_norm": 0.2955837845802307,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 88255
    },
    {
      "epoch": 78.17537643932684,
      "grad_norm": 0.2114020437002182,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 88260
    },
    {
      "epoch": 78.17980513728963,
      "grad_norm": 0.27439990639686584,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 88265
    },
    {
      "epoch": 78.18423383525244,
      "grad_norm": 0.21725498139858246,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 88270
    },
    {
      "epoch": 78.18866253321524,
      "grad_norm": 0.2442336231470108,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 88275
    },
    {
      "epoch": 78.19309123117803,
      "grad_norm": 0.2618340253829956,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 88280
    },
    {
      "epoch": 78.19751992914084,
      "grad_norm": 0.25004422664642334,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 88285
    },
    {
      "epoch": 78.20194862710363,
      "grad_norm": 0.2553393840789795,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 88290
    },
    {
      "epoch": 78.20637732506643,
      "grad_norm": 0.23919129371643066,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 88295
    },
    {
      "epoch": 78.21080602302924,
      "grad_norm": 0.2585943043231964,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 88300
    },
    {
      "epoch": 78.21523472099203,
      "grad_norm": 0.2585223317146301,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 88305
    },
    {
      "epoch": 78.21966341895482,
      "grad_norm": 0.22464485466480255,
      "learning_rate": 1e-05,
      "loss": 0.9155,
      "step": 88310
    },
    {
      "epoch": 78.22409211691763,
      "grad_norm": 0.22997532784938812,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 88315
    },
    {
      "epoch": 78.22852081488043,
      "grad_norm": 0.24111147224903107,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 88320
    },
    {
      "epoch": 78.23294951284322,
      "grad_norm": 0.25492793321609497,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 88325
    },
    {
      "epoch": 78.23737821080603,
      "grad_norm": 0.23459693789482117,
      "learning_rate": 1e-05,
      "loss": 0.9208,
      "step": 88330
    },
    {
      "epoch": 78.24180690876882,
      "grad_norm": 0.2433767020702362,
      "learning_rate": 1e-05,
      "loss": 1.0233,
      "step": 88335
    },
    {
      "epoch": 78.24623560673162,
      "grad_norm": 0.23111091554164886,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 88340
    },
    {
      "epoch": 78.25066430469442,
      "grad_norm": 0.2255808711051941,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 88345
    },
    {
      "epoch": 78.25509300265722,
      "grad_norm": 0.2177070677280426,
      "learning_rate": 1e-05,
      "loss": 0.9154,
      "step": 88350
    },
    {
      "epoch": 78.25952170062001,
      "grad_norm": 0.22173449397087097,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 88355
    },
    {
      "epoch": 78.26395039858282,
      "grad_norm": 0.22008760273456573,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 88360
    },
    {
      "epoch": 78.26837909654562,
      "grad_norm": 0.22661566734313965,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 88365
    },
    {
      "epoch": 78.27280779450841,
      "grad_norm": 0.2508631646633148,
      "learning_rate": 1e-05,
      "loss": 0.9017,
      "step": 88370
    },
    {
      "epoch": 78.27723649247122,
      "grad_norm": 0.25236210227012634,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 88375
    },
    {
      "epoch": 78.28166519043401,
      "grad_norm": 0.20679466426372528,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 88380
    },
    {
      "epoch": 78.2860938883968,
      "grad_norm": 0.21976430714130402,
      "learning_rate": 1e-05,
      "loss": 0.9033,
      "step": 88385
    },
    {
      "epoch": 78.29052258635961,
      "grad_norm": 0.3356921374797821,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 88390
    },
    {
      "epoch": 78.29495128432241,
      "grad_norm": 0.25843536853790283,
      "learning_rate": 1e-05,
      "loss": 0.9223,
      "step": 88395
    },
    {
      "epoch": 78.2993799822852,
      "grad_norm": 0.22746503353118896,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 88400
    },
    {
      "epoch": 78.30380868024801,
      "grad_norm": 0.22590909898281097,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 88405
    },
    {
      "epoch": 78.3082373782108,
      "grad_norm": 0.2101229727268219,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 88410
    },
    {
      "epoch": 78.3126660761736,
      "grad_norm": 0.25208696722984314,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 88415
    },
    {
      "epoch": 78.31709477413641,
      "grad_norm": 0.2421363741159439,
      "learning_rate": 1e-05,
      "loss": 0.933,
      "step": 88420
    },
    {
      "epoch": 78.3215234720992,
      "grad_norm": 0.22763632237911224,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 88425
    },
    {
      "epoch": 78.325952170062,
      "grad_norm": 0.2504916787147522,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 88430
    },
    {
      "epoch": 78.3303808680248,
      "grad_norm": 0.24096882343292236,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 88435
    },
    {
      "epoch": 78.3348095659876,
      "grad_norm": 0.22622767090797424,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 88440
    },
    {
      "epoch": 78.33923826395039,
      "grad_norm": 0.2609744966030121,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 88445
    },
    {
      "epoch": 78.3436669619132,
      "grad_norm": 0.2167573720216751,
      "learning_rate": 1e-05,
      "loss": 0.9244,
      "step": 88450
    },
    {
      "epoch": 78.348095659876,
      "grad_norm": 0.1996101289987564,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 88455
    },
    {
      "epoch": 78.35252435783879,
      "grad_norm": 0.22567802667617798,
      "learning_rate": 1e-05,
      "loss": 0.9113,
      "step": 88460
    },
    {
      "epoch": 78.3569530558016,
      "grad_norm": 0.2525363266468048,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 88465
    },
    {
      "epoch": 78.36138175376439,
      "grad_norm": 0.23195043206214905,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 88470
    },
    {
      "epoch": 78.36581045172719,
      "grad_norm": 0.21244372427463531,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 88475
    },
    {
      "epoch": 78.37023914969,
      "grad_norm": 0.25897324085235596,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 88480
    },
    {
      "epoch": 78.37466784765279,
      "grad_norm": 0.27224600315093994,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 88485
    },
    {
      "epoch": 78.3790965456156,
      "grad_norm": 0.27718424797058105,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 88490
    },
    {
      "epoch": 78.38352524357839,
      "grad_norm": 0.26623815298080444,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 88495
    },
    {
      "epoch": 78.38795394154118,
      "grad_norm": 0.23037631809711456,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 88500
    },
    {
      "epoch": 78.39238263950399,
      "grad_norm": 0.22158603370189667,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 88505
    },
    {
      "epoch": 78.39681133746679,
      "grad_norm": 0.22727711498737335,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 88510
    },
    {
      "epoch": 78.40124003542958,
      "grad_norm": 0.2231655716896057,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 88515
    },
    {
      "epoch": 78.40566873339239,
      "grad_norm": 0.25223758816719055,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 88520
    },
    {
      "epoch": 78.41009743135518,
      "grad_norm": 0.24551533162593842,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 88525
    },
    {
      "epoch": 78.41452612931798,
      "grad_norm": 0.266607403755188,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 88530
    },
    {
      "epoch": 78.41895482728079,
      "grad_norm": 0.234235018491745,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 88535
    },
    {
      "epoch": 78.42338352524358,
      "grad_norm": 0.22250519692897797,
      "learning_rate": 1e-05,
      "loss": 0.9062,
      "step": 88540
    },
    {
      "epoch": 78.42781222320637,
      "grad_norm": 0.2604774534702301,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 88545
    },
    {
      "epoch": 78.43224092116918,
      "grad_norm": 0.24216750264167786,
      "learning_rate": 1e-05,
      "loss": 0.9265,
      "step": 88550
    },
    {
      "epoch": 78.43666961913198,
      "grad_norm": 0.22254213690757751,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 88555
    },
    {
      "epoch": 78.44109831709477,
      "grad_norm": 0.2351011037826538,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 88560
    },
    {
      "epoch": 78.44552701505758,
      "grad_norm": 0.23696167767047882,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 88565
    },
    {
      "epoch": 78.44995571302037,
      "grad_norm": 0.2576257586479187,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 88570
    },
    {
      "epoch": 78.45438441098317,
      "grad_norm": 0.24720187485218048,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 88575
    },
    {
      "epoch": 78.45881310894598,
      "grad_norm": 0.238107830286026,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 88580
    },
    {
      "epoch": 78.46324180690877,
      "grad_norm": 0.3204800486564636,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 88585
    },
    {
      "epoch": 78.46767050487156,
      "grad_norm": 0.24230603873729706,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 88590
    },
    {
      "epoch": 78.47209920283437,
      "grad_norm": 0.21399644017219543,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 88595
    },
    {
      "epoch": 78.47652790079717,
      "grad_norm": 0.2422972172498703,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 88600
    },
    {
      "epoch": 78.48095659875996,
      "grad_norm": 0.24130190908908844,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 88605
    },
    {
      "epoch": 78.48538529672277,
      "grad_norm": 0.25818291306495667,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 88610
    },
    {
      "epoch": 78.48981399468556,
      "grad_norm": 0.28703704476356506,
      "learning_rate": 1e-05,
      "loss": 1.0217,
      "step": 88615
    },
    {
      "epoch": 78.49424269264836,
      "grad_norm": 0.22022441029548645,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 88620
    },
    {
      "epoch": 78.49867139061116,
      "grad_norm": 0.25715476274490356,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 88625
    },
    {
      "epoch": 78.50310008857396,
      "grad_norm": 0.2681129574775696,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 88630
    },
    {
      "epoch": 78.50752878653675,
      "grad_norm": 0.24399589002132416,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 88635
    },
    {
      "epoch": 78.51195748449956,
      "grad_norm": 0.28183767199516296,
      "learning_rate": 1e-05,
      "loss": 0.9325,
      "step": 88640
    },
    {
      "epoch": 78.51638618246236,
      "grad_norm": 0.27202925086021423,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 88645
    },
    {
      "epoch": 78.52081488042515,
      "grad_norm": 0.2646104693412781,
      "learning_rate": 1e-05,
      "loss": 0.899,
      "step": 88650
    },
    {
      "epoch": 78.52524357838796,
      "grad_norm": 0.24787229299545288,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 88655
    },
    {
      "epoch": 78.52967227635075,
      "grad_norm": 0.25754332542419434,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 88660
    },
    {
      "epoch": 78.53410097431355,
      "grad_norm": 0.24591226875782013,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 88665
    },
    {
      "epoch": 78.53852967227635,
      "grad_norm": 0.267474502325058,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 88670
    },
    {
      "epoch": 78.54295837023915,
      "grad_norm": 0.2907086908817291,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 88675
    },
    {
      "epoch": 78.54738706820194,
      "grad_norm": 0.2724979817867279,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 88680
    },
    {
      "epoch": 78.55181576616475,
      "grad_norm": 0.2637103497982025,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 88685
    },
    {
      "epoch": 78.55624446412754,
      "grad_norm": 0.2124089002609253,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 88690
    },
    {
      "epoch": 78.56067316209034,
      "grad_norm": 0.2602483034133911,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 88695
    },
    {
      "epoch": 78.56510186005315,
      "grad_norm": 0.23150505125522614,
      "learning_rate": 1e-05,
      "loss": 0.9217,
      "step": 88700
    },
    {
      "epoch": 78.56953055801594,
      "grad_norm": 0.2326960265636444,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 88705
    },
    {
      "epoch": 78.57395925597874,
      "grad_norm": 0.24118410050868988,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 88710
    },
    {
      "epoch": 78.57838795394154,
      "grad_norm": 0.24178457260131836,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 88715
    },
    {
      "epoch": 78.58281665190434,
      "grad_norm": 0.23243384063243866,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 88720
    },
    {
      "epoch": 78.58724534986715,
      "grad_norm": 0.24737320840358734,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 88725
    },
    {
      "epoch": 78.59167404782994,
      "grad_norm": 0.2641957104206085,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 88730
    },
    {
      "epoch": 78.59610274579273,
      "grad_norm": 0.27690625190734863,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 88735
    },
    {
      "epoch": 78.60053144375554,
      "grad_norm": 0.23812811076641083,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 88740
    },
    {
      "epoch": 78.60496014171834,
      "grad_norm": 0.22169415652751923,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 88745
    },
    {
      "epoch": 78.60938883968113,
      "grad_norm": 0.2443220168352127,
      "learning_rate": 1e-05,
      "loss": 0.924,
      "step": 88750
    },
    {
      "epoch": 78.61381753764394,
      "grad_norm": 0.2266269475221634,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 88755
    },
    {
      "epoch": 78.61824623560673,
      "grad_norm": 0.2168651968240738,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 88760
    },
    {
      "epoch": 78.62267493356953,
      "grad_norm": 0.2248748391866684,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 88765
    },
    {
      "epoch": 78.62710363153234,
      "grad_norm": 0.26974332332611084,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 88770
    },
    {
      "epoch": 78.63153232949513,
      "grad_norm": 0.2683171033859253,
      "learning_rate": 1e-05,
      "loss": 1.0393,
      "step": 88775
    },
    {
      "epoch": 78.63596102745792,
      "grad_norm": 0.23703140020370483,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 88780
    },
    {
      "epoch": 78.64038972542073,
      "grad_norm": 0.21404774487018585,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 88785
    },
    {
      "epoch": 78.64481842338353,
      "grad_norm": 0.23793858289718628,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 88790
    },
    {
      "epoch": 78.64924712134632,
      "grad_norm": 0.23274613916873932,
      "learning_rate": 1e-05,
      "loss": 1.0216,
      "step": 88795
    },
    {
      "epoch": 78.65367581930913,
      "grad_norm": 0.24821344017982483,
      "learning_rate": 1e-05,
      "loss": 0.8647,
      "step": 88800
    },
    {
      "epoch": 78.65810451727192,
      "grad_norm": 0.2392818033695221,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 88805
    },
    {
      "epoch": 78.66253321523472,
      "grad_norm": 0.290387898683548,
      "learning_rate": 1e-05,
      "loss": 0.875,
      "step": 88810
    },
    {
      "epoch": 78.66696191319753,
      "grad_norm": 0.24052433669567108,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 88815
    },
    {
      "epoch": 78.67139061116032,
      "grad_norm": 0.24712617695331573,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 88820
    },
    {
      "epoch": 78.67581930912311,
      "grad_norm": 0.24385656416416168,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 88825
    },
    {
      "epoch": 78.68024800708592,
      "grad_norm": 0.22187002003192902,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 88830
    },
    {
      "epoch": 78.68467670504872,
      "grad_norm": 0.27289679646492004,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 88835
    },
    {
      "epoch": 78.68910540301151,
      "grad_norm": 0.2206193059682846,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 88840
    },
    {
      "epoch": 78.69353410097432,
      "grad_norm": 0.209039106965065,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 88845
    },
    {
      "epoch": 78.69796279893711,
      "grad_norm": 0.24443653225898743,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 88850
    },
    {
      "epoch": 78.7023914968999,
      "grad_norm": 0.23161883652210236,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 88855
    },
    {
      "epoch": 78.70682019486271,
      "grad_norm": 0.21250134706497192,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 88860
    },
    {
      "epoch": 78.71124889282551,
      "grad_norm": 0.2488052397966385,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 88865
    },
    {
      "epoch": 78.7156775907883,
      "grad_norm": 0.2144518792629242,
      "learning_rate": 1e-05,
      "loss": 0.8974,
      "step": 88870
    },
    {
      "epoch": 78.72010628875111,
      "grad_norm": 0.23893864452838898,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 88875
    },
    {
      "epoch": 78.7245349867139,
      "grad_norm": 0.24880045652389526,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 88880
    },
    {
      "epoch": 78.7289636846767,
      "grad_norm": 0.25378575921058655,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 88885
    },
    {
      "epoch": 78.73339238263951,
      "grad_norm": 0.21793386340141296,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 88890
    },
    {
      "epoch": 78.7378210806023,
      "grad_norm": 0.24666571617126465,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 88895
    },
    {
      "epoch": 78.7422497785651,
      "grad_norm": 0.21823570132255554,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 88900
    },
    {
      "epoch": 78.7466784765279,
      "grad_norm": 0.2418294996023178,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 88905
    },
    {
      "epoch": 78.7511071744907,
      "grad_norm": 0.20839330554008484,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 88910
    },
    {
      "epoch": 78.75553587245349,
      "grad_norm": 0.2121295928955078,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 88915
    },
    {
      "epoch": 78.7599645704163,
      "grad_norm": 0.2436417043209076,
      "learning_rate": 1e-05,
      "loss": 1.0352,
      "step": 88920
    },
    {
      "epoch": 78.7643932683791,
      "grad_norm": 0.2525129020214081,
      "learning_rate": 1e-05,
      "loss": 1.0188,
      "step": 88925
    },
    {
      "epoch": 78.76882196634189,
      "grad_norm": 0.2668607831001282,
      "learning_rate": 1e-05,
      "loss": 0.9202,
      "step": 88930
    },
    {
      "epoch": 78.7732506643047,
      "grad_norm": 0.2485848218202591,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 88935
    },
    {
      "epoch": 78.77767936226749,
      "grad_norm": 0.3061036467552185,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 88940
    },
    {
      "epoch": 78.78210806023029,
      "grad_norm": 0.23484215140342712,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 88945
    },
    {
      "epoch": 78.7865367581931,
      "grad_norm": 0.22473669052124023,
      "learning_rate": 1e-05,
      "loss": 0.9184,
      "step": 88950
    },
    {
      "epoch": 78.79096545615589,
      "grad_norm": 0.2281215339899063,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 88955
    },
    {
      "epoch": 78.79539415411868,
      "grad_norm": 0.2707948684692383,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 88960
    },
    {
      "epoch": 78.79982285208149,
      "grad_norm": 0.22884972393512726,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 88965
    },
    {
      "epoch": 78.80425155004428,
      "grad_norm": 0.238788440823555,
      "learning_rate": 1e-05,
      "loss": 0.9935,
      "step": 88970
    },
    {
      "epoch": 78.8086802480071,
      "grad_norm": 0.320576936006546,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 88975
    },
    {
      "epoch": 78.81310894596989,
      "grad_norm": 0.2474735826253891,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 88980
    },
    {
      "epoch": 78.81753764393268,
      "grad_norm": 0.2555488049983978,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 88985
    },
    {
      "epoch": 78.82196634189549,
      "grad_norm": 0.2540493309497833,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 88990
    },
    {
      "epoch": 78.82639503985828,
      "grad_norm": 0.2371339201927185,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 88995
    },
    {
      "epoch": 78.83082373782108,
      "grad_norm": 0.23167012631893158,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 89000
    },
    {
      "epoch": 78.83525243578389,
      "grad_norm": 0.23635420203208923,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 89005
    },
    {
      "epoch": 78.83968113374668,
      "grad_norm": 0.2356945425271988,
      "learning_rate": 1e-05,
      "loss": 0.9286,
      "step": 89010
    },
    {
      "epoch": 78.84410983170947,
      "grad_norm": 0.22508975863456726,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 89015
    },
    {
      "epoch": 78.84853852967228,
      "grad_norm": 0.2304401993751526,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 89020
    },
    {
      "epoch": 78.85296722763508,
      "grad_norm": 0.2298077642917633,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 89025
    },
    {
      "epoch": 78.85739592559787,
      "grad_norm": 0.32508131861686707,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 89030
    },
    {
      "epoch": 78.86182462356068,
      "grad_norm": 0.22980186343193054,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 89035
    },
    {
      "epoch": 78.86625332152347,
      "grad_norm": 0.23897279798984528,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 89040
    },
    {
      "epoch": 78.87068201948627,
      "grad_norm": 0.2118704915046692,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 89045
    },
    {
      "epoch": 78.87511071744908,
      "grad_norm": 0.21781527996063232,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 89050
    },
    {
      "epoch": 78.87953941541187,
      "grad_norm": 0.22839125990867615,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 89055
    },
    {
      "epoch": 78.88396811337466,
      "grad_norm": 0.26798373460769653,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 89060
    },
    {
      "epoch": 78.88839681133747,
      "grad_norm": 0.2478044480085373,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 89065
    },
    {
      "epoch": 78.89282550930027,
      "grad_norm": 0.23872819542884827,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 89070
    },
    {
      "epoch": 78.89725420726306,
      "grad_norm": 0.21047410368919373,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 89075
    },
    {
      "epoch": 78.90168290522587,
      "grad_norm": 0.26323169469833374,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 89080
    },
    {
      "epoch": 78.90611160318866,
      "grad_norm": 0.2525649070739746,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 89085
    },
    {
      "epoch": 78.91054030115146,
      "grad_norm": 0.23054446280002594,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 89090
    },
    {
      "epoch": 78.91496899911427,
      "grad_norm": 0.2454458326101303,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 89095
    },
    {
      "epoch": 78.91939769707706,
      "grad_norm": 0.21247033774852753,
      "learning_rate": 1e-05,
      "loss": 0.9143,
      "step": 89100
    },
    {
      "epoch": 78.92382639503985,
      "grad_norm": 0.24538767337799072,
      "learning_rate": 1e-05,
      "loss": 0.8958,
      "step": 89105
    },
    {
      "epoch": 78.92825509300266,
      "grad_norm": 0.20474044978618622,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 89110
    },
    {
      "epoch": 78.93268379096546,
      "grad_norm": 0.25186586380004883,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 89115
    },
    {
      "epoch": 78.93711248892825,
      "grad_norm": 0.19835858047008514,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 89120
    },
    {
      "epoch": 78.94154118689106,
      "grad_norm": 0.22860778868198395,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 89125
    },
    {
      "epoch": 78.94596988485385,
      "grad_norm": 0.26187822222709656,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 89130
    },
    {
      "epoch": 78.95039858281665,
      "grad_norm": 0.2351790964603424,
      "learning_rate": 1e-05,
      "loss": 0.9237,
      "step": 89135
    },
    {
      "epoch": 78.95482728077945,
      "grad_norm": 0.27152693271636963,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 89140
    },
    {
      "epoch": 78.95925597874225,
      "grad_norm": 0.24576568603515625,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 89145
    },
    {
      "epoch": 78.96368467670504,
      "grad_norm": 0.21399107575416565,
      "learning_rate": 1e-05,
      "loss": 0.9155,
      "step": 89150
    },
    {
      "epoch": 78.96811337466785,
      "grad_norm": 0.24784275889396667,
      "learning_rate": 1e-05,
      "loss": 0.9176,
      "step": 89155
    },
    {
      "epoch": 78.97254207263065,
      "grad_norm": 0.26372650265693665,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 89160
    },
    {
      "epoch": 78.97697077059344,
      "grad_norm": 0.2719539403915405,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 89165
    },
    {
      "epoch": 78.98139946855625,
      "grad_norm": 0.247012197971344,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 89170
    },
    {
      "epoch": 78.98582816651904,
      "grad_norm": 0.24623580276966095,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 89175
    },
    {
      "epoch": 78.99025686448184,
      "grad_norm": 0.27150431275367737,
      "learning_rate": 1e-05,
      "loss": 0.9408,
      "step": 89180
    },
    {
      "epoch": 78.99468556244464,
      "grad_norm": 0.23377108573913574,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 89185
    },
    {
      "epoch": 78.99911426040744,
      "grad_norm": 0.22719402611255646,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 89190
    },
    {
      "epoch": 79.00354295837023,
      "grad_norm": 0.2191123515367508,
      "learning_rate": 1e-05,
      "loss": 1.0303,
      "step": 89195
    },
    {
      "epoch": 79.00797165633304,
      "grad_norm": 0.19615575671195984,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 89200
    },
    {
      "epoch": 79.01240035429583,
      "grad_norm": 0.23900577425956726,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 89205
    },
    {
      "epoch": 79.01682905225863,
      "grad_norm": 0.24274350702762604,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 89210
    },
    {
      "epoch": 79.02125775022144,
      "grad_norm": 0.23424094915390015,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 89215
    },
    {
      "epoch": 79.02568644818423,
      "grad_norm": 0.29628610610961914,
      "learning_rate": 1e-05,
      "loss": 0.8842,
      "step": 89220
    },
    {
      "epoch": 79.03011514614704,
      "grad_norm": 0.26593416929244995,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 89225
    },
    {
      "epoch": 79.03454384410983,
      "grad_norm": 0.2791079580783844,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 89230
    },
    {
      "epoch": 79.03897254207263,
      "grad_norm": 0.2200455665588379,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 89235
    },
    {
      "epoch": 79.04340124003544,
      "grad_norm": 0.25736767053604126,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 89240
    },
    {
      "epoch": 79.04782993799823,
      "grad_norm": 0.24660871922969818,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 89245
    },
    {
      "epoch": 79.05225863596102,
      "grad_norm": 0.23847870528697968,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 89250
    },
    {
      "epoch": 79.05668733392383,
      "grad_norm": 0.23797206580638885,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 89255
    },
    {
      "epoch": 79.06111603188663,
      "grad_norm": 0.24937275052070618,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 89260
    },
    {
      "epoch": 79.06554472984942,
      "grad_norm": 0.24184469878673553,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 89265
    },
    {
      "epoch": 79.06997342781223,
      "grad_norm": 0.27600204944610596,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 89270
    },
    {
      "epoch": 79.07440212577502,
      "grad_norm": 0.24106377363204956,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 89275
    },
    {
      "epoch": 79.07883082373782,
      "grad_norm": 0.27480489015579224,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 89280
    },
    {
      "epoch": 79.08325952170063,
      "grad_norm": 0.22212105989456177,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 89285
    },
    {
      "epoch": 79.08768821966342,
      "grad_norm": 0.26639440655708313,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 89290
    },
    {
      "epoch": 79.09211691762621,
      "grad_norm": 0.2756419777870178,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 89295
    },
    {
      "epoch": 79.09654561558902,
      "grad_norm": 0.2521350085735321,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 89300
    },
    {
      "epoch": 79.10097431355182,
      "grad_norm": 0.25528621673583984,
      "learning_rate": 1e-05,
      "loss": 0.9108,
      "step": 89305
    },
    {
      "epoch": 79.10540301151461,
      "grad_norm": 0.25457897782325745,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 89310
    },
    {
      "epoch": 79.10983170947742,
      "grad_norm": 0.24718253314495087,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 89315
    },
    {
      "epoch": 79.11426040744021,
      "grad_norm": 0.22772423923015594,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 89320
    },
    {
      "epoch": 79.118689105403,
      "grad_norm": 0.19612038135528564,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 89325
    },
    {
      "epoch": 79.12311780336582,
      "grad_norm": 0.2685089409351349,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 89330
    },
    {
      "epoch": 79.12754650132861,
      "grad_norm": 0.2330874800682068,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 89335
    },
    {
      "epoch": 79.1319751992914,
      "grad_norm": 0.2725229859352112,
      "learning_rate": 1e-05,
      "loss": 0.9184,
      "step": 89340
    },
    {
      "epoch": 79.13640389725421,
      "grad_norm": 0.24011482298374176,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 89345
    },
    {
      "epoch": 79.140832595217,
      "grad_norm": 0.24903951585292816,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 89350
    },
    {
      "epoch": 79.1452612931798,
      "grad_norm": 0.24760718643665314,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 89355
    },
    {
      "epoch": 79.14968999114261,
      "grad_norm": 0.27038514614105225,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 89360
    },
    {
      "epoch": 79.1541186891054,
      "grad_norm": 0.19365015625953674,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 89365
    },
    {
      "epoch": 79.1585473870682,
      "grad_norm": 0.24248088896274567,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 89370
    },
    {
      "epoch": 79.162976085031,
      "grad_norm": 0.272960901260376,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 89375
    },
    {
      "epoch": 79.1674047829938,
      "grad_norm": 0.22384007275104523,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 89380
    },
    {
      "epoch": 79.1718334809566,
      "grad_norm": 0.19956332445144653,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 89385
    },
    {
      "epoch": 79.1762621789194,
      "grad_norm": 0.2596125900745392,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 89390
    },
    {
      "epoch": 79.1806908768822,
      "grad_norm": 0.2166665643453598,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 89395
    },
    {
      "epoch": 79.18511957484499,
      "grad_norm": 0.22437381744384766,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 89400
    },
    {
      "epoch": 79.1895482728078,
      "grad_norm": 0.25173407793045044,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 89405
    },
    {
      "epoch": 79.19397697077059,
      "grad_norm": 0.21334417164325714,
      "learning_rate": 1e-05,
      "loss": 0.8795,
      "step": 89410
    },
    {
      "epoch": 79.19840566873339,
      "grad_norm": 0.2467496395111084,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 89415
    },
    {
      "epoch": 79.2028343666962,
      "grad_norm": 0.23701106011867523,
      "learning_rate": 1e-05,
      "loss": 1.0608,
      "step": 89420
    },
    {
      "epoch": 79.20726306465899,
      "grad_norm": 0.2276698350906372,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 89425
    },
    {
      "epoch": 79.21169176262178,
      "grad_norm": 0.2324695736169815,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 89430
    },
    {
      "epoch": 79.21612046058459,
      "grad_norm": 0.27067631483078003,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 89435
    },
    {
      "epoch": 79.22054915854739,
      "grad_norm": 0.25942954421043396,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 89440
    },
    {
      "epoch": 79.22497785651018,
      "grad_norm": 0.2337646782398224,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 89445
    },
    {
      "epoch": 79.22940655447299,
      "grad_norm": 0.2337169498205185,
      "learning_rate": 1e-05,
      "loss": 0.9335,
      "step": 89450
    },
    {
      "epoch": 79.23383525243578,
      "grad_norm": 0.23848369717597961,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 89455
    },
    {
      "epoch": 79.23826395039858,
      "grad_norm": 0.23774364590644836,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 89460
    },
    {
      "epoch": 79.24269264836138,
      "grad_norm": 0.21404683589935303,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 89465
    },
    {
      "epoch": 79.24712134632418,
      "grad_norm": 0.2269837111234665,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 89470
    },
    {
      "epoch": 79.25155004428699,
      "grad_norm": 0.2135545313358307,
      "learning_rate": 1e-05,
      "loss": 0.9245,
      "step": 89475
    },
    {
      "epoch": 79.25597874224978,
      "grad_norm": 0.23130248486995697,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 89480
    },
    {
      "epoch": 79.26040744021257,
      "grad_norm": 0.29714348912239075,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 89485
    },
    {
      "epoch": 79.26483613817538,
      "grad_norm": 0.2454877495765686,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 89490
    },
    {
      "epoch": 79.26926483613818,
      "grad_norm": 0.21029037237167358,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 89495
    },
    {
      "epoch": 79.27369353410097,
      "grad_norm": 0.265084445476532,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 89500
    },
    {
      "epoch": 79.27812223206378,
      "grad_norm": 0.282025009393692,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 89505
    },
    {
      "epoch": 79.28255093002657,
      "grad_norm": 0.24618028104305267,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 89510
    },
    {
      "epoch": 79.28697962798937,
      "grad_norm": 0.21301627159118652,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 89515
    },
    {
      "epoch": 79.29140832595218,
      "grad_norm": 0.22592532634735107,
      "learning_rate": 1e-05,
      "loss": 0.9138,
      "step": 89520
    },
    {
      "epoch": 79.29583702391497,
      "grad_norm": 0.21918508410453796,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 89525
    },
    {
      "epoch": 79.30026572187776,
      "grad_norm": 0.24886052310466766,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 89530
    },
    {
      "epoch": 79.30469441984057,
      "grad_norm": 0.26327425241470337,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 89535
    },
    {
      "epoch": 79.30912311780337,
      "grad_norm": 0.23520930111408234,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 89540
    },
    {
      "epoch": 79.31355181576616,
      "grad_norm": 0.2018265277147293,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 89545
    },
    {
      "epoch": 79.31798051372897,
      "grad_norm": 0.2244083285331726,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 89550
    },
    {
      "epoch": 79.32240921169176,
      "grad_norm": 0.2708916664123535,
      "learning_rate": 1e-05,
      "loss": 0.9346,
      "step": 89555
    },
    {
      "epoch": 79.32683790965456,
      "grad_norm": 0.23382501304149628,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 89560
    },
    {
      "epoch": 79.33126660761737,
      "grad_norm": 0.22604914009571075,
      "learning_rate": 1e-05,
      "loss": 0.933,
      "step": 89565
    },
    {
      "epoch": 79.33569530558016,
      "grad_norm": 0.2737271785736084,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 89570
    },
    {
      "epoch": 79.34012400354295,
      "grad_norm": 0.21776720881462097,
      "learning_rate": 1e-05,
      "loss": 0.9279,
      "step": 89575
    },
    {
      "epoch": 79.34455270150576,
      "grad_norm": 0.25181880593299866,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 89580
    },
    {
      "epoch": 79.34898139946856,
      "grad_norm": 0.24273978173732758,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 89585
    },
    {
      "epoch": 79.35341009743135,
      "grad_norm": 0.26046350598335266,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 89590
    },
    {
      "epoch": 79.35783879539416,
      "grad_norm": 0.2510797083377838,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 89595
    },
    {
      "epoch": 79.36226749335695,
      "grad_norm": 0.22702626883983612,
      "learning_rate": 1e-05,
      "loss": 0.942,
      "step": 89600
    },
    {
      "epoch": 79.36669619131975,
      "grad_norm": 0.28290918469429016,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 89605
    },
    {
      "epoch": 79.37112488928256,
      "grad_norm": 0.23685164749622345,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 89610
    },
    {
      "epoch": 79.37555358724535,
      "grad_norm": 0.25025680661201477,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 89615
    },
    {
      "epoch": 79.37998228520814,
      "grad_norm": 0.2567918598651886,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 89620
    },
    {
      "epoch": 79.38441098317095,
      "grad_norm": 0.2924160361289978,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 89625
    },
    {
      "epoch": 79.38883968113375,
      "grad_norm": 0.30412012338638306,
      "learning_rate": 1e-05,
      "loss": 0.8983,
      "step": 89630
    },
    {
      "epoch": 79.39326837909654,
      "grad_norm": 0.20977526903152466,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 89635
    },
    {
      "epoch": 79.39769707705935,
      "grad_norm": 0.2971060872077942,
      "learning_rate": 1e-05,
      "loss": 0.9303,
      "step": 89640
    },
    {
      "epoch": 79.40212577502214,
      "grad_norm": 0.22963666915893555,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 89645
    },
    {
      "epoch": 79.40655447298494,
      "grad_norm": 0.2330242544412613,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 89650
    },
    {
      "epoch": 79.41098317094774,
      "grad_norm": 0.21970553696155548,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 89655
    },
    {
      "epoch": 79.41541186891054,
      "grad_norm": 0.22707703709602356,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 89660
    },
    {
      "epoch": 79.41984056687333,
      "grad_norm": 0.21161366999149323,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 89665
    },
    {
      "epoch": 79.42426926483614,
      "grad_norm": 0.2178703248500824,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 89670
    },
    {
      "epoch": 79.42869796279894,
      "grad_norm": 0.23282155394554138,
      "learning_rate": 1e-05,
      "loss": 0.9196,
      "step": 89675
    },
    {
      "epoch": 79.43312666076173,
      "grad_norm": 0.2014344334602356,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 89680
    },
    {
      "epoch": 79.43755535872454,
      "grad_norm": 0.2859559953212738,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 89685
    },
    {
      "epoch": 79.44198405668733,
      "grad_norm": 0.23912128806114197,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 89690
    },
    {
      "epoch": 79.44641275465013,
      "grad_norm": 0.23863229155540466,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 89695
    },
    {
      "epoch": 79.45084145261293,
      "grad_norm": 0.23816151916980743,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 89700
    },
    {
      "epoch": 79.45527015057573,
      "grad_norm": 0.2225727140903473,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 89705
    },
    {
      "epoch": 79.45969884853854,
      "grad_norm": 0.24674266576766968,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 89710
    },
    {
      "epoch": 79.46412754650133,
      "grad_norm": 0.23209114372730255,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 89715
    },
    {
      "epoch": 79.46855624446412,
      "grad_norm": 0.24304057657718658,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 89720
    },
    {
      "epoch": 79.47298494242693,
      "grad_norm": 0.2184872329235077,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 89725
    },
    {
      "epoch": 79.47741364038973,
      "grad_norm": 0.23874977231025696,
      "learning_rate": 1e-05,
      "loss": 0.9177,
      "step": 89730
    },
    {
      "epoch": 79.48184233835252,
      "grad_norm": 0.24227285385131836,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 89735
    },
    {
      "epoch": 79.48627103631533,
      "grad_norm": 0.23455865681171417,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 89740
    },
    {
      "epoch": 79.49069973427812,
      "grad_norm": 0.197706401348114,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 89745
    },
    {
      "epoch": 79.49512843224092,
      "grad_norm": 0.2293943166732788,
      "learning_rate": 1e-05,
      "loss": 0.9367,
      "step": 89750
    },
    {
      "epoch": 79.49955713020373,
      "grad_norm": 0.24075208604335785,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 89755
    },
    {
      "epoch": 79.50398582816652,
      "grad_norm": 0.28364071249961853,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 89760
    },
    {
      "epoch": 79.50841452612931,
      "grad_norm": 0.24988830089569092,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 89765
    },
    {
      "epoch": 79.51284322409212,
      "grad_norm": 0.23506776988506317,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 89770
    },
    {
      "epoch": 79.51727192205492,
      "grad_norm": 0.2698654234409332,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 89775
    },
    {
      "epoch": 79.52170062001771,
      "grad_norm": 0.24132831394672394,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 89780
    },
    {
      "epoch": 79.52612931798052,
      "grad_norm": 0.25501906871795654,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 89785
    },
    {
      "epoch": 79.53055801594331,
      "grad_norm": 0.250794917345047,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 89790
    },
    {
      "epoch": 79.53498671390611,
      "grad_norm": 0.20638127624988556,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 89795
    },
    {
      "epoch": 79.53941541186892,
      "grad_norm": 0.20616641640663147,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 89800
    },
    {
      "epoch": 79.54384410983171,
      "grad_norm": 0.25351670384407043,
      "learning_rate": 1e-05,
      "loss": 0.922,
      "step": 89805
    },
    {
      "epoch": 79.5482728077945,
      "grad_norm": 0.22439146041870117,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 89810
    },
    {
      "epoch": 79.55270150575731,
      "grad_norm": 0.22937269508838654,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 89815
    },
    {
      "epoch": 79.5571302037201,
      "grad_norm": 0.22090381383895874,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 89820
    },
    {
      "epoch": 79.5615589016829,
      "grad_norm": 0.25759562849998474,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 89825
    },
    {
      "epoch": 79.56598759964571,
      "grad_norm": 0.2240026593208313,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 89830
    },
    {
      "epoch": 79.5704162976085,
      "grad_norm": 0.27752652764320374,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 89835
    },
    {
      "epoch": 79.5748449955713,
      "grad_norm": 0.2386729121208191,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 89840
    },
    {
      "epoch": 79.5792736935341,
      "grad_norm": 0.22522301971912384,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 89845
    },
    {
      "epoch": 79.5837023914969,
      "grad_norm": 0.23391591012477875,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 89850
    },
    {
      "epoch": 79.5881310894597,
      "grad_norm": 0.22013649344444275,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 89855
    },
    {
      "epoch": 79.5925597874225,
      "grad_norm": 0.2454390972852707,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 89860
    },
    {
      "epoch": 79.5969884853853,
      "grad_norm": 0.21828554570674896,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 89865
    },
    {
      "epoch": 79.60141718334809,
      "grad_norm": 0.24717754125595093,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 89870
    },
    {
      "epoch": 79.6058458813109,
      "grad_norm": 0.23066185414791107,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 89875
    },
    {
      "epoch": 79.61027457927369,
      "grad_norm": 0.24032637476921082,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 89880
    },
    {
      "epoch": 79.61470327723649,
      "grad_norm": 0.23051051795482635,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 89885
    },
    {
      "epoch": 79.6191319751993,
      "grad_norm": 0.23454685509204865,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 89890
    },
    {
      "epoch": 79.62356067316209,
      "grad_norm": 0.2410600483417511,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 89895
    },
    {
      "epoch": 79.62798937112488,
      "grad_norm": 0.23189964890480042,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 89900
    },
    {
      "epoch": 79.63241806908769,
      "grad_norm": 0.26159584522247314,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 89905
    },
    {
      "epoch": 79.63684676705049,
      "grad_norm": 0.23677869141101837,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 89910
    },
    {
      "epoch": 79.64127546501328,
      "grad_norm": 0.21713300049304962,
      "learning_rate": 1e-05,
      "loss": 1.0329,
      "step": 89915
    },
    {
      "epoch": 79.64570416297609,
      "grad_norm": 0.20646561682224274,
      "learning_rate": 1e-05,
      "loss": 0.9233,
      "step": 89920
    },
    {
      "epoch": 79.65013286093888,
      "grad_norm": 0.23404309153556824,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 89925
    },
    {
      "epoch": 79.65456155890168,
      "grad_norm": 0.20942646265029907,
      "learning_rate": 1e-05,
      "loss": 0.9286,
      "step": 89930
    },
    {
      "epoch": 79.65899025686448,
      "grad_norm": 0.2503487169742584,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 89935
    },
    {
      "epoch": 79.66341895482728,
      "grad_norm": 0.2490573674440384,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 89940
    },
    {
      "epoch": 79.66784765279007,
      "grad_norm": 0.2281889170408249,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 89945
    },
    {
      "epoch": 79.67227635075288,
      "grad_norm": 0.23662573099136353,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 89950
    },
    {
      "epoch": 79.67670504871568,
      "grad_norm": 0.217342808842659,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 89955
    },
    {
      "epoch": 79.68113374667848,
      "grad_norm": 0.2023911029100418,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 89960
    },
    {
      "epoch": 79.68556244464128,
      "grad_norm": 0.23905080556869507,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 89965
    },
    {
      "epoch": 79.68999114260407,
      "grad_norm": 0.2180573046207428,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 89970
    },
    {
      "epoch": 79.69441984056688,
      "grad_norm": 0.21884189546108246,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 89975
    },
    {
      "epoch": 79.69884853852967,
      "grad_norm": 0.23229235410690308,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 89980
    },
    {
      "epoch": 79.70327723649247,
      "grad_norm": 0.25508442521095276,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 89985
    },
    {
      "epoch": 79.70770593445528,
      "grad_norm": 0.25490033626556396,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 89990
    },
    {
      "epoch": 79.71213463241807,
      "grad_norm": 0.25662222504615784,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 89995
    },
    {
      "epoch": 79.71656333038086,
      "grad_norm": 0.20283883810043335,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 90000
    },
    {
      "epoch": 79.72099202834367,
      "grad_norm": 0.2433825433254242,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 90005
    },
    {
      "epoch": 79.72542072630647,
      "grad_norm": 0.23120273649692535,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 90010
    },
    {
      "epoch": 79.72984942426926,
      "grad_norm": 0.20766118168830872,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 90015
    },
    {
      "epoch": 79.73427812223207,
      "grad_norm": 0.21366922557353973,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 90020
    },
    {
      "epoch": 79.73870682019486,
      "grad_norm": 0.2469257116317749,
      "learning_rate": 1e-05,
      "loss": 1.0226,
      "step": 90025
    },
    {
      "epoch": 79.74313551815766,
      "grad_norm": 0.23615744709968567,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 90030
    },
    {
      "epoch": 79.74756421612047,
      "grad_norm": 0.23682944476604462,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 90035
    },
    {
      "epoch": 79.75199291408326,
      "grad_norm": 0.27958759665489197,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 90040
    },
    {
      "epoch": 79.75642161204605,
      "grad_norm": 0.23270191252231598,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 90045
    },
    {
      "epoch": 79.76085031000886,
      "grad_norm": 0.237318754196167,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 90050
    },
    {
      "epoch": 79.76527900797166,
      "grad_norm": 0.25557467341423035,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 90055
    },
    {
      "epoch": 79.76970770593445,
      "grad_norm": 0.30670565366744995,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 90060
    },
    {
      "epoch": 79.77413640389726,
      "grad_norm": 0.2577807903289795,
      "learning_rate": 1e-05,
      "loss": 0.891,
      "step": 90065
    },
    {
      "epoch": 79.77856510186005,
      "grad_norm": 0.2242809236049652,
      "learning_rate": 1e-05,
      "loss": 0.9247,
      "step": 90070
    },
    {
      "epoch": 79.78299379982285,
      "grad_norm": 0.21872539818286896,
      "learning_rate": 1e-05,
      "loss": 0.9392,
      "step": 90075
    },
    {
      "epoch": 79.78742249778566,
      "grad_norm": 0.21904049813747406,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 90080
    },
    {
      "epoch": 79.79185119574845,
      "grad_norm": 0.28020399808883667,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 90085
    },
    {
      "epoch": 79.79627989371124,
      "grad_norm": 0.228740856051445,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 90090
    },
    {
      "epoch": 79.80070859167405,
      "grad_norm": 0.23158478736877441,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 90095
    },
    {
      "epoch": 79.80513728963685,
      "grad_norm": 0.24134615063667297,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 90100
    },
    {
      "epoch": 79.80956598759964,
      "grad_norm": 0.2543979287147522,
      "learning_rate": 1e-05,
      "loss": 0.9247,
      "step": 90105
    },
    {
      "epoch": 79.81399468556245,
      "grad_norm": 0.23991595208644867,
      "learning_rate": 1e-05,
      "loss": 0.9196,
      "step": 90110
    },
    {
      "epoch": 79.81842338352524,
      "grad_norm": 0.2548162341117859,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 90115
    },
    {
      "epoch": 79.82285208148804,
      "grad_norm": 0.2417423576116562,
      "learning_rate": 1e-05,
      "loss": 0.9245,
      "step": 90120
    },
    {
      "epoch": 79.82728077945085,
      "grad_norm": 0.2563302218914032,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 90125
    },
    {
      "epoch": 79.83170947741364,
      "grad_norm": 0.2482224851846695,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 90130
    },
    {
      "epoch": 79.83613817537643,
      "grad_norm": 0.2457963079214096,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 90135
    },
    {
      "epoch": 79.84056687333924,
      "grad_norm": 0.21073241531848907,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 90140
    },
    {
      "epoch": 79.84499557130204,
      "grad_norm": 0.29018402099609375,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 90145
    },
    {
      "epoch": 79.84942426926483,
      "grad_norm": 0.24830599129199982,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 90150
    },
    {
      "epoch": 79.85385296722764,
      "grad_norm": 0.2795458137989044,
      "learning_rate": 1e-05,
      "loss": 0.9176,
      "step": 90155
    },
    {
      "epoch": 79.85828166519043,
      "grad_norm": 0.23765401542186737,
      "learning_rate": 1e-05,
      "loss": 0.9392,
      "step": 90160
    },
    {
      "epoch": 79.86271036315323,
      "grad_norm": 0.22332780063152313,
      "learning_rate": 1e-05,
      "loss": 0.9001,
      "step": 90165
    },
    {
      "epoch": 79.86713906111603,
      "grad_norm": 0.2536269426345825,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 90170
    },
    {
      "epoch": 79.87156775907883,
      "grad_norm": 0.27211257815361023,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 90175
    },
    {
      "epoch": 79.87599645704162,
      "grad_norm": 0.2315283715724945,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 90180
    },
    {
      "epoch": 79.88042515500443,
      "grad_norm": 0.2842007279396057,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 90185
    },
    {
      "epoch": 79.88485385296723,
      "grad_norm": 0.27521586418151855,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 90190
    },
    {
      "epoch": 79.88928255093003,
      "grad_norm": 0.25162190198898315,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 90195
    },
    {
      "epoch": 79.89371124889283,
      "grad_norm": 0.2608662545681,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 90200
    },
    {
      "epoch": 79.89813994685562,
      "grad_norm": 0.2480461448431015,
      "learning_rate": 1e-05,
      "loss": 0.9236,
      "step": 90205
    },
    {
      "epoch": 79.90256864481843,
      "grad_norm": 0.1931961476802826,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 90210
    },
    {
      "epoch": 79.90699734278122,
      "grad_norm": 0.22209890186786652,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 90215
    },
    {
      "epoch": 79.91142604074402,
      "grad_norm": 0.2419925332069397,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 90220
    },
    {
      "epoch": 79.91585473870683,
      "grad_norm": 0.2507255971431732,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 90225
    },
    {
      "epoch": 79.92028343666962,
      "grad_norm": 0.2477208822965622,
      "learning_rate": 1e-05,
      "loss": 1.015,
      "step": 90230
    },
    {
      "epoch": 79.92471213463241,
      "grad_norm": 0.24340468645095825,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 90235
    },
    {
      "epoch": 79.92914083259522,
      "grad_norm": 0.21622225642204285,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 90240
    },
    {
      "epoch": 79.93356953055802,
      "grad_norm": 0.25533327460289,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 90245
    },
    {
      "epoch": 79.93799822852081,
      "grad_norm": 0.22512692213058472,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 90250
    },
    {
      "epoch": 79.94242692648362,
      "grad_norm": 0.19293341040611267,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 90255
    },
    {
      "epoch": 79.94685562444641,
      "grad_norm": 0.24189257621765137,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 90260
    },
    {
      "epoch": 79.95128432240921,
      "grad_norm": 0.2521813213825226,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 90265
    },
    {
      "epoch": 79.95571302037202,
      "grad_norm": 0.2447579801082611,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 90270
    },
    {
      "epoch": 79.96014171833481,
      "grad_norm": 0.23347577452659607,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 90275
    },
    {
      "epoch": 79.9645704162976,
      "grad_norm": 0.22960013151168823,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 90280
    },
    {
      "epoch": 79.96899911426041,
      "grad_norm": 0.30896326899528503,
      "learning_rate": 1e-05,
      "loss": 0.9058,
      "step": 90285
    },
    {
      "epoch": 79.9734278122232,
      "grad_norm": 0.2722979485988617,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 90290
    },
    {
      "epoch": 79.977856510186,
      "grad_norm": 0.2933483421802521,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 90295
    },
    {
      "epoch": 79.98228520814881,
      "grad_norm": 0.2473868429660797,
      "learning_rate": 1e-05,
      "loss": 1.0145,
      "step": 90300
    },
    {
      "epoch": 79.9867139061116,
      "grad_norm": 0.22241902351379395,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 90305
    },
    {
      "epoch": 79.9911426040744,
      "grad_norm": 0.21211975812911987,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 90310
    },
    {
      "epoch": 79.9955713020372,
      "grad_norm": 0.21840862929821014,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 90315
    },
    {
      "epoch": 80.0,
      "grad_norm": 0.2095823436975479,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 90320
    },
    {
      "epoch": 80.0044286979628,
      "grad_norm": 0.26246175169944763,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 90325
    },
    {
      "epoch": 80.0088573959256,
      "grad_norm": 0.2380438596010208,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 90330
    },
    {
      "epoch": 80.0132860938884,
      "grad_norm": 0.21253980696201324,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 90335
    },
    {
      "epoch": 80.01771479185119,
      "grad_norm": 0.24193133413791656,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 90340
    },
    {
      "epoch": 80.022143489814,
      "grad_norm": 0.2500888407230377,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 90345
    },
    {
      "epoch": 80.0265721877768,
      "grad_norm": 0.24459004402160645,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 90350
    },
    {
      "epoch": 80.03100088573959,
      "grad_norm": 0.22864751517772675,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 90355
    },
    {
      "epoch": 80.0354295837024,
      "grad_norm": 0.26301053166389465,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 90360
    },
    {
      "epoch": 80.03985828166519,
      "grad_norm": 0.21500445902347565,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 90365
    },
    {
      "epoch": 80.04428697962798,
      "grad_norm": 0.223270446062088,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 90370
    },
    {
      "epoch": 80.04871567759079,
      "grad_norm": 0.24247729778289795,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 90375
    },
    {
      "epoch": 80.05314437555359,
      "grad_norm": 0.20041173696517944,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 90380
    },
    {
      "epoch": 80.05757307351638,
      "grad_norm": 0.23628394305706024,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 90385
    },
    {
      "epoch": 80.06200177147919,
      "grad_norm": 0.2290562391281128,
      "learning_rate": 1e-05,
      "loss": 1.0344,
      "step": 90390
    },
    {
      "epoch": 80.06643046944198,
      "grad_norm": 0.2636483311653137,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 90395
    },
    {
      "epoch": 80.07085916740478,
      "grad_norm": 0.23545511066913605,
      "learning_rate": 1e-05,
      "loss": 0.9231,
      "step": 90400
    },
    {
      "epoch": 80.07528786536759,
      "grad_norm": 0.23498377203941345,
      "learning_rate": 1e-05,
      "loss": 0.9105,
      "step": 90405
    },
    {
      "epoch": 80.07971656333038,
      "grad_norm": 0.23322664201259613,
      "learning_rate": 1e-05,
      "loss": 1.0555,
      "step": 90410
    },
    {
      "epoch": 80.08414526129317,
      "grad_norm": 0.2805613875389099,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 90415
    },
    {
      "epoch": 80.08857395925598,
      "grad_norm": 0.25593459606170654,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 90420
    },
    {
      "epoch": 80.09300265721878,
      "grad_norm": 0.28242868185043335,
      "learning_rate": 1e-05,
      "loss": 0.9045,
      "step": 90425
    },
    {
      "epoch": 80.09743135518157,
      "grad_norm": 0.2630789279937744,
      "learning_rate": 1e-05,
      "loss": 0.8962,
      "step": 90430
    },
    {
      "epoch": 80.10186005314438,
      "grad_norm": 0.23096276819705963,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 90435
    },
    {
      "epoch": 80.10628875110717,
      "grad_norm": 0.24531184136867523,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 90440
    },
    {
      "epoch": 80.11071744906998,
      "grad_norm": 0.2260182648897171,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 90445
    },
    {
      "epoch": 80.11514614703277,
      "grad_norm": 0.24155649542808533,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 90450
    },
    {
      "epoch": 80.11957484499557,
      "grad_norm": 0.2517586648464203,
      "learning_rate": 1e-05,
      "loss": 0.8965,
      "step": 90455
    },
    {
      "epoch": 80.12400354295838,
      "grad_norm": 0.22883270680904388,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 90460
    },
    {
      "epoch": 80.12843224092117,
      "grad_norm": 0.26872262358665466,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 90465
    },
    {
      "epoch": 80.13286093888397,
      "grad_norm": 0.2595265805721283,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 90470
    },
    {
      "epoch": 80.13728963684677,
      "grad_norm": 0.25981834530830383,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 90475
    },
    {
      "epoch": 80.14171833480957,
      "grad_norm": 0.276508092880249,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 90480
    },
    {
      "epoch": 80.14614703277236,
      "grad_norm": 0.2871341109275818,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 90485
    },
    {
      "epoch": 80.15057573073517,
      "grad_norm": 0.2640523314476013,
      "learning_rate": 1e-05,
      "loss": 1.0249,
      "step": 90490
    },
    {
      "epoch": 80.15500442869796,
      "grad_norm": 0.25753283500671387,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 90495
    },
    {
      "epoch": 80.15943312666076,
      "grad_norm": 0.23141242563724518,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 90500
    },
    {
      "epoch": 80.16386182462357,
      "grad_norm": 0.2090873420238495,
      "learning_rate": 1e-05,
      "loss": 1.0254,
      "step": 90505
    },
    {
      "epoch": 80.16829052258636,
      "grad_norm": 0.21610578894615173,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 90510
    },
    {
      "epoch": 80.17271922054915,
      "grad_norm": 0.24537496268749237,
      "learning_rate": 1e-05,
      "loss": 0.9364,
      "step": 90515
    },
    {
      "epoch": 80.17714791851196,
      "grad_norm": 0.2676410675048828,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 90520
    },
    {
      "epoch": 80.18157661647476,
      "grad_norm": 0.2525486946105957,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 90525
    },
    {
      "epoch": 80.18600531443755,
      "grad_norm": 0.22568634152412415,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 90530
    },
    {
      "epoch": 80.19043401240036,
      "grad_norm": 0.21275459229946136,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 90535
    },
    {
      "epoch": 80.19486271036315,
      "grad_norm": 0.22479788959026337,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 90540
    },
    {
      "epoch": 80.19929140832595,
      "grad_norm": 0.23335857689380646,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 90545
    },
    {
      "epoch": 80.20372010628876,
      "grad_norm": 0.24292509257793427,
      "learning_rate": 1e-05,
      "loss": 0.9094,
      "step": 90550
    },
    {
      "epoch": 80.20814880425155,
      "grad_norm": 0.26341915130615234,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 90555
    },
    {
      "epoch": 80.21257750221434,
      "grad_norm": 0.22146227955818176,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 90560
    },
    {
      "epoch": 80.21700620017715,
      "grad_norm": 0.2333306521177292,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 90565
    },
    {
      "epoch": 80.22143489813995,
      "grad_norm": 0.25761678814888,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 90570
    },
    {
      "epoch": 80.22586359610274,
      "grad_norm": 0.2671818733215332,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 90575
    },
    {
      "epoch": 80.23029229406555,
      "grad_norm": 0.25359222292900085,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 90580
    },
    {
      "epoch": 80.23472099202834,
      "grad_norm": 0.26131007075309753,
      "learning_rate": 1e-05,
      "loss": 1.0042,
      "step": 90585
    },
    {
      "epoch": 80.23914968999114,
      "grad_norm": 0.22572430968284607,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 90590
    },
    {
      "epoch": 80.24357838795395,
      "grad_norm": 0.2223578542470932,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 90595
    },
    {
      "epoch": 80.24800708591674,
      "grad_norm": 0.2399706095457077,
      "learning_rate": 1e-05,
      "loss": 0.929,
      "step": 90600
    },
    {
      "epoch": 80.25243578387953,
      "grad_norm": 0.2179998755455017,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 90605
    },
    {
      "epoch": 80.25686448184234,
      "grad_norm": 0.222193643450737,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 90610
    },
    {
      "epoch": 80.26129317980514,
      "grad_norm": 0.26568669080734253,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 90615
    },
    {
      "epoch": 80.26572187776793,
      "grad_norm": 0.2616262137889862,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 90620
    },
    {
      "epoch": 80.27015057573074,
      "grad_norm": 0.2509314715862274,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 90625
    },
    {
      "epoch": 80.27457927369353,
      "grad_norm": 0.2570199966430664,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 90630
    },
    {
      "epoch": 80.27900797165633,
      "grad_norm": 0.23955181241035461,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 90635
    },
    {
      "epoch": 80.28343666961914,
      "grad_norm": 0.26595553755760193,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 90640
    },
    {
      "epoch": 80.28786536758193,
      "grad_norm": 0.2420101910829544,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 90645
    },
    {
      "epoch": 80.29229406554472,
      "grad_norm": 0.23126855492591858,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 90650
    },
    {
      "epoch": 80.29672276350753,
      "grad_norm": 0.2859993577003479,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 90655
    },
    {
      "epoch": 80.30115146147033,
      "grad_norm": 0.21770113706588745,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 90660
    },
    {
      "epoch": 80.30558015943312,
      "grad_norm": 0.2505124807357788,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 90665
    },
    {
      "epoch": 80.31000885739593,
      "grad_norm": 0.2358206808567047,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 90670
    },
    {
      "epoch": 80.31443755535872,
      "grad_norm": 0.2855980396270752,
      "learning_rate": 1e-05,
      "loss": 0.9283,
      "step": 90675
    },
    {
      "epoch": 80.31886625332152,
      "grad_norm": 0.31953322887420654,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 90680
    },
    {
      "epoch": 80.32329495128432,
      "grad_norm": 0.24071930348873138,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 90685
    },
    {
      "epoch": 80.32772364924712,
      "grad_norm": 0.2715437710285187,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 90690
    },
    {
      "epoch": 80.33215234720993,
      "grad_norm": 0.22301892936229706,
      "learning_rate": 1e-05,
      "loss": 0.9114,
      "step": 90695
    },
    {
      "epoch": 80.33658104517272,
      "grad_norm": 0.2870946526527405,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 90700
    },
    {
      "epoch": 80.34100974313552,
      "grad_norm": 0.2254539430141449,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 90705
    },
    {
      "epoch": 80.34543844109832,
      "grad_norm": 0.22369804978370667,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 90710
    },
    {
      "epoch": 80.34986713906112,
      "grad_norm": 0.2669138014316559,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 90715
    },
    {
      "epoch": 80.35429583702391,
      "grad_norm": 0.26017987728118896,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 90720
    },
    {
      "epoch": 80.35872453498672,
      "grad_norm": 0.261029452085495,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 90725
    },
    {
      "epoch": 80.36315323294951,
      "grad_norm": 0.2585115134716034,
      "learning_rate": 1e-05,
      "loss": 0.9213,
      "step": 90730
    },
    {
      "epoch": 80.36758193091231,
      "grad_norm": 0.22768008708953857,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 90735
    },
    {
      "epoch": 80.37201062887512,
      "grad_norm": 0.2629387676715851,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 90740
    },
    {
      "epoch": 80.37643932683791,
      "grad_norm": 0.24866102635860443,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 90745
    },
    {
      "epoch": 80.3808680248007,
      "grad_norm": 0.27576547861099243,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 90750
    },
    {
      "epoch": 80.38529672276351,
      "grad_norm": 0.35673296451568604,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 90755
    },
    {
      "epoch": 80.38972542072631,
      "grad_norm": 0.2631607949733734,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 90760
    },
    {
      "epoch": 80.3941541186891,
      "grad_norm": 0.24194203317165375,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 90765
    },
    {
      "epoch": 80.39858281665191,
      "grad_norm": 0.2552587687969208,
      "learning_rate": 1e-05,
      "loss": 0.8925,
      "step": 90770
    },
    {
      "epoch": 80.4030115146147,
      "grad_norm": 0.2479856163263321,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 90775
    },
    {
      "epoch": 80.4074402125775,
      "grad_norm": 0.2398465871810913,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 90780
    },
    {
      "epoch": 80.4118689105403,
      "grad_norm": 0.21170029044151306,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 90785
    },
    {
      "epoch": 80.4162976085031,
      "grad_norm": 0.2215103656053543,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 90790
    },
    {
      "epoch": 80.4207263064659,
      "grad_norm": 0.22934865951538086,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 90795
    },
    {
      "epoch": 80.4251550044287,
      "grad_norm": 0.2242850959300995,
      "learning_rate": 1e-05,
      "loss": 0.8819,
      "step": 90800
    },
    {
      "epoch": 80.4295837023915,
      "grad_norm": 0.21365399658679962,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 90805
    },
    {
      "epoch": 80.43401240035429,
      "grad_norm": 0.20895269513130188,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 90810
    },
    {
      "epoch": 80.4384410983171,
      "grad_norm": 0.22954924404621124,
      "learning_rate": 1e-05,
      "loss": 0.93,
      "step": 90815
    },
    {
      "epoch": 80.4428697962799,
      "grad_norm": 0.2231500893831253,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 90820
    },
    {
      "epoch": 80.44729849424269,
      "grad_norm": 0.24777980148792267,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 90825
    },
    {
      "epoch": 80.4517271922055,
      "grad_norm": 0.27447614073753357,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 90830
    },
    {
      "epoch": 80.45615589016829,
      "grad_norm": 0.23682816326618195,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 90835
    },
    {
      "epoch": 80.46058458813108,
      "grad_norm": 0.2200881987810135,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 90840
    },
    {
      "epoch": 80.46501328609389,
      "grad_norm": 0.26538461446762085,
      "learning_rate": 1e-05,
      "loss": 0.9198,
      "step": 90845
    },
    {
      "epoch": 80.46944198405669,
      "grad_norm": 0.2557761073112488,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 90850
    },
    {
      "epoch": 80.47387068201948,
      "grad_norm": 0.23829787969589233,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 90855
    },
    {
      "epoch": 80.47829937998229,
      "grad_norm": 0.20655357837677002,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 90860
    },
    {
      "epoch": 80.48272807794508,
      "grad_norm": 0.23293107748031616,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 90865
    },
    {
      "epoch": 80.48715677590788,
      "grad_norm": 0.2538190484046936,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 90870
    },
    {
      "epoch": 80.49158547387069,
      "grad_norm": 0.24415796995162964,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 90875
    },
    {
      "epoch": 80.49601417183348,
      "grad_norm": 0.24092614650726318,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 90880
    },
    {
      "epoch": 80.50044286979627,
      "grad_norm": 0.2633575201034546,
      "learning_rate": 1e-05,
      "loss": 0.8947,
      "step": 90885
    },
    {
      "epoch": 80.50487156775908,
      "grad_norm": 0.32758960127830505,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 90890
    },
    {
      "epoch": 80.50930026572188,
      "grad_norm": 0.276058554649353,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 90895
    },
    {
      "epoch": 80.51372896368467,
      "grad_norm": 0.2731495499610901,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 90900
    },
    {
      "epoch": 80.51815766164748,
      "grad_norm": 0.270124226808548,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 90905
    },
    {
      "epoch": 80.52258635961027,
      "grad_norm": 0.23477540910243988,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 90910
    },
    {
      "epoch": 80.52701505757307,
      "grad_norm": 0.2700737714767456,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 90915
    },
    {
      "epoch": 80.53144375553588,
      "grad_norm": 0.2259398251771927,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 90920
    },
    {
      "epoch": 80.53587245349867,
      "grad_norm": 0.2582458257675171,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 90925
    },
    {
      "epoch": 80.54030115146146,
      "grad_norm": 0.31537526845932007,
      "learning_rate": 1e-05,
      "loss": 0.9122,
      "step": 90930
    },
    {
      "epoch": 80.54472984942427,
      "grad_norm": 0.2454194724559784,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 90935
    },
    {
      "epoch": 80.54915854738707,
      "grad_norm": 0.26461490988731384,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 90940
    },
    {
      "epoch": 80.55358724534987,
      "grad_norm": 0.2461141049861908,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 90945
    },
    {
      "epoch": 80.55801594331267,
      "grad_norm": 0.24520158767700195,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 90950
    },
    {
      "epoch": 80.56244464127546,
      "grad_norm": 0.33144229650497437,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 90955
    },
    {
      "epoch": 80.56687333923827,
      "grad_norm": 0.2548491060733795,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 90960
    },
    {
      "epoch": 80.57130203720106,
      "grad_norm": 0.26169726252555847,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 90965
    },
    {
      "epoch": 80.57573073516386,
      "grad_norm": 0.21496829390525818,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 90970
    },
    {
      "epoch": 80.58015943312667,
      "grad_norm": 0.258695125579834,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 90975
    },
    {
      "epoch": 80.58458813108946,
      "grad_norm": 0.22384880483150482,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 90980
    },
    {
      "epoch": 80.58901682905226,
      "grad_norm": 0.22364690899848938,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 90985
    },
    {
      "epoch": 80.59344552701506,
      "grad_norm": 0.20511779189109802,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 90990
    },
    {
      "epoch": 80.59787422497786,
      "grad_norm": 0.25335246324539185,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 90995
    },
    {
      "epoch": 80.60230292294065,
      "grad_norm": 0.2724718749523163,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 91000
    },
    {
      "epoch": 80.60673162090346,
      "grad_norm": 0.24197956919670105,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 91005
    },
    {
      "epoch": 80.61116031886625,
      "grad_norm": 0.2098158895969391,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 91010
    },
    {
      "epoch": 80.61558901682905,
      "grad_norm": 0.27841877937316895,
      "learning_rate": 1e-05,
      "loss": 1.0012,
      "step": 91015
    },
    {
      "epoch": 80.62001771479186,
      "grad_norm": 0.23239122331142426,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 91020
    },
    {
      "epoch": 80.62444641275465,
      "grad_norm": 0.20474544167518616,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 91025
    },
    {
      "epoch": 80.62887511071744,
      "grad_norm": 0.21133720874786377,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 91030
    },
    {
      "epoch": 80.63330380868025,
      "grad_norm": 0.2813125252723694,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 91035
    },
    {
      "epoch": 80.63773250664305,
      "grad_norm": 0.2161625325679779,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 91040
    },
    {
      "epoch": 80.64216120460584,
      "grad_norm": 0.25862497091293335,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 91045
    },
    {
      "epoch": 80.64658990256865,
      "grad_norm": 0.24200892448425293,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 91050
    },
    {
      "epoch": 80.65101860053144,
      "grad_norm": 0.2496143877506256,
      "learning_rate": 1e-05,
      "loss": 0.9023,
      "step": 91055
    },
    {
      "epoch": 80.65544729849424,
      "grad_norm": 0.2194894254207611,
      "learning_rate": 1e-05,
      "loss": 0.9187,
      "step": 91060
    },
    {
      "epoch": 80.65987599645705,
      "grad_norm": 0.23861615359783173,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 91065
    },
    {
      "epoch": 80.66430469441984,
      "grad_norm": 0.25838911533355713,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 91070
    },
    {
      "epoch": 80.66873339238263,
      "grad_norm": 0.2183622568845749,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 91075
    },
    {
      "epoch": 80.67316209034544,
      "grad_norm": 0.231674462556839,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 91080
    },
    {
      "epoch": 80.67759078830824,
      "grad_norm": 0.22098885476589203,
      "learning_rate": 1e-05,
      "loss": 1.0258,
      "step": 91085
    },
    {
      "epoch": 80.68201948627103,
      "grad_norm": 0.23202075064182281,
      "learning_rate": 1e-05,
      "loss": 0.904,
      "step": 91090
    },
    {
      "epoch": 80.68644818423384,
      "grad_norm": 0.24226707220077515,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 91095
    },
    {
      "epoch": 80.69087688219663,
      "grad_norm": 0.24776986241340637,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 91100
    },
    {
      "epoch": 80.69530558015943,
      "grad_norm": 0.2096671462059021,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 91105
    },
    {
      "epoch": 80.69973427812224,
      "grad_norm": 0.2444484382867813,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 91110
    },
    {
      "epoch": 80.70416297608503,
      "grad_norm": 0.27736368775367737,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 91115
    },
    {
      "epoch": 80.70859167404782,
      "grad_norm": 0.261273056268692,
      "learning_rate": 1e-05,
      "loss": 0.8861,
      "step": 91120
    },
    {
      "epoch": 80.71302037201063,
      "grad_norm": 0.2523120641708374,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 91125
    },
    {
      "epoch": 80.71744906997343,
      "grad_norm": 0.2804732024669647,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 91130
    },
    {
      "epoch": 80.72187776793622,
      "grad_norm": 0.2127504199743271,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 91135
    },
    {
      "epoch": 80.72630646589903,
      "grad_norm": 0.2552631199359894,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 91140
    },
    {
      "epoch": 80.73073516386182,
      "grad_norm": 0.24905171990394592,
      "learning_rate": 1e-05,
      "loss": 0.9161,
      "step": 91145
    },
    {
      "epoch": 80.73516386182462,
      "grad_norm": 0.29543524980545044,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 91150
    },
    {
      "epoch": 80.73959255978743,
      "grad_norm": 0.25018495321273804,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 91155
    },
    {
      "epoch": 80.74402125775022,
      "grad_norm": 0.22952282428741455,
      "learning_rate": 1e-05,
      "loss": 0.9277,
      "step": 91160
    },
    {
      "epoch": 80.74844995571301,
      "grad_norm": 0.21172106266021729,
      "learning_rate": 1e-05,
      "loss": 0.9241,
      "step": 91165
    },
    {
      "epoch": 80.75287865367582,
      "grad_norm": 0.2110438197851181,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 91170
    },
    {
      "epoch": 80.75730735163862,
      "grad_norm": 0.21945476531982422,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 91175
    },
    {
      "epoch": 80.76173604960141,
      "grad_norm": 0.23098935186862946,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 91180
    },
    {
      "epoch": 80.76616474756422,
      "grad_norm": 0.25565969944000244,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 91185
    },
    {
      "epoch": 80.77059344552701,
      "grad_norm": 0.24035532772541046,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 91190
    },
    {
      "epoch": 80.77502214348982,
      "grad_norm": 0.24948714673519135,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 91195
    },
    {
      "epoch": 80.77945084145261,
      "grad_norm": 0.2062489241361618,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 91200
    },
    {
      "epoch": 80.78387953941541,
      "grad_norm": 0.22135037183761597,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 91205
    },
    {
      "epoch": 80.78830823737822,
      "grad_norm": 0.1966964602470398,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 91210
    },
    {
      "epoch": 80.79273693534101,
      "grad_norm": 0.2449837476015091,
      "learning_rate": 1e-05,
      "loss": 1.037,
      "step": 91215
    },
    {
      "epoch": 80.7971656333038,
      "grad_norm": 0.23168903589248657,
      "learning_rate": 1e-05,
      "loss": 1.0224,
      "step": 91220
    },
    {
      "epoch": 80.80159433126661,
      "grad_norm": 0.2194792479276657,
      "learning_rate": 1e-05,
      "loss": 0.9061,
      "step": 91225
    },
    {
      "epoch": 80.80602302922941,
      "grad_norm": 0.25971657037734985,
      "learning_rate": 1e-05,
      "loss": 0.9181,
      "step": 91230
    },
    {
      "epoch": 80.8104517271922,
      "grad_norm": 0.2402009814977646,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 91235
    },
    {
      "epoch": 80.81488042515501,
      "grad_norm": 0.21228161454200745,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 91240
    },
    {
      "epoch": 80.8193091231178,
      "grad_norm": 0.23742033541202545,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 91245
    },
    {
      "epoch": 80.8237378210806,
      "grad_norm": 0.24568568170070648,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 91250
    },
    {
      "epoch": 80.8281665190434,
      "grad_norm": 0.24253027141094208,
      "learning_rate": 1e-05,
      "loss": 0.9258,
      "step": 91255
    },
    {
      "epoch": 80.8325952170062,
      "grad_norm": 0.276459664106369,
      "learning_rate": 1e-05,
      "loss": 0.9799,
      "step": 91260
    },
    {
      "epoch": 80.837023914969,
      "grad_norm": 0.22146469354629517,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 91265
    },
    {
      "epoch": 80.8414526129318,
      "grad_norm": 0.22102241218090057,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 91270
    },
    {
      "epoch": 80.8458813108946,
      "grad_norm": 0.2480599731206894,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 91275
    },
    {
      "epoch": 80.85031000885739,
      "grad_norm": 0.24597778916358948,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 91280
    },
    {
      "epoch": 80.8547387068202,
      "grad_norm": 0.24654646217823029,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 91285
    },
    {
      "epoch": 80.859167404783,
      "grad_norm": 0.2694201171398163,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 91290
    },
    {
      "epoch": 80.86359610274579,
      "grad_norm": 0.23365439474582672,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 91295
    },
    {
      "epoch": 80.8680248007086,
      "grad_norm": 0.21915245056152344,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 91300
    },
    {
      "epoch": 80.87245349867139,
      "grad_norm": 0.22658202052116394,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 91305
    },
    {
      "epoch": 80.87688219663418,
      "grad_norm": 0.23377378284931183,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 91310
    },
    {
      "epoch": 80.881310894597,
      "grad_norm": 0.2150438129901886,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 91315
    },
    {
      "epoch": 80.88573959255979,
      "grad_norm": 0.2552729547023773,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 91320
    },
    {
      "epoch": 80.89016829052258,
      "grad_norm": 0.2692682445049286,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 91325
    },
    {
      "epoch": 80.89459698848539,
      "grad_norm": 0.21927419304847717,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 91330
    },
    {
      "epoch": 80.89902568644818,
      "grad_norm": 0.2481800615787506,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 91335
    },
    {
      "epoch": 80.90345438441098,
      "grad_norm": 0.2540706992149353,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 91340
    },
    {
      "epoch": 80.90788308237379,
      "grad_norm": 0.2587836980819702,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 91345
    },
    {
      "epoch": 80.91231178033658,
      "grad_norm": 0.23256359994411469,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 91350
    },
    {
      "epoch": 80.91674047829937,
      "grad_norm": 0.24111385643482208,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 91355
    },
    {
      "epoch": 80.92116917626218,
      "grad_norm": 0.27664461731910706,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 91360
    },
    {
      "epoch": 80.92559787422498,
      "grad_norm": 0.22027702629566193,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 91365
    },
    {
      "epoch": 80.93002657218777,
      "grad_norm": 0.21982650458812714,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 91370
    },
    {
      "epoch": 80.93445527015058,
      "grad_norm": 0.23025935888290405,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 91375
    },
    {
      "epoch": 80.93888396811337,
      "grad_norm": 0.21099212765693665,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 91380
    },
    {
      "epoch": 80.94331266607617,
      "grad_norm": 0.26262176036834717,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 91385
    },
    {
      "epoch": 80.94774136403898,
      "grad_norm": 0.24932602047920227,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 91390
    },
    {
      "epoch": 80.95217006200177,
      "grad_norm": 0.22671513259410858,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 91395
    },
    {
      "epoch": 80.95659875996456,
      "grad_norm": 0.27160924673080444,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 91400
    },
    {
      "epoch": 80.96102745792737,
      "grad_norm": 0.2533956468105316,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 91405
    },
    {
      "epoch": 80.96545615589017,
      "grad_norm": 0.2485702484846115,
      "learning_rate": 1e-05,
      "loss": 0.9161,
      "step": 91410
    },
    {
      "epoch": 80.96988485385296,
      "grad_norm": 0.258137047290802,
      "learning_rate": 1e-05,
      "loss": 0.9176,
      "step": 91415
    },
    {
      "epoch": 80.97431355181577,
      "grad_norm": 0.2436603605747223,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 91420
    },
    {
      "epoch": 80.97874224977856,
      "grad_norm": 0.29182374477386475,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 91425
    },
    {
      "epoch": 80.98317094774137,
      "grad_norm": 0.26369819045066833,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 91430
    },
    {
      "epoch": 80.98759964570417,
      "grad_norm": 0.322521835565567,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 91435
    },
    {
      "epoch": 80.99202834366696,
      "grad_norm": 0.2459186315536499,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 91440
    },
    {
      "epoch": 80.99645704162977,
      "grad_norm": 0.26624155044555664,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 91445
    },
    {
      "epoch": 81.00088573959256,
      "grad_norm": 0.2507934272289276,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 91450
    },
    {
      "epoch": 81.00531443755536,
      "grad_norm": 0.227080300450325,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 91455
    },
    {
      "epoch": 81.00974313551816,
      "grad_norm": 0.24502043426036835,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 91460
    },
    {
      "epoch": 81.01417183348096,
      "grad_norm": 0.285030335187912,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 91465
    },
    {
      "epoch": 81.01860053144375,
      "grad_norm": 0.3109884560108185,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 91470
    },
    {
      "epoch": 81.02302922940656,
      "grad_norm": 0.24894285202026367,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 91475
    },
    {
      "epoch": 81.02745792736935,
      "grad_norm": 0.24787092208862305,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 91480
    },
    {
      "epoch": 81.03188662533215,
      "grad_norm": 0.3745787739753723,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 91485
    },
    {
      "epoch": 81.03631532329496,
      "grad_norm": 0.30479496717453003,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 91490
    },
    {
      "epoch": 81.04074402125775,
      "grad_norm": 0.24336491525173187,
      "learning_rate": 1e-05,
      "loss": 0.8926,
      "step": 91495
    },
    {
      "epoch": 81.04517271922055,
      "grad_norm": 0.26006007194519043,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 91500
    },
    {
      "epoch": 81.04960141718335,
      "grad_norm": 0.24037177860736847,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 91505
    },
    {
      "epoch": 81.05403011514615,
      "grad_norm": 0.2424057424068451,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 91510
    },
    {
      "epoch": 81.05845881310894,
      "grad_norm": 0.19677449762821198,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 91515
    },
    {
      "epoch": 81.06288751107175,
      "grad_norm": 0.22905278205871582,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 91520
    },
    {
      "epoch": 81.06731620903454,
      "grad_norm": 0.2423352301120758,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 91525
    },
    {
      "epoch": 81.07174490699734,
      "grad_norm": 0.20317932963371277,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 91530
    },
    {
      "epoch": 81.07617360496015,
      "grad_norm": 0.23584222793579102,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 91535
    },
    {
      "epoch": 81.08060230292294,
      "grad_norm": 0.24752803146839142,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 91540
    },
    {
      "epoch": 81.08503100088573,
      "grad_norm": 0.2781928479671478,
      "learning_rate": 1e-05,
      "loss": 0.9779,
      "step": 91545
    },
    {
      "epoch": 81.08945969884854,
      "grad_norm": 0.27298372983932495,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 91550
    },
    {
      "epoch": 81.09388839681134,
      "grad_norm": 0.25532519817352295,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 91555
    },
    {
      "epoch": 81.09831709477413,
      "grad_norm": 0.25626635551452637,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 91560
    },
    {
      "epoch": 81.10274579273694,
      "grad_norm": 0.3022654950618744,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 91565
    },
    {
      "epoch": 81.10717449069973,
      "grad_norm": 0.24486123025417328,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 91570
    },
    {
      "epoch": 81.11160318866253,
      "grad_norm": 0.23280680179595947,
      "learning_rate": 1e-05,
      "loss": 0.9295,
      "step": 91575
    },
    {
      "epoch": 81.11603188662534,
      "grad_norm": 0.25300851464271545,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 91580
    },
    {
      "epoch": 81.12046058458813,
      "grad_norm": 0.2673119008541107,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 91585
    },
    {
      "epoch": 81.12488928255092,
      "grad_norm": 0.24592231214046478,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 91590
    },
    {
      "epoch": 81.12931798051373,
      "grad_norm": 0.33494842052459717,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 91595
    },
    {
      "epoch": 81.13374667847653,
      "grad_norm": 0.25529637932777405,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 91600
    },
    {
      "epoch": 81.13817537643932,
      "grad_norm": 0.247398242354393,
      "learning_rate": 1e-05,
      "loss": 0.9342,
      "step": 91605
    },
    {
      "epoch": 81.14260407440213,
      "grad_norm": 0.28469541668891907,
      "learning_rate": 1e-05,
      "loss": 0.9214,
      "step": 91610
    },
    {
      "epoch": 81.14703277236492,
      "grad_norm": 0.2281649261713028,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 91615
    },
    {
      "epoch": 81.15146147032772,
      "grad_norm": 0.2777380347251892,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 91620
    },
    {
      "epoch": 81.15589016829053,
      "grad_norm": 0.24761246144771576,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 91625
    },
    {
      "epoch": 81.16031886625332,
      "grad_norm": 0.2413511574268341,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 91630
    },
    {
      "epoch": 81.16474756421611,
      "grad_norm": 0.26133501529693604,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 91635
    },
    {
      "epoch": 81.16917626217892,
      "grad_norm": 0.24164548516273499,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 91640
    },
    {
      "epoch": 81.17360496014172,
      "grad_norm": 0.21848039329051971,
      "learning_rate": 1e-05,
      "loss": 0.9214,
      "step": 91645
    },
    {
      "epoch": 81.17803365810451,
      "grad_norm": 0.24051550030708313,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 91650
    },
    {
      "epoch": 81.18246235606732,
      "grad_norm": 0.21918097138404846,
      "learning_rate": 1e-05,
      "loss": 0.9024,
      "step": 91655
    },
    {
      "epoch": 81.18689105403011,
      "grad_norm": 0.21730661392211914,
      "learning_rate": 1e-05,
      "loss": 0.9212,
      "step": 91660
    },
    {
      "epoch": 81.1913197519929,
      "grad_norm": 0.247291699051857,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 91665
    },
    {
      "epoch": 81.19574844995572,
      "grad_norm": 0.2425602674484253,
      "learning_rate": 1e-05,
      "loss": 0.9311,
      "step": 91670
    },
    {
      "epoch": 81.20017714791851,
      "grad_norm": 0.2484261393547058,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 91675
    },
    {
      "epoch": 81.20460584588132,
      "grad_norm": 0.2652241289615631,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 91680
    },
    {
      "epoch": 81.20903454384411,
      "grad_norm": 0.28946951031684875,
      "learning_rate": 1e-05,
      "loss": 0.9835,
      "step": 91685
    },
    {
      "epoch": 81.2134632418069,
      "grad_norm": 0.21521642804145813,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 91690
    },
    {
      "epoch": 81.21789193976971,
      "grad_norm": 0.23349493741989136,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 91695
    },
    {
      "epoch": 81.22232063773251,
      "grad_norm": 0.2843078374862671,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 91700
    },
    {
      "epoch": 81.2267493356953,
      "grad_norm": 0.23122236132621765,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 91705
    },
    {
      "epoch": 81.23117803365811,
      "grad_norm": 0.23589713871479034,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 91710
    },
    {
      "epoch": 81.2356067316209,
      "grad_norm": 0.23940306901931763,
      "learning_rate": 1e-05,
      "loss": 1.0633,
      "step": 91715
    },
    {
      "epoch": 81.2400354295837,
      "grad_norm": 0.2338026463985443,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 91720
    },
    {
      "epoch": 81.24446412754651,
      "grad_norm": 0.25521382689476013,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 91725
    },
    {
      "epoch": 81.2488928255093,
      "grad_norm": 0.20579414069652557,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 91730
    },
    {
      "epoch": 81.2533215234721,
      "grad_norm": 0.203611359000206,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 91735
    },
    {
      "epoch": 81.2577502214349,
      "grad_norm": 0.26565149426460266,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 91740
    },
    {
      "epoch": 81.2621789193977,
      "grad_norm": 0.25646114349365234,
      "learning_rate": 1e-05,
      "loss": 0.9139,
      "step": 91745
    },
    {
      "epoch": 81.26660761736049,
      "grad_norm": 0.24975556135177612,
      "learning_rate": 1e-05,
      "loss": 1.0507,
      "step": 91750
    },
    {
      "epoch": 81.2710363153233,
      "grad_norm": 0.21961677074432373,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 91755
    },
    {
      "epoch": 81.2754650132861,
      "grad_norm": 0.2566573917865753,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 91760
    },
    {
      "epoch": 81.27989371124889,
      "grad_norm": 0.25011637806892395,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 91765
    },
    {
      "epoch": 81.2843224092117,
      "grad_norm": 0.26534557342529297,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 91770
    },
    {
      "epoch": 81.28875110717449,
      "grad_norm": 0.21518675982952118,
      "learning_rate": 1e-05,
      "loss": 0.9075,
      "step": 91775
    },
    {
      "epoch": 81.29317980513729,
      "grad_norm": 0.24369928240776062,
      "learning_rate": 1e-05,
      "loss": 1.0214,
      "step": 91780
    },
    {
      "epoch": 81.2976085031001,
      "grad_norm": 0.23437944054603577,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 91785
    },
    {
      "epoch": 81.30203720106289,
      "grad_norm": 0.20320825278759003,
      "learning_rate": 1e-05,
      "loss": 0.8983,
      "step": 91790
    },
    {
      "epoch": 81.30646589902568,
      "grad_norm": 0.2202541083097458,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 91795
    },
    {
      "epoch": 81.31089459698849,
      "grad_norm": 0.2626502513885498,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 91800
    },
    {
      "epoch": 81.31532329495128,
      "grad_norm": 0.2656176686286926,
      "learning_rate": 1e-05,
      "loss": 0.8879,
      "step": 91805
    },
    {
      "epoch": 81.31975199291408,
      "grad_norm": 0.22513507306575775,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 91810
    },
    {
      "epoch": 81.32418069087689,
      "grad_norm": 0.2819122076034546,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 91815
    },
    {
      "epoch": 81.32860938883968,
      "grad_norm": 0.21301336586475372,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 91820
    },
    {
      "epoch": 81.33303808680247,
      "grad_norm": 0.2654283046722412,
      "learning_rate": 1e-05,
      "loss": 0.9005,
      "step": 91825
    },
    {
      "epoch": 81.33746678476528,
      "grad_norm": 0.22072476148605347,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 91830
    },
    {
      "epoch": 81.34189548272808,
      "grad_norm": 0.3169809579849243,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 91835
    },
    {
      "epoch": 81.34632418069087,
      "grad_norm": 0.22610099613666534,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 91840
    },
    {
      "epoch": 81.35075287865368,
      "grad_norm": 0.24698369204998016,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 91845
    },
    {
      "epoch": 81.35518157661647,
      "grad_norm": 0.21200230717658997,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 91850
    },
    {
      "epoch": 81.35961027457927,
      "grad_norm": 0.2316744178533554,
      "learning_rate": 1e-05,
      "loss": 1.0299,
      "step": 91855
    },
    {
      "epoch": 81.36403897254208,
      "grad_norm": 0.2111215889453888,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 91860
    },
    {
      "epoch": 81.36846767050487,
      "grad_norm": 0.2498284876346588,
      "learning_rate": 1e-05,
      "loss": 0.9221,
      "step": 91865
    },
    {
      "epoch": 81.37289636846766,
      "grad_norm": 0.2121538519859314,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 91870
    },
    {
      "epoch": 81.37732506643047,
      "grad_norm": 0.2108762115240097,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 91875
    },
    {
      "epoch": 81.38175376439327,
      "grad_norm": 0.2326587438583374,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 91880
    },
    {
      "epoch": 81.38618246235606,
      "grad_norm": 0.24290013313293457,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 91885
    },
    {
      "epoch": 81.39061116031887,
      "grad_norm": 0.2500825524330139,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 91890
    },
    {
      "epoch": 81.39503985828166,
      "grad_norm": 0.23301725089550018,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 91895
    },
    {
      "epoch": 81.39946855624446,
      "grad_norm": 0.22314879298210144,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 91900
    },
    {
      "epoch": 81.40389725420727,
      "grad_norm": 0.24765098094940186,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 91905
    },
    {
      "epoch": 81.40832595217006,
      "grad_norm": 0.23086926341056824,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 91910
    },
    {
      "epoch": 81.41275465013285,
      "grad_norm": 0.2578636109828949,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 91915
    },
    {
      "epoch": 81.41718334809566,
      "grad_norm": 0.24778832495212555,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 91920
    },
    {
      "epoch": 81.42161204605846,
      "grad_norm": 0.27392515540122986,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 91925
    },
    {
      "epoch": 81.42604074402126,
      "grad_norm": 0.23272523283958435,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 91930
    },
    {
      "epoch": 81.43046944198406,
      "grad_norm": 0.2507994472980499,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 91935
    },
    {
      "epoch": 81.43489813994685,
      "grad_norm": 0.25610029697418213,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 91940
    },
    {
      "epoch": 81.43932683790966,
      "grad_norm": 0.27585524320602417,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 91945
    },
    {
      "epoch": 81.44375553587246,
      "grad_norm": 0.2214675396680832,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 91950
    },
    {
      "epoch": 81.44818423383525,
      "grad_norm": 0.22733256220817566,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 91955
    },
    {
      "epoch": 81.45261293179806,
      "grad_norm": 0.2225317507982254,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 91960
    },
    {
      "epoch": 81.45704162976085,
      "grad_norm": 0.2192356288433075,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 91965
    },
    {
      "epoch": 81.46147032772365,
      "grad_norm": 0.19329208135604858,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 91970
    },
    {
      "epoch": 81.46589902568645,
      "grad_norm": 0.23062781989574432,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 91975
    },
    {
      "epoch": 81.47032772364925,
      "grad_norm": 0.24189935624599457,
      "learning_rate": 1e-05,
      "loss": 0.92,
      "step": 91980
    },
    {
      "epoch": 81.47475642161204,
      "grad_norm": 0.22949936985969543,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 91985
    },
    {
      "epoch": 81.47918511957485,
      "grad_norm": 0.2188076674938202,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 91990
    },
    {
      "epoch": 81.48361381753764,
      "grad_norm": 0.21806061267852783,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 91995
    },
    {
      "epoch": 81.48804251550044,
      "grad_norm": 0.2329736351966858,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 92000
    },
    {
      "epoch": 81.49247121346325,
      "grad_norm": 0.21016301214694977,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 92005
    },
    {
      "epoch": 81.49689991142604,
      "grad_norm": 0.20761898159980774,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 92010
    },
    {
      "epoch": 81.50132860938884,
      "grad_norm": 0.2447303831577301,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 92015
    },
    {
      "epoch": 81.50575730735164,
      "grad_norm": 0.23071740567684174,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 92020
    },
    {
      "epoch": 81.51018600531444,
      "grad_norm": 0.23061630129814148,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 92025
    },
    {
      "epoch": 81.51461470327723,
      "grad_norm": 0.26496872305870056,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 92030
    },
    {
      "epoch": 81.51904340124004,
      "grad_norm": 0.23935557901859283,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 92035
    },
    {
      "epoch": 81.52347209920283,
      "grad_norm": 0.24850115180015564,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 92040
    },
    {
      "epoch": 81.52790079716563,
      "grad_norm": 0.24320080876350403,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 92045
    },
    {
      "epoch": 81.53232949512844,
      "grad_norm": 0.2559666633605957,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 92050
    },
    {
      "epoch": 81.53675819309123,
      "grad_norm": 0.19619694352149963,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 92055
    },
    {
      "epoch": 81.54118689105402,
      "grad_norm": 0.23495841026306152,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 92060
    },
    {
      "epoch": 81.54561558901683,
      "grad_norm": 0.24964475631713867,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 92065
    },
    {
      "epoch": 81.55004428697963,
      "grad_norm": 0.24080830812454224,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 92070
    },
    {
      "epoch": 81.55447298494242,
      "grad_norm": 0.24716900289058685,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 92075
    },
    {
      "epoch": 81.55890168290523,
      "grad_norm": 0.32225799560546875,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 92080
    },
    {
      "epoch": 81.56333038086802,
      "grad_norm": 0.23737460374832153,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 92085
    },
    {
      "epoch": 81.56775907883082,
      "grad_norm": 0.2512807250022888,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 92090
    },
    {
      "epoch": 81.57218777679363,
      "grad_norm": 0.22749358415603638,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 92095
    },
    {
      "epoch": 81.57661647475642,
      "grad_norm": 0.26936012506484985,
      "learning_rate": 1e-05,
      "loss": 0.9301,
      "step": 92100
    },
    {
      "epoch": 81.58104517271921,
      "grad_norm": 0.22237396240234375,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 92105
    },
    {
      "epoch": 81.58547387068202,
      "grad_norm": 0.22897861897945404,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 92110
    },
    {
      "epoch": 81.58990256864482,
      "grad_norm": 0.24843649566173553,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 92115
    },
    {
      "epoch": 81.59433126660761,
      "grad_norm": 0.23942862451076508,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 92120
    },
    {
      "epoch": 81.59875996457042,
      "grad_norm": 0.24777622520923615,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 92125
    },
    {
      "epoch": 81.60318866253321,
      "grad_norm": 0.25789037346839905,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 92130
    },
    {
      "epoch": 81.60761736049601,
      "grad_norm": 0.25624892115592957,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 92135
    },
    {
      "epoch": 81.61204605845882,
      "grad_norm": 0.23858697712421417,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 92140
    },
    {
      "epoch": 81.61647475642161,
      "grad_norm": 0.26931270956993103,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 92145
    },
    {
      "epoch": 81.6209034543844,
      "grad_norm": 0.26362141966819763,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 92150
    },
    {
      "epoch": 81.62533215234721,
      "grad_norm": 0.23770059645175934,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 92155
    },
    {
      "epoch": 81.62976085031,
      "grad_norm": 0.22525376081466675,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 92160
    },
    {
      "epoch": 81.63418954827281,
      "grad_norm": 0.21832843124866486,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 92165
    },
    {
      "epoch": 81.63861824623561,
      "grad_norm": 0.2812003493309021,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 92170
    },
    {
      "epoch": 81.6430469441984,
      "grad_norm": 0.21997977793216705,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 92175
    },
    {
      "epoch": 81.64747564216121,
      "grad_norm": 0.23646268248558044,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 92180
    },
    {
      "epoch": 81.651904340124,
      "grad_norm": 0.23823480308055878,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 92185
    },
    {
      "epoch": 81.6563330380868,
      "grad_norm": 0.25054866075515747,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 92190
    },
    {
      "epoch": 81.66076173604961,
      "grad_norm": 0.2461187094449997,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 92195
    },
    {
      "epoch": 81.6651904340124,
      "grad_norm": 0.24883274734020233,
      "learning_rate": 1e-05,
      "loss": 0.8884,
      "step": 92200
    },
    {
      "epoch": 81.6696191319752,
      "grad_norm": 0.238087460398674,
      "learning_rate": 1e-05,
      "loss": 1.045,
      "step": 92205
    },
    {
      "epoch": 81.674047829938,
      "grad_norm": 0.2064228355884552,
      "learning_rate": 1e-05,
      "loss": 0.9205,
      "step": 92210
    },
    {
      "epoch": 81.6784765279008,
      "grad_norm": 0.2297319769859314,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 92215
    },
    {
      "epoch": 81.68290522586359,
      "grad_norm": 0.2751140892505646,
      "learning_rate": 1e-05,
      "loss": 0.9188,
      "step": 92220
    },
    {
      "epoch": 81.6873339238264,
      "grad_norm": 0.23701654374599457,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 92225
    },
    {
      "epoch": 81.6917626217892,
      "grad_norm": 0.2428114414215088,
      "learning_rate": 1e-05,
      "loss": 0.921,
      "step": 92230
    },
    {
      "epoch": 81.69619131975199,
      "grad_norm": 0.24080325663089752,
      "learning_rate": 1e-05,
      "loss": 0.9408,
      "step": 92235
    },
    {
      "epoch": 81.7006200177148,
      "grad_norm": 0.2512313425540924,
      "learning_rate": 1e-05,
      "loss": 1.0007,
      "step": 92240
    },
    {
      "epoch": 81.70504871567759,
      "grad_norm": 0.2967431843280792,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 92245
    },
    {
      "epoch": 81.70947741364039,
      "grad_norm": 0.20880401134490967,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 92250
    },
    {
      "epoch": 81.7139061116032,
      "grad_norm": 0.22423481941223145,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 92255
    },
    {
      "epoch": 81.71833480956599,
      "grad_norm": 0.20792345702648163,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 92260
    },
    {
      "epoch": 81.72276350752878,
      "grad_norm": 0.2642989754676819,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 92265
    },
    {
      "epoch": 81.72719220549159,
      "grad_norm": 0.2577972114086151,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 92270
    },
    {
      "epoch": 81.73162090345438,
      "grad_norm": 0.22807720303535461,
      "learning_rate": 1e-05,
      "loss": 0.9197,
      "step": 92275
    },
    {
      "epoch": 81.73604960141718,
      "grad_norm": 0.30161944031715393,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 92280
    },
    {
      "epoch": 81.74047829937999,
      "grad_norm": 0.2140309363603592,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 92285
    },
    {
      "epoch": 81.74490699734278,
      "grad_norm": 0.24424193799495697,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 92290
    },
    {
      "epoch": 81.74933569530558,
      "grad_norm": 0.23796679079532623,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 92295
    },
    {
      "epoch": 81.75376439326838,
      "grad_norm": 0.2624052166938782,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 92300
    },
    {
      "epoch": 81.75819309123118,
      "grad_norm": 0.24641534686088562,
      "learning_rate": 1e-05,
      "loss": 0.9177,
      "step": 92305
    },
    {
      "epoch": 81.76262178919397,
      "grad_norm": 0.22633585333824158,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 92310
    },
    {
      "epoch": 81.76705048715678,
      "grad_norm": 0.2430175095796585,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 92315
    },
    {
      "epoch": 81.77147918511957,
      "grad_norm": 0.2352379709482193,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 92320
    },
    {
      "epoch": 81.77590788308237,
      "grad_norm": 0.27295732498168945,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 92325
    },
    {
      "epoch": 81.78033658104518,
      "grad_norm": 0.2678602635860443,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 92330
    },
    {
      "epoch": 81.78476527900797,
      "grad_norm": 0.26329970359802246,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 92335
    },
    {
      "epoch": 81.78919397697076,
      "grad_norm": 0.21536113321781158,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 92340
    },
    {
      "epoch": 81.79362267493357,
      "grad_norm": 0.21278207004070282,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 92345
    },
    {
      "epoch": 81.79805137289637,
      "grad_norm": 0.2980216443538666,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 92350
    },
    {
      "epoch": 81.80248007085916,
      "grad_norm": 0.23702271282672882,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 92355
    },
    {
      "epoch": 81.80690876882197,
      "grad_norm": 0.2431996762752533,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 92360
    },
    {
      "epoch": 81.81133746678476,
      "grad_norm": 0.2558686137199402,
      "learning_rate": 1e-05,
      "loss": 0.8983,
      "step": 92365
    },
    {
      "epoch": 81.81576616474756,
      "grad_norm": 0.26055800914764404,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 92370
    },
    {
      "epoch": 81.82019486271037,
      "grad_norm": 0.2423030436038971,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 92375
    },
    {
      "epoch": 81.82462356067316,
      "grad_norm": 0.2459389865398407,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 92380
    },
    {
      "epoch": 81.82905225863595,
      "grad_norm": 0.21564634144306183,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 92385
    },
    {
      "epoch": 81.83348095659876,
      "grad_norm": 0.2186041921377182,
      "learning_rate": 1e-05,
      "loss": 0.9346,
      "step": 92390
    },
    {
      "epoch": 81.83790965456156,
      "grad_norm": 0.21956850588321686,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 92395
    },
    {
      "epoch": 81.84233835252435,
      "grad_norm": 0.2123612016439438,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 92400
    },
    {
      "epoch": 81.84676705048716,
      "grad_norm": 0.21563979983329773,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 92405
    },
    {
      "epoch": 81.85119574844995,
      "grad_norm": 0.2123672068119049,
      "learning_rate": 1e-05,
      "loss": 0.9117,
      "step": 92410
    },
    {
      "epoch": 81.85562444641276,
      "grad_norm": 0.240903839468956,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 92415
    },
    {
      "epoch": 81.86005314437556,
      "grad_norm": 0.2425391674041748,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 92420
    },
    {
      "epoch": 81.86448184233835,
      "grad_norm": 0.23530447483062744,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 92425
    },
    {
      "epoch": 81.86891054030116,
      "grad_norm": 0.24704478681087494,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 92430
    },
    {
      "epoch": 81.87333923826395,
      "grad_norm": 0.272141695022583,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 92435
    },
    {
      "epoch": 81.87776793622675,
      "grad_norm": 0.2595696747303009,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 92440
    },
    {
      "epoch": 81.88219663418955,
      "grad_norm": 0.22840823233127594,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 92445
    },
    {
      "epoch": 81.88662533215235,
      "grad_norm": 0.24744726717472076,
      "learning_rate": 1e-05,
      "loss": 1.0123,
      "step": 92450
    },
    {
      "epoch": 81.89105403011514,
      "grad_norm": 0.2935553789138794,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 92455
    },
    {
      "epoch": 81.89548272807795,
      "grad_norm": 0.2323647290468216,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 92460
    },
    {
      "epoch": 81.89991142604075,
      "grad_norm": 0.2983042299747467,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 92465
    },
    {
      "epoch": 81.90434012400354,
      "grad_norm": 0.2761507034301758,
      "learning_rate": 1e-05,
      "loss": 0.9873,
      "step": 92470
    },
    {
      "epoch": 81.90876882196635,
      "grad_norm": 0.25779053568840027,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 92475
    },
    {
      "epoch": 81.91319751992914,
      "grad_norm": 0.28125467896461487,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 92480
    },
    {
      "epoch": 81.91762621789194,
      "grad_norm": 0.23626072704792023,
      "learning_rate": 1e-05,
      "loss": 1.025,
      "step": 92485
    },
    {
      "epoch": 81.92205491585474,
      "grad_norm": 0.2601568102836609,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 92490
    },
    {
      "epoch": 81.92648361381754,
      "grad_norm": 0.25193536281585693,
      "learning_rate": 1e-05,
      "loss": 0.9012,
      "step": 92495
    },
    {
      "epoch": 81.93091231178033,
      "grad_norm": 0.20401793718338013,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 92500
    },
    {
      "epoch": 81.93534100974314,
      "grad_norm": 0.22088439762592316,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 92505
    },
    {
      "epoch": 81.93976970770593,
      "grad_norm": 0.21642133593559265,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 92510
    },
    {
      "epoch": 81.94419840566873,
      "grad_norm": 0.23719359934329987,
      "learning_rate": 1e-05,
      "loss": 0.909,
      "step": 92515
    },
    {
      "epoch": 81.94862710363154,
      "grad_norm": 0.22414977848529816,
      "learning_rate": 1e-05,
      "loss": 0.9215,
      "step": 92520
    },
    {
      "epoch": 81.95305580159433,
      "grad_norm": 0.23038838803768158,
      "learning_rate": 1e-05,
      "loss": 0.9348,
      "step": 92525
    },
    {
      "epoch": 81.95748449955713,
      "grad_norm": 0.26599398255348206,
      "learning_rate": 1e-05,
      "loss": 0.911,
      "step": 92530
    },
    {
      "epoch": 81.96191319751993,
      "grad_norm": 0.25525933504104614,
      "learning_rate": 1e-05,
      "loss": 0.9149,
      "step": 92535
    },
    {
      "epoch": 81.96634189548273,
      "grad_norm": 0.26162856817245483,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 92540
    },
    {
      "epoch": 81.97077059344552,
      "grad_norm": 0.2528301477432251,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 92545
    },
    {
      "epoch": 81.97519929140833,
      "grad_norm": 0.25905710458755493,
      "learning_rate": 1e-05,
      "loss": 0.9333,
      "step": 92550
    },
    {
      "epoch": 81.97962798937112,
      "grad_norm": 0.2095557302236557,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 92555
    },
    {
      "epoch": 81.98405668733392,
      "grad_norm": 0.24877482652664185,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 92560
    },
    {
      "epoch": 81.98848538529673,
      "grad_norm": 0.23846156895160675,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 92565
    },
    {
      "epoch": 81.99291408325952,
      "grad_norm": 0.24456891417503357,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 92570
    },
    {
      "epoch": 81.99734278122232,
      "grad_norm": 0.24258312582969666,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 92575
    },
    {
      "epoch": 82.00177147918512,
      "grad_norm": 0.22928327322006226,
      "learning_rate": 1e-05,
      "loss": 0.9258,
      "step": 92580
    },
    {
      "epoch": 82.00620017714792,
      "grad_norm": 0.2241503894329071,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 92585
    },
    {
      "epoch": 82.01062887511071,
      "grad_norm": 0.2845040559768677,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 92590
    },
    {
      "epoch": 82.01505757307352,
      "grad_norm": 0.2193719744682312,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 92595
    },
    {
      "epoch": 82.01948627103631,
      "grad_norm": 0.2309652864933014,
      "learning_rate": 1e-05,
      "loss": 0.9165,
      "step": 92600
    },
    {
      "epoch": 82.02391496899911,
      "grad_norm": 0.2453942745923996,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 92605
    },
    {
      "epoch": 82.02834366696192,
      "grad_norm": 0.21675848960876465,
      "learning_rate": 1e-05,
      "loss": 0.9219,
      "step": 92610
    },
    {
      "epoch": 82.03277236492471,
      "grad_norm": 0.2201230376958847,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 92615
    },
    {
      "epoch": 82.0372010628875,
      "grad_norm": 0.2498418390750885,
      "learning_rate": 1e-05,
      "loss": 1.0335,
      "step": 92620
    },
    {
      "epoch": 82.04162976085031,
      "grad_norm": 0.245519757270813,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 92625
    },
    {
      "epoch": 82.0460584588131,
      "grad_norm": 0.2237943708896637,
      "learning_rate": 1e-05,
      "loss": 0.9307,
      "step": 92630
    },
    {
      "epoch": 82.0504871567759,
      "grad_norm": 0.21586014330387115,
      "learning_rate": 1e-05,
      "loss": 0.9114,
      "step": 92635
    },
    {
      "epoch": 82.05491585473871,
      "grad_norm": 0.271984726190567,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 92640
    },
    {
      "epoch": 82.0593445527015,
      "grad_norm": 0.2107381373643875,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 92645
    },
    {
      "epoch": 82.0637732506643,
      "grad_norm": 0.2336064577102661,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 92650
    },
    {
      "epoch": 82.0682019486271,
      "grad_norm": 0.24420031905174255,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 92655
    },
    {
      "epoch": 82.0726306465899,
      "grad_norm": 0.2063104510307312,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 92660
    },
    {
      "epoch": 82.07705934455271,
      "grad_norm": 0.27653226256370544,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 92665
    },
    {
      "epoch": 82.0814880425155,
      "grad_norm": 0.3089624345302582,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 92670
    },
    {
      "epoch": 82.0859167404783,
      "grad_norm": 0.20853734016418457,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 92675
    },
    {
      "epoch": 82.0903454384411,
      "grad_norm": 0.2841624617576599,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 92680
    },
    {
      "epoch": 82.0947741364039,
      "grad_norm": 0.26382946968078613,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 92685
    },
    {
      "epoch": 82.0992028343667,
      "grad_norm": 0.29224419593811035,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 92690
    },
    {
      "epoch": 82.1036315323295,
      "grad_norm": 0.30479520559310913,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 92695
    },
    {
      "epoch": 82.1080602302923,
      "grad_norm": 0.2462264448404312,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 92700
    },
    {
      "epoch": 82.11248892825509,
      "grad_norm": 0.23337511718273163,
      "learning_rate": 1e-05,
      "loss": 1.0242,
      "step": 92705
    },
    {
      "epoch": 82.1169176262179,
      "grad_norm": 0.2221749722957611,
      "learning_rate": 1e-05,
      "loss": 0.9077,
      "step": 92710
    },
    {
      "epoch": 82.12134632418069,
      "grad_norm": 0.23320898413658142,
      "learning_rate": 1e-05,
      "loss": 0.9975,
      "step": 92715
    },
    {
      "epoch": 82.12577502214349,
      "grad_norm": 0.21928885579109192,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 92720
    },
    {
      "epoch": 82.1302037201063,
      "grad_norm": 0.23455867171287537,
      "learning_rate": 1e-05,
      "loss": 0.914,
      "step": 92725
    },
    {
      "epoch": 82.13463241806909,
      "grad_norm": 0.2031598538160324,
      "learning_rate": 1e-05,
      "loss": 0.9223,
      "step": 92730
    },
    {
      "epoch": 82.13906111603188,
      "grad_norm": 0.19815897941589355,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 92735
    },
    {
      "epoch": 82.14348981399469,
      "grad_norm": 0.23133572936058044,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 92740
    },
    {
      "epoch": 82.14791851195749,
      "grad_norm": 0.2101556956768036,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 92745
    },
    {
      "epoch": 82.15234720992028,
      "grad_norm": 0.2910807728767395,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 92750
    },
    {
      "epoch": 82.15677590788309,
      "grad_norm": 0.3049667179584503,
      "learning_rate": 1e-05,
      "loss": 0.9044,
      "step": 92755
    },
    {
      "epoch": 82.16120460584588,
      "grad_norm": 0.2579787075519562,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 92760
    },
    {
      "epoch": 82.16563330380868,
      "grad_norm": 0.23370757699012756,
      "learning_rate": 1e-05,
      "loss": 1.0367,
      "step": 92765
    },
    {
      "epoch": 82.17006200177148,
      "grad_norm": 0.28080523014068604,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 92770
    },
    {
      "epoch": 82.17449069973428,
      "grad_norm": 0.20041103661060333,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 92775
    },
    {
      "epoch": 82.17891939769707,
      "grad_norm": 0.2114713042974472,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 92780
    },
    {
      "epoch": 82.18334809565988,
      "grad_norm": 0.2356327325105667,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 92785
    },
    {
      "epoch": 82.18777679362267,
      "grad_norm": 0.21957969665527344,
      "learning_rate": 1e-05,
      "loss": 0.9249,
      "step": 92790
    },
    {
      "epoch": 82.19220549158547,
      "grad_norm": 0.24526815116405487,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 92795
    },
    {
      "epoch": 82.19663418954828,
      "grad_norm": 0.2780296504497528,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 92800
    },
    {
      "epoch": 82.20106288751107,
      "grad_norm": 0.22559769451618195,
      "learning_rate": 1e-05,
      "loss": 0.9091,
      "step": 92805
    },
    {
      "epoch": 82.20549158547387,
      "grad_norm": 0.19436782598495483,
      "learning_rate": 1e-05,
      "loss": 0.923,
      "step": 92810
    },
    {
      "epoch": 82.20992028343667,
      "grad_norm": 0.2156790792942047,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 92815
    },
    {
      "epoch": 82.21434898139947,
      "grad_norm": 0.22508729994297028,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 92820
    },
    {
      "epoch": 82.21877767936226,
      "grad_norm": 0.26046520471572876,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 92825
    },
    {
      "epoch": 82.22320637732507,
      "grad_norm": 0.21837759017944336,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 92830
    },
    {
      "epoch": 82.22763507528786,
      "grad_norm": 0.22372931241989136,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 92835
    },
    {
      "epoch": 82.23206377325066,
      "grad_norm": 0.2550649642944336,
      "learning_rate": 1e-05,
      "loss": 0.9324,
      "step": 92840
    },
    {
      "epoch": 82.23649247121347,
      "grad_norm": 0.28876209259033203,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 92845
    },
    {
      "epoch": 82.24092116917626,
      "grad_norm": 0.24127979576587677,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 92850
    },
    {
      "epoch": 82.24534986713905,
      "grad_norm": 0.2675110399723053,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 92855
    },
    {
      "epoch": 82.24977856510186,
      "grad_norm": 0.2507060766220093,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 92860
    },
    {
      "epoch": 82.25420726306466,
      "grad_norm": 0.2154734879732132,
      "learning_rate": 1e-05,
      "loss": 0.9834,
      "step": 92865
    },
    {
      "epoch": 82.25863596102745,
      "grad_norm": 0.2733849585056305,
      "learning_rate": 1e-05,
      "loss": 0.929,
      "step": 92870
    },
    {
      "epoch": 82.26306465899026,
      "grad_norm": 0.22636304795742035,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 92875
    },
    {
      "epoch": 82.26749335695305,
      "grad_norm": 0.22808566689491272,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 92880
    },
    {
      "epoch": 82.27192205491585,
      "grad_norm": 0.2521180212497711,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 92885
    },
    {
      "epoch": 82.27635075287866,
      "grad_norm": 0.23331858217716217,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 92890
    },
    {
      "epoch": 82.28077945084145,
      "grad_norm": 0.23508606851100922,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 92895
    },
    {
      "epoch": 82.28520814880426,
      "grad_norm": 0.26364055275917053,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 92900
    },
    {
      "epoch": 82.28963684676705,
      "grad_norm": 0.2509511709213257,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 92905
    },
    {
      "epoch": 82.29406554472985,
      "grad_norm": 0.2176615446805954,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 92910
    },
    {
      "epoch": 82.29849424269266,
      "grad_norm": 0.23476454615592957,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 92915
    },
    {
      "epoch": 82.30292294065545,
      "grad_norm": 0.24876607954502106,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 92920
    },
    {
      "epoch": 82.30735163861824,
      "grad_norm": 0.22676807641983032,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 92925
    },
    {
      "epoch": 82.31178033658105,
      "grad_norm": 0.24459494650363922,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 92930
    },
    {
      "epoch": 82.31620903454385,
      "grad_norm": 0.25257933139801025,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 92935
    },
    {
      "epoch": 82.32063773250664,
      "grad_norm": 0.2784808874130249,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 92940
    },
    {
      "epoch": 82.32506643046945,
      "grad_norm": 0.25922703742980957,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 92945
    },
    {
      "epoch": 82.32949512843224,
      "grad_norm": 0.2753784954547882,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 92950
    },
    {
      "epoch": 82.33392382639504,
      "grad_norm": 0.24441808462142944,
      "learning_rate": 1e-05,
      "loss": 0.9194,
      "step": 92955
    },
    {
      "epoch": 82.33835252435784,
      "grad_norm": 0.2293798178434372,
      "learning_rate": 1e-05,
      "loss": 0.9408,
      "step": 92960
    },
    {
      "epoch": 82.34278122232064,
      "grad_norm": 0.22352881729602814,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 92965
    },
    {
      "epoch": 82.34720992028343,
      "grad_norm": 0.1953272670507431,
      "learning_rate": 1e-05,
      "loss": 0.9234,
      "step": 92970
    },
    {
      "epoch": 82.35163861824624,
      "grad_norm": 0.2378494292497635,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 92975
    },
    {
      "epoch": 82.35606731620904,
      "grad_norm": 0.2331896424293518,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 92980
    },
    {
      "epoch": 82.36049601417183,
      "grad_norm": 0.2640392482280731,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 92985
    },
    {
      "epoch": 82.36492471213464,
      "grad_norm": 0.2554363012313843,
      "learning_rate": 1e-05,
      "loss": 0.9307,
      "step": 92990
    },
    {
      "epoch": 82.36935341009743,
      "grad_norm": 0.2527042031288147,
      "learning_rate": 1e-05,
      "loss": 0.9137,
      "step": 92995
    },
    {
      "epoch": 82.37378210806023,
      "grad_norm": 0.21290934085845947,
      "learning_rate": 1e-05,
      "loss": 0.9142,
      "step": 93000
    },
    {
      "epoch": 82.37821080602303,
      "grad_norm": 0.23322202265262604,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 93005
    },
    {
      "epoch": 82.38263950398583,
      "grad_norm": 0.20906901359558105,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 93010
    },
    {
      "epoch": 82.38706820194862,
      "grad_norm": 0.2555759847164154,
      "learning_rate": 1e-05,
      "loss": 0.9333,
      "step": 93015
    },
    {
      "epoch": 82.39149689991143,
      "grad_norm": 0.2137509137392044,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 93020
    },
    {
      "epoch": 82.39592559787422,
      "grad_norm": 0.2556888163089752,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 93025
    },
    {
      "epoch": 82.40035429583702,
      "grad_norm": 0.23329058289527893,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 93030
    },
    {
      "epoch": 82.40478299379983,
      "grad_norm": 0.2291184961795807,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 93035
    },
    {
      "epoch": 82.40921169176262,
      "grad_norm": 0.24417750537395477,
      "learning_rate": 1e-05,
      "loss": 0.9088,
      "step": 93040
    },
    {
      "epoch": 82.41364038972542,
      "grad_norm": 0.2221381962299347,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 93045
    },
    {
      "epoch": 82.41806908768822,
      "grad_norm": 0.213862344622612,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 93050
    },
    {
      "epoch": 82.42249778565102,
      "grad_norm": 0.20365607738494873,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 93055
    },
    {
      "epoch": 82.42692648361381,
      "grad_norm": 0.21590612828731537,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 93060
    },
    {
      "epoch": 82.43135518157662,
      "grad_norm": 0.23220698535442352,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 93065
    },
    {
      "epoch": 82.43578387953941,
      "grad_norm": 0.20819173753261566,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 93070
    },
    {
      "epoch": 82.44021257750221,
      "grad_norm": 0.21610568463802338,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 93075
    },
    {
      "epoch": 82.44464127546502,
      "grad_norm": 0.2327726185321808,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 93080
    },
    {
      "epoch": 82.44906997342781,
      "grad_norm": 0.2299305647611618,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 93085
    },
    {
      "epoch": 82.4534986713906,
      "grad_norm": 0.2253926545381546,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 93090
    },
    {
      "epoch": 82.45792736935341,
      "grad_norm": 0.21596622467041016,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 93095
    },
    {
      "epoch": 82.46235606731621,
      "grad_norm": 0.25354430079460144,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 93100
    },
    {
      "epoch": 82.466784765279,
      "grad_norm": 0.2254207283258438,
      "learning_rate": 1e-05,
      "loss": 1.0501,
      "step": 93105
    },
    {
      "epoch": 82.47121346324181,
      "grad_norm": 0.2142503261566162,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 93110
    },
    {
      "epoch": 82.4756421612046,
      "grad_norm": 0.2229224294424057,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 93115
    },
    {
      "epoch": 82.4800708591674,
      "grad_norm": 0.22385510802268982,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 93120
    },
    {
      "epoch": 82.4844995571302,
      "grad_norm": 0.25829800963401794,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 93125
    },
    {
      "epoch": 82.488928255093,
      "grad_norm": 0.32452359795570374,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 93130
    },
    {
      "epoch": 82.4933569530558,
      "grad_norm": 0.28407981991767883,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 93135
    },
    {
      "epoch": 82.4977856510186,
      "grad_norm": 0.23401610553264618,
      "learning_rate": 1e-05,
      "loss": 1.0103,
      "step": 93140
    },
    {
      "epoch": 82.5022143489814,
      "grad_norm": 0.22621333599090576,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 93145
    },
    {
      "epoch": 82.5066430469442,
      "grad_norm": 0.2164662480354309,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 93150
    },
    {
      "epoch": 82.511071744907,
      "grad_norm": 0.2804012894630432,
      "learning_rate": 1e-05,
      "loss": 0.9033,
      "step": 93155
    },
    {
      "epoch": 82.5155004428698,
      "grad_norm": 0.2612646222114563,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 93160
    },
    {
      "epoch": 82.5199291408326,
      "grad_norm": 0.24112051725387573,
      "learning_rate": 1e-05,
      "loss": 0.9298,
      "step": 93165
    },
    {
      "epoch": 82.5243578387954,
      "grad_norm": 0.24598760902881622,
      "learning_rate": 1e-05,
      "loss": 0.9122,
      "step": 93170
    },
    {
      "epoch": 82.52878653675819,
      "grad_norm": 0.25218650698661804,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 93175
    },
    {
      "epoch": 82.533215234721,
      "grad_norm": 0.2694810926914215,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 93180
    },
    {
      "epoch": 82.53764393268379,
      "grad_norm": 0.2419082224369049,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 93185
    },
    {
      "epoch": 82.54207263064659,
      "grad_norm": 0.2429601550102234,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 93190
    },
    {
      "epoch": 82.5465013286094,
      "grad_norm": 0.26009324193000793,
      "learning_rate": 1e-05,
      "loss": 0.9127,
      "step": 93195
    },
    {
      "epoch": 82.55093002657219,
      "grad_norm": 0.2199859917163849,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 93200
    },
    {
      "epoch": 82.55535872453498,
      "grad_norm": 0.2532244622707367,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 93205
    },
    {
      "epoch": 82.55978742249779,
      "grad_norm": 0.2014007419347763,
      "learning_rate": 1e-05,
      "loss": 0.9214,
      "step": 93210
    },
    {
      "epoch": 82.56421612046059,
      "grad_norm": 0.2239326685667038,
      "learning_rate": 1e-05,
      "loss": 0.9221,
      "step": 93215
    },
    {
      "epoch": 82.56864481842338,
      "grad_norm": 0.23541061580181122,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 93220
    },
    {
      "epoch": 82.57307351638619,
      "grad_norm": 0.22436967492103577,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 93225
    },
    {
      "epoch": 82.57750221434898,
      "grad_norm": 0.24005074799060822,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 93230
    },
    {
      "epoch": 82.58193091231178,
      "grad_norm": 0.2823718786239624,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 93235
    },
    {
      "epoch": 82.58635961027458,
      "grad_norm": 0.25429126620292664,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 93240
    },
    {
      "epoch": 82.59078830823738,
      "grad_norm": 0.28728052973747253,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 93245
    },
    {
      "epoch": 82.59521700620017,
      "grad_norm": 0.30380699038505554,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 93250
    },
    {
      "epoch": 82.59964570416298,
      "grad_norm": 0.23408547043800354,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 93255
    },
    {
      "epoch": 82.60407440212578,
      "grad_norm": 0.27851173281669617,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 93260
    },
    {
      "epoch": 82.60850310008857,
      "grad_norm": 0.2508356273174286,
      "learning_rate": 1e-05,
      "loss": 0.9166,
      "step": 93265
    },
    {
      "epoch": 82.61293179805138,
      "grad_norm": 0.22674061357975006,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 93270
    },
    {
      "epoch": 82.61736049601417,
      "grad_norm": 0.25120002031326294,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 93275
    },
    {
      "epoch": 82.62178919397697,
      "grad_norm": 0.1947123408317566,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 93280
    },
    {
      "epoch": 82.62621789193977,
      "grad_norm": 0.20119597017765045,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 93285
    },
    {
      "epoch": 82.63064658990257,
      "grad_norm": 0.26322153210639954,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 93290
    },
    {
      "epoch": 82.63507528786536,
      "grad_norm": 0.2715175747871399,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 93295
    },
    {
      "epoch": 82.63950398582817,
      "grad_norm": 0.2481713742017746,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 93300
    },
    {
      "epoch": 82.64393268379096,
      "grad_norm": 0.2643562853336334,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 93305
    },
    {
      "epoch": 82.64836138175376,
      "grad_norm": 0.2414681613445282,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 93310
    },
    {
      "epoch": 82.65279007971657,
      "grad_norm": 0.2412794977426529,
      "learning_rate": 1e-05,
      "loss": 0.9093,
      "step": 93315
    },
    {
      "epoch": 82.65721877767936,
      "grad_norm": 0.24661731719970703,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 93320
    },
    {
      "epoch": 82.66164747564216,
      "grad_norm": 0.23826812207698822,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 93325
    },
    {
      "epoch": 82.66607617360496,
      "grad_norm": 0.25095582008361816,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 93330
    },
    {
      "epoch": 82.67050487156776,
      "grad_norm": 0.2329203337430954,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 93335
    },
    {
      "epoch": 82.67493356953055,
      "grad_norm": 0.24053002893924713,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 93340
    },
    {
      "epoch": 82.67936226749336,
      "grad_norm": 0.2403680831193924,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 93345
    },
    {
      "epoch": 82.68379096545615,
      "grad_norm": 0.2531846761703491,
      "learning_rate": 1e-05,
      "loss": 0.9255,
      "step": 93350
    },
    {
      "epoch": 82.68821966341895,
      "grad_norm": 0.24978692829608917,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 93355
    },
    {
      "epoch": 82.69264836138176,
      "grad_norm": 0.20021769404411316,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 93360
    },
    {
      "epoch": 82.69707705934455,
      "grad_norm": 0.2459782212972641,
      "learning_rate": 1e-05,
      "loss": 0.9287,
      "step": 93365
    },
    {
      "epoch": 82.70150575730734,
      "grad_norm": 0.2232067734003067,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 93370
    },
    {
      "epoch": 82.70593445527015,
      "grad_norm": 0.19907306134700775,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 93375
    },
    {
      "epoch": 82.71036315323295,
      "grad_norm": 0.2596239447593689,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 93380
    },
    {
      "epoch": 82.71479185119574,
      "grad_norm": 0.22871831059455872,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 93385
    },
    {
      "epoch": 82.71922054915855,
      "grad_norm": 0.2362879067659378,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 93390
    },
    {
      "epoch": 82.72364924712134,
      "grad_norm": 0.24570442736148834,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 93395
    },
    {
      "epoch": 82.72807794508415,
      "grad_norm": 0.2295689433813095,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 93400
    },
    {
      "epoch": 82.73250664304695,
      "grad_norm": 0.2715066075325012,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 93405
    },
    {
      "epoch": 82.73693534100974,
      "grad_norm": 0.2211691439151764,
      "learning_rate": 1e-05,
      "loss": 1.0296,
      "step": 93410
    },
    {
      "epoch": 82.74136403897255,
      "grad_norm": 0.22630882263183594,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 93415
    },
    {
      "epoch": 82.74579273693534,
      "grad_norm": 0.20364220440387726,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 93420
    },
    {
      "epoch": 82.75022143489814,
      "grad_norm": 0.23906847834587097,
      "learning_rate": 1e-05,
      "loss": 0.9306,
      "step": 93425
    },
    {
      "epoch": 82.75465013286095,
      "grad_norm": 0.22127152979373932,
      "learning_rate": 1e-05,
      "loss": 0.9094,
      "step": 93430
    },
    {
      "epoch": 82.75907883082374,
      "grad_norm": 0.24635757505893707,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 93435
    },
    {
      "epoch": 82.76350752878653,
      "grad_norm": 0.23902717232704163,
      "learning_rate": 1e-05,
      "loss": 1.0056,
      "step": 93440
    },
    {
      "epoch": 82.76793622674934,
      "grad_norm": 0.23260647058486938,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 93445
    },
    {
      "epoch": 82.77236492471214,
      "grad_norm": 0.22645121812820435,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 93450
    },
    {
      "epoch": 82.77679362267493,
      "grad_norm": 0.25291764736175537,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 93455
    },
    {
      "epoch": 82.78122232063774,
      "grad_norm": 0.23050011694431305,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 93460
    },
    {
      "epoch": 82.78565101860053,
      "grad_norm": 0.24984940886497498,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 93465
    },
    {
      "epoch": 82.79007971656333,
      "grad_norm": 0.22395296394824982,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 93470
    },
    {
      "epoch": 82.79450841452613,
      "grad_norm": 0.2991333603858948,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 93475
    },
    {
      "epoch": 82.79893711248893,
      "grad_norm": 0.20330600440502167,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 93480
    },
    {
      "epoch": 82.80336581045172,
      "grad_norm": 0.26736918091773987,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 93485
    },
    {
      "epoch": 82.80779450841453,
      "grad_norm": 0.2058844417333603,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 93490
    },
    {
      "epoch": 82.81222320637733,
      "grad_norm": 0.27888891100883484,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 93495
    },
    {
      "epoch": 82.81665190434012,
      "grad_norm": 0.2519075274467468,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 93500
    },
    {
      "epoch": 82.82108060230293,
      "grad_norm": 0.25055989623069763,
      "learning_rate": 1e-05,
      "loss": 0.9121,
      "step": 93505
    },
    {
      "epoch": 82.82550930026572,
      "grad_norm": 0.23881596326828003,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 93510
    },
    {
      "epoch": 82.82993799822852,
      "grad_norm": 0.2660433053970337,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 93515
    },
    {
      "epoch": 82.83436669619132,
      "grad_norm": 0.2356128990650177,
      "learning_rate": 1e-05,
      "loss": 0.9004,
      "step": 93520
    },
    {
      "epoch": 82.83879539415412,
      "grad_norm": 0.27228793501853943,
      "learning_rate": 1e-05,
      "loss": 1.0454,
      "step": 93525
    },
    {
      "epoch": 82.84322409211691,
      "grad_norm": 0.27038222551345825,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 93530
    },
    {
      "epoch": 82.84765279007972,
      "grad_norm": 0.23006685078144073,
      "learning_rate": 1e-05,
      "loss": 0.9283,
      "step": 93535
    },
    {
      "epoch": 82.85208148804251,
      "grad_norm": 0.22513146698474884,
      "learning_rate": 1e-05,
      "loss": 0.8931,
      "step": 93540
    },
    {
      "epoch": 82.85651018600531,
      "grad_norm": 0.25537872314453125,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 93545
    },
    {
      "epoch": 82.86093888396812,
      "grad_norm": 0.216476172208786,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 93550
    },
    {
      "epoch": 82.86536758193091,
      "grad_norm": 0.29048407077789307,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 93555
    },
    {
      "epoch": 82.8697962798937,
      "grad_norm": 0.24756154417991638,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 93560
    },
    {
      "epoch": 82.87422497785651,
      "grad_norm": 0.26578348875045776,
      "learning_rate": 1e-05,
      "loss": 0.9311,
      "step": 93565
    },
    {
      "epoch": 82.87865367581931,
      "grad_norm": 0.20526890456676483,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 93570
    },
    {
      "epoch": 82.8830823737821,
      "grad_norm": 0.24398256838321686,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 93575
    },
    {
      "epoch": 82.88751107174491,
      "grad_norm": 0.240494504570961,
      "learning_rate": 1e-05,
      "loss": 1.0197,
      "step": 93580
    },
    {
      "epoch": 82.8919397697077,
      "grad_norm": 0.2334831953048706,
      "learning_rate": 1e-05,
      "loss": 0.9274,
      "step": 93585
    },
    {
      "epoch": 82.8963684676705,
      "grad_norm": 0.2473529726266861,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 93590
    },
    {
      "epoch": 82.9007971656333,
      "grad_norm": 0.2314014732837677,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 93595
    },
    {
      "epoch": 82.9052258635961,
      "grad_norm": 0.2326183021068573,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 93600
    },
    {
      "epoch": 82.9096545615589,
      "grad_norm": 0.23426249623298645,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 93605
    },
    {
      "epoch": 82.9140832595217,
      "grad_norm": 0.2544894516468048,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 93610
    },
    {
      "epoch": 82.9185119574845,
      "grad_norm": 0.23566783964633942,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 93615
    },
    {
      "epoch": 82.92294065544729,
      "grad_norm": 0.2213720679283142,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 93620
    },
    {
      "epoch": 82.9273693534101,
      "grad_norm": 0.22739677131175995,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 93625
    },
    {
      "epoch": 82.9317980513729,
      "grad_norm": 0.23768272995948792,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 93630
    },
    {
      "epoch": 82.9362267493357,
      "grad_norm": 0.274625301361084,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 93635
    },
    {
      "epoch": 82.9406554472985,
      "grad_norm": 0.21489015221595764,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 93640
    },
    {
      "epoch": 82.94508414526129,
      "grad_norm": 0.22081969678401947,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 93645
    },
    {
      "epoch": 82.9495128432241,
      "grad_norm": 0.2381548285484314,
      "learning_rate": 1e-05,
      "loss": 0.9147,
      "step": 93650
    },
    {
      "epoch": 82.9539415411869,
      "grad_norm": 0.25227224826812744,
      "learning_rate": 1e-05,
      "loss": 0.9287,
      "step": 93655
    },
    {
      "epoch": 82.95837023914969,
      "grad_norm": 0.23421943187713623,
      "learning_rate": 1e-05,
      "loss": 0.8847,
      "step": 93660
    },
    {
      "epoch": 82.9627989371125,
      "grad_norm": 0.23901128768920898,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 93665
    },
    {
      "epoch": 82.96722763507529,
      "grad_norm": 0.2223140150308609,
      "learning_rate": 1e-05,
      "loss": 0.9138,
      "step": 93670
    },
    {
      "epoch": 82.97165633303808,
      "grad_norm": 0.23102019727230072,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 93675
    },
    {
      "epoch": 82.97608503100089,
      "grad_norm": 0.23231561481952667,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 93680
    },
    {
      "epoch": 82.98051372896369,
      "grad_norm": 0.265850692987442,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 93685
    },
    {
      "epoch": 82.98494242692648,
      "grad_norm": 0.24320149421691895,
      "learning_rate": 1e-05,
      "loss": 0.9245,
      "step": 93690
    },
    {
      "epoch": 82.98937112488929,
      "grad_norm": 0.21519111096858978,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 93695
    },
    {
      "epoch": 82.99379982285208,
      "grad_norm": 0.25483617186546326,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 93700
    },
    {
      "epoch": 82.99822852081488,
      "grad_norm": 0.2806563973426819,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 93705
    },
    {
      "epoch": 83.00265721877768,
      "grad_norm": 0.2656966745853424,
      "learning_rate": 1e-05,
      "loss": 0.9243,
      "step": 93710
    },
    {
      "epoch": 83.00708591674048,
      "grad_norm": 0.20566770434379578,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 93715
    },
    {
      "epoch": 83.01151461470327,
      "grad_norm": 0.2320093810558319,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 93720
    },
    {
      "epoch": 83.01594331266608,
      "grad_norm": 0.24047455191612244,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 93725
    },
    {
      "epoch": 83.02037201062888,
      "grad_norm": 0.28820961713790894,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 93730
    },
    {
      "epoch": 83.02480070859167,
      "grad_norm": 0.2528837025165558,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 93735
    },
    {
      "epoch": 83.02922940655448,
      "grad_norm": 0.21906864643096924,
      "learning_rate": 1e-05,
      "loss": 0.9894,
      "step": 93740
    },
    {
      "epoch": 83.03365810451727,
      "grad_norm": 0.24371162056922913,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 93745
    },
    {
      "epoch": 83.03808680248007,
      "grad_norm": 0.25481653213500977,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 93750
    },
    {
      "epoch": 83.04251550044287,
      "grad_norm": 0.25346407294273376,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 93755
    },
    {
      "epoch": 83.04694419840567,
      "grad_norm": 0.22388868033885956,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 93760
    },
    {
      "epoch": 83.05137289636846,
      "grad_norm": 0.2326165735721588,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 93765
    },
    {
      "epoch": 83.05580159433127,
      "grad_norm": 0.23829403519630432,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 93770
    },
    {
      "epoch": 83.06023029229407,
      "grad_norm": 0.254862904548645,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 93775
    },
    {
      "epoch": 83.06465899025686,
      "grad_norm": 0.261135071516037,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 93780
    },
    {
      "epoch": 83.06908768821967,
      "grad_norm": 0.2482091784477234,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 93785
    },
    {
      "epoch": 83.07351638618246,
      "grad_norm": 0.24871626496315002,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 93790
    },
    {
      "epoch": 83.07794508414526,
      "grad_norm": 0.22859255969524384,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 93795
    },
    {
      "epoch": 83.08237378210806,
      "grad_norm": 0.2156241238117218,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 93800
    },
    {
      "epoch": 83.08680248007086,
      "grad_norm": 0.24705450236797333,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 93805
    },
    {
      "epoch": 83.09123117803365,
      "grad_norm": 0.253366619348526,
      "learning_rate": 1e-05,
      "loss": 1.0136,
      "step": 93810
    },
    {
      "epoch": 83.09565987599646,
      "grad_norm": 0.26688405871391296,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 93815
    },
    {
      "epoch": 83.10008857395925,
      "grad_norm": 0.24295754730701447,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 93820
    },
    {
      "epoch": 83.10451727192205,
      "grad_norm": 0.2648477852344513,
      "learning_rate": 1e-05,
      "loss": 0.917,
      "step": 93825
    },
    {
      "epoch": 83.10894596988486,
      "grad_norm": 0.24447417259216309,
      "learning_rate": 1e-05,
      "loss": 0.9163,
      "step": 93830
    },
    {
      "epoch": 83.11337466784765,
      "grad_norm": 0.23296914994716644,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 93835
    },
    {
      "epoch": 83.11780336581045,
      "grad_norm": 0.24273428320884705,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 93840
    },
    {
      "epoch": 83.12223206377325,
      "grad_norm": 0.24386471509933472,
      "learning_rate": 1e-05,
      "loss": 0.9255,
      "step": 93845
    },
    {
      "epoch": 83.12666076173605,
      "grad_norm": 0.2202557623386383,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 93850
    },
    {
      "epoch": 83.13108945969884,
      "grad_norm": 0.21652165055274963,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 93855
    },
    {
      "epoch": 83.13551815766165,
      "grad_norm": 0.2537614405155182,
      "learning_rate": 1e-05,
      "loss": 0.9244,
      "step": 93860
    },
    {
      "epoch": 83.13994685562444,
      "grad_norm": 0.21981799602508545,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 93865
    },
    {
      "epoch": 83.14437555358724,
      "grad_norm": 0.2640995383262634,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 93870
    },
    {
      "epoch": 83.14880425155005,
      "grad_norm": 0.2844601571559906,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 93875
    },
    {
      "epoch": 83.15323294951284,
      "grad_norm": 0.2336452305316925,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 93880
    },
    {
      "epoch": 83.15766164747565,
      "grad_norm": 0.23424705862998962,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 93885
    },
    {
      "epoch": 83.16209034543844,
      "grad_norm": 0.22237670421600342,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 93890
    },
    {
      "epoch": 83.16651904340124,
      "grad_norm": 0.256944477558136,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 93895
    },
    {
      "epoch": 83.17094774136405,
      "grad_norm": 0.27956265211105347,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 93900
    },
    {
      "epoch": 83.17537643932684,
      "grad_norm": 0.2760850787162781,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 93905
    },
    {
      "epoch": 83.17980513728963,
      "grad_norm": 0.24100790917873383,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 93910
    },
    {
      "epoch": 83.18423383525244,
      "grad_norm": 0.2767733335494995,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 93915
    },
    {
      "epoch": 83.18866253321524,
      "grad_norm": 0.2566892206668854,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 93920
    },
    {
      "epoch": 83.19309123117803,
      "grad_norm": 0.26374760270118713,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 93925
    },
    {
      "epoch": 83.19751992914084,
      "grad_norm": 0.25351467728614807,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 93930
    },
    {
      "epoch": 83.20194862710363,
      "grad_norm": 0.2797525227069855,
      "learning_rate": 1e-05,
      "loss": 0.9338,
      "step": 93935
    },
    {
      "epoch": 83.20637732506643,
      "grad_norm": 0.2216639667749405,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 93940
    },
    {
      "epoch": 83.21080602302924,
      "grad_norm": 0.2510758936405182,
      "learning_rate": 1e-05,
      "loss": 0.9177,
      "step": 93945
    },
    {
      "epoch": 83.21523472099203,
      "grad_norm": 0.2510535717010498,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 93950
    },
    {
      "epoch": 83.21966341895482,
      "grad_norm": 0.20791025459766388,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 93955
    },
    {
      "epoch": 83.22409211691763,
      "grad_norm": 0.2299337387084961,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 93960
    },
    {
      "epoch": 83.22852081488043,
      "grad_norm": 0.24523498117923737,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 93965
    },
    {
      "epoch": 83.23294951284322,
      "grad_norm": 0.2189374417066574,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 93970
    },
    {
      "epoch": 83.23737821080603,
      "grad_norm": 0.2211942970752716,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 93975
    },
    {
      "epoch": 83.24180690876882,
      "grad_norm": 0.27817586064338684,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 93980
    },
    {
      "epoch": 83.24623560673162,
      "grad_norm": 0.24003823101520538,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 93985
    },
    {
      "epoch": 83.25066430469442,
      "grad_norm": 0.2670159637928009,
      "learning_rate": 1e-05,
      "loss": 0.9194,
      "step": 93990
    },
    {
      "epoch": 83.25509300265722,
      "grad_norm": 0.24465975165367126,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 93995
    },
    {
      "epoch": 83.25952170062001,
      "grad_norm": 0.23428300023078918,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 94000
    },
    {
      "epoch": 83.26395039858282,
      "grad_norm": 0.26530906558036804,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 94005
    },
    {
      "epoch": 83.26837909654562,
      "grad_norm": 0.23970292508602142,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 94010
    },
    {
      "epoch": 83.27280779450841,
      "grad_norm": 0.23912519216537476,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 94015
    },
    {
      "epoch": 83.27723649247122,
      "grad_norm": 0.22111675143241882,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 94020
    },
    {
      "epoch": 83.28166519043401,
      "grad_norm": 0.2533017694950104,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 94025
    },
    {
      "epoch": 83.2860938883968,
      "grad_norm": 0.2356625348329544,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 94030
    },
    {
      "epoch": 83.29052258635961,
      "grad_norm": 0.2418094277381897,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 94035
    },
    {
      "epoch": 83.29495128432241,
      "grad_norm": 0.22982174158096313,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 94040
    },
    {
      "epoch": 83.2993799822852,
      "grad_norm": 0.23485109210014343,
      "learning_rate": 1e-05,
      "loss": 0.9082,
      "step": 94045
    },
    {
      "epoch": 83.30380868024801,
      "grad_norm": 0.24822627007961273,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 94050
    },
    {
      "epoch": 83.3082373782108,
      "grad_norm": 0.22795900702476501,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 94055
    },
    {
      "epoch": 83.3126660761736,
      "grad_norm": 0.24357326328754425,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 94060
    },
    {
      "epoch": 83.31709477413641,
      "grad_norm": 0.20118311047554016,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 94065
    },
    {
      "epoch": 83.3215234720992,
      "grad_norm": 0.23784692585468292,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 94070
    },
    {
      "epoch": 83.325952170062,
      "grad_norm": 0.2454223781824112,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 94075
    },
    {
      "epoch": 83.3303808680248,
      "grad_norm": 0.21749012172222137,
      "learning_rate": 1e-05,
      "loss": 0.896,
      "step": 94080
    },
    {
      "epoch": 83.3348095659876,
      "grad_norm": 0.2591759264469147,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 94085
    },
    {
      "epoch": 83.33923826395039,
      "grad_norm": 0.22938373684883118,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 94090
    },
    {
      "epoch": 83.3436669619132,
      "grad_norm": 0.23867255449295044,
      "learning_rate": 1e-05,
      "loss": 0.9295,
      "step": 94095
    },
    {
      "epoch": 83.348095659876,
      "grad_norm": 0.26197946071624756,
      "learning_rate": 1e-05,
      "loss": 0.9188,
      "step": 94100
    },
    {
      "epoch": 83.35252435783879,
      "grad_norm": 0.19717928767204285,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 94105
    },
    {
      "epoch": 83.3569530558016,
      "grad_norm": 0.243445485830307,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 94110
    },
    {
      "epoch": 83.36138175376439,
      "grad_norm": 0.2510448694229126,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 94115
    },
    {
      "epoch": 83.36581045172719,
      "grad_norm": 0.24246296286582947,
      "learning_rate": 1e-05,
      "loss": 0.9116,
      "step": 94120
    },
    {
      "epoch": 83.37023914969,
      "grad_norm": 0.23425616323947906,
      "learning_rate": 1e-05,
      "loss": 0.9252,
      "step": 94125
    },
    {
      "epoch": 83.37466784765279,
      "grad_norm": 0.27216097712516785,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 94130
    },
    {
      "epoch": 83.3790965456156,
      "grad_norm": 0.22336341440677643,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 94135
    },
    {
      "epoch": 83.38352524357839,
      "grad_norm": 0.23352289199829102,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 94140
    },
    {
      "epoch": 83.38795394154118,
      "grad_norm": 0.2487468123435974,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 94145
    },
    {
      "epoch": 83.39238263950399,
      "grad_norm": 0.23621197044849396,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 94150
    },
    {
      "epoch": 83.39681133746679,
      "grad_norm": 0.2535061240196228,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 94155
    },
    {
      "epoch": 83.40124003542958,
      "grad_norm": 0.21085460484027863,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 94160
    },
    {
      "epoch": 83.40566873339239,
      "grad_norm": 0.22095757722854614,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 94165
    },
    {
      "epoch": 83.41009743135518,
      "grad_norm": 0.23923417925834656,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 94170
    },
    {
      "epoch": 83.41452612931798,
      "grad_norm": 0.20255383849143982,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 94175
    },
    {
      "epoch": 83.41895482728079,
      "grad_norm": 0.2439180463552475,
      "learning_rate": 1e-05,
      "loss": 0.9317,
      "step": 94180
    },
    {
      "epoch": 83.42338352524358,
      "grad_norm": 0.21478743851184845,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 94185
    },
    {
      "epoch": 83.42781222320637,
      "grad_norm": 0.2719917297363281,
      "learning_rate": 1e-05,
      "loss": 1.0214,
      "step": 94190
    },
    {
      "epoch": 83.43224092116918,
      "grad_norm": 0.24778911471366882,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 94195
    },
    {
      "epoch": 83.43666961913198,
      "grad_norm": 0.23322463035583496,
      "learning_rate": 1e-05,
      "loss": 0.909,
      "step": 94200
    },
    {
      "epoch": 83.44109831709477,
      "grad_norm": 0.2065475881099701,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 94205
    },
    {
      "epoch": 83.44552701505758,
      "grad_norm": 0.31307050585746765,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 94210
    },
    {
      "epoch": 83.44995571302037,
      "grad_norm": 0.2713836431503296,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 94215
    },
    {
      "epoch": 83.45438441098317,
      "grad_norm": 0.2739722430706024,
      "learning_rate": 1e-05,
      "loss": 0.9119,
      "step": 94220
    },
    {
      "epoch": 83.45881310894598,
      "grad_norm": 0.2169104665517807,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 94225
    },
    {
      "epoch": 83.46324180690877,
      "grad_norm": 0.2187582552433014,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 94230
    },
    {
      "epoch": 83.46767050487156,
      "grad_norm": 0.2652219235897064,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 94235
    },
    {
      "epoch": 83.47209920283437,
      "grad_norm": 0.2385750412940979,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 94240
    },
    {
      "epoch": 83.47652790079717,
      "grad_norm": 0.23065853118896484,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 94245
    },
    {
      "epoch": 83.48095659875996,
      "grad_norm": 0.23643366992473602,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 94250
    },
    {
      "epoch": 83.48538529672277,
      "grad_norm": 0.21341638267040253,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 94255
    },
    {
      "epoch": 83.48981399468556,
      "grad_norm": 0.23498867452144623,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 94260
    },
    {
      "epoch": 83.49424269264836,
      "grad_norm": 0.2745988965034485,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 94265
    },
    {
      "epoch": 83.49867139061116,
      "grad_norm": 0.23934786021709442,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 94270
    },
    {
      "epoch": 83.50310008857396,
      "grad_norm": 0.2456146478652954,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 94275
    },
    {
      "epoch": 83.50752878653675,
      "grad_norm": 0.2496112436056137,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 94280
    },
    {
      "epoch": 83.51195748449956,
      "grad_norm": 0.23185831308364868,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 94285
    },
    {
      "epoch": 83.51638618246236,
      "grad_norm": 0.22603899240493774,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 94290
    },
    {
      "epoch": 83.52081488042515,
      "grad_norm": 0.2061179131269455,
      "learning_rate": 1e-05,
      "loss": 0.9108,
      "step": 94295
    },
    {
      "epoch": 83.52524357838796,
      "grad_norm": 0.22852736711502075,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 94300
    },
    {
      "epoch": 83.52967227635075,
      "grad_norm": 0.25372546911239624,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 94305
    },
    {
      "epoch": 83.53410097431355,
      "grad_norm": 0.25917112827301025,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 94310
    },
    {
      "epoch": 83.53852967227635,
      "grad_norm": 0.2512149512767792,
      "learning_rate": 1e-05,
      "loss": 1.0308,
      "step": 94315
    },
    {
      "epoch": 83.54295837023915,
      "grad_norm": 0.256344199180603,
      "learning_rate": 1e-05,
      "loss": 0.8993,
      "step": 94320
    },
    {
      "epoch": 83.54738706820194,
      "grad_norm": 0.23940466344356537,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 94325
    },
    {
      "epoch": 83.55181576616475,
      "grad_norm": 0.2747702896595001,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 94330
    },
    {
      "epoch": 83.55624446412754,
      "grad_norm": 0.2620646059513092,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 94335
    },
    {
      "epoch": 83.56067316209034,
      "grad_norm": 0.24633489549160004,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 94340
    },
    {
      "epoch": 83.56510186005315,
      "grad_norm": 0.2337767630815506,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 94345
    },
    {
      "epoch": 83.56953055801594,
      "grad_norm": 0.2599075734615326,
      "learning_rate": 1e-05,
      "loss": 0.9184,
      "step": 94350
    },
    {
      "epoch": 83.57395925597874,
      "grad_norm": 0.2704962491989136,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 94355
    },
    {
      "epoch": 83.57838795394154,
      "grad_norm": 0.25836795568466187,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 94360
    },
    {
      "epoch": 83.58281665190434,
      "grad_norm": 0.25076961517333984,
      "learning_rate": 1e-05,
      "loss": 1.0115,
      "step": 94365
    },
    {
      "epoch": 83.58724534986715,
      "grad_norm": 0.2208089828491211,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 94370
    },
    {
      "epoch": 83.59167404782994,
      "grad_norm": 0.23884716629981995,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 94375
    },
    {
      "epoch": 83.59610274579273,
      "grad_norm": 0.22609737515449524,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 94380
    },
    {
      "epoch": 83.60053144375554,
      "grad_norm": 0.2637893557548523,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 94385
    },
    {
      "epoch": 83.60496014171834,
      "grad_norm": 0.24326424300670624,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 94390
    },
    {
      "epoch": 83.60938883968113,
      "grad_norm": 0.23057204484939575,
      "learning_rate": 1e-05,
      "loss": 1.0339,
      "step": 94395
    },
    {
      "epoch": 83.61381753764394,
      "grad_norm": 0.23118382692337036,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 94400
    },
    {
      "epoch": 83.61824623560673,
      "grad_norm": 0.2289682924747467,
      "learning_rate": 1e-05,
      "loss": 1.02,
      "step": 94405
    },
    {
      "epoch": 83.62267493356953,
      "grad_norm": 0.21582165360450745,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 94410
    },
    {
      "epoch": 83.62710363153234,
      "grad_norm": 0.27156198024749756,
      "learning_rate": 1e-05,
      "loss": 0.9057,
      "step": 94415
    },
    {
      "epoch": 83.63153232949513,
      "grad_norm": 0.23715059459209442,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 94420
    },
    {
      "epoch": 83.63596102745792,
      "grad_norm": 0.24554996192455292,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 94425
    },
    {
      "epoch": 83.64038972542073,
      "grad_norm": 0.19621340930461884,
      "learning_rate": 1e-05,
      "loss": 0.9038,
      "step": 94430
    },
    {
      "epoch": 83.64481842338353,
      "grad_norm": 0.22292792797088623,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 94435
    },
    {
      "epoch": 83.64924712134632,
      "grad_norm": 0.21607856452465057,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 94440
    },
    {
      "epoch": 83.65367581930913,
      "grad_norm": 0.24264638125896454,
      "learning_rate": 1e-05,
      "loss": 0.8949,
      "step": 94445
    },
    {
      "epoch": 83.65810451727192,
      "grad_norm": 0.2808072865009308,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 94450
    },
    {
      "epoch": 83.66253321523472,
      "grad_norm": 0.2280391901731491,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 94455
    },
    {
      "epoch": 83.66696191319753,
      "grad_norm": 0.2014925479888916,
      "learning_rate": 1e-05,
      "loss": 0.9296,
      "step": 94460
    },
    {
      "epoch": 83.67139061116032,
      "grad_norm": 0.26185154914855957,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 94465
    },
    {
      "epoch": 83.67581930912311,
      "grad_norm": 0.29537293314933777,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 94470
    },
    {
      "epoch": 83.68024800708592,
      "grad_norm": 0.2430686205625534,
      "learning_rate": 1e-05,
      "loss": 0.9215,
      "step": 94475
    },
    {
      "epoch": 83.68467670504872,
      "grad_norm": 0.25594595074653625,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 94480
    },
    {
      "epoch": 83.68910540301151,
      "grad_norm": 0.28057706356048584,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 94485
    },
    {
      "epoch": 83.69353410097432,
      "grad_norm": 0.26596999168395996,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 94490
    },
    {
      "epoch": 83.69796279893711,
      "grad_norm": 0.21645207703113556,
      "learning_rate": 1e-05,
      "loss": 0.9324,
      "step": 94495
    },
    {
      "epoch": 83.7023914968999,
      "grad_norm": 0.2167506068944931,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 94500
    },
    {
      "epoch": 83.70682019486271,
      "grad_norm": 0.26335132122039795,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 94505
    },
    {
      "epoch": 83.71124889282551,
      "grad_norm": 0.24410569667816162,
      "learning_rate": 1e-05,
      "loss": 0.9318,
      "step": 94510
    },
    {
      "epoch": 83.7156775907883,
      "grad_norm": 0.27268674969673157,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 94515
    },
    {
      "epoch": 83.72010628875111,
      "grad_norm": 0.26459160447120667,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 94520
    },
    {
      "epoch": 83.7245349867139,
      "grad_norm": 0.26562801003456116,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 94525
    },
    {
      "epoch": 83.7289636846767,
      "grad_norm": 0.22361956536769867,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 94530
    },
    {
      "epoch": 83.73339238263951,
      "grad_norm": 0.2498777210712433,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 94535
    },
    {
      "epoch": 83.7378210806023,
      "grad_norm": 0.23832964897155762,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 94540
    },
    {
      "epoch": 83.7422497785651,
      "grad_norm": 0.22808273136615753,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 94545
    },
    {
      "epoch": 83.7466784765279,
      "grad_norm": 0.22089675068855286,
      "learning_rate": 1e-05,
      "loss": 0.9192,
      "step": 94550
    },
    {
      "epoch": 83.7511071744907,
      "grad_norm": 0.21950238943099976,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 94555
    },
    {
      "epoch": 83.75553587245349,
      "grad_norm": 0.23859283328056335,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 94560
    },
    {
      "epoch": 83.7599645704163,
      "grad_norm": 0.26677677035331726,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 94565
    },
    {
      "epoch": 83.7643932683791,
      "grad_norm": 0.2582817077636719,
      "learning_rate": 1e-05,
      "loss": 1.0173,
      "step": 94570
    },
    {
      "epoch": 83.76882196634189,
      "grad_norm": 0.2499431073665619,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 94575
    },
    {
      "epoch": 83.7732506643047,
      "grad_norm": 0.26266103982925415,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 94580
    },
    {
      "epoch": 83.77767936226749,
      "grad_norm": 0.2051333785057068,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 94585
    },
    {
      "epoch": 83.78210806023029,
      "grad_norm": 0.25427547097206116,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 94590
    },
    {
      "epoch": 83.7865367581931,
      "grad_norm": 0.2582823634147644,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 94595
    },
    {
      "epoch": 83.79096545615589,
      "grad_norm": 0.2203664630651474,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 94600
    },
    {
      "epoch": 83.79539415411868,
      "grad_norm": 0.25477251410484314,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 94605
    },
    {
      "epoch": 83.79982285208149,
      "grad_norm": 0.23443010449409485,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 94610
    },
    {
      "epoch": 83.80425155004428,
      "grad_norm": 0.2462361454963684,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 94615
    },
    {
      "epoch": 83.8086802480071,
      "grad_norm": 0.2674030065536499,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 94620
    },
    {
      "epoch": 83.81310894596989,
      "grad_norm": 0.28179067373275757,
      "learning_rate": 1e-05,
      "loss": 0.8903,
      "step": 94625
    },
    {
      "epoch": 83.81753764393268,
      "grad_norm": 0.2698670029640198,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 94630
    },
    {
      "epoch": 83.82196634189549,
      "grad_norm": 0.2474033683538437,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 94635
    },
    {
      "epoch": 83.82639503985828,
      "grad_norm": 0.2378651648759842,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 94640
    },
    {
      "epoch": 83.83082373782108,
      "grad_norm": 0.23646245896816254,
      "learning_rate": 1e-05,
      "loss": 0.8945,
      "step": 94645
    },
    {
      "epoch": 83.83525243578389,
      "grad_norm": 0.2506885230541229,
      "learning_rate": 1e-05,
      "loss": 0.9129,
      "step": 94650
    },
    {
      "epoch": 83.83968113374668,
      "grad_norm": 0.21488212049007416,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 94655
    },
    {
      "epoch": 83.84410983170947,
      "grad_norm": 0.25113147497177124,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 94660
    },
    {
      "epoch": 83.84853852967228,
      "grad_norm": 0.24635741114616394,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 94665
    },
    {
      "epoch": 83.85296722763508,
      "grad_norm": 0.21752731502056122,
      "learning_rate": 1e-05,
      "loss": 0.9149,
      "step": 94670
    },
    {
      "epoch": 83.85739592559787,
      "grad_norm": 0.24392277002334595,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 94675
    },
    {
      "epoch": 83.86182462356068,
      "grad_norm": 0.22871124744415283,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 94680
    },
    {
      "epoch": 83.86625332152347,
      "grad_norm": 0.23201842606067657,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 94685
    },
    {
      "epoch": 83.87068201948627,
      "grad_norm": 0.21385346353054047,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 94690
    },
    {
      "epoch": 83.87511071744908,
      "grad_norm": 0.22885695099830627,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 94695
    },
    {
      "epoch": 83.87953941541187,
      "grad_norm": 0.25592973828315735,
      "learning_rate": 1e-05,
      "loss": 0.9357,
      "step": 94700
    },
    {
      "epoch": 83.88396811337466,
      "grad_norm": 0.22758303582668304,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 94705
    },
    {
      "epoch": 83.88839681133747,
      "grad_norm": 0.23584698140621185,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 94710
    },
    {
      "epoch": 83.89282550930027,
      "grad_norm": 0.24014510214328766,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 94715
    },
    {
      "epoch": 83.89725420726306,
      "grad_norm": 0.2617188096046448,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 94720
    },
    {
      "epoch": 83.90168290522587,
      "grad_norm": 0.27377843856811523,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 94725
    },
    {
      "epoch": 83.90611160318866,
      "grad_norm": 0.2842259407043457,
      "learning_rate": 1e-05,
      "loss": 0.9074,
      "step": 94730
    },
    {
      "epoch": 83.91054030115146,
      "grad_norm": 0.23000213503837585,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 94735
    },
    {
      "epoch": 83.91496899911427,
      "grad_norm": 0.2568821609020233,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 94740
    },
    {
      "epoch": 83.91939769707706,
      "grad_norm": 0.2215203195810318,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 94745
    },
    {
      "epoch": 83.92382639503985,
      "grad_norm": 0.2594284415245056,
      "learning_rate": 1e-05,
      "loss": 0.9153,
      "step": 94750
    },
    {
      "epoch": 83.92825509300266,
      "grad_norm": 0.23628057539463043,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 94755
    },
    {
      "epoch": 83.93268379096546,
      "grad_norm": 0.24657033383846283,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 94760
    },
    {
      "epoch": 83.93711248892825,
      "grad_norm": 0.21617963910102844,
      "learning_rate": 1e-05,
      "loss": 0.9157,
      "step": 94765
    },
    {
      "epoch": 83.94154118689106,
      "grad_norm": 0.23649059236049652,
      "learning_rate": 1e-05,
      "loss": 0.9187,
      "step": 94770
    },
    {
      "epoch": 83.94596988485385,
      "grad_norm": 0.2079278528690338,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 94775
    },
    {
      "epoch": 83.95039858281665,
      "grad_norm": 0.2660117745399475,
      "learning_rate": 1e-05,
      "loss": 0.8715,
      "step": 94780
    },
    {
      "epoch": 83.95482728077945,
      "grad_norm": 0.23003382980823517,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 94785
    },
    {
      "epoch": 83.95925597874225,
      "grad_norm": 0.24406704306602478,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 94790
    },
    {
      "epoch": 83.96368467670504,
      "grad_norm": 0.2565998435020447,
      "learning_rate": 1e-05,
      "loss": 0.9286,
      "step": 94795
    },
    {
      "epoch": 83.96811337466785,
      "grad_norm": 0.2575805187225342,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 94800
    },
    {
      "epoch": 83.97254207263065,
      "grad_norm": 0.23764467239379883,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 94805
    },
    {
      "epoch": 83.97697077059344,
      "grad_norm": 0.23078009486198425,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 94810
    },
    {
      "epoch": 83.98139946855625,
      "grad_norm": 0.21822692453861237,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 94815
    },
    {
      "epoch": 83.98582816651904,
      "grad_norm": 0.2532300651073456,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 94820
    },
    {
      "epoch": 83.99025686448184,
      "grad_norm": 0.30720821022987366,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 94825
    },
    {
      "epoch": 83.99468556244464,
      "grad_norm": 0.22327812016010284,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 94830
    },
    {
      "epoch": 83.99911426040744,
      "grad_norm": 0.23577070236206055,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 94835
    },
    {
      "epoch": 84.00354295837023,
      "grad_norm": 0.2950497567653656,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 94840
    },
    {
      "epoch": 84.00797165633304,
      "grad_norm": 0.2588575780391693,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 94845
    },
    {
      "epoch": 84.01240035429583,
      "grad_norm": 0.28334105014801025,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 94850
    },
    {
      "epoch": 84.01682905225863,
      "grad_norm": 0.24211524426937103,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 94855
    },
    {
      "epoch": 84.02125775022144,
      "grad_norm": 0.24798712134361267,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 94860
    },
    {
      "epoch": 84.02568644818423,
      "grad_norm": 0.24365146458148956,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 94865
    },
    {
      "epoch": 84.03011514614704,
      "grad_norm": 0.2494594305753708,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 94870
    },
    {
      "epoch": 84.03454384410983,
      "grad_norm": 0.2192710041999817,
      "learning_rate": 1e-05,
      "loss": 0.9408,
      "step": 94875
    },
    {
      "epoch": 84.03897254207263,
      "grad_norm": 0.2324598878622055,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 94880
    },
    {
      "epoch": 84.04340124003544,
      "grad_norm": 0.27968695759773254,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 94885
    },
    {
      "epoch": 84.04782993799823,
      "grad_norm": 0.22181417047977448,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 94890
    },
    {
      "epoch": 84.05225863596102,
      "grad_norm": 0.24373731017112732,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 94895
    },
    {
      "epoch": 84.05668733392383,
      "grad_norm": 0.2466449737548828,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 94900
    },
    {
      "epoch": 84.06111603188663,
      "grad_norm": 0.2175285816192627,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 94905
    },
    {
      "epoch": 84.06554472984942,
      "grad_norm": 0.2768596410751343,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 94910
    },
    {
      "epoch": 84.06997342781223,
      "grad_norm": 0.25873762369155884,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 94915
    },
    {
      "epoch": 84.07440212577502,
      "grad_norm": 0.2660638391971588,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 94920
    },
    {
      "epoch": 84.07883082373782,
      "grad_norm": 0.22319315373897552,
      "learning_rate": 1e-05,
      "loss": 0.9054,
      "step": 94925
    },
    {
      "epoch": 84.08325952170063,
      "grad_norm": 0.21766650676727295,
      "learning_rate": 1e-05,
      "loss": 0.9298,
      "step": 94930
    },
    {
      "epoch": 84.08768821966342,
      "grad_norm": 0.2293868064880371,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 94935
    },
    {
      "epoch": 84.09211691762621,
      "grad_norm": 0.27747952938079834,
      "learning_rate": 1e-05,
      "loss": 0.9053,
      "step": 94940
    },
    {
      "epoch": 84.09654561558902,
      "grad_norm": 0.23462530970573425,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 94945
    },
    {
      "epoch": 84.10097431355182,
      "grad_norm": 0.2313680350780487,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 94950
    },
    {
      "epoch": 84.10540301151461,
      "grad_norm": 0.21589159965515137,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 94955
    },
    {
      "epoch": 84.10983170947742,
      "grad_norm": 0.2361680269241333,
      "learning_rate": 1e-05,
      "loss": 0.9753,
      "step": 94960
    },
    {
      "epoch": 84.11426040744021,
      "grad_norm": 0.26228246092796326,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 94965
    },
    {
      "epoch": 84.118689105403,
      "grad_norm": 0.21194227039813995,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 94970
    },
    {
      "epoch": 84.12311780336582,
      "grad_norm": 0.241623193025589,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 94975
    },
    {
      "epoch": 84.12754650132861,
      "grad_norm": 0.2385462522506714,
      "learning_rate": 1e-05,
      "loss": 0.907,
      "step": 94980
    },
    {
      "epoch": 84.1319751992914,
      "grad_norm": 0.22319334745407104,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 94985
    },
    {
      "epoch": 84.13640389725421,
      "grad_norm": 0.21094316244125366,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 94990
    },
    {
      "epoch": 84.140832595217,
      "grad_norm": 0.21485401690006256,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 94995
    },
    {
      "epoch": 84.1452612931798,
      "grad_norm": 0.22340507805347443,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 95000
    },
    {
      "epoch": 84.14968999114261,
      "grad_norm": 0.22193661332130432,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 95005
    },
    {
      "epoch": 84.1541186891054,
      "grad_norm": 0.191301167011261,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 95010
    },
    {
      "epoch": 84.1585473870682,
      "grad_norm": 0.25105011463165283,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 95015
    },
    {
      "epoch": 84.162976085031,
      "grad_norm": 0.2406686395406723,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 95020
    },
    {
      "epoch": 84.1674047829938,
      "grad_norm": 0.245299831032753,
      "learning_rate": 1e-05,
      "loss": 0.8933,
      "step": 95025
    },
    {
      "epoch": 84.1718334809566,
      "grad_norm": 0.22139288485050201,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 95030
    },
    {
      "epoch": 84.1762621789194,
      "grad_norm": 0.2491191178560257,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 95035
    },
    {
      "epoch": 84.1806908768822,
      "grad_norm": 0.2570236623287201,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 95040
    },
    {
      "epoch": 84.18511957484499,
      "grad_norm": 0.22704187035560608,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 95045
    },
    {
      "epoch": 84.1895482728078,
      "grad_norm": 0.2254067063331604,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 95050
    },
    {
      "epoch": 84.19397697077059,
      "grad_norm": 0.22757205367088318,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 95055
    },
    {
      "epoch": 84.19840566873339,
      "grad_norm": 0.22878535091876984,
      "learning_rate": 1e-05,
      "loss": 0.9237,
      "step": 95060
    },
    {
      "epoch": 84.2028343666962,
      "grad_norm": 0.24961993098258972,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 95065
    },
    {
      "epoch": 84.20726306465899,
      "grad_norm": 0.2246078997850418,
      "learning_rate": 1e-05,
      "loss": 0.9265,
      "step": 95070
    },
    {
      "epoch": 84.21169176262178,
      "grad_norm": 0.2078251987695694,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 95075
    },
    {
      "epoch": 84.21612046058459,
      "grad_norm": 0.2782115936279297,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 95080
    },
    {
      "epoch": 84.22054915854739,
      "grad_norm": 0.24469760060310364,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 95085
    },
    {
      "epoch": 84.22497785651018,
      "grad_norm": 0.26080989837646484,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 95090
    },
    {
      "epoch": 84.22940655447299,
      "grad_norm": 0.2590102553367615,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 95095
    },
    {
      "epoch": 84.23383525243578,
      "grad_norm": 0.18541674315929413,
      "learning_rate": 1e-05,
      "loss": 0.9878,
      "step": 95100
    },
    {
      "epoch": 84.23826395039858,
      "grad_norm": 0.20707251131534576,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 95105
    },
    {
      "epoch": 84.24269264836138,
      "grad_norm": 0.24196644127368927,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 95110
    },
    {
      "epoch": 84.24712134632418,
      "grad_norm": 0.2021612972021103,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 95115
    },
    {
      "epoch": 84.25155004428699,
      "grad_norm": 0.2330324798822403,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 95120
    },
    {
      "epoch": 84.25597874224978,
      "grad_norm": 0.2529261112213135,
      "learning_rate": 1e-05,
      "loss": 0.925,
      "step": 95125
    },
    {
      "epoch": 84.26040744021257,
      "grad_norm": 0.24427860975265503,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 95130
    },
    {
      "epoch": 84.26483613817538,
      "grad_norm": 0.21987900137901306,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 95135
    },
    {
      "epoch": 84.26926483613818,
      "grad_norm": 0.25686851143836975,
      "learning_rate": 1e-05,
      "loss": 0.8946,
      "step": 95140
    },
    {
      "epoch": 84.27369353410097,
      "grad_norm": 0.22783900797367096,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 95145
    },
    {
      "epoch": 84.27812223206378,
      "grad_norm": 0.2344811111688614,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 95150
    },
    {
      "epoch": 84.28255093002657,
      "grad_norm": 0.22524075210094452,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 95155
    },
    {
      "epoch": 84.28697962798937,
      "grad_norm": 0.24430370330810547,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 95160
    },
    {
      "epoch": 84.29140832595218,
      "grad_norm": 0.22766785323619843,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 95165
    },
    {
      "epoch": 84.29583702391497,
      "grad_norm": 0.20383374392986298,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 95170
    },
    {
      "epoch": 84.30026572187776,
      "grad_norm": 0.2632075250148773,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 95175
    },
    {
      "epoch": 84.30469441984057,
      "grad_norm": 0.2478143870830536,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 95180
    },
    {
      "epoch": 84.30912311780337,
      "grad_norm": 0.23032645881175995,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 95185
    },
    {
      "epoch": 84.31355181576616,
      "grad_norm": 0.2212352305650711,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 95190
    },
    {
      "epoch": 84.31798051372897,
      "grad_norm": 0.2680801451206207,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 95195
    },
    {
      "epoch": 84.32240921169176,
      "grad_norm": 0.22448107600212097,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 95200
    },
    {
      "epoch": 84.32683790965456,
      "grad_norm": 0.24958157539367676,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 95205
    },
    {
      "epoch": 84.33126660761737,
      "grad_norm": 0.22682936489582062,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 95210
    },
    {
      "epoch": 84.33569530558016,
      "grad_norm": 0.25163504481315613,
      "learning_rate": 1e-05,
      "loss": 0.906,
      "step": 95215
    },
    {
      "epoch": 84.34012400354295,
      "grad_norm": 0.232279434800148,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 95220
    },
    {
      "epoch": 84.34455270150576,
      "grad_norm": 0.2613610625267029,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 95225
    },
    {
      "epoch": 84.34898139946856,
      "grad_norm": 0.26100119948387146,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 95230
    },
    {
      "epoch": 84.35341009743135,
      "grad_norm": 0.24040307104587555,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 95235
    },
    {
      "epoch": 84.35783879539416,
      "grad_norm": 0.21100692451000214,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 95240
    },
    {
      "epoch": 84.36226749335695,
      "grad_norm": 0.2626730501651764,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 95245
    },
    {
      "epoch": 84.36669619131975,
      "grad_norm": 0.2319963276386261,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 95250
    },
    {
      "epoch": 84.37112488928256,
      "grad_norm": 0.28629860281944275,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 95255
    },
    {
      "epoch": 84.37555358724535,
      "grad_norm": 0.23362088203430176,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 95260
    },
    {
      "epoch": 84.37998228520814,
      "grad_norm": 0.22681240737438202,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 95265
    },
    {
      "epoch": 84.38441098317095,
      "grad_norm": 0.23317331075668335,
      "learning_rate": 1e-05,
      "loss": 0.9432,
      "step": 95270
    },
    {
      "epoch": 84.38883968113375,
      "grad_norm": 0.27958372235298157,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 95275
    },
    {
      "epoch": 84.39326837909654,
      "grad_norm": 0.2536323368549347,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 95280
    },
    {
      "epoch": 84.39769707705935,
      "grad_norm": 0.2526127099990845,
      "learning_rate": 1e-05,
      "loss": 0.9188,
      "step": 95285
    },
    {
      "epoch": 84.40212577502214,
      "grad_norm": 0.2667582035064697,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 95290
    },
    {
      "epoch": 84.40655447298494,
      "grad_norm": 0.2597736716270447,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 95295
    },
    {
      "epoch": 84.41098317094774,
      "grad_norm": 0.2338431179523468,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 95300
    },
    {
      "epoch": 84.41541186891054,
      "grad_norm": 0.25917288661003113,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 95305
    },
    {
      "epoch": 84.41984056687333,
      "grad_norm": 0.23737460374832153,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 95310
    },
    {
      "epoch": 84.42426926483614,
      "grad_norm": 0.2870638370513916,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 95315
    },
    {
      "epoch": 84.42869796279894,
      "grad_norm": 0.21002182364463806,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 95320
    },
    {
      "epoch": 84.43312666076173,
      "grad_norm": 0.20469854772090912,
      "learning_rate": 1e-05,
      "loss": 0.9184,
      "step": 95325
    },
    {
      "epoch": 84.43755535872454,
      "grad_norm": 0.21518981456756592,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 95330
    },
    {
      "epoch": 84.44198405668733,
      "grad_norm": 0.20124612748622894,
      "learning_rate": 1e-05,
      "loss": 0.9052,
      "step": 95335
    },
    {
      "epoch": 84.44641275465013,
      "grad_norm": 0.21300576627254486,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 95340
    },
    {
      "epoch": 84.45084145261293,
      "grad_norm": 0.248631551861763,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 95345
    },
    {
      "epoch": 84.45527015057573,
      "grad_norm": 0.27155691385269165,
      "learning_rate": 1e-05,
      "loss": 1.022,
      "step": 95350
    },
    {
      "epoch": 84.45969884853854,
      "grad_norm": 0.217047780752182,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 95355
    },
    {
      "epoch": 84.46412754650133,
      "grad_norm": 0.2551300525665283,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 95360
    },
    {
      "epoch": 84.46855624446412,
      "grad_norm": 0.21841205656528473,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 95365
    },
    {
      "epoch": 84.47298494242693,
      "grad_norm": 0.22561630606651306,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 95370
    },
    {
      "epoch": 84.47741364038973,
      "grad_norm": 0.23209480941295624,
      "learning_rate": 1e-05,
      "loss": 0.9179,
      "step": 95375
    },
    {
      "epoch": 84.48184233835252,
      "grad_norm": 0.24146901071071625,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 95380
    },
    {
      "epoch": 84.48627103631533,
      "grad_norm": 0.23416675627231598,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 95385
    },
    {
      "epoch": 84.49069973427812,
      "grad_norm": 0.2584945559501648,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 95390
    },
    {
      "epoch": 84.49512843224092,
      "grad_norm": 0.2838519215583801,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 95395
    },
    {
      "epoch": 84.49955713020373,
      "grad_norm": 0.23502430319786072,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 95400
    },
    {
      "epoch": 84.50398582816652,
      "grad_norm": 0.242192804813385,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 95405
    },
    {
      "epoch": 84.50841452612931,
      "grad_norm": 0.2141350954771042,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 95410
    },
    {
      "epoch": 84.51284322409212,
      "grad_norm": 0.23970328271389008,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 95415
    },
    {
      "epoch": 84.51727192205492,
      "grad_norm": 0.25437837839126587,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 95420
    },
    {
      "epoch": 84.52170062001771,
      "grad_norm": 0.24400384724140167,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 95425
    },
    {
      "epoch": 84.52612931798052,
      "grad_norm": 0.22580833733081818,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 95430
    },
    {
      "epoch": 84.53055801594331,
      "grad_norm": 0.2961694598197937,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 95435
    },
    {
      "epoch": 84.53498671390611,
      "grad_norm": 0.25675418972969055,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 95440
    },
    {
      "epoch": 84.53941541186892,
      "grad_norm": 0.2608996331691742,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 95445
    },
    {
      "epoch": 84.54384410983171,
      "grad_norm": 0.2342691272497177,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 95450
    },
    {
      "epoch": 84.5482728077945,
      "grad_norm": 0.26692405343055725,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 95455
    },
    {
      "epoch": 84.55270150575731,
      "grad_norm": 0.2560400664806366,
      "learning_rate": 1e-05,
      "loss": 0.9066,
      "step": 95460
    },
    {
      "epoch": 84.5571302037201,
      "grad_norm": 0.19967405498027802,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 95465
    },
    {
      "epoch": 84.5615589016829,
      "grad_norm": 0.2428506314754486,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 95470
    },
    {
      "epoch": 84.56598759964571,
      "grad_norm": 0.2687271237373352,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 95475
    },
    {
      "epoch": 84.5704162976085,
      "grad_norm": 0.27449697256088257,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 95480
    },
    {
      "epoch": 84.5748449955713,
      "grad_norm": 0.23221999406814575,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 95485
    },
    {
      "epoch": 84.5792736935341,
      "grad_norm": 0.27246710658073425,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 95490
    },
    {
      "epoch": 84.5837023914969,
      "grad_norm": 0.2390279322862625,
      "learning_rate": 1e-05,
      "loss": 0.9048,
      "step": 95495
    },
    {
      "epoch": 84.5881310894597,
      "grad_norm": 0.2245926409959793,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 95500
    },
    {
      "epoch": 84.5925597874225,
      "grad_norm": 0.23420579731464386,
      "learning_rate": 1e-05,
      "loss": 0.9333,
      "step": 95505
    },
    {
      "epoch": 84.5969884853853,
      "grad_norm": 0.2726210653781891,
      "learning_rate": 1e-05,
      "loss": 0.93,
      "step": 95510
    },
    {
      "epoch": 84.60141718334809,
      "grad_norm": 0.2940177321434021,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 95515
    },
    {
      "epoch": 84.6058458813109,
      "grad_norm": 0.2457304149866104,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 95520
    },
    {
      "epoch": 84.61027457927369,
      "grad_norm": 0.2744705379009247,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 95525
    },
    {
      "epoch": 84.61470327723649,
      "grad_norm": 0.29199880361557007,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 95530
    },
    {
      "epoch": 84.6191319751993,
      "grad_norm": 0.2516320049762726,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 95535
    },
    {
      "epoch": 84.62356067316209,
      "grad_norm": 0.21507993340492249,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 95540
    },
    {
      "epoch": 84.62798937112488,
      "grad_norm": 0.22222648561000824,
      "learning_rate": 1e-05,
      "loss": 0.8921,
      "step": 95545
    },
    {
      "epoch": 84.63241806908769,
      "grad_norm": 0.2148786187171936,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 95550
    },
    {
      "epoch": 84.63684676705049,
      "grad_norm": 0.20781093835830688,
      "learning_rate": 1e-05,
      "loss": 0.9093,
      "step": 95555
    },
    {
      "epoch": 84.64127546501328,
      "grad_norm": 0.22792086005210876,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 95560
    },
    {
      "epoch": 84.64570416297609,
      "grad_norm": 0.23570506274700165,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 95565
    },
    {
      "epoch": 84.65013286093888,
      "grad_norm": 0.2219143658876419,
      "learning_rate": 1e-05,
      "loss": 0.9119,
      "step": 95570
    },
    {
      "epoch": 84.65456155890168,
      "grad_norm": 0.22343456745147705,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 95575
    },
    {
      "epoch": 84.65899025686448,
      "grad_norm": 0.2881627678871155,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 95580
    },
    {
      "epoch": 84.66341895482728,
      "grad_norm": 0.2373683899641037,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 95585
    },
    {
      "epoch": 84.66784765279007,
      "grad_norm": 0.24828515946865082,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 95590
    },
    {
      "epoch": 84.67227635075288,
      "grad_norm": 0.2688041925430298,
      "learning_rate": 1e-05,
      "loss": 0.9231,
      "step": 95595
    },
    {
      "epoch": 84.67670504871568,
      "grad_norm": 0.24422608315944672,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 95600
    },
    {
      "epoch": 84.68113374667848,
      "grad_norm": 0.21485407650470734,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 95605
    },
    {
      "epoch": 84.68556244464128,
      "grad_norm": 0.2476443201303482,
      "learning_rate": 1e-05,
      "loss": 0.9179,
      "step": 95610
    },
    {
      "epoch": 84.68999114260407,
      "grad_norm": 0.2540789246559143,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 95615
    },
    {
      "epoch": 84.69441984056688,
      "grad_norm": 0.24090464413166046,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 95620
    },
    {
      "epoch": 84.69884853852967,
      "grad_norm": 0.22515863180160522,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 95625
    },
    {
      "epoch": 84.70327723649247,
      "grad_norm": 0.22733783721923828,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 95630
    },
    {
      "epoch": 84.70770593445528,
      "grad_norm": 0.23291243612766266,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 95635
    },
    {
      "epoch": 84.71213463241807,
      "grad_norm": 0.22671736776828766,
      "learning_rate": 1e-05,
      "loss": 0.9795,
      "step": 95640
    },
    {
      "epoch": 84.71656333038086,
      "grad_norm": 0.20079156756401062,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 95645
    },
    {
      "epoch": 84.72099202834367,
      "grad_norm": 0.2567376494407654,
      "learning_rate": 1e-05,
      "loss": 0.9325,
      "step": 95650
    },
    {
      "epoch": 84.72542072630647,
      "grad_norm": 0.2568581700325012,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 95655
    },
    {
      "epoch": 84.72984942426926,
      "grad_norm": 0.2080429345369339,
      "learning_rate": 1e-05,
      "loss": 0.9275,
      "step": 95660
    },
    {
      "epoch": 84.73427812223207,
      "grad_norm": 0.2557322680950165,
      "learning_rate": 1e-05,
      "loss": 1.0084,
      "step": 95665
    },
    {
      "epoch": 84.73870682019486,
      "grad_norm": 0.24420638382434845,
      "learning_rate": 1e-05,
      "loss": 0.9299,
      "step": 95670
    },
    {
      "epoch": 84.74313551815766,
      "grad_norm": 0.2776748239994049,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 95675
    },
    {
      "epoch": 84.74756421612047,
      "grad_norm": 0.20128783583641052,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 95680
    },
    {
      "epoch": 84.75199291408326,
      "grad_norm": 0.21968212723731995,
      "learning_rate": 1e-05,
      "loss": 1.0026,
      "step": 95685
    },
    {
      "epoch": 84.75642161204605,
      "grad_norm": 0.21135146915912628,
      "learning_rate": 1e-05,
      "loss": 0.8722,
      "step": 95690
    },
    {
      "epoch": 84.76085031000886,
      "grad_norm": 0.23909050226211548,
      "learning_rate": 1e-05,
      "loss": 0.9348,
      "step": 95695
    },
    {
      "epoch": 84.76527900797166,
      "grad_norm": 0.2672107219696045,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 95700
    },
    {
      "epoch": 84.76970770593445,
      "grad_norm": 0.3292006254196167,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 95705
    },
    {
      "epoch": 84.77413640389726,
      "grad_norm": 0.2799969017505646,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 95710
    },
    {
      "epoch": 84.77856510186005,
      "grad_norm": 0.24608798325061798,
      "learning_rate": 1e-05,
      "loss": 0.9136,
      "step": 95715
    },
    {
      "epoch": 84.78299379982285,
      "grad_norm": 0.2224859744310379,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 95720
    },
    {
      "epoch": 84.78742249778566,
      "grad_norm": 0.2026747465133667,
      "learning_rate": 1e-05,
      "loss": 1.011,
      "step": 95725
    },
    {
      "epoch": 84.79185119574845,
      "grad_norm": 0.25361397862434387,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 95730
    },
    {
      "epoch": 84.79627989371124,
      "grad_norm": 0.2583954930305481,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 95735
    },
    {
      "epoch": 84.80070859167405,
      "grad_norm": 0.23851393163204193,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 95740
    },
    {
      "epoch": 84.80513728963685,
      "grad_norm": 0.23354066908359528,
      "learning_rate": 1e-05,
      "loss": 0.9162,
      "step": 95745
    },
    {
      "epoch": 84.80956598759964,
      "grad_norm": 0.26972946524620056,
      "learning_rate": 1e-05,
      "loss": 0.9165,
      "step": 95750
    },
    {
      "epoch": 84.81399468556245,
      "grad_norm": 0.226934015750885,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 95755
    },
    {
      "epoch": 84.81842338352524,
      "grad_norm": 0.2299937605857849,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 95760
    },
    {
      "epoch": 84.82285208148804,
      "grad_norm": 0.24781367182731628,
      "learning_rate": 1e-05,
      "loss": 0.9037,
      "step": 95765
    },
    {
      "epoch": 84.82728077945085,
      "grad_norm": 0.2466375082731247,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 95770
    },
    {
      "epoch": 84.83170947741364,
      "grad_norm": 0.2405197024345398,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 95775
    },
    {
      "epoch": 84.83613817537643,
      "grad_norm": 0.2225402444601059,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 95780
    },
    {
      "epoch": 84.84056687333924,
      "grad_norm": 0.2650313973426819,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 95785
    },
    {
      "epoch": 84.84499557130204,
      "grad_norm": 0.2287360280752182,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 95790
    },
    {
      "epoch": 84.84942426926483,
      "grad_norm": 0.26884597539901733,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 95795
    },
    {
      "epoch": 84.85385296722764,
      "grad_norm": 0.32081130146980286,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 95800
    },
    {
      "epoch": 84.85828166519043,
      "grad_norm": 0.22239120304584503,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 95805
    },
    {
      "epoch": 84.86271036315323,
      "grad_norm": 0.24696296453475952,
      "learning_rate": 1e-05,
      "loss": 0.941,
      "step": 95810
    },
    {
      "epoch": 84.86713906111603,
      "grad_norm": 0.23134391009807587,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 95815
    },
    {
      "epoch": 84.87156775907883,
      "grad_norm": 0.24801728129386902,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 95820
    },
    {
      "epoch": 84.87599645704162,
      "grad_norm": 0.2218467891216278,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 95825
    },
    {
      "epoch": 84.88042515500443,
      "grad_norm": 0.27515849471092224,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 95830
    },
    {
      "epoch": 84.88485385296723,
      "grad_norm": 0.2518330514431,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 95835
    },
    {
      "epoch": 84.88928255093003,
      "grad_norm": 0.23716948926448822,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 95840
    },
    {
      "epoch": 84.89371124889283,
      "grad_norm": 0.2406175583600998,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 95845
    },
    {
      "epoch": 84.89813994685562,
      "grad_norm": 0.22526101768016815,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 95850
    },
    {
      "epoch": 84.90256864481843,
      "grad_norm": 0.2963843047618866,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 95855
    },
    {
      "epoch": 84.90699734278122,
      "grad_norm": 0.2801845967769623,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 95860
    },
    {
      "epoch": 84.91142604074402,
      "grad_norm": 0.2423602044582367,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 95865
    },
    {
      "epoch": 84.91585473870683,
      "grad_norm": 0.23360104858875275,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 95870
    },
    {
      "epoch": 84.92028343666962,
      "grad_norm": 0.25409355759620667,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 95875
    },
    {
      "epoch": 84.92471213463241,
      "grad_norm": 0.23014232516288757,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 95880
    },
    {
      "epoch": 84.92914083259522,
      "grad_norm": 0.23911939561367035,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 95885
    },
    {
      "epoch": 84.93356953055802,
      "grad_norm": 0.2504749596118927,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 95890
    },
    {
      "epoch": 84.93799822852081,
      "grad_norm": 0.29032817482948303,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 95895
    },
    {
      "epoch": 84.94242692648362,
      "grad_norm": 0.2515168786048889,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 95900
    },
    {
      "epoch": 84.94685562444641,
      "grad_norm": 0.24183066189289093,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 95905
    },
    {
      "epoch": 84.95128432240921,
      "grad_norm": 0.23755010962486267,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 95910
    },
    {
      "epoch": 84.95571302037202,
      "grad_norm": 0.2761787474155426,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 95915
    },
    {
      "epoch": 84.96014171833481,
      "grad_norm": 0.22239238023757935,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 95920
    },
    {
      "epoch": 84.9645704162976,
      "grad_norm": 0.23220498859882355,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 95925
    },
    {
      "epoch": 84.96899911426041,
      "grad_norm": 0.2367704063653946,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 95930
    },
    {
      "epoch": 84.9734278122232,
      "grad_norm": 0.2466348558664322,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 95935
    },
    {
      "epoch": 84.977856510186,
      "grad_norm": 0.26395756006240845,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 95940
    },
    {
      "epoch": 84.98228520814881,
      "grad_norm": 0.250715047121048,
      "learning_rate": 1e-05,
      "loss": 0.9223,
      "step": 95945
    },
    {
      "epoch": 84.9867139061116,
      "grad_norm": 0.24061135947704315,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 95950
    },
    {
      "epoch": 84.9911426040744,
      "grad_norm": 0.21292370557785034,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 95955
    },
    {
      "epoch": 84.9955713020372,
      "grad_norm": 0.21280573308467865,
      "learning_rate": 1e-05,
      "loss": 0.9301,
      "step": 95960
    },
    {
      "epoch": 85.0,
      "grad_norm": 0.2240002304315567,
      "learning_rate": 1e-05,
      "loss": 0.8722,
      "step": 95965
    },
    {
      "epoch": 85.0044286979628,
      "grad_norm": 0.2056228518486023,
      "learning_rate": 1e-05,
      "loss": 0.9105,
      "step": 95970
    },
    {
      "epoch": 85.0088573959256,
      "grad_norm": 0.22989049553871155,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 95975
    },
    {
      "epoch": 85.0132860938884,
      "grad_norm": 0.2428256869316101,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 95980
    },
    {
      "epoch": 85.01771479185119,
      "grad_norm": 0.2553693652153015,
      "learning_rate": 1e-05,
      "loss": 0.9109,
      "step": 95985
    },
    {
      "epoch": 85.022143489814,
      "grad_norm": 0.24397438764572144,
      "learning_rate": 1e-05,
      "loss": 0.9185,
      "step": 95990
    },
    {
      "epoch": 85.0265721877768,
      "grad_norm": 0.2233012616634369,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 95995
    },
    {
      "epoch": 85.03100088573959,
      "grad_norm": 0.2178582102060318,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 96000
    },
    {
      "epoch": 85.0354295837024,
      "grad_norm": 0.236038476228714,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 96005
    },
    {
      "epoch": 85.03985828166519,
      "grad_norm": 0.24216914176940918,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 96010
    },
    {
      "epoch": 85.04428697962798,
      "grad_norm": 0.23506732285022736,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 96015
    },
    {
      "epoch": 85.04871567759079,
      "grad_norm": 0.222891703248024,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 96020
    },
    {
      "epoch": 85.05314437555359,
      "grad_norm": 0.20837712287902832,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 96025
    },
    {
      "epoch": 85.05757307351638,
      "grad_norm": 0.2472689002752304,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 96030
    },
    {
      "epoch": 85.06200177147919,
      "grad_norm": 0.22774197161197662,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 96035
    },
    {
      "epoch": 85.06643046944198,
      "grad_norm": 0.2266194075345993,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 96040
    },
    {
      "epoch": 85.07085916740478,
      "grad_norm": 0.2301270216703415,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 96045
    },
    {
      "epoch": 85.07528786536759,
      "grad_norm": 0.2130812257528305,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 96050
    },
    {
      "epoch": 85.07971656333038,
      "grad_norm": 0.23036104440689087,
      "learning_rate": 1e-05,
      "loss": 0.9318,
      "step": 96055
    },
    {
      "epoch": 85.08414526129317,
      "grad_norm": 0.25373438000679016,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 96060
    },
    {
      "epoch": 85.08857395925598,
      "grad_norm": 0.24780772626399994,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 96065
    },
    {
      "epoch": 85.09300265721878,
      "grad_norm": 0.21206624805927277,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 96070
    },
    {
      "epoch": 85.09743135518157,
      "grad_norm": 0.24895332753658295,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 96075
    },
    {
      "epoch": 85.10186005314438,
      "grad_norm": 0.2327793836593628,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 96080
    },
    {
      "epoch": 85.10628875110717,
      "grad_norm": 0.24390637874603271,
      "learning_rate": 1e-05,
      "loss": 0.9221,
      "step": 96085
    },
    {
      "epoch": 85.11071744906998,
      "grad_norm": 0.26170188188552856,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 96090
    },
    {
      "epoch": 85.11514614703277,
      "grad_norm": 0.22147339582443237,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 96095
    },
    {
      "epoch": 85.11957484499557,
      "grad_norm": 0.23095647990703583,
      "learning_rate": 1e-05,
      "loss": 0.9283,
      "step": 96100
    },
    {
      "epoch": 85.12400354295838,
      "grad_norm": 0.2348744124174118,
      "learning_rate": 1e-05,
      "loss": 0.9203,
      "step": 96105
    },
    {
      "epoch": 85.12843224092117,
      "grad_norm": 0.2612001299858093,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 96110
    },
    {
      "epoch": 85.13286093888397,
      "grad_norm": 0.23263311386108398,
      "learning_rate": 1e-05,
      "loss": 0.9096,
      "step": 96115
    },
    {
      "epoch": 85.13728963684677,
      "grad_norm": 0.25108346343040466,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 96120
    },
    {
      "epoch": 85.14171833480957,
      "grad_norm": 0.27531638741493225,
      "learning_rate": 1e-05,
      "loss": 1.0082,
      "step": 96125
    },
    {
      "epoch": 85.14614703277236,
      "grad_norm": 0.23817220330238342,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 96130
    },
    {
      "epoch": 85.15057573073517,
      "grad_norm": 0.3062964379787445,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 96135
    },
    {
      "epoch": 85.15500442869796,
      "grad_norm": 0.2517683506011963,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 96140
    },
    {
      "epoch": 85.15943312666076,
      "grad_norm": 0.2597475051879883,
      "learning_rate": 1e-05,
      "loss": 0.9067,
      "step": 96145
    },
    {
      "epoch": 85.16386182462357,
      "grad_norm": 0.2368151992559433,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 96150
    },
    {
      "epoch": 85.16829052258636,
      "grad_norm": 0.2859891355037689,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 96155
    },
    {
      "epoch": 85.17271922054915,
      "grad_norm": 0.25269755721092224,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 96160
    },
    {
      "epoch": 85.17714791851196,
      "grad_norm": 0.22500145435333252,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 96165
    },
    {
      "epoch": 85.18157661647476,
      "grad_norm": 0.26541993021965027,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 96170
    },
    {
      "epoch": 85.18600531443755,
      "grad_norm": 0.22606979310512543,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 96175
    },
    {
      "epoch": 85.19043401240036,
      "grad_norm": 0.30259039998054504,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 96180
    },
    {
      "epoch": 85.19486271036315,
      "grad_norm": 0.21107599139213562,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 96185
    },
    {
      "epoch": 85.19929140832595,
      "grad_norm": 0.20577754080295563,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 96190
    },
    {
      "epoch": 85.20372010628876,
      "grad_norm": 0.2331273853778839,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 96195
    },
    {
      "epoch": 85.20814880425155,
      "grad_norm": 0.25989672541618347,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 96200
    },
    {
      "epoch": 85.21257750221434,
      "grad_norm": 0.2498919665813446,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 96205
    },
    {
      "epoch": 85.21700620017715,
      "grad_norm": 0.24295833706855774,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 96210
    },
    {
      "epoch": 85.22143489813995,
      "grad_norm": 0.21256954967975616,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 96215
    },
    {
      "epoch": 85.22586359610274,
      "grad_norm": 0.22360825538635254,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 96220
    },
    {
      "epoch": 85.23029229406555,
      "grad_norm": 0.24023771286010742,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 96225
    },
    {
      "epoch": 85.23472099202834,
      "grad_norm": 0.27968940138816833,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 96230
    },
    {
      "epoch": 85.23914968999114,
      "grad_norm": 0.26634642481803894,
      "learning_rate": 1e-05,
      "loss": 0.9274,
      "step": 96235
    },
    {
      "epoch": 85.24357838795395,
      "grad_norm": 0.22590051591396332,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 96240
    },
    {
      "epoch": 85.24800708591674,
      "grad_norm": 0.2178240716457367,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 96245
    },
    {
      "epoch": 85.25243578387953,
      "grad_norm": 0.22034253180027008,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 96250
    },
    {
      "epoch": 85.25686448184234,
      "grad_norm": 0.2205260992050171,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 96255
    },
    {
      "epoch": 85.26129317980514,
      "grad_norm": 0.2659783363342285,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 96260
    },
    {
      "epoch": 85.26572187776793,
      "grad_norm": 0.2397063970565796,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 96265
    },
    {
      "epoch": 85.27015057573074,
      "grad_norm": 0.24733999371528625,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 96270
    },
    {
      "epoch": 85.27457927369353,
      "grad_norm": 0.24266697466373444,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 96275
    },
    {
      "epoch": 85.27900797165633,
      "grad_norm": 0.25073230266571045,
      "learning_rate": 1e-05,
      "loss": 0.9021,
      "step": 96280
    },
    {
      "epoch": 85.28343666961914,
      "grad_norm": 0.23984003067016602,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 96285
    },
    {
      "epoch": 85.28786536758193,
      "grad_norm": 0.24378043413162231,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 96290
    },
    {
      "epoch": 85.29229406554472,
      "grad_norm": 0.21318432688713074,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 96295
    },
    {
      "epoch": 85.29672276350753,
      "grad_norm": 0.23227250576019287,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 96300
    },
    {
      "epoch": 85.30115146147033,
      "grad_norm": 0.23576156795024872,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 96305
    },
    {
      "epoch": 85.30558015943312,
      "grad_norm": 0.22098015248775482,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 96310
    },
    {
      "epoch": 85.31000885739593,
      "grad_norm": 0.23480024933815002,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 96315
    },
    {
      "epoch": 85.31443755535872,
      "grad_norm": 0.231546089053154,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 96320
    },
    {
      "epoch": 85.31886625332152,
      "grad_norm": 0.2620023190975189,
      "learning_rate": 1e-05,
      "loss": 1.0364,
      "step": 96325
    },
    {
      "epoch": 85.32329495128432,
      "grad_norm": 0.2844887673854828,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 96330
    },
    {
      "epoch": 85.32772364924712,
      "grad_norm": 0.2536701560020447,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 96335
    },
    {
      "epoch": 85.33215234720993,
      "grad_norm": 0.23318342864513397,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 96340
    },
    {
      "epoch": 85.33658104517272,
      "grad_norm": 0.26637378334999084,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 96345
    },
    {
      "epoch": 85.34100974313552,
      "grad_norm": 0.25212958455085754,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 96350
    },
    {
      "epoch": 85.34543844109832,
      "grad_norm": 0.26395702362060547,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 96355
    },
    {
      "epoch": 85.34986713906112,
      "grad_norm": 0.29097980260849,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 96360
    },
    {
      "epoch": 85.35429583702391,
      "grad_norm": 0.22208768129348755,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 96365
    },
    {
      "epoch": 85.35872453498672,
      "grad_norm": 0.25918999314308167,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 96370
    },
    {
      "epoch": 85.36315323294951,
      "grad_norm": 0.23392996191978455,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 96375
    },
    {
      "epoch": 85.36758193091231,
      "grad_norm": 0.24971581995487213,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 96380
    },
    {
      "epoch": 85.37201062887512,
      "grad_norm": 0.24059641361236572,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 96385
    },
    {
      "epoch": 85.37643932683791,
      "grad_norm": 0.29464736580848694,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 96390
    },
    {
      "epoch": 85.3808680248007,
      "grad_norm": 0.30348721146583557,
      "learning_rate": 1e-05,
      "loss": 0.9004,
      "step": 96395
    },
    {
      "epoch": 85.38529672276351,
      "grad_norm": 0.19624000787734985,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 96400
    },
    {
      "epoch": 85.38972542072631,
      "grad_norm": 0.25620174407958984,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 96405
    },
    {
      "epoch": 85.3941541186891,
      "grad_norm": 0.25557562708854675,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 96410
    },
    {
      "epoch": 85.39858281665191,
      "grad_norm": 0.23023800551891327,
      "learning_rate": 1e-05,
      "loss": 0.895,
      "step": 96415
    },
    {
      "epoch": 85.4030115146147,
      "grad_norm": 0.21636566519737244,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 96420
    },
    {
      "epoch": 85.4074402125775,
      "grad_norm": 0.26349443197250366,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 96425
    },
    {
      "epoch": 85.4118689105403,
      "grad_norm": 0.21954220533370972,
      "learning_rate": 1e-05,
      "loss": 0.9134,
      "step": 96430
    },
    {
      "epoch": 85.4162976085031,
      "grad_norm": 0.23321829736232758,
      "learning_rate": 1e-05,
      "loss": 0.9248,
      "step": 96435
    },
    {
      "epoch": 85.4207263064659,
      "grad_norm": 0.22055462002754211,
      "learning_rate": 1e-05,
      "loss": 0.889,
      "step": 96440
    },
    {
      "epoch": 85.4251550044287,
      "grad_norm": 0.22671271860599518,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 96445
    },
    {
      "epoch": 85.4295837023915,
      "grad_norm": 0.24289150536060333,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 96450
    },
    {
      "epoch": 85.43401240035429,
      "grad_norm": 0.20306511223316193,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 96455
    },
    {
      "epoch": 85.4384410983171,
      "grad_norm": 0.22065265476703644,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 96460
    },
    {
      "epoch": 85.4428697962799,
      "grad_norm": 0.24153415858745575,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 96465
    },
    {
      "epoch": 85.44729849424269,
      "grad_norm": 0.24186073243618011,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 96470
    },
    {
      "epoch": 85.4517271922055,
      "grad_norm": 0.26280611753463745,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 96475
    },
    {
      "epoch": 85.45615589016829,
      "grad_norm": 0.2052849531173706,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 96480
    },
    {
      "epoch": 85.46058458813108,
      "grad_norm": 0.2617191672325134,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 96485
    },
    {
      "epoch": 85.46501328609389,
      "grad_norm": 0.22841627895832062,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 96490
    },
    {
      "epoch": 85.46944198405669,
      "grad_norm": 0.22078357636928558,
      "learning_rate": 1e-05,
      "loss": 0.9802,
      "step": 96495
    },
    {
      "epoch": 85.47387068201948,
      "grad_norm": 0.18779288232326508,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 96500
    },
    {
      "epoch": 85.47829937998229,
      "grad_norm": 0.2376197725534439,
      "learning_rate": 1e-05,
      "loss": 0.9269,
      "step": 96505
    },
    {
      "epoch": 85.48272807794508,
      "grad_norm": 0.2139972299337387,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 96510
    },
    {
      "epoch": 85.48715677590788,
      "grad_norm": 0.23659105598926544,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 96515
    },
    {
      "epoch": 85.49158547387069,
      "grad_norm": 0.20469188690185547,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 96520
    },
    {
      "epoch": 85.49601417183348,
      "grad_norm": 0.2353823482990265,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 96525
    },
    {
      "epoch": 85.50044286979627,
      "grad_norm": 0.21777375042438507,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 96530
    },
    {
      "epoch": 85.50487156775908,
      "grad_norm": 0.29329055547714233,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 96535
    },
    {
      "epoch": 85.50930026572188,
      "grad_norm": 0.25493407249450684,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 96540
    },
    {
      "epoch": 85.51372896368467,
      "grad_norm": 0.26193639636039734,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 96545
    },
    {
      "epoch": 85.51815766164748,
      "grad_norm": 0.22499775886535645,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 96550
    },
    {
      "epoch": 85.52258635961027,
      "grad_norm": 0.2297782003879547,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 96555
    },
    {
      "epoch": 85.52701505757307,
      "grad_norm": 0.2217954397201538,
      "learning_rate": 1e-05,
      "loss": 0.9229,
      "step": 96560
    },
    {
      "epoch": 85.53144375553588,
      "grad_norm": 0.20019471645355225,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 96565
    },
    {
      "epoch": 85.53587245349867,
      "grad_norm": 0.2243320196866989,
      "learning_rate": 1e-05,
      "loss": 0.9115,
      "step": 96570
    },
    {
      "epoch": 85.54030115146146,
      "grad_norm": 0.2591068744659424,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 96575
    },
    {
      "epoch": 85.54472984942427,
      "grad_norm": 0.2729327082633972,
      "learning_rate": 1e-05,
      "loss": 1.027,
      "step": 96580
    },
    {
      "epoch": 85.54915854738707,
      "grad_norm": 0.20644141733646393,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 96585
    },
    {
      "epoch": 85.55358724534987,
      "grad_norm": 0.24437689781188965,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 96590
    },
    {
      "epoch": 85.55801594331267,
      "grad_norm": 0.21154636144638062,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 96595
    },
    {
      "epoch": 85.56244464127546,
      "grad_norm": 0.3021708130836487,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 96600
    },
    {
      "epoch": 85.56687333923827,
      "grad_norm": 0.24401721358299255,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 96605
    },
    {
      "epoch": 85.57130203720106,
      "grad_norm": 0.23233330249786377,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 96610
    },
    {
      "epoch": 85.57573073516386,
      "grad_norm": 0.23136001825332642,
      "learning_rate": 1e-05,
      "loss": 0.9115,
      "step": 96615
    },
    {
      "epoch": 85.58015943312667,
      "grad_norm": 0.22464518249034882,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 96620
    },
    {
      "epoch": 85.58458813108946,
      "grad_norm": 0.2325783371925354,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 96625
    },
    {
      "epoch": 85.58901682905226,
      "grad_norm": 0.2508772313594818,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 96630
    },
    {
      "epoch": 85.59344552701506,
      "grad_norm": 0.24181824922561646,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 96635
    },
    {
      "epoch": 85.59787422497786,
      "grad_norm": 0.29301080107688904,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 96640
    },
    {
      "epoch": 85.60230292294065,
      "grad_norm": 0.21493034064769745,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 96645
    },
    {
      "epoch": 85.60673162090346,
      "grad_norm": 0.24663220345973969,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 96650
    },
    {
      "epoch": 85.61116031886625,
      "grad_norm": 0.2454022467136383,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 96655
    },
    {
      "epoch": 85.61558901682905,
      "grad_norm": 0.21560049057006836,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 96660
    },
    {
      "epoch": 85.62001771479186,
      "grad_norm": 0.23938271403312683,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 96665
    },
    {
      "epoch": 85.62444641275465,
      "grad_norm": 0.2568552494049072,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 96670
    },
    {
      "epoch": 85.62887511071744,
      "grad_norm": 0.21122068166732788,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 96675
    },
    {
      "epoch": 85.63330380868025,
      "grad_norm": 0.2101432830095291,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 96680
    },
    {
      "epoch": 85.63773250664305,
      "grad_norm": 0.25863781571388245,
      "learning_rate": 1e-05,
      "loss": 1.014,
      "step": 96685
    },
    {
      "epoch": 85.64216120460584,
      "grad_norm": 0.28838083148002625,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 96690
    },
    {
      "epoch": 85.64658990256865,
      "grad_norm": 0.22984611988067627,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 96695
    },
    {
      "epoch": 85.65101860053144,
      "grad_norm": 0.25642403960227966,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 96700
    },
    {
      "epoch": 85.65544729849424,
      "grad_norm": 0.2519679367542267,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 96705
    },
    {
      "epoch": 85.65987599645705,
      "grad_norm": 0.23742642998695374,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 96710
    },
    {
      "epoch": 85.66430469441984,
      "grad_norm": 0.20871853828430176,
      "learning_rate": 1e-05,
      "loss": 0.9324,
      "step": 96715
    },
    {
      "epoch": 85.66873339238263,
      "grad_norm": 0.2628812789916992,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 96720
    },
    {
      "epoch": 85.67316209034544,
      "grad_norm": 0.2525511682033539,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 96725
    },
    {
      "epoch": 85.67759078830824,
      "grad_norm": 0.2440093755722046,
      "learning_rate": 1e-05,
      "loss": 0.9312,
      "step": 96730
    },
    {
      "epoch": 85.68201948627103,
      "grad_norm": 0.25228914618492126,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 96735
    },
    {
      "epoch": 85.68644818423384,
      "grad_norm": 0.21076780557632446,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 96740
    },
    {
      "epoch": 85.69087688219663,
      "grad_norm": 0.21402721107006073,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 96745
    },
    {
      "epoch": 85.69530558015943,
      "grad_norm": 0.23448315262794495,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 96750
    },
    {
      "epoch": 85.69973427812224,
      "grad_norm": 0.2593778669834137,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 96755
    },
    {
      "epoch": 85.70416297608503,
      "grad_norm": 0.20808227360248566,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 96760
    },
    {
      "epoch": 85.70859167404782,
      "grad_norm": 0.21210364997386932,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 96765
    },
    {
      "epoch": 85.71302037201063,
      "grad_norm": 0.23420847952365875,
      "learning_rate": 1e-05,
      "loss": 0.9033,
      "step": 96770
    },
    {
      "epoch": 85.71744906997343,
      "grad_norm": 0.2520560026168823,
      "learning_rate": 1e-05,
      "loss": 0.9093,
      "step": 96775
    },
    {
      "epoch": 85.72187776793622,
      "grad_norm": 0.24139678478240967,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 96780
    },
    {
      "epoch": 85.72630646589903,
      "grad_norm": 0.23454616963863373,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 96785
    },
    {
      "epoch": 85.73073516386182,
      "grad_norm": 0.2702997922897339,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 96790
    },
    {
      "epoch": 85.73516386182462,
      "grad_norm": 0.2382647842168808,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 96795
    },
    {
      "epoch": 85.73959255978743,
      "grad_norm": 0.24645884335041046,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 96800
    },
    {
      "epoch": 85.74402125775022,
      "grad_norm": 0.23619645833969116,
      "learning_rate": 1e-05,
      "loss": 0.9183,
      "step": 96805
    },
    {
      "epoch": 85.74844995571301,
      "grad_norm": 0.2531241476535797,
      "learning_rate": 1e-05,
      "loss": 0.906,
      "step": 96810
    },
    {
      "epoch": 85.75287865367582,
      "grad_norm": 0.2335454374551773,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 96815
    },
    {
      "epoch": 85.75730735163862,
      "grad_norm": 0.2425515502691269,
      "learning_rate": 1e-05,
      "loss": 0.9607,
      "step": 96820
    },
    {
      "epoch": 85.76173604960141,
      "grad_norm": 0.2611462473869324,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 96825
    },
    {
      "epoch": 85.76616474756422,
      "grad_norm": 0.24901819229125977,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 96830
    },
    {
      "epoch": 85.77059344552701,
      "grad_norm": 0.2651823163032532,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 96835
    },
    {
      "epoch": 85.77502214348982,
      "grad_norm": 0.2842780649662018,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 96840
    },
    {
      "epoch": 85.77945084145261,
      "grad_norm": 0.23318541049957275,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 96845
    },
    {
      "epoch": 85.78387953941541,
      "grad_norm": 0.2438599020242691,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 96850
    },
    {
      "epoch": 85.78830823737822,
      "grad_norm": 0.24094738066196442,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 96855
    },
    {
      "epoch": 85.79273693534101,
      "grad_norm": 0.22020454704761505,
      "learning_rate": 1e-05,
      "loss": 0.9158,
      "step": 96860
    },
    {
      "epoch": 85.7971656333038,
      "grad_norm": 0.30342698097229004,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 96865
    },
    {
      "epoch": 85.80159433126661,
      "grad_norm": 0.2090945988893509,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 96870
    },
    {
      "epoch": 85.80602302922941,
      "grad_norm": 0.2712833285331726,
      "learning_rate": 1e-05,
      "loss": 0.9099,
      "step": 96875
    },
    {
      "epoch": 85.8104517271922,
      "grad_norm": 0.2247336506843567,
      "learning_rate": 1e-05,
      "loss": 1.0278,
      "step": 96880
    },
    {
      "epoch": 85.81488042515501,
      "grad_norm": 0.20719477534294128,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 96885
    },
    {
      "epoch": 85.8193091231178,
      "grad_norm": 0.24392910301685333,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 96890
    },
    {
      "epoch": 85.8237378210806,
      "grad_norm": 0.20210695266723633,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 96895
    },
    {
      "epoch": 85.8281665190434,
      "grad_norm": 0.22325649857521057,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 96900
    },
    {
      "epoch": 85.8325952170062,
      "grad_norm": 0.22377164661884308,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 96905
    },
    {
      "epoch": 85.837023914969,
      "grad_norm": 0.2175963819026947,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 96910
    },
    {
      "epoch": 85.8414526129318,
      "grad_norm": 0.23284812271595,
      "learning_rate": 1e-05,
      "loss": 0.9192,
      "step": 96915
    },
    {
      "epoch": 85.8458813108946,
      "grad_norm": 0.2144775688648224,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 96920
    },
    {
      "epoch": 85.85031000885739,
      "grad_norm": 0.2345629632472992,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 96925
    },
    {
      "epoch": 85.8547387068202,
      "grad_norm": 0.22753940522670746,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 96930
    },
    {
      "epoch": 85.859167404783,
      "grad_norm": 0.22238260507583618,
      "learning_rate": 1e-05,
      "loss": 0.8913,
      "step": 96935
    },
    {
      "epoch": 85.86359610274579,
      "grad_norm": 0.2365919053554535,
      "learning_rate": 1e-05,
      "loss": 0.9111,
      "step": 96940
    },
    {
      "epoch": 85.8680248007086,
      "grad_norm": 0.2639627158641815,
      "learning_rate": 1e-05,
      "loss": 0.8692,
      "step": 96945
    },
    {
      "epoch": 85.87245349867139,
      "grad_norm": 0.24952369928359985,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 96950
    },
    {
      "epoch": 85.87688219663418,
      "grad_norm": 0.2549598217010498,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 96955
    },
    {
      "epoch": 85.881310894597,
      "grad_norm": 0.2707740068435669,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 96960
    },
    {
      "epoch": 85.88573959255979,
      "grad_norm": 0.20950205624103546,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 96965
    },
    {
      "epoch": 85.89016829052258,
      "grad_norm": 0.26541775465011597,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 96970
    },
    {
      "epoch": 85.89459698848539,
      "grad_norm": 0.22506029903888702,
      "learning_rate": 1e-05,
      "loss": 0.8782,
      "step": 96975
    },
    {
      "epoch": 85.89902568644818,
      "grad_norm": 0.24413637816905975,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 96980
    },
    {
      "epoch": 85.90345438441098,
      "grad_norm": 0.2097465693950653,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 96985
    },
    {
      "epoch": 85.90788308237379,
      "grad_norm": 0.21111370623111725,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 96990
    },
    {
      "epoch": 85.91231178033658,
      "grad_norm": 0.2513578534126282,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 96995
    },
    {
      "epoch": 85.91674047829937,
      "grad_norm": 0.2248516082763672,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 97000
    },
    {
      "epoch": 85.92116917626218,
      "grad_norm": 0.22819189727306366,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 97005
    },
    {
      "epoch": 85.92559787422498,
      "grad_norm": 0.25540363788604736,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 97010
    },
    {
      "epoch": 85.93002657218777,
      "grad_norm": 0.20727913081645966,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 97015
    },
    {
      "epoch": 85.93445527015058,
      "grad_norm": 0.20069310069084167,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 97020
    },
    {
      "epoch": 85.93888396811337,
      "grad_norm": 0.2319866120815277,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 97025
    },
    {
      "epoch": 85.94331266607617,
      "grad_norm": 0.23785686492919922,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 97030
    },
    {
      "epoch": 85.94774136403898,
      "grad_norm": 0.26096266508102417,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 97035
    },
    {
      "epoch": 85.95217006200177,
      "grad_norm": 0.22606976330280304,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 97040
    },
    {
      "epoch": 85.95659875996456,
      "grad_norm": 0.2725220024585724,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 97045
    },
    {
      "epoch": 85.96102745792737,
      "grad_norm": 0.2425740361213684,
      "learning_rate": 1e-05,
      "loss": 0.9095,
      "step": 97050
    },
    {
      "epoch": 85.96545615589017,
      "grad_norm": 0.2293911725282669,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 97055
    },
    {
      "epoch": 85.96988485385296,
      "grad_norm": 0.2514573037624359,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 97060
    },
    {
      "epoch": 85.97431355181577,
      "grad_norm": 0.24325360357761383,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 97065
    },
    {
      "epoch": 85.97874224977856,
      "grad_norm": 0.2408817559480667,
      "learning_rate": 1e-05,
      "loss": 0.8835,
      "step": 97070
    },
    {
      "epoch": 85.98317094774137,
      "grad_norm": 0.19104577600955963,
      "learning_rate": 1e-05,
      "loss": 0.9023,
      "step": 97075
    },
    {
      "epoch": 85.98759964570417,
      "grad_norm": 0.2191479653120041,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 97080
    },
    {
      "epoch": 85.99202834366696,
      "grad_norm": 0.24199625849723816,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 97085
    },
    {
      "epoch": 85.99645704162977,
      "grad_norm": 0.23012059926986694,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 97090
    },
    {
      "epoch": 86.00088573959256,
      "grad_norm": 0.24371026456356049,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 97095
    },
    {
      "epoch": 86.00531443755536,
      "grad_norm": 0.2351166009902954,
      "learning_rate": 1e-05,
      "loss": 0.9161,
      "step": 97100
    },
    {
      "epoch": 86.00974313551816,
      "grad_norm": 0.2213047295808792,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 97105
    },
    {
      "epoch": 86.01417183348096,
      "grad_norm": 0.2363177090883255,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 97110
    },
    {
      "epoch": 86.01860053144375,
      "grad_norm": 0.2377803921699524,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 97115
    },
    {
      "epoch": 86.02302922940656,
      "grad_norm": 0.24732117354869843,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 97120
    },
    {
      "epoch": 86.02745792736935,
      "grad_norm": 0.28680679202079773,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 97125
    },
    {
      "epoch": 86.03188662533215,
      "grad_norm": 0.2635520398616791,
      "learning_rate": 1e-05,
      "loss": 0.9787,
      "step": 97130
    },
    {
      "epoch": 86.03631532329496,
      "grad_norm": 0.22710517048835754,
      "learning_rate": 1e-05,
      "loss": 0.9144,
      "step": 97135
    },
    {
      "epoch": 86.04074402125775,
      "grad_norm": 0.2916337251663208,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 97140
    },
    {
      "epoch": 86.04517271922055,
      "grad_norm": 0.20640406012535095,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 97145
    },
    {
      "epoch": 86.04960141718335,
      "grad_norm": 0.22454360127449036,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 97150
    },
    {
      "epoch": 86.05403011514615,
      "grad_norm": 0.22561374306678772,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 97155
    },
    {
      "epoch": 86.05845881310894,
      "grad_norm": 0.2779489755630493,
      "learning_rate": 1e-05,
      "loss": 0.921,
      "step": 97160
    },
    {
      "epoch": 86.06288751107175,
      "grad_norm": 0.23810631036758423,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 97165
    },
    {
      "epoch": 86.06731620903454,
      "grad_norm": 0.22184278070926666,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 97170
    },
    {
      "epoch": 86.07174490699734,
      "grad_norm": 0.21872155368328094,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 97175
    },
    {
      "epoch": 86.07617360496015,
      "grad_norm": 0.22361136972904205,
      "learning_rate": 1e-05,
      "loss": 0.9432,
      "step": 97180
    },
    {
      "epoch": 86.08060230292294,
      "grad_norm": 0.2417464703321457,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 97185
    },
    {
      "epoch": 86.08503100088573,
      "grad_norm": 0.3094818890094757,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 97190
    },
    {
      "epoch": 86.08945969884854,
      "grad_norm": 0.25736093521118164,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 97195
    },
    {
      "epoch": 86.09388839681134,
      "grad_norm": 0.23641172051429749,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 97200
    },
    {
      "epoch": 86.09831709477413,
      "grad_norm": 0.2542100250720978,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 97205
    },
    {
      "epoch": 86.10274579273694,
      "grad_norm": 0.2423756867647171,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 97210
    },
    {
      "epoch": 86.10717449069973,
      "grad_norm": 0.23302805423736572,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 97215
    },
    {
      "epoch": 86.11160318866253,
      "grad_norm": 0.2346411794424057,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 97220
    },
    {
      "epoch": 86.11603188662534,
      "grad_norm": 0.24297769367694855,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 97225
    },
    {
      "epoch": 86.12046058458813,
      "grad_norm": 0.2443009912967682,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 97230
    },
    {
      "epoch": 86.12488928255092,
      "grad_norm": 0.23632077872753143,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 97235
    },
    {
      "epoch": 86.12931798051373,
      "grad_norm": 0.21805907785892487,
      "learning_rate": 1e-05,
      "loss": 1.0009,
      "step": 97240
    },
    {
      "epoch": 86.13374667847653,
      "grad_norm": 0.294930636882782,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 97245
    },
    {
      "epoch": 86.13817537643932,
      "grad_norm": 0.26988664269447327,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 97250
    },
    {
      "epoch": 86.14260407440213,
      "grad_norm": 0.24270285665988922,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 97255
    },
    {
      "epoch": 86.14703277236492,
      "grad_norm": 0.298213392496109,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 97260
    },
    {
      "epoch": 86.15146147032772,
      "grad_norm": 0.22539810836315155,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 97265
    },
    {
      "epoch": 86.15589016829053,
      "grad_norm": 0.2500755488872528,
      "learning_rate": 1e-05,
      "loss": 1.0194,
      "step": 97270
    },
    {
      "epoch": 86.16031886625332,
      "grad_norm": 0.2239968478679657,
      "learning_rate": 1e-05,
      "loss": 1.0032,
      "step": 97275
    },
    {
      "epoch": 86.16474756421611,
      "grad_norm": 0.2412382960319519,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 97280
    },
    {
      "epoch": 86.16917626217892,
      "grad_norm": 0.24195455014705658,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 97285
    },
    {
      "epoch": 86.17360496014172,
      "grad_norm": 0.25617069005966187,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 97290
    },
    {
      "epoch": 86.17803365810451,
      "grad_norm": 0.24802014231681824,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 97295
    },
    {
      "epoch": 86.18246235606732,
      "grad_norm": 0.29795777797698975,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 97300
    },
    {
      "epoch": 86.18689105403011,
      "grad_norm": 0.26427629590034485,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 97305
    },
    {
      "epoch": 86.1913197519929,
      "grad_norm": 0.2762966752052307,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 97310
    },
    {
      "epoch": 86.19574844995572,
      "grad_norm": 0.24291293323040009,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 97315
    },
    {
      "epoch": 86.20017714791851,
      "grad_norm": 0.2579694390296936,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 97320
    },
    {
      "epoch": 86.20460584588132,
      "grad_norm": 0.2655971944332123,
      "learning_rate": 1e-05,
      "loss": 0.9113,
      "step": 97325
    },
    {
      "epoch": 86.20903454384411,
      "grad_norm": 0.21086034178733826,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 97330
    },
    {
      "epoch": 86.2134632418069,
      "grad_norm": 0.23919431865215302,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 97335
    },
    {
      "epoch": 86.21789193976971,
      "grad_norm": 0.24885676801204681,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 97340
    },
    {
      "epoch": 86.22232063773251,
      "grad_norm": 0.25179407000541687,
      "learning_rate": 1e-05,
      "loss": 0.915,
      "step": 97345
    },
    {
      "epoch": 86.2267493356953,
      "grad_norm": 0.23263004422187805,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 97350
    },
    {
      "epoch": 86.23117803365811,
      "grad_norm": 0.21684429049491882,
      "learning_rate": 1e-05,
      "loss": 0.918,
      "step": 97355
    },
    {
      "epoch": 86.2356067316209,
      "grad_norm": 0.2501088082790375,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 97360
    },
    {
      "epoch": 86.2400354295837,
      "grad_norm": 0.265235036611557,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 97365
    },
    {
      "epoch": 86.24446412754651,
      "grad_norm": 0.3170413672924042,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 97370
    },
    {
      "epoch": 86.2488928255093,
      "grad_norm": 0.22767452895641327,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 97375
    },
    {
      "epoch": 86.2533215234721,
      "grad_norm": 0.29185739159584045,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 97380
    },
    {
      "epoch": 86.2577502214349,
      "grad_norm": 0.23806607723236084,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 97385
    },
    {
      "epoch": 86.2621789193977,
      "grad_norm": 0.2752698063850403,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 97390
    },
    {
      "epoch": 86.26660761736049,
      "grad_norm": 0.226464182138443,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 97395
    },
    {
      "epoch": 86.2710363153233,
      "grad_norm": 0.2214711308479309,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 97400
    },
    {
      "epoch": 86.2754650132861,
      "grad_norm": 0.22089965641498566,
      "learning_rate": 1e-05,
      "loss": 0.9187,
      "step": 97405
    },
    {
      "epoch": 86.27989371124889,
      "grad_norm": 0.2389654815196991,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 97410
    },
    {
      "epoch": 86.2843224092117,
      "grad_norm": 0.2774227261543274,
      "learning_rate": 1e-05,
      "loss": 1.0066,
      "step": 97415
    },
    {
      "epoch": 86.28875110717449,
      "grad_norm": 0.3148418962955475,
      "learning_rate": 1e-05,
      "loss": 0.9009,
      "step": 97420
    },
    {
      "epoch": 86.29317980513729,
      "grad_norm": 0.28654780983924866,
      "learning_rate": 1e-05,
      "loss": 0.918,
      "step": 97425
    },
    {
      "epoch": 86.2976085031001,
      "grad_norm": 0.2448236495256424,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 97430
    },
    {
      "epoch": 86.30203720106289,
      "grad_norm": 0.26177218556404114,
      "learning_rate": 1e-05,
      "loss": 0.9077,
      "step": 97435
    },
    {
      "epoch": 86.30646589902568,
      "grad_norm": 0.2374694049358368,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 97440
    },
    {
      "epoch": 86.31089459698849,
      "grad_norm": 0.23216263949871063,
      "learning_rate": 1e-05,
      "loss": 1.0455,
      "step": 97445
    },
    {
      "epoch": 86.31532329495128,
      "grad_norm": 0.2625068426132202,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 97450
    },
    {
      "epoch": 86.31975199291408,
      "grad_norm": 0.23084689676761627,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 97455
    },
    {
      "epoch": 86.32418069087689,
      "grad_norm": 0.2509269118309021,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 97460
    },
    {
      "epoch": 86.32860938883968,
      "grad_norm": 0.26281052827835083,
      "learning_rate": 1e-05,
      "loss": 1.0168,
      "step": 97465
    },
    {
      "epoch": 86.33303808680247,
      "grad_norm": 0.2245265543460846,
      "learning_rate": 1e-05,
      "loss": 0.9154,
      "step": 97470
    },
    {
      "epoch": 86.33746678476528,
      "grad_norm": 0.24464735388755798,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 97475
    },
    {
      "epoch": 86.34189548272808,
      "grad_norm": 0.2500230371952057,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 97480
    },
    {
      "epoch": 86.34632418069087,
      "grad_norm": 0.2138415426015854,
      "learning_rate": 1e-05,
      "loss": 1.0278,
      "step": 97485
    },
    {
      "epoch": 86.35075287865368,
      "grad_norm": 0.2069804072380066,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 97490
    },
    {
      "epoch": 86.35518157661647,
      "grad_norm": 0.2307499647140503,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 97495
    },
    {
      "epoch": 86.35961027457927,
      "grad_norm": 0.22184212505817413,
      "learning_rate": 1e-05,
      "loss": 0.9307,
      "step": 97500
    },
    {
      "epoch": 86.36403897254208,
      "grad_norm": 0.23166102170944214,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 97505
    },
    {
      "epoch": 86.36846767050487,
      "grad_norm": 0.2785622775554657,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 97510
    },
    {
      "epoch": 86.37289636846766,
      "grad_norm": 0.23424622416496277,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 97515
    },
    {
      "epoch": 86.37732506643047,
      "grad_norm": 0.2433195263147354,
      "learning_rate": 1e-05,
      "loss": 1.0421,
      "step": 97520
    },
    {
      "epoch": 86.38175376439327,
      "grad_norm": 0.2580808103084564,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 97525
    },
    {
      "epoch": 86.38618246235606,
      "grad_norm": 0.26988548040390015,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 97530
    },
    {
      "epoch": 86.39061116031887,
      "grad_norm": 0.337377667427063,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 97535
    },
    {
      "epoch": 86.39503985828166,
      "grad_norm": 0.2984538674354553,
      "learning_rate": 1e-05,
      "loss": 0.9319,
      "step": 97540
    },
    {
      "epoch": 86.39946855624446,
      "grad_norm": 0.2478247433900833,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 97545
    },
    {
      "epoch": 86.40389725420727,
      "grad_norm": 0.2710384428501129,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 97550
    },
    {
      "epoch": 86.40832595217006,
      "grad_norm": 0.2437439113855362,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 97555
    },
    {
      "epoch": 86.41275465013285,
      "grad_norm": 0.28736335039138794,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 97560
    },
    {
      "epoch": 86.41718334809566,
      "grad_norm": 0.2324373722076416,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 97565
    },
    {
      "epoch": 86.42161204605846,
      "grad_norm": 0.2412739396095276,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 97570
    },
    {
      "epoch": 86.42604074402126,
      "grad_norm": 0.24981698393821716,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 97575
    },
    {
      "epoch": 86.43046944198406,
      "grad_norm": 0.2564948499202728,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 97580
    },
    {
      "epoch": 86.43489813994685,
      "grad_norm": 0.27079471945762634,
      "learning_rate": 1e-05,
      "loss": 0.9305,
      "step": 97585
    },
    {
      "epoch": 86.43932683790966,
      "grad_norm": 0.2288156896829605,
      "learning_rate": 1e-05,
      "loss": 0.8891,
      "step": 97590
    },
    {
      "epoch": 86.44375553587246,
      "grad_norm": 0.2102874368429184,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 97595
    },
    {
      "epoch": 86.44818423383525,
      "grad_norm": 0.23175379633903503,
      "learning_rate": 1e-05,
      "loss": 1.0107,
      "step": 97600
    },
    {
      "epoch": 86.45261293179806,
      "grad_norm": 0.2514692544937134,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 97605
    },
    {
      "epoch": 86.45704162976085,
      "grad_norm": 0.22433802485466003,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 97610
    },
    {
      "epoch": 86.46147032772365,
      "grad_norm": 0.24838802218437195,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 97615
    },
    {
      "epoch": 86.46589902568645,
      "grad_norm": 0.2515776455402374,
      "learning_rate": 1e-05,
      "loss": 0.9214,
      "step": 97620
    },
    {
      "epoch": 86.47032772364925,
      "grad_norm": 0.2791557312011719,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 97625
    },
    {
      "epoch": 86.47475642161204,
      "grad_norm": 0.28195664286613464,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 97630
    },
    {
      "epoch": 86.47918511957485,
      "grad_norm": 0.22671641409397125,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 97635
    },
    {
      "epoch": 86.48361381753764,
      "grad_norm": 0.219867542386055,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 97640
    },
    {
      "epoch": 86.48804251550044,
      "grad_norm": 0.21709716320037842,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 97645
    },
    {
      "epoch": 86.49247121346325,
      "grad_norm": 0.22821044921875,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 97650
    },
    {
      "epoch": 86.49689991142604,
      "grad_norm": 0.22244662046432495,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 97655
    },
    {
      "epoch": 86.50132860938884,
      "grad_norm": 0.22795841097831726,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 97660
    },
    {
      "epoch": 86.50575730735164,
      "grad_norm": 0.25087156891822815,
      "learning_rate": 1e-05,
      "loss": 0.942,
      "step": 97665
    },
    {
      "epoch": 86.51018600531444,
      "grad_norm": 0.2219347506761551,
      "learning_rate": 1e-05,
      "loss": 0.9134,
      "step": 97670
    },
    {
      "epoch": 86.51461470327723,
      "grad_norm": 0.2347313016653061,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 97675
    },
    {
      "epoch": 86.51904340124004,
      "grad_norm": 0.2769928574562073,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 97680
    },
    {
      "epoch": 86.52347209920283,
      "grad_norm": 0.2632846236228943,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 97685
    },
    {
      "epoch": 86.52790079716563,
      "grad_norm": 0.2277977466583252,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 97690
    },
    {
      "epoch": 86.53232949512844,
      "grad_norm": 0.21311065554618835,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 97695
    },
    {
      "epoch": 86.53675819309123,
      "grad_norm": 0.22003811597824097,
      "learning_rate": 1e-05,
      "loss": 0.9305,
      "step": 97700
    },
    {
      "epoch": 86.54118689105402,
      "grad_norm": 0.21697616577148438,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 97705
    },
    {
      "epoch": 86.54561558901683,
      "grad_norm": 0.24771209061145782,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 97710
    },
    {
      "epoch": 86.55004428697963,
      "grad_norm": 0.27790436148643494,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 97715
    },
    {
      "epoch": 86.55447298494242,
      "grad_norm": 0.21535947918891907,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 97720
    },
    {
      "epoch": 86.55890168290523,
      "grad_norm": 0.23929442465305328,
      "learning_rate": 1e-05,
      "loss": 0.9303,
      "step": 97725
    },
    {
      "epoch": 86.56333038086802,
      "grad_norm": 0.225034698843956,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 97730
    },
    {
      "epoch": 86.56775907883082,
      "grad_norm": 0.24152566492557526,
      "learning_rate": 1e-05,
      "loss": 0.9137,
      "step": 97735
    },
    {
      "epoch": 86.57218777679363,
      "grad_norm": 0.22860662639141083,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 97740
    },
    {
      "epoch": 86.57661647475642,
      "grad_norm": 0.23329578340053558,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 97745
    },
    {
      "epoch": 86.58104517271921,
      "grad_norm": 0.23391984403133392,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 97750
    },
    {
      "epoch": 86.58547387068202,
      "grad_norm": 0.2556324303150177,
      "learning_rate": 1e-05,
      "loss": 0.8946,
      "step": 97755
    },
    {
      "epoch": 86.58990256864482,
      "grad_norm": 0.26190054416656494,
      "learning_rate": 1e-05,
      "loss": 0.9195,
      "step": 97760
    },
    {
      "epoch": 86.59433126660761,
      "grad_norm": 0.2326575666666031,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 97765
    },
    {
      "epoch": 86.59875996457042,
      "grad_norm": 0.2641197741031647,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 97770
    },
    {
      "epoch": 86.60318866253321,
      "grad_norm": 0.2541007995605469,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 97775
    },
    {
      "epoch": 86.60761736049601,
      "grad_norm": 0.22842933237552643,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 97780
    },
    {
      "epoch": 86.61204605845882,
      "grad_norm": 0.2426004260778427,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 97785
    },
    {
      "epoch": 86.61647475642161,
      "grad_norm": 0.2489362359046936,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 97790
    },
    {
      "epoch": 86.6209034543844,
      "grad_norm": 0.2868337035179138,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 97795
    },
    {
      "epoch": 86.62533215234721,
      "grad_norm": 0.20236356556415558,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 97800
    },
    {
      "epoch": 86.62976085031,
      "grad_norm": 0.20789262652397156,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 97805
    },
    {
      "epoch": 86.63418954827281,
      "grad_norm": 0.23282887041568756,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 97810
    },
    {
      "epoch": 86.63861824623561,
      "grad_norm": 0.2431809902191162,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 97815
    },
    {
      "epoch": 86.6430469441984,
      "grad_norm": 0.25418025255203247,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 97820
    },
    {
      "epoch": 86.64747564216121,
      "grad_norm": 0.25638049840927124,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 97825
    },
    {
      "epoch": 86.651904340124,
      "grad_norm": 0.2520405352115631,
      "learning_rate": 1e-05,
      "loss": 0.926,
      "step": 97830
    },
    {
      "epoch": 86.6563330380868,
      "grad_norm": 0.24076220393180847,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 97835
    },
    {
      "epoch": 86.66076173604961,
      "grad_norm": 0.2553439736366272,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 97840
    },
    {
      "epoch": 86.6651904340124,
      "grad_norm": 0.23596803843975067,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 97845
    },
    {
      "epoch": 86.6696191319752,
      "grad_norm": 0.23852095007896423,
      "learning_rate": 1e-05,
      "loss": 1.0229,
      "step": 97850
    },
    {
      "epoch": 86.674047829938,
      "grad_norm": 0.2198798656463623,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 97855
    },
    {
      "epoch": 86.6784765279008,
      "grad_norm": 0.2180701345205307,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 97860
    },
    {
      "epoch": 86.68290522586359,
      "grad_norm": 0.23095166683197021,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 97865
    },
    {
      "epoch": 86.6873339238264,
      "grad_norm": 0.2211650311946869,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 97870
    },
    {
      "epoch": 86.6917626217892,
      "grad_norm": 0.2256811410188675,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 97875
    },
    {
      "epoch": 86.69619131975199,
      "grad_norm": 0.21950854361057281,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 97880
    },
    {
      "epoch": 86.7006200177148,
      "grad_norm": 0.23210366070270538,
      "learning_rate": 1e-05,
      "loss": 0.9182,
      "step": 97885
    },
    {
      "epoch": 86.70504871567759,
      "grad_norm": 0.25355851650238037,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 97890
    },
    {
      "epoch": 86.70947741364039,
      "grad_norm": 0.2949579656124115,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 97895
    },
    {
      "epoch": 86.7139061116032,
      "grad_norm": 0.2376047819852829,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 97900
    },
    {
      "epoch": 86.71833480956599,
      "grad_norm": 0.20439501106739044,
      "learning_rate": 1e-05,
      "loss": 1.022,
      "step": 97905
    },
    {
      "epoch": 86.72276350752878,
      "grad_norm": 0.2237773984670639,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 97910
    },
    {
      "epoch": 86.72719220549159,
      "grad_norm": 0.21593715250492096,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 97915
    },
    {
      "epoch": 86.73162090345438,
      "grad_norm": 0.23074491322040558,
      "learning_rate": 1e-05,
      "loss": 0.9203,
      "step": 97920
    },
    {
      "epoch": 86.73604960141718,
      "grad_norm": 0.21759459376335144,
      "learning_rate": 1e-05,
      "loss": 1.0382,
      "step": 97925
    },
    {
      "epoch": 86.74047829937999,
      "grad_norm": 0.24667830765247345,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 97930
    },
    {
      "epoch": 86.74490699734278,
      "grad_norm": 0.2486148178577423,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 97935
    },
    {
      "epoch": 86.74933569530558,
      "grad_norm": 0.21728363633155823,
      "learning_rate": 1e-05,
      "loss": 0.9239,
      "step": 97940
    },
    {
      "epoch": 86.75376439326838,
      "grad_norm": 0.2204359620809555,
      "learning_rate": 1e-05,
      "loss": 0.9251,
      "step": 97945
    },
    {
      "epoch": 86.75819309123118,
      "grad_norm": 0.24216429889202118,
      "learning_rate": 1e-05,
      "loss": 0.8985,
      "step": 97950
    },
    {
      "epoch": 86.76262178919397,
      "grad_norm": 0.26172107458114624,
      "learning_rate": 1e-05,
      "loss": 1.0401,
      "step": 97955
    },
    {
      "epoch": 86.76705048715678,
      "grad_norm": 0.2145540416240692,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 97960
    },
    {
      "epoch": 86.77147918511957,
      "grad_norm": 0.22526490688323975,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 97965
    },
    {
      "epoch": 86.77590788308237,
      "grad_norm": 0.22993145883083344,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 97970
    },
    {
      "epoch": 86.78033658104518,
      "grad_norm": 0.24860502779483795,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 97975
    },
    {
      "epoch": 86.78476527900797,
      "grad_norm": 0.23333078622817993,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 97980
    },
    {
      "epoch": 86.78919397697076,
      "grad_norm": 0.2229924350976944,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 97985
    },
    {
      "epoch": 86.79362267493357,
      "grad_norm": 0.25023791193962097,
      "learning_rate": 1e-05,
      "loss": 1.0038,
      "step": 97990
    },
    {
      "epoch": 86.79805137289637,
      "grad_norm": 0.22050027549266815,
      "learning_rate": 1e-05,
      "loss": 0.8818,
      "step": 97995
    },
    {
      "epoch": 86.80248007085916,
      "grad_norm": 0.23029173910617828,
      "learning_rate": 1e-05,
      "loss": 1.0337,
      "step": 98000
    },
    {
      "epoch": 86.80690876882197,
      "grad_norm": 0.18844686448574066,
      "learning_rate": 1e-05,
      "loss": 1.0224,
      "step": 98005
    },
    {
      "epoch": 86.81133746678476,
      "grad_norm": 0.2100907415151596,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 98010
    },
    {
      "epoch": 86.81576616474756,
      "grad_norm": 0.2497255802154541,
      "learning_rate": 1e-05,
      "loss": 0.9287,
      "step": 98015
    },
    {
      "epoch": 86.82019486271037,
      "grad_norm": 0.22252006828784943,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 98020
    },
    {
      "epoch": 86.82462356067316,
      "grad_norm": 0.25199493765830994,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 98025
    },
    {
      "epoch": 86.82905225863595,
      "grad_norm": 0.2313881665468216,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 98030
    },
    {
      "epoch": 86.83348095659876,
      "grad_norm": 0.2628002166748047,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 98035
    },
    {
      "epoch": 86.83790965456156,
      "grad_norm": 0.25987300276756287,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 98040
    },
    {
      "epoch": 86.84233835252435,
      "grad_norm": 0.23391568660736084,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 98045
    },
    {
      "epoch": 86.84676705048716,
      "grad_norm": 0.2406986951828003,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 98050
    },
    {
      "epoch": 86.85119574844995,
      "grad_norm": 0.24355019629001617,
      "learning_rate": 1e-05,
      "loss": 0.9284,
      "step": 98055
    },
    {
      "epoch": 86.85562444641276,
      "grad_norm": 0.21033534407615662,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 98060
    },
    {
      "epoch": 86.86005314437556,
      "grad_norm": 0.22992965579032898,
      "learning_rate": 1e-05,
      "loss": 0.9159,
      "step": 98065
    },
    {
      "epoch": 86.86448184233835,
      "grad_norm": 0.2060347944498062,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 98070
    },
    {
      "epoch": 86.86891054030116,
      "grad_norm": 0.23132586479187012,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 98075
    },
    {
      "epoch": 86.87333923826395,
      "grad_norm": 0.21193315088748932,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 98080
    },
    {
      "epoch": 86.87776793622675,
      "grad_norm": 0.2275550812482834,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 98085
    },
    {
      "epoch": 86.88219663418955,
      "grad_norm": 0.21882900595664978,
      "learning_rate": 1e-05,
      "loss": 0.9979,
      "step": 98090
    },
    {
      "epoch": 86.88662533215235,
      "grad_norm": 0.21325819194316864,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 98095
    },
    {
      "epoch": 86.89105403011514,
      "grad_norm": 0.2026692032814026,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 98100
    },
    {
      "epoch": 86.89548272807795,
      "grad_norm": 0.226902037858963,
      "learning_rate": 1e-05,
      "loss": 1.0039,
      "step": 98105
    },
    {
      "epoch": 86.89991142604075,
      "grad_norm": 0.20858511328697205,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 98110
    },
    {
      "epoch": 86.90434012400354,
      "grad_norm": 0.24579277634620667,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 98115
    },
    {
      "epoch": 86.90876882196635,
      "grad_norm": 0.22798758745193481,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 98120
    },
    {
      "epoch": 86.91319751992914,
      "grad_norm": 0.24776817858219147,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 98125
    },
    {
      "epoch": 86.91762621789194,
      "grad_norm": 0.230184867978096,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 98130
    },
    {
      "epoch": 86.92205491585474,
      "grad_norm": 0.235616073012352,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 98135
    },
    {
      "epoch": 86.92648361381754,
      "grad_norm": 0.25655829906463623,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 98140
    },
    {
      "epoch": 86.93091231178033,
      "grad_norm": 0.20392318069934845,
      "learning_rate": 1e-05,
      "loss": 0.9208,
      "step": 98145
    },
    {
      "epoch": 86.93534100974314,
      "grad_norm": 0.22574551403522491,
      "learning_rate": 1e-05,
      "loss": 0.9295,
      "step": 98150
    },
    {
      "epoch": 86.93976970770593,
      "grad_norm": 0.22375118732452393,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 98155
    },
    {
      "epoch": 86.94419840566873,
      "grad_norm": 0.2875608503818512,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 98160
    },
    {
      "epoch": 86.94862710363154,
      "grad_norm": 0.2933945655822754,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 98165
    },
    {
      "epoch": 86.95305580159433,
      "grad_norm": 0.25477349758148193,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 98170
    },
    {
      "epoch": 86.95748449955713,
      "grad_norm": 0.24826818704605103,
      "learning_rate": 1e-05,
      "loss": 0.9083,
      "step": 98175
    },
    {
      "epoch": 86.96191319751993,
      "grad_norm": 0.22769761085510254,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 98180
    },
    {
      "epoch": 86.96634189548273,
      "grad_norm": 0.23774100840091705,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 98185
    },
    {
      "epoch": 86.97077059344552,
      "grad_norm": 0.3010922968387604,
      "learning_rate": 1e-05,
      "loss": 0.906,
      "step": 98190
    },
    {
      "epoch": 86.97519929140833,
      "grad_norm": 0.2281743437051773,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 98195
    },
    {
      "epoch": 86.97962798937112,
      "grad_norm": 0.25352758169174194,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 98200
    },
    {
      "epoch": 86.98405668733392,
      "grad_norm": 0.24387317895889282,
      "learning_rate": 1e-05,
      "loss": 0.9255,
      "step": 98205
    },
    {
      "epoch": 86.98848538529673,
      "grad_norm": 0.22576482594013214,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 98210
    },
    {
      "epoch": 86.99291408325952,
      "grad_norm": 0.2359921783208847,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 98215
    },
    {
      "epoch": 86.99734278122232,
      "grad_norm": 0.2172839492559433,
      "learning_rate": 1e-05,
      "loss": 0.9176,
      "step": 98220
    },
    {
      "epoch": 87.00177147918512,
      "grad_norm": 0.22878891229629517,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 98225
    },
    {
      "epoch": 87.00620017714792,
      "grad_norm": 0.23352956771850586,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 98230
    },
    {
      "epoch": 87.01062887511071,
      "grad_norm": 0.2274307906627655,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 98235
    },
    {
      "epoch": 87.01505757307352,
      "grad_norm": 0.24780115485191345,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 98240
    },
    {
      "epoch": 87.01948627103631,
      "grad_norm": 0.23804190754890442,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 98245
    },
    {
      "epoch": 87.02391496899911,
      "grad_norm": 0.26680079102516174,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 98250
    },
    {
      "epoch": 87.02834366696192,
      "grad_norm": 0.27159383893013,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 98255
    },
    {
      "epoch": 87.03277236492471,
      "grad_norm": 0.2324649691581726,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 98260
    },
    {
      "epoch": 87.0372010628875,
      "grad_norm": 0.2546885907649994,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 98265
    },
    {
      "epoch": 87.04162976085031,
      "grad_norm": 0.2547507882118225,
      "learning_rate": 1e-05,
      "loss": 0.9996,
      "step": 98270
    },
    {
      "epoch": 87.0460584588131,
      "grad_norm": 0.2586901783943176,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 98275
    },
    {
      "epoch": 87.0504871567759,
      "grad_norm": 0.2597416639328003,
      "learning_rate": 1e-05,
      "loss": 0.9057,
      "step": 98280
    },
    {
      "epoch": 87.05491585473871,
      "grad_norm": 0.27232998609542847,
      "learning_rate": 1e-05,
      "loss": 0.9948,
      "step": 98285
    },
    {
      "epoch": 87.0593445527015,
      "grad_norm": 0.1909365952014923,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 98290
    },
    {
      "epoch": 87.0637732506643,
      "grad_norm": 0.20015062391757965,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 98295
    },
    {
      "epoch": 87.0682019486271,
      "grad_norm": 0.23849888145923615,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 98300
    },
    {
      "epoch": 87.0726306465899,
      "grad_norm": 0.2270723283290863,
      "learning_rate": 1e-05,
      "loss": 1.0301,
      "step": 98305
    },
    {
      "epoch": 87.07705934455271,
      "grad_norm": 0.23387481272220612,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 98310
    },
    {
      "epoch": 87.0814880425155,
      "grad_norm": 0.27684634923934937,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 98315
    },
    {
      "epoch": 87.0859167404783,
      "grad_norm": 0.23975835740566254,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 98320
    },
    {
      "epoch": 87.0903454384411,
      "grad_norm": 0.27283918857574463,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 98325
    },
    {
      "epoch": 87.0947741364039,
      "grad_norm": 0.23532938957214355,
      "learning_rate": 1e-05,
      "loss": 1.0117,
      "step": 98330
    },
    {
      "epoch": 87.0992028343667,
      "grad_norm": 0.2911739647388458,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 98335
    },
    {
      "epoch": 87.1036315323295,
      "grad_norm": 0.24351496994495392,
      "learning_rate": 1e-05,
      "loss": 0.9295,
      "step": 98340
    },
    {
      "epoch": 87.1080602302923,
      "grad_norm": 0.23794305324554443,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 98345
    },
    {
      "epoch": 87.11248892825509,
      "grad_norm": 0.23275792598724365,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 98350
    },
    {
      "epoch": 87.1169176262179,
      "grad_norm": 0.24212241172790527,
      "learning_rate": 1e-05,
      "loss": 0.9096,
      "step": 98355
    },
    {
      "epoch": 87.12134632418069,
      "grad_norm": 0.22095568478107452,
      "learning_rate": 1e-05,
      "loss": 0.9113,
      "step": 98360
    },
    {
      "epoch": 87.12577502214349,
      "grad_norm": 0.20863617956638336,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 98365
    },
    {
      "epoch": 87.1302037201063,
      "grad_norm": 0.21951468288898468,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 98370
    },
    {
      "epoch": 87.13463241806909,
      "grad_norm": 0.22290784120559692,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 98375
    },
    {
      "epoch": 87.13906111603188,
      "grad_norm": 0.22808930277824402,
      "learning_rate": 1e-05,
      "loss": 1.0361,
      "step": 98380
    },
    {
      "epoch": 87.14348981399469,
      "grad_norm": 0.2338210493326187,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 98385
    },
    {
      "epoch": 87.14791851195749,
      "grad_norm": 0.26032376289367676,
      "learning_rate": 1e-05,
      "loss": 0.9348,
      "step": 98390
    },
    {
      "epoch": 87.15234720992028,
      "grad_norm": 0.2538096010684967,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 98395
    },
    {
      "epoch": 87.15677590788309,
      "grad_norm": 0.21519728004932404,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 98400
    },
    {
      "epoch": 87.16120460584588,
      "grad_norm": 0.2721410393714905,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 98405
    },
    {
      "epoch": 87.16563330380868,
      "grad_norm": 0.24952802062034607,
      "learning_rate": 1e-05,
      "loss": 0.9156,
      "step": 98410
    },
    {
      "epoch": 87.17006200177148,
      "grad_norm": 0.23069506883621216,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 98415
    },
    {
      "epoch": 87.17449069973428,
      "grad_norm": 0.2460014373064041,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 98420
    },
    {
      "epoch": 87.17891939769707,
      "grad_norm": 0.22733376920223236,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 98425
    },
    {
      "epoch": 87.18334809565988,
      "grad_norm": 0.275743305683136,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 98430
    },
    {
      "epoch": 87.18777679362267,
      "grad_norm": 0.24603989720344543,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 98435
    },
    {
      "epoch": 87.19220549158547,
      "grad_norm": 0.21844574809074402,
      "learning_rate": 1e-05,
      "loss": 0.9016,
      "step": 98440
    },
    {
      "epoch": 87.19663418954828,
      "grad_norm": 0.23349353671073914,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 98445
    },
    {
      "epoch": 87.20106288751107,
      "grad_norm": 0.2166444957256317,
      "learning_rate": 1e-05,
      "loss": 0.9212,
      "step": 98450
    },
    {
      "epoch": 87.20549158547387,
      "grad_norm": 0.2517101466655731,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 98455
    },
    {
      "epoch": 87.20992028343667,
      "grad_norm": 0.21668767929077148,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 98460
    },
    {
      "epoch": 87.21434898139947,
      "grad_norm": 0.19566695392131805,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 98465
    },
    {
      "epoch": 87.21877767936226,
      "grad_norm": 0.2491917759180069,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 98470
    },
    {
      "epoch": 87.22320637732507,
      "grad_norm": 0.23572002351284027,
      "learning_rate": 1e-05,
      "loss": 0.9346,
      "step": 98475
    },
    {
      "epoch": 87.22763507528786,
      "grad_norm": 0.20109450817108154,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 98480
    },
    {
      "epoch": 87.23206377325066,
      "grad_norm": 0.25902730226516724,
      "learning_rate": 1e-05,
      "loss": 0.9119,
      "step": 98485
    },
    {
      "epoch": 87.23649247121347,
      "grad_norm": 0.24835005402565002,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 98490
    },
    {
      "epoch": 87.24092116917626,
      "grad_norm": 0.2484293133020401,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 98495
    },
    {
      "epoch": 87.24534986713905,
      "grad_norm": 0.22117173671722412,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 98500
    },
    {
      "epoch": 87.24977856510186,
      "grad_norm": 0.20593170821666718,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 98505
    },
    {
      "epoch": 87.25420726306466,
      "grad_norm": 0.22248238325119019,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 98510
    },
    {
      "epoch": 87.25863596102745,
      "grad_norm": 0.2633008658885956,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 98515
    },
    {
      "epoch": 87.26306465899026,
      "grad_norm": 0.2506774663925171,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 98520
    },
    {
      "epoch": 87.26749335695305,
      "grad_norm": 0.28996458649635315,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 98525
    },
    {
      "epoch": 87.27192205491585,
      "grad_norm": 0.21098190546035767,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 98530
    },
    {
      "epoch": 87.27635075287866,
      "grad_norm": 0.23079082369804382,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 98535
    },
    {
      "epoch": 87.28077945084145,
      "grad_norm": 0.2727959454059601,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 98540
    },
    {
      "epoch": 87.28520814880426,
      "grad_norm": 0.2653674781322479,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 98545
    },
    {
      "epoch": 87.28963684676705,
      "grad_norm": 0.2709977924823761,
      "learning_rate": 1e-05,
      "loss": 1.0632,
      "step": 98550
    },
    {
      "epoch": 87.29406554472985,
      "grad_norm": 0.27767375111579895,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 98555
    },
    {
      "epoch": 87.29849424269266,
      "grad_norm": 0.3140869140625,
      "learning_rate": 1e-05,
      "loss": 0.9264,
      "step": 98560
    },
    {
      "epoch": 87.30292294065545,
      "grad_norm": 0.23338013887405396,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 98565
    },
    {
      "epoch": 87.30735163861824,
      "grad_norm": 0.25078892707824707,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 98570
    },
    {
      "epoch": 87.31178033658105,
      "grad_norm": 0.24289509654045105,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 98575
    },
    {
      "epoch": 87.31620903454385,
      "grad_norm": 0.24767766892910004,
      "learning_rate": 1e-05,
      "loss": 0.9244,
      "step": 98580
    },
    {
      "epoch": 87.32063773250664,
      "grad_norm": 0.24860763549804688,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 98585
    },
    {
      "epoch": 87.32506643046945,
      "grad_norm": 0.22739925980567932,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 98590
    },
    {
      "epoch": 87.32949512843224,
      "grad_norm": 0.27999404072761536,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 98595
    },
    {
      "epoch": 87.33392382639504,
      "grad_norm": 0.23120373487472534,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 98600
    },
    {
      "epoch": 87.33835252435784,
      "grad_norm": 0.2214059978723526,
      "learning_rate": 1e-05,
      "loss": 1.0179,
      "step": 98605
    },
    {
      "epoch": 87.34278122232064,
      "grad_norm": 0.25575026869773865,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 98610
    },
    {
      "epoch": 87.34720992028343,
      "grad_norm": 0.20689494907855988,
      "learning_rate": 1e-05,
      "loss": 0.9231,
      "step": 98615
    },
    {
      "epoch": 87.35163861824624,
      "grad_norm": 0.23888923227787018,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 98620
    },
    {
      "epoch": 87.35606731620904,
      "grad_norm": 0.2670289874076843,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 98625
    },
    {
      "epoch": 87.36049601417183,
      "grad_norm": 0.29295748472213745,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 98630
    },
    {
      "epoch": 87.36492471213464,
      "grad_norm": 0.24720269441604614,
      "learning_rate": 1e-05,
      "loss": 0.9303,
      "step": 98635
    },
    {
      "epoch": 87.36935341009743,
      "grad_norm": 0.23388172686100006,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 98640
    },
    {
      "epoch": 87.37378210806023,
      "grad_norm": 0.21629522740840912,
      "learning_rate": 1e-05,
      "loss": 1.0148,
      "step": 98645
    },
    {
      "epoch": 87.37821080602303,
      "grad_norm": 0.25006499886512756,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 98650
    },
    {
      "epoch": 87.38263950398583,
      "grad_norm": 0.22009262442588806,
      "learning_rate": 1e-05,
      "loss": 0.9104,
      "step": 98655
    },
    {
      "epoch": 87.38706820194862,
      "grad_norm": 0.2552126348018646,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 98660
    },
    {
      "epoch": 87.39149689991143,
      "grad_norm": 0.2860468924045563,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 98665
    },
    {
      "epoch": 87.39592559787422,
      "grad_norm": 0.24340319633483887,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 98670
    },
    {
      "epoch": 87.40035429583702,
      "grad_norm": 0.22523324191570282,
      "learning_rate": 1e-05,
      "loss": 0.9245,
      "step": 98675
    },
    {
      "epoch": 87.40478299379983,
      "grad_norm": 0.2452406883239746,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 98680
    },
    {
      "epoch": 87.40921169176262,
      "grad_norm": 0.25363871455192566,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 98685
    },
    {
      "epoch": 87.41364038972542,
      "grad_norm": 0.22660808265209198,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 98690
    },
    {
      "epoch": 87.41806908768822,
      "grad_norm": 0.2833593785762787,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 98695
    },
    {
      "epoch": 87.42249778565102,
      "grad_norm": 0.2170831710100174,
      "learning_rate": 1e-05,
      "loss": 0.8968,
      "step": 98700
    },
    {
      "epoch": 87.42692648361381,
      "grad_norm": 0.24000315368175507,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 98705
    },
    {
      "epoch": 87.43135518157662,
      "grad_norm": 0.23038068413734436,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 98710
    },
    {
      "epoch": 87.43578387953941,
      "grad_norm": 0.24463027715682983,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 98715
    },
    {
      "epoch": 87.44021257750221,
      "grad_norm": 0.25267237424850464,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 98720
    },
    {
      "epoch": 87.44464127546502,
      "grad_norm": 0.21883666515350342,
      "learning_rate": 1e-05,
      "loss": 0.9311,
      "step": 98725
    },
    {
      "epoch": 87.44906997342781,
      "grad_norm": 0.23028209805488586,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 98730
    },
    {
      "epoch": 87.4534986713906,
      "grad_norm": 0.2352771759033203,
      "learning_rate": 1e-05,
      "loss": 0.9045,
      "step": 98735
    },
    {
      "epoch": 87.45792736935341,
      "grad_norm": 0.234528049826622,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 98740
    },
    {
      "epoch": 87.46235606731621,
      "grad_norm": 0.2326662689447403,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 98745
    },
    {
      "epoch": 87.466784765279,
      "grad_norm": 0.25211647152900696,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 98750
    },
    {
      "epoch": 87.47121346324181,
      "grad_norm": 0.21470589935779572,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 98755
    },
    {
      "epoch": 87.4756421612046,
      "grad_norm": 0.29836538434028625,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 98760
    },
    {
      "epoch": 87.4800708591674,
      "grad_norm": 0.2584453225135803,
      "learning_rate": 1e-05,
      "loss": 0.9122,
      "step": 98765
    },
    {
      "epoch": 87.4844995571302,
      "grad_norm": 0.20180445909500122,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 98770
    },
    {
      "epoch": 87.488928255093,
      "grad_norm": 0.24020785093307495,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 98775
    },
    {
      "epoch": 87.4933569530558,
      "grad_norm": 0.22691568732261658,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 98780
    },
    {
      "epoch": 87.4977856510186,
      "grad_norm": 0.2516210973262787,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 98785
    },
    {
      "epoch": 87.5022143489814,
      "grad_norm": 0.22353962063789368,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 98790
    },
    {
      "epoch": 87.5066430469442,
      "grad_norm": 0.26746803522109985,
      "learning_rate": 1e-05,
      "loss": 1.0186,
      "step": 98795
    },
    {
      "epoch": 87.511071744907,
      "grad_norm": 0.27643832564353943,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 98800
    },
    {
      "epoch": 87.5155004428698,
      "grad_norm": 0.2607962191104889,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 98805
    },
    {
      "epoch": 87.5199291408326,
      "grad_norm": 0.25980764627456665,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 98810
    },
    {
      "epoch": 87.5243578387954,
      "grad_norm": 0.2470128983259201,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 98815
    },
    {
      "epoch": 87.52878653675819,
      "grad_norm": 0.24734635651111603,
      "learning_rate": 1e-05,
      "loss": 1.0,
      "step": 98820
    },
    {
      "epoch": 87.533215234721,
      "grad_norm": 0.24346600472927094,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 98825
    },
    {
      "epoch": 87.53764393268379,
      "grad_norm": 0.2174825370311737,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 98830
    },
    {
      "epoch": 87.54207263064659,
      "grad_norm": 0.24301636219024658,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 98835
    },
    {
      "epoch": 87.5465013286094,
      "grad_norm": 0.2106200009584427,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 98840
    },
    {
      "epoch": 87.55093002657219,
      "grad_norm": 0.2418515682220459,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 98845
    },
    {
      "epoch": 87.55535872453498,
      "grad_norm": 0.25311174988746643,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 98850
    },
    {
      "epoch": 87.55978742249779,
      "grad_norm": 0.19423292577266693,
      "learning_rate": 1e-05,
      "loss": 0.9168,
      "step": 98855
    },
    {
      "epoch": 87.56421612046059,
      "grad_norm": 0.24214698374271393,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 98860
    },
    {
      "epoch": 87.56864481842338,
      "grad_norm": 0.20991046726703644,
      "learning_rate": 1e-05,
      "loss": 0.9279,
      "step": 98865
    },
    {
      "epoch": 87.57307351638619,
      "grad_norm": 0.2326023131608963,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 98870
    },
    {
      "epoch": 87.57750221434898,
      "grad_norm": 0.26416683197021484,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 98875
    },
    {
      "epoch": 87.58193091231178,
      "grad_norm": 0.21936079859733582,
      "learning_rate": 1e-05,
      "loss": 0.9007,
      "step": 98880
    },
    {
      "epoch": 87.58635961027458,
      "grad_norm": 0.23663152754306793,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 98885
    },
    {
      "epoch": 87.59078830823738,
      "grad_norm": 0.2777262330055237,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 98890
    },
    {
      "epoch": 87.59521700620017,
      "grad_norm": 0.22472350299358368,
      "learning_rate": 1e-05,
      "loss": 0.9095,
      "step": 98895
    },
    {
      "epoch": 87.59964570416298,
      "grad_norm": 0.3032849431037903,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 98900
    },
    {
      "epoch": 87.60407440212578,
      "grad_norm": 0.30397355556488037,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 98905
    },
    {
      "epoch": 87.60850310008857,
      "grad_norm": 0.29726454615592957,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 98910
    },
    {
      "epoch": 87.61293179805138,
      "grad_norm": 0.30529820919036865,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 98915
    },
    {
      "epoch": 87.61736049601417,
      "grad_norm": 0.25008270144462585,
      "learning_rate": 1e-05,
      "loss": 1.0253,
      "step": 98920
    },
    {
      "epoch": 87.62178919397697,
      "grad_norm": 0.247336283326149,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 98925
    },
    {
      "epoch": 87.62621789193977,
      "grad_norm": 0.21041858196258545,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 98930
    },
    {
      "epoch": 87.63064658990257,
      "grad_norm": 0.20626723766326904,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 98935
    },
    {
      "epoch": 87.63507528786536,
      "grad_norm": 0.2532220184803009,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 98940
    },
    {
      "epoch": 87.63950398582817,
      "grad_norm": 0.24755412340164185,
      "learning_rate": 1e-05,
      "loss": 0.9176,
      "step": 98945
    },
    {
      "epoch": 87.64393268379096,
      "grad_norm": 0.23819641768932343,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 98950
    },
    {
      "epoch": 87.64836138175376,
      "grad_norm": 0.22585979104042053,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 98955
    },
    {
      "epoch": 87.65279007971657,
      "grad_norm": 0.28521087765693665,
      "learning_rate": 1e-05,
      "loss": 0.9105,
      "step": 98960
    },
    {
      "epoch": 87.65721877767936,
      "grad_norm": 0.22838784754276276,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 98965
    },
    {
      "epoch": 87.66164747564216,
      "grad_norm": 0.2483397126197815,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 98970
    },
    {
      "epoch": 87.66607617360496,
      "grad_norm": 0.23880240321159363,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 98975
    },
    {
      "epoch": 87.67050487156776,
      "grad_norm": 0.31222718954086304,
      "learning_rate": 1e-05,
      "loss": 1.0046,
      "step": 98980
    },
    {
      "epoch": 87.67493356953055,
      "grad_norm": 0.25406596064567566,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 98985
    },
    {
      "epoch": 87.67936226749336,
      "grad_norm": 0.2309110462665558,
      "learning_rate": 1e-05,
      "loss": 0.93,
      "step": 98990
    },
    {
      "epoch": 87.68379096545615,
      "grad_norm": 0.2634873390197754,
      "learning_rate": 1e-05,
      "loss": 0.9242,
      "step": 98995
    },
    {
      "epoch": 87.68821966341895,
      "grad_norm": 0.2261311262845993,
      "learning_rate": 1e-05,
      "loss": 0.9652,
      "step": 99000
    },
    {
      "epoch": 87.69264836138176,
      "grad_norm": 0.2464703917503357,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 99005
    },
    {
      "epoch": 87.69707705934455,
      "grad_norm": 0.22071000933647156,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 99010
    },
    {
      "epoch": 87.70150575730734,
      "grad_norm": 0.2115013599395752,
      "learning_rate": 1e-05,
      "loss": 0.9237,
      "step": 99015
    },
    {
      "epoch": 87.70593445527015,
      "grad_norm": 0.21408239006996155,
      "learning_rate": 1e-05,
      "loss": 0.9131,
      "step": 99020
    },
    {
      "epoch": 87.71036315323295,
      "grad_norm": 0.23356370627880096,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 99025
    },
    {
      "epoch": 87.71479185119574,
      "grad_norm": 0.20626841485500336,
      "learning_rate": 1e-05,
      "loss": 1.0212,
      "step": 99030
    },
    {
      "epoch": 87.71922054915855,
      "grad_norm": 0.26001638174057007,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 99035
    },
    {
      "epoch": 87.72364924712134,
      "grad_norm": 0.2766408920288086,
      "learning_rate": 1e-05,
      "loss": 1.0407,
      "step": 99040
    },
    {
      "epoch": 87.72807794508415,
      "grad_norm": 0.23147602379322052,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 99045
    },
    {
      "epoch": 87.73250664304695,
      "grad_norm": 0.2566184401512146,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 99050
    },
    {
      "epoch": 87.73693534100974,
      "grad_norm": 0.22532761096954346,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 99055
    },
    {
      "epoch": 87.74136403897255,
      "grad_norm": 0.21976672112941742,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 99060
    },
    {
      "epoch": 87.74579273693534,
      "grad_norm": 0.25288182497024536,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 99065
    },
    {
      "epoch": 87.75022143489814,
      "grad_norm": 0.23161187767982483,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 99070
    },
    {
      "epoch": 87.75465013286095,
      "grad_norm": 0.23595760762691498,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 99075
    },
    {
      "epoch": 87.75907883082374,
      "grad_norm": 0.2467581033706665,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 99080
    },
    {
      "epoch": 87.76350752878653,
      "grad_norm": 0.21384133398532867,
      "learning_rate": 1e-05,
      "loss": 0.9007,
      "step": 99085
    },
    {
      "epoch": 87.76793622674934,
      "grad_norm": 0.2870183289051056,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 99090
    },
    {
      "epoch": 87.77236492471214,
      "grad_norm": 0.22450602054595947,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 99095
    },
    {
      "epoch": 87.77679362267493,
      "grad_norm": 0.24323642253875732,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 99100
    },
    {
      "epoch": 87.78122232063774,
      "grad_norm": 0.26496610045433044,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 99105
    },
    {
      "epoch": 87.78565101860053,
      "grad_norm": 0.23802362382411957,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 99110
    },
    {
      "epoch": 87.79007971656333,
      "grad_norm": 0.2635282278060913,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 99115
    },
    {
      "epoch": 87.79450841452613,
      "grad_norm": 0.22189420461654663,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 99120
    },
    {
      "epoch": 87.79893711248893,
      "grad_norm": 0.24880361557006836,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 99125
    },
    {
      "epoch": 87.80336581045172,
      "grad_norm": 0.24588751792907715,
      "learning_rate": 1e-05,
      "loss": 0.9066,
      "step": 99130
    },
    {
      "epoch": 87.80779450841453,
      "grad_norm": 0.25989171862602234,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 99135
    },
    {
      "epoch": 87.81222320637733,
      "grad_norm": 0.2163795381784439,
      "learning_rate": 1e-05,
      "loss": 0.9202,
      "step": 99140
    },
    {
      "epoch": 87.81665190434012,
      "grad_norm": 0.24971000850200653,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 99145
    },
    {
      "epoch": 87.82108060230293,
      "grad_norm": 0.22685950994491577,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 99150
    },
    {
      "epoch": 87.82550930026572,
      "grad_norm": 0.23735003173351288,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 99155
    },
    {
      "epoch": 87.82993799822852,
      "grad_norm": 0.2772241234779358,
      "learning_rate": 1e-05,
      "loss": 0.926,
      "step": 99160
    },
    {
      "epoch": 87.83436669619132,
      "grad_norm": 0.21943655610084534,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 99165
    },
    {
      "epoch": 87.83879539415412,
      "grad_norm": 0.241880863904953,
      "learning_rate": 1e-05,
      "loss": 0.9348,
      "step": 99170
    },
    {
      "epoch": 87.84322409211691,
      "grad_norm": 0.26835131645202637,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 99175
    },
    {
      "epoch": 87.84765279007972,
      "grad_norm": 0.2499963492155075,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 99180
    },
    {
      "epoch": 87.85208148804251,
      "grad_norm": 0.21494348347187042,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 99185
    },
    {
      "epoch": 87.85651018600531,
      "grad_norm": 0.2531016767024994,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 99190
    },
    {
      "epoch": 87.86093888396812,
      "grad_norm": 0.2302526831626892,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 99195
    },
    {
      "epoch": 87.86536758193091,
      "grad_norm": 0.24988315999507904,
      "learning_rate": 1e-05,
      "loss": 0.9632,
      "step": 99200
    },
    {
      "epoch": 87.8697962798937,
      "grad_norm": 0.2678731083869934,
      "learning_rate": 1e-05,
      "loss": 0.8867,
      "step": 99205
    },
    {
      "epoch": 87.87422497785651,
      "grad_norm": 0.23875993490219116,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 99210
    },
    {
      "epoch": 87.87865367581931,
      "grad_norm": 0.21770165860652924,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 99215
    },
    {
      "epoch": 87.8830823737821,
      "grad_norm": 0.22409257292747498,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 99220
    },
    {
      "epoch": 87.88751107174491,
      "grad_norm": 0.24944385886192322,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 99225
    },
    {
      "epoch": 87.8919397697077,
      "grad_norm": 0.24507126212120056,
      "learning_rate": 1e-05,
      "loss": 0.9004,
      "step": 99230
    },
    {
      "epoch": 87.8963684676705,
      "grad_norm": 0.24941736459732056,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 99235
    },
    {
      "epoch": 87.9007971656333,
      "grad_norm": 0.2289191484451294,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 99240
    },
    {
      "epoch": 87.9052258635961,
      "grad_norm": 0.20393109321594238,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 99245
    },
    {
      "epoch": 87.9096545615589,
      "grad_norm": 0.26088833808898926,
      "learning_rate": 1e-05,
      "loss": 1.0182,
      "step": 99250
    },
    {
      "epoch": 87.9140832595217,
      "grad_norm": 0.259015828371048,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 99255
    },
    {
      "epoch": 87.9185119574845,
      "grad_norm": 0.27033761143684387,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 99260
    },
    {
      "epoch": 87.92294065544729,
      "grad_norm": 0.19690492749214172,
      "learning_rate": 1e-05,
      "loss": 0.9556,
      "step": 99265
    },
    {
      "epoch": 87.9273693534101,
      "grad_norm": 0.2180897295475006,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 99270
    },
    {
      "epoch": 87.9317980513729,
      "grad_norm": 0.25377869606018066,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 99275
    },
    {
      "epoch": 87.9362267493357,
      "grad_norm": 0.2251621037721634,
      "learning_rate": 1e-05,
      "loss": 0.8883,
      "step": 99280
    },
    {
      "epoch": 87.9406554472985,
      "grad_norm": 0.25489184260368347,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 99285
    },
    {
      "epoch": 87.94508414526129,
      "grad_norm": 0.21676458418369293,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 99290
    },
    {
      "epoch": 87.9495128432241,
      "grad_norm": 0.2695954442024231,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 99295
    },
    {
      "epoch": 87.9539415411869,
      "grad_norm": 0.2721954882144928,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 99300
    },
    {
      "epoch": 87.95837023914969,
      "grad_norm": 0.22035151720046997,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 99305
    },
    {
      "epoch": 87.9627989371125,
      "grad_norm": 0.20629064738750458,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 99310
    },
    {
      "epoch": 87.96722763507529,
      "grad_norm": 0.25649625062942505,
      "learning_rate": 1e-05,
      "loss": 0.9918,
      "step": 99315
    },
    {
      "epoch": 87.97165633303808,
      "grad_norm": 0.2524641156196594,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 99320
    },
    {
      "epoch": 87.97608503100089,
      "grad_norm": 0.24857378005981445,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 99325
    },
    {
      "epoch": 87.98051372896369,
      "grad_norm": 0.23496681451797485,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 99330
    },
    {
      "epoch": 87.98494242692648,
      "grad_norm": 0.2633059620857239,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 99335
    },
    {
      "epoch": 87.98937112488929,
      "grad_norm": 0.2400464117527008,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 99340
    },
    {
      "epoch": 87.99379982285208,
      "grad_norm": 0.23970000445842743,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 99345
    },
    {
      "epoch": 87.99822852081488,
      "grad_norm": 0.24878472089767456,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 99350
    },
    {
      "epoch": 88.00265721877768,
      "grad_norm": 0.3312890827655792,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 99355
    },
    {
      "epoch": 88.00708591674048,
      "grad_norm": 0.253065824508667,
      "learning_rate": 1e-05,
      "loss": 0.8935,
      "step": 99360
    },
    {
      "epoch": 88.01151461470327,
      "grad_norm": 0.26019906997680664,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 99365
    },
    {
      "epoch": 88.01594331266608,
      "grad_norm": 0.22965823113918304,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 99370
    },
    {
      "epoch": 88.02037201062888,
      "grad_norm": 0.23158767819404602,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 99375
    },
    {
      "epoch": 88.02480070859167,
      "grad_norm": 0.24661968648433685,
      "learning_rate": 1e-05,
      "loss": 0.922,
      "step": 99380
    },
    {
      "epoch": 88.02922940655448,
      "grad_norm": 0.26257190108299255,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 99385
    },
    {
      "epoch": 88.03365810451727,
      "grad_norm": 0.27869170904159546,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 99390
    },
    {
      "epoch": 88.03808680248007,
      "grad_norm": 0.2570676803588867,
      "learning_rate": 1e-05,
      "loss": 0.8894,
      "step": 99395
    },
    {
      "epoch": 88.04251550044287,
      "grad_norm": 0.25156939029693604,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 99400
    },
    {
      "epoch": 88.04694419840567,
      "grad_norm": 0.21452170610427856,
      "learning_rate": 1e-05,
      "loss": 0.8991,
      "step": 99405
    },
    {
      "epoch": 88.05137289636846,
      "grad_norm": 0.23744866251945496,
      "learning_rate": 1e-05,
      "loss": 0.8967,
      "step": 99410
    },
    {
      "epoch": 88.05580159433127,
      "grad_norm": 0.2338162511587143,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 99415
    },
    {
      "epoch": 88.06023029229407,
      "grad_norm": 0.250712126493454,
      "learning_rate": 1e-05,
      "loss": 0.9116,
      "step": 99420
    },
    {
      "epoch": 88.06465899025686,
      "grad_norm": 0.22308313846588135,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 99425
    },
    {
      "epoch": 88.06908768821967,
      "grad_norm": 0.21670743823051453,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 99430
    },
    {
      "epoch": 88.07351638618246,
      "grad_norm": 0.23527945578098297,
      "learning_rate": 1e-05,
      "loss": 0.9809,
      "step": 99435
    },
    {
      "epoch": 88.07794508414526,
      "grad_norm": 0.23743298649787903,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 99440
    },
    {
      "epoch": 88.08237378210806,
      "grad_norm": 0.2462809681892395,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 99445
    },
    {
      "epoch": 88.08680248007086,
      "grad_norm": 0.26181820034980774,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 99450
    },
    {
      "epoch": 88.09123117803365,
      "grad_norm": 0.23750481009483337,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 99455
    },
    {
      "epoch": 88.09565987599646,
      "grad_norm": 0.27057144045829773,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 99460
    },
    {
      "epoch": 88.10008857395925,
      "grad_norm": 0.2528458535671234,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 99465
    },
    {
      "epoch": 88.10451727192205,
      "grad_norm": 0.2581286132335663,
      "learning_rate": 1e-05,
      "loss": 0.9413,
      "step": 99470
    },
    {
      "epoch": 88.10894596988486,
      "grad_norm": 0.24771030247211456,
      "learning_rate": 1e-05,
      "loss": 0.9127,
      "step": 99475
    },
    {
      "epoch": 88.11337466784765,
      "grad_norm": 0.28207796812057495,
      "learning_rate": 1e-05,
      "loss": 0.908,
      "step": 99480
    },
    {
      "epoch": 88.11780336581045,
      "grad_norm": 0.25008776783943176,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 99485
    },
    {
      "epoch": 88.12223206377325,
      "grad_norm": 0.22862808406352997,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 99490
    },
    {
      "epoch": 88.12666076173605,
      "grad_norm": 0.22968505322933197,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 99495
    },
    {
      "epoch": 88.13108945969884,
      "grad_norm": 0.2297797054052353,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 99500
    },
    {
      "epoch": 88.13551815766165,
      "grad_norm": 0.2316872924566269,
      "learning_rate": 1e-05,
      "loss": 1.0568,
      "step": 99505
    },
    {
      "epoch": 88.13994685562444,
      "grad_norm": 0.2860967516899109,
      "learning_rate": 1e-05,
      "loss": 0.9026,
      "step": 99510
    },
    {
      "epoch": 88.14437555358724,
      "grad_norm": 0.22373518347740173,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 99515
    },
    {
      "epoch": 88.14880425155005,
      "grad_norm": 0.2924445569515228,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 99520
    },
    {
      "epoch": 88.15323294951284,
      "grad_norm": 0.24673527479171753,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 99525
    },
    {
      "epoch": 88.15766164747565,
      "grad_norm": 0.2519305348396301,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 99530
    },
    {
      "epoch": 88.16209034543844,
      "grad_norm": 0.2540664076805115,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 99535
    },
    {
      "epoch": 88.16651904340124,
      "grad_norm": 0.28010979294776917,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 99540
    },
    {
      "epoch": 88.17094774136405,
      "grad_norm": 0.2572025954723358,
      "learning_rate": 1e-05,
      "loss": 0.9543,
      "step": 99545
    },
    {
      "epoch": 88.17537643932684,
      "grad_norm": 0.2626935541629791,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 99550
    },
    {
      "epoch": 88.17980513728963,
      "grad_norm": 0.23510734736919403,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 99555
    },
    {
      "epoch": 88.18423383525244,
      "grad_norm": 0.2211681753396988,
      "learning_rate": 1e-05,
      "loss": 0.9962,
      "step": 99560
    },
    {
      "epoch": 88.18866253321524,
      "grad_norm": 0.25278133153915405,
      "learning_rate": 1e-05,
      "loss": 1.0079,
      "step": 99565
    },
    {
      "epoch": 88.19309123117803,
      "grad_norm": 0.2636018395423889,
      "learning_rate": 1e-05,
      "loss": 0.9203,
      "step": 99570
    },
    {
      "epoch": 88.19751992914084,
      "grad_norm": 0.29537665843963623,
      "learning_rate": 1e-05,
      "loss": 1.0384,
      "step": 99575
    },
    {
      "epoch": 88.20194862710363,
      "grad_norm": 0.2552962601184845,
      "learning_rate": 1e-05,
      "loss": 0.974,
      "step": 99580
    },
    {
      "epoch": 88.20637732506643,
      "grad_norm": 0.2730923891067505,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 99585
    },
    {
      "epoch": 88.21080602302924,
      "grad_norm": 0.27667036652565,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 99590
    },
    {
      "epoch": 88.21523472099203,
      "grad_norm": 0.23600807785987854,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 99595
    },
    {
      "epoch": 88.21966341895482,
      "grad_norm": 0.23295293748378754,
      "learning_rate": 1e-05,
      "loss": 0.9267,
      "step": 99600
    },
    {
      "epoch": 88.22409211691763,
      "grad_norm": 0.22966763377189636,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 99605
    },
    {
      "epoch": 88.22852081488043,
      "grad_norm": 0.23855608701705933,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 99610
    },
    {
      "epoch": 88.23294951284322,
      "grad_norm": 0.2854846119880676,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 99615
    },
    {
      "epoch": 88.23737821080603,
      "grad_norm": 0.23175206780433655,
      "learning_rate": 1e-05,
      "loss": 0.8973,
      "step": 99620
    },
    {
      "epoch": 88.24180690876882,
      "grad_norm": 0.23563654720783234,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 99625
    },
    {
      "epoch": 88.24623560673162,
      "grad_norm": 0.2592059373855591,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 99630
    },
    {
      "epoch": 88.25066430469442,
      "grad_norm": 0.27779364585876465,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 99635
    },
    {
      "epoch": 88.25509300265722,
      "grad_norm": 0.21161754429340363,
      "learning_rate": 1e-05,
      "loss": 0.9338,
      "step": 99640
    },
    {
      "epoch": 88.25952170062001,
      "grad_norm": 0.2226719707250595,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 99645
    },
    {
      "epoch": 88.26395039858282,
      "grad_norm": 0.2077556699514389,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 99650
    },
    {
      "epoch": 88.26837909654562,
      "grad_norm": 0.2503184676170349,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 99655
    },
    {
      "epoch": 88.27280779450841,
      "grad_norm": 0.26173096895217896,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 99660
    },
    {
      "epoch": 88.27723649247122,
      "grad_norm": 0.24127070605754852,
      "learning_rate": 1e-05,
      "loss": 0.9147,
      "step": 99665
    },
    {
      "epoch": 88.28166519043401,
      "grad_norm": 0.26643872261047363,
      "learning_rate": 1e-05,
      "loss": 0.9091,
      "step": 99670
    },
    {
      "epoch": 88.2860938883968,
      "grad_norm": 0.22323083877563477,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 99675
    },
    {
      "epoch": 88.29052258635961,
      "grad_norm": 0.2844523787498474,
      "learning_rate": 1e-05,
      "loss": 0.902,
      "step": 99680
    },
    {
      "epoch": 88.29495128432241,
      "grad_norm": 0.25690388679504395,
      "learning_rate": 1e-05,
      "loss": 0.9819,
      "step": 99685
    },
    {
      "epoch": 88.2993799822852,
      "grad_norm": 0.2573966681957245,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 99690
    },
    {
      "epoch": 88.30380868024801,
      "grad_norm": 0.20467689633369446,
      "learning_rate": 1e-05,
      "loss": 0.8746,
      "step": 99695
    },
    {
      "epoch": 88.3082373782108,
      "grad_norm": 0.2291315495967865,
      "learning_rate": 1e-05,
      "loss": 0.9294,
      "step": 99700
    },
    {
      "epoch": 88.3126660761736,
      "grad_norm": 0.2611003816127777,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 99705
    },
    {
      "epoch": 88.31709477413641,
      "grad_norm": 0.22807160019874573,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 99710
    },
    {
      "epoch": 88.3215234720992,
      "grad_norm": 0.20707808434963226,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 99715
    },
    {
      "epoch": 88.325952170062,
      "grad_norm": 0.23027831315994263,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 99720
    },
    {
      "epoch": 88.3303808680248,
      "grad_norm": 0.22548729181289673,
      "learning_rate": 1e-05,
      "loss": 0.9848,
      "step": 99725
    },
    {
      "epoch": 88.3348095659876,
      "grad_norm": 0.2556377649307251,
      "learning_rate": 1e-05,
      "loss": 0.9176,
      "step": 99730
    },
    {
      "epoch": 88.33923826395039,
      "grad_norm": 0.2506229281425476,
      "learning_rate": 1e-05,
      "loss": 0.9219,
      "step": 99735
    },
    {
      "epoch": 88.3436669619132,
      "grad_norm": 0.20109428465366364,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 99740
    },
    {
      "epoch": 88.348095659876,
      "grad_norm": 0.2601589560508728,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 99745
    },
    {
      "epoch": 88.35252435783879,
      "grad_norm": 0.23489879071712494,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 99750
    },
    {
      "epoch": 88.3569530558016,
      "grad_norm": 0.2710605263710022,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 99755
    },
    {
      "epoch": 88.36138175376439,
      "grad_norm": 0.2892511188983917,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 99760
    },
    {
      "epoch": 88.36581045172719,
      "grad_norm": 0.25111764669418335,
      "learning_rate": 1e-05,
      "loss": 0.9576,
      "step": 99765
    },
    {
      "epoch": 88.37023914969,
      "grad_norm": 0.24800343811511993,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 99770
    },
    {
      "epoch": 88.37466784765279,
      "grad_norm": 0.24540822207927704,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 99775
    },
    {
      "epoch": 88.3790965456156,
      "grad_norm": 0.21082238852977753,
      "learning_rate": 1e-05,
      "loss": 0.9087,
      "step": 99780
    },
    {
      "epoch": 88.38352524357839,
      "grad_norm": 0.21847842633724213,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 99785
    },
    {
      "epoch": 88.38795394154118,
      "grad_norm": 0.24009904265403748,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 99790
    },
    {
      "epoch": 88.39238263950399,
      "grad_norm": 0.23809295892715454,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 99795
    },
    {
      "epoch": 88.39681133746679,
      "grad_norm": 0.24205884337425232,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 99800
    },
    {
      "epoch": 88.40124003542958,
      "grad_norm": 0.2568868398666382,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 99805
    },
    {
      "epoch": 88.40566873339239,
      "grad_norm": 0.2555282413959503,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 99810
    },
    {
      "epoch": 88.41009743135518,
      "grad_norm": 0.23748254776000977,
      "learning_rate": 1e-05,
      "loss": 0.9319,
      "step": 99815
    },
    {
      "epoch": 88.41452612931798,
      "grad_norm": 0.24691760540008545,
      "learning_rate": 1e-05,
      "loss": 0.9241,
      "step": 99820
    },
    {
      "epoch": 88.41895482728079,
      "grad_norm": 0.28536152839660645,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 99825
    },
    {
      "epoch": 88.42338352524358,
      "grad_norm": 0.217886820435524,
      "learning_rate": 1e-05,
      "loss": 0.9235,
      "step": 99830
    },
    {
      "epoch": 88.42781222320637,
      "grad_norm": 0.2699647545814514,
      "learning_rate": 1e-05,
      "loss": 0.9237,
      "step": 99835
    },
    {
      "epoch": 88.43224092116918,
      "grad_norm": 0.2297172099351883,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 99840
    },
    {
      "epoch": 88.43666961913198,
      "grad_norm": 0.23674717545509338,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 99845
    },
    {
      "epoch": 88.44109831709477,
      "grad_norm": 0.22396232187747955,
      "learning_rate": 1e-05,
      "loss": 0.9278,
      "step": 99850
    },
    {
      "epoch": 88.44552701505758,
      "grad_norm": 0.1978616565465927,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 99855
    },
    {
      "epoch": 88.44995571302037,
      "grad_norm": 0.22385986149311066,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 99860
    },
    {
      "epoch": 88.45438441098317,
      "grad_norm": 0.1950816959142685,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 99865
    },
    {
      "epoch": 88.45881310894598,
      "grad_norm": 0.1992218941450119,
      "learning_rate": 1e-05,
      "loss": 0.9235,
      "step": 99870
    },
    {
      "epoch": 88.46324180690877,
      "grad_norm": 0.23768770694732666,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 99875
    },
    {
      "epoch": 88.46767050487156,
      "grad_norm": 0.2011580467224121,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 99880
    },
    {
      "epoch": 88.47209920283437,
      "grad_norm": 0.21293315291404724,
      "learning_rate": 1e-05,
      "loss": 0.9228,
      "step": 99885
    },
    {
      "epoch": 88.47652790079717,
      "grad_norm": 0.22952774167060852,
      "learning_rate": 1e-05,
      "loss": 0.9249,
      "step": 99890
    },
    {
      "epoch": 88.48095659875996,
      "grad_norm": 0.2631010115146637,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 99895
    },
    {
      "epoch": 88.48538529672277,
      "grad_norm": 0.21053150296211243,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 99900
    },
    {
      "epoch": 88.48981399468556,
      "grad_norm": 0.22930829226970673,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 99905
    },
    {
      "epoch": 88.49424269264836,
      "grad_norm": 0.2742099463939667,
      "learning_rate": 1e-05,
      "loss": 1.039,
      "step": 99910
    },
    {
      "epoch": 88.49867139061116,
      "grad_norm": 0.26076045632362366,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 99915
    },
    {
      "epoch": 88.50310008857396,
      "grad_norm": 0.22787722945213318,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 99920
    },
    {
      "epoch": 88.50752878653675,
      "grad_norm": 0.24036970734596252,
      "learning_rate": 1e-05,
      "loss": 0.9528,
      "step": 99925
    },
    {
      "epoch": 88.51195748449956,
      "grad_norm": 0.2343783974647522,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 99930
    },
    {
      "epoch": 88.51638618246236,
      "grad_norm": 0.24172183871269226,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 99935
    },
    {
      "epoch": 88.52081488042515,
      "grad_norm": 0.26128506660461426,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 99940
    },
    {
      "epoch": 88.52524357838796,
      "grad_norm": 0.27385762333869934,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 99945
    },
    {
      "epoch": 88.52967227635075,
      "grad_norm": 0.26363852620124817,
      "learning_rate": 1e-05,
      "loss": 0.9205,
      "step": 99950
    },
    {
      "epoch": 88.53410097431355,
      "grad_norm": 0.29527363181114197,
      "learning_rate": 1e-05,
      "loss": 1.0199,
      "step": 99955
    },
    {
      "epoch": 88.53852967227635,
      "grad_norm": 0.2481592297554016,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 99960
    },
    {
      "epoch": 88.54295837023915,
      "grad_norm": 0.30731603503227234,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 99965
    },
    {
      "epoch": 88.54738706820194,
      "grad_norm": 0.27373579144477844,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 99970
    },
    {
      "epoch": 88.55181576616475,
      "grad_norm": 0.23890335857868195,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 99975
    },
    {
      "epoch": 88.55624446412754,
      "grad_norm": 0.2325640767812729,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 99980
    },
    {
      "epoch": 88.56067316209034,
      "grad_norm": 0.2493053525686264,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 99985
    },
    {
      "epoch": 88.56510186005315,
      "grad_norm": 0.22135327756404877,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 99990
    },
    {
      "epoch": 88.56953055801594,
      "grad_norm": 0.22047559916973114,
      "learning_rate": 1e-05,
      "loss": 1.0289,
      "step": 99995
    },
    {
      "epoch": 88.57395925597874,
      "grad_norm": 0.22702765464782715,
      "learning_rate": 1e-05,
      "loss": 0.9203,
      "step": 100000
    },
    {
      "epoch": 88.57838795394154,
      "grad_norm": 0.21299892663955688,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 100005
    },
    {
      "epoch": 88.58281665190434,
      "grad_norm": 0.24758689105510712,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 100010
    },
    {
      "epoch": 88.58724534986715,
      "grad_norm": 0.2863371670246124,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 100015
    },
    {
      "epoch": 88.59167404782994,
      "grad_norm": 0.25844478607177734,
      "learning_rate": 1e-05,
      "loss": 0.9247,
      "step": 100020
    },
    {
      "epoch": 88.59610274579273,
      "grad_norm": 0.27765244245529175,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 100025
    },
    {
      "epoch": 88.60053144375554,
      "grad_norm": 0.29093942046165466,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 100030
    },
    {
      "epoch": 88.60496014171834,
      "grad_norm": 0.3137884736061096,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 100035
    },
    {
      "epoch": 88.60938883968113,
      "grad_norm": 0.30530211329460144,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 100040
    },
    {
      "epoch": 88.61381753764394,
      "grad_norm": 0.27440449595451355,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 100045
    },
    {
      "epoch": 88.61824623560673,
      "grad_norm": 0.27623388171195984,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 100050
    },
    {
      "epoch": 88.62267493356953,
      "grad_norm": 0.2780875265598297,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 100055
    },
    {
      "epoch": 88.62710363153234,
      "grad_norm": 0.2932929992675781,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 100060
    },
    {
      "epoch": 88.63153232949513,
      "grad_norm": 0.2343708574771881,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 100065
    },
    {
      "epoch": 88.63596102745792,
      "grad_norm": 0.20966751873493195,
      "learning_rate": 1e-05,
      "loss": 0.9222,
      "step": 100070
    },
    {
      "epoch": 88.64038972542073,
      "grad_norm": 0.24885550141334534,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 100075
    },
    {
      "epoch": 88.64481842338353,
      "grad_norm": 0.29896649718284607,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 100080
    },
    {
      "epoch": 88.64924712134632,
      "grad_norm": 0.238017275929451,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 100085
    },
    {
      "epoch": 88.65367581930913,
      "grad_norm": 0.2601117789745331,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 100090
    },
    {
      "epoch": 88.65810451727192,
      "grad_norm": 0.25714123249053955,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 100095
    },
    {
      "epoch": 88.66253321523472,
      "grad_norm": 0.24347330629825592,
      "learning_rate": 1e-05,
      "loss": 0.8983,
      "step": 100100
    },
    {
      "epoch": 88.66696191319753,
      "grad_norm": 0.24118779599666595,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 100105
    },
    {
      "epoch": 88.67139061116032,
      "grad_norm": 0.2053503841161728,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 100110
    },
    {
      "epoch": 88.67581930912311,
      "grad_norm": 0.20410113036632538,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 100115
    },
    {
      "epoch": 88.68024800708592,
      "grad_norm": 0.20267923176288605,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 100120
    },
    {
      "epoch": 88.68467670504872,
      "grad_norm": 0.23829081654548645,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 100125
    },
    {
      "epoch": 88.68910540301151,
      "grad_norm": 0.22634829580783844,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 100130
    },
    {
      "epoch": 88.69353410097432,
      "grad_norm": 0.25587794184684753,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 100135
    },
    {
      "epoch": 88.69796279893711,
      "grad_norm": 0.22228361666202545,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 100140
    },
    {
      "epoch": 88.7023914968999,
      "grad_norm": 0.2489021271467209,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 100145
    },
    {
      "epoch": 88.70682019486271,
      "grad_norm": 0.2050846666097641,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 100150
    },
    {
      "epoch": 88.71124889282551,
      "grad_norm": 0.28455400466918945,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 100155
    },
    {
      "epoch": 88.7156775907883,
      "grad_norm": 0.2465212345123291,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 100160
    },
    {
      "epoch": 88.72010628875111,
      "grad_norm": 0.22401882708072662,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 100165
    },
    {
      "epoch": 88.7245349867139,
      "grad_norm": 0.23985005915164948,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 100170
    },
    {
      "epoch": 88.7289636846767,
      "grad_norm": 0.21802686154842377,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 100175
    },
    {
      "epoch": 88.73339238263951,
      "grad_norm": 0.26026055216789246,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 100180
    },
    {
      "epoch": 88.7378210806023,
      "grad_norm": 0.22029007971286774,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 100185
    },
    {
      "epoch": 88.7422497785651,
      "grad_norm": 0.2248016893863678,
      "learning_rate": 1e-05,
      "loss": 0.909,
      "step": 100190
    },
    {
      "epoch": 88.7466784765279,
      "grad_norm": 0.2388961911201477,
      "learning_rate": 1e-05,
      "loss": 0.9287,
      "step": 100195
    },
    {
      "epoch": 88.7511071744907,
      "grad_norm": 0.2143191248178482,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 100200
    },
    {
      "epoch": 88.75553587245349,
      "grad_norm": 0.2264724224805832,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 100205
    },
    {
      "epoch": 88.7599645704163,
      "grad_norm": 0.22710593044757843,
      "learning_rate": 1e-05,
      "loss": 0.9229,
      "step": 100210
    },
    {
      "epoch": 88.7643932683791,
      "grad_norm": 0.2247508019208908,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 100215
    },
    {
      "epoch": 88.76882196634189,
      "grad_norm": 0.29041588306427,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 100220
    },
    {
      "epoch": 88.7732506643047,
      "grad_norm": 0.2267577201128006,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 100225
    },
    {
      "epoch": 88.77767936226749,
      "grad_norm": 0.19997918605804443,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 100230
    },
    {
      "epoch": 88.78210806023029,
      "grad_norm": 0.26032817363739014,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 100235
    },
    {
      "epoch": 88.7865367581931,
      "grad_norm": 0.269196093082428,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 100240
    },
    {
      "epoch": 88.79096545615589,
      "grad_norm": 0.24599489569664001,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 100245
    },
    {
      "epoch": 88.79539415411868,
      "grad_norm": 0.2428746521472931,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 100250
    },
    {
      "epoch": 88.79982285208149,
      "grad_norm": 0.2269238531589508,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 100255
    },
    {
      "epoch": 88.80425155004428,
      "grad_norm": 0.22356781363487244,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 100260
    },
    {
      "epoch": 88.8086802480071,
      "grad_norm": 0.21390698850154877,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 100265
    },
    {
      "epoch": 88.81310894596989,
      "grad_norm": 0.24619929492473602,
      "learning_rate": 1e-05,
      "loss": 0.9059,
      "step": 100270
    },
    {
      "epoch": 88.81753764393268,
      "grad_norm": 0.29885417222976685,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 100275
    },
    {
      "epoch": 88.82196634189549,
      "grad_norm": 0.2692887783050537,
      "learning_rate": 1e-05,
      "loss": 0.99,
      "step": 100280
    },
    {
      "epoch": 88.82639503985828,
      "grad_norm": 0.2062106728553772,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 100285
    },
    {
      "epoch": 88.83082373782108,
      "grad_norm": 0.22567851841449738,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 100290
    },
    {
      "epoch": 88.83525243578389,
      "grad_norm": 0.2369624674320221,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 100295
    },
    {
      "epoch": 88.83968113374668,
      "grad_norm": 0.23463581502437592,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 100300
    },
    {
      "epoch": 88.84410983170947,
      "grad_norm": 0.2131398469209671,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 100305
    },
    {
      "epoch": 88.84853852967228,
      "grad_norm": 0.22814461588859558,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 100310
    },
    {
      "epoch": 88.85296722763508,
      "grad_norm": 0.24123625457286835,
      "learning_rate": 1e-05,
      "loss": 0.9091,
      "step": 100315
    },
    {
      "epoch": 88.85739592559787,
      "grad_norm": 0.23458488285541534,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 100320
    },
    {
      "epoch": 88.86182462356068,
      "grad_norm": 0.2231205850839615,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 100325
    },
    {
      "epoch": 88.86625332152347,
      "grad_norm": 0.26181745529174805,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 100330
    },
    {
      "epoch": 88.87068201948627,
      "grad_norm": 0.2155875861644745,
      "learning_rate": 1e-05,
      "loss": 0.9196,
      "step": 100335
    },
    {
      "epoch": 88.87511071744908,
      "grad_norm": 0.21791820228099823,
      "learning_rate": 1e-05,
      "loss": 0.9088,
      "step": 100340
    },
    {
      "epoch": 88.87953941541187,
      "grad_norm": 0.22697681188583374,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 100345
    },
    {
      "epoch": 88.88396811337466,
      "grad_norm": 0.2161789983510971,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 100350
    },
    {
      "epoch": 88.88839681133747,
      "grad_norm": 0.26885396242141724,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 100355
    },
    {
      "epoch": 88.89282550930027,
      "grad_norm": 0.26077714562416077,
      "learning_rate": 1e-05,
      "loss": 0.9295,
      "step": 100360
    },
    {
      "epoch": 88.89725420726306,
      "grad_norm": 0.2938379943370819,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 100365
    },
    {
      "epoch": 88.90168290522587,
      "grad_norm": 0.2660319209098816,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 100370
    },
    {
      "epoch": 88.90611160318866,
      "grad_norm": 0.24503764510154724,
      "learning_rate": 1e-05,
      "loss": 0.9201,
      "step": 100375
    },
    {
      "epoch": 88.91054030115146,
      "grad_norm": 0.2380249798297882,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 100380
    },
    {
      "epoch": 88.91496899911427,
      "grad_norm": 0.25015366077423096,
      "learning_rate": 1e-05,
      "loss": 1.0392,
      "step": 100385
    },
    {
      "epoch": 88.91939769707706,
      "grad_norm": 0.2481878399848938,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 100390
    },
    {
      "epoch": 88.92382639503985,
      "grad_norm": 0.24644330143928528,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 100395
    },
    {
      "epoch": 88.92825509300266,
      "grad_norm": 0.2675034999847412,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 100400
    },
    {
      "epoch": 88.93268379096546,
      "grad_norm": 0.21234582364559174,
      "learning_rate": 1e-05,
      "loss": 0.9998,
      "step": 100405
    },
    {
      "epoch": 88.93711248892825,
      "grad_norm": 0.2354971170425415,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 100410
    },
    {
      "epoch": 88.94154118689106,
      "grad_norm": 0.21754314005374908,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 100415
    },
    {
      "epoch": 88.94596988485385,
      "grad_norm": 0.24992431700229645,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 100420
    },
    {
      "epoch": 88.95039858281665,
      "grad_norm": 0.2641519606113434,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 100425
    },
    {
      "epoch": 88.95482728077945,
      "grad_norm": 0.24620196223258972,
      "learning_rate": 1e-05,
      "loss": 0.981,
      "step": 100430
    },
    {
      "epoch": 88.95925597874225,
      "grad_norm": 0.25087955594062805,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 100435
    },
    {
      "epoch": 88.96368467670504,
      "grad_norm": 0.217938631772995,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 100440
    },
    {
      "epoch": 88.96811337466785,
      "grad_norm": 0.23816604912281036,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 100445
    },
    {
      "epoch": 88.97254207263065,
      "grad_norm": 0.2639926075935364,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 100450
    },
    {
      "epoch": 88.97697077059344,
      "grad_norm": 0.27573248744010925,
      "learning_rate": 1e-05,
      "loss": 0.9396,
      "step": 100455
    },
    {
      "epoch": 88.98139946855625,
      "grad_norm": 0.24717433750629425,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 100460
    },
    {
      "epoch": 88.98582816651904,
      "grad_norm": 0.2584714889526367,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 100465
    },
    {
      "epoch": 88.99025686448184,
      "grad_norm": 0.2804734408855438,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 100470
    },
    {
      "epoch": 88.99468556244464,
      "grad_norm": 0.2646898627281189,
      "learning_rate": 1e-05,
      "loss": 0.9181,
      "step": 100475
    },
    {
      "epoch": 88.99911426040744,
      "grad_norm": 0.26005032658576965,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 100480
    },
    {
      "epoch": 89.00354295837023,
      "grad_norm": 0.22763916850090027,
      "learning_rate": 1e-05,
      "loss": 0.9263,
      "step": 100485
    },
    {
      "epoch": 89.00797165633304,
      "grad_norm": 0.2225470393896103,
      "learning_rate": 1e-05,
      "loss": 0.9038,
      "step": 100490
    },
    {
      "epoch": 89.01240035429583,
      "grad_norm": 0.28239724040031433,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 100495
    },
    {
      "epoch": 89.01682905225863,
      "grad_norm": 0.22660139203071594,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 100500
    },
    {
      "epoch": 89.02125775022144,
      "grad_norm": 0.2755179703235626,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 100505
    },
    {
      "epoch": 89.02568644818423,
      "grad_norm": 0.2388431429862976,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 100510
    },
    {
      "epoch": 89.03011514614704,
      "grad_norm": 0.2778083384037018,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 100515
    },
    {
      "epoch": 89.03454384410983,
      "grad_norm": 0.22600561380386353,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 100520
    },
    {
      "epoch": 89.03897254207263,
      "grad_norm": 0.23254676163196564,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 100525
    },
    {
      "epoch": 89.04340124003544,
      "grad_norm": 0.26569318771362305,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 100530
    },
    {
      "epoch": 89.04782993799823,
      "grad_norm": 0.23208293318748474,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 100535
    },
    {
      "epoch": 89.05225863596102,
      "grad_norm": 0.22812440991401672,
      "learning_rate": 1e-05,
      "loss": 1.0451,
      "step": 100540
    },
    {
      "epoch": 89.05668733392383,
      "grad_norm": 0.2540033161640167,
      "learning_rate": 1e-05,
      "loss": 0.9297,
      "step": 100545
    },
    {
      "epoch": 89.06111603188663,
      "grad_norm": 0.25510844588279724,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 100550
    },
    {
      "epoch": 89.06554472984942,
      "grad_norm": 0.22745384275913239,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 100555
    },
    {
      "epoch": 89.06997342781223,
      "grad_norm": 0.21563567221164703,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 100560
    },
    {
      "epoch": 89.07440212577502,
      "grad_norm": 0.24625031650066376,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 100565
    },
    {
      "epoch": 89.07883082373782,
      "grad_norm": 0.22668561339378357,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 100570
    },
    {
      "epoch": 89.08325952170063,
      "grad_norm": 0.21900923550128937,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 100575
    },
    {
      "epoch": 89.08768821966342,
      "grad_norm": 0.24717196822166443,
      "learning_rate": 1e-05,
      "loss": 0.933,
      "step": 100580
    },
    {
      "epoch": 89.09211691762621,
      "grad_norm": 0.25774258375167847,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 100585
    },
    {
      "epoch": 89.09654561558902,
      "grad_norm": 0.30301162600517273,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 100590
    },
    {
      "epoch": 89.10097431355182,
      "grad_norm": 0.24284137785434723,
      "learning_rate": 1e-05,
      "loss": 0.9287,
      "step": 100595
    },
    {
      "epoch": 89.10540301151461,
      "grad_norm": 0.25048893690109253,
      "learning_rate": 1e-05,
      "loss": 0.8973,
      "step": 100600
    },
    {
      "epoch": 89.10983170947742,
      "grad_norm": 0.27334481477737427,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 100605
    },
    {
      "epoch": 89.11426040744021,
      "grad_norm": 0.2897074222564697,
      "learning_rate": 1e-05,
      "loss": 0.9165,
      "step": 100610
    },
    {
      "epoch": 89.118689105403,
      "grad_norm": 0.2764134705066681,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 100615
    },
    {
      "epoch": 89.12311780336582,
      "grad_norm": 0.2903238832950592,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 100620
    },
    {
      "epoch": 89.12754650132861,
      "grad_norm": 0.24302980303764343,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 100625
    },
    {
      "epoch": 89.1319751992914,
      "grad_norm": 0.3113268315792084,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 100630
    },
    {
      "epoch": 89.13640389725421,
      "grad_norm": 0.214813694357872,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 100635
    },
    {
      "epoch": 89.140832595217,
      "grad_norm": 0.23509356379508972,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 100640
    },
    {
      "epoch": 89.1452612931798,
      "grad_norm": 0.23588000237941742,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 100645
    },
    {
      "epoch": 89.14968999114261,
      "grad_norm": 0.22670619189739227,
      "learning_rate": 1e-05,
      "loss": 1.0286,
      "step": 100650
    },
    {
      "epoch": 89.1541186891054,
      "grad_norm": 0.2659050226211548,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 100655
    },
    {
      "epoch": 89.1585473870682,
      "grad_norm": 0.23737093806266785,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 100660
    },
    {
      "epoch": 89.162976085031,
      "grad_norm": 0.24317654967308044,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 100665
    },
    {
      "epoch": 89.1674047829938,
      "grad_norm": 0.24173079431056976,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 100670
    },
    {
      "epoch": 89.1718334809566,
      "grad_norm": 0.25410574674606323,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 100675
    },
    {
      "epoch": 89.1762621789194,
      "grad_norm": 0.2642264664173126,
      "learning_rate": 1e-05,
      "loss": 1.0312,
      "step": 100680
    },
    {
      "epoch": 89.1806908768822,
      "grad_norm": 0.2311367243528366,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 100685
    },
    {
      "epoch": 89.18511957484499,
      "grad_norm": 0.22274249792099,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 100690
    },
    {
      "epoch": 89.1895482728078,
      "grad_norm": 0.22816705703735352,
      "learning_rate": 1e-05,
      "loss": 0.9219,
      "step": 100695
    },
    {
      "epoch": 89.19397697077059,
      "grad_norm": 0.22125868499279022,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 100700
    },
    {
      "epoch": 89.19840566873339,
      "grad_norm": 0.24668489396572113,
      "learning_rate": 1e-05,
      "loss": 0.9073,
      "step": 100705
    },
    {
      "epoch": 89.2028343666962,
      "grad_norm": 0.20868946611881256,
      "learning_rate": 1e-05,
      "loss": 0.9137,
      "step": 100710
    },
    {
      "epoch": 89.20726306465899,
      "grad_norm": 0.2574494183063507,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 100715
    },
    {
      "epoch": 89.21169176262178,
      "grad_norm": 0.28075528144836426,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 100720
    },
    {
      "epoch": 89.21612046058459,
      "grad_norm": 0.22525160014629364,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 100725
    },
    {
      "epoch": 89.22054915854739,
      "grad_norm": 0.2189497947692871,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 100730
    },
    {
      "epoch": 89.22497785651018,
      "grad_norm": 0.2552158534526825,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 100735
    },
    {
      "epoch": 89.22940655447299,
      "grad_norm": 0.2355300486087799,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 100740
    },
    {
      "epoch": 89.23383525243578,
      "grad_norm": 0.23161034286022186,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 100745
    },
    {
      "epoch": 89.23826395039858,
      "grad_norm": 0.24285735189914703,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 100750
    },
    {
      "epoch": 89.24269264836138,
      "grad_norm": 0.24322988092899323,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 100755
    },
    {
      "epoch": 89.24712134632418,
      "grad_norm": 0.2284044325351715,
      "learning_rate": 1e-05,
      "loss": 0.8853,
      "step": 100760
    },
    {
      "epoch": 89.25155004428699,
      "grad_norm": 0.21393656730651855,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 100765
    },
    {
      "epoch": 89.25597874224978,
      "grad_norm": 0.2537887394428253,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 100770
    },
    {
      "epoch": 89.26040744021257,
      "grad_norm": 0.24150004982948303,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 100775
    },
    {
      "epoch": 89.26483613817538,
      "grad_norm": 0.22063104808330536,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 100780
    },
    {
      "epoch": 89.26926483613818,
      "grad_norm": 0.250154584646225,
      "learning_rate": 1e-05,
      "loss": 0.9092,
      "step": 100785
    },
    {
      "epoch": 89.27369353410097,
      "grad_norm": 0.2574433982372284,
      "learning_rate": 1e-05,
      "loss": 0.9318,
      "step": 100790
    },
    {
      "epoch": 89.27812223206378,
      "grad_norm": 0.21480466425418854,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 100795
    },
    {
      "epoch": 89.28255093002657,
      "grad_norm": 0.2599238157272339,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 100800
    },
    {
      "epoch": 89.28697962798937,
      "grad_norm": 0.23576489090919495,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 100805
    },
    {
      "epoch": 89.29140832595218,
      "grad_norm": 0.2433391511440277,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 100810
    },
    {
      "epoch": 89.29583702391497,
      "grad_norm": 0.19983533024787903,
      "learning_rate": 1e-05,
      "loss": 0.898,
      "step": 100815
    },
    {
      "epoch": 89.30026572187776,
      "grad_norm": 0.2664582133293152,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 100820
    },
    {
      "epoch": 89.30469441984057,
      "grad_norm": 0.20401927828788757,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 100825
    },
    {
      "epoch": 89.30912311780337,
      "grad_norm": 0.22246390581130981,
      "learning_rate": 1e-05,
      "loss": 0.9065,
      "step": 100830
    },
    {
      "epoch": 89.31355181576616,
      "grad_norm": 0.2207382321357727,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 100835
    },
    {
      "epoch": 89.31798051372897,
      "grad_norm": 0.20225244760513306,
      "learning_rate": 1e-05,
      "loss": 0.9245,
      "step": 100840
    },
    {
      "epoch": 89.32240921169176,
      "grad_norm": 0.23379398882389069,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 100845
    },
    {
      "epoch": 89.32683790965456,
      "grad_norm": 0.24309130012989044,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 100850
    },
    {
      "epoch": 89.33126660761737,
      "grad_norm": 0.22226004302501678,
      "learning_rate": 1e-05,
      "loss": 0.9364,
      "step": 100855
    },
    {
      "epoch": 89.33569530558016,
      "grad_norm": 0.2278764247894287,
      "learning_rate": 1e-05,
      "loss": 0.9046,
      "step": 100860
    },
    {
      "epoch": 89.34012400354295,
      "grad_norm": 0.2293863594532013,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 100865
    },
    {
      "epoch": 89.34455270150576,
      "grad_norm": 0.22405551373958588,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 100870
    },
    {
      "epoch": 89.34898139946856,
      "grad_norm": 0.25798124074935913,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 100875
    },
    {
      "epoch": 89.35341009743135,
      "grad_norm": 0.21349021792411804,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 100880
    },
    {
      "epoch": 89.35783879539416,
      "grad_norm": 0.26878902316093445,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 100885
    },
    {
      "epoch": 89.36226749335695,
      "grad_norm": 0.2652953267097473,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 100890
    },
    {
      "epoch": 89.36669619131975,
      "grad_norm": 0.30907854437828064,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 100895
    },
    {
      "epoch": 89.37112488928256,
      "grad_norm": 0.2491472363471985,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 100900
    },
    {
      "epoch": 89.37555358724535,
      "grad_norm": 0.22202803194522858,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 100905
    },
    {
      "epoch": 89.37998228520814,
      "grad_norm": 0.2564280033111572,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 100910
    },
    {
      "epoch": 89.38441098317095,
      "grad_norm": 0.22271765768527985,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 100915
    },
    {
      "epoch": 89.38883968113375,
      "grad_norm": 0.2571192979812622,
      "learning_rate": 1e-05,
      "loss": 0.9194,
      "step": 100920
    },
    {
      "epoch": 89.39326837909654,
      "grad_norm": 0.2326420545578003,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 100925
    },
    {
      "epoch": 89.39769707705935,
      "grad_norm": 0.21721535921096802,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 100930
    },
    {
      "epoch": 89.40212577502214,
      "grad_norm": 0.19836705923080444,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 100935
    },
    {
      "epoch": 89.40655447298494,
      "grad_norm": 0.24120672047138214,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 100940
    },
    {
      "epoch": 89.41098317094774,
      "grad_norm": 0.21410706639289856,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 100945
    },
    {
      "epoch": 89.41541186891054,
      "grad_norm": 0.2139817774295807,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 100950
    },
    {
      "epoch": 89.41984056687333,
      "grad_norm": 0.2646258771419525,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 100955
    },
    {
      "epoch": 89.42426926483614,
      "grad_norm": 0.2344912439584732,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 100960
    },
    {
      "epoch": 89.42869796279894,
      "grad_norm": 0.2454075962305069,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 100965
    },
    {
      "epoch": 89.43312666076173,
      "grad_norm": 0.24018558859825134,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 100970
    },
    {
      "epoch": 89.43755535872454,
      "grad_norm": 0.2437729835510254,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 100975
    },
    {
      "epoch": 89.44198405668733,
      "grad_norm": 0.24139168858528137,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 100980
    },
    {
      "epoch": 89.44641275465013,
      "grad_norm": 0.3099621832370758,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 100985
    },
    {
      "epoch": 89.45084145261293,
      "grad_norm": 0.2625819742679596,
      "learning_rate": 1e-05,
      "loss": 0.9261,
      "step": 100990
    },
    {
      "epoch": 89.45527015057573,
      "grad_norm": 0.24205012619495392,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 100995
    },
    {
      "epoch": 89.45969884853854,
      "grad_norm": 0.29166045784950256,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 101000
    },
    {
      "epoch": 89.46412754650133,
      "grad_norm": 0.23231516778469086,
      "learning_rate": 1e-05,
      "loss": 0.9088,
      "step": 101005
    },
    {
      "epoch": 89.46855624446412,
      "grad_norm": 0.22558072209358215,
      "learning_rate": 1e-05,
      "loss": 0.9286,
      "step": 101010
    },
    {
      "epoch": 89.47298494242693,
      "grad_norm": 0.30114856362342834,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 101015
    },
    {
      "epoch": 89.47741364038973,
      "grad_norm": 0.24628520011901855,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 101020
    },
    {
      "epoch": 89.48184233835252,
      "grad_norm": 0.21452312171459198,
      "learning_rate": 1e-05,
      "loss": 0.918,
      "step": 101025
    },
    {
      "epoch": 89.48627103631533,
      "grad_norm": 0.25621384382247925,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 101030
    },
    {
      "epoch": 89.49069973427812,
      "grad_norm": 0.20334744453430176,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 101035
    },
    {
      "epoch": 89.49512843224092,
      "grad_norm": 0.219874769449234,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 101040
    },
    {
      "epoch": 89.49955713020373,
      "grad_norm": 0.21250200271606445,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 101045
    },
    {
      "epoch": 89.50398582816652,
      "grad_norm": 0.2596737742424011,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 101050
    },
    {
      "epoch": 89.50841452612931,
      "grad_norm": 0.27819231152534485,
      "learning_rate": 1e-05,
      "loss": 0.9187,
      "step": 101055
    },
    {
      "epoch": 89.51284322409212,
      "grad_norm": 0.2287324219942093,
      "learning_rate": 1e-05,
      "loss": 0.8767,
      "step": 101060
    },
    {
      "epoch": 89.51727192205492,
      "grad_norm": 0.24213194847106934,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 101065
    },
    {
      "epoch": 89.52170062001771,
      "grad_norm": 0.2717299163341522,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 101070
    },
    {
      "epoch": 89.52612931798052,
      "grad_norm": 0.23695769906044006,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 101075
    },
    {
      "epoch": 89.53055801594331,
      "grad_norm": 0.24120908975601196,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 101080
    },
    {
      "epoch": 89.53498671390611,
      "grad_norm": 0.25059378147125244,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 101085
    },
    {
      "epoch": 89.53941541186892,
      "grad_norm": 0.25819337368011475,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 101090
    },
    {
      "epoch": 89.54384410983171,
      "grad_norm": 0.2337779700756073,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 101095
    },
    {
      "epoch": 89.5482728077945,
      "grad_norm": 0.22248628735542297,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 101100
    },
    {
      "epoch": 89.55270150575731,
      "grad_norm": 0.29555508494377136,
      "learning_rate": 1e-05,
      "loss": 0.9367,
      "step": 101105
    },
    {
      "epoch": 89.5571302037201,
      "grad_norm": 0.26678186655044556,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 101110
    },
    {
      "epoch": 89.5615589016829,
      "grad_norm": 0.20290780067443848,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 101115
    },
    {
      "epoch": 89.56598759964571,
      "grad_norm": 0.2417866438627243,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 101120
    },
    {
      "epoch": 89.5704162976085,
      "grad_norm": 0.25751861929893494,
      "learning_rate": 1e-05,
      "loss": 0.9297,
      "step": 101125
    },
    {
      "epoch": 89.5748449955713,
      "grad_norm": 0.22269615530967712,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 101130
    },
    {
      "epoch": 89.5792736935341,
      "grad_norm": 0.23602698743343353,
      "learning_rate": 1e-05,
      "loss": 0.9104,
      "step": 101135
    },
    {
      "epoch": 89.5837023914969,
      "grad_norm": 0.27046141028404236,
      "learning_rate": 1e-05,
      "loss": 0.9257,
      "step": 101140
    },
    {
      "epoch": 89.5881310894597,
      "grad_norm": 0.2539267838001251,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 101145
    },
    {
      "epoch": 89.5925597874225,
      "grad_norm": 0.28185129165649414,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 101150
    },
    {
      "epoch": 89.5969884853853,
      "grad_norm": 0.2545854449272156,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 101155
    },
    {
      "epoch": 89.60141718334809,
      "grad_norm": 0.23088335990905762,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 101160
    },
    {
      "epoch": 89.6058458813109,
      "grad_norm": 0.23168882727622986,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 101165
    },
    {
      "epoch": 89.61027457927369,
      "grad_norm": 0.23449388146400452,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 101170
    },
    {
      "epoch": 89.61470327723649,
      "grad_norm": 0.26414287090301514,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 101175
    },
    {
      "epoch": 89.6191319751993,
      "grad_norm": 0.31195399165153503,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 101180
    },
    {
      "epoch": 89.62356067316209,
      "grad_norm": 0.22076132893562317,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 101185
    },
    {
      "epoch": 89.62798937112488,
      "grad_norm": 0.23083002865314484,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 101190
    },
    {
      "epoch": 89.63241806908769,
      "grad_norm": 0.20712360739707947,
      "learning_rate": 1e-05,
      "loss": 0.9201,
      "step": 101195
    },
    {
      "epoch": 89.63684676705049,
      "grad_norm": 0.21525146067142487,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 101200
    },
    {
      "epoch": 89.64127546501328,
      "grad_norm": 0.2378852814435959,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 101205
    },
    {
      "epoch": 89.64570416297609,
      "grad_norm": 0.24321012198925018,
      "learning_rate": 1e-05,
      "loss": 0.9338,
      "step": 101210
    },
    {
      "epoch": 89.65013286093888,
      "grad_norm": 0.2689436972141266,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 101215
    },
    {
      "epoch": 89.65456155890168,
      "grad_norm": 0.21890833973884583,
      "learning_rate": 1e-05,
      "loss": 0.9169,
      "step": 101220
    },
    {
      "epoch": 89.65899025686448,
      "grad_norm": 0.23341307044029236,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 101225
    },
    {
      "epoch": 89.66341895482728,
      "grad_norm": 0.2196146845817566,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 101230
    },
    {
      "epoch": 89.66784765279007,
      "grad_norm": 0.30614590644836426,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 101235
    },
    {
      "epoch": 89.67227635075288,
      "grad_norm": 0.31138527393341064,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 101240
    },
    {
      "epoch": 89.67670504871568,
      "grad_norm": 0.28225719928741455,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 101245
    },
    {
      "epoch": 89.68113374667848,
      "grad_norm": 0.2552628815174103,
      "learning_rate": 1e-05,
      "loss": 0.9577,
      "step": 101250
    },
    {
      "epoch": 89.68556244464128,
      "grad_norm": 0.21399152278900146,
      "learning_rate": 1e-05,
      "loss": 0.9959,
      "step": 101255
    },
    {
      "epoch": 89.68999114260407,
      "grad_norm": 0.2505783438682556,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 101260
    },
    {
      "epoch": 89.69441984056688,
      "grad_norm": 0.2451978623867035,
      "learning_rate": 1e-05,
      "loss": 0.8823,
      "step": 101265
    },
    {
      "epoch": 89.69884853852967,
      "grad_norm": 0.2342648208141327,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 101270
    },
    {
      "epoch": 89.70327723649247,
      "grad_norm": 0.2624475359916687,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 101275
    },
    {
      "epoch": 89.70770593445528,
      "grad_norm": 0.23838311433792114,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 101280
    },
    {
      "epoch": 89.71213463241807,
      "grad_norm": 0.3192882835865021,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 101285
    },
    {
      "epoch": 89.71656333038086,
      "grad_norm": 0.26912227272987366,
      "learning_rate": 1e-05,
      "loss": 1.0317,
      "step": 101290
    },
    {
      "epoch": 89.72099202834367,
      "grad_norm": 0.21121801435947418,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 101295
    },
    {
      "epoch": 89.72542072630647,
      "grad_norm": 0.2362804114818573,
      "learning_rate": 1e-05,
      "loss": 0.9161,
      "step": 101300
    },
    {
      "epoch": 89.72984942426926,
      "grad_norm": 0.22550930082798004,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 101305
    },
    {
      "epoch": 89.73427812223207,
      "grad_norm": 0.3003152906894684,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 101310
    },
    {
      "epoch": 89.73870682019486,
      "grad_norm": 0.21749617159366608,
      "learning_rate": 1e-05,
      "loss": 0.934,
      "step": 101315
    },
    {
      "epoch": 89.74313551815766,
      "grad_norm": 0.2600935101509094,
      "learning_rate": 1e-05,
      "loss": 0.9903,
      "step": 101320
    },
    {
      "epoch": 89.74756421612047,
      "grad_norm": 0.27593857049942017,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 101325
    },
    {
      "epoch": 89.75199291408326,
      "grad_norm": 0.25915616750717163,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 101330
    },
    {
      "epoch": 89.75642161204605,
      "grad_norm": 0.2346881777048111,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 101335
    },
    {
      "epoch": 89.76085031000886,
      "grad_norm": 0.20665375888347626,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 101340
    },
    {
      "epoch": 89.76527900797166,
      "grad_norm": 0.23834674060344696,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 101345
    },
    {
      "epoch": 89.76970770593445,
      "grad_norm": 0.2480219155550003,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 101350
    },
    {
      "epoch": 89.77413640389726,
      "grad_norm": 0.22663390636444092,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 101355
    },
    {
      "epoch": 89.77856510186005,
      "grad_norm": 0.2249283343553543,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 101360
    },
    {
      "epoch": 89.78299379982285,
      "grad_norm": 0.27349337935447693,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 101365
    },
    {
      "epoch": 89.78742249778566,
      "grad_norm": 0.2681344449520111,
      "learning_rate": 1e-05,
      "loss": 0.914,
      "step": 101370
    },
    {
      "epoch": 89.79185119574845,
      "grad_norm": 0.2435312569141388,
      "learning_rate": 1e-05,
      "loss": 0.9333,
      "step": 101375
    },
    {
      "epoch": 89.79627989371124,
      "grad_norm": 0.22463159263134003,
      "learning_rate": 1e-05,
      "loss": 0.9237,
      "step": 101380
    },
    {
      "epoch": 89.80070859167405,
      "grad_norm": 0.22619637846946716,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 101385
    },
    {
      "epoch": 89.80513728963685,
      "grad_norm": 0.2531736493110657,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 101390
    },
    {
      "epoch": 89.80956598759964,
      "grad_norm": 0.26272615790367126,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 101395
    },
    {
      "epoch": 89.81399468556245,
      "grad_norm": 0.21618276834487915,
      "learning_rate": 1e-05,
      "loss": 0.9045,
      "step": 101400
    },
    {
      "epoch": 89.81842338352524,
      "grad_norm": 0.2515418231487274,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 101405
    },
    {
      "epoch": 89.82285208148804,
      "grad_norm": 0.25076019763946533,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 101410
    },
    {
      "epoch": 89.82728077945085,
      "grad_norm": 0.2517538368701935,
      "learning_rate": 1e-05,
      "loss": 0.9265,
      "step": 101415
    },
    {
      "epoch": 89.83170947741364,
      "grad_norm": 0.20044861733913422,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 101420
    },
    {
      "epoch": 89.83613817537643,
      "grad_norm": 0.23109887540340424,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 101425
    },
    {
      "epoch": 89.84056687333924,
      "grad_norm": 0.24378788471221924,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 101430
    },
    {
      "epoch": 89.84499557130204,
      "grad_norm": 0.20190908014774323,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 101435
    },
    {
      "epoch": 89.84942426926483,
      "grad_norm": 0.24551331996917725,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 101440
    },
    {
      "epoch": 89.85385296722764,
      "grad_norm": 0.2267090380191803,
      "learning_rate": 1e-05,
      "loss": 0.8884,
      "step": 101445
    },
    {
      "epoch": 89.85828166519043,
      "grad_norm": 0.2133764922618866,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 101450
    },
    {
      "epoch": 89.86271036315323,
      "grad_norm": 0.21808011829853058,
      "learning_rate": 1e-05,
      "loss": 1.0096,
      "step": 101455
    },
    {
      "epoch": 89.86713906111603,
      "grad_norm": 0.23564226925373077,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 101460
    },
    {
      "epoch": 89.87156775907883,
      "grad_norm": 0.2676661014556885,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 101465
    },
    {
      "epoch": 89.87599645704162,
      "grad_norm": 0.2114715725183487,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 101470
    },
    {
      "epoch": 89.88042515500443,
      "grad_norm": 0.22617237269878387,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 101475
    },
    {
      "epoch": 89.88485385296723,
      "grad_norm": 0.22413133084774017,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 101480
    },
    {
      "epoch": 89.88928255093003,
      "grad_norm": 0.28835514187812805,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 101485
    },
    {
      "epoch": 89.89371124889283,
      "grad_norm": 0.26613354682922363,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 101490
    },
    {
      "epoch": 89.89813994685562,
      "grad_norm": 0.2634512782096863,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 101495
    },
    {
      "epoch": 89.90256864481843,
      "grad_norm": 0.24423576891422272,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 101500
    },
    {
      "epoch": 89.90699734278122,
      "grad_norm": 0.2354712188243866,
      "learning_rate": 1e-05,
      "loss": 0.9243,
      "step": 101505
    },
    {
      "epoch": 89.91142604074402,
      "grad_norm": 0.2415490448474884,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 101510
    },
    {
      "epoch": 89.91585473870683,
      "grad_norm": 0.21499145030975342,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 101515
    },
    {
      "epoch": 89.92028343666962,
      "grad_norm": 0.21010562777519226,
      "learning_rate": 1e-05,
      "loss": 0.9115,
      "step": 101520
    },
    {
      "epoch": 89.92471213463241,
      "grad_norm": 0.21463963389396667,
      "learning_rate": 1e-05,
      "loss": 0.9165,
      "step": 101525
    },
    {
      "epoch": 89.92914083259522,
      "grad_norm": 0.2589956223964691,
      "learning_rate": 1e-05,
      "loss": 0.8772,
      "step": 101530
    },
    {
      "epoch": 89.93356953055802,
      "grad_norm": 0.23022598028182983,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 101535
    },
    {
      "epoch": 89.93799822852081,
      "grad_norm": 0.24445129930973053,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 101540
    },
    {
      "epoch": 89.94242692648362,
      "grad_norm": 0.2590886950492859,
      "learning_rate": 1e-05,
      "loss": 0.9039,
      "step": 101545
    },
    {
      "epoch": 89.94685562444641,
      "grad_norm": 0.2533506453037262,
      "learning_rate": 1e-05,
      "loss": 0.9011,
      "step": 101550
    },
    {
      "epoch": 89.95128432240921,
      "grad_norm": 0.24357374012470245,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 101555
    },
    {
      "epoch": 89.95571302037202,
      "grad_norm": 0.23195527493953705,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 101560
    },
    {
      "epoch": 89.96014171833481,
      "grad_norm": 0.2404184639453888,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 101565
    },
    {
      "epoch": 89.9645704162976,
      "grad_norm": 0.21428319811820984,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 101570
    },
    {
      "epoch": 89.96899911426041,
      "grad_norm": 0.2308656871318817,
      "learning_rate": 1e-05,
      "loss": 0.8916,
      "step": 101575
    },
    {
      "epoch": 89.9734278122232,
      "grad_norm": 0.2555184066295624,
      "learning_rate": 1e-05,
      "loss": 1.0202,
      "step": 101580
    },
    {
      "epoch": 89.977856510186,
      "grad_norm": 0.2408762276172638,
      "learning_rate": 1e-05,
      "loss": 0.9182,
      "step": 101585
    },
    {
      "epoch": 89.98228520814881,
      "grad_norm": 0.23161953687667847,
      "learning_rate": 1e-05,
      "loss": 0.9095,
      "step": 101590
    },
    {
      "epoch": 89.9867139061116,
      "grad_norm": 0.2480221390724182,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 101595
    },
    {
      "epoch": 89.9911426040744,
      "grad_norm": 0.23990701138973236,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 101600
    },
    {
      "epoch": 89.9955713020372,
      "grad_norm": 0.27477511763572693,
      "learning_rate": 1e-05,
      "loss": 0.933,
      "step": 101605
    },
    {
      "epoch": 90.0,
      "grad_norm": 0.2638186514377594,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 101610
    },
    {
      "epoch": 90.0044286979628,
      "grad_norm": 0.21731531620025635,
      "learning_rate": 1e-05,
      "loss": 0.9111,
      "step": 101615
    },
    {
      "epoch": 90.0088573959256,
      "grad_norm": 0.22683370113372803,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 101620
    },
    {
      "epoch": 90.0132860938884,
      "grad_norm": 0.21489375829696655,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 101625
    },
    {
      "epoch": 90.01771479185119,
      "grad_norm": 0.21699175238609314,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 101630
    },
    {
      "epoch": 90.022143489814,
      "grad_norm": 0.2774849832057953,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 101635
    },
    {
      "epoch": 90.0265721877768,
      "grad_norm": 0.23622387647628784,
      "learning_rate": 1e-05,
      "loss": 1.0085,
      "step": 101640
    },
    {
      "epoch": 90.03100088573959,
      "grad_norm": 0.2810623049736023,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 101645
    },
    {
      "epoch": 90.0354295837024,
      "grad_norm": 0.27120575308799744,
      "learning_rate": 1e-05,
      "loss": 0.9153,
      "step": 101650
    },
    {
      "epoch": 90.03985828166519,
      "grad_norm": 0.22164461016654968,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 101655
    },
    {
      "epoch": 90.04428697962798,
      "grad_norm": 0.2398461103439331,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 101660
    },
    {
      "epoch": 90.04871567759079,
      "grad_norm": 0.2734074294567108,
      "learning_rate": 1e-05,
      "loss": 0.8878,
      "step": 101665
    },
    {
      "epoch": 90.05314437555359,
      "grad_norm": 0.2700138986110687,
      "learning_rate": 1e-05,
      "loss": 0.9974,
      "step": 101670
    },
    {
      "epoch": 90.05757307351638,
      "grad_norm": 0.29804757237434387,
      "learning_rate": 1e-05,
      "loss": 0.9271,
      "step": 101675
    },
    {
      "epoch": 90.06200177147919,
      "grad_norm": 0.2095632404088974,
      "learning_rate": 1e-05,
      "loss": 0.8943,
      "step": 101680
    },
    {
      "epoch": 90.06643046944198,
      "grad_norm": 0.22926340997219086,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 101685
    },
    {
      "epoch": 90.07085916740478,
      "grad_norm": 0.23530203104019165,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 101690
    },
    {
      "epoch": 90.07528786536759,
      "grad_norm": 0.21834242343902588,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 101695
    },
    {
      "epoch": 90.07971656333038,
      "grad_norm": 0.25071537494659424,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 101700
    },
    {
      "epoch": 90.08414526129317,
      "grad_norm": 0.32564565539360046,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 101705
    },
    {
      "epoch": 90.08857395925598,
      "grad_norm": 0.2299775630235672,
      "learning_rate": 1e-05,
      "loss": 1.0234,
      "step": 101710
    },
    {
      "epoch": 90.09300265721878,
      "grad_norm": 0.23191912472248077,
      "learning_rate": 1e-05,
      "loss": 0.9342,
      "step": 101715
    },
    {
      "epoch": 90.09743135518157,
      "grad_norm": 0.22682884335517883,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 101720
    },
    {
      "epoch": 90.10186005314438,
      "grad_norm": 0.2016848772764206,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 101725
    },
    {
      "epoch": 90.10628875110717,
      "grad_norm": 0.22069627046585083,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 101730
    },
    {
      "epoch": 90.11071744906998,
      "grad_norm": 0.23564662039279938,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 101735
    },
    {
      "epoch": 90.11514614703277,
      "grad_norm": 0.2687149941921234,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 101740
    },
    {
      "epoch": 90.11957484499557,
      "grad_norm": 0.27545130252838135,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 101745
    },
    {
      "epoch": 90.12400354295838,
      "grad_norm": 0.2593679428100586,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 101750
    },
    {
      "epoch": 90.12843224092117,
      "grad_norm": 0.2439153641462326,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 101755
    },
    {
      "epoch": 90.13286093888397,
      "grad_norm": 0.22765451669692993,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 101760
    },
    {
      "epoch": 90.13728963684677,
      "grad_norm": 0.2597343921661377,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 101765
    },
    {
      "epoch": 90.14171833480957,
      "grad_norm": 0.24106836318969727,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 101770
    },
    {
      "epoch": 90.14614703277236,
      "grad_norm": 0.2886425256729126,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 101775
    },
    {
      "epoch": 90.15057573073517,
      "grad_norm": 0.2757726013660431,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 101780
    },
    {
      "epoch": 90.15500442869796,
      "grad_norm": 0.2336672991514206,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 101785
    },
    {
      "epoch": 90.15943312666076,
      "grad_norm": 0.24112758040428162,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 101790
    },
    {
      "epoch": 90.16386182462357,
      "grad_norm": 0.22675307095050812,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 101795
    },
    {
      "epoch": 90.16829052258636,
      "grad_norm": 0.29166585206985474,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 101800
    },
    {
      "epoch": 90.17271922054915,
      "grad_norm": 0.2473752200603485,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 101805
    },
    {
      "epoch": 90.17714791851196,
      "grad_norm": 0.25430354475975037,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 101810
    },
    {
      "epoch": 90.18157661647476,
      "grad_norm": 0.2628345489501953,
      "learning_rate": 1e-05,
      "loss": 0.9522,
      "step": 101815
    },
    {
      "epoch": 90.18600531443755,
      "grad_norm": 0.2560077905654907,
      "learning_rate": 1e-05,
      "loss": 0.9028,
      "step": 101820
    },
    {
      "epoch": 90.19043401240036,
      "grad_norm": 0.2639038562774658,
      "learning_rate": 1e-05,
      "loss": 0.9118,
      "step": 101825
    },
    {
      "epoch": 90.19486271036315,
      "grad_norm": 0.22030985355377197,
      "learning_rate": 1e-05,
      "loss": 0.9123,
      "step": 101830
    },
    {
      "epoch": 90.19929140832595,
      "grad_norm": 0.24887403845787048,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 101835
    },
    {
      "epoch": 90.20372010628876,
      "grad_norm": 0.22400911152362823,
      "learning_rate": 1e-05,
      "loss": 0.8926,
      "step": 101840
    },
    {
      "epoch": 90.20814880425155,
      "grad_norm": 0.2544662356376648,
      "learning_rate": 1e-05,
      "loss": 0.9169,
      "step": 101845
    },
    {
      "epoch": 90.21257750221434,
      "grad_norm": 0.25758492946624756,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 101850
    },
    {
      "epoch": 90.21700620017715,
      "grad_norm": 0.22494804859161377,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 101855
    },
    {
      "epoch": 90.22143489813995,
      "grad_norm": 0.3007393479347229,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 101860
    },
    {
      "epoch": 90.22586359610274,
      "grad_norm": 0.2298215627670288,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 101865
    },
    {
      "epoch": 90.23029229406555,
      "grad_norm": 0.22728070616722107,
      "learning_rate": 1e-05,
      "loss": 0.9144,
      "step": 101870
    },
    {
      "epoch": 90.23472099202834,
      "grad_norm": 0.21045230329036713,
      "learning_rate": 1e-05,
      "loss": 0.9209,
      "step": 101875
    },
    {
      "epoch": 90.23914968999114,
      "grad_norm": 0.22615285217761993,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 101880
    },
    {
      "epoch": 90.24357838795395,
      "grad_norm": 0.2574142515659332,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 101885
    },
    {
      "epoch": 90.24800708591674,
      "grad_norm": 0.24605898559093475,
      "learning_rate": 1e-05,
      "loss": 0.9924,
      "step": 101890
    },
    {
      "epoch": 90.25243578387953,
      "grad_norm": 0.23315763473510742,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 101895
    },
    {
      "epoch": 90.25686448184234,
      "grad_norm": 0.24179145693778992,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 101900
    },
    {
      "epoch": 90.26129317980514,
      "grad_norm": 0.2404242306947708,
      "learning_rate": 1e-05,
      "loss": 0.9189,
      "step": 101905
    },
    {
      "epoch": 90.26572187776793,
      "grad_norm": 0.26178833842277527,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 101910
    },
    {
      "epoch": 90.27015057573074,
      "grad_norm": 0.24088487029075623,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 101915
    },
    {
      "epoch": 90.27457927369353,
      "grad_norm": 0.24002672731876373,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 101920
    },
    {
      "epoch": 90.27900797165633,
      "grad_norm": 0.2565065920352936,
      "learning_rate": 1e-05,
      "loss": 0.9227,
      "step": 101925
    },
    {
      "epoch": 90.28343666961914,
      "grad_norm": 0.24138912558555603,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 101930
    },
    {
      "epoch": 90.28786536758193,
      "grad_norm": 0.2719273567199707,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 101935
    },
    {
      "epoch": 90.29229406554472,
      "grad_norm": 0.20821243524551392,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 101940
    },
    {
      "epoch": 90.29672276350753,
      "grad_norm": 0.22810816764831543,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 101945
    },
    {
      "epoch": 90.30115146147033,
      "grad_norm": 0.23754476010799408,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 101950
    },
    {
      "epoch": 90.30558015943312,
      "grad_norm": 0.21767400205135345,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 101955
    },
    {
      "epoch": 90.31000885739593,
      "grad_norm": 0.2229224145412445,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 101960
    },
    {
      "epoch": 90.31443755535872,
      "grad_norm": 0.24212396144866943,
      "learning_rate": 1e-05,
      "loss": 0.9265,
      "step": 101965
    },
    {
      "epoch": 90.31886625332152,
      "grad_norm": 0.26202520728111267,
      "learning_rate": 1e-05,
      "loss": 0.9184,
      "step": 101970
    },
    {
      "epoch": 90.32329495128432,
      "grad_norm": 0.2544514536857605,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 101975
    },
    {
      "epoch": 90.32772364924712,
      "grad_norm": 0.24189475178718567,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 101980
    },
    {
      "epoch": 90.33215234720993,
      "grad_norm": 0.22791410982608795,
      "learning_rate": 1e-05,
      "loss": 0.9166,
      "step": 101985
    },
    {
      "epoch": 90.33658104517272,
      "grad_norm": 0.25980496406555176,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 101990
    },
    {
      "epoch": 90.34100974313552,
      "grad_norm": 0.2330830991268158,
      "learning_rate": 1e-05,
      "loss": 1.0184,
      "step": 101995
    },
    {
      "epoch": 90.34543844109832,
      "grad_norm": 0.25266340374946594,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 102000
    },
    {
      "epoch": 90.34986713906112,
      "grad_norm": 0.22988548874855042,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 102005
    },
    {
      "epoch": 90.35429583702391,
      "grad_norm": 0.2298308163881302,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 102010
    },
    {
      "epoch": 90.35872453498672,
      "grad_norm": 0.25408226251602173,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 102015
    },
    {
      "epoch": 90.36315323294951,
      "grad_norm": 0.24033193290233612,
      "learning_rate": 1e-05,
      "loss": 1.0222,
      "step": 102020
    },
    {
      "epoch": 90.36758193091231,
      "grad_norm": 0.22981905937194824,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 102025
    },
    {
      "epoch": 90.37201062887512,
      "grad_norm": 0.22549563646316528,
      "learning_rate": 1e-05,
      "loss": 1.0316,
      "step": 102030
    },
    {
      "epoch": 90.37643932683791,
      "grad_norm": 0.2371857762336731,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 102035
    },
    {
      "epoch": 90.3808680248007,
      "grad_norm": 0.2649405002593994,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 102040
    },
    {
      "epoch": 90.38529672276351,
      "grad_norm": 0.25831612944602966,
      "learning_rate": 1e-05,
      "loss": 0.9348,
      "step": 102045
    },
    {
      "epoch": 90.38972542072631,
      "grad_norm": 0.2512674331665039,
      "learning_rate": 1e-05,
      "loss": 0.9246,
      "step": 102050
    },
    {
      "epoch": 90.3941541186891,
      "grad_norm": 0.24431277811527252,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 102055
    },
    {
      "epoch": 90.39858281665191,
      "grad_norm": 0.20237493515014648,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 102060
    },
    {
      "epoch": 90.4030115146147,
      "grad_norm": 0.24072594940662384,
      "learning_rate": 1e-05,
      "loss": 0.9385,
      "step": 102065
    },
    {
      "epoch": 90.4074402125775,
      "grad_norm": 0.22452762722969055,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 102070
    },
    {
      "epoch": 90.4118689105403,
      "grad_norm": 0.2582663893699646,
      "learning_rate": 1e-05,
      "loss": 0.9247,
      "step": 102075
    },
    {
      "epoch": 90.4162976085031,
      "grad_norm": 0.2441079318523407,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 102080
    },
    {
      "epoch": 90.4207263064659,
      "grad_norm": 0.22792191803455353,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 102085
    },
    {
      "epoch": 90.4251550044287,
      "grad_norm": 0.29045307636260986,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 102090
    },
    {
      "epoch": 90.4295837023915,
      "grad_norm": 0.23156116902828217,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 102095
    },
    {
      "epoch": 90.43401240035429,
      "grad_norm": 0.21465878188610077,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 102100
    },
    {
      "epoch": 90.4384410983171,
      "grad_norm": 0.22731177508831024,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 102105
    },
    {
      "epoch": 90.4428697962799,
      "grad_norm": 0.2856611907482147,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 102110
    },
    {
      "epoch": 90.44729849424269,
      "grad_norm": 0.2671787142753601,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 102115
    },
    {
      "epoch": 90.4517271922055,
      "grad_norm": 0.2776314616203308,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 102120
    },
    {
      "epoch": 90.45615589016829,
      "grad_norm": 0.22591732442378998,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 102125
    },
    {
      "epoch": 90.46058458813108,
      "grad_norm": 0.22665360569953918,
      "learning_rate": 1e-05,
      "loss": 0.9751,
      "step": 102130
    },
    {
      "epoch": 90.46501328609389,
      "grad_norm": 0.21546079218387604,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 102135
    },
    {
      "epoch": 90.46944198405669,
      "grad_norm": 0.2616395056247711,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 102140
    },
    {
      "epoch": 90.47387068201948,
      "grad_norm": 0.27524682879447937,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 102145
    },
    {
      "epoch": 90.47829937998229,
      "grad_norm": 0.25979822874069214,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 102150
    },
    {
      "epoch": 90.48272807794508,
      "grad_norm": 0.23312847316265106,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 102155
    },
    {
      "epoch": 90.48715677590788,
      "grad_norm": 0.2386724352836609,
      "learning_rate": 1e-05,
      "loss": 0.971,
      "step": 102160
    },
    {
      "epoch": 90.49158547387069,
      "grad_norm": 0.20194567739963531,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 102165
    },
    {
      "epoch": 90.49601417183348,
      "grad_norm": 0.2279941290616989,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 102170
    },
    {
      "epoch": 90.50044286979627,
      "grad_norm": 0.24980227649211884,
      "learning_rate": 1e-05,
      "loss": 0.9357,
      "step": 102175
    },
    {
      "epoch": 90.50487156775908,
      "grad_norm": 0.26709261536598206,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 102180
    },
    {
      "epoch": 90.50930026572188,
      "grad_norm": 0.23109747469425201,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 102185
    },
    {
      "epoch": 90.51372896368467,
      "grad_norm": 0.2044699341058731,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 102190
    },
    {
      "epoch": 90.51815766164748,
      "grad_norm": 0.23382556438446045,
      "learning_rate": 1e-05,
      "loss": 0.9493,
      "step": 102195
    },
    {
      "epoch": 90.52258635961027,
      "grad_norm": 0.2366105318069458,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 102200
    },
    {
      "epoch": 90.52701505757307,
      "grad_norm": 0.23021741211414337,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 102205
    },
    {
      "epoch": 90.53144375553588,
      "grad_norm": 0.2401573657989502,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 102210
    },
    {
      "epoch": 90.53587245349867,
      "grad_norm": 0.21014755964279175,
      "learning_rate": 1e-05,
      "loss": 1.0152,
      "step": 102215
    },
    {
      "epoch": 90.54030115146146,
      "grad_norm": 0.2495400309562683,
      "learning_rate": 1e-05,
      "loss": 1.0276,
      "step": 102220
    },
    {
      "epoch": 90.54472984942427,
      "grad_norm": 0.24934224784374237,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 102225
    },
    {
      "epoch": 90.54915854738707,
      "grad_norm": 0.21330714225769043,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 102230
    },
    {
      "epoch": 90.55358724534987,
      "grad_norm": 0.2508140206336975,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 102235
    },
    {
      "epoch": 90.55801594331267,
      "grad_norm": 0.22362473607063293,
      "learning_rate": 1e-05,
      "loss": 0.9522,
      "step": 102240
    },
    {
      "epoch": 90.56244464127546,
      "grad_norm": 0.3272494971752167,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 102245
    },
    {
      "epoch": 90.56687333923827,
      "grad_norm": 0.23910930752754211,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 102250
    },
    {
      "epoch": 90.57130203720106,
      "grad_norm": 0.2771513760089874,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 102255
    },
    {
      "epoch": 90.57573073516386,
      "grad_norm": 0.2514110505580902,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 102260
    },
    {
      "epoch": 90.58015943312667,
      "grad_norm": 0.22452853620052338,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 102265
    },
    {
      "epoch": 90.58458813108946,
      "grad_norm": 0.22370469570159912,
      "learning_rate": 1e-05,
      "loss": 1.0365,
      "step": 102270
    },
    {
      "epoch": 90.58901682905226,
      "grad_norm": 0.2513195872306824,
      "learning_rate": 1e-05,
      "loss": 0.9891,
      "step": 102275
    },
    {
      "epoch": 90.59344552701506,
      "grad_norm": 0.24970132112503052,
      "learning_rate": 1e-05,
      "loss": 0.9202,
      "step": 102280
    },
    {
      "epoch": 90.59787422497786,
      "grad_norm": 0.24576792120933533,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 102285
    },
    {
      "epoch": 90.60230292294065,
      "grad_norm": 0.23058372735977173,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 102290
    },
    {
      "epoch": 90.60673162090346,
      "grad_norm": 0.22476179897785187,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 102295
    },
    {
      "epoch": 90.61116031886625,
      "grad_norm": 0.2078927457332611,
      "learning_rate": 1e-05,
      "loss": 0.9759,
      "step": 102300
    },
    {
      "epoch": 90.61558901682905,
      "grad_norm": 0.2763863503932953,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 102305
    },
    {
      "epoch": 90.62001771479186,
      "grad_norm": 0.20340348780155182,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 102310
    },
    {
      "epoch": 90.62444641275465,
      "grad_norm": 0.2515465021133423,
      "learning_rate": 1e-05,
      "loss": 0.924,
      "step": 102315
    },
    {
      "epoch": 90.62887511071744,
      "grad_norm": 0.23996025323867798,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 102320
    },
    {
      "epoch": 90.63330380868025,
      "grad_norm": 0.242412731051445,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 102325
    },
    {
      "epoch": 90.63773250664305,
      "grad_norm": 0.27623438835144043,
      "learning_rate": 1e-05,
      "loss": 0.9135,
      "step": 102330
    },
    {
      "epoch": 90.64216120460584,
      "grad_norm": 0.288758248090744,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 102335
    },
    {
      "epoch": 90.64658990256865,
      "grad_norm": 0.27220550179481506,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 102340
    },
    {
      "epoch": 90.65101860053144,
      "grad_norm": 0.2263362556695938,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 102345
    },
    {
      "epoch": 90.65544729849424,
      "grad_norm": 0.24110384285449982,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 102350
    },
    {
      "epoch": 90.65987599645705,
      "grad_norm": 0.2227737307548523,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 102355
    },
    {
      "epoch": 90.66430469441984,
      "grad_norm": 0.23778122663497925,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 102360
    },
    {
      "epoch": 90.66873339238263,
      "grad_norm": 0.263927161693573,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 102365
    },
    {
      "epoch": 90.67316209034544,
      "grad_norm": 0.2805740535259247,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 102370
    },
    {
      "epoch": 90.67759078830824,
      "grad_norm": 0.22144073247909546,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 102375
    },
    {
      "epoch": 90.68201948627103,
      "grad_norm": 0.2424430102109909,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 102380
    },
    {
      "epoch": 90.68644818423384,
      "grad_norm": 0.22817647457122803,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 102385
    },
    {
      "epoch": 90.69087688219663,
      "grad_norm": 0.21668201684951782,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 102390
    },
    {
      "epoch": 90.69530558015943,
      "grad_norm": 0.26588642597198486,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 102395
    },
    {
      "epoch": 90.69973427812224,
      "grad_norm": 0.27106720209121704,
      "learning_rate": 1e-05,
      "loss": 0.8924,
      "step": 102400
    },
    {
      "epoch": 90.70416297608503,
      "grad_norm": 0.2208392322063446,
      "learning_rate": 1e-05,
      "loss": 0.929,
      "step": 102405
    },
    {
      "epoch": 90.70859167404782,
      "grad_norm": 0.24484282732009888,
      "learning_rate": 1e-05,
      "loss": 0.9604,
      "step": 102410
    },
    {
      "epoch": 90.71302037201063,
      "grad_norm": 0.21779116988182068,
      "learning_rate": 1e-05,
      "loss": 0.942,
      "step": 102415
    },
    {
      "epoch": 90.71744906997343,
      "grad_norm": 0.1969458907842636,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 102420
    },
    {
      "epoch": 90.72187776793622,
      "grad_norm": 0.23428089916706085,
      "learning_rate": 1e-05,
      "loss": 0.9936,
      "step": 102425
    },
    {
      "epoch": 90.72630646589903,
      "grad_norm": 0.2588891386985779,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 102430
    },
    {
      "epoch": 90.73073516386182,
      "grad_norm": 0.27162402868270874,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 102435
    },
    {
      "epoch": 90.73516386182462,
      "grad_norm": 0.3209723234176636,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 102440
    },
    {
      "epoch": 90.73959255978743,
      "grad_norm": 0.22253888845443726,
      "learning_rate": 1e-05,
      "loss": 0.9991,
      "step": 102445
    },
    {
      "epoch": 90.74402125775022,
      "grad_norm": 0.21653956174850464,
      "learning_rate": 1e-05,
      "loss": 0.9397,
      "step": 102450
    },
    {
      "epoch": 90.74844995571301,
      "grad_norm": 0.23073752224445343,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 102455
    },
    {
      "epoch": 90.75287865367582,
      "grad_norm": 0.2643356919288635,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 102460
    },
    {
      "epoch": 90.75730735163862,
      "grad_norm": 0.2735397517681122,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 102465
    },
    {
      "epoch": 90.76173604960141,
      "grad_norm": 0.2693506181240082,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 102470
    },
    {
      "epoch": 90.76616474756422,
      "grad_norm": 0.22383420169353485,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 102475
    },
    {
      "epoch": 90.77059344552701,
      "grad_norm": 0.29040536284446716,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 102480
    },
    {
      "epoch": 90.77502214348982,
      "grad_norm": 0.2927846610546112,
      "learning_rate": 1e-05,
      "loss": 0.9171,
      "step": 102485
    },
    {
      "epoch": 90.77945084145261,
      "grad_norm": 0.24086327850818634,
      "learning_rate": 1e-05,
      "loss": 0.9063,
      "step": 102490
    },
    {
      "epoch": 90.78387953941541,
      "grad_norm": 0.2180607169866562,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 102495
    },
    {
      "epoch": 90.78830823737822,
      "grad_norm": 0.19970956444740295,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 102500
    },
    {
      "epoch": 90.79273693534101,
      "grad_norm": 0.20911867916584015,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 102505
    },
    {
      "epoch": 90.7971656333038,
      "grad_norm": 0.22589075565338135,
      "learning_rate": 1e-05,
      "loss": 1.0589,
      "step": 102510
    },
    {
      "epoch": 90.80159433126661,
      "grad_norm": 0.2891215980052948,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 102515
    },
    {
      "epoch": 90.80602302922941,
      "grad_norm": 0.25613537430763245,
      "learning_rate": 1e-05,
      "loss": 0.9122,
      "step": 102520
    },
    {
      "epoch": 90.8104517271922,
      "grad_norm": 0.237107053399086,
      "learning_rate": 1e-05,
      "loss": 0.9184,
      "step": 102525
    },
    {
      "epoch": 90.81488042515501,
      "grad_norm": 0.2320411503314972,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 102530
    },
    {
      "epoch": 90.8193091231178,
      "grad_norm": 0.2441605031490326,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 102535
    },
    {
      "epoch": 90.8237378210806,
      "grad_norm": 0.23499517142772675,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 102540
    },
    {
      "epoch": 90.8281665190434,
      "grad_norm": 0.2542986571788788,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 102545
    },
    {
      "epoch": 90.8325952170062,
      "grad_norm": 0.2669646739959717,
      "learning_rate": 1e-05,
      "loss": 0.9338,
      "step": 102550
    },
    {
      "epoch": 90.837023914969,
      "grad_norm": 0.2450842261314392,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 102555
    },
    {
      "epoch": 90.8414526129318,
      "grad_norm": 0.2577531635761261,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 102560
    },
    {
      "epoch": 90.8458813108946,
      "grad_norm": 0.237412229180336,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 102565
    },
    {
      "epoch": 90.85031000885739,
      "grad_norm": 0.25039517879486084,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 102570
    },
    {
      "epoch": 90.8547387068202,
      "grad_norm": 0.19879670441150665,
      "learning_rate": 1e-05,
      "loss": 0.9303,
      "step": 102575
    },
    {
      "epoch": 90.859167404783,
      "grad_norm": 0.2328568994998932,
      "learning_rate": 1e-05,
      "loss": 0.9911,
      "step": 102580
    },
    {
      "epoch": 90.86359610274579,
      "grad_norm": 0.22340106964111328,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 102585
    },
    {
      "epoch": 90.8680248007086,
      "grad_norm": 0.23357562720775604,
      "learning_rate": 1e-05,
      "loss": 0.925,
      "step": 102590
    },
    {
      "epoch": 90.87245349867139,
      "grad_norm": 0.2421547919511795,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 102595
    },
    {
      "epoch": 90.87688219663418,
      "grad_norm": 0.2209460288286209,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 102600
    },
    {
      "epoch": 90.881310894597,
      "grad_norm": 0.20997555553913116,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 102605
    },
    {
      "epoch": 90.88573959255979,
      "grad_norm": 0.21754190325737,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 102610
    },
    {
      "epoch": 90.89016829052258,
      "grad_norm": 0.2394261509180069,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 102615
    },
    {
      "epoch": 90.89459698848539,
      "grad_norm": 0.27301013469696045,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 102620
    },
    {
      "epoch": 90.89902568644818,
      "grad_norm": 0.2613600790500641,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 102625
    },
    {
      "epoch": 90.90345438441098,
      "grad_norm": 0.24907121062278748,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 102630
    },
    {
      "epoch": 90.90788308237379,
      "grad_norm": 0.2370993047952652,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 102635
    },
    {
      "epoch": 90.91231178033658,
      "grad_norm": 0.24002397060394287,
      "learning_rate": 1e-05,
      "loss": 1.0187,
      "step": 102640
    },
    {
      "epoch": 90.91674047829937,
      "grad_norm": 0.2537846565246582,
      "learning_rate": 1e-05,
      "loss": 0.9964,
      "step": 102645
    },
    {
      "epoch": 90.92116917626218,
      "grad_norm": 0.249927818775177,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 102650
    },
    {
      "epoch": 90.92559787422498,
      "grad_norm": 0.2988022267818451,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 102655
    },
    {
      "epoch": 90.93002657218777,
      "grad_norm": 0.24389055371284485,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 102660
    },
    {
      "epoch": 90.93445527015058,
      "grad_norm": 0.19886188209056854,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 102665
    },
    {
      "epoch": 90.93888396811337,
      "grad_norm": 0.23499402403831482,
      "learning_rate": 1e-05,
      "loss": 0.9163,
      "step": 102670
    },
    {
      "epoch": 90.94331266607617,
      "grad_norm": 0.2173004448413849,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 102675
    },
    {
      "epoch": 90.94774136403898,
      "grad_norm": 0.26165562868118286,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 102680
    },
    {
      "epoch": 90.95217006200177,
      "grad_norm": 0.2698575556278229,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 102685
    },
    {
      "epoch": 90.95659875996456,
      "grad_norm": 0.25355783104896545,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 102690
    },
    {
      "epoch": 90.96102745792737,
      "grad_norm": 0.2638607919216156,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 102695
    },
    {
      "epoch": 90.96545615589017,
      "grad_norm": 0.2440202832221985,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 102700
    },
    {
      "epoch": 90.96988485385296,
      "grad_norm": 0.2394225150346756,
      "learning_rate": 1e-05,
      "loss": 0.9004,
      "step": 102705
    },
    {
      "epoch": 90.97431355181577,
      "grad_norm": 0.22733670473098755,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 102710
    },
    {
      "epoch": 90.97874224977856,
      "grad_norm": 0.21324612200260162,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 102715
    },
    {
      "epoch": 90.98317094774137,
      "grad_norm": 0.23697666823863983,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 102720
    },
    {
      "epoch": 90.98759964570417,
      "grad_norm": 0.20261825621128082,
      "learning_rate": 1e-05,
      "loss": 0.9817,
      "step": 102725
    },
    {
      "epoch": 90.99202834366696,
      "grad_norm": 0.2591460943222046,
      "learning_rate": 1e-05,
      "loss": 0.9189,
      "step": 102730
    },
    {
      "epoch": 90.99645704162977,
      "grad_norm": 0.22383613884449005,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 102735
    },
    {
      "epoch": 91.00088573959256,
      "grad_norm": 0.27633705735206604,
      "learning_rate": 1e-05,
      "loss": 0.9932,
      "step": 102740
    },
    {
      "epoch": 91.00531443755536,
      "grad_norm": 0.20688210427761078,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 102745
    },
    {
      "epoch": 91.00974313551816,
      "grad_norm": 0.22417804598808289,
      "learning_rate": 1e-05,
      "loss": 1.0074,
      "step": 102750
    },
    {
      "epoch": 91.01417183348096,
      "grad_norm": 0.21591448783874512,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 102755
    },
    {
      "epoch": 91.01860053144375,
      "grad_norm": 0.2580453157424927,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 102760
    },
    {
      "epoch": 91.02302922940656,
      "grad_norm": 0.24250510334968567,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 102765
    },
    {
      "epoch": 91.02745792736935,
      "grad_norm": 0.25079652667045593,
      "learning_rate": 1e-05,
      "loss": 0.8957,
      "step": 102770
    },
    {
      "epoch": 91.03188662533215,
      "grad_norm": 0.3197646737098694,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 102775
    },
    {
      "epoch": 91.03631532329496,
      "grad_norm": 0.26770463585853577,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 102780
    },
    {
      "epoch": 91.04074402125775,
      "grad_norm": 0.2663167417049408,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 102785
    },
    {
      "epoch": 91.04517271922055,
      "grad_norm": 0.24982136487960815,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 102790
    },
    {
      "epoch": 91.04960141718335,
      "grad_norm": 0.22635824978351593,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 102795
    },
    {
      "epoch": 91.05403011514615,
      "grad_norm": 0.2758108079433441,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 102800
    },
    {
      "epoch": 91.05845881310894,
      "grad_norm": 0.26177066564559937,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 102805
    },
    {
      "epoch": 91.06288751107175,
      "grad_norm": 0.22039005160331726,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 102810
    },
    {
      "epoch": 91.06731620903454,
      "grad_norm": 0.2507549226284027,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 102815
    },
    {
      "epoch": 91.07174490699734,
      "grad_norm": 0.2560892701148987,
      "learning_rate": 1e-05,
      "loss": 0.9197,
      "step": 102820
    },
    {
      "epoch": 91.07617360496015,
      "grad_norm": 0.23397120833396912,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 102825
    },
    {
      "epoch": 91.08060230292294,
      "grad_norm": 0.24060940742492676,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 102830
    },
    {
      "epoch": 91.08503100088573,
      "grad_norm": 0.27812033891677856,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 102835
    },
    {
      "epoch": 91.08945969884854,
      "grad_norm": 0.2784745693206787,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 102840
    },
    {
      "epoch": 91.09388839681134,
      "grad_norm": 0.22309961915016174,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 102845
    },
    {
      "epoch": 91.09831709477413,
      "grad_norm": 0.23336555063724518,
      "learning_rate": 1e-05,
      "loss": 0.915,
      "step": 102850
    },
    {
      "epoch": 91.10274579273694,
      "grad_norm": 0.22197680175304413,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 102855
    },
    {
      "epoch": 91.10717449069973,
      "grad_norm": 0.22243289649486542,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 102860
    },
    {
      "epoch": 91.11160318866253,
      "grad_norm": 0.23436890542507172,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 102865
    },
    {
      "epoch": 91.11603188662534,
      "grad_norm": 0.24831825494766235,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 102870
    },
    {
      "epoch": 91.12046058458813,
      "grad_norm": 0.20246507227420807,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 102875
    },
    {
      "epoch": 91.12488928255092,
      "grad_norm": 0.2302592545747757,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 102880
    },
    {
      "epoch": 91.12931798051373,
      "grad_norm": 0.23203761875629425,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 102885
    },
    {
      "epoch": 91.13374667847653,
      "grad_norm": 0.2576983869075775,
      "learning_rate": 1e-05,
      "loss": 0.9167,
      "step": 102890
    },
    {
      "epoch": 91.13817537643932,
      "grad_norm": 0.2207961231470108,
      "learning_rate": 1e-05,
      "loss": 0.9111,
      "step": 102895
    },
    {
      "epoch": 91.14260407440213,
      "grad_norm": 0.23966224491596222,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 102900
    },
    {
      "epoch": 91.14703277236492,
      "grad_norm": 0.23793980479240417,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 102905
    },
    {
      "epoch": 91.15146147032772,
      "grad_norm": 0.23741868138313293,
      "learning_rate": 1e-05,
      "loss": 1.025,
      "step": 102910
    },
    {
      "epoch": 91.15589016829053,
      "grad_norm": 0.2640037536621094,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 102915
    },
    {
      "epoch": 91.16031886625332,
      "grad_norm": 0.22840575873851776,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 102920
    },
    {
      "epoch": 91.16474756421611,
      "grad_norm": 0.24525302648544312,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 102925
    },
    {
      "epoch": 91.16917626217892,
      "grad_norm": 0.26970311999320984,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 102930
    },
    {
      "epoch": 91.17360496014172,
      "grad_norm": 0.25400611758232117,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 102935
    },
    {
      "epoch": 91.17803365810451,
      "grad_norm": 0.296526700258255,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 102940
    },
    {
      "epoch": 91.18246235606732,
      "grad_norm": 0.25308018922805786,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 102945
    },
    {
      "epoch": 91.18689105403011,
      "grad_norm": 0.2912212908267975,
      "learning_rate": 1e-05,
      "loss": 1.0263,
      "step": 102950
    },
    {
      "epoch": 91.1913197519929,
      "grad_norm": 0.21558740735054016,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 102955
    },
    {
      "epoch": 91.19574844995572,
      "grad_norm": 0.20958438515663147,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 102960
    },
    {
      "epoch": 91.20017714791851,
      "grad_norm": 0.2251177579164505,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 102965
    },
    {
      "epoch": 91.20460584588132,
      "grad_norm": 0.2489888072013855,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 102970
    },
    {
      "epoch": 91.20903454384411,
      "grad_norm": 0.2482336163520813,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 102975
    },
    {
      "epoch": 91.2134632418069,
      "grad_norm": 0.20183423161506653,
      "learning_rate": 1e-05,
      "loss": 0.9906,
      "step": 102980
    },
    {
      "epoch": 91.21789193976971,
      "grad_norm": 0.22801020741462708,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 102985
    },
    {
      "epoch": 91.22232063773251,
      "grad_norm": 0.2383076697587967,
      "learning_rate": 1e-05,
      "loss": 0.9673,
      "step": 102990
    },
    {
      "epoch": 91.2267493356953,
      "grad_norm": 0.2368904948234558,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 102995
    },
    {
      "epoch": 91.23117803365811,
      "grad_norm": 0.23279932141304016,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 103000
    },
    {
      "epoch": 91.2356067316209,
      "grad_norm": 0.23190240561962128,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 103005
    },
    {
      "epoch": 91.2400354295837,
      "grad_norm": 0.2907813489437103,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 103010
    },
    {
      "epoch": 91.24446412754651,
      "grad_norm": 0.24953655898571014,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 103015
    },
    {
      "epoch": 91.2488928255093,
      "grad_norm": 0.22680160403251648,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 103020
    },
    {
      "epoch": 91.2533215234721,
      "grad_norm": 0.21623431146144867,
      "learning_rate": 1e-05,
      "loss": 1.0282,
      "step": 103025
    },
    {
      "epoch": 91.2577502214349,
      "grad_norm": 0.2570224404335022,
      "learning_rate": 1e-05,
      "loss": 0.9344,
      "step": 103030
    },
    {
      "epoch": 91.2621789193977,
      "grad_norm": 0.24587789177894592,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 103035
    },
    {
      "epoch": 91.26660761736049,
      "grad_norm": 0.2315250188112259,
      "learning_rate": 1e-05,
      "loss": 0.9264,
      "step": 103040
    },
    {
      "epoch": 91.2710363153233,
      "grad_norm": 0.20012202858924866,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 103045
    },
    {
      "epoch": 91.2754650132861,
      "grad_norm": 0.2329656034708023,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 103050
    },
    {
      "epoch": 91.27989371124889,
      "grad_norm": 0.22873006761074066,
      "learning_rate": 1e-05,
      "loss": 0.9109,
      "step": 103055
    },
    {
      "epoch": 91.2843224092117,
      "grad_norm": 0.2502482533454895,
      "learning_rate": 1e-05,
      "loss": 0.982,
      "step": 103060
    },
    {
      "epoch": 91.28875110717449,
      "grad_norm": 0.23776748776435852,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 103065
    },
    {
      "epoch": 91.29317980513729,
      "grad_norm": 0.2268352508544922,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 103070
    },
    {
      "epoch": 91.2976085031001,
      "grad_norm": 0.22304129600524902,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 103075
    },
    {
      "epoch": 91.30203720106289,
      "grad_norm": 0.237280935049057,
      "learning_rate": 1e-05,
      "loss": 0.9208,
      "step": 103080
    },
    {
      "epoch": 91.30646589902568,
      "grad_norm": 0.21677251160144806,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 103085
    },
    {
      "epoch": 91.31089459698849,
      "grad_norm": 0.25403955578804016,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 103090
    },
    {
      "epoch": 91.31532329495128,
      "grad_norm": 0.219672292470932,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 103095
    },
    {
      "epoch": 91.31975199291408,
      "grad_norm": 0.2725236117839813,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 103100
    },
    {
      "epoch": 91.32418069087689,
      "grad_norm": 0.2217569649219513,
      "learning_rate": 1e-05,
      "loss": 0.991,
      "step": 103105
    },
    {
      "epoch": 91.32860938883968,
      "grad_norm": 0.1981177031993866,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 103110
    },
    {
      "epoch": 91.33303808680247,
      "grad_norm": 0.23574258387088776,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 103115
    },
    {
      "epoch": 91.33746678476528,
      "grad_norm": 0.2810885012149811,
      "learning_rate": 1e-05,
      "loss": 0.9269,
      "step": 103120
    },
    {
      "epoch": 91.34189548272808,
      "grad_norm": 0.2744520306587219,
      "learning_rate": 1e-05,
      "loss": 0.9803,
      "step": 103125
    },
    {
      "epoch": 91.34632418069087,
      "grad_norm": 0.24710234999656677,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 103130
    },
    {
      "epoch": 91.35075287865368,
      "grad_norm": 0.1955631971359253,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 103135
    },
    {
      "epoch": 91.35518157661647,
      "grad_norm": 0.22885887324810028,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 103140
    },
    {
      "epoch": 91.35961027457927,
      "grad_norm": 0.25404462218284607,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 103145
    },
    {
      "epoch": 91.36403897254208,
      "grad_norm": 0.20895420014858246,
      "learning_rate": 1e-05,
      "loss": 1.0192,
      "step": 103150
    },
    {
      "epoch": 91.36846767050487,
      "grad_norm": 0.23463614284992218,
      "learning_rate": 1e-05,
      "loss": 0.9735,
      "step": 103155
    },
    {
      "epoch": 91.37289636846766,
      "grad_norm": 0.2608257830142975,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 103160
    },
    {
      "epoch": 91.37732506643047,
      "grad_norm": 0.25690919160842896,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 103165
    },
    {
      "epoch": 91.38175376439327,
      "grad_norm": 0.2564966678619385,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 103170
    },
    {
      "epoch": 91.38618246235606,
      "grad_norm": 0.24132072925567627,
      "learning_rate": 1e-05,
      "loss": 0.911,
      "step": 103175
    },
    {
      "epoch": 91.39061116031887,
      "grad_norm": 0.24077676236629486,
      "learning_rate": 1e-05,
      "loss": 0.9215,
      "step": 103180
    },
    {
      "epoch": 91.39503985828166,
      "grad_norm": 0.23886670172214508,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 103185
    },
    {
      "epoch": 91.39946855624446,
      "grad_norm": 0.2572554349899292,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 103190
    },
    {
      "epoch": 91.40389725420727,
      "grad_norm": 0.2579529285430908,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 103195
    },
    {
      "epoch": 91.40832595217006,
      "grad_norm": 0.23599202930927277,
      "learning_rate": 1e-05,
      "loss": 1.0052,
      "step": 103200
    },
    {
      "epoch": 91.41275465013285,
      "grad_norm": 0.2726353704929352,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 103205
    },
    {
      "epoch": 91.41718334809566,
      "grad_norm": 0.2780470848083496,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 103210
    },
    {
      "epoch": 91.42161204605846,
      "grad_norm": 0.25613561272621155,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 103215
    },
    {
      "epoch": 91.42604074402126,
      "grad_norm": 0.24017487466335297,
      "learning_rate": 1e-05,
      "loss": 0.9252,
      "step": 103220
    },
    {
      "epoch": 91.43046944198406,
      "grad_norm": 0.21221084892749786,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 103225
    },
    {
      "epoch": 91.43489813994685,
      "grad_norm": 0.24472211301326752,
      "learning_rate": 1e-05,
      "loss": 0.9629,
      "step": 103230
    },
    {
      "epoch": 91.43932683790966,
      "grad_norm": 0.2382994145154953,
      "learning_rate": 1e-05,
      "loss": 0.913,
      "step": 103235
    },
    {
      "epoch": 91.44375553587246,
      "grad_norm": 0.2526755928993225,
      "learning_rate": 1e-05,
      "loss": 0.9195,
      "step": 103240
    },
    {
      "epoch": 91.44818423383525,
      "grad_norm": 0.22146645188331604,
      "learning_rate": 1e-05,
      "loss": 0.9206,
      "step": 103245
    },
    {
      "epoch": 91.45261293179806,
      "grad_norm": 0.2645628750324249,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 103250
    },
    {
      "epoch": 91.45704162976085,
      "grad_norm": 0.25690531730651855,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 103255
    },
    {
      "epoch": 91.46147032772365,
      "grad_norm": 0.2326193004846573,
      "learning_rate": 1e-05,
      "loss": 0.9074,
      "step": 103260
    },
    {
      "epoch": 91.46589902568645,
      "grad_norm": 0.23041969537734985,
      "learning_rate": 1e-05,
      "loss": 0.8958,
      "step": 103265
    },
    {
      "epoch": 91.47032772364925,
      "grad_norm": 0.2243490368127823,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 103270
    },
    {
      "epoch": 91.47475642161204,
      "grad_norm": 0.2456541359424591,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 103275
    },
    {
      "epoch": 91.47918511957485,
      "grad_norm": 0.22051627933979034,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 103280
    },
    {
      "epoch": 91.48361381753764,
      "grad_norm": 0.22589299082756042,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 103285
    },
    {
      "epoch": 91.48804251550044,
      "grad_norm": 0.2465701550245285,
      "learning_rate": 1e-05,
      "loss": 0.9252,
      "step": 103290
    },
    {
      "epoch": 91.49247121346325,
      "grad_norm": 0.23816026747226715,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 103295
    },
    {
      "epoch": 91.49689991142604,
      "grad_norm": 0.25310397148132324,
      "learning_rate": 1e-05,
      "loss": 0.8909,
      "step": 103300
    },
    {
      "epoch": 91.50132860938884,
      "grad_norm": 0.24437053501605988,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 103305
    },
    {
      "epoch": 91.50575730735164,
      "grad_norm": 0.286076158285141,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 103310
    },
    {
      "epoch": 91.51018600531444,
      "grad_norm": 0.2814585864543915,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 103315
    },
    {
      "epoch": 91.51461470327723,
      "grad_norm": 0.2804497182369232,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 103320
    },
    {
      "epoch": 91.51904340124004,
      "grad_norm": 0.2686281204223633,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 103325
    },
    {
      "epoch": 91.52347209920283,
      "grad_norm": 0.253391295671463,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 103330
    },
    {
      "epoch": 91.52790079716563,
      "grad_norm": 0.28184524178504944,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 103335
    },
    {
      "epoch": 91.53232949512844,
      "grad_norm": 0.24850156903266907,
      "learning_rate": 1e-05,
      "loss": 0.9348,
      "step": 103340
    },
    {
      "epoch": 91.53675819309123,
      "grad_norm": 0.2278163731098175,
      "learning_rate": 1e-05,
      "loss": 0.9952,
      "step": 103345
    },
    {
      "epoch": 91.54118689105402,
      "grad_norm": 0.2612732946872711,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 103350
    },
    {
      "epoch": 91.54561558901683,
      "grad_norm": 0.22379198670387268,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 103355
    },
    {
      "epoch": 91.55004428697963,
      "grad_norm": 0.21157947182655334,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 103360
    },
    {
      "epoch": 91.55447298494242,
      "grad_norm": 0.22921548783779144,
      "learning_rate": 1e-05,
      "loss": 1.0296,
      "step": 103365
    },
    {
      "epoch": 91.55890168290523,
      "grad_norm": 0.2334277629852295,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 103370
    },
    {
      "epoch": 91.56333038086802,
      "grad_norm": 0.27004989981651306,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 103375
    },
    {
      "epoch": 91.56775907883082,
      "grad_norm": 0.24896419048309326,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 103380
    },
    {
      "epoch": 91.57218777679363,
      "grad_norm": 0.2145649939775467,
      "learning_rate": 1e-05,
      "loss": 0.9314,
      "step": 103385
    },
    {
      "epoch": 91.57661647475642,
      "grad_norm": 0.2485453188419342,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 103390
    },
    {
      "epoch": 91.58104517271921,
      "grad_norm": 0.2488047480583191,
      "learning_rate": 1e-05,
      "loss": 0.936,
      "step": 103395
    },
    {
      "epoch": 91.58547387068202,
      "grad_norm": 0.22245216369628906,
      "learning_rate": 1e-05,
      "loss": 0.9187,
      "step": 103400
    },
    {
      "epoch": 91.58990256864482,
      "grad_norm": 0.21726730465888977,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 103405
    },
    {
      "epoch": 91.59433126660761,
      "grad_norm": 0.2124769389629364,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 103410
    },
    {
      "epoch": 91.59875996457042,
      "grad_norm": 0.22882641851902008,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 103415
    },
    {
      "epoch": 91.60318866253321,
      "grad_norm": 0.21799640357494354,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 103420
    },
    {
      "epoch": 91.60761736049601,
      "grad_norm": 0.21663933992385864,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 103425
    },
    {
      "epoch": 91.61204605845882,
      "grad_norm": 0.2153746485710144,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 103430
    },
    {
      "epoch": 91.61647475642161,
      "grad_norm": 0.227844700217247,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 103435
    },
    {
      "epoch": 91.6209034543844,
      "grad_norm": 0.1955043524503708,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 103440
    },
    {
      "epoch": 91.62533215234721,
      "grad_norm": 0.23712210357189178,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 103445
    },
    {
      "epoch": 91.62976085031,
      "grad_norm": 0.22838705778121948,
      "learning_rate": 1e-05,
      "loss": 0.9297,
      "step": 103450
    },
    {
      "epoch": 91.63418954827281,
      "grad_norm": 0.265532910823822,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 103455
    },
    {
      "epoch": 91.63861824623561,
      "grad_norm": 0.259996235370636,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 103460
    },
    {
      "epoch": 91.6430469441984,
      "grad_norm": 0.25401797890663147,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 103465
    },
    {
      "epoch": 91.64747564216121,
      "grad_norm": 0.2826327681541443,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 103470
    },
    {
      "epoch": 91.651904340124,
      "grad_norm": 0.21522057056427002,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 103475
    },
    {
      "epoch": 91.6563330380868,
      "grad_norm": 0.2449258714914322,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 103480
    },
    {
      "epoch": 91.66076173604961,
      "grad_norm": 0.2270217090845108,
      "learning_rate": 1e-05,
      "loss": 0.914,
      "step": 103485
    },
    {
      "epoch": 91.6651904340124,
      "grad_norm": 0.23899859189987183,
      "learning_rate": 1e-05,
      "loss": 0.9193,
      "step": 103490
    },
    {
      "epoch": 91.6696191319752,
      "grad_norm": 0.26138898730278015,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 103495
    },
    {
      "epoch": 91.674047829938,
      "grad_norm": 0.2310044914484024,
      "learning_rate": 1e-05,
      "loss": 0.9162,
      "step": 103500
    },
    {
      "epoch": 91.6784765279008,
      "grad_norm": 0.24325542151927948,
      "learning_rate": 1e-05,
      "loss": 0.9016,
      "step": 103505
    },
    {
      "epoch": 91.68290522586359,
      "grad_norm": 0.2700601816177368,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 103510
    },
    {
      "epoch": 91.6873339238264,
      "grad_norm": 0.2320859283208847,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 103515
    },
    {
      "epoch": 91.6917626217892,
      "grad_norm": 0.21339771151542664,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 103520
    },
    {
      "epoch": 91.69619131975199,
      "grad_norm": 0.2141464501619339,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 103525
    },
    {
      "epoch": 91.7006200177148,
      "grad_norm": 0.2622971832752228,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 103530
    },
    {
      "epoch": 91.70504871567759,
      "grad_norm": 0.2728383243083954,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 103535
    },
    {
      "epoch": 91.70947741364039,
      "grad_norm": 0.27427875995635986,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 103540
    },
    {
      "epoch": 91.7139061116032,
      "grad_norm": 0.19493287801742554,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 103545
    },
    {
      "epoch": 91.71833480956599,
      "grad_norm": 0.23417465388774872,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 103550
    },
    {
      "epoch": 91.72276350752878,
      "grad_norm": 0.2390751987695694,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 103555
    },
    {
      "epoch": 91.72719220549159,
      "grad_norm": 0.2476937621831894,
      "learning_rate": 1e-05,
      "loss": 0.9995,
      "step": 103560
    },
    {
      "epoch": 91.73162090345438,
      "grad_norm": 0.22002314031124115,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 103565
    },
    {
      "epoch": 91.73604960141718,
      "grad_norm": 0.23944716155529022,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 103570
    },
    {
      "epoch": 91.74047829937999,
      "grad_norm": 0.24306169152259827,
      "learning_rate": 1e-05,
      "loss": 0.9487,
      "step": 103575
    },
    {
      "epoch": 91.74490699734278,
      "grad_norm": 0.25580257177352905,
      "learning_rate": 1e-05,
      "loss": 0.9136,
      "step": 103580
    },
    {
      "epoch": 91.74933569530558,
      "grad_norm": 0.22761861979961395,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 103585
    },
    {
      "epoch": 91.75376439326838,
      "grad_norm": 0.23008909821510315,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 103590
    },
    {
      "epoch": 91.75819309123118,
      "grad_norm": 0.22540812194347382,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 103595
    },
    {
      "epoch": 91.76262178919397,
      "grad_norm": 0.2673172056674957,
      "learning_rate": 1e-05,
      "loss": 0.9972,
      "step": 103600
    },
    {
      "epoch": 91.76705048715678,
      "grad_norm": 0.2583169937133789,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 103605
    },
    {
      "epoch": 91.77147918511957,
      "grad_norm": 0.2722247838973999,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 103610
    },
    {
      "epoch": 91.77590788308237,
      "grad_norm": 0.22237414121627808,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 103615
    },
    {
      "epoch": 91.78033658104518,
      "grad_norm": 0.28582963347435,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 103620
    },
    {
      "epoch": 91.78476527900797,
      "grad_norm": 0.26985061168670654,
      "learning_rate": 1e-05,
      "loss": 0.9357,
      "step": 103625
    },
    {
      "epoch": 91.78919397697076,
      "grad_norm": 0.26526930928230286,
      "learning_rate": 1e-05,
      "loss": 1.0225,
      "step": 103630
    },
    {
      "epoch": 91.79362267493357,
      "grad_norm": 0.24199633300304413,
      "learning_rate": 1e-05,
      "loss": 0.9949,
      "step": 103635
    },
    {
      "epoch": 91.79805137289637,
      "grad_norm": 0.2448570430278778,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 103640
    },
    {
      "epoch": 91.80248007085916,
      "grad_norm": 0.22262941300868988,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 103645
    },
    {
      "epoch": 91.80690876882197,
      "grad_norm": 0.20818395912647247,
      "learning_rate": 1e-05,
      "loss": 0.9157,
      "step": 103650
    },
    {
      "epoch": 91.81133746678476,
      "grad_norm": 0.28213945031166077,
      "learning_rate": 1e-05,
      "loss": 0.921,
      "step": 103655
    },
    {
      "epoch": 91.81576616474756,
      "grad_norm": 0.24637234210968018,
      "learning_rate": 1e-05,
      "loss": 0.9128,
      "step": 103660
    },
    {
      "epoch": 91.82019486271037,
      "grad_norm": 0.2656781077384949,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 103665
    },
    {
      "epoch": 91.82462356067316,
      "grad_norm": 0.24238321185112,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 103670
    },
    {
      "epoch": 91.82905225863595,
      "grad_norm": 0.25143301486968994,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 103675
    },
    {
      "epoch": 91.83348095659876,
      "grad_norm": 0.2636754512786865,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 103680
    },
    {
      "epoch": 91.83790965456156,
      "grad_norm": 0.27992525696754456,
      "learning_rate": 1e-05,
      "loss": 0.9658,
      "step": 103685
    },
    {
      "epoch": 91.84233835252435,
      "grad_norm": 0.21452893316745758,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 103690
    },
    {
      "epoch": 91.84676705048716,
      "grad_norm": 0.23102730512619019,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 103695
    },
    {
      "epoch": 91.85119574844995,
      "grad_norm": 0.2500423789024353,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 103700
    },
    {
      "epoch": 91.85562444641276,
      "grad_norm": 0.21993277966976166,
      "learning_rate": 1e-05,
      "loss": 0.9203,
      "step": 103705
    },
    {
      "epoch": 91.86005314437556,
      "grad_norm": 0.2363894283771515,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 103710
    },
    {
      "epoch": 91.86448184233835,
      "grad_norm": 0.2259645015001297,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 103715
    },
    {
      "epoch": 91.86891054030116,
      "grad_norm": 0.22022132575511932,
      "learning_rate": 1e-05,
      "loss": 0.9805,
      "step": 103720
    },
    {
      "epoch": 91.87333923826395,
      "grad_norm": 0.2187359780073166,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 103725
    },
    {
      "epoch": 91.87776793622675,
      "grad_norm": 0.23153722286224365,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 103730
    },
    {
      "epoch": 91.88219663418955,
      "grad_norm": 0.26593056321144104,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 103735
    },
    {
      "epoch": 91.88662533215235,
      "grad_norm": 0.24761411547660828,
      "learning_rate": 1e-05,
      "loss": 0.9157,
      "step": 103740
    },
    {
      "epoch": 91.89105403011514,
      "grad_norm": 0.19420082867145538,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 103745
    },
    {
      "epoch": 91.89548272807795,
      "grad_norm": 0.23348109424114227,
      "learning_rate": 1e-05,
      "loss": 1.0093,
      "step": 103750
    },
    {
      "epoch": 91.89991142604075,
      "grad_norm": 0.23166079819202423,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 103755
    },
    {
      "epoch": 91.90434012400354,
      "grad_norm": 0.23135671019554138,
      "learning_rate": 1e-05,
      "loss": 0.912,
      "step": 103760
    },
    {
      "epoch": 91.90876882196635,
      "grad_norm": 0.2203824818134308,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 103765
    },
    {
      "epoch": 91.91319751992914,
      "grad_norm": 0.23650586605072021,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 103770
    },
    {
      "epoch": 91.91762621789194,
      "grad_norm": 0.19482700526714325,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 103775
    },
    {
      "epoch": 91.92205491585474,
      "grad_norm": 0.21504123508930206,
      "learning_rate": 1e-05,
      "loss": 1.0577,
      "step": 103780
    },
    {
      "epoch": 91.92648361381754,
      "grad_norm": 0.21181507408618927,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 103785
    },
    {
      "epoch": 91.93091231178033,
      "grad_norm": 0.20825058221817017,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 103790
    },
    {
      "epoch": 91.93534100974314,
      "grad_norm": 0.2213062196969986,
      "learning_rate": 1e-05,
      "loss": 0.9218,
      "step": 103795
    },
    {
      "epoch": 91.93976970770593,
      "grad_norm": 0.22234788537025452,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 103800
    },
    {
      "epoch": 91.94419840566873,
      "grad_norm": 0.21452941000461578,
      "learning_rate": 1e-05,
      "loss": 0.9097,
      "step": 103805
    },
    {
      "epoch": 91.94862710363154,
      "grad_norm": 0.22836118936538696,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 103810
    },
    {
      "epoch": 91.95305580159433,
      "grad_norm": 0.22469556331634521,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 103815
    },
    {
      "epoch": 91.95748449955713,
      "grad_norm": 0.23338736593723297,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 103820
    },
    {
      "epoch": 91.96191319751993,
      "grad_norm": 0.24340717494487762,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 103825
    },
    {
      "epoch": 91.96634189548273,
      "grad_norm": 0.22474537789821625,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 103830
    },
    {
      "epoch": 91.97077059344552,
      "grad_norm": 0.27064454555511475,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 103835
    },
    {
      "epoch": 91.97519929140833,
      "grad_norm": 0.30823981761932373,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 103840
    },
    {
      "epoch": 91.97962798937112,
      "grad_norm": 0.24128656089305878,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 103845
    },
    {
      "epoch": 91.98405668733392,
      "grad_norm": 0.2586424648761749,
      "learning_rate": 1e-05,
      "loss": 0.8864,
      "step": 103850
    },
    {
      "epoch": 91.98848538529673,
      "grad_norm": 0.2608751654624939,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 103855
    },
    {
      "epoch": 91.99291408325952,
      "grad_norm": 0.26205989718437195,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 103860
    },
    {
      "epoch": 91.99734278122232,
      "grad_norm": 0.22395013272762299,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 103865
    },
    {
      "epoch": 92.00177147918512,
      "grad_norm": 0.23495995998382568,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 103870
    },
    {
      "epoch": 92.00620017714792,
      "grad_norm": 0.24605832993984222,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 103875
    },
    {
      "epoch": 92.01062887511071,
      "grad_norm": 0.24388107657432556,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 103880
    },
    {
      "epoch": 92.01505757307352,
      "grad_norm": 0.22551771998405457,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 103885
    },
    {
      "epoch": 92.01948627103631,
      "grad_norm": 0.20739533007144928,
      "learning_rate": 1e-05,
      "loss": 0.9462,
      "step": 103890
    },
    {
      "epoch": 92.02391496899911,
      "grad_norm": 0.208740696310997,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 103895
    },
    {
      "epoch": 92.02834366696192,
      "grad_norm": 0.24853219091892242,
      "learning_rate": 1e-05,
      "loss": 0.8905,
      "step": 103900
    },
    {
      "epoch": 92.03277236492471,
      "grad_norm": 0.23374173045158386,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 103905
    },
    {
      "epoch": 92.0372010628875,
      "grad_norm": 0.25845223665237427,
      "learning_rate": 1e-05,
      "loss": 0.942,
      "step": 103910
    },
    {
      "epoch": 92.04162976085031,
      "grad_norm": 0.242400661110878,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 103915
    },
    {
      "epoch": 92.0460584588131,
      "grad_norm": 0.2653234899044037,
      "learning_rate": 1e-05,
      "loss": 0.8951,
      "step": 103920
    },
    {
      "epoch": 92.0504871567759,
      "grad_norm": 0.23645661771297455,
      "learning_rate": 1e-05,
      "loss": 1.0128,
      "step": 103925
    },
    {
      "epoch": 92.05491585473871,
      "grad_norm": 0.2464582622051239,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 103930
    },
    {
      "epoch": 92.0593445527015,
      "grad_norm": 0.23376061022281647,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 103935
    },
    {
      "epoch": 92.0637732506643,
      "grad_norm": 0.26240673661231995,
      "learning_rate": 1e-05,
      "loss": 0.919,
      "step": 103940
    },
    {
      "epoch": 92.0682019486271,
      "grad_norm": 0.26038098335266113,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 103945
    },
    {
      "epoch": 92.0726306465899,
      "grad_norm": 0.21223558485507965,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 103950
    },
    {
      "epoch": 92.07705934455271,
      "grad_norm": 0.23779959976673126,
      "learning_rate": 1e-05,
      "loss": 0.9981,
      "step": 103955
    },
    {
      "epoch": 92.0814880425155,
      "grad_norm": 0.27635443210601807,
      "learning_rate": 1e-05,
      "loss": 0.9162,
      "step": 103960
    },
    {
      "epoch": 92.0859167404783,
      "grad_norm": 0.2691574990749359,
      "learning_rate": 1e-05,
      "loss": 1.0295,
      "step": 103965
    },
    {
      "epoch": 92.0903454384411,
      "grad_norm": 0.24721093475818634,
      "learning_rate": 1e-05,
      "loss": 0.9039,
      "step": 103970
    },
    {
      "epoch": 92.0947741364039,
      "grad_norm": 0.28207874298095703,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 103975
    },
    {
      "epoch": 92.0992028343667,
      "grad_norm": 0.2782837152481079,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 103980
    },
    {
      "epoch": 92.1036315323295,
      "grad_norm": 0.25969675183296204,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 103985
    },
    {
      "epoch": 92.1080602302923,
      "grad_norm": 0.2584705054759979,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 103990
    },
    {
      "epoch": 92.11248892825509,
      "grad_norm": 0.20821960270404816,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 103995
    },
    {
      "epoch": 92.1169176262179,
      "grad_norm": 0.2077736109495163,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 104000
    },
    {
      "epoch": 92.12134632418069,
      "grad_norm": 0.23634015023708344,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 104005
    },
    {
      "epoch": 92.12577502214349,
      "grad_norm": 0.23706485331058502,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 104010
    },
    {
      "epoch": 92.1302037201063,
      "grad_norm": 0.25284987688064575,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 104015
    },
    {
      "epoch": 92.13463241806909,
      "grad_norm": 0.237601637840271,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 104020
    },
    {
      "epoch": 92.13906111603188,
      "grad_norm": 0.23123103380203247,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 104025
    },
    {
      "epoch": 92.14348981399469,
      "grad_norm": 0.2782042920589447,
      "learning_rate": 1e-05,
      "loss": 0.925,
      "step": 104030
    },
    {
      "epoch": 92.14791851195749,
      "grad_norm": 0.2409246861934662,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 104035
    },
    {
      "epoch": 92.15234720992028,
      "grad_norm": 0.2444685846567154,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 104040
    },
    {
      "epoch": 92.15677590788309,
      "grad_norm": 0.24705785512924194,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 104045
    },
    {
      "epoch": 92.16120460584588,
      "grad_norm": 0.24171793460845947,
      "learning_rate": 1e-05,
      "loss": 0.8968,
      "step": 104050
    },
    {
      "epoch": 92.16563330380868,
      "grad_norm": 0.2305198609828949,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 104055
    },
    {
      "epoch": 92.17006200177148,
      "grad_norm": 0.26496249437332153,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 104060
    },
    {
      "epoch": 92.17449069973428,
      "grad_norm": 0.2260299175977707,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 104065
    },
    {
      "epoch": 92.17891939769707,
      "grad_norm": 0.22638362646102905,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 104070
    },
    {
      "epoch": 92.18334809565988,
      "grad_norm": 0.2787269353866577,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 104075
    },
    {
      "epoch": 92.18777679362267,
      "grad_norm": 0.22965730726718903,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 104080
    },
    {
      "epoch": 92.19220549158547,
      "grad_norm": 0.24454183876514435,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 104085
    },
    {
      "epoch": 92.19663418954828,
      "grad_norm": 0.21229122579097748,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 104090
    },
    {
      "epoch": 92.20106288751107,
      "grad_norm": 0.24226590991020203,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 104095
    },
    {
      "epoch": 92.20549158547387,
      "grad_norm": 0.21497997641563416,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 104100
    },
    {
      "epoch": 92.20992028343667,
      "grad_norm": 0.20268036425113678,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 104105
    },
    {
      "epoch": 92.21434898139947,
      "grad_norm": 0.28298458456993103,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 104110
    },
    {
      "epoch": 92.21877767936226,
      "grad_norm": 0.28500646352767944,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 104115
    },
    {
      "epoch": 92.22320637732507,
      "grad_norm": 0.23234178125858307,
      "learning_rate": 1e-05,
      "loss": 0.9196,
      "step": 104120
    },
    {
      "epoch": 92.22763507528786,
      "grad_norm": 0.23796726763248444,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 104125
    },
    {
      "epoch": 92.23206377325066,
      "grad_norm": 0.25222048163414,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 104130
    },
    {
      "epoch": 92.23649247121347,
      "grad_norm": 0.3183692395687103,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 104135
    },
    {
      "epoch": 92.24092116917626,
      "grad_norm": 0.27457526326179504,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 104140
    },
    {
      "epoch": 92.24534986713905,
      "grad_norm": 0.23743785917758942,
      "learning_rate": 1e-05,
      "loss": 0.9333,
      "step": 104145
    },
    {
      "epoch": 92.24977856510186,
      "grad_norm": 0.2591595947742462,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 104150
    },
    {
      "epoch": 92.25420726306466,
      "grad_norm": 0.24182827770709991,
      "learning_rate": 1e-05,
      "loss": 0.9306,
      "step": 104155
    },
    {
      "epoch": 92.25863596102745,
      "grad_norm": 0.25132352113723755,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 104160
    },
    {
      "epoch": 92.26306465899026,
      "grad_norm": 0.264423131942749,
      "learning_rate": 1e-05,
      "loss": 0.9707,
      "step": 104165
    },
    {
      "epoch": 92.26749335695305,
      "grad_norm": 0.24574121832847595,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 104170
    },
    {
      "epoch": 92.27192205491585,
      "grad_norm": 0.20907630026340485,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 104175
    },
    {
      "epoch": 92.27635075287866,
      "grad_norm": 0.23272904753684998,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 104180
    },
    {
      "epoch": 92.28077945084145,
      "grad_norm": 0.22407276928424835,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 104185
    },
    {
      "epoch": 92.28520814880426,
      "grad_norm": 0.23466186225414276,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 104190
    },
    {
      "epoch": 92.28963684676705,
      "grad_norm": 0.24560891091823578,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 104195
    },
    {
      "epoch": 92.29406554472985,
      "grad_norm": 0.22800418734550476,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 104200
    },
    {
      "epoch": 92.29849424269266,
      "grad_norm": 0.22940276563167572,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 104205
    },
    {
      "epoch": 92.30292294065545,
      "grad_norm": 0.19750966131687164,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 104210
    },
    {
      "epoch": 92.30735163861824,
      "grad_norm": 0.24489665031433105,
      "learning_rate": 1e-05,
      "loss": 0.9922,
      "step": 104215
    },
    {
      "epoch": 92.31178033658105,
      "grad_norm": 0.2556707561016083,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 104220
    },
    {
      "epoch": 92.31620903454385,
      "grad_norm": 0.250402569770813,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 104225
    },
    {
      "epoch": 92.32063773250664,
      "grad_norm": 0.23689934611320496,
      "learning_rate": 1e-05,
      "loss": 0.9476,
      "step": 104230
    },
    {
      "epoch": 92.32506643046945,
      "grad_norm": 0.24054846167564392,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 104235
    },
    {
      "epoch": 92.32949512843224,
      "grad_norm": 0.2402598112821579,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 104240
    },
    {
      "epoch": 92.33392382639504,
      "grad_norm": 0.22055096924304962,
      "learning_rate": 1e-05,
      "loss": 0.9982,
      "step": 104245
    },
    {
      "epoch": 92.33835252435784,
      "grad_norm": 0.24660301208496094,
      "learning_rate": 1e-05,
      "loss": 0.9989,
      "step": 104250
    },
    {
      "epoch": 92.34278122232064,
      "grad_norm": 0.2818809747695923,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 104255
    },
    {
      "epoch": 92.34720992028343,
      "grad_norm": 0.21271498501300812,
      "learning_rate": 1e-05,
      "loss": 0.9904,
      "step": 104260
    },
    {
      "epoch": 92.35163861824624,
      "grad_norm": 0.22835443913936615,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 104265
    },
    {
      "epoch": 92.35606731620904,
      "grad_norm": 0.25345247983932495,
      "learning_rate": 1e-05,
      "loss": 0.9174,
      "step": 104270
    },
    {
      "epoch": 92.36049601417183,
      "grad_norm": 0.21460489928722382,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 104275
    },
    {
      "epoch": 92.36492471213464,
      "grad_norm": 0.2641546130180359,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 104280
    },
    {
      "epoch": 92.36935341009743,
      "grad_norm": 0.23895984888076782,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 104285
    },
    {
      "epoch": 92.37378210806023,
      "grad_norm": 0.23548202216625214,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 104290
    },
    {
      "epoch": 92.37821080602303,
      "grad_norm": 0.2439844310283661,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 104295
    },
    {
      "epoch": 92.38263950398583,
      "grad_norm": 0.21296456456184387,
      "learning_rate": 1e-05,
      "loss": 0.913,
      "step": 104300
    },
    {
      "epoch": 92.38706820194862,
      "grad_norm": 0.22155778110027313,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 104305
    },
    {
      "epoch": 92.39149689991143,
      "grad_norm": 0.3749118745326996,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 104310
    },
    {
      "epoch": 92.39592559787422,
      "grad_norm": 0.27496710419654846,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 104315
    },
    {
      "epoch": 92.40035429583702,
      "grad_norm": 0.22657912969589233,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 104320
    },
    {
      "epoch": 92.40478299379983,
      "grad_norm": 0.23495355248451233,
      "learning_rate": 1e-05,
      "loss": 0.8912,
      "step": 104325
    },
    {
      "epoch": 92.40921169176262,
      "grad_norm": 0.25046372413635254,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 104330
    },
    {
      "epoch": 92.41364038972542,
      "grad_norm": 0.20248231291770935,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 104335
    },
    {
      "epoch": 92.41806908768822,
      "grad_norm": 0.2215609848499298,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 104340
    },
    {
      "epoch": 92.42249778565102,
      "grad_norm": 0.20338699221611023,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 104345
    },
    {
      "epoch": 92.42692648361381,
      "grad_norm": 0.23027391731739044,
      "learning_rate": 1e-05,
      "loss": 0.9075,
      "step": 104350
    },
    {
      "epoch": 92.43135518157662,
      "grad_norm": 0.2456299215555191,
      "learning_rate": 1e-05,
      "loss": 0.8935,
      "step": 104355
    },
    {
      "epoch": 92.43578387953941,
      "grad_norm": 0.234702467918396,
      "learning_rate": 1e-05,
      "loss": 0.918,
      "step": 104360
    },
    {
      "epoch": 92.44021257750221,
      "grad_norm": 0.23205651342868805,
      "learning_rate": 1e-05,
      "loss": 0.8771,
      "step": 104365
    },
    {
      "epoch": 92.44464127546502,
      "grad_norm": 0.25196555256843567,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 104370
    },
    {
      "epoch": 92.44906997342781,
      "grad_norm": 0.22299695014953613,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 104375
    },
    {
      "epoch": 92.4534986713906,
      "grad_norm": 0.2515668272972107,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 104380
    },
    {
      "epoch": 92.45792736935341,
      "grad_norm": 0.22013632953166962,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 104385
    },
    {
      "epoch": 92.46235606731621,
      "grad_norm": 0.2658098042011261,
      "learning_rate": 1e-05,
      "loss": 0.906,
      "step": 104390
    },
    {
      "epoch": 92.466784765279,
      "grad_norm": 0.23746754229068756,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 104395
    },
    {
      "epoch": 92.47121346324181,
      "grad_norm": 0.22035254538059235,
      "learning_rate": 1e-05,
      "loss": 0.9205,
      "step": 104400
    },
    {
      "epoch": 92.4756421612046,
      "grad_norm": 0.23228588700294495,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 104405
    },
    {
      "epoch": 92.4800708591674,
      "grad_norm": 0.22178740799427032,
      "learning_rate": 1e-05,
      "loss": 0.8872,
      "step": 104410
    },
    {
      "epoch": 92.4844995571302,
      "grad_norm": 0.23507189750671387,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 104415
    },
    {
      "epoch": 92.488928255093,
      "grad_norm": 0.29131391644477844,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 104420
    },
    {
      "epoch": 92.4933569530558,
      "grad_norm": 0.26209887862205505,
      "learning_rate": 1e-05,
      "loss": 1.0061,
      "step": 104425
    },
    {
      "epoch": 92.4977856510186,
      "grad_norm": 0.2290084809064865,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 104430
    },
    {
      "epoch": 92.5022143489814,
      "grad_norm": 0.25928211212158203,
      "learning_rate": 1e-05,
      "loss": 0.9851,
      "step": 104435
    },
    {
      "epoch": 92.5066430469442,
      "grad_norm": 0.24726904928684235,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 104440
    },
    {
      "epoch": 92.511071744907,
      "grad_norm": 0.2822262644767761,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 104445
    },
    {
      "epoch": 92.5155004428698,
      "grad_norm": 0.22445835173130035,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 104450
    },
    {
      "epoch": 92.5199291408326,
      "grad_norm": 0.22521314024925232,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 104455
    },
    {
      "epoch": 92.5243578387954,
      "grad_norm": 0.24269069731235504,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 104460
    },
    {
      "epoch": 92.52878653675819,
      "grad_norm": 0.23524372279644012,
      "learning_rate": 1e-05,
      "loss": 0.9318,
      "step": 104465
    },
    {
      "epoch": 92.533215234721,
      "grad_norm": 0.26258307695388794,
      "learning_rate": 1e-05,
      "loss": 0.9063,
      "step": 104470
    },
    {
      "epoch": 92.53764393268379,
      "grad_norm": 0.22780494391918182,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 104475
    },
    {
      "epoch": 92.54207263064659,
      "grad_norm": 0.23488205671310425,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 104480
    },
    {
      "epoch": 92.5465013286094,
      "grad_norm": 0.24419263005256653,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 104485
    },
    {
      "epoch": 92.55093002657219,
      "grad_norm": 0.22354215383529663,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 104490
    },
    {
      "epoch": 92.55535872453498,
      "grad_norm": 0.27550143003463745,
      "learning_rate": 1e-05,
      "loss": 0.9752,
      "step": 104495
    },
    {
      "epoch": 92.55978742249779,
      "grad_norm": 0.2394014596939087,
      "learning_rate": 1e-05,
      "loss": 0.9462,
      "step": 104500
    },
    {
      "epoch": 92.56421612046059,
      "grad_norm": 0.25666239857673645,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 104505
    },
    {
      "epoch": 92.56864481842338,
      "grad_norm": 0.22383306920528412,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 104510
    },
    {
      "epoch": 92.57307351638619,
      "grad_norm": 0.2127399742603302,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 104515
    },
    {
      "epoch": 92.57750221434898,
      "grad_norm": 0.2735644578933716,
      "learning_rate": 1e-05,
      "loss": 0.9033,
      "step": 104520
    },
    {
      "epoch": 92.58193091231178,
      "grad_norm": 0.2298261821269989,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 104525
    },
    {
      "epoch": 92.58635961027458,
      "grad_norm": 0.2576485574245453,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 104530
    },
    {
      "epoch": 92.59078830823738,
      "grad_norm": 0.2730349600315094,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 104535
    },
    {
      "epoch": 92.59521700620017,
      "grad_norm": 0.20015253126621246,
      "learning_rate": 1e-05,
      "loss": 0.9338,
      "step": 104540
    },
    {
      "epoch": 92.59964570416298,
      "grad_norm": 0.2402438223361969,
      "learning_rate": 1e-05,
      "loss": 0.927,
      "step": 104545
    },
    {
      "epoch": 92.60407440212578,
      "grad_norm": 0.2422506958246231,
      "learning_rate": 1e-05,
      "loss": 0.9222,
      "step": 104550
    },
    {
      "epoch": 92.60850310008857,
      "grad_norm": 0.23408442735671997,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 104555
    },
    {
      "epoch": 92.61293179805138,
      "grad_norm": 0.25045037269592285,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 104560
    },
    {
      "epoch": 92.61736049601417,
      "grad_norm": 0.28587305545806885,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 104565
    },
    {
      "epoch": 92.62178919397697,
      "grad_norm": 0.2303403615951538,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 104570
    },
    {
      "epoch": 92.62621789193977,
      "grad_norm": 0.22380352020263672,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 104575
    },
    {
      "epoch": 92.63064658990257,
      "grad_norm": 0.25911083817481995,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 104580
    },
    {
      "epoch": 92.63507528786536,
      "grad_norm": 0.2241116464138031,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 104585
    },
    {
      "epoch": 92.63950398582817,
      "grad_norm": 0.2572423815727234,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 104590
    },
    {
      "epoch": 92.64393268379096,
      "grad_norm": 0.26465338468551636,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 104595
    },
    {
      "epoch": 92.64836138175376,
      "grad_norm": 0.24486985802650452,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 104600
    },
    {
      "epoch": 92.65279007971657,
      "grad_norm": 0.20631392300128937,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 104605
    },
    {
      "epoch": 92.65721877767936,
      "grad_norm": 0.19914941489696503,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 104610
    },
    {
      "epoch": 92.66164747564216,
      "grad_norm": 0.22169391810894012,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 104615
    },
    {
      "epoch": 92.66607617360496,
      "grad_norm": 0.24760858714580536,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 104620
    },
    {
      "epoch": 92.67050487156776,
      "grad_norm": 0.2333918660879135,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 104625
    },
    {
      "epoch": 92.67493356953055,
      "grad_norm": 0.24727186560630798,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 104630
    },
    {
      "epoch": 92.67936226749336,
      "grad_norm": 0.2489270269870758,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 104635
    },
    {
      "epoch": 92.68379096545615,
      "grad_norm": 0.2481164187192917,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 104640
    },
    {
      "epoch": 92.68821966341895,
      "grad_norm": 0.23372292518615723,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 104645
    },
    {
      "epoch": 92.69264836138176,
      "grad_norm": 0.21750441193580627,
      "learning_rate": 1e-05,
      "loss": 0.9057,
      "step": 104650
    },
    {
      "epoch": 92.69707705934455,
      "grad_norm": 0.20124943554401398,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 104655
    },
    {
      "epoch": 92.70150575730734,
      "grad_norm": 0.2387171983718872,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 104660
    },
    {
      "epoch": 92.70593445527015,
      "grad_norm": 0.2116619199514389,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 104665
    },
    {
      "epoch": 92.71036315323295,
      "grad_norm": 0.2503810226917267,
      "learning_rate": 1e-05,
      "loss": 1.0141,
      "step": 104670
    },
    {
      "epoch": 92.71479185119574,
      "grad_norm": 0.23251168429851532,
      "learning_rate": 1e-05,
      "loss": 1.005,
      "step": 104675
    },
    {
      "epoch": 92.71922054915855,
      "grad_norm": 0.21455849707126617,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 104680
    },
    {
      "epoch": 92.72364924712134,
      "grad_norm": 0.23423728346824646,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 104685
    },
    {
      "epoch": 92.72807794508415,
      "grad_norm": 0.19891810417175293,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 104690
    },
    {
      "epoch": 92.73250664304695,
      "grad_norm": 0.27176955342292786,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 104695
    },
    {
      "epoch": 92.73693534100974,
      "grad_norm": 0.2571922242641449,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 104700
    },
    {
      "epoch": 92.74136403897255,
      "grad_norm": 0.27428558468818665,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 104705
    },
    {
      "epoch": 92.74579273693534,
      "grad_norm": 0.20511743426322937,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 104710
    },
    {
      "epoch": 92.75022143489814,
      "grad_norm": 0.22100993990898132,
      "learning_rate": 1e-05,
      "loss": 1.022,
      "step": 104715
    },
    {
      "epoch": 92.75465013286095,
      "grad_norm": 0.2564641237258911,
      "learning_rate": 1e-05,
      "loss": 0.888,
      "step": 104720
    },
    {
      "epoch": 92.75907883082374,
      "grad_norm": 0.23297759890556335,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 104725
    },
    {
      "epoch": 92.76350752878653,
      "grad_norm": 0.2554158568382263,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 104730
    },
    {
      "epoch": 92.76793622674934,
      "grad_norm": 0.25017249584198,
      "learning_rate": 1e-05,
      "loss": 1.0114,
      "step": 104735
    },
    {
      "epoch": 92.77236492471214,
      "grad_norm": 0.22326292097568512,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 104740
    },
    {
      "epoch": 92.77679362267493,
      "grad_norm": 0.24096561968326569,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 104745
    },
    {
      "epoch": 92.78122232063774,
      "grad_norm": 0.23591643571853638,
      "learning_rate": 1e-05,
      "loss": 0.9367,
      "step": 104750
    },
    {
      "epoch": 92.78565101860053,
      "grad_norm": 0.20652945339679718,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 104755
    },
    {
      "epoch": 92.79007971656333,
      "grad_norm": 0.2492186576128006,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 104760
    },
    {
      "epoch": 92.79450841452613,
      "grad_norm": 0.21970874071121216,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 104765
    },
    {
      "epoch": 92.79893711248893,
      "grad_norm": 0.24520668387413025,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 104770
    },
    {
      "epoch": 92.80336581045172,
      "grad_norm": 0.22851429879665375,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 104775
    },
    {
      "epoch": 92.80779450841453,
      "grad_norm": 0.28227218985557556,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 104780
    },
    {
      "epoch": 92.81222320637733,
      "grad_norm": 0.24791565537452698,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 104785
    },
    {
      "epoch": 92.81665190434012,
      "grad_norm": 0.40673360228538513,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 104790
    },
    {
      "epoch": 92.82108060230293,
      "grad_norm": 0.24803990125656128,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 104795
    },
    {
      "epoch": 92.82550930026572,
      "grad_norm": 0.22825345396995544,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 104800
    },
    {
      "epoch": 92.82993799822852,
      "grad_norm": 0.26857370138168335,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 104805
    },
    {
      "epoch": 92.83436669619132,
      "grad_norm": 0.21720892190933228,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 104810
    },
    {
      "epoch": 92.83879539415412,
      "grad_norm": 0.22743439674377441,
      "learning_rate": 1e-05,
      "loss": 0.9408,
      "step": 104815
    },
    {
      "epoch": 92.84322409211691,
      "grad_norm": 0.21095409989356995,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 104820
    },
    {
      "epoch": 92.84765279007972,
      "grad_norm": 0.2132745087146759,
      "learning_rate": 1e-05,
      "loss": 0.9697,
      "step": 104825
    },
    {
      "epoch": 92.85208148804251,
      "grad_norm": 0.23872998356819153,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 104830
    },
    {
      "epoch": 92.85651018600531,
      "grad_norm": 0.2352243810892105,
      "learning_rate": 1e-05,
      "loss": 1.0017,
      "step": 104835
    },
    {
      "epoch": 92.86093888396812,
      "grad_norm": 0.22757011651992798,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 104840
    },
    {
      "epoch": 92.86536758193091,
      "grad_norm": 0.2358512133359909,
      "learning_rate": 1e-05,
      "loss": 0.9058,
      "step": 104845
    },
    {
      "epoch": 92.8697962798937,
      "grad_norm": 0.2236064225435257,
      "learning_rate": 1e-05,
      "loss": 1.0125,
      "step": 104850
    },
    {
      "epoch": 92.87422497785651,
      "grad_norm": 0.2655746042728424,
      "learning_rate": 1e-05,
      "loss": 0.9726,
      "step": 104855
    },
    {
      "epoch": 92.87865367581931,
      "grad_norm": 0.2977094352245331,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 104860
    },
    {
      "epoch": 92.8830823737821,
      "grad_norm": 0.2825170159339905,
      "learning_rate": 1e-05,
      "loss": 1.0069,
      "step": 104865
    },
    {
      "epoch": 92.88751107174491,
      "grad_norm": 0.23948055505752563,
      "learning_rate": 1e-05,
      "loss": 0.9296,
      "step": 104870
    },
    {
      "epoch": 92.8919397697077,
      "grad_norm": 0.2645384967327118,
      "learning_rate": 1e-05,
      "loss": 0.9211,
      "step": 104875
    },
    {
      "epoch": 92.8963684676705,
      "grad_norm": 0.27442029118537903,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 104880
    },
    {
      "epoch": 92.9007971656333,
      "grad_norm": 0.24493233859539032,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 104885
    },
    {
      "epoch": 92.9052258635961,
      "grad_norm": 0.2219218909740448,
      "learning_rate": 1e-05,
      "loss": 0.8975,
      "step": 104890
    },
    {
      "epoch": 92.9096545615589,
      "grad_norm": 0.23851463198661804,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 104895
    },
    {
      "epoch": 92.9140832595217,
      "grad_norm": 0.21424821019172668,
      "learning_rate": 1e-05,
      "loss": 0.9231,
      "step": 104900
    },
    {
      "epoch": 92.9185119574845,
      "grad_norm": 0.22765296697616577,
      "learning_rate": 1e-05,
      "loss": 1.0331,
      "step": 104905
    },
    {
      "epoch": 92.92294065544729,
      "grad_norm": 0.25646573305130005,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 104910
    },
    {
      "epoch": 92.9273693534101,
      "grad_norm": 0.21691830456256866,
      "learning_rate": 1e-05,
      "loss": 0.8911,
      "step": 104915
    },
    {
      "epoch": 92.9317980513729,
      "grad_norm": 0.21906930208206177,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 104920
    },
    {
      "epoch": 92.9362267493357,
      "grad_norm": 0.2170998603105545,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 104925
    },
    {
      "epoch": 92.9406554472985,
      "grad_norm": 0.23521636426448822,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 104930
    },
    {
      "epoch": 92.94508414526129,
      "grad_norm": 0.2122250497341156,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 104935
    },
    {
      "epoch": 92.9495128432241,
      "grad_norm": 0.22160866856575012,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 104940
    },
    {
      "epoch": 92.9539415411869,
      "grad_norm": 0.22128285467624664,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 104945
    },
    {
      "epoch": 92.95837023914969,
      "grad_norm": 0.27809882164001465,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 104950
    },
    {
      "epoch": 92.9627989371125,
      "grad_norm": 0.21179543435573578,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 104955
    },
    {
      "epoch": 92.96722763507529,
      "grad_norm": 0.27684488892555237,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 104960
    },
    {
      "epoch": 92.97165633303808,
      "grad_norm": 0.2391420155763626,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 104965
    },
    {
      "epoch": 92.97608503100089,
      "grad_norm": 0.3003121018409729,
      "learning_rate": 1e-05,
      "loss": 0.9357,
      "step": 104970
    },
    {
      "epoch": 92.98051372896369,
      "grad_norm": 0.22289247810840607,
      "learning_rate": 1e-05,
      "loss": 0.8927,
      "step": 104975
    },
    {
      "epoch": 92.98494242692648,
      "grad_norm": 0.19042231142520905,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 104980
    },
    {
      "epoch": 92.98937112488929,
      "grad_norm": 0.22295452654361725,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 104985
    },
    {
      "epoch": 92.99379982285208,
      "grad_norm": 0.2865106463432312,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 104990
    },
    {
      "epoch": 92.99822852081488,
      "grad_norm": 0.22310854494571686,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 104995
    },
    {
      "epoch": 93.00265721877768,
      "grad_norm": 0.2363872081041336,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 105000
    },
    {
      "epoch": 93.00708591674048,
      "grad_norm": 0.20139619708061218,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 105005
    },
    {
      "epoch": 93.01151461470327,
      "grad_norm": 0.20604220032691956,
      "learning_rate": 1e-05,
      "loss": 0.9272,
      "step": 105010
    },
    {
      "epoch": 93.01594331266608,
      "grad_norm": 0.23860906064510345,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 105015
    },
    {
      "epoch": 93.02037201062888,
      "grad_norm": 0.24045388400554657,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 105020
    },
    {
      "epoch": 93.02480070859167,
      "grad_norm": 0.20502442121505737,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 105025
    },
    {
      "epoch": 93.02922940655448,
      "grad_norm": 0.2512694001197815,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 105030
    },
    {
      "epoch": 93.03365810451727,
      "grad_norm": 0.2656692862510681,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 105035
    },
    {
      "epoch": 93.03808680248007,
      "grad_norm": 0.25519657135009766,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 105040
    },
    {
      "epoch": 93.04251550044287,
      "grad_norm": 0.269862562417984,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 105045
    },
    {
      "epoch": 93.04694419840567,
      "grad_norm": 0.24477478861808777,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 105050
    },
    {
      "epoch": 93.05137289636846,
      "grad_norm": 0.25380614399909973,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 105055
    },
    {
      "epoch": 93.05580159433127,
      "grad_norm": 0.2589641511440277,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 105060
    },
    {
      "epoch": 93.06023029229407,
      "grad_norm": 0.22753207385540009,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 105065
    },
    {
      "epoch": 93.06465899025686,
      "grad_norm": 0.24029399454593658,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 105070
    },
    {
      "epoch": 93.06908768821967,
      "grad_norm": 0.2508775293827057,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 105075
    },
    {
      "epoch": 93.07351638618246,
      "grad_norm": 0.26097723841667175,
      "learning_rate": 1e-05,
      "loss": 0.9108,
      "step": 105080
    },
    {
      "epoch": 93.07794508414526,
      "grad_norm": 0.24411065876483917,
      "learning_rate": 1e-05,
      "loss": 0.9048,
      "step": 105085
    },
    {
      "epoch": 93.08237378210806,
      "grad_norm": 0.2114243507385254,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 105090
    },
    {
      "epoch": 93.08680248007086,
      "grad_norm": 0.24687489867210388,
      "learning_rate": 1e-05,
      "loss": 0.9103,
      "step": 105095
    },
    {
      "epoch": 93.09123117803365,
      "grad_norm": 0.2059982866048813,
      "learning_rate": 1e-05,
      "loss": 1.0134,
      "step": 105100
    },
    {
      "epoch": 93.09565987599646,
      "grad_norm": 0.23935602605342865,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 105105
    },
    {
      "epoch": 93.10008857395925,
      "grad_norm": 0.24573317170143127,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 105110
    },
    {
      "epoch": 93.10451727192205,
      "grad_norm": 0.2513274848461151,
      "learning_rate": 1e-05,
      "loss": 0.9293,
      "step": 105115
    },
    {
      "epoch": 93.10894596988486,
      "grad_norm": 0.24034105241298676,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 105120
    },
    {
      "epoch": 93.11337466784765,
      "grad_norm": 0.24127021431922913,
      "learning_rate": 1e-05,
      "loss": 0.9176,
      "step": 105125
    },
    {
      "epoch": 93.11780336581045,
      "grad_norm": 0.2650061845779419,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 105130
    },
    {
      "epoch": 93.12223206377325,
      "grad_norm": 0.25877442955970764,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 105135
    },
    {
      "epoch": 93.12666076173605,
      "grad_norm": 0.20887874066829681,
      "learning_rate": 1e-05,
      "loss": 0.9101,
      "step": 105140
    },
    {
      "epoch": 93.13108945969884,
      "grad_norm": 0.22704558074474335,
      "learning_rate": 1e-05,
      "loss": 1.0058,
      "step": 105145
    },
    {
      "epoch": 93.13551815766165,
      "grad_norm": 0.26891300082206726,
      "learning_rate": 1e-05,
      "loss": 0.9137,
      "step": 105150
    },
    {
      "epoch": 93.13994685562444,
      "grad_norm": 0.2265508770942688,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 105155
    },
    {
      "epoch": 93.14437555358724,
      "grad_norm": 0.29639148712158203,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 105160
    },
    {
      "epoch": 93.14880425155005,
      "grad_norm": 0.2827858030796051,
      "learning_rate": 1e-05,
      "loss": 0.9459,
      "step": 105165
    },
    {
      "epoch": 93.15323294951284,
      "grad_norm": 0.26621493697166443,
      "learning_rate": 1e-05,
      "loss": 0.9465,
      "step": 105170
    },
    {
      "epoch": 93.15766164747565,
      "grad_norm": 0.34894272685050964,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 105175
    },
    {
      "epoch": 93.16209034543844,
      "grad_norm": 0.30075308680534363,
      "learning_rate": 1e-05,
      "loss": 0.9078,
      "step": 105180
    },
    {
      "epoch": 93.16651904340124,
      "grad_norm": 0.3316897451877594,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 105185
    },
    {
      "epoch": 93.17094774136405,
      "grad_norm": 0.23709797859191895,
      "learning_rate": 1e-05,
      "loss": 0.909,
      "step": 105190
    },
    {
      "epoch": 93.17537643932684,
      "grad_norm": 0.2581119239330292,
      "learning_rate": 1e-05,
      "loss": 0.993,
      "step": 105195
    },
    {
      "epoch": 93.17980513728963,
      "grad_norm": 0.2426549792289734,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 105200
    },
    {
      "epoch": 93.18423383525244,
      "grad_norm": 0.2955203950405121,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 105205
    },
    {
      "epoch": 93.18866253321524,
      "grad_norm": 0.2558835744857788,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 105210
    },
    {
      "epoch": 93.19309123117803,
      "grad_norm": 0.2346285581588745,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 105215
    },
    {
      "epoch": 93.19751992914084,
      "grad_norm": 0.2578800618648529,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 105220
    },
    {
      "epoch": 93.20194862710363,
      "grad_norm": 0.24137237668037415,
      "learning_rate": 1e-05,
      "loss": 0.9178,
      "step": 105225
    },
    {
      "epoch": 93.20637732506643,
      "grad_norm": 0.23343440890312195,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 105230
    },
    {
      "epoch": 93.21080602302924,
      "grad_norm": 0.2665894329547882,
      "learning_rate": 1e-05,
      "loss": 1.0001,
      "step": 105235
    },
    {
      "epoch": 93.21523472099203,
      "grad_norm": 0.27070286870002747,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 105240
    },
    {
      "epoch": 93.21966341895482,
      "grad_norm": 0.2480572909116745,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 105245
    },
    {
      "epoch": 93.22409211691763,
      "grad_norm": 0.22742024064064026,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 105250
    },
    {
      "epoch": 93.22852081488043,
      "grad_norm": 0.21622078120708466,
      "learning_rate": 1e-05,
      "loss": 0.9526,
      "step": 105255
    },
    {
      "epoch": 93.23294951284322,
      "grad_norm": 0.230107843875885,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 105260
    },
    {
      "epoch": 93.23737821080603,
      "grad_norm": 0.2354736626148224,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 105265
    },
    {
      "epoch": 93.24180690876882,
      "grad_norm": 0.2273092418909073,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 105270
    },
    {
      "epoch": 93.24623560673162,
      "grad_norm": 0.22728396952152252,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 105275
    },
    {
      "epoch": 93.25066430469442,
      "grad_norm": 0.2526763677597046,
      "learning_rate": 1e-05,
      "loss": 1.0011,
      "step": 105280
    },
    {
      "epoch": 93.25509300265722,
      "grad_norm": 0.22754980623722076,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 105285
    },
    {
      "epoch": 93.25952170062001,
      "grad_norm": 0.237117737531662,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 105290
    },
    {
      "epoch": 93.26395039858282,
      "grad_norm": 0.25008267164230347,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 105295
    },
    {
      "epoch": 93.26837909654562,
      "grad_norm": 0.2122572809457779,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 105300
    },
    {
      "epoch": 93.27280779450841,
      "grad_norm": 0.26850852370262146,
      "learning_rate": 1e-05,
      "loss": 1.0009,
      "step": 105305
    },
    {
      "epoch": 93.27723649247122,
      "grad_norm": 0.2170819491147995,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 105310
    },
    {
      "epoch": 93.28166519043401,
      "grad_norm": 0.25903019309043884,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 105315
    },
    {
      "epoch": 93.2860938883968,
      "grad_norm": 0.22077183425426483,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 105320
    },
    {
      "epoch": 93.29052258635961,
      "grad_norm": 0.27361154556274414,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 105325
    },
    {
      "epoch": 93.29495128432241,
      "grad_norm": 0.2323404997587204,
      "learning_rate": 1e-05,
      "loss": 0.9986,
      "step": 105330
    },
    {
      "epoch": 93.2993799822852,
      "grad_norm": 0.23912915587425232,
      "learning_rate": 1e-05,
      "loss": 0.9274,
      "step": 105335
    },
    {
      "epoch": 93.30380868024801,
      "grad_norm": 0.23637878894805908,
      "learning_rate": 1e-05,
      "loss": 0.9253,
      "step": 105340
    },
    {
      "epoch": 93.3082373782108,
      "grad_norm": 0.21239687502384186,
      "learning_rate": 1e-05,
      "loss": 0.9722,
      "step": 105345
    },
    {
      "epoch": 93.3126660761736,
      "grad_norm": 0.2532808780670166,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 105350
    },
    {
      "epoch": 93.31709477413641,
      "grad_norm": 0.24801155924797058,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 105355
    },
    {
      "epoch": 93.3215234720992,
      "grad_norm": 0.2455122470855713,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 105360
    },
    {
      "epoch": 93.325952170062,
      "grad_norm": 0.2233913689851761,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 105365
    },
    {
      "epoch": 93.3303808680248,
      "grad_norm": 0.22580552101135254,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 105370
    },
    {
      "epoch": 93.3348095659876,
      "grad_norm": 0.2244798094034195,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 105375
    },
    {
      "epoch": 93.33923826395039,
      "grad_norm": 0.2265198677778244,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 105380
    },
    {
      "epoch": 93.3436669619132,
      "grad_norm": 0.26863136887550354,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 105385
    },
    {
      "epoch": 93.348095659876,
      "grad_norm": 0.23423950374126434,
      "learning_rate": 1e-05,
      "loss": 0.9271,
      "step": 105390
    },
    {
      "epoch": 93.35252435783879,
      "grad_norm": 0.2464640587568283,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 105395
    },
    {
      "epoch": 93.3569530558016,
      "grad_norm": 0.19595874845981598,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 105400
    },
    {
      "epoch": 93.36138175376439,
      "grad_norm": 0.23236393928527832,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 105405
    },
    {
      "epoch": 93.36581045172719,
      "grad_norm": 0.22964251041412354,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 105410
    },
    {
      "epoch": 93.37023914969,
      "grad_norm": 0.22863388061523438,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 105415
    },
    {
      "epoch": 93.37466784765279,
      "grad_norm": 0.24817189574241638,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 105420
    },
    {
      "epoch": 93.3790965456156,
      "grad_norm": 0.2519422769546509,
      "learning_rate": 1e-05,
      "loss": 0.8984,
      "step": 105425
    },
    {
      "epoch": 93.38352524357839,
      "grad_norm": 0.24377983808517456,
      "learning_rate": 1e-05,
      "loss": 0.936,
      "step": 105430
    },
    {
      "epoch": 93.38795394154118,
      "grad_norm": 0.253544420003891,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 105435
    },
    {
      "epoch": 93.39238263950399,
      "grad_norm": 0.24527321755886078,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 105440
    },
    {
      "epoch": 93.39681133746679,
      "grad_norm": 0.23329472541809082,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 105445
    },
    {
      "epoch": 93.40124003542958,
      "grad_norm": 0.23848466575145721,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 105450
    },
    {
      "epoch": 93.40566873339239,
      "grad_norm": 0.23122332990169525,
      "learning_rate": 1e-05,
      "loss": 0.8782,
      "step": 105455
    },
    {
      "epoch": 93.41009743135518,
      "grad_norm": 0.2294558584690094,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 105460
    },
    {
      "epoch": 93.41452612931798,
      "grad_norm": 0.24524912238121033,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 105465
    },
    {
      "epoch": 93.41895482728079,
      "grad_norm": 0.23014891147613525,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 105470
    },
    {
      "epoch": 93.42338352524358,
      "grad_norm": 0.1919795572757721,
      "learning_rate": 1e-05,
      "loss": 0.8927,
      "step": 105475
    },
    {
      "epoch": 93.42781222320637,
      "grad_norm": 0.25757303833961487,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 105480
    },
    {
      "epoch": 93.43224092116918,
      "grad_norm": 0.3006926476955414,
      "learning_rate": 1e-05,
      "loss": 0.9243,
      "step": 105485
    },
    {
      "epoch": 93.43666961913198,
      "grad_norm": 0.23858633637428284,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 105490
    },
    {
      "epoch": 93.44109831709477,
      "grad_norm": 0.2525339424610138,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 105495
    },
    {
      "epoch": 93.44552701505758,
      "grad_norm": 0.19873204827308655,
      "learning_rate": 1e-05,
      "loss": 0.9053,
      "step": 105500
    },
    {
      "epoch": 93.44995571302037,
      "grad_norm": 0.2233034074306488,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 105505
    },
    {
      "epoch": 93.45438441098317,
      "grad_norm": 0.2588280141353607,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 105510
    },
    {
      "epoch": 93.45881310894598,
      "grad_norm": 0.19022412598133087,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 105515
    },
    {
      "epoch": 93.46324180690877,
      "grad_norm": 0.24633122980594635,
      "learning_rate": 1e-05,
      "loss": 1.0333,
      "step": 105520
    },
    {
      "epoch": 93.46767050487156,
      "grad_norm": 0.24296921491622925,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 105525
    },
    {
      "epoch": 93.47209920283437,
      "grad_norm": 0.2565198838710785,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 105530
    },
    {
      "epoch": 93.47652790079717,
      "grad_norm": 0.23200194537639618,
      "learning_rate": 1e-05,
      "loss": 0.9549,
      "step": 105535
    },
    {
      "epoch": 93.48095659875996,
      "grad_norm": 0.2639910578727722,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 105540
    },
    {
      "epoch": 93.48538529672277,
      "grad_norm": 0.1978563517332077,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 105545
    },
    {
      "epoch": 93.48981399468556,
      "grad_norm": 0.21637532114982605,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 105550
    },
    {
      "epoch": 93.49424269264836,
      "grad_norm": 0.21952396631240845,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 105555
    },
    {
      "epoch": 93.49867139061116,
      "grad_norm": 0.24132317304611206,
      "learning_rate": 1e-05,
      "loss": 1.0003,
      "step": 105560
    },
    {
      "epoch": 93.50310008857396,
      "grad_norm": 0.23598463833332062,
      "learning_rate": 1e-05,
      "loss": 0.9879,
      "step": 105565
    },
    {
      "epoch": 93.50752878653675,
      "grad_norm": 0.22310227155685425,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 105570
    },
    {
      "epoch": 93.51195748449956,
      "grad_norm": 0.2703414261341095,
      "learning_rate": 1e-05,
      "loss": 0.8837,
      "step": 105575
    },
    {
      "epoch": 93.51638618246236,
      "grad_norm": 0.2524678409099579,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 105580
    },
    {
      "epoch": 93.52081488042515,
      "grad_norm": 0.22786295413970947,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 105585
    },
    {
      "epoch": 93.52524357838796,
      "grad_norm": 0.2373133897781372,
      "learning_rate": 1e-05,
      "loss": 1.0109,
      "step": 105590
    },
    {
      "epoch": 93.52967227635075,
      "grad_norm": 0.24242708086967468,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 105595
    },
    {
      "epoch": 93.53410097431355,
      "grad_norm": 0.2709553837776184,
      "learning_rate": 1e-05,
      "loss": 0.9227,
      "step": 105600
    },
    {
      "epoch": 93.53852967227635,
      "grad_norm": 0.21862784028053284,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 105605
    },
    {
      "epoch": 93.54295837023915,
      "grad_norm": 0.30443358421325684,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 105610
    },
    {
      "epoch": 93.54738706820194,
      "grad_norm": 0.2625056207180023,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 105615
    },
    {
      "epoch": 93.55181576616475,
      "grad_norm": 0.2663037180900574,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 105620
    },
    {
      "epoch": 93.55624446412754,
      "grad_norm": 0.25701725482940674,
      "learning_rate": 1e-05,
      "loss": 0.9721,
      "step": 105625
    },
    {
      "epoch": 93.56067316209034,
      "grad_norm": 0.27689310908317566,
      "learning_rate": 1e-05,
      "loss": 0.9211,
      "step": 105630
    },
    {
      "epoch": 93.56510186005315,
      "grad_norm": 0.2511526942253113,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 105635
    },
    {
      "epoch": 93.56953055801594,
      "grad_norm": 0.23363229632377625,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 105640
    },
    {
      "epoch": 93.57395925597874,
      "grad_norm": 0.21774157881736755,
      "learning_rate": 1e-05,
      "loss": 0.9634,
      "step": 105645
    },
    {
      "epoch": 93.57838795394154,
      "grad_norm": 0.266116738319397,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 105650
    },
    {
      "epoch": 93.58281665190434,
      "grad_norm": 0.26490187644958496,
      "learning_rate": 1e-05,
      "loss": 0.9246,
      "step": 105655
    },
    {
      "epoch": 93.58724534986715,
      "grad_norm": 0.24928782880306244,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 105660
    },
    {
      "epoch": 93.59167404782994,
      "grad_norm": 0.26242780685424805,
      "learning_rate": 1e-05,
      "loss": 0.9034,
      "step": 105665
    },
    {
      "epoch": 93.59610274579273,
      "grad_norm": 0.22840522229671478,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 105670
    },
    {
      "epoch": 93.60053144375554,
      "grad_norm": 0.3233090937137604,
      "learning_rate": 1e-05,
      "loss": 0.9167,
      "step": 105675
    },
    {
      "epoch": 93.60496014171834,
      "grad_norm": 0.28963232040405273,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 105680
    },
    {
      "epoch": 93.60938883968113,
      "grad_norm": 0.21576011180877686,
      "learning_rate": 1e-05,
      "loss": 0.9347,
      "step": 105685
    },
    {
      "epoch": 93.61381753764394,
      "grad_norm": 0.2524646818637848,
      "learning_rate": 1e-05,
      "loss": 0.8925,
      "step": 105690
    },
    {
      "epoch": 93.61824623560673,
      "grad_norm": 0.23083454370498657,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 105695
    },
    {
      "epoch": 93.62267493356953,
      "grad_norm": 0.2314504086971283,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 105700
    },
    {
      "epoch": 93.62710363153234,
      "grad_norm": 0.2916317284107208,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 105705
    },
    {
      "epoch": 93.63153232949513,
      "grad_norm": 0.24366340041160583,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 105710
    },
    {
      "epoch": 93.63596102745792,
      "grad_norm": 0.2564623951911926,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 105715
    },
    {
      "epoch": 93.64038972542073,
      "grad_norm": 0.23970985412597656,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 105720
    },
    {
      "epoch": 93.64481842338353,
      "grad_norm": 0.25679343938827515,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 105725
    },
    {
      "epoch": 93.64924712134632,
      "grad_norm": 0.24072374403476715,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 105730
    },
    {
      "epoch": 93.65367581930913,
      "grad_norm": 0.23277287185192108,
      "learning_rate": 1e-05,
      "loss": 0.9206,
      "step": 105735
    },
    {
      "epoch": 93.65810451727192,
      "grad_norm": 0.2334434539079666,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 105740
    },
    {
      "epoch": 93.66253321523472,
      "grad_norm": 0.26355889439582825,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 105745
    },
    {
      "epoch": 93.66696191319753,
      "grad_norm": 0.236064150929451,
      "learning_rate": 1e-05,
      "loss": 0.9357,
      "step": 105750
    },
    {
      "epoch": 93.67139061116032,
      "grad_norm": 0.21392065286636353,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 105755
    },
    {
      "epoch": 93.67581930912311,
      "grad_norm": 0.2361832857131958,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 105760
    },
    {
      "epoch": 93.68024800708592,
      "grad_norm": 0.23423390090465546,
      "learning_rate": 1e-05,
      "loss": 0.912,
      "step": 105765
    },
    {
      "epoch": 93.68467670504872,
      "grad_norm": 0.22610947489738464,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 105770
    },
    {
      "epoch": 93.68910540301151,
      "grad_norm": 0.27814966440200806,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 105775
    },
    {
      "epoch": 93.69353410097432,
      "grad_norm": 0.2127348929643631,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 105780
    },
    {
      "epoch": 93.69796279893711,
      "grad_norm": 0.2242303192615509,
      "learning_rate": 1e-05,
      "loss": 0.9073,
      "step": 105785
    },
    {
      "epoch": 93.7023914968999,
      "grad_norm": 0.21777324378490448,
      "learning_rate": 1e-05,
      "loss": 0.9335,
      "step": 105790
    },
    {
      "epoch": 93.70682019486271,
      "grad_norm": 0.24466101825237274,
      "learning_rate": 1e-05,
      "loss": 0.9227,
      "step": 105795
    },
    {
      "epoch": 93.71124889282551,
      "grad_norm": 0.2629692256450653,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 105800
    },
    {
      "epoch": 93.7156775907883,
      "grad_norm": 0.26391953229904175,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 105805
    },
    {
      "epoch": 93.72010628875111,
      "grad_norm": 0.2962143123149872,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 105810
    },
    {
      "epoch": 93.7245349867139,
      "grad_norm": 0.26831114292144775,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 105815
    },
    {
      "epoch": 93.7289636846767,
      "grad_norm": 0.1945819854736328,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 105820
    },
    {
      "epoch": 93.73339238263951,
      "grad_norm": 0.21918925642967224,
      "learning_rate": 1e-05,
      "loss": 0.9195,
      "step": 105825
    },
    {
      "epoch": 93.7378210806023,
      "grad_norm": 0.22030341625213623,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 105830
    },
    {
      "epoch": 93.7422497785651,
      "grad_norm": 0.23286330699920654,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 105835
    },
    {
      "epoch": 93.7466784765279,
      "grad_norm": 0.22802424430847168,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 105840
    },
    {
      "epoch": 93.7511071744907,
      "grad_norm": 0.232124462723732,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 105845
    },
    {
      "epoch": 93.75553587245349,
      "grad_norm": 0.23087060451507568,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 105850
    },
    {
      "epoch": 93.7599645704163,
      "grad_norm": 0.22931510210037231,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 105855
    },
    {
      "epoch": 93.7643932683791,
      "grad_norm": 0.20915593206882477,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 105860
    },
    {
      "epoch": 93.76882196634189,
      "grad_norm": 0.30223989486694336,
      "learning_rate": 1e-05,
      "loss": 1.0422,
      "step": 105865
    },
    {
      "epoch": 93.7732506643047,
      "grad_norm": 0.2437208592891693,
      "learning_rate": 1e-05,
      "loss": 0.9818,
      "step": 105870
    },
    {
      "epoch": 93.77767936226749,
      "grad_norm": 0.2636796236038208,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 105875
    },
    {
      "epoch": 93.78210806023029,
      "grad_norm": 0.23928864300251007,
      "learning_rate": 1e-05,
      "loss": 1.0097,
      "step": 105880
    },
    {
      "epoch": 93.7865367581931,
      "grad_norm": 0.2815961241722107,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 105885
    },
    {
      "epoch": 93.79096545615589,
      "grad_norm": 0.2229508012533188,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 105890
    },
    {
      "epoch": 93.79539415411868,
      "grad_norm": 0.24366120994091034,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 105895
    },
    {
      "epoch": 93.79982285208149,
      "grad_norm": 0.2436981350183487,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 105900
    },
    {
      "epoch": 93.80425155004428,
      "grad_norm": 0.21499690413475037,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 105905
    },
    {
      "epoch": 93.8086802480071,
      "grad_norm": 0.23913350701332092,
      "learning_rate": 1e-05,
      "loss": 0.9148,
      "step": 105910
    },
    {
      "epoch": 93.81310894596989,
      "grad_norm": 0.23672091960906982,
      "learning_rate": 1e-05,
      "loss": 1.0014,
      "step": 105915
    },
    {
      "epoch": 93.81753764393268,
      "grad_norm": 0.25560393929481506,
      "learning_rate": 1e-05,
      "loss": 0.8931,
      "step": 105920
    },
    {
      "epoch": 93.82196634189549,
      "grad_norm": 0.23304376006126404,
      "learning_rate": 1e-05,
      "loss": 0.8914,
      "step": 105925
    },
    {
      "epoch": 93.82639503985828,
      "grad_norm": 0.2780101001262665,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 105930
    },
    {
      "epoch": 93.83082373782108,
      "grad_norm": 0.20780281722545624,
      "learning_rate": 1e-05,
      "loss": 0.954,
      "step": 105935
    },
    {
      "epoch": 93.83525243578389,
      "grad_norm": 0.29038724303245544,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 105940
    },
    {
      "epoch": 93.83968113374668,
      "grad_norm": 0.24847756326198578,
      "learning_rate": 1e-05,
      "loss": 0.9289,
      "step": 105945
    },
    {
      "epoch": 93.84410983170947,
      "grad_norm": 0.2558285593986511,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 105950
    },
    {
      "epoch": 93.84853852967228,
      "grad_norm": 0.2624794542789459,
      "learning_rate": 1e-05,
      "loss": 0.9129,
      "step": 105955
    },
    {
      "epoch": 93.85296722763508,
      "grad_norm": 0.2644163966178894,
      "learning_rate": 1e-05,
      "loss": 1.0023,
      "step": 105960
    },
    {
      "epoch": 93.85739592559787,
      "grad_norm": 0.2528930604457855,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 105965
    },
    {
      "epoch": 93.86182462356068,
      "grad_norm": 0.2019290030002594,
      "learning_rate": 1e-05,
      "loss": 0.9758,
      "step": 105970
    },
    {
      "epoch": 93.86625332152347,
      "grad_norm": 0.24584545195102692,
      "learning_rate": 1e-05,
      "loss": 0.9177,
      "step": 105975
    },
    {
      "epoch": 93.87068201948627,
      "grad_norm": 0.23608802258968353,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 105980
    },
    {
      "epoch": 93.87511071744908,
      "grad_norm": 0.25936681032180786,
      "learning_rate": 1e-05,
      "loss": 0.9227,
      "step": 105985
    },
    {
      "epoch": 93.87953941541187,
      "grad_norm": 0.2223770022392273,
      "learning_rate": 1e-05,
      "loss": 0.9843,
      "step": 105990
    },
    {
      "epoch": 93.88396811337466,
      "grad_norm": 0.2358471006155014,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 105995
    },
    {
      "epoch": 93.88839681133747,
      "grad_norm": 0.23934701085090637,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 106000
    },
    {
      "epoch": 93.89282550930027,
      "grad_norm": 0.24968120455741882,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 106005
    },
    {
      "epoch": 93.89725420726306,
      "grad_norm": 0.23409950733184814,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 106010
    },
    {
      "epoch": 93.90168290522587,
      "grad_norm": 0.2168843001127243,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 106015
    },
    {
      "epoch": 93.90611160318866,
      "grad_norm": 0.2087348997592926,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 106020
    },
    {
      "epoch": 93.91054030115146,
      "grad_norm": 0.23356500267982483,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 106025
    },
    {
      "epoch": 93.91496899911427,
      "grad_norm": 0.220457524061203,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 106030
    },
    {
      "epoch": 93.91939769707706,
      "grad_norm": 0.21549856662750244,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 106035
    },
    {
      "epoch": 93.92382639503985,
      "grad_norm": 0.21504388749599457,
      "learning_rate": 1e-05,
      "loss": 0.9082,
      "step": 106040
    },
    {
      "epoch": 93.92825509300266,
      "grad_norm": 0.23265616595745087,
      "learning_rate": 1e-05,
      "loss": 0.8788,
      "step": 106045
    },
    {
      "epoch": 93.93268379096546,
      "grad_norm": 0.28221890330314636,
      "learning_rate": 1e-05,
      "loss": 0.951,
      "step": 106050
    },
    {
      "epoch": 93.93711248892825,
      "grad_norm": 0.2596142888069153,
      "learning_rate": 1e-05,
      "loss": 0.9273,
      "step": 106055
    },
    {
      "epoch": 93.94154118689106,
      "grad_norm": 0.21485969424247742,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 106060
    },
    {
      "epoch": 93.94596988485385,
      "grad_norm": 0.22859859466552734,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 106065
    },
    {
      "epoch": 93.95039858281665,
      "grad_norm": 0.2548113465309143,
      "learning_rate": 1e-05,
      "loss": 0.9416,
      "step": 106070
    },
    {
      "epoch": 93.95482728077945,
      "grad_norm": 0.258844792842865,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 106075
    },
    {
      "epoch": 93.95925597874225,
      "grad_norm": 0.22398611903190613,
      "learning_rate": 1e-05,
      "loss": 0.9123,
      "step": 106080
    },
    {
      "epoch": 93.96368467670504,
      "grad_norm": 0.24999378621578217,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 106085
    },
    {
      "epoch": 93.96811337466785,
      "grad_norm": 0.22970613837242126,
      "learning_rate": 1e-05,
      "loss": 0.9062,
      "step": 106090
    },
    {
      "epoch": 93.97254207263065,
      "grad_norm": 0.24498316645622253,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 106095
    },
    {
      "epoch": 93.97697077059344,
      "grad_norm": 0.2519926428794861,
      "learning_rate": 1e-05,
      "loss": 1.036,
      "step": 106100
    },
    {
      "epoch": 93.98139946855625,
      "grad_norm": 0.24293173849582672,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 106105
    },
    {
      "epoch": 93.98582816651904,
      "grad_norm": 0.24342729151248932,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 106110
    },
    {
      "epoch": 93.99025686448184,
      "grad_norm": 0.23912575840950012,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 106115
    },
    {
      "epoch": 93.99468556244464,
      "grad_norm": 0.25066015124320984,
      "learning_rate": 1e-05,
      "loss": 0.8752,
      "step": 106120
    },
    {
      "epoch": 93.99911426040744,
      "grad_norm": 0.24286554753780365,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 106125
    },
    {
      "epoch": 94.00354295837023,
      "grad_norm": 0.32619595527648926,
      "learning_rate": 1e-05,
      "loss": 0.9093,
      "step": 106130
    },
    {
      "epoch": 94.00797165633304,
      "grad_norm": 0.2892949879169464,
      "learning_rate": 1e-05,
      "loss": 0.8868,
      "step": 106135
    },
    {
      "epoch": 94.01240035429583,
      "grad_norm": 0.2920706272125244,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 106140
    },
    {
      "epoch": 94.01682905225863,
      "grad_norm": 0.22681845724582672,
      "learning_rate": 1e-05,
      "loss": 0.9482,
      "step": 106145
    },
    {
      "epoch": 94.02125775022144,
      "grad_norm": 0.2509005069732666,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 106150
    },
    {
      "epoch": 94.02568644818423,
      "grad_norm": 0.24970906972885132,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 106155
    },
    {
      "epoch": 94.03011514614704,
      "grad_norm": 0.2244836837053299,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 106160
    },
    {
      "epoch": 94.03454384410983,
      "grad_norm": 0.28171905875205994,
      "learning_rate": 1e-05,
      "loss": 0.9938,
      "step": 106165
    },
    {
      "epoch": 94.03897254207263,
      "grad_norm": 0.26539576053619385,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 106170
    },
    {
      "epoch": 94.04340124003544,
      "grad_norm": 0.29274851083755493,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 106175
    },
    {
      "epoch": 94.04782993799823,
      "grad_norm": 0.22678150236606598,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 106180
    },
    {
      "epoch": 94.05225863596102,
      "grad_norm": 0.24295900762081146,
      "learning_rate": 1e-05,
      "loss": 0.9742,
      "step": 106185
    },
    {
      "epoch": 94.05668733392383,
      "grad_norm": 0.2369834929704666,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 106190
    },
    {
      "epoch": 94.06111603188663,
      "grad_norm": 0.2000340223312378,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 106195
    },
    {
      "epoch": 94.06554472984942,
      "grad_norm": 0.20921172201633453,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 106200
    },
    {
      "epoch": 94.06997342781223,
      "grad_norm": 0.24889475107192993,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 106205
    },
    {
      "epoch": 94.07440212577502,
      "grad_norm": 0.2599664628505707,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 106210
    },
    {
      "epoch": 94.07883082373782,
      "grad_norm": 0.251973420381546,
      "learning_rate": 1e-05,
      "loss": 1.0244,
      "step": 106215
    },
    {
      "epoch": 94.08325952170063,
      "grad_norm": 0.26440638303756714,
      "learning_rate": 1e-05,
      "loss": 0.9379,
      "step": 106220
    },
    {
      "epoch": 94.08768821966342,
      "grad_norm": 0.2862977087497711,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 106225
    },
    {
      "epoch": 94.09211691762621,
      "grad_norm": 0.21448568999767303,
      "learning_rate": 1e-05,
      "loss": 0.9159,
      "step": 106230
    },
    {
      "epoch": 94.09654561558902,
      "grad_norm": 0.2629851698875427,
      "learning_rate": 1e-05,
      "loss": 0.9383,
      "step": 106235
    },
    {
      "epoch": 94.10097431355182,
      "grad_norm": 0.24043430387973785,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 106240
    },
    {
      "epoch": 94.10540301151461,
      "grad_norm": 0.24246253073215485,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 106245
    },
    {
      "epoch": 94.10983170947742,
      "grad_norm": 0.2435413897037506,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 106250
    },
    {
      "epoch": 94.11426040744021,
      "grad_norm": 0.27490028738975525,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 106255
    },
    {
      "epoch": 94.118689105403,
      "grad_norm": 0.22965127229690552,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 106260
    },
    {
      "epoch": 94.12311780336582,
      "grad_norm": 0.22054623067378998,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 106265
    },
    {
      "epoch": 94.12754650132861,
      "grad_norm": 0.23733578622341156,
      "learning_rate": 1e-05,
      "loss": 0.8998,
      "step": 106270
    },
    {
      "epoch": 94.1319751992914,
      "grad_norm": 0.24123911559581757,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 106275
    },
    {
      "epoch": 94.13640389725421,
      "grad_norm": 0.2411148101091385,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 106280
    },
    {
      "epoch": 94.140832595217,
      "grad_norm": 0.23350247740745544,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 106285
    },
    {
      "epoch": 94.1452612931798,
      "grad_norm": 0.2606596350669861,
      "learning_rate": 1e-05,
      "loss": 0.9471,
      "step": 106290
    },
    {
      "epoch": 94.14968999114261,
      "grad_norm": 0.24810384213924408,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 106295
    },
    {
      "epoch": 94.1541186891054,
      "grad_norm": 0.2267456352710724,
      "learning_rate": 1e-05,
      "loss": 0.9558,
      "step": 106300
    },
    {
      "epoch": 94.1585473870682,
      "grad_norm": 0.25738993287086487,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 106305
    },
    {
      "epoch": 94.162976085031,
      "grad_norm": 0.20813482999801636,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 106310
    },
    {
      "epoch": 94.1674047829938,
      "grad_norm": 0.24512743949890137,
      "learning_rate": 1e-05,
      "loss": 0.9263,
      "step": 106315
    },
    {
      "epoch": 94.1718334809566,
      "grad_norm": 0.2616995871067047,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 106320
    },
    {
      "epoch": 94.1762621789194,
      "grad_norm": 0.23294930160045624,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 106325
    },
    {
      "epoch": 94.1806908768822,
      "grad_norm": 0.21492715179920197,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 106330
    },
    {
      "epoch": 94.18511957484499,
      "grad_norm": 0.2580125629901886,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 106335
    },
    {
      "epoch": 94.1895482728078,
      "grad_norm": 0.24805280566215515,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 106340
    },
    {
      "epoch": 94.19397697077059,
      "grad_norm": 0.2524789571762085,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 106345
    },
    {
      "epoch": 94.19840566873339,
      "grad_norm": 0.22692793607711792,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 106350
    },
    {
      "epoch": 94.2028343666962,
      "grad_norm": 0.2800218462944031,
      "learning_rate": 1e-05,
      "loss": 1.0008,
      "step": 106355
    },
    {
      "epoch": 94.20726306465899,
      "grad_norm": 0.26655757427215576,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 106360
    },
    {
      "epoch": 94.21169176262178,
      "grad_norm": 0.2496543526649475,
      "learning_rate": 1e-05,
      "loss": 0.9298,
      "step": 106365
    },
    {
      "epoch": 94.21612046058459,
      "grad_norm": 0.2739472985267639,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 106370
    },
    {
      "epoch": 94.22054915854739,
      "grad_norm": 0.2369619607925415,
      "learning_rate": 1e-05,
      "loss": 0.9946,
      "step": 106375
    },
    {
      "epoch": 94.22497785651018,
      "grad_norm": 0.2599835693836212,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 106380
    },
    {
      "epoch": 94.22940655447299,
      "grad_norm": 0.270287424325943,
      "learning_rate": 1e-05,
      "loss": 1.0174,
      "step": 106385
    },
    {
      "epoch": 94.23383525243578,
      "grad_norm": 0.24475695192813873,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 106390
    },
    {
      "epoch": 94.23826395039858,
      "grad_norm": 0.2421625852584839,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 106395
    },
    {
      "epoch": 94.24269264836138,
      "grad_norm": 0.2789970338344574,
      "learning_rate": 1e-05,
      "loss": 0.8944,
      "step": 106400
    },
    {
      "epoch": 94.24712134632418,
      "grad_norm": 0.24444790184497833,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 106405
    },
    {
      "epoch": 94.25155004428699,
      "grad_norm": 0.28055405616760254,
      "learning_rate": 1e-05,
      "loss": 0.9747,
      "step": 106410
    },
    {
      "epoch": 94.25597874224978,
      "grad_norm": 0.2829570770263672,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 106415
    },
    {
      "epoch": 94.26040744021257,
      "grad_norm": 0.20914830267429352,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 106420
    },
    {
      "epoch": 94.26483613817538,
      "grad_norm": 0.25046849250793457,
      "learning_rate": 1e-05,
      "loss": 1.0129,
      "step": 106425
    },
    {
      "epoch": 94.26926483613818,
      "grad_norm": 0.25098326802253723,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 106430
    },
    {
      "epoch": 94.27369353410097,
      "grad_norm": 0.23200610280036926,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 106435
    },
    {
      "epoch": 94.27812223206378,
      "grad_norm": 0.24208246171474457,
      "learning_rate": 1e-05,
      "loss": 0.9244,
      "step": 106440
    },
    {
      "epoch": 94.28255093002657,
      "grad_norm": 0.25149649381637573,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 106445
    },
    {
      "epoch": 94.28697962798937,
      "grad_norm": 0.22865618765354156,
      "learning_rate": 1e-05,
      "loss": 0.9148,
      "step": 106450
    },
    {
      "epoch": 94.29140832595218,
      "grad_norm": 0.21656499803066254,
      "learning_rate": 1e-05,
      "loss": 0.985,
      "step": 106455
    },
    {
      "epoch": 94.29583702391497,
      "grad_norm": 0.20786502957344055,
      "learning_rate": 1e-05,
      "loss": 0.9096,
      "step": 106460
    },
    {
      "epoch": 94.30026572187776,
      "grad_norm": 0.23585541546344757,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 106465
    },
    {
      "epoch": 94.30469441984057,
      "grad_norm": 0.24068616330623627,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 106470
    },
    {
      "epoch": 94.30912311780337,
      "grad_norm": 0.2645930349826813,
      "learning_rate": 1e-05,
      "loss": 0.8932,
      "step": 106475
    },
    {
      "epoch": 94.31355181576616,
      "grad_norm": 0.22868798673152924,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 106480
    },
    {
      "epoch": 94.31798051372897,
      "grad_norm": 0.22378195822238922,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 106485
    },
    {
      "epoch": 94.32240921169176,
      "grad_norm": 0.2546638250350952,
      "learning_rate": 1e-05,
      "loss": 0.92,
      "step": 106490
    },
    {
      "epoch": 94.32683790965456,
      "grad_norm": 0.26925382018089294,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 106495
    },
    {
      "epoch": 94.33126660761737,
      "grad_norm": 0.23870424926280975,
      "learning_rate": 1e-05,
      "loss": 0.9454,
      "step": 106500
    },
    {
      "epoch": 94.33569530558016,
      "grad_norm": 0.28356173634529114,
      "learning_rate": 1e-05,
      "loss": 0.9072,
      "step": 106505
    },
    {
      "epoch": 94.34012400354295,
      "grad_norm": 0.21108368039131165,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 106510
    },
    {
      "epoch": 94.34455270150576,
      "grad_norm": 0.2537795305252075,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 106515
    },
    {
      "epoch": 94.34898139946856,
      "grad_norm": 0.25566232204437256,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 106520
    },
    {
      "epoch": 94.35341009743135,
      "grad_norm": 0.24104613065719604,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 106525
    },
    {
      "epoch": 94.35783879539416,
      "grad_norm": 0.2654995322227478,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 106530
    },
    {
      "epoch": 94.36226749335695,
      "grad_norm": 0.23405615985393524,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 106535
    },
    {
      "epoch": 94.36669619131975,
      "grad_norm": 0.25690433382987976,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 106540
    },
    {
      "epoch": 94.37112488928256,
      "grad_norm": 0.2330116629600525,
      "learning_rate": 1e-05,
      "loss": 0.917,
      "step": 106545
    },
    {
      "epoch": 94.37555358724535,
      "grad_norm": 0.2403959035873413,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 106550
    },
    {
      "epoch": 94.37998228520814,
      "grad_norm": 0.24891816079616547,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 106555
    },
    {
      "epoch": 94.38441098317095,
      "grad_norm": 0.27309247851371765,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 106560
    },
    {
      "epoch": 94.38883968113375,
      "grad_norm": 0.23420576751232147,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 106565
    },
    {
      "epoch": 94.39326837909654,
      "grad_norm": 0.22360877692699432,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 106570
    },
    {
      "epoch": 94.39769707705935,
      "grad_norm": 0.21367734670639038,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 106575
    },
    {
      "epoch": 94.40212577502214,
      "grad_norm": 0.2309485524892807,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 106580
    },
    {
      "epoch": 94.40655447298494,
      "grad_norm": 0.2124474197626114,
      "learning_rate": 1e-05,
      "loss": 0.8815,
      "step": 106585
    },
    {
      "epoch": 94.41098317094774,
      "grad_norm": 0.2443016618490219,
      "learning_rate": 1e-05,
      "loss": 0.9478,
      "step": 106590
    },
    {
      "epoch": 94.41541186891054,
      "grad_norm": 0.2841915488243103,
      "learning_rate": 1e-05,
      "loss": 1.0113,
      "step": 106595
    },
    {
      "epoch": 94.41984056687333,
      "grad_norm": 0.23119981586933136,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 106600
    },
    {
      "epoch": 94.42426926483614,
      "grad_norm": 0.2255610227584839,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 106605
    },
    {
      "epoch": 94.42869796279894,
      "grad_norm": 0.25441014766693115,
      "learning_rate": 1e-05,
      "loss": 0.9836,
      "step": 106610
    },
    {
      "epoch": 94.43312666076173,
      "grad_norm": 0.24843120574951172,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 106615
    },
    {
      "epoch": 94.43755535872454,
      "grad_norm": 0.20645496249198914,
      "learning_rate": 1e-05,
      "loss": 0.9121,
      "step": 106620
    },
    {
      "epoch": 94.44198405668733,
      "grad_norm": 0.2380525916814804,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 106625
    },
    {
      "epoch": 94.44641275465013,
      "grad_norm": 0.24944616854190826,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 106630
    },
    {
      "epoch": 94.45084145261293,
      "grad_norm": 0.2314627766609192,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 106635
    },
    {
      "epoch": 94.45527015057573,
      "grad_norm": 0.23358173668384552,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 106640
    },
    {
      "epoch": 94.45969884853854,
      "grad_norm": 0.22349950671195984,
      "learning_rate": 1e-05,
      "loss": 0.919,
      "step": 106645
    },
    {
      "epoch": 94.46412754650133,
      "grad_norm": 0.2273334115743637,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 106650
    },
    {
      "epoch": 94.46855624446412,
      "grad_norm": 0.2579975426197052,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 106655
    },
    {
      "epoch": 94.47298494242693,
      "grad_norm": 0.2125396430492401,
      "learning_rate": 1e-05,
      "loss": 0.9255,
      "step": 106660
    },
    {
      "epoch": 94.47741364038973,
      "grad_norm": 0.24439695477485657,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 106665
    },
    {
      "epoch": 94.48184233835252,
      "grad_norm": 0.22889472544193268,
      "learning_rate": 1e-05,
      "loss": 0.9267,
      "step": 106670
    },
    {
      "epoch": 94.48627103631533,
      "grad_norm": 0.21434274315834045,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 106675
    },
    {
      "epoch": 94.49069973427812,
      "grad_norm": 0.23859034478664398,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 106680
    },
    {
      "epoch": 94.49512843224092,
      "grad_norm": 0.2425820231437683,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 106685
    },
    {
      "epoch": 94.49955713020373,
      "grad_norm": 0.24613337218761444,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 106690
    },
    {
      "epoch": 94.50398582816652,
      "grad_norm": 0.25566110014915466,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 106695
    },
    {
      "epoch": 94.50841452612931,
      "grad_norm": 0.23636366426944733,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 106700
    },
    {
      "epoch": 94.51284322409212,
      "grad_norm": 0.27085795998573303,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 106705
    },
    {
      "epoch": 94.51727192205492,
      "grad_norm": 0.2568426728248596,
      "learning_rate": 1e-05,
      "loss": 0.9003,
      "step": 106710
    },
    {
      "epoch": 94.52170062001771,
      "grad_norm": 0.22434687614440918,
      "learning_rate": 1e-05,
      "loss": 0.9249,
      "step": 106715
    },
    {
      "epoch": 94.52612931798052,
      "grad_norm": 0.2627421021461487,
      "learning_rate": 1e-05,
      "loss": 0.9065,
      "step": 106720
    },
    {
      "epoch": 94.53055801594331,
      "grad_norm": 0.25969594717025757,
      "learning_rate": 1e-05,
      "loss": 0.9905,
      "step": 106725
    },
    {
      "epoch": 94.53498671390611,
      "grad_norm": 0.2693764865398407,
      "learning_rate": 1e-05,
      "loss": 0.9444,
      "step": 106730
    },
    {
      "epoch": 94.53941541186892,
      "grad_norm": 0.24594071507453918,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 106735
    },
    {
      "epoch": 94.54384410983171,
      "grad_norm": 0.2511425018310547,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 106740
    },
    {
      "epoch": 94.5482728077945,
      "grad_norm": 0.23891757428646088,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 106745
    },
    {
      "epoch": 94.55270150575731,
      "grad_norm": 0.23691214621067047,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 106750
    },
    {
      "epoch": 94.5571302037201,
      "grad_norm": 0.21230220794677734,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 106755
    },
    {
      "epoch": 94.5615589016829,
      "grad_norm": 0.20601458847522736,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 106760
    },
    {
      "epoch": 94.56598759964571,
      "grad_norm": 0.228900745511055,
      "learning_rate": 1e-05,
      "loss": 0.9294,
      "step": 106765
    },
    {
      "epoch": 94.5704162976085,
      "grad_norm": 0.2278706282377243,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 106770
    },
    {
      "epoch": 94.5748449955713,
      "grad_norm": 0.21343880891799927,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 106775
    },
    {
      "epoch": 94.5792736935341,
      "grad_norm": 0.26100000739097595,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 106780
    },
    {
      "epoch": 94.5837023914969,
      "grad_norm": 0.2264852076768875,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 106785
    },
    {
      "epoch": 94.5881310894597,
      "grad_norm": 0.27163198590278625,
      "learning_rate": 1e-05,
      "loss": 0.9255,
      "step": 106790
    },
    {
      "epoch": 94.5925597874225,
      "grad_norm": 0.2259998619556427,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 106795
    },
    {
      "epoch": 94.5969884853853,
      "grad_norm": 0.23439352214336395,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 106800
    },
    {
      "epoch": 94.60141718334809,
      "grad_norm": 0.2967948615550995,
      "learning_rate": 1e-05,
      "loss": 0.9522,
      "step": 106805
    },
    {
      "epoch": 94.6058458813109,
      "grad_norm": 0.22962042689323425,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 106810
    },
    {
      "epoch": 94.61027457927369,
      "grad_norm": 0.24918319284915924,
      "learning_rate": 1e-05,
      "loss": 0.9863,
      "step": 106815
    },
    {
      "epoch": 94.61470327723649,
      "grad_norm": 0.2544427216053009,
      "learning_rate": 1e-05,
      "loss": 0.9427,
      "step": 106820
    },
    {
      "epoch": 94.6191319751993,
      "grad_norm": 0.24607641994953156,
      "learning_rate": 1e-05,
      "loss": 0.9621,
      "step": 106825
    },
    {
      "epoch": 94.62356067316209,
      "grad_norm": 0.21463237702846527,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 106830
    },
    {
      "epoch": 94.62798937112488,
      "grad_norm": 0.2559424042701721,
      "learning_rate": 1e-05,
      "loss": 0.9419,
      "step": 106835
    },
    {
      "epoch": 94.63241806908769,
      "grad_norm": 0.24596081674098969,
      "learning_rate": 1e-05,
      "loss": 0.9514,
      "step": 106840
    },
    {
      "epoch": 94.63684676705049,
      "grad_norm": 0.22070500254631042,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 106845
    },
    {
      "epoch": 94.64127546501328,
      "grad_norm": 0.21979346871376038,
      "learning_rate": 1e-05,
      "loss": 0.906,
      "step": 106850
    },
    {
      "epoch": 94.64570416297609,
      "grad_norm": 0.22034062445163727,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 106855
    },
    {
      "epoch": 94.65013286093888,
      "grad_norm": 0.23303696513175964,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 106860
    },
    {
      "epoch": 94.65456155890168,
      "grad_norm": 0.24786536395549774,
      "learning_rate": 1e-05,
      "loss": 0.9215,
      "step": 106865
    },
    {
      "epoch": 94.65899025686448,
      "grad_norm": 0.25682350993156433,
      "learning_rate": 1e-05,
      "loss": 0.975,
      "step": 106870
    },
    {
      "epoch": 94.66341895482728,
      "grad_norm": 0.20970618724822998,
      "learning_rate": 1e-05,
      "loss": 0.9085,
      "step": 106875
    },
    {
      "epoch": 94.66784765279007,
      "grad_norm": 0.24832530319690704,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 106880
    },
    {
      "epoch": 94.67227635075288,
      "grad_norm": 0.2522828280925751,
      "learning_rate": 1e-05,
      "loss": 0.9189,
      "step": 106885
    },
    {
      "epoch": 94.67670504871568,
      "grad_norm": 0.2097586691379547,
      "learning_rate": 1e-05,
      "loss": 0.9591,
      "step": 106890
    },
    {
      "epoch": 94.68113374667848,
      "grad_norm": 0.2314344048500061,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 106895
    },
    {
      "epoch": 94.68556244464128,
      "grad_norm": 0.2404661476612091,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 106900
    },
    {
      "epoch": 94.68999114260407,
      "grad_norm": 0.21006403863430023,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 106905
    },
    {
      "epoch": 94.69441984056688,
      "grad_norm": 0.2337801307439804,
      "learning_rate": 1e-05,
      "loss": 0.9207,
      "step": 106910
    },
    {
      "epoch": 94.69884853852967,
      "grad_norm": 0.24181610345840454,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 106915
    },
    {
      "epoch": 94.70327723649247,
      "grad_norm": 0.24205538630485535,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 106920
    },
    {
      "epoch": 94.70770593445528,
      "grad_norm": 0.23540495336055756,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 106925
    },
    {
      "epoch": 94.71213463241807,
      "grad_norm": 0.23933860659599304,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 106930
    },
    {
      "epoch": 94.71656333038086,
      "grad_norm": 0.21771498024463654,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 106935
    },
    {
      "epoch": 94.72099202834367,
      "grad_norm": 0.2311660200357437,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 106940
    },
    {
      "epoch": 94.72542072630647,
      "grad_norm": 0.257680743932724,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 106945
    },
    {
      "epoch": 94.72984942426926,
      "grad_norm": 0.19089046120643616,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 106950
    },
    {
      "epoch": 94.73427812223207,
      "grad_norm": 0.23095518350601196,
      "learning_rate": 1e-05,
      "loss": 0.9248,
      "step": 106955
    },
    {
      "epoch": 94.73870682019486,
      "grad_norm": 0.21291907131671906,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 106960
    },
    {
      "epoch": 94.74313551815766,
      "grad_norm": 0.28700071573257446,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 106965
    },
    {
      "epoch": 94.74756421612047,
      "grad_norm": 0.2774430215358734,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 106970
    },
    {
      "epoch": 94.75199291408326,
      "grad_norm": 0.22784988582134247,
      "learning_rate": 1e-05,
      "loss": 0.9443,
      "step": 106975
    },
    {
      "epoch": 94.75642161204605,
      "grad_norm": 0.25632473826408386,
      "learning_rate": 1e-05,
      "loss": 0.922,
      "step": 106980
    },
    {
      "epoch": 94.76085031000886,
      "grad_norm": 0.2301356941461563,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 106985
    },
    {
      "epoch": 94.76527900797166,
      "grad_norm": 0.24693381786346436,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 106990
    },
    {
      "epoch": 94.76970770593445,
      "grad_norm": 0.24424348771572113,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 106995
    },
    {
      "epoch": 94.77413640389726,
      "grad_norm": 0.27526140213012695,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 107000
    },
    {
      "epoch": 94.77856510186005,
      "grad_norm": 0.25445759296417236,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 107005
    },
    {
      "epoch": 94.78299379982285,
      "grad_norm": 0.25785014033317566,
      "learning_rate": 1e-05,
      "loss": 0.9154,
      "step": 107010
    },
    {
      "epoch": 94.78742249778566,
      "grad_norm": 0.20772771537303925,
      "learning_rate": 1e-05,
      "loss": 0.9856,
      "step": 107015
    },
    {
      "epoch": 94.79185119574845,
      "grad_norm": 0.24703574180603027,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 107020
    },
    {
      "epoch": 94.79627989371124,
      "grad_norm": 0.26185283064842224,
      "learning_rate": 1e-05,
      "loss": 1.0057,
      "step": 107025
    },
    {
      "epoch": 94.80070859167405,
      "grad_norm": 0.24025486409664154,
      "learning_rate": 1e-05,
      "loss": 0.9589,
      "step": 107030
    },
    {
      "epoch": 94.80513728963685,
      "grad_norm": 0.23892425000667572,
      "learning_rate": 1e-05,
      "loss": 0.9934,
      "step": 107035
    },
    {
      "epoch": 94.80956598759964,
      "grad_norm": 0.2291329950094223,
      "learning_rate": 1e-05,
      "loss": 0.923,
      "step": 107040
    },
    {
      "epoch": 94.81399468556245,
      "grad_norm": 0.23601289093494415,
      "learning_rate": 1e-05,
      "loss": 0.9333,
      "step": 107045
    },
    {
      "epoch": 94.81842338352524,
      "grad_norm": 0.27586621046066284,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 107050
    },
    {
      "epoch": 94.82285208148804,
      "grad_norm": 0.294270783662796,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 107055
    },
    {
      "epoch": 94.82728077945085,
      "grad_norm": 0.2527262568473816,
      "learning_rate": 1e-05,
      "loss": 0.98,
      "step": 107060
    },
    {
      "epoch": 94.83170947741364,
      "grad_norm": 0.26242318749427795,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 107065
    },
    {
      "epoch": 94.83613817537643,
      "grad_norm": 0.22645775973796844,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 107070
    },
    {
      "epoch": 94.84056687333924,
      "grad_norm": 0.19579118490219116,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 107075
    },
    {
      "epoch": 94.84499557130204,
      "grad_norm": 0.23198047280311584,
      "learning_rate": 1e-05,
      "loss": 0.9229,
      "step": 107080
    },
    {
      "epoch": 94.84942426926483,
      "grad_norm": 0.2218148112297058,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 107085
    },
    {
      "epoch": 94.85385296722764,
      "grad_norm": 0.2478971779346466,
      "learning_rate": 1e-05,
      "loss": 0.9188,
      "step": 107090
    },
    {
      "epoch": 94.85828166519043,
      "grad_norm": 0.20794397592544556,
      "learning_rate": 1e-05,
      "loss": 0.9862,
      "step": 107095
    },
    {
      "epoch": 94.86271036315323,
      "grad_norm": 0.21862350404262543,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 107100
    },
    {
      "epoch": 94.86713906111603,
      "grad_norm": 0.20600490272045135,
      "learning_rate": 1e-05,
      "loss": 1.0267,
      "step": 107105
    },
    {
      "epoch": 94.87156775907883,
      "grad_norm": 0.2510509788990021,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 107110
    },
    {
      "epoch": 94.87599645704162,
      "grad_norm": 0.23539741337299347,
      "learning_rate": 1e-05,
      "loss": 0.9295,
      "step": 107115
    },
    {
      "epoch": 94.88042515500443,
      "grad_norm": 0.21304330229759216,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 107120
    },
    {
      "epoch": 94.88485385296723,
      "grad_norm": 0.24177055060863495,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 107125
    },
    {
      "epoch": 94.88928255093003,
      "grad_norm": 0.2605949640274048,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 107130
    },
    {
      "epoch": 94.89371124889283,
      "grad_norm": 0.21455734968185425,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 107135
    },
    {
      "epoch": 94.89813994685562,
      "grad_norm": 0.24048122763633728,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 107140
    },
    {
      "epoch": 94.90256864481843,
      "grad_norm": 0.2430018186569214,
      "learning_rate": 1e-05,
      "loss": 1.0224,
      "step": 107145
    },
    {
      "epoch": 94.90699734278122,
      "grad_norm": 0.26060742139816284,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 107150
    },
    {
      "epoch": 94.91142604074402,
      "grad_norm": 0.2363378405570984,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 107155
    },
    {
      "epoch": 94.91585473870683,
      "grad_norm": 0.2722887396812439,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 107160
    },
    {
      "epoch": 94.92028343666962,
      "grad_norm": 0.2661963999271393,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 107165
    },
    {
      "epoch": 94.92471213463241,
      "grad_norm": 0.2966618537902832,
      "learning_rate": 1e-05,
      "loss": 0.9396,
      "step": 107170
    },
    {
      "epoch": 94.92914083259522,
      "grad_norm": 0.25535956025123596,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 107175
    },
    {
      "epoch": 94.93356953055802,
      "grad_norm": 0.2727159559726715,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 107180
    },
    {
      "epoch": 94.93799822852081,
      "grad_norm": 0.2500159442424774,
      "learning_rate": 1e-05,
      "loss": 0.8967,
      "step": 107185
    },
    {
      "epoch": 94.94242692648362,
      "grad_norm": 0.24730458855628967,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 107190
    },
    {
      "epoch": 94.94685562444641,
      "grad_norm": 0.27276644110679626,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 107195
    },
    {
      "epoch": 94.95128432240921,
      "grad_norm": 0.24198442697525024,
      "learning_rate": 1e-05,
      "loss": 0.9289,
      "step": 107200
    },
    {
      "epoch": 94.95571302037202,
      "grad_norm": 0.25210312008857727,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 107205
    },
    {
      "epoch": 94.96014171833481,
      "grad_norm": 0.21360740065574646,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 107210
    },
    {
      "epoch": 94.9645704162976,
      "grad_norm": 0.2179337441921234,
      "learning_rate": 1e-05,
      "loss": 0.9223,
      "step": 107215
    },
    {
      "epoch": 94.96899911426041,
      "grad_norm": 0.27676576375961304,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 107220
    },
    {
      "epoch": 94.9734278122232,
      "grad_norm": 0.2532136142253876,
      "learning_rate": 1e-05,
      "loss": 0.9522,
      "step": 107225
    },
    {
      "epoch": 94.977856510186,
      "grad_norm": 0.24171559512615204,
      "learning_rate": 1e-05,
      "loss": 0.9324,
      "step": 107230
    },
    {
      "epoch": 94.98228520814881,
      "grad_norm": 0.23634073138237,
      "learning_rate": 1e-05,
      "loss": 0.9639,
      "step": 107235
    },
    {
      "epoch": 94.9867139061116,
      "grad_norm": 0.2309700846672058,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 107240
    },
    {
      "epoch": 94.9911426040744,
      "grad_norm": 0.25484704971313477,
      "learning_rate": 1e-05,
      "loss": 0.983,
      "step": 107245
    },
    {
      "epoch": 94.9955713020372,
      "grad_norm": 0.2663779556751251,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 107250
    },
    {
      "epoch": 95.0,
      "grad_norm": 0.25913217663764954,
      "learning_rate": 1e-05,
      "loss": 0.8762,
      "step": 107255
    },
    {
      "epoch": 95.0044286979628,
      "grad_norm": 0.23530063033103943,
      "learning_rate": 1e-05,
      "loss": 0.9112,
      "step": 107260
    },
    {
      "epoch": 95.0088573959256,
      "grad_norm": 0.3243725895881653,
      "learning_rate": 1e-05,
      "loss": 0.9339,
      "step": 107265
    },
    {
      "epoch": 95.0132860938884,
      "grad_norm": 0.26743388175964355,
      "learning_rate": 1e-05,
      "loss": 0.9193,
      "step": 107270
    },
    {
      "epoch": 95.01771479185119,
      "grad_norm": 0.24291455745697021,
      "learning_rate": 1e-05,
      "loss": 0.8946,
      "step": 107275
    },
    {
      "epoch": 95.022143489814,
      "grad_norm": 0.27711138129234314,
      "learning_rate": 1e-05,
      "loss": 1.0305,
      "step": 107280
    },
    {
      "epoch": 95.0265721877768,
      "grad_norm": 0.2239644080400467,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 107285
    },
    {
      "epoch": 95.03100088573959,
      "grad_norm": 0.21524082124233246,
      "learning_rate": 1e-05,
      "loss": 0.9087,
      "step": 107290
    },
    {
      "epoch": 95.0354295837024,
      "grad_norm": 0.22020091116428375,
      "learning_rate": 1e-05,
      "loss": 0.9123,
      "step": 107295
    },
    {
      "epoch": 95.03985828166519,
      "grad_norm": 0.28928342461586,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 107300
    },
    {
      "epoch": 95.04428697962798,
      "grad_norm": 0.2070544809103012,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 107305
    },
    {
      "epoch": 95.04871567759079,
      "grad_norm": 0.24191558361053467,
      "learning_rate": 1e-05,
      "loss": 0.9635,
      "step": 107310
    },
    {
      "epoch": 95.05314437555359,
      "grad_norm": 0.23150593042373657,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 107315
    },
    {
      "epoch": 95.05757307351638,
      "grad_norm": 0.2508734166622162,
      "learning_rate": 1e-05,
      "loss": 1.004,
      "step": 107320
    },
    {
      "epoch": 95.06200177147919,
      "grad_norm": 0.2241988629102707,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 107325
    },
    {
      "epoch": 95.06643046944198,
      "grad_norm": 0.24449783563613892,
      "learning_rate": 1e-05,
      "loss": 0.901,
      "step": 107330
    },
    {
      "epoch": 95.07085916740478,
      "grad_norm": 0.20451109111309052,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 107335
    },
    {
      "epoch": 95.07528786536759,
      "grad_norm": 0.21989353001117706,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 107340
    },
    {
      "epoch": 95.07971656333038,
      "grad_norm": 0.22003479301929474,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 107345
    },
    {
      "epoch": 95.08414526129317,
      "grad_norm": 0.2669506072998047,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 107350
    },
    {
      "epoch": 95.08857395925598,
      "grad_norm": 0.2667923867702484,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 107355
    },
    {
      "epoch": 95.09300265721878,
      "grad_norm": 0.2533407509326935,
      "learning_rate": 1e-05,
      "loss": 0.9138,
      "step": 107360
    },
    {
      "epoch": 95.09743135518157,
      "grad_norm": 0.24007335305213928,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 107365
    },
    {
      "epoch": 95.10186005314438,
      "grad_norm": 0.2502859830856323,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 107370
    },
    {
      "epoch": 95.10628875110717,
      "grad_norm": 0.22984546422958374,
      "learning_rate": 1e-05,
      "loss": 0.9317,
      "step": 107375
    },
    {
      "epoch": 95.11071744906998,
      "grad_norm": 0.21428455412387848,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 107380
    },
    {
      "epoch": 95.11514614703277,
      "grad_norm": 0.27482813596725464,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 107385
    },
    {
      "epoch": 95.11957484499557,
      "grad_norm": 0.25785624980926514,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 107390
    },
    {
      "epoch": 95.12400354295838,
      "grad_norm": 0.33070603013038635,
      "learning_rate": 1e-05,
      "loss": 0.9371,
      "step": 107395
    },
    {
      "epoch": 95.12843224092117,
      "grad_norm": 0.23416925966739655,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 107400
    },
    {
      "epoch": 95.13286093888397,
      "grad_norm": 0.24412092566490173,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 107405
    },
    {
      "epoch": 95.13728963684677,
      "grad_norm": 0.23255768418312073,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 107410
    },
    {
      "epoch": 95.14171833480957,
      "grad_norm": 0.21100811660289764,
      "learning_rate": 1e-05,
      "loss": 0.9389,
      "step": 107415
    },
    {
      "epoch": 95.14614703277236,
      "grad_norm": 0.22685614228248596,
      "learning_rate": 1e-05,
      "loss": 0.9866,
      "step": 107420
    },
    {
      "epoch": 95.15057573073517,
      "grad_norm": 0.23177024722099304,
      "learning_rate": 1e-05,
      "loss": 0.9223,
      "step": 107425
    },
    {
      "epoch": 95.15500442869796,
      "grad_norm": 0.23917986452579498,
      "learning_rate": 1e-05,
      "loss": 0.9876,
      "step": 107430
    },
    {
      "epoch": 95.15943312666076,
      "grad_norm": 0.24278783798217773,
      "learning_rate": 1e-05,
      "loss": 0.9396,
      "step": 107435
    },
    {
      "epoch": 95.16386182462357,
      "grad_norm": 0.21225441992282867,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 107440
    },
    {
      "epoch": 95.16829052258636,
      "grad_norm": 0.2684895098209381,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 107445
    },
    {
      "epoch": 95.17271922054915,
      "grad_norm": 0.21584945917129517,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 107450
    },
    {
      "epoch": 95.17714791851196,
      "grad_norm": 0.25267452001571655,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 107455
    },
    {
      "epoch": 95.18157661647476,
      "grad_norm": 0.23381614685058594,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 107460
    },
    {
      "epoch": 95.18600531443755,
      "grad_norm": 0.2509293854236603,
      "learning_rate": 1e-05,
      "loss": 0.9354,
      "step": 107465
    },
    {
      "epoch": 95.19043401240036,
      "grad_norm": 0.288047194480896,
      "learning_rate": 1e-05,
      "loss": 0.965,
      "step": 107470
    },
    {
      "epoch": 95.19486271036315,
      "grad_norm": 0.3397163152694702,
      "learning_rate": 1e-05,
      "loss": 0.9213,
      "step": 107475
    },
    {
      "epoch": 95.19929140832595,
      "grad_norm": 0.297993004322052,
      "learning_rate": 1e-05,
      "loss": 1.0076,
      "step": 107480
    },
    {
      "epoch": 95.20372010628876,
      "grad_norm": 0.2647121548652649,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 107485
    },
    {
      "epoch": 95.20814880425155,
      "grad_norm": 0.29293352365493774,
      "learning_rate": 1e-05,
      "loss": 0.9325,
      "step": 107490
    },
    {
      "epoch": 95.21257750221434,
      "grad_norm": 0.2722911238670349,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 107495
    },
    {
      "epoch": 95.21700620017715,
      "grad_norm": 0.256132036447525,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 107500
    },
    {
      "epoch": 95.22143489813995,
      "grad_norm": 0.2437218278646469,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 107505
    },
    {
      "epoch": 95.22586359610274,
      "grad_norm": 0.2889432907104492,
      "learning_rate": 1e-05,
      "loss": 0.9672,
      "step": 107510
    },
    {
      "epoch": 95.23029229406555,
      "grad_norm": 0.26617172360420227,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 107515
    },
    {
      "epoch": 95.23472099202834,
      "grad_norm": 0.2752954065799713,
      "learning_rate": 1e-05,
      "loss": 0.9123,
      "step": 107520
    },
    {
      "epoch": 95.23914968999114,
      "grad_norm": 0.2673304080963135,
      "learning_rate": 1e-05,
      "loss": 0.9757,
      "step": 107525
    },
    {
      "epoch": 95.24357838795395,
      "grad_norm": 0.2583892047405243,
      "learning_rate": 1e-05,
      "loss": 0.9431,
      "step": 107530
    },
    {
      "epoch": 95.24800708591674,
      "grad_norm": 0.21743465960025787,
      "learning_rate": 1e-05,
      "loss": 1.0182,
      "step": 107535
    },
    {
      "epoch": 95.25243578387953,
      "grad_norm": 0.23750510811805725,
      "learning_rate": 1e-05,
      "loss": 0.8856,
      "step": 107540
    },
    {
      "epoch": 95.25686448184234,
      "grad_norm": 0.22443650662899017,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 107545
    },
    {
      "epoch": 95.26129317980514,
      "grad_norm": 0.22520148754119873,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 107550
    },
    {
      "epoch": 95.26572187776793,
      "grad_norm": 0.23910659551620483,
      "learning_rate": 1e-05,
      "loss": 1.031,
      "step": 107555
    },
    {
      "epoch": 95.27015057573074,
      "grad_norm": 0.21996942162513733,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 107560
    },
    {
      "epoch": 95.27457927369353,
      "grad_norm": 0.22658292949199677,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 107565
    },
    {
      "epoch": 95.27900797165633,
      "grad_norm": 0.2718266546726227,
      "learning_rate": 1e-05,
      "loss": 0.9423,
      "step": 107570
    },
    {
      "epoch": 95.28343666961914,
      "grad_norm": 0.21327579021453857,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 107575
    },
    {
      "epoch": 95.28786536758193,
      "grad_norm": 0.22494789958000183,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 107580
    },
    {
      "epoch": 95.29229406554472,
      "grad_norm": 0.20625858008861542,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 107585
    },
    {
      "epoch": 95.29672276350753,
      "grad_norm": 0.19615961611270905,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 107590
    },
    {
      "epoch": 95.30115146147033,
      "grad_norm": 0.2181224524974823,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 107595
    },
    {
      "epoch": 95.30558015943312,
      "grad_norm": 0.26227763295173645,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 107600
    },
    {
      "epoch": 95.31000885739593,
      "grad_norm": 0.24930843710899353,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 107605
    },
    {
      "epoch": 95.31443755535872,
      "grad_norm": 0.2209462672472,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 107610
    },
    {
      "epoch": 95.31886625332152,
      "grad_norm": 0.2423831820487976,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 107615
    },
    {
      "epoch": 95.32329495128432,
      "grad_norm": 0.27176910638809204,
      "learning_rate": 1e-05,
      "loss": 0.9171,
      "step": 107620
    },
    {
      "epoch": 95.32772364924712,
      "grad_norm": 0.227650985121727,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 107625
    },
    {
      "epoch": 95.33215234720993,
      "grad_norm": 0.26221513748168945,
      "learning_rate": 1e-05,
      "loss": 0.9259,
      "step": 107630
    },
    {
      "epoch": 95.33658104517272,
      "grad_norm": 0.26837992668151855,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 107635
    },
    {
      "epoch": 95.34100974313552,
      "grad_norm": 0.2633305490016937,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 107640
    },
    {
      "epoch": 95.34543844109832,
      "grad_norm": 0.2688346803188324,
      "learning_rate": 1e-05,
      "loss": 0.9791,
      "step": 107645
    },
    {
      "epoch": 95.34986713906112,
      "grad_norm": 0.2908957004547119,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 107650
    },
    {
      "epoch": 95.35429583702391,
      "grad_norm": 0.23492327332496643,
      "learning_rate": 1e-05,
      "loss": 0.9536,
      "step": 107655
    },
    {
      "epoch": 95.35872453498672,
      "grad_norm": 0.2329270839691162,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 107660
    },
    {
      "epoch": 95.36315323294951,
      "grad_norm": 0.2525000274181366,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 107665
    },
    {
      "epoch": 95.36758193091231,
      "grad_norm": 0.2484455555677414,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 107670
    },
    {
      "epoch": 95.37201062887512,
      "grad_norm": 0.2335076630115509,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 107675
    },
    {
      "epoch": 95.37643932683791,
      "grad_norm": 0.2261621057987213,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 107680
    },
    {
      "epoch": 95.3808680248007,
      "grad_norm": 0.24132534861564636,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 107685
    },
    {
      "epoch": 95.38529672276351,
      "grad_norm": 0.2227826863527298,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 107690
    },
    {
      "epoch": 95.38972542072631,
      "grad_norm": 0.2079281508922577,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 107695
    },
    {
      "epoch": 95.3941541186891,
      "grad_norm": 0.25957751274108887,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 107700
    },
    {
      "epoch": 95.39858281665191,
      "grad_norm": 0.21455520391464233,
      "learning_rate": 1e-05,
      "loss": 0.9516,
      "step": 107705
    },
    {
      "epoch": 95.4030115146147,
      "grad_norm": 0.21068847179412842,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 107710
    },
    {
      "epoch": 95.4074402125775,
      "grad_norm": 0.2353878915309906,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 107715
    },
    {
      "epoch": 95.4118689105403,
      "grad_norm": 0.2761279046535492,
      "learning_rate": 1e-05,
      "loss": 0.9186,
      "step": 107720
    },
    {
      "epoch": 95.4162976085031,
      "grad_norm": 0.27247875928878784,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 107725
    },
    {
      "epoch": 95.4207263064659,
      "grad_norm": 0.2290761023759842,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 107730
    },
    {
      "epoch": 95.4251550044287,
      "grad_norm": 0.24237966537475586,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 107735
    },
    {
      "epoch": 95.4295837023915,
      "grad_norm": 0.23053774237632751,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 107740
    },
    {
      "epoch": 95.43401240035429,
      "grad_norm": 0.23504970967769623,
      "learning_rate": 1e-05,
      "loss": 0.8922,
      "step": 107745
    },
    {
      "epoch": 95.4384410983171,
      "grad_norm": 0.22553934156894684,
      "learning_rate": 1e-05,
      "loss": 0.9138,
      "step": 107750
    },
    {
      "epoch": 95.4428697962799,
      "grad_norm": 0.2617259621620178,
      "learning_rate": 1e-05,
      "loss": 1.0025,
      "step": 107755
    },
    {
      "epoch": 95.44729849424269,
      "grad_norm": 0.2620776295661926,
      "learning_rate": 1e-05,
      "loss": 0.919,
      "step": 107760
    },
    {
      "epoch": 95.4517271922055,
      "grad_norm": 0.27812597155570984,
      "learning_rate": 1e-05,
      "loss": 0.9638,
      "step": 107765
    },
    {
      "epoch": 95.45615589016829,
      "grad_norm": 0.27106449007987976,
      "learning_rate": 1e-05,
      "loss": 0.9288,
      "step": 107770
    },
    {
      "epoch": 95.46058458813108,
      "grad_norm": 0.2544442415237427,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 107775
    },
    {
      "epoch": 95.46501328609389,
      "grad_norm": 0.2651505768299103,
      "learning_rate": 1e-05,
      "loss": 0.9219,
      "step": 107780
    },
    {
      "epoch": 95.46944198405669,
      "grad_norm": 0.27600473165512085,
      "learning_rate": 1e-05,
      "loss": 0.9928,
      "step": 107785
    },
    {
      "epoch": 95.47387068201948,
      "grad_norm": 0.24292562901973724,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 107790
    },
    {
      "epoch": 95.47829937998229,
      "grad_norm": 0.24555571377277374,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 107795
    },
    {
      "epoch": 95.48272807794508,
      "grad_norm": 0.2511807978153229,
      "learning_rate": 1e-05,
      "loss": 0.8852,
      "step": 107800
    },
    {
      "epoch": 95.48715677590788,
      "grad_norm": 0.20451007783412933,
      "learning_rate": 1e-05,
      "loss": 0.9561,
      "step": 107805
    },
    {
      "epoch": 95.49158547387069,
      "grad_norm": 0.20194494724273682,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 107810
    },
    {
      "epoch": 95.49601417183348,
      "grad_norm": 0.24525058269500732,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 107815
    },
    {
      "epoch": 95.50044286979627,
      "grad_norm": 0.2273806780576706,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 107820
    },
    {
      "epoch": 95.50487156775908,
      "grad_norm": 0.3020380437374115,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 107825
    },
    {
      "epoch": 95.50930026572188,
      "grad_norm": 0.2678582966327667,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 107830
    },
    {
      "epoch": 95.51372896368467,
      "grad_norm": 0.2440209686756134,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 107835
    },
    {
      "epoch": 95.51815766164748,
      "grad_norm": 0.22443720698356628,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 107840
    },
    {
      "epoch": 95.52258635961027,
      "grad_norm": 0.21010196208953857,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 107845
    },
    {
      "epoch": 95.52701505757307,
      "grad_norm": 0.23201937973499298,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 107850
    },
    {
      "epoch": 95.53144375553588,
      "grad_norm": 0.2587655782699585,
      "learning_rate": 1e-05,
      "loss": 0.9845,
      "step": 107855
    },
    {
      "epoch": 95.53587245349867,
      "grad_norm": 0.23483608663082123,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 107860
    },
    {
      "epoch": 95.54030115146146,
      "grad_norm": 0.2619555592536926,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 107865
    },
    {
      "epoch": 95.54472984942427,
      "grad_norm": 0.21851742267608643,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 107870
    },
    {
      "epoch": 95.54915854738707,
      "grad_norm": 0.23866872489452362,
      "learning_rate": 1e-05,
      "loss": 0.9107,
      "step": 107875
    },
    {
      "epoch": 95.55358724534987,
      "grad_norm": 0.2349606454372406,
      "learning_rate": 1e-05,
      "loss": 0.9908,
      "step": 107880
    },
    {
      "epoch": 95.55801594331267,
      "grad_norm": 0.24345038831233978,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 107885
    },
    {
      "epoch": 95.56244464127546,
      "grad_norm": 0.23698939383029938,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 107890
    },
    {
      "epoch": 95.56687333923827,
      "grad_norm": 0.2121095508337021,
      "learning_rate": 1e-05,
      "loss": 0.9571,
      "step": 107895
    },
    {
      "epoch": 95.57130203720106,
      "grad_norm": 0.2416294664144516,
      "learning_rate": 1e-05,
      "loss": 0.9191,
      "step": 107900
    },
    {
      "epoch": 95.57573073516386,
      "grad_norm": 0.23848575353622437,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 107905
    },
    {
      "epoch": 95.58015943312667,
      "grad_norm": 0.24208682775497437,
      "learning_rate": 1e-05,
      "loss": 0.9788,
      "step": 107910
    },
    {
      "epoch": 95.58458813108946,
      "grad_norm": 0.23229193687438965,
      "learning_rate": 1e-05,
      "loss": 0.9583,
      "step": 107915
    },
    {
      "epoch": 95.58901682905226,
      "grad_norm": 0.21544359624385834,
      "learning_rate": 1e-05,
      "loss": 0.9483,
      "step": 107920
    },
    {
      "epoch": 95.59344552701506,
      "grad_norm": 0.2173500657081604,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 107925
    },
    {
      "epoch": 95.59787422497786,
      "grad_norm": 0.24962668120861053,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 107930
    },
    {
      "epoch": 95.60230292294065,
      "grad_norm": 0.23697854578495026,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 107935
    },
    {
      "epoch": 95.60673162090346,
      "grad_norm": 0.2456091195344925,
      "learning_rate": 1e-05,
      "loss": 0.999,
      "step": 107940
    },
    {
      "epoch": 95.61116031886625,
      "grad_norm": 0.2261052429676056,
      "learning_rate": 1e-05,
      "loss": 1.0016,
      "step": 107945
    },
    {
      "epoch": 95.61558901682905,
      "grad_norm": 0.21750393509864807,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 107950
    },
    {
      "epoch": 95.62001771479186,
      "grad_norm": 0.2327343225479126,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 107955
    },
    {
      "epoch": 95.62444641275465,
      "grad_norm": 0.21743425726890564,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 107960
    },
    {
      "epoch": 95.62887511071744,
      "grad_norm": 0.21500687301158905,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 107965
    },
    {
      "epoch": 95.63330380868025,
      "grad_norm": 0.2827708125114441,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 107970
    },
    {
      "epoch": 95.63773250664305,
      "grad_norm": 0.22060155868530273,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 107975
    },
    {
      "epoch": 95.64216120460584,
      "grad_norm": 0.23911650478839874,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 107980
    },
    {
      "epoch": 95.64658990256865,
      "grad_norm": 0.25105273723602295,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 107985
    },
    {
      "epoch": 95.65101860053144,
      "grad_norm": 0.24852430820465088,
      "learning_rate": 1e-05,
      "loss": 0.9274,
      "step": 107990
    },
    {
      "epoch": 95.65544729849424,
      "grad_norm": 0.26660463213920593,
      "learning_rate": 1e-05,
      "loss": 0.9086,
      "step": 107995
    },
    {
      "epoch": 95.65987599645705,
      "grad_norm": 0.27217134833335876,
      "learning_rate": 1e-05,
      "loss": 0.9179,
      "step": 108000
    },
    {
      "epoch": 95.66430469441984,
      "grad_norm": 0.257432222366333,
      "learning_rate": 1e-05,
      "loss": 0.9696,
      "step": 108005
    },
    {
      "epoch": 95.66873339238263,
      "grad_norm": 0.2769724428653717,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 108010
    },
    {
      "epoch": 95.67316209034544,
      "grad_norm": 0.24498045444488525,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 108015
    },
    {
      "epoch": 95.67759078830824,
      "grad_norm": 0.21995790302753448,
      "learning_rate": 1e-05,
      "loss": 0.9239,
      "step": 108020
    },
    {
      "epoch": 95.68201948627103,
      "grad_norm": 0.2099144607782364,
      "learning_rate": 1e-05,
      "loss": 0.9099,
      "step": 108025
    },
    {
      "epoch": 95.68644818423384,
      "grad_norm": 0.22911645472049713,
      "learning_rate": 1e-05,
      "loss": 0.9346,
      "step": 108030
    },
    {
      "epoch": 95.69087688219663,
      "grad_norm": 0.2082708179950714,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 108035
    },
    {
      "epoch": 95.69530558015943,
      "grad_norm": 0.22762998938560486,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 108040
    },
    {
      "epoch": 95.69973427812224,
      "grad_norm": 0.2905156910419464,
      "learning_rate": 1e-05,
      "loss": 0.9076,
      "step": 108045
    },
    {
      "epoch": 95.70416297608503,
      "grad_norm": 0.2250833511352539,
      "learning_rate": 1e-05,
      "loss": 0.8849,
      "step": 108050
    },
    {
      "epoch": 95.70859167404782,
      "grad_norm": 0.23311945796012878,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 108055
    },
    {
      "epoch": 95.71302037201063,
      "grad_norm": 0.24047549068927765,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 108060
    },
    {
      "epoch": 95.71744906997343,
      "grad_norm": 0.25582608580589294,
      "learning_rate": 1e-05,
      "loss": 0.9352,
      "step": 108065
    },
    {
      "epoch": 95.72187776793622,
      "grad_norm": 0.2521994411945343,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 108070
    },
    {
      "epoch": 95.72630646589903,
      "grad_norm": 0.21439537405967712,
      "learning_rate": 1e-05,
      "loss": 0.9691,
      "step": 108075
    },
    {
      "epoch": 95.73073516386182,
      "grad_norm": 0.2501664161682129,
      "learning_rate": 1e-05,
      "loss": 0.968,
      "step": 108080
    },
    {
      "epoch": 95.73516386182462,
      "grad_norm": 0.19152836501598358,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 108085
    },
    {
      "epoch": 95.73959255978743,
      "grad_norm": 0.22520753741264343,
      "learning_rate": 1e-05,
      "loss": 0.9308,
      "step": 108090
    },
    {
      "epoch": 95.74402125775022,
      "grad_norm": 0.23537737131118774,
      "learning_rate": 1e-05,
      "loss": 0.9874,
      "step": 108095
    },
    {
      "epoch": 95.74844995571301,
      "grad_norm": 0.22484339773654938,
      "learning_rate": 1e-05,
      "loss": 0.9251,
      "step": 108100
    },
    {
      "epoch": 95.75287865367582,
      "grad_norm": 0.23144637048244476,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 108105
    },
    {
      "epoch": 95.75730735163862,
      "grad_norm": 0.25670066475868225,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 108110
    },
    {
      "epoch": 95.76173604960141,
      "grad_norm": 0.24359601736068726,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 108115
    },
    {
      "epoch": 95.76616474756422,
      "grad_norm": 0.2958217263221741,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 108120
    },
    {
      "epoch": 95.77059344552701,
      "grad_norm": 0.2801280915737152,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 108125
    },
    {
      "epoch": 95.77502214348982,
      "grad_norm": 0.34061455726623535,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 108130
    },
    {
      "epoch": 95.77945084145261,
      "grad_norm": 0.2528132200241089,
      "learning_rate": 1e-05,
      "loss": 1.0364,
      "step": 108135
    },
    {
      "epoch": 95.78387953941541,
      "grad_norm": 0.2196105420589447,
      "learning_rate": 1e-05,
      "loss": 0.944,
      "step": 108140
    },
    {
      "epoch": 95.78830823737822,
      "grad_norm": 0.23865923285484314,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 108145
    },
    {
      "epoch": 95.79273693534101,
      "grad_norm": 0.24523451924324036,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 108150
    },
    {
      "epoch": 95.7971656333038,
      "grad_norm": 0.2544254958629608,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 108155
    },
    {
      "epoch": 95.80159433126661,
      "grad_norm": 0.22589778900146484,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 108160
    },
    {
      "epoch": 95.80602302922941,
      "grad_norm": 0.21877838671207428,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 108165
    },
    {
      "epoch": 95.8104517271922,
      "grad_norm": 0.2545568645000458,
      "learning_rate": 1e-05,
      "loss": 0.9712,
      "step": 108170
    },
    {
      "epoch": 95.81488042515501,
      "grad_norm": 0.22206686437129974,
      "learning_rate": 1e-05,
      "loss": 0.9766,
      "step": 108175
    },
    {
      "epoch": 95.8193091231178,
      "grad_norm": 0.22488637268543243,
      "learning_rate": 1e-05,
      "loss": 0.8899,
      "step": 108180
    },
    {
      "epoch": 95.8237378210806,
      "grad_norm": 0.2672382891178131,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 108185
    },
    {
      "epoch": 95.8281665190434,
      "grad_norm": 0.23874084651470184,
      "learning_rate": 1e-05,
      "loss": 0.989,
      "step": 108190
    },
    {
      "epoch": 95.8325952170062,
      "grad_norm": 0.27910977602005005,
      "learning_rate": 1e-05,
      "loss": 0.9121,
      "step": 108195
    },
    {
      "epoch": 95.837023914969,
      "grad_norm": 0.21587279438972473,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 108200
    },
    {
      "epoch": 95.8414526129318,
      "grad_norm": 0.25613000988960266,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 108205
    },
    {
      "epoch": 95.8458813108946,
      "grad_norm": 0.2395486682653427,
      "learning_rate": 1e-05,
      "loss": 0.9275,
      "step": 108210
    },
    {
      "epoch": 95.85031000885739,
      "grad_norm": 0.22896182537078857,
      "learning_rate": 1e-05,
      "loss": 0.9826,
      "step": 108215
    },
    {
      "epoch": 95.8547387068202,
      "grad_norm": 0.1961967796087265,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 108220
    },
    {
      "epoch": 95.859167404783,
      "grad_norm": 0.2177097350358963,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 108225
    },
    {
      "epoch": 95.86359610274579,
      "grad_norm": 0.2459237426519394,
      "learning_rate": 1e-05,
      "loss": 0.9112,
      "step": 108230
    },
    {
      "epoch": 95.8680248007086,
      "grad_norm": 0.21894758939743042,
      "learning_rate": 1e-05,
      "loss": 0.9189,
      "step": 108235
    },
    {
      "epoch": 95.87245349867139,
      "grad_norm": 0.2109079658985138,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 108240
    },
    {
      "epoch": 95.87688219663418,
      "grad_norm": 0.22265484929084778,
      "learning_rate": 1e-05,
      "loss": 1.0252,
      "step": 108245
    },
    {
      "epoch": 95.881310894597,
      "grad_norm": 0.21388401091098785,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 108250
    },
    {
      "epoch": 95.88573959255979,
      "grad_norm": 0.2093137949705124,
      "learning_rate": 1e-05,
      "loss": 1.0204,
      "step": 108255
    },
    {
      "epoch": 95.89016829052258,
      "grad_norm": 0.2581399977207184,
      "learning_rate": 1e-05,
      "loss": 0.9222,
      "step": 108260
    },
    {
      "epoch": 95.89459698848539,
      "grad_norm": 0.2453896701335907,
      "learning_rate": 1e-05,
      "loss": 0.9378,
      "step": 108265
    },
    {
      "epoch": 95.89902568644818,
      "grad_norm": 0.23279815912246704,
      "learning_rate": 1e-05,
      "loss": 1.0243,
      "step": 108270
    },
    {
      "epoch": 95.90345438441098,
      "grad_norm": 0.24917802214622498,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 108275
    },
    {
      "epoch": 95.90788308237379,
      "grad_norm": 0.23274897038936615,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 108280
    },
    {
      "epoch": 95.91231178033658,
      "grad_norm": 0.23741154372692108,
      "learning_rate": 1e-05,
      "loss": 0.9256,
      "step": 108285
    },
    {
      "epoch": 95.91674047829937,
      "grad_norm": 0.26576706767082214,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 108290
    },
    {
      "epoch": 95.92116917626218,
      "grad_norm": 0.25688600540161133,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 108295
    },
    {
      "epoch": 95.92559787422498,
      "grad_norm": 0.22089515626430511,
      "learning_rate": 1e-05,
      "loss": 0.9804,
      "step": 108300
    },
    {
      "epoch": 95.93002657218777,
      "grad_norm": 0.22952593863010406,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 108305
    },
    {
      "epoch": 95.93445527015058,
      "grad_norm": 0.25194287300109863,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 108310
    },
    {
      "epoch": 95.93888396811337,
      "grad_norm": 0.24947448074817657,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 108315
    },
    {
      "epoch": 95.94331266607617,
      "grad_norm": 0.23501215875148773,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 108320
    },
    {
      "epoch": 95.94774136403898,
      "grad_norm": 0.23731659352779388,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 108325
    },
    {
      "epoch": 95.95217006200177,
      "grad_norm": 0.24125874042510986,
      "learning_rate": 1e-05,
      "loss": 0.9554,
      "step": 108330
    },
    {
      "epoch": 95.95659875996456,
      "grad_norm": 0.2147240787744522,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 108335
    },
    {
      "epoch": 95.96102745792737,
      "grad_norm": 0.26002317667007446,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 108340
    },
    {
      "epoch": 95.96545615589017,
      "grad_norm": 0.26153072714805603,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 108345
    },
    {
      "epoch": 95.96988485385296,
      "grad_norm": 0.22740699350833893,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 108350
    },
    {
      "epoch": 95.97431355181577,
      "grad_norm": 0.3171957731246948,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 108355
    },
    {
      "epoch": 95.97874224977856,
      "grad_norm": 0.2792864143848419,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 108360
    },
    {
      "epoch": 95.98317094774137,
      "grad_norm": 0.21778006851673126,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 108365
    },
    {
      "epoch": 95.98759964570417,
      "grad_norm": 0.2647404968738556,
      "learning_rate": 1e-05,
      "loss": 0.932,
      "step": 108370
    },
    {
      "epoch": 95.99202834366696,
      "grad_norm": 0.23767496645450592,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 108375
    },
    {
      "epoch": 95.99645704162977,
      "grad_norm": 0.2408391386270523,
      "learning_rate": 1e-05,
      "loss": 0.9021,
      "step": 108380
    },
    {
      "epoch": 96.00088573959256,
      "grad_norm": 0.2284712940454483,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 108385
    },
    {
      "epoch": 96.00531443755536,
      "grad_norm": 0.2424333393573761,
      "learning_rate": 1e-05,
      "loss": 0.9535,
      "step": 108390
    },
    {
      "epoch": 96.00974313551816,
      "grad_norm": 0.24135784804821014,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 108395
    },
    {
      "epoch": 96.01417183348096,
      "grad_norm": 0.21310147643089294,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 108400
    },
    {
      "epoch": 96.01860053144375,
      "grad_norm": 0.27662280201911926,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 108405
    },
    {
      "epoch": 96.02302922940656,
      "grad_norm": 0.255436509847641,
      "learning_rate": 1e-05,
      "loss": 0.9737,
      "step": 108410
    },
    {
      "epoch": 96.02745792736935,
      "grad_norm": 0.25066521763801575,
      "learning_rate": 1e-05,
      "loss": 0.9239,
      "step": 108415
    },
    {
      "epoch": 96.03188662533215,
      "grad_norm": 0.26973816752433777,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 108420
    },
    {
      "epoch": 96.03631532329496,
      "grad_norm": 0.25600868463516235,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 108425
    },
    {
      "epoch": 96.04074402125775,
      "grad_norm": 0.2159801870584488,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 108430
    },
    {
      "epoch": 96.04517271922055,
      "grad_norm": 0.239048033952713,
      "learning_rate": 1e-05,
      "loss": 0.9366,
      "step": 108435
    },
    {
      "epoch": 96.04960141718335,
      "grad_norm": 0.2514578700065613,
      "learning_rate": 1e-05,
      "loss": 0.9154,
      "step": 108440
    },
    {
      "epoch": 96.05403011514615,
      "grad_norm": 0.23884740471839905,
      "learning_rate": 1e-05,
      "loss": 0.9859,
      "step": 108445
    },
    {
      "epoch": 96.05845881310894,
      "grad_norm": 0.23057234287261963,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 108450
    },
    {
      "epoch": 96.06288751107175,
      "grad_norm": 0.2376694679260254,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 108455
    },
    {
      "epoch": 96.06731620903454,
      "grad_norm": 0.20688025653362274,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 108460
    },
    {
      "epoch": 96.07174490699734,
      "grad_norm": 0.24544085562229156,
      "learning_rate": 1e-05,
      "loss": 0.9267,
      "step": 108465
    },
    {
      "epoch": 96.07617360496015,
      "grad_norm": 0.24091307818889618,
      "learning_rate": 1e-05,
      "loss": 0.9258,
      "step": 108470
    },
    {
      "epoch": 96.08060230292294,
      "grad_norm": 0.28245028853416443,
      "learning_rate": 1e-05,
      "loss": 0.9342,
      "step": 108475
    },
    {
      "epoch": 96.08503100088573,
      "grad_norm": 0.3136764168739319,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 108480
    },
    {
      "epoch": 96.08945969884854,
      "grad_norm": 0.24032048881053925,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 108485
    },
    {
      "epoch": 96.09388839681134,
      "grad_norm": 0.23953594267368317,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 108490
    },
    {
      "epoch": 96.09831709477413,
      "grad_norm": 0.22534404695034027,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 108495
    },
    {
      "epoch": 96.10274579273694,
      "grad_norm": 0.27071499824523926,
      "learning_rate": 1e-05,
      "loss": 0.9944,
      "step": 108500
    },
    {
      "epoch": 96.10717449069973,
      "grad_norm": 0.2184804379940033,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 108505
    },
    {
      "epoch": 96.11160318866253,
      "grad_norm": 0.2290293276309967,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 108510
    },
    {
      "epoch": 96.11603188662534,
      "grad_norm": 0.31282541155815125,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 108515
    },
    {
      "epoch": 96.12046058458813,
      "grad_norm": 0.354139506816864,
      "learning_rate": 1e-05,
      "loss": 0.9563,
      "step": 108520
    },
    {
      "epoch": 96.12488928255092,
      "grad_norm": 0.2839699983596802,
      "learning_rate": 1e-05,
      "loss": 0.9133,
      "step": 108525
    },
    {
      "epoch": 96.12931798051373,
      "grad_norm": 0.2797127366065979,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 108530
    },
    {
      "epoch": 96.13374667847653,
      "grad_norm": 0.22523237764835358,
      "learning_rate": 1e-05,
      "loss": 0.9985,
      "step": 108535
    },
    {
      "epoch": 96.13817537643932,
      "grad_norm": 0.22221854329109192,
      "learning_rate": 1e-05,
      "loss": 0.9882,
      "step": 108540
    },
    {
      "epoch": 96.14260407440213,
      "grad_norm": 0.24611735343933105,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 108545
    },
    {
      "epoch": 96.14703277236492,
      "grad_norm": 0.2309277355670929,
      "learning_rate": 1e-05,
      "loss": 0.9606,
      "step": 108550
    },
    {
      "epoch": 96.15146147032772,
      "grad_norm": 0.22810816764831543,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 108555
    },
    {
      "epoch": 96.15589016829053,
      "grad_norm": 0.30302348732948303,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 108560
    },
    {
      "epoch": 96.16031886625332,
      "grad_norm": 0.24275289475917816,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 108565
    },
    {
      "epoch": 96.16474756421611,
      "grad_norm": 0.23894831538200378,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 108570
    },
    {
      "epoch": 96.16917626217892,
      "grad_norm": 0.23714321851730347,
      "learning_rate": 1e-05,
      "loss": 0.9745,
      "step": 108575
    },
    {
      "epoch": 96.17360496014172,
      "grad_norm": 0.23956306278705597,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 108580
    },
    {
      "epoch": 96.17803365810451,
      "grad_norm": 0.29898393154144287,
      "learning_rate": 1e-05,
      "loss": 0.9297,
      "step": 108585
    },
    {
      "epoch": 96.18246235606732,
      "grad_norm": 0.28933021426200867,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 108590
    },
    {
      "epoch": 96.18689105403011,
      "grad_norm": 0.2901092767715454,
      "learning_rate": 1e-05,
      "loss": 0.9557,
      "step": 108595
    },
    {
      "epoch": 96.1913197519929,
      "grad_norm": 0.21013809740543365,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 108600
    },
    {
      "epoch": 96.19574844995572,
      "grad_norm": 0.2515837550163269,
      "learning_rate": 1e-05,
      "loss": 0.9376,
      "step": 108605
    },
    {
      "epoch": 96.20017714791851,
      "grad_norm": 0.26505985856056213,
      "learning_rate": 1e-05,
      "loss": 0.9318,
      "step": 108610
    },
    {
      "epoch": 96.20460584588132,
      "grad_norm": 0.2377777397632599,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 108615
    },
    {
      "epoch": 96.20903454384411,
      "grad_norm": 0.25668391585350037,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 108620
    },
    {
      "epoch": 96.2134632418069,
      "grad_norm": 0.2132730931043625,
      "learning_rate": 1e-05,
      "loss": 0.939,
      "step": 108625
    },
    {
      "epoch": 96.21789193976971,
      "grad_norm": 0.22132623195648193,
      "learning_rate": 1e-05,
      "loss": 0.933,
      "step": 108630
    },
    {
      "epoch": 96.22232063773251,
      "grad_norm": 0.24276478588581085,
      "learning_rate": 1e-05,
      "loss": 0.9574,
      "step": 108635
    },
    {
      "epoch": 96.2267493356953,
      "grad_norm": 0.21171914041042328,
      "learning_rate": 1e-05,
      "loss": 0.949,
      "step": 108640
    },
    {
      "epoch": 96.23117803365811,
      "grad_norm": 0.22050949931144714,
      "learning_rate": 1e-05,
      "loss": 0.9307,
      "step": 108645
    },
    {
      "epoch": 96.2356067316209,
      "grad_norm": 0.27296534180641174,
      "learning_rate": 1e-05,
      "loss": 0.9602,
      "step": 108650
    },
    {
      "epoch": 96.2400354295837,
      "grad_norm": 0.22463339567184448,
      "learning_rate": 1e-05,
      "loss": 0.9883,
      "step": 108655
    },
    {
      "epoch": 96.24446412754651,
      "grad_norm": 0.25902196764945984,
      "learning_rate": 1e-05,
      "loss": 0.9347,
      "step": 108660
    },
    {
      "epoch": 96.2488928255093,
      "grad_norm": 0.26889732480049133,
      "learning_rate": 1e-05,
      "loss": 0.9234,
      "step": 108665
    },
    {
      "epoch": 96.2533215234721,
      "grad_norm": 0.24156594276428223,
      "learning_rate": 1e-05,
      "loss": 0.9146,
      "step": 108670
    },
    {
      "epoch": 96.2577502214349,
      "grad_norm": 0.2529046833515167,
      "learning_rate": 1e-05,
      "loss": 0.9293,
      "step": 108675
    },
    {
      "epoch": 96.2621789193977,
      "grad_norm": 0.2482517957687378,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 108680
    },
    {
      "epoch": 96.26660761736049,
      "grad_norm": 0.22952386736869812,
      "learning_rate": 1e-05,
      "loss": 0.9489,
      "step": 108685
    },
    {
      "epoch": 96.2710363153233,
      "grad_norm": 0.20501549541950226,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 108690
    },
    {
      "epoch": 96.2754650132861,
      "grad_norm": 0.26863911747932434,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 108695
    },
    {
      "epoch": 96.27989371124889,
      "grad_norm": 0.2132173627614975,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 108700
    },
    {
      "epoch": 96.2843224092117,
      "grad_norm": 0.26510581374168396,
      "learning_rate": 1e-05,
      "loss": 1.0121,
      "step": 108705
    },
    {
      "epoch": 96.28875110717449,
      "grad_norm": 0.2395889163017273,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 108710
    },
    {
      "epoch": 96.29317980513729,
      "grad_norm": 0.2426411509513855,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 108715
    },
    {
      "epoch": 96.2976085031001,
      "grad_norm": 0.2448132038116455,
      "learning_rate": 1e-05,
      "loss": 1.0326,
      "step": 108720
    },
    {
      "epoch": 96.30203720106289,
      "grad_norm": 0.2125806361436844,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 108725
    },
    {
      "epoch": 96.30646589902568,
      "grad_norm": 0.23368892073631287,
      "learning_rate": 1e-05,
      "loss": 1.0088,
      "step": 108730
    },
    {
      "epoch": 96.31089459698849,
      "grad_norm": 0.2467987835407257,
      "learning_rate": 1e-05,
      "loss": 0.9842,
      "step": 108735
    },
    {
      "epoch": 96.31532329495128,
      "grad_norm": 0.22907590866088867,
      "learning_rate": 1e-05,
      "loss": 0.9523,
      "step": 108740
    },
    {
      "epoch": 96.31975199291408,
      "grad_norm": 0.2262767106294632,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 108745
    },
    {
      "epoch": 96.32418069087689,
      "grad_norm": 0.24563902616500854,
      "learning_rate": 1e-05,
      "loss": 0.9201,
      "step": 108750
    },
    {
      "epoch": 96.32860938883968,
      "grad_norm": 0.22593539953231812,
      "learning_rate": 1e-05,
      "loss": 0.9203,
      "step": 108755
    },
    {
      "epoch": 96.33303808680247,
      "grad_norm": 0.2171163111925125,
      "learning_rate": 1e-05,
      "loss": 0.9384,
      "step": 108760
    },
    {
      "epoch": 96.33746678476528,
      "grad_norm": 0.22219586372375488,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 108765
    },
    {
      "epoch": 96.34189548272808,
      "grad_norm": 0.21997058391571045,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 108770
    },
    {
      "epoch": 96.34632418069087,
      "grad_norm": 0.2191559076309204,
      "learning_rate": 1e-05,
      "loss": 0.995,
      "step": 108775
    },
    {
      "epoch": 96.35075287865368,
      "grad_norm": 0.3007301092147827,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 108780
    },
    {
      "epoch": 96.35518157661647,
      "grad_norm": 0.24937856197357178,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 108785
    },
    {
      "epoch": 96.35961027457927,
      "grad_norm": 0.21745456755161285,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 108790
    },
    {
      "epoch": 96.36403897254208,
      "grad_norm": 0.21041302382946014,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 108795
    },
    {
      "epoch": 96.36846767050487,
      "grad_norm": 0.2228080779314041,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 108800
    },
    {
      "epoch": 96.37289636846766,
      "grad_norm": 0.27315106987953186,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 108805
    },
    {
      "epoch": 96.37732506643047,
      "grad_norm": 0.26474303007125854,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 108810
    },
    {
      "epoch": 96.38175376439327,
      "grad_norm": 0.264626681804657,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 108815
    },
    {
      "epoch": 96.38618246235606,
      "grad_norm": 0.24691453576087952,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 108820
    },
    {
      "epoch": 96.39061116031887,
      "grad_norm": 0.23327571153640747,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 108825
    },
    {
      "epoch": 96.39503985828166,
      "grad_norm": 0.2663676142692566,
      "learning_rate": 1e-05,
      "loss": 0.9942,
      "step": 108830
    },
    {
      "epoch": 96.39946855624446,
      "grad_norm": 0.2329336255788803,
      "learning_rate": 1e-05,
      "loss": 0.9587,
      "step": 108835
    },
    {
      "epoch": 96.40389725420727,
      "grad_norm": 0.23028676211833954,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 108840
    },
    {
      "epoch": 96.40832595217006,
      "grad_norm": 0.23600134253501892,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 108845
    },
    {
      "epoch": 96.41275465013285,
      "grad_norm": 0.27145883440971375,
      "learning_rate": 1e-05,
      "loss": 0.9625,
      "step": 108850
    },
    {
      "epoch": 96.41718334809566,
      "grad_norm": 0.25177595019340515,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 108855
    },
    {
      "epoch": 96.42161204605846,
      "grad_norm": 0.2219705581665039,
      "learning_rate": 1e-05,
      "loss": 0.9831,
      "step": 108860
    },
    {
      "epoch": 96.42604074402126,
      "grad_norm": 0.24424809217453003,
      "learning_rate": 1e-05,
      "loss": 0.9252,
      "step": 108865
    },
    {
      "epoch": 96.43046944198406,
      "grad_norm": 0.26454147696495056,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 108870
    },
    {
      "epoch": 96.43489813994685,
      "grad_norm": 0.21250733733177185,
      "learning_rate": 1e-05,
      "loss": 0.9567,
      "step": 108875
    },
    {
      "epoch": 96.43932683790966,
      "grad_norm": 0.221794992685318,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 108880
    },
    {
      "epoch": 96.44375553587246,
      "grad_norm": 0.2635557949542999,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 108885
    },
    {
      "epoch": 96.44818423383525,
      "grad_norm": 0.38913199305534363,
      "learning_rate": 1e-05,
      "loss": 0.879,
      "step": 108890
    },
    {
      "epoch": 96.45261293179806,
      "grad_norm": 0.2503507435321808,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 108895
    },
    {
      "epoch": 96.45704162976085,
      "grad_norm": 0.2502439022064209,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 108900
    },
    {
      "epoch": 96.46147032772365,
      "grad_norm": 0.22168058156967163,
      "learning_rate": 1e-05,
      "loss": 1.0167,
      "step": 108905
    },
    {
      "epoch": 96.46589902568645,
      "grad_norm": 0.22931894659996033,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 108910
    },
    {
      "epoch": 96.47032772364925,
      "grad_norm": 0.26465457677841187,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 108915
    },
    {
      "epoch": 96.47475642161204,
      "grad_norm": 0.22159187495708466,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 108920
    },
    {
      "epoch": 96.47918511957485,
      "grad_norm": 0.22609472274780273,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 108925
    },
    {
      "epoch": 96.48361381753764,
      "grad_norm": 0.23655027151107788,
      "learning_rate": 1e-05,
      "loss": 0.9686,
      "step": 108930
    },
    {
      "epoch": 96.48804251550044,
      "grad_norm": 0.2643718421459198,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 108935
    },
    {
      "epoch": 96.49247121346325,
      "grad_norm": 0.2730221748352051,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 108940
    },
    {
      "epoch": 96.49689991142604,
      "grad_norm": 0.26589077711105347,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 108945
    },
    {
      "epoch": 96.50132860938884,
      "grad_norm": 0.22781260311603546,
      "learning_rate": 1e-05,
      "loss": 0.9272,
      "step": 108950
    },
    {
      "epoch": 96.50575730735164,
      "grad_norm": 0.25082796812057495,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 108955
    },
    {
      "epoch": 96.51018600531444,
      "grad_norm": 0.22763484716415405,
      "learning_rate": 1e-05,
      "loss": 0.9219,
      "step": 108960
    },
    {
      "epoch": 96.51461470327723,
      "grad_norm": 0.2302742600440979,
      "learning_rate": 1e-05,
      "loss": 0.9141,
      "step": 108965
    },
    {
      "epoch": 96.51904340124004,
      "grad_norm": 0.23829132318496704,
      "learning_rate": 1e-05,
      "loss": 0.9347,
      "step": 108970
    },
    {
      "epoch": 96.52347209920283,
      "grad_norm": 0.23597604036331177,
      "learning_rate": 1e-05,
      "loss": 0.931,
      "step": 108975
    },
    {
      "epoch": 96.52790079716563,
      "grad_norm": 0.3194771409034729,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 108980
    },
    {
      "epoch": 96.53232949512844,
      "grad_norm": 0.21260182559490204,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 108985
    },
    {
      "epoch": 96.53675819309123,
      "grad_norm": 0.28577062487602234,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 108990
    },
    {
      "epoch": 96.54118689105402,
      "grad_norm": 0.26052534580230713,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 108995
    },
    {
      "epoch": 96.54561558901683,
      "grad_norm": 0.2302294224500656,
      "learning_rate": 1e-05,
      "loss": 0.9305,
      "step": 109000
    },
    {
      "epoch": 96.55004428697963,
      "grad_norm": 0.21933409571647644,
      "learning_rate": 1e-05,
      "loss": 0.8796,
      "step": 109005
    },
    {
      "epoch": 96.55447298494242,
      "grad_norm": 0.26347604393959045,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 109010
    },
    {
      "epoch": 96.55890168290523,
      "grad_norm": 0.2808287739753723,
      "learning_rate": 1e-05,
      "loss": 0.9496,
      "step": 109015
    },
    {
      "epoch": 96.56333038086802,
      "grad_norm": 0.22767126560211182,
      "learning_rate": 1e-05,
      "loss": 1.0236,
      "step": 109020
    },
    {
      "epoch": 96.56775907883082,
      "grad_norm": 0.2757602632045746,
      "learning_rate": 1e-05,
      "loss": 1.036,
      "step": 109025
    },
    {
      "epoch": 96.57218777679363,
      "grad_norm": 0.21776150166988373,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 109030
    },
    {
      "epoch": 96.57661647475642,
      "grad_norm": 0.22101013362407684,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 109035
    },
    {
      "epoch": 96.58104517271921,
      "grad_norm": 0.2152671217918396,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 109040
    },
    {
      "epoch": 96.58547387068202,
      "grad_norm": 0.22448362410068512,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 109045
    },
    {
      "epoch": 96.58990256864482,
      "grad_norm": 0.21453110873699188,
      "learning_rate": 1e-05,
      "loss": 0.9205,
      "step": 109050
    },
    {
      "epoch": 96.59433126660761,
      "grad_norm": 0.24533586204051971,
      "learning_rate": 1e-05,
      "loss": 0.9349,
      "step": 109055
    },
    {
      "epoch": 96.59875996457042,
      "grad_norm": 0.23309175670146942,
      "learning_rate": 1e-05,
      "loss": 0.9885,
      "step": 109060
    },
    {
      "epoch": 96.60318866253321,
      "grad_norm": 0.22769691050052643,
      "learning_rate": 1e-05,
      "loss": 0.988,
      "step": 109065
    },
    {
      "epoch": 96.60761736049601,
      "grad_norm": 0.21319161355495453,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 109070
    },
    {
      "epoch": 96.61204605845882,
      "grad_norm": 0.2137531042098999,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 109075
    },
    {
      "epoch": 96.61647475642161,
      "grad_norm": 0.2274094820022583,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 109080
    },
    {
      "epoch": 96.6209034543844,
      "grad_norm": 0.2536349296569824,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 109085
    },
    {
      "epoch": 96.62533215234721,
      "grad_norm": 0.25103116035461426,
      "learning_rate": 1e-05,
      "loss": 0.9781,
      "step": 109090
    },
    {
      "epoch": 96.62976085031,
      "grad_norm": 0.22480247914791107,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 109095
    },
    {
      "epoch": 96.63418954827281,
      "grad_norm": 0.22571085393428802,
      "learning_rate": 1e-05,
      "loss": 0.926,
      "step": 109100
    },
    {
      "epoch": 96.63861824623561,
      "grad_norm": 0.23392663896083832,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 109105
    },
    {
      "epoch": 96.6430469441984,
      "grad_norm": 0.21833255887031555,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 109110
    },
    {
      "epoch": 96.64747564216121,
      "grad_norm": 0.2714911103248596,
      "learning_rate": 1e-05,
      "loss": 0.8766,
      "step": 109115
    },
    {
      "epoch": 96.651904340124,
      "grad_norm": 0.24225836992263794,
      "learning_rate": 1e-05,
      "loss": 0.9294,
      "step": 109120
    },
    {
      "epoch": 96.6563330380868,
      "grad_norm": 0.2602677047252655,
      "learning_rate": 1e-05,
      "loss": 0.9301,
      "step": 109125
    },
    {
      "epoch": 96.66076173604961,
      "grad_norm": 0.2385934293270111,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 109130
    },
    {
      "epoch": 96.6651904340124,
      "grad_norm": 0.23226875066757202,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 109135
    },
    {
      "epoch": 96.6696191319752,
      "grad_norm": 0.2980876564979553,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 109140
    },
    {
      "epoch": 96.674047829938,
      "grad_norm": 0.23006302118301392,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 109145
    },
    {
      "epoch": 96.6784765279008,
      "grad_norm": 0.2630012333393097,
      "learning_rate": 1e-05,
      "loss": 0.9436,
      "step": 109150
    },
    {
      "epoch": 96.68290522586359,
      "grad_norm": 0.22954444587230682,
      "learning_rate": 1e-05,
      "loss": 0.9099,
      "step": 109155
    },
    {
      "epoch": 96.6873339238264,
      "grad_norm": 0.22115427255630493,
      "learning_rate": 1e-05,
      "loss": 0.9148,
      "step": 109160
    },
    {
      "epoch": 96.6917626217892,
      "grad_norm": 0.22107937932014465,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 109165
    },
    {
      "epoch": 96.69619131975199,
      "grad_norm": 0.2343861609697342,
      "learning_rate": 1e-05,
      "loss": 0.8955,
      "step": 109170
    },
    {
      "epoch": 96.7006200177148,
      "grad_norm": 0.24237355589866638,
      "learning_rate": 1e-05,
      "loss": 0.9139,
      "step": 109175
    },
    {
      "epoch": 96.70504871567759,
      "grad_norm": 0.2518700957298279,
      "learning_rate": 1e-05,
      "loss": 0.9076,
      "step": 109180
    },
    {
      "epoch": 96.70947741364039,
      "grad_norm": 0.24206562340259552,
      "learning_rate": 1e-05,
      "loss": 0.976,
      "step": 109185
    },
    {
      "epoch": 96.7139061116032,
      "grad_norm": 0.233285591006279,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 109190
    },
    {
      "epoch": 96.71833480956599,
      "grad_norm": 0.22143955528736115,
      "learning_rate": 1e-05,
      "loss": 0.9229,
      "step": 109195
    },
    {
      "epoch": 96.72276350752878,
      "grad_norm": 0.22123315930366516,
      "learning_rate": 1e-05,
      "loss": 0.9175,
      "step": 109200
    },
    {
      "epoch": 96.72719220549159,
      "grad_norm": 0.2531883716583252,
      "learning_rate": 1e-05,
      "loss": 0.9481,
      "step": 109205
    },
    {
      "epoch": 96.73162090345438,
      "grad_norm": 0.23167569935321808,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 109210
    },
    {
      "epoch": 96.73604960141718,
      "grad_norm": 0.19870179891586304,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 109215
    },
    {
      "epoch": 96.74047829937999,
      "grad_norm": 0.2643187642097473,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 109220
    },
    {
      "epoch": 96.74490699734278,
      "grad_norm": 0.24785612523555756,
      "learning_rate": 1e-05,
      "loss": 0.9291,
      "step": 109225
    },
    {
      "epoch": 96.74933569530558,
      "grad_norm": 0.22520042955875397,
      "learning_rate": 1e-05,
      "loss": 0.9955,
      "step": 109230
    },
    {
      "epoch": 96.75376439326838,
      "grad_norm": 0.22258657217025757,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 109235
    },
    {
      "epoch": 96.75819309123118,
      "grad_norm": 0.22151513397693634,
      "learning_rate": 1e-05,
      "loss": 0.9701,
      "step": 109240
    },
    {
      "epoch": 96.76262178919397,
      "grad_norm": 0.24933257699012756,
      "learning_rate": 1e-05,
      "loss": 0.9597,
      "step": 109245
    },
    {
      "epoch": 96.76705048715678,
      "grad_norm": 0.23289336264133453,
      "learning_rate": 1e-05,
      "loss": 1.0009,
      "step": 109250
    },
    {
      "epoch": 96.77147918511957,
      "grad_norm": 0.2526431083679199,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 109255
    },
    {
      "epoch": 96.77590788308237,
      "grad_norm": 0.23179958760738373,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 109260
    },
    {
      "epoch": 96.78033658104518,
      "grad_norm": 0.2722504436969757,
      "learning_rate": 1e-05,
      "loss": 0.9425,
      "step": 109265
    },
    {
      "epoch": 96.78476527900797,
      "grad_norm": 0.20657408237457275,
      "learning_rate": 1e-05,
      "loss": 0.8949,
      "step": 109270
    },
    {
      "epoch": 96.78919397697076,
      "grad_norm": 0.23214073479175568,
      "learning_rate": 1e-05,
      "loss": 0.9263,
      "step": 109275
    },
    {
      "epoch": 96.79362267493357,
      "grad_norm": 0.2209925353527069,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 109280
    },
    {
      "epoch": 96.79805137289637,
      "grad_norm": 0.28586846590042114,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 109285
    },
    {
      "epoch": 96.80248007085916,
      "grad_norm": 0.2213120013475418,
      "learning_rate": 1e-05,
      "loss": 1.018,
      "step": 109290
    },
    {
      "epoch": 96.80690876882197,
      "grad_norm": 0.22843529284000397,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 109295
    },
    {
      "epoch": 96.81133746678476,
      "grad_norm": 0.22046613693237305,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 109300
    },
    {
      "epoch": 96.81576616474756,
      "grad_norm": 0.23868194222450256,
      "learning_rate": 1e-05,
      "loss": 0.9902,
      "step": 109305
    },
    {
      "epoch": 96.82019486271037,
      "grad_norm": 0.22923843562602997,
      "learning_rate": 1e-05,
      "loss": 1.0222,
      "step": 109310
    },
    {
      "epoch": 96.82462356067316,
      "grad_norm": 0.23581412434577942,
      "learning_rate": 1e-05,
      "loss": 0.9077,
      "step": 109315
    },
    {
      "epoch": 96.82905225863595,
      "grad_norm": 0.2506400942802429,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 109320
    },
    {
      "epoch": 96.83348095659876,
      "grad_norm": 0.22032801806926727,
      "learning_rate": 1e-05,
      "loss": 0.9326,
      "step": 109325
    },
    {
      "epoch": 96.83790965456156,
      "grad_norm": 0.25650304555892944,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 109330
    },
    {
      "epoch": 96.84233835252435,
      "grad_norm": 0.2222817987203598,
      "learning_rate": 1e-05,
      "loss": 1.0027,
      "step": 109335
    },
    {
      "epoch": 96.84676705048716,
      "grad_norm": 0.20385071635246277,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 109340
    },
    {
      "epoch": 96.85119574844995,
      "grad_norm": 0.2320050150156021,
      "learning_rate": 1e-05,
      "loss": 0.9614,
      "step": 109345
    },
    {
      "epoch": 96.85562444641276,
      "grad_norm": 0.24511975049972534,
      "learning_rate": 1e-05,
      "loss": 0.9126,
      "step": 109350
    },
    {
      "epoch": 96.86005314437556,
      "grad_norm": 0.22876961529254913,
      "learning_rate": 1e-05,
      "loss": 0.9137,
      "step": 109355
    },
    {
      "epoch": 96.86448184233835,
      "grad_norm": 0.22456419467926025,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 109360
    },
    {
      "epoch": 96.86891054030116,
      "grad_norm": 0.22563810646533966,
      "learning_rate": 1e-05,
      "loss": 0.9646,
      "step": 109365
    },
    {
      "epoch": 96.87333923826395,
      "grad_norm": 0.21834054589271545,
      "learning_rate": 1e-05,
      "loss": 0.9681,
      "step": 109370
    },
    {
      "epoch": 96.87776793622675,
      "grad_norm": 0.24215298891067505,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 109375
    },
    {
      "epoch": 96.88219663418955,
      "grad_norm": 0.20790977776050568,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 109380
    },
    {
      "epoch": 96.88662533215235,
      "grad_norm": 0.22327399253845215,
      "learning_rate": 1e-05,
      "loss": 1.0133,
      "step": 109385
    },
    {
      "epoch": 96.89105403011514,
      "grad_norm": 0.23436813056468964,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 109390
    },
    {
      "epoch": 96.89548272807795,
      "grad_norm": 0.23443818092346191,
      "learning_rate": 1e-05,
      "loss": 0.9248,
      "step": 109395
    },
    {
      "epoch": 96.89991142604075,
      "grad_norm": 0.22904512286186218,
      "learning_rate": 1e-05,
      "loss": 0.9247,
      "step": 109400
    },
    {
      "epoch": 96.90434012400354,
      "grad_norm": 0.21700811386108398,
      "learning_rate": 1e-05,
      "loss": 0.947,
      "step": 109405
    },
    {
      "epoch": 96.90876882196635,
      "grad_norm": 0.2270621657371521,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 109410
    },
    {
      "epoch": 96.91319751992914,
      "grad_norm": 0.21223436295986176,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 109415
    },
    {
      "epoch": 96.91762621789194,
      "grad_norm": 0.250649094581604,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 109420
    },
    {
      "epoch": 96.92205491585474,
      "grad_norm": 0.2699147164821625,
      "learning_rate": 1e-05,
      "loss": 0.9001,
      "step": 109425
    },
    {
      "epoch": 96.92648361381754,
      "grad_norm": 0.23613402247428894,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 109430
    },
    {
      "epoch": 96.93091231178033,
      "grad_norm": 0.23313196003437042,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 109435
    },
    {
      "epoch": 96.93534100974314,
      "grad_norm": 0.2752964496612549,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 109440
    },
    {
      "epoch": 96.93976970770593,
      "grad_norm": 0.2277020961046219,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 109445
    },
    {
      "epoch": 96.94419840566873,
      "grad_norm": 0.20685577392578125,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 109450
    },
    {
      "epoch": 96.94862710363154,
      "grad_norm": 0.24853739142417908,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 109455
    },
    {
      "epoch": 96.95305580159433,
      "grad_norm": 0.2623380720615387,
      "learning_rate": 1e-05,
      "loss": 0.9246,
      "step": 109460
    },
    {
      "epoch": 96.95748449955713,
      "grad_norm": 0.2784663736820221,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 109465
    },
    {
      "epoch": 96.96191319751993,
      "grad_norm": 0.23758426308631897,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 109470
    },
    {
      "epoch": 96.96634189548273,
      "grad_norm": 0.21574918925762177,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 109475
    },
    {
      "epoch": 96.97077059344552,
      "grad_norm": 0.2676204442977905,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 109480
    },
    {
      "epoch": 96.97519929140833,
      "grad_norm": 0.28029686212539673,
      "learning_rate": 1e-05,
      "loss": 0.9219,
      "step": 109485
    },
    {
      "epoch": 96.97962798937112,
      "grad_norm": 0.2208881825208664,
      "learning_rate": 1e-05,
      "loss": 0.9678,
      "step": 109490
    },
    {
      "epoch": 96.98405668733392,
      "grad_norm": 0.2544923722743988,
      "learning_rate": 1e-05,
      "loss": 0.929,
      "step": 109495
    },
    {
      "epoch": 96.98848538529673,
      "grad_norm": 0.25113755464553833,
      "learning_rate": 1e-05,
      "loss": 1.0231,
      "step": 109500
    },
    {
      "epoch": 96.99291408325952,
      "grad_norm": 0.20882739126682281,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 109505
    },
    {
      "epoch": 96.99734278122232,
      "grad_norm": 0.2533929646015167,
      "learning_rate": 1e-05,
      "loss": 0.9264,
      "step": 109510
    },
    {
      "epoch": 97.00177147918512,
      "grad_norm": 0.21742448210716248,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 109515
    },
    {
      "epoch": 97.00620017714792,
      "grad_norm": 0.224729984998703,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 109520
    },
    {
      "epoch": 97.01062887511071,
      "grad_norm": 0.25263383984565735,
      "learning_rate": 1e-05,
      "loss": 0.978,
      "step": 109525
    },
    {
      "epoch": 97.01505757307352,
      "grad_norm": 0.26445722579956055,
      "learning_rate": 1e-05,
      "loss": 0.9437,
      "step": 109530
    },
    {
      "epoch": 97.01948627103631,
      "grad_norm": 0.2620391249656677,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 109535
    },
    {
      "epoch": 97.02391496899911,
      "grad_norm": 0.27834469079971313,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 109540
    },
    {
      "epoch": 97.02834366696192,
      "grad_norm": 0.2653108239173889,
      "learning_rate": 1e-05,
      "loss": 0.9507,
      "step": 109545
    },
    {
      "epoch": 97.03277236492471,
      "grad_norm": 0.2766575813293457,
      "learning_rate": 1e-05,
      "loss": 0.9464,
      "step": 109550
    },
    {
      "epoch": 97.0372010628875,
      "grad_norm": 0.24768273532390594,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 109555
    },
    {
      "epoch": 97.04162976085031,
      "grad_norm": 0.2635209262371063,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 109560
    },
    {
      "epoch": 97.0460584588131,
      "grad_norm": 0.250985711812973,
      "learning_rate": 1e-05,
      "loss": 0.9704,
      "step": 109565
    },
    {
      "epoch": 97.0504871567759,
      "grad_norm": 0.24129335582256317,
      "learning_rate": 1e-05,
      "loss": 0.9374,
      "step": 109570
    },
    {
      "epoch": 97.05491585473871,
      "grad_norm": 0.26707589626312256,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 109575
    },
    {
      "epoch": 97.0593445527015,
      "grad_norm": 0.22685092687606812,
      "learning_rate": 1e-05,
      "loss": 0.9433,
      "step": 109580
    },
    {
      "epoch": 97.0637732506643,
      "grad_norm": 0.23330271244049072,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 109585
    },
    {
      "epoch": 97.0682019486271,
      "grad_norm": 0.23260751366615295,
      "learning_rate": 1e-05,
      "loss": 0.9395,
      "step": 109590
    },
    {
      "epoch": 97.0726306465899,
      "grad_norm": 0.2356194704771042,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 109595
    },
    {
      "epoch": 97.07705934455271,
      "grad_norm": 0.2609013319015503,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 109600
    },
    {
      "epoch": 97.0814880425155,
      "grad_norm": 0.2565682530403137,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 109605
    },
    {
      "epoch": 97.0859167404783,
      "grad_norm": 0.25871917605400085,
      "learning_rate": 1e-05,
      "loss": 0.9823,
      "step": 109610
    },
    {
      "epoch": 97.0903454384411,
      "grad_norm": 0.27744966745376587,
      "learning_rate": 1e-05,
      "loss": 1.0201,
      "step": 109615
    },
    {
      "epoch": 97.0947741364039,
      "grad_norm": 0.25573158264160156,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 109620
    },
    {
      "epoch": 97.0992028343667,
      "grad_norm": 0.2509702444076538,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 109625
    },
    {
      "epoch": 97.1036315323295,
      "grad_norm": 0.23970910906791687,
      "learning_rate": 1e-05,
      "loss": 0.9485,
      "step": 109630
    },
    {
      "epoch": 97.1080602302923,
      "grad_norm": 0.2555277645587921,
      "learning_rate": 1e-05,
      "loss": 0.9764,
      "step": 109635
    },
    {
      "epoch": 97.11248892825509,
      "grad_norm": 0.2591082453727722,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 109640
    },
    {
      "epoch": 97.1169176262179,
      "grad_norm": 0.2916388213634491,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 109645
    },
    {
      "epoch": 97.12134632418069,
      "grad_norm": 0.2743247151374817,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 109650
    },
    {
      "epoch": 97.12577502214349,
      "grad_norm": 0.22339360415935516,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 109655
    },
    {
      "epoch": 97.1302037201063,
      "grad_norm": 0.23839640617370605,
      "learning_rate": 1e-05,
      "loss": 0.9101,
      "step": 109660
    },
    {
      "epoch": 97.13463241806909,
      "grad_norm": 0.2508106827735901,
      "learning_rate": 1e-05,
      "loss": 0.9462,
      "step": 109665
    },
    {
      "epoch": 97.13906111603188,
      "grad_norm": 0.22281931340694427,
      "learning_rate": 1e-05,
      "loss": 0.94,
      "step": 109670
    },
    {
      "epoch": 97.14348981399469,
      "grad_norm": 0.2207503765821457,
      "learning_rate": 1e-05,
      "loss": 0.9695,
      "step": 109675
    },
    {
      "epoch": 97.14791851195749,
      "grad_norm": 0.20828008651733398,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 109680
    },
    {
      "epoch": 97.15234720992028,
      "grad_norm": 0.20707137882709503,
      "learning_rate": 1e-05,
      "loss": 0.9343,
      "step": 109685
    },
    {
      "epoch": 97.15677590788309,
      "grad_norm": 0.23699553310871124,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 109690
    },
    {
      "epoch": 97.16120460584588,
      "grad_norm": 0.22275662422180176,
      "learning_rate": 1e-05,
      "loss": 0.9442,
      "step": 109695
    },
    {
      "epoch": 97.16563330380868,
      "grad_norm": 0.2669040262699127,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 109700
    },
    {
      "epoch": 97.17006200177148,
      "grad_norm": 0.23328854143619537,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 109705
    },
    {
      "epoch": 97.17449069973428,
      "grad_norm": 0.26291340589523315,
      "learning_rate": 1e-05,
      "loss": 0.9331,
      "step": 109710
    },
    {
      "epoch": 97.17891939769707,
      "grad_norm": 0.24099422991275787,
      "learning_rate": 1e-05,
      "loss": 0.9189,
      "step": 109715
    },
    {
      "epoch": 97.18334809565988,
      "grad_norm": 0.21687135100364685,
      "learning_rate": 1e-05,
      "loss": 0.9761,
      "step": 109720
    },
    {
      "epoch": 97.18777679362267,
      "grad_norm": 0.24583666026592255,
      "learning_rate": 1e-05,
      "loss": 0.9293,
      "step": 109725
    },
    {
      "epoch": 97.19220549158547,
      "grad_norm": 0.2456425279378891,
      "learning_rate": 1e-05,
      "loss": 0.9218,
      "step": 109730
    },
    {
      "epoch": 97.19663418954828,
      "grad_norm": 0.2676035165786743,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 109735
    },
    {
      "epoch": 97.20106288751107,
      "grad_norm": 0.22334642708301544,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 109740
    },
    {
      "epoch": 97.20549158547387,
      "grad_norm": 0.21113184094429016,
      "learning_rate": 1e-05,
      "loss": 0.8868,
      "step": 109745
    },
    {
      "epoch": 97.20992028343667,
      "grad_norm": 0.21757923066616058,
      "learning_rate": 1e-05,
      "loss": 0.9253,
      "step": 109750
    },
    {
      "epoch": 97.21434898139947,
      "grad_norm": 0.24147436022758484,
      "learning_rate": 1e-05,
      "loss": 0.9246,
      "step": 109755
    },
    {
      "epoch": 97.21877767936226,
      "grad_norm": 0.21963462233543396,
      "learning_rate": 1e-05,
      "loss": 0.9108,
      "step": 109760
    },
    {
      "epoch": 97.22320637732507,
      "grad_norm": 0.22466474771499634,
      "learning_rate": 1e-05,
      "loss": 0.9356,
      "step": 109765
    },
    {
      "epoch": 97.22763507528786,
      "grad_norm": 0.24098114669322968,
      "learning_rate": 1e-05,
      "loss": 0.9888,
      "step": 109770
    },
    {
      "epoch": 97.23206377325066,
      "grad_norm": 0.24827177822589874,
      "learning_rate": 1e-05,
      "loss": 0.9387,
      "step": 109775
    },
    {
      "epoch": 97.23649247121347,
      "grad_norm": 0.2679842710494995,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 109780
    },
    {
      "epoch": 97.24092116917626,
      "grad_norm": 0.2706936001777649,
      "learning_rate": 1e-05,
      "loss": 0.9174,
      "step": 109785
    },
    {
      "epoch": 97.24534986713905,
      "grad_norm": 0.21122415363788605,
      "learning_rate": 1e-05,
      "loss": 0.9731,
      "step": 109790
    },
    {
      "epoch": 97.24977856510186,
      "grad_norm": 0.22289395332336426,
      "learning_rate": 1e-05,
      "loss": 0.9281,
      "step": 109795
    },
    {
      "epoch": 97.25420726306466,
      "grad_norm": 0.22101697325706482,
      "learning_rate": 1e-05,
      "loss": 0.9499,
      "step": 109800
    },
    {
      "epoch": 97.25863596102745,
      "grad_norm": 0.19991667568683624,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 109805
    },
    {
      "epoch": 97.26306465899026,
      "grad_norm": 0.24198384582996368,
      "learning_rate": 1e-05,
      "loss": 0.9259,
      "step": 109810
    },
    {
      "epoch": 97.26749335695305,
      "grad_norm": 0.2563733756542206,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 109815
    },
    {
      "epoch": 97.27192205491585,
      "grad_norm": 0.21383938193321228,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 109820
    },
    {
      "epoch": 97.27635075287866,
      "grad_norm": 0.2876136898994446,
      "learning_rate": 1e-05,
      "loss": 0.963,
      "step": 109825
    },
    {
      "epoch": 97.28077945084145,
      "grad_norm": 0.2394861876964569,
      "learning_rate": 1e-05,
      "loss": 0.9827,
      "step": 109830
    },
    {
      "epoch": 97.28520814880426,
      "grad_norm": 0.2696128785610199,
      "learning_rate": 1e-05,
      "loss": 0.9943,
      "step": 109835
    },
    {
      "epoch": 97.28963684676705,
      "grad_norm": 0.22445590794086456,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 109840
    },
    {
      "epoch": 97.29406554472985,
      "grad_norm": 0.25492721796035767,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 109845
    },
    {
      "epoch": 97.29849424269266,
      "grad_norm": 0.24470648169517517,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 109850
    },
    {
      "epoch": 97.30292294065545,
      "grad_norm": 0.23633340001106262,
      "learning_rate": 1e-05,
      "loss": 1.0098,
      "step": 109855
    },
    {
      "epoch": 97.30735163861824,
      "grad_norm": 0.23683921992778778,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 109860
    },
    {
      "epoch": 97.31178033658105,
      "grad_norm": 0.24632786214351654,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 109865
    },
    {
      "epoch": 97.31620903454385,
      "grad_norm": 0.23267455399036407,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 109870
    },
    {
      "epoch": 97.32063773250664,
      "grad_norm": 0.3037605583667755,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 109875
    },
    {
      "epoch": 97.32506643046945,
      "grad_norm": 0.23018823564052582,
      "learning_rate": 1e-05,
      "loss": 1.0022,
      "step": 109880
    },
    {
      "epoch": 97.32949512843224,
      "grad_norm": 0.2253151684999466,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 109885
    },
    {
      "epoch": 97.33392382639504,
      "grad_norm": 0.25150591135025024,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 109890
    },
    {
      "epoch": 97.33835252435784,
      "grad_norm": 0.20476333796977997,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 109895
    },
    {
      "epoch": 97.34278122232064,
      "grad_norm": 0.2290610820055008,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 109900
    },
    {
      "epoch": 97.34720992028343,
      "grad_norm": 0.21188905835151672,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 109905
    },
    {
      "epoch": 97.35163861824624,
      "grad_norm": 0.2743215560913086,
      "learning_rate": 1e-05,
      "loss": 0.9495,
      "step": 109910
    },
    {
      "epoch": 97.35606731620904,
      "grad_norm": 0.2437146157026291,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 109915
    },
    {
      "epoch": 97.36049601417183,
      "grad_norm": 0.21139484643936157,
      "learning_rate": 1e-05,
      "loss": 0.9296,
      "step": 109920
    },
    {
      "epoch": 97.36492471213464,
      "grad_norm": 0.2499586045742035,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 109925
    },
    {
      "epoch": 97.36935341009743,
      "grad_norm": 0.25432199239730835,
      "learning_rate": 1e-05,
      "loss": 0.9822,
      "step": 109930
    },
    {
      "epoch": 97.37378210806023,
      "grad_norm": 0.21627236902713776,
      "learning_rate": 1e-05,
      "loss": 0.925,
      "step": 109935
    },
    {
      "epoch": 97.37821080602303,
      "grad_norm": 0.22862394154071808,
      "learning_rate": 1e-05,
      "loss": 0.9209,
      "step": 109940
    },
    {
      "epoch": 97.38263950398583,
      "grad_norm": 0.2629219591617584,
      "learning_rate": 1e-05,
      "loss": 0.8942,
      "step": 109945
    },
    {
      "epoch": 97.38706820194862,
      "grad_norm": 0.2441430687904358,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 109950
    },
    {
      "epoch": 97.39149689991143,
      "grad_norm": 0.24243900179862976,
      "learning_rate": 1e-05,
      "loss": 0.8985,
      "step": 109955
    },
    {
      "epoch": 97.39592559787422,
      "grad_norm": 0.22874903678894043,
      "learning_rate": 1e-05,
      "loss": 0.9008,
      "step": 109960
    },
    {
      "epoch": 97.40035429583702,
      "grad_norm": 0.24870111048221588,
      "learning_rate": 1e-05,
      "loss": 0.9321,
      "step": 109965
    },
    {
      "epoch": 97.40478299379983,
      "grad_norm": 0.2535886764526367,
      "learning_rate": 1e-05,
      "loss": 1.0632,
      "step": 109970
    },
    {
      "epoch": 97.40921169176262,
      "grad_norm": 0.2632834017276764,
      "learning_rate": 1e-05,
      "loss": 0.9168,
      "step": 109975
    },
    {
      "epoch": 97.41364038972542,
      "grad_norm": 0.22288323938846588,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 109980
    },
    {
      "epoch": 97.41806908768822,
      "grad_norm": 0.22108377516269684,
      "learning_rate": 1e-05,
      "loss": 0.9278,
      "step": 109985
    },
    {
      "epoch": 97.42249778565102,
      "grad_norm": 0.21149767935276031,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 109990
    },
    {
      "epoch": 97.42692648361381,
      "grad_norm": 0.288916677236557,
      "learning_rate": 1e-05,
      "loss": 0.9739,
      "step": 109995
    },
    {
      "epoch": 97.43135518157662,
      "grad_norm": 0.2233758270740509,
      "learning_rate": 1e-05,
      "loss": 0.9993,
      "step": 110000
    },
    {
      "epoch": 97.43578387953941,
      "grad_norm": 0.2551714777946472,
      "learning_rate": 1e-05,
      "loss": 0.9335,
      "step": 110005
    },
    {
      "epoch": 97.44021257750221,
      "grad_norm": 0.25014278292655945,
      "learning_rate": 1e-05,
      "loss": 0.8779,
      "step": 110010
    },
    {
      "epoch": 97.44464127546502,
      "grad_norm": 0.23546725511550903,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 110015
    },
    {
      "epoch": 97.44906997342781,
      "grad_norm": 0.231979101896286,
      "learning_rate": 1e-05,
      "loss": 1.073,
      "step": 110020
    },
    {
      "epoch": 97.4534986713906,
      "grad_norm": 0.2500385642051697,
      "learning_rate": 1e-05,
      "loss": 0.9047,
      "step": 110025
    },
    {
      "epoch": 97.45792736935341,
      "grad_norm": 0.24851708114147186,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 110030
    },
    {
      "epoch": 97.46235606731621,
      "grad_norm": 0.22333583235740662,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 110035
    },
    {
      "epoch": 97.466784765279,
      "grad_norm": 0.2142449915409088,
      "learning_rate": 1e-05,
      "loss": 0.9599,
      "step": 110040
    },
    {
      "epoch": 97.47121346324181,
      "grad_norm": 0.19926738739013672,
      "learning_rate": 1e-05,
      "loss": 0.8815,
      "step": 110045
    },
    {
      "epoch": 97.4756421612046,
      "grad_norm": 0.24198172986507416,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 110050
    },
    {
      "epoch": 97.4800708591674,
      "grad_norm": 0.279943585395813,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 110055
    },
    {
      "epoch": 97.4844995571302,
      "grad_norm": 0.23222343623638153,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 110060
    },
    {
      "epoch": 97.488928255093,
      "grad_norm": 0.22931066155433655,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 110065
    },
    {
      "epoch": 97.4933569530558,
      "grad_norm": 0.23459193110466003,
      "learning_rate": 1e-05,
      "loss": 0.9931,
      "step": 110070
    },
    {
      "epoch": 97.4977856510186,
      "grad_norm": 0.22001387178897858,
      "learning_rate": 1e-05,
      "loss": 0.888,
      "step": 110075
    },
    {
      "epoch": 97.5022143489814,
      "grad_norm": 0.2374526709318161,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 110080
    },
    {
      "epoch": 97.5066430469442,
      "grad_norm": 0.25915834307670593,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 110085
    },
    {
      "epoch": 97.511071744907,
      "grad_norm": 0.24076178669929504,
      "learning_rate": 1e-05,
      "loss": 0.9312,
      "step": 110090
    },
    {
      "epoch": 97.5155004428698,
      "grad_norm": 0.25095003843307495,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 110095
    },
    {
      "epoch": 97.5199291408326,
      "grad_norm": 0.240245521068573,
      "learning_rate": 1e-05,
      "loss": 0.9448,
      "step": 110100
    },
    {
      "epoch": 97.5243578387954,
      "grad_norm": 0.25898298621177673,
      "learning_rate": 1e-05,
      "loss": 0.9449,
      "step": 110105
    },
    {
      "epoch": 97.52878653675819,
      "grad_norm": 0.22489018738269806,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 110110
    },
    {
      "epoch": 97.533215234721,
      "grad_norm": 0.2503131330013275,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 110115
    },
    {
      "epoch": 97.53764393268379,
      "grad_norm": 0.22982026636600494,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 110120
    },
    {
      "epoch": 97.54207263064659,
      "grad_norm": 0.25346243381500244,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 110125
    },
    {
      "epoch": 97.5465013286094,
      "grad_norm": 0.26209741830825806,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 110130
    },
    {
      "epoch": 97.55093002657219,
      "grad_norm": 0.25933605432510376,
      "learning_rate": 1e-05,
      "loss": 0.9435,
      "step": 110135
    },
    {
      "epoch": 97.55535872453498,
      "grad_norm": 0.2322559803724289,
      "learning_rate": 1e-05,
      "loss": 0.9068,
      "step": 110140
    },
    {
      "epoch": 97.55978742249779,
      "grad_norm": 0.26609495282173157,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 110145
    },
    {
      "epoch": 97.56421612046059,
      "grad_norm": 0.272363543510437,
      "learning_rate": 1e-05,
      "loss": 0.9559,
      "step": 110150
    },
    {
      "epoch": 97.56864481842338,
      "grad_norm": 0.22329014539718628,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 110155
    },
    {
      "epoch": 97.57307351638619,
      "grad_norm": 0.2261374443769455,
      "learning_rate": 1e-05,
      "loss": 0.9881,
      "step": 110160
    },
    {
      "epoch": 97.57750221434898,
      "grad_norm": 0.3081692159175873,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 110165
    },
    {
      "epoch": 97.58193091231178,
      "grad_norm": 0.2734352946281433,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 110170
    },
    {
      "epoch": 97.58635961027458,
      "grad_norm": 0.23925837874412537,
      "learning_rate": 1e-05,
      "loss": 0.9593,
      "step": 110175
    },
    {
      "epoch": 97.59078830823738,
      "grad_norm": 0.2325543761253357,
      "learning_rate": 1e-05,
      "loss": 0.9509,
      "step": 110180
    },
    {
      "epoch": 97.59521700620017,
      "grad_norm": 0.25276368856430054,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 110185
    },
    {
      "epoch": 97.59964570416298,
      "grad_norm": 0.24771519005298615,
      "learning_rate": 1e-05,
      "loss": 0.9214,
      "step": 110190
    },
    {
      "epoch": 97.60407440212578,
      "grad_norm": 0.23299753665924072,
      "learning_rate": 1e-05,
      "loss": 0.9424,
      "step": 110195
    },
    {
      "epoch": 97.60850310008857,
      "grad_norm": 0.2554718255996704,
      "learning_rate": 1e-05,
      "loss": 0.9318,
      "step": 110200
    },
    {
      "epoch": 97.61293179805138,
      "grad_norm": 0.2444372922182083,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 110205
    },
    {
      "epoch": 97.61736049601417,
      "grad_norm": 0.25932085514068604,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 110210
    },
    {
      "epoch": 97.62178919397697,
      "grad_norm": 0.24134407937526703,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 110215
    },
    {
      "epoch": 97.62621789193977,
      "grad_norm": 0.2226153314113617,
      "learning_rate": 1e-05,
      "loss": 0.9312,
      "step": 110220
    },
    {
      "epoch": 97.63064658990257,
      "grad_norm": 0.2217981517314911,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 110225
    },
    {
      "epoch": 97.63507528786536,
      "grad_norm": 0.1999123990535736,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 110230
    },
    {
      "epoch": 97.63950398582817,
      "grad_norm": 0.18966734409332275,
      "learning_rate": 1e-05,
      "loss": 0.9453,
      "step": 110235
    },
    {
      "epoch": 97.64393268379096,
      "grad_norm": 0.2350645512342453,
      "learning_rate": 1e-05,
      "loss": 0.9705,
      "step": 110240
    },
    {
      "epoch": 97.64836138175376,
      "grad_norm": 0.23140601813793182,
      "learning_rate": 1e-05,
      "loss": 0.9728,
      "step": 110245
    },
    {
      "epoch": 97.65279007971657,
      "grad_norm": 0.28096091747283936,
      "learning_rate": 1e-05,
      "loss": 0.961,
      "step": 110250
    },
    {
      "epoch": 97.65721877767936,
      "grad_norm": 0.24578042328357697,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 110255
    },
    {
      "epoch": 97.66164747564216,
      "grad_norm": 0.20488551259040833,
      "learning_rate": 1e-05,
      "loss": 0.9616,
      "step": 110260
    },
    {
      "epoch": 97.66607617360496,
      "grad_norm": 0.2229185849428177,
      "learning_rate": 1e-05,
      "loss": 0.8908,
      "step": 110265
    },
    {
      "epoch": 97.67050487156776,
      "grad_norm": 0.3444502353668213,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 110270
    },
    {
      "epoch": 97.67493356953055,
      "grad_norm": 0.23644956946372986,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 110275
    },
    {
      "epoch": 97.67936226749336,
      "grad_norm": 0.22033430635929108,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 110280
    },
    {
      "epoch": 97.68379096545615,
      "grad_norm": 0.22529464960098267,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 110285
    },
    {
      "epoch": 97.68821966341895,
      "grad_norm": 0.2736489772796631,
      "learning_rate": 1e-05,
      "loss": 0.9303,
      "step": 110290
    },
    {
      "epoch": 97.69264836138176,
      "grad_norm": 0.24621880054473877,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 110295
    },
    {
      "epoch": 97.69707705934455,
      "grad_norm": 0.236724853515625,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 110300
    },
    {
      "epoch": 97.70150575730734,
      "grad_norm": 0.2793537974357605,
      "learning_rate": 1e-05,
      "loss": 0.943,
      "step": 110305
    },
    {
      "epoch": 97.70593445527015,
      "grad_norm": 0.22544656693935394,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 110310
    },
    {
      "epoch": 97.71036315323295,
      "grad_norm": 0.2598043382167816,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 110315
    },
    {
      "epoch": 97.71479185119574,
      "grad_norm": 0.26090654730796814,
      "learning_rate": 1e-05,
      "loss": 0.9765,
      "step": 110320
    },
    {
      "epoch": 97.71922054915855,
      "grad_norm": 0.19919486343860626,
      "learning_rate": 1e-05,
      "loss": 0.9386,
      "step": 110325
    },
    {
      "epoch": 97.72364924712134,
      "grad_norm": 0.25627952814102173,
      "learning_rate": 1e-05,
      "loss": 0.9506,
      "step": 110330
    },
    {
      "epoch": 97.72807794508415,
      "grad_norm": 0.2522720694541931,
      "learning_rate": 1e-05,
      "loss": 0.9429,
      "step": 110335
    },
    {
      "epoch": 97.73250664304695,
      "grad_norm": 0.21583019196987152,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 110340
    },
    {
      "epoch": 97.73693534100974,
      "grad_norm": 0.23461996018886566,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 110345
    },
    {
      "epoch": 97.74136403897255,
      "grad_norm": 0.20691999793052673,
      "learning_rate": 1e-05,
      "loss": 0.956,
      "step": 110350
    },
    {
      "epoch": 97.74579273693534,
      "grad_norm": 0.26187989115715027,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 110355
    },
    {
      "epoch": 97.75022143489814,
      "grad_norm": 0.20601873099803925,
      "learning_rate": 1e-05,
      "loss": 0.9857,
      "step": 110360
    },
    {
      "epoch": 97.75465013286095,
      "grad_norm": 0.23647361993789673,
      "learning_rate": 1e-05,
      "loss": 0.9168,
      "step": 110365
    },
    {
      "epoch": 97.75907883082374,
      "grad_norm": 0.22742019593715668,
      "learning_rate": 1e-05,
      "loss": 0.8856,
      "step": 110370
    },
    {
      "epoch": 97.76350752878653,
      "grad_norm": 0.26356250047683716,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 110375
    },
    {
      "epoch": 97.76793622674934,
      "grad_norm": 0.26778408885002136,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 110380
    },
    {
      "epoch": 97.77236492471214,
      "grad_norm": 0.24719949066638947,
      "learning_rate": 1e-05,
      "loss": 0.9233,
      "step": 110385
    },
    {
      "epoch": 97.77679362267493,
      "grad_norm": 0.21829064190387726,
      "learning_rate": 1e-05,
      "loss": 0.9144,
      "step": 110390
    },
    {
      "epoch": 97.78122232063774,
      "grad_norm": 0.21765105426311493,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 110395
    },
    {
      "epoch": 97.78565101860053,
      "grad_norm": 0.24503357708454132,
      "learning_rate": 1e-05,
      "loss": 0.948,
      "step": 110400
    },
    {
      "epoch": 97.79007971656333,
      "grad_norm": 0.23801587522029877,
      "learning_rate": 1e-05,
      "loss": 0.9793,
      "step": 110405
    },
    {
      "epoch": 97.79450841452613,
      "grad_norm": 0.284613698720932,
      "learning_rate": 1e-05,
      "loss": 0.9332,
      "step": 110410
    },
    {
      "epoch": 97.79893711248893,
      "grad_norm": 0.2725047767162323,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 110415
    },
    {
      "epoch": 97.80336581045172,
      "grad_norm": 0.2405172884464264,
      "learning_rate": 1e-05,
      "loss": 0.9555,
      "step": 110420
    },
    {
      "epoch": 97.80779450841453,
      "grad_norm": 0.23448525369167328,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 110425
    },
    {
      "epoch": 97.81222320637733,
      "grad_norm": 0.24444708228111267,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 110430
    },
    {
      "epoch": 97.81665190434012,
      "grad_norm": 0.23331406712532043,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 110435
    },
    {
      "epoch": 97.82108060230293,
      "grad_norm": 0.25338447093963623,
      "learning_rate": 1e-05,
      "loss": 0.9525,
      "step": 110440
    },
    {
      "epoch": 97.82550930026572,
      "grad_norm": 0.22169671952724457,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 110445
    },
    {
      "epoch": 97.82993799822852,
      "grad_norm": 0.2375779002904892,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 110450
    },
    {
      "epoch": 97.83436669619132,
      "grad_norm": 0.2070441097021103,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 110455
    },
    {
      "epoch": 97.83879539415412,
      "grad_norm": 0.24972273409366608,
      "learning_rate": 1e-05,
      "loss": 0.9736,
      "step": 110460
    },
    {
      "epoch": 97.84322409211691,
      "grad_norm": 0.22928331792354584,
      "learning_rate": 1e-05,
      "loss": 0.8943,
      "step": 110465
    },
    {
      "epoch": 97.84765279007972,
      "grad_norm": 0.2580093741416931,
      "learning_rate": 1e-05,
      "loss": 0.9242,
      "step": 110470
    },
    {
      "epoch": 97.85208148804251,
      "grad_norm": 0.24549056589603424,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 110475
    },
    {
      "epoch": 97.85651018600531,
      "grad_norm": 0.2445269674062729,
      "learning_rate": 1e-05,
      "loss": 0.9971,
      "step": 110480
    },
    {
      "epoch": 97.86093888396812,
      "grad_norm": 0.2873865067958832,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 110485
    },
    {
      "epoch": 97.86536758193091,
      "grad_norm": 0.2622067630290985,
      "learning_rate": 1e-05,
      "loss": 0.9377,
      "step": 110490
    },
    {
      "epoch": 97.8697962798937,
      "grad_norm": 0.21422092616558075,
      "learning_rate": 1e-05,
      "loss": 0.962,
      "step": 110495
    },
    {
      "epoch": 97.87422497785651,
      "grad_norm": 0.24483653903007507,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 110500
    },
    {
      "epoch": 97.87865367581931,
      "grad_norm": 0.24253413081169128,
      "learning_rate": 1e-05,
      "loss": 1.0047,
      "step": 110505
    },
    {
      "epoch": 97.8830823737821,
      "grad_norm": 0.2257661074399948,
      "learning_rate": 1e-05,
      "loss": 0.95,
      "step": 110510
    },
    {
      "epoch": 97.88751107174491,
      "grad_norm": 0.27211010456085205,
      "learning_rate": 1e-05,
      "loss": 0.9837,
      "step": 110515
    },
    {
      "epoch": 97.8919397697077,
      "grad_norm": 0.27671968936920166,
      "learning_rate": 1e-05,
      "loss": 0.9929,
      "step": 110520
    },
    {
      "epoch": 97.8963684676705,
      "grad_norm": 0.22439676523208618,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 110525
    },
    {
      "epoch": 97.9007971656333,
      "grad_norm": 0.2122560292482376,
      "learning_rate": 1e-05,
      "loss": 0.9373,
      "step": 110530
    },
    {
      "epoch": 97.9052258635961,
      "grad_norm": 0.2398727834224701,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 110535
    },
    {
      "epoch": 97.9096545615589,
      "grad_norm": 0.2522470951080322,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 110540
    },
    {
      "epoch": 97.9140832595217,
      "grad_norm": 0.22640301287174225,
      "learning_rate": 1e-05,
      "loss": 0.9038,
      "step": 110545
    },
    {
      "epoch": 97.9185119574845,
      "grad_norm": 0.23604297637939453,
      "learning_rate": 1e-05,
      "loss": 0.9596,
      "step": 110550
    },
    {
      "epoch": 97.92294065544729,
      "grad_norm": 0.27733513712882996,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 110555
    },
    {
      "epoch": 97.9273693534101,
      "grad_norm": 0.22241398692131042,
      "learning_rate": 1e-05,
      "loss": 0.9359,
      "step": 110560
    },
    {
      "epoch": 97.9317980513729,
      "grad_norm": 0.247050940990448,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 110565
    },
    {
      "epoch": 97.9362267493357,
      "grad_norm": 0.238209068775177,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 110570
    },
    {
      "epoch": 97.9406554472985,
      "grad_norm": 0.237433522939682,
      "learning_rate": 1e-05,
      "loss": 0.9951,
      "step": 110575
    },
    {
      "epoch": 97.94508414526129,
      "grad_norm": 0.20334464311599731,
      "learning_rate": 1e-05,
      "loss": 0.9219,
      "step": 110580
    },
    {
      "epoch": 97.9495128432241,
      "grad_norm": 0.20749704539775848,
      "learning_rate": 1e-05,
      "loss": 0.8973,
      "step": 110585
    },
    {
      "epoch": 97.9539415411869,
      "grad_norm": 0.2957179844379425,
      "learning_rate": 1e-05,
      "loss": 0.9547,
      "step": 110590
    },
    {
      "epoch": 97.95837023914969,
      "grad_norm": 0.2433343380689621,
      "learning_rate": 1e-05,
      "loss": 0.9777,
      "step": 110595
    },
    {
      "epoch": 97.9627989371125,
      "grad_norm": 0.22279171645641327,
      "learning_rate": 1e-05,
      "loss": 0.9135,
      "step": 110600
    },
    {
      "epoch": 97.96722763507529,
      "grad_norm": 0.22962026298046112,
      "learning_rate": 1e-05,
      "loss": 0.9683,
      "step": 110605
    },
    {
      "epoch": 97.97165633303808,
      "grad_norm": 0.2327713966369629,
      "learning_rate": 1e-05,
      "loss": 0.9933,
      "step": 110610
    },
    {
      "epoch": 97.97608503100089,
      "grad_norm": 0.305185467004776,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 110615
    },
    {
      "epoch": 97.98051372896369,
      "grad_norm": 0.24828708171844482,
      "learning_rate": 1e-05,
      "loss": 0.973,
      "step": 110620
    },
    {
      "epoch": 97.98494242692648,
      "grad_norm": 0.24489706754684448,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 110625
    },
    {
      "epoch": 97.98937112488929,
      "grad_norm": 0.21437722444534302,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 110630
    },
    {
      "epoch": 97.99379982285208,
      "grad_norm": 0.24583928287029266,
      "learning_rate": 1e-05,
      "loss": 0.9716,
      "step": 110635
    },
    {
      "epoch": 97.99822852081488,
      "grad_norm": 0.21689701080322266,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 110640
    },
    {
      "epoch": 98.00265721877768,
      "grad_norm": 0.24357813596725464,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 110645
    },
    {
      "epoch": 98.00708591674048,
      "grad_norm": 0.2195167988538742,
      "learning_rate": 1e-05,
      "loss": 0.955,
      "step": 110650
    },
    {
      "epoch": 98.01151461470327,
      "grad_norm": 0.2015455663204193,
      "learning_rate": 1e-05,
      "loss": 0.9651,
      "step": 110655
    },
    {
      "epoch": 98.01594331266608,
      "grad_norm": 0.2187841385602951,
      "learning_rate": 1e-05,
      "loss": 0.925,
      "step": 110660
    },
    {
      "epoch": 98.02037201062888,
      "grad_norm": 0.2483312338590622,
      "learning_rate": 1e-05,
      "loss": 0.9202,
      "step": 110665
    },
    {
      "epoch": 98.02480070859167,
      "grad_norm": 0.22294673323631287,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 110670
    },
    {
      "epoch": 98.02922940655448,
      "grad_norm": 0.24270865321159363,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 110675
    },
    {
      "epoch": 98.03365810451727,
      "grad_norm": 0.24967585504055023,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 110680
    },
    {
      "epoch": 98.03808680248007,
      "grad_norm": 0.2395699918270111,
      "learning_rate": 1e-05,
      "loss": 0.9636,
      "step": 110685
    },
    {
      "epoch": 98.04251550044287,
      "grad_norm": 0.20785701274871826,
      "learning_rate": 1e-05,
      "loss": 0.9649,
      "step": 110690
    },
    {
      "epoch": 98.04694419840567,
      "grad_norm": 0.2575901448726654,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 110695
    },
    {
      "epoch": 98.05137289636846,
      "grad_norm": 0.2463057041168213,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 110700
    },
    {
      "epoch": 98.05580159433127,
      "grad_norm": 0.23454691469669342,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 110705
    },
    {
      "epoch": 98.06023029229407,
      "grad_norm": 0.20412150025367737,
      "learning_rate": 1e-05,
      "loss": 0.9537,
      "step": 110710
    },
    {
      "epoch": 98.06465899025686,
      "grad_norm": 0.1934882253408432,
      "learning_rate": 1e-05,
      "loss": 0.915,
      "step": 110715
    },
    {
      "epoch": 98.06908768821967,
      "grad_norm": 0.24537554383277893,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 110720
    },
    {
      "epoch": 98.07351638618246,
      "grad_norm": 0.2861257493495941,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 110725
    },
    {
      "epoch": 98.07794508414526,
      "grad_norm": 0.29009395837783813,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 110730
    },
    {
      "epoch": 98.08237378210806,
      "grad_norm": 0.26506465673446655,
      "learning_rate": 1e-05,
      "loss": 0.9657,
      "step": 110735
    },
    {
      "epoch": 98.08680248007086,
      "grad_norm": 0.27491116523742676,
      "learning_rate": 1e-05,
      "loss": 0.9711,
      "step": 110740
    },
    {
      "epoch": 98.09123117803365,
      "grad_norm": 0.2540963292121887,
      "learning_rate": 1e-05,
      "loss": 0.9341,
      "step": 110745
    },
    {
      "epoch": 98.09565987599646,
      "grad_norm": 0.2668173015117645,
      "learning_rate": 1e-05,
      "loss": 0.9492,
      "step": 110750
    },
    {
      "epoch": 98.10008857395925,
      "grad_norm": 0.25814470648765564,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 110755
    },
    {
      "epoch": 98.10451727192205,
      "grad_norm": 0.3089388906955719,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 110760
    },
    {
      "epoch": 98.10894596988486,
      "grad_norm": 0.2397197037935257,
      "learning_rate": 1e-05,
      "loss": 0.9169,
      "step": 110765
    },
    {
      "epoch": 98.11337466784765,
      "grad_norm": 0.25080448389053345,
      "learning_rate": 1e-05,
      "loss": 1.006,
      "step": 110770
    },
    {
      "epoch": 98.11780336581045,
      "grad_norm": 0.2382594496011734,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 110775
    },
    {
      "epoch": 98.12223206377325,
      "grad_norm": 0.22635255753993988,
      "learning_rate": 1e-05,
      "loss": 0.8874,
      "step": 110780
    },
    {
      "epoch": 98.12666076173605,
      "grad_norm": 0.2467639446258545,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 110785
    },
    {
      "epoch": 98.13108945969884,
      "grad_norm": 0.2402334064245224,
      "learning_rate": 1e-05,
      "loss": 0.9088,
      "step": 110790
    },
    {
      "epoch": 98.13551815766165,
      "grad_norm": 0.20824818313121796,
      "learning_rate": 1e-05,
      "loss": 0.9316,
      "step": 110795
    },
    {
      "epoch": 98.13994685562444,
      "grad_norm": 0.21272899210453033,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 110800
    },
    {
      "epoch": 98.14437555358724,
      "grad_norm": 0.2517433762550354,
      "learning_rate": 1e-05,
      "loss": 0.921,
      "step": 110805
    },
    {
      "epoch": 98.14880425155005,
      "grad_norm": 0.2900422215461731,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 110810
    },
    {
      "epoch": 98.15323294951284,
      "grad_norm": 0.26187029480934143,
      "learning_rate": 1e-05,
      "loss": 0.92,
      "step": 110815
    },
    {
      "epoch": 98.15766164747565,
      "grad_norm": 0.21175190806388855,
      "learning_rate": 1e-05,
      "loss": 0.8952,
      "step": 110820
    },
    {
      "epoch": 98.16209034543844,
      "grad_norm": 0.2889721393585205,
      "learning_rate": 1e-05,
      "loss": 0.9208,
      "step": 110825
    },
    {
      "epoch": 98.16651904340124,
      "grad_norm": 0.2230461984872818,
      "learning_rate": 1e-05,
      "loss": 0.9714,
      "step": 110830
    },
    {
      "epoch": 98.17094774136405,
      "grad_norm": 0.23982511460781097,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 110835
    },
    {
      "epoch": 98.17537643932684,
      "grad_norm": 0.226076140999794,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 110840
    },
    {
      "epoch": 98.17980513728963,
      "grad_norm": 0.23692019283771515,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 110845
    },
    {
      "epoch": 98.18423383525244,
      "grad_norm": 0.23556871712207794,
      "learning_rate": 1e-05,
      "loss": 1.013,
      "step": 110850
    },
    {
      "epoch": 98.18866253321524,
      "grad_norm": 0.23961105942726135,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 110855
    },
    {
      "epoch": 98.19309123117803,
      "grad_norm": 0.23254182934761047,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 110860
    },
    {
      "epoch": 98.19751992914084,
      "grad_norm": 0.24206198751926422,
      "learning_rate": 1e-05,
      "loss": 0.9729,
      "step": 110865
    },
    {
      "epoch": 98.20194862710363,
      "grad_norm": 0.24894015491008759,
      "learning_rate": 1e-05,
      "loss": 0.9469,
      "step": 110870
    },
    {
      "epoch": 98.20637732506643,
      "grad_norm": 0.2574262320995331,
      "learning_rate": 1e-05,
      "loss": 0.9475,
      "step": 110875
    },
    {
      "epoch": 98.21080602302924,
      "grad_norm": 0.27776750922203064,
      "learning_rate": 1e-05,
      "loss": 0.9662,
      "step": 110880
    },
    {
      "epoch": 98.21523472099203,
      "grad_norm": 0.2227364480495453,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 110885
    },
    {
      "epoch": 98.21966341895482,
      "grad_norm": 0.24638865888118744,
      "learning_rate": 1e-05,
      "loss": 0.9116,
      "step": 110890
    },
    {
      "epoch": 98.22409211691763,
      "grad_norm": 0.2511344254016876,
      "learning_rate": 1e-05,
      "loss": 0.9451,
      "step": 110895
    },
    {
      "epoch": 98.22852081488043,
      "grad_norm": 0.2079719603061676,
      "learning_rate": 1e-05,
      "loss": 0.9457,
      "step": 110900
    },
    {
      "epoch": 98.23294951284322,
      "grad_norm": 0.2462829202413559,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 110905
    },
    {
      "epoch": 98.23737821080603,
      "grad_norm": 0.24128328263759613,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 110910
    },
    {
      "epoch": 98.24180690876882,
      "grad_norm": 0.278601735830307,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 110915
    },
    {
      "epoch": 98.24623560673162,
      "grad_norm": 0.2555367946624756,
      "learning_rate": 1e-05,
      "loss": 0.9733,
      "step": 110920
    },
    {
      "epoch": 98.25066430469442,
      "grad_norm": 0.22950394451618195,
      "learning_rate": 1e-05,
      "loss": 0.9582,
      "step": 110925
    },
    {
      "epoch": 98.25509300265722,
      "grad_norm": 0.23852285742759705,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 110930
    },
    {
      "epoch": 98.25952170062001,
      "grad_norm": 0.2170538455247879,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 110935
    },
    {
      "epoch": 98.26395039858282,
      "grad_norm": 0.22956758737564087,
      "learning_rate": 1e-05,
      "loss": 0.9858,
      "step": 110940
    },
    {
      "epoch": 98.26837909654562,
      "grad_norm": 0.24847343564033508,
      "learning_rate": 1e-05,
      "loss": 1.0357,
      "step": 110945
    },
    {
      "epoch": 98.27280779450841,
      "grad_norm": 0.23312512040138245,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 110950
    },
    {
      "epoch": 98.27723649247122,
      "grad_norm": 0.2557574212551117,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 110955
    },
    {
      "epoch": 98.28166519043401,
      "grad_norm": 0.2281859815120697,
      "learning_rate": 1e-05,
      "loss": 0.9452,
      "step": 110960
    },
    {
      "epoch": 98.2860938883968,
      "grad_norm": 0.25699982047080994,
      "learning_rate": 1e-05,
      "loss": 1.03,
      "step": 110965
    },
    {
      "epoch": 98.29052258635961,
      "grad_norm": 0.3048710227012634,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 110970
    },
    {
      "epoch": 98.29495128432241,
      "grad_norm": 0.22279497981071472,
      "learning_rate": 1e-05,
      "loss": 0.9304,
      "step": 110975
    },
    {
      "epoch": 98.2993799822852,
      "grad_norm": 0.2391633689403534,
      "learning_rate": 1e-05,
      "loss": 0.9353,
      "step": 110980
    },
    {
      "epoch": 98.30380868024801,
      "grad_norm": 0.23048844933509827,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 110985
    },
    {
      "epoch": 98.3082373782108,
      "grad_norm": 0.2232014685869217,
      "learning_rate": 1e-05,
      "loss": 0.9627,
      "step": 110990
    },
    {
      "epoch": 98.3126660761736,
      "grad_norm": 0.2609648108482361,
      "learning_rate": 1e-05,
      "loss": 0.9334,
      "step": 110995
    },
    {
      "epoch": 98.31709477413641,
      "grad_norm": 0.2280598133802414,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 111000
    },
    {
      "epoch": 98.3215234720992,
      "grad_norm": 0.21310418844223022,
      "learning_rate": 1e-05,
      "loss": 0.96,
      "step": 111005
    },
    {
      "epoch": 98.325952170062,
      "grad_norm": 0.2580728232860565,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 111010
    },
    {
      "epoch": 98.3303808680248,
      "grad_norm": 0.2343798577785492,
      "learning_rate": 1e-05,
      "loss": 0.9081,
      "step": 111015
    },
    {
      "epoch": 98.3348095659876,
      "grad_norm": 0.23639167845249176,
      "learning_rate": 1e-05,
      "loss": 0.8776,
      "step": 111020
    },
    {
      "epoch": 98.33923826395039,
      "grad_norm": 0.24484464526176453,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 111025
    },
    {
      "epoch": 98.3436669619132,
      "grad_norm": 0.22854070365428925,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 111030
    },
    {
      "epoch": 98.348095659876,
      "grad_norm": 0.2719861567020416,
      "learning_rate": 1e-05,
      "loss": 0.9306,
      "step": 111035
    },
    {
      "epoch": 98.35252435783879,
      "grad_norm": 0.21270538866519928,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 111040
    },
    {
      "epoch": 98.3569530558016,
      "grad_norm": 0.2241915911436081,
      "learning_rate": 1e-05,
      "loss": 0.9233,
      "step": 111045
    },
    {
      "epoch": 98.36138175376439,
      "grad_norm": 0.2666899859905243,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 111050
    },
    {
      "epoch": 98.36581045172719,
      "grad_norm": 0.25463852286338806,
      "learning_rate": 1e-05,
      "loss": 0.9434,
      "step": 111055
    },
    {
      "epoch": 98.37023914969,
      "grad_norm": 0.20345906913280487,
      "learning_rate": 1e-05,
      "loss": 0.9868,
      "step": 111060
    },
    {
      "epoch": 98.37466784765279,
      "grad_norm": 0.2599492073059082,
      "learning_rate": 1e-05,
      "loss": 0.9399,
      "step": 111065
    },
    {
      "epoch": 98.3790965456156,
      "grad_norm": 0.23631832003593445,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 111070
    },
    {
      "epoch": 98.38352524357839,
      "grad_norm": 0.22881107032299042,
      "learning_rate": 1e-05,
      "loss": 1.0154,
      "step": 111075
    },
    {
      "epoch": 98.38795394154118,
      "grad_norm": 0.24219910800457,
      "learning_rate": 1e-05,
      "loss": 0.9947,
      "step": 111080
    },
    {
      "epoch": 98.39238263950399,
      "grad_norm": 0.2311176061630249,
      "learning_rate": 1e-05,
      "loss": 0.9421,
      "step": 111085
    },
    {
      "epoch": 98.39681133746679,
      "grad_norm": 0.24359175562858582,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 111090
    },
    {
      "epoch": 98.40124003542958,
      "grad_norm": 0.2252773642539978,
      "learning_rate": 1e-05,
      "loss": 0.9901,
      "step": 111095
    },
    {
      "epoch": 98.40566873339239,
      "grad_norm": 0.23360174894332886,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 111100
    },
    {
      "epoch": 98.41009743135518,
      "grad_norm": 0.2345714271068573,
      "learning_rate": 1e-05,
      "loss": 0.9631,
      "step": 111105
    },
    {
      "epoch": 98.41452612931798,
      "grad_norm": 0.2597193717956543,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 111110
    },
    {
      "epoch": 98.41895482728079,
      "grad_norm": 0.2260092943906784,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 111115
    },
    {
      "epoch": 98.42338352524358,
      "grad_norm": 0.25709378719329834,
      "learning_rate": 1e-05,
      "loss": 0.9213,
      "step": 111120
    },
    {
      "epoch": 98.42781222320637,
      "grad_norm": 0.2524168789386749,
      "learning_rate": 1e-05,
      "loss": 0.9337,
      "step": 111125
    },
    {
      "epoch": 98.43224092116918,
      "grad_norm": 0.2275114357471466,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 111130
    },
    {
      "epoch": 98.43666961913198,
      "grad_norm": 0.20537328720092773,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 111135
    },
    {
      "epoch": 98.44109831709477,
      "grad_norm": 0.2553630471229553,
      "learning_rate": 1e-05,
      "loss": 0.9091,
      "step": 111140
    },
    {
      "epoch": 98.44552701505758,
      "grad_norm": 0.20277798175811768,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 111145
    },
    {
      "epoch": 98.44995571302037,
      "grad_norm": 0.23031894862651825,
      "learning_rate": 1e-05,
      "loss": 0.8984,
      "step": 111150
    },
    {
      "epoch": 98.45438441098317,
      "grad_norm": 0.23661047220230103,
      "learning_rate": 1e-05,
      "loss": 0.9552,
      "step": 111155
    },
    {
      "epoch": 98.45881310894598,
      "grad_norm": 0.21697628498077393,
      "learning_rate": 1e-05,
      "loss": 0.9578,
      "step": 111160
    },
    {
      "epoch": 98.46324180690877,
      "grad_norm": 0.3396170139312744,
      "learning_rate": 1e-05,
      "loss": 0.9665,
      "step": 111165
    },
    {
      "epoch": 98.46767050487156,
      "grad_norm": 0.2101564109325409,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 111170
    },
    {
      "epoch": 98.47209920283437,
      "grad_norm": 0.28520822525024414,
      "learning_rate": 1e-05,
      "loss": 0.9655,
      "step": 111175
    },
    {
      "epoch": 98.47652790079717,
      "grad_norm": 0.2421087920665741,
      "learning_rate": 1e-05,
      "loss": 0.9274,
      "step": 111180
    },
    {
      "epoch": 98.48095659875996,
      "grad_norm": 0.205408975481987,
      "learning_rate": 1e-05,
      "loss": 0.8917,
      "step": 111185
    },
    {
      "epoch": 98.48538529672277,
      "grad_norm": 0.2246294468641281,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 111190
    },
    {
      "epoch": 98.48981399468556,
      "grad_norm": 0.23292651772499084,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 111195
    },
    {
      "epoch": 98.49424269264836,
      "grad_norm": 0.23167160153388977,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 111200
    },
    {
      "epoch": 98.49867139061116,
      "grad_norm": 0.2443193942308426,
      "learning_rate": 1e-05,
      "loss": 0.9092,
      "step": 111205
    },
    {
      "epoch": 98.50310008857396,
      "grad_norm": 0.2544289529323578,
      "learning_rate": 1e-05,
      "loss": 0.9123,
      "step": 111210
    },
    {
      "epoch": 98.50752878653675,
      "grad_norm": 0.22717690467834473,
      "learning_rate": 1e-05,
      "loss": 0.927,
      "step": 111215
    },
    {
      "epoch": 98.51195748449956,
      "grad_norm": 0.2539583444595337,
      "learning_rate": 1e-05,
      "loss": 0.9961,
      "step": 111220
    },
    {
      "epoch": 98.51638618246236,
      "grad_norm": 0.25394365191459656,
      "learning_rate": 1e-05,
      "loss": 0.967,
      "step": 111225
    },
    {
      "epoch": 98.52081488042515,
      "grad_norm": 0.2882280647754669,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 111230
    },
    {
      "epoch": 98.52524357838796,
      "grad_norm": 0.2492353916168213,
      "learning_rate": 1e-05,
      "loss": 0.9414,
      "step": 111235
    },
    {
      "epoch": 98.52967227635075,
      "grad_norm": 0.270648330450058,
      "learning_rate": 1e-05,
      "loss": 0.9358,
      "step": 111240
    },
    {
      "epoch": 98.53410097431355,
      "grad_norm": 0.2774631977081299,
      "learning_rate": 1e-05,
      "loss": 0.997,
      "step": 111245
    },
    {
      "epoch": 98.53852967227635,
      "grad_norm": 0.2577580511569977,
      "learning_rate": 1e-05,
      "loss": 0.9568,
      "step": 111250
    },
    {
      "epoch": 98.54295837023915,
      "grad_norm": 0.26933014392852783,
      "learning_rate": 1e-05,
      "loss": 0.9203,
      "step": 111255
    },
    {
      "epoch": 98.54738706820194,
      "grad_norm": 0.22621290385723114,
      "learning_rate": 1e-05,
      "loss": 0.9412,
      "step": 111260
    },
    {
      "epoch": 98.55181576616475,
      "grad_norm": 0.23177401721477509,
      "learning_rate": 1e-05,
      "loss": 0.9204,
      "step": 111265
    },
    {
      "epoch": 98.55624446412754,
      "grad_norm": 0.24310125410556793,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 111270
    },
    {
      "epoch": 98.56067316209034,
      "grad_norm": 0.270818293094635,
      "learning_rate": 1e-05,
      "loss": 1.0087,
      "step": 111275
    },
    {
      "epoch": 98.56510186005315,
      "grad_norm": 0.23837019503116608,
      "learning_rate": 1e-05,
      "loss": 0.8923,
      "step": 111280
    },
    {
      "epoch": 98.56953055801594,
      "grad_norm": 0.2304750233888626,
      "learning_rate": 1e-05,
      "loss": 0.9264,
      "step": 111285
    },
    {
      "epoch": 98.57395925597874,
      "grad_norm": 0.2091333419084549,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 111290
    },
    {
      "epoch": 98.57838795394154,
      "grad_norm": 0.2561667263507843,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 111295
    },
    {
      "epoch": 98.58281665190434,
      "grad_norm": 0.23644684255123138,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 111300
    },
    {
      "epoch": 98.58724534986715,
      "grad_norm": 0.26437947154045105,
      "learning_rate": 1e-05,
      "loss": 0.9754,
      "step": 111305
    },
    {
      "epoch": 98.59167404782994,
      "grad_norm": 0.2136402279138565,
      "learning_rate": 1e-05,
      "loss": 0.9994,
      "step": 111310
    },
    {
      "epoch": 98.59610274579273,
      "grad_norm": 0.2597300708293915,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 111315
    },
    {
      "epoch": 98.60053144375554,
      "grad_norm": 0.20606687664985657,
      "learning_rate": 1e-05,
      "loss": 0.9363,
      "step": 111320
    },
    {
      "epoch": 98.60496014171834,
      "grad_norm": 0.2137267142534256,
      "learning_rate": 1e-05,
      "loss": 0.9213,
      "step": 111325
    },
    {
      "epoch": 98.60938883968113,
      "grad_norm": 0.21220698952674866,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 111330
    },
    {
      "epoch": 98.61381753764394,
      "grad_norm": 0.2315533310174942,
      "learning_rate": 1e-05,
      "loss": 0.9479,
      "step": 111335
    },
    {
      "epoch": 98.61824623560673,
      "grad_norm": 0.23546794056892395,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 111340
    },
    {
      "epoch": 98.62267493356953,
      "grad_norm": 0.23706048727035522,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 111345
    },
    {
      "epoch": 98.62710363153234,
      "grad_norm": 0.2531529366970062,
      "learning_rate": 1e-05,
      "loss": 0.9898,
      "step": 111350
    },
    {
      "epoch": 98.63153232949513,
      "grad_norm": 0.23595298826694489,
      "learning_rate": 1e-05,
      "loss": 0.9521,
      "step": 111355
    },
    {
      "epoch": 98.63596102745792,
      "grad_norm": 0.22604885697364807,
      "learning_rate": 1e-05,
      "loss": 0.941,
      "step": 111360
    },
    {
      "epoch": 98.64038972542073,
      "grad_norm": 0.20171397924423218,
      "learning_rate": 1e-05,
      "loss": 0.9661,
      "step": 111365
    },
    {
      "epoch": 98.64481842338353,
      "grad_norm": 0.22373943030834198,
      "learning_rate": 1e-05,
      "loss": 0.9508,
      "step": 111370
    },
    {
      "epoch": 98.64924712134632,
      "grad_norm": 0.24752505123615265,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 111375
    },
    {
      "epoch": 98.65367581930913,
      "grad_norm": 0.24248725175857544,
      "learning_rate": 1e-05,
      "loss": 0.9581,
      "step": 111380
    },
    {
      "epoch": 98.65810451727192,
      "grad_norm": 0.26691681146621704,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 111385
    },
    {
      "epoch": 98.66253321523472,
      "grad_norm": 0.2707720994949341,
      "learning_rate": 1e-05,
      "loss": 0.9426,
      "step": 111390
    },
    {
      "epoch": 98.66696191319753,
      "grad_norm": 0.25114381313323975,
      "learning_rate": 1e-05,
      "loss": 0.9844,
      "step": 111395
    },
    {
      "epoch": 98.67139061116032,
      "grad_norm": 0.23242361843585968,
      "learning_rate": 1e-05,
      "loss": 0.9653,
      "step": 111400
    },
    {
      "epoch": 98.67581930912311,
      "grad_norm": 0.25180405378341675,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 111405
    },
    {
      "epoch": 98.68024800708592,
      "grad_norm": 0.21613353490829468,
      "learning_rate": 1e-05,
      "loss": 0.8961,
      "step": 111410
    },
    {
      "epoch": 98.68467670504872,
      "grad_norm": 0.274187296628952,
      "learning_rate": 1e-05,
      "loss": 0.9182,
      "step": 111415
    },
    {
      "epoch": 98.68910540301151,
      "grad_norm": 0.24361048638820648,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 111420
    },
    {
      "epoch": 98.69353410097432,
      "grad_norm": 0.2177826166152954,
      "learning_rate": 1e-05,
      "loss": 0.9584,
      "step": 111425
    },
    {
      "epoch": 98.69796279893711,
      "grad_norm": 0.2366824746131897,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 111430
    },
    {
      "epoch": 98.7023914968999,
      "grad_norm": 0.2475898712873459,
      "learning_rate": 1e-05,
      "loss": 1.0256,
      "step": 111435
    },
    {
      "epoch": 98.70682019486271,
      "grad_norm": 0.25118812918663025,
      "learning_rate": 1e-05,
      "loss": 0.8879,
      "step": 111440
    },
    {
      "epoch": 98.71124889282551,
      "grad_norm": 0.23733948171138763,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 111445
    },
    {
      "epoch": 98.7156775907883,
      "grad_norm": 0.2434430718421936,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 111450
    },
    {
      "epoch": 98.72010628875111,
      "grad_norm": 0.24628238379955292,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 111455
    },
    {
      "epoch": 98.7245349867139,
      "grad_norm": 0.22682726383209229,
      "learning_rate": 1e-05,
      "loss": 0.9092,
      "step": 111460
    },
    {
      "epoch": 98.7289636846767,
      "grad_norm": 0.19881866872310638,
      "learning_rate": 1e-05,
      "loss": 0.9289,
      "step": 111465
    },
    {
      "epoch": 98.73339238263951,
      "grad_norm": 0.22952796518802643,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 111470
    },
    {
      "epoch": 98.7378210806023,
      "grad_norm": 0.23233212530612946,
      "learning_rate": 1e-05,
      "loss": 0.9109,
      "step": 111475
    },
    {
      "epoch": 98.7422497785651,
      "grad_norm": 0.22479300200939178,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 111480
    },
    {
      "epoch": 98.7466784765279,
      "grad_norm": 0.2576109766960144,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 111485
    },
    {
      "epoch": 98.7511071744907,
      "grad_norm": 0.2226061075925827,
      "learning_rate": 1e-05,
      "loss": 0.9762,
      "step": 111490
    },
    {
      "epoch": 98.75553587245349,
      "grad_norm": 0.25576964020729065,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 111495
    },
    {
      "epoch": 98.7599645704163,
      "grad_norm": 0.23511266708374023,
      "learning_rate": 1e-05,
      "loss": 0.9461,
      "step": 111500
    },
    {
      "epoch": 98.7643932683791,
      "grad_norm": 0.27317604422569275,
      "learning_rate": 1e-05,
      "loss": 1.0073,
      "step": 111505
    },
    {
      "epoch": 98.76882196634189,
      "grad_norm": 0.2240634560585022,
      "learning_rate": 1e-05,
      "loss": 0.9605,
      "step": 111510
    },
    {
      "epoch": 98.7732506643047,
      "grad_norm": 0.2444314807653427,
      "learning_rate": 1e-05,
      "loss": 0.9369,
      "step": 111515
    },
    {
      "epoch": 98.77767936226749,
      "grad_norm": 0.27551326155662537,
      "learning_rate": 1e-05,
      "loss": 0.9382,
      "step": 111520
    },
    {
      "epoch": 98.78210806023029,
      "grad_norm": 0.26869821548461914,
      "learning_rate": 1e-05,
      "loss": 0.9248,
      "step": 111525
    },
    {
      "epoch": 98.7865367581931,
      "grad_norm": 0.26883891224861145,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 111530
    },
    {
      "epoch": 98.79096545615589,
      "grad_norm": 0.2614055871963501,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 111535
    },
    {
      "epoch": 98.79539415411868,
      "grad_norm": 0.2303207814693451,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 111540
    },
    {
      "epoch": 98.79982285208149,
      "grad_norm": 0.2326149344444275,
      "learning_rate": 1e-05,
      "loss": 0.9146,
      "step": 111545
    },
    {
      "epoch": 98.80425155004428,
      "grad_norm": 0.2920328974723816,
      "learning_rate": 1e-05,
      "loss": 0.9668,
      "step": 111550
    },
    {
      "epoch": 98.8086802480071,
      "grad_norm": 0.23487339913845062,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 111555
    },
    {
      "epoch": 98.81310894596989,
      "grad_norm": 0.22838394343852997,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 111560
    },
    {
      "epoch": 98.81753764393268,
      "grad_norm": 0.2282736897468567,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 111565
    },
    {
      "epoch": 98.82196634189549,
      "grad_norm": 0.28193390369415283,
      "learning_rate": 1e-05,
      "loss": 0.9698,
      "step": 111570
    },
    {
      "epoch": 98.82639503985828,
      "grad_norm": 0.2746627926826477,
      "learning_rate": 1e-05,
      "loss": 0.9101,
      "step": 111575
    },
    {
      "epoch": 98.83082373782108,
      "grad_norm": 0.22043448686599731,
      "learning_rate": 1e-05,
      "loss": 0.9887,
      "step": 111580
    },
    {
      "epoch": 98.83525243578389,
      "grad_norm": 0.255452960729599,
      "learning_rate": 1e-05,
      "loss": 0.9613,
      "step": 111585
    },
    {
      "epoch": 98.83968113374668,
      "grad_norm": 0.24972356855869293,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 111590
    },
    {
      "epoch": 98.84410983170947,
      "grad_norm": 0.22377607226371765,
      "learning_rate": 1e-05,
      "loss": 0.9875,
      "step": 111595
    },
    {
      "epoch": 98.84853852967228,
      "grad_norm": 0.21315820515155792,
      "learning_rate": 1e-05,
      "loss": 0.9598,
      "step": 111600
    },
    {
      "epoch": 98.85296722763508,
      "grad_norm": 0.2499481439590454,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 111605
    },
    {
      "epoch": 98.85739592559787,
      "grad_norm": 0.2549728453159332,
      "learning_rate": 1e-05,
      "loss": 0.9467,
      "step": 111610
    },
    {
      "epoch": 98.86182462356068,
      "grad_norm": 0.24899548292160034,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 111615
    },
    {
      "epoch": 98.86625332152347,
      "grad_norm": 0.23861071467399597,
      "learning_rate": 1e-05,
      "loss": 0.9518,
      "step": 111620
    },
    {
      "epoch": 98.87068201948627,
      "grad_norm": 0.22890697419643402,
      "learning_rate": 1e-05,
      "loss": 0.9168,
      "step": 111625
    },
    {
      "epoch": 98.87511071744908,
      "grad_norm": 0.24206499755382538,
      "learning_rate": 1e-05,
      "loss": 0.926,
      "step": 111630
    },
    {
      "epoch": 98.87953941541187,
      "grad_norm": 0.28654757142066956,
      "learning_rate": 1e-05,
      "loss": 0.9368,
      "step": 111635
    },
    {
      "epoch": 98.88396811337466,
      "grad_norm": 0.26713982224464417,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 111640
    },
    {
      "epoch": 98.88839681133747,
      "grad_norm": 0.22974300384521484,
      "learning_rate": 1e-05,
      "loss": 0.9542,
      "step": 111645
    },
    {
      "epoch": 98.89282550930027,
      "grad_norm": 0.19291585683822632,
      "learning_rate": 1e-05,
      "loss": 0.9675,
      "step": 111650
    },
    {
      "epoch": 98.89725420726306,
      "grad_norm": 0.23455610871315002,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 111655
    },
    {
      "epoch": 98.90168290522587,
      "grad_norm": 0.2292855978012085,
      "learning_rate": 1e-05,
      "loss": 0.9755,
      "step": 111660
    },
    {
      "epoch": 98.90611160318866,
      "grad_norm": 0.24286925792694092,
      "learning_rate": 1e-05,
      "loss": 0.8819,
      "step": 111665
    },
    {
      "epoch": 98.91054030115146,
      "grad_norm": 0.2769407331943512,
      "learning_rate": 1e-05,
      "loss": 0.913,
      "step": 111670
    },
    {
      "epoch": 98.91496899911427,
      "grad_norm": 0.2178068310022354,
      "learning_rate": 1e-05,
      "loss": 0.9706,
      "step": 111675
    },
    {
      "epoch": 98.91939769707706,
      "grad_norm": 0.21476486325263977,
      "learning_rate": 1e-05,
      "loss": 0.9411,
      "step": 111680
    },
    {
      "epoch": 98.92382639503985,
      "grad_norm": 0.2701799273490906,
      "learning_rate": 1e-05,
      "loss": 0.9388,
      "step": 111685
    },
    {
      "epoch": 98.92825509300266,
      "grad_norm": 0.24470177292823792,
      "learning_rate": 1e-05,
      "loss": 0.9907,
      "step": 111690
    },
    {
      "epoch": 98.93268379096546,
      "grad_norm": 0.2537175714969635,
      "learning_rate": 1e-05,
      "loss": 0.9867,
      "step": 111695
    },
    {
      "epoch": 98.93711248892825,
      "grad_norm": 0.2429252713918686,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 111700
    },
    {
      "epoch": 98.94154118689106,
      "grad_norm": 0.23320193588733673,
      "learning_rate": 1e-05,
      "loss": 0.9957,
      "step": 111705
    },
    {
      "epoch": 98.94596988485385,
      "grad_norm": 0.22364431619644165,
      "learning_rate": 1e-05,
      "loss": 1.018,
      "step": 111710
    },
    {
      "epoch": 98.95039858281665,
      "grad_norm": 0.22956855595111847,
      "learning_rate": 1e-05,
      "loss": 0.9015,
      "step": 111715
    },
    {
      "epoch": 98.95482728077945,
      "grad_norm": 0.2372560352087021,
      "learning_rate": 1e-05,
      "loss": 0.9534,
      "step": 111720
    },
    {
      "epoch": 98.95925597874225,
      "grad_norm": 0.23301735520362854,
      "learning_rate": 1e-05,
      "loss": 0.9504,
      "step": 111725
    },
    {
      "epoch": 98.96368467670504,
      "grad_norm": 0.21001306176185608,
      "learning_rate": 1e-05,
      "loss": 0.9498,
      "step": 111730
    },
    {
      "epoch": 98.96811337466785,
      "grad_norm": 0.29205936193466187,
      "learning_rate": 1e-05,
      "loss": 0.9501,
      "step": 111735
    },
    {
      "epoch": 98.97254207263065,
      "grad_norm": 0.23656350374221802,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 111740
    },
    {
      "epoch": 98.97697077059344,
      "grad_norm": 0.2577022910118103,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 111745
    },
    {
      "epoch": 98.98139946855625,
      "grad_norm": 0.26302826404571533,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 111750
    },
    {
      "epoch": 98.98582816651904,
      "grad_norm": 0.23785418272018433,
      "learning_rate": 1e-05,
      "loss": 0.9703,
      "step": 111755
    },
    {
      "epoch": 98.99025686448184,
      "grad_norm": 0.25688380002975464,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 111760
    },
    {
      "epoch": 98.99468556244464,
      "grad_norm": 0.28728407621383667,
      "learning_rate": 1e-05,
      "loss": 0.9432,
      "step": 111765
    },
    {
      "epoch": 98.99911426040744,
      "grad_norm": 0.29873979091644287,
      "learning_rate": 1e-05,
      "loss": 0.9407,
      "step": 111770
    },
    {
      "epoch": 99.00354295837023,
      "grad_norm": 0.25227078795433044,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 111775
    },
    {
      "epoch": 99.00797165633304,
      "grad_norm": 0.23644119501113892,
      "learning_rate": 1e-05,
      "loss": 0.9748,
      "step": 111780
    },
    {
      "epoch": 99.01240035429583,
      "grad_norm": 0.2506011724472046,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 111785
    },
    {
      "epoch": 99.01682905225863,
      "grad_norm": 0.21115323901176453,
      "learning_rate": 1e-05,
      "loss": 0.9524,
      "step": 111790
    },
    {
      "epoch": 99.02125775022144,
      "grad_norm": 0.2390095293521881,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 111795
    },
    {
      "epoch": 99.02568644818423,
      "grad_norm": 0.22407297790050507,
      "learning_rate": 1e-05,
      "loss": 0.9548,
      "step": 111800
    },
    {
      "epoch": 99.03011514614704,
      "grad_norm": 0.23390407860279083,
      "learning_rate": 1e-05,
      "loss": 0.945,
      "step": 111805
    },
    {
      "epoch": 99.03454384410983,
      "grad_norm": 0.2735450565814972,
      "learning_rate": 1e-05,
      "loss": 0.9973,
      "step": 111810
    },
    {
      "epoch": 99.03897254207263,
      "grad_norm": 0.26146021485328674,
      "learning_rate": 1e-05,
      "loss": 0.9204,
      "step": 111815
    },
    {
      "epoch": 99.04340124003544,
      "grad_norm": 0.2431093007326126,
      "learning_rate": 1e-05,
      "loss": 0.941,
      "step": 111820
    },
    {
      "epoch": 99.04782993799823,
      "grad_norm": 0.20601999759674072,
      "learning_rate": 1e-05,
      "loss": 0.9473,
      "step": 111825
    },
    {
      "epoch": 99.05225863596102,
      "grad_norm": 0.20047248899936676,
      "learning_rate": 1e-05,
      "loss": 0.9909,
      "step": 111830
    },
    {
      "epoch": 99.05668733392383,
      "grad_norm": 0.21753501892089844,
      "learning_rate": 1e-05,
      "loss": 0.9322,
      "step": 111835
    },
    {
      "epoch": 99.06111603188663,
      "grad_norm": 0.22059334814548492,
      "learning_rate": 1e-05,
      "loss": 0.9626,
      "step": 111840
    },
    {
      "epoch": 99.06554472984942,
      "grad_norm": 0.24513216316699982,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 111845
    },
    {
      "epoch": 99.06997342781223,
      "grad_norm": 0.24849770963191986,
      "learning_rate": 1e-05,
      "loss": 0.9517,
      "step": 111850
    },
    {
      "epoch": 99.07440212577502,
      "grad_norm": 0.23517654836177826,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 111855
    },
    {
      "epoch": 99.07883082373782,
      "grad_norm": 0.2622279226779938,
      "learning_rate": 1e-05,
      "loss": 0.9167,
      "step": 111860
    },
    {
      "epoch": 99.08325952170063,
      "grad_norm": 0.27186161279678345,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 111865
    },
    {
      "epoch": 99.08768821966342,
      "grad_norm": 0.24114683270454407,
      "learning_rate": 1e-05,
      "loss": 0.9692,
      "step": 111870
    },
    {
      "epoch": 99.09211691762621,
      "grad_norm": 0.2248872071504593,
      "learning_rate": 1e-05,
      "loss": 0.9172,
      "step": 111875
    },
    {
      "epoch": 99.09654561558902,
      "grad_norm": 0.2622697353363037,
      "learning_rate": 1e-05,
      "loss": 0.9725,
      "step": 111880
    },
    {
      "epoch": 99.10097431355182,
      "grad_norm": 0.2753954231739044,
      "learning_rate": 1e-05,
      "loss": 0.9361,
      "step": 111885
    },
    {
      "epoch": 99.10540301151461,
      "grad_norm": 0.2478536069393158,
      "learning_rate": 1e-05,
      "loss": 0.9105,
      "step": 111890
    },
    {
      "epoch": 99.10983170947742,
      "grad_norm": 0.2597898840904236,
      "learning_rate": 1e-05,
      "loss": 0.9983,
      "step": 111895
    },
    {
      "epoch": 99.11426040744021,
      "grad_norm": 0.2667137682437897,
      "learning_rate": 1e-05,
      "loss": 0.9529,
      "step": 111900
    },
    {
      "epoch": 99.118689105403,
      "grad_norm": 0.21300847828388214,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 111905
    },
    {
      "epoch": 99.12311780336582,
      "grad_norm": 0.24673312902450562,
      "learning_rate": 1e-05,
      "loss": 0.9727,
      "step": 111910
    },
    {
      "epoch": 99.12754650132861,
      "grad_norm": 0.2489415407180786,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 111915
    },
    {
      "epoch": 99.1319751992914,
      "grad_norm": 0.2588604688644409,
      "learning_rate": 1e-05,
      "loss": 0.9682,
      "step": 111920
    },
    {
      "epoch": 99.13640389725421,
      "grad_norm": 0.22797806560993195,
      "learning_rate": 1e-05,
      "loss": 0.9179,
      "step": 111925
    },
    {
      "epoch": 99.140832595217,
      "grad_norm": 0.25321805477142334,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 111930
    },
    {
      "epoch": 99.1452612931798,
      "grad_norm": 0.23397018015384674,
      "learning_rate": 1e-05,
      "loss": 0.9544,
      "step": 111935
    },
    {
      "epoch": 99.14968999114261,
      "grad_norm": 0.23838578164577484,
      "learning_rate": 1e-05,
      "loss": 0.9282,
      "step": 111940
    },
    {
      "epoch": 99.1541186891054,
      "grad_norm": 0.23428472876548767,
      "learning_rate": 1e-05,
      "loss": 0.9677,
      "step": 111945
    },
    {
      "epoch": 99.1585473870682,
      "grad_norm": 0.22358359396457672,
      "learning_rate": 1e-05,
      "loss": 0.9785,
      "step": 111950
    },
    {
      "epoch": 99.162976085031,
      "grad_norm": 0.21769684553146362,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 111955
    },
    {
      "epoch": 99.1674047829938,
      "grad_norm": 0.23592406511306763,
      "learning_rate": 1e-05,
      "loss": 0.911,
      "step": 111960
    },
    {
      "epoch": 99.1718334809566,
      "grad_norm": 0.21641094982624054,
      "learning_rate": 1e-05,
      "loss": 0.9853,
      "step": 111965
    },
    {
      "epoch": 99.1762621789194,
      "grad_norm": 0.21236218512058258,
      "learning_rate": 1e-05,
      "loss": 0.9484,
      "step": 111970
    },
    {
      "epoch": 99.1806908768822,
      "grad_norm": 0.2305438369512558,
      "learning_rate": 1e-05,
      "loss": 0.9566,
      "step": 111975
    },
    {
      "epoch": 99.18511957484499,
      "grad_norm": 0.19450995326042175,
      "learning_rate": 1e-05,
      "loss": 0.9264,
      "step": 111980
    },
    {
      "epoch": 99.1895482728078,
      "grad_norm": 0.21670381724834442,
      "learning_rate": 1e-05,
      "loss": 0.9463,
      "step": 111985
    },
    {
      "epoch": 99.19397697077059,
      "grad_norm": 0.2027820497751236,
      "learning_rate": 1e-05,
      "loss": 0.9732,
      "step": 111990
    },
    {
      "epoch": 99.19840566873339,
      "grad_norm": 0.26259005069732666,
      "learning_rate": 1e-05,
      "loss": 0.9072,
      "step": 111995
    },
    {
      "epoch": 99.2028343666962,
      "grad_norm": 0.26879674196243286,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 112000
    },
    {
      "epoch": 99.20726306465899,
      "grad_norm": 0.278833270072937,
      "learning_rate": 1e-05,
      "loss": 0.9717,
      "step": 112005
    },
    {
      "epoch": 99.21169176262178,
      "grad_norm": 0.2328919917345047,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 112010
    },
    {
      "epoch": 99.21612046058459,
      "grad_norm": 0.22976909577846527,
      "learning_rate": 1e-05,
      "loss": 0.9472,
      "step": 112015
    },
    {
      "epoch": 99.22054915854739,
      "grad_norm": 0.23653466999530792,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 112020
    },
    {
      "epoch": 99.22497785651018,
      "grad_norm": 0.2941906452178955,
      "learning_rate": 1e-05,
      "loss": 0.9693,
      "step": 112025
    },
    {
      "epoch": 99.22940655447299,
      "grad_norm": 0.3187143802642822,
      "learning_rate": 1e-05,
      "loss": 0.9447,
      "step": 112030
    },
    {
      "epoch": 99.23383525243578,
      "grad_norm": 0.208413764834404,
      "learning_rate": 1e-05,
      "loss": 0.9776,
      "step": 112035
    },
    {
      "epoch": 99.23826395039858,
      "grad_norm": 0.2234906703233719,
      "learning_rate": 1e-05,
      "loss": 0.9477,
      "step": 112040
    },
    {
      "epoch": 99.24269264836138,
      "grad_norm": 0.21417900919914246,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 112045
    },
    {
      "epoch": 99.24712134632418,
      "grad_norm": 0.23505137860774994,
      "learning_rate": 1e-05,
      "loss": 0.9573,
      "step": 112050
    },
    {
      "epoch": 99.25155004428699,
      "grad_norm": 0.2561967670917511,
      "learning_rate": 1e-05,
      "loss": 0.9643,
      "step": 112055
    },
    {
      "epoch": 99.25597874224978,
      "grad_norm": 0.322628378868103,
      "learning_rate": 1e-05,
      "loss": 0.9659,
      "step": 112060
    },
    {
      "epoch": 99.26040744021257,
      "grad_norm": 0.263893723487854,
      "learning_rate": 1e-05,
      "loss": 1.0116,
      "step": 112065
    },
    {
      "epoch": 99.26483613817538,
      "grad_norm": 0.21519190073013306,
      "learning_rate": 1e-05,
      "loss": 0.9539,
      "step": 112070
    },
    {
      "epoch": 99.26926483613818,
      "grad_norm": 0.2471558153629303,
      "learning_rate": 1e-05,
      "loss": 0.9166,
      "step": 112075
    },
    {
      "epoch": 99.27369353410097,
      "grad_norm": 0.2374553233385086,
      "learning_rate": 1e-05,
      "loss": 0.9997,
      "step": 112080
    },
    {
      "epoch": 99.27812223206378,
      "grad_norm": 0.21322056651115417,
      "learning_rate": 1e-05,
      "loss": 0.9087,
      "step": 112085
    },
    {
      "epoch": 99.28255093002657,
      "grad_norm": 0.22902484238147736,
      "learning_rate": 1e-05,
      "loss": 0.9987,
      "step": 112090
    },
    {
      "epoch": 99.28697962798937,
      "grad_norm": 0.25176161527633667,
      "learning_rate": 1e-05,
      "loss": 0.9427,
      "step": 112095
    },
    {
      "epoch": 99.29140832595218,
      "grad_norm": 0.27048763632774353,
      "learning_rate": 1e-05,
      "loss": 0.9079,
      "step": 112100
    },
    {
      "epoch": 99.29583702391497,
      "grad_norm": 0.21228544414043427,
      "learning_rate": 1e-05,
      "loss": 0.928,
      "step": 112105
    },
    {
      "epoch": 99.30026572187776,
      "grad_norm": 0.23203589022159576,
      "learning_rate": 1e-05,
      "loss": 0.9279,
      "step": 112110
    },
    {
      "epoch": 99.30469441984057,
      "grad_norm": 0.24339962005615234,
      "learning_rate": 1e-05,
      "loss": 0.9438,
      "step": 112115
    },
    {
      "epoch": 99.30912311780337,
      "grad_norm": 0.25905856490135193,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 112120
    },
    {
      "epoch": 99.31355181576616,
      "grad_norm": 0.25077489018440247,
      "learning_rate": 1e-05,
      "loss": 0.9287,
      "step": 112125
    },
    {
      "epoch": 99.31798051372897,
      "grad_norm": 0.2527613937854767,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 112130
    },
    {
      "epoch": 99.32240921169176,
      "grad_norm": 0.2319108545780182,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 112135
    },
    {
      "epoch": 99.32683790965456,
      "grad_norm": 0.21724678575992584,
      "learning_rate": 1e-05,
      "loss": 0.936,
      "step": 112140
    },
    {
      "epoch": 99.33126660761737,
      "grad_norm": 0.2282632291316986,
      "learning_rate": 1e-05,
      "loss": 0.9216,
      "step": 112145
    },
    {
      "epoch": 99.33569530558016,
      "grad_norm": 0.27804258465766907,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 112150
    },
    {
      "epoch": 99.34012400354295,
      "grad_norm": 0.2042091190814972,
      "learning_rate": 1e-05,
      "loss": 0.895,
      "step": 112155
    },
    {
      "epoch": 99.34455270150576,
      "grad_norm": 0.23103472590446472,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 112160
    },
    {
      "epoch": 99.34898139946856,
      "grad_norm": 0.24889393150806427,
      "learning_rate": 1e-05,
      "loss": 0.9139,
      "step": 112165
    },
    {
      "epoch": 99.35341009743135,
      "grad_norm": 0.2449347823858261,
      "learning_rate": 1e-05,
      "loss": 0.9778,
      "step": 112170
    },
    {
      "epoch": 99.35783879539416,
      "grad_norm": 0.22189085185527802,
      "learning_rate": 1e-05,
      "loss": 0.9187,
      "step": 112175
    },
    {
      "epoch": 99.36226749335695,
      "grad_norm": 0.2285684198141098,
      "learning_rate": 1e-05,
      "loss": 0.996,
      "step": 112180
    },
    {
      "epoch": 99.36669619131975,
      "grad_norm": 0.21446222066879272,
      "learning_rate": 1e-05,
      "loss": 0.9572,
      "step": 112185
    },
    {
      "epoch": 99.37112488928256,
      "grad_norm": 0.2525879144668579,
      "learning_rate": 1e-05,
      "loss": 0.9396,
      "step": 112190
    },
    {
      "epoch": 99.37555358724535,
      "grad_norm": 0.2445988804101944,
      "learning_rate": 1e-05,
      "loss": 0.8767,
      "step": 112195
    },
    {
      "epoch": 99.37998228520814,
      "grad_norm": 0.21096929907798767,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 112200
    },
    {
      "epoch": 99.38441098317095,
      "grad_norm": 0.24791206419467926,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 112205
    },
    {
      "epoch": 99.38883968113375,
      "grad_norm": 0.21883061528205872,
      "learning_rate": 1e-05,
      "loss": 0.9689,
      "step": 112210
    },
    {
      "epoch": 99.39326837909654,
      "grad_norm": 0.23737597465515137,
      "learning_rate": 1e-05,
      "loss": 0.908,
      "step": 112215
    },
    {
      "epoch": 99.39769707705935,
      "grad_norm": 0.2720600664615631,
      "learning_rate": 1e-05,
      "loss": 0.9186,
      "step": 112220
    },
    {
      "epoch": 99.40212577502214,
      "grad_norm": 0.21748831868171692,
      "learning_rate": 1e-05,
      "loss": 1.002,
      "step": 112225
    },
    {
      "epoch": 99.40655447298494,
      "grad_norm": 0.23587162792682648,
      "learning_rate": 1e-05,
      "loss": 0.9685,
      "step": 112230
    },
    {
      "epoch": 99.41098317094774,
      "grad_norm": 0.22226890921592712,
      "learning_rate": 1e-05,
      "loss": 0.9153,
      "step": 112235
    },
    {
      "epoch": 99.41541186891054,
      "grad_norm": 0.2393627166748047,
      "learning_rate": 1e-05,
      "loss": 0.8845,
      "step": 112240
    },
    {
      "epoch": 99.41984056687333,
      "grad_norm": 0.2564423978328705,
      "learning_rate": 1e-05,
      "loss": 0.9091,
      "step": 112245
    },
    {
      "epoch": 99.42426926483614,
      "grad_norm": 0.22174178063869476,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 112250
    },
    {
      "epoch": 99.42869796279894,
      "grad_norm": 0.23430350422859192,
      "learning_rate": 1e-05,
      "loss": 0.9137,
      "step": 112255
    },
    {
      "epoch": 99.43312666076173,
      "grad_norm": 0.22487905621528625,
      "learning_rate": 1e-05,
      "loss": 0.9852,
      "step": 112260
    },
    {
      "epoch": 99.43755535872454,
      "grad_norm": 0.22497770190238953,
      "learning_rate": 1e-05,
      "loss": 0.8792,
      "step": 112265
    },
    {
      "epoch": 99.44198405668733,
      "grad_norm": 0.23379693925380707,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 112270
    },
    {
      "epoch": 99.44641275465013,
      "grad_norm": 0.22368568181991577,
      "learning_rate": 1e-05,
      "loss": 0.964,
      "step": 112275
    },
    {
      "epoch": 99.45084145261293,
      "grad_norm": 0.2857246994972229,
      "learning_rate": 1e-05,
      "loss": 0.9872,
      "step": 112280
    },
    {
      "epoch": 99.45527015057573,
      "grad_norm": 0.22969003021717072,
      "learning_rate": 1e-05,
      "loss": 0.9663,
      "step": 112285
    },
    {
      "epoch": 99.45969884853854,
      "grad_norm": 0.2787075936794281,
      "learning_rate": 1e-05,
      "loss": 0.9694,
      "step": 112290
    },
    {
      "epoch": 99.46412754650133,
      "grad_norm": 0.22288420796394348,
      "learning_rate": 1e-05,
      "loss": 0.9825,
      "step": 112295
    },
    {
      "epoch": 99.46855624446412,
      "grad_norm": 0.234920933842659,
      "learning_rate": 1e-05,
      "loss": 1.0295,
      "step": 112300
    },
    {
      "epoch": 99.47298494242693,
      "grad_norm": 0.23380663990974426,
      "learning_rate": 1e-05,
      "loss": 0.9801,
      "step": 112305
    },
    {
      "epoch": 99.47741364038973,
      "grad_norm": 0.216245636343956,
      "learning_rate": 1e-05,
      "loss": 0.9718,
      "step": 112310
    },
    {
      "epoch": 99.48184233835252,
      "grad_norm": 0.19571220874786377,
      "learning_rate": 1e-05,
      "loss": 0.9275,
      "step": 112315
    },
    {
      "epoch": 99.48627103631533,
      "grad_norm": 0.29585444927215576,
      "learning_rate": 1e-05,
      "loss": 0.9914,
      "step": 112320
    },
    {
      "epoch": 99.49069973427812,
      "grad_norm": 0.2280639111995697,
      "learning_rate": 1e-05,
      "loss": 0.9618,
      "step": 112325
    },
    {
      "epoch": 99.49512843224092,
      "grad_norm": 0.24458849430084229,
      "learning_rate": 1e-05,
      "loss": 0.9298,
      "step": 112330
    },
    {
      "epoch": 99.49955713020373,
      "grad_norm": 0.2576400339603424,
      "learning_rate": 1e-05,
      "loss": 0.9656,
      "step": 112335
    },
    {
      "epoch": 99.50398582816652,
      "grad_norm": 0.28934964537620544,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 112340
    },
    {
      "epoch": 99.50841452612931,
      "grad_norm": 0.33302006125450134,
      "learning_rate": 1e-05,
      "loss": 0.9409,
      "step": 112345
    },
    {
      "epoch": 99.51284322409212,
      "grad_norm": 0.26159700751304626,
      "learning_rate": 1e-05,
      "loss": 0.907,
      "step": 112350
    },
    {
      "epoch": 99.51727192205492,
      "grad_norm": 0.22977440059185028,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 112355
    },
    {
      "epoch": 99.52170062001771,
      "grad_norm": 0.23559938371181488,
      "learning_rate": 1e-05,
      "loss": 0.9889,
      "step": 112360
    },
    {
      "epoch": 99.52612931798052,
      "grad_norm": 0.24109657108783722,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 112365
    },
    {
      "epoch": 99.53055801594331,
      "grad_norm": 0.252225786447525,
      "learning_rate": 1e-05,
      "loss": 0.8816,
      "step": 112370
    },
    {
      "epoch": 99.53498671390611,
      "grad_norm": 0.23753182590007782,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 112375
    },
    {
      "epoch": 99.53941541186892,
      "grad_norm": 0.25965628027915955,
      "learning_rate": 1e-05,
      "loss": 0.9302,
      "step": 112380
    },
    {
      "epoch": 99.54384410983171,
      "grad_norm": 0.246441051363945,
      "learning_rate": 1e-05,
      "loss": 0.9417,
      "step": 112385
    },
    {
      "epoch": 99.5482728077945,
      "grad_norm": 0.25906872749328613,
      "learning_rate": 1e-05,
      "loss": 0.8966,
      "step": 112390
    },
    {
      "epoch": 99.55270150575731,
      "grad_norm": 0.2492685765028,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 112395
    },
    {
      "epoch": 99.5571302037201,
      "grad_norm": 0.22617076337337494,
      "learning_rate": 1e-05,
      "loss": 0.8938,
      "step": 112400
    },
    {
      "epoch": 99.5615589016829,
      "grad_norm": 0.20683042705059052,
      "learning_rate": 1e-05,
      "loss": 0.9327,
      "step": 112405
    },
    {
      "epoch": 99.56598759964571,
      "grad_norm": 0.24301917850971222,
      "learning_rate": 1e-05,
      "loss": 0.9243,
      "step": 112410
    },
    {
      "epoch": 99.5704162976085,
      "grad_norm": 0.23414640128612518,
      "learning_rate": 1e-05,
      "loss": 0.9106,
      "step": 112415
    },
    {
      "epoch": 99.5748449955713,
      "grad_norm": 0.22861480712890625,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 112420
    },
    {
      "epoch": 99.5792736935341,
      "grad_norm": 0.227803036570549,
      "learning_rate": 1e-05,
      "loss": 0.9774,
      "step": 112425
    },
    {
      "epoch": 99.5837023914969,
      "grad_norm": 0.2459019273519516,
      "learning_rate": 1e-05,
      "loss": 0.925,
      "step": 112430
    },
    {
      "epoch": 99.5881310894597,
      "grad_norm": 0.21924693882465363,
      "learning_rate": 1e-05,
      "loss": 0.9497,
      "step": 112435
    },
    {
      "epoch": 99.5925597874225,
      "grad_norm": 0.21996010839939117,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 112440
    },
    {
      "epoch": 99.5969884853853,
      "grad_norm": 0.20834368467330933,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 112445
    },
    {
      "epoch": 99.60141718334809,
      "grad_norm": 0.274662047624588,
      "learning_rate": 1e-05,
      "loss": 0.9511,
      "step": 112450
    },
    {
      "epoch": 99.6058458813109,
      "grad_norm": 0.22213467955589294,
      "learning_rate": 1e-05,
      "loss": 0.9247,
      "step": 112455
    },
    {
      "epoch": 99.61027457927369,
      "grad_norm": 0.26723361015319824,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 112460
    },
    {
      "epoch": 99.61470327723649,
      "grad_norm": 0.20168496668338776,
      "learning_rate": 1e-05,
      "loss": 0.9796,
      "step": 112465
    },
    {
      "epoch": 99.6191319751993,
      "grad_norm": 0.24060934782028198,
      "learning_rate": 1e-05,
      "loss": 0.9719,
      "step": 112470
    },
    {
      "epoch": 99.62356067316209,
      "grad_norm": 0.24771174788475037,
      "learning_rate": 1e-05,
      "loss": 0.884,
      "step": 112475
    },
    {
      "epoch": 99.62798937112488,
      "grad_norm": 0.24702011048793793,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 112480
    },
    {
      "epoch": 99.63241806908769,
      "grad_norm": 0.20212148129940033,
      "learning_rate": 1e-05,
      "loss": 0.885,
      "step": 112485
    },
    {
      "epoch": 99.63684676705049,
      "grad_norm": 0.2540283501148224,
      "learning_rate": 1e-05,
      "loss": 0.9958,
      "step": 112490
    },
    {
      "epoch": 99.64127546501328,
      "grad_norm": 0.23832114040851593,
      "learning_rate": 1e-05,
      "loss": 0.9773,
      "step": 112495
    },
    {
      "epoch": 99.64570416297609,
      "grad_norm": 0.273372083902359,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 112500
    },
    {
      "epoch": 99.65013286093888,
      "grad_norm": 0.2506163418292999,
      "learning_rate": 1e-05,
      "loss": 0.9893,
      "step": 112505
    },
    {
      "epoch": 99.65456155890168,
      "grad_norm": 0.23988185822963715,
      "learning_rate": 1e-05,
      "loss": 0.9775,
      "step": 112510
    },
    {
      "epoch": 99.65899025686448,
      "grad_norm": 0.2652115821838379,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 112515
    },
    {
      "epoch": 99.66341895482728,
      "grad_norm": 0.25321099162101746,
      "learning_rate": 1e-05,
      "loss": 0.9647,
      "step": 112520
    },
    {
      "epoch": 99.66784765279007,
      "grad_norm": 0.2707461714744568,
      "learning_rate": 1e-05,
      "loss": 0.9669,
      "step": 112525
    },
    {
      "epoch": 99.67227635075288,
      "grad_norm": 0.3014930784702301,
      "learning_rate": 1e-05,
      "loss": 0.9861,
      "step": 112530
    },
    {
      "epoch": 99.67670504871568,
      "grad_norm": 0.26914748549461365,
      "learning_rate": 1e-05,
      "loss": 0.9641,
      "step": 112535
    },
    {
      "epoch": 99.68113374667848,
      "grad_norm": 0.25054702162742615,
      "learning_rate": 1e-05,
      "loss": 1.0029,
      "step": 112540
    },
    {
      "epoch": 99.68556244464128,
      "grad_norm": 0.22802293300628662,
      "learning_rate": 1e-05,
      "loss": 0.9743,
      "step": 112545
    },
    {
      "epoch": 99.68999114260407,
      "grad_norm": 0.2554066777229309,
      "learning_rate": 1e-05,
      "loss": 0.9179,
      "step": 112550
    },
    {
      "epoch": 99.69441984056688,
      "grad_norm": 0.22374007105827332,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 112555
    },
    {
      "epoch": 99.69884853852967,
      "grad_norm": 0.21790727972984314,
      "learning_rate": 1e-05,
      "loss": 0.915,
      "step": 112560
    },
    {
      "epoch": 99.70327723649247,
      "grad_norm": 0.2810305655002594,
      "learning_rate": 1e-05,
      "loss": 0.9575,
      "step": 112565
    },
    {
      "epoch": 99.70770593445528,
      "grad_norm": 0.21818692982196808,
      "learning_rate": 1e-05,
      "loss": 0.9254,
      "step": 112570
    },
    {
      "epoch": 99.71213463241807,
      "grad_norm": 0.22811481356620789,
      "learning_rate": 1e-05,
      "loss": 0.9336,
      "step": 112575
    },
    {
      "epoch": 99.71656333038086,
      "grad_norm": 0.21748724579811096,
      "learning_rate": 1e-05,
      "loss": 0.9446,
      "step": 112580
    },
    {
      "epoch": 99.72099202834367,
      "grad_norm": 0.23890122771263123,
      "learning_rate": 1e-05,
      "loss": 0.9124,
      "step": 112585
    },
    {
      "epoch": 99.72542072630647,
      "grad_norm": 0.23146294057369232,
      "learning_rate": 1e-05,
      "loss": 1.0018,
      "step": 112590
    },
    {
      "epoch": 99.72984942426926,
      "grad_norm": 0.22594179213047028,
      "learning_rate": 1e-05,
      "loss": 0.9204,
      "step": 112595
    },
    {
      "epoch": 99.73427812223207,
      "grad_norm": 0.26333120465278625,
      "learning_rate": 1e-05,
      "loss": 0.9954,
      "step": 112600
    },
    {
      "epoch": 99.73870682019486,
      "grad_norm": 0.2790357768535614,
      "learning_rate": 1e-05,
      "loss": 0.9807,
      "step": 112605
    },
    {
      "epoch": 99.74313551815766,
      "grad_norm": 0.24205732345581055,
      "learning_rate": 1e-05,
      "loss": 0.9401,
      "step": 112610
    },
    {
      "epoch": 99.74756421612047,
      "grad_norm": 0.2575497031211853,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 112615
    },
    {
      "epoch": 99.75199291408326,
      "grad_norm": 0.26168960332870483,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 112620
    },
    {
      "epoch": 99.75642161204605,
      "grad_norm": 0.25716209411621094,
      "learning_rate": 1e-05,
      "loss": 0.919,
      "step": 112625
    },
    {
      "epoch": 99.76085031000886,
      "grad_norm": 0.24601739645004272,
      "learning_rate": 1e-05,
      "loss": 1.0241,
      "step": 112630
    },
    {
      "epoch": 99.76527900797166,
      "grad_norm": 0.2519760727882385,
      "learning_rate": 1e-05,
      "loss": 0.9403,
      "step": 112635
    },
    {
      "epoch": 99.76970770593445,
      "grad_norm": 0.23660080134868622,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 112640
    },
    {
      "epoch": 99.77413640389726,
      "grad_norm": 0.2316771298646927,
      "learning_rate": 1e-05,
      "loss": 0.9763,
      "step": 112645
    },
    {
      "epoch": 99.77856510186005,
      "grad_norm": 0.208352729678154,
      "learning_rate": 1e-05,
      "loss": 0.953,
      "step": 112650
    },
    {
      "epoch": 99.78299379982285,
      "grad_norm": 0.1979127824306488,
      "learning_rate": 1e-05,
      "loss": 0.9153,
      "step": 112655
    },
    {
      "epoch": 99.78742249778566,
      "grad_norm": 0.2557685971260071,
      "learning_rate": 1e-05,
      "loss": 0.937,
      "step": 112660
    },
    {
      "epoch": 99.79185119574845,
      "grad_norm": 0.26532456278800964,
      "learning_rate": 1e-05,
      "loss": 0.9622,
      "step": 112665
    },
    {
      "epoch": 99.79627989371124,
      "grad_norm": 0.274320125579834,
      "learning_rate": 1e-05,
      "loss": 0.9422,
      "step": 112670
    },
    {
      "epoch": 99.80070859167405,
      "grad_norm": 0.21255089342594147,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 112675
    },
    {
      "epoch": 99.80513728963685,
      "grad_norm": 0.2488599419593811,
      "learning_rate": 1e-05,
      "loss": 0.9351,
      "step": 112680
    },
    {
      "epoch": 99.80956598759964,
      "grad_norm": 0.2480822503566742,
      "learning_rate": 1e-05,
      "loss": 0.8982,
      "step": 112685
    },
    {
      "epoch": 99.81399468556245,
      "grad_norm": 0.21427126228809357,
      "learning_rate": 1e-05,
      "loss": 0.9398,
      "step": 112690
    },
    {
      "epoch": 99.81842338352524,
      "grad_norm": 0.24526908993721008,
      "learning_rate": 1e-05,
      "loss": 0.9645,
      "step": 112695
    },
    {
      "epoch": 99.82285208148804,
      "grad_norm": 0.2674933671951294,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 112700
    },
    {
      "epoch": 99.82728077945085,
      "grad_norm": 0.23999366164207458,
      "learning_rate": 1e-05,
      "loss": 0.9212,
      "step": 112705
    },
    {
      "epoch": 99.83170947741364,
      "grad_norm": 0.2285846322774887,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 112710
    },
    {
      "epoch": 99.83613817537643,
      "grad_norm": 0.2737208306789398,
      "learning_rate": 1e-05,
      "loss": 0.946,
      "step": 112715
    },
    {
      "epoch": 99.84056687333924,
      "grad_norm": 0.23447902500629425,
      "learning_rate": 1e-05,
      "loss": 0.9794,
      "step": 112720
    },
    {
      "epoch": 99.84499557130204,
      "grad_norm": 0.22516512870788574,
      "learning_rate": 1e-05,
      "loss": 0.9601,
      "step": 112725
    },
    {
      "epoch": 99.84942426926483,
      "grad_norm": 0.24350744485855103,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 112730
    },
    {
      "epoch": 99.85385296722764,
      "grad_norm": 0.23664674162864685,
      "learning_rate": 1e-05,
      "loss": 0.9145,
      "step": 112735
    },
    {
      "epoch": 99.85828166519043,
      "grad_norm": 0.22188521921634674,
      "learning_rate": 1e-05,
      "loss": 0.9249,
      "step": 112740
    },
    {
      "epoch": 99.86271036315323,
      "grad_norm": 0.23321318626403809,
      "learning_rate": 1e-05,
      "loss": 0.9664,
      "step": 112745
    },
    {
      "epoch": 99.86713906111603,
      "grad_norm": 0.19657748937606812,
      "learning_rate": 1e-05,
      "loss": 0.9016,
      "step": 112750
    },
    {
      "epoch": 99.87156775907883,
      "grad_norm": 0.22899463772773743,
      "learning_rate": 1e-05,
      "loss": 0.9513,
      "step": 112755
    },
    {
      "epoch": 99.87599645704162,
      "grad_norm": 0.21229495108127594,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 112760
    },
    {
      "epoch": 99.88042515500443,
      "grad_norm": 0.21518702805042267,
      "learning_rate": 1e-05,
      "loss": 0.897,
      "step": 112765
    },
    {
      "epoch": 99.88485385296723,
      "grad_norm": 0.22340236604213715,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 112770
    },
    {
      "epoch": 99.88928255093003,
      "grad_norm": 0.2334798276424408,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 112775
    },
    {
      "epoch": 99.89371124889283,
      "grad_norm": 0.24507346749305725,
      "learning_rate": 1e-05,
      "loss": 0.9744,
      "step": 112780
    },
    {
      "epoch": 99.89813994685562,
      "grad_norm": 0.23902076482772827,
      "learning_rate": 1e-05,
      "loss": 0.957,
      "step": 112785
    },
    {
      "epoch": 99.90256864481843,
      "grad_norm": 0.19866837561130524,
      "learning_rate": 1e-05,
      "loss": 0.9494,
      "step": 112790
    },
    {
      "epoch": 99.90699734278122,
      "grad_norm": 0.198320671916008,
      "learning_rate": 1e-05,
      "loss": 0.9916,
      "step": 112795
    },
    {
      "epoch": 99.91142604074402,
      "grad_norm": 0.23699766397476196,
      "learning_rate": 1e-05,
      "loss": 0.9839,
      "step": 112800
    },
    {
      "epoch": 99.91585473870683,
      "grad_norm": 0.22235526144504547,
      "learning_rate": 1e-05,
      "loss": 0.9812,
      "step": 112805
    },
    {
      "epoch": 99.92028343666962,
      "grad_norm": 0.2261887937784195,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 112810
    },
    {
      "epoch": 99.92471213463241,
      "grad_norm": 0.296207457780838,
      "learning_rate": 1e-05,
      "loss": 0.9708,
      "step": 112815
    },
    {
      "epoch": 99.92914083259522,
      "grad_norm": 0.2781381607055664,
      "learning_rate": 1e-05,
      "loss": 0.9372,
      "step": 112820
    },
    {
      "epoch": 99.93356953055802,
      "grad_norm": 0.27711695432662964,
      "learning_rate": 1e-05,
      "loss": 0.8998,
      "step": 112825
    },
    {
      "epoch": 99.93799822852081,
      "grad_norm": 0.2417774349451065,
      "learning_rate": 1e-05,
      "loss": 0.9276,
      "step": 112830
    },
    {
      "epoch": 99.94242692648362,
      "grad_norm": 0.2478482574224472,
      "learning_rate": 1e-05,
      "loss": 0.9142,
      "step": 112835
    },
    {
      "epoch": 99.94685562444641,
      "grad_norm": 0.2155621200799942,
      "learning_rate": 1e-05,
      "loss": 0.9532,
      "step": 112840
    },
    {
      "epoch": 99.95128432240921,
      "grad_norm": 0.23429962992668152,
      "learning_rate": 1e-05,
      "loss": 0.986,
      "step": 112845
    },
    {
      "epoch": 99.95571302037202,
      "grad_norm": 0.24005959928035736,
      "learning_rate": 1e-05,
      "loss": 0.9488,
      "step": 112850
    },
    {
      "epoch": 99.96014171833481,
      "grad_norm": 0.262634813785553,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 112855
    },
    {
      "epoch": 99.9645704162976,
      "grad_norm": 0.17670851945877075,
      "learning_rate": 1e-05,
      "loss": 0.9217,
      "step": 112860
    },
    {
      "epoch": 99.96899911426041,
      "grad_norm": 0.21186046302318573,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 112865
    },
    {
      "epoch": 99.9734278122232,
      "grad_norm": 0.23945198953151703,
      "learning_rate": 1e-05,
      "loss": 0.9768,
      "step": 112870
    },
    {
      "epoch": 99.977856510186,
      "grad_norm": 0.22392725944519043,
      "learning_rate": 1e-05,
      "loss": 0.9533,
      "step": 112875
    },
    {
      "epoch": 99.98228520814881,
      "grad_norm": 0.2557377219200134,
      "learning_rate": 1e-05,
      "loss": 0.9783,
      "step": 112880
    },
    {
      "epoch": 99.9867139061116,
      "grad_norm": 0.23206056654453278,
      "learning_rate": 1e-05,
      "loss": 1.0533,
      "step": 112885
    },
    {
      "epoch": 99.9911426040744,
      "grad_norm": 0.23726427555084229,
      "learning_rate": 1e-05,
      "loss": 0.9674,
      "step": 112890
    },
    {
      "epoch": 99.9955713020372,
      "grad_norm": 0.2533961832523346,
      "learning_rate": 1e-05,
      "loss": 0.9782,
      "step": 112895
    },
    {
      "epoch": 100.0,
      "grad_norm": 0.25743329524993896,
      "learning_rate": 1e-05,
      "loss": 0.9841,
      "step": 112900
    },
    {
      "epoch": 100.0,
      "step": 112900,
      "total_flos": 1.1509983892450836e+19,
      "train_loss": 0.9787479164116778,
      "train_runtime": 471297.9641,
      "train_samples_per_second": 30.672,
      "train_steps_per_second": 0.24
    }
  ],
  "logging_steps": 5,
  "max_steps": 112900,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 100,
  "save_steps": 12000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1509983892450836e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}