{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.4040126536763131,
  "eval_steps": 500,
  "global_step": 100000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 4.040126536763131e-05,
      "grad_norm": 34776792.0,
      "learning_rate": 1.6e-08,
      "loss": 1112513.0,
      "step": 10
    },
    {
      "epoch": 8.080253073526263e-05,
      "grad_norm": 8441178.0,
      "learning_rate": 3.2e-08,
      "loss": 1301709.2,
      "step": 20
    },
    {
      "epoch": 0.00012120379610289395,
      "grad_norm": 12542169.0,
      "learning_rate": 4.8e-08,
      "loss": 1175913.9,
      "step": 30
    },
    {
      "epoch": 0.00016160506147052525,
      "grad_norm": 42386256.0,
      "learning_rate": 6.4e-08,
      "loss": 1313873.9,
      "step": 40
    },
    {
      "epoch": 0.00020200632683815657,
      "grad_norm": 33162978.0,
      "learning_rate": 8e-08,
      "loss": 1142017.2,
      "step": 50
    },
    {
      "epoch": 0.0002424075922057879,
      "grad_norm": 52537692.0,
      "learning_rate": 9.6e-08,
      "loss": 1040597.1,
      "step": 60
    },
    {
      "epoch": 0.0002828088575734192,
      "grad_norm": 17375308.0,
      "learning_rate": 1.1200000000000001e-07,
      "loss": 1494345.4,
      "step": 70
    },
    {
      "epoch": 0.0003232101229410505,
      "grad_norm": 95491040.0,
      "learning_rate": 1.28e-07,
      "loss": 1404674.8,
      "step": 80
    },
    {
      "epoch": 0.0003636113883086818,
      "grad_norm": 13118159.0,
      "learning_rate": 1.4400000000000002e-07,
      "loss": 1204737.6,
      "step": 90
    },
    {
      "epoch": 0.00040401265367631315,
      "grad_norm": 18581922.0,
      "learning_rate": 1.6e-07,
      "loss": 783122.2,
      "step": 100
    },
    {
      "epoch": 0.00044441391904394446,
      "grad_norm": 15600268.0,
      "learning_rate": 1.7600000000000001e-07,
      "loss": 1342865.9,
      "step": 110
    },
    {
      "epoch": 0.0004848151844115758,
      "grad_norm": 5757326.5,
      "learning_rate": 1.92e-07,
      "loss": 998477.5,
      "step": 120
    },
    {
      "epoch": 0.000525216449779207,
      "grad_norm": 37603152.0,
      "learning_rate": 2.08e-07,
      "loss": 1113327.2,
      "step": 130
    },
    {
      "epoch": 0.0005656177151468384,
      "grad_norm": 3987340.5,
      "learning_rate": 2.2400000000000002e-07,
      "loss": 815339.55,
      "step": 140
    },
    {
      "epoch": 0.0006060189805144697,
      "grad_norm": 17012036.0,
      "learning_rate": 2.4000000000000003e-07,
      "loss": 799237.55,
      "step": 150
    },
    {
      "epoch": 0.000646420245882101,
      "grad_norm": 15123844.0,
      "learning_rate": 2.56e-07,
      "loss": 795830.9,
      "step": 160
    },
    {
      "epoch": 0.0006868215112497323,
      "grad_norm": 13699358.0,
      "learning_rate": 2.72e-07,
      "loss": 675074.05,
      "step": 170
    },
    {
      "epoch": 0.0007272227766173637,
      "grad_norm": 5762779.5,
      "learning_rate": 2.8800000000000004e-07,
      "loss": 397814.875,
      "step": 180
    },
    {
      "epoch": 0.000767624041984995,
      "grad_norm": 3545711.25,
      "learning_rate": 3.04e-07,
      "loss": 278445.225,
      "step": 190
    },
    {
      "epoch": 0.0008080253073526263,
      "grad_norm": 22483670.0,
      "learning_rate": 3.2e-07,
      "loss": 350395.65,
      "step": 200
    },
    {
      "epoch": 0.0008484265727202576,
      "grad_norm": 3111300.75,
      "learning_rate": 3.36e-07,
      "loss": 215067.95,
      "step": 210
    },
    {
      "epoch": 0.0008888278380878889,
      "grad_norm": 4985113.0,
      "learning_rate": 3.5200000000000003e-07,
      "loss": 160062.15,
      "step": 220
    },
    {
      "epoch": 0.0009292291034555202,
      "grad_norm": 10360142.0,
      "learning_rate": 3.68e-07,
      "loss": 177085.5,
      "step": 230
    },
    {
      "epoch": 0.0009696303688231516,
      "grad_norm": 4212897.0,
      "learning_rate": 3.84e-07,
      "loss": 156687.3375,
      "step": 240
    },
    {
      "epoch": 0.0010100316341907828,
      "grad_norm": 2492394.75,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 218777.15,
      "step": 250
    },
    {
      "epoch": 0.001050432899558414,
      "grad_norm": 1271087.375,
      "learning_rate": 4.16e-07,
      "loss": 74811.1875,
      "step": 260
    },
    {
      "epoch": 0.0010908341649260454,
      "grad_norm": 11904261.0,
      "learning_rate": 4.3200000000000006e-07,
      "loss": 78403.6125,
      "step": 270
    },
    {
      "epoch": 0.0011312354302936767,
      "grad_norm": 1911185.875,
      "learning_rate": 4.4800000000000004e-07,
      "loss": 112190.75,
      "step": 280
    },
    {
      "epoch": 0.001171636695661308,
      "grad_norm": 1619918.75,
      "learning_rate": 4.64e-07,
      "loss": 33613.9688,
      "step": 290
    },
    {
      "epoch": 0.0012120379610289394,
      "grad_norm": 1365019.625,
      "learning_rate": 4.800000000000001e-07,
      "loss": 32836.45,
      "step": 300
    },
    {
      "epoch": 0.0012524392263965707,
      "grad_norm": 1140846.75,
      "learning_rate": 4.96e-07,
      "loss": 38892.625,
      "step": 310
    },
    {
      "epoch": 0.001292840491764202,
      "grad_norm": 950318.8125,
      "learning_rate": 5.12e-07,
      "loss": 43860.9844,
      "step": 320
    },
    {
      "epoch": 0.0013332417571318333,
      "grad_norm": 191924.40625,
      "learning_rate": 5.280000000000001e-07,
      "loss": 19675.0312,
      "step": 330
    },
    {
      "epoch": 0.0013736430224994647,
      "grad_norm": 354602.96875,
      "learning_rate": 5.44e-07,
      "loss": 17043.5047,
      "step": 340
    },
    {
      "epoch": 0.001414044287867096,
      "grad_norm": 64334.99609375,
      "learning_rate": 5.6e-07,
      "loss": 6989.507,
      "step": 350
    },
    {
      "epoch": 0.0014544455532347273,
      "grad_norm": 29138.234375,
      "learning_rate": 5.760000000000001e-07,
      "loss": 2483.216,
      "step": 360
    },
    {
      "epoch": 0.0014948468186023586,
      "grad_norm": 4881.9619140625,
      "learning_rate": 5.920000000000001e-07,
      "loss": 4704.4551,
      "step": 370
    },
    {
      "epoch": 0.00153524808396999,
      "grad_norm": 73014.8671875,
      "learning_rate": 6.08e-07,
      "loss": 1492.4164,
      "step": 380
    },
    {
      "epoch": 0.0015756493493376213,
      "grad_norm": 86664.8203125,
      "learning_rate": 6.24e-07,
      "loss": 1407.5383,
      "step": 390
    },
    {
      "epoch": 0.0016160506147052526,
      "grad_norm": 91074.265625,
      "learning_rate": 6.4e-07,
      "loss": 932.89,
      "step": 400
    },
    {
      "epoch": 0.001656451880072884,
      "grad_norm": 5759.06396484375,
      "learning_rate": 6.560000000000002e-07,
      "loss": 856.515,
      "step": 410
    },
    {
      "epoch": 0.0016968531454405152,
      "grad_norm": 1454.9195556640625,
      "learning_rate": 6.72e-07,
      "loss": 419.0969,
      "step": 420
    },
    {
      "epoch": 0.0017372544108081465,
      "grad_norm": 673.9178466796875,
      "learning_rate": 6.88e-07,
      "loss": 621.3434,
      "step": 430
    },
    {
      "epoch": 0.0017776556761757779,
      "grad_norm": 1015.685546875,
      "learning_rate": 7.040000000000001e-07,
      "loss": 498.2878,
      "step": 440
    },
    {
      "epoch": 0.0018180569415434092,
      "grad_norm": 6716.2080078125,
      "learning_rate": 7.2e-07,
      "loss": 556.3153,
      "step": 450
    },
    {
      "epoch": 0.0018584582069110405,
      "grad_norm": 1468.1954345703125,
      "learning_rate": 7.36e-07,
      "loss": 488.196,
      "step": 460
    },
    {
      "epoch": 0.0018988594722786718,
      "grad_norm": 5355.9794921875,
      "learning_rate": 7.520000000000001e-07,
      "loss": 553.4161,
      "step": 470
    },
    {
      "epoch": 0.0019392607376463031,
      "grad_norm": 1276.7708740234375,
      "learning_rate": 7.68e-07,
      "loss": 575.7385,
      "step": 480
    },
    {
      "epoch": 0.0019796620030139342,
      "grad_norm": 1997.7783203125,
      "learning_rate": 7.84e-07,
      "loss": 449.5601,
      "step": 490
    },
    {
      "epoch": 0.0020200632683815656,
      "grad_norm": 1234.7044677734375,
      "learning_rate": 8.000000000000001e-07,
      "loss": 454.7207,
      "step": 500
    },
    {
      "epoch": 0.002060464533749197,
      "grad_norm": 1192.5169677734375,
      "learning_rate": 8.160000000000001e-07,
      "loss": 459.7037,
      "step": 510
    },
    {
      "epoch": 0.002100865799116828,
      "grad_norm": 1192.2779541015625,
      "learning_rate": 8.32e-07,
      "loss": 406.1772,
      "step": 520
    },
    {
      "epoch": 0.0021412670644844595,
      "grad_norm": 1168.502685546875,
      "learning_rate": 8.480000000000001e-07,
      "loss": 520.4339,
      "step": 530
    },
    {
      "epoch": 0.002181668329852091,
      "grad_norm": 1580.9521484375,
      "learning_rate": 8.640000000000001e-07,
      "loss": 282.541,
      "step": 540
    },
    {
      "epoch": 0.002222069595219722,
      "grad_norm": 12831.9287109375,
      "learning_rate": 8.8e-07,
      "loss": 531.3956,
      "step": 550
    },
    {
      "epoch": 0.0022624708605873535,
      "grad_norm": 6975.4140625,
      "learning_rate": 8.960000000000001e-07,
      "loss": 377.4974,
      "step": 560
    },
    {
      "epoch": 0.002302872125954985,
      "grad_norm": 3642.0546875,
      "learning_rate": 9.120000000000001e-07,
      "loss": 325.7552,
      "step": 570
    },
    {
      "epoch": 0.002343273391322616,
      "grad_norm": 1394.17822265625,
      "learning_rate": 9.28e-07,
      "loss": 450.4489,
      "step": 580
    },
    {
      "epoch": 0.0023836746566902474,
      "grad_norm": 6102.92626953125,
      "learning_rate": 9.440000000000001e-07,
      "loss": 581.4113,
      "step": 590
    },
    {
      "epoch": 0.0024240759220578788,
      "grad_norm": 942.1673583984375,
      "learning_rate": 9.600000000000001e-07,
      "loss": 327.3541,
      "step": 600
    },
    {
      "epoch": 0.00246447718742551,
      "grad_norm": 7506.9111328125,
      "learning_rate": 9.76e-07,
      "loss": 562.9349,
      "step": 610
    },
    {
      "epoch": 0.0025048784527931414,
      "grad_norm": 828.693603515625,
      "learning_rate": 9.92e-07,
      "loss": 480.7408,
      "step": 620
    },
    {
      "epoch": 0.0025452797181607727,
      "grad_norm": 1101.0645751953125,
      "learning_rate": 1.0080000000000001e-06,
      "loss": 510.9642,
      "step": 630
    },
    {
      "epoch": 0.002585680983528404,
      "grad_norm": 4109.21728515625,
      "learning_rate": 1.024e-06,
      "loss": 380.4708,
      "step": 640
    },
    {
      "epoch": 0.0026260822488960354,
      "grad_norm": 1035.9979248046875,
      "learning_rate": 1.04e-06,
      "loss": 438.6869,
      "step": 650
    },
    {
      "epoch": 0.0026664835142636667,
      "grad_norm": 857.4368896484375,
      "learning_rate": 1.0560000000000001e-06,
      "loss": 333.5929,
      "step": 660
    },
    {
      "epoch": 0.002706884779631298,
      "grad_norm": 861.3447875976562,
      "learning_rate": 1.072e-06,
      "loss": 369.1007,
      "step": 670
    },
    {
      "epoch": 0.0027472860449989293,
      "grad_norm": 1937.781005859375,
      "learning_rate": 1.088e-06,
      "loss": 467.2207,
      "step": 680
    },
    {
      "epoch": 0.0027876873103665606,
      "grad_norm": 11980.4130859375,
      "learning_rate": 1.1040000000000001e-06,
      "loss": 461.9273,
      "step": 690
    },
    {
      "epoch": 0.002828088575734192,
      "grad_norm": 1039.8837890625,
      "learning_rate": 1.12e-06,
      "loss": 376.5904,
      "step": 700
    },
    {
      "epoch": 0.0028684898411018233,
      "grad_norm": 2001.8681640625,
      "learning_rate": 1.1360000000000002e-06,
      "loss": 397.0597,
      "step": 710
    },
    {
      "epoch": 0.0029088911064694546,
      "grad_norm": 4143.15087890625,
      "learning_rate": 1.1520000000000002e-06,
      "loss": 524.1538,
      "step": 720
    },
    {
      "epoch": 0.002949292371837086,
      "grad_norm": 1036.8919677734375,
      "learning_rate": 1.168e-06,
      "loss": 367.9916,
      "step": 730
    },
    {
      "epoch": 0.0029896936372047172,
      "grad_norm": 8330.5068359375,
      "learning_rate": 1.1840000000000002e-06,
      "loss": 394.5002,
      "step": 740
    },
    {
      "epoch": 0.0030300949025723486,
      "grad_norm": 2307.091064453125,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 295.3556,
      "step": 750
    },
    {
      "epoch": 0.00307049616793998,
      "grad_norm": 1893.0789794921875,
      "learning_rate": 1.216e-06,
      "loss": 265.5479,
      "step": 760
    },
    {
      "epoch": 0.003110897433307611,
      "grad_norm": 2477.446533203125,
      "learning_rate": 1.2320000000000002e-06,
      "loss": 534.005,
      "step": 770
    },
    {
      "epoch": 0.0031512986986752425,
      "grad_norm": 916.7561645507812,
      "learning_rate": 1.248e-06,
      "loss": 334.4991,
      "step": 780
    },
    {
      "epoch": 0.003191699964042874,
      "grad_norm": 13106.6064453125,
      "learning_rate": 1.2640000000000003e-06,
      "loss": 464.4099,
      "step": 790
    },
    {
      "epoch": 0.003232101229410505,
      "grad_norm": 1100.711669921875,
      "learning_rate": 1.28e-06,
      "loss": 302.2139,
      "step": 800
    },
    {
      "epoch": 0.0032725024947781365,
      "grad_norm": 2172.443603515625,
      "learning_rate": 1.296e-06,
      "loss": 389.6274,
      "step": 810
    },
    {
      "epoch": 0.003312903760145768,
      "grad_norm": 1690.6312255859375,
      "learning_rate": 1.3120000000000003e-06,
      "loss": 361.3152,
      "step": 820
    },
    {
      "epoch": 0.003353305025513399,
      "grad_norm": 924.0753173828125,
      "learning_rate": 1.328e-06,
      "loss": 247.9785,
      "step": 830
    },
    {
      "epoch": 0.0033937062908810304,
      "grad_norm": 1367.6988525390625,
      "learning_rate": 1.344e-06,
      "loss": 320.1396,
      "step": 840
    },
    {
      "epoch": 0.0034341075562486618,
      "grad_norm": 731.0768432617188,
      "learning_rate": 1.3600000000000001e-06,
      "loss": 394.588,
      "step": 850
    },
    {
      "epoch": 0.003474508821616293,
      "grad_norm": 788.498046875,
      "learning_rate": 1.376e-06,
      "loss": 430.4721,
      "step": 860
    },
    {
      "epoch": 0.0035149100869839244,
      "grad_norm": 989.5845336914062,
      "learning_rate": 1.392e-06,
      "loss": 343.7414,
      "step": 870
    },
    {
      "epoch": 0.0035553113523515557,
      "grad_norm": 661.3421020507812,
      "learning_rate": 1.4080000000000001e-06,
      "loss": 382.3532,
      "step": 880
    },
    {
      "epoch": 0.003595712617719187,
      "grad_norm": 831.7238159179688,
      "learning_rate": 1.424e-06,
      "loss": 442.5967,
      "step": 890
    },
    {
      "epoch": 0.0036361138830868184,
      "grad_norm": 1258.17236328125,
      "learning_rate": 1.44e-06,
      "loss": 329.9442,
      "step": 900
    },
    {
      "epoch": 0.0036765151484544497,
      "grad_norm": 1166.2576904296875,
      "learning_rate": 1.4560000000000001e-06,
      "loss": 332.621,
      "step": 910
    },
    {
      "epoch": 0.003716916413822081,
      "grad_norm": 877.9427490234375,
      "learning_rate": 1.472e-06,
      "loss": 464.8878,
      "step": 920
    },
    {
      "epoch": 0.0037573176791897123,
      "grad_norm": 1044.3133544921875,
      "learning_rate": 1.488e-06,
      "loss": 292.7577,
      "step": 930
    },
    {
      "epoch": 0.0037977189445573436,
      "grad_norm": 1089.46728515625,
      "learning_rate": 1.5040000000000001e-06,
      "loss": 460.0909,
      "step": 940
    },
    {
      "epoch": 0.003838120209924975,
      "grad_norm": 797.8550415039062,
      "learning_rate": 1.52e-06,
      "loss": 346.0366,
      "step": 950
    },
    {
      "epoch": 0.0038785214752926063,
      "grad_norm": 798.5408325195312,
      "learning_rate": 1.536e-06,
      "loss": 310.8797,
      "step": 960
    },
    {
      "epoch": 0.003918922740660237,
      "grad_norm": 2262.073974609375,
      "learning_rate": 1.5520000000000001e-06,
      "loss": 370.4966,
      "step": 970
    },
    {
      "epoch": 0.0039593240060278685,
      "grad_norm": 2735.94482421875,
      "learning_rate": 1.568e-06,
      "loss": 323.6035,
      "step": 980
    },
    {
      "epoch": 0.0039997252713955,
      "grad_norm": 646.0745849609375,
      "learning_rate": 1.5840000000000002e-06,
      "loss": 339.8437,
      "step": 990
    },
    {
      "epoch": 0.004040126536763131,
      "grad_norm": 1748.4075927734375,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 317.4156,
      "step": 1000
    },
    {
      "epoch": 0.0040805278021307624,
      "grad_norm": 780.6202392578125,
      "learning_rate": 1.616e-06,
      "loss": 296.7254,
      "step": 1010
    },
    {
      "epoch": 0.004120929067498394,
      "grad_norm": 1253.2996826171875,
      "learning_rate": 1.6320000000000002e-06,
      "loss": 384.1156,
      "step": 1020
    },
    {
      "epoch": 0.004161330332866025,
      "grad_norm": 1055.1898193359375,
      "learning_rate": 1.6480000000000001e-06,
      "loss": 397.7781,
      "step": 1030
    },
    {
      "epoch": 0.004201731598233656,
      "grad_norm": 803.1446533203125,
      "learning_rate": 1.664e-06,
      "loss": 359.3899,
      "step": 1040
    },
    {
      "epoch": 0.004242132863601288,
      "grad_norm": 798.8219604492188,
      "learning_rate": 1.6800000000000002e-06,
      "loss": 381.0167,
      "step": 1050
    },
    {
      "epoch": 0.004282534128968919,
      "grad_norm": 3304.813232421875,
      "learning_rate": 1.6960000000000002e-06,
      "loss": 391.1282,
      "step": 1060
    },
    {
      "epoch": 0.00432293539433655,
      "grad_norm": 2308.608154296875,
      "learning_rate": 1.712e-06,
      "loss": 453.3599,
      "step": 1070
    },
    {
      "epoch": 0.004363336659704182,
      "grad_norm": 744.8621215820312,
      "learning_rate": 1.7280000000000002e-06,
      "loss": 306.6811,
      "step": 1080
    },
    {
      "epoch": 0.004403737925071813,
      "grad_norm": 5821.24267578125,
      "learning_rate": 1.7440000000000002e-06,
      "loss": 608.5389,
      "step": 1090
    },
    {
      "epoch": 0.004444139190439444,
      "grad_norm": 865.3998413085938,
      "learning_rate": 1.76e-06,
      "loss": 242.1831,
      "step": 1100
    },
    {
      "epoch": 0.004484540455807076,
      "grad_norm": 955.5283203125,
      "learning_rate": 1.7760000000000002e-06,
      "loss": 406.3821,
      "step": 1110
    },
    {
      "epoch": 0.004524941721174707,
      "grad_norm": 1257.974853515625,
      "learning_rate": 1.7920000000000002e-06,
      "loss": 337.8827,
      "step": 1120
    },
    {
      "epoch": 0.004565342986542338,
      "grad_norm": 848.3350830078125,
      "learning_rate": 1.808e-06,
      "loss": 383.7443,
      "step": 1130
    },
    {
      "epoch": 0.00460574425190997,
      "grad_norm": 2609.029541015625,
      "learning_rate": 1.8240000000000002e-06,
      "loss": 394.2932,
      "step": 1140
    },
    {
      "epoch": 0.004646145517277601,
      "grad_norm": 1440.8753662109375,
      "learning_rate": 1.8400000000000002e-06,
      "loss": 433.8044,
      "step": 1150
    },
    {
      "epoch": 0.004686546782645232,
      "grad_norm": 1531.273681640625,
      "learning_rate": 1.856e-06,
      "loss": 424.2002,
      "step": 1160
    },
    {
      "epoch": 0.0047269480480128636,
      "grad_norm": 2845.63623046875,
      "learning_rate": 1.8720000000000002e-06,
      "loss": 343.2191,
      "step": 1170
    },
    {
      "epoch": 0.004767349313380495,
      "grad_norm": 679.157470703125,
      "learning_rate": 1.8880000000000002e-06,
      "loss": 327.4853,
      "step": 1180
    },
    {
      "epoch": 0.004807750578748126,
      "grad_norm": 1685.8907470703125,
      "learning_rate": 1.9040000000000003e-06,
      "loss": 343.0018,
      "step": 1190
    },
    {
      "epoch": 0.0048481518441157575,
      "grad_norm": 961.8662719726562,
      "learning_rate": 1.9200000000000003e-06,
      "loss": 345.5244,
      "step": 1200
    },
    {
      "epoch": 0.004888553109483389,
      "grad_norm": 1039.9063720703125,
      "learning_rate": 1.936e-06,
      "loss": 286.9379,
      "step": 1210
    },
    {
      "epoch": 0.00492895437485102,
      "grad_norm": 915.8055419921875,
      "learning_rate": 1.952e-06,
      "loss": 291.1856,
      "step": 1220
    },
    {
      "epoch": 0.0049693556402186515,
      "grad_norm": 2453.09716796875,
      "learning_rate": 1.968e-06,
      "loss": 357.5581,
      "step": 1230
    },
    {
      "epoch": 0.005009756905586283,
      "grad_norm": 1281.601318359375,
      "learning_rate": 1.984e-06,
      "loss": 370.0242,
      "step": 1240
    },
    {
      "epoch": 0.005050158170953914,
      "grad_norm": 1927.247802734375,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 420.42,
      "step": 1250
    },
    {
      "epoch": 0.0050905594363215454,
      "grad_norm": 1159.7850341796875,
      "learning_rate": 2.0160000000000003e-06,
      "loss": 381.0742,
      "step": 1260
    },
    {
      "epoch": 0.005130960701689177,
      "grad_norm": 1102.625244140625,
      "learning_rate": 2.032e-06,
      "loss": 441.1049,
      "step": 1270
    },
    {
      "epoch": 0.005171361967056808,
      "grad_norm": 1439.484130859375,
      "learning_rate": 2.048e-06,
      "loss": 277.274,
      "step": 1280
    },
    {
      "epoch": 0.005211763232424439,
      "grad_norm": 5610.380859375,
      "learning_rate": 2.064e-06,
      "loss": 421.7488,
      "step": 1290
    },
    {
      "epoch": 0.005252164497792071,
      "grad_norm": 982.2139892578125,
      "learning_rate": 2.08e-06,
      "loss": 340.8586,
      "step": 1300
    },
    {
      "epoch": 0.005292565763159702,
      "grad_norm": 791.9103393554688,
      "learning_rate": 2.0960000000000003e-06,
      "loss": 363.6415,
      "step": 1310
    },
    {
      "epoch": 0.005332967028527333,
      "grad_norm": 0.0,
      "learning_rate": 2.1120000000000003e-06,
      "loss": 443.9375,
      "step": 1320
    },
    {
      "epoch": 0.005373368293894965,
      "grad_norm": 1269.4005126953125,
      "learning_rate": 2.128e-06,
      "loss": 352.3543,
      "step": 1330
    },
    {
      "epoch": 0.005413769559262596,
      "grad_norm": 1225.1302490234375,
      "learning_rate": 2.144e-06,
      "loss": 357.3038,
      "step": 1340
    },
    {
      "epoch": 0.005454170824630227,
      "grad_norm": 827.63720703125,
      "learning_rate": 2.16e-06,
      "loss": 303.4535,
      "step": 1350
    },
    {
      "epoch": 0.005494572089997859,
      "grad_norm": 848.7166137695312,
      "learning_rate": 2.176e-06,
      "loss": 268.09,
      "step": 1360
    },
    {
      "epoch": 0.00553497335536549,
      "grad_norm": 1410.2564697265625,
      "learning_rate": 2.1920000000000004e-06,
      "loss": 389.759,
      "step": 1370
    },
    {
      "epoch": 0.005575374620733121,
      "grad_norm": 1602.010009765625,
      "learning_rate": 2.2080000000000003e-06,
      "loss": 412.8606,
      "step": 1380
    },
    {
      "epoch": 0.005615775886100753,
      "grad_norm": 1337.7384033203125,
      "learning_rate": 2.2240000000000002e-06,
      "loss": 428.547,
      "step": 1390
    },
    {
      "epoch": 0.005656177151468384,
      "grad_norm": 867.4197387695312,
      "learning_rate": 2.24e-06,
      "loss": 378.6157,
      "step": 1400
    },
    {
      "epoch": 0.005696578416836015,
      "grad_norm": 731.271728515625,
      "learning_rate": 2.256e-06,
      "loss": 374.0744,
      "step": 1410
    },
    {
      "epoch": 0.0057369796822036466,
      "grad_norm": 2343.665771484375,
      "learning_rate": 2.2720000000000004e-06,
      "loss": 379.0267,
      "step": 1420
    },
    {
      "epoch": 0.005777380947571278,
      "grad_norm": 1540.48388671875,
      "learning_rate": 2.2880000000000004e-06,
      "loss": 382.6617,
      "step": 1430
    },
    {
      "epoch": 0.005817782212938909,
      "grad_norm": 1533.5958251953125,
      "learning_rate": 2.3040000000000003e-06,
      "loss": 396.0985,
      "step": 1440
    },
    {
      "epoch": 0.0058581834783065405,
      "grad_norm": 723.0481567382812,
      "learning_rate": 2.3200000000000002e-06,
      "loss": 308.0563,
      "step": 1450
    },
    {
      "epoch": 0.005898584743674172,
      "grad_norm": 2209.6083984375,
      "learning_rate": 2.336e-06,
      "loss": 424.7519,
      "step": 1460
    },
    {
      "epoch": 0.005938986009041803,
      "grad_norm": 566.4544067382812,
      "learning_rate": 2.352e-06,
      "loss": 248.9047,
      "step": 1470
    },
    {
      "epoch": 0.0059793872744094345,
      "grad_norm": 868.7781982421875,
      "learning_rate": 2.3680000000000005e-06,
      "loss": 257.8537,
      "step": 1480
    },
    {
      "epoch": 0.006019788539777066,
      "grad_norm": 732.210205078125,
      "learning_rate": 2.3840000000000004e-06,
      "loss": 304.146,
      "step": 1490
    },
    {
      "epoch": 0.006060189805144697,
      "grad_norm": 946.9075317382812,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 239.7318,
      "step": 1500
    },
    {
      "epoch": 0.0061005910705123284,
      "grad_norm": 13990.4541015625,
      "learning_rate": 2.4160000000000002e-06,
      "loss": 317.3043,
      "step": 1510
    },
    {
      "epoch": 0.00614099233587996,
      "grad_norm": 1053.9326171875,
      "learning_rate": 2.432e-06,
      "loss": 306.3782,
      "step": 1520
    },
    {
      "epoch": 0.006181393601247591,
      "grad_norm": 1479.0665283203125,
      "learning_rate": 2.448e-06,
      "loss": 321.1809,
      "step": 1530
    },
    {
      "epoch": 0.006221794866615222,
      "grad_norm": 0.0,
      "learning_rate": 2.4640000000000005e-06,
      "loss": 272.3122,
      "step": 1540
    },
    {
      "epoch": 0.006262196131982854,
      "grad_norm": 836.43994140625,
      "learning_rate": 2.4800000000000004e-06,
      "loss": 276.743,
      "step": 1550
    },
    {
      "epoch": 0.006302597397350485,
      "grad_norm": 1576.191650390625,
      "learning_rate": 2.496e-06,
      "loss": 293.9756,
      "step": 1560
    },
    {
      "epoch": 0.006342998662718116,
      "grad_norm": 1241.993408203125,
      "learning_rate": 2.512e-06,
      "loss": 324.3042,
      "step": 1570
    },
    {
      "epoch": 0.006383399928085748,
      "grad_norm": 3129.779052734375,
      "learning_rate": 2.5280000000000006e-06,
      "loss": 282.3287,
      "step": 1580
    },
    {
      "epoch": 0.006423801193453379,
      "grad_norm": 824.8917846679688,
      "learning_rate": 2.5440000000000005e-06,
      "loss": 332.7758,
      "step": 1590
    },
    {
      "epoch": 0.00646420245882101,
      "grad_norm": 1429.83154296875,
      "learning_rate": 2.56e-06,
      "loss": 341.1543,
      "step": 1600
    },
    {
      "epoch": 0.006504603724188642,
      "grad_norm": 1096.0162353515625,
      "learning_rate": 2.576e-06,
      "loss": 370.956,
      "step": 1610
    },
    {
      "epoch": 0.006545004989556273,
      "grad_norm": 2190.672607421875,
      "learning_rate": 2.592e-06,
      "loss": 393.7615,
      "step": 1620
    },
    {
      "epoch": 0.006585406254923904,
      "grad_norm": 659.6456909179688,
      "learning_rate": 2.608e-06,
      "loss": 371.8131,
      "step": 1630
    },
    {
      "epoch": 0.006625807520291536,
      "grad_norm": 1114.4139404296875,
      "learning_rate": 2.6240000000000006e-06,
      "loss": 387.3432,
      "step": 1640
    },
    {
      "epoch": 0.006666208785659167,
      "grad_norm": 1945.1435546875,
      "learning_rate": 2.64e-06,
      "loss": 419.9613,
      "step": 1650
    },
    {
      "epoch": 0.006706610051026798,
      "grad_norm": 1134.198486328125,
      "learning_rate": 2.656e-06,
      "loss": 261.7552,
      "step": 1660
    },
    {
      "epoch": 0.0067470113163944296,
      "grad_norm": 625.1832885742188,
      "learning_rate": 2.672e-06,
      "loss": 348.9073,
      "step": 1670
    },
    {
      "epoch": 0.006787412581762061,
      "grad_norm": 1558.600830078125,
      "learning_rate": 2.688e-06,
      "loss": 329.5783,
      "step": 1680
    },
    {
      "epoch": 0.006827813847129692,
      "grad_norm": 903.2876586914062,
      "learning_rate": 2.704e-06,
      "loss": 288.3402,
      "step": 1690
    },
    {
      "epoch": 0.0068682151124973235,
      "grad_norm": 4166.87158203125,
      "learning_rate": 2.7200000000000002e-06,
      "loss": 357.7317,
      "step": 1700
    },
    {
      "epoch": 0.006908616377864955,
      "grad_norm": 784.2251586914062,
      "learning_rate": 2.736e-06,
      "loss": 318.5063,
      "step": 1710
    },
    {
      "epoch": 0.006949017643232586,
      "grad_norm": 5029.16552734375,
      "learning_rate": 2.752e-06,
      "loss": 442.4195,
      "step": 1720
    },
    {
      "epoch": 0.0069894189086002175,
      "grad_norm": 781.5169677734375,
      "learning_rate": 2.768e-06,
      "loss": 306.0392,
      "step": 1730
    },
    {
      "epoch": 0.007029820173967849,
      "grad_norm": 1048.294921875,
      "learning_rate": 2.784e-06,
      "loss": 359.3845,
      "step": 1740
    },
    {
      "epoch": 0.00707022143933548,
      "grad_norm": 1155.3671875,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 359.3311,
      "step": 1750
    },
    {
      "epoch": 0.0071106227047031114,
      "grad_norm": 857.3798217773438,
      "learning_rate": 2.8160000000000002e-06,
      "loss": 391.5449,
      "step": 1760
    },
    {
      "epoch": 0.007151023970070743,
      "grad_norm": 933.0081787109375,
      "learning_rate": 2.832e-06,
      "loss": 297.32,
      "step": 1770
    },
    {
      "epoch": 0.007191425235438374,
      "grad_norm": 1060.47314453125,
      "learning_rate": 2.848e-06,
      "loss": 411.9166,
      "step": 1780
    },
    {
      "epoch": 0.007231826500806005,
      "grad_norm": 707.4050903320312,
      "learning_rate": 2.864e-06,
      "loss": 238.9362,
      "step": 1790
    },
    {
      "epoch": 0.007272227766173637,
      "grad_norm": 526.6619873046875,
      "learning_rate": 2.88e-06,
      "loss": 290.1179,
      "step": 1800
    },
    {
      "epoch": 0.007312629031541268,
      "grad_norm": 829.2198486328125,
      "learning_rate": 2.8960000000000003e-06,
      "loss": 347.0842,
      "step": 1810
    },
    {
      "epoch": 0.007353030296908899,
      "grad_norm": 716.2144165039062,
      "learning_rate": 2.9120000000000002e-06,
      "loss": 295.0978,
      "step": 1820
    },
    {
      "epoch": 0.007393431562276531,
      "grad_norm": 1365.7164306640625,
      "learning_rate": 2.928e-06,
      "loss": 399.1327,
      "step": 1830
    },
    {
      "epoch": 0.007433832827644162,
      "grad_norm": 1474.3541259765625,
      "learning_rate": 2.944e-06,
      "loss": 249.1963,
      "step": 1840
    },
    {
      "epoch": 0.007474234093011793,
      "grad_norm": 958.6668090820312,
      "learning_rate": 2.96e-06,
      "loss": 405.7022,
      "step": 1850
    },
    {
      "epoch": 0.007514635358379425,
      "grad_norm": 928.7461547851562,
      "learning_rate": 2.976e-06,
      "loss": 392.5513,
      "step": 1860
    },
    {
      "epoch": 0.007555036623747056,
      "grad_norm": 0.0,
      "learning_rate": 2.9920000000000003e-06,
      "loss": 235.2757,
      "step": 1870
    },
    {
      "epoch": 0.007595437889114687,
      "grad_norm": 13806.5771484375,
      "learning_rate": 3.0080000000000003e-06,
      "loss": 338.4041,
      "step": 1880
    },
    {
      "epoch": 0.007635839154482319,
      "grad_norm": 2829.76513671875,
      "learning_rate": 3.024e-06,
      "loss": 365.0503,
      "step": 1890
    },
    {
      "epoch": 0.00767624041984995,
      "grad_norm": 1206.556640625,
      "learning_rate": 3.04e-06,
      "loss": 372.7524,
      "step": 1900
    },
    {
      "epoch": 0.007716641685217581,
      "grad_norm": 1795.8287353515625,
      "learning_rate": 3.056e-06,
      "loss": 355.855,
      "step": 1910
    },
    {
      "epoch": 0.0077570429505852126,
      "grad_norm": 733.5438232421875,
      "learning_rate": 3.072e-06,
      "loss": 369.4895,
      "step": 1920
    },
    {
      "epoch": 0.007797444215952844,
      "grad_norm": 1936.3282470703125,
      "learning_rate": 3.0880000000000003e-06,
      "loss": 331.1103,
      "step": 1930
    },
    {
      "epoch": 0.007837845481320474,
      "grad_norm": 2228.966552734375,
      "learning_rate": 3.1040000000000003e-06,
      "loss": 441.473,
      "step": 1940
    },
    {
      "epoch": 0.007878246746688106,
      "grad_norm": 1527.3521728515625,
      "learning_rate": 3.12e-06,
      "loss": 286.5484,
      "step": 1950
    },
    {
      "epoch": 0.007918648012055737,
      "grad_norm": 1155.413330078125,
      "learning_rate": 3.136e-06,
      "loss": 325.0518,
      "step": 1960
    },
    {
      "epoch": 0.007959049277423368,
      "grad_norm": 5305.16845703125,
      "learning_rate": 3.152e-06,
      "loss": 426.8875,
      "step": 1970
    },
    {
      "epoch": 0.007999450542791,
      "grad_norm": 497.835693359375,
      "learning_rate": 3.1680000000000004e-06,
      "loss": 305.8856,
      "step": 1980
    },
    {
      "epoch": 0.008039851808158631,
      "grad_norm": 950.1622924804688,
      "learning_rate": 3.1840000000000003e-06,
      "loss": 298.3484,
      "step": 1990
    },
    {
      "epoch": 0.008080253073526262,
      "grad_norm": 977.10546875,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 385.7309,
      "step": 2000
    },
    {
      "epoch": 0.008120654338893894,
      "grad_norm": 1347.1170654296875,
      "learning_rate": 3.216e-06,
      "loss": 311.8849,
      "step": 2010
    },
    {
      "epoch": 0.008161055604261525,
      "grad_norm": 7115.32470703125,
      "learning_rate": 3.232e-06,
      "loss": 310.0182,
      "step": 2020
    },
    {
      "epoch": 0.008201456869629156,
      "grad_norm": 979.557861328125,
      "learning_rate": 3.248e-06,
      "loss": 378.8968,
      "step": 2030
    },
    {
      "epoch": 0.008241858134996788,
      "grad_norm": 713.5515747070312,
      "learning_rate": 3.2640000000000004e-06,
      "loss": 333.8391,
      "step": 2040
    },
    {
      "epoch": 0.008282259400364419,
      "grad_norm": 1746.4747314453125,
      "learning_rate": 3.2800000000000004e-06,
      "loss": 211.6272,
      "step": 2050
    },
    {
      "epoch": 0.00832266066573205,
      "grad_norm": 3959.4423828125,
      "learning_rate": 3.2960000000000003e-06,
      "loss": 418.0941,
      "step": 2060
    },
    {
      "epoch": 0.008363061931099681,
      "grad_norm": 1449.259765625,
      "learning_rate": 3.3120000000000002e-06,
      "loss": 366.1021,
      "step": 2070
    },
    {
      "epoch": 0.008403463196467313,
      "grad_norm": 686.3818359375,
      "learning_rate": 3.328e-06,
      "loss": 317.5829,
      "step": 2080
    },
    {
      "epoch": 0.008443864461834944,
      "grad_norm": 1431.3048095703125,
      "learning_rate": 3.344e-06,
      "loss": 341.1351,
      "step": 2090
    },
    {
      "epoch": 0.008484265727202575,
      "grad_norm": 1141.1195068359375,
      "learning_rate": 3.3600000000000004e-06,
      "loss": 284.9502,
      "step": 2100
    },
    {
      "epoch": 0.008524666992570207,
      "grad_norm": 583.6392211914062,
      "learning_rate": 3.3760000000000004e-06,
      "loss": 246.5204,
      "step": 2110
    },
    {
      "epoch": 0.008565068257937838,
      "grad_norm": 1383.4288330078125,
      "learning_rate": 3.3920000000000003e-06,
      "loss": 335.3397,
      "step": 2120
    },
    {
      "epoch": 0.00860546952330547,
      "grad_norm": 720.5582275390625,
      "learning_rate": 3.4080000000000002e-06,
      "loss": 300.3475,
      "step": 2130
    },
    {
      "epoch": 0.0086458707886731,
      "grad_norm": 1018.2730712890625,
      "learning_rate": 3.424e-06,
      "loss": 364.7046,
      "step": 2140
    },
    {
      "epoch": 0.008686272054040732,
      "grad_norm": 836.5111694335938,
      "learning_rate": 3.44e-06,
      "loss": 215.2347,
      "step": 2150
    },
    {
      "epoch": 0.008726673319408363,
      "grad_norm": 1106.7752685546875,
      "learning_rate": 3.4560000000000005e-06,
      "loss": 392.8156,
      "step": 2160
    },
    {
      "epoch": 0.008767074584775995,
      "grad_norm": 838.0346069335938,
      "learning_rate": 3.4720000000000004e-06,
      "loss": 328.1276,
      "step": 2170
    },
    {
      "epoch": 0.008807475850143626,
      "grad_norm": 870.4414672851562,
      "learning_rate": 3.4880000000000003e-06,
      "loss": 313.8927,
      "step": 2180
    },
    {
      "epoch": 0.008847877115511257,
      "grad_norm": 2859.623046875,
      "learning_rate": 3.5040000000000002e-06,
      "loss": 241.9949,
      "step": 2190
    },
    {
      "epoch": 0.008888278380878889,
      "grad_norm": 971.9884033203125,
      "learning_rate": 3.52e-06,
      "loss": 349.5033,
      "step": 2200
    },
    {
      "epoch": 0.00892867964624652,
      "grad_norm": 670.4691772460938,
      "learning_rate": 3.5360000000000005e-06,
      "loss": 239.4784,
      "step": 2210
    },
    {
      "epoch": 0.008969080911614151,
      "grad_norm": 784.5670166015625,
      "learning_rate": 3.5520000000000005e-06,
      "loss": 310.9449,
      "step": 2220
    },
    {
      "epoch": 0.009009482176981783,
      "grad_norm": 915.0491333007812,
      "learning_rate": 3.5680000000000004e-06,
      "loss": 413.7835,
      "step": 2230
    },
    {
      "epoch": 0.009049883442349414,
      "grad_norm": 893.2843627929688,
      "learning_rate": 3.5840000000000003e-06,
      "loss": 365.2546,
      "step": 2240
    },
    {
      "epoch": 0.009090284707717045,
      "grad_norm": 1068.404296875,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 302.3378,
      "step": 2250
    },
    {
      "epoch": 0.009130685973084677,
      "grad_norm": 1454.05859375,
      "learning_rate": 3.616e-06,
      "loss": 285.7942,
      "step": 2260
    },
    {
      "epoch": 0.009171087238452308,
      "grad_norm": 3212.861083984375,
      "learning_rate": 3.6320000000000005e-06,
      "loss": 296.5446,
      "step": 2270
    },
    {
      "epoch": 0.00921148850381994,
      "grad_norm": 700.9654541015625,
      "learning_rate": 3.6480000000000005e-06,
      "loss": 277.7706,
      "step": 2280
    },
    {
      "epoch": 0.00925188976918757,
      "grad_norm": 601.0996704101562,
      "learning_rate": 3.6640000000000004e-06,
      "loss": 299.2936,
      "step": 2290
    },
    {
      "epoch": 0.009292291034555202,
      "grad_norm": 1073.6248779296875,
      "learning_rate": 3.6800000000000003e-06,
      "loss": 227.1634,
      "step": 2300
    },
    {
      "epoch": 0.009332692299922833,
      "grad_norm": 803.8406372070312,
      "learning_rate": 3.6960000000000003e-06,
      "loss": 344.3006,
      "step": 2310
    },
    {
      "epoch": 0.009373093565290464,
      "grad_norm": 940.9983520507812,
      "learning_rate": 3.712e-06,
      "loss": 266.4591,
      "step": 2320
    },
    {
      "epoch": 0.009413494830658096,
      "grad_norm": 13408.8291015625,
      "learning_rate": 3.7280000000000006e-06,
      "loss": 295.9835,
      "step": 2330
    },
    {
      "epoch": 0.009453896096025727,
      "grad_norm": 1009.2761840820312,
      "learning_rate": 3.7440000000000005e-06,
      "loss": 292.3968,
      "step": 2340
    },
    {
      "epoch": 0.009494297361393358,
      "grad_norm": 556.4901733398438,
      "learning_rate": 3.7600000000000004e-06,
      "loss": 262.7075,
      "step": 2350
    },
    {
      "epoch": 0.00953469862676099,
      "grad_norm": 913.5816040039062,
      "learning_rate": 3.7760000000000004e-06,
      "loss": 508.4715,
      "step": 2360
    },
    {
      "epoch": 0.009575099892128621,
      "grad_norm": 1582.836181640625,
      "learning_rate": 3.7920000000000003e-06,
      "loss": 321.3247,
      "step": 2370
    },
    {
      "epoch": 0.009615501157496252,
      "grad_norm": 1292.945068359375,
      "learning_rate": 3.8080000000000006e-06,
      "loss": 266.3162,
      "step": 2380
    },
    {
      "epoch": 0.009655902422863884,
      "grad_norm": 741.0372314453125,
      "learning_rate": 3.824e-06,
      "loss": 333.5866,
      "step": 2390
    },
    {
      "epoch": 0.009696303688231515,
      "grad_norm": 594.6494750976562,
      "learning_rate": 3.8400000000000005e-06,
      "loss": 370.6466,
      "step": 2400
    },
    {
      "epoch": 0.009736704953599146,
      "grad_norm": 671.8253173828125,
      "learning_rate": 3.856e-06,
      "loss": 176.7593,
      "step": 2410
    },
    {
      "epoch": 0.009777106218966778,
      "grad_norm": 1318.59912109375,
      "learning_rate": 3.872e-06,
      "loss": 296.8816,
      "step": 2420
    },
    {
      "epoch": 0.009817507484334409,
      "grad_norm": 4490.33837890625,
      "learning_rate": 3.888e-06,
      "loss": 353.2209,
      "step": 2430
    },
    {
      "epoch": 0.00985790874970204,
      "grad_norm": 1145.6590576171875,
      "learning_rate": 3.904e-06,
      "loss": 232.9775,
      "step": 2440
    },
    {
      "epoch": 0.009898310015069672,
      "grad_norm": 669.175537109375,
      "learning_rate": 3.920000000000001e-06,
      "loss": 245.2131,
      "step": 2450
    },
    {
      "epoch": 0.009938711280437303,
      "grad_norm": 945.0543212890625,
      "learning_rate": 3.936e-06,
      "loss": 189.2121,
      "step": 2460
    },
    {
      "epoch": 0.009979112545804934,
      "grad_norm": 871.2230224609375,
      "learning_rate": 3.9520000000000004e-06,
      "loss": 354.6668,
      "step": 2470
    },
    {
      "epoch": 0.010019513811172566,
      "grad_norm": 1143.5745849609375,
      "learning_rate": 3.968e-06,
      "loss": 309.6017,
      "step": 2480
    },
    {
      "epoch": 0.010059915076540197,
      "grad_norm": 921.1688842773438,
      "learning_rate": 3.984e-06,
      "loss": 393.8875,
      "step": 2490
    },
    {
      "epoch": 0.010100316341907828,
      "grad_norm": 670.0267333984375,
      "learning_rate": 4.000000000000001e-06,
      "loss": 187.9601,
      "step": 2500
    },
    {
      "epoch": 0.01014071760727546,
      "grad_norm": 906.67626953125,
      "learning_rate": 4.016e-06,
      "loss": 263.1041,
      "step": 2510
    },
    {
      "epoch": 0.010181118872643091,
      "grad_norm": 501.8714599609375,
      "learning_rate": 4.0320000000000005e-06,
      "loss": 249.8009,
      "step": 2520
    },
    {
      "epoch": 0.010221520138010722,
      "grad_norm": 1012.7957153320312,
      "learning_rate": 4.048e-06,
      "loss": 215.6589,
      "step": 2530
    },
    {
      "epoch": 0.010261921403378354,
      "grad_norm": 1067.8087158203125,
      "learning_rate": 4.064e-06,
      "loss": 277.4538,
      "step": 2540
    },
    {
      "epoch": 0.010302322668745985,
      "grad_norm": 821.8876342773438,
      "learning_rate": 4.08e-06,
      "loss": 357.288,
      "step": 2550
    },
    {
      "epoch": 0.010342723934113616,
      "grad_norm": 695.7543334960938,
      "learning_rate": 4.096e-06,
      "loss": 192.3841,
      "step": 2560
    },
    {
      "epoch": 0.010383125199481247,
      "grad_norm": 2217.32080078125,
      "learning_rate": 4.112000000000001e-06,
      "loss": 311.4946,
      "step": 2570
    },
    {
      "epoch": 0.010423526464848879,
      "grad_norm": 1252.262451171875,
      "learning_rate": 4.128e-06,
      "loss": 299.7565,
      "step": 2580
    },
    {
      "epoch": 0.01046392773021651,
      "grad_norm": 1227.412109375,
      "learning_rate": 4.1440000000000005e-06,
      "loss": 258.2722,
      "step": 2590
    },
    {
      "epoch": 0.010504328995584141,
      "grad_norm": 1460.778076171875,
      "learning_rate": 4.16e-06,
      "loss": 310.2179,
      "step": 2600
    },
    {
      "epoch": 0.010544730260951773,
      "grad_norm": 3191.095947265625,
      "learning_rate": 4.176e-06,
      "loss": 276.1908,
      "step": 2610
    },
    {
      "epoch": 0.010585131526319404,
      "grad_norm": 841.0155029296875,
      "learning_rate": 4.192000000000001e-06,
      "loss": 409.2921,
      "step": 2620
    },
    {
      "epoch": 0.010625532791687035,
      "grad_norm": 979.8448486328125,
      "learning_rate": 4.208e-06,
      "loss": 370.0784,
      "step": 2630
    },
    {
      "epoch": 0.010665934057054667,
      "grad_norm": 779.2421264648438,
      "learning_rate": 4.2240000000000006e-06,
      "loss": 327.6988,
      "step": 2640
    },
    {
      "epoch": 0.010706335322422298,
      "grad_norm": 1958.304443359375,
      "learning_rate": 4.24e-06,
      "loss": 270.3837,
      "step": 2650
    },
    {
      "epoch": 0.01074673658778993,
      "grad_norm": 612.7789916992188,
      "learning_rate": 4.256e-06,
      "loss": 208.5096,
      "step": 2660
    },
    {
      "epoch": 0.01078713785315756,
      "grad_norm": 913.646484375,
      "learning_rate": 4.272000000000001e-06,
      "loss": 326.1478,
      "step": 2670
    },
    {
      "epoch": 0.010827539118525192,
      "grad_norm": 609.9149780273438,
      "learning_rate": 4.288e-06,
      "loss": 323.9391,
      "step": 2680
    },
    {
      "epoch": 0.010867940383892823,
      "grad_norm": 4146.9677734375,
      "learning_rate": 4.304000000000001e-06,
      "loss": 331.8204,
      "step": 2690
    },
    {
      "epoch": 0.010908341649260455,
      "grad_norm": 1144.696044921875,
      "learning_rate": 4.32e-06,
      "loss": 344.9819,
      "step": 2700
    },
    {
      "epoch": 0.010948742914628086,
      "grad_norm": 1466.951904296875,
      "learning_rate": 4.3360000000000005e-06,
      "loss": 364.1662,
      "step": 2710
    },
    {
      "epoch": 0.010989144179995717,
      "grad_norm": 955.5458984375,
      "learning_rate": 4.352e-06,
      "loss": 284.9516,
      "step": 2720
    },
    {
      "epoch": 0.011029545445363349,
      "grad_norm": 591.5704345703125,
      "learning_rate": 4.368e-06,
      "loss": 279.7713,
      "step": 2730
    },
    {
      "epoch": 0.01106994671073098,
      "grad_norm": 1922.548095703125,
      "learning_rate": 4.384000000000001e-06,
      "loss": 269.3951,
      "step": 2740
    },
    {
      "epoch": 0.011110347976098611,
      "grad_norm": 705.4920654296875,
      "learning_rate": 4.4e-06,
      "loss": 202.7344,
      "step": 2750
    },
    {
      "epoch": 0.011150749241466243,
      "grad_norm": 6034.74951171875,
      "learning_rate": 4.416000000000001e-06,
      "loss": 319.9652,
      "step": 2760
    },
    {
      "epoch": 0.011191150506833874,
      "grad_norm": 1474.4591064453125,
      "learning_rate": 4.432e-06,
      "loss": 239.9669,
      "step": 2770
    },
    {
      "epoch": 0.011231551772201505,
      "grad_norm": 998.685302734375,
      "learning_rate": 4.4480000000000004e-06,
      "loss": 254.0082,
      "step": 2780
    },
    {
      "epoch": 0.011271953037569137,
      "grad_norm": 1730.3958740234375,
      "learning_rate": 4.464000000000001e-06,
      "loss": 285.7142,
      "step": 2790
    },
    {
      "epoch": 0.011312354302936768,
      "grad_norm": 666.0034790039062,
      "learning_rate": 4.48e-06,
      "loss": 243.7346,
      "step": 2800
    },
    {
      "epoch": 0.0113527555683044,
      "grad_norm": 1185.6009521484375,
      "learning_rate": 4.496000000000001e-06,
      "loss": 243.6461,
      "step": 2810
    },
    {
      "epoch": 0.01139315683367203,
      "grad_norm": 902.0358276367188,
      "learning_rate": 4.512e-06,
      "loss": 193.6116,
      "step": 2820
    },
    {
      "epoch": 0.011433558099039662,
      "grad_norm": 1310.3072509765625,
      "learning_rate": 4.5280000000000005e-06,
      "loss": 327.7516,
      "step": 2830
    },
    {
      "epoch": 0.011473959364407293,
      "grad_norm": 693.1176147460938,
      "learning_rate": 4.544000000000001e-06,
      "loss": 239.3185,
      "step": 2840
    },
    {
      "epoch": 0.011514360629774924,
      "grad_norm": 968.5886840820312,
      "learning_rate": 4.56e-06,
      "loss": 287.3573,
      "step": 2850
    },
    {
      "epoch": 0.011554761895142556,
      "grad_norm": 705.2077026367188,
      "learning_rate": 4.576000000000001e-06,
      "loss": 282.05,
      "step": 2860
    },
    {
      "epoch": 0.011595163160510187,
      "grad_norm": 1975.849365234375,
      "learning_rate": 4.592e-06,
      "loss": 296.7342,
      "step": 2870
    },
    {
      "epoch": 0.011635564425877818,
      "grad_norm": 746.0858764648438,
      "learning_rate": 4.608000000000001e-06,
      "loss": 204.9099,
      "step": 2880
    },
    {
      "epoch": 0.01167596569124545,
      "grad_norm": 1130.648681640625,
      "learning_rate": 4.624e-06,
      "loss": 342.3003,
      "step": 2890
    },
    {
      "epoch": 0.011716366956613081,
      "grad_norm": 1225.900634765625,
      "learning_rate": 4.6400000000000005e-06,
      "loss": 302.3731,
      "step": 2900
    },
    {
      "epoch": 0.011756768221980712,
      "grad_norm": 965.9699096679688,
      "learning_rate": 4.656000000000001e-06,
      "loss": 403.5073,
      "step": 2910
    },
    {
      "epoch": 0.011797169487348344,
      "grad_norm": 700.615966796875,
      "learning_rate": 4.672e-06,
      "loss": 220.3576,
      "step": 2920
    },
    {
      "epoch": 0.011837570752715975,
      "grad_norm": 744.499267578125,
      "learning_rate": 4.688000000000001e-06,
      "loss": 227.1823,
      "step": 2930
    },
    {
      "epoch": 0.011877972018083606,
      "grad_norm": 588.2945556640625,
      "learning_rate": 4.704e-06,
      "loss": 294.2805,
      "step": 2940
    },
    {
      "epoch": 0.011918373283451238,
      "grad_norm": 8179.02490234375,
      "learning_rate": 4.7200000000000005e-06,
      "loss": 360.6105,
      "step": 2950
    },
    {
      "epoch": 0.011958774548818869,
      "grad_norm": 4027.3515625,
      "learning_rate": 4.736000000000001e-06,
      "loss": 286.0583,
      "step": 2960
    },
    {
      "epoch": 0.0119991758141865,
      "grad_norm": 8620.037109375,
      "learning_rate": 4.752e-06,
      "loss": 338.6114,
      "step": 2970
    },
    {
      "epoch": 0.012039577079554132,
      "grad_norm": 1030.6036376953125,
      "learning_rate": 4.768000000000001e-06,
      "loss": 293.7208,
      "step": 2980
    },
    {
      "epoch": 0.012079978344921763,
      "grad_norm": 2113.48095703125,
      "learning_rate": 4.784e-06,
      "loss": 338.7042,
      "step": 2990
    },
    {
      "epoch": 0.012120379610289394,
      "grad_norm": 664.2764892578125,
      "learning_rate": 4.800000000000001e-06,
      "loss": 180.6681,
      "step": 3000
    },
    {
      "epoch": 0.012160780875657026,
      "grad_norm": 2052.64111328125,
      "learning_rate": 4.816e-06,
      "loss": 269.5311,
      "step": 3010
    },
    {
      "epoch": 0.012201182141024657,
      "grad_norm": 646.8975219726562,
      "learning_rate": 4.8320000000000005e-06,
      "loss": 199.238,
      "step": 3020
    },
    {
      "epoch": 0.012241583406392288,
      "grad_norm": 1087.690673828125,
      "learning_rate": 4.848000000000001e-06,
      "loss": 300.9158,
      "step": 3030
    },
    {
      "epoch": 0.01228198467175992,
      "grad_norm": 1012.8238525390625,
      "learning_rate": 4.864e-06,
      "loss": 405.1533,
      "step": 3040
    },
    {
      "epoch": 0.01232238593712755,
      "grad_norm": 1519.6015625,
      "learning_rate": 4.880000000000001e-06,
      "loss": 340.4835,
      "step": 3050
    },
    {
      "epoch": 0.012362787202495182,
      "grad_norm": 787.8037109375,
      "learning_rate": 4.896e-06,
      "loss": 249.336,
      "step": 3060
    },
    {
      "epoch": 0.012403188467862813,
      "grad_norm": 1030.4306640625,
      "learning_rate": 4.9120000000000006e-06,
      "loss": 248.3195,
      "step": 3070
    },
    {
      "epoch": 0.012443589733230445,
      "grad_norm": 1620.253173828125,
      "learning_rate": 4.928000000000001e-06,
      "loss": 210.1467,
      "step": 3080
    },
    {
      "epoch": 0.012483990998598076,
      "grad_norm": 2037.361328125,
      "learning_rate": 4.9440000000000004e-06,
      "loss": 331.2714,
      "step": 3090
    },
    {
      "epoch": 0.012524392263965707,
      "grad_norm": 4152.18701171875,
      "learning_rate": 4.960000000000001e-06,
      "loss": 369.5478,
      "step": 3100
    },
    {
      "epoch": 0.012564793529333339,
      "grad_norm": 751.2320556640625,
      "learning_rate": 4.976e-06,
      "loss": 339.7245,
      "step": 3110
    },
    {
      "epoch": 0.01260519479470097,
      "grad_norm": 1463.79833984375,
      "learning_rate": 4.992e-06,
      "loss": 349.4508,
      "step": 3120
    },
    {
      "epoch": 0.012645596060068601,
      "grad_norm": 931.794189453125,
      "learning_rate": 5.008000000000001e-06,
      "loss": 219.2512,
      "step": 3130
    },
    {
      "epoch": 0.012685997325436233,
      "grad_norm": 587.0802001953125,
      "learning_rate": 5.024e-06,
      "loss": 249.6072,
      "step": 3140
    },
    {
      "epoch": 0.012726398590803864,
      "grad_norm": 4629.9833984375,
      "learning_rate": 5.04e-06,
      "loss": 274.248,
      "step": 3150
    },
    {
      "epoch": 0.012766799856171495,
      "grad_norm": 1912.8997802734375,
      "learning_rate": 5.056000000000001e-06,
      "loss": 265.1541,
      "step": 3160
    },
    {
      "epoch": 0.012807201121539127,
      "grad_norm": 714.0695190429688,
      "learning_rate": 5.072e-06,
      "loss": 189.4337,
      "step": 3170
    },
    {
      "epoch": 0.012847602386906758,
      "grad_norm": 1278.3006591796875,
      "learning_rate": 5.088000000000001e-06,
      "loss": 368.549,
      "step": 3180
    },
    {
      "epoch": 0.01288800365227439,
      "grad_norm": 749.0835571289062,
      "learning_rate": 5.104e-06,
      "loss": 228.1246,
      "step": 3190
    },
    {
      "epoch": 0.01292840491764202,
      "grad_norm": 0.0,
      "learning_rate": 5.12e-06,
      "loss": 227.3482,
      "step": 3200
    },
    {
      "epoch": 0.012968806183009652,
      "grad_norm": 888.6599731445312,
      "learning_rate": 5.136e-06,
      "loss": 310.4809,
      "step": 3210
    },
    {
      "epoch": 0.013009207448377283,
      "grad_norm": 760.9104614257812,
      "learning_rate": 5.152e-06,
      "loss": 229.1965,
      "step": 3220
    },
    {
      "epoch": 0.013049608713744915,
      "grad_norm": 1096.23291015625,
      "learning_rate": 5.168000000000001e-06,
      "loss": 325.4543,
      "step": 3230
    },
    {
      "epoch": 0.013090009979112546,
      "grad_norm": 2388.175537109375,
      "learning_rate": 5.184e-06,
      "loss": 270.4536,
      "step": 3240
    },
    {
      "epoch": 0.013130411244480177,
      "grad_norm": 7733.79296875,
      "learning_rate": 5.2e-06,
      "loss": 364.0487,
      "step": 3250
    },
    {
      "epoch": 0.013170812509847809,
      "grad_norm": 1177.9654541015625,
      "learning_rate": 5.216e-06,
      "loss": 272.8813,
      "step": 3260
    },
    {
      "epoch": 0.01321121377521544,
      "grad_norm": 1095.551025390625,
      "learning_rate": 5.232e-06,
      "loss": 356.0726,
      "step": 3270
    },
    {
      "epoch": 0.013251615040583071,
      "grad_norm": 1486.8251953125,
      "learning_rate": 5.248000000000001e-06,
      "loss": 269.6024,
      "step": 3280
    },
    {
      "epoch": 0.013292016305950703,
      "grad_norm": 1649.369384765625,
      "learning_rate": 5.264e-06,
      "loss": 406.2197,
      "step": 3290
    },
    {
      "epoch": 0.013332417571318334,
      "grad_norm": 901.133544921875,
      "learning_rate": 5.28e-06,
      "loss": 334.5719,
      "step": 3300
    },
    {
      "epoch": 0.013372818836685965,
      "grad_norm": 1080.5753173828125,
      "learning_rate": 5.296e-06,
      "loss": 256.0706,
      "step": 3310
    },
    {
      "epoch": 0.013413220102053596,
      "grad_norm": 903.6046752929688,
      "learning_rate": 5.312e-06,
      "loss": 229.897,
      "step": 3320
    },
    {
      "epoch": 0.013453621367421228,
      "grad_norm": 984.148193359375,
      "learning_rate": 5.328000000000001e-06,
      "loss": 268.5443,
      "step": 3330
    },
    {
      "epoch": 0.013494022632788859,
      "grad_norm": 1167.3309326171875,
      "learning_rate": 5.344e-06,
      "loss": 339.0956,
      "step": 3340
    },
    {
      "epoch": 0.01353442389815649,
      "grad_norm": 2445.7763671875,
      "learning_rate": 5.36e-06,
      "loss": 237.0434,
      "step": 3350
    },
    {
      "epoch": 0.013574825163524122,
      "grad_norm": 814.3775634765625,
      "learning_rate": 5.376e-06,
      "loss": 182.9125,
      "step": 3360
    },
    {
      "epoch": 0.013615226428891753,
      "grad_norm": 883.6320190429688,
      "learning_rate": 5.392e-06,
      "loss": 346.1443,
      "step": 3370
    },
    {
      "epoch": 0.013655627694259384,
      "grad_norm": 788.3290405273438,
      "learning_rate": 5.408e-06,
      "loss": 143.6525,
      "step": 3380
    },
    {
      "epoch": 0.013696028959627016,
      "grad_norm": 1109.12744140625,
      "learning_rate": 5.424e-06,
      "loss": 293.3582,
      "step": 3390
    },
    {
      "epoch": 0.013736430224994647,
      "grad_norm": 968.6015014648438,
      "learning_rate": 5.4400000000000004e-06,
      "loss": 228.7989,
      "step": 3400
    },
    {
      "epoch": 0.013776831490362278,
      "grad_norm": 1213.0040283203125,
      "learning_rate": 5.456e-06,
      "loss": 432.3414,
      "step": 3410
    },
    {
      "epoch": 0.01381723275572991,
      "grad_norm": 1326.287109375,
      "learning_rate": 5.472e-06,
      "loss": 332.6496,
      "step": 3420
    },
    {
      "epoch": 0.013857634021097541,
      "grad_norm": 6259.54248046875,
      "learning_rate": 5.488e-06,
      "loss": 271.7205,
      "step": 3430
    },
    {
      "epoch": 0.013898035286465172,
      "grad_norm": 708.2656860351562,
      "learning_rate": 5.504e-06,
      "loss": 180.486,
      "step": 3440
    },
    {
      "epoch": 0.013938436551832804,
      "grad_norm": 685.2205200195312,
      "learning_rate": 5.5200000000000005e-06,
      "loss": 268.5384,
      "step": 3450
    },
    {
      "epoch": 0.013978837817200435,
      "grad_norm": 1125.2498779296875,
      "learning_rate": 5.536e-06,
      "loss": 288.8181,
      "step": 3460
    },
    {
      "epoch": 0.014019239082568066,
      "grad_norm": 1228.8125,
      "learning_rate": 5.552e-06,
      "loss": 244.6971,
      "step": 3470
    },
    {
      "epoch": 0.014059640347935698,
      "grad_norm": 2762.309814453125,
      "learning_rate": 5.568e-06,
      "loss": 273.8453,
      "step": 3480
    },
    {
      "epoch": 0.014100041613303329,
      "grad_norm": 945.8040161132812,
      "learning_rate": 5.584e-06,
      "loss": 240.8805,
      "step": 3490
    },
    {
      "epoch": 0.01414044287867096,
      "grad_norm": 2207.40185546875,
      "learning_rate": 5.600000000000001e-06,
      "loss": 165.5966,
      "step": 3500
    },
    {
      "epoch": 0.014180844144038592,
      "grad_norm": 622.099609375,
      "learning_rate": 5.616e-06,
      "loss": 202.4418,
      "step": 3510
    },
    {
      "epoch": 0.014221245409406223,
      "grad_norm": 1473.064208984375,
      "learning_rate": 5.6320000000000005e-06,
      "loss": 229.2896,
      "step": 3520
    },
    {
      "epoch": 0.014261646674773854,
      "grad_norm": 1057.5018310546875,
      "learning_rate": 5.648e-06,
      "loss": 211.9801,
      "step": 3530
    },
    {
      "epoch": 0.014302047940141486,
      "grad_norm": 4774.0947265625,
      "learning_rate": 5.664e-06,
      "loss": 204.8896,
      "step": 3540
    },
    {
      "epoch": 0.014342449205509117,
      "grad_norm": 676.964111328125,
      "learning_rate": 5.68e-06,
      "loss": 200.4312,
      "step": 3550
    },
    {
      "epoch": 0.014382850470876748,
      "grad_norm": 571.7015380859375,
      "learning_rate": 5.696e-06,
      "loss": 275.7348,
      "step": 3560
    },
    {
      "epoch": 0.01442325173624438,
      "grad_norm": 1923.3953857421875,
      "learning_rate": 5.7120000000000005e-06,
      "loss": 224.3793,
      "step": 3570
    },
    {
      "epoch": 0.01446365300161201,
      "grad_norm": 541.0182495117188,
      "learning_rate": 5.728e-06,
      "loss": 155.7405,
      "step": 3580
    },
    {
      "epoch": 0.014504054266979642,
      "grad_norm": 1030.9990234375,
      "learning_rate": 5.744e-06,
      "loss": 225.919,
      "step": 3590
    },
    {
      "epoch": 0.014544455532347273,
      "grad_norm": 1258.4619140625,
      "learning_rate": 5.76e-06,
      "loss": 242.7457,
      "step": 3600
    },
    {
      "epoch": 0.014584856797714905,
      "grad_norm": 1085.32177734375,
      "learning_rate": 5.776e-06,
      "loss": 251.3692,
      "step": 3610
    },
    {
      "epoch": 0.014625258063082536,
      "grad_norm": 880.9969482421875,
      "learning_rate": 5.792000000000001e-06,
      "loss": 228.1045,
      "step": 3620
    },
    {
      "epoch": 0.014665659328450167,
      "grad_norm": 616.6618041992188,
      "learning_rate": 5.808e-06,
      "loss": 216.2711,
      "step": 3630
    },
    {
      "epoch": 0.014706060593817799,
      "grad_norm": 2865.4169921875,
      "learning_rate": 5.8240000000000005e-06,
      "loss": 253.2795,
      "step": 3640
    },
    {
      "epoch": 0.01474646185918543,
      "grad_norm": 941.0968627929688,
      "learning_rate": 5.84e-06,
      "loss": 268.1058,
      "step": 3650
    },
    {
      "epoch": 0.014786863124553061,
      "grad_norm": 1495.3153076171875,
      "learning_rate": 5.856e-06,
      "loss": 262.6222,
      "step": 3660
    },
    {
      "epoch": 0.014827264389920693,
      "grad_norm": 1153.8773193359375,
      "learning_rate": 5.872000000000001e-06,
      "loss": 314.2995,
      "step": 3670
    },
    {
      "epoch": 0.014867665655288324,
      "grad_norm": 3272.336669921875,
      "learning_rate": 5.888e-06,
      "loss": 275.6237,
      "step": 3680
    },
    {
      "epoch": 0.014908066920655955,
      "grad_norm": 1363.574462890625,
      "learning_rate": 5.9040000000000006e-06,
      "loss": 408.7946,
      "step": 3690
    },
    {
      "epoch": 0.014948468186023587,
      "grad_norm": 1989.05810546875,
      "learning_rate": 5.92e-06,
      "loss": 347.1884,
      "step": 3700
    },
    {
      "epoch": 0.014988869451391218,
      "grad_norm": 893.3648071289062,
      "learning_rate": 5.9360000000000004e-06,
      "loss": 231.8197,
      "step": 3710
    },
    {
      "epoch": 0.01502927071675885,
      "grad_norm": 2015.3258056640625,
      "learning_rate": 5.952e-06,
      "loss": 241.9273,
      "step": 3720
    },
    {
      "epoch": 0.01506967198212648,
      "grad_norm": 1097.1138916015625,
      "learning_rate": 5.968e-06,
      "loss": 289.7548,
      "step": 3730
    },
    {
      "epoch": 0.015110073247494112,
      "grad_norm": 1198.3504638671875,
      "learning_rate": 5.984000000000001e-06,
      "loss": 257.8569,
      "step": 3740
    },
    {
      "epoch": 0.015150474512861743,
      "grad_norm": 2764.9599609375,
      "learning_rate": 6e-06,
      "loss": 394.4565,
      "step": 3750
    },
    {
      "epoch": 0.015190875778229375,
      "grad_norm": 758.348876953125,
      "learning_rate": 6.0160000000000005e-06,
      "loss": 249.2975,
      "step": 3760
    },
    {
      "epoch": 0.015231277043597006,
      "grad_norm": 2277.13818359375,
      "learning_rate": 6.032e-06,
      "loss": 294.9143,
      "step": 3770
    },
    {
      "epoch": 0.015271678308964637,
      "grad_norm": 2135.546142578125,
      "learning_rate": 6.048e-06,
      "loss": 235.1545,
      "step": 3780
    },
    {
      "epoch": 0.015312079574332269,
      "grad_norm": 1097.388916015625,
      "learning_rate": 6.064000000000001e-06,
      "loss": 241.006,
      "step": 3790
    },
    {
      "epoch": 0.0153524808396999,
      "grad_norm": 943.5238647460938,
      "learning_rate": 6.08e-06,
      "loss": 174.3143,
      "step": 3800
    },
    {
      "epoch": 0.015392882105067531,
      "grad_norm": 1239.0521240234375,
      "learning_rate": 6.096000000000001e-06,
      "loss": 307.3331,
      "step": 3810
    },
    {
      "epoch": 0.015433283370435162,
      "grad_norm": 643.6400146484375,
      "learning_rate": 6.112e-06,
      "loss": 217.4633,
      "step": 3820
    },
    {
      "epoch": 0.015473684635802794,
      "grad_norm": 1569.7869873046875,
      "learning_rate": 6.1280000000000005e-06,
      "loss": 322.4538,
      "step": 3830
    },
    {
      "epoch": 0.015514085901170425,
      "grad_norm": 3763.859130859375,
      "learning_rate": 6.144e-06,
      "loss": 182.1174,
      "step": 3840
    },
    {
      "epoch": 0.015554487166538056,
      "grad_norm": 717.2794799804688,
      "learning_rate": 6.16e-06,
      "loss": 207.8513,
      "step": 3850
    },
    {
      "epoch": 0.015594888431905688,
      "grad_norm": 1418.3277587890625,
      "learning_rate": 6.176000000000001e-06,
      "loss": 254.4121,
      "step": 3860
    },
    {
      "epoch": 0.01563528969727332,
      "grad_norm": 1528.9119873046875,
      "learning_rate": 6.192e-06,
      "loss": 223.6123,
      "step": 3870
    },
    {
      "epoch": 0.01567569096264095,
      "grad_norm": 544.6751708984375,
      "learning_rate": 6.2080000000000005e-06,
      "loss": 222.1293,
      "step": 3880
    },
    {
      "epoch": 0.01571609222800858,
      "grad_norm": 754.9981079101562,
      "learning_rate": 6.224e-06,
      "loss": 255.1399,
      "step": 3890
    },
    {
      "epoch": 0.01575649349337621,
      "grad_norm": 1966.5140380859375,
      "learning_rate": 6.24e-06,
      "loss": 275.0173,
      "step": 3900
    },
    {
      "epoch": 0.015796894758743844,
      "grad_norm": 810.1495361328125,
      "learning_rate": 6.256000000000001e-06,
      "loss": 219.6732,
      "step": 3910
    },
    {
      "epoch": 0.015837296024111474,
      "grad_norm": 3056.527099609375,
      "learning_rate": 6.272e-06,
      "loss": 184.322,
      "step": 3920
    },
    {
      "epoch": 0.015877697289479107,
      "grad_norm": 2820.2568359375,
      "learning_rate": 6.288000000000001e-06,
      "loss": 235.3666,
      "step": 3930
    },
    {
      "epoch": 0.015918098554846737,
      "grad_norm": 1320.5225830078125,
      "learning_rate": 6.304e-06,
      "loss": 220.3165,
      "step": 3940
    },
    {
      "epoch": 0.01595849982021437,
      "grad_norm": 933.7466430664062,
      "learning_rate": 6.3200000000000005e-06,
      "loss": 307.4496,
      "step": 3950
    },
    {
      "epoch": 0.015998901085582,
      "grad_norm": 867.0122680664062,
      "learning_rate": 6.336000000000001e-06,
      "loss": 263.1927,
      "step": 3960
    },
    {
      "epoch": 0.016039302350949632,
      "grad_norm": 747.3133544921875,
      "learning_rate": 6.352e-06,
      "loss": 306.4962,
      "step": 3970
    },
    {
      "epoch": 0.016079703616317262,
      "grad_norm": 711.7617797851562,
      "learning_rate": 6.368000000000001e-06,
      "loss": 243.1562,
      "step": 3980
    },
    {
      "epoch": 0.016120104881684895,
      "grad_norm": 732.2003173828125,
      "learning_rate": 6.384e-06,
      "loss": 231.337,
      "step": 3990
    },
    {
      "epoch": 0.016160506147052525,
      "grad_norm": 582.39990234375,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 273.9559,
      "step": 4000
    },
    {
      "epoch": 0.016200907412420158,
      "grad_norm": 572.5735473632812,
      "learning_rate": 6.416e-06,
      "loss": 212.5563,
      "step": 4010
    },
    {
      "epoch": 0.016241308677787787,
      "grad_norm": 0.0,
      "learning_rate": 6.432e-06,
      "loss": 224.8545,
      "step": 4020
    },
    {
      "epoch": 0.01628170994315542,
      "grad_norm": 887.3201293945312,
      "learning_rate": 6.448000000000001e-06,
      "loss": 271.5942,
      "step": 4030
    },
    {
      "epoch": 0.01632211120852305,
      "grad_norm": 1036.5596923828125,
      "learning_rate": 6.464e-06,
      "loss": 324.4769,
      "step": 4040
    },
    {
      "epoch": 0.016362512473890683,
      "grad_norm": 1676.9654541015625,
      "learning_rate": 6.480000000000001e-06,
      "loss": 221.9381,
      "step": 4050
    },
    {
      "epoch": 0.016402913739258312,
      "grad_norm": 1006.4035034179688,
      "learning_rate": 6.496e-06,
      "loss": 330.1907,
      "step": 4060
    },
    {
      "epoch": 0.016443315004625945,
      "grad_norm": 1340.500244140625,
      "learning_rate": 6.5120000000000005e-06,
      "loss": 241.6059,
      "step": 4070
    },
    {
      "epoch": 0.016483716269993575,
      "grad_norm": 824.397705078125,
      "learning_rate": 6.528000000000001e-06,
      "loss": 237.4955,
      "step": 4080
    },
    {
      "epoch": 0.016524117535361208,
      "grad_norm": 623.558837890625,
      "learning_rate": 6.544e-06,
      "loss": 338.69,
      "step": 4090
    },
    {
      "epoch": 0.016564518800728838,
      "grad_norm": 1017.6536865234375,
      "learning_rate": 6.560000000000001e-06,
      "loss": 250.0294,
      "step": 4100
    },
    {
      "epoch": 0.01660492006609647,
      "grad_norm": 708.3771362304688,
      "learning_rate": 6.576e-06,
      "loss": 201.3914,
      "step": 4110
    },
    {
      "epoch": 0.0166453213314641,
      "grad_norm": 609.1826171875,
      "learning_rate": 6.592000000000001e-06,
      "loss": 320.5673,
      "step": 4120
    },
    {
      "epoch": 0.016685722596831733,
      "grad_norm": 833.898193359375,
      "learning_rate": 6.608000000000001e-06,
      "loss": 214.4177,
      "step": 4130
    },
    {
      "epoch": 0.016726123862199363,
      "grad_norm": 1081.476318359375,
      "learning_rate": 6.6240000000000004e-06,
      "loss": 264.2287,
      "step": 4140
    },
    {
      "epoch": 0.016766525127566996,
      "grad_norm": 1495.6732177734375,
      "learning_rate": 6.640000000000001e-06,
      "loss": 315.7932,
      "step": 4150
    },
    {
      "epoch": 0.016806926392934626,
      "grad_norm": 838.46630859375,
      "learning_rate": 6.656e-06,
      "loss": 280.6871,
      "step": 4160
    },
    {
      "epoch": 0.01684732765830226,
      "grad_norm": 1784.915283203125,
      "learning_rate": 6.672000000000001e-06,
      "loss": 342.9094,
      "step": 4170
    },
    {
      "epoch": 0.016887728923669888,
      "grad_norm": 2573.862548828125,
      "learning_rate": 6.688e-06,
      "loss": 242.5726,
      "step": 4180
    },
    {
      "epoch": 0.01692813018903752,
      "grad_norm": 604.2163696289062,
      "learning_rate": 6.7040000000000005e-06,
      "loss": 243.26,
      "step": 4190
    },
    {
      "epoch": 0.01696853145440515,
      "grad_norm": 1083.6295166015625,
      "learning_rate": 6.720000000000001e-06,
      "loss": 174.146,
      "step": 4200
    },
    {
      "epoch": 0.017008932719772784,
      "grad_norm": 844.8646850585938,
      "learning_rate": 6.736e-06,
      "loss": 270.4487,
      "step": 4210
    },
    {
      "epoch": 0.017049333985140414,
      "grad_norm": 555.3699340820312,
      "learning_rate": 6.752000000000001e-06,
      "loss": 185.4077,
      "step": 4220
    },
    {
      "epoch": 0.017089735250508047,
      "grad_norm": 1184.4735107421875,
      "learning_rate": 6.768e-06,
      "loss": 287.9112,
      "step": 4230
    },
    {
      "epoch": 0.017130136515875676,
      "grad_norm": 699.8986206054688,
      "learning_rate": 6.784000000000001e-06,
      "loss": 312.1473,
      "step": 4240
    },
    {
      "epoch": 0.01717053778124331,
      "grad_norm": 1370.1610107421875,
      "learning_rate": 6.800000000000001e-06,
      "loss": 267.4715,
      "step": 4250
    },
    {
      "epoch": 0.01721093904661094,
      "grad_norm": 735.01123046875,
      "learning_rate": 6.8160000000000005e-06,
      "loss": 259.6105,
      "step": 4260
    },
    {
      "epoch": 0.017251340311978572,
      "grad_norm": 768.6294555664062,
      "learning_rate": 6.832000000000001e-06,
      "loss": 195.3645,
      "step": 4270
    },
    {
      "epoch": 0.0172917415773462,
      "grad_norm": 1129.20068359375,
      "learning_rate": 6.848e-06,
      "loss": 333.3923,
      "step": 4280
    },
    {
      "epoch": 0.017332142842713835,
      "grad_norm": 620.0527954101562,
      "learning_rate": 6.864000000000001e-06,
      "loss": 140.4836,
      "step": 4290
    },
    {
      "epoch": 0.017372544108081464,
      "grad_norm": 1524.297119140625,
      "learning_rate": 6.88e-06,
      "loss": 234.0078,
      "step": 4300
    },
    {
      "epoch": 0.017412945373449097,
      "grad_norm": 1074.130126953125,
      "learning_rate": 6.8960000000000006e-06,
      "loss": 333.4636,
      "step": 4310
    },
    {
      "epoch": 0.017453346638816727,
      "grad_norm": 1849.30419921875,
      "learning_rate": 6.912000000000001e-06,
      "loss": 250.6815,
      "step": 4320
    },
    {
      "epoch": 0.01749374790418436,
      "grad_norm": 953.8714599609375,
      "learning_rate": 6.928e-06,
      "loss": 168.5187,
      "step": 4330
    },
    {
      "epoch": 0.01753414916955199,
      "grad_norm": 1707.04296875,
      "learning_rate": 6.944000000000001e-06,
      "loss": 247.184,
      "step": 4340
    },
    {
      "epoch": 0.017574550434919622,
      "grad_norm": 1027.0791015625,
      "learning_rate": 6.96e-06,
      "loss": 212.3976,
      "step": 4350
    },
    {
      "epoch": 0.017614951700287252,
      "grad_norm": 1421.16650390625,
      "learning_rate": 6.976000000000001e-06,
      "loss": 196.0206,
      "step": 4360
    },
    {
      "epoch": 0.017655352965654885,
      "grad_norm": 2190.223388671875,
      "learning_rate": 6.992000000000001e-06,
      "loss": 245.279,
      "step": 4370
    },
    {
      "epoch": 0.017695754231022515,
      "grad_norm": 607.2709350585938,
      "learning_rate": 7.0080000000000005e-06,
      "loss": 269.9326,
      "step": 4380
    },
    {
      "epoch": 0.017736155496390148,
      "grad_norm": 905.8397216796875,
      "learning_rate": 7.024000000000001e-06,
      "loss": 197.5017,
      "step": 4390
    },
    {
      "epoch": 0.017776556761757777,
      "grad_norm": 829.648681640625,
      "learning_rate": 7.04e-06,
      "loss": 306.7443,
      "step": 4400
    },
    {
      "epoch": 0.01781695802712541,
      "grad_norm": 750.1204223632812,
      "learning_rate": 7.056000000000001e-06,
      "loss": 208.2544,
      "step": 4410
    },
    {
      "epoch": 0.01785735929249304,
      "grad_norm": 1103.8912353515625,
      "learning_rate": 7.072000000000001e-06,
      "loss": 208.2772,
      "step": 4420
    },
    {
      "epoch": 0.017897760557860673,
      "grad_norm": 2012.5870361328125,
      "learning_rate": 7.088000000000001e-06,
      "loss": 227.8844,
      "step": 4430
    },
    {
      "epoch": 0.017938161823228303,
      "grad_norm": 1063.1541748046875,
      "learning_rate": 7.104000000000001e-06,
      "loss": 197.1388,
      "step": 4440
    },
    {
      "epoch": 0.017978563088595936,
      "grad_norm": 1133.552978515625,
      "learning_rate": 7.1200000000000004e-06,
      "loss": 212.4223,
      "step": 4450
    },
    {
      "epoch": 0.018018964353963565,
      "grad_norm": 808.1385498046875,
      "learning_rate": 7.136000000000001e-06,
      "loss": 224.3651,
      "step": 4460
    },
    {
      "epoch": 0.0180593656193312,
      "grad_norm": 1333.4466552734375,
      "learning_rate": 7.152e-06,
      "loss": 213.2375,
      "step": 4470
    },
    {
      "epoch": 0.018099766884698828,
      "grad_norm": 1058.0499267578125,
      "learning_rate": 7.168000000000001e-06,
      "loss": 239.6389,
      "step": 4480
    },
    {
      "epoch": 0.01814016815006646,
      "grad_norm": 1192.045654296875,
      "learning_rate": 7.184000000000001e-06,
      "loss": 201.7774,
      "step": 4490
    },
    {
      "epoch": 0.01818056941543409,
      "grad_norm": 761.6637573242188,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 239.5527,
      "step": 4500
    },
    {
      "epoch": 0.018220970680801724,
      "grad_norm": 1080.089111328125,
      "learning_rate": 7.216000000000001e-06,
      "loss": 220.7039,
      "step": 4510
    },
    {
      "epoch": 0.018261371946169353,
      "grad_norm": 900.0491943359375,
      "learning_rate": 7.232e-06,
      "loss": 202.6598,
      "step": 4520
    },
    {
      "epoch": 0.018301773211536986,
      "grad_norm": 1005.7420043945312,
      "learning_rate": 7.248000000000001e-06,
      "loss": 226.7294,
      "step": 4530
    },
    {
      "epoch": 0.018342174476904616,
      "grad_norm": 942.6360473632812,
      "learning_rate": 7.264000000000001e-06,
      "loss": 188.8948,
      "step": 4540
    },
    {
      "epoch": 0.01838257574227225,
      "grad_norm": 968.3139038085938,
      "learning_rate": 7.280000000000001e-06,
      "loss": 313.6362,
      "step": 4550
    },
    {
      "epoch": 0.01842297700763988,
      "grad_norm": 1409.2060546875,
      "learning_rate": 7.296000000000001e-06,
      "loss": 322.2942,
      "step": 4560
    },
    {
      "epoch": 0.01846337827300751,
      "grad_norm": 814.0696411132812,
      "learning_rate": 7.3120000000000005e-06,
      "loss": 222.029,
      "step": 4570
    },
    {
      "epoch": 0.01850377953837514,
      "grad_norm": 809.8764038085938,
      "learning_rate": 7.328000000000001e-06,
      "loss": 273.6918,
      "step": 4580
    },
    {
      "epoch": 0.018544180803742774,
      "grad_norm": 513.40625,
      "learning_rate": 7.344000000000001e-06,
      "loss": 307.8005,
      "step": 4590
    },
    {
      "epoch": 0.018584582069110404,
      "grad_norm": 4324.00146484375,
      "learning_rate": 7.360000000000001e-06,
      "loss": 303.7309,
      "step": 4600
    },
    {
      "epoch": 0.018624983334478037,
      "grad_norm": 836.23193359375,
      "learning_rate": 7.376000000000001e-06,
      "loss": 204.3285,
      "step": 4610
    },
    {
      "epoch": 0.018665384599845666,
      "grad_norm": 579.7659301757812,
      "learning_rate": 7.3920000000000005e-06,
      "loss": 175.3677,
      "step": 4620
    },
    {
      "epoch": 0.0187057858652133,
      "grad_norm": 1128.6890869140625,
      "learning_rate": 7.408000000000001e-06,
      "loss": 183.0995,
      "step": 4630
    },
    {
      "epoch": 0.01874618713058093,
      "grad_norm": 1351.5643310546875,
      "learning_rate": 7.424e-06,
      "loss": 184.3664,
      "step": 4640
    },
    {
      "epoch": 0.018786588395948562,
      "grad_norm": 2014.1749267578125,
      "learning_rate": 7.440000000000001e-06,
      "loss": 270.0711,
      "step": 4650
    },
    {
      "epoch": 0.01882698966131619,
      "grad_norm": 1736.61279296875,
      "learning_rate": 7.456000000000001e-06,
      "loss": 288.5561,
      "step": 4660
    },
    {
      "epoch": 0.018867390926683825,
      "grad_norm": 725.383544921875,
      "learning_rate": 7.472000000000001e-06,
      "loss": 279.8696,
      "step": 4670
    },
    {
      "epoch": 0.018907792192051454,
      "grad_norm": 1754.6114501953125,
      "learning_rate": 7.488000000000001e-06,
      "loss": 201.6906,
      "step": 4680
    },
    {
      "epoch": 0.018948193457419087,
      "grad_norm": 1377.6466064453125,
      "learning_rate": 7.5040000000000005e-06,
      "loss": 200.6483,
      "step": 4690
    },
    {
      "epoch": 0.018988594722786717,
      "grad_norm": 2007.0408935546875,
      "learning_rate": 7.520000000000001e-06,
      "loss": 232.2962,
      "step": 4700
    },
    {
      "epoch": 0.01902899598815435,
      "grad_norm": 1615.7125244140625,
      "learning_rate": 7.536000000000001e-06,
      "loss": 245.0112,
      "step": 4710
    },
    {
      "epoch": 0.01906939725352198,
      "grad_norm": 0.0,
      "learning_rate": 7.552000000000001e-06,
      "loss": 189.8007,
      "step": 4720
    },
    {
      "epoch": 0.019109798518889613,
      "grad_norm": 1523.142333984375,
      "learning_rate": 7.568000000000001e-06,
      "loss": 194.2404,
      "step": 4730
    },
    {
      "epoch": 0.019150199784257242,
      "grad_norm": 2107.94677734375,
      "learning_rate": 7.5840000000000006e-06,
      "loss": 254.0392,
      "step": 4740
    },
    {
      "epoch": 0.019190601049624875,
      "grad_norm": 858.9860229492188,
      "learning_rate": 7.600000000000001e-06,
      "loss": 260.6963,
      "step": 4750
    },
    {
      "epoch": 0.019231002314992505,
      "grad_norm": 801.8906860351562,
      "learning_rate": 7.616000000000001e-06,
      "loss": 210.6171,
      "step": 4760
    },
    {
      "epoch": 0.019271403580360138,
      "grad_norm": 763.8599243164062,
      "learning_rate": 7.632e-06,
      "loss": 244.6692,
      "step": 4770
    },
    {
      "epoch": 0.019311804845727767,
      "grad_norm": 1143.01171875,
      "learning_rate": 7.648e-06,
      "loss": 232.0638,
      "step": 4780
    },
    {
      "epoch": 0.0193522061110954,
      "grad_norm": 473.067626953125,
      "learning_rate": 7.664e-06,
      "loss": 163.634,
      "step": 4790
    },
    {
      "epoch": 0.01939260737646303,
      "grad_norm": 738.6350708007812,
      "learning_rate": 7.680000000000001e-06,
      "loss": 171.6407,
      "step": 4800
    },
    {
      "epoch": 0.019433008641830663,
      "grad_norm": 707.3241577148438,
      "learning_rate": 7.696e-06,
      "loss": 225.7683,
      "step": 4810
    },
    {
      "epoch": 0.019473409907198293,
      "grad_norm": 1966.50927734375,
      "learning_rate": 7.712e-06,
      "loss": 259.4346,
      "step": 4820
    },
    {
      "epoch": 0.019513811172565926,
      "grad_norm": 2337.94580078125,
      "learning_rate": 7.728000000000001e-06,
      "loss": 210.5883,
      "step": 4830
    },
    {
      "epoch": 0.019554212437933555,
      "grad_norm": 1208.361083984375,
      "learning_rate": 7.744e-06,
      "loss": 166.7386,
      "step": 4840
    },
    {
      "epoch": 0.01959461370330119,
      "grad_norm": 1413.30126953125,
      "learning_rate": 7.76e-06,
      "loss": 291.6949,
      "step": 4850
    },
    {
      "epoch": 0.019635014968668818,
      "grad_norm": 422.9515380859375,
      "learning_rate": 7.776e-06,
      "loss": 161.5541,
      "step": 4860
    },
    {
      "epoch": 0.01967541623403645,
      "grad_norm": 2810.00927734375,
      "learning_rate": 7.792000000000001e-06,
      "loss": 287.8005,
      "step": 4870
    },
    {
      "epoch": 0.01971581749940408,
      "grad_norm": 1965.0252685546875,
      "learning_rate": 7.808e-06,
      "loss": 194.2817,
      "step": 4880
    },
    {
      "epoch": 0.019756218764771714,
      "grad_norm": 1410.799560546875,
      "learning_rate": 7.824e-06,
      "loss": 180.6805,
      "step": 4890
    },
    {
      "epoch": 0.019796620030139343,
      "grad_norm": 857.7700805664062,
      "learning_rate": 7.840000000000001e-06,
      "loss": 295.3197,
      "step": 4900
    },
    {
      "epoch": 0.019837021295506976,
      "grad_norm": 1902.400146484375,
      "learning_rate": 7.856e-06,
      "loss": 278.1652,
      "step": 4910
    },
    {
      "epoch": 0.019877422560874606,
      "grad_norm": 897.0117797851562,
      "learning_rate": 7.872e-06,
      "loss": 154.4284,
      "step": 4920
    },
    {
      "epoch": 0.01991782382624224,
      "grad_norm": 1253.6666259765625,
      "learning_rate": 7.888e-06,
      "loss": 261.9078,
      "step": 4930
    },
    {
      "epoch": 0.01995822509160987,
      "grad_norm": 928.166748046875,
      "learning_rate": 7.904000000000001e-06,
      "loss": 260.5916,
      "step": 4940
    },
    {
      "epoch": 0.0199986263569775,
      "grad_norm": 1244.7052001953125,
      "learning_rate": 7.92e-06,
      "loss": 216.8228,
      "step": 4950
    },
    {
      "epoch": 0.02003902762234513,
      "grad_norm": 1885.830078125,
      "learning_rate": 7.936e-06,
      "loss": 138.4373,
      "step": 4960
    },
    {
      "epoch": 0.020079428887712764,
      "grad_norm": 10677.4921875,
      "learning_rate": 7.952000000000001e-06,
      "loss": 370.5018,
      "step": 4970
    },
    {
      "epoch": 0.020119830153080394,
      "grad_norm": 1309.8223876953125,
      "learning_rate": 7.968e-06,
      "loss": 235.4404,
      "step": 4980
    },
    {
      "epoch": 0.020160231418448027,
      "grad_norm": 633.9755249023438,
      "learning_rate": 7.984e-06,
      "loss": 255.8174,
      "step": 4990
    },
    {
      "epoch": 0.020200632683815656,
      "grad_norm": 1383.577880859375,
      "learning_rate": 8.000000000000001e-06,
      "loss": 177.3417,
      "step": 5000
    },
    {
      "epoch": 0.02024103394918329,
      "grad_norm": 1378.894287109375,
      "learning_rate": 8.016e-06,
      "loss": 225.4421,
      "step": 5010
    },
    {
      "epoch": 0.02028143521455092,
      "grad_norm": 1220.368896484375,
      "learning_rate": 8.032e-06,
      "loss": 147.1734,
      "step": 5020
    },
    {
      "epoch": 0.020321836479918552,
      "grad_norm": 1136.2757568359375,
      "learning_rate": 8.048e-06,
      "loss": 257.1214,
      "step": 5030
    },
    {
      "epoch": 0.020362237745286182,
      "grad_norm": 1104.6431884765625,
      "learning_rate": 8.064000000000001e-06,
      "loss": 246.424,
      "step": 5040
    },
    {
      "epoch": 0.020402639010653815,
      "grad_norm": 1294.03857421875,
      "learning_rate": 8.08e-06,
      "loss": 184.7015,
      "step": 5050
    },
    {
      "epoch": 0.020443040276021444,
      "grad_norm": 0.0,
      "learning_rate": 8.096e-06,
      "loss": 274.0798,
      "step": 5060
    },
    {
      "epoch": 0.020483441541389077,
      "grad_norm": 1032.7880859375,
      "learning_rate": 8.112000000000001e-06,
      "loss": 209.0563,
      "step": 5070
    },
    {
      "epoch": 0.020523842806756707,
      "grad_norm": 2536.4736328125,
      "learning_rate": 8.128e-06,
      "loss": 229.0236,
      "step": 5080
    },
    {
      "epoch": 0.02056424407212434,
      "grad_norm": 1960.7764892578125,
      "learning_rate": 8.144e-06,
      "loss": 290.2731,
      "step": 5090
    },
    {
      "epoch": 0.02060464533749197,
      "grad_norm": 544.7542114257812,
      "learning_rate": 8.16e-06,
      "loss": 217.8538,
      "step": 5100
    },
    {
      "epoch": 0.020645046602859603,
      "grad_norm": 881.3969116210938,
      "learning_rate": 8.176000000000001e-06,
      "loss": 253.8835,
      "step": 5110
    },
    {
      "epoch": 0.020685447868227232,
      "grad_norm": 933.0445556640625,
      "learning_rate": 8.192e-06,
      "loss": 199.286,
      "step": 5120
    },
    {
      "epoch": 0.020725849133594865,
      "grad_norm": 2709.8310546875,
      "learning_rate": 8.208e-06,
      "loss": 182.7853,
      "step": 5130
    },
    {
      "epoch": 0.020766250398962495,
      "grad_norm": 641.6781616210938,
      "learning_rate": 8.224000000000001e-06,
      "loss": 179.5877,
      "step": 5140
    },
    {
      "epoch": 0.020806651664330128,
      "grad_norm": 785.1431274414062,
      "learning_rate": 8.24e-06,
      "loss": 156.9568,
      "step": 5150
    },
    {
      "epoch": 0.020847052929697758,
      "grad_norm": 755.727783203125,
      "learning_rate": 8.256e-06,
      "loss": 220.8312,
      "step": 5160
    },
    {
      "epoch": 0.02088745419506539,
      "grad_norm": 1180.8179931640625,
      "learning_rate": 8.272000000000001e-06,
      "loss": 246.3089,
      "step": 5170
    },
    {
      "epoch": 0.02092785546043302,
      "grad_norm": 2293.158203125,
      "learning_rate": 8.288000000000001e-06,
      "loss": 205.6415,
      "step": 5180
    },
    {
      "epoch": 0.020968256725800653,
      "grad_norm": 842.48828125,
      "learning_rate": 8.304e-06,
      "loss": 249.575,
      "step": 5190
    },
    {
      "epoch": 0.021008657991168283,
      "grad_norm": 790.164794921875,
      "learning_rate": 8.32e-06,
      "loss": 245.8863,
      "step": 5200
    },
    {
      "epoch": 0.021049059256535916,
      "grad_norm": 1180.41650390625,
      "learning_rate": 8.336000000000001e-06,
      "loss": 221.1177,
      "step": 5210
    },
    {
      "epoch": 0.021089460521903546,
      "grad_norm": 999.559326171875,
      "learning_rate": 8.352e-06,
      "loss": 220.637,
      "step": 5220
    },
    {
      "epoch": 0.02112986178727118,
      "grad_norm": 1407.5498046875,
      "learning_rate": 8.368e-06,
      "loss": 245.8358,
      "step": 5230
    },
    {
      "epoch": 0.021170263052638808,
      "grad_norm": 1201.23486328125,
      "learning_rate": 8.384000000000001e-06,
      "loss": 303.6925,
      "step": 5240
    },
    {
      "epoch": 0.02121066431800644,
      "grad_norm": 3144.137939453125,
      "learning_rate": 8.400000000000001e-06,
      "loss": 256.0646,
      "step": 5250
    },
    {
      "epoch": 0.02125106558337407,
      "grad_norm": 1443.5968017578125,
      "learning_rate": 8.416e-06,
      "loss": 278.0506,
      "step": 5260
    },
    {
      "epoch": 0.021291466848741704,
      "grad_norm": 1761.7537841796875,
      "learning_rate": 8.432e-06,
      "loss": 213.0941,
      "step": 5270
    },
    {
      "epoch": 0.021331868114109333,
      "grad_norm": 1119.3565673828125,
      "learning_rate": 8.448000000000001e-06,
      "loss": 253.8438,
      "step": 5280
    },
    {
      "epoch": 0.021372269379476967,
      "grad_norm": 1161.480224609375,
      "learning_rate": 8.464e-06,
      "loss": 240.5102,
      "step": 5290
    },
    {
      "epoch": 0.021412670644844596,
      "grad_norm": 932.6578979492188,
      "learning_rate": 8.48e-06,
      "loss": 269.217,
      "step": 5300
    },
    {
      "epoch": 0.02145307191021223,
      "grad_norm": 2974.05712890625,
      "learning_rate": 8.496000000000001e-06,
      "loss": 275.3746,
      "step": 5310
    },
    {
      "epoch": 0.02149347317557986,
      "grad_norm": 1162.4228515625,
      "learning_rate": 8.512e-06,
      "loss": 203.4607,
      "step": 5320
    },
    {
      "epoch": 0.021533874440947492,
      "grad_norm": 818.4716186523438,
      "learning_rate": 8.528e-06,
      "loss": 232.1441,
      "step": 5330
    },
    {
      "epoch": 0.02157427570631512,
      "grad_norm": 1251.9176025390625,
      "learning_rate": 8.544000000000002e-06,
      "loss": 200.7606,
      "step": 5340
    },
    {
      "epoch": 0.021614676971682754,
      "grad_norm": 675.2981567382812,
      "learning_rate": 8.560000000000001e-06,
      "loss": 222.9481,
      "step": 5350
    },
    {
      "epoch": 0.021655078237050384,
      "grad_norm": 722.9965209960938,
      "learning_rate": 8.576e-06,
      "loss": 295.3248,
      "step": 5360
    },
    {
      "epoch": 0.021695479502418017,
      "grad_norm": 787.6085815429688,
      "learning_rate": 8.592e-06,
      "loss": 237.5749,
      "step": 5370
    },
    {
      "epoch": 0.021735880767785647,
      "grad_norm": 1618.6192626953125,
      "learning_rate": 8.608000000000001e-06,
      "loss": 273.413,
      "step": 5380
    },
    {
      "epoch": 0.02177628203315328,
      "grad_norm": 2153.603271484375,
      "learning_rate": 8.624e-06,
      "loss": 251.1006,
      "step": 5390
    },
    {
      "epoch": 0.02181668329852091,
      "grad_norm": 3269.96044921875,
      "learning_rate": 8.64e-06,
      "loss": 306.5505,
      "step": 5400
    },
    {
      "epoch": 0.021857084563888542,
      "grad_norm": 735.1317138671875,
      "learning_rate": 8.656000000000001e-06,
      "loss": 251.5622,
      "step": 5410
    },
    {
      "epoch": 0.021897485829256172,
      "grad_norm": 3821.17919921875,
      "learning_rate": 8.672000000000001e-06,
      "loss": 254.9936,
      "step": 5420
    },
    {
      "epoch": 0.021937887094623805,
      "grad_norm": 6532.201171875,
      "learning_rate": 8.688e-06,
      "loss": 255.663,
      "step": 5430
    },
    {
      "epoch": 0.021978288359991435,
      "grad_norm": 1257.637939453125,
      "learning_rate": 8.704e-06,
      "loss": 236.9894,
      "step": 5440
    },
    {
      "epoch": 0.022018689625359068,
      "grad_norm": 18393.888671875,
      "learning_rate": 8.720000000000001e-06,
      "loss": 342.399,
      "step": 5450
    },
    {
      "epoch": 0.022059090890726697,
      "grad_norm": 916.3475341796875,
      "learning_rate": 8.736e-06,
      "loss": 219.2116,
      "step": 5460
    },
    {
      "epoch": 0.02209949215609433,
      "grad_norm": 1160.2779541015625,
      "learning_rate": 8.752e-06,
      "loss": 218.9973,
      "step": 5470
    },
    {
      "epoch": 0.02213989342146196,
      "grad_norm": 2570.178466796875,
      "learning_rate": 8.768000000000001e-06,
      "loss": 243.9221,
      "step": 5480
    },
    {
      "epoch": 0.022180294686829593,
      "grad_norm": 1173.486572265625,
      "learning_rate": 8.784000000000001e-06,
      "loss": 181.3593,
      "step": 5490
    },
    {
      "epoch": 0.022220695952197222,
      "grad_norm": 2153.557373046875,
      "learning_rate": 8.8e-06,
      "loss": 254.7195,
      "step": 5500
    },
    {
      "epoch": 0.022261097217564856,
      "grad_norm": 3733.31201171875,
      "learning_rate": 8.816000000000002e-06,
      "loss": 196.0029,
      "step": 5510
    },
    {
      "epoch": 0.022301498482932485,
      "grad_norm": 912.1559448242188,
      "learning_rate": 8.832000000000001e-06,
      "loss": 213.8326,
      "step": 5520
    },
    {
      "epoch": 0.022341899748300118,
      "grad_norm": 4412.5341796875,
      "learning_rate": 8.848e-06,
      "loss": 189.7024,
      "step": 5530
    },
    {
      "epoch": 0.022382301013667748,
      "grad_norm": 2686.481201171875,
      "learning_rate": 8.864e-06,
      "loss": 243.9844,
      "step": 5540
    },
    {
      "epoch": 0.02242270227903538,
      "grad_norm": 2003.0455322265625,
      "learning_rate": 8.880000000000001e-06,
      "loss": 245.3063,
      "step": 5550
    },
    {
      "epoch": 0.02246310354440301,
      "grad_norm": 873.5650634765625,
      "learning_rate": 8.896000000000001e-06,
      "loss": 196.9391,
      "step": 5560
    },
    {
      "epoch": 0.022503504809770643,
      "grad_norm": 936.9183349609375,
      "learning_rate": 8.912e-06,
      "loss": 120.0026,
      "step": 5570
    },
    {
      "epoch": 0.022543906075138273,
      "grad_norm": 3665.57470703125,
      "learning_rate": 8.928000000000002e-06,
      "loss": 215.1941,
      "step": 5580
    },
    {
      "epoch": 0.022584307340505906,
      "grad_norm": 502.3744201660156,
      "learning_rate": 8.944000000000001e-06,
      "loss": 245.2761,
      "step": 5590
    },
    {
      "epoch": 0.022624708605873536,
      "grad_norm": 801.8453979492188,
      "learning_rate": 8.96e-06,
      "loss": 184.7434,
      "step": 5600
    },
    {
      "epoch": 0.02266510987124117,
      "grad_norm": 6697.9013671875,
      "learning_rate": 8.976e-06,
      "loss": 217.284,
      "step": 5610
    },
    {
      "epoch": 0.0227055111366088,
      "grad_norm": 1576.22265625,
      "learning_rate": 8.992000000000001e-06,
      "loss": 196.8912,
      "step": 5620
    },
    {
      "epoch": 0.02274591240197643,
      "grad_norm": 1674.2708740234375,
      "learning_rate": 9.008e-06,
      "loss": 277.6959,
      "step": 5630
    },
    {
      "epoch": 0.02278631366734406,
      "grad_norm": 889.8017578125,
      "learning_rate": 9.024e-06,
      "loss": 252.6676,
      "step": 5640
    },
    {
      "epoch": 0.022826714932711694,
      "grad_norm": 2106.39208984375,
      "learning_rate": 9.040000000000002e-06,
      "loss": 230.2455,
      "step": 5650
    },
    {
      "epoch": 0.022867116198079324,
      "grad_norm": 1100.5885009765625,
      "learning_rate": 9.056000000000001e-06,
      "loss": 248.682,
      "step": 5660
    },
    {
      "epoch": 0.022907517463446957,
      "grad_norm": 1121.548828125,
      "learning_rate": 9.072e-06,
      "loss": 192.1396,
      "step": 5670
    },
    {
      "epoch": 0.022947918728814586,
      "grad_norm": 525.61181640625,
      "learning_rate": 9.088000000000002e-06,
      "loss": 176.5126,
      "step": 5680
    },
    {
      "epoch": 0.02298831999418222,
      "grad_norm": 0.0,
      "learning_rate": 9.104000000000001e-06,
      "loss": 241.5197,
      "step": 5690
    },
    {
      "epoch": 0.02302872125954985,
      "grad_norm": 979.9448852539062,
      "learning_rate": 9.12e-06,
      "loss": 203.9497,
      "step": 5700
    },
    {
      "epoch": 0.023069122524917482,
      "grad_norm": 1121.1590576171875,
      "learning_rate": 9.136e-06,
      "loss": 210.5354,
      "step": 5710
    },
    {
      "epoch": 0.02310952379028511,
      "grad_norm": 528.6953125,
      "learning_rate": 9.152000000000001e-06,
      "loss": 183.6194,
      "step": 5720
    },
    {
      "epoch": 0.023149925055652745,
      "grad_norm": 747.2396850585938,
      "learning_rate": 9.168000000000001e-06,
      "loss": 158.1538,
      "step": 5730
    },
    {
      "epoch": 0.023190326321020374,
      "grad_norm": 706.2687377929688,
      "learning_rate": 9.184e-06,
      "loss": 196.5268,
      "step": 5740
    },
    {
      "epoch": 0.023230727586388007,
      "grad_norm": 1605.8697509765625,
      "learning_rate": 9.200000000000002e-06,
      "loss": 224.3516,
      "step": 5750
    },
    {
      "epoch": 0.023271128851755637,
      "grad_norm": 2184.8427734375,
      "learning_rate": 9.216000000000001e-06,
      "loss": 309.77,
      "step": 5760
    },
    {
      "epoch": 0.02331153011712327,
      "grad_norm": 1893.63037109375,
      "learning_rate": 9.232e-06,
      "loss": 201.3919,
      "step": 5770
    },
    {
      "epoch": 0.0233519313824909,
      "grad_norm": 871.6980590820312,
      "learning_rate": 9.248e-06,
      "loss": 128.403,
      "step": 5780
    },
    {
      "epoch": 0.023392332647858533,
      "grad_norm": 747.3956298828125,
      "learning_rate": 9.264000000000001e-06,
      "loss": 267.743,
      "step": 5790
    },
    {
      "epoch": 0.023432733913226162,
      "grad_norm": 920.072265625,
      "learning_rate": 9.280000000000001e-06,
      "loss": 222.4892,
      "step": 5800
    },
    {
      "epoch": 0.023473135178593795,
      "grad_norm": 923.3662109375,
      "learning_rate": 9.296e-06,
      "loss": 175.1828,
      "step": 5810
    },
    {
      "epoch": 0.023513536443961425,
      "grad_norm": 1141.5255126953125,
      "learning_rate": 9.312000000000002e-06,
      "loss": 212.2976,
      "step": 5820
    },
    {
      "epoch": 0.023553937709329058,
      "grad_norm": 803.1054077148438,
      "learning_rate": 9.328000000000001e-06,
      "loss": 167.2026,
      "step": 5830
    },
    {
      "epoch": 0.023594338974696687,
      "grad_norm": 1214.9140625,
      "learning_rate": 9.344e-06,
      "loss": 285.092,
      "step": 5840
    },
    {
      "epoch": 0.02363474024006432,
      "grad_norm": 1920.0303955078125,
      "learning_rate": 9.360000000000002e-06,
      "loss": 366.2662,
      "step": 5850
    },
    {
      "epoch": 0.02367514150543195,
      "grad_norm": 1020.25537109375,
      "learning_rate": 9.376000000000001e-06,
      "loss": 258.3467,
      "step": 5860
    },
    {
      "epoch": 0.023715542770799583,
      "grad_norm": 1090.7298583984375,
      "learning_rate": 9.392000000000001e-06,
      "loss": 214.8844,
      "step": 5870
    },
    {
      "epoch": 0.023755944036167213,
      "grad_norm": 551.9674072265625,
      "learning_rate": 9.408e-06,
      "loss": 194.8219,
      "step": 5880
    },
    {
      "epoch": 0.023796345301534846,
      "grad_norm": 1726.1187744140625,
      "learning_rate": 9.424000000000002e-06,
      "loss": 202.5622,
      "step": 5890
    },
    {
      "epoch": 0.023836746566902475,
      "grad_norm": 2476.4296875,
      "learning_rate": 9.440000000000001e-06,
      "loss": 308.0325,
      "step": 5900
    },
    {
      "epoch": 0.02387714783227011,
      "grad_norm": 1048.78369140625,
      "learning_rate": 9.456e-06,
      "loss": 230.1398,
      "step": 5910
    },
    {
      "epoch": 0.023917549097637738,
      "grad_norm": 1003.924560546875,
      "learning_rate": 9.472000000000002e-06,
      "loss": 143.4615,
      "step": 5920
    },
    {
      "epoch": 0.02395795036300537,
      "grad_norm": 2788.75830078125,
      "learning_rate": 9.488000000000001e-06,
      "loss": 196.0169,
      "step": 5930
    },
    {
      "epoch": 0.023998351628373,
      "grad_norm": 1144.117431640625,
      "learning_rate": 9.504e-06,
      "loss": 173.9962,
      "step": 5940
    },
    {
      "epoch": 0.024038752893740634,
      "grad_norm": 1300.0675048828125,
      "learning_rate": 9.52e-06,
      "loss": 205.8077,
      "step": 5950
    },
    {
      "epoch": 0.024079154159108263,
      "grad_norm": 831.2354125976562,
      "learning_rate": 9.536000000000002e-06,
      "loss": 187.0024,
      "step": 5960
    },
    {
      "epoch": 0.024119555424475896,
      "grad_norm": 1625.884765625,
      "learning_rate": 9.552000000000001e-06,
      "loss": 211.5945,
      "step": 5970
    },
    {
      "epoch": 0.024159956689843526,
      "grad_norm": 1912.547607421875,
      "learning_rate": 9.568e-06,
      "loss": 227.0418,
      "step": 5980
    },
    {
      "epoch": 0.02420035795521116,
      "grad_norm": 1300.2178955078125,
      "learning_rate": 9.584000000000002e-06,
      "loss": 110.0356,
      "step": 5990
    },
    {
      "epoch": 0.02424075922057879,
      "grad_norm": 1286.5931396484375,
      "learning_rate": 9.600000000000001e-06,
      "loss": 207.4342,
      "step": 6000
    },
    {
      "epoch": 0.02428116048594642,
      "grad_norm": 889.315673828125,
      "learning_rate": 9.616e-06,
      "loss": 166.3021,
      "step": 6010
    },
    {
      "epoch": 0.02432156175131405,
      "grad_norm": 1677.3450927734375,
      "learning_rate": 9.632e-06,
      "loss": 176.0853,
      "step": 6020
    },
    {
      "epoch": 0.024361963016681684,
      "grad_norm": 2593.000244140625,
      "learning_rate": 9.648000000000001e-06,
      "loss": 210.7468,
      "step": 6030
    },
    {
      "epoch": 0.024402364282049314,
      "grad_norm": 2517.197998046875,
      "learning_rate": 9.664000000000001e-06,
      "loss": 157.0118,
      "step": 6040
    },
    {
      "epoch": 0.024442765547416947,
      "grad_norm": 867.0703735351562,
      "learning_rate": 9.68e-06,
      "loss": 214.9123,
      "step": 6050
    },
    {
      "epoch": 0.024483166812784576,
      "grad_norm": 684.4923095703125,
      "learning_rate": 9.696000000000002e-06,
      "loss": 222.2375,
      "step": 6060
    },
    {
      "epoch": 0.02452356807815221,
      "grad_norm": 1864.6201171875,
      "learning_rate": 9.712e-06,
      "loss": 218.7184,
      "step": 6070
    },
    {
      "epoch": 0.02456396934351984,
      "grad_norm": 4889.07763671875,
      "learning_rate": 9.728e-06,
      "loss": 225.6674,
      "step": 6080
    },
    {
      "epoch": 0.02460437060888747,
      "grad_norm": 1293.3353271484375,
      "learning_rate": 9.744000000000002e-06,
      "loss": 224.6604,
      "step": 6090
    },
    {
      "epoch": 0.0246447718742551,
      "grad_norm": 1205.1356201171875,
      "learning_rate": 9.760000000000001e-06,
      "loss": 262.9507,
      "step": 6100
    },
    {
      "epoch": 0.02468517313962273,
      "grad_norm": 1262.5826416015625,
      "learning_rate": 9.776000000000001e-06,
      "loss": 227.2488,
      "step": 6110
    },
    {
      "epoch": 0.024725574404990364,
      "grad_norm": 1142.44921875,
      "learning_rate": 9.792e-06,
      "loss": 346.6811,
      "step": 6120
    },
    {
      "epoch": 0.024765975670357994,
      "grad_norm": 1234.95166015625,
      "learning_rate": 9.808000000000002e-06,
      "loss": 210.1194,
      "step": 6130
    },
    {
      "epoch": 0.024806376935725627,
      "grad_norm": 2154.517822265625,
      "learning_rate": 9.824000000000001e-06,
      "loss": 212.564,
      "step": 6140
    },
    {
      "epoch": 0.024846778201093257,
      "grad_norm": 699.7747192382812,
      "learning_rate": 9.84e-06,
      "loss": 208.7139,
      "step": 6150
    },
    {
      "epoch": 0.02488717946646089,
      "grad_norm": 720.7115478515625,
      "learning_rate": 9.856000000000002e-06,
      "loss": 201.0257,
      "step": 6160
    },
    {
      "epoch": 0.02492758073182852,
      "grad_norm": 1159.8109130859375,
      "learning_rate": 9.872e-06,
      "loss": 197.7215,
      "step": 6170
    },
    {
      "epoch": 0.024967981997196152,
      "grad_norm": 675.26904296875,
      "learning_rate": 9.888000000000001e-06,
      "loss": 220.4793,
      "step": 6180
    },
    {
      "epoch": 0.025008383262563782,
      "grad_norm": 1005.2100830078125,
      "learning_rate": 9.904e-06,
      "loss": 205.4046,
      "step": 6190
    },
    {
      "epoch": 0.025048784527931415,
      "grad_norm": 727.637939453125,
      "learning_rate": 9.920000000000002e-06,
      "loss": 248.9305,
      "step": 6200
    },
    {
      "epoch": 0.025089185793299044,
      "grad_norm": 8848.505859375,
      "learning_rate": 9.936000000000001e-06,
      "loss": 261.5944,
      "step": 6210
    },
    {
      "epoch": 0.025129587058666678,
      "grad_norm": 915.894287109375,
      "learning_rate": 9.952e-06,
      "loss": 211.3287,
      "step": 6220
    },
    {
      "epoch": 0.025169988324034307,
      "grad_norm": 844.5236206054688,
      "learning_rate": 9.968000000000002e-06,
      "loss": 183.3922,
      "step": 6230
    },
    {
      "epoch": 0.02521038958940194,
      "grad_norm": 924.010009765625,
      "learning_rate": 9.984e-06,
      "loss": 170.826,
      "step": 6240
    },
    {
      "epoch": 0.02525079085476957,
      "grad_norm": 516.2547607421875,
      "learning_rate": 1e-05,
      "loss": 248.4559,
      "step": 6250
    },
    {
      "epoch": 0.025291192120137203,
      "grad_norm": 1103.63330078125,
      "learning_rate": 1.0016000000000002e-05,
      "loss": 223.1157,
      "step": 6260
    },
    {
      "epoch": 0.025331593385504832,
      "grad_norm": 4242.97314453125,
      "learning_rate": 1.0032000000000002e-05,
      "loss": 320.5385,
      "step": 6270
    },
    {
      "epoch": 0.025371994650872465,
      "grad_norm": 1703.0921630859375,
      "learning_rate": 1.0048e-05,
      "loss": 208.4248,
      "step": 6280
    },
    {
      "epoch": 0.025412395916240095,
      "grad_norm": 842.8690795898438,
      "learning_rate": 1.0064e-05,
      "loss": 147.7861,
      "step": 6290
    },
    {
      "epoch": 0.025452797181607728,
      "grad_norm": 1763.817626953125,
      "learning_rate": 1.008e-05,
      "loss": 180.2249,
      "step": 6300
    },
    {
      "epoch": 0.025493198446975358,
      "grad_norm": 631.6640014648438,
      "learning_rate": 1.0096000000000001e-05,
      "loss": 141.9704,
      "step": 6310
    },
    {
      "epoch": 0.02553359971234299,
      "grad_norm": 2309.67724609375,
      "learning_rate": 1.0112000000000002e-05,
      "loss": 219.5726,
      "step": 6320
    },
    {
      "epoch": 0.02557400097771062,
      "grad_norm": 3058.4248046875,
      "learning_rate": 1.0128e-05,
      "loss": 271.9829,
      "step": 6330
    },
    {
      "epoch": 0.025614402243078253,
      "grad_norm": 778.4915161132812,
      "learning_rate": 1.0144e-05,
      "loss": 241.9638,
      "step": 6340
    },
    {
      "epoch": 0.025654803508445883,
      "grad_norm": 682.7378540039062,
      "learning_rate": 1.0160000000000001e-05,
      "loss": 254.1633,
      "step": 6350
    },
    {
      "epoch": 0.025695204773813516,
      "grad_norm": 1311.454833984375,
      "learning_rate": 1.0176000000000002e-05,
      "loss": 221.548,
      "step": 6360
    },
    {
      "epoch": 0.025735606039181146,
      "grad_norm": 1173.302734375,
      "learning_rate": 1.0192000000000002e-05,
      "loss": 192.5287,
      "step": 6370
    },
    {
      "epoch": 0.02577600730454878,
      "grad_norm": 1387.337646484375,
      "learning_rate": 1.0208e-05,
      "loss": 262.8041,
      "step": 6380
    },
    {
      "epoch": 0.025816408569916408,
      "grad_norm": 4407.6435546875,
      "learning_rate": 1.0224e-05,
      "loss": 219.8955,
      "step": 6390
    },
    {
      "epoch": 0.02585680983528404,
      "grad_norm": 621.7880249023438,
      "learning_rate": 1.024e-05,
      "loss": 183.8228,
      "step": 6400
    },
    {
      "epoch": 0.02589721110065167,
      "grad_norm": 809.182373046875,
      "learning_rate": 1.0256000000000001e-05,
      "loss": 189.9909,
      "step": 6410
    },
    {
      "epoch": 0.025937612366019304,
      "grad_norm": 2138.258056640625,
      "learning_rate": 1.0272e-05,
      "loss": 188.2057,
      "step": 6420
    },
    {
      "epoch": 0.025978013631386934,
      "grad_norm": 847.4774169921875,
      "learning_rate": 1.0288e-05,
      "loss": 198.1693,
      "step": 6430
    },
    {
      "epoch": 0.026018414896754567,
      "grad_norm": 3683.830322265625,
      "learning_rate": 1.0304e-05,
      "loss": 252.0563,
      "step": 6440
    },
    {
      "epoch": 0.026058816162122196,
      "grad_norm": 1519.5914306640625,
      "learning_rate": 1.0320000000000001e-05,
      "loss": 251.5904,
      "step": 6450
    },
    {
      "epoch": 0.02609921742748983,
      "grad_norm": 803.3671875,
      "learning_rate": 1.0336000000000002e-05,
      "loss": 153.3266,
      "step": 6460
    },
    {
      "epoch": 0.02613961869285746,
      "grad_norm": 826.2311401367188,
      "learning_rate": 1.0352e-05,
      "loss": 229.4645,
      "step": 6470
    },
    {
      "epoch": 0.026180019958225092,
      "grad_norm": 1242.169677734375,
      "learning_rate": 1.0368e-05,
      "loss": 257.6692,
      "step": 6480
    },
    {
      "epoch": 0.02622042122359272,
      "grad_norm": 3527.55712890625,
      "learning_rate": 1.0384000000000001e-05,
      "loss": 203.5762,
      "step": 6490
    },
    {
      "epoch": 0.026260822488960354,
      "grad_norm": 936.0517578125,
      "learning_rate": 1.04e-05,
      "loss": 235.4074,
      "step": 6500
    },
    {
      "epoch": 0.026301223754327984,
      "grad_norm": 735.2395629882812,
      "learning_rate": 1.0416000000000002e-05,
      "loss": 206.3603,
      "step": 6510
    },
    {
      "epoch": 0.026341625019695617,
      "grad_norm": 955.069091796875,
      "learning_rate": 1.0432e-05,
      "loss": 211.4242,
      "step": 6520
    },
    {
      "epoch": 0.026382026285063247,
      "grad_norm": 2535.602783203125,
      "learning_rate": 1.0448e-05,
      "loss": 245.856,
      "step": 6530
    },
    {
      "epoch": 0.02642242755043088,
      "grad_norm": 1577.4102783203125,
      "learning_rate": 1.0464e-05,
      "loss": 240.6815,
      "step": 6540
    },
    {
      "epoch": 0.02646282881579851,
      "grad_norm": 3635.015380859375,
      "learning_rate": 1.0480000000000001e-05,
      "loss": 171.5102,
      "step": 6550
    },
    {
      "epoch": 0.026503230081166142,
      "grad_norm": 1024.40966796875,
      "learning_rate": 1.0496000000000003e-05,
      "loss": 196.577,
      "step": 6560
    },
    {
      "epoch": 0.026543631346533772,
      "grad_norm": 4052.225341796875,
      "learning_rate": 1.0512e-05,
      "loss": 164.7473,
      "step": 6570
    },
    {
      "epoch": 0.026584032611901405,
      "grad_norm": 861.9880981445312,
      "learning_rate": 1.0528e-05,
      "loss": 198.6812,
      "step": 6580
    },
    {
      "epoch": 0.026624433877269035,
      "grad_norm": 1004.9078979492188,
      "learning_rate": 1.0544000000000001e-05,
      "loss": 124.6632,
      "step": 6590
    },
    {
      "epoch": 0.026664835142636668,
      "grad_norm": 815.4194946289062,
      "learning_rate": 1.056e-05,
      "loss": 252.8283,
      "step": 6600
    },
    {
      "epoch": 0.026705236408004297,
      "grad_norm": 1178.261962890625,
      "learning_rate": 1.0576000000000002e-05,
      "loss": 239.9308,
      "step": 6610
    },
    {
      "epoch": 0.02674563767337193,
      "grad_norm": 855.1956787109375,
      "learning_rate": 1.0592e-05,
      "loss": 170.0014,
      "step": 6620
    },
    {
      "epoch": 0.02678603893873956,
      "grad_norm": 764.4215087890625,
      "learning_rate": 1.0608e-05,
      "loss": 311.9611,
      "step": 6630
    },
    {
      "epoch": 0.026826440204107193,
      "grad_norm": 1226.867919921875,
      "learning_rate": 1.0624e-05,
      "loss": 164.5296,
      "step": 6640
    },
    {
      "epoch": 0.026866841469474823,
      "grad_norm": 793.0070190429688,
      "learning_rate": 1.0640000000000001e-05,
      "loss": 157.1765,
      "step": 6650
    },
    {
      "epoch": 0.026907242734842456,
      "grad_norm": 2848.79052734375,
      "learning_rate": 1.0656000000000003e-05,
      "loss": 178.3442,
      "step": 6660
    },
    {
      "epoch": 0.026947644000210085,
      "grad_norm": 1390.5682373046875,
      "learning_rate": 1.0672e-05,
      "loss": 203.0393,
      "step": 6670
    },
    {
      "epoch": 0.026988045265577718,
      "grad_norm": 912.458984375,
      "learning_rate": 1.0688e-05,
      "loss": 163.6808,
      "step": 6680
    },
    {
      "epoch": 0.027028446530945348,
      "grad_norm": 1008.1384887695312,
      "learning_rate": 1.0704000000000001e-05,
      "loss": 207.8135,
      "step": 6690
    },
    {
      "epoch": 0.02706884779631298,
      "grad_norm": 6595.03466796875,
      "learning_rate": 1.072e-05,
      "loss": 269.4763,
      "step": 6700
    },
    {
      "epoch": 0.02710924906168061,
      "grad_norm": 836.4190673828125,
      "learning_rate": 1.0736000000000002e-05,
      "loss": 244.2324,
      "step": 6710
    },
    {
      "epoch": 0.027149650327048244,
      "grad_norm": 2060.159423828125,
      "learning_rate": 1.0752e-05,
      "loss": 192.5937,
      "step": 6720
    },
    {
      "epoch": 0.027190051592415873,
      "grad_norm": 3539.124755859375,
      "learning_rate": 1.0768000000000001e-05,
      "loss": 257.8765,
      "step": 6730
    },
    {
      "epoch": 0.027230452857783506,
      "grad_norm": 1152.4991455078125,
      "learning_rate": 1.0784e-05,
      "loss": 215.0623,
      "step": 6740
    },
    {
      "epoch": 0.027270854123151136,
      "grad_norm": 1865.2408447265625,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 298.6338,
      "step": 6750
    },
    {
      "epoch": 0.02731125538851877,
      "grad_norm": 3113.653564453125,
      "learning_rate": 1.0816e-05,
      "loss": 212.6075,
      "step": 6760
    },
    {
      "epoch": 0.0273516566538864,
      "grad_norm": 723.0384521484375,
      "learning_rate": 1.0832e-05,
      "loss": 176.7559,
      "step": 6770
    },
    {
      "epoch": 0.02739205791925403,
      "grad_norm": 1553.911376953125,
      "learning_rate": 1.0848e-05,
      "loss": 214.2935,
      "step": 6780
    },
    {
      "epoch": 0.02743245918462166,
      "grad_norm": 2617.25927734375,
      "learning_rate": 1.0864000000000001e-05,
      "loss": 208.5479,
      "step": 6790
    },
    {
      "epoch": 0.027472860449989294,
      "grad_norm": 2721.330078125,
      "learning_rate": 1.0880000000000001e-05,
      "loss": 247.9051,
      "step": 6800
    },
    {
      "epoch": 0.027513261715356924,
      "grad_norm": 1277.49365234375,
      "learning_rate": 1.0896e-05,
      "loss": 188.6469,
      "step": 6810
    },
    {
      "epoch": 0.027553662980724557,
      "grad_norm": 1097.37451171875,
      "learning_rate": 1.0912e-05,
      "loss": 165.6317,
      "step": 6820
    },
    {
      "epoch": 0.027594064246092186,
      "grad_norm": 2809.764404296875,
      "learning_rate": 1.0928000000000001e-05,
      "loss": 207.2922,
      "step": 6830
    },
    {
      "epoch": 0.02763446551145982,
      "grad_norm": 3215.823974609375,
      "learning_rate": 1.0944e-05,
      "loss": 145.3388,
      "step": 6840
    },
    {
      "epoch": 0.02767486677682745,
      "grad_norm": 896.5782470703125,
      "learning_rate": 1.0960000000000002e-05,
      "loss": 246.8312,
      "step": 6850
    },
    {
      "epoch": 0.027715268042195082,
      "grad_norm": 1490.93408203125,
      "learning_rate": 1.0976e-05,
      "loss": 167.422,
      "step": 6860
    },
    {
      "epoch": 0.02775566930756271,
      "grad_norm": 1855.150390625,
      "learning_rate": 1.0992e-05,
      "loss": 218.5458,
      "step": 6870
    },
    {
      "epoch": 0.027796070572930345,
      "grad_norm": 885.1539306640625,
      "learning_rate": 1.1008e-05,
      "loss": 247.277,
      "step": 6880
    },
    {
      "epoch": 0.027836471838297974,
      "grad_norm": 988.1691284179688,
      "learning_rate": 1.1024000000000002e-05,
      "loss": 275.7784,
      "step": 6890
    },
    {
      "epoch": 0.027876873103665607,
      "grad_norm": 1423.39404296875,
      "learning_rate": 1.1040000000000001e-05,
      "loss": 155.3902,
      "step": 6900
    },
    {
      "epoch": 0.027917274369033237,
      "grad_norm": 1528.7322998046875,
      "learning_rate": 1.1056e-05,
      "loss": 199.039,
      "step": 6910
    },
    {
      "epoch": 0.02795767563440087,
      "grad_norm": 744.6835327148438,
      "learning_rate": 1.1072e-05,
      "loss": 170.9844,
      "step": 6920
    },
    {
      "epoch": 0.0279980768997685,
      "grad_norm": 1452.251708984375,
      "learning_rate": 1.1088000000000001e-05,
      "loss": 159.8676,
      "step": 6930
    },
    {
      "epoch": 0.028038478165136133,
      "grad_norm": 1314.5618896484375,
      "learning_rate": 1.1104e-05,
      "loss": 181.9609,
      "step": 6940
    },
    {
      "epoch": 0.028078879430503762,
      "grad_norm": 906.5917358398438,
      "learning_rate": 1.1120000000000002e-05,
      "loss": 178.3718,
      "step": 6950
    },
    {
      "epoch": 0.028119280695871395,
      "grad_norm": 1324.80810546875,
      "learning_rate": 1.1136e-05,
      "loss": 143.8825,
      "step": 6960
    },
    {
      "epoch": 0.028159681961239025,
      "grad_norm": 0.0,
      "learning_rate": 1.1152000000000001e-05,
      "loss": 118.0358,
      "step": 6970
    },
    {
      "epoch": 0.028200083226606658,
      "grad_norm": 4537.2822265625,
      "learning_rate": 1.1168e-05,
      "loss": 205.2892,
      "step": 6980
    },
    {
      "epoch": 0.028240484491974287,
      "grad_norm": 1045.489501953125,
      "learning_rate": 1.1184000000000002e-05,
      "loss": 196.2545,
      "step": 6990
    },
    {
      "epoch": 0.02828088575734192,
      "grad_norm": 814.6027221679688,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 188.3368,
      "step": 7000
    },
    {
      "epoch": 0.02832128702270955,
      "grad_norm": 823.8950805664062,
      "learning_rate": 1.1216e-05,
      "loss": 188.974,
      "step": 7010
    },
    {
      "epoch": 0.028361688288077183,
      "grad_norm": 1544.5411376953125,
      "learning_rate": 1.1232e-05,
      "loss": 230.2053,
      "step": 7020
    },
    {
      "epoch": 0.028402089553444813,
      "grad_norm": 646.723388671875,
      "learning_rate": 1.1248000000000001e-05,
      "loss": 128.2589,
      "step": 7030
    },
    {
      "epoch": 0.028442490818812446,
      "grad_norm": 1701.8121337890625,
      "learning_rate": 1.1264000000000001e-05,
      "loss": 180.4897,
      "step": 7040
    },
    {
      "epoch": 0.028482892084180075,
      "grad_norm": 1359.3916015625,
      "learning_rate": 1.128e-05,
      "loss": 204.192,
      "step": 7050
    },
    {
      "epoch": 0.02852329334954771,
      "grad_norm": 1082.42236328125,
      "learning_rate": 1.1296e-05,
      "loss": 232.6115,
      "step": 7060
    },
    {
      "epoch": 0.028563694614915338,
      "grad_norm": 1729.7169189453125,
      "learning_rate": 1.1312000000000001e-05,
      "loss": 141.8797,
      "step": 7070
    },
    {
      "epoch": 0.02860409588028297,
      "grad_norm": 1071.9970703125,
      "learning_rate": 1.1328e-05,
      "loss": 191.3511,
      "step": 7080
    },
    {
      "epoch": 0.0286444971456506,
      "grad_norm": 1277.10595703125,
      "learning_rate": 1.1344000000000002e-05,
      "loss": 200.7731,
      "step": 7090
    },
    {
      "epoch": 0.028684898411018234,
      "grad_norm": 668.3596801757812,
      "learning_rate": 1.136e-05,
      "loss": 255.4492,
      "step": 7100
    },
    {
      "epoch": 0.028725299676385863,
      "grad_norm": 778.4135131835938,
      "learning_rate": 1.1376000000000001e-05,
      "loss": 232.344,
      "step": 7110
    },
    {
      "epoch": 0.028765700941753496,
      "grad_norm": 839.8685302734375,
      "learning_rate": 1.1392e-05,
      "loss": 212.6378,
      "step": 7120
    },
    {
      "epoch": 0.028806102207121126,
      "grad_norm": 1015.4122924804688,
      "learning_rate": 1.1408000000000002e-05,
      "loss": 171.4305,
      "step": 7130
    },
    {
      "epoch": 0.02884650347248876,
      "grad_norm": 1297.2659912109375,
      "learning_rate": 1.1424000000000001e-05,
      "loss": 210.3192,
      "step": 7140
    },
    {
      "epoch": 0.02888690473785639,
      "grad_norm": 1470.431396484375,
      "learning_rate": 1.144e-05,
      "loss": 179.2978,
      "step": 7150
    },
    {
      "epoch": 0.02892730600322402,
      "grad_norm": 1462.60302734375,
      "learning_rate": 1.1456e-05,
      "loss": 243.1692,
      "step": 7160
    },
    {
      "epoch": 0.02896770726859165,
      "grad_norm": 1914.7281494140625,
      "learning_rate": 1.1472000000000001e-05,
      "loss": 178.1436,
      "step": 7170
    },
    {
      "epoch": 0.029008108533959284,
      "grad_norm": 0.0,
      "learning_rate": 1.1488e-05,
      "loss": 220.6732,
      "step": 7180
    },
    {
      "epoch": 0.029048509799326914,
      "grad_norm": 1843.9134521484375,
      "learning_rate": 1.1504000000000002e-05,
      "loss": 240.3133,
      "step": 7190
    },
    {
      "epoch": 0.029088911064694547,
      "grad_norm": 7561.37255859375,
      "learning_rate": 1.152e-05,
      "loss": 199.7405,
      "step": 7200
    },
    {
      "epoch": 0.029129312330062176,
      "grad_norm": 1293.88330078125,
      "learning_rate": 1.1536000000000001e-05,
      "loss": 180.5538,
      "step": 7210
    },
    {
      "epoch": 0.02916971359542981,
      "grad_norm": 504.95745849609375,
      "learning_rate": 1.1552e-05,
      "loss": 202.5211,
      "step": 7220
    },
    {
      "epoch": 0.02921011486079744,
      "grad_norm": 2889.349365234375,
      "learning_rate": 1.1568000000000002e-05,
      "loss": 236.6316,
      "step": 7230
    },
    {
      "epoch": 0.029250516126165072,
      "grad_norm": 741.410888671875,
      "learning_rate": 1.1584000000000001e-05,
      "loss": 168.3202,
      "step": 7240
    },
    {
      "epoch": 0.029290917391532702,
      "grad_norm": 2157.1884765625,
      "learning_rate": 1.16e-05,
      "loss": 203.7105,
      "step": 7250
    },
    {
      "epoch": 0.029331318656900335,
      "grad_norm": 2404.79150390625,
      "learning_rate": 1.1616e-05,
      "loss": 205.5438,
      "step": 7260
    },
    {
      "epoch": 0.029371719922267964,
      "grad_norm": 0.0,
      "learning_rate": 1.1632000000000001e-05,
      "loss": 141.7996,
      "step": 7270
    },
    {
      "epoch": 0.029412121187635597,
      "grad_norm": 651.517578125,
      "learning_rate": 1.1648000000000001e-05,
      "loss": 213.4516,
      "step": 7280
    },
    {
      "epoch": 0.029452522453003227,
      "grad_norm": 1541.439208984375,
      "learning_rate": 1.1664000000000002e-05,
      "loss": 212.3698,
      "step": 7290
    },
    {
      "epoch": 0.02949292371837086,
      "grad_norm": 2294.57373046875,
      "learning_rate": 1.168e-05,
      "loss": 161.3065,
      "step": 7300
    },
    {
      "epoch": 0.02953332498373849,
      "grad_norm": 1858.2723388671875,
      "learning_rate": 1.1696000000000001e-05,
      "loss": 259.5773,
      "step": 7310
    },
    {
      "epoch": 0.029573726249106123,
      "grad_norm": 1009.5340576171875,
      "learning_rate": 1.1712e-05,
      "loss": 248.1431,
      "step": 7320
    },
    {
      "epoch": 0.029614127514473752,
      "grad_norm": 3950.963623046875,
      "learning_rate": 1.1728000000000002e-05,
      "loss": 214.9634,
      "step": 7330
    },
    {
      "epoch": 0.029654528779841385,
      "grad_norm": 2193.398681640625,
      "learning_rate": 1.1744000000000001e-05,
      "loss": 181.1218,
      "step": 7340
    },
    {
      "epoch": 0.029694930045209015,
      "grad_norm": 944.3338012695312,
      "learning_rate": 1.1760000000000001e-05,
      "loss": 136.3482,
      "step": 7350
    },
    {
      "epoch": 0.029735331310576648,
      "grad_norm": 1111.0726318359375,
      "learning_rate": 1.1776e-05,
      "loss": 205.5683,
      "step": 7360
    },
    {
      "epoch": 0.029775732575944278,
      "grad_norm": 1929.471435546875,
      "learning_rate": 1.1792000000000002e-05,
      "loss": 147.8906,
      "step": 7370
    },
    {
      "epoch": 0.02981613384131191,
      "grad_norm": 1500.8133544921875,
      "learning_rate": 1.1808000000000001e-05,
      "loss": 176.9189,
      "step": 7380
    },
    {
      "epoch": 0.02985653510667954,
      "grad_norm": 1397.207763671875,
      "learning_rate": 1.1824e-05,
      "loss": 206.1869,
      "step": 7390
    },
    {
      "epoch": 0.029896936372047173,
      "grad_norm": 1064.505126953125,
      "learning_rate": 1.184e-05,
      "loss": 285.7666,
      "step": 7400
    },
    {
      "epoch": 0.029937337637414803,
      "grad_norm": 1040.721923828125,
      "learning_rate": 1.1856000000000001e-05,
      "loss": 200.1409,
      "step": 7410
    },
    {
      "epoch": 0.029977738902782436,
      "grad_norm": 1411.9283447265625,
      "learning_rate": 1.1872000000000001e-05,
      "loss": 197.3061,
      "step": 7420
    },
    {
      "epoch": 0.030018140168150065,
      "grad_norm": 1552.750244140625,
      "learning_rate": 1.1888000000000002e-05,
      "loss": 131.5654,
      "step": 7430
    },
    {
      "epoch": 0.0300585414335177,
      "grad_norm": 4135.3154296875,
      "learning_rate": 1.1904e-05,
      "loss": 158.1736,
      "step": 7440
    },
    {
      "epoch": 0.030098942698885328,
      "grad_norm": 1020.1656494140625,
      "learning_rate": 1.1920000000000001e-05,
      "loss": 202.9766,
      "step": 7450
    },
    {
      "epoch": 0.03013934396425296,
      "grad_norm": 1060.5179443359375,
      "learning_rate": 1.1936e-05,
      "loss": 231.8231,
      "step": 7460
    },
    {
      "epoch": 0.03017974522962059,
      "grad_norm": 1135.9185791015625,
      "learning_rate": 1.1952000000000002e-05,
      "loss": 180.0088,
      "step": 7470
    },
    {
      "epoch": 0.030220146494988224,
      "grad_norm": 1562.1593017578125,
      "learning_rate": 1.1968000000000001e-05,
      "loss": 194.9986,
      "step": 7480
    },
    {
      "epoch": 0.030260547760355853,
      "grad_norm": 1475.9083251953125,
      "learning_rate": 1.1984e-05,
      "loss": 190.9672,
      "step": 7490
    },
    {
      "epoch": 0.030300949025723486,
      "grad_norm": 2843.638427734375,
      "learning_rate": 1.2e-05,
      "loss": 208.5317,
      "step": 7500
    },
    {
      "epoch": 0.030341350291091116,
      "grad_norm": 13340.033203125,
      "learning_rate": 1.2016000000000002e-05,
      "loss": 290.147,
      "step": 7510
    },
    {
      "epoch": 0.03038175155645875,
      "grad_norm": 1923.64990234375,
      "learning_rate": 1.2032000000000001e-05,
      "loss": 187.0977,
      "step": 7520
    },
    {
      "epoch": 0.03042215282182638,
      "grad_norm": 753.6450805664062,
      "learning_rate": 1.2048000000000002e-05,
      "loss": 173.7488,
      "step": 7530
    },
    {
      "epoch": 0.030462554087194012,
      "grad_norm": 1671.169189453125,
      "learning_rate": 1.2064e-05,
      "loss": 234.1393,
      "step": 7540
    },
    {
      "epoch": 0.03050295535256164,
      "grad_norm": 4448.47607421875,
      "learning_rate": 1.2080000000000001e-05,
      "loss": 148.8285,
      "step": 7550
    },
    {
      "epoch": 0.030543356617929274,
      "grad_norm": 1697.405029296875,
      "learning_rate": 1.2096e-05,
      "loss": 165.7743,
      "step": 7560
    },
    {
      "epoch": 0.030583757883296904,
      "grad_norm": 1653.86669921875,
      "learning_rate": 1.2112000000000002e-05,
      "loss": 128.8316,
      "step": 7570
    },
    {
      "epoch": 0.030624159148664537,
      "grad_norm": 2187.434814453125,
      "learning_rate": 1.2128000000000001e-05,
      "loss": 309.9175,
      "step": 7580
    },
    {
      "epoch": 0.030664560414032167,
      "grad_norm": 648.2998657226562,
      "learning_rate": 1.2144000000000001e-05,
      "loss": 260.0035,
      "step": 7590
    },
    {
      "epoch": 0.0307049616793998,
      "grad_norm": 1619.0577392578125,
      "learning_rate": 1.216e-05,
      "loss": 269.7444,
      "step": 7600
    },
    {
      "epoch": 0.03074536294476743,
      "grad_norm": 793.527587890625,
      "learning_rate": 1.2176000000000002e-05,
      "loss": 253.9378,
      "step": 7610
    },
    {
      "epoch": 0.030785764210135062,
      "grad_norm": 618.2582397460938,
      "learning_rate": 1.2192000000000001e-05,
      "loss": 177.5981,
      "step": 7620
    },
    {
      "epoch": 0.030826165475502692,
      "grad_norm": 1379.3106689453125,
      "learning_rate": 1.2208000000000002e-05,
      "loss": 209.9215,
      "step": 7630
    },
    {
      "epoch": 0.030866566740870325,
      "grad_norm": 640.5135498046875,
      "learning_rate": 1.2224e-05,
      "loss": 150.5801,
      "step": 7640
    },
    {
      "epoch": 0.030906968006237955,
      "grad_norm": 1602.00390625,
      "learning_rate": 1.2240000000000001e-05,
      "loss": 207.2341,
      "step": 7650
    },
    {
      "epoch": 0.030947369271605588,
      "grad_norm": 804.4661865234375,
      "learning_rate": 1.2256000000000001e-05,
      "loss": 132.286,
      "step": 7660
    },
    {
      "epoch": 0.030987770536973217,
      "grad_norm": 1168.7255859375,
      "learning_rate": 1.2272000000000002e-05,
      "loss": 159.3607,
      "step": 7670
    },
    {
      "epoch": 0.03102817180234085,
      "grad_norm": 5299.14990234375,
      "learning_rate": 1.2288e-05,
      "loss": 247.6179,
      "step": 7680
    },
    {
      "epoch": 0.03106857306770848,
      "grad_norm": 2897.222900390625,
      "learning_rate": 1.2304000000000001e-05,
      "loss": 210.1399,
      "step": 7690
    },
    {
      "epoch": 0.031108974333076113,
      "grad_norm": 1037.8817138671875,
      "learning_rate": 1.232e-05,
      "loss": 177.3573,
      "step": 7700
    },
    {
      "epoch": 0.031149375598443742,
      "grad_norm": 1748.6580810546875,
      "learning_rate": 1.2336000000000002e-05,
      "loss": 200.1308,
      "step": 7710
    },
    {
      "epoch": 0.031189776863811376,
      "grad_norm": 1623.1146240234375,
      "learning_rate": 1.2352000000000001e-05,
      "loss": 138.4404,
      "step": 7720
    },
    {
      "epoch": 0.031230178129179005,
      "grad_norm": 658.6551513671875,
      "learning_rate": 1.2368e-05,
      "loss": 187.0388,
      "step": 7730
    },
    {
      "epoch": 0.03127057939454664,
      "grad_norm": 851.2140502929688,
      "learning_rate": 1.2384e-05,
      "loss": 121.0485,
      "step": 7740
    },
    {
      "epoch": 0.03131098065991427,
      "grad_norm": 1161.504150390625,
      "learning_rate": 1.2400000000000002e-05,
      "loss": 187.6907,
      "step": 7750
    },
    {
      "epoch": 0.0313513819252819,
      "grad_norm": 2363.146728515625,
      "learning_rate": 1.2416000000000001e-05,
      "loss": 243.086,
      "step": 7760
    },
    {
      "epoch": 0.03139178319064953,
      "grad_norm": 1060.61376953125,
      "learning_rate": 1.2432000000000002e-05,
      "loss": 159.8972,
      "step": 7770
    },
    {
      "epoch": 0.03143218445601716,
      "grad_norm": 1024.3526611328125,
      "learning_rate": 1.2448e-05,
      "loss": 197.9344,
      "step": 7780
    },
    {
      "epoch": 0.031472585721384796,
      "grad_norm": 1298.6092529296875,
      "learning_rate": 1.2464000000000001e-05,
      "loss": 179.0248,
      "step": 7790
    },
    {
      "epoch": 0.03151298698675242,
      "grad_norm": 1501.115966796875,
      "learning_rate": 1.248e-05,
      "loss": 199.7626,
      "step": 7800
    },
    {
      "epoch": 0.031553388252120056,
      "grad_norm": 645.8001098632812,
      "learning_rate": 1.2496000000000002e-05,
      "loss": 140.9509,
      "step": 7810
    },
    {
      "epoch": 0.03159378951748769,
      "grad_norm": 1474.071533203125,
      "learning_rate": 1.2512000000000002e-05,
      "loss": 225.0743,
      "step": 7820
    },
    {
      "epoch": 0.03163419078285532,
      "grad_norm": 578.9594116210938,
      "learning_rate": 1.2528e-05,
      "loss": 149.4478,
      "step": 7830
    },
    {
      "epoch": 0.03167459204822295,
      "grad_norm": 1799.5963134765625,
      "learning_rate": 1.2544e-05,
      "loss": 190.2897,
      "step": 7840
    },
    {
      "epoch": 0.03171499331359058,
      "grad_norm": 1678.9425048828125,
      "learning_rate": 1.2560000000000002e-05,
      "loss": 201.671,
      "step": 7850
    },
    {
      "epoch": 0.031755394578958214,
      "grad_norm": 1604.530517578125,
      "learning_rate": 1.2576000000000001e-05,
      "loss": 160.7375,
      "step": 7860
    },
    {
      "epoch": 0.03179579584432585,
      "grad_norm": 1029.1220703125,
      "learning_rate": 1.2592000000000002e-05,
      "loss": 168.309,
      "step": 7870
    },
    {
      "epoch": 0.03183619710969347,
      "grad_norm": 5791.7275390625,
      "learning_rate": 1.2608e-05,
      "loss": 317.029,
      "step": 7880
    },
    {
      "epoch": 0.031876598375061106,
      "grad_norm": 1256.510986328125,
      "learning_rate": 1.2624000000000001e-05,
      "loss": 145.226,
      "step": 7890
    },
    {
      "epoch": 0.03191699964042874,
      "grad_norm": 1424.3720703125,
      "learning_rate": 1.2640000000000001e-05,
      "loss": 123.2829,
      "step": 7900
    },
    {
      "epoch": 0.03195740090579637,
      "grad_norm": 1358.22314453125,
      "learning_rate": 1.2656000000000002e-05,
      "loss": 211.4159,
      "step": 7910
    },
    {
      "epoch": 0.031997802171164,
      "grad_norm": 1109.47509765625,
      "learning_rate": 1.2672000000000002e-05,
      "loss": 211.996,
      "step": 7920
    },
    {
      "epoch": 0.03203820343653163,
      "grad_norm": 1332.3016357421875,
      "learning_rate": 1.2688e-05,
      "loss": 202.7615,
      "step": 7930
    },
    {
      "epoch": 0.032078604701899265,
      "grad_norm": 0.0,
      "learning_rate": 1.2704e-05,
      "loss": 149.2255,
      "step": 7940
    },
    {
      "epoch": 0.0321190059672669,
      "grad_norm": 913.5186157226562,
      "learning_rate": 1.2720000000000002e-05,
      "loss": 229.9926,
      "step": 7950
    },
    {
      "epoch": 0.032159407232634524,
      "grad_norm": 1214.4002685546875,
      "learning_rate": 1.2736000000000001e-05,
      "loss": 262.6403,
      "step": 7960
    },
    {
      "epoch": 0.03219980849800216,
      "grad_norm": 2469.748291015625,
      "learning_rate": 1.2752e-05,
      "loss": 255.2624,
      "step": 7970
    },
    {
      "epoch": 0.03224020976336979,
      "grad_norm": 1674.9351806640625,
      "learning_rate": 1.2768e-05,
      "loss": 203.4937,
      "step": 7980
    },
    {
      "epoch": 0.03228061102873742,
      "grad_norm": 1442.7392578125,
      "learning_rate": 1.2784000000000002e-05,
      "loss": 174.4651,
      "step": 7990
    },
    {
      "epoch": 0.03232101229410505,
      "grad_norm": 697.1456909179688,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 216.5391,
      "step": 8000
    },
    {
      "epoch": 0.03236141355947268,
      "grad_norm": 2099.24853515625,
      "learning_rate": 1.2816000000000002e-05,
      "loss": 210.8427,
      "step": 8010
    },
    {
      "epoch": 0.032401814824840315,
      "grad_norm": 3176.245361328125,
      "learning_rate": 1.2832e-05,
      "loss": 305.0184,
      "step": 8020
    },
    {
      "epoch": 0.03244221609020795,
      "grad_norm": 1037.900146484375,
      "learning_rate": 1.2848e-05,
      "loss": 224.8694,
      "step": 8030
    },
    {
      "epoch": 0.032482617355575574,
      "grad_norm": 474.0095520019531,
      "learning_rate": 1.2864e-05,
      "loss": 206.4488,
      "step": 8040
    },
    {
      "epoch": 0.03252301862094321,
      "grad_norm": 961.8276977539062,
      "learning_rate": 1.2880000000000002e-05,
      "loss": 174.9908,
      "step": 8050
    },
    {
      "epoch": 0.03256341988631084,
      "grad_norm": 1525.29833984375,
      "learning_rate": 1.2896000000000002e-05,
      "loss": 256.3985,
      "step": 8060
    },
    {
      "epoch": 0.03260382115167847,
      "grad_norm": 13022.603515625,
      "learning_rate": 1.2912e-05,
      "loss": 171.9165,
      "step": 8070
    },
    {
      "epoch": 0.0326442224170461,
      "grad_norm": 1341.8682861328125,
      "learning_rate": 1.2928e-05,
      "loss": 196.2449,
      "step": 8080
    },
    {
      "epoch": 0.03268462368241373,
      "grad_norm": 637.4829711914062,
      "learning_rate": 1.2944000000000002e-05,
      "loss": 154.6469,
      "step": 8090
    },
    {
      "epoch": 0.032725024947781366,
      "grad_norm": 1549.2552490234375,
      "learning_rate": 1.2960000000000001e-05,
      "loss": 132.3086,
      "step": 8100
    },
    {
      "epoch": 0.032765426213149,
      "grad_norm": 1383.0369873046875,
      "learning_rate": 1.2976000000000002e-05,
      "loss": 216.4039,
      "step": 8110
    },
    {
      "epoch": 0.032805827478516625,
      "grad_norm": 1574.9188232421875,
      "learning_rate": 1.2992e-05,
      "loss": 191.3513,
      "step": 8120
    },
    {
      "epoch": 0.03284622874388426,
      "grad_norm": 1016.8265380859375,
      "learning_rate": 1.3008e-05,
      "loss": 191.888,
      "step": 8130
    },
    {
      "epoch": 0.03288663000925189,
      "grad_norm": 579.7612915039062,
      "learning_rate": 1.3024000000000001e-05,
      "loss": 148.2349,
      "step": 8140
    },
    {
      "epoch": 0.032927031274619524,
      "grad_norm": 688.9622192382812,
      "learning_rate": 1.3040000000000002e-05,
      "loss": 131.6262,
      "step": 8150
    },
    {
      "epoch": 0.03296743253998715,
      "grad_norm": 1008.6786499023438,
      "learning_rate": 1.3056000000000002e-05,
      "loss": 191.4088,
      "step": 8160
    },
    {
      "epoch": 0.03300783380535478,
      "grad_norm": 731.140380859375,
      "learning_rate": 1.3072e-05,
      "loss": 144.9651,
      "step": 8170
    },
    {
      "epoch": 0.033048235070722416,
      "grad_norm": 1061.424072265625,
      "learning_rate": 1.3088e-05,
      "loss": 198.5589,
      "step": 8180
    },
    {
      "epoch": 0.03308863633609005,
      "grad_norm": 1009.1111450195312,
      "learning_rate": 1.3104000000000002e-05,
      "loss": 173.0443,
      "step": 8190
    },
    {
      "epoch": 0.033129037601457675,
      "grad_norm": 1960.4769287109375,
      "learning_rate": 1.3120000000000001e-05,
      "loss": 243.3992,
      "step": 8200
    },
    {
      "epoch": 0.03316943886682531,
      "grad_norm": 1036.068359375,
      "learning_rate": 1.3136000000000003e-05,
      "loss": 179.2266,
      "step": 8210
    },
    {
      "epoch": 0.03320984013219294,
      "grad_norm": 477.9182434082031,
      "learning_rate": 1.3152e-05,
      "loss": 132.2955,
      "step": 8220
    },
    {
      "epoch": 0.033250241397560575,
      "grad_norm": 997.7559814453125,
      "learning_rate": 1.3168e-05,
      "loss": 176.5007,
      "step": 8230
    },
    {
      "epoch": 0.0332906426629282,
      "grad_norm": 980.5240478515625,
      "learning_rate": 1.3184000000000001e-05,
      "loss": 114.6417,
      "step": 8240
    },
    {
      "epoch": 0.033331043928295834,
      "grad_norm": 732.45263671875,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 196.549,
      "step": 8250
    },
    {
      "epoch": 0.03337144519366347,
      "grad_norm": 785.9083862304688,
      "learning_rate": 1.3216000000000002e-05,
      "loss": 178.3419,
      "step": 8260
    },
    {
      "epoch": 0.0334118464590311,
      "grad_norm": 1148.8909912109375,
      "learning_rate": 1.3232e-05,
      "loss": 219.1049,
      "step": 8270
    },
    {
      "epoch": 0.033452247724398726,
      "grad_norm": 1013.8929443359375,
      "learning_rate": 1.3248000000000001e-05,
      "loss": 245.6775,
      "step": 8280
    },
    {
      "epoch": 0.03349264898976636,
      "grad_norm": 942.855224609375,
      "learning_rate": 1.3264000000000002e-05,
      "loss": 217.4367,
      "step": 8290
    },
    {
      "epoch": 0.03353305025513399,
      "grad_norm": 524.8524780273438,
      "learning_rate": 1.3280000000000002e-05,
      "loss": 174.6055,
      "step": 8300
    },
    {
      "epoch": 0.033573451520501625,
      "grad_norm": 876.1190185546875,
      "learning_rate": 1.3296e-05,
      "loss": 164.101,
      "step": 8310
    },
    {
      "epoch": 0.03361385278586925,
      "grad_norm": 6020.8359375,
      "learning_rate": 1.3312e-05,
      "loss": 213.8552,
      "step": 8320
    },
    {
      "epoch": 0.033654254051236884,
      "grad_norm": 731.1875,
      "learning_rate": 1.3328e-05,
      "loss": 177.7584,
      "step": 8330
    },
    {
      "epoch": 0.03369465531660452,
      "grad_norm": 1150.075439453125,
      "learning_rate": 1.3344000000000001e-05,
      "loss": 167.518,
      "step": 8340
    },
    {
      "epoch": 0.03373505658197215,
      "grad_norm": 2308.036865234375,
      "learning_rate": 1.3360000000000003e-05,
      "loss": 216.0022,
      "step": 8350
    },
    {
      "epoch": 0.033775457847339777,
      "grad_norm": 784.7696533203125,
      "learning_rate": 1.3376e-05,
      "loss": 186.3424,
      "step": 8360
    },
    {
      "epoch": 0.03381585911270741,
      "grad_norm": 1114.6754150390625,
      "learning_rate": 1.3392e-05,
      "loss": 161.8721,
      "step": 8370
    },
    {
      "epoch": 0.03385626037807504,
      "grad_norm": 1241.662841796875,
      "learning_rate": 1.3408000000000001e-05,
      "loss": 147.7201,
      "step": 8380
    },
    {
      "epoch": 0.033896661643442676,
      "grad_norm": 3314.600341796875,
      "learning_rate": 1.3424000000000002e-05,
      "loss": 234.9966,
      "step": 8390
    },
    {
      "epoch": 0.0339370629088103,
      "grad_norm": 2304.822021484375,
      "learning_rate": 1.3440000000000002e-05,
      "loss": 154.348,
      "step": 8400
    },
    {
      "epoch": 0.033977464174177935,
      "grad_norm": 1355.9027099609375,
      "learning_rate": 1.3456e-05,
      "loss": 208.0225,
      "step": 8410
    },
    {
      "epoch": 0.03401786543954557,
      "grad_norm": 865.4819946289062,
      "learning_rate": 1.3472e-05,
      "loss": 162.2137,
      "step": 8420
    },
    {
      "epoch": 0.0340582667049132,
      "grad_norm": 649.958984375,
      "learning_rate": 1.3488e-05,
      "loss": 178.5345,
      "step": 8430
    },
    {
      "epoch": 0.03409866797028083,
      "grad_norm": 875.6942749023438,
      "learning_rate": 1.3504000000000001e-05,
      "loss": 167.0072,
      "step": 8440
    },
    {
      "epoch": 0.03413906923564846,
      "grad_norm": 967.6959228515625,
      "learning_rate": 1.3520000000000003e-05,
      "loss": 111.3185,
      "step": 8450
    },
    {
      "epoch": 0.03417947050101609,
      "grad_norm": 946.614501953125,
      "learning_rate": 1.3536e-05,
      "loss": 189.8445,
      "step": 8460
    },
    {
      "epoch": 0.034219871766383726,
      "grad_norm": 645.7300415039062,
      "learning_rate": 1.3552e-05,
      "loss": 181.4957,
      "step": 8470
    },
    {
      "epoch": 0.03426027303175135,
      "grad_norm": 1362.599365234375,
      "learning_rate": 1.3568000000000001e-05,
      "loss": 198.2257,
      "step": 8480
    },
    {
      "epoch": 0.034300674297118985,
      "grad_norm": 752.5654907226562,
      "learning_rate": 1.3584000000000002e-05,
      "loss": 231.436,
      "step": 8490
    },
    {
      "epoch": 0.03434107556248662,
      "grad_norm": 518.3107299804688,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 198.5842,
      "step": 8500
    },
    {
      "epoch": 0.03438147682785425,
      "grad_norm": 953.03076171875,
      "learning_rate": 1.3616e-05,
      "loss": 238.2004,
      "step": 8510
    },
    {
      "epoch": 0.03442187809322188,
      "grad_norm": 880.9703979492188,
      "learning_rate": 1.3632000000000001e-05,
      "loss": 184.2565,
      "step": 8520
    },
    {
      "epoch": 0.03446227935858951,
      "grad_norm": 1198.8533935546875,
      "learning_rate": 1.3648e-05,
      "loss": 188.4894,
      "step": 8530
    },
    {
      "epoch": 0.034502680623957144,
      "grad_norm": 1765.7196044921875,
      "learning_rate": 1.3664000000000002e-05,
      "loss": 149.972,
      "step": 8540
    },
    {
      "epoch": 0.03454308188932478,
      "grad_norm": 2390.854736328125,
      "learning_rate": 1.3680000000000003e-05,
      "loss": 244.8083,
      "step": 8550
    },
    {
      "epoch": 0.0345834831546924,
      "grad_norm": 1242.086181640625,
      "learning_rate": 1.3696e-05,
      "loss": 113.8188,
      "step": 8560
    },
    {
      "epoch": 0.034623884420060036,
      "grad_norm": 838.2288818359375,
      "learning_rate": 1.3712e-05,
      "loss": 210.1455,
      "step": 8570
    },
    {
      "epoch": 0.03466428568542767,
      "grad_norm": 1526.849365234375,
      "learning_rate": 1.3728000000000001e-05,
      "loss": 176.4923,
      "step": 8580
    },
    {
      "epoch": 0.0347046869507953,
      "grad_norm": 3328.768310546875,
      "learning_rate": 1.3744000000000003e-05,
      "loss": 143.1111,
      "step": 8590
    },
    {
      "epoch": 0.03474508821616293,
      "grad_norm": 949.486328125,
      "learning_rate": 1.376e-05,
      "loss": 343.4464,
      "step": 8600
    },
    {
      "epoch": 0.03478548948153056,
      "grad_norm": 934.4647827148438,
      "learning_rate": 1.3776e-05,
      "loss": 200.4245,
      "step": 8610
    },
    {
      "epoch": 0.034825890746898194,
      "grad_norm": 1757.977294921875,
      "learning_rate": 1.3792000000000001e-05,
      "loss": 137.9326,
      "step": 8620
    },
    {
      "epoch": 0.03486629201226583,
      "grad_norm": 2243.847900390625,
      "learning_rate": 1.3808e-05,
      "loss": 198.7237,
      "step": 8630
    },
    {
      "epoch": 0.034906693277633453,
      "grad_norm": 2031.08984375,
      "learning_rate": 1.3824000000000002e-05,
      "loss": 201.4117,
      "step": 8640
    },
    {
      "epoch": 0.03494709454300109,
      "grad_norm": 2345.795654296875,
      "learning_rate": 1.384e-05,
      "loss": 166.4765,
      "step": 8650
    },
    {
      "epoch": 0.03498749580836872,
      "grad_norm": 861.6460571289062,
      "learning_rate": 1.3856e-05,
      "loss": 175.7778,
      "step": 8660
    },
    {
      "epoch": 0.03502789707373635,
      "grad_norm": 1131.3941650390625,
      "learning_rate": 1.3872e-05,
      "loss": 159.2677,
      "step": 8670
    },
    {
      "epoch": 0.03506829833910398,
      "grad_norm": 1215.5858154296875,
      "learning_rate": 1.3888000000000002e-05,
      "loss": 93.1331,
      "step": 8680
    },
    {
      "epoch": 0.03510869960447161,
      "grad_norm": 1228.4119873046875,
      "learning_rate": 1.3904000000000003e-05,
      "loss": 124.4559,
      "step": 8690
    },
    {
      "epoch": 0.035149100869839245,
      "grad_norm": 953.3961181640625,
      "learning_rate": 1.392e-05,
      "loss": 259.4703,
      "step": 8700
    },
    {
      "epoch": 0.03518950213520688,
      "grad_norm": 1472.0211181640625,
      "learning_rate": 1.3936e-05,
      "loss": 195.8783,
      "step": 8710
    },
    {
      "epoch": 0.035229903400574504,
      "grad_norm": 1071.033935546875,
      "learning_rate": 1.3952000000000001e-05,
      "loss": 207.037,
      "step": 8720
    },
    {
      "epoch": 0.03527030466594214,
      "grad_norm": 4129.77880859375,
      "learning_rate": 1.3968e-05,
      "loss": 202.4777,
      "step": 8730
    },
    {
      "epoch": 0.03531070593130977,
      "grad_norm": 3947.674560546875,
      "learning_rate": 1.3984000000000002e-05,
      "loss": 249.0112,
      "step": 8740
    },
    {
      "epoch": 0.0353511071966774,
      "grad_norm": 1019.8095092773438,
      "learning_rate": 1.4e-05,
      "loss": 159.161,
      "step": 8750
    },
    {
      "epoch": 0.03539150846204503,
      "grad_norm": 2045.958740234375,
      "learning_rate": 1.4016000000000001e-05,
      "loss": 166.8638,
      "step": 8760
    },
    {
      "epoch": 0.03543190972741266,
      "grad_norm": 569.2305908203125,
      "learning_rate": 1.4032e-05,
      "loss": 147.5823,
      "step": 8770
    },
    {
      "epoch": 0.035472310992780295,
      "grad_norm": 1683.003173828125,
      "learning_rate": 1.4048000000000002e-05,
      "loss": 178.0475,
      "step": 8780
    },
    {
      "epoch": 0.03551271225814793,
      "grad_norm": 1133.3314208984375,
      "learning_rate": 1.4064000000000003e-05,
      "loss": 179.5424,
      "step": 8790
    },
    {
      "epoch": 0.035553113523515555,
      "grad_norm": 996.33642578125,
      "learning_rate": 1.408e-05,
      "loss": 184.4815,
      "step": 8800
    },
    {
      "epoch": 0.03559351478888319,
      "grad_norm": 4099.81396484375,
      "learning_rate": 1.4096e-05,
      "loss": 217.5045,
      "step": 8810
    },
    {
      "epoch": 0.03563391605425082,
      "grad_norm": 2777.29052734375,
      "learning_rate": 1.4112000000000001e-05,
      "loss": 145.5609,
      "step": 8820
    },
    {
      "epoch": 0.035674317319618454,
      "grad_norm": 897.3009033203125,
      "learning_rate": 1.4128000000000001e-05,
      "loss": 220.9953,
      "step": 8830
    },
    {
      "epoch": 0.03571471858498608,
      "grad_norm": 3256.197509765625,
      "learning_rate": 1.4144000000000002e-05,
      "loss": 153.7076,
      "step": 8840
    },
    {
      "epoch": 0.03575511985035371,
      "grad_norm": 1044.2850341796875,
      "learning_rate": 1.416e-05,
      "loss": 160.3179,
      "step": 8850
    },
    {
      "epoch": 0.035795521115721346,
      "grad_norm": 598.4876098632812,
      "learning_rate": 1.4176000000000001e-05,
      "loss": 164.7267,
      "step": 8860
    },
    {
      "epoch": 0.03583592238108898,
      "grad_norm": 1391.9383544921875,
      "learning_rate": 1.4192e-05,
      "loss": 147.3058,
      "step": 8870
    },
    {
      "epoch": 0.035876323646456605,
      "grad_norm": 654.5458374023438,
      "learning_rate": 1.4208000000000002e-05,
      "loss": 191.4131,
      "step": 8880
    },
    {
      "epoch": 0.03591672491182424,
      "grad_norm": 1340.3818359375,
      "learning_rate": 1.4224000000000003e-05,
      "loss": 210.29,
      "step": 8890
    },
    {
      "epoch": 0.03595712617719187,
      "grad_norm": 1099.5445556640625,
      "learning_rate": 1.4240000000000001e-05,
      "loss": 168.8799,
      "step": 8900
    },
    {
      "epoch": 0.035997527442559504,
      "grad_norm": 1962.413330078125,
      "learning_rate": 1.4256e-05,
      "loss": 166.7184,
      "step": 8910
    },
    {
      "epoch": 0.03603792870792713,
      "grad_norm": 1163.4332275390625,
      "learning_rate": 1.4272000000000002e-05,
      "loss": 170.5609,
      "step": 8920
    },
    {
      "epoch": 0.036078329973294763,
      "grad_norm": 1148.7928466796875,
      "learning_rate": 1.4288000000000001e-05,
      "loss": 200.6612,
      "step": 8930
    },
    {
      "epoch": 0.0361187312386624,
      "grad_norm": 899.7924194335938,
      "learning_rate": 1.4304e-05,
      "loss": 120.8179,
      "step": 8940
    },
    {
      "epoch": 0.03615913250403003,
      "grad_norm": 1276.4134521484375,
      "learning_rate": 1.432e-05,
      "loss": 205.1565,
      "step": 8950
    },
    {
      "epoch": 0.036199533769397656,
      "grad_norm": 4312.658203125,
      "learning_rate": 1.4336000000000001e-05,
      "loss": 262.0594,
      "step": 8960
    },
    {
      "epoch": 0.03623993503476529,
      "grad_norm": 7287.62451171875,
      "learning_rate": 1.4352e-05,
      "loss": 169.9519,
      "step": 8970
    },
    {
      "epoch": 0.03628033630013292,
      "grad_norm": 791.3291625976562,
      "learning_rate": 1.4368000000000002e-05,
      "loss": 158.2487,
      "step": 8980
    },
    {
      "epoch": 0.036320737565500555,
      "grad_norm": 2023.2037353515625,
      "learning_rate": 1.4384e-05,
      "loss": 183.3372,
      "step": 8990
    },
    {
      "epoch": 0.03636113883086818,
      "grad_norm": 3376.461669921875,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 195.7389,
      "step": 9000
    },
    {
      "epoch": 0.036401540096235814,
      "grad_norm": 1903.6873779296875,
      "learning_rate": 1.4416e-05,
      "loss": 205.4889,
      "step": 9010
    },
    {
      "epoch": 0.03644194136160345,
      "grad_norm": 2198.595458984375,
      "learning_rate": 1.4432000000000002e-05,
      "loss": 116.7384,
      "step": 9020
    },
    {
      "epoch": 0.03648234262697108,
      "grad_norm": 1408.8551025390625,
      "learning_rate": 1.4448000000000001e-05,
      "loss": 92.1331,
      "step": 9030
    },
    {
      "epoch": 0.036522743892338706,
      "grad_norm": 1082.4267578125,
      "learning_rate": 1.4464e-05,
      "loss": 161.9581,
      "step": 9040
    },
    {
      "epoch": 0.03656314515770634,
      "grad_norm": 933.0946655273438,
      "learning_rate": 1.448e-05,
      "loss": 162.6531,
      "step": 9050
    },
    {
      "epoch": 0.03660354642307397,
      "grad_norm": 1909.8857421875,
      "learning_rate": 1.4496000000000001e-05,
      "loss": 312.7636,
      "step": 9060
    },
    {
      "epoch": 0.036643947688441605,
      "grad_norm": 1491.857666015625,
      "learning_rate": 1.4512000000000001e-05,
      "loss": 190.8777,
      "step": 9070
    },
    {
      "epoch": 0.03668434895380923,
      "grad_norm": 894.1411743164062,
      "learning_rate": 1.4528000000000002e-05,
      "loss": 142.2706,
      "step": 9080
    },
    {
      "epoch": 0.036724750219176865,
      "grad_norm": 611.9476318359375,
      "learning_rate": 1.4544e-05,
      "loss": 141.9291,
      "step": 9090
    },
    {
      "epoch": 0.0367651514845445,
      "grad_norm": 1880.233642578125,
      "learning_rate": 1.4560000000000001e-05,
      "loss": 234.3097,
      "step": 9100
    },
    {
      "epoch": 0.03680555274991213,
      "grad_norm": 1328.005615234375,
      "learning_rate": 1.4576e-05,
      "loss": 102.8834,
      "step": 9110
    },
    {
      "epoch": 0.03684595401527976,
      "grad_norm": 1829.908447265625,
      "learning_rate": 1.4592000000000002e-05,
      "loss": 132.0794,
      "step": 9120
    },
    {
      "epoch": 0.03688635528064739,
      "grad_norm": 708.6975708007812,
      "learning_rate": 1.4608000000000001e-05,
      "loss": 126.2266,
      "step": 9130
    },
    {
      "epoch": 0.03692675654601502,
      "grad_norm": 1220.538818359375,
      "learning_rate": 1.4624000000000001e-05,
      "loss": 160.287,
      "step": 9140
    },
    {
      "epoch": 0.03696715781138265,
      "grad_norm": 813.824462890625,
      "learning_rate": 1.464e-05,
      "loss": 175.8625,
      "step": 9150
    },
    {
      "epoch": 0.03700755907675028,
      "grad_norm": 1006.5848388671875,
      "learning_rate": 1.4656000000000002e-05,
      "loss": 234.0254,
      "step": 9160
    },
    {
      "epoch": 0.037047960342117915,
      "grad_norm": 2413.237548828125,
      "learning_rate": 1.4672000000000001e-05,
      "loss": 184.3562,
      "step": 9170
    },
    {
      "epoch": 0.03708836160748555,
      "grad_norm": 1156.4200439453125,
      "learning_rate": 1.4688000000000002e-05,
      "loss": 141.5659,
      "step": 9180
    },
    {
      "epoch": 0.037128762872853174,
      "grad_norm": 1558.5284423828125,
      "learning_rate": 1.4704e-05,
      "loss": 179.8484,
      "step": 9190
    },
    {
      "epoch": 0.03716916413822081,
      "grad_norm": 5858.7548828125,
      "learning_rate": 1.4720000000000001e-05,
      "loss": 173.0041,
      "step": 9200
    },
    {
      "epoch": 0.03720956540358844,
      "grad_norm": 5371.80078125,
      "learning_rate": 1.4736000000000001e-05,
      "loss": 214.0791,
      "step": 9210
    },
    {
      "epoch": 0.037249966668956074,
      "grad_norm": 712.1119995117188,
      "learning_rate": 1.4752000000000002e-05,
      "loss": 150.8345,
      "step": 9220
    },
    {
      "epoch": 0.0372903679343237,
      "grad_norm": 2370.50927734375,
      "learning_rate": 1.4768e-05,
      "loss": 173.9066,
      "step": 9230
    },
    {
      "epoch": 0.03733076919969133,
      "grad_norm": 1736.8262939453125,
      "learning_rate": 1.4784000000000001e-05,
      "loss": 194.8434,
      "step": 9240
    },
    {
      "epoch": 0.037371170465058966,
      "grad_norm": 1515.369140625,
      "learning_rate": 1.48e-05,
      "loss": 195.5914,
      "step": 9250
    },
    {
      "epoch": 0.0374115717304266,
      "grad_norm": 3758.433837890625,
      "learning_rate": 1.4816000000000002e-05,
      "loss": 138.8154,
      "step": 9260
    },
    {
      "epoch": 0.037451972995794225,
      "grad_norm": 452.4622802734375,
      "learning_rate": 1.4832000000000001e-05,
      "loss": 200.1029,
      "step": 9270
    },
    {
      "epoch": 0.03749237426116186,
      "grad_norm": 726.9000854492188,
      "learning_rate": 1.4848e-05,
      "loss": 179.7058,
      "step": 9280
    },
    {
      "epoch": 0.03753277552652949,
      "grad_norm": 1493.676513671875,
      "learning_rate": 1.4864e-05,
      "loss": 172.9092,
      "step": 9290
    },
    {
      "epoch": 0.037573176791897124,
      "grad_norm": 1207.8115234375,
      "learning_rate": 1.4880000000000002e-05,
      "loss": 178.3551,
      "step": 9300
    },
    {
      "epoch": 0.03761357805726475,
      "grad_norm": 896.19873046875,
      "learning_rate": 1.4896000000000001e-05,
      "loss": 156.0574,
      "step": 9310
    },
    {
      "epoch": 0.03765397932263238,
      "grad_norm": 968.2182006835938,
      "learning_rate": 1.4912000000000002e-05,
      "loss": 232.6957,
      "step": 9320
    },
    {
      "epoch": 0.037694380588000016,
      "grad_norm": 1797.7032470703125,
      "learning_rate": 1.4928e-05,
      "loss": 254.2054,
      "step": 9330
    },
    {
      "epoch": 0.03773478185336765,
      "grad_norm": 1541.966064453125,
      "learning_rate": 1.4944000000000001e-05,
      "loss": 164.4436,
      "step": 9340
    },
    {
      "epoch": 0.037775183118735275,
      "grad_norm": 829.641845703125,
      "learning_rate": 1.496e-05,
      "loss": 98.9679,
      "step": 9350
    },
    {
      "epoch": 0.03781558438410291,
      "grad_norm": 1358.08544921875,
      "learning_rate": 1.4976000000000002e-05,
      "loss": 229.9947,
      "step": 9360
    },
    {
      "epoch": 0.03785598564947054,
      "grad_norm": 1063.0125732421875,
      "learning_rate": 1.4992000000000001e-05,
      "loss": 138.6031,
      "step": 9370
    },
    {
      "epoch": 0.037896386914838175,
      "grad_norm": 2268.275390625,
      "learning_rate": 1.5008000000000001e-05,
      "loss": 206.4531,
      "step": 9380
    },
    {
      "epoch": 0.0379367881802058,
      "grad_norm": 2438.62841796875,
      "learning_rate": 1.5024e-05,
      "loss": 133.9822,
      "step": 9390
    },
    {
      "epoch": 0.037977189445573434,
      "grad_norm": 661.8216552734375,
      "learning_rate": 1.5040000000000002e-05,
      "loss": 162.2787,
      "step": 9400
    },
    {
      "epoch": 0.03801759071094107,
      "grad_norm": 1721.645751953125,
      "learning_rate": 1.5056000000000001e-05,
      "loss": 202.9203,
      "step": 9410
    },
    {
      "epoch": 0.0380579919763087,
      "grad_norm": 1209.0455322265625,
      "learning_rate": 1.5072000000000002e-05,
      "loss": 140.194,
      "step": 9420
    },
    {
      "epoch": 0.038098393241676326,
      "grad_norm": 850.0849609375,
      "learning_rate": 1.5088e-05,
      "loss": 126.9117,
      "step": 9430
    },
    {
      "epoch": 0.03813879450704396,
      "grad_norm": 2595.15283203125,
      "learning_rate": 1.5104000000000001e-05,
      "loss": 189.6362,
      "step": 9440
    },
    {
      "epoch": 0.03817919577241159,
      "grad_norm": 2151.7373046875,
      "learning_rate": 1.5120000000000001e-05,
      "loss": 177.4334,
      "step": 9450
    },
    {
      "epoch": 0.038219597037779225,
      "grad_norm": 853.5826416015625,
      "learning_rate": 1.5136000000000002e-05,
      "loss": 211.8327,
      "step": 9460
    },
    {
      "epoch": 0.03825999830314685,
      "grad_norm": 1072.6729736328125,
      "learning_rate": 1.5152000000000002e-05,
      "loss": 201.9211,
      "step": 9470
    },
    {
      "epoch": 0.038300399568514484,
      "grad_norm": 2053.216552734375,
      "learning_rate": 1.5168000000000001e-05,
      "loss": 196.2717,
      "step": 9480
    },
    {
      "epoch": 0.03834080083388212,
      "grad_norm": 1001.5113525390625,
      "learning_rate": 1.5184e-05,
      "loss": 182.2864,
      "step": 9490
    },
    {
      "epoch": 0.03838120209924975,
      "grad_norm": 1147.33203125,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 194.6276,
      "step": 9500
    },
    {
      "epoch": 0.03842160336461738,
      "grad_norm": 1260.394775390625,
      "learning_rate": 1.5216000000000001e-05,
      "loss": 91.9645,
      "step": 9510
    },
    {
      "epoch": 0.03846200462998501,
      "grad_norm": 1672.2742919921875,
      "learning_rate": 1.5232000000000003e-05,
      "loss": 177.3926,
      "step": 9520
    },
    {
      "epoch": 0.03850240589535264,
      "grad_norm": 7829.462890625,
      "learning_rate": 1.5248e-05,
      "loss": 176.5193,
      "step": 9530
    },
    {
      "epoch": 0.038542807160720276,
      "grad_norm": 732.9152221679688,
      "learning_rate": 1.5264e-05,
      "loss": 118.7326,
      "step": 9540
    },
    {
      "epoch": 0.0385832084260879,
      "grad_norm": 889.5083618164062,
      "learning_rate": 1.5280000000000003e-05,
      "loss": 138.2893,
      "step": 9550
    },
    {
      "epoch": 0.038623609691455535,
      "grad_norm": 1869.059326171875,
      "learning_rate": 1.5296e-05,
      "loss": 173.2165,
      "step": 9560
    },
    {
      "epoch": 0.03866401095682317,
      "grad_norm": 6519.04150390625,
      "learning_rate": 1.5312000000000002e-05,
      "loss": 177.7631,
      "step": 9570
    },
    {
      "epoch": 0.0387044122221908,
      "grad_norm": 1132.0162353515625,
      "learning_rate": 1.5328e-05,
      "loss": 153.1753,
      "step": 9580
    },
    {
      "epoch": 0.03874481348755843,
      "grad_norm": 961.602294921875,
      "learning_rate": 1.5344e-05,
      "loss": 131.0789,
      "step": 9590
    },
    {
      "epoch": 0.03878521475292606,
      "grad_norm": 976.28466796875,
      "learning_rate": 1.5360000000000002e-05,
      "loss": 173.8198,
      "step": 9600
    },
    {
      "epoch": 0.03882561601829369,
      "grad_norm": 714.9140014648438,
      "learning_rate": 1.5376000000000003e-05,
      "loss": 128.8499,
      "step": 9610
    },
    {
      "epoch": 0.038866017283661326,
      "grad_norm": 1191.9742431640625,
      "learning_rate": 1.5392e-05,
      "loss": 161.6513,
      "step": 9620
    },
    {
      "epoch": 0.03890641854902895,
      "grad_norm": 0.0,
      "learning_rate": 1.5408000000000002e-05,
      "loss": 183.5057,
      "step": 9630
    },
    {
      "epoch": 0.038946819814396585,
      "grad_norm": 1573.1395263671875,
      "learning_rate": 1.5424e-05,
      "loss": 152.1399,
      "step": 9640
    },
    {
      "epoch": 0.03898722107976422,
      "grad_norm": 999.6334838867188,
      "learning_rate": 1.544e-05,
      "loss": 162.3157,
      "step": 9650
    },
    {
      "epoch": 0.03902762234513185,
      "grad_norm": 0.0,
      "learning_rate": 1.5456000000000002e-05,
      "loss": 147.9127,
      "step": 9660
    },
    {
      "epoch": 0.03906802361049948,
      "grad_norm": 1467.85205078125,
      "learning_rate": 1.5472e-05,
      "loss": 146.4466,
      "step": 9670
    },
    {
      "epoch": 0.03910842487586711,
      "grad_norm": 1301.513427734375,
      "learning_rate": 1.5488e-05,
      "loss": 130.0043,
      "step": 9680
    },
    {
      "epoch": 0.039148826141234744,
      "grad_norm": 1770.3184814453125,
      "learning_rate": 1.5504000000000003e-05,
      "loss": 176.4808,
      "step": 9690
    },
    {
      "epoch": 0.03918922740660238,
      "grad_norm": 1669.854736328125,
      "learning_rate": 1.552e-05,
      "loss": 186.6205,
      "step": 9700
    },
    {
      "epoch": 0.03922962867197,
      "grad_norm": 638.5211791992188,
      "learning_rate": 1.5536e-05,
      "loss": 127.193,
      "step": 9710
    },
    {
      "epoch": 0.039270029937337636,
      "grad_norm": 3371.012939453125,
      "learning_rate": 1.5552e-05,
      "loss": 183.313,
      "step": 9720
    },
    {
      "epoch": 0.03931043120270527,
      "grad_norm": 1039.69677734375,
      "learning_rate": 1.5568e-05,
      "loss": 107.9096,
      "step": 9730
    },
    {
      "epoch": 0.0393508324680729,
      "grad_norm": 784.8140258789062,
      "learning_rate": 1.5584000000000002e-05,
      "loss": 161.9457,
      "step": 9740
    },
    {
      "epoch": 0.03939123373344053,
      "grad_norm": 5022.111328125,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 199.7558,
      "step": 9750
    },
    {
      "epoch": 0.03943163499880816,
      "grad_norm": 2041.7452392578125,
      "learning_rate": 1.5616e-05,
      "loss": 166.1727,
      "step": 9760
    },
    {
      "epoch": 0.039472036264175794,
      "grad_norm": 1846.129638671875,
      "learning_rate": 1.5632000000000002e-05,
      "loss": 174.9207,
      "step": 9770
    },
    {
      "epoch": 0.03951243752954343,
      "grad_norm": 1014.9839477539062,
      "learning_rate": 1.5648e-05,
      "loss": 120.4058,
      "step": 9780
    },
    {
      "epoch": 0.039552838794911054,
      "grad_norm": 2188.939697265625,
      "learning_rate": 1.5664e-05,
      "loss": 190.7619,
      "step": 9790
    },
    {
      "epoch": 0.03959324006027869,
      "grad_norm": 992.56689453125,
      "learning_rate": 1.5680000000000002e-05,
      "loss": 127.4463,
      "step": 9800
    },
    {
      "epoch": 0.03963364132564632,
      "grad_norm": 2232.902099609375,
      "learning_rate": 1.5696000000000004e-05,
      "loss": 212.9937,
      "step": 9810
    },
    {
      "epoch": 0.03967404259101395,
      "grad_norm": 569.5333862304688,
      "learning_rate": 1.5712e-05,
      "loss": 165.9236,
      "step": 9820
    },
    {
      "epoch": 0.03971444385638158,
      "grad_norm": 454.3460998535156,
      "learning_rate": 1.5728000000000003e-05,
      "loss": 135.0038,
      "step": 9830
    },
    {
      "epoch": 0.03975484512174921,
      "grad_norm": 2677.75,
      "learning_rate": 1.5744e-05,
      "loss": 202.8767,
      "step": 9840
    },
    {
      "epoch": 0.039795246387116845,
      "grad_norm": 2137.688720703125,
      "learning_rate": 1.576e-05,
      "loss": 147.8878,
      "step": 9850
    },
    {
      "epoch": 0.03983564765248448,
      "grad_norm": 1873.88330078125,
      "learning_rate": 1.5776e-05,
      "loss": 201.978,
      "step": 9860
    },
    {
      "epoch": 0.039876048917852104,
      "grad_norm": 1263.277099609375,
      "learning_rate": 1.5792e-05,
      "loss": 179.6682,
      "step": 9870
    },
    {
      "epoch": 0.03991645018321974,
      "grad_norm": 569.78466796875,
      "learning_rate": 1.5808000000000002e-05,
      "loss": 131.1906,
      "step": 9880
    },
    {
      "epoch": 0.03995685144858737,
      "grad_norm": 458.5001220703125,
      "learning_rate": 1.5824000000000003e-05,
      "loss": 102.6573,
      "step": 9890
    },
    {
      "epoch": 0.039997252713955,
      "grad_norm": 727.7654418945312,
      "learning_rate": 1.584e-05,
      "loss": 172.3399,
      "step": 9900
    },
    {
      "epoch": 0.04003765397932263,
      "grad_norm": 540.7371215820312,
      "learning_rate": 1.5856e-05,
      "loss": 140.1403,
      "step": 9910
    },
    {
      "epoch": 0.04007805524469026,
      "grad_norm": 1218.4239501953125,
      "learning_rate": 1.5872e-05,
      "loss": 198.2595,
      "step": 9920
    },
    {
      "epoch": 0.040118456510057895,
      "grad_norm": 2273.38427734375,
      "learning_rate": 1.5888e-05,
      "loss": 224.8911,
      "step": 9930
    },
    {
      "epoch": 0.04015885777542553,
      "grad_norm": 803.8103637695312,
      "learning_rate": 1.5904000000000002e-05,
      "loss": 84.3416,
      "step": 9940
    },
    {
      "epoch": 0.040199259040793155,
      "grad_norm": 1237.755615234375,
      "learning_rate": 1.5920000000000003e-05,
      "loss": 125.1505,
      "step": 9950
    },
    {
      "epoch": 0.04023966030616079,
      "grad_norm": 957.0908813476562,
      "learning_rate": 1.5936e-05,
      "loss": 170.3722,
      "step": 9960
    },
    {
      "epoch": 0.04028006157152842,
      "grad_norm": 1235.8572998046875,
      "learning_rate": 1.5952000000000002e-05,
      "loss": 188.307,
      "step": 9970
    },
    {
      "epoch": 0.040320462836896054,
      "grad_norm": 1665.431884765625,
      "learning_rate": 1.5968e-05,
      "loss": 173.6013,
      "step": 9980
    },
    {
      "epoch": 0.04036086410226368,
      "grad_norm": 944.884521484375,
      "learning_rate": 1.5984e-05,
      "loss": 172.7963,
      "step": 9990
    },
    {
      "epoch": 0.04040126536763131,
      "grad_norm": 817.8704223632812,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 121.9647,
      "step": 10000
    },
    {
      "epoch": 0.040441666632998946,
      "grad_norm": 531.5571899414062,
      "learning_rate": 1.6016e-05,
      "loss": 144.545,
      "step": 10010
    },
    {
      "epoch": 0.04048206789836658,
      "grad_norm": 1372.50048828125,
      "learning_rate": 1.6032e-05,
      "loss": 167.1301,
      "step": 10020
    },
    {
      "epoch": 0.040522469163734205,
      "grad_norm": 1101.1865234375,
      "learning_rate": 1.6048000000000003e-05,
      "loss": 196.8536,
      "step": 10030
    },
    {
      "epoch": 0.04056287042910184,
      "grad_norm": 1433.8011474609375,
      "learning_rate": 1.6064e-05,
      "loss": 157.4351,
      "step": 10040
    },
    {
      "epoch": 0.04060327169446947,
      "grad_norm": 2047.3099365234375,
      "learning_rate": 1.6080000000000002e-05,
      "loss": 127.7889,
      "step": 10050
    },
    {
      "epoch": 0.040643672959837104,
      "grad_norm": 0.0,
      "learning_rate": 1.6096e-05,
      "loss": 134.3584,
      "step": 10060
    },
    {
      "epoch": 0.04068407422520473,
      "grad_norm": 2167.48486328125,
      "learning_rate": 1.6112e-05,
      "loss": 161.7745,
      "step": 10070
    },
    {
      "epoch": 0.040724475490572364,
      "grad_norm": 2235.814697265625,
      "learning_rate": 1.6128000000000002e-05,
      "loss": 210.6461,
      "step": 10080
    },
    {
      "epoch": 0.04076487675594,
      "grad_norm": 766.3438110351562,
      "learning_rate": 1.6144000000000003e-05,
      "loss": 166.4439,
      "step": 10090
    },
    {
      "epoch": 0.04080527802130763,
      "grad_norm": 1047.2254638671875,
      "learning_rate": 1.616e-05,
      "loss": 162.5282,
      "step": 10100
    },
    {
      "epoch": 0.040845679286675256,
      "grad_norm": 1334.6319580078125,
      "learning_rate": 1.6176e-05,
      "loss": 172.6197,
      "step": 10110
    },
    {
      "epoch": 0.04088608055204289,
      "grad_norm": 796.6995849609375,
      "learning_rate": 1.6192e-05,
      "loss": 150.1012,
      "step": 10120
    },
    {
      "epoch": 0.04092648181741052,
      "grad_norm": 1244.7584228515625,
      "learning_rate": 1.6208e-05,
      "loss": 172.9664,
      "step": 10130
    },
    {
      "epoch": 0.040966883082778155,
      "grad_norm": 1606.3883056640625,
      "learning_rate": 1.6224000000000003e-05,
      "loss": 173.0676,
      "step": 10140
    },
    {
      "epoch": 0.04100728434814578,
      "grad_norm": 877.404541015625,
      "learning_rate": 1.6240000000000004e-05,
      "loss": 167.6871,
      "step": 10150
    },
    {
      "epoch": 0.041047685613513414,
      "grad_norm": 2475.5068359375,
      "learning_rate": 1.6256e-05,
      "loss": 172.3827,
      "step": 10160
    },
    {
      "epoch": 0.04108808687888105,
      "grad_norm": 1478.54638671875,
      "learning_rate": 1.6272000000000003e-05,
      "loss": 180.1601,
      "step": 10170
    },
    {
      "epoch": 0.04112848814424868,
      "grad_norm": 521.4425659179688,
      "learning_rate": 1.6288e-05,
      "loss": 167.5962,
      "step": 10180
    },
    {
      "epoch": 0.041168889409616306,
      "grad_norm": 2434.41064453125,
      "learning_rate": 1.6304000000000002e-05,
      "loss": 200.3995,
      "step": 10190
    },
    {
      "epoch": 0.04120929067498394,
      "grad_norm": 685.0830078125,
      "learning_rate": 1.632e-05,
      "loss": 149.9868,
      "step": 10200
    },
    {
      "epoch": 0.04124969194035157,
      "grad_norm": 997.065185546875,
      "learning_rate": 1.6336e-05,
      "loss": 152.3204,
      "step": 10210
    },
    {
      "epoch": 0.041290093205719205,
      "grad_norm": 1296.7747802734375,
      "learning_rate": 1.6352000000000002e-05,
      "loss": 187.0858,
      "step": 10220
    },
    {
      "epoch": 0.04133049447108683,
      "grad_norm": 757.8452758789062,
      "learning_rate": 1.6368000000000003e-05,
      "loss": 119.7687,
      "step": 10230
    },
    {
      "epoch": 0.041370895736454465,
      "grad_norm": 1303.6944580078125,
      "learning_rate": 1.6384e-05,
      "loss": 164.7086,
      "step": 10240
    },
    {
      "epoch": 0.0414112970018221,
      "grad_norm": 1585.5537109375,
      "learning_rate": 1.64e-05,
      "loss": 258.1144,
      "step": 10250
    },
    {
      "epoch": 0.04145169826718973,
      "grad_norm": 836.6436157226562,
      "learning_rate": 1.6416e-05,
      "loss": 155.0704,
      "step": 10260
    },
    {
      "epoch": 0.04149209953255736,
      "grad_norm": 918.627197265625,
      "learning_rate": 1.6432e-05,
      "loss": 148.9556,
      "step": 10270
    },
    {
      "epoch": 0.04153250079792499,
      "grad_norm": 1133.152099609375,
      "learning_rate": 1.6448000000000002e-05,
      "loss": 189.2886,
      "step": 10280
    },
    {
      "epoch": 0.04157290206329262,
      "grad_norm": 725.5909423828125,
      "learning_rate": 1.6464000000000004e-05,
      "loss": 102.5979,
      "step": 10290
    },
    {
      "epoch": 0.041613303328660256,
      "grad_norm": 2373.759765625,
      "learning_rate": 1.648e-05,
      "loss": 194.0122,
      "step": 10300
    },
    {
      "epoch": 0.04165370459402788,
      "grad_norm": 1110.9903564453125,
      "learning_rate": 1.6496e-05,
      "loss": 122.7091,
      "step": 10310
    },
    {
      "epoch": 0.041694105859395515,
      "grad_norm": 875.1665649414062,
      "learning_rate": 1.6512e-05,
      "loss": 159.7906,
      "step": 10320
    },
    {
      "epoch": 0.04173450712476315,
      "grad_norm": 601.5936279296875,
      "learning_rate": 1.6528e-05,
      "loss": 125.9924,
      "step": 10330
    },
    {
      "epoch": 0.04177490839013078,
      "grad_norm": 929.2774658203125,
      "learning_rate": 1.6544000000000003e-05,
      "loss": 128.1777,
      "step": 10340
    },
    {
      "epoch": 0.04181530965549841,
      "grad_norm": 1600.3740234375,
      "learning_rate": 1.656e-05,
      "loss": 142.6078,
      "step": 10350
    },
    {
      "epoch": 0.04185571092086604,
      "grad_norm": 1814.5093994140625,
      "learning_rate": 1.6576000000000002e-05,
      "loss": 188.1647,
      "step": 10360
    },
    {
      "epoch": 0.041896112186233674,
      "grad_norm": 3294.983642578125,
      "learning_rate": 1.6592000000000003e-05,
      "loss": 237.4234,
      "step": 10370
    },
    {
      "epoch": 0.04193651345160131,
      "grad_norm": 815.1757202148438,
      "learning_rate": 1.6608e-05,
      "loss": 210.1955,
      "step": 10380
    },
    {
      "epoch": 0.04197691471696893,
      "grad_norm": 396.14996337890625,
      "learning_rate": 1.6624000000000002e-05,
      "loss": 129.1369,
      "step": 10390
    },
    {
      "epoch": 0.042017315982336566,
      "grad_norm": 862.579833984375,
      "learning_rate": 1.664e-05,
      "loss": 179.3043,
      "step": 10400
    },
    {
      "epoch": 0.0420577172477042,
      "grad_norm": 889.371337890625,
      "learning_rate": 1.6656e-05,
      "loss": 152.2917,
      "step": 10410
    },
    {
      "epoch": 0.04209811851307183,
      "grad_norm": 606.1854858398438,
      "learning_rate": 1.6672000000000002e-05,
      "loss": 149.5471,
      "step": 10420
    },
    {
      "epoch": 0.04213851977843946,
      "grad_norm": 2273.623779296875,
      "learning_rate": 1.6688000000000004e-05,
      "loss": 167.0708,
      "step": 10430
    },
    {
      "epoch": 0.04217892104380709,
      "grad_norm": 929.865966796875,
      "learning_rate": 1.6704e-05,
      "loss": 190.4482,
      "step": 10440
    },
    {
      "epoch": 0.042219322309174724,
      "grad_norm": 1227.5264892578125,
      "learning_rate": 1.672e-05,
      "loss": 170.9611,
      "step": 10450
    },
    {
      "epoch": 0.04225972357454236,
      "grad_norm": 3041.83544921875,
      "learning_rate": 1.6736e-05,
      "loss": 145.8744,
      "step": 10460
    },
    {
      "epoch": 0.04230012483990998,
      "grad_norm": 682.5531616210938,
      "learning_rate": 1.6752e-05,
      "loss": 112.5866,
      "step": 10470
    },
    {
      "epoch": 0.042340526105277616,
      "grad_norm": 864.705322265625,
      "learning_rate": 1.6768000000000003e-05,
      "loss": 127.5083,
      "step": 10480
    },
    {
      "epoch": 0.04238092737064525,
      "grad_norm": 1205.29150390625,
      "learning_rate": 1.6784e-05,
      "loss": 157.477,
      "step": 10490
    },
    {
      "epoch": 0.04242132863601288,
      "grad_norm": 1558.863525390625,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 169.6953,
      "step": 10500
    },
    {
      "epoch": 0.04246172990138051,
      "grad_norm": 792.3493041992188,
      "learning_rate": 1.6816e-05,
      "loss": 150.5652,
      "step": 10510
    },
    {
      "epoch": 0.04250213116674814,
      "grad_norm": 1993.4556884765625,
      "learning_rate": 1.6832e-05,
      "loss": 166.2245,
      "step": 10520
    },
    {
      "epoch": 0.042542532432115775,
      "grad_norm": 662.0936889648438,
      "learning_rate": 1.6848000000000002e-05,
      "loss": 161.7556,
      "step": 10530
    },
    {
      "epoch": 0.04258293369748341,
      "grad_norm": 671.510009765625,
      "learning_rate": 1.6864e-05,
      "loss": 144.911,
      "step": 10540
    },
    {
      "epoch": 0.042623334962851034,
      "grad_norm": 552.993896484375,
      "learning_rate": 1.688e-05,
      "loss": 113.3128,
      "step": 10550
    },
    {
      "epoch": 0.04266373622821867,
      "grad_norm": 2240.13720703125,
      "learning_rate": 1.6896000000000002e-05,
      "loss": 146.852,
      "step": 10560
    },
    {
      "epoch": 0.0427041374935863,
      "grad_norm": 1047.0406494140625,
      "learning_rate": 1.6912000000000003e-05,
      "loss": 131.8557,
      "step": 10570
    },
    {
      "epoch": 0.04274453875895393,
      "grad_norm": 1396.8720703125,
      "learning_rate": 1.6928e-05,
      "loss": 163.8682,
      "step": 10580
    },
    {
      "epoch": 0.04278494002432156,
      "grad_norm": 2435.2705078125,
      "learning_rate": 1.6944e-05,
      "loss": 170.1234,
      "step": 10590
    },
    {
      "epoch": 0.04282534128968919,
      "grad_norm": 1863.5042724609375,
      "learning_rate": 1.696e-05,
      "loss": 146.7008,
      "step": 10600
    },
    {
      "epoch": 0.042865742555056825,
      "grad_norm": 1173.83447265625,
      "learning_rate": 1.6976e-05,
      "loss": 197.7816,
      "step": 10610
    },
    {
      "epoch": 0.04290614382042446,
      "grad_norm": 2275.920166015625,
      "learning_rate": 1.6992000000000003e-05,
      "loss": 206.3002,
      "step": 10620
    },
    {
      "epoch": 0.042946545085792084,
      "grad_norm": 1100.142578125,
      "learning_rate": 1.7008000000000004e-05,
      "loss": 140.3349,
      "step": 10630
    },
    {
      "epoch": 0.04298694635115972,
      "grad_norm": 1740.51953125,
      "learning_rate": 1.7024e-05,
      "loss": 174.8,
      "step": 10640
    },
    {
      "epoch": 0.04302734761652735,
      "grad_norm": 1941.865234375,
      "learning_rate": 1.704e-05,
      "loss": 176.8449,
      "step": 10650
    },
    {
      "epoch": 0.043067748881894984,
      "grad_norm": 1190.6832275390625,
      "learning_rate": 1.7056e-05,
      "loss": 186.9511,
      "step": 10660
    },
    {
      "epoch": 0.04310815014726261,
      "grad_norm": 696.5484008789062,
      "learning_rate": 1.7072000000000002e-05,
      "loss": 144.1866,
      "step": 10670
    },
    {
      "epoch": 0.04314855141263024,
      "grad_norm": 861.3687744140625,
      "learning_rate": 1.7088000000000003e-05,
      "loss": 150.022,
      "step": 10680
    },
    {
      "epoch": 0.043188952677997876,
      "grad_norm": 659.2498779296875,
      "learning_rate": 1.7104e-05,
      "loss": 132.2025,
      "step": 10690
    },
    {
      "epoch": 0.04322935394336551,
      "grad_norm": 1642.438232421875,
      "learning_rate": 1.7120000000000002e-05,
      "loss": 218.317,
      "step": 10700
    },
    {
      "epoch": 0.043269755208733135,
      "grad_norm": 620.3860473632812,
      "learning_rate": 1.7136e-05,
      "loss": 173.4349,
      "step": 10710
    },
    {
      "epoch": 0.04331015647410077,
      "grad_norm": 1452.962890625,
      "learning_rate": 1.7152e-05,
      "loss": 203.3664,
      "step": 10720
    },
    {
      "epoch": 0.0433505577394684,
      "grad_norm": 2354.88916015625,
      "learning_rate": 1.7168000000000002e-05,
      "loss": 187.4407,
      "step": 10730
    },
    {
      "epoch": 0.043390959004836034,
      "grad_norm": 1276.4913330078125,
      "learning_rate": 1.7184e-05,
      "loss": 145.4733,
      "step": 10740
    },
    {
      "epoch": 0.04343136027020366,
      "grad_norm": 695.041259765625,
      "learning_rate": 1.72e-05,
      "loss": 189.1464,
      "step": 10750
    },
    {
      "epoch": 0.04347176153557129,
      "grad_norm": 14365.31640625,
      "learning_rate": 1.7216000000000003e-05,
      "loss": 225.0812,
      "step": 10760
    },
    {
      "epoch": 0.043512162800938926,
      "grad_norm": 1351.369140625,
      "learning_rate": 1.7232000000000004e-05,
      "loss": 179.7414,
      "step": 10770
    },
    {
      "epoch": 0.04355256406630656,
      "grad_norm": 1129.187255859375,
      "learning_rate": 1.7248e-05,
      "loss": 153.1924,
      "step": 10780
    },
    {
      "epoch": 0.043592965331674186,
      "grad_norm": 1236.0352783203125,
      "learning_rate": 1.7264e-05,
      "loss": 167.2036,
      "step": 10790
    },
    {
      "epoch": 0.04363336659704182,
      "grad_norm": 1314.908447265625,
      "learning_rate": 1.728e-05,
      "loss": 155.7452,
      "step": 10800
    },
    {
      "epoch": 0.04367376786240945,
      "grad_norm": 1064.8416748046875,
      "learning_rate": 1.7296000000000002e-05,
      "loss": 249.5876,
      "step": 10810
    },
    {
      "epoch": 0.043714169127777085,
      "grad_norm": 1580.3046875,
      "learning_rate": 1.7312000000000003e-05,
      "loss": 183.0618,
      "step": 10820
    },
    {
      "epoch": 0.04375457039314471,
      "grad_norm": 799.4830322265625,
      "learning_rate": 1.7328e-05,
      "loss": 193.176,
      "step": 10830
    },
    {
      "epoch": 0.043794971658512344,
      "grad_norm": 3630.85693359375,
      "learning_rate": 1.7344000000000002e-05,
      "loss": 169.7418,
      "step": 10840
    },
    {
      "epoch": 0.04383537292387998,
      "grad_norm": 1167.7203369140625,
      "learning_rate": 1.736e-05,
      "loss": 201.4246,
      "step": 10850
    },
    {
      "epoch": 0.04387577418924761,
      "grad_norm": 888.1072998046875,
      "learning_rate": 1.7376e-05,
      "loss": 151.4057,
      "step": 10860
    },
    {
      "epoch": 0.043916175454615236,
      "grad_norm": 2073.16455078125,
      "learning_rate": 1.7392000000000002e-05,
      "loss": 187.2557,
      "step": 10870
    },
    {
      "epoch": 0.04395657671998287,
      "grad_norm": 1258.6611328125,
      "learning_rate": 1.7408e-05,
      "loss": 182.0184,
      "step": 10880
    },
    {
      "epoch": 0.0439969779853505,
      "grad_norm": 759.2626953125,
      "learning_rate": 1.7424e-05,
      "loss": 180.3921,
      "step": 10890
    },
    {
      "epoch": 0.044037379250718135,
      "grad_norm": 1016.7139282226562,
      "learning_rate": 1.7440000000000002e-05,
      "loss": 156.4396,
      "step": 10900
    },
    {
      "epoch": 0.04407778051608576,
      "grad_norm": 1495.8121337890625,
      "learning_rate": 1.7456e-05,
      "loss": 120.8279,
      "step": 10910
    },
    {
      "epoch": 0.044118181781453394,
      "grad_norm": 782.62109375,
      "learning_rate": 1.7472e-05,
      "loss": 135.7859,
      "step": 10920
    },
    {
      "epoch": 0.04415858304682103,
      "grad_norm": 1541.3516845703125,
      "learning_rate": 1.7488e-05,
      "loss": 142.1355,
      "step": 10930
    },
    {
      "epoch": 0.04419898431218866,
      "grad_norm": 1355.412109375,
      "learning_rate": 1.7504e-05,
      "loss": 165.6635,
      "step": 10940
    },
    {
      "epoch": 0.04423938557755629,
      "grad_norm": 974.6658325195312,
      "learning_rate": 1.752e-05,
      "loss": 121.126,
      "step": 10950
    },
    {
      "epoch": 0.04427978684292392,
      "grad_norm": 1273.1082763671875,
      "learning_rate": 1.7536000000000003e-05,
      "loss": 102.4414,
      "step": 10960
    },
    {
      "epoch": 0.04432018810829155,
      "grad_norm": 269.8813781738281,
      "learning_rate": 1.7552e-05,
      "loss": 127.8815,
      "step": 10970
    },
    {
      "epoch": 0.044360589373659186,
      "grad_norm": 2854.266357421875,
      "learning_rate": 1.7568000000000002e-05,
      "loss": 220.2031,
      "step": 10980
    },
    {
      "epoch": 0.04440099063902681,
      "grad_norm": 808.132080078125,
      "learning_rate": 1.7584e-05,
      "loss": 291.7426,
      "step": 10990
    },
    {
      "epoch": 0.044441391904394445,
      "grad_norm": 1838.3631591796875,
      "learning_rate": 1.76e-05,
      "loss": 170.3808,
      "step": 11000
    },
    {
      "epoch": 0.04448179316976208,
      "grad_norm": 0.0,
      "learning_rate": 1.7616000000000002e-05,
      "loss": 160.7307,
      "step": 11010
    },
    {
      "epoch": 0.04452219443512971,
      "grad_norm": 1464.164306640625,
      "learning_rate": 1.7632000000000003e-05,
      "loss": 132.0214,
      "step": 11020
    },
    {
      "epoch": 0.04456259570049734,
      "grad_norm": 1470.7813720703125,
      "learning_rate": 1.7648e-05,
      "loss": 177.7017,
      "step": 11030
    },
    {
      "epoch": 0.04460299696586497,
      "grad_norm": 673.0449829101562,
      "learning_rate": 1.7664000000000002e-05,
      "loss": 166.1588,
      "step": 11040
    },
    {
      "epoch": 0.0446433982312326,
      "grad_norm": 450.8736572265625,
      "learning_rate": 1.768e-05,
      "loss": 135.9254,
      "step": 11050
    },
    {
      "epoch": 0.044683799496600236,
      "grad_norm": 694.109130859375,
      "learning_rate": 1.7696e-05,
      "loss": 116.8292,
      "step": 11060
    },
    {
      "epoch": 0.04472420076196786,
      "grad_norm": 549.8363647460938,
      "learning_rate": 1.7712000000000003e-05,
      "loss": 115.2245,
      "step": 11070
    },
    {
      "epoch": 0.044764602027335496,
      "grad_norm": 1329.90087890625,
      "learning_rate": 1.7728e-05,
      "loss": 151.094,
      "step": 11080
    },
    {
      "epoch": 0.04480500329270313,
      "grad_norm": 1619.731201171875,
      "learning_rate": 1.7744e-05,
      "loss": 125.6026,
      "step": 11090
    },
    {
      "epoch": 0.04484540455807076,
      "grad_norm": 1591.700927734375,
      "learning_rate": 1.7760000000000003e-05,
      "loss": 210.2778,
      "step": 11100
    },
    {
      "epoch": 0.04488580582343839,
      "grad_norm": 1928.7462158203125,
      "learning_rate": 1.7776e-05,
      "loss": 168.4507,
      "step": 11110
    },
    {
      "epoch": 0.04492620708880602,
      "grad_norm": 1136.9661865234375,
      "learning_rate": 1.7792000000000002e-05,
      "loss": 159.5933,
      "step": 11120
    },
    {
      "epoch": 0.044966608354173654,
      "grad_norm": 1273.693603515625,
      "learning_rate": 1.7808e-05,
      "loss": 90.969,
      "step": 11130
    },
    {
      "epoch": 0.04500700961954129,
      "grad_norm": 6502.5478515625,
      "learning_rate": 1.7824e-05,
      "loss": 200.0657,
      "step": 11140
    },
    {
      "epoch": 0.04504741088490891,
      "grad_norm": 1432.283447265625,
      "learning_rate": 1.7840000000000002e-05,
      "loss": 167.2854,
      "step": 11150
    },
    {
      "epoch": 0.045087812150276546,
      "grad_norm": 1637.12890625,
      "learning_rate": 1.7856000000000003e-05,
      "loss": 151.2584,
      "step": 11160
    },
    {
      "epoch": 0.04512821341564418,
      "grad_norm": 3664.730224609375,
      "learning_rate": 1.7872e-05,
      "loss": 139.8826,
      "step": 11170
    },
    {
      "epoch": 0.04516861468101181,
      "grad_norm": 1181.7783203125,
      "learning_rate": 1.7888000000000002e-05,
      "loss": 125.8161,
      "step": 11180
    },
    {
      "epoch": 0.04520901594637944,
      "grad_norm": 3673.306884765625,
      "learning_rate": 1.7904e-05,
      "loss": 134.7778,
      "step": 11190
    },
    {
      "epoch": 0.04524941721174707,
      "grad_norm": 2812.187744140625,
      "learning_rate": 1.792e-05,
      "loss": 253.2139,
      "step": 11200
    },
    {
      "epoch": 0.045289818477114704,
      "grad_norm": 937.6162109375,
      "learning_rate": 1.7936000000000002e-05,
      "loss": 151.9196,
      "step": 11210
    },
    {
      "epoch": 0.04533021974248234,
      "grad_norm": 1033.5677490234375,
      "learning_rate": 1.7952e-05,
      "loss": 184.6166,
      "step": 11220
    },
    {
      "epoch": 0.045370621007849964,
      "grad_norm": 1049.357421875,
      "learning_rate": 1.7968e-05,
      "loss": 143.1292,
      "step": 11230
    },
    {
      "epoch": 0.0454110222732176,
      "grad_norm": 2090.681640625,
      "learning_rate": 1.7984000000000003e-05,
      "loss": 237.326,
      "step": 11240
    },
    {
      "epoch": 0.04545142353858523,
      "grad_norm": 588.9541625976562,
      "learning_rate": 1.8e-05,
      "loss": 136.8592,
      "step": 11250
    },
    {
      "epoch": 0.04549182480395286,
      "grad_norm": 989.3211059570312,
      "learning_rate": 1.8016e-05,
      "loss": 173.3406,
      "step": 11260
    },
    {
      "epoch": 0.04553222606932049,
      "grad_norm": 1291.71875,
      "learning_rate": 1.8032e-05,
      "loss": 161.0808,
      "step": 11270
    },
    {
      "epoch": 0.04557262733468812,
      "grad_norm": 820.1669921875,
      "learning_rate": 1.8048e-05,
      "loss": 197.2681,
      "step": 11280
    },
    {
      "epoch": 0.045613028600055755,
      "grad_norm": 3350.81591796875,
      "learning_rate": 1.8064000000000002e-05,
      "loss": 206.3674,
      "step": 11290
    },
    {
      "epoch": 0.04565342986542339,
      "grad_norm": 1087.4700927734375,
      "learning_rate": 1.8080000000000003e-05,
      "loss": 133.9378,
      "step": 11300
    },
    {
      "epoch": 0.045693831130791014,
      "grad_norm": 880.30419921875,
      "learning_rate": 1.8096e-05,
      "loss": 164.019,
      "step": 11310
    },
    {
      "epoch": 0.04573423239615865,
      "grad_norm": 962.7153930664062,
      "learning_rate": 1.8112000000000002e-05,
      "loss": 178.859,
      "step": 11320
    },
    {
      "epoch": 0.04577463366152628,
      "grad_norm": 6233.5029296875,
      "learning_rate": 1.8128e-05,
      "loss": 187.2053,
      "step": 11330
    },
    {
      "epoch": 0.04581503492689391,
      "grad_norm": 1161.468994140625,
      "learning_rate": 1.8144e-05,
      "loss": 169.2743,
      "step": 11340
    },
    {
      "epoch": 0.04585543619226154,
      "grad_norm": 1120.290283203125,
      "learning_rate": 1.8160000000000002e-05,
      "loss": 172.2986,
      "step": 11350
    },
    {
      "epoch": 0.04589583745762917,
      "grad_norm": 3448.97705078125,
      "learning_rate": 1.8176000000000004e-05,
      "loss": 206.6683,
      "step": 11360
    },
    {
      "epoch": 0.045936238722996806,
      "grad_norm": 1253.4058837890625,
      "learning_rate": 1.8192e-05,
      "loss": 139.1167,
      "step": 11370
    },
    {
      "epoch": 0.04597663998836444,
      "grad_norm": 924.9556884765625,
      "learning_rate": 1.8208000000000003e-05,
      "loss": 148.5544,
      "step": 11380
    },
    {
      "epoch": 0.046017041253732065,
      "grad_norm": 660.0718994140625,
      "learning_rate": 1.8224e-05,
      "loss": 126.323,
      "step": 11390
    },
    {
      "epoch": 0.0460574425190997,
      "grad_norm": 822.0247802734375,
      "learning_rate": 1.824e-05,
      "loss": 146.354,
      "step": 11400
    },
    {
      "epoch": 0.04609784378446733,
      "grad_norm": 1849.636474609375,
      "learning_rate": 1.8256e-05,
      "loss": 131.2848,
      "step": 11410
    },
    {
      "epoch": 0.046138245049834964,
      "grad_norm": 4517.93212890625,
      "learning_rate": 1.8272e-05,
      "loss": 184.1627,
      "step": 11420
    },
    {
      "epoch": 0.04617864631520259,
      "grad_norm": 1306.6082763671875,
      "learning_rate": 1.8288000000000002e-05,
      "loss": 152.2419,
      "step": 11430
    },
    {
      "epoch": 0.04621904758057022,
      "grad_norm": 913.3738403320312,
      "learning_rate": 1.8304000000000003e-05,
      "loss": 174.6116,
      "step": 11440
    },
    {
      "epoch": 0.046259448845937856,
      "grad_norm": 801.408447265625,
      "learning_rate": 1.832e-05,
      "loss": 161.1145,
      "step": 11450
    },
    {
      "epoch": 0.04629985011130549,
      "grad_norm": 1175.022216796875,
      "learning_rate": 1.8336000000000002e-05,
      "loss": 183.3758,
      "step": 11460
    },
    {
      "epoch": 0.046340251376673115,
      "grad_norm": 1856.43701171875,
      "learning_rate": 1.8352e-05,
      "loss": 161.8469,
      "step": 11470
    },
    {
      "epoch": 0.04638065264204075,
      "grad_norm": 1363.8883056640625,
      "learning_rate": 1.8368e-05,
      "loss": 202.2804,
      "step": 11480
    },
    {
      "epoch": 0.04642105390740838,
      "grad_norm": 1238.9725341796875,
      "learning_rate": 1.8384000000000002e-05,
      "loss": 142.8043,
      "step": 11490
    },
    {
      "epoch": 0.046461455172776014,
      "grad_norm": 1129.2506103515625,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 153.3768,
      "step": 11500
    },
    {
      "epoch": 0.04650185643814364,
      "grad_norm": 642.8306274414062,
      "learning_rate": 1.8416e-05,
      "loss": 152.1908,
      "step": 11510
    },
    {
      "epoch": 0.046542257703511274,
      "grad_norm": 480.4953918457031,
      "learning_rate": 1.8432000000000002e-05,
      "loss": 151.4173,
      "step": 11520
    },
    {
      "epoch": 0.04658265896887891,
      "grad_norm": 702.656005859375,
      "learning_rate": 1.8448e-05,
      "loss": 131.2011,
      "step": 11530
    },
    {
      "epoch": 0.04662306023424654,
      "grad_norm": 1275.0462646484375,
      "learning_rate": 1.8464e-05,
      "loss": 171.1476,
      "step": 11540
    },
    {
      "epoch": 0.046663461499614166,
      "grad_norm": 1706.020263671875,
      "learning_rate": 1.8480000000000003e-05,
      "loss": 194.7064,
      "step": 11550
    },
    {
      "epoch": 0.0467038627649818,
      "grad_norm": 1688.6947021484375,
      "learning_rate": 1.8496e-05,
      "loss": 244.8772,
      "step": 11560
    },
    {
      "epoch": 0.04674426403034943,
      "grad_norm": 1043.65087890625,
      "learning_rate": 1.8512e-05,
      "loss": 149.636,
      "step": 11570
    },
    {
      "epoch": 0.046784665295717065,
      "grad_norm": 614.2929077148438,
      "learning_rate": 1.8528000000000003e-05,
      "loss": 90.4073,
      "step": 11580
    },
    {
      "epoch": 0.04682506656108469,
      "grad_norm": 1882.9847412109375,
      "learning_rate": 1.8544e-05,
      "loss": 142.7772,
      "step": 11590
    },
    {
      "epoch": 0.046865467826452324,
      "grad_norm": 1458.3114013671875,
      "learning_rate": 1.8560000000000002e-05,
      "loss": 128.3059,
      "step": 11600
    },
    {
      "epoch": 0.04690586909181996,
      "grad_norm": 1554.3670654296875,
      "learning_rate": 1.8576e-05,
      "loss": 92.2401,
      "step": 11610
    },
    {
      "epoch": 0.04694627035718759,
      "grad_norm": 1737.34130859375,
      "learning_rate": 1.8592e-05,
      "loss": 151.1104,
      "step": 11620
    },
    {
      "epoch": 0.046986671622555216,
      "grad_norm": 1246.5958251953125,
      "learning_rate": 1.8608000000000002e-05,
      "loss": 189.7216,
      "step": 11630
    },
    {
      "epoch": 0.04702707288792285,
      "grad_norm": 1355.0965576171875,
      "learning_rate": 1.8624000000000003e-05,
      "loss": 164.5095,
      "step": 11640
    },
    {
      "epoch": 0.04706747415329048,
      "grad_norm": 1271.6949462890625,
      "learning_rate": 1.864e-05,
      "loss": 168.5206,
      "step": 11650
    },
    {
      "epoch": 0.047107875418658116,
      "grad_norm": 1827.9058837890625,
      "learning_rate": 1.8656000000000002e-05,
      "loss": 92.2439,
      "step": 11660
    },
    {
      "epoch": 0.04714827668402574,
      "grad_norm": 497.3401184082031,
      "learning_rate": 1.8672e-05,
      "loss": 77.0604,
      "step": 11670
    },
    {
      "epoch": 0.047188677949393375,
      "grad_norm": 969.7373046875,
      "learning_rate": 1.8688e-05,
      "loss": 244.0555,
      "step": 11680
    },
    {
      "epoch": 0.04722907921476101,
      "grad_norm": 2381.234130859375,
      "learning_rate": 1.8704000000000003e-05,
      "loss": 249.9872,
      "step": 11690
    },
    {
      "epoch": 0.04726948048012864,
      "grad_norm": 1113.687744140625,
      "learning_rate": 1.8720000000000004e-05,
      "loss": 140.6708,
      "step": 11700
    },
    {
      "epoch": 0.04730988174549627,
      "grad_norm": 1018.4200439453125,
      "learning_rate": 1.8736e-05,
      "loss": 123.3354,
      "step": 11710
    },
    {
      "epoch": 0.0473502830108639,
      "grad_norm": 1131.4271240234375,
      "learning_rate": 1.8752000000000003e-05,
      "loss": 151.4623,
      "step": 11720
    },
    {
      "epoch": 0.04739068427623153,
      "grad_norm": 727.86376953125,
      "learning_rate": 1.8768e-05,
      "loss": 161.0078,
      "step": 11730
    },
    {
      "epoch": 0.047431085541599166,
      "grad_norm": 1448.9930419921875,
      "learning_rate": 1.8784000000000002e-05,
      "loss": 132.0155,
      "step": 11740
    },
    {
      "epoch": 0.04747148680696679,
      "grad_norm": 1065.7197265625,
      "learning_rate": 1.88e-05,
      "loss": 178.1806,
      "step": 11750
    },
    {
      "epoch": 0.047511888072334425,
      "grad_norm": 888.4298706054688,
      "learning_rate": 1.8816e-05,
      "loss": 99.7927,
      "step": 11760
    },
    {
      "epoch": 0.04755228933770206,
      "grad_norm": 3456.236083984375,
      "learning_rate": 1.8832000000000002e-05,
      "loss": 172.4764,
      "step": 11770
    },
    {
      "epoch": 0.04759269060306969,
      "grad_norm": 1027.8367919921875,
      "learning_rate": 1.8848000000000003e-05,
      "loss": 96.5219,
      "step": 11780
    },
    {
      "epoch": 0.04763309186843732,
      "grad_norm": 2561.343505859375,
      "learning_rate": 1.8864e-05,
      "loss": 152.5547,
      "step": 11790
    },
    {
      "epoch": 0.04767349313380495,
      "grad_norm": 726.5477294921875,
      "learning_rate": 1.8880000000000002e-05,
      "loss": 163.8758,
      "step": 11800
    },
    {
      "epoch": 0.047713894399172584,
      "grad_norm": 497.03021240234375,
      "learning_rate": 1.8896e-05,
      "loss": 165.0003,
      "step": 11810
    },
    {
      "epoch": 0.04775429566454022,
      "grad_norm": 1021.0319213867188,
      "learning_rate": 1.8912e-05,
      "loss": 144.085,
      "step": 11820
    },
    {
      "epoch": 0.04779469692990784,
      "grad_norm": 803.744384765625,
      "learning_rate": 1.8928000000000002e-05,
      "loss": 163.5293,
      "step": 11830
    },
    {
      "epoch": 0.047835098195275476,
      "grad_norm": 1352.6136474609375,
      "learning_rate": 1.8944000000000004e-05,
      "loss": 156.9807,
      "step": 11840
    },
    {
      "epoch": 0.04787549946064311,
      "grad_norm": 1498.0391845703125,
      "learning_rate": 1.896e-05,
      "loss": 136.0139,
      "step": 11850
    },
    {
      "epoch": 0.04791590072601074,
      "grad_norm": 2074.9453125,
      "learning_rate": 1.8976000000000003e-05,
      "loss": 135.7337,
      "step": 11860
    },
    {
      "epoch": 0.04795630199137837,
      "grad_norm": 1958.399658203125,
      "learning_rate": 1.8992e-05,
      "loss": 169.1383,
      "step": 11870
    },
    {
      "epoch": 0.047996703256746,
      "grad_norm": 1390.4888916015625,
      "learning_rate": 1.9008e-05,
      "loss": 195.9214,
      "step": 11880
    },
    {
      "epoch": 0.048037104522113634,
      "grad_norm": 1233.4976806640625,
      "learning_rate": 1.9024000000000003e-05,
      "loss": 185.4625,
      "step": 11890
    },
    {
      "epoch": 0.04807750578748127,
      "grad_norm": 7298.0615234375,
      "learning_rate": 1.904e-05,
      "loss": 208.103,
      "step": 11900
    },
    {
      "epoch": 0.04811790705284889,
      "grad_norm": 1131.549560546875,
      "learning_rate": 1.9056000000000002e-05,
      "loss": 130.0675,
      "step": 11910
    },
    {
      "epoch": 0.048158308318216526,
      "grad_norm": 4298.8173828125,
      "learning_rate": 1.9072000000000003e-05,
      "loss": 130.8222,
      "step": 11920
    },
    {
      "epoch": 0.04819870958358416,
      "grad_norm": 1270.93505859375,
      "learning_rate": 1.9088e-05,
      "loss": 138.4447,
      "step": 11930
    },
    {
      "epoch": 0.04823911084895179,
      "grad_norm": 2340.214111328125,
      "learning_rate": 1.9104000000000002e-05,
      "loss": 176.73,
      "step": 11940
    },
    {
      "epoch": 0.04827951211431942,
      "grad_norm": 1349.122314453125,
      "learning_rate": 1.912e-05,
      "loss": 209.7494,
      "step": 11950
    },
    {
      "epoch": 0.04831991337968705,
      "grad_norm": 1326.8984375,
      "learning_rate": 1.9136e-05,
      "loss": 117.3182,
      "step": 11960
    },
    {
      "epoch": 0.048360314645054685,
      "grad_norm": 868.340576171875,
      "learning_rate": 1.9152000000000002e-05,
      "loss": 108.6576,
      "step": 11970
    },
    {
      "epoch": 0.04840071591042232,
      "grad_norm": 754.5211791992188,
      "learning_rate": 1.9168000000000004e-05,
      "loss": 160.192,
      "step": 11980
    },
    {
      "epoch": 0.048441117175789944,
      "grad_norm": 2791.62744140625,
      "learning_rate": 1.9184e-05,
      "loss": 207.5689,
      "step": 11990
    },
    {
      "epoch": 0.04848151844115758,
      "grad_norm": 653.2196655273438,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 171.7243,
      "step": 12000
    },
    {
      "epoch": 0.04852191970652521,
      "grad_norm": 1092.5750732421875,
      "learning_rate": 1.9216e-05,
      "loss": 157.921,
      "step": 12010
    },
    {
      "epoch": 0.04856232097189284,
      "grad_norm": 1124.6910400390625,
      "learning_rate": 1.9232e-05,
      "loss": 152.1631,
      "step": 12020
    },
    {
      "epoch": 0.04860272223726047,
      "grad_norm": 996.9329223632812,
      "learning_rate": 1.9248000000000003e-05,
      "loss": 199.5053,
      "step": 12030
    },
    {
      "epoch": 0.0486431235026281,
      "grad_norm": 2939.33447265625,
      "learning_rate": 1.9264e-05,
      "loss": 125.7709,
      "step": 12040
    },
    {
      "epoch": 0.048683524767995735,
      "grad_norm": 620.4122924804688,
      "learning_rate": 1.9280000000000002e-05,
      "loss": 155.2206,
      "step": 12050
    },
    {
      "epoch": 0.04872392603336337,
      "grad_norm": 987.931884765625,
      "learning_rate": 1.9296000000000003e-05,
      "loss": 144.0866,
      "step": 12060
    },
    {
      "epoch": 0.048764327298730994,
      "grad_norm": 1881.5980224609375,
      "learning_rate": 1.9312e-05,
      "loss": 145.1766,
      "step": 12070
    },
    {
      "epoch": 0.04880472856409863,
      "grad_norm": 2606.857177734375,
      "learning_rate": 1.9328000000000002e-05,
      "loss": 121.8912,
      "step": 12080
    },
    {
      "epoch": 0.04884512982946626,
      "grad_norm": 1632.544677734375,
      "learning_rate": 1.9344e-05,
      "loss": 233.9964,
      "step": 12090
    },
    {
      "epoch": 0.048885531094833894,
      "grad_norm": 1396.0145263671875,
      "learning_rate": 1.936e-05,
      "loss": 199.8086,
      "step": 12100
    },
    {
      "epoch": 0.04892593236020152,
      "grad_norm": 2512.758056640625,
      "learning_rate": 1.9376000000000002e-05,
      "loss": 200.7336,
      "step": 12110
    },
    {
      "epoch": 0.04896633362556915,
      "grad_norm": 1487.132080078125,
      "learning_rate": 1.9392000000000003e-05,
      "loss": 192.9431,
      "step": 12120
    },
    {
      "epoch": 0.049006734890936786,
      "grad_norm": 1734.700439453125,
      "learning_rate": 1.9408e-05,
      "loss": 163.1206,
      "step": 12130
    },
    {
      "epoch": 0.04904713615630442,
      "grad_norm": 898.8317260742188,
      "learning_rate": 1.9424e-05,
      "loss": 161.1301,
      "step": 12140
    },
    {
      "epoch": 0.049087537421672045,
      "grad_norm": 1437.3121337890625,
      "learning_rate": 1.944e-05,
      "loss": 177.5109,
      "step": 12150
    },
    {
      "epoch": 0.04912793868703968,
      "grad_norm": 5169.5986328125,
      "learning_rate": 1.9456e-05,
      "loss": 210.6177,
      "step": 12160
    },
    {
      "epoch": 0.04916833995240731,
      "grad_norm": 1230.391357421875,
      "learning_rate": 1.9472000000000003e-05,
      "loss": 187.2195,
      "step": 12170
    },
    {
      "epoch": 0.04920874121777494,
      "grad_norm": 3862.664794921875,
      "learning_rate": 1.9488000000000004e-05,
      "loss": 177.8196,
      "step": 12180
    },
    {
      "epoch": 0.04924914248314257,
      "grad_norm": 1136.101318359375,
      "learning_rate": 1.9504e-05,
      "loss": 177.7476,
      "step": 12190
    },
    {
      "epoch": 0.0492895437485102,
      "grad_norm": 2034.5399169921875,
      "learning_rate": 1.9520000000000003e-05,
      "loss": 108.4205,
      "step": 12200
    },
    {
      "epoch": 0.049329945013877836,
      "grad_norm": 3904.319091796875,
      "learning_rate": 1.9536e-05,
      "loss": 125.3061,
      "step": 12210
    },
    {
      "epoch": 0.04937034627924546,
      "grad_norm": 1803.0955810546875,
      "learning_rate": 1.9552000000000002e-05,
      "loss": 128.3449,
      "step": 12220
    },
    {
      "epoch": 0.049410747544613096,
      "grad_norm": 1118.046875,
      "learning_rate": 1.9568000000000003e-05,
      "loss": 199.2001,
      "step": 12230
    },
    {
      "epoch": 0.04945114880998073,
      "grad_norm": 958.41259765625,
      "learning_rate": 1.9584e-05,
      "loss": 194.366,
      "step": 12240
    },
    {
      "epoch": 0.04949155007534836,
      "grad_norm": 4219.087890625,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 181.3084,
      "step": 12250
    },
    {
      "epoch": 0.04953195134071599,
      "grad_norm": 743.9143676757812,
      "learning_rate": 1.9616000000000003e-05,
      "loss": 211.1867,
      "step": 12260
    },
    {
      "epoch": 0.04957235260608362,
      "grad_norm": 1083.905029296875,
      "learning_rate": 1.9632e-05,
      "loss": 133.8096,
      "step": 12270
    },
    {
      "epoch": 0.049612753871451254,
      "grad_norm": 906.1483764648438,
      "learning_rate": 1.9648000000000002e-05,
      "loss": 171.5658,
      "step": 12280
    },
    {
      "epoch": 0.04965315513681889,
      "grad_norm": 717.866455078125,
      "learning_rate": 1.9664e-05,
      "loss": 113.4374,
      "step": 12290
    },
    {
      "epoch": 0.04969355640218651,
      "grad_norm": 1161.523681640625,
      "learning_rate": 1.968e-05,
      "loss": 173.7262,
      "step": 12300
    },
    {
      "epoch": 0.049733957667554146,
      "grad_norm": 1218.2979736328125,
      "learning_rate": 1.9696000000000003e-05,
      "loss": 121.5765,
      "step": 12310
    },
    {
      "epoch": 0.04977435893292178,
      "grad_norm": 1055.20166015625,
      "learning_rate": 1.9712000000000004e-05,
      "loss": 144.3302,
      "step": 12320
    },
    {
      "epoch": 0.04981476019828941,
      "grad_norm": 1507.555419921875,
      "learning_rate": 1.9728e-05,
      "loss": 144.8477,
      "step": 12330
    },
    {
      "epoch": 0.04985516146365704,
      "grad_norm": 1124.2626953125,
      "learning_rate": 1.9744e-05,
      "loss": 216.2754,
      "step": 12340
    },
    {
      "epoch": 0.04989556272902467,
      "grad_norm": 1329.7135009765625,
      "learning_rate": 1.976e-05,
      "loss": 169.4856,
      "step": 12350
    },
    {
      "epoch": 0.049935963994392304,
      "grad_norm": 1700.2542724609375,
      "learning_rate": 1.9776000000000002e-05,
      "loss": 160.8507,
      "step": 12360
    },
    {
      "epoch": 0.04997636525975994,
      "grad_norm": 5681.53662109375,
      "learning_rate": 1.9792000000000003e-05,
      "loss": 175.2828,
      "step": 12370
    },
    {
      "epoch": 0.050016766525127564,
      "grad_norm": 1659.982177734375,
      "learning_rate": 1.9808e-05,
      "loss": 159.9759,
      "step": 12380
    },
    {
      "epoch": 0.0500571677904952,
      "grad_norm": 2237.371337890625,
      "learning_rate": 1.9824000000000002e-05,
      "loss": 230.2747,
      "step": 12390
    },
    {
      "epoch": 0.05009756905586283,
      "grad_norm": 12513.3671875,
      "learning_rate": 1.9840000000000003e-05,
      "loss": 155.1298,
      "step": 12400
    },
    {
      "epoch": 0.05013797032123046,
      "grad_norm": 3976.13232421875,
      "learning_rate": 1.9856e-05,
      "loss": 173.9067,
      "step": 12410
    },
    {
      "epoch": 0.05017837158659809,
      "grad_norm": 5326.07080078125,
      "learning_rate": 1.9872000000000002e-05,
      "loss": 141.2005,
      "step": 12420
    },
    {
      "epoch": 0.05021877285196572,
      "grad_norm": 1620.393310546875,
      "learning_rate": 1.9888e-05,
      "loss": 192.9733,
      "step": 12430
    },
    {
      "epoch": 0.050259174117333355,
      "grad_norm": 3446.421142578125,
      "learning_rate": 1.9904e-05,
      "loss": 244.3542,
      "step": 12440
    },
    {
      "epoch": 0.05029957538270099,
      "grad_norm": 2199.027099609375,
      "learning_rate": 1.9920000000000002e-05,
      "loss": 140.0264,
      "step": 12450
    },
    {
      "epoch": 0.050339976648068614,
      "grad_norm": 2481.56494140625,
      "learning_rate": 1.9936000000000004e-05,
      "loss": 115.8152,
      "step": 12460
    },
    {
      "epoch": 0.05038037791343625,
      "grad_norm": 691.91748046875,
      "learning_rate": 1.9952e-05,
      "loss": 70.8518,
      "step": 12470
    },
    {
      "epoch": 0.05042077917880388,
      "grad_norm": 717.9995727539062,
      "learning_rate": 1.9968e-05,
      "loss": 198.1902,
      "step": 12480
    },
    {
      "epoch": 0.05046118044417151,
      "grad_norm": 790.2597045898438,
      "learning_rate": 1.9984e-05,
      "loss": 171.2203,
      "step": 12490
    },
    {
      "epoch": 0.05050158170953914,
      "grad_norm": 1000.937255859375,
      "learning_rate": 2e-05,
      "loss": 124.132,
      "step": 12500
    },
    {
      "epoch": 0.05054198297490677,
      "grad_norm": 2628.76708984375,
      "learning_rate": 2.0016e-05,
      "loss": 137.946,
      "step": 12510
    },
    {
      "epoch": 0.050582384240274406,
      "grad_norm": 3296.44287109375,
      "learning_rate": 2.0032000000000004e-05,
      "loss": 138.9817,
      "step": 12520
    },
    {
      "epoch": 0.05062278550564204,
      "grad_norm": 1251.9384765625,
      "learning_rate": 2.0048000000000002e-05,
      "loss": 116.2065,
      "step": 12530
    },
    {
      "epoch": 0.050663186771009665,
      "grad_norm": 3689.34814453125,
      "learning_rate": 2.0064000000000003e-05,
      "loss": 190.9615,
      "step": 12540
    },
    {
      "epoch": 0.0507035880363773,
      "grad_norm": 1161.31787109375,
      "learning_rate": 2.008e-05,
      "loss": 165.3648,
      "step": 12550
    },
    {
      "epoch": 0.05074398930174493,
      "grad_norm": 1477.417724609375,
      "learning_rate": 2.0096e-05,
      "loss": 181.9141,
      "step": 12560
    },
    {
      "epoch": 0.050784390567112564,
      "grad_norm": 699.610595703125,
      "learning_rate": 2.0112000000000003e-05,
      "loss": 209.1569,
      "step": 12570
    },
    {
      "epoch": 0.05082479183248019,
      "grad_norm": 520.3313598632812,
      "learning_rate": 2.0128e-05,
      "loss": 140.6631,
      "step": 12580
    },
    {
      "epoch": 0.05086519309784782,
      "grad_norm": 443.720947265625,
      "learning_rate": 2.0144000000000002e-05,
      "loss": 185.0845,
      "step": 12590
    },
    {
      "epoch": 0.050905594363215456,
      "grad_norm": 905.438232421875,
      "learning_rate": 2.016e-05,
      "loss": 113.3908,
      "step": 12600
    },
    {
      "epoch": 0.05094599562858309,
      "grad_norm": 1029.767578125,
      "learning_rate": 2.0176e-05,
      "loss": 161.9168,
      "step": 12610
    },
    {
      "epoch": 0.050986396893950715,
      "grad_norm": 809.9927978515625,
      "learning_rate": 2.0192000000000003e-05,
      "loss": 196.2517,
      "step": 12620
    },
    {
      "epoch": 0.05102679815931835,
      "grad_norm": 1245.3720703125,
      "learning_rate": 2.0208e-05,
      "loss": 162.7455,
      "step": 12630
    },
    {
      "epoch": 0.05106719942468598,
      "grad_norm": 461.35711669921875,
      "learning_rate": 2.0224000000000005e-05,
      "loss": 197.058,
      "step": 12640
    },
    {
      "epoch": 0.051107600690053615,
      "grad_norm": 0.0,
      "learning_rate": 2.0240000000000003e-05,
      "loss": 159.2665,
      "step": 12650
    },
    {
      "epoch": 0.05114800195542124,
      "grad_norm": 1341.0728759765625,
      "learning_rate": 2.0256e-05,
      "loss": 153.8318,
      "step": 12660
    },
    {
      "epoch": 0.051188403220788874,
      "grad_norm": 2662.817626953125,
      "learning_rate": 2.0272000000000002e-05,
      "loss": 164.7002,
      "step": 12670
    },
    {
      "epoch": 0.05122880448615651,
      "grad_norm": 2191.09423828125,
      "learning_rate": 2.0288e-05,
      "loss": 146.5947,
      "step": 12680
    },
    {
      "epoch": 0.05126920575152414,
      "grad_norm": 1247.1414794921875,
      "learning_rate": 2.0304000000000004e-05,
      "loss": 175.3468,
      "step": 12690
    },
    {
      "epoch": 0.051309607016891766,
      "grad_norm": 728.5390625,
      "learning_rate": 2.0320000000000002e-05,
      "loss": 133.0505,
      "step": 12700
    },
    {
      "epoch": 0.0513500082822594,
      "grad_norm": 830.8687744140625,
      "learning_rate": 2.0336e-05,
      "loss": 108.7042,
      "step": 12710
    },
    {
      "epoch": 0.05139040954762703,
      "grad_norm": 879.9666137695312,
      "learning_rate": 2.0352000000000004e-05,
      "loss": 129.6018,
      "step": 12720
    },
    {
      "epoch": 0.051430810812994665,
      "grad_norm": 1011.3085327148438,
      "learning_rate": 2.0368000000000002e-05,
      "loss": 173.0924,
      "step": 12730
    },
    {
      "epoch": 0.05147121207836229,
      "grad_norm": 1851.60107421875,
      "learning_rate": 2.0384000000000003e-05,
      "loss": 145.7644,
      "step": 12740
    },
    {
      "epoch": 0.051511613343729924,
      "grad_norm": 663.5894165039062,
      "learning_rate": 2.04e-05,
      "loss": 121.5047,
      "step": 12750
    },
    {
      "epoch": 0.05155201460909756,
      "grad_norm": 714.2340698242188,
      "learning_rate": 2.0416e-05,
      "loss": 182.6054,
      "step": 12760
    },
    {
      "epoch": 0.05159241587446519,
      "grad_norm": 3369.689697265625,
      "learning_rate": 2.0432000000000004e-05,
      "loss": 197.5232,
      "step": 12770
    },
    {
      "epoch": 0.051632817139832816,
      "grad_norm": 1169.3397216796875,
      "learning_rate": 2.0448e-05,
      "loss": 167.386,
      "step": 12780
    },
    {
      "epoch": 0.05167321840520045,
      "grad_norm": 875.116943359375,
      "learning_rate": 2.0464000000000003e-05,
      "loss": 174.5069,
      "step": 12790
    },
    {
      "epoch": 0.05171361967056808,
      "grad_norm": 2895.494140625,
      "learning_rate": 2.048e-05,
      "loss": 175.5416,
      "step": 12800
    },
    {
      "epoch": 0.051754020935935716,
      "grad_norm": 1204.8529052734375,
      "learning_rate": 2.0496e-05,
      "loss": 232.9305,
      "step": 12810
    },
    {
      "epoch": 0.05179442220130334,
      "grad_norm": 966.132080078125,
      "learning_rate": 2.0512000000000003e-05,
      "loss": 155.4237,
      "step": 12820
    },
    {
      "epoch": 0.051834823466670975,
      "grad_norm": 1232.181884765625,
      "learning_rate": 2.0528e-05,
      "loss": 200.3393,
      "step": 12830
    },
    {
      "epoch": 0.05187522473203861,
      "grad_norm": 794.1149291992188,
      "learning_rate": 2.0544e-05,
      "loss": 143.9225,
      "step": 12840
    },
    {
      "epoch": 0.05191562599740624,
      "grad_norm": 911.0679321289062,
      "learning_rate": 2.0560000000000003e-05,
      "loss": 130.7181,
      "step": 12850
    },
    {
      "epoch": 0.05195602726277387,
      "grad_norm": 577.2883911132812,
      "learning_rate": 2.0576e-05,
      "loss": 118.0697,
      "step": 12860
    },
    {
      "epoch": 0.0519964285281415,
      "grad_norm": 1354.333740234375,
      "learning_rate": 2.0592000000000002e-05,
      "loss": 154.2893,
      "step": 12870
    },
    {
      "epoch": 0.05203682979350913,
      "grad_norm": 2138.07958984375,
      "learning_rate": 2.0608e-05,
      "loss": 151.4154,
      "step": 12880
    },
    {
      "epoch": 0.052077231058876766,
      "grad_norm": 4152.9033203125,
      "learning_rate": 2.0624e-05,
      "loss": 155.5429,
      "step": 12890
    },
    {
      "epoch": 0.05211763232424439,
      "grad_norm": 714.933349609375,
      "learning_rate": 2.0640000000000002e-05,
      "loss": 143.2872,
      "step": 12900
    },
    {
      "epoch": 0.052158033589612025,
      "grad_norm": 759.1387329101562,
      "learning_rate": 2.0656e-05,
      "loss": 151.3617,
      "step": 12910
    },
    {
      "epoch": 0.05219843485497966,
      "grad_norm": 1528.81494140625,
      "learning_rate": 2.0672000000000005e-05,
      "loss": 180.9476,
      "step": 12920
    },
    {
      "epoch": 0.05223883612034729,
      "grad_norm": 1480.501220703125,
      "learning_rate": 2.0688000000000003e-05,
      "loss": 162.473,
      "step": 12930
    },
    {
      "epoch": 0.05227923738571492,
      "grad_norm": 1521.9293212890625,
      "learning_rate": 2.0704e-05,
      "loss": 166.9442,
      "step": 12940
    },
    {
      "epoch": 0.05231963865108255,
      "grad_norm": 2496.98828125,
      "learning_rate": 2.072e-05,
      "loss": 167.1679,
      "step": 12950
    },
    {
      "epoch": 0.052360039916450184,
      "grad_norm": 1867.0335693359375,
      "learning_rate": 2.0736e-05,
      "loss": 207.2721,
      "step": 12960
    },
    {
      "epoch": 0.05240044118181782,
      "grad_norm": 626.9522094726562,
      "learning_rate": 2.0752000000000004e-05,
      "loss": 144.8393,
      "step": 12970
    },
    {
      "epoch": 0.05244084244718544,
      "grad_norm": 870.9942016601562,
      "learning_rate": 2.0768000000000002e-05,
      "loss": 115.7043,
      "step": 12980
    },
    {
      "epoch": 0.052481243712553076,
      "grad_norm": 652.1517944335938,
      "learning_rate": 2.0784e-05,
      "loss": 96.7211,
      "step": 12990
    },
    {
      "epoch": 0.05252164497792071,
      "grad_norm": 1137.64697265625,
      "learning_rate": 2.08e-05,
      "loss": 135.8071,
      "step": 13000
    },
    {
      "epoch": 0.05256204624328834,
      "grad_norm": 1930.1585693359375,
      "learning_rate": 2.0816000000000002e-05,
      "loss": 136.4251,
      "step": 13010
    },
    {
      "epoch": 0.05260244750865597,
      "grad_norm": 907.9906005859375,
      "learning_rate": 2.0832000000000003e-05,
      "loss": 89.2023,
      "step": 13020
    },
    {
      "epoch": 0.0526428487740236,
      "grad_norm": 558.8703002929688,
      "learning_rate": 2.0848e-05,
      "loss": 147.8288,
      "step": 13030
    },
    {
      "epoch": 0.052683250039391234,
      "grad_norm": 2412.71484375,
      "learning_rate": 2.0864e-05,
      "loss": 137.4368,
      "step": 13040
    },
    {
      "epoch": 0.05272365130475887,
      "grad_norm": 1965.6383056640625,
      "learning_rate": 2.0880000000000003e-05,
      "loss": 105.5261,
      "step": 13050
    },
    {
      "epoch": 0.05276405257012649,
      "grad_norm": 960.9400634765625,
      "learning_rate": 2.0896e-05,
      "loss": 120.9263,
      "step": 13060
    },
    {
      "epoch": 0.052804453835494126,
      "grad_norm": 0.0,
      "learning_rate": 2.0912000000000002e-05,
      "loss": 208.6261,
      "step": 13070
    },
    {
      "epoch": 0.05284485510086176,
      "grad_norm": 807.88623046875,
      "learning_rate": 2.0928e-05,
      "loss": 135.5652,
      "step": 13080
    },
    {
      "epoch": 0.05288525636622939,
      "grad_norm": 683.6365356445312,
      "learning_rate": 2.0944e-05,
      "loss": 122.0361,
      "step": 13090
    },
    {
      "epoch": 0.05292565763159702,
      "grad_norm": 6482.541015625,
      "learning_rate": 2.0960000000000003e-05,
      "loss": 118.127,
      "step": 13100
    },
    {
      "epoch": 0.05296605889696465,
      "grad_norm": 861.0405883789062,
      "learning_rate": 2.0976e-05,
      "loss": 127.4016,
      "step": 13110
    },
    {
      "epoch": 0.053006460162332285,
      "grad_norm": 2566.6220703125,
      "learning_rate": 2.0992000000000005e-05,
      "loss": 116.446,
      "step": 13120
    },
    {
      "epoch": 0.05304686142769992,
      "grad_norm": 1192.2772216796875,
      "learning_rate": 2.1008000000000003e-05,
      "loss": 202.4242,
      "step": 13130
    },
    {
      "epoch": 0.053087262693067544,
      "grad_norm": 598.492431640625,
      "learning_rate": 2.1024e-05,
      "loss": 169.9234,
      "step": 13140
    },
    {
      "epoch": 0.05312766395843518,
      "grad_norm": 532.2695922851562,
      "learning_rate": 2.1040000000000002e-05,
      "loss": 151.1845,
      "step": 13150
    },
    {
      "epoch": 0.05316806522380281,
      "grad_norm": 2847.472412109375,
      "learning_rate": 2.1056e-05,
      "loss": 222.5728,
      "step": 13160
    },
    {
      "epoch": 0.05320846648917044,
      "grad_norm": 3605.068603515625,
      "learning_rate": 2.1072000000000004e-05,
      "loss": 154.8115,
      "step": 13170
    },
    {
      "epoch": 0.05324886775453807,
      "grad_norm": 2527.835205078125,
      "learning_rate": 2.1088000000000002e-05,
      "loss": 216.4981,
      "step": 13180
    },
    {
      "epoch": 0.0532892690199057,
      "grad_norm": 559.9857177734375,
      "learning_rate": 2.1104e-05,
      "loss": 106.2085,
      "step": 13190
    },
    {
      "epoch": 0.053329670285273335,
      "grad_norm": 1436.94287109375,
      "learning_rate": 2.112e-05,
      "loss": 127.6425,
      "step": 13200
    },
    {
      "epoch": 0.05337007155064097,
      "grad_norm": 821.7449340820312,
      "learning_rate": 2.1136000000000002e-05,
      "loss": 244.2373,
      "step": 13210
    },
    {
      "epoch": 0.053410472816008595,
      "grad_norm": 887.7293090820312,
      "learning_rate": 2.1152000000000003e-05,
      "loss": 103.6638,
      "step": 13220
    },
    {
      "epoch": 0.05345087408137623,
      "grad_norm": 574.5474243164062,
      "learning_rate": 2.1168e-05,
      "loss": 172.9593,
      "step": 13230
    },
    {
      "epoch": 0.05349127534674386,
      "grad_norm": 1248.0147705078125,
      "learning_rate": 2.1184e-05,
      "loss": 152.9556,
      "step": 13240
    },
    {
      "epoch": 0.053531676612111494,
      "grad_norm": 964.110595703125,
      "learning_rate": 2.1200000000000004e-05,
      "loss": 188.2181,
      "step": 13250
    },
    {
      "epoch": 0.05357207787747912,
      "grad_norm": 796.8040161132812,
      "learning_rate": 2.1216e-05,
      "loss": 113.264,
      "step": 13260
    },
    {
      "epoch": 0.05361247914284675,
      "grad_norm": 1286.59765625,
      "learning_rate": 2.1232000000000003e-05,
      "loss": 122.0433,
      "step": 13270
    },
    {
      "epoch": 0.053652880408214386,
      "grad_norm": 1039.0003662109375,
      "learning_rate": 2.1248e-05,
      "loss": 166.3482,
      "step": 13280
    },
    {
      "epoch": 0.05369328167358202,
      "grad_norm": 1605.037841796875,
      "learning_rate": 2.1264000000000002e-05,
      "loss": 119.0407,
      "step": 13290
    },
    {
      "epoch": 0.053733682938949645,
      "grad_norm": 1566.1156005859375,
      "learning_rate": 2.1280000000000003e-05,
      "loss": 159.9284,
      "step": 13300
    },
    {
      "epoch": 0.05377408420431728,
      "grad_norm": 905.0177612304688,
      "learning_rate": 2.1296e-05,
      "loss": 98.9591,
      "step": 13310
    },
    {
      "epoch": 0.05381448546968491,
      "grad_norm": 1059.0621337890625,
      "learning_rate": 2.1312000000000005e-05,
      "loss": 124.6446,
      "step": 13320
    },
    {
      "epoch": 0.053854886735052544,
      "grad_norm": 717.8817138671875,
      "learning_rate": 2.1328000000000003e-05,
      "loss": 219.626,
      "step": 13330
    },
    {
      "epoch": 0.05389528800042017,
      "grad_norm": 1320.574951171875,
      "learning_rate": 2.1344e-05,
      "loss": 176.7211,
      "step": 13340
    },
    {
      "epoch": 0.0539356892657878,
      "grad_norm": 1039.6083984375,
      "learning_rate": 2.1360000000000002e-05,
      "loss": 130.3765,
      "step": 13350
    },
    {
      "epoch": 0.053976090531155436,
      "grad_norm": 1116.202880859375,
      "learning_rate": 2.1376e-05,
      "loss": 197.231,
      "step": 13360
    },
    {
      "epoch": 0.05401649179652307,
      "grad_norm": 2312.27587890625,
      "learning_rate": 2.1392000000000005e-05,
      "loss": 192.5465,
      "step": 13370
    },
    {
      "epoch": 0.054056893061890696,
      "grad_norm": 3889.453857421875,
      "learning_rate": 2.1408000000000002e-05,
      "loss": 123.0288,
      "step": 13380
    },
    {
      "epoch": 0.05409729432725833,
      "grad_norm": 1394.5799560546875,
      "learning_rate": 2.1424e-05,
      "loss": 139.7589,
      "step": 13390
    },
    {
      "epoch": 0.05413769559262596,
      "grad_norm": 686.0022583007812,
      "learning_rate": 2.144e-05,
      "loss": 101.6074,
      "step": 13400
    },
    {
      "epoch": 0.054178096857993595,
      "grad_norm": 3230.813720703125,
      "learning_rate": 2.1456000000000003e-05,
      "loss": 147.0386,
      "step": 13410
    },
    {
      "epoch": 0.05421849812336122,
      "grad_norm": 1445.752197265625,
      "learning_rate": 2.1472000000000004e-05,
      "loss": 178.1156,
      "step": 13420
    },
    {
      "epoch": 0.054258899388728854,
      "grad_norm": 1217.630126953125,
      "learning_rate": 2.1488e-05,
      "loss": 107.1757,
      "step": 13430
    },
    {
      "epoch": 0.05429930065409649,
      "grad_norm": 910.73583984375,
      "learning_rate": 2.1504e-05,
      "loss": 150.5691,
      "step": 13440
    },
    {
      "epoch": 0.05433970191946412,
      "grad_norm": 1758.7811279296875,
      "learning_rate": 2.1520000000000004e-05,
      "loss": 169.052,
      "step": 13450
    },
    {
      "epoch": 0.054380103184831746,
      "grad_norm": 1926.9754638671875,
      "learning_rate": 2.1536000000000002e-05,
      "loss": 121.4231,
      "step": 13460
    },
    {
      "epoch": 0.05442050445019938,
      "grad_norm": 2126.060791015625,
      "learning_rate": 2.1552e-05,
      "loss": 166.1807,
      "step": 13470
    },
    {
      "epoch": 0.05446090571556701,
      "grad_norm": 779.2872924804688,
      "learning_rate": 2.1568e-05,
      "loss": 148.9455,
      "step": 13480
    },
    {
      "epoch": 0.054501306980934645,
      "grad_norm": 971.3914794921875,
      "learning_rate": 2.1584000000000002e-05,
      "loss": 173.3893,
      "step": 13490
    },
    {
      "epoch": 0.05454170824630227,
      "grad_norm": 1498.48486328125,
      "learning_rate": 2.1600000000000003e-05,
      "loss": 200.5015,
      "step": 13500
    },
    {
      "epoch": 0.054582109511669905,
      "grad_norm": 1800.676025390625,
      "learning_rate": 2.1616e-05,
      "loss": 144.656,
      "step": 13510
    },
    {
      "epoch": 0.05462251077703754,
      "grad_norm": 952.250244140625,
      "learning_rate": 2.1632e-05,
      "loss": 117.9983,
      "step": 13520
    },
    {
      "epoch": 0.05466291204240517,
      "grad_norm": 1999.3441162109375,
      "learning_rate": 2.1648000000000003e-05,
      "loss": 106.9742,
      "step": 13530
    },
    {
      "epoch": 0.0547033133077728,
      "grad_norm": 1853.0484619140625,
      "learning_rate": 2.1664e-05,
      "loss": 170.8819,
      "step": 13540
    },
    {
      "epoch": 0.05474371457314043,
      "grad_norm": 1101.6063232421875,
      "learning_rate": 2.1680000000000002e-05,
      "loss": 184.5558,
      "step": 13550
    },
    {
      "epoch": 0.05478411583850806,
      "grad_norm": 1025.219970703125,
      "learning_rate": 2.1696e-05,
      "loss": 141.2452,
      "step": 13560
    },
    {
      "epoch": 0.054824517103875696,
      "grad_norm": 1053.7066650390625,
      "learning_rate": 2.1711999999999998e-05,
      "loss": 158.2792,
      "step": 13570
    },
    {
      "epoch": 0.05486491836924332,
      "grad_norm": 0.0,
      "learning_rate": 2.1728000000000003e-05,
      "loss": 83.3861,
      "step": 13580
    },
    {
      "epoch": 0.054905319634610955,
      "grad_norm": 744.6858520507812,
      "learning_rate": 2.1744e-05,
      "loss": 165.1833,
      "step": 13590
    },
    {
      "epoch": 0.05494572089997859,
      "grad_norm": 2354.499267578125,
      "learning_rate": 2.1760000000000002e-05,
      "loss": 121.2317,
      "step": 13600
    },
    {
      "epoch": 0.05498612216534622,
      "grad_norm": 1683.3133544921875,
      "learning_rate": 2.1776000000000003e-05,
      "loss": 231.1734,
      "step": 13610
    },
    {
      "epoch": 0.05502652343071385,
      "grad_norm": 792.2965698242188,
      "learning_rate": 2.1792e-05,
      "loss": 115.8473,
      "step": 13620
    },
    {
      "epoch": 0.05506692469608148,
      "grad_norm": 1531.6292724609375,
      "learning_rate": 2.1808000000000002e-05,
      "loss": 161.9549,
      "step": 13630
    },
    {
      "epoch": 0.05510732596144911,
      "grad_norm": 1182.44091796875,
      "learning_rate": 2.1824e-05,
      "loss": 119.7171,
      "step": 13640
    },
    {
      "epoch": 0.055147727226816746,
      "grad_norm": 1206.6705322265625,
      "learning_rate": 2.1840000000000004e-05,
      "loss": 184.1771,
      "step": 13650
    },
    {
      "epoch": 0.05518812849218437,
      "grad_norm": 629.7221069335938,
      "learning_rate": 2.1856000000000002e-05,
      "loss": 158.6814,
      "step": 13660
    },
    {
      "epoch": 0.055228529757552006,
      "grad_norm": 769.7032470703125,
      "learning_rate": 2.1872e-05,
      "loss": 133.69,
      "step": 13670
    },
    {
      "epoch": 0.05526893102291964,
      "grad_norm": 910.8077392578125,
      "learning_rate": 2.1888e-05,
      "loss": 104.3274,
      "step": 13680
    },
    {
      "epoch": 0.05530933228828727,
      "grad_norm": 1321.2825927734375,
      "learning_rate": 2.1904000000000002e-05,
      "loss": 150.9695,
      "step": 13690
    },
    {
      "epoch": 0.0553497335536549,
      "grad_norm": 692.1464233398438,
      "learning_rate": 2.1920000000000004e-05,
      "loss": 154.0631,
      "step": 13700
    },
    {
      "epoch": 0.05539013481902253,
      "grad_norm": 1960.6553955078125,
      "learning_rate": 2.1936e-05,
      "loss": 172.6004,
      "step": 13710
    },
    {
      "epoch": 0.055430536084390164,
      "grad_norm": 937.15625,
      "learning_rate": 2.1952e-05,
      "loss": 116.4846,
      "step": 13720
    },
    {
      "epoch": 0.0554709373497578,
      "grad_norm": 1401.888671875,
      "learning_rate": 2.1968000000000004e-05,
      "loss": 143.6478,
      "step": 13730
    },
    {
      "epoch": 0.05551133861512542,
      "grad_norm": 1163.2547607421875,
      "learning_rate": 2.1984e-05,
      "loss": 158.8097,
      "step": 13740
    },
    {
      "epoch": 0.055551739880493056,
      "grad_norm": 878.194091796875,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 136.7762,
      "step": 13750
    },
    {
      "epoch": 0.05559214114586069,
      "grad_norm": 2160.6083984375,
      "learning_rate": 2.2016e-05,
      "loss": 133.5667,
      "step": 13760
    },
    {
      "epoch": 0.05563254241122832,
      "grad_norm": 1110.6484375,
      "learning_rate": 2.2032e-05,
      "loss": 144.071,
      "step": 13770
    },
    {
      "epoch": 0.05567294367659595,
      "grad_norm": 759.3908081054688,
      "learning_rate": 2.2048000000000003e-05,
      "loss": 124.2756,
      "step": 13780
    },
    {
      "epoch": 0.05571334494196358,
      "grad_norm": 689.4682006835938,
      "learning_rate": 2.2064e-05,
      "loss": 132.506,
      "step": 13790
    },
    {
      "epoch": 0.055753746207331215,
      "grad_norm": 1468.38232421875,
      "learning_rate": 2.2080000000000002e-05,
      "loss": 172.1625,
      "step": 13800
    },
    {
      "epoch": 0.05579414747269885,
      "grad_norm": 1024.220947265625,
      "learning_rate": 2.2096000000000003e-05,
      "loss": 110.2222,
      "step": 13810
    },
    {
      "epoch": 0.055834548738066474,
      "grad_norm": 1557.86669921875,
      "learning_rate": 2.2112e-05,
      "loss": 154.4878,
      "step": 13820
    },
    {
      "epoch": 0.05587495000343411,
      "grad_norm": 410.6898498535156,
      "learning_rate": 2.2128000000000002e-05,
      "loss": 149.6926,
      "step": 13830
    },
    {
      "epoch": 0.05591535126880174,
      "grad_norm": 1283.35791015625,
      "learning_rate": 2.2144e-05,
      "loss": 121.5034,
      "step": 13840
    },
    {
      "epoch": 0.05595575253416937,
      "grad_norm": 908.16748046875,
      "learning_rate": 2.2160000000000005e-05,
      "loss": 156.9578,
      "step": 13850
    },
    {
      "epoch": 0.055996153799537,
      "grad_norm": 1568.7562255859375,
      "learning_rate": 2.2176000000000002e-05,
      "loss": 130.5618,
      "step": 13860
    },
    {
      "epoch": 0.05603655506490463,
      "grad_norm": 949.5377807617188,
      "learning_rate": 2.2192e-05,
      "loss": 185.1214,
      "step": 13870
    },
    {
      "epoch": 0.056076956330272265,
      "grad_norm": 1054.2724609375,
      "learning_rate": 2.2208e-05,
      "loss": 139.6275,
      "step": 13880
    },
    {
      "epoch": 0.0561173575956399,
      "grad_norm": 816.1144409179688,
      "learning_rate": 2.2224000000000003e-05,
      "loss": 89.5711,
      "step": 13890
    },
    {
      "epoch": 0.056157758861007524,
      "grad_norm": 913.6422729492188,
      "learning_rate": 2.2240000000000004e-05,
      "loss": 107.2711,
      "step": 13900
    },
    {
      "epoch": 0.05619816012637516,
      "grad_norm": 684.166015625,
      "learning_rate": 2.2256000000000002e-05,
      "loss": 131.4961,
      "step": 13910
    },
    {
      "epoch": 0.05623856139174279,
      "grad_norm": 3857.287841796875,
      "learning_rate": 2.2272e-05,
      "loss": 179.3707,
      "step": 13920
    },
    {
      "epoch": 0.05627896265711042,
      "grad_norm": 1287.8927001953125,
      "learning_rate": 2.2288000000000004e-05,
      "loss": 126.2159,
      "step": 13930
    },
    {
      "epoch": 0.05631936392247805,
      "grad_norm": 753.49853515625,
      "learning_rate": 2.2304000000000002e-05,
      "loss": 127.1484,
      "step": 13940
    },
    {
      "epoch": 0.05635976518784568,
      "grad_norm": 1118.16748046875,
      "learning_rate": 2.2320000000000003e-05,
      "loss": 165.6699,
      "step": 13950
    },
    {
      "epoch": 0.056400166453213316,
      "grad_norm": 1042.443115234375,
      "learning_rate": 2.2336e-05,
      "loss": 142.2855,
      "step": 13960
    },
    {
      "epoch": 0.05644056771858095,
      "grad_norm": 1004.0928955078125,
      "learning_rate": 2.2352e-05,
      "loss": 127.3151,
      "step": 13970
    },
    {
      "epoch": 0.056480968983948575,
      "grad_norm": 1794.01171875,
      "learning_rate": 2.2368000000000003e-05,
      "loss": 136.5507,
      "step": 13980
    },
    {
      "epoch": 0.05652137024931621,
      "grad_norm": 487.4826354980469,
      "learning_rate": 2.2384e-05,
      "loss": 164.9086,
      "step": 13990
    },
    {
      "epoch": 0.05656177151468384,
      "grad_norm": 751.5004272460938,
      "learning_rate": 2.2400000000000002e-05,
      "loss": 114.4318,
      "step": 14000
    },
    {
      "epoch": 0.056602172780051474,
      "grad_norm": 1770.594482421875,
      "learning_rate": 2.2416000000000004e-05,
      "loss": 202.0991,
      "step": 14010
    },
    {
      "epoch": 0.0566425740454191,
      "grad_norm": 1120.473388671875,
      "learning_rate": 2.2432e-05,
      "loss": 152.7142,
      "step": 14020
    },
    {
      "epoch": 0.05668297531078673,
      "grad_norm": 520.49365234375,
      "learning_rate": 2.2448000000000003e-05,
      "loss": 137.2922,
      "step": 14030
    },
    {
      "epoch": 0.056723376576154366,
      "grad_norm": 1072.8531494140625,
      "learning_rate": 2.2464e-05,
      "loss": 147.3355,
      "step": 14040
    },
    {
      "epoch": 0.056763777841522,
      "grad_norm": 1459.0318603515625,
      "learning_rate": 2.2480000000000005e-05,
      "loss": 144.517,
      "step": 14050
    },
    {
      "epoch": 0.056804179106889625,
      "grad_norm": 1281.28271484375,
      "learning_rate": 2.2496000000000003e-05,
      "loss": 153.5128,
      "step": 14060
    },
    {
      "epoch": 0.05684458037225726,
      "grad_norm": 2782.97900390625,
      "learning_rate": 2.2512e-05,
      "loss": 180.4696,
      "step": 14070
    },
    {
      "epoch": 0.05688498163762489,
      "grad_norm": 736.7996826171875,
      "learning_rate": 2.2528000000000002e-05,
      "loss": 123.2284,
      "step": 14080
    },
    {
      "epoch": 0.056925382902992525,
      "grad_norm": 3690.111083984375,
      "learning_rate": 2.2544000000000003e-05,
      "loss": 116.4326,
      "step": 14090
    },
    {
      "epoch": 0.05696578416836015,
      "grad_norm": 4582.87841796875,
      "learning_rate": 2.256e-05,
      "loss": 205.8677,
      "step": 14100
    },
    {
      "epoch": 0.057006185433727784,
      "grad_norm": 2828.5712890625,
      "learning_rate": 2.2576000000000002e-05,
      "loss": 121.0329,
      "step": 14110
    },
    {
      "epoch": 0.05704658669909542,
      "grad_norm": 910.3086547851562,
      "learning_rate": 2.2592e-05,
      "loss": 127.2261,
      "step": 14120
    },
    {
      "epoch": 0.05708698796446305,
      "grad_norm": 1229.384033203125,
      "learning_rate": 2.2608000000000004e-05,
      "loss": 150.3236,
      "step": 14130
    },
    {
      "epoch": 0.057127389229830676,
      "grad_norm": 2388.891845703125,
      "learning_rate": 2.2624000000000002e-05,
      "loss": 202.8478,
      "step": 14140
    },
    {
      "epoch": 0.05716779049519831,
      "grad_norm": 820.1349487304688,
      "learning_rate": 2.264e-05,
      "loss": 139.7399,
      "step": 14150
    },
    {
      "epoch": 0.05720819176056594,
      "grad_norm": 1131.6744384765625,
      "learning_rate": 2.2656e-05,
      "loss": 119.5409,
      "step": 14160
    },
    {
      "epoch": 0.057248593025933575,
      "grad_norm": 1692.20947265625,
      "learning_rate": 2.2672e-05,
      "loss": 136.8943,
      "step": 14170
    },
    {
      "epoch": 0.0572889942913012,
      "grad_norm": 841.63330078125,
      "learning_rate": 2.2688000000000004e-05,
      "loss": 112.7383,
      "step": 14180
    },
    {
      "epoch": 0.057329395556668834,
      "grad_norm": 993.0922241210938,
      "learning_rate": 2.2704e-05,
      "loss": 139.6173,
      "step": 14190
    },
    {
      "epoch": 0.05736979682203647,
      "grad_norm": 848.125,
      "learning_rate": 2.272e-05,
      "loss": 117.9697,
      "step": 14200
    },
    {
      "epoch": 0.0574101980874041,
      "grad_norm": 1794.74755859375,
      "learning_rate": 2.2736000000000004e-05,
      "loss": 120.4493,
      "step": 14210
    },
    {
      "epoch": 0.057450599352771727,
      "grad_norm": 700.1529541015625,
      "learning_rate": 2.2752000000000002e-05,
      "loss": 111.5371,
      "step": 14220
    },
    {
      "epoch": 0.05749100061813936,
      "grad_norm": 2567.41845703125,
      "learning_rate": 2.2768000000000003e-05,
      "loss": 151.0815,
      "step": 14230
    },
    {
      "epoch": 0.05753140188350699,
      "grad_norm": 1424.2322998046875,
      "learning_rate": 2.2784e-05,
      "loss": 173.4974,
      "step": 14240
    },
    {
      "epoch": 0.057571803148874626,
      "grad_norm": 1377.6383056640625,
      "learning_rate": 2.28e-05,
      "loss": 135.3481,
      "step": 14250
    },
    {
      "epoch": 0.05761220441424225,
      "grad_norm": 1382.552734375,
      "learning_rate": 2.2816000000000003e-05,
      "loss": 148.4613,
      "step": 14260
    },
    {
      "epoch": 0.057652605679609885,
      "grad_norm": 478.0409240722656,
      "learning_rate": 2.2832e-05,
      "loss": 115.8696,
      "step": 14270
    },
    {
      "epoch": 0.05769300694497752,
      "grad_norm": 588.9566040039062,
      "learning_rate": 2.2848000000000002e-05,
      "loss": 149.4777,
      "step": 14280
    },
    {
      "epoch": 0.05773340821034515,
      "grad_norm": 620.9132690429688,
      "learning_rate": 2.2864000000000003e-05,
      "loss": 196.3389,
      "step": 14290
    },
    {
      "epoch": 0.05777380947571278,
      "grad_norm": 3938.6494140625,
      "learning_rate": 2.288e-05,
      "loss": 172.2783,
      "step": 14300
    },
    {
      "epoch": 0.05781421074108041,
      "grad_norm": 3169.12548828125,
      "learning_rate": 2.2896000000000002e-05,
      "loss": 128.8295,
      "step": 14310
    },
    {
      "epoch": 0.05785461200644804,
      "grad_norm": 1205.0302734375,
      "learning_rate": 2.2912e-05,
      "loss": 149.3006,
      "step": 14320
    },
    {
      "epoch": 0.057895013271815676,
      "grad_norm": 880.8743286132812,
      "learning_rate": 2.2928000000000005e-05,
      "loss": 100.6335,
      "step": 14330
    },
    {
      "epoch": 0.0579354145371833,
      "grad_norm": 411.00128173828125,
      "learning_rate": 2.2944000000000003e-05,
      "loss": 167.0317,
      "step": 14340
    },
    {
      "epoch": 0.057975815802550935,
      "grad_norm": 438.6155700683594,
      "learning_rate": 2.296e-05,
      "loss": 170.8691,
      "step": 14350
    },
    {
      "epoch": 0.05801621706791857,
      "grad_norm": 2452.795654296875,
      "learning_rate": 2.2976e-05,
      "loss": 159.6383,
      "step": 14360
    },
    {
      "epoch": 0.0580566183332862,
      "grad_norm": 3605.11181640625,
      "learning_rate": 2.2992e-05,
      "loss": 128.4706,
      "step": 14370
    },
    {
      "epoch": 0.05809701959865383,
      "grad_norm": 1165.4371337890625,
      "learning_rate": 2.3008000000000004e-05,
      "loss": 163.2181,
      "step": 14380
    },
    {
      "epoch": 0.05813742086402146,
      "grad_norm": 3038.6328125,
      "learning_rate": 2.3024000000000002e-05,
      "loss": 176.579,
      "step": 14390
    },
    {
      "epoch": 0.058177822129389094,
      "grad_norm": 810.4943237304688,
      "learning_rate": 2.304e-05,
      "loss": 105.8301,
      "step": 14400
    },
    {
      "epoch": 0.05821822339475673,
      "grad_norm": 2276.8701171875,
      "learning_rate": 2.3056000000000004e-05,
      "loss": 169.4745,
      "step": 14410
    },
    {
      "epoch": 0.05825862466012435,
      "grad_norm": 706.0416259765625,
      "learning_rate": 2.3072000000000002e-05,
      "loss": 132.1325,
      "step": 14420
    },
    {
      "epoch": 0.058299025925491986,
      "grad_norm": 723.7369995117188,
      "learning_rate": 2.3088000000000003e-05,
      "loss": 106.6888,
      "step": 14430
    },
    {
      "epoch": 0.05833942719085962,
      "grad_norm": 609.4017944335938,
      "learning_rate": 2.3104e-05,
      "loss": 125.4262,
      "step": 14440
    },
    {
      "epoch": 0.05837982845622725,
      "grad_norm": 1099.2354736328125,
      "learning_rate": 2.312e-05,
      "loss": 143.0599,
      "step": 14450
    },
    {
      "epoch": 0.05842022972159488,
      "grad_norm": 978.5392456054688,
      "learning_rate": 2.3136000000000003e-05,
      "loss": 135.362,
      "step": 14460
    },
    {
      "epoch": 0.05846063098696251,
      "grad_norm": 893.01806640625,
      "learning_rate": 2.3152e-05,
      "loss": 108.5644,
      "step": 14470
    },
    {
      "epoch": 0.058501032252330144,
      "grad_norm": 783.920166015625,
      "learning_rate": 2.3168000000000002e-05,
      "loss": 202.1595,
      "step": 14480
    },
    {
      "epoch": 0.05854143351769778,
      "grad_norm": 1331.897705078125,
      "learning_rate": 2.3184000000000004e-05,
      "loss": 144.4532,
      "step": 14490
    },
    {
      "epoch": 0.058581834783065403,
      "grad_norm": 1102.8529052734375,
      "learning_rate": 2.32e-05,
      "loss": 122.3522,
      "step": 14500
    },
    {
      "epoch": 0.05862223604843304,
      "grad_norm": 743.5985717773438,
      "learning_rate": 2.3216000000000003e-05,
      "loss": 99.6895,
      "step": 14510
    },
    {
      "epoch": 0.05866263731380067,
      "grad_norm": 1235.1953125,
      "learning_rate": 2.3232e-05,
      "loss": 179.0568,
      "step": 14520
    },
    {
      "epoch": 0.0587030385791683,
      "grad_norm": 698.5330810546875,
      "learning_rate": 2.3248000000000005e-05,
      "loss": 152.631,
      "step": 14530
    },
    {
      "epoch": 0.05874343984453593,
      "grad_norm": 826.0693359375,
      "learning_rate": 2.3264000000000003e-05,
      "loss": 102.5496,
      "step": 14540
    },
    {
      "epoch": 0.05878384110990356,
      "grad_norm": 931.8894653320312,
      "learning_rate": 2.328e-05,
      "loss": 104.5592,
      "step": 14550
    },
    {
      "epoch": 0.058824242375271195,
      "grad_norm": 1267.3544921875,
      "learning_rate": 2.3296000000000002e-05,
      "loss": 145.6881,
      "step": 14560
    },
    {
      "epoch": 0.05886464364063883,
      "grad_norm": 710.9752197265625,
      "learning_rate": 2.3312e-05,
      "loss": 102.4729,
      "step": 14570
    },
    {
      "epoch": 0.058905044906006454,
      "grad_norm": 1418.5509033203125,
      "learning_rate": 2.3328000000000004e-05,
      "loss": 123.5127,
      "step": 14580
    },
    {
      "epoch": 0.05894544617137409,
      "grad_norm": 2174.474853515625,
      "learning_rate": 2.3344000000000002e-05,
      "loss": 120.0803,
      "step": 14590
    },
    {
      "epoch": 0.05898584743674172,
      "grad_norm": 2546.129150390625,
      "learning_rate": 2.336e-05,
      "loss": 174.6094,
      "step": 14600
    },
    {
      "epoch": 0.05902624870210935,
      "grad_norm": 2337.811767578125,
      "learning_rate": 2.3376000000000005e-05,
      "loss": 121.787,
      "step": 14610
    },
    {
      "epoch": 0.05906664996747698,
      "grad_norm": 607.4607543945312,
      "learning_rate": 2.3392000000000002e-05,
      "loss": 145.2568,
      "step": 14620
    },
    {
      "epoch": 0.05910705123284461,
      "grad_norm": 1109.9788818359375,
      "learning_rate": 2.3408000000000004e-05,
      "loss": 92.7968,
      "step": 14630
    },
    {
      "epoch": 0.059147452498212245,
      "grad_norm": 996.027587890625,
      "learning_rate": 2.3424e-05,
      "loss": 146.8611,
      "step": 14640
    },
    {
      "epoch": 0.05918785376357988,
      "grad_norm": 1124.2379150390625,
      "learning_rate": 2.344e-05,
      "loss": 106.3015,
      "step": 14650
    },
    {
      "epoch": 0.059228255028947505,
      "grad_norm": 674.2631225585938,
      "learning_rate": 2.3456000000000004e-05,
      "loss": 150.5028,
      "step": 14660
    },
    {
      "epoch": 0.05926865629431514,
      "grad_norm": 1879.044677734375,
      "learning_rate": 2.3472e-05,
      "loss": 128.7712,
      "step": 14670
    },
    {
      "epoch": 0.05930905755968277,
      "grad_norm": 1457.988037109375,
      "learning_rate": 2.3488000000000003e-05,
      "loss": 147.8928,
      "step": 14680
    },
    {
      "epoch": 0.059349458825050404,
      "grad_norm": 1219.433349609375,
      "learning_rate": 2.3504000000000004e-05,
      "loss": 131.0105,
      "step": 14690
    },
    {
      "epoch": 0.05938986009041803,
      "grad_norm": 467.0386047363281,
      "learning_rate": 2.3520000000000002e-05,
      "loss": 151.1394,
      "step": 14700
    },
    {
      "epoch": 0.05943026135578566,
      "grad_norm": 2050.65234375,
      "learning_rate": 2.3536000000000003e-05,
      "loss": 129.0073,
      "step": 14710
    },
    {
      "epoch": 0.059470662621153296,
      "grad_norm": 1917.9736328125,
      "learning_rate": 2.3552e-05,
      "loss": 165.6747,
      "step": 14720
    },
    {
      "epoch": 0.05951106388652093,
      "grad_norm": 14780.8095703125,
      "learning_rate": 2.3568e-05,
      "loss": 180.8651,
      "step": 14730
    },
    {
      "epoch": 0.059551465151888555,
      "grad_norm": 1709.83349609375,
      "learning_rate": 2.3584000000000003e-05,
      "loss": 131.5537,
      "step": 14740
    },
    {
      "epoch": 0.05959186641725619,
      "grad_norm": 476.5604248046875,
      "learning_rate": 2.36e-05,
      "loss": 115.247,
      "step": 14750
    },
    {
      "epoch": 0.05963226768262382,
      "grad_norm": 1123.6815185546875,
      "learning_rate": 2.3616000000000002e-05,
      "loss": 145.8521,
      "step": 14760
    },
    {
      "epoch": 0.059672668947991454,
      "grad_norm": 802.3085327148438,
      "learning_rate": 2.3632e-05,
      "loss": 137.058,
      "step": 14770
    },
    {
      "epoch": 0.05971307021335908,
      "grad_norm": 571.6912231445312,
      "learning_rate": 2.3648e-05,
      "loss": 108.2743,
      "step": 14780
    },
    {
      "epoch": 0.059753471478726713,
      "grad_norm": 1714.322265625,
      "learning_rate": 2.3664000000000002e-05,
      "loss": 150.8154,
      "step": 14790
    },
    {
      "epoch": 0.05979387274409435,
      "grad_norm": 1163.4627685546875,
      "learning_rate": 2.368e-05,
      "loss": 155.5478,
      "step": 14800
    },
    {
      "epoch": 0.05983427400946198,
      "grad_norm": 668.583740234375,
      "learning_rate": 2.3696000000000005e-05,
      "loss": 143.8596,
      "step": 14810
    },
    {
      "epoch": 0.059874675274829606,
      "grad_norm": 598.8736572265625,
      "learning_rate": 2.3712000000000003e-05,
      "loss": 132.4404,
      "step": 14820
    },
    {
      "epoch": 0.05991507654019724,
      "grad_norm": 1189.759033203125,
      "learning_rate": 2.3728e-05,
      "loss": 109.6164,
      "step": 14830
    },
    {
      "epoch": 0.05995547780556487,
      "grad_norm": 1057.4130859375,
      "learning_rate": 2.3744000000000002e-05,
      "loss": 129.6899,
      "step": 14840
    },
    {
      "epoch": 0.059995879070932505,
      "grad_norm": 463.6476135253906,
      "learning_rate": 2.376e-05,
      "loss": 154.4845,
      "step": 14850
    },
    {
      "epoch": 0.06003628033630013,
      "grad_norm": 2989.97265625,
      "learning_rate": 2.3776000000000004e-05,
      "loss": 113.0495,
      "step": 14860
    },
    {
      "epoch": 0.060076681601667764,
      "grad_norm": 1022.0186157226562,
      "learning_rate": 2.3792000000000002e-05,
      "loss": 114.5988,
      "step": 14870
    },
    {
      "epoch": 0.0601170828670354,
      "grad_norm": 800.594970703125,
      "learning_rate": 2.3808e-05,
      "loss": 120.2548,
      "step": 14880
    },
    {
      "epoch": 0.06015748413240303,
      "grad_norm": 893.33447265625,
      "learning_rate": 2.3824000000000004e-05,
      "loss": 138.4012,
      "step": 14890
    },
    {
      "epoch": 0.060197885397770656,
      "grad_norm": 766.477783203125,
      "learning_rate": 2.3840000000000002e-05,
      "loss": 152.4039,
      "step": 14900
    },
    {
      "epoch": 0.06023828666313829,
      "grad_norm": 4025.67822265625,
      "learning_rate": 2.3856000000000003e-05,
      "loss": 144.6702,
      "step": 14910
    },
    {
      "epoch": 0.06027868792850592,
      "grad_norm": 1157.56884765625,
      "learning_rate": 2.3872e-05,
      "loss": 182.9655,
      "step": 14920
    },
    {
      "epoch": 0.060319089193873555,
      "grad_norm": 2603.474853515625,
      "learning_rate": 2.3888e-05,
      "loss": 172.9417,
      "step": 14930
    },
    {
      "epoch": 0.06035949045924118,
      "grad_norm": 1721.3291015625,
      "learning_rate": 2.3904000000000004e-05,
      "loss": 148.0654,
      "step": 14940
    },
    {
      "epoch": 0.060399891724608815,
      "grad_norm": 1430.939208984375,
      "learning_rate": 2.392e-05,
      "loss": 132.7668,
      "step": 14950
    },
    {
      "epoch": 0.06044029298997645,
      "grad_norm": 853.9814453125,
      "learning_rate": 2.3936000000000003e-05,
      "loss": 67.8114,
      "step": 14960
    },
    {
      "epoch": 0.06048069425534408,
      "grad_norm": 3689.06005859375,
      "learning_rate": 2.3952e-05,
      "loss": 192.2166,
      "step": 14970
    },
    {
      "epoch": 0.06052109552071171,
      "grad_norm": 481.86920166015625,
      "learning_rate": 2.3968e-05,
      "loss": 117.9959,
      "step": 14980
    },
    {
      "epoch": 0.06056149678607934,
      "grad_norm": 683.6334228515625,
      "learning_rate": 2.3984000000000003e-05,
      "loss": 100.6381,
      "step": 14990
    },
    {
      "epoch": 0.06060189805144697,
      "grad_norm": 1204.35693359375,
      "learning_rate": 2.4e-05,
      "loss": 128.2924,
      "step": 15000
    },
    {
      "epoch": 0.060642299316814606,
      "grad_norm": 1228.6927490234375,
      "learning_rate": 2.4016000000000005e-05,
      "loss": 125.4803,
      "step": 15010
    },
    {
      "epoch": 0.06068270058218223,
      "grad_norm": 1097.30126953125,
      "learning_rate": 2.4032000000000003e-05,
      "loss": 122.2645,
      "step": 15020
    },
    {
      "epoch": 0.060723101847549865,
      "grad_norm": 680.966796875,
      "learning_rate": 2.4048e-05,
      "loss": 135.325,
      "step": 15030
    },
    {
      "epoch": 0.0607635031129175,
      "grad_norm": 1410.129638671875,
      "learning_rate": 2.4064000000000002e-05,
      "loss": 115.6653,
      "step": 15040
    },
    {
      "epoch": 0.06080390437828513,
      "grad_norm": 830.8237915039062,
      "learning_rate": 2.408e-05,
      "loss": 146.791,
      "step": 15050
    },
    {
      "epoch": 0.06084430564365276,
      "grad_norm": 1142.181640625,
      "learning_rate": 2.4096000000000004e-05,
      "loss": 113.4309,
      "step": 15060
    },
    {
      "epoch": 0.06088470690902039,
      "grad_norm": 845.2885131835938,
      "learning_rate": 2.4112000000000002e-05,
      "loss": 142.3753,
      "step": 15070
    },
    {
      "epoch": 0.060925108174388024,
      "grad_norm": 1506.6373291015625,
      "learning_rate": 2.4128e-05,
      "loss": 177.1799,
      "step": 15080
    },
    {
      "epoch": 0.06096550943975566,
      "grad_norm": 1998.8515625,
      "learning_rate": 2.4144000000000005e-05,
      "loss": 147.3931,
      "step": 15090
    },
    {
      "epoch": 0.06100591070512328,
      "grad_norm": 1360.8604736328125,
      "learning_rate": 2.4160000000000002e-05,
      "loss": 156.8285,
      "step": 15100
    },
    {
      "epoch": 0.061046311970490916,
      "grad_norm": 1141.12109375,
      "learning_rate": 2.4176000000000004e-05,
      "loss": 124.2817,
      "step": 15110
    },
    {
      "epoch": 0.06108671323585855,
      "grad_norm": 4913.79443359375,
      "learning_rate": 2.4192e-05,
      "loss": 154.2291,
      "step": 15120
    },
    {
      "epoch": 0.06112711450122618,
      "grad_norm": 1053.5574951171875,
      "learning_rate": 2.4208e-05,
      "loss": 83.8442,
      "step": 15130
    },
    {
      "epoch": 0.06116751576659381,
      "grad_norm": 1806.4326171875,
      "learning_rate": 2.4224000000000004e-05,
      "loss": 154.9223,
      "step": 15140
    },
    {
      "epoch": 0.06120791703196144,
      "grad_norm": 655.29443359375,
      "learning_rate": 2.4240000000000002e-05,
      "loss": 135.3248,
      "step": 15150
    },
    {
      "epoch": 0.061248318297329074,
      "grad_norm": 1480.966064453125,
      "learning_rate": 2.4256000000000003e-05,
      "loss": 172.8818,
      "step": 15160
    },
    {
      "epoch": 0.06128871956269671,
      "grad_norm": 10371.9072265625,
      "learning_rate": 2.4272e-05,
      "loss": 150.4807,
      "step": 15170
    },
    {
      "epoch": 0.06132912082806433,
      "grad_norm": 3907.77001953125,
      "learning_rate": 2.4288000000000002e-05,
      "loss": 197.0451,
      "step": 15180
    },
    {
      "epoch": 0.061369522093431966,
      "grad_norm": 1567.521484375,
      "learning_rate": 2.4304000000000003e-05,
      "loss": 132.6108,
      "step": 15190
    },
    {
      "epoch": 0.0614099233587996,
      "grad_norm": 877.543212890625,
      "learning_rate": 2.432e-05,
      "loss": 152.2519,
      "step": 15200
    },
    {
      "epoch": 0.061450324624167225,
      "grad_norm": 1100.1990966796875,
      "learning_rate": 2.4336000000000006e-05,
      "loss": 152.2165,
      "step": 15210
    },
    {
      "epoch": 0.06149072588953486,
      "grad_norm": 1301.2376708984375,
      "learning_rate": 2.4352000000000003e-05,
      "loss": 172.2335,
      "step": 15220
    },
    {
      "epoch": 0.06153112715490249,
      "grad_norm": 776.1826171875,
      "learning_rate": 2.4368e-05,
      "loss": 109.9598,
      "step": 15230
    },
    {
      "epoch": 0.061571528420270125,
      "grad_norm": 1602.8883056640625,
      "learning_rate": 2.4384000000000002e-05,
      "loss": 95.1105,
      "step": 15240
    },
    {
      "epoch": 0.06161192968563775,
      "grad_norm": 929.6801147460938,
      "learning_rate": 2.44e-05,
      "loss": 165.9625,
      "step": 15250
    },
    {
      "epoch": 0.061652330951005384,
      "grad_norm": 332.2320251464844,
      "learning_rate": 2.4416000000000005e-05,
      "loss": 75.9045,
      "step": 15260
    },
    {
      "epoch": 0.06169273221637302,
      "grad_norm": 2452.326416015625,
      "learning_rate": 2.4432000000000003e-05,
      "loss": 244.5978,
      "step": 15270
    },
    {
      "epoch": 0.06173313348174065,
      "grad_norm": 1301.530029296875,
      "learning_rate": 2.4448e-05,
      "loss": 111.0449,
      "step": 15280
    },
    {
      "epoch": 0.061773534747108276,
      "grad_norm": 696.3073120117188,
      "learning_rate": 2.4464000000000005e-05,
      "loss": 88.0738,
      "step": 15290
    },
    {
      "epoch": 0.06181393601247591,
      "grad_norm": 1441.196533203125,
      "learning_rate": 2.4480000000000003e-05,
      "loss": 132.6508,
      "step": 15300
    },
    {
      "epoch": 0.06185433727784354,
      "grad_norm": 833.18896484375,
      "learning_rate": 2.4496000000000004e-05,
      "loss": 140.9942,
      "step": 15310
    },
    {
      "epoch": 0.061894738543211175,
      "grad_norm": 1413.989013671875,
      "learning_rate": 2.4512000000000002e-05,
      "loss": 132.8459,
      "step": 15320
    },
    {
      "epoch": 0.0619351398085788,
      "grad_norm": 1399.9921875,
      "learning_rate": 2.4528e-05,
      "loss": 104.1776,
      "step": 15330
    },
    {
      "epoch": 0.061975541073946434,
      "grad_norm": 2461.732177734375,
      "learning_rate": 2.4544000000000004e-05,
      "loss": 157.2997,
      "step": 15340
    },
    {
      "epoch": 0.06201594233931407,
      "grad_norm": 645.6864624023438,
      "learning_rate": 2.4560000000000002e-05,
      "loss": 135.3236,
      "step": 15350
    },
    {
      "epoch": 0.0620563436046817,
      "grad_norm": 687.8350219726562,
      "learning_rate": 2.4576e-05,
      "loss": 101.6606,
      "step": 15360
    },
    {
      "epoch": 0.06209674487004933,
      "grad_norm": 1433.7965087890625,
      "learning_rate": 2.4592e-05,
      "loss": 119.4751,
      "step": 15370
    },
    {
      "epoch": 0.06213714613541696,
      "grad_norm": 375.80035400390625,
      "learning_rate": 2.4608000000000002e-05,
      "loss": 110.7543,
      "step": 15380
    },
    {
      "epoch": 0.06217754740078459,
      "grad_norm": 1924.688720703125,
      "learning_rate": 2.4624000000000003e-05,
      "loss": 134.0393,
      "step": 15390
    },
    {
      "epoch": 0.062217948666152226,
      "grad_norm": 1919.5198974609375,
      "learning_rate": 2.464e-05,
      "loss": 148.155,
      "step": 15400
    },
    {
      "epoch": 0.06225834993151985,
      "grad_norm": 952.2554931640625,
      "learning_rate": 2.4656e-05,
      "loss": 147.3355,
      "step": 15410
    },
    {
      "epoch": 0.062298751196887485,
      "grad_norm": 1131.790771484375,
      "learning_rate": 2.4672000000000004e-05,
      "loss": 128.157,
      "step": 15420
    },
    {
      "epoch": 0.06233915246225512,
      "grad_norm": 919.5127563476562,
      "learning_rate": 2.4688e-05,
      "loss": 115.3047,
      "step": 15430
    },
    {
      "epoch": 0.06237955372762275,
      "grad_norm": 2218.905029296875,
      "learning_rate": 2.4704000000000003e-05,
      "loss": 201.5949,
      "step": 15440
    },
    {
      "epoch": 0.06241995499299038,
      "grad_norm": 1855.6712646484375,
      "learning_rate": 2.472e-05,
      "loss": 134.0892,
      "step": 15450
    },
    {
      "epoch": 0.06246035625835801,
      "grad_norm": 877.10009765625,
      "learning_rate": 2.4736e-05,
      "loss": 102.3566,
      "step": 15460
    },
    {
      "epoch": 0.06250075752372564,
      "grad_norm": 1202.427734375,
      "learning_rate": 2.4752000000000003e-05,
      "loss": 109.7552,
      "step": 15470
    },
    {
      "epoch": 0.06254115878909328,
      "grad_norm": 1207.779296875,
      "learning_rate": 2.4768e-05,
      "loss": 101.6074,
      "step": 15480
    },
    {
      "epoch": 0.0625815600544609,
      "grad_norm": 1002.058349609375,
      "learning_rate": 2.4784000000000005e-05,
      "loss": 159.0241,
      "step": 15490
    },
    {
      "epoch": 0.06262196131982854,
      "grad_norm": 1376.6588134765625,
      "learning_rate": 2.4800000000000003e-05,
      "loss": 129.3523,
      "step": 15500
    },
    {
      "epoch": 0.06266236258519617,
      "grad_norm": 7330.33447265625,
      "learning_rate": 2.4816e-05,
      "loss": 124.5064,
      "step": 15510
    },
    {
      "epoch": 0.0627027638505638,
      "grad_norm": 1264.3929443359375,
      "learning_rate": 2.4832000000000002e-05,
      "loss": 122.9997,
      "step": 15520
    },
    {
      "epoch": 0.06274316511593143,
      "grad_norm": 2313.05712890625,
      "learning_rate": 2.4848e-05,
      "loss": 127.3592,
      "step": 15530
    },
    {
      "epoch": 0.06278356638129906,
      "grad_norm": 813.3026123046875,
      "learning_rate": 2.4864000000000005e-05,
      "loss": 111.5016,
      "step": 15540
    },
    {
      "epoch": 0.06282396764666669,
      "grad_norm": 1202.1077880859375,
      "learning_rate": 2.4880000000000002e-05,
      "loss": 130.1115,
      "step": 15550
    },
    {
      "epoch": 0.06286436891203433,
      "grad_norm": 1124.310791015625,
      "learning_rate": 2.4896e-05,
      "loss": 113.2873,
      "step": 15560
    },
    {
      "epoch": 0.06290477017740195,
      "grad_norm": 635.4754638671875,
      "learning_rate": 2.4912e-05,
      "loss": 143.046,
      "step": 15570
    },
    {
      "epoch": 0.06294517144276959,
      "grad_norm": 721.7821655273438,
      "learning_rate": 2.4928000000000003e-05,
      "loss": 136.0892,
      "step": 15580
    },
    {
      "epoch": 0.06298557270813722,
      "grad_norm": 2478.818115234375,
      "learning_rate": 2.4944000000000004e-05,
      "loss": 192.2085,
      "step": 15590
    },
    {
      "epoch": 0.06302597397350485,
      "grad_norm": 1114.921630859375,
      "learning_rate": 2.496e-05,
      "loss": 176.0056,
      "step": 15600
    },
    {
      "epoch": 0.06306637523887249,
      "grad_norm": 1268.53076171875,
      "learning_rate": 2.4976e-05,
      "loss": 136.4438,
      "step": 15610
    },
    {
      "epoch": 0.06310677650424011,
      "grad_norm": 1675.6795654296875,
      "learning_rate": 2.4992000000000004e-05,
      "loss": 137.4553,
      "step": 15620
    },
    {
      "epoch": 0.06314717776960774,
      "grad_norm": 1147.9560546875,
      "learning_rate": 2.5008000000000002e-05,
      "loss": 138.8533,
      "step": 15630
    },
    {
      "epoch": 0.06318757903497538,
      "grad_norm": 875.7190551757812,
      "learning_rate": 2.5024000000000003e-05,
      "loss": 162.8821,
      "step": 15640
    },
    {
      "epoch": 0.063227980300343,
      "grad_norm": 0.0,
      "learning_rate": 2.504e-05,
      "loss": 114.5405,
      "step": 15650
    },
    {
      "epoch": 0.06326838156571064,
      "grad_norm": 867.88037109375,
      "learning_rate": 2.5056e-05,
      "loss": 150.2994,
      "step": 15660
    },
    {
      "epoch": 0.06330878283107827,
      "grad_norm": 1961.737548828125,
      "learning_rate": 2.5072000000000003e-05,
      "loss": 99.0726,
      "step": 15670
    },
    {
      "epoch": 0.0633491840964459,
      "grad_norm": 989.47265625,
      "learning_rate": 2.5088e-05,
      "loss": 161.8524,
      "step": 15680
    },
    {
      "epoch": 0.06338958536181354,
      "grad_norm": 1500.89697265625,
      "learning_rate": 2.5104000000000002e-05,
      "loss": 162.7008,
      "step": 15690
    },
    {
      "epoch": 0.06342998662718116,
      "grad_norm": 899.3739013671875,
      "learning_rate": 2.5120000000000003e-05,
      "loss": 129.2148,
      "step": 15700
    },
    {
      "epoch": 0.06347038789254879,
      "grad_norm": 458.4581298828125,
      "learning_rate": 2.5136e-05,
      "loss": 105.5956,
      "step": 15710
    },
    {
      "epoch": 0.06351078915791643,
      "grad_norm": 974.6414184570312,
      "learning_rate": 2.5152000000000002e-05,
      "loss": 156.5464,
      "step": 15720
    },
    {
      "epoch": 0.06355119042328405,
      "grad_norm": 2536.4658203125,
      "learning_rate": 2.5168e-05,
      "loss": 109.3993,
      "step": 15730
    },
    {
      "epoch": 0.0635915916886517,
      "grad_norm": 2482.249267578125,
      "learning_rate": 2.5184000000000005e-05,
      "loss": 140.73,
      "step": 15740
    },
    {
      "epoch": 0.06363199295401932,
      "grad_norm": 2715.053955078125,
      "learning_rate": 2.5200000000000003e-05,
      "loss": 146.4766,
      "step": 15750
    },
    {
      "epoch": 0.06367239421938695,
      "grad_norm": 877.90771484375,
      "learning_rate": 2.5216e-05,
      "loss": 122.6491,
      "step": 15760
    },
    {
      "epoch": 0.06371279548475459,
      "grad_norm": 808.40625,
      "learning_rate": 2.5232e-05,
      "loss": 138.159,
      "step": 15770
    },
    {
      "epoch": 0.06375319675012221,
      "grad_norm": 1289.734130859375,
      "learning_rate": 2.5248000000000003e-05,
      "loss": 138.8296,
      "step": 15780
    },
    {
      "epoch": 0.06379359801548984,
      "grad_norm": 2172.106201171875,
      "learning_rate": 2.5264000000000004e-05,
      "loss": 144.9031,
      "step": 15790
    },
    {
      "epoch": 0.06383399928085748,
      "grad_norm": 930.3880004882812,
      "learning_rate": 2.5280000000000002e-05,
      "loss": 160.7876,
      "step": 15800
    },
    {
      "epoch": 0.0638744005462251,
      "grad_norm": 1657.616455078125,
      "learning_rate": 2.5296e-05,
      "loss": 111.3719,
      "step": 15810
    },
    {
      "epoch": 0.06391480181159274,
      "grad_norm": 791.59765625,
      "learning_rate": 2.5312000000000004e-05,
      "loss": 83.945,
      "step": 15820
    },
    {
      "epoch": 0.06395520307696037,
      "grad_norm": 3115.993896484375,
      "learning_rate": 2.5328000000000002e-05,
      "loss": 122.3074,
      "step": 15830
    },
    {
      "epoch": 0.063995604342328,
      "grad_norm": 1407.626220703125,
      "learning_rate": 2.5344000000000003e-05,
      "loss": 87.4488,
      "step": 15840
    },
    {
      "epoch": 0.06403600560769564,
      "grad_norm": 1312.5579833984375,
      "learning_rate": 2.536e-05,
      "loss": 126.2786,
      "step": 15850
    },
    {
      "epoch": 0.06407640687306326,
      "grad_norm": 1124.7440185546875,
      "learning_rate": 2.5376e-05,
      "loss": 147.5409,
      "step": 15860
    },
    {
      "epoch": 0.06411680813843089,
      "grad_norm": 1031.2725830078125,
      "learning_rate": 2.5392000000000004e-05,
      "loss": 130.0712,
      "step": 15870
    },
    {
      "epoch": 0.06415720940379853,
      "grad_norm": 1127.12548828125,
      "learning_rate": 2.5408e-05,
      "loss": 86.7857,
      "step": 15880
    },
    {
      "epoch": 0.06419761066916616,
      "grad_norm": 1454.6871337890625,
      "learning_rate": 2.5424000000000003e-05,
      "loss": 225.9277,
      "step": 15890
    },
    {
      "epoch": 0.0642380119345338,
      "grad_norm": 680.7732543945312,
      "learning_rate": 2.5440000000000004e-05,
      "loss": 114.08,
      "step": 15900
    },
    {
      "epoch": 0.06427841319990142,
      "grad_norm": 1291.77685546875,
      "learning_rate": 2.5456e-05,
      "loss": 112.0439,
      "step": 15910
    },
    {
      "epoch": 0.06431881446526905,
      "grad_norm": 870.5249633789062,
      "learning_rate": 2.5472000000000003e-05,
      "loss": 90.3746,
      "step": 15920
    },
    {
      "epoch": 0.06435921573063669,
      "grad_norm": 966.7274169921875,
      "learning_rate": 2.5488e-05,
      "loss": 180.7504,
      "step": 15930
    },
    {
      "epoch": 0.06439961699600431,
      "grad_norm": 584.5242919921875,
      "learning_rate": 2.5504e-05,
      "loss": 93.3127,
      "step": 15940
    },
    {
      "epoch": 0.06444001826137194,
      "grad_norm": 684.8132934570312,
      "learning_rate": 2.5520000000000003e-05,
      "loss": 104.0317,
      "step": 15950
    },
    {
      "epoch": 0.06448041952673958,
      "grad_norm": 736.9375610351562,
      "learning_rate": 2.5536e-05,
      "loss": 130.1245,
      "step": 15960
    },
    {
      "epoch": 0.0645208207921072,
      "grad_norm": 1334.9195556640625,
      "learning_rate": 2.5552000000000002e-05,
      "loss": 158.7703,
      "step": 15970
    },
    {
      "epoch": 0.06456122205747485,
      "grad_norm": 1718.863525390625,
      "learning_rate": 2.5568000000000003e-05,
      "loss": 165.1619,
      "step": 15980
    },
    {
      "epoch": 0.06460162332284247,
      "grad_norm": 1156.9222412109375,
      "learning_rate": 2.5584e-05,
      "loss": 158.3269,
      "step": 15990
    },
    {
      "epoch": 0.0646420245882101,
      "grad_norm": 1795.82177734375,
      "learning_rate": 2.5600000000000002e-05,
      "loss": 182.6038,
      "step": 16000
    },
    {
      "epoch": 0.06468242585357774,
      "grad_norm": 9040.69921875,
      "learning_rate": 2.5616e-05,
      "loss": 197.5475,
      "step": 16010
    },
    {
      "epoch": 0.06472282711894536,
      "grad_norm": 2413.524658203125,
      "learning_rate": 2.5632000000000005e-05,
      "loss": 150.8373,
      "step": 16020
    },
    {
      "epoch": 0.06476322838431299,
      "grad_norm": 1966.7882080078125,
      "learning_rate": 2.5648000000000002e-05,
      "loss": 163.2591,
      "step": 16030
    },
    {
      "epoch": 0.06480362964968063,
      "grad_norm": 985.5374145507812,
      "learning_rate": 2.5664e-05,
      "loss": 167.9098,
      "step": 16040
    },
    {
      "epoch": 0.06484403091504826,
      "grad_norm": 1213.7303466796875,
      "learning_rate": 2.568e-05,
      "loss": 120.5761,
      "step": 16050
    },
    {
      "epoch": 0.0648844321804159,
      "grad_norm": 2820.966796875,
      "learning_rate": 2.5696e-05,
      "loss": 107.9031,
      "step": 16060
    },
    {
      "epoch": 0.06492483344578352,
      "grad_norm": 1509.03076171875,
      "learning_rate": 2.5712000000000004e-05,
      "loss": 147.851,
      "step": 16070
    },
    {
      "epoch": 0.06496523471115115,
      "grad_norm": 5092.55859375,
      "learning_rate": 2.5728e-05,
      "loss": 107.7917,
      "step": 16080
    },
    {
      "epoch": 0.06500563597651879,
      "grad_norm": 1273.028076171875,
      "learning_rate": 2.5744e-05,
      "loss": 135.458,
      "step": 16090
    },
    {
      "epoch": 0.06504603724188641,
      "grad_norm": 6080.85498046875,
      "learning_rate": 2.5760000000000004e-05,
      "loss": 138.0135,
      "step": 16100
    },
    {
      "epoch": 0.06508643850725404,
      "grad_norm": 1283.4329833984375,
      "learning_rate": 2.5776000000000002e-05,
      "loss": 113.1424,
      "step": 16110
    },
    {
      "epoch": 0.06512683977262168,
      "grad_norm": 2675.70166015625,
      "learning_rate": 2.5792000000000003e-05,
      "loss": 106.4951,
      "step": 16120
    },
    {
      "epoch": 0.0651672410379893,
      "grad_norm": 8512.3154296875,
      "learning_rate": 2.5808e-05,
      "loss": 168.7911,
      "step": 16130
    },
    {
      "epoch": 0.06520764230335695,
      "grad_norm": 1965.6282958984375,
      "learning_rate": 2.5824e-05,
      "loss": 193.8709,
      "step": 16140
    },
    {
      "epoch": 0.06524804356872457,
      "grad_norm": 893.007080078125,
      "learning_rate": 2.5840000000000003e-05,
      "loss": 117.2167,
      "step": 16150
    },
    {
      "epoch": 0.0652884448340922,
      "grad_norm": 1275.0057373046875,
      "learning_rate": 2.5856e-05,
      "loss": 113.1999,
      "step": 16160
    },
    {
      "epoch": 0.06532884609945984,
      "grad_norm": 1035.0467529296875,
      "learning_rate": 2.5872000000000002e-05,
      "loss": 157.7083,
      "step": 16170
    },
    {
      "epoch": 0.06536924736482747,
      "grad_norm": 1248.9114990234375,
      "learning_rate": 2.5888000000000004e-05,
      "loss": 112.5848,
      "step": 16180
    },
    {
      "epoch": 0.06540964863019509,
      "grad_norm": 1170.686767578125,
      "learning_rate": 2.5904e-05,
      "loss": 167.6834,
      "step": 16190
    },
    {
      "epoch": 0.06545004989556273,
      "grad_norm": 974.4927978515625,
      "learning_rate": 2.5920000000000003e-05,
      "loss": 119.0438,
      "step": 16200
    },
    {
      "epoch": 0.06549045116093036,
      "grad_norm": 1922.8992919921875,
      "learning_rate": 2.5936e-05,
      "loss": 171.7924,
      "step": 16210
    },
    {
      "epoch": 0.065530852426298,
      "grad_norm": 772.7111206054688,
      "learning_rate": 2.5952000000000005e-05,
      "loss": 123.9591,
      "step": 16220
    },
    {
      "epoch": 0.06557125369166562,
      "grad_norm": 1295.742431640625,
      "learning_rate": 2.5968000000000003e-05,
      "loss": 167.7593,
      "step": 16230
    },
    {
      "epoch": 0.06561165495703325,
      "grad_norm": 1234.3642578125,
      "learning_rate": 2.5984e-05,
      "loss": 122.131,
      "step": 16240
    },
    {
      "epoch": 0.06565205622240089,
      "grad_norm": 435.2963562011719,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 104.5241,
      "step": 16250
    },
    {
      "epoch": 0.06569245748776852,
      "grad_norm": 489.53704833984375,
      "learning_rate": 2.6016e-05,
      "loss": 190.6924,
      "step": 16260
    },
    {
      "epoch": 0.06573285875313614,
      "grad_norm": 646.9764404296875,
      "learning_rate": 2.6032000000000004e-05,
      "loss": 138.386,
      "step": 16270
    },
    {
      "epoch": 0.06577326001850378,
      "grad_norm": 3126.28369140625,
      "learning_rate": 2.6048000000000002e-05,
      "loss": 199.9261,
      "step": 16280
    },
    {
      "epoch": 0.06581366128387141,
      "grad_norm": 879.2015991210938,
      "learning_rate": 2.6064e-05,
      "loss": 125.0898,
      "step": 16290
    },
    {
      "epoch": 0.06585406254923905,
      "grad_norm": 1662.759033203125,
      "learning_rate": 2.6080000000000004e-05,
      "loss": 180.1892,
      "step": 16300
    },
    {
      "epoch": 0.06589446381460667,
      "grad_norm": 1649.497802734375,
      "learning_rate": 2.6096000000000002e-05,
      "loss": 115.2371,
      "step": 16310
    },
    {
      "epoch": 0.0659348650799743,
      "grad_norm": 923.7296142578125,
      "learning_rate": 2.6112000000000003e-05,
      "loss": 120.7811,
      "step": 16320
    },
    {
      "epoch": 0.06597526634534194,
      "grad_norm": 661.9632568359375,
      "learning_rate": 2.6128e-05,
      "loss": 147.1547,
      "step": 16330
    },
    {
      "epoch": 0.06601566761070957,
      "grad_norm": 2440.39892578125,
      "learning_rate": 2.6144e-05,
      "loss": 134.8431,
      "step": 16340
    },
    {
      "epoch": 0.06605606887607719,
      "grad_norm": 1195.9329833984375,
      "learning_rate": 2.6160000000000004e-05,
      "loss": 198.3232,
      "step": 16350
    },
    {
      "epoch": 0.06609647014144483,
      "grad_norm": 674.6731567382812,
      "learning_rate": 2.6176e-05,
      "loss": 105.0343,
      "step": 16360
    },
    {
      "epoch": 0.06613687140681246,
      "grad_norm": 623.1482543945312,
      "learning_rate": 2.6192000000000003e-05,
      "loss": 110.9122,
      "step": 16370
    },
    {
      "epoch": 0.0661772726721801,
      "grad_norm": 1151.586181640625,
      "learning_rate": 2.6208000000000004e-05,
      "loss": 133.3356,
      "step": 16380
    },
    {
      "epoch": 0.06621767393754772,
      "grad_norm": 2473.778076171875,
      "learning_rate": 2.6224e-05,
      "loss": 129.0639,
      "step": 16390
    },
    {
      "epoch": 0.06625807520291535,
      "grad_norm": 1522.5828857421875,
      "learning_rate": 2.6240000000000003e-05,
      "loss": 101.4512,
      "step": 16400
    },
    {
      "epoch": 0.06629847646828299,
      "grad_norm": 3793.374755859375,
      "learning_rate": 2.6256e-05,
      "loss": 134.3651,
      "step": 16410
    },
    {
      "epoch": 0.06633887773365062,
      "grad_norm": 1190.8206787109375,
      "learning_rate": 2.6272000000000005e-05,
      "loss": 120.1487,
      "step": 16420
    },
    {
      "epoch": 0.06637927899901824,
      "grad_norm": 0.0,
      "learning_rate": 2.6288000000000003e-05,
      "loss": 71.3246,
      "step": 16430
    },
    {
      "epoch": 0.06641968026438588,
      "grad_norm": 746.3397216796875,
      "learning_rate": 2.6304e-05,
      "loss": 102.9548,
      "step": 16440
    },
    {
      "epoch": 0.06646008152975351,
      "grad_norm": 505.4109802246094,
      "learning_rate": 2.6320000000000002e-05,
      "loss": 109.599,
      "step": 16450
    },
    {
      "epoch": 0.06650048279512115,
      "grad_norm": 466.3865661621094,
      "learning_rate": 2.6336e-05,
      "loss": 121.0319,
      "step": 16460
    },
    {
      "epoch": 0.06654088406048878,
      "grad_norm": 1312.0987548828125,
      "learning_rate": 2.6352000000000005e-05,
      "loss": 124.488,
      "step": 16470
    },
    {
      "epoch": 0.0665812853258564,
      "grad_norm": 984.8582763671875,
      "learning_rate": 2.6368000000000002e-05,
      "loss": 99.0475,
      "step": 16480
    },
    {
      "epoch": 0.06662168659122404,
      "grad_norm": 1430.27001953125,
      "learning_rate": 2.6384e-05,
      "loss": 100.6167,
      "step": 16490
    },
    {
      "epoch": 0.06666208785659167,
      "grad_norm": 986.4356079101562,
      "learning_rate": 2.6400000000000005e-05,
      "loss": 130.1774,
      "step": 16500
    },
    {
      "epoch": 0.0667024891219593,
      "grad_norm": 684.648193359375,
      "learning_rate": 2.6416000000000003e-05,
      "loss": 127.4885,
      "step": 16510
    },
    {
      "epoch": 0.06674289038732693,
      "grad_norm": 866.86474609375,
      "learning_rate": 2.6432000000000004e-05,
      "loss": 134.0104,
      "step": 16520
    },
    {
      "epoch": 0.06678329165269456,
      "grad_norm": 761.9561157226562,
      "learning_rate": 2.6448e-05,
      "loss": 117.3864,
      "step": 16530
    },
    {
      "epoch": 0.0668236929180622,
      "grad_norm": 0.0,
      "learning_rate": 2.6464e-05,
      "loss": 89.1632,
      "step": 16540
    },
    {
      "epoch": 0.06686409418342983,
      "grad_norm": 1040.6204833984375,
      "learning_rate": 2.6480000000000004e-05,
      "loss": 147.5284,
      "step": 16550
    },
    {
      "epoch": 0.06690449544879745,
      "grad_norm": 1022.326171875,
      "learning_rate": 2.6496000000000002e-05,
      "loss": 119.413,
      "step": 16560
    },
    {
      "epoch": 0.06694489671416509,
      "grad_norm": 641.7581176757812,
      "learning_rate": 2.6512e-05,
      "loss": 118.3933,
      "step": 16570
    },
    {
      "epoch": 0.06698529797953272,
      "grad_norm": 1253.9312744140625,
      "learning_rate": 2.6528000000000004e-05,
      "loss": 125.6518,
      "step": 16580
    },
    {
      "epoch": 0.06702569924490034,
      "grad_norm": 2859.3115234375,
      "learning_rate": 2.6544000000000002e-05,
      "loss": 151.1297,
      "step": 16590
    },
    {
      "epoch": 0.06706610051026798,
      "grad_norm": 643.5469360351562,
      "learning_rate": 2.6560000000000003e-05,
      "loss": 150.4048,
      "step": 16600
    },
    {
      "epoch": 0.06710650177563561,
      "grad_norm": 457.0594177246094,
      "learning_rate": 2.6576e-05,
      "loss": 117.0777,
      "step": 16610
    },
    {
      "epoch": 0.06714690304100325,
      "grad_norm": 688.000244140625,
      "learning_rate": 2.6592e-05,
      "loss": 156.2777,
      "step": 16620
    },
    {
      "epoch": 0.06718730430637088,
      "grad_norm": 1273.36572265625,
      "learning_rate": 2.6608000000000003e-05,
      "loss": 219.1801,
      "step": 16630
    },
    {
      "epoch": 0.0672277055717385,
      "grad_norm": 1638.51708984375,
      "learning_rate": 2.6624e-05,
      "loss": 185.9331,
      "step": 16640
    },
    {
      "epoch": 0.06726810683710614,
      "grad_norm": 851.9219970703125,
      "learning_rate": 2.6640000000000002e-05,
      "loss": 118.9226,
      "step": 16650
    },
    {
      "epoch": 0.06730850810247377,
      "grad_norm": 1012.8102416992188,
      "learning_rate": 2.6656e-05,
      "loss": 117.7325,
      "step": 16660
    },
    {
      "epoch": 0.0673489093678414,
      "grad_norm": 752.0963745117188,
      "learning_rate": 2.6672e-05,
      "loss": 129.8979,
      "step": 16670
    },
    {
      "epoch": 0.06738931063320903,
      "grad_norm": 652.9700927734375,
      "learning_rate": 2.6688000000000003e-05,
      "loss": 126.5764,
      "step": 16680
    },
    {
      "epoch": 0.06742971189857666,
      "grad_norm": 498.1106872558594,
      "learning_rate": 2.6704e-05,
      "loss": 118.9791,
      "step": 16690
    },
    {
      "epoch": 0.0674701131639443,
      "grad_norm": 660.4872436523438,
      "learning_rate": 2.6720000000000005e-05,
      "loss": 117.8117,
      "step": 16700
    },
    {
      "epoch": 0.06751051442931193,
      "grad_norm": 773.5398559570312,
      "learning_rate": 2.6736000000000003e-05,
      "loss": 126.5417,
      "step": 16710
    },
    {
      "epoch": 0.06755091569467955,
      "grad_norm": 701.9429321289062,
      "learning_rate": 2.6752e-05,
      "loss": 121.2665,
      "step": 16720
    },
    {
      "epoch": 0.06759131696004719,
      "grad_norm": 1298.2679443359375,
      "learning_rate": 2.6768000000000002e-05,
      "loss": 134.9247,
      "step": 16730
    },
    {
      "epoch": 0.06763171822541482,
      "grad_norm": 733.1492919921875,
      "learning_rate": 2.6784e-05,
      "loss": 104.2357,
      "step": 16740
    },
    {
      "epoch": 0.06767211949078245,
      "grad_norm": 1005.172607421875,
      "learning_rate": 2.6800000000000004e-05,
      "loss": 110.5115,
      "step": 16750
    },
    {
      "epoch": 0.06771252075615009,
      "grad_norm": 897.8380737304688,
      "learning_rate": 2.6816000000000002e-05,
      "loss": 105.6745,
      "step": 16760
    },
    {
      "epoch": 0.06775292202151771,
      "grad_norm": 891.0289916992188,
      "learning_rate": 2.6832e-05,
      "loss": 127.119,
      "step": 16770
    },
    {
      "epoch": 0.06779332328688535,
      "grad_norm": 1331.1082763671875,
      "learning_rate": 2.6848000000000005e-05,
      "loss": 146.7941,
      "step": 16780
    },
    {
      "epoch": 0.06783372455225298,
      "grad_norm": 2302.458740234375,
      "learning_rate": 2.6864000000000002e-05,
      "loss": 147.356,
      "step": 16790
    },
    {
      "epoch": 0.0678741258176206,
      "grad_norm": 1848.271240234375,
      "learning_rate": 2.6880000000000004e-05,
      "loss": 140.58,
      "step": 16800
    },
    {
      "epoch": 0.06791452708298824,
      "grad_norm": 1468.3052978515625,
      "learning_rate": 2.6896e-05,
      "loss": 200.7853,
      "step": 16810
    },
    {
      "epoch": 0.06795492834835587,
      "grad_norm": 1507.734130859375,
      "learning_rate": 2.6912e-05,
      "loss": 133.0001,
      "step": 16820
    },
    {
      "epoch": 0.0679953296137235,
      "grad_norm": 1028.371337890625,
      "learning_rate": 2.6928000000000004e-05,
      "loss": 122.7306,
      "step": 16830
    },
    {
      "epoch": 0.06803573087909114,
      "grad_norm": 1497.8377685546875,
      "learning_rate": 2.6944e-05,
      "loss": 158.8021,
      "step": 16840
    },
    {
      "epoch": 0.06807613214445876,
      "grad_norm": 2305.415283203125,
      "learning_rate": 2.6960000000000003e-05,
      "loss": 159.5207,
      "step": 16850
    },
    {
      "epoch": 0.0681165334098264,
      "grad_norm": 2324.898193359375,
      "learning_rate": 2.6976e-05,
      "loss": 137.9295,
      "step": 16860
    },
    {
      "epoch": 0.06815693467519403,
      "grad_norm": 2173.98876953125,
      "learning_rate": 2.6992000000000002e-05,
      "loss": 158.221,
      "step": 16870
    },
    {
      "epoch": 0.06819733594056165,
      "grad_norm": 3569.12158203125,
      "learning_rate": 2.7008000000000003e-05,
      "loss": 145.5339,
      "step": 16880
    },
    {
      "epoch": 0.0682377372059293,
      "grad_norm": 1188.2027587890625,
      "learning_rate": 2.7024e-05,
      "loss": 114.8195,
      "step": 16890
    },
    {
      "epoch": 0.06827813847129692,
      "grad_norm": 1106.9266357421875,
      "learning_rate": 2.7040000000000005e-05,
      "loss": 169.873,
      "step": 16900
    },
    {
      "epoch": 0.06831853973666455,
      "grad_norm": 1759.7666015625,
      "learning_rate": 2.7056000000000003e-05,
      "loss": 115.8787,
      "step": 16910
    },
    {
      "epoch": 0.06835894100203219,
      "grad_norm": 3340.66162109375,
      "learning_rate": 2.7072e-05,
      "loss": 188.1911,
      "step": 16920
    },
    {
      "epoch": 0.06839934226739981,
      "grad_norm": 1076.770751953125,
      "learning_rate": 2.7088000000000002e-05,
      "loss": 132.0581,
      "step": 16930
    },
    {
      "epoch": 0.06843974353276745,
      "grad_norm": 516.5988159179688,
      "learning_rate": 2.7104e-05,
      "loss": 136.9663,
      "step": 16940
    },
    {
      "epoch": 0.06848014479813508,
      "grad_norm": 630.5826416015625,
      "learning_rate": 2.7120000000000005e-05,
      "loss": 99.7604,
      "step": 16950
    },
    {
      "epoch": 0.0685205460635027,
      "grad_norm": 1387.2393798828125,
      "learning_rate": 2.7136000000000002e-05,
      "loss": 161.1518,
      "step": 16960
    },
    {
      "epoch": 0.06856094732887034,
      "grad_norm": 2331.400390625,
      "learning_rate": 2.7152e-05,
      "loss": 170.4191,
      "step": 16970
    },
    {
      "epoch": 0.06860134859423797,
      "grad_norm": 812.1047973632812,
      "learning_rate": 2.7168000000000005e-05,
      "loss": 143.006,
      "step": 16980
    },
    {
      "epoch": 0.0686417498596056,
      "grad_norm": 2469.12255859375,
      "learning_rate": 2.7184000000000003e-05,
      "loss": 165.8849,
      "step": 16990
    },
    {
      "epoch": 0.06868215112497324,
      "grad_norm": 964.4491577148438,
      "learning_rate": 2.7200000000000004e-05,
      "loss": 244.0513,
      "step": 17000
    },
    {
      "epoch": 0.06872255239034086,
      "grad_norm": 1405.3626708984375,
      "learning_rate": 2.7216e-05,
      "loss": 119.3549,
      "step": 17010
    },
    {
      "epoch": 0.0687629536557085,
      "grad_norm": 0.0,
      "learning_rate": 2.7232e-05,
      "loss": 117.0388,
      "step": 17020
    },
    {
      "epoch": 0.06880335492107613,
      "grad_norm": 1727.89794921875,
      "learning_rate": 2.7248000000000004e-05,
      "loss": 164.1464,
      "step": 17030
    },
    {
      "epoch": 0.06884375618644376,
      "grad_norm": 1437.021728515625,
      "learning_rate": 2.7264000000000002e-05,
      "loss": 156.7928,
      "step": 17040
    },
    {
      "epoch": 0.0688841574518114,
      "grad_norm": 710.775634765625,
      "learning_rate": 2.7280000000000003e-05,
      "loss": 141.683,
      "step": 17050
    },
    {
      "epoch": 0.06892455871717902,
      "grad_norm": 879.23193359375,
      "learning_rate": 2.7296e-05,
      "loss": 115.3712,
      "step": 17060
    },
    {
      "epoch": 0.06896495998254665,
      "grad_norm": 1486.0762939453125,
      "learning_rate": 2.7312000000000002e-05,
      "loss": 107.1092,
      "step": 17070
    },
    {
      "epoch": 0.06900536124791429,
      "grad_norm": 1041.4608154296875,
      "learning_rate": 2.7328000000000003e-05,
      "loss": 93.719,
      "step": 17080
    },
    {
      "epoch": 0.06904576251328191,
      "grad_norm": 641.7068481445312,
      "learning_rate": 2.7344e-05,
      "loss": 94.0493,
      "step": 17090
    },
    {
      "epoch": 0.06908616377864955,
      "grad_norm": 1755.1153564453125,
      "learning_rate": 2.7360000000000006e-05,
      "loss": 127.0733,
      "step": 17100
    },
    {
      "epoch": 0.06912656504401718,
      "grad_norm": 659.0975952148438,
      "learning_rate": 2.7376000000000004e-05,
      "loss": 139.6171,
      "step": 17110
    },
    {
      "epoch": 0.0691669663093848,
      "grad_norm": 1058.2906494140625,
      "learning_rate": 2.7392e-05,
      "loss": 82.3711,
      "step": 17120
    },
    {
      "epoch": 0.06920736757475245,
      "grad_norm": 1285.0167236328125,
      "learning_rate": 2.7408000000000003e-05,
      "loss": 124.321,
      "step": 17130
    },
    {
      "epoch": 0.06924776884012007,
      "grad_norm": 1735.9765625,
      "learning_rate": 2.7424e-05,
      "loss": 125.6664,
      "step": 17140
    },
    {
      "epoch": 0.0692881701054877,
      "grad_norm": 9220.4453125,
      "learning_rate": 2.7440000000000005e-05,
      "loss": 156.938,
      "step": 17150
    },
    {
      "epoch": 0.06932857137085534,
      "grad_norm": 2777.53759765625,
      "learning_rate": 2.7456000000000003e-05,
      "loss": 115.4183,
      "step": 17160
    },
    {
      "epoch": 0.06936897263622296,
      "grad_norm": 1371.195068359375,
      "learning_rate": 2.7472e-05,
      "loss": 116.4201,
      "step": 17170
    },
    {
      "epoch": 0.0694093739015906,
      "grad_norm": 10435.0751953125,
      "learning_rate": 2.7488000000000005e-05,
      "loss": 142.1004,
      "step": 17180
    },
    {
      "epoch": 0.06944977516695823,
      "grad_norm": 1178.654296875,
      "learning_rate": 2.7504000000000003e-05,
      "loss": 181.0073,
      "step": 17190
    },
    {
      "epoch": 0.06949017643232586,
      "grad_norm": 654.7691040039062,
      "learning_rate": 2.752e-05,
      "loss": 93.7593,
      "step": 17200
    },
    {
      "epoch": 0.0695305776976935,
      "grad_norm": 1095.8475341796875,
      "learning_rate": 2.7536000000000002e-05,
      "loss": 109.7081,
      "step": 17210
    },
    {
      "epoch": 0.06957097896306112,
      "grad_norm": 2671.902587890625,
      "learning_rate": 2.7552e-05,
      "loss": 144.6699,
      "step": 17220
    },
    {
      "epoch": 0.06961138022842875,
      "grad_norm": 1229.1751708984375,
      "learning_rate": 2.7568000000000004e-05,
      "loss": 138.7873,
      "step": 17230
    },
    {
      "epoch": 0.06965178149379639,
      "grad_norm": 1349.5089111328125,
      "learning_rate": 2.7584000000000002e-05,
      "loss": 94.9966,
      "step": 17240
    },
    {
      "epoch": 0.06969218275916401,
      "grad_norm": 676.7572021484375,
      "learning_rate": 2.76e-05,
      "loss": 121.9994,
      "step": 17250
    },
    {
      "epoch": 0.06973258402453165,
      "grad_norm": 512.7169189453125,
      "learning_rate": 2.7616e-05,
      "loss": 80.4757,
      "step": 17260
    },
    {
      "epoch": 0.06977298528989928,
      "grad_norm": 591.3970947265625,
      "learning_rate": 2.7632000000000002e-05,
      "loss": 120.3442,
      "step": 17270
    },
    {
      "epoch": 0.06981338655526691,
      "grad_norm": 0.0,
      "learning_rate": 2.7648000000000004e-05,
      "loss": 111.9648,
      "step": 17280
    },
    {
      "epoch": 0.06985378782063455,
      "grad_norm": 904.6334228515625,
      "learning_rate": 2.7664e-05,
      "loss": 164.6575,
      "step": 17290
    },
    {
      "epoch": 0.06989418908600217,
      "grad_norm": 2712.48291015625,
      "learning_rate": 2.768e-05,
      "loss": 103.8473,
      "step": 17300
    },
    {
      "epoch": 0.0699345903513698,
      "grad_norm": 975.9140625,
      "learning_rate": 2.7696000000000004e-05,
      "loss": 132.7137,
      "step": 17310
    },
    {
      "epoch": 0.06997499161673744,
      "grad_norm": 844.25537109375,
      "learning_rate": 2.7712e-05,
      "loss": 134.3907,
      "step": 17320
    },
    {
      "epoch": 0.07001539288210507,
      "grad_norm": 558.0436401367188,
      "learning_rate": 2.7728000000000003e-05,
      "loss": 123.1448,
      "step": 17330
    },
    {
      "epoch": 0.0700557941474727,
      "grad_norm": 1072.6103515625,
      "learning_rate": 2.7744e-05,
      "loss": 133.5322,
      "step": 17340
    },
    {
      "epoch": 0.07009619541284033,
      "grad_norm": 605.5867919921875,
      "learning_rate": 2.7760000000000002e-05,
      "loss": 141.9054,
      "step": 17350
    },
    {
      "epoch": 0.07013659667820796,
      "grad_norm": 891.19384765625,
      "learning_rate": 2.7776000000000003e-05,
      "loss": 141.5447,
      "step": 17360
    },
    {
      "epoch": 0.0701769979435756,
      "grad_norm": 2915.88818359375,
      "learning_rate": 2.7792e-05,
      "loss": 171.8047,
      "step": 17370
    },
    {
      "epoch": 0.07021739920894322,
      "grad_norm": 1861.5186767578125,
      "learning_rate": 2.7808000000000005e-05,
      "loss": 164.1231,
      "step": 17380
    },
    {
      "epoch": 0.07025780047431085,
      "grad_norm": 998.0330810546875,
      "learning_rate": 2.7824000000000003e-05,
      "loss": 115.6295,
      "step": 17390
    },
    {
      "epoch": 0.07029820173967849,
      "grad_norm": 1189.415771484375,
      "learning_rate": 2.784e-05,
      "loss": 70.8924,
      "step": 17400
    },
    {
      "epoch": 0.07033860300504612,
      "grad_norm": 1055.349853515625,
      "learning_rate": 2.7856000000000002e-05,
      "loss": 190.4664,
      "step": 17410
    },
    {
      "epoch": 0.07037900427041376,
      "grad_norm": 1523.332275390625,
      "learning_rate": 2.7872e-05,
      "loss": 143.7663,
      "step": 17420
    },
    {
      "epoch": 0.07041940553578138,
      "grad_norm": 1106.2191162109375,
      "learning_rate": 2.7888000000000005e-05,
      "loss": 112.4859,
      "step": 17430
    },
    {
      "epoch": 0.07045980680114901,
      "grad_norm": 1430.1102294921875,
      "learning_rate": 2.7904000000000003e-05,
      "loss": 111.35,
      "step": 17440
    },
    {
      "epoch": 0.07050020806651665,
      "grad_norm": 1060.981201171875,
      "learning_rate": 2.792e-05,
      "loss": 89.6806,
      "step": 17450
    },
    {
      "epoch": 0.07054060933188427,
      "grad_norm": 2935.448486328125,
      "learning_rate": 2.7936e-05,
      "loss": 133.8329,
      "step": 17460
    },
    {
      "epoch": 0.0705810105972519,
      "grad_norm": 2200.762451171875,
      "learning_rate": 2.7952000000000003e-05,
      "loss": 145.5165,
      "step": 17470
    },
    {
      "epoch": 0.07062141186261954,
      "grad_norm": 1097.909423828125,
      "learning_rate": 2.7968000000000004e-05,
      "loss": 123.1423,
      "step": 17480
    },
    {
      "epoch": 0.07066181312798717,
      "grad_norm": 2040.2109375,
      "learning_rate": 2.7984000000000002e-05,
      "loss": 112.1275,
      "step": 17490
    },
    {
      "epoch": 0.0707022143933548,
      "grad_norm": 645.7501831054688,
      "learning_rate": 2.8e-05,
      "loss": 110.6064,
      "step": 17500
    },
    {
      "epoch": 0.07074261565872243,
      "grad_norm": 13496.923828125,
      "learning_rate": 2.8016000000000004e-05,
      "loss": 167.6134,
      "step": 17510
    },
    {
      "epoch": 0.07078301692409006,
      "grad_norm": 1220.3197021484375,
      "learning_rate": 2.8032000000000002e-05,
      "loss": 91.5186,
      "step": 17520
    },
    {
      "epoch": 0.0708234181894577,
      "grad_norm": 1471.6046142578125,
      "learning_rate": 2.8048000000000003e-05,
      "loss": 123.2698,
      "step": 17530
    },
    {
      "epoch": 0.07086381945482532,
      "grad_norm": 565.1813354492188,
      "learning_rate": 2.8064e-05,
      "loss": 108.927,
      "step": 17540
    },
    {
      "epoch": 0.07090422072019295,
      "grad_norm": 1448.7821044921875,
      "learning_rate": 2.8080000000000002e-05,
      "loss": 195.6281,
      "step": 17550
    },
    {
      "epoch": 0.07094462198556059,
      "grad_norm": 864.647216796875,
      "learning_rate": 2.8096000000000003e-05,
      "loss": 164.469,
      "step": 17560
    },
    {
      "epoch": 0.07098502325092822,
      "grad_norm": 605.3051147460938,
      "learning_rate": 2.8112e-05,
      "loss": 86.6841,
      "step": 17570
    },
    {
      "epoch": 0.07102542451629586,
      "grad_norm": 816.9525756835938,
      "learning_rate": 2.8128000000000006e-05,
      "loss": 132.6632,
      "step": 17580
    },
    {
      "epoch": 0.07106582578166348,
      "grad_norm": 1012.05419921875,
      "learning_rate": 2.8144000000000004e-05,
      "loss": 124.3037,
      "step": 17590
    },
    {
      "epoch": 0.07110622704703111,
      "grad_norm": 748.9439086914062,
      "learning_rate": 2.816e-05,
      "loss": 181.5007,
      "step": 17600
    },
    {
      "epoch": 0.07114662831239875,
      "grad_norm": 2104.114990234375,
      "learning_rate": 2.8176000000000003e-05,
      "loss": 130.2049,
      "step": 17610
    },
    {
      "epoch": 0.07118702957776638,
      "grad_norm": 0.0,
      "learning_rate": 2.8192e-05,
      "loss": 148.4061,
      "step": 17620
    },
    {
      "epoch": 0.071227430843134,
      "grad_norm": 751.892578125,
      "learning_rate": 2.8208000000000005e-05,
      "loss": 99.2189,
      "step": 17630
    },
    {
      "epoch": 0.07126783210850164,
      "grad_norm": 829.3687744140625,
      "learning_rate": 2.8224000000000003e-05,
      "loss": 100.3081,
      "step": 17640
    },
    {
      "epoch": 0.07130823337386927,
      "grad_norm": 1362.5450439453125,
      "learning_rate": 2.824e-05,
      "loss": 163.4869,
      "step": 17650
    },
    {
      "epoch": 0.07134863463923691,
      "grad_norm": 736.9421997070312,
      "learning_rate": 2.8256000000000002e-05,
      "loss": 114.6114,
      "step": 17660
    },
    {
      "epoch": 0.07138903590460453,
      "grad_norm": 0.0,
      "learning_rate": 2.8272000000000003e-05,
      "loss": 152.363,
      "step": 17670
    },
    {
      "epoch": 0.07142943716997216,
      "grad_norm": 1884.3018798828125,
      "learning_rate": 2.8288000000000004e-05,
      "loss": 181.776,
      "step": 17680
    },
    {
      "epoch": 0.0714698384353398,
      "grad_norm": 702.0150756835938,
      "learning_rate": 2.8304000000000002e-05,
      "loss": 138.0082,
      "step": 17690
    },
    {
      "epoch": 0.07151023970070743,
      "grad_norm": 1243.4521484375,
      "learning_rate": 2.832e-05,
      "loss": 126.1238,
      "step": 17700
    },
    {
      "epoch": 0.07155064096607505,
      "grad_norm": 2849.708740234375,
      "learning_rate": 2.8336000000000004e-05,
      "loss": 124.8339,
      "step": 17710
    },
    {
      "epoch": 0.07159104223144269,
      "grad_norm": 814.1820068359375,
      "learning_rate": 2.8352000000000002e-05,
      "loss": 100.6486,
      "step": 17720
    },
    {
      "epoch": 0.07163144349681032,
      "grad_norm": 1201.3582763671875,
      "learning_rate": 2.8368000000000004e-05,
      "loss": 145.7217,
      "step": 17730
    },
    {
      "epoch": 0.07167184476217796,
      "grad_norm": 571.5975952148438,
      "learning_rate": 2.8384e-05,
      "loss": 103.9509,
      "step": 17740
    },
    {
      "epoch": 0.07171224602754558,
      "grad_norm": 896.2073364257812,
      "learning_rate": 2.8400000000000003e-05,
      "loss": 155.8821,
      "step": 17750
    },
    {
      "epoch": 0.07175264729291321,
      "grad_norm": 962.5536499023438,
      "learning_rate": 2.8416000000000004e-05,
      "loss": 92.6593,
      "step": 17760
    },
    {
      "epoch": 0.07179304855828085,
      "grad_norm": 1393.9190673828125,
      "learning_rate": 2.8432e-05,
      "loss": 168.3829,
      "step": 17770
    },
    {
      "epoch": 0.07183344982364848,
      "grad_norm": 672.989501953125,
      "learning_rate": 2.8448000000000006e-05,
      "loss": 158.9301,
      "step": 17780
    },
    {
      "epoch": 0.0718738510890161,
      "grad_norm": 687.2365112304688,
      "learning_rate": 2.8464000000000004e-05,
      "loss": 79.7871,
      "step": 17790
    },
    {
      "epoch": 0.07191425235438374,
      "grad_norm": 738.2789916992188,
      "learning_rate": 2.8480000000000002e-05,
      "loss": 108.1994,
      "step": 17800
    },
    {
      "epoch": 0.07195465361975137,
      "grad_norm": 1095.77099609375,
      "learning_rate": 2.8496000000000003e-05,
      "loss": 106.8416,
      "step": 17810
    },
    {
      "epoch": 0.07199505488511901,
      "grad_norm": 326.4542236328125,
      "learning_rate": 2.8512e-05,
      "loss": 104.4799,
      "step": 17820
    },
    {
      "epoch": 0.07203545615048663,
      "grad_norm": 2892.407958984375,
      "learning_rate": 2.8528e-05,
      "loss": 94.0657,
      "step": 17830
    },
    {
      "epoch": 0.07207585741585426,
      "grad_norm": 1634.0128173828125,
      "learning_rate": 2.8544000000000003e-05,
      "loss": 119.6491,
      "step": 17840
    },
    {
      "epoch": 0.0721162586812219,
      "grad_norm": 2394.59033203125,
      "learning_rate": 2.856e-05,
      "loss": 160.5872,
      "step": 17850
    },
    {
      "epoch": 0.07215665994658953,
      "grad_norm": 995.382080078125,
      "learning_rate": 2.8576000000000002e-05,
      "loss": 89.2815,
      "step": 17860
    },
    {
      "epoch": 0.07219706121195715,
      "grad_norm": 562.4957275390625,
      "learning_rate": 2.8592000000000003e-05,
      "loss": 117.9278,
      "step": 17870
    },
    {
      "epoch": 0.0722374624773248,
      "grad_norm": 1529.1700439453125,
      "learning_rate": 2.8608e-05,
      "loss": 144.6818,
      "step": 17880
    },
    {
      "epoch": 0.07227786374269242,
      "grad_norm": 1666.5615234375,
      "learning_rate": 2.8624000000000002e-05,
      "loss": 154.7127,
      "step": 17890
    },
    {
      "epoch": 0.07231826500806006,
      "grad_norm": 1106.76416015625,
      "learning_rate": 2.864e-05,
      "loss": 121.5894,
      "step": 17900
    },
    {
      "epoch": 0.07235866627342769,
      "grad_norm": 1048.846435546875,
      "learning_rate": 2.8656000000000005e-05,
      "loss": 118.475,
      "step": 17910
    },
    {
      "epoch": 0.07239906753879531,
      "grad_norm": 1723.060546875,
      "learning_rate": 2.8672000000000003e-05,
      "loss": 81.5143,
      "step": 17920
    },
    {
      "epoch": 0.07243946880416295,
      "grad_norm": 827.5028076171875,
      "learning_rate": 2.8688e-05,
      "loss": 120.4572,
      "step": 17930
    },
    {
      "epoch": 0.07247987006953058,
      "grad_norm": 2439.23193359375,
      "learning_rate": 2.8704e-05,
      "loss": 150.0546,
      "step": 17940
    },
    {
      "epoch": 0.0725202713348982,
      "grad_norm": 1679.714111328125,
      "learning_rate": 2.8720000000000003e-05,
      "loss": 113.1306,
      "step": 17950
    },
    {
      "epoch": 0.07256067260026584,
      "grad_norm": 1051.8074951171875,
      "learning_rate": 2.8736000000000004e-05,
      "loss": 175.98,
      "step": 17960
    },
    {
      "epoch": 0.07260107386563347,
      "grad_norm": 708.1478881835938,
      "learning_rate": 2.8752000000000002e-05,
      "loss": 120.6407,
      "step": 17970
    },
    {
      "epoch": 0.07264147513100111,
      "grad_norm": 1140.8126220703125,
      "learning_rate": 2.8768e-05,
      "loss": 100.2487,
      "step": 17980
    },
    {
      "epoch": 0.07268187639636874,
      "grad_norm": 1086.7423095703125,
      "learning_rate": 2.8784000000000004e-05,
      "loss": 88.6672,
      "step": 17990
    },
    {
      "epoch": 0.07272227766173636,
      "grad_norm": 2896.58935546875,
      "learning_rate": 2.8800000000000002e-05,
      "loss": 123.9438,
      "step": 18000
    },
    {
      "epoch": 0.072762678927104,
      "grad_norm": 781.7880249023438,
      "learning_rate": 2.8816000000000003e-05,
      "loss": 128.9763,
      "step": 18010
    },
    {
      "epoch": 0.07280308019247163,
      "grad_norm": 1596.1004638671875,
      "learning_rate": 2.8832e-05,
      "loss": 103.9027,
      "step": 18020
    },
    {
      "epoch": 0.07284348145783925,
      "grad_norm": 1522.1275634765625,
      "learning_rate": 2.8848e-05,
      "loss": 82.2587,
      "step": 18030
    },
    {
      "epoch": 0.0728838827232069,
      "grad_norm": 1198.4852294921875,
      "learning_rate": 2.8864000000000004e-05,
      "loss": 78.4048,
      "step": 18040
    },
    {
      "epoch": 0.07292428398857452,
      "grad_norm": 1041.67578125,
      "learning_rate": 2.888e-05,
      "loss": 130.7603,
      "step": 18050
    },
    {
      "epoch": 0.07296468525394216,
      "grad_norm": 1674.17919921875,
      "learning_rate": 2.8896000000000003e-05,
      "loss": 133.4344,
      "step": 18060
    },
    {
      "epoch": 0.07300508651930979,
      "grad_norm": 1320.9654541015625,
      "learning_rate": 2.8912000000000004e-05,
      "loss": 101.3308,
      "step": 18070
    },
    {
      "epoch": 0.07304548778467741,
      "grad_norm": 2332.47265625,
      "learning_rate": 2.8928e-05,
      "loss": 104.6813,
      "step": 18080
    },
    {
      "epoch": 0.07308588905004505,
      "grad_norm": 677.6102294921875,
      "learning_rate": 2.8944000000000003e-05,
      "loss": 111.6524,
      "step": 18090
    },
    {
      "epoch": 0.07312629031541268,
      "grad_norm": 446.9483337402344,
      "learning_rate": 2.896e-05,
      "loss": 79.7397,
      "step": 18100
    },
    {
      "epoch": 0.0731666915807803,
      "grad_norm": 903.9324951171875,
      "learning_rate": 2.8976000000000005e-05,
      "loss": 106.6063,
      "step": 18110
    },
    {
      "epoch": 0.07320709284614794,
      "grad_norm": 825.9483642578125,
      "learning_rate": 2.8992000000000003e-05,
      "loss": 119.8426,
      "step": 18120
    },
    {
      "epoch": 0.07324749411151557,
      "grad_norm": 871.2711791992188,
      "learning_rate": 2.9008e-05,
      "loss": 147.6799,
      "step": 18130
    },
    {
      "epoch": 0.07328789537688321,
      "grad_norm": 1512.626220703125,
      "learning_rate": 2.9024000000000002e-05,
      "loss": 100.2563,
      "step": 18140
    },
    {
      "epoch": 0.07332829664225084,
      "grad_norm": 2183.07421875,
      "learning_rate": 2.9040000000000003e-05,
      "loss": 128.243,
      "step": 18150
    },
    {
      "epoch": 0.07336869790761846,
      "grad_norm": 577.5823974609375,
      "learning_rate": 2.9056000000000004e-05,
      "loss": 104.5627,
      "step": 18160
    },
    {
      "epoch": 0.0734090991729861,
      "grad_norm": 1346.950927734375,
      "learning_rate": 2.9072000000000002e-05,
      "loss": 131.9779,
      "step": 18170
    },
    {
      "epoch": 0.07344950043835373,
      "grad_norm": 897.385498046875,
      "learning_rate": 2.9088e-05,
      "loss": 139.202,
      "step": 18180
    },
    {
      "epoch": 0.07348990170372136,
      "grad_norm": 957.313232421875,
      "learning_rate": 2.9104000000000005e-05,
      "loss": 129.68,
      "step": 18190
    },
    {
      "epoch": 0.073530302969089,
      "grad_norm": 644.3162231445312,
      "learning_rate": 2.9120000000000002e-05,
      "loss": 126.1235,
      "step": 18200
    },
    {
      "epoch": 0.07357070423445662,
      "grad_norm": 812.66943359375,
      "learning_rate": 2.9136000000000004e-05,
      "loss": 144.1799,
      "step": 18210
    },
    {
      "epoch": 0.07361110549982426,
      "grad_norm": 1055.3414306640625,
      "learning_rate": 2.9152e-05,
      "loss": 129.42,
      "step": 18220
    },
    {
      "epoch": 0.07365150676519189,
      "grad_norm": 2815.364013671875,
      "learning_rate": 2.9168e-05,
      "loss": 130.298,
      "step": 18230
    },
    {
      "epoch": 0.07369190803055951,
      "grad_norm": 2299.622314453125,
      "learning_rate": 2.9184000000000004e-05,
      "loss": 153.6496,
      "step": 18240
    },
    {
      "epoch": 0.07373230929592715,
      "grad_norm": 1223.2135009765625,
      "learning_rate": 2.92e-05,
      "loss": 131.9549,
      "step": 18250
    },
    {
      "epoch": 0.07377271056129478,
      "grad_norm": 800.2896728515625,
      "learning_rate": 2.9216000000000003e-05,
      "loss": 96.7813,
      "step": 18260
    },
    {
      "epoch": 0.0738131118266624,
      "grad_norm": 1550.8492431640625,
      "learning_rate": 2.9232000000000004e-05,
      "loss": 134.3318,
      "step": 18270
    },
    {
      "epoch": 0.07385351309203005,
      "grad_norm": 1408.572021484375,
      "learning_rate": 2.9248000000000002e-05,
      "loss": 169.4472,
      "step": 18280
    },
    {
      "epoch": 0.07389391435739767,
      "grad_norm": 775.6466674804688,
      "learning_rate": 2.9264000000000003e-05,
      "loss": 139.8296,
      "step": 18290
    },
    {
      "epoch": 0.0739343156227653,
      "grad_norm": 3935.3876953125,
      "learning_rate": 2.928e-05,
      "loss": 197.3509,
      "step": 18300
    },
    {
      "epoch": 0.07397471688813294,
      "grad_norm": 1000.7813720703125,
      "learning_rate": 2.9296000000000005e-05,
      "loss": 124.4853,
      "step": 18310
    },
    {
      "epoch": 0.07401511815350056,
      "grad_norm": 746.0226440429688,
      "learning_rate": 2.9312000000000003e-05,
      "loss": 139.8521,
      "step": 18320
    },
    {
      "epoch": 0.0740555194188682,
      "grad_norm": 1989.065673828125,
      "learning_rate": 2.9328e-05,
      "loss": 163.3705,
      "step": 18330
    },
    {
      "epoch": 0.07409592068423583,
      "grad_norm": 1917.332763671875,
      "learning_rate": 2.9344000000000002e-05,
      "loss": 127.1422,
      "step": 18340
    },
    {
      "epoch": 0.07413632194960346,
      "grad_norm": 1084.9027099609375,
      "learning_rate": 2.9360000000000003e-05,
      "loss": 67.5235,
      "step": 18350
    },
    {
      "epoch": 0.0741767232149711,
      "grad_norm": 629.862548828125,
      "learning_rate": 2.9376000000000005e-05,
      "loss": 92.0765,
      "step": 18360
    },
    {
      "epoch": 0.07421712448033872,
      "grad_norm": 772.4987182617188,
      "learning_rate": 2.9392000000000003e-05,
      "loss": 188.4281,
      "step": 18370
    },
    {
      "epoch": 0.07425752574570635,
      "grad_norm": 977.8203125,
      "learning_rate": 2.9408e-05,
      "loss": 112.2548,
      "step": 18380
    },
    {
      "epoch": 0.07429792701107399,
      "grad_norm": 996.1343383789062,
      "learning_rate": 2.9424000000000005e-05,
      "loss": 125.9146,
      "step": 18390
    },
    {
      "epoch": 0.07433832827644161,
      "grad_norm": 1767.5286865234375,
      "learning_rate": 2.9440000000000003e-05,
      "loss": 121.1109,
      "step": 18400
    },
    {
      "epoch": 0.07437872954180925,
      "grad_norm": 4251.8291015625,
      "learning_rate": 2.9456000000000004e-05,
      "loss": 177.3741,
      "step": 18410
    },
    {
      "epoch": 0.07441913080717688,
      "grad_norm": 923.5038452148438,
      "learning_rate": 2.9472000000000002e-05,
      "loss": 98.2529,
      "step": 18420
    },
    {
      "epoch": 0.07445953207254451,
      "grad_norm": 1615.978271484375,
      "learning_rate": 2.9488e-05,
      "loss": 139.2994,
      "step": 18430
    },
    {
      "epoch": 0.07449993333791215,
      "grad_norm": 1293.6497802734375,
      "learning_rate": 2.9504000000000004e-05,
      "loss": 67.7334,
      "step": 18440
    },
    {
      "epoch": 0.07454033460327977,
      "grad_norm": 952.5794677734375,
      "learning_rate": 2.9520000000000002e-05,
      "loss": 98.4205,
      "step": 18450
    },
    {
      "epoch": 0.0745807358686474,
      "grad_norm": 843.2684936523438,
      "learning_rate": 2.9536e-05,
      "loss": 133.0427,
      "step": 18460
    },
    {
      "epoch": 0.07462113713401504,
      "grad_norm": 1032.7259521484375,
      "learning_rate": 2.9552000000000004e-05,
      "loss": 115.7002,
      "step": 18470
    },
    {
      "epoch": 0.07466153839938267,
      "grad_norm": 709.962890625,
      "learning_rate": 2.9568000000000002e-05,
      "loss": 93.0021,
      "step": 18480
    },
    {
      "epoch": 0.0747019396647503,
      "grad_norm": 1049.282470703125,
      "learning_rate": 2.9584000000000003e-05,
      "loss": 115.9874,
      "step": 18490
    },
    {
      "epoch": 0.07474234093011793,
      "grad_norm": 946.2427978515625,
      "learning_rate": 2.96e-05,
      "loss": 103.3239,
      "step": 18500
    },
    {
      "epoch": 0.07478274219548556,
      "grad_norm": 813.2142333984375,
      "learning_rate": 2.9616e-05,
      "loss": 96.9168,
      "step": 18510
    },
    {
      "epoch": 0.0748231434608532,
      "grad_norm": 702.8202514648438,
      "learning_rate": 2.9632000000000004e-05,
      "loss": 96.5812,
      "step": 18520
    },
    {
      "epoch": 0.07486354472622082,
      "grad_norm": 995.9627685546875,
      "learning_rate": 2.9648e-05,
      "loss": 97.2412,
      "step": 18530
    },
    {
      "epoch": 0.07490394599158845,
      "grad_norm": 3526.63671875,
      "learning_rate": 2.9664000000000003e-05,
      "loss": 192.7317,
      "step": 18540
    },
    {
      "epoch": 0.07494434725695609,
      "grad_norm": 455.59088134765625,
      "learning_rate": 2.9680000000000004e-05,
      "loss": 131.787,
      "step": 18550
    },
    {
      "epoch": 0.07498474852232372,
      "grad_norm": 1115.4947509765625,
      "learning_rate": 2.9696e-05,
      "loss": 86.6757,
      "step": 18560
    },
    {
      "epoch": 0.07502514978769136,
      "grad_norm": 947.726318359375,
      "learning_rate": 2.9712000000000003e-05,
      "loss": 96.3201,
      "step": 18570
    },
    {
      "epoch": 0.07506555105305898,
      "grad_norm": 1747.0982666015625,
      "learning_rate": 2.9728e-05,
      "loss": 120.4169,
      "step": 18580
    },
    {
      "epoch": 0.07510595231842661,
      "grad_norm": 678.9027099609375,
      "learning_rate": 2.9744000000000005e-05,
      "loss": 72.0365,
      "step": 18590
    },
    {
      "epoch": 0.07514635358379425,
      "grad_norm": 1209.59326171875,
      "learning_rate": 2.9760000000000003e-05,
      "loss": 199.8874,
      "step": 18600
    },
    {
      "epoch": 0.07518675484916187,
      "grad_norm": 658.3634643554688,
      "learning_rate": 2.9776e-05,
      "loss": 175.686,
      "step": 18610
    },
    {
      "epoch": 0.0752271561145295,
      "grad_norm": 748.1829833984375,
      "learning_rate": 2.9792000000000002e-05,
      "loss": 145.6304,
      "step": 18620
    },
    {
      "epoch": 0.07526755737989714,
      "grad_norm": 693.3621215820312,
      "learning_rate": 2.9808e-05,
      "loss": 120.1813,
      "step": 18630
    },
    {
      "epoch": 0.07530795864526477,
      "grad_norm": 731.7698364257812,
      "learning_rate": 2.9824000000000004e-05,
      "loss": 167.9795,
      "step": 18640
    },
    {
      "epoch": 0.0753483599106324,
      "grad_norm": 813.2443237304688,
      "learning_rate": 2.9840000000000002e-05,
      "loss": 73.1504,
      "step": 18650
    },
    {
      "epoch": 0.07538876117600003,
      "grad_norm": 2292.02490234375,
      "learning_rate": 2.9856e-05,
      "loss": 132.4692,
      "step": 18660
    },
    {
      "epoch": 0.07542916244136766,
      "grad_norm": 1452.90087890625,
      "learning_rate": 2.9872000000000005e-05,
      "loss": 134.5143,
      "step": 18670
    },
    {
      "epoch": 0.0754695637067353,
      "grad_norm": 918.1961669921875,
      "learning_rate": 2.9888000000000003e-05,
      "loss": 106.5664,
      "step": 18680
    },
    {
      "epoch": 0.07550996497210292,
      "grad_norm": 1377.7470703125,
      "learning_rate": 2.9904000000000004e-05,
      "loss": 89.391,
      "step": 18690
    },
    {
      "epoch": 0.07555036623747055,
      "grad_norm": 1466.1744384765625,
      "learning_rate": 2.992e-05,
      "loss": 124.1821,
      "step": 18700
    },
    {
      "epoch": 0.07559076750283819,
      "grad_norm": 1014.8499755859375,
      "learning_rate": 2.9936e-05,
      "loss": 185.3473,
      "step": 18710
    },
    {
      "epoch": 0.07563116876820582,
      "grad_norm": 1184.2947998046875,
      "learning_rate": 2.9952000000000004e-05,
      "loss": 125.8145,
      "step": 18720
    },
    {
      "epoch": 0.07567157003357346,
      "grad_norm": 774.492919921875,
      "learning_rate": 2.9968000000000002e-05,
      "loss": 120.7197,
      "step": 18730
    },
    {
      "epoch": 0.07571197129894108,
      "grad_norm": 1086.3082275390625,
      "learning_rate": 2.9984000000000003e-05,
      "loss": 126.5244,
      "step": 18740
    },
    {
      "epoch": 0.07575237256430871,
      "grad_norm": 937.5267944335938,
      "learning_rate": 3.0000000000000004e-05,
      "loss": 118.7993,
      "step": 18750
    },
    {
      "epoch": 0.07579277382967635,
      "grad_norm": 1027.5623779296875,
      "learning_rate": 3.0016000000000002e-05,
      "loss": 94.4266,
      "step": 18760
    },
    {
      "epoch": 0.07583317509504398,
      "grad_norm": 516.3690185546875,
      "learning_rate": 3.0032000000000003e-05,
      "loss": 120.8037,
      "step": 18770
    },
    {
      "epoch": 0.0758735763604116,
      "grad_norm": 1685.0865478515625,
      "learning_rate": 3.0048e-05,
      "loss": 106.1391,
      "step": 18780
    },
    {
      "epoch": 0.07591397762577924,
      "grad_norm": 0.0,
      "learning_rate": 3.0064000000000006e-05,
      "loss": 74.0085,
      "step": 18790
    },
    {
      "epoch": 0.07595437889114687,
      "grad_norm": 1456.856689453125,
      "learning_rate": 3.0080000000000003e-05,
      "loss": 118.443,
      "step": 18800
    },
    {
      "epoch": 0.07599478015651451,
      "grad_norm": 2234.4599609375,
      "learning_rate": 3.0096e-05,
      "loss": 171.6478,
      "step": 18810
    },
    {
      "epoch": 0.07603518142188213,
      "grad_norm": 976.5721435546875,
      "learning_rate": 3.0112000000000002e-05,
      "loss": 155.1822,
      "step": 18820
    },
    {
      "epoch": 0.07607558268724976,
      "grad_norm": 1938.18603515625,
      "learning_rate": 3.0128e-05,
      "loss": 139.4546,
      "step": 18830
    },
    {
      "epoch": 0.0761159839526174,
      "grad_norm": 621.24755859375,
      "learning_rate": 3.0144000000000005e-05,
      "loss": 84.2352,
      "step": 18840
    },
    {
      "epoch": 0.07615638521798503,
      "grad_norm": 599.5986938476562,
      "learning_rate": 3.0160000000000003e-05,
      "loss": 139.9842,
      "step": 18850
    },
    {
      "epoch": 0.07619678648335265,
      "grad_norm": 1418.4581298828125,
      "learning_rate": 3.0176e-05,
      "loss": 128.9917,
      "step": 18860
    },
    {
      "epoch": 0.07623718774872029,
      "grad_norm": 1401.3021240234375,
      "learning_rate": 3.0192000000000005e-05,
      "loss": 104.3214,
      "step": 18870
    },
    {
      "epoch": 0.07627758901408792,
      "grad_norm": 748.1585083007812,
      "learning_rate": 3.0208000000000003e-05,
      "loss": 119.8759,
      "step": 18880
    },
    {
      "epoch": 0.07631799027945556,
      "grad_norm": 1339.5333251953125,
      "learning_rate": 3.0224000000000004e-05,
      "loss": 115.9315,
      "step": 18890
    },
    {
      "epoch": 0.07635839154482318,
      "grad_norm": 1126.16552734375,
      "learning_rate": 3.0240000000000002e-05,
      "loss": 145.7979,
      "step": 18900
    },
    {
      "epoch": 0.07639879281019081,
      "grad_norm": 808.7857055664062,
      "learning_rate": 3.0256e-05,
      "loss": 127.7257,
      "step": 18910
    },
    {
      "epoch": 0.07643919407555845,
      "grad_norm": 1016.2286987304688,
      "learning_rate": 3.0272000000000004e-05,
      "loss": 228.2659,
      "step": 18920
    },
    {
      "epoch": 0.07647959534092608,
      "grad_norm": 1210.65771484375,
      "learning_rate": 3.0288000000000002e-05,
      "loss": 118.8449,
      "step": 18930
    },
    {
      "epoch": 0.0765199966062937,
      "grad_norm": 759.594970703125,
      "learning_rate": 3.0304000000000003e-05,
      "loss": 73.6208,
      "step": 18940
    },
    {
      "epoch": 0.07656039787166134,
      "grad_norm": 805.1461181640625,
      "learning_rate": 3.032e-05,
      "loss": 100.1516,
      "step": 18950
    },
    {
      "epoch": 0.07660079913702897,
      "grad_norm": 1020.590576171875,
      "learning_rate": 3.0336000000000002e-05,
      "loss": 130.3593,
      "step": 18960
    },
    {
      "epoch": 0.07664120040239661,
      "grad_norm": 281.2948303222656,
      "learning_rate": 3.0352000000000003e-05,
      "loss": 87.3999,
      "step": 18970
    },
    {
      "epoch": 0.07668160166776423,
      "grad_norm": 1126.6138916015625,
      "learning_rate": 3.0368e-05,
      "loss": 125.5814,
      "step": 18980
    },
    {
      "epoch": 0.07672200293313186,
      "grad_norm": 3934.1298828125,
      "learning_rate": 3.0384000000000006e-05,
      "loss": 179.5769,
      "step": 18990
    },
    {
      "epoch": 0.0767624041984995,
      "grad_norm": 1989.8963623046875,
      "learning_rate": 3.0400000000000004e-05,
      "loss": 164.1262,
      "step": 19000
    },
    {
      "epoch": 0.07680280546386713,
      "grad_norm": 891.9097290039062,
      "learning_rate": 3.0416e-05,
      "loss": 78.4145,
      "step": 19010
    },
    {
      "epoch": 0.07684320672923475,
      "grad_norm": 1621.536865234375,
      "learning_rate": 3.0432000000000003e-05,
      "loss": 97.0821,
      "step": 19020
    },
    {
      "epoch": 0.0768836079946024,
      "grad_norm": 1369.31591796875,
      "learning_rate": 3.0448e-05,
      "loss": 97.2216,
      "step": 19030
    },
    {
      "epoch": 0.07692400925997002,
      "grad_norm": 1075.7518310546875,
      "learning_rate": 3.0464000000000005e-05,
      "loss": 130.9594,
      "step": 19040
    },
    {
      "epoch": 0.07696441052533766,
      "grad_norm": 1702.1077880859375,
      "learning_rate": 3.0480000000000003e-05,
      "loss": 157.67,
      "step": 19050
    },
    {
      "epoch": 0.07700481179070529,
      "grad_norm": 1624.005126953125,
      "learning_rate": 3.0496e-05,
      "loss": 180.6171,
      "step": 19060
    },
    {
      "epoch": 0.07704521305607291,
      "grad_norm": 1812.77099609375,
      "learning_rate": 3.0512000000000005e-05,
      "loss": 115.4045,
      "step": 19070
    },
    {
      "epoch": 0.07708561432144055,
      "grad_norm": 823.732177734375,
      "learning_rate": 3.0528e-05,
      "loss": 105.2697,
      "step": 19080
    },
    {
      "epoch": 0.07712601558680818,
      "grad_norm": 900.1708984375,
      "learning_rate": 3.0544e-05,
      "loss": 118.3052,
      "step": 19090
    },
    {
      "epoch": 0.0771664168521758,
      "grad_norm": 1073.2254638671875,
      "learning_rate": 3.0560000000000006e-05,
      "loss": 161.0425,
      "step": 19100
    },
    {
      "epoch": 0.07720681811754344,
      "grad_norm": 932.1373901367188,
      "learning_rate": 3.0576e-05,
      "loss": 124.8353,
      "step": 19110
    },
    {
      "epoch": 0.07724721938291107,
      "grad_norm": 992.2178344726562,
      "learning_rate": 3.0592e-05,
      "loss": 87.1603,
      "step": 19120
    },
    {
      "epoch": 0.07728762064827871,
      "grad_norm": 655.614013671875,
      "learning_rate": 3.0608e-05,
      "loss": 148.0067,
      "step": 19130
    },
    {
      "epoch": 0.07732802191364634,
      "grad_norm": 566.0341186523438,
      "learning_rate": 3.0624000000000004e-05,
      "loss": 104.9753,
      "step": 19140
    },
    {
      "epoch": 0.07736842317901396,
      "grad_norm": 604.9459838867188,
      "learning_rate": 3.0640000000000005e-05,
      "loss": 80.1844,
      "step": 19150
    },
    {
      "epoch": 0.0774088244443816,
      "grad_norm": 939.7496948242188,
      "learning_rate": 3.0656e-05,
      "loss": 78.9237,
      "step": 19160
    },
    {
      "epoch": 0.07744922570974923,
      "grad_norm": 1111.1923828125,
      "learning_rate": 3.067200000000001e-05,
      "loss": 70.4472,
      "step": 19170
    },
    {
      "epoch": 0.07748962697511685,
      "grad_norm": 1667.745361328125,
      "learning_rate": 3.0688e-05,
      "loss": 109.166,
      "step": 19180
    },
    {
      "epoch": 0.0775300282404845,
      "grad_norm": 1046.6304931640625,
      "learning_rate": 3.0704e-05,
      "loss": 165.926,
      "step": 19190
    },
    {
      "epoch": 0.07757042950585212,
      "grad_norm": 1761.15869140625,
      "learning_rate": 3.0720000000000004e-05,
      "loss": 129.2139,
      "step": 19200
    },
    {
      "epoch": 0.07761083077121976,
      "grad_norm": 1684.56884765625,
      "learning_rate": 3.0736e-05,
      "loss": 154.1343,
      "step": 19210
    },
    {
      "epoch": 0.07765123203658739,
      "grad_norm": 551.8289794921875,
      "learning_rate": 3.0752000000000006e-05,
      "loss": 172.8532,
      "step": 19220
    },
    {
      "epoch": 0.07769163330195501,
      "grad_norm": 1006.9168701171875,
      "learning_rate": 3.0768e-05,
      "loss": 110.0696,
      "step": 19230
    },
    {
      "epoch": 0.07773203456732265,
      "grad_norm": 1449.43994140625,
      "learning_rate": 3.0784e-05,
      "loss": 160.3563,
      "step": 19240
    },
    {
      "epoch": 0.07777243583269028,
      "grad_norm": 994.652587890625,
      "learning_rate": 3.08e-05,
      "loss": 106.7819,
      "step": 19250
    },
    {
      "epoch": 0.0778128370980579,
      "grad_norm": 1494.919189453125,
      "learning_rate": 3.0816000000000004e-05,
      "loss": 154.2932,
      "step": 19260
    },
    {
      "epoch": 0.07785323836342554,
      "grad_norm": 3010.73974609375,
      "learning_rate": 3.0832000000000006e-05,
      "loss": 115.7749,
      "step": 19270
    },
    {
      "epoch": 0.07789363962879317,
      "grad_norm": 1167.75634765625,
      "learning_rate": 3.0848e-05,
      "loss": 84.5835,
      "step": 19280
    },
    {
      "epoch": 0.07793404089416081,
      "grad_norm": 512.6953735351562,
      "learning_rate": 3.0864e-05,
      "loss": 70.4986,
      "step": 19290
    },
    {
      "epoch": 0.07797444215952844,
      "grad_norm": 1121.130126953125,
      "learning_rate": 3.088e-05,
      "loss": 131.0057,
      "step": 19300
    },
    {
      "epoch": 0.07801484342489606,
      "grad_norm": 821.2005004882812,
      "learning_rate": 3.0896000000000004e-05,
      "loss": 93.5886,
      "step": 19310
    },
    {
      "epoch": 0.0780552446902637,
      "grad_norm": 19915.904296875,
      "learning_rate": 3.0912000000000005e-05,
      "loss": 210.1925,
      "step": 19320
    },
    {
      "epoch": 0.07809564595563133,
      "grad_norm": 970.84033203125,
      "learning_rate": 3.0928e-05,
      "loss": 158.7296,
      "step": 19330
    },
    {
      "epoch": 0.07813604722099896,
      "grad_norm": 2203.380859375,
      "learning_rate": 3.0944e-05,
      "loss": 124.2448,
      "step": 19340
    },
    {
      "epoch": 0.0781764484863666,
      "grad_norm": 1706.534423828125,
      "learning_rate": 3.096e-05,
      "loss": 112.4453,
      "step": 19350
    },
    {
      "epoch": 0.07821684975173422,
      "grad_norm": 558.5722045898438,
      "learning_rate": 3.0976e-05,
      "loss": 158.2485,
      "step": 19360
    },
    {
      "epoch": 0.07825725101710186,
      "grad_norm": 1104.03125,
      "learning_rate": 3.0992000000000004e-05,
      "loss": 112.5157,
      "step": 19370
    },
    {
      "epoch": 0.07829765228246949,
      "grad_norm": 3149.04296875,
      "learning_rate": 3.1008000000000005e-05,
      "loss": 172.0443,
      "step": 19380
    },
    {
      "epoch": 0.07833805354783711,
      "grad_norm": 1571.8192138671875,
      "learning_rate": 3.1024e-05,
      "loss": 162.6578,
      "step": 19390
    },
    {
      "epoch": 0.07837845481320475,
      "grad_norm": 1629.932373046875,
      "learning_rate": 3.104e-05,
      "loss": 164.0212,
      "step": 19400
    },
    {
      "epoch": 0.07841885607857238,
      "grad_norm": 2302.69140625,
      "learning_rate": 3.1056e-05,
      "loss": 138.8319,
      "step": 19410
    },
    {
      "epoch": 0.07845925734394,
      "grad_norm": 1597.150634765625,
      "learning_rate": 3.1072e-05,
      "loss": 127.0055,
      "step": 19420
    },
    {
      "epoch": 0.07849965860930765,
      "grad_norm": 905.0975952148438,
      "learning_rate": 3.1088000000000005e-05,
      "loss": 124.8965,
      "step": 19430
    },
    {
      "epoch": 0.07854005987467527,
      "grad_norm": 756.5381469726562,
      "learning_rate": 3.1104e-05,
      "loss": 111.2599,
      "step": 19440
    },
    {
      "epoch": 0.07858046114004291,
      "grad_norm": 3848.724609375,
      "learning_rate": 3.112e-05,
      "loss": 138.655,
      "step": 19450
    },
    {
      "epoch": 0.07862086240541054,
      "grad_norm": 344.63385009765625,
      "learning_rate": 3.1136e-05,
      "loss": 85.4582,
      "step": 19460
    },
    {
      "epoch": 0.07866126367077816,
      "grad_norm": 996.605712890625,
      "learning_rate": 3.1152e-05,
      "loss": 133.7001,
      "step": 19470
    },
    {
      "epoch": 0.0787016649361458,
      "grad_norm": 1711.14892578125,
      "learning_rate": 3.1168000000000004e-05,
      "loss": 95.2133,
      "step": 19480
    },
    {
      "epoch": 0.07874206620151343,
      "grad_norm": 1539.0604248046875,
      "learning_rate": 3.1184e-05,
      "loss": 105.0518,
      "step": 19490
    },
    {
      "epoch": 0.07878246746688106,
      "grad_norm": 744.2188720703125,
      "learning_rate": 3.1200000000000006e-05,
      "loss": 112.2348,
      "step": 19500
    },
    {
      "epoch": 0.0788228687322487,
      "grad_norm": 665.140380859375,
      "learning_rate": 3.1216e-05,
      "loss": 105.9944,
      "step": 19510
    },
    {
      "epoch": 0.07886326999761632,
      "grad_norm": 1196.1968994140625,
      "learning_rate": 3.1232e-05,
      "loss": 102.4042,
      "step": 19520
    },
    {
      "epoch": 0.07890367126298396,
      "grad_norm": 1342.2335205078125,
      "learning_rate": 3.1248e-05,
      "loss": 124.4353,
      "step": 19530
    },
    {
      "epoch": 0.07894407252835159,
      "grad_norm": 2917.825439453125,
      "learning_rate": 3.1264000000000004e-05,
      "loss": 119.8374,
      "step": 19540
    },
    {
      "epoch": 0.07898447379371921,
      "grad_norm": 1537.9954833984375,
      "learning_rate": 3.1280000000000005e-05,
      "loss": 120.8902,
      "step": 19550
    },
    {
      "epoch": 0.07902487505908685,
      "grad_norm": 713.2342529296875,
      "learning_rate": 3.1296e-05,
      "loss": 123.1111,
      "step": 19560
    },
    {
      "epoch": 0.07906527632445448,
      "grad_norm": 1007.7623901367188,
      "learning_rate": 3.131200000000001e-05,
      "loss": 122.3046,
      "step": 19570
    },
    {
      "epoch": 0.07910567758982211,
      "grad_norm": 744.2635498046875,
      "learning_rate": 3.1328e-05,
      "loss": 88.612,
      "step": 19580
    },
    {
      "epoch": 0.07914607885518975,
      "grad_norm": 1206.9375,
      "learning_rate": 3.1344000000000003e-05,
      "loss": 81.3717,
      "step": 19590
    },
    {
      "epoch": 0.07918648012055737,
      "grad_norm": 1274.3399658203125,
      "learning_rate": 3.1360000000000005e-05,
      "loss": 157.8111,
      "step": 19600
    },
    {
      "epoch": 0.07922688138592501,
      "grad_norm": 696.0009765625,
      "learning_rate": 3.1376e-05,
      "loss": 112.9236,
      "step": 19610
    },
    {
      "epoch": 0.07926728265129264,
      "grad_norm": 1086.2701416015625,
      "learning_rate": 3.139200000000001e-05,
      "loss": 103.9647,
      "step": 19620
    },
    {
      "epoch": 0.07930768391666027,
      "grad_norm": 1140.2965087890625,
      "learning_rate": 3.1408e-05,
      "loss": 92.4705,
      "step": 19630
    },
    {
      "epoch": 0.0793480851820279,
      "grad_norm": 1569.0645751953125,
      "learning_rate": 3.1424e-05,
      "loss": 121.3736,
      "step": 19640
    },
    {
      "epoch": 0.07938848644739553,
      "grad_norm": 1136.5538330078125,
      "learning_rate": 3.1440000000000004e-05,
      "loss": 112.8915,
      "step": 19650
    },
    {
      "epoch": 0.07942888771276316,
      "grad_norm": 740.44091796875,
      "learning_rate": 3.1456000000000005e-05,
      "loss": 121.2627,
      "step": 19660
    },
    {
      "epoch": 0.0794692889781308,
      "grad_norm": 1086.4205322265625,
      "learning_rate": 3.1472000000000006e-05,
      "loss": 119.6377,
      "step": 19670
    },
    {
      "epoch": 0.07950969024349842,
      "grad_norm": 943.4683227539062,
      "learning_rate": 3.1488e-05,
      "loss": 95.2023,
      "step": 19680
    },
    {
      "epoch": 0.07955009150886606,
      "grad_norm": 991.89892578125,
      "learning_rate": 3.1504e-05,
      "loss": 178.4546,
      "step": 19690
    },
    {
      "epoch": 0.07959049277423369,
      "grad_norm": 421.7826232910156,
      "learning_rate": 3.152e-05,
      "loss": 100.0158,
      "step": 19700
    },
    {
      "epoch": 0.07963089403960132,
      "grad_norm": 1290.488525390625,
      "learning_rate": 3.1536000000000004e-05,
      "loss": 104.956,
      "step": 19710
    },
    {
      "epoch": 0.07967129530496896,
      "grad_norm": 1131.934326171875,
      "learning_rate": 3.1552e-05,
      "loss": 100.3167,
      "step": 19720
    },
    {
      "epoch": 0.07971169657033658,
      "grad_norm": 1422.13525390625,
      "learning_rate": 3.1568e-05,
      "loss": 139.9084,
      "step": 19730
    },
    {
      "epoch": 0.07975209783570421,
      "grad_norm": 1051.194580078125,
      "learning_rate": 3.1584e-05,
      "loss": 109.0689,
      "step": 19740
    },
    {
      "epoch": 0.07979249910107185,
      "grad_norm": 1163.12548828125,
      "learning_rate": 3.16e-05,
      "loss": 89.2682,
      "step": 19750
    },
    {
      "epoch": 0.07983290036643947,
      "grad_norm": 1967.873291015625,
      "learning_rate": 3.1616000000000004e-05,
      "loss": 128.6821,
      "step": 19760
    },
    {
      "epoch": 0.07987330163180711,
      "grad_norm": 1103.2349853515625,
      "learning_rate": 3.1632e-05,
      "loss": 147.6279,
      "step": 19770
    },
    {
      "epoch": 0.07991370289717474,
      "grad_norm": 1125.0118408203125,
      "learning_rate": 3.1648000000000006e-05,
      "loss": 92.0829,
      "step": 19780
    },
    {
      "epoch": 0.07995410416254237,
      "grad_norm": 1415.862060546875,
      "learning_rate": 3.1664e-05,
      "loss": 111.4338,
      "step": 19790
    },
    {
      "epoch": 0.07999450542791,
      "grad_norm": 1246.4283447265625,
      "learning_rate": 3.168e-05,
      "loss": 83.8045,
      "step": 19800
    },
    {
      "epoch": 0.08003490669327763,
      "grad_norm": 1319.5565185546875,
      "learning_rate": 3.1696e-05,
      "loss": 138.7009,
      "step": 19810
    },
    {
      "epoch": 0.08007530795864526,
      "grad_norm": 539.1595458984375,
      "learning_rate": 3.1712e-05,
      "loss": 143.3671,
      "step": 19820
    },
    {
      "epoch": 0.0801157092240129,
      "grad_norm": 2494.538818359375,
      "learning_rate": 3.1728000000000005e-05,
      "loss": 186.8914,
      "step": 19830
    },
    {
      "epoch": 0.08015611048938052,
      "grad_norm": 1490.8349609375,
      "learning_rate": 3.1744e-05,
      "loss": 178.6781,
      "step": 19840
    },
    {
      "epoch": 0.08019651175474816,
      "grad_norm": 1049.079833984375,
      "learning_rate": 3.176e-05,
      "loss": 146.0306,
      "step": 19850
    },
    {
      "epoch": 0.08023691302011579,
      "grad_norm": 1048.5006103515625,
      "learning_rate": 3.1776e-05,
      "loss": 73.0426,
      "step": 19860
    },
    {
      "epoch": 0.08027731428548342,
      "grad_norm": 962.3606567382812,
      "learning_rate": 3.1792e-05,
      "loss": 88.1,
      "step": 19870
    },
    {
      "epoch": 0.08031771555085106,
      "grad_norm": 962.2769775390625,
      "learning_rate": 3.1808000000000004e-05,
      "loss": 144.6858,
      "step": 19880
    },
    {
      "epoch": 0.08035811681621868,
      "grad_norm": 517.8065185546875,
      "learning_rate": 3.1824e-05,
      "loss": 135.4217,
      "step": 19890
    },
    {
      "epoch": 0.08039851808158631,
      "grad_norm": 0.0,
      "learning_rate": 3.184000000000001e-05,
      "loss": 102.9744,
      "step": 19900
    },
    {
      "epoch": 0.08043891934695395,
      "grad_norm": 2071.30810546875,
      "learning_rate": 3.1856e-05,
      "loss": 223.0694,
      "step": 19910
    },
    {
      "epoch": 0.08047932061232158,
      "grad_norm": 726.2340087890625,
      "learning_rate": 3.1872e-05,
      "loss": 85.1156,
      "step": 19920
    },
    {
      "epoch": 0.08051972187768922,
      "grad_norm": 1996.00390625,
      "learning_rate": 3.1888000000000004e-05,
      "loss": 90.4812,
      "step": 19930
    },
    {
      "epoch": 0.08056012314305684,
      "grad_norm": 665.4238891601562,
      "learning_rate": 3.1904000000000005e-05,
      "loss": 82.7014,
      "step": 19940
    },
    {
      "epoch": 0.08060052440842447,
      "grad_norm": 724.19970703125,
      "learning_rate": 3.1920000000000006e-05,
      "loss": 107.1705,
      "step": 19950
    },
    {
      "epoch": 0.08064092567379211,
      "grad_norm": 1721.2724609375,
      "learning_rate": 3.1936e-05,
      "loss": 129.9797,
      "step": 19960
    },
    {
      "epoch": 0.08068132693915973,
      "grad_norm": 3038.216796875,
      "learning_rate": 3.1952e-05,
      "loss": 177.2808,
      "step": 19970
    },
    {
      "epoch": 0.08072172820452736,
      "grad_norm": 1314.772705078125,
      "learning_rate": 3.1968e-05,
      "loss": 113.6141,
      "step": 19980
    },
    {
      "epoch": 0.080762129469895,
      "grad_norm": 785.627685546875,
      "learning_rate": 3.1984000000000004e-05,
      "loss": 89.7999,
      "step": 19990
    },
    {
      "epoch": 0.08080253073526263,
      "grad_norm": 745.5064697265625,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 118.3001,
      "step": 20000
    },
    {
      "epoch": 0.08084293200063027,
      "grad_norm": 6274.64111328125,
      "learning_rate": 3.2016e-05,
      "loss": 188.7399,
      "step": 20010
    },
    {
      "epoch": 0.08088333326599789,
      "grad_norm": 613.6302490234375,
      "learning_rate": 3.2032e-05,
      "loss": 105.8049,
      "step": 20020
    },
    {
      "epoch": 0.08092373453136552,
      "grad_norm": 408.5493469238281,
      "learning_rate": 3.2048e-05,
      "loss": 114.5295,
      "step": 20030
    },
    {
      "epoch": 0.08096413579673316,
      "grad_norm": 1796.3682861328125,
      "learning_rate": 3.2064e-05,
      "loss": 145.9985,
      "step": 20040
    },
    {
      "epoch": 0.08100453706210078,
      "grad_norm": 750.31396484375,
      "learning_rate": 3.2080000000000005e-05,
      "loss": 119.3285,
      "step": 20050
    },
    {
      "epoch": 0.08104493832746841,
      "grad_norm": 664.3645629882812,
      "learning_rate": 3.2096000000000006e-05,
      "loss": 124.7819,
      "step": 20060
    },
    {
      "epoch": 0.08108533959283605,
      "grad_norm": 3907.92431640625,
      "learning_rate": 3.2112e-05,
      "loss": 184.4771,
      "step": 20070
    },
    {
      "epoch": 0.08112574085820368,
      "grad_norm": 1053.4244384765625,
      "learning_rate": 3.2128e-05,
      "loss": 169.1388,
      "step": 20080
    },
    {
      "epoch": 0.08116614212357132,
      "grad_norm": 1770.0511474609375,
      "learning_rate": 3.2144e-05,
      "loss": 109.763,
      "step": 20090
    },
    {
      "epoch": 0.08120654338893894,
      "grad_norm": 2079.87548828125,
      "learning_rate": 3.2160000000000004e-05,
      "loss": 118.6706,
      "step": 20100
    },
    {
      "epoch": 0.08124694465430657,
      "grad_norm": 1446.0218505859375,
      "learning_rate": 3.2176000000000005e-05,
      "loss": 129.1161,
      "step": 20110
    },
    {
      "epoch": 0.08128734591967421,
      "grad_norm": 944.9737548828125,
      "learning_rate": 3.2192e-05,
      "loss": 145.1025,
      "step": 20120
    },
    {
      "epoch": 0.08132774718504183,
      "grad_norm": 489.6378173828125,
      "learning_rate": 3.2208e-05,
      "loss": 143.1353,
      "step": 20130
    },
    {
      "epoch": 0.08136814845040946,
      "grad_norm": 4239.8681640625,
      "learning_rate": 3.2224e-05,
      "loss": 146.636,
      "step": 20140
    },
    {
      "epoch": 0.0814085497157771,
      "grad_norm": 375.4095153808594,
      "learning_rate": 3.224e-05,
      "loss": 119.1396,
      "step": 20150
    },
    {
      "epoch": 0.08144895098114473,
      "grad_norm": 678.9210205078125,
      "learning_rate": 3.2256000000000004e-05,
      "loss": 114.3637,
      "step": 20160
    },
    {
      "epoch": 0.08148935224651237,
      "grad_norm": 1916.13671875,
      "learning_rate": 3.2272e-05,
      "loss": 107.4577,
      "step": 20170
    },
    {
      "epoch": 0.08152975351188,
      "grad_norm": 1574.0010986328125,
      "learning_rate": 3.228800000000001e-05,
      "loss": 148.9322,
      "step": 20180
    },
    {
      "epoch": 0.08157015477724762,
      "grad_norm": 1359.7548828125,
      "learning_rate": 3.2304e-05,
      "loss": 115.4104,
      "step": 20190
    },
    {
      "epoch": 0.08161055604261526,
      "grad_norm": 838.4428100585938,
      "learning_rate": 3.232e-05,
      "loss": 116.7437,
      "step": 20200
    },
    {
      "epoch": 0.08165095730798289,
      "grad_norm": 653.6634521484375,
      "learning_rate": 3.2336000000000003e-05,
      "loss": 110.3881,
      "step": 20210
    },
    {
      "epoch": 0.08169135857335051,
      "grad_norm": 947.4215698242188,
      "learning_rate": 3.2352e-05,
      "loss": 116.1792,
      "step": 20220
    },
    {
      "epoch": 0.08173175983871815,
      "grad_norm": 3714.198974609375,
      "learning_rate": 3.2368000000000006e-05,
      "loss": 164.5668,
      "step": 20230
    },
    {
      "epoch": 0.08177216110408578,
      "grad_norm": 0.0,
      "learning_rate": 3.2384e-05,
      "loss": 127.3948,
      "step": 20240
    },
    {
      "epoch": 0.08181256236945342,
      "grad_norm": 899.4036865234375,
      "learning_rate": 3.24e-05,
      "loss": 109.6105,
      "step": 20250
    },
    {
      "epoch": 0.08185296363482104,
      "grad_norm": 1221.5169677734375,
      "learning_rate": 3.2416e-05,
      "loss": 108.7112,
      "step": 20260
    },
    {
      "epoch": 0.08189336490018867,
      "grad_norm": 945.672607421875,
      "learning_rate": 3.2432000000000004e-05,
      "loss": 110.9536,
      "step": 20270
    },
    {
      "epoch": 0.08193376616555631,
      "grad_norm": 659.3902587890625,
      "learning_rate": 3.2448000000000005e-05,
      "loss": 145.62,
      "step": 20280
    },
    {
      "epoch": 0.08197416743092394,
      "grad_norm": 936.4209594726562,
      "learning_rate": 3.2464e-05,
      "loss": 99.8831,
      "step": 20290
    },
    {
      "epoch": 0.08201456869629156,
      "grad_norm": 6973.3740234375,
      "learning_rate": 3.248000000000001e-05,
      "loss": 113.0091,
      "step": 20300
    },
    {
      "epoch": 0.0820549699616592,
      "grad_norm": 1618.539794921875,
      "learning_rate": 3.2496e-05,
      "loss": 175.2063,
      "step": 20310
    },
    {
      "epoch": 0.08209537122702683,
      "grad_norm": 1092.4254150390625,
      "learning_rate": 3.2512e-05,
      "loss": 95.937,
      "step": 20320
    },
    {
      "epoch": 0.08213577249239447,
      "grad_norm": 1008.83544921875,
      "learning_rate": 3.2528000000000004e-05,
      "loss": 123.3814,
      "step": 20330
    },
    {
      "epoch": 0.0821761737577621,
      "grad_norm": 811.6151123046875,
      "learning_rate": 3.2544000000000006e-05,
      "loss": 111.5735,
      "step": 20340
    },
    {
      "epoch": 0.08221657502312972,
      "grad_norm": 2305.8076171875,
      "learning_rate": 3.256e-05,
      "loss": 105.6014,
      "step": 20350
    },
    {
      "epoch": 0.08225697628849736,
      "grad_norm": 874.2246704101562,
      "learning_rate": 3.2576e-05,
      "loss": 84.1637,
      "step": 20360
    },
    {
      "epoch": 0.08229737755386499,
      "grad_norm": 5405.212890625,
      "learning_rate": 3.2592e-05,
      "loss": 155.2143,
      "step": 20370
    },
    {
      "epoch": 0.08233777881923261,
      "grad_norm": 2860.93994140625,
      "learning_rate": 3.2608000000000004e-05,
      "loss": 112.6364,
      "step": 20380
    },
    {
      "epoch": 0.08237818008460025,
      "grad_norm": 3501.953125,
      "learning_rate": 3.2624000000000005e-05,
      "loss": 117.0012,
      "step": 20390
    },
    {
      "epoch": 0.08241858134996788,
      "grad_norm": 786.2649536132812,
      "learning_rate": 3.264e-05,
      "loss": 118.425,
      "step": 20400
    },
    {
      "epoch": 0.08245898261533552,
      "grad_norm": 5292.2001953125,
      "learning_rate": 3.2656e-05,
      "loss": 100.0973,
      "step": 20410
    },
    {
      "epoch": 0.08249938388070314,
      "grad_norm": 1418.3062744140625,
      "learning_rate": 3.2672e-05,
      "loss": 155.9431,
      "step": 20420
    },
    {
      "epoch": 0.08253978514607077,
      "grad_norm": 905.533935546875,
      "learning_rate": 3.2688e-05,
      "loss": 98.626,
      "step": 20430
    },
    {
      "epoch": 0.08258018641143841,
      "grad_norm": 3069.26416015625,
      "learning_rate": 3.2704000000000004e-05,
      "loss": 118.547,
      "step": 20440
    },
    {
      "epoch": 0.08262058767680604,
      "grad_norm": 2428.871826171875,
      "learning_rate": 3.272e-05,
      "loss": 119.8318,
      "step": 20450
    },
    {
      "epoch": 0.08266098894217366,
      "grad_norm": 411.2120361328125,
      "learning_rate": 3.2736000000000006e-05,
      "loss": 114.0648,
      "step": 20460
    },
    {
      "epoch": 0.0827013902075413,
      "grad_norm": 1001.8212890625,
      "learning_rate": 3.2752e-05,
      "loss": 158.7806,
      "step": 20470
    },
    {
      "epoch": 0.08274179147290893,
      "grad_norm": 736.3692016601562,
      "learning_rate": 3.2768e-05,
      "loss": 97.0208,
      "step": 20480
    },
    {
      "epoch": 0.08278219273827657,
      "grad_norm": 1338.68994140625,
      "learning_rate": 3.2784e-05,
      "loss": 145.6164,
      "step": 20490
    },
    {
      "epoch": 0.0828225940036442,
      "grad_norm": 2277.1337890625,
      "learning_rate": 3.28e-05,
      "loss": 156.1081,
      "step": 20500
    },
    {
      "epoch": 0.08286299526901182,
      "grad_norm": 639.820556640625,
      "learning_rate": 3.2816000000000006e-05,
      "loss": 148.0952,
      "step": 20510
    },
    {
      "epoch": 0.08290339653437946,
      "grad_norm": 976.7066650390625,
      "learning_rate": 3.2832e-05,
      "loss": 88.4951,
      "step": 20520
    },
    {
      "epoch": 0.08294379779974709,
      "grad_norm": 920.2750244140625,
      "learning_rate": 3.2848e-05,
      "loss": 140.128,
      "step": 20530
    },
    {
      "epoch": 0.08298419906511471,
      "grad_norm": 604.2228393554688,
      "learning_rate": 3.2864e-05,
      "loss": 91.421,
      "step": 20540
    },
    {
      "epoch": 0.08302460033048235,
      "grad_norm": 1255.64013671875,
      "learning_rate": 3.2880000000000004e-05,
      "loss": 114.2171,
      "step": 20550
    },
    {
      "epoch": 0.08306500159584998,
      "grad_norm": 1299.9122314453125,
      "learning_rate": 3.2896000000000005e-05,
      "loss": 102.5171,
      "step": 20560
    },
    {
      "epoch": 0.08310540286121762,
      "grad_norm": 926.244384765625,
      "learning_rate": 3.2912e-05,
      "loss": 64.6635,
      "step": 20570
    },
    {
      "epoch": 0.08314580412658525,
      "grad_norm": 982.9434814453125,
      "learning_rate": 3.292800000000001e-05,
      "loss": 145.3727,
      "step": 20580
    },
    {
      "epoch": 0.08318620539195287,
      "grad_norm": 685.9315795898438,
      "learning_rate": 3.2944e-05,
      "loss": 137.4475,
      "step": 20590
    },
    {
      "epoch": 0.08322660665732051,
      "grad_norm": 750.7125244140625,
      "learning_rate": 3.296e-05,
      "loss": 162.0429,
      "step": 20600
    },
    {
      "epoch": 0.08326700792268814,
      "grad_norm": 1121.966064453125,
      "learning_rate": 3.2976000000000004e-05,
      "loss": 163.8441,
      "step": 20610
    },
    {
      "epoch": 0.08330740918805576,
      "grad_norm": 2098.244384765625,
      "learning_rate": 3.2992e-05,
      "loss": 140.6999,
      "step": 20620
    },
    {
      "epoch": 0.0833478104534234,
      "grad_norm": 1297.5299072265625,
      "learning_rate": 3.3008000000000007e-05,
      "loss": 161.9485,
      "step": 20630
    },
    {
      "epoch": 0.08338821171879103,
      "grad_norm": 764.8720703125,
      "learning_rate": 3.3024e-05,
      "loss": 131.4767,
      "step": 20640
    },
    {
      "epoch": 0.08342861298415867,
      "grad_norm": 3363.512939453125,
      "learning_rate": 3.304e-05,
      "loss": 101.7747,
      "step": 20650
    },
    {
      "epoch": 0.0834690142495263,
      "grad_norm": 4445.4267578125,
      "learning_rate": 3.3056e-05,
      "loss": 174.0515,
      "step": 20660
    },
    {
      "epoch": 0.08350941551489392,
      "grad_norm": 1621.59228515625,
      "learning_rate": 3.3072000000000005e-05,
      "loss": 162.773,
      "step": 20670
    },
    {
      "epoch": 0.08354981678026156,
      "grad_norm": 758.7988891601562,
      "learning_rate": 3.3088000000000006e-05,
      "loss": 110.7945,
      "step": 20680
    },
    {
      "epoch": 0.08359021804562919,
      "grad_norm": 1453.5654296875,
      "learning_rate": 3.3104e-05,
      "loss": 124.9009,
      "step": 20690
    },
    {
      "epoch": 0.08363061931099681,
      "grad_norm": 1057.9686279296875,
      "learning_rate": 3.312e-05,
      "loss": 149.3531,
      "step": 20700
    },
    {
      "epoch": 0.08367102057636445,
      "grad_norm": 805.1429443359375,
      "learning_rate": 3.3136e-05,
      "loss": 150.6182,
      "step": 20710
    },
    {
      "epoch": 0.08371142184173208,
      "grad_norm": 963.612060546875,
      "learning_rate": 3.3152000000000004e-05,
      "loss": 112.0466,
      "step": 20720
    },
    {
      "epoch": 0.08375182310709972,
      "grad_norm": 832.54296875,
      "learning_rate": 3.3168000000000005e-05,
      "loss": 135.5144,
      "step": 20730
    },
    {
      "epoch": 0.08379222437246735,
      "grad_norm": 8515.330078125,
      "learning_rate": 3.3184000000000006e-05,
      "loss": 197.9367,
      "step": 20740
    },
    {
      "epoch": 0.08383262563783497,
      "grad_norm": 5629.37890625,
      "learning_rate": 3.32e-05,
      "loss": 80.24,
      "step": 20750
    },
    {
      "epoch": 0.08387302690320261,
      "grad_norm": 1472.2364501953125,
      "learning_rate": 3.3216e-05,
      "loss": 125.4584,
      "step": 20760
    },
    {
      "epoch": 0.08391342816857024,
      "grad_norm": 5512.04736328125,
      "learning_rate": 3.3232e-05,
      "loss": 111.62,
      "step": 20770
    },
    {
      "epoch": 0.08395382943393787,
      "grad_norm": 759.9108276367188,
      "learning_rate": 3.3248000000000004e-05,
      "loss": 129.376,
      "step": 20780
    },
    {
      "epoch": 0.0839942306993055,
      "grad_norm": 670.1220092773438,
      "learning_rate": 3.3264000000000005e-05,
      "loss": 116.9635,
      "step": 20790
    },
    {
      "epoch": 0.08403463196467313,
      "grad_norm": 3641.0625,
      "learning_rate": 3.328e-05,
      "loss": 161.1426,
      "step": 20800
    },
    {
      "epoch": 0.08407503323004077,
      "grad_norm": 3545.739013671875,
      "learning_rate": 3.3296e-05,
      "loss": 140.4873,
      "step": 20810
    },
    {
      "epoch": 0.0841154344954084,
      "grad_norm": 1007.53564453125,
      "learning_rate": 3.3312e-05,
      "loss": 94.6542,
      "step": 20820
    },
    {
      "epoch": 0.08415583576077602,
      "grad_norm": 993.787109375,
      "learning_rate": 3.3328000000000003e-05,
      "loss": 132.4424,
      "step": 20830
    },
    {
      "epoch": 0.08419623702614366,
      "grad_norm": 1444.1688232421875,
      "learning_rate": 3.3344000000000005e-05,
      "loss": 132.11,
      "step": 20840
    },
    {
      "epoch": 0.08423663829151129,
      "grad_norm": 785.0534057617188,
      "learning_rate": 3.336e-05,
      "loss": 77.0269,
      "step": 20850
    },
    {
      "epoch": 0.08427703955687892,
      "grad_norm": 847.0296020507812,
      "learning_rate": 3.337600000000001e-05,
      "loss": 87.6466,
      "step": 20860
    },
    {
      "epoch": 0.08431744082224656,
      "grad_norm": 1261.2034912109375,
      "learning_rate": 3.3392e-05,
      "loss": 90.1927,
      "step": 20870
    },
    {
      "epoch": 0.08435784208761418,
      "grad_norm": 1592.902099609375,
      "learning_rate": 3.3408e-05,
      "loss": 159.0212,
      "step": 20880
    },
    {
      "epoch": 0.08439824335298182,
      "grad_norm": 933.5139770507812,
      "learning_rate": 3.3424000000000004e-05,
      "loss": 109.2687,
      "step": 20890
    },
    {
      "epoch": 0.08443864461834945,
      "grad_norm": 2483.051513671875,
      "learning_rate": 3.344e-05,
      "loss": 133.2125,
      "step": 20900
    },
    {
      "epoch": 0.08447904588371707,
      "grad_norm": 1393.5037841796875,
      "learning_rate": 3.3456000000000006e-05,
      "loss": 215.5028,
      "step": 20910
    },
    {
      "epoch": 0.08451944714908471,
      "grad_norm": 3210.046142578125,
      "learning_rate": 3.3472e-05,
      "loss": 112.5895,
      "step": 20920
    },
    {
      "epoch": 0.08455984841445234,
      "grad_norm": 867.4534301757812,
      "learning_rate": 3.3488e-05,
      "loss": 117.9855,
      "step": 20930
    },
    {
      "epoch": 0.08460024967981997,
      "grad_norm": 897.9375,
      "learning_rate": 3.3504e-05,
      "loss": 116.4767,
      "step": 20940
    },
    {
      "epoch": 0.0846406509451876,
      "grad_norm": 829.8159790039062,
      "learning_rate": 3.3520000000000004e-05,
      "loss": 117.4122,
      "step": 20950
    },
    {
      "epoch": 0.08468105221055523,
      "grad_norm": 496.8394470214844,
      "learning_rate": 3.3536000000000006e-05,
      "loss": 100.3099,
      "step": 20960
    },
    {
      "epoch": 0.08472145347592287,
      "grad_norm": 1295.52783203125,
      "learning_rate": 3.3552e-05,
      "loss": 130.5201,
      "step": 20970
    },
    {
      "epoch": 0.0847618547412905,
      "grad_norm": 1697.2413330078125,
      "learning_rate": 3.3568e-05,
      "loss": 81.13,
      "step": 20980
    },
    {
      "epoch": 0.08480225600665812,
      "grad_norm": 642.0286254882812,
      "learning_rate": 3.3584e-05,
      "loss": 124.2179,
      "step": 20990
    },
    {
      "epoch": 0.08484265727202576,
      "grad_norm": 825.5877075195312,
      "learning_rate": 3.3600000000000004e-05,
      "loss": 93.1462,
      "step": 21000
    },
    {
      "epoch": 0.08488305853739339,
      "grad_norm": 2915.330078125,
      "learning_rate": 3.3616000000000005e-05,
      "loss": 183.7277,
      "step": 21010
    },
    {
      "epoch": 0.08492345980276102,
      "grad_norm": 5314.201171875,
      "learning_rate": 3.3632e-05,
      "loss": 161.8029,
      "step": 21020
    },
    {
      "epoch": 0.08496386106812866,
      "grad_norm": 309.57415771484375,
      "learning_rate": 3.3648e-05,
      "loss": 97.7818,
      "step": 21030
    },
    {
      "epoch": 0.08500426233349628,
      "grad_norm": 727.4566040039062,
      "learning_rate": 3.3664e-05,
      "loss": 84.2195,
      "step": 21040
    },
    {
      "epoch": 0.08504466359886392,
      "grad_norm": 1173.5078125,
      "learning_rate": 3.368e-05,
      "loss": 110.4723,
      "step": 21050
    },
    {
      "epoch": 0.08508506486423155,
      "grad_norm": 1007.157958984375,
      "learning_rate": 3.3696000000000004e-05,
      "loss": 79.8195,
      "step": 21060
    },
    {
      "epoch": 0.08512546612959918,
      "grad_norm": 647.617431640625,
      "learning_rate": 3.3712000000000005e-05,
      "loss": 159.9282,
      "step": 21070
    },
    {
      "epoch": 0.08516586739496682,
      "grad_norm": 784.1207275390625,
      "learning_rate": 3.3728e-05,
      "loss": 120.1622,
      "step": 21080
    },
    {
      "epoch": 0.08520626866033444,
      "grad_norm": 5701.87451171875,
      "learning_rate": 3.3744e-05,
      "loss": 126.8908,
      "step": 21090
    },
    {
      "epoch": 0.08524666992570207,
      "grad_norm": 1457.8411865234375,
      "learning_rate": 3.376e-05,
      "loss": 111.7649,
      "step": 21100
    },
    {
      "epoch": 0.08528707119106971,
      "grad_norm": 1022.2379760742188,
      "learning_rate": 3.3776e-05,
      "loss": 116.9885,
      "step": 21110
    },
    {
      "epoch": 0.08532747245643733,
      "grad_norm": 665.683837890625,
      "learning_rate": 3.3792000000000004e-05,
      "loss": 124.5699,
      "step": 21120
    },
    {
      "epoch": 0.08536787372180497,
      "grad_norm": 565.2042846679688,
      "learning_rate": 3.3808e-05,
      "loss": 128.3419,
      "step": 21130
    },
    {
      "epoch": 0.0854082749871726,
      "grad_norm": 1426.2344970703125,
      "learning_rate": 3.382400000000001e-05,
      "loss": 169.1157,
      "step": 21140
    },
    {
      "epoch": 0.08544867625254023,
      "grad_norm": 1249.49267578125,
      "learning_rate": 3.384e-05,
      "loss": 109.0338,
      "step": 21150
    },
    {
      "epoch": 0.08548907751790787,
      "grad_norm": 712.49609375,
      "learning_rate": 3.3856e-05,
      "loss": 85.7494,
      "step": 21160
    },
    {
      "epoch": 0.08552947878327549,
      "grad_norm": 692.7472534179688,
      "learning_rate": 3.3872000000000004e-05,
      "loss": 111.5182,
      "step": 21170
    },
    {
      "epoch": 0.08556988004864312,
      "grad_norm": 441.89892578125,
      "learning_rate": 3.3888e-05,
      "loss": 99.3837,
      "step": 21180
    },
    {
      "epoch": 0.08561028131401076,
      "grad_norm": 945.43603515625,
      "learning_rate": 3.3904000000000006e-05,
      "loss": 138.1833,
      "step": 21190
    },
    {
      "epoch": 0.08565068257937838,
      "grad_norm": 1521.3837890625,
      "learning_rate": 3.392e-05,
      "loss": 123.8109,
      "step": 21200
    },
    {
      "epoch": 0.08569108384474602,
      "grad_norm": 2716.091796875,
      "learning_rate": 3.3936e-05,
      "loss": 176.2317,
      "step": 21210
    },
    {
      "epoch": 0.08573148511011365,
      "grad_norm": 2309.829833984375,
      "learning_rate": 3.3952e-05,
      "loss": 166.3631,
      "step": 21220
    },
    {
      "epoch": 0.08577188637548128,
      "grad_norm": 1705.5728759765625,
      "learning_rate": 3.3968000000000004e-05,
      "loss": 127.6594,
      "step": 21230
    },
    {
      "epoch": 0.08581228764084892,
      "grad_norm": 388.3826599121094,
      "learning_rate": 3.3984000000000005e-05,
      "loss": 91.3603,
      "step": 21240
    },
    {
      "epoch": 0.08585268890621654,
      "grad_norm": 1309.0146484375,
      "learning_rate": 3.4e-05,
      "loss": 95.5241,
      "step": 21250
    },
    {
      "epoch": 0.08589309017158417,
      "grad_norm": 1299.4666748046875,
      "learning_rate": 3.401600000000001e-05,
      "loss": 138.5965,
      "step": 21260
    },
    {
      "epoch": 0.08593349143695181,
      "grad_norm": 1387.5,
      "learning_rate": 3.4032e-05,
      "loss": 112.3464,
      "step": 21270
    },
    {
      "epoch": 0.08597389270231943,
      "grad_norm": 1501.5421142578125,
      "learning_rate": 3.4048e-05,
      "loss": 121.9169,
      "step": 21280
    },
    {
      "epoch": 0.08601429396768706,
      "grad_norm": 7527.328125,
      "learning_rate": 3.4064000000000005e-05,
      "loss": 171.4414,
      "step": 21290
    },
    {
      "epoch": 0.0860546952330547,
      "grad_norm": 991.0069580078125,
      "learning_rate": 3.408e-05,
      "loss": 113.5109,
      "step": 21300
    },
    {
      "epoch": 0.08609509649842233,
      "grad_norm": 1593.8912353515625,
      "learning_rate": 3.409600000000001e-05,
      "loss": 119.5967,
      "step": 21310
    },
    {
      "epoch": 0.08613549776378997,
      "grad_norm": 949.4871826171875,
      "learning_rate": 3.4112e-05,
      "loss": 102.3903,
      "step": 21320
    },
    {
      "epoch": 0.0861758990291576,
      "grad_norm": 753.6689453125,
      "learning_rate": 3.4128e-05,
      "loss": 103.7823,
      "step": 21330
    },
    {
      "epoch": 0.08621630029452522,
      "grad_norm": 1338.2713623046875,
      "learning_rate": 3.4144000000000004e-05,
      "loss": 130.913,
      "step": 21340
    },
    {
      "epoch": 0.08625670155989286,
      "grad_norm": 973.3551635742188,
      "learning_rate": 3.4160000000000005e-05,
      "loss": 106.8074,
      "step": 21350
    },
    {
      "epoch": 0.08629710282526049,
      "grad_norm": 916.3033447265625,
      "learning_rate": 3.4176000000000006e-05,
      "loss": 103.7848,
      "step": 21360
    },
    {
      "epoch": 0.08633750409062811,
      "grad_norm": 810.3303833007812,
      "learning_rate": 3.4192e-05,
      "loss": 105.5932,
      "step": 21370
    },
    {
      "epoch": 0.08637790535599575,
      "grad_norm": 1097.08203125,
      "learning_rate": 3.4208e-05,
      "loss": 100.8104,
      "step": 21380
    },
    {
      "epoch": 0.08641830662136338,
      "grad_norm": 4735.88671875,
      "learning_rate": 3.4224e-05,
      "loss": 127.2736,
      "step": 21390
    },
    {
      "epoch": 0.08645870788673102,
      "grad_norm": 1749.6431884765625,
      "learning_rate": 3.4240000000000004e-05,
      "loss": 163.8167,
      "step": 21400
    },
    {
      "epoch": 0.08649910915209864,
      "grad_norm": 730.6497802734375,
      "learning_rate": 3.4256000000000005e-05,
      "loss": 152.2738,
      "step": 21410
    },
    {
      "epoch": 0.08653951041746627,
      "grad_norm": 882.814208984375,
      "learning_rate": 3.4272e-05,
      "loss": 102.3541,
      "step": 21420
    },
    {
      "epoch": 0.08657991168283391,
      "grad_norm": 1553.72216796875,
      "learning_rate": 3.4288e-05,
      "loss": 101.4461,
      "step": 21430
    },
    {
      "epoch": 0.08662031294820154,
      "grad_norm": 1559.2188720703125,
      "learning_rate": 3.4304e-05,
      "loss": 153.7359,
      "step": 21440
    },
    {
      "epoch": 0.08666071421356916,
      "grad_norm": 684.1407470703125,
      "learning_rate": 3.4320000000000003e-05,
      "loss": 108.5924,
      "step": 21450
    },
    {
      "epoch": 0.0867011154789368,
      "grad_norm": 3555.82177734375,
      "learning_rate": 3.4336000000000005e-05,
      "loss": 123.4738,
      "step": 21460
    },
    {
      "epoch": 0.08674151674430443,
      "grad_norm": 1726.3648681640625,
      "learning_rate": 3.4352000000000006e-05,
      "loss": 167.0885,
      "step": 21470
    },
    {
      "epoch": 0.08678191800967207,
      "grad_norm": 1249.389404296875,
      "learning_rate": 3.4368e-05,
      "loss": 67.5671,
      "step": 21480
    },
    {
      "epoch": 0.0868223192750397,
      "grad_norm": 1911.92431640625,
      "learning_rate": 3.4384e-05,
      "loss": 163.4095,
      "step": 21490
    },
    {
      "epoch": 0.08686272054040732,
      "grad_norm": 1018.9912109375,
      "learning_rate": 3.44e-05,
      "loss": 91.9574,
      "step": 21500
    },
    {
      "epoch": 0.08690312180577496,
      "grad_norm": 675.4153442382812,
      "learning_rate": 3.4416000000000004e-05,
      "loss": 121.4217,
      "step": 21510
    },
    {
      "epoch": 0.08694352307114259,
      "grad_norm": 1455.37451171875,
      "learning_rate": 3.4432000000000005e-05,
      "loss": 106.3664,
      "step": 21520
    },
    {
      "epoch": 0.08698392433651021,
      "grad_norm": 1519.950927734375,
      "learning_rate": 3.4448e-05,
      "loss": 118.7567,
      "step": 21530
    },
    {
      "epoch": 0.08702432560187785,
      "grad_norm": 1685.112548828125,
      "learning_rate": 3.446400000000001e-05,
      "loss": 80.7093,
      "step": 21540
    },
    {
      "epoch": 0.08706472686724548,
      "grad_norm": 848.3253784179688,
      "learning_rate": 3.448e-05,
      "loss": 123.2339,
      "step": 21550
    },
    {
      "epoch": 0.08710512813261312,
      "grad_norm": 3590.259765625,
      "learning_rate": 3.4496e-05,
      "loss": 131.1099,
      "step": 21560
    },
    {
      "epoch": 0.08714552939798074,
      "grad_norm": 1122.024169921875,
      "learning_rate": 3.4512000000000004e-05,
      "loss": 78.0208,
      "step": 21570
    },
    {
      "epoch": 0.08718593066334837,
      "grad_norm": 483.8023376464844,
      "learning_rate": 3.4528e-05,
      "loss": 119.6661,
      "step": 21580
    },
    {
      "epoch": 0.08722633192871601,
      "grad_norm": 793.3316650390625,
      "learning_rate": 3.454400000000001e-05,
      "loss": 118.78,
      "step": 21590
    },
    {
      "epoch": 0.08726673319408364,
      "grad_norm": 1123.4295654296875,
      "learning_rate": 3.456e-05,
      "loss": 133.3474,
      "step": 21600
    },
    {
      "epoch": 0.08730713445945126,
      "grad_norm": 718.1134033203125,
      "learning_rate": 3.4576e-05,
      "loss": 96.3448,
      "step": 21610
    },
    {
      "epoch": 0.0873475357248189,
      "grad_norm": 997.1589965820312,
      "learning_rate": 3.4592000000000004e-05,
      "loss": 106.7913,
      "step": 21620
    },
    {
      "epoch": 0.08738793699018653,
      "grad_norm": 533.8572998046875,
      "learning_rate": 3.4608000000000005e-05,
      "loss": 175.2727,
      "step": 21630
    },
    {
      "epoch": 0.08742833825555417,
      "grad_norm": 1356.1878662109375,
      "learning_rate": 3.4624000000000006e-05,
      "loss": 93.2392,
      "step": 21640
    },
    {
      "epoch": 0.0874687395209218,
      "grad_norm": 1554.261962890625,
      "learning_rate": 3.464e-05,
      "loss": 122.4991,
      "step": 21650
    },
    {
      "epoch": 0.08750914078628942,
      "grad_norm": 905.5225219726562,
      "learning_rate": 3.4656e-05,
      "loss": 106.4889,
      "step": 21660
    },
    {
      "epoch": 0.08754954205165706,
      "grad_norm": 484.3202209472656,
      "learning_rate": 3.4672e-05,
      "loss": 119.5927,
      "step": 21670
    },
    {
      "epoch": 0.08758994331702469,
      "grad_norm": 1324.7713623046875,
      "learning_rate": 3.4688000000000004e-05,
      "loss": 88.9022,
      "step": 21680
    },
    {
      "epoch": 0.08763034458239231,
      "grad_norm": 863.8072509765625,
      "learning_rate": 3.4704000000000005e-05,
      "loss": 123.0851,
      "step": 21690
    },
    {
      "epoch": 0.08767074584775995,
      "grad_norm": 345.5253601074219,
      "learning_rate": 3.472e-05,
      "loss": 139.7284,
      "step": 21700
    },
    {
      "epoch": 0.08771114711312758,
      "grad_norm": 1755.52001953125,
      "learning_rate": 3.4736e-05,
      "loss": 127.2942,
      "step": 21710
    },
    {
      "epoch": 0.08775154837849522,
      "grad_norm": 959.5047607421875,
      "learning_rate": 3.4752e-05,
      "loss": 115.2856,
      "step": 21720
    },
    {
      "epoch": 0.08779194964386285,
      "grad_norm": 599.0006103515625,
      "learning_rate": 3.4768e-05,
      "loss": 86.4051,
      "step": 21730
    },
    {
      "epoch": 0.08783235090923047,
      "grad_norm": 1719.8614501953125,
      "learning_rate": 3.4784000000000004e-05,
      "loss": 84.124,
      "step": 21740
    },
    {
      "epoch": 0.08787275217459811,
      "grad_norm": 744.1102905273438,
      "learning_rate": 3.4800000000000006e-05,
      "loss": 106.3017,
      "step": 21750
    },
    {
      "epoch": 0.08791315343996574,
      "grad_norm": 1090.3604736328125,
      "learning_rate": 3.4816e-05,
      "loss": 81.3375,
      "step": 21760
    },
    {
      "epoch": 0.08795355470533336,
      "grad_norm": 570.5980834960938,
      "learning_rate": 3.4832e-05,
      "loss": 100.2312,
      "step": 21770
    },
    {
      "epoch": 0.087993955970701,
      "grad_norm": 591.20947265625,
      "learning_rate": 3.4848e-05,
      "loss": 74.5655,
      "step": 21780
    },
    {
      "epoch": 0.08803435723606863,
      "grad_norm": 675.690185546875,
      "learning_rate": 3.4864000000000004e-05,
      "loss": 120.0532,
      "step": 21790
    },
    {
      "epoch": 0.08807475850143627,
      "grad_norm": 1083.811767578125,
      "learning_rate": 3.4880000000000005e-05,
      "loss": 161.6993,
      "step": 21800
    },
    {
      "epoch": 0.0881151597668039,
      "grad_norm": 1041.786376953125,
      "learning_rate": 3.4896e-05,
      "loss": 131.3204,
      "step": 21810
    },
    {
      "epoch": 0.08815556103217152,
      "grad_norm": 1283.564453125,
      "learning_rate": 3.4912e-05,
      "loss": 79.6838,
      "step": 21820
    },
    {
      "epoch": 0.08819596229753916,
      "grad_norm": 2376.157958984375,
      "learning_rate": 3.4928e-05,
      "loss": 119.8821,
      "step": 21830
    },
    {
      "epoch": 0.08823636356290679,
      "grad_norm": 623.5429077148438,
      "learning_rate": 3.4944e-05,
      "loss": 116.1281,
      "step": 21840
    },
    {
      "epoch": 0.08827676482827441,
      "grad_norm": 770.1627807617188,
      "learning_rate": 3.4960000000000004e-05,
      "loss": 60.8863,
      "step": 21850
    },
    {
      "epoch": 0.08831716609364205,
      "grad_norm": 1002.646240234375,
      "learning_rate": 3.4976e-05,
      "loss": 113.2037,
      "step": 21860
    },
    {
      "epoch": 0.08835756735900968,
      "grad_norm": 1004.6346435546875,
      "learning_rate": 3.4992000000000006e-05,
      "loss": 110.553,
      "step": 21870
    },
    {
      "epoch": 0.08839796862437732,
      "grad_norm": 1741.873046875,
      "learning_rate": 3.5008e-05,
      "loss": 111.8646,
      "step": 21880
    },
    {
      "epoch": 0.08843836988974495,
      "grad_norm": 1661.6435546875,
      "learning_rate": 3.5024e-05,
      "loss": 86.3073,
      "step": 21890
    },
    {
      "epoch": 0.08847877115511257,
      "grad_norm": 940.521728515625,
      "learning_rate": 3.504e-05,
      "loss": 89.9938,
      "step": 21900
    },
    {
      "epoch": 0.08851917242048021,
      "grad_norm": 1818.4632568359375,
      "learning_rate": 3.5056e-05,
      "loss": 138.901,
      "step": 21910
    },
    {
      "epoch": 0.08855957368584784,
      "grad_norm": 4558.5595703125,
      "learning_rate": 3.5072000000000006e-05,
      "loss": 95.3903,
      "step": 21920
    },
    {
      "epoch": 0.08859997495121547,
      "grad_norm": 2346.676025390625,
      "learning_rate": 3.5088e-05,
      "loss": 102.4624,
      "step": 21930
    },
    {
      "epoch": 0.0886403762165831,
      "grad_norm": 784.5299682617188,
      "learning_rate": 3.5104e-05,
      "loss": 110.0189,
      "step": 21940
    },
    {
      "epoch": 0.08868077748195073,
      "grad_norm": 2423.447265625,
      "learning_rate": 3.512e-05,
      "loss": 102.8573,
      "step": 21950
    },
    {
      "epoch": 0.08872117874731837,
      "grad_norm": 1321.034912109375,
      "learning_rate": 3.5136000000000004e-05,
      "loss": 162.9776,
      "step": 21960
    },
    {
      "epoch": 0.088761580012686,
      "grad_norm": 644.175048828125,
      "learning_rate": 3.5152000000000005e-05,
      "loss": 151.2046,
      "step": 21970
    },
    {
      "epoch": 0.08880198127805362,
      "grad_norm": 769.9696044921875,
      "learning_rate": 3.5168e-05,
      "loss": 102.919,
      "step": 21980
    },
    {
      "epoch": 0.08884238254342126,
      "grad_norm": 0.0,
      "learning_rate": 3.518400000000001e-05,
      "loss": 84.6088,
      "step": 21990
    },
    {
      "epoch": 0.08888278380878889,
      "grad_norm": 617.4421997070312,
      "learning_rate": 3.52e-05,
      "loss": 109.1384,
      "step": 22000
    },
    {
      "epoch": 0.08892318507415652,
      "grad_norm": 853.436279296875,
      "learning_rate": 3.5216e-05,
      "loss": 104.1218,
      "step": 22010
    },
    {
      "epoch": 0.08896358633952416,
      "grad_norm": 4421.9814453125,
      "learning_rate": 3.5232000000000004e-05,
      "loss": 65.627,
      "step": 22020
    },
    {
      "epoch": 0.08900398760489178,
      "grad_norm": 1764.05224609375,
      "learning_rate": 3.5248000000000005e-05,
      "loss": 142.3967,
      "step": 22030
    },
    {
      "epoch": 0.08904438887025942,
      "grad_norm": 1308.7003173828125,
      "learning_rate": 3.5264000000000007e-05,
      "loss": 102.8396,
      "step": 22040
    },
    {
      "epoch": 0.08908479013562705,
      "grad_norm": 1276.386474609375,
      "learning_rate": 3.528e-05,
      "loss": 83.7799,
      "step": 22050
    },
    {
      "epoch": 0.08912519140099467,
      "grad_norm": 1775.8756103515625,
      "learning_rate": 3.5296e-05,
      "loss": 130.5492,
      "step": 22060
    },
    {
      "epoch": 0.08916559266636231,
      "grad_norm": 742.988525390625,
      "learning_rate": 3.5312000000000003e-05,
      "loss": 103.372,
      "step": 22070
    },
    {
      "epoch": 0.08920599393172994,
      "grad_norm": 4215.49609375,
      "learning_rate": 3.5328000000000005e-05,
      "loss": 142.1393,
      "step": 22080
    },
    {
      "epoch": 0.08924639519709757,
      "grad_norm": 2923.902099609375,
      "learning_rate": 3.5344000000000006e-05,
      "loss": 157.1858,
      "step": 22090
    },
    {
      "epoch": 0.0892867964624652,
      "grad_norm": 1210.239501953125,
      "learning_rate": 3.536e-05,
      "loss": 130.1175,
      "step": 22100
    },
    {
      "epoch": 0.08932719772783283,
      "grad_norm": 2710.9931640625,
      "learning_rate": 3.5376e-05,
      "loss": 128.473,
      "step": 22110
    },
    {
      "epoch": 0.08936759899320047,
      "grad_norm": 1313.605712890625,
      "learning_rate": 3.5392e-05,
      "loss": 105.5896,
      "step": 22120
    },
    {
      "epoch": 0.0894080002585681,
      "grad_norm": 1979.692626953125,
      "learning_rate": 3.5408000000000004e-05,
      "loss": 122.7433,
      "step": 22130
    },
    {
      "epoch": 0.08944840152393572,
      "grad_norm": 878.58642578125,
      "learning_rate": 3.5424000000000005e-05,
      "loss": 97.7734,
      "step": 22140
    },
    {
      "epoch": 0.08948880278930336,
      "grad_norm": 1813.173828125,
      "learning_rate": 3.5440000000000006e-05,
      "loss": 125.1185,
      "step": 22150
    },
    {
      "epoch": 0.08952920405467099,
      "grad_norm": 1052.124267578125,
      "learning_rate": 3.5456e-05,
      "loss": 80.2057,
      "step": 22160
    },
    {
      "epoch": 0.08956960532003862,
      "grad_norm": 1051.0555419921875,
      "learning_rate": 3.5472e-05,
      "loss": 117.2363,
      "step": 22170
    },
    {
      "epoch": 0.08961000658540626,
      "grad_norm": 567.1451416015625,
      "learning_rate": 3.5488e-05,
      "loss": 90.6433,
      "step": 22180
    },
    {
      "epoch": 0.08965040785077388,
      "grad_norm": 1285.5130615234375,
      "learning_rate": 3.5504e-05,
      "loss": 114.9006,
      "step": 22190
    },
    {
      "epoch": 0.08969080911614152,
      "grad_norm": 1761.4788818359375,
      "learning_rate": 3.5520000000000006e-05,
      "loss": 107.4548,
      "step": 22200
    },
    {
      "epoch": 0.08973121038150915,
      "grad_norm": 922.4437866210938,
      "learning_rate": 3.5536e-05,
      "loss": 118.1805,
      "step": 22210
    },
    {
      "epoch": 0.08977161164687678,
      "grad_norm": 1029.1673583984375,
      "learning_rate": 3.5552e-05,
      "loss": 168.9282,
      "step": 22220
    },
    {
      "epoch": 0.08981201291224442,
      "grad_norm": 438.4372863769531,
      "learning_rate": 3.5568e-05,
      "loss": 81.5482,
      "step": 22230
    },
    {
      "epoch": 0.08985241417761204,
      "grad_norm": 587.6930541992188,
      "learning_rate": 3.5584000000000004e-05,
      "loss": 147.9191,
      "step": 22240
    },
    {
      "epoch": 0.08989281544297967,
      "grad_norm": 1436.885498046875,
      "learning_rate": 3.5600000000000005e-05,
      "loss": 141.406,
      "step": 22250
    },
    {
      "epoch": 0.08993321670834731,
      "grad_norm": 835.5823974609375,
      "learning_rate": 3.5616e-05,
      "loss": 103.0261,
      "step": 22260
    },
    {
      "epoch": 0.08997361797371493,
      "grad_norm": 604.3268432617188,
      "learning_rate": 3.563200000000001e-05,
      "loss": 114.0826,
      "step": 22270
    },
    {
      "epoch": 0.09001401923908257,
      "grad_norm": 694.15869140625,
      "learning_rate": 3.5648e-05,
      "loss": 81.4152,
      "step": 22280
    },
    {
      "epoch": 0.0900544205044502,
      "grad_norm": 412.0992736816406,
      "learning_rate": 3.5664e-05,
      "loss": 139.5334,
      "step": 22290
    },
    {
      "epoch": 0.09009482176981783,
      "grad_norm": 1818.29296875,
      "learning_rate": 3.5680000000000004e-05,
      "loss": 155.0708,
      "step": 22300
    },
    {
      "epoch": 0.09013522303518547,
      "grad_norm": 635.8440551757812,
      "learning_rate": 3.5696e-05,
      "loss": 100.3289,
      "step": 22310
    },
    {
      "epoch": 0.09017562430055309,
      "grad_norm": 358.85845947265625,
      "learning_rate": 3.5712000000000006e-05,
      "loss": 153.7336,
      "step": 22320
    },
    {
      "epoch": 0.09021602556592072,
      "grad_norm": 1078.0045166015625,
      "learning_rate": 3.5728e-05,
      "loss": 104.7138,
      "step": 22330
    },
    {
      "epoch": 0.09025642683128836,
      "grad_norm": 2302.094482421875,
      "learning_rate": 3.5744e-05,
      "loss": 104.76,
      "step": 22340
    },
    {
      "epoch": 0.09029682809665598,
      "grad_norm": 545.4147338867188,
      "learning_rate": 3.576e-05,
      "loss": 107.9992,
      "step": 22350
    },
    {
      "epoch": 0.09033722936202362,
      "grad_norm": 1128.086669921875,
      "learning_rate": 3.5776000000000004e-05,
      "loss": 95.0454,
      "step": 22360
    },
    {
      "epoch": 0.09037763062739125,
      "grad_norm": 1193.4002685546875,
      "learning_rate": 3.5792000000000006e-05,
      "loss": 112.622,
      "step": 22370
    },
    {
      "epoch": 0.09041803189275888,
      "grad_norm": 939.8592529296875,
      "learning_rate": 3.5808e-05,
      "loss": 104.9512,
      "step": 22380
    },
    {
      "epoch": 0.09045843315812652,
      "grad_norm": 720.6575317382812,
      "learning_rate": 3.5824e-05,
      "loss": 139.2956,
      "step": 22390
    },
    {
      "epoch": 0.09049883442349414,
      "grad_norm": 941.1958618164062,
      "learning_rate": 3.584e-05,
      "loss": 99.4069,
      "step": 22400
    },
    {
      "epoch": 0.09053923568886177,
      "grad_norm": 1065.10498046875,
      "learning_rate": 3.5856000000000004e-05,
      "loss": 107.6034,
      "step": 22410
    },
    {
      "epoch": 0.09057963695422941,
      "grad_norm": 668.1845092773438,
      "learning_rate": 3.5872000000000005e-05,
      "loss": 113.4124,
      "step": 22420
    },
    {
      "epoch": 0.09062003821959703,
      "grad_norm": 792.0282592773438,
      "learning_rate": 3.5888000000000006e-05,
      "loss": 110.8783,
      "step": 22430
    },
    {
      "epoch": 0.09066043948496467,
      "grad_norm": 0.0,
      "learning_rate": 3.5904e-05,
      "loss": 85.3758,
      "step": 22440
    },
    {
      "epoch": 0.0907008407503323,
      "grad_norm": 1627.4656982421875,
      "learning_rate": 3.592e-05,
      "loss": 135.8512,
      "step": 22450
    },
    {
      "epoch": 0.09074124201569993,
      "grad_norm": 17669.587890625,
      "learning_rate": 3.5936e-05,
      "loss": 167.7172,
      "step": 22460
    },
    {
      "epoch": 0.09078164328106757,
      "grad_norm": 830.4048461914062,
      "learning_rate": 3.5952000000000004e-05,
      "loss": 128.9894,
      "step": 22470
    },
    {
      "epoch": 0.0908220445464352,
      "grad_norm": 609.1981811523438,
      "learning_rate": 3.5968000000000005e-05,
      "loss": 131.7272,
      "step": 22480
    },
    {
      "epoch": 0.09086244581180282,
      "grad_norm": 829.3265991210938,
      "learning_rate": 3.5984e-05,
      "loss": 121.7255,
      "step": 22490
    },
    {
      "epoch": 0.09090284707717046,
      "grad_norm": 1340.07958984375,
      "learning_rate": 3.6e-05,
      "loss": 79.694,
      "step": 22500
    },
    {
      "epoch": 0.09094324834253809,
      "grad_norm": 1699.5284423828125,
      "learning_rate": 3.6016e-05,
      "loss": 96.8572,
      "step": 22510
    },
    {
      "epoch": 0.09098364960790573,
      "grad_norm": 1204.338623046875,
      "learning_rate": 3.6032e-05,
      "loss": 91.7086,
      "step": 22520
    },
    {
      "epoch": 0.09102405087327335,
      "grad_norm": 1220.1192626953125,
      "learning_rate": 3.6048000000000005e-05,
      "loss": 92.4644,
      "step": 22530
    },
    {
      "epoch": 0.09106445213864098,
      "grad_norm": 1748.9383544921875,
      "learning_rate": 3.6064e-05,
      "loss": 150.2315,
      "step": 22540
    },
    {
      "epoch": 0.09110485340400862,
      "grad_norm": 1452.05078125,
      "learning_rate": 3.608000000000001e-05,
      "loss": 85.5033,
      "step": 22550
    },
    {
      "epoch": 0.09114525466937624,
      "grad_norm": 678.501953125,
      "learning_rate": 3.6096e-05,
      "loss": 130.9142,
      "step": 22560
    },
    {
      "epoch": 0.09118565593474387,
      "grad_norm": 565.4130859375,
      "learning_rate": 3.6112e-05,
      "loss": 150.4763,
      "step": 22570
    },
    {
      "epoch": 0.09122605720011151,
      "grad_norm": 769.1754760742188,
      "learning_rate": 3.6128000000000004e-05,
      "loss": 125.7674,
      "step": 22580
    },
    {
      "epoch": 0.09126645846547914,
      "grad_norm": 881.0234375,
      "learning_rate": 3.6144e-05,
      "loss": 83.7712,
      "step": 22590
    },
    {
      "epoch": 0.09130685973084678,
      "grad_norm": 622.296142578125,
      "learning_rate": 3.6160000000000006e-05,
      "loss": 148.5823,
      "step": 22600
    },
    {
      "epoch": 0.0913472609962144,
      "grad_norm": 983.2738647460938,
      "learning_rate": 3.6176e-05,
      "loss": 107.8446,
      "step": 22610
    },
    {
      "epoch": 0.09138766226158203,
      "grad_norm": 887.962890625,
      "learning_rate": 3.6192e-05,
      "loss": 121.2446,
      "step": 22620
    },
    {
      "epoch": 0.09142806352694967,
      "grad_norm": 903.2112426757812,
      "learning_rate": 3.6208e-05,
      "loss": 122.9406,
      "step": 22630
    },
    {
      "epoch": 0.0914684647923173,
      "grad_norm": 401.5851135253906,
      "learning_rate": 3.6224000000000004e-05,
      "loss": 99.0492,
      "step": 22640
    },
    {
      "epoch": 0.09150886605768492,
      "grad_norm": 553.1171875,
      "learning_rate": 3.6240000000000005e-05,
      "loss": 122.498,
      "step": 22650
    },
    {
      "epoch": 0.09154926732305256,
      "grad_norm": 4576.8271484375,
      "learning_rate": 3.6256e-05,
      "loss": 138.7712,
      "step": 22660
    },
    {
      "epoch": 0.09158966858842019,
      "grad_norm": 603.9403686523438,
      "learning_rate": 3.627200000000001e-05,
      "loss": 124.1618,
      "step": 22670
    },
    {
      "epoch": 0.09163006985378783,
      "grad_norm": 1222.1373291015625,
      "learning_rate": 3.6288e-05,
      "loss": 108.6959,
      "step": 22680
    },
    {
      "epoch": 0.09167047111915545,
      "grad_norm": 766.6356201171875,
      "learning_rate": 3.6304000000000003e-05,
      "loss": 90.1852,
      "step": 22690
    },
    {
      "epoch": 0.09171087238452308,
      "grad_norm": 777.71240234375,
      "learning_rate": 3.6320000000000005e-05,
      "loss": 137.761,
      "step": 22700
    },
    {
      "epoch": 0.09175127364989072,
      "grad_norm": 0.0,
      "learning_rate": 3.6336e-05,
      "loss": 102.6547,
      "step": 22710
    },
    {
      "epoch": 0.09179167491525834,
      "grad_norm": 566.0398559570312,
      "learning_rate": 3.635200000000001e-05,
      "loss": 85.4628,
      "step": 22720
    },
    {
      "epoch": 0.09183207618062597,
      "grad_norm": 1419.7137451171875,
      "learning_rate": 3.6368e-05,
      "loss": 137.7339,
      "step": 22730
    },
    {
      "epoch": 0.09187247744599361,
      "grad_norm": 851.448486328125,
      "learning_rate": 3.6384e-05,
      "loss": 98.0729,
      "step": 22740
    },
    {
      "epoch": 0.09191287871136124,
      "grad_norm": 724.6092529296875,
      "learning_rate": 3.6400000000000004e-05,
      "loss": 54.0254,
      "step": 22750
    },
    {
      "epoch": 0.09195327997672888,
      "grad_norm": 1501.339599609375,
      "learning_rate": 3.6416000000000005e-05,
      "loss": 153.3385,
      "step": 22760
    },
    {
      "epoch": 0.0919936812420965,
      "grad_norm": 905.32177734375,
      "learning_rate": 3.6432000000000006e-05,
      "loss": 117.4171,
      "step": 22770
    },
    {
      "epoch": 0.09203408250746413,
      "grad_norm": 2430.184326171875,
      "learning_rate": 3.6448e-05,
      "loss": 111.6443,
      "step": 22780
    },
    {
      "epoch": 0.09207448377283177,
      "grad_norm": 992.3556518554688,
      "learning_rate": 3.6464e-05,
      "loss": 71.0798,
      "step": 22790
    },
    {
      "epoch": 0.0921148850381994,
      "grad_norm": 1678.62744140625,
      "learning_rate": 3.648e-05,
      "loss": 109.5951,
      "step": 22800
    },
    {
      "epoch": 0.09215528630356702,
      "grad_norm": 1451.3145751953125,
      "learning_rate": 3.6496000000000004e-05,
      "loss": 149.5164,
      "step": 22810
    },
    {
      "epoch": 0.09219568756893466,
      "grad_norm": 3507.716552734375,
      "learning_rate": 3.6512e-05,
      "loss": 90.6843,
      "step": 22820
    },
    {
      "epoch": 0.09223608883430229,
      "grad_norm": 825.1586303710938,
      "learning_rate": 3.652800000000001e-05,
      "loss": 92.9351,
      "step": 22830
    },
    {
      "epoch": 0.09227649009966993,
      "grad_norm": 1350.14306640625,
      "learning_rate": 3.6544e-05,
      "loss": 131.2104,
      "step": 22840
    },
    {
      "epoch": 0.09231689136503755,
      "grad_norm": 558.4495849609375,
      "learning_rate": 3.656e-05,
      "loss": 106.5891,
      "step": 22850
    },
    {
      "epoch": 0.09235729263040518,
      "grad_norm": 1416.712158203125,
      "learning_rate": 3.6576000000000004e-05,
      "loss": 134.2764,
      "step": 22860
    },
    {
      "epoch": 0.09239769389577282,
      "grad_norm": 883.6400756835938,
      "learning_rate": 3.6592e-05,
      "loss": 94.5502,
      "step": 22870
    },
    {
      "epoch": 0.09243809516114045,
      "grad_norm": 589.9857788085938,
      "learning_rate": 3.6608000000000006e-05,
      "loss": 65.831,
      "step": 22880
    },
    {
      "epoch": 0.09247849642650807,
      "grad_norm": 2002.652587890625,
      "learning_rate": 3.6624e-05,
      "loss": 136.2,
      "step": 22890
    },
    {
      "epoch": 0.09251889769187571,
      "grad_norm": 753.989501953125,
      "learning_rate": 3.664e-05,
      "loss": 104.6583,
      "step": 22900
    },
    {
      "epoch": 0.09255929895724334,
      "grad_norm": 615.0721435546875,
      "learning_rate": 3.6656e-05,
      "loss": 146.1161,
      "step": 22910
    },
    {
      "epoch": 0.09259970022261098,
      "grad_norm": 1668.8343505859375,
      "learning_rate": 3.6672000000000004e-05,
      "loss": 139.9676,
      "step": 22920
    },
    {
      "epoch": 0.0926401014879786,
      "grad_norm": 883.7720336914062,
      "learning_rate": 3.6688000000000005e-05,
      "loss": 132.2069,
      "step": 22930
    },
    {
      "epoch": 0.09268050275334623,
      "grad_norm": 5342.50244140625,
      "learning_rate": 3.6704e-05,
      "loss": 98.5303,
      "step": 22940
    },
    {
      "epoch": 0.09272090401871387,
      "grad_norm": 888.3287963867188,
      "learning_rate": 3.672000000000001e-05,
      "loss": 97.6094,
      "step": 22950
    },
    {
      "epoch": 0.0927613052840815,
      "grad_norm": 519.9688110351562,
      "learning_rate": 3.6736e-05,
      "loss": 110.704,
      "step": 22960
    },
    {
      "epoch": 0.09280170654944912,
      "grad_norm": 851.119140625,
      "learning_rate": 3.6752e-05,
      "loss": 122.4055,
      "step": 22970
    },
    {
      "epoch": 0.09284210781481676,
      "grad_norm": 1095.8951416015625,
      "learning_rate": 3.6768000000000004e-05,
      "loss": 85.4359,
      "step": 22980
    },
    {
      "epoch": 0.09288250908018439,
      "grad_norm": 1443.4359130859375,
      "learning_rate": 3.6784e-05,
      "loss": 105.7661,
      "step": 22990
    },
    {
      "epoch": 0.09292291034555203,
      "grad_norm": 1374.320068359375,
      "learning_rate": 3.680000000000001e-05,
      "loss": 84.9598,
      "step": 23000
    },
    {
      "epoch": 0.09296331161091966,
      "grad_norm": 1480.087890625,
      "learning_rate": 3.6816e-05,
      "loss": 86.5958,
      "step": 23010
    },
    {
      "epoch": 0.09300371287628728,
      "grad_norm": 1209.684326171875,
      "learning_rate": 3.6832e-05,
      "loss": 67.3631,
      "step": 23020
    },
    {
      "epoch": 0.09304411414165492,
      "grad_norm": 815.2809448242188,
      "learning_rate": 3.6848000000000004e-05,
      "loss": 140.1031,
      "step": 23030
    },
    {
      "epoch": 0.09308451540702255,
      "grad_norm": 1305.93994140625,
      "learning_rate": 3.6864000000000005e-05,
      "loss": 116.6758,
      "step": 23040
    },
    {
      "epoch": 0.09312491667239017,
      "grad_norm": 599.248046875,
      "learning_rate": 3.6880000000000006e-05,
      "loss": 282.2803,
      "step": 23050
    },
    {
      "epoch": 0.09316531793775781,
      "grad_norm": 695.9918823242188,
      "learning_rate": 3.6896e-05,
      "loss": 99.9409,
      "step": 23060
    },
    {
      "epoch": 0.09320571920312544,
      "grad_norm": 4307.1064453125,
      "learning_rate": 3.6912e-05,
      "loss": 114.5274,
      "step": 23070
    },
    {
      "epoch": 0.09324612046849308,
      "grad_norm": 1717.5181884765625,
      "learning_rate": 3.6928e-05,
      "loss": 120.5985,
      "step": 23080
    },
    {
      "epoch": 0.0932865217338607,
      "grad_norm": 1048.98095703125,
      "learning_rate": 3.6944000000000004e-05,
      "loss": 144.147,
      "step": 23090
    },
    {
      "epoch": 0.09332692299922833,
      "grad_norm": 451.38031005859375,
      "learning_rate": 3.6960000000000005e-05,
      "loss": 113.492,
      "step": 23100
    },
    {
      "epoch": 0.09336732426459597,
      "grad_norm": 307.1217346191406,
      "learning_rate": 3.6976e-05,
      "loss": 112.2755,
      "step": 23110
    },
    {
      "epoch": 0.0934077255299636,
      "grad_norm": 532.1742553710938,
      "learning_rate": 3.6992e-05,
      "loss": 81.0992,
      "step": 23120
    },
    {
      "epoch": 0.09344812679533122,
      "grad_norm": 1730.8272705078125,
      "learning_rate": 3.7008e-05,
      "loss": 133.7006,
      "step": 23130
    },
    {
      "epoch": 0.09348852806069886,
      "grad_norm": 671.4747314453125,
      "learning_rate": 3.7024e-05,
      "loss": 84.7934,
      "step": 23140
    },
    {
      "epoch": 0.09352892932606649,
      "grad_norm": 2103.27880859375,
      "learning_rate": 3.7040000000000005e-05,
      "loss": 173.953,
      "step": 23150
    },
    {
      "epoch": 0.09356933059143413,
      "grad_norm": 3488.06787109375,
      "learning_rate": 3.7056000000000006e-05,
      "loss": 128.9973,
      "step": 23160
    },
    {
      "epoch": 0.09360973185680176,
      "grad_norm": 819.060791015625,
      "learning_rate": 3.7072e-05,
      "loss": 62.6777,
      "step": 23170
    },
    {
      "epoch": 0.09365013312216938,
      "grad_norm": 905.0873413085938,
      "learning_rate": 3.7088e-05,
      "loss": 148.4846,
      "step": 23180
    },
    {
      "epoch": 0.09369053438753702,
      "grad_norm": 987.261474609375,
      "learning_rate": 3.7104e-05,
      "loss": 132.8778,
      "step": 23190
    },
    {
      "epoch": 0.09373093565290465,
      "grad_norm": 1234.3077392578125,
      "learning_rate": 3.7120000000000004e-05,
      "loss": 124.4064,
      "step": 23200
    },
    {
      "epoch": 0.09377133691827227,
      "grad_norm": 1419.2791748046875,
      "learning_rate": 3.7136000000000005e-05,
      "loss": 126.3721,
      "step": 23210
    },
    {
      "epoch": 0.09381173818363991,
      "grad_norm": 917.65771484375,
      "learning_rate": 3.7152e-05,
      "loss": 130.0472,
      "step": 23220
    },
    {
      "epoch": 0.09385213944900754,
      "grad_norm": 1203.3936767578125,
      "learning_rate": 3.716800000000001e-05,
      "loss": 180.4214,
      "step": 23230
    },
    {
      "epoch": 0.09389254071437518,
      "grad_norm": 631.2800903320312,
      "learning_rate": 3.7184e-05,
      "loss": 77.2428,
      "step": 23240
    },
    {
      "epoch": 0.0939329419797428,
      "grad_norm": 1260.2119140625,
      "learning_rate": 3.72e-05,
      "loss": 61.4884,
      "step": 23250
    },
    {
      "epoch": 0.09397334324511043,
      "grad_norm": 515.2831420898438,
      "learning_rate": 3.7216000000000004e-05,
      "loss": 105.8688,
      "step": 23260
    },
    {
      "epoch": 0.09401374451047807,
      "grad_norm": 374.6453857421875,
      "learning_rate": 3.7232e-05,
      "loss": 62.7642,
      "step": 23270
    },
    {
      "epoch": 0.0940541457758457,
      "grad_norm": 790.422607421875,
      "learning_rate": 3.7248000000000007e-05,
      "loss": 103.0694,
      "step": 23280
    },
    {
      "epoch": 0.09409454704121333,
      "grad_norm": 442.69647216796875,
      "learning_rate": 3.7264e-05,
      "loss": 73.2971,
      "step": 23290
    },
    {
      "epoch": 0.09413494830658097,
      "grad_norm": 1085.193115234375,
      "learning_rate": 3.728e-05,
      "loss": 98.9435,
      "step": 23300
    },
    {
      "epoch": 0.09417534957194859,
      "grad_norm": 536.7901000976562,
      "learning_rate": 3.7296000000000003e-05,
      "loss": 80.0706,
      "step": 23310
    },
    {
      "epoch": 0.09421575083731623,
      "grad_norm": 1735.353759765625,
      "learning_rate": 3.7312000000000005e-05,
      "loss": 116.9445,
      "step": 23320
    },
    {
      "epoch": 0.09425615210268386,
      "grad_norm": 1051.9068603515625,
      "learning_rate": 3.7328000000000006e-05,
      "loss": 107.8367,
      "step": 23330
    },
    {
      "epoch": 0.09429655336805148,
      "grad_norm": 678.4794921875,
      "learning_rate": 3.7344e-05,
      "loss": 101.073,
      "step": 23340
    },
    {
      "epoch": 0.09433695463341912,
      "grad_norm": 1435.1505126953125,
      "learning_rate": 3.736000000000001e-05,
      "loss": 91.6877,
      "step": 23350
    },
    {
      "epoch": 0.09437735589878675,
      "grad_norm": 940.3088989257812,
      "learning_rate": 3.7376e-05,
      "loss": 114.8032,
      "step": 23360
    },
    {
      "epoch": 0.09441775716415438,
      "grad_norm": 740.4794311523438,
      "learning_rate": 3.7392000000000004e-05,
      "loss": 86.9696,
      "step": 23370
    },
    {
      "epoch": 0.09445815842952202,
      "grad_norm": 2469.025634765625,
      "learning_rate": 3.7408000000000005e-05,
      "loss": 120.204,
      "step": 23380
    },
    {
      "epoch": 0.09449855969488964,
      "grad_norm": 856.591796875,
      "learning_rate": 3.7424e-05,
      "loss": 62.1455,
      "step": 23390
    },
    {
      "epoch": 0.09453896096025728,
      "grad_norm": 1139.650390625,
      "learning_rate": 3.744000000000001e-05,
      "loss": 174.0881,
      "step": 23400
    },
    {
      "epoch": 0.09457936222562491,
      "grad_norm": 1099.4644775390625,
      "learning_rate": 3.7456e-05,
      "loss": 116.2295,
      "step": 23410
    },
    {
      "epoch": 0.09461976349099253,
      "grad_norm": 955.2962646484375,
      "learning_rate": 3.7472e-05,
      "loss": 94.5716,
      "step": 23420
    },
    {
      "epoch": 0.09466016475636017,
      "grad_norm": 960.6361694335938,
      "learning_rate": 3.7488000000000004e-05,
      "loss": 73.6774,
      "step": 23430
    },
    {
      "epoch": 0.0947005660217278,
      "grad_norm": 728.4270629882812,
      "learning_rate": 3.7504000000000005e-05,
      "loss": 110.0601,
      "step": 23440
    },
    {
      "epoch": 0.09474096728709543,
      "grad_norm": 879.024658203125,
      "learning_rate": 3.752e-05,
      "loss": 126.9347,
      "step": 23450
    },
    {
      "epoch": 0.09478136855246307,
      "grad_norm": 607.6163940429688,
      "learning_rate": 3.7536e-05,
      "loss": 66.8819,
      "step": 23460
    },
    {
      "epoch": 0.09482176981783069,
      "grad_norm": 1249.79931640625,
      "learning_rate": 3.7552e-05,
      "loss": 109.3968,
      "step": 23470
    },
    {
      "epoch": 0.09486217108319833,
      "grad_norm": 820.2130126953125,
      "learning_rate": 3.7568000000000004e-05,
      "loss": 77.7926,
      "step": 23480
    },
    {
      "epoch": 0.09490257234856596,
      "grad_norm": 2280.809326171875,
      "learning_rate": 3.7584000000000005e-05,
      "loss": 138.4351,
      "step": 23490
    },
    {
      "epoch": 0.09494297361393358,
      "grad_norm": 1270.8955078125,
      "learning_rate": 3.76e-05,
      "loss": 141.0211,
      "step": 23500
    },
    {
      "epoch": 0.09498337487930122,
      "grad_norm": 768.01953125,
      "learning_rate": 3.7616e-05,
      "loss": 99.7928,
      "step": 23510
    },
    {
      "epoch": 0.09502377614466885,
      "grad_norm": 1329.7403564453125,
      "learning_rate": 3.7632e-05,
      "loss": 117.8346,
      "step": 23520
    },
    {
      "epoch": 0.09506417741003648,
      "grad_norm": 1108.1868896484375,
      "learning_rate": 3.7648e-05,
      "loss": 108.9952,
      "step": 23530
    },
    {
      "epoch": 0.09510457867540412,
      "grad_norm": 1093.89794921875,
      "learning_rate": 3.7664000000000004e-05,
      "loss": 86.9773,
      "step": 23540
    },
    {
      "epoch": 0.09514497994077174,
      "grad_norm": 1512.74951171875,
      "learning_rate": 3.768e-05,
      "loss": 117.1909,
      "step": 23550
    },
    {
      "epoch": 0.09518538120613938,
      "grad_norm": 701.7789916992188,
      "learning_rate": 3.7696000000000006e-05,
      "loss": 109.6943,
      "step": 23560
    },
    {
      "epoch": 0.09522578247150701,
      "grad_norm": 2115.947265625,
      "learning_rate": 3.7712e-05,
      "loss": 105.5409,
      "step": 23570
    },
    {
      "epoch": 0.09526618373687464,
      "grad_norm": 467.70703125,
      "learning_rate": 3.7728e-05,
      "loss": 79.5947,
      "step": 23580
    },
    {
      "epoch": 0.09530658500224228,
      "grad_norm": 736.294921875,
      "learning_rate": 3.7744e-05,
      "loss": 70.7979,
      "step": 23590
    },
    {
      "epoch": 0.0953469862676099,
      "grad_norm": 1899.92529296875,
      "learning_rate": 3.7760000000000004e-05,
      "loss": 131.5578,
      "step": 23600
    },
    {
      "epoch": 0.09538738753297753,
      "grad_norm": 848.14453125,
      "learning_rate": 3.7776000000000006e-05,
      "loss": 96.7246,
      "step": 23610
    },
    {
      "epoch": 0.09542778879834517,
      "grad_norm": 1070.447509765625,
      "learning_rate": 3.7792e-05,
      "loss": 119.4036,
      "step": 23620
    },
    {
      "epoch": 0.0954681900637128,
      "grad_norm": 1277.6422119140625,
      "learning_rate": 3.780800000000001e-05,
      "loss": 168.7875,
      "step": 23630
    },
    {
      "epoch": 0.09550859132908043,
      "grad_norm": 1350.6763916015625,
      "learning_rate": 3.7824e-05,
      "loss": 110.022,
      "step": 23640
    },
    {
      "epoch": 0.09554899259444806,
      "grad_norm": 2100.522705078125,
      "learning_rate": 3.7840000000000004e-05,
      "loss": 166.9165,
      "step": 23650
    },
    {
      "epoch": 0.09558939385981569,
      "grad_norm": 632.5545043945312,
      "learning_rate": 3.7856000000000005e-05,
      "loss": 102.5553,
      "step": 23660
    },
    {
      "epoch": 0.09562979512518333,
      "grad_norm": 1013.6865844726562,
      "learning_rate": 3.7872e-05,
      "loss": 115.3168,
      "step": 23670
    },
    {
      "epoch": 0.09567019639055095,
      "grad_norm": 834.7869262695312,
      "learning_rate": 3.788800000000001e-05,
      "loss": 154.9165,
      "step": 23680
    },
    {
      "epoch": 0.09571059765591858,
      "grad_norm": 854.434814453125,
      "learning_rate": 3.7904e-05,
      "loss": 92.1741,
      "step": 23690
    },
    {
      "epoch": 0.09575099892128622,
      "grad_norm": 899.4342041015625,
      "learning_rate": 3.792e-05,
      "loss": 121.1676,
      "step": 23700
    },
    {
      "epoch": 0.09579140018665384,
      "grad_norm": 489.4569091796875,
      "learning_rate": 3.7936000000000004e-05,
      "loss": 114.8512,
      "step": 23710
    },
    {
      "epoch": 0.09583180145202148,
      "grad_norm": 1061.2698974609375,
      "learning_rate": 3.7952000000000005e-05,
      "loss": 104.255,
      "step": 23720
    },
    {
      "epoch": 0.09587220271738911,
      "grad_norm": 2563.587158203125,
      "learning_rate": 3.7968000000000006e-05,
      "loss": 97.8892,
      "step": 23730
    },
    {
      "epoch": 0.09591260398275674,
      "grad_norm": 1684.666015625,
      "learning_rate": 3.7984e-05,
      "loss": 142.1081,
      "step": 23740
    },
    {
      "epoch": 0.09595300524812438,
      "grad_norm": 1298.2633056640625,
      "learning_rate": 3.8e-05,
      "loss": 108.3703,
      "step": 23750
    },
    {
      "epoch": 0.095993406513492,
      "grad_norm": 639.5582275390625,
      "learning_rate": 3.8016e-05,
      "loss": 103.8809,
      "step": 23760
    },
    {
      "epoch": 0.09603380777885963,
      "grad_norm": 884.7063598632812,
      "learning_rate": 3.8032000000000004e-05,
      "loss": 142.9607,
      "step": 23770
    },
    {
      "epoch": 0.09607420904422727,
      "grad_norm": 835.0308837890625,
      "learning_rate": 3.8048000000000006e-05,
      "loss": 98.5731,
      "step": 23780
    },
    {
      "epoch": 0.0961146103095949,
      "grad_norm": 1084.726806640625,
      "learning_rate": 3.8064e-05,
      "loss": 135.3973,
      "step": 23790
    },
    {
      "epoch": 0.09615501157496253,
      "grad_norm": 684.0401000976562,
      "learning_rate": 3.808e-05,
      "loss": 119.6357,
      "step": 23800
    },
    {
      "epoch": 0.09619541284033016,
      "grad_norm": 546.32421875,
      "learning_rate": 3.8096e-05,
      "loss": 89.1974,
      "step": 23810
    },
    {
      "epoch": 0.09623581410569779,
      "grad_norm": 661.587646484375,
      "learning_rate": 3.8112000000000004e-05,
      "loss": 111.5153,
      "step": 23820
    },
    {
      "epoch": 0.09627621537106543,
      "grad_norm": 1184.2430419921875,
      "learning_rate": 3.8128000000000005e-05,
      "loss": 151.34,
      "step": 23830
    },
    {
      "epoch": 0.09631661663643305,
      "grad_norm": 694.1065673828125,
      "learning_rate": 3.8144000000000006e-05,
      "loss": 130.5082,
      "step": 23840
    },
    {
      "epoch": 0.09635701790180068,
      "grad_norm": 1578.1820068359375,
      "learning_rate": 3.816e-05,
      "loss": 78.125,
      "step": 23850
    },
    {
      "epoch": 0.09639741916716832,
      "grad_norm": 810.7494506835938,
      "learning_rate": 3.8176e-05,
      "loss": 115.6929,
      "step": 23860
    },
    {
      "epoch": 0.09643782043253595,
      "grad_norm": 806.3323364257812,
      "learning_rate": 3.8192e-05,
      "loss": 88.014,
      "step": 23870
    },
    {
      "epoch": 0.09647822169790359,
      "grad_norm": 677.9561157226562,
      "learning_rate": 3.8208000000000004e-05,
      "loss": 92.2052,
      "step": 23880
    },
    {
      "epoch": 0.09651862296327121,
      "grad_norm": 843.2612915039062,
      "learning_rate": 3.8224000000000005e-05,
      "loss": 117.2728,
      "step": 23890
    },
    {
      "epoch": 0.09655902422863884,
      "grad_norm": 560.440673828125,
      "learning_rate": 3.824e-05,
      "loss": 156.6519,
      "step": 23900
    },
    {
      "epoch": 0.09659942549400648,
      "grad_norm": 2791.057373046875,
      "learning_rate": 3.8256e-05,
      "loss": 160.5463,
      "step": 23910
    },
    {
      "epoch": 0.0966398267593741,
      "grad_norm": 1825.56494140625,
      "learning_rate": 3.8272e-05,
      "loss": 76.3217,
      "step": 23920
    },
    {
      "epoch": 0.09668022802474173,
      "grad_norm": 756.5723876953125,
      "learning_rate": 3.8288000000000003e-05,
      "loss": 114.9041,
      "step": 23930
    },
    {
      "epoch": 0.09672062929010937,
      "grad_norm": 735.3499755859375,
      "learning_rate": 3.8304000000000005e-05,
      "loss": 125.154,
      "step": 23940
    },
    {
      "epoch": 0.096761030555477,
      "grad_norm": 1043.668701171875,
      "learning_rate": 3.832e-05,
      "loss": 118.3253,
      "step": 23950
    },
    {
      "epoch": 0.09680143182084464,
      "grad_norm": 694.8841552734375,
      "learning_rate": 3.833600000000001e-05,
      "loss": 103.1498,
      "step": 23960
    },
    {
      "epoch": 0.09684183308621226,
      "grad_norm": 1475.5992431640625,
      "learning_rate": 3.8352e-05,
      "loss": 121.2389,
      "step": 23970
    },
    {
      "epoch": 0.09688223435157989,
      "grad_norm": 506.0126953125,
      "learning_rate": 3.8368e-05,
      "loss": 80.8196,
      "step": 23980
    },
    {
      "epoch": 0.09692263561694753,
      "grad_norm": 885.1693115234375,
      "learning_rate": 3.8384000000000004e-05,
      "loss": 104.6312,
      "step": 23990
    },
    {
      "epoch": 0.09696303688231515,
      "grad_norm": 1443.1024169921875,
      "learning_rate": 3.8400000000000005e-05,
      "loss": 134.2576,
      "step": 24000
    },
    {
      "epoch": 0.09700343814768278,
      "grad_norm": 1300.169921875,
      "learning_rate": 3.8416000000000006e-05,
      "loss": 136.2243,
      "step": 24010
    },
    {
      "epoch": 0.09704383941305042,
      "grad_norm": 606.1035766601562,
      "learning_rate": 3.8432e-05,
      "loss": 129.7206,
      "step": 24020
    },
    {
      "epoch": 0.09708424067841805,
      "grad_norm": 1067.2373046875,
      "learning_rate": 3.844800000000001e-05,
      "loss": 117.3039,
      "step": 24030
    },
    {
      "epoch": 0.09712464194378569,
      "grad_norm": 827.10302734375,
      "learning_rate": 3.8464e-05,
      "loss": 133.863,
      "step": 24040
    },
    {
      "epoch": 0.09716504320915331,
      "grad_norm": 1323.8310546875,
      "learning_rate": 3.8480000000000004e-05,
      "loss": 97.7097,
      "step": 24050
    },
    {
      "epoch": 0.09720544447452094,
      "grad_norm": 694.2848510742188,
      "learning_rate": 3.8496000000000005e-05,
      "loss": 98.1374,
      "step": 24060
    },
    {
      "epoch": 0.09724584573988858,
      "grad_norm": 389.7241516113281,
      "learning_rate": 3.8512e-05,
      "loss": 56.3713,
      "step": 24070
    },
    {
      "epoch": 0.0972862470052562,
      "grad_norm": 1979.0894775390625,
      "learning_rate": 3.8528e-05,
      "loss": 135.2374,
      "step": 24080
    },
    {
      "epoch": 0.09732664827062383,
      "grad_norm": 920.7020263671875,
      "learning_rate": 3.8544e-05,
      "loss": 84.8128,
      "step": 24090
    },
    {
      "epoch": 0.09736704953599147,
      "grad_norm": 650.2858276367188,
      "learning_rate": 3.8560000000000004e-05,
      "loss": 124.8343,
      "step": 24100
    },
    {
      "epoch": 0.0974074508013591,
      "grad_norm": 748.7391357421875,
      "learning_rate": 3.8576000000000005e-05,
      "loss": 105.6386,
      "step": 24110
    },
    {
      "epoch": 0.09744785206672674,
      "grad_norm": 1407.361328125,
      "learning_rate": 3.8592000000000006e-05,
      "loss": 88.1447,
      "step": 24120
    },
    {
      "epoch": 0.09748825333209436,
      "grad_norm": 823.1781616210938,
      "learning_rate": 3.8608e-05,
      "loss": 94.2701,
      "step": 24130
    },
    {
      "epoch": 0.09752865459746199,
      "grad_norm": 1753.80712890625,
      "learning_rate": 3.8624e-05,
      "loss": 112.8359,
      "step": 24140
    },
    {
      "epoch": 0.09756905586282963,
      "grad_norm": 1930.0491943359375,
      "learning_rate": 3.864e-05,
      "loss": 108.6234,
      "step": 24150
    },
    {
      "epoch": 0.09760945712819726,
      "grad_norm": 720.0437622070312,
      "learning_rate": 3.8656000000000004e-05,
      "loss": 91.7267,
      "step": 24160
    },
    {
      "epoch": 0.09764985839356488,
      "grad_norm": 1543.5885009765625,
      "learning_rate": 3.8672000000000005e-05,
      "loss": 86.1541,
      "step": 24170
    },
    {
      "epoch": 0.09769025965893252,
      "grad_norm": 1376.3941650390625,
      "learning_rate": 3.8688e-05,
      "loss": 139.0348,
      "step": 24180
    },
    {
      "epoch": 0.09773066092430015,
      "grad_norm": 1557.67919921875,
      "learning_rate": 3.8704e-05,
      "loss": 105.9557,
      "step": 24190
    },
    {
      "epoch": 0.09777106218966779,
      "grad_norm": 1324.2451171875,
      "learning_rate": 3.872e-05,
      "loss": 102.7443,
      "step": 24200
    },
    {
      "epoch": 0.09781146345503541,
      "grad_norm": 921.9620361328125,
      "learning_rate": 3.8736e-05,
      "loss": 143.7736,
      "step": 24210
    },
    {
      "epoch": 0.09785186472040304,
      "grad_norm": 738.0506591796875,
      "learning_rate": 3.8752000000000004e-05,
      "loss": 104.4059,
      "step": 24220
    },
    {
      "epoch": 0.09789226598577068,
      "grad_norm": 3766.658447265625,
      "learning_rate": 3.8768e-05,
      "loss": 158.2326,
      "step": 24230
    },
    {
      "epoch": 0.0979326672511383,
      "grad_norm": 653.2822875976562,
      "learning_rate": 3.878400000000001e-05,
      "loss": 81.5535,
      "step": 24240
    },
    {
      "epoch": 0.09797306851650593,
      "grad_norm": 550.072265625,
      "learning_rate": 3.88e-05,
      "loss": 168.4727,
      "step": 24250
    },
    {
      "epoch": 0.09801346978187357,
      "grad_norm": 1663.3287353515625,
      "learning_rate": 3.8816e-05,
      "loss": 155.4968,
      "step": 24260
    },
    {
      "epoch": 0.0980538710472412,
      "grad_norm": 2497.78466796875,
      "learning_rate": 3.8832000000000004e-05,
      "loss": 90.2777,
      "step": 24270
    },
    {
      "epoch": 0.09809427231260884,
      "grad_norm": 3979.058837890625,
      "learning_rate": 3.8848e-05,
      "loss": 113.6011,
      "step": 24280
    },
    {
      "epoch": 0.09813467357797646,
      "grad_norm": 1241.6263427734375,
      "learning_rate": 3.8864000000000006e-05,
      "loss": 99.367,
      "step": 24290
    },
    {
      "epoch": 0.09817507484334409,
      "grad_norm": 650.6756591796875,
      "learning_rate": 3.888e-05,
      "loss": 86.7288,
      "step": 24300
    },
    {
      "epoch": 0.09821547610871173,
      "grad_norm": 764.9371948242188,
      "learning_rate": 3.8896e-05,
      "loss": 113.2154,
      "step": 24310
    },
    {
      "epoch": 0.09825587737407936,
      "grad_norm": 787.2350463867188,
      "learning_rate": 3.8912e-05,
      "loss": 99.8835,
      "step": 24320
    },
    {
      "epoch": 0.09829627863944698,
      "grad_norm": 998.755126953125,
      "learning_rate": 3.8928000000000004e-05,
      "loss": 94.1024,
      "step": 24330
    },
    {
      "epoch": 0.09833667990481462,
      "grad_norm": 388.8599853515625,
      "learning_rate": 3.8944000000000005e-05,
      "loss": 61.2629,
      "step": 24340
    },
    {
      "epoch": 0.09837708117018225,
      "grad_norm": 525.0775756835938,
      "learning_rate": 3.896e-05,
      "loss": 79.6292,
      "step": 24350
    },
    {
      "epoch": 0.09841748243554987,
      "grad_norm": 1979.02294921875,
      "learning_rate": 3.897600000000001e-05,
      "loss": 157.633,
      "step": 24360
    },
    {
      "epoch": 0.09845788370091751,
      "grad_norm": 598.3854370117188,
      "learning_rate": 3.8992e-05,
      "loss": 96.9426,
      "step": 24370
    },
    {
      "epoch": 0.09849828496628514,
      "grad_norm": 1230.67724609375,
      "learning_rate": 3.9008e-05,
      "loss": 129.7387,
      "step": 24380
    },
    {
      "epoch": 0.09853868623165278,
      "grad_norm": 1653.3104248046875,
      "learning_rate": 3.9024000000000004e-05,
      "loss": 131.7615,
      "step": 24390
    },
    {
      "epoch": 0.0985790874970204,
      "grad_norm": 907.8214111328125,
      "learning_rate": 3.9040000000000006e-05,
      "loss": 127.1972,
      "step": 24400
    },
    {
      "epoch": 0.09861948876238803,
      "grad_norm": 3581.931640625,
      "learning_rate": 3.905600000000001e-05,
      "loss": 154.1143,
      "step": 24410
    },
    {
      "epoch": 0.09865989002775567,
      "grad_norm": 1160.47119140625,
      "learning_rate": 3.9072e-05,
      "loss": 133.1434,
      "step": 24420
    },
    {
      "epoch": 0.0987002912931233,
      "grad_norm": 890.6558227539062,
      "learning_rate": 3.9088e-05,
      "loss": 78.0879,
      "step": 24430
    },
    {
      "epoch": 0.09874069255849093,
      "grad_norm": 744.0023803710938,
      "learning_rate": 3.9104000000000004e-05,
      "loss": 158.2902,
      "step": 24440
    },
    {
      "epoch": 0.09878109382385857,
      "grad_norm": 1092.13671875,
      "learning_rate": 3.9120000000000005e-05,
      "loss": 129.2463,
      "step": 24450
    },
    {
      "epoch": 0.09882149508922619,
      "grad_norm": 783.1102905273438,
      "learning_rate": 3.9136000000000006e-05,
      "loss": 118.658,
      "step": 24460
    },
    {
      "epoch": 0.09886189635459383,
      "grad_norm": 2806.539306640625,
      "learning_rate": 3.9152e-05,
      "loss": 103.9075,
      "step": 24470
    },
    {
      "epoch": 0.09890229761996146,
      "grad_norm": 603.8475341796875,
      "learning_rate": 3.9168e-05,
      "loss": 76.5813,
      "step": 24480
    },
    {
      "epoch": 0.09894269888532908,
      "grad_norm": 1004.6978149414062,
      "learning_rate": 3.9184e-05,
      "loss": 161.1362,
      "step": 24490
    },
    {
      "epoch": 0.09898310015069672,
      "grad_norm": 1597.204833984375,
      "learning_rate": 3.9200000000000004e-05,
      "loss": 136.4567,
      "step": 24500
    },
    {
      "epoch": 0.09902350141606435,
      "grad_norm": 930.0423583984375,
      "learning_rate": 3.9216000000000005e-05,
      "loss": 114.5231,
      "step": 24510
    },
    {
      "epoch": 0.09906390268143198,
      "grad_norm": 2183.127685546875,
      "learning_rate": 3.9232000000000007e-05,
      "loss": 102.2725,
      "step": 24520
    },
    {
      "epoch": 0.09910430394679962,
      "grad_norm": 653.5531005859375,
      "learning_rate": 3.9248e-05,
      "loss": 93.1283,
      "step": 24530
    },
    {
      "epoch": 0.09914470521216724,
      "grad_norm": 1036.9739990234375,
      "learning_rate": 3.9264e-05,
      "loss": 120.2268,
      "step": 24540
    },
    {
      "epoch": 0.09918510647753488,
      "grad_norm": 1612.5570068359375,
      "learning_rate": 3.9280000000000003e-05,
      "loss": 94.6924,
      "step": 24550
    },
    {
      "epoch": 0.09922550774290251,
      "grad_norm": 863.7626342773438,
      "learning_rate": 3.9296000000000005e-05,
      "loss": 120.5359,
      "step": 24560
    },
    {
      "epoch": 0.09926590900827013,
      "grad_norm": 1776.318115234375,
      "learning_rate": 3.9312000000000006e-05,
      "loss": 179.9252,
      "step": 24570
    },
    {
      "epoch": 0.09930631027363777,
      "grad_norm": 665.2869262695312,
      "learning_rate": 3.9328e-05,
      "loss": 83.8321,
      "step": 24580
    },
    {
      "epoch": 0.0993467115390054,
      "grad_norm": 892.1969604492188,
      "learning_rate": 3.9344e-05,
      "loss": 146.0414,
      "step": 24590
    },
    {
      "epoch": 0.09938711280437303,
      "grad_norm": 0.0,
      "learning_rate": 3.936e-05,
      "loss": 133.4863,
      "step": 24600
    },
    {
      "epoch": 0.09942751406974067,
      "grad_norm": 1188.8670654296875,
      "learning_rate": 3.9376000000000004e-05,
      "loss": 101.8091,
      "step": 24610
    },
    {
      "epoch": 0.09946791533510829,
      "grad_norm": 737.1953735351562,
      "learning_rate": 3.9392000000000005e-05,
      "loss": 74.0432,
      "step": 24620
    },
    {
      "epoch": 0.09950831660047593,
      "grad_norm": 490.60430908203125,
      "learning_rate": 3.9408e-05,
      "loss": 82.5474,
      "step": 24630
    },
    {
      "epoch": 0.09954871786584356,
      "grad_norm": 1418.0870361328125,
      "learning_rate": 3.942400000000001e-05,
      "loss": 115.8092,
      "step": 24640
    },
    {
      "epoch": 0.09958911913121118,
      "grad_norm": 1001.0704956054688,
      "learning_rate": 3.944e-05,
      "loss": 74.5635,
      "step": 24650
    },
    {
      "epoch": 0.09962952039657882,
      "grad_norm": 1704.5621337890625,
      "learning_rate": 3.9456e-05,
      "loss": 119.8563,
      "step": 24660
    },
    {
      "epoch": 0.09966992166194645,
      "grad_norm": 1012.4752197265625,
      "learning_rate": 3.9472000000000004e-05,
      "loss": 124.5904,
      "step": 24670
    },
    {
      "epoch": 0.09971032292731408,
      "grad_norm": 1744.2069091796875,
      "learning_rate": 3.9488e-05,
      "loss": 114.0528,
      "step": 24680
    },
    {
      "epoch": 0.09975072419268172,
      "grad_norm": 895.2362670898438,
      "learning_rate": 3.950400000000001e-05,
      "loss": 128.9408,
      "step": 24690
    },
    {
      "epoch": 0.09979112545804934,
      "grad_norm": 713.849609375,
      "learning_rate": 3.952e-05,
      "loss": 79.1376,
      "step": 24700
    },
    {
      "epoch": 0.09983152672341698,
      "grad_norm": 966.8638305664062,
      "learning_rate": 3.9536e-05,
      "loss": 103.0292,
      "step": 24710
    },
    {
      "epoch": 0.09987192798878461,
      "grad_norm": 881.3704223632812,
      "learning_rate": 3.9552000000000003e-05,
      "loss": 82.8581,
      "step": 24720
    },
    {
      "epoch": 0.09991232925415224,
      "grad_norm": 555.9190063476562,
      "learning_rate": 3.9568000000000005e-05,
      "loss": 101.4556,
      "step": 24730
    },
    {
      "epoch": 0.09995273051951988,
      "grad_norm": 1946.76025390625,
      "learning_rate": 3.9584000000000006e-05,
      "loss": 112.9839,
      "step": 24740
    },
    {
      "epoch": 0.0999931317848875,
      "grad_norm": 1399.944091796875,
      "learning_rate": 3.96e-05,
      "loss": 161.2461,
      "step": 24750
    },
    {
      "epoch": 0.10003353305025513,
      "grad_norm": 492.08251953125,
      "learning_rate": 3.9616e-05,
      "loss": 85.7148,
      "step": 24760
    },
    {
      "epoch": 0.10007393431562277,
      "grad_norm": 698.5414428710938,
      "learning_rate": 3.9632e-05,
      "loss": 81.156,
      "step": 24770
    },
    {
      "epoch": 0.1001143355809904,
      "grad_norm": 1429.7156982421875,
      "learning_rate": 3.9648000000000004e-05,
      "loss": 78.8263,
      "step": 24780
    },
    {
      "epoch": 0.10015473684635803,
      "grad_norm": 2266.67041015625,
      "learning_rate": 3.9664000000000005e-05,
      "loss": 133.6976,
      "step": 24790
    },
    {
      "epoch": 0.10019513811172566,
      "grad_norm": 2114.505615234375,
      "learning_rate": 3.9680000000000006e-05,
      "loss": 157.514,
      "step": 24800
    },
    {
      "epoch": 0.10023553937709329,
      "grad_norm": 1892.30615234375,
      "learning_rate": 3.9696e-05,
      "loss": 112.1205,
      "step": 24810
    },
    {
      "epoch": 0.10027594064246093,
      "grad_norm": 1739.9906005859375,
      "learning_rate": 3.9712e-05,
      "loss": 124.3941,
      "step": 24820
    },
    {
      "epoch": 0.10031634190782855,
      "grad_norm": 1360.4404296875,
      "learning_rate": 3.9728e-05,
      "loss": 170.2335,
      "step": 24830
    },
    {
      "epoch": 0.10035674317319618,
      "grad_norm": 887.6372680664062,
      "learning_rate": 3.9744000000000004e-05,
      "loss": 135.4758,
      "step": 24840
    },
    {
      "epoch": 0.10039714443856382,
      "grad_norm": 356.5849914550781,
      "learning_rate": 3.9760000000000006e-05,
      "loss": 93.1374,
      "step": 24850
    },
    {
      "epoch": 0.10043754570393144,
      "grad_norm": 1182.120849609375,
      "learning_rate": 3.9776e-05,
      "loss": 147.9866,
      "step": 24860
    },
    {
      "epoch": 0.10047794696929908,
      "grad_norm": 1378.6016845703125,
      "learning_rate": 3.9792e-05,
      "loss": 116.0618,
      "step": 24870
    },
    {
      "epoch": 0.10051834823466671,
      "grad_norm": 1231.1202392578125,
      "learning_rate": 3.9808e-05,
      "loss": 113.3535,
      "step": 24880
    },
    {
      "epoch": 0.10055874950003434,
      "grad_norm": 2073.397705078125,
      "learning_rate": 3.9824000000000004e-05,
      "loss": 101.1435,
      "step": 24890
    },
    {
      "epoch": 0.10059915076540198,
      "grad_norm": 700.2265014648438,
      "learning_rate": 3.9840000000000005e-05,
      "loss": 119.9656,
      "step": 24900
    },
    {
      "epoch": 0.1006395520307696,
      "grad_norm": 1094.663818359375,
      "learning_rate": 3.9856e-05,
      "loss": 96.4916,
      "step": 24910
    },
    {
      "epoch": 0.10067995329613723,
      "grad_norm": 776.6048583984375,
      "learning_rate": 3.987200000000001e-05,
      "loss": 76.0012,
      "step": 24920
    },
    {
      "epoch": 0.10072035456150487,
      "grad_norm": 711.3641357421875,
      "learning_rate": 3.9888e-05,
      "loss": 103.8389,
      "step": 24930
    },
    {
      "epoch": 0.1007607558268725,
      "grad_norm": 1021.636474609375,
      "learning_rate": 3.9904e-05,
      "loss": 105.0635,
      "step": 24940
    },
    {
      "epoch": 0.10080115709224013,
      "grad_norm": 1410.3927001953125,
      "learning_rate": 3.9920000000000004e-05,
      "loss": 93.3259,
      "step": 24950
    },
    {
      "epoch": 0.10084155835760776,
      "grad_norm": 692.6359252929688,
      "learning_rate": 3.9936e-05,
      "loss": 110.1186,
      "step": 24960
    },
    {
      "epoch": 0.10088195962297539,
      "grad_norm": 672.6102905273438,
      "learning_rate": 3.9952000000000006e-05,
      "loss": 89.1554,
      "step": 24970
    },
    {
      "epoch": 0.10092236088834303,
      "grad_norm": 2117.183349609375,
      "learning_rate": 3.9968e-05,
      "loss": 142.3189,
      "step": 24980
    },
    {
      "epoch": 0.10096276215371065,
      "grad_norm": 937.2538452148438,
      "learning_rate": 3.9984e-05,
      "loss": 103.7949,
      "step": 24990
    },
    {
      "epoch": 0.10100316341907828,
      "grad_norm": 495.8382568359375,
      "learning_rate": 4e-05,
      "loss": 105.3346,
      "step": 25000
    },
    {
      "epoch": 0.10104356468444592,
      "grad_norm": 5377.6513671875,
      "learning_rate": 3.999999980504486e-05,
      "loss": 163.8216,
      "step": 25010
    },
    {
      "epoch": 0.10108396594981355,
      "grad_norm": 1024.221435546875,
      "learning_rate": 3.999999922017941e-05,
      "loss": 167.4767,
      "step": 25020
    },
    {
      "epoch": 0.10112436721518119,
      "grad_norm": 2024.415283203125,
      "learning_rate": 3.999999824540369e-05,
      "loss": 107.0843,
      "step": 25030
    },
    {
      "epoch": 0.10116476848054881,
      "grad_norm": 1808.91455078125,
      "learning_rate": 3.9999996880717705e-05,
      "loss": 75.9993,
      "step": 25040
    },
    {
      "epoch": 0.10120516974591644,
      "grad_norm": 1401.970458984375,
      "learning_rate": 3.999999512612149e-05,
      "loss": 118.6832,
      "step": 25050
    },
    {
      "epoch": 0.10124557101128408,
      "grad_norm": 1190.476806640625,
      "learning_rate": 3.9999992981615066e-05,
      "loss": 116.5726,
      "step": 25060
    },
    {
      "epoch": 0.1012859722766517,
      "grad_norm": 517.6707763671875,
      "learning_rate": 3.999999044719848e-05,
      "loss": 101.7224,
      "step": 25070
    },
    {
      "epoch": 0.10132637354201933,
      "grad_norm": 514.5786743164062,
      "learning_rate": 3.9999987522871786e-05,
      "loss": 83.3145,
      "step": 25080
    },
    {
      "epoch": 0.10136677480738697,
      "grad_norm": 1905.0189208984375,
      "learning_rate": 3.999998420863504e-05,
      "loss": 116.2477,
      "step": 25090
    },
    {
      "epoch": 0.1014071760727546,
      "grad_norm": 2310.874267578125,
      "learning_rate": 3.9999980504488305e-05,
      "loss": 92.526,
      "step": 25100
    },
    {
      "epoch": 0.10144757733812224,
      "grad_norm": 907.70166015625,
      "learning_rate": 3.999997641043165e-05,
      "loss": 128.3388,
      "step": 25110
    },
    {
      "epoch": 0.10148797860348986,
      "grad_norm": 440.4619140625,
      "learning_rate": 3.999997192646517e-05,
      "loss": 105.6848,
      "step": 25120
    },
    {
      "epoch": 0.10152837986885749,
      "grad_norm": 688.9473266601562,
      "learning_rate": 3.999996705258893e-05,
      "loss": 97.8082,
      "step": 25130
    },
    {
      "epoch": 0.10156878113422513,
      "grad_norm": 472.28729248046875,
      "learning_rate": 3.9999961788803034e-05,
      "loss": 174.5959,
      "step": 25140
    },
    {
      "epoch": 0.10160918239959275,
      "grad_norm": 919.4473266601562,
      "learning_rate": 3.999995613510759e-05,
      "loss": 138.1659,
      "step": 25150
    },
    {
      "epoch": 0.10164958366496038,
      "grad_norm": 1194.282470703125,
      "learning_rate": 3.9999950091502706e-05,
      "loss": 138.3285,
      "step": 25160
    },
    {
      "epoch": 0.10168998493032802,
      "grad_norm": 1675.2808837890625,
      "learning_rate": 3.9999943657988496e-05,
      "loss": 139.3541,
      "step": 25170
    },
    {
      "epoch": 0.10173038619569565,
      "grad_norm": 995.9146728515625,
      "learning_rate": 3.9999936834565085e-05,
      "loss": 106.5323,
      "step": 25180
    },
    {
      "epoch": 0.10177078746106329,
      "grad_norm": 1579.7122802734375,
      "learning_rate": 3.9999929621232614e-05,
      "loss": 113.5721,
      "step": 25190
    },
    {
      "epoch": 0.10181118872643091,
      "grad_norm": 1105.6669921875,
      "learning_rate": 3.9999922017991213e-05,
      "loss": 117.8805,
      "step": 25200
    },
    {
      "epoch": 0.10185158999179854,
      "grad_norm": 888.4319458007812,
      "learning_rate": 3.999991402484104e-05,
      "loss": 94.7415,
      "step": 25210
    },
    {
      "epoch": 0.10189199125716618,
      "grad_norm": 1027.7147216796875,
      "learning_rate": 3.999990564178225e-05,
      "loss": 62.6618,
      "step": 25220
    },
    {
      "epoch": 0.1019323925225338,
      "grad_norm": 799.4996337890625,
      "learning_rate": 3.9999896868814995e-05,
      "loss": 116.6909,
      "step": 25230
    },
    {
      "epoch": 0.10197279378790143,
      "grad_norm": 3651.783203125,
      "learning_rate": 3.999988770593946e-05,
      "loss": 165.9793,
      "step": 25240
    },
    {
      "epoch": 0.10201319505326907,
      "grad_norm": 729.583740234375,
      "learning_rate": 3.999987815315581e-05,
      "loss": 95.7285,
      "step": 25250
    },
    {
      "epoch": 0.1020535963186367,
      "grad_norm": 634.2191162109375,
      "learning_rate": 3.9999868210464244e-05,
      "loss": 109.2163,
      "step": 25260
    },
    {
      "epoch": 0.10209399758400434,
      "grad_norm": 683.3226318359375,
      "learning_rate": 3.9999857877864945e-05,
      "loss": 125.2095,
      "step": 25270
    },
    {
      "epoch": 0.10213439884937196,
      "grad_norm": 952.6441650390625,
      "learning_rate": 3.999984715535813e-05,
      "loss": 96.8702,
      "step": 25280
    },
    {
      "epoch": 0.10217480011473959,
      "grad_norm": 612.0934448242188,
      "learning_rate": 3.999983604294399e-05,
      "loss": 98.7964,
      "step": 25290
    },
    {
      "epoch": 0.10221520138010723,
      "grad_norm": 668.8614501953125,
      "learning_rate": 3.999982454062275e-05,
      "loss": 79.0043,
      "step": 25300
    },
    {
      "epoch": 0.10225560264547486,
      "grad_norm": 1596.1842041015625,
      "learning_rate": 3.999981264839464e-05,
      "loss": 103.495,
      "step": 25310
    },
    {
      "epoch": 0.10229600391084248,
      "grad_norm": 675.1773681640625,
      "learning_rate": 3.999980036625989e-05,
      "loss": 157.9035,
      "step": 25320
    },
    {
      "epoch": 0.10233640517621012,
      "grad_norm": 1616.4547119140625,
      "learning_rate": 3.9999787694218724e-05,
      "loss": 93.8177,
      "step": 25330
    },
    {
      "epoch": 0.10237680644157775,
      "grad_norm": 834.3451538085938,
      "learning_rate": 3.99997746322714e-05,
      "loss": 137.3178,
      "step": 25340
    },
    {
      "epoch": 0.10241720770694539,
      "grad_norm": 1074.680419921875,
      "learning_rate": 3.999976118041818e-05,
      "loss": 106.7518,
      "step": 25350
    },
    {
      "epoch": 0.10245760897231301,
      "grad_norm": 432.718505859375,
      "learning_rate": 3.999974733865932e-05,
      "loss": 110.611,
      "step": 25360
    },
    {
      "epoch": 0.10249801023768064,
      "grad_norm": 1014.7579956054688,
      "learning_rate": 3.999973310699509e-05,
      "loss": 137.3919,
      "step": 25370
    },
    {
      "epoch": 0.10253841150304828,
      "grad_norm": 1398.2919921875,
      "learning_rate": 3.999971848542576e-05,
      "loss": 129.0789,
      "step": 25380
    },
    {
      "epoch": 0.1025788127684159,
      "grad_norm": 2139.510498046875,
      "learning_rate": 3.999970347395162e-05,
      "loss": 93.033,
      "step": 25390
    },
    {
      "epoch": 0.10261921403378353,
      "grad_norm": 1227.2335205078125,
      "learning_rate": 3.9999688072572966e-05,
      "loss": 158.6131,
      "step": 25400
    },
    {
      "epoch": 0.10265961529915117,
      "grad_norm": 444.8744201660156,
      "learning_rate": 3.99996722812901e-05,
      "loss": 104.8071,
      "step": 25410
    },
    {
      "epoch": 0.1027000165645188,
      "grad_norm": 761.33837890625,
      "learning_rate": 3.9999656100103325e-05,
      "loss": 97.1906,
      "step": 25420
    },
    {
      "epoch": 0.10274041782988644,
      "grad_norm": 1000.1937255859375,
      "learning_rate": 3.999963952901296e-05,
      "loss": 83.2851,
      "step": 25430
    },
    {
      "epoch": 0.10278081909525406,
      "grad_norm": 1454.795166015625,
      "learning_rate": 3.999962256801932e-05,
      "loss": 125.3475,
      "step": 25440
    },
    {
      "epoch": 0.10282122036062169,
      "grad_norm": 620.0244750976562,
      "learning_rate": 3.999960521712274e-05,
      "loss": 151.7363,
      "step": 25450
    },
    {
      "epoch": 0.10286162162598933,
      "grad_norm": 1074.2821044921875,
      "learning_rate": 3.999958747632357e-05,
      "loss": 117.8375,
      "step": 25460
    },
    {
      "epoch": 0.10290202289135696,
      "grad_norm": 2790.96533203125,
      "learning_rate": 3.9999569345622135e-05,
      "loss": 123.3024,
      "step": 25470
    },
    {
      "epoch": 0.10294242415672458,
      "grad_norm": 2994.655517578125,
      "learning_rate": 3.999955082501881e-05,
      "loss": 92.7528,
      "step": 25480
    },
    {
      "epoch": 0.10298282542209222,
      "grad_norm": 445.1771545410156,
      "learning_rate": 3.999953191451394e-05,
      "loss": 88.779,
      "step": 25490
    },
    {
      "epoch": 0.10302322668745985,
      "grad_norm": 1553.27490234375,
      "learning_rate": 3.9999512614107894e-05,
      "loss": 153.5818,
      "step": 25500
    },
    {
      "epoch": 0.10306362795282749,
      "grad_norm": 2650.843017578125,
      "learning_rate": 3.999949292380106e-05,
      "loss": 93.9213,
      "step": 25510
    },
    {
      "epoch": 0.10310402921819511,
      "grad_norm": 654.7503662109375,
      "learning_rate": 3.9999472843593816e-05,
      "loss": 133.6554,
      "step": 25520
    },
    {
      "epoch": 0.10314443048356274,
      "grad_norm": 398.31951904296875,
      "learning_rate": 3.999945237348655e-05,
      "loss": 167.9717,
      "step": 25530
    },
    {
      "epoch": 0.10318483174893038,
      "grad_norm": 1162.1407470703125,
      "learning_rate": 3.999943151347967e-05,
      "loss": 124.5522,
      "step": 25540
    },
    {
      "epoch": 0.103225233014298,
      "grad_norm": 1724.40283203125,
      "learning_rate": 3.999941026357356e-05,
      "loss": 122.637,
      "step": 25550
    },
    {
      "epoch": 0.10326563427966563,
      "grad_norm": 731.8903198242188,
      "learning_rate": 3.999938862376866e-05,
      "loss": 67.4073,
      "step": 25560
    },
    {
      "epoch": 0.10330603554503327,
      "grad_norm": 0.0,
      "learning_rate": 3.999936659406539e-05,
      "loss": 68.4163,
      "step": 25570
    },
    {
      "epoch": 0.1033464368104009,
      "grad_norm": 705.5543212890625,
      "learning_rate": 3.999934417446416e-05,
      "loss": 83.734,
      "step": 25580
    },
    {
      "epoch": 0.10338683807576854,
      "grad_norm": 1163.376220703125,
      "learning_rate": 3.9999321364965415e-05,
      "loss": 128.0824,
      "step": 25590
    },
    {
      "epoch": 0.10342723934113617,
      "grad_norm": 1845.739990234375,
      "learning_rate": 3.9999298165569614e-05,
      "loss": 109.6928,
      "step": 25600
    },
    {
      "epoch": 0.10346764060650379,
      "grad_norm": 509.751220703125,
      "learning_rate": 3.9999274576277196e-05,
      "loss": 97.6254,
      "step": 25610
    },
    {
      "epoch": 0.10350804187187143,
      "grad_norm": 1141.1207275390625,
      "learning_rate": 3.999925059708863e-05,
      "loss": 89.0088,
      "step": 25620
    },
    {
      "epoch": 0.10354844313723906,
      "grad_norm": 1681.5274658203125,
      "learning_rate": 3.999922622800437e-05,
      "loss": 107.7538,
      "step": 25630
    },
    {
      "epoch": 0.10358884440260668,
      "grad_norm": 2417.304931640625,
      "learning_rate": 3.99992014690249e-05,
      "loss": 169.7357,
      "step": 25640
    },
    {
      "epoch": 0.10362924566797432,
      "grad_norm": 1126.2418212890625,
      "learning_rate": 3.99991763201507e-05,
      "loss": 111.1401,
      "step": 25650
    },
    {
      "epoch": 0.10366964693334195,
      "grad_norm": 1027.2210693359375,
      "learning_rate": 3.999915078138226e-05,
      "loss": 116.2014,
      "step": 25660
    },
    {
      "epoch": 0.10371004819870959,
      "grad_norm": 1453.3272705078125,
      "learning_rate": 3.999912485272008e-05,
      "loss": 103.9699,
      "step": 25670
    },
    {
      "epoch": 0.10375044946407722,
      "grad_norm": 457.8889465332031,
      "learning_rate": 3.9999098534164675e-05,
      "loss": 69.5038,
      "step": 25680
    },
    {
      "epoch": 0.10379085072944484,
      "grad_norm": 531.3936767578125,
      "learning_rate": 3.999907182571654e-05,
      "loss": 106.4484,
      "step": 25690
    },
    {
      "epoch": 0.10383125199481248,
      "grad_norm": 3379.6875,
      "learning_rate": 3.99990447273762e-05,
      "loss": 134.128,
      "step": 25700
    },
    {
      "epoch": 0.10387165326018011,
      "grad_norm": 1140.1044921875,
      "learning_rate": 3.999901723914419e-05,
      "loss": 123.5596,
      "step": 25710
    },
    {
      "epoch": 0.10391205452554773,
      "grad_norm": 459.4917907714844,
      "learning_rate": 3.999898936102104e-05,
      "loss": 62.9321,
      "step": 25720
    },
    {
      "epoch": 0.10395245579091537,
      "grad_norm": 1177.872314453125,
      "learning_rate": 3.9998961093007295e-05,
      "loss": 143.5161,
      "step": 25730
    },
    {
      "epoch": 0.103992857056283,
      "grad_norm": 873.7769165039062,
      "learning_rate": 3.9998932435103513e-05,
      "loss": 93.7114,
      "step": 25740
    },
    {
      "epoch": 0.10403325832165064,
      "grad_norm": 1600.9781494140625,
      "learning_rate": 3.999890338731025e-05,
      "loss": 103.5482,
      "step": 25750
    },
    {
      "epoch": 0.10407365958701827,
      "grad_norm": 4216.92724609375,
      "learning_rate": 3.999887394962806e-05,
      "loss": 121.3336,
      "step": 25760
    },
    {
      "epoch": 0.10411406085238589,
      "grad_norm": 2037.351806640625,
      "learning_rate": 3.999884412205753e-05,
      "loss": 112.482,
      "step": 25770
    },
    {
      "epoch": 0.10415446211775353,
      "grad_norm": 1124.732177734375,
      "learning_rate": 3.9998813904599234e-05,
      "loss": 79.0539,
      "step": 25780
    },
    {
      "epoch": 0.10419486338312116,
      "grad_norm": 860.0994262695312,
      "learning_rate": 3.999878329725377e-05,
      "loss": 134.404,
      "step": 25790
    },
    {
      "epoch": 0.10423526464848878,
      "grad_norm": 538.1453247070312,
      "learning_rate": 3.9998752300021736e-05,
      "loss": 148.6289,
      "step": 25800
    },
    {
      "epoch": 0.10427566591385642,
      "grad_norm": 441.525634765625,
      "learning_rate": 3.999872091290372e-05,
      "loss": 86.7214,
      "step": 25810
    },
    {
      "epoch": 0.10431606717922405,
      "grad_norm": 660.6888427734375,
      "learning_rate": 3.999868913590034e-05,
      "loss": 135.1893,
      "step": 25820
    },
    {
      "epoch": 0.10435646844459169,
      "grad_norm": 567.1407470703125,
      "learning_rate": 3.999865696901223e-05,
      "loss": 122.7115,
      "step": 25830
    },
    {
      "epoch": 0.10439686970995932,
      "grad_norm": 755.7587890625,
      "learning_rate": 3.9998624412240004e-05,
      "loss": 125.0915,
      "step": 25840
    },
    {
      "epoch": 0.10443727097532694,
      "grad_norm": 837.2969360351562,
      "learning_rate": 3.999859146558429e-05,
      "loss": 86.853,
      "step": 25850
    },
    {
      "epoch": 0.10447767224069458,
      "grad_norm": 694.6524658203125,
      "learning_rate": 3.999855812904575e-05,
      "loss": 91.0812,
      "step": 25860
    },
    {
      "epoch": 0.10451807350606221,
      "grad_norm": 0.0,
      "learning_rate": 3.999852440262502e-05,
      "loss": 89.327,
      "step": 25870
    },
    {
      "epoch": 0.10455847477142984,
      "grad_norm": 1052.8719482421875,
      "learning_rate": 3.999849028632276e-05,
      "loss": 119.6729,
      "step": 25880
    },
    {
      "epoch": 0.10459887603679748,
      "grad_norm": 1102.21875,
      "learning_rate": 3.9998455780139635e-05,
      "loss": 112.6542,
      "step": 25890
    },
    {
      "epoch": 0.1046392773021651,
      "grad_norm": 1052.8319091796875,
      "learning_rate": 3.999842088407633e-05,
      "loss": 110.996,
      "step": 25900
    },
    {
      "epoch": 0.10467967856753274,
      "grad_norm": 2375.400146484375,
      "learning_rate": 3.99983855981335e-05,
      "loss": 110.829,
      "step": 25910
    },
    {
      "epoch": 0.10472007983290037,
      "grad_norm": 1199.724609375,
      "learning_rate": 3.9998349922311856e-05,
      "loss": 71.8274,
      "step": 25920
    },
    {
      "epoch": 0.104760481098268,
      "grad_norm": 581.4923706054688,
      "learning_rate": 3.999831385661208e-05,
      "loss": 82.232,
      "step": 25930
    },
    {
      "epoch": 0.10480088236363563,
      "grad_norm": 635.7989501953125,
      "learning_rate": 3.999827740103487e-05,
      "loss": 127.8533,
      "step": 25940
    },
    {
      "epoch": 0.10484128362900326,
      "grad_norm": 2129.2431640625,
      "learning_rate": 3.9998240555580954e-05,
      "loss": 129.5675,
      "step": 25950
    },
    {
      "epoch": 0.10488168489437089,
      "grad_norm": 1494.1981201171875,
      "learning_rate": 3.999820332025105e-05,
      "loss": 107.2558,
      "step": 25960
    },
    {
      "epoch": 0.10492208615973853,
      "grad_norm": 1150.99560546875,
      "learning_rate": 3.999816569504587e-05,
      "loss": 116.159,
      "step": 25970
    },
    {
      "epoch": 0.10496248742510615,
      "grad_norm": 1286.3641357421875,
      "learning_rate": 3.9998127679966154e-05,
      "loss": 122.6338,
      "step": 25980
    },
    {
      "epoch": 0.10500288869047379,
      "grad_norm": 1373.02685546875,
      "learning_rate": 3.999808927501264e-05,
      "loss": 114.1198,
      "step": 25990
    },
    {
      "epoch": 0.10504328995584142,
      "grad_norm": 915.142333984375,
      "learning_rate": 3.999805048018609e-05,
      "loss": 100.8753,
      "step": 26000
    },
    {
      "epoch": 0.10508369122120904,
      "grad_norm": 1715.4393310546875,
      "learning_rate": 3.9998011295487236e-05,
      "loss": 88.3989,
      "step": 26010
    },
    {
      "epoch": 0.10512409248657668,
      "grad_norm": 721.6860961914062,
      "learning_rate": 3.999797172091687e-05,
      "loss": 62.4881,
      "step": 26020
    },
    {
      "epoch": 0.10516449375194431,
      "grad_norm": 642.193359375,
      "learning_rate": 3.9997931756475744e-05,
      "loss": 96.4464,
      "step": 26030
    },
    {
      "epoch": 0.10520489501731194,
      "grad_norm": 3102.461181640625,
      "learning_rate": 3.9997891402164644e-05,
      "loss": 111.057,
      "step": 26040
    },
    {
      "epoch": 0.10524529628267958,
      "grad_norm": 1137.797119140625,
      "learning_rate": 3.999785065798435e-05,
      "loss": 66.1138,
      "step": 26050
    },
    {
      "epoch": 0.1052856975480472,
      "grad_norm": 1059.8470458984375,
      "learning_rate": 3.999780952393566e-05,
      "loss": 110.6024,
      "step": 26060
    },
    {
      "epoch": 0.10532609881341484,
      "grad_norm": 1064.2550048828125,
      "learning_rate": 3.999776800001939e-05,
      "loss": 101.4643,
      "step": 26070
    },
    {
      "epoch": 0.10536650007878247,
      "grad_norm": 1279.454833984375,
      "learning_rate": 3.9997726086236325e-05,
      "loss": 125.3454,
      "step": 26080
    },
    {
      "epoch": 0.1054069013441501,
      "grad_norm": 1695.6534423828125,
      "learning_rate": 3.999768378258731e-05,
      "loss": 103.3218,
      "step": 26090
    },
    {
      "epoch": 0.10544730260951773,
      "grad_norm": 841.222412109375,
      "learning_rate": 3.999764108907314e-05,
      "loss": 68.5106,
      "step": 26100
    },
    {
      "epoch": 0.10548770387488536,
      "grad_norm": 1536.8846435546875,
      "learning_rate": 3.999759800569467e-05,
      "loss": 94.1219,
      "step": 26110
    },
    {
      "epoch": 0.10552810514025299,
      "grad_norm": 1104.2398681640625,
      "learning_rate": 3.999755453245272e-05,
      "loss": 61.2582,
      "step": 26120
    },
    {
      "epoch": 0.10556850640562063,
      "grad_norm": 1145.83056640625,
      "learning_rate": 3.999751066934816e-05,
      "loss": 87.8465,
      "step": 26130
    },
    {
      "epoch": 0.10560890767098825,
      "grad_norm": 896.1638793945312,
      "learning_rate": 3.999746641638183e-05,
      "loss": 77.771,
      "step": 26140
    },
    {
      "epoch": 0.10564930893635589,
      "grad_norm": 1831.322509765625,
      "learning_rate": 3.9997421773554596e-05,
      "loss": 162.7093,
      "step": 26150
    },
    {
      "epoch": 0.10568971020172352,
      "grad_norm": 495.71588134765625,
      "learning_rate": 3.9997376740867334e-05,
      "loss": 122.5265,
      "step": 26160
    },
    {
      "epoch": 0.10573011146709115,
      "grad_norm": 1053.8997802734375,
      "learning_rate": 3.9997331318320906e-05,
      "loss": 120.7888,
      "step": 26170
    },
    {
      "epoch": 0.10577051273245879,
      "grad_norm": 826.036865234375,
      "learning_rate": 3.9997285505916215e-05,
      "loss": 84.5706,
      "step": 26180
    },
    {
      "epoch": 0.10581091399782641,
      "grad_norm": 580.434814453125,
      "learning_rate": 3.999723930365415e-05,
      "loss": 74.2667,
      "step": 26190
    },
    {
      "epoch": 0.10585131526319404,
      "grad_norm": 299.5496520996094,
      "learning_rate": 3.999719271153561e-05,
      "loss": 92.8395,
      "step": 26200
    },
    {
      "epoch": 0.10589171652856168,
      "grad_norm": 1334.234375,
      "learning_rate": 3.9997145729561505e-05,
      "loss": 100.1445,
      "step": 26210
    },
    {
      "epoch": 0.1059321177939293,
      "grad_norm": 796.839599609375,
      "learning_rate": 3.999709835773274e-05,
      "loss": 144.5433,
      "step": 26220
    },
    {
      "epoch": 0.10597251905929694,
      "grad_norm": 778.1115112304688,
      "learning_rate": 3.9997050596050255e-05,
      "loss": 108.5845,
      "step": 26230
    },
    {
      "epoch": 0.10601292032466457,
      "grad_norm": 6468.21044921875,
      "learning_rate": 3.999700244451497e-05,
      "loss": 81.2048,
      "step": 26240
    },
    {
      "epoch": 0.1060533215900322,
      "grad_norm": 671.6563720703125,
      "learning_rate": 3.999695390312783e-05,
      "loss": 54.482,
      "step": 26250
    },
    {
      "epoch": 0.10609372285539984,
      "grad_norm": 782.2870483398438,
      "learning_rate": 3.999690497188978e-05,
      "loss": 49.7613,
      "step": 26260
    },
    {
      "epoch": 0.10613412412076746,
      "grad_norm": 1402.428466796875,
      "learning_rate": 3.999685565080176e-05,
      "loss": 118.4322,
      "step": 26270
    },
    {
      "epoch": 0.10617452538613509,
      "grad_norm": 563.5989990234375,
      "learning_rate": 3.999680593986475e-05,
      "loss": 84.3666,
      "step": 26280
    },
    {
      "epoch": 0.10621492665150273,
      "grad_norm": 928.5982055664062,
      "learning_rate": 3.999675583907972e-05,
      "loss": 134.5598,
      "step": 26290
    },
    {
      "epoch": 0.10625532791687035,
      "grad_norm": 1692.81640625,
      "learning_rate": 3.999670534844763e-05,
      "loss": 126.942,
      "step": 26300
    },
    {
      "epoch": 0.106295729182238,
      "grad_norm": 3591.871826171875,
      "learning_rate": 3.9996654467969485e-05,
      "loss": 92.5221,
      "step": 26310
    },
    {
      "epoch": 0.10633613044760562,
      "grad_norm": 2086.629150390625,
      "learning_rate": 3.999660319764626e-05,
      "loss": 85.6505,
      "step": 26320
    },
    {
      "epoch": 0.10637653171297325,
      "grad_norm": 1024.0364990234375,
      "learning_rate": 3.9996551537478965e-05,
      "loss": 165.9923,
      "step": 26330
    },
    {
      "epoch": 0.10641693297834089,
      "grad_norm": 2205.33154296875,
      "learning_rate": 3.99964994874686e-05,
      "loss": 112.9845,
      "step": 26340
    },
    {
      "epoch": 0.10645733424370851,
      "grad_norm": 1261.9781494140625,
      "learning_rate": 3.9996447047616185e-05,
      "loss": 142.4679,
      "step": 26350
    },
    {
      "epoch": 0.10649773550907614,
      "grad_norm": 1418.99755859375,
      "learning_rate": 3.999639421792274e-05,
      "loss": 83.0303,
      "step": 26360
    },
    {
      "epoch": 0.10653813677444378,
      "grad_norm": 3501.380615234375,
      "learning_rate": 3.999634099838929e-05,
      "loss": 78.1086,
      "step": 26370
    },
    {
      "epoch": 0.1065785380398114,
      "grad_norm": 534.0908813476562,
      "learning_rate": 3.9996287389016876e-05,
      "loss": 111.2012,
      "step": 26380
    },
    {
      "epoch": 0.10661893930517904,
      "grad_norm": 771.1531982421875,
      "learning_rate": 3.999623338980655e-05,
      "loss": 147.8283,
      "step": 26390
    },
    {
      "epoch": 0.10665934057054667,
      "grad_norm": 927.004638671875,
      "learning_rate": 3.999617900075936e-05,
      "loss": 87.6993,
      "step": 26400
    },
    {
      "epoch": 0.1066997418359143,
      "grad_norm": 1738.247802734375,
      "learning_rate": 3.9996124221876364e-05,
      "loss": 117.6841,
      "step": 26410
    },
    {
      "epoch": 0.10674014310128194,
      "grad_norm": 1035.49609375,
      "learning_rate": 3.9996069053158626e-05,
      "loss": 182.0309,
      "step": 26420
    },
    {
      "epoch": 0.10678054436664956,
      "grad_norm": 1327.6573486328125,
      "learning_rate": 3.9996013494607234e-05,
      "loss": 107.2708,
      "step": 26430
    },
    {
      "epoch": 0.10682094563201719,
      "grad_norm": 1198.87255859375,
      "learning_rate": 3.999595754622326e-05,
      "loss": 130.0626,
      "step": 26440
    },
    {
      "epoch": 0.10686134689738483,
      "grad_norm": 0.0,
      "learning_rate": 3.99959012080078e-05,
      "loss": 86.2307,
      "step": 26450
    },
    {
      "epoch": 0.10690174816275246,
      "grad_norm": 1337.4168701171875,
      "learning_rate": 3.999584447996196e-05,
      "loss": 122.7532,
      "step": 26460
    },
    {
      "epoch": 0.1069421494281201,
      "grad_norm": 813.623046875,
      "learning_rate": 3.9995787362086824e-05,
      "loss": 61.1589,
      "step": 26470
    },
    {
      "epoch": 0.10698255069348772,
      "grad_norm": 1643.9227294921875,
      "learning_rate": 3.9995729854383526e-05,
      "loss": 112.8391,
      "step": 26480
    },
    {
      "epoch": 0.10702295195885535,
      "grad_norm": 1001.5737915039062,
      "learning_rate": 3.999567195685318e-05,
      "loss": 137.1343,
      "step": 26490
    },
    {
      "epoch": 0.10706335322422299,
      "grad_norm": 768.2099609375,
      "learning_rate": 3.999561366949691e-05,
      "loss": 101.1197,
      "step": 26500
    },
    {
      "epoch": 0.10710375448959061,
      "grad_norm": 1358.92041015625,
      "learning_rate": 3.9995554992315863e-05,
      "loss": 128.1427,
      "step": 26510
    },
    {
      "epoch": 0.10714415575495824,
      "grad_norm": 683.9628295898438,
      "learning_rate": 3.999549592531118e-05,
      "loss": 77.3999,
      "step": 26520
    },
    {
      "epoch": 0.10718455702032588,
      "grad_norm": 1114.36376953125,
      "learning_rate": 3.9995436468484e-05,
      "loss": 122.6443,
      "step": 26530
    },
    {
      "epoch": 0.1072249582856935,
      "grad_norm": 867.2285766601562,
      "learning_rate": 3.999537662183549e-05,
      "loss": 89.0948,
      "step": 26540
    },
    {
      "epoch": 0.10726535955106115,
      "grad_norm": 1265.471435546875,
      "learning_rate": 3.9995316385366825e-05,
      "loss": 90.6688,
      "step": 26550
    },
    {
      "epoch": 0.10730576081642877,
      "grad_norm": 1176.1046142578125,
      "learning_rate": 3.999525575907918e-05,
      "loss": 68.6942,
      "step": 26560
    },
    {
      "epoch": 0.1073461620817964,
      "grad_norm": 1456.6717529296875,
      "learning_rate": 3.999519474297372e-05,
      "loss": 94.2505,
      "step": 26570
    },
    {
      "epoch": 0.10738656334716404,
      "grad_norm": 942.738037109375,
      "learning_rate": 3.9995133337051645e-05,
      "loss": 158.1714,
      "step": 26580
    },
    {
      "epoch": 0.10742696461253166,
      "grad_norm": 624.3585205078125,
      "learning_rate": 3.999507154131415e-05,
      "loss": 128.1745,
      "step": 26590
    },
    {
      "epoch": 0.10746736587789929,
      "grad_norm": 1895.4283447265625,
      "learning_rate": 3.999500935576245e-05,
      "loss": 126.8323,
      "step": 26600
    },
    {
      "epoch": 0.10750776714326693,
      "grad_norm": 1229.0679931640625,
      "learning_rate": 3.999494678039774e-05,
      "loss": 116.0763,
      "step": 26610
    },
    {
      "epoch": 0.10754816840863456,
      "grad_norm": 417.828125,
      "learning_rate": 3.999488381522125e-05,
      "loss": 73.9217,
      "step": 26620
    },
    {
      "epoch": 0.1075885696740022,
      "grad_norm": 520.174560546875,
      "learning_rate": 3.999482046023421e-05,
      "loss": 99.6772,
      "step": 26630
    },
    {
      "epoch": 0.10762897093936982,
      "grad_norm": 1039.8192138671875,
      "learning_rate": 3.9994756715437846e-05,
      "loss": 143.6114,
      "step": 26640
    },
    {
      "epoch": 0.10766937220473745,
      "grad_norm": 6462.5771484375,
      "learning_rate": 3.9994692580833406e-05,
      "loss": 98.8883,
      "step": 26650
    },
    {
      "epoch": 0.10770977347010509,
      "grad_norm": 2129.985107421875,
      "learning_rate": 3.999462805642214e-05,
      "loss": 106.4062,
      "step": 26660
    },
    {
      "epoch": 0.10775017473547271,
      "grad_norm": 2644.1337890625,
      "learning_rate": 3.999456314220531e-05,
      "loss": 123.5466,
      "step": 26670
    },
    {
      "epoch": 0.10779057600084034,
      "grad_norm": 6350.77001953125,
      "learning_rate": 3.9994497838184173e-05,
      "loss": 96.4797,
      "step": 26680
    },
    {
      "epoch": 0.10783097726620798,
      "grad_norm": 1997.462890625,
      "learning_rate": 3.9994432144360014e-05,
      "loss": 107.3564,
      "step": 26690
    },
    {
      "epoch": 0.1078713785315756,
      "grad_norm": 1046.4324951171875,
      "learning_rate": 3.99943660607341e-05,
      "loss": 95.4632,
      "step": 26700
    },
    {
      "epoch": 0.10791177979694325,
      "grad_norm": 1082.221435546875,
      "learning_rate": 3.9994299587307724e-05,
      "loss": 73.5173,
      "step": 26710
    },
    {
      "epoch": 0.10795218106231087,
      "grad_norm": 475.9582824707031,
      "learning_rate": 3.999423272408219e-05,
      "loss": 118.0872,
      "step": 26720
    },
    {
      "epoch": 0.1079925823276785,
      "grad_norm": 2606.221435546875,
      "learning_rate": 3.9994165471058795e-05,
      "loss": 142.7932,
      "step": 26730
    },
    {
      "epoch": 0.10803298359304614,
      "grad_norm": 987.0201416015625,
      "learning_rate": 3.999409782823884e-05,
      "loss": 121.8413,
      "step": 26740
    },
    {
      "epoch": 0.10807338485841377,
      "grad_norm": 979.9131469726562,
      "learning_rate": 3.999402979562367e-05,
      "loss": 107.7124,
      "step": 26750
    },
    {
      "epoch": 0.10811378612378139,
      "grad_norm": 751.9520263671875,
      "learning_rate": 3.9993961373214585e-05,
      "loss": 91.4504,
      "step": 26760
    },
    {
      "epoch": 0.10815418738914903,
      "grad_norm": 466.2114562988281,
      "learning_rate": 3.9993892561012935e-05,
      "loss": 129.0346,
      "step": 26770
    },
    {
      "epoch": 0.10819458865451666,
      "grad_norm": 1181.390380859375,
      "learning_rate": 3.999382335902005e-05,
      "loss": 125.1326,
      "step": 26780
    },
    {
      "epoch": 0.1082349899198843,
      "grad_norm": 914.0978393554688,
      "learning_rate": 3.999375376723729e-05,
      "loss": 117.911,
      "step": 26790
    },
    {
      "epoch": 0.10827539118525192,
      "grad_norm": 1160.3109130859375,
      "learning_rate": 3.9993683785666e-05,
      "loss": 94.7705,
      "step": 26800
    },
    {
      "epoch": 0.10831579245061955,
      "grad_norm": 766.2025146484375,
      "learning_rate": 3.999361341430756e-05,
      "loss": 103.3526,
      "step": 26810
    },
    {
      "epoch": 0.10835619371598719,
      "grad_norm": 2005.7353515625,
      "learning_rate": 3.999354265316333e-05,
      "loss": 106.9616,
      "step": 26820
    },
    {
      "epoch": 0.10839659498135482,
      "grad_norm": 2113.0107421875,
      "learning_rate": 3.999347150223469e-05,
      "loss": 117.2856,
      "step": 26830
    },
    {
      "epoch": 0.10843699624672244,
      "grad_norm": 1407.716064453125,
      "learning_rate": 3.999339996152303e-05,
      "loss": 72.8703,
      "step": 26840
    },
    {
      "epoch": 0.10847739751209008,
      "grad_norm": 332.46832275390625,
      "learning_rate": 3.999332803102974e-05,
      "loss": 97.1084,
      "step": 26850
    },
    {
      "epoch": 0.10851779877745771,
      "grad_norm": 519.922119140625,
      "learning_rate": 3.999325571075624e-05,
      "loss": 98.0303,
      "step": 26860
    },
    {
      "epoch": 0.10855820004282535,
      "grad_norm": 401.651611328125,
      "learning_rate": 3.999318300070392e-05,
      "loss": 77.1453,
      "step": 26870
    },
    {
      "epoch": 0.10859860130819297,
      "grad_norm": 946.8197631835938,
      "learning_rate": 3.999310990087421e-05,
      "loss": 96.2507,
      "step": 26880
    },
    {
      "epoch": 0.1086390025735606,
      "grad_norm": 916.1583251953125,
      "learning_rate": 3.999303641126852e-05,
      "loss": 92.2482,
      "step": 26890
    },
    {
      "epoch": 0.10867940383892824,
      "grad_norm": 1386.3487548828125,
      "learning_rate": 3.999296253188829e-05,
      "loss": 168.3209,
      "step": 26900
    },
    {
      "epoch": 0.10871980510429587,
      "grad_norm": 1362.5413818359375,
      "learning_rate": 3.999288826273497e-05,
      "loss": 168.2882,
      "step": 26910
    },
    {
      "epoch": 0.10876020636966349,
      "grad_norm": 481.9889221191406,
      "learning_rate": 3.999281360381e-05,
      "loss": 83.7807,
      "step": 26920
    },
    {
      "epoch": 0.10880060763503113,
      "grad_norm": 785.8662719726562,
      "learning_rate": 3.999273855511483e-05,
      "loss": 123.8795,
      "step": 26930
    },
    {
      "epoch": 0.10884100890039876,
      "grad_norm": 958.2007446289062,
      "learning_rate": 3.999266311665094e-05,
      "loss": 75.8988,
      "step": 26940
    },
    {
      "epoch": 0.1088814101657664,
      "grad_norm": 676.339111328125,
      "learning_rate": 3.999258728841977e-05,
      "loss": 131.9098,
      "step": 26950
    },
    {
      "epoch": 0.10892181143113402,
      "grad_norm": 920.4100952148438,
      "learning_rate": 3.999251107042284e-05,
      "loss": 75.1306,
      "step": 26960
    },
    {
      "epoch": 0.10896221269650165,
      "grad_norm": 495.47821044921875,
      "learning_rate": 3.99924344626616e-05,
      "loss": 101.7487,
      "step": 26970
    },
    {
      "epoch": 0.10900261396186929,
      "grad_norm": 1311.8775634765625,
      "learning_rate": 3.999235746513757e-05,
      "loss": 94.5948,
      "step": 26980
    },
    {
      "epoch": 0.10904301522723692,
      "grad_norm": 706.6832275390625,
      "learning_rate": 3.999228007785222e-05,
      "loss": 110.3509,
      "step": 26990
    },
    {
      "epoch": 0.10908341649260454,
      "grad_norm": 350.2680358886719,
      "learning_rate": 3.999220230080709e-05,
      "loss": 82.0634,
      "step": 27000
    },
    {
      "epoch": 0.10912381775797218,
      "grad_norm": 863.8991088867188,
      "learning_rate": 3.999212413400368e-05,
      "loss": 110.7725,
      "step": 27010
    },
    {
      "epoch": 0.10916421902333981,
      "grad_norm": 1048.016845703125,
      "learning_rate": 3.999204557744352e-05,
      "loss": 111.5144,
      "step": 27020
    },
    {
      "epoch": 0.10920462028870745,
      "grad_norm": 555.6709594726562,
      "learning_rate": 3.999196663112813e-05,
      "loss": 167.0167,
      "step": 27030
    },
    {
      "epoch": 0.10924502155407508,
      "grad_norm": 695.7076416015625,
      "learning_rate": 3.9991887295059065e-05,
      "loss": 103.8333,
      "step": 27040
    },
    {
      "epoch": 0.1092854228194427,
      "grad_norm": 619.1875,
      "learning_rate": 3.999180756923787e-05,
      "loss": 171.0366,
      "step": 27050
    },
    {
      "epoch": 0.10932582408481034,
      "grad_norm": 649.6815185546875,
      "learning_rate": 3.999172745366609e-05,
      "loss": 98.1232,
      "step": 27060
    },
    {
      "epoch": 0.10936622535017797,
      "grad_norm": 709.5027465820312,
      "learning_rate": 3.999164694834529e-05,
      "loss": 130.3714,
      "step": 27070
    },
    {
      "epoch": 0.1094066266155456,
      "grad_norm": 1100.6815185546875,
      "learning_rate": 3.999156605327704e-05,
      "loss": 106.8556,
      "step": 27080
    },
    {
      "epoch": 0.10944702788091323,
      "grad_norm": 726.9677124023438,
      "learning_rate": 3.999148476846292e-05,
      "loss": 86.2317,
      "step": 27090
    },
    {
      "epoch": 0.10948742914628086,
      "grad_norm": 560.1898803710938,
      "learning_rate": 3.9991403093904505e-05,
      "loss": 171.8191,
      "step": 27100
    },
    {
      "epoch": 0.1095278304116485,
      "grad_norm": 1026.121337890625,
      "learning_rate": 3.99913210296034e-05,
      "loss": 89.2664,
      "step": 27110
    },
    {
      "epoch": 0.10956823167701613,
      "grad_norm": 1080.0189208984375,
      "learning_rate": 3.99912385755612e-05,
      "loss": 61.5915,
      "step": 27120
    },
    {
      "epoch": 0.10960863294238375,
      "grad_norm": 630.165771484375,
      "learning_rate": 3.9991155731779506e-05,
      "loss": 97.5475,
      "step": 27130
    },
    {
      "epoch": 0.10964903420775139,
      "grad_norm": 479.4427795410156,
      "learning_rate": 3.999107249825994e-05,
      "loss": 89.1548,
      "step": 27140
    },
    {
      "epoch": 0.10968943547311902,
      "grad_norm": 750.473876953125,
      "learning_rate": 3.999098887500413e-05,
      "loss": 105.348,
      "step": 27150
    },
    {
      "epoch": 0.10972983673848664,
      "grad_norm": 459.2093811035156,
      "learning_rate": 3.999090486201369e-05,
      "loss": 111.195,
      "step": 27160
    },
    {
      "epoch": 0.10977023800385428,
      "grad_norm": 956.58544921875,
      "learning_rate": 3.999082045929028e-05,
      "loss": 150.9714,
      "step": 27170
    },
    {
      "epoch": 0.10981063926922191,
      "grad_norm": 699.5205078125,
      "learning_rate": 3.999073566683552e-05,
      "loss": 87.5247,
      "step": 27180
    },
    {
      "epoch": 0.10985104053458955,
      "grad_norm": 944.4476928710938,
      "learning_rate": 3.999065048465108e-05,
      "loss": 116.8029,
      "step": 27190
    },
    {
      "epoch": 0.10989144179995718,
      "grad_norm": 837.1326904296875,
      "learning_rate": 3.9990564912738626e-05,
      "loss": 75.7266,
      "step": 27200
    },
    {
      "epoch": 0.1099318430653248,
      "grad_norm": 778.825927734375,
      "learning_rate": 3.999047895109981e-05,
      "loss": 100.6665,
      "step": 27210
    },
    {
      "epoch": 0.10997224433069244,
      "grad_norm": 1490.7548828125,
      "learning_rate": 3.999039259973632e-05,
      "loss": 126.6425,
      "step": 27220
    },
    {
      "epoch": 0.11001264559606007,
      "grad_norm": 452.08575439453125,
      "learning_rate": 3.999030585864983e-05,
      "loss": 87.2329,
      "step": 27230
    },
    {
      "epoch": 0.1100530468614277,
      "grad_norm": 1867.3792724609375,
      "learning_rate": 3.999021872784203e-05,
      "loss": 94.7032,
      "step": 27240
    },
    {
      "epoch": 0.11009344812679533,
      "grad_norm": 620.962158203125,
      "learning_rate": 3.9990131207314634e-05,
      "loss": 72.1691,
      "step": 27250
    },
    {
      "epoch": 0.11013384939216296,
      "grad_norm": 932.4627075195312,
      "learning_rate": 3.9990043297069335e-05,
      "loss": 101.9201,
      "step": 27260
    },
    {
      "epoch": 0.1101742506575306,
      "grad_norm": 1095.8741455078125,
      "learning_rate": 3.998995499710785e-05,
      "loss": 100.2253,
      "step": 27270
    },
    {
      "epoch": 0.11021465192289823,
      "grad_norm": 1374.950927734375,
      "learning_rate": 3.99898663074319e-05,
      "loss": 81.8334,
      "step": 27280
    },
    {
      "epoch": 0.11025505318826585,
      "grad_norm": 2230.595947265625,
      "learning_rate": 3.9989777228043216e-05,
      "loss": 95.7221,
      "step": 27290
    },
    {
      "epoch": 0.11029545445363349,
      "grad_norm": 1166.83447265625,
      "learning_rate": 3.998968775894354e-05,
      "loss": 114.2452,
      "step": 27300
    },
    {
      "epoch": 0.11033585571900112,
      "grad_norm": 705.6727905273438,
      "learning_rate": 3.9989597900134594e-05,
      "loss": 69.3813,
      "step": 27310
    },
    {
      "epoch": 0.11037625698436875,
      "grad_norm": 455.673583984375,
      "learning_rate": 3.998950765161816e-05,
      "loss": 145.5359,
      "step": 27320
    },
    {
      "epoch": 0.11041665824973639,
      "grad_norm": 1001.2382202148438,
      "learning_rate": 3.9989417013395975e-05,
      "loss": 106.7314,
      "step": 27330
    },
    {
      "epoch": 0.11045705951510401,
      "grad_norm": 1855.0413818359375,
      "learning_rate": 3.998932598546982e-05,
      "loss": 152.2735,
      "step": 27340
    },
    {
      "epoch": 0.11049746078047164,
      "grad_norm": 1260.8231201171875,
      "learning_rate": 3.998923456784146e-05,
      "loss": 136.3262,
      "step": 27350
    },
    {
      "epoch": 0.11053786204583928,
      "grad_norm": 888.2510375976562,
      "learning_rate": 3.998914276051269e-05,
      "loss": 115.822,
      "step": 27360
    },
    {
      "epoch": 0.1105782633112069,
      "grad_norm": 779.9644775390625,
      "learning_rate": 3.9989050563485276e-05,
      "loss": 163.1847,
      "step": 27370
    },
    {
      "epoch": 0.11061866457657454,
      "grad_norm": 531.3817749023438,
      "learning_rate": 3.998895797676103e-05,
      "loss": 86.5874,
      "step": 27380
    },
    {
      "epoch": 0.11065906584194217,
      "grad_norm": 574.7570190429688,
      "learning_rate": 3.9988865000341764e-05,
      "loss": 109.047,
      "step": 27390
    },
    {
      "epoch": 0.1106994671073098,
      "grad_norm": 603.2306518554688,
      "learning_rate": 3.998877163422929e-05,
      "loss": 114.8451,
      "step": 27400
    },
    {
      "epoch": 0.11073986837267744,
      "grad_norm": 1424.3984375,
      "learning_rate": 3.9988677878425414e-05,
      "loss": 116.6163,
      "step": 27410
    },
    {
      "epoch": 0.11078026963804506,
      "grad_norm": 3271.53369140625,
      "learning_rate": 3.998858373293198e-05,
      "loss": 124.9541,
      "step": 27420
    },
    {
      "epoch": 0.11082067090341269,
      "grad_norm": 1137.7222900390625,
      "learning_rate": 3.99884891977508e-05,
      "loss": 98.2931,
      "step": 27430
    },
    {
      "epoch": 0.11086107216878033,
      "grad_norm": 1314.94091796875,
      "learning_rate": 3.998839427288375e-05,
      "loss": 84.932,
      "step": 27440
    },
    {
      "epoch": 0.11090147343414795,
      "grad_norm": 981.7919311523438,
      "learning_rate": 3.998829895833265e-05,
      "loss": 121.3453,
      "step": 27450
    },
    {
      "epoch": 0.1109418746995156,
      "grad_norm": 778.75146484375,
      "learning_rate": 3.9988203254099373e-05,
      "loss": 75.516,
      "step": 27460
    },
    {
      "epoch": 0.11098227596488322,
      "grad_norm": 664.9393310546875,
      "learning_rate": 3.9988107160185785e-05,
      "loss": 113.4883,
      "step": 27470
    },
    {
      "epoch": 0.11102267723025085,
      "grad_norm": 2451.9697265625,
      "learning_rate": 3.998801067659376e-05,
      "loss": 108.3004,
      "step": 27480
    },
    {
      "epoch": 0.11106307849561849,
      "grad_norm": 645.4609375,
      "learning_rate": 3.998791380332517e-05,
      "loss": 87.737,
      "step": 27490
    },
    {
      "epoch": 0.11110347976098611,
      "grad_norm": 693.7180786132812,
      "learning_rate": 3.998781654038192e-05,
      "loss": 110.6939,
      "step": 27500
    },
    {
      "epoch": 0.11114388102635374,
      "grad_norm": 618.6651611328125,
      "learning_rate": 3.998771888776589e-05,
      "loss": 63.0742,
      "step": 27510
    },
    {
      "epoch": 0.11118428229172138,
      "grad_norm": 162.7868194580078,
      "learning_rate": 3.998762084547899e-05,
      "loss": 104.0014,
      "step": 27520
    },
    {
      "epoch": 0.111224683557089,
      "grad_norm": 653.8965454101562,
      "learning_rate": 3.9987522413523135e-05,
      "loss": 96.1498,
      "step": 27530
    },
    {
      "epoch": 0.11126508482245664,
      "grad_norm": 2130.4833984375,
      "learning_rate": 3.998742359190023e-05,
      "loss": 95.9789,
      "step": 27540
    },
    {
      "epoch": 0.11130548608782427,
      "grad_norm": 548.673583984375,
      "learning_rate": 3.998732438061222e-05,
      "loss": 137.2213,
      "step": 27550
    },
    {
      "epoch": 0.1113458873531919,
      "grad_norm": 745.030029296875,
      "learning_rate": 3.998722477966103e-05,
      "loss": 69.7114,
      "step": 27560
    },
    {
      "epoch": 0.11138628861855954,
      "grad_norm": 1384.381591796875,
      "learning_rate": 3.99871247890486e-05,
      "loss": 121.4083,
      "step": 27570
    },
    {
      "epoch": 0.11142668988392716,
      "grad_norm": 1384.610107421875,
      "learning_rate": 3.998702440877689e-05,
      "loss": 100.7045,
      "step": 27580
    },
    {
      "epoch": 0.11146709114929479,
      "grad_norm": 880.5833129882812,
      "learning_rate": 3.998692363884784e-05,
      "loss": 86.9645,
      "step": 27590
    },
    {
      "epoch": 0.11150749241466243,
      "grad_norm": 1050.4068603515625,
      "learning_rate": 3.998682247926343e-05,
      "loss": 160.0109,
      "step": 27600
    },
    {
      "epoch": 0.11154789368003006,
      "grad_norm": 2584.7294921875,
      "learning_rate": 3.998672093002562e-05,
      "loss": 109.3766,
      "step": 27610
    },
    {
      "epoch": 0.1115882949453977,
      "grad_norm": 1057.3648681640625,
      "learning_rate": 3.99866189911364e-05,
      "loss": 116.4955,
      "step": 27620
    },
    {
      "epoch": 0.11162869621076532,
      "grad_norm": 1380.513916015625,
      "learning_rate": 3.998651666259775e-05,
      "loss": 86.7301,
      "step": 27630
    },
    {
      "epoch": 0.11166909747613295,
      "grad_norm": 567.40869140625,
      "learning_rate": 3.998641394441167e-05,
      "loss": 68.0413,
      "step": 27640
    },
    {
      "epoch": 0.11170949874150059,
      "grad_norm": 1073.0047607421875,
      "learning_rate": 3.998631083658016e-05,
      "loss": 121.8724,
      "step": 27650
    },
    {
      "epoch": 0.11174990000686821,
      "grad_norm": 990.1610717773438,
      "learning_rate": 3.9986207339105235e-05,
      "loss": 97.6038,
      "step": 27660
    },
    {
      "epoch": 0.11179030127223584,
      "grad_norm": 630.7728881835938,
      "learning_rate": 3.99861034519889e-05,
      "loss": 96.9841,
      "step": 27670
    },
    {
      "epoch": 0.11183070253760348,
      "grad_norm": 1043.8477783203125,
      "learning_rate": 3.99859991752332e-05,
      "loss": 95.7591,
      "step": 27680
    },
    {
      "epoch": 0.1118711038029711,
      "grad_norm": 1339.642578125,
      "learning_rate": 3.998589450884014e-05,
      "loss": 86.0877,
      "step": 27690
    },
    {
      "epoch": 0.11191150506833875,
      "grad_norm": 1462.45751953125,
      "learning_rate": 3.998578945281179e-05,
      "loss": 105.2841,
      "step": 27700
    },
    {
      "epoch": 0.11195190633370637,
      "grad_norm": 1582.08251953125,
      "learning_rate": 3.998568400715018e-05,
      "loss": 93.621,
      "step": 27710
    },
    {
      "epoch": 0.111992307599074,
      "grad_norm": 853.9618530273438,
      "learning_rate": 3.998557817185737e-05,
      "loss": 67.4083,
      "step": 27720
    },
    {
      "epoch": 0.11203270886444164,
      "grad_norm": 4256.23876953125,
      "learning_rate": 3.998547194693543e-05,
      "loss": 128.0626,
      "step": 27730
    },
    {
      "epoch": 0.11207311012980926,
      "grad_norm": 687.2172241210938,
      "learning_rate": 3.9985365332386424e-05,
      "loss": 101.9704,
      "step": 27740
    },
    {
      "epoch": 0.11211351139517689,
      "grad_norm": 542.3268432617188,
      "learning_rate": 3.998525832821242e-05,
      "loss": 112.3464,
      "step": 27750
    },
    {
      "epoch": 0.11215391266054453,
      "grad_norm": 920.6582641601562,
      "learning_rate": 3.998515093441553e-05,
      "loss": 78.368,
      "step": 27760
    },
    {
      "epoch": 0.11219431392591216,
      "grad_norm": 968.70849609375,
      "learning_rate": 3.998504315099783e-05,
      "loss": 77.5449,
      "step": 27770
    },
    {
      "epoch": 0.1122347151912798,
      "grad_norm": 814.1373291015625,
      "learning_rate": 3.998493497796142e-05,
      "loss": 86.3444,
      "step": 27780
    },
    {
      "epoch": 0.11227511645664742,
      "grad_norm": 652.4821166992188,
      "learning_rate": 3.998482641530842e-05,
      "loss": 80.1591,
      "step": 27790
    },
    {
      "epoch": 0.11231551772201505,
      "grad_norm": 1460.57373046875,
      "learning_rate": 3.998471746304094e-05,
      "loss": 133.1036,
      "step": 27800
    },
    {
      "epoch": 0.11235591898738269,
      "grad_norm": 1677.220947265625,
      "learning_rate": 3.99846081211611e-05,
      "loss": 122.758,
      "step": 27810
    },
    {
      "epoch": 0.11239632025275031,
      "grad_norm": 1221.714111328125,
      "learning_rate": 3.998449838967104e-05,
      "loss": 97.5468,
      "step": 27820
    },
    {
      "epoch": 0.11243672151811794,
      "grad_norm": 1173.8179931640625,
      "learning_rate": 3.9984388268572894e-05,
      "loss": 116.5459,
      "step": 27830
    },
    {
      "epoch": 0.11247712278348558,
      "grad_norm": 1088.1904296875,
      "learning_rate": 3.998427775786881e-05,
      "loss": 70.6181,
      "step": 27840
    },
    {
      "epoch": 0.1125175240488532,
      "grad_norm": 749.2481079101562,
      "learning_rate": 3.998416685756094e-05,
      "loss": 146.586,
      "step": 27850
    },
    {
      "epoch": 0.11255792531422085,
      "grad_norm": 799.7578735351562,
      "learning_rate": 3.998405556765145e-05,
      "loss": 117.1241,
      "step": 27860
    },
    {
      "epoch": 0.11259832657958847,
      "grad_norm": 758.5236206054688,
      "learning_rate": 3.9983943888142505e-05,
      "loss": 90.8864,
      "step": 27870
    },
    {
      "epoch": 0.1126387278449561,
      "grad_norm": 3251.1787109375,
      "learning_rate": 3.998383181903629e-05,
      "loss": 110.5195,
      "step": 27880
    },
    {
      "epoch": 0.11267912911032374,
      "grad_norm": 805.9417114257812,
      "learning_rate": 3.9983719360334985e-05,
      "loss": 68.4453,
      "step": 27890
    },
    {
      "epoch": 0.11271953037569137,
      "grad_norm": 721.3700561523438,
      "learning_rate": 3.9983606512040786e-05,
      "loss": 108.295,
      "step": 27900
    },
    {
      "epoch": 0.11275993164105899,
      "grad_norm": 805.6432495117188,
      "learning_rate": 3.998349327415588e-05,
      "loss": 72.9431,
      "step": 27910
    },
    {
      "epoch": 0.11280033290642663,
      "grad_norm": 1314.1534423828125,
      "learning_rate": 3.998337964668249e-05,
      "loss": 89.8111,
      "step": 27920
    },
    {
      "epoch": 0.11284073417179426,
      "grad_norm": 616.1529541015625,
      "learning_rate": 3.998326562962283e-05,
      "loss": 94.4337,
      "step": 27930
    },
    {
      "epoch": 0.1128811354371619,
      "grad_norm": 3171.981201171875,
      "learning_rate": 3.998315122297911e-05,
      "loss": 99.0224,
      "step": 27940
    },
    {
      "epoch": 0.11292153670252952,
      "grad_norm": 817.9891967773438,
      "learning_rate": 3.9983036426753574e-05,
      "loss": 162.2938,
      "step": 27950
    },
    {
      "epoch": 0.11296193796789715,
      "grad_norm": 1005.9824829101562,
      "learning_rate": 3.998292124094845e-05,
      "loss": 87.353,
      "step": 27960
    },
    {
      "epoch": 0.11300233923326479,
      "grad_norm": 773.955810546875,
      "learning_rate": 3.9982805665566e-05,
      "loss": 137.2771,
      "step": 27970
    },
    {
      "epoch": 0.11304274049863242,
      "grad_norm": 2464.9912109375,
      "learning_rate": 3.998268970060846e-05,
      "loss": 78.9946,
      "step": 27980
    },
    {
      "epoch": 0.11308314176400004,
      "grad_norm": 985.4806518554688,
      "learning_rate": 3.998257334607809e-05,
      "loss": 152.6931,
      "step": 27990
    },
    {
      "epoch": 0.11312354302936768,
      "grad_norm": 1514.001953125,
      "learning_rate": 3.998245660197717e-05,
      "loss": 101.2416,
      "step": 28000
    },
    {
      "epoch": 0.11316394429473531,
      "grad_norm": 784.7289428710938,
      "learning_rate": 3.9982339468307974e-05,
      "loss": 98.5727,
      "step": 28010
    },
    {
      "epoch": 0.11320434556010295,
      "grad_norm": 1584.1505126953125,
      "learning_rate": 3.998222194507277e-05,
      "loss": 90.057,
      "step": 28020
    },
    {
      "epoch": 0.11324474682547057,
      "grad_norm": 489.11767578125,
      "learning_rate": 3.9982104032273875e-05,
      "loss": 112.8647,
      "step": 28030
    },
    {
      "epoch": 0.1132851480908382,
      "grad_norm": 988.647705078125,
      "learning_rate": 3.998198572991357e-05,
      "loss": 118.9399,
      "step": 28040
    },
    {
      "epoch": 0.11332554935620584,
      "grad_norm": 1578.1778564453125,
      "learning_rate": 3.998186703799417e-05,
      "loss": 133.543,
      "step": 28050
    },
    {
      "epoch": 0.11336595062157347,
      "grad_norm": 1059.1217041015625,
      "learning_rate": 3.998174795651798e-05,
      "loss": 78.5302,
      "step": 28060
    },
    {
      "epoch": 0.11340635188694109,
      "grad_norm": 1273.186279296875,
      "learning_rate": 3.998162848548733e-05,
      "loss": 176.5455,
      "step": 28070
    },
    {
      "epoch": 0.11344675315230873,
      "grad_norm": 642.8786010742188,
      "learning_rate": 3.998150862490453e-05,
      "loss": 83.828,
      "step": 28080
    },
    {
      "epoch": 0.11348715441767636,
      "grad_norm": 1636.2587890625,
      "learning_rate": 3.998138837477195e-05,
      "loss": 184.4266,
      "step": 28090
    },
    {
      "epoch": 0.113527555683044,
      "grad_norm": 778.3223876953125,
      "learning_rate": 3.998126773509191e-05,
      "loss": 118.4609,
      "step": 28100
    },
    {
      "epoch": 0.11356795694841162,
      "grad_norm": 882.5321044921875,
      "learning_rate": 3.998114670586678e-05,
      "loss": 98.4019,
      "step": 28110
    },
    {
      "epoch": 0.11360835821377925,
      "grad_norm": 2727.796630859375,
      "learning_rate": 3.998102528709889e-05,
      "loss": 99.3129,
      "step": 28120
    },
    {
      "epoch": 0.11364875947914689,
      "grad_norm": 925.4190063476562,
      "learning_rate": 3.998090347879063e-05,
      "loss": 99.0337,
      "step": 28130
    },
    {
      "epoch": 0.11368916074451452,
      "grad_norm": 765.3026123046875,
      "learning_rate": 3.998078128094437e-05,
      "loss": 93.1959,
      "step": 28140
    },
    {
      "epoch": 0.11372956200988214,
      "grad_norm": 675.1519165039062,
      "learning_rate": 3.99806586935625e-05,
      "loss": 79.8641,
      "step": 28150
    },
    {
      "epoch": 0.11376996327524978,
      "grad_norm": 854.69921875,
      "learning_rate": 3.9980535716647394e-05,
      "loss": 121.8348,
      "step": 28160
    },
    {
      "epoch": 0.11381036454061741,
      "grad_norm": 759.4384155273438,
      "learning_rate": 3.998041235020146e-05,
      "loss": 103.8514,
      "step": 28170
    },
    {
      "epoch": 0.11385076580598505,
      "grad_norm": 1522.1177978515625,
      "learning_rate": 3.99802885942271e-05,
      "loss": 110.5345,
      "step": 28180
    },
    {
      "epoch": 0.11389116707135268,
      "grad_norm": 1211.49072265625,
      "learning_rate": 3.998016444872673e-05,
      "loss": 119.6895,
      "step": 28190
    },
    {
      "epoch": 0.1139315683367203,
      "grad_norm": 1361.937255859375,
      "learning_rate": 3.998003991370277e-05,
      "loss": 92.2974,
      "step": 28200
    },
    {
      "epoch": 0.11397196960208794,
      "grad_norm": 818.15380859375,
      "learning_rate": 3.9979914989157634e-05,
      "loss": 87.3831,
      "step": 28210
    },
    {
      "epoch": 0.11401237086745557,
      "grad_norm": 793.6883544921875,
      "learning_rate": 3.997978967509378e-05,
      "loss": 77.2384,
      "step": 28220
    },
    {
      "epoch": 0.1140527721328232,
      "grad_norm": 1113.99267578125,
      "learning_rate": 3.997966397151364e-05,
      "loss": 92.0928,
      "step": 28230
    },
    {
      "epoch": 0.11409317339819083,
      "grad_norm": 824.6795043945312,
      "learning_rate": 3.997953787841965e-05,
      "loss": 112.8881,
      "step": 28240
    },
    {
      "epoch": 0.11413357466355846,
      "grad_norm": 890.2943725585938,
      "learning_rate": 3.99794113958143e-05,
      "loss": 94.8603,
      "step": 28250
    },
    {
      "epoch": 0.1141739759289261,
      "grad_norm": 859.4837036132812,
      "learning_rate": 3.997928452370003e-05,
      "loss": 85.9278,
      "step": 28260
    },
    {
      "epoch": 0.11421437719429373,
      "grad_norm": 412.1380615234375,
      "learning_rate": 3.997915726207932e-05,
      "loss": 137.5683,
      "step": 28270
    },
    {
      "epoch": 0.11425477845966135,
      "grad_norm": 4836.41845703125,
      "learning_rate": 3.9979029610954664e-05,
      "loss": 159.7229,
      "step": 28280
    },
    {
      "epoch": 0.11429517972502899,
      "grad_norm": 1094.9395751953125,
      "learning_rate": 3.997890157032853e-05,
      "loss": 94.2975,
      "step": 28290
    },
    {
      "epoch": 0.11433558099039662,
      "grad_norm": 1191.2569580078125,
      "learning_rate": 3.997877314020343e-05,
      "loss": 75.3415,
      "step": 28300
    },
    {
      "epoch": 0.11437598225576424,
      "grad_norm": 915.8956298828125,
      "learning_rate": 3.9978644320581856e-05,
      "loss": 102.0962,
      "step": 28310
    },
    {
      "epoch": 0.11441638352113188,
      "grad_norm": 797.2686767578125,
      "learning_rate": 3.997851511146633e-05,
      "loss": 98.4335,
      "step": 28320
    },
    {
      "epoch": 0.11445678478649951,
      "grad_norm": 565.2205810546875,
      "learning_rate": 3.997838551285936e-05,
      "loss": 110.0382,
      "step": 28330
    },
    {
      "epoch": 0.11449718605186715,
      "grad_norm": 656.094482421875,
      "learning_rate": 3.9978255524763494e-05,
      "loss": 68.3598,
      "step": 28340
    },
    {
      "epoch": 0.11453758731723478,
      "grad_norm": 2733.058837890625,
      "learning_rate": 3.9978125147181235e-05,
      "loss": 134.7628,
      "step": 28350
    },
    {
      "epoch": 0.1145779885826024,
      "grad_norm": 508.31982421875,
      "learning_rate": 3.997799438011515e-05,
      "loss": 74.2418,
      "step": 28360
    },
    {
      "epoch": 0.11461838984797004,
      "grad_norm": 1044.4010009765625,
      "learning_rate": 3.9977863223567774e-05,
      "loss": 101.4478,
      "step": 28370
    },
    {
      "epoch": 0.11465879111333767,
      "grad_norm": 888.8732299804688,
      "learning_rate": 3.997773167754167e-05,
      "loss": 95.4757,
      "step": 28380
    },
    {
      "epoch": 0.1146991923787053,
      "grad_norm": 869.9403076171875,
      "learning_rate": 3.9977599742039404e-05,
      "loss": 101.585,
      "step": 28390
    },
    {
      "epoch": 0.11473959364407293,
      "grad_norm": 450.99981689453125,
      "learning_rate": 3.9977467417063544e-05,
      "loss": 102.3021,
      "step": 28400
    },
    {
      "epoch": 0.11477999490944056,
      "grad_norm": 1323.034423828125,
      "learning_rate": 3.9977334702616676e-05,
      "loss": 105.8671,
      "step": 28410
    },
    {
      "epoch": 0.1148203961748082,
      "grad_norm": 2043.1444091796875,
      "learning_rate": 3.997720159870137e-05,
      "loss": 101.5552,
      "step": 28420
    },
    {
      "epoch": 0.11486079744017583,
      "grad_norm": 1640.767822265625,
      "learning_rate": 3.997706810532025e-05,
      "loss": 107.5822,
      "step": 28430
    },
    {
      "epoch": 0.11490119870554345,
      "grad_norm": 643.0208740234375,
      "learning_rate": 3.9976934222475893e-05,
      "loss": 118.5226,
      "step": 28440
    },
    {
      "epoch": 0.11494159997091109,
      "grad_norm": 1271.7794189453125,
      "learning_rate": 3.997679995017092e-05,
      "loss": 59.5527,
      "step": 28450
    },
    {
      "epoch": 0.11498200123627872,
      "grad_norm": 1369.9146728515625,
      "learning_rate": 3.997666528840795e-05,
      "loss": 82.6735,
      "step": 28460
    },
    {
      "epoch": 0.11502240250164635,
      "grad_norm": 601.714599609375,
      "learning_rate": 3.99765302371896e-05,
      "loss": 82.4955,
      "step": 28470
    },
    {
      "epoch": 0.11506280376701399,
      "grad_norm": 1195.427001953125,
      "learning_rate": 3.9976394796518514e-05,
      "loss": 88.4767,
      "step": 28480
    },
    {
      "epoch": 0.11510320503238161,
      "grad_norm": 457.9436340332031,
      "learning_rate": 3.997625896639733e-05,
      "loss": 132.1176,
      "step": 28490
    },
    {
      "epoch": 0.11514360629774925,
      "grad_norm": 1108.717529296875,
      "learning_rate": 3.9976122746828684e-05,
      "loss": 130.0687,
      "step": 28500
    },
    {
      "epoch": 0.11518400756311688,
      "grad_norm": 1652.4522705078125,
      "learning_rate": 3.997598613781525e-05,
      "loss": 113.9347,
      "step": 28510
    },
    {
      "epoch": 0.1152244088284845,
      "grad_norm": 717.0134887695312,
      "learning_rate": 3.997584913935967e-05,
      "loss": 76.4573,
      "step": 28520
    },
    {
      "epoch": 0.11526481009385214,
      "grad_norm": 928.9058227539062,
      "learning_rate": 3.997571175146463e-05,
      "loss": 70.7099,
      "step": 28530
    },
    {
      "epoch": 0.11530521135921977,
      "grad_norm": 1893.921142578125,
      "learning_rate": 3.997557397413281e-05,
      "loss": 94.3326,
      "step": 28540
    },
    {
      "epoch": 0.1153456126245874,
      "grad_norm": 938.5257568359375,
      "learning_rate": 3.9975435807366895e-05,
      "loss": 117.9883,
      "step": 28550
    },
    {
      "epoch": 0.11538601388995504,
      "grad_norm": 2122.823486328125,
      "learning_rate": 3.997529725116957e-05,
      "loss": 82.8352,
      "step": 28560
    },
    {
      "epoch": 0.11542641515532266,
      "grad_norm": 1137.513671875,
      "learning_rate": 3.9975158305543536e-05,
      "loss": 121.7476,
      "step": 28570
    },
    {
      "epoch": 0.1154668164206903,
      "grad_norm": 964.6400756835938,
      "learning_rate": 3.9975018970491515e-05,
      "loss": 94.4993,
      "step": 28580
    },
    {
      "epoch": 0.11550721768605793,
      "grad_norm": 1079.1875,
      "learning_rate": 3.997487924601621e-05,
      "loss": 136.5345,
      "step": 28590
    },
    {
      "epoch": 0.11554761895142555,
      "grad_norm": 0.0,
      "learning_rate": 3.997473913212036e-05,
      "loss": 98.4004,
      "step": 28600
    },
    {
      "epoch": 0.1155880202167932,
      "grad_norm": 823.25830078125,
      "learning_rate": 3.9974598628806675e-05,
      "loss": 117.2009,
      "step": 28610
    },
    {
      "epoch": 0.11562842148216082,
      "grad_norm": 702.5995483398438,
      "learning_rate": 3.99744577360779e-05,
      "loss": 99.8464,
      "step": 28620
    },
    {
      "epoch": 0.11566882274752845,
      "grad_norm": 1018.5477294921875,
      "learning_rate": 3.9974316453936806e-05,
      "loss": 73.2368,
      "step": 28630
    },
    {
      "epoch": 0.11570922401289609,
      "grad_norm": 824.5650634765625,
      "learning_rate": 3.997417478238612e-05,
      "loss": 71.0984,
      "step": 28640
    },
    {
      "epoch": 0.11574962527826371,
      "grad_norm": 871.6005859375,
      "learning_rate": 3.9974032721428615e-05,
      "loss": 150.4988,
      "step": 28650
    },
    {
      "epoch": 0.11579002654363135,
      "grad_norm": 1187.3568115234375,
      "learning_rate": 3.9973890271067056e-05,
      "loss": 138.0883,
      "step": 28660
    },
    {
      "epoch": 0.11583042780899898,
      "grad_norm": 1488.8284912109375,
      "learning_rate": 3.997374743130423e-05,
      "loss": 119.0744,
      "step": 28670
    },
    {
      "epoch": 0.1158708290743666,
      "grad_norm": 1320.722900390625,
      "learning_rate": 3.9973604202142906e-05,
      "loss": 119.735,
      "step": 28680
    },
    {
      "epoch": 0.11591123033973424,
      "grad_norm": 2015.63818359375,
      "learning_rate": 3.997346058358589e-05,
      "loss": 130.6859,
      "step": 28690
    },
    {
      "epoch": 0.11595163160510187,
      "grad_norm": 1006.3882446289062,
      "learning_rate": 3.997331657563598e-05,
      "loss": 120.2585,
      "step": 28700
    },
    {
      "epoch": 0.1159920328704695,
      "grad_norm": 1254.012451171875,
      "learning_rate": 3.997317217829598e-05,
      "loss": 105.0112,
      "step": 28710
    },
    {
      "epoch": 0.11603243413583714,
      "grad_norm": 1289.52001953125,
      "learning_rate": 3.99730273915687e-05,
      "loss": 115.2411,
      "step": 28720
    },
    {
      "epoch": 0.11607283540120476,
      "grad_norm": 3730.40771484375,
      "learning_rate": 3.997288221545697e-05,
      "loss": 89.3371,
      "step": 28730
    },
    {
      "epoch": 0.1161132366665724,
      "grad_norm": 1199.5682373046875,
      "learning_rate": 3.997273664996361e-05,
      "loss": 116.5537,
      "step": 28740
    },
    {
      "epoch": 0.11615363793194003,
      "grad_norm": 372.0835876464844,
      "learning_rate": 3.9972590695091476e-05,
      "loss": 56.8714,
      "step": 28750
    },
    {
      "epoch": 0.11619403919730766,
      "grad_norm": 1533.7314453125,
      "learning_rate": 3.997244435084341e-05,
      "loss": 81.515,
      "step": 28760
    },
    {
      "epoch": 0.1162344404626753,
      "grad_norm": 3192.15185546875,
      "learning_rate": 3.997229761722225e-05,
      "loss": 127.6431,
      "step": 28770
    },
    {
      "epoch": 0.11627484172804292,
      "grad_norm": 1288.7010498046875,
      "learning_rate": 3.997215049423086e-05,
      "loss": 85.945,
      "step": 28780
    },
    {
      "epoch": 0.11631524299341055,
      "grad_norm": 1837.347900390625,
      "learning_rate": 3.9972002981872124e-05,
      "loss": 126.8788,
      "step": 28790
    },
    {
      "epoch": 0.11635564425877819,
      "grad_norm": 1303.0589599609375,
      "learning_rate": 3.9971855080148906e-05,
      "loss": 112.7729,
      "step": 28800
    },
    {
      "epoch": 0.11639604552414581,
      "grad_norm": 2085.722900390625,
      "learning_rate": 3.997170678906409e-05,
      "loss": 134.1721,
      "step": 28810
    },
    {
      "epoch": 0.11643644678951345,
      "grad_norm": 512.4556274414062,
      "learning_rate": 3.997155810862057e-05,
      "loss": 76.3151,
      "step": 28820
    },
    {
      "epoch": 0.11647684805488108,
      "grad_norm": 727.9221801757812,
      "learning_rate": 3.997140903882124e-05,
      "loss": 76.4606,
      "step": 28830
    },
    {
      "epoch": 0.1165172493202487,
      "grad_norm": 1693.4381103515625,
      "learning_rate": 3.9971259579669006e-05,
      "loss": 103.796,
      "step": 28840
    },
    {
      "epoch": 0.11655765058561635,
      "grad_norm": 1555.8154296875,
      "learning_rate": 3.997110973116679e-05,
      "loss": 112.5202,
      "step": 28850
    },
    {
      "epoch": 0.11659805185098397,
      "grad_norm": 585.7073974609375,
      "learning_rate": 3.99709594933175e-05,
      "loss": 81.8929,
      "step": 28860
    },
    {
      "epoch": 0.1166384531163516,
      "grad_norm": 1181.9959716796875,
      "learning_rate": 3.997080886612408e-05,
      "loss": 112.615,
      "step": 28870
    },
    {
      "epoch": 0.11667885438171924,
      "grad_norm": 720.1932373046875,
      "learning_rate": 3.9970657849589465e-05,
      "loss": 72.8144,
      "step": 28880
    },
    {
      "epoch": 0.11671925564708686,
      "grad_norm": 1048.8651123046875,
      "learning_rate": 3.9970506443716586e-05,
      "loss": 112.9906,
      "step": 28890
    },
    {
      "epoch": 0.1167596569124545,
      "grad_norm": 1059.72998046875,
      "learning_rate": 3.99703546485084e-05,
      "loss": 149.952,
      "step": 28900
    },
    {
      "epoch": 0.11680005817782213,
      "grad_norm": 1071.949951171875,
      "learning_rate": 3.997020246396787e-05,
      "loss": 111.094,
      "step": 28910
    },
    {
      "epoch": 0.11684045944318976,
      "grad_norm": 1332.3616943359375,
      "learning_rate": 3.9970049890097965e-05,
      "loss": 88.1347,
      "step": 28920
    },
    {
      "epoch": 0.1168808607085574,
      "grad_norm": 437.4418029785156,
      "learning_rate": 3.996989692690165e-05,
      "loss": 60.3282,
      "step": 28930
    },
    {
      "epoch": 0.11692126197392502,
      "grad_norm": 419.84619140625,
      "learning_rate": 3.996974357438192e-05,
      "loss": 115.4472,
      "step": 28940
    },
    {
      "epoch": 0.11696166323929265,
      "grad_norm": 864.1251220703125,
      "learning_rate": 3.996958983254175e-05,
      "loss": 145.9911,
      "step": 28950
    },
    {
      "epoch": 0.11700206450466029,
      "grad_norm": 953.606689453125,
      "learning_rate": 3.996943570138416e-05,
      "loss": 117.4569,
      "step": 28960
    },
    {
      "epoch": 0.11704246577002791,
      "grad_norm": 845.4664916992188,
      "learning_rate": 3.996928118091213e-05,
      "loss": 69.0099,
      "step": 28970
    },
    {
      "epoch": 0.11708286703539555,
      "grad_norm": 1636.659423828125,
      "learning_rate": 3.996912627112868e-05,
      "loss": 120.3927,
      "step": 28980
    },
    {
      "epoch": 0.11712326830076318,
      "grad_norm": 1072.338623046875,
      "learning_rate": 3.996897097203684e-05,
      "loss": 119.216,
      "step": 28990
    },
    {
      "epoch": 0.11716366956613081,
      "grad_norm": 674.6630859375,
      "learning_rate": 3.9968815283639625e-05,
      "loss": 73.2614,
      "step": 29000
    },
    {
      "epoch": 0.11720407083149845,
      "grad_norm": 616.6441650390625,
      "learning_rate": 3.996865920594007e-05,
      "loss": 123.0924,
      "step": 29010
    },
    {
      "epoch": 0.11724447209686607,
      "grad_norm": 1423.77978515625,
      "learning_rate": 3.996850273894124e-05,
      "loss": 138.7357,
      "step": 29020
    },
    {
      "epoch": 0.1172848733622337,
      "grad_norm": 1283.6065673828125,
      "learning_rate": 3.996834588264615e-05,
      "loss": 80.1142,
      "step": 29030
    },
    {
      "epoch": 0.11732527462760134,
      "grad_norm": 1390.6195068359375,
      "learning_rate": 3.9968188637057886e-05,
      "loss": 112.3068,
      "step": 29040
    },
    {
      "epoch": 0.11736567589296897,
      "grad_norm": 6142.37744140625,
      "learning_rate": 3.99680310021795e-05,
      "loss": 119.8403,
      "step": 29050
    },
    {
      "epoch": 0.1174060771583366,
      "grad_norm": 497.8596496582031,
      "learning_rate": 3.9967872978014074e-05,
      "loss": 57.4572,
      "step": 29060
    },
    {
      "epoch": 0.11744647842370423,
      "grad_norm": 476.202880859375,
      "learning_rate": 3.996771456456468e-05,
      "loss": 99.7678,
      "step": 29070
    },
    {
      "epoch": 0.11748687968907186,
      "grad_norm": 957.2471313476562,
      "learning_rate": 3.996755576183442e-05,
      "loss": 120.2677,
      "step": 29080
    },
    {
      "epoch": 0.1175272809544395,
      "grad_norm": 606.5353393554688,
      "learning_rate": 3.9967396569826374e-05,
      "loss": 90.1209,
      "step": 29090
    },
    {
      "epoch": 0.11756768221980712,
      "grad_norm": 0.0,
      "learning_rate": 3.996723698854365e-05,
      "loss": 62.9719,
      "step": 29100
    },
    {
      "epoch": 0.11760808348517475,
      "grad_norm": 853.1898193359375,
      "learning_rate": 3.996707701798936e-05,
      "loss": 109.8049,
      "step": 29110
    },
    {
      "epoch": 0.11764848475054239,
      "grad_norm": 1235.3507080078125,
      "learning_rate": 3.9966916658166625e-05,
      "loss": 88.3993,
      "step": 29120
    },
    {
      "epoch": 0.11768888601591002,
      "grad_norm": 733.3147583007812,
      "learning_rate": 3.996675590907857e-05,
      "loss": 68.5436,
      "step": 29130
    },
    {
      "epoch": 0.11772928728127766,
      "grad_norm": 1256.930908203125,
      "learning_rate": 3.996659477072833e-05,
      "loss": 96.9386,
      "step": 29140
    },
    {
      "epoch": 0.11776968854664528,
      "grad_norm": 818.8697509765625,
      "learning_rate": 3.996643324311905e-05,
      "loss": 120.3725,
      "step": 29150
    },
    {
      "epoch": 0.11781008981201291,
      "grad_norm": 744.5905151367188,
      "learning_rate": 3.9966271326253874e-05,
      "loss": 79.7646,
      "step": 29160
    },
    {
      "epoch": 0.11785049107738055,
      "grad_norm": 924.5302734375,
      "learning_rate": 3.996610902013595e-05,
      "loss": 119.0584,
      "step": 29170
    },
    {
      "epoch": 0.11789089234274817,
      "grad_norm": 755.6444091796875,
      "learning_rate": 3.996594632476846e-05,
      "loss": 86.5377,
      "step": 29180
    },
    {
      "epoch": 0.1179312936081158,
      "grad_norm": 709.6162719726562,
      "learning_rate": 3.996578324015456e-05,
      "loss": 152.3329,
      "step": 29190
    },
    {
      "epoch": 0.11797169487348344,
      "grad_norm": 635.0053100585938,
      "learning_rate": 3.996561976629744e-05,
      "loss": 67.3558,
      "step": 29200
    },
    {
      "epoch": 0.11801209613885107,
      "grad_norm": 3188.703857421875,
      "learning_rate": 3.996545590320029e-05,
      "loss": 132.2558,
      "step": 29210
    },
    {
      "epoch": 0.1180524974042187,
      "grad_norm": 1246.9461669921875,
      "learning_rate": 3.996529165086629e-05,
      "loss": 92.1603,
      "step": 29220
    },
    {
      "epoch": 0.11809289866958633,
      "grad_norm": 525.8442993164062,
      "learning_rate": 3.9965127009298655e-05,
      "loss": 113.832,
      "step": 29230
    },
    {
      "epoch": 0.11813329993495396,
      "grad_norm": 1278.482421875,
      "learning_rate": 3.9964961978500586e-05,
      "loss": 119.3593,
      "step": 29240
    },
    {
      "epoch": 0.1181737012003216,
      "grad_norm": 586.2559204101562,
      "learning_rate": 3.9964796558475314e-05,
      "loss": 124.3714,
      "step": 29250
    },
    {
      "epoch": 0.11821410246568922,
      "grad_norm": 857.21533203125,
      "learning_rate": 3.996463074922604e-05,
      "loss": 74.2636,
      "step": 29260
    },
    {
      "epoch": 0.11825450373105685,
      "grad_norm": 1271.8468017578125,
      "learning_rate": 3.996446455075602e-05,
      "loss": 98.3933,
      "step": 29270
    },
    {
      "epoch": 0.11829490499642449,
      "grad_norm": 910.466064453125,
      "learning_rate": 3.996429796306848e-05,
      "loss": 98.6038,
      "step": 29280
    },
    {
      "epoch": 0.11833530626179212,
      "grad_norm": 947.049072265625,
      "learning_rate": 3.996413098616668e-05,
      "loss": 146.7748,
      "step": 29290
    },
    {
      "epoch": 0.11837570752715976,
      "grad_norm": 2612.84521484375,
      "learning_rate": 3.9963963620053865e-05,
      "loss": 119.463,
      "step": 29300
    },
    {
      "epoch": 0.11841610879252738,
      "grad_norm": 694.37255859375,
      "learning_rate": 3.99637958647333e-05,
      "loss": 125.3401,
      "step": 29310
    },
    {
      "epoch": 0.11845651005789501,
      "grad_norm": 731.656494140625,
      "learning_rate": 3.996362772020826e-05,
      "loss": 68.2026,
      "step": 29320
    },
    {
      "epoch": 0.11849691132326265,
      "grad_norm": 1356.3507080078125,
      "learning_rate": 3.9963459186482014e-05,
      "loss": 122.4604,
      "step": 29330
    },
    {
      "epoch": 0.11853731258863028,
      "grad_norm": 1672.551513671875,
      "learning_rate": 3.9963290263557856e-05,
      "loss": 134.1063,
      "step": 29340
    },
    {
      "epoch": 0.1185777138539979,
      "grad_norm": 2280.984619140625,
      "learning_rate": 3.996312095143908e-05,
      "loss": 124.3759,
      "step": 29350
    },
    {
      "epoch": 0.11861811511936554,
      "grad_norm": 687.0208129882812,
      "learning_rate": 3.996295125012898e-05,
      "loss": 105.2226,
      "step": 29360
    },
    {
      "epoch": 0.11865851638473317,
      "grad_norm": 830.426513671875,
      "learning_rate": 3.9962781159630865e-05,
      "loss": 87.1865,
      "step": 29370
    },
    {
      "epoch": 0.11869891765010081,
      "grad_norm": 1512.3504638671875,
      "learning_rate": 3.9962610679948065e-05,
      "loss": 105.9469,
      "step": 29380
    },
    {
      "epoch": 0.11873931891546843,
      "grad_norm": 955.3963012695312,
      "learning_rate": 3.9962439811083875e-05,
      "loss": 136.482,
      "step": 29390
    },
    {
      "epoch": 0.11877972018083606,
      "grad_norm": 398.2100830078125,
      "learning_rate": 3.9962268553041656e-05,
      "loss": 96.7592,
      "step": 29400
    },
    {
      "epoch": 0.1188201214462037,
      "grad_norm": 1317.3018798828125,
      "learning_rate": 3.996209690582473e-05,
      "loss": 90.4329,
      "step": 29410
    },
    {
      "epoch": 0.11886052271157133,
      "grad_norm": 1313.802490234375,
      "learning_rate": 3.996192486943645e-05,
      "loss": 119.6198,
      "step": 29420
    },
    {
      "epoch": 0.11890092397693895,
      "grad_norm": 1340.90869140625,
      "learning_rate": 3.996175244388017e-05,
      "loss": 120.3501,
      "step": 29430
    },
    {
      "epoch": 0.11894132524230659,
      "grad_norm": 683.0484008789062,
      "learning_rate": 3.9961579629159244e-05,
      "loss": 107.2445,
      "step": 29440
    },
    {
      "epoch": 0.11898172650767422,
      "grad_norm": 804.3728637695312,
      "learning_rate": 3.9961406425277045e-05,
      "loss": 97.0491,
      "step": 29450
    },
    {
      "epoch": 0.11902212777304186,
      "grad_norm": 591.4122314453125,
      "learning_rate": 3.9961232832236956e-05,
      "loss": 99.1442,
      "step": 29460
    },
    {
      "epoch": 0.11906252903840948,
      "grad_norm": 1971.5877685546875,
      "learning_rate": 3.9961058850042345e-05,
      "loss": 99.8225,
      "step": 29470
    },
    {
      "epoch": 0.11910293030377711,
      "grad_norm": 1043.689208984375,
      "learning_rate": 3.9960884478696627e-05,
      "loss": 90.9582,
      "step": 29480
    },
    {
      "epoch": 0.11914333156914475,
      "grad_norm": 2496.54541015625,
      "learning_rate": 3.996070971820319e-05,
      "loss": 131.3534,
      "step": 29490
    },
    {
      "epoch": 0.11918373283451238,
      "grad_norm": 736.2053833007812,
      "learning_rate": 3.9960534568565436e-05,
      "loss": 120.754,
      "step": 29500
    },
    {
      "epoch": 0.11922413409988,
      "grad_norm": 1127.8834228515625,
      "learning_rate": 3.996035902978679e-05,
      "loss": 117.7929,
      "step": 29510
    },
    {
      "epoch": 0.11926453536524764,
      "grad_norm": 747.2788696289062,
      "learning_rate": 3.996018310187066e-05,
      "loss": 84.6329,
      "step": 29520
    },
    {
      "epoch": 0.11930493663061527,
      "grad_norm": 1040.839599609375,
      "learning_rate": 3.9960006784820485e-05,
      "loss": 147.8331,
      "step": 29530
    },
    {
      "epoch": 0.11934533789598291,
      "grad_norm": 2474.502197265625,
      "learning_rate": 3.99598300786397e-05,
      "loss": 118.2935,
      "step": 29540
    },
    {
      "epoch": 0.11938573916135053,
      "grad_norm": 1262.937255859375,
      "learning_rate": 3.995965298333176e-05,
      "loss": 85.3698,
      "step": 29550
    },
    {
      "epoch": 0.11942614042671816,
      "grad_norm": 833.2108154296875,
      "learning_rate": 3.99594754989001e-05,
      "loss": 100.4102,
      "step": 29560
    },
    {
      "epoch": 0.1194665416920858,
      "grad_norm": 1125.703125,
      "learning_rate": 3.9959297625348196e-05,
      "loss": 115.4946,
      "step": 29570
    },
    {
      "epoch": 0.11950694295745343,
      "grad_norm": 741.2850341796875,
      "learning_rate": 3.995911936267951e-05,
      "loss": 90.9823,
      "step": 29580
    },
    {
      "epoch": 0.11954734422282105,
      "grad_norm": 465.7984313964844,
      "learning_rate": 3.995894071089751e-05,
      "loss": 95.7353,
      "step": 29590
    },
    {
      "epoch": 0.1195877454881887,
      "grad_norm": 1323.34130859375,
      "learning_rate": 3.995876167000569e-05,
      "loss": 72.345,
      "step": 29600
    },
    {
      "epoch": 0.11962814675355632,
      "grad_norm": 2258.849609375,
      "learning_rate": 3.9958582240007536e-05,
      "loss": 131.1186,
      "step": 29610
    },
    {
      "epoch": 0.11966854801892396,
      "grad_norm": 854.4385375976562,
      "learning_rate": 3.995840242090655e-05,
      "loss": 95.1046,
      "step": 29620
    },
    {
      "epoch": 0.11970894928429159,
      "grad_norm": 394.0882873535156,
      "learning_rate": 3.995822221270622e-05,
      "loss": 112.4525,
      "step": 29630
    },
    {
      "epoch": 0.11974935054965921,
      "grad_norm": 636.0965576171875,
      "learning_rate": 3.9958041615410085e-05,
      "loss": 63.1489,
      "step": 29640
    },
    {
      "epoch": 0.11978975181502685,
      "grad_norm": 1257.3487548828125,
      "learning_rate": 3.995786062902165e-05,
      "loss": 154.1359,
      "step": 29650
    },
    {
      "epoch": 0.11983015308039448,
      "grad_norm": 997.6132202148438,
      "learning_rate": 3.995767925354445e-05,
      "loss": 77.0885,
      "step": 29660
    },
    {
      "epoch": 0.1198705543457621,
      "grad_norm": 939.9737548828125,
      "learning_rate": 3.9957497488982014e-05,
      "loss": 101.9474,
      "step": 29670
    },
    {
      "epoch": 0.11991095561112974,
      "grad_norm": 1560.003173828125,
      "learning_rate": 3.995731533533789e-05,
      "loss": 128.4176,
      "step": 29680
    },
    {
      "epoch": 0.11995135687649737,
      "grad_norm": 1200.610107421875,
      "learning_rate": 3.995713279261563e-05,
      "loss": 105.8525,
      "step": 29690
    },
    {
      "epoch": 0.11999175814186501,
      "grad_norm": 579.2079467773438,
      "learning_rate": 3.995694986081879e-05,
      "loss": 96.4495,
      "step": 29700
    },
    {
      "epoch": 0.12003215940723264,
      "grad_norm": 8644.3876953125,
      "learning_rate": 3.9956766539950934e-05,
      "loss": 148.03,
      "step": 29710
    },
    {
      "epoch": 0.12007256067260026,
      "grad_norm": 4799.1953125,
      "learning_rate": 3.995658283001564e-05,
      "loss": 122.2243,
      "step": 29720
    },
    {
      "epoch": 0.1201129619379679,
      "grad_norm": 953.9501342773438,
      "learning_rate": 3.99563987310165e-05,
      "loss": 112.8426,
      "step": 29730
    },
    {
      "epoch": 0.12015336320333553,
      "grad_norm": 590.100830078125,
      "learning_rate": 3.995621424295709e-05,
      "loss": 102.4715,
      "step": 29740
    },
    {
      "epoch": 0.12019376446870315,
      "grad_norm": 3277.8310546875,
      "learning_rate": 3.9956029365841005e-05,
      "loss": 101.0477,
      "step": 29750
    },
    {
      "epoch": 0.1202341657340708,
      "grad_norm": 404.7799072265625,
      "learning_rate": 3.995584409967185e-05,
      "loss": 113.8865,
      "step": 29760
    },
    {
      "epoch": 0.12027456699943842,
      "grad_norm": 1674.560791015625,
      "learning_rate": 3.9955658444453244e-05,
      "loss": 111.887,
      "step": 29770
    },
    {
      "epoch": 0.12031496826480606,
      "grad_norm": 1182.077880859375,
      "learning_rate": 3.995547240018881e-05,
      "loss": 109.1132,
      "step": 29780
    },
    {
      "epoch": 0.12035536953017369,
      "grad_norm": 401.8653869628906,
      "learning_rate": 3.995528596688216e-05,
      "loss": 82.7655,
      "step": 29790
    },
    {
      "epoch": 0.12039577079554131,
      "grad_norm": 2734.860107421875,
      "learning_rate": 3.995509914453694e-05,
      "loss": 104.0152,
      "step": 29800
    },
    {
      "epoch": 0.12043617206090895,
      "grad_norm": 696.2372436523438,
      "learning_rate": 3.995491193315678e-05,
      "loss": 67.8705,
      "step": 29810
    },
    {
      "epoch": 0.12047657332627658,
      "grad_norm": 1432.671142578125,
      "learning_rate": 3.9954724332745344e-05,
      "loss": 119.9143,
      "step": 29820
    },
    {
      "epoch": 0.1205169745916442,
      "grad_norm": 696.959228515625,
      "learning_rate": 3.9954536343306286e-05,
      "loss": 93.7465,
      "step": 29830
    },
    {
      "epoch": 0.12055737585701184,
      "grad_norm": 1257.0362548828125,
      "learning_rate": 3.995434796484326e-05,
      "loss": 89.61,
      "step": 29840
    },
    {
      "epoch": 0.12059777712237947,
      "grad_norm": 1463.5260009765625,
      "learning_rate": 3.9954159197359956e-05,
      "loss": 84.8807,
      "step": 29850
    },
    {
      "epoch": 0.12063817838774711,
      "grad_norm": 579.013916015625,
      "learning_rate": 3.9953970040860045e-05,
      "loss": 74.2993,
      "step": 29860
    },
    {
      "epoch": 0.12067857965311474,
      "grad_norm": 880.2040405273438,
      "learning_rate": 3.9953780495347214e-05,
      "loss": 80.5523,
      "step": 29870
    },
    {
      "epoch": 0.12071898091848236,
      "grad_norm": 1162.419677734375,
      "learning_rate": 3.995359056082516e-05,
      "loss": 86.788,
      "step": 29880
    },
    {
      "epoch": 0.12075938218385,
      "grad_norm": 1339.3079833984375,
      "learning_rate": 3.9953400237297584e-05,
      "loss": 96.6253,
      "step": 29890
    },
    {
      "epoch": 0.12079978344921763,
      "grad_norm": 717.9099731445312,
      "learning_rate": 3.99532095247682e-05,
      "loss": 83.6678,
      "step": 29900
    },
    {
      "epoch": 0.12084018471458526,
      "grad_norm": 935.6298217773438,
      "learning_rate": 3.995301842324072e-05,
      "loss": 134.8532,
      "step": 29910
    },
    {
      "epoch": 0.1208805859799529,
      "grad_norm": 474.3695068359375,
      "learning_rate": 3.9952826932718874e-05,
      "loss": 114.1753,
      "step": 29920
    },
    {
      "epoch": 0.12092098724532052,
      "grad_norm": 1465.263671875,
      "learning_rate": 3.995263505320639e-05,
      "loss": 191.371,
      "step": 29930
    },
    {
      "epoch": 0.12096138851068816,
      "grad_norm": 653.8214111328125,
      "learning_rate": 3.9952442784707025e-05,
      "loss": 103.5274,
      "step": 29940
    },
    {
      "epoch": 0.12100178977605579,
      "grad_norm": 1105.3265380859375,
      "learning_rate": 3.995225012722451e-05,
      "loss": 128.0836,
      "step": 29950
    },
    {
      "epoch": 0.12104219104142341,
      "grad_norm": 1911.07861328125,
      "learning_rate": 3.995205708076261e-05,
      "loss": 129.5159,
      "step": 29960
    },
    {
      "epoch": 0.12108259230679105,
      "grad_norm": 665.4229125976562,
      "learning_rate": 3.995186364532507e-05,
      "loss": 108.7585,
      "step": 29970
    },
    {
      "epoch": 0.12112299357215868,
      "grad_norm": 1236.4959716796875,
      "learning_rate": 3.99516698209157e-05,
      "loss": 109.1846,
      "step": 29980
    },
    {
      "epoch": 0.1211633948375263,
      "grad_norm": 2250.654541015625,
      "learning_rate": 3.9951475607538234e-05,
      "loss": 77.2249,
      "step": 29990
    },
    {
      "epoch": 0.12120379610289395,
      "grad_norm": 868.831298828125,
      "learning_rate": 3.9951281005196486e-05,
      "loss": 144.552,
      "step": 30000
    },
    {
      "epoch": 0.12124419736826157,
      "grad_norm": 448.3783874511719,
      "learning_rate": 3.995108601389425e-05,
      "loss": 90.4089,
      "step": 30010
    },
    {
      "epoch": 0.12128459863362921,
      "grad_norm": 945.7691040039062,
      "learning_rate": 3.995089063363531e-05,
      "loss": 109.7032,
      "step": 30020
    },
    {
      "epoch": 0.12132499989899684,
      "grad_norm": 915.0947265625,
      "learning_rate": 3.9950694864423496e-05,
      "loss": 152.643,
      "step": 30030
    },
    {
      "epoch": 0.12136540116436446,
      "grad_norm": 700.6856079101562,
      "learning_rate": 3.995049870626261e-05,
      "loss": 87.6623,
      "step": 30040
    },
    {
      "epoch": 0.1214058024297321,
      "grad_norm": 798.1749267578125,
      "learning_rate": 3.9950302159156476e-05,
      "loss": 123.6986,
      "step": 30050
    },
    {
      "epoch": 0.12144620369509973,
      "grad_norm": 1099.6778564453125,
      "learning_rate": 3.9950105223108935e-05,
      "loss": 125.2038,
      "step": 30060
    },
    {
      "epoch": 0.12148660496046736,
      "grad_norm": 781.4649658203125,
      "learning_rate": 3.994990789812383e-05,
      "loss": 121.0432,
      "step": 30070
    },
    {
      "epoch": 0.121527006225835,
      "grad_norm": 260.9813537597656,
      "learning_rate": 3.994971018420498e-05,
      "loss": 107.953,
      "step": 30080
    },
    {
      "epoch": 0.12156740749120262,
      "grad_norm": 2069.24169921875,
      "learning_rate": 3.9949512081356275e-05,
      "loss": 117.141,
      "step": 30090
    },
    {
      "epoch": 0.12160780875657026,
      "grad_norm": 1008.3120727539062,
      "learning_rate": 3.9949313589581555e-05,
      "loss": 78.568,
      "step": 30100
    },
    {
      "epoch": 0.12164821002193789,
      "grad_norm": 1447.808349609375,
      "learning_rate": 3.99491147088847e-05,
      "loss": 93.7204,
      "step": 30110
    },
    {
      "epoch": 0.12168861128730551,
      "grad_norm": 699.3298950195312,
      "learning_rate": 3.994891543926958e-05,
      "loss": 114.6168,
      "step": 30120
    },
    {
      "epoch": 0.12172901255267315,
      "grad_norm": 2293.560302734375,
      "learning_rate": 3.9948715780740077e-05,
      "loss": 80.4595,
      "step": 30130
    },
    {
      "epoch": 0.12176941381804078,
      "grad_norm": 879.6803588867188,
      "learning_rate": 3.99485157333001e-05,
      "loss": 83.2679,
      "step": 30140
    },
    {
      "epoch": 0.12180981508340841,
      "grad_norm": 1878.105224609375,
      "learning_rate": 3.9948315296953535e-05,
      "loss": 104.4954,
      "step": 30150
    },
    {
      "epoch": 0.12185021634877605,
      "grad_norm": 940.0166625976562,
      "learning_rate": 3.9948114471704296e-05,
      "loss": 128.4855,
      "step": 30160
    },
    {
      "epoch": 0.12189061761414367,
      "grad_norm": 545.82861328125,
      "learning_rate": 3.9947913257556285e-05,
      "loss": 112.7048,
      "step": 30170
    },
    {
      "epoch": 0.12193101887951131,
      "grad_norm": 709.3935546875,
      "learning_rate": 3.9947711654513445e-05,
      "loss": 92.9257,
      "step": 30180
    },
    {
      "epoch": 0.12197142014487894,
      "grad_norm": 2208.292236328125,
      "learning_rate": 3.99475096625797e-05,
      "loss": 110.0978,
      "step": 30190
    },
    {
      "epoch": 0.12201182141024657,
      "grad_norm": 939.6221923828125,
      "learning_rate": 3.994730728175897e-05,
      "loss": 81.1035,
      "step": 30200
    },
    {
      "epoch": 0.1220522226756142,
      "grad_norm": 754.7445678710938,
      "learning_rate": 3.994710451205523e-05,
      "loss": 76.923,
      "step": 30210
    },
    {
      "epoch": 0.12209262394098183,
      "grad_norm": 710.9456787109375,
      "learning_rate": 3.994690135347241e-05,
      "loss": 81.77,
      "step": 30220
    },
    {
      "epoch": 0.12213302520634946,
      "grad_norm": 7868.96240234375,
      "learning_rate": 3.9946697806014476e-05,
      "loss": 131.3129,
      "step": 30230
    },
    {
      "epoch": 0.1221734264717171,
      "grad_norm": 914.0556030273438,
      "learning_rate": 3.99464938696854e-05,
      "loss": 76.2861,
      "step": 30240
    },
    {
      "epoch": 0.12221382773708472,
      "grad_norm": 835.4411010742188,
      "learning_rate": 3.994628954448916e-05,
      "loss": 73.5062,
      "step": 30250
    },
    {
      "epoch": 0.12225422900245236,
      "grad_norm": 1167.3447265625,
      "learning_rate": 3.994608483042974e-05,
      "loss": 102.8286,
      "step": 30260
    },
    {
      "epoch": 0.12229463026781999,
      "grad_norm": 631.1087646484375,
      "learning_rate": 3.9945879727511126e-05,
      "loss": 58.7113,
      "step": 30270
    },
    {
      "epoch": 0.12233503153318762,
      "grad_norm": 445.20159912109375,
      "learning_rate": 3.9945674235737314e-05,
      "loss": 70.526,
      "step": 30280
    },
    {
      "epoch": 0.12237543279855526,
      "grad_norm": 284.5147705078125,
      "learning_rate": 3.994546835511232e-05,
      "loss": 73.7212,
      "step": 30290
    },
    {
      "epoch": 0.12241583406392288,
      "grad_norm": 1401.1563720703125,
      "learning_rate": 3.994526208564014e-05,
      "loss": 93.3938,
      "step": 30300
    },
    {
      "epoch": 0.12245623532929051,
      "grad_norm": 2154.647705078125,
      "learning_rate": 3.994505542732482e-05,
      "loss": 158.8047,
      "step": 30310
    },
    {
      "epoch": 0.12249663659465815,
      "grad_norm": 5086.50927734375,
      "learning_rate": 3.994484838017037e-05,
      "loss": 93.4497,
      "step": 30320
    },
    {
      "epoch": 0.12253703786002577,
      "grad_norm": 782.09521484375,
      "learning_rate": 3.994464094418083e-05,
      "loss": 104.2793,
      "step": 30330
    },
    {
      "epoch": 0.12257743912539341,
      "grad_norm": 2351.08056640625,
      "learning_rate": 3.994443311936025e-05,
      "loss": 120.7483,
      "step": 30340
    },
    {
      "epoch": 0.12261784039076104,
      "grad_norm": 1276.39111328125,
      "learning_rate": 3.994422490571268e-05,
      "loss": 80.3845,
      "step": 30350
    },
    {
      "epoch": 0.12265824165612867,
      "grad_norm": 940.2988891601562,
      "learning_rate": 3.994401630324217e-05,
      "loss": 133.5265,
      "step": 30360
    },
    {
      "epoch": 0.1226986429214963,
      "grad_norm": 690.517578125,
      "learning_rate": 3.99438073119528e-05,
      "loss": 130.6995,
      "step": 30370
    },
    {
      "epoch": 0.12273904418686393,
      "grad_norm": 490.5731201171875,
      "learning_rate": 3.994359793184864e-05,
      "loss": 91.125,
      "step": 30380
    },
    {
      "epoch": 0.12277944545223156,
      "grad_norm": 439.6798400878906,
      "learning_rate": 3.994338816293377e-05,
      "loss": 83.8165,
      "step": 30390
    },
    {
      "epoch": 0.1228198467175992,
      "grad_norm": 751.61767578125,
      "learning_rate": 3.994317800521228e-05,
      "loss": 100.241,
      "step": 30400
    },
    {
      "epoch": 0.12286024798296682,
      "grad_norm": 429.4967041015625,
      "learning_rate": 3.9942967458688267e-05,
      "loss": 90.6561,
      "step": 30410
    },
    {
      "epoch": 0.12290064924833445,
      "grad_norm": 701.9464721679688,
      "learning_rate": 3.9942756523365835e-05,
      "loss": 57.0881,
      "step": 30420
    },
    {
      "epoch": 0.12294105051370209,
      "grad_norm": 1101.32958984375,
      "learning_rate": 3.99425451992491e-05,
      "loss": 105.1508,
      "step": 30430
    },
    {
      "epoch": 0.12298145177906972,
      "grad_norm": 976.6168212890625,
      "learning_rate": 3.994233348634218e-05,
      "loss": 102.866,
      "step": 30440
    },
    {
      "epoch": 0.12302185304443736,
      "grad_norm": 747.2569580078125,
      "learning_rate": 3.99421213846492e-05,
      "loss": 59.548,
      "step": 30450
    },
    {
      "epoch": 0.12306225430980498,
      "grad_norm": 1975.991455078125,
      "learning_rate": 3.9941908894174295e-05,
      "loss": 91.4762,
      "step": 30460
    },
    {
      "epoch": 0.12310265557517261,
      "grad_norm": 1650.881103515625,
      "learning_rate": 3.994169601492161e-05,
      "loss": 111.3243,
      "step": 30470
    },
    {
      "epoch": 0.12314305684054025,
      "grad_norm": 1898.3350830078125,
      "learning_rate": 3.994148274689529e-05,
      "loss": 121.5767,
      "step": 30480
    },
    {
      "epoch": 0.12318345810590788,
      "grad_norm": 0.0,
      "learning_rate": 3.99412690900995e-05,
      "loss": 133.1026,
      "step": 30490
    },
    {
      "epoch": 0.1232238593712755,
      "grad_norm": 803.773193359375,
      "learning_rate": 3.994105504453841e-05,
      "loss": 108.3969,
      "step": 30500
    },
    {
      "epoch": 0.12326426063664314,
      "grad_norm": 1622.7261962890625,
      "learning_rate": 3.994084061021618e-05,
      "loss": 109.2424,
      "step": 30510
    },
    {
      "epoch": 0.12330466190201077,
      "grad_norm": 1043.625732421875,
      "learning_rate": 3.9940625787136995e-05,
      "loss": 99.5337,
      "step": 30520
    },
    {
      "epoch": 0.12334506316737841,
      "grad_norm": 1230.0400390625,
      "learning_rate": 3.9940410575305044e-05,
      "loss": 80.0279,
      "step": 30530
    },
    {
      "epoch": 0.12338546443274603,
      "grad_norm": 954.072021484375,
      "learning_rate": 3.9940194974724524e-05,
      "loss": 104.5978,
      "step": 30540
    },
    {
      "epoch": 0.12342586569811366,
      "grad_norm": 660.2078247070312,
      "learning_rate": 3.993997898539964e-05,
      "loss": 113.9359,
      "step": 30550
    },
    {
      "epoch": 0.1234662669634813,
      "grad_norm": 1597.9298095703125,
      "learning_rate": 3.993976260733459e-05,
      "loss": 76.1934,
      "step": 30560
    },
    {
      "epoch": 0.12350666822884893,
      "grad_norm": 1095.2469482421875,
      "learning_rate": 3.993954584053361e-05,
      "loss": 92.5734,
      "step": 30570
    },
    {
      "epoch": 0.12354706949421655,
      "grad_norm": 2675.254638671875,
      "learning_rate": 3.993932868500092e-05,
      "loss": 98.4438,
      "step": 30580
    },
    {
      "epoch": 0.12358747075958419,
      "grad_norm": 1733.7291259765625,
      "learning_rate": 3.9939111140740754e-05,
      "loss": 96.593,
      "step": 30590
    },
    {
      "epoch": 0.12362787202495182,
      "grad_norm": 516.620361328125,
      "learning_rate": 3.993889320775735e-05,
      "loss": 85.3212,
      "step": 30600
    },
    {
      "epoch": 0.12366827329031946,
      "grad_norm": 1090.36083984375,
      "learning_rate": 3.993867488605495e-05,
      "loss": 92.5375,
      "step": 30610
    },
    {
      "epoch": 0.12370867455568708,
      "grad_norm": 1519.6614990234375,
      "learning_rate": 3.993845617563782e-05,
      "loss": 158.0534,
      "step": 30620
    },
    {
      "epoch": 0.12374907582105471,
      "grad_norm": 1311.4442138671875,
      "learning_rate": 3.993823707651023e-05,
      "loss": 113.0447,
      "step": 30630
    },
    {
      "epoch": 0.12378947708642235,
      "grad_norm": 1757.02685546875,
      "learning_rate": 3.993801758867643e-05,
      "loss": 156.7644,
      "step": 30640
    },
    {
      "epoch": 0.12382987835178998,
      "grad_norm": 1286.7274169921875,
      "learning_rate": 3.993779771214072e-05,
      "loss": 102.2056,
      "step": 30650
    },
    {
      "epoch": 0.1238702796171576,
      "grad_norm": 530.4484252929688,
      "learning_rate": 3.9937577446907376e-05,
      "loss": 81.6175,
      "step": 30660
    },
    {
      "epoch": 0.12391068088252524,
      "grad_norm": 624.5010375976562,
      "learning_rate": 3.99373567929807e-05,
      "loss": 70.0852,
      "step": 30670
    },
    {
      "epoch": 0.12395108214789287,
      "grad_norm": 512.4428100585938,
      "learning_rate": 3.9937135750364983e-05,
      "loss": 61.0621,
      "step": 30680
    },
    {
      "epoch": 0.12399148341326051,
      "grad_norm": 662.2406616210938,
      "learning_rate": 3.9936914319064543e-05,
      "loss": 134.4542,
      "step": 30690
    },
    {
      "epoch": 0.12403188467862813,
      "grad_norm": 2490.583251953125,
      "learning_rate": 3.9936692499083696e-05,
      "loss": 89.6221,
      "step": 30700
    },
    {
      "epoch": 0.12407228594399576,
      "grad_norm": 1335.3441162109375,
      "learning_rate": 3.9936470290426765e-05,
      "loss": 87.1301,
      "step": 30710
    },
    {
      "epoch": 0.1241126872093634,
      "grad_norm": 2422.685546875,
      "learning_rate": 3.993624769309808e-05,
      "loss": 79.4778,
      "step": 30720
    },
    {
      "epoch": 0.12415308847473103,
      "grad_norm": 1162.318603515625,
      "learning_rate": 3.9936024707101984e-05,
      "loss": 47.9735,
      "step": 30730
    },
    {
      "epoch": 0.12419348974009865,
      "grad_norm": 2508.990234375,
      "learning_rate": 3.993580133244282e-05,
      "loss": 121.7656,
      "step": 30740
    },
    {
      "epoch": 0.1242338910054663,
      "grad_norm": 1584.348876953125,
      "learning_rate": 3.993557756912495e-05,
      "loss": 67.5902,
      "step": 30750
    },
    {
      "epoch": 0.12427429227083392,
      "grad_norm": 726.6798095703125,
      "learning_rate": 3.9935353417152724e-05,
      "loss": 75.2805,
      "step": 30760
    },
    {
      "epoch": 0.12431469353620156,
      "grad_norm": 784.2379150390625,
      "learning_rate": 3.9935128876530524e-05,
      "loss": 89.1682,
      "step": 30770
    },
    {
      "epoch": 0.12435509480156919,
      "grad_norm": 1024.3555908203125,
      "learning_rate": 3.9934903947262726e-05,
      "loss": 89.0586,
      "step": 30780
    },
    {
      "epoch": 0.12439549606693681,
      "grad_norm": 1700.198486328125,
      "learning_rate": 3.99346786293537e-05,
      "loss": 109.834,
      "step": 30790
    },
    {
      "epoch": 0.12443589733230445,
      "grad_norm": 3284.315185546875,
      "learning_rate": 3.993445292280787e-05,
      "loss": 153.4986,
      "step": 30800
    },
    {
      "epoch": 0.12447629859767208,
      "grad_norm": 1563.0628662109375,
      "learning_rate": 3.9934226827629603e-05,
      "loss": 82.126,
      "step": 30810
    },
    {
      "epoch": 0.1245166998630397,
      "grad_norm": 739.8726196289062,
      "learning_rate": 3.993400034382333e-05,
      "loss": 82.8279,
      "step": 30820
    },
    {
      "epoch": 0.12455710112840734,
      "grad_norm": 1393.825439453125,
      "learning_rate": 3.993377347139345e-05,
      "loss": 108.3987,
      "step": 30830
    },
    {
      "epoch": 0.12459750239377497,
      "grad_norm": 836.917724609375,
      "learning_rate": 3.99335462103444e-05,
      "loss": 108.2581,
      "step": 30840
    },
    {
      "epoch": 0.12463790365914261,
      "grad_norm": 1300.4273681640625,
      "learning_rate": 3.99333185606806e-05,
      "loss": 65.6786,
      "step": 30850
    },
    {
      "epoch": 0.12467830492451024,
      "grad_norm": 276.4493713378906,
      "learning_rate": 3.99330905224065e-05,
      "loss": 133.912,
      "step": 30860
    },
    {
      "epoch": 0.12471870618987786,
      "grad_norm": 712.7859497070312,
      "learning_rate": 3.993286209552652e-05,
      "loss": 115.7652,
      "step": 30870
    },
    {
      "epoch": 0.1247591074552455,
      "grad_norm": 1046.329345703125,
      "learning_rate": 3.9932633280045154e-05,
      "loss": 83.9093,
      "step": 30880
    },
    {
      "epoch": 0.12479950872061313,
      "grad_norm": 708.8577880859375,
      "learning_rate": 3.993240407596682e-05,
      "loss": 137.295,
      "step": 30890
    },
    {
      "epoch": 0.12483990998598075,
      "grad_norm": 1173.4384765625,
      "learning_rate": 3.993217448329602e-05,
      "loss": 65.7867,
      "step": 30900
    },
    {
      "epoch": 0.1248803112513484,
      "grad_norm": 589.5326538085938,
      "learning_rate": 3.993194450203721e-05,
      "loss": 93.5474,
      "step": 30910
    },
    {
      "epoch": 0.12492071251671602,
      "grad_norm": 809.4989013671875,
      "learning_rate": 3.993171413219489e-05,
      "loss": 76.3629,
      "step": 30920
    },
    {
      "epoch": 0.12496111378208366,
      "grad_norm": 1225.9090576171875,
      "learning_rate": 3.9931483373773524e-05,
      "loss": 80.0511,
      "step": 30930
    },
    {
      "epoch": 0.12500151504745127,
      "grad_norm": 442.1229248046875,
      "learning_rate": 3.9931252226777646e-05,
      "loss": 133.8095,
      "step": 30940
    },
    {
      "epoch": 0.1250419163128189,
      "grad_norm": 1054.5335693359375,
      "learning_rate": 3.993102069121173e-05,
      "loss": 81.4998,
      "step": 30950
    },
    {
      "epoch": 0.12508231757818655,
      "grad_norm": 1194.748046875,
      "learning_rate": 3.9930788767080316e-05,
      "loss": 175.7166,
      "step": 30960
    },
    {
      "epoch": 0.1251227188435542,
      "grad_norm": 434.2280578613281,
      "learning_rate": 3.993055645438791e-05,
      "loss": 99.1189,
      "step": 30970
    },
    {
      "epoch": 0.1251631201089218,
      "grad_norm": 734.3633422851562,
      "learning_rate": 3.993032375313904e-05,
      "loss": 80.5113,
      "step": 30980
    },
    {
      "epoch": 0.12520352137428944,
      "grad_norm": 1156.7469482421875,
      "learning_rate": 3.993009066333826e-05,
      "loss": 90.6556,
      "step": 30990
    },
    {
      "epoch": 0.12524392263965708,
      "grad_norm": 979.235107421875,
      "learning_rate": 3.992985718499009e-05,
      "loss": 108.3393,
      "step": 31000
    },
    {
      "epoch": 0.1252843239050247,
      "grad_norm": 570.459716796875,
      "learning_rate": 3.99296233180991e-05,
      "loss": 71.2451,
      "step": 31010
    },
    {
      "epoch": 0.12532472517039234,
      "grad_norm": 1419.055908203125,
      "learning_rate": 3.9929389062669834e-05,
      "loss": 105.6378,
      "step": 31020
    },
    {
      "epoch": 0.12536512643575998,
      "grad_norm": 762.352294921875,
      "learning_rate": 3.9929154418706876e-05,
      "loss": 114.2663,
      "step": 31030
    },
    {
      "epoch": 0.1254055277011276,
      "grad_norm": 1120.8670654296875,
      "learning_rate": 3.992891938621479e-05,
      "loss": 135.846,
      "step": 31040
    },
    {
      "epoch": 0.12544592896649523,
      "grad_norm": 1000.6475219726562,
      "learning_rate": 3.992868396519816e-05,
      "loss": 78.1399,
      "step": 31050
    },
    {
      "epoch": 0.12548633023186287,
      "grad_norm": 1076.135986328125,
      "learning_rate": 3.9928448155661576e-05,
      "loss": 101.4686,
      "step": 31060
    },
    {
      "epoch": 0.12552673149723048,
      "grad_norm": 1245.489501953125,
      "learning_rate": 3.992821195760964e-05,
      "loss": 98.2108,
      "step": 31070
    },
    {
      "epoch": 0.12556713276259812,
      "grad_norm": 1279.958984375,
      "learning_rate": 3.992797537104695e-05,
      "loss": 89.94,
      "step": 31080
    },
    {
      "epoch": 0.12560753402796576,
      "grad_norm": 1093.621337890625,
      "learning_rate": 3.9927738395978116e-05,
      "loss": 122.3186,
      "step": 31090
    },
    {
      "epoch": 0.12564793529333337,
      "grad_norm": 567.3983154296875,
      "learning_rate": 3.9927501032407775e-05,
      "loss": 98.0507,
      "step": 31100
    },
    {
      "epoch": 0.125688336558701,
      "grad_norm": 961.0360107421875,
      "learning_rate": 3.992726328034053e-05,
      "loss": 86.6894,
      "step": 31110
    },
    {
      "epoch": 0.12572873782406865,
      "grad_norm": 1218.6414794921875,
      "learning_rate": 3.9927025139781023e-05,
      "loss": 82.9272,
      "step": 31120
    },
    {
      "epoch": 0.1257691390894363,
      "grad_norm": 531.4141845703125,
      "learning_rate": 3.9926786610733917e-05,
      "loss": 95.1389,
      "step": 31130
    },
    {
      "epoch": 0.1258095403548039,
      "grad_norm": 1093.076904296875,
      "learning_rate": 3.9926547693203836e-05,
      "loss": 96.6621,
      "step": 31140
    },
    {
      "epoch": 0.12584994162017155,
      "grad_norm": 548.7486572265625,
      "learning_rate": 3.9926308387195456e-05,
      "loss": 88.0375,
      "step": 31150
    },
    {
      "epoch": 0.12589034288553919,
      "grad_norm": 1444.5079345703125,
      "learning_rate": 3.992606869271343e-05,
      "loss": 107.03,
      "step": 31160
    },
    {
      "epoch": 0.1259307441509068,
      "grad_norm": 1084.5355224609375,
      "learning_rate": 3.992582860976244e-05,
      "loss": 104.2385,
      "step": 31170
    },
    {
      "epoch": 0.12597114541627444,
      "grad_norm": 587.4788208007812,
      "learning_rate": 3.9925588138347155e-05,
      "loss": 58.3109,
      "step": 31180
    },
    {
      "epoch": 0.12601154668164208,
      "grad_norm": 1209.0523681640625,
      "learning_rate": 3.992534727847228e-05,
      "loss": 58.8434,
      "step": 31190
    },
    {
      "epoch": 0.1260519479470097,
      "grad_norm": 1135.3553466796875,
      "learning_rate": 3.99251060301425e-05,
      "loss": 114.3812,
      "step": 31200
    },
    {
      "epoch": 0.12609234921237733,
      "grad_norm": 2140.721435546875,
      "learning_rate": 3.992486439336252e-05,
      "loss": 82.2324,
      "step": 31210
    },
    {
      "epoch": 0.12613275047774497,
      "grad_norm": 2814.8818359375,
      "learning_rate": 3.992462236813704e-05,
      "loss": 141.9477,
      "step": 31220
    },
    {
      "epoch": 0.12617315174311258,
      "grad_norm": 742.0789794921875,
      "learning_rate": 3.99243799544708e-05,
      "loss": 117.9271,
      "step": 31230
    },
    {
      "epoch": 0.12621355300848022,
      "grad_norm": 647.2166748046875,
      "learning_rate": 3.9924137152368516e-05,
      "loss": 76.7173,
      "step": 31240
    },
    {
      "epoch": 0.12625395427384786,
      "grad_norm": 2315.521728515625,
      "learning_rate": 3.9923893961834914e-05,
      "loss": 104.3867,
      "step": 31250
    },
    {
      "epoch": 0.12629435553921547,
      "grad_norm": 2408.8681640625,
      "learning_rate": 3.9923650382874744e-05,
      "loss": 139.8789,
      "step": 31260
    },
    {
      "epoch": 0.12633475680458311,
      "grad_norm": 631.905029296875,
      "learning_rate": 3.9923406415492755e-05,
      "loss": 69.1908,
      "step": 31270
    },
    {
      "epoch": 0.12637515806995075,
      "grad_norm": 1183.33056640625,
      "learning_rate": 3.99231620596937e-05,
      "loss": 66.9473,
      "step": 31280
    },
    {
      "epoch": 0.1264155593353184,
      "grad_norm": 657.8339233398438,
      "learning_rate": 3.9922917315482344e-05,
      "loss": 129.2818,
      "step": 31290
    },
    {
      "epoch": 0.126455960600686,
      "grad_norm": 635.81982421875,
      "learning_rate": 3.9922672182863456e-05,
      "loss": 87.3411,
      "step": 31300
    },
    {
      "epoch": 0.12649636186605365,
      "grad_norm": 1390.1109619140625,
      "learning_rate": 3.992242666184181e-05,
      "loss": 95.1961,
      "step": 31310
    },
    {
      "epoch": 0.1265367631314213,
      "grad_norm": 735.389892578125,
      "learning_rate": 3.9922180752422214e-05,
      "loss": 104.7503,
      "step": 31320
    },
    {
      "epoch": 0.1265771643967889,
      "grad_norm": 904.2584838867188,
      "learning_rate": 3.9921934454609435e-05,
      "loss": 59.2481,
      "step": 31330
    },
    {
      "epoch": 0.12661756566215654,
      "grad_norm": 927.343505859375,
      "learning_rate": 3.99216877684083e-05,
      "loss": 77.0618,
      "step": 31340
    },
    {
      "epoch": 0.12665796692752418,
      "grad_norm": 1172.96044921875,
      "learning_rate": 3.9921440693823596e-05,
      "loss": 79.6245,
      "step": 31350
    },
    {
      "epoch": 0.1266983681928918,
      "grad_norm": 1871.51416015625,
      "learning_rate": 3.992119323086015e-05,
      "loss": 74.9882,
      "step": 31360
    },
    {
      "epoch": 0.12673876945825943,
      "grad_norm": 257.1997985839844,
      "learning_rate": 3.9920945379522784e-05,
      "loss": 95.8861,
      "step": 31370
    },
    {
      "epoch": 0.12677917072362707,
      "grad_norm": 1125.3067626953125,
      "learning_rate": 3.992069713981634e-05,
      "loss": 113.5839,
      "step": 31380
    },
    {
      "epoch": 0.12681957198899468,
      "grad_norm": 4032.64111328125,
      "learning_rate": 3.992044851174564e-05,
      "loss": 124.356,
      "step": 31390
    },
    {
      "epoch": 0.12685997325436232,
      "grad_norm": 605.8938598632812,
      "learning_rate": 3.992019949531555e-05,
      "loss": 75.3891,
      "step": 31400
    },
    {
      "epoch": 0.12690037451972996,
      "grad_norm": 4289.74853515625,
      "learning_rate": 3.991995009053091e-05,
      "loss": 131.748,
      "step": 31410
    },
    {
      "epoch": 0.12694077578509758,
      "grad_norm": 926.93603515625,
      "learning_rate": 3.9919700297396585e-05,
      "loss": 88.6814,
      "step": 31420
    },
    {
      "epoch": 0.12698117705046522,
      "grad_norm": 413.9977111816406,
      "learning_rate": 3.991945011591745e-05,
      "loss": 111.6733,
      "step": 31430
    },
    {
      "epoch": 0.12702157831583286,
      "grad_norm": 778.385009765625,
      "learning_rate": 3.9919199546098377e-05,
      "loss": 64.7624,
      "step": 31440
    },
    {
      "epoch": 0.1270619795812005,
      "grad_norm": 2869.706787109375,
      "learning_rate": 3.991894858794426e-05,
      "loss": 112.8549,
      "step": 31450
    },
    {
      "epoch": 0.1271023808465681,
      "grad_norm": 2792.096923828125,
      "learning_rate": 3.991869724145998e-05,
      "loss": 97.986,
      "step": 31460
    },
    {
      "epoch": 0.12714278211193575,
      "grad_norm": 1144.2340087890625,
      "learning_rate": 3.991844550665045e-05,
      "loss": 93.456,
      "step": 31470
    },
    {
      "epoch": 0.1271831833773034,
      "grad_norm": 1057.1517333984375,
      "learning_rate": 3.991819338352057e-05,
      "loss": 78.6856,
      "step": 31480
    },
    {
      "epoch": 0.127223584642671,
      "grad_norm": 1895.421630859375,
      "learning_rate": 3.991794087207524e-05,
      "loss": 104.1628,
      "step": 31490
    },
    {
      "epoch": 0.12726398590803864,
      "grad_norm": 1157.536865234375,
      "learning_rate": 3.991768797231941e-05,
      "loss": 101.4132,
      "step": 31500
    },
    {
      "epoch": 0.12730438717340628,
      "grad_norm": 575.575439453125,
      "learning_rate": 3.9917434684258e-05,
      "loss": 130.7384,
      "step": 31510
    },
    {
      "epoch": 0.1273447884387739,
      "grad_norm": 1335.8681640625,
      "learning_rate": 3.9917181007895946e-05,
      "loss": 81.0369,
      "step": 31520
    },
    {
      "epoch": 0.12738518970414153,
      "grad_norm": 3140.3173828125,
      "learning_rate": 3.9916926943238185e-05,
      "loss": 104.6779,
      "step": 31530
    },
    {
      "epoch": 0.12742559096950917,
      "grad_norm": 390.1804504394531,
      "learning_rate": 3.991667249028969e-05,
      "loss": 117.5998,
      "step": 31540
    },
    {
      "epoch": 0.12746599223487678,
      "grad_norm": 1342.5087890625,
      "learning_rate": 3.9916417649055404e-05,
      "loss": 94.7564,
      "step": 31550
    },
    {
      "epoch": 0.12750639350024442,
      "grad_norm": 441.471435546875,
      "learning_rate": 3.991616241954031e-05,
      "loss": 124.2194,
      "step": 31560
    },
    {
      "epoch": 0.12754679476561206,
      "grad_norm": 922.2943115234375,
      "learning_rate": 3.9915906801749365e-05,
      "loss": 127.6032,
      "step": 31570
    },
    {
      "epoch": 0.12758719603097968,
      "grad_norm": 569.3106689453125,
      "learning_rate": 3.991565079568757e-05,
      "loss": 81.4049,
      "step": 31580
    },
    {
      "epoch": 0.12762759729634732,
      "grad_norm": 1510.0428466796875,
      "learning_rate": 3.991539440135991e-05,
      "loss": 89.6174,
      "step": 31590
    },
    {
      "epoch": 0.12766799856171496,
      "grad_norm": 447.491943359375,
      "learning_rate": 3.9915137618771386e-05,
      "loss": 74.3884,
      "step": 31600
    },
    {
      "epoch": 0.1277083998270826,
      "grad_norm": 466.8428649902344,
      "learning_rate": 3.9914880447927e-05,
      "loss": 86.6112,
      "step": 31610
    },
    {
      "epoch": 0.1277488010924502,
      "grad_norm": 1096.8792724609375,
      "learning_rate": 3.991462288883176e-05,
      "loss": 105.1563,
      "step": 31620
    },
    {
      "epoch": 0.12778920235781785,
      "grad_norm": 622.0668334960938,
      "learning_rate": 3.99143649414907e-05,
      "loss": 101.307,
      "step": 31630
    },
    {
      "epoch": 0.1278296036231855,
      "grad_norm": 939.9150390625,
      "learning_rate": 3.9914106605908845e-05,
      "loss": 119.6522,
      "step": 31640
    },
    {
      "epoch": 0.1278700048885531,
      "grad_norm": 1054.1038818359375,
      "learning_rate": 3.991384788209123e-05,
      "loss": 131.3634,
      "step": 31650
    },
    {
      "epoch": 0.12791040615392074,
      "grad_norm": 1325.1695556640625,
      "learning_rate": 3.991358877004289e-05,
      "loss": 136.0962,
      "step": 31660
    },
    {
      "epoch": 0.12795080741928838,
      "grad_norm": 1035.624267578125,
      "learning_rate": 3.991332926976888e-05,
      "loss": 110.5461,
      "step": 31670
    },
    {
      "epoch": 0.127991208684656,
      "grad_norm": 951.2764892578125,
      "learning_rate": 3.9913069381274274e-05,
      "loss": 72.4459,
      "step": 31680
    },
    {
      "epoch": 0.12803160995002363,
      "grad_norm": 1553.06787109375,
      "learning_rate": 3.991280910456413e-05,
      "loss": 107.8237,
      "step": 31690
    },
    {
      "epoch": 0.12807201121539127,
      "grad_norm": 379.1010437011719,
      "learning_rate": 3.991254843964352e-05,
      "loss": 117.5021,
      "step": 31700
    },
    {
      "epoch": 0.12811241248075889,
      "grad_norm": 1347.34423828125,
      "learning_rate": 3.9912287386517524e-05,
      "loss": 83.7117,
      "step": 31710
    },
    {
      "epoch": 0.12815281374612653,
      "grad_norm": 683.2993774414062,
      "learning_rate": 3.9912025945191233e-05,
      "loss": 113.7659,
      "step": 31720
    },
    {
      "epoch": 0.12819321501149417,
      "grad_norm": 857.4117431640625,
      "learning_rate": 3.991176411566974e-05,
      "loss": 81.4052,
      "step": 31730
    },
    {
      "epoch": 0.12823361627686178,
      "grad_norm": 3744.839111328125,
      "learning_rate": 3.991150189795816e-05,
      "loss": 138.6356,
      "step": 31740
    },
    {
      "epoch": 0.12827401754222942,
      "grad_norm": 1018.3668823242188,
      "learning_rate": 3.99112392920616e-05,
      "loss": 90.8482,
      "step": 31750
    },
    {
      "epoch": 0.12831441880759706,
      "grad_norm": 730.134033203125,
      "learning_rate": 3.991097629798519e-05,
      "loss": 118.8588,
      "step": 31760
    },
    {
      "epoch": 0.1283548200729647,
      "grad_norm": 1990.3084716796875,
      "learning_rate": 3.991071291573403e-05,
      "loss": 118.1557,
      "step": 31770
    },
    {
      "epoch": 0.1283952213383323,
      "grad_norm": 757.5863647460938,
      "learning_rate": 3.991044914531327e-05,
      "loss": 55.718,
      "step": 31780
    },
    {
      "epoch": 0.12843562260369995,
      "grad_norm": 439.8123779296875,
      "learning_rate": 3.991018498672806e-05,
      "loss": 66.3055,
      "step": 31790
    },
    {
      "epoch": 0.1284760238690676,
      "grad_norm": 758.9248046875,
      "learning_rate": 3.990992043998354e-05,
      "loss": 101.9324,
      "step": 31800
    },
    {
      "epoch": 0.1285164251344352,
      "grad_norm": 2390.60791015625,
      "learning_rate": 3.990965550508488e-05,
      "loss": 113.7801,
      "step": 31810
    },
    {
      "epoch": 0.12855682639980284,
      "grad_norm": 766.4974365234375,
      "learning_rate": 3.990939018203723e-05,
      "loss": 106.3282,
      "step": 31820
    },
    {
      "epoch": 0.12859722766517048,
      "grad_norm": 1003.3472900390625,
      "learning_rate": 3.990912447084576e-05,
      "loss": 93.9784,
      "step": 31830
    },
    {
      "epoch": 0.1286376289305381,
      "grad_norm": 1004.0360107421875,
      "learning_rate": 3.990885837151567e-05,
      "loss": 110.0491,
      "step": 31840
    },
    {
      "epoch": 0.12867803019590573,
      "grad_norm": 647.4901733398438,
      "learning_rate": 3.990859188405213e-05,
      "loss": 89.2669,
      "step": 31850
    },
    {
      "epoch": 0.12871843146127337,
      "grad_norm": 600.5079345703125,
      "learning_rate": 3.990832500846034e-05,
      "loss": 82.5083,
      "step": 31860
    },
    {
      "epoch": 0.128758832726641,
      "grad_norm": 566.7688598632812,
      "learning_rate": 3.990805774474551e-05,
      "loss": 76.502,
      "step": 31870
    },
    {
      "epoch": 0.12879923399200863,
      "grad_norm": 1008.888671875,
      "learning_rate": 3.990779009291284e-05,
      "loss": 71.2898,
      "step": 31880
    },
    {
      "epoch": 0.12883963525737627,
      "grad_norm": 957.80615234375,
      "learning_rate": 3.9907522052967556e-05,
      "loss": 85.8995,
      "step": 31890
    },
    {
      "epoch": 0.12888003652274388,
      "grad_norm": 877.7742309570312,
      "learning_rate": 3.990725362491488e-05,
      "loss": 103.1037,
      "step": 31900
    },
    {
      "epoch": 0.12892043778811152,
      "grad_norm": 1224.771728515625,
      "learning_rate": 3.990698480876005e-05,
      "loss": 63.2698,
      "step": 31910
    },
    {
      "epoch": 0.12896083905347916,
      "grad_norm": 1240.038818359375,
      "learning_rate": 3.9906715604508295e-05,
      "loss": 109.5601,
      "step": 31920
    },
    {
      "epoch": 0.1290012403188468,
      "grad_norm": 1355.4737548828125,
      "learning_rate": 3.990644601216487e-05,
      "loss": 105.8995,
      "step": 31930
    },
    {
      "epoch": 0.1290416415842144,
      "grad_norm": 549.8170166015625,
      "learning_rate": 3.990617603173504e-05,
      "loss": 61.5239,
      "step": 31940
    },
    {
      "epoch": 0.12908204284958205,
      "grad_norm": 976.7488403320312,
      "learning_rate": 3.9905905663224054e-05,
      "loss": 84.4975,
      "step": 31950
    },
    {
      "epoch": 0.1291224441149497,
      "grad_norm": 776.4035034179688,
      "learning_rate": 3.9905634906637185e-05,
      "loss": 66.8786,
      "step": 31960
    },
    {
      "epoch": 0.1291628453803173,
      "grad_norm": 441.0339050292969,
      "learning_rate": 3.9905363761979724e-05,
      "loss": 77.6645,
      "step": 31970
    },
    {
      "epoch": 0.12920324664568494,
      "grad_norm": 391.274169921875,
      "learning_rate": 3.9905092229256945e-05,
      "loss": 56.2057,
      "step": 31980
    },
    {
      "epoch": 0.12924364791105258,
      "grad_norm": 1062.639892578125,
      "learning_rate": 3.9904820308474145e-05,
      "loss": 114.6927,
      "step": 31990
    },
    {
      "epoch": 0.1292840491764202,
      "grad_norm": 1814.6654052734375,
      "learning_rate": 3.9904547999636625e-05,
      "loss": 110.2025,
      "step": 32000
    },
    {
      "epoch": 0.12932445044178784,
      "grad_norm": 756.2318725585938,
      "learning_rate": 3.9904275302749696e-05,
      "loss": 132.7259,
      "step": 32010
    },
    {
      "epoch": 0.12936485170715548,
      "grad_norm": 1501.403076171875,
      "learning_rate": 3.990400221781867e-05,
      "loss": 113.8047,
      "step": 32020
    },
    {
      "epoch": 0.1294052529725231,
      "grad_norm": 1061.5772705078125,
      "learning_rate": 3.990372874484887e-05,
      "loss": 88.4707,
      "step": 32030
    },
    {
      "epoch": 0.12944565423789073,
      "grad_norm": 697.8121337890625,
      "learning_rate": 3.9903454883845645e-05,
      "loss": 98.3409,
      "step": 32040
    },
    {
      "epoch": 0.12948605550325837,
      "grad_norm": 1300.261962890625,
      "learning_rate": 3.9903180634814304e-05,
      "loss": 71.091,
      "step": 32050
    },
    {
      "epoch": 0.12952645676862598,
      "grad_norm": 4183.81982421875,
      "learning_rate": 3.990290599776022e-05,
      "loss": 96.3512,
      "step": 32060
    },
    {
      "epoch": 0.12956685803399362,
      "grad_norm": 1181.3817138671875,
      "learning_rate": 3.990263097268873e-05,
      "loss": 98.9867,
      "step": 32070
    },
    {
      "epoch": 0.12960725929936126,
      "grad_norm": 551.7296752929688,
      "learning_rate": 3.990235555960521e-05,
      "loss": 160.0591,
      "step": 32080
    },
    {
      "epoch": 0.1296476605647289,
      "grad_norm": 629.874267578125,
      "learning_rate": 3.990207975851502e-05,
      "loss": 92.8718,
      "step": 32090
    },
    {
      "epoch": 0.1296880618300965,
      "grad_norm": 1450.927001953125,
      "learning_rate": 3.990180356942353e-05,
      "loss": 90.074,
      "step": 32100
    },
    {
      "epoch": 0.12972846309546415,
      "grad_norm": 1250.63916015625,
      "learning_rate": 3.990152699233614e-05,
      "loss": 72.6941,
      "step": 32110
    },
    {
      "epoch": 0.1297688643608318,
      "grad_norm": 1635.5433349609375,
      "learning_rate": 3.990125002725824e-05,
      "loss": 94.4942,
      "step": 32120
    },
    {
      "epoch": 0.1298092656261994,
      "grad_norm": 692.6416625976562,
      "learning_rate": 3.990097267419522e-05,
      "loss": 83.3047,
      "step": 32130
    },
    {
      "epoch": 0.12984966689156704,
      "grad_norm": 360.87469482421875,
      "learning_rate": 3.990069493315249e-05,
      "loss": 48.6074,
      "step": 32140
    },
    {
      "epoch": 0.12989006815693468,
      "grad_norm": 953.3905639648438,
      "learning_rate": 3.990041680413547e-05,
      "loss": 70.155,
      "step": 32150
    },
    {
      "epoch": 0.1299304694223023,
      "grad_norm": 1337.5194091796875,
      "learning_rate": 3.9900138287149575e-05,
      "loss": 135.2556,
      "step": 32160
    },
    {
      "epoch": 0.12997087068766994,
      "grad_norm": 1706.7265625,
      "learning_rate": 3.989985938220025e-05,
      "loss": 138.5848,
      "step": 32170
    },
    {
      "epoch": 0.13001127195303758,
      "grad_norm": 848.6858520507812,
      "learning_rate": 3.9899580089292904e-05,
      "loss": 88.3094,
      "step": 32180
    },
    {
      "epoch": 0.1300516732184052,
      "grad_norm": 1716.445556640625,
      "learning_rate": 3.9899300408433005e-05,
      "loss": 93.1287,
      "step": 32190
    },
    {
      "epoch": 0.13009207448377283,
      "grad_norm": 803.143798828125,
      "learning_rate": 3.989902033962601e-05,
      "loss": 79.1566,
      "step": 32200
    },
    {
      "epoch": 0.13013247574914047,
      "grad_norm": 643.987548828125,
      "learning_rate": 3.989873988287736e-05,
      "loss": 85.9153,
      "step": 32210
    },
    {
      "epoch": 0.13017287701450808,
      "grad_norm": 1115.3116455078125,
      "learning_rate": 3.989845903819253e-05,
      "loss": 97.8489,
      "step": 32220
    },
    {
      "epoch": 0.13021327827987572,
      "grad_norm": 822.09912109375,
      "learning_rate": 3.9898177805577e-05,
      "loss": 87.0368,
      "step": 32230
    },
    {
      "epoch": 0.13025367954524336,
      "grad_norm": 2172.662353515625,
      "learning_rate": 3.989789618503624e-05,
      "loss": 95.4219,
      "step": 32240
    },
    {
      "epoch": 0.130294080810611,
      "grad_norm": 406.08599853515625,
      "learning_rate": 3.989761417657577e-05,
      "loss": 99.9701,
      "step": 32250
    },
    {
      "epoch": 0.1303344820759786,
      "grad_norm": 984.9737548828125,
      "learning_rate": 3.989733178020105e-05,
      "loss": 57.6613,
      "step": 32260
    },
    {
      "epoch": 0.13037488334134625,
      "grad_norm": 1696.062744140625,
      "learning_rate": 3.989704899591761e-05,
      "loss": 68.0745,
      "step": 32270
    },
    {
      "epoch": 0.1304152846067139,
      "grad_norm": 1011.5084838867188,
      "learning_rate": 3.989676582373096e-05,
      "loss": 98.9419,
      "step": 32280
    },
    {
      "epoch": 0.1304556858720815,
      "grad_norm": 1103.0570068359375,
      "learning_rate": 3.989648226364661e-05,
      "loss": 134.7057,
      "step": 32290
    },
    {
      "epoch": 0.13049608713744915,
      "grad_norm": 261.1528015136719,
      "learning_rate": 3.98961983156701e-05,
      "loss": 55.7983,
      "step": 32300
    },
    {
      "epoch": 0.13053648840281679,
      "grad_norm": 1446.6749267578125,
      "learning_rate": 3.9895913979806955e-05,
      "loss": 109.7331,
      "step": 32310
    },
    {
      "epoch": 0.1305768896681844,
      "grad_norm": 822.2714233398438,
      "learning_rate": 3.9895629256062726e-05,
      "loss": 101.0507,
      "step": 32320
    },
    {
      "epoch": 0.13061729093355204,
      "grad_norm": 1948.6011962890625,
      "learning_rate": 3.989534414444296e-05,
      "loss": 111.6279,
      "step": 32330
    },
    {
      "epoch": 0.13065769219891968,
      "grad_norm": 609.5797729492188,
      "learning_rate": 3.989505864495322e-05,
      "loss": 67.6002,
      "step": 32340
    },
    {
      "epoch": 0.1306980934642873,
      "grad_norm": 6199.38720703125,
      "learning_rate": 3.989477275759907e-05,
      "loss": 101.5113,
      "step": 32350
    },
    {
      "epoch": 0.13073849472965493,
      "grad_norm": 0.0,
      "learning_rate": 3.989448648238608e-05,
      "loss": 123.2018,
      "step": 32360
    },
    {
      "epoch": 0.13077889599502257,
      "grad_norm": 606.24658203125,
      "learning_rate": 3.989419981931984e-05,
      "loss": 65.7179,
      "step": 32370
    },
    {
      "epoch": 0.13081929726039018,
      "grad_norm": 520.7660522460938,
      "learning_rate": 3.989391276840592e-05,
      "loss": 104.3083,
      "step": 32380
    },
    {
      "epoch": 0.13085969852575782,
      "grad_norm": 0.0,
      "learning_rate": 3.989362532964994e-05,
      "loss": 73.2306,
      "step": 32390
    },
    {
      "epoch": 0.13090009979112546,
      "grad_norm": 935.7091064453125,
      "learning_rate": 3.989333750305749e-05,
      "loss": 116.7558,
      "step": 32400
    },
    {
      "epoch": 0.1309405010564931,
      "grad_norm": 1432.7255859375,
      "learning_rate": 3.9893049288634174e-05,
      "loss": 124.7905,
      "step": 32410
    },
    {
      "epoch": 0.13098090232186071,
      "grad_norm": 776.0060424804688,
      "learning_rate": 3.989276068638563e-05,
      "loss": 83.9067,
      "step": 32420
    },
    {
      "epoch": 0.13102130358722835,
      "grad_norm": 1112.199951171875,
      "learning_rate": 3.9892471696317467e-05,
      "loss": 107.2018,
      "step": 32430
    },
    {
      "epoch": 0.131061704852596,
      "grad_norm": 849.1611328125,
      "learning_rate": 3.9892182318435336e-05,
      "loss": 91.3857,
      "step": 32440
    },
    {
      "epoch": 0.1311021061179636,
      "grad_norm": 1403.359130859375,
      "learning_rate": 3.989189255274487e-05,
      "loss": 112.8682,
      "step": 32450
    },
    {
      "epoch": 0.13114250738333125,
      "grad_norm": 912.6363525390625,
      "learning_rate": 3.989160239925171e-05,
      "loss": 104.4206,
      "step": 32460
    },
    {
      "epoch": 0.1311829086486989,
      "grad_norm": 2311.5087890625,
      "learning_rate": 3.989131185796153e-05,
      "loss": 105.4599,
      "step": 32470
    },
    {
      "epoch": 0.1312233099140665,
      "grad_norm": 817.505859375,
      "learning_rate": 3.989102092887997e-05,
      "loss": 69.9134,
      "step": 32480
    },
    {
      "epoch": 0.13126371117943414,
      "grad_norm": 1116.56298828125,
      "learning_rate": 3.9890729612012726e-05,
      "loss": 94.8134,
      "step": 32490
    },
    {
      "epoch": 0.13130411244480178,
      "grad_norm": 1570.620361328125,
      "learning_rate": 3.989043790736547e-05,
      "loss": 131.5534,
      "step": 32500
    },
    {
      "epoch": 0.1313445137101694,
      "grad_norm": 1480.2958984375,
      "learning_rate": 3.989014581494388e-05,
      "loss": 104.3328,
      "step": 32510
    },
    {
      "epoch": 0.13138491497553703,
      "grad_norm": 1166.97998046875,
      "learning_rate": 3.9889853334753666e-05,
      "loss": 82.1462,
      "step": 32520
    },
    {
      "epoch": 0.13142531624090467,
      "grad_norm": 950.3305053710938,
      "learning_rate": 3.988956046680051e-05,
      "loss": 148.301,
      "step": 32530
    },
    {
      "epoch": 0.13146571750627228,
      "grad_norm": 1937.57080078125,
      "learning_rate": 3.9889267211090145e-05,
      "loss": 125.095,
      "step": 32540
    },
    {
      "epoch": 0.13150611877163992,
      "grad_norm": 808.1902465820312,
      "learning_rate": 3.988897356762827e-05,
      "loss": 84.5435,
      "step": 32550
    },
    {
      "epoch": 0.13154652003700756,
      "grad_norm": 2491.928955078125,
      "learning_rate": 3.988867953642062e-05,
      "loss": 136.9711,
      "step": 32560
    },
    {
      "epoch": 0.1315869213023752,
      "grad_norm": 383.66302490234375,
      "learning_rate": 3.9888385117472914e-05,
      "loss": 76.6657,
      "step": 32570
    },
    {
      "epoch": 0.13162732256774282,
      "grad_norm": 1040.90087890625,
      "learning_rate": 3.9888090310790904e-05,
      "loss": 61.5886,
      "step": 32580
    },
    {
      "epoch": 0.13166772383311046,
      "grad_norm": 782.1383666992188,
      "learning_rate": 3.9887795116380336e-05,
      "loss": 104.2784,
      "step": 32590
    },
    {
      "epoch": 0.1317081250984781,
      "grad_norm": 1465.7843017578125,
      "learning_rate": 3.988749953424696e-05,
      "loss": 90.0144,
      "step": 32600
    },
    {
      "epoch": 0.1317485263638457,
      "grad_norm": 661.722412109375,
      "learning_rate": 3.988720356439655e-05,
      "loss": 123.0394,
      "step": 32610
    },
    {
      "epoch": 0.13178892762921335,
      "grad_norm": 1469.308837890625,
      "learning_rate": 3.988690720683486e-05,
      "loss": 110.255,
      "step": 32620
    },
    {
      "epoch": 0.131829328894581,
      "grad_norm": 491.6518859863281,
      "learning_rate": 3.988661046156768e-05,
      "loss": 114.0818,
      "step": 32630
    },
    {
      "epoch": 0.1318697301599486,
      "grad_norm": 610.9449462890625,
      "learning_rate": 3.988631332860079e-05,
      "loss": 78.5732,
      "step": 32640
    },
    {
      "epoch": 0.13191013142531624,
      "grad_norm": 762.2991333007812,
      "learning_rate": 3.988601580793998e-05,
      "loss": 104.1977,
      "step": 32650
    },
    {
      "epoch": 0.13195053269068388,
      "grad_norm": 1247.912109375,
      "learning_rate": 3.988571789959106e-05,
      "loss": 108.5621,
      "step": 32660
    },
    {
      "epoch": 0.1319909339560515,
      "grad_norm": 742.3600463867188,
      "learning_rate": 3.988541960355982e-05,
      "loss": 77.3082,
      "step": 32670
    },
    {
      "epoch": 0.13203133522141913,
      "grad_norm": 780.1296997070312,
      "learning_rate": 3.98851209198521e-05,
      "loss": 74.177,
      "step": 32680
    },
    {
      "epoch": 0.13207173648678677,
      "grad_norm": 605.8449096679688,
      "learning_rate": 3.988482184847371e-05,
      "loss": 77.8987,
      "step": 32690
    },
    {
      "epoch": 0.13211213775215438,
      "grad_norm": 705.7808227539062,
      "learning_rate": 3.988452238943047e-05,
      "loss": 78.1375,
      "step": 32700
    },
    {
      "epoch": 0.13215253901752202,
      "grad_norm": 1150.5770263671875,
      "learning_rate": 3.9884222542728236e-05,
      "loss": 82.5151,
      "step": 32710
    },
    {
      "epoch": 0.13219294028288966,
      "grad_norm": 1019.019775390625,
      "learning_rate": 3.988392230837285e-05,
      "loss": 121.0599,
      "step": 32720
    },
    {
      "epoch": 0.1322333415482573,
      "grad_norm": 797.7850341796875,
      "learning_rate": 3.988362168637015e-05,
      "loss": 98.1945,
      "step": 32730
    },
    {
      "epoch": 0.13227374281362492,
      "grad_norm": 2180.442138671875,
      "learning_rate": 3.9883320676726015e-05,
      "loss": 79.0123,
      "step": 32740
    },
    {
      "epoch": 0.13231414407899256,
      "grad_norm": 1715.8033447265625,
      "learning_rate": 3.988301927944631e-05,
      "loss": 110.9263,
      "step": 32750
    },
    {
      "epoch": 0.1323545453443602,
      "grad_norm": 739.5059204101562,
      "learning_rate": 3.988271749453691e-05,
      "loss": 111.0484,
      "step": 32760
    },
    {
      "epoch": 0.1323949466097278,
      "grad_norm": 866.54833984375,
      "learning_rate": 3.988241532200369e-05,
      "loss": 95.5286,
      "step": 32770
    },
    {
      "epoch": 0.13243534787509545,
      "grad_norm": 1200.807373046875,
      "learning_rate": 3.9882112761852544e-05,
      "loss": 85.0713,
      "step": 32780
    },
    {
      "epoch": 0.1324757491404631,
      "grad_norm": 2249.028564453125,
      "learning_rate": 3.9881809814089376e-05,
      "loss": 111.7555,
      "step": 32790
    },
    {
      "epoch": 0.1325161504058307,
      "grad_norm": 678.5682983398438,
      "learning_rate": 3.9881506478720095e-05,
      "loss": 79.8005,
      "step": 32800
    },
    {
      "epoch": 0.13255655167119834,
      "grad_norm": 1730.4722900390625,
      "learning_rate": 3.9881202755750604e-05,
      "loss": 110.668,
      "step": 32810
    },
    {
      "epoch": 0.13259695293656598,
      "grad_norm": 547.8741455078125,
      "learning_rate": 3.988089864518683e-05,
      "loss": 82.1663,
      "step": 32820
    },
    {
      "epoch": 0.1326373542019336,
      "grad_norm": 678.1209106445312,
      "learning_rate": 3.988059414703471e-05,
      "loss": 92.4558,
      "step": 32830
    },
    {
      "epoch": 0.13267775546730123,
      "grad_norm": 1880.7825927734375,
      "learning_rate": 3.9880289261300166e-05,
      "loss": 167.3718,
      "step": 32840
    },
    {
      "epoch": 0.13271815673266887,
      "grad_norm": 1045.5400390625,
      "learning_rate": 3.987998398798914e-05,
      "loss": 141.326,
      "step": 32850
    },
    {
      "epoch": 0.13275855799803649,
      "grad_norm": 1256.3192138671875,
      "learning_rate": 3.987967832710761e-05,
      "loss": 123.7676,
      "step": 32860
    },
    {
      "epoch": 0.13279895926340413,
      "grad_norm": 415.38787841796875,
      "learning_rate": 3.98793722786615e-05,
      "loss": 85.7358,
      "step": 32870
    },
    {
      "epoch": 0.13283936052877177,
      "grad_norm": 2784.51953125,
      "learning_rate": 3.9879065842656796e-05,
      "loss": 88.3061,
      "step": 32880
    },
    {
      "epoch": 0.1328797617941394,
      "grad_norm": 1133.65771484375,
      "learning_rate": 3.987875901909947e-05,
      "loss": 114.6212,
      "step": 32890
    },
    {
      "epoch": 0.13292016305950702,
      "grad_norm": 654.9711303710938,
      "learning_rate": 3.9878451807995496e-05,
      "loss": 75.8217,
      "step": 32900
    },
    {
      "epoch": 0.13296056432487466,
      "grad_norm": 1000.1675415039062,
      "learning_rate": 3.987814420935088e-05,
      "loss": 73.452,
      "step": 32910
    },
    {
      "epoch": 0.1330009655902423,
      "grad_norm": 804.9745483398438,
      "learning_rate": 3.987783622317161e-05,
      "loss": 66.1093,
      "step": 32920
    },
    {
      "epoch": 0.1330413668556099,
      "grad_norm": 1595.5555419921875,
      "learning_rate": 3.987752784946368e-05,
      "loss": 130.9591,
      "step": 32930
    },
    {
      "epoch": 0.13308176812097755,
      "grad_norm": 724.1141357421875,
      "learning_rate": 3.9877219088233115e-05,
      "loss": 111.874,
      "step": 32940
    },
    {
      "epoch": 0.1331221693863452,
      "grad_norm": 1259.160400390625,
      "learning_rate": 3.987690993948594e-05,
      "loss": 110.3312,
      "step": 32950
    },
    {
      "epoch": 0.1331625706517128,
      "grad_norm": 468.20782470703125,
      "learning_rate": 3.9876600403228154e-05,
      "loss": 84.0473,
      "step": 32960
    },
    {
      "epoch": 0.13320297191708044,
      "grad_norm": 525.40087890625,
      "learning_rate": 3.987629047946582e-05,
      "loss": 88.113,
      "step": 32970
    },
    {
      "epoch": 0.13324337318244808,
      "grad_norm": 728.6627197265625,
      "learning_rate": 3.9875980168204976e-05,
      "loss": 94.1612,
      "step": 32980
    },
    {
      "epoch": 0.1332837744478157,
      "grad_norm": 685.3392333984375,
      "learning_rate": 3.987566946945166e-05,
      "loss": 82.9419,
      "step": 32990
    },
    {
      "epoch": 0.13332417571318333,
      "grad_norm": 1111.914306640625,
      "learning_rate": 3.987535838321193e-05,
      "loss": 101.6496,
      "step": 33000
    },
    {
      "epoch": 0.13336457697855097,
      "grad_norm": 488.5706481933594,
      "learning_rate": 3.987504690949186e-05,
      "loss": 93.6909,
      "step": 33010
    },
    {
      "epoch": 0.1334049782439186,
      "grad_norm": 513.7379150390625,
      "learning_rate": 3.987473504829752e-05,
      "loss": 75.1137,
      "step": 33020
    },
    {
      "epoch": 0.13344537950928623,
      "grad_norm": 913.1249389648438,
      "learning_rate": 3.987442279963499e-05,
      "loss": 70.1443,
      "step": 33030
    },
    {
      "epoch": 0.13348578077465387,
      "grad_norm": 517.8516845703125,
      "learning_rate": 3.9874110163510345e-05,
      "loss": 109.9103,
      "step": 33040
    },
    {
      "epoch": 0.1335261820400215,
      "grad_norm": 1112.8758544921875,
      "learning_rate": 3.98737971399297e-05,
      "loss": 80.9284,
      "step": 33050
    },
    {
      "epoch": 0.13356658330538912,
      "grad_norm": 909.37451171875,
      "learning_rate": 3.987348372889915e-05,
      "loss": 109.8047,
      "step": 33060
    },
    {
      "epoch": 0.13360698457075676,
      "grad_norm": 1172.3194580078125,
      "learning_rate": 3.9873169930424796e-05,
      "loss": 82.3379,
      "step": 33070
    },
    {
      "epoch": 0.1336473858361244,
      "grad_norm": 1014.2901000976562,
      "learning_rate": 3.987285574451276e-05,
      "loss": 114.6765,
      "step": 33080
    },
    {
      "epoch": 0.133687787101492,
      "grad_norm": 282.9718322753906,
      "learning_rate": 3.987254117116918e-05,
      "loss": 105.4184,
      "step": 33090
    },
    {
      "epoch": 0.13372818836685965,
      "grad_norm": 702.3818359375,
      "learning_rate": 3.987222621040017e-05,
      "loss": 124.3384,
      "step": 33100
    },
    {
      "epoch": 0.1337685896322273,
      "grad_norm": 706.9837036132812,
      "learning_rate": 3.987191086221189e-05,
      "loss": 101.213,
      "step": 33110
    },
    {
      "epoch": 0.1338089908975949,
      "grad_norm": 1250.2989501953125,
      "learning_rate": 3.9871595126610466e-05,
      "loss": 90.6203,
      "step": 33120
    },
    {
      "epoch": 0.13384939216296254,
      "grad_norm": 2235.422607421875,
      "learning_rate": 3.987127900360207e-05,
      "loss": 96.2993,
      "step": 33130
    },
    {
      "epoch": 0.13388979342833018,
      "grad_norm": 960.5076904296875,
      "learning_rate": 3.9870962493192856e-05,
      "loss": 93.8741,
      "step": 33140
    },
    {
      "epoch": 0.1339301946936978,
      "grad_norm": 998.1218872070312,
      "learning_rate": 3.9870645595389e-05,
      "loss": 75.5594,
      "step": 33150
    },
    {
      "epoch": 0.13397059595906544,
      "grad_norm": 516.9995727539062,
      "learning_rate": 3.987032831019668e-05,
      "loss": 113.7387,
      "step": 33160
    },
    {
      "epoch": 0.13401099722443308,
      "grad_norm": 1901.6185302734375,
      "learning_rate": 3.987001063762208e-05,
      "loss": 86.161,
      "step": 33170
    },
    {
      "epoch": 0.1340513984898007,
      "grad_norm": 753.5205078125,
      "learning_rate": 3.986969257767139e-05,
      "loss": 73.1547,
      "step": 33180
    },
    {
      "epoch": 0.13409179975516833,
      "grad_norm": 1136.2593994140625,
      "learning_rate": 3.9869374130350805e-05,
      "loss": 81.2451,
      "step": 33190
    },
    {
      "epoch": 0.13413220102053597,
      "grad_norm": 1040.9864501953125,
      "learning_rate": 3.986905529566655e-05,
      "loss": 81.6566,
      "step": 33200
    },
    {
      "epoch": 0.1341726022859036,
      "grad_norm": 828.9725952148438,
      "learning_rate": 3.986873607362484e-05,
      "loss": 65.9397,
      "step": 33210
    },
    {
      "epoch": 0.13421300355127122,
      "grad_norm": 596.030517578125,
      "learning_rate": 3.986841646423188e-05,
      "loss": 64.6862,
      "step": 33220
    },
    {
      "epoch": 0.13425340481663886,
      "grad_norm": 326.7821350097656,
      "learning_rate": 3.9868096467493924e-05,
      "loss": 99.5571,
      "step": 33230
    },
    {
      "epoch": 0.1342938060820065,
      "grad_norm": 1642.4222412109375,
      "learning_rate": 3.9867776083417186e-05,
      "loss": 84.424,
      "step": 33240
    },
    {
      "epoch": 0.1343342073473741,
      "grad_norm": 1301.66552734375,
      "learning_rate": 3.986745531200793e-05,
      "loss": 90.8854,
      "step": 33250
    },
    {
      "epoch": 0.13437460861274175,
      "grad_norm": 1154.9903564453125,
      "learning_rate": 3.9867134153272404e-05,
      "loss": 81.8859,
      "step": 33260
    },
    {
      "epoch": 0.1344150098781094,
      "grad_norm": 660.017578125,
      "learning_rate": 3.9866812607216875e-05,
      "loss": 104.3724,
      "step": 33270
    },
    {
      "epoch": 0.134455411143477,
      "grad_norm": 671.1018676757812,
      "learning_rate": 3.9866490673847596e-05,
      "loss": 96.5107,
      "step": 33280
    },
    {
      "epoch": 0.13449581240884464,
      "grad_norm": 594.5077514648438,
      "learning_rate": 3.9866168353170855e-05,
      "loss": 87.6706,
      "step": 33290
    },
    {
      "epoch": 0.13453621367421228,
      "grad_norm": 764.1627197265625,
      "learning_rate": 3.986584564519294e-05,
      "loss": 94.7507,
      "step": 33300
    },
    {
      "epoch": 0.1345766149395799,
      "grad_norm": 1440.6705322265625,
      "learning_rate": 3.9865522549920135e-05,
      "loss": 89.43,
      "step": 33310
    },
    {
      "epoch": 0.13461701620494754,
      "grad_norm": 509.2161560058594,
      "learning_rate": 3.986519906735874e-05,
      "loss": 127.8514,
      "step": 33320
    },
    {
      "epoch": 0.13465741747031518,
      "grad_norm": 330.7436218261719,
      "learning_rate": 3.986487519751506e-05,
      "loss": 71.8705,
      "step": 33330
    },
    {
      "epoch": 0.1346978187356828,
      "grad_norm": 3115.505859375,
      "learning_rate": 3.9864550940395413e-05,
      "loss": 168.864,
      "step": 33340
    },
    {
      "epoch": 0.13473822000105043,
      "grad_norm": 708.8053588867188,
      "learning_rate": 3.9864226296006114e-05,
      "loss": 99.1782,
      "step": 33350
    },
    {
      "epoch": 0.13477862126641807,
      "grad_norm": 827.0519409179688,
      "learning_rate": 3.986390126435351e-05,
      "loss": 88.3666,
      "step": 33360
    },
    {
      "epoch": 0.1348190225317857,
      "grad_norm": 348.74884033203125,
      "learning_rate": 3.986357584544391e-05,
      "loss": 79.9038,
      "step": 33370
    },
    {
      "epoch": 0.13485942379715332,
      "grad_norm": 809.9948120117188,
      "learning_rate": 3.986325003928367e-05,
      "loss": 75.5851,
      "step": 33380
    },
    {
      "epoch": 0.13489982506252096,
      "grad_norm": 530.9696044921875,
      "learning_rate": 3.986292384587916e-05,
      "loss": 74.4206,
      "step": 33390
    },
    {
      "epoch": 0.1349402263278886,
      "grad_norm": 840.42236328125,
      "learning_rate": 3.986259726523671e-05,
      "loss": 98.9562,
      "step": 33400
    },
    {
      "epoch": 0.1349806275932562,
      "grad_norm": 362.3224182128906,
      "learning_rate": 3.9862270297362704e-05,
      "loss": 74.7036,
      "step": 33410
    },
    {
      "epoch": 0.13502102885862385,
      "grad_norm": 4263.97314453125,
      "learning_rate": 3.986194294226351e-05,
      "loss": 132.8398,
      "step": 33420
    },
    {
      "epoch": 0.1350614301239915,
      "grad_norm": 982.0714111328125,
      "learning_rate": 3.986161519994552e-05,
      "loss": 112.7037,
      "step": 33430
    },
    {
      "epoch": 0.1351018313893591,
      "grad_norm": 961.3419799804688,
      "learning_rate": 3.986128707041511e-05,
      "loss": 95.7062,
      "step": 33440
    },
    {
      "epoch": 0.13514223265472675,
      "grad_norm": 749.6942138671875,
      "learning_rate": 3.9860958553678694e-05,
      "loss": 115.0838,
      "step": 33450
    },
    {
      "epoch": 0.13518263392009439,
      "grad_norm": 0.0,
      "learning_rate": 3.986062964974265e-05,
      "loss": 68.6302,
      "step": 33460
    },
    {
      "epoch": 0.135223035185462,
      "grad_norm": 1634.578125,
      "learning_rate": 3.9860300358613416e-05,
      "loss": 94.7061,
      "step": 33470
    },
    {
      "epoch": 0.13526343645082964,
      "grad_norm": 504.92449951171875,
      "learning_rate": 3.98599706802974e-05,
      "loss": 69.5252,
      "step": 33480
    },
    {
      "epoch": 0.13530383771619728,
      "grad_norm": 2182.119140625,
      "learning_rate": 3.985964061480103e-05,
      "loss": 91.9078,
      "step": 33490
    },
    {
      "epoch": 0.1353442389815649,
      "grad_norm": 669.2144775390625,
      "learning_rate": 3.985931016213074e-05,
      "loss": 65.255,
      "step": 33500
    },
    {
      "epoch": 0.13538464024693253,
      "grad_norm": 678.09521484375,
      "learning_rate": 3.985897932229298e-05,
      "loss": 130.8218,
      "step": 33510
    },
    {
      "epoch": 0.13542504151230017,
      "grad_norm": 860.627685546875,
      "learning_rate": 3.985864809529419e-05,
      "loss": 136.829,
      "step": 33520
    },
    {
      "epoch": 0.13546544277766778,
      "grad_norm": 691.2222290039062,
      "learning_rate": 3.985831648114083e-05,
      "loss": 106.527,
      "step": 33530
    },
    {
      "epoch": 0.13550584404303542,
      "grad_norm": 677.6774291992188,
      "learning_rate": 3.985798447983937e-05,
      "loss": 84.5683,
      "step": 33540
    },
    {
      "epoch": 0.13554624530840306,
      "grad_norm": 463.0068359375,
      "learning_rate": 3.9857652091396276e-05,
      "loss": 71.5883,
      "step": 33550
    },
    {
      "epoch": 0.1355866465737707,
      "grad_norm": 1355.86083984375,
      "learning_rate": 3.985731931581804e-05,
      "loss": 75.3525,
      "step": 33560
    },
    {
      "epoch": 0.13562704783913831,
      "grad_norm": 595.2366333007812,
      "learning_rate": 3.985698615311113e-05,
      "loss": 72.4136,
      "step": 33570
    },
    {
      "epoch": 0.13566744910450595,
      "grad_norm": 838.0628662109375,
      "learning_rate": 3.985665260328205e-05,
      "loss": 111.8475,
      "step": 33580
    },
    {
      "epoch": 0.1357078503698736,
      "grad_norm": 607.656005859375,
      "learning_rate": 3.985631866633731e-05,
      "loss": 92.9903,
      "step": 33590
    },
    {
      "epoch": 0.1357482516352412,
      "grad_norm": 802.7066650390625,
      "learning_rate": 3.9855984342283414e-05,
      "loss": 97.0624,
      "step": 33600
    },
    {
      "epoch": 0.13578865290060885,
      "grad_norm": 1092.105712890625,
      "learning_rate": 3.9855649631126884e-05,
      "loss": 117.6566,
      "step": 33610
    },
    {
      "epoch": 0.1358290541659765,
      "grad_norm": 450.5210266113281,
      "learning_rate": 3.985531453287424e-05,
      "loss": 92.6769,
      "step": 33620
    },
    {
      "epoch": 0.1358694554313441,
      "grad_norm": 439.0806884765625,
      "learning_rate": 3.985497904753201e-05,
      "loss": 84.2845,
      "step": 33630
    },
    {
      "epoch": 0.13590985669671174,
      "grad_norm": 1899.7283935546875,
      "learning_rate": 3.9854643175106756e-05,
      "loss": 102.3803,
      "step": 33640
    },
    {
      "epoch": 0.13595025796207938,
      "grad_norm": 767.927490234375,
      "learning_rate": 3.9854306915605e-05,
      "loss": 146.8131,
      "step": 33650
    },
    {
      "epoch": 0.135990659227447,
      "grad_norm": 864.8515014648438,
      "learning_rate": 3.985397026903331e-05,
      "loss": 94.1728,
      "step": 33660
    },
    {
      "epoch": 0.13603106049281463,
      "grad_norm": 1093.5802001953125,
      "learning_rate": 3.985363323539825e-05,
      "loss": 107.6846,
      "step": 33670
    },
    {
      "epoch": 0.13607146175818227,
      "grad_norm": 1755.8154296875,
      "learning_rate": 3.9853295814706395e-05,
      "loss": 113.3186,
      "step": 33680
    },
    {
      "epoch": 0.13611186302354988,
      "grad_norm": 992.8015747070312,
      "learning_rate": 3.98529580069643e-05,
      "loss": 107.5117,
      "step": 33690
    },
    {
      "epoch": 0.13615226428891752,
      "grad_norm": 536.5291137695312,
      "learning_rate": 3.985261981217858e-05,
      "loss": 72.3382,
      "step": 33700
    },
    {
      "epoch": 0.13619266555428516,
      "grad_norm": 1494.7418212890625,
      "learning_rate": 3.985228123035582e-05,
      "loss": 69.5633,
      "step": 33710
    },
    {
      "epoch": 0.1362330668196528,
      "grad_norm": 362.628173828125,
      "learning_rate": 3.98519422615026e-05,
      "loss": 127.147,
      "step": 33720
    },
    {
      "epoch": 0.13627346808502042,
      "grad_norm": 915.39208984375,
      "learning_rate": 3.985160290562556e-05,
      "loss": 134.0353,
      "step": 33730
    },
    {
      "epoch": 0.13631386935038806,
      "grad_norm": 1054.449462890625,
      "learning_rate": 3.98512631627313e-05,
      "loss": 89.7184,
      "step": 33740
    },
    {
      "epoch": 0.1363542706157557,
      "grad_norm": 892.0343017578125,
      "learning_rate": 3.985092303282645e-05,
      "loss": 68.4411,
      "step": 33750
    },
    {
      "epoch": 0.1363946718811233,
      "grad_norm": 1595.01953125,
      "learning_rate": 3.985058251591762e-05,
      "loss": 92.7033,
      "step": 33760
    },
    {
      "epoch": 0.13643507314649095,
      "grad_norm": 1104.977783203125,
      "learning_rate": 3.985024161201147e-05,
      "loss": 91.0335,
      "step": 33770
    },
    {
      "epoch": 0.1364754744118586,
      "grad_norm": 0.0,
      "learning_rate": 3.984990032111465e-05,
      "loss": 58.1744,
      "step": 33780
    },
    {
      "epoch": 0.1365158756772262,
      "grad_norm": 1564.0908203125,
      "learning_rate": 3.984955864323379e-05,
      "loss": 75.7117,
      "step": 33790
    },
    {
      "epoch": 0.13655627694259384,
      "grad_norm": 700.7978515625,
      "learning_rate": 3.984921657837557e-05,
      "loss": 108.2256,
      "step": 33800
    },
    {
      "epoch": 0.13659667820796148,
      "grad_norm": 1223.709716796875,
      "learning_rate": 3.9848874126546654e-05,
      "loss": 150.9498,
      "step": 33810
    },
    {
      "epoch": 0.1366370794733291,
      "grad_norm": 732.4114379882812,
      "learning_rate": 3.984853128775372e-05,
      "loss": 60.9993,
      "step": 33820
    },
    {
      "epoch": 0.13667748073869673,
      "grad_norm": 1338.5892333984375,
      "learning_rate": 3.9848188062003445e-05,
      "loss": 123.3062,
      "step": 33830
    },
    {
      "epoch": 0.13671788200406437,
      "grad_norm": 696.5016479492188,
      "learning_rate": 3.984784444930253e-05,
      "loss": 101.0902,
      "step": 33840
    },
    {
      "epoch": 0.13675828326943198,
      "grad_norm": 565.2360229492188,
      "learning_rate": 3.9847500449657666e-05,
      "loss": 88.9729,
      "step": 33850
    },
    {
      "epoch": 0.13679868453479962,
      "grad_norm": 788.9739379882812,
      "learning_rate": 3.984715606307556e-05,
      "loss": 79.289,
      "step": 33860
    },
    {
      "epoch": 0.13683908580016726,
      "grad_norm": 531.5402221679688,
      "learning_rate": 3.984681128956294e-05,
      "loss": 70.041,
      "step": 33870
    },
    {
      "epoch": 0.1368794870655349,
      "grad_norm": 541.3567504882812,
      "learning_rate": 3.9846466129126506e-05,
      "loss": 142.128,
      "step": 33880
    },
    {
      "epoch": 0.13691988833090252,
      "grad_norm": 640.6295166015625,
      "learning_rate": 3.9846120581773e-05,
      "loss": 108.9354,
      "step": 33890
    },
    {
      "epoch": 0.13696028959627016,
      "grad_norm": 474.4267578125,
      "learning_rate": 3.984577464750916e-05,
      "loss": 83.6102,
      "step": 33900
    },
    {
      "epoch": 0.1370006908616378,
      "grad_norm": 769.0298461914062,
      "learning_rate": 3.984542832634172e-05,
      "loss": 86.5954,
      "step": 33910
    },
    {
      "epoch": 0.1370410921270054,
      "grad_norm": 1042.3206787109375,
      "learning_rate": 3.984508161827743e-05,
      "loss": 131.0708,
      "step": 33920
    },
    {
      "epoch": 0.13708149339237305,
      "grad_norm": 1229.37158203125,
      "learning_rate": 3.984473452332307e-05,
      "loss": 72.523,
      "step": 33930
    },
    {
      "epoch": 0.1371218946577407,
      "grad_norm": 1179.4063720703125,
      "learning_rate": 3.9844387041485396e-05,
      "loss": 121.1243,
      "step": 33940
    },
    {
      "epoch": 0.1371622959231083,
      "grad_norm": 616.3158569335938,
      "learning_rate": 3.9844039172771174e-05,
      "loss": 74.466,
      "step": 33950
    },
    {
      "epoch": 0.13720269718847594,
      "grad_norm": 1129.1138916015625,
      "learning_rate": 3.984369091718719e-05,
      "loss": 115.7123,
      "step": 33960
    },
    {
      "epoch": 0.13724309845384358,
      "grad_norm": 907.6840209960938,
      "learning_rate": 3.984334227474023e-05,
      "loss": 114.4695,
      "step": 33970
    },
    {
      "epoch": 0.1372834997192112,
      "grad_norm": 2617.643798828125,
      "learning_rate": 3.984299324543711e-05,
      "loss": 113.5119,
      "step": 33980
    },
    {
      "epoch": 0.13732390098457883,
      "grad_norm": 1317.3724365234375,
      "learning_rate": 3.984264382928461e-05,
      "loss": 111.3625,
      "step": 33990
    },
    {
      "epoch": 0.13736430224994647,
      "grad_norm": 1482.2108154296875,
      "learning_rate": 3.9842294026289565e-05,
      "loss": 123.0276,
      "step": 34000
    },
    {
      "epoch": 0.13740470351531409,
      "grad_norm": 471.9527282714844,
      "learning_rate": 3.984194383645877e-05,
      "loss": 71.2953,
      "step": 34010
    },
    {
      "epoch": 0.13744510478068173,
      "grad_norm": 645.5573120117188,
      "learning_rate": 3.984159325979907e-05,
      "loss": 117.5256,
      "step": 34020
    },
    {
      "epoch": 0.13748550604604937,
      "grad_norm": 644.1488037109375,
      "learning_rate": 3.9841242296317294e-05,
      "loss": 54.6307,
      "step": 34030
    },
    {
      "epoch": 0.137525907311417,
      "grad_norm": 1674.2548828125,
      "learning_rate": 3.984089094602028e-05,
      "loss": 61.2714,
      "step": 34040
    },
    {
      "epoch": 0.13756630857678462,
      "grad_norm": 1002.239501953125,
      "learning_rate": 3.984053920891489e-05,
      "loss": 78.2601,
      "step": 34050
    },
    {
      "epoch": 0.13760670984215226,
      "grad_norm": 785.2839965820312,
      "learning_rate": 3.9840187085007965e-05,
      "loss": 103.6112,
      "step": 34060
    },
    {
      "epoch": 0.1376471111075199,
      "grad_norm": 914.4097290039062,
      "learning_rate": 3.983983457430639e-05,
      "loss": 66.841,
      "step": 34070
    },
    {
      "epoch": 0.1376875123728875,
      "grad_norm": 681.90234375,
      "learning_rate": 3.983948167681701e-05,
      "loss": 68.6914,
      "step": 34080
    },
    {
      "epoch": 0.13772791363825515,
      "grad_norm": 2546.71337890625,
      "learning_rate": 3.983912839254673e-05,
      "loss": 108.741,
      "step": 34090
    },
    {
      "epoch": 0.1377683149036228,
      "grad_norm": 1170.43212890625,
      "learning_rate": 3.983877472150243e-05,
      "loss": 73.4186,
      "step": 34100
    },
    {
      "epoch": 0.1378087161689904,
      "grad_norm": 921.9115600585938,
      "learning_rate": 3.9838420663691e-05,
      "loss": 100.3894,
      "step": 34110
    },
    {
      "epoch": 0.13784911743435804,
      "grad_norm": 1105.57421875,
      "learning_rate": 3.9838066219119354e-05,
      "loss": 90.6614,
      "step": 34120
    },
    {
      "epoch": 0.13788951869972568,
      "grad_norm": 884.2128295898438,
      "learning_rate": 3.983771138779438e-05,
      "loss": 80.733,
      "step": 34130
    },
    {
      "epoch": 0.1379299199650933,
      "grad_norm": 4304.98828125,
      "learning_rate": 3.983735616972301e-05,
      "loss": 152.8355,
      "step": 34140
    },
    {
      "epoch": 0.13797032123046093,
      "grad_norm": 831.826904296875,
      "learning_rate": 3.983700056491218e-05,
      "loss": 66.2968,
      "step": 34150
    },
    {
      "epoch": 0.13801072249582857,
      "grad_norm": 691.4739990234375,
      "learning_rate": 3.9836644573368804e-05,
      "loss": 133.1252,
      "step": 34160
    },
    {
      "epoch": 0.1380511237611962,
      "grad_norm": 926.5187377929688,
      "learning_rate": 3.9836288195099834e-05,
      "loss": 113.0467,
      "step": 34170
    },
    {
      "epoch": 0.13809152502656383,
      "grad_norm": 808.3482055664062,
      "learning_rate": 3.983593143011221e-05,
      "loss": 97.4411,
      "step": 34180
    },
    {
      "epoch": 0.13813192629193147,
      "grad_norm": 1823.265869140625,
      "learning_rate": 3.983557427841289e-05,
      "loss": 136.7448,
      "step": 34190
    },
    {
      "epoch": 0.1381723275572991,
      "grad_norm": 1061.7825927734375,
      "learning_rate": 3.9835216740008835e-05,
      "loss": 81.9278,
      "step": 34200
    },
    {
      "epoch": 0.13821272882266672,
      "grad_norm": 1279.1239013671875,
      "learning_rate": 3.983485881490702e-05,
      "loss": 125.9145,
      "step": 34210
    },
    {
      "epoch": 0.13825313008803436,
      "grad_norm": 713.9881591796875,
      "learning_rate": 3.983450050311442e-05,
      "loss": 72.8081,
      "step": 34220
    },
    {
      "epoch": 0.138293531353402,
      "grad_norm": 902.9114990234375,
      "learning_rate": 3.9834141804638024e-05,
      "loss": 95.5154,
      "step": 34230
    },
    {
      "epoch": 0.1383339326187696,
      "grad_norm": 675.9998779296875,
      "learning_rate": 3.983378271948482e-05,
      "loss": 70.6422,
      "step": 34240
    },
    {
      "epoch": 0.13837433388413725,
      "grad_norm": 1091.7891845703125,
      "learning_rate": 3.983342324766181e-05,
      "loss": 88.8903,
      "step": 34250
    },
    {
      "epoch": 0.1384147351495049,
      "grad_norm": 647.6904907226562,
      "learning_rate": 3.9833063389176005e-05,
      "loss": 136.1893,
      "step": 34260
    },
    {
      "epoch": 0.1384551364148725,
      "grad_norm": 544.426025390625,
      "learning_rate": 3.983270314403442e-05,
      "loss": 87.8933,
      "step": 34270
    },
    {
      "epoch": 0.13849553768024014,
      "grad_norm": 845.7183227539062,
      "learning_rate": 3.983234251224407e-05,
      "loss": 89.7384,
      "step": 34280
    },
    {
      "epoch": 0.13853593894560778,
      "grad_norm": 0.0,
      "learning_rate": 3.9831981493812e-05,
      "loss": 63.4833,
      "step": 34290
    },
    {
      "epoch": 0.1385763402109754,
      "grad_norm": 1032.6282958984375,
      "learning_rate": 3.9831620088745236e-05,
      "loss": 120.7541,
      "step": 34300
    },
    {
      "epoch": 0.13861674147634304,
      "grad_norm": 1177.107421875,
      "learning_rate": 3.983125829705083e-05,
      "loss": 93.1693,
      "step": 34310
    },
    {
      "epoch": 0.13865714274171068,
      "grad_norm": 1217.580322265625,
      "learning_rate": 3.983089611873583e-05,
      "loss": 79.4667,
      "step": 34320
    },
    {
      "epoch": 0.1386975440070783,
      "grad_norm": 5233.41015625,
      "learning_rate": 3.9830533553807306e-05,
      "loss": 117.9349,
      "step": 34330
    },
    {
      "epoch": 0.13873794527244593,
      "grad_norm": 536.6962890625,
      "learning_rate": 3.9830170602272316e-05,
      "loss": 94.761,
      "step": 34340
    },
    {
      "epoch": 0.13877834653781357,
      "grad_norm": 567.7264404296875,
      "learning_rate": 3.982980726413795e-05,
      "loss": 79.9903,
      "step": 34350
    },
    {
      "epoch": 0.1388187478031812,
      "grad_norm": 514.4133911132812,
      "learning_rate": 3.9829443539411275e-05,
      "loss": 106.1553,
      "step": 34360
    },
    {
      "epoch": 0.13885914906854882,
      "grad_norm": 724.4990234375,
      "learning_rate": 3.9829079428099386e-05,
      "loss": 101.8466,
      "step": 34370
    },
    {
      "epoch": 0.13889955033391646,
      "grad_norm": 776.9714965820312,
      "learning_rate": 3.982871493020939e-05,
      "loss": 89.8777,
      "step": 34380
    },
    {
      "epoch": 0.1389399515992841,
      "grad_norm": 986.380126953125,
      "learning_rate": 3.9828350045748395e-05,
      "loss": 99.2931,
      "step": 34390
    },
    {
      "epoch": 0.1389803528646517,
      "grad_norm": 1103.84228515625,
      "learning_rate": 3.9827984774723495e-05,
      "loss": 84.1699,
      "step": 34400
    },
    {
      "epoch": 0.13902075413001935,
      "grad_norm": 1649.4866943359375,
      "learning_rate": 3.982761911714183e-05,
      "loss": 86.1248,
      "step": 34410
    },
    {
      "epoch": 0.139061155395387,
      "grad_norm": 1313.41015625,
      "learning_rate": 3.982725307301052e-05,
      "loss": 109.6458,
      "step": 34420
    },
    {
      "epoch": 0.1391015566607546,
      "grad_norm": 1034.6004638671875,
      "learning_rate": 3.98268866423367e-05,
      "loss": 95.0798,
      "step": 34430
    },
    {
      "epoch": 0.13914195792612225,
      "grad_norm": 1234.4898681640625,
      "learning_rate": 3.9826519825127524e-05,
      "loss": 119.6586,
      "step": 34440
    },
    {
      "epoch": 0.13918235919148988,
      "grad_norm": 1234.220947265625,
      "learning_rate": 3.9826152621390137e-05,
      "loss": 64.2341,
      "step": 34450
    },
    {
      "epoch": 0.1392227604568575,
      "grad_norm": 948.7850952148438,
      "learning_rate": 3.982578503113169e-05,
      "loss": 82.6322,
      "step": 34460
    },
    {
      "epoch": 0.13926316172222514,
      "grad_norm": 552.3663330078125,
      "learning_rate": 3.982541705435936e-05,
      "loss": 134.0263,
      "step": 34470
    },
    {
      "epoch": 0.13930356298759278,
      "grad_norm": 1042.3890380859375,
      "learning_rate": 3.982504869108033e-05,
      "loss": 77.6071,
      "step": 34480
    },
    {
      "epoch": 0.1393439642529604,
      "grad_norm": 883.0326538085938,
      "learning_rate": 3.982467994130175e-05,
      "loss": 84.257,
      "step": 34490
    },
    {
      "epoch": 0.13938436551832803,
      "grad_norm": 917.8054809570312,
      "learning_rate": 3.982431080503084e-05,
      "loss": 82.6933,
      "step": 34500
    },
    {
      "epoch": 0.13942476678369567,
      "grad_norm": 594.1129760742188,
      "learning_rate": 3.982394128227478e-05,
      "loss": 93.3724,
      "step": 34510
    },
    {
      "epoch": 0.1394651680490633,
      "grad_norm": 554.779541015625,
      "learning_rate": 3.982357137304078e-05,
      "loss": 118.062,
      "step": 34520
    },
    {
      "epoch": 0.13950556931443092,
      "grad_norm": 2090.916259765625,
      "learning_rate": 3.982320107733605e-05,
      "loss": 92.5961,
      "step": 34530
    },
    {
      "epoch": 0.13954597057979856,
      "grad_norm": 888.49951171875,
      "learning_rate": 3.982283039516781e-05,
      "loss": 88.7859,
      "step": 34540
    },
    {
      "epoch": 0.1395863718451662,
      "grad_norm": 580.6609497070312,
      "learning_rate": 3.982245932654328e-05,
      "loss": 127.9387,
      "step": 34550
    },
    {
      "epoch": 0.13962677311053381,
      "grad_norm": 1474.858154296875,
      "learning_rate": 3.982208787146971e-05,
      "loss": 109.5851,
      "step": 34560
    },
    {
      "epoch": 0.13966717437590145,
      "grad_norm": 432.7795104980469,
      "learning_rate": 3.9821716029954326e-05,
      "loss": 76.8793,
      "step": 34570
    },
    {
      "epoch": 0.1397075756412691,
      "grad_norm": 563.7490844726562,
      "learning_rate": 3.9821343802004386e-05,
      "loss": 71.9206,
      "step": 34580
    },
    {
      "epoch": 0.1397479769066367,
      "grad_norm": 649.1707763671875,
      "learning_rate": 3.982097118762714e-05,
      "loss": 88.6161,
      "step": 34590
    },
    {
      "epoch": 0.13978837817200435,
      "grad_norm": 858.9306030273438,
      "learning_rate": 3.982059818682986e-05,
      "loss": 102.4193,
      "step": 34600
    },
    {
      "epoch": 0.13982877943737199,
      "grad_norm": 1089.40673828125,
      "learning_rate": 3.982022479961981e-05,
      "loss": 74.4667,
      "step": 34610
    },
    {
      "epoch": 0.1398691807027396,
      "grad_norm": 1060.2073974609375,
      "learning_rate": 3.981985102600428e-05,
      "loss": 106.4542,
      "step": 34620
    },
    {
      "epoch": 0.13990958196810724,
      "grad_norm": 1195.9647216796875,
      "learning_rate": 3.981947686599054e-05,
      "loss": 58.7589,
      "step": 34630
    },
    {
      "epoch": 0.13994998323347488,
      "grad_norm": 1027.3399658203125,
      "learning_rate": 3.98191023195859e-05,
      "loss": 110.5383,
      "step": 34640
    },
    {
      "epoch": 0.1399903844988425,
      "grad_norm": 598.2379150390625,
      "learning_rate": 3.9818727386797656e-05,
      "loss": 58.4869,
      "step": 34650
    },
    {
      "epoch": 0.14003078576421013,
      "grad_norm": 811.7263793945312,
      "learning_rate": 3.981835206763312e-05,
      "loss": 112.3262,
      "step": 34660
    },
    {
      "epoch": 0.14007118702957777,
      "grad_norm": 686.4317016601562,
      "learning_rate": 3.981797636209961e-05,
      "loss": 99.8144,
      "step": 34670
    },
    {
      "epoch": 0.1401115882949454,
      "grad_norm": 1130.1160888671875,
      "learning_rate": 3.981760027020444e-05,
      "loss": 106.6619,
      "step": 34680
    },
    {
      "epoch": 0.14015198956031302,
      "grad_norm": 1811.7294921875,
      "learning_rate": 3.981722379195496e-05,
      "loss": 137.9515,
      "step": 34690
    },
    {
      "epoch": 0.14019239082568066,
      "grad_norm": 2056.862060546875,
      "learning_rate": 3.981684692735849e-05,
      "loss": 127.8664,
      "step": 34700
    },
    {
      "epoch": 0.1402327920910483,
      "grad_norm": 2816.014892578125,
      "learning_rate": 3.9816469676422395e-05,
      "loss": 112.5819,
      "step": 34710
    },
    {
      "epoch": 0.14027319335641592,
      "grad_norm": 587.19580078125,
      "learning_rate": 3.981609203915402e-05,
      "loss": 79.4191,
      "step": 34720
    },
    {
      "epoch": 0.14031359462178356,
      "grad_norm": 1270.5882568359375,
      "learning_rate": 3.9815714015560724e-05,
      "loss": 84.8151,
      "step": 34730
    },
    {
      "epoch": 0.1403539958871512,
      "grad_norm": 717.61767578125,
      "learning_rate": 3.981533560564988e-05,
      "loss": 101.6747,
      "step": 34740
    },
    {
      "epoch": 0.1403943971525188,
      "grad_norm": 1222.3599853515625,
      "learning_rate": 3.9814956809428876e-05,
      "loss": 117.1959,
      "step": 34750
    },
    {
      "epoch": 0.14043479841788645,
      "grad_norm": 1309.3331298828125,
      "learning_rate": 3.981457762690508e-05,
      "loss": 93.242,
      "step": 34760
    },
    {
      "epoch": 0.1404751996832541,
      "grad_norm": 981.174560546875,
      "learning_rate": 3.98141980580859e-05,
      "loss": 97.2162,
      "step": 34770
    },
    {
      "epoch": 0.1405156009486217,
      "grad_norm": 1109.3221435546875,
      "learning_rate": 3.981381810297872e-05,
      "loss": 99.3464,
      "step": 34780
    },
    {
      "epoch": 0.14055600221398934,
      "grad_norm": 1016.7904663085938,
      "learning_rate": 3.9813437761590954e-05,
      "loss": 100.2344,
      "step": 34790
    },
    {
      "epoch": 0.14059640347935698,
      "grad_norm": 461.59075927734375,
      "learning_rate": 3.981305703393002e-05,
      "loss": 77.6536,
      "step": 34800
    },
    {
      "epoch": 0.1406368047447246,
      "grad_norm": 806.1221923828125,
      "learning_rate": 3.981267592000334e-05,
      "loss": 84.1041,
      "step": 34810
    },
    {
      "epoch": 0.14067720601009223,
      "grad_norm": 1290.621337890625,
      "learning_rate": 3.981229441981834e-05,
      "loss": 115.2158,
      "step": 34820
    },
    {
      "epoch": 0.14071760727545987,
      "grad_norm": 1207.3558349609375,
      "learning_rate": 3.9811912533382465e-05,
      "loss": 94.0456,
      "step": 34830
    },
    {
      "epoch": 0.1407580085408275,
      "grad_norm": 618.889892578125,
      "learning_rate": 3.981153026070315e-05,
      "loss": 103.8425,
      "step": 34840
    },
    {
      "epoch": 0.14079840980619512,
      "grad_norm": 1754.47265625,
      "learning_rate": 3.981114760178786e-05,
      "loss": 101.4771,
      "step": 34850
    },
    {
      "epoch": 0.14083881107156276,
      "grad_norm": 1202.9140625,
      "learning_rate": 3.981076455664405e-05,
      "loss": 92.0938,
      "step": 34860
    },
    {
      "epoch": 0.1408792123369304,
      "grad_norm": 505.58685302734375,
      "learning_rate": 3.981038112527918e-05,
      "loss": 71.7284,
      "step": 34870
    },
    {
      "epoch": 0.14091961360229802,
      "grad_norm": 1087.3023681640625,
      "learning_rate": 3.980999730770072e-05,
      "loss": 72.5711,
      "step": 34880
    },
    {
      "epoch": 0.14096001486766566,
      "grad_norm": 437.0929260253906,
      "learning_rate": 3.9809613103916185e-05,
      "loss": 105.24,
      "step": 34890
    },
    {
      "epoch": 0.1410004161330333,
      "grad_norm": 569.6231689453125,
      "learning_rate": 3.980922851393303e-05,
      "loss": 88.6582,
      "step": 34900
    },
    {
      "epoch": 0.1410408173984009,
      "grad_norm": 1146.248779296875,
      "learning_rate": 3.980884353775877e-05,
      "loss": 104.9999,
      "step": 34910
    },
    {
      "epoch": 0.14108121866376855,
      "grad_norm": 3283.828857421875,
      "learning_rate": 3.980845817540091e-05,
      "loss": 80.5895,
      "step": 34920
    },
    {
      "epoch": 0.1411216199291362,
      "grad_norm": 629.7263793945312,
      "learning_rate": 3.980807242686695e-05,
      "loss": 58.2123,
      "step": 34930
    },
    {
      "epoch": 0.1411620211945038,
      "grad_norm": 594.664794921875,
      "learning_rate": 3.980768629216443e-05,
      "loss": 148.3967,
      "step": 34940
    },
    {
      "epoch": 0.14120242245987144,
      "grad_norm": 1153.6336669921875,
      "learning_rate": 3.980729977130086e-05,
      "loss": 120.5844,
      "step": 34950
    },
    {
      "epoch": 0.14124282372523908,
      "grad_norm": 2773.668701171875,
      "learning_rate": 3.9806912864283796e-05,
      "loss": 91.5852,
      "step": 34960
    },
    {
      "epoch": 0.1412832249906067,
      "grad_norm": 590.5272216796875,
      "learning_rate": 3.9806525571120754e-05,
      "loss": 91.9213,
      "step": 34970
    },
    {
      "epoch": 0.14132362625597433,
      "grad_norm": 951.1868286132812,
      "learning_rate": 3.9806137891819306e-05,
      "loss": 118.9212,
      "step": 34980
    },
    {
      "epoch": 0.14136402752134197,
      "grad_norm": 761.1874389648438,
      "learning_rate": 3.9805749826387005e-05,
      "loss": 89.8664,
      "step": 34990
    },
    {
      "epoch": 0.1414044287867096,
      "grad_norm": 1326.7030029296875,
      "learning_rate": 3.980536137483141e-05,
      "loss": 220.9611,
      "step": 35000
    },
    {
      "epoch": 0.14144483005207723,
      "grad_norm": 385.3703918457031,
      "learning_rate": 3.98049725371601e-05,
      "loss": 79.5324,
      "step": 35010
    },
    {
      "epoch": 0.14148523131744487,
      "grad_norm": 1560.7586669921875,
      "learning_rate": 3.980458331338065e-05,
      "loss": 100.8703,
      "step": 35020
    },
    {
      "epoch": 0.1415256325828125,
      "grad_norm": 949.5518188476562,
      "learning_rate": 3.980419370350066e-05,
      "loss": 106.5335,
      "step": 35030
    },
    {
      "epoch": 0.14156603384818012,
      "grad_norm": 1331.5462646484375,
      "learning_rate": 3.980380370752771e-05,
      "loss": 92.9983,
      "step": 35040
    },
    {
      "epoch": 0.14160643511354776,
      "grad_norm": 1191.44140625,
      "learning_rate": 3.980341332546942e-05,
      "loss": 92.3872,
      "step": 35050
    },
    {
      "epoch": 0.1416468363789154,
      "grad_norm": 1959.1849365234375,
      "learning_rate": 3.9803022557333387e-05,
      "loss": 102.082,
      "step": 35060
    },
    {
      "epoch": 0.141687237644283,
      "grad_norm": 559.7910766601562,
      "learning_rate": 3.9802631403127234e-05,
      "loss": 74.2511,
      "step": 35070
    },
    {
      "epoch": 0.14172763890965065,
      "grad_norm": 1271.74169921875,
      "learning_rate": 3.980223986285859e-05,
      "loss": 138.9416,
      "step": 35080
    },
    {
      "epoch": 0.1417680401750183,
      "grad_norm": 900.1123046875,
      "learning_rate": 3.980184793653508e-05,
      "loss": 78.2254,
      "step": 35090
    },
    {
      "epoch": 0.1418084414403859,
      "grad_norm": 0.0,
      "learning_rate": 3.9801455624164354e-05,
      "loss": 103.8371,
      "step": 35100
    },
    {
      "epoch": 0.14184884270575354,
      "grad_norm": 4405.7216796875,
      "learning_rate": 3.980106292575405e-05,
      "loss": 95.1343,
      "step": 35110
    },
    {
      "epoch": 0.14188924397112118,
      "grad_norm": 2287.119384765625,
      "learning_rate": 3.980066984131184e-05,
      "loss": 156.3899,
      "step": 35120
    },
    {
      "epoch": 0.1419296452364888,
      "grad_norm": 915.5543823242188,
      "learning_rate": 3.980027637084537e-05,
      "loss": 110.9924,
      "step": 35130
    },
    {
      "epoch": 0.14197004650185643,
      "grad_norm": 299.7638244628906,
      "learning_rate": 3.979988251436232e-05,
      "loss": 101.3539,
      "step": 35140
    },
    {
      "epoch": 0.14201044776722407,
      "grad_norm": 467.0106506347656,
      "learning_rate": 3.979948827187036e-05,
      "loss": 89.7661,
      "step": 35150
    },
    {
      "epoch": 0.14205084903259171,
      "grad_norm": 916.4078369140625,
      "learning_rate": 3.9799093643377195e-05,
      "loss": 102.0699,
      "step": 35160
    },
    {
      "epoch": 0.14209125029795933,
      "grad_norm": 946.13232421875,
      "learning_rate": 3.97986986288905e-05,
      "loss": 108.2325,
      "step": 35170
    },
    {
      "epoch": 0.14213165156332697,
      "grad_norm": 1187.6005859375,
      "learning_rate": 3.9798303228417975e-05,
      "loss": 77.135,
      "step": 35180
    },
    {
      "epoch": 0.1421720528286946,
      "grad_norm": 761.8079833984375,
      "learning_rate": 3.9797907441967345e-05,
      "loss": 191.0544,
      "step": 35190
    },
    {
      "epoch": 0.14221245409406222,
      "grad_norm": 527.9906005859375,
      "learning_rate": 3.979751126954632e-05,
      "loss": 84.1937,
      "step": 35200
    },
    {
      "epoch": 0.14225285535942986,
      "grad_norm": 433.1168212890625,
      "learning_rate": 3.979711471116261e-05,
      "loss": 63.9944,
      "step": 35210
    },
    {
      "epoch": 0.1422932566247975,
      "grad_norm": 771.2758178710938,
      "learning_rate": 3.9796717766823966e-05,
      "loss": 102.1717,
      "step": 35220
    },
    {
      "epoch": 0.1423336578901651,
      "grad_norm": 1156.9407958984375,
      "learning_rate": 3.979632043653811e-05,
      "loss": 82.1883,
      "step": 35230
    },
    {
      "epoch": 0.14237405915553275,
      "grad_norm": 1199.1309814453125,
      "learning_rate": 3.9795922720312796e-05,
      "loss": 99.0348,
      "step": 35240
    },
    {
      "epoch": 0.1424144604209004,
      "grad_norm": 2005.9031982421875,
      "learning_rate": 3.979552461815578e-05,
      "loss": 94.9322,
      "step": 35250
    },
    {
      "epoch": 0.142454861686268,
      "grad_norm": 781.2155151367188,
      "learning_rate": 3.9795126130074824e-05,
      "loss": 80.2976,
      "step": 35260
    },
    {
      "epoch": 0.14249526295163564,
      "grad_norm": 900.551025390625,
      "learning_rate": 3.9794727256077685e-05,
      "loss": 111.1041,
      "step": 35270
    },
    {
      "epoch": 0.14253566421700328,
      "grad_norm": 2165.27294921875,
      "learning_rate": 3.979432799617215e-05,
      "loss": 122.0848,
      "step": 35280
    },
    {
      "epoch": 0.1425760654823709,
      "grad_norm": 2192.3740234375,
      "learning_rate": 3.9793928350366e-05,
      "loss": 77.1153,
      "step": 35290
    },
    {
      "epoch": 0.14261646674773854,
      "grad_norm": 1193.210205078125,
      "learning_rate": 3.979352831866702e-05,
      "loss": 63.2352,
      "step": 35300
    },
    {
      "epoch": 0.14265686801310618,
      "grad_norm": 679.7105712890625,
      "learning_rate": 3.979312790108303e-05,
      "loss": 108.7417,
      "step": 35310
    },
    {
      "epoch": 0.14269726927847382,
      "grad_norm": 3321.63671875,
      "learning_rate": 3.979272709762181e-05,
      "loss": 98.3625,
      "step": 35320
    },
    {
      "epoch": 0.14273767054384143,
      "grad_norm": 520.6390380859375,
      "learning_rate": 3.979232590829119e-05,
      "loss": 88.9054,
      "step": 35330
    },
    {
      "epoch": 0.14277807180920907,
      "grad_norm": 755.5736083984375,
      "learning_rate": 3.9791924333098986e-05,
      "loss": 118.9642,
      "step": 35340
    },
    {
      "epoch": 0.1428184730745767,
      "grad_norm": 1435.6326904296875,
      "learning_rate": 3.979152237205302e-05,
      "loss": 60.0594,
      "step": 35350
    },
    {
      "epoch": 0.14285887433994432,
      "grad_norm": 687.8348388671875,
      "learning_rate": 3.979112002516115e-05,
      "loss": 99.3451,
      "step": 35360
    },
    {
      "epoch": 0.14289927560531196,
      "grad_norm": 3126.065673828125,
      "learning_rate": 3.979071729243119e-05,
      "loss": 94.5208,
      "step": 35370
    },
    {
      "epoch": 0.1429396768706796,
      "grad_norm": 627.322265625,
      "learning_rate": 3.979031417387102e-05,
      "loss": 96.7639,
      "step": 35380
    },
    {
      "epoch": 0.1429800781360472,
      "grad_norm": 1383.9814453125,
      "learning_rate": 3.978991066948848e-05,
      "loss": 121.3856,
      "step": 35390
    },
    {
      "epoch": 0.14302047940141485,
      "grad_norm": 1195.60498046875,
      "learning_rate": 3.9789506779291445e-05,
      "loss": 100.726,
      "step": 35400
    },
    {
      "epoch": 0.1430608806667825,
      "grad_norm": 1437.9139404296875,
      "learning_rate": 3.978910250328779e-05,
      "loss": 97.5714,
      "step": 35410
    },
    {
      "epoch": 0.1431012819321501,
      "grad_norm": 1709.1278076171875,
      "learning_rate": 3.978869784148539e-05,
      "loss": 90.859,
      "step": 35420
    },
    {
      "epoch": 0.14314168319751774,
      "grad_norm": 724.835693359375,
      "learning_rate": 3.978829279389214e-05,
      "loss": 110.8027,
      "step": 35430
    },
    {
      "epoch": 0.14318208446288538,
      "grad_norm": 3041.685791015625,
      "learning_rate": 3.978788736051593e-05,
      "loss": 77.8094,
      "step": 35440
    },
    {
      "epoch": 0.143222485728253,
      "grad_norm": 1632.168701171875,
      "learning_rate": 3.978748154136467e-05,
      "loss": 101.8852,
      "step": 35450
    },
    {
      "epoch": 0.14326288699362064,
      "grad_norm": 0.0,
      "learning_rate": 3.978707533644627e-05,
      "loss": 79.8863,
      "step": 35460
    },
    {
      "epoch": 0.14330328825898828,
      "grad_norm": 1186.8582763671875,
      "learning_rate": 3.978666874576865e-05,
      "loss": 83.3348,
      "step": 35470
    },
    {
      "epoch": 0.14334368952435592,
      "grad_norm": 1080.42919921875,
      "learning_rate": 3.978626176933974e-05,
      "loss": 82.4089,
      "step": 35480
    },
    {
      "epoch": 0.14338409078972353,
      "grad_norm": 658.2520751953125,
      "learning_rate": 3.978585440716746e-05,
      "loss": 102.263,
      "step": 35490
    },
    {
      "epoch": 0.14342449205509117,
      "grad_norm": 924.6647338867188,
      "learning_rate": 3.978544665925977e-05,
      "loss": 90.3248,
      "step": 35500
    },
    {
      "epoch": 0.1434648933204588,
      "grad_norm": 1277.648681640625,
      "learning_rate": 3.978503852562461e-05,
      "loss": 60.2407,
      "step": 35510
    },
    {
      "epoch": 0.14350529458582642,
      "grad_norm": 1398.44775390625,
      "learning_rate": 3.9784630006269935e-05,
      "loss": 91.2831,
      "step": 35520
    },
    {
      "epoch": 0.14354569585119406,
      "grad_norm": 637.0263671875,
      "learning_rate": 3.9784221101203715e-05,
      "loss": 85.0435,
      "step": 35530
    },
    {
      "epoch": 0.1435860971165617,
      "grad_norm": 2696.323486328125,
      "learning_rate": 3.978381181043392e-05,
      "loss": 95.1324,
      "step": 35540
    },
    {
      "epoch": 0.1436264983819293,
      "grad_norm": 914.337890625,
      "learning_rate": 3.978340213396853e-05,
      "loss": 89.1903,
      "step": 35550
    },
    {
      "epoch": 0.14366689964729695,
      "grad_norm": 569.9886474609375,
      "learning_rate": 3.978299207181552e-05,
      "loss": 146.3416,
      "step": 35560
    },
    {
      "epoch": 0.1437073009126646,
      "grad_norm": 2463.093017578125,
      "learning_rate": 3.97825816239829e-05,
      "loss": 85.1585,
      "step": 35570
    },
    {
      "epoch": 0.1437477021780322,
      "grad_norm": 888.9117431640625,
      "learning_rate": 3.978217079047867e-05,
      "loss": 77.3447,
      "step": 35580
    },
    {
      "epoch": 0.14378810344339985,
      "grad_norm": 860.64013671875,
      "learning_rate": 3.9781759571310835e-05,
      "loss": 87.8478,
      "step": 35590
    },
    {
      "epoch": 0.14382850470876749,
      "grad_norm": 1036.18994140625,
      "learning_rate": 3.9781347966487415e-05,
      "loss": 111.4597,
      "step": 35600
    },
    {
      "epoch": 0.1438689059741351,
      "grad_norm": 547.6618041992188,
      "learning_rate": 3.978093597601643e-05,
      "loss": 76.2303,
      "step": 35610
    },
    {
      "epoch": 0.14390930723950274,
      "grad_norm": 1933.931640625,
      "learning_rate": 3.978052359990591e-05,
      "loss": 89.3041,
      "step": 35620
    },
    {
      "epoch": 0.14394970850487038,
      "grad_norm": 360.5538330078125,
      "learning_rate": 3.97801108381639e-05,
      "loss": 85.9847,
      "step": 35630
    },
    {
      "epoch": 0.14399010977023802,
      "grad_norm": 1371.674560546875,
      "learning_rate": 3.977969769079845e-05,
      "loss": 92.9651,
      "step": 35640
    },
    {
      "epoch": 0.14403051103560563,
      "grad_norm": 3026.696044921875,
      "learning_rate": 3.977928415781761e-05,
      "loss": 86.0992,
      "step": 35650
    },
    {
      "epoch": 0.14407091230097327,
      "grad_norm": 1090.5291748046875,
      "learning_rate": 3.977887023922944e-05,
      "loss": 98.6011,
      "step": 35660
    },
    {
      "epoch": 0.1441113135663409,
      "grad_norm": 886.7594604492188,
      "learning_rate": 3.977845593504201e-05,
      "loss": 128.8346,
      "step": 35670
    },
    {
      "epoch": 0.14415171483170852,
      "grad_norm": 372.66937255859375,
      "learning_rate": 3.97780412452634e-05,
      "loss": 80.6952,
      "step": 35680
    },
    {
      "epoch": 0.14419211609707616,
      "grad_norm": 569.3165283203125,
      "learning_rate": 3.97776261699017e-05,
      "loss": 82.1124,
      "step": 35690
    },
    {
      "epoch": 0.1442325173624438,
      "grad_norm": 495.3749694824219,
      "learning_rate": 3.977721070896499e-05,
      "loss": 63.9636,
      "step": 35700
    },
    {
      "epoch": 0.14427291862781141,
      "grad_norm": 1659.83544921875,
      "learning_rate": 3.977679486246137e-05,
      "loss": 150.3122,
      "step": 35710
    },
    {
      "epoch": 0.14431331989317905,
      "grad_norm": 933.7442626953125,
      "learning_rate": 3.977637863039896e-05,
      "loss": 85.469,
      "step": 35720
    },
    {
      "epoch": 0.1443537211585467,
      "grad_norm": 2418.2099609375,
      "learning_rate": 3.9775962012785863e-05,
      "loss": 76.3617,
      "step": 35730
    },
    {
      "epoch": 0.1443941224239143,
      "grad_norm": 908.3840942382812,
      "learning_rate": 3.9775545009630204e-05,
      "loss": 98.9815,
      "step": 35740
    },
    {
      "epoch": 0.14443452368928195,
      "grad_norm": 593.0587158203125,
      "learning_rate": 3.977512762094012e-05,
      "loss": 106.191,
      "step": 35750
    },
    {
      "epoch": 0.1444749249546496,
      "grad_norm": 1349.186279296875,
      "learning_rate": 3.977470984672374e-05,
      "loss": 124.571,
      "step": 35760
    },
    {
      "epoch": 0.1445153262200172,
      "grad_norm": 399.5314025878906,
      "learning_rate": 3.977429168698921e-05,
      "loss": 107.339,
      "step": 35770
    },
    {
      "epoch": 0.14455572748538484,
      "grad_norm": 867.3482055664062,
      "learning_rate": 3.9773873141744684e-05,
      "loss": 123.2388,
      "step": 35780
    },
    {
      "epoch": 0.14459612875075248,
      "grad_norm": 377.1903991699219,
      "learning_rate": 3.9773454210998315e-05,
      "loss": 116.9524,
      "step": 35790
    },
    {
      "epoch": 0.14463653001612012,
      "grad_norm": 851.285888671875,
      "learning_rate": 3.977303489475828e-05,
      "loss": 108.5521,
      "step": 35800
    },
    {
      "epoch": 0.14467693128148773,
      "grad_norm": 1076.890869140625,
      "learning_rate": 3.977261519303275e-05,
      "loss": 128.3607,
      "step": 35810
    },
    {
      "epoch": 0.14471733254685537,
      "grad_norm": 412.141845703125,
      "learning_rate": 3.9772195105829914e-05,
      "loss": 90.6862,
      "step": 35820
    },
    {
      "epoch": 0.144757733812223,
      "grad_norm": 959.1841430664062,
      "learning_rate": 3.977177463315795e-05,
      "loss": 77.2501,
      "step": 35830
    },
    {
      "epoch": 0.14479813507759062,
      "grad_norm": 839.197509765625,
      "learning_rate": 3.977135377502506e-05,
      "loss": 81.871,
      "step": 35840
    },
    {
      "epoch": 0.14483853634295826,
      "grad_norm": 977.5892944335938,
      "learning_rate": 3.9770932531439445e-05,
      "loss": 97.2153,
      "step": 35850
    },
    {
      "epoch": 0.1448789376083259,
      "grad_norm": 3265.607177734375,
      "learning_rate": 3.977051090240933e-05,
      "loss": 85.7729,
      "step": 35860
    },
    {
      "epoch": 0.14491933887369352,
      "grad_norm": 535.1461791992188,
      "learning_rate": 3.977008888794291e-05,
      "loss": 81.5642,
      "step": 35870
    },
    {
      "epoch": 0.14495974013906116,
      "grad_norm": 624.2498779296875,
      "learning_rate": 3.976966648804845e-05,
      "loss": 149.5731,
      "step": 35880
    },
    {
      "epoch": 0.1450001414044288,
      "grad_norm": 1168.02734375,
      "learning_rate": 3.976924370273415e-05,
      "loss": 98.4266,
      "step": 35890
    },
    {
      "epoch": 0.1450405426697964,
      "grad_norm": 1551.1861572265625,
      "learning_rate": 3.976882053200827e-05,
      "loss": 79.1619,
      "step": 35900
    },
    {
      "epoch": 0.14508094393516405,
      "grad_norm": 634.694580078125,
      "learning_rate": 3.976839697587906e-05,
      "loss": 67.8042,
      "step": 35910
    },
    {
      "epoch": 0.1451213452005317,
      "grad_norm": 602.9379272460938,
      "learning_rate": 3.9767973034354764e-05,
      "loss": 83.8792,
      "step": 35920
    },
    {
      "epoch": 0.1451617464658993,
      "grad_norm": 2905.12451171875,
      "learning_rate": 3.9767548707443665e-05,
      "loss": 120.9742,
      "step": 35930
    },
    {
      "epoch": 0.14520214773126694,
      "grad_norm": 910.666015625,
      "learning_rate": 3.976712399515402e-05,
      "loss": 111.3549,
      "step": 35940
    },
    {
      "epoch": 0.14524254899663458,
      "grad_norm": 1786.91796875,
      "learning_rate": 3.976669889749412e-05,
      "loss": 72.8603,
      "step": 35950
    },
    {
      "epoch": 0.14528295026200222,
      "grad_norm": 1702.6033935546875,
      "learning_rate": 3.9766273414472254e-05,
      "loss": 72.2166,
      "step": 35960
    },
    {
      "epoch": 0.14532335152736983,
      "grad_norm": 1203.12548828125,
      "learning_rate": 3.97658475460967e-05,
      "loss": 85.1111,
      "step": 35970
    },
    {
      "epoch": 0.14536375279273747,
      "grad_norm": 1354.2686767578125,
      "learning_rate": 3.976542129237578e-05,
      "loss": 82.6227,
      "step": 35980
    },
    {
      "epoch": 0.1454041540581051,
      "grad_norm": 646.3790893554688,
      "learning_rate": 3.976499465331779e-05,
      "loss": 83.9092,
      "step": 35990
    },
    {
      "epoch": 0.14544455532347272,
      "grad_norm": 1774.5269775390625,
      "learning_rate": 3.976456762893106e-05,
      "loss": 147.7749,
      "step": 36000
    },
    {
      "epoch": 0.14548495658884036,
      "grad_norm": 484.8998718261719,
      "learning_rate": 3.976414021922391e-05,
      "loss": 69.3422,
      "step": 36010
    },
    {
      "epoch": 0.145525357854208,
      "grad_norm": 803.2476806640625,
      "learning_rate": 3.9763712424204664e-05,
      "loss": 81.9522,
      "step": 36020
    },
    {
      "epoch": 0.14556575911957562,
      "grad_norm": 390.46014404296875,
      "learning_rate": 3.976328424388167e-05,
      "loss": 58.7185,
      "step": 36030
    },
    {
      "epoch": 0.14560616038494326,
      "grad_norm": 552.03076171875,
      "learning_rate": 3.9762855678263274e-05,
      "loss": 125.9961,
      "step": 36040
    },
    {
      "epoch": 0.1456465616503109,
      "grad_norm": 849.45361328125,
      "learning_rate": 3.976242672735784e-05,
      "loss": 96.4432,
      "step": 36050
    },
    {
      "epoch": 0.1456869629156785,
      "grad_norm": 995.2199096679688,
      "learning_rate": 3.9761997391173715e-05,
      "loss": 121.193,
      "step": 36060
    },
    {
      "epoch": 0.14572736418104615,
      "grad_norm": 645.1380004882812,
      "learning_rate": 3.976156766971928e-05,
      "loss": 99.5118,
      "step": 36070
    },
    {
      "epoch": 0.1457677654464138,
      "grad_norm": 494.75238037109375,
      "learning_rate": 3.976113756300291e-05,
      "loss": 77.7827,
      "step": 36080
    },
    {
      "epoch": 0.1458081667117814,
      "grad_norm": 616.9432983398438,
      "learning_rate": 3.976070707103299e-05,
      "loss": 99.2099,
      "step": 36090
    },
    {
      "epoch": 0.14584856797714904,
      "grad_norm": 894.1851196289062,
      "learning_rate": 3.976027619381791e-05,
      "loss": 80.5238,
      "step": 36100
    },
    {
      "epoch": 0.14588896924251668,
      "grad_norm": 1238.346923828125,
      "learning_rate": 3.975984493136607e-05,
      "loss": 75.3991,
      "step": 36110
    },
    {
      "epoch": 0.14592937050788432,
      "grad_norm": 1124.619873046875,
      "learning_rate": 3.975941328368588e-05,
      "loss": 139.1967,
      "step": 36120
    },
    {
      "epoch": 0.14596977177325193,
      "grad_norm": 0.0,
      "learning_rate": 3.975898125078576e-05,
      "loss": 86.8455,
      "step": 36130
    },
    {
      "epoch": 0.14601017303861957,
      "grad_norm": 999.146728515625,
      "learning_rate": 3.9758548832674126e-05,
      "loss": 98.2229,
      "step": 36140
    },
    {
      "epoch": 0.1460505743039872,
      "grad_norm": 921.8599243164062,
      "learning_rate": 3.975811602935941e-05,
      "loss": 65.2062,
      "step": 36150
    },
    {
      "epoch": 0.14609097556935483,
      "grad_norm": 1060.4775390625,
      "learning_rate": 3.9757682840850045e-05,
      "loss": 73.0853,
      "step": 36160
    },
    {
      "epoch": 0.14613137683472247,
      "grad_norm": 1437.3382568359375,
      "learning_rate": 3.975724926715449e-05,
      "loss": 103.5526,
      "step": 36170
    },
    {
      "epoch": 0.1461717781000901,
      "grad_norm": 1258.572509765625,
      "learning_rate": 3.9756815308281185e-05,
      "loss": 97.9677,
      "step": 36180
    },
    {
      "epoch": 0.14621217936545772,
      "grad_norm": 767.1698608398438,
      "learning_rate": 3.975638096423859e-05,
      "loss": 148.626,
      "step": 36190
    },
    {
      "epoch": 0.14625258063082536,
      "grad_norm": 1137.6505126953125,
      "learning_rate": 3.975594623503518e-05,
      "loss": 72.6446,
      "step": 36200
    },
    {
      "epoch": 0.146292981896193,
      "grad_norm": 733.9754638671875,
      "learning_rate": 3.975551112067943e-05,
      "loss": 76.0892,
      "step": 36210
    },
    {
      "epoch": 0.1463333831615606,
      "grad_norm": 0.0,
      "learning_rate": 3.975507562117982e-05,
      "loss": 79.5784,
      "step": 36220
    },
    {
      "epoch": 0.14637378442692825,
      "grad_norm": 1067.524169921875,
      "learning_rate": 3.975463973654483e-05,
      "loss": 95.0418,
      "step": 36230
    },
    {
      "epoch": 0.1464141856922959,
      "grad_norm": 1821.6085205078125,
      "learning_rate": 3.975420346678298e-05,
      "loss": 85.3464,
      "step": 36240
    },
    {
      "epoch": 0.1464545869576635,
      "grad_norm": 477.71600341796875,
      "learning_rate": 3.9753766811902756e-05,
      "loss": 79.2371,
      "step": 36250
    },
    {
      "epoch": 0.14649498822303114,
      "grad_norm": 668.6638793945312,
      "learning_rate": 3.975332977191268e-05,
      "loss": 92.1396,
      "step": 36260
    },
    {
      "epoch": 0.14653538948839878,
      "grad_norm": 0.0,
      "learning_rate": 3.9752892346821274e-05,
      "loss": 166.9924,
      "step": 36270
    },
    {
      "epoch": 0.14657579075376642,
      "grad_norm": 1038.5552978515625,
      "learning_rate": 3.975245453663706e-05,
      "loss": 95.1147,
      "step": 36280
    },
    {
      "epoch": 0.14661619201913403,
      "grad_norm": 1050.910400390625,
      "learning_rate": 3.9752016341368574e-05,
      "loss": 92.2121,
      "step": 36290
    },
    {
      "epoch": 0.14665659328450167,
      "grad_norm": 1757.861572265625,
      "learning_rate": 3.9751577761024366e-05,
      "loss": 72.9227,
      "step": 36300
    },
    {
      "epoch": 0.14669699454986931,
      "grad_norm": 1091.1439208984375,
      "learning_rate": 3.9751138795612976e-05,
      "loss": 114.0233,
      "step": 36310
    },
    {
      "epoch": 0.14673739581523693,
      "grad_norm": 253.02093505859375,
      "learning_rate": 3.975069944514296e-05,
      "loss": 84.1495,
      "step": 36320
    },
    {
      "epoch": 0.14677779708060457,
      "grad_norm": 1279.52490234375,
      "learning_rate": 3.97502597096229e-05,
      "loss": 110.2362,
      "step": 36330
    },
    {
      "epoch": 0.1468181983459722,
      "grad_norm": 856.653076171875,
      "learning_rate": 3.9749819589061354e-05,
      "loss": 75.3621,
      "step": 36340
    },
    {
      "epoch": 0.14685859961133982,
      "grad_norm": 726.4605712890625,
      "learning_rate": 3.974937908346691e-05,
      "loss": 84.8673,
      "step": 36350
    },
    {
      "epoch": 0.14689900087670746,
      "grad_norm": 1209.525634765625,
      "learning_rate": 3.974893819284815e-05,
      "loss": 77.0584,
      "step": 36360
    },
    {
      "epoch": 0.1469394021420751,
      "grad_norm": 1527.44580078125,
      "learning_rate": 3.9748496917213675e-05,
      "loss": 119.0866,
      "step": 36370
    },
    {
      "epoch": 0.1469798034074427,
      "grad_norm": 665.8195190429688,
      "learning_rate": 3.9748055256572084e-05,
      "loss": 76.9333,
      "step": 36380
    },
    {
      "epoch": 0.14702020467281035,
      "grad_norm": 408.0297546386719,
      "learning_rate": 3.974761321093199e-05,
      "loss": 99.8797,
      "step": 36390
    },
    {
      "epoch": 0.147060605938178,
      "grad_norm": 617.4054565429688,
      "learning_rate": 3.974717078030201e-05,
      "loss": 71.3116,
      "step": 36400
    },
    {
      "epoch": 0.1471010072035456,
      "grad_norm": 715.342529296875,
      "learning_rate": 3.974672796469077e-05,
      "loss": 107.3487,
      "step": 36410
    },
    {
      "epoch": 0.14714140846891324,
      "grad_norm": 639.6741943359375,
      "learning_rate": 3.97462847641069e-05,
      "loss": 65.2551,
      "step": 36420
    },
    {
      "epoch": 0.14718180973428088,
      "grad_norm": 0.0,
      "learning_rate": 3.974584117855904e-05,
      "loss": 106.0524,
      "step": 36430
    },
    {
      "epoch": 0.14722221099964852,
      "grad_norm": 1352.6854248046875,
      "learning_rate": 3.974539720805585e-05,
      "loss": 86.2377,
      "step": 36440
    },
    {
      "epoch": 0.14726261226501614,
      "grad_norm": 1535.055419921875,
      "learning_rate": 3.9744952852605965e-05,
      "loss": 90.6992,
      "step": 36450
    },
    {
      "epoch": 0.14730301353038378,
      "grad_norm": 986.7384643554688,
      "learning_rate": 3.974450811221806e-05,
      "loss": 92.4026,
      "step": 36460
    },
    {
      "epoch": 0.14734341479575142,
      "grad_norm": 887.9044799804688,
      "learning_rate": 3.974406298690081e-05,
      "loss": 74.2204,
      "step": 36470
    },
    {
      "epoch": 0.14738381606111903,
      "grad_norm": 392.92437744140625,
      "learning_rate": 3.974361747666288e-05,
      "loss": 79.7651,
      "step": 36480
    },
    {
      "epoch": 0.14742421732648667,
      "grad_norm": 1977.6578369140625,
      "learning_rate": 3.974317158151297e-05,
      "loss": 88.8437,
      "step": 36490
    },
    {
      "epoch": 0.1474646185918543,
      "grad_norm": 650.1098022460938,
      "learning_rate": 3.974272530145976e-05,
      "loss": 64.0652,
      "step": 36500
    },
    {
      "epoch": 0.14750501985722192,
      "grad_norm": 555.5321044921875,
      "learning_rate": 3.974227863651196e-05,
      "loss": 95.1656,
      "step": 36510
    },
    {
      "epoch": 0.14754542112258956,
      "grad_norm": 1186.1986083984375,
      "learning_rate": 3.974183158667827e-05,
      "loss": 104.5347,
      "step": 36520
    },
    {
      "epoch": 0.1475858223879572,
      "grad_norm": 804.430908203125,
      "learning_rate": 3.9741384151967416e-05,
      "loss": 78.5228,
      "step": 36530
    },
    {
      "epoch": 0.1476262236533248,
      "grad_norm": 724.3908081054688,
      "learning_rate": 3.974093633238811e-05,
      "loss": 95.4657,
      "step": 36540
    },
    {
      "epoch": 0.14766662491869245,
      "grad_norm": 657.1284790039062,
      "learning_rate": 3.974048812794908e-05,
      "loss": 95.8921,
      "step": 36550
    },
    {
      "epoch": 0.1477070261840601,
      "grad_norm": 813.1570434570312,
      "learning_rate": 3.974003953865908e-05,
      "loss": 87.4508,
      "step": 36560
    },
    {
      "epoch": 0.1477474274494277,
      "grad_norm": 1045.587890625,
      "learning_rate": 3.973959056452685e-05,
      "loss": 116.7263,
      "step": 36570
    },
    {
      "epoch": 0.14778782871479534,
      "grad_norm": 680.8571166992188,
      "learning_rate": 3.9739141205561126e-05,
      "loss": 117.5743,
      "step": 36580
    },
    {
      "epoch": 0.14782822998016298,
      "grad_norm": 584.37158203125,
      "learning_rate": 3.973869146177069e-05,
      "loss": 79.1421,
      "step": 36590
    },
    {
      "epoch": 0.1478686312455306,
      "grad_norm": 1341.7860107421875,
      "learning_rate": 3.973824133316431e-05,
      "loss": 97.5215,
      "step": 36600
    },
    {
      "epoch": 0.14790903251089824,
      "grad_norm": 829.0403442382812,
      "learning_rate": 3.973779081975074e-05,
      "loss": 109.4106,
      "step": 36610
    },
    {
      "epoch": 0.14794943377626588,
      "grad_norm": 435.8904724121094,
      "learning_rate": 3.9737339921538775e-05,
      "loss": 94.5021,
      "step": 36620
    },
    {
      "epoch": 0.14798983504163352,
      "grad_norm": 1048.1324462890625,
      "learning_rate": 3.973688863853721e-05,
      "loss": 105.4422,
      "step": 36630
    },
    {
      "epoch": 0.14803023630700113,
      "grad_norm": 1104.1217041015625,
      "learning_rate": 3.973643697075485e-05,
      "loss": 86.1049,
      "step": 36640
    },
    {
      "epoch": 0.14807063757236877,
      "grad_norm": 2705.364501953125,
      "learning_rate": 3.973598491820048e-05,
      "loss": 125.3173,
      "step": 36650
    },
    {
      "epoch": 0.1481110388377364,
      "grad_norm": 1945.0257568359375,
      "learning_rate": 3.9735532480882925e-05,
      "loss": 110.4482,
      "step": 36660
    },
    {
      "epoch": 0.14815144010310402,
      "grad_norm": 1239.58056640625,
      "learning_rate": 3.9735079658811006e-05,
      "loss": 102.1776,
      "step": 36670
    },
    {
      "epoch": 0.14819184136847166,
      "grad_norm": 751.5692138671875,
      "learning_rate": 3.973462645199355e-05,
      "loss": 88.416,
      "step": 36680
    },
    {
      "epoch": 0.1482322426338393,
      "grad_norm": 738.1787719726562,
      "learning_rate": 3.9734172860439385e-05,
      "loss": 72.5949,
      "step": 36690
    },
    {
      "epoch": 0.1482726438992069,
      "grad_norm": 442.69232177734375,
      "learning_rate": 3.973371888415736e-05,
      "loss": 91.564,
      "step": 36700
    },
    {
      "epoch": 0.14831304516457455,
      "grad_norm": 2007.3953857421875,
      "learning_rate": 3.9733264523156337e-05,
      "loss": 136.281,
      "step": 36710
    },
    {
      "epoch": 0.1483534464299422,
      "grad_norm": 453.4363708496094,
      "learning_rate": 3.973280977744515e-05,
      "loss": 111.2205,
      "step": 36720
    },
    {
      "epoch": 0.1483938476953098,
      "grad_norm": 912.5247192382812,
      "learning_rate": 3.9732354647032687e-05,
      "loss": 86.6152,
      "step": 36730
    },
    {
      "epoch": 0.14843424896067745,
      "grad_norm": 412.4578857421875,
      "learning_rate": 3.97318991319278e-05,
      "loss": 80.3192,
      "step": 36740
    },
    {
      "epoch": 0.14847465022604509,
      "grad_norm": 879.3831176757812,
      "learning_rate": 3.973144323213939e-05,
      "loss": 64.8996,
      "step": 36750
    },
    {
      "epoch": 0.1485150514914127,
      "grad_norm": 697.5357666015625,
      "learning_rate": 3.9730986947676336e-05,
      "loss": 152.6258,
      "step": 36760
    },
    {
      "epoch": 0.14855545275678034,
      "grad_norm": 546.4622802734375,
      "learning_rate": 3.973053027854753e-05,
      "loss": 78.2268,
      "step": 36770
    },
    {
      "epoch": 0.14859585402214798,
      "grad_norm": 0.0,
      "learning_rate": 3.973007322476188e-05,
      "loss": 75.756,
      "step": 36780
    },
    {
      "epoch": 0.14863625528751562,
      "grad_norm": 1019.0073852539062,
      "learning_rate": 3.9729615786328286e-05,
      "loss": 110.6772,
      "step": 36790
    },
    {
      "epoch": 0.14867665655288323,
      "grad_norm": 548.605224609375,
      "learning_rate": 3.972915796325569e-05,
      "loss": 66.7377,
      "step": 36800
    },
    {
      "epoch": 0.14871705781825087,
      "grad_norm": 877.7902221679688,
      "learning_rate": 3.972869975555299e-05,
      "loss": 82.7605,
      "step": 36810
    },
    {
      "epoch": 0.1487574590836185,
      "grad_norm": 1418.376708984375,
      "learning_rate": 3.972824116322913e-05,
      "loss": 96.2335,
      "step": 36820
    },
    {
      "epoch": 0.14879786034898612,
      "grad_norm": 2107.466064453125,
      "learning_rate": 3.9727782186293066e-05,
      "loss": 99.8524,
      "step": 36830
    },
    {
      "epoch": 0.14883826161435376,
      "grad_norm": 863.6719360351562,
      "learning_rate": 3.972732282475372e-05,
      "loss": 100.959,
      "step": 36840
    },
    {
      "epoch": 0.1488786628797214,
      "grad_norm": 400.9438171386719,
      "learning_rate": 3.9726863078620066e-05,
      "loss": 70.2841,
      "step": 36850
    },
    {
      "epoch": 0.14891906414508901,
      "grad_norm": 588.9529418945312,
      "learning_rate": 3.972640294790106e-05,
      "loss": 118.53,
      "step": 36860
    },
    {
      "epoch": 0.14895946541045665,
      "grad_norm": 764.5951538085938,
      "learning_rate": 3.9725942432605666e-05,
      "loss": 110.678,
      "step": 36870
    },
    {
      "epoch": 0.1489998666758243,
      "grad_norm": 999.9136352539062,
      "learning_rate": 3.972548153274287e-05,
      "loss": 86.9159,
      "step": 36880
    },
    {
      "epoch": 0.1490402679411919,
      "grad_norm": 739.2926635742188,
      "learning_rate": 3.972502024832166e-05,
      "loss": 117.9995,
      "step": 36890
    },
    {
      "epoch": 0.14908066920655955,
      "grad_norm": 1673.72900390625,
      "learning_rate": 3.972455857935102e-05,
      "loss": 57.6239,
      "step": 36900
    },
    {
      "epoch": 0.1491210704719272,
      "grad_norm": 721.6724243164062,
      "learning_rate": 3.972409652583996e-05,
      "loss": 66.6377,
      "step": 36910
    },
    {
      "epoch": 0.1491614717372948,
      "grad_norm": 580.5721435546875,
      "learning_rate": 3.9723634087797486e-05,
      "loss": 85.9561,
      "step": 36920
    },
    {
      "epoch": 0.14920187300266244,
      "grad_norm": 416.51556396484375,
      "learning_rate": 3.9723171265232606e-05,
      "loss": 111.8571,
      "step": 36930
    },
    {
      "epoch": 0.14924227426803008,
      "grad_norm": 809.1591186523438,
      "learning_rate": 3.972270805815435e-05,
      "loss": 93.3519,
      "step": 36940
    },
    {
      "epoch": 0.14928267553339772,
      "grad_norm": 522.8436279296875,
      "learning_rate": 3.9722244466571745e-05,
      "loss": 102.2965,
      "step": 36950
    },
    {
      "epoch": 0.14932307679876533,
      "grad_norm": 1699.3441162109375,
      "learning_rate": 3.972178049049383e-05,
      "loss": 125.5619,
      "step": 36960
    },
    {
      "epoch": 0.14936347806413297,
      "grad_norm": 1413.349609375,
      "learning_rate": 3.972131612992965e-05,
      "loss": 119.3396,
      "step": 36970
    },
    {
      "epoch": 0.1494038793295006,
      "grad_norm": 520.0859375,
      "learning_rate": 3.972085138488826e-05,
      "loss": 101.6701,
      "step": 36980
    },
    {
      "epoch": 0.14944428059486822,
      "grad_norm": 3968.024658203125,
      "learning_rate": 3.972038625537873e-05,
      "loss": 122.6143,
      "step": 36990
    },
    {
      "epoch": 0.14948468186023586,
      "grad_norm": 579.20751953125,
      "learning_rate": 3.97199207414101e-05,
      "loss": 85.4002,
      "step": 37000
    },
    {
      "epoch": 0.1495250831256035,
      "grad_norm": 578.19140625,
      "learning_rate": 3.9719454842991474e-05,
      "loss": 81.768,
      "step": 37010
    },
    {
      "epoch": 0.14956548439097112,
      "grad_norm": 1352.609619140625,
      "learning_rate": 3.9718988560131915e-05,
      "loss": 94.6862,
      "step": 37020
    },
    {
      "epoch": 0.14960588565633876,
      "grad_norm": 1129.244140625,
      "learning_rate": 3.971852189284053e-05,
      "loss": 98.3099,
      "step": 37030
    },
    {
      "epoch": 0.1496462869217064,
      "grad_norm": 886.0817260742188,
      "learning_rate": 3.9718054841126406e-05,
      "loss": 74.3526,
      "step": 37040
    },
    {
      "epoch": 0.149686688187074,
      "grad_norm": 792.4911499023438,
      "learning_rate": 3.9717587404998645e-05,
      "loss": 82.4121,
      "step": 37050
    },
    {
      "epoch": 0.14972708945244165,
      "grad_norm": 1291.1729736328125,
      "learning_rate": 3.971711958446638e-05,
      "loss": 125.0775,
      "step": 37060
    },
    {
      "epoch": 0.1497674907178093,
      "grad_norm": 312.7253723144531,
      "learning_rate": 3.9716651379538704e-05,
      "loss": 94.9803,
      "step": 37070
    },
    {
      "epoch": 0.1498078919831769,
      "grad_norm": 459.3520812988281,
      "learning_rate": 3.971618279022477e-05,
      "loss": 71.5689,
      "step": 37080
    },
    {
      "epoch": 0.14984829324854454,
      "grad_norm": 508.1872863769531,
      "learning_rate": 3.9715713816533695e-05,
      "loss": 55.3603,
      "step": 37090
    },
    {
      "epoch": 0.14988869451391218,
      "grad_norm": 1485.495849609375,
      "learning_rate": 3.971524445847463e-05,
      "loss": 112.9238,
      "step": 37100
    },
    {
      "epoch": 0.14992909577927982,
      "grad_norm": 974.6215209960938,
      "learning_rate": 3.971477471605672e-05,
      "loss": 102.8172,
      "step": 37110
    },
    {
      "epoch": 0.14996949704464743,
      "grad_norm": 3471.12646484375,
      "learning_rate": 3.9714304589289134e-05,
      "loss": 99.5517,
      "step": 37120
    },
    {
      "epoch": 0.15000989831001507,
      "grad_norm": 1056.742919921875,
      "learning_rate": 3.971383407818103e-05,
      "loss": 131.3219,
      "step": 37130
    },
    {
      "epoch": 0.1500502995753827,
      "grad_norm": 3554.474609375,
      "learning_rate": 3.971336318274158e-05,
      "loss": 134.7771,
      "step": 37140
    },
    {
      "epoch": 0.15009070084075032,
      "grad_norm": 358.5300598144531,
      "learning_rate": 3.971289190297997e-05,
      "loss": 95.0274,
      "step": 37150
    },
    {
      "epoch": 0.15013110210611796,
      "grad_norm": 1012.35888671875,
      "learning_rate": 3.971242023890537e-05,
      "loss": 99.5953,
      "step": 37160
    },
    {
      "epoch": 0.1501715033714856,
      "grad_norm": 862.522705078125,
      "learning_rate": 3.9711948190526997e-05,
      "loss": 89.4212,
      "step": 37170
    },
    {
      "epoch": 0.15021190463685322,
      "grad_norm": 744.9771118164062,
      "learning_rate": 3.9711475757854045e-05,
      "loss": 80.1172,
      "step": 37180
    },
    {
      "epoch": 0.15025230590222086,
      "grad_norm": 971.1306762695312,
      "learning_rate": 3.971100294089573e-05,
      "loss": 92.3551,
      "step": 37190
    },
    {
      "epoch": 0.1502927071675885,
      "grad_norm": 667.2974853515625,
      "learning_rate": 3.971052973966126e-05,
      "loss": 119.2706,
      "step": 37200
    },
    {
      "epoch": 0.1503331084329561,
      "grad_norm": 988.1494750976562,
      "learning_rate": 3.9710056154159865e-05,
      "loss": 112.7863,
      "step": 37210
    },
    {
      "epoch": 0.15037350969832375,
      "grad_norm": 602.7747192382812,
      "learning_rate": 3.970958218440078e-05,
      "loss": 109.6781,
      "step": 37220
    },
    {
      "epoch": 0.1504139109636914,
      "grad_norm": 591.7127685546875,
      "learning_rate": 3.970910783039324e-05,
      "loss": 63.6022,
      "step": 37230
    },
    {
      "epoch": 0.150454312229059,
      "grad_norm": 439.0905456542969,
      "learning_rate": 3.97086330921465e-05,
      "loss": 59.3322,
      "step": 37240
    },
    {
      "epoch": 0.15049471349442664,
      "grad_norm": 605.6318969726562,
      "learning_rate": 3.970815796966981e-05,
      "loss": 112.7581,
      "step": 37250
    },
    {
      "epoch": 0.15053511475979428,
      "grad_norm": 692.8350830078125,
      "learning_rate": 3.970768246297243e-05,
      "loss": 97.8905,
      "step": 37260
    },
    {
      "epoch": 0.15057551602516192,
      "grad_norm": 881.8389282226562,
      "learning_rate": 3.970720657206363e-05,
      "loss": 111.7944,
      "step": 37270
    },
    {
      "epoch": 0.15061591729052953,
      "grad_norm": 799.2363891601562,
      "learning_rate": 3.97067302969527e-05,
      "loss": 95.6336,
      "step": 37280
    },
    {
      "epoch": 0.15065631855589717,
      "grad_norm": 828.990478515625,
      "learning_rate": 3.970625363764892e-05,
      "loss": 90.0103,
      "step": 37290
    },
    {
      "epoch": 0.1506967198212648,
      "grad_norm": 943.7299194335938,
      "learning_rate": 3.970577659416158e-05,
      "loss": 60.9479,
      "step": 37300
    },
    {
      "epoch": 0.15073712108663243,
      "grad_norm": 560.2476196289062,
      "learning_rate": 3.970529916649997e-05,
      "loss": 61.7219,
      "step": 37310
    },
    {
      "epoch": 0.15077752235200007,
      "grad_norm": 773.2211303710938,
      "learning_rate": 3.9704821354673414e-05,
      "loss": 47.1061,
      "step": 37320
    },
    {
      "epoch": 0.1508179236173677,
      "grad_norm": 1033.5302734375,
      "learning_rate": 3.9704343158691226e-05,
      "loss": 112.5848,
      "step": 37330
    },
    {
      "epoch": 0.15085832488273532,
      "grad_norm": 778.7824096679688,
      "learning_rate": 3.970386457856271e-05,
      "loss": 58.9093,
      "step": 37340
    },
    {
      "epoch": 0.15089872614810296,
      "grad_norm": 431.2569274902344,
      "learning_rate": 3.9703385614297224e-05,
      "loss": 47.7874,
      "step": 37350
    },
    {
      "epoch": 0.1509391274134706,
      "grad_norm": 799.5265502929688,
      "learning_rate": 3.9702906265904085e-05,
      "loss": 70.1433,
      "step": 37360
    },
    {
      "epoch": 0.1509795286788382,
      "grad_norm": 407.71075439453125,
      "learning_rate": 3.9702426533392645e-05,
      "loss": 100.7932,
      "step": 37370
    },
    {
      "epoch": 0.15101992994420585,
      "grad_norm": 1554.0538330078125,
      "learning_rate": 3.9701946416772254e-05,
      "loss": 100.1079,
      "step": 37380
    },
    {
      "epoch": 0.1510603312095735,
      "grad_norm": 790.8369140625,
      "learning_rate": 3.970146591605228e-05,
      "loss": 77.1758,
      "step": 37390
    },
    {
      "epoch": 0.1511007324749411,
      "grad_norm": 796.0833129882812,
      "learning_rate": 3.9700985031242076e-05,
      "loss": 89.7363,
      "step": 37400
    },
    {
      "epoch": 0.15114113374030874,
      "grad_norm": 359.7783203125,
      "learning_rate": 3.970050376235104e-05,
      "loss": 80.3151,
      "step": 37410
    },
    {
      "epoch": 0.15118153500567638,
      "grad_norm": 904.0203857421875,
      "learning_rate": 3.970002210938853e-05,
      "loss": 121.1745,
      "step": 37420
    },
    {
      "epoch": 0.15122193627104402,
      "grad_norm": 702.6942138671875,
      "learning_rate": 3.969954007236395e-05,
      "loss": 53.5863,
      "step": 37430
    },
    {
      "epoch": 0.15126233753641163,
      "grad_norm": 1143.8529052734375,
      "learning_rate": 3.969905765128669e-05,
      "loss": 74.2955,
      "step": 37440
    },
    {
      "epoch": 0.15130273880177927,
      "grad_norm": 795.322021484375,
      "learning_rate": 3.969857484616616e-05,
      "loss": 77.7138,
      "step": 37450
    },
    {
      "epoch": 0.15134314006714691,
      "grad_norm": 1021.1033325195312,
      "learning_rate": 3.969809165701177e-05,
      "loss": 111.0508,
      "step": 37460
    },
    {
      "epoch": 0.15138354133251453,
      "grad_norm": 521.082275390625,
      "learning_rate": 3.969760808383295e-05,
      "loss": 83.8422,
      "step": 37470
    },
    {
      "epoch": 0.15142394259788217,
      "grad_norm": 778.8450927734375,
      "learning_rate": 3.969712412663912e-05,
      "loss": 60.9541,
      "step": 37480
    },
    {
      "epoch": 0.1514643438632498,
      "grad_norm": 593.9297485351562,
      "learning_rate": 3.969663978543971e-05,
      "loss": 69.2896,
      "step": 37490
    },
    {
      "epoch": 0.15150474512861742,
      "grad_norm": 609.1443481445312,
      "learning_rate": 3.9696155060244166e-05,
      "loss": 111.5514,
      "step": 37500
    },
    {
      "epoch": 0.15154514639398506,
      "grad_norm": 730.1421508789062,
      "learning_rate": 3.969566995106194e-05,
      "loss": 81.852,
      "step": 37510
    },
    {
      "epoch": 0.1515855476593527,
      "grad_norm": 831.3905029296875,
      "learning_rate": 3.969518445790248e-05,
      "loss": 80.2276,
      "step": 37520
    },
    {
      "epoch": 0.1516259489247203,
      "grad_norm": 1291.9847412109375,
      "learning_rate": 3.9694698580775275e-05,
      "loss": 99.0928,
      "step": 37530
    },
    {
      "epoch": 0.15166635019008795,
      "grad_norm": 809.3699340820312,
      "learning_rate": 3.969421231968977e-05,
      "loss": 78.3106,
      "step": 37540
    },
    {
      "epoch": 0.1517067514554556,
      "grad_norm": 2072.676025390625,
      "learning_rate": 3.9693725674655465e-05,
      "loss": 85.1074,
      "step": 37550
    },
    {
      "epoch": 0.1517471527208232,
      "grad_norm": 1137.106689453125,
      "learning_rate": 3.969323864568183e-05,
      "loss": 93.6903,
      "step": 37560
    },
    {
      "epoch": 0.15178755398619084,
      "grad_norm": 426.4228515625,
      "learning_rate": 3.9692751232778376e-05,
      "loss": 83.9971,
      "step": 37570
    },
    {
      "epoch": 0.15182795525155848,
      "grad_norm": 567.2477416992188,
      "learning_rate": 3.96922634359546e-05,
      "loss": 73.9459,
      "step": 37580
    },
    {
      "epoch": 0.15186835651692612,
      "grad_norm": 965.7251586914062,
      "learning_rate": 3.969177525522e-05,
      "loss": 67.762,
      "step": 37590
    },
    {
      "epoch": 0.15190875778229374,
      "grad_norm": 1557.0831298828125,
      "learning_rate": 3.969128669058411e-05,
      "loss": 105.5141,
      "step": 37600
    },
    {
      "epoch": 0.15194915904766138,
      "grad_norm": 730.3154907226562,
      "learning_rate": 3.969079774205645e-05,
      "loss": 52.5085,
      "step": 37610
    },
    {
      "epoch": 0.15198956031302902,
      "grad_norm": 424.4344177246094,
      "learning_rate": 3.969030840964655e-05,
      "loss": 81.0125,
      "step": 37620
    },
    {
      "epoch": 0.15202996157839663,
      "grad_norm": 1289.1795654296875,
      "learning_rate": 3.968981869336395e-05,
      "loss": 118.6801,
      "step": 37630
    },
    {
      "epoch": 0.15207036284376427,
      "grad_norm": 768.481201171875,
      "learning_rate": 3.968932859321819e-05,
      "loss": 116.7715,
      "step": 37640
    },
    {
      "epoch": 0.1521107641091319,
      "grad_norm": 1433.2896728515625,
      "learning_rate": 3.9688838109218836e-05,
      "loss": 99.1868,
      "step": 37650
    },
    {
      "epoch": 0.15215116537449952,
      "grad_norm": 1053.1134033203125,
      "learning_rate": 3.968834724137545e-05,
      "loss": 64.6867,
      "step": 37660
    },
    {
      "epoch": 0.15219156663986716,
      "grad_norm": 1071.5689697265625,
      "learning_rate": 3.96878559896976e-05,
      "loss": 77.4557,
      "step": 37670
    },
    {
      "epoch": 0.1522319679052348,
      "grad_norm": 763.2518920898438,
      "learning_rate": 3.9687364354194854e-05,
      "loss": 102.5508,
      "step": 37680
    },
    {
      "epoch": 0.1522723691706024,
      "grad_norm": 698.8189697265625,
      "learning_rate": 3.968687233487681e-05,
      "loss": 58.931,
      "step": 37690
    },
    {
      "epoch": 0.15231277043597005,
      "grad_norm": 1018.3699951171875,
      "learning_rate": 3.968637993175305e-05,
      "loss": 118.7861,
      "step": 37700
    },
    {
      "epoch": 0.1523531717013377,
      "grad_norm": 557.1495361328125,
      "learning_rate": 3.9685887144833185e-05,
      "loss": 75.3155,
      "step": 37710
    },
    {
      "epoch": 0.1523935729667053,
      "grad_norm": 0.0,
      "learning_rate": 3.9685393974126804e-05,
      "loss": 75.7861,
      "step": 37720
    },
    {
      "epoch": 0.15243397423207294,
      "grad_norm": 697.90185546875,
      "learning_rate": 3.9684900419643544e-05,
      "loss": 46.2014,
      "step": 37730
    },
    {
      "epoch": 0.15247437549744058,
      "grad_norm": 1075.672119140625,
      "learning_rate": 3.968440648139301e-05,
      "loss": 83.1796,
      "step": 37740
    },
    {
      "epoch": 0.15251477676280822,
      "grad_norm": 786.6807861328125,
      "learning_rate": 3.968391215938484e-05,
      "loss": 68.6527,
      "step": 37750
    },
    {
      "epoch": 0.15255517802817584,
      "grad_norm": 923.9205932617188,
      "learning_rate": 3.968341745362867e-05,
      "loss": 108.3222,
      "step": 37760
    },
    {
      "epoch": 0.15259557929354348,
      "grad_norm": 702.1288452148438,
      "learning_rate": 3.968292236413414e-05,
      "loss": 140.3841,
      "step": 37770
    },
    {
      "epoch": 0.15263598055891112,
      "grad_norm": 916.8606567382812,
      "learning_rate": 3.96824268909109e-05,
      "loss": 76.2463,
      "step": 37780
    },
    {
      "epoch": 0.15267638182427873,
      "grad_norm": 715.5382080078125,
      "learning_rate": 3.9681931033968625e-05,
      "loss": 83.2818,
      "step": 37790
    },
    {
      "epoch": 0.15271678308964637,
      "grad_norm": 668.6615600585938,
      "learning_rate": 3.9681434793316966e-05,
      "loss": 57.9715,
      "step": 37800
    },
    {
      "epoch": 0.152757184355014,
      "grad_norm": 879.0608520507812,
      "learning_rate": 3.96809381689656e-05,
      "loss": 108.9632,
      "step": 37810
    },
    {
      "epoch": 0.15279758562038162,
      "grad_norm": 1151.613037109375,
      "learning_rate": 3.9680441160924225e-05,
      "loss": 104.5107,
      "step": 37820
    },
    {
      "epoch": 0.15283798688574926,
      "grad_norm": 1173.372314453125,
      "learning_rate": 3.96799437692025e-05,
      "loss": 104.231,
      "step": 37830
    },
    {
      "epoch": 0.1528783881511169,
      "grad_norm": 2017.4984130859375,
      "learning_rate": 3.967944599381015e-05,
      "loss": 105.15,
      "step": 37840
    },
    {
      "epoch": 0.1529187894164845,
      "grad_norm": 1668.8818359375,
      "learning_rate": 3.9678947834756864e-05,
      "loss": 104.0688,
      "step": 37850
    },
    {
      "epoch": 0.15295919068185215,
      "grad_norm": 597.96337890625,
      "learning_rate": 3.967844929205236e-05,
      "loss": 78.81,
      "step": 37860
    },
    {
      "epoch": 0.1529995919472198,
      "grad_norm": 646.5809936523438,
      "learning_rate": 3.9677950365706365e-05,
      "loss": 72.663,
      "step": 37870
    },
    {
      "epoch": 0.1530399932125874,
      "grad_norm": 1005.3717651367188,
      "learning_rate": 3.967745105572858e-05,
      "loss": 123.4354,
      "step": 37880
    },
    {
      "epoch": 0.15308039447795505,
      "grad_norm": 674.1689453125,
      "learning_rate": 3.967695136212877e-05,
      "loss": 74.8123,
      "step": 37890
    },
    {
      "epoch": 0.15312079574332269,
      "grad_norm": 919.642822265625,
      "learning_rate": 3.967645128491666e-05,
      "loss": 55.2663,
      "step": 37900
    },
    {
      "epoch": 0.15316119700869033,
      "grad_norm": 559.356689453125,
      "learning_rate": 3.967595082410199e-05,
      "loss": 81.8104,
      "step": 37910
    },
    {
      "epoch": 0.15320159827405794,
      "grad_norm": 852.5032958984375,
      "learning_rate": 3.967544997969454e-05,
      "loss": 72.4133,
      "step": 37920
    },
    {
      "epoch": 0.15324199953942558,
      "grad_norm": 760.2708129882812,
      "learning_rate": 3.967494875170406e-05,
      "loss": 113.9829,
      "step": 37930
    },
    {
      "epoch": 0.15328240080479322,
      "grad_norm": 472.9176025390625,
      "learning_rate": 3.967444714014032e-05,
      "loss": 96.0468,
      "step": 37940
    },
    {
      "epoch": 0.15332280207016083,
      "grad_norm": 771.8394165039062,
      "learning_rate": 3.9673945145013114e-05,
      "loss": 81.0025,
      "step": 37950
    },
    {
      "epoch": 0.15336320333552847,
      "grad_norm": 472.99627685546875,
      "learning_rate": 3.967344276633222e-05,
      "loss": 81.9558,
      "step": 37960
    },
    {
      "epoch": 0.1534036046008961,
      "grad_norm": 555.433837890625,
      "learning_rate": 3.9672940004107426e-05,
      "loss": 103.9633,
      "step": 37970
    },
    {
      "epoch": 0.15344400586626372,
      "grad_norm": 860.8916625976562,
      "learning_rate": 3.967243685834854e-05,
      "loss": 104.1778,
      "step": 37980
    },
    {
      "epoch": 0.15348440713163136,
      "grad_norm": 648.06787109375,
      "learning_rate": 3.967193332906537e-05,
      "loss": 76.7259,
      "step": 37990
    },
    {
      "epoch": 0.153524808396999,
      "grad_norm": 1496.262451171875,
      "learning_rate": 3.967142941626772e-05,
      "loss": 76.2057,
      "step": 38000
    },
    {
      "epoch": 0.15356520966236661,
      "grad_norm": 1186.5738525390625,
      "learning_rate": 3.9670925119965434e-05,
      "loss": 90.1337,
      "step": 38010
    },
    {
      "epoch": 0.15360561092773425,
      "grad_norm": 628.0262451171875,
      "learning_rate": 3.9670420440168335e-05,
      "loss": 96.4808,
      "step": 38020
    },
    {
      "epoch": 0.1536460121931019,
      "grad_norm": 1244.613037109375,
      "learning_rate": 3.9669915376886265e-05,
      "loss": 83.5602,
      "step": 38030
    },
    {
      "epoch": 0.1536864134584695,
      "grad_norm": 1034.5682373046875,
      "learning_rate": 3.966940993012907e-05,
      "loss": 84.4719,
      "step": 38040
    },
    {
      "epoch": 0.15372681472383715,
      "grad_norm": 630.8580322265625,
      "learning_rate": 3.96689040999066e-05,
      "loss": 73.0619,
      "step": 38050
    },
    {
      "epoch": 0.1537672159892048,
      "grad_norm": 765.167724609375,
      "learning_rate": 3.966839788622872e-05,
      "loss": 106.9954,
      "step": 38060
    },
    {
      "epoch": 0.15380761725457243,
      "grad_norm": 0.0,
      "learning_rate": 3.966789128910529e-05,
      "loss": 49.2205,
      "step": 38070
    },
    {
      "epoch": 0.15384801851994004,
      "grad_norm": 869.7047729492188,
      "learning_rate": 3.966738430854619e-05,
      "loss": 94.2237,
      "step": 38080
    },
    {
      "epoch": 0.15388841978530768,
      "grad_norm": 412.8327941894531,
      "learning_rate": 3.966687694456132e-05,
      "loss": 92.6531,
      "step": 38090
    },
    {
      "epoch": 0.15392882105067532,
      "grad_norm": 994.3251342773438,
      "learning_rate": 3.966636919716056e-05,
      "loss": 97.2214,
      "step": 38100
    },
    {
      "epoch": 0.15396922231604293,
      "grad_norm": 1117.16650390625,
      "learning_rate": 3.966586106635379e-05,
      "loss": 86.1158,
      "step": 38110
    },
    {
      "epoch": 0.15400962358141057,
      "grad_norm": 568.3110961914062,
      "learning_rate": 3.966535255215095e-05,
      "loss": 111.1136,
      "step": 38120
    },
    {
      "epoch": 0.1540500248467782,
      "grad_norm": 1117.4755859375,
      "learning_rate": 3.966484365456193e-05,
      "loss": 82.1834,
      "step": 38130
    },
    {
      "epoch": 0.15409042611214582,
      "grad_norm": 926.09326171875,
      "learning_rate": 3.966433437359667e-05,
      "loss": 92.5847,
      "step": 38140
    },
    {
      "epoch": 0.15413082737751346,
      "grad_norm": 760.3319702148438,
      "learning_rate": 3.9663824709265075e-05,
      "loss": 81.4818,
      "step": 38150
    },
    {
      "epoch": 0.1541712286428811,
      "grad_norm": 694.4853515625,
      "learning_rate": 3.96633146615771e-05,
      "loss": 83.545,
      "step": 38160
    },
    {
      "epoch": 0.15421162990824872,
      "grad_norm": 606.398193359375,
      "learning_rate": 3.9662804230542676e-05,
      "loss": 83.9181,
      "step": 38170
    },
    {
      "epoch": 0.15425203117361636,
      "grad_norm": 901.7685546875,
      "learning_rate": 3.966229341617177e-05,
      "loss": 77.0491,
      "step": 38180
    },
    {
      "epoch": 0.154292432438984,
      "grad_norm": 830.6808471679688,
      "learning_rate": 3.9661782218474325e-05,
      "loss": 70.1705,
      "step": 38190
    },
    {
      "epoch": 0.1543328337043516,
      "grad_norm": 1575.902099609375,
      "learning_rate": 3.966127063746031e-05,
      "loss": 61.3311,
      "step": 38200
    },
    {
      "epoch": 0.15437323496971925,
      "grad_norm": 844.6079711914062,
      "learning_rate": 3.966075867313971e-05,
      "loss": 88.1307,
      "step": 38210
    },
    {
      "epoch": 0.1544136362350869,
      "grad_norm": 413.93865966796875,
      "learning_rate": 3.966024632552249e-05,
      "loss": 95.2208,
      "step": 38220
    },
    {
      "epoch": 0.15445403750045453,
      "grad_norm": 789.9360961914062,
      "learning_rate": 3.965973359461865e-05,
      "loss": 105.475,
      "step": 38230
    },
    {
      "epoch": 0.15449443876582214,
      "grad_norm": 0.0,
      "learning_rate": 3.965922048043818e-05,
      "loss": 127.8588,
      "step": 38240
    },
    {
      "epoch": 0.15453484003118978,
      "grad_norm": 3267.522216796875,
      "learning_rate": 3.965870698299109e-05,
      "loss": 126.6062,
      "step": 38250
    },
    {
      "epoch": 0.15457524129655742,
      "grad_norm": 1106.333251953125,
      "learning_rate": 3.965819310228738e-05,
      "loss": 93.0975,
      "step": 38260
    },
    {
      "epoch": 0.15461564256192503,
      "grad_norm": 1589.9356689453125,
      "learning_rate": 3.965767883833708e-05,
      "loss": 113.15,
      "step": 38270
    },
    {
      "epoch": 0.15465604382729267,
      "grad_norm": 1526.2000732421875,
      "learning_rate": 3.965716419115021e-05,
      "loss": 96.7691,
      "step": 38280
    },
    {
      "epoch": 0.1546964450926603,
      "grad_norm": 736.6803588867188,
      "learning_rate": 3.96566491607368e-05,
      "loss": 96.5812,
      "step": 38290
    },
    {
      "epoch": 0.15473684635802792,
      "grad_norm": 495.99993896484375,
      "learning_rate": 3.96561337471069e-05,
      "loss": 91.3958,
      "step": 38300
    },
    {
      "epoch": 0.15477724762339556,
      "grad_norm": 1220.4888916015625,
      "learning_rate": 3.965561795027054e-05,
      "loss": 86.7831,
      "step": 38310
    },
    {
      "epoch": 0.1548176488887632,
      "grad_norm": 511.2290344238281,
      "learning_rate": 3.9655101770237805e-05,
      "loss": 126.2356,
      "step": 38320
    },
    {
      "epoch": 0.15485805015413082,
      "grad_norm": 993.008544921875,
      "learning_rate": 3.965458520701874e-05,
      "loss": 82.0525,
      "step": 38330
    },
    {
      "epoch": 0.15489845141949846,
      "grad_norm": 491.7477722167969,
      "learning_rate": 3.965406826062341e-05,
      "loss": 73.5542,
      "step": 38340
    },
    {
      "epoch": 0.1549388526848661,
      "grad_norm": 975.9268188476562,
      "learning_rate": 3.96535509310619e-05,
      "loss": 62.1843,
      "step": 38350
    },
    {
      "epoch": 0.1549792539502337,
      "grad_norm": 1564.1531982421875,
      "learning_rate": 3.96530332183443e-05,
      "loss": 76.3985,
      "step": 38360
    },
    {
      "epoch": 0.15501965521560135,
      "grad_norm": 1053.3935546875,
      "learning_rate": 3.96525151224807e-05,
      "loss": 70.3877,
      "step": 38370
    },
    {
      "epoch": 0.155060056480969,
      "grad_norm": 820.1409912109375,
      "learning_rate": 3.96519966434812e-05,
      "loss": 93.3208,
      "step": 38380
    },
    {
      "epoch": 0.15510045774633663,
      "grad_norm": 428.53887939453125,
      "learning_rate": 3.965147778135591e-05,
      "loss": 78.4831,
      "step": 38390
    },
    {
      "epoch": 0.15514085901170424,
      "grad_norm": 626.5733032226562,
      "learning_rate": 3.965095853611494e-05,
      "loss": 100.7439,
      "step": 38400
    },
    {
      "epoch": 0.15518126027707188,
      "grad_norm": 709.6735229492188,
      "learning_rate": 3.9650438907768414e-05,
      "loss": 148.4095,
      "step": 38410
    },
    {
      "epoch": 0.15522166154243952,
      "grad_norm": 899.0986938476562,
      "learning_rate": 3.964991889632647e-05,
      "loss": 86.0226,
      "step": 38420
    },
    {
      "epoch": 0.15526206280780713,
      "grad_norm": 1556.7742919921875,
      "learning_rate": 3.964939850179923e-05,
      "loss": 74.1751,
      "step": 38430
    },
    {
      "epoch": 0.15530246407317477,
      "grad_norm": 298.58319091796875,
      "learning_rate": 3.964887772419687e-05,
      "loss": 107.9748,
      "step": 38440
    },
    {
      "epoch": 0.1553428653385424,
      "grad_norm": 656.981689453125,
      "learning_rate": 3.9648356563529506e-05,
      "loss": 115.6758,
      "step": 38450
    },
    {
      "epoch": 0.15538326660391003,
      "grad_norm": 917.2551879882812,
      "learning_rate": 3.964783501980732e-05,
      "loss": 72.7824,
      "step": 38460
    },
    {
      "epoch": 0.15542366786927767,
      "grad_norm": 312.0767822265625,
      "learning_rate": 3.9647313093040475e-05,
      "loss": 93.3208,
      "step": 38470
    },
    {
      "epoch": 0.1554640691346453,
      "grad_norm": 919.5922241210938,
      "learning_rate": 3.964679078323915e-05,
      "loss": 116.9953,
      "step": 38480
    },
    {
      "epoch": 0.15550447040001292,
      "grad_norm": 898.2864379882812,
      "learning_rate": 3.9646268090413516e-05,
      "loss": 89.4504,
      "step": 38490
    },
    {
      "epoch": 0.15554487166538056,
      "grad_norm": 948.1696166992188,
      "learning_rate": 3.964574501457378e-05,
      "loss": 128.4789,
      "step": 38500
    },
    {
      "epoch": 0.1555852729307482,
      "grad_norm": 2223.70166015625,
      "learning_rate": 3.964522155573012e-05,
      "loss": 104.3688,
      "step": 38510
    },
    {
      "epoch": 0.1556256741961158,
      "grad_norm": 705.4492797851562,
      "learning_rate": 3.964469771389276e-05,
      "loss": 97.0524,
      "step": 38520
    },
    {
      "epoch": 0.15566607546148345,
      "grad_norm": 1810.4759521484375,
      "learning_rate": 3.96441734890719e-05,
      "loss": 72.1592,
      "step": 38530
    },
    {
      "epoch": 0.1557064767268511,
      "grad_norm": 500.4573974609375,
      "learning_rate": 3.964364888127777e-05,
      "loss": 98.4273,
      "step": 38540
    },
    {
      "epoch": 0.15574687799221873,
      "grad_norm": 335.0638122558594,
      "learning_rate": 3.9643123890520584e-05,
      "loss": 70.7148,
      "step": 38550
    },
    {
      "epoch": 0.15578727925758634,
      "grad_norm": 1567.4873046875,
      "learning_rate": 3.9642598516810593e-05,
      "loss": 90.5238,
      "step": 38560
    },
    {
      "epoch": 0.15582768052295398,
      "grad_norm": 552.8950805664062,
      "learning_rate": 3.9642072760158024e-05,
      "loss": 108.9904,
      "step": 38570
    },
    {
      "epoch": 0.15586808178832162,
      "grad_norm": 634.5554809570312,
      "learning_rate": 3.964154662057314e-05,
      "loss": 60.6461,
      "step": 38580
    },
    {
      "epoch": 0.15590848305368923,
      "grad_norm": 1294.1488037109375,
      "learning_rate": 3.9641020098066185e-05,
      "loss": 93.5097,
      "step": 38590
    },
    {
      "epoch": 0.15594888431905687,
      "grad_norm": 1212.355712890625,
      "learning_rate": 3.964049319264744e-05,
      "loss": 108.3358,
      "step": 38600
    },
    {
      "epoch": 0.15598928558442451,
      "grad_norm": 1027.73974609375,
      "learning_rate": 3.963996590432716e-05,
      "loss": 101.4828,
      "step": 38610
    },
    {
      "epoch": 0.15602968684979213,
      "grad_norm": 816.5595703125,
      "learning_rate": 3.963943823311564e-05,
      "loss": 77.334,
      "step": 38620
    },
    {
      "epoch": 0.15607008811515977,
      "grad_norm": 497.77679443359375,
      "learning_rate": 3.9638910179023156e-05,
      "loss": 114.117,
      "step": 38630
    },
    {
      "epoch": 0.1561104893805274,
      "grad_norm": 731.276123046875,
      "learning_rate": 3.963838174206001e-05,
      "loss": 91.6992,
      "step": 38640
    },
    {
      "epoch": 0.15615089064589502,
      "grad_norm": 533.4052124023438,
      "learning_rate": 3.963785292223651e-05,
      "loss": 96.2654,
      "step": 38650
    },
    {
      "epoch": 0.15619129191126266,
      "grad_norm": 421.8385009765625,
      "learning_rate": 3.9637323719562936e-05,
      "loss": 66.5563,
      "step": 38660
    },
    {
      "epoch": 0.1562316931766303,
      "grad_norm": 3177.8974609375,
      "learning_rate": 3.963679413404964e-05,
      "loss": 131.4963,
      "step": 38670
    },
    {
      "epoch": 0.1562720944419979,
      "grad_norm": 771.8431396484375,
      "learning_rate": 3.963626416570693e-05,
      "loss": 102.0437,
      "step": 38680
    },
    {
      "epoch": 0.15631249570736555,
      "grad_norm": 965.1478881835938,
      "learning_rate": 3.963573381454515e-05,
      "loss": 93.6955,
      "step": 38690
    },
    {
      "epoch": 0.1563528969727332,
      "grad_norm": 706.6978149414062,
      "learning_rate": 3.963520308057462e-05,
      "loss": 96.3812,
      "step": 38700
    },
    {
      "epoch": 0.15639329823810083,
      "grad_norm": 1077.8919677734375,
      "learning_rate": 3.96346719638057e-05,
      "loss": 55.8834,
      "step": 38710
    },
    {
      "epoch": 0.15643369950346844,
      "grad_norm": 2407.916748046875,
      "learning_rate": 3.963414046424874e-05,
      "loss": 59.8068,
      "step": 38720
    },
    {
      "epoch": 0.15647410076883608,
      "grad_norm": 1492.0802001953125,
      "learning_rate": 3.96336085819141e-05,
      "loss": 88.5377,
      "step": 38730
    },
    {
      "epoch": 0.15651450203420372,
      "grad_norm": 1077.629150390625,
      "learning_rate": 3.9633076316812155e-05,
      "loss": 86.7247,
      "step": 38740
    },
    {
      "epoch": 0.15655490329957134,
      "grad_norm": 468.01129150390625,
      "learning_rate": 3.9632543668953284e-05,
      "loss": 93.5541,
      "step": 38750
    },
    {
      "epoch": 0.15659530456493898,
      "grad_norm": 1229.2720947265625,
      "learning_rate": 3.9632010638347865e-05,
      "loss": 70.9075,
      "step": 38760
    },
    {
      "epoch": 0.15663570583030662,
      "grad_norm": 1041.6185302734375,
      "learning_rate": 3.9631477225006285e-05,
      "loss": 112.2093,
      "step": 38770
    },
    {
      "epoch": 0.15667610709567423,
      "grad_norm": 811.9631958007812,
      "learning_rate": 3.963094342893896e-05,
      "loss": 61.5546,
      "step": 38780
    },
    {
      "epoch": 0.15671650836104187,
      "grad_norm": 959.6410522460938,
      "learning_rate": 3.963040925015628e-05,
      "loss": 96.6481,
      "step": 38790
    },
    {
      "epoch": 0.1567569096264095,
      "grad_norm": 364.67559814453125,
      "learning_rate": 3.962987468866866e-05,
      "loss": 65.4956,
      "step": 38800
    },
    {
      "epoch": 0.15679731089177712,
      "grad_norm": 1041.04150390625,
      "learning_rate": 3.9629339744486534e-05,
      "loss": 81.0789,
      "step": 38810
    },
    {
      "epoch": 0.15683771215714476,
      "grad_norm": 1525.4169921875,
      "learning_rate": 3.962880441762032e-05,
      "loss": 116.3364,
      "step": 38820
    },
    {
      "epoch": 0.1568781134225124,
      "grad_norm": 820.9378662109375,
      "learning_rate": 3.962826870808046e-05,
      "loss": 102.4911,
      "step": 38830
    },
    {
      "epoch": 0.15691851468788,
      "grad_norm": 564.9578247070312,
      "learning_rate": 3.96277326158774e-05,
      "loss": 71.7791,
      "step": 38840
    },
    {
      "epoch": 0.15695891595324765,
      "grad_norm": 426.0314025878906,
      "learning_rate": 3.962719614102158e-05,
      "loss": 87.1656,
      "step": 38850
    },
    {
      "epoch": 0.1569993172186153,
      "grad_norm": 1478.522705078125,
      "learning_rate": 3.9626659283523475e-05,
      "loss": 75.4472,
      "step": 38860
    },
    {
      "epoch": 0.15703971848398293,
      "grad_norm": 519.1206665039062,
      "learning_rate": 3.9626122043393535e-05,
      "loss": 90.5628,
      "step": 38870
    },
    {
      "epoch": 0.15708011974935054,
      "grad_norm": 369.3790283203125,
      "learning_rate": 3.9625584420642245e-05,
      "loss": 83.9716,
      "step": 38880
    },
    {
      "epoch": 0.15712052101471818,
      "grad_norm": 951.3013916015625,
      "learning_rate": 3.962504641528009e-05,
      "loss": 65.3877,
      "step": 38890
    },
    {
      "epoch": 0.15716092228008582,
      "grad_norm": 523.937744140625,
      "learning_rate": 3.962450802731754e-05,
      "loss": 60.1187,
      "step": 38900
    },
    {
      "epoch": 0.15720132354545344,
      "grad_norm": 1172.847900390625,
      "learning_rate": 3.962396925676511e-05,
      "loss": 85.6999,
      "step": 38910
    },
    {
      "epoch": 0.15724172481082108,
      "grad_norm": 676.1847534179688,
      "learning_rate": 3.962343010363329e-05,
      "loss": 98.4259,
      "step": 38920
    },
    {
      "epoch": 0.15728212607618872,
      "grad_norm": 1541.6544189453125,
      "learning_rate": 3.9622890567932605e-05,
      "loss": 85.6831,
      "step": 38930
    },
    {
      "epoch": 0.15732252734155633,
      "grad_norm": 1251.5394287109375,
      "learning_rate": 3.962235064967356e-05,
      "loss": 114.6008,
      "step": 38940
    },
    {
      "epoch": 0.15736292860692397,
      "grad_norm": 858.481201171875,
      "learning_rate": 3.962181034886668e-05,
      "loss": 80.0586,
      "step": 38950
    },
    {
      "epoch": 0.1574033298722916,
      "grad_norm": 737.1397705078125,
      "learning_rate": 3.9621269665522516e-05,
      "loss": 85.137,
      "step": 38960
    },
    {
      "epoch": 0.15744373113765922,
      "grad_norm": 771.252685546875,
      "learning_rate": 3.9620728599651596e-05,
      "loss": 107.0303,
      "step": 38970
    },
    {
      "epoch": 0.15748413240302686,
      "grad_norm": 555.4755249023438,
      "learning_rate": 3.9620187151264474e-05,
      "loss": 94.9436,
      "step": 38980
    },
    {
      "epoch": 0.1575245336683945,
      "grad_norm": 2882.533203125,
      "learning_rate": 3.961964532037169e-05,
      "loss": 102.9361,
      "step": 38990
    },
    {
      "epoch": 0.1575649349337621,
      "grad_norm": 562.4039306640625,
      "learning_rate": 3.9619103106983835e-05,
      "loss": 123.3879,
      "step": 39000
    },
    {
      "epoch": 0.15760533619912975,
      "grad_norm": 1785.08740234375,
      "learning_rate": 3.961856051111146e-05,
      "loss": 98.0155,
      "step": 39010
    },
    {
      "epoch": 0.1576457374644974,
      "grad_norm": 763.029296875,
      "learning_rate": 3.961801753276514e-05,
      "loss": 41.8298,
      "step": 39020
    },
    {
      "epoch": 0.15768613872986503,
      "grad_norm": 1000.0222778320312,
      "learning_rate": 3.9617474171955475e-05,
      "loss": 94.5578,
      "step": 39030
    },
    {
      "epoch": 0.15772653999523265,
      "grad_norm": 334.7162780761719,
      "learning_rate": 3.961693042869305e-05,
      "loss": 86.058,
      "step": 39040
    },
    {
      "epoch": 0.15776694126060029,
      "grad_norm": 1190.070556640625,
      "learning_rate": 3.961638630298847e-05,
      "loss": 98.6637,
      "step": 39050
    },
    {
      "epoch": 0.15780734252596793,
      "grad_norm": 629.3197021484375,
      "learning_rate": 3.9615841794852336e-05,
      "loss": 83.3572,
      "step": 39060
    },
    {
      "epoch": 0.15784774379133554,
      "grad_norm": 460.8150329589844,
      "learning_rate": 3.9615296904295264e-05,
      "loss": 122.8573,
      "step": 39070
    },
    {
      "epoch": 0.15788814505670318,
      "grad_norm": 604.5160522460938,
      "learning_rate": 3.961475163132789e-05,
      "loss": 55.7113,
      "step": 39080
    },
    {
      "epoch": 0.15792854632207082,
      "grad_norm": 1380.890380859375,
      "learning_rate": 3.961420597596082e-05,
      "loss": 121.6424,
      "step": 39090
    },
    {
      "epoch": 0.15796894758743843,
      "grad_norm": 808.0025634765625,
      "learning_rate": 3.961365993820471e-05,
      "loss": 66.9209,
      "step": 39100
    },
    {
      "epoch": 0.15800934885280607,
      "grad_norm": 3276.390625,
      "learning_rate": 3.961311351807022e-05,
      "loss": 82.583,
      "step": 39110
    },
    {
      "epoch": 0.1580497501181737,
      "grad_norm": 2217.059326171875,
      "learning_rate": 3.961256671556796e-05,
      "loss": 89.6204,
      "step": 39120
    },
    {
      "epoch": 0.15809015138354132,
      "grad_norm": 1654.140380859375,
      "learning_rate": 3.961201953070863e-05,
      "loss": 64.8732,
      "step": 39130
    },
    {
      "epoch": 0.15813055264890896,
      "grad_norm": 532.6248168945312,
      "learning_rate": 3.961147196350288e-05,
      "loss": 98.7978,
      "step": 39140
    },
    {
      "epoch": 0.1581709539142766,
      "grad_norm": 809.4042358398438,
      "learning_rate": 3.9610924013961376e-05,
      "loss": 69.8393,
      "step": 39150
    },
    {
      "epoch": 0.15821135517964421,
      "grad_norm": 918.3334350585938,
      "learning_rate": 3.9610375682094824e-05,
      "loss": 77.7073,
      "step": 39160
    },
    {
      "epoch": 0.15825175644501185,
      "grad_norm": 0.0,
      "learning_rate": 3.96098269679139e-05,
      "loss": 61.9318,
      "step": 39170
    },
    {
      "epoch": 0.1582921577103795,
      "grad_norm": 401.17279052734375,
      "learning_rate": 3.96092778714293e-05,
      "loss": 114.9751,
      "step": 39180
    },
    {
      "epoch": 0.15833255897574713,
      "grad_norm": 731.6635131835938,
      "learning_rate": 3.9608728392651734e-05,
      "loss": 53.0921,
      "step": 39190
    },
    {
      "epoch": 0.15837296024111475,
      "grad_norm": 1872.025390625,
      "learning_rate": 3.960817853159192e-05,
      "loss": 80.1266,
      "step": 39200
    },
    {
      "epoch": 0.1584133615064824,
      "grad_norm": 764.140869140625,
      "learning_rate": 3.960762828826056e-05,
      "loss": 84.0326,
      "step": 39210
    },
    {
      "epoch": 0.15845376277185003,
      "grad_norm": 877.96484375,
      "learning_rate": 3.96070776626684e-05,
      "loss": 82.8266,
      "step": 39220
    },
    {
      "epoch": 0.15849416403721764,
      "grad_norm": 363.5792541503906,
      "learning_rate": 3.9606526654826154e-05,
      "loss": 79.819,
      "step": 39230
    },
    {
      "epoch": 0.15853456530258528,
      "grad_norm": 555.1956787109375,
      "learning_rate": 3.960597526474459e-05,
      "loss": 68.749,
      "step": 39240
    },
    {
      "epoch": 0.15857496656795292,
      "grad_norm": 543.0640869140625,
      "learning_rate": 3.9605423492434444e-05,
      "loss": 53.4244,
      "step": 39250
    },
    {
      "epoch": 0.15861536783332053,
      "grad_norm": 538.7412109375,
      "learning_rate": 3.9604871337906466e-05,
      "loss": 79.1591,
      "step": 39260
    },
    {
      "epoch": 0.15865576909868817,
      "grad_norm": 1329.620849609375,
      "learning_rate": 3.960431880117143e-05,
      "loss": 79.2483,
      "step": 39270
    },
    {
      "epoch": 0.1586961703640558,
      "grad_norm": 521.2180786132812,
      "learning_rate": 3.96037658822401e-05,
      "loss": 84.4006,
      "step": 39280
    },
    {
      "epoch": 0.15873657162942342,
      "grad_norm": 1234.4501953125,
      "learning_rate": 3.960321258112328e-05,
      "loss": 56.2984,
      "step": 39290
    },
    {
      "epoch": 0.15877697289479106,
      "grad_norm": 1980.7930908203125,
      "learning_rate": 3.960265889783173e-05,
      "loss": 89.2887,
      "step": 39300
    },
    {
      "epoch": 0.1588173741601587,
      "grad_norm": 1109.7982177734375,
      "learning_rate": 3.9602104832376244e-05,
      "loss": 67.8054,
      "step": 39310
    },
    {
      "epoch": 0.15885777542552632,
      "grad_norm": 1189.190185546875,
      "learning_rate": 3.960155038476764e-05,
      "loss": 72.3154,
      "step": 39320
    },
    {
      "epoch": 0.15889817669089396,
      "grad_norm": 687.5368041992188,
      "learning_rate": 3.9600995555016715e-05,
      "loss": 102.9218,
      "step": 39330
    },
    {
      "epoch": 0.1589385779562616,
      "grad_norm": 1048.0631103515625,
      "learning_rate": 3.960044034313429e-05,
      "loss": 79.893,
      "step": 39340
    },
    {
      "epoch": 0.15897897922162924,
      "grad_norm": 1082.3621826171875,
      "learning_rate": 3.9599884749131195e-05,
      "loss": 114.1062,
      "step": 39350
    },
    {
      "epoch": 0.15901938048699685,
      "grad_norm": 1691.8948974609375,
      "learning_rate": 3.9599328773018255e-05,
      "loss": 106.8196,
      "step": 39360
    },
    {
      "epoch": 0.1590597817523645,
      "grad_norm": 1052.7442626953125,
      "learning_rate": 3.9598772414806306e-05,
      "loss": 109.5173,
      "step": 39370
    },
    {
      "epoch": 0.15910018301773213,
      "grad_norm": 1856.539306640625,
      "learning_rate": 3.95982156745062e-05,
      "loss": 64.213,
      "step": 39380
    },
    {
      "epoch": 0.15914058428309974,
      "grad_norm": 652.4766845703125,
      "learning_rate": 3.95976585521288e-05,
      "loss": 99.9336,
      "step": 39390
    },
    {
      "epoch": 0.15918098554846738,
      "grad_norm": 677.7192993164062,
      "learning_rate": 3.959710104768494e-05,
      "loss": 85.7536,
      "step": 39400
    },
    {
      "epoch": 0.15922138681383502,
      "grad_norm": 232.5833740234375,
      "learning_rate": 3.9596543161185515e-05,
      "loss": 93.3603,
      "step": 39410
    },
    {
      "epoch": 0.15926178807920263,
      "grad_norm": 1655.010986328125,
      "learning_rate": 3.959598489264139e-05,
      "loss": 82.8985,
      "step": 39420
    },
    {
      "epoch": 0.15930218934457027,
      "grad_norm": 1230.93603515625,
      "learning_rate": 3.959542624206346e-05,
      "loss": 121.071,
      "step": 39430
    },
    {
      "epoch": 0.1593425906099379,
      "grad_norm": 397.9216003417969,
      "learning_rate": 3.9594867209462594e-05,
      "loss": 82.405,
      "step": 39440
    },
    {
      "epoch": 0.15938299187530552,
      "grad_norm": 816.2092895507812,
      "learning_rate": 3.959430779484971e-05,
      "loss": 98.675,
      "step": 39450
    },
    {
      "epoch": 0.15942339314067316,
      "grad_norm": 953.708740234375,
      "learning_rate": 3.9593747998235696e-05,
      "loss": 118.3295,
      "step": 39460
    },
    {
      "epoch": 0.1594637944060408,
      "grad_norm": 1187.2166748046875,
      "learning_rate": 3.9593187819631496e-05,
      "loss": 77.116,
      "step": 39470
    },
    {
      "epoch": 0.15950419567140842,
      "grad_norm": 1949.8765869140625,
      "learning_rate": 3.9592627259048e-05,
      "loss": 94.8703,
      "step": 39480
    },
    {
      "epoch": 0.15954459693677606,
      "grad_norm": 745.1921997070312,
      "learning_rate": 3.9592066316496155e-05,
      "loss": 76.3544,
      "step": 39490
    },
    {
      "epoch": 0.1595849982021437,
      "grad_norm": 570.1115112304688,
      "learning_rate": 3.959150499198688e-05,
      "loss": 69.0455,
      "step": 39500
    },
    {
      "epoch": 0.1596253994675113,
      "grad_norm": 1056.903076171875,
      "learning_rate": 3.9590943285531146e-05,
      "loss": 80.1547,
      "step": 39510
    },
    {
      "epoch": 0.15966580073287895,
      "grad_norm": 982.5983276367188,
      "learning_rate": 3.959038119713987e-05,
      "loss": 63.0331,
      "step": 39520
    },
    {
      "epoch": 0.1597062019982466,
      "grad_norm": 689.6868896484375,
      "learning_rate": 3.958981872682404e-05,
      "loss": 103.1534,
      "step": 39530
    },
    {
      "epoch": 0.15974660326361423,
      "grad_norm": 688.78564453125,
      "learning_rate": 3.95892558745946e-05,
      "loss": 77.986,
      "step": 39540
    },
    {
      "epoch": 0.15978700452898184,
      "grad_norm": 590.9727783203125,
      "learning_rate": 3.958869264046253e-05,
      "loss": 100.1267,
      "step": 39550
    },
    {
      "epoch": 0.15982740579434948,
      "grad_norm": 833.3731079101562,
      "learning_rate": 3.958812902443882e-05,
      "loss": 88.936,
      "step": 39560
    },
    {
      "epoch": 0.15986780705971712,
      "grad_norm": 529.066650390625,
      "learning_rate": 3.958756502653444e-05,
      "loss": 86.5306,
      "step": 39570
    },
    {
      "epoch": 0.15990820832508473,
      "grad_norm": 2491.70166015625,
      "learning_rate": 3.95870006467604e-05,
      "loss": 113.7662,
      "step": 39580
    },
    {
      "epoch": 0.15994860959045237,
      "grad_norm": 890.7277221679688,
      "learning_rate": 3.9586435885127705e-05,
      "loss": 109.6661,
      "step": 39590
    },
    {
      "epoch": 0.15998901085582,
      "grad_norm": 2090.965576171875,
      "learning_rate": 3.958587074164735e-05,
      "loss": 66.3457,
      "step": 39600
    },
    {
      "epoch": 0.16002941212118763,
      "grad_norm": 971.6329956054688,
      "learning_rate": 3.958530521633036e-05,
      "loss": 92.5265,
      "step": 39610
    },
    {
      "epoch": 0.16006981338655527,
      "grad_norm": 377.7248840332031,
      "learning_rate": 3.958473930918777e-05,
      "loss": 96.4538,
      "step": 39620
    },
    {
      "epoch": 0.1601102146519229,
      "grad_norm": 472.4529724121094,
      "learning_rate": 3.95841730202306e-05,
      "loss": 117.1553,
      "step": 39630
    },
    {
      "epoch": 0.16015061591729052,
      "grad_norm": 5134.74853515625,
      "learning_rate": 3.958360634946989e-05,
      "loss": 120.4135,
      "step": 39640
    },
    {
      "epoch": 0.16019101718265816,
      "grad_norm": 1356.9932861328125,
      "learning_rate": 3.9583039296916704e-05,
      "loss": 89.7805,
      "step": 39650
    },
    {
      "epoch": 0.1602314184480258,
      "grad_norm": 4443.1279296875,
      "learning_rate": 3.958247186258208e-05,
      "loss": 108.7569,
      "step": 39660
    },
    {
      "epoch": 0.1602718197133934,
      "grad_norm": 782.8939208984375,
      "learning_rate": 3.9581904046477076e-05,
      "loss": 78.7764,
      "step": 39670
    },
    {
      "epoch": 0.16031222097876105,
      "grad_norm": 736.0662841796875,
      "learning_rate": 3.958133584861278e-05,
      "loss": 76.8847,
      "step": 39680
    },
    {
      "epoch": 0.1603526222441287,
      "grad_norm": 458.4825744628906,
      "learning_rate": 3.958076726900026e-05,
      "loss": 77.1281,
      "step": 39690
    },
    {
      "epoch": 0.16039302350949633,
      "grad_norm": 786.83740234375,
      "learning_rate": 3.95801983076506e-05,
      "loss": 54.9982,
      "step": 39700
    },
    {
      "epoch": 0.16043342477486394,
      "grad_norm": 1158.18017578125,
      "learning_rate": 3.957962896457489e-05,
      "loss": 71.0925,
      "step": 39710
    },
    {
      "epoch": 0.16047382604023158,
      "grad_norm": 844.1268310546875,
      "learning_rate": 3.957905923978424e-05,
      "loss": 118.5867,
      "step": 39720
    },
    {
      "epoch": 0.16051422730559922,
      "grad_norm": 1144.5777587890625,
      "learning_rate": 3.9578489133289745e-05,
      "loss": 75.6427,
      "step": 39730
    },
    {
      "epoch": 0.16055462857096683,
      "grad_norm": 2612.0439453125,
      "learning_rate": 3.9577918645102524e-05,
      "loss": 96.4201,
      "step": 39740
    },
    {
      "epoch": 0.16059502983633447,
      "grad_norm": 1410.67724609375,
      "learning_rate": 3.9577347775233705e-05,
      "loss": 108.2059,
      "step": 39750
    },
    {
      "epoch": 0.16063543110170211,
      "grad_norm": 730.7840576171875,
      "learning_rate": 3.957677652369441e-05,
      "loss": 75.2652,
      "step": 39760
    },
    {
      "epoch": 0.16067583236706973,
      "grad_norm": 1091.61376953125,
      "learning_rate": 3.957620489049577e-05,
      "loss": 109.8948,
      "step": 39770
    },
    {
      "epoch": 0.16071623363243737,
      "grad_norm": 3437.56884765625,
      "learning_rate": 3.957563287564895e-05,
      "loss": 105.7026,
      "step": 39780
    },
    {
      "epoch": 0.160756634897805,
      "grad_norm": 1045.112548828125,
      "learning_rate": 3.957506047916508e-05,
      "loss": 54.9407,
      "step": 39790
    },
    {
      "epoch": 0.16079703616317262,
      "grad_norm": 2067.13623046875,
      "learning_rate": 3.9574487701055326e-05,
      "loss": 84.4046,
      "step": 39800
    },
    {
      "epoch": 0.16083743742854026,
      "grad_norm": 868.0408935546875,
      "learning_rate": 3.9573914541330865e-05,
      "loss": 75.3392,
      "step": 39810
    },
    {
      "epoch": 0.1608778386939079,
      "grad_norm": 946.1988525390625,
      "learning_rate": 3.957334100000286e-05,
      "loss": 79.963,
      "step": 39820
    },
    {
      "epoch": 0.1609182399592755,
      "grad_norm": 951.0245361328125,
      "learning_rate": 3.95727670770825e-05,
      "loss": 88.2257,
      "step": 39830
    },
    {
      "epoch": 0.16095864122464315,
      "grad_norm": 780.5066528320312,
      "learning_rate": 3.957219277258096e-05,
      "loss": 111.4322,
      "step": 39840
    },
    {
      "epoch": 0.1609990424900108,
      "grad_norm": 1591.3392333984375,
      "learning_rate": 3.957161808650944e-05,
      "loss": 66.2167,
      "step": 39850
    },
    {
      "epoch": 0.16103944375537843,
      "grad_norm": 658.369140625,
      "learning_rate": 3.957104301887916e-05,
      "loss": 58.5095,
      "step": 39860
    },
    {
      "epoch": 0.16107984502074604,
      "grad_norm": 904.9728393554688,
      "learning_rate": 3.957046756970132e-05,
      "loss": 98.7982,
      "step": 39870
    },
    {
      "epoch": 0.16112024628611368,
      "grad_norm": 531.2388916015625,
      "learning_rate": 3.9569891738987136e-05,
      "loss": 74.5442,
      "step": 39880
    },
    {
      "epoch": 0.16116064755148132,
      "grad_norm": 551.3512573242188,
      "learning_rate": 3.9569315526747843e-05,
      "loss": 86.7883,
      "step": 39890
    },
    {
      "epoch": 0.16120104881684894,
      "grad_norm": 638.0982055664062,
      "learning_rate": 3.9568738932994665e-05,
      "loss": 89.8119,
      "step": 39900
    },
    {
      "epoch": 0.16124145008221658,
      "grad_norm": 642.2186279296875,
      "learning_rate": 3.9568161957738844e-05,
      "loss": 107.5101,
      "step": 39910
    },
    {
      "epoch": 0.16128185134758422,
      "grad_norm": 563.9918212890625,
      "learning_rate": 3.9567584600991635e-05,
      "loss": 82.3985,
      "step": 39920
    },
    {
      "epoch": 0.16132225261295183,
      "grad_norm": 1653.6661376953125,
      "learning_rate": 3.9567006862764286e-05,
      "loss": 126.2073,
      "step": 39930
    },
    {
      "epoch": 0.16136265387831947,
      "grad_norm": 491.9596862792969,
      "learning_rate": 3.9566428743068074e-05,
      "loss": 78.9053,
      "step": 39940
    },
    {
      "epoch": 0.1614030551436871,
      "grad_norm": 1005.4078979492188,
      "learning_rate": 3.9565850241914246e-05,
      "loss": 117.193,
      "step": 39950
    },
    {
      "epoch": 0.16144345640905472,
      "grad_norm": 2084.00634765625,
      "learning_rate": 3.9565271359314107e-05,
      "loss": 105.7904,
      "step": 39960
    },
    {
      "epoch": 0.16148385767442236,
      "grad_norm": 1149.814453125,
      "learning_rate": 3.9564692095278924e-05,
      "loss": 133.7218,
      "step": 39970
    },
    {
      "epoch": 0.16152425893979,
      "grad_norm": 1500.60107421875,
      "learning_rate": 3.956411244981999e-05,
      "loss": 106.1676,
      "step": 39980
    },
    {
      "epoch": 0.1615646602051576,
      "grad_norm": 530.439208984375,
      "learning_rate": 3.9563532422948625e-05,
      "loss": 114.8235,
      "step": 39990
    },
    {
      "epoch": 0.16160506147052525,
      "grad_norm": 766.109619140625,
      "learning_rate": 3.9562952014676116e-05,
      "loss": 86.8458,
      "step": 40000
    },
    {
      "epoch": 0.1616454627358929,
      "grad_norm": 495.8179016113281,
      "learning_rate": 3.956237122501379e-05,
      "loss": 62.4111,
      "step": 40010
    },
    {
      "epoch": 0.16168586400126053,
      "grad_norm": 634.97412109375,
      "learning_rate": 3.956179005397296e-05,
      "loss": 82.6757,
      "step": 40020
    },
    {
      "epoch": 0.16172626526662814,
      "grad_norm": 928.5135498046875,
      "learning_rate": 3.956120850156496e-05,
      "loss": 76.7512,
      "step": 40030
    },
    {
      "epoch": 0.16176666653199578,
      "grad_norm": 909.52294921875,
      "learning_rate": 3.9560626567801136e-05,
      "loss": 79.9613,
      "step": 40040
    },
    {
      "epoch": 0.16180706779736342,
      "grad_norm": 980.2033081054688,
      "learning_rate": 3.9560044252692826e-05,
      "loss": 104.3447,
      "step": 40050
    },
    {
      "epoch": 0.16184746906273104,
      "grad_norm": 2851.51611328125,
      "learning_rate": 3.955946155625138e-05,
      "loss": 88.5099,
      "step": 40060
    },
    {
      "epoch": 0.16188787032809868,
      "grad_norm": 445.9934387207031,
      "learning_rate": 3.955887847848816e-05,
      "loss": 66.0215,
      "step": 40070
    },
    {
      "epoch": 0.16192827159346632,
      "grad_norm": 1758.5360107421875,
      "learning_rate": 3.9558295019414534e-05,
      "loss": 82.5284,
      "step": 40080
    },
    {
      "epoch": 0.16196867285883393,
      "grad_norm": 831.5187377929688,
      "learning_rate": 3.9557711179041887e-05,
      "loss": 70.7222,
      "step": 40090
    },
    {
      "epoch": 0.16200907412420157,
      "grad_norm": 2261.695556640625,
      "learning_rate": 3.955712695738158e-05,
      "loss": 78.2804,
      "step": 40100
    },
    {
      "epoch": 0.1620494753895692,
      "grad_norm": 720.5111694335938,
      "learning_rate": 3.955654235444502e-05,
      "loss": 74.9033,
      "step": 40110
    },
    {
      "epoch": 0.16208987665493682,
      "grad_norm": 484.18115234375,
      "learning_rate": 3.95559573702436e-05,
      "loss": 96.4327,
      "step": 40120
    },
    {
      "epoch": 0.16213027792030446,
      "grad_norm": 1393.7919921875,
      "learning_rate": 3.955537200478872e-05,
      "loss": 101.8992,
      "step": 40130
    },
    {
      "epoch": 0.1621706791856721,
      "grad_norm": 627.5079345703125,
      "learning_rate": 3.955478625809179e-05,
      "loss": 122.9893,
      "step": 40140
    },
    {
      "epoch": 0.1622110804510397,
      "grad_norm": 695.4521484375,
      "learning_rate": 3.955420013016424e-05,
      "loss": 81.3099,
      "step": 40150
    },
    {
      "epoch": 0.16225148171640735,
      "grad_norm": 1384.3963623046875,
      "learning_rate": 3.9553613621017495e-05,
      "loss": 112.9681,
      "step": 40160
    },
    {
      "epoch": 0.162291882981775,
      "grad_norm": 823.0811767578125,
      "learning_rate": 3.955302673066298e-05,
      "loss": 100.5564,
      "step": 40170
    },
    {
      "epoch": 0.16233228424714263,
      "grad_norm": 690.856201171875,
      "learning_rate": 3.955243945911214e-05,
      "loss": 89.7016,
      "step": 40180
    },
    {
      "epoch": 0.16237268551251025,
      "grad_norm": 1174.603759765625,
      "learning_rate": 3.955185180637643e-05,
      "loss": 79.1691,
      "step": 40190
    },
    {
      "epoch": 0.16241308677787789,
      "grad_norm": 641.5059814453125,
      "learning_rate": 3.955126377246731e-05,
      "loss": 106.7659,
      "step": 40200
    },
    {
      "epoch": 0.16245348804324553,
      "grad_norm": 3953.796630859375,
      "learning_rate": 3.955067535739623e-05,
      "loss": 97.4286,
      "step": 40210
    },
    {
      "epoch": 0.16249388930861314,
      "grad_norm": 560.7224731445312,
      "learning_rate": 3.955008656117467e-05,
      "loss": 83.4753,
      "step": 40220
    },
    {
      "epoch": 0.16253429057398078,
      "grad_norm": 577.6102294921875,
      "learning_rate": 3.9549497383814105e-05,
      "loss": 64.1252,
      "step": 40230
    },
    {
      "epoch": 0.16257469183934842,
      "grad_norm": 825.94677734375,
      "learning_rate": 3.954890782532602e-05,
      "loss": 80.1885,
      "step": 40240
    },
    {
      "epoch": 0.16261509310471603,
      "grad_norm": 973.9566040039062,
      "learning_rate": 3.9548317885721925e-05,
      "loss": 60.6026,
      "step": 40250
    },
    {
      "epoch": 0.16265549437008367,
      "grad_norm": 384.3602600097656,
      "learning_rate": 3.9547727565013295e-05,
      "loss": 78.6402,
      "step": 40260
    },
    {
      "epoch": 0.1626958956354513,
      "grad_norm": 817.73388671875,
      "learning_rate": 3.954713686321166e-05,
      "loss": 165.5563,
      "step": 40270
    },
    {
      "epoch": 0.16273629690081892,
      "grad_norm": 450.3658447265625,
      "learning_rate": 3.954654578032853e-05,
      "loss": 120.2511,
      "step": 40280
    },
    {
      "epoch": 0.16277669816618656,
      "grad_norm": 239.74362182617188,
      "learning_rate": 3.954595431637542e-05,
      "loss": 64.2798,
      "step": 40290
    },
    {
      "epoch": 0.1628170994315542,
      "grad_norm": 980.2670288085938,
      "learning_rate": 3.954536247136387e-05,
      "loss": 77.4414,
      "step": 40300
    },
    {
      "epoch": 0.16285750069692181,
      "grad_norm": 2538.709228515625,
      "learning_rate": 3.954477024530542e-05,
      "loss": 121.8563,
      "step": 40310
    },
    {
      "epoch": 0.16289790196228945,
      "grad_norm": 1247.4947509765625,
      "learning_rate": 3.954417763821161e-05,
      "loss": 80.4332,
      "step": 40320
    },
    {
      "epoch": 0.1629383032276571,
      "grad_norm": 730.9105834960938,
      "learning_rate": 3.9543584650093994e-05,
      "loss": 140.6749,
      "step": 40330
    },
    {
      "epoch": 0.16297870449302473,
      "grad_norm": 1664.7508544921875,
      "learning_rate": 3.954299128096413e-05,
      "loss": 109.0711,
      "step": 40340
    },
    {
      "epoch": 0.16301910575839235,
      "grad_norm": 770.0411987304688,
      "learning_rate": 3.95423975308336e-05,
      "loss": 62.409,
      "step": 40350
    },
    {
      "epoch": 0.16305950702376,
      "grad_norm": 733.1558227539062,
      "learning_rate": 3.9541803399713956e-05,
      "loss": 97.6732,
      "step": 40360
    },
    {
      "epoch": 0.16309990828912763,
      "grad_norm": 1060.8912353515625,
      "learning_rate": 3.9541208887616805e-05,
      "loss": 69.0565,
      "step": 40370
    },
    {
      "epoch": 0.16314030955449524,
      "grad_norm": 356.4098815917969,
      "learning_rate": 3.954061399455372e-05,
      "loss": 105.4898,
      "step": 40380
    },
    {
      "epoch": 0.16318071081986288,
      "grad_norm": 504.13482666015625,
      "learning_rate": 3.95400187205363e-05,
      "loss": 61.4338,
      "step": 40390
    },
    {
      "epoch": 0.16322111208523052,
      "grad_norm": 682.7866821289062,
      "learning_rate": 3.9539423065576165e-05,
      "loss": 115.8884,
      "step": 40400
    },
    {
      "epoch": 0.16326151335059813,
      "grad_norm": 737.2675170898438,
      "learning_rate": 3.9538827029684916e-05,
      "loss": 55.9793,
      "step": 40410
    },
    {
      "epoch": 0.16330191461596577,
      "grad_norm": 924.0701293945312,
      "learning_rate": 3.9538230612874174e-05,
      "loss": 103.0369,
      "step": 40420
    },
    {
      "epoch": 0.1633423158813334,
      "grad_norm": 1058.978759765625,
      "learning_rate": 3.953763381515556e-05,
      "loss": 103.6735,
      "step": 40430
    },
    {
      "epoch": 0.16338271714670102,
      "grad_norm": 1186.2822265625,
      "learning_rate": 3.953703663654072e-05,
      "loss": 106.1324,
      "step": 40440
    },
    {
      "epoch": 0.16342311841206866,
      "grad_norm": 825.0392456054688,
      "learning_rate": 3.95364390770413e-05,
      "loss": 81.3327,
      "step": 40450
    },
    {
      "epoch": 0.1634635196774363,
      "grad_norm": 584.6072998046875,
      "learning_rate": 3.9535841136668936e-05,
      "loss": 76.001,
      "step": 40460
    },
    {
      "epoch": 0.16350392094280392,
      "grad_norm": 915.310791015625,
      "learning_rate": 3.953524281543529e-05,
      "loss": 82.7482,
      "step": 40470
    },
    {
      "epoch": 0.16354432220817156,
      "grad_norm": 1052.021240234375,
      "learning_rate": 3.9534644113352036e-05,
      "loss": 86.3224,
      "step": 40480
    },
    {
      "epoch": 0.1635847234735392,
      "grad_norm": 688.9476928710938,
      "learning_rate": 3.953404503043083e-05,
      "loss": 79.8042,
      "step": 40490
    },
    {
      "epoch": 0.16362512473890684,
      "grad_norm": 2522.06787109375,
      "learning_rate": 3.9533445566683364e-05,
      "loss": 120.3851,
      "step": 40500
    },
    {
      "epoch": 0.16366552600427445,
      "grad_norm": 531.6077880859375,
      "learning_rate": 3.9532845722121315e-05,
      "loss": 41.9876,
      "step": 40510
    },
    {
      "epoch": 0.1637059272696421,
      "grad_norm": 696.3869018554688,
      "learning_rate": 3.953224549675638e-05,
      "loss": 94.6229,
      "step": 40520
    },
    {
      "epoch": 0.16374632853500973,
      "grad_norm": 724.3345336914062,
      "learning_rate": 3.9531644890600276e-05,
      "loss": 99.2432,
      "step": 40530
    },
    {
      "epoch": 0.16378672980037734,
      "grad_norm": 1059.6839599609375,
      "learning_rate": 3.953104390366469e-05,
      "loss": 71.4181,
      "step": 40540
    },
    {
      "epoch": 0.16382713106574498,
      "grad_norm": 774.9991455078125,
      "learning_rate": 3.953044253596135e-05,
      "loss": 80.0983,
      "step": 40550
    },
    {
      "epoch": 0.16386753233111262,
      "grad_norm": 491.90277099609375,
      "learning_rate": 3.952984078750198e-05,
      "loss": 90.0716,
      "step": 40560
    },
    {
      "epoch": 0.16390793359648023,
      "grad_norm": 814.5154418945312,
      "learning_rate": 3.9529238658298304e-05,
      "loss": 81.0831,
      "step": 40570
    },
    {
      "epoch": 0.16394833486184787,
      "grad_norm": 980.4241943359375,
      "learning_rate": 3.952863614836207e-05,
      "loss": 62.4222,
      "step": 40580
    },
    {
      "epoch": 0.1639887361272155,
      "grad_norm": 2784.486572265625,
      "learning_rate": 3.952803325770501e-05,
      "loss": 108.2839,
      "step": 40590
    },
    {
      "epoch": 0.16402913739258312,
      "grad_norm": 1452.8868408203125,
      "learning_rate": 3.95274299863389e-05,
      "loss": 82.7274,
      "step": 40600
    },
    {
      "epoch": 0.16406953865795076,
      "grad_norm": 523.1763916015625,
      "learning_rate": 3.952682633427548e-05,
      "loss": 72.2552,
      "step": 40610
    },
    {
      "epoch": 0.1641099399233184,
      "grad_norm": 1814.376220703125,
      "learning_rate": 3.952622230152654e-05,
      "loss": 104.2498,
      "step": 40620
    },
    {
      "epoch": 0.16415034118868602,
      "grad_norm": 1456.19140625,
      "learning_rate": 3.952561788810384e-05,
      "loss": 111.9931,
      "step": 40630
    },
    {
      "epoch": 0.16419074245405366,
      "grad_norm": 418.4021911621094,
      "learning_rate": 3.952501309401916e-05,
      "loss": 82.0589,
      "step": 40640
    },
    {
      "epoch": 0.1642311437194213,
      "grad_norm": 1160.2156982421875,
      "learning_rate": 3.95244079192843e-05,
      "loss": 131.2082,
      "step": 40650
    },
    {
      "epoch": 0.16427154498478894,
      "grad_norm": 979.8446044921875,
      "learning_rate": 3.952380236391106e-05,
      "loss": 89.2304,
      "step": 40660
    },
    {
      "epoch": 0.16431194625015655,
      "grad_norm": 1009.5330200195312,
      "learning_rate": 3.952319642791124e-05,
      "loss": 54.3505,
      "step": 40670
    },
    {
      "epoch": 0.1643523475155242,
      "grad_norm": 515.9607543945312,
      "learning_rate": 3.9522590111296646e-05,
      "loss": 86.7991,
      "step": 40680
    },
    {
      "epoch": 0.16439274878089183,
      "grad_norm": 958.9117431640625,
      "learning_rate": 3.952198341407911e-05,
      "loss": 94.9694,
      "step": 40690
    },
    {
      "epoch": 0.16443315004625944,
      "grad_norm": 973.7232055664062,
      "learning_rate": 3.9521376336270466e-05,
      "loss": 101.5517,
      "step": 40700
    },
    {
      "epoch": 0.16447355131162708,
      "grad_norm": 2472.23193359375,
      "learning_rate": 3.952076887788253e-05,
      "loss": 113.2867,
      "step": 40710
    },
    {
      "epoch": 0.16451395257699472,
      "grad_norm": 1117.849609375,
      "learning_rate": 3.952016103892716e-05,
      "loss": 63.5003,
      "step": 40720
    },
    {
      "epoch": 0.16455435384236233,
      "grad_norm": 427.4268493652344,
      "learning_rate": 3.95195528194162e-05,
      "loss": 112.6997,
      "step": 40730
    },
    {
      "epoch": 0.16459475510772997,
      "grad_norm": 227.0601806640625,
      "learning_rate": 3.951894421936151e-05,
      "loss": 82.3097,
      "step": 40740
    },
    {
      "epoch": 0.1646351563730976,
      "grad_norm": 3456.140625,
      "learning_rate": 3.951833523877495e-05,
      "loss": 101.9074,
      "step": 40750
    },
    {
      "epoch": 0.16467555763846523,
      "grad_norm": 1029.93603515625,
      "learning_rate": 3.95177258776684e-05,
      "loss": 80.9789,
      "step": 40760
    },
    {
      "epoch": 0.16471595890383287,
      "grad_norm": 983.2781372070312,
      "learning_rate": 3.951711613605374e-05,
      "loss": 66.6682,
      "step": 40770
    },
    {
      "epoch": 0.1647563601692005,
      "grad_norm": 1014.9578247070312,
      "learning_rate": 3.9516506013942836e-05,
      "loss": 89.8433,
      "step": 40780
    },
    {
      "epoch": 0.16479676143456812,
      "grad_norm": 900.0442504882812,
      "learning_rate": 3.951589551134761e-05,
      "loss": 98.0265,
      "step": 40790
    },
    {
      "epoch": 0.16483716269993576,
      "grad_norm": 759.8152465820312,
      "learning_rate": 3.9515284628279954e-05,
      "loss": 50.9888,
      "step": 40800
    },
    {
      "epoch": 0.1648775639653034,
      "grad_norm": 822.179443359375,
      "learning_rate": 3.9514673364751776e-05,
      "loss": 107.9123,
      "step": 40810
    },
    {
      "epoch": 0.16491796523067104,
      "grad_norm": 710.6921997070312,
      "learning_rate": 3.9514061720775e-05,
      "loss": 117.4759,
      "step": 40820
    },
    {
      "epoch": 0.16495836649603865,
      "grad_norm": 818.1751098632812,
      "learning_rate": 3.9513449696361535e-05,
      "loss": 80.6732,
      "step": 40830
    },
    {
      "epoch": 0.1649987677614063,
      "grad_norm": 1064.0673828125,
      "learning_rate": 3.951283729152332e-05,
      "loss": 76.4911,
      "step": 40840
    },
    {
      "epoch": 0.16503916902677393,
      "grad_norm": 662.7578735351562,
      "learning_rate": 3.951222450627231e-05,
      "loss": 48.1669,
      "step": 40850
    },
    {
      "epoch": 0.16507957029214154,
      "grad_norm": 545.164794921875,
      "learning_rate": 3.951161134062042e-05,
      "loss": 90.6871,
      "step": 40860
    },
    {
      "epoch": 0.16511997155750918,
      "grad_norm": 1219.935302734375,
      "learning_rate": 3.951099779457963e-05,
      "loss": 56.5449,
      "step": 40870
    },
    {
      "epoch": 0.16516037282287682,
      "grad_norm": 988.223876953125,
      "learning_rate": 3.95103838681619e-05,
      "loss": 90.2754,
      "step": 40880
    },
    {
      "epoch": 0.16520077408824443,
      "grad_norm": 719.2081298828125,
      "learning_rate": 3.9509769561379184e-05,
      "loss": 89.6676,
      "step": 40890
    },
    {
      "epoch": 0.16524117535361207,
      "grad_norm": 874.89404296875,
      "learning_rate": 3.9509154874243466e-05,
      "loss": 59.4812,
      "step": 40900
    },
    {
      "epoch": 0.16528157661897971,
      "grad_norm": 701.1567993164062,
      "learning_rate": 3.950853980676673e-05,
      "loss": 98.7121,
      "step": 40910
    },
    {
      "epoch": 0.16532197788434733,
      "grad_norm": 1197.994140625,
      "learning_rate": 3.950792435896097e-05,
      "loss": 112.145,
      "step": 40920
    },
    {
      "epoch": 0.16536237914971497,
      "grad_norm": 4444.24267578125,
      "learning_rate": 3.950730853083818e-05,
      "loss": 104.277,
      "step": 40930
    },
    {
      "epoch": 0.1654027804150826,
      "grad_norm": 1147.2232666015625,
      "learning_rate": 3.950669232241036e-05,
      "loss": 132.6901,
      "step": 40940
    },
    {
      "epoch": 0.16544318168045022,
      "grad_norm": 1110.708984375,
      "learning_rate": 3.950607573368954e-05,
      "loss": 59.1241,
      "step": 40950
    },
    {
      "epoch": 0.16548358294581786,
      "grad_norm": 1008.7105712890625,
      "learning_rate": 3.950545876468773e-05,
      "loss": 75.2638,
      "step": 40960
    },
    {
      "epoch": 0.1655239842111855,
      "grad_norm": 2469.8623046875,
      "learning_rate": 3.9504841415416955e-05,
      "loss": 85.3266,
      "step": 40970
    },
    {
      "epoch": 0.16556438547655314,
      "grad_norm": 1274.08447265625,
      "learning_rate": 3.950422368588926e-05,
      "loss": 73.8704,
      "step": 40980
    },
    {
      "epoch": 0.16560478674192075,
      "grad_norm": 1128.9522705078125,
      "learning_rate": 3.950360557611668e-05,
      "loss": 73.6916,
      "step": 40990
    },
    {
      "epoch": 0.1656451880072884,
      "grad_norm": 491.25958251953125,
      "learning_rate": 3.950298708611127e-05,
      "loss": 76.6751,
      "step": 41000
    },
    {
      "epoch": 0.16568558927265603,
      "grad_norm": 625.8621826171875,
      "learning_rate": 3.950236821588508e-05,
      "loss": 80.7306,
      "step": 41010
    },
    {
      "epoch": 0.16572599053802364,
      "grad_norm": 629.7647094726562,
      "learning_rate": 3.9501748965450186e-05,
      "loss": 75.1716,
      "step": 41020
    },
    {
      "epoch": 0.16576639180339128,
      "grad_norm": 1378.2105712890625,
      "learning_rate": 3.950112933481866e-05,
      "loss": 73.2715,
      "step": 41030
    },
    {
      "epoch": 0.16580679306875892,
      "grad_norm": 571.427734375,
      "learning_rate": 3.950050932400257e-05,
      "loss": 91.563,
      "step": 41040
    },
    {
      "epoch": 0.16584719433412654,
      "grad_norm": 890.5443115234375,
      "learning_rate": 3.949988893301401e-05,
      "loss": 125.3012,
      "step": 41050
    },
    {
      "epoch": 0.16588759559949418,
      "grad_norm": 1563.917236328125,
      "learning_rate": 3.9499268161865085e-05,
      "loss": 65.5101,
      "step": 41060
    },
    {
      "epoch": 0.16592799686486182,
      "grad_norm": 614.747802734375,
      "learning_rate": 3.949864701056788e-05,
      "loss": 87.1772,
      "step": 41070
    },
    {
      "epoch": 0.16596839813022943,
      "grad_norm": 642.1237182617188,
      "learning_rate": 3.9498025479134516e-05,
      "loss": 77.2061,
      "step": 41080
    },
    {
      "epoch": 0.16600879939559707,
      "grad_norm": 613.6666259765625,
      "learning_rate": 3.9497403567577114e-05,
      "loss": 115.0317,
      "step": 41090
    },
    {
      "epoch": 0.1660492006609647,
      "grad_norm": 468.08660888671875,
      "learning_rate": 3.949678127590778e-05,
      "loss": 74.2069,
      "step": 41100
    },
    {
      "epoch": 0.16608960192633232,
      "grad_norm": 513.4039306640625,
      "learning_rate": 3.949615860413866e-05,
      "loss": 60.0802,
      "step": 41110
    },
    {
      "epoch": 0.16613000319169996,
      "grad_norm": 931.3645629882812,
      "learning_rate": 3.94955355522819e-05,
      "loss": 143.4764,
      "step": 41120
    },
    {
      "epoch": 0.1661704044570676,
      "grad_norm": 2047.0240478515625,
      "learning_rate": 3.9494912120349626e-05,
      "loss": 55.3324,
      "step": 41130
    },
    {
      "epoch": 0.16621080572243524,
      "grad_norm": 533.386474609375,
      "learning_rate": 3.949428830835401e-05,
      "loss": 55.0412,
      "step": 41140
    },
    {
      "epoch": 0.16625120698780285,
      "grad_norm": 620.4298095703125,
      "learning_rate": 3.9493664116307204e-05,
      "loss": 95.4496,
      "step": 41150
    },
    {
      "epoch": 0.1662916082531705,
      "grad_norm": 641.894287109375,
      "learning_rate": 3.9493039544221375e-05,
      "loss": 103.0947,
      "step": 41160
    },
    {
      "epoch": 0.16633200951853813,
      "grad_norm": 677.8861694335938,
      "learning_rate": 3.949241459210871e-05,
      "loss": 66.4296,
      "step": 41170
    },
    {
      "epoch": 0.16637241078390574,
      "grad_norm": 683.670654296875,
      "learning_rate": 3.949178925998139e-05,
      "loss": 123.1877,
      "step": 41180
    },
    {
      "epoch": 0.16641281204927338,
      "grad_norm": 448.2669372558594,
      "learning_rate": 3.9491163547851604e-05,
      "loss": 79.1134,
      "step": 41190
    },
    {
      "epoch": 0.16645321331464102,
      "grad_norm": 6343.84619140625,
      "learning_rate": 3.949053745573155e-05,
      "loss": 127.7989,
      "step": 41200
    },
    {
      "epoch": 0.16649361458000864,
      "grad_norm": 1064.6607666015625,
      "learning_rate": 3.9489910983633426e-05,
      "loss": 114.2296,
      "step": 41210
    },
    {
      "epoch": 0.16653401584537628,
      "grad_norm": 980.7378540039062,
      "learning_rate": 3.9489284131569456e-05,
      "loss": 85.9728,
      "step": 41220
    },
    {
      "epoch": 0.16657441711074392,
      "grad_norm": 480.8191833496094,
      "learning_rate": 3.948865689955186e-05,
      "loss": 71.3556,
      "step": 41230
    },
    {
      "epoch": 0.16661481837611153,
      "grad_norm": 1177.986083984375,
      "learning_rate": 3.948802928759287e-05,
      "loss": 83.3197,
      "step": 41240
    },
    {
      "epoch": 0.16665521964147917,
      "grad_norm": 865.0421752929688,
      "learning_rate": 3.948740129570471e-05,
      "loss": 76.694,
      "step": 41250
    },
    {
      "epoch": 0.1666956209068468,
      "grad_norm": 835.8969116210938,
      "learning_rate": 3.948677292389963e-05,
      "loss": 74.6189,
      "step": 41260
    },
    {
      "epoch": 0.16673602217221442,
      "grad_norm": 474.4497375488281,
      "learning_rate": 3.948614417218988e-05,
      "loss": 81.0856,
      "step": 41270
    },
    {
      "epoch": 0.16677642343758206,
      "grad_norm": 2981.888671875,
      "learning_rate": 3.948551504058771e-05,
      "loss": 175.6042,
      "step": 41280
    },
    {
      "epoch": 0.1668168247029497,
      "grad_norm": 615.6951904296875,
      "learning_rate": 3.94848855291054e-05,
      "loss": 89.9799,
      "step": 41290
    },
    {
      "epoch": 0.16685722596831734,
      "grad_norm": 1236.52587890625,
      "learning_rate": 3.948425563775521e-05,
      "loss": 110.5213,
      "step": 41300
    },
    {
      "epoch": 0.16689762723368495,
      "grad_norm": 926.8430786132812,
      "learning_rate": 3.948362536654943e-05,
      "loss": 81.6385,
      "step": 41310
    },
    {
      "epoch": 0.1669380284990526,
      "grad_norm": 606.0703125,
      "learning_rate": 3.948299471550034e-05,
      "loss": 82.6857,
      "step": 41320
    },
    {
      "epoch": 0.16697842976442023,
      "grad_norm": 1497.1026611328125,
      "learning_rate": 3.9482363684620247e-05,
      "loss": 109.4494,
      "step": 41330
    },
    {
      "epoch": 0.16701883102978785,
      "grad_norm": 1270.447021484375,
      "learning_rate": 3.9481732273921435e-05,
      "loss": 101.7404,
      "step": 41340
    },
    {
      "epoch": 0.16705923229515549,
      "grad_norm": 1176.3333740234375,
      "learning_rate": 3.948110048341622e-05,
      "loss": 92.2997,
      "step": 41350
    },
    {
      "epoch": 0.16709963356052313,
      "grad_norm": 2200.978759765625,
      "learning_rate": 3.9480468313116925e-05,
      "loss": 121.3717,
      "step": 41360
    },
    {
      "epoch": 0.16714003482589074,
      "grad_norm": 577.7105712890625,
      "learning_rate": 3.947983576303587e-05,
      "loss": 74.5596,
      "step": 41370
    },
    {
      "epoch": 0.16718043609125838,
      "grad_norm": 949.1942749023438,
      "learning_rate": 3.947920283318539e-05,
      "loss": 83.1184,
      "step": 41380
    },
    {
      "epoch": 0.16722083735662602,
      "grad_norm": 1375.0670166015625,
      "learning_rate": 3.947856952357782e-05,
      "loss": 116.7619,
      "step": 41390
    },
    {
      "epoch": 0.16726123862199363,
      "grad_norm": 467.704345703125,
      "learning_rate": 3.9477935834225503e-05,
      "loss": 96.9709,
      "step": 41400
    },
    {
      "epoch": 0.16730163988736127,
      "grad_norm": 944.2130126953125,
      "learning_rate": 3.947730176514081e-05,
      "loss": 82.2528,
      "step": 41410
    },
    {
      "epoch": 0.1673420411527289,
      "grad_norm": 791.8900756835938,
      "learning_rate": 3.947666731633609e-05,
      "loss": 71.4929,
      "step": 41420
    },
    {
      "epoch": 0.16738244241809652,
      "grad_norm": 557.81884765625,
      "learning_rate": 3.947603248782371e-05,
      "loss": 76.7843,
      "step": 41430
    },
    {
      "epoch": 0.16742284368346416,
      "grad_norm": 681.4508666992188,
      "learning_rate": 3.947539727961605e-05,
      "loss": 91.5363,
      "step": 41440
    },
    {
      "epoch": 0.1674632449488318,
      "grad_norm": 1442.1290283203125,
      "learning_rate": 3.947476169172549e-05,
      "loss": 93.9871,
      "step": 41450
    },
    {
      "epoch": 0.16750364621419944,
      "grad_norm": 483.75927734375,
      "learning_rate": 3.947412572416443e-05,
      "loss": 65.8295,
      "step": 41460
    },
    {
      "epoch": 0.16754404747956705,
      "grad_norm": 5170.841796875,
      "learning_rate": 3.947348937694526e-05,
      "loss": 119.9971,
      "step": 41470
    },
    {
      "epoch": 0.1675844487449347,
      "grad_norm": 0.0,
      "learning_rate": 3.947285265008039e-05,
      "loss": 72.5246,
      "step": 41480
    },
    {
      "epoch": 0.16762485001030233,
      "grad_norm": 832.8370361328125,
      "learning_rate": 3.9472215543582234e-05,
      "loss": 66.7058,
      "step": 41490
    },
    {
      "epoch": 0.16766525127566995,
      "grad_norm": 1178.047607421875,
      "learning_rate": 3.9471578057463206e-05,
      "loss": 82.1243,
      "step": 41500
    },
    {
      "epoch": 0.1677056525410376,
      "grad_norm": 1412.372314453125,
      "learning_rate": 3.9470940191735745e-05,
      "loss": 69.2213,
      "step": 41510
    },
    {
      "epoch": 0.16774605380640523,
      "grad_norm": 945.0978393554688,
      "learning_rate": 3.947030194641228e-05,
      "loss": 79.4089,
      "step": 41520
    },
    {
      "epoch": 0.16778645507177284,
      "grad_norm": 1217.2481689453125,
      "learning_rate": 3.946966332150525e-05,
      "loss": 116.153,
      "step": 41530
    },
    {
      "epoch": 0.16782685633714048,
      "grad_norm": 883.7046508789062,
      "learning_rate": 3.9469024317027115e-05,
      "loss": 78.1752,
      "step": 41540
    },
    {
      "epoch": 0.16786725760250812,
      "grad_norm": 528.1052856445312,
      "learning_rate": 3.9468384932990324e-05,
      "loss": 93.4256,
      "step": 41550
    },
    {
      "epoch": 0.16790765886787573,
      "grad_norm": 972.721923828125,
      "learning_rate": 3.9467745169407346e-05,
      "loss": 104.534,
      "step": 41560
    },
    {
      "epoch": 0.16794806013324337,
      "grad_norm": 2014.378173828125,
      "learning_rate": 3.946710502629065e-05,
      "loss": 113.3028,
      "step": 41570
    },
    {
      "epoch": 0.167988461398611,
      "grad_norm": 585.617431640625,
      "learning_rate": 3.946646450365273e-05,
      "loss": 97.4658,
      "step": 41580
    },
    {
      "epoch": 0.16802886266397862,
      "grad_norm": 1128.84521484375,
      "learning_rate": 3.9465823601506055e-05,
      "loss": 83.0689,
      "step": 41590
    },
    {
      "epoch": 0.16806926392934626,
      "grad_norm": 1114.514404296875,
      "learning_rate": 3.946518231986313e-05,
      "loss": 78.9807,
      "step": 41600
    },
    {
      "epoch": 0.1681096651947139,
      "grad_norm": 738.8482055664062,
      "learning_rate": 3.946454065873645e-05,
      "loss": 112.8025,
      "step": 41610
    },
    {
      "epoch": 0.16815006646008154,
      "grad_norm": 786.415771484375,
      "learning_rate": 3.946389861813854e-05,
      "loss": 84.9417,
      "step": 41620
    },
    {
      "epoch": 0.16819046772544916,
      "grad_norm": 293.24774169921875,
      "learning_rate": 3.946325619808189e-05,
      "loss": 89.6034,
      "step": 41630
    },
    {
      "epoch": 0.1682308689908168,
      "grad_norm": 860.4002075195312,
      "learning_rate": 3.9462613398579044e-05,
      "loss": 70.1078,
      "step": 41640
    },
    {
      "epoch": 0.16827127025618444,
      "grad_norm": 470.266357421875,
      "learning_rate": 3.9461970219642535e-05,
      "loss": 98.0849,
      "step": 41650
    },
    {
      "epoch": 0.16831167152155205,
      "grad_norm": 240.69097900390625,
      "learning_rate": 3.946132666128489e-05,
      "loss": 78.9933,
      "step": 41660
    },
    {
      "epoch": 0.1683520727869197,
      "grad_norm": 681.0687866210938,
      "learning_rate": 3.946068272351867e-05,
      "loss": 108.567,
      "step": 41670
    },
    {
      "epoch": 0.16839247405228733,
      "grad_norm": 1583.38330078125,
      "learning_rate": 3.946003840635642e-05,
      "loss": 100.0662,
      "step": 41680
    },
    {
      "epoch": 0.16843287531765494,
      "grad_norm": 1083.94677734375,
      "learning_rate": 3.94593937098107e-05,
      "loss": 86.575,
      "step": 41690
    },
    {
      "epoch": 0.16847327658302258,
      "grad_norm": 556.242431640625,
      "learning_rate": 3.945874863389408e-05,
      "loss": 69.5942,
      "step": 41700
    },
    {
      "epoch": 0.16851367784839022,
      "grad_norm": 749.9948120117188,
      "learning_rate": 3.9458103178619146e-05,
      "loss": 91.4834,
      "step": 41710
    },
    {
      "epoch": 0.16855407911375783,
      "grad_norm": 626.6065063476562,
      "learning_rate": 3.945745734399846e-05,
      "loss": 103.6504,
      "step": 41720
    },
    {
      "epoch": 0.16859448037912547,
      "grad_norm": 762.7034301757812,
      "learning_rate": 3.945681113004463e-05,
      "loss": 129.9065,
      "step": 41730
    },
    {
      "epoch": 0.1686348816444931,
      "grad_norm": 456.9073486328125,
      "learning_rate": 3.945616453677025e-05,
      "loss": 71.9278,
      "step": 41740
    },
    {
      "epoch": 0.16867528290986072,
      "grad_norm": 1217.7987060546875,
      "learning_rate": 3.945551756418794e-05,
      "loss": 99.0016,
      "step": 41750
    },
    {
      "epoch": 0.16871568417522836,
      "grad_norm": 983.9970092773438,
      "learning_rate": 3.945487021231028e-05,
      "loss": 82.6868,
      "step": 41760
    },
    {
      "epoch": 0.168756085440596,
      "grad_norm": 614.2739868164062,
      "learning_rate": 3.9454222481149916e-05,
      "loss": 69.2636,
      "step": 41770
    },
    {
      "epoch": 0.16879648670596364,
      "grad_norm": 797.482421875,
      "learning_rate": 3.945357437071947e-05,
      "loss": 88.3623,
      "step": 41780
    },
    {
      "epoch": 0.16883688797133126,
      "grad_norm": 1216.2550048828125,
      "learning_rate": 3.9452925881031574e-05,
      "loss": 69.5762,
      "step": 41790
    },
    {
      "epoch": 0.1688772892366989,
      "grad_norm": 706.0321044921875,
      "learning_rate": 3.9452277012098875e-05,
      "loss": 83.3594,
      "step": 41800
    },
    {
      "epoch": 0.16891769050206654,
      "grad_norm": 888.8679809570312,
      "learning_rate": 3.945162776393402e-05,
      "loss": 84.315,
      "step": 41810
    },
    {
      "epoch": 0.16895809176743415,
      "grad_norm": 846.63037109375,
      "learning_rate": 3.9450978136549665e-05,
      "loss": 62.6609,
      "step": 41820
    },
    {
      "epoch": 0.1689984930328018,
      "grad_norm": 588.7283935546875,
      "learning_rate": 3.9450328129958484e-05,
      "loss": 88.6878,
      "step": 41830
    },
    {
      "epoch": 0.16903889429816943,
      "grad_norm": 510.836669921875,
      "learning_rate": 3.9449677744173135e-05,
      "loss": 80.6699,
      "step": 41840
    },
    {
      "epoch": 0.16907929556353704,
      "grad_norm": 524.7201538085938,
      "learning_rate": 3.9449026979206305e-05,
      "loss": 58.7747,
      "step": 41850
    },
    {
      "epoch": 0.16911969682890468,
      "grad_norm": 9646.47265625,
      "learning_rate": 3.9448375835070685e-05,
      "loss": 143.0488,
      "step": 41860
    },
    {
      "epoch": 0.16916009809427232,
      "grad_norm": 1948.435302734375,
      "learning_rate": 3.944772431177896e-05,
      "loss": 85.102,
      "step": 41870
    },
    {
      "epoch": 0.16920049935963993,
      "grad_norm": 809.4441528320312,
      "learning_rate": 3.9447072409343844e-05,
      "loss": 82.9807,
      "step": 41880
    },
    {
      "epoch": 0.16924090062500757,
      "grad_norm": 711.6243896484375,
      "learning_rate": 3.944642012777804e-05,
      "loss": 77.7331,
      "step": 41890
    },
    {
      "epoch": 0.1692813018903752,
      "grad_norm": 450.212646484375,
      "learning_rate": 3.9445767467094256e-05,
      "loss": 77.5103,
      "step": 41900
    },
    {
      "epoch": 0.16932170315574283,
      "grad_norm": 1139.9300537109375,
      "learning_rate": 3.944511442730523e-05,
      "loss": 127.2431,
      "step": 41910
    },
    {
      "epoch": 0.16936210442111047,
      "grad_norm": 681.1736450195312,
      "learning_rate": 3.9444461008423687e-05,
      "loss": 63.0341,
      "step": 41920
    },
    {
      "epoch": 0.1694025056864781,
      "grad_norm": 794.4819946289062,
      "learning_rate": 3.944380721046236e-05,
      "loss": 75.632,
      "step": 41930
    },
    {
      "epoch": 0.16944290695184575,
      "grad_norm": 1029.8173828125,
      "learning_rate": 3.944315303343401e-05,
      "loss": 76.0773,
      "step": 41940
    },
    {
      "epoch": 0.16948330821721336,
      "grad_norm": 661.2575073242188,
      "learning_rate": 3.9442498477351376e-05,
      "loss": 117.2325,
      "step": 41950
    },
    {
      "epoch": 0.169523709482581,
      "grad_norm": 383.8749084472656,
      "learning_rate": 3.944184354222722e-05,
      "loss": 80.0125,
      "step": 41960
    },
    {
      "epoch": 0.16956411074794864,
      "grad_norm": 3768.85986328125,
      "learning_rate": 3.9441188228074326e-05,
      "loss": 113.0336,
      "step": 41970
    },
    {
      "epoch": 0.16960451201331625,
      "grad_norm": 969.9005737304688,
      "learning_rate": 3.944053253490546e-05,
      "loss": 75.1191,
      "step": 41980
    },
    {
      "epoch": 0.1696449132786839,
      "grad_norm": 759.4833984375,
      "learning_rate": 3.943987646273339e-05,
      "loss": 120.5944,
      "step": 41990
    },
    {
      "epoch": 0.16968531454405153,
      "grad_norm": 414.0032958984375,
      "learning_rate": 3.943922001157093e-05,
      "loss": 61.2086,
      "step": 42000
    },
    {
      "epoch": 0.16972571580941914,
      "grad_norm": 577.0447387695312,
      "learning_rate": 3.9438563181430863e-05,
      "loss": 66.7768,
      "step": 42010
    },
    {
      "epoch": 0.16976611707478678,
      "grad_norm": 462.7989501953125,
      "learning_rate": 3.9437905972326e-05,
      "loss": 49.0065,
      "step": 42020
    },
    {
      "epoch": 0.16980651834015442,
      "grad_norm": 2454.510986328125,
      "learning_rate": 3.9437248384269155e-05,
      "loss": 74.5437,
      "step": 42030
    },
    {
      "epoch": 0.16984691960552203,
      "grad_norm": 0.0,
      "learning_rate": 3.943659041727314e-05,
      "loss": 63.6807,
      "step": 42040
    },
    {
      "epoch": 0.16988732087088967,
      "grad_norm": 507.4505920410156,
      "learning_rate": 3.94359320713508e-05,
      "loss": 84.9203,
      "step": 42050
    },
    {
      "epoch": 0.16992772213625731,
      "grad_norm": 490.3003234863281,
      "learning_rate": 3.943527334651495e-05,
      "loss": 50.9804,
      "step": 42060
    },
    {
      "epoch": 0.16996812340162493,
      "grad_norm": 1047.548095703125,
      "learning_rate": 3.9434614242778435e-05,
      "loss": 82.9078,
      "step": 42070
    },
    {
      "epoch": 0.17000852466699257,
      "grad_norm": 1060.2462158203125,
      "learning_rate": 3.9433954760154116e-05,
      "loss": 101.4438,
      "step": 42080
    },
    {
      "epoch": 0.1700489259323602,
      "grad_norm": 1283.112060546875,
      "learning_rate": 3.9433294898654846e-05,
      "loss": 90.8724,
      "step": 42090
    },
    {
      "epoch": 0.17008932719772785,
      "grad_norm": 701.982177734375,
      "learning_rate": 3.943263465829348e-05,
      "loss": 64.7551,
      "step": 42100
    },
    {
      "epoch": 0.17012972846309546,
      "grad_norm": 505.5760498046875,
      "learning_rate": 3.94319740390829e-05,
      "loss": 48.8886,
      "step": 42110
    },
    {
      "epoch": 0.1701701297284631,
      "grad_norm": 1101.0406494140625,
      "learning_rate": 3.943131304103599e-05,
      "loss": 87.4703,
      "step": 42120
    },
    {
      "epoch": 0.17021053099383074,
      "grad_norm": 1292.386474609375,
      "learning_rate": 3.9430651664165616e-05,
      "loss": 113.6191,
      "step": 42130
    },
    {
      "epoch": 0.17025093225919835,
      "grad_norm": 374.6224060058594,
      "learning_rate": 3.942998990848469e-05,
      "loss": 76.7611,
      "step": 42140
    },
    {
      "epoch": 0.170291333524566,
      "grad_norm": 1148.2353515625,
      "learning_rate": 3.942932777400611e-05,
      "loss": 78.2513,
      "step": 42150
    },
    {
      "epoch": 0.17033173478993363,
      "grad_norm": 652.7380981445312,
      "learning_rate": 3.942866526074277e-05,
      "loss": 68.5083,
      "step": 42160
    },
    {
      "epoch": 0.17037213605530124,
      "grad_norm": 1101.5650634765625,
      "learning_rate": 3.942800236870761e-05,
      "loss": 83.1009,
      "step": 42170
    },
    {
      "epoch": 0.17041253732066888,
      "grad_norm": 765.6564331054688,
      "learning_rate": 3.942733909791354e-05,
      "loss": 81.3225,
      "step": 42180
    },
    {
      "epoch": 0.17045293858603652,
      "grad_norm": 647.0240478515625,
      "learning_rate": 3.942667544837349e-05,
      "loss": 113.6144,
      "step": 42190
    },
    {
      "epoch": 0.17049333985140414,
      "grad_norm": 644.6279296875,
      "learning_rate": 3.9426011420100405e-05,
      "loss": 82.8448,
      "step": 42200
    },
    {
      "epoch": 0.17053374111677178,
      "grad_norm": 849.5108642578125,
      "learning_rate": 3.942534701310722e-05,
      "loss": 75.6162,
      "step": 42210
    },
    {
      "epoch": 0.17057414238213942,
      "grad_norm": 869.8896484375,
      "learning_rate": 3.94246822274069e-05,
      "loss": 107.6029,
      "step": 42220
    },
    {
      "epoch": 0.17061454364750703,
      "grad_norm": 1466.9735107421875,
      "learning_rate": 3.9424017063012394e-05,
      "loss": 73.4565,
      "step": 42230
    },
    {
      "epoch": 0.17065494491287467,
      "grad_norm": 934.4044799804688,
      "learning_rate": 3.942335151993668e-05,
      "loss": 86.7369,
      "step": 42240
    },
    {
      "epoch": 0.1706953461782423,
      "grad_norm": 1067.3001708984375,
      "learning_rate": 3.942268559819272e-05,
      "loss": 82.3458,
      "step": 42250
    },
    {
      "epoch": 0.17073574744360995,
      "grad_norm": 1136.6522216796875,
      "learning_rate": 3.9422019297793516e-05,
      "loss": 77.6294,
      "step": 42260
    },
    {
      "epoch": 0.17077614870897756,
      "grad_norm": 696.1351928710938,
      "learning_rate": 3.942135261875204e-05,
      "loss": 118.3667,
      "step": 42270
    },
    {
      "epoch": 0.1708165499743452,
      "grad_norm": 593.81103515625,
      "learning_rate": 3.94206855610813e-05,
      "loss": 88.9259,
      "step": 42280
    },
    {
      "epoch": 0.17085695123971284,
      "grad_norm": 1247.7071533203125,
      "learning_rate": 3.9420018124794294e-05,
      "loss": 79.2447,
      "step": 42290
    },
    {
      "epoch": 0.17089735250508045,
      "grad_norm": 1543.30322265625,
      "learning_rate": 3.941935030990403e-05,
      "loss": 88.1879,
      "step": 42300
    },
    {
      "epoch": 0.1709377537704481,
      "grad_norm": 616.7864379882812,
      "learning_rate": 3.941868211642355e-05,
      "loss": 81.0888,
      "step": 42310
    },
    {
      "epoch": 0.17097815503581573,
      "grad_norm": 504.2239685058594,
      "learning_rate": 3.941801354436585e-05,
      "loss": 79.8196,
      "step": 42320
    },
    {
      "epoch": 0.17101855630118334,
      "grad_norm": 646.478271484375,
      "learning_rate": 3.941734459374399e-05,
      "loss": 79.9383,
      "step": 42330
    },
    {
      "epoch": 0.17105895756655098,
      "grad_norm": 304.4427490234375,
      "learning_rate": 3.941667526457099e-05,
      "loss": 79.6843,
      "step": 42340
    },
    {
      "epoch": 0.17109935883191862,
      "grad_norm": 2277.253662109375,
      "learning_rate": 3.9416005556859914e-05,
      "loss": 91.1325,
      "step": 42350
    },
    {
      "epoch": 0.17113976009728624,
      "grad_norm": 596.6697387695312,
      "learning_rate": 3.9415335470623816e-05,
      "loss": 100.1752,
      "step": 42360
    },
    {
      "epoch": 0.17118016136265388,
      "grad_norm": 2080.851806640625,
      "learning_rate": 3.941466500587575e-05,
      "loss": 111.6894,
      "step": 42370
    },
    {
      "epoch": 0.17122056262802152,
      "grad_norm": 497.0038146972656,
      "learning_rate": 3.9413994162628804e-05,
      "loss": 117.063,
      "step": 42380
    },
    {
      "epoch": 0.17126096389338913,
      "grad_norm": 1601.8934326171875,
      "learning_rate": 3.941332294089604e-05,
      "loss": 79.9856,
      "step": 42390
    },
    {
      "epoch": 0.17130136515875677,
      "grad_norm": 638.1102294921875,
      "learning_rate": 3.941265134069055e-05,
      "loss": 89.9926,
      "step": 42400
    },
    {
      "epoch": 0.1713417664241244,
      "grad_norm": 731.0360107421875,
      "learning_rate": 3.941197936202543e-05,
      "loss": 96.968,
      "step": 42410
    },
    {
      "epoch": 0.17138216768949205,
      "grad_norm": 696.0232543945312,
      "learning_rate": 3.941130700491378e-05,
      "loss": 69.7123,
      "step": 42420
    },
    {
      "epoch": 0.17142256895485966,
      "grad_norm": 1154.012939453125,
      "learning_rate": 3.9410634269368706e-05,
      "loss": 98.8974,
      "step": 42430
    },
    {
      "epoch": 0.1714629702202273,
      "grad_norm": 548.6651000976562,
      "learning_rate": 3.940996115540332e-05,
      "loss": 90.9087,
      "step": 42440
    },
    {
      "epoch": 0.17150337148559494,
      "grad_norm": 1186.6876220703125,
      "learning_rate": 3.9409287663030754e-05,
      "loss": 98.7666,
      "step": 42450
    },
    {
      "epoch": 0.17154377275096255,
      "grad_norm": 1163.623291015625,
      "learning_rate": 3.9408613792264125e-05,
      "loss": 111.3392,
      "step": 42460
    },
    {
      "epoch": 0.1715841740163302,
      "grad_norm": 2000.774658203125,
      "learning_rate": 3.940793954311659e-05,
      "loss": 96.1054,
      "step": 42470
    },
    {
      "epoch": 0.17162457528169783,
      "grad_norm": 1090.0098876953125,
      "learning_rate": 3.940726491560127e-05,
      "loss": 81.3527,
      "step": 42480
    },
    {
      "epoch": 0.17166497654706545,
      "grad_norm": 414.6957092285156,
      "learning_rate": 3.9406589909731335e-05,
      "loss": 50.2307,
      "step": 42490
    },
    {
      "epoch": 0.17170537781243309,
      "grad_norm": 707.9476318359375,
      "learning_rate": 3.940591452551993e-05,
      "loss": 112.7422,
      "step": 42500
    },
    {
      "epoch": 0.17174577907780073,
      "grad_norm": 499.1845703125,
      "learning_rate": 3.940523876298024e-05,
      "loss": 79.7069,
      "step": 42510
    },
    {
      "epoch": 0.17178618034316834,
      "grad_norm": 876.81298828125,
      "learning_rate": 3.940456262212543e-05,
      "loss": 93.4067,
      "step": 42520
    },
    {
      "epoch": 0.17182658160853598,
      "grad_norm": 732.292236328125,
      "learning_rate": 3.940388610296868e-05,
      "loss": 66.0287,
      "step": 42530
    },
    {
      "epoch": 0.17186698287390362,
      "grad_norm": 875.480712890625,
      "learning_rate": 3.9403209205523173e-05,
      "loss": 78.4949,
      "step": 42540
    },
    {
      "epoch": 0.17190738413927123,
      "grad_norm": 677.6840209960938,
      "learning_rate": 3.940253192980212e-05,
      "loss": 92.2969,
      "step": 42550
    },
    {
      "epoch": 0.17194778540463887,
      "grad_norm": 1275.6361083984375,
      "learning_rate": 3.9401854275818716e-05,
      "loss": 86.797,
      "step": 42560
    },
    {
      "epoch": 0.1719881866700065,
      "grad_norm": 857.1155395507812,
      "learning_rate": 3.9401176243586177e-05,
      "loss": 74.4845,
      "step": 42570
    },
    {
      "epoch": 0.17202858793537412,
      "grad_norm": 3165.652587890625,
      "learning_rate": 3.9400497833117716e-05,
      "loss": 116.9501,
      "step": 42580
    },
    {
      "epoch": 0.17206898920074176,
      "grad_norm": 486.62353515625,
      "learning_rate": 3.939981904442657e-05,
      "loss": 76.6197,
      "step": 42590
    },
    {
      "epoch": 0.1721093904661094,
      "grad_norm": 891.3754272460938,
      "learning_rate": 3.939913987752595e-05,
      "loss": 89.9513,
      "step": 42600
    },
    {
      "epoch": 0.17214979173147704,
      "grad_norm": 594.1891479492188,
      "learning_rate": 3.9398460332429115e-05,
      "loss": 91.0488,
      "step": 42610
    },
    {
      "epoch": 0.17219019299684465,
      "grad_norm": 429.0885009765625,
      "learning_rate": 3.9397780409149314e-05,
      "loss": 49.7918,
      "step": 42620
    },
    {
      "epoch": 0.1722305942622123,
      "grad_norm": 1666.359375,
      "learning_rate": 3.9397100107699795e-05,
      "loss": 97.0718,
      "step": 42630
    },
    {
      "epoch": 0.17227099552757993,
      "grad_norm": 603.205322265625,
      "learning_rate": 3.939641942809382e-05,
      "loss": 88.019,
      "step": 42640
    },
    {
      "epoch": 0.17231139679294755,
      "grad_norm": 716.9115600585938,
      "learning_rate": 3.939573837034466e-05,
      "loss": 104.4141,
      "step": 42650
    },
    {
      "epoch": 0.1723517980583152,
      "grad_norm": 755.2896118164062,
      "learning_rate": 3.9395056934465604e-05,
      "loss": 104.734,
      "step": 42660
    },
    {
      "epoch": 0.17239219932368283,
      "grad_norm": 438.7043151855469,
      "learning_rate": 3.939437512046993e-05,
      "loss": 83.8911,
      "step": 42670
    },
    {
      "epoch": 0.17243260058905044,
      "grad_norm": 1080.4783935546875,
      "learning_rate": 3.939369292837092e-05,
      "loss": 80.1764,
      "step": 42680
    },
    {
      "epoch": 0.17247300185441808,
      "grad_norm": 1368.8602294921875,
      "learning_rate": 3.939301035818188e-05,
      "loss": 76.1026,
      "step": 42690
    },
    {
      "epoch": 0.17251340311978572,
      "grad_norm": 1755.056640625,
      "learning_rate": 3.939232740991612e-05,
      "loss": 102.755,
      "step": 42700
    },
    {
      "epoch": 0.17255380438515333,
      "grad_norm": 1074.890869140625,
      "learning_rate": 3.939164408358696e-05,
      "loss": 110.8715,
      "step": 42710
    },
    {
      "epoch": 0.17259420565052097,
      "grad_norm": 749.5791015625,
      "learning_rate": 3.939096037920771e-05,
      "loss": 83.5525,
      "step": 42720
    },
    {
      "epoch": 0.1726346069158886,
      "grad_norm": 882.516357421875,
      "learning_rate": 3.939027629679171e-05,
      "loss": 90.7301,
      "step": 42730
    },
    {
      "epoch": 0.17267500818125622,
      "grad_norm": 6177.97705078125,
      "learning_rate": 3.938959183635228e-05,
      "loss": 101.8573,
      "step": 42740
    },
    {
      "epoch": 0.17271540944662386,
      "grad_norm": 1139.997314453125,
      "learning_rate": 3.9388906997902784e-05,
      "loss": 85.6628,
      "step": 42750
    },
    {
      "epoch": 0.1727558107119915,
      "grad_norm": 1481.9365234375,
      "learning_rate": 3.938822178145656e-05,
      "loss": 114.4302,
      "step": 42760
    },
    {
      "epoch": 0.17279621197735914,
      "grad_norm": 1003.7747192382812,
      "learning_rate": 3.938753618702697e-05,
      "loss": 107.0744,
      "step": 42770
    },
    {
      "epoch": 0.17283661324272676,
      "grad_norm": 778.5792236328125,
      "learning_rate": 3.938685021462738e-05,
      "loss": 75.9139,
      "step": 42780
    },
    {
      "epoch": 0.1728770145080944,
      "grad_norm": 596.6504516601562,
      "learning_rate": 3.938616386427117e-05,
      "loss": 95.3925,
      "step": 42790
    },
    {
      "epoch": 0.17291741577346204,
      "grad_norm": 346.0478820800781,
      "learning_rate": 3.938547713597171e-05,
      "loss": 64.3553,
      "step": 42800
    },
    {
      "epoch": 0.17295781703882965,
      "grad_norm": 659.998291015625,
      "learning_rate": 3.938479002974239e-05,
      "loss": 79.2351,
      "step": 42810
    },
    {
      "epoch": 0.1729982183041973,
      "grad_norm": 1590.90283203125,
      "learning_rate": 3.938410254559661e-05,
      "loss": 111.3108,
      "step": 42820
    },
    {
      "epoch": 0.17303861956956493,
      "grad_norm": 580.7548217773438,
      "learning_rate": 3.938341468354778e-05,
      "loss": 78.8128,
      "step": 42830
    },
    {
      "epoch": 0.17307902083493254,
      "grad_norm": 1085.07373046875,
      "learning_rate": 3.938272644360929e-05,
      "loss": 142.478,
      "step": 42840
    },
    {
      "epoch": 0.17311942210030018,
      "grad_norm": 446.0859375,
      "learning_rate": 3.938203782579457e-05,
      "loss": 73.5577,
      "step": 42850
    },
    {
      "epoch": 0.17315982336566782,
      "grad_norm": 452.1142578125,
      "learning_rate": 3.9381348830117055e-05,
      "loss": 93.8503,
      "step": 42860
    },
    {
      "epoch": 0.17320022463103543,
      "grad_norm": 1228.706298828125,
      "learning_rate": 3.938065945659016e-05,
      "loss": 89.0013,
      "step": 42870
    },
    {
      "epoch": 0.17324062589640307,
      "grad_norm": 830.8073120117188,
      "learning_rate": 3.937996970522733e-05,
      "loss": 82.8595,
      "step": 42880
    },
    {
      "epoch": 0.1732810271617707,
      "grad_norm": 730.0189208984375,
      "learning_rate": 3.937927957604201e-05,
      "loss": 50.0603,
      "step": 42890
    },
    {
      "epoch": 0.17332142842713832,
      "grad_norm": 706.650634765625,
      "learning_rate": 3.937858906904766e-05,
      "loss": 44.7563,
      "step": 42900
    },
    {
      "epoch": 0.17336182969250596,
      "grad_norm": 1158.515380859375,
      "learning_rate": 3.937789818425774e-05,
      "loss": 125.2809,
      "step": 42910
    },
    {
      "epoch": 0.1734022309578736,
      "grad_norm": 834.2371215820312,
      "learning_rate": 3.9377206921685725e-05,
      "loss": 93.5468,
      "step": 42920
    },
    {
      "epoch": 0.17344263222324124,
      "grad_norm": 609.667236328125,
      "learning_rate": 3.9376515281345076e-05,
      "loss": 92.9755,
      "step": 42930
    },
    {
      "epoch": 0.17348303348860886,
      "grad_norm": 701.4984130859375,
      "learning_rate": 3.937582326324929e-05,
      "loss": 89.169,
      "step": 42940
    },
    {
      "epoch": 0.1735234347539765,
      "grad_norm": 0.0,
      "learning_rate": 3.937513086741185e-05,
      "loss": 84.4199,
      "step": 42950
    },
    {
      "epoch": 0.17356383601934414,
      "grad_norm": 755.2521362304688,
      "learning_rate": 3.937443809384626e-05,
      "loss": 87.8608,
      "step": 42960
    },
    {
      "epoch": 0.17360423728471175,
      "grad_norm": 240.83729553222656,
      "learning_rate": 3.9373744942566025e-05,
      "loss": 59.1297,
      "step": 42970
    },
    {
      "epoch": 0.1736446385500794,
      "grad_norm": 513.7739868164062,
      "learning_rate": 3.937305141358466e-05,
      "loss": 75.2643,
      "step": 42980
    },
    {
      "epoch": 0.17368503981544703,
      "grad_norm": 667.5625610351562,
      "learning_rate": 3.937235750691569e-05,
      "loss": 98.8688,
      "step": 42990
    },
    {
      "epoch": 0.17372544108081464,
      "grad_norm": 540.6917724609375,
      "learning_rate": 3.9371663222572625e-05,
      "loss": 91.6689,
      "step": 43000
    },
    {
      "epoch": 0.17376584234618228,
      "grad_norm": 1611.7796630859375,
      "learning_rate": 3.937096856056902e-05,
      "loss": 77.2374,
      "step": 43010
    },
    {
      "epoch": 0.17380624361154992,
      "grad_norm": 1174.986572265625,
      "learning_rate": 3.93702735209184e-05,
      "loss": 88.8681,
      "step": 43020
    },
    {
      "epoch": 0.17384664487691753,
      "grad_norm": 1794.541259765625,
      "learning_rate": 3.936957810363434e-05,
      "loss": 123.8373,
      "step": 43030
    },
    {
      "epoch": 0.17388704614228517,
      "grad_norm": 643.811279296875,
      "learning_rate": 3.936888230873037e-05,
      "loss": 79.1278,
      "step": 43040
    },
    {
      "epoch": 0.1739274474076528,
      "grad_norm": 1467.1026611328125,
      "learning_rate": 3.936818613622008e-05,
      "loss": 112.4248,
      "step": 43050
    },
    {
      "epoch": 0.17396784867302043,
      "grad_norm": 595.2394409179688,
      "learning_rate": 3.936748958611702e-05,
      "loss": 94.1707,
      "step": 43060
    },
    {
      "epoch": 0.17400824993838807,
      "grad_norm": 2667.404052734375,
      "learning_rate": 3.936679265843478e-05,
      "loss": 115.8994,
      "step": 43070
    },
    {
      "epoch": 0.1740486512037557,
      "grad_norm": 1556.4005126953125,
      "learning_rate": 3.936609535318695e-05,
      "loss": 94.7888,
      "step": 43080
    },
    {
      "epoch": 0.17408905246912335,
      "grad_norm": 621.27099609375,
      "learning_rate": 3.936539767038712e-05,
      "loss": 77.0604,
      "step": 43090
    },
    {
      "epoch": 0.17412945373449096,
      "grad_norm": 711.2188720703125,
      "learning_rate": 3.9364699610048894e-05,
      "loss": 89.3644,
      "step": 43100
    },
    {
      "epoch": 0.1741698549998586,
      "grad_norm": 662.9762573242188,
      "learning_rate": 3.9364001172185875e-05,
      "loss": 57.3206,
      "step": 43110
    },
    {
      "epoch": 0.17421025626522624,
      "grad_norm": 1898.1923828125,
      "learning_rate": 3.936330235681169e-05,
      "loss": 103.0097,
      "step": 43120
    },
    {
      "epoch": 0.17425065753059385,
      "grad_norm": 682.8529052734375,
      "learning_rate": 3.936260316393995e-05,
      "loss": 108.5161,
      "step": 43130
    },
    {
      "epoch": 0.1742910587959615,
      "grad_norm": 369.9712829589844,
      "learning_rate": 3.936190359358429e-05,
      "loss": 92.947,
      "step": 43140
    },
    {
      "epoch": 0.17433146006132913,
      "grad_norm": 757.6825561523438,
      "learning_rate": 3.936120364575836e-05,
      "loss": 70.8851,
      "step": 43150
    },
    {
      "epoch": 0.17437186132669674,
      "grad_norm": 719.6384887695312,
      "learning_rate": 3.9360503320475795e-05,
      "loss": 94.8329,
      "step": 43160
    },
    {
      "epoch": 0.17441226259206438,
      "grad_norm": 917.0863647460938,
      "learning_rate": 3.935980261775025e-05,
      "loss": 95.4134,
      "step": 43170
    },
    {
      "epoch": 0.17445266385743202,
      "grad_norm": 795.0466918945312,
      "learning_rate": 3.935910153759538e-05,
      "loss": 75.6606,
      "step": 43180
    },
    {
      "epoch": 0.17449306512279963,
      "grad_norm": 618.247802734375,
      "learning_rate": 3.9358400080024867e-05,
      "loss": 86.7528,
      "step": 43190
    },
    {
      "epoch": 0.17453346638816727,
      "grad_norm": 850.3892211914062,
      "learning_rate": 3.935769824505238e-05,
      "loss": 118.1322,
      "step": 43200
    },
    {
      "epoch": 0.17457386765353491,
      "grad_norm": 578.2403564453125,
      "learning_rate": 3.935699603269159e-05,
      "loss": 81.528,
      "step": 43210
    },
    {
      "epoch": 0.17461426891890253,
      "grad_norm": 625.4942626953125,
      "learning_rate": 3.935629344295621e-05,
      "loss": 87.5598,
      "step": 43220
    },
    {
      "epoch": 0.17465467018427017,
      "grad_norm": 2594.657470703125,
      "learning_rate": 3.935559047585991e-05,
      "loss": 75.9269,
      "step": 43230
    },
    {
      "epoch": 0.1746950714496378,
      "grad_norm": 841.3003540039062,
      "learning_rate": 3.935488713141641e-05,
      "loss": 95.3032,
      "step": 43240
    },
    {
      "epoch": 0.17473547271500545,
      "grad_norm": 807.5640258789062,
      "learning_rate": 3.935418340963943e-05,
      "loss": 95.9208,
      "step": 43250
    },
    {
      "epoch": 0.17477587398037306,
      "grad_norm": 1481.22509765625,
      "learning_rate": 3.935347931054267e-05,
      "loss": 71.7565,
      "step": 43260
    },
    {
      "epoch": 0.1748162752457407,
      "grad_norm": 1714.5614013671875,
      "learning_rate": 3.9352774834139875e-05,
      "loss": 98.7609,
      "step": 43270
    },
    {
      "epoch": 0.17485667651110834,
      "grad_norm": 801.7985229492188,
      "learning_rate": 3.9352069980444764e-05,
      "loss": 103.7962,
      "step": 43280
    },
    {
      "epoch": 0.17489707777647595,
      "grad_norm": 970.6455078125,
      "learning_rate": 3.9351364749471095e-05,
      "loss": 66.2405,
      "step": 43290
    },
    {
      "epoch": 0.1749374790418436,
      "grad_norm": 665.2471313476562,
      "learning_rate": 3.93506591412326e-05,
      "loss": 62.2794,
      "step": 43300
    },
    {
      "epoch": 0.17497788030721123,
      "grad_norm": 1091.2325439453125,
      "learning_rate": 3.9349953155743046e-05,
      "loss": 75.6549,
      "step": 43310
    },
    {
      "epoch": 0.17501828157257884,
      "grad_norm": 819.3821411132812,
      "learning_rate": 3.934924679301619e-05,
      "loss": 78.266,
      "step": 43320
    },
    {
      "epoch": 0.17505868283794648,
      "grad_norm": 0.0,
      "learning_rate": 3.934854005306581e-05,
      "loss": 108.1789,
      "step": 43330
    },
    {
      "epoch": 0.17509908410331412,
      "grad_norm": 283.76165771484375,
      "learning_rate": 3.934783293590568e-05,
      "loss": 67.2186,
      "step": 43340
    },
    {
      "epoch": 0.17513948536868174,
      "grad_norm": 577.4725341796875,
      "learning_rate": 3.9347125441549585e-05,
      "loss": 72.1054,
      "step": 43350
    },
    {
      "epoch": 0.17517988663404938,
      "grad_norm": 737.1785278320312,
      "learning_rate": 3.9346417570011316e-05,
      "loss": 112.302,
      "step": 43360
    },
    {
      "epoch": 0.17522028789941702,
      "grad_norm": 3394.2783203125,
      "learning_rate": 3.934570932130468e-05,
      "loss": 117.2829,
      "step": 43370
    },
    {
      "epoch": 0.17526068916478463,
      "grad_norm": 893.6383666992188,
      "learning_rate": 3.934500069544348e-05,
      "loss": 79.7919,
      "step": 43380
    },
    {
      "epoch": 0.17530109043015227,
      "grad_norm": 835.4736938476562,
      "learning_rate": 3.934429169244154e-05,
      "loss": 87.5427,
      "step": 43390
    },
    {
      "epoch": 0.1753414916955199,
      "grad_norm": 416.30816650390625,
      "learning_rate": 3.934358231231266e-05,
      "loss": 95.4932,
      "step": 43400
    },
    {
      "epoch": 0.17538189296088755,
      "grad_norm": 764.5811157226562,
      "learning_rate": 3.934287255507069e-05,
      "loss": 68.0534,
      "step": 43410
    },
    {
      "epoch": 0.17542229422625516,
      "grad_norm": 404.3819274902344,
      "learning_rate": 3.9342162420729464e-05,
      "loss": 65.4451,
      "step": 43420
    },
    {
      "epoch": 0.1754626954916228,
      "grad_norm": 862.0521240234375,
      "learning_rate": 3.9341451909302814e-05,
      "loss": 96.0394,
      "step": 43430
    },
    {
      "epoch": 0.17550309675699044,
      "grad_norm": 1187.9178466796875,
      "learning_rate": 3.934074102080461e-05,
      "loss": 85.2422,
      "step": 43440
    },
    {
      "epoch": 0.17554349802235805,
      "grad_norm": 914.9650268554688,
      "learning_rate": 3.9340029755248707e-05,
      "loss": 85.6274,
      "step": 43450
    },
    {
      "epoch": 0.1755838992877257,
      "grad_norm": 528.726318359375,
      "learning_rate": 3.933931811264896e-05,
      "loss": 87.7742,
      "step": 43460
    },
    {
      "epoch": 0.17562430055309333,
      "grad_norm": 575.98095703125,
      "learning_rate": 3.933860609301925e-05,
      "loss": 92.3105,
      "step": 43470
    },
    {
      "epoch": 0.17566470181846094,
      "grad_norm": 656.6376953125,
      "learning_rate": 3.9337893696373456e-05,
      "loss": 50.5727,
      "step": 43480
    },
    {
      "epoch": 0.17570510308382858,
      "grad_norm": 599.2984619140625,
      "learning_rate": 3.933718092272547e-05,
      "loss": 80.5968,
      "step": 43490
    },
    {
      "epoch": 0.17574550434919622,
      "grad_norm": 794.0865478515625,
      "learning_rate": 3.9336467772089195e-05,
      "loss": 77.5349,
      "step": 43500
    },
    {
      "epoch": 0.17578590561456384,
      "grad_norm": 1055.6077880859375,
      "learning_rate": 3.9335754244478514e-05,
      "loss": 96.4366,
      "step": 43510
    },
    {
      "epoch": 0.17582630687993148,
      "grad_norm": 378.260009765625,
      "learning_rate": 3.9335040339907354e-05,
      "loss": 57.8697,
      "step": 43520
    },
    {
      "epoch": 0.17586670814529912,
      "grad_norm": 536.3444213867188,
      "learning_rate": 3.933432605838963e-05,
      "loss": 72.2509,
      "step": 43530
    },
    {
      "epoch": 0.17590710941066673,
      "grad_norm": 2800.318603515625,
      "learning_rate": 3.933361139993926e-05,
      "loss": 102.6901,
      "step": 43540
    },
    {
      "epoch": 0.17594751067603437,
      "grad_norm": 637.18701171875,
      "learning_rate": 3.933289636457019e-05,
      "loss": 53.6399,
      "step": 43550
    },
    {
      "epoch": 0.175987911941402,
      "grad_norm": 835.831787109375,
      "learning_rate": 3.933218095229634e-05,
      "loss": 109.055,
      "step": 43560
    },
    {
      "epoch": 0.17602831320676965,
      "grad_norm": 671.2997436523438,
      "learning_rate": 3.933146516313169e-05,
      "loss": 73.9181,
      "step": 43570
    },
    {
      "epoch": 0.17606871447213726,
      "grad_norm": 916.1385498046875,
      "learning_rate": 3.933074899709016e-05,
      "loss": 137.5969,
      "step": 43580
    },
    {
      "epoch": 0.1761091157375049,
      "grad_norm": 548.91943359375,
      "learning_rate": 3.933003245418572e-05,
      "loss": 77.7896,
      "step": 43590
    },
    {
      "epoch": 0.17614951700287254,
      "grad_norm": 1265.389404296875,
      "learning_rate": 3.932931553443235e-05,
      "loss": 82.4361,
      "step": 43600
    },
    {
      "epoch": 0.17618991826824015,
      "grad_norm": 1766.3031005859375,
      "learning_rate": 3.9328598237844035e-05,
      "loss": 82.3127,
      "step": 43610
    },
    {
      "epoch": 0.1762303195336078,
      "grad_norm": 833.5023803710938,
      "learning_rate": 3.9327880564434735e-05,
      "loss": 115.4179,
      "step": 43620
    },
    {
      "epoch": 0.17627072079897543,
      "grad_norm": 1055.803955078125,
      "learning_rate": 3.932716251421845e-05,
      "loss": 67.2918,
      "step": 43630
    },
    {
      "epoch": 0.17631112206434305,
      "grad_norm": 898.137939453125,
      "learning_rate": 3.932644408720919e-05,
      "loss": 99.6032,
      "step": 43640
    },
    {
      "epoch": 0.17635152332971069,
      "grad_norm": 2444.6591796875,
      "learning_rate": 3.932572528342094e-05,
      "loss": 100.4514,
      "step": 43650
    },
    {
      "epoch": 0.17639192459507833,
      "grad_norm": 1445.8916015625,
      "learning_rate": 3.932500610286774e-05,
      "loss": 70.1387,
      "step": 43660
    },
    {
      "epoch": 0.17643232586044594,
      "grad_norm": 927.367919921875,
      "learning_rate": 3.9324286545563595e-05,
      "loss": 93.5744,
      "step": 43670
    },
    {
      "epoch": 0.17647272712581358,
      "grad_norm": 459.9096984863281,
      "learning_rate": 3.932356661152253e-05,
      "loss": 80.8082,
      "step": 43680
    },
    {
      "epoch": 0.17651312839118122,
      "grad_norm": 2630.177490234375,
      "learning_rate": 3.9322846300758585e-05,
      "loss": 79.219,
      "step": 43690
    },
    {
      "epoch": 0.17655352965654883,
      "grad_norm": 0.0,
      "learning_rate": 3.9322125613285805e-05,
      "loss": 60.184,
      "step": 43700
    },
    {
      "epoch": 0.17659393092191647,
      "grad_norm": 883.2347412109375,
      "learning_rate": 3.9321404549118236e-05,
      "loss": 74.5576,
      "step": 43710
    },
    {
      "epoch": 0.1766343321872841,
      "grad_norm": 650.3308715820312,
      "learning_rate": 3.9320683108269945e-05,
      "loss": 103.2459,
      "step": 43720
    },
    {
      "epoch": 0.17667473345265175,
      "grad_norm": 555.4349365234375,
      "learning_rate": 3.9319961290754985e-05,
      "loss": 64.6827,
      "step": 43730
    },
    {
      "epoch": 0.17671513471801936,
      "grad_norm": 310.9591064453125,
      "learning_rate": 3.931923909658744e-05,
      "loss": 77.393,
      "step": 43740
    },
    {
      "epoch": 0.176755535983387,
      "grad_norm": 919.4443969726562,
      "learning_rate": 3.931851652578137e-05,
      "loss": 100.588,
      "step": 43750
    },
    {
      "epoch": 0.17679593724875464,
      "grad_norm": 673.5975341796875,
      "learning_rate": 3.931779357835088e-05,
      "loss": 92.4325,
      "step": 43760
    },
    {
      "epoch": 0.17683633851412225,
      "grad_norm": 1112.824951171875,
      "learning_rate": 3.9317070254310056e-05,
      "loss": 74.7196,
      "step": 43770
    },
    {
      "epoch": 0.1768767397794899,
      "grad_norm": 1260.9169921875,
      "learning_rate": 3.931634655367301e-05,
      "loss": 73.1931,
      "step": 43780
    },
    {
      "epoch": 0.17691714104485753,
      "grad_norm": 0.0,
      "learning_rate": 3.931562247645384e-05,
      "loss": 89.5991,
      "step": 43790
    },
    {
      "epoch": 0.17695754231022515,
      "grad_norm": 335.5714111328125,
      "learning_rate": 3.9314898022666656e-05,
      "loss": 76.3316,
      "step": 43800
    },
    {
      "epoch": 0.1769979435755928,
      "grad_norm": 840.8833618164062,
      "learning_rate": 3.9314173192325603e-05,
      "loss": 131.7811,
      "step": 43810
    },
    {
      "epoch": 0.17703834484096043,
      "grad_norm": 823.3009643554688,
      "learning_rate": 3.9313447985444796e-05,
      "loss": 86.1308,
      "step": 43820
    },
    {
      "epoch": 0.17707874610632804,
      "grad_norm": 627.9525146484375,
      "learning_rate": 3.931272240203838e-05,
      "loss": 70.9252,
      "step": 43830
    },
    {
      "epoch": 0.17711914737169568,
      "grad_norm": 781.281982421875,
      "learning_rate": 3.93119964421205e-05,
      "loss": 66.6196,
      "step": 43840
    },
    {
      "epoch": 0.17715954863706332,
      "grad_norm": 878.00048828125,
      "learning_rate": 3.9311270105705296e-05,
      "loss": 96.4722,
      "step": 43850
    },
    {
      "epoch": 0.17719994990243093,
      "grad_norm": 634.25244140625,
      "learning_rate": 3.931054339280695e-05,
      "loss": 83.8569,
      "step": 43860
    },
    {
      "epoch": 0.17724035116779857,
      "grad_norm": 3098.12255859375,
      "learning_rate": 3.930981630343962e-05,
      "loss": 103.4742,
      "step": 43870
    },
    {
      "epoch": 0.1772807524331662,
      "grad_norm": 655.8944091796875,
      "learning_rate": 3.9309088837617464e-05,
      "loss": 97.8886,
      "step": 43880
    },
    {
      "epoch": 0.17732115369853385,
      "grad_norm": 849.6776733398438,
      "learning_rate": 3.930836099535469e-05,
      "loss": 75.4539,
      "step": 43890
    },
    {
      "epoch": 0.17736155496390146,
      "grad_norm": 593.6189575195312,
      "learning_rate": 3.930763277666548e-05,
      "loss": 96.333,
      "step": 43900
    },
    {
      "epoch": 0.1774019562292691,
      "grad_norm": 0.0,
      "learning_rate": 3.9306904181564025e-05,
      "loss": 46.8365,
      "step": 43910
    },
    {
      "epoch": 0.17744235749463674,
      "grad_norm": 654.2417602539062,
      "learning_rate": 3.930617521006454e-05,
      "loss": 70.6252,
      "step": 43920
    },
    {
      "epoch": 0.17748275876000436,
      "grad_norm": 984.5786743164062,
      "learning_rate": 3.9305445862181225e-05,
      "loss": 79.7624,
      "step": 43930
    },
    {
      "epoch": 0.177523160025372,
      "grad_norm": 1487.3980712890625,
      "learning_rate": 3.93047161379283e-05,
      "loss": 100.9878,
      "step": 43940
    },
    {
      "epoch": 0.17756356129073964,
      "grad_norm": 725.6891479492188,
      "learning_rate": 3.9303986037320004e-05,
      "loss": 75.8188,
      "step": 43950
    },
    {
      "epoch": 0.17760396255610725,
      "grad_norm": 410.4052734375,
      "learning_rate": 3.9303255560370554e-05,
      "loss": 93.6783,
      "step": 43960
    },
    {
      "epoch": 0.1776443638214749,
      "grad_norm": 746.7178344726562,
      "learning_rate": 3.93025247070942e-05,
      "loss": 67.6668,
      "step": 43970
    },
    {
      "epoch": 0.17768476508684253,
      "grad_norm": 586.8406982421875,
      "learning_rate": 3.930179347750519e-05,
      "loss": 74.0758,
      "step": 43980
    },
    {
      "epoch": 0.17772516635221014,
      "grad_norm": 555.885009765625,
      "learning_rate": 3.930106187161778e-05,
      "loss": 79.881,
      "step": 43990
    },
    {
      "epoch": 0.17776556761757778,
      "grad_norm": 1578.247314453125,
      "learning_rate": 3.930032988944623e-05,
      "loss": 101.9592,
      "step": 44000
    },
    {
      "epoch": 0.17780596888294542,
      "grad_norm": 890.3057861328125,
      "learning_rate": 3.9299597531004816e-05,
      "loss": 72.5262,
      "step": 44010
    },
    {
      "epoch": 0.17784637014831303,
      "grad_norm": 896.8656616210938,
      "learning_rate": 3.929886479630781e-05,
      "loss": 113.3441,
      "step": 44020
    },
    {
      "epoch": 0.17788677141368067,
      "grad_norm": 355.8651123046875,
      "learning_rate": 3.92981316853695e-05,
      "loss": 70.6483,
      "step": 44030
    },
    {
      "epoch": 0.1779271726790483,
      "grad_norm": 547.337890625,
      "learning_rate": 3.929739819820418e-05,
      "loss": 131.6473,
      "step": 44040
    },
    {
      "epoch": 0.17796757394441595,
      "grad_norm": 1019.341064453125,
      "learning_rate": 3.9296664334826135e-05,
      "loss": 51.182,
      "step": 44050
    },
    {
      "epoch": 0.17800797520978356,
      "grad_norm": 1067.612548828125,
      "learning_rate": 3.9295930095249695e-05,
      "loss": 117.0978,
      "step": 44060
    },
    {
      "epoch": 0.1780483764751512,
      "grad_norm": 832.2264404296875,
      "learning_rate": 3.9295195479489165e-05,
      "loss": 63.1411,
      "step": 44070
    },
    {
      "epoch": 0.17808877774051884,
      "grad_norm": 605.8679809570312,
      "learning_rate": 3.9294460487558865e-05,
      "loss": 97.1942,
      "step": 44080
    },
    {
      "epoch": 0.17812917900588646,
      "grad_norm": 546.1611938476562,
      "learning_rate": 3.929372511947311e-05,
      "loss": 77.0141,
      "step": 44090
    },
    {
      "epoch": 0.1781695802712541,
      "grad_norm": 1067.419921875,
      "learning_rate": 3.929298937524627e-05,
      "loss": 266.8003,
      "step": 44100
    },
    {
      "epoch": 0.17820998153662174,
      "grad_norm": 428.70965576171875,
      "learning_rate": 3.929225325489266e-05,
      "loss": 76.3916,
      "step": 44110
    },
    {
      "epoch": 0.17825038280198935,
      "grad_norm": 632.5858154296875,
      "learning_rate": 3.929151675842664e-05,
      "loss": 58.2997,
      "step": 44120
    },
    {
      "epoch": 0.178290784067357,
      "grad_norm": 895.70654296875,
      "learning_rate": 3.929077988586257e-05,
      "loss": 93.6626,
      "step": 44130
    },
    {
      "epoch": 0.17833118533272463,
      "grad_norm": 776.591552734375,
      "learning_rate": 3.9290042637214815e-05,
      "loss": 81.59,
      "step": 44140
    },
    {
      "epoch": 0.17837158659809224,
      "grad_norm": 1873.4609375,
      "learning_rate": 3.9289305012497745e-05,
      "loss": 100.9556,
      "step": 44150
    },
    {
      "epoch": 0.17841198786345988,
      "grad_norm": 761.1587524414062,
      "learning_rate": 3.928856701172575e-05,
      "loss": 80.4297,
      "step": 44160
    },
    {
      "epoch": 0.17845238912882752,
      "grad_norm": 805.0286254882812,
      "learning_rate": 3.928782863491321e-05,
      "loss": 76.5413,
      "step": 44170
    },
    {
      "epoch": 0.17849279039419513,
      "grad_norm": 711.6513671875,
      "learning_rate": 3.928708988207452e-05,
      "loss": 51.0569,
      "step": 44180
    },
    {
      "epoch": 0.17853319165956277,
      "grad_norm": 1271.82568359375,
      "learning_rate": 3.928635075322407e-05,
      "loss": 73.9333,
      "step": 44190
    },
    {
      "epoch": 0.1785735929249304,
      "grad_norm": 864.5879516601562,
      "learning_rate": 3.92856112483763e-05,
      "loss": 50.6767,
      "step": 44200
    },
    {
      "epoch": 0.17861399419029805,
      "grad_norm": 958.3281860351562,
      "learning_rate": 3.9284871367545595e-05,
      "loss": 88.4344,
      "step": 44210
    },
    {
      "epoch": 0.17865439545566567,
      "grad_norm": 1121.1851806640625,
      "learning_rate": 3.9284131110746404e-05,
      "loss": 100.7003,
      "step": 44220
    },
    {
      "epoch": 0.1786947967210333,
      "grad_norm": 685.8429565429688,
      "learning_rate": 3.928339047799315e-05,
      "loss": 73.3415,
      "step": 44230
    },
    {
      "epoch": 0.17873519798640095,
      "grad_norm": 2012.408935546875,
      "learning_rate": 3.928264946930027e-05,
      "loss": 81.0713,
      "step": 44240
    },
    {
      "epoch": 0.17877559925176856,
      "grad_norm": 1126.882568359375,
      "learning_rate": 3.92819080846822e-05,
      "loss": 70.0383,
      "step": 44250
    },
    {
      "epoch": 0.1788160005171362,
      "grad_norm": 861.8523559570312,
      "learning_rate": 3.928116632415342e-05,
      "loss": 91.0342,
      "step": 44260
    },
    {
      "epoch": 0.17885640178250384,
      "grad_norm": 1251.7584228515625,
      "learning_rate": 3.928042418772837e-05,
      "loss": 138.0804,
      "step": 44270
    },
    {
      "epoch": 0.17889680304787145,
      "grad_norm": 1022.703857421875,
      "learning_rate": 3.927968167542152e-05,
      "loss": 82.487,
      "step": 44280
    },
    {
      "epoch": 0.1789372043132391,
      "grad_norm": 1013.7217407226562,
      "learning_rate": 3.9278938787247354e-05,
      "loss": 71.4098,
      "step": 44290
    },
    {
      "epoch": 0.17897760557860673,
      "grad_norm": 1512.89990234375,
      "learning_rate": 3.927819552322035e-05,
      "loss": 90.1114,
      "step": 44300
    },
    {
      "epoch": 0.17901800684397434,
      "grad_norm": 365.7040100097656,
      "learning_rate": 3.9277451883355005e-05,
      "loss": 59.6052,
      "step": 44310
    },
    {
      "epoch": 0.17905840810934198,
      "grad_norm": 4800.70263671875,
      "learning_rate": 3.9276707867665805e-05,
      "loss": 81.1484,
      "step": 44320
    },
    {
      "epoch": 0.17909880937470962,
      "grad_norm": 537.82568359375,
      "learning_rate": 3.927596347616726e-05,
      "loss": 85.3282,
      "step": 44330
    },
    {
      "epoch": 0.17913921064007723,
      "grad_norm": 815.4710693359375,
      "learning_rate": 3.9275218708873887e-05,
      "loss": 72.1017,
      "step": 44340
    },
    {
      "epoch": 0.17917961190544487,
      "grad_norm": 1025.384765625,
      "learning_rate": 3.92744735658002e-05,
      "loss": 132.5589,
      "step": 44350
    },
    {
      "epoch": 0.17922001317081251,
      "grad_norm": 359.1304016113281,
      "learning_rate": 3.9273728046960726e-05,
      "loss": 82.7631,
      "step": 44360
    },
    {
      "epoch": 0.17926041443618015,
      "grad_norm": 1203.0955810546875,
      "learning_rate": 3.927298215237e-05,
      "loss": 66.8553,
      "step": 44370
    },
    {
      "epoch": 0.17930081570154777,
      "grad_norm": 689.65478515625,
      "learning_rate": 3.927223588204257e-05,
      "loss": 66.0106,
      "step": 44380
    },
    {
      "epoch": 0.1793412169669154,
      "grad_norm": 832.0293579101562,
      "learning_rate": 3.927148923599297e-05,
      "loss": 67.1883,
      "step": 44390
    },
    {
      "epoch": 0.17938161823228305,
      "grad_norm": 1041.6201171875,
      "learning_rate": 3.927074221423578e-05,
      "loss": 67.549,
      "step": 44400
    },
    {
      "epoch": 0.17942201949765066,
      "grad_norm": 566.5820922851562,
      "learning_rate": 3.9269994816785535e-05,
      "loss": 75.9646,
      "step": 44410
    },
    {
      "epoch": 0.1794624207630183,
      "grad_norm": 1437.30029296875,
      "learning_rate": 3.9269247043656825e-05,
      "loss": 107.2695,
      "step": 44420
    },
    {
      "epoch": 0.17950282202838594,
      "grad_norm": 663.2697143554688,
      "learning_rate": 3.926849889486423e-05,
      "loss": 72.5427,
      "step": 44430
    },
    {
      "epoch": 0.17954322329375355,
      "grad_norm": 1476.98974609375,
      "learning_rate": 3.926775037042232e-05,
      "loss": 84.6121,
      "step": 44440
    },
    {
      "epoch": 0.1795836245591212,
      "grad_norm": 629.34130859375,
      "learning_rate": 3.92670014703457e-05,
      "loss": 86.1632,
      "step": 44450
    },
    {
      "epoch": 0.17962402582448883,
      "grad_norm": 900.3147583007812,
      "learning_rate": 3.926625219464897e-05,
      "loss": 61.9779,
      "step": 44460
    },
    {
      "epoch": 0.17966442708985644,
      "grad_norm": 444.7907409667969,
      "learning_rate": 3.926550254334673e-05,
      "loss": 71.5317,
      "step": 44470
    },
    {
      "epoch": 0.17970482835522408,
      "grad_norm": 1078.9156494140625,
      "learning_rate": 3.92647525164536e-05,
      "loss": 74.6889,
      "step": 44480
    },
    {
      "epoch": 0.17974522962059172,
      "grad_norm": 1081.789794921875,
      "learning_rate": 3.92640021139842e-05,
      "loss": 104.8707,
      "step": 44490
    },
    {
      "epoch": 0.17978563088595934,
      "grad_norm": 642.1022338867188,
      "learning_rate": 3.926325133595317e-05,
      "loss": 65.5858,
      "step": 44500
    },
    {
      "epoch": 0.17982603215132698,
      "grad_norm": 1025.814208984375,
      "learning_rate": 3.926250018237513e-05,
      "loss": 86.6868,
      "step": 44510
    },
    {
      "epoch": 0.17986643341669462,
      "grad_norm": 2212.697021484375,
      "learning_rate": 3.926174865326473e-05,
      "loss": 110.0421,
      "step": 44520
    },
    {
      "epoch": 0.17990683468206226,
      "grad_norm": 890.4210815429688,
      "learning_rate": 3.926099674863663e-05,
      "loss": 58.2174,
      "step": 44530
    },
    {
      "epoch": 0.17994723594742987,
      "grad_norm": 1108.4072265625,
      "learning_rate": 3.926024446850548e-05,
      "loss": 82.1017,
      "step": 44540
    },
    {
      "epoch": 0.1799876372127975,
      "grad_norm": 1117.5262451171875,
      "learning_rate": 3.925949181288595e-05,
      "loss": 48.2448,
      "step": 44550
    },
    {
      "epoch": 0.18002803847816515,
      "grad_norm": 665.634521484375,
      "learning_rate": 3.92587387817927e-05,
      "loss": 67.9298,
      "step": 44560
    },
    {
      "epoch": 0.18006843974353276,
      "grad_norm": 1632.61328125,
      "learning_rate": 3.9257985375240435e-05,
      "loss": 108.7696,
      "step": 44570
    },
    {
      "epoch": 0.1801088410089004,
      "grad_norm": 790.5283813476562,
      "learning_rate": 3.925723159324383e-05,
      "loss": 98.9906,
      "step": 44580
    },
    {
      "epoch": 0.18014924227426804,
      "grad_norm": 500.9121398925781,
      "learning_rate": 3.925647743581758e-05,
      "loss": 68.065,
      "step": 44590
    },
    {
      "epoch": 0.18018964353963565,
      "grad_norm": 327.4957580566406,
      "learning_rate": 3.925572290297638e-05,
      "loss": 79.8086,
      "step": 44600
    },
    {
      "epoch": 0.1802300448050033,
      "grad_norm": 786.4631958007812,
      "learning_rate": 3.9254967994734954e-05,
      "loss": 87.2137,
      "step": 44610
    },
    {
      "epoch": 0.18027044607037093,
      "grad_norm": 743.5513305664062,
      "learning_rate": 3.925421271110802e-05,
      "loss": 96.5469,
      "step": 44620
    },
    {
      "epoch": 0.18031084733573854,
      "grad_norm": 1273.8321533203125,
      "learning_rate": 3.925345705211029e-05,
      "loss": 109.8698,
      "step": 44630
    },
    {
      "epoch": 0.18035124860110618,
      "grad_norm": 876.0645751953125,
      "learning_rate": 3.92527010177565e-05,
      "loss": 73.0859,
      "step": 44640
    },
    {
      "epoch": 0.18039164986647382,
      "grad_norm": 811.0771484375,
      "learning_rate": 3.9251944608061395e-05,
      "loss": 54.302,
      "step": 44650
    },
    {
      "epoch": 0.18043205113184144,
      "grad_norm": 1616.77685546875,
      "learning_rate": 3.9251187823039715e-05,
      "loss": 101.0898,
      "step": 44660
    },
    {
      "epoch": 0.18047245239720908,
      "grad_norm": 517.669189453125,
      "learning_rate": 3.925043066270622e-05,
      "loss": 55.0952,
      "step": 44670
    },
    {
      "epoch": 0.18051285366257672,
      "grad_norm": 445.7159118652344,
      "learning_rate": 3.9249673127075675e-05,
      "loss": 76.503,
      "step": 44680
    },
    {
      "epoch": 0.18055325492794436,
      "grad_norm": 1248.5576171875,
      "learning_rate": 3.924891521616283e-05,
      "loss": 91.7174,
      "step": 44690
    },
    {
      "epoch": 0.18059365619331197,
      "grad_norm": 721.5814819335938,
      "learning_rate": 3.924815692998248e-05,
      "loss": 82.0018,
      "step": 44700
    },
    {
      "epoch": 0.1806340574586796,
      "grad_norm": 2136.960693359375,
      "learning_rate": 3.9247398268549405e-05,
      "loss": 85.2536,
      "step": 44710
    },
    {
      "epoch": 0.18067445872404725,
      "grad_norm": 509.6750183105469,
      "learning_rate": 3.9246639231878384e-05,
      "loss": 102.8554,
      "step": 44720
    },
    {
      "epoch": 0.18071485998941486,
      "grad_norm": 865.4503173828125,
      "learning_rate": 3.924587981998422e-05,
      "loss": 83.089,
      "step": 44730
    },
    {
      "epoch": 0.1807552612547825,
      "grad_norm": 921.2974243164062,
      "learning_rate": 3.924512003288173e-05,
      "loss": 85.8162,
      "step": 44740
    },
    {
      "epoch": 0.18079566252015014,
      "grad_norm": 1070.1435546875,
      "learning_rate": 3.924435987058571e-05,
      "loss": 64.5532,
      "step": 44750
    },
    {
      "epoch": 0.18083606378551775,
      "grad_norm": 1231.571044921875,
      "learning_rate": 3.924359933311099e-05,
      "loss": 77.4203,
      "step": 44760
    },
    {
      "epoch": 0.1808764650508854,
      "grad_norm": 943.9061279296875,
      "learning_rate": 3.9242838420472393e-05,
      "loss": 104.9452,
      "step": 44770
    },
    {
      "epoch": 0.18091686631625303,
      "grad_norm": 963.1873779296875,
      "learning_rate": 3.9242077132684755e-05,
      "loss": 47.5799,
      "step": 44780
    },
    {
      "epoch": 0.18095726758162065,
      "grad_norm": 648.9423828125,
      "learning_rate": 3.9241315469762915e-05,
      "loss": 64.1059,
      "step": 44790
    },
    {
      "epoch": 0.18099766884698829,
      "grad_norm": 454.11474609375,
      "learning_rate": 3.924055343172172e-05,
      "loss": 94.2031,
      "step": 44800
    },
    {
      "epoch": 0.18103807011235593,
      "grad_norm": 713.5806274414062,
      "learning_rate": 3.9239791018576034e-05,
      "loss": 82.6018,
      "step": 44810
    },
    {
      "epoch": 0.18107847137772354,
      "grad_norm": 497.6768798828125,
      "learning_rate": 3.923902823034072e-05,
      "loss": 60.9297,
      "step": 44820
    },
    {
      "epoch": 0.18111887264309118,
      "grad_norm": 369.3504333496094,
      "learning_rate": 3.923826506703064e-05,
      "loss": 77.6983,
      "step": 44830
    },
    {
      "epoch": 0.18115927390845882,
      "grad_norm": 633.0171508789062,
      "learning_rate": 3.923750152866068e-05,
      "loss": 65.9964,
      "step": 44840
    },
    {
      "epoch": 0.18119967517382646,
      "grad_norm": 1098.6373291015625,
      "learning_rate": 3.9236737615245726e-05,
      "loss": 82.9418,
      "step": 44850
    },
    {
      "epoch": 0.18124007643919407,
      "grad_norm": 819.2840576171875,
      "learning_rate": 3.923597332680067e-05,
      "loss": 84.5484,
      "step": 44860
    },
    {
      "epoch": 0.1812804777045617,
      "grad_norm": 656.0801391601562,
      "learning_rate": 3.9235208663340404e-05,
      "loss": 93.1484,
      "step": 44870
    },
    {
      "epoch": 0.18132087896992935,
      "grad_norm": 2753.8642578125,
      "learning_rate": 3.923444362487985e-05,
      "loss": 54.1425,
      "step": 44880
    },
    {
      "epoch": 0.18136128023529696,
      "grad_norm": 1034.2335205078125,
      "learning_rate": 3.923367821143391e-05,
      "loss": 85.2575,
      "step": 44890
    },
    {
      "epoch": 0.1814016815006646,
      "grad_norm": 650.7776489257812,
      "learning_rate": 3.923291242301751e-05,
      "loss": 83.4574,
      "step": 44900
    },
    {
      "epoch": 0.18144208276603224,
      "grad_norm": 1090.718505859375,
      "learning_rate": 3.923214625964558e-05,
      "loss": 48.8431,
      "step": 44910
    },
    {
      "epoch": 0.18148248403139985,
      "grad_norm": 678.7628784179688,
      "learning_rate": 3.923137972133306e-05,
      "loss": 75.1513,
      "step": 44920
    },
    {
      "epoch": 0.1815228852967675,
      "grad_norm": 852.8782348632812,
      "learning_rate": 3.923061280809489e-05,
      "loss": 70.6103,
      "step": 44930
    },
    {
      "epoch": 0.18156328656213513,
      "grad_norm": 637.3016357421875,
      "learning_rate": 3.922984551994602e-05,
      "loss": 93.514,
      "step": 44940
    },
    {
      "epoch": 0.18160368782750275,
      "grad_norm": 600.9539184570312,
      "learning_rate": 3.922907785690142e-05,
      "loss": 43.0151,
      "step": 44950
    },
    {
      "epoch": 0.1816440890928704,
      "grad_norm": 1205.2518310546875,
      "learning_rate": 3.922830981897604e-05,
      "loss": 82.0866,
      "step": 44960
    },
    {
      "epoch": 0.18168449035823803,
      "grad_norm": 1370.759765625,
      "learning_rate": 3.922754140618486e-05,
      "loss": 63.6987,
      "step": 44970
    },
    {
      "epoch": 0.18172489162360564,
      "grad_norm": 664.2175903320312,
      "learning_rate": 3.9226772618542874e-05,
      "loss": 85.1669,
      "step": 44980
    },
    {
      "epoch": 0.18176529288897328,
      "grad_norm": 473.1387634277344,
      "learning_rate": 3.9226003456065045e-05,
      "loss": 64.7944,
      "step": 44990
    },
    {
      "epoch": 0.18180569415434092,
      "grad_norm": 971.3717651367188,
      "learning_rate": 3.922523391876638e-05,
      "loss": 82.8255,
      "step": 45000
    },
    {
      "epoch": 0.18184609541970856,
      "grad_norm": 791.2940063476562,
      "learning_rate": 3.9224464006661885e-05,
      "loss": 92.3966,
      "step": 45010
    },
    {
      "epoch": 0.18188649668507617,
      "grad_norm": 2711.322265625,
      "learning_rate": 3.922369371976657e-05,
      "loss": 97.3031,
      "step": 45020
    },
    {
      "epoch": 0.1819268979504438,
      "grad_norm": 500.21954345703125,
      "learning_rate": 3.9222923058095445e-05,
      "loss": 87.3782,
      "step": 45030
    },
    {
      "epoch": 0.18196729921581145,
      "grad_norm": 977.1084594726562,
      "learning_rate": 3.922215202166354e-05,
      "loss": 61.2811,
      "step": 45040
    },
    {
      "epoch": 0.18200770048117906,
      "grad_norm": 710.3567504882812,
      "learning_rate": 3.922138061048588e-05,
      "loss": 94.9242,
      "step": 45050
    },
    {
      "epoch": 0.1820481017465467,
      "grad_norm": 866.1929931640625,
      "learning_rate": 3.922060882457751e-05,
      "loss": 95.5972,
      "step": 45060
    },
    {
      "epoch": 0.18208850301191434,
      "grad_norm": 495.1076965332031,
      "learning_rate": 3.921983666395348e-05,
      "loss": 112.3219,
      "step": 45070
    },
    {
      "epoch": 0.18212890427728196,
      "grad_norm": 582.1326293945312,
      "learning_rate": 3.921906412862884e-05,
      "loss": 86.3041,
      "step": 45080
    },
    {
      "epoch": 0.1821693055426496,
      "grad_norm": 728.2086791992188,
      "learning_rate": 3.921829121861865e-05,
      "loss": 78.3341,
      "step": 45090
    },
    {
      "epoch": 0.18220970680801724,
      "grad_norm": 956.2645263671875,
      "learning_rate": 3.9217517933937974e-05,
      "loss": 94.1187,
      "step": 45100
    },
    {
      "epoch": 0.18225010807338485,
      "grad_norm": 1114.2762451171875,
      "learning_rate": 3.9216744274601895e-05,
      "loss": 84.1699,
      "step": 45110
    },
    {
      "epoch": 0.1822905093387525,
      "grad_norm": 477.52532958984375,
      "learning_rate": 3.921597024062549e-05,
      "loss": 110.3044,
      "step": 45120
    },
    {
      "epoch": 0.18233091060412013,
      "grad_norm": 801.48828125,
      "learning_rate": 3.921519583202386e-05,
      "loss": 65.7045,
      "step": 45130
    },
    {
      "epoch": 0.18237131186948774,
      "grad_norm": 406.8154602050781,
      "learning_rate": 3.921442104881209e-05,
      "loss": 65.1984,
      "step": 45140
    },
    {
      "epoch": 0.18241171313485538,
      "grad_norm": 597.41162109375,
      "learning_rate": 3.9213645891005284e-05,
      "loss": 74.5257,
      "step": 45150
    },
    {
      "epoch": 0.18245211440022302,
      "grad_norm": 1153.83837890625,
      "learning_rate": 3.921287035861857e-05,
      "loss": 94.8159,
      "step": 45160
    },
    {
      "epoch": 0.18249251566559066,
      "grad_norm": 621.4681396484375,
      "learning_rate": 3.921209445166705e-05,
      "loss": 58.4845,
      "step": 45170
    },
    {
      "epoch": 0.18253291693095827,
      "grad_norm": 1161.484130859375,
      "learning_rate": 3.9211318170165854e-05,
      "loss": 77.8721,
      "step": 45180
    },
    {
      "epoch": 0.1825733181963259,
      "grad_norm": 1004.8511352539062,
      "learning_rate": 3.9210541514130126e-05,
      "loss": 80.6697,
      "step": 45190
    },
    {
      "epoch": 0.18261371946169355,
      "grad_norm": 974.8493041992188,
      "learning_rate": 3.9209764483575e-05,
      "loss": 111.7471,
      "step": 45200
    },
    {
      "epoch": 0.18265412072706116,
      "grad_norm": 810.5325927734375,
      "learning_rate": 3.9208987078515625e-05,
      "loss": 81.0349,
      "step": 45210
    },
    {
      "epoch": 0.1826945219924288,
      "grad_norm": 514.47998046875,
      "learning_rate": 3.9208209298967156e-05,
      "loss": 49.2597,
      "step": 45220
    },
    {
      "epoch": 0.18273492325779644,
      "grad_norm": 864.6641235351562,
      "learning_rate": 3.920743114494476e-05,
      "loss": 81.931,
      "step": 45230
    },
    {
      "epoch": 0.18277532452316406,
      "grad_norm": 1110.948974609375,
      "learning_rate": 3.92066526164636e-05,
      "loss": 70.5063,
      "step": 45240
    },
    {
      "epoch": 0.1828157257885317,
      "grad_norm": 775.122314453125,
      "learning_rate": 3.9205873713538864e-05,
      "loss": 74.6872,
      "step": 45250
    },
    {
      "epoch": 0.18285612705389934,
      "grad_norm": 1156.91064453125,
      "learning_rate": 3.9205094436185734e-05,
      "loss": 76.9457,
      "step": 45260
    },
    {
      "epoch": 0.18289652831926695,
      "grad_norm": 1151.217041015625,
      "learning_rate": 3.9204314784419396e-05,
      "loss": 82.7603,
      "step": 45270
    },
    {
      "epoch": 0.1829369295846346,
      "grad_norm": 1386.2657470703125,
      "learning_rate": 3.9203534758255056e-05,
      "loss": 96.6438,
      "step": 45280
    },
    {
      "epoch": 0.18297733085000223,
      "grad_norm": 991.702880859375,
      "learning_rate": 3.920275435770791e-05,
      "loss": 75.2325,
      "step": 45290
    },
    {
      "epoch": 0.18301773211536984,
      "grad_norm": 782.2591552734375,
      "learning_rate": 3.92019735827932e-05,
      "loss": 73.0687,
      "step": 45300
    },
    {
      "epoch": 0.18305813338073748,
      "grad_norm": 850.5282592773438,
      "learning_rate": 3.920119243352612e-05,
      "loss": 60.5605,
      "step": 45310
    },
    {
      "epoch": 0.18309853464610512,
      "grad_norm": 1552.738037109375,
      "learning_rate": 3.9200410909921903e-05,
      "loss": 65.0465,
      "step": 45320
    },
    {
      "epoch": 0.18313893591147276,
      "grad_norm": 1523.2933349609375,
      "learning_rate": 3.91996290119958e-05,
      "loss": 82.3612,
      "step": 45330
    },
    {
      "epoch": 0.18317933717684037,
      "grad_norm": 498.4686279296875,
      "learning_rate": 3.919884673976304e-05,
      "loss": 79.3921,
      "step": 45340
    },
    {
      "epoch": 0.183219738442208,
      "grad_norm": 1048.4293212890625,
      "learning_rate": 3.9198064093238874e-05,
      "loss": 109.0177,
      "step": 45350
    },
    {
      "epoch": 0.18326013970757565,
      "grad_norm": 1083.177734375,
      "learning_rate": 3.9197281072438576e-05,
      "loss": 112.6941,
      "step": 45360
    },
    {
      "epoch": 0.18330054097294327,
      "grad_norm": 726.718505859375,
      "learning_rate": 3.919649767737739e-05,
      "loss": 70.6048,
      "step": 45370
    },
    {
      "epoch": 0.1833409422383109,
      "grad_norm": 895.0009765625,
      "learning_rate": 3.9195713908070606e-05,
      "loss": 73.5802,
      "step": 45380
    },
    {
      "epoch": 0.18338134350367855,
      "grad_norm": 878.7771606445312,
      "learning_rate": 3.919492976453349e-05,
      "loss": 114.3825,
      "step": 45390
    },
    {
      "epoch": 0.18342174476904616,
      "grad_norm": 1377.43896484375,
      "learning_rate": 3.9194145246781336e-05,
      "loss": 103.3703,
      "step": 45400
    },
    {
      "epoch": 0.1834621460344138,
      "grad_norm": 650.5206909179688,
      "learning_rate": 3.919336035482944e-05,
      "loss": 86.7826,
      "step": 45410
    },
    {
      "epoch": 0.18350254729978144,
      "grad_norm": 680.419189453125,
      "learning_rate": 3.9192575088693104e-05,
      "loss": 89.4733,
      "step": 45420
    },
    {
      "epoch": 0.18354294856514905,
      "grad_norm": 864.0675048828125,
      "learning_rate": 3.9191789448387634e-05,
      "loss": 74.6113,
      "step": 45430
    },
    {
      "epoch": 0.1835833498305167,
      "grad_norm": 359.46331787109375,
      "learning_rate": 3.919100343392835e-05,
      "loss": 61.1938,
      "step": 45440
    },
    {
      "epoch": 0.18362375109588433,
      "grad_norm": 260.6347351074219,
      "learning_rate": 3.919021704533058e-05,
      "loss": 101.0921,
      "step": 45450
    },
    {
      "epoch": 0.18366415236125194,
      "grad_norm": 714.068603515625,
      "learning_rate": 3.918943028260964e-05,
      "loss": 91.7542,
      "step": 45460
    },
    {
      "epoch": 0.18370455362661958,
      "grad_norm": 1035.927490234375,
      "learning_rate": 3.918864314578087e-05,
      "loss": 93.5492,
      "step": 45470
    },
    {
      "epoch": 0.18374495489198722,
      "grad_norm": 1542.9630126953125,
      "learning_rate": 3.918785563485964e-05,
      "loss": 106.8872,
      "step": 45480
    },
    {
      "epoch": 0.18378535615735486,
      "grad_norm": 806.6215209960938,
      "learning_rate": 3.918706774986128e-05,
      "loss": 92.174,
      "step": 45490
    },
    {
      "epoch": 0.18382575742272247,
      "grad_norm": 764.0255737304688,
      "learning_rate": 3.918627949080115e-05,
      "loss": 100.8685,
      "step": 45500
    },
    {
      "epoch": 0.18386615868809011,
      "grad_norm": 836.418212890625,
      "learning_rate": 3.918549085769464e-05,
      "loss": 75.9819,
      "step": 45510
    },
    {
      "epoch": 0.18390655995345775,
      "grad_norm": 670.6646728515625,
      "learning_rate": 3.918470185055709e-05,
      "loss": 63.5554,
      "step": 45520
    },
    {
      "epoch": 0.18394696121882537,
      "grad_norm": 1930.96435546875,
      "learning_rate": 3.91839124694039e-05,
      "loss": 109.9029,
      "step": 45530
    },
    {
      "epoch": 0.183987362484193,
      "grad_norm": 1163.8868408203125,
      "learning_rate": 3.918312271425047e-05,
      "loss": 85.8055,
      "step": 45540
    },
    {
      "epoch": 0.18402776374956065,
      "grad_norm": 786.4589233398438,
      "learning_rate": 3.9182332585112186e-05,
      "loss": 82.8965,
      "step": 45550
    },
    {
      "epoch": 0.18406816501492826,
      "grad_norm": 845.56298828125,
      "learning_rate": 3.918154208200445e-05,
      "loss": 123.4536,
      "step": 45560
    },
    {
      "epoch": 0.1841085662802959,
      "grad_norm": 677.166259765625,
      "learning_rate": 3.918075120494267e-05,
      "loss": 87.6609,
      "step": 45570
    },
    {
      "epoch": 0.18414896754566354,
      "grad_norm": 1088.3443603515625,
      "learning_rate": 3.9179959953942283e-05,
      "loss": 119.1445,
      "step": 45580
    },
    {
      "epoch": 0.18418936881103115,
      "grad_norm": 1323.2698974609375,
      "learning_rate": 3.917916832901869e-05,
      "loss": 88.3157,
      "step": 45590
    },
    {
      "epoch": 0.1842297700763988,
      "grad_norm": 1136.53759765625,
      "learning_rate": 3.917837633018734e-05,
      "loss": 93.4334,
      "step": 45600
    },
    {
      "epoch": 0.18427017134176643,
      "grad_norm": 808.8739624023438,
      "learning_rate": 3.917758395746368e-05,
      "loss": 65.8987,
      "step": 45610
    },
    {
      "epoch": 0.18431057260713404,
      "grad_norm": 850.0430908203125,
      "learning_rate": 3.9176791210863136e-05,
      "loss": 154.3409,
      "step": 45620
    },
    {
      "epoch": 0.18435097387250168,
      "grad_norm": 1013.5833740234375,
      "learning_rate": 3.917599809040117e-05,
      "loss": 119.8127,
      "step": 45630
    },
    {
      "epoch": 0.18439137513786932,
      "grad_norm": 1381.6162109375,
      "learning_rate": 3.917520459609326e-05,
      "loss": 86.2685,
      "step": 45640
    },
    {
      "epoch": 0.18443177640323694,
      "grad_norm": 307.11041259765625,
      "learning_rate": 3.9174410727954856e-05,
      "loss": 66.6993,
      "step": 45650
    },
    {
      "epoch": 0.18447217766860458,
      "grad_norm": 1673.486083984375,
      "learning_rate": 3.917361648600145e-05,
      "loss": 88.8105,
      "step": 45660
    },
    {
      "epoch": 0.18451257893397222,
      "grad_norm": 1405.7908935546875,
      "learning_rate": 3.9172821870248524e-05,
      "loss": 88.505,
      "step": 45670
    },
    {
      "epoch": 0.18455298019933986,
      "grad_norm": 954.478271484375,
      "learning_rate": 3.917202688071155e-05,
      "loss": 58.216,
      "step": 45680
    },
    {
      "epoch": 0.18459338146470747,
      "grad_norm": 1570.70703125,
      "learning_rate": 3.917123151740605e-05,
      "loss": 102.5098,
      "step": 45690
    },
    {
      "epoch": 0.1846337827300751,
      "grad_norm": 806.4921264648438,
      "learning_rate": 3.917043578034752e-05,
      "loss": 60.7017,
      "step": 45700
    },
    {
      "epoch": 0.18467418399544275,
      "grad_norm": 2194.05908203125,
      "learning_rate": 3.9169639669551476e-05,
      "loss": 67.0707,
      "step": 45710
    },
    {
      "epoch": 0.18471458526081036,
      "grad_norm": 1100.316162109375,
      "learning_rate": 3.9168843185033434e-05,
      "loss": 59.0874,
      "step": 45720
    },
    {
      "epoch": 0.184754986526178,
      "grad_norm": 1363.7559814453125,
      "learning_rate": 3.9168046326808934e-05,
      "loss": 65.7012,
      "step": 45730
    },
    {
      "epoch": 0.18479538779154564,
      "grad_norm": 704.1640014648438,
      "learning_rate": 3.9167249094893495e-05,
      "loss": 60.0356,
      "step": 45740
    },
    {
      "epoch": 0.18483578905691325,
      "grad_norm": 275.0506591796875,
      "learning_rate": 3.916645148930267e-05,
      "loss": 92.7959,
      "step": 45750
    },
    {
      "epoch": 0.1848761903222809,
      "grad_norm": 422.1753845214844,
      "learning_rate": 3.9165653510052004e-05,
      "loss": 83.3863,
      "step": 45760
    },
    {
      "epoch": 0.18491659158764853,
      "grad_norm": 612.3901977539062,
      "learning_rate": 3.9164855157157055e-05,
      "loss": 64.9848,
      "step": 45770
    },
    {
      "epoch": 0.18495699285301614,
      "grad_norm": 1202.4676513671875,
      "learning_rate": 3.916405643063339e-05,
      "loss": 80.2309,
      "step": 45780
    },
    {
      "epoch": 0.18499739411838378,
      "grad_norm": 452.2706298828125,
      "learning_rate": 3.9163257330496576e-05,
      "loss": 53.2638,
      "step": 45790
    },
    {
      "epoch": 0.18503779538375142,
      "grad_norm": 466.4120178222656,
      "learning_rate": 3.91624578567622e-05,
      "loss": 72.9452,
      "step": 45800
    },
    {
      "epoch": 0.18507819664911904,
      "grad_norm": 1306.1180419921875,
      "learning_rate": 3.9161658009445834e-05,
      "loss": 110.0645,
      "step": 45810
    },
    {
      "epoch": 0.18511859791448668,
      "grad_norm": 980.823974609375,
      "learning_rate": 3.9160857788563085e-05,
      "loss": 105.6,
      "step": 45820
    },
    {
      "epoch": 0.18515899917985432,
      "grad_norm": 313.7354431152344,
      "learning_rate": 3.9160057194129544e-05,
      "loss": 57.1493,
      "step": 45830
    },
    {
      "epoch": 0.18519940044522196,
      "grad_norm": 578.3187866210938,
      "learning_rate": 3.915925622616083e-05,
      "loss": 78.0657,
      "step": 45840
    },
    {
      "epoch": 0.18523980171058957,
      "grad_norm": 960.7958984375,
      "learning_rate": 3.915845488467255e-05,
      "loss": 80.249,
      "step": 45850
    },
    {
      "epoch": 0.1852802029759572,
      "grad_norm": 739.0011596679688,
      "learning_rate": 3.915765316968033e-05,
      "loss": 96.2227,
      "step": 45860
    },
    {
      "epoch": 0.18532060424132485,
      "grad_norm": 924.9442749023438,
      "learning_rate": 3.91568510811998e-05,
      "loss": 47.8373,
      "step": 45870
    },
    {
      "epoch": 0.18536100550669246,
      "grad_norm": 935.7515258789062,
      "learning_rate": 3.915604861924659e-05,
      "loss": 65.6091,
      "step": 45880
    },
    {
      "epoch": 0.1854014067720601,
      "grad_norm": 1007.1822509765625,
      "learning_rate": 3.915524578383635e-05,
      "loss": 62.2943,
      "step": 45890
    },
    {
      "epoch": 0.18544180803742774,
      "grad_norm": 569.2777099609375,
      "learning_rate": 3.9154442574984735e-05,
      "loss": 72.1652,
      "step": 45900
    },
    {
      "epoch": 0.18548220930279535,
      "grad_norm": 564.3727416992188,
      "learning_rate": 3.91536389927074e-05,
      "loss": 83.1148,
      "step": 45910
    },
    {
      "epoch": 0.185522610568163,
      "grad_norm": 1044.73779296875,
      "learning_rate": 3.9152835037020013e-05,
      "loss": 79.5553,
      "step": 45920
    },
    {
      "epoch": 0.18556301183353063,
      "grad_norm": 1104.0501708984375,
      "learning_rate": 3.915203070793825e-05,
      "loss": 75.241,
      "step": 45930
    },
    {
      "epoch": 0.18560341309889825,
      "grad_norm": 1001.4999389648438,
      "learning_rate": 3.9151226005477786e-05,
      "loss": 100.4497,
      "step": 45940
    },
    {
      "epoch": 0.18564381436426589,
      "grad_norm": 1332.5491943359375,
      "learning_rate": 3.91504209296543e-05,
      "loss": 103.7096,
      "step": 45950
    },
    {
      "epoch": 0.18568421562963353,
      "grad_norm": 1234.172119140625,
      "learning_rate": 3.914961548048351e-05,
      "loss": 74.3093,
      "step": 45960
    },
    {
      "epoch": 0.18572461689500114,
      "grad_norm": 1303.2974853515625,
      "learning_rate": 3.91488096579811e-05,
      "loss": 74.6654,
      "step": 45970
    },
    {
      "epoch": 0.18576501816036878,
      "grad_norm": 2073.360107421875,
      "learning_rate": 3.914800346216279e-05,
      "loss": 68.3085,
      "step": 45980
    },
    {
      "epoch": 0.18580541942573642,
      "grad_norm": 485.2785339355469,
      "learning_rate": 3.9147196893044305e-05,
      "loss": 68.9047,
      "step": 45990
    },
    {
      "epoch": 0.18584582069110406,
      "grad_norm": 1112.5963134765625,
      "learning_rate": 3.9146389950641345e-05,
      "loss": 69.142,
      "step": 46000
    },
    {
      "epoch": 0.18588622195647167,
      "grad_norm": 1303.632080078125,
      "learning_rate": 3.914558263496966e-05,
      "loss": 65.932,
      "step": 46010
    },
    {
      "epoch": 0.1859266232218393,
      "grad_norm": 855.09423828125,
      "learning_rate": 3.914477494604499e-05,
      "loss": 48.3135,
      "step": 46020
    },
    {
      "epoch": 0.18596702448720695,
      "grad_norm": 1216.570556640625,
      "learning_rate": 3.914396688388308e-05,
      "loss": 48.3435,
      "step": 46030
    },
    {
      "epoch": 0.18600742575257456,
      "grad_norm": 711.9722900390625,
      "learning_rate": 3.914315844849966e-05,
      "loss": 86.1562,
      "step": 46040
    },
    {
      "epoch": 0.1860478270179422,
      "grad_norm": 703.8426513671875,
      "learning_rate": 3.914234963991053e-05,
      "loss": 102.8623,
      "step": 46050
    },
    {
      "epoch": 0.18608822828330984,
      "grad_norm": 1274.6202392578125,
      "learning_rate": 3.914154045813143e-05,
      "loss": 79.2652,
      "step": 46060
    },
    {
      "epoch": 0.18612862954867745,
      "grad_norm": 709.7684936523438,
      "learning_rate": 3.914073090317814e-05,
      "loss": 117.7384,
      "step": 46070
    },
    {
      "epoch": 0.1861690308140451,
      "grad_norm": 422.33416748046875,
      "learning_rate": 3.913992097506645e-05,
      "loss": 82.4315,
      "step": 46080
    },
    {
      "epoch": 0.18620943207941273,
      "grad_norm": 0.0,
      "learning_rate": 3.9139110673812155e-05,
      "loss": 64.9715,
      "step": 46090
    },
    {
      "epoch": 0.18624983334478035,
      "grad_norm": 1187.5250244140625,
      "learning_rate": 3.913829999943103e-05,
      "loss": 66.9666,
      "step": 46100
    },
    {
      "epoch": 0.186290234610148,
      "grad_norm": 1188.3201904296875,
      "learning_rate": 3.91374889519389e-05,
      "loss": 121.7604,
      "step": 46110
    },
    {
      "epoch": 0.18633063587551563,
      "grad_norm": 1243.7998046875,
      "learning_rate": 3.913667753135157e-05,
      "loss": 106.2775,
      "step": 46120
    },
    {
      "epoch": 0.18637103714088324,
      "grad_norm": 1100.300048828125,
      "learning_rate": 3.913586573768486e-05,
      "loss": 67.3994,
      "step": 46130
    },
    {
      "epoch": 0.18641143840625088,
      "grad_norm": 580.4454345703125,
      "learning_rate": 3.91350535709546e-05,
      "loss": 79.131,
      "step": 46140
    },
    {
      "epoch": 0.18645183967161852,
      "grad_norm": 1536.91650390625,
      "learning_rate": 3.9134241031176614e-05,
      "loss": 74.0721,
      "step": 46150
    },
    {
      "epoch": 0.18649224093698616,
      "grad_norm": 1414.6435546875,
      "learning_rate": 3.9133428118366746e-05,
      "loss": 91.2891,
      "step": 46160
    },
    {
      "epoch": 0.18653264220235377,
      "grad_norm": 4572.24169921875,
      "learning_rate": 3.9132614832540854e-05,
      "loss": 122.7679,
      "step": 46170
    },
    {
      "epoch": 0.1865730434677214,
      "grad_norm": 670.1581420898438,
      "learning_rate": 3.913180117371478e-05,
      "loss": 69.168,
      "step": 46180
    },
    {
      "epoch": 0.18661344473308905,
      "grad_norm": 1101.5228271484375,
      "learning_rate": 3.913098714190439e-05,
      "loss": 86.6942,
      "step": 46190
    },
    {
      "epoch": 0.18665384599845666,
      "grad_norm": 890.9598999023438,
      "learning_rate": 3.913017273712556e-05,
      "loss": 94.4553,
      "step": 46200
    },
    {
      "epoch": 0.1866942472638243,
      "grad_norm": 1160.6763916015625,
      "learning_rate": 3.9129357959394166e-05,
      "loss": 175.9972,
      "step": 46210
    },
    {
      "epoch": 0.18673464852919194,
      "grad_norm": 561.4642333984375,
      "learning_rate": 3.9128542808726084e-05,
      "loss": 61.6344,
      "step": 46220
    },
    {
      "epoch": 0.18677504979455956,
      "grad_norm": 2253.88818359375,
      "learning_rate": 3.912772728513722e-05,
      "loss": 80.0333,
      "step": 46230
    },
    {
      "epoch": 0.1868154510599272,
      "grad_norm": 0.0,
      "learning_rate": 3.912691138864346e-05,
      "loss": 88.1769,
      "step": 46240
    },
    {
      "epoch": 0.18685585232529484,
      "grad_norm": 0.0,
      "learning_rate": 3.912609511926071e-05,
      "loss": 75.8361,
      "step": 46250
    },
    {
      "epoch": 0.18689625359066245,
      "grad_norm": 1239.348876953125,
      "learning_rate": 3.912527847700489e-05,
      "loss": 103.7301,
      "step": 46260
    },
    {
      "epoch": 0.1869366548560301,
      "grad_norm": 754.636474609375,
      "learning_rate": 3.912446146189193e-05,
      "loss": 81.214,
      "step": 46270
    },
    {
      "epoch": 0.18697705612139773,
      "grad_norm": 602.5750732421875,
      "learning_rate": 3.912364407393774e-05,
      "loss": 68.1354,
      "step": 46280
    },
    {
      "epoch": 0.18701745738676534,
      "grad_norm": 1123.705322265625,
      "learning_rate": 3.912282631315827e-05,
      "loss": 87.4514,
      "step": 46290
    },
    {
      "epoch": 0.18705785865213298,
      "grad_norm": 913.4268188476562,
      "learning_rate": 3.912200817956945e-05,
      "loss": 77.5673,
      "step": 46300
    },
    {
      "epoch": 0.18709825991750062,
      "grad_norm": 1332.4677734375,
      "learning_rate": 3.912118967318724e-05,
      "loss": 96.7064,
      "step": 46310
    },
    {
      "epoch": 0.18713866118286826,
      "grad_norm": 861.162841796875,
      "learning_rate": 3.912037079402759e-05,
      "loss": 65.3486,
      "step": 46320
    },
    {
      "epoch": 0.18717906244823587,
      "grad_norm": 568.9047241210938,
      "learning_rate": 3.911955154210646e-05,
      "loss": 47.261,
      "step": 46330
    },
    {
      "epoch": 0.1872194637136035,
      "grad_norm": 523.025634765625,
      "learning_rate": 3.9118731917439844e-05,
      "loss": 52.6651,
      "step": 46340
    },
    {
      "epoch": 0.18725986497897115,
      "grad_norm": 1224.9736328125,
      "learning_rate": 3.9117911920043695e-05,
      "loss": 92.0736,
      "step": 46350
    },
    {
      "epoch": 0.18730026624433876,
      "grad_norm": 824.1813354492188,
      "learning_rate": 3.9117091549934014e-05,
      "loss": 75.8925,
      "step": 46360
    },
    {
      "epoch": 0.1873406675097064,
      "grad_norm": 2099.2509765625,
      "learning_rate": 3.9116270807126794e-05,
      "loss": 88.4477,
      "step": 46370
    },
    {
      "epoch": 0.18738106877507404,
      "grad_norm": 724.2691650390625,
      "learning_rate": 3.9115449691638027e-05,
      "loss": 94.5531,
      "step": 46380
    },
    {
      "epoch": 0.18742147004044166,
      "grad_norm": 922.3722534179688,
      "learning_rate": 3.911462820348374e-05,
      "loss": 75.4537,
      "step": 46390
    },
    {
      "epoch": 0.1874618713058093,
      "grad_norm": 829.753173828125,
      "learning_rate": 3.9113806342679926e-05,
      "loss": 73.5001,
      "step": 46400
    },
    {
      "epoch": 0.18750227257117694,
      "grad_norm": 1990.830078125,
      "learning_rate": 3.911298410924262e-05,
      "loss": 136.7985,
      "step": 46410
    },
    {
      "epoch": 0.18754267383654455,
      "grad_norm": 400.29083251953125,
      "learning_rate": 3.911216150318784e-05,
      "loss": 77.2953,
      "step": 46420
    },
    {
      "epoch": 0.1875830751019122,
      "grad_norm": 1748.0518798828125,
      "learning_rate": 3.911133852453164e-05,
      "loss": 74.3119,
      "step": 46430
    },
    {
      "epoch": 0.18762347636727983,
      "grad_norm": 828.0217895507812,
      "learning_rate": 3.911051517329005e-05,
      "loss": 65.2073,
      "step": 46440
    },
    {
      "epoch": 0.18766387763264744,
      "grad_norm": 853.9519653320312,
      "learning_rate": 3.910969144947913e-05,
      "loss": 94.6754,
      "step": 46450
    },
    {
      "epoch": 0.18770427889801508,
      "grad_norm": 508.40576171875,
      "learning_rate": 3.910886735311494e-05,
      "loss": 50.4667,
      "step": 46460
    },
    {
      "epoch": 0.18774468016338272,
      "grad_norm": 441.7326965332031,
      "learning_rate": 3.910804288421355e-05,
      "loss": 108.5341,
      "step": 46470
    },
    {
      "epoch": 0.18778508142875036,
      "grad_norm": 795.6704711914062,
      "learning_rate": 3.910721804279101e-05,
      "loss": 103.7186,
      "step": 46480
    },
    {
      "epoch": 0.18782548269411797,
      "grad_norm": 386.6521301269531,
      "learning_rate": 3.910639282886343e-05,
      "loss": 82.726,
      "step": 46490
    },
    {
      "epoch": 0.1878658839594856,
      "grad_norm": 823.4021606445312,
      "learning_rate": 3.9105567242446874e-05,
      "loss": 56.3026,
      "step": 46500
    },
    {
      "epoch": 0.18790628522485325,
      "grad_norm": 1041.70361328125,
      "learning_rate": 3.9104741283557456e-05,
      "loss": 49.9035,
      "step": 46510
    },
    {
      "epoch": 0.18794668649022087,
      "grad_norm": 1130.276611328125,
      "learning_rate": 3.910391495221127e-05,
      "loss": 56.8048,
      "step": 46520
    },
    {
      "epoch": 0.1879870877555885,
      "grad_norm": 1045.17822265625,
      "learning_rate": 3.9103088248424425e-05,
      "loss": 80.3857,
      "step": 46530
    },
    {
      "epoch": 0.18802748902095615,
      "grad_norm": 702.6853637695312,
      "learning_rate": 3.910226117221305e-05,
      "loss": 64.5383,
      "step": 46540
    },
    {
      "epoch": 0.18806789028632376,
      "grad_norm": 615.02197265625,
      "learning_rate": 3.9101433723593245e-05,
      "loss": 66.2929,
      "step": 46550
    },
    {
      "epoch": 0.1881082915516914,
      "grad_norm": 585.1364135742188,
      "learning_rate": 3.910060590258116e-05,
      "loss": 60.8131,
      "step": 46560
    },
    {
      "epoch": 0.18814869281705904,
      "grad_norm": 1662.8502197265625,
      "learning_rate": 3.909977770919293e-05,
      "loss": 59.843,
      "step": 46570
    },
    {
      "epoch": 0.18818909408242665,
      "grad_norm": 641.4120483398438,
      "learning_rate": 3.9098949143444704e-05,
      "loss": 146.1808,
      "step": 46580
    },
    {
      "epoch": 0.1882294953477943,
      "grad_norm": 924.935791015625,
      "learning_rate": 3.909812020535262e-05,
      "loss": 68.6732,
      "step": 46590
    },
    {
      "epoch": 0.18826989661316193,
      "grad_norm": 573.886962890625,
      "learning_rate": 3.9097290894932866e-05,
      "loss": 97.9553,
      "step": 46600
    },
    {
      "epoch": 0.18831029787852954,
      "grad_norm": 476.7160339355469,
      "learning_rate": 3.909646121220158e-05,
      "loss": 103.4015,
      "step": 46610
    },
    {
      "epoch": 0.18835069914389718,
      "grad_norm": 693.3080444335938,
      "learning_rate": 3.9095631157174956e-05,
      "loss": 58.9891,
      "step": 46620
    },
    {
      "epoch": 0.18839110040926482,
      "grad_norm": 1162.9171142578125,
      "learning_rate": 3.909480072986918e-05,
      "loss": 71.2949,
      "step": 46630
    },
    {
      "epoch": 0.18843150167463246,
      "grad_norm": 395.7523193359375,
      "learning_rate": 3.9093969930300424e-05,
      "loss": 102.4105,
      "step": 46640
    },
    {
      "epoch": 0.18847190294000007,
      "grad_norm": 1078.971435546875,
      "learning_rate": 3.909313875848489e-05,
      "loss": 85.173,
      "step": 46650
    },
    {
      "epoch": 0.18851230420536771,
      "grad_norm": 75.84223175048828,
      "learning_rate": 3.9092307214438795e-05,
      "loss": 72.504,
      "step": 46660
    },
    {
      "epoch": 0.18855270547073535,
      "grad_norm": 657.8626708984375,
      "learning_rate": 3.909147529817834e-05,
      "loss": 56.7639,
      "step": 46670
    },
    {
      "epoch": 0.18859310673610297,
      "grad_norm": 825.9349365234375,
      "learning_rate": 3.909064300971975e-05,
      "loss": 93.713,
      "step": 46680
    },
    {
      "epoch": 0.1886335080014706,
      "grad_norm": 847.5474243164062,
      "learning_rate": 3.908981034907923e-05,
      "loss": 61.4545,
      "step": 46690
    },
    {
      "epoch": 0.18867390926683825,
      "grad_norm": 549.8899536132812,
      "learning_rate": 3.9088977316273044e-05,
      "loss": 75.5487,
      "step": 46700
    },
    {
      "epoch": 0.18871431053220586,
      "grad_norm": 408.15753173828125,
      "learning_rate": 3.908814391131741e-05,
      "loss": 132.707,
      "step": 46710
    },
    {
      "epoch": 0.1887547117975735,
      "grad_norm": 572.0899658203125,
      "learning_rate": 3.908731013422859e-05,
      "loss": 107.9505,
      "step": 46720
    },
    {
      "epoch": 0.18879511306294114,
      "grad_norm": 1054.216064453125,
      "learning_rate": 3.908647598502282e-05,
      "loss": 78.4994,
      "step": 46730
    },
    {
      "epoch": 0.18883551432830875,
      "grad_norm": 869.3137817382812,
      "learning_rate": 3.908564146371639e-05,
      "loss": 64.2145,
      "step": 46740
    },
    {
      "epoch": 0.1888759155936764,
      "grad_norm": 4045.420654296875,
      "learning_rate": 3.908480657032554e-05,
      "loss": 80.4099,
      "step": 46750
    },
    {
      "epoch": 0.18891631685904403,
      "grad_norm": 1174.1817626953125,
      "learning_rate": 3.9083971304866566e-05,
      "loss": 63.7517,
      "step": 46760
    },
    {
      "epoch": 0.18895671812441164,
      "grad_norm": 1713.68896484375,
      "learning_rate": 3.908313566735575e-05,
      "loss": 104.0484,
      "step": 46770
    },
    {
      "epoch": 0.18899711938977928,
      "grad_norm": 3929.40087890625,
      "learning_rate": 3.908229965780937e-05,
      "loss": 114.1248,
      "step": 46780
    },
    {
      "epoch": 0.18903752065514692,
      "grad_norm": 1123.20263671875,
      "learning_rate": 3.908146327624374e-05,
      "loss": 97.4944,
      "step": 46790
    },
    {
      "epoch": 0.18907792192051456,
      "grad_norm": 893.2198486328125,
      "learning_rate": 3.908062652267516e-05,
      "loss": 62.3123,
      "step": 46800
    },
    {
      "epoch": 0.18911832318588218,
      "grad_norm": 1673.565185546875,
      "learning_rate": 3.907978939711995e-05,
      "loss": 70.2366,
      "step": 46810
    },
    {
      "epoch": 0.18915872445124982,
      "grad_norm": 944.0204467773438,
      "learning_rate": 3.907895189959441e-05,
      "loss": 65.0089,
      "step": 46820
    },
    {
      "epoch": 0.18919912571661746,
      "grad_norm": 491.1232604980469,
      "learning_rate": 3.907811403011488e-05,
      "loss": 71.3438,
      "step": 46830
    },
    {
      "epoch": 0.18923952698198507,
      "grad_norm": 865.6963500976562,
      "learning_rate": 3.9077275788697704e-05,
      "loss": 95.7543,
      "step": 46840
    },
    {
      "epoch": 0.1892799282473527,
      "grad_norm": 251.933349609375,
      "learning_rate": 3.907643717535921e-05,
      "loss": 68.4161,
      "step": 46850
    },
    {
      "epoch": 0.18932032951272035,
      "grad_norm": 1765.4664306640625,
      "learning_rate": 3.9075598190115756e-05,
      "loss": 68.8602,
      "step": 46860
    },
    {
      "epoch": 0.18936073077808796,
      "grad_norm": 1550.0150146484375,
      "learning_rate": 3.9074758832983685e-05,
      "loss": 93.8396,
      "step": 46870
    },
    {
      "epoch": 0.1894011320434556,
      "grad_norm": 1457.9483642578125,
      "learning_rate": 3.9073919103979383e-05,
      "loss": 79.1825,
      "step": 46880
    },
    {
      "epoch": 0.18944153330882324,
      "grad_norm": 624.1560668945312,
      "learning_rate": 3.90730790031192e-05,
      "loss": 91.3734,
      "step": 46890
    },
    {
      "epoch": 0.18948193457419085,
      "grad_norm": 1067.4862060546875,
      "learning_rate": 3.9072238530419525e-05,
      "loss": 103.1383,
      "step": 46900
    },
    {
      "epoch": 0.1895223358395585,
      "grad_norm": 426.91864013671875,
      "learning_rate": 3.907139768589674e-05,
      "loss": 75.1453,
      "step": 46910
    },
    {
      "epoch": 0.18956273710492613,
      "grad_norm": 1307.3526611328125,
      "learning_rate": 3.907055646956724e-05,
      "loss": 80.5333,
      "step": 46920
    },
    {
      "epoch": 0.18960313837029374,
      "grad_norm": 818.2450561523438,
      "learning_rate": 3.9069714881447416e-05,
      "loss": 66.8826,
      "step": 46930
    },
    {
      "epoch": 0.18964353963566138,
      "grad_norm": 452.9537353515625,
      "learning_rate": 3.906887292155369e-05,
      "loss": 92.7734,
      "step": 46940
    },
    {
      "epoch": 0.18968394090102902,
      "grad_norm": 1140.84765625,
      "learning_rate": 3.906803058990247e-05,
      "loss": 84.7196,
      "step": 46950
    },
    {
      "epoch": 0.18972434216639666,
      "grad_norm": 1345.89453125,
      "learning_rate": 3.9067187886510173e-05,
      "loss": 77.4215,
      "step": 46960
    },
    {
      "epoch": 0.18976474343176428,
      "grad_norm": 636.783203125,
      "learning_rate": 3.9066344811393226e-05,
      "loss": 66.3102,
      "step": 46970
    },
    {
      "epoch": 0.18980514469713192,
      "grad_norm": 902.4142456054688,
      "learning_rate": 3.906550136456808e-05,
      "loss": 69.7095,
      "step": 46980
    },
    {
      "epoch": 0.18984554596249956,
      "grad_norm": 1144.4429931640625,
      "learning_rate": 3.9064657546051175e-05,
      "loss": 71.0341,
      "step": 46990
    },
    {
      "epoch": 0.18988594722786717,
      "grad_norm": 1247.4844970703125,
      "learning_rate": 3.9063813355858944e-05,
      "loss": 89.9072,
      "step": 47000
    },
    {
      "epoch": 0.1899263484932348,
      "grad_norm": 1052.9495849609375,
      "learning_rate": 3.906296879400786e-05,
      "loss": 97.3577,
      "step": 47010
    },
    {
      "epoch": 0.18996674975860245,
      "grad_norm": 1629.934814453125,
      "learning_rate": 3.906212386051439e-05,
      "loss": 81.06,
      "step": 47020
    },
    {
      "epoch": 0.19000715102397006,
      "grad_norm": 238.38658142089844,
      "learning_rate": 3.9061278555394995e-05,
      "loss": 60.9792,
      "step": 47030
    },
    {
      "epoch": 0.1900475522893377,
      "grad_norm": 767.5965576171875,
      "learning_rate": 3.906043287866617e-05,
      "loss": 92.079,
      "step": 47040
    },
    {
      "epoch": 0.19008795355470534,
      "grad_norm": 880.5455322265625,
      "learning_rate": 3.905958683034438e-05,
      "loss": 71.8466,
      "step": 47050
    },
    {
      "epoch": 0.19012835482007295,
      "grad_norm": 425.3398132324219,
      "learning_rate": 3.905874041044614e-05,
      "loss": 77.9913,
      "step": 47060
    },
    {
      "epoch": 0.1901687560854406,
      "grad_norm": 2035.0830078125,
      "learning_rate": 3.9057893618987946e-05,
      "loss": 72.4353,
      "step": 47070
    },
    {
      "epoch": 0.19020915735080823,
      "grad_norm": 568.90478515625,
      "learning_rate": 3.90570464559863e-05,
      "loss": 72.8722,
      "step": 47080
    },
    {
      "epoch": 0.19024955861617585,
      "grad_norm": 818.4541625976562,
      "learning_rate": 3.905619892145772e-05,
      "loss": 75.254,
      "step": 47090
    },
    {
      "epoch": 0.19028995988154349,
      "grad_norm": 646.7007446289062,
      "learning_rate": 3.9055351015418735e-05,
      "loss": 84.8752,
      "step": 47100
    },
    {
      "epoch": 0.19033036114691113,
      "grad_norm": 1430.896240234375,
      "learning_rate": 3.905450273788587e-05,
      "loss": 65.986,
      "step": 47110
    },
    {
      "epoch": 0.19037076241227877,
      "grad_norm": 610.7198486328125,
      "learning_rate": 3.9053654088875667e-05,
      "loss": 77.2479,
      "step": 47120
    },
    {
      "epoch": 0.19041116367764638,
      "grad_norm": 946.6858520507812,
      "learning_rate": 3.905280506840466e-05,
      "loss": 137.4436,
      "step": 47130
    },
    {
      "epoch": 0.19045156494301402,
      "grad_norm": 927.581787109375,
      "learning_rate": 3.905195567648942e-05,
      "loss": 71.1932,
      "step": 47140
    },
    {
      "epoch": 0.19049196620838166,
      "grad_norm": 871.3401489257812,
      "learning_rate": 3.905110591314649e-05,
      "loss": 89.0064,
      "step": 47150
    },
    {
      "epoch": 0.19053236747374927,
      "grad_norm": 585.0065307617188,
      "learning_rate": 3.905025577839244e-05,
      "loss": 101.181,
      "step": 47160
    },
    {
      "epoch": 0.1905727687391169,
      "grad_norm": 587.296142578125,
      "learning_rate": 3.904940527224385e-05,
      "loss": 53.9529,
      "step": 47170
    },
    {
      "epoch": 0.19061317000448455,
      "grad_norm": 580.9244995117188,
      "learning_rate": 3.90485543947173e-05,
      "loss": 92.6486,
      "step": 47180
    },
    {
      "epoch": 0.19065357126985216,
      "grad_norm": 904.3273315429688,
      "learning_rate": 3.904770314582937e-05,
      "loss": 78.1508,
      "step": 47190
    },
    {
      "epoch": 0.1906939725352198,
      "grad_norm": 950.0264892578125,
      "learning_rate": 3.9046851525596656e-05,
      "loss": 64.4661,
      "step": 47200
    },
    {
      "epoch": 0.19073437380058744,
      "grad_norm": 1081.5543212890625,
      "learning_rate": 3.904599953403577e-05,
      "loss": 120.9151,
      "step": 47210
    },
    {
      "epoch": 0.19077477506595505,
      "grad_norm": 700.6988525390625,
      "learning_rate": 3.904514717116332e-05,
      "loss": 85.0513,
      "step": 47220
    },
    {
      "epoch": 0.1908151763313227,
      "grad_norm": 793.6326293945312,
      "learning_rate": 3.9044294436995924e-05,
      "loss": 103.1733,
      "step": 47230
    },
    {
      "epoch": 0.19085557759669033,
      "grad_norm": 952.3103637695312,
      "learning_rate": 3.9043441331550195e-05,
      "loss": 106.587,
      "step": 47240
    },
    {
      "epoch": 0.19089597886205795,
      "grad_norm": 1237.8389892578125,
      "learning_rate": 3.9042587854842776e-05,
      "loss": 71.8429,
      "step": 47250
    },
    {
      "epoch": 0.1909363801274256,
      "grad_norm": 876.2874755859375,
      "learning_rate": 3.9041734006890304e-05,
      "loss": 52.6092,
      "step": 47260
    },
    {
      "epoch": 0.19097678139279323,
      "grad_norm": 847.1505126953125,
      "learning_rate": 3.9040879787709426e-05,
      "loss": 106.2469,
      "step": 47270
    },
    {
      "epoch": 0.19101718265816087,
      "grad_norm": 514.6716918945312,
      "learning_rate": 3.904002519731679e-05,
      "loss": 71.6665,
      "step": 47280
    },
    {
      "epoch": 0.19105758392352848,
      "grad_norm": 1077.2454833984375,
      "learning_rate": 3.903917023572907e-05,
      "loss": 92.8363,
      "step": 47290
    },
    {
      "epoch": 0.19109798518889612,
      "grad_norm": 825.1358032226562,
      "learning_rate": 3.9038314902962906e-05,
      "loss": 69.5302,
      "step": 47300
    },
    {
      "epoch": 0.19113838645426376,
      "grad_norm": 415.63623046875,
      "learning_rate": 3.9037459199035004e-05,
      "loss": 85.427,
      "step": 47310
    },
    {
      "epoch": 0.19117878771963137,
      "grad_norm": 665.2693481445312,
      "learning_rate": 3.903660312396203e-05,
      "loss": 71.0723,
      "step": 47320
    },
    {
      "epoch": 0.191219188984999,
      "grad_norm": 913.37548828125,
      "learning_rate": 3.903574667776068e-05,
      "loss": 60.3247,
      "step": 47330
    },
    {
      "epoch": 0.19125959025036665,
      "grad_norm": 963.267822265625,
      "learning_rate": 3.903488986044765e-05,
      "loss": 81.5899,
      "step": 47340
    },
    {
      "epoch": 0.19129999151573426,
      "grad_norm": 1512.7403564453125,
      "learning_rate": 3.9034032672039634e-05,
      "loss": 85.3047,
      "step": 47350
    },
    {
      "epoch": 0.1913403927811019,
      "grad_norm": 689.649658203125,
      "learning_rate": 3.9033175112553354e-05,
      "loss": 90.7539,
      "step": 47360
    },
    {
      "epoch": 0.19138079404646954,
      "grad_norm": 303.6385192871094,
      "learning_rate": 3.9032317182005536e-05,
      "loss": 73.4,
      "step": 47370
    },
    {
      "epoch": 0.19142119531183716,
      "grad_norm": 780.4526977539062,
      "learning_rate": 3.903145888041289e-05,
      "loss": 80.5795,
      "step": 47380
    },
    {
      "epoch": 0.1914615965772048,
      "grad_norm": 737.2525634765625,
      "learning_rate": 3.903060020779215e-05,
      "loss": 55.1407,
      "step": 47390
    },
    {
      "epoch": 0.19150199784257244,
      "grad_norm": 616.6204833984375,
      "learning_rate": 3.902974116416006e-05,
      "loss": 71.6661,
      "step": 47400
    },
    {
      "epoch": 0.19154239910794005,
      "grad_norm": 385.3418884277344,
      "learning_rate": 3.902888174953338e-05,
      "loss": 66.3275,
      "step": 47410
    },
    {
      "epoch": 0.1915828003733077,
      "grad_norm": 641.3906860351562,
      "learning_rate": 3.902802196392885e-05,
      "loss": 51.0823,
      "step": 47420
    },
    {
      "epoch": 0.19162320163867533,
      "grad_norm": 1668.456787109375,
      "learning_rate": 3.902716180736323e-05,
      "loss": 74.4394,
      "step": 47430
    },
    {
      "epoch": 0.19166360290404297,
      "grad_norm": 634.2432250976562,
      "learning_rate": 3.90263012798533e-05,
      "loss": 73.6734,
      "step": 47440
    },
    {
      "epoch": 0.19170400416941058,
      "grad_norm": 1996.33056640625,
      "learning_rate": 3.902544038141583e-05,
      "loss": 99.414,
      "step": 47450
    },
    {
      "epoch": 0.19174440543477822,
      "grad_norm": 622.3128662109375,
      "learning_rate": 3.9024579112067604e-05,
      "loss": 52.0503,
      "step": 47460
    },
    {
      "epoch": 0.19178480670014586,
      "grad_norm": 1247.3145751953125,
      "learning_rate": 3.9023717471825416e-05,
      "loss": 100.0393,
      "step": 47470
    },
    {
      "epoch": 0.19182520796551347,
      "grad_norm": 629.420166015625,
      "learning_rate": 3.902285546070606e-05,
      "loss": 111.175,
      "step": 47480
    },
    {
      "epoch": 0.1918656092308811,
      "grad_norm": 1420.0347900390625,
      "learning_rate": 3.902199307872634e-05,
      "loss": 104.5429,
      "step": 47490
    },
    {
      "epoch": 0.19190601049624875,
      "grad_norm": 521.70703125,
      "learning_rate": 3.9021130325903076e-05,
      "loss": 61.8919,
      "step": 47500
    },
    {
      "epoch": 0.19194641176161636,
      "grad_norm": 991.524169921875,
      "learning_rate": 3.902026720225308e-05,
      "loss": 117.9224,
      "step": 47510
    },
    {
      "epoch": 0.191986813026984,
      "grad_norm": 791.923583984375,
      "learning_rate": 3.901940370779318e-05,
      "loss": 77.5792,
      "step": 47520
    },
    {
      "epoch": 0.19202721429235164,
      "grad_norm": 540.6439819335938,
      "learning_rate": 3.901853984254023e-05,
      "loss": 92.2672,
      "step": 47530
    },
    {
      "epoch": 0.19206761555771926,
      "grad_norm": 461.5845947265625,
      "learning_rate": 3.901767560651104e-05,
      "loss": 80.2478,
      "step": 47540
    },
    {
      "epoch": 0.1921080168230869,
      "grad_norm": 662.9263916015625,
      "learning_rate": 3.901681099972247e-05,
      "loss": 63.0723,
      "step": 47550
    },
    {
      "epoch": 0.19214841808845454,
      "grad_norm": 779.2219848632812,
      "learning_rate": 3.901594602219138e-05,
      "loss": 82.2809,
      "step": 47560
    },
    {
      "epoch": 0.19218881935382215,
      "grad_norm": 1538.7470703125,
      "learning_rate": 3.901508067393464e-05,
      "loss": 75.5992,
      "step": 47570
    },
    {
      "epoch": 0.1922292206191898,
      "grad_norm": 1130.8165283203125,
      "learning_rate": 3.901421495496911e-05,
      "loss": 84.9875,
      "step": 47580
    },
    {
      "epoch": 0.19226962188455743,
      "grad_norm": 2149.62744140625,
      "learning_rate": 3.901334886531167e-05,
      "loss": 93.2036,
      "step": 47590
    },
    {
      "epoch": 0.19231002314992507,
      "grad_norm": 395.6761169433594,
      "learning_rate": 3.90124824049792e-05,
      "loss": 97.5525,
      "step": 47600
    },
    {
      "epoch": 0.19235042441529268,
      "grad_norm": 1826.5126953125,
      "learning_rate": 3.901161557398861e-05,
      "loss": 118.2228,
      "step": 47610
    },
    {
      "epoch": 0.19239082568066032,
      "grad_norm": 1124.4610595703125,
      "learning_rate": 3.901074837235677e-05,
      "loss": 100.9521,
      "step": 47620
    },
    {
      "epoch": 0.19243122694602796,
      "grad_norm": 840.7826538085938,
      "learning_rate": 3.900988080010061e-05,
      "loss": 86.1119,
      "step": 47630
    },
    {
      "epoch": 0.19247162821139557,
      "grad_norm": 1439.2591552734375,
      "learning_rate": 3.900901285723704e-05,
      "loss": 46.7816,
      "step": 47640
    },
    {
      "epoch": 0.1925120294767632,
      "grad_norm": 1832.960205078125,
      "learning_rate": 3.900814454378297e-05,
      "loss": 96.7867,
      "step": 47650
    },
    {
      "epoch": 0.19255243074213085,
      "grad_norm": 1689.2100830078125,
      "learning_rate": 3.900727585975535e-05,
      "loss": 75.5526,
      "step": 47660
    },
    {
      "epoch": 0.19259283200749847,
      "grad_norm": 1058.6866455078125,
      "learning_rate": 3.900640680517109e-05,
      "loss": 65.278,
      "step": 47670
    },
    {
      "epoch": 0.1926332332728661,
      "grad_norm": 466.8570556640625,
      "learning_rate": 3.900553738004715e-05,
      "loss": 78.7198,
      "step": 47680
    },
    {
      "epoch": 0.19267363453823375,
      "grad_norm": 451.8070373535156,
      "learning_rate": 3.9004667584400474e-05,
      "loss": 76.8182,
      "step": 47690
    },
    {
      "epoch": 0.19271403580360136,
      "grad_norm": 1532.7535400390625,
      "learning_rate": 3.900379741824802e-05,
      "loss": 81.1931,
      "step": 47700
    },
    {
      "epoch": 0.192754437068969,
      "grad_norm": 942.5155639648438,
      "learning_rate": 3.900292688160674e-05,
      "loss": 66.7955,
      "step": 47710
    },
    {
      "epoch": 0.19279483833433664,
      "grad_norm": 1023.271484375,
      "learning_rate": 3.9002055974493626e-05,
      "loss": 64.7008,
      "step": 47720
    },
    {
      "epoch": 0.19283523959970425,
      "grad_norm": 926.4970092773438,
      "learning_rate": 3.900118469692565e-05,
      "loss": 132.3741,
      "step": 47730
    },
    {
      "epoch": 0.1928756408650719,
      "grad_norm": 573.7406616210938,
      "learning_rate": 3.90003130489198e-05,
      "loss": 81.473,
      "step": 47740
    },
    {
      "epoch": 0.19291604213043953,
      "grad_norm": 1699.2728271484375,
      "learning_rate": 3.899944103049305e-05,
      "loss": 84.5563,
      "step": 47750
    },
    {
      "epoch": 0.19295644339580717,
      "grad_norm": 796.8876342773438,
      "learning_rate": 3.8998568641662425e-05,
      "loss": 70.3057,
      "step": 47760
    },
    {
      "epoch": 0.19299684466117478,
      "grad_norm": 348.61224365234375,
      "learning_rate": 3.899769588244493e-05,
      "loss": 89.7806,
      "step": 47770
    },
    {
      "epoch": 0.19303724592654242,
      "grad_norm": 1488.8883056640625,
      "learning_rate": 3.899682275285756e-05,
      "loss": 60.286,
      "step": 47780
    },
    {
      "epoch": 0.19307764719191006,
      "grad_norm": 616.420654296875,
      "learning_rate": 3.8995949252917354e-05,
      "loss": 70.8484,
      "step": 47790
    },
    {
      "epoch": 0.19311804845727767,
      "grad_norm": 524.6080932617188,
      "learning_rate": 3.899507538264134e-05,
      "loss": 59.4811,
      "step": 47800
    },
    {
      "epoch": 0.19315844972264531,
      "grad_norm": 822.7127075195312,
      "learning_rate": 3.899420114204655e-05,
      "loss": 75.0543,
      "step": 47810
    },
    {
      "epoch": 0.19319885098801295,
      "grad_norm": 2325.68408203125,
      "learning_rate": 3.899332653115004e-05,
      "loss": 46.1322,
      "step": 47820
    },
    {
      "epoch": 0.19323925225338057,
      "grad_norm": 577.6055908203125,
      "learning_rate": 3.899245154996884e-05,
      "loss": 53.3111,
      "step": 47830
    },
    {
      "epoch": 0.1932796535187482,
      "grad_norm": 592.7323608398438,
      "learning_rate": 3.8991576198520015e-05,
      "loss": 141.9589,
      "step": 47840
    },
    {
      "epoch": 0.19332005478411585,
      "grad_norm": 1331.01220703125,
      "learning_rate": 3.899070047682064e-05,
      "loss": 91.7902,
      "step": 47850
    },
    {
      "epoch": 0.19336045604948346,
      "grad_norm": 2341.49755859375,
      "learning_rate": 3.898982438488778e-05,
      "loss": 98.7416,
      "step": 47860
    },
    {
      "epoch": 0.1934008573148511,
      "grad_norm": 689.9047241210938,
      "learning_rate": 3.898894792273853e-05,
      "loss": 83.9316,
      "step": 47870
    },
    {
      "epoch": 0.19344125858021874,
      "grad_norm": 684.3204956054688,
      "learning_rate": 3.898807109038995e-05,
      "loss": 67.1209,
      "step": 47880
    },
    {
      "epoch": 0.19348165984558635,
      "grad_norm": 866.631103515625,
      "learning_rate": 3.898719388785915e-05,
      "loss": 76.8427,
      "step": 47890
    },
    {
      "epoch": 0.193522061110954,
      "grad_norm": 2371.7421875,
      "learning_rate": 3.898631631516323e-05,
      "loss": 84.4593,
      "step": 47900
    },
    {
      "epoch": 0.19356246237632163,
      "grad_norm": 1004.8216552734375,
      "learning_rate": 3.89854383723193e-05,
      "loss": 113.4156,
      "step": 47910
    },
    {
      "epoch": 0.19360286364168927,
      "grad_norm": 895.5128784179688,
      "learning_rate": 3.898456005934447e-05,
      "loss": 72.0463,
      "step": 47920
    },
    {
      "epoch": 0.19364326490705688,
      "grad_norm": 2553.209228515625,
      "learning_rate": 3.8983681376255876e-05,
      "loss": 116.5803,
      "step": 47930
    },
    {
      "epoch": 0.19368366617242452,
      "grad_norm": 514.206298828125,
      "learning_rate": 3.898280232307064e-05,
      "loss": 62.542,
      "step": 47940
    },
    {
      "epoch": 0.19372406743779216,
      "grad_norm": 3847.76806640625,
      "learning_rate": 3.898192289980589e-05,
      "loss": 100.5998,
      "step": 47950
    },
    {
      "epoch": 0.19376446870315978,
      "grad_norm": 710.794677734375,
      "learning_rate": 3.898104310647879e-05,
      "loss": 84.6892,
      "step": 47960
    },
    {
      "epoch": 0.19380486996852742,
      "grad_norm": 244.08517456054688,
      "learning_rate": 3.8980162943106486e-05,
      "loss": 66.8276,
      "step": 47970
    },
    {
      "epoch": 0.19384527123389506,
      "grad_norm": 1008.19677734375,
      "learning_rate": 3.897928240970612e-05,
      "loss": 83.5675,
      "step": 47980
    },
    {
      "epoch": 0.19388567249926267,
      "grad_norm": 1166.093505859375,
      "learning_rate": 3.8978401506294885e-05,
      "loss": 105.6205,
      "step": 47990
    },
    {
      "epoch": 0.1939260737646303,
      "grad_norm": 645.5625,
      "learning_rate": 3.897752023288993e-05,
      "loss": 105.2789,
      "step": 48000
    },
    {
      "epoch": 0.19396647502999795,
      "grad_norm": 679.8927612304688,
      "learning_rate": 3.8976638589508456e-05,
      "loss": 51.4738,
      "step": 48010
    },
    {
      "epoch": 0.19400687629536556,
      "grad_norm": 599.9944458007812,
      "learning_rate": 3.897575657616764e-05,
      "loss": 76.4371,
      "step": 48020
    },
    {
      "epoch": 0.1940472775607332,
      "grad_norm": 4461.6572265625,
      "learning_rate": 3.897487419288468e-05,
      "loss": 124.2049,
      "step": 48030
    },
    {
      "epoch": 0.19408767882610084,
      "grad_norm": 1515.529296875,
      "learning_rate": 3.8973991439676783e-05,
      "loss": 87.721,
      "step": 48040
    },
    {
      "epoch": 0.19412808009146845,
      "grad_norm": 669.8505249023438,
      "learning_rate": 3.897310831656115e-05,
      "loss": 60.2474,
      "step": 48050
    },
    {
      "epoch": 0.1941684813568361,
      "grad_norm": 697.6897583007812,
      "learning_rate": 3.897222482355501e-05,
      "loss": 96.5461,
      "step": 48060
    },
    {
      "epoch": 0.19420888262220373,
      "grad_norm": 517.4088134765625,
      "learning_rate": 3.897134096067557e-05,
      "loss": 70.281,
      "step": 48070
    },
    {
      "epoch": 0.19424928388757137,
      "grad_norm": 645.1104736328125,
      "learning_rate": 3.8970456727940076e-05,
      "loss": 89.9621,
      "step": 48080
    },
    {
      "epoch": 0.19428968515293898,
      "grad_norm": 534.7449340820312,
      "learning_rate": 3.896957212536576e-05,
      "loss": 81.962,
      "step": 48090
    },
    {
      "epoch": 0.19433008641830662,
      "grad_norm": 1291.22119140625,
      "learning_rate": 3.896868715296987e-05,
      "loss": 108.5878,
      "step": 48100
    },
    {
      "epoch": 0.19437048768367426,
      "grad_norm": 1313.7838134765625,
      "learning_rate": 3.896780181076966e-05,
      "loss": 57.1638,
      "step": 48110
    },
    {
      "epoch": 0.19441088894904188,
      "grad_norm": 1689.31640625,
      "learning_rate": 3.8966916098782377e-05,
      "loss": 71.2753,
      "step": 48120
    },
    {
      "epoch": 0.19445129021440952,
      "grad_norm": 628.2205810546875,
      "learning_rate": 3.896603001702531e-05,
      "loss": 64.059,
      "step": 48130
    },
    {
      "epoch": 0.19449169147977716,
      "grad_norm": 1219.47509765625,
      "learning_rate": 3.896514356551572e-05,
      "loss": 95.7702,
      "step": 48140
    },
    {
      "epoch": 0.19453209274514477,
      "grad_norm": 1684.8138427734375,
      "learning_rate": 3.896425674427089e-05,
      "loss": 63.9552,
      "step": 48150
    },
    {
      "epoch": 0.1945724940105124,
      "grad_norm": 926.5546264648438,
      "learning_rate": 3.896336955330812e-05,
      "loss": 115.7368,
      "step": 48160
    },
    {
      "epoch": 0.19461289527588005,
      "grad_norm": 2358.445556640625,
      "learning_rate": 3.8962481992644694e-05,
      "loss": 122.3302,
      "step": 48170
    },
    {
      "epoch": 0.19465329654124766,
      "grad_norm": 1648.923583984375,
      "learning_rate": 3.896159406229792e-05,
      "loss": 62.5087,
      "step": 48180
    },
    {
      "epoch": 0.1946936978066153,
      "grad_norm": 637.2569580078125,
      "learning_rate": 3.896070576228511e-05,
      "loss": 77.8548,
      "step": 48190
    },
    {
      "epoch": 0.19473409907198294,
      "grad_norm": 1143.2896728515625,
      "learning_rate": 3.895981709262357e-05,
      "loss": 60.9295,
      "step": 48200
    },
    {
      "epoch": 0.19477450033735055,
      "grad_norm": 2155.3466796875,
      "learning_rate": 3.895892805333064e-05,
      "loss": 89.5178,
      "step": 48210
    },
    {
      "epoch": 0.1948149016027182,
      "grad_norm": 1476.8765869140625,
      "learning_rate": 3.895803864442365e-05,
      "loss": 95.1555,
      "step": 48220
    },
    {
      "epoch": 0.19485530286808583,
      "grad_norm": 601.9876708984375,
      "learning_rate": 3.895714886591993e-05,
      "loss": 86.8787,
      "step": 48230
    },
    {
      "epoch": 0.19489570413345347,
      "grad_norm": 701.3627319335938,
      "learning_rate": 3.895625871783684e-05,
      "loss": 68.1065,
      "step": 48240
    },
    {
      "epoch": 0.19493610539882109,
      "grad_norm": 1250.448486328125,
      "learning_rate": 3.895536820019172e-05,
      "loss": 91.6428,
      "step": 48250
    },
    {
      "epoch": 0.19497650666418873,
      "grad_norm": 993.2598876953125,
      "learning_rate": 3.895447731300194e-05,
      "loss": 61.6161,
      "step": 48260
    },
    {
      "epoch": 0.19501690792955637,
      "grad_norm": 542.6673583984375,
      "learning_rate": 3.8953586056284866e-05,
      "loss": 49.7327,
      "step": 48270
    },
    {
      "epoch": 0.19505730919492398,
      "grad_norm": 1102.58642578125,
      "learning_rate": 3.895269443005788e-05,
      "loss": 70.2154,
      "step": 48280
    },
    {
      "epoch": 0.19509771046029162,
      "grad_norm": 1155.5128173828125,
      "learning_rate": 3.895180243433835e-05,
      "loss": 77.5229,
      "step": 48290
    },
    {
      "epoch": 0.19513811172565926,
      "grad_norm": 856.4344482421875,
      "learning_rate": 3.895091006914368e-05,
      "loss": 79.453,
      "step": 48300
    },
    {
      "epoch": 0.19517851299102687,
      "grad_norm": 1338.512451171875,
      "learning_rate": 3.8950017334491256e-05,
      "loss": 102.0866,
      "step": 48310
    },
    {
      "epoch": 0.1952189142563945,
      "grad_norm": 943.7992553710938,
      "learning_rate": 3.894912423039849e-05,
      "loss": 106.8494,
      "step": 48320
    },
    {
      "epoch": 0.19525931552176215,
      "grad_norm": 477.94769287109375,
      "learning_rate": 3.89482307568828e-05,
      "loss": 74.899,
      "step": 48330
    },
    {
      "epoch": 0.19529971678712976,
      "grad_norm": 2688.3466796875,
      "learning_rate": 3.894733691396159e-05,
      "loss": 68.6513,
      "step": 48340
    },
    {
      "epoch": 0.1953401180524974,
      "grad_norm": 783.2159423828125,
      "learning_rate": 3.894644270165229e-05,
      "loss": 111.6442,
      "step": 48350
    },
    {
      "epoch": 0.19538051931786504,
      "grad_norm": 667.3715209960938,
      "learning_rate": 3.894554811997234e-05,
      "loss": 70.8076,
      "step": 48360
    },
    {
      "epoch": 0.19542092058323265,
      "grad_norm": 912.3576049804688,
      "learning_rate": 3.894465316893917e-05,
      "loss": 83.1459,
      "step": 48370
    },
    {
      "epoch": 0.1954613218486003,
      "grad_norm": 830.3079833984375,
      "learning_rate": 3.894375784857023e-05,
      "loss": 106.7063,
      "step": 48380
    },
    {
      "epoch": 0.19550172311396793,
      "grad_norm": 1187.99560546875,
      "learning_rate": 3.894286215888299e-05,
      "loss": 92.7125,
      "step": 48390
    },
    {
      "epoch": 0.19554212437933557,
      "grad_norm": 2235.258056640625,
      "learning_rate": 3.894196609989489e-05,
      "loss": 77.8964,
      "step": 48400
    },
    {
      "epoch": 0.1955825256447032,
      "grad_norm": 1120.1141357421875,
      "learning_rate": 3.894106967162341e-05,
      "loss": 75.0723,
      "step": 48410
    },
    {
      "epoch": 0.19562292691007083,
      "grad_norm": 1820.933837890625,
      "learning_rate": 3.894017287408603e-05,
      "loss": 74.4507,
      "step": 48420
    },
    {
      "epoch": 0.19566332817543847,
      "grad_norm": 911.1453247070312,
      "learning_rate": 3.8939275707300226e-05,
      "loss": 59.5158,
      "step": 48430
    },
    {
      "epoch": 0.19570372944080608,
      "grad_norm": 1145.5992431640625,
      "learning_rate": 3.8938378171283485e-05,
      "loss": 66.2683,
      "step": 48440
    },
    {
      "epoch": 0.19574413070617372,
      "grad_norm": 1104.505859375,
      "learning_rate": 3.893748026605332e-05,
      "loss": 64.669,
      "step": 48450
    },
    {
      "epoch": 0.19578453197154136,
      "grad_norm": 1642.31396484375,
      "learning_rate": 3.893658199162723e-05,
      "loss": 86.1956,
      "step": 48460
    },
    {
      "epoch": 0.19582493323690897,
      "grad_norm": 2475.31787109375,
      "learning_rate": 3.893568334802272e-05,
      "loss": 104.0921,
      "step": 48470
    },
    {
      "epoch": 0.1958653345022766,
      "grad_norm": 952.88134765625,
      "learning_rate": 3.893478433525731e-05,
      "loss": 87.7983,
      "step": 48480
    },
    {
      "epoch": 0.19590573576764425,
      "grad_norm": 1892.2713623046875,
      "learning_rate": 3.8933884953348545e-05,
      "loss": 88.2686,
      "step": 48490
    },
    {
      "epoch": 0.19594613703301186,
      "grad_norm": 908.8019409179688,
      "learning_rate": 3.893298520231393e-05,
      "loss": 101.3182,
      "step": 48500
    },
    {
      "epoch": 0.1959865382983795,
      "grad_norm": 1779.0821533203125,
      "learning_rate": 3.893208508217103e-05,
      "loss": 91.2895,
      "step": 48510
    },
    {
      "epoch": 0.19602693956374714,
      "grad_norm": 823.1393432617188,
      "learning_rate": 3.893118459293738e-05,
      "loss": 79.2318,
      "step": 48520
    },
    {
      "epoch": 0.19606734082911476,
      "grad_norm": 1146.0291748046875,
      "learning_rate": 3.893028373463054e-05,
      "loss": 60.2765,
      "step": 48530
    },
    {
      "epoch": 0.1961077420944824,
      "grad_norm": 907.8261108398438,
      "learning_rate": 3.8929382507268074e-05,
      "loss": 81.0746,
      "step": 48540
    },
    {
      "epoch": 0.19614814335985004,
      "grad_norm": 542.8344116210938,
      "learning_rate": 3.892848091086755e-05,
      "loss": 71.9454,
      "step": 48550
    },
    {
      "epoch": 0.19618854462521768,
      "grad_norm": 735.962890625,
      "learning_rate": 3.8927578945446544e-05,
      "loss": 81.1646,
      "step": 48560
    },
    {
      "epoch": 0.1962289458905853,
      "grad_norm": 608.4637451171875,
      "learning_rate": 3.892667661102265e-05,
      "loss": 43.2689,
      "step": 48570
    },
    {
      "epoch": 0.19626934715595293,
      "grad_norm": 494.1917419433594,
      "learning_rate": 3.892577390761344e-05,
      "loss": 119.9042,
      "step": 48580
    },
    {
      "epoch": 0.19630974842132057,
      "grad_norm": 1613.7349853515625,
      "learning_rate": 3.892487083523653e-05,
      "loss": 134.0927,
      "step": 48590
    },
    {
      "epoch": 0.19635014968668818,
      "grad_norm": 570.557373046875,
      "learning_rate": 3.892396739390952e-05,
      "loss": 70.8616,
      "step": 48600
    },
    {
      "epoch": 0.19639055095205582,
      "grad_norm": 1168.809326171875,
      "learning_rate": 3.892306358365002e-05,
      "loss": 106.0381,
      "step": 48610
    },
    {
      "epoch": 0.19643095221742346,
      "grad_norm": 1149.137451171875,
      "learning_rate": 3.8922159404475653e-05,
      "loss": 74.9864,
      "step": 48620
    },
    {
      "epoch": 0.19647135348279107,
      "grad_norm": 830.5051879882812,
      "learning_rate": 3.8921254856404056e-05,
      "loss": 91.8877,
      "step": 48630
    },
    {
      "epoch": 0.1965117547481587,
      "grad_norm": 1049.21142578125,
      "learning_rate": 3.892034993945284e-05,
      "loss": 65.3357,
      "step": 48640
    },
    {
      "epoch": 0.19655215601352635,
      "grad_norm": 885.1357421875,
      "learning_rate": 3.891944465363968e-05,
      "loss": 74.8494,
      "step": 48650
    },
    {
      "epoch": 0.19659255727889396,
      "grad_norm": 1449.2069091796875,
      "learning_rate": 3.891853899898219e-05,
      "loss": 106.2978,
      "step": 48660
    },
    {
      "epoch": 0.1966329585442616,
      "grad_norm": 1177.130615234375,
      "learning_rate": 3.8917632975498046e-05,
      "loss": 59.5166,
      "step": 48670
    },
    {
      "epoch": 0.19667335980962924,
      "grad_norm": 903.5010986328125,
      "learning_rate": 3.891672658320491e-05,
      "loss": 47.0828,
      "step": 48680
    },
    {
      "epoch": 0.19671376107499686,
      "grad_norm": 527.488037109375,
      "learning_rate": 3.891581982212045e-05,
      "loss": 44.1152,
      "step": 48690
    },
    {
      "epoch": 0.1967541623403645,
      "grad_norm": 555.2778930664062,
      "learning_rate": 3.891491269226234e-05,
      "loss": 58.6444,
      "step": 48700
    },
    {
      "epoch": 0.19679456360573214,
      "grad_norm": 1518.626953125,
      "learning_rate": 3.891400519364827e-05,
      "loss": 98.2571,
      "step": 48710
    },
    {
      "epoch": 0.19683496487109975,
      "grad_norm": 1145.939697265625,
      "learning_rate": 3.891309732629593e-05,
      "loss": 83.2956,
      "step": 48720
    },
    {
      "epoch": 0.1968753661364674,
      "grad_norm": 509.7560729980469,
      "learning_rate": 3.891218909022302e-05,
      "loss": 57.1039,
      "step": 48730
    },
    {
      "epoch": 0.19691576740183503,
      "grad_norm": 1032.6080322265625,
      "learning_rate": 3.891128048544726e-05,
      "loss": 82.644,
      "step": 48740
    },
    {
      "epoch": 0.19695616866720267,
      "grad_norm": 1114.884033203125,
      "learning_rate": 3.891037151198634e-05,
      "loss": 130.455,
      "step": 48750
    },
    {
      "epoch": 0.19699656993257028,
      "grad_norm": 848.599365234375,
      "learning_rate": 3.890946216985799e-05,
      "loss": 107.0122,
      "step": 48760
    },
    {
      "epoch": 0.19703697119793792,
      "grad_norm": 635.4305419921875,
      "learning_rate": 3.890855245907994e-05,
      "loss": 95.4124,
      "step": 48770
    },
    {
      "epoch": 0.19707737246330556,
      "grad_norm": 744.4996948242188,
      "learning_rate": 3.890764237966993e-05,
      "loss": 63.5169,
      "step": 48780
    },
    {
      "epoch": 0.19711777372867317,
      "grad_norm": 1917.751708984375,
      "learning_rate": 3.8906731931645694e-05,
      "loss": 121.8446,
      "step": 48790
    },
    {
      "epoch": 0.1971581749940408,
      "grad_norm": 2296.08544921875,
      "learning_rate": 3.8905821115024994e-05,
      "loss": 91.3989,
      "step": 48800
    },
    {
      "epoch": 0.19719857625940845,
      "grad_norm": 571.256591796875,
      "learning_rate": 3.8904909929825565e-05,
      "loss": 54.8591,
      "step": 48810
    },
    {
      "epoch": 0.19723897752477607,
      "grad_norm": 582.6764526367188,
      "learning_rate": 3.890399837606519e-05,
      "loss": 88.5333,
      "step": 48820
    },
    {
      "epoch": 0.1972793787901437,
      "grad_norm": 1239.0958251953125,
      "learning_rate": 3.8903086453761634e-05,
      "loss": 96.0011,
      "step": 48830
    },
    {
      "epoch": 0.19731978005551135,
      "grad_norm": 1710.357421875,
      "learning_rate": 3.890217416293267e-05,
      "loss": 91.3379,
      "step": 48840
    },
    {
      "epoch": 0.19736018132087896,
      "grad_norm": 791.1911010742188,
      "learning_rate": 3.89012615035961e-05,
      "loss": 37.7967,
      "step": 48850
    },
    {
      "epoch": 0.1974005825862466,
      "grad_norm": 636.7476196289062,
      "learning_rate": 3.89003484757697e-05,
      "loss": 50.4115,
      "step": 48860
    },
    {
      "epoch": 0.19744098385161424,
      "grad_norm": 665.1348876953125,
      "learning_rate": 3.889943507947128e-05,
      "loss": 62.8492,
      "step": 48870
    },
    {
      "epoch": 0.19748138511698185,
      "grad_norm": 565.6461181640625,
      "learning_rate": 3.889852131471863e-05,
      "loss": 81.9738,
      "step": 48880
    },
    {
      "epoch": 0.1975217863823495,
      "grad_norm": 439.4663391113281,
      "learning_rate": 3.8897607181529594e-05,
      "loss": 56.1482,
      "step": 48890
    },
    {
      "epoch": 0.19756218764771713,
      "grad_norm": 701.4642333984375,
      "learning_rate": 3.889669267992197e-05,
      "loss": 74.9441,
      "step": 48900
    },
    {
      "epoch": 0.19760258891308477,
      "grad_norm": 519.2825927734375,
      "learning_rate": 3.889577780991359e-05,
      "loss": 64.2358,
      "step": 48910
    },
    {
      "epoch": 0.19764299017845238,
      "grad_norm": 998.2957153320312,
      "learning_rate": 3.8894862571522297e-05,
      "loss": 78.8985,
      "step": 48920
    },
    {
      "epoch": 0.19768339144382002,
      "grad_norm": 772.8052978515625,
      "learning_rate": 3.889394696476593e-05,
      "loss": 70.7561,
      "step": 48930
    },
    {
      "epoch": 0.19772379270918766,
      "grad_norm": 1149.341064453125,
      "learning_rate": 3.8893030989662336e-05,
      "loss": 83.4886,
      "step": 48940
    },
    {
      "epoch": 0.19776419397455527,
      "grad_norm": 473.2572021484375,
      "learning_rate": 3.889211464622938e-05,
      "loss": 82.0341,
      "step": 48950
    },
    {
      "epoch": 0.19780459523992291,
      "grad_norm": 1460.015869140625,
      "learning_rate": 3.889119793448493e-05,
      "loss": 88.2184,
      "step": 48960
    },
    {
      "epoch": 0.19784499650529055,
      "grad_norm": 542.736328125,
      "learning_rate": 3.889028085444684e-05,
      "loss": 59.861,
      "step": 48970
    },
    {
      "epoch": 0.19788539777065817,
      "grad_norm": 1630.9713134765625,
      "learning_rate": 3.8889363406133e-05,
      "loss": 92.3278,
      "step": 48980
    },
    {
      "epoch": 0.1979257990360258,
      "grad_norm": 891.9635620117188,
      "learning_rate": 3.88884455895613e-05,
      "loss": 101.8944,
      "step": 48990
    },
    {
      "epoch": 0.19796620030139345,
      "grad_norm": 1021.063720703125,
      "learning_rate": 3.888752740474962e-05,
      "loss": 103.379,
      "step": 49000
    },
    {
      "epoch": 0.19800660156676106,
      "grad_norm": 1191.407470703125,
      "learning_rate": 3.888660885171588e-05,
      "loss": 87.3679,
      "step": 49010
    },
    {
      "epoch": 0.1980470028321287,
      "grad_norm": 1310.1839599609375,
      "learning_rate": 3.8885689930477974e-05,
      "loss": 89.4196,
      "step": 49020
    },
    {
      "epoch": 0.19808740409749634,
      "grad_norm": 476.1018981933594,
      "learning_rate": 3.8884770641053815e-05,
      "loss": 80.5733,
      "step": 49030
    },
    {
      "epoch": 0.19812780536286395,
      "grad_norm": 905.9243774414062,
      "learning_rate": 3.888385098346134e-05,
      "loss": 70.9268,
      "step": 49040
    },
    {
      "epoch": 0.1981682066282316,
      "grad_norm": 411.78668212890625,
      "learning_rate": 3.888293095771846e-05,
      "loss": 87.6161,
      "step": 49050
    },
    {
      "epoch": 0.19820860789359923,
      "grad_norm": 1124.540771484375,
      "learning_rate": 3.888201056384312e-05,
      "loss": 106.0178,
      "step": 49060
    },
    {
      "epoch": 0.19824900915896687,
      "grad_norm": 481.7277526855469,
      "learning_rate": 3.888108980185326e-05,
      "loss": 99.5968,
      "step": 49070
    },
    {
      "epoch": 0.19828941042433448,
      "grad_norm": 1168.4268798828125,
      "learning_rate": 3.888016867176684e-05,
      "loss": 81.1295,
      "step": 49080
    },
    {
      "epoch": 0.19832981168970212,
      "grad_norm": 179.3350830078125,
      "learning_rate": 3.88792471736018e-05,
      "loss": 78.5757,
      "step": 49090
    },
    {
      "epoch": 0.19837021295506976,
      "grad_norm": 554.9774169921875,
      "learning_rate": 3.887832530737613e-05,
      "loss": 72.1782,
      "step": 49100
    },
    {
      "epoch": 0.19841061422043738,
      "grad_norm": 495.8739318847656,
      "learning_rate": 3.887740307310779e-05,
      "loss": 54.6817,
      "step": 49110
    },
    {
      "epoch": 0.19845101548580502,
      "grad_norm": 920.6334838867188,
      "learning_rate": 3.887648047081475e-05,
      "loss": 69.6979,
      "step": 49120
    },
    {
      "epoch": 0.19849141675117266,
      "grad_norm": 3227.0537109375,
      "learning_rate": 3.887555750051501e-05,
      "loss": 100.7972,
      "step": 49130
    },
    {
      "epoch": 0.19853181801654027,
      "grad_norm": 1286.3974609375,
      "learning_rate": 3.887463416222655e-05,
      "loss": 66.7063,
      "step": 49140
    },
    {
      "epoch": 0.1985722192819079,
      "grad_norm": 890.7499389648438,
      "learning_rate": 3.887371045596739e-05,
      "loss": 100.4346,
      "step": 49150
    },
    {
      "epoch": 0.19861262054727555,
      "grad_norm": 679.1954956054688,
      "learning_rate": 3.887278638175552e-05,
      "loss": 95.4674,
      "step": 49160
    },
    {
      "epoch": 0.19865302181264316,
      "grad_norm": 778.8389892578125,
      "learning_rate": 3.8871861939608976e-05,
      "loss": 93.5247,
      "step": 49170
    },
    {
      "epoch": 0.1986934230780108,
      "grad_norm": 3962.55322265625,
      "learning_rate": 3.8870937129545756e-05,
      "loss": 94.9488,
      "step": 49180
    },
    {
      "epoch": 0.19873382434337844,
      "grad_norm": 627.3102416992188,
      "learning_rate": 3.88700119515839e-05,
      "loss": 43.7665,
      "step": 49190
    },
    {
      "epoch": 0.19877422560874605,
      "grad_norm": 1436.988037109375,
      "learning_rate": 3.886908640574145e-05,
      "loss": 110.0446,
      "step": 49200
    },
    {
      "epoch": 0.1988146268741137,
      "grad_norm": 733.1195068359375,
      "learning_rate": 3.8868160492036444e-05,
      "loss": 66.3317,
      "step": 49210
    },
    {
      "epoch": 0.19885502813948133,
      "grad_norm": 710.966552734375,
      "learning_rate": 3.886723421048694e-05,
      "loss": 69.9744,
      "step": 49220
    },
    {
      "epoch": 0.19889542940484897,
      "grad_norm": 542.9668579101562,
      "learning_rate": 3.886630756111099e-05,
      "loss": 112.02,
      "step": 49230
    },
    {
      "epoch": 0.19893583067021658,
      "grad_norm": 813.4812622070312,
      "learning_rate": 3.886538054392665e-05,
      "loss": 149.1125,
      "step": 49240
    },
    {
      "epoch": 0.19897623193558422,
      "grad_norm": 286.7843933105469,
      "learning_rate": 3.886445315895202e-05,
      "loss": 70.4966,
      "step": 49250
    },
    {
      "epoch": 0.19901663320095186,
      "grad_norm": 605.0004272460938,
      "learning_rate": 3.886352540620516e-05,
      "loss": 62.9089,
      "step": 49260
    },
    {
      "epoch": 0.19905703446631948,
      "grad_norm": 495.3343200683594,
      "learning_rate": 3.886259728570416e-05,
      "loss": 68.2199,
      "step": 49270
    },
    {
      "epoch": 0.19909743573168712,
      "grad_norm": 1497.2340087890625,
      "learning_rate": 3.8861668797467114e-05,
      "loss": 87.3006,
      "step": 49280
    },
    {
      "epoch": 0.19913783699705476,
      "grad_norm": 599.7035522460938,
      "learning_rate": 3.886073994151212e-05,
      "loss": 54.236,
      "step": 49290
    },
    {
      "epoch": 0.19917823826242237,
      "grad_norm": 568.5770874023438,
      "learning_rate": 3.8859810717857296e-05,
      "loss": 56.1212,
      "step": 49300
    },
    {
      "epoch": 0.19921863952779,
      "grad_norm": 638.6181640625,
      "learning_rate": 3.8858881126520745e-05,
      "loss": 65.3909,
      "step": 49310
    },
    {
      "epoch": 0.19925904079315765,
      "grad_norm": 1189.181884765625,
      "learning_rate": 3.8857951167520605e-05,
      "loss": 69.2821,
      "step": 49320
    },
    {
      "epoch": 0.19929944205852526,
      "grad_norm": 1266.6331787109375,
      "learning_rate": 3.885702084087499e-05,
      "loss": 82.3138,
      "step": 49330
    },
    {
      "epoch": 0.1993398433238929,
      "grad_norm": 579.02685546875,
      "learning_rate": 3.885609014660205e-05,
      "loss": 99.8087,
      "step": 49340
    },
    {
      "epoch": 0.19938024458926054,
      "grad_norm": 665.1373901367188,
      "learning_rate": 3.885515908471993e-05,
      "loss": 97.9025,
      "step": 49350
    },
    {
      "epoch": 0.19942064585462815,
      "grad_norm": 461.8253173828125,
      "learning_rate": 3.885422765524677e-05,
      "loss": 59.6303,
      "step": 49360
    },
    {
      "epoch": 0.1994610471199958,
      "grad_norm": 512.667236328125,
      "learning_rate": 3.885329585820074e-05,
      "loss": 52.7883,
      "step": 49370
    },
    {
      "epoch": 0.19950144838536343,
      "grad_norm": 549.1116943359375,
      "learning_rate": 3.8852363693599996e-05,
      "loss": 71.6235,
      "step": 49380
    },
    {
      "epoch": 0.19954184965073107,
      "grad_norm": 2068.1806640625,
      "learning_rate": 3.885143116146272e-05,
      "loss": 162.5312,
      "step": 49390
    },
    {
      "epoch": 0.19958225091609869,
      "grad_norm": 1925.5687255859375,
      "learning_rate": 3.885049826180709e-05,
      "loss": 86.9825,
      "step": 49400
    },
    {
      "epoch": 0.19962265218146633,
      "grad_norm": 804.568359375,
      "learning_rate": 3.8849564994651284e-05,
      "loss": 51.5701,
      "step": 49410
    },
    {
      "epoch": 0.19966305344683397,
      "grad_norm": 709.6029052734375,
      "learning_rate": 3.884863136001351e-05,
      "loss": 57.2898,
      "step": 49420
    },
    {
      "epoch": 0.19970345471220158,
      "grad_norm": 1091.1815185546875,
      "learning_rate": 3.884769735791196e-05,
      "loss": 138.4531,
      "step": 49430
    },
    {
      "epoch": 0.19974385597756922,
      "grad_norm": 366.7632141113281,
      "learning_rate": 3.8846762988364845e-05,
      "loss": 71.2259,
      "step": 49440
    },
    {
      "epoch": 0.19978425724293686,
      "grad_norm": 247.80860900878906,
      "learning_rate": 3.884582825139038e-05,
      "loss": 67.1933,
      "step": 49450
    },
    {
      "epoch": 0.19982465850830447,
      "grad_norm": 1035.06103515625,
      "learning_rate": 3.88448931470068e-05,
      "loss": 74.0027,
      "step": 49460
    },
    {
      "epoch": 0.1998650597736721,
      "grad_norm": 1263.3199462890625,
      "learning_rate": 3.884395767523232e-05,
      "loss": 49.4541,
      "step": 49470
    },
    {
      "epoch": 0.19990546103903975,
      "grad_norm": 681.6302490234375,
      "learning_rate": 3.884302183608519e-05,
      "loss": 63.8364,
      "step": 49480
    },
    {
      "epoch": 0.19994586230440736,
      "grad_norm": 604.8681640625,
      "learning_rate": 3.884208562958364e-05,
      "loss": 61.3758,
      "step": 49490
    },
    {
      "epoch": 0.199986263569775,
      "grad_norm": 838.427490234375,
      "learning_rate": 3.884114905574594e-05,
      "loss": 63.9567,
      "step": 49500
    },
    {
      "epoch": 0.20002666483514264,
      "grad_norm": 2336.60693359375,
      "learning_rate": 3.884021211459033e-05,
      "loss": 102.7997,
      "step": 49510
    },
    {
      "epoch": 0.20006706610051025,
      "grad_norm": 590.4444580078125,
      "learning_rate": 3.8839274806135095e-05,
      "loss": 90.1219,
      "step": 49520
    },
    {
      "epoch": 0.2001074673658779,
      "grad_norm": 924.6959228515625,
      "learning_rate": 3.883833713039849e-05,
      "loss": 50.8415,
      "step": 49530
    },
    {
      "epoch": 0.20014786863124553,
      "grad_norm": 677.3268432617188,
      "learning_rate": 3.8837399087398803e-05,
      "loss": 75.5059,
      "step": 49540
    },
    {
      "epoch": 0.20018826989661317,
      "grad_norm": 402.6632385253906,
      "learning_rate": 3.883646067715433e-05,
      "loss": 76.6058,
      "step": 49550
    },
    {
      "epoch": 0.2002286711619808,
      "grad_norm": 1494.4140625,
      "learning_rate": 3.8835521899683345e-05,
      "loss": 115.2886,
      "step": 49560
    },
    {
      "epoch": 0.20026907242734843,
      "grad_norm": 1487.2633056640625,
      "learning_rate": 3.883458275500417e-05,
      "loss": 89.6929,
      "step": 49570
    },
    {
      "epoch": 0.20030947369271607,
      "grad_norm": 500.047119140625,
      "learning_rate": 3.883364324313512e-05,
      "loss": 72.081,
      "step": 49580
    },
    {
      "epoch": 0.20034987495808368,
      "grad_norm": 827.5897827148438,
      "learning_rate": 3.883270336409448e-05,
      "loss": 64.2101,
      "step": 49590
    },
    {
      "epoch": 0.20039027622345132,
      "grad_norm": 878.098388671875,
      "learning_rate": 3.8831763117900605e-05,
      "loss": 90.874,
      "step": 49600
    },
    {
      "epoch": 0.20043067748881896,
      "grad_norm": 701.7119140625,
      "learning_rate": 3.883082250457181e-05,
      "loss": 85.7707,
      "step": 49610
    },
    {
      "epoch": 0.20047107875418657,
      "grad_norm": 750.2167358398438,
      "learning_rate": 3.882988152412643e-05,
      "loss": 105.379,
      "step": 49620
    },
    {
      "epoch": 0.2005114800195542,
      "grad_norm": 1229.0809326171875,
      "learning_rate": 3.882894017658283e-05,
      "loss": 82.8475,
      "step": 49630
    },
    {
      "epoch": 0.20055188128492185,
      "grad_norm": 745.6799926757812,
      "learning_rate": 3.882799846195933e-05,
      "loss": 66.1066,
      "step": 49640
    },
    {
      "epoch": 0.20059228255028946,
      "grad_norm": 866.1677856445312,
      "learning_rate": 3.8827056380274316e-05,
      "loss": 68.2047,
      "step": 49650
    },
    {
      "epoch": 0.2006326838156571,
      "grad_norm": 1317.9329833984375,
      "learning_rate": 3.882611393154614e-05,
      "loss": 89.9241,
      "step": 49660
    },
    {
      "epoch": 0.20067308508102474,
      "grad_norm": 607.7474365234375,
      "learning_rate": 3.882517111579319e-05,
      "loss": 81.1579,
      "step": 49670
    },
    {
      "epoch": 0.20071348634639236,
      "grad_norm": 1269.857666015625,
      "learning_rate": 3.882422793303382e-05,
      "loss": 70.4856,
      "step": 49680
    },
    {
      "epoch": 0.20075388761176,
      "grad_norm": 800.6294555664062,
      "learning_rate": 3.882328438328645e-05,
      "loss": 73.7126,
      "step": 49690
    },
    {
      "epoch": 0.20079428887712764,
      "grad_norm": 1065.0087890625,
      "learning_rate": 3.8822340466569456e-05,
      "loss": 84.4078,
      "step": 49700
    },
    {
      "epoch": 0.20083469014249528,
      "grad_norm": 393.25213623046875,
      "learning_rate": 3.882139618290125e-05,
      "loss": 82.6332,
      "step": 49710
    },
    {
      "epoch": 0.2008750914078629,
      "grad_norm": 671.2138671875,
      "learning_rate": 3.882045153230022e-05,
      "loss": 75.602,
      "step": 49720
    },
    {
      "epoch": 0.20091549267323053,
      "grad_norm": 2600.48828125,
      "learning_rate": 3.881950651478481e-05,
      "loss": 101.0363,
      "step": 49730
    },
    {
      "epoch": 0.20095589393859817,
      "grad_norm": 997.9617919921875,
      "learning_rate": 3.8818561130373436e-05,
      "loss": 77.1079,
      "step": 49740
    },
    {
      "epoch": 0.20099629520396578,
      "grad_norm": 725.0342407226562,
      "learning_rate": 3.8817615379084514e-05,
      "loss": 86.4621,
      "step": 49750
    },
    {
      "epoch": 0.20103669646933342,
      "grad_norm": 812.4722900390625,
      "learning_rate": 3.8816669260936494e-05,
      "loss": 109.4252,
      "step": 49760
    },
    {
      "epoch": 0.20107709773470106,
      "grad_norm": 610.0919799804688,
      "learning_rate": 3.881572277594782e-05,
      "loss": 116.1996,
      "step": 49770
    },
    {
      "epoch": 0.20111749900006867,
      "grad_norm": 841.7056884765625,
      "learning_rate": 3.8814775924136945e-05,
      "loss": 100.2897,
      "step": 49780
    },
    {
      "epoch": 0.2011579002654363,
      "grad_norm": 1134.255615234375,
      "learning_rate": 3.8813828705522325e-05,
      "loss": 66.0785,
      "step": 49790
    },
    {
      "epoch": 0.20119830153080395,
      "grad_norm": 805.7269897460938,
      "learning_rate": 3.881288112012243e-05,
      "loss": 97.9667,
      "step": 49800
    },
    {
      "epoch": 0.20123870279617156,
      "grad_norm": 1975.1326904296875,
      "learning_rate": 3.8811933167955734e-05,
      "loss": 128.7955,
      "step": 49810
    },
    {
      "epoch": 0.2012791040615392,
      "grad_norm": 1214.29345703125,
      "learning_rate": 3.8810984849040715e-05,
      "loss": 88.5912,
      "step": 49820
    },
    {
      "epoch": 0.20131950532690684,
      "grad_norm": 1285.6263427734375,
      "learning_rate": 3.8810036163395855e-05,
      "loss": 56.1671,
      "step": 49830
    },
    {
      "epoch": 0.20135990659227446,
      "grad_norm": 748.7781982421875,
      "learning_rate": 3.880908711103966e-05,
      "loss": 56.235,
      "step": 49840
    },
    {
      "epoch": 0.2014003078576421,
      "grad_norm": 453.6629638671875,
      "learning_rate": 3.8808137691990634e-05,
      "loss": 119.4733,
      "step": 49850
    },
    {
      "epoch": 0.20144070912300974,
      "grad_norm": 251.61802673339844,
      "learning_rate": 3.880718790626728e-05,
      "loss": 81.0512,
      "step": 49860
    },
    {
      "epoch": 0.20148111038837738,
      "grad_norm": 583.4518432617188,
      "learning_rate": 3.8806237753888104e-05,
      "loss": 47.1915,
      "step": 49870
    },
    {
      "epoch": 0.201521511653745,
      "grad_norm": 886.7020874023438,
      "learning_rate": 3.880528723487165e-05,
      "loss": 59.9802,
      "step": 49880
    },
    {
      "epoch": 0.20156191291911263,
      "grad_norm": 621.6359252929688,
      "learning_rate": 3.880433634923644e-05,
      "loss": 77.426,
      "step": 49890
    },
    {
      "epoch": 0.20160231418448027,
      "grad_norm": 517.2048950195312,
      "learning_rate": 3.880338509700101e-05,
      "loss": 57.5223,
      "step": 49900
    },
    {
      "epoch": 0.20164271544984788,
      "grad_norm": 603.9962768554688,
      "learning_rate": 3.88024334781839e-05,
      "loss": 68.5799,
      "step": 49910
    },
    {
      "epoch": 0.20168311671521552,
      "grad_norm": 367.4395446777344,
      "learning_rate": 3.880148149280368e-05,
      "loss": 128.3774,
      "step": 49920
    },
    {
      "epoch": 0.20172351798058316,
      "grad_norm": 506.17791748046875,
      "learning_rate": 3.8800529140878894e-05,
      "loss": 66.3896,
      "step": 49930
    },
    {
      "epoch": 0.20176391924595077,
      "grad_norm": 600.4085083007812,
      "learning_rate": 3.879957642242811e-05,
      "loss": 84.5182,
      "step": 49940
    },
    {
      "epoch": 0.2018043205113184,
      "grad_norm": 424.294921875,
      "learning_rate": 3.8798623337469914e-05,
      "loss": 104.3975,
      "step": 49950
    },
    {
      "epoch": 0.20184472177668605,
      "grad_norm": 1428.9373779296875,
      "learning_rate": 3.879766988602287e-05,
      "loss": 81.3248,
      "step": 49960
    },
    {
      "epoch": 0.20188512304205367,
      "grad_norm": 449.1412048339844,
      "learning_rate": 3.879671606810558e-05,
      "loss": 69.7736,
      "step": 49970
    },
    {
      "epoch": 0.2019255243074213,
      "grad_norm": 695.0242309570312,
      "learning_rate": 3.879576188373663e-05,
      "loss": 89.6673,
      "step": 49980
    },
    {
      "epoch": 0.20196592557278895,
      "grad_norm": 1224.8419189453125,
      "learning_rate": 3.879480733293462e-05,
      "loss": 143.0869,
      "step": 49990
    },
    {
      "epoch": 0.20200632683815656,
      "grad_norm": 499.9820556640625,
      "learning_rate": 3.879385241571817e-05,
      "loss": 68.873,
      "step": 50000
    },
    {
      "epoch": 0.2020467281035242,
      "grad_norm": 908.8736572265625,
      "learning_rate": 3.879289713210589e-05,
      "loss": 99.5502,
      "step": 50010
    },
    {
      "epoch": 0.20208712936889184,
      "grad_norm": 424.8169250488281,
      "learning_rate": 3.8791941482116404e-05,
      "loss": 86.1843,
      "step": 50020
    },
    {
      "epoch": 0.20212753063425948,
      "grad_norm": 1957.55029296875,
      "learning_rate": 3.879098546576835e-05,
      "loss": 87.1757,
      "step": 50030
    },
    {
      "epoch": 0.2021679318996271,
      "grad_norm": 1173.882568359375,
      "learning_rate": 3.879002908308036e-05,
      "loss": 114.1178,
      "step": 50040
    },
    {
      "epoch": 0.20220833316499473,
      "grad_norm": 2099.638916015625,
      "learning_rate": 3.878907233407107e-05,
      "loss": 84.3291,
      "step": 50050
    },
    {
      "epoch": 0.20224873443036237,
      "grad_norm": 460.4996643066406,
      "learning_rate": 3.878811521875915e-05,
      "loss": 61.1567,
      "step": 50060
    },
    {
      "epoch": 0.20228913569572998,
      "grad_norm": 400.74761962890625,
      "learning_rate": 3.878715773716325e-05,
      "loss": 79.2012,
      "step": 50070
    },
    {
      "epoch": 0.20232953696109762,
      "grad_norm": 1148.2376708984375,
      "learning_rate": 3.8786199889302035e-05,
      "loss": 73.1642,
      "step": 50080
    },
    {
      "epoch": 0.20236993822646526,
      "grad_norm": 1028.424072265625,
      "learning_rate": 3.8785241675194175e-05,
      "loss": 73.1485,
      "step": 50090
    },
    {
      "epoch": 0.20241033949183287,
      "grad_norm": 832.64697265625,
      "learning_rate": 3.878428309485837e-05,
      "loss": 70.0815,
      "step": 50100
    },
    {
      "epoch": 0.20245074075720051,
      "grad_norm": 1009.1151733398438,
      "learning_rate": 3.878332414831329e-05,
      "loss": 74.1212,
      "step": 50110
    },
    {
      "epoch": 0.20249114202256815,
      "grad_norm": 651.4840698242188,
      "learning_rate": 3.878236483557763e-05,
      "loss": 101.4476,
      "step": 50120
    },
    {
      "epoch": 0.20253154328793577,
      "grad_norm": 852.27978515625,
      "learning_rate": 3.878140515667011e-05,
      "loss": 122.2347,
      "step": 50130
    },
    {
      "epoch": 0.2025719445533034,
      "grad_norm": 675.4700927734375,
      "learning_rate": 3.8780445111609414e-05,
      "loss": 87.2724,
      "step": 50140
    },
    {
      "epoch": 0.20261234581867105,
      "grad_norm": 901.0217895507812,
      "learning_rate": 3.877948470041428e-05,
      "loss": 86.2449,
      "step": 50150
    },
    {
      "epoch": 0.20265274708403866,
      "grad_norm": 1145.986083984375,
      "learning_rate": 3.877852392310342e-05,
      "loss": 90.7556,
      "step": 50160
    },
    {
      "epoch": 0.2026931483494063,
      "grad_norm": 373.2372131347656,
      "learning_rate": 3.877756277969557e-05,
      "loss": 81.8044,
      "step": 50170
    },
    {
      "epoch": 0.20273354961477394,
      "grad_norm": 822.2389526367188,
      "learning_rate": 3.877660127020947e-05,
      "loss": 72.069,
      "step": 50180
    },
    {
      "epoch": 0.20277395088014158,
      "grad_norm": 687.273193359375,
      "learning_rate": 3.877563939466385e-05,
      "loss": 56.9589,
      "step": 50190
    },
    {
      "epoch": 0.2028143521455092,
      "grad_norm": 887.560791015625,
      "learning_rate": 3.8774677153077485e-05,
      "loss": 67.9246,
      "step": 50200
    },
    {
      "epoch": 0.20285475341087683,
      "grad_norm": 526.19287109375,
      "learning_rate": 3.8773714545469116e-05,
      "loss": 90.1091,
      "step": 50210
    },
    {
      "epoch": 0.20289515467624447,
      "grad_norm": 1176.3887939453125,
      "learning_rate": 3.8772751571857525e-05,
      "loss": 91.4007,
      "step": 50220
    },
    {
      "epoch": 0.20293555594161208,
      "grad_norm": 1239.13037109375,
      "learning_rate": 3.877178823226147e-05,
      "loss": 117.7525,
      "step": 50230
    },
    {
      "epoch": 0.20297595720697972,
      "grad_norm": 521.9329833984375,
      "learning_rate": 3.877082452669974e-05,
      "loss": 103.9068,
      "step": 50240
    },
    {
      "epoch": 0.20301635847234736,
      "grad_norm": 807.4822998046875,
      "learning_rate": 3.876986045519112e-05,
      "loss": 62.8522,
      "step": 50250
    },
    {
      "epoch": 0.20305675973771498,
      "grad_norm": 386.10101318359375,
      "learning_rate": 3.876889601775441e-05,
      "loss": 68.9867,
      "step": 50260
    },
    {
      "epoch": 0.20309716100308262,
      "grad_norm": 1182.9283447265625,
      "learning_rate": 3.876793121440841e-05,
      "loss": 88.6812,
      "step": 50270
    },
    {
      "epoch": 0.20313756226845026,
      "grad_norm": 698.7742309570312,
      "learning_rate": 3.876696604517193e-05,
      "loss": 77.0877,
      "step": 50280
    },
    {
      "epoch": 0.20317796353381787,
      "grad_norm": 366.74517822265625,
      "learning_rate": 3.876600051006378e-05,
      "loss": 51.5957,
      "step": 50290
    },
    {
      "epoch": 0.2032183647991855,
      "grad_norm": 834.828857421875,
      "learning_rate": 3.8765034609102794e-05,
      "loss": 64.3455,
      "step": 50300
    },
    {
      "epoch": 0.20325876606455315,
      "grad_norm": 1612.7303466796875,
      "learning_rate": 3.876406834230779e-05,
      "loss": 90.6349,
      "step": 50310
    },
    {
      "epoch": 0.20329916732992076,
      "grad_norm": 986.4638671875,
      "learning_rate": 3.8763101709697625e-05,
      "loss": 92.7768,
      "step": 50320
    },
    {
      "epoch": 0.2033395685952884,
      "grad_norm": 1258.14404296875,
      "learning_rate": 3.876213471129112e-05,
      "loss": 98.2655,
      "step": 50330
    },
    {
      "epoch": 0.20337996986065604,
      "grad_norm": 480.3702087402344,
      "learning_rate": 3.876116734710714e-05,
      "loss": 46.751,
      "step": 50340
    },
    {
      "epoch": 0.20342037112602368,
      "grad_norm": 128.0701141357422,
      "learning_rate": 3.8760199617164556e-05,
      "loss": 87.2124,
      "step": 50350
    },
    {
      "epoch": 0.2034607723913913,
      "grad_norm": 753.6339721679688,
      "learning_rate": 3.8759231521482215e-05,
      "loss": 52.4494,
      "step": 50360
    },
    {
      "epoch": 0.20350117365675893,
      "grad_norm": 717.4784545898438,
      "learning_rate": 3.875826306007899e-05,
      "loss": 82.5807,
      "step": 50370
    },
    {
      "epoch": 0.20354157492212657,
      "grad_norm": 817.7188110351562,
      "learning_rate": 3.875729423297378e-05,
      "loss": 61.2956,
      "step": 50380
    },
    {
      "epoch": 0.20358197618749418,
      "grad_norm": 794.8400268554688,
      "learning_rate": 3.875632504018546e-05,
      "loss": 75.8191,
      "step": 50390
    },
    {
      "epoch": 0.20362237745286182,
      "grad_norm": 1094.506591796875,
      "learning_rate": 3.875535548173292e-05,
      "loss": 61.9088,
      "step": 50400
    },
    {
      "epoch": 0.20366277871822946,
      "grad_norm": 1352.447998046875,
      "learning_rate": 3.8754385557635076e-05,
      "loss": 91.2527,
      "step": 50410
    },
    {
      "epoch": 0.20370317998359708,
      "grad_norm": 474.88079833984375,
      "learning_rate": 3.875341526791083e-05,
      "loss": 66.049,
      "step": 50420
    },
    {
      "epoch": 0.20374358124896472,
      "grad_norm": 1338.16748046875,
      "learning_rate": 3.87524446125791e-05,
      "loss": 70.1112,
      "step": 50430
    },
    {
      "epoch": 0.20378398251433236,
      "grad_norm": 1756.7215576171875,
      "learning_rate": 3.8751473591658805e-05,
      "loss": 91.4397,
      "step": 50440
    },
    {
      "epoch": 0.20382438377969997,
      "grad_norm": 1858.53173828125,
      "learning_rate": 3.8750502205168876e-05,
      "loss": 70.961,
      "step": 50450
    },
    {
      "epoch": 0.2038647850450676,
      "grad_norm": 637.2379150390625,
      "learning_rate": 3.874953045312825e-05,
      "loss": 73.8202,
      "step": 50460
    },
    {
      "epoch": 0.20390518631043525,
      "grad_norm": 1144.32763671875,
      "learning_rate": 3.8748558335555885e-05,
      "loss": 129.5399,
      "step": 50470
    },
    {
      "epoch": 0.20394558757580286,
      "grad_norm": 1619.4119873046875,
      "learning_rate": 3.8747585852470716e-05,
      "loss": 100.1211,
      "step": 50480
    },
    {
      "epoch": 0.2039859888411705,
      "grad_norm": 742.9066162109375,
      "learning_rate": 3.874661300389171e-05,
      "loss": 78.3759,
      "step": 50490
    },
    {
      "epoch": 0.20402639010653814,
      "grad_norm": 663.3787841796875,
      "learning_rate": 3.874563978983784e-05,
      "loss": 70.6641,
      "step": 50500
    },
    {
      "epoch": 0.20406679137190578,
      "grad_norm": 792.5663452148438,
      "learning_rate": 3.874466621032806e-05,
      "loss": 74.8681,
      "step": 50510
    },
    {
      "epoch": 0.2041071926372734,
      "grad_norm": 356.4153747558594,
      "learning_rate": 3.874369226538136e-05,
      "loss": 48.0686,
      "step": 50520
    },
    {
      "epoch": 0.20414759390264103,
      "grad_norm": 848.6477661132812,
      "learning_rate": 3.874271795501674e-05,
      "loss": 75.6694,
      "step": 50530
    },
    {
      "epoch": 0.20418799516800867,
      "grad_norm": 968.4424438476562,
      "learning_rate": 3.8741743279253176e-05,
      "loss": 85.1814,
      "step": 50540
    },
    {
      "epoch": 0.20422839643337629,
      "grad_norm": 906.9072265625,
      "learning_rate": 3.874076823810968e-05,
      "loss": 102.1449,
      "step": 50550
    },
    {
      "epoch": 0.20426879769874393,
      "grad_norm": 1824.4039306640625,
      "learning_rate": 3.873979283160526e-05,
      "loss": 78.7749,
      "step": 50560
    },
    {
      "epoch": 0.20430919896411157,
      "grad_norm": 2528.183349609375,
      "learning_rate": 3.8738817059758936e-05,
      "loss": 76.1273,
      "step": 50570
    },
    {
      "epoch": 0.20434960022947918,
      "grad_norm": 552.7752685546875,
      "learning_rate": 3.873784092258972e-05,
      "loss": 55.3903,
      "step": 50580
    },
    {
      "epoch": 0.20439000149484682,
      "grad_norm": 845.9926147460938,
      "learning_rate": 3.873686442011665e-05,
      "loss": 66.7958,
      "step": 50590
    },
    {
      "epoch": 0.20443040276021446,
      "grad_norm": 1158.8406982421875,
      "learning_rate": 3.873588755235876e-05,
      "loss": 71.1778,
      "step": 50600
    },
    {
      "epoch": 0.20447080402558207,
      "grad_norm": 640.7486572265625,
      "learning_rate": 3.87349103193351e-05,
      "loss": 91.4142,
      "step": 50610
    },
    {
      "epoch": 0.2045112052909497,
      "grad_norm": 1400.6976318359375,
      "learning_rate": 3.8733932721064714e-05,
      "loss": 87.4391,
      "step": 50620
    },
    {
      "epoch": 0.20455160655631735,
      "grad_norm": 552.7567749023438,
      "learning_rate": 3.873295475756667e-05,
      "loss": 78.1331,
      "step": 50630
    },
    {
      "epoch": 0.20459200782168496,
      "grad_norm": 564.9442138671875,
      "learning_rate": 3.873197642886002e-05,
      "loss": 85.3637,
      "step": 50640
    },
    {
      "epoch": 0.2046324090870526,
      "grad_norm": 626.0609130859375,
      "learning_rate": 3.873099773496385e-05,
      "loss": 99.6655,
      "step": 50650
    },
    {
      "epoch": 0.20467281035242024,
      "grad_norm": 425.89886474609375,
      "learning_rate": 3.873001867589723e-05,
      "loss": 58.2747,
      "step": 50660
    },
    {
      "epoch": 0.20471321161778788,
      "grad_norm": 699.8375854492188,
      "learning_rate": 3.872903925167927e-05,
      "loss": 70.8648,
      "step": 50670
    },
    {
      "epoch": 0.2047536128831555,
      "grad_norm": 1484.742919921875,
      "learning_rate": 3.8728059462329035e-05,
      "loss": 55.6887,
      "step": 50680
    },
    {
      "epoch": 0.20479401414852313,
      "grad_norm": 1168.312255859375,
      "learning_rate": 3.872707930786564e-05,
      "loss": 105.1602,
      "step": 50690
    },
    {
      "epoch": 0.20483441541389077,
      "grad_norm": 942.3574829101562,
      "learning_rate": 3.872609878830819e-05,
      "loss": 88.1229,
      "step": 50700
    },
    {
      "epoch": 0.2048748166792584,
      "grad_norm": 527.11572265625,
      "learning_rate": 3.872511790367581e-05,
      "loss": 73.1579,
      "step": 50710
    },
    {
      "epoch": 0.20491521794462603,
      "grad_norm": 569.9839477539062,
      "learning_rate": 3.872413665398761e-05,
      "loss": 100.4503,
      "step": 50720
    },
    {
      "epoch": 0.20495561920999367,
      "grad_norm": 712.02880859375,
      "learning_rate": 3.872315503926273e-05,
      "loss": 67.1564,
      "step": 50730
    },
    {
      "epoch": 0.20499602047536128,
      "grad_norm": 894.4396362304688,
      "learning_rate": 3.87221730595203e-05,
      "loss": 61.3889,
      "step": 50740
    },
    {
      "epoch": 0.20503642174072892,
      "grad_norm": 1530.17724609375,
      "learning_rate": 3.872119071477947e-05,
      "loss": 82.5945,
      "step": 50750
    },
    {
      "epoch": 0.20507682300609656,
      "grad_norm": 529.53515625,
      "learning_rate": 3.8720208005059386e-05,
      "loss": 68.7454,
      "step": 50760
    },
    {
      "epoch": 0.20511722427146417,
      "grad_norm": 674.3864135742188,
      "learning_rate": 3.871922493037921e-05,
      "loss": 86.0123,
      "step": 50770
    },
    {
      "epoch": 0.2051576255368318,
      "grad_norm": 1044.22021484375,
      "learning_rate": 3.8718241490758106e-05,
      "loss": 92.6389,
      "step": 50780
    },
    {
      "epoch": 0.20519802680219945,
      "grad_norm": 853.1271362304688,
      "learning_rate": 3.871725768621526e-05,
      "loss": 66.9134,
      "step": 50790
    },
    {
      "epoch": 0.20523842806756706,
      "grad_norm": 583.3541259765625,
      "learning_rate": 3.871627351676982e-05,
      "loss": 64.4168,
      "step": 50800
    },
    {
      "epoch": 0.2052788293329347,
      "grad_norm": 573.4091796875,
      "learning_rate": 3.8715288982441e-05,
      "loss": 76.0434,
      "step": 50810
    },
    {
      "epoch": 0.20531923059830234,
      "grad_norm": 379.55615234375,
      "learning_rate": 3.8714304083247984e-05,
      "loss": 51.8694,
      "step": 50820
    },
    {
      "epoch": 0.20535963186366998,
      "grad_norm": 832.7188110351562,
      "learning_rate": 3.871331881920998e-05,
      "loss": 58.693,
      "step": 50830
    },
    {
      "epoch": 0.2054000331290376,
      "grad_norm": 2003.7796630859375,
      "learning_rate": 3.871233319034619e-05,
      "loss": 75.3883,
      "step": 50840
    },
    {
      "epoch": 0.20544043439440524,
      "grad_norm": 882.434814453125,
      "learning_rate": 3.871134719667583e-05,
      "loss": 103.4826,
      "step": 50850
    },
    {
      "epoch": 0.20548083565977288,
      "grad_norm": 796.6503295898438,
      "learning_rate": 3.8710360838218125e-05,
      "loss": 103.9689,
      "step": 50860
    },
    {
      "epoch": 0.2055212369251405,
      "grad_norm": 844.0625610351562,
      "learning_rate": 3.870937411499229e-05,
      "loss": 85.5435,
      "step": 50870
    },
    {
      "epoch": 0.20556163819050813,
      "grad_norm": 1352.2198486328125,
      "learning_rate": 3.870838702701759e-05,
      "loss": 89.8294,
      "step": 50880
    },
    {
      "epoch": 0.20560203945587577,
      "grad_norm": 1182.203369140625,
      "learning_rate": 3.8707399574313246e-05,
      "loss": 72.2934,
      "step": 50890
    },
    {
      "epoch": 0.20564244072124338,
      "grad_norm": 541.3485107421875,
      "learning_rate": 3.870641175689852e-05,
      "loss": 72.552,
      "step": 50900
    },
    {
      "epoch": 0.20568284198661102,
      "grad_norm": 773.8545532226562,
      "learning_rate": 3.870542357479266e-05,
      "loss": 99.4841,
      "step": 50910
    },
    {
      "epoch": 0.20572324325197866,
      "grad_norm": 521.749755859375,
      "learning_rate": 3.870443502801494e-05,
      "loss": 71.8828,
      "step": 50920
    },
    {
      "epoch": 0.20576364451734627,
      "grad_norm": 651.1212768554688,
      "learning_rate": 3.870344611658464e-05,
      "loss": 113.8711,
      "step": 50930
    },
    {
      "epoch": 0.2058040457827139,
      "grad_norm": 718.744140625,
      "learning_rate": 3.870245684052101e-05,
      "loss": 98.3473,
      "step": 50940
    },
    {
      "epoch": 0.20584444704808155,
      "grad_norm": 568.1864013671875,
      "learning_rate": 3.870146719984337e-05,
      "loss": 99.5937,
      "step": 50950
    },
    {
      "epoch": 0.20588484831344916,
      "grad_norm": 5688.7666015625,
      "learning_rate": 3.870047719457099e-05,
      "loss": 114.7975,
      "step": 50960
    },
    {
      "epoch": 0.2059252495788168,
      "grad_norm": 1301.0987548828125,
      "learning_rate": 3.869948682472318e-05,
      "loss": 65.0882,
      "step": 50970
    },
    {
      "epoch": 0.20596565084418444,
      "grad_norm": 1446.244873046875,
      "learning_rate": 3.8698496090319257e-05,
      "loss": 77.6484,
      "step": 50980
    },
    {
      "epoch": 0.20600605210955208,
      "grad_norm": 682.64306640625,
      "learning_rate": 3.869750499137851e-05,
      "loss": 59.3151,
      "step": 50990
    },
    {
      "epoch": 0.2060464533749197,
      "grad_norm": 1040.8548583984375,
      "learning_rate": 3.869651352792029e-05,
      "loss": 89.0893,
      "step": 51000
    },
    {
      "epoch": 0.20608685464028734,
      "grad_norm": 894.2227172851562,
      "learning_rate": 3.869552169996391e-05,
      "loss": 104.3016,
      "step": 51010
    },
    {
      "epoch": 0.20612725590565498,
      "grad_norm": 1589.9327392578125,
      "learning_rate": 3.869452950752871e-05,
      "loss": 64.4769,
      "step": 51020
    },
    {
      "epoch": 0.2061676571710226,
      "grad_norm": 2216.80126953125,
      "learning_rate": 3.869353695063403e-05,
      "loss": 75.212,
      "step": 51030
    },
    {
      "epoch": 0.20620805843639023,
      "grad_norm": 1083.3333740234375,
      "learning_rate": 3.8692544029299225e-05,
      "loss": 67.3579,
      "step": 51040
    },
    {
      "epoch": 0.20624845970175787,
      "grad_norm": 1293.1751708984375,
      "learning_rate": 3.869155074354365e-05,
      "loss": 83.4344,
      "step": 51050
    },
    {
      "epoch": 0.20628886096712548,
      "grad_norm": 1822.27587890625,
      "learning_rate": 3.869055709338667e-05,
      "loss": 114.4531,
      "step": 51060
    },
    {
      "epoch": 0.20632926223249312,
      "grad_norm": 767.5370483398438,
      "learning_rate": 3.8689563078847656e-05,
      "loss": 71.1274,
      "step": 51070
    },
    {
      "epoch": 0.20636966349786076,
      "grad_norm": 781.1602783203125,
      "learning_rate": 3.868856869994599e-05,
      "loss": 75.8399,
      "step": 51080
    },
    {
      "epoch": 0.20641006476322837,
      "grad_norm": 947.2127075195312,
      "learning_rate": 3.868757395670105e-05,
      "loss": 116.0168,
      "step": 51090
    },
    {
      "epoch": 0.206450466028596,
      "grad_norm": 1369.5260009765625,
      "learning_rate": 3.8686578849132244e-05,
      "loss": 93.0958,
      "step": 51100
    },
    {
      "epoch": 0.20649086729396365,
      "grad_norm": 949.6033935546875,
      "learning_rate": 3.8685583377258955e-05,
      "loss": 59.0235,
      "step": 51110
    },
    {
      "epoch": 0.20653126855933127,
      "grad_norm": 528.279296875,
      "learning_rate": 3.86845875411006e-05,
      "loss": 78.9101,
      "step": 51120
    },
    {
      "epoch": 0.2065716698246989,
      "grad_norm": 1002.0960083007812,
      "learning_rate": 3.8683591340676596e-05,
      "loss": 93.5453,
      "step": 51130
    },
    {
      "epoch": 0.20661207109006655,
      "grad_norm": 651.449462890625,
      "learning_rate": 3.868259477600636e-05,
      "loss": 86.5347,
      "step": 51140
    },
    {
      "epoch": 0.20665247235543419,
      "grad_norm": 851.9552001953125,
      "learning_rate": 3.868159784710931e-05,
      "loss": 110.314,
      "step": 51150
    },
    {
      "epoch": 0.2066928736208018,
      "grad_norm": 632.49951171875,
      "learning_rate": 3.8680600554004905e-05,
      "loss": 70.093,
      "step": 51160
    },
    {
      "epoch": 0.20673327488616944,
      "grad_norm": 987.5447998046875,
      "learning_rate": 3.867960289671257e-05,
      "loss": 73.0326,
      "step": 51170
    },
    {
      "epoch": 0.20677367615153708,
      "grad_norm": 1574.3292236328125,
      "learning_rate": 3.867860487525175e-05,
      "loss": 109.6682,
      "step": 51180
    },
    {
      "epoch": 0.2068140774169047,
      "grad_norm": 530.6597900390625,
      "learning_rate": 3.8677606489641924e-05,
      "loss": 69.3783,
      "step": 51190
    },
    {
      "epoch": 0.20685447868227233,
      "grad_norm": 1183.2138671875,
      "learning_rate": 3.867660773990254e-05,
      "loss": 102.358,
      "step": 51200
    },
    {
      "epoch": 0.20689487994763997,
      "grad_norm": 623.4737548828125,
      "learning_rate": 3.867560862605307e-05,
      "loss": 63.161,
      "step": 51210
    },
    {
      "epoch": 0.20693528121300758,
      "grad_norm": 302.43426513671875,
      "learning_rate": 3.8674609148112996e-05,
      "loss": 76.0289,
      "step": 51220
    },
    {
      "epoch": 0.20697568247837522,
      "grad_norm": 532.8798217773438,
      "learning_rate": 3.86736093061018e-05,
      "loss": 115.1544,
      "step": 51230
    },
    {
      "epoch": 0.20701608374374286,
      "grad_norm": 871.456787109375,
      "learning_rate": 3.8672609100038986e-05,
      "loss": 78.027,
      "step": 51240
    },
    {
      "epoch": 0.20705648500911047,
      "grad_norm": 733.9722290039062,
      "learning_rate": 3.8671608529944035e-05,
      "loss": 89.2671,
      "step": 51250
    },
    {
      "epoch": 0.20709688627447811,
      "grad_norm": 589.8917236328125,
      "learning_rate": 3.867060759583647e-05,
      "loss": 61.1345,
      "step": 51260
    },
    {
      "epoch": 0.20713728753984575,
      "grad_norm": 550.5474243164062,
      "learning_rate": 3.866960629773579e-05,
      "loss": 76.382,
      "step": 51270
    },
    {
      "epoch": 0.20717768880521337,
      "grad_norm": 612.1178588867188,
      "learning_rate": 3.866860463566153e-05,
      "loss": 70.8993,
      "step": 51280
    },
    {
      "epoch": 0.207218090070581,
      "grad_norm": 1227.650634765625,
      "learning_rate": 3.866760260963321e-05,
      "loss": 54.2812,
      "step": 51290
    },
    {
      "epoch": 0.20725849133594865,
      "grad_norm": 629.446533203125,
      "learning_rate": 3.8666600219670365e-05,
      "loss": 86.5742,
      "step": 51300
    },
    {
      "epoch": 0.2072988926013163,
      "grad_norm": 643.613525390625,
      "learning_rate": 3.866559746579254e-05,
      "loss": 62.8,
      "step": 51310
    },
    {
      "epoch": 0.2073392938666839,
      "grad_norm": 1438.744140625,
      "learning_rate": 3.866459434801928e-05,
      "loss": 68.6836,
      "step": 51320
    },
    {
      "epoch": 0.20737969513205154,
      "grad_norm": 446.6057434082031,
      "learning_rate": 3.8663590866370147e-05,
      "loss": 67.867,
      "step": 51330
    },
    {
      "epoch": 0.20742009639741918,
      "grad_norm": 733.7489624023438,
      "learning_rate": 3.86625870208647e-05,
      "loss": 88.8869,
      "step": 51340
    },
    {
      "epoch": 0.2074604976627868,
      "grad_norm": 1379.1614990234375,
      "learning_rate": 3.866158281152251e-05,
      "loss": 61.0902,
      "step": 51350
    },
    {
      "epoch": 0.20750089892815443,
      "grad_norm": 1295.2818603515625,
      "learning_rate": 3.8660578238363156e-05,
      "loss": 55.145,
      "step": 51360
    },
    {
      "epoch": 0.20754130019352207,
      "grad_norm": 1555.087158203125,
      "learning_rate": 3.865957330140622e-05,
      "loss": 85.2966,
      "step": 51370
    },
    {
      "epoch": 0.20758170145888968,
      "grad_norm": 1380.683837890625,
      "learning_rate": 3.86585680006713e-05,
      "loss": 81.9897,
      "step": 51380
    },
    {
      "epoch": 0.20762210272425732,
      "grad_norm": 670.623291015625,
      "learning_rate": 3.865756233617799e-05,
      "loss": 101.2558,
      "step": 51390
    },
    {
      "epoch": 0.20766250398962496,
      "grad_norm": 602.7833251953125,
      "learning_rate": 3.8656556307945894e-05,
      "loss": 57.1718,
      "step": 51400
    },
    {
      "epoch": 0.20770290525499258,
      "grad_norm": 1327.6904296875,
      "learning_rate": 3.8655549915994626e-05,
      "loss": 82.1553,
      "step": 51410
    },
    {
      "epoch": 0.20774330652036022,
      "grad_norm": 1156.4627685546875,
      "learning_rate": 3.8654543160343816e-05,
      "loss": 101.3851,
      "step": 51420
    },
    {
      "epoch": 0.20778370778572786,
      "grad_norm": 1305.48974609375,
      "learning_rate": 3.8653536041013075e-05,
      "loss": 82.6139,
      "step": 51430
    },
    {
      "epoch": 0.20782410905109547,
      "grad_norm": 755.2139892578125,
      "learning_rate": 3.865252855802205e-05,
      "loss": 49.8036,
      "step": 51440
    },
    {
      "epoch": 0.2078645103164631,
      "grad_norm": 902.1483764648438,
      "learning_rate": 3.865152071139038e-05,
      "loss": 74.7251,
      "step": 51450
    },
    {
      "epoch": 0.20790491158183075,
      "grad_norm": 910.2382202148438,
      "learning_rate": 3.8650512501137704e-05,
      "loss": 59.3921,
      "step": 51460
    },
    {
      "epoch": 0.2079453128471984,
      "grad_norm": 1980.9892578125,
      "learning_rate": 3.8649503927283686e-05,
      "loss": 114.7764,
      "step": 51470
    },
    {
      "epoch": 0.207985714112566,
      "grad_norm": 922.23583984375,
      "learning_rate": 3.864849498984799e-05,
      "loss": 60.0026,
      "step": 51480
    },
    {
      "epoch": 0.20802611537793364,
      "grad_norm": 565.62548828125,
      "learning_rate": 3.864748568885029e-05,
      "loss": 101.4692,
      "step": 51490
    },
    {
      "epoch": 0.20806651664330128,
      "grad_norm": 476.4505615234375,
      "learning_rate": 3.864647602431025e-05,
      "loss": 59.0585,
      "step": 51500
    },
    {
      "epoch": 0.2081069179086689,
      "grad_norm": 794.0309448242188,
      "learning_rate": 3.864546599624756e-05,
      "loss": 75.9052,
      "step": 51510
    },
    {
      "epoch": 0.20814731917403653,
      "grad_norm": 659.0484619140625,
      "learning_rate": 3.864445560468191e-05,
      "loss": 94.3838,
      "step": 51520
    },
    {
      "epoch": 0.20818772043940417,
      "grad_norm": 1586.227294921875,
      "learning_rate": 3.8643444849633e-05,
      "loss": 109.3399,
      "step": 51530
    },
    {
      "epoch": 0.20822812170477178,
      "grad_norm": 415.051513671875,
      "learning_rate": 3.864243373112053e-05,
      "loss": 52.1959,
      "step": 51540
    },
    {
      "epoch": 0.20826852297013942,
      "grad_norm": 1006.6847534179688,
      "learning_rate": 3.864142224916422e-05,
      "loss": 77.8924,
      "step": 51550
    },
    {
      "epoch": 0.20830892423550706,
      "grad_norm": 642.0609130859375,
      "learning_rate": 3.864041040378379e-05,
      "loss": 48.1212,
      "step": 51560
    },
    {
      "epoch": 0.20834932550087468,
      "grad_norm": 704.0679931640625,
      "learning_rate": 3.863939819499896e-05,
      "loss": 59.1892,
      "step": 51570
    },
    {
      "epoch": 0.20838972676624232,
      "grad_norm": 866.2320556640625,
      "learning_rate": 3.8638385622829464e-05,
      "loss": 109.2826,
      "step": 51580
    },
    {
      "epoch": 0.20843012803160996,
      "grad_norm": 560.8289794921875,
      "learning_rate": 3.863737268729504e-05,
      "loss": 59.0756,
      "step": 51590
    },
    {
      "epoch": 0.20847052929697757,
      "grad_norm": 1119.5093994140625,
      "learning_rate": 3.863635938841545e-05,
      "loss": 63.5763,
      "step": 51600
    },
    {
      "epoch": 0.2085109305623452,
      "grad_norm": 484.5401306152344,
      "learning_rate": 3.8635345726210435e-05,
      "loss": 100.063,
      "step": 51610
    },
    {
      "epoch": 0.20855133182771285,
      "grad_norm": 1919.9112548828125,
      "learning_rate": 3.863433170069976e-05,
      "loss": 65.146,
      "step": 51620
    },
    {
      "epoch": 0.20859173309308046,
      "grad_norm": 329.201904296875,
      "learning_rate": 3.86333173119032e-05,
      "loss": 79.1359,
      "step": 51630
    },
    {
      "epoch": 0.2086321343584481,
      "grad_norm": 1020.9974365234375,
      "learning_rate": 3.863230255984052e-05,
      "loss": 43.7593,
      "step": 51640
    },
    {
      "epoch": 0.20867253562381574,
      "grad_norm": 394.7405700683594,
      "learning_rate": 3.863128744453152e-05,
      "loss": 59.4796,
      "step": 51650
    },
    {
      "epoch": 0.20871293688918338,
      "grad_norm": 674.1074829101562,
      "learning_rate": 3.863027196599596e-05,
      "loss": 76.5259,
      "step": 51660
    },
    {
      "epoch": 0.208753338154551,
      "grad_norm": 1158.9613037109375,
      "learning_rate": 3.8629256124253675e-05,
      "loss": 83.799,
      "step": 51670
    },
    {
      "epoch": 0.20879373941991863,
      "grad_norm": 998.66064453125,
      "learning_rate": 3.8628239919324435e-05,
      "loss": 50.7926,
      "step": 51680
    },
    {
      "epoch": 0.20883414068528627,
      "grad_norm": 1256.56396484375,
      "learning_rate": 3.862722335122808e-05,
      "loss": 94.6407,
      "step": 51690
    },
    {
      "epoch": 0.20887454195065389,
      "grad_norm": 0.0,
      "learning_rate": 3.862620641998441e-05,
      "loss": 70.0929,
      "step": 51700
    },
    {
      "epoch": 0.20891494321602153,
      "grad_norm": 926.8525390625,
      "learning_rate": 3.862518912561326e-05,
      "loss": 67.6057,
      "step": 51710
    },
    {
      "epoch": 0.20895534448138917,
      "grad_norm": 326.4804382324219,
      "learning_rate": 3.862417146813445e-05,
      "loss": 61.7977,
      "step": 51720
    },
    {
      "epoch": 0.20899574574675678,
      "grad_norm": 477.6756896972656,
      "learning_rate": 3.862315344756784e-05,
      "loss": 92.2266,
      "step": 51730
    },
    {
      "epoch": 0.20903614701212442,
      "grad_norm": 405.753173828125,
      "learning_rate": 3.862213506393326e-05,
      "loss": 70.3351,
      "step": 51740
    },
    {
      "epoch": 0.20907654827749206,
      "grad_norm": 1080.689453125,
      "learning_rate": 3.862111631725057e-05,
      "loss": 96.0838,
      "step": 51750
    },
    {
      "epoch": 0.20911694954285967,
      "grad_norm": 627.3723754882812,
      "learning_rate": 3.862009720753963e-05,
      "loss": 93.8915,
      "step": 51760
    },
    {
      "epoch": 0.2091573508082273,
      "grad_norm": 820.82763671875,
      "learning_rate": 3.8619077734820314e-05,
      "loss": 61.6034,
      "step": 51770
    },
    {
      "epoch": 0.20919775207359495,
      "grad_norm": 842.6414184570312,
      "learning_rate": 3.861805789911248e-05,
      "loss": 153.1677,
      "step": 51780
    },
    {
      "epoch": 0.20923815333896256,
      "grad_norm": 714.5660400390625,
      "learning_rate": 3.8617037700436034e-05,
      "loss": 57.6857,
      "step": 51790
    },
    {
      "epoch": 0.2092785546043302,
      "grad_norm": 627.9758911132812,
      "learning_rate": 3.861601713881085e-05,
      "loss": 55.2468,
      "step": 51800
    },
    {
      "epoch": 0.20931895586969784,
      "grad_norm": 1113.80712890625,
      "learning_rate": 3.8614996214256826e-05,
      "loss": 89.2621,
      "step": 51810
    },
    {
      "epoch": 0.20935935713506548,
      "grad_norm": 531.0305786132812,
      "learning_rate": 3.861397492679387e-05,
      "loss": 61.7609,
      "step": 51820
    },
    {
      "epoch": 0.2093997584004331,
      "grad_norm": 1602.581787109375,
      "learning_rate": 3.861295327644189e-05,
      "loss": 76.7713,
      "step": 51830
    },
    {
      "epoch": 0.20944015966580073,
      "grad_norm": 1647.34130859375,
      "learning_rate": 3.8611931263220794e-05,
      "loss": 88.0155,
      "step": 51840
    },
    {
      "epoch": 0.20948056093116837,
      "grad_norm": 1438.4656982421875,
      "learning_rate": 3.861090888715052e-05,
      "loss": 94.1094,
      "step": 51850
    },
    {
      "epoch": 0.209520962196536,
      "grad_norm": 1173.360107421875,
      "learning_rate": 3.8609886148251006e-05,
      "loss": 135.0007,
      "step": 51860
    },
    {
      "epoch": 0.20956136346190363,
      "grad_norm": 1216.895751953125,
      "learning_rate": 3.8608863046542164e-05,
      "loss": 61.3545,
      "step": 51870
    },
    {
      "epoch": 0.20960176472727127,
      "grad_norm": 852.8516845703125,
      "learning_rate": 3.860783958204397e-05,
      "loss": 90.4901,
      "step": 51880
    },
    {
      "epoch": 0.20964216599263888,
      "grad_norm": 452.8900146484375,
      "learning_rate": 3.860681575477636e-05,
      "loss": 62.5983,
      "step": 51890
    },
    {
      "epoch": 0.20968256725800652,
      "grad_norm": 377.25396728515625,
      "learning_rate": 3.8605791564759296e-05,
      "loss": 59.1545,
      "step": 51900
    },
    {
      "epoch": 0.20972296852337416,
      "grad_norm": 1327.0433349609375,
      "learning_rate": 3.8604767012012746e-05,
      "loss": 90.5913,
      "step": 51910
    },
    {
      "epoch": 0.20976336978874177,
      "grad_norm": 653.3995971679688,
      "learning_rate": 3.8603742096556687e-05,
      "loss": 86.1637,
      "step": 51920
    },
    {
      "epoch": 0.2098037710541094,
      "grad_norm": 574.94873046875,
      "learning_rate": 3.8602716818411093e-05,
      "loss": 74.9849,
      "step": 51930
    },
    {
      "epoch": 0.20984417231947705,
      "grad_norm": 793.8121337890625,
      "learning_rate": 3.8601691177595964e-05,
      "loss": 71.8904,
      "step": 51940
    },
    {
      "epoch": 0.20988457358484466,
      "grad_norm": 1037.7066650390625,
      "learning_rate": 3.860066517413129e-05,
      "loss": 84.2814,
      "step": 51950
    },
    {
      "epoch": 0.2099249748502123,
      "grad_norm": 623.3878784179688,
      "learning_rate": 3.859963880803706e-05,
      "loss": 63.5703,
      "step": 51960
    },
    {
      "epoch": 0.20996537611557994,
      "grad_norm": 418.7212219238281,
      "learning_rate": 3.859861207933331e-05,
      "loss": 64.8058,
      "step": 51970
    },
    {
      "epoch": 0.21000577738094758,
      "grad_norm": 913.1371459960938,
      "learning_rate": 3.8597584988040034e-05,
      "loss": 55.0011,
      "step": 51980
    },
    {
      "epoch": 0.2100461786463152,
      "grad_norm": 1014.57861328125,
      "learning_rate": 3.859655753417726e-05,
      "loss": 82.6761,
      "step": 51990
    },
    {
      "epoch": 0.21008657991168284,
      "grad_norm": 930.0344848632812,
      "learning_rate": 3.859552971776503e-05,
      "loss": 86.5342,
      "step": 52000
    },
    {
      "epoch": 0.21012698117705048,
      "grad_norm": 1351.761474609375,
      "learning_rate": 3.8594501538823374e-05,
      "loss": 62.6794,
      "step": 52010
    },
    {
      "epoch": 0.2101673824424181,
      "grad_norm": 786.5313720703125,
      "learning_rate": 3.8593472997372336e-05,
      "loss": 77.6322,
      "step": 52020
    },
    {
      "epoch": 0.21020778370778573,
      "grad_norm": 1234.1259765625,
      "learning_rate": 3.8592444093431976e-05,
      "loss": 62.3295,
      "step": 52030
    },
    {
      "epoch": 0.21024818497315337,
      "grad_norm": 829.6130981445312,
      "learning_rate": 3.859141482702233e-05,
      "loss": 84.1906,
      "step": 52040
    },
    {
      "epoch": 0.21028858623852098,
      "grad_norm": 488.3553771972656,
      "learning_rate": 3.8590385198163495e-05,
      "loss": 65.9482,
      "step": 52050
    },
    {
      "epoch": 0.21032898750388862,
      "grad_norm": 1383.439208984375,
      "learning_rate": 3.858935520687553e-05,
      "loss": 84.0547,
      "step": 52060
    },
    {
      "epoch": 0.21036938876925626,
      "grad_norm": 555.8527221679688,
      "learning_rate": 3.858832485317851e-05,
      "loss": 82.3044,
      "step": 52070
    },
    {
      "epoch": 0.21040979003462387,
      "grad_norm": 410.3157958984375,
      "learning_rate": 3.8587294137092526e-05,
      "loss": 68.6609,
      "step": 52080
    },
    {
      "epoch": 0.2104501912999915,
      "grad_norm": 867.0938720703125,
      "learning_rate": 3.858626305863767e-05,
      "loss": 86.5969,
      "step": 52090
    },
    {
      "epoch": 0.21049059256535915,
      "grad_norm": 1007.9744262695312,
      "learning_rate": 3.8585231617834054e-05,
      "loss": 87.7424,
      "step": 52100
    },
    {
      "epoch": 0.21053099383072676,
      "grad_norm": 855.5218505859375,
      "learning_rate": 3.8584199814701774e-05,
      "loss": 67.6049,
      "step": 52110
    },
    {
      "epoch": 0.2105713950960944,
      "grad_norm": 1032.388427734375,
      "learning_rate": 3.8583167649260956e-05,
      "loss": 81.8633,
      "step": 52120
    },
    {
      "epoch": 0.21061179636146204,
      "grad_norm": 846.8680419921875,
      "learning_rate": 3.858213512153171e-05,
      "loss": 122.2433,
      "step": 52130
    },
    {
      "epoch": 0.21065219762682968,
      "grad_norm": 658.4915161132812,
      "learning_rate": 3.858110223153418e-05,
      "loss": 43.3111,
      "step": 52140
    },
    {
      "epoch": 0.2106925988921973,
      "grad_norm": 494.8357238769531,
      "learning_rate": 3.8580068979288495e-05,
      "loss": 81.8783,
      "step": 52150
    },
    {
      "epoch": 0.21073300015756494,
      "grad_norm": 657.0929565429688,
      "learning_rate": 3.8579035364814793e-05,
      "loss": 95.437,
      "step": 52160
    },
    {
      "epoch": 0.21077340142293258,
      "grad_norm": 607.5389404296875,
      "learning_rate": 3.857800138813324e-05,
      "loss": 99.3467,
      "step": 52170
    },
    {
      "epoch": 0.2108138026883002,
      "grad_norm": 1181.9508056640625,
      "learning_rate": 3.857696704926398e-05,
      "loss": 71.005,
      "step": 52180
    },
    {
      "epoch": 0.21085420395366783,
      "grad_norm": 888.6665649414062,
      "learning_rate": 3.857593234822718e-05,
      "loss": 97.3609,
      "step": 52190
    },
    {
      "epoch": 0.21089460521903547,
      "grad_norm": 0.0,
      "learning_rate": 3.857489728504303e-05,
      "loss": 54.3101,
      "step": 52200
    },
    {
      "epoch": 0.21093500648440308,
      "grad_norm": 316.5389099121094,
      "learning_rate": 3.857386185973168e-05,
      "loss": 108.8267,
      "step": 52210
    },
    {
      "epoch": 0.21097540774977072,
      "grad_norm": 838.7566528320312,
      "learning_rate": 3.857282607231334e-05,
      "loss": 69.4257,
      "step": 52220
    },
    {
      "epoch": 0.21101580901513836,
      "grad_norm": 485.1976623535156,
      "learning_rate": 3.857178992280819e-05,
      "loss": 60.2769,
      "step": 52230
    },
    {
      "epoch": 0.21105621028050597,
      "grad_norm": 528.6210327148438,
      "learning_rate": 3.857075341123643e-05,
      "loss": 63.1721,
      "step": 52240
    },
    {
      "epoch": 0.2110966115458736,
      "grad_norm": 664.318115234375,
      "learning_rate": 3.856971653761827e-05,
      "loss": 81.4066,
      "step": 52250
    },
    {
      "epoch": 0.21113701281124125,
      "grad_norm": 818.0326538085938,
      "learning_rate": 3.856867930197393e-05,
      "loss": 106.0535,
      "step": 52260
    },
    {
      "epoch": 0.21117741407660887,
      "grad_norm": 646.04638671875,
      "learning_rate": 3.8567641704323636e-05,
      "loss": 82.7102,
      "step": 52270
    },
    {
      "epoch": 0.2112178153419765,
      "grad_norm": 515.3419799804688,
      "learning_rate": 3.8566603744687595e-05,
      "loss": 60.0256,
      "step": 52280
    },
    {
      "epoch": 0.21125821660734415,
      "grad_norm": 570.6306762695312,
      "learning_rate": 3.856556542308606e-05,
      "loss": 118.7118,
      "step": 52290
    },
    {
      "epoch": 0.21129861787271179,
      "grad_norm": 588.5262451171875,
      "learning_rate": 3.8564526739539266e-05,
      "loss": 87.5939,
      "step": 52300
    },
    {
      "epoch": 0.2113390191380794,
      "grad_norm": 433.0246887207031,
      "learning_rate": 3.856348769406747e-05,
      "loss": 57.9482,
      "step": 52310
    },
    {
      "epoch": 0.21137942040344704,
      "grad_norm": 856.47216796875,
      "learning_rate": 3.856244828669092e-05,
      "loss": 78.945,
      "step": 52320
    },
    {
      "epoch": 0.21141982166881468,
      "grad_norm": 381.0574951171875,
      "learning_rate": 3.856140851742989e-05,
      "loss": 48.9479,
      "step": 52330
    },
    {
      "epoch": 0.2114602229341823,
      "grad_norm": 1054.4591064453125,
      "learning_rate": 3.856036838630464e-05,
      "loss": 101.4416,
      "step": 52340
    },
    {
      "epoch": 0.21150062419954993,
      "grad_norm": 488.54656982421875,
      "learning_rate": 3.855932789333546e-05,
      "loss": 94.751,
      "step": 52350
    },
    {
      "epoch": 0.21154102546491757,
      "grad_norm": 1185.2191162109375,
      "learning_rate": 3.855828703854262e-05,
      "loss": 103.5967,
      "step": 52360
    },
    {
      "epoch": 0.21158142673028518,
      "grad_norm": 958.7376708984375,
      "learning_rate": 3.8557245821946414e-05,
      "loss": 84.2619,
      "step": 52370
    },
    {
      "epoch": 0.21162182799565282,
      "grad_norm": 808.8856811523438,
      "learning_rate": 3.8556204243567156e-05,
      "loss": 138.5501,
      "step": 52380
    },
    {
      "epoch": 0.21166222926102046,
      "grad_norm": 712.3560791015625,
      "learning_rate": 3.855516230342514e-05,
      "loss": 71.5041,
      "step": 52390
    },
    {
      "epoch": 0.21170263052638807,
      "grad_norm": 976.9653930664062,
      "learning_rate": 3.8554120001540684e-05,
      "loss": 64.854,
      "step": 52400
    },
    {
      "epoch": 0.21174303179175571,
      "grad_norm": 1233.3726806640625,
      "learning_rate": 3.8553077337934114e-05,
      "loss": 65.5629,
      "step": 52410
    },
    {
      "epoch": 0.21178343305712335,
      "grad_norm": 502.90521240234375,
      "learning_rate": 3.855203431262574e-05,
      "loss": 73.8202,
      "step": 52420
    },
    {
      "epoch": 0.21182383432249097,
      "grad_norm": 975.7722778320312,
      "learning_rate": 3.85509909256359e-05,
      "loss": 103.7028,
      "step": 52430
    },
    {
      "epoch": 0.2118642355878586,
      "grad_norm": 1887.445068359375,
      "learning_rate": 3.854994717698495e-05,
      "loss": 108.6903,
      "step": 52440
    },
    {
      "epoch": 0.21190463685322625,
      "grad_norm": 1207.58203125,
      "learning_rate": 3.8548903066693234e-05,
      "loss": 60.712,
      "step": 52450
    },
    {
      "epoch": 0.2119450381185939,
      "grad_norm": 774.343017578125,
      "learning_rate": 3.854785859478109e-05,
      "loss": 68.7223,
      "step": 52460
    },
    {
      "epoch": 0.2119854393839615,
      "grad_norm": 659.7357788085938,
      "learning_rate": 3.85468137612689e-05,
      "loss": 50.9767,
      "step": 52470
    },
    {
      "epoch": 0.21202584064932914,
      "grad_norm": 601.4818115234375,
      "learning_rate": 3.8545768566177034e-05,
      "loss": 63.6934,
      "step": 52480
    },
    {
      "epoch": 0.21206624191469678,
      "grad_norm": 1886.767333984375,
      "learning_rate": 3.8544723009525855e-05,
      "loss": 75.3949,
      "step": 52490
    },
    {
      "epoch": 0.2121066431800644,
      "grad_norm": 621.9034423828125,
      "learning_rate": 3.854367709133575e-05,
      "loss": 67.8432,
      "step": 52500
    },
    {
      "epoch": 0.21214704444543203,
      "grad_norm": 966.6736450195312,
      "learning_rate": 3.854263081162712e-05,
      "loss": 94.2898,
      "step": 52510
    },
    {
      "epoch": 0.21218744571079967,
      "grad_norm": 649.41162109375,
      "learning_rate": 3.854158417042035e-05,
      "loss": 80.9697,
      "step": 52520
    },
    {
      "epoch": 0.21222784697616728,
      "grad_norm": 616.9508666992188,
      "learning_rate": 3.854053716773586e-05,
      "loss": 43.8091,
      "step": 52530
    },
    {
      "epoch": 0.21226824824153492,
      "grad_norm": 714.7644653320312,
      "learning_rate": 3.8539489803594044e-05,
      "loss": 62.8353,
      "step": 52540
    },
    {
      "epoch": 0.21230864950690256,
      "grad_norm": 2239.86865234375,
      "learning_rate": 3.853844207801533e-05,
      "loss": 59.3144,
      "step": 52550
    },
    {
      "epoch": 0.21234905077227018,
      "grad_norm": 999.2901611328125,
      "learning_rate": 3.853739399102014e-05,
      "loss": 67.6813,
      "step": 52560
    },
    {
      "epoch": 0.21238945203763782,
      "grad_norm": 983.7887573242188,
      "learning_rate": 3.8536345542628925e-05,
      "loss": 72.0771,
      "step": 52570
    },
    {
      "epoch": 0.21242985330300546,
      "grad_norm": 2161.25634765625,
      "learning_rate": 3.85352967328621e-05,
      "loss": 97.4594,
      "step": 52580
    },
    {
      "epoch": 0.21247025456837307,
      "grad_norm": 323.06915283203125,
      "learning_rate": 3.8534247561740124e-05,
      "loss": 79.5232,
      "step": 52590
    },
    {
      "epoch": 0.2125106558337407,
      "grad_norm": 605.979248046875,
      "learning_rate": 3.853319802928345e-05,
      "loss": 87.567,
      "step": 52600
    },
    {
      "epoch": 0.21255105709910835,
      "grad_norm": 1341.1494140625,
      "learning_rate": 3.853214813551254e-05,
      "loss": 83.6259,
      "step": 52610
    },
    {
      "epoch": 0.212591458364476,
      "grad_norm": 652.4749755859375,
      "learning_rate": 3.8531097880447854e-05,
      "loss": 61.8694,
      "step": 52620
    },
    {
      "epoch": 0.2126318596298436,
      "grad_norm": 1503.4381103515625,
      "learning_rate": 3.853004726410988e-05,
      "loss": 101.6983,
      "step": 52630
    },
    {
      "epoch": 0.21267226089521124,
      "grad_norm": 1674.4990234375,
      "learning_rate": 3.852899628651909e-05,
      "loss": 95.5688,
      "step": 52640
    },
    {
      "epoch": 0.21271266216057888,
      "grad_norm": 196.89166259765625,
      "learning_rate": 3.852794494769599e-05,
      "loss": 41.1336,
      "step": 52650
    },
    {
      "epoch": 0.2127530634259465,
      "grad_norm": 1766.521240234375,
      "learning_rate": 3.8526893247661056e-05,
      "loss": 82.6849,
      "step": 52660
    },
    {
      "epoch": 0.21279346469131413,
      "grad_norm": 744.296142578125,
      "learning_rate": 3.85258411864348e-05,
      "loss": 101.108,
      "step": 52670
    },
    {
      "epoch": 0.21283386595668177,
      "grad_norm": 627.980712890625,
      "learning_rate": 3.852478876403773e-05,
      "loss": 103.0711,
      "step": 52680
    },
    {
      "epoch": 0.21287426722204938,
      "grad_norm": 1504.383056640625,
      "learning_rate": 3.8523735980490375e-05,
      "loss": 93.4536,
      "step": 52690
    },
    {
      "epoch": 0.21291466848741702,
      "grad_norm": 1216.134033203125,
      "learning_rate": 3.8522682835813244e-05,
      "loss": 71.4415,
      "step": 52700
    },
    {
      "epoch": 0.21295506975278466,
      "grad_norm": 537.6961059570312,
      "learning_rate": 3.852162933002687e-05,
      "loss": 61.1766,
      "step": 52710
    },
    {
      "epoch": 0.21299547101815228,
      "grad_norm": 490.43096923828125,
      "learning_rate": 3.8520575463151805e-05,
      "loss": 54.5532,
      "step": 52720
    },
    {
      "epoch": 0.21303587228351992,
      "grad_norm": 859.5098876953125,
      "learning_rate": 3.851952123520859e-05,
      "loss": 90.5423,
      "step": 52730
    },
    {
      "epoch": 0.21307627354888756,
      "grad_norm": 1086.8525390625,
      "learning_rate": 3.851846664621776e-05,
      "loss": 58.3408,
      "step": 52740
    },
    {
      "epoch": 0.21311667481425517,
      "grad_norm": 696.736328125,
      "learning_rate": 3.85174116961999e-05,
      "loss": 113.3828,
      "step": 52750
    },
    {
      "epoch": 0.2131570760796228,
      "grad_norm": 2076.949462890625,
      "learning_rate": 3.851635638517556e-05,
      "loss": 63.6123,
      "step": 52760
    },
    {
      "epoch": 0.21319747734499045,
      "grad_norm": 498.77569580078125,
      "learning_rate": 3.851530071316532e-05,
      "loss": 80.4989,
      "step": 52770
    },
    {
      "epoch": 0.2132378786103581,
      "grad_norm": 845.3126831054688,
      "learning_rate": 3.851424468018976e-05,
      "loss": 72.9964,
      "step": 52780
    },
    {
      "epoch": 0.2132782798757257,
      "grad_norm": 2801.711181640625,
      "learning_rate": 3.851318828626947e-05,
      "loss": 144.1869,
      "step": 52790
    },
    {
      "epoch": 0.21331868114109334,
      "grad_norm": 325.9803161621094,
      "learning_rate": 3.851213153142505e-05,
      "loss": 58.3193,
      "step": 52800
    },
    {
      "epoch": 0.21335908240646098,
      "grad_norm": 1257.360107421875,
      "learning_rate": 3.851107441567708e-05,
      "loss": 62.4607,
      "step": 52810
    },
    {
      "epoch": 0.2133994836718286,
      "grad_norm": 1192.9422607421875,
      "learning_rate": 3.85100169390462e-05,
      "loss": 77.8515,
      "step": 52820
    },
    {
      "epoch": 0.21343988493719623,
      "grad_norm": 911.9278564453125,
      "learning_rate": 3.8508959101553e-05,
      "loss": 87.0119,
      "step": 52830
    },
    {
      "epoch": 0.21348028620256387,
      "grad_norm": 663.7138671875,
      "learning_rate": 3.850790090321812e-05,
      "loss": 78.5091,
      "step": 52840
    },
    {
      "epoch": 0.21352068746793149,
      "grad_norm": 660.0030517578125,
      "learning_rate": 3.850684234406217e-05,
      "loss": 45.0362,
      "step": 52850
    },
    {
      "epoch": 0.21356108873329913,
      "grad_norm": 905.1046142578125,
      "learning_rate": 3.850578342410581e-05,
      "loss": 45.9832,
      "step": 52860
    },
    {
      "epoch": 0.21360148999866677,
      "grad_norm": 867.9483642578125,
      "learning_rate": 3.8504724143369674e-05,
      "loss": 53.6359,
      "step": 52870
    },
    {
      "epoch": 0.21364189126403438,
      "grad_norm": 476.4436340332031,
      "learning_rate": 3.850366450187442e-05,
      "loss": 91.5648,
      "step": 52880
    },
    {
      "epoch": 0.21368229252940202,
      "grad_norm": 839.8441162109375,
      "learning_rate": 3.850260449964069e-05,
      "loss": 85.5399,
      "step": 52890
    },
    {
      "epoch": 0.21372269379476966,
      "grad_norm": 611.8558349609375,
      "learning_rate": 3.850154413668916e-05,
      "loss": 73.8346,
      "step": 52900
    },
    {
      "epoch": 0.21376309506013727,
      "grad_norm": 688.4009399414062,
      "learning_rate": 3.8500483413040514e-05,
      "loss": 66.8995,
      "step": 52910
    },
    {
      "epoch": 0.2138034963255049,
      "grad_norm": 1049.5130615234375,
      "learning_rate": 3.849942232871541e-05,
      "loss": 104.3818,
      "step": 52920
    },
    {
      "epoch": 0.21384389759087255,
      "grad_norm": 533.0537109375,
      "learning_rate": 3.849836088373454e-05,
      "loss": 70.7031,
      "step": 52930
    },
    {
      "epoch": 0.2138842988562402,
      "grad_norm": 533.7456665039062,
      "learning_rate": 3.849729907811861e-05,
      "loss": 81.8711,
      "step": 52940
    },
    {
      "epoch": 0.2139247001216078,
      "grad_norm": 1017.4571533203125,
      "learning_rate": 3.8496236911888306e-05,
      "loss": 71.7599,
      "step": 52950
    },
    {
      "epoch": 0.21396510138697544,
      "grad_norm": 2414.6064453125,
      "learning_rate": 3.8495174385064335e-05,
      "loss": 107.8708,
      "step": 52960
    },
    {
      "epoch": 0.21400550265234308,
      "grad_norm": 415.21685791015625,
      "learning_rate": 3.849411149766743e-05,
      "loss": 87.1992,
      "step": 52970
    },
    {
      "epoch": 0.2140459039177107,
      "grad_norm": 1234.713134765625,
      "learning_rate": 3.849304824971829e-05,
      "loss": 62.1045,
      "step": 52980
    },
    {
      "epoch": 0.21408630518307833,
      "grad_norm": 842.6171264648438,
      "learning_rate": 3.849198464123766e-05,
      "loss": 113.1145,
      "step": 52990
    },
    {
      "epoch": 0.21412670644844597,
      "grad_norm": 841.9878540039062,
      "learning_rate": 3.849092067224627e-05,
      "loss": 56.3479,
      "step": 53000
    },
    {
      "epoch": 0.2141671077138136,
      "grad_norm": 1213.228271484375,
      "learning_rate": 3.848985634276486e-05,
      "loss": 87.3512,
      "step": 53010
    },
    {
      "epoch": 0.21420750897918123,
      "grad_norm": 1443.3297119140625,
      "learning_rate": 3.848879165281417e-05,
      "loss": 52.2816,
      "step": 53020
    },
    {
      "epoch": 0.21424791024454887,
      "grad_norm": 1061.485107421875,
      "learning_rate": 3.848772660241499e-05,
      "loss": 86.4943,
      "step": 53030
    },
    {
      "epoch": 0.21428831150991648,
      "grad_norm": 624.9254150390625,
      "learning_rate": 3.848666119158804e-05,
      "loss": 63.9003,
      "step": 53040
    },
    {
      "epoch": 0.21432871277528412,
      "grad_norm": 1313.3116455078125,
      "learning_rate": 3.848559542035412e-05,
      "loss": 69.6492,
      "step": 53050
    },
    {
      "epoch": 0.21436911404065176,
      "grad_norm": 1044.481689453125,
      "learning_rate": 3.848452928873401e-05,
      "loss": 79.7042,
      "step": 53060
    },
    {
      "epoch": 0.21440951530601937,
      "grad_norm": 989.2664184570312,
      "learning_rate": 3.848346279674847e-05,
      "loss": 105.1888,
      "step": 53070
    },
    {
      "epoch": 0.214449916571387,
      "grad_norm": 902.6455688476562,
      "learning_rate": 3.848239594441831e-05,
      "loss": 90.2427,
      "step": 53080
    },
    {
      "epoch": 0.21449031783675465,
      "grad_norm": 2429.779541015625,
      "learning_rate": 3.848132873176434e-05,
      "loss": 85.9649,
      "step": 53090
    },
    {
      "epoch": 0.2145307191021223,
      "grad_norm": 658.7740478515625,
      "learning_rate": 3.848026115880734e-05,
      "loss": 52.3369,
      "step": 53100
    },
    {
      "epoch": 0.2145711203674899,
      "grad_norm": 743.6819458007812,
      "learning_rate": 3.847919322556814e-05,
      "loss": 112.591,
      "step": 53110
    },
    {
      "epoch": 0.21461152163285754,
      "grad_norm": 0.0,
      "learning_rate": 3.847812493206754e-05,
      "loss": 59.0467,
      "step": 53120
    },
    {
      "epoch": 0.21465192289822518,
      "grad_norm": 4304.6552734375,
      "learning_rate": 3.84770562783264e-05,
      "loss": 78.0041,
      "step": 53130
    },
    {
      "epoch": 0.2146923241635928,
      "grad_norm": 735.5576782226562,
      "learning_rate": 3.847598726436553e-05,
      "loss": 63.3402,
      "step": 53140
    },
    {
      "epoch": 0.21473272542896044,
      "grad_norm": 766.4822998046875,
      "learning_rate": 3.847491789020576e-05,
      "loss": 149.0328,
      "step": 53150
    },
    {
      "epoch": 0.21477312669432808,
      "grad_norm": 682.111572265625,
      "learning_rate": 3.8473848155867976e-05,
      "loss": 84.1817,
      "step": 53160
    },
    {
      "epoch": 0.2148135279596957,
      "grad_norm": 791.5773315429688,
      "learning_rate": 3.8472778061373e-05,
      "loss": 74.4928,
      "step": 53170
    },
    {
      "epoch": 0.21485392922506333,
      "grad_norm": 898.5239868164062,
      "learning_rate": 3.8471707606741706e-05,
      "loss": 66.1085,
      "step": 53180
    },
    {
      "epoch": 0.21489433049043097,
      "grad_norm": 621.9130249023438,
      "learning_rate": 3.8470636791994965e-05,
      "loss": 75.2301,
      "step": 53190
    },
    {
      "epoch": 0.21493473175579858,
      "grad_norm": 1194.8358154296875,
      "learning_rate": 3.8469565617153646e-05,
      "loss": 95.7897,
      "step": 53200
    },
    {
      "epoch": 0.21497513302116622,
      "grad_norm": 3260.6416015625,
      "learning_rate": 3.846849408223865e-05,
      "loss": 61.8074,
      "step": 53210
    },
    {
      "epoch": 0.21501553428653386,
      "grad_norm": 1330.064208984375,
      "learning_rate": 3.846742218727084e-05,
      "loss": 80.6778,
      "step": 53220
    },
    {
      "epoch": 0.21505593555190147,
      "grad_norm": 768.6731567382812,
      "learning_rate": 3.846634993227113e-05,
      "loss": 98.0611,
      "step": 53230
    },
    {
      "epoch": 0.2150963368172691,
      "grad_norm": 726.4506225585938,
      "learning_rate": 3.846527731726042e-05,
      "loss": 56.8013,
      "step": 53240
    },
    {
      "epoch": 0.21513673808263675,
      "grad_norm": 595.484130859375,
      "learning_rate": 3.846420434225962e-05,
      "loss": 68.5684,
      "step": 53250
    },
    {
      "epoch": 0.2151771393480044,
      "grad_norm": 2421.9501953125,
      "learning_rate": 3.846313100728965e-05,
      "loss": 90.2925,
      "step": 53260
    },
    {
      "epoch": 0.215217540613372,
      "grad_norm": 1195.77880859375,
      "learning_rate": 3.846205731237144e-05,
      "loss": 55.6472,
      "step": 53270
    },
    {
      "epoch": 0.21525794187873964,
      "grad_norm": 556.8364868164062,
      "learning_rate": 3.846098325752592e-05,
      "loss": 67.6011,
      "step": 53280
    },
    {
      "epoch": 0.21529834314410728,
      "grad_norm": 688.4712524414062,
      "learning_rate": 3.845990884277402e-05,
      "loss": 79.6667,
      "step": 53290
    },
    {
      "epoch": 0.2153387444094749,
      "grad_norm": 906.529541015625,
      "learning_rate": 3.8458834068136704e-05,
      "loss": 70.9961,
      "step": 53300
    },
    {
      "epoch": 0.21537914567484254,
      "grad_norm": 5431.75146484375,
      "learning_rate": 3.8457758933634905e-05,
      "loss": 139.834,
      "step": 53310
    },
    {
      "epoch": 0.21541954694021018,
      "grad_norm": 556.4915771484375,
      "learning_rate": 3.84566834392896e-05,
      "loss": 69.6828,
      "step": 53320
    },
    {
      "epoch": 0.2154599482055778,
      "grad_norm": 783.346923828125,
      "learning_rate": 3.8455607585121745e-05,
      "loss": 91.7018,
      "step": 53330
    },
    {
      "epoch": 0.21550034947094543,
      "grad_norm": 444.45556640625,
      "learning_rate": 3.8454531371152315e-05,
      "loss": 64.6532,
      "step": 53340
    },
    {
      "epoch": 0.21554075073631307,
      "grad_norm": 1354.439453125,
      "learning_rate": 3.84534547974023e-05,
      "loss": 92.3306,
      "step": 53350
    },
    {
      "epoch": 0.21558115200168068,
      "grad_norm": 593.5985717773438,
      "learning_rate": 3.8452377863892686e-05,
      "loss": 63.3151,
      "step": 53360
    },
    {
      "epoch": 0.21562155326704832,
      "grad_norm": 868.5457763671875,
      "learning_rate": 3.845130057064446e-05,
      "loss": 64.5144,
      "step": 53370
    },
    {
      "epoch": 0.21566195453241596,
      "grad_norm": 708.3560180664062,
      "learning_rate": 3.8450222917678626e-05,
      "loss": 77.0622,
      "step": 53380
    },
    {
      "epoch": 0.21570235579778357,
      "grad_norm": 1146.61376953125,
      "learning_rate": 3.84491449050162e-05,
      "loss": 96.9242,
      "step": 53390
    },
    {
      "epoch": 0.2157427570631512,
      "grad_norm": 501.0816345214844,
      "learning_rate": 3.8448066532678206e-05,
      "loss": 72.2287,
      "step": 53400
    },
    {
      "epoch": 0.21578315832851885,
      "grad_norm": 1105.1387939453125,
      "learning_rate": 3.8446987800685646e-05,
      "loss": 120.5814,
      "step": 53410
    },
    {
      "epoch": 0.2158235595938865,
      "grad_norm": 696.0546875,
      "learning_rate": 3.844590870905957e-05,
      "loss": 85.6093,
      "step": 53420
    },
    {
      "epoch": 0.2158639608592541,
      "grad_norm": 1029.6751708984375,
      "learning_rate": 3.8444829257821e-05,
      "loss": 106.5893,
      "step": 53430
    },
    {
      "epoch": 0.21590436212462175,
      "grad_norm": 1049.6455078125,
      "learning_rate": 3.844374944699099e-05,
      "loss": 57.9386,
      "step": 53440
    },
    {
      "epoch": 0.21594476338998939,
      "grad_norm": 322.9332580566406,
      "learning_rate": 3.844266927659059e-05,
      "loss": 80.7071,
      "step": 53450
    },
    {
      "epoch": 0.215985164655357,
      "grad_norm": 649.258056640625,
      "learning_rate": 3.8441588746640856e-05,
      "loss": 60.9057,
      "step": 53460
    },
    {
      "epoch": 0.21602556592072464,
      "grad_norm": 703.34326171875,
      "learning_rate": 3.8440507857162856e-05,
      "loss": 65.4921,
      "step": 53470
    },
    {
      "epoch": 0.21606596718609228,
      "grad_norm": 451.3043518066406,
      "learning_rate": 3.843942660817767e-05,
      "loss": 73.565,
      "step": 53480
    },
    {
      "epoch": 0.2161063684514599,
      "grad_norm": 1561.3681640625,
      "learning_rate": 3.8438344999706355e-05,
      "loss": 77.2143,
      "step": 53490
    },
    {
      "epoch": 0.21614676971682753,
      "grad_norm": 613.5264282226562,
      "learning_rate": 3.8437263031770015e-05,
      "loss": 67.6038,
      "step": 53500
    },
    {
      "epoch": 0.21618717098219517,
      "grad_norm": 580.0654296875,
      "learning_rate": 3.843618070438974e-05,
      "loss": 80.3659,
      "step": 53510
    },
    {
      "epoch": 0.21622757224756278,
      "grad_norm": 411.0973205566406,
      "learning_rate": 3.8435098017586635e-05,
      "loss": 57.7491,
      "step": 53520
    },
    {
      "epoch": 0.21626797351293042,
      "grad_norm": 515.7644653320312,
      "learning_rate": 3.84340149713818e-05,
      "loss": 96.419,
      "step": 53530
    },
    {
      "epoch": 0.21630837477829806,
      "grad_norm": 1984.08740234375,
      "learning_rate": 3.843293156579635e-05,
      "loss": 80.8797,
      "step": 53540
    },
    {
      "epoch": 0.21634877604366567,
      "grad_norm": 536.7109375,
      "learning_rate": 3.843184780085142e-05,
      "loss": 88.4045,
      "step": 53550
    },
    {
      "epoch": 0.21638917730903331,
      "grad_norm": 916.5641479492188,
      "learning_rate": 3.843076367656811e-05,
      "loss": 85.3823,
      "step": 53560
    },
    {
      "epoch": 0.21642957857440095,
      "grad_norm": 573.3203735351562,
      "learning_rate": 3.842967919296759e-05,
      "loss": 62.3376,
      "step": 53570
    },
    {
      "epoch": 0.2164699798397686,
      "grad_norm": 717.7938842773438,
      "learning_rate": 3.842859435007098e-05,
      "loss": 63.3645,
      "step": 53580
    },
    {
      "epoch": 0.2165103811051362,
      "grad_norm": 641.4676513671875,
      "learning_rate": 3.8427509147899436e-05,
      "loss": 81.8669,
      "step": 53590
    },
    {
      "epoch": 0.21655078237050385,
      "grad_norm": 790.5074462890625,
      "learning_rate": 3.842642358647411e-05,
      "loss": 67.9639,
      "step": 53600
    },
    {
      "epoch": 0.2165911836358715,
      "grad_norm": 1648.721435546875,
      "learning_rate": 3.842533766581618e-05,
      "loss": 83.7934,
      "step": 53610
    },
    {
      "epoch": 0.2166315849012391,
      "grad_norm": 722.0618896484375,
      "learning_rate": 3.842425138594679e-05,
      "loss": 64.1934,
      "step": 53620
    },
    {
      "epoch": 0.21667198616660674,
      "grad_norm": 702.053955078125,
      "learning_rate": 3.842316474688715e-05,
      "loss": 174.5111,
      "step": 53630
    },
    {
      "epoch": 0.21671238743197438,
      "grad_norm": 868.510986328125,
      "learning_rate": 3.8422077748658415e-05,
      "loss": 55.1778,
      "step": 53640
    },
    {
      "epoch": 0.216752788697342,
      "grad_norm": 644.6915893554688,
      "learning_rate": 3.842099039128179e-05,
      "loss": 74.3898,
      "step": 53650
    },
    {
      "epoch": 0.21679318996270963,
      "grad_norm": 1687.4537353515625,
      "learning_rate": 3.8419902674778484e-05,
      "loss": 79.454,
      "step": 53660
    },
    {
      "epoch": 0.21683359122807727,
      "grad_norm": 581.4046020507812,
      "learning_rate": 3.8418814599169684e-05,
      "loss": 80.527,
      "step": 53670
    },
    {
      "epoch": 0.21687399249344488,
      "grad_norm": 0.0,
      "learning_rate": 3.841772616447662e-05,
      "loss": 59.3345,
      "step": 53680
    },
    {
      "epoch": 0.21691439375881252,
      "grad_norm": 747.6123046875,
      "learning_rate": 3.84166373707205e-05,
      "loss": 64.1358,
      "step": 53690
    },
    {
      "epoch": 0.21695479502418016,
      "grad_norm": 893.1629638671875,
      "learning_rate": 3.8415548217922546e-05,
      "loss": 96.7892,
      "step": 53700
    },
    {
      "epoch": 0.21699519628954778,
      "grad_norm": 815.9344482421875,
      "learning_rate": 3.8414458706104e-05,
      "loss": 76.8955,
      "step": 53710
    },
    {
      "epoch": 0.21703559755491542,
      "grad_norm": 977.8809204101562,
      "learning_rate": 3.8413368835286104e-05,
      "loss": 110.5061,
      "step": 53720
    },
    {
      "epoch": 0.21707599882028306,
      "grad_norm": 925.52392578125,
      "learning_rate": 3.84122786054901e-05,
      "loss": 98.8299,
      "step": 53730
    },
    {
      "epoch": 0.2171164000856507,
      "grad_norm": 865.4520263671875,
      "learning_rate": 3.8411188016737245e-05,
      "loss": 79.2513,
      "step": 53740
    },
    {
      "epoch": 0.2171568013510183,
      "grad_norm": 981.2018432617188,
      "learning_rate": 3.841009706904881e-05,
      "loss": 65.9173,
      "step": 53750
    },
    {
      "epoch": 0.21719720261638595,
      "grad_norm": 531.883056640625,
      "learning_rate": 3.8409005762446046e-05,
      "loss": 53.9927,
      "step": 53760
    },
    {
      "epoch": 0.2172376038817536,
      "grad_norm": 1539.730712890625,
      "learning_rate": 3.840791409695024e-05,
      "loss": 87.3034,
      "step": 53770
    },
    {
      "epoch": 0.2172780051471212,
      "grad_norm": 846.1365356445312,
      "learning_rate": 3.840682207258267e-05,
      "loss": 57.685,
      "step": 53780
    },
    {
      "epoch": 0.21731840641248884,
      "grad_norm": 906.4349365234375,
      "learning_rate": 3.8405729689364625e-05,
      "loss": 94.1517,
      "step": 53790
    },
    {
      "epoch": 0.21735880767785648,
      "grad_norm": 1619.25634765625,
      "learning_rate": 3.840463694731741e-05,
      "loss": 102.3085,
      "step": 53800
    },
    {
      "epoch": 0.2173992089432241,
      "grad_norm": 700.124267578125,
      "learning_rate": 3.840354384646232e-05,
      "loss": 81.1631,
      "step": 53810
    },
    {
      "epoch": 0.21743961020859173,
      "grad_norm": 759.2460327148438,
      "learning_rate": 3.840245038682067e-05,
      "loss": 63.1743,
      "step": 53820
    },
    {
      "epoch": 0.21748001147395937,
      "grad_norm": 474.08856201171875,
      "learning_rate": 3.8401356568413766e-05,
      "loss": 86.7374,
      "step": 53830
    },
    {
      "epoch": 0.21752041273932698,
      "grad_norm": 1223.6085205078125,
      "learning_rate": 3.840026239126295e-05,
      "loss": 98.9601,
      "step": 53840
    },
    {
      "epoch": 0.21756081400469462,
      "grad_norm": 701.7997436523438,
      "learning_rate": 3.839916785538955e-05,
      "loss": 113.6616,
      "step": 53850
    },
    {
      "epoch": 0.21760121527006226,
      "grad_norm": 535.2505493164062,
      "learning_rate": 3.839807296081489e-05,
      "loss": 53.6258,
      "step": 53860
    },
    {
      "epoch": 0.21764161653542988,
      "grad_norm": 498.7857360839844,
      "learning_rate": 3.839697770756033e-05,
      "loss": 45.3512,
      "step": 53870
    },
    {
      "epoch": 0.21768201780079752,
      "grad_norm": 1155.1322021484375,
      "learning_rate": 3.839588209564721e-05,
      "loss": 91.6865,
      "step": 53880
    },
    {
      "epoch": 0.21772241906616516,
      "grad_norm": 509.37603759765625,
      "learning_rate": 3.839478612509691e-05,
      "loss": 68.8964,
      "step": 53890
    },
    {
      "epoch": 0.2177628203315328,
      "grad_norm": 986.0476684570312,
      "learning_rate": 3.839368979593079e-05,
      "loss": 77.7923,
      "step": 53900
    },
    {
      "epoch": 0.2178032215969004,
      "grad_norm": 503.35791015625,
      "learning_rate": 3.83925931081702e-05,
      "loss": 45.5724,
      "step": 53910
    },
    {
      "epoch": 0.21784362286226805,
      "grad_norm": 612.2771606445312,
      "learning_rate": 3.8391496061836545e-05,
      "loss": 64.179,
      "step": 53920
    },
    {
      "epoch": 0.2178840241276357,
      "grad_norm": 693.4111938476562,
      "learning_rate": 3.8390398656951204e-05,
      "loss": 59.0524,
      "step": 53930
    },
    {
      "epoch": 0.2179244253930033,
      "grad_norm": 3350.630126953125,
      "learning_rate": 3.838930089353558e-05,
      "loss": 124.8092,
      "step": 53940
    },
    {
      "epoch": 0.21796482665837094,
      "grad_norm": 388.133056640625,
      "learning_rate": 3.838820277161106e-05,
      "loss": 57.8169,
      "step": 53950
    },
    {
      "epoch": 0.21800522792373858,
      "grad_norm": 512.2550659179688,
      "learning_rate": 3.8387104291199066e-05,
      "loss": 58.9273,
      "step": 53960
    },
    {
      "epoch": 0.2180456291891062,
      "grad_norm": 429.9571228027344,
      "learning_rate": 3.8386005452321e-05,
      "loss": 67.8202,
      "step": 53970
    },
    {
      "epoch": 0.21808603045447383,
      "grad_norm": 1184.2342529296875,
      "learning_rate": 3.83849062549983e-05,
      "loss": 95.5747,
      "step": 53980
    },
    {
      "epoch": 0.21812643171984147,
      "grad_norm": 1229.4407958984375,
      "learning_rate": 3.838380669925238e-05,
      "loss": 64.7428,
      "step": 53990
    },
    {
      "epoch": 0.21816683298520909,
      "grad_norm": 926.0938110351562,
      "learning_rate": 3.838270678510469e-05,
      "loss": 56.7303,
      "step": 54000
    },
    {
      "epoch": 0.21820723425057673,
      "grad_norm": 776.5170288085938,
      "learning_rate": 3.8381606512576664e-05,
      "loss": 67.6247,
      "step": 54010
    },
    {
      "epoch": 0.21824763551594437,
      "grad_norm": 756.7637939453125,
      "learning_rate": 3.838050588168976e-05,
      "loss": 74.2062,
      "step": 54020
    },
    {
      "epoch": 0.21828803678131198,
      "grad_norm": 1093.98291015625,
      "learning_rate": 3.837940489246543e-05,
      "loss": 67.1495,
      "step": 54030
    },
    {
      "epoch": 0.21832843804667962,
      "grad_norm": 1240.094970703125,
      "learning_rate": 3.8378303544925134e-05,
      "loss": 61.8741,
      "step": 54040
    },
    {
      "epoch": 0.21836883931204726,
      "grad_norm": 696.2560424804688,
      "learning_rate": 3.837720183909035e-05,
      "loss": 103.6368,
      "step": 54050
    },
    {
      "epoch": 0.2184092405774149,
      "grad_norm": 632.3446044921875,
      "learning_rate": 3.8376099774982553e-05,
      "loss": 80.5212,
      "step": 54060
    },
    {
      "epoch": 0.2184496418427825,
      "grad_norm": 665.3270874023438,
      "learning_rate": 3.8374997352623234e-05,
      "loss": 55.7393,
      "step": 54070
    },
    {
      "epoch": 0.21849004310815015,
      "grad_norm": 1013.0740966796875,
      "learning_rate": 3.837389457203388e-05,
      "loss": 78.6451,
      "step": 54080
    },
    {
      "epoch": 0.2185304443735178,
      "grad_norm": 814.3826293945312,
      "learning_rate": 3.8372791433236e-05,
      "loss": 87.6565,
      "step": 54090
    },
    {
      "epoch": 0.2185708456388854,
      "grad_norm": 1004.903564453125,
      "learning_rate": 3.8371687936251085e-05,
      "loss": 74.4121,
      "step": 54100
    },
    {
      "epoch": 0.21861124690425304,
      "grad_norm": 609.3934326171875,
      "learning_rate": 3.837058408110066e-05,
      "loss": 92.1348,
      "step": 54110
    },
    {
      "epoch": 0.21865164816962068,
      "grad_norm": 498.446044921875,
      "learning_rate": 3.836947986780623e-05,
      "loss": 140.8018,
      "step": 54120
    },
    {
      "epoch": 0.2186920494349883,
      "grad_norm": 1687.71435546875,
      "learning_rate": 3.836837529638934e-05,
      "loss": 105.1114,
      "step": 54130
    },
    {
      "epoch": 0.21873245070035593,
      "grad_norm": 759.4395751953125,
      "learning_rate": 3.836727036687151e-05,
      "loss": 73.0021,
      "step": 54140
    },
    {
      "epoch": 0.21877285196572357,
      "grad_norm": 866.75341796875,
      "learning_rate": 3.83661650792743e-05,
      "loss": 78.3296,
      "step": 54150
    },
    {
      "epoch": 0.2188132532310912,
      "grad_norm": 1064.7254638671875,
      "learning_rate": 3.8365059433619236e-05,
      "loss": 70.7086,
      "step": 54160
    },
    {
      "epoch": 0.21885365449645883,
      "grad_norm": 480.29156494140625,
      "learning_rate": 3.836395342992789e-05,
      "loss": 52.3391,
      "step": 54170
    },
    {
      "epoch": 0.21889405576182647,
      "grad_norm": 465.5809326171875,
      "learning_rate": 3.8362847068221816e-05,
      "loss": 77.8524,
      "step": 54180
    },
    {
      "epoch": 0.21893445702719408,
      "grad_norm": 853.2993774414062,
      "learning_rate": 3.836174034852258e-05,
      "loss": 79.3692,
      "step": 54190
    },
    {
      "epoch": 0.21897485829256172,
      "grad_norm": 781.8079223632812,
      "learning_rate": 3.8360633270851765e-05,
      "loss": 64.2943,
      "step": 54200
    },
    {
      "epoch": 0.21901525955792936,
      "grad_norm": 704.2810668945312,
      "learning_rate": 3.835952583523095e-05,
      "loss": 76.673,
      "step": 54210
    },
    {
      "epoch": 0.219055660823297,
      "grad_norm": 1497.0550537109375,
      "learning_rate": 3.835841804168174e-05,
      "loss": 75.4406,
      "step": 54220
    },
    {
      "epoch": 0.2190960620886646,
      "grad_norm": 1838.40869140625,
      "learning_rate": 3.8357309890225696e-05,
      "loss": 102.6016,
      "step": 54230
    },
    {
      "epoch": 0.21913646335403225,
      "grad_norm": 469.5345153808594,
      "learning_rate": 3.8356201380884465e-05,
      "loss": 63.3835,
      "step": 54240
    },
    {
      "epoch": 0.2191768646193999,
      "grad_norm": 1066.0787353515625,
      "learning_rate": 3.835509251367963e-05,
      "loss": 77.8217,
      "step": 54250
    },
    {
      "epoch": 0.2192172658847675,
      "grad_norm": 1482.1439208984375,
      "learning_rate": 3.835398328863281e-05,
      "loss": 77.4048,
      "step": 54260
    },
    {
      "epoch": 0.21925766715013514,
      "grad_norm": 811.32080078125,
      "learning_rate": 3.835287370576564e-05,
      "loss": 76.4093,
      "step": 54270
    },
    {
      "epoch": 0.21929806841550278,
      "grad_norm": 643.56787109375,
      "learning_rate": 3.8351763765099755e-05,
      "loss": 101.0972,
      "step": 54280
    },
    {
      "epoch": 0.2193384696808704,
      "grad_norm": 552.455810546875,
      "learning_rate": 3.835065346665679e-05,
      "loss": 77.881,
      "step": 54290
    },
    {
      "epoch": 0.21937887094623804,
      "grad_norm": 1201.8634033203125,
      "learning_rate": 3.8349542810458374e-05,
      "loss": 65.2546,
      "step": 54300
    },
    {
      "epoch": 0.21941927221160568,
      "grad_norm": 160.37318420410156,
      "learning_rate": 3.834843179652618e-05,
      "loss": 63.4198,
      "step": 54310
    },
    {
      "epoch": 0.2194596734769733,
      "grad_norm": 1110.407470703125,
      "learning_rate": 3.834732042488186e-05,
      "loss": 56.3331,
      "step": 54320
    },
    {
      "epoch": 0.21950007474234093,
      "grad_norm": 863.4337768554688,
      "learning_rate": 3.834620869554709e-05,
      "loss": 66.6363,
      "step": 54330
    },
    {
      "epoch": 0.21954047600770857,
      "grad_norm": 1122.90576171875,
      "learning_rate": 3.834509660854353e-05,
      "loss": 68.35,
      "step": 54340
    },
    {
      "epoch": 0.21958087727307618,
      "grad_norm": 1227.4989013671875,
      "learning_rate": 3.834398416389287e-05,
      "loss": 75.4078,
      "step": 54350
    },
    {
      "epoch": 0.21962127853844382,
      "grad_norm": 855.5853271484375,
      "learning_rate": 3.834287136161679e-05,
      "loss": 87.4391,
      "step": 54360
    },
    {
      "epoch": 0.21966167980381146,
      "grad_norm": 1111.5548095703125,
      "learning_rate": 3.8341758201736995e-05,
      "loss": 82.4853,
      "step": 54370
    },
    {
      "epoch": 0.2197020810691791,
      "grad_norm": 639.7208862304688,
      "learning_rate": 3.8340644684275184e-05,
      "loss": 76.4388,
      "step": 54380
    },
    {
      "epoch": 0.2197424823345467,
      "grad_norm": 661.3689575195312,
      "learning_rate": 3.833953080925305e-05,
      "loss": 43.4275,
      "step": 54390
    },
    {
      "epoch": 0.21978288359991435,
      "grad_norm": 515.0191040039062,
      "learning_rate": 3.8338416576692335e-05,
      "loss": 58.5762,
      "step": 54400
    },
    {
      "epoch": 0.219823284865282,
      "grad_norm": 2175.806396484375,
      "learning_rate": 3.833730198661473e-05,
      "loss": 102.4807,
      "step": 54410
    },
    {
      "epoch": 0.2198636861306496,
      "grad_norm": 1177.308837890625,
      "learning_rate": 3.8336187039042e-05,
      "loss": 103.9685,
      "step": 54420
    },
    {
      "epoch": 0.21990408739601724,
      "grad_norm": 2354.74072265625,
      "learning_rate": 3.833507173399586e-05,
      "loss": 83.4062,
      "step": 54430
    },
    {
      "epoch": 0.21994448866138488,
      "grad_norm": 483.1558532714844,
      "learning_rate": 3.8333956071498047e-05,
      "loss": 81.7757,
      "step": 54440
    },
    {
      "epoch": 0.2199848899267525,
      "grad_norm": 1119.9464111328125,
      "learning_rate": 3.833284005157033e-05,
      "loss": 62.2444,
      "step": 54450
    },
    {
      "epoch": 0.22002529119212014,
      "grad_norm": 1096.0882568359375,
      "learning_rate": 3.833172367423445e-05,
      "loss": 76.2725,
      "step": 54460
    },
    {
      "epoch": 0.22006569245748778,
      "grad_norm": 853.37646484375,
      "learning_rate": 3.833060693951219e-05,
      "loss": 60.5453,
      "step": 54470
    },
    {
      "epoch": 0.2201060937228554,
      "grad_norm": 874.2672729492188,
      "learning_rate": 3.83294898474253e-05,
      "loss": 86.5318,
      "step": 54480
    },
    {
      "epoch": 0.22014649498822303,
      "grad_norm": 2054.371826171875,
      "learning_rate": 3.832837239799558e-05,
      "loss": 135.8652,
      "step": 54490
    },
    {
      "epoch": 0.22018689625359067,
      "grad_norm": 506.741943359375,
      "learning_rate": 3.8327254591244795e-05,
      "loss": 68.7912,
      "step": 54500
    },
    {
      "epoch": 0.22022729751895828,
      "grad_norm": 524.73193359375,
      "learning_rate": 3.832613642719475e-05,
      "loss": 36.3728,
      "step": 54510
    },
    {
      "epoch": 0.22026769878432592,
      "grad_norm": 1817.7291259765625,
      "learning_rate": 3.832501790586724e-05,
      "loss": 54.225,
      "step": 54520
    },
    {
      "epoch": 0.22030810004969356,
      "grad_norm": 1189.2894287109375,
      "learning_rate": 3.832389902728407e-05,
      "loss": 100.2454,
      "step": 54530
    },
    {
      "epoch": 0.2203485013150612,
      "grad_norm": 280.9716491699219,
      "learning_rate": 3.832277979146706e-05,
      "loss": 47.7974,
      "step": 54540
    },
    {
      "epoch": 0.2203889025804288,
      "grad_norm": 985.0027465820312,
      "learning_rate": 3.8321660198438013e-05,
      "loss": 94.2698,
      "step": 54550
    },
    {
      "epoch": 0.22042930384579645,
      "grad_norm": 1097.96875,
      "learning_rate": 3.832054024821877e-05,
      "loss": 62.8399,
      "step": 54560
    },
    {
      "epoch": 0.2204697051111641,
      "grad_norm": 904.7685546875,
      "learning_rate": 3.831941994083118e-05,
      "loss": 62.1448,
      "step": 54570
    },
    {
      "epoch": 0.2205101063765317,
      "grad_norm": 569.3748779296875,
      "learning_rate": 3.831829927629705e-05,
      "loss": 47.2207,
      "step": 54580
    },
    {
      "epoch": 0.22055050764189935,
      "grad_norm": 888.6369018554688,
      "learning_rate": 3.831717825463825e-05,
      "loss": 117.8203,
      "step": 54590
    },
    {
      "epoch": 0.22059090890726699,
      "grad_norm": 813.579833984375,
      "learning_rate": 3.831605687587663e-05,
      "loss": 107.6004,
      "step": 54600
    },
    {
      "epoch": 0.2206313101726346,
      "grad_norm": 686.66845703125,
      "learning_rate": 3.831493514003405e-05,
      "loss": 53.9199,
      "step": 54610
    },
    {
      "epoch": 0.22067171143800224,
      "grad_norm": 1741.1114501953125,
      "learning_rate": 3.8313813047132384e-05,
      "loss": 61.756,
      "step": 54620
    },
    {
      "epoch": 0.22071211270336988,
      "grad_norm": 1898.2445068359375,
      "learning_rate": 3.8312690597193494e-05,
      "loss": 75.3093,
      "step": 54630
    },
    {
      "epoch": 0.2207525139687375,
      "grad_norm": 1430.97509765625,
      "learning_rate": 3.8311567790239284e-05,
      "loss": 78.689,
      "step": 54640
    },
    {
      "epoch": 0.22079291523410513,
      "grad_norm": 453.10400390625,
      "learning_rate": 3.831044462629163e-05,
      "loss": 72.2116,
      "step": 54650
    },
    {
      "epoch": 0.22083331649947277,
      "grad_norm": 977.4793090820312,
      "learning_rate": 3.830932110537242e-05,
      "loss": 82.8747,
      "step": 54660
    },
    {
      "epoch": 0.22087371776484038,
      "grad_norm": 523.4397583007812,
      "learning_rate": 3.830819722750358e-05,
      "loss": 127.7064,
      "step": 54670
    },
    {
      "epoch": 0.22091411903020802,
      "grad_norm": 1129.9034423828125,
      "learning_rate": 3.830707299270701e-05,
      "loss": 69.922,
      "step": 54680
    },
    {
      "epoch": 0.22095452029557566,
      "grad_norm": 657.69580078125,
      "learning_rate": 3.830594840100463e-05,
      "loss": 63.3711,
      "step": 54690
    },
    {
      "epoch": 0.22099492156094327,
      "grad_norm": 1762.107177734375,
      "learning_rate": 3.830482345241835e-05,
      "loss": 61.1276,
      "step": 54700
    },
    {
      "epoch": 0.22103532282631091,
      "grad_norm": 1589.8001708984375,
      "learning_rate": 3.830369814697013e-05,
      "loss": 99.1144,
      "step": 54710
    },
    {
      "epoch": 0.22107572409167855,
      "grad_norm": 764.2601928710938,
      "learning_rate": 3.830257248468187e-05,
      "loss": 72.5749,
      "step": 54720
    },
    {
      "epoch": 0.2211161253570462,
      "grad_norm": 1804.06640625,
      "learning_rate": 3.8301446465575554e-05,
      "loss": 85.9519,
      "step": 54730
    },
    {
      "epoch": 0.2211565266224138,
      "grad_norm": 561.2105102539062,
      "learning_rate": 3.8300320089673106e-05,
      "loss": 57.1239,
      "step": 54740
    },
    {
      "epoch": 0.22119692788778145,
      "grad_norm": 545.8154907226562,
      "learning_rate": 3.82991933569965e-05,
      "loss": 67.4139,
      "step": 54750
    },
    {
      "epoch": 0.2212373291531491,
      "grad_norm": 1868.6494140625,
      "learning_rate": 3.82980662675677e-05,
      "loss": 115.0834,
      "step": 54760
    },
    {
      "epoch": 0.2212777304185167,
      "grad_norm": 856.7229614257812,
      "learning_rate": 3.829693882140867e-05,
      "loss": 84.5909,
      "step": 54770
    },
    {
      "epoch": 0.22131813168388434,
      "grad_norm": 547.6751098632812,
      "learning_rate": 3.8295811018541406e-05,
      "loss": 54.9399,
      "step": 54780
    },
    {
      "epoch": 0.22135853294925198,
      "grad_norm": 805.74072265625,
      "learning_rate": 3.829468285898789e-05,
      "loss": 91.6851,
      "step": 54790
    },
    {
      "epoch": 0.2213989342146196,
      "grad_norm": 781.6884155273438,
      "learning_rate": 3.82935543427701e-05,
      "loss": 66.7251,
      "step": 54800
    },
    {
      "epoch": 0.22143933547998723,
      "grad_norm": 1233.2275390625,
      "learning_rate": 3.829242546991006e-05,
      "loss": 81.143,
      "step": 54810
    },
    {
      "epoch": 0.22147973674535487,
      "grad_norm": 951.1732177734375,
      "learning_rate": 3.8291296240429766e-05,
      "loss": 63.9419,
      "step": 54820
    },
    {
      "epoch": 0.22152013801072248,
      "grad_norm": 814.5828857421875,
      "learning_rate": 3.8290166654351236e-05,
      "loss": 63.3376,
      "step": 54830
    },
    {
      "epoch": 0.22156053927609012,
      "grad_norm": 456.1623840332031,
      "learning_rate": 3.828903671169649e-05,
      "loss": 84.4635,
      "step": 54840
    },
    {
      "epoch": 0.22160094054145776,
      "grad_norm": 1096.0460205078125,
      "learning_rate": 3.8287906412487555e-05,
      "loss": 104.485,
      "step": 54850
    },
    {
      "epoch": 0.22164134180682538,
      "grad_norm": 625.3572998046875,
      "learning_rate": 3.828677575674647e-05,
      "loss": 65.4626,
      "step": 54860
    },
    {
      "epoch": 0.22168174307219302,
      "grad_norm": 1885.492919921875,
      "learning_rate": 3.828564474449527e-05,
      "loss": 68.5083,
      "step": 54870
    },
    {
      "epoch": 0.22172214433756066,
      "grad_norm": 558.2794799804688,
      "learning_rate": 3.8284513375756024e-05,
      "loss": 61.8752,
      "step": 54880
    },
    {
      "epoch": 0.2217625456029283,
      "grad_norm": 529.7760009765625,
      "learning_rate": 3.828338165055077e-05,
      "loss": 86.1901,
      "step": 54890
    },
    {
      "epoch": 0.2218029468682959,
      "grad_norm": 726.2923583984375,
      "learning_rate": 3.828224956890158e-05,
      "loss": 66.4456,
      "step": 54900
    },
    {
      "epoch": 0.22184334813366355,
      "grad_norm": 1140.76416015625,
      "learning_rate": 3.828111713083052e-05,
      "loss": 68.4928,
      "step": 54910
    },
    {
      "epoch": 0.2218837493990312,
      "grad_norm": 774.8322143554688,
      "learning_rate": 3.827998433635967e-05,
      "loss": 63.8003,
      "step": 54920
    },
    {
      "epoch": 0.2219241506643988,
      "grad_norm": 929.5682373046875,
      "learning_rate": 3.8278851185511115e-05,
      "loss": 112.3734,
      "step": 54930
    },
    {
      "epoch": 0.22196455192976644,
      "grad_norm": 825.6917724609375,
      "learning_rate": 3.8277717678306946e-05,
      "loss": 94.8121,
      "step": 54940
    },
    {
      "epoch": 0.22200495319513408,
      "grad_norm": 762.4610595703125,
      "learning_rate": 3.8276583814769265e-05,
      "loss": 55.5008,
      "step": 54950
    },
    {
      "epoch": 0.2220453544605017,
      "grad_norm": 577.0433959960938,
      "learning_rate": 3.827544959492017e-05,
      "loss": 80.0855,
      "step": 54960
    },
    {
      "epoch": 0.22208575572586933,
      "grad_norm": 1125.5518798828125,
      "learning_rate": 3.827431501878177e-05,
      "loss": 85.5211,
      "step": 54970
    },
    {
      "epoch": 0.22212615699123697,
      "grad_norm": 797.3460083007812,
      "learning_rate": 3.8273180086376195e-05,
      "loss": 91.1367,
      "step": 54980
    },
    {
      "epoch": 0.22216655825660458,
      "grad_norm": 1483.161865234375,
      "learning_rate": 3.827204479772557e-05,
      "loss": 51.6119,
      "step": 54990
    },
    {
      "epoch": 0.22220695952197222,
      "grad_norm": 1471.6492919921875,
      "learning_rate": 3.827090915285202e-05,
      "loss": 137.4107,
      "step": 55000
    },
    {
      "epoch": 0.22224736078733986,
      "grad_norm": 593.1961669921875,
      "learning_rate": 3.826977315177769e-05,
      "loss": 58.0393,
      "step": 55010
    },
    {
      "epoch": 0.22228776205270748,
      "grad_norm": 1406.79541015625,
      "learning_rate": 3.8268636794524724e-05,
      "loss": 81.1842,
      "step": 55020
    },
    {
      "epoch": 0.22232816331807512,
      "grad_norm": 1183.783203125,
      "learning_rate": 3.826750008111529e-05,
      "loss": 73.7566,
      "step": 55030
    },
    {
      "epoch": 0.22236856458344276,
      "grad_norm": 718.0703735351562,
      "learning_rate": 3.826636301157152e-05,
      "loss": 70.2362,
      "step": 55040
    },
    {
      "epoch": 0.2224089658488104,
      "grad_norm": 381.4212646484375,
      "learning_rate": 3.8265225585915616e-05,
      "loss": 63.0713,
      "step": 55050
    },
    {
      "epoch": 0.222449367114178,
      "grad_norm": 0.0,
      "learning_rate": 3.826408780416973e-05,
      "loss": 61.6525,
      "step": 55060
    },
    {
      "epoch": 0.22248976837954565,
      "grad_norm": 458.7480773925781,
      "learning_rate": 3.8262949666356046e-05,
      "loss": 48.9388,
      "step": 55070
    },
    {
      "epoch": 0.2225301696449133,
      "grad_norm": 785.8255004882812,
      "learning_rate": 3.826181117249676e-05,
      "loss": 61.7339,
      "step": 55080
    },
    {
      "epoch": 0.2225705709102809,
      "grad_norm": 790.076171875,
      "learning_rate": 3.8260672322614054e-05,
      "loss": 82.4103,
      "step": 55090
    },
    {
      "epoch": 0.22261097217564854,
      "grad_norm": 998.904296875,
      "learning_rate": 3.825953311673015e-05,
      "loss": 86.9125,
      "step": 55100
    },
    {
      "epoch": 0.22265137344101618,
      "grad_norm": 661.4468994140625,
      "learning_rate": 3.825839355486724e-05,
      "loss": 73.2599,
      "step": 55110
    },
    {
      "epoch": 0.2226917747063838,
      "grad_norm": 419.6131286621094,
      "learning_rate": 3.825725363704755e-05,
      "loss": 85.9492,
      "step": 55120
    },
    {
      "epoch": 0.22273217597175143,
      "grad_norm": 1006.8649291992188,
      "learning_rate": 3.82561133632933e-05,
      "loss": 76.4517,
      "step": 55130
    },
    {
      "epoch": 0.22277257723711907,
      "grad_norm": 595.6953125,
      "learning_rate": 3.825497273362673e-05,
      "loss": 67.5577,
      "step": 55140
    },
    {
      "epoch": 0.22281297850248669,
      "grad_norm": 559.351806640625,
      "learning_rate": 3.825383174807006e-05,
      "loss": 89.7158,
      "step": 55150
    },
    {
      "epoch": 0.22285337976785433,
      "grad_norm": 1137.574462890625,
      "learning_rate": 3.825269040664554e-05,
      "loss": 123.4338,
      "step": 55160
    },
    {
      "epoch": 0.22289378103322197,
      "grad_norm": 800.9525756835938,
      "learning_rate": 3.825154870937543e-05,
      "loss": 63.4805,
      "step": 55170
    },
    {
      "epoch": 0.22293418229858958,
      "grad_norm": 0.0,
      "learning_rate": 3.825040665628198e-05,
      "loss": 67.7657,
      "step": 55180
    },
    {
      "epoch": 0.22297458356395722,
      "grad_norm": 667.6600952148438,
      "learning_rate": 3.824926424738745e-05,
      "loss": 74.2919,
      "step": 55190
    },
    {
      "epoch": 0.22301498482932486,
      "grad_norm": 668.782958984375,
      "learning_rate": 3.824812148271413e-05,
      "loss": 60.5619,
      "step": 55200
    },
    {
      "epoch": 0.2230553860946925,
      "grad_norm": 883.674560546875,
      "learning_rate": 3.824697836228428e-05,
      "loss": 61.6603,
      "step": 55210
    },
    {
      "epoch": 0.2230957873600601,
      "grad_norm": 635.1038818359375,
      "learning_rate": 3.824583488612019e-05,
      "loss": 72.0887,
      "step": 55220
    },
    {
      "epoch": 0.22313618862542775,
      "grad_norm": 1303.10595703125,
      "learning_rate": 3.824469105424416e-05,
      "loss": 59.0078,
      "step": 55230
    },
    {
      "epoch": 0.2231765898907954,
      "grad_norm": 611.309814453125,
      "learning_rate": 3.824354686667848e-05,
      "loss": 64.6639,
      "step": 55240
    },
    {
      "epoch": 0.223216991156163,
      "grad_norm": 1070.7742919921875,
      "learning_rate": 3.8242402323445464e-05,
      "loss": 77.4441,
      "step": 55250
    },
    {
      "epoch": 0.22325739242153064,
      "grad_norm": 694.8192138671875,
      "learning_rate": 3.824125742456742e-05,
      "loss": 69.6013,
      "step": 55260
    },
    {
      "epoch": 0.22329779368689828,
      "grad_norm": 752.5867919921875,
      "learning_rate": 3.824011217006668e-05,
      "loss": 81.2646,
      "step": 55270
    },
    {
      "epoch": 0.2233381949522659,
      "grad_norm": 547.7276611328125,
      "learning_rate": 3.823896655996556e-05,
      "loss": 96.0991,
      "step": 55280
    },
    {
      "epoch": 0.22337859621763353,
      "grad_norm": 612.2052612304688,
      "learning_rate": 3.823782059428639e-05,
      "loss": 66.8349,
      "step": 55290
    },
    {
      "epoch": 0.22341899748300117,
      "grad_norm": 1166.7867431640625,
      "learning_rate": 3.823667427305152e-05,
      "loss": 79.9102,
      "step": 55300
    },
    {
      "epoch": 0.2234593987483688,
      "grad_norm": 985.7227172851562,
      "learning_rate": 3.8235527596283294e-05,
      "loss": 76.2051,
      "step": 55310
    },
    {
      "epoch": 0.22349980001373643,
      "grad_norm": 1459.9608154296875,
      "learning_rate": 3.823438056400408e-05,
      "loss": 96.8804,
      "step": 55320
    },
    {
      "epoch": 0.22354020127910407,
      "grad_norm": 584.8526000976562,
      "learning_rate": 3.823323317623622e-05,
      "loss": 37.4984,
      "step": 55330
    },
    {
      "epoch": 0.22358060254447168,
      "grad_norm": 1119.935546875,
      "learning_rate": 3.823208543300209e-05,
      "loss": 101.0891,
      "step": 55340
    },
    {
      "epoch": 0.22362100380983932,
      "grad_norm": 2631.894775390625,
      "learning_rate": 3.8230937334324075e-05,
      "loss": 102.7161,
      "step": 55350
    },
    {
      "epoch": 0.22366140507520696,
      "grad_norm": 1560.7054443359375,
      "learning_rate": 3.822978888022455e-05,
      "loss": 106.3137,
      "step": 55360
    },
    {
      "epoch": 0.2237018063405746,
      "grad_norm": 715.1233520507812,
      "learning_rate": 3.82286400707259e-05,
      "loss": 67.4711,
      "step": 55370
    },
    {
      "epoch": 0.2237422076059422,
      "grad_norm": 993.5302734375,
      "learning_rate": 3.8227490905850534e-05,
      "loss": 80.1368,
      "step": 55380
    },
    {
      "epoch": 0.22378260887130985,
      "grad_norm": 548.71484375,
      "learning_rate": 3.8226341385620845e-05,
      "loss": 68.046,
      "step": 55390
    },
    {
      "epoch": 0.2238230101366775,
      "grad_norm": 819.5445556640625,
      "learning_rate": 3.822519151005925e-05,
      "loss": 81.1588,
      "step": 55400
    },
    {
      "epoch": 0.2238634114020451,
      "grad_norm": 516.8148193359375,
      "learning_rate": 3.822404127918816e-05,
      "loss": 78.6651,
      "step": 55410
    },
    {
      "epoch": 0.22390381266741274,
      "grad_norm": 489.1547546386719,
      "learning_rate": 3.8222890693030007e-05,
      "loss": 104.5522,
      "step": 55420
    },
    {
      "epoch": 0.22394421393278038,
      "grad_norm": 440.1009216308594,
      "learning_rate": 3.8221739751607205e-05,
      "loss": 79.4987,
      "step": 55430
    },
    {
      "epoch": 0.223984615198148,
      "grad_norm": 583.6029663085938,
      "learning_rate": 3.822058845494222e-05,
      "loss": 69.6008,
      "step": 55440
    },
    {
      "epoch": 0.22402501646351564,
      "grad_norm": 528.2589111328125,
      "learning_rate": 3.8219436803057484e-05,
      "loss": 79.5607,
      "step": 55450
    },
    {
      "epoch": 0.22406541772888328,
      "grad_norm": 413.80364990234375,
      "learning_rate": 3.821828479597543e-05,
      "loss": 47.3397,
      "step": 55460
    },
    {
      "epoch": 0.2241058189942509,
      "grad_norm": 461.31805419921875,
      "learning_rate": 3.8217132433718556e-05,
      "loss": 79.1027,
      "step": 55470
    },
    {
      "epoch": 0.22414622025961853,
      "grad_norm": 686.8890991210938,
      "learning_rate": 3.821597971630929e-05,
      "loss": 77.6598,
      "step": 55480
    },
    {
      "epoch": 0.22418662152498617,
      "grad_norm": 711.459716796875,
      "learning_rate": 3.821482664377013e-05,
      "loss": 67.9017,
      "step": 55490
    },
    {
      "epoch": 0.22422702279035378,
      "grad_norm": 5779.49951171875,
      "learning_rate": 3.821367321612354e-05,
      "loss": 112.8159,
      "step": 55500
    },
    {
      "epoch": 0.22426742405572142,
      "grad_norm": 1230.7401123046875,
      "learning_rate": 3.821251943339202e-05,
      "loss": 87.1288,
      "step": 55510
    },
    {
      "epoch": 0.22430782532108906,
      "grad_norm": 412.5945129394531,
      "learning_rate": 3.821136529559806e-05,
      "loss": 78.4689,
      "step": 55520
    },
    {
      "epoch": 0.2243482265864567,
      "grad_norm": 644.5408325195312,
      "learning_rate": 3.821021080276415e-05,
      "loss": 69.6584,
      "step": 55530
    },
    {
      "epoch": 0.2243886278518243,
      "grad_norm": 530.8232421875,
      "learning_rate": 3.8209055954912815e-05,
      "loss": 80.4994,
      "step": 55540
    },
    {
      "epoch": 0.22442902911719195,
      "grad_norm": 754.1405639648438,
      "learning_rate": 3.820790075206655e-05,
      "loss": 96.8612,
      "step": 55550
    },
    {
      "epoch": 0.2244694303825596,
      "grad_norm": 706.6307373046875,
      "learning_rate": 3.820674519424789e-05,
      "loss": 62.1917,
      "step": 55560
    },
    {
      "epoch": 0.2245098316479272,
      "grad_norm": 949.218505859375,
      "learning_rate": 3.820558928147935e-05,
      "loss": 75.9506,
      "step": 55570
    },
    {
      "epoch": 0.22455023291329484,
      "grad_norm": 817.1376342773438,
      "learning_rate": 3.820443301378348e-05,
      "loss": 43.3707,
      "step": 55580
    },
    {
      "epoch": 0.22459063417866248,
      "grad_norm": 702.3113403320312,
      "learning_rate": 3.820327639118282e-05,
      "loss": 94.1257,
      "step": 55590
    },
    {
      "epoch": 0.2246310354440301,
      "grad_norm": 1354.1820068359375,
      "learning_rate": 3.8202119413699914e-05,
      "loss": 51.7388,
      "step": 55600
    },
    {
      "epoch": 0.22467143670939774,
      "grad_norm": 1198.4644775390625,
      "learning_rate": 3.820096208135732e-05,
      "loss": 91.0793,
      "step": 55610
    },
    {
      "epoch": 0.22471183797476538,
      "grad_norm": 521.9857788085938,
      "learning_rate": 3.8199804394177594e-05,
      "loss": 48.2309,
      "step": 55620
    },
    {
      "epoch": 0.224752239240133,
      "grad_norm": 500.62518310546875,
      "learning_rate": 3.819864635218332e-05,
      "loss": 91.5175,
      "step": 55630
    },
    {
      "epoch": 0.22479264050550063,
      "grad_norm": 592.45361328125,
      "learning_rate": 3.819748795539706e-05,
      "loss": 86.7297,
      "step": 55640
    },
    {
      "epoch": 0.22483304177086827,
      "grad_norm": 719.0573120117188,
      "learning_rate": 3.8196329203841404e-05,
      "loss": 102.5109,
      "step": 55650
    },
    {
      "epoch": 0.22487344303623588,
      "grad_norm": 1039.817626953125,
      "learning_rate": 3.819517009753894e-05,
      "loss": 82.0633,
      "step": 55660
    },
    {
      "epoch": 0.22491384430160352,
      "grad_norm": 584.3021240234375,
      "learning_rate": 3.8194010636512274e-05,
      "loss": 67.5605,
      "step": 55670
    },
    {
      "epoch": 0.22495424556697116,
      "grad_norm": 365.3063659667969,
      "learning_rate": 3.8192850820784e-05,
      "loss": 76.2773,
      "step": 55680
    },
    {
      "epoch": 0.2249946468323388,
      "grad_norm": 683.2793579101562,
      "learning_rate": 3.8191690650376734e-05,
      "loss": 111.8646,
      "step": 55690
    },
    {
      "epoch": 0.2250350480977064,
      "grad_norm": 5094.06884765625,
      "learning_rate": 3.819053012531309e-05,
      "loss": 98.7676,
      "step": 55700
    },
    {
      "epoch": 0.22507544936307405,
      "grad_norm": 1392.592041015625,
      "learning_rate": 3.8189369245615695e-05,
      "loss": 97.9336,
      "step": 55710
    },
    {
      "epoch": 0.2251158506284417,
      "grad_norm": 1905.1131591796875,
      "learning_rate": 3.818820801130719e-05,
      "loss": 76.434,
      "step": 55720
    },
    {
      "epoch": 0.2251562518938093,
      "grad_norm": 1060.8299560546875,
      "learning_rate": 3.81870464224102e-05,
      "loss": 76.4275,
      "step": 55730
    },
    {
      "epoch": 0.22519665315917695,
      "grad_norm": 697.6439819335938,
      "learning_rate": 3.818588447894738e-05,
      "loss": 74.6944,
      "step": 55740
    },
    {
      "epoch": 0.22523705442454459,
      "grad_norm": 1250.4781494140625,
      "learning_rate": 3.8184722180941376e-05,
      "loss": 85.4563,
      "step": 55750
    },
    {
      "epoch": 0.2252774556899122,
      "grad_norm": 889.2294311523438,
      "learning_rate": 3.818355952841485e-05,
      "loss": 86.7714,
      "step": 55760
    },
    {
      "epoch": 0.22531785695527984,
      "grad_norm": 442.0536804199219,
      "learning_rate": 3.818239652139047e-05,
      "loss": 64.0727,
      "step": 55770
    },
    {
      "epoch": 0.22535825822064748,
      "grad_norm": 449.8352966308594,
      "learning_rate": 3.818123315989091e-05,
      "loss": 92.5837,
      "step": 55780
    },
    {
      "epoch": 0.2253986594860151,
      "grad_norm": 450.0088195800781,
      "learning_rate": 3.818006944393885e-05,
      "loss": 78.8138,
      "step": 55790
    },
    {
      "epoch": 0.22543906075138273,
      "grad_norm": 486.0877685546875,
      "learning_rate": 3.817890537355698e-05,
      "loss": 58.394,
      "step": 55800
    },
    {
      "epoch": 0.22547946201675037,
      "grad_norm": 472.1658630371094,
      "learning_rate": 3.817774094876798e-05,
      "loss": 48.9125,
      "step": 55810
    },
    {
      "epoch": 0.22551986328211798,
      "grad_norm": 475.9318542480469,
      "learning_rate": 3.8176576169594574e-05,
      "loss": 84.1555,
      "step": 55820
    },
    {
      "epoch": 0.22556026454748562,
      "grad_norm": 847.432861328125,
      "learning_rate": 3.817541103605945e-05,
      "loss": 80.2289,
      "step": 55830
    },
    {
      "epoch": 0.22560066581285326,
      "grad_norm": 814.4562377929688,
      "learning_rate": 3.817424554818533e-05,
      "loss": 56.8917,
      "step": 55840
    },
    {
      "epoch": 0.2256410670782209,
      "grad_norm": 1087.533203125,
      "learning_rate": 3.817307970599494e-05,
      "loss": 61.4748,
      "step": 55850
    },
    {
      "epoch": 0.22568146834358851,
      "grad_norm": 1206.31005859375,
      "learning_rate": 3.817191350951101e-05,
      "loss": 81.1293,
      "step": 55860
    },
    {
      "epoch": 0.22572186960895615,
      "grad_norm": 1083.26904296875,
      "learning_rate": 3.817074695875626e-05,
      "loss": 83.8608,
      "step": 55870
    },
    {
      "epoch": 0.2257622708743238,
      "grad_norm": 584.4603271484375,
      "learning_rate": 3.816958005375344e-05,
      "loss": 67.0806,
      "step": 55880
    },
    {
      "epoch": 0.2258026721396914,
      "grad_norm": 1291.0843505859375,
      "learning_rate": 3.816841279452532e-05,
      "loss": 90.554,
      "step": 55890
    },
    {
      "epoch": 0.22584307340505905,
      "grad_norm": 831.9024047851562,
      "learning_rate": 3.816724518109463e-05,
      "loss": 65.832,
      "step": 55900
    },
    {
      "epoch": 0.2258834746704267,
      "grad_norm": 407.05072021484375,
      "learning_rate": 3.8166077213484135e-05,
      "loss": 58.9891,
      "step": 55910
    },
    {
      "epoch": 0.2259238759357943,
      "grad_norm": 1658.80908203125,
      "learning_rate": 3.816490889171662e-05,
      "loss": 65.7529,
      "step": 55920
    },
    {
      "epoch": 0.22596427720116194,
      "grad_norm": 1227.1304931640625,
      "learning_rate": 3.816374021581485e-05,
      "loss": 92.7391,
      "step": 55930
    },
    {
      "epoch": 0.22600467846652958,
      "grad_norm": 703.5914916992188,
      "learning_rate": 3.816257118580161e-05,
      "loss": 82.3,
      "step": 55940
    },
    {
      "epoch": 0.2260450797318972,
      "grad_norm": 529.1463012695312,
      "learning_rate": 3.816140180169971e-05,
      "loss": 62.4928,
      "step": 55950
    },
    {
      "epoch": 0.22608548099726483,
      "grad_norm": 769.5684814453125,
      "learning_rate": 3.816023206353192e-05,
      "loss": 74.4622,
      "step": 55960
    },
    {
      "epoch": 0.22612588226263247,
      "grad_norm": 784.1138916015625,
      "learning_rate": 3.815906197132106e-05,
      "loss": 58.6365,
      "step": 55970
    },
    {
      "epoch": 0.22616628352800008,
      "grad_norm": 1663.0970458984375,
      "learning_rate": 3.815789152508993e-05,
      "loss": 58.3804,
      "step": 55980
    },
    {
      "epoch": 0.22620668479336772,
      "grad_norm": 806.343505859375,
      "learning_rate": 3.815672072486136e-05,
      "loss": 51.7544,
      "step": 55990
    },
    {
      "epoch": 0.22624708605873536,
      "grad_norm": 295.89154052734375,
      "learning_rate": 3.815554957065818e-05,
      "loss": 68.6234,
      "step": 56000
    },
    {
      "epoch": 0.226287487324103,
      "grad_norm": 492.88128662109375,
      "learning_rate": 3.8154378062503207e-05,
      "loss": 83.6571,
      "step": 56010
    },
    {
      "epoch": 0.22632788858947062,
      "grad_norm": 598.7699584960938,
      "learning_rate": 3.815320620041929e-05,
      "loss": 63.2561,
      "step": 56020
    },
    {
      "epoch": 0.22636828985483826,
      "grad_norm": 1166.81201171875,
      "learning_rate": 3.815203398442927e-05,
      "loss": 66.7785,
      "step": 56030
    },
    {
      "epoch": 0.2264086911202059,
      "grad_norm": 1060.6319580078125,
      "learning_rate": 3.8150861414556e-05,
      "loss": 65.2264,
      "step": 56040
    },
    {
      "epoch": 0.2264490923855735,
      "grad_norm": 738.6588745117188,
      "learning_rate": 3.814968849082234e-05,
      "loss": 81.1179,
      "step": 56050
    },
    {
      "epoch": 0.22648949365094115,
      "grad_norm": 1409.0303955078125,
      "learning_rate": 3.8148515213251166e-05,
      "loss": 93.0806,
      "step": 56060
    },
    {
      "epoch": 0.2265298949163088,
      "grad_norm": 389.2867736816406,
      "learning_rate": 3.814734158186534e-05,
      "loss": 69.5129,
      "step": 56070
    },
    {
      "epoch": 0.2265702961816764,
      "grad_norm": 873.1412963867188,
      "learning_rate": 3.8146167596687746e-05,
      "loss": 87.3668,
      "step": 56080
    },
    {
      "epoch": 0.22661069744704404,
      "grad_norm": 988.9115600585938,
      "learning_rate": 3.814499325774128e-05,
      "loss": 70.7732,
      "step": 56090
    },
    {
      "epoch": 0.22665109871241168,
      "grad_norm": 1032.6561279296875,
      "learning_rate": 3.814381856504882e-05,
      "loss": 73.4399,
      "step": 56100
    },
    {
      "epoch": 0.2266914999777793,
      "grad_norm": 689.4199829101562,
      "learning_rate": 3.814264351863328e-05,
      "loss": 53.0507,
      "step": 56110
    },
    {
      "epoch": 0.22673190124314693,
      "grad_norm": 363.88037109375,
      "learning_rate": 3.814146811851757e-05,
      "loss": 54.2716,
      "step": 56120
    },
    {
      "epoch": 0.22677230250851457,
      "grad_norm": 979.6790771484375,
      "learning_rate": 3.814029236472459e-05,
      "loss": 68.8041,
      "step": 56130
    },
    {
      "epoch": 0.22681270377388218,
      "grad_norm": 1053.1512451171875,
      "learning_rate": 3.813911625727727e-05,
      "loss": 65.6413,
      "step": 56140
    },
    {
      "epoch": 0.22685310503924982,
      "grad_norm": 832.77099609375,
      "learning_rate": 3.813793979619854e-05,
      "loss": 56.1802,
      "step": 56150
    },
    {
      "epoch": 0.22689350630461746,
      "grad_norm": 738.0537719726562,
      "learning_rate": 3.813676298151135e-05,
      "loss": 97.4371,
      "step": 56160
    },
    {
      "epoch": 0.2269339075699851,
      "grad_norm": 1048.9244384765625,
      "learning_rate": 3.8135585813238616e-05,
      "loss": 101.326,
      "step": 56170
    },
    {
      "epoch": 0.22697430883535272,
      "grad_norm": 1091.7122802734375,
      "learning_rate": 3.81344082914033e-05,
      "loss": 62.9711,
      "step": 56180
    },
    {
      "epoch": 0.22701471010072036,
      "grad_norm": 656.7578735351562,
      "learning_rate": 3.8133230416028355e-05,
      "loss": 67.8833,
      "step": 56190
    },
    {
      "epoch": 0.227055111366088,
      "grad_norm": 1725.5526123046875,
      "learning_rate": 3.813205218713676e-05,
      "loss": 106.0663,
      "step": 56200
    },
    {
      "epoch": 0.2270955126314556,
      "grad_norm": 558.1640625,
      "learning_rate": 3.813087360475146e-05,
      "loss": 49.1143,
      "step": 56210
    },
    {
      "epoch": 0.22713591389682325,
      "grad_norm": 1560.0738525390625,
      "learning_rate": 3.812969466889545e-05,
      "loss": 69.2833,
      "step": 56220
    },
    {
      "epoch": 0.2271763151621909,
      "grad_norm": 1940.1322021484375,
      "learning_rate": 3.812851537959171e-05,
      "loss": 77.8127,
      "step": 56230
    },
    {
      "epoch": 0.2272167164275585,
      "grad_norm": 1342.005615234375,
      "learning_rate": 3.8127335736863227e-05,
      "loss": 69.5098,
      "step": 56240
    },
    {
      "epoch": 0.22725711769292614,
      "grad_norm": 768.77490234375,
      "learning_rate": 3.812615574073301e-05,
      "loss": 63.5844,
      "step": 56250
    },
    {
      "epoch": 0.22729751895829378,
      "grad_norm": 1894.3037109375,
      "learning_rate": 3.812497539122404e-05,
      "loss": 92.5692,
      "step": 56260
    },
    {
      "epoch": 0.2273379202236614,
      "grad_norm": 886.0888061523438,
      "learning_rate": 3.812379468835935e-05,
      "loss": 66.3142,
      "step": 56270
    },
    {
      "epoch": 0.22737832148902903,
      "grad_norm": 2326.173095703125,
      "learning_rate": 3.812261363216195e-05,
      "loss": 81.2702,
      "step": 56280
    },
    {
      "epoch": 0.22741872275439667,
      "grad_norm": 601.2302856445312,
      "learning_rate": 3.812143222265487e-05,
      "loss": 83.3885,
      "step": 56290
    },
    {
      "epoch": 0.22745912401976429,
      "grad_norm": 715.7938842773438,
      "learning_rate": 3.8120250459861144e-05,
      "loss": 60.7751,
      "step": 56300
    },
    {
      "epoch": 0.22749952528513193,
      "grad_norm": 1360.266845703125,
      "learning_rate": 3.81190683438038e-05,
      "loss": 111.9645,
      "step": 56310
    },
    {
      "epoch": 0.22753992655049957,
      "grad_norm": 874.0000610351562,
      "learning_rate": 3.811788587450589e-05,
      "loss": 61.2237,
      "step": 56320
    },
    {
      "epoch": 0.2275803278158672,
      "grad_norm": 531.2874755859375,
      "learning_rate": 3.8116703051990464e-05,
      "loss": 84.936,
      "step": 56330
    },
    {
      "epoch": 0.22762072908123482,
      "grad_norm": 1022.2213745117188,
      "learning_rate": 3.811551987628059e-05,
      "loss": 52.4177,
      "step": 56340
    },
    {
      "epoch": 0.22766113034660246,
      "grad_norm": 813.2898559570312,
      "learning_rate": 3.811433634739933e-05,
      "loss": 49.6749,
      "step": 56350
    },
    {
      "epoch": 0.2277015316119701,
      "grad_norm": 661.9342651367188,
      "learning_rate": 3.8113152465369756e-05,
      "loss": 67.0093,
      "step": 56360
    },
    {
      "epoch": 0.2277419328773377,
      "grad_norm": 1208.7891845703125,
      "learning_rate": 3.8111968230214945e-05,
      "loss": 74.5439,
      "step": 56370
    },
    {
      "epoch": 0.22778233414270535,
      "grad_norm": 859.9384155273438,
      "learning_rate": 3.811078364195799e-05,
      "loss": 43.7246,
      "step": 56380
    },
    {
      "epoch": 0.227822735408073,
      "grad_norm": 825.8929443359375,
      "learning_rate": 3.810959870062199e-05,
      "loss": 79.4211,
      "step": 56390
    },
    {
      "epoch": 0.2278631366734406,
      "grad_norm": 1221.9461669921875,
      "learning_rate": 3.810841340623004e-05,
      "loss": 95.7748,
      "step": 56400
    },
    {
      "epoch": 0.22790353793880824,
      "grad_norm": 1074.3507080078125,
      "learning_rate": 3.810722775880524e-05,
      "loss": 67.8669,
      "step": 56410
    },
    {
      "epoch": 0.22794393920417588,
      "grad_norm": 706.3327026367188,
      "learning_rate": 3.810604175837072e-05,
      "loss": 93.8343,
      "step": 56420
    },
    {
      "epoch": 0.2279843404695435,
      "grad_norm": 1404.4757080078125,
      "learning_rate": 3.8104855404949585e-05,
      "loss": 68.0535,
      "step": 56430
    },
    {
      "epoch": 0.22802474173491113,
      "grad_norm": 732.0487670898438,
      "learning_rate": 3.810366869856498e-05,
      "loss": 96.2141,
      "step": 56440
    },
    {
      "epoch": 0.22806514300027877,
      "grad_norm": 1142.020751953125,
      "learning_rate": 3.8102481639240035e-05,
      "loss": 67.8034,
      "step": 56450
    },
    {
      "epoch": 0.2281055442656464,
      "grad_norm": 646.2639770507812,
      "learning_rate": 3.810129422699789e-05,
      "loss": 91.0126,
      "step": 56460
    },
    {
      "epoch": 0.22814594553101403,
      "grad_norm": 1303.6097412109375,
      "learning_rate": 3.810010646186169e-05,
      "loss": 97.9156,
      "step": 56470
    },
    {
      "epoch": 0.22818634679638167,
      "grad_norm": 1019.5330810546875,
      "learning_rate": 3.80989183438546e-05,
      "loss": 69.5221,
      "step": 56480
    },
    {
      "epoch": 0.2282267480617493,
      "grad_norm": 389.3905029296875,
      "learning_rate": 3.809772987299978e-05,
      "loss": 125.0458,
      "step": 56490
    },
    {
      "epoch": 0.22826714932711692,
      "grad_norm": 720.518310546875,
      "learning_rate": 3.809654104932039e-05,
      "loss": 121.4447,
      "step": 56500
    },
    {
      "epoch": 0.22830755059248456,
      "grad_norm": 901.7028198242188,
      "learning_rate": 3.809535187283962e-05,
      "loss": 49.731,
      "step": 56510
    },
    {
      "epoch": 0.2283479518578522,
      "grad_norm": 469.4713134765625,
      "learning_rate": 3.8094162343580654e-05,
      "loss": 98.3911,
      "step": 56520
    },
    {
      "epoch": 0.2283883531232198,
      "grad_norm": 1055.3323974609375,
      "learning_rate": 3.809297246156668e-05,
      "loss": 75.417,
      "step": 56530
    },
    {
      "epoch": 0.22842875438858745,
      "grad_norm": 632.37109375,
      "learning_rate": 3.8091782226820886e-05,
      "loss": 95.9724,
      "step": 56540
    },
    {
      "epoch": 0.2284691556539551,
      "grad_norm": 1114.0428466796875,
      "learning_rate": 3.809059163936648e-05,
      "loss": 50.7466,
      "step": 56550
    },
    {
      "epoch": 0.2285095569193227,
      "grad_norm": 1918.535400390625,
      "learning_rate": 3.808940069922669e-05,
      "loss": 91.4348,
      "step": 56560
    },
    {
      "epoch": 0.22854995818469034,
      "grad_norm": 453.5932922363281,
      "learning_rate": 3.808820940642471e-05,
      "loss": 65.904,
      "step": 56570
    },
    {
      "epoch": 0.22859035945005798,
      "grad_norm": 699.6660766601562,
      "learning_rate": 3.8087017760983774e-05,
      "loss": 67.6307,
      "step": 56580
    },
    {
      "epoch": 0.2286307607154256,
      "grad_norm": 954.3672485351562,
      "learning_rate": 3.808582576292712e-05,
      "loss": 78.6105,
      "step": 56590
    },
    {
      "epoch": 0.22867116198079324,
      "grad_norm": 991.3224487304688,
      "learning_rate": 3.8084633412277974e-05,
      "loss": 72.0767,
      "step": 56600
    },
    {
      "epoch": 0.22871156324616088,
      "grad_norm": 896.9384765625,
      "learning_rate": 3.808344070905959e-05,
      "loss": 39.0718,
      "step": 56610
    },
    {
      "epoch": 0.2287519645115285,
      "grad_norm": 1196.9996337890625,
      "learning_rate": 3.808224765329523e-05,
      "loss": 99.872,
      "step": 56620
    },
    {
      "epoch": 0.22879236577689613,
      "grad_norm": 1548.3662109375,
      "learning_rate": 3.808105424500814e-05,
      "loss": 107.1882,
      "step": 56630
    },
    {
      "epoch": 0.22883276704226377,
      "grad_norm": 1136.67822265625,
      "learning_rate": 3.807986048422158e-05,
      "loss": 84.6685,
      "step": 56640
    },
    {
      "epoch": 0.2288731683076314,
      "grad_norm": 1998.874267578125,
      "learning_rate": 3.807866637095884e-05,
      "loss": 92.7216,
      "step": 56650
    },
    {
      "epoch": 0.22891356957299902,
      "grad_norm": 973.369873046875,
      "learning_rate": 3.807747190524318e-05,
      "loss": 84.9332,
      "step": 56660
    },
    {
      "epoch": 0.22895397083836666,
      "grad_norm": 684.3551025390625,
      "learning_rate": 3.8076277087097915e-05,
      "loss": 56.3043,
      "step": 56670
    },
    {
      "epoch": 0.2289943721037343,
      "grad_norm": 929.8072509765625,
      "learning_rate": 3.807508191654632e-05,
      "loss": 100.4173,
      "step": 56680
    },
    {
      "epoch": 0.2290347733691019,
      "grad_norm": 375.4898986816406,
      "learning_rate": 3.807388639361168e-05,
      "loss": 70.284,
      "step": 56690
    },
    {
      "epoch": 0.22907517463446955,
      "grad_norm": 1525.9503173828125,
      "learning_rate": 3.8072690518317334e-05,
      "loss": 55.3537,
      "step": 56700
    },
    {
      "epoch": 0.2291155758998372,
      "grad_norm": 792.93896484375,
      "learning_rate": 3.807149429068658e-05,
      "loss": 71.9927,
      "step": 56710
    },
    {
      "epoch": 0.2291559771652048,
      "grad_norm": 931.7693481445312,
      "learning_rate": 3.807029771074274e-05,
      "loss": 96.6875,
      "step": 56720
    },
    {
      "epoch": 0.22919637843057244,
      "grad_norm": 1130.6885986328125,
      "learning_rate": 3.806910077850914e-05,
      "loss": 80.8834,
      "step": 56730
    },
    {
      "epoch": 0.22923677969594008,
      "grad_norm": 365.8529052734375,
      "learning_rate": 3.806790349400912e-05,
      "loss": 64.6014,
      "step": 56740
    },
    {
      "epoch": 0.2292771809613077,
      "grad_norm": 5867.74560546875,
      "learning_rate": 3.806670585726602e-05,
      "loss": 98.2985,
      "step": 56750
    },
    {
      "epoch": 0.22931758222667534,
      "grad_norm": 962.7188110351562,
      "learning_rate": 3.806550786830319e-05,
      "loss": 99.5331,
      "step": 56760
    },
    {
      "epoch": 0.22935798349204298,
      "grad_norm": 413.5105895996094,
      "learning_rate": 3.806430952714398e-05,
      "loss": 63.9609,
      "step": 56770
    },
    {
      "epoch": 0.2293983847574106,
      "grad_norm": 979.9683837890625,
      "learning_rate": 3.806311083381175e-05,
      "loss": 92.7886,
      "step": 56780
    },
    {
      "epoch": 0.22943878602277823,
      "grad_norm": 1042.4696044921875,
      "learning_rate": 3.806191178832989e-05,
      "loss": 52.4567,
      "step": 56790
    },
    {
      "epoch": 0.22947918728814587,
      "grad_norm": 1003.2159423828125,
      "learning_rate": 3.806071239072175e-05,
      "loss": 44.2825,
      "step": 56800
    },
    {
      "epoch": 0.2295195885535135,
      "grad_norm": 1484.490966796875,
      "learning_rate": 3.8059512641010726e-05,
      "loss": 89.2549,
      "step": 56810
    },
    {
      "epoch": 0.22955998981888112,
      "grad_norm": 1329.6519775390625,
      "learning_rate": 3.80583125392202e-05,
      "loss": 63.9567,
      "step": 56820
    },
    {
      "epoch": 0.22960039108424876,
      "grad_norm": 667.831298828125,
      "learning_rate": 3.805711208537358e-05,
      "loss": 66.5759,
      "step": 56830
    },
    {
      "epoch": 0.2296407923496164,
      "grad_norm": 1251.3240966796875,
      "learning_rate": 3.805591127949426e-05,
      "loss": 95.0565,
      "step": 56840
    },
    {
      "epoch": 0.229681193614984,
      "grad_norm": 832.2625122070312,
      "learning_rate": 3.805471012160566e-05,
      "loss": 93.1982,
      "step": 56850
    },
    {
      "epoch": 0.22972159488035165,
      "grad_norm": 501.2583312988281,
      "learning_rate": 3.8053508611731185e-05,
      "loss": 55.0099,
      "step": 56860
    },
    {
      "epoch": 0.2297619961457193,
      "grad_norm": 766.9327392578125,
      "learning_rate": 3.805230674989427e-05,
      "loss": 85.5932,
      "step": 56870
    },
    {
      "epoch": 0.2298023974110869,
      "grad_norm": 712.4401245117188,
      "learning_rate": 3.805110453611834e-05,
      "loss": 86.5868,
      "step": 56880
    },
    {
      "epoch": 0.22984279867645455,
      "grad_norm": 1190.645263671875,
      "learning_rate": 3.804990197042683e-05,
      "loss": 86.9735,
      "step": 56890
    },
    {
      "epoch": 0.22988319994182219,
      "grad_norm": 908.5149536132812,
      "learning_rate": 3.804869905284319e-05,
      "loss": 66.2249,
      "step": 56900
    },
    {
      "epoch": 0.2299236012071898,
      "grad_norm": 739.5358276367188,
      "learning_rate": 3.8047495783390874e-05,
      "loss": 74.1746,
      "step": 56910
    },
    {
      "epoch": 0.22996400247255744,
      "grad_norm": 414.0364990234375,
      "learning_rate": 3.804629216209333e-05,
      "loss": 65.4793,
      "step": 56920
    },
    {
      "epoch": 0.23000440373792508,
      "grad_norm": 1083.235595703125,
      "learning_rate": 3.8045088188974026e-05,
      "loss": 118.2313,
      "step": 56930
    },
    {
      "epoch": 0.2300448050032927,
      "grad_norm": 969.8745727539062,
      "learning_rate": 3.804388386405645e-05,
      "loss": 81.8473,
      "step": 56940
    },
    {
      "epoch": 0.23008520626866033,
      "grad_norm": 727.8637084960938,
      "learning_rate": 3.804267918736406e-05,
      "loss": 83.6049,
      "step": 56950
    },
    {
      "epoch": 0.23012560753402797,
      "grad_norm": 1320.610595703125,
      "learning_rate": 3.8041474158920356e-05,
      "loss": 95.4228,
      "step": 56960
    },
    {
      "epoch": 0.2301660087993956,
      "grad_norm": 699.66796875,
      "learning_rate": 3.804026877874882e-05,
      "loss": 83.4832,
      "step": 56970
    },
    {
      "epoch": 0.23020641006476322,
      "grad_norm": 968.6633911132812,
      "learning_rate": 3.803906304687296e-05,
      "loss": 78.6762,
      "step": 56980
    },
    {
      "epoch": 0.23024681133013086,
      "grad_norm": 408.0656433105469,
      "learning_rate": 3.803785696331627e-05,
      "loss": 79.3933,
      "step": 56990
    },
    {
      "epoch": 0.2302872125954985,
      "grad_norm": 360.78814697265625,
      "learning_rate": 3.803665052810228e-05,
      "loss": 64.7716,
      "step": 57000
    },
    {
      "epoch": 0.23032761386086611,
      "grad_norm": 607.16748046875,
      "learning_rate": 3.803544374125449e-05,
      "loss": 60.6444,
      "step": 57010
    },
    {
      "epoch": 0.23036801512623375,
      "grad_norm": 761.7095947265625,
      "learning_rate": 3.803423660279646e-05,
      "loss": 66.7402,
      "step": 57020
    },
    {
      "epoch": 0.2304084163916014,
      "grad_norm": 603.35009765625,
      "learning_rate": 3.803302911275169e-05,
      "loss": 84.5132,
      "step": 57030
    },
    {
      "epoch": 0.230448817656969,
      "grad_norm": 1848.899169921875,
      "learning_rate": 3.803182127114374e-05,
      "loss": 61.8791,
      "step": 57040
    },
    {
      "epoch": 0.23048921892233665,
      "grad_norm": 559.4889526367188,
      "learning_rate": 3.803061307799614e-05,
      "loss": 90.565,
      "step": 57050
    },
    {
      "epoch": 0.2305296201877043,
      "grad_norm": 487.5803527832031,
      "learning_rate": 3.802940453333246e-05,
      "loss": 83.0407,
      "step": 57060
    },
    {
      "epoch": 0.2305700214530719,
      "grad_norm": 913.366455078125,
      "learning_rate": 3.802819563717626e-05,
      "loss": 112.3049,
      "step": 57070
    },
    {
      "epoch": 0.23061042271843954,
      "grad_norm": 1263.8419189453125,
      "learning_rate": 3.8026986389551105e-05,
      "loss": 85.9298,
      "step": 57080
    },
    {
      "epoch": 0.23065082398380718,
      "grad_norm": 305.4476623535156,
      "learning_rate": 3.802577679048057e-05,
      "loss": 45.9334,
      "step": 57090
    },
    {
      "epoch": 0.2306912252491748,
      "grad_norm": 593.847900390625,
      "learning_rate": 3.802456683998823e-05,
      "loss": 53.0649,
      "step": 57100
    },
    {
      "epoch": 0.23073162651454243,
      "grad_norm": 1953.02001953125,
      "learning_rate": 3.802335653809768e-05,
      "loss": 81.2339,
      "step": 57110
    },
    {
      "epoch": 0.23077202777991007,
      "grad_norm": 1080.8931884765625,
      "learning_rate": 3.802214588483252e-05,
      "loss": 81.1776,
      "step": 57120
    },
    {
      "epoch": 0.2308124290452777,
      "grad_norm": 396.26373291015625,
      "learning_rate": 3.8020934880216344e-05,
      "loss": 71.585,
      "step": 57130
    },
    {
      "epoch": 0.23085283031064532,
      "grad_norm": 1105.419189453125,
      "learning_rate": 3.801972352427277e-05,
      "loss": 102.6832,
      "step": 57140
    },
    {
      "epoch": 0.23089323157601296,
      "grad_norm": 795.7288818359375,
      "learning_rate": 3.801851181702541e-05,
      "loss": 48.1124,
      "step": 57150
    },
    {
      "epoch": 0.2309336328413806,
      "grad_norm": 1058.8551025390625,
      "learning_rate": 3.8017299758497875e-05,
      "loss": 79.2187,
      "step": 57160
    },
    {
      "epoch": 0.23097403410674822,
      "grad_norm": 1670.0687255859375,
      "learning_rate": 3.801608734871381e-05,
      "loss": 56.1381,
      "step": 57170
    },
    {
      "epoch": 0.23101443537211586,
      "grad_norm": 427.2680358886719,
      "learning_rate": 3.8014874587696846e-05,
      "loss": 96.5046,
      "step": 57180
    },
    {
      "epoch": 0.2310548366374835,
      "grad_norm": 677.8304443359375,
      "learning_rate": 3.8013661475470634e-05,
      "loss": 64.1499,
      "step": 57190
    },
    {
      "epoch": 0.2310952379028511,
      "grad_norm": 1529.733642578125,
      "learning_rate": 3.801244801205881e-05,
      "loss": 87.2802,
      "step": 57200
    },
    {
      "epoch": 0.23113563916821875,
      "grad_norm": 0.0,
      "learning_rate": 3.801123419748504e-05,
      "loss": 58.6589,
      "step": 57210
    },
    {
      "epoch": 0.2311760404335864,
      "grad_norm": 1265.4381103515625,
      "learning_rate": 3.801002003177299e-05,
      "loss": 71.0298,
      "step": 57220
    },
    {
      "epoch": 0.231216441698954,
      "grad_norm": 1491.0811767578125,
      "learning_rate": 3.800880551494632e-05,
      "loss": 86.0877,
      "step": 57230
    },
    {
      "epoch": 0.23125684296432164,
      "grad_norm": 557.4287719726562,
      "learning_rate": 3.800759064702873e-05,
      "loss": 93.9744,
      "step": 57240
    },
    {
      "epoch": 0.23129724422968928,
      "grad_norm": 326.94873046875,
      "learning_rate": 3.800637542804387e-05,
      "loss": 56.5683,
      "step": 57250
    },
    {
      "epoch": 0.2313376454950569,
      "grad_norm": 733.343505859375,
      "learning_rate": 3.800515985801546e-05,
      "loss": 60.8453,
      "step": 57260
    },
    {
      "epoch": 0.23137804676042453,
      "grad_norm": 883.8818359375,
      "learning_rate": 3.800394393696719e-05,
      "loss": 102.1983,
      "step": 57270
    },
    {
      "epoch": 0.23141844802579217,
      "grad_norm": 553.3629150390625,
      "learning_rate": 3.800272766492277e-05,
      "loss": 59.822,
      "step": 57280
    },
    {
      "epoch": 0.2314588492911598,
      "grad_norm": 1296.9984130859375,
      "learning_rate": 3.800151104190589e-05,
      "loss": 96.5139,
      "step": 57290
    },
    {
      "epoch": 0.23149925055652743,
      "grad_norm": 1746.950439453125,
      "learning_rate": 3.8000294067940295e-05,
      "loss": 84.3049,
      "step": 57300
    },
    {
      "epoch": 0.23153965182189506,
      "grad_norm": 372.6632080078125,
      "learning_rate": 3.79990767430497e-05,
      "loss": 115.2568,
      "step": 57310
    },
    {
      "epoch": 0.2315800530872627,
      "grad_norm": 1053.0244140625,
      "learning_rate": 3.799785906725784e-05,
      "loss": 109.0187,
      "step": 57320
    },
    {
      "epoch": 0.23162045435263032,
      "grad_norm": 0.0,
      "learning_rate": 3.7996641040588443e-05,
      "loss": 50.7161,
      "step": 57330
    },
    {
      "epoch": 0.23166085561799796,
      "grad_norm": 487.5233459472656,
      "learning_rate": 3.7995422663065264e-05,
      "loss": 65.7552,
      "step": 57340
    },
    {
      "epoch": 0.2317012568833656,
      "grad_norm": 549.9506225585938,
      "learning_rate": 3.799420393471206e-05,
      "loss": 94.2068,
      "step": 57350
    },
    {
      "epoch": 0.2317416581487332,
      "grad_norm": 2816.931884765625,
      "learning_rate": 3.799298485555259e-05,
      "loss": 76.761,
      "step": 57360
    },
    {
      "epoch": 0.23178205941410085,
      "grad_norm": 794.5421752929688,
      "learning_rate": 3.799176542561061e-05,
      "loss": 72.8375,
      "step": 57370
    },
    {
      "epoch": 0.2318224606794685,
      "grad_norm": 614.9068603515625,
      "learning_rate": 3.7990545644909905e-05,
      "loss": 60.9909,
      "step": 57380
    },
    {
      "epoch": 0.2318628619448361,
      "grad_norm": 961.1927490234375,
      "learning_rate": 3.798932551347424e-05,
      "loss": 83.393,
      "step": 57390
    },
    {
      "epoch": 0.23190326321020374,
      "grad_norm": 672.9813232421875,
      "learning_rate": 3.798810503132742e-05,
      "loss": 78.9836,
      "step": 57400
    },
    {
      "epoch": 0.23194366447557138,
      "grad_norm": 821.4522705078125,
      "learning_rate": 3.798688419849324e-05,
      "loss": 94.6433,
      "step": 57410
    },
    {
      "epoch": 0.231984065740939,
      "grad_norm": 415.84783935546875,
      "learning_rate": 3.798566301499548e-05,
      "loss": 98.24,
      "step": 57420
    },
    {
      "epoch": 0.23202446700630663,
      "grad_norm": 792.7560424804688,
      "learning_rate": 3.798444148085796e-05,
      "loss": 55.6949,
      "step": 57430
    },
    {
      "epoch": 0.23206486827167427,
      "grad_norm": 419.20770263671875,
      "learning_rate": 3.79832195961045e-05,
      "loss": 77.6947,
      "step": 57440
    },
    {
      "epoch": 0.23210526953704191,
      "grad_norm": 1155.12451171875,
      "learning_rate": 3.7981997360758917e-05,
      "loss": 93.3328,
      "step": 57450
    },
    {
      "epoch": 0.23214567080240953,
      "grad_norm": 1187.1175537109375,
      "learning_rate": 3.7980774774845035e-05,
      "loss": 81.4386,
      "step": 57460
    },
    {
      "epoch": 0.23218607206777717,
      "grad_norm": 660.607421875,
      "learning_rate": 3.797955183838669e-05,
      "loss": 105.6275,
      "step": 57470
    },
    {
      "epoch": 0.2322264733331448,
      "grad_norm": 649.1986694335938,
      "learning_rate": 3.797832855140773e-05,
      "loss": 73.3256,
      "step": 57480
    },
    {
      "epoch": 0.23226687459851242,
      "grad_norm": 415.9481201171875,
      "learning_rate": 3.797710491393199e-05,
      "loss": 72.9341,
      "step": 57490
    },
    {
      "epoch": 0.23230727586388006,
      "grad_norm": 1390.0479736328125,
      "learning_rate": 3.7975880925983345e-05,
      "loss": 86.1443,
      "step": 57500
    },
    {
      "epoch": 0.2323476771292477,
      "grad_norm": 721.8440551757812,
      "learning_rate": 3.7974656587585645e-05,
      "loss": 86.9662,
      "step": 57510
    },
    {
      "epoch": 0.2323880783946153,
      "grad_norm": 1014.0465698242188,
      "learning_rate": 3.7973431898762757e-05,
      "loss": 89.9118,
      "step": 57520
    },
    {
      "epoch": 0.23242847965998295,
      "grad_norm": 499.45050048828125,
      "learning_rate": 3.797220685953856e-05,
      "loss": 66.7644,
      "step": 57530
    },
    {
      "epoch": 0.2324688809253506,
      "grad_norm": 544.9583129882812,
      "learning_rate": 3.7970981469936936e-05,
      "loss": 39.842,
      "step": 57540
    },
    {
      "epoch": 0.2325092821907182,
      "grad_norm": 752.9443969726562,
      "learning_rate": 3.7969755729981776e-05,
      "loss": 52.3351,
      "step": 57550
    },
    {
      "epoch": 0.23254968345608584,
      "grad_norm": 666.3173217773438,
      "learning_rate": 3.796852963969699e-05,
      "loss": 86.4459,
      "step": 57560
    },
    {
      "epoch": 0.23259008472145348,
      "grad_norm": 882.573974609375,
      "learning_rate": 3.796730319910645e-05,
      "loss": 93.6017,
      "step": 57570
    },
    {
      "epoch": 0.2326304859868211,
      "grad_norm": 556.37109375,
      "learning_rate": 3.796607640823409e-05,
      "loss": 97.7774,
      "step": 57580
    },
    {
      "epoch": 0.23267088725218874,
      "grad_norm": 1168.2757568359375,
      "learning_rate": 3.7964849267103824e-05,
      "loss": 62.6684,
      "step": 57590
    },
    {
      "epoch": 0.23271128851755638,
      "grad_norm": 601.9468383789062,
      "learning_rate": 3.796362177573957e-05,
      "loss": 92.6433,
      "step": 57600
    },
    {
      "epoch": 0.23275168978292402,
      "grad_norm": 1131.87353515625,
      "learning_rate": 3.796239393416526e-05,
      "loss": 80.5643,
      "step": 57610
    },
    {
      "epoch": 0.23279209104829163,
      "grad_norm": 459.4023742675781,
      "learning_rate": 3.796116574240484e-05,
      "loss": 62.8203,
      "step": 57620
    },
    {
      "epoch": 0.23283249231365927,
      "grad_norm": 1018.473388671875,
      "learning_rate": 3.795993720048224e-05,
      "loss": 61.0011,
      "step": 57630
    },
    {
      "epoch": 0.2328728935790269,
      "grad_norm": 1023.8245849609375,
      "learning_rate": 3.795870830842142e-05,
      "loss": 76.0714,
      "step": 57640
    },
    {
      "epoch": 0.23291329484439452,
      "grad_norm": 667.9993286132812,
      "learning_rate": 3.795747906624634e-05,
      "loss": 81.131,
      "step": 57650
    },
    {
      "epoch": 0.23295369610976216,
      "grad_norm": 939.4752807617188,
      "learning_rate": 3.795624947398096e-05,
      "loss": 65.7513,
      "step": 57660
    },
    {
      "epoch": 0.2329940973751298,
      "grad_norm": 811.994384765625,
      "learning_rate": 3.795501953164924e-05,
      "loss": 69.7315,
      "step": 57670
    },
    {
      "epoch": 0.2330344986404974,
      "grad_norm": 863.0499877929688,
      "learning_rate": 3.795378923927518e-05,
      "loss": 110.3658,
      "step": 57680
    },
    {
      "epoch": 0.23307489990586505,
      "grad_norm": 885.8748779296875,
      "learning_rate": 3.795255859688276e-05,
      "loss": 93.2206,
      "step": 57690
    },
    {
      "epoch": 0.2331153011712327,
      "grad_norm": 708.9866943359375,
      "learning_rate": 3.7951327604495957e-05,
      "loss": 62.6562,
      "step": 57700
    },
    {
      "epoch": 0.2331557024366003,
      "grad_norm": 640.904052734375,
      "learning_rate": 3.7950096262138784e-05,
      "loss": 51.269,
      "step": 57710
    },
    {
      "epoch": 0.23319610370196794,
      "grad_norm": 695.4147338867188,
      "learning_rate": 3.794886456983524e-05,
      "loss": 101.4676,
      "step": 57720
    },
    {
      "epoch": 0.23323650496733558,
      "grad_norm": 817.9987182617188,
      "learning_rate": 3.7947632527609345e-05,
      "loss": 83.4514,
      "step": 57730
    },
    {
      "epoch": 0.2332769062327032,
      "grad_norm": 2216.619873046875,
      "learning_rate": 3.7946400135485115e-05,
      "loss": 76.9581,
      "step": 57740
    },
    {
      "epoch": 0.23331730749807084,
      "grad_norm": 1862.0076904296875,
      "learning_rate": 3.794516739348657e-05,
      "loss": 77.3212,
      "step": 57750
    },
    {
      "epoch": 0.23335770876343848,
      "grad_norm": 510.77276611328125,
      "learning_rate": 3.794393430163776e-05,
      "loss": 51.2993,
      "step": 57760
    },
    {
      "epoch": 0.2333981100288061,
      "grad_norm": 278.53564453125,
      "learning_rate": 3.7942700859962694e-05,
      "loss": 87.068,
      "step": 57770
    },
    {
      "epoch": 0.23343851129417373,
      "grad_norm": 1577.158935546875,
      "learning_rate": 3.794146706848545e-05,
      "loss": 114.5216,
      "step": 57780
    },
    {
      "epoch": 0.23347891255954137,
      "grad_norm": 1093.2125244140625,
      "learning_rate": 3.794023292723006e-05,
      "loss": 72.6221,
      "step": 57790
    },
    {
      "epoch": 0.233519313824909,
      "grad_norm": 1251.1124267578125,
      "learning_rate": 3.7938998436220604e-05,
      "loss": 86.3804,
      "step": 57800
    },
    {
      "epoch": 0.23355971509027662,
      "grad_norm": 3054.21240234375,
      "learning_rate": 3.793776359548113e-05,
      "loss": 87.1267,
      "step": 57810
    },
    {
      "epoch": 0.23360011635564426,
      "grad_norm": 1642.425537109375,
      "learning_rate": 3.793652840503572e-05,
      "loss": 53.5206,
      "step": 57820
    },
    {
      "epoch": 0.2336405176210119,
      "grad_norm": 997.0510864257812,
      "learning_rate": 3.793529286490846e-05,
      "loss": 80.6299,
      "step": 57830
    },
    {
      "epoch": 0.2336809188863795,
      "grad_norm": 304.63812255859375,
      "learning_rate": 3.7934056975123424e-05,
      "loss": 68.2548,
      "step": 57840
    },
    {
      "epoch": 0.23372132015174715,
      "grad_norm": 576.9412231445312,
      "learning_rate": 3.793282073570472e-05,
      "loss": 100.3076,
      "step": 57850
    },
    {
      "epoch": 0.2337617214171148,
      "grad_norm": 967.371826171875,
      "learning_rate": 3.7931584146676444e-05,
      "loss": 103.107,
      "step": 57860
    },
    {
      "epoch": 0.2338021226824824,
      "grad_norm": 851.3161010742188,
      "learning_rate": 3.793034720806269e-05,
      "loss": 59.3635,
      "step": 57870
    },
    {
      "epoch": 0.23384252394785005,
      "grad_norm": 748.4859008789062,
      "learning_rate": 3.792910991988761e-05,
      "loss": 77.2267,
      "step": 57880
    },
    {
      "epoch": 0.23388292521321769,
      "grad_norm": 1159.9666748046875,
      "learning_rate": 3.792787228217529e-05,
      "loss": 96.6831,
      "step": 57890
    },
    {
      "epoch": 0.2339233264785853,
      "grad_norm": 1524.46240234375,
      "learning_rate": 3.792663429494986e-05,
      "loss": 71.0455,
      "step": 57900
    },
    {
      "epoch": 0.23396372774395294,
      "grad_norm": 2358.887939453125,
      "learning_rate": 3.792539595823548e-05,
      "loss": 70.4084,
      "step": 57910
    },
    {
      "epoch": 0.23400412900932058,
      "grad_norm": 779.4794921875,
      "learning_rate": 3.792415727205627e-05,
      "loss": 73.8759,
      "step": 57920
    },
    {
      "epoch": 0.2340445302746882,
      "grad_norm": 1307.2642822265625,
      "learning_rate": 3.7922918236436384e-05,
      "loss": 107.4841,
      "step": 57930
    },
    {
      "epoch": 0.23408493154005583,
      "grad_norm": 1434.2913818359375,
      "learning_rate": 3.7921678851399984e-05,
      "loss": 56.1901,
      "step": 57940
    },
    {
      "epoch": 0.23412533280542347,
      "grad_norm": 524.6749877929688,
      "learning_rate": 3.792043911697122e-05,
      "loss": 66.964,
      "step": 57950
    },
    {
      "epoch": 0.2341657340707911,
      "grad_norm": 756.17822265625,
      "learning_rate": 3.791919903317428e-05,
      "loss": 55.6833,
      "step": 57960
    },
    {
      "epoch": 0.23420613533615872,
      "grad_norm": 943.431396484375,
      "learning_rate": 3.7917958600033326e-05,
      "loss": 85.9044,
      "step": 57970
    },
    {
      "epoch": 0.23424653660152636,
      "grad_norm": 856.8779907226562,
      "learning_rate": 3.791671781757254e-05,
      "loss": 53.0595,
      "step": 57980
    },
    {
      "epoch": 0.234286937866894,
      "grad_norm": 432.1573181152344,
      "learning_rate": 3.7915476685816124e-05,
      "loss": 55.1902,
      "step": 57990
    },
    {
      "epoch": 0.23432733913226161,
      "grad_norm": 1029.0245361328125,
      "learning_rate": 3.791423520478826e-05,
      "loss": 79.4546,
      "step": 58000
    },
    {
      "epoch": 0.23436774039762925,
      "grad_norm": 949.74609375,
      "learning_rate": 3.791299337451316e-05,
      "loss": 62.7363,
      "step": 58010
    },
    {
      "epoch": 0.2344081416629969,
      "grad_norm": 2479.85888671875,
      "learning_rate": 3.791175119501503e-05,
      "loss": 100.7351,
      "step": 58020
    },
    {
      "epoch": 0.2344485429283645,
      "grad_norm": 876.0036010742188,
      "learning_rate": 3.791050866631809e-05,
      "loss": 57.7869,
      "step": 58030
    },
    {
      "epoch": 0.23448894419373215,
      "grad_norm": 1590.4725341796875,
      "learning_rate": 3.790926578844657e-05,
      "loss": 109.7307,
      "step": 58040
    },
    {
      "epoch": 0.2345293454590998,
      "grad_norm": 633.0686645507812,
      "learning_rate": 3.790802256142468e-05,
      "loss": 72.1402,
      "step": 58050
    },
    {
      "epoch": 0.2345697467244674,
      "grad_norm": 700.5449829101562,
      "learning_rate": 3.790677898527668e-05,
      "loss": 106.0299,
      "step": 58060
    },
    {
      "epoch": 0.23461014798983504,
      "grad_norm": 851.0189208984375,
      "learning_rate": 3.79055350600268e-05,
      "loss": 84.5764,
      "step": 58070
    },
    {
      "epoch": 0.23465054925520268,
      "grad_norm": 615.2695922851562,
      "learning_rate": 3.79042907856993e-05,
      "loss": 88.8549,
      "step": 58080
    },
    {
      "epoch": 0.2346909505205703,
      "grad_norm": 552.6085205078125,
      "learning_rate": 3.790304616231843e-05,
      "loss": 81.1972,
      "step": 58090
    },
    {
      "epoch": 0.23473135178593793,
      "grad_norm": 307.0838928222656,
      "learning_rate": 3.790180118990845e-05,
      "loss": 48.328,
      "step": 58100
    },
    {
      "epoch": 0.23477175305130557,
      "grad_norm": 529.787841796875,
      "learning_rate": 3.7900555868493656e-05,
      "loss": 42.8784,
      "step": 58110
    },
    {
      "epoch": 0.2348121543166732,
      "grad_norm": 378.6830139160156,
      "learning_rate": 3.78993101980983e-05,
      "loss": 50.0545,
      "step": 58120
    },
    {
      "epoch": 0.23485255558204082,
      "grad_norm": 1031.341064453125,
      "learning_rate": 3.789806417874668e-05,
      "loss": 55.2613,
      "step": 58130
    },
    {
      "epoch": 0.23489295684740846,
      "grad_norm": 585.3533325195312,
      "learning_rate": 3.789681781046308e-05,
      "loss": 74.903,
      "step": 58140
    },
    {
      "epoch": 0.2349333581127761,
      "grad_norm": 5360.59130859375,
      "learning_rate": 3.789557109327181e-05,
      "loss": 74.3554,
      "step": 58150
    },
    {
      "epoch": 0.23497375937814372,
      "grad_norm": 848.1822509765625,
      "learning_rate": 3.789432402719716e-05,
      "loss": 66.7466,
      "step": 58160
    },
    {
      "epoch": 0.23501416064351136,
      "grad_norm": 967.7154541015625,
      "learning_rate": 3.7893076612263454e-05,
      "loss": 100.5138,
      "step": 58170
    },
    {
      "epoch": 0.235054561908879,
      "grad_norm": 565.66845703125,
      "learning_rate": 3.7891828848495006e-05,
      "loss": 49.7677,
      "step": 58180
    },
    {
      "epoch": 0.2350949631742466,
      "grad_norm": 437.8297424316406,
      "learning_rate": 3.789058073591615e-05,
      "loss": 60.5386,
      "step": 58190
    },
    {
      "epoch": 0.23513536443961425,
      "grad_norm": 997.5803833007812,
      "learning_rate": 3.78893322745512e-05,
      "loss": 66.6851,
      "step": 58200
    },
    {
      "epoch": 0.2351757657049819,
      "grad_norm": 1354.269775390625,
      "learning_rate": 3.7888083464424513e-05,
      "loss": 81.9496,
      "step": 58210
    },
    {
      "epoch": 0.2352161669703495,
      "grad_norm": 885.1353149414062,
      "learning_rate": 3.788683430556043e-05,
      "loss": 64.9704,
      "step": 58220
    },
    {
      "epoch": 0.23525656823571714,
      "grad_norm": 1185.8641357421875,
      "learning_rate": 3.78855847979833e-05,
      "loss": 71.7671,
      "step": 58230
    },
    {
      "epoch": 0.23529696950108478,
      "grad_norm": 1009.7171630859375,
      "learning_rate": 3.7884334941717494e-05,
      "loss": 113.77,
      "step": 58240
    },
    {
      "epoch": 0.2353373707664524,
      "grad_norm": 5088.71240234375,
      "learning_rate": 3.788308473678737e-05,
      "loss": 168.055,
      "step": 58250
    },
    {
      "epoch": 0.23537777203182003,
      "grad_norm": 506.412109375,
      "learning_rate": 3.78818341832173e-05,
      "loss": 90.3233,
      "step": 58260
    },
    {
      "epoch": 0.23541817329718767,
      "grad_norm": 486.355712890625,
      "learning_rate": 3.788058328103166e-05,
      "loss": 96.4967,
      "step": 58270
    },
    {
      "epoch": 0.2354585745625553,
      "grad_norm": 651.5347290039062,
      "learning_rate": 3.787933203025485e-05,
      "loss": 43.4667,
      "step": 58280
    },
    {
      "epoch": 0.23549897582792292,
      "grad_norm": 678.38330078125,
      "learning_rate": 3.787808043091126e-05,
      "loss": 79.4518,
      "step": 58290
    },
    {
      "epoch": 0.23553937709329056,
      "grad_norm": 248.24075317382812,
      "learning_rate": 3.787682848302528e-05,
      "loss": 80.3003,
      "step": 58300
    },
    {
      "epoch": 0.2355797783586582,
      "grad_norm": 1245.1475830078125,
      "learning_rate": 3.787557618662133e-05,
      "loss": 61.9618,
      "step": 58310
    },
    {
      "epoch": 0.23562017962402582,
      "grad_norm": 1014.1390380859375,
      "learning_rate": 3.787432354172381e-05,
      "loss": 67.9585,
      "step": 58320
    },
    {
      "epoch": 0.23566058088939346,
      "grad_norm": 3150.63330078125,
      "learning_rate": 3.787307054835716e-05,
      "loss": 93.1764,
      "step": 58330
    },
    {
      "epoch": 0.2357009821547611,
      "grad_norm": 898.7249145507812,
      "learning_rate": 3.7871817206545805e-05,
      "loss": 77.5884,
      "step": 58340
    },
    {
      "epoch": 0.2357413834201287,
      "grad_norm": 635.70166015625,
      "learning_rate": 3.787056351631416e-05,
      "loss": 66.2552,
      "step": 58350
    },
    {
      "epoch": 0.23578178468549635,
      "grad_norm": 3590.01318359375,
      "learning_rate": 3.786930947768668e-05,
      "loss": 104.977,
      "step": 58360
    },
    {
      "epoch": 0.235822185950864,
      "grad_norm": 490.37164306640625,
      "learning_rate": 3.7868055090687814e-05,
      "loss": 74.0476,
      "step": 58370
    },
    {
      "epoch": 0.2358625872162316,
      "grad_norm": 920.75146484375,
      "learning_rate": 3.786680035534202e-05,
      "loss": 65.383,
      "step": 58380
    },
    {
      "epoch": 0.23590298848159924,
      "grad_norm": 548.8573608398438,
      "learning_rate": 3.7865545271673744e-05,
      "loss": 66.729,
      "step": 58390
    },
    {
      "epoch": 0.23594338974696688,
      "grad_norm": 436.5902099609375,
      "learning_rate": 3.786428983970748e-05,
      "loss": 81.463,
      "step": 58400
    },
    {
      "epoch": 0.2359837910123345,
      "grad_norm": 1367.1328125,
      "learning_rate": 3.7863034059467676e-05,
      "loss": 92.1821,
      "step": 58410
    },
    {
      "epoch": 0.23602419227770213,
      "grad_norm": 478.23712158203125,
      "learning_rate": 3.786177793097883e-05,
      "loss": 71.7802,
      "step": 58420
    },
    {
      "epoch": 0.23606459354306977,
      "grad_norm": 1330.28076171875,
      "learning_rate": 3.786052145426543e-05,
      "loss": 66.2717,
      "step": 58430
    },
    {
      "epoch": 0.2361049948084374,
      "grad_norm": 682.5220947265625,
      "learning_rate": 3.7859264629351965e-05,
      "loss": 69.2333,
      "step": 58440
    },
    {
      "epoch": 0.23614539607380503,
      "grad_norm": 290.62176513671875,
      "learning_rate": 3.785800745626294e-05,
      "loss": 68.7359,
      "step": 58450
    },
    {
      "epoch": 0.23618579733917267,
      "grad_norm": 1322.16552734375,
      "learning_rate": 3.785674993502287e-05,
      "loss": 92.5002,
      "step": 58460
    },
    {
      "epoch": 0.2362261986045403,
      "grad_norm": 464.6538391113281,
      "learning_rate": 3.785549206565626e-05,
      "loss": 64.4816,
      "step": 58470
    },
    {
      "epoch": 0.23626659986990792,
      "grad_norm": 493.19158935546875,
      "learning_rate": 3.785423384818765e-05,
      "loss": 65.7234,
      "step": 58480
    },
    {
      "epoch": 0.23630700113527556,
      "grad_norm": 1342.987060546875,
      "learning_rate": 3.7852975282641555e-05,
      "loss": 72.9235,
      "step": 58490
    },
    {
      "epoch": 0.2363474024006432,
      "grad_norm": 1095.0755615234375,
      "learning_rate": 3.785171636904252e-05,
      "loss": 67.7989,
      "step": 58500
    },
    {
      "epoch": 0.2363878036660108,
      "grad_norm": 581.5772094726562,
      "learning_rate": 3.785045710741507e-05,
      "loss": 71.6452,
      "step": 58510
    },
    {
      "epoch": 0.23642820493137845,
      "grad_norm": 900.9180908203125,
      "learning_rate": 3.7849197497783775e-05,
      "loss": 72.144,
      "step": 58520
    },
    {
      "epoch": 0.2364686061967461,
      "grad_norm": 350.960693359375,
      "learning_rate": 3.784793754017319e-05,
      "loss": 89.0088,
      "step": 58530
    },
    {
      "epoch": 0.2365090074621137,
      "grad_norm": 0.0,
      "learning_rate": 3.7846677234607874e-05,
      "loss": 96.5052,
      "step": 58540
    },
    {
      "epoch": 0.23654940872748134,
      "grad_norm": 0.0,
      "learning_rate": 3.7845416581112394e-05,
      "loss": 51.9279,
      "step": 58550
    },
    {
      "epoch": 0.23658980999284898,
      "grad_norm": 3187.671142578125,
      "learning_rate": 3.784415557971133e-05,
      "loss": 93.1097,
      "step": 58560
    },
    {
      "epoch": 0.2366302112582166,
      "grad_norm": 1416.762939453125,
      "learning_rate": 3.784289423042927e-05,
      "loss": 94.1914,
      "step": 58570
    },
    {
      "epoch": 0.23667061252358423,
      "grad_norm": 884.7224731445312,
      "learning_rate": 3.784163253329079e-05,
      "loss": 98.9293,
      "step": 58580
    },
    {
      "epoch": 0.23671101378895187,
      "grad_norm": 1644.8016357421875,
      "learning_rate": 3.7840370488320514e-05,
      "loss": 139.2957,
      "step": 58590
    },
    {
      "epoch": 0.23675141505431951,
      "grad_norm": 820.919921875,
      "learning_rate": 3.7839108095543016e-05,
      "loss": 71.7127,
      "step": 58600
    },
    {
      "epoch": 0.23679181631968713,
      "grad_norm": 708.4010009765625,
      "learning_rate": 3.783784535498293e-05,
      "loss": 55.7633,
      "step": 58610
    },
    {
      "epoch": 0.23683221758505477,
      "grad_norm": 1582.913818359375,
      "learning_rate": 3.7836582266664864e-05,
      "loss": 92.6639,
      "step": 58620
    },
    {
      "epoch": 0.2368726188504224,
      "grad_norm": 834.7432861328125,
      "learning_rate": 3.783531883061345e-05,
      "loss": 82.2903,
      "step": 58630
    },
    {
      "epoch": 0.23691302011579002,
      "grad_norm": 605.028076171875,
      "learning_rate": 3.7834055046853297e-05,
      "loss": 65.3869,
      "step": 58640
    },
    {
      "epoch": 0.23695342138115766,
      "grad_norm": 2071.533447265625,
      "learning_rate": 3.783279091540907e-05,
      "loss": 76.7022,
      "step": 58650
    },
    {
      "epoch": 0.2369938226465253,
      "grad_norm": 768.4856567382812,
      "learning_rate": 3.783152643630541e-05,
      "loss": 56.1115,
      "step": 58660
    },
    {
      "epoch": 0.2370342239118929,
      "grad_norm": 617.2097778320312,
      "learning_rate": 3.783026160956695e-05,
      "loss": 56.9971,
      "step": 58670
    },
    {
      "epoch": 0.23707462517726055,
      "grad_norm": 665.7985229492188,
      "learning_rate": 3.782899643521836e-05,
      "loss": 64.8416,
      "step": 58680
    },
    {
      "epoch": 0.2371150264426282,
      "grad_norm": 860.6461181640625,
      "learning_rate": 3.782773091328431e-05,
      "loss": 54.3539,
      "step": 58690
    },
    {
      "epoch": 0.2371554277079958,
      "grad_norm": 1088.1025390625,
      "learning_rate": 3.782646504378947e-05,
      "loss": 55.3993,
      "step": 58700
    },
    {
      "epoch": 0.23719582897336344,
      "grad_norm": 916.010498046875,
      "learning_rate": 3.782519882675851e-05,
      "loss": 72.0945,
      "step": 58710
    },
    {
      "epoch": 0.23723623023873108,
      "grad_norm": 428.01141357421875,
      "learning_rate": 3.782393226221613e-05,
      "loss": 52.974,
      "step": 58720
    },
    {
      "epoch": 0.2372766315040987,
      "grad_norm": 1858.70703125,
      "learning_rate": 3.7822665350187006e-05,
      "loss": 111.1884,
      "step": 58730
    },
    {
      "epoch": 0.23731703276946634,
      "grad_norm": 1709.9642333984375,
      "learning_rate": 3.782139809069585e-05,
      "loss": 92.5917,
      "step": 58740
    },
    {
      "epoch": 0.23735743403483398,
      "grad_norm": 1273.9083251953125,
      "learning_rate": 3.782013048376736e-05,
      "loss": 75.8697,
      "step": 58750
    },
    {
      "epoch": 0.23739783530020162,
      "grad_norm": 816.621826171875,
      "learning_rate": 3.7818862529426255e-05,
      "loss": 91.2924,
      "step": 58760
    },
    {
      "epoch": 0.23743823656556923,
      "grad_norm": 566.1441650390625,
      "learning_rate": 3.781759422769725e-05,
      "loss": 61.1575,
      "step": 58770
    },
    {
      "epoch": 0.23747863783093687,
      "grad_norm": 368.1184997558594,
      "learning_rate": 3.781632557860507e-05,
      "loss": 55.6892,
      "step": 58780
    },
    {
      "epoch": 0.2375190390963045,
      "grad_norm": 1110.0902099609375,
      "learning_rate": 3.7815056582174455e-05,
      "loss": 83.4707,
      "step": 58790
    },
    {
      "epoch": 0.23755944036167212,
      "grad_norm": 765.1774291992188,
      "learning_rate": 3.781378723843014e-05,
      "loss": 77.6408,
      "step": 58800
    },
    {
      "epoch": 0.23759984162703976,
      "grad_norm": 1125.1318359375,
      "learning_rate": 3.781251754739686e-05,
      "loss": 83.5344,
      "step": 58810
    },
    {
      "epoch": 0.2376402428924074,
      "grad_norm": 615.6629028320312,
      "learning_rate": 3.781124750909939e-05,
      "loss": 51.5137,
      "step": 58820
    },
    {
      "epoch": 0.237680644157775,
      "grad_norm": 702.6609497070312,
      "learning_rate": 3.7809977123562486e-05,
      "loss": 82.915,
      "step": 58830
    },
    {
      "epoch": 0.23772104542314265,
      "grad_norm": 962.4164428710938,
      "learning_rate": 3.78087063908109e-05,
      "loss": 80.9654,
      "step": 58840
    },
    {
      "epoch": 0.2377614466885103,
      "grad_norm": 205.25384521484375,
      "learning_rate": 3.7807435310869415e-05,
      "loss": 87.7431,
      "step": 58850
    },
    {
      "epoch": 0.2378018479538779,
      "grad_norm": 573.6884765625,
      "learning_rate": 3.780616388376281e-05,
      "loss": 75.8221,
      "step": 58860
    },
    {
      "epoch": 0.23784224921924554,
      "grad_norm": 1432.1715087890625,
      "learning_rate": 3.780489210951588e-05,
      "loss": 77.8596,
      "step": 58870
    },
    {
      "epoch": 0.23788265048461318,
      "grad_norm": 437.041015625,
      "learning_rate": 3.7803619988153404e-05,
      "loss": 73.7691,
      "step": 58880
    },
    {
      "epoch": 0.2379230517499808,
      "grad_norm": 1854.2607421875,
      "learning_rate": 3.780234751970019e-05,
      "loss": 54.8822,
      "step": 58890
    },
    {
      "epoch": 0.23796345301534844,
      "grad_norm": 1220.783203125,
      "learning_rate": 3.7801074704181046e-05,
      "loss": 89.3865,
      "step": 58900
    },
    {
      "epoch": 0.23800385428071608,
      "grad_norm": 1539.205322265625,
      "learning_rate": 3.77998015416208e-05,
      "loss": 88.4137,
      "step": 58910
    },
    {
      "epoch": 0.23804425554608372,
      "grad_norm": 572.49609375,
      "learning_rate": 3.779852803204424e-05,
      "loss": 48.2826,
      "step": 58920
    },
    {
      "epoch": 0.23808465681145133,
      "grad_norm": 812.40966796875,
      "learning_rate": 3.779725417547622e-05,
      "loss": 79.3678,
      "step": 58930
    },
    {
      "epoch": 0.23812505807681897,
      "grad_norm": 553.4939575195312,
      "learning_rate": 3.7795979971941573e-05,
      "loss": 59.0748,
      "step": 58940
    },
    {
      "epoch": 0.2381654593421866,
      "grad_norm": 650.33251953125,
      "learning_rate": 3.779470542146513e-05,
      "loss": 61.1547,
      "step": 58950
    },
    {
      "epoch": 0.23820586060755422,
      "grad_norm": 1057.7850341796875,
      "learning_rate": 3.779343052407174e-05,
      "loss": 55.8551,
      "step": 58960
    },
    {
      "epoch": 0.23824626187292186,
      "grad_norm": 616.528076171875,
      "learning_rate": 3.779215527978626e-05,
      "loss": 77.092,
      "step": 58970
    },
    {
      "epoch": 0.2382866631382895,
      "grad_norm": 733.9642333984375,
      "learning_rate": 3.7790879688633565e-05,
      "loss": 81.81,
      "step": 58980
    },
    {
      "epoch": 0.2383270644036571,
      "grad_norm": 751.3402099609375,
      "learning_rate": 3.77896037506385e-05,
      "loss": 142.6629,
      "step": 58990
    },
    {
      "epoch": 0.23836746566902475,
      "grad_norm": 1286.316162109375,
      "learning_rate": 3.778832746582596e-05,
      "loss": 99.7319,
      "step": 59000
    },
    {
      "epoch": 0.2384078669343924,
      "grad_norm": 672.05712890625,
      "learning_rate": 3.778705083422081e-05,
      "loss": 77.2239,
      "step": 59010
    },
    {
      "epoch": 0.23844826819976,
      "grad_norm": 1207.39892578125,
      "learning_rate": 3.7785773855847944e-05,
      "loss": 64.5921,
      "step": 59020
    },
    {
      "epoch": 0.23848866946512765,
      "grad_norm": 1469.94580078125,
      "learning_rate": 3.7784496530732264e-05,
      "loss": 66.1509,
      "step": 59030
    },
    {
      "epoch": 0.23852907073049529,
      "grad_norm": 630.2105102539062,
      "learning_rate": 3.778321885889867e-05,
      "loss": 75.3852,
      "step": 59040
    },
    {
      "epoch": 0.2385694719958629,
      "grad_norm": 601.9887084960938,
      "learning_rate": 3.778194084037207e-05,
      "loss": 74.5009,
      "step": 59050
    },
    {
      "epoch": 0.23860987326123054,
      "grad_norm": 940.4525756835938,
      "learning_rate": 3.778066247517737e-05,
      "loss": 57.7681,
      "step": 59060
    },
    {
      "epoch": 0.23865027452659818,
      "grad_norm": 361.8607482910156,
      "learning_rate": 3.7779383763339505e-05,
      "loss": 57.5467,
      "step": 59070
    },
    {
      "epoch": 0.23869067579196582,
      "grad_norm": 383.443359375,
      "learning_rate": 3.7778104704883405e-05,
      "loss": 107.8757,
      "step": 59080
    },
    {
      "epoch": 0.23873107705733343,
      "grad_norm": 575.1685180664062,
      "learning_rate": 3.7776825299834e-05,
      "loss": 63.5811,
      "step": 59090
    },
    {
      "epoch": 0.23877147832270107,
      "grad_norm": 978.732177734375,
      "learning_rate": 3.777554554821623e-05,
      "loss": 88.5463,
      "step": 59100
    },
    {
      "epoch": 0.2388118795880687,
      "grad_norm": 629.1954956054688,
      "learning_rate": 3.777426545005505e-05,
      "loss": 57.9253,
      "step": 59110
    },
    {
      "epoch": 0.23885228085343632,
      "grad_norm": 402.6628723144531,
      "learning_rate": 3.777298500537542e-05,
      "loss": 104.7018,
      "step": 59120
    },
    {
      "epoch": 0.23889268211880396,
      "grad_norm": 416.7928161621094,
      "learning_rate": 3.7771704214202287e-05,
      "loss": 80.7156,
      "step": 59130
    },
    {
      "epoch": 0.2389330833841716,
      "grad_norm": 612.1146240234375,
      "learning_rate": 3.7770423076560635e-05,
      "loss": 63.2797,
      "step": 59140
    },
    {
      "epoch": 0.23897348464953921,
      "grad_norm": 779.5694580078125,
      "learning_rate": 3.776914159247544e-05,
      "loss": 79.3714,
      "step": 59150
    },
    {
      "epoch": 0.23901388591490685,
      "grad_norm": 450.3630676269531,
      "learning_rate": 3.776785976197168e-05,
      "loss": 64.3583,
      "step": 59160
    },
    {
      "epoch": 0.2390542871802745,
      "grad_norm": 236.486328125,
      "learning_rate": 3.776657758507434e-05,
      "loss": 66.5476,
      "step": 59170
    },
    {
      "epoch": 0.2390946884456421,
      "grad_norm": 1080.7608642578125,
      "learning_rate": 3.776529506180843e-05,
      "loss": 39.3434,
      "step": 59180
    },
    {
      "epoch": 0.23913508971100975,
      "grad_norm": 1653.2723388671875,
      "learning_rate": 3.776401219219894e-05,
      "loss": 84.2639,
      "step": 59190
    },
    {
      "epoch": 0.2391754909763774,
      "grad_norm": 595.5156860351562,
      "learning_rate": 3.7762728976270897e-05,
      "loss": 73.5355,
      "step": 59200
    },
    {
      "epoch": 0.239215892241745,
      "grad_norm": 1844.938232421875,
      "learning_rate": 3.77614454140493e-05,
      "loss": 73.5855,
      "step": 59210
    },
    {
      "epoch": 0.23925629350711264,
      "grad_norm": 553.8212890625,
      "learning_rate": 3.776016150555918e-05,
      "loss": 61.2807,
      "step": 59220
    },
    {
      "epoch": 0.23929669477248028,
      "grad_norm": 887.1659545898438,
      "learning_rate": 3.775887725082557e-05,
      "loss": 56.4401,
      "step": 59230
    },
    {
      "epoch": 0.23933709603784792,
      "grad_norm": 925.3824462890625,
      "learning_rate": 3.7757592649873503e-05,
      "loss": 97.8124,
      "step": 59240
    },
    {
      "epoch": 0.23937749730321553,
      "grad_norm": 669.0916748046875,
      "learning_rate": 3.7756307702728026e-05,
      "loss": 83.3682,
      "step": 59250
    },
    {
      "epoch": 0.23941789856858317,
      "grad_norm": 510.4156799316406,
      "learning_rate": 3.7755022409414195e-05,
      "loss": 74.0489,
      "step": 59260
    },
    {
      "epoch": 0.2394582998339508,
      "grad_norm": 832.69580078125,
      "learning_rate": 3.775373676995705e-05,
      "loss": 55.0958,
      "step": 59270
    },
    {
      "epoch": 0.23949870109931842,
      "grad_norm": 1248.660888671875,
      "learning_rate": 3.775245078438168e-05,
      "loss": 59.6337,
      "step": 59280
    },
    {
      "epoch": 0.23953910236468606,
      "grad_norm": 962.0632934570312,
      "learning_rate": 3.775116445271313e-05,
      "loss": 54.9472,
      "step": 59290
    },
    {
      "epoch": 0.2395795036300537,
      "grad_norm": 1186.34130859375,
      "learning_rate": 3.77498777749765e-05,
      "loss": 73.6589,
      "step": 59300
    },
    {
      "epoch": 0.23961990489542132,
      "grad_norm": 554.897705078125,
      "learning_rate": 3.7748590751196854e-05,
      "loss": 77.572,
      "step": 59310
    },
    {
      "epoch": 0.23966030616078896,
      "grad_norm": 1017.129638671875,
      "learning_rate": 3.7747303381399304e-05,
      "loss": 76.4657,
      "step": 59320
    },
    {
      "epoch": 0.2397007074261566,
      "grad_norm": 3422.4931640625,
      "learning_rate": 3.7746015665608934e-05,
      "loss": 89.2775,
      "step": 59330
    },
    {
      "epoch": 0.2397411086915242,
      "grad_norm": 2242.440185546875,
      "learning_rate": 3.774472760385085e-05,
      "loss": 99.2422,
      "step": 59340
    },
    {
      "epoch": 0.23978150995689185,
      "grad_norm": 636.38134765625,
      "learning_rate": 3.7743439196150166e-05,
      "loss": 74.0525,
      "step": 59350
    },
    {
      "epoch": 0.2398219112222595,
      "grad_norm": 733.93701171875,
      "learning_rate": 3.774215044253201e-05,
      "loss": 69.8146,
      "step": 59360
    },
    {
      "epoch": 0.2398623124876271,
      "grad_norm": 1365.994140625,
      "learning_rate": 3.774086134302148e-05,
      "loss": 96.8566,
      "step": 59370
    },
    {
      "epoch": 0.23990271375299474,
      "grad_norm": 828.5872192382812,
      "learning_rate": 3.773957189764373e-05,
      "loss": 66.3401,
      "step": 59380
    },
    {
      "epoch": 0.23994311501836238,
      "grad_norm": 607.1014404296875,
      "learning_rate": 3.77382821064239e-05,
      "loss": 93.6759,
      "step": 59390
    },
    {
      "epoch": 0.23998351628373002,
      "grad_norm": 1687.7762451171875,
      "learning_rate": 3.773699196938712e-05,
      "loss": 89.2844,
      "step": 59400
    },
    {
      "epoch": 0.24002391754909763,
      "grad_norm": 822.9414672851562,
      "learning_rate": 3.7735701486558555e-05,
      "loss": 45.2697,
      "step": 59410
    },
    {
      "epoch": 0.24006431881446527,
      "grad_norm": 473.8529357910156,
      "learning_rate": 3.773441065796335e-05,
      "loss": 57.0944,
      "step": 59420
    },
    {
      "epoch": 0.2401047200798329,
      "grad_norm": 1023.8804321289062,
      "learning_rate": 3.7733119483626694e-05,
      "loss": 62.7734,
      "step": 59430
    },
    {
      "epoch": 0.24014512134520052,
      "grad_norm": 525.4328002929688,
      "learning_rate": 3.7731827963573734e-05,
      "loss": 83.7604,
      "step": 59440
    },
    {
      "epoch": 0.24018552261056816,
      "grad_norm": 706.7175903320312,
      "learning_rate": 3.7730536097829655e-05,
      "loss": 54.1134,
      "step": 59450
    },
    {
      "epoch": 0.2402259238759358,
      "grad_norm": 956.0654907226562,
      "learning_rate": 3.7729243886419656e-05,
      "loss": 57.8804,
      "step": 59460
    },
    {
      "epoch": 0.24026632514130342,
      "grad_norm": 444.929931640625,
      "learning_rate": 3.772795132936891e-05,
      "loss": 42.6263,
      "step": 59470
    },
    {
      "epoch": 0.24030672640667106,
      "grad_norm": 1101.1414794921875,
      "learning_rate": 3.7726658426702636e-05,
      "loss": 91.3343,
      "step": 59480
    },
    {
      "epoch": 0.2403471276720387,
      "grad_norm": 1087.0360107421875,
      "learning_rate": 3.7725365178446024e-05,
      "loss": 68.0614,
      "step": 59490
    },
    {
      "epoch": 0.2403875289374063,
      "grad_norm": 806.5647583007812,
      "learning_rate": 3.7724071584624296e-05,
      "loss": 70.7971,
      "step": 59500
    },
    {
      "epoch": 0.24042793020277395,
      "grad_norm": 911.4544067382812,
      "learning_rate": 3.772277764526267e-05,
      "loss": 85.7331,
      "step": 59510
    },
    {
      "epoch": 0.2404683314681416,
      "grad_norm": 469.47430419921875,
      "learning_rate": 3.772148336038636e-05,
      "loss": 51.1957,
      "step": 59520
    },
    {
      "epoch": 0.2405087327335092,
      "grad_norm": 324.8692932128906,
      "learning_rate": 3.772018873002061e-05,
      "loss": 67.4201,
      "step": 59530
    },
    {
      "epoch": 0.24054913399887684,
      "grad_norm": 585.5924682617188,
      "learning_rate": 3.7718893754190665e-05,
      "loss": 80.5849,
      "step": 59540
    },
    {
      "epoch": 0.24058953526424448,
      "grad_norm": 538.8850708007812,
      "learning_rate": 3.7717598432921766e-05,
      "loss": 61.0507,
      "step": 59550
    },
    {
      "epoch": 0.24062993652961212,
      "grad_norm": 716.21044921875,
      "learning_rate": 3.771630276623915e-05,
      "loss": 49.5787,
      "step": 59560
    },
    {
      "epoch": 0.24067033779497973,
      "grad_norm": 728.6638793945312,
      "learning_rate": 3.77150067541681e-05,
      "loss": 65.5644,
      "step": 59570
    },
    {
      "epoch": 0.24071073906034737,
      "grad_norm": 380.37115478515625,
      "learning_rate": 3.7713710396733866e-05,
      "loss": 53.3074,
      "step": 59580
    },
    {
      "epoch": 0.240751140325715,
      "grad_norm": 418.3934326171875,
      "learning_rate": 3.771241369396174e-05,
      "loss": 61.8226,
      "step": 59590
    },
    {
      "epoch": 0.24079154159108263,
      "grad_norm": 672.7009887695312,
      "learning_rate": 3.7711116645876984e-05,
      "loss": 57.1527,
      "step": 59600
    },
    {
      "epoch": 0.24083194285645027,
      "grad_norm": 681.2256469726562,
      "learning_rate": 3.770981925250489e-05,
      "loss": 98.6313,
      "step": 59610
    },
    {
      "epoch": 0.2408723441218179,
      "grad_norm": 658.5431518554688,
      "learning_rate": 3.7708521513870756e-05,
      "loss": 74.5584,
      "step": 59620
    },
    {
      "epoch": 0.24091274538718552,
      "grad_norm": 426.46783447265625,
      "learning_rate": 3.7707223429999874e-05,
      "loss": 67.2407,
      "step": 59630
    },
    {
      "epoch": 0.24095314665255316,
      "grad_norm": 1107.43310546875,
      "learning_rate": 3.770592500091755e-05,
      "loss": 66.4551,
      "step": 59640
    },
    {
      "epoch": 0.2409935479179208,
      "grad_norm": 346.4035949707031,
      "learning_rate": 3.7704626226649106e-05,
      "loss": 80.4863,
      "step": 59650
    },
    {
      "epoch": 0.2410339491832884,
      "grad_norm": 631.7588500976562,
      "learning_rate": 3.7703327107219866e-05,
      "loss": 79.9864,
      "step": 59660
    },
    {
      "epoch": 0.24107435044865605,
      "grad_norm": 560.8602294921875,
      "learning_rate": 3.770202764265514e-05,
      "loss": 64.0211,
      "step": 59670
    },
    {
      "epoch": 0.2411147517140237,
      "grad_norm": 664.188232421875,
      "learning_rate": 3.7700727832980275e-05,
      "loss": 56.0255,
      "step": 59680
    },
    {
      "epoch": 0.2411551529793913,
      "grad_norm": 456.9137268066406,
      "learning_rate": 3.769942767822061e-05,
      "loss": 69.6144,
      "step": 59690
    },
    {
      "epoch": 0.24119555424475894,
      "grad_norm": 1441.332763671875,
      "learning_rate": 3.769812717840149e-05,
      "loss": 49.5474,
      "step": 59700
    },
    {
      "epoch": 0.24123595551012658,
      "grad_norm": 893.2605590820312,
      "learning_rate": 3.7696826333548265e-05,
      "loss": 83.8505,
      "step": 59710
    },
    {
      "epoch": 0.24127635677549422,
      "grad_norm": 1629.607421875,
      "learning_rate": 3.7695525143686305e-05,
      "loss": 84.8785,
      "step": 59720
    },
    {
      "epoch": 0.24131675804086183,
      "grad_norm": 472.76495361328125,
      "learning_rate": 3.7694223608840966e-05,
      "loss": 75.3418,
      "step": 59730
    },
    {
      "epoch": 0.24135715930622947,
      "grad_norm": 1648.9901123046875,
      "learning_rate": 3.7692921729037636e-05,
      "loss": 71.5154,
      "step": 59740
    },
    {
      "epoch": 0.24139756057159711,
      "grad_norm": 602.0360717773438,
      "learning_rate": 3.769161950430168e-05,
      "loss": 67.145,
      "step": 59750
    },
    {
      "epoch": 0.24143796183696473,
      "grad_norm": 299.7599182128906,
      "learning_rate": 3.7690316934658497e-05,
      "loss": 53.0463,
      "step": 59760
    },
    {
      "epoch": 0.24147836310233237,
      "grad_norm": 681.124755859375,
      "learning_rate": 3.768901402013348e-05,
      "loss": 85.4751,
      "step": 59770
    },
    {
      "epoch": 0.2415187643677,
      "grad_norm": 942.6168212890625,
      "learning_rate": 3.7687710760752026e-05,
      "loss": 92.2869,
      "step": 59780
    },
    {
      "epoch": 0.24155916563306762,
      "grad_norm": 723.7786865234375,
      "learning_rate": 3.768640715653955e-05,
      "loss": 85.5628,
      "step": 59790
    },
    {
      "epoch": 0.24159956689843526,
      "grad_norm": 0.0,
      "learning_rate": 3.768510320752145e-05,
      "loss": 59.9159,
      "step": 59800
    },
    {
      "epoch": 0.2416399681638029,
      "grad_norm": 2249.366455078125,
      "learning_rate": 3.768379891372316e-05,
      "loss": 72.3002,
      "step": 59810
    },
    {
      "epoch": 0.2416803694291705,
      "grad_norm": 956.3267822265625,
      "learning_rate": 3.768249427517011e-05,
      "loss": 44.7645,
      "step": 59820
    },
    {
      "epoch": 0.24172077069453815,
      "grad_norm": 4947.322265625,
      "learning_rate": 3.7681189291887734e-05,
      "loss": 111.1965,
      "step": 59830
    },
    {
      "epoch": 0.2417611719599058,
      "grad_norm": 1276.848876953125,
      "learning_rate": 3.767988396390146e-05,
      "loss": 64.6901,
      "step": 59840
    },
    {
      "epoch": 0.2418015732252734,
      "grad_norm": 1012.7622680664062,
      "learning_rate": 3.7678578291236756e-05,
      "loss": 79.9848,
      "step": 59850
    },
    {
      "epoch": 0.24184197449064104,
      "grad_norm": 789.6127319335938,
      "learning_rate": 3.767727227391906e-05,
      "loss": 89.3822,
      "step": 59860
    },
    {
      "epoch": 0.24188237575600868,
      "grad_norm": 1364.601318359375,
      "learning_rate": 3.7675965911973846e-05,
      "loss": 84.0104,
      "step": 59870
    },
    {
      "epoch": 0.24192277702137632,
      "grad_norm": 1051.7496337890625,
      "learning_rate": 3.767465920542657e-05,
      "loss": 95.6532,
      "step": 59880
    },
    {
      "epoch": 0.24196317828674394,
      "grad_norm": 976.9489135742188,
      "learning_rate": 3.767335215430271e-05,
      "loss": 83.0921,
      "step": 59890
    },
    {
      "epoch": 0.24200357955211158,
      "grad_norm": 846.85107421875,
      "learning_rate": 3.767204475862777e-05,
      "loss": 79.4891,
      "step": 59900
    },
    {
      "epoch": 0.24204398081747922,
      "grad_norm": 667.676025390625,
      "learning_rate": 3.76707370184272e-05,
      "loss": 75.6804,
      "step": 59910
    },
    {
      "epoch": 0.24208438208284683,
      "grad_norm": 1053.187255859375,
      "learning_rate": 3.766942893372652e-05,
      "loss": 96.0024,
      "step": 59920
    },
    {
      "epoch": 0.24212478334821447,
      "grad_norm": 2702.04736328125,
      "learning_rate": 3.766812050455123e-05,
      "loss": 77.5523,
      "step": 59930
    },
    {
      "epoch": 0.2421651846135821,
      "grad_norm": 1904.9150390625,
      "learning_rate": 3.766681173092683e-05,
      "loss": 107.33,
      "step": 59940
    },
    {
      "epoch": 0.24220558587894972,
      "grad_norm": 3805.7177734375,
      "learning_rate": 3.766550261287884e-05,
      "loss": 91.3345,
      "step": 59950
    },
    {
      "epoch": 0.24224598714431736,
      "grad_norm": 913.11328125,
      "learning_rate": 3.766419315043278e-05,
      "loss": 84.3102,
      "step": 59960
    },
    {
      "epoch": 0.242286388409685,
      "grad_norm": 679.9208374023438,
      "learning_rate": 3.7662883343614184e-05,
      "loss": 68.6057,
      "step": 59970
    },
    {
      "epoch": 0.2423267896750526,
      "grad_norm": 1311.968505859375,
      "learning_rate": 3.766157319244858e-05,
      "loss": 107.963,
      "step": 59980
    },
    {
      "epoch": 0.24236719094042025,
      "grad_norm": 674.0853271484375,
      "learning_rate": 3.766026269696152e-05,
      "loss": 83.4591,
      "step": 59990
    },
    {
      "epoch": 0.2424075922057879,
      "grad_norm": 759.6544189453125,
      "learning_rate": 3.7658951857178544e-05,
      "loss": 91.7942,
      "step": 60000
    },
    {
      "epoch": 0.2424479934711555,
      "grad_norm": 4039.81005859375,
      "learning_rate": 3.765764067312521e-05,
      "loss": 82.161,
      "step": 60010
    },
    {
      "epoch": 0.24248839473652314,
      "grad_norm": 800.8172607421875,
      "learning_rate": 3.7656329144827076e-05,
      "loss": 68.165,
      "step": 60020
    },
    {
      "epoch": 0.24252879600189078,
      "grad_norm": 579.8431396484375,
      "learning_rate": 3.765501727230972e-05,
      "loss": 88.252,
      "step": 60030
    },
    {
      "epoch": 0.24256919726725842,
      "grad_norm": 625.5407104492188,
      "learning_rate": 3.765370505559871e-05,
      "loss": 57.5803,
      "step": 60040
    },
    {
      "epoch": 0.24260959853262604,
      "grad_norm": 679.8961791992188,
      "learning_rate": 3.765239249471964e-05,
      "loss": 65.6697,
      "step": 60050
    },
    {
      "epoch": 0.24264999979799368,
      "grad_norm": 380.233154296875,
      "learning_rate": 3.7651079589698075e-05,
      "loss": 66.8145,
      "step": 60060
    },
    {
      "epoch": 0.24269040106336132,
      "grad_norm": 1168.4925537109375,
      "learning_rate": 3.764976634055963e-05,
      "loss": 91.2174,
      "step": 60070
    },
    {
      "epoch": 0.24273080232872893,
      "grad_norm": 979.0671997070312,
      "learning_rate": 3.764845274732992e-05,
      "loss": 86.8945,
      "step": 60080
    },
    {
      "epoch": 0.24277120359409657,
      "grad_norm": 587.3710327148438,
      "learning_rate": 3.7647138810034526e-05,
      "loss": 88.6826,
      "step": 60090
    },
    {
      "epoch": 0.2428116048594642,
      "grad_norm": 796.7279663085938,
      "learning_rate": 3.764582452869907e-05,
      "loss": 78.2054,
      "step": 60100
    },
    {
      "epoch": 0.24285200612483182,
      "grad_norm": 1938.2862548828125,
      "learning_rate": 3.7644509903349186e-05,
      "loss": 69.0388,
      "step": 60110
    },
    {
      "epoch": 0.24289240739019946,
      "grad_norm": 934.1618041992188,
      "learning_rate": 3.7643194934010494e-05,
      "loss": 87.1884,
      "step": 60120
    },
    {
      "epoch": 0.2429328086555671,
      "grad_norm": 586.398193359375,
      "learning_rate": 3.7641879620708636e-05,
      "loss": 80.0338,
      "step": 60130
    },
    {
      "epoch": 0.2429732099209347,
      "grad_norm": 738.1567993164062,
      "learning_rate": 3.764056396346925e-05,
      "loss": 67.8395,
      "step": 60140
    },
    {
      "epoch": 0.24301361118630235,
      "grad_norm": 1113.212890625,
      "learning_rate": 3.763924796231799e-05,
      "loss": 73.3919,
      "step": 60150
    },
    {
      "epoch": 0.24305401245167,
      "grad_norm": 586.4918823242188,
      "learning_rate": 3.763793161728051e-05,
      "loss": 53.713,
      "step": 60160
    },
    {
      "epoch": 0.2430944137170376,
      "grad_norm": 1516.4075927734375,
      "learning_rate": 3.763661492838247e-05,
      "loss": 58.0344,
      "step": 60170
    },
    {
      "epoch": 0.24313481498240525,
      "grad_norm": 791.6630249023438,
      "learning_rate": 3.763529789564955e-05,
      "loss": 62.3867,
      "step": 60180
    },
    {
      "epoch": 0.24317521624777289,
      "grad_norm": 966.2651977539062,
      "learning_rate": 3.7633980519107406e-05,
      "loss": 79.9166,
      "step": 60190
    },
    {
      "epoch": 0.24321561751314053,
      "grad_norm": 635.9515380859375,
      "learning_rate": 3.763266279878174e-05,
      "loss": 92.245,
      "step": 60200
    },
    {
      "epoch": 0.24325601877850814,
      "grad_norm": 775.6006469726562,
      "learning_rate": 3.763134473469824e-05,
      "loss": 55.5144,
      "step": 60210
    },
    {
      "epoch": 0.24329642004387578,
      "grad_norm": 683.8140258789062,
      "learning_rate": 3.7630026326882586e-05,
      "loss": 46.8155,
      "step": 60220
    },
    {
      "epoch": 0.24333682130924342,
      "grad_norm": 348.5704345703125,
      "learning_rate": 3.76287075753605e-05,
      "loss": 97.6181,
      "step": 60230
    },
    {
      "epoch": 0.24337722257461103,
      "grad_norm": 3307.7119140625,
      "learning_rate": 3.762738848015768e-05,
      "loss": 58.3651,
      "step": 60240
    },
    {
      "epoch": 0.24341762383997867,
      "grad_norm": 851.7776489257812,
      "learning_rate": 3.7626069041299847e-05,
      "loss": 71.8492,
      "step": 60250
    },
    {
      "epoch": 0.2434580251053463,
      "grad_norm": 983.8527221679688,
      "learning_rate": 3.7624749258812726e-05,
      "loss": 99.8106,
      "step": 60260
    },
    {
      "epoch": 0.24349842637071392,
      "grad_norm": 529.5291748046875,
      "learning_rate": 3.762342913272204e-05,
      "loss": 58.98,
      "step": 60270
    },
    {
      "epoch": 0.24353882763608156,
      "grad_norm": 689.0316772460938,
      "learning_rate": 3.762210866305354e-05,
      "loss": 78.8336,
      "step": 60280
    },
    {
      "epoch": 0.2435792289014492,
      "grad_norm": 468.4165344238281,
      "learning_rate": 3.762078784983294e-05,
      "loss": 52.1385,
      "step": 60290
    },
    {
      "epoch": 0.24361963016681681,
      "grad_norm": 794.3234252929688,
      "learning_rate": 3.7619466693086025e-05,
      "loss": 70.7858,
      "step": 60300
    },
    {
      "epoch": 0.24366003143218445,
      "grad_norm": 526.5389404296875,
      "learning_rate": 3.761814519283853e-05,
      "loss": 52.2118,
      "step": 60310
    },
    {
      "epoch": 0.2437004326975521,
      "grad_norm": 689.9909057617188,
      "learning_rate": 3.761682334911623e-05,
      "loss": 52.7487,
      "step": 60320
    },
    {
      "epoch": 0.2437408339629197,
      "grad_norm": 543.2650146484375,
      "learning_rate": 3.761550116194488e-05,
      "loss": 72.3932,
      "step": 60330
    },
    {
      "epoch": 0.24378123522828735,
      "grad_norm": 622.240966796875,
      "learning_rate": 3.7614178631350274e-05,
      "loss": 56.8141,
      "step": 60340
    },
    {
      "epoch": 0.243821636493655,
      "grad_norm": 674.28369140625,
      "learning_rate": 3.761285575735818e-05,
      "loss": 63.2067,
      "step": 60350
    },
    {
      "epoch": 0.24386203775902263,
      "grad_norm": 1530.883544921875,
      "learning_rate": 3.7611532539994405e-05,
      "loss": 91.0302,
      "step": 60360
    },
    {
      "epoch": 0.24390243902439024,
      "grad_norm": 751.9327392578125,
      "learning_rate": 3.7610208979284724e-05,
      "loss": 61.8245,
      "step": 60370
    },
    {
      "epoch": 0.24394284028975788,
      "grad_norm": 934.640869140625,
      "learning_rate": 3.7608885075254965e-05,
      "loss": 61.9268,
      "step": 60380
    },
    {
      "epoch": 0.24398324155512552,
      "grad_norm": 917.923583984375,
      "learning_rate": 3.760756082793092e-05,
      "loss": 55.6999,
      "step": 60390
    },
    {
      "epoch": 0.24402364282049313,
      "grad_norm": 518.9017944335938,
      "learning_rate": 3.7606236237338406e-05,
      "loss": 57.7457,
      "step": 60400
    },
    {
      "epoch": 0.24406404408586077,
      "grad_norm": 589.4769287109375,
      "learning_rate": 3.7604911303503255e-05,
      "loss": 45.9069,
      "step": 60410
    },
    {
      "epoch": 0.2441044453512284,
      "grad_norm": 930.5416870117188,
      "learning_rate": 3.7603586026451296e-05,
      "loss": 77.5556,
      "step": 60420
    },
    {
      "epoch": 0.24414484661659602,
      "grad_norm": 1381.807373046875,
      "learning_rate": 3.760226040620837e-05,
      "loss": 113.4836,
      "step": 60430
    },
    {
      "epoch": 0.24418524788196366,
      "grad_norm": 783.2221069335938,
      "learning_rate": 3.760093444280031e-05,
      "loss": 62.2471,
      "step": 60440
    },
    {
      "epoch": 0.2442256491473313,
      "grad_norm": 676.7279663085938,
      "learning_rate": 3.7599608136252975e-05,
      "loss": 82.4463,
      "step": 60450
    },
    {
      "epoch": 0.24426605041269892,
      "grad_norm": 639.694091796875,
      "learning_rate": 3.759828148659221e-05,
      "loss": 74.3016,
      "step": 60460
    },
    {
      "epoch": 0.24430645167806656,
      "grad_norm": 464.94781494140625,
      "learning_rate": 3.759695449384389e-05,
      "loss": 50.5467,
      "step": 60470
    },
    {
      "epoch": 0.2443468529434342,
      "grad_norm": 1204.313720703125,
      "learning_rate": 3.7595627158033895e-05,
      "loss": 68.3696,
      "step": 60480
    },
    {
      "epoch": 0.2443872542088018,
      "grad_norm": 702.4818725585938,
      "learning_rate": 3.759429947918808e-05,
      "loss": 89.8446,
      "step": 60490
    },
    {
      "epoch": 0.24442765547416945,
      "grad_norm": 854.7379760742188,
      "learning_rate": 3.759297145733234e-05,
      "loss": 65.2083,
      "step": 60500
    },
    {
      "epoch": 0.2444680567395371,
      "grad_norm": 435.7986145019531,
      "learning_rate": 3.7591643092492554e-05,
      "loss": 49.3518,
      "step": 60510
    },
    {
      "epoch": 0.24450845800490473,
      "grad_norm": 1312.701171875,
      "learning_rate": 3.759031438469464e-05,
      "loss": 85.8196,
      "step": 60520
    },
    {
      "epoch": 0.24454885927027234,
      "grad_norm": 1406.1134033203125,
      "learning_rate": 3.7588985333964486e-05,
      "loss": 64.5632,
      "step": 60530
    },
    {
      "epoch": 0.24458926053563998,
      "grad_norm": 985.60400390625,
      "learning_rate": 3.758765594032801e-05,
      "loss": 117.3803,
      "step": 60540
    },
    {
      "epoch": 0.24462966180100762,
      "grad_norm": 1442.556396484375,
      "learning_rate": 3.7586326203811124e-05,
      "loss": 108.1851,
      "step": 60550
    },
    {
      "epoch": 0.24467006306637523,
      "grad_norm": 749.6279296875,
      "learning_rate": 3.758499612443976e-05,
      "loss": 72.7907,
      "step": 60560
    },
    {
      "epoch": 0.24471046433174287,
      "grad_norm": 360.9629821777344,
      "learning_rate": 3.758366570223984e-05,
      "loss": 41.274,
      "step": 60570
    },
    {
      "epoch": 0.2447508655971105,
      "grad_norm": 1088.8790283203125,
      "learning_rate": 3.758233493723731e-05,
      "loss": 52.9848,
      "step": 60580
    },
    {
      "epoch": 0.24479126686247812,
      "grad_norm": 662.2022705078125,
      "learning_rate": 3.7581003829458104e-05,
      "loss": 78.0779,
      "step": 60590
    },
    {
      "epoch": 0.24483166812784576,
      "grad_norm": 636.2974853515625,
      "learning_rate": 3.757967237892818e-05,
      "loss": 55.419,
      "step": 60600
    },
    {
      "epoch": 0.2448720693932134,
      "grad_norm": 622.9697265625,
      "learning_rate": 3.757834058567348e-05,
      "loss": 63.7335,
      "step": 60610
    },
    {
      "epoch": 0.24491247065858102,
      "grad_norm": 291.7939758300781,
      "learning_rate": 3.757700844971999e-05,
      "loss": 32.439,
      "step": 60620
    },
    {
      "epoch": 0.24495287192394866,
      "grad_norm": 421.67889404296875,
      "learning_rate": 3.7575675971093674e-05,
      "loss": 107.8974,
      "step": 60630
    },
    {
      "epoch": 0.2449932731893163,
      "grad_norm": 681.85595703125,
      "learning_rate": 3.75743431498205e-05,
      "loss": 92.9533,
      "step": 60640
    },
    {
      "epoch": 0.2450336744546839,
      "grad_norm": 633.4293823242188,
      "learning_rate": 3.757300998592646e-05,
      "loss": 51.9228,
      "step": 60650
    },
    {
      "epoch": 0.24507407572005155,
      "grad_norm": 600.5732421875,
      "learning_rate": 3.757167647943755e-05,
      "loss": 38.8882,
      "step": 60660
    },
    {
      "epoch": 0.2451144769854192,
      "grad_norm": 597.346435546875,
      "learning_rate": 3.757034263037975e-05,
      "loss": 79.2042,
      "step": 60670
    },
    {
      "epoch": 0.24515487825078683,
      "grad_norm": 1310.50537109375,
      "learning_rate": 3.756900843877908e-05,
      "loss": 79.4892,
      "step": 60680
    },
    {
      "epoch": 0.24519527951615444,
      "grad_norm": 1101.57861328125,
      "learning_rate": 3.756767390466154e-05,
      "loss": 72.6483,
      "step": 60690
    },
    {
      "epoch": 0.24523568078152208,
      "grad_norm": 1607.89599609375,
      "learning_rate": 3.756633902805316e-05,
      "loss": 89.7742,
      "step": 60700
    },
    {
      "epoch": 0.24527608204688972,
      "grad_norm": 901.9148559570312,
      "learning_rate": 3.756500380897995e-05,
      "loss": 69.8062,
      "step": 60710
    },
    {
      "epoch": 0.24531648331225733,
      "grad_norm": 1015.1190795898438,
      "learning_rate": 3.756366824746795e-05,
      "loss": 97.6566,
      "step": 60720
    },
    {
      "epoch": 0.24535688457762497,
      "grad_norm": 1168.988037109375,
      "learning_rate": 3.75623323435432e-05,
      "loss": 83.2587,
      "step": 60730
    },
    {
      "epoch": 0.2453972858429926,
      "grad_norm": 1274.6507568359375,
      "learning_rate": 3.7560996097231736e-05,
      "loss": 75.8513,
      "step": 60740
    },
    {
      "epoch": 0.24543768710836023,
      "grad_norm": 1174.125,
      "learning_rate": 3.755965950855961e-05,
      "loss": 68.9883,
      "step": 60750
    },
    {
      "epoch": 0.24547808837372787,
      "grad_norm": 1353.937744140625,
      "learning_rate": 3.755832257755289e-05,
      "loss": 57.9202,
      "step": 60760
    },
    {
      "epoch": 0.2455184896390955,
      "grad_norm": 1133.187744140625,
      "learning_rate": 3.7556985304237625e-05,
      "loss": 108.1759,
      "step": 60770
    },
    {
      "epoch": 0.24555889090446312,
      "grad_norm": 664.1755981445312,
      "learning_rate": 3.755564768863989e-05,
      "loss": 71.9331,
      "step": 60780
    },
    {
      "epoch": 0.24559929216983076,
      "grad_norm": 812.7107543945312,
      "learning_rate": 3.7554309730785765e-05,
      "loss": 59.1399,
      "step": 60790
    },
    {
      "epoch": 0.2456396934351984,
      "grad_norm": 2424.87744140625,
      "learning_rate": 3.7552971430701344e-05,
      "loss": 72.2935,
      "step": 60800
    },
    {
      "epoch": 0.245680094700566,
      "grad_norm": 613.9686889648438,
      "learning_rate": 3.75516327884127e-05,
      "loss": 58.8767,
      "step": 60810
    },
    {
      "epoch": 0.24572049596593365,
      "grad_norm": 1093.13525390625,
      "learning_rate": 3.755029380394594e-05,
      "loss": 84.119,
      "step": 60820
    },
    {
      "epoch": 0.2457608972313013,
      "grad_norm": 906.2931518554688,
      "learning_rate": 3.7548954477327166e-05,
      "loss": 96.3142,
      "step": 60830
    },
    {
      "epoch": 0.2458012984966689,
      "grad_norm": 1076.640380859375,
      "learning_rate": 3.754761480858249e-05,
      "loss": 77.1947,
      "step": 60840
    },
    {
      "epoch": 0.24584169976203654,
      "grad_norm": 558.1048583984375,
      "learning_rate": 3.7546274797738034e-05,
      "loss": 57.5041,
      "step": 60850
    },
    {
      "epoch": 0.24588210102740418,
      "grad_norm": 1044.107177734375,
      "learning_rate": 3.7544934444819915e-05,
      "loss": 67.853,
      "step": 60860
    },
    {
      "epoch": 0.24592250229277182,
      "grad_norm": 317.490966796875,
      "learning_rate": 3.754359374985426e-05,
      "loss": 80.8347,
      "step": 60870
    },
    {
      "epoch": 0.24596290355813943,
      "grad_norm": 1043.4903564453125,
      "learning_rate": 3.7542252712867214e-05,
      "loss": 104.5762,
      "step": 60880
    },
    {
      "epoch": 0.24600330482350707,
      "grad_norm": 795.8134765625,
      "learning_rate": 3.7540911333884926e-05,
      "loss": 78.7376,
      "step": 60890
    },
    {
      "epoch": 0.24604370608887471,
      "grad_norm": 1080.662841796875,
      "learning_rate": 3.7539569612933536e-05,
      "loss": 105.4763,
      "step": 60900
    },
    {
      "epoch": 0.24608410735424233,
      "grad_norm": 568.9183349609375,
      "learning_rate": 3.753822755003921e-05,
      "loss": 94.6521,
      "step": 60910
    },
    {
      "epoch": 0.24612450861960997,
      "grad_norm": 805.295654296875,
      "learning_rate": 3.753688514522811e-05,
      "loss": 50.2088,
      "step": 60920
    },
    {
      "epoch": 0.2461649098849776,
      "grad_norm": 735.4624633789062,
      "learning_rate": 3.75355423985264e-05,
      "loss": 82.6926,
      "step": 60930
    },
    {
      "epoch": 0.24620531115034522,
      "grad_norm": 3377.027587890625,
      "learning_rate": 3.753419930996026e-05,
      "loss": 88.3906,
      "step": 60940
    },
    {
      "epoch": 0.24624571241571286,
      "grad_norm": 636.9912109375,
      "learning_rate": 3.7532855879555887e-05,
      "loss": 77.4918,
      "step": 60950
    },
    {
      "epoch": 0.2462861136810805,
      "grad_norm": 547.3155517578125,
      "learning_rate": 3.753151210733946e-05,
      "loss": 65.1002,
      "step": 60960
    },
    {
      "epoch": 0.2463265149464481,
      "grad_norm": 957.2113647460938,
      "learning_rate": 3.753016799333717e-05,
      "loss": 65.5449,
      "step": 60970
    },
    {
      "epoch": 0.24636691621181575,
      "grad_norm": 656.7042236328125,
      "learning_rate": 3.752882353757524e-05,
      "loss": 65.8079,
      "step": 60980
    },
    {
      "epoch": 0.2464073174771834,
      "grad_norm": 594.960205078125,
      "learning_rate": 3.752747874007987e-05,
      "loss": 64.1638,
      "step": 60990
    },
    {
      "epoch": 0.246447718742551,
      "grad_norm": 309.1339111328125,
      "learning_rate": 3.7526133600877275e-05,
      "loss": 65.9514,
      "step": 61000
    },
    {
      "epoch": 0.24648812000791864,
      "grad_norm": 1143.3817138671875,
      "learning_rate": 3.7524788119993687e-05,
      "loss": 138.5115,
      "step": 61010
    },
    {
      "epoch": 0.24652852127328628,
      "grad_norm": 833.0191650390625,
      "learning_rate": 3.752344229745532e-05,
      "loss": 106.2841,
      "step": 61020
    },
    {
      "epoch": 0.24656892253865392,
      "grad_norm": 537.9629516601562,
      "learning_rate": 3.7522096133288434e-05,
      "loss": 65.1042,
      "step": 61030
    },
    {
      "epoch": 0.24660932380402154,
      "grad_norm": 1437.601806640625,
      "learning_rate": 3.752074962751926e-05,
      "loss": 111.3172,
      "step": 61040
    },
    {
      "epoch": 0.24664972506938918,
      "grad_norm": 746.3345336914062,
      "learning_rate": 3.751940278017405e-05,
      "loss": 48.527,
      "step": 61050
    },
    {
      "epoch": 0.24669012633475682,
      "grad_norm": 455.0274353027344,
      "learning_rate": 3.751805559127907e-05,
      "loss": 59.7064,
      "step": 61060
    },
    {
      "epoch": 0.24673052760012443,
      "grad_norm": 928.3031616210938,
      "learning_rate": 3.751670806086058e-05,
      "loss": 52.3294,
      "step": 61070
    },
    {
      "epoch": 0.24677092886549207,
      "grad_norm": 766.8455810546875,
      "learning_rate": 3.7515360188944835e-05,
      "loss": 67.7144,
      "step": 61080
    },
    {
      "epoch": 0.2468113301308597,
      "grad_norm": 1098.980224609375,
      "learning_rate": 3.751401197555813e-05,
      "loss": 76.7141,
      "step": 61090
    },
    {
      "epoch": 0.24685173139622732,
      "grad_norm": 522.4241943359375,
      "learning_rate": 3.751266342072675e-05,
      "loss": 72.6551,
      "step": 61100
    },
    {
      "epoch": 0.24689213266159496,
      "grad_norm": 1027.0843505859375,
      "learning_rate": 3.751131452447697e-05,
      "loss": 96.9705,
      "step": 61110
    },
    {
      "epoch": 0.2469325339269626,
      "grad_norm": 1297.1356201171875,
      "learning_rate": 3.7509965286835106e-05,
      "loss": 97.9248,
      "step": 61120
    },
    {
      "epoch": 0.2469729351923302,
      "grad_norm": 740.5809326171875,
      "learning_rate": 3.7508615707827456e-05,
      "loss": 77.4702,
      "step": 61130
    },
    {
      "epoch": 0.24701333645769785,
      "grad_norm": 187.4073486328125,
      "learning_rate": 3.750726578748032e-05,
      "loss": 75.4316,
      "step": 61140
    },
    {
      "epoch": 0.2470537377230655,
      "grad_norm": 769.25634765625,
      "learning_rate": 3.750591552582003e-05,
      "loss": 100.9645,
      "step": 61150
    },
    {
      "epoch": 0.2470941389884331,
      "grad_norm": 582.0955810546875,
      "learning_rate": 3.750456492287291e-05,
      "loss": 94.5516,
      "step": 61160
    },
    {
      "epoch": 0.24713454025380074,
      "grad_norm": 518.8898315429688,
      "learning_rate": 3.7503213978665275e-05,
      "loss": 107.41,
      "step": 61170
    },
    {
      "epoch": 0.24717494151916838,
      "grad_norm": 808.5994873046875,
      "learning_rate": 3.750186269322348e-05,
      "loss": 56.5193,
      "step": 61180
    },
    {
      "epoch": 0.24721534278453602,
      "grad_norm": 818.251953125,
      "learning_rate": 3.750051106657386e-05,
      "loss": 64.9534,
      "step": 61190
    },
    {
      "epoch": 0.24725574404990364,
      "grad_norm": 1727.6558837890625,
      "learning_rate": 3.749915909874276e-05,
      "loss": 67.6505,
      "step": 61200
    },
    {
      "epoch": 0.24729614531527128,
      "grad_norm": 482.91314697265625,
      "learning_rate": 3.749780678975655e-05,
      "loss": 66.262,
      "step": 61210
    },
    {
      "epoch": 0.24733654658063892,
      "grad_norm": 445.8804626464844,
      "learning_rate": 3.7496454139641584e-05,
      "loss": 59.1876,
      "step": 61220
    },
    {
      "epoch": 0.24737694784600653,
      "grad_norm": 817.4906005859375,
      "learning_rate": 3.749510114842424e-05,
      "loss": 79.4554,
      "step": 61230
    },
    {
      "epoch": 0.24741734911137417,
      "grad_norm": 244.64735412597656,
      "learning_rate": 3.7493747816130887e-05,
      "loss": 66.7083,
      "step": 61240
    },
    {
      "epoch": 0.2474577503767418,
      "grad_norm": 790.3543701171875,
      "learning_rate": 3.749239414278792e-05,
      "loss": 71.0745,
      "step": 61250
    },
    {
      "epoch": 0.24749815164210942,
      "grad_norm": 1867.8568115234375,
      "learning_rate": 3.749104012842172e-05,
      "loss": 87.3792,
      "step": 61260
    },
    {
      "epoch": 0.24753855290747706,
      "grad_norm": 1511.1171875,
      "learning_rate": 3.748968577305869e-05,
      "loss": 89.4993,
      "step": 61270
    },
    {
      "epoch": 0.2475789541728447,
      "grad_norm": 1624.3121337890625,
      "learning_rate": 3.748833107672523e-05,
      "loss": 77.4677,
      "step": 61280
    },
    {
      "epoch": 0.2476193554382123,
      "grad_norm": 602.2093505859375,
      "learning_rate": 3.748697603944775e-05,
      "loss": 52.443,
      "step": 61290
    },
    {
      "epoch": 0.24765975670357995,
      "grad_norm": 270.013671875,
      "learning_rate": 3.7485620661252676e-05,
      "loss": 47.6834,
      "step": 61300
    },
    {
      "epoch": 0.2477001579689476,
      "grad_norm": 1774.5087890625,
      "learning_rate": 3.748426494216643e-05,
      "loss": 78.0292,
      "step": 61310
    },
    {
      "epoch": 0.2477405592343152,
      "grad_norm": 1068.1976318359375,
      "learning_rate": 3.748290888221542e-05,
      "loss": 95.4075,
      "step": 61320
    },
    {
      "epoch": 0.24778096049968285,
      "grad_norm": 737.3121337890625,
      "learning_rate": 3.748155248142611e-05,
      "loss": 83.9512,
      "step": 61330
    },
    {
      "epoch": 0.24782136176505049,
      "grad_norm": 760.8573608398438,
      "learning_rate": 3.748019573982493e-05,
      "loss": 73.3414,
      "step": 61340
    },
    {
      "epoch": 0.24786176303041813,
      "grad_norm": 665.1573486328125,
      "learning_rate": 3.747883865743834e-05,
      "loss": 106.5665,
      "step": 61350
    },
    {
      "epoch": 0.24790216429578574,
      "grad_norm": 657.4424438476562,
      "learning_rate": 3.747748123429279e-05,
      "loss": 57.0647,
      "step": 61360
    },
    {
      "epoch": 0.24794256556115338,
      "grad_norm": 946.54248046875,
      "learning_rate": 3.7476123470414745e-05,
      "loss": 69.0039,
      "step": 61370
    },
    {
      "epoch": 0.24798296682652102,
      "grad_norm": 1425.1268310546875,
      "learning_rate": 3.747476536583068e-05,
      "loss": 80.416,
      "step": 61380
    },
    {
      "epoch": 0.24802336809188863,
      "grad_norm": 659.6389770507812,
      "learning_rate": 3.747340692056706e-05,
      "loss": 82.6614,
      "step": 61390
    },
    {
      "epoch": 0.24806376935725627,
      "grad_norm": 908.4437866210938,
      "learning_rate": 3.7472048134650376e-05,
      "loss": 105.7427,
      "step": 61400
    },
    {
      "epoch": 0.2481041706226239,
      "grad_norm": 577.42724609375,
      "learning_rate": 3.747068900810712e-05,
      "loss": 67.8526,
      "step": 61410
    },
    {
      "epoch": 0.24814457188799152,
      "grad_norm": 1340.310791015625,
      "learning_rate": 3.7469329540963795e-05,
      "loss": 57.7734,
      "step": 61420
    },
    {
      "epoch": 0.24818497315335916,
      "grad_norm": 2040.929443359375,
      "learning_rate": 3.746796973324689e-05,
      "loss": 87.0531,
      "step": 61430
    },
    {
      "epoch": 0.2482253744187268,
      "grad_norm": 1341.835693359375,
      "learning_rate": 3.746660958498292e-05,
      "loss": 88.1157,
      "step": 61440
    },
    {
      "epoch": 0.24826577568409441,
      "grad_norm": 682.9309692382812,
      "learning_rate": 3.746524909619841e-05,
      "loss": 68.3139,
      "step": 61450
    },
    {
      "epoch": 0.24830617694946205,
      "grad_norm": 875.2620849609375,
      "learning_rate": 3.746388826691987e-05,
      "loss": 65.5418,
      "step": 61460
    },
    {
      "epoch": 0.2483465782148297,
      "grad_norm": 0.0,
      "learning_rate": 3.746252709717384e-05,
      "loss": 51.6307,
      "step": 61470
    },
    {
      "epoch": 0.2483869794801973,
      "grad_norm": 624.052978515625,
      "learning_rate": 3.746116558698686e-05,
      "loss": 56.3622,
      "step": 61480
    },
    {
      "epoch": 0.24842738074556495,
      "grad_norm": 633.302001953125,
      "learning_rate": 3.745980373638546e-05,
      "loss": 52.7323,
      "step": 61490
    },
    {
      "epoch": 0.2484677820109326,
      "grad_norm": 1113.2095947265625,
      "learning_rate": 3.745844154539619e-05,
      "loss": 96.0454,
      "step": 61500
    },
    {
      "epoch": 0.24850818327630023,
      "grad_norm": 766.4423217773438,
      "learning_rate": 3.745707901404563e-05,
      "loss": 87.217,
      "step": 61510
    },
    {
      "epoch": 0.24854858454166784,
      "grad_norm": 1302.5953369140625,
      "learning_rate": 3.7455716142360324e-05,
      "loss": 100.6767,
      "step": 61520
    },
    {
      "epoch": 0.24858898580703548,
      "grad_norm": 322.5316467285156,
      "learning_rate": 3.7454352930366834e-05,
      "loss": 85.5547,
      "step": 61530
    },
    {
      "epoch": 0.24862938707240312,
      "grad_norm": 809.63916015625,
      "learning_rate": 3.7452989378091755e-05,
      "loss": 67.3807,
      "step": 61540
    },
    {
      "epoch": 0.24866978833777073,
      "grad_norm": 513.2466430664062,
      "learning_rate": 3.7451625485561665e-05,
      "loss": 50.81,
      "step": 61550
    },
    {
      "epoch": 0.24871018960313837,
      "grad_norm": 3258.237060546875,
      "learning_rate": 3.7450261252803146e-05,
      "loss": 105.0272,
      "step": 61560
    },
    {
      "epoch": 0.248750590868506,
      "grad_norm": 745.373046875,
      "learning_rate": 3.74488966798428e-05,
      "loss": 40.7696,
      "step": 61570
    },
    {
      "epoch": 0.24879099213387362,
      "grad_norm": 652.2501831054688,
      "learning_rate": 3.744753176670724e-05,
      "loss": 70.7613,
      "step": 61580
    },
    {
      "epoch": 0.24883139339924126,
      "grad_norm": 619.707763671875,
      "learning_rate": 3.744616651342305e-05,
      "loss": 86.4871,
      "step": 61590
    },
    {
      "epoch": 0.2488717946646089,
      "grad_norm": 2102.8876953125,
      "learning_rate": 3.7444800920016875e-05,
      "loss": 79.4103,
      "step": 61600
    },
    {
      "epoch": 0.24891219592997652,
      "grad_norm": 1047.6876220703125,
      "learning_rate": 3.744343498651532e-05,
      "loss": 60.9901,
      "step": 61610
    },
    {
      "epoch": 0.24895259719534416,
      "grad_norm": 762.4063110351562,
      "learning_rate": 3.744206871294502e-05,
      "loss": 84.9273,
      "step": 61620
    },
    {
      "epoch": 0.2489929984607118,
      "grad_norm": 595.7750244140625,
      "learning_rate": 3.744070209933261e-05,
      "loss": 78.6703,
      "step": 61630
    },
    {
      "epoch": 0.2490333997260794,
      "grad_norm": 920.8694458007812,
      "learning_rate": 3.743933514570473e-05,
      "loss": 74.1275,
      "step": 61640
    },
    {
      "epoch": 0.24907380099144705,
      "grad_norm": 802.4329223632812,
      "learning_rate": 3.743796785208804e-05,
      "loss": 51.6497,
      "step": 61650
    },
    {
      "epoch": 0.2491142022568147,
      "grad_norm": 921.7340087890625,
      "learning_rate": 3.743660021850919e-05,
      "loss": 53.6544,
      "step": 61660
    },
    {
      "epoch": 0.24915460352218233,
      "grad_norm": 992.1342163085938,
      "learning_rate": 3.743523224499483e-05,
      "loss": 69.4302,
      "step": 61670
    },
    {
      "epoch": 0.24919500478754994,
      "grad_norm": 1299.7554931640625,
      "learning_rate": 3.743386393157165e-05,
      "loss": 64.1878,
      "step": 61680
    },
    {
      "epoch": 0.24923540605291758,
      "grad_norm": 500.22796630859375,
      "learning_rate": 3.743249527826632e-05,
      "loss": 36.1513,
      "step": 61690
    },
    {
      "epoch": 0.24927580731828522,
      "grad_norm": 325.031005859375,
      "learning_rate": 3.7431126285105516e-05,
      "loss": 138.1334,
      "step": 61700
    },
    {
      "epoch": 0.24931620858365283,
      "grad_norm": 630.5838623046875,
      "learning_rate": 3.742975695211593e-05,
      "loss": 88.5154,
      "step": 61710
    },
    {
      "epoch": 0.24935660984902047,
      "grad_norm": 534.2533569335938,
      "learning_rate": 3.742838727932426e-05,
      "loss": 61.4114,
      "step": 61720
    },
    {
      "epoch": 0.2493970111143881,
      "grad_norm": 437.1034851074219,
      "learning_rate": 3.74270172667572e-05,
      "loss": 93.259,
      "step": 61730
    },
    {
      "epoch": 0.24943741237975572,
      "grad_norm": 530.0656127929688,
      "learning_rate": 3.742564691444147e-05,
      "loss": 81.5553,
      "step": 61740
    },
    {
      "epoch": 0.24947781364512336,
      "grad_norm": 577.9969482421875,
      "learning_rate": 3.7424276222403795e-05,
      "loss": 70.7233,
      "step": 61750
    },
    {
      "epoch": 0.249518214910491,
      "grad_norm": 1053.4208984375,
      "learning_rate": 3.742290519067087e-05,
      "loss": 90.1125,
      "step": 61760
    },
    {
      "epoch": 0.24955861617585862,
      "grad_norm": 633.0479736328125,
      "learning_rate": 3.742153381926945e-05,
      "loss": 60.1169,
      "step": 61770
    },
    {
      "epoch": 0.24959901744122626,
      "grad_norm": 548.5878295898438,
      "learning_rate": 3.742016210822624e-05,
      "loss": 88.1457,
      "step": 61780
    },
    {
      "epoch": 0.2496394187065939,
      "grad_norm": 368.5179748535156,
      "learning_rate": 3.741879005756802e-05,
      "loss": 101.9935,
      "step": 61790
    },
    {
      "epoch": 0.2496798199719615,
      "grad_norm": 2975.427978515625,
      "learning_rate": 3.741741766732151e-05,
      "loss": 93.7819,
      "step": 61800
    },
    {
      "epoch": 0.24972022123732915,
      "grad_norm": 767.5460205078125,
      "learning_rate": 3.741604493751348e-05,
      "loss": 76.1625,
      "step": 61810
    },
    {
      "epoch": 0.2497606225026968,
      "grad_norm": 704.7868041992188,
      "learning_rate": 3.741467186817068e-05,
      "loss": 133.6862,
      "step": 61820
    },
    {
      "epoch": 0.24980102376806443,
      "grad_norm": 1289.1790771484375,
      "learning_rate": 3.7413298459319897e-05,
      "loss": 91.6314,
      "step": 61830
    },
    {
      "epoch": 0.24984142503343204,
      "grad_norm": 815.9835815429688,
      "learning_rate": 3.741192471098789e-05,
      "loss": 66.9851,
      "step": 61840
    },
    {
      "epoch": 0.24988182629879968,
      "grad_norm": 683.07763671875,
      "learning_rate": 3.741055062320145e-05,
      "loss": 36.704,
      "step": 61850
    },
    {
      "epoch": 0.24992222756416732,
      "grad_norm": 1541.2674560546875,
      "learning_rate": 3.740917619598736e-05,
      "loss": 79.5865,
      "step": 61860
    },
    {
      "epoch": 0.24996262882953493,
      "grad_norm": 738.2477416992188,
      "learning_rate": 3.740780142937241e-05,
      "loss": 57.1532,
      "step": 61870
    },
    {
      "epoch": 0.25000303009490255,
      "grad_norm": 403.5694885253906,
      "learning_rate": 3.7406426323383416e-05,
      "loss": 87.1524,
      "step": 61880
    },
    {
      "epoch": 0.2500434313602702,
      "grad_norm": 779.46533203125,
      "learning_rate": 3.740505087804718e-05,
      "loss": 53.8653,
      "step": 61890
    },
    {
      "epoch": 0.2500838326256378,
      "grad_norm": 359.40716552734375,
      "learning_rate": 3.740367509339052e-05,
      "loss": 87.3346,
      "step": 61900
    },
    {
      "epoch": 0.25012423389100547,
      "grad_norm": 1014.7713012695312,
      "learning_rate": 3.740229896944025e-05,
      "loss": 61.4974,
      "step": 61910
    },
    {
      "epoch": 0.2501646351563731,
      "grad_norm": 870.2494506835938,
      "learning_rate": 3.74009225062232e-05,
      "loss": 93.1416,
      "step": 61920
    },
    {
      "epoch": 0.25020503642174075,
      "grad_norm": 578.4398193359375,
      "learning_rate": 3.739954570376621e-05,
      "loss": 50.1696,
      "step": 61930
    },
    {
      "epoch": 0.2502454376871084,
      "grad_norm": 1939.190185546875,
      "learning_rate": 3.7398168562096115e-05,
      "loss": 128.2365,
      "step": 61940
    },
    {
      "epoch": 0.25028583895247597,
      "grad_norm": 838.7545776367188,
      "learning_rate": 3.7396791081239766e-05,
      "loss": 55.7641,
      "step": 61950
    },
    {
      "epoch": 0.2503262402178436,
      "grad_norm": 1380.188232421875,
      "learning_rate": 3.7395413261224026e-05,
      "loss": 58.8139,
      "step": 61960
    },
    {
      "epoch": 0.25036664148321125,
      "grad_norm": 657.556884765625,
      "learning_rate": 3.739403510207574e-05,
      "loss": 40.2833,
      "step": 61970
    },
    {
      "epoch": 0.2504070427485789,
      "grad_norm": 827.7333984375,
      "learning_rate": 3.7392656603821794e-05,
      "loss": 62.6232,
      "step": 61980
    },
    {
      "epoch": 0.25044744401394653,
      "grad_norm": 933.279541015625,
      "learning_rate": 3.7391277766489044e-05,
      "loss": 63.5383,
      "step": 61990
    },
    {
      "epoch": 0.25048784527931417,
      "grad_norm": 1004.0429077148438,
      "learning_rate": 3.738989859010438e-05,
      "loss": 101.3109,
      "step": 62000
    },
    {
      "epoch": 0.25052824654468175,
      "grad_norm": 0.0,
      "learning_rate": 3.73885190746947e-05,
      "loss": 65.8957,
      "step": 62010
    },
    {
      "epoch": 0.2505686478100494,
      "grad_norm": 806.5186767578125,
      "learning_rate": 3.738713922028688e-05,
      "loss": 53.8241,
      "step": 62020
    },
    {
      "epoch": 0.25060904907541703,
      "grad_norm": 2066.307861328125,
      "learning_rate": 3.7385759026907836e-05,
      "loss": 77.2466,
      "step": 62030
    },
    {
      "epoch": 0.2506494503407847,
      "grad_norm": 1385.8570556640625,
      "learning_rate": 3.738437849458446e-05,
      "loss": 93.8371,
      "step": 62040
    },
    {
      "epoch": 0.2506898516061523,
      "grad_norm": 1187.6527099609375,
      "learning_rate": 3.738299762334368e-05,
      "loss": 66.5861,
      "step": 62050
    },
    {
      "epoch": 0.25073025287151995,
      "grad_norm": 567.3513793945312,
      "learning_rate": 3.738161641321242e-05,
      "loss": 59.9558,
      "step": 62060
    },
    {
      "epoch": 0.2507706541368876,
      "grad_norm": 611.0079345703125,
      "learning_rate": 3.7380234864217584e-05,
      "loss": 84.0864,
      "step": 62070
    },
    {
      "epoch": 0.2508110554022552,
      "grad_norm": 646.5040283203125,
      "learning_rate": 3.737885297638613e-05,
      "loss": 70.7648,
      "step": 62080
    },
    {
      "epoch": 0.2508514566676228,
      "grad_norm": 855.0535278320312,
      "learning_rate": 3.737747074974498e-05,
      "loss": 76.4832,
      "step": 62090
    },
    {
      "epoch": 0.25089185793299046,
      "grad_norm": 560.4104614257812,
      "learning_rate": 3.737608818432111e-05,
      "loss": 61.4092,
      "step": 62100
    },
    {
      "epoch": 0.2509322591983581,
      "grad_norm": 695.6069946289062,
      "learning_rate": 3.737470528014145e-05,
      "loss": 62.4705,
      "step": 62110
    },
    {
      "epoch": 0.25097266046372574,
      "grad_norm": 646.8819580078125,
      "learning_rate": 3.7373322037232956e-05,
      "loss": 53.1848,
      "step": 62120
    },
    {
      "epoch": 0.2510130617290934,
      "grad_norm": 1300.523681640625,
      "learning_rate": 3.737193845562261e-05,
      "loss": 79.2234,
      "step": 62130
    },
    {
      "epoch": 0.25105346299446096,
      "grad_norm": 917.3502197265625,
      "learning_rate": 3.737055453533738e-05,
      "loss": 49.7496,
      "step": 62140
    },
    {
      "epoch": 0.2510938642598286,
      "grad_norm": 1223.5814208984375,
      "learning_rate": 3.7369170276404245e-05,
      "loss": 79.9229,
      "step": 62150
    },
    {
      "epoch": 0.25113426552519624,
      "grad_norm": 823.5051879882812,
      "learning_rate": 3.7367785678850196e-05,
      "loss": 65.1247,
      "step": 62160
    },
    {
      "epoch": 0.2511746667905639,
      "grad_norm": 1042.2567138671875,
      "learning_rate": 3.736640074270222e-05,
      "loss": 62.8723,
      "step": 62170
    },
    {
      "epoch": 0.2512150680559315,
      "grad_norm": 1149.8233642578125,
      "learning_rate": 3.736501546798732e-05,
      "loss": 69.4781,
      "step": 62180
    },
    {
      "epoch": 0.25125546932129916,
      "grad_norm": 905.4100341796875,
      "learning_rate": 3.7363629854732506e-05,
      "loss": 66.374,
      "step": 62190
    },
    {
      "epoch": 0.25129587058666675,
      "grad_norm": 1982.888916015625,
      "learning_rate": 3.736224390296479e-05,
      "loss": 89.6081,
      "step": 62200
    },
    {
      "epoch": 0.2513362718520344,
      "grad_norm": 725.7110595703125,
      "learning_rate": 3.736085761271119e-05,
      "loss": 64.456,
      "step": 62210
    },
    {
      "epoch": 0.251376673117402,
      "grad_norm": 1051.466796875,
      "learning_rate": 3.735947098399873e-05,
      "loss": 60.7936,
      "step": 62220
    },
    {
      "epoch": 0.25141707438276967,
      "grad_norm": 702.5640258789062,
      "learning_rate": 3.7358084016854445e-05,
      "loss": 44.4488,
      "step": 62230
    },
    {
      "epoch": 0.2514574756481373,
      "grad_norm": 1065.532958984375,
      "learning_rate": 3.735669671130538e-05,
      "loss": 60.1263,
      "step": 62240
    },
    {
      "epoch": 0.25149787691350495,
      "grad_norm": 872.7839965820312,
      "learning_rate": 3.735530906737857e-05,
      "loss": 67.0821,
      "step": 62250
    },
    {
      "epoch": 0.2515382781788726,
      "grad_norm": 906.0625,
      "learning_rate": 3.735392108510108e-05,
      "loss": 65.652,
      "step": 62260
    },
    {
      "epoch": 0.25157867944424017,
      "grad_norm": 697.1912841796875,
      "learning_rate": 3.735253276449997e-05,
      "loss": 100.2691,
      "step": 62270
    },
    {
      "epoch": 0.2516190807096078,
      "grad_norm": 1175.5914306640625,
      "learning_rate": 3.735114410560229e-05,
      "loss": 69.7755,
      "step": 62280
    },
    {
      "epoch": 0.25165948197497545,
      "grad_norm": 1274.5322265625,
      "learning_rate": 3.734975510843512e-05,
      "loss": 74.5015,
      "step": 62290
    },
    {
      "epoch": 0.2516998832403431,
      "grad_norm": 442.2524108886719,
      "learning_rate": 3.734836577302555e-05,
      "loss": 66.8449,
      "step": 62300
    },
    {
      "epoch": 0.25174028450571073,
      "grad_norm": 466.80865478515625,
      "learning_rate": 3.734697609940066e-05,
      "loss": 79.3407,
      "step": 62310
    },
    {
      "epoch": 0.25178068577107837,
      "grad_norm": 192.1374969482422,
      "learning_rate": 3.7345586087587535e-05,
      "loss": 61.5628,
      "step": 62320
    },
    {
      "epoch": 0.25182108703644596,
      "grad_norm": 2281.552978515625,
      "learning_rate": 3.734419573761328e-05,
      "loss": 72.6444,
      "step": 62330
    },
    {
      "epoch": 0.2518614883018136,
      "grad_norm": 821.1915893554688,
      "learning_rate": 3.7342805049504994e-05,
      "loss": 111.8539,
      "step": 62340
    },
    {
      "epoch": 0.25190188956718124,
      "grad_norm": 576.8593139648438,
      "learning_rate": 3.73414140232898e-05,
      "loss": 63.514,
      "step": 62350
    },
    {
      "epoch": 0.2519422908325489,
      "grad_norm": 285.91815185546875,
      "learning_rate": 3.7340022658994816e-05,
      "loss": 74.2678,
      "step": 62360
    },
    {
      "epoch": 0.2519826920979165,
      "grad_norm": 779.8338012695312,
      "learning_rate": 3.733863095664716e-05,
      "loss": 55.4725,
      "step": 62370
    },
    {
      "epoch": 0.25202309336328416,
      "grad_norm": 601.2800903320312,
      "learning_rate": 3.733723891627396e-05,
      "loss": 57.6946,
      "step": 62380
    },
    {
      "epoch": 0.2520634946286518,
      "grad_norm": 1605.4139404296875,
      "learning_rate": 3.733584653790237e-05,
      "loss": 88.7472,
      "step": 62390
    },
    {
      "epoch": 0.2521038958940194,
      "grad_norm": 755.4089965820312,
      "learning_rate": 3.7334453821559534e-05,
      "loss": 52.6565,
      "step": 62400
    },
    {
      "epoch": 0.252144297159387,
      "grad_norm": 696.624755859375,
      "learning_rate": 3.733306076727258e-05,
      "loss": 65.9678,
      "step": 62410
    },
    {
      "epoch": 0.25218469842475466,
      "grad_norm": 1728.781982421875,
      "learning_rate": 3.733166737506869e-05,
      "loss": 65.9936,
      "step": 62420
    },
    {
      "epoch": 0.2522250996901223,
      "grad_norm": 329.3301086425781,
      "learning_rate": 3.733027364497502e-05,
      "loss": 93.0544,
      "step": 62430
    },
    {
      "epoch": 0.25226550095548994,
      "grad_norm": 928.8203735351562,
      "learning_rate": 3.732887957701874e-05,
      "loss": 76.0958,
      "step": 62440
    },
    {
      "epoch": 0.2523059022208576,
      "grad_norm": 765.6958618164062,
      "learning_rate": 3.732748517122704e-05,
      "loss": 53.0726,
      "step": 62450
    },
    {
      "epoch": 0.25234630348622517,
      "grad_norm": 510.4883117675781,
      "learning_rate": 3.732609042762709e-05,
      "loss": 58.0227,
      "step": 62460
    },
    {
      "epoch": 0.2523867047515928,
      "grad_norm": 980.6634521484375,
      "learning_rate": 3.732469534624609e-05,
      "loss": 104.429,
      "step": 62470
    },
    {
      "epoch": 0.25242710601696045,
      "grad_norm": 1157.6436767578125,
      "learning_rate": 3.732329992711123e-05,
      "loss": 91.0166,
      "step": 62480
    },
    {
      "epoch": 0.2524675072823281,
      "grad_norm": 946.9705200195312,
      "learning_rate": 3.732190417024972e-05,
      "loss": 89.0631,
      "step": 62490
    },
    {
      "epoch": 0.2525079085476957,
      "grad_norm": 719.9625244140625,
      "learning_rate": 3.732050807568878e-05,
      "loss": 80.3477,
      "step": 62500
    },
    {
      "epoch": 0.25254830981306337,
      "grad_norm": 636.4144897460938,
      "learning_rate": 3.731911164345561e-05,
      "loss": 56.9953,
      "step": 62510
    },
    {
      "epoch": 0.25258871107843095,
      "grad_norm": 370.7101135253906,
      "learning_rate": 3.731771487357744e-05,
      "loss": 66.2618,
      "step": 62520
    },
    {
      "epoch": 0.2526291123437986,
      "grad_norm": 892.8953247070312,
      "learning_rate": 3.731631776608151e-05,
      "loss": 74.3697,
      "step": 62530
    },
    {
      "epoch": 0.25266951360916623,
      "grad_norm": 421.4278259277344,
      "learning_rate": 3.731492032099504e-05,
      "loss": 100.9044,
      "step": 62540
    },
    {
      "epoch": 0.25270991487453387,
      "grad_norm": 969.6364135742188,
      "learning_rate": 3.731352253834529e-05,
      "loss": 94.6507,
      "step": 62550
    },
    {
      "epoch": 0.2527503161399015,
      "grad_norm": 855.2815551757812,
      "learning_rate": 3.73121244181595e-05,
      "loss": 120.8158,
      "step": 62560
    },
    {
      "epoch": 0.25279071740526915,
      "grad_norm": 3310.03515625,
      "learning_rate": 3.731072596046493e-05,
      "loss": 86.2226,
      "step": 62570
    },
    {
      "epoch": 0.2528311186706368,
      "grad_norm": 393.9659423828125,
      "learning_rate": 3.7309327165288857e-05,
      "loss": 77.2755,
      "step": 62580
    },
    {
      "epoch": 0.2528715199360044,
      "grad_norm": 702.1710815429688,
      "learning_rate": 3.730792803265853e-05,
      "loss": 47.471,
      "step": 62590
    },
    {
      "epoch": 0.252911921201372,
      "grad_norm": 2308.98095703125,
      "learning_rate": 3.7306528562601245e-05,
      "loss": 64.0197,
      "step": 62600
    },
    {
      "epoch": 0.25295232246673965,
      "grad_norm": 1097.486083984375,
      "learning_rate": 3.7305128755144265e-05,
      "loss": 120.0064,
      "step": 62610
    },
    {
      "epoch": 0.2529927237321073,
      "grad_norm": 750.9093627929688,
      "learning_rate": 3.73037286103149e-05,
      "loss": 101.2599,
      "step": 62620
    },
    {
      "epoch": 0.25303312499747493,
      "grad_norm": 502.05645751953125,
      "learning_rate": 3.730232812814043e-05,
      "loss": 68.1188,
      "step": 62630
    },
    {
      "epoch": 0.2530735262628426,
      "grad_norm": 1803.06005859375,
      "learning_rate": 3.730092730864816e-05,
      "loss": 70.3783,
      "step": 62640
    },
    {
      "epoch": 0.25311392752821016,
      "grad_norm": 1725.7425537109375,
      "learning_rate": 3.729952615186542e-05,
      "loss": 69.9886,
      "step": 62650
    },
    {
      "epoch": 0.2531543287935778,
      "grad_norm": 656.0050048828125,
      "learning_rate": 3.72981246578195e-05,
      "loss": 116.2143,
      "step": 62660
    },
    {
      "epoch": 0.25319473005894544,
      "grad_norm": 1667.094970703125,
      "learning_rate": 3.729672282653774e-05,
      "loss": 81.6159,
      "step": 62670
    },
    {
      "epoch": 0.2532351313243131,
      "grad_norm": 1342.62890625,
      "learning_rate": 3.729532065804746e-05,
      "loss": 102.0702,
      "step": 62680
    },
    {
      "epoch": 0.2532755325896807,
      "grad_norm": 648.2364501953125,
      "learning_rate": 3.7293918152376004e-05,
      "loss": 67.3058,
      "step": 62690
    },
    {
      "epoch": 0.25331593385504836,
      "grad_norm": 773.669921875,
      "learning_rate": 3.72925153095507e-05,
      "loss": 69.3494,
      "step": 62700
    },
    {
      "epoch": 0.253356335120416,
      "grad_norm": 616.6019897460938,
      "learning_rate": 3.729111212959891e-05,
      "loss": 49.2931,
      "step": 62710
    },
    {
      "epoch": 0.2533967363857836,
      "grad_norm": 1789.956787109375,
      "learning_rate": 3.7289708612547995e-05,
      "loss": 44.2154,
      "step": 62720
    },
    {
      "epoch": 0.2534371376511512,
      "grad_norm": 603.3187255859375,
      "learning_rate": 3.72883047584253e-05,
      "loss": 65.0651,
      "step": 62730
    },
    {
      "epoch": 0.25347753891651886,
      "grad_norm": 669.8085327148438,
      "learning_rate": 3.728690056725821e-05,
      "loss": 99.6179,
      "step": 62740
    },
    {
      "epoch": 0.2535179401818865,
      "grad_norm": 1055.7017822265625,
      "learning_rate": 3.72854960390741e-05,
      "loss": 86.5876,
      "step": 62750
    },
    {
      "epoch": 0.25355834144725414,
      "grad_norm": 323.4556884765625,
      "learning_rate": 3.728409117390033e-05,
      "loss": 72.2166,
      "step": 62760
    },
    {
      "epoch": 0.2535987427126218,
      "grad_norm": 3452.567626953125,
      "learning_rate": 3.7282685971764316e-05,
      "loss": 188.2608,
      "step": 62770
    },
    {
      "epoch": 0.25363914397798937,
      "grad_norm": 684.7014770507812,
      "learning_rate": 3.7281280432693436e-05,
      "loss": 58.8837,
      "step": 62780
    },
    {
      "epoch": 0.253679545243357,
      "grad_norm": 240.87582397460938,
      "learning_rate": 3.72798745567151e-05,
      "loss": 63.6725,
      "step": 62790
    },
    {
      "epoch": 0.25371994650872465,
      "grad_norm": 1372.149658203125,
      "learning_rate": 3.727846834385671e-05,
      "loss": 104.2431,
      "step": 62800
    },
    {
      "epoch": 0.2537603477740923,
      "grad_norm": 1499.7823486328125,
      "learning_rate": 3.727706179414568e-05,
      "loss": 102.7563,
      "step": 62810
    },
    {
      "epoch": 0.2538007490394599,
      "grad_norm": 661.4194946289062,
      "learning_rate": 3.7275654907609444e-05,
      "loss": 45.8049,
      "step": 62820
    },
    {
      "epoch": 0.25384115030482757,
      "grad_norm": 399.1147766113281,
      "learning_rate": 3.727424768427542e-05,
      "loss": 74.2699,
      "step": 62830
    },
    {
      "epoch": 0.25388155157019515,
      "grad_norm": 908.5316772460938,
      "learning_rate": 3.727284012417104e-05,
      "loss": 71.9168,
      "step": 62840
    },
    {
      "epoch": 0.2539219528355628,
      "grad_norm": 636.9518432617188,
      "learning_rate": 3.727143222732375e-05,
      "loss": 71.2355,
      "step": 62850
    },
    {
      "epoch": 0.25396235410093043,
      "grad_norm": 1124.2080078125,
      "learning_rate": 3.7270023993761e-05,
      "loss": 104.8447,
      "step": 62860
    },
    {
      "epoch": 0.25400275536629807,
      "grad_norm": 1469.802001953125,
      "learning_rate": 3.726861542351024e-05,
      "loss": 56.2755,
      "step": 62870
    },
    {
      "epoch": 0.2540431566316657,
      "grad_norm": 532.6864624023438,
      "learning_rate": 3.726720651659893e-05,
      "loss": 70.7993,
      "step": 62880
    },
    {
      "epoch": 0.25408355789703335,
      "grad_norm": 783.446044921875,
      "learning_rate": 3.726579727305454e-05,
      "loss": 47.7729,
      "step": 62890
    },
    {
      "epoch": 0.254123959162401,
      "grad_norm": 1531.939697265625,
      "learning_rate": 3.726438769290454e-05,
      "loss": 85.5388,
      "step": 62900
    },
    {
      "epoch": 0.2541643604277686,
      "grad_norm": 571.3388061523438,
      "learning_rate": 3.726297777617642e-05,
      "loss": 94.1963,
      "step": 62910
    },
    {
      "epoch": 0.2542047616931362,
      "grad_norm": 1117.3424072265625,
      "learning_rate": 3.7261567522897656e-05,
      "loss": 80.618,
      "step": 62920
    },
    {
      "epoch": 0.25424516295850386,
      "grad_norm": 613.401611328125,
      "learning_rate": 3.726015693309575e-05,
      "loss": 67.6146,
      "step": 62930
    },
    {
      "epoch": 0.2542855642238715,
      "grad_norm": 194.6875762939453,
      "learning_rate": 3.7258746006798194e-05,
      "loss": 70.3354,
      "step": 62940
    },
    {
      "epoch": 0.25432596548923914,
      "grad_norm": 586.9131469726562,
      "learning_rate": 3.72573347440325e-05,
      "loss": 91.6146,
      "step": 62950
    },
    {
      "epoch": 0.2543663667546068,
      "grad_norm": 886.0733032226562,
      "learning_rate": 3.725592314482619e-05,
      "loss": 69.5728,
      "step": 62960
    },
    {
      "epoch": 0.25440676801997436,
      "grad_norm": 887.9442138671875,
      "learning_rate": 3.725451120920676e-05,
      "loss": 77.3229,
      "step": 62970
    },
    {
      "epoch": 0.254447169285342,
      "grad_norm": 617.3834228515625,
      "learning_rate": 3.725309893720176e-05,
      "loss": 56.8363,
      "step": 62980
    },
    {
      "epoch": 0.25448757055070964,
      "grad_norm": 105.37464141845703,
      "learning_rate": 3.725168632883871e-05,
      "loss": 45.1374,
      "step": 62990
    },
    {
      "epoch": 0.2545279718160773,
      "grad_norm": 1526.1341552734375,
      "learning_rate": 3.725027338414515e-05,
      "loss": 105.5693,
      "step": 63000
    },
    {
      "epoch": 0.2545683730814449,
      "grad_norm": 1037.0675048828125,
      "learning_rate": 3.7248860103148636e-05,
      "loss": 74.4302,
      "step": 63010
    },
    {
      "epoch": 0.25460877434681256,
      "grad_norm": 436.3997802734375,
      "learning_rate": 3.724744648587671e-05,
      "loss": 47.4108,
      "step": 63020
    },
    {
      "epoch": 0.2546491756121802,
      "grad_norm": 812.2733764648438,
      "learning_rate": 3.724603253235694e-05,
      "loss": 58.7146,
      "step": 63030
    },
    {
      "epoch": 0.2546895768775478,
      "grad_norm": 764.3171997070312,
      "learning_rate": 3.724461824261688e-05,
      "loss": 84.978,
      "step": 63040
    },
    {
      "epoch": 0.2547299781429154,
      "grad_norm": 605.2657470703125,
      "learning_rate": 3.724320361668412e-05,
      "loss": 63.5626,
      "step": 63050
    },
    {
      "epoch": 0.25477037940828307,
      "grad_norm": 607.5908813476562,
      "learning_rate": 3.7241788654586215e-05,
      "loss": 72.101,
      "step": 63060
    },
    {
      "epoch": 0.2548107806736507,
      "grad_norm": 550.0535888671875,
      "learning_rate": 3.724037335635076e-05,
      "loss": 58.6045,
      "step": 63070
    },
    {
      "epoch": 0.25485118193901835,
      "grad_norm": 756.3518676757812,
      "learning_rate": 3.7238957722005364e-05,
      "loss": 79.5881,
      "step": 63080
    },
    {
      "epoch": 0.254891583204386,
      "grad_norm": 736.576904296875,
      "learning_rate": 3.723754175157761e-05,
      "loss": 43.176,
      "step": 63090
    },
    {
      "epoch": 0.25493198446975357,
      "grad_norm": 1498.8287353515625,
      "learning_rate": 3.72361254450951e-05,
      "loss": 71.7575,
      "step": 63100
    },
    {
      "epoch": 0.2549723857351212,
      "grad_norm": 989.7176513671875,
      "learning_rate": 3.7234708802585455e-05,
      "loss": 62.0116,
      "step": 63110
    },
    {
      "epoch": 0.25501278700048885,
      "grad_norm": 663.2655029296875,
      "learning_rate": 3.723329182407629e-05,
      "loss": 64.7273,
      "step": 63120
    },
    {
      "epoch": 0.2550531882658565,
      "grad_norm": 528.9407958984375,
      "learning_rate": 3.723187450959523e-05,
      "loss": 60.0449,
      "step": 63130
    },
    {
      "epoch": 0.25509358953122413,
      "grad_norm": 1840.953857421875,
      "learning_rate": 3.7230456859169894e-05,
      "loss": 86.6429,
      "step": 63140
    },
    {
      "epoch": 0.25513399079659177,
      "grad_norm": 707.001708984375,
      "learning_rate": 3.7229038872827936e-05,
      "loss": 65.0906,
      "step": 63150
    },
    {
      "epoch": 0.25517439206195935,
      "grad_norm": 584.7863159179688,
      "learning_rate": 3.7227620550597e-05,
      "loss": 44.3554,
      "step": 63160
    },
    {
      "epoch": 0.255214793327327,
      "grad_norm": 316.122314453125,
      "learning_rate": 3.722620189250473e-05,
      "loss": 84.5047,
      "step": 63170
    },
    {
      "epoch": 0.25525519459269463,
      "grad_norm": 710.6033935546875,
      "learning_rate": 3.722478289857878e-05,
      "loss": 48.4606,
      "step": 63180
    },
    {
      "epoch": 0.2552955958580623,
      "grad_norm": 1002.5296020507812,
      "learning_rate": 3.722336356884682e-05,
      "loss": 79.9188,
      "step": 63190
    },
    {
      "epoch": 0.2553359971234299,
      "grad_norm": 868.291748046875,
      "learning_rate": 3.722194390333653e-05,
      "loss": 82.2959,
      "step": 63200
    },
    {
      "epoch": 0.25537639838879755,
      "grad_norm": 528.7097778320312,
      "learning_rate": 3.722052390207557e-05,
      "loss": 69.8529,
      "step": 63210
    },
    {
      "epoch": 0.2554167996541652,
      "grad_norm": 429.8751525878906,
      "learning_rate": 3.721910356509163e-05,
      "loss": 99.49,
      "step": 63220
    },
    {
      "epoch": 0.2554572009195328,
      "grad_norm": 500.3701477050781,
      "learning_rate": 3.7217682892412404e-05,
      "loss": 66.9053,
      "step": 63230
    },
    {
      "epoch": 0.2554976021849004,
      "grad_norm": 415.3155212402344,
      "learning_rate": 3.7216261884065585e-05,
      "loss": 75.5311,
      "step": 63240
    },
    {
      "epoch": 0.25553800345026806,
      "grad_norm": 934.2700805664062,
      "learning_rate": 3.721484054007888e-05,
      "loss": 74.0346,
      "step": 63250
    },
    {
      "epoch": 0.2555784047156357,
      "grad_norm": 1153.2728271484375,
      "learning_rate": 3.721341886047999e-05,
      "loss": 78.737,
      "step": 63260
    },
    {
      "epoch": 0.25561880598100334,
      "grad_norm": 1125.8875732421875,
      "learning_rate": 3.7211996845296636e-05,
      "loss": 74.579,
      "step": 63270
    },
    {
      "epoch": 0.255659207246371,
      "grad_norm": 561.3126220703125,
      "learning_rate": 3.7210574494556546e-05,
      "loss": 111.5302,
      "step": 63280
    },
    {
      "epoch": 0.25569960851173856,
      "grad_norm": 967.9246215820312,
      "learning_rate": 3.7209151808287447e-05,
      "loss": 70.2826,
      "step": 63290
    },
    {
      "epoch": 0.2557400097771062,
      "grad_norm": 702.910400390625,
      "learning_rate": 3.720772878651707e-05,
      "loss": 87.1197,
      "step": 63300
    },
    {
      "epoch": 0.25578041104247384,
      "grad_norm": 1672.2962646484375,
      "learning_rate": 3.7206305429273164e-05,
      "loss": 77.6456,
      "step": 63310
    },
    {
      "epoch": 0.2558208123078415,
      "grad_norm": 980.3726806640625,
      "learning_rate": 3.720488173658347e-05,
      "loss": 80.2127,
      "step": 63320
    },
    {
      "epoch": 0.2558612135732091,
      "grad_norm": 1012.27880859375,
      "learning_rate": 3.720345770847575e-05,
      "loss": 60.8711,
      "step": 63330
    },
    {
      "epoch": 0.25590161483857676,
      "grad_norm": 1901.3438720703125,
      "learning_rate": 3.7202033344977774e-05,
      "loss": 74.989,
      "step": 63340
    },
    {
      "epoch": 0.2559420161039444,
      "grad_norm": 1224.521484375,
      "learning_rate": 3.72006086461173e-05,
      "loss": 45.6287,
      "step": 63350
    },
    {
      "epoch": 0.255982417369312,
      "grad_norm": 586.5023193359375,
      "learning_rate": 3.7199183611922095e-05,
      "loss": 83.5081,
      "step": 63360
    },
    {
      "epoch": 0.2560228186346796,
      "grad_norm": 1021.5157470703125,
      "learning_rate": 3.719775824241996e-05,
      "loss": 55.7175,
      "step": 63370
    },
    {
      "epoch": 0.25606321990004727,
      "grad_norm": 1121.025146484375,
      "learning_rate": 3.719633253763867e-05,
      "loss": 92.7057,
      "step": 63380
    },
    {
      "epoch": 0.2561036211654149,
      "grad_norm": 721.4158935546875,
      "learning_rate": 3.719490649760603e-05,
      "loss": 77.5612,
      "step": 63390
    },
    {
      "epoch": 0.25614402243078255,
      "grad_norm": 0.0,
      "learning_rate": 3.7193480122349824e-05,
      "loss": 66.09,
      "step": 63400
    },
    {
      "epoch": 0.2561844236961502,
      "grad_norm": 998.5385131835938,
      "learning_rate": 3.7192053411897876e-05,
      "loss": 77.9643,
      "step": 63410
    },
    {
      "epoch": 0.25622482496151777,
      "grad_norm": 1532.927001953125,
      "learning_rate": 3.7190626366278e-05,
      "loss": 105.0879,
      "step": 63420
    },
    {
      "epoch": 0.2562652262268854,
      "grad_norm": 1900.2274169921875,
      "learning_rate": 3.718919898551801e-05,
      "loss": 70.104,
      "step": 63430
    },
    {
      "epoch": 0.25630562749225305,
      "grad_norm": 675.3201293945312,
      "learning_rate": 3.718777126964573e-05,
      "loss": 80.4306,
      "step": 63440
    },
    {
      "epoch": 0.2563460287576207,
      "grad_norm": 599.4599609375,
      "learning_rate": 3.718634321868901e-05,
      "loss": 77.3058,
      "step": 63450
    },
    {
      "epoch": 0.25638643002298833,
      "grad_norm": 2024.2564697265625,
      "learning_rate": 3.718491483267568e-05,
      "loss": 126.2976,
      "step": 63460
    },
    {
      "epoch": 0.25642683128835597,
      "grad_norm": 451.1012268066406,
      "learning_rate": 3.7183486111633585e-05,
      "loss": 80.9007,
      "step": 63470
    },
    {
      "epoch": 0.25646723255372356,
      "grad_norm": 560.759765625,
      "learning_rate": 3.7182057055590576e-05,
      "loss": 107.118,
      "step": 63480
    },
    {
      "epoch": 0.2565076338190912,
      "grad_norm": 898.7105102539062,
      "learning_rate": 3.7180627664574524e-05,
      "loss": 90.2963,
      "step": 63490
    },
    {
      "epoch": 0.25654803508445884,
      "grad_norm": 579.5004272460938,
      "learning_rate": 3.717919793861329e-05,
      "loss": 63.7845,
      "step": 63500
    },
    {
      "epoch": 0.2565884363498265,
      "grad_norm": 809.5593872070312,
      "learning_rate": 3.717776787773475e-05,
      "loss": 76.0838,
      "step": 63510
    },
    {
      "epoch": 0.2566288376151941,
      "grad_norm": 491.0308532714844,
      "learning_rate": 3.717633748196678e-05,
      "loss": 71.3151,
      "step": 63520
    },
    {
      "epoch": 0.25666923888056176,
      "grad_norm": 513.8741455078125,
      "learning_rate": 3.717490675133727e-05,
      "loss": 82.5305,
      "step": 63530
    },
    {
      "epoch": 0.2567096401459294,
      "grad_norm": 682.1354370117188,
      "learning_rate": 3.7173475685874105e-05,
      "loss": 58.0706,
      "step": 63540
    },
    {
      "epoch": 0.256750041411297,
      "grad_norm": 783.206298828125,
      "learning_rate": 3.7172044285605195e-05,
      "loss": 53.3748,
      "step": 63550
    },
    {
      "epoch": 0.2567904426766646,
      "grad_norm": 755.7714233398438,
      "learning_rate": 3.717061255055844e-05,
      "loss": 76.3537,
      "step": 63560
    },
    {
      "epoch": 0.25683084394203226,
      "grad_norm": 1231.5157470703125,
      "learning_rate": 3.7169180480761754e-05,
      "loss": 78.1751,
      "step": 63570
    },
    {
      "epoch": 0.2568712452073999,
      "grad_norm": 1083.4071044921875,
      "learning_rate": 3.7167748076243046e-05,
      "loss": 70.2778,
      "step": 63580
    },
    {
      "epoch": 0.25691164647276754,
      "grad_norm": 594.28564453125,
      "learning_rate": 3.716631533703026e-05,
      "loss": 64.8809,
      "step": 63590
    },
    {
      "epoch": 0.2569520477381352,
      "grad_norm": 930.3058471679688,
      "learning_rate": 3.7164882263151315e-05,
      "loss": 79.9638,
      "step": 63600
    },
    {
      "epoch": 0.25699244900350277,
      "grad_norm": 496.72998046875,
      "learning_rate": 3.716344885463416e-05,
      "loss": 80.5776,
      "step": 63610
    },
    {
      "epoch": 0.2570328502688704,
      "grad_norm": 641.936279296875,
      "learning_rate": 3.716201511150673e-05,
      "loss": 66.4667,
      "step": 63620
    },
    {
      "epoch": 0.25707325153423805,
      "grad_norm": 541.667236328125,
      "learning_rate": 3.716058103379697e-05,
      "loss": 72.8316,
      "step": 63630
    },
    {
      "epoch": 0.2571136527996057,
      "grad_norm": 522.942138671875,
      "learning_rate": 3.715914662153286e-05,
      "loss": 81.5492,
      "step": 63640
    },
    {
      "epoch": 0.2571540540649733,
      "grad_norm": 813.4447631835938,
      "learning_rate": 3.715771187474235e-05,
      "loss": 64.2315,
      "step": 63650
    },
    {
      "epoch": 0.25719445533034097,
      "grad_norm": 543.0736083984375,
      "learning_rate": 3.7156276793453414e-05,
      "loss": 56.0146,
      "step": 63660
    },
    {
      "epoch": 0.2572348565957086,
      "grad_norm": 622.9854125976562,
      "learning_rate": 3.715484137769402e-05,
      "loss": 72.0535,
      "step": 63670
    },
    {
      "epoch": 0.2572752578610762,
      "grad_norm": 382.42767333984375,
      "learning_rate": 3.7153405627492175e-05,
      "loss": 78.4168,
      "step": 63680
    },
    {
      "epoch": 0.25731565912644383,
      "grad_norm": 746.149658203125,
      "learning_rate": 3.715196954287585e-05,
      "loss": 47.0814,
      "step": 63690
    },
    {
      "epoch": 0.25735606039181147,
      "grad_norm": 828.6617431640625,
      "learning_rate": 3.715053312387305e-05,
      "loss": 83.7009,
      "step": 63700
    },
    {
      "epoch": 0.2573964616571791,
      "grad_norm": 1059.0740966796875,
      "learning_rate": 3.7149096370511776e-05,
      "loss": 81.3558,
      "step": 63710
    },
    {
      "epoch": 0.25743686292254675,
      "grad_norm": 2411.71142578125,
      "learning_rate": 3.714765928282004e-05,
      "loss": 96.2945,
      "step": 63720
    },
    {
      "epoch": 0.2574772641879144,
      "grad_norm": 867.219482421875,
      "learning_rate": 3.714622186082585e-05,
      "loss": 82.7037,
      "step": 63730
    },
    {
      "epoch": 0.257517665453282,
      "grad_norm": 865.6116333007812,
      "learning_rate": 3.7144784104557246e-05,
      "loss": 77.5183,
      "step": 63740
    },
    {
      "epoch": 0.2575580667186496,
      "grad_norm": 736.4535522460938,
      "learning_rate": 3.714334601404225e-05,
      "loss": 58.0193,
      "step": 63750
    },
    {
      "epoch": 0.25759846798401725,
      "grad_norm": 792.9779052734375,
      "learning_rate": 3.714190758930889e-05,
      "loss": 79.0416,
      "step": 63760
    },
    {
      "epoch": 0.2576388692493849,
      "grad_norm": 539.277099609375,
      "learning_rate": 3.7140468830385227e-05,
      "loss": 64.7736,
      "step": 63770
    },
    {
      "epoch": 0.25767927051475253,
      "grad_norm": 1489.8095703125,
      "learning_rate": 3.713902973729929e-05,
      "loss": 93.2199,
      "step": 63780
    },
    {
      "epoch": 0.2577196717801202,
      "grad_norm": 295.87420654296875,
      "learning_rate": 3.713759031007915e-05,
      "loss": 85.7172,
      "step": 63790
    },
    {
      "epoch": 0.25776007304548776,
      "grad_norm": 1169.1015625,
      "learning_rate": 3.713615054875286e-05,
      "loss": 70.2371,
      "step": 63800
    },
    {
      "epoch": 0.2578004743108554,
      "grad_norm": 1383.2265625,
      "learning_rate": 3.7134710453348496e-05,
      "loss": 74.2921,
      "step": 63810
    },
    {
      "epoch": 0.25784087557622304,
      "grad_norm": 633.1953735351562,
      "learning_rate": 3.713327002389413e-05,
      "loss": 53.2052,
      "step": 63820
    },
    {
      "epoch": 0.2578812768415907,
      "grad_norm": 1091.2930908203125,
      "learning_rate": 3.713182926041785e-05,
      "loss": 97.7233,
      "step": 63830
    },
    {
      "epoch": 0.2579216781069583,
      "grad_norm": 835.3092651367188,
      "learning_rate": 3.7130388162947726e-05,
      "loss": 68.7806,
      "step": 63840
    },
    {
      "epoch": 0.25796207937232596,
      "grad_norm": 811.8614501953125,
      "learning_rate": 3.712894673151187e-05,
      "loss": 76.9277,
      "step": 63850
    },
    {
      "epoch": 0.2580024806376936,
      "grad_norm": 873.2691040039062,
      "learning_rate": 3.712750496613838e-05,
      "loss": 56.6611,
      "step": 63860
    },
    {
      "epoch": 0.2580428819030612,
      "grad_norm": 534.2119140625,
      "learning_rate": 3.712606286685536e-05,
      "loss": 91.2972,
      "step": 63870
    },
    {
      "epoch": 0.2580832831684288,
      "grad_norm": 2092.973876953125,
      "learning_rate": 3.712462043369093e-05,
      "loss": 65.7608,
      "step": 63880
    },
    {
      "epoch": 0.25812368443379646,
      "grad_norm": 1082.7713623046875,
      "learning_rate": 3.712317766667321e-05,
      "loss": 61.7201,
      "step": 63890
    },
    {
      "epoch": 0.2581640856991641,
      "grad_norm": 540.9574584960938,
      "learning_rate": 3.712173456583033e-05,
      "loss": 43.2988,
      "step": 63900
    },
    {
      "epoch": 0.25820448696453174,
      "grad_norm": 719.4511108398438,
      "learning_rate": 3.712029113119041e-05,
      "loss": 43.7259,
      "step": 63910
    },
    {
      "epoch": 0.2582448882298994,
      "grad_norm": 837.5140380859375,
      "learning_rate": 3.7118847362781605e-05,
      "loss": 116.4185,
      "step": 63920
    },
    {
      "epoch": 0.25828528949526697,
      "grad_norm": 1147.0135498046875,
      "learning_rate": 3.711740326063206e-05,
      "loss": 68.8039,
      "step": 63930
    },
    {
      "epoch": 0.2583256907606346,
      "grad_norm": 322.7746887207031,
      "learning_rate": 3.711595882476992e-05,
      "loss": 56.783,
      "step": 63940
    },
    {
      "epoch": 0.25836609202600225,
      "grad_norm": 561.376953125,
      "learning_rate": 3.711451405522335e-05,
      "loss": 56.7755,
      "step": 63950
    },
    {
      "epoch": 0.2584064932913699,
      "grad_norm": 1156.663818359375,
      "learning_rate": 3.711306895202052e-05,
      "loss": 94.3347,
      "step": 63960
    },
    {
      "epoch": 0.2584468945567375,
      "grad_norm": 511.6494445800781,
      "learning_rate": 3.7111623515189605e-05,
      "loss": 53.324,
      "step": 63970
    },
    {
      "epoch": 0.25848729582210517,
      "grad_norm": 855.0453491210938,
      "learning_rate": 3.7110177744758774e-05,
      "loss": 77.657,
      "step": 63980
    },
    {
      "epoch": 0.25852769708747275,
      "grad_norm": 664.17578125,
      "learning_rate": 3.7108731640756224e-05,
      "loss": 57.8758,
      "step": 63990
    },
    {
      "epoch": 0.2585680983528404,
      "grad_norm": 785.47607421875,
      "learning_rate": 3.710728520321014e-05,
      "loss": 56.7535,
      "step": 64000
    },
    {
      "epoch": 0.25860849961820803,
      "grad_norm": 868.9013061523438,
      "learning_rate": 3.710583843214872e-05,
      "loss": 72.5036,
      "step": 64010
    },
    {
      "epoch": 0.25864890088357567,
      "grad_norm": 1320.7911376953125,
      "learning_rate": 3.7104391327600176e-05,
      "loss": 77.4738,
      "step": 64020
    },
    {
      "epoch": 0.2586893021489433,
      "grad_norm": 2751.525390625,
      "learning_rate": 3.710294388959272e-05,
      "loss": 64.6033,
      "step": 64030
    },
    {
      "epoch": 0.25872970341431095,
      "grad_norm": 518.192138671875,
      "learning_rate": 3.7101496118154557e-05,
      "loss": 63.9527,
      "step": 64040
    },
    {
      "epoch": 0.2587701046796786,
      "grad_norm": 556.1539306640625,
      "learning_rate": 3.7100048013313933e-05,
      "loss": 69.788,
      "step": 64050
    },
    {
      "epoch": 0.2588105059450462,
      "grad_norm": 910.7564086914062,
      "learning_rate": 3.709859957509907e-05,
      "loss": 83.9675,
      "step": 64060
    },
    {
      "epoch": 0.2588509072104138,
      "grad_norm": 799.0169677734375,
      "learning_rate": 3.70971508035382e-05,
      "loss": 81.879,
      "step": 64070
    },
    {
      "epoch": 0.25889130847578146,
      "grad_norm": 344.5417785644531,
      "learning_rate": 3.7095701698659576e-05,
      "loss": 62.578,
      "step": 64080
    },
    {
      "epoch": 0.2589317097411491,
      "grad_norm": 988.5333251953125,
      "learning_rate": 3.709425226049145e-05,
      "loss": 70.7378,
      "step": 64090
    },
    {
      "epoch": 0.25897211100651674,
      "grad_norm": 1221.0657958984375,
      "learning_rate": 3.709280248906206e-05,
      "loss": 99.6785,
      "step": 64100
    },
    {
      "epoch": 0.2590125122718844,
      "grad_norm": 850.9677124023438,
      "learning_rate": 3.70913523843997e-05,
      "loss": 103.1617,
      "step": 64110
    },
    {
      "epoch": 0.25905291353725196,
      "grad_norm": 689.6639404296875,
      "learning_rate": 3.708990194653262e-05,
      "loss": 61.4117,
      "step": 64120
    },
    {
      "epoch": 0.2590933148026196,
      "grad_norm": 577.7689208984375,
      "learning_rate": 3.708845117548911e-05,
      "loss": 97.5373,
      "step": 64130
    },
    {
      "epoch": 0.25913371606798724,
      "grad_norm": 963.0814208984375,
      "learning_rate": 3.708700007129744e-05,
      "loss": 74.6624,
      "step": 64140
    },
    {
      "epoch": 0.2591741173333549,
      "grad_norm": 586.4309692382812,
      "learning_rate": 3.7085548633985906e-05,
      "loss": 66.786,
      "step": 64150
    },
    {
      "epoch": 0.2592145185987225,
      "grad_norm": 1598.16455078125,
      "learning_rate": 3.708409686358281e-05,
      "loss": 58.5813,
      "step": 64160
    },
    {
      "epoch": 0.25925491986409016,
      "grad_norm": 395.3716735839844,
      "learning_rate": 3.708264476011645e-05,
      "loss": 65.7176,
      "step": 64170
    },
    {
      "epoch": 0.2592953211294578,
      "grad_norm": 951.0369873046875,
      "learning_rate": 3.708119232361513e-05,
      "loss": 83.8157,
      "step": 64180
    },
    {
      "epoch": 0.2593357223948254,
      "grad_norm": 868.982421875,
      "learning_rate": 3.707973955410717e-05,
      "loss": 110.5201,
      "step": 64190
    },
    {
      "epoch": 0.259376123660193,
      "grad_norm": 1175.660400390625,
      "learning_rate": 3.707828645162091e-05,
      "loss": 59.3471,
      "step": 64200
    },
    {
      "epoch": 0.25941652492556067,
      "grad_norm": 587.8705444335938,
      "learning_rate": 3.7076833016184646e-05,
      "loss": 66.6157,
      "step": 64210
    },
    {
      "epoch": 0.2594569261909283,
      "grad_norm": 882.6228637695312,
      "learning_rate": 3.707537924782673e-05,
      "loss": 49.927,
      "step": 64220
    },
    {
      "epoch": 0.25949732745629595,
      "grad_norm": 404.48907470703125,
      "learning_rate": 3.7073925146575514e-05,
      "loss": 100.524,
      "step": 64230
    },
    {
      "epoch": 0.2595377287216636,
      "grad_norm": 3163.03076171875,
      "learning_rate": 3.707247071245933e-05,
      "loss": 71.4904,
      "step": 64240
    },
    {
      "epoch": 0.25957812998703117,
      "grad_norm": 982.10400390625,
      "learning_rate": 3.707101594550655e-05,
      "loss": 77.9145,
      "step": 64250
    },
    {
      "epoch": 0.2596185312523988,
      "grad_norm": 720.5392456054688,
      "learning_rate": 3.706956084574552e-05,
      "loss": 75.0282,
      "step": 64260
    },
    {
      "epoch": 0.25965893251776645,
      "grad_norm": 754.7142333984375,
      "learning_rate": 3.706810541320462e-05,
      "loss": 53.8455,
      "step": 64270
    },
    {
      "epoch": 0.2596993337831341,
      "grad_norm": 1012.333984375,
      "learning_rate": 3.706664964791221e-05,
      "loss": 75.8129,
      "step": 64280
    },
    {
      "epoch": 0.25973973504850173,
      "grad_norm": 1253.0430908203125,
      "learning_rate": 3.7065193549896676e-05,
      "loss": 87.2013,
      "step": 64290
    },
    {
      "epoch": 0.25978013631386937,
      "grad_norm": 384.3373718261719,
      "learning_rate": 3.706373711918641e-05,
      "loss": 53.5726,
      "step": 64300
    },
    {
      "epoch": 0.25982053757923695,
      "grad_norm": 0.0,
      "learning_rate": 3.706228035580981e-05,
      "loss": 53.8202,
      "step": 64310
    },
    {
      "epoch": 0.2598609388446046,
      "grad_norm": 452.86566162109375,
      "learning_rate": 3.7060823259795256e-05,
      "loss": 88.486,
      "step": 64320
    },
    {
      "epoch": 0.25990134010997223,
      "grad_norm": 1422.01025390625,
      "learning_rate": 3.705936583117118e-05,
      "loss": 94.5883,
      "step": 64330
    },
    {
      "epoch": 0.2599417413753399,
      "grad_norm": 410.7811584472656,
      "learning_rate": 3.7057908069965984e-05,
      "loss": 73.8817,
      "step": 64340
    },
    {
      "epoch": 0.2599821426407075,
      "grad_norm": 943.582763671875,
      "learning_rate": 3.705644997620809e-05,
      "loss": 65.7515,
      "step": 64350
    },
    {
      "epoch": 0.26002254390607515,
      "grad_norm": 836.2688598632812,
      "learning_rate": 3.705499154992592e-05,
      "loss": 93.7108,
      "step": 64360
    },
    {
      "epoch": 0.2600629451714428,
      "grad_norm": 915.242919921875,
      "learning_rate": 3.705353279114791e-05,
      "loss": 76.5149,
      "step": 64370
    },
    {
      "epoch": 0.2601033464368104,
      "grad_norm": 964.8082885742188,
      "learning_rate": 3.7052073699902494e-05,
      "loss": 69.2267,
      "step": 64380
    },
    {
      "epoch": 0.260143747702178,
      "grad_norm": 715.5172119140625,
      "learning_rate": 3.705061427621813e-05,
      "loss": 64.467,
      "step": 64390
    },
    {
      "epoch": 0.26018414896754566,
      "grad_norm": 868.9583129882812,
      "learning_rate": 3.7049154520123256e-05,
      "loss": 77.0933,
      "step": 64400
    },
    {
      "epoch": 0.2602245502329133,
      "grad_norm": 485.1432800292969,
      "learning_rate": 3.7047694431646334e-05,
      "loss": 68.6032,
      "step": 64410
    },
    {
      "epoch": 0.26026495149828094,
      "grad_norm": 731.9483642578125,
      "learning_rate": 3.704623401081584e-05,
      "loss": 44.2681,
      "step": 64420
    },
    {
      "epoch": 0.2603053527636486,
      "grad_norm": 504.07818603515625,
      "learning_rate": 3.7044773257660234e-05,
      "loss": 59.9306,
      "step": 64430
    },
    {
      "epoch": 0.26034575402901616,
      "grad_norm": 872.7208862304688,
      "learning_rate": 3.7043312172207996e-05,
      "loss": 93.4973,
      "step": 64440
    },
    {
      "epoch": 0.2603861552943838,
      "grad_norm": 1535.181640625,
      "learning_rate": 3.7041850754487623e-05,
      "loss": 58.0155,
      "step": 64450
    },
    {
      "epoch": 0.26042655655975144,
      "grad_norm": 554.6906127929688,
      "learning_rate": 3.704038900452758e-05,
      "loss": 68.4291,
      "step": 64460
    },
    {
      "epoch": 0.2604669578251191,
      "grad_norm": 561.7522583007812,
      "learning_rate": 3.7038926922356395e-05,
      "loss": 70.7031,
      "step": 64470
    },
    {
      "epoch": 0.2605073590904867,
      "grad_norm": 1062.995361328125,
      "learning_rate": 3.703746450800255e-05,
      "loss": 55.4427,
      "step": 64480
    },
    {
      "epoch": 0.26054776035585436,
      "grad_norm": 796.6412963867188,
      "learning_rate": 3.703600176149457e-05,
      "loss": 93.2851,
      "step": 64490
    },
    {
      "epoch": 0.260588161621222,
      "grad_norm": 961.7485961914062,
      "learning_rate": 3.703453868286096e-05,
      "loss": 60.0274,
      "step": 64500
    },
    {
      "epoch": 0.2606285628865896,
      "grad_norm": 638.6123657226562,
      "learning_rate": 3.703307527213024e-05,
      "loss": 85.6026,
      "step": 64510
    },
    {
      "epoch": 0.2606689641519572,
      "grad_norm": 1373.725830078125,
      "learning_rate": 3.7031611529330956e-05,
      "loss": 82.0446,
      "step": 64520
    },
    {
      "epoch": 0.26070936541732487,
      "grad_norm": 1274.551513671875,
      "learning_rate": 3.703014745449164e-05,
      "loss": 91.4852,
      "step": 64530
    },
    {
      "epoch": 0.2607497666826925,
      "grad_norm": 805.552001953125,
      "learning_rate": 3.702868304764083e-05,
      "loss": 67.7787,
      "step": 64540
    },
    {
      "epoch": 0.26079016794806015,
      "grad_norm": 1122.33837890625,
      "learning_rate": 3.702721830880707e-05,
      "loss": 140.0442,
      "step": 64550
    },
    {
      "epoch": 0.2608305692134278,
      "grad_norm": 1279.6610107421875,
      "learning_rate": 3.702575323801893e-05,
      "loss": 102.1003,
      "step": 64560
    },
    {
      "epoch": 0.26087097047879537,
      "grad_norm": 540.7999267578125,
      "learning_rate": 3.702428783530497e-05,
      "loss": 65.1975,
      "step": 64570
    },
    {
      "epoch": 0.260911371744163,
      "grad_norm": 576.8917846679688,
      "learning_rate": 3.7022822100693746e-05,
      "loss": 82.8855,
      "step": 64580
    },
    {
      "epoch": 0.26095177300953065,
      "grad_norm": 469.16571044921875,
      "learning_rate": 3.702135603421385e-05,
      "loss": 67.3132,
      "step": 64590
    },
    {
      "epoch": 0.2609921742748983,
      "grad_norm": 452.0544738769531,
      "learning_rate": 3.701988963589384e-05,
      "loss": 104.5678,
      "step": 64600
    },
    {
      "epoch": 0.26103257554026593,
      "grad_norm": 476.0960998535156,
      "learning_rate": 3.701842290576233e-05,
      "loss": 82.0795,
      "step": 64610
    },
    {
      "epoch": 0.26107297680563357,
      "grad_norm": 1001.22314453125,
      "learning_rate": 3.70169558438479e-05,
      "loss": 97.6005,
      "step": 64620
    },
    {
      "epoch": 0.26111337807100116,
      "grad_norm": 1013.3632202148438,
      "learning_rate": 3.701548845017915e-05,
      "loss": 77.4689,
      "step": 64630
    },
    {
      "epoch": 0.2611537793363688,
      "grad_norm": 640.3134765625,
      "learning_rate": 3.7014020724784703e-05,
      "loss": 103.3352,
      "step": 64640
    },
    {
      "epoch": 0.26119418060173644,
      "grad_norm": 870.0180053710938,
      "learning_rate": 3.701255266769316e-05,
      "loss": 63.6067,
      "step": 64650
    },
    {
      "epoch": 0.2612345818671041,
      "grad_norm": 224.58753967285156,
      "learning_rate": 3.701108427893314e-05,
      "loss": 50.96,
      "step": 64660
    },
    {
      "epoch": 0.2612749831324717,
      "grad_norm": 1057.7576904296875,
      "learning_rate": 3.7009615558533275e-05,
      "loss": 49.2088,
      "step": 64670
    },
    {
      "epoch": 0.26131538439783936,
      "grad_norm": 899.7709350585938,
      "learning_rate": 3.70081465065222e-05,
      "loss": 88.2398,
      "step": 64680
    },
    {
      "epoch": 0.261355785663207,
      "grad_norm": 400.64068603515625,
      "learning_rate": 3.7006677122928546e-05,
      "loss": 86.5793,
      "step": 64690
    },
    {
      "epoch": 0.2613961869285746,
      "grad_norm": 534.3073120117188,
      "learning_rate": 3.7005207407780975e-05,
      "loss": 86.3469,
      "step": 64700
    },
    {
      "epoch": 0.2614365881939422,
      "grad_norm": 836.5368041992188,
      "learning_rate": 3.7003737361108124e-05,
      "loss": 67.9904,
      "step": 64710
    },
    {
      "epoch": 0.26147698945930986,
      "grad_norm": 614.4078369140625,
      "learning_rate": 3.700226698293866e-05,
      "loss": 38.8319,
      "step": 64720
    },
    {
      "epoch": 0.2615173907246775,
      "grad_norm": 999.4403686523438,
      "learning_rate": 3.7000796273301254e-05,
      "loss": 76.9349,
      "step": 64730
    },
    {
      "epoch": 0.26155779199004514,
      "grad_norm": 1237.0614013671875,
      "learning_rate": 3.6999325232224564e-05,
      "loss": 61.3102,
      "step": 64740
    },
    {
      "epoch": 0.2615981932554128,
      "grad_norm": 1526.6424560546875,
      "learning_rate": 3.699785385973728e-05,
      "loss": 75.9351,
      "step": 64750
    },
    {
      "epoch": 0.26163859452078037,
      "grad_norm": 1123.9716796875,
      "learning_rate": 3.6996382155868084e-05,
      "loss": 53.3472,
      "step": 64760
    },
    {
      "epoch": 0.261678995786148,
      "grad_norm": 681.2954711914062,
      "learning_rate": 3.699491012064567e-05,
      "loss": 80.2444,
      "step": 64770
    },
    {
      "epoch": 0.26171939705151565,
      "grad_norm": 490.217529296875,
      "learning_rate": 3.6993437754098734e-05,
      "loss": 74.4671,
      "step": 64780
    },
    {
      "epoch": 0.2617597983168833,
      "grad_norm": 731.5593872070312,
      "learning_rate": 3.6991965056255976e-05,
      "loss": 76.7609,
      "step": 64790
    },
    {
      "epoch": 0.2618001995822509,
      "grad_norm": 443.29791259765625,
      "learning_rate": 3.6990492027146114e-05,
      "loss": 53.8047,
      "step": 64800
    },
    {
      "epoch": 0.26184060084761857,
      "grad_norm": 1002.4287109375,
      "learning_rate": 3.698901866679786e-05,
      "loss": 75.2404,
      "step": 64810
    },
    {
      "epoch": 0.2618810021129862,
      "grad_norm": 0.0,
      "learning_rate": 3.698754497523994e-05,
      "loss": 73.0152,
      "step": 64820
    },
    {
      "epoch": 0.2619214033783538,
      "grad_norm": 466.6201171875,
      "learning_rate": 3.6986070952501085e-05,
      "loss": 58.1316,
      "step": 64830
    },
    {
      "epoch": 0.26196180464372143,
      "grad_norm": 515.5542602539062,
      "learning_rate": 3.698459659861003e-05,
      "loss": 84.0784,
      "step": 64840
    },
    {
      "epoch": 0.26200220590908907,
      "grad_norm": 540.0292358398438,
      "learning_rate": 3.698312191359553e-05,
      "loss": 76.5521,
      "step": 64850
    },
    {
      "epoch": 0.2620426071744567,
      "grad_norm": 611.6865844726562,
      "learning_rate": 3.698164689748631e-05,
      "loss": 57.7677,
      "step": 64860
    },
    {
      "epoch": 0.26208300843982435,
      "grad_norm": 545.26416015625,
      "learning_rate": 3.6980171550311156e-05,
      "loss": 58.3426,
      "step": 64870
    },
    {
      "epoch": 0.262123409705192,
      "grad_norm": 954.208984375,
      "learning_rate": 3.6978695872098806e-05,
      "loss": 115.1971,
      "step": 64880
    },
    {
      "epoch": 0.2621638109705596,
      "grad_norm": 882.8229370117188,
      "learning_rate": 3.697721986287804e-05,
      "loss": 54.8908,
      "step": 64890
    },
    {
      "epoch": 0.2622042122359272,
      "grad_norm": 857.4375610351562,
      "learning_rate": 3.697574352267764e-05,
      "loss": 78.1034,
      "step": 64900
    },
    {
      "epoch": 0.26224461350129485,
      "grad_norm": 549.2964477539062,
      "learning_rate": 3.697426685152637e-05,
      "loss": 57.651,
      "step": 64910
    },
    {
      "epoch": 0.2622850147666625,
      "grad_norm": 983.4130249023438,
      "learning_rate": 3.697278984945304e-05,
      "loss": 68.4534,
      "step": 64920
    },
    {
      "epoch": 0.26232541603203013,
      "grad_norm": 612.8244018554688,
      "learning_rate": 3.6971312516486426e-05,
      "loss": 95.7964,
      "step": 64930
    },
    {
      "epoch": 0.2623658172973978,
      "grad_norm": 713.4525146484375,
      "learning_rate": 3.6969834852655334e-05,
      "loss": 64.7911,
      "step": 64940
    },
    {
      "epoch": 0.26240621856276536,
      "grad_norm": 3631.888671875,
      "learning_rate": 3.696835685798858e-05,
      "loss": 86.5216,
      "step": 64950
    },
    {
      "epoch": 0.262446619828133,
      "grad_norm": 1668.7647705078125,
      "learning_rate": 3.696687853251497e-05,
      "loss": 108.1082,
      "step": 64960
    },
    {
      "epoch": 0.26248702109350064,
      "grad_norm": 621.1637573242188,
      "learning_rate": 3.696539987626334e-05,
      "loss": 69.7315,
      "step": 64970
    },
    {
      "epoch": 0.2625274223588683,
      "grad_norm": 670.1963500976562,
      "learning_rate": 3.696392088926248e-05,
      "loss": 94.8227,
      "step": 64980
    },
    {
      "epoch": 0.2625678236242359,
      "grad_norm": 395.003662109375,
      "learning_rate": 3.696244157154128e-05,
      "loss": 55.8034,
      "step": 64990
    },
    {
      "epoch": 0.26260822488960356,
      "grad_norm": 789.5413208007812,
      "learning_rate": 3.696096192312852e-05,
      "loss": 90.5435,
      "step": 65000
    },
    {
      "epoch": 0.2626486261549712,
      "grad_norm": 598.6563720703125,
      "learning_rate": 3.695948194405309e-05,
      "loss": 60.1081,
      "step": 65010
    },
    {
      "epoch": 0.2626890274203388,
      "grad_norm": 290.1271057128906,
      "learning_rate": 3.6958001634343825e-05,
      "loss": 80.8175,
      "step": 65020
    },
    {
      "epoch": 0.2627294286857064,
      "grad_norm": 553.1530151367188,
      "learning_rate": 3.695652099402959e-05,
      "loss": 40.2118,
      "step": 65030
    },
    {
      "epoch": 0.26276982995107406,
      "grad_norm": 2150.581298828125,
      "learning_rate": 3.695504002313924e-05,
      "loss": 82.7208,
      "step": 65040
    },
    {
      "epoch": 0.2628102312164417,
      "grad_norm": 837.5015258789062,
      "learning_rate": 3.6953558721701666e-05,
      "loss": 78.5257,
      "step": 65050
    },
    {
      "epoch": 0.26285063248180934,
      "grad_norm": 428.619140625,
      "learning_rate": 3.6952077089745735e-05,
      "loss": 84.7003,
      "step": 65060
    },
    {
      "epoch": 0.262891033747177,
      "grad_norm": 990.3151245117188,
      "learning_rate": 3.695059512730033e-05,
      "loss": 78.2993,
      "step": 65070
    },
    {
      "epoch": 0.26293143501254457,
      "grad_norm": 502.7200927734375,
      "learning_rate": 3.694911283439435e-05,
      "loss": 88.1456,
      "step": 65080
    },
    {
      "epoch": 0.2629718362779122,
      "grad_norm": 1459.866943359375,
      "learning_rate": 3.6947630211056684e-05,
      "loss": 58.0296,
      "step": 65090
    },
    {
      "epoch": 0.26301223754327985,
      "grad_norm": 583.2791748046875,
      "learning_rate": 3.6946147257316247e-05,
      "loss": 63.4767,
      "step": 65100
    },
    {
      "epoch": 0.2630526388086475,
      "grad_norm": 572.3054809570312,
      "learning_rate": 3.6944663973201945e-05,
      "loss": 50.8657,
      "step": 65110
    },
    {
      "epoch": 0.2630930400740151,
      "grad_norm": 576.8928833007812,
      "learning_rate": 3.694318035874269e-05,
      "loss": 58.8146,
      "step": 65120
    },
    {
      "epoch": 0.26313344133938277,
      "grad_norm": 699.8406982421875,
      "learning_rate": 3.694169641396741e-05,
      "loss": 65.0462,
      "step": 65130
    },
    {
      "epoch": 0.2631738426047504,
      "grad_norm": 545.7052001953125,
      "learning_rate": 3.6940212138905043e-05,
      "loss": 64.3889,
      "step": 65140
    },
    {
      "epoch": 0.263214243870118,
      "grad_norm": 966.1185302734375,
      "learning_rate": 3.6938727533584515e-05,
      "loss": 56.8199,
      "step": 65150
    },
    {
      "epoch": 0.26325464513548563,
      "grad_norm": 418.309326171875,
      "learning_rate": 3.6937242598034776e-05,
      "loss": 53.9059,
      "step": 65160
    },
    {
      "epoch": 0.26329504640085327,
      "grad_norm": 0.0,
      "learning_rate": 3.693575733228477e-05,
      "loss": 71.7491,
      "step": 65170
    },
    {
      "epoch": 0.2633354476662209,
      "grad_norm": 626.2059326171875,
      "learning_rate": 3.6934271736363455e-05,
      "loss": 43.8876,
      "step": 65180
    },
    {
      "epoch": 0.26337584893158855,
      "grad_norm": 472.5262756347656,
      "learning_rate": 3.69327858102998e-05,
      "loss": 93.9019,
      "step": 65190
    },
    {
      "epoch": 0.2634162501969562,
      "grad_norm": 1224.383544921875,
      "learning_rate": 3.6931299554122754e-05,
      "loss": 64.856,
      "step": 65200
    },
    {
      "epoch": 0.2634566514623238,
      "grad_norm": 1208.1915283203125,
      "learning_rate": 3.692981296786132e-05,
      "loss": 88.3372,
      "step": 65210
    },
    {
      "epoch": 0.2634970527276914,
      "grad_norm": 838.8875122070312,
      "learning_rate": 3.692832605154446e-05,
      "loss": 68.6252,
      "step": 65220
    },
    {
      "epoch": 0.26353745399305906,
      "grad_norm": 2221.710693359375,
      "learning_rate": 3.692683880520117e-05,
      "loss": 79.5289,
      "step": 65230
    },
    {
      "epoch": 0.2635778552584267,
      "grad_norm": 1237.591064453125,
      "learning_rate": 3.6925351228860445e-05,
      "loss": 77.699,
      "step": 65240
    },
    {
      "epoch": 0.26361825652379434,
      "grad_norm": 549.4052734375,
      "learning_rate": 3.6923863322551284e-05,
      "loss": 67.4263,
      "step": 65250
    },
    {
      "epoch": 0.263658657789162,
      "grad_norm": 614.0408935546875,
      "learning_rate": 3.692237508630269e-05,
      "loss": 55.3787,
      "step": 65260
    },
    {
      "epoch": 0.26369905905452956,
      "grad_norm": 1487.8175048828125,
      "learning_rate": 3.6920886520143684e-05,
      "loss": 56.1122,
      "step": 65270
    },
    {
      "epoch": 0.2637394603198972,
      "grad_norm": 468.63720703125,
      "learning_rate": 3.6919397624103284e-05,
      "loss": 60.9861,
      "step": 65280
    },
    {
      "epoch": 0.26377986158526484,
      "grad_norm": 1053.509033203125,
      "learning_rate": 3.691790839821052e-05,
      "loss": 47.5022,
      "step": 65290
    },
    {
      "epoch": 0.2638202628506325,
      "grad_norm": 731.3139038085938,
      "learning_rate": 3.6916418842494416e-05,
      "loss": 56.3728,
      "step": 65300
    },
    {
      "epoch": 0.2638606641160001,
      "grad_norm": 652.5587768554688,
      "learning_rate": 3.691492895698402e-05,
      "loss": 57.3569,
      "step": 65310
    },
    {
      "epoch": 0.26390106538136776,
      "grad_norm": 672.2579345703125,
      "learning_rate": 3.691343874170838e-05,
      "loss": 87.096,
      "step": 65320
    },
    {
      "epoch": 0.2639414666467354,
      "grad_norm": 789.9293212890625,
      "learning_rate": 3.6911948196696533e-05,
      "loss": 58.2702,
      "step": 65330
    },
    {
      "epoch": 0.263981867912103,
      "grad_norm": 682.2620239257812,
      "learning_rate": 3.691045732197756e-05,
      "loss": 77.8977,
      "step": 65340
    },
    {
      "epoch": 0.2640222691774706,
      "grad_norm": 985.931884765625,
      "learning_rate": 3.690896611758051e-05,
      "loss": 65.8317,
      "step": 65350
    },
    {
      "epoch": 0.26406267044283827,
      "grad_norm": 732.6767578125,
      "learning_rate": 3.690747458353446e-05,
      "loss": 88.3303,
      "step": 65360
    },
    {
      "epoch": 0.2641030717082059,
      "grad_norm": 416.494140625,
      "learning_rate": 3.6905982719868493e-05,
      "loss": 82.3161,
      "step": 65370
    },
    {
      "epoch": 0.26414347297357355,
      "grad_norm": 656.120361328125,
      "learning_rate": 3.6904490526611684e-05,
      "loss": 89.1459,
      "step": 65380
    },
    {
      "epoch": 0.2641838742389412,
      "grad_norm": 381.7622985839844,
      "learning_rate": 3.690299800379313e-05,
      "loss": 47.3332,
      "step": 65390
    },
    {
      "epoch": 0.26422427550430877,
      "grad_norm": 578.4981079101562,
      "learning_rate": 3.6901505151441935e-05,
      "loss": 96.7956,
      "step": 65400
    },
    {
      "epoch": 0.2642646767696764,
      "grad_norm": 952.101318359375,
      "learning_rate": 3.690001196958719e-05,
      "loss": 92.0501,
      "step": 65410
    },
    {
      "epoch": 0.26430507803504405,
      "grad_norm": 681.7398071289062,
      "learning_rate": 3.6898518458258006e-05,
      "loss": 47.1945,
      "step": 65420
    },
    {
      "epoch": 0.2643454793004117,
      "grad_norm": 132.1608123779297,
      "learning_rate": 3.689702461748351e-05,
      "loss": 81.5882,
      "step": 65430
    },
    {
      "epoch": 0.26438588056577933,
      "grad_norm": 1409.0716552734375,
      "learning_rate": 3.689553044729282e-05,
      "loss": 96.7558,
      "step": 65440
    },
    {
      "epoch": 0.26442628183114697,
      "grad_norm": 580.634521484375,
      "learning_rate": 3.689403594771506e-05,
      "loss": 65.5121,
      "step": 65450
    },
    {
      "epoch": 0.2644666830965146,
      "grad_norm": 1463.0916748046875,
      "learning_rate": 3.689254111877938e-05,
      "loss": 69.6791,
      "step": 65460
    },
    {
      "epoch": 0.2645070843618822,
      "grad_norm": 1450.8865966796875,
      "learning_rate": 3.6891045960514904e-05,
      "loss": 67.0916,
      "step": 65470
    },
    {
      "epoch": 0.26454748562724983,
      "grad_norm": 592.7374267578125,
      "learning_rate": 3.6889550472950804e-05,
      "loss": 79.5903,
      "step": 65480
    },
    {
      "epoch": 0.2645878868926175,
      "grad_norm": 473.9434509277344,
      "learning_rate": 3.688805465611621e-05,
      "loss": 62.0096,
      "step": 65490
    },
    {
      "epoch": 0.2646282881579851,
      "grad_norm": 664.6771240234375,
      "learning_rate": 3.6886558510040305e-05,
      "loss": 47.8189,
      "step": 65500
    },
    {
      "epoch": 0.26466868942335275,
      "grad_norm": 791.1522216796875,
      "learning_rate": 3.6885062034752244e-05,
      "loss": 67.1695,
      "step": 65510
    },
    {
      "epoch": 0.2647090906887204,
      "grad_norm": 638.1292114257812,
      "learning_rate": 3.6883565230281205e-05,
      "loss": 51.381,
      "step": 65520
    },
    {
      "epoch": 0.264749491954088,
      "grad_norm": 521.1875,
      "learning_rate": 3.688206809665637e-05,
      "loss": 51.432,
      "step": 65530
    },
    {
      "epoch": 0.2647898932194556,
      "grad_norm": 658.73095703125,
      "learning_rate": 3.688057063390693e-05,
      "loss": 56.7711,
      "step": 65540
    },
    {
      "epoch": 0.26483029448482326,
      "grad_norm": 562.1188354492188,
      "learning_rate": 3.687907284206207e-05,
      "loss": 54.7807,
      "step": 65550
    },
    {
      "epoch": 0.2648706957501909,
      "grad_norm": 474.15313720703125,
      "learning_rate": 3.6877574721151e-05,
      "loss": 86.3629,
      "step": 65560
    },
    {
      "epoch": 0.26491109701555854,
      "grad_norm": 870.3119506835938,
      "learning_rate": 3.687607627120291e-05,
      "loss": 76.91,
      "step": 65570
    },
    {
      "epoch": 0.2649514982809262,
      "grad_norm": 495.1028747558594,
      "learning_rate": 3.6874577492247036e-05,
      "loss": 103.5563,
      "step": 65580
    },
    {
      "epoch": 0.26499189954629376,
      "grad_norm": 775.1270751953125,
      "learning_rate": 3.687307838431258e-05,
      "loss": 63.2456,
      "step": 65590
    },
    {
      "epoch": 0.2650323008116614,
      "grad_norm": 761.67333984375,
      "learning_rate": 3.687157894742878e-05,
      "loss": 92.0111,
      "step": 65600
    },
    {
      "epoch": 0.26507270207702904,
      "grad_norm": 579.744873046875,
      "learning_rate": 3.687007918162486e-05,
      "loss": 52.8847,
      "step": 65610
    },
    {
      "epoch": 0.2651131033423967,
      "grad_norm": 929.838134765625,
      "learning_rate": 3.686857908693006e-05,
      "loss": 52.4094,
      "step": 65620
    },
    {
      "epoch": 0.2651535046077643,
      "grad_norm": 833.6460571289062,
      "learning_rate": 3.6867078663373624e-05,
      "loss": 86.709,
      "step": 65630
    },
    {
      "epoch": 0.26519390587313196,
      "grad_norm": 485.7236633300781,
      "learning_rate": 3.686557791098481e-05,
      "loss": 71.4412,
      "step": 65640
    },
    {
      "epoch": 0.2652343071384996,
      "grad_norm": 1216.905029296875,
      "learning_rate": 3.6864076829792865e-05,
      "loss": 87.901,
      "step": 65650
    },
    {
      "epoch": 0.2652747084038672,
      "grad_norm": 923.685546875,
      "learning_rate": 3.686257541982706e-05,
      "loss": 85.9243,
      "step": 65660
    },
    {
      "epoch": 0.2653151096692348,
      "grad_norm": 951.8165283203125,
      "learning_rate": 3.6861073681116674e-05,
      "loss": 81.3739,
      "step": 65670
    },
    {
      "epoch": 0.26535551093460247,
      "grad_norm": 342.7435607910156,
      "learning_rate": 3.685957161369098e-05,
      "loss": 68.2152,
      "step": 65680
    },
    {
      "epoch": 0.2653959121999701,
      "grad_norm": 2444.807373046875,
      "learning_rate": 3.685806921757925e-05,
      "loss": 86.507,
      "step": 65690
    },
    {
      "epoch": 0.26543631346533775,
      "grad_norm": 1393.3995361328125,
      "learning_rate": 3.685656649281078e-05,
      "loss": 67.6843,
      "step": 65700
    },
    {
      "epoch": 0.2654767147307054,
      "grad_norm": 1030.308837890625,
      "learning_rate": 3.6855063439414866e-05,
      "loss": 49.6981,
      "step": 65710
    },
    {
      "epoch": 0.26551711599607297,
      "grad_norm": 722.55810546875,
      "learning_rate": 3.685356005742082e-05,
      "loss": 86.8199,
      "step": 65720
    },
    {
      "epoch": 0.2655575172614406,
      "grad_norm": 2299.416259765625,
      "learning_rate": 3.685205634685794e-05,
      "loss": 115.0299,
      "step": 65730
    },
    {
      "epoch": 0.26559791852680825,
      "grad_norm": 761.451904296875,
      "learning_rate": 3.685055230775554e-05,
      "loss": 57.6311,
      "step": 65740
    },
    {
      "epoch": 0.2656383197921759,
      "grad_norm": 679.90625,
      "learning_rate": 3.684904794014296e-05,
      "loss": 71.5767,
      "step": 65750
    },
    {
      "epoch": 0.26567872105754353,
      "grad_norm": 1391.4635009765625,
      "learning_rate": 3.6847543244049505e-05,
      "loss": 48.7501,
      "step": 65760
    },
    {
      "epoch": 0.26571912232291117,
      "grad_norm": 581.217041015625,
      "learning_rate": 3.684603821950452e-05,
      "loss": 80.6224,
      "step": 65770
    },
    {
      "epoch": 0.2657595235882788,
      "grad_norm": 960.2913818359375,
      "learning_rate": 3.6844532866537355e-05,
      "loss": 106.6197,
      "step": 65780
    },
    {
      "epoch": 0.2657999248536464,
      "grad_norm": 1181.9364013671875,
      "learning_rate": 3.684302718517734e-05,
      "loss": 74.9867,
      "step": 65790
    },
    {
      "epoch": 0.26584032611901404,
      "grad_norm": 480.9149475097656,
      "learning_rate": 3.684152117545385e-05,
      "loss": 91.1819,
      "step": 65800
    },
    {
      "epoch": 0.2658807273843817,
      "grad_norm": 1295.970458984375,
      "learning_rate": 3.684001483739623e-05,
      "loss": 88.7823,
      "step": 65810
    },
    {
      "epoch": 0.2659211286497493,
      "grad_norm": 447.7242431640625,
      "learning_rate": 3.6838508171033846e-05,
      "loss": 68.8692,
      "step": 65820
    },
    {
      "epoch": 0.26596152991511696,
      "grad_norm": 556.4803466796875,
      "learning_rate": 3.6837001176396084e-05,
      "loss": 69.6128,
      "step": 65830
    },
    {
      "epoch": 0.2660019311804846,
      "grad_norm": 856.51513671875,
      "learning_rate": 3.683549385351231e-05,
      "loss": 55.587,
      "step": 65840
    },
    {
      "epoch": 0.2660423324458522,
      "grad_norm": 908.120849609375,
      "learning_rate": 3.6833986202411914e-05,
      "loss": 69.7311,
      "step": 65850
    },
    {
      "epoch": 0.2660827337112198,
      "grad_norm": 640.2660522460938,
      "learning_rate": 3.68324782231243e-05,
      "loss": 70.6052,
      "step": 65860
    },
    {
      "epoch": 0.26612313497658746,
      "grad_norm": 524.6282348632812,
      "learning_rate": 3.683096991567885e-05,
      "loss": 49.6814,
      "step": 65870
    },
    {
      "epoch": 0.2661635362419551,
      "grad_norm": 1259.0841064453125,
      "learning_rate": 3.682946128010498e-05,
      "loss": 66.3145,
      "step": 65880
    },
    {
      "epoch": 0.26620393750732274,
      "grad_norm": 1364.9793701171875,
      "learning_rate": 3.682795231643209e-05,
      "loss": 84.0134,
      "step": 65890
    },
    {
      "epoch": 0.2662443387726904,
      "grad_norm": 1976.0228271484375,
      "learning_rate": 3.682644302468961e-05,
      "loss": 84.8792,
      "step": 65900
    },
    {
      "epoch": 0.26628474003805797,
      "grad_norm": 616.8114013671875,
      "learning_rate": 3.682493340490697e-05,
      "loss": 98.148,
      "step": 65910
    },
    {
      "epoch": 0.2663251413034256,
      "grad_norm": 1352.3936767578125,
      "learning_rate": 3.6823423457113575e-05,
      "loss": 70.8769,
      "step": 65920
    },
    {
      "epoch": 0.26636554256879325,
      "grad_norm": 833.4444580078125,
      "learning_rate": 3.682191318133889e-05,
      "loss": 83.6237,
      "step": 65930
    },
    {
      "epoch": 0.2664059438341609,
      "grad_norm": 276.592529296875,
      "learning_rate": 3.6820402577612336e-05,
      "loss": 72.8676,
      "step": 65940
    },
    {
      "epoch": 0.2664463450995285,
      "grad_norm": 521.146728515625,
      "learning_rate": 3.681889164596339e-05,
      "loss": 38.219,
      "step": 65950
    },
    {
      "epoch": 0.26648674636489617,
      "grad_norm": 1227.583740234375,
      "learning_rate": 3.6817380386421475e-05,
      "loss": 56.357,
      "step": 65960
    },
    {
      "epoch": 0.2665271476302638,
      "grad_norm": 928.7647094726562,
      "learning_rate": 3.681586879901608e-05,
      "loss": 72.9631,
      "step": 65970
    },
    {
      "epoch": 0.2665675488956314,
      "grad_norm": 645.942138671875,
      "learning_rate": 3.6814356883776665e-05,
      "loss": 83.9413,
      "step": 65980
    },
    {
      "epoch": 0.26660795016099903,
      "grad_norm": 1118.24365234375,
      "learning_rate": 3.681284464073271e-05,
      "loss": 78.2679,
      "step": 65990
    },
    {
      "epoch": 0.26664835142636667,
      "grad_norm": 427.3819580078125,
      "learning_rate": 3.6811332069913685e-05,
      "loss": 47.7537,
      "step": 66000
    },
    {
      "epoch": 0.2666887526917343,
      "grad_norm": 1441.3486328125,
      "learning_rate": 3.680981917134909e-05,
      "loss": 66.2791,
      "step": 66010
    },
    {
      "epoch": 0.26672915395710195,
      "grad_norm": 621.1433715820312,
      "learning_rate": 3.6808305945068424e-05,
      "loss": 113.461,
      "step": 66020
    },
    {
      "epoch": 0.2667695552224696,
      "grad_norm": 705.9483642578125,
      "learning_rate": 3.6806792391101166e-05,
      "loss": 81.4384,
      "step": 66030
    },
    {
      "epoch": 0.2668099564878372,
      "grad_norm": 2083.895751953125,
      "learning_rate": 3.6805278509476844e-05,
      "loss": 65.3315,
      "step": 66040
    },
    {
      "epoch": 0.2668503577532048,
      "grad_norm": 1001.9884033203125,
      "learning_rate": 3.680376430022497e-05,
      "loss": 50.324,
      "step": 66050
    },
    {
      "epoch": 0.26689075901857245,
      "grad_norm": 618.3958740234375,
      "learning_rate": 3.680224976337505e-05,
      "loss": 66.6406,
      "step": 66060
    },
    {
      "epoch": 0.2669311602839401,
      "grad_norm": 531.05712890625,
      "learning_rate": 3.680073489895663e-05,
      "loss": 69.8654,
      "step": 66070
    },
    {
      "epoch": 0.26697156154930773,
      "grad_norm": 780.5286254882812,
      "learning_rate": 3.679921970699923e-05,
      "loss": 63.6973,
      "step": 66080
    },
    {
      "epoch": 0.2670119628146754,
      "grad_norm": 1168.5731201171875,
      "learning_rate": 3.679770418753239e-05,
      "loss": 64.4696,
      "step": 66090
    },
    {
      "epoch": 0.267052364080043,
      "grad_norm": 461.4474182128906,
      "learning_rate": 3.679618834058566e-05,
      "loss": 63.9347,
      "step": 66100
    },
    {
      "epoch": 0.2670927653454106,
      "grad_norm": 317.3716125488281,
      "learning_rate": 3.6794672166188595e-05,
      "loss": 74.5739,
      "step": 66110
    },
    {
      "epoch": 0.26713316661077824,
      "grad_norm": 397.1957092285156,
      "learning_rate": 3.679315566437074e-05,
      "loss": 56.8725,
      "step": 66120
    },
    {
      "epoch": 0.2671735678761459,
      "grad_norm": 505.0995788574219,
      "learning_rate": 3.679163883516168e-05,
      "loss": 84.3939,
      "step": 66130
    },
    {
      "epoch": 0.2672139691415135,
      "grad_norm": 3382.34033203125,
      "learning_rate": 3.6790121678590975e-05,
      "loss": 71.4125,
      "step": 66140
    },
    {
      "epoch": 0.26725437040688116,
      "grad_norm": 391.44189453125,
      "learning_rate": 3.6788604194688205e-05,
      "loss": 55.7094,
      "step": 66150
    },
    {
      "epoch": 0.2672947716722488,
      "grad_norm": 562.6239013671875,
      "learning_rate": 3.6787086383482946e-05,
      "loss": 60.2443,
      "step": 66160
    },
    {
      "epoch": 0.2673351729376164,
      "grad_norm": 754.8357543945312,
      "learning_rate": 3.6785568245004796e-05,
      "loss": 69.7856,
      "step": 66170
    },
    {
      "epoch": 0.267375574202984,
      "grad_norm": 623.0692138671875,
      "learning_rate": 3.6784049779283355e-05,
      "loss": 56.0616,
      "step": 66180
    },
    {
      "epoch": 0.26741597546835166,
      "grad_norm": 1186.1983642578125,
      "learning_rate": 3.678253098634822e-05,
      "loss": 78.3309,
      "step": 66190
    },
    {
      "epoch": 0.2674563767337193,
      "grad_norm": 398.3062438964844,
      "learning_rate": 3.678101186622901e-05,
      "loss": 64.1084,
      "step": 66200
    },
    {
      "epoch": 0.26749677799908694,
      "grad_norm": 1136.1226806640625,
      "learning_rate": 3.6779492418955324e-05,
      "loss": 57.2094,
      "step": 66210
    },
    {
      "epoch": 0.2675371792644546,
      "grad_norm": 658.5238647460938,
      "learning_rate": 3.67779726445568e-05,
      "loss": 96.4897,
      "step": 66220
    },
    {
      "epoch": 0.26757758052982217,
      "grad_norm": 653.9174194335938,
      "learning_rate": 3.677645254306306e-05,
      "loss": 61.6924,
      "step": 66230
    },
    {
      "epoch": 0.2676179817951898,
      "grad_norm": 707.5989990234375,
      "learning_rate": 3.6774932114503743e-05,
      "loss": 71.5184,
      "step": 66240
    },
    {
      "epoch": 0.26765838306055745,
      "grad_norm": 1067.349609375,
      "learning_rate": 3.6773411358908486e-05,
      "loss": 86.1329,
      "step": 66250
    },
    {
      "epoch": 0.2676987843259251,
      "grad_norm": 1056.942138671875,
      "learning_rate": 3.677189027630694e-05,
      "loss": 57.0639,
      "step": 66260
    },
    {
      "epoch": 0.2677391855912927,
      "grad_norm": 604.6203002929688,
      "learning_rate": 3.6770368866728756e-05,
      "loss": 70.5156,
      "step": 66270
    },
    {
      "epoch": 0.26777958685666037,
      "grad_norm": 1239.1727294921875,
      "learning_rate": 3.6768847130203595e-05,
      "loss": 68.9191,
      "step": 66280
    },
    {
      "epoch": 0.267819988122028,
      "grad_norm": 1437.4583740234375,
      "learning_rate": 3.6767325066761136e-05,
      "loss": 98.3323,
      "step": 66290
    },
    {
      "epoch": 0.2678603893873956,
      "grad_norm": 696.0053100585938,
      "learning_rate": 3.676580267643103e-05,
      "loss": 123.7425,
      "step": 66300
    },
    {
      "epoch": 0.26790079065276323,
      "grad_norm": 660.8330688476562,
      "learning_rate": 3.676427995924297e-05,
      "loss": 44.3089,
      "step": 66310
    },
    {
      "epoch": 0.26794119191813087,
      "grad_norm": 0.0,
      "learning_rate": 3.6762756915226654e-05,
      "loss": 60.4803,
      "step": 66320
    },
    {
      "epoch": 0.2679815931834985,
      "grad_norm": 326.1693115234375,
      "learning_rate": 3.676123354441175e-05,
      "loss": 61.5102,
      "step": 66330
    },
    {
      "epoch": 0.26802199444886615,
      "grad_norm": 409.2984924316406,
      "learning_rate": 3.6759709846827977e-05,
      "loss": 91.4111,
      "step": 66340
    },
    {
      "epoch": 0.2680623957142338,
      "grad_norm": 866.605224609375,
      "learning_rate": 3.675818582250503e-05,
      "loss": 72.9242,
      "step": 66350
    },
    {
      "epoch": 0.2681027969796014,
      "grad_norm": 1267.880615234375,
      "learning_rate": 3.6756661471472626e-05,
      "loss": 95.2002,
      "step": 66360
    },
    {
      "epoch": 0.268143198244969,
      "grad_norm": 591.977783203125,
      "learning_rate": 3.675513679376047e-05,
      "loss": 60.2687,
      "step": 66370
    },
    {
      "epoch": 0.26818359951033666,
      "grad_norm": 1351.8251953125,
      "learning_rate": 3.67536117893983e-05,
      "loss": 62.6659,
      "step": 66380
    },
    {
      "epoch": 0.2682240007757043,
      "grad_norm": 874.5169677734375,
      "learning_rate": 3.675208645841584e-05,
      "loss": 66.4748,
      "step": 66390
    },
    {
      "epoch": 0.26826440204107194,
      "grad_norm": 4129.86474609375,
      "learning_rate": 3.675056080084284e-05,
      "loss": 109.5643,
      "step": 66400
    },
    {
      "epoch": 0.2683048033064396,
      "grad_norm": 681.3414916992188,
      "learning_rate": 3.674903481670902e-05,
      "loss": 76.4744,
      "step": 66410
    },
    {
      "epoch": 0.2683452045718072,
      "grad_norm": 989.4423828125,
      "learning_rate": 3.6747508506044156e-05,
      "loss": 56.9486,
      "step": 66420
    },
    {
      "epoch": 0.2683856058371748,
      "grad_norm": 644.98681640625,
      "learning_rate": 3.6745981868877986e-05,
      "loss": 107.4632,
      "step": 66430
    },
    {
      "epoch": 0.26842600710254244,
      "grad_norm": 709.885009765625,
      "learning_rate": 3.674445490524027e-05,
      "loss": 83.2144,
      "step": 66440
    },
    {
      "epoch": 0.2684664083679101,
      "grad_norm": 268.4350280761719,
      "learning_rate": 3.67429276151608e-05,
      "loss": 44.282,
      "step": 66450
    },
    {
      "epoch": 0.2685068096332777,
      "grad_norm": 972.099365234375,
      "learning_rate": 3.674139999866933e-05,
      "loss": 68.6298,
      "step": 66460
    },
    {
      "epoch": 0.26854721089864536,
      "grad_norm": 1137.8095703125,
      "learning_rate": 3.6739872055795646e-05,
      "loss": 49.5315,
      "step": 66470
    },
    {
      "epoch": 0.268587612164013,
      "grad_norm": 1797.35595703125,
      "learning_rate": 3.673834378656954e-05,
      "loss": 54.9943,
      "step": 66480
    },
    {
      "epoch": 0.2686280134293806,
      "grad_norm": 668.4421997070312,
      "learning_rate": 3.67368151910208e-05,
      "loss": 92.2195,
      "step": 66490
    },
    {
      "epoch": 0.2686684146947482,
      "grad_norm": 848.138671875,
      "learning_rate": 3.673528626917924e-05,
      "loss": 105.8378,
      "step": 66500
    },
    {
      "epoch": 0.26870881596011587,
      "grad_norm": 599.9146728515625,
      "learning_rate": 3.673375702107465e-05,
      "loss": 71.8781,
      "step": 66510
    },
    {
      "epoch": 0.2687492172254835,
      "grad_norm": 665.2034301757812,
      "learning_rate": 3.673222744673686e-05,
      "loss": 66.1195,
      "step": 66520
    },
    {
      "epoch": 0.26878961849085115,
      "grad_norm": 1226.78515625,
      "learning_rate": 3.673069754619567e-05,
      "loss": 81.691,
      "step": 66530
    },
    {
      "epoch": 0.2688300197562188,
      "grad_norm": 676.3750610351562,
      "learning_rate": 3.672916731948092e-05,
      "loss": 66.9422,
      "step": 66540
    },
    {
      "epoch": 0.26887042102158637,
      "grad_norm": 990.93994140625,
      "learning_rate": 3.672763676662245e-05,
      "loss": 80.2395,
      "step": 66550
    },
    {
      "epoch": 0.268910822286954,
      "grad_norm": 2785.474853515625,
      "learning_rate": 3.672610588765008e-05,
      "loss": 85.2976,
      "step": 66560
    },
    {
      "epoch": 0.26895122355232165,
      "grad_norm": 777.316162109375,
      "learning_rate": 3.672457468259367e-05,
      "loss": 69.1201,
      "step": 66570
    },
    {
      "epoch": 0.2689916248176893,
      "grad_norm": 911.9228515625,
      "learning_rate": 3.6723043151483066e-05,
      "loss": 73.0671,
      "step": 66580
    },
    {
      "epoch": 0.26903202608305693,
      "grad_norm": 1047.5543212890625,
      "learning_rate": 3.6721511294348124e-05,
      "loss": 96.2293,
      "step": 66590
    },
    {
      "epoch": 0.26907242734842457,
      "grad_norm": 0.0,
      "learning_rate": 3.671997911121871e-05,
      "loss": 67.165,
      "step": 66600
    },
    {
      "epoch": 0.2691128286137922,
      "grad_norm": 1073.5423583984375,
      "learning_rate": 3.67184466021247e-05,
      "loss": 83.9561,
      "step": 66610
    },
    {
      "epoch": 0.2691532298791598,
      "grad_norm": 599.5601806640625,
      "learning_rate": 3.6716913767095964e-05,
      "loss": 60.168,
      "step": 66620
    },
    {
      "epoch": 0.26919363114452743,
      "grad_norm": 626.68701171875,
      "learning_rate": 3.671538060616239e-05,
      "loss": 50.7169,
      "step": 66630
    },
    {
      "epoch": 0.2692340324098951,
      "grad_norm": 2009.712158203125,
      "learning_rate": 3.671384711935386e-05,
      "loss": 71.0345,
      "step": 66640
    },
    {
      "epoch": 0.2692744336752627,
      "grad_norm": 805.1333618164062,
      "learning_rate": 3.6712313306700276e-05,
      "loss": 102.1495,
      "step": 66650
    },
    {
      "epoch": 0.26931483494063035,
      "grad_norm": 950.0916137695312,
      "learning_rate": 3.6710779168231535e-05,
      "loss": 79.8175,
      "step": 66660
    },
    {
      "epoch": 0.269355236205998,
      "grad_norm": 436.6391906738281,
      "learning_rate": 3.670924470397756e-05,
      "loss": 56.7998,
      "step": 66670
    },
    {
      "epoch": 0.2693956374713656,
      "grad_norm": 601.5894165039062,
      "learning_rate": 3.6707709913968254e-05,
      "loss": 69.9057,
      "step": 66680
    },
    {
      "epoch": 0.2694360387367332,
      "grad_norm": 910.7144165039062,
      "learning_rate": 3.6706174798233536e-05,
      "loss": 62.8246,
      "step": 66690
    },
    {
      "epoch": 0.26947644000210086,
      "grad_norm": 683.0995483398438,
      "learning_rate": 3.670463935680335e-05,
      "loss": 82.9362,
      "step": 66700
    },
    {
      "epoch": 0.2695168412674685,
      "grad_norm": 510.84930419921875,
      "learning_rate": 3.670310358970762e-05,
      "loss": 50.4376,
      "step": 66710
    },
    {
      "epoch": 0.26955724253283614,
      "grad_norm": 555.572998046875,
      "learning_rate": 3.670156749697627e-05,
      "loss": 65.4187,
      "step": 66720
    },
    {
      "epoch": 0.2695976437982038,
      "grad_norm": 564.907470703125,
      "learning_rate": 3.670003107863928e-05,
      "loss": 84.8135,
      "step": 66730
    },
    {
      "epoch": 0.2696380450635714,
      "grad_norm": 1052.8568115234375,
      "learning_rate": 3.6698494334726575e-05,
      "loss": 101.7041,
      "step": 66740
    },
    {
      "epoch": 0.269678446328939,
      "grad_norm": 246.42845153808594,
      "learning_rate": 3.6696957265268134e-05,
      "loss": 48.1932,
      "step": 66750
    },
    {
      "epoch": 0.26971884759430664,
      "grad_norm": 1020.048583984375,
      "learning_rate": 3.6695419870293915e-05,
      "loss": 100.3386,
      "step": 66760
    },
    {
      "epoch": 0.2697592488596743,
      "grad_norm": 1148.8536376953125,
      "learning_rate": 3.669388214983388e-05,
      "loss": 59.7867,
      "step": 66770
    },
    {
      "epoch": 0.2697996501250419,
      "grad_norm": 382.64501953125,
      "learning_rate": 3.669234410391803e-05,
      "loss": 44.2975,
      "step": 66780
    },
    {
      "epoch": 0.26984005139040956,
      "grad_norm": 1493.79296875,
      "learning_rate": 3.669080573257633e-05,
      "loss": 62.6379,
      "step": 66790
    },
    {
      "epoch": 0.2698804526557772,
      "grad_norm": 369.2339172363281,
      "learning_rate": 3.668926703583878e-05,
      "loss": 54.9538,
      "step": 66800
    },
    {
      "epoch": 0.2699208539211448,
      "grad_norm": 716.9329223632812,
      "learning_rate": 3.668772801373538e-05,
      "loss": 82.7792,
      "step": 66810
    },
    {
      "epoch": 0.2699612551865124,
      "grad_norm": 882.6861572265625,
      "learning_rate": 3.6686188666296135e-05,
      "loss": 75.0362,
      "step": 66820
    },
    {
      "epoch": 0.27000165645188007,
      "grad_norm": 359.2427978515625,
      "learning_rate": 3.668464899355105e-05,
      "loss": 88.4607,
      "step": 66830
    },
    {
      "epoch": 0.2700420577172477,
      "grad_norm": 941.4803466796875,
      "learning_rate": 3.668310899553014e-05,
      "loss": 63.2962,
      "step": 66840
    },
    {
      "epoch": 0.27008245898261535,
      "grad_norm": 839.5689697265625,
      "learning_rate": 3.668156867226343e-05,
      "loss": 51.2695,
      "step": 66850
    },
    {
      "epoch": 0.270122860247983,
      "grad_norm": 950.5919189453125,
      "learning_rate": 3.668002802378094e-05,
      "loss": 79.6808,
      "step": 66860
    },
    {
      "epoch": 0.27016326151335057,
      "grad_norm": 794.5095825195312,
      "learning_rate": 3.6678487050112735e-05,
      "loss": 60.239,
      "step": 66870
    },
    {
      "epoch": 0.2702036627787182,
      "grad_norm": 300.521240234375,
      "learning_rate": 3.667694575128883e-05,
      "loss": 78.1244,
      "step": 66880
    },
    {
      "epoch": 0.27024406404408585,
      "grad_norm": 1067.5,
      "learning_rate": 3.667540412733928e-05,
      "loss": 56.125,
      "step": 66890
    },
    {
      "epoch": 0.2702844653094535,
      "grad_norm": 496.9468688964844,
      "learning_rate": 3.667386217829415e-05,
      "loss": 46.4149,
      "step": 66900
    },
    {
      "epoch": 0.27032486657482113,
      "grad_norm": 670.4450073242188,
      "learning_rate": 3.667231990418348e-05,
      "loss": 70.8287,
      "step": 66910
    },
    {
      "epoch": 0.27036526784018877,
      "grad_norm": 787.8795776367188,
      "learning_rate": 3.667077730503736e-05,
      "loss": 57.6352,
      "step": 66920
    },
    {
      "epoch": 0.2704056691055564,
      "grad_norm": 542.7733764648438,
      "learning_rate": 3.666923438088585e-05,
      "loss": 55.8081,
      "step": 66930
    },
    {
      "epoch": 0.270446070370924,
      "grad_norm": 568.5580444335938,
      "learning_rate": 3.6667691131759034e-05,
      "loss": 87.649,
      "step": 66940
    },
    {
      "epoch": 0.27048647163629164,
      "grad_norm": 1132.9769287109375,
      "learning_rate": 3.6666147557687e-05,
      "loss": 102.21,
      "step": 66950
    },
    {
      "epoch": 0.2705268729016593,
      "grad_norm": 1040.704345703125,
      "learning_rate": 3.6664603658699836e-05,
      "loss": 74.1205,
      "step": 66960
    },
    {
      "epoch": 0.2705672741670269,
      "grad_norm": 593.2798461914062,
      "learning_rate": 3.666305943482765e-05,
      "loss": 64.604,
      "step": 66970
    },
    {
      "epoch": 0.27060767543239456,
      "grad_norm": 1937.970458984375,
      "learning_rate": 3.666151488610053e-05,
      "loss": 87.1729,
      "step": 66980
    },
    {
      "epoch": 0.2706480766977622,
      "grad_norm": 746.2176513671875,
      "learning_rate": 3.6659970012548606e-05,
      "loss": 72.0054,
      "step": 66990
    },
    {
      "epoch": 0.2706884779631298,
      "grad_norm": 0.0,
      "learning_rate": 3.665842481420199e-05,
      "loss": 66.9999,
      "step": 67000
    },
    {
      "epoch": 0.2707288792284974,
      "grad_norm": 725.7628784179688,
      "learning_rate": 3.665687929109081e-05,
      "loss": 87.2939,
      "step": 67010
    },
    {
      "epoch": 0.27076928049386506,
      "grad_norm": 1472.5765380859375,
      "learning_rate": 3.6655333443245184e-05,
      "loss": 61.8437,
      "step": 67020
    },
    {
      "epoch": 0.2708096817592327,
      "grad_norm": 2046.2059326171875,
      "learning_rate": 3.6653787270695266e-05,
      "loss": 67.9492,
      "step": 67030
    },
    {
      "epoch": 0.27085008302460034,
      "grad_norm": 1022.05517578125,
      "learning_rate": 3.665224077347119e-05,
      "loss": 65.6357,
      "step": 67040
    },
    {
      "epoch": 0.270890484289968,
      "grad_norm": 1938.1826171875,
      "learning_rate": 3.66506939516031e-05,
      "loss": 51.326,
      "step": 67050
    },
    {
      "epoch": 0.27093088555533557,
      "grad_norm": 1422.065185546875,
      "learning_rate": 3.664914680512117e-05,
      "loss": 56.8821,
      "step": 67060
    },
    {
      "epoch": 0.2709712868207032,
      "grad_norm": 644.83349609375,
      "learning_rate": 3.664759933405554e-05,
      "loss": 59.517,
      "step": 67070
    },
    {
      "epoch": 0.27101168808607085,
      "grad_norm": 537.67578125,
      "learning_rate": 3.6646051538436395e-05,
      "loss": 43.3435,
      "step": 67080
    },
    {
      "epoch": 0.2710520893514385,
      "grad_norm": 485.49444580078125,
      "learning_rate": 3.6644503418293905e-05,
      "loss": 80.7018,
      "step": 67090
    },
    {
      "epoch": 0.2710924906168061,
      "grad_norm": 975.3925170898438,
      "learning_rate": 3.6642954973658256e-05,
      "loss": 99.1135,
      "step": 67100
    },
    {
      "epoch": 0.27113289188217377,
      "grad_norm": 980.0559692382812,
      "learning_rate": 3.664140620455962e-05,
      "loss": 77.6215,
      "step": 67110
    },
    {
      "epoch": 0.2711732931475414,
      "grad_norm": 532.5707397460938,
      "learning_rate": 3.6639857111028215e-05,
      "loss": 60.3496,
      "step": 67120
    },
    {
      "epoch": 0.271213694412909,
      "grad_norm": 524.780517578125,
      "learning_rate": 3.663830769309423e-05,
      "loss": 88.2819,
      "step": 67130
    },
    {
      "epoch": 0.27125409567827663,
      "grad_norm": 691.8131713867188,
      "learning_rate": 3.663675795078786e-05,
      "loss": 103.3954,
      "step": 67140
    },
    {
      "epoch": 0.27129449694364427,
      "grad_norm": 1298.9471435546875,
      "learning_rate": 3.663520788413933e-05,
      "loss": 91.5792,
      "step": 67150
    },
    {
      "epoch": 0.2713348982090119,
      "grad_norm": 783.9149169921875,
      "learning_rate": 3.663365749317886e-05,
      "loss": 47.3599,
      "step": 67160
    },
    {
      "epoch": 0.27137529947437955,
      "grad_norm": 2600.474853515625,
      "learning_rate": 3.663210677793668e-05,
      "loss": 72.9183,
      "step": 67170
    },
    {
      "epoch": 0.2714157007397472,
      "grad_norm": 728.5089111328125,
      "learning_rate": 3.6630555738443e-05,
      "loss": 58.8554,
      "step": 67180
    },
    {
      "epoch": 0.2714561020051148,
      "grad_norm": 1014.21435546875,
      "learning_rate": 3.662900437472809e-05,
      "loss": 61.5755,
      "step": 67190
    },
    {
      "epoch": 0.2714965032704824,
      "grad_norm": 557.588623046875,
      "learning_rate": 3.662745268682217e-05,
      "loss": 71.5919,
      "step": 67200
    },
    {
      "epoch": 0.27153690453585005,
      "grad_norm": 539.9105834960938,
      "learning_rate": 3.6625900674755503e-05,
      "loss": 70.7316,
      "step": 67210
    },
    {
      "epoch": 0.2715773058012177,
      "grad_norm": 980.9306030273438,
      "learning_rate": 3.662434833855835e-05,
      "loss": 62.9494,
      "step": 67220
    },
    {
      "epoch": 0.27161770706658533,
      "grad_norm": 646.4301147460938,
      "learning_rate": 3.662279567826096e-05,
      "loss": 56.7494,
      "step": 67230
    },
    {
      "epoch": 0.271658108331953,
      "grad_norm": 573.99072265625,
      "learning_rate": 3.66212426938936e-05,
      "loss": 84.0111,
      "step": 67240
    },
    {
      "epoch": 0.2716985095973206,
      "grad_norm": 1240.050048828125,
      "learning_rate": 3.6619689385486566e-05,
      "loss": 74.4962,
      "step": 67250
    },
    {
      "epoch": 0.2717389108626882,
      "grad_norm": 776.9219970703125,
      "learning_rate": 3.6618135753070136e-05,
      "loss": 65.1254,
      "step": 67260
    },
    {
      "epoch": 0.27177931212805584,
      "grad_norm": 828.0167846679688,
      "learning_rate": 3.661658179667459e-05,
      "loss": 74.3977,
      "step": 67270
    },
    {
      "epoch": 0.2718197133934235,
      "grad_norm": 785.6934814453125,
      "learning_rate": 3.6615027516330226e-05,
      "loss": 87.0982,
      "step": 67280
    },
    {
      "epoch": 0.2718601146587911,
      "grad_norm": 464.1484680175781,
      "learning_rate": 3.6613472912067345e-05,
      "loss": 45.3541,
      "step": 67290
    },
    {
      "epoch": 0.27190051592415876,
      "grad_norm": 463.8283386230469,
      "learning_rate": 3.661191798391626e-05,
      "loss": 77.3246,
      "step": 67300
    },
    {
      "epoch": 0.2719409171895264,
      "grad_norm": 1396.08984375,
      "learning_rate": 3.661036273190727e-05,
      "loss": 70.516,
      "step": 67310
    },
    {
      "epoch": 0.271981318454894,
      "grad_norm": 665.4253540039062,
      "learning_rate": 3.660880715607072e-05,
      "loss": 83.0413,
      "step": 67320
    },
    {
      "epoch": 0.2720217197202616,
      "grad_norm": 501.6424560546875,
      "learning_rate": 3.6607251256436925e-05,
      "loss": 68.0622,
      "step": 67330
    },
    {
      "epoch": 0.27206212098562926,
      "grad_norm": 608.7156982421875,
      "learning_rate": 3.660569503303621e-05,
      "loss": 56.7109,
      "step": 67340
    },
    {
      "epoch": 0.2721025222509969,
      "grad_norm": 482.20233154296875,
      "learning_rate": 3.660413848589892e-05,
      "loss": 68.6953,
      "step": 67350
    },
    {
      "epoch": 0.27214292351636454,
      "grad_norm": 738.3167114257812,
      "learning_rate": 3.6602581615055406e-05,
      "loss": 65.3163,
      "step": 67360
    },
    {
      "epoch": 0.2721833247817322,
      "grad_norm": 357.5082092285156,
      "learning_rate": 3.660102442053601e-05,
      "loss": 73.2594,
      "step": 67370
    },
    {
      "epoch": 0.27222372604709977,
      "grad_norm": 775.5992431640625,
      "learning_rate": 3.659946690237111e-05,
      "loss": 74.1652,
      "step": 67380
    },
    {
      "epoch": 0.2722641273124674,
      "grad_norm": 926.7460327148438,
      "learning_rate": 3.659790906059105e-05,
      "loss": 67.9021,
      "step": 67390
    },
    {
      "epoch": 0.27230452857783505,
      "grad_norm": 614.5941162109375,
      "learning_rate": 3.65963508952262e-05,
      "loss": 89.038,
      "step": 67400
    },
    {
      "epoch": 0.2723449298432027,
      "grad_norm": 503.3794250488281,
      "learning_rate": 3.659479240630695e-05,
      "loss": 123.8151,
      "step": 67410
    },
    {
      "epoch": 0.2723853311085703,
      "grad_norm": 680.7290649414062,
      "learning_rate": 3.659323359386368e-05,
      "loss": 68.3755,
      "step": 67420
    },
    {
      "epoch": 0.27242573237393797,
      "grad_norm": 876.765869140625,
      "learning_rate": 3.659167445792677e-05,
      "loss": 69.3434,
      "step": 67430
    },
    {
      "epoch": 0.2724661336393056,
      "grad_norm": 466.19525146484375,
      "learning_rate": 3.659011499852664e-05,
      "loss": 70.4111,
      "step": 67440
    },
    {
      "epoch": 0.2725065349046732,
      "grad_norm": 1058.6571044921875,
      "learning_rate": 3.658855521569367e-05,
      "loss": 60.2211,
      "step": 67450
    },
    {
      "epoch": 0.27254693617004083,
      "grad_norm": 516.0714111328125,
      "learning_rate": 3.6586995109458266e-05,
      "loss": 70.7459,
      "step": 67460
    },
    {
      "epoch": 0.27258733743540847,
      "grad_norm": 452.1062927246094,
      "learning_rate": 3.6585434679850866e-05,
      "loss": 44.2226,
      "step": 67470
    },
    {
      "epoch": 0.2726277387007761,
      "grad_norm": 719.1903076171875,
      "learning_rate": 3.6583873926901866e-05,
      "loss": 64.644,
      "step": 67480
    },
    {
      "epoch": 0.27266813996614375,
      "grad_norm": 807.54052734375,
      "learning_rate": 3.658231285064172e-05,
      "loss": 65.3155,
      "step": 67490
    },
    {
      "epoch": 0.2727085412315114,
      "grad_norm": 656.8856201171875,
      "learning_rate": 3.658075145110083e-05,
      "loss": 58.9538,
      "step": 67500
    },
    {
      "epoch": 0.272748942496879,
      "grad_norm": 619.9434814453125,
      "learning_rate": 3.657918972830967e-05,
      "loss": 75.0069,
      "step": 67510
    },
    {
      "epoch": 0.2727893437622466,
      "grad_norm": 970.5076293945312,
      "learning_rate": 3.657762768229867e-05,
      "loss": 48.6334,
      "step": 67520
    },
    {
      "epoch": 0.27282974502761426,
      "grad_norm": 243.0670928955078,
      "learning_rate": 3.6576065313098276e-05,
      "loss": 77.6509,
      "step": 67530
    },
    {
      "epoch": 0.2728701462929819,
      "grad_norm": 420.6954345703125,
      "learning_rate": 3.657450262073896e-05,
      "loss": 51.6488,
      "step": 67540
    },
    {
      "epoch": 0.27291054755834954,
      "grad_norm": 1002.0394287109375,
      "learning_rate": 3.657293960525118e-05,
      "loss": 68.1972,
      "step": 67550
    },
    {
      "epoch": 0.2729509488237172,
      "grad_norm": 814.0970458984375,
      "learning_rate": 3.657137626666541e-05,
      "loss": 48.7616,
      "step": 67560
    },
    {
      "epoch": 0.2729913500890848,
      "grad_norm": 674.52978515625,
      "learning_rate": 3.656981260501213e-05,
      "loss": 67.1653,
      "step": 67570
    },
    {
      "epoch": 0.2730317513544524,
      "grad_norm": 740.9395751953125,
      "learning_rate": 3.656824862032182e-05,
      "loss": 67.1278,
      "step": 67580
    },
    {
      "epoch": 0.27307215261982004,
      "grad_norm": 467.32293701171875,
      "learning_rate": 3.6566684312624966e-05,
      "loss": 66.5664,
      "step": 67590
    },
    {
      "epoch": 0.2731125538851877,
      "grad_norm": 1745.7122802734375,
      "learning_rate": 3.6565119681952086e-05,
      "loss": 79.5672,
      "step": 67600
    },
    {
      "epoch": 0.2731529551505553,
      "grad_norm": 1980.6866455078125,
      "learning_rate": 3.6563554728333664e-05,
      "loss": 61.6905,
      "step": 67610
    },
    {
      "epoch": 0.27319335641592296,
      "grad_norm": 650.4952392578125,
      "learning_rate": 3.6561989451800215e-05,
      "loss": 66.1603,
      "step": 67620
    },
    {
      "epoch": 0.2732337576812906,
      "grad_norm": 501.155517578125,
      "learning_rate": 3.656042385238225e-05,
      "loss": 60.0304,
      "step": 67630
    },
    {
      "epoch": 0.2732741589466582,
      "grad_norm": 437.7528991699219,
      "learning_rate": 3.655885793011031e-05,
      "loss": 118.2731,
      "step": 67640
    },
    {
      "epoch": 0.2733145602120258,
      "grad_norm": 601.8849487304688,
      "learning_rate": 3.6557291685014896e-05,
      "loss": 73.234,
      "step": 67650
    },
    {
      "epoch": 0.27335496147739347,
      "grad_norm": 1878.159912109375,
      "learning_rate": 3.655572511712656e-05,
      "loss": 96.2345,
      "step": 67660
    },
    {
      "epoch": 0.2733953627427611,
      "grad_norm": 1368.6416015625,
      "learning_rate": 3.655415822647584e-05,
      "loss": 87.2413,
      "step": 67670
    },
    {
      "epoch": 0.27343576400812875,
      "grad_norm": 760.669677734375,
      "learning_rate": 3.655259101309328e-05,
      "loss": 70.731,
      "step": 67680
    },
    {
      "epoch": 0.2734761652734964,
      "grad_norm": 915.2348022460938,
      "learning_rate": 3.655102347700944e-05,
      "loss": 87.2867,
      "step": 67690
    },
    {
      "epoch": 0.27351656653886397,
      "grad_norm": 710.090087890625,
      "learning_rate": 3.654945561825488e-05,
      "loss": 74.8702,
      "step": 67700
    },
    {
      "epoch": 0.2735569678042316,
      "grad_norm": 1369.951171875,
      "learning_rate": 3.6547887436860164e-05,
      "loss": 81.5384,
      "step": 67710
    },
    {
      "epoch": 0.27359736906959925,
      "grad_norm": 1007.9266967773438,
      "learning_rate": 3.654631893285585e-05,
      "loss": 56.8456,
      "step": 67720
    },
    {
      "epoch": 0.2736377703349669,
      "grad_norm": 447.4447326660156,
      "learning_rate": 3.6544750106272534e-05,
      "loss": 42.0591,
      "step": 67730
    },
    {
      "epoch": 0.27367817160033453,
      "grad_norm": 568.894287109375,
      "learning_rate": 3.6543180957140804e-05,
      "loss": 80.5745,
      "step": 67740
    },
    {
      "epoch": 0.27371857286570217,
      "grad_norm": 843.6287231445312,
      "learning_rate": 3.654161148549124e-05,
      "loss": 90.9321,
      "step": 67750
    },
    {
      "epoch": 0.2737589741310698,
      "grad_norm": 1305.992919921875,
      "learning_rate": 3.654004169135444e-05,
      "loss": 71.9361,
      "step": 67760
    },
    {
      "epoch": 0.2737993753964374,
      "grad_norm": 798.0366821289062,
      "learning_rate": 3.653847157476101e-05,
      "loss": 49.0851,
      "step": 67770
    },
    {
      "epoch": 0.27383977666180503,
      "grad_norm": 863.2374877929688,
      "learning_rate": 3.653690113574158e-05,
      "loss": 84.5428,
      "step": 67780
    },
    {
      "epoch": 0.2738801779271727,
      "grad_norm": 584.3214111328125,
      "learning_rate": 3.6535330374326734e-05,
      "loss": 59.1151,
      "step": 67790
    },
    {
      "epoch": 0.2739205791925403,
      "grad_norm": 805.32568359375,
      "learning_rate": 3.6533759290547114e-05,
      "loss": 90.4222,
      "step": 67800
    },
    {
      "epoch": 0.27396098045790795,
      "grad_norm": 531.2085571289062,
      "learning_rate": 3.653218788443334e-05,
      "loss": 41.2134,
      "step": 67810
    },
    {
      "epoch": 0.2740013817232756,
      "grad_norm": 496.3345642089844,
      "learning_rate": 3.653061615601606e-05,
      "loss": 52.7027,
      "step": 67820
    },
    {
      "epoch": 0.2740417829886432,
      "grad_norm": 987.3363647460938,
      "learning_rate": 3.652904410532589e-05,
      "loss": 62.067,
      "step": 67830
    },
    {
      "epoch": 0.2740821842540108,
      "grad_norm": 778.4385986328125,
      "learning_rate": 3.6527471732393515e-05,
      "loss": 69.8173,
      "step": 67840
    },
    {
      "epoch": 0.27412258551937846,
      "grad_norm": 565.9769897460938,
      "learning_rate": 3.652589903724956e-05,
      "loss": 66.5563,
      "step": 67850
    },
    {
      "epoch": 0.2741629867847461,
      "grad_norm": 1377.9952392578125,
      "learning_rate": 3.65243260199247e-05,
      "loss": 58.3849,
      "step": 67860
    },
    {
      "epoch": 0.27420338805011374,
      "grad_norm": 641.1226196289062,
      "learning_rate": 3.652275268044959e-05,
      "loss": 106.0223,
      "step": 67870
    },
    {
      "epoch": 0.2742437893154814,
      "grad_norm": 1087.67236328125,
      "learning_rate": 3.6521179018854914e-05,
      "loss": 64.6192,
      "step": 67880
    },
    {
      "epoch": 0.274284190580849,
      "grad_norm": 1090.8638916015625,
      "learning_rate": 3.651960503517135e-05,
      "loss": 99.0607,
      "step": 67890
    },
    {
      "epoch": 0.2743245918462166,
      "grad_norm": 299.8224792480469,
      "learning_rate": 3.651803072942957e-05,
      "loss": 45.6526,
      "step": 67900
    },
    {
      "epoch": 0.27436499311158424,
      "grad_norm": 1958.9107666015625,
      "learning_rate": 3.6516456101660286e-05,
      "loss": 119.4297,
      "step": 67910
    },
    {
      "epoch": 0.2744053943769519,
      "grad_norm": 1904.2401123046875,
      "learning_rate": 3.651488115189419e-05,
      "loss": 86.1005,
      "step": 67920
    },
    {
      "epoch": 0.2744457956423195,
      "grad_norm": 1208.603271484375,
      "learning_rate": 3.651330588016198e-05,
      "loss": 75.8416,
      "step": 67930
    },
    {
      "epoch": 0.27448619690768716,
      "grad_norm": 868.4407958984375,
      "learning_rate": 3.6511730286494366e-05,
      "loss": 51.1043,
      "step": 67940
    },
    {
      "epoch": 0.2745265981730548,
      "grad_norm": 997.5648193359375,
      "learning_rate": 3.651015437092208e-05,
      "loss": 54.7892,
      "step": 67950
    },
    {
      "epoch": 0.2745669994384224,
      "grad_norm": 651.9278564453125,
      "learning_rate": 3.650857813347582e-05,
      "loss": 70.1957,
      "step": 67960
    },
    {
      "epoch": 0.27460740070379,
      "grad_norm": 788.9302368164062,
      "learning_rate": 3.6507001574186346e-05,
      "loss": 110.4476,
      "step": 67970
    },
    {
      "epoch": 0.27464780196915767,
      "grad_norm": 737.9695434570312,
      "learning_rate": 3.650542469308437e-05,
      "loss": 75.0885,
      "step": 67980
    },
    {
      "epoch": 0.2746882032345253,
      "grad_norm": 0.0,
      "learning_rate": 3.6503847490200636e-05,
      "loss": 76.1971,
      "step": 67990
    },
    {
      "epoch": 0.27472860449989295,
      "grad_norm": 494.2500915527344,
      "learning_rate": 3.6502269965565904e-05,
      "loss": 51.3936,
      "step": 68000
    },
    {
      "epoch": 0.2747690057652606,
      "grad_norm": 775.4039916992188,
      "learning_rate": 3.650069211921093e-05,
      "loss": 51.5934,
      "step": 68010
    },
    {
      "epoch": 0.27480940703062817,
      "grad_norm": 541.7156982421875,
      "learning_rate": 3.649911395116646e-05,
      "loss": 51.5515,
      "step": 68020
    },
    {
      "epoch": 0.2748498082959958,
      "grad_norm": 745.6041870117188,
      "learning_rate": 3.649753546146327e-05,
      "loss": 48.3755,
      "step": 68030
    },
    {
      "epoch": 0.27489020956136345,
      "grad_norm": 969.8783569335938,
      "learning_rate": 3.6495956650132126e-05,
      "loss": 95.7821,
      "step": 68040
    },
    {
      "epoch": 0.2749306108267311,
      "grad_norm": 513.9328002929688,
      "learning_rate": 3.6494377517203825e-05,
      "loss": 77.0404,
      "step": 68050
    },
    {
      "epoch": 0.27497101209209873,
      "grad_norm": 396.3102111816406,
      "learning_rate": 3.649279806270914e-05,
      "loss": 63.6906,
      "step": 68060
    },
    {
      "epoch": 0.27501141335746637,
      "grad_norm": 519.8453979492188,
      "learning_rate": 3.6491218286678867e-05,
      "loss": 66.4816,
      "step": 68070
    },
    {
      "epoch": 0.275051814622834,
      "grad_norm": 1306.8505859375,
      "learning_rate": 3.648963818914379e-05,
      "loss": 55.9493,
      "step": 68080
    },
    {
      "epoch": 0.2750922158882016,
      "grad_norm": 610.3496704101562,
      "learning_rate": 3.6488057770134746e-05,
      "loss": 54.7979,
      "step": 68090
    },
    {
      "epoch": 0.27513261715356924,
      "grad_norm": 779.8862915039062,
      "learning_rate": 3.648647702968252e-05,
      "loss": 81.0502,
      "step": 68100
    },
    {
      "epoch": 0.2751730184189369,
      "grad_norm": 871.0150146484375,
      "learning_rate": 3.6484895967817935e-05,
      "loss": 66.1679,
      "step": 68110
    },
    {
      "epoch": 0.2752134196843045,
      "grad_norm": 635.9258422851562,
      "learning_rate": 3.6483314584571815e-05,
      "loss": 41.4136,
      "step": 68120
    },
    {
      "epoch": 0.27525382094967216,
      "grad_norm": 360.8929138183594,
      "learning_rate": 3.6481732879974995e-05,
      "loss": 73.421,
      "step": 68130
    },
    {
      "epoch": 0.2752942222150398,
      "grad_norm": 674.0567016601562,
      "learning_rate": 3.64801508540583e-05,
      "loss": 68.4341,
      "step": 68140
    },
    {
      "epoch": 0.2753346234804074,
      "grad_norm": 957.514892578125,
      "learning_rate": 3.647856850685259e-05,
      "loss": 80.5406,
      "step": 68150
    },
    {
      "epoch": 0.275375024745775,
      "grad_norm": 2177.1552734375,
      "learning_rate": 3.6476985838388694e-05,
      "loss": 76.5557,
      "step": 68160
    },
    {
      "epoch": 0.27541542601114266,
      "grad_norm": 749.5941162109375,
      "learning_rate": 3.647540284869748e-05,
      "loss": 72.2992,
      "step": 68170
    },
    {
      "epoch": 0.2754558272765103,
      "grad_norm": 736.7278442382812,
      "learning_rate": 3.647381953780981e-05,
      "loss": 94.8278,
      "step": 68180
    },
    {
      "epoch": 0.27549622854187794,
      "grad_norm": 895.4473876953125,
      "learning_rate": 3.647223590575654e-05,
      "loss": 63.1256,
      "step": 68190
    },
    {
      "epoch": 0.2755366298072456,
      "grad_norm": 361.90374755859375,
      "learning_rate": 3.647065195256855e-05,
      "loss": 82.1848,
      "step": 68200
    },
    {
      "epoch": 0.2755770310726132,
      "grad_norm": 746.0802001953125,
      "learning_rate": 3.6469067678276726e-05,
      "loss": 78.7891,
      "step": 68210
    },
    {
      "epoch": 0.2756174323379808,
      "grad_norm": 260.87823486328125,
      "learning_rate": 3.6467483082911945e-05,
      "loss": 53.1604,
      "step": 68220
    },
    {
      "epoch": 0.27565783360334845,
      "grad_norm": 544.7657470703125,
      "learning_rate": 3.646589816650511e-05,
      "loss": 79.9751,
      "step": 68230
    },
    {
      "epoch": 0.2756982348687161,
      "grad_norm": 709.5982666015625,
      "learning_rate": 3.64643129290871e-05,
      "loss": 64.3202,
      "step": 68240
    },
    {
      "epoch": 0.2757386361340837,
      "grad_norm": 650.8275756835938,
      "learning_rate": 3.6462727370688844e-05,
      "loss": 55.9401,
      "step": 68250
    },
    {
      "epoch": 0.27577903739945137,
      "grad_norm": 303.5444030761719,
      "learning_rate": 3.6461141491341235e-05,
      "loss": 100.8449,
      "step": 68260
    },
    {
      "epoch": 0.275819438664819,
      "grad_norm": 456.7483825683594,
      "learning_rate": 3.6459555291075204e-05,
      "loss": 78.0596,
      "step": 68270
    },
    {
      "epoch": 0.2758598399301866,
      "grad_norm": 457.44207763671875,
      "learning_rate": 3.6457968769921664e-05,
      "loss": 59.3378,
      "step": 68280
    },
    {
      "epoch": 0.27590024119555423,
      "grad_norm": 1555.9522705078125,
      "learning_rate": 3.645638192791155e-05,
      "loss": 79.8157,
      "step": 68290
    },
    {
      "epoch": 0.27594064246092187,
      "grad_norm": 822.7843627929688,
      "learning_rate": 3.64547947650758e-05,
      "loss": 71.4093,
      "step": 68300
    },
    {
      "epoch": 0.2759810437262895,
      "grad_norm": 948.1829833984375,
      "learning_rate": 3.6453207281445356e-05,
      "loss": 98.2951,
      "step": 68310
    },
    {
      "epoch": 0.27602144499165715,
      "grad_norm": 1063.466796875,
      "learning_rate": 3.6451619477051165e-05,
      "loss": 81.6967,
      "step": 68320
    },
    {
      "epoch": 0.2760618462570248,
      "grad_norm": 790.401123046875,
      "learning_rate": 3.645003135192418e-05,
      "loss": 107.049,
      "step": 68330
    },
    {
      "epoch": 0.2761022475223924,
      "grad_norm": 1264.2486572265625,
      "learning_rate": 3.6448442906095365e-05,
      "loss": 91.9635,
      "step": 68340
    },
    {
      "epoch": 0.27614264878776,
      "grad_norm": 358.43896484375,
      "learning_rate": 3.644685413959569e-05,
      "loss": 62.9964,
      "step": 68350
    },
    {
      "epoch": 0.27618305005312765,
      "grad_norm": 1053.30615234375,
      "learning_rate": 3.644526505245612e-05,
      "loss": 105.667,
      "step": 68360
    },
    {
      "epoch": 0.2762234513184953,
      "grad_norm": 741.2691040039062,
      "learning_rate": 3.6443675644707645e-05,
      "loss": 65.3842,
      "step": 68370
    },
    {
      "epoch": 0.27626385258386293,
      "grad_norm": 614.4652709960938,
      "learning_rate": 3.6442085916381256e-05,
      "loss": 63.0738,
      "step": 68380
    },
    {
      "epoch": 0.2763042538492306,
      "grad_norm": 545.9593505859375,
      "learning_rate": 3.644049586750792e-05,
      "loss": 76.6601,
      "step": 68390
    },
    {
      "epoch": 0.2763446551145982,
      "grad_norm": 351.7081298828125,
      "learning_rate": 3.6438905498118665e-05,
      "loss": 80.7119,
      "step": 68400
    },
    {
      "epoch": 0.2763850563799658,
      "grad_norm": 450.5742492675781,
      "learning_rate": 3.643731480824448e-05,
      "loss": 56.7873,
      "step": 68410
    },
    {
      "epoch": 0.27642545764533344,
      "grad_norm": 833.8705444335938,
      "learning_rate": 3.6435723797916384e-05,
      "loss": 52.5591,
      "step": 68420
    },
    {
      "epoch": 0.2764658589107011,
      "grad_norm": 304.5091857910156,
      "learning_rate": 3.643413246716539e-05,
      "loss": 35.8782,
      "step": 68430
    },
    {
      "epoch": 0.2765062601760687,
      "grad_norm": 386.6527404785156,
      "learning_rate": 3.643254081602252e-05,
      "loss": 66.0746,
      "step": 68440
    },
    {
      "epoch": 0.27654666144143636,
      "grad_norm": 1705.518798828125,
      "learning_rate": 3.6430948844518806e-05,
      "loss": 92.2069,
      "step": 68450
    },
    {
      "epoch": 0.276587062706804,
      "grad_norm": 807.954833984375,
      "learning_rate": 3.6429356552685285e-05,
      "loss": 53.035,
      "step": 68460
    },
    {
      "epoch": 0.2766274639721716,
      "grad_norm": 818.7935180664062,
      "learning_rate": 3.6427763940553004e-05,
      "loss": 57.0737,
      "step": 68470
    },
    {
      "epoch": 0.2766678652375392,
      "grad_norm": 510.9322204589844,
      "learning_rate": 3.6426171008153e-05,
      "loss": 71.6756,
      "step": 68480
    },
    {
      "epoch": 0.27670826650290686,
      "grad_norm": 898.836181640625,
      "learning_rate": 3.6424577755516344e-05,
      "loss": 73.5406,
      "step": 68490
    },
    {
      "epoch": 0.2767486677682745,
      "grad_norm": 512.6793823242188,
      "learning_rate": 3.642298418267408e-05,
      "loss": 81.9066,
      "step": 68500
    },
    {
      "epoch": 0.27678906903364214,
      "grad_norm": 597.5086669921875,
      "learning_rate": 3.642139028965729e-05,
      "loss": 43.1931,
      "step": 68510
    },
    {
      "epoch": 0.2768294702990098,
      "grad_norm": 1361.612548828125,
      "learning_rate": 3.6419796076497045e-05,
      "loss": 69.7019,
      "step": 68520
    },
    {
      "epoch": 0.2768698715643774,
      "grad_norm": 8374.60546875,
      "learning_rate": 3.6418201543224414e-05,
      "loss": 79.9838,
      "step": 68530
    },
    {
      "epoch": 0.276910272829745,
      "grad_norm": 1369.9517822265625,
      "learning_rate": 3.64166066898705e-05,
      "loss": 96.7081,
      "step": 68540
    },
    {
      "epoch": 0.27695067409511265,
      "grad_norm": 564.1013793945312,
      "learning_rate": 3.641501151646638e-05,
      "loss": 82.5782,
      "step": 68550
    },
    {
      "epoch": 0.2769910753604803,
      "grad_norm": 1302.3148193359375,
      "learning_rate": 3.641341602304316e-05,
      "loss": 66.468,
      "step": 68560
    },
    {
      "epoch": 0.27703147662584793,
      "grad_norm": 1007.4530639648438,
      "learning_rate": 3.6411820209631944e-05,
      "loss": 61.811,
      "step": 68570
    },
    {
      "epoch": 0.27707187789121557,
      "grad_norm": 2957.66455078125,
      "learning_rate": 3.641022407626385e-05,
      "loss": 69.4076,
      "step": 68580
    },
    {
      "epoch": 0.2771122791565832,
      "grad_norm": 1141.3900146484375,
      "learning_rate": 3.640862762296999e-05,
      "loss": 55.4128,
      "step": 68590
    },
    {
      "epoch": 0.2771526804219508,
      "grad_norm": 1451.926513671875,
      "learning_rate": 3.6407030849781475e-05,
      "loss": 75.5169,
      "step": 68600
    },
    {
      "epoch": 0.27719308168731843,
      "grad_norm": 824.8275146484375,
      "learning_rate": 3.6405433756729456e-05,
      "loss": 50.1131,
      "step": 68610
    },
    {
      "epoch": 0.27723348295268607,
      "grad_norm": 549.1823120117188,
      "learning_rate": 3.640383634384505e-05,
      "loss": 50.6561,
      "step": 68620
    },
    {
      "epoch": 0.2772738842180537,
      "grad_norm": 459.22216796875,
      "learning_rate": 3.6402238611159424e-05,
      "loss": 92.3009,
      "step": 68630
    },
    {
      "epoch": 0.27731428548342135,
      "grad_norm": 651.8291625976562,
      "learning_rate": 3.64006405587037e-05,
      "loss": 63.6321,
      "step": 68640
    },
    {
      "epoch": 0.277354686748789,
      "grad_norm": 1424.467041015625,
      "learning_rate": 3.639904218650905e-05,
      "loss": 72.4273,
      "step": 68650
    },
    {
      "epoch": 0.2773950880141566,
      "grad_norm": 768.332763671875,
      "learning_rate": 3.639744349460663e-05,
      "loss": 57.5873,
      "step": 68660
    },
    {
      "epoch": 0.2774354892795242,
      "grad_norm": 657.2222290039062,
      "learning_rate": 3.639584448302761e-05,
      "loss": 52.1681,
      "step": 68670
    },
    {
      "epoch": 0.27747589054489186,
      "grad_norm": 692.5430297851562,
      "learning_rate": 3.639424515180315e-05,
      "loss": 61.6016,
      "step": 68680
    },
    {
      "epoch": 0.2775162918102595,
      "grad_norm": 700.7813110351562,
      "learning_rate": 3.639264550096445e-05,
      "loss": 86.2401,
      "step": 68690
    },
    {
      "epoch": 0.27755669307562714,
      "grad_norm": 354.1077575683594,
      "learning_rate": 3.639104553054268e-05,
      "loss": 84.5295,
      "step": 68700
    },
    {
      "epoch": 0.2775970943409948,
      "grad_norm": 1245.1407470703125,
      "learning_rate": 3.6389445240569043e-05,
      "loss": 67.8237,
      "step": 68710
    },
    {
      "epoch": 0.2776374956063624,
      "grad_norm": 736.7760620117188,
      "learning_rate": 3.6387844631074726e-05,
      "loss": 61.6653,
      "step": 68720
    },
    {
      "epoch": 0.27767789687173,
      "grad_norm": 282.4181213378906,
      "learning_rate": 3.638624370209095e-05,
      "loss": 59.5905,
      "step": 68730
    },
    {
      "epoch": 0.27771829813709764,
      "grad_norm": 407.526123046875,
      "learning_rate": 3.638464245364891e-05,
      "loss": 70.27,
      "step": 68740
    },
    {
      "epoch": 0.2777586994024653,
      "grad_norm": 1361.0655517578125,
      "learning_rate": 3.638304088577984e-05,
      "loss": 68.5818,
      "step": 68750
    },
    {
      "epoch": 0.2777991006678329,
      "grad_norm": 749.9307250976562,
      "learning_rate": 3.6381438998514945e-05,
      "loss": 58.5423,
      "step": 68760
    },
    {
      "epoch": 0.27783950193320056,
      "grad_norm": 986.7581787109375,
      "learning_rate": 3.637983679188547e-05,
      "loss": 89.6241,
      "step": 68770
    },
    {
      "epoch": 0.2778799031985682,
      "grad_norm": 441.9813232421875,
      "learning_rate": 3.6378234265922636e-05,
      "loss": 52.9904,
      "step": 68780
    },
    {
      "epoch": 0.2779203044639358,
      "grad_norm": 871.7626342773438,
      "learning_rate": 3.6376631420657695e-05,
      "loss": 85.0174,
      "step": 68790
    },
    {
      "epoch": 0.2779607057293034,
      "grad_norm": 1010.308837890625,
      "learning_rate": 3.637502825612189e-05,
      "loss": 70.3216,
      "step": 68800
    },
    {
      "epoch": 0.27800110699467107,
      "grad_norm": 394.0765380859375,
      "learning_rate": 3.637342477234649e-05,
      "loss": 54.1732,
      "step": 68810
    },
    {
      "epoch": 0.2780415082600387,
      "grad_norm": 1378.3243408203125,
      "learning_rate": 3.637182096936274e-05,
      "loss": 78.4019,
      "step": 68820
    },
    {
      "epoch": 0.27808190952540635,
      "grad_norm": 807.8096313476562,
      "learning_rate": 3.63702168472019e-05,
      "loss": 70.5899,
      "step": 68830
    },
    {
      "epoch": 0.278122310790774,
      "grad_norm": 748.4146728515625,
      "learning_rate": 3.636861240589527e-05,
      "loss": 62.313,
      "step": 68840
    },
    {
      "epoch": 0.2781627120561416,
      "grad_norm": 1081.0704345703125,
      "learning_rate": 3.63670076454741e-05,
      "loss": 77.3886,
      "step": 68850
    },
    {
      "epoch": 0.2782031133215092,
      "grad_norm": 1138.8304443359375,
      "learning_rate": 3.63654025659697e-05,
      "loss": 89.0992,
      "step": 68860
    },
    {
      "epoch": 0.27824351458687685,
      "grad_norm": 1212.7421875,
      "learning_rate": 3.636379716741335e-05,
      "loss": 53.7292,
      "step": 68870
    },
    {
      "epoch": 0.2782839158522445,
      "grad_norm": 656.9420776367188,
      "learning_rate": 3.6362191449836346e-05,
      "loss": 68.1364,
      "step": 68880
    },
    {
      "epoch": 0.27832431711761213,
      "grad_norm": 702.285400390625,
      "learning_rate": 3.6360585413269995e-05,
      "loss": 54.8042,
      "step": 68890
    },
    {
      "epoch": 0.27836471838297977,
      "grad_norm": 1127.06591796875,
      "learning_rate": 3.6358979057745615e-05,
      "loss": 59.6537,
      "step": 68900
    },
    {
      "epoch": 0.2784051196483474,
      "grad_norm": 1014.2935180664062,
      "learning_rate": 3.635737238329451e-05,
      "loss": 58.5084,
      "step": 68910
    },
    {
      "epoch": 0.278445520913715,
      "grad_norm": 724.390625,
      "learning_rate": 3.6355765389948015e-05,
      "loss": 85.7793,
      "step": 68920
    },
    {
      "epoch": 0.27848592217908263,
      "grad_norm": 1114.0927734375,
      "learning_rate": 3.635415807773745e-05,
      "loss": 84.4643,
      "step": 68930
    },
    {
      "epoch": 0.2785263234444503,
      "grad_norm": 676.5823364257812,
      "learning_rate": 3.635255044669415e-05,
      "loss": 70.0983,
      "step": 68940
    },
    {
      "epoch": 0.2785667247098179,
      "grad_norm": 1191.34619140625,
      "learning_rate": 3.635094249684946e-05,
      "loss": 57.6529,
      "step": 68950
    },
    {
      "epoch": 0.27860712597518555,
      "grad_norm": 1170.3385009765625,
      "learning_rate": 3.6349334228234736e-05,
      "loss": 65.0031,
      "step": 68960
    },
    {
      "epoch": 0.2786475272405532,
      "grad_norm": 862.3743896484375,
      "learning_rate": 3.634772564088132e-05,
      "loss": 50.383,
      "step": 68970
    },
    {
      "epoch": 0.2786879285059208,
      "grad_norm": 211.3525848388672,
      "learning_rate": 3.6346116734820575e-05,
      "loss": 59.9885,
      "step": 68980
    },
    {
      "epoch": 0.2787283297712884,
      "grad_norm": 1631.1240234375,
      "learning_rate": 3.634450751008387e-05,
      "loss": 49.5971,
      "step": 68990
    },
    {
      "epoch": 0.27876873103665606,
      "grad_norm": 648.0172729492188,
      "learning_rate": 3.634289796670257e-05,
      "loss": 68.3455,
      "step": 69000
    },
    {
      "epoch": 0.2788091323020237,
      "grad_norm": 1328.7872314453125,
      "learning_rate": 3.634128810470807e-05,
      "loss": 78.5994,
      "step": 69010
    },
    {
      "epoch": 0.27884953356739134,
      "grad_norm": 1701.229248046875,
      "learning_rate": 3.633967792413174e-05,
      "loss": 78.8696,
      "step": 69020
    },
    {
      "epoch": 0.278889934832759,
      "grad_norm": 937.852783203125,
      "learning_rate": 3.6338067425004975e-05,
      "loss": 90.5087,
      "step": 69030
    },
    {
      "epoch": 0.2789303360981266,
      "grad_norm": 506.93182373046875,
      "learning_rate": 3.633645660735918e-05,
      "loss": 88.4482,
      "step": 69040
    },
    {
      "epoch": 0.2789707373634942,
      "grad_norm": 837.9589233398438,
      "learning_rate": 3.633484547122575e-05,
      "loss": 67.4671,
      "step": 69050
    },
    {
      "epoch": 0.27901113862886184,
      "grad_norm": 354.0655822753906,
      "learning_rate": 3.63332340166361e-05,
      "loss": 61.7967,
      "step": 69060
    },
    {
      "epoch": 0.2790515398942295,
      "grad_norm": 471.1148681640625,
      "learning_rate": 3.633162224362164e-05,
      "loss": 38.5641,
      "step": 69070
    },
    {
      "epoch": 0.2790919411595971,
      "grad_norm": 1305.4991455078125,
      "learning_rate": 3.63300101522138e-05,
      "loss": 109.4982,
      "step": 69080
    },
    {
      "epoch": 0.27913234242496476,
      "grad_norm": 1326.2255859375,
      "learning_rate": 3.6328397742444003e-05,
      "loss": 60.515,
      "step": 69090
    },
    {
      "epoch": 0.2791727436903324,
      "grad_norm": 674.3216552734375,
      "learning_rate": 3.632678501434368e-05,
      "loss": 77.833,
      "step": 69100
    },
    {
      "epoch": 0.2792131449557,
      "grad_norm": 798.8903198242188,
      "learning_rate": 3.632517196794429e-05,
      "loss": 67.0676,
      "step": 69110
    },
    {
      "epoch": 0.27925354622106763,
      "grad_norm": 662.2749633789062,
      "learning_rate": 3.632355860327725e-05,
      "loss": 70.8592,
      "step": 69120
    },
    {
      "epoch": 0.27929394748643527,
      "grad_norm": 559.8394165039062,
      "learning_rate": 3.632194492037404e-05,
      "loss": 55.1575,
      "step": 69130
    },
    {
      "epoch": 0.2793343487518029,
      "grad_norm": 406.7459716796875,
      "learning_rate": 3.632033091926612e-05,
      "loss": 67.0519,
      "step": 69140
    },
    {
      "epoch": 0.27937475001717055,
      "grad_norm": 1365.2215576171875,
      "learning_rate": 3.631871659998494e-05,
      "loss": 78.6769,
      "step": 69150
    },
    {
      "epoch": 0.2794151512825382,
      "grad_norm": 2076.132080078125,
      "learning_rate": 3.6317101962561974e-05,
      "loss": 114.1054,
      "step": 69160
    },
    {
      "epoch": 0.27945555254790583,
      "grad_norm": 582.463623046875,
      "learning_rate": 3.6315487007028706e-05,
      "loss": 79.3306,
      "step": 69170
    },
    {
      "epoch": 0.2794959538132734,
      "grad_norm": 803.3694458007812,
      "learning_rate": 3.631387173341662e-05,
      "loss": 88.376,
      "step": 69180
    },
    {
      "epoch": 0.27953635507864105,
      "grad_norm": 403.2090148925781,
      "learning_rate": 3.631225614175721e-05,
      "loss": 84.4055,
      "step": 69190
    },
    {
      "epoch": 0.2795767563440087,
      "grad_norm": 682.4560546875,
      "learning_rate": 3.6310640232081966e-05,
      "loss": 112.4025,
      "step": 69200
    },
    {
      "epoch": 0.27961715760937633,
      "grad_norm": 549.6954345703125,
      "learning_rate": 3.630902400442239e-05,
      "loss": 70.8671,
      "step": 69210
    },
    {
      "epoch": 0.27965755887474397,
      "grad_norm": 880.340087890625,
      "learning_rate": 3.630740745881e-05,
      "loss": 78.0043,
      "step": 69220
    },
    {
      "epoch": 0.2796979601401116,
      "grad_norm": 924.3558349609375,
      "learning_rate": 3.6305790595276303e-05,
      "loss": 59.7715,
      "step": 69230
    },
    {
      "epoch": 0.2797383614054792,
      "grad_norm": 0.0,
      "learning_rate": 3.630417341385282e-05,
      "loss": 47.4674,
      "step": 69240
    },
    {
      "epoch": 0.27977876267084684,
      "grad_norm": 1742.8243408203125,
      "learning_rate": 3.630255591457108e-05,
      "loss": 101.035,
      "step": 69250
    },
    {
      "epoch": 0.2798191639362145,
      "grad_norm": 2068.00048828125,
      "learning_rate": 3.6300938097462634e-05,
      "loss": 60.5279,
      "step": 69260
    },
    {
      "epoch": 0.2798595652015821,
      "grad_norm": 984.5831298828125,
      "learning_rate": 3.6299319962559e-05,
      "loss": 87.9492,
      "step": 69270
    },
    {
      "epoch": 0.27989996646694976,
      "grad_norm": 775.4241333007812,
      "learning_rate": 3.629770150989173e-05,
      "loss": 105.0423,
      "step": 69280
    },
    {
      "epoch": 0.2799403677323174,
      "grad_norm": 941.29248046875,
      "learning_rate": 3.629608273949238e-05,
      "loss": 77.3444,
      "step": 69290
    },
    {
      "epoch": 0.279980768997685,
      "grad_norm": 809.008544921875,
      "learning_rate": 3.62944636513925e-05,
      "loss": 54.6458,
      "step": 69300
    },
    {
      "epoch": 0.2800211702630526,
      "grad_norm": 848.7319946289062,
      "learning_rate": 3.629284424562367e-05,
      "loss": 82.872,
      "step": 69310
    },
    {
      "epoch": 0.28006157152842026,
      "grad_norm": 483.484130859375,
      "learning_rate": 3.629122452221746e-05,
      "loss": 79.1142,
      "step": 69320
    },
    {
      "epoch": 0.2801019727937879,
      "grad_norm": 295.2940979003906,
      "learning_rate": 3.6289604481205424e-05,
      "loss": 71.1907,
      "step": 69330
    },
    {
      "epoch": 0.28014237405915554,
      "grad_norm": 685.3955688476562,
      "learning_rate": 3.628798412261917e-05,
      "loss": 65.1816,
      "step": 69340
    },
    {
      "epoch": 0.2801827753245232,
      "grad_norm": 611.7601318359375,
      "learning_rate": 3.628636344649028e-05,
      "loss": 104.939,
      "step": 69350
    },
    {
      "epoch": 0.2802231765898908,
      "grad_norm": 1643.69677734375,
      "learning_rate": 3.628474245285035e-05,
      "loss": 66.5051,
      "step": 69360
    },
    {
      "epoch": 0.2802635778552584,
      "grad_norm": 894.060546875,
      "learning_rate": 3.628312114173098e-05,
      "loss": 57.6963,
      "step": 69370
    },
    {
      "epoch": 0.28030397912062605,
      "grad_norm": 570.6958618164062,
      "learning_rate": 3.628149951316378e-05,
      "loss": 56.4009,
      "step": 69380
    },
    {
      "epoch": 0.2803443803859937,
      "grad_norm": 662.741455078125,
      "learning_rate": 3.627987756718037e-05,
      "loss": 64.0986,
      "step": 69390
    },
    {
      "epoch": 0.2803847816513613,
      "grad_norm": 932.7979125976562,
      "learning_rate": 3.6278255303812366e-05,
      "loss": 83.0091,
      "step": 69400
    },
    {
      "epoch": 0.28042518291672897,
      "grad_norm": 541.6932983398438,
      "learning_rate": 3.627663272309139e-05,
      "loss": 72.5464,
      "step": 69410
    },
    {
      "epoch": 0.2804655841820966,
      "grad_norm": 1677.179443359375,
      "learning_rate": 3.627500982504908e-05,
      "loss": 79.6893,
      "step": 69420
    },
    {
      "epoch": 0.2805059854474642,
      "grad_norm": 921.6911010742188,
      "learning_rate": 3.6273386609717076e-05,
      "loss": 89.6619,
      "step": 69430
    },
    {
      "epoch": 0.28054638671283183,
      "grad_norm": 1072.14111328125,
      "learning_rate": 3.627176307712702e-05,
      "loss": 56.191,
      "step": 69440
    },
    {
      "epoch": 0.28058678797819947,
      "grad_norm": 885.6234130859375,
      "learning_rate": 3.627013922731056e-05,
      "loss": 70.9728,
      "step": 69450
    },
    {
      "epoch": 0.2806271892435671,
      "grad_norm": 1424.7335205078125,
      "learning_rate": 3.626851506029937e-05,
      "loss": 60.7801,
      "step": 69460
    },
    {
      "epoch": 0.28066759050893475,
      "grad_norm": 1208.6298828125,
      "learning_rate": 3.6266890576125095e-05,
      "loss": 68.1118,
      "step": 69470
    },
    {
      "epoch": 0.2807079917743024,
      "grad_norm": 2981.10107421875,
      "learning_rate": 3.6265265774819415e-05,
      "loss": 83.2845,
      "step": 69480
    },
    {
      "epoch": 0.28074839303967003,
      "grad_norm": 630.0430297851562,
      "learning_rate": 3.626364065641401e-05,
      "loss": 49.1042,
      "step": 69490
    },
    {
      "epoch": 0.2807887943050376,
      "grad_norm": 621.1134643554688,
      "learning_rate": 3.6262015220940556e-05,
      "loss": 118.1,
      "step": 69500
    },
    {
      "epoch": 0.28082919557040525,
      "grad_norm": 781.4208984375,
      "learning_rate": 3.626038946843074e-05,
      "loss": 64.6514,
      "step": 69510
    },
    {
      "epoch": 0.2808695968357729,
      "grad_norm": 446.54052734375,
      "learning_rate": 3.625876339891626e-05,
      "loss": 90.1646,
      "step": 69520
    },
    {
      "epoch": 0.28090999810114053,
      "grad_norm": 615.22900390625,
      "learning_rate": 3.625713701242882e-05,
      "loss": 69.1315,
      "step": 69530
    },
    {
      "epoch": 0.2809503993665082,
      "grad_norm": 791.39501953125,
      "learning_rate": 3.6255510309000126e-05,
      "loss": 74.8812,
      "step": 69540
    },
    {
      "epoch": 0.2809908006318758,
      "grad_norm": 995.6783447265625,
      "learning_rate": 3.625388328866188e-05,
      "loss": 68.0451,
      "step": 69550
    },
    {
      "epoch": 0.2810312018972434,
      "grad_norm": 277.51214599609375,
      "learning_rate": 3.625225595144582e-05,
      "loss": 85.4179,
      "step": 69560
    },
    {
      "epoch": 0.28107160316261104,
      "grad_norm": 1276.77685546875,
      "learning_rate": 3.6250628297383664e-05,
      "loss": 65.1418,
      "step": 69570
    },
    {
      "epoch": 0.2811120044279787,
      "grad_norm": 551.7674560546875,
      "learning_rate": 3.624900032650714e-05,
      "loss": 73.1631,
      "step": 69580
    },
    {
      "epoch": 0.2811524056933463,
      "grad_norm": 577.58544921875,
      "learning_rate": 3.624737203884798e-05,
      "loss": 50.5589,
      "step": 69590
    },
    {
      "epoch": 0.28119280695871396,
      "grad_norm": 1459.6531982421875,
      "learning_rate": 3.624574343443794e-05,
      "loss": 108.8682,
      "step": 69600
    },
    {
      "epoch": 0.2812332082240816,
      "grad_norm": 866.807861328125,
      "learning_rate": 3.624411451330878e-05,
      "loss": 69.1802,
      "step": 69610
    },
    {
      "epoch": 0.2812736094894492,
      "grad_norm": 476.4073791503906,
      "learning_rate": 3.624248527549224e-05,
      "loss": 53.6986,
      "step": 69620
    },
    {
      "epoch": 0.2813140107548168,
      "grad_norm": 829.8703002929688,
      "learning_rate": 3.624085572102009e-05,
      "loss": 69.2045,
      "step": 69630
    },
    {
      "epoch": 0.28135441202018446,
      "grad_norm": 836.7081298828125,
      "learning_rate": 3.6239225849924086e-05,
      "loss": 75.2816,
      "step": 69640
    },
    {
      "epoch": 0.2813948132855521,
      "grad_norm": 838.9620971679688,
      "learning_rate": 3.623759566223602e-05,
      "loss": 79.1252,
      "step": 69650
    },
    {
      "epoch": 0.28143521455091974,
      "grad_norm": 1362.0731201171875,
      "learning_rate": 3.6235965157987665e-05,
      "loss": 72.6706,
      "step": 69660
    },
    {
      "epoch": 0.2814756158162874,
      "grad_norm": 597.159912109375,
      "learning_rate": 3.623433433721081e-05,
      "loss": 72.3076,
      "step": 69670
    },
    {
      "epoch": 0.281516017081655,
      "grad_norm": 471.72637939453125,
      "learning_rate": 3.6232703199937246e-05,
      "loss": 39.1326,
      "step": 69680
    },
    {
      "epoch": 0.2815564183470226,
      "grad_norm": 645.277099609375,
      "learning_rate": 3.6231071746198784e-05,
      "loss": 97.0042,
      "step": 69690
    },
    {
      "epoch": 0.28159681961239025,
      "grad_norm": 985.9708862304688,
      "learning_rate": 3.6229439976027225e-05,
      "loss": 62.6094,
      "step": 69700
    },
    {
      "epoch": 0.2816372208777579,
      "grad_norm": 535.4151611328125,
      "learning_rate": 3.622780788945437e-05,
      "loss": 75.3714,
      "step": 69710
    },
    {
      "epoch": 0.28167762214312553,
      "grad_norm": 381.4093017578125,
      "learning_rate": 3.622617548651205e-05,
      "loss": 93.5277,
      "step": 69720
    },
    {
      "epoch": 0.28171802340849317,
      "grad_norm": 681.9583740234375,
      "learning_rate": 3.6224542767232076e-05,
      "loss": 66.7195,
      "step": 69730
    },
    {
      "epoch": 0.2817584246738608,
      "grad_norm": 748.0338745117188,
      "learning_rate": 3.62229097316463e-05,
      "loss": 112.8168,
      "step": 69740
    },
    {
      "epoch": 0.2817988259392284,
      "grad_norm": 588.4625854492188,
      "learning_rate": 3.6221276379786534e-05,
      "loss": 85.1326,
      "step": 69750
    },
    {
      "epoch": 0.28183922720459603,
      "grad_norm": 489.44464111328125,
      "learning_rate": 3.621964271168464e-05,
      "loss": 84.1481,
      "step": 69760
    },
    {
      "epoch": 0.2818796284699637,
      "grad_norm": 875.048828125,
      "learning_rate": 3.621800872737247e-05,
      "loss": 62.7425,
      "step": 69770
    },
    {
      "epoch": 0.2819200297353313,
      "grad_norm": 923.087646484375,
      "learning_rate": 3.6216374426881854e-05,
      "loss": 67.2313,
      "step": 69780
    },
    {
      "epoch": 0.28196043100069895,
      "grad_norm": 1275.936279296875,
      "learning_rate": 3.6214739810244684e-05,
      "loss": 69.6002,
      "step": 69790
    },
    {
      "epoch": 0.2820008322660666,
      "grad_norm": 778.4428100585938,
      "learning_rate": 3.62131048774928e-05,
      "loss": 58.0094,
      "step": 69800
    },
    {
      "epoch": 0.28204123353143423,
      "grad_norm": 868.2858276367188,
      "learning_rate": 3.62114696286581e-05,
      "loss": 86.843,
      "step": 69810
    },
    {
      "epoch": 0.2820816347968018,
      "grad_norm": 390.36212158203125,
      "learning_rate": 3.6209834063772446e-05,
      "loss": 36.7542,
      "step": 69820
    },
    {
      "epoch": 0.28212203606216946,
      "grad_norm": 660.957763671875,
      "learning_rate": 3.6208198182867737e-05,
      "loss": 66.1943,
      "step": 69830
    },
    {
      "epoch": 0.2821624373275371,
      "grad_norm": 882.1265869140625,
      "learning_rate": 3.620656198597586e-05,
      "loss": 115.4422,
      "step": 69840
    },
    {
      "epoch": 0.28220283859290474,
      "grad_norm": 762.0680541992188,
      "learning_rate": 3.620492547312871e-05,
      "loss": 62.1258,
      "step": 69850
    },
    {
      "epoch": 0.2822432398582724,
      "grad_norm": 1419.7911376953125,
      "learning_rate": 3.62032886443582e-05,
      "loss": 59.704,
      "step": 69860
    },
    {
      "epoch": 0.28228364112364,
      "grad_norm": 1616.5609130859375,
      "learning_rate": 3.620165149969623e-05,
      "loss": 102.7321,
      "step": 69870
    },
    {
      "epoch": 0.2823240423890076,
      "grad_norm": 544.9016723632812,
      "learning_rate": 3.620001403917472e-05,
      "loss": 105.496,
      "step": 69880
    },
    {
      "epoch": 0.28236444365437524,
      "grad_norm": 887.2398071289062,
      "learning_rate": 3.6198376262825606e-05,
      "loss": 88.13,
      "step": 69890
    },
    {
      "epoch": 0.2824048449197429,
      "grad_norm": 1018.4188842773438,
      "learning_rate": 3.61967381706808e-05,
      "loss": 75.7496,
      "step": 69900
    },
    {
      "epoch": 0.2824452461851105,
      "grad_norm": 677.3698120117188,
      "learning_rate": 3.619509976277225e-05,
      "loss": 83.6201,
      "step": 69910
    },
    {
      "epoch": 0.28248564745047816,
      "grad_norm": 816.182861328125,
      "learning_rate": 3.619346103913189e-05,
      "loss": 51.5256,
      "step": 69920
    },
    {
      "epoch": 0.2825260487158458,
      "grad_norm": 779.9092407226562,
      "learning_rate": 3.6191821999791665e-05,
      "loss": 46.279,
      "step": 69930
    },
    {
      "epoch": 0.2825664499812134,
      "grad_norm": 862.4857788085938,
      "learning_rate": 3.619018264478354e-05,
      "loss": 64.6215,
      "step": 69940
    },
    {
      "epoch": 0.282606851246581,
      "grad_norm": 1061.5059814453125,
      "learning_rate": 3.6188542974139466e-05,
      "loss": 70.1019,
      "step": 69950
    },
    {
      "epoch": 0.28264725251194867,
      "grad_norm": 980.0887451171875,
      "learning_rate": 3.618690298789142e-05,
      "loss": 57.2568,
      "step": 69960
    },
    {
      "epoch": 0.2826876537773163,
      "grad_norm": 683.5441284179688,
      "learning_rate": 3.618526268607136e-05,
      "loss": 103.2784,
      "step": 69970
    },
    {
      "epoch": 0.28272805504268395,
      "grad_norm": 731.2699584960938,
      "learning_rate": 3.618362206871127e-05,
      "loss": 72.0942,
      "step": 69980
    },
    {
      "epoch": 0.2827684563080516,
      "grad_norm": 914.0678100585938,
      "learning_rate": 3.6181981135843134e-05,
      "loss": 45.8803,
      "step": 69990
    },
    {
      "epoch": 0.2828088575734192,
      "grad_norm": 1745.98486328125,
      "learning_rate": 3.6180339887498953e-05,
      "loss": 70.0135,
      "step": 70000
    },
    {
      "epoch": 0.2828492588387868,
      "grad_norm": 538.3794555664062,
      "learning_rate": 3.617869832371071e-05,
      "loss": 87.8444,
      "step": 70010
    },
    {
      "epoch": 0.28288966010415445,
      "grad_norm": 1619.2493896484375,
      "learning_rate": 3.617705644451041e-05,
      "loss": 60.7661,
      "step": 70020
    },
    {
      "epoch": 0.2829300613695221,
      "grad_norm": 870.5992431640625,
      "learning_rate": 3.617541424993007e-05,
      "loss": 61.3241,
      "step": 70030
    },
    {
      "epoch": 0.28297046263488973,
      "grad_norm": 293.04705810546875,
      "learning_rate": 3.617377174000171e-05,
      "loss": 77.9364,
      "step": 70040
    },
    {
      "epoch": 0.28301086390025737,
      "grad_norm": 715.2342529296875,
      "learning_rate": 3.617212891475733e-05,
      "loss": 102.4682,
      "step": 70050
    },
    {
      "epoch": 0.283051265165625,
      "grad_norm": 621.2023315429688,
      "learning_rate": 3.6170485774228976e-05,
      "loss": 67.3741,
      "step": 70060
    },
    {
      "epoch": 0.2830916664309926,
      "grad_norm": 952.3382568359375,
      "learning_rate": 3.6168842318448684e-05,
      "loss": 83.3982,
      "step": 70070
    },
    {
      "epoch": 0.28313206769636023,
      "grad_norm": 489.0035705566406,
      "learning_rate": 3.6167198547448476e-05,
      "loss": 67.6127,
      "step": 70080
    },
    {
      "epoch": 0.2831724689617279,
      "grad_norm": 1169.779296875,
      "learning_rate": 3.616555446126041e-05,
      "loss": 62.9801,
      "step": 70090
    },
    {
      "epoch": 0.2832128702270955,
      "grad_norm": 601.5595092773438,
      "learning_rate": 3.6163910059916544e-05,
      "loss": 77.4872,
      "step": 70100
    },
    {
      "epoch": 0.28325327149246315,
      "grad_norm": 621.1578369140625,
      "learning_rate": 3.6162265343448924e-05,
      "loss": 68.0312,
      "step": 70110
    },
    {
      "epoch": 0.2832936727578308,
      "grad_norm": 530.8627319335938,
      "learning_rate": 3.616062031188962e-05,
      "loss": 56.4951,
      "step": 70120
    },
    {
      "epoch": 0.2833340740231984,
      "grad_norm": 786.9353637695312,
      "learning_rate": 3.615897496527071e-05,
      "loss": 80.8605,
      "step": 70130
    },
    {
      "epoch": 0.283374475288566,
      "grad_norm": 1017.2835083007812,
      "learning_rate": 3.615732930362426e-05,
      "loss": 80.876,
      "step": 70140
    },
    {
      "epoch": 0.28341487655393366,
      "grad_norm": 886.339111328125,
      "learning_rate": 3.615568332698235e-05,
      "loss": 83.4627,
      "step": 70150
    },
    {
      "epoch": 0.2834552778193013,
      "grad_norm": 617.3817749023438,
      "learning_rate": 3.6154037035377084e-05,
      "loss": 73.36,
      "step": 70160
    },
    {
      "epoch": 0.28349567908466894,
      "grad_norm": 974.203125,
      "learning_rate": 3.6152390428840546e-05,
      "loss": 63.3565,
      "step": 70170
    },
    {
      "epoch": 0.2835360803500366,
      "grad_norm": 1039.22216796875,
      "learning_rate": 3.6150743507404845e-05,
      "loss": 65.9822,
      "step": 70180
    },
    {
      "epoch": 0.2835764816154042,
      "grad_norm": 760.4146118164062,
      "learning_rate": 3.614909627110208e-05,
      "loss": 45.3543,
      "step": 70190
    },
    {
      "epoch": 0.2836168828807718,
      "grad_norm": 377.31842041015625,
      "learning_rate": 3.614744871996437e-05,
      "loss": 45.3849,
      "step": 70200
    },
    {
      "epoch": 0.28365728414613944,
      "grad_norm": 1098.9931640625,
      "learning_rate": 3.614580085402383e-05,
      "loss": 63.57,
      "step": 70210
    },
    {
      "epoch": 0.2836976854115071,
      "grad_norm": 1463.590576171875,
      "learning_rate": 3.6144152673312596e-05,
      "loss": 59.9706,
      "step": 70220
    },
    {
      "epoch": 0.2837380866768747,
      "grad_norm": 609.68896484375,
      "learning_rate": 3.6142504177862796e-05,
      "loss": 72.5341,
      "step": 70230
    },
    {
      "epoch": 0.28377848794224236,
      "grad_norm": 559.6856079101562,
      "learning_rate": 3.614085536770656e-05,
      "loss": 52.6119,
      "step": 70240
    },
    {
      "epoch": 0.28381888920761,
      "grad_norm": 695.3104248046875,
      "learning_rate": 3.613920624287604e-05,
      "loss": 63.0411,
      "step": 70250
    },
    {
      "epoch": 0.2838592904729776,
      "grad_norm": 281.5528259277344,
      "learning_rate": 3.613755680340339e-05,
      "loss": 71.9128,
      "step": 70260
    },
    {
      "epoch": 0.28389969173834523,
      "grad_norm": 812.212890625,
      "learning_rate": 3.6135907049320757e-05,
      "loss": 54.3615,
      "step": 70270
    },
    {
      "epoch": 0.28394009300371287,
      "grad_norm": 205.86573791503906,
      "learning_rate": 3.6134256980660306e-05,
      "loss": 54.9781,
      "step": 70280
    },
    {
      "epoch": 0.2839804942690805,
      "grad_norm": 707.3431396484375,
      "learning_rate": 3.613260659745422e-05,
      "loss": 90.469,
      "step": 70290
    },
    {
      "epoch": 0.28402089553444815,
      "grad_norm": 2262.424072265625,
      "learning_rate": 3.613095589973465e-05,
      "loss": 77.3084,
      "step": 70300
    },
    {
      "epoch": 0.2840612967998158,
      "grad_norm": 675.3787841796875,
      "learning_rate": 3.61293048875338e-05,
      "loss": 75.9326,
      "step": 70310
    },
    {
      "epoch": 0.28410169806518343,
      "grad_norm": 767.5772094726562,
      "learning_rate": 3.612765356088384e-05,
      "loss": 73.561,
      "step": 70320
    },
    {
      "epoch": 0.284142099330551,
      "grad_norm": 284.82177734375,
      "learning_rate": 3.612600191981697e-05,
      "loss": 56.5622,
      "step": 70330
    },
    {
      "epoch": 0.28418250059591865,
      "grad_norm": 402.4296569824219,
      "learning_rate": 3.61243499643654e-05,
      "loss": 92.6953,
      "step": 70340
    },
    {
      "epoch": 0.2842229018612863,
      "grad_norm": 1076.490234375,
      "learning_rate": 3.612269769456132e-05,
      "loss": 78.2423,
      "step": 70350
    },
    {
      "epoch": 0.28426330312665393,
      "grad_norm": 1266.93798828125,
      "learning_rate": 3.612104511043694e-05,
      "loss": 55.861,
      "step": 70360
    },
    {
      "epoch": 0.2843037043920216,
      "grad_norm": 1214.197265625,
      "learning_rate": 3.61193922120245e-05,
      "loss": 93.7413,
      "step": 70370
    },
    {
      "epoch": 0.2843441056573892,
      "grad_norm": 561.1460571289062,
      "learning_rate": 3.61177389993562e-05,
      "loss": 50.5075,
      "step": 70380
    },
    {
      "epoch": 0.2843845069227568,
      "grad_norm": 531.0656127929688,
      "learning_rate": 3.611608547246429e-05,
      "loss": 89.8019,
      "step": 70390
    },
    {
      "epoch": 0.28442490818812444,
      "grad_norm": 856.7954711914062,
      "learning_rate": 3.611443163138099e-05,
      "loss": 69.8518,
      "step": 70400
    },
    {
      "epoch": 0.2844653094534921,
      "grad_norm": 359.9229736328125,
      "learning_rate": 3.611277747613855e-05,
      "loss": 49.7166,
      "step": 70410
    },
    {
      "epoch": 0.2845057107188597,
      "grad_norm": 1801.592041015625,
      "learning_rate": 3.611112300676921e-05,
      "loss": 45.1849,
      "step": 70420
    },
    {
      "epoch": 0.28454611198422736,
      "grad_norm": 443.1506042480469,
      "learning_rate": 3.6109468223305244e-05,
      "loss": 68.0052,
      "step": 70430
    },
    {
      "epoch": 0.284586513249595,
      "grad_norm": 267.1731872558594,
      "learning_rate": 3.61078131257789e-05,
      "loss": 62.3781,
      "step": 70440
    },
    {
      "epoch": 0.2846269145149626,
      "grad_norm": 767.2887573242188,
      "learning_rate": 3.6106157714222436e-05,
      "loss": 56.1048,
      "step": 70450
    },
    {
      "epoch": 0.2846673157803302,
      "grad_norm": 774.2536010742188,
      "learning_rate": 3.610450198866815e-05,
      "loss": 95.8227,
      "step": 70460
    },
    {
      "epoch": 0.28470771704569786,
      "grad_norm": 1884.04443359375,
      "learning_rate": 3.61028459491483e-05,
      "loss": 70.254,
      "step": 70470
    },
    {
      "epoch": 0.2847481183110655,
      "grad_norm": 693.6580200195312,
      "learning_rate": 3.6101189595695174e-05,
      "loss": 52.211,
      "step": 70480
    },
    {
      "epoch": 0.28478851957643314,
      "grad_norm": 318.2597351074219,
      "learning_rate": 3.609953292834107e-05,
      "loss": 88.7317,
      "step": 70490
    },
    {
      "epoch": 0.2848289208418008,
      "grad_norm": 437.4230651855469,
      "learning_rate": 3.609787594711828e-05,
      "loss": 67.7461,
      "step": 70500
    },
    {
      "epoch": 0.2848693221071684,
      "grad_norm": 549.7716674804688,
      "learning_rate": 3.609621865205912e-05,
      "loss": 68.1587,
      "step": 70510
    },
    {
      "epoch": 0.284909723372536,
      "grad_norm": 830.4805297851562,
      "learning_rate": 3.609456104319589e-05,
      "loss": 90.8575,
      "step": 70520
    },
    {
      "epoch": 0.28495012463790365,
      "grad_norm": 321.4366149902344,
      "learning_rate": 3.609290312056089e-05,
      "loss": 80.5304,
      "step": 70530
    },
    {
      "epoch": 0.2849905259032713,
      "grad_norm": 1150.2452392578125,
      "learning_rate": 3.609124488418647e-05,
      "loss": 94.2768,
      "step": 70540
    },
    {
      "epoch": 0.2850309271686389,
      "grad_norm": 1228.3502197265625,
      "learning_rate": 3.608958633410495e-05,
      "loss": 102.8754,
      "step": 70550
    },
    {
      "epoch": 0.28507132843400657,
      "grad_norm": 1073.699951171875,
      "learning_rate": 3.608792747034866e-05,
      "loss": 67.9786,
      "step": 70560
    },
    {
      "epoch": 0.2851117296993742,
      "grad_norm": 853.513916015625,
      "learning_rate": 3.608626829294994e-05,
      "loss": 33.2278,
      "step": 70570
    },
    {
      "epoch": 0.2851521309647418,
      "grad_norm": 821.870361328125,
      "learning_rate": 3.608460880194113e-05,
      "loss": 55.8737,
      "step": 70580
    },
    {
      "epoch": 0.28519253223010943,
      "grad_norm": 634.192138671875,
      "learning_rate": 3.60829489973546e-05,
      "loss": 58.3211,
      "step": 70590
    },
    {
      "epoch": 0.28523293349547707,
      "grad_norm": 1230.9725341796875,
      "learning_rate": 3.6081288879222696e-05,
      "loss": 62.6132,
      "step": 70600
    },
    {
      "epoch": 0.2852733347608447,
      "grad_norm": 786.7669067382812,
      "learning_rate": 3.607962844757779e-05,
      "loss": 85.2633,
      "step": 70610
    },
    {
      "epoch": 0.28531373602621235,
      "grad_norm": 592.4337158203125,
      "learning_rate": 3.607796770245224e-05,
      "loss": 70.8817,
      "step": 70620
    },
    {
      "epoch": 0.28535413729158,
      "grad_norm": 1594.3375244140625,
      "learning_rate": 3.607630664387844e-05,
      "loss": 73.1753,
      "step": 70630
    },
    {
      "epoch": 0.28539453855694763,
      "grad_norm": 700.613037109375,
      "learning_rate": 3.6074645271888765e-05,
      "loss": 65.9696,
      "step": 70640
    },
    {
      "epoch": 0.2854349398223152,
      "grad_norm": 640.7323608398438,
      "learning_rate": 3.60729835865156e-05,
      "loss": 95.231,
      "step": 70650
    },
    {
      "epoch": 0.28547534108768285,
      "grad_norm": 749.009033203125,
      "learning_rate": 3.607132158779135e-05,
      "loss": 54.9461,
      "step": 70660
    },
    {
      "epoch": 0.2855157423530505,
      "grad_norm": 447.9740905761719,
      "learning_rate": 3.6069659275748404e-05,
      "loss": 70.5396,
      "step": 70670
    },
    {
      "epoch": 0.28555614361841813,
      "grad_norm": 1128.191162109375,
      "learning_rate": 3.606799665041918e-05,
      "loss": 60.2394,
      "step": 70680
    },
    {
      "epoch": 0.2855965448837858,
      "grad_norm": 796.18408203125,
      "learning_rate": 3.60663337118361e-05,
      "loss": 76.6932,
      "step": 70690
    },
    {
      "epoch": 0.2856369461491534,
      "grad_norm": 643.4987182617188,
      "learning_rate": 3.606467046003156e-05,
      "loss": 55.9977,
      "step": 70700
    },
    {
      "epoch": 0.285677347414521,
      "grad_norm": 1437.8402099609375,
      "learning_rate": 3.6063006895038004e-05,
      "loss": 71.8034,
      "step": 70710
    },
    {
      "epoch": 0.28571774867988864,
      "grad_norm": 445.0003662109375,
      "learning_rate": 3.6061343016887856e-05,
      "loss": 67.1427,
      "step": 70720
    },
    {
      "epoch": 0.2857581499452563,
      "grad_norm": 305.48779296875,
      "learning_rate": 3.605967882561356e-05,
      "loss": 52.1864,
      "step": 70730
    },
    {
      "epoch": 0.2857985512106239,
      "grad_norm": 667.2313232421875,
      "learning_rate": 3.6058014321247556e-05,
      "loss": 64.2683,
      "step": 70740
    },
    {
      "epoch": 0.28583895247599156,
      "grad_norm": 418.2080993652344,
      "learning_rate": 3.6056349503822295e-05,
      "loss": 73.9068,
      "step": 70750
    },
    {
      "epoch": 0.2858793537413592,
      "grad_norm": 7662.9677734375,
      "learning_rate": 3.605468437337023e-05,
      "loss": 93.897,
      "step": 70760
    },
    {
      "epoch": 0.2859197550067268,
      "grad_norm": 824.5432739257812,
      "learning_rate": 3.605301892992383e-05,
      "loss": 82.0532,
      "step": 70770
    },
    {
      "epoch": 0.2859601562720944,
      "grad_norm": 396.4519958496094,
      "learning_rate": 3.605135317351556e-05,
      "loss": 143.2694,
      "step": 70780
    },
    {
      "epoch": 0.28600055753746206,
      "grad_norm": 595.1510009765625,
      "learning_rate": 3.604968710417791e-05,
      "loss": 77.5454,
      "step": 70790
    },
    {
      "epoch": 0.2860409588028297,
      "grad_norm": 560.7315673828125,
      "learning_rate": 3.604802072194334e-05,
      "loss": 54.1081,
      "step": 70800
    },
    {
      "epoch": 0.28608136006819734,
      "grad_norm": 514.366455078125,
      "learning_rate": 3.604635402684434e-05,
      "loss": 88.6278,
      "step": 70810
    },
    {
      "epoch": 0.286121761333565,
      "grad_norm": 1537.122802734375,
      "learning_rate": 3.60446870189134e-05,
      "loss": 77.5263,
      "step": 70820
    },
    {
      "epoch": 0.2861621625989326,
      "grad_norm": 556.6965942382812,
      "learning_rate": 3.604301969818304e-05,
      "loss": 63.4476,
      "step": 70830
    },
    {
      "epoch": 0.2862025638643002,
      "grad_norm": 1061.7496337890625,
      "learning_rate": 3.604135206468574e-05,
      "loss": 55.0528,
      "step": 70840
    },
    {
      "epoch": 0.28624296512966785,
      "grad_norm": 916.4915771484375,
      "learning_rate": 3.603968411845402e-05,
      "loss": 64.2837,
      "step": 70850
    },
    {
      "epoch": 0.2862833663950355,
      "grad_norm": 970.0208740234375,
      "learning_rate": 3.603801585952042e-05,
      "loss": 88.3489,
      "step": 70860
    },
    {
      "epoch": 0.28632376766040313,
      "grad_norm": 469.9666442871094,
      "learning_rate": 3.603634728791743e-05,
      "loss": 60.465,
      "step": 70870
    },
    {
      "epoch": 0.28636416892577077,
      "grad_norm": 417.34771728515625,
      "learning_rate": 3.6034678403677595e-05,
      "loss": 43.7073,
      "step": 70880
    },
    {
      "epoch": 0.2864045701911384,
      "grad_norm": 715.1119384765625,
      "learning_rate": 3.603300920683345e-05,
      "loss": 57.8118,
      "step": 70890
    },
    {
      "epoch": 0.286444971456506,
      "grad_norm": 1027.1759033203125,
      "learning_rate": 3.6031339697417535e-05,
      "loss": 77.0941,
      "step": 70900
    },
    {
      "epoch": 0.28648537272187363,
      "grad_norm": 6179.7158203125,
      "learning_rate": 3.60296698754624e-05,
      "loss": 66.7193,
      "step": 70910
    },
    {
      "epoch": 0.2865257739872413,
      "grad_norm": 1350.91259765625,
      "learning_rate": 3.6027999741000596e-05,
      "loss": 64.565,
      "step": 70920
    },
    {
      "epoch": 0.2865661752526089,
      "grad_norm": 1316.241455078125,
      "learning_rate": 3.602632929406469e-05,
      "loss": 70.2296,
      "step": 70930
    },
    {
      "epoch": 0.28660657651797655,
      "grad_norm": 605.462646484375,
      "learning_rate": 3.602465853468724e-05,
      "loss": 81.5859,
      "step": 70940
    },
    {
      "epoch": 0.2866469777833442,
      "grad_norm": 964.7286987304688,
      "learning_rate": 3.6022987462900824e-05,
      "loss": 66.2818,
      "step": 70950
    },
    {
      "epoch": 0.28668737904871183,
      "grad_norm": 455.8790588378906,
      "learning_rate": 3.602131607873801e-05,
      "loss": 97.039,
      "step": 70960
    },
    {
      "epoch": 0.2867277803140794,
      "grad_norm": 1274.9256591796875,
      "learning_rate": 3.601964438223141e-05,
      "loss": 85.475,
      "step": 70970
    },
    {
      "epoch": 0.28676818157944706,
      "grad_norm": 2069.627685546875,
      "learning_rate": 3.601797237341358e-05,
      "loss": 103.252,
      "step": 70980
    },
    {
      "epoch": 0.2868085828448147,
      "grad_norm": 718.9536743164062,
      "learning_rate": 3.6016300052317135e-05,
      "loss": 67.6074,
      "step": 70990
    },
    {
      "epoch": 0.28684898411018234,
      "grad_norm": 755.44873046875,
      "learning_rate": 3.601462741897467e-05,
      "loss": 55.8639,
      "step": 71000
    },
    {
      "epoch": 0.28688938537555,
      "grad_norm": 585.0748291015625,
      "learning_rate": 3.6012954473418806e-05,
      "loss": 74.2655,
      "step": 71010
    },
    {
      "epoch": 0.2869297866409176,
      "grad_norm": 611.1162109375,
      "learning_rate": 3.601128121568215e-05,
      "loss": 72.3695,
      "step": 71020
    },
    {
      "epoch": 0.2869701879062852,
      "grad_norm": 527.9596557617188,
      "learning_rate": 3.600960764579732e-05,
      "loss": 55.1402,
      "step": 71030
    },
    {
      "epoch": 0.28701058917165284,
      "grad_norm": 466.97265625,
      "learning_rate": 3.6007933763796946e-05,
      "loss": 48.7595,
      "step": 71040
    },
    {
      "epoch": 0.2870509904370205,
      "grad_norm": 872.8701782226562,
      "learning_rate": 3.6006259569713665e-05,
      "loss": 68.9266,
      "step": 71050
    },
    {
      "epoch": 0.2870913917023881,
      "grad_norm": 1329.21044921875,
      "learning_rate": 3.600458506358011e-05,
      "loss": 60.3758,
      "step": 71060
    },
    {
      "epoch": 0.28713179296775576,
      "grad_norm": 789.2830810546875,
      "learning_rate": 3.6002910245428936e-05,
      "loss": 46.2742,
      "step": 71070
    },
    {
      "epoch": 0.2871721942331234,
      "grad_norm": 1676.2977294921875,
      "learning_rate": 3.6001235115292784e-05,
      "loss": 60.3875,
      "step": 71080
    },
    {
      "epoch": 0.287212595498491,
      "grad_norm": 543.3878173828125,
      "learning_rate": 3.599955967320432e-05,
      "loss": 45.1755,
      "step": 71090
    },
    {
      "epoch": 0.2872529967638586,
      "grad_norm": 2713.2490234375,
      "learning_rate": 3.5997883919196193e-05,
      "loss": 80.3693,
      "step": 71100
    },
    {
      "epoch": 0.28729339802922627,
      "grad_norm": 212.3776092529297,
      "learning_rate": 3.5996207853301084e-05,
      "loss": 58.0897,
      "step": 71110
    },
    {
      "epoch": 0.2873337992945939,
      "grad_norm": 613.86767578125,
      "learning_rate": 3.599453147555167e-05,
      "loss": 40.5792,
      "step": 71120
    },
    {
      "epoch": 0.28737420055996155,
      "grad_norm": 752.4508056640625,
      "learning_rate": 3.5992854785980635e-05,
      "loss": 111.4546,
      "step": 71130
    },
    {
      "epoch": 0.2874146018253292,
      "grad_norm": 739.1114501953125,
      "learning_rate": 3.599117778462066e-05,
      "loss": 78.3358,
      "step": 71140
    },
    {
      "epoch": 0.2874550030906968,
      "grad_norm": 1881.143798828125,
      "learning_rate": 3.598950047150444e-05,
      "loss": 114.0669,
      "step": 71150
    },
    {
      "epoch": 0.2874954043560644,
      "grad_norm": 781.6092529296875,
      "learning_rate": 3.598782284666467e-05,
      "loss": 91.2981,
      "step": 71160
    },
    {
      "epoch": 0.28753580562143205,
      "grad_norm": 635.6123657226562,
      "learning_rate": 3.5986144910134074e-05,
      "loss": 44.6738,
      "step": 71170
    },
    {
      "epoch": 0.2875762068867997,
      "grad_norm": 714.689208984375,
      "learning_rate": 3.5984466661945346e-05,
      "loss": 94.8069,
      "step": 71180
    },
    {
      "epoch": 0.28761660815216733,
      "grad_norm": 1838.5892333984375,
      "learning_rate": 3.598278810213121e-05,
      "loss": 102.966,
      "step": 71190
    },
    {
      "epoch": 0.28765700941753497,
      "grad_norm": 723.8616333007812,
      "learning_rate": 3.59811092307244e-05,
      "loss": 69.6785,
      "step": 71200
    },
    {
      "epoch": 0.2876974106829026,
      "grad_norm": 1040.573974609375,
      "learning_rate": 3.5979430047757634e-05,
      "loss": 61.6997,
      "step": 71210
    },
    {
      "epoch": 0.2877378119482702,
      "grad_norm": 1204.720458984375,
      "learning_rate": 3.597775055326365e-05,
      "loss": 59.2142,
      "step": 71220
    },
    {
      "epoch": 0.28777821321363783,
      "grad_norm": 1090.214599609375,
      "learning_rate": 3.5976070747275194e-05,
      "loss": 52.0799,
      "step": 71230
    },
    {
      "epoch": 0.2878186144790055,
      "grad_norm": 318.8838806152344,
      "learning_rate": 3.597439062982501e-05,
      "loss": 75.3177,
      "step": 71240
    },
    {
      "epoch": 0.2878590157443731,
      "grad_norm": 2042.7188720703125,
      "learning_rate": 3.597271020094586e-05,
      "loss": 96.1377,
      "step": 71250
    },
    {
      "epoch": 0.28789941700974075,
      "grad_norm": 280.46185302734375,
      "learning_rate": 3.59710294606705e-05,
      "loss": 79.229,
      "step": 71260
    },
    {
      "epoch": 0.2879398182751084,
      "grad_norm": 601.186767578125,
      "learning_rate": 3.59693484090317e-05,
      "loss": 96.3455,
      "step": 71270
    },
    {
      "epoch": 0.28798021954047603,
      "grad_norm": 658.820556640625,
      "learning_rate": 3.596766704606223e-05,
      "loss": 102.5423,
      "step": 71280
    },
    {
      "epoch": 0.2880206208058436,
      "grad_norm": 1202.2777099609375,
      "learning_rate": 3.596598537179487e-05,
      "loss": 68.3238,
      "step": 71290
    },
    {
      "epoch": 0.28806102207121126,
      "grad_norm": 434.55828857421875,
      "learning_rate": 3.596430338626241e-05,
      "loss": 63.3015,
      "step": 71300
    },
    {
      "epoch": 0.2881014233365789,
      "grad_norm": 895.1974487304688,
      "learning_rate": 3.5962621089497634e-05,
      "loss": 60.5648,
      "step": 71310
    },
    {
      "epoch": 0.28814182460194654,
      "grad_norm": 458.9768371582031,
      "learning_rate": 3.596093848153334e-05,
      "loss": 59.3895,
      "step": 71320
    },
    {
      "epoch": 0.2881822258673142,
      "grad_norm": 903.0233764648438,
      "learning_rate": 3.595925556240233e-05,
      "loss": 72.1123,
      "step": 71330
    },
    {
      "epoch": 0.2882226271326818,
      "grad_norm": 1241.1566162109375,
      "learning_rate": 3.595757233213742e-05,
      "loss": 60.014,
      "step": 71340
    },
    {
      "epoch": 0.2882630283980494,
      "grad_norm": 665.7273559570312,
      "learning_rate": 3.5955888790771426e-05,
      "loss": 72.169,
      "step": 71350
    },
    {
      "epoch": 0.28830342966341704,
      "grad_norm": 780.5765991210938,
      "learning_rate": 3.5954204938337156e-05,
      "loss": 70.9737,
      "step": 71360
    },
    {
      "epoch": 0.2883438309287847,
      "grad_norm": 769.46728515625,
      "learning_rate": 3.5952520774867454e-05,
      "loss": 66.6491,
      "step": 71370
    },
    {
      "epoch": 0.2883842321941523,
      "grad_norm": 876.2539672851562,
      "learning_rate": 3.5950836300395146e-05,
      "loss": 54.9499,
      "step": 71380
    },
    {
      "epoch": 0.28842463345951996,
      "grad_norm": 894.6849365234375,
      "learning_rate": 3.5949151514953074e-05,
      "loss": 63.0425,
      "step": 71390
    },
    {
      "epoch": 0.2884650347248876,
      "grad_norm": 847.6256103515625,
      "learning_rate": 3.594746641857407e-05,
      "loss": 73.8627,
      "step": 71400
    },
    {
      "epoch": 0.2885054359902552,
      "grad_norm": 959.1056518554688,
      "learning_rate": 3.594578101129101e-05,
      "loss": 91.9577,
      "step": 71410
    },
    {
      "epoch": 0.28854583725562283,
      "grad_norm": 566.6630859375,
      "learning_rate": 3.594409529313674e-05,
      "loss": 65.078,
      "step": 71420
    },
    {
      "epoch": 0.28858623852099047,
      "grad_norm": 769.6134033203125,
      "learning_rate": 3.594240926414412e-05,
      "loss": 71.833,
      "step": 71430
    },
    {
      "epoch": 0.2886266397863581,
      "grad_norm": 825.9446411132812,
      "learning_rate": 3.594072292434602e-05,
      "loss": 58.1939,
      "step": 71440
    },
    {
      "epoch": 0.28866704105172575,
      "grad_norm": 1975.863525390625,
      "learning_rate": 3.593903627377533e-05,
      "loss": 84.32,
      "step": 71450
    },
    {
      "epoch": 0.2887074423170934,
      "grad_norm": 729.3330078125,
      "learning_rate": 3.593734931246491e-05,
      "loss": 46.2589,
      "step": 71460
    },
    {
      "epoch": 0.28874784358246103,
      "grad_norm": 824.732666015625,
      "learning_rate": 3.593566204044767e-05,
      "loss": 55.354,
      "step": 71470
    },
    {
      "epoch": 0.2887882448478286,
      "grad_norm": 739.683349609375,
      "learning_rate": 3.5933974457756494e-05,
      "loss": 67.6329,
      "step": 71480
    },
    {
      "epoch": 0.28882864611319625,
      "grad_norm": 685.531005859375,
      "learning_rate": 3.593228656442427e-05,
      "loss": 73.9774,
      "step": 71490
    },
    {
      "epoch": 0.2888690473785639,
      "grad_norm": 1414.84814453125,
      "learning_rate": 3.5930598360483926e-05,
      "loss": 58.4536,
      "step": 71500
    },
    {
      "epoch": 0.28890944864393153,
      "grad_norm": 992.2027587890625,
      "learning_rate": 3.592890984596837e-05,
      "loss": 60.3957,
      "step": 71510
    },
    {
      "epoch": 0.2889498499092992,
      "grad_norm": 1208.000244140625,
      "learning_rate": 3.592722102091051e-05,
      "loss": 64.489,
      "step": 71520
    },
    {
      "epoch": 0.2889902511746668,
      "grad_norm": 1417.686279296875,
      "learning_rate": 3.592553188534328e-05,
      "loss": 82.0553,
      "step": 71530
    },
    {
      "epoch": 0.2890306524400344,
      "grad_norm": 432.866943359375,
      "learning_rate": 3.59238424392996e-05,
      "loss": 50.9258,
      "step": 71540
    },
    {
      "epoch": 0.28907105370540204,
      "grad_norm": 982.0166625976562,
      "learning_rate": 3.592215268281242e-05,
      "loss": 55.183,
      "step": 71550
    },
    {
      "epoch": 0.2891114549707697,
      "grad_norm": 462.43292236328125,
      "learning_rate": 3.592046261591467e-05,
      "loss": 75.8433,
      "step": 71560
    },
    {
      "epoch": 0.2891518562361373,
      "grad_norm": 1216.3463134765625,
      "learning_rate": 3.5918772238639304e-05,
      "loss": 77.9336,
      "step": 71570
    },
    {
      "epoch": 0.28919225750150496,
      "grad_norm": 2263.478271484375,
      "learning_rate": 3.591708155101928e-05,
      "loss": 67.5811,
      "step": 71580
    },
    {
      "epoch": 0.2892326587668726,
      "grad_norm": 998.69873046875,
      "learning_rate": 3.591539055308756e-05,
      "loss": 72.0668,
      "step": 71590
    },
    {
      "epoch": 0.28927306003224024,
      "grad_norm": 782.7774047851562,
      "learning_rate": 3.591369924487711e-05,
      "loss": 84.08,
      "step": 71600
    },
    {
      "epoch": 0.2893134612976078,
      "grad_norm": 746.9514770507812,
      "learning_rate": 3.59120076264209e-05,
      "loss": 39.7181,
      "step": 71610
    },
    {
      "epoch": 0.28935386256297546,
      "grad_norm": 480.704345703125,
      "learning_rate": 3.59103156977519e-05,
      "loss": 75.6742,
      "step": 71620
    },
    {
      "epoch": 0.2893942638283431,
      "grad_norm": 823.4539794921875,
      "learning_rate": 3.590862345890311e-05,
      "loss": 79.9501,
      "step": 71630
    },
    {
      "epoch": 0.28943466509371074,
      "grad_norm": 3223.1875,
      "learning_rate": 3.590693090990752e-05,
      "loss": 100.8133,
      "step": 71640
    },
    {
      "epoch": 0.2894750663590784,
      "grad_norm": 270.4209289550781,
      "learning_rate": 3.590523805079812e-05,
      "loss": 52.6623,
      "step": 71650
    },
    {
      "epoch": 0.289515467624446,
      "grad_norm": 896.4325561523438,
      "learning_rate": 3.590354488160791e-05,
      "loss": 81.3142,
      "step": 71660
    },
    {
      "epoch": 0.2895558688898136,
      "grad_norm": 295.3815612792969,
      "learning_rate": 3.5901851402369905e-05,
      "loss": 71.8267,
      "step": 71670
    },
    {
      "epoch": 0.28959627015518125,
      "grad_norm": 982.9450073242188,
      "learning_rate": 3.590015761311713e-05,
      "loss": 64.7104,
      "step": 71680
    },
    {
      "epoch": 0.2896366714205489,
      "grad_norm": 1250.133056640625,
      "learning_rate": 3.5898463513882584e-05,
      "loss": 105.7186,
      "step": 71690
    },
    {
      "epoch": 0.2896770726859165,
      "grad_norm": 321.7370300292969,
      "learning_rate": 3.589676910469932e-05,
      "loss": 54.8599,
      "step": 71700
    },
    {
      "epoch": 0.28971747395128417,
      "grad_norm": 935.1970825195312,
      "learning_rate": 3.589507438560034e-05,
      "loss": 88.217,
      "step": 71710
    },
    {
      "epoch": 0.2897578752166518,
      "grad_norm": 1932.10693359375,
      "learning_rate": 3.589337935661871e-05,
      "loss": 97.3057,
      "step": 71720
    },
    {
      "epoch": 0.2897982764820194,
      "grad_norm": 1101.5948486328125,
      "learning_rate": 3.589168401778747e-05,
      "loss": 85.077,
      "step": 71730
    },
    {
      "epoch": 0.28983867774738703,
      "grad_norm": 1101.748779296875,
      "learning_rate": 3.588998836913967e-05,
      "loss": 87.2077,
      "step": 71740
    },
    {
      "epoch": 0.28987907901275467,
      "grad_norm": 1247.614013671875,
      "learning_rate": 3.5888292410708364e-05,
      "loss": 65.8963,
      "step": 71750
    },
    {
      "epoch": 0.2899194802781223,
      "grad_norm": 644.5755615234375,
      "learning_rate": 3.588659614252662e-05,
      "loss": 48.0149,
      "step": 71760
    },
    {
      "epoch": 0.28995988154348995,
      "grad_norm": 538.0592041015625,
      "learning_rate": 3.5884899564627504e-05,
      "loss": 56.984,
      "step": 71770
    },
    {
      "epoch": 0.2900002828088576,
      "grad_norm": 988.62646484375,
      "learning_rate": 3.588320267704409e-05,
      "loss": 77.014,
      "step": 71780
    },
    {
      "epoch": 0.29004068407422523,
      "grad_norm": 681.0162353515625,
      "learning_rate": 3.588150547980946e-05,
      "loss": 63.0402,
      "step": 71790
    },
    {
      "epoch": 0.2900810853395928,
      "grad_norm": 1037.8740234375,
      "learning_rate": 3.587980797295671e-05,
      "loss": 67.6628,
      "step": 71800
    },
    {
      "epoch": 0.29012148660496045,
      "grad_norm": 2027.07568359375,
      "learning_rate": 3.587811015651893e-05,
      "loss": 45.8344,
      "step": 71810
    },
    {
      "epoch": 0.2901618878703281,
      "grad_norm": 728.1589965820312,
      "learning_rate": 3.587641203052921e-05,
      "loss": 73.745,
      "step": 71820
    },
    {
      "epoch": 0.29020228913569573,
      "grad_norm": 612.0015869140625,
      "learning_rate": 3.587471359502066e-05,
      "loss": 57.5115,
      "step": 71830
    },
    {
      "epoch": 0.2902426904010634,
      "grad_norm": 1017.7273559570312,
      "learning_rate": 3.587301485002641e-05,
      "loss": 64.8747,
      "step": 71840
    },
    {
      "epoch": 0.290283091666431,
      "grad_norm": 1304.318115234375,
      "learning_rate": 3.587131579557956e-05,
      "loss": 83.5981,
      "step": 71850
    },
    {
      "epoch": 0.2903234929317986,
      "grad_norm": 724.4077758789062,
      "learning_rate": 3.5869616431713235e-05,
      "loss": 55.5681,
      "step": 71860
    },
    {
      "epoch": 0.29036389419716624,
      "grad_norm": 571.9274291992188,
      "learning_rate": 3.586791675846057e-05,
      "loss": 89.1574,
      "step": 71870
    },
    {
      "epoch": 0.2904042954625339,
      "grad_norm": 321.2984313964844,
      "learning_rate": 3.586621677585469e-05,
      "loss": 70.9063,
      "step": 71880
    },
    {
      "epoch": 0.2904446967279015,
      "grad_norm": 348.6775207519531,
      "learning_rate": 3.586451648392875e-05,
      "loss": 45.3349,
      "step": 71890
    },
    {
      "epoch": 0.29048509799326916,
      "grad_norm": 463.2661437988281,
      "learning_rate": 3.58628158827159e-05,
      "loss": 70.9898,
      "step": 71900
    },
    {
      "epoch": 0.2905254992586368,
      "grad_norm": 370.6292419433594,
      "learning_rate": 3.586111497224928e-05,
      "loss": 57.8408,
      "step": 71910
    },
    {
      "epoch": 0.29056590052400444,
      "grad_norm": 907.023193359375,
      "learning_rate": 3.585941375256206e-05,
      "loss": 81.2636,
      "step": 71920
    },
    {
      "epoch": 0.290606301789372,
      "grad_norm": 1051.9798583984375,
      "learning_rate": 3.58577122236874e-05,
      "loss": 66.2477,
      "step": 71930
    },
    {
      "epoch": 0.29064670305473966,
      "grad_norm": 763.0676879882812,
      "learning_rate": 3.585601038565848e-05,
      "loss": 62.0955,
      "step": 71940
    },
    {
      "epoch": 0.2906871043201073,
      "grad_norm": 473.68792724609375,
      "learning_rate": 3.585430823850847e-05,
      "loss": 83.7084,
      "step": 71950
    },
    {
      "epoch": 0.29072750558547494,
      "grad_norm": 742.2813110351562,
      "learning_rate": 3.5852605782270566e-05,
      "loss": 66.4683,
      "step": 71960
    },
    {
      "epoch": 0.2907679068508426,
      "grad_norm": 1462.05517578125,
      "learning_rate": 3.585090301697795e-05,
      "loss": 97.0444,
      "step": 71970
    },
    {
      "epoch": 0.2908083081162102,
      "grad_norm": 553.8416137695312,
      "learning_rate": 3.584919994266382e-05,
      "loss": 47.8547,
      "step": 71980
    },
    {
      "epoch": 0.2908487093815778,
      "grad_norm": 1666.500244140625,
      "learning_rate": 3.584749655936137e-05,
      "loss": 105.8823,
      "step": 71990
    },
    {
      "epoch": 0.29088911064694545,
      "grad_norm": 1440.423095703125,
      "learning_rate": 3.584579286710382e-05,
      "loss": 91.9585,
      "step": 72000
    },
    {
      "epoch": 0.2909295119123131,
      "grad_norm": 500.3939514160156,
      "learning_rate": 3.5844088865924376e-05,
      "loss": 55.625,
      "step": 72010
    },
    {
      "epoch": 0.29096991317768073,
      "grad_norm": 654.8592529296875,
      "learning_rate": 3.584238455585626e-05,
      "loss": 69.709,
      "step": 72020
    },
    {
      "epoch": 0.29101031444304837,
      "grad_norm": 395.8098449707031,
      "learning_rate": 3.5840679936932714e-05,
      "loss": 49.6149,
      "step": 72030
    },
    {
      "epoch": 0.291050715708416,
      "grad_norm": 667.9511108398438,
      "learning_rate": 3.5838975009186945e-05,
      "loss": 99.7107,
      "step": 72040
    },
    {
      "epoch": 0.2910911169737836,
      "grad_norm": 964.6243896484375,
      "learning_rate": 3.583726977265222e-05,
      "loss": 92.4027,
      "step": 72050
    },
    {
      "epoch": 0.29113151823915123,
      "grad_norm": 1799.1485595703125,
      "learning_rate": 3.583556422736175e-05,
      "loss": 123.6818,
      "step": 72060
    },
    {
      "epoch": 0.2911719195045189,
      "grad_norm": 562.0403442382812,
      "learning_rate": 3.583385837334881e-05,
      "loss": 37.064,
      "step": 72070
    },
    {
      "epoch": 0.2912123207698865,
      "grad_norm": 562.592529296875,
      "learning_rate": 3.5832152210646646e-05,
      "loss": 59.6503,
      "step": 72080
    },
    {
      "epoch": 0.29125272203525415,
      "grad_norm": 1167.1383056640625,
      "learning_rate": 3.583044573928853e-05,
      "loss": 77.4025,
      "step": 72090
    },
    {
      "epoch": 0.2912931233006218,
      "grad_norm": 2110.247802734375,
      "learning_rate": 3.5828738959307715e-05,
      "loss": 67.5149,
      "step": 72100
    },
    {
      "epoch": 0.29133352456598943,
      "grad_norm": 680.3276977539062,
      "learning_rate": 3.582703187073749e-05,
      "loss": 60.1068,
      "step": 72110
    },
    {
      "epoch": 0.291373925831357,
      "grad_norm": 495.9038391113281,
      "learning_rate": 3.582532447361114e-05,
      "loss": 64.1526,
      "step": 72120
    },
    {
      "epoch": 0.29141432709672466,
      "grad_norm": 968.1295166015625,
      "learning_rate": 3.5823616767961924e-05,
      "loss": 58.9892,
      "step": 72130
    },
    {
      "epoch": 0.2914547283620923,
      "grad_norm": 337.22491455078125,
      "learning_rate": 3.5821908753823163e-05,
      "loss": 54.9471,
      "step": 72140
    },
    {
      "epoch": 0.29149512962745994,
      "grad_norm": 1134.8824462890625,
      "learning_rate": 3.582020043122814e-05,
      "loss": 100.2467,
      "step": 72150
    },
    {
      "epoch": 0.2915355308928276,
      "grad_norm": 1638.614013671875,
      "learning_rate": 3.581849180021017e-05,
      "loss": 74.3465,
      "step": 72160
    },
    {
      "epoch": 0.2915759321581952,
      "grad_norm": 1058.03515625,
      "learning_rate": 3.581678286080256e-05,
      "loss": 90.0985,
      "step": 72170
    },
    {
      "epoch": 0.2916163334235628,
      "grad_norm": 329.9512634277344,
      "learning_rate": 3.581507361303862e-05,
      "loss": 48.0506,
      "step": 72180
    },
    {
      "epoch": 0.29165673468893044,
      "grad_norm": 653.0698852539062,
      "learning_rate": 3.5813364056951676e-05,
      "loss": 92.9112,
      "step": 72190
    },
    {
      "epoch": 0.2916971359542981,
      "grad_norm": 934.0549926757812,
      "learning_rate": 3.5811654192575064e-05,
      "loss": 50.3939,
      "step": 72200
    },
    {
      "epoch": 0.2917375372196657,
      "grad_norm": 562.1233520507812,
      "learning_rate": 3.580994401994211e-05,
      "loss": 82.1733,
      "step": 72210
    },
    {
      "epoch": 0.29177793848503336,
      "grad_norm": 485.84515380859375,
      "learning_rate": 3.5808233539086155e-05,
      "loss": 49.1954,
      "step": 72220
    },
    {
      "epoch": 0.291818339750401,
      "grad_norm": 1089.2716064453125,
      "learning_rate": 3.580652275004055e-05,
      "loss": 60.4214,
      "step": 72230
    },
    {
      "epoch": 0.29185874101576864,
      "grad_norm": 1303.6624755859375,
      "learning_rate": 3.580481165283865e-05,
      "loss": 101.3685,
      "step": 72240
    },
    {
      "epoch": 0.2918991422811362,
      "grad_norm": 951.2037963867188,
      "learning_rate": 3.580310024751381e-05,
      "loss": 82.3328,
      "step": 72250
    },
    {
      "epoch": 0.29193954354650387,
      "grad_norm": 659.6070556640625,
      "learning_rate": 3.5801388534099396e-05,
      "loss": 85.988,
      "step": 72260
    },
    {
      "epoch": 0.2919799448118715,
      "grad_norm": 524.1322631835938,
      "learning_rate": 3.5799676512628775e-05,
      "loss": 49.4976,
      "step": 72270
    },
    {
      "epoch": 0.29202034607723915,
      "grad_norm": 814.81591796875,
      "learning_rate": 3.579796418313532e-05,
      "loss": 51.3425,
      "step": 72280
    },
    {
      "epoch": 0.2920607473426068,
      "grad_norm": 313.1018371582031,
      "learning_rate": 3.5796251545652425e-05,
      "loss": 86.0607,
      "step": 72290
    },
    {
      "epoch": 0.2921011486079744,
      "grad_norm": 528.6152954101562,
      "learning_rate": 3.579453860021348e-05,
      "loss": 61.0077,
      "step": 72300
    },
    {
      "epoch": 0.292141549873342,
      "grad_norm": 1399.322998046875,
      "learning_rate": 3.579282534685186e-05,
      "loss": 99.063,
      "step": 72310
    },
    {
      "epoch": 0.29218195113870965,
      "grad_norm": 2749.59130859375,
      "learning_rate": 3.579111178560099e-05,
      "loss": 66.6921,
      "step": 72320
    },
    {
      "epoch": 0.2922223524040773,
      "grad_norm": 759.787841796875,
      "learning_rate": 3.578939791649426e-05,
      "loss": 102.9235,
      "step": 72330
    },
    {
      "epoch": 0.29226275366944493,
      "grad_norm": 730.5199584960938,
      "learning_rate": 3.5787683739565096e-05,
      "loss": 55.3452,
      "step": 72340
    },
    {
      "epoch": 0.29230315493481257,
      "grad_norm": 707.774658203125,
      "learning_rate": 3.57859692548469e-05,
      "loss": 81.0743,
      "step": 72350
    },
    {
      "epoch": 0.2923435562001802,
      "grad_norm": 1520.0377197265625,
      "learning_rate": 3.578425446237312e-05,
      "loss": 78.8926,
      "step": 72360
    },
    {
      "epoch": 0.2923839574655478,
      "grad_norm": 1314.706298828125,
      "learning_rate": 3.578253936217716e-05,
      "loss": 84.5639,
      "step": 72370
    },
    {
      "epoch": 0.29242435873091543,
      "grad_norm": 315.541748046875,
      "learning_rate": 3.578082395429247e-05,
      "loss": 62.6152,
      "step": 72380
    },
    {
      "epoch": 0.2924647599962831,
      "grad_norm": 801.6674194335938,
      "learning_rate": 3.57791082387525e-05,
      "loss": 47.1351,
      "step": 72390
    },
    {
      "epoch": 0.2925051612616507,
      "grad_norm": 1115.9136962890625,
      "learning_rate": 3.577739221559069e-05,
      "loss": 61.5287,
      "step": 72400
    },
    {
      "epoch": 0.29254556252701835,
      "grad_norm": 1236.024658203125,
      "learning_rate": 3.577567588484049e-05,
      "loss": 62.6104,
      "step": 72410
    },
    {
      "epoch": 0.292585963792386,
      "grad_norm": 508.2314147949219,
      "learning_rate": 3.577395924653537e-05,
      "loss": 53.6619,
      "step": 72420
    },
    {
      "epoch": 0.29262636505775363,
      "grad_norm": 645.4722290039062,
      "learning_rate": 3.577224230070879e-05,
      "loss": 78.5961,
      "step": 72430
    },
    {
      "epoch": 0.2926667663231212,
      "grad_norm": 1093.773681640625,
      "learning_rate": 3.577052504739423e-05,
      "loss": 81.6924,
      "step": 72440
    },
    {
      "epoch": 0.29270716758848886,
      "grad_norm": 799.0197143554688,
      "learning_rate": 3.5768807486625166e-05,
      "loss": 76.3117,
      "step": 72450
    },
    {
      "epoch": 0.2927475688538565,
      "grad_norm": 462.7156677246094,
      "learning_rate": 3.576708961843508e-05,
      "loss": 64.3802,
      "step": 72460
    },
    {
      "epoch": 0.29278797011922414,
      "grad_norm": 687.6199951171875,
      "learning_rate": 3.576537144285747e-05,
      "loss": 62.2381,
      "step": 72470
    },
    {
      "epoch": 0.2928283713845918,
      "grad_norm": 793.7186889648438,
      "learning_rate": 3.576365295992582e-05,
      "loss": 73.4337,
      "step": 72480
    },
    {
      "epoch": 0.2928687726499594,
      "grad_norm": 704.6602783203125,
      "learning_rate": 3.576193416967364e-05,
      "loss": 79.5488,
      "step": 72490
    },
    {
      "epoch": 0.292909173915327,
      "grad_norm": 447.6377258300781,
      "learning_rate": 3.576021507213444e-05,
      "loss": 55.2884,
      "step": 72500
    },
    {
      "epoch": 0.29294957518069464,
      "grad_norm": 1093.624267578125,
      "learning_rate": 3.575849566734174e-05,
      "loss": 55.568,
      "step": 72510
    },
    {
      "epoch": 0.2929899764460623,
      "grad_norm": 863.1414184570312,
      "learning_rate": 3.5756775955329045e-05,
      "loss": 52.4301,
      "step": 72520
    },
    {
      "epoch": 0.2930303777114299,
      "grad_norm": 726.0432739257812,
      "learning_rate": 3.575505593612989e-05,
      "loss": 42.6705,
      "step": 72530
    },
    {
      "epoch": 0.29307077897679756,
      "grad_norm": 426.0909423828125,
      "learning_rate": 3.575333560977782e-05,
      "loss": 57.5002,
      "step": 72540
    },
    {
      "epoch": 0.2931111802421652,
      "grad_norm": 985.9451293945312,
      "learning_rate": 3.5751614976306347e-05,
      "loss": 71.6082,
      "step": 72550
    },
    {
      "epoch": 0.29315158150753284,
      "grad_norm": 810.1474609375,
      "learning_rate": 3.574989403574904e-05,
      "loss": 103.1282,
      "step": 72560
    },
    {
      "epoch": 0.29319198277290043,
      "grad_norm": 467.8982849121094,
      "learning_rate": 3.574817278813943e-05,
      "loss": 78.4023,
      "step": 72570
    },
    {
      "epoch": 0.29323238403826807,
      "grad_norm": 615.1262817382812,
      "learning_rate": 3.574645123351109e-05,
      "loss": 33.2686,
      "step": 72580
    },
    {
      "epoch": 0.2932727853036357,
      "grad_norm": 2113.16845703125,
      "learning_rate": 3.5744729371897584e-05,
      "loss": 77.6977,
      "step": 72590
    },
    {
      "epoch": 0.29331318656900335,
      "grad_norm": 1216.817626953125,
      "learning_rate": 3.574300720333247e-05,
      "loss": 56.1763,
      "step": 72600
    },
    {
      "epoch": 0.293353587834371,
      "grad_norm": 1282.3033447265625,
      "learning_rate": 3.574128472784932e-05,
      "loss": 49.9721,
      "step": 72610
    },
    {
      "epoch": 0.29339398909973863,
      "grad_norm": 502.74530029296875,
      "learning_rate": 3.573956194548172e-05,
      "loss": 43.6453,
      "step": 72620
    },
    {
      "epoch": 0.2934343903651062,
      "grad_norm": 842.286865234375,
      "learning_rate": 3.573783885626326e-05,
      "loss": 75.6466,
      "step": 72630
    },
    {
      "epoch": 0.29347479163047385,
      "grad_norm": 757.489501953125,
      "learning_rate": 3.573611546022753e-05,
      "loss": 100.9482,
      "step": 72640
    },
    {
      "epoch": 0.2935151928958415,
      "grad_norm": 525.7161254882812,
      "learning_rate": 3.5734391757408123e-05,
      "loss": 100.7201,
      "step": 72650
    },
    {
      "epoch": 0.29355559416120913,
      "grad_norm": 556.4071655273438,
      "learning_rate": 3.5732667747838654e-05,
      "loss": 55.8495,
      "step": 72660
    },
    {
      "epoch": 0.2935959954265768,
      "grad_norm": 629.6777954101562,
      "learning_rate": 3.573094343155272e-05,
      "loss": 68.8656,
      "step": 72670
    },
    {
      "epoch": 0.2936363966919444,
      "grad_norm": 315.14923095703125,
      "learning_rate": 3.572921880858395e-05,
      "loss": 58.5177,
      "step": 72680
    },
    {
      "epoch": 0.293676797957312,
      "grad_norm": 844.389892578125,
      "learning_rate": 3.572749387896596e-05,
      "loss": 58.0804,
      "step": 72690
    },
    {
      "epoch": 0.29371719922267964,
      "grad_norm": 829.093017578125,
      "learning_rate": 3.572576864273238e-05,
      "loss": 57.9544,
      "step": 72700
    },
    {
      "epoch": 0.2937576004880473,
      "grad_norm": 807.382568359375,
      "learning_rate": 3.572404309991685e-05,
      "loss": 81.5726,
      "step": 72710
    },
    {
      "epoch": 0.2937980017534149,
      "grad_norm": 1963.383544921875,
      "learning_rate": 3.5722317250553e-05,
      "loss": 170.714,
      "step": 72720
    },
    {
      "epoch": 0.29383840301878256,
      "grad_norm": 692.9369506835938,
      "learning_rate": 3.572059109467447e-05,
      "loss": 80.3831,
      "step": 72730
    },
    {
      "epoch": 0.2938788042841502,
      "grad_norm": 729.177490234375,
      "learning_rate": 3.5718864632314936e-05,
      "loss": 66.1752,
      "step": 72740
    },
    {
      "epoch": 0.29391920554951784,
      "grad_norm": 2613.15380859375,
      "learning_rate": 3.571713786350804e-05,
      "loss": 134.1205,
      "step": 72750
    },
    {
      "epoch": 0.2939596068148854,
      "grad_norm": 909.3247680664062,
      "learning_rate": 3.571541078828745e-05,
      "loss": 71.8222,
      "step": 72760
    },
    {
      "epoch": 0.29400000808025306,
      "grad_norm": 649.9159545898438,
      "learning_rate": 3.5713683406686834e-05,
      "loss": 76.7148,
      "step": 72770
    },
    {
      "epoch": 0.2940404093456207,
      "grad_norm": 413.2486267089844,
      "learning_rate": 3.571195571873988e-05,
      "loss": 67.5009,
      "step": 72780
    },
    {
      "epoch": 0.29408081061098834,
      "grad_norm": 471.48583984375,
      "learning_rate": 3.571022772448024e-05,
      "loss": 72.5633,
      "step": 72790
    },
    {
      "epoch": 0.294121211876356,
      "grad_norm": 337.2598571777344,
      "learning_rate": 3.570849942394164e-05,
      "loss": 84.3592,
      "step": 72800
    },
    {
      "epoch": 0.2941616131417236,
      "grad_norm": 1427.0047607421875,
      "learning_rate": 3.570677081715775e-05,
      "loss": 65.8118,
      "step": 72810
    },
    {
      "epoch": 0.2942020144070912,
      "grad_norm": 1616.63134765625,
      "learning_rate": 3.5705041904162274e-05,
      "loss": 63.8992,
      "step": 72820
    },
    {
      "epoch": 0.29424241567245885,
      "grad_norm": 4020.565185546875,
      "learning_rate": 3.5703312684988924e-05,
      "loss": 103.3089,
      "step": 72830
    },
    {
      "epoch": 0.2942828169378265,
      "grad_norm": 737.0076293945312,
      "learning_rate": 3.570158315967141e-05,
      "loss": 75.0354,
      "step": 72840
    },
    {
      "epoch": 0.2943232182031941,
      "grad_norm": 407.1966857910156,
      "learning_rate": 3.569985332824345e-05,
      "loss": 46.7118,
      "step": 72850
    },
    {
      "epoch": 0.29436361946856177,
      "grad_norm": 1618.404052734375,
      "learning_rate": 3.569812319073876e-05,
      "loss": 109.4477,
      "step": 72860
    },
    {
      "epoch": 0.2944040207339294,
      "grad_norm": 726.337646484375,
      "learning_rate": 3.5696392747191084e-05,
      "loss": 52.4097,
      "step": 72870
    },
    {
      "epoch": 0.29444442199929705,
      "grad_norm": 1107.2373046875,
      "learning_rate": 3.569466199763414e-05,
      "loss": 68.6211,
      "step": 72880
    },
    {
      "epoch": 0.29448482326466463,
      "grad_norm": 1022.0324096679688,
      "learning_rate": 3.569293094210169e-05,
      "loss": 51.532,
      "step": 72890
    },
    {
      "epoch": 0.29452522453003227,
      "grad_norm": 289.0499267578125,
      "learning_rate": 3.569119958062747e-05,
      "loss": 48.7533,
      "step": 72900
    },
    {
      "epoch": 0.2945656257953999,
      "grad_norm": 1667.3704833984375,
      "learning_rate": 3.568946791324523e-05,
      "loss": 89.5318,
      "step": 72910
    },
    {
      "epoch": 0.29460602706076755,
      "grad_norm": 422.2791442871094,
      "learning_rate": 3.5687735939988745e-05,
      "loss": 93.2784,
      "step": 72920
    },
    {
      "epoch": 0.2946464283261352,
      "grad_norm": 894.854248046875,
      "learning_rate": 3.5686003660891763e-05,
      "loss": 69.9056,
      "step": 72930
    },
    {
      "epoch": 0.29468682959150283,
      "grad_norm": 435.1180114746094,
      "learning_rate": 3.568427107598807e-05,
      "loss": 43.495,
      "step": 72940
    },
    {
      "epoch": 0.2947272308568704,
      "grad_norm": 742.6530151367188,
      "learning_rate": 3.568253818531143e-05,
      "loss": 79.4282,
      "step": 72950
    },
    {
      "epoch": 0.29476763212223805,
      "grad_norm": 2197.233642578125,
      "learning_rate": 3.568080498889564e-05,
      "loss": 79.8292,
      "step": 72960
    },
    {
      "epoch": 0.2948080333876057,
      "grad_norm": 1193.3448486328125,
      "learning_rate": 3.567907148677448e-05,
      "loss": 95.2717,
      "step": 72970
    },
    {
      "epoch": 0.29484843465297333,
      "grad_norm": 1066.093994140625,
      "learning_rate": 3.567733767898176e-05,
      "loss": 74.4719,
      "step": 72980
    },
    {
      "epoch": 0.294888835918341,
      "grad_norm": 525.0219116210938,
      "learning_rate": 3.567560356555126e-05,
      "loss": 49.5146,
      "step": 72990
    },
    {
      "epoch": 0.2949292371837086,
      "grad_norm": 1088.7265625,
      "learning_rate": 3.56738691465168e-05,
      "loss": 55.829,
      "step": 73000
    },
    {
      "epoch": 0.2949696384490762,
      "grad_norm": 931.0994873046875,
      "learning_rate": 3.5672134421912186e-05,
      "loss": 82.1505,
      "step": 73010
    },
    {
      "epoch": 0.29501003971444384,
      "grad_norm": 481.8676452636719,
      "learning_rate": 3.567039939177125e-05,
      "loss": 72.3018,
      "step": 73020
    },
    {
      "epoch": 0.2950504409798115,
      "grad_norm": 1044.1207275390625,
      "learning_rate": 3.566866405612781e-05,
      "loss": 63.6777,
      "step": 73030
    },
    {
      "epoch": 0.2950908422451791,
      "grad_norm": 945.3455200195312,
      "learning_rate": 3.56669284150157e-05,
      "loss": 65.7787,
      "step": 73040
    },
    {
      "epoch": 0.29513124351054676,
      "grad_norm": 391.0890197753906,
      "learning_rate": 3.566519246846875e-05,
      "loss": 93.8222,
      "step": 73050
    },
    {
      "epoch": 0.2951716447759144,
      "grad_norm": 319.558349609375,
      "learning_rate": 3.566345621652081e-05,
      "loss": 42.0089,
      "step": 73060
    },
    {
      "epoch": 0.29521204604128204,
      "grad_norm": 0.0,
      "learning_rate": 3.566171965920573e-05,
      "loss": 72.8675,
      "step": 73070
    },
    {
      "epoch": 0.2952524473066496,
      "grad_norm": 1246.403076171875,
      "learning_rate": 3.565998279655736e-05,
      "loss": 85.1066,
      "step": 73080
    },
    {
      "epoch": 0.29529284857201726,
      "grad_norm": 484.09002685546875,
      "learning_rate": 3.565824562860956e-05,
      "loss": 68.7531,
      "step": 73090
    },
    {
      "epoch": 0.2953332498373849,
      "grad_norm": 709.6483764648438,
      "learning_rate": 3.56565081553962e-05,
      "loss": 70.2082,
      "step": 73100
    },
    {
      "epoch": 0.29537365110275254,
      "grad_norm": 500.7955322265625,
      "learning_rate": 3.565477037695116e-05,
      "loss": 61.7862,
      "step": 73110
    },
    {
      "epoch": 0.2954140523681202,
      "grad_norm": 970.36572265625,
      "learning_rate": 3.56530322933083e-05,
      "loss": 64.6293,
      "step": 73120
    },
    {
      "epoch": 0.2954544536334878,
      "grad_norm": 875.14306640625,
      "learning_rate": 3.565129390450152e-05,
      "loss": 79.7622,
      "step": 73130
    },
    {
      "epoch": 0.2954948548988554,
      "grad_norm": 2398.982421875,
      "learning_rate": 3.564955521056471e-05,
      "loss": 75.0798,
      "step": 73140
    },
    {
      "epoch": 0.29553525616422305,
      "grad_norm": 559.6068115234375,
      "learning_rate": 3.5647816211531765e-05,
      "loss": 52.8689,
      "step": 73150
    },
    {
      "epoch": 0.2955756574295907,
      "grad_norm": 573.7216186523438,
      "learning_rate": 3.5646076907436586e-05,
      "loss": 73.3646,
      "step": 73160
    },
    {
      "epoch": 0.29561605869495833,
      "grad_norm": 761.6354370117188,
      "learning_rate": 3.5644337298313086e-05,
      "loss": 83.2933,
      "step": 73170
    },
    {
      "epoch": 0.29565645996032597,
      "grad_norm": 475.8253173828125,
      "learning_rate": 3.5642597384195166e-05,
      "loss": 110.5668,
      "step": 73180
    },
    {
      "epoch": 0.2956968612256936,
      "grad_norm": 823.4723510742188,
      "learning_rate": 3.564085716511677e-05,
      "loss": 52.4007,
      "step": 73190
    },
    {
      "epoch": 0.2957372624910612,
      "grad_norm": 2419.842041015625,
      "learning_rate": 3.5639116641111804e-05,
      "loss": 80.5333,
      "step": 73200
    },
    {
      "epoch": 0.29577766375642883,
      "grad_norm": 746.0689697265625,
      "learning_rate": 3.563737581221421e-05,
      "loss": 77.6496,
      "step": 73210
    },
    {
      "epoch": 0.2958180650217965,
      "grad_norm": 850.6876831054688,
      "learning_rate": 3.563563467845792e-05,
      "loss": 56.607,
      "step": 73220
    },
    {
      "epoch": 0.2958584662871641,
      "grad_norm": 854.9337158203125,
      "learning_rate": 3.563389323987688e-05,
      "loss": 106.5573,
      "step": 73230
    },
    {
      "epoch": 0.29589886755253175,
      "grad_norm": 1438.9986572265625,
      "learning_rate": 3.563215149650505e-05,
      "loss": 101.9418,
      "step": 73240
    },
    {
      "epoch": 0.2959392688178994,
      "grad_norm": 720.9024047851562,
      "learning_rate": 3.563040944837638e-05,
      "loss": 83.7277,
      "step": 73250
    },
    {
      "epoch": 0.29597967008326703,
      "grad_norm": 1647.9295654296875,
      "learning_rate": 3.562866709552483e-05,
      "loss": 56.4929,
      "step": 73260
    },
    {
      "epoch": 0.2960200713486346,
      "grad_norm": 614.6683349609375,
      "learning_rate": 3.562692443798436e-05,
      "loss": 53.4081,
      "step": 73270
    },
    {
      "epoch": 0.29606047261400226,
      "grad_norm": 1394.190185546875,
      "learning_rate": 3.562518147578896e-05,
      "loss": 84.1321,
      "step": 73280
    },
    {
      "epoch": 0.2961008738793699,
      "grad_norm": 955.5640869140625,
      "learning_rate": 3.56234382089726e-05,
      "loss": 70.5042,
      "step": 73290
    },
    {
      "epoch": 0.29614127514473754,
      "grad_norm": 0.0,
      "learning_rate": 3.5621694637569263e-05,
      "loss": 47.5671,
      "step": 73300
    },
    {
      "epoch": 0.2961816764101052,
      "grad_norm": 470.16400146484375,
      "learning_rate": 3.561995076161296e-05,
      "loss": 62.6391,
      "step": 73310
    },
    {
      "epoch": 0.2962220776754728,
      "grad_norm": 819.5801391601562,
      "learning_rate": 3.5618206581137664e-05,
      "loss": 45.0165,
      "step": 73320
    },
    {
      "epoch": 0.2962624789408404,
      "grad_norm": 590.9551391601562,
      "learning_rate": 3.5616462096177396e-05,
      "loss": 55.2823,
      "step": 73330
    },
    {
      "epoch": 0.29630288020620804,
      "grad_norm": 958.3380737304688,
      "learning_rate": 3.561471730676616e-05,
      "loss": 100.6481,
      "step": 73340
    },
    {
      "epoch": 0.2963432814715757,
      "grad_norm": 460.4815368652344,
      "learning_rate": 3.561297221293797e-05,
      "loss": 37.1382,
      "step": 73350
    },
    {
      "epoch": 0.2963836827369433,
      "grad_norm": 706.3472290039062,
      "learning_rate": 3.561122681472684e-05,
      "loss": 65.2587,
      "step": 73360
    },
    {
      "epoch": 0.29642408400231096,
      "grad_norm": 876.3798217773438,
      "learning_rate": 3.560948111216682e-05,
      "loss": 63.1272,
      "step": 73370
    },
    {
      "epoch": 0.2964644852676786,
      "grad_norm": 758.551025390625,
      "learning_rate": 3.560773510529192e-05,
      "loss": 101.7248,
      "step": 73380
    },
    {
      "epoch": 0.29650488653304624,
      "grad_norm": 1736.4576416015625,
      "learning_rate": 3.560598879413619e-05,
      "loss": 83.9242,
      "step": 73390
    },
    {
      "epoch": 0.2965452877984138,
      "grad_norm": 453.7065734863281,
      "learning_rate": 3.560424217873368e-05,
      "loss": 54.8189,
      "step": 73400
    },
    {
      "epoch": 0.29658568906378147,
      "grad_norm": 1004.9888305664062,
      "learning_rate": 3.560249525911842e-05,
      "loss": 65.9622,
      "step": 73410
    },
    {
      "epoch": 0.2966260903291491,
      "grad_norm": 704.4619140625,
      "learning_rate": 3.56007480353245e-05,
      "loss": 53.3604,
      "step": 73420
    },
    {
      "epoch": 0.29666649159451675,
      "grad_norm": 544.6813354492188,
      "learning_rate": 3.559900050738596e-05,
      "loss": 57.0338,
      "step": 73430
    },
    {
      "epoch": 0.2967068928598844,
      "grad_norm": 809.4789428710938,
      "learning_rate": 3.559725267533686e-05,
      "loss": 63.4728,
      "step": 73440
    },
    {
      "epoch": 0.296747294125252,
      "grad_norm": 1127.5126953125,
      "learning_rate": 3.559550453921131e-05,
      "loss": 60.555,
      "step": 73450
    },
    {
      "epoch": 0.2967876953906196,
      "grad_norm": 1059.709228515625,
      "learning_rate": 3.559375609904336e-05,
      "loss": 77.2083,
      "step": 73460
    },
    {
      "epoch": 0.29682809665598725,
      "grad_norm": 526.6636352539062,
      "learning_rate": 3.559200735486711e-05,
      "loss": 50.3187,
      "step": 73470
    },
    {
      "epoch": 0.2968684979213549,
      "grad_norm": 937.5421752929688,
      "learning_rate": 3.559025830671664e-05,
      "loss": 70.7462,
      "step": 73480
    },
    {
      "epoch": 0.29690889918672253,
      "grad_norm": 1234.875,
      "learning_rate": 3.558850895462607e-05,
      "loss": 78.2883,
      "step": 73490
    },
    {
      "epoch": 0.29694930045209017,
      "grad_norm": 407.6966857910156,
      "learning_rate": 3.5586759298629486e-05,
      "loss": 40.0754,
      "step": 73500
    },
    {
      "epoch": 0.2969897017174578,
      "grad_norm": 1737.535888671875,
      "learning_rate": 3.5585009338761005e-05,
      "loss": 64.8734,
      "step": 73510
    },
    {
      "epoch": 0.2970301029828254,
      "grad_norm": 335.2982482910156,
      "learning_rate": 3.5583259075054746e-05,
      "loss": 58.7745,
      "step": 73520
    },
    {
      "epoch": 0.29707050424819303,
      "grad_norm": 642.6002197265625,
      "learning_rate": 3.5581508507544825e-05,
      "loss": 99.0578,
      "step": 73530
    },
    {
      "epoch": 0.2971109055135607,
      "grad_norm": 107.51019287109375,
      "learning_rate": 3.5579757636265377e-05,
      "loss": 55.8075,
      "step": 73540
    },
    {
      "epoch": 0.2971513067789283,
      "grad_norm": 650.3108520507812,
      "learning_rate": 3.557800646125053e-05,
      "loss": 83.543,
      "step": 73550
    },
    {
      "epoch": 0.29719170804429595,
      "grad_norm": 650.6173706054688,
      "learning_rate": 3.557625498253443e-05,
      "loss": 61.2145,
      "step": 73560
    },
    {
      "epoch": 0.2972321093096636,
      "grad_norm": 1853.13037109375,
      "learning_rate": 3.5574503200151213e-05,
      "loss": 101.2655,
      "step": 73570
    },
    {
      "epoch": 0.29727251057503123,
      "grad_norm": 846.0449829101562,
      "learning_rate": 3.557275111413505e-05,
      "loss": 104.6277,
      "step": 73580
    },
    {
      "epoch": 0.2973129118403988,
      "grad_norm": 472.6047058105469,
      "learning_rate": 3.557099872452008e-05,
      "loss": 83.1568,
      "step": 73590
    },
    {
      "epoch": 0.29735331310576646,
      "grad_norm": 655.8888549804688,
      "learning_rate": 3.5569246031340474e-05,
      "loss": 81.285,
      "step": 73600
    },
    {
      "epoch": 0.2973937143711341,
      "grad_norm": 833.2941284179688,
      "learning_rate": 3.5567493034630395e-05,
      "loss": 88.6944,
      "step": 73610
    },
    {
      "epoch": 0.29743411563650174,
      "grad_norm": 969.6983642578125,
      "learning_rate": 3.5565739734424034e-05,
      "loss": 67.491,
      "step": 73620
    },
    {
      "epoch": 0.2974745169018694,
      "grad_norm": 482.8932189941406,
      "learning_rate": 3.5563986130755557e-05,
      "loss": 52.3097,
      "step": 73630
    },
    {
      "epoch": 0.297514918167237,
      "grad_norm": 594.0313110351562,
      "learning_rate": 3.556223222365916e-05,
      "loss": 60.7423,
      "step": 73640
    },
    {
      "epoch": 0.2975553194326046,
      "grad_norm": 1269.3704833984375,
      "learning_rate": 3.556047801316903e-05,
      "loss": 86.5832,
      "step": 73650
    },
    {
      "epoch": 0.29759572069797224,
      "grad_norm": 527.30615234375,
      "learning_rate": 3.555872349931938e-05,
      "loss": 80.0871,
      "step": 73660
    },
    {
      "epoch": 0.2976361219633399,
      "grad_norm": 1384.492431640625,
      "learning_rate": 3.5556968682144395e-05,
      "loss": 56.7632,
      "step": 73670
    },
    {
      "epoch": 0.2976765232287075,
      "grad_norm": 686.9051513671875,
      "learning_rate": 3.5555213561678305e-05,
      "loss": 59.716,
      "step": 73680
    },
    {
      "epoch": 0.29771692449407516,
      "grad_norm": 965.8775634765625,
      "learning_rate": 3.555345813795531e-05,
      "loss": 86.5317,
      "step": 73690
    },
    {
      "epoch": 0.2977573257594428,
      "grad_norm": 502.6318054199219,
      "learning_rate": 3.5551702411009645e-05,
      "loss": 40.2029,
      "step": 73700
    },
    {
      "epoch": 0.29779772702481044,
      "grad_norm": 1723.965576171875,
      "learning_rate": 3.5549946380875536e-05,
      "loss": 66.3792,
      "step": 73710
    },
    {
      "epoch": 0.29783812829017803,
      "grad_norm": 859.0875854492188,
      "learning_rate": 3.554819004758721e-05,
      "loss": 62.8533,
      "step": 73720
    },
    {
      "epoch": 0.29787852955554567,
      "grad_norm": 994.7597045898438,
      "learning_rate": 3.554643341117892e-05,
      "loss": 69.4177,
      "step": 73730
    },
    {
      "epoch": 0.2979189308209133,
      "grad_norm": 1833.2977294921875,
      "learning_rate": 3.5544676471684906e-05,
      "loss": 108.2813,
      "step": 73740
    },
    {
      "epoch": 0.29795933208628095,
      "grad_norm": 513.6215209960938,
      "learning_rate": 3.554291922913942e-05,
      "loss": 64.1931,
      "step": 73750
    },
    {
      "epoch": 0.2979997333516486,
      "grad_norm": 458.6168212890625,
      "learning_rate": 3.554116168357673e-05,
      "loss": 49.7643,
      "step": 73760
    },
    {
      "epoch": 0.29804013461701623,
      "grad_norm": 634.9912109375,
      "learning_rate": 3.5539403835031075e-05,
      "loss": 47.9218,
      "step": 73770
    },
    {
      "epoch": 0.2980805358823838,
      "grad_norm": 1256.944091796875,
      "learning_rate": 3.553764568353676e-05,
      "loss": 66.0002,
      "step": 73780
    },
    {
      "epoch": 0.29812093714775145,
      "grad_norm": 776.3186645507812,
      "learning_rate": 3.553588722912803e-05,
      "loss": 76.021,
      "step": 73790
    },
    {
      "epoch": 0.2981613384131191,
      "grad_norm": 755.112548828125,
      "learning_rate": 3.553412847183919e-05,
      "loss": 75.2325,
      "step": 73800
    },
    {
      "epoch": 0.29820173967848673,
      "grad_norm": 896.2506103515625,
      "learning_rate": 3.5532369411704505e-05,
      "loss": 55.7502,
      "step": 73810
    },
    {
      "epoch": 0.2982421409438544,
      "grad_norm": 754.2654418945312,
      "learning_rate": 3.5530610048758295e-05,
      "loss": 71.5779,
      "step": 73820
    },
    {
      "epoch": 0.298282542209222,
      "grad_norm": 678.224365234375,
      "learning_rate": 3.552885038303484e-05,
      "loss": 45.6851,
      "step": 73830
    },
    {
      "epoch": 0.2983229434745896,
      "grad_norm": 925.0071411132812,
      "learning_rate": 3.552709041456845e-05,
      "loss": 54.6353,
      "step": 73840
    },
    {
      "epoch": 0.29836334473995724,
      "grad_norm": 1991.972412109375,
      "learning_rate": 3.552533014339344e-05,
      "loss": 86.6209,
      "step": 73850
    },
    {
      "epoch": 0.2984037460053249,
      "grad_norm": 2509.668212890625,
      "learning_rate": 3.552356956954413e-05,
      "loss": 122.7409,
      "step": 73860
    },
    {
      "epoch": 0.2984441472706925,
      "grad_norm": 898.805908203125,
      "learning_rate": 3.552180869305483e-05,
      "loss": 57.9041,
      "step": 73870
    },
    {
      "epoch": 0.29848454853606016,
      "grad_norm": 789.3969116210938,
      "learning_rate": 3.552004751395989e-05,
      "loss": 55.5541,
      "step": 73880
    },
    {
      "epoch": 0.2985249498014278,
      "grad_norm": 1287.8016357421875,
      "learning_rate": 3.551828603229363e-05,
      "loss": 67.9635,
      "step": 73890
    },
    {
      "epoch": 0.29856535106679544,
      "grad_norm": 2758.351806640625,
      "learning_rate": 3.551652424809039e-05,
      "loss": 65.6716,
      "step": 73900
    },
    {
      "epoch": 0.298605752332163,
      "grad_norm": 676.35595703125,
      "learning_rate": 3.551476216138453e-05,
      "loss": 53.9801,
      "step": 73910
    },
    {
      "epoch": 0.29864615359753066,
      "grad_norm": 1682.7109375,
      "learning_rate": 3.551299977221038e-05,
      "loss": 105.8357,
      "step": 73920
    },
    {
      "epoch": 0.2986865548628983,
      "grad_norm": 485.17279052734375,
      "learning_rate": 3.551123708060233e-05,
      "loss": 46.7262,
      "step": 73930
    },
    {
      "epoch": 0.29872695612826594,
      "grad_norm": 1494.2843017578125,
      "learning_rate": 3.550947408659471e-05,
      "loss": 113.5479,
      "step": 73940
    },
    {
      "epoch": 0.2987673573936336,
      "grad_norm": 495.2820129394531,
      "learning_rate": 3.550771079022192e-05,
      "loss": 50.9028,
      "step": 73950
    },
    {
      "epoch": 0.2988077586590012,
      "grad_norm": 1544.240234375,
      "learning_rate": 3.550594719151832e-05,
      "loss": 56.2776,
      "step": 73960
    },
    {
      "epoch": 0.2988481599243688,
      "grad_norm": 451.206787109375,
      "learning_rate": 3.550418329051829e-05,
      "loss": 71.9325,
      "step": 73970
    },
    {
      "epoch": 0.29888856118973645,
      "grad_norm": 442.68548583984375,
      "learning_rate": 3.550241908725624e-05,
      "loss": 63.0581,
      "step": 73980
    },
    {
      "epoch": 0.2989289624551041,
      "grad_norm": 519.7935180664062,
      "learning_rate": 3.550065458176653e-05,
      "loss": 51.5123,
      "step": 73990
    },
    {
      "epoch": 0.2989693637204717,
      "grad_norm": 497.5556945800781,
      "learning_rate": 3.549888977408359e-05,
      "loss": 61.3014,
      "step": 74000
    },
    {
      "epoch": 0.29900976498583937,
      "grad_norm": 700.3739013671875,
      "learning_rate": 3.5497124664241816e-05,
      "loss": 81.6645,
      "step": 74010
    },
    {
      "epoch": 0.299050166251207,
      "grad_norm": 497.64599609375,
      "learning_rate": 3.549535925227562e-05,
      "loss": 48.5412,
      "step": 74020
    },
    {
      "epoch": 0.29909056751657465,
      "grad_norm": 499.0395812988281,
      "learning_rate": 3.549359353821941e-05,
      "loss": 59.0803,
      "step": 74030
    },
    {
      "epoch": 0.29913096878194223,
      "grad_norm": 1075.3427734375,
      "learning_rate": 3.5491827522107624e-05,
      "loss": 67.4793,
      "step": 74040
    },
    {
      "epoch": 0.29917137004730987,
      "grad_norm": 627.8690185546875,
      "learning_rate": 3.5490061203974676e-05,
      "loss": 69.5395,
      "step": 74050
    },
    {
      "epoch": 0.2992117713126775,
      "grad_norm": 1320.747314453125,
      "learning_rate": 3.548829458385502e-05,
      "loss": 81.1121,
      "step": 74060
    },
    {
      "epoch": 0.29925217257804515,
      "grad_norm": 751.7114868164062,
      "learning_rate": 3.548652766178308e-05,
      "loss": 49.1731,
      "step": 74070
    },
    {
      "epoch": 0.2992925738434128,
      "grad_norm": 974.225830078125,
      "learning_rate": 3.5484760437793316e-05,
      "loss": 79.8373,
      "step": 74080
    },
    {
      "epoch": 0.29933297510878043,
      "grad_norm": 2224.293701171875,
      "learning_rate": 3.5482992911920174e-05,
      "loss": 65.8389,
      "step": 74090
    },
    {
      "epoch": 0.299373376374148,
      "grad_norm": 1174.2479248046875,
      "learning_rate": 3.548122508419811e-05,
      "loss": 108.1642,
      "step": 74100
    },
    {
      "epoch": 0.29941377763951565,
      "grad_norm": 264.214599609375,
      "learning_rate": 3.547945695466159e-05,
      "loss": 74.0446,
      "step": 74110
    },
    {
      "epoch": 0.2994541789048833,
      "grad_norm": 1260.7275390625,
      "learning_rate": 3.5477688523345095e-05,
      "loss": 120.0823,
      "step": 74120
    },
    {
      "epoch": 0.29949458017025093,
      "grad_norm": 886.28515625,
      "learning_rate": 3.547591979028309e-05,
      "loss": 79.4073,
      "step": 74130
    },
    {
      "epoch": 0.2995349814356186,
      "grad_norm": 1063.769775390625,
      "learning_rate": 3.5474150755510065e-05,
      "loss": 60.8101,
      "step": 74140
    },
    {
      "epoch": 0.2995753827009862,
      "grad_norm": 761.662109375,
      "learning_rate": 3.54723814190605e-05,
      "loss": 67.1386,
      "step": 74150
    },
    {
      "epoch": 0.2996157839663538,
      "grad_norm": 901.34521484375,
      "learning_rate": 3.54706117809689e-05,
      "loss": 73.7574,
      "step": 74160
    },
    {
      "epoch": 0.29965618523172144,
      "grad_norm": 1164.415771484375,
      "learning_rate": 3.546884184126975e-05,
      "loss": 84.1045,
      "step": 74170
    },
    {
      "epoch": 0.2996965864970891,
      "grad_norm": 808.2579956054688,
      "learning_rate": 3.546707159999756e-05,
      "loss": 77.337,
      "step": 74180
    },
    {
      "epoch": 0.2997369877624567,
      "grad_norm": 238.32205200195312,
      "learning_rate": 3.5465301057186864e-05,
      "loss": 38.9205,
      "step": 74190
    },
    {
      "epoch": 0.29977738902782436,
      "grad_norm": 651.8353881835938,
      "learning_rate": 3.5463530212872145e-05,
      "loss": 55.3188,
      "step": 74200
    },
    {
      "epoch": 0.299817790293192,
      "grad_norm": 323.20458984375,
      "learning_rate": 3.546175906708795e-05,
      "loss": 42.7316,
      "step": 74210
    },
    {
      "epoch": 0.29985819155855964,
      "grad_norm": 469.6799011230469,
      "learning_rate": 3.545998761986881e-05,
      "loss": 59.3492,
      "step": 74220
    },
    {
      "epoch": 0.2998985928239272,
      "grad_norm": 2659.387451171875,
      "learning_rate": 3.545821587124924e-05,
      "loss": 56.3482,
      "step": 74230
    },
    {
      "epoch": 0.29993899408929486,
      "grad_norm": 893.9676513671875,
      "learning_rate": 3.54564438212638e-05,
      "loss": 67.2324,
      "step": 74240
    },
    {
      "epoch": 0.2999793953546625,
      "grad_norm": 499.2391662597656,
      "learning_rate": 3.5454671469947024e-05,
      "loss": 50.6774,
      "step": 74250
    },
    {
      "epoch": 0.30001979662003014,
      "grad_norm": 738.5967407226562,
      "learning_rate": 3.5452898817333474e-05,
      "loss": 82.5717,
      "step": 74260
    },
    {
      "epoch": 0.3000601978853978,
      "grad_norm": 384.5932922363281,
      "learning_rate": 3.545112586345771e-05,
      "loss": 54.1004,
      "step": 74270
    },
    {
      "epoch": 0.3001005991507654,
      "grad_norm": 1085.3819580078125,
      "learning_rate": 3.544935260835429e-05,
      "loss": 77.678,
      "step": 74280
    },
    {
      "epoch": 0.300141000416133,
      "grad_norm": 436.23541259765625,
      "learning_rate": 3.5447579052057776e-05,
      "loss": 60.4012,
      "step": 74290
    },
    {
      "epoch": 0.30018140168150065,
      "grad_norm": 505.67169189453125,
      "learning_rate": 3.544580519460277e-05,
      "loss": 70.9453,
      "step": 74300
    },
    {
      "epoch": 0.3002218029468683,
      "grad_norm": 1916.5804443359375,
      "learning_rate": 3.5444031036023837e-05,
      "loss": 81.2167,
      "step": 74310
    },
    {
      "epoch": 0.30026220421223593,
      "grad_norm": 835.12158203125,
      "learning_rate": 3.5442256576355564e-05,
      "loss": 80.3109,
      "step": 74320
    },
    {
      "epoch": 0.30030260547760357,
      "grad_norm": 746.8428344726562,
      "learning_rate": 3.544048181563255e-05,
      "loss": 123.7258,
      "step": 74330
    },
    {
      "epoch": 0.3003430067429712,
      "grad_norm": 900.0025024414062,
      "learning_rate": 3.5438706753889396e-05,
      "loss": 80.3383,
      "step": 74340
    },
    {
      "epoch": 0.30038340800833885,
      "grad_norm": 634.4719848632812,
      "learning_rate": 3.54369313911607e-05,
      "loss": 80.1079,
      "step": 74350
    },
    {
      "epoch": 0.30042380927370643,
      "grad_norm": 1878.78466796875,
      "learning_rate": 3.543515572748108e-05,
      "loss": 66.6467,
      "step": 74360
    },
    {
      "epoch": 0.3004642105390741,
      "grad_norm": 749.2390747070312,
      "learning_rate": 3.5433379762885165e-05,
      "loss": 82.4085,
      "step": 74370
    },
    {
      "epoch": 0.3005046118044417,
      "grad_norm": 3661.9140625,
      "learning_rate": 3.543160349740755e-05,
      "loss": 69.3461,
      "step": 74380
    },
    {
      "epoch": 0.30054501306980935,
      "grad_norm": 276.2242736816406,
      "learning_rate": 3.542982693108289e-05,
      "loss": 74.4839,
      "step": 74390
    },
    {
      "epoch": 0.300585414335177,
      "grad_norm": 2024.406005859375,
      "learning_rate": 3.542805006394581e-05,
      "loss": 75.6991,
      "step": 74400
    },
    {
      "epoch": 0.30062581560054463,
      "grad_norm": 1021.9595336914062,
      "learning_rate": 3.5426272896030944e-05,
      "loss": 77.9724,
      "step": 74410
    },
    {
      "epoch": 0.3006662168659122,
      "grad_norm": 353.10577392578125,
      "learning_rate": 3.5424495427372946e-05,
      "loss": 74.5174,
      "step": 74420
    },
    {
      "epoch": 0.30070661813127986,
      "grad_norm": 814.4290161132812,
      "learning_rate": 3.5422717658006475e-05,
      "loss": 77.0783,
      "step": 74430
    },
    {
      "epoch": 0.3007470193966475,
      "grad_norm": 1428.374267578125,
      "learning_rate": 3.542093958796618e-05,
      "loss": 71.8454,
      "step": 74440
    },
    {
      "epoch": 0.30078742066201514,
      "grad_norm": 358.1316833496094,
      "learning_rate": 3.541916121728673e-05,
      "loss": 41.539,
      "step": 74450
    },
    {
      "epoch": 0.3008278219273828,
      "grad_norm": 610.126220703125,
      "learning_rate": 3.541738254600279e-05,
      "loss": 58.5676,
      "step": 74460
    },
    {
      "epoch": 0.3008682231927504,
      "grad_norm": 570.5245971679688,
      "learning_rate": 3.541560357414904e-05,
      "loss": 58.6159,
      "step": 74470
    },
    {
      "epoch": 0.300908624458118,
      "grad_norm": 795.262939453125,
      "learning_rate": 3.5413824301760165e-05,
      "loss": 82.1075,
      "step": 74480
    },
    {
      "epoch": 0.30094902572348564,
      "grad_norm": 949.2442016601562,
      "learning_rate": 3.541204472887085e-05,
      "loss": 64.7017,
      "step": 74490
    },
    {
      "epoch": 0.3009894269888533,
      "grad_norm": 506.1993713378906,
      "learning_rate": 3.541026485551579e-05,
      "loss": 52.474,
      "step": 74500
    },
    {
      "epoch": 0.3010298282542209,
      "grad_norm": 966.4857788085938,
      "learning_rate": 3.540848468172968e-05,
      "loss": 90.3593,
      "step": 74510
    },
    {
      "epoch": 0.30107022951958856,
      "grad_norm": 864.8375854492188,
      "learning_rate": 3.540670420754722e-05,
      "loss": 69.907,
      "step": 74520
    },
    {
      "epoch": 0.3011106307849562,
      "grad_norm": 426.7903137207031,
      "learning_rate": 3.540492343300314e-05,
      "loss": 96.3889,
      "step": 74530
    },
    {
      "epoch": 0.30115103205032384,
      "grad_norm": 573.3229370117188,
      "learning_rate": 3.540314235813215e-05,
      "loss": 66.8149,
      "step": 74540
    },
    {
      "epoch": 0.3011914333156914,
      "grad_norm": 583.5394897460938,
      "learning_rate": 3.540136098296896e-05,
      "loss": 66.5317,
      "step": 74550
    },
    {
      "epoch": 0.30123183458105907,
      "grad_norm": 1171.0955810546875,
      "learning_rate": 3.5399579307548314e-05,
      "loss": 74.4664,
      "step": 74560
    },
    {
      "epoch": 0.3012722358464267,
      "grad_norm": 771.04345703125,
      "learning_rate": 3.539779733190494e-05,
      "loss": 59.7181,
      "step": 74570
    },
    {
      "epoch": 0.30131263711179435,
      "grad_norm": 1087.257080078125,
      "learning_rate": 3.539601505607358e-05,
      "loss": 61.8407,
      "step": 74580
    },
    {
      "epoch": 0.301353038377162,
      "grad_norm": 571.8591918945312,
      "learning_rate": 3.5394232480088986e-05,
      "loss": 61.0519,
      "step": 74590
    },
    {
      "epoch": 0.3013934396425296,
      "grad_norm": 1148.0616455078125,
      "learning_rate": 3.5392449603985894e-05,
      "loss": 84.9272,
      "step": 74600
    },
    {
      "epoch": 0.3014338409078972,
      "grad_norm": 669.8700561523438,
      "learning_rate": 3.539066642779907e-05,
      "loss": 72.5787,
      "step": 74610
    },
    {
      "epoch": 0.30147424217326485,
      "grad_norm": 533.347900390625,
      "learning_rate": 3.538888295156329e-05,
      "loss": 61.7806,
      "step": 74620
    },
    {
      "epoch": 0.3015146434386325,
      "grad_norm": 1645.3468017578125,
      "learning_rate": 3.538709917531331e-05,
      "loss": 59.6045,
      "step": 74630
    },
    {
      "epoch": 0.30155504470400013,
      "grad_norm": 1048.553955078125,
      "learning_rate": 3.538531509908391e-05,
      "loss": 67.4957,
      "step": 74640
    },
    {
      "epoch": 0.30159544596936777,
      "grad_norm": 629.9532470703125,
      "learning_rate": 3.538353072290988e-05,
      "loss": 75.4748,
      "step": 74650
    },
    {
      "epoch": 0.3016358472347354,
      "grad_norm": 561.3018798828125,
      "learning_rate": 3.538174604682599e-05,
      "loss": 63.2835,
      "step": 74660
    },
    {
      "epoch": 0.30167624850010305,
      "grad_norm": 690.0833129882812,
      "learning_rate": 3.537996107086704e-05,
      "loss": 86.1682,
      "step": 74670
    },
    {
      "epoch": 0.30171664976547063,
      "grad_norm": 634.8307495117188,
      "learning_rate": 3.537817579506783e-05,
      "loss": 65.4851,
      "step": 74680
    },
    {
      "epoch": 0.3017570510308383,
      "grad_norm": 1294.4716796875,
      "learning_rate": 3.537639021946317e-05,
      "loss": 77.3135,
      "step": 74690
    },
    {
      "epoch": 0.3017974522962059,
      "grad_norm": 1165.074462890625,
      "learning_rate": 3.5374604344087866e-05,
      "loss": 58.4666,
      "step": 74700
    },
    {
      "epoch": 0.30183785356157355,
      "grad_norm": 833.4439086914062,
      "learning_rate": 3.5372818168976734e-05,
      "loss": 47.627,
      "step": 74710
    },
    {
      "epoch": 0.3018782548269412,
      "grad_norm": 1249.32763671875,
      "learning_rate": 3.53710316941646e-05,
      "loss": 62.6458,
      "step": 74720
    },
    {
      "epoch": 0.30191865609230883,
      "grad_norm": 466.3185119628906,
      "learning_rate": 3.5369244919686284e-05,
      "loss": 62.2887,
      "step": 74730
    },
    {
      "epoch": 0.3019590573576764,
      "grad_norm": 1040.11669921875,
      "learning_rate": 3.536745784557663e-05,
      "loss": 66.7786,
      "step": 74740
    },
    {
      "epoch": 0.30199945862304406,
      "grad_norm": 980.9957885742188,
      "learning_rate": 3.536567047187047e-05,
      "loss": 68.8723,
      "step": 74750
    },
    {
      "epoch": 0.3020398598884117,
      "grad_norm": 927.2369384765625,
      "learning_rate": 3.536388279860266e-05,
      "loss": 69.8665,
      "step": 74760
    },
    {
      "epoch": 0.30208026115377934,
      "grad_norm": 902.037353515625,
      "learning_rate": 3.536209482580804e-05,
      "loss": 92.5639,
      "step": 74770
    },
    {
      "epoch": 0.302120662419147,
      "grad_norm": 831.7162475585938,
      "learning_rate": 3.536030655352147e-05,
      "loss": 67.2394,
      "step": 74780
    },
    {
      "epoch": 0.3021610636845146,
      "grad_norm": 381.0028381347656,
      "learning_rate": 3.535851798177782e-05,
      "loss": 58.4229,
      "step": 74790
    },
    {
      "epoch": 0.3022014649498822,
      "grad_norm": 1212.741455078125,
      "learning_rate": 3.535672911061196e-05,
      "loss": 55.6143,
      "step": 74800
    },
    {
      "epoch": 0.30224186621524984,
      "grad_norm": 400.9530944824219,
      "learning_rate": 3.535493994005874e-05,
      "loss": 55.2473,
      "step": 74810
    },
    {
      "epoch": 0.3022822674806175,
      "grad_norm": 631.3129272460938,
      "learning_rate": 3.535315047015308e-05,
      "loss": 49.2248,
      "step": 74820
    },
    {
      "epoch": 0.3023226687459851,
      "grad_norm": 1094.7774658203125,
      "learning_rate": 3.535136070092984e-05,
      "loss": 80.757,
      "step": 74830
    },
    {
      "epoch": 0.30236307001135276,
      "grad_norm": 753.2235107421875,
      "learning_rate": 3.5349570632423925e-05,
      "loss": 80.3722,
      "step": 74840
    },
    {
      "epoch": 0.3024034712767204,
      "grad_norm": 490.4667053222656,
      "learning_rate": 3.534778026467022e-05,
      "loss": 67.9342,
      "step": 74850
    },
    {
      "epoch": 0.30244387254208804,
      "grad_norm": 1227.3990478515625,
      "learning_rate": 3.534598959770364e-05,
      "loss": 112.9435,
      "step": 74860
    },
    {
      "epoch": 0.30248427380745563,
      "grad_norm": 414.38543701171875,
      "learning_rate": 3.5344198631559096e-05,
      "loss": 60.2385,
      "step": 74870
    },
    {
      "epoch": 0.30252467507282327,
      "grad_norm": 1657.522216796875,
      "learning_rate": 3.5342407366271495e-05,
      "loss": 54.7299,
      "step": 74880
    },
    {
      "epoch": 0.3025650763381909,
      "grad_norm": 1603.0582275390625,
      "learning_rate": 3.534061580187577e-05,
      "loss": 79.5543,
      "step": 74890
    },
    {
      "epoch": 0.30260547760355855,
      "grad_norm": 674.0008544921875,
      "learning_rate": 3.5338823938406834e-05,
      "loss": 56.2018,
      "step": 74900
    },
    {
      "epoch": 0.3026458788689262,
      "grad_norm": 1061.078125,
      "learning_rate": 3.533703177589964e-05,
      "loss": 65.559,
      "step": 74910
    },
    {
      "epoch": 0.30268628013429383,
      "grad_norm": 466.67864990234375,
      "learning_rate": 3.53352393143891e-05,
      "loss": 54.7194,
      "step": 74920
    },
    {
      "epoch": 0.3027266813996614,
      "grad_norm": 419.76861572265625,
      "learning_rate": 3.5333446553910184e-05,
      "loss": 71.4791,
      "step": 74930
    },
    {
      "epoch": 0.30276708266502905,
      "grad_norm": 1148.4244384765625,
      "learning_rate": 3.533165349449783e-05,
      "loss": 107.1067,
      "step": 74940
    },
    {
      "epoch": 0.3028074839303967,
      "grad_norm": 1184.2965087890625,
      "learning_rate": 3.5329860136187e-05,
      "loss": 63.7784,
      "step": 74950
    },
    {
      "epoch": 0.30284788519576433,
      "grad_norm": 552.2979125976562,
      "learning_rate": 3.5328066479012655e-05,
      "loss": 69.7092,
      "step": 74960
    },
    {
      "epoch": 0.302888286461132,
      "grad_norm": 405.5337829589844,
      "learning_rate": 3.5326272523009754e-05,
      "loss": 60.7039,
      "step": 74970
    },
    {
      "epoch": 0.3029286877264996,
      "grad_norm": 963.63330078125,
      "learning_rate": 3.532447826821329e-05,
      "loss": 44.4223,
      "step": 74980
    },
    {
      "epoch": 0.30296908899186725,
      "grad_norm": 574.0205078125,
      "learning_rate": 3.532268371465823e-05,
      "loss": 66.2384,
      "step": 74990
    },
    {
      "epoch": 0.30300949025723484,
      "grad_norm": 607.9288940429688,
      "learning_rate": 3.532088886237956e-05,
      "loss": 53.3745,
      "step": 75000
    },
    {
      "epoch": 0.3030498915226025,
      "grad_norm": 1956.13623046875,
      "learning_rate": 3.531909371141228e-05,
      "loss": 109.1442,
      "step": 75010
    },
    {
      "epoch": 0.3030902927879701,
      "grad_norm": 1171.8389892578125,
      "learning_rate": 3.531729826179138e-05,
      "loss": 62.4617,
      "step": 75020
    },
    {
      "epoch": 0.30313069405333776,
      "grad_norm": 1014.7931518554688,
      "learning_rate": 3.531550251355186e-05,
      "loss": 58.9204,
      "step": 75030
    },
    {
      "epoch": 0.3031710953187054,
      "grad_norm": 1173.664306640625,
      "learning_rate": 3.531370646672874e-05,
      "loss": 59.5735,
      "step": 75040
    },
    {
      "epoch": 0.30321149658407304,
      "grad_norm": 913.2959594726562,
      "learning_rate": 3.5311910121357016e-05,
      "loss": 62.6237,
      "step": 75050
    },
    {
      "epoch": 0.3032518978494406,
      "grad_norm": 637.90185546875,
      "learning_rate": 3.531011347747173e-05,
      "loss": 68.8141,
      "step": 75060
    },
    {
      "epoch": 0.30329229911480826,
      "grad_norm": 1701.0555419921875,
      "learning_rate": 3.53083165351079e-05,
      "loss": 122.4263,
      "step": 75070
    },
    {
      "epoch": 0.3033327003801759,
      "grad_norm": 656.2587280273438,
      "learning_rate": 3.530651929430055e-05,
      "loss": 64.5342,
      "step": 75080
    },
    {
      "epoch": 0.30337310164554354,
      "grad_norm": 884.7515869140625,
      "learning_rate": 3.5304721755084734e-05,
      "loss": 45.98,
      "step": 75090
    },
    {
      "epoch": 0.3034135029109112,
      "grad_norm": 628.9338989257812,
      "learning_rate": 3.530292391749549e-05,
      "loss": 63.3537,
      "step": 75100
    },
    {
      "epoch": 0.3034539041762788,
      "grad_norm": 512.0625,
      "learning_rate": 3.530112578156786e-05,
      "loss": 79.4432,
      "step": 75110
    },
    {
      "epoch": 0.3034943054416464,
      "grad_norm": 695.7696533203125,
      "learning_rate": 3.529932734733691e-05,
      "loss": 82.5149,
      "step": 75120
    },
    {
      "epoch": 0.30353470670701405,
      "grad_norm": 704.0012817382812,
      "learning_rate": 3.529752861483769e-05,
      "loss": 77.3432,
      "step": 75130
    },
    {
      "epoch": 0.3035751079723817,
      "grad_norm": 930.8369750976562,
      "learning_rate": 3.529572958410528e-05,
      "loss": 43.9781,
      "step": 75140
    },
    {
      "epoch": 0.3036155092377493,
      "grad_norm": 1269.88720703125,
      "learning_rate": 3.529393025517475e-05,
      "loss": 72.7654,
      "step": 75150
    },
    {
      "epoch": 0.30365591050311697,
      "grad_norm": 739.17578125,
      "learning_rate": 3.529213062808116e-05,
      "loss": 43.866,
      "step": 75160
    },
    {
      "epoch": 0.3036963117684846,
      "grad_norm": 1014.5413818359375,
      "learning_rate": 3.5290330702859624e-05,
      "loss": 64.9485,
      "step": 75170
    },
    {
      "epoch": 0.30373671303385225,
      "grad_norm": 440.79931640625,
      "learning_rate": 3.528853047954521e-05,
      "loss": 65.0976,
      "step": 75180
    },
    {
      "epoch": 0.30377711429921983,
      "grad_norm": 861.1629028320312,
      "learning_rate": 3.5286729958173036e-05,
      "loss": 91.4645,
      "step": 75190
    },
    {
      "epoch": 0.30381751556458747,
      "grad_norm": 574.344482421875,
      "learning_rate": 3.528492913877818e-05,
      "loss": 65.2671,
      "step": 75200
    },
    {
      "epoch": 0.3038579168299551,
      "grad_norm": 993.7310791015625,
      "learning_rate": 3.528312802139577e-05,
      "loss": 73.2417,
      "step": 75210
    },
    {
      "epoch": 0.30389831809532275,
      "grad_norm": 735.0557250976562,
      "learning_rate": 3.5281326606060905e-05,
      "loss": 73.2255,
      "step": 75220
    },
    {
      "epoch": 0.3039387193606904,
      "grad_norm": 0.0,
      "learning_rate": 3.5279524892808714e-05,
      "loss": 49.4044,
      "step": 75230
    },
    {
      "epoch": 0.30397912062605803,
      "grad_norm": 721.876220703125,
      "learning_rate": 3.5277722881674314e-05,
      "loss": 71.4891,
      "step": 75240
    },
    {
      "epoch": 0.3040195218914256,
      "grad_norm": 816.6832885742188,
      "learning_rate": 3.527592057269285e-05,
      "loss": 44.2561,
      "step": 75250
    },
    {
      "epoch": 0.30405992315679325,
      "grad_norm": 619.421630859375,
      "learning_rate": 3.527411796589944e-05,
      "loss": 64.0264,
      "step": 75260
    },
    {
      "epoch": 0.3041003244221609,
      "grad_norm": 0.0,
      "learning_rate": 3.5272315061329236e-05,
      "loss": 52.1535,
      "step": 75270
    },
    {
      "epoch": 0.30414072568752853,
      "grad_norm": 571.986572265625,
      "learning_rate": 3.52705118590174e-05,
      "loss": 50.209,
      "step": 75280
    },
    {
      "epoch": 0.3041811269528962,
      "grad_norm": 402.1851501464844,
      "learning_rate": 3.5268708358999064e-05,
      "loss": 62.3056,
      "step": 75290
    },
    {
      "epoch": 0.3042215282182638,
      "grad_norm": 503.4267272949219,
      "learning_rate": 3.52669045613094e-05,
      "loss": 44.4806,
      "step": 75300
    },
    {
      "epoch": 0.30426192948363145,
      "grad_norm": 685.234130859375,
      "learning_rate": 3.5265100465983564e-05,
      "loss": 63.299,
      "step": 75310
    },
    {
      "epoch": 0.30430233074899904,
      "grad_norm": 2023.0379638671875,
      "learning_rate": 3.526329607305675e-05,
      "loss": 57.7,
      "step": 75320
    },
    {
      "epoch": 0.3043427320143667,
      "grad_norm": 1253.1976318359375,
      "learning_rate": 3.526149138256411e-05,
      "loss": 61.8799,
      "step": 75330
    },
    {
      "epoch": 0.3043831332797343,
      "grad_norm": 493.9773864746094,
      "learning_rate": 3.525968639454084e-05,
      "loss": 64.0784,
      "step": 75340
    },
    {
      "epoch": 0.30442353454510196,
      "grad_norm": 1406.955322265625,
      "learning_rate": 3.525788110902213e-05,
      "loss": 77.3711,
      "step": 75350
    },
    {
      "epoch": 0.3044639358104696,
      "grad_norm": 1954.9525146484375,
      "learning_rate": 3.525607552604317e-05,
      "loss": 72.2056,
      "step": 75360
    },
    {
      "epoch": 0.30450433707583724,
      "grad_norm": 2237.93115234375,
      "learning_rate": 3.525426964563916e-05,
      "loss": 78.1073,
      "step": 75370
    },
    {
      "epoch": 0.3045447383412048,
      "grad_norm": 966.1798095703125,
      "learning_rate": 3.525246346784532e-05,
      "loss": 79.0288,
      "step": 75380
    },
    {
      "epoch": 0.30458513960657246,
      "grad_norm": 691.7649536132812,
      "learning_rate": 3.525065699269684e-05,
      "loss": 77.6586,
      "step": 75390
    },
    {
      "epoch": 0.3046255408719401,
      "grad_norm": 638.4725952148438,
      "learning_rate": 3.524885022022896e-05,
      "loss": 67.9005,
      "step": 75400
    },
    {
      "epoch": 0.30466594213730774,
      "grad_norm": 994.4598388671875,
      "learning_rate": 3.5247043150476895e-05,
      "loss": 66.9641,
      "step": 75410
    },
    {
      "epoch": 0.3047063434026754,
      "grad_norm": 849.9085693359375,
      "learning_rate": 3.5245235783475866e-05,
      "loss": 51.2462,
      "step": 75420
    },
    {
      "epoch": 0.304746744668043,
      "grad_norm": 569.6253051757812,
      "learning_rate": 3.524342811926112e-05,
      "loss": 53.9249,
      "step": 75430
    },
    {
      "epoch": 0.3047871459334106,
      "grad_norm": 1156.7808837890625,
      "learning_rate": 3.524162015786789e-05,
      "loss": 73.3264,
      "step": 75440
    },
    {
      "epoch": 0.30482754719877825,
      "grad_norm": 1112.2142333984375,
      "learning_rate": 3.523981189933144e-05,
      "loss": 74.5475,
      "step": 75450
    },
    {
      "epoch": 0.3048679484641459,
      "grad_norm": 857.4349365234375,
      "learning_rate": 3.5238003343687005e-05,
      "loss": 89.3386,
      "step": 75460
    },
    {
      "epoch": 0.30490834972951353,
      "grad_norm": 591.4561767578125,
      "learning_rate": 3.523619449096985e-05,
      "loss": 34.5415,
      "step": 75470
    },
    {
      "epoch": 0.30494875099488117,
      "grad_norm": 791.2875366210938,
      "learning_rate": 3.523438534121524e-05,
      "loss": 56.9031,
      "step": 75480
    },
    {
      "epoch": 0.3049891522602488,
      "grad_norm": 719.8794555664062,
      "learning_rate": 3.523257589445845e-05,
      "loss": 53.7153,
      "step": 75490
    },
    {
      "epoch": 0.30502955352561645,
      "grad_norm": 1207.1207275390625,
      "learning_rate": 3.523076615073474e-05,
      "loss": 93.9197,
      "step": 75500
    },
    {
      "epoch": 0.30506995479098403,
      "grad_norm": 747.3038330078125,
      "learning_rate": 3.522895611007941e-05,
      "loss": 91.8553,
      "step": 75510
    },
    {
      "epoch": 0.3051103560563517,
      "grad_norm": 646.2911987304688,
      "learning_rate": 3.522714577252773e-05,
      "loss": 51.9658,
      "step": 75520
    },
    {
      "epoch": 0.3051507573217193,
      "grad_norm": 528.4061279296875,
      "learning_rate": 3.5225335138115016e-05,
      "loss": 61.4797,
      "step": 75530
    },
    {
      "epoch": 0.30519115858708695,
      "grad_norm": 714.3599243164062,
      "learning_rate": 3.522352420687655e-05,
      "loss": 60.6513,
      "step": 75540
    },
    {
      "epoch": 0.3052315598524546,
      "grad_norm": 1063.8582763671875,
      "learning_rate": 3.522171297884764e-05,
      "loss": 84.7672,
      "step": 75550
    },
    {
      "epoch": 0.30527196111782223,
      "grad_norm": 539.630126953125,
      "learning_rate": 3.52199014540636e-05,
      "loss": 83.8034,
      "step": 75560
    },
    {
      "epoch": 0.3053123623831898,
      "grad_norm": 853.978515625,
      "learning_rate": 3.5218089632559744e-05,
      "loss": 58.7987,
      "step": 75570
    },
    {
      "epoch": 0.30535276364855746,
      "grad_norm": 271.160888671875,
      "learning_rate": 3.52162775143714e-05,
      "loss": 43.0782,
      "step": 75580
    },
    {
      "epoch": 0.3053931649139251,
      "grad_norm": 465.7930908203125,
      "learning_rate": 3.521446509953389e-05,
      "loss": 61.485,
      "step": 75590
    },
    {
      "epoch": 0.30543356617929274,
      "grad_norm": 978.6604614257812,
      "learning_rate": 3.521265238808255e-05,
      "loss": 49.8255,
      "step": 75600
    },
    {
      "epoch": 0.3054739674446604,
      "grad_norm": 1103.0828857421875,
      "learning_rate": 3.521083938005272e-05,
      "loss": 78.9844,
      "step": 75610
    },
    {
      "epoch": 0.305514368710028,
      "grad_norm": 750.4556884765625,
      "learning_rate": 3.520902607547974e-05,
      "loss": 57.8552,
      "step": 75620
    },
    {
      "epoch": 0.30555476997539566,
      "grad_norm": 667.4697265625,
      "learning_rate": 3.520721247439897e-05,
      "loss": 37.519,
      "step": 75630
    },
    {
      "epoch": 0.30559517124076324,
      "grad_norm": 1199.414306640625,
      "learning_rate": 3.520539857684577e-05,
      "loss": 91.3736,
      "step": 75640
    },
    {
      "epoch": 0.3056355725061309,
      "grad_norm": 583.2481689453125,
      "learning_rate": 3.520358438285548e-05,
      "loss": 86.4529,
      "step": 75650
    },
    {
      "epoch": 0.3056759737714985,
      "grad_norm": 1052.4718017578125,
      "learning_rate": 3.5201769892463506e-05,
      "loss": 66.1742,
      "step": 75660
    },
    {
      "epoch": 0.30571637503686616,
      "grad_norm": 515.3801879882812,
      "learning_rate": 3.519995510570519e-05,
      "loss": 87.5196,
      "step": 75670
    },
    {
      "epoch": 0.3057567763022338,
      "grad_norm": 666.802490234375,
      "learning_rate": 3.519814002261593e-05,
      "loss": 55.1974,
      "step": 75680
    },
    {
      "epoch": 0.30579717756760144,
      "grad_norm": 973.0885009765625,
      "learning_rate": 3.5196324643231094e-05,
      "loss": 66.9718,
      "step": 75690
    },
    {
      "epoch": 0.305837578832969,
      "grad_norm": 1037.1934814453125,
      "learning_rate": 3.51945089675861e-05,
      "loss": 54.3903,
      "step": 75700
    },
    {
      "epoch": 0.30587798009833667,
      "grad_norm": 1361.24560546875,
      "learning_rate": 3.5192692995716324e-05,
      "loss": 57.3873,
      "step": 75710
    },
    {
      "epoch": 0.3059183813637043,
      "grad_norm": 409.15191650390625,
      "learning_rate": 3.519087672765717e-05,
      "loss": 47.4263,
      "step": 75720
    },
    {
      "epoch": 0.30595878262907195,
      "grad_norm": 472.4361877441406,
      "learning_rate": 3.518906016344406e-05,
      "loss": 66.8334,
      "step": 75730
    },
    {
      "epoch": 0.3059991838944396,
      "grad_norm": 229.70162963867188,
      "learning_rate": 3.5187243303112406e-05,
      "loss": 60.1793,
      "step": 75740
    },
    {
      "epoch": 0.3060395851598072,
      "grad_norm": 493.8519287109375,
      "learning_rate": 3.518542614669762e-05,
      "loss": 53.0308,
      "step": 75750
    },
    {
      "epoch": 0.3060799864251748,
      "grad_norm": 646.2508544921875,
      "learning_rate": 3.518360869423514e-05,
      "loss": 89.02,
      "step": 75760
    },
    {
      "epoch": 0.30612038769054245,
      "grad_norm": 0.0,
      "learning_rate": 3.518179094576038e-05,
      "loss": 63.6022,
      "step": 75770
    },
    {
      "epoch": 0.3061607889559101,
      "grad_norm": 1185.7301025390625,
      "learning_rate": 3.5179972901308794e-05,
      "loss": 60.0821,
      "step": 75780
    },
    {
      "epoch": 0.30620119022127773,
      "grad_norm": 1166.6944580078125,
      "learning_rate": 3.5178154560915825e-05,
      "loss": 91.2704,
      "step": 75790
    },
    {
      "epoch": 0.30624159148664537,
      "grad_norm": 1487.95166015625,
      "learning_rate": 3.5176335924616916e-05,
      "loss": 91.0684,
      "step": 75800
    },
    {
      "epoch": 0.306281992752013,
      "grad_norm": 524.3199462890625,
      "learning_rate": 3.517451699244752e-05,
      "loss": 108.7739,
      "step": 75810
    },
    {
      "epoch": 0.30632239401738065,
      "grad_norm": 0.0,
      "learning_rate": 3.517269776444311e-05,
      "loss": 70.3399,
      "step": 75820
    },
    {
      "epoch": 0.30636279528274823,
      "grad_norm": 1462.861572265625,
      "learning_rate": 3.5170878240639145e-05,
      "loss": 90.5969,
      "step": 75830
    },
    {
      "epoch": 0.3064031965481159,
      "grad_norm": 309.7550964355469,
      "learning_rate": 3.516905842107109e-05,
      "loss": 48.8648,
      "step": 75840
    },
    {
      "epoch": 0.3064435978134835,
      "grad_norm": 1339.1041259765625,
      "learning_rate": 3.5167238305774444e-05,
      "loss": 56.7237,
      "step": 75850
    },
    {
      "epoch": 0.30648399907885115,
      "grad_norm": 1145.357177734375,
      "learning_rate": 3.516541789478467e-05,
      "loss": 93.9895,
      "step": 75860
    },
    {
      "epoch": 0.3065244003442188,
      "grad_norm": 585.9578857421875,
      "learning_rate": 3.516359718813727e-05,
      "loss": 53.5185,
      "step": 75870
    },
    {
      "epoch": 0.30656480160958643,
      "grad_norm": 633.2238159179688,
      "learning_rate": 3.516177618586773e-05,
      "loss": 50.6192,
      "step": 75880
    },
    {
      "epoch": 0.306605202874954,
      "grad_norm": 664.184326171875,
      "learning_rate": 3.5159954888011564e-05,
      "loss": 85.1204,
      "step": 75890
    },
    {
      "epoch": 0.30664560414032166,
      "grad_norm": 1013.7833862304688,
      "learning_rate": 3.515813329460427e-05,
      "loss": 48.1728,
      "step": 75900
    },
    {
      "epoch": 0.3066860054056893,
      "grad_norm": 904.5157470703125,
      "learning_rate": 3.5156311405681366e-05,
      "loss": 40.3763,
      "step": 75910
    },
    {
      "epoch": 0.30672640667105694,
      "grad_norm": 571.832275390625,
      "learning_rate": 3.5154489221278366e-05,
      "loss": 76.1373,
      "step": 75920
    },
    {
      "epoch": 0.3067668079364246,
      "grad_norm": 1582.66015625,
      "learning_rate": 3.51526667414308e-05,
      "loss": 60.2339,
      "step": 75930
    },
    {
      "epoch": 0.3068072092017922,
      "grad_norm": 799.6014404296875,
      "learning_rate": 3.515084396617419e-05,
      "loss": 77.2938,
      "step": 75940
    },
    {
      "epoch": 0.3068476104671598,
      "grad_norm": 670.2860107421875,
      "learning_rate": 3.514902089554408e-05,
      "loss": 83.6992,
      "step": 75950
    },
    {
      "epoch": 0.30688801173252744,
      "grad_norm": 515.1392822265625,
      "learning_rate": 3.5147197529576e-05,
      "loss": 70.0786,
      "step": 75960
    },
    {
      "epoch": 0.3069284129978951,
      "grad_norm": 711.9067993164062,
      "learning_rate": 3.514537386830552e-05,
      "loss": 53.2691,
      "step": 75970
    },
    {
      "epoch": 0.3069688142632627,
      "grad_norm": 1451.2913818359375,
      "learning_rate": 3.5143549911768166e-05,
      "loss": 69.1946,
      "step": 75980
    },
    {
      "epoch": 0.30700921552863036,
      "grad_norm": 539.037841796875,
      "learning_rate": 3.514172565999952e-05,
      "loss": 76.1712,
      "step": 75990
    },
    {
      "epoch": 0.307049616793998,
      "grad_norm": 811.0957641601562,
      "learning_rate": 3.513990111303513e-05,
      "loss": 80.0229,
      "step": 76000
    },
    {
      "epoch": 0.30709001805936564,
      "grad_norm": 866.5407104492188,
      "learning_rate": 3.5138076270910575e-05,
      "loss": 66.0944,
      "step": 76010
    },
    {
      "epoch": 0.30713041932473323,
      "grad_norm": 797.3560180664062,
      "learning_rate": 3.513625113366144e-05,
      "loss": 47.7251,
      "step": 76020
    },
    {
      "epoch": 0.30717082059010087,
      "grad_norm": 495.70745849609375,
      "learning_rate": 3.513442570132328e-05,
      "loss": 55.8337,
      "step": 76030
    },
    {
      "epoch": 0.3072112218554685,
      "grad_norm": 1171.0638427734375,
      "learning_rate": 3.513259997393171e-05,
      "loss": 65.7174,
      "step": 76040
    },
    {
      "epoch": 0.30725162312083615,
      "grad_norm": 978.8792114257812,
      "learning_rate": 3.51307739515223e-05,
      "loss": 66.6673,
      "step": 76050
    },
    {
      "epoch": 0.3072920243862038,
      "grad_norm": 1636.1409912109375,
      "learning_rate": 3.512894763413068e-05,
      "loss": 77.5981,
      "step": 76060
    },
    {
      "epoch": 0.30733242565157143,
      "grad_norm": 803.0050048828125,
      "learning_rate": 3.5127121021792425e-05,
      "loss": 66.8965,
      "step": 76070
    },
    {
      "epoch": 0.307372826916939,
      "grad_norm": 745.1494750976562,
      "learning_rate": 3.512529411454316e-05,
      "loss": 69.0238,
      "step": 76080
    },
    {
      "epoch": 0.30741322818230665,
      "grad_norm": 547.9581298828125,
      "learning_rate": 3.51234669124185e-05,
      "loss": 44.1149,
      "step": 76090
    },
    {
      "epoch": 0.3074536294476743,
      "grad_norm": 1483.5592041015625,
      "learning_rate": 3.512163941545407e-05,
      "loss": 73.7918,
      "step": 76100
    },
    {
      "epoch": 0.30749403071304193,
      "grad_norm": 605.3894653320312,
      "learning_rate": 3.511981162368549e-05,
      "loss": 71.8262,
      "step": 76110
    },
    {
      "epoch": 0.3075344319784096,
      "grad_norm": 687.6273803710938,
      "learning_rate": 3.5117983537148395e-05,
      "loss": 71.0229,
      "step": 76120
    },
    {
      "epoch": 0.3075748332437772,
      "grad_norm": 1335.818115234375,
      "learning_rate": 3.511615515587843e-05,
      "loss": 71.5045,
      "step": 76130
    },
    {
      "epoch": 0.30761523450914485,
      "grad_norm": 822.9049682617188,
      "learning_rate": 3.5114326479911244e-05,
      "loss": 58.5649,
      "step": 76140
    },
    {
      "epoch": 0.30765563577451244,
      "grad_norm": 729.94384765625,
      "learning_rate": 3.5112497509282474e-05,
      "loss": 61.9444,
      "step": 76150
    },
    {
      "epoch": 0.3076960370398801,
      "grad_norm": 1042.3271484375,
      "learning_rate": 3.511066824402779e-05,
      "loss": 92.8846,
      "step": 76160
    },
    {
      "epoch": 0.3077364383052477,
      "grad_norm": 794.2461547851562,
      "learning_rate": 3.510883868418284e-05,
      "loss": 72.1416,
      "step": 76170
    },
    {
      "epoch": 0.30777683957061536,
      "grad_norm": 984.2166748046875,
      "learning_rate": 3.5107008829783314e-05,
      "loss": 73.4033,
      "step": 76180
    },
    {
      "epoch": 0.307817240835983,
      "grad_norm": 513.4242553710938,
      "learning_rate": 3.510517868086487e-05,
      "loss": 53.9028,
      "step": 76190
    },
    {
      "epoch": 0.30785764210135064,
      "grad_norm": 797.1355590820312,
      "learning_rate": 3.5103348237463184e-05,
      "loss": 68.143,
      "step": 76200
    },
    {
      "epoch": 0.3078980433667182,
      "grad_norm": 370.9480285644531,
      "learning_rate": 3.510151749961395e-05,
      "loss": 109.3378,
      "step": 76210
    },
    {
      "epoch": 0.30793844463208586,
      "grad_norm": 606.9356689453125,
      "learning_rate": 3.509968646735287e-05,
      "loss": 58.9858,
      "step": 76220
    },
    {
      "epoch": 0.3079788458974535,
      "grad_norm": 894.0355224609375,
      "learning_rate": 3.509785514071562e-05,
      "loss": 87.4262,
      "step": 76230
    },
    {
      "epoch": 0.30801924716282114,
      "grad_norm": 1086.481689453125,
      "learning_rate": 3.50960235197379e-05,
      "loss": 71.8501,
      "step": 76240
    },
    {
      "epoch": 0.3080596484281888,
      "grad_norm": 499.71099853515625,
      "learning_rate": 3.5094191604455446e-05,
      "loss": 53.5725,
      "step": 76250
    },
    {
      "epoch": 0.3081000496935564,
      "grad_norm": 692.6753540039062,
      "learning_rate": 3.509235939490394e-05,
      "loss": 83.9878,
      "step": 76260
    },
    {
      "epoch": 0.308140450958924,
      "grad_norm": 751.8206176757812,
      "learning_rate": 3.509052689111913e-05,
      "loss": 88.5014,
      "step": 76270
    },
    {
      "epoch": 0.30818085222429165,
      "grad_norm": 567.2142333984375,
      "learning_rate": 3.5088694093136726e-05,
      "loss": 53.7215,
      "step": 76280
    },
    {
      "epoch": 0.3082212534896593,
      "grad_norm": 944.86376953125,
      "learning_rate": 3.508686100099246e-05,
      "loss": 87.2339,
      "step": 76290
    },
    {
      "epoch": 0.3082616547550269,
      "grad_norm": 872.0805053710938,
      "learning_rate": 3.508502761472208e-05,
      "loss": 80.3885,
      "step": 76300
    },
    {
      "epoch": 0.30830205602039457,
      "grad_norm": 715.2340087890625,
      "learning_rate": 3.508319393436131e-05,
      "loss": 86.7982,
      "step": 76310
    },
    {
      "epoch": 0.3083424572857622,
      "grad_norm": 1295.7491455078125,
      "learning_rate": 3.5081359959945916e-05,
      "loss": 65.9816,
      "step": 76320
    },
    {
      "epoch": 0.30838285855112985,
      "grad_norm": 2658.11962890625,
      "learning_rate": 3.5079525691511644e-05,
      "loss": 46.9333,
      "step": 76330
    },
    {
      "epoch": 0.30842325981649743,
      "grad_norm": 754.1270141601562,
      "learning_rate": 3.507769112909425e-05,
      "loss": 72.6826,
      "step": 76340
    },
    {
      "epoch": 0.30846366108186507,
      "grad_norm": 646.848876953125,
      "learning_rate": 3.507585627272951e-05,
      "loss": 51.2794,
      "step": 76350
    },
    {
      "epoch": 0.3085040623472327,
      "grad_norm": 644.6250610351562,
      "learning_rate": 3.507402112245319e-05,
      "loss": 54.5868,
      "step": 76360
    },
    {
      "epoch": 0.30854446361260035,
      "grad_norm": 2552.768798828125,
      "learning_rate": 3.507218567830107e-05,
      "loss": 94.0588,
      "step": 76370
    },
    {
      "epoch": 0.308584864877968,
      "grad_norm": 402.1734924316406,
      "learning_rate": 3.507034994030892e-05,
      "loss": 59.7754,
      "step": 76380
    },
    {
      "epoch": 0.30862526614333563,
      "grad_norm": 1175.5374755859375,
      "learning_rate": 3.506851390851255e-05,
      "loss": 65.8783,
      "step": 76390
    },
    {
      "epoch": 0.3086656674087032,
      "grad_norm": 819.6211547851562,
      "learning_rate": 3.5066677582947744e-05,
      "loss": 88.8508,
      "step": 76400
    },
    {
      "epoch": 0.30870606867407085,
      "grad_norm": 697.6337890625,
      "learning_rate": 3.5064840963650295e-05,
      "loss": 49.8864,
      "step": 76410
    },
    {
      "epoch": 0.3087464699394385,
      "grad_norm": 438.434814453125,
      "learning_rate": 3.5063004050656016e-05,
      "loss": 84.5187,
      "step": 76420
    },
    {
      "epoch": 0.30878687120480613,
      "grad_norm": 894.0114135742188,
      "learning_rate": 3.506116684400072e-05,
      "loss": 63.094,
      "step": 76430
    },
    {
      "epoch": 0.3088272724701738,
      "grad_norm": 544.4618530273438,
      "learning_rate": 3.505932934372022e-05,
      "loss": 63.0377,
      "step": 76440
    },
    {
      "epoch": 0.3088676737355414,
      "grad_norm": 485.47479248046875,
      "learning_rate": 3.505749154985035e-05,
      "loss": 50.5823,
      "step": 76450
    },
    {
      "epoch": 0.30890807500090905,
      "grad_norm": 914.7135620117188,
      "learning_rate": 3.505565346242692e-05,
      "loss": 78.1928,
      "step": 76460
    },
    {
      "epoch": 0.30894847626627664,
      "grad_norm": 694.2228393554688,
      "learning_rate": 3.5053815081485776e-05,
      "loss": 66.9965,
      "step": 76470
    },
    {
      "epoch": 0.3089888775316443,
      "grad_norm": 900.4125366210938,
      "learning_rate": 3.505197640706276e-05,
      "loss": 59.6299,
      "step": 76480
    },
    {
      "epoch": 0.3090292787970119,
      "grad_norm": 899.3259887695312,
      "learning_rate": 3.505013743919372e-05,
      "loss": 49.0054,
      "step": 76490
    },
    {
      "epoch": 0.30906968006237956,
      "grad_norm": 506.0701599121094,
      "learning_rate": 3.504829817791449e-05,
      "loss": 67.3172,
      "step": 76500
    },
    {
      "epoch": 0.3091100813277472,
      "grad_norm": 488.9270324707031,
      "learning_rate": 3.5046458623260946e-05,
      "loss": 64.5393,
      "step": 76510
    },
    {
      "epoch": 0.30915048259311484,
      "grad_norm": 532.37939453125,
      "learning_rate": 3.5044618775268944e-05,
      "loss": 77.8112,
      "step": 76520
    },
    {
      "epoch": 0.3091908838584824,
      "grad_norm": 1935.016845703125,
      "learning_rate": 3.5042778633974355e-05,
      "loss": 81.3636,
      "step": 76530
    },
    {
      "epoch": 0.30923128512385006,
      "grad_norm": 1644.5169677734375,
      "learning_rate": 3.504093819941305e-05,
      "loss": 75.7931,
      "step": 76540
    },
    {
      "epoch": 0.3092716863892177,
      "grad_norm": 980.7655029296875,
      "learning_rate": 3.503909747162091e-05,
      "loss": 70.2966,
      "step": 76550
    },
    {
      "epoch": 0.30931208765458534,
      "grad_norm": 1009.39599609375,
      "learning_rate": 3.503725645063383e-05,
      "loss": 123.4905,
      "step": 76560
    },
    {
      "epoch": 0.309352488919953,
      "grad_norm": 980.9332275390625,
      "learning_rate": 3.5035415136487685e-05,
      "loss": 60.888,
      "step": 76570
    },
    {
      "epoch": 0.3093928901853206,
      "grad_norm": 764.3687133789062,
      "learning_rate": 3.503357352921839e-05,
      "loss": 68.6264,
      "step": 76580
    },
    {
      "epoch": 0.3094332914506882,
      "grad_norm": 1286.886962890625,
      "learning_rate": 3.503173162886183e-05,
      "loss": 54.0793,
      "step": 76590
    },
    {
      "epoch": 0.30947369271605585,
      "grad_norm": 1331.6363525390625,
      "learning_rate": 3.5029889435453924e-05,
      "loss": 85.7386,
      "step": 76600
    },
    {
      "epoch": 0.3095140939814235,
      "grad_norm": 393.58111572265625,
      "learning_rate": 3.5028046949030584e-05,
      "loss": 74.7227,
      "step": 76610
    },
    {
      "epoch": 0.30955449524679113,
      "grad_norm": 512.5623168945312,
      "learning_rate": 3.5026204169627744e-05,
      "loss": 97.1083,
      "step": 76620
    },
    {
      "epoch": 0.30959489651215877,
      "grad_norm": 1050.2479248046875,
      "learning_rate": 3.50243610972813e-05,
      "loss": 61.3017,
      "step": 76630
    },
    {
      "epoch": 0.3096352977775264,
      "grad_norm": 682.5951538085938,
      "learning_rate": 3.502251773202722e-05,
      "loss": 77.5261,
      "step": 76640
    },
    {
      "epoch": 0.30967569904289405,
      "grad_norm": 1385.592529296875,
      "learning_rate": 3.5020674073901406e-05,
      "loss": 96.6229,
      "step": 76650
    },
    {
      "epoch": 0.30971610030826163,
      "grad_norm": 1093.2525634765625,
      "learning_rate": 3.501883012293983e-05,
      "loss": 82.5706,
      "step": 76660
    },
    {
      "epoch": 0.3097565015736293,
      "grad_norm": 1153.8555908203125,
      "learning_rate": 3.501698587917842e-05,
      "loss": 49.4748,
      "step": 76670
    },
    {
      "epoch": 0.3097969028389969,
      "grad_norm": 424.3147888183594,
      "learning_rate": 3.501514134265315e-05,
      "loss": 65.4982,
      "step": 76680
    },
    {
      "epoch": 0.30983730410436455,
      "grad_norm": 1009.4806518554688,
      "learning_rate": 3.501329651339996e-05,
      "loss": 64.0963,
      "step": 76690
    },
    {
      "epoch": 0.3098777053697322,
      "grad_norm": 1129.088623046875,
      "learning_rate": 3.501145139145483e-05,
      "loss": 110.2901,
      "step": 76700
    },
    {
      "epoch": 0.30991810663509983,
      "grad_norm": 426.0953369140625,
      "learning_rate": 3.500960597685372e-05,
      "loss": 73.1976,
      "step": 76710
    },
    {
      "epoch": 0.3099585079004674,
      "grad_norm": 0.0,
      "learning_rate": 3.500776026963262e-05,
      "loss": 55.8843,
      "step": 76720
    },
    {
      "epoch": 0.30999890916583506,
      "grad_norm": 387.7458190917969,
      "learning_rate": 3.50059142698275e-05,
      "loss": 43.3823,
      "step": 76730
    },
    {
      "epoch": 0.3100393104312027,
      "grad_norm": 823.2538452148438,
      "learning_rate": 3.500406797747436e-05,
      "loss": 71.2642,
      "step": 76740
    },
    {
      "epoch": 0.31007971169657034,
      "grad_norm": 1283.1015625,
      "learning_rate": 3.5002221392609196e-05,
      "loss": 54.9085,
      "step": 76750
    },
    {
      "epoch": 0.310120112961938,
      "grad_norm": 1642.7064208984375,
      "learning_rate": 3.5000374515268e-05,
      "loss": 102.8734,
      "step": 76760
    },
    {
      "epoch": 0.3101605142273056,
      "grad_norm": 732.9937133789062,
      "learning_rate": 3.499852734548677e-05,
      "loss": 57.9155,
      "step": 76770
    },
    {
      "epoch": 0.31020091549267326,
      "grad_norm": 1375.7432861328125,
      "learning_rate": 3.4996679883301535e-05,
      "loss": 65.2121,
      "step": 76780
    },
    {
      "epoch": 0.31024131675804084,
      "grad_norm": 434.67779541015625,
      "learning_rate": 3.49948321287483e-05,
      "loss": 71.201,
      "step": 76790
    },
    {
      "epoch": 0.3102817180234085,
      "grad_norm": 1775.1773681640625,
      "learning_rate": 3.49929840818631e-05,
      "loss": 68.5134,
      "step": 76800
    },
    {
      "epoch": 0.3103221192887761,
      "grad_norm": 531.1075439453125,
      "learning_rate": 3.499113574268196e-05,
      "loss": 50.1119,
      "step": 76810
    },
    {
      "epoch": 0.31036252055414376,
      "grad_norm": 1148.52001953125,
      "learning_rate": 3.49892871112409e-05,
      "loss": 77.1326,
      "step": 76820
    },
    {
      "epoch": 0.3104029218195114,
      "grad_norm": 1055.23828125,
      "learning_rate": 3.498743818757598e-05,
      "loss": 62.7596,
      "step": 76830
    },
    {
      "epoch": 0.31044332308487904,
      "grad_norm": 885.1285400390625,
      "learning_rate": 3.498558897172324e-05,
      "loss": 75.788,
      "step": 76840
    },
    {
      "epoch": 0.3104837243502466,
      "grad_norm": 733.3975830078125,
      "learning_rate": 3.4983739463718706e-05,
      "loss": 51.2205,
      "step": 76850
    },
    {
      "epoch": 0.31052412561561427,
      "grad_norm": 620.9288940429688,
      "learning_rate": 3.498188966359848e-05,
      "loss": 93.8355,
      "step": 76860
    },
    {
      "epoch": 0.3105645268809819,
      "grad_norm": 1757.525146484375,
      "learning_rate": 3.498003957139859e-05,
      "loss": 63.8485,
      "step": 76870
    },
    {
      "epoch": 0.31060492814634955,
      "grad_norm": 162.37403869628906,
      "learning_rate": 3.4978189187155114e-05,
      "loss": 62.836,
      "step": 76880
    },
    {
      "epoch": 0.3106453294117172,
      "grad_norm": 573.8180541992188,
      "learning_rate": 3.4976338510904134e-05,
      "loss": 50.8696,
      "step": 76890
    },
    {
      "epoch": 0.3106857306770848,
      "grad_norm": 734.051025390625,
      "learning_rate": 3.4974487542681724e-05,
      "loss": 46.2981,
      "step": 76900
    },
    {
      "epoch": 0.3107261319424524,
      "grad_norm": 1082.28955078125,
      "learning_rate": 3.497263628252397e-05,
      "loss": 70.5763,
      "step": 76910
    },
    {
      "epoch": 0.31076653320782005,
      "grad_norm": 742.8665161132812,
      "learning_rate": 3.497078473046697e-05,
      "loss": 62.9141,
      "step": 76920
    },
    {
      "epoch": 0.3108069344731877,
      "grad_norm": 758.2858276367188,
      "learning_rate": 3.49689328865468e-05,
      "loss": 57.3749,
      "step": 76930
    },
    {
      "epoch": 0.31084733573855533,
      "grad_norm": 478.9095764160156,
      "learning_rate": 3.496708075079959e-05,
      "loss": 36.4947,
      "step": 76940
    },
    {
      "epoch": 0.31088773700392297,
      "grad_norm": 969.2333374023438,
      "learning_rate": 3.496522832326143e-05,
      "loss": 104.6406,
      "step": 76950
    },
    {
      "epoch": 0.3109281382692906,
      "grad_norm": 642.496826171875,
      "learning_rate": 3.496337560396844e-05,
      "loss": 54.6326,
      "step": 76960
    },
    {
      "epoch": 0.31096853953465825,
      "grad_norm": 979.9578857421875,
      "learning_rate": 3.496152259295673e-05,
      "loss": 70.7781,
      "step": 76970
    },
    {
      "epoch": 0.31100894080002583,
      "grad_norm": 1154.487060546875,
      "learning_rate": 3.495966929026244e-05,
      "loss": 53.4658,
      "step": 76980
    },
    {
      "epoch": 0.3110493420653935,
      "grad_norm": 1025.4051513671875,
      "learning_rate": 3.49578156959217e-05,
      "loss": 58.9349,
      "step": 76990
    },
    {
      "epoch": 0.3110897433307611,
      "grad_norm": 1516.22998046875,
      "learning_rate": 3.495596180997064e-05,
      "loss": 108.4065,
      "step": 77000
    },
    {
      "epoch": 0.31113014459612875,
      "grad_norm": 1075.0015869140625,
      "learning_rate": 3.495410763244541e-05,
      "loss": 81.8675,
      "step": 77010
    },
    {
      "epoch": 0.3111705458614964,
      "grad_norm": 1219.5714111328125,
      "learning_rate": 3.4952253163382144e-05,
      "loss": 64.139,
      "step": 77020
    },
    {
      "epoch": 0.31121094712686403,
      "grad_norm": 854.6863403320312,
      "learning_rate": 3.4950398402817006e-05,
      "loss": 93.2717,
      "step": 77030
    },
    {
      "epoch": 0.3112513483922316,
      "grad_norm": 487.3167724609375,
      "learning_rate": 3.4948543350786156e-05,
      "loss": 46.6558,
      "step": 77040
    },
    {
      "epoch": 0.31129174965759926,
      "grad_norm": 181.5683135986328,
      "learning_rate": 3.494668800732575e-05,
      "loss": 64.1349,
      "step": 77050
    },
    {
      "epoch": 0.3113321509229669,
      "grad_norm": 940.066650390625,
      "learning_rate": 3.4944832372471977e-05,
      "loss": 87.6811,
      "step": 77060
    },
    {
      "epoch": 0.31137255218833454,
      "grad_norm": 636.5015869140625,
      "learning_rate": 3.4942976446261e-05,
      "loss": 68.7378,
      "step": 77070
    },
    {
      "epoch": 0.3114129534537022,
      "grad_norm": 477.9764709472656,
      "learning_rate": 3.494112022872901e-05,
      "loss": 66.3613,
      "step": 77080
    },
    {
      "epoch": 0.3114533547190698,
      "grad_norm": 1191.1737060546875,
      "learning_rate": 3.493926371991218e-05,
      "loss": 70.3055,
      "step": 77090
    },
    {
      "epoch": 0.31149375598443746,
      "grad_norm": 873.9033203125,
      "learning_rate": 3.493740691984672e-05,
      "loss": 72.6238,
      "step": 77100
    },
    {
      "epoch": 0.31153415724980504,
      "grad_norm": 656.4010620117188,
      "learning_rate": 3.4935549828568807e-05,
      "loss": 63.6683,
      "step": 77110
    },
    {
      "epoch": 0.3115745585151727,
      "grad_norm": 533.2958374023438,
      "learning_rate": 3.493369244611467e-05,
      "loss": 75.7154,
      "step": 77120
    },
    {
      "epoch": 0.3116149597805403,
      "grad_norm": 477.51409912109375,
      "learning_rate": 3.493183477252051e-05,
      "loss": 47.95,
      "step": 77130
    },
    {
      "epoch": 0.31165536104590796,
      "grad_norm": 808.3378295898438,
      "learning_rate": 3.4929976807822546e-05,
      "loss": 54.0265,
      "step": 77140
    },
    {
      "epoch": 0.3116957623112756,
      "grad_norm": 540.817626953125,
      "learning_rate": 3.4928118552056994e-05,
      "loss": 39.5284,
      "step": 77150
    },
    {
      "epoch": 0.31173616357664324,
      "grad_norm": 0.0,
      "learning_rate": 3.492626000526008e-05,
      "loss": 78.5489,
      "step": 77160
    },
    {
      "epoch": 0.31177656484201083,
      "grad_norm": 1253.980712890625,
      "learning_rate": 3.492440116746805e-05,
      "loss": 62.1076,
      "step": 77170
    },
    {
      "epoch": 0.31181696610737847,
      "grad_norm": 427.8959045410156,
      "learning_rate": 3.492254203871714e-05,
      "loss": 84.0081,
      "step": 77180
    },
    {
      "epoch": 0.3118573673727461,
      "grad_norm": 1483.2845458984375,
      "learning_rate": 3.4920682619043584e-05,
      "loss": 66.6233,
      "step": 77190
    },
    {
      "epoch": 0.31189776863811375,
      "grad_norm": 578.9376831054688,
      "learning_rate": 3.4918822908483645e-05,
      "loss": 61.4817,
      "step": 77200
    },
    {
      "epoch": 0.3119381699034814,
      "grad_norm": 415.4118957519531,
      "learning_rate": 3.491696290707357e-05,
      "loss": 69.8285,
      "step": 77210
    },
    {
      "epoch": 0.31197857116884903,
      "grad_norm": 1217.6763916015625,
      "learning_rate": 3.491510261484962e-05,
      "loss": 77.4011,
      "step": 77220
    },
    {
      "epoch": 0.3120189724342166,
      "grad_norm": 1282.310302734375,
      "learning_rate": 3.4913242031848064e-05,
      "loss": 70.3646,
      "step": 77230
    },
    {
      "epoch": 0.31205937369958425,
      "grad_norm": 478.583251953125,
      "learning_rate": 3.4911381158105185e-05,
      "loss": 75.6116,
      "step": 77240
    },
    {
      "epoch": 0.3120997749649519,
      "grad_norm": 778.688232421875,
      "learning_rate": 3.4909519993657244e-05,
      "loss": 62.7415,
      "step": 77250
    },
    {
      "epoch": 0.31214017623031953,
      "grad_norm": 1136.1414794921875,
      "learning_rate": 3.490765853854054e-05,
      "loss": 106.0001,
      "step": 77260
    },
    {
      "epoch": 0.3121805774956872,
      "grad_norm": 640.33203125,
      "learning_rate": 3.490579679279136e-05,
      "loss": 51.8542,
      "step": 77270
    },
    {
      "epoch": 0.3122209787610548,
      "grad_norm": 750.8095092773438,
      "learning_rate": 3.4903934756445995e-05,
      "loss": 43.308,
      "step": 77280
    },
    {
      "epoch": 0.31226138002642245,
      "grad_norm": 1000.5927734375,
      "learning_rate": 3.490207242954075e-05,
      "loss": 66.9694,
      "step": 77290
    },
    {
      "epoch": 0.31230178129179004,
      "grad_norm": 788.5052490234375,
      "learning_rate": 3.4900209812111927e-05,
      "loss": 72.0724,
      "step": 77300
    },
    {
      "epoch": 0.3123421825571577,
      "grad_norm": 1133.585693359375,
      "learning_rate": 3.489834690419584e-05,
      "loss": 81.6365,
      "step": 77310
    },
    {
      "epoch": 0.3123825838225253,
      "grad_norm": 701.0135498046875,
      "learning_rate": 3.489648370582882e-05,
      "loss": 92.826,
      "step": 77320
    },
    {
      "epoch": 0.31242298508789296,
      "grad_norm": 754.3093872070312,
      "learning_rate": 3.489462021704717e-05,
      "loss": 75.8383,
      "step": 77330
    },
    {
      "epoch": 0.3124633863532606,
      "grad_norm": 523.960693359375,
      "learning_rate": 3.489275643788724e-05,
      "loss": 76.4888,
      "step": 77340
    },
    {
      "epoch": 0.31250378761862824,
      "grad_norm": 859.30029296875,
      "learning_rate": 3.489089236838535e-05,
      "loss": 96.7169,
      "step": 77350
    },
    {
      "epoch": 0.3125441888839958,
      "grad_norm": 3732.568603515625,
      "learning_rate": 3.488902800857785e-05,
      "loss": 71.5349,
      "step": 77360
    },
    {
      "epoch": 0.31258459014936346,
      "grad_norm": 763.2684936523438,
      "learning_rate": 3.488716335850108e-05,
      "loss": 56.7226,
      "step": 77370
    },
    {
      "epoch": 0.3126249914147311,
      "grad_norm": 479.9824523925781,
      "learning_rate": 3.4885298418191405e-05,
      "loss": 69.6836,
      "step": 77380
    },
    {
      "epoch": 0.31266539268009874,
      "grad_norm": 1107.8890380859375,
      "learning_rate": 3.488343318768516e-05,
      "loss": 73.3432,
      "step": 77390
    },
    {
      "epoch": 0.3127057939454664,
      "grad_norm": 517.3402099609375,
      "learning_rate": 3.488156766701873e-05,
      "loss": 90.4566,
      "step": 77400
    },
    {
      "epoch": 0.312746195210834,
      "grad_norm": 727.0076293945312,
      "learning_rate": 3.487970185622848e-05,
      "loss": 111.965,
      "step": 77410
    },
    {
      "epoch": 0.31278659647620166,
      "grad_norm": 507.585693359375,
      "learning_rate": 3.487783575535078e-05,
      "loss": 39.8812,
      "step": 77420
    },
    {
      "epoch": 0.31282699774156925,
      "grad_norm": 1095.2391357421875,
      "learning_rate": 3.487596936442201e-05,
      "loss": 65.871,
      "step": 77430
    },
    {
      "epoch": 0.3128673990069369,
      "grad_norm": 462.9928283691406,
      "learning_rate": 3.487410268347856e-05,
      "loss": 57.1373,
      "step": 77440
    },
    {
      "epoch": 0.3129078002723045,
      "grad_norm": 531.0853271484375,
      "learning_rate": 3.487223571255682e-05,
      "loss": 87.5273,
      "step": 77450
    },
    {
      "epoch": 0.31294820153767217,
      "grad_norm": 648.5484619140625,
      "learning_rate": 3.4870368451693184e-05,
      "loss": 71.222,
      "step": 77460
    },
    {
      "epoch": 0.3129886028030398,
      "grad_norm": 512.2035522460938,
      "learning_rate": 3.486850090092407e-05,
      "loss": 78.9093,
      "step": 77470
    },
    {
      "epoch": 0.31302900406840745,
      "grad_norm": 549.501220703125,
      "learning_rate": 3.486663306028587e-05,
      "loss": 120.1285,
      "step": 77480
    },
    {
      "epoch": 0.31306940533377503,
      "grad_norm": 826.307861328125,
      "learning_rate": 3.4864764929815e-05,
      "loss": 39.0094,
      "step": 77490
    },
    {
      "epoch": 0.31310980659914267,
      "grad_norm": 1610.810791015625,
      "learning_rate": 3.4862896509547886e-05,
      "loss": 107.4009,
      "step": 77500
    },
    {
      "epoch": 0.3131502078645103,
      "grad_norm": 1580.2742919921875,
      "learning_rate": 3.4861027799520956e-05,
      "loss": 126.6358,
      "step": 77510
    },
    {
      "epoch": 0.31319060912987795,
      "grad_norm": 1006.2523193359375,
      "learning_rate": 3.4859158799770635e-05,
      "loss": 55.5261,
      "step": 77520
    },
    {
      "epoch": 0.3132310103952456,
      "grad_norm": 463.68115234375,
      "learning_rate": 3.4857289510333365e-05,
      "loss": 43.9542,
      "step": 77530
    },
    {
      "epoch": 0.31327141166061323,
      "grad_norm": 1687.350341796875,
      "learning_rate": 3.485541993124559e-05,
      "loss": 55.4924,
      "step": 77540
    },
    {
      "epoch": 0.3133118129259808,
      "grad_norm": 844.6075439453125,
      "learning_rate": 3.485355006254375e-05,
      "loss": 69.7897,
      "step": 77550
    },
    {
      "epoch": 0.31335221419134845,
      "grad_norm": 1155.069580078125,
      "learning_rate": 3.4851679904264314e-05,
      "loss": 60.9398,
      "step": 77560
    },
    {
      "epoch": 0.3133926154567161,
      "grad_norm": 444.0328369140625,
      "learning_rate": 3.4849809456443725e-05,
      "loss": 43.1102,
      "step": 77570
    },
    {
      "epoch": 0.31343301672208373,
      "grad_norm": 1202.2822265625,
      "learning_rate": 3.484793871911845e-05,
      "loss": 90.6961,
      "step": 77580
    },
    {
      "epoch": 0.3134734179874514,
      "grad_norm": 1158.4188232421875,
      "learning_rate": 3.4846067692324976e-05,
      "loss": 63.8845,
      "step": 77590
    },
    {
      "epoch": 0.313513819252819,
      "grad_norm": 1004.656982421875,
      "learning_rate": 3.484419637609977e-05,
      "loss": 89.7987,
      "step": 77600
    },
    {
      "epoch": 0.31355422051818665,
      "grad_norm": 365.622802734375,
      "learning_rate": 3.48423247704793e-05,
      "loss": 58.2506,
      "step": 77610
    },
    {
      "epoch": 0.31359462178355424,
      "grad_norm": 534.121337890625,
      "learning_rate": 3.484045287550007e-05,
      "loss": 59.851,
      "step": 77620
    },
    {
      "epoch": 0.3136350230489219,
      "grad_norm": 564.9100341796875,
      "learning_rate": 3.4838580691198584e-05,
      "loss": 77.1259,
      "step": 77630
    },
    {
      "epoch": 0.3136754243142895,
      "grad_norm": 953.2900390625,
      "learning_rate": 3.4836708217611316e-05,
      "loss": 68.8026,
      "step": 77640
    },
    {
      "epoch": 0.31371582557965716,
      "grad_norm": 951.2789916992188,
      "learning_rate": 3.4834835454774784e-05,
      "loss": 59.8535,
      "step": 77650
    },
    {
      "epoch": 0.3137562268450248,
      "grad_norm": 1164.3463134765625,
      "learning_rate": 3.48329624027255e-05,
      "loss": 68.9722,
      "step": 77660
    },
    {
      "epoch": 0.31379662811039244,
      "grad_norm": 1070.16162109375,
      "learning_rate": 3.4831089061499975e-05,
      "loss": 45.6516,
      "step": 77670
    },
    {
      "epoch": 0.31383702937576,
      "grad_norm": 473.43255615234375,
      "learning_rate": 3.482921543113474e-05,
      "loss": 87.9103,
      "step": 77680
    },
    {
      "epoch": 0.31387743064112766,
      "grad_norm": 570.1047973632812,
      "learning_rate": 3.4827341511666315e-05,
      "loss": 59.4656,
      "step": 77690
    },
    {
      "epoch": 0.3139178319064953,
      "grad_norm": 865.6242065429688,
      "learning_rate": 3.482546730313122e-05,
      "loss": 46.6567,
      "step": 77700
    },
    {
      "epoch": 0.31395823317186294,
      "grad_norm": 771.6071166992188,
      "learning_rate": 3.482359280556602e-05,
      "loss": 82.0451,
      "step": 77710
    },
    {
      "epoch": 0.3139986344372306,
      "grad_norm": 896.2112426757812,
      "learning_rate": 3.482171801900725e-05,
      "loss": 90.4533,
      "step": 77720
    },
    {
      "epoch": 0.3140390357025982,
      "grad_norm": 1196.485107421875,
      "learning_rate": 3.481984294349145e-05,
      "loss": 64.6757,
      "step": 77730
    },
    {
      "epoch": 0.31407943696796586,
      "grad_norm": 807.33349609375,
      "learning_rate": 3.4817967579055176e-05,
      "loss": 75.7475,
      "step": 77740
    },
    {
      "epoch": 0.31411983823333345,
      "grad_norm": 1021.001220703125,
      "learning_rate": 3.4816091925735e-05,
      "loss": 70.141,
      "step": 77750
    },
    {
      "epoch": 0.3141602394987011,
      "grad_norm": 662.2337036132812,
      "learning_rate": 3.481421598356749e-05,
      "loss": 71.5092,
      "step": 77760
    },
    {
      "epoch": 0.31420064076406873,
      "grad_norm": 1262.9063720703125,
      "learning_rate": 3.4812339752589206e-05,
      "loss": 72.7259,
      "step": 77770
    },
    {
      "epoch": 0.31424104202943637,
      "grad_norm": 399.2818908691406,
      "learning_rate": 3.481046323283674e-05,
      "loss": 51.4161,
      "step": 77780
    },
    {
      "epoch": 0.314281443294804,
      "grad_norm": 1037.9417724609375,
      "learning_rate": 3.480858642434666e-05,
      "loss": 80.6908,
      "step": 77790
    },
    {
      "epoch": 0.31432184456017165,
      "grad_norm": 1524.025634765625,
      "learning_rate": 3.4806709327155564e-05,
      "loss": 68.7706,
      "step": 77800
    },
    {
      "epoch": 0.31436224582553923,
      "grad_norm": 1578.2554931640625,
      "learning_rate": 3.480483194130005e-05,
      "loss": 51.718,
      "step": 77810
    },
    {
      "epoch": 0.3144026470909069,
      "grad_norm": 713.9732666015625,
      "learning_rate": 3.480295426681671e-05,
      "loss": 63.0997,
      "step": 77820
    },
    {
      "epoch": 0.3144430483562745,
      "grad_norm": 542.3390502929688,
      "learning_rate": 3.480107630374217e-05,
      "loss": 51.8624,
      "step": 77830
    },
    {
      "epoch": 0.31448344962164215,
      "grad_norm": 561.2479858398438,
      "learning_rate": 3.479919805211301e-05,
      "loss": 42.934,
      "step": 77840
    },
    {
      "epoch": 0.3145238508870098,
      "grad_norm": 1051.489990234375,
      "learning_rate": 3.4797319511965875e-05,
      "loss": 80.7784,
      "step": 77850
    },
    {
      "epoch": 0.31456425215237743,
      "grad_norm": 1260.8740234375,
      "learning_rate": 3.479544068333737e-05,
      "loss": 87.1448,
      "step": 77860
    },
    {
      "epoch": 0.314604653417745,
      "grad_norm": 724.090087890625,
      "learning_rate": 3.479356156626414e-05,
      "loss": 53.8852,
      "step": 77870
    },
    {
      "epoch": 0.31464505468311266,
      "grad_norm": 1093.593505859375,
      "learning_rate": 3.479168216078281e-05,
      "loss": 61.8862,
      "step": 77880
    },
    {
      "epoch": 0.3146854559484803,
      "grad_norm": 875.7823486328125,
      "learning_rate": 3.478980246693001e-05,
      "loss": 84.8797,
      "step": 77890
    },
    {
      "epoch": 0.31472585721384794,
      "grad_norm": 1626.4224853515625,
      "learning_rate": 3.478792248474241e-05,
      "loss": 85.2176,
      "step": 77900
    },
    {
      "epoch": 0.3147662584792156,
      "grad_norm": 746.38134765625,
      "learning_rate": 3.478604221425665e-05,
      "loss": 56.6662,
      "step": 77910
    },
    {
      "epoch": 0.3148066597445832,
      "grad_norm": 805.35498046875,
      "learning_rate": 3.478416165550938e-05,
      "loss": 87.0488,
      "step": 77920
    },
    {
      "epoch": 0.31484706100995086,
      "grad_norm": 634.9224243164062,
      "learning_rate": 3.478228080853726e-05,
      "loss": 63.2272,
      "step": 77930
    },
    {
      "epoch": 0.31488746227531844,
      "grad_norm": 627.2695922851562,
      "learning_rate": 3.478039967337697e-05,
      "loss": 84.5405,
      "step": 77940
    },
    {
      "epoch": 0.3149278635406861,
      "grad_norm": 1291.2822265625,
      "learning_rate": 3.477851825006518e-05,
      "loss": 69.2081,
      "step": 77950
    },
    {
      "epoch": 0.3149682648060537,
      "grad_norm": 0.0,
      "learning_rate": 3.4776636538638565e-05,
      "loss": 58.2442,
      "step": 77960
    },
    {
      "epoch": 0.31500866607142136,
      "grad_norm": 693.0581665039062,
      "learning_rate": 3.477475453913381e-05,
      "loss": 51.7055,
      "step": 77970
    },
    {
      "epoch": 0.315049067336789,
      "grad_norm": 416.64129638671875,
      "learning_rate": 3.477287225158762e-05,
      "loss": 41.2215,
      "step": 77980
    },
    {
      "epoch": 0.31508946860215664,
      "grad_norm": 744.0357666015625,
      "learning_rate": 3.477098967603667e-05,
      "loss": 71.1339,
      "step": 77990
    },
    {
      "epoch": 0.3151298698675242,
      "grad_norm": 2952.7724609375,
      "learning_rate": 3.4769106812517685e-05,
      "loss": 79.9298,
      "step": 78000
    },
    {
      "epoch": 0.31517027113289187,
      "grad_norm": 365.7235412597656,
      "learning_rate": 3.476722366106734e-05,
      "loss": 48.3076,
      "step": 78010
    },
    {
      "epoch": 0.3152106723982595,
      "grad_norm": 806.73876953125,
      "learning_rate": 3.476534022172238e-05,
      "loss": 58.0707,
      "step": 78020
    },
    {
      "epoch": 0.31525107366362715,
      "grad_norm": 520.0816650390625,
      "learning_rate": 3.4763456494519505e-05,
      "loss": 84.0921,
      "step": 78030
    },
    {
      "epoch": 0.3152914749289948,
      "grad_norm": 986.6627197265625,
      "learning_rate": 3.476157247949545e-05,
      "loss": 55.3089,
      "step": 78040
    },
    {
      "epoch": 0.3153318761943624,
      "grad_norm": 754.4058837890625,
      "learning_rate": 3.475968817668694e-05,
      "loss": 69.4051,
      "step": 78050
    },
    {
      "epoch": 0.31537227745973007,
      "grad_norm": 559.5108642578125,
      "learning_rate": 3.47578035861307e-05,
      "loss": 40.0251,
      "step": 78060
    },
    {
      "epoch": 0.31541267872509765,
      "grad_norm": 329.8301086425781,
      "learning_rate": 3.475591870786349e-05,
      "loss": 56.468,
      "step": 78070
    },
    {
      "epoch": 0.3154530799904653,
      "grad_norm": 768.6515502929688,
      "learning_rate": 3.4754033541922054e-05,
      "loss": 57.4969,
      "step": 78080
    },
    {
      "epoch": 0.31549348125583293,
      "grad_norm": 616.1010131835938,
      "learning_rate": 3.475214808834313e-05,
      "loss": 57.8262,
      "step": 78090
    },
    {
      "epoch": 0.31553388252120057,
      "grad_norm": 655.2771606445312,
      "learning_rate": 3.475026234716348e-05,
      "loss": 67.711,
      "step": 78100
    },
    {
      "epoch": 0.3155742837865682,
      "grad_norm": 1178.9937744140625,
      "learning_rate": 3.474837631841988e-05,
      "loss": 81.5335,
      "step": 78110
    },
    {
      "epoch": 0.31561468505193585,
      "grad_norm": 536.2959594726562,
      "learning_rate": 3.474649000214909e-05,
      "loss": 51.0797,
      "step": 78120
    },
    {
      "epoch": 0.31565508631730343,
      "grad_norm": 752.1845092773438,
      "learning_rate": 3.474460339838788e-05,
      "loss": 56.8901,
      "step": 78130
    },
    {
      "epoch": 0.3156954875826711,
      "grad_norm": 634.9712524414062,
      "learning_rate": 3.474271650717303e-05,
      "loss": 77.3995,
      "step": 78140
    },
    {
      "epoch": 0.3157358888480387,
      "grad_norm": 882.2861938476562,
      "learning_rate": 3.474082932854135e-05,
      "loss": 56.3155,
      "step": 78150
    },
    {
      "epoch": 0.31577629011340635,
      "grad_norm": 916.5146484375,
      "learning_rate": 3.47389418625296e-05,
      "loss": 67.5177,
      "step": 78160
    },
    {
      "epoch": 0.315816691378774,
      "grad_norm": 571.06298828125,
      "learning_rate": 3.4737054109174596e-05,
      "loss": 68.8433,
      "step": 78170
    },
    {
      "epoch": 0.31585709264414163,
      "grad_norm": 1100.0029296875,
      "learning_rate": 3.473516606851313e-05,
      "loss": 66.3951,
      "step": 78180
    },
    {
      "epoch": 0.3158974939095092,
      "grad_norm": 924.810791015625,
      "learning_rate": 3.473327774058201e-05,
      "loss": 85.4612,
      "step": 78190
    },
    {
      "epoch": 0.31593789517487686,
      "grad_norm": 1606.62890625,
      "learning_rate": 3.473138912541807e-05,
      "loss": 107.2161,
      "step": 78200
    },
    {
      "epoch": 0.3159782964402445,
      "grad_norm": 969.39892578125,
      "learning_rate": 3.472950022305811e-05,
      "loss": 51.2587,
      "step": 78210
    },
    {
      "epoch": 0.31601869770561214,
      "grad_norm": 520.0035400390625,
      "learning_rate": 3.472761103353895e-05,
      "loss": 109.1551,
      "step": 78220
    },
    {
      "epoch": 0.3160590989709798,
      "grad_norm": 434.693603515625,
      "learning_rate": 3.472572155689744e-05,
      "loss": 66.7653,
      "step": 78230
    },
    {
      "epoch": 0.3160995002363474,
      "grad_norm": 845.1107788085938,
      "learning_rate": 3.4723831793170406e-05,
      "loss": 53.3309,
      "step": 78240
    },
    {
      "epoch": 0.31613990150171506,
      "grad_norm": 435.4742431640625,
      "learning_rate": 3.4721941742394694e-05,
      "loss": 47.1871,
      "step": 78250
    },
    {
      "epoch": 0.31618030276708264,
      "grad_norm": 1057.7083740234375,
      "learning_rate": 3.472005140460714e-05,
      "loss": 51.1896,
      "step": 78260
    },
    {
      "epoch": 0.3162207040324503,
      "grad_norm": 621.1998901367188,
      "learning_rate": 3.471816077984461e-05,
      "loss": 62.4693,
      "step": 78270
    },
    {
      "epoch": 0.3162611052978179,
      "grad_norm": 950.8551635742188,
      "learning_rate": 3.4716269868143956e-05,
      "loss": 79.8929,
      "step": 78280
    },
    {
      "epoch": 0.31630150656318556,
      "grad_norm": 913.15625,
      "learning_rate": 3.4714378669542046e-05,
      "loss": 55.8315,
      "step": 78290
    },
    {
      "epoch": 0.3163419078285532,
      "grad_norm": 1112.23046875,
      "learning_rate": 3.471248718407575e-05,
      "loss": 81.3037,
      "step": 78300
    },
    {
      "epoch": 0.31638230909392084,
      "grad_norm": 645.7435302734375,
      "learning_rate": 3.471059541178194e-05,
      "loss": 45.1448,
      "step": 78310
    },
    {
      "epoch": 0.31642271035928843,
      "grad_norm": 480.93511962890625,
      "learning_rate": 3.4708703352697496e-05,
      "loss": 59.0489,
      "step": 78320
    },
    {
      "epoch": 0.31646311162465607,
      "grad_norm": 1059.6514892578125,
      "learning_rate": 3.4706811006859315e-05,
      "loss": 68.5006,
      "step": 78330
    },
    {
      "epoch": 0.3165035128900237,
      "grad_norm": 1040.41748046875,
      "learning_rate": 3.470491837430428e-05,
      "loss": 95.524,
      "step": 78340
    },
    {
      "epoch": 0.31654391415539135,
      "grad_norm": 352.0108337402344,
      "learning_rate": 3.470302545506929e-05,
      "loss": 49.154,
      "step": 78350
    },
    {
      "epoch": 0.316584315420759,
      "grad_norm": 794.4562377929688,
      "learning_rate": 3.4701132249191245e-05,
      "loss": 69.4265,
      "step": 78360
    },
    {
      "epoch": 0.31662471668612663,
      "grad_norm": 1108.3975830078125,
      "learning_rate": 3.469923875670706e-05,
      "loss": 88.2527,
      "step": 78370
    },
    {
      "epoch": 0.31666511795149427,
      "grad_norm": 754.4819946289062,
      "learning_rate": 3.469734497765365e-05,
      "loss": 74.1756,
      "step": 78380
    },
    {
      "epoch": 0.31670551921686185,
      "grad_norm": 554.4060668945312,
      "learning_rate": 3.469545091206793e-05,
      "loss": 48.7816,
      "step": 78390
    },
    {
      "epoch": 0.3167459204822295,
      "grad_norm": 670.9364013671875,
      "learning_rate": 3.469355655998683e-05,
      "loss": 74.2816,
      "step": 78400
    },
    {
      "epoch": 0.31678632174759713,
      "grad_norm": 548.6668701171875,
      "learning_rate": 3.4691661921447284e-05,
      "loss": 72.3777,
      "step": 78410
    },
    {
      "epoch": 0.3168267230129648,
      "grad_norm": 1143.9090576171875,
      "learning_rate": 3.468976699648622e-05,
      "loss": 70.6955,
      "step": 78420
    },
    {
      "epoch": 0.3168671242783324,
      "grad_norm": 230.33255004882812,
      "learning_rate": 3.4687871785140586e-05,
      "loss": 51.944,
      "step": 78430
    },
    {
      "epoch": 0.31690752554370005,
      "grad_norm": 294.7291259765625,
      "learning_rate": 3.4685976287447326e-05,
      "loss": 59.3842,
      "step": 78440
    },
    {
      "epoch": 0.31694792680906764,
      "grad_norm": 595.3283081054688,
      "learning_rate": 3.4684080503443405e-05,
      "loss": 83.9593,
      "step": 78450
    },
    {
      "epoch": 0.3169883280744353,
      "grad_norm": 1100.2291259765625,
      "learning_rate": 3.468218443316577e-05,
      "loss": 82.4289,
      "step": 78460
    },
    {
      "epoch": 0.3170287293398029,
      "grad_norm": 1326.83740234375,
      "learning_rate": 3.46802880766514e-05,
      "loss": 70.2116,
      "step": 78470
    },
    {
      "epoch": 0.31706913060517056,
      "grad_norm": 594.4410400390625,
      "learning_rate": 3.467839143393724e-05,
      "loss": 69.3761,
      "step": 78480
    },
    {
      "epoch": 0.3171095318705382,
      "grad_norm": 511.62786865234375,
      "learning_rate": 3.46764945050603e-05,
      "loss": 75.8792,
      "step": 78490
    },
    {
      "epoch": 0.31714993313590584,
      "grad_norm": 2378.63134765625,
      "learning_rate": 3.467459729005753e-05,
      "loss": 72.8961,
      "step": 78500
    },
    {
      "epoch": 0.3171903344012734,
      "grad_norm": 1395.0682373046875,
      "learning_rate": 3.467269978896594e-05,
      "loss": 102.6495,
      "step": 78510
    },
    {
      "epoch": 0.31723073566664106,
      "grad_norm": 473.0601501464844,
      "learning_rate": 3.467080200182251e-05,
      "loss": 54.5584,
      "step": 78520
    },
    {
      "epoch": 0.3172711369320087,
      "grad_norm": 349.14813232421875,
      "learning_rate": 3.4668903928664234e-05,
      "loss": 68.1884,
      "step": 78530
    },
    {
      "epoch": 0.31731153819737634,
      "grad_norm": 932.8010864257812,
      "learning_rate": 3.4667005569528134e-05,
      "loss": 48.9466,
      "step": 78540
    },
    {
      "epoch": 0.317351939462744,
      "grad_norm": 1845.6824951171875,
      "learning_rate": 3.466510692445121e-05,
      "loss": 62.2658,
      "step": 78550
    },
    {
      "epoch": 0.3173923407281116,
      "grad_norm": 1455.640380859375,
      "learning_rate": 3.4663207993470466e-05,
      "loss": 56.2996,
      "step": 78560
    },
    {
      "epoch": 0.31743274199347926,
      "grad_norm": 447.1221618652344,
      "learning_rate": 3.466130877662294e-05,
      "loss": 48.1869,
      "step": 78570
    },
    {
      "epoch": 0.31747314325884685,
      "grad_norm": 901.009521484375,
      "learning_rate": 3.465940927394565e-05,
      "loss": 68.94,
      "step": 78580
    },
    {
      "epoch": 0.3175135445242145,
      "grad_norm": 1162.469482421875,
      "learning_rate": 3.465750948547563e-05,
      "loss": 56.6787,
      "step": 78590
    },
    {
      "epoch": 0.3175539457895821,
      "grad_norm": 1017.1430053710938,
      "learning_rate": 3.465560941124992e-05,
      "loss": 42.6346,
      "step": 78600
    },
    {
      "epoch": 0.31759434705494977,
      "grad_norm": 474.6357727050781,
      "learning_rate": 3.4653709051305546e-05,
      "loss": 45.265,
      "step": 78610
    },
    {
      "epoch": 0.3176347483203174,
      "grad_norm": 772.208251953125,
      "learning_rate": 3.465180840567958e-05,
      "loss": 60.2214,
      "step": 78620
    },
    {
      "epoch": 0.31767514958568505,
      "grad_norm": 716.8950805664062,
      "learning_rate": 3.4649907474409074e-05,
      "loss": 78.6945,
      "step": 78630
    },
    {
      "epoch": 0.31771555085105263,
      "grad_norm": 782.5782470703125,
      "learning_rate": 3.464800625753107e-05,
      "loss": 55.895,
      "step": 78640
    },
    {
      "epoch": 0.31775595211642027,
      "grad_norm": 1307.5980224609375,
      "learning_rate": 3.464610475508264e-05,
      "loss": 85.9435,
      "step": 78650
    },
    {
      "epoch": 0.3177963533817879,
      "grad_norm": 625.3878784179688,
      "learning_rate": 3.464420296710086e-05,
      "loss": 45.7682,
      "step": 78660
    },
    {
      "epoch": 0.31783675464715555,
      "grad_norm": 662.2884521484375,
      "learning_rate": 3.46423008936228e-05,
      "loss": 75.9776,
      "step": 78670
    },
    {
      "epoch": 0.3178771559125232,
      "grad_norm": 797.3787841796875,
      "learning_rate": 3.464039853468555e-05,
      "loss": 93.6652,
      "step": 78680
    },
    {
      "epoch": 0.31791755717789083,
      "grad_norm": 2007.70947265625,
      "learning_rate": 3.4638495890326194e-05,
      "loss": 119.6598,
      "step": 78690
    },
    {
      "epoch": 0.31795795844325847,
      "grad_norm": 929.6837768554688,
      "learning_rate": 3.4636592960581825e-05,
      "loss": 60.2757,
      "step": 78700
    },
    {
      "epoch": 0.31799835970862605,
      "grad_norm": 1003.391845703125,
      "learning_rate": 3.463468974548954e-05,
      "loss": 55.3076,
      "step": 78710
    },
    {
      "epoch": 0.3180387609739937,
      "grad_norm": 664.688720703125,
      "learning_rate": 3.463278624508644e-05,
      "loss": 73.8371,
      "step": 78720
    },
    {
      "epoch": 0.31807916223936133,
      "grad_norm": 759.0806884765625,
      "learning_rate": 3.463088245940965e-05,
      "loss": 113.9012,
      "step": 78730
    },
    {
      "epoch": 0.318119563504729,
      "grad_norm": 765.0343017578125,
      "learning_rate": 3.4628978388496266e-05,
      "loss": 54.457,
      "step": 78740
    },
    {
      "epoch": 0.3181599647700966,
      "grad_norm": 542.20654296875,
      "learning_rate": 3.462707403238341e-05,
      "loss": 62.4326,
      "step": 78750
    },
    {
      "epoch": 0.31820036603546425,
      "grad_norm": 0.0,
      "learning_rate": 3.4625169391108224e-05,
      "loss": 59.6151,
      "step": 78760
    },
    {
      "epoch": 0.31824076730083184,
      "grad_norm": 823.3811645507812,
      "learning_rate": 3.4623264464707834e-05,
      "loss": 64.7899,
      "step": 78770
    },
    {
      "epoch": 0.3182811685661995,
      "grad_norm": 1177.2208251953125,
      "learning_rate": 3.462135925321937e-05,
      "loss": 84.4173,
      "step": 78780
    },
    {
      "epoch": 0.3183215698315671,
      "grad_norm": 422.4059143066406,
      "learning_rate": 3.461945375667998e-05,
      "loss": 84.0472,
      "step": 78790
    },
    {
      "epoch": 0.31836197109693476,
      "grad_norm": 1313.7181396484375,
      "learning_rate": 3.461754797512681e-05,
      "loss": 72.8415,
      "step": 78800
    },
    {
      "epoch": 0.3184023723623024,
      "grad_norm": 672.3056640625,
      "learning_rate": 3.4615641908597016e-05,
      "loss": 51.1723,
      "step": 78810
    },
    {
      "epoch": 0.31844277362767004,
      "grad_norm": 493.34539794921875,
      "learning_rate": 3.461373555712776e-05,
      "loss": 61.0822,
      "step": 78820
    },
    {
      "epoch": 0.3184831748930376,
      "grad_norm": 1001.4664916992188,
      "learning_rate": 3.4611828920756204e-05,
      "loss": 47.7915,
      "step": 78830
    },
    {
      "epoch": 0.31852357615840526,
      "grad_norm": 1193.3172607421875,
      "learning_rate": 3.4609921999519525e-05,
      "loss": 91.9756,
      "step": 78840
    },
    {
      "epoch": 0.3185639774237729,
      "grad_norm": 861.0563354492188,
      "learning_rate": 3.46080147934549e-05,
      "loss": 56.0792,
      "step": 78850
    },
    {
      "epoch": 0.31860437868914054,
      "grad_norm": 358.73297119140625,
      "learning_rate": 3.46061073025995e-05,
      "loss": 51.1715,
      "step": 78860
    },
    {
      "epoch": 0.3186447799545082,
      "grad_norm": 549.15673828125,
      "learning_rate": 3.4604199526990514e-05,
      "loss": 81.9061,
      "step": 78870
    },
    {
      "epoch": 0.3186851812198758,
      "grad_norm": 428.2076721191406,
      "learning_rate": 3.460229146666514e-05,
      "loss": 72.709,
      "step": 78880
    },
    {
      "epoch": 0.31872558248524346,
      "grad_norm": 793.41748046875,
      "learning_rate": 3.460038312166058e-05,
      "loss": 61.257,
      "step": 78890
    },
    {
      "epoch": 0.31876598375061105,
      "grad_norm": 425.1347351074219,
      "learning_rate": 3.4598474492014036e-05,
      "loss": 67.8388,
      "step": 78900
    },
    {
      "epoch": 0.3188063850159787,
      "grad_norm": 1619.48876953125,
      "learning_rate": 3.459656557776271e-05,
      "loss": 67.4973,
      "step": 78910
    },
    {
      "epoch": 0.31884678628134633,
      "grad_norm": 809.5908203125,
      "learning_rate": 3.459465637894383e-05,
      "loss": 64.2227,
      "step": 78920
    },
    {
      "epoch": 0.31888718754671397,
      "grad_norm": 1020.9970092773438,
      "learning_rate": 3.4592746895594604e-05,
      "loss": 53.4182,
      "step": 78930
    },
    {
      "epoch": 0.3189275888120816,
      "grad_norm": 736.4168090820312,
      "learning_rate": 3.459083712775226e-05,
      "loss": 68.4549,
      "step": 78940
    },
    {
      "epoch": 0.31896799007744925,
      "grad_norm": 611.6744384765625,
      "learning_rate": 3.458892707545405e-05,
      "loss": 82.075,
      "step": 78950
    },
    {
      "epoch": 0.31900839134281683,
      "grad_norm": 786.9500732421875,
      "learning_rate": 3.4587016738737186e-05,
      "loss": 59.6807,
      "step": 78960
    },
    {
      "epoch": 0.3190487926081845,
      "grad_norm": 669.5952758789062,
      "learning_rate": 3.4585106117638916e-05,
      "loss": 42.0674,
      "step": 78970
    },
    {
      "epoch": 0.3190891938735521,
      "grad_norm": 1109.5372314453125,
      "learning_rate": 3.45831952121965e-05,
      "loss": 89.6373,
      "step": 78980
    },
    {
      "epoch": 0.31912959513891975,
      "grad_norm": 1470.2607421875,
      "learning_rate": 3.458128402244719e-05,
      "loss": 75.5235,
      "step": 78990
    },
    {
      "epoch": 0.3191699964042874,
      "grad_norm": 543.140625,
      "learning_rate": 3.457937254842823e-05,
      "loss": 60.188,
      "step": 79000
    },
    {
      "epoch": 0.31921039766965503,
      "grad_norm": 932.54296875,
      "learning_rate": 3.457746079017691e-05,
      "loss": 48.5002,
      "step": 79010
    },
    {
      "epoch": 0.3192507989350226,
      "grad_norm": 661.5245971679688,
      "learning_rate": 3.457554874773047e-05,
      "loss": 84.905,
      "step": 79020
    },
    {
      "epoch": 0.31929120020039026,
      "grad_norm": 591.1777954101562,
      "learning_rate": 3.457363642112622e-05,
      "loss": 67.2805,
      "step": 79030
    },
    {
      "epoch": 0.3193316014657579,
      "grad_norm": 910.734130859375,
      "learning_rate": 3.457172381040141e-05,
      "loss": 98.3831,
      "step": 79040
    },
    {
      "epoch": 0.31937200273112554,
      "grad_norm": 733.234375,
      "learning_rate": 3.4569810915593356e-05,
      "loss": 74.8495,
      "step": 79050
    },
    {
      "epoch": 0.3194124039964932,
      "grad_norm": 597.4061889648438,
      "learning_rate": 3.456789773673933e-05,
      "loss": 41.353,
      "step": 79060
    },
    {
      "epoch": 0.3194528052618608,
      "grad_norm": 466.7047424316406,
      "learning_rate": 3.4565984273876635e-05,
      "loss": 56.502,
      "step": 79070
    },
    {
      "epoch": 0.31949320652722846,
      "grad_norm": 1494.1431884765625,
      "learning_rate": 3.456407052704258e-05,
      "loss": 70.4316,
      "step": 79080
    },
    {
      "epoch": 0.31953360779259604,
      "grad_norm": 761.2359008789062,
      "learning_rate": 3.456215649627447e-05,
      "loss": 52.0162,
      "step": 79090
    },
    {
      "epoch": 0.3195740090579637,
      "grad_norm": 988.449951171875,
      "learning_rate": 3.456024218160963e-05,
      "loss": 57.3673,
      "step": 79100
    },
    {
      "epoch": 0.3196144103233313,
      "grad_norm": 629.7996215820312,
      "learning_rate": 3.455832758308536e-05,
      "loss": 91.0117,
      "step": 79110
    },
    {
      "epoch": 0.31965481158869896,
      "grad_norm": 314.0284118652344,
      "learning_rate": 3.455641270073901e-05,
      "loss": 30.0743,
      "step": 79120
    },
    {
      "epoch": 0.3196952128540666,
      "grad_norm": 524.0881958007812,
      "learning_rate": 3.4554497534607895e-05,
      "loss": 74.0941,
      "step": 79130
    },
    {
      "epoch": 0.31973561411943424,
      "grad_norm": 798.7915649414062,
      "learning_rate": 3.455258208472936e-05,
      "loss": 69.5489,
      "step": 79140
    },
    {
      "epoch": 0.3197760153848018,
      "grad_norm": 1003.20361328125,
      "learning_rate": 3.455066635114074e-05,
      "loss": 66.3634,
      "step": 79150
    },
    {
      "epoch": 0.31981641665016947,
      "grad_norm": 757.2113037109375,
      "learning_rate": 3.4548750333879395e-05,
      "loss": 73.0947,
      "step": 79160
    },
    {
      "epoch": 0.3198568179155371,
      "grad_norm": 2536.526611328125,
      "learning_rate": 3.454683403298266e-05,
      "loss": 107.1402,
      "step": 79170
    },
    {
      "epoch": 0.31989721918090475,
      "grad_norm": 1102.933837890625,
      "learning_rate": 3.4544917448487915e-05,
      "loss": 40.2471,
      "step": 79180
    },
    {
      "epoch": 0.3199376204462724,
      "grad_norm": 523.2879638671875,
      "learning_rate": 3.454300058043252e-05,
      "loss": 56.0137,
      "step": 79190
    },
    {
      "epoch": 0.31997802171164,
      "grad_norm": 1049.5115966796875,
      "learning_rate": 3.454108342885384e-05,
      "loss": 88.8176,
      "step": 79200
    },
    {
      "epoch": 0.32001842297700767,
      "grad_norm": 1003.3671264648438,
      "learning_rate": 3.453916599378925e-05,
      "loss": 62.6265,
      "step": 79210
    },
    {
      "epoch": 0.32005882424237525,
      "grad_norm": 1479.1932373046875,
      "learning_rate": 3.453724827527613e-05,
      "loss": 86.5033,
      "step": 79220
    },
    {
      "epoch": 0.3200992255077429,
      "grad_norm": 793.9293823242188,
      "learning_rate": 3.453533027335188e-05,
      "loss": 72.8704,
      "step": 79230
    },
    {
      "epoch": 0.32013962677311053,
      "grad_norm": 738.2943725585938,
      "learning_rate": 3.453341198805388e-05,
      "loss": 69.7305,
      "step": 79240
    },
    {
      "epoch": 0.32018002803847817,
      "grad_norm": 609.8193969726562,
      "learning_rate": 3.4531493419419525e-05,
      "loss": 63.8748,
      "step": 79250
    },
    {
      "epoch": 0.3202204293038458,
      "grad_norm": 747.0555419921875,
      "learning_rate": 3.452957456748622e-05,
      "loss": 72.1068,
      "step": 79260
    },
    {
      "epoch": 0.32026083056921345,
      "grad_norm": 1100.9898681640625,
      "learning_rate": 3.4527655432291384e-05,
      "loss": 87.2401,
      "step": 79270
    },
    {
      "epoch": 0.32030123183458103,
      "grad_norm": 874.8444213867188,
      "learning_rate": 3.4525736013872433e-05,
      "loss": 64.6822,
      "step": 79280
    },
    {
      "epoch": 0.3203416330999487,
      "grad_norm": 552.16015625,
      "learning_rate": 3.4523816312266773e-05,
      "loss": 57.4605,
      "step": 79290
    },
    {
      "epoch": 0.3203820343653163,
      "grad_norm": 5363.94482421875,
      "learning_rate": 3.4521896327511836e-05,
      "loss": 94.8203,
      "step": 79300
    },
    {
      "epoch": 0.32042243563068395,
      "grad_norm": 499.93145751953125,
      "learning_rate": 3.451997605964506e-05,
      "loss": 96.6472,
      "step": 79310
    },
    {
      "epoch": 0.3204628368960516,
      "grad_norm": 1525.2781982421875,
      "learning_rate": 3.451805550870387e-05,
      "loss": 46.9949,
      "step": 79320
    },
    {
      "epoch": 0.32050323816141923,
      "grad_norm": 871.80322265625,
      "learning_rate": 3.4516134674725723e-05,
      "loss": 63.2105,
      "step": 79330
    },
    {
      "epoch": 0.3205436394267868,
      "grad_norm": 562.4641723632812,
      "learning_rate": 3.4514213557748046e-05,
      "loss": 74.4778,
      "step": 79340
    },
    {
      "epoch": 0.32058404069215446,
      "grad_norm": 942.83447265625,
      "learning_rate": 3.4512292157808306e-05,
      "loss": 81.0777,
      "step": 79350
    },
    {
      "epoch": 0.3206244419575221,
      "grad_norm": 1135.0953369140625,
      "learning_rate": 3.4510370474943956e-05,
      "loss": 62.4802,
      "step": 79360
    },
    {
      "epoch": 0.32066484322288974,
      "grad_norm": 766.571044921875,
      "learning_rate": 3.450844850919247e-05,
      "loss": 65.9881,
      "step": 79370
    },
    {
      "epoch": 0.3207052444882574,
      "grad_norm": 927.8790283203125,
      "learning_rate": 3.450652626059131e-05,
      "loss": 67.4498,
      "step": 79380
    },
    {
      "epoch": 0.320745645753625,
      "grad_norm": 1346.9913330078125,
      "learning_rate": 3.4504603729177945e-05,
      "loss": 71.3516,
      "step": 79390
    },
    {
      "epoch": 0.32078604701899266,
      "grad_norm": 595.7301635742188,
      "learning_rate": 3.450268091498987e-05,
      "loss": 88.8037,
      "step": 79400
    },
    {
      "epoch": 0.32082644828436024,
      "grad_norm": 913.6329956054688,
      "learning_rate": 3.4500757818064565e-05,
      "loss": 65.262,
      "step": 79410
    },
    {
      "epoch": 0.3208668495497279,
      "grad_norm": 1271.197509765625,
      "learning_rate": 3.4498834438439516e-05,
      "loss": 103.464,
      "step": 79420
    },
    {
      "epoch": 0.3209072508150955,
      "grad_norm": 891.8692626953125,
      "learning_rate": 3.4496910776152226e-05,
      "loss": 75.6412,
      "step": 79430
    },
    {
      "epoch": 0.32094765208046316,
      "grad_norm": 835.5440673828125,
      "learning_rate": 3.44949868312402e-05,
      "loss": 80.8117,
      "step": 79440
    },
    {
      "epoch": 0.3209880533458308,
      "grad_norm": 821.7725219726562,
      "learning_rate": 3.4493062603740934e-05,
      "loss": 81.0348,
      "step": 79450
    },
    {
      "epoch": 0.32102845461119844,
      "grad_norm": 579.9627685546875,
      "learning_rate": 3.449113809369196e-05,
      "loss": 77.9407,
      "step": 79460
    },
    {
      "epoch": 0.32106885587656603,
      "grad_norm": 436.1594543457031,
      "learning_rate": 3.448921330113079e-05,
      "loss": 35.9948,
      "step": 79470
    },
    {
      "epoch": 0.32110925714193367,
      "grad_norm": 500.90777587890625,
      "learning_rate": 3.448728822609494e-05,
      "loss": 75.6861,
      "step": 79480
    },
    {
      "epoch": 0.3211496584073013,
      "grad_norm": 808.7592163085938,
      "learning_rate": 3.448536286862195e-05,
      "loss": 65.08,
      "step": 79490
    },
    {
      "epoch": 0.32119005967266895,
      "grad_norm": 428.0213317871094,
      "learning_rate": 3.4483437228749356e-05,
      "loss": 62.6515,
      "step": 79500
    },
    {
      "epoch": 0.3212304609380366,
      "grad_norm": 669.585693359375,
      "learning_rate": 3.448151130651469e-05,
      "loss": 66.1477,
      "step": 79510
    },
    {
      "epoch": 0.32127086220340423,
      "grad_norm": 461.6833190917969,
      "learning_rate": 3.4479585101955506e-05,
      "loss": 72.4844,
      "step": 79520
    },
    {
      "epoch": 0.32131126346877187,
      "grad_norm": 0.0,
      "learning_rate": 3.4477658615109365e-05,
      "loss": 71.0492,
      "step": 79530
    },
    {
      "epoch": 0.32135166473413945,
      "grad_norm": 636.3035888671875,
      "learning_rate": 3.447573184601381e-05,
      "loss": 55.4109,
      "step": 79540
    },
    {
      "epoch": 0.3213920659995071,
      "grad_norm": 1040.3365478515625,
      "learning_rate": 3.447380479470641e-05,
      "loss": 64.308,
      "step": 79550
    },
    {
      "epoch": 0.32143246726487473,
      "grad_norm": 1166.4844970703125,
      "learning_rate": 3.4471877461224735e-05,
      "loss": 49.8759,
      "step": 79560
    },
    {
      "epoch": 0.3214728685302424,
      "grad_norm": 618.626708984375,
      "learning_rate": 3.446994984560636e-05,
      "loss": 48.319,
      "step": 79570
    },
    {
      "epoch": 0.32151326979561,
      "grad_norm": 990.6760864257812,
      "learning_rate": 3.4468021947888855e-05,
      "loss": 68.7862,
      "step": 79580
    },
    {
      "epoch": 0.32155367106097765,
      "grad_norm": 417.9886474609375,
      "learning_rate": 3.4466093768109825e-05,
      "loss": 58.8706,
      "step": 79590
    },
    {
      "epoch": 0.32159407232634524,
      "grad_norm": 575.5824584960938,
      "learning_rate": 3.4464165306306845e-05,
      "loss": 73.3827,
      "step": 79600
    },
    {
      "epoch": 0.3216344735917129,
      "grad_norm": 943.4527587890625,
      "learning_rate": 3.446223656251751e-05,
      "loss": 56.8044,
      "step": 79610
    },
    {
      "epoch": 0.3216748748570805,
      "grad_norm": 383.214599609375,
      "learning_rate": 3.4460307536779434e-05,
      "loss": 55.9928,
      "step": 79620
    },
    {
      "epoch": 0.32171527612244816,
      "grad_norm": 960.987548828125,
      "learning_rate": 3.4458378229130214e-05,
      "loss": 54.2587,
      "step": 79630
    },
    {
      "epoch": 0.3217556773878158,
      "grad_norm": 720.4596557617188,
      "learning_rate": 3.4456448639607476e-05,
      "loss": 60.2574,
      "step": 79640
    },
    {
      "epoch": 0.32179607865318344,
      "grad_norm": 860.6263427734375,
      "learning_rate": 3.4454518768248816e-05,
      "loss": 70.9943,
      "step": 79650
    },
    {
      "epoch": 0.321836479918551,
      "grad_norm": 1033.0247802734375,
      "learning_rate": 3.445258861509188e-05,
      "loss": 82.2327,
      "step": 79660
    },
    {
      "epoch": 0.32187688118391866,
      "grad_norm": 739.957763671875,
      "learning_rate": 3.4450658180174286e-05,
      "loss": 43.3071,
      "step": 79670
    },
    {
      "epoch": 0.3219172824492863,
      "grad_norm": 422.6085510253906,
      "learning_rate": 3.4448727463533666e-05,
      "loss": 46.9807,
      "step": 79680
    },
    {
      "epoch": 0.32195768371465394,
      "grad_norm": 470.52569580078125,
      "learning_rate": 3.4446796465207665e-05,
      "loss": 84.7386,
      "step": 79690
    },
    {
      "epoch": 0.3219980849800216,
      "grad_norm": 333.7068786621094,
      "learning_rate": 3.444486518523394e-05,
      "loss": 59.9149,
      "step": 79700
    },
    {
      "epoch": 0.3220384862453892,
      "grad_norm": 1886.801513671875,
      "learning_rate": 3.4442933623650124e-05,
      "loss": 55.9101,
      "step": 79710
    },
    {
      "epoch": 0.32207888751075686,
      "grad_norm": 725.7935180664062,
      "learning_rate": 3.444100178049389e-05,
      "loss": 62.0386,
      "step": 79720
    },
    {
      "epoch": 0.32211928877612445,
      "grad_norm": 519.2236328125,
      "learning_rate": 3.4439069655802875e-05,
      "loss": 43.5637,
      "step": 79730
    },
    {
      "epoch": 0.3221596900414921,
      "grad_norm": 915.39208984375,
      "learning_rate": 3.443713724961478e-05,
      "loss": 77.7298,
      "step": 79740
    },
    {
      "epoch": 0.3222000913068597,
      "grad_norm": 1126.2271728515625,
      "learning_rate": 3.4435204561967244e-05,
      "loss": 45.7563,
      "step": 79750
    },
    {
      "epoch": 0.32224049257222737,
      "grad_norm": 1261.709228515625,
      "learning_rate": 3.443327159289798e-05,
      "loss": 79.6718,
      "step": 79760
    },
    {
      "epoch": 0.322280893837595,
      "grad_norm": 1179.6519775390625,
      "learning_rate": 3.443133834244465e-05,
      "loss": 71.1249,
      "step": 79770
    },
    {
      "epoch": 0.32232129510296265,
      "grad_norm": 279.9851989746094,
      "learning_rate": 3.4429404810644944e-05,
      "loss": 47.8068,
      "step": 79780
    },
    {
      "epoch": 0.32236169636833023,
      "grad_norm": 411.3631591796875,
      "learning_rate": 3.4427470997536567e-05,
      "loss": 60.9555,
      "step": 79790
    },
    {
      "epoch": 0.32240209763369787,
      "grad_norm": 423.3763732910156,
      "learning_rate": 3.442553690315722e-05,
      "loss": 72.2827,
      "step": 79800
    },
    {
      "epoch": 0.3224424988990655,
      "grad_norm": 814.5885620117188,
      "learning_rate": 3.4423602527544594e-05,
      "loss": 96.0732,
      "step": 79810
    },
    {
      "epoch": 0.32248290016443315,
      "grad_norm": 434.79327392578125,
      "learning_rate": 3.442166787073642e-05,
      "loss": 36.7295,
      "step": 79820
    },
    {
      "epoch": 0.3225233014298008,
      "grad_norm": 317.3850402832031,
      "learning_rate": 3.4419732932770394e-05,
      "loss": 49.3777,
      "step": 79830
    },
    {
      "epoch": 0.32256370269516843,
      "grad_norm": 637.21435546875,
      "learning_rate": 3.441779771368426e-05,
      "loss": 56.3814,
      "step": 79840
    },
    {
      "epoch": 0.32260410396053607,
      "grad_norm": 654.25927734375,
      "learning_rate": 3.4415862213515735e-05,
      "loss": 45.2075,
      "step": 79850
    },
    {
      "epoch": 0.32264450522590365,
      "grad_norm": 1865.554931640625,
      "learning_rate": 3.4413926432302554e-05,
      "loss": 70.6747,
      "step": 79860
    },
    {
      "epoch": 0.3226849064912713,
      "grad_norm": 845.0151977539062,
      "learning_rate": 3.441199037008246e-05,
      "loss": 53.3493,
      "step": 79870
    },
    {
      "epoch": 0.32272530775663893,
      "grad_norm": 0.0,
      "learning_rate": 3.441005402689319e-05,
      "loss": 50.3135,
      "step": 79880
    },
    {
      "epoch": 0.3227657090220066,
      "grad_norm": 726.01123046875,
      "learning_rate": 3.4408117402772494e-05,
      "loss": 107.5813,
      "step": 79890
    },
    {
      "epoch": 0.3228061102873742,
      "grad_norm": 622.0538940429688,
      "learning_rate": 3.440618049775814e-05,
      "loss": 45.5656,
      "step": 79900
    },
    {
      "epoch": 0.32284651155274185,
      "grad_norm": 610.9111328125,
      "learning_rate": 3.440424331188788e-05,
      "loss": 52.3628,
      "step": 79910
    },
    {
      "epoch": 0.32288691281810944,
      "grad_norm": 655.2313232421875,
      "learning_rate": 3.4402305845199475e-05,
      "loss": 93.0792,
      "step": 79920
    },
    {
      "epoch": 0.3229273140834771,
      "grad_norm": 739.9714965820312,
      "learning_rate": 3.4400368097730705e-05,
      "loss": 51.3259,
      "step": 79930
    },
    {
      "epoch": 0.3229677153488447,
      "grad_norm": 1402.1278076171875,
      "learning_rate": 3.439843006951935e-05,
      "loss": 84.88,
      "step": 79940
    },
    {
      "epoch": 0.32300811661421236,
      "grad_norm": 383.4258728027344,
      "learning_rate": 3.439649176060318e-05,
      "loss": 73.4971,
      "step": 79950
    },
    {
      "epoch": 0.32304851787958,
      "grad_norm": 553.3306274414062,
      "learning_rate": 3.439455317102e-05,
      "loss": 64.7679,
      "step": 79960
    },
    {
      "epoch": 0.32308891914494764,
      "grad_norm": 677.3023071289062,
      "learning_rate": 3.439261430080759e-05,
      "loss": 77.7907,
      "step": 79970
    },
    {
      "epoch": 0.3231293204103152,
      "grad_norm": 1300.3046875,
      "learning_rate": 3.439067515000375e-05,
      "loss": 60.2865,
      "step": 79980
    },
    {
      "epoch": 0.32316972167568286,
      "grad_norm": 734.296142578125,
      "learning_rate": 3.4388735718646294e-05,
      "loss": 94.4354,
      "step": 79990
    },
    {
      "epoch": 0.3232101229410505,
      "grad_norm": 543.9745483398438,
      "learning_rate": 3.438679600677303e-05,
      "loss": 43.9692,
      "step": 80000
    },
    {
      "epoch": 0.32325052420641814,
      "grad_norm": 590.18994140625,
      "learning_rate": 3.438485601442176e-05,
      "loss": 83.3607,
      "step": 80010
    },
    {
      "epoch": 0.3232909254717858,
      "grad_norm": 840.5568237304688,
      "learning_rate": 3.438291574163032e-05,
      "loss": 46.152,
      "step": 80020
    },
    {
      "epoch": 0.3233313267371534,
      "grad_norm": 746.9127197265625,
      "learning_rate": 3.4380975188436547e-05,
      "loss": 66.6191,
      "step": 80030
    },
    {
      "epoch": 0.32337172800252106,
      "grad_norm": 1017.9415893554688,
      "learning_rate": 3.437903435487825e-05,
      "loss": 65.5659,
      "step": 80040
    },
    {
      "epoch": 0.32341212926788865,
      "grad_norm": 1068.95654296875,
      "learning_rate": 3.437709324099327e-05,
      "loss": 67.7499,
      "step": 80050
    },
    {
      "epoch": 0.3234525305332563,
      "grad_norm": 460.35491943359375,
      "learning_rate": 3.4375151846819456e-05,
      "loss": 60.3891,
      "step": 80060
    },
    {
      "epoch": 0.32349293179862393,
      "grad_norm": 750.3103637695312,
      "learning_rate": 3.437321017239466e-05,
      "loss": 71.0767,
      "step": 80070
    },
    {
      "epoch": 0.32353333306399157,
      "grad_norm": 514.40234375,
      "learning_rate": 3.4371268217756734e-05,
      "loss": 86.9577,
      "step": 80080
    },
    {
      "epoch": 0.3235737343293592,
      "grad_norm": 784.3499145507812,
      "learning_rate": 3.4369325982943536e-05,
      "loss": 41.5145,
      "step": 80090
    },
    {
      "epoch": 0.32361413559472685,
      "grad_norm": 993.7069702148438,
      "learning_rate": 3.4367383467992926e-05,
      "loss": 67.2231,
      "step": 80100
    },
    {
      "epoch": 0.32365453686009443,
      "grad_norm": 269.11529541015625,
      "learning_rate": 3.436544067294278e-05,
      "loss": 60.3321,
      "step": 80110
    },
    {
      "epoch": 0.3236949381254621,
      "grad_norm": 1333.7032470703125,
      "learning_rate": 3.436349759783097e-05,
      "loss": 80.0917,
      "step": 80120
    },
    {
      "epoch": 0.3237353393908297,
      "grad_norm": 642.1954956054688,
      "learning_rate": 3.4361554242695384e-05,
      "loss": 54.0837,
      "step": 80130
    },
    {
      "epoch": 0.32377574065619735,
      "grad_norm": 1311.14453125,
      "learning_rate": 3.43596106075739e-05,
      "loss": 88.6644,
      "step": 80140
    },
    {
      "epoch": 0.323816141921565,
      "grad_norm": 981.5884399414062,
      "learning_rate": 3.4357666692504415e-05,
      "loss": 66.0389,
      "step": 80150
    },
    {
      "epoch": 0.32385654318693263,
      "grad_norm": 673.2181396484375,
      "learning_rate": 3.4355722497524826e-05,
      "loss": 45.9753,
      "step": 80160
    },
    {
      "epoch": 0.3238969444523003,
      "grad_norm": 325.1015930175781,
      "learning_rate": 3.4353778022673035e-05,
      "loss": 65.172,
      "step": 80170
    },
    {
      "epoch": 0.32393734571766786,
      "grad_norm": 2092.246826171875,
      "learning_rate": 3.4351833267986956e-05,
      "loss": 87.7155,
      "step": 80180
    },
    {
      "epoch": 0.3239777469830355,
      "grad_norm": 422.2587585449219,
      "learning_rate": 3.434988823350449e-05,
      "loss": 73.1039,
      "step": 80190
    },
    {
      "epoch": 0.32401814824840314,
      "grad_norm": 947.8048706054688,
      "learning_rate": 3.434794291926358e-05,
      "loss": 83.3786,
      "step": 80200
    },
    {
      "epoch": 0.3240585495137708,
      "grad_norm": 737.5042114257812,
      "learning_rate": 3.434599732530212e-05,
      "loss": 52.9744,
      "step": 80210
    },
    {
      "epoch": 0.3240989507791384,
      "grad_norm": 1040.675048828125,
      "learning_rate": 3.434405145165807e-05,
      "loss": 88.4331,
      "step": 80220
    },
    {
      "epoch": 0.32413935204450606,
      "grad_norm": 748.6989135742188,
      "learning_rate": 3.434210529836934e-05,
      "loss": 75.7019,
      "step": 80230
    },
    {
      "epoch": 0.32417975330987364,
      "grad_norm": 292.437255859375,
      "learning_rate": 3.434015886547389e-05,
      "loss": 50.4198,
      "step": 80240
    },
    {
      "epoch": 0.3242201545752413,
      "grad_norm": 574.4251708984375,
      "learning_rate": 3.433821215300966e-05,
      "loss": 57.2804,
      "step": 80250
    },
    {
      "epoch": 0.3242605558406089,
      "grad_norm": 997.3341064453125,
      "learning_rate": 3.4336265161014596e-05,
      "loss": 65.6375,
      "step": 80260
    },
    {
      "epoch": 0.32430095710597656,
      "grad_norm": 617.52392578125,
      "learning_rate": 3.433431788952667e-05,
      "loss": 57.1869,
      "step": 80270
    },
    {
      "epoch": 0.3243413583713442,
      "grad_norm": 1012.9528198242188,
      "learning_rate": 3.433237033858384e-05,
      "loss": 90.5776,
      "step": 80280
    },
    {
      "epoch": 0.32438175963671184,
      "grad_norm": 616.6028442382812,
      "learning_rate": 3.4330422508224064e-05,
      "loss": 50.6281,
      "step": 80290
    },
    {
      "epoch": 0.3244221609020794,
      "grad_norm": 867.1862182617188,
      "learning_rate": 3.432847439848532e-05,
      "loss": 57.7962,
      "step": 80300
    },
    {
      "epoch": 0.32446256216744707,
      "grad_norm": 295.4292907714844,
      "learning_rate": 3.43265260094056e-05,
      "loss": 71.0389,
      "step": 80310
    },
    {
      "epoch": 0.3245029634328147,
      "grad_norm": 436.0542297363281,
      "learning_rate": 3.432457734102287e-05,
      "loss": 59.4424,
      "step": 80320
    },
    {
      "epoch": 0.32454336469818235,
      "grad_norm": 1088.5181884765625,
      "learning_rate": 3.4322628393375144e-05,
      "loss": 64.7318,
      "step": 80330
    },
    {
      "epoch": 0.32458376596355,
      "grad_norm": 441.3780517578125,
      "learning_rate": 3.4320679166500386e-05,
      "loss": 57.3388,
      "step": 80340
    },
    {
      "epoch": 0.3246241672289176,
      "grad_norm": 413.42926025390625,
      "learning_rate": 3.4318729660436624e-05,
      "loss": 108.3097,
      "step": 80350
    },
    {
      "epoch": 0.32466456849428527,
      "grad_norm": 669.157958984375,
      "learning_rate": 3.431677987522186e-05,
      "loss": 65.8818,
      "step": 80360
    },
    {
      "epoch": 0.32470496975965285,
      "grad_norm": 1144.414794921875,
      "learning_rate": 3.4314829810894095e-05,
      "loss": 59.8718,
      "step": 80370
    },
    {
      "epoch": 0.3247453710250205,
      "grad_norm": 467.1025085449219,
      "learning_rate": 3.431287946749136e-05,
      "loss": 56.5579,
      "step": 80380
    },
    {
      "epoch": 0.32478577229038813,
      "grad_norm": 852.761474609375,
      "learning_rate": 3.4310928845051656e-05,
      "loss": 58.1927,
      "step": 80390
    },
    {
      "epoch": 0.32482617355575577,
      "grad_norm": 1139.1510009765625,
      "learning_rate": 3.430897794361304e-05,
      "loss": 68.5792,
      "step": 80400
    },
    {
      "epoch": 0.3248665748211234,
      "grad_norm": 645.5872802734375,
      "learning_rate": 3.430702676321353e-05,
      "loss": 40.7631,
      "step": 80410
    },
    {
      "epoch": 0.32490697608649105,
      "grad_norm": 466.1105041503906,
      "learning_rate": 3.4305075303891165e-05,
      "loss": 73.6079,
      "step": 80420
    },
    {
      "epoch": 0.32494737735185864,
      "grad_norm": 1006.1669311523438,
      "learning_rate": 3.430312356568399e-05,
      "loss": 89.6406,
      "step": 80430
    },
    {
      "epoch": 0.3249877786172263,
      "grad_norm": 610.352294921875,
      "learning_rate": 3.430117154863006e-05,
      "loss": 38.2407,
      "step": 80440
    },
    {
      "epoch": 0.3250281798825939,
      "grad_norm": 535.408203125,
      "learning_rate": 3.4299219252767436e-05,
      "loss": 52.4986,
      "step": 80450
    },
    {
      "epoch": 0.32506858114796156,
      "grad_norm": 634.6419067382812,
      "learning_rate": 3.429726667813416e-05,
      "loss": 57.0195,
      "step": 80460
    },
    {
      "epoch": 0.3251089824133292,
      "grad_norm": 657.6990966796875,
      "learning_rate": 3.429531382476832e-05,
      "loss": 51.0841,
      "step": 80470
    },
    {
      "epoch": 0.32514938367869683,
      "grad_norm": 693.49853515625,
      "learning_rate": 3.429336069270796e-05,
      "loss": 67.1961,
      "step": 80480
    },
    {
      "epoch": 0.3251897849440645,
      "grad_norm": 1512.3751220703125,
      "learning_rate": 3.429140728199119e-05,
      "loss": 76.1084,
      "step": 80490
    },
    {
      "epoch": 0.32523018620943206,
      "grad_norm": 1344.2879638671875,
      "learning_rate": 3.428945359265607e-05,
      "loss": 81.8786,
      "step": 80500
    },
    {
      "epoch": 0.3252705874747997,
      "grad_norm": 1220.5660400390625,
      "learning_rate": 3.42874996247407e-05,
      "loss": 76.1141,
      "step": 80510
    },
    {
      "epoch": 0.32531098874016734,
      "grad_norm": 761.2682495117188,
      "learning_rate": 3.4285545378283165e-05,
      "loss": 62.3988,
      "step": 80520
    },
    {
      "epoch": 0.325351390005535,
      "grad_norm": 692.607421875,
      "learning_rate": 3.428359085332157e-05,
      "loss": 77.3975,
      "step": 80530
    },
    {
      "epoch": 0.3253917912709026,
      "grad_norm": 845.8311157226562,
      "learning_rate": 3.4281636049894014e-05,
      "loss": 58.645,
      "step": 80540
    },
    {
      "epoch": 0.32543219253627026,
      "grad_norm": 783.7555541992188,
      "learning_rate": 3.427968096803862e-05,
      "loss": 72.8846,
      "step": 80550
    },
    {
      "epoch": 0.32547259380163784,
      "grad_norm": 889.4086303710938,
      "learning_rate": 3.4277725607793486e-05,
      "loss": 68.3054,
      "step": 80560
    },
    {
      "epoch": 0.3255129950670055,
      "grad_norm": 442.7275695800781,
      "learning_rate": 3.4275769969196745e-05,
      "loss": 63.8601,
      "step": 80570
    },
    {
      "epoch": 0.3255533963323731,
      "grad_norm": 1197.6492919921875,
      "learning_rate": 3.427381405228651e-05,
      "loss": 64.7808,
      "step": 80580
    },
    {
      "epoch": 0.32559379759774076,
      "grad_norm": 534.90771484375,
      "learning_rate": 3.427185785710093e-05,
      "loss": 47.0797,
      "step": 80590
    },
    {
      "epoch": 0.3256341988631084,
      "grad_norm": 1945.9188232421875,
      "learning_rate": 3.426990138367813e-05,
      "loss": 74.8174,
      "step": 80600
    },
    {
      "epoch": 0.32567460012847604,
      "grad_norm": 978.9138793945312,
      "learning_rate": 3.426794463205626e-05,
      "loss": 52.3134,
      "step": 80610
    },
    {
      "epoch": 0.32571500139384363,
      "grad_norm": 1130.74267578125,
      "learning_rate": 3.426598760227346e-05,
      "loss": 56.3229,
      "step": 80620
    },
    {
      "epoch": 0.32575540265921127,
      "grad_norm": 486.0097351074219,
      "learning_rate": 3.426403029436789e-05,
      "loss": 51.0282,
      "step": 80630
    },
    {
      "epoch": 0.3257958039245789,
      "grad_norm": 1250.3785400390625,
      "learning_rate": 3.42620727083777e-05,
      "loss": 66.0362,
      "step": 80640
    },
    {
      "epoch": 0.32583620518994655,
      "grad_norm": 1186.8968505859375,
      "learning_rate": 3.426011484434106e-05,
      "loss": 58.2137,
      "step": 80650
    },
    {
      "epoch": 0.3258766064553142,
      "grad_norm": 1575.198974609375,
      "learning_rate": 3.425815670229614e-05,
      "loss": 49.037,
      "step": 80660
    },
    {
      "epoch": 0.32591700772068183,
      "grad_norm": 387.7889709472656,
      "learning_rate": 3.425619828228112e-05,
      "loss": 51.2633,
      "step": 80670
    },
    {
      "epoch": 0.32595740898604947,
      "grad_norm": 977.7369995117188,
      "learning_rate": 3.425423958433418e-05,
      "loss": 58.8048,
      "step": 80680
    },
    {
      "epoch": 0.32599781025141705,
      "grad_norm": 721.7234497070312,
      "learning_rate": 3.425228060849349e-05,
      "loss": 60.1916,
      "step": 80690
    },
    {
      "epoch": 0.3260382115167847,
      "grad_norm": 738.7984619140625,
      "learning_rate": 3.425032135479725e-05,
      "loss": 58.4803,
      "step": 80700
    },
    {
      "epoch": 0.32607861278215233,
      "grad_norm": 620.6669921875,
      "learning_rate": 3.424836182328367e-05,
      "loss": 75.7753,
      "step": 80710
    },
    {
      "epoch": 0.32611901404752,
      "grad_norm": 1548.1915283203125,
      "learning_rate": 3.4246402013990935e-05,
      "loss": 91.6949,
      "step": 80720
    },
    {
      "epoch": 0.3261594153128876,
      "grad_norm": 72.1471176147461,
      "learning_rate": 3.4244441926957254e-05,
      "loss": 67.6949,
      "step": 80730
    },
    {
      "epoch": 0.32619981657825525,
      "grad_norm": 921.9493408203125,
      "learning_rate": 3.424248156222085e-05,
      "loss": 81.8886,
      "step": 80740
    },
    {
      "epoch": 0.32624021784362284,
      "grad_norm": 640.4423217773438,
      "learning_rate": 3.424052091981994e-05,
      "loss": 57.0548,
      "step": 80750
    },
    {
      "epoch": 0.3262806191089905,
      "grad_norm": 535.4073486328125,
      "learning_rate": 3.4238559999792726e-05,
      "loss": 57.7797,
      "step": 80760
    },
    {
      "epoch": 0.3263210203743581,
      "grad_norm": 619.300048828125,
      "learning_rate": 3.423659880217747e-05,
      "loss": 66.9942,
      "step": 80770
    },
    {
      "epoch": 0.32636142163972576,
      "grad_norm": 1028.4447021484375,
      "learning_rate": 3.4234637327012384e-05,
      "loss": 55.0906,
      "step": 80780
    },
    {
      "epoch": 0.3264018229050934,
      "grad_norm": 459.99017333984375,
      "learning_rate": 3.423267557433572e-05,
      "loss": 51.9673,
      "step": 80790
    },
    {
      "epoch": 0.32644222417046104,
      "grad_norm": 0.0,
      "learning_rate": 3.4230713544185715e-05,
      "loss": 74.0897,
      "step": 80800
    },
    {
      "epoch": 0.3264826254358287,
      "grad_norm": 600.255126953125,
      "learning_rate": 3.4228751236600615e-05,
      "loss": 42.6606,
      "step": 80810
    },
    {
      "epoch": 0.32652302670119626,
      "grad_norm": 1884.208740234375,
      "learning_rate": 3.4226788651618695e-05,
      "loss": 76.6577,
      "step": 80820
    },
    {
      "epoch": 0.3265634279665639,
      "grad_norm": 995.8324584960938,
      "learning_rate": 3.42248257892782e-05,
      "loss": 71.5253,
      "step": 80830
    },
    {
      "epoch": 0.32660382923193154,
      "grad_norm": 530.7216186523438,
      "learning_rate": 3.422286264961741e-05,
      "loss": 45.6298,
      "step": 80840
    },
    {
      "epoch": 0.3266442304972992,
      "grad_norm": 890.0387573242188,
      "learning_rate": 3.422089923267458e-05,
      "loss": 68.8483,
      "step": 80850
    },
    {
      "epoch": 0.3266846317626668,
      "grad_norm": 761.2548828125,
      "learning_rate": 3.421893553848801e-05,
      "loss": 62.3427,
      "step": 80860
    },
    {
      "epoch": 0.32672503302803446,
      "grad_norm": 437.8416442871094,
      "learning_rate": 3.421697156709596e-05,
      "loss": 70.8058,
      "step": 80870
    },
    {
      "epoch": 0.32676543429340205,
      "grad_norm": 167.2902374267578,
      "learning_rate": 3.421500731853674e-05,
      "loss": 43.4283,
      "step": 80880
    },
    {
      "epoch": 0.3268058355587697,
      "grad_norm": 622.623291015625,
      "learning_rate": 3.421304279284862e-05,
      "loss": 39.4079,
      "step": 80890
    },
    {
      "epoch": 0.3268462368241373,
      "grad_norm": 1559.4071044921875,
      "learning_rate": 3.421107799006992e-05,
      "loss": 71.1375,
      "step": 80900
    },
    {
      "epoch": 0.32688663808950497,
      "grad_norm": 817.6478271484375,
      "learning_rate": 3.420911291023894e-05,
      "loss": 87.4325,
      "step": 80910
    },
    {
      "epoch": 0.3269270393548726,
      "grad_norm": 1013.1358032226562,
      "learning_rate": 3.4207147553393996e-05,
      "loss": 57.3408,
      "step": 80920
    },
    {
      "epoch": 0.32696744062024025,
      "grad_norm": 797.0433349609375,
      "learning_rate": 3.420518191957339e-05,
      "loss": 95.546,
      "step": 80930
    },
    {
      "epoch": 0.32700784188560783,
      "grad_norm": 647.9711303710938,
      "learning_rate": 3.4203216008815446e-05,
      "loss": 66.2914,
      "step": 80940
    },
    {
      "epoch": 0.32704824315097547,
      "grad_norm": 585.9962768554688,
      "learning_rate": 3.42012498211585e-05,
      "loss": 52.315,
      "step": 80950
    },
    {
      "epoch": 0.3270886444163431,
      "grad_norm": 1960.798095703125,
      "learning_rate": 3.419928335664087e-05,
      "loss": 60.6582,
      "step": 80960
    },
    {
      "epoch": 0.32712904568171075,
      "grad_norm": 866.8156127929688,
      "learning_rate": 3.4197316615300904e-05,
      "loss": 69.2536,
      "step": 80970
    },
    {
      "epoch": 0.3271694469470784,
      "grad_norm": 1049.0009765625,
      "learning_rate": 3.419534959717694e-05,
      "loss": 86.5613,
      "step": 80980
    },
    {
      "epoch": 0.32720984821244603,
      "grad_norm": 1512.5225830078125,
      "learning_rate": 3.419338230230733e-05,
      "loss": 88.7681,
      "step": 80990
    },
    {
      "epoch": 0.32725024947781367,
      "grad_norm": 522.5618286132812,
      "learning_rate": 3.419141473073042e-05,
      "loss": 45.2407,
      "step": 81000
    },
    {
      "epoch": 0.32729065074318126,
      "grad_norm": 1057.761474609375,
      "learning_rate": 3.418944688248458e-05,
      "loss": 73.3249,
      "step": 81010
    },
    {
      "epoch": 0.3273310520085489,
      "grad_norm": 405.7344665527344,
      "learning_rate": 3.4187478757608166e-05,
      "loss": 58.5253,
      "step": 81020
    },
    {
      "epoch": 0.32737145327391654,
      "grad_norm": 227.71484375,
      "learning_rate": 3.418551035613954e-05,
      "loss": 68.0555,
      "step": 81030
    },
    {
      "epoch": 0.3274118545392842,
      "grad_norm": 752.8635864257812,
      "learning_rate": 3.41835416781171e-05,
      "loss": 77.9322,
      "step": 81040
    },
    {
      "epoch": 0.3274522558046518,
      "grad_norm": 580.7243041992188,
      "learning_rate": 3.4181572723579205e-05,
      "loss": 60.2561,
      "step": 81050
    },
    {
      "epoch": 0.32749265707001946,
      "grad_norm": 755.0991821289062,
      "learning_rate": 3.417960349256425e-05,
      "loss": 50.1816,
      "step": 81060
    },
    {
      "epoch": 0.32753305833538704,
      "grad_norm": 875.7639770507812,
      "learning_rate": 3.417763398511063e-05,
      "loss": 47.6605,
      "step": 81070
    },
    {
      "epoch": 0.3275734596007547,
      "grad_norm": 575.2279052734375,
      "learning_rate": 3.417566420125673e-05,
      "loss": 74.7226,
      "step": 81080
    },
    {
      "epoch": 0.3276138608661223,
      "grad_norm": 652.4024047851562,
      "learning_rate": 3.417369414104096e-05,
      "loss": 41.3797,
      "step": 81090
    },
    {
      "epoch": 0.32765426213148996,
      "grad_norm": 683.2135620117188,
      "learning_rate": 3.417172380450173e-05,
      "loss": 55.1849,
      "step": 81100
    },
    {
      "epoch": 0.3276946633968576,
      "grad_norm": 488.3472595214844,
      "learning_rate": 3.416975319167744e-05,
      "loss": 78.153,
      "step": 81110
    },
    {
      "epoch": 0.32773506466222524,
      "grad_norm": 1091.7830810546875,
      "learning_rate": 3.416778230260652e-05,
      "loss": 42.6646,
      "step": 81120
    },
    {
      "epoch": 0.3277754659275929,
      "grad_norm": 2867.489990234375,
      "learning_rate": 3.416581113732739e-05,
      "loss": 77.2699,
      "step": 81130
    },
    {
      "epoch": 0.32781586719296046,
      "grad_norm": 431.8386535644531,
      "learning_rate": 3.416383969587848e-05,
      "loss": 84.1893,
      "step": 81140
    },
    {
      "epoch": 0.3278562684583281,
      "grad_norm": 1082.2401123046875,
      "learning_rate": 3.4161867978298225e-05,
      "loss": 68.9859,
      "step": 81150
    },
    {
      "epoch": 0.32789666972369574,
      "grad_norm": 543.3303833007812,
      "learning_rate": 3.415989598462506e-05,
      "loss": 59.1727,
      "step": 81160
    },
    {
      "epoch": 0.3279370709890634,
      "grad_norm": 682.0264282226562,
      "learning_rate": 3.415792371489743e-05,
      "loss": 67.5518,
      "step": 81170
    },
    {
      "epoch": 0.327977472254431,
      "grad_norm": 884.221435546875,
      "learning_rate": 3.415595116915379e-05,
      "loss": 62.808,
      "step": 81180
    },
    {
      "epoch": 0.32801787351979866,
      "grad_norm": 1019.9883422851562,
      "learning_rate": 3.41539783474326e-05,
      "loss": 67.86,
      "step": 81190
    },
    {
      "epoch": 0.32805827478516625,
      "grad_norm": 601.5031127929688,
      "learning_rate": 3.415200524977231e-05,
      "loss": 81.0774,
      "step": 81200
    },
    {
      "epoch": 0.3280986760505339,
      "grad_norm": 543.6842651367188,
      "learning_rate": 3.415003187621139e-05,
      "loss": 59.2001,
      "step": 81210
    },
    {
      "epoch": 0.32813907731590153,
      "grad_norm": 1308.574951171875,
      "learning_rate": 3.414805822678831e-05,
      "loss": 39.9627,
      "step": 81220
    },
    {
      "epoch": 0.32817947858126917,
      "grad_norm": 543.0343017578125,
      "learning_rate": 3.4146084301541565e-05,
      "loss": 59.6952,
      "step": 81230
    },
    {
      "epoch": 0.3282198798466368,
      "grad_norm": 1221.7008056640625,
      "learning_rate": 3.414411010050962e-05,
      "loss": 72.1673,
      "step": 81240
    },
    {
      "epoch": 0.32826028111200445,
      "grad_norm": 2868.123046875,
      "learning_rate": 3.4142135623730954e-05,
      "loss": 115.6474,
      "step": 81250
    },
    {
      "epoch": 0.32830068237737203,
      "grad_norm": 887.4540405273438,
      "learning_rate": 3.4140160871244076e-05,
      "loss": 43.3435,
      "step": 81260
    },
    {
      "epoch": 0.3283410836427397,
      "grad_norm": 814.0238647460938,
      "learning_rate": 3.413818584308749e-05,
      "loss": 72.639,
      "step": 81270
    },
    {
      "epoch": 0.3283814849081073,
      "grad_norm": 592.52490234375,
      "learning_rate": 3.413621053929969e-05,
      "loss": 52.7047,
      "step": 81280
    },
    {
      "epoch": 0.32842188617347495,
      "grad_norm": 197.75149536132812,
      "learning_rate": 3.4134234959919185e-05,
      "loss": 49.6054,
      "step": 81290
    },
    {
      "epoch": 0.3284622874388426,
      "grad_norm": 1070.3717041015625,
      "learning_rate": 3.41322591049845e-05,
      "loss": 103.2167,
      "step": 81300
    },
    {
      "epoch": 0.32850268870421023,
      "grad_norm": 512.3439331054688,
      "learning_rate": 3.413028297453414e-05,
      "loss": 79.9242,
      "step": 81310
    },
    {
      "epoch": 0.3285430899695779,
      "grad_norm": 594.7864990234375,
      "learning_rate": 3.4128306568606644e-05,
      "loss": 53.6589,
      "step": 81320
    },
    {
      "epoch": 0.32858349123494546,
      "grad_norm": 645.3168334960938,
      "learning_rate": 3.4126329887240536e-05,
      "loss": 64.8506,
      "step": 81330
    },
    {
      "epoch": 0.3286238925003131,
      "grad_norm": 555.6690063476562,
      "learning_rate": 3.412435293047435e-05,
      "loss": 81.0256,
      "step": 81340
    },
    {
      "epoch": 0.32866429376568074,
      "grad_norm": 1569.737548828125,
      "learning_rate": 3.4122375698346636e-05,
      "loss": 56.2626,
      "step": 81350
    },
    {
      "epoch": 0.3287046950310484,
      "grad_norm": 573.2012329101562,
      "learning_rate": 3.412039819089593e-05,
      "loss": 81.6873,
      "step": 81360
    },
    {
      "epoch": 0.328745096296416,
      "grad_norm": 1500.382568359375,
      "learning_rate": 3.41184204081608e-05,
      "loss": 47.1895,
      "step": 81370
    },
    {
      "epoch": 0.32878549756178366,
      "grad_norm": 542.472900390625,
      "learning_rate": 3.41164423501798e-05,
      "loss": 57.6808,
      "step": 81380
    },
    {
      "epoch": 0.32882589882715124,
      "grad_norm": 2190.964111328125,
      "learning_rate": 3.4114464016991476e-05,
      "loss": 60.6285,
      "step": 81390
    },
    {
      "epoch": 0.3288663000925189,
      "grad_norm": 626.0020751953125,
      "learning_rate": 3.411248540863442e-05,
      "loss": 57.284,
      "step": 81400
    },
    {
      "epoch": 0.3289067013578865,
      "grad_norm": 1373.947265625,
      "learning_rate": 3.411050652514719e-05,
      "loss": 65.5403,
      "step": 81410
    },
    {
      "epoch": 0.32894710262325416,
      "grad_norm": 636.7581176757812,
      "learning_rate": 3.410852736656837e-05,
      "loss": 52.3231,
      "step": 81420
    },
    {
      "epoch": 0.3289875038886218,
      "grad_norm": 276.45831298828125,
      "learning_rate": 3.410654793293654e-05,
      "loss": 66.9052,
      "step": 81430
    },
    {
      "epoch": 0.32902790515398944,
      "grad_norm": 1264.7098388671875,
      "learning_rate": 3.410456822429031e-05,
      "loss": 63.3931,
      "step": 81440
    },
    {
      "epoch": 0.3290683064193571,
      "grad_norm": 1080.706787109375,
      "learning_rate": 3.410258824066825e-05,
      "loss": 65.6225,
      "step": 81450
    },
    {
      "epoch": 0.32910870768472467,
      "grad_norm": 1004.5601806640625,
      "learning_rate": 3.4100607982108975e-05,
      "loss": 70.5807,
      "step": 81460
    },
    {
      "epoch": 0.3291491089500923,
      "grad_norm": 981.4774169921875,
      "learning_rate": 3.409862744865109e-05,
      "loss": 79.1536,
      "step": 81470
    },
    {
      "epoch": 0.32918951021545995,
      "grad_norm": 1003.9555053710938,
      "learning_rate": 3.4096646640333205e-05,
      "loss": 48.9285,
      "step": 81480
    },
    {
      "epoch": 0.3292299114808276,
      "grad_norm": 524.7885131835938,
      "learning_rate": 3.4094665557193934e-05,
      "loss": 73.5914,
      "step": 81490
    },
    {
      "epoch": 0.3292703127461952,
      "grad_norm": 1038.293212890625,
      "learning_rate": 3.4092684199271896e-05,
      "loss": 69.5635,
      "step": 81500
    },
    {
      "epoch": 0.32931071401156287,
      "grad_norm": 1191.782958984375,
      "learning_rate": 3.409070256660573e-05,
      "loss": 72.7884,
      "step": 81510
    },
    {
      "epoch": 0.32935111527693045,
      "grad_norm": 738.4837646484375,
      "learning_rate": 3.408872065923406e-05,
      "loss": 78.2964,
      "step": 81520
    },
    {
      "epoch": 0.3293915165422981,
      "grad_norm": 321.0873718261719,
      "learning_rate": 3.408673847719553e-05,
      "loss": 67.1254,
      "step": 81530
    },
    {
      "epoch": 0.32943191780766573,
      "grad_norm": 707.3953247070312,
      "learning_rate": 3.408475602052878e-05,
      "loss": 63.5638,
      "step": 81540
    },
    {
      "epoch": 0.32947231907303337,
      "grad_norm": 721.4423828125,
      "learning_rate": 3.408277328927246e-05,
      "loss": 63.4857,
      "step": 81550
    },
    {
      "epoch": 0.329512720338401,
      "grad_norm": 1489.64697265625,
      "learning_rate": 3.408079028346523e-05,
      "loss": 61.2157,
      "step": 81560
    },
    {
      "epoch": 0.32955312160376865,
      "grad_norm": 1054.1998291015625,
      "learning_rate": 3.407880700314574e-05,
      "loss": 74.2502,
      "step": 81570
    },
    {
      "epoch": 0.32959352286913624,
      "grad_norm": 1091.3291015625,
      "learning_rate": 3.407682344835266e-05,
      "loss": 67.4396,
      "step": 81580
    },
    {
      "epoch": 0.3296339241345039,
      "grad_norm": 744.8306274414062,
      "learning_rate": 3.407483961912465e-05,
      "loss": 62.6582,
      "step": 81590
    },
    {
      "epoch": 0.3296743253998715,
      "grad_norm": 1556.3101806640625,
      "learning_rate": 3.407285551550041e-05,
      "loss": 66.0962,
      "step": 81600
    },
    {
      "epoch": 0.32971472666523916,
      "grad_norm": 218.52691650390625,
      "learning_rate": 3.40708711375186e-05,
      "loss": 79.0563,
      "step": 81610
    },
    {
      "epoch": 0.3297551279306068,
      "grad_norm": 561.5374755859375,
      "learning_rate": 3.4068886485217915e-05,
      "loss": 43.3707,
      "step": 81620
    },
    {
      "epoch": 0.32979552919597444,
      "grad_norm": 691.798095703125,
      "learning_rate": 3.406690155863704e-05,
      "loss": 61.0591,
      "step": 81630
    },
    {
      "epoch": 0.3298359304613421,
      "grad_norm": 489.8292236328125,
      "learning_rate": 3.406491635781468e-05,
      "loss": 70.3329,
      "step": 81640
    },
    {
      "epoch": 0.32987633172670966,
      "grad_norm": 719.1077880859375,
      "learning_rate": 3.406293088278953e-05,
      "loss": 55.4561,
      "step": 81650
    },
    {
      "epoch": 0.3299167329920773,
      "grad_norm": 306.428955078125,
      "learning_rate": 3.406094513360031e-05,
      "loss": 57.011,
      "step": 81660
    },
    {
      "epoch": 0.32995713425744494,
      "grad_norm": 1282.330810546875,
      "learning_rate": 3.4058959110285724e-05,
      "loss": 55.3932,
      "step": 81670
    },
    {
      "epoch": 0.3299975355228126,
      "grad_norm": 1125.1302490234375,
      "learning_rate": 3.4056972812884495e-05,
      "loss": 92.1705,
      "step": 81680
    },
    {
      "epoch": 0.3300379367881802,
      "grad_norm": 883.2679443359375,
      "learning_rate": 3.405498624143533e-05,
      "loss": 95.1925,
      "step": 81690
    },
    {
      "epoch": 0.33007833805354786,
      "grad_norm": 1473.376953125,
      "learning_rate": 3.405299939597699e-05,
      "loss": 72.0129,
      "step": 81700
    },
    {
      "epoch": 0.33011873931891544,
      "grad_norm": 679.3895874023438,
      "learning_rate": 3.405101227654818e-05,
      "loss": 99.1354,
      "step": 81710
    },
    {
      "epoch": 0.3301591405842831,
      "grad_norm": 1817.0472412109375,
      "learning_rate": 3.404902488318766e-05,
      "loss": 67.3122,
      "step": 81720
    },
    {
      "epoch": 0.3301995418496507,
      "grad_norm": 451.7236328125,
      "learning_rate": 3.404703721593416e-05,
      "loss": 75.9052,
      "step": 81730
    },
    {
      "epoch": 0.33023994311501836,
      "grad_norm": 685.8228759765625,
      "learning_rate": 3.404504927482644e-05,
      "loss": 83.0571,
      "step": 81740
    },
    {
      "epoch": 0.330280344380386,
      "grad_norm": 917.3994750976562,
      "learning_rate": 3.4043061059903254e-05,
      "loss": 67.7444,
      "step": 81750
    },
    {
      "epoch": 0.33032074564575364,
      "grad_norm": 808.7835083007812,
      "learning_rate": 3.404107257120336e-05,
      "loss": 69.5583,
      "step": 81760
    },
    {
      "epoch": 0.3303611469111213,
      "grad_norm": 1228.276123046875,
      "learning_rate": 3.4039083808765534e-05,
      "loss": 74.209,
      "step": 81770
    },
    {
      "epoch": 0.33040154817648887,
      "grad_norm": 519.683349609375,
      "learning_rate": 3.403709477262853e-05,
      "loss": 98.1551,
      "step": 81780
    },
    {
      "epoch": 0.3304419494418565,
      "grad_norm": 729.7907104492188,
      "learning_rate": 3.403510546283115e-05,
      "loss": 71.8872,
      "step": 81790
    },
    {
      "epoch": 0.33048235070722415,
      "grad_norm": 1474.07470703125,
      "learning_rate": 3.403311587941215e-05,
      "loss": 79.9147,
      "step": 81800
    },
    {
      "epoch": 0.3305227519725918,
      "grad_norm": 272.7594299316406,
      "learning_rate": 3.403112602241034e-05,
      "loss": 70.2182,
      "step": 81810
    },
    {
      "epoch": 0.33056315323795943,
      "grad_norm": 565.2523803710938,
      "learning_rate": 3.40291358918645e-05,
      "loss": 58.0919,
      "step": 81820
    },
    {
      "epoch": 0.33060355450332707,
      "grad_norm": 679.3093872070312,
      "learning_rate": 3.402714548781344e-05,
      "loss": 77.7463,
      "step": 81830
    },
    {
      "epoch": 0.33064395576869465,
      "grad_norm": 0.0,
      "learning_rate": 3.402515481029595e-05,
      "loss": 72.2586,
      "step": 81840
    },
    {
      "epoch": 0.3306843570340623,
      "grad_norm": 1301.266845703125,
      "learning_rate": 3.402316385935085e-05,
      "loss": 75.1626,
      "step": 81850
    },
    {
      "epoch": 0.33072475829942993,
      "grad_norm": 981.5969848632812,
      "learning_rate": 3.402117263501695e-05,
      "loss": 83.7884,
      "step": 81860
    },
    {
      "epoch": 0.3307651595647976,
      "grad_norm": 508.7431335449219,
      "learning_rate": 3.4019181137333066e-05,
      "loss": 72.0551,
      "step": 81870
    },
    {
      "epoch": 0.3308055608301652,
      "grad_norm": 850.6770629882812,
      "learning_rate": 3.4017189366338036e-05,
      "loss": 89.4061,
      "step": 81880
    },
    {
      "epoch": 0.33084596209553285,
      "grad_norm": 715.669677734375,
      "learning_rate": 3.401519732207068e-05,
      "loss": 29.283,
      "step": 81890
    },
    {
      "epoch": 0.33088636336090044,
      "grad_norm": 0.0,
      "learning_rate": 3.401320500456984e-05,
      "loss": 72.7661,
      "step": 81900
    },
    {
      "epoch": 0.3309267646262681,
      "grad_norm": 735.201904296875,
      "learning_rate": 3.401121241387435e-05,
      "loss": 74.2545,
      "step": 81910
    },
    {
      "epoch": 0.3309671658916357,
      "grad_norm": 879.0210571289062,
      "learning_rate": 3.400921955002306e-05,
      "loss": 73.9921,
      "step": 81920
    },
    {
      "epoch": 0.33100756715700336,
      "grad_norm": 704.4776611328125,
      "learning_rate": 3.4007226413054824e-05,
      "loss": 44.1779,
      "step": 81930
    },
    {
      "epoch": 0.331047968422371,
      "grad_norm": 1078.0277099609375,
      "learning_rate": 3.400523300300849e-05,
      "loss": 83.9579,
      "step": 81940
    },
    {
      "epoch": 0.33108836968773864,
      "grad_norm": 876.3001098632812,
      "learning_rate": 3.4003239319922935e-05,
      "loss": 55.4064,
      "step": 81950
    },
    {
      "epoch": 0.3311287709531063,
      "grad_norm": 466.29736328125,
      "learning_rate": 3.4001245363837025e-05,
      "loss": 65.0786,
      "step": 81960
    },
    {
      "epoch": 0.33116917221847386,
      "grad_norm": 797.1023559570312,
      "learning_rate": 3.3999251134789624e-05,
      "loss": 56.1711,
      "step": 81970
    },
    {
      "epoch": 0.3312095734838415,
      "grad_norm": 1126.777587890625,
      "learning_rate": 3.3997256632819616e-05,
      "loss": 86.8346,
      "step": 81980
    },
    {
      "epoch": 0.33124997474920914,
      "grad_norm": 519.679931640625,
      "learning_rate": 3.399526185796588e-05,
      "loss": 53.7663,
      "step": 81990
    },
    {
      "epoch": 0.3312903760145768,
      "grad_norm": 1049.772705078125,
      "learning_rate": 3.399326681026731e-05,
      "loss": 68.3721,
      "step": 82000
    },
    {
      "epoch": 0.3313307772799444,
      "grad_norm": 972.539794921875,
      "learning_rate": 3.399127148976281e-05,
      "loss": 84.3961,
      "step": 82010
    },
    {
      "epoch": 0.33137117854531206,
      "grad_norm": 533.8385620117188,
      "learning_rate": 3.398927589649125e-05,
      "loss": 28.2353,
      "step": 82020
    },
    {
      "epoch": 0.33141157981067965,
      "grad_norm": 432.9517822265625,
      "learning_rate": 3.3987280030491564e-05,
      "loss": 54.7922,
      "step": 82030
    },
    {
      "epoch": 0.3314519810760473,
      "grad_norm": 1321.9207763671875,
      "learning_rate": 3.398528389180265e-05,
      "loss": 65.6759,
      "step": 82040
    },
    {
      "epoch": 0.3314923823414149,
      "grad_norm": 1934.19677734375,
      "learning_rate": 3.398328748046343e-05,
      "loss": 72.9202,
      "step": 82050
    },
    {
      "epoch": 0.33153278360678257,
      "grad_norm": 973.4382934570312,
      "learning_rate": 3.3981290796512825e-05,
      "loss": 57.7526,
      "step": 82060
    },
    {
      "epoch": 0.3315731848721502,
      "grad_norm": 567.466064453125,
      "learning_rate": 3.397929383998975e-05,
      "loss": 66.2216,
      "step": 82070
    },
    {
      "epoch": 0.33161358613751785,
      "grad_norm": 848.7359008789062,
      "learning_rate": 3.3977296610933145e-05,
      "loss": 59.2886,
      "step": 82080
    },
    {
      "epoch": 0.33165398740288543,
      "grad_norm": 702.6904296875,
      "learning_rate": 3.397529910938195e-05,
      "loss": 58.9801,
      "step": 82090
    },
    {
      "epoch": 0.33169438866825307,
      "grad_norm": 295.2632751464844,
      "learning_rate": 3.3973301335375104e-05,
      "loss": 63.8799,
      "step": 82100
    },
    {
      "epoch": 0.3317347899336207,
      "grad_norm": 492.2831115722656,
      "learning_rate": 3.3971303288951554e-05,
      "loss": 98.2528,
      "step": 82110
    },
    {
      "epoch": 0.33177519119898835,
      "grad_norm": 977.1430053710938,
      "learning_rate": 3.3969304970150255e-05,
      "loss": 53.2605,
      "step": 82120
    },
    {
      "epoch": 0.331815592464356,
      "grad_norm": 761.702392578125,
      "learning_rate": 3.396730637901016e-05,
      "loss": 77.0182,
      "step": 82130
    },
    {
      "epoch": 0.33185599372972363,
      "grad_norm": 656.1929321289062,
      "learning_rate": 3.396530751557024e-05,
      "loss": 55.2026,
      "step": 82140
    },
    {
      "epoch": 0.33189639499509127,
      "grad_norm": 516.7476196289062,
      "learning_rate": 3.396330837986946e-05,
      "loss": 77.5115,
      "step": 82150
    },
    {
      "epoch": 0.33193679626045886,
      "grad_norm": 394.3022155761719,
      "learning_rate": 3.396130897194679e-05,
      "loss": 59.2189,
      "step": 82160
    },
    {
      "epoch": 0.3319771975258265,
      "grad_norm": 707.18994140625,
      "learning_rate": 3.395930929184122e-05,
      "loss": 47.8635,
      "step": 82170
    },
    {
      "epoch": 0.33201759879119414,
      "grad_norm": 1019.1130981445312,
      "learning_rate": 3.395730933959172e-05,
      "loss": 63.8909,
      "step": 82180
    },
    {
      "epoch": 0.3320580000565618,
      "grad_norm": 530.5923461914062,
      "learning_rate": 3.3955309115237296e-05,
      "loss": 67.0644,
      "step": 82190
    },
    {
      "epoch": 0.3320984013219294,
      "grad_norm": 600.5137329101562,
      "learning_rate": 3.395330861881693e-05,
      "loss": 65.8614,
      "step": 82200
    },
    {
      "epoch": 0.33213880258729706,
      "grad_norm": 871.7723388671875,
      "learning_rate": 3.3951307850369634e-05,
      "loss": 41.2921,
      "step": 82210
    },
    {
      "epoch": 0.33217920385266464,
      "grad_norm": 593.4152221679688,
      "learning_rate": 3.394930680993441e-05,
      "loss": 95.3884,
      "step": 82220
    },
    {
      "epoch": 0.3322196051180323,
      "grad_norm": 390.56927490234375,
      "learning_rate": 3.394730549755027e-05,
      "loss": 56.8083,
      "step": 82230
    },
    {
      "epoch": 0.3322600063833999,
      "grad_norm": 474.56964111328125,
      "learning_rate": 3.3945303913256216e-05,
      "loss": 70.1066,
      "step": 82240
    },
    {
      "epoch": 0.33230040764876756,
      "grad_norm": 882.9507446289062,
      "learning_rate": 3.39433020570913e-05,
      "loss": 78.8813,
      "step": 82250
    },
    {
      "epoch": 0.3323408089141352,
      "grad_norm": 771.1679077148438,
      "learning_rate": 3.394129992909452e-05,
      "loss": 63.4995,
      "step": 82260
    },
    {
      "epoch": 0.33238121017950284,
      "grad_norm": 2786.767578125,
      "learning_rate": 3.3939297529304926e-05,
      "loss": 59.4498,
      "step": 82270
    },
    {
      "epoch": 0.3324216114448705,
      "grad_norm": 758.8515014648438,
      "learning_rate": 3.393729485776154e-05,
      "loss": 94.3925,
      "step": 82280
    },
    {
      "epoch": 0.33246201271023806,
      "grad_norm": 937.47998046875,
      "learning_rate": 3.393529191450342e-05,
      "loss": 68.8819,
      "step": 82290
    },
    {
      "epoch": 0.3325024139756057,
      "grad_norm": 418.58563232421875,
      "learning_rate": 3.393328869956962e-05,
      "loss": 47.6533,
      "step": 82300
    },
    {
      "epoch": 0.33254281524097334,
      "grad_norm": 407.52972412109375,
      "learning_rate": 3.393128521299917e-05,
      "loss": 54.5192,
      "step": 82310
    },
    {
      "epoch": 0.332583216506341,
      "grad_norm": 1072.966552734375,
      "learning_rate": 3.392928145483115e-05,
      "loss": 71.5673,
      "step": 82320
    },
    {
      "epoch": 0.3326236177717086,
      "grad_norm": 636.4874877929688,
      "learning_rate": 3.392727742510462e-05,
      "loss": 80.9786,
      "step": 82330
    },
    {
      "epoch": 0.33266401903707626,
      "grad_norm": 801.964599609375,
      "learning_rate": 3.3925273123858644e-05,
      "loss": 66.5421,
      "step": 82340
    },
    {
      "epoch": 0.33270442030244385,
      "grad_norm": 614.4639282226562,
      "learning_rate": 3.3923268551132294e-05,
      "loss": 56.6249,
      "step": 82350
    },
    {
      "epoch": 0.3327448215678115,
      "grad_norm": 2463.41162109375,
      "learning_rate": 3.392126370696466e-05,
      "loss": 95.9889,
      "step": 82360
    },
    {
      "epoch": 0.33278522283317913,
      "grad_norm": 950.767822265625,
      "learning_rate": 3.391925859139482e-05,
      "loss": 64.5511,
      "step": 82370
    },
    {
      "epoch": 0.33282562409854677,
      "grad_norm": 968.8400268554688,
      "learning_rate": 3.391725320446187e-05,
      "loss": 72.9015,
      "step": 82380
    },
    {
      "epoch": 0.3328660253639144,
      "grad_norm": 481.743408203125,
      "learning_rate": 3.3915247546204905e-05,
      "loss": 75.9669,
      "step": 82390
    },
    {
      "epoch": 0.33290642662928205,
      "grad_norm": 925.4773559570312,
      "learning_rate": 3.391324161666302e-05,
      "loss": 78.0991,
      "step": 82400
    },
    {
      "epoch": 0.33294682789464963,
      "grad_norm": 1083.5574951171875,
      "learning_rate": 3.3911235415875326e-05,
      "loss": 94.1073,
      "step": 82410
    },
    {
      "epoch": 0.3329872291600173,
      "grad_norm": 0.0,
      "learning_rate": 3.390922894388094e-05,
      "loss": 48.1652,
      "step": 82420
    },
    {
      "epoch": 0.3330276304253849,
      "grad_norm": 1400.068359375,
      "learning_rate": 3.390722220071897e-05,
      "loss": 72.396,
      "step": 82430
    },
    {
      "epoch": 0.33306803169075255,
      "grad_norm": 856.3434448242188,
      "learning_rate": 3.390521518642855e-05,
      "loss": 79.0431,
      "step": 82440
    },
    {
      "epoch": 0.3331084329561202,
      "grad_norm": 553.9498901367188,
      "learning_rate": 3.39032079010488e-05,
      "loss": 61.321,
      "step": 82450
    },
    {
      "epoch": 0.33314883422148783,
      "grad_norm": 830.5960083007812,
      "learning_rate": 3.390120034461886e-05,
      "loss": 100.5,
      "step": 82460
    },
    {
      "epoch": 0.3331892354868555,
      "grad_norm": 873.7688598632812,
      "learning_rate": 3.389919251717785e-05,
      "loss": 105.3553,
      "step": 82470
    },
    {
      "epoch": 0.33322963675222306,
      "grad_norm": 773.9730834960938,
      "learning_rate": 3.3897184418764925e-05,
      "loss": 87.4409,
      "step": 82480
    },
    {
      "epoch": 0.3332700380175907,
      "grad_norm": 694.1939086914062,
      "learning_rate": 3.389517604941924e-05,
      "loss": 86.3784,
      "step": 82490
    },
    {
      "epoch": 0.33331043928295834,
      "grad_norm": 1546.2618408203125,
      "learning_rate": 3.3893167409179945e-05,
      "loss": 61.1646,
      "step": 82500
    },
    {
      "epoch": 0.333350840548326,
      "grad_norm": 1124.92724609375,
      "learning_rate": 3.389115849808621e-05,
      "loss": 70.7375,
      "step": 82510
    },
    {
      "epoch": 0.3333912418136936,
      "grad_norm": 627.1702270507812,
      "learning_rate": 3.3889149316177167e-05,
      "loss": 82.8139,
      "step": 82520
    },
    {
      "epoch": 0.33343164307906126,
      "grad_norm": 484.9052429199219,
      "learning_rate": 3.388713986349202e-05,
      "loss": 63.3153,
      "step": 82530
    },
    {
      "epoch": 0.33347204434442884,
      "grad_norm": 1087.6514892578125,
      "learning_rate": 3.388513014006993e-05,
      "loss": 43.2902,
      "step": 82540
    },
    {
      "epoch": 0.3335124456097965,
      "grad_norm": 875.129150390625,
      "learning_rate": 3.388312014595008e-05,
      "loss": 82.1861,
      "step": 82550
    },
    {
      "epoch": 0.3335528468751641,
      "grad_norm": 628.6128540039062,
      "learning_rate": 3.3881109881171656e-05,
      "loss": 58.0868,
      "step": 82560
    },
    {
      "epoch": 0.33359324814053176,
      "grad_norm": 570.00048828125,
      "learning_rate": 3.387909934577384e-05,
      "loss": 57.2228,
      "step": 82570
    },
    {
      "epoch": 0.3336336494058994,
      "grad_norm": 590.1167602539062,
      "learning_rate": 3.387708853979585e-05,
      "loss": 47.8389,
      "step": 82580
    },
    {
      "epoch": 0.33367405067126704,
      "grad_norm": 801.5050659179688,
      "learning_rate": 3.3875077463276865e-05,
      "loss": 82.9828,
      "step": 82590
    },
    {
      "epoch": 0.3337144519366347,
      "grad_norm": 802.7755126953125,
      "learning_rate": 3.38730661162561e-05,
      "loss": 69.2401,
      "step": 82600
    },
    {
      "epoch": 0.33375485320200227,
      "grad_norm": 711.4531860351562,
      "learning_rate": 3.387105449877278e-05,
      "loss": 44.5337,
      "step": 82610
    },
    {
      "epoch": 0.3337952544673699,
      "grad_norm": 371.0378112792969,
      "learning_rate": 3.38690426108661e-05,
      "loss": 38.7474,
      "step": 82620
    },
    {
      "epoch": 0.33383565573273755,
      "grad_norm": 429.09185791015625,
      "learning_rate": 3.3867030452575296e-05,
      "loss": 70.3189,
      "step": 82630
    },
    {
      "epoch": 0.3338760569981052,
      "grad_norm": 760.4613037109375,
      "learning_rate": 3.3865018023939595e-05,
      "loss": 71.0347,
      "step": 82640
    },
    {
      "epoch": 0.3339164582634728,
      "grad_norm": 411.7153625488281,
      "learning_rate": 3.386300532499823e-05,
      "loss": 76.0749,
      "step": 82650
    },
    {
      "epoch": 0.33395685952884047,
      "grad_norm": 305.81219482421875,
      "learning_rate": 3.386099235579044e-05,
      "loss": 50.1637,
      "step": 82660
    },
    {
      "epoch": 0.33399726079420805,
      "grad_norm": 1149.036376953125,
      "learning_rate": 3.385897911635547e-05,
      "loss": 72.353,
      "step": 82670
    },
    {
      "epoch": 0.3340376620595757,
      "grad_norm": 386.5274353027344,
      "learning_rate": 3.385696560673257e-05,
      "loss": 60.3233,
      "step": 82680
    },
    {
      "epoch": 0.33407806332494333,
      "grad_norm": 1266.0382080078125,
      "learning_rate": 3.385495182696098e-05,
      "loss": 57.865,
      "step": 82690
    },
    {
      "epoch": 0.33411846459031097,
      "grad_norm": 673.6781616210938,
      "learning_rate": 3.3852937777079976e-05,
      "loss": 39.5594,
      "step": 82700
    },
    {
      "epoch": 0.3341588658556786,
      "grad_norm": 2739.0546875,
      "learning_rate": 3.385092345712882e-05,
      "loss": 54.4652,
      "step": 82710
    },
    {
      "epoch": 0.33419926712104625,
      "grad_norm": 1225.4815673828125,
      "learning_rate": 3.3848908867146784e-05,
      "loss": 77.8722,
      "step": 82720
    },
    {
      "epoch": 0.33423966838641384,
      "grad_norm": 362.6470031738281,
      "learning_rate": 3.3846894007173135e-05,
      "loss": 77.0374,
      "step": 82730
    },
    {
      "epoch": 0.3342800696517815,
      "grad_norm": 1833.700439453125,
      "learning_rate": 3.384487887724716e-05,
      "loss": 79.7389,
      "step": 82740
    },
    {
      "epoch": 0.3343204709171491,
      "grad_norm": 835.5281372070312,
      "learning_rate": 3.384286347740814e-05,
      "loss": 57.7872,
      "step": 82750
    },
    {
      "epoch": 0.33436087218251676,
      "grad_norm": 1015.665771484375,
      "learning_rate": 3.3840847807695367e-05,
      "loss": 72.3746,
      "step": 82760
    },
    {
      "epoch": 0.3344012734478844,
      "grad_norm": 876.6022338867188,
      "learning_rate": 3.383883186814815e-05,
      "loss": 63.0179,
      "step": 82770
    },
    {
      "epoch": 0.33444167471325204,
      "grad_norm": 615.6849975585938,
      "learning_rate": 3.3836815658805776e-05,
      "loss": 56.2009,
      "step": 82780
    },
    {
      "epoch": 0.3344820759786197,
      "grad_norm": 1607.5733642578125,
      "learning_rate": 3.383479917970756e-05,
      "loss": 71.4537,
      "step": 82790
    },
    {
      "epoch": 0.33452247724398726,
      "grad_norm": 527.6516723632812,
      "learning_rate": 3.3832782430892806e-05,
      "loss": 65.4176,
      "step": 82800
    },
    {
      "epoch": 0.3345628785093549,
      "grad_norm": 647.039794921875,
      "learning_rate": 3.383076541240084e-05,
      "loss": 66.098,
      "step": 82810
    },
    {
      "epoch": 0.33460327977472254,
      "grad_norm": 511.5931396484375,
      "learning_rate": 3.3828748124270983e-05,
      "loss": 74.0146,
      "step": 82820
    },
    {
      "epoch": 0.3346436810400902,
      "grad_norm": 791.4042358398438,
      "learning_rate": 3.3826730566542555e-05,
      "loss": 60.8105,
      "step": 82830
    },
    {
      "epoch": 0.3346840823054578,
      "grad_norm": 609.68603515625,
      "learning_rate": 3.3824712739254904e-05,
      "loss": 64.9001,
      "step": 82840
    },
    {
      "epoch": 0.33472448357082546,
      "grad_norm": 795.0293579101562,
      "learning_rate": 3.382269464244736e-05,
      "loss": 60.0467,
      "step": 82850
    },
    {
      "epoch": 0.33476488483619304,
      "grad_norm": 555.2614135742188,
      "learning_rate": 3.382067627615926e-05,
      "loss": 79.6582,
      "step": 82860
    },
    {
      "epoch": 0.3348052861015607,
      "grad_norm": 534.7796630859375,
      "learning_rate": 3.381865764042997e-05,
      "loss": 73.6634,
      "step": 82870
    },
    {
      "epoch": 0.3348456873669283,
      "grad_norm": 2460.16162109375,
      "learning_rate": 3.381663873529883e-05,
      "loss": 62.1055,
      "step": 82880
    },
    {
      "epoch": 0.33488608863229596,
      "grad_norm": 524.8576049804688,
      "learning_rate": 3.3814619560805205e-05,
      "loss": 123.9045,
      "step": 82890
    },
    {
      "epoch": 0.3349264898976636,
      "grad_norm": 1630.5303955078125,
      "learning_rate": 3.381260011698846e-05,
      "loss": 51.2895,
      "step": 82900
    },
    {
      "epoch": 0.33496689116303124,
      "grad_norm": 561.4447631835938,
      "learning_rate": 3.3810580403887965e-05,
      "loss": 73.8346,
      "step": 82910
    },
    {
      "epoch": 0.3350072924283989,
      "grad_norm": 598.4013671875,
      "learning_rate": 3.3808560421543094e-05,
      "loss": 60.0628,
      "step": 82920
    },
    {
      "epoch": 0.33504769369376647,
      "grad_norm": 324.10302734375,
      "learning_rate": 3.380654016999323e-05,
      "loss": 58.894,
      "step": 82930
    },
    {
      "epoch": 0.3350880949591341,
      "grad_norm": 504.5718688964844,
      "learning_rate": 3.3804519649277754e-05,
      "loss": 62.397,
      "step": 82940
    },
    {
      "epoch": 0.33512849622450175,
      "grad_norm": 707.5982666015625,
      "learning_rate": 3.3802498859436066e-05,
      "loss": 69.2411,
      "step": 82950
    },
    {
      "epoch": 0.3351688974898694,
      "grad_norm": 1631.22119140625,
      "learning_rate": 3.3800477800507555e-05,
      "loss": 77.0283,
      "step": 82960
    },
    {
      "epoch": 0.33520929875523703,
      "grad_norm": 1262.4522705078125,
      "learning_rate": 3.379845647253162e-05,
      "loss": 56.2125,
      "step": 82970
    },
    {
      "epoch": 0.33524970002060467,
      "grad_norm": 680.508544921875,
      "learning_rate": 3.3796434875547675e-05,
      "loss": 51.1179,
      "step": 82980
    },
    {
      "epoch": 0.33529010128597225,
      "grad_norm": 1367.9285888671875,
      "learning_rate": 3.379441300959513e-05,
      "loss": 78.3301,
      "step": 82990
    },
    {
      "epoch": 0.3353305025513399,
      "grad_norm": 1376.3389892578125,
      "learning_rate": 3.37923908747134e-05,
      "loss": 65.1142,
      "step": 83000
    },
    {
      "epoch": 0.33537090381670753,
      "grad_norm": 701.1461791992188,
      "learning_rate": 3.379036847094191e-05,
      "loss": 72.6158,
      "step": 83010
    },
    {
      "epoch": 0.3354113050820752,
      "grad_norm": 436.153076171875,
      "learning_rate": 3.378834579832008e-05,
      "loss": 79.1066,
      "step": 83020
    },
    {
      "epoch": 0.3354517063474428,
      "grad_norm": 664.0464477539062,
      "learning_rate": 3.378632285688736e-05,
      "loss": 75.7545,
      "step": 83030
    },
    {
      "epoch": 0.33549210761281045,
      "grad_norm": 1134.675048828125,
      "learning_rate": 3.378429964668318e-05,
      "loss": 58.1786,
      "step": 83040
    },
    {
      "epoch": 0.33553250887817804,
      "grad_norm": 1004.5962524414062,
      "learning_rate": 3.378227616774697e-05,
      "loss": 51.7707,
      "step": 83050
    },
    {
      "epoch": 0.3355729101435457,
      "grad_norm": 977.3733520507812,
      "learning_rate": 3.37802524201182e-05,
      "loss": 66.7884,
      "step": 83060
    },
    {
      "epoch": 0.3356133114089133,
      "grad_norm": 600.7833862304688,
      "learning_rate": 3.377822840383632e-05,
      "loss": 94.3781,
      "step": 83070
    },
    {
      "epoch": 0.33565371267428096,
      "grad_norm": 634.0911254882812,
      "learning_rate": 3.377620411894077e-05,
      "loss": 56.2981,
      "step": 83080
    },
    {
      "epoch": 0.3356941139396486,
      "grad_norm": 494.5894470214844,
      "learning_rate": 3.377417956547103e-05,
      "loss": 92.2448,
      "step": 83090
    },
    {
      "epoch": 0.33573451520501624,
      "grad_norm": 1046.20703125,
      "learning_rate": 3.377215474346657e-05,
      "loss": 60.8228,
      "step": 83100
    },
    {
      "epoch": 0.3357749164703839,
      "grad_norm": 2085.396484375,
      "learning_rate": 3.377012965296687e-05,
      "loss": 59.7961,
      "step": 83110
    },
    {
      "epoch": 0.33581531773575146,
      "grad_norm": 755.2464599609375,
      "learning_rate": 3.3768104294011394e-05,
      "loss": 53.3835,
      "step": 83120
    },
    {
      "epoch": 0.3358557190011191,
      "grad_norm": 768.9041137695312,
      "learning_rate": 3.376607866663964e-05,
      "loss": 57.641,
      "step": 83130
    },
    {
      "epoch": 0.33589612026648674,
      "grad_norm": 1092.0604248046875,
      "learning_rate": 3.3764052770891095e-05,
      "loss": 73.4515,
      "step": 83140
    },
    {
      "epoch": 0.3359365215318544,
      "grad_norm": 1306.6480712890625,
      "learning_rate": 3.376202660680526e-05,
      "loss": 74.138,
      "step": 83150
    },
    {
      "epoch": 0.335976922797222,
      "grad_norm": 215.6219024658203,
      "learning_rate": 3.376000017442162e-05,
      "loss": 69.0981,
      "step": 83160
    },
    {
      "epoch": 0.33601732406258966,
      "grad_norm": 464.001220703125,
      "learning_rate": 3.37579734737797e-05,
      "loss": 41.6006,
      "step": 83170
    },
    {
      "epoch": 0.33605772532795725,
      "grad_norm": 2117.4033203125,
      "learning_rate": 3.3755946504919e-05,
      "loss": 115.2802,
      "step": 83180
    },
    {
      "epoch": 0.3360981265933249,
      "grad_norm": 1764.74462890625,
      "learning_rate": 3.375391926787905e-05,
      "loss": 60.6559,
      "step": 83190
    },
    {
      "epoch": 0.3361385278586925,
      "grad_norm": 1227.6932373046875,
      "learning_rate": 3.375189176269935e-05,
      "loss": 66.6953,
      "step": 83200
    },
    {
      "epoch": 0.33617892912406017,
      "grad_norm": 828.9819946289062,
      "learning_rate": 3.374986398941944e-05,
      "loss": 66.9964,
      "step": 83210
    },
    {
      "epoch": 0.3362193303894278,
      "grad_norm": 1059.575927734375,
      "learning_rate": 3.374783594807887e-05,
      "loss": 73.6725,
      "step": 83220
    },
    {
      "epoch": 0.33625973165479545,
      "grad_norm": 346.5843811035156,
      "learning_rate": 3.3745807638717144e-05,
      "loss": 63.1543,
      "step": 83230
    },
    {
      "epoch": 0.3363001329201631,
      "grad_norm": 5886.8798828125,
      "learning_rate": 3.374377906137383e-05,
      "loss": 53.1895,
      "step": 83240
    },
    {
      "epoch": 0.33634053418553067,
      "grad_norm": 1697.4791259765625,
      "learning_rate": 3.3741750216088465e-05,
      "loss": 74.849,
      "step": 83250
    },
    {
      "epoch": 0.3363809354508983,
      "grad_norm": 559.0037231445312,
      "learning_rate": 3.37397211029006e-05,
      "loss": 70.1515,
      "step": 83260
    },
    {
      "epoch": 0.33642133671626595,
      "grad_norm": 549.3245239257812,
      "learning_rate": 3.373769172184981e-05,
      "loss": 68.2745,
      "step": 83270
    },
    {
      "epoch": 0.3364617379816336,
      "grad_norm": 530.4989013671875,
      "learning_rate": 3.3735662072975635e-05,
      "loss": 59.415,
      "step": 83280
    },
    {
      "epoch": 0.33650213924700123,
      "grad_norm": 598.5599365234375,
      "learning_rate": 3.373363215631766e-05,
      "loss": 49.4548,
      "step": 83290
    },
    {
      "epoch": 0.33654254051236887,
      "grad_norm": 1466.6553955078125,
      "learning_rate": 3.373160197191546e-05,
      "loss": 67.0955,
      "step": 83300
    },
    {
      "epoch": 0.33658294177773646,
      "grad_norm": 442.7725830078125,
      "learning_rate": 3.3729571519808606e-05,
      "loss": 56.7274,
      "step": 83310
    },
    {
      "epoch": 0.3366233430431041,
      "grad_norm": 477.98785400390625,
      "learning_rate": 3.372754080003669e-05,
      "loss": 64.1001,
      "step": 83320
    },
    {
      "epoch": 0.33666374430847174,
      "grad_norm": 673.998046875,
      "learning_rate": 3.372550981263929e-05,
      "loss": 64.573,
      "step": 83330
    },
    {
      "epoch": 0.3367041455738394,
      "grad_norm": 3385.6376953125,
      "learning_rate": 3.3723478557656016e-05,
      "loss": 72.427,
      "step": 83340
    },
    {
      "epoch": 0.336744546839207,
      "grad_norm": 401.60125732421875,
      "learning_rate": 3.3721447035126464e-05,
      "loss": 70.8086,
      "step": 83350
    },
    {
      "epoch": 0.33678494810457466,
      "grad_norm": 442.1596374511719,
      "learning_rate": 3.371941524509024e-05,
      "loss": 62.7455,
      "step": 83360
    },
    {
      "epoch": 0.33682534936994224,
      "grad_norm": 1584.1695556640625,
      "learning_rate": 3.371738318758694e-05,
      "loss": 82.1702,
      "step": 83370
    },
    {
      "epoch": 0.3368657506353099,
      "grad_norm": 1419.6942138671875,
      "learning_rate": 3.37153508626562e-05,
      "loss": 80.0828,
      "step": 83380
    },
    {
      "epoch": 0.3369061519006775,
      "grad_norm": 1289.188720703125,
      "learning_rate": 3.3713318270337643e-05,
      "loss": 73.5477,
      "step": 83390
    },
    {
      "epoch": 0.33694655316604516,
      "grad_norm": 816.6846923828125,
      "learning_rate": 3.3711285410670876e-05,
      "loss": 49.8017,
      "step": 83400
    },
    {
      "epoch": 0.3369869544314128,
      "grad_norm": 641.5718383789062,
      "learning_rate": 3.370925228369554e-05,
      "loss": 85.3183,
      "step": 83410
    },
    {
      "epoch": 0.33702735569678044,
      "grad_norm": 514.9337768554688,
      "learning_rate": 3.370721888945127e-05,
      "loss": 51.9517,
      "step": 83420
    },
    {
      "epoch": 0.3370677569621481,
      "grad_norm": 484.6968078613281,
      "learning_rate": 3.370518522797772e-05,
      "loss": 39.9968,
      "step": 83430
    },
    {
      "epoch": 0.33710815822751566,
      "grad_norm": 710.4841918945312,
      "learning_rate": 3.370315129931453e-05,
      "loss": 53.6119,
      "step": 83440
    },
    {
      "epoch": 0.3371485594928833,
      "grad_norm": 228.62632751464844,
      "learning_rate": 3.370111710350134e-05,
      "loss": 48.0815,
      "step": 83450
    },
    {
      "epoch": 0.33718896075825094,
      "grad_norm": 1313.511474609375,
      "learning_rate": 3.369908264057783e-05,
      "loss": 78.3585,
      "step": 83460
    },
    {
      "epoch": 0.3372293620236186,
      "grad_norm": 882.86279296875,
      "learning_rate": 3.3697047910583635e-05,
      "loss": 67.3662,
      "step": 83470
    },
    {
      "epoch": 0.3372697632889862,
      "grad_norm": 1051.37451171875,
      "learning_rate": 3.369501291355845e-05,
      "loss": 54.169,
      "step": 83480
    },
    {
      "epoch": 0.33731016455435386,
      "grad_norm": 495.60162353515625,
      "learning_rate": 3.369297764954194e-05,
      "loss": 65.4812,
      "step": 83490
    },
    {
      "epoch": 0.33735056581972145,
      "grad_norm": 0.0,
      "learning_rate": 3.369094211857378e-05,
      "loss": 83.5426,
      "step": 83500
    },
    {
      "epoch": 0.3373909670850891,
      "grad_norm": 588.1001586914062,
      "learning_rate": 3.3688906320693645e-05,
      "loss": 41.9696,
      "step": 83510
    },
    {
      "epoch": 0.33743136835045673,
      "grad_norm": 772.1838989257812,
      "learning_rate": 3.3686870255941246e-05,
      "loss": 67.5083,
      "step": 83520
    },
    {
      "epoch": 0.33747176961582437,
      "grad_norm": 599.001708984375,
      "learning_rate": 3.368483392435626e-05,
      "loss": 81.6238,
      "step": 83530
    },
    {
      "epoch": 0.337512170881192,
      "grad_norm": 865.5444946289062,
      "learning_rate": 3.368279732597839e-05,
      "loss": 63.9727,
      "step": 83540
    },
    {
      "epoch": 0.33755257214655965,
      "grad_norm": 600.300048828125,
      "learning_rate": 3.368076046084734e-05,
      "loss": 79.3259,
      "step": 83550
    },
    {
      "epoch": 0.3375929734119273,
      "grad_norm": 3054.122802734375,
      "learning_rate": 3.3678723329002826e-05,
      "loss": 84.2376,
      "step": 83560
    },
    {
      "epoch": 0.3376333746772949,
      "grad_norm": 0.0,
      "learning_rate": 3.367668593048456e-05,
      "loss": 98.9598,
      "step": 83570
    },
    {
      "epoch": 0.3376737759426625,
      "grad_norm": 2259.979248046875,
      "learning_rate": 3.367464826533225e-05,
      "loss": 77.0516,
      "step": 83580
    },
    {
      "epoch": 0.33771417720803015,
      "grad_norm": 864.2886352539062,
      "learning_rate": 3.3672610333585645e-05,
      "loss": 53.8932,
      "step": 83590
    },
    {
      "epoch": 0.3377545784733978,
      "grad_norm": 884.65087890625,
      "learning_rate": 3.3670572135284456e-05,
      "loss": 64.9674,
      "step": 83600
    },
    {
      "epoch": 0.33779497973876543,
      "grad_norm": 361.47686767578125,
      "learning_rate": 3.366853367046843e-05,
      "loss": 52.3785,
      "step": 83610
    },
    {
      "epoch": 0.3378353810041331,
      "grad_norm": 375.3967590332031,
      "learning_rate": 3.3666494939177295e-05,
      "loss": 76.4976,
      "step": 83620
    },
    {
      "epoch": 0.33787578226950066,
      "grad_norm": 454.41827392578125,
      "learning_rate": 3.366445594145081e-05,
      "loss": 62.219,
      "step": 83630
    },
    {
      "epoch": 0.3379161835348683,
      "grad_norm": 1941.0068359375,
      "learning_rate": 3.366241667732872e-05,
      "loss": 96.1385,
      "step": 83640
    },
    {
      "epoch": 0.33795658480023594,
      "grad_norm": 597.4854125976562,
      "learning_rate": 3.366037714685078e-05,
      "loss": 44.2322,
      "step": 83650
    },
    {
      "epoch": 0.3379969860656036,
      "grad_norm": 790.761962890625,
      "learning_rate": 3.365833735005676e-05,
      "loss": 42.7057,
      "step": 83660
    },
    {
      "epoch": 0.3380373873309712,
      "grad_norm": 608.0296020507812,
      "learning_rate": 3.365629728698642e-05,
      "loss": 56.8395,
      "step": 83670
    },
    {
      "epoch": 0.33807778859633886,
      "grad_norm": 1275.1829833984375,
      "learning_rate": 3.365425695767953e-05,
      "loss": 92.5405,
      "step": 83680
    },
    {
      "epoch": 0.33811818986170644,
      "grad_norm": 1744.1668701171875,
      "learning_rate": 3.365221636217588e-05,
      "loss": 57.669,
      "step": 83690
    },
    {
      "epoch": 0.3381585911270741,
      "grad_norm": 2922.179443359375,
      "learning_rate": 3.3650175500515235e-05,
      "loss": 58.4586,
      "step": 83700
    },
    {
      "epoch": 0.3381989923924417,
      "grad_norm": 976.9667358398438,
      "learning_rate": 3.364813437273739e-05,
      "loss": 68.5843,
      "step": 83710
    },
    {
      "epoch": 0.33823939365780936,
      "grad_norm": 773.35888671875,
      "learning_rate": 3.3646092978882144e-05,
      "loss": 63.5441,
      "step": 83720
    },
    {
      "epoch": 0.338279794923177,
      "grad_norm": 474.0544128417969,
      "learning_rate": 3.364405131898929e-05,
      "loss": 39.5866,
      "step": 83730
    },
    {
      "epoch": 0.33832019618854464,
      "grad_norm": 2066.063720703125,
      "learning_rate": 3.364200939309863e-05,
      "loss": 71.9654,
      "step": 83740
    },
    {
      "epoch": 0.3383605974539123,
      "grad_norm": 531.757568359375,
      "learning_rate": 3.363996720124997e-05,
      "loss": 97.3059,
      "step": 83750
    },
    {
      "epoch": 0.33840099871927987,
      "grad_norm": 498.4417419433594,
      "learning_rate": 3.363792474348313e-05,
      "loss": 84.403,
      "step": 83760
    },
    {
      "epoch": 0.3384413999846475,
      "grad_norm": 551.0558471679688,
      "learning_rate": 3.363588201983792e-05,
      "loss": 106.5099,
      "step": 83770
    },
    {
      "epoch": 0.33848180125001515,
      "grad_norm": 555.5409545898438,
      "learning_rate": 3.363383903035419e-05,
      "loss": 51.1536,
      "step": 83780
    },
    {
      "epoch": 0.3385222025153828,
      "grad_norm": 477.4742431640625,
      "learning_rate": 3.363179577507173e-05,
      "loss": 49.2929,
      "step": 83790
    },
    {
      "epoch": 0.3385626037807504,
      "grad_norm": 506.8771057128906,
      "learning_rate": 3.36297522540304e-05,
      "loss": 62.7076,
      "step": 83800
    },
    {
      "epoch": 0.33860300504611807,
      "grad_norm": 1209.50634765625,
      "learning_rate": 3.362770846727003e-05,
      "loss": 100.1989,
      "step": 83810
    },
    {
      "epoch": 0.33864340631148565,
      "grad_norm": 314.1358337402344,
      "learning_rate": 3.362566441483046e-05,
      "loss": 68.2915,
      "step": 83820
    },
    {
      "epoch": 0.3386838075768533,
      "grad_norm": 575.9443969726562,
      "learning_rate": 3.362362009675156e-05,
      "loss": 65.2,
      "step": 83830
    },
    {
      "epoch": 0.33872420884222093,
      "grad_norm": 745.6015014648438,
      "learning_rate": 3.362157551307317e-05,
      "loss": 75.2707,
      "step": 83840
    },
    {
      "epoch": 0.33876461010758857,
      "grad_norm": 658.1475830078125,
      "learning_rate": 3.361953066383515e-05,
      "loss": 55.5822,
      "step": 83850
    },
    {
      "epoch": 0.3388050113729562,
      "grad_norm": 811.0732421875,
      "learning_rate": 3.3617485549077365e-05,
      "loss": 67.6831,
      "step": 83860
    },
    {
      "epoch": 0.33884541263832385,
      "grad_norm": 914.1915893554688,
      "learning_rate": 3.3615440168839693e-05,
      "loss": 68.3131,
      "step": 83870
    },
    {
      "epoch": 0.3388858139036915,
      "grad_norm": 990.7426147460938,
      "learning_rate": 3.3613394523162e-05,
      "loss": 71.9516,
      "step": 83880
    },
    {
      "epoch": 0.3389262151690591,
      "grad_norm": 2585.360107421875,
      "learning_rate": 3.361134861208419e-05,
      "loss": 55.0049,
      "step": 83890
    },
    {
      "epoch": 0.3389666164344267,
      "grad_norm": 621.7271728515625,
      "learning_rate": 3.360930243564611e-05,
      "loss": 60.3423,
      "step": 83900
    },
    {
      "epoch": 0.33900701769979436,
      "grad_norm": 423.0601806640625,
      "learning_rate": 3.360725599388768e-05,
      "loss": 101.5376,
      "step": 83910
    },
    {
      "epoch": 0.339047418965162,
      "grad_norm": 509.084228515625,
      "learning_rate": 3.360520928684879e-05,
      "loss": 69.3444,
      "step": 83920
    },
    {
      "epoch": 0.33908782023052964,
      "grad_norm": 332.6028747558594,
      "learning_rate": 3.3603162314569334e-05,
      "loss": 36.2318,
      "step": 83930
    },
    {
      "epoch": 0.3391282214958973,
      "grad_norm": 1342.31201171875,
      "learning_rate": 3.360111507708923e-05,
      "loss": 98.992,
      "step": 83940
    },
    {
      "epoch": 0.33916862276126486,
      "grad_norm": 774.0115356445312,
      "learning_rate": 3.3599067574448385e-05,
      "loss": 32.5285,
      "step": 83950
    },
    {
      "epoch": 0.3392090240266325,
      "grad_norm": 459.2820129394531,
      "learning_rate": 3.3597019806686724e-05,
      "loss": 72.3013,
      "step": 83960
    },
    {
      "epoch": 0.33924942529200014,
      "grad_norm": 1240.106201171875,
      "learning_rate": 3.359497177384415e-05,
      "loss": 107.1222,
      "step": 83970
    },
    {
      "epoch": 0.3392898265573678,
      "grad_norm": 791.4290771484375,
      "learning_rate": 3.3592923475960604e-05,
      "loss": 73.2466,
      "step": 83980
    },
    {
      "epoch": 0.3393302278227354,
      "grad_norm": 854.4566040039062,
      "learning_rate": 3.3590874913076024e-05,
      "loss": 60.4601,
      "step": 83990
    },
    {
      "epoch": 0.33937062908810306,
      "grad_norm": 380.6976623535156,
      "learning_rate": 3.3588826085230336e-05,
      "loss": 63.7296,
      "step": 84000
    },
    {
      "epoch": 0.33941103035347064,
      "grad_norm": 372.28118896484375,
      "learning_rate": 3.3586776992463486e-05,
      "loss": 45.1759,
      "step": 84010
    },
    {
      "epoch": 0.3394514316188383,
      "grad_norm": 518.045654296875,
      "learning_rate": 3.358472763481542e-05,
      "loss": 80.7986,
      "step": 84020
    },
    {
      "epoch": 0.3394918328842059,
      "grad_norm": 1669.1104736328125,
      "learning_rate": 3.35826780123261e-05,
      "loss": 83.3932,
      "step": 84030
    },
    {
      "epoch": 0.33953223414957356,
      "grad_norm": 799.4876708984375,
      "learning_rate": 3.358062812503548e-05,
      "loss": 63.6164,
      "step": 84040
    },
    {
      "epoch": 0.3395726354149412,
      "grad_norm": 1021.8041381835938,
      "learning_rate": 3.357857797298353e-05,
      "loss": 54.8351,
      "step": 84050
    },
    {
      "epoch": 0.33961303668030884,
      "grad_norm": 1971.704345703125,
      "learning_rate": 3.35765275562102e-05,
      "loss": 57.4328,
      "step": 84060
    },
    {
      "epoch": 0.3396534379456765,
      "grad_norm": 461.5122985839844,
      "learning_rate": 3.357447687475548e-05,
      "loss": 56.3414,
      "step": 84070
    },
    {
      "epoch": 0.33969383921104407,
      "grad_norm": 1216.770263671875,
      "learning_rate": 3.357242592865934e-05,
      "loss": 57.114,
      "step": 84080
    },
    {
      "epoch": 0.3397342404764117,
      "grad_norm": 578.294189453125,
      "learning_rate": 3.357037471796178e-05,
      "loss": 56.4494,
      "step": 84090
    },
    {
      "epoch": 0.33977464174177935,
      "grad_norm": 642.9100952148438,
      "learning_rate": 3.356832324270277e-05,
      "loss": 66.4275,
      "step": 84100
    },
    {
      "epoch": 0.339815043007147,
      "grad_norm": 756.030029296875,
      "learning_rate": 3.356627150292231e-05,
      "loss": 80.5683,
      "step": 84110
    },
    {
      "epoch": 0.33985544427251463,
      "grad_norm": 575.1790161132812,
      "learning_rate": 3.356421949866041e-05,
      "loss": 69.7805,
      "step": 84120
    },
    {
      "epoch": 0.33989584553788227,
      "grad_norm": 2446.708740234375,
      "learning_rate": 3.356216722995706e-05,
      "loss": 61.9794,
      "step": 84130
    },
    {
      "epoch": 0.33993624680324985,
      "grad_norm": 699.87255859375,
      "learning_rate": 3.356011469685229e-05,
      "loss": 52.9517,
      "step": 84140
    },
    {
      "epoch": 0.3399766480686175,
      "grad_norm": 759.2429809570312,
      "learning_rate": 3.355806189938609e-05,
      "loss": 56.9572,
      "step": 84150
    },
    {
      "epoch": 0.34001704933398513,
      "grad_norm": 680.115478515625,
      "learning_rate": 3.35560088375985e-05,
      "loss": 59.5063,
      "step": 84160
    },
    {
      "epoch": 0.3400574505993528,
      "grad_norm": 1015.9598388671875,
      "learning_rate": 3.3553955511529534e-05,
      "loss": 63.287,
      "step": 84170
    },
    {
      "epoch": 0.3400978518647204,
      "grad_norm": 336.59588623046875,
      "learning_rate": 3.355190192121923e-05,
      "loss": 59.212,
      "step": 84180
    },
    {
      "epoch": 0.34013825313008805,
      "grad_norm": 1549.30810546875,
      "learning_rate": 3.3549848066707626e-05,
      "loss": 74.1864,
      "step": 84190
    },
    {
      "epoch": 0.3401786543954557,
      "grad_norm": 570.6986083984375,
      "learning_rate": 3.354779394803475e-05,
      "loss": 77.9518,
      "step": 84200
    },
    {
      "epoch": 0.3402190556608233,
      "grad_norm": 649.0238647460938,
      "learning_rate": 3.354573956524066e-05,
      "loss": 64.5057,
      "step": 84210
    },
    {
      "epoch": 0.3402594569261909,
      "grad_norm": 1301.7386474609375,
      "learning_rate": 3.3543684918365405e-05,
      "loss": 55.2643,
      "step": 84220
    },
    {
      "epoch": 0.34029985819155856,
      "grad_norm": 759.0762329101562,
      "learning_rate": 3.354163000744903e-05,
      "loss": 63.5302,
      "step": 84230
    },
    {
      "epoch": 0.3403402594569262,
      "grad_norm": 1009.452392578125,
      "learning_rate": 3.3539574832531617e-05,
      "loss": 48.2601,
      "step": 84240
    },
    {
      "epoch": 0.34038066072229384,
      "grad_norm": 591.0625,
      "learning_rate": 3.3537519393653216e-05,
      "loss": 53.8696,
      "step": 84250
    },
    {
      "epoch": 0.3404210619876615,
      "grad_norm": 1154.417724609375,
      "learning_rate": 3.353546369085391e-05,
      "loss": 94.1478,
      "step": 84260
    },
    {
      "epoch": 0.34046146325302906,
      "grad_norm": 735.6290283203125,
      "learning_rate": 3.3533407724173765e-05,
      "loss": 52.151,
      "step": 84270
    },
    {
      "epoch": 0.3405018645183967,
      "grad_norm": 1594.7706298828125,
      "learning_rate": 3.353135149365288e-05,
      "loss": 63.1956,
      "step": 84280
    },
    {
      "epoch": 0.34054226578376434,
      "grad_norm": 692.4290161132812,
      "learning_rate": 3.352929499933132e-05,
      "loss": 74.3355,
      "step": 84290
    },
    {
      "epoch": 0.340582667049132,
      "grad_norm": 1480.5269775390625,
      "learning_rate": 3.352723824124919e-05,
      "loss": 84.1563,
      "step": 84300
    },
    {
      "epoch": 0.3406230683144996,
      "grad_norm": 420.89239501953125,
      "learning_rate": 3.352518121944659e-05,
      "loss": 86.6826,
      "step": 84310
    },
    {
      "epoch": 0.34066346957986726,
      "grad_norm": 1095.89990234375,
      "learning_rate": 3.3523123933963614e-05,
      "loss": 85.8979,
      "step": 84320
    },
    {
      "epoch": 0.34070387084523485,
      "grad_norm": 546.4743041992188,
      "learning_rate": 3.352106638484038e-05,
      "loss": 62.4009,
      "step": 84330
    },
    {
      "epoch": 0.3407442721106025,
      "grad_norm": 808.0443725585938,
      "learning_rate": 3.351900857211699e-05,
      "loss": 61.3366,
      "step": 84340
    },
    {
      "epoch": 0.3407846733759701,
      "grad_norm": 2018.72314453125,
      "learning_rate": 3.351695049583357e-05,
      "loss": 63.5511,
      "step": 84350
    },
    {
      "epoch": 0.34082507464133777,
      "grad_norm": 1057.6278076171875,
      "learning_rate": 3.351489215603024e-05,
      "loss": 79.8961,
      "step": 84360
    },
    {
      "epoch": 0.3408654759067054,
      "grad_norm": 1301.7979736328125,
      "learning_rate": 3.351283355274714e-05,
      "loss": 82.8797,
      "step": 84370
    },
    {
      "epoch": 0.34090587717207305,
      "grad_norm": 529.9147338867188,
      "learning_rate": 3.351077468602438e-05,
      "loss": 66.2892,
      "step": 84380
    },
    {
      "epoch": 0.3409462784374407,
      "grad_norm": 870.2681274414062,
      "learning_rate": 3.350871555590212e-05,
      "loss": 69.3231,
      "step": 84390
    },
    {
      "epoch": 0.34098667970280827,
      "grad_norm": 1227.81298828125,
      "learning_rate": 3.350665616242049e-05,
      "loss": 62.5176,
      "step": 84400
    },
    {
      "epoch": 0.3410270809681759,
      "grad_norm": 829.910888671875,
      "learning_rate": 3.350459650561964e-05,
      "loss": 62.6331,
      "step": 84410
    },
    {
      "epoch": 0.34106748223354355,
      "grad_norm": 636.0309448242188,
      "learning_rate": 3.3502536585539746e-05,
      "loss": 42.9555,
      "step": 84420
    },
    {
      "epoch": 0.3411078834989112,
      "grad_norm": 856.1690673828125,
      "learning_rate": 3.350047640222094e-05,
      "loss": 56.321,
      "step": 84430
    },
    {
      "epoch": 0.34114828476427883,
      "grad_norm": 283.14874267578125,
      "learning_rate": 3.349841595570339e-05,
      "loss": 56.6846,
      "step": 84440
    },
    {
      "epoch": 0.34118868602964647,
      "grad_norm": 670.0524291992188,
      "learning_rate": 3.3496355246027276e-05,
      "loss": 56.418,
      "step": 84450
    },
    {
      "epoch": 0.34122908729501406,
      "grad_norm": 834.9964599609375,
      "learning_rate": 3.349429427323277e-05,
      "loss": 56.699,
      "step": 84460
    },
    {
      "epoch": 0.3412694885603817,
      "grad_norm": 2523.474853515625,
      "learning_rate": 3.349223303736005e-05,
      "loss": 88.8671,
      "step": 84470
    },
    {
      "epoch": 0.34130988982574934,
      "grad_norm": 618.8658447265625,
      "learning_rate": 3.34901715384493e-05,
      "loss": 60.285,
      "step": 84480
    },
    {
      "epoch": 0.341350291091117,
      "grad_norm": 683.580078125,
      "learning_rate": 3.3488109776540704e-05,
      "loss": 67.6615,
      "step": 84490
    },
    {
      "epoch": 0.3413906923564846,
      "grad_norm": 1387.2510986328125,
      "learning_rate": 3.3486047751674465e-05,
      "loss": 62.8798,
      "step": 84500
    },
    {
      "epoch": 0.34143109362185226,
      "grad_norm": 523.8330078125,
      "learning_rate": 3.348398546389079e-05,
      "loss": 43.4702,
      "step": 84510
    },
    {
      "epoch": 0.3414714948872199,
      "grad_norm": 972.6190795898438,
      "learning_rate": 3.3481922913229875e-05,
      "loss": 64.8293,
      "step": 84520
    },
    {
      "epoch": 0.3415118961525875,
      "grad_norm": 1023.8384399414062,
      "learning_rate": 3.347986009973193e-05,
      "loss": 80.5796,
      "step": 84530
    },
    {
      "epoch": 0.3415522974179551,
      "grad_norm": 306.61041259765625,
      "learning_rate": 3.3477797023437176e-05,
      "loss": 71.0203,
      "step": 84540
    },
    {
      "epoch": 0.34159269868332276,
      "grad_norm": 272.7121887207031,
      "learning_rate": 3.3475733684385815e-05,
      "loss": 60.9458,
      "step": 84550
    },
    {
      "epoch": 0.3416330999486904,
      "grad_norm": 648.0752563476562,
      "learning_rate": 3.3473670082618105e-05,
      "loss": 64.1713,
      "step": 84560
    },
    {
      "epoch": 0.34167350121405804,
      "grad_norm": 522.2056274414062,
      "learning_rate": 3.347160621817425e-05,
      "loss": 53.4062,
      "step": 84570
    },
    {
      "epoch": 0.3417139024794257,
      "grad_norm": 460.1097717285156,
      "learning_rate": 3.34695420910945e-05,
      "loss": 64.7991,
      "step": 84580
    },
    {
      "epoch": 0.34175430374479326,
      "grad_norm": 254.0565185546875,
      "learning_rate": 3.3467477701419095e-05,
      "loss": 42.2828,
      "step": 84590
    },
    {
      "epoch": 0.3417947050101609,
      "grad_norm": 964.922119140625,
      "learning_rate": 3.3465413049188276e-05,
      "loss": 61.1818,
      "step": 84600
    },
    {
      "epoch": 0.34183510627552854,
      "grad_norm": 1123.0037841796875,
      "learning_rate": 3.34633481344423e-05,
      "loss": 87.6611,
      "step": 84610
    },
    {
      "epoch": 0.3418755075408962,
      "grad_norm": 740.2823486328125,
      "learning_rate": 3.346128295722142e-05,
      "loss": 49.8443,
      "step": 84620
    },
    {
      "epoch": 0.3419159088062638,
      "grad_norm": 495.0751647949219,
      "learning_rate": 3.3459217517565896e-05,
      "loss": 79.6292,
      "step": 84630
    },
    {
      "epoch": 0.34195631007163146,
      "grad_norm": 970.404052734375,
      "learning_rate": 3.3457151815516e-05,
      "loss": 91.5376,
      "step": 84640
    },
    {
      "epoch": 0.34199671133699905,
      "grad_norm": 611.1907958984375,
      "learning_rate": 3.3455085851112e-05,
      "loss": 65.6393,
      "step": 84650
    },
    {
      "epoch": 0.3420371126023667,
      "grad_norm": 605.852294921875,
      "learning_rate": 3.345301962439417e-05,
      "loss": 60.7457,
      "step": 84660
    },
    {
      "epoch": 0.34207751386773433,
      "grad_norm": 1289.8677978515625,
      "learning_rate": 3.34509531354028e-05,
      "loss": 69.1039,
      "step": 84670
    },
    {
      "epoch": 0.34211791513310197,
      "grad_norm": 914.4951171875,
      "learning_rate": 3.344888638417817e-05,
      "loss": 62.5575,
      "step": 84680
    },
    {
      "epoch": 0.3421583163984696,
      "grad_norm": 518.4591674804688,
      "learning_rate": 3.3446819370760577e-05,
      "loss": 55.3549,
      "step": 84690
    },
    {
      "epoch": 0.34219871766383725,
      "grad_norm": 586.8419799804688,
      "learning_rate": 3.3444752095190326e-05,
      "loss": 65.0346,
      "step": 84700
    },
    {
      "epoch": 0.3422391189292049,
      "grad_norm": 640.1392822265625,
      "learning_rate": 3.34426845575077e-05,
      "loss": 56.1609,
      "step": 84710
    },
    {
      "epoch": 0.3422795201945725,
      "grad_norm": 680.588134765625,
      "learning_rate": 3.344061675775303e-05,
      "loss": 62.718,
      "step": 84720
    },
    {
      "epoch": 0.3423199214599401,
      "grad_norm": 848.21044921875,
      "learning_rate": 3.34385486959666e-05,
      "loss": 49.1798,
      "step": 84730
    },
    {
      "epoch": 0.34236032272530775,
      "grad_norm": 1049.4501953125,
      "learning_rate": 3.343648037218876e-05,
      "loss": 86.3738,
      "step": 84740
    },
    {
      "epoch": 0.3424007239906754,
      "grad_norm": 915.064453125,
      "learning_rate": 3.343441178645981e-05,
      "loss": 89.3447,
      "step": 84750
    },
    {
      "epoch": 0.34244112525604303,
      "grad_norm": 293.6428527832031,
      "learning_rate": 3.3432342938820084e-05,
      "loss": 54.0937,
      "step": 84760
    },
    {
      "epoch": 0.3424815265214107,
      "grad_norm": 1054.6556396484375,
      "learning_rate": 3.3430273829309925e-05,
      "loss": 65.8907,
      "step": 84770
    },
    {
      "epoch": 0.34252192778677826,
      "grad_norm": 632.8988037109375,
      "learning_rate": 3.342820445796966e-05,
      "loss": 48.7645,
      "step": 84780
    },
    {
      "epoch": 0.3425623290521459,
      "grad_norm": 835.9781494140625,
      "learning_rate": 3.342613482483963e-05,
      "loss": 58.9518,
      "step": 84790
    },
    {
      "epoch": 0.34260273031751354,
      "grad_norm": 490.72906494140625,
      "learning_rate": 3.342406492996019e-05,
      "loss": 72.7595,
      "step": 84800
    },
    {
      "epoch": 0.3426431315828812,
      "grad_norm": 931.8734741210938,
      "learning_rate": 3.34219947733717e-05,
      "loss": 56.5696,
      "step": 84810
    },
    {
      "epoch": 0.3426835328482488,
      "grad_norm": 1742.132568359375,
      "learning_rate": 3.3419924355114505e-05,
      "loss": 61.6062,
      "step": 84820
    },
    {
      "epoch": 0.34272393411361646,
      "grad_norm": 774.6658935546875,
      "learning_rate": 3.341785367522898e-05,
      "loss": 65.5544,
      "step": 84830
    },
    {
      "epoch": 0.3427643353789841,
      "grad_norm": 1044.997802734375,
      "learning_rate": 3.341578273375548e-05,
      "loss": 60.1936,
      "step": 84840
    },
    {
      "epoch": 0.3428047366443517,
      "grad_norm": 371.9859619140625,
      "learning_rate": 3.3413711530734404e-05,
      "loss": 53.2397,
      "step": 84850
    },
    {
      "epoch": 0.3428451379097193,
      "grad_norm": 535.1820678710938,
      "learning_rate": 3.341164006620611e-05,
      "loss": 83.0636,
      "step": 84860
    },
    {
      "epoch": 0.34288553917508696,
      "grad_norm": 1121.55029296875,
      "learning_rate": 3.340956834021099e-05,
      "loss": 68.9959,
      "step": 84870
    },
    {
      "epoch": 0.3429259404404546,
      "grad_norm": 783.1961059570312,
      "learning_rate": 3.340749635278942e-05,
      "loss": 62.8407,
      "step": 84880
    },
    {
      "epoch": 0.34296634170582224,
      "grad_norm": 680.4360961914062,
      "learning_rate": 3.3405424103981815e-05,
      "loss": 51.7818,
      "step": 84890
    },
    {
      "epoch": 0.3430067429711899,
      "grad_norm": 655.9916381835938,
      "learning_rate": 3.340335159382857e-05,
      "loss": 73.2568,
      "step": 84900
    },
    {
      "epoch": 0.34304714423655747,
      "grad_norm": 728.5745239257812,
      "learning_rate": 3.340127882237008e-05,
      "loss": 49.0942,
      "step": 84910
    },
    {
      "epoch": 0.3430875455019251,
      "grad_norm": 1191.2718505859375,
      "learning_rate": 3.339920578964676e-05,
      "loss": 76.6716,
      "step": 84920
    },
    {
      "epoch": 0.34312794676729275,
      "grad_norm": 656.1549072265625,
      "learning_rate": 3.339713249569902e-05,
      "loss": 64.1306,
      "step": 84930
    },
    {
      "epoch": 0.3431683480326604,
      "grad_norm": 883.1411743164062,
      "learning_rate": 3.339505894056729e-05,
      "loss": 61.9963,
      "step": 84940
    },
    {
      "epoch": 0.343208749298028,
      "grad_norm": 4564.13525390625,
      "learning_rate": 3.339298512429199e-05,
      "loss": 129.574,
      "step": 84950
    },
    {
      "epoch": 0.34324915056339567,
      "grad_norm": 726.3573608398438,
      "learning_rate": 3.339091104691355e-05,
      "loss": 70.7135,
      "step": 84960
    },
    {
      "epoch": 0.34328955182876325,
      "grad_norm": 435.5312805175781,
      "learning_rate": 3.3388836708472404e-05,
      "loss": 48.515,
      "step": 84970
    },
    {
      "epoch": 0.3433299530941309,
      "grad_norm": 852.5233154296875,
      "learning_rate": 3.3386762109009e-05,
      "loss": 38.0021,
      "step": 84980
    },
    {
      "epoch": 0.34337035435949853,
      "grad_norm": 741.4345092773438,
      "learning_rate": 3.338468724856377e-05,
      "loss": 43.5169,
      "step": 84990
    },
    {
      "epoch": 0.34341075562486617,
      "grad_norm": 523.470947265625,
      "learning_rate": 3.3382612127177166e-05,
      "loss": 89.3195,
      "step": 85000
    },
    {
      "epoch": 0.3434511568902338,
      "grad_norm": 903.5706176757812,
      "learning_rate": 3.338053674488966e-05,
      "loss": 64.1985,
      "step": 85010
    },
    {
      "epoch": 0.34349155815560145,
      "grad_norm": 694.2612915039062,
      "learning_rate": 3.3378461101741693e-05,
      "loss": 66.603,
      "step": 85020
    },
    {
      "epoch": 0.3435319594209691,
      "grad_norm": 696.1839599609375,
      "learning_rate": 3.3376385197773737e-05,
      "loss": 72.5089,
      "step": 85030
    },
    {
      "epoch": 0.3435723606863367,
      "grad_norm": 804.7989501953125,
      "learning_rate": 3.337430903302627e-05,
      "loss": 76.6745,
      "step": 85040
    },
    {
      "epoch": 0.3436127619517043,
      "grad_norm": 446.0153503417969,
      "learning_rate": 3.337223260753977e-05,
      "loss": 77.8143,
      "step": 85050
    },
    {
      "epoch": 0.34365316321707196,
      "grad_norm": 757.8631591796875,
      "learning_rate": 3.33701559213547e-05,
      "loss": 103.2349,
      "step": 85060
    },
    {
      "epoch": 0.3436935644824396,
      "grad_norm": 1979.259033203125,
      "learning_rate": 3.336807897451156e-05,
      "loss": 68.4003,
      "step": 85070
    },
    {
      "epoch": 0.34373396574780724,
      "grad_norm": 1280.52587890625,
      "learning_rate": 3.336600176705083e-05,
      "loss": 66.167,
      "step": 85080
    },
    {
      "epoch": 0.3437743670131749,
      "grad_norm": 698.9182739257812,
      "learning_rate": 3.336392429901303e-05,
      "loss": 59.4369,
      "step": 85090
    },
    {
      "epoch": 0.34381476827854246,
      "grad_norm": 1795.8751220703125,
      "learning_rate": 3.336184657043864e-05,
      "loss": 58.2561,
      "step": 85100
    },
    {
      "epoch": 0.3438551695439101,
      "grad_norm": 1120.764892578125,
      "learning_rate": 3.335976858136816e-05,
      "loss": 69.6759,
      "step": 85110
    },
    {
      "epoch": 0.34389557080927774,
      "grad_norm": 525.7089233398438,
      "learning_rate": 3.335769033184213e-05,
      "loss": 45.6366,
      "step": 85120
    },
    {
      "epoch": 0.3439359720746454,
      "grad_norm": 754.0464477539062,
      "learning_rate": 3.3355611821901046e-05,
      "loss": 62.9835,
      "step": 85130
    },
    {
      "epoch": 0.343976373340013,
      "grad_norm": 540.761962890625,
      "learning_rate": 3.335353305158543e-05,
      "loss": 80.4276,
      "step": 85140
    },
    {
      "epoch": 0.34401677460538066,
      "grad_norm": 628.1510009765625,
      "learning_rate": 3.335145402093582e-05,
      "loss": 43.522,
      "step": 85150
    },
    {
      "epoch": 0.34405717587074824,
      "grad_norm": 1078.826416015625,
      "learning_rate": 3.3349374729992725e-05,
      "loss": 59.7352,
      "step": 85160
    },
    {
      "epoch": 0.3440975771361159,
      "grad_norm": 748.0184936523438,
      "learning_rate": 3.3347295178796707e-05,
      "loss": 78.6812,
      "step": 85170
    },
    {
      "epoch": 0.3441379784014835,
      "grad_norm": 670.2138671875,
      "learning_rate": 3.334521536738829e-05,
      "loss": 35.4985,
      "step": 85180
    },
    {
      "epoch": 0.34417837966685116,
      "grad_norm": 2402.2900390625,
      "learning_rate": 3.334313529580804e-05,
      "loss": 93.7723,
      "step": 85190
    },
    {
      "epoch": 0.3442187809322188,
      "grad_norm": 780.775634765625,
      "learning_rate": 3.334105496409649e-05,
      "loss": 57.9252,
      "step": 85200
    },
    {
      "epoch": 0.34425918219758644,
      "grad_norm": 836.4694213867188,
      "learning_rate": 3.333897437229421e-05,
      "loss": 100.2123,
      "step": 85210
    },
    {
      "epoch": 0.3442995834629541,
      "grad_norm": 1626.2457275390625,
      "learning_rate": 3.333689352044175e-05,
      "loss": 69.8935,
      "step": 85220
    },
    {
      "epoch": 0.34433998472832167,
      "grad_norm": 931.92236328125,
      "learning_rate": 3.3334812408579696e-05,
      "loss": 61.816,
      "step": 85230
    },
    {
      "epoch": 0.3443803859936893,
      "grad_norm": 1187.04541015625,
      "learning_rate": 3.3332731036748604e-05,
      "loss": 61.946,
      "step": 85240
    },
    {
      "epoch": 0.34442078725905695,
      "grad_norm": 1133.5804443359375,
      "learning_rate": 3.333064940498905e-05,
      "loss": 66.1367,
      "step": 85250
    },
    {
      "epoch": 0.3444611885244246,
      "grad_norm": 639.8355102539062,
      "learning_rate": 3.332856751334163e-05,
      "loss": 105.5586,
      "step": 85260
    },
    {
      "epoch": 0.34450158978979223,
      "grad_norm": 1008.3250732421875,
      "learning_rate": 3.3326485361846924e-05,
      "loss": 52.8669,
      "step": 85270
    },
    {
      "epoch": 0.34454199105515987,
      "grad_norm": 1149.5870361328125,
      "learning_rate": 3.3324402950545524e-05,
      "loss": 93.5141,
      "step": 85280
    },
    {
      "epoch": 0.34458239232052745,
      "grad_norm": 541.025146484375,
      "learning_rate": 3.3322320279478025e-05,
      "loss": 51.9097,
      "step": 85290
    },
    {
      "epoch": 0.3446227935858951,
      "grad_norm": 925.6935424804688,
      "learning_rate": 3.332023734868504e-05,
      "loss": 58.3896,
      "step": 85300
    },
    {
      "epoch": 0.34466319485126273,
      "grad_norm": 693.147216796875,
      "learning_rate": 3.3318154158207164e-05,
      "loss": 111.9228,
      "step": 85310
    },
    {
      "epoch": 0.3447035961166304,
      "grad_norm": 0.0,
      "learning_rate": 3.3316070708085014e-05,
      "loss": 49.1727,
      "step": 85320
    },
    {
      "epoch": 0.344743997381998,
      "grad_norm": 2374.637451171875,
      "learning_rate": 3.3313986998359213e-05,
      "loss": 85.6452,
      "step": 85330
    },
    {
      "epoch": 0.34478439864736565,
      "grad_norm": 593.0366821289062,
      "learning_rate": 3.3311903029070384e-05,
      "loss": 56.1827,
      "step": 85340
    },
    {
      "epoch": 0.3448247999127333,
      "grad_norm": 583.0576782226562,
      "learning_rate": 3.330981880025915e-05,
      "loss": 35.6842,
      "step": 85350
    },
    {
      "epoch": 0.3448652011781009,
      "grad_norm": 1863.508056640625,
      "learning_rate": 3.3307734311966144e-05,
      "loss": 93.4643,
      "step": 85360
    },
    {
      "epoch": 0.3449056024434685,
      "grad_norm": 744.82763671875,
      "learning_rate": 3.330564956423201e-05,
      "loss": 69.3355,
      "step": 85370
    },
    {
      "epoch": 0.34494600370883616,
      "grad_norm": 1112.11279296875,
      "learning_rate": 3.330356455709739e-05,
      "loss": 63.4886,
      "step": 85380
    },
    {
      "epoch": 0.3449864049742038,
      "grad_norm": 581.6898803710938,
      "learning_rate": 3.3301479290602925e-05,
      "loss": 70.5977,
      "step": 85390
    },
    {
      "epoch": 0.34502680623957144,
      "grad_norm": 311.9263916015625,
      "learning_rate": 3.329939376478927e-05,
      "loss": 61.0018,
      "step": 85400
    },
    {
      "epoch": 0.3450672075049391,
      "grad_norm": 482.0248107910156,
      "learning_rate": 3.329730797969709e-05,
      "loss": 62.5195,
      "step": 85410
    },
    {
      "epoch": 0.34510760877030666,
      "grad_norm": 814.2288208007812,
      "learning_rate": 3.329522193536705e-05,
      "loss": 70.7657,
      "step": 85420
    },
    {
      "epoch": 0.3451480100356743,
      "grad_norm": 923.7318725585938,
      "learning_rate": 3.3293135631839806e-05,
      "loss": 46.5404,
      "step": 85430
    },
    {
      "epoch": 0.34518841130104194,
      "grad_norm": 913.3489990234375,
      "learning_rate": 3.329104906915604e-05,
      "loss": 60.2891,
      "step": 85440
    },
    {
      "epoch": 0.3452288125664096,
      "grad_norm": 2197.94384765625,
      "learning_rate": 3.328896224735644e-05,
      "loss": 108.9677,
      "step": 85450
    },
    {
      "epoch": 0.3452692138317772,
      "grad_norm": 636.381103515625,
      "learning_rate": 3.328687516648167e-05,
      "loss": 73.8755,
      "step": 85460
    },
    {
      "epoch": 0.34530961509714486,
      "grad_norm": 948.1078491210938,
      "learning_rate": 3.328478782657243e-05,
      "loss": 58.9634,
      "step": 85470
    },
    {
      "epoch": 0.34535001636251245,
      "grad_norm": 397.9957580566406,
      "learning_rate": 3.328270022766941e-05,
      "loss": 77.2741,
      "step": 85480
    },
    {
      "epoch": 0.3453904176278801,
      "grad_norm": 1208.4564208984375,
      "learning_rate": 3.3280612369813305e-05,
      "loss": 67.3984,
      "step": 85490
    },
    {
      "epoch": 0.3454308188932477,
      "grad_norm": 953.3515014648438,
      "learning_rate": 3.3278524253044834e-05,
      "loss": 63.2876,
      "step": 85500
    },
    {
      "epoch": 0.34547122015861537,
      "grad_norm": 663.9226684570312,
      "learning_rate": 3.327643587740469e-05,
      "loss": 62.1283,
      "step": 85510
    },
    {
      "epoch": 0.345511621423983,
      "grad_norm": 599.0784301757812,
      "learning_rate": 3.3274347242933606e-05,
      "loss": 65.7558,
      "step": 85520
    },
    {
      "epoch": 0.34555202268935065,
      "grad_norm": 1658.9404296875,
      "learning_rate": 3.327225834967227e-05,
      "loss": 50.9672,
      "step": 85530
    },
    {
      "epoch": 0.3455924239547183,
      "grad_norm": 1039.7579345703125,
      "learning_rate": 3.327016919766144e-05,
      "loss": 68.3731,
      "step": 85540
    },
    {
      "epoch": 0.34563282522008587,
      "grad_norm": 802.1036376953125,
      "learning_rate": 3.3268079786941825e-05,
      "loss": 85.3128,
      "step": 85550
    },
    {
      "epoch": 0.3456732264854535,
      "grad_norm": 2582.881103515625,
      "learning_rate": 3.326599011755416e-05,
      "loss": 84.6592,
      "step": 85560
    },
    {
      "epoch": 0.34571362775082115,
      "grad_norm": 1083.4248046875,
      "learning_rate": 3.326390018953919e-05,
      "loss": 59.0829,
      "step": 85570
    },
    {
      "epoch": 0.3457540290161888,
      "grad_norm": 580.1563720703125,
      "learning_rate": 3.326181000293766e-05,
      "loss": 61.7299,
      "step": 85580
    },
    {
      "epoch": 0.34579443028155643,
      "grad_norm": 847.4132080078125,
      "learning_rate": 3.325971955779031e-05,
      "loss": 71.2915,
      "step": 85590
    },
    {
      "epoch": 0.34583483154692407,
      "grad_norm": 621.8236083984375,
      "learning_rate": 3.325762885413791e-05,
      "loss": 70.941,
      "step": 85600
    },
    {
      "epoch": 0.34587523281229166,
      "grad_norm": 626.3765869140625,
      "learning_rate": 3.32555378920212e-05,
      "loss": 57.7146,
      "step": 85610
    },
    {
      "epoch": 0.3459156340776593,
      "grad_norm": 1419.3797607421875,
      "learning_rate": 3.325344667148095e-05,
      "loss": 71.6652,
      "step": 85620
    },
    {
      "epoch": 0.34595603534302694,
      "grad_norm": 1376.2755126953125,
      "learning_rate": 3.325135519255795e-05,
      "loss": 103.9367,
      "step": 85630
    },
    {
      "epoch": 0.3459964366083946,
      "grad_norm": 526.369140625,
      "learning_rate": 3.3249263455292954e-05,
      "loss": 42.4211,
      "step": 85640
    },
    {
      "epoch": 0.3460368378737622,
      "grad_norm": 554.4691772460938,
      "learning_rate": 3.324717145972674e-05,
      "loss": 82.0677,
      "step": 85650
    },
    {
      "epoch": 0.34607723913912986,
      "grad_norm": 609.2090454101562,
      "learning_rate": 3.32450792059001e-05,
      "loss": 52.1391,
      "step": 85660
    },
    {
      "epoch": 0.3461176404044975,
      "grad_norm": 561.9303588867188,
      "learning_rate": 3.3242986693853824e-05,
      "loss": 50.3393,
      "step": 85670
    },
    {
      "epoch": 0.3461580416698651,
      "grad_norm": 1017.120849609375,
      "learning_rate": 3.32408939236287e-05,
      "loss": 66.1674,
      "step": 85680
    },
    {
      "epoch": 0.3461984429352327,
      "grad_norm": 737.3256225585938,
      "learning_rate": 3.323880089526554e-05,
      "loss": 56.035,
      "step": 85690
    },
    {
      "epoch": 0.34623884420060036,
      "grad_norm": 769.2302856445312,
      "learning_rate": 3.323670760880513e-05,
      "loss": 79.3797,
      "step": 85700
    },
    {
      "epoch": 0.346279245465968,
      "grad_norm": 628.8507080078125,
      "learning_rate": 3.3234614064288297e-05,
      "loss": 53.0274,
      "step": 85710
    },
    {
      "epoch": 0.34631964673133564,
      "grad_norm": 970.371337890625,
      "learning_rate": 3.323252026175585e-05,
      "loss": 79.7288,
      "step": 85720
    },
    {
      "epoch": 0.3463600479967033,
      "grad_norm": 591.3244018554688,
      "learning_rate": 3.323042620124861e-05,
      "loss": 55.2425,
      "step": 85730
    },
    {
      "epoch": 0.34640044926207086,
      "grad_norm": 2088.962158203125,
      "learning_rate": 3.32283318828074e-05,
      "loss": 100.4404,
      "step": 85740
    },
    {
      "epoch": 0.3464408505274385,
      "grad_norm": 1535.4835205078125,
      "learning_rate": 3.322623730647304e-05,
      "loss": 76.5266,
      "step": 85750
    },
    {
      "epoch": 0.34648125179280614,
      "grad_norm": 614.2904052734375,
      "learning_rate": 3.322414247228638e-05,
      "loss": 58.9267,
      "step": 85760
    },
    {
      "epoch": 0.3465216530581738,
      "grad_norm": 1063.307373046875,
      "learning_rate": 3.322204738028826e-05,
      "loss": 49.7642,
      "step": 85770
    },
    {
      "epoch": 0.3465620543235414,
      "grad_norm": 449.1618957519531,
      "learning_rate": 3.321995203051951e-05,
      "loss": 51.4166,
      "step": 85780
    },
    {
      "epoch": 0.34660245558890906,
      "grad_norm": 883.9197387695312,
      "learning_rate": 3.321785642302099e-05,
      "loss": 101.6042,
      "step": 85790
    },
    {
      "epoch": 0.34664285685427665,
      "grad_norm": 1039.1463623046875,
      "learning_rate": 3.3215760557833556e-05,
      "loss": 56.6844,
      "step": 85800
    },
    {
      "epoch": 0.3466832581196443,
      "grad_norm": 425.7856750488281,
      "learning_rate": 3.3213664434998065e-05,
      "loss": 74.9194,
      "step": 85810
    },
    {
      "epoch": 0.34672365938501193,
      "grad_norm": 754.9216918945312,
      "learning_rate": 3.3211568054555384e-05,
      "loss": 67.509,
      "step": 85820
    },
    {
      "epoch": 0.34676406065037957,
      "grad_norm": 495.1691589355469,
      "learning_rate": 3.320947141654639e-05,
      "loss": 54.1455,
      "step": 85830
    },
    {
      "epoch": 0.3468044619157472,
      "grad_norm": 1658.7933349609375,
      "learning_rate": 3.320737452101194e-05,
      "loss": 71.4703,
      "step": 85840
    },
    {
      "epoch": 0.34684486318111485,
      "grad_norm": 1947.14501953125,
      "learning_rate": 3.3205277367992924e-05,
      "loss": 81.6708,
      "step": 85850
    },
    {
      "epoch": 0.3468852644464825,
      "grad_norm": 485.7032775878906,
      "learning_rate": 3.3203179957530235e-05,
      "loss": 72.3338,
      "step": 85860
    },
    {
      "epoch": 0.3469256657118501,
      "grad_norm": 1791.69140625,
      "learning_rate": 3.320108228966475e-05,
      "loss": 71.3995,
      "step": 85870
    },
    {
      "epoch": 0.3469660669772177,
      "grad_norm": 489.6419982910156,
      "learning_rate": 3.319898436443737e-05,
      "loss": 46.9738,
      "step": 85880
    },
    {
      "epoch": 0.34700646824258535,
      "grad_norm": 485.36065673828125,
      "learning_rate": 3.319688618188899e-05,
      "loss": 41.8324,
      "step": 85890
    },
    {
      "epoch": 0.347046869507953,
      "grad_norm": 1790.6739501953125,
      "learning_rate": 3.319478774206053e-05,
      "loss": 98.1587,
      "step": 85900
    },
    {
      "epoch": 0.34708727077332063,
      "grad_norm": 592.9716186523438,
      "learning_rate": 3.319268904499288e-05,
      "loss": 76.073,
      "step": 85910
    },
    {
      "epoch": 0.3471276720386883,
      "grad_norm": 1569.42431640625,
      "learning_rate": 3.3190590090726966e-05,
      "loss": 76.2268,
      "step": 85920
    },
    {
      "epoch": 0.34716807330405586,
      "grad_norm": 348.3356628417969,
      "learning_rate": 3.318849087930371e-05,
      "loss": 41.0145,
      "step": 85930
    },
    {
      "epoch": 0.3472084745694235,
      "grad_norm": 3468.679931640625,
      "learning_rate": 3.3186391410764033e-05,
      "loss": 68.0374,
      "step": 85940
    },
    {
      "epoch": 0.34724887583479114,
      "grad_norm": 432.0105285644531,
      "learning_rate": 3.3184291685148866e-05,
      "loss": 59.3231,
      "step": 85950
    },
    {
      "epoch": 0.3472892771001588,
      "grad_norm": 609.4529418945312,
      "learning_rate": 3.3182191702499146e-05,
      "loss": 95.9846,
      "step": 85960
    },
    {
      "epoch": 0.3473296783655264,
      "grad_norm": 1003.2774047851562,
      "learning_rate": 3.318009146285582e-05,
      "loss": 55.1918,
      "step": 85970
    },
    {
      "epoch": 0.34737007963089406,
      "grad_norm": 914.4887084960938,
      "learning_rate": 3.317799096625981e-05,
      "loss": 55.2846,
      "step": 85980
    },
    {
      "epoch": 0.3474104808962617,
      "grad_norm": 1035.424560546875,
      "learning_rate": 3.317589021275209e-05,
      "loss": 74.6195,
      "step": 85990
    },
    {
      "epoch": 0.3474508821616293,
      "grad_norm": 324.5082092285156,
      "learning_rate": 3.317378920237361e-05,
      "loss": 56.005,
      "step": 86000
    },
    {
      "epoch": 0.3474912834269969,
      "grad_norm": 1099.5167236328125,
      "learning_rate": 3.317168793516533e-05,
      "loss": 53.5784,
      "step": 86010
    },
    {
      "epoch": 0.34753168469236456,
      "grad_norm": 431.3697509765625,
      "learning_rate": 3.31695864111682e-05,
      "loss": 46.6206,
      "step": 86020
    },
    {
      "epoch": 0.3475720859577322,
      "grad_norm": 688.5933837890625,
      "learning_rate": 3.316748463042321e-05,
      "loss": 70.581,
      "step": 86030
    },
    {
      "epoch": 0.34761248722309984,
      "grad_norm": 625.9749145507812,
      "learning_rate": 3.316538259297133e-05,
      "loss": 78.3036,
      "step": 86040
    },
    {
      "epoch": 0.3476528884884675,
      "grad_norm": 1391.593994140625,
      "learning_rate": 3.316328029885353e-05,
      "loss": 75.0877,
      "step": 86050
    },
    {
      "epoch": 0.34769328975383507,
      "grad_norm": 423.77862548828125,
      "learning_rate": 3.3161177748110816e-05,
      "loss": 46.4944,
      "step": 86060
    },
    {
      "epoch": 0.3477336910192027,
      "grad_norm": 488.5755310058594,
      "learning_rate": 3.315907494078416e-05,
      "loss": 44.6107,
      "step": 86070
    },
    {
      "epoch": 0.34777409228457035,
      "grad_norm": 1501.919921875,
      "learning_rate": 3.315697187691456e-05,
      "loss": 123.4812,
      "step": 86080
    },
    {
      "epoch": 0.347814493549938,
      "grad_norm": 732.31884765625,
      "learning_rate": 3.315486855654302e-05,
      "loss": 48.9343,
      "step": 86090
    },
    {
      "epoch": 0.3478548948153056,
      "grad_norm": 342.4243469238281,
      "learning_rate": 3.315276497971055e-05,
      "loss": 52.2912,
      "step": 86100
    },
    {
      "epoch": 0.34789529608067327,
      "grad_norm": 1203.0311279296875,
      "learning_rate": 3.315066114645815e-05,
      "loss": 51.8036,
      "step": 86110
    },
    {
      "epoch": 0.34793569734604085,
      "grad_norm": 356.6678466796875,
      "learning_rate": 3.314855705682685e-05,
      "loss": 70.2863,
      "step": 86120
    },
    {
      "epoch": 0.3479760986114085,
      "grad_norm": 655.4275512695312,
      "learning_rate": 3.314645271085765e-05,
      "loss": 64.5816,
      "step": 86130
    },
    {
      "epoch": 0.34801649987677613,
      "grad_norm": 1261.5318603515625,
      "learning_rate": 3.3144348108591594e-05,
      "loss": 67.8096,
      "step": 86140
    },
    {
      "epoch": 0.34805690114214377,
      "grad_norm": 460.6929626464844,
      "learning_rate": 3.314224325006969e-05,
      "loss": 59.9346,
      "step": 86150
    },
    {
      "epoch": 0.3480973024075114,
      "grad_norm": 1493.7012939453125,
      "learning_rate": 3.3140138135333004e-05,
      "loss": 70.3878,
      "step": 86160
    },
    {
      "epoch": 0.34813770367287905,
      "grad_norm": 451.81988525390625,
      "learning_rate": 3.313803276442255e-05,
      "loss": 63.3886,
      "step": 86170
    },
    {
      "epoch": 0.3481781049382467,
      "grad_norm": 768.2365112304688,
      "learning_rate": 3.313592713737939e-05,
      "loss": 75.5211,
      "step": 86180
    },
    {
      "epoch": 0.3482185062036143,
      "grad_norm": 949.534912109375,
      "learning_rate": 3.3133821254244564e-05,
      "loss": 79.6422,
      "step": 86190
    },
    {
      "epoch": 0.3482589074689819,
      "grad_norm": 871.8472900390625,
      "learning_rate": 3.3131715115059134e-05,
      "loss": 59.8607,
      "step": 86200
    },
    {
      "epoch": 0.34829930873434956,
      "grad_norm": 1112.0130615234375,
      "learning_rate": 3.3129608719864154e-05,
      "loss": 53.979,
      "step": 86210
    },
    {
      "epoch": 0.3483397099997172,
      "grad_norm": 399.10418701171875,
      "learning_rate": 3.312750206870069e-05,
      "loss": 61.8568,
      "step": 86220
    },
    {
      "epoch": 0.34838011126508484,
      "grad_norm": 7023.32861328125,
      "learning_rate": 3.312539516160982e-05,
      "loss": 73.2986,
      "step": 86230
    },
    {
      "epoch": 0.3484205125304525,
      "grad_norm": 2396.18701171875,
      "learning_rate": 3.312328799863261e-05,
      "loss": 114.8817,
      "step": 86240
    },
    {
      "epoch": 0.34846091379582006,
      "grad_norm": 876.801513671875,
      "learning_rate": 3.312118057981015e-05,
      "loss": 77.4313,
      "step": 86250
    },
    {
      "epoch": 0.3485013150611877,
      "grad_norm": 485.1725769042969,
      "learning_rate": 3.311907290518352e-05,
      "loss": 53.1866,
      "step": 86260
    },
    {
      "epoch": 0.34854171632655534,
      "grad_norm": 844.3080444335938,
      "learning_rate": 3.31169649747938e-05,
      "loss": 58.6329,
      "step": 86270
    },
    {
      "epoch": 0.348582117591923,
      "grad_norm": 2072.34130859375,
      "learning_rate": 3.3114856788682105e-05,
      "loss": 79.7277,
      "step": 86280
    },
    {
      "epoch": 0.3486225188572906,
      "grad_norm": 791.1768798828125,
      "learning_rate": 3.311274834688951e-05,
      "loss": 81.8647,
      "step": 86290
    },
    {
      "epoch": 0.34866292012265826,
      "grad_norm": 617.4157104492188,
      "learning_rate": 3.3110639649457153e-05,
      "loss": 59.8066,
      "step": 86300
    },
    {
      "epoch": 0.3487033213880259,
      "grad_norm": 670.8255615234375,
      "learning_rate": 3.310853069642611e-05,
      "loss": 64.7944,
      "step": 86310
    },
    {
      "epoch": 0.3487437226533935,
      "grad_norm": 825.6277465820312,
      "learning_rate": 3.310642148783752e-05,
      "loss": 72.0322,
      "step": 86320
    },
    {
      "epoch": 0.3487841239187611,
      "grad_norm": 316.1906433105469,
      "learning_rate": 3.31043120237325e-05,
      "loss": 51.1401,
      "step": 86330
    },
    {
      "epoch": 0.34882452518412876,
      "grad_norm": 1181.583740234375,
      "learning_rate": 3.310220230415217e-05,
      "loss": 60.8622,
      "step": 86340
    },
    {
      "epoch": 0.3488649264494964,
      "grad_norm": 366.9600524902344,
      "learning_rate": 3.3100092329137654e-05,
      "loss": 51.5616,
      "step": 86350
    },
    {
      "epoch": 0.34890532771486404,
      "grad_norm": 1895.227783203125,
      "learning_rate": 3.30979820987301e-05,
      "loss": 66.7041,
      "step": 86360
    },
    {
      "epoch": 0.3489457289802317,
      "grad_norm": 1102.731201171875,
      "learning_rate": 3.3095871612970636e-05,
      "loss": 50.2599,
      "step": 86370
    },
    {
      "epoch": 0.34898613024559927,
      "grad_norm": 836.0551147460938,
      "learning_rate": 3.3093760871900414e-05,
      "loss": 59.316,
      "step": 86380
    },
    {
      "epoch": 0.3490265315109669,
      "grad_norm": 774.7025756835938,
      "learning_rate": 3.3091649875560584e-05,
      "loss": 76.6823,
      "step": 86390
    },
    {
      "epoch": 0.34906693277633455,
      "grad_norm": 800.423095703125,
      "learning_rate": 3.30895386239923e-05,
      "loss": 68.0474,
      "step": 86400
    },
    {
      "epoch": 0.3491073340417022,
      "grad_norm": 1548.18115234375,
      "learning_rate": 3.308742711723672e-05,
      "loss": 93.2495,
      "step": 86410
    },
    {
      "epoch": 0.34914773530706983,
      "grad_norm": 1229.29345703125,
      "learning_rate": 3.308531535533501e-05,
      "loss": 55.2976,
      "step": 86420
    },
    {
      "epoch": 0.34918813657243747,
      "grad_norm": 430.7838134765625,
      "learning_rate": 3.308320333832835e-05,
      "loss": 68.637,
      "step": 86430
    },
    {
      "epoch": 0.34922853783780505,
      "grad_norm": 630.6207885742188,
      "learning_rate": 3.308109106625789e-05,
      "loss": 78.5927,
      "step": 86440
    },
    {
      "epoch": 0.3492689391031727,
      "grad_norm": 395.88885498046875,
      "learning_rate": 3.307897853916483e-05,
      "loss": 72.9264,
      "step": 86450
    },
    {
      "epoch": 0.34930934036854033,
      "grad_norm": 787.671630859375,
      "learning_rate": 3.307686575709036e-05,
      "loss": 62.1931,
      "step": 86460
    },
    {
      "epoch": 0.349349741633908,
      "grad_norm": 816.1442260742188,
      "learning_rate": 3.3074752720075644e-05,
      "loss": 58.2558,
      "step": 86470
    },
    {
      "epoch": 0.3493901428992756,
      "grad_norm": 780.057861328125,
      "learning_rate": 3.30726394281619e-05,
      "loss": 49.8044,
      "step": 86480
    },
    {
      "epoch": 0.34943054416464325,
      "grad_norm": 694.2754516601562,
      "learning_rate": 3.307052588139032e-05,
      "loss": 82.6303,
      "step": 86490
    },
    {
      "epoch": 0.3494709454300109,
      "grad_norm": 641.3462524414062,
      "learning_rate": 3.3068412079802114e-05,
      "loss": 42.0418,
      "step": 86500
    },
    {
      "epoch": 0.3495113466953785,
      "grad_norm": 376.70440673828125,
      "learning_rate": 3.306629802343848e-05,
      "loss": 51.2786,
      "step": 86510
    },
    {
      "epoch": 0.3495517479607461,
      "grad_norm": 395.3877868652344,
      "learning_rate": 3.306418371234064e-05,
      "loss": 154.0648,
      "step": 86520
    },
    {
      "epoch": 0.34959214922611376,
      "grad_norm": 916.4523315429688,
      "learning_rate": 3.306206914654981e-05,
      "loss": 85.2943,
      "step": 86530
    },
    {
      "epoch": 0.3496325504914814,
      "grad_norm": 573.927978515625,
      "learning_rate": 3.305995432610722e-05,
      "loss": 51.809,
      "step": 86540
    },
    {
      "epoch": 0.34967295175684904,
      "grad_norm": 1681.1505126953125,
      "learning_rate": 3.30578392510541e-05,
      "loss": 61.1669,
      "step": 86550
    },
    {
      "epoch": 0.3497133530222167,
      "grad_norm": 1356.93017578125,
      "learning_rate": 3.305572392143168e-05,
      "loss": 130.7669,
      "step": 86560
    },
    {
      "epoch": 0.34975375428758426,
      "grad_norm": 379.07769775390625,
      "learning_rate": 3.3053608337281194e-05,
      "loss": 68.2804,
      "step": 86570
    },
    {
      "epoch": 0.3497941555529519,
      "grad_norm": 1200.1802978515625,
      "learning_rate": 3.30514924986439e-05,
      "loss": 84.0009,
      "step": 86580
    },
    {
      "epoch": 0.34983455681831954,
      "grad_norm": 1230.31591796875,
      "learning_rate": 3.3049376405561046e-05,
      "loss": 85.864,
      "step": 86590
    },
    {
      "epoch": 0.3498749580836872,
      "grad_norm": 1067.477294921875,
      "learning_rate": 3.304726005807386e-05,
      "loss": 77.4034,
      "step": 86600
    },
    {
      "epoch": 0.3499153593490548,
      "grad_norm": 649.9188232421875,
      "learning_rate": 3.304514345622364e-05,
      "loss": 48.0631,
      "step": 86610
    },
    {
      "epoch": 0.34995576061442246,
      "grad_norm": 1626.9434814453125,
      "learning_rate": 3.3043026600051624e-05,
      "loss": 60.3829,
      "step": 86620
    },
    {
      "epoch": 0.3499961618797901,
      "grad_norm": 713.0986328125,
      "learning_rate": 3.304090948959909e-05,
      "loss": 40.1923,
      "step": 86630
    },
    {
      "epoch": 0.3500365631451577,
      "grad_norm": 630.3682250976562,
      "learning_rate": 3.3038792124907325e-05,
      "loss": 69.6386,
      "step": 86640
    },
    {
      "epoch": 0.3500769644105253,
      "grad_norm": 3280.472900390625,
      "learning_rate": 3.303667450601758e-05,
      "loss": 75.8308,
      "step": 86650
    },
    {
      "epoch": 0.35011736567589297,
      "grad_norm": 839.1499633789062,
      "learning_rate": 3.303455663297116e-05,
      "loss": 64.2162,
      "step": 86660
    },
    {
      "epoch": 0.3501577669412606,
      "grad_norm": 729.3599853515625,
      "learning_rate": 3.303243850580934e-05,
      "loss": 71.651,
      "step": 86670
    },
    {
      "epoch": 0.35019816820662825,
      "grad_norm": 463.62384033203125,
      "learning_rate": 3.303032012457343e-05,
      "loss": 45.6416,
      "step": 86680
    },
    {
      "epoch": 0.3502385694719959,
      "grad_norm": 745.6293334960938,
      "learning_rate": 3.3028201489304716e-05,
      "loss": 69.6891,
      "step": 86690
    },
    {
      "epoch": 0.35027897073736347,
      "grad_norm": 1560.3077392578125,
      "learning_rate": 3.3026082600044506e-05,
      "loss": 66.435,
      "step": 86700
    },
    {
      "epoch": 0.3503193720027311,
      "grad_norm": 1437.30517578125,
      "learning_rate": 3.3023963456834115e-05,
      "loss": 93.1862,
      "step": 86710
    },
    {
      "epoch": 0.35035977326809875,
      "grad_norm": 476.25872802734375,
      "learning_rate": 3.302184405971485e-05,
      "loss": 82.2367,
      "step": 86720
    },
    {
      "epoch": 0.3504001745334664,
      "grad_norm": 1436.0726318359375,
      "learning_rate": 3.301972440872803e-05,
      "loss": 121.6115,
      "step": 86730
    },
    {
      "epoch": 0.35044057579883403,
      "grad_norm": 1528.564208984375,
      "learning_rate": 3.3017604503914976e-05,
      "loss": 55.2463,
      "step": 86740
    },
    {
      "epoch": 0.35048097706420167,
      "grad_norm": 399.49444580078125,
      "learning_rate": 3.301548434531702e-05,
      "loss": 49.284,
      "step": 86750
    },
    {
      "epoch": 0.35052137832956926,
      "grad_norm": 613.2311401367188,
      "learning_rate": 3.30133639329755e-05,
      "loss": 90.4632,
      "step": 86760
    },
    {
      "epoch": 0.3505617795949369,
      "grad_norm": 747.5999145507812,
      "learning_rate": 3.3011243266931745e-05,
      "loss": 53.8135,
      "step": 86770
    },
    {
      "epoch": 0.35060218086030454,
      "grad_norm": 1396.3726806640625,
      "learning_rate": 3.300912234722711e-05,
      "loss": 73.096,
      "step": 86780
    },
    {
      "epoch": 0.3506425821256722,
      "grad_norm": 562.4843139648438,
      "learning_rate": 3.300700117390294e-05,
      "loss": 91.8843,
      "step": 86790
    },
    {
      "epoch": 0.3506829833910398,
      "grad_norm": 684.2686157226562,
      "learning_rate": 3.300487974700058e-05,
      "loss": 38.9725,
      "step": 86800
    },
    {
      "epoch": 0.35072338465640746,
      "grad_norm": 904.96923828125,
      "learning_rate": 3.3002758066561394e-05,
      "loss": 60.236,
      "step": 86810
    },
    {
      "epoch": 0.3507637859217751,
      "grad_norm": 1734.596923828125,
      "learning_rate": 3.300063613262675e-05,
      "loss": 89.0508,
      "step": 86820
    },
    {
      "epoch": 0.3508041871871427,
      "grad_norm": 272.29534912109375,
      "learning_rate": 3.2998513945238e-05,
      "loss": 60.678,
      "step": 86830
    },
    {
      "epoch": 0.3508445884525103,
      "grad_norm": 693.614990234375,
      "learning_rate": 3.299639150443654e-05,
      "loss": 53.3913,
      "step": 86840
    },
    {
      "epoch": 0.35088498971787796,
      "grad_norm": 2130.129638671875,
      "learning_rate": 3.299426881026374e-05,
      "loss": 65.3584,
      "step": 86850
    },
    {
      "epoch": 0.3509253909832456,
      "grad_norm": 347.66632080078125,
      "learning_rate": 3.299214586276096e-05,
      "loss": 70.7766,
      "step": 86860
    },
    {
      "epoch": 0.35096579224861324,
      "grad_norm": 1611.8804931640625,
      "learning_rate": 3.2990022661969626e-05,
      "loss": 86.1092,
      "step": 86870
    },
    {
      "epoch": 0.3510061935139809,
      "grad_norm": 715.7470703125,
      "learning_rate": 3.2987899207931105e-05,
      "loss": 72.2611,
      "step": 86880
    },
    {
      "epoch": 0.35104659477934846,
      "grad_norm": 668.6033325195312,
      "learning_rate": 3.29857755006868e-05,
      "loss": 86.1486,
      "step": 86890
    },
    {
      "epoch": 0.3510869960447161,
      "grad_norm": 384.3507385253906,
      "learning_rate": 3.298365154027812e-05,
      "loss": 46.282,
      "step": 86900
    },
    {
      "epoch": 0.35112739731008374,
      "grad_norm": 488.5640869140625,
      "learning_rate": 3.298152732674647e-05,
      "loss": 42.0539,
      "step": 86910
    },
    {
      "epoch": 0.3511677985754514,
      "grad_norm": 1133.824462890625,
      "learning_rate": 3.2979402860133264e-05,
      "loss": 66.0079,
      "step": 86920
    },
    {
      "epoch": 0.351208199840819,
      "grad_norm": 788.2798461914062,
      "learning_rate": 3.297727814047991e-05,
      "loss": 51.0111,
      "step": 86930
    },
    {
      "epoch": 0.35124860110618666,
      "grad_norm": 488.20965576171875,
      "learning_rate": 3.297515316782784e-05,
      "loss": 50.3032,
      "step": 86940
    },
    {
      "epoch": 0.3512890023715543,
      "grad_norm": 794.2489624023438,
      "learning_rate": 3.297302794221849e-05,
      "loss": 65.2345,
      "step": 86950
    },
    {
      "epoch": 0.3513294036369219,
      "grad_norm": 939.7693481445312,
      "learning_rate": 3.2970902463693264e-05,
      "loss": 82.1904,
      "step": 86960
    },
    {
      "epoch": 0.35136980490228953,
      "grad_norm": 334.984375,
      "learning_rate": 3.296877673229362e-05,
      "loss": 75.9347,
      "step": 86970
    },
    {
      "epoch": 0.35141020616765717,
      "grad_norm": 1295.0634765625,
      "learning_rate": 3.2966650748061004e-05,
      "loss": 78.6464,
      "step": 86980
    },
    {
      "epoch": 0.3514506074330248,
      "grad_norm": 273.43988037109375,
      "learning_rate": 3.2964524511036856e-05,
      "loss": 44.0533,
      "step": 86990
    },
    {
      "epoch": 0.35149100869839245,
      "grad_norm": 690.1515502929688,
      "learning_rate": 3.2962398021262623e-05,
      "loss": 56.9621,
      "step": 87000
    },
    {
      "epoch": 0.3515314099637601,
      "grad_norm": 715.7372436523438,
      "learning_rate": 3.2960271278779764e-05,
      "loss": 69.2657,
      "step": 87010
    },
    {
      "epoch": 0.3515718112291277,
      "grad_norm": 443.57598876953125,
      "learning_rate": 3.295814428362975e-05,
      "loss": 60.3319,
      "step": 87020
    },
    {
      "epoch": 0.3516122124944953,
      "grad_norm": 566.7713623046875,
      "learning_rate": 3.295601703585404e-05,
      "loss": 39.7271,
      "step": 87030
    },
    {
      "epoch": 0.35165261375986295,
      "grad_norm": 1608.1795654296875,
      "learning_rate": 3.2953889535494114e-05,
      "loss": 56.2998,
      "step": 87040
    },
    {
      "epoch": 0.3516930150252306,
      "grad_norm": 1522.18505859375,
      "learning_rate": 3.295176178259143e-05,
      "loss": 71.0216,
      "step": 87050
    },
    {
      "epoch": 0.35173341629059823,
      "grad_norm": 899.77294921875,
      "learning_rate": 3.294963377718749e-05,
      "loss": 66.3725,
      "step": 87060
    },
    {
      "epoch": 0.3517738175559659,
      "grad_norm": 425.5625,
      "learning_rate": 3.294750551932377e-05,
      "loss": 83.7303,
      "step": 87070
    },
    {
      "epoch": 0.35181421882133346,
      "grad_norm": 1519.5233154296875,
      "learning_rate": 3.294537700904177e-05,
      "loss": 51.6581,
      "step": 87080
    },
    {
      "epoch": 0.3518546200867011,
      "grad_norm": 389.0169677734375,
      "learning_rate": 3.294324824638297e-05,
      "loss": 88.9213,
      "step": 87090
    },
    {
      "epoch": 0.35189502135206874,
      "grad_norm": 791.9541625976562,
      "learning_rate": 3.294111923138889e-05,
      "loss": 64.3564,
      "step": 87100
    },
    {
      "epoch": 0.3519354226174364,
      "grad_norm": 1049.439208984375,
      "learning_rate": 3.2938989964101026e-05,
      "loss": 91.8242,
      "step": 87110
    },
    {
      "epoch": 0.351975823882804,
      "grad_norm": 1153.3204345703125,
      "learning_rate": 3.293686044456089e-05,
      "loss": 77.8267,
      "step": 87120
    },
    {
      "epoch": 0.35201622514817166,
      "grad_norm": 607.4779052734375,
      "learning_rate": 3.293473067281e-05,
      "loss": 48.7235,
      "step": 87130
    },
    {
      "epoch": 0.3520566264135393,
      "grad_norm": 516.1886596679688,
      "learning_rate": 3.293260064888988e-05,
      "loss": 73.5357,
      "step": 87140
    },
    {
      "epoch": 0.3520970276789069,
      "grad_norm": 231.58335876464844,
      "learning_rate": 3.293047037284205e-05,
      "loss": 40.6186,
      "step": 87150
    },
    {
      "epoch": 0.3521374289442745,
      "grad_norm": 897.0509033203125,
      "learning_rate": 3.292833984470804e-05,
      "loss": 81.0548,
      "step": 87160
    },
    {
      "epoch": 0.35217783020964216,
      "grad_norm": 512.804443359375,
      "learning_rate": 3.2926209064529384e-05,
      "loss": 106.7156,
      "step": 87170
    },
    {
      "epoch": 0.3522182314750098,
      "grad_norm": 498.0970458984375,
      "learning_rate": 3.292407803234763e-05,
      "loss": 67.021,
      "step": 87180
    },
    {
      "epoch": 0.35225863274037744,
      "grad_norm": 656.015625,
      "learning_rate": 3.292194674820433e-05,
      "loss": 74.4482,
      "step": 87190
    },
    {
      "epoch": 0.3522990340057451,
      "grad_norm": 757.9091186523438,
      "learning_rate": 3.2919815212141025e-05,
      "loss": 58.3926,
      "step": 87200
    },
    {
      "epoch": 0.35233943527111267,
      "grad_norm": 521.5278930664062,
      "learning_rate": 3.2917683424199255e-05,
      "loss": 76.3875,
      "step": 87210
    },
    {
      "epoch": 0.3523798365364803,
      "grad_norm": 1062.794677734375,
      "learning_rate": 3.291555138442062e-05,
      "loss": 62.0553,
      "step": 87220
    },
    {
      "epoch": 0.35242023780184795,
      "grad_norm": 464.8687438964844,
      "learning_rate": 3.291341909284664e-05,
      "loss": 57.1597,
      "step": 87230
    },
    {
      "epoch": 0.3524606390672156,
      "grad_norm": 1125.7791748046875,
      "learning_rate": 3.291128654951892e-05,
      "loss": 64.3481,
      "step": 87240
    },
    {
      "epoch": 0.3525010403325832,
      "grad_norm": 1071.926513671875,
      "learning_rate": 3.290915375447902e-05,
      "loss": 78.0845,
      "step": 87250
    },
    {
      "epoch": 0.35254144159795087,
      "grad_norm": 1123.571533203125,
      "learning_rate": 3.290702070776851e-05,
      "loss": 57.5495,
      "step": 87260
    },
    {
      "epoch": 0.3525818428633185,
      "grad_norm": 991.531005859375,
      "learning_rate": 3.2904887409429e-05,
      "loss": 51.8717,
      "step": 87270
    },
    {
      "epoch": 0.3526222441286861,
      "grad_norm": 1023.5750122070312,
      "learning_rate": 3.2902753859502056e-05,
      "loss": 40.4581,
      "step": 87280
    },
    {
      "epoch": 0.35266264539405373,
      "grad_norm": 751.7762451171875,
      "learning_rate": 3.290062005802929e-05,
      "loss": 61.994,
      "step": 87290
    },
    {
      "epoch": 0.35270304665942137,
      "grad_norm": 250.3064422607422,
      "learning_rate": 3.289848600505229e-05,
      "loss": 54.284,
      "step": 87300
    },
    {
      "epoch": 0.352743447924789,
      "grad_norm": 519.025146484375,
      "learning_rate": 3.289635170061267e-05,
      "loss": 82.9329,
      "step": 87310
    },
    {
      "epoch": 0.35278384919015665,
      "grad_norm": 849.8499755859375,
      "learning_rate": 3.289421714475203e-05,
      "loss": 72.7599,
      "step": 87320
    },
    {
      "epoch": 0.3528242504555243,
      "grad_norm": 1280.5404052734375,
      "learning_rate": 3.289208233751199e-05,
      "loss": 60.5242,
      "step": 87330
    },
    {
      "epoch": 0.3528646517208919,
      "grad_norm": 854.2972412109375,
      "learning_rate": 3.288994727893416e-05,
      "loss": 73.9743,
      "step": 87340
    },
    {
      "epoch": 0.3529050529862595,
      "grad_norm": 1820.94482421875,
      "learning_rate": 3.2887811969060184e-05,
      "loss": 43.6074,
      "step": 87350
    },
    {
      "epoch": 0.35294545425162716,
      "grad_norm": 745.350341796875,
      "learning_rate": 3.288567640793167e-05,
      "loss": 118.5294,
      "step": 87360
    },
    {
      "epoch": 0.3529858555169948,
      "grad_norm": 1059.162109375,
      "learning_rate": 3.288354059559026e-05,
      "loss": 62.4829,
      "step": 87370
    },
    {
      "epoch": 0.35302625678236244,
      "grad_norm": 267.9883728027344,
      "learning_rate": 3.28814045320776e-05,
      "loss": 47.2061,
      "step": 87380
    },
    {
      "epoch": 0.3530666580477301,
      "grad_norm": 870.7761840820312,
      "learning_rate": 3.287926821743532e-05,
      "loss": 68.1026,
      "step": 87390
    },
    {
      "epoch": 0.35310705931309766,
      "grad_norm": 1524.5283203125,
      "learning_rate": 3.287713165170508e-05,
      "loss": 70.0073,
      "step": 87400
    },
    {
      "epoch": 0.3531474605784653,
      "grad_norm": 1669.3951416015625,
      "learning_rate": 3.2874994834928524e-05,
      "loss": 42.2512,
      "step": 87410
    },
    {
      "epoch": 0.35318786184383294,
      "grad_norm": 862.4916381835938,
      "learning_rate": 3.2872857767147316e-05,
      "loss": 54.6232,
      "step": 87420
    },
    {
      "epoch": 0.3532282631092006,
      "grad_norm": 201.9671630859375,
      "learning_rate": 3.2870720448403127e-05,
      "loss": 68.1916,
      "step": 87430
    },
    {
      "epoch": 0.3532686643745682,
      "grad_norm": 894.3649291992188,
      "learning_rate": 3.286858287873761e-05,
      "loss": 84.1714,
      "step": 87440
    },
    {
      "epoch": 0.35330906563993586,
      "grad_norm": 512.4931640625,
      "learning_rate": 3.286644505819244e-05,
      "loss": 33.4036,
      "step": 87450
    },
    {
      "epoch": 0.3533494669053035,
      "grad_norm": 853.0204467773438,
      "learning_rate": 3.286430698680931e-05,
      "loss": 50.9634,
      "step": 87460
    },
    {
      "epoch": 0.3533898681706711,
      "grad_norm": 2792.16455078125,
      "learning_rate": 3.286216866462988e-05,
      "loss": 67.5983,
      "step": 87470
    },
    {
      "epoch": 0.3534302694360387,
      "grad_norm": 479.12408447265625,
      "learning_rate": 3.286003009169586e-05,
      "loss": 43.4239,
      "step": 87480
    },
    {
      "epoch": 0.35347067070140636,
      "grad_norm": 424.5600891113281,
      "learning_rate": 3.285789126804893e-05,
      "loss": 89.1872,
      "step": 87490
    },
    {
      "epoch": 0.353511071966774,
      "grad_norm": 921.9595947265625,
      "learning_rate": 3.285575219373079e-05,
      "loss": 38.4102,
      "step": 87500
    },
    {
      "epoch": 0.35355147323214164,
      "grad_norm": 765.3430786132812,
      "learning_rate": 3.285361286878314e-05,
      "loss": 76.0972,
      "step": 87510
    },
    {
      "epoch": 0.3535918744975093,
      "grad_norm": 561.0750122070312,
      "learning_rate": 3.2851473293247694e-05,
      "loss": 52.4396,
      "step": 87520
    },
    {
      "epoch": 0.35363227576287687,
      "grad_norm": 505.8731994628906,
      "learning_rate": 3.2849333467166156e-05,
      "loss": 81.7981,
      "step": 87530
    },
    {
      "epoch": 0.3536726770282445,
      "grad_norm": 926.2390747070312,
      "learning_rate": 3.284719339058025e-05,
      "loss": 50.5313,
      "step": 87540
    },
    {
      "epoch": 0.35371307829361215,
      "grad_norm": 870.8189697265625,
      "learning_rate": 3.284505306353169e-05,
      "loss": 66.9499,
      "step": 87550
    },
    {
      "epoch": 0.3537534795589798,
      "grad_norm": 860.2401733398438,
      "learning_rate": 3.284291248606221e-05,
      "loss": 81.4307,
      "step": 87560
    },
    {
      "epoch": 0.35379388082434743,
      "grad_norm": 335.7196044921875,
      "learning_rate": 3.284077165821354e-05,
      "loss": 69.8784,
      "step": 87570
    },
    {
      "epoch": 0.35383428208971507,
      "grad_norm": 1089.861083984375,
      "learning_rate": 3.2838630580027416e-05,
      "loss": 54.2658,
      "step": 87580
    },
    {
      "epoch": 0.3538746833550827,
      "grad_norm": 546.7323608398438,
      "learning_rate": 3.2836489251545576e-05,
      "loss": 46.5775,
      "step": 87590
    },
    {
      "epoch": 0.3539150846204503,
      "grad_norm": 283.970458984375,
      "learning_rate": 3.2834347672809776e-05,
      "loss": 70.2304,
      "step": 87600
    },
    {
      "epoch": 0.35395548588581793,
      "grad_norm": 841.40576171875,
      "learning_rate": 3.283220584386175e-05,
      "loss": 63.9377,
      "step": 87610
    },
    {
      "epoch": 0.3539958871511856,
      "grad_norm": 597.7101440429688,
      "learning_rate": 3.283006376474327e-05,
      "loss": 58.1834,
      "step": 87620
    },
    {
      "epoch": 0.3540362884165532,
      "grad_norm": 344.36767578125,
      "learning_rate": 3.2827921435496097e-05,
      "loss": 51.8543,
      "step": 87630
    },
    {
      "epoch": 0.35407668968192085,
      "grad_norm": 778.3963623046875,
      "learning_rate": 3.2825778856161984e-05,
      "loss": 83.5701,
      "step": 87640
    },
    {
      "epoch": 0.3541170909472885,
      "grad_norm": 414.31817626953125,
      "learning_rate": 3.2823636026782715e-05,
      "loss": 43.3595,
      "step": 87650
    },
    {
      "epoch": 0.3541574922126561,
      "grad_norm": 983.6788330078125,
      "learning_rate": 3.282149294740005e-05,
      "loss": 58.7744,
      "step": 87660
    },
    {
      "epoch": 0.3541978934780237,
      "grad_norm": 613.5213012695312,
      "learning_rate": 3.2819349618055784e-05,
      "loss": 62.683,
      "step": 87670
    },
    {
      "epoch": 0.35423829474339136,
      "grad_norm": 985.743408203125,
      "learning_rate": 3.28172060387917e-05,
      "loss": 63.0019,
      "step": 87680
    },
    {
      "epoch": 0.354278696008759,
      "grad_norm": 1007.1724243164062,
      "learning_rate": 3.2815062209649585e-05,
      "loss": 76.2185,
      "step": 87690
    },
    {
      "epoch": 0.35431909727412664,
      "grad_norm": 2079.40087890625,
      "learning_rate": 3.281291813067123e-05,
      "loss": 104.9009,
      "step": 87700
    },
    {
      "epoch": 0.3543594985394943,
      "grad_norm": 1013.1952514648438,
      "learning_rate": 3.2810773801898445e-05,
      "loss": 70.3131,
      "step": 87710
    },
    {
      "epoch": 0.35439989980486186,
      "grad_norm": 685.0365600585938,
      "learning_rate": 3.2808629223373026e-05,
      "loss": 68.3621,
      "step": 87720
    },
    {
      "epoch": 0.3544403010702295,
      "grad_norm": 1584.983642578125,
      "learning_rate": 3.280648439513679e-05,
      "loss": 82.7203,
      "step": 87730
    },
    {
      "epoch": 0.35448070233559714,
      "grad_norm": 607.6953735351562,
      "learning_rate": 3.2804339317231545e-05,
      "loss": 53.9031,
      "step": 87740
    },
    {
      "epoch": 0.3545211036009648,
      "grad_norm": 3307.65234375,
      "learning_rate": 3.2802193989699116e-05,
      "loss": 83.5846,
      "step": 87750
    },
    {
      "epoch": 0.3545615048663324,
      "grad_norm": 792.563720703125,
      "learning_rate": 3.2800048412581315e-05,
      "loss": 49.468,
      "step": 87760
    },
    {
      "epoch": 0.35460190613170006,
      "grad_norm": 404.6004333496094,
      "learning_rate": 3.279790258591999e-05,
      "loss": 60.5075,
      "step": 87770
    },
    {
      "epoch": 0.3546423073970677,
      "grad_norm": 1089.806640625,
      "learning_rate": 3.279575650975696e-05,
      "loss": 61.289,
      "step": 87780
    },
    {
      "epoch": 0.3546827086624353,
      "grad_norm": 544.2926635742188,
      "learning_rate": 3.279361018413407e-05,
      "loss": 55.3457,
      "step": 87790
    },
    {
      "epoch": 0.3547231099278029,
      "grad_norm": 518.1968383789062,
      "learning_rate": 3.2791463609093164e-05,
      "loss": 77.1913,
      "step": 87800
    },
    {
      "epoch": 0.35476351119317057,
      "grad_norm": 690.2503662109375,
      "learning_rate": 3.278931678467609e-05,
      "loss": 87.9437,
      "step": 87810
    },
    {
      "epoch": 0.3548039124585382,
      "grad_norm": 540.9448852539062,
      "learning_rate": 3.27871697109247e-05,
      "loss": 60.3074,
      "step": 87820
    },
    {
      "epoch": 0.35484431372390585,
      "grad_norm": 465.333984375,
      "learning_rate": 3.2785022387880854e-05,
      "loss": 64.1311,
      "step": 87830
    },
    {
      "epoch": 0.3548847149892735,
      "grad_norm": 807.2095947265625,
      "learning_rate": 3.278287481558641e-05,
      "loss": 70.1779,
      "step": 87840
    },
    {
      "epoch": 0.35492511625464107,
      "grad_norm": 627.5873413085938,
      "learning_rate": 3.278072699408324e-05,
      "loss": 72.7112,
      "step": 87850
    },
    {
      "epoch": 0.3549655175200087,
      "grad_norm": 3926.626953125,
      "learning_rate": 3.2778578923413226e-05,
      "loss": 61.1929,
      "step": 87860
    },
    {
      "epoch": 0.35500591878537635,
      "grad_norm": 671.0523071289062,
      "learning_rate": 3.2776430603618225e-05,
      "loss": 71.2781,
      "step": 87870
    },
    {
      "epoch": 0.355046320050744,
      "grad_norm": 821.95947265625,
      "learning_rate": 3.277428203474014e-05,
      "loss": 65.0558,
      "step": 87880
    },
    {
      "epoch": 0.35508672131611163,
      "grad_norm": 275.886474609375,
      "learning_rate": 3.277213321682085e-05,
      "loss": 41.9689,
      "step": 87890
    },
    {
      "epoch": 0.35512712258147927,
      "grad_norm": 592.1325073242188,
      "learning_rate": 3.276998414990225e-05,
      "loss": 78.3772,
      "step": 87900
    },
    {
      "epoch": 0.3551675238468469,
      "grad_norm": 463.1005859375,
      "learning_rate": 3.276783483402623e-05,
      "loss": 84.9977,
      "step": 87910
    },
    {
      "epoch": 0.3552079251122145,
      "grad_norm": 750.5401611328125,
      "learning_rate": 3.27656852692347e-05,
      "loss": 33.9231,
      "step": 87920
    },
    {
      "epoch": 0.35524832637758214,
      "grad_norm": 456.62890625,
      "learning_rate": 3.276353545556956e-05,
      "loss": 65.4011,
      "step": 87930
    },
    {
      "epoch": 0.3552887276429498,
      "grad_norm": 1070.725830078125,
      "learning_rate": 3.276138539307273e-05,
      "loss": 54.5551,
      "step": 87940
    },
    {
      "epoch": 0.3553291289083174,
      "grad_norm": 547.5213623046875,
      "learning_rate": 3.275923508178611e-05,
      "loss": 92.1273,
      "step": 87950
    },
    {
      "epoch": 0.35536953017368506,
      "grad_norm": 3793.371337890625,
      "learning_rate": 3.275708452175164e-05,
      "loss": 86.2895,
      "step": 87960
    },
    {
      "epoch": 0.3554099314390527,
      "grad_norm": 532.2041625976562,
      "learning_rate": 3.2754933713011245e-05,
      "loss": 62.463,
      "step": 87970
    },
    {
      "epoch": 0.3554503327044203,
      "grad_norm": 380.2835693359375,
      "learning_rate": 3.275278265560684e-05,
      "loss": 57.4501,
      "step": 87980
    },
    {
      "epoch": 0.3554907339697879,
      "grad_norm": 411.328369140625,
      "learning_rate": 3.275063134958038e-05,
      "loss": 66.6874,
      "step": 87990
    },
    {
      "epoch": 0.35553113523515556,
      "grad_norm": 804.7635498046875,
      "learning_rate": 3.27484797949738e-05,
      "loss": 56.3235,
      "step": 88000
    },
    {
      "epoch": 0.3555715365005232,
      "grad_norm": 354.7542419433594,
      "learning_rate": 3.274632799182904e-05,
      "loss": 49.5995,
      "step": 88010
    },
    {
      "epoch": 0.35561193776589084,
      "grad_norm": 1237.0531005859375,
      "learning_rate": 3.274417594018805e-05,
      "loss": 59.8406,
      "step": 88020
    },
    {
      "epoch": 0.3556523390312585,
      "grad_norm": 0.0,
      "learning_rate": 3.2742023640092785e-05,
      "loss": 54.7096,
      "step": 88030
    },
    {
      "epoch": 0.35569274029662606,
      "grad_norm": 1007.9598388671875,
      "learning_rate": 3.2739871091585216e-05,
      "loss": 78.0177,
      "step": 88040
    },
    {
      "epoch": 0.3557331415619937,
      "grad_norm": 811.9549560546875,
      "learning_rate": 3.27377182947073e-05,
      "loss": 52.268,
      "step": 88050
    },
    {
      "epoch": 0.35577354282736134,
      "grad_norm": 741.9793701171875,
      "learning_rate": 3.2735565249501005e-05,
      "loss": 78.049,
      "step": 88060
    },
    {
      "epoch": 0.355813944092729,
      "grad_norm": 1019.4302368164062,
      "learning_rate": 3.2733411956008314e-05,
      "loss": 59.1742,
      "step": 88070
    },
    {
      "epoch": 0.3558543453580966,
      "grad_norm": 1513.0257568359375,
      "learning_rate": 3.27312584142712e-05,
      "loss": 82.382,
      "step": 88080
    },
    {
      "epoch": 0.35589474662346426,
      "grad_norm": 1924.9989013671875,
      "learning_rate": 3.2729104624331643e-05,
      "loss": 66.7991,
      "step": 88090
    },
    {
      "epoch": 0.3559351478888319,
      "grad_norm": 439.1302185058594,
      "learning_rate": 3.272695058623165e-05,
      "loss": 73.8741,
      "step": 88100
    },
    {
      "epoch": 0.3559755491541995,
      "grad_norm": 534.2994995117188,
      "learning_rate": 3.272479630001319e-05,
      "loss": 69.6003,
      "step": 88110
    },
    {
      "epoch": 0.35601595041956713,
      "grad_norm": 1121.692626953125,
      "learning_rate": 3.272264176571828e-05,
      "loss": 67.2641,
      "step": 88120
    },
    {
      "epoch": 0.35605635168493477,
      "grad_norm": 732.5870361328125,
      "learning_rate": 3.272048698338892e-05,
      "loss": 94.3675,
      "step": 88130
    },
    {
      "epoch": 0.3560967529503024,
      "grad_norm": 765.2777099609375,
      "learning_rate": 3.271833195306711e-05,
      "loss": 54.8232,
      "step": 88140
    },
    {
      "epoch": 0.35613715421567005,
      "grad_norm": 1179.7957763671875,
      "learning_rate": 3.271617667479489e-05,
      "loss": 105.0848,
      "step": 88150
    },
    {
      "epoch": 0.3561775554810377,
      "grad_norm": 1157.9771728515625,
      "learning_rate": 3.271402114861424e-05,
      "loss": 105.5937,
      "step": 88160
    },
    {
      "epoch": 0.3562179567464053,
      "grad_norm": 701.4404907226562,
      "learning_rate": 3.271186537456721e-05,
      "loss": 50.7279,
      "step": 88170
    },
    {
      "epoch": 0.3562583580117729,
      "grad_norm": 722.4993896484375,
      "learning_rate": 3.270970935269582e-05,
      "loss": 51.4809,
      "step": 88180
    },
    {
      "epoch": 0.35629875927714055,
      "grad_norm": 1130.43408203125,
      "learning_rate": 3.27075530830421e-05,
      "loss": 60.8365,
      "step": 88190
    },
    {
      "epoch": 0.3563391605425082,
      "grad_norm": 1566.13427734375,
      "learning_rate": 3.270539656564809e-05,
      "loss": 63.4244,
      "step": 88200
    },
    {
      "epoch": 0.35637956180787583,
      "grad_norm": 599.3772583007812,
      "learning_rate": 3.270323980055583e-05,
      "loss": 58.7716,
      "step": 88210
    },
    {
      "epoch": 0.3564199630732435,
      "grad_norm": 2744.5830078125,
      "learning_rate": 3.270108278780738e-05,
      "loss": 94.2214,
      "step": 88220
    },
    {
      "epoch": 0.35646036433861106,
      "grad_norm": 621.2742309570312,
      "learning_rate": 3.2698925527444775e-05,
      "loss": 65.0944,
      "step": 88230
    },
    {
      "epoch": 0.3565007656039787,
      "grad_norm": 720.038818359375,
      "learning_rate": 3.269676801951008e-05,
      "loss": 108.5621,
      "step": 88240
    },
    {
      "epoch": 0.35654116686934634,
      "grad_norm": 2238.180419921875,
      "learning_rate": 3.2694610264045355e-05,
      "loss": 59.7955,
      "step": 88250
    },
    {
      "epoch": 0.356581568134714,
      "grad_norm": 478.5419921875,
      "learning_rate": 3.269245226109267e-05,
      "loss": 41.493,
      "step": 88260
    },
    {
      "epoch": 0.3566219694000816,
      "grad_norm": 315.3074951171875,
      "learning_rate": 3.269029401069409e-05,
      "loss": 46.2351,
      "step": 88270
    },
    {
      "epoch": 0.35666237066544926,
      "grad_norm": 363.5831604003906,
      "learning_rate": 3.2688135512891696e-05,
      "loss": 52.8951,
      "step": 88280
    },
    {
      "epoch": 0.3567027719308169,
      "grad_norm": 240.89109802246094,
      "learning_rate": 3.268597676772757e-05,
      "loss": 58.7392,
      "step": 88290
    },
    {
      "epoch": 0.3567431731961845,
      "grad_norm": 657.1890869140625,
      "learning_rate": 3.268381777524379e-05,
      "loss": 61.5428,
      "step": 88300
    },
    {
      "epoch": 0.3567835744615521,
      "grad_norm": 1022.2767944335938,
      "learning_rate": 3.268165853548247e-05,
      "loss": 114.9119,
      "step": 88310
    },
    {
      "epoch": 0.35682397572691976,
      "grad_norm": 1009.5117797851562,
      "learning_rate": 3.2679499048485665e-05,
      "loss": 72.1214,
      "step": 88320
    },
    {
      "epoch": 0.3568643769922874,
      "grad_norm": 748.3343505859375,
      "learning_rate": 3.267733931429551e-05,
      "loss": 48.6923,
      "step": 88330
    },
    {
      "epoch": 0.35690477825765504,
      "grad_norm": 1002.4199829101562,
      "learning_rate": 3.2675179332954094e-05,
      "loss": 87.0557,
      "step": 88340
    },
    {
      "epoch": 0.3569451795230227,
      "grad_norm": 543.9052734375,
      "learning_rate": 3.267301910450353e-05,
      "loss": 68.6447,
      "step": 88350
    },
    {
      "epoch": 0.35698558078839027,
      "grad_norm": 585.1415405273438,
      "learning_rate": 3.267085862898594e-05,
      "loss": 60.3965,
      "step": 88360
    },
    {
      "epoch": 0.3570259820537579,
      "grad_norm": 664.0950317382812,
      "learning_rate": 3.266869790644344e-05,
      "loss": 57.0627,
      "step": 88370
    },
    {
      "epoch": 0.35706638331912555,
      "grad_norm": 455.95172119140625,
      "learning_rate": 3.266653693691814e-05,
      "loss": 59.9233,
      "step": 88380
    },
    {
      "epoch": 0.3571067845844932,
      "grad_norm": 621.0262451171875,
      "learning_rate": 3.266437572045219e-05,
      "loss": 63.693,
      "step": 88390
    },
    {
      "epoch": 0.3571471858498608,
      "grad_norm": 1385.278076171875,
      "learning_rate": 3.266221425708771e-05,
      "loss": 56.2835,
      "step": 88400
    },
    {
      "epoch": 0.35718758711522847,
      "grad_norm": 479.0565490722656,
      "learning_rate": 3.266005254686686e-05,
      "loss": 62.6534,
      "step": 88410
    },
    {
      "epoch": 0.3572279883805961,
      "grad_norm": 901.646484375,
      "learning_rate": 3.265789058983175e-05,
      "loss": 55.1813,
      "step": 88420
    },
    {
      "epoch": 0.3572683896459637,
      "grad_norm": 837.3994750976562,
      "learning_rate": 3.265572838602455e-05,
      "loss": 65.0508,
      "step": 88430
    },
    {
      "epoch": 0.35730879091133133,
      "grad_norm": 1183.2374267578125,
      "learning_rate": 3.265356593548741e-05,
      "loss": 60.1707,
      "step": 88440
    },
    {
      "epoch": 0.35734919217669897,
      "grad_norm": 556.451416015625,
      "learning_rate": 3.265140323826249e-05,
      "loss": 56.4609,
      "step": 88450
    },
    {
      "epoch": 0.3573895934420666,
      "grad_norm": 938.9976196289062,
      "learning_rate": 3.264924029439195e-05,
      "loss": 85.0968,
      "step": 88460
    },
    {
      "epoch": 0.35742999470743425,
      "grad_norm": 1263.639404296875,
      "learning_rate": 3.264707710391796e-05,
      "loss": 55.0165,
      "step": 88470
    },
    {
      "epoch": 0.3574703959728019,
      "grad_norm": 419.4474182128906,
      "learning_rate": 3.264491366688269e-05,
      "loss": 47.9041,
      "step": 88480
    },
    {
      "epoch": 0.3575107972381695,
      "grad_norm": 1374.3382568359375,
      "learning_rate": 3.264274998332831e-05,
      "loss": 62.4838,
      "step": 88490
    },
    {
      "epoch": 0.3575511985035371,
      "grad_norm": 1557.05078125,
      "learning_rate": 3.264058605329702e-05,
      "loss": 59.3793,
      "step": 88500
    },
    {
      "epoch": 0.35759159976890476,
      "grad_norm": 843.0123291015625,
      "learning_rate": 3.2638421876831e-05,
      "loss": 75.1318,
      "step": 88510
    },
    {
      "epoch": 0.3576320010342724,
      "grad_norm": 518.8107299804688,
      "learning_rate": 3.2636257453972424e-05,
      "loss": 74.0245,
      "step": 88520
    },
    {
      "epoch": 0.35767240229964004,
      "grad_norm": 714.6971435546875,
      "learning_rate": 3.2634092784763515e-05,
      "loss": 74.727,
      "step": 88530
    },
    {
      "epoch": 0.3577128035650077,
      "grad_norm": 1000.6592407226562,
      "learning_rate": 3.2631927869246456e-05,
      "loss": 75.3208,
      "step": 88540
    },
    {
      "epoch": 0.35775320483037526,
      "grad_norm": 361.2498779296875,
      "learning_rate": 3.2629762707463466e-05,
      "loss": 52.9834,
      "step": 88550
    },
    {
      "epoch": 0.3577936060957429,
      "grad_norm": 305.6169128417969,
      "learning_rate": 3.2627597299456746e-05,
      "loss": 69.2028,
      "step": 88560
    },
    {
      "epoch": 0.35783400736111054,
      "grad_norm": 413.1781311035156,
      "learning_rate": 3.262543164526852e-05,
      "loss": 62.5969,
      "step": 88570
    },
    {
      "epoch": 0.3578744086264782,
      "grad_norm": 572.3367309570312,
      "learning_rate": 3.2623265744941e-05,
      "loss": 76.3019,
      "step": 88580
    },
    {
      "epoch": 0.3579148098918458,
      "grad_norm": 896.4129638671875,
      "learning_rate": 3.262109959851642e-05,
      "loss": 58.3669,
      "step": 88590
    },
    {
      "epoch": 0.35795521115721346,
      "grad_norm": 1135.0291748046875,
      "learning_rate": 3.2618933206036994e-05,
      "loss": 64.4839,
      "step": 88600
    },
    {
      "epoch": 0.3579956124225811,
      "grad_norm": 4038.096435546875,
      "learning_rate": 3.2616766567544976e-05,
      "loss": 94.1078,
      "step": 88610
    },
    {
      "epoch": 0.3580360136879487,
      "grad_norm": 469.45233154296875,
      "learning_rate": 3.26145996830826e-05,
      "loss": 68.6764,
      "step": 88620
    },
    {
      "epoch": 0.3580764149533163,
      "grad_norm": 483.4333801269531,
      "learning_rate": 3.261243255269211e-05,
      "loss": 37.7474,
      "step": 88630
    },
    {
      "epoch": 0.35811681621868396,
      "grad_norm": 1233.3927001953125,
      "learning_rate": 3.2610265176415746e-05,
      "loss": 68.6271,
      "step": 88640
    },
    {
      "epoch": 0.3581572174840516,
      "grad_norm": 624.0841064453125,
      "learning_rate": 3.260809755429578e-05,
      "loss": 65.5295,
      "step": 88650
    },
    {
      "epoch": 0.35819761874941924,
      "grad_norm": 818.8887939453125,
      "learning_rate": 3.260592968637445e-05,
      "loss": 86.7369,
      "step": 88660
    },
    {
      "epoch": 0.3582380200147869,
      "grad_norm": 2144.048828125,
      "learning_rate": 3.260376157269404e-05,
      "loss": 77.7664,
      "step": 88670
    },
    {
      "epoch": 0.35827842128015447,
      "grad_norm": 284.1769104003906,
      "learning_rate": 3.2601593213296805e-05,
      "loss": 68.8628,
      "step": 88680
    },
    {
      "epoch": 0.3583188225455221,
      "grad_norm": 327.7435302734375,
      "learning_rate": 3.259942460822503e-05,
      "loss": 61.975,
      "step": 88690
    },
    {
      "epoch": 0.35835922381088975,
      "grad_norm": 985.4323120117188,
      "learning_rate": 3.2597255757520976e-05,
      "loss": 63.7146,
      "step": 88700
    },
    {
      "epoch": 0.3583996250762574,
      "grad_norm": 466.5003662109375,
      "learning_rate": 3.2595086661226943e-05,
      "loss": 54.2732,
      "step": 88710
    },
    {
      "epoch": 0.35844002634162503,
      "grad_norm": 750.10498046875,
      "learning_rate": 3.25929173193852e-05,
      "loss": 67.922,
      "step": 88720
    },
    {
      "epoch": 0.35848042760699267,
      "grad_norm": 617.0565795898438,
      "learning_rate": 3.259074773203806e-05,
      "loss": 68.2847,
      "step": 88730
    },
    {
      "epoch": 0.3585208288723603,
      "grad_norm": 519.65673828125,
      "learning_rate": 3.2588577899227814e-05,
      "loss": 90.388,
      "step": 88740
    },
    {
      "epoch": 0.3585612301377279,
      "grad_norm": 626.7796020507812,
      "learning_rate": 3.258640782099675e-05,
      "loss": 58.2107,
      "step": 88750
    },
    {
      "epoch": 0.35860163140309553,
      "grad_norm": 940.4680786132812,
      "learning_rate": 3.258423749738719e-05,
      "loss": 67.1222,
      "step": 88760
    },
    {
      "epoch": 0.3586420326684632,
      "grad_norm": 1334.208984375,
      "learning_rate": 3.258206692844145e-05,
      "loss": 53.6078,
      "step": 88770
    },
    {
      "epoch": 0.3586824339338308,
      "grad_norm": 1046.6566162109375,
      "learning_rate": 3.2579896114201826e-05,
      "loss": 76.479,
      "step": 88780
    },
    {
      "epoch": 0.35872283519919845,
      "grad_norm": 311.5898742675781,
      "learning_rate": 3.257772505471065e-05,
      "loss": 83.0157,
      "step": 88790
    },
    {
      "epoch": 0.3587632364645661,
      "grad_norm": 1352.649169921875,
      "learning_rate": 3.257555375001026e-05,
      "loss": 65.9161,
      "step": 88800
    },
    {
      "epoch": 0.3588036377299337,
      "grad_norm": 1243.841796875,
      "learning_rate": 3.257338220014297e-05,
      "loss": 72.3628,
      "step": 88810
    },
    {
      "epoch": 0.3588440389953013,
      "grad_norm": 3334.228759765625,
      "learning_rate": 3.257121040515112e-05,
      "loss": 106.4708,
      "step": 88820
    },
    {
      "epoch": 0.35888444026066896,
      "grad_norm": 927.7748413085938,
      "learning_rate": 3.2569038365077045e-05,
      "loss": 50.691,
      "step": 88830
    },
    {
      "epoch": 0.3589248415260366,
      "grad_norm": 657.2202758789062,
      "learning_rate": 3.2566866079963104e-05,
      "loss": 83.4183,
      "step": 88840
    },
    {
      "epoch": 0.35896524279140424,
      "grad_norm": 1189.534423828125,
      "learning_rate": 3.256469354985163e-05,
      "loss": 85.9154,
      "step": 88850
    },
    {
      "epoch": 0.3590056440567719,
      "grad_norm": 950.003662109375,
      "learning_rate": 3.2562520774785e-05,
      "loss": 76.2324,
      "step": 88860
    },
    {
      "epoch": 0.35904604532213946,
      "grad_norm": 284.270263671875,
      "learning_rate": 3.256034775480555e-05,
      "loss": 61.8749,
      "step": 88870
    },
    {
      "epoch": 0.3590864465875071,
      "grad_norm": 972.6361694335938,
      "learning_rate": 3.2558174489955656e-05,
      "loss": 74.3974,
      "step": 88880
    },
    {
      "epoch": 0.35912684785287474,
      "grad_norm": 525.1915893554688,
      "learning_rate": 3.2556000980277686e-05,
      "loss": 51.2244,
      "step": 88890
    },
    {
      "epoch": 0.3591672491182424,
      "grad_norm": 702.22412109375,
      "learning_rate": 3.255382722581401e-05,
      "loss": 52.7797,
      "step": 88900
    },
    {
      "epoch": 0.35920765038361,
      "grad_norm": 4037.07958984375,
      "learning_rate": 3.2551653226607016e-05,
      "loss": 89.1209,
      "step": 88910
    },
    {
      "epoch": 0.35924805164897766,
      "grad_norm": 710.5283203125,
      "learning_rate": 3.2549478982699074e-05,
      "loss": 53.0275,
      "step": 88920
    },
    {
      "epoch": 0.3592884529143453,
      "grad_norm": 548.4154663085938,
      "learning_rate": 3.254730449413258e-05,
      "loss": 66.1228,
      "step": 88930
    },
    {
      "epoch": 0.3593288541797129,
      "grad_norm": 722.6907348632812,
      "learning_rate": 3.2545129760949924e-05,
      "loss": 60.9345,
      "step": 88940
    },
    {
      "epoch": 0.3593692554450805,
      "grad_norm": 747.479248046875,
      "learning_rate": 3.254295478319351e-05,
      "loss": 55.8477,
      "step": 88950
    },
    {
      "epoch": 0.35940965671044817,
      "grad_norm": 924.00927734375,
      "learning_rate": 3.254077956090573e-05,
      "loss": 49.4122,
      "step": 88960
    },
    {
      "epoch": 0.3594500579758158,
      "grad_norm": 279.5837097167969,
      "learning_rate": 3.2538604094128994e-05,
      "loss": 82.6696,
      "step": 88970
    },
    {
      "epoch": 0.35949045924118345,
      "grad_norm": 784.2312622070312,
      "learning_rate": 3.253642838290572e-05,
      "loss": 88.6956,
      "step": 88980
    },
    {
      "epoch": 0.3595308605065511,
      "grad_norm": 1277.6805419921875,
      "learning_rate": 3.2534252427278316e-05,
      "loss": 48.7821,
      "step": 88990
    },
    {
      "epoch": 0.35957126177191867,
      "grad_norm": 835.9051513671875,
      "learning_rate": 3.253207622728921e-05,
      "loss": 64.9844,
      "step": 89000
    },
    {
      "epoch": 0.3596116630372863,
      "grad_norm": 326.860107421875,
      "learning_rate": 3.252989978298083e-05,
      "loss": 66.8233,
      "step": 89010
    },
    {
      "epoch": 0.35965206430265395,
      "grad_norm": 1021.7041015625,
      "learning_rate": 3.25277230943956e-05,
      "loss": 67.8086,
      "step": 89020
    },
    {
      "epoch": 0.3596924655680216,
      "grad_norm": 816.9478149414062,
      "learning_rate": 3.2525546161575964e-05,
      "loss": 71.5213,
      "step": 89030
    },
    {
      "epoch": 0.35973286683338923,
      "grad_norm": 746.73291015625,
      "learning_rate": 3.2523368984564345e-05,
      "loss": 58.0566,
      "step": 89040
    },
    {
      "epoch": 0.35977326809875687,
      "grad_norm": 604.9473876953125,
      "learning_rate": 3.252119156340321e-05,
      "loss": 63.2392,
      "step": 89050
    },
    {
      "epoch": 0.3598136693641245,
      "grad_norm": 431.6227111816406,
      "learning_rate": 3.2519013898134994e-05,
      "loss": 83.0441,
      "step": 89060
    },
    {
      "epoch": 0.3598540706294921,
      "grad_norm": 446.3559875488281,
      "learning_rate": 3.2516835988802155e-05,
      "loss": 57.1256,
      "step": 89070
    },
    {
      "epoch": 0.35989447189485974,
      "grad_norm": 688.2459106445312,
      "learning_rate": 3.251465783544716e-05,
      "loss": 53.325,
      "step": 89080
    },
    {
      "epoch": 0.3599348731602274,
      "grad_norm": 638.8283081054688,
      "learning_rate": 3.2512479438112464e-05,
      "loss": 42.3715,
      "step": 89090
    },
    {
      "epoch": 0.359975274425595,
      "grad_norm": 1033.72998046875,
      "learning_rate": 3.2510300796840546e-05,
      "loss": 73.0672,
      "step": 89100
    },
    {
      "epoch": 0.36001567569096266,
      "grad_norm": 1512.4454345703125,
      "learning_rate": 3.2508121911673866e-05,
      "loss": 63.4775,
      "step": 89110
    },
    {
      "epoch": 0.3600560769563303,
      "grad_norm": 584.1983642578125,
      "learning_rate": 3.250594278265491e-05,
      "loss": 57.5051,
      "step": 89120
    },
    {
      "epoch": 0.3600964782216979,
      "grad_norm": 0.0,
      "learning_rate": 3.250376340982616e-05,
      "loss": 62.77,
      "step": 89130
    },
    {
      "epoch": 0.3601368794870655,
      "grad_norm": 312.9271545410156,
      "learning_rate": 3.250158379323011e-05,
      "loss": 51.5549,
      "step": 89140
    },
    {
      "epoch": 0.36017728075243316,
      "grad_norm": 898.7651977539062,
      "learning_rate": 3.249940393290925e-05,
      "loss": 68.3641,
      "step": 89150
    },
    {
      "epoch": 0.3602176820178008,
      "grad_norm": 1063.464599609375,
      "learning_rate": 3.249722382890607e-05,
      "loss": 45.5237,
      "step": 89160
    },
    {
      "epoch": 0.36025808328316844,
      "grad_norm": 1576.947265625,
      "learning_rate": 3.249504348126308e-05,
      "loss": 54.5762,
      "step": 89170
    },
    {
      "epoch": 0.3602984845485361,
      "grad_norm": 738.2628784179688,
      "learning_rate": 3.249286289002278e-05,
      "loss": 52.6362,
      "step": 89180
    },
    {
      "epoch": 0.36033888581390366,
      "grad_norm": 977.0211181640625,
      "learning_rate": 3.2490682055227695e-05,
      "loss": 79.1429,
      "step": 89190
    },
    {
      "epoch": 0.3603792870792713,
      "grad_norm": 606.0816040039062,
      "learning_rate": 3.248850097692032e-05,
      "loss": 55.5943,
      "step": 89200
    },
    {
      "epoch": 0.36041968834463894,
      "grad_norm": 518.8965454101562,
      "learning_rate": 3.2486319655143196e-05,
      "loss": 78.7807,
      "step": 89210
    },
    {
      "epoch": 0.3604600896100066,
      "grad_norm": 314.8681335449219,
      "learning_rate": 3.248413808993884e-05,
      "loss": 48.6494,
      "step": 89220
    },
    {
      "epoch": 0.3605004908753742,
      "grad_norm": 808.1603393554688,
      "learning_rate": 3.248195628134979e-05,
      "loss": 55.6867,
      "step": 89230
    },
    {
      "epoch": 0.36054089214074186,
      "grad_norm": 549.8880004882812,
      "learning_rate": 3.2479774229418565e-05,
      "loss": 49.6709,
      "step": 89240
    },
    {
      "epoch": 0.3605812934061095,
      "grad_norm": 796.691650390625,
      "learning_rate": 3.247759193418773e-05,
      "loss": 58.077,
      "step": 89250
    },
    {
      "epoch": 0.3606216946714771,
      "grad_norm": 456.6281433105469,
      "learning_rate": 3.2475409395699805e-05,
      "loss": 61.6451,
      "step": 89260
    },
    {
      "epoch": 0.36066209593684473,
      "grad_norm": 348.8264465332031,
      "learning_rate": 3.2473226613997355e-05,
      "loss": 100.175,
      "step": 89270
    },
    {
      "epoch": 0.36070249720221237,
      "grad_norm": 778.9163208007812,
      "learning_rate": 3.247104358912293e-05,
      "loss": 92.9797,
      "step": 89280
    },
    {
      "epoch": 0.36074289846758,
      "grad_norm": 647.578857421875,
      "learning_rate": 3.2468860321119095e-05,
      "loss": 89.2201,
      "step": 89290
    },
    {
      "epoch": 0.36078329973294765,
      "grad_norm": 8475.5908203125,
      "learning_rate": 3.246667681002841e-05,
      "loss": 101.6309,
      "step": 89300
    },
    {
      "epoch": 0.3608237009983153,
      "grad_norm": 1095.387451171875,
      "learning_rate": 3.2464493055893436e-05,
      "loss": 79.6601,
      "step": 89310
    },
    {
      "epoch": 0.3608641022636829,
      "grad_norm": 345.1563720703125,
      "learning_rate": 3.246230905875675e-05,
      "loss": 59.279,
      "step": 89320
    },
    {
      "epoch": 0.3609045035290505,
      "grad_norm": 740.0067749023438,
      "learning_rate": 3.246012481866093e-05,
      "loss": 58.2936,
      "step": 89330
    },
    {
      "epoch": 0.36094490479441815,
      "grad_norm": 943.1046142578125,
      "learning_rate": 3.245794033564857e-05,
      "loss": 57.9797,
      "step": 89340
    },
    {
      "epoch": 0.3609853060597858,
      "grad_norm": 0.0,
      "learning_rate": 3.245575560976225e-05,
      "loss": 73.7864,
      "step": 89350
    },
    {
      "epoch": 0.36102570732515343,
      "grad_norm": 547.4214477539062,
      "learning_rate": 3.2453570641044565e-05,
      "loss": 47.4376,
      "step": 89360
    },
    {
      "epoch": 0.3610661085905211,
      "grad_norm": 746.9415283203125,
      "learning_rate": 3.24513854295381e-05,
      "loss": 37.7824,
      "step": 89370
    },
    {
      "epoch": 0.3611065098558887,
      "grad_norm": 581.8740234375,
      "learning_rate": 3.244919997528546e-05,
      "loss": 78.0141,
      "step": 89380
    },
    {
      "epoch": 0.3611469111212563,
      "grad_norm": 525.6729125976562,
      "learning_rate": 3.2447014278329275e-05,
      "loss": 65.2984,
      "step": 89390
    },
    {
      "epoch": 0.36118731238662394,
      "grad_norm": 1312.8070068359375,
      "learning_rate": 3.244482833871213e-05,
      "loss": 51.3537,
      "step": 89400
    },
    {
      "epoch": 0.3612277136519916,
      "grad_norm": 330.7095031738281,
      "learning_rate": 3.2442642156476653e-05,
      "loss": 35.5988,
      "step": 89410
    },
    {
      "epoch": 0.3612681149173592,
      "grad_norm": 744.0631103515625,
      "learning_rate": 3.244045573166545e-05,
      "loss": 57.9905,
      "step": 89420
    },
    {
      "epoch": 0.36130851618272686,
      "grad_norm": 5193.68212890625,
      "learning_rate": 3.243826906432117e-05,
      "loss": 76.9366,
      "step": 89430
    },
    {
      "epoch": 0.3613489174480945,
      "grad_norm": 1730.064697265625,
      "learning_rate": 3.2436082154486426e-05,
      "loss": 84.2516,
      "step": 89440
    },
    {
      "epoch": 0.3613893187134621,
      "grad_norm": 442.8031921386719,
      "learning_rate": 3.243389500220386e-05,
      "loss": 53.5648,
      "step": 89450
    },
    {
      "epoch": 0.3614297199788297,
      "grad_norm": 1522.3472900390625,
      "learning_rate": 3.243170760751611e-05,
      "loss": 88.243,
      "step": 89460
    },
    {
      "epoch": 0.36147012124419736,
      "grad_norm": 490.6719055175781,
      "learning_rate": 3.242951997046581e-05,
      "loss": 71.2257,
      "step": 89470
    },
    {
      "epoch": 0.361510522509565,
      "grad_norm": 0.0,
      "learning_rate": 3.242733209109563e-05,
      "loss": 46.0571,
      "step": 89480
    },
    {
      "epoch": 0.36155092377493264,
      "grad_norm": 703.7340698242188,
      "learning_rate": 3.242514396944821e-05,
      "loss": 40.7495,
      "step": 89490
    },
    {
      "epoch": 0.3615913250403003,
      "grad_norm": 754.905029296875,
      "learning_rate": 3.242295560556621e-05,
      "loss": 80.2023,
      "step": 89500
    },
    {
      "epoch": 0.36163172630566787,
      "grad_norm": 1635.7506103515625,
      "learning_rate": 3.24207669994923e-05,
      "loss": 70.7424,
      "step": 89510
    },
    {
      "epoch": 0.3616721275710355,
      "grad_norm": 0.0,
      "learning_rate": 3.2418578151269135e-05,
      "loss": 70.3404,
      "step": 89520
    },
    {
      "epoch": 0.36171252883640315,
      "grad_norm": 793.4153442382812,
      "learning_rate": 3.24163890609394e-05,
      "loss": 74.2491,
      "step": 89530
    },
    {
      "epoch": 0.3617529301017708,
      "grad_norm": 949.236083984375,
      "learning_rate": 3.2414199728545767e-05,
      "loss": 48.3229,
      "step": 89540
    },
    {
      "epoch": 0.3617933313671384,
      "grad_norm": 552.69091796875,
      "learning_rate": 3.2412010154130914e-05,
      "loss": 50.1861,
      "step": 89550
    },
    {
      "epoch": 0.36183373263250607,
      "grad_norm": 2303.78564453125,
      "learning_rate": 3.240982033773754e-05,
      "loss": 91.6548,
      "step": 89560
    },
    {
      "epoch": 0.3618741338978737,
      "grad_norm": 279.16015625,
      "learning_rate": 3.2407630279408326e-05,
      "loss": 32.6592,
      "step": 89570
    },
    {
      "epoch": 0.3619145351632413,
      "grad_norm": 892.2996826171875,
      "learning_rate": 3.240543997918598e-05,
      "loss": 59.1878,
      "step": 89580
    },
    {
      "epoch": 0.36195493642860893,
      "grad_norm": 757.2686767578125,
      "learning_rate": 3.240324943711318e-05,
      "loss": 56.4908,
      "step": 89590
    },
    {
      "epoch": 0.36199533769397657,
      "grad_norm": 304.4239196777344,
      "learning_rate": 3.240105865323266e-05,
      "loss": 34.8889,
      "step": 89600
    },
    {
      "epoch": 0.3620357389593442,
      "grad_norm": 532.8265380859375,
      "learning_rate": 3.23988676275871e-05,
      "loss": 107.3717,
      "step": 89610
    },
    {
      "epoch": 0.36207614022471185,
      "grad_norm": 1257.88720703125,
      "learning_rate": 3.239667636021925e-05,
      "loss": 75.3513,
      "step": 89620
    },
    {
      "epoch": 0.3621165414900795,
      "grad_norm": 907.2530517578125,
      "learning_rate": 3.239448485117181e-05,
      "loss": 60.0257,
      "step": 89630
    },
    {
      "epoch": 0.3621569427554471,
      "grad_norm": 495.6878662109375,
      "learning_rate": 3.23922931004875e-05,
      "loss": 95.7158,
      "step": 89640
    },
    {
      "epoch": 0.3621973440208147,
      "grad_norm": 723.4539184570312,
      "learning_rate": 3.239010110820906e-05,
      "loss": 50.1088,
      "step": 89650
    },
    {
      "epoch": 0.36223774528618236,
      "grad_norm": 611.675537109375,
      "learning_rate": 3.2387908874379213e-05,
      "loss": 60.4733,
      "step": 89660
    },
    {
      "epoch": 0.36227814655155,
      "grad_norm": 960.843994140625,
      "learning_rate": 3.2385716399040706e-05,
      "loss": 63.4884,
      "step": 89670
    },
    {
      "epoch": 0.36231854781691764,
      "grad_norm": 2095.677734375,
      "learning_rate": 3.238352368223629e-05,
      "loss": 81.218,
      "step": 89680
    },
    {
      "epoch": 0.3623589490822853,
      "grad_norm": 748.8034057617188,
      "learning_rate": 3.23813307240087e-05,
      "loss": 87.9378,
      "step": 89690
    },
    {
      "epoch": 0.3623993503476529,
      "grad_norm": 858.1019287109375,
      "learning_rate": 3.23791375244007e-05,
      "loss": 51.7633,
      "step": 89700
    },
    {
      "epoch": 0.3624397516130205,
      "grad_norm": 553.71826171875,
      "learning_rate": 3.237694408345503e-05,
      "loss": 80.8901,
      "step": 89710
    },
    {
      "epoch": 0.36248015287838814,
      "grad_norm": 741.5399780273438,
      "learning_rate": 3.2374750401214466e-05,
      "loss": 54.558,
      "step": 89720
    },
    {
      "epoch": 0.3625205541437558,
      "grad_norm": 545.9320678710938,
      "learning_rate": 3.2372556477721766e-05,
      "loss": 50.4829,
      "step": 89730
    },
    {
      "epoch": 0.3625609554091234,
      "grad_norm": 1244.79052734375,
      "learning_rate": 3.2370362313019725e-05,
      "loss": 66.7102,
      "step": 89740
    },
    {
      "epoch": 0.36260135667449106,
      "grad_norm": 1156.6531982421875,
      "learning_rate": 3.2368167907151086e-05,
      "loss": 58.8571,
      "step": 89750
    },
    {
      "epoch": 0.3626417579398587,
      "grad_norm": 1276.3740234375,
      "learning_rate": 3.236597326015865e-05,
      "loss": 72.6798,
      "step": 89760
    },
    {
      "epoch": 0.3626821592052263,
      "grad_norm": 768.4413452148438,
      "learning_rate": 3.23637783720852e-05,
      "loss": 59.6834,
      "step": 89770
    },
    {
      "epoch": 0.3627225604705939,
      "grad_norm": 723.568359375,
      "learning_rate": 3.236158324297353e-05,
      "loss": 60.3245,
      "step": 89780
    },
    {
      "epoch": 0.36276296173596156,
      "grad_norm": 753.067626953125,
      "learning_rate": 3.235938787286642e-05,
      "loss": 52.7935,
      "step": 89790
    },
    {
      "epoch": 0.3628033630013292,
      "grad_norm": 1123.5992431640625,
      "learning_rate": 3.235719226180669e-05,
      "loss": 43.8822,
      "step": 89800
    },
    {
      "epoch": 0.36284376426669684,
      "grad_norm": 841.1117553710938,
      "learning_rate": 3.2354996409837136e-05,
      "loss": 83.7583,
      "step": 89810
    },
    {
      "epoch": 0.3628841655320645,
      "grad_norm": 758.0658569335938,
      "learning_rate": 3.2352800317000555e-05,
      "loss": 61.8449,
      "step": 89820
    },
    {
      "epoch": 0.36292456679743207,
      "grad_norm": 745.8280639648438,
      "learning_rate": 3.235060398333978e-05,
      "loss": 51.7937,
      "step": 89830
    },
    {
      "epoch": 0.3629649680627997,
      "grad_norm": 0.0,
      "learning_rate": 3.234840740889762e-05,
      "loss": 60.1308,
      "step": 89840
    },
    {
      "epoch": 0.36300536932816735,
      "grad_norm": 422.6933288574219,
      "learning_rate": 3.23462105937169e-05,
      "loss": 53.8212,
      "step": 89850
    },
    {
      "epoch": 0.363045770593535,
      "grad_norm": 954.4671630859375,
      "learning_rate": 3.234401353784045e-05,
      "loss": 69.0127,
      "step": 89860
    },
    {
      "epoch": 0.36308617185890263,
      "grad_norm": 661.1810913085938,
      "learning_rate": 3.23418162413111e-05,
      "loss": 50.7474,
      "step": 89870
    },
    {
      "epoch": 0.36312657312427027,
      "grad_norm": 997.407958984375,
      "learning_rate": 3.233961870417169e-05,
      "loss": 88.6371,
      "step": 89880
    },
    {
      "epoch": 0.3631669743896379,
      "grad_norm": 502.9741516113281,
      "learning_rate": 3.233742092646506e-05,
      "loss": 42.5384,
      "step": 89890
    },
    {
      "epoch": 0.3632073756550055,
      "grad_norm": 779.2553100585938,
      "learning_rate": 3.2335222908234054e-05,
      "loss": 98.033,
      "step": 89900
    },
    {
      "epoch": 0.36324777692037313,
      "grad_norm": 1040.43359375,
      "learning_rate": 3.233302464952153e-05,
      "loss": 73.3634,
      "step": 89910
    },
    {
      "epoch": 0.3632881781857408,
      "grad_norm": 663.7528076171875,
      "learning_rate": 3.233082615037034e-05,
      "loss": 69.4142,
      "step": 89920
    },
    {
      "epoch": 0.3633285794511084,
      "grad_norm": 447.36468505859375,
      "learning_rate": 3.232862741082335e-05,
      "loss": 47.4241,
      "step": 89930
    },
    {
      "epoch": 0.36336898071647605,
      "grad_norm": 1314.697509765625,
      "learning_rate": 3.232642843092341e-05,
      "loss": 47.8006,
      "step": 89940
    },
    {
      "epoch": 0.3634093819818437,
      "grad_norm": 783.3645629882812,
      "learning_rate": 3.232422921071341e-05,
      "loss": 78.2489,
      "step": 89950
    },
    {
      "epoch": 0.3634497832472113,
      "grad_norm": 571.39111328125,
      "learning_rate": 3.2322029750236224e-05,
      "loss": 60.6246,
      "step": 89960
    },
    {
      "epoch": 0.3634901845125789,
      "grad_norm": 573.41259765625,
      "learning_rate": 3.2319830049534714e-05,
      "loss": 64.5672,
      "step": 89970
    },
    {
      "epoch": 0.36353058577794656,
      "grad_norm": 651.1929931640625,
      "learning_rate": 3.2317630108651775e-05,
      "loss": 82.1077,
      "step": 89980
    },
    {
      "epoch": 0.3635709870433142,
      "grad_norm": 841.3027954101562,
      "learning_rate": 3.231542992763029e-05,
      "loss": 48.8317,
      "step": 89990
    },
    {
      "epoch": 0.36361138830868184,
      "grad_norm": 1385.8458251953125,
      "learning_rate": 3.2313229506513167e-05,
      "loss": 82.3096,
      "step": 90000
    },
    {
      "epoch": 0.3636517895740495,
      "grad_norm": 633.955078125,
      "learning_rate": 3.231102884534329e-05,
      "loss": 83.0627,
      "step": 90010
    },
    {
      "epoch": 0.3636921908394171,
      "grad_norm": 873.541259765625,
      "learning_rate": 3.2308827944163576e-05,
      "loss": 82.5397,
      "step": 90020
    },
    {
      "epoch": 0.3637325921047847,
      "grad_norm": 738.3284912109375,
      "learning_rate": 3.230662680301692e-05,
      "loss": 110.8826,
      "step": 90030
    },
    {
      "epoch": 0.36377299337015234,
      "grad_norm": 799.2305297851562,
      "learning_rate": 3.2304425421946234e-05,
      "loss": 63.8086,
      "step": 90040
    },
    {
      "epoch": 0.36381339463552,
      "grad_norm": 1250.10400390625,
      "learning_rate": 3.230222380099445e-05,
      "loss": 78.0658,
      "step": 90050
    },
    {
      "epoch": 0.3638537959008876,
      "grad_norm": 1647.9473876953125,
      "learning_rate": 3.230002194020447e-05,
      "loss": 83.1571,
      "step": 90060
    },
    {
      "epoch": 0.36389419716625526,
      "grad_norm": 606.3142700195312,
      "learning_rate": 3.229781983961923e-05,
      "loss": 58.3433,
      "step": 90070
    },
    {
      "epoch": 0.3639345984316229,
      "grad_norm": 836.625244140625,
      "learning_rate": 3.229561749928166e-05,
      "loss": 60.6996,
      "step": 90080
    },
    {
      "epoch": 0.3639749996969905,
      "grad_norm": 459.7673034667969,
      "learning_rate": 3.22934149192347e-05,
      "loss": 55.3213,
      "step": 90090
    },
    {
      "epoch": 0.3640154009623581,
      "grad_norm": 681.5406494140625,
      "learning_rate": 3.229121209952129e-05,
      "loss": 57.4772,
      "step": 90100
    },
    {
      "epoch": 0.36405580222772577,
      "grad_norm": 593.47998046875,
      "learning_rate": 3.2289009040184375e-05,
      "loss": 69.6571,
      "step": 90110
    },
    {
      "epoch": 0.3640962034930934,
      "grad_norm": 646.6085205078125,
      "learning_rate": 3.2286805741266895e-05,
      "loss": 104.5698,
      "step": 90120
    },
    {
      "epoch": 0.36413660475846105,
      "grad_norm": 870.090576171875,
      "learning_rate": 3.228460220281181e-05,
      "loss": 81.7248,
      "step": 90130
    },
    {
      "epoch": 0.3641770060238287,
      "grad_norm": 647.8412475585938,
      "learning_rate": 3.2282398424862086e-05,
      "loss": 61.7405,
      "step": 90140
    },
    {
      "epoch": 0.36421740728919627,
      "grad_norm": 1034.1387939453125,
      "learning_rate": 3.228019440746068e-05,
      "loss": 60.8793,
      "step": 90150
    },
    {
      "epoch": 0.3642578085545639,
      "grad_norm": 603.2088012695312,
      "learning_rate": 3.2277990150650554e-05,
      "loss": 57.9373,
      "step": 90160
    },
    {
      "epoch": 0.36429820981993155,
      "grad_norm": 651.9238891601562,
      "learning_rate": 3.22757856544747e-05,
      "loss": 61.3861,
      "step": 90170
    },
    {
      "epoch": 0.3643386110852992,
      "grad_norm": 821.35009765625,
      "learning_rate": 3.227358091897608e-05,
      "loss": 91.5335,
      "step": 90180
    },
    {
      "epoch": 0.36437901235066683,
      "grad_norm": 570.3350830078125,
      "learning_rate": 3.227137594419768e-05,
      "loss": 53.6288,
      "step": 90190
    },
    {
      "epoch": 0.36441941361603447,
      "grad_norm": 1010.9404296875,
      "learning_rate": 3.2269170730182486e-05,
      "loss": 61.6317,
      "step": 90200
    },
    {
      "epoch": 0.3644598148814021,
      "grad_norm": 1174.55126953125,
      "learning_rate": 3.22669652769735e-05,
      "loss": 41.7424,
      "step": 90210
    },
    {
      "epoch": 0.3645002161467697,
      "grad_norm": 1207.1864013671875,
      "learning_rate": 3.22647595846137e-05,
      "loss": 73.6048,
      "step": 90220
    },
    {
      "epoch": 0.36454061741213734,
      "grad_norm": 532.3197631835938,
      "learning_rate": 3.2262553653146106e-05,
      "loss": 40.4853,
      "step": 90230
    },
    {
      "epoch": 0.364581018677505,
      "grad_norm": 310.5439453125,
      "learning_rate": 3.2260347482613714e-05,
      "loss": 57.3422,
      "step": 90240
    },
    {
      "epoch": 0.3646214199428726,
      "grad_norm": 915.779052734375,
      "learning_rate": 3.2258141073059533e-05,
      "loss": 89.975,
      "step": 90250
    },
    {
      "epoch": 0.36466182120824026,
      "grad_norm": 483.0838928222656,
      "learning_rate": 3.225593442452658e-05,
      "loss": 68.5153,
      "step": 90260
    },
    {
      "epoch": 0.3647022224736079,
      "grad_norm": 965.7793579101562,
      "learning_rate": 3.225372753705788e-05,
      "loss": 66.1839,
      "step": 90270
    },
    {
      "epoch": 0.3647426237389755,
      "grad_norm": 809.3526000976562,
      "learning_rate": 3.225152041069645e-05,
      "loss": 91.8366,
      "step": 90280
    },
    {
      "epoch": 0.3647830250043431,
      "grad_norm": 659.0712890625,
      "learning_rate": 3.224931304548532e-05,
      "loss": 66.1223,
      "step": 90290
    },
    {
      "epoch": 0.36482342626971076,
      "grad_norm": 1612.8360595703125,
      "learning_rate": 3.224710544146753e-05,
      "loss": 74.4651,
      "step": 90300
    },
    {
      "epoch": 0.3648638275350784,
      "grad_norm": 1003.6226196289062,
      "learning_rate": 3.224489759868612e-05,
      "loss": 49.3529,
      "step": 90310
    },
    {
      "epoch": 0.36490422880044604,
      "grad_norm": 2766.993896484375,
      "learning_rate": 3.224268951718411e-05,
      "loss": 77.647,
      "step": 90320
    },
    {
      "epoch": 0.3649446300658137,
      "grad_norm": 2475.29443359375,
      "learning_rate": 3.224048119700458e-05,
      "loss": 107.4289,
      "step": 90330
    },
    {
      "epoch": 0.3649850313311813,
      "grad_norm": 1772.4407958984375,
      "learning_rate": 3.223827263819056e-05,
      "loss": 81.8948,
      "step": 90340
    },
    {
      "epoch": 0.3650254325965489,
      "grad_norm": 1427.0467529296875,
      "learning_rate": 3.223606384078512e-05,
      "loss": 79.2289,
      "step": 90350
    },
    {
      "epoch": 0.36506583386191654,
      "grad_norm": 765.2987060546875,
      "learning_rate": 3.223385480483131e-05,
      "loss": 87.3908,
      "step": 90360
    },
    {
      "epoch": 0.3651062351272842,
      "grad_norm": 445.6130676269531,
      "learning_rate": 3.223164553037221e-05,
      "loss": 45.1544,
      "step": 90370
    },
    {
      "epoch": 0.3651466363926518,
      "grad_norm": 2689.060302734375,
      "learning_rate": 3.2229436017450876e-05,
      "loss": 61.4555,
      "step": 90380
    },
    {
      "epoch": 0.36518703765801946,
      "grad_norm": 5151.96533203125,
      "learning_rate": 3.222722626611039e-05,
      "loss": 94.6698,
      "step": 90390
    },
    {
      "epoch": 0.3652274389233871,
      "grad_norm": 1440.62353515625,
      "learning_rate": 3.222501627639384e-05,
      "loss": 77.1921,
      "step": 90400
    },
    {
      "epoch": 0.3652678401887547,
      "grad_norm": 547.9785766601562,
      "learning_rate": 3.222280604834429e-05,
      "loss": 52.1105,
      "step": 90410
    },
    {
      "epoch": 0.36530824145412233,
      "grad_norm": 355.76617431640625,
      "learning_rate": 3.222059558200486e-05,
      "loss": 94.4201,
      "step": 90420
    },
    {
      "epoch": 0.36534864271948997,
      "grad_norm": 989.900146484375,
      "learning_rate": 3.221838487741862e-05,
      "loss": 41.2473,
      "step": 90430
    },
    {
      "epoch": 0.3653890439848576,
      "grad_norm": 1165.8541259765625,
      "learning_rate": 3.2216173934628674e-05,
      "loss": 55.0149,
      "step": 90440
    },
    {
      "epoch": 0.36542944525022525,
      "grad_norm": 914.6465454101562,
      "learning_rate": 3.221396275367813e-05,
      "loss": 74.3536,
      "step": 90450
    },
    {
      "epoch": 0.3654698465155929,
      "grad_norm": 425.96435546875,
      "learning_rate": 3.2211751334610094e-05,
      "loss": 69.3202,
      "step": 90460
    },
    {
      "epoch": 0.3655102477809605,
      "grad_norm": 267.4327392578125,
      "learning_rate": 3.220953967746768e-05,
      "loss": 74.8355,
      "step": 90470
    },
    {
      "epoch": 0.3655506490463281,
      "grad_norm": 327.8559265136719,
      "learning_rate": 3.2207327782294e-05,
      "loss": 85.0147,
      "step": 90480
    },
    {
      "epoch": 0.36559105031169575,
      "grad_norm": 381.49578857421875,
      "learning_rate": 3.2205115649132185e-05,
      "loss": 65.0611,
      "step": 90490
    },
    {
      "epoch": 0.3656314515770634,
      "grad_norm": 960.8960571289062,
      "learning_rate": 3.220290327802536e-05,
      "loss": 63.5097,
      "step": 90500
    },
    {
      "epoch": 0.36567185284243103,
      "grad_norm": 415.2254333496094,
      "learning_rate": 3.2200690669016645e-05,
      "loss": 65.3691,
      "step": 90510
    },
    {
      "epoch": 0.3657122541077987,
      "grad_norm": 361.4107666015625,
      "learning_rate": 3.219847782214918e-05,
      "loss": 42.4056,
      "step": 90520
    },
    {
      "epoch": 0.3657526553731663,
      "grad_norm": 503.8821105957031,
      "learning_rate": 3.219626473746613e-05,
      "loss": 71.6832,
      "step": 90530
    },
    {
      "epoch": 0.3657930566385339,
      "grad_norm": 802.1369018554688,
      "learning_rate": 3.21940514150106e-05,
      "loss": 50.1877,
      "step": 90540
    },
    {
      "epoch": 0.36583345790390154,
      "grad_norm": 2277.003173828125,
      "learning_rate": 3.2191837854825766e-05,
      "loss": 75.0608,
      "step": 90550
    },
    {
      "epoch": 0.3658738591692692,
      "grad_norm": 1255.205322265625,
      "learning_rate": 3.218962405695478e-05,
      "loss": 61.8157,
      "step": 90560
    },
    {
      "epoch": 0.3659142604346368,
      "grad_norm": 426.26446533203125,
      "learning_rate": 3.2187410021440786e-05,
      "loss": 64.9316,
      "step": 90570
    },
    {
      "epoch": 0.36595466170000446,
      "grad_norm": 1173.32763671875,
      "learning_rate": 3.218519574832697e-05,
      "loss": 69.6508,
      "step": 90580
    },
    {
      "epoch": 0.3659950629653721,
      "grad_norm": 1251.942138671875,
      "learning_rate": 3.218298123765649e-05,
      "loss": 63.1345,
      "step": 90590
    },
    {
      "epoch": 0.3660354642307397,
      "grad_norm": 0.0,
      "learning_rate": 3.218076648947251e-05,
      "loss": 35.3188,
      "step": 90600
    },
    {
      "epoch": 0.3660758654961073,
      "grad_norm": 531.1837768554688,
      "learning_rate": 3.217855150381822e-05,
      "loss": 65.0267,
      "step": 90610
    },
    {
      "epoch": 0.36611626676147496,
      "grad_norm": 924.5548706054688,
      "learning_rate": 3.217633628073681e-05,
      "loss": 60.6692,
      "step": 90620
    },
    {
      "epoch": 0.3661566680268426,
      "grad_norm": 990.2822875976562,
      "learning_rate": 3.217412082027144e-05,
      "loss": 60.5045,
      "step": 90630
    },
    {
      "epoch": 0.36619706929221024,
      "grad_norm": 969.4849243164062,
      "learning_rate": 3.217190512246532e-05,
      "loss": 89.893,
      "step": 90640
    },
    {
      "epoch": 0.3662374705575779,
      "grad_norm": 575.37255859375,
      "learning_rate": 3.216968918736164e-05,
      "loss": 61.3946,
      "step": 90650
    },
    {
      "epoch": 0.3662778718229455,
      "grad_norm": 370.7169189453125,
      "learning_rate": 3.2167473015003616e-05,
      "loss": 40.7734,
      "step": 90660
    },
    {
      "epoch": 0.3663182730883131,
      "grad_norm": 859.6712646484375,
      "learning_rate": 3.216525660543444e-05,
      "loss": 60.6101,
      "step": 90670
    },
    {
      "epoch": 0.36635867435368075,
      "grad_norm": 460.1180419921875,
      "learning_rate": 3.216303995869731e-05,
      "loss": 74.8931,
      "step": 90680
    },
    {
      "epoch": 0.3663990756190484,
      "grad_norm": 522.7005004882812,
      "learning_rate": 3.2160823074835464e-05,
      "loss": 67.9626,
      "step": 90690
    },
    {
      "epoch": 0.366439476884416,
      "grad_norm": 645.8761596679688,
      "learning_rate": 3.215860595389211e-05,
      "loss": 65.7363,
      "step": 90700
    },
    {
      "epoch": 0.36647987814978367,
      "grad_norm": 436.8887634277344,
      "learning_rate": 3.215638859591048e-05,
      "loss": 99.0599,
      "step": 90710
    },
    {
      "epoch": 0.3665202794151513,
      "grad_norm": 783.563720703125,
      "learning_rate": 3.215417100093378e-05,
      "loss": 73.7595,
      "step": 90720
    },
    {
      "epoch": 0.3665606806805189,
      "grad_norm": 1526.1766357421875,
      "learning_rate": 3.215195316900527e-05,
      "loss": 80.5374,
      "step": 90730
    },
    {
      "epoch": 0.36660108194588653,
      "grad_norm": 1934.50048828125,
      "learning_rate": 3.2149735100168176e-05,
      "loss": 62.5631,
      "step": 90740
    },
    {
      "epoch": 0.36664148321125417,
      "grad_norm": 376.3502197265625,
      "learning_rate": 3.214751679446574e-05,
      "loss": 86.4771,
      "step": 90750
    },
    {
      "epoch": 0.3666818844766218,
      "grad_norm": 0.0,
      "learning_rate": 3.214529825194121e-05,
      "loss": 81.0698,
      "step": 90760
    },
    {
      "epoch": 0.36672228574198945,
      "grad_norm": 641.5331420898438,
      "learning_rate": 3.214307947263783e-05,
      "loss": 64.7252,
      "step": 90770
    },
    {
      "epoch": 0.3667626870073571,
      "grad_norm": 759.74755859375,
      "learning_rate": 3.2140860456598877e-05,
      "loss": 49.2482,
      "step": 90780
    },
    {
      "epoch": 0.3668030882727247,
      "grad_norm": 2156.34423828125,
      "learning_rate": 3.213864120386759e-05,
      "loss": 52.424,
      "step": 90790
    },
    {
      "epoch": 0.3668434895380923,
      "grad_norm": 510.70782470703125,
      "learning_rate": 3.213642171448725e-05,
      "loss": 36.5959,
      "step": 90800
    },
    {
      "epoch": 0.36688389080345996,
      "grad_norm": 630.8547973632812,
      "learning_rate": 3.213420198850111e-05,
      "loss": 49.5609,
      "step": 90810
    },
    {
      "epoch": 0.3669242920688276,
      "grad_norm": 1101.6739501953125,
      "learning_rate": 3.213198202595247e-05,
      "loss": 53.2187,
      "step": 90820
    },
    {
      "epoch": 0.36696469333419524,
      "grad_norm": 1123.56494140625,
      "learning_rate": 3.212976182688458e-05,
      "loss": 55.8884,
      "step": 90830
    },
    {
      "epoch": 0.3670050945995629,
      "grad_norm": 814.5668334960938,
      "learning_rate": 3.212754139134075e-05,
      "loss": 74.1858,
      "step": 90840
    },
    {
      "epoch": 0.3670454958649305,
      "grad_norm": 850.3318481445312,
      "learning_rate": 3.212532071936425e-05,
      "loss": 68.7365,
      "step": 90850
    },
    {
      "epoch": 0.3670858971302981,
      "grad_norm": 656.654541015625,
      "learning_rate": 3.2123099810998385e-05,
      "loss": 48.3398,
      "step": 90860
    },
    {
      "epoch": 0.36712629839566574,
      "grad_norm": 448.7389831542969,
      "learning_rate": 3.212087866628644e-05,
      "loss": 69.4263,
      "step": 90870
    },
    {
      "epoch": 0.3671666996610334,
      "grad_norm": 930.092529296875,
      "learning_rate": 3.211865728527173e-05,
      "loss": 59.9901,
      "step": 90880
    },
    {
      "epoch": 0.367207100926401,
      "grad_norm": 423.372314453125,
      "learning_rate": 3.211643566799756e-05,
      "loss": 47.8891,
      "step": 90890
    },
    {
      "epoch": 0.36724750219176866,
      "grad_norm": 634.6875610351562,
      "learning_rate": 3.2114213814507235e-05,
      "loss": 25.0227,
      "step": 90900
    },
    {
      "epoch": 0.3672879034571363,
      "grad_norm": 858.253662109375,
      "learning_rate": 3.211199172484407e-05,
      "loss": 72.241,
      "step": 90910
    },
    {
      "epoch": 0.3673283047225039,
      "grad_norm": 840.0797119140625,
      "learning_rate": 3.21097693990514e-05,
      "loss": 46.978,
      "step": 90920
    },
    {
      "epoch": 0.3673687059878715,
      "grad_norm": 1786.7987060546875,
      "learning_rate": 3.210754683717253e-05,
      "loss": 72.6414,
      "step": 90930
    },
    {
      "epoch": 0.36740910725323916,
      "grad_norm": 0.0,
      "learning_rate": 3.2105324039250814e-05,
      "loss": 52.4552,
      "step": 90940
    },
    {
      "epoch": 0.3674495085186068,
      "grad_norm": 987.6989135742188,
      "learning_rate": 3.210310100532956e-05,
      "loss": 77.9075,
      "step": 90950
    },
    {
      "epoch": 0.36748990978397444,
      "grad_norm": 611.7899169921875,
      "learning_rate": 3.210087773545214e-05,
      "loss": 45.6186,
      "step": 90960
    },
    {
      "epoch": 0.3675303110493421,
      "grad_norm": 459.25738525390625,
      "learning_rate": 3.209865422966186e-05,
      "loss": 57.6455,
      "step": 90970
    },
    {
      "epoch": 0.3675707123147097,
      "grad_norm": 723.4088745117188,
      "learning_rate": 3.20964304880021e-05,
      "loss": 54.5551,
      "step": 90980
    },
    {
      "epoch": 0.3676111135800773,
      "grad_norm": 784.4147338867188,
      "learning_rate": 3.209420651051619e-05,
      "loss": 71.0863,
      "step": 90990
    },
    {
      "epoch": 0.36765151484544495,
      "grad_norm": 479.0218200683594,
      "learning_rate": 3.2091982297247505e-05,
      "loss": 55.232,
      "step": 91000
    },
    {
      "epoch": 0.3676919161108126,
      "grad_norm": 887.7474365234375,
      "learning_rate": 3.2089757848239395e-05,
      "loss": 76.9433,
      "step": 91010
    },
    {
      "epoch": 0.36773231737618023,
      "grad_norm": 562.9622802734375,
      "learning_rate": 3.208753316353523e-05,
      "loss": 68.3854,
      "step": 91020
    },
    {
      "epoch": 0.36777271864154787,
      "grad_norm": 478.1982421875,
      "learning_rate": 3.2085308243178386e-05,
      "loss": 61.2692,
      "step": 91030
    },
    {
      "epoch": 0.3678131199069155,
      "grad_norm": 432.77459716796875,
      "learning_rate": 3.208308308721224e-05,
      "loss": 95.8793,
      "step": 91040
    },
    {
      "epoch": 0.3678535211722831,
      "grad_norm": 684.714111328125,
      "learning_rate": 3.2080857695680156e-05,
      "loss": 97.4408,
      "step": 91050
    },
    {
      "epoch": 0.36789392243765073,
      "grad_norm": 766.7606201171875,
      "learning_rate": 3.2078632068625534e-05,
      "loss": 86.9324,
      "step": 91060
    },
    {
      "epoch": 0.3679343237030184,
      "grad_norm": 475.80517578125,
      "learning_rate": 3.207640620609177e-05,
      "loss": 51.0135,
      "step": 91070
    },
    {
      "epoch": 0.367974724968386,
      "grad_norm": 651.9415283203125,
      "learning_rate": 3.2074180108122255e-05,
      "loss": 56.6108,
      "step": 91080
    },
    {
      "epoch": 0.36801512623375365,
      "grad_norm": 604.5901489257812,
      "learning_rate": 3.207195377476037e-05,
      "loss": 64.7571,
      "step": 91090
    },
    {
      "epoch": 0.3680555274991213,
      "grad_norm": 940.1326293945312,
      "learning_rate": 3.2069727206049536e-05,
      "loss": 41.8473,
      "step": 91100
    },
    {
      "epoch": 0.3680959287644889,
      "grad_norm": 799.6327514648438,
      "learning_rate": 3.2067500402033154e-05,
      "loss": 65.0775,
      "step": 91110
    },
    {
      "epoch": 0.3681363300298565,
      "grad_norm": 275.72406005859375,
      "learning_rate": 3.206527336275464e-05,
      "loss": 106.8737,
      "step": 91120
    },
    {
      "epoch": 0.36817673129522416,
      "grad_norm": 1342.920166015625,
      "learning_rate": 3.206304608825741e-05,
      "loss": 68.7097,
      "step": 91130
    },
    {
      "epoch": 0.3682171325605918,
      "grad_norm": 627.02783203125,
      "learning_rate": 3.206081857858489e-05,
      "loss": 65.4008,
      "step": 91140
    },
    {
      "epoch": 0.36825753382595944,
      "grad_norm": 1053.357666015625,
      "learning_rate": 3.20585908337805e-05,
      "loss": 44.5238,
      "step": 91150
    },
    {
      "epoch": 0.3682979350913271,
      "grad_norm": 383.9593200683594,
      "learning_rate": 3.205636285388767e-05,
      "loss": 53.7099,
      "step": 91160
    },
    {
      "epoch": 0.3683383363566947,
      "grad_norm": 287.8760986328125,
      "learning_rate": 3.205413463894984e-05,
      "loss": 78.2327,
      "step": 91170
    },
    {
      "epoch": 0.3683787376220623,
      "grad_norm": 694.3605346679688,
      "learning_rate": 3.2051906189010456e-05,
      "loss": 77.24,
      "step": 91180
    },
    {
      "epoch": 0.36841913888742994,
      "grad_norm": 893.4785766601562,
      "learning_rate": 3.204967750411295e-05,
      "loss": 40.0852,
      "step": 91190
    },
    {
      "epoch": 0.3684595401527976,
      "grad_norm": 1242.676513671875,
      "learning_rate": 3.2047448584300775e-05,
      "loss": 56.3566,
      "step": 91200
    },
    {
      "epoch": 0.3684999414181652,
      "grad_norm": 453.7666320800781,
      "learning_rate": 3.204521942961739e-05,
      "loss": 51.7536,
      "step": 91210
    },
    {
      "epoch": 0.36854034268353286,
      "grad_norm": 752.3657836914062,
      "learning_rate": 3.2042990040106255e-05,
      "loss": 67.9474,
      "step": 91220
    },
    {
      "epoch": 0.3685807439489005,
      "grad_norm": 0.0,
      "learning_rate": 3.204076041581082e-05,
      "loss": 59.873,
      "step": 91230
    },
    {
      "epoch": 0.3686211452142681,
      "grad_norm": 923.444580078125,
      "learning_rate": 3.2038530556774574e-05,
      "loss": 54.3655,
      "step": 91240
    },
    {
      "epoch": 0.3686615464796357,
      "grad_norm": 761.899169921875,
      "learning_rate": 3.203630046304097e-05,
      "loss": 65.4671,
      "step": 91250
    },
    {
      "epoch": 0.36870194774500337,
      "grad_norm": 309.4624938964844,
      "learning_rate": 3.203407013465349e-05,
      "loss": 66.5682,
      "step": 91260
    },
    {
      "epoch": 0.368742349010371,
      "grad_norm": 670.3491821289062,
      "learning_rate": 3.2031839571655625e-05,
      "loss": 87.206,
      "step": 91270
    },
    {
      "epoch": 0.36878275027573865,
      "grad_norm": 915.1228637695312,
      "learning_rate": 3.2029608774090845e-05,
      "loss": 71.0715,
      "step": 91280
    },
    {
      "epoch": 0.3688231515411063,
      "grad_norm": 728.1142578125,
      "learning_rate": 3.202737774200265e-05,
      "loss": 76.3794,
      "step": 91290
    },
    {
      "epoch": 0.36886355280647387,
      "grad_norm": 1065.0478515625,
      "learning_rate": 3.202514647543454e-05,
      "loss": 54.8924,
      "step": 91300
    },
    {
      "epoch": 0.3689039540718415,
      "grad_norm": 516.5827026367188,
      "learning_rate": 3.2022914974429995e-05,
      "loss": 67.8348,
      "step": 91310
    },
    {
      "epoch": 0.36894435533720915,
      "grad_norm": 794.9502563476562,
      "learning_rate": 3.2020683239032545e-05,
      "loss": 55.1017,
      "step": 91320
    },
    {
      "epoch": 0.3689847566025768,
      "grad_norm": 1035.4910888671875,
      "learning_rate": 3.2018451269285677e-05,
      "loss": 75.9056,
      "step": 91330
    },
    {
      "epoch": 0.36902515786794443,
      "grad_norm": 586.384765625,
      "learning_rate": 3.201621906523293e-05,
      "loss": 46.815,
      "step": 91340
    },
    {
      "epoch": 0.36906555913331207,
      "grad_norm": 2527.49609375,
      "learning_rate": 3.201398662691779e-05,
      "loss": 98.0505,
      "step": 91350
    },
    {
      "epoch": 0.3691059603986797,
      "grad_norm": 678.8090209960938,
      "learning_rate": 3.2011753954383805e-05,
      "loss": 64.6653,
      "step": 91360
    },
    {
      "epoch": 0.3691463616640473,
      "grad_norm": 695.07470703125,
      "learning_rate": 3.200952104767448e-05,
      "loss": 39.9693,
      "step": 91370
    },
    {
      "epoch": 0.36918676292941494,
      "grad_norm": 322.0420227050781,
      "learning_rate": 3.200728790683338e-05,
      "loss": 67.0857,
      "step": 91380
    },
    {
      "epoch": 0.3692271641947826,
      "grad_norm": 729.5950317382812,
      "learning_rate": 3.2005054531904006e-05,
      "loss": 87.5696,
      "step": 91390
    },
    {
      "epoch": 0.3692675654601502,
      "grad_norm": 509.17108154296875,
      "learning_rate": 3.200282092292991e-05,
      "loss": 64.9555,
      "step": 91400
    },
    {
      "epoch": 0.36930796672551786,
      "grad_norm": 1967.5986328125,
      "learning_rate": 3.200058707995465e-05,
      "loss": 58.8442,
      "step": 91410
    },
    {
      "epoch": 0.3693483679908855,
      "grad_norm": 914.3970947265625,
      "learning_rate": 3.1998353003021766e-05,
      "loss": 38.4076,
      "step": 91420
    },
    {
      "epoch": 0.3693887692562531,
      "grad_norm": 810.2129516601562,
      "learning_rate": 3.199611869217481e-05,
      "loss": 66.5225,
      "step": 91430
    },
    {
      "epoch": 0.3694291705216207,
      "grad_norm": 1145.6221923828125,
      "learning_rate": 3.1993884147457345e-05,
      "loss": 62.7686,
      "step": 91440
    },
    {
      "epoch": 0.36946957178698836,
      "grad_norm": 496.845703125,
      "learning_rate": 3.199164936891293e-05,
      "loss": 87.6221,
      "step": 91450
    },
    {
      "epoch": 0.369509973052356,
      "grad_norm": 420.1404113769531,
      "learning_rate": 3.198941435658514e-05,
      "loss": 56.4746,
      "step": 91460
    },
    {
      "epoch": 0.36955037431772364,
      "grad_norm": 629.4871826171875,
      "learning_rate": 3.1987179110517546e-05,
      "loss": 56.3853,
      "step": 91470
    },
    {
      "epoch": 0.3695907755830913,
      "grad_norm": 625.9359741210938,
      "learning_rate": 3.198494363075372e-05,
      "loss": 75.2549,
      "step": 91480
    },
    {
      "epoch": 0.3696311768484589,
      "grad_norm": 868.8189086914062,
      "learning_rate": 3.198270791733725e-05,
      "loss": 68.2451,
      "step": 91490
    },
    {
      "epoch": 0.3696715781138265,
      "grad_norm": 607.7655029296875,
      "learning_rate": 3.198047197031172e-05,
      "loss": 69.5034,
      "step": 91500
    },
    {
      "epoch": 0.36971197937919414,
      "grad_norm": 1387.5396728515625,
      "learning_rate": 3.197823578972072e-05,
      "loss": 85.6114,
      "step": 91510
    },
    {
      "epoch": 0.3697523806445618,
      "grad_norm": 464.6197814941406,
      "learning_rate": 3.1975999375607854e-05,
      "loss": 70.0409,
      "step": 91520
    },
    {
      "epoch": 0.3697927819099294,
      "grad_norm": 885.7677001953125,
      "learning_rate": 3.1973762728016705e-05,
      "loss": 77.555,
      "step": 91530
    },
    {
      "epoch": 0.36983318317529706,
      "grad_norm": 1390.1436767578125,
      "learning_rate": 3.1971525846990886e-05,
      "loss": 101.7281,
      "step": 91540
    },
    {
      "epoch": 0.3698735844406647,
      "grad_norm": 598.4069213867188,
      "learning_rate": 3.1969288732574015e-05,
      "loss": 69.8319,
      "step": 91550
    },
    {
      "epoch": 0.3699139857060323,
      "grad_norm": 491.0005187988281,
      "learning_rate": 3.196705138480969e-05,
      "loss": 46.9911,
      "step": 91560
    },
    {
      "epoch": 0.36995438697139993,
      "grad_norm": 2121.340576171875,
      "learning_rate": 3.196481380374154e-05,
      "loss": 70.583,
      "step": 91570
    },
    {
      "epoch": 0.36999478823676757,
      "grad_norm": 686.2039184570312,
      "learning_rate": 3.1962575989413185e-05,
      "loss": 71.3919,
      "step": 91580
    },
    {
      "epoch": 0.3700351895021352,
      "grad_norm": 1019.8297119140625,
      "learning_rate": 3.196033794186826e-05,
      "loss": 41.0625,
      "step": 91590
    },
    {
      "epoch": 0.37007559076750285,
      "grad_norm": 882.1055297851562,
      "learning_rate": 3.195809966115038e-05,
      "loss": 56.5875,
      "step": 91600
    },
    {
      "epoch": 0.3701159920328705,
      "grad_norm": 1562.2581787109375,
      "learning_rate": 3.1955861147303194e-05,
      "loss": 48.7799,
      "step": 91610
    },
    {
      "epoch": 0.3701563932982381,
      "grad_norm": 542.5730590820312,
      "learning_rate": 3.195362240037034e-05,
      "loss": 51.144,
      "step": 91620
    },
    {
      "epoch": 0.3701967945636057,
      "grad_norm": 602.9548950195312,
      "learning_rate": 3.1951383420395456e-05,
      "loss": 46.5236,
      "step": 91630
    },
    {
      "epoch": 0.37023719582897335,
      "grad_norm": 653.1578369140625,
      "learning_rate": 3.194914420742221e-05,
      "loss": 70.5503,
      "step": 91640
    },
    {
      "epoch": 0.370277597094341,
      "grad_norm": 539.1424560546875,
      "learning_rate": 3.194690476149425e-05,
      "loss": 42.9432,
      "step": 91650
    },
    {
      "epoch": 0.37031799835970863,
      "grad_norm": 1839.1611328125,
      "learning_rate": 3.194466508265522e-05,
      "loss": 62.7661,
      "step": 91660
    },
    {
      "epoch": 0.3703583996250763,
      "grad_norm": 387.30133056640625,
      "learning_rate": 3.1942425170948795e-05,
      "loss": 50.549,
      "step": 91670
    },
    {
      "epoch": 0.3703988008904439,
      "grad_norm": 657.3198852539062,
      "learning_rate": 3.194018502641864e-05,
      "loss": 48.9035,
      "step": 91680
    },
    {
      "epoch": 0.3704392021558115,
      "grad_norm": 1220.2711181640625,
      "learning_rate": 3.193794464910844e-05,
      "loss": 50.7038,
      "step": 91690
    },
    {
      "epoch": 0.37047960342117914,
      "grad_norm": 956.4788818359375,
      "learning_rate": 3.193570403906186e-05,
      "loss": 71.9509,
      "step": 91700
    },
    {
      "epoch": 0.3705200046865468,
      "grad_norm": 331.8046875,
      "learning_rate": 3.193346319632258e-05,
      "loss": 64.4131,
      "step": 91710
    },
    {
      "epoch": 0.3705604059519144,
      "grad_norm": 1147.4146728515625,
      "learning_rate": 3.193122212093429e-05,
      "loss": 68.8306,
      "step": 91720
    },
    {
      "epoch": 0.37060080721728206,
      "grad_norm": 540.6961059570312,
      "learning_rate": 3.192898081294069e-05,
      "loss": 78.8963,
      "step": 91730
    },
    {
      "epoch": 0.3706412084826497,
      "grad_norm": 554.406005859375,
      "learning_rate": 3.192673927238547e-05,
      "loss": 54.2971,
      "step": 91740
    },
    {
      "epoch": 0.3706816097480173,
      "grad_norm": 1999.7381591796875,
      "learning_rate": 3.1924497499312317e-05,
      "loss": 96.5161,
      "step": 91750
    },
    {
      "epoch": 0.3707220110133849,
      "grad_norm": 467.8526306152344,
      "learning_rate": 3.1922255493764956e-05,
      "loss": 57.5308,
      "step": 91760
    },
    {
      "epoch": 0.37076241227875256,
      "grad_norm": 1122.2821044921875,
      "learning_rate": 3.192001325578708e-05,
      "loss": 54.3601,
      "step": 91770
    },
    {
      "epoch": 0.3708028135441202,
      "grad_norm": 699.6976318359375,
      "learning_rate": 3.1917770785422406e-05,
      "loss": 51.2669,
      "step": 91780
    },
    {
      "epoch": 0.37084321480948784,
      "grad_norm": 1274.6541748046875,
      "learning_rate": 3.1915528082714664e-05,
      "loss": 59.7874,
      "step": 91790
    },
    {
      "epoch": 0.3708836160748555,
      "grad_norm": 600.6614379882812,
      "learning_rate": 3.191328514770757e-05,
      "loss": 72.3069,
      "step": 91800
    },
    {
      "epoch": 0.3709240173402231,
      "grad_norm": 1041.9267578125,
      "learning_rate": 3.1911041980444836e-05,
      "loss": 99.4359,
      "step": 91810
    },
    {
      "epoch": 0.3709644186055907,
      "grad_norm": 436.3846435546875,
      "learning_rate": 3.190879858097021e-05,
      "loss": 45.6987,
      "step": 91820
    },
    {
      "epoch": 0.37100481987095835,
      "grad_norm": 628.568115234375,
      "learning_rate": 3.190655494932742e-05,
      "loss": 92.0801,
      "step": 91830
    },
    {
      "epoch": 0.371045221136326,
      "grad_norm": 1433.095703125,
      "learning_rate": 3.190431108556022e-05,
      "loss": 49.9664,
      "step": 91840
    },
    {
      "epoch": 0.3710856224016936,
      "grad_norm": 1095.739501953125,
      "learning_rate": 3.190206698971235e-05,
      "loss": 51.1511,
      "step": 91850
    },
    {
      "epoch": 0.37112602366706127,
      "grad_norm": 1261.6505126953125,
      "learning_rate": 3.189982266182755e-05,
      "loss": 46.6542,
      "step": 91860
    },
    {
      "epoch": 0.3711664249324289,
      "grad_norm": 443.4308166503906,
      "learning_rate": 3.189757810194958e-05,
      "loss": 59.0583,
      "step": 91870
    },
    {
      "epoch": 0.3712068261977965,
      "grad_norm": 1636.2860107421875,
      "learning_rate": 3.18953333101222e-05,
      "loss": 62.8327,
      "step": 91880
    },
    {
      "epoch": 0.37124722746316413,
      "grad_norm": 648.6049194335938,
      "learning_rate": 3.189308828638917e-05,
      "loss": 60.4501,
      "step": 91890
    },
    {
      "epoch": 0.37128762872853177,
      "grad_norm": 839.2045288085938,
      "learning_rate": 3.189084303079427e-05,
      "loss": 39.8968,
      "step": 91900
    },
    {
      "epoch": 0.3713280299938994,
      "grad_norm": 927.9185180664062,
      "learning_rate": 3.1888597543381255e-05,
      "loss": 70.0525,
      "step": 91910
    },
    {
      "epoch": 0.37136843125926705,
      "grad_norm": 1390.071533203125,
      "learning_rate": 3.1886351824193916e-05,
      "loss": 46.6219,
      "step": 91920
    },
    {
      "epoch": 0.3714088325246347,
      "grad_norm": 2126.73486328125,
      "learning_rate": 3.188410587327602e-05,
      "loss": 60.301,
      "step": 91930
    },
    {
      "epoch": 0.3714492337900023,
      "grad_norm": 605.2738037109375,
      "learning_rate": 3.188185969067137e-05,
      "loss": 62.8989,
      "step": 91940
    },
    {
      "epoch": 0.3714896350553699,
      "grad_norm": 977.8110961914062,
      "learning_rate": 3.187961327642374e-05,
      "loss": 62.2956,
      "step": 91950
    },
    {
      "epoch": 0.37153003632073756,
      "grad_norm": 464.37969970703125,
      "learning_rate": 3.1877366630576945e-05,
      "loss": 52.0955,
      "step": 91960
    },
    {
      "epoch": 0.3715704375861052,
      "grad_norm": 710.4423217773438,
      "learning_rate": 3.1875119753174766e-05,
      "loss": 90.9263,
      "step": 91970
    },
    {
      "epoch": 0.37161083885147284,
      "grad_norm": 1177.0816650390625,
      "learning_rate": 3.187287264426101e-05,
      "loss": 89.1915,
      "step": 91980
    },
    {
      "epoch": 0.3716512401168405,
      "grad_norm": 356.5927429199219,
      "learning_rate": 3.18706253038795e-05,
      "loss": 41.8411,
      "step": 91990
    },
    {
      "epoch": 0.3716916413822081,
      "grad_norm": 539.5660400390625,
      "learning_rate": 3.1868377732074034e-05,
      "loss": 57.2294,
      "step": 92000
    },
    {
      "epoch": 0.3717320426475757,
      "grad_norm": 644.1307983398438,
      "learning_rate": 3.1866129928888425e-05,
      "loss": 68.3304,
      "step": 92010
    },
    {
      "epoch": 0.37177244391294334,
      "grad_norm": 610.5741577148438,
      "learning_rate": 3.186388189436652e-05,
      "loss": 49.4517,
      "step": 92020
    },
    {
      "epoch": 0.371812845178311,
      "grad_norm": 502.1660461425781,
      "learning_rate": 3.186163362855212e-05,
      "loss": 56.3577,
      "step": 92030
    },
    {
      "epoch": 0.3718532464436786,
      "grad_norm": 506.15301513671875,
      "learning_rate": 3.185938513148906e-05,
      "loss": 56.0064,
      "step": 92040
    },
    {
      "epoch": 0.37189364770904626,
      "grad_norm": 498.75146484375,
      "learning_rate": 3.185713640322119e-05,
      "loss": 56.7962,
      "step": 92050
    },
    {
      "epoch": 0.3719340489744139,
      "grad_norm": 972.7112426757812,
      "learning_rate": 3.185488744379234e-05,
      "loss": 70.0248,
      "step": 92060
    },
    {
      "epoch": 0.3719744502397815,
      "grad_norm": 934.3443603515625,
      "learning_rate": 3.185263825324635e-05,
      "loss": 71.4783,
      "step": 92070
    },
    {
      "epoch": 0.3720148515051491,
      "grad_norm": 1344.581298828125,
      "learning_rate": 3.185038883162708e-05,
      "loss": 85.8356,
      "step": 92080
    },
    {
      "epoch": 0.37205525277051676,
      "grad_norm": 883.6851196289062,
      "learning_rate": 3.184813917897838e-05,
      "loss": 58.2341,
      "step": 92090
    },
    {
      "epoch": 0.3720956540358844,
      "grad_norm": 564.726318359375,
      "learning_rate": 3.18458892953441e-05,
      "loss": 54.6441,
      "step": 92100
    },
    {
      "epoch": 0.37213605530125204,
      "grad_norm": 431.2422180175781,
      "learning_rate": 3.1843639180768115e-05,
      "loss": 76.4143,
      "step": 92110
    },
    {
      "epoch": 0.3721764565666197,
      "grad_norm": 497.0868225097656,
      "learning_rate": 3.184138883529429e-05,
      "loss": 60.8222,
      "step": 92120
    },
    {
      "epoch": 0.3722168578319873,
      "grad_norm": 863.14208984375,
      "learning_rate": 3.183913825896649e-05,
      "loss": 93.3023,
      "step": 92130
    },
    {
      "epoch": 0.3722572590973549,
      "grad_norm": 752.2738647460938,
      "learning_rate": 3.1836887451828595e-05,
      "loss": 77.4681,
      "step": 92140
    },
    {
      "epoch": 0.37229766036272255,
      "grad_norm": 302.57965087890625,
      "learning_rate": 3.183463641392448e-05,
      "loss": 74.6553,
      "step": 92150
    },
    {
      "epoch": 0.3723380616280902,
      "grad_norm": 598.8102416992188,
      "learning_rate": 3.183238514529804e-05,
      "loss": 56.2187,
      "step": 92160
    },
    {
      "epoch": 0.37237846289345783,
      "grad_norm": 1294.8486328125,
      "learning_rate": 3.183013364599316e-05,
      "loss": 51.3917,
      "step": 92170
    },
    {
      "epoch": 0.37241886415882547,
      "grad_norm": 826.580810546875,
      "learning_rate": 3.1827881916053734e-05,
      "loss": 53.6744,
      "step": 92180
    },
    {
      "epoch": 0.3724592654241931,
      "grad_norm": 687.9397583007812,
      "learning_rate": 3.182562995552366e-05,
      "loss": 68.3652,
      "step": 92190
    },
    {
      "epoch": 0.3724996666895607,
      "grad_norm": 269.9486999511719,
      "learning_rate": 3.182337776444684e-05,
      "loss": 57.6726,
      "step": 92200
    },
    {
      "epoch": 0.37254006795492833,
      "grad_norm": 844.561767578125,
      "learning_rate": 3.182112534286719e-05,
      "loss": 64.4658,
      "step": 92210
    },
    {
      "epoch": 0.372580469220296,
      "grad_norm": 569.0249633789062,
      "learning_rate": 3.181887269082861e-05,
      "loss": 73.5815,
      "step": 92220
    },
    {
      "epoch": 0.3726208704856636,
      "grad_norm": 1159.74853515625,
      "learning_rate": 3.1816619808375016e-05,
      "loss": 66.0471,
      "step": 92230
    },
    {
      "epoch": 0.37266127175103125,
      "grad_norm": 561.364013671875,
      "learning_rate": 3.1814366695550346e-05,
      "loss": 58.5301,
      "step": 92240
    },
    {
      "epoch": 0.3727016730163989,
      "grad_norm": 1479.8369140625,
      "learning_rate": 3.181211335239851e-05,
      "loss": 63.9735,
      "step": 92250
    },
    {
      "epoch": 0.3727420742817665,
      "grad_norm": 1265.6650390625,
      "learning_rate": 3.1809859778963445e-05,
      "loss": 47.1746,
      "step": 92260
    },
    {
      "epoch": 0.3727824755471341,
      "grad_norm": 552.6262817382812,
      "learning_rate": 3.180760597528908e-05,
      "loss": 58.6089,
      "step": 92270
    },
    {
      "epoch": 0.37282287681250176,
      "grad_norm": 943.9286499023438,
      "learning_rate": 3.180535194141936e-05,
      "loss": 83.6766,
      "step": 92280
    },
    {
      "epoch": 0.3728632780778694,
      "grad_norm": 863.5250244140625,
      "learning_rate": 3.180309767739823e-05,
      "loss": 73.9203,
      "step": 92290
    },
    {
      "epoch": 0.37290367934323704,
      "grad_norm": 0.0,
      "learning_rate": 3.1800843183269625e-05,
      "loss": 49.264,
      "step": 92300
    },
    {
      "epoch": 0.3729440806086047,
      "grad_norm": 599.714111328125,
      "learning_rate": 3.179858845907752e-05,
      "loss": 68.3465,
      "step": 92310
    },
    {
      "epoch": 0.3729844818739723,
      "grad_norm": 883.240966796875,
      "learning_rate": 3.179633350486584e-05,
      "loss": 76.5089,
      "step": 92320
    },
    {
      "epoch": 0.3730248831393399,
      "grad_norm": 558.8479614257812,
      "learning_rate": 3.179407832067858e-05,
      "loss": 83.0287,
      "step": 92330
    },
    {
      "epoch": 0.37306528440470754,
      "grad_norm": 1013.3880004882812,
      "learning_rate": 3.179182290655969e-05,
      "loss": 81.3642,
      "step": 92340
    },
    {
      "epoch": 0.3731056856700752,
      "grad_norm": 433.5400695800781,
      "learning_rate": 3.178956726255314e-05,
      "loss": 49.3599,
      "step": 92350
    },
    {
      "epoch": 0.3731460869354428,
      "grad_norm": 978.0261840820312,
      "learning_rate": 3.1787311388702894e-05,
      "loss": 57.0135,
      "step": 92360
    },
    {
      "epoch": 0.37318648820081046,
      "grad_norm": 1010.7636108398438,
      "learning_rate": 3.178505528505296e-05,
      "loss": 66.5352,
      "step": 92370
    },
    {
      "epoch": 0.3732268894661781,
      "grad_norm": 241.92910766601562,
      "learning_rate": 3.17827989516473e-05,
      "loss": 60.54,
      "step": 92380
    },
    {
      "epoch": 0.3732672907315457,
      "grad_norm": 665.576416015625,
      "learning_rate": 3.1780542388529906e-05,
      "loss": 77.0888,
      "step": 92390
    },
    {
      "epoch": 0.3733076919969133,
      "grad_norm": 715.9092407226562,
      "learning_rate": 3.177828559574477e-05,
      "loss": 63.322,
      "step": 92400
    },
    {
      "epoch": 0.37334809326228097,
      "grad_norm": 544.0072021484375,
      "learning_rate": 3.1776028573335906e-05,
      "loss": 60.1967,
      "step": 92410
    },
    {
      "epoch": 0.3733884945276486,
      "grad_norm": 856.43701171875,
      "learning_rate": 3.177377132134729e-05,
      "loss": 73.3287,
      "step": 92420
    },
    {
      "epoch": 0.37342889579301625,
      "grad_norm": 749.0526733398438,
      "learning_rate": 3.177151383982295e-05,
      "loss": 59.487,
      "step": 92430
    },
    {
      "epoch": 0.3734692970583839,
      "grad_norm": 560.5958862304688,
      "learning_rate": 3.176925612880688e-05,
      "loss": 55.7059,
      "step": 92440
    },
    {
      "epoch": 0.3735096983237515,
      "grad_norm": 604.20556640625,
      "learning_rate": 3.176699818834311e-05,
      "loss": 55.1843,
      "step": 92450
    },
    {
      "epoch": 0.3735500995891191,
      "grad_norm": 398.9888000488281,
      "learning_rate": 3.1764740018475646e-05,
      "loss": 75.0972,
      "step": 92460
    },
    {
      "epoch": 0.37359050085448675,
      "grad_norm": 466.8522033691406,
      "learning_rate": 3.1762481619248516e-05,
      "loss": 46.4847,
      "step": 92470
    },
    {
      "epoch": 0.3736309021198544,
      "grad_norm": 548.5853271484375,
      "learning_rate": 3.1760222990705756e-05,
      "loss": 80.4183,
      "step": 92480
    },
    {
      "epoch": 0.37367130338522203,
      "grad_norm": 709.9428100585938,
      "learning_rate": 3.175796413289139e-05,
      "loss": 56.5794,
      "step": 92490
    },
    {
      "epoch": 0.37371170465058967,
      "grad_norm": 694.28076171875,
      "learning_rate": 3.1755705045849465e-05,
      "loss": 65.296,
      "step": 92500
    },
    {
      "epoch": 0.3737521059159573,
      "grad_norm": 1514.2996826171875,
      "learning_rate": 3.175344572962402e-05,
      "loss": 99.0674,
      "step": 92510
    },
    {
      "epoch": 0.3737925071813249,
      "grad_norm": 1029.3662109375,
      "learning_rate": 3.175118618425909e-05,
      "loss": 75.531,
      "step": 92520
    },
    {
      "epoch": 0.37383290844669254,
      "grad_norm": 1189.4278564453125,
      "learning_rate": 3.174892640979875e-05,
      "loss": 81.1586,
      "step": 92530
    },
    {
      "epoch": 0.3738733097120602,
      "grad_norm": 513.4951782226562,
      "learning_rate": 3.174666640628702e-05,
      "loss": 86.4052,
      "step": 92540
    },
    {
      "epoch": 0.3739137109774278,
      "grad_norm": 467.4132080078125,
      "learning_rate": 3.174440617376799e-05,
      "loss": 52.4841,
      "step": 92550
    },
    {
      "epoch": 0.37395411224279546,
      "grad_norm": 628.9441528320312,
      "learning_rate": 3.1742145712285725e-05,
      "loss": 70.6902,
      "step": 92560
    },
    {
      "epoch": 0.3739945135081631,
      "grad_norm": 816.73828125,
      "learning_rate": 3.173988502188428e-05,
      "loss": 97.6481,
      "step": 92570
    },
    {
      "epoch": 0.3740349147735307,
      "grad_norm": 1033.385986328125,
      "learning_rate": 3.173762410260772e-05,
      "loss": 78.7853,
      "step": 92580
    },
    {
      "epoch": 0.3740753160388983,
      "grad_norm": 466.20404052734375,
      "learning_rate": 3.173536295450014e-05,
      "loss": 47.6014,
      "step": 92590
    },
    {
      "epoch": 0.37411571730426596,
      "grad_norm": 337.6639099121094,
      "learning_rate": 3.173310157760563e-05,
      "loss": 53.6203,
      "step": 92600
    },
    {
      "epoch": 0.3741561185696336,
      "grad_norm": 803.3235473632812,
      "learning_rate": 3.173083997196825e-05,
      "loss": 66.1968,
      "step": 92610
    },
    {
      "epoch": 0.37419651983500124,
      "grad_norm": 960.7337036132812,
      "learning_rate": 3.172857813763211e-05,
      "loss": 30.3207,
      "step": 92620
    },
    {
      "epoch": 0.3742369211003689,
      "grad_norm": 782.6198120117188,
      "learning_rate": 3.172631607464131e-05,
      "loss": 81.5086,
      "step": 92630
    },
    {
      "epoch": 0.3742773223657365,
      "grad_norm": 542.2578735351562,
      "learning_rate": 3.1724053783039935e-05,
      "loss": 42.5299,
      "step": 92640
    },
    {
      "epoch": 0.3743177236311041,
      "grad_norm": 134.53843688964844,
      "learning_rate": 3.172179126287209e-05,
      "loss": 61.441,
      "step": 92650
    },
    {
      "epoch": 0.37435812489647174,
      "grad_norm": 511.2257080078125,
      "learning_rate": 3.1719528514181894e-05,
      "loss": 70.0888,
      "step": 92660
    },
    {
      "epoch": 0.3743985261618394,
      "grad_norm": 484.2981262207031,
      "learning_rate": 3.1717265537013454e-05,
      "loss": 76.5403,
      "step": 92670
    },
    {
      "epoch": 0.374438927427207,
      "grad_norm": 782.7688598632812,
      "learning_rate": 3.1715002331410886e-05,
      "loss": 52.4916,
      "step": 92680
    },
    {
      "epoch": 0.37447932869257466,
      "grad_norm": 410.0548400878906,
      "learning_rate": 3.1712738897418325e-05,
      "loss": 40.2399,
      "step": 92690
    },
    {
      "epoch": 0.3745197299579423,
      "grad_norm": 345.84930419921875,
      "learning_rate": 3.1710475235079885e-05,
      "loss": 43.5013,
      "step": 92700
    },
    {
      "epoch": 0.3745601312233099,
      "grad_norm": 1913.61669921875,
      "learning_rate": 3.17082113444397e-05,
      "loss": 82.2697,
      "step": 92710
    },
    {
      "epoch": 0.37460053248867753,
      "grad_norm": 1220.21923828125,
      "learning_rate": 3.1705947225541915e-05,
      "loss": 64.3923,
      "step": 92720
    },
    {
      "epoch": 0.37464093375404517,
      "grad_norm": 468.8391418457031,
      "learning_rate": 3.170368287843065e-05,
      "loss": 65.5118,
      "step": 92730
    },
    {
      "epoch": 0.3746813350194128,
      "grad_norm": 716.503173828125,
      "learning_rate": 3.1701418303150067e-05,
      "loss": 75.1031,
      "step": 92740
    },
    {
      "epoch": 0.37472173628478045,
      "grad_norm": 731.4243774414062,
      "learning_rate": 3.169915349974432e-05,
      "loss": 71.5022,
      "step": 92750
    },
    {
      "epoch": 0.3747621375501481,
      "grad_norm": 752.8480834960938,
      "learning_rate": 3.169688846825754e-05,
      "loss": 42.9594,
      "step": 92760
    },
    {
      "epoch": 0.37480253881551573,
      "grad_norm": 967.5347290039062,
      "learning_rate": 3.16946232087339e-05,
      "loss": 56.4926,
      "step": 92770
    },
    {
      "epoch": 0.3748429400808833,
      "grad_norm": 1119.449951171875,
      "learning_rate": 3.169235772121756e-05,
      "loss": 76.5434,
      "step": 92780
    },
    {
      "epoch": 0.37488334134625095,
      "grad_norm": 410.9073486328125,
      "learning_rate": 3.169009200575268e-05,
      "loss": 57.9905,
      "step": 92790
    },
    {
      "epoch": 0.3749237426116186,
      "grad_norm": 550.1514892578125,
      "learning_rate": 3.1687826062383444e-05,
      "loss": 61.867,
      "step": 92800
    },
    {
      "epoch": 0.37496414387698623,
      "grad_norm": 246.54542541503906,
      "learning_rate": 3.168555989115402e-05,
      "loss": 60.9686,
      "step": 92810
    },
    {
      "epoch": 0.3750045451423539,
      "grad_norm": 586.08837890625,
      "learning_rate": 3.1683293492108595e-05,
      "loss": 50.8065,
      "step": 92820
    },
    {
      "epoch": 0.3750449464077215,
      "grad_norm": 884.8882446289062,
      "learning_rate": 3.168102686529135e-05,
      "loss": 48.4871,
      "step": 92830
    },
    {
      "epoch": 0.3750853476730891,
      "grad_norm": 761.7444458007812,
      "learning_rate": 3.167876001074646e-05,
      "loss": 67.6586,
      "step": 92840
    },
    {
      "epoch": 0.37512574893845674,
      "grad_norm": 372.81304931640625,
      "learning_rate": 3.167649292851814e-05,
      "loss": 57.7765,
      "step": 92850
    },
    {
      "epoch": 0.3751661502038244,
      "grad_norm": 193.35227966308594,
      "learning_rate": 3.167422561865058e-05,
      "loss": 50.352,
      "step": 92860
    },
    {
      "epoch": 0.375206551469192,
      "grad_norm": 967.7341918945312,
      "learning_rate": 3.167195808118798e-05,
      "loss": 69.5428,
      "step": 92870
    },
    {
      "epoch": 0.37524695273455966,
      "grad_norm": 574.5850830078125,
      "learning_rate": 3.166969031617455e-05,
      "loss": 53.5927,
      "step": 92880
    },
    {
      "epoch": 0.3752873539999273,
      "grad_norm": 1133.9346923828125,
      "learning_rate": 3.166742232365449e-05,
      "loss": 69.1476,
      "step": 92890
    },
    {
      "epoch": 0.3753277552652949,
      "grad_norm": 547.107421875,
      "learning_rate": 3.166515410367203e-05,
      "loss": 67.2989,
      "step": 92900
    },
    {
      "epoch": 0.3753681565306625,
      "grad_norm": 921.1433715820312,
      "learning_rate": 3.166288565627139e-05,
      "loss": 75.7809,
      "step": 92910
    },
    {
      "epoch": 0.37540855779603016,
      "grad_norm": 380.323486328125,
      "learning_rate": 3.166061698149679e-05,
      "loss": 56.84,
      "step": 92920
    },
    {
      "epoch": 0.3754489590613978,
      "grad_norm": 598.1117553710938,
      "learning_rate": 3.165834807939245e-05,
      "loss": 61.9656,
      "step": 92930
    },
    {
      "epoch": 0.37548936032676544,
      "grad_norm": 235.85311889648438,
      "learning_rate": 3.165607895000262e-05,
      "loss": 52.8294,
      "step": 92940
    },
    {
      "epoch": 0.3755297615921331,
      "grad_norm": 992.2584228515625,
      "learning_rate": 3.1653809593371526e-05,
      "loss": 65.1801,
      "step": 92950
    },
    {
      "epoch": 0.3755701628575007,
      "grad_norm": 912.9456176757812,
      "learning_rate": 3.165154000954341e-05,
      "loss": 73.9485,
      "step": 92960
    },
    {
      "epoch": 0.3756105641228683,
      "grad_norm": 483.7486572265625,
      "learning_rate": 3.164927019856253e-05,
      "loss": 55.6523,
      "step": 92970
    },
    {
      "epoch": 0.37565096538823595,
      "grad_norm": 597.1483764648438,
      "learning_rate": 3.1647000160473126e-05,
      "loss": 77.8391,
      "step": 92980
    },
    {
      "epoch": 0.3756913666536036,
      "grad_norm": 1421.8509521484375,
      "learning_rate": 3.164472989531946e-05,
      "loss": 44.8926,
      "step": 92990
    },
    {
      "epoch": 0.3757317679189712,
      "grad_norm": 944.5368041992188,
      "learning_rate": 3.1642459403145794e-05,
      "loss": 81.4513,
      "step": 93000
    },
    {
      "epoch": 0.37577216918433887,
      "grad_norm": 1083.89794921875,
      "learning_rate": 3.164018868399638e-05,
      "loss": 65.1026,
      "step": 93010
    },
    {
      "epoch": 0.3758125704497065,
      "grad_norm": 796.9815673828125,
      "learning_rate": 3.16379177379155e-05,
      "loss": 80.6928,
      "step": 93020
    },
    {
      "epoch": 0.3758529717150741,
      "grad_norm": 516.5150756835938,
      "learning_rate": 3.163564656494742e-05,
      "loss": 74.6458,
      "step": 93030
    },
    {
      "epoch": 0.37589337298044173,
      "grad_norm": 695.0779418945312,
      "learning_rate": 3.163337516513642e-05,
      "loss": 73.3617,
      "step": 93040
    },
    {
      "epoch": 0.37593377424580937,
      "grad_norm": 661.1953125,
      "learning_rate": 3.1631103538526774e-05,
      "loss": 44.4488,
      "step": 93050
    },
    {
      "epoch": 0.375974175511177,
      "grad_norm": 1036.6097412109375,
      "learning_rate": 3.162883168516279e-05,
      "loss": 81.1605,
      "step": 93060
    },
    {
      "epoch": 0.37601457677654465,
      "grad_norm": 1192.6729736328125,
      "learning_rate": 3.1626559605088734e-05,
      "loss": 107.4897,
      "step": 93070
    },
    {
      "epoch": 0.3760549780419123,
      "grad_norm": 594.5126953125,
      "learning_rate": 3.1624287298348923e-05,
      "loss": 66.0238,
      "step": 93080
    },
    {
      "epoch": 0.37609537930727993,
      "grad_norm": 461.3974609375,
      "learning_rate": 3.1622014764987637e-05,
      "loss": 62.9703,
      "step": 93090
    },
    {
      "epoch": 0.3761357805726475,
      "grad_norm": 220.75631713867188,
      "learning_rate": 3.1619742005049204e-05,
      "loss": 45.0191,
      "step": 93100
    },
    {
      "epoch": 0.37617618183801516,
      "grad_norm": 892.0763549804688,
      "learning_rate": 3.16174690185779e-05,
      "loss": 78.6,
      "step": 93110
    },
    {
      "epoch": 0.3762165831033828,
      "grad_norm": 1821.3922119140625,
      "learning_rate": 3.161519580561807e-05,
      "loss": 59.5127,
      "step": 93120
    },
    {
      "epoch": 0.37625698436875044,
      "grad_norm": 762.4219360351562,
      "learning_rate": 3.1612922366214014e-05,
      "loss": 117.8003,
      "step": 93130
    },
    {
      "epoch": 0.3762973856341181,
      "grad_norm": 660.0880126953125,
      "learning_rate": 3.1610648700410057e-05,
      "loss": 74.9537,
      "step": 93140
    },
    {
      "epoch": 0.3763377868994857,
      "grad_norm": 708.5904541015625,
      "learning_rate": 3.160837480825052e-05,
      "loss": 69.8171,
      "step": 93150
    },
    {
      "epoch": 0.3763781881648533,
      "grad_norm": 272.2256774902344,
      "learning_rate": 3.160610068977975e-05,
      "loss": 32.8539,
      "step": 93160
    },
    {
      "epoch": 0.37641858943022094,
      "grad_norm": 416.69927978515625,
      "learning_rate": 3.1603826345042064e-05,
      "loss": 55.2995,
      "step": 93170
    },
    {
      "epoch": 0.3764589906955886,
      "grad_norm": 756.4736328125,
      "learning_rate": 3.160155177408181e-05,
      "loss": 50.5459,
      "step": 93180
    },
    {
      "epoch": 0.3764993919609562,
      "grad_norm": 1576.5340576171875,
      "learning_rate": 3.159927697694334e-05,
      "loss": 63.1412,
      "step": 93190
    },
    {
      "epoch": 0.37653979322632386,
      "grad_norm": 622.9821166992188,
      "learning_rate": 3.1597001953670985e-05,
      "loss": 37.7336,
      "step": 93200
    },
    {
      "epoch": 0.3765801944916915,
      "grad_norm": 1226.30224609375,
      "learning_rate": 3.159472670430911e-05,
      "loss": 71.5446,
      "step": 93210
    },
    {
      "epoch": 0.3766205957570591,
      "grad_norm": 768.5546264648438,
      "learning_rate": 3.159245122890207e-05,
      "loss": 61.1742,
      "step": 93220
    },
    {
      "epoch": 0.3766609970224267,
      "grad_norm": 1447.1888427734375,
      "learning_rate": 3.159017552749422e-05,
      "loss": 74.656,
      "step": 93230
    },
    {
      "epoch": 0.37670139828779436,
      "grad_norm": 432.21624755859375,
      "learning_rate": 3.158789960012993e-05,
      "loss": 87.346,
      "step": 93240
    },
    {
      "epoch": 0.376741799553162,
      "grad_norm": 1397.281005859375,
      "learning_rate": 3.1585623446853574e-05,
      "loss": 86.6955,
      "step": 93250
    },
    {
      "epoch": 0.37678220081852964,
      "grad_norm": 203.1778564453125,
      "learning_rate": 3.158334706770953e-05,
      "loss": 61.4963,
      "step": 93260
    },
    {
      "epoch": 0.3768226020838973,
      "grad_norm": 396.96685791015625,
      "learning_rate": 3.158107046274217e-05,
      "loss": 24.4406,
      "step": 93270
    },
    {
      "epoch": 0.3768630033492649,
      "grad_norm": 512.5032958984375,
      "learning_rate": 3.157879363199588e-05,
      "loss": 65.7688,
      "step": 93280
    },
    {
      "epoch": 0.3769034046146325,
      "grad_norm": 797.4293823242188,
      "learning_rate": 3.1576516575515034e-05,
      "loss": 48.7656,
      "step": 93290
    },
    {
      "epoch": 0.37694380588000015,
      "grad_norm": 1280.97705078125,
      "learning_rate": 3.157423929334405e-05,
      "loss": 59.6893,
      "step": 93300
    },
    {
      "epoch": 0.3769842071453678,
      "grad_norm": 719.6341552734375,
      "learning_rate": 3.157196178552731e-05,
      "loss": 54.0297,
      "step": 93310
    },
    {
      "epoch": 0.37702460841073543,
      "grad_norm": 513.1243896484375,
      "learning_rate": 3.156968405210921e-05,
      "loss": 51.1877,
      "step": 93320
    },
    {
      "epoch": 0.37706500967610307,
      "grad_norm": 1609.5087890625,
      "learning_rate": 3.156740609313417e-05,
      "loss": 56.3624,
      "step": 93330
    },
    {
      "epoch": 0.3771054109414707,
      "grad_norm": 900.366943359375,
      "learning_rate": 3.156512790864659e-05,
      "loss": 82.3196,
      "step": 93340
    },
    {
      "epoch": 0.3771458122068383,
      "grad_norm": 1422.4246826171875,
      "learning_rate": 3.1562849498690894e-05,
      "loss": 57.0498,
      "step": 93350
    },
    {
      "epoch": 0.37718621347220593,
      "grad_norm": 633.8297119140625,
      "learning_rate": 3.1560570863311486e-05,
      "loss": 51.3569,
      "step": 93360
    },
    {
      "epoch": 0.3772266147375736,
      "grad_norm": 1020.2067260742188,
      "learning_rate": 3.15582920025528e-05,
      "loss": 68.3464,
      "step": 93370
    },
    {
      "epoch": 0.3772670160029412,
      "grad_norm": 452.5904235839844,
      "learning_rate": 3.1556012916459264e-05,
      "loss": 69.8498,
      "step": 93380
    },
    {
      "epoch": 0.37730741726830885,
      "grad_norm": 504.51922607421875,
      "learning_rate": 3.15537336050753e-05,
      "loss": 44.5778,
      "step": 93390
    },
    {
      "epoch": 0.3773478185336765,
      "grad_norm": 1517.9036865234375,
      "learning_rate": 3.155145406844535e-05,
      "loss": 79.1883,
      "step": 93400
    },
    {
      "epoch": 0.37738821979904413,
      "grad_norm": 1100.0196533203125,
      "learning_rate": 3.154917430661387e-05,
      "loss": 57.3691,
      "step": 93410
    },
    {
      "epoch": 0.3774286210644117,
      "grad_norm": 890.3381958007812,
      "learning_rate": 3.154689431962528e-05,
      "loss": 70.4406,
      "step": 93420
    },
    {
      "epoch": 0.37746902232977936,
      "grad_norm": 1641.3446044921875,
      "learning_rate": 3.1544614107524044e-05,
      "loss": 67.4326,
      "step": 93430
    },
    {
      "epoch": 0.377509423595147,
      "grad_norm": 1637.2950439453125,
      "learning_rate": 3.154233367035461e-05,
      "loss": 46.8634,
      "step": 93440
    },
    {
      "epoch": 0.37754982486051464,
      "grad_norm": 740.50048828125,
      "learning_rate": 3.154005300816144e-05,
      "loss": 55.2992,
      "step": 93450
    },
    {
      "epoch": 0.3775902261258823,
      "grad_norm": 541.7504272460938,
      "learning_rate": 3.153777212098899e-05,
      "loss": 48.3187,
      "step": 93460
    },
    {
      "epoch": 0.3776306273912499,
      "grad_norm": 757.1552734375,
      "learning_rate": 3.1535491008881735e-05,
      "loss": 61.7585,
      "step": 93470
    },
    {
      "epoch": 0.3776710286566175,
      "grad_norm": 748.1531372070312,
      "learning_rate": 3.153320967188415e-05,
      "loss": 53.4171,
      "step": 93480
    },
    {
      "epoch": 0.37771142992198514,
      "grad_norm": 1016.3422241210938,
      "learning_rate": 3.1530928110040696e-05,
      "loss": 44.028,
      "step": 93490
    },
    {
      "epoch": 0.3777518311873528,
      "grad_norm": 1564.3062744140625,
      "learning_rate": 3.1528646323395865e-05,
      "loss": 54.3597,
      "step": 93500
    },
    {
      "epoch": 0.3777922324527204,
      "grad_norm": 563.5382690429688,
      "learning_rate": 3.152636431199414e-05,
      "loss": 72.5935,
      "step": 93510
    },
    {
      "epoch": 0.37783263371808806,
      "grad_norm": 491.2622375488281,
      "learning_rate": 3.152408207588001e-05,
      "loss": 48.6152,
      "step": 93520
    },
    {
      "epoch": 0.3778730349834557,
      "grad_norm": 1588.01513671875,
      "learning_rate": 3.152179961509797e-05,
      "loss": 101.9268,
      "step": 93530
    },
    {
      "epoch": 0.3779134362488233,
      "grad_norm": 601.212158203125,
      "learning_rate": 3.1519516929692506e-05,
      "loss": 47.7366,
      "step": 93540
    },
    {
      "epoch": 0.3779538375141909,
      "grad_norm": 628.07177734375,
      "learning_rate": 3.1517234019708136e-05,
      "loss": 59.8987,
      "step": 93550
    },
    {
      "epoch": 0.37799423877955857,
      "grad_norm": 680.4913940429688,
      "learning_rate": 3.151495088518936e-05,
      "loss": 49.9818,
      "step": 93560
    },
    {
      "epoch": 0.3780346400449262,
      "grad_norm": 820.2808227539062,
      "learning_rate": 3.1512667526180686e-05,
      "loss": 53.5809,
      "step": 93570
    },
    {
      "epoch": 0.37807504131029385,
      "grad_norm": 923.7725219726562,
      "learning_rate": 3.1510383942726626e-05,
      "loss": 60.1639,
      "step": 93580
    },
    {
      "epoch": 0.3781154425756615,
      "grad_norm": 618.584716796875,
      "learning_rate": 3.1508100134871705e-05,
      "loss": 76.269,
      "step": 93590
    },
    {
      "epoch": 0.3781558438410291,
      "grad_norm": 201.3993377685547,
      "learning_rate": 3.150581610266046e-05,
      "loss": 36.1293,
      "step": 93600
    },
    {
      "epoch": 0.3781962451063967,
      "grad_norm": 2918.939453125,
      "learning_rate": 3.150353184613739e-05,
      "loss": 86.6181,
      "step": 93610
    },
    {
      "epoch": 0.37823664637176435,
      "grad_norm": 748.3023071289062,
      "learning_rate": 3.150124736534705e-05,
      "loss": 42.7009,
      "step": 93620
    },
    {
      "epoch": 0.378277047637132,
      "grad_norm": 377.7673034667969,
      "learning_rate": 3.149896266033398e-05,
      "loss": 44.2963,
      "step": 93630
    },
    {
      "epoch": 0.37831744890249963,
      "grad_norm": 1187.8922119140625,
      "learning_rate": 3.149667773114271e-05,
      "loss": 78.8231,
      "step": 93640
    },
    {
      "epoch": 0.37835785016786727,
      "grad_norm": 882.2860107421875,
      "learning_rate": 3.1494392577817775e-05,
      "loss": 70.4679,
      "step": 93650
    },
    {
      "epoch": 0.3783982514332349,
      "grad_norm": 799.963134765625,
      "learning_rate": 3.149210720040375e-05,
      "loss": 49.0757,
      "step": 93660
    },
    {
      "epoch": 0.3784386526986025,
      "grad_norm": 1207.2667236328125,
      "learning_rate": 3.148982159894518e-05,
      "loss": 65.3828,
      "step": 93670
    },
    {
      "epoch": 0.37847905396397014,
      "grad_norm": 511.77978515625,
      "learning_rate": 3.1487535773486624e-05,
      "loss": 49.8612,
      "step": 93680
    },
    {
      "epoch": 0.3785194552293378,
      "grad_norm": 339.2344970703125,
      "learning_rate": 3.148524972407263e-05,
      "loss": 36.7039,
      "step": 93690
    },
    {
      "epoch": 0.3785598564947054,
      "grad_norm": 1724.865966796875,
      "learning_rate": 3.14829634507478e-05,
      "loss": 81.2211,
      "step": 93700
    },
    {
      "epoch": 0.37860025776007306,
      "grad_norm": 1382.4368896484375,
      "learning_rate": 3.148067695355667e-05,
      "loss": 83.5194,
      "step": 93710
    },
    {
      "epoch": 0.3786406590254407,
      "grad_norm": 618.5198364257812,
      "learning_rate": 3.1478390232543835e-05,
      "loss": 61.2203,
      "step": 93720
    },
    {
      "epoch": 0.37868106029080834,
      "grad_norm": 514.2406005859375,
      "learning_rate": 3.147610328775388e-05,
      "loss": 73.8755,
      "step": 93730
    },
    {
      "epoch": 0.3787214615561759,
      "grad_norm": 478.9051818847656,
      "learning_rate": 3.1473816119231374e-05,
      "loss": 63.9839,
      "step": 93740
    },
    {
      "epoch": 0.37876186282154356,
      "grad_norm": 649.34521484375,
      "learning_rate": 3.147152872702092e-05,
      "loss": 53.3547,
      "step": 93750
    },
    {
      "epoch": 0.3788022640869112,
      "grad_norm": 583.468994140625,
      "learning_rate": 3.146924111116711e-05,
      "loss": 60.0845,
      "step": 93760
    },
    {
      "epoch": 0.37884266535227884,
      "grad_norm": 1076.1488037109375,
      "learning_rate": 3.146695327171454e-05,
      "loss": 52.8453,
      "step": 93770
    },
    {
      "epoch": 0.3788830666176465,
      "grad_norm": 640.3786010742188,
      "learning_rate": 3.146466520870781e-05,
      "loss": 88.6203,
      "step": 93780
    },
    {
      "epoch": 0.3789234678830141,
      "grad_norm": 546.0957641601562,
      "learning_rate": 3.146237692219153e-05,
      "loss": 73.0292,
      "step": 93790
    },
    {
      "epoch": 0.3789638691483817,
      "grad_norm": 662.7147216796875,
      "learning_rate": 3.146008841221031e-05,
      "loss": 81.7544,
      "step": 93800
    },
    {
      "epoch": 0.37900427041374934,
      "grad_norm": 746.7833251953125,
      "learning_rate": 3.145779967880877e-05,
      "loss": 64.7103,
      "step": 93810
    },
    {
      "epoch": 0.379044671679117,
      "grad_norm": 1038.830810546875,
      "learning_rate": 3.1455510722031516e-05,
      "loss": 82.4053,
      "step": 93820
    },
    {
      "epoch": 0.3790850729444846,
      "grad_norm": 643.62353515625,
      "learning_rate": 3.145322154192319e-05,
      "loss": 70.2069,
      "step": 93830
    },
    {
      "epoch": 0.37912547420985226,
      "grad_norm": 1289.2843017578125,
      "learning_rate": 3.145093213852842e-05,
      "loss": 65.6404,
      "step": 93840
    },
    {
      "epoch": 0.3791658754752199,
      "grad_norm": 790.8561401367188,
      "learning_rate": 3.144864251189182e-05,
      "loss": 121.4687,
      "step": 93850
    },
    {
      "epoch": 0.3792062767405875,
      "grad_norm": 392.8358154296875,
      "learning_rate": 3.144635266205804e-05,
      "loss": 51.6023,
      "step": 93860
    },
    {
      "epoch": 0.37924667800595513,
      "grad_norm": 264.0917663574219,
      "learning_rate": 3.1444062589071735e-05,
      "loss": 56.2863,
      "step": 93870
    },
    {
      "epoch": 0.37928707927132277,
      "grad_norm": 614.1214599609375,
      "learning_rate": 3.1441772292977534e-05,
      "loss": 64.174,
      "step": 93880
    },
    {
      "epoch": 0.3793274805366904,
      "grad_norm": 508.02008056640625,
      "learning_rate": 3.1439481773820086e-05,
      "loss": 71.7199,
      "step": 93890
    },
    {
      "epoch": 0.37936788180205805,
      "grad_norm": 404.8476257324219,
      "learning_rate": 3.143719103164405e-05,
      "loss": 42.7474,
      "step": 93900
    },
    {
      "epoch": 0.3794082830674257,
      "grad_norm": 391.021728515625,
      "learning_rate": 3.143490006649409e-05,
      "loss": 74.7712,
      "step": 93910
    },
    {
      "epoch": 0.37944868433279333,
      "grad_norm": 822.2650756835938,
      "learning_rate": 3.143260887841487e-05,
      "loss": 57.2424,
      "step": 93920
    },
    {
      "epoch": 0.3794890855981609,
      "grad_norm": 727.6133422851562,
      "learning_rate": 3.1430317467451045e-05,
      "loss": 53.5675,
      "step": 93930
    },
    {
      "epoch": 0.37952948686352855,
      "grad_norm": 485.53179931640625,
      "learning_rate": 3.1428025833647306e-05,
      "loss": 63.1455,
      "step": 93940
    },
    {
      "epoch": 0.3795698881288962,
      "grad_norm": 574.5487060546875,
      "learning_rate": 3.1425733977048313e-05,
      "loss": 67.5347,
      "step": 93950
    },
    {
      "epoch": 0.37961028939426383,
      "grad_norm": 1004.76171875,
      "learning_rate": 3.142344189769876e-05,
      "loss": 75.701,
      "step": 93960
    },
    {
      "epoch": 0.3796506906596315,
      "grad_norm": 857.5787963867188,
      "learning_rate": 3.142114959564332e-05,
      "loss": 56.6185,
      "step": 93970
    },
    {
      "epoch": 0.3796910919249991,
      "grad_norm": 666.8101806640625,
      "learning_rate": 3.141885707092669e-05,
      "loss": 41.2738,
      "step": 93980
    },
    {
      "epoch": 0.3797314931903667,
      "grad_norm": 548.367919921875,
      "learning_rate": 3.141656432359356e-05,
      "loss": 66.6917,
      "step": 93990
    },
    {
      "epoch": 0.37977189445573434,
      "grad_norm": 550.247802734375,
      "learning_rate": 3.141427135368864e-05,
      "loss": 83.6517,
      "step": 94000
    },
    {
      "epoch": 0.379812295721102,
      "grad_norm": 547.6636352539062,
      "learning_rate": 3.141197816125662e-05,
      "loss": 41.3346,
      "step": 94010
    },
    {
      "epoch": 0.3798526969864696,
      "grad_norm": 880.8660278320312,
      "learning_rate": 3.14096847463422e-05,
      "loss": 50.6206,
      "step": 94020
    },
    {
      "epoch": 0.37989309825183726,
      "grad_norm": 488.7218017578125,
      "learning_rate": 3.140739110899011e-05,
      "loss": 54.8867,
      "step": 94030
    },
    {
      "epoch": 0.3799334995172049,
      "grad_norm": 977.4651489257812,
      "learning_rate": 3.140509724924506e-05,
      "loss": 66.8367,
      "step": 94040
    },
    {
      "epoch": 0.37997390078257254,
      "grad_norm": 763.7493286132812,
      "learning_rate": 3.1402803167151766e-05,
      "loss": 90.5319,
      "step": 94050
    },
    {
      "epoch": 0.3800143020479401,
      "grad_norm": 674.5718383789062,
      "learning_rate": 3.140050886275496e-05,
      "loss": 56.2098,
      "step": 94060
    },
    {
      "epoch": 0.38005470331330776,
      "grad_norm": 797.9072875976562,
      "learning_rate": 3.1398214336099345e-05,
      "loss": 73.0468,
      "step": 94070
    },
    {
      "epoch": 0.3800951045786754,
      "grad_norm": 523.2122192382812,
      "learning_rate": 3.139591958722969e-05,
      "loss": 72.2402,
      "step": 94080
    },
    {
      "epoch": 0.38013550584404304,
      "grad_norm": 98.30523681640625,
      "learning_rate": 3.139362461619071e-05,
      "loss": 31.6787,
      "step": 94090
    },
    {
      "epoch": 0.3801759071094107,
      "grad_norm": 1349.02978515625,
      "learning_rate": 3.139132942302715e-05,
      "loss": 57.2611,
      "step": 94100
    },
    {
      "epoch": 0.3802163083747783,
      "grad_norm": 1889.5108642578125,
      "learning_rate": 3.138903400778376e-05,
      "loss": 76.0786,
      "step": 94110
    },
    {
      "epoch": 0.3802567096401459,
      "grad_norm": 455.8885498046875,
      "learning_rate": 3.1386738370505296e-05,
      "loss": 45.5989,
      "step": 94120
    },
    {
      "epoch": 0.38029711090551355,
      "grad_norm": 864.2047729492188,
      "learning_rate": 3.138444251123649e-05,
      "loss": 102.4932,
      "step": 94130
    },
    {
      "epoch": 0.3803375121708812,
      "grad_norm": 1400.811767578125,
      "learning_rate": 3.138214643002213e-05,
      "loss": 72.045,
      "step": 94140
    },
    {
      "epoch": 0.3803779134362488,
      "grad_norm": 869.3282470703125,
      "learning_rate": 3.1379850126906966e-05,
      "loss": 46.5034,
      "step": 94150
    },
    {
      "epoch": 0.38041831470161647,
      "grad_norm": 1013.1737670898438,
      "learning_rate": 3.137755360193575e-05,
      "loss": 54.6387,
      "step": 94160
    },
    {
      "epoch": 0.3804587159669841,
      "grad_norm": 841.7120361328125,
      "learning_rate": 3.137525685515329e-05,
      "loss": 46.6986,
      "step": 94170
    },
    {
      "epoch": 0.3804991172323517,
      "grad_norm": 1022.562744140625,
      "learning_rate": 3.137295988660432e-05,
      "loss": 70.4712,
      "step": 94180
    },
    {
      "epoch": 0.38053951849771933,
      "grad_norm": 719.1082153320312,
      "learning_rate": 3.1370662696333653e-05,
      "loss": 87.8841,
      "step": 94190
    },
    {
      "epoch": 0.38057991976308697,
      "grad_norm": 499.92938232421875,
      "learning_rate": 3.136836528438606e-05,
      "loss": 48.2856,
      "step": 94200
    },
    {
      "epoch": 0.3806203210284546,
      "grad_norm": 791.6148681640625,
      "learning_rate": 3.136606765080634e-05,
      "loss": 74.0039,
      "step": 94210
    },
    {
      "epoch": 0.38066072229382225,
      "grad_norm": 452.509521484375,
      "learning_rate": 3.136376979563927e-05,
      "loss": 66.7968,
      "step": 94220
    },
    {
      "epoch": 0.3807011235591899,
      "grad_norm": 4187.9775390625,
      "learning_rate": 3.136147171892966e-05,
      "loss": 66.3752,
      "step": 94230
    },
    {
      "epoch": 0.38074152482455753,
      "grad_norm": 461.0445251464844,
      "learning_rate": 3.135917342072231e-05,
      "loss": 65.7219,
      "step": 94240
    },
    {
      "epoch": 0.3807819260899251,
      "grad_norm": 767.9695434570312,
      "learning_rate": 3.135687490106203e-05,
      "loss": 105.8939,
      "step": 94250
    },
    {
      "epoch": 0.38082232735529276,
      "grad_norm": 651.4767456054688,
      "learning_rate": 3.135457615999362e-05,
      "loss": 42.9474,
      "step": 94260
    },
    {
      "epoch": 0.3808627286206604,
      "grad_norm": 379.09381103515625,
      "learning_rate": 3.13522771975619e-05,
      "loss": 54.3321,
      "step": 94270
    },
    {
      "epoch": 0.38090312988602804,
      "grad_norm": 984.194580078125,
      "learning_rate": 3.13499780138117e-05,
      "loss": 53.293,
      "step": 94280
    },
    {
      "epoch": 0.3809435311513957,
      "grad_norm": 619.849609375,
      "learning_rate": 3.134767860878784e-05,
      "loss": 65.771,
      "step": 94290
    },
    {
      "epoch": 0.3809839324167633,
      "grad_norm": 468.43536376953125,
      "learning_rate": 3.134537898253514e-05,
      "loss": 78.9491,
      "step": 94300
    },
    {
      "epoch": 0.3810243336821309,
      "grad_norm": 941.4468994140625,
      "learning_rate": 3.1343079135098425e-05,
      "loss": 51.5337,
      "step": 94310
    },
    {
      "epoch": 0.38106473494749854,
      "grad_norm": 1798.6435546875,
      "learning_rate": 3.134077906652255e-05,
      "loss": 56.2025,
      "step": 94320
    },
    {
      "epoch": 0.3811051362128662,
      "grad_norm": 1016.700927734375,
      "learning_rate": 3.133847877685235e-05,
      "loss": 48.8304,
      "step": 94330
    },
    {
      "epoch": 0.3811455374782338,
      "grad_norm": 680.2740478515625,
      "learning_rate": 3.133617826613266e-05,
      "loss": 73.8818,
      "step": 94340
    },
    {
      "epoch": 0.38118593874360146,
      "grad_norm": 628.7955932617188,
      "learning_rate": 3.1333877534408345e-05,
      "loss": 102.1594,
      "step": 94350
    },
    {
      "epoch": 0.3812263400089691,
      "grad_norm": 829.4243774414062,
      "learning_rate": 3.133157658172425e-05,
      "loss": 78.945,
      "step": 94360
    },
    {
      "epoch": 0.3812667412743367,
      "grad_norm": 1169.0318603515625,
      "learning_rate": 3.1329275408125245e-05,
      "loss": 73.4227,
      "step": 94370
    },
    {
      "epoch": 0.3813071425397043,
      "grad_norm": 984.260498046875,
      "learning_rate": 3.132697401365617e-05,
      "loss": 59.6905,
      "step": 94380
    },
    {
      "epoch": 0.38134754380507196,
      "grad_norm": 373.3670654296875,
      "learning_rate": 3.132467239836191e-05,
      "loss": 52.9543,
      "step": 94390
    },
    {
      "epoch": 0.3813879450704396,
      "grad_norm": 1990.4619140625,
      "learning_rate": 3.132237056228733e-05,
      "loss": 56.968,
      "step": 94400
    },
    {
      "epoch": 0.38142834633580724,
      "grad_norm": 1599.738525390625,
      "learning_rate": 3.132006850547732e-05,
      "loss": 55.0064,
      "step": 94410
    },
    {
      "epoch": 0.3814687476011749,
      "grad_norm": 566.6906127929688,
      "learning_rate": 3.131776622797673e-05,
      "loss": 73.1165,
      "step": 94420
    },
    {
      "epoch": 0.3815091488665425,
      "grad_norm": 1275.9754638671875,
      "learning_rate": 3.131546372983047e-05,
      "loss": 73.1124,
      "step": 94430
    },
    {
      "epoch": 0.3815495501319101,
      "grad_norm": 368.97845458984375,
      "learning_rate": 3.131316101108341e-05,
      "loss": 66.7757,
      "step": 94440
    },
    {
      "epoch": 0.38158995139727775,
      "grad_norm": 675.3721923828125,
      "learning_rate": 3.131085807178046e-05,
      "loss": 52.4319,
      "step": 94450
    },
    {
      "epoch": 0.3816303526626454,
      "grad_norm": 3169.10498046875,
      "learning_rate": 3.13085549119665e-05,
      "loss": 49.8318,
      "step": 94460
    },
    {
      "epoch": 0.38167075392801303,
      "grad_norm": 1114.07763671875,
      "learning_rate": 3.130625153168645e-05,
      "loss": 76.9605,
      "step": 94470
    },
    {
      "epoch": 0.38171115519338067,
      "grad_norm": 793.2512817382812,
      "learning_rate": 3.13039479309852e-05,
      "loss": 61.0932,
      "step": 94480
    },
    {
      "epoch": 0.3817515564587483,
      "grad_norm": 1011.723388671875,
      "learning_rate": 3.130164410990767e-05,
      "loss": 59.3565,
      "step": 94490
    },
    {
      "epoch": 0.3817919577241159,
      "grad_norm": 943.5010375976562,
      "learning_rate": 3.129934006849876e-05,
      "loss": 69.2316,
      "step": 94500
    },
    {
      "epoch": 0.38183235898948353,
      "grad_norm": 555.880126953125,
      "learning_rate": 3.1297035806803407e-05,
      "loss": 76.105,
      "step": 94510
    },
    {
      "epoch": 0.3818727602548512,
      "grad_norm": 1616.5235595703125,
      "learning_rate": 3.129473132486652e-05,
      "loss": 84.9985,
      "step": 94520
    },
    {
      "epoch": 0.3819131615202188,
      "grad_norm": 1294.511962890625,
      "learning_rate": 3.1292426622733036e-05,
      "loss": 43.3877,
      "step": 94530
    },
    {
      "epoch": 0.38195356278558645,
      "grad_norm": 636.7005004882812,
      "learning_rate": 3.1290121700447876e-05,
      "loss": 30.9998,
      "step": 94540
    },
    {
      "epoch": 0.3819939640509541,
      "grad_norm": 1390.4415283203125,
      "learning_rate": 3.128781655805599e-05,
      "loss": 59.5266,
      "step": 94550
    },
    {
      "epoch": 0.38203436531632173,
      "grad_norm": 924.8335571289062,
      "learning_rate": 3.12855111956023e-05,
      "loss": 35.2753,
      "step": 94560
    },
    {
      "epoch": 0.3820747665816893,
      "grad_norm": 3251.146484375,
      "learning_rate": 3.128320561313176e-05,
      "loss": 87.2722,
      "step": 94570
    },
    {
      "epoch": 0.38211516784705696,
      "grad_norm": 307.5578308105469,
      "learning_rate": 3.1280899810689316e-05,
      "loss": 79.3565,
      "step": 94580
    },
    {
      "epoch": 0.3821555691124246,
      "grad_norm": 838.0941772460938,
      "learning_rate": 3.127859378831992e-05,
      "loss": 77.5458,
      "step": 94590
    },
    {
      "epoch": 0.38219597037779224,
      "grad_norm": 760.473388671875,
      "learning_rate": 3.1276287546068536e-05,
      "loss": 53.5369,
      "step": 94600
    },
    {
      "epoch": 0.3822363716431599,
      "grad_norm": 761.6842041015625,
      "learning_rate": 3.1273981083980126e-05,
      "loss": 77.6941,
      "step": 94610
    },
    {
      "epoch": 0.3822767729085275,
      "grad_norm": 323.1531677246094,
      "learning_rate": 3.127167440209964e-05,
      "loss": 62.5255,
      "step": 94620
    },
    {
      "epoch": 0.3823171741738951,
      "grad_norm": 1118.4254150390625,
      "learning_rate": 3.1269367500472065e-05,
      "loss": 63.3706,
      "step": 94630
    },
    {
      "epoch": 0.38235757543926274,
      "grad_norm": 1080.7705078125,
      "learning_rate": 3.126706037914237e-05,
      "loss": 37.4864,
      "step": 94640
    },
    {
      "epoch": 0.3823979767046304,
      "grad_norm": 670.8095092773438,
      "learning_rate": 3.126475303815554e-05,
      "loss": 53.1566,
      "step": 94650
    },
    {
      "epoch": 0.382438377969998,
      "grad_norm": 1319.5672607421875,
      "learning_rate": 3.1262445477556536e-05,
      "loss": 58.251,
      "step": 94660
    },
    {
      "epoch": 0.38247877923536566,
      "grad_norm": 454.59344482421875,
      "learning_rate": 3.126013769739036e-05,
      "loss": 48.4365,
      "step": 94670
    },
    {
      "epoch": 0.3825191805007333,
      "grad_norm": 1183.15869140625,
      "learning_rate": 3.1257829697702004e-05,
      "loss": 86.8193,
      "step": 94680
    },
    {
      "epoch": 0.3825595817661009,
      "grad_norm": 570.905517578125,
      "learning_rate": 3.125552147853646e-05,
      "loss": 62.5943,
      "step": 94690
    },
    {
      "epoch": 0.3825999830314685,
      "grad_norm": 379.4037780761719,
      "learning_rate": 3.125321303993874e-05,
      "loss": 55.0205,
      "step": 94700
    },
    {
      "epoch": 0.38264038429683617,
      "grad_norm": 1375.26953125,
      "learning_rate": 3.125090438195383e-05,
      "loss": 160.7165,
      "step": 94710
    },
    {
      "epoch": 0.3826807855622038,
      "grad_norm": 587.9031372070312,
      "learning_rate": 3.124859550462675e-05,
      "loss": 43.147,
      "step": 94720
    },
    {
      "epoch": 0.38272118682757145,
      "grad_norm": 3379.8837890625,
      "learning_rate": 3.124628640800251e-05,
      "loss": 110.5884,
      "step": 94730
    },
    {
      "epoch": 0.3827615880929391,
      "grad_norm": 891.32373046875,
      "learning_rate": 3.124397709212613e-05,
      "loss": 52.807,
      "step": 94740
    },
    {
      "epoch": 0.3828019893583067,
      "grad_norm": 1796.2423095703125,
      "learning_rate": 3.124166755704261e-05,
      "loss": 59.8244,
      "step": 94750
    },
    {
      "epoch": 0.3828423906236743,
      "grad_norm": 629.0004272460938,
      "learning_rate": 3.123935780279701e-05,
      "loss": 59.133,
      "step": 94760
    },
    {
      "epoch": 0.38288279188904195,
      "grad_norm": 718.2772216796875,
      "learning_rate": 3.123704782943434e-05,
      "loss": 53.9646,
      "step": 94770
    },
    {
      "epoch": 0.3829231931544096,
      "grad_norm": 980.1384887695312,
      "learning_rate": 3.123473763699964e-05,
      "loss": 65.5862,
      "step": 94780
    },
    {
      "epoch": 0.38296359441977723,
      "grad_norm": 986.5202026367188,
      "learning_rate": 3.1232427225537935e-05,
      "loss": 56.9099,
      "step": 94790
    },
    {
      "epoch": 0.38300399568514487,
      "grad_norm": 653.0430908203125,
      "learning_rate": 3.123011659509429e-05,
      "loss": 78.1085,
      "step": 94800
    },
    {
      "epoch": 0.3830443969505125,
      "grad_norm": 670.354736328125,
      "learning_rate": 3.122780574571373e-05,
      "loss": 60.5578,
      "step": 94810
    },
    {
      "epoch": 0.3830847982158801,
      "grad_norm": 432.60552978515625,
      "learning_rate": 3.122549467744133e-05,
      "loss": 88.2522,
      "step": 94820
    },
    {
      "epoch": 0.38312519948124774,
      "grad_norm": 587.0864868164062,
      "learning_rate": 3.122318339032212e-05,
      "loss": 63.6049,
      "step": 94830
    },
    {
      "epoch": 0.3831656007466154,
      "grad_norm": 1073.345947265625,
      "learning_rate": 3.122087188440118e-05,
      "loss": 62.4557,
      "step": 94840
    },
    {
      "epoch": 0.383206002011983,
      "grad_norm": 681.5462036132812,
      "learning_rate": 3.121856015972356e-05,
      "loss": 49.4943,
      "step": 94850
    },
    {
      "epoch": 0.38324640327735066,
      "grad_norm": 467.7748718261719,
      "learning_rate": 3.121624821633434e-05,
      "loss": 46.3031,
      "step": 94860
    },
    {
      "epoch": 0.3832868045427183,
      "grad_norm": 809.7584838867188,
      "learning_rate": 3.121393605427858e-05,
      "loss": 58.0714,
      "step": 94870
    },
    {
      "epoch": 0.38332720580808594,
      "grad_norm": 171.03561401367188,
      "learning_rate": 3.1211623673601366e-05,
      "loss": 76.0863,
      "step": 94880
    },
    {
      "epoch": 0.3833676070734535,
      "grad_norm": 392.6013488769531,
      "learning_rate": 3.120931107434778e-05,
      "loss": 45.0487,
      "step": 94890
    },
    {
      "epoch": 0.38340800833882116,
      "grad_norm": 462.5611572265625,
      "learning_rate": 3.1206998256562894e-05,
      "loss": 68.4658,
      "step": 94900
    },
    {
      "epoch": 0.3834484096041888,
      "grad_norm": 521.8003540039062,
      "learning_rate": 3.120468522029182e-05,
      "loss": 92.6952,
      "step": 94910
    },
    {
      "epoch": 0.38348881086955644,
      "grad_norm": 831.3837280273438,
      "learning_rate": 3.1202371965579625e-05,
      "loss": 71.1433,
      "step": 94920
    },
    {
      "epoch": 0.3835292121349241,
      "grad_norm": 0.0,
      "learning_rate": 3.120005849247143e-05,
      "loss": 69.1638,
      "step": 94930
    },
    {
      "epoch": 0.3835696134002917,
      "grad_norm": 501.0518493652344,
      "learning_rate": 3.119774480101233e-05,
      "loss": 52.1852,
      "step": 94940
    },
    {
      "epoch": 0.3836100146656593,
      "grad_norm": 565.007080078125,
      "learning_rate": 3.119543089124743e-05,
      "loss": 82.6033,
      "step": 94950
    },
    {
      "epoch": 0.38365041593102694,
      "grad_norm": 489.13934326171875,
      "learning_rate": 3.119311676322183e-05,
      "loss": 41.9466,
      "step": 94960
    },
    {
      "epoch": 0.3836908171963946,
      "grad_norm": 1597.20947265625,
      "learning_rate": 3.1190802416980664e-05,
      "loss": 57.1254,
      "step": 94970
    },
    {
      "epoch": 0.3837312184617622,
      "grad_norm": 492.63336181640625,
      "learning_rate": 3.118848785256905e-05,
      "loss": 55.7734,
      "step": 94980
    },
    {
      "epoch": 0.38377161972712986,
      "grad_norm": 315.7362365722656,
      "learning_rate": 3.118617307003209e-05,
      "loss": 60.9176,
      "step": 94990
    },
    {
      "epoch": 0.3838120209924975,
      "grad_norm": 777.3889770507812,
      "learning_rate": 3.1183858069414936e-05,
      "loss": 60.9691,
      "step": 95000
    },
    {
      "epoch": 0.3838524222578651,
      "grad_norm": 945.9274291992188,
      "learning_rate": 3.118154285076271e-05,
      "loss": 44.9719,
      "step": 95010
    },
    {
      "epoch": 0.38389282352323273,
      "grad_norm": 969.2196655273438,
      "learning_rate": 3.1179227414120556e-05,
      "loss": 103.0269,
      "step": 95020
    },
    {
      "epoch": 0.38393322478860037,
      "grad_norm": 1758.84033203125,
      "learning_rate": 3.11769117595336e-05,
      "loss": 94.0152,
      "step": 95030
    },
    {
      "epoch": 0.383973626053968,
      "grad_norm": 750.189697265625,
      "learning_rate": 3.1174595887047e-05,
      "loss": 59.8756,
      "step": 95040
    },
    {
      "epoch": 0.38401402731933565,
      "grad_norm": 743.2469482421875,
      "learning_rate": 3.11722797967059e-05,
      "loss": 72.4545,
      "step": 95050
    },
    {
      "epoch": 0.3840544285847033,
      "grad_norm": 746.4964599609375,
      "learning_rate": 3.1169963488555445e-05,
      "loss": 85.1586,
      "step": 95060
    },
    {
      "epoch": 0.38409482985007093,
      "grad_norm": 662.7686767578125,
      "learning_rate": 3.116764696264081e-05,
      "loss": 61.581,
      "step": 95070
    },
    {
      "epoch": 0.3841352311154385,
      "grad_norm": 280.2574157714844,
      "learning_rate": 3.116533021900714e-05,
      "loss": 55.689,
      "step": 95080
    },
    {
      "epoch": 0.38417563238080615,
      "grad_norm": 0.0,
      "learning_rate": 3.116301325769961e-05,
      "loss": 46.6048,
      "step": 95090
    },
    {
      "epoch": 0.3842160336461738,
      "grad_norm": 691.7823486328125,
      "learning_rate": 3.11606960787634e-05,
      "loss": 55.2978,
      "step": 95100
    },
    {
      "epoch": 0.38425643491154143,
      "grad_norm": 492.3890686035156,
      "learning_rate": 3.1158378682243666e-05,
      "loss": 73.2431,
      "step": 95110
    },
    {
      "epoch": 0.3842968361769091,
      "grad_norm": 622.8359375,
      "learning_rate": 3.1156061068185595e-05,
      "loss": 54.8487,
      "step": 95120
    },
    {
      "epoch": 0.3843372374422767,
      "grad_norm": 857.6041259765625,
      "learning_rate": 3.115374323663437e-05,
      "loss": 79.2417,
      "step": 95130
    },
    {
      "epoch": 0.3843776387076443,
      "grad_norm": 616.6744384765625,
      "learning_rate": 3.115142518763518e-05,
      "loss": 79.7958,
      "step": 95140
    },
    {
      "epoch": 0.38441803997301194,
      "grad_norm": 1210.3330078125,
      "learning_rate": 3.1149106921233216e-05,
      "loss": 125.1222,
      "step": 95150
    },
    {
      "epoch": 0.3844584412383796,
      "grad_norm": 768.7312622070312,
      "learning_rate": 3.114678843747367e-05,
      "loss": 59.6823,
      "step": 95160
    },
    {
      "epoch": 0.3844988425037472,
      "grad_norm": 537.5147705078125,
      "learning_rate": 3.114446973640175e-05,
      "loss": 41.4512,
      "step": 95170
    },
    {
      "epoch": 0.38453924376911486,
      "grad_norm": 1011.1067504882812,
      "learning_rate": 3.114215081806265e-05,
      "loss": 48.0649,
      "step": 95180
    },
    {
      "epoch": 0.3845796450344825,
      "grad_norm": 247.35574340820312,
      "learning_rate": 3.113983168250158e-05,
      "loss": 101.9663,
      "step": 95190
    },
    {
      "epoch": 0.38462004629985014,
      "grad_norm": 588.1903686523438,
      "learning_rate": 3.113751232976376e-05,
      "loss": 62.9706,
      "step": 95200
    },
    {
      "epoch": 0.3846604475652177,
      "grad_norm": 567.7507934570312,
      "learning_rate": 3.1135192759894403e-05,
      "loss": 70.8379,
      "step": 95210
    },
    {
      "epoch": 0.38470084883058536,
      "grad_norm": 835.3895263671875,
      "learning_rate": 3.1132872972938735e-05,
      "loss": 29.8401,
      "step": 95220
    },
    {
      "epoch": 0.384741250095953,
      "grad_norm": 347.2232971191406,
      "learning_rate": 3.113055296894197e-05,
      "loss": 38.3244,
      "step": 95230
    },
    {
      "epoch": 0.38478165136132064,
      "grad_norm": 1113.609619140625,
      "learning_rate": 3.112823274794935e-05,
      "loss": 58.9926,
      "step": 95240
    },
    {
      "epoch": 0.3848220526266883,
      "grad_norm": 791.2014770507812,
      "learning_rate": 3.11259123100061e-05,
      "loss": 63.5421,
      "step": 95250
    },
    {
      "epoch": 0.3848624538920559,
      "grad_norm": 805.3345336914062,
      "learning_rate": 3.112359165515746e-05,
      "loss": 39.3376,
      "step": 95260
    },
    {
      "epoch": 0.3849028551574235,
      "grad_norm": 724.2696533203125,
      "learning_rate": 3.112127078344868e-05,
      "loss": 110.7945,
      "step": 95270
    },
    {
      "epoch": 0.38494325642279115,
      "grad_norm": 491.9793701171875,
      "learning_rate": 3.111894969492499e-05,
      "loss": 42.779,
      "step": 95280
    },
    {
      "epoch": 0.3849836576881588,
      "grad_norm": 685.3145141601562,
      "learning_rate": 3.1116628389631655e-05,
      "loss": 85.3884,
      "step": 95290
    },
    {
      "epoch": 0.3850240589535264,
      "grad_norm": 534.1943969726562,
      "learning_rate": 3.1114306867613935e-05,
      "loss": 52.4637,
      "step": 95300
    },
    {
      "epoch": 0.38506446021889407,
      "grad_norm": 832.2371215820312,
      "learning_rate": 3.111198512891707e-05,
      "loss": 51.0535,
      "step": 95310
    },
    {
      "epoch": 0.3851048614842617,
      "grad_norm": 2605.079345703125,
      "learning_rate": 3.110966317358634e-05,
      "loss": 75.6956,
      "step": 95320
    },
    {
      "epoch": 0.3851452627496293,
      "grad_norm": 1315.3677978515625,
      "learning_rate": 3.1107341001667007e-05,
      "loss": 67.2918,
      "step": 95330
    },
    {
      "epoch": 0.38518566401499693,
      "grad_norm": 776.8063354492188,
      "learning_rate": 3.110501861320434e-05,
      "loss": 52.1419,
      "step": 95340
    },
    {
      "epoch": 0.38522606528036457,
      "grad_norm": 677.2888793945312,
      "learning_rate": 3.110269600824362e-05,
      "loss": 67.7612,
      "step": 95350
    },
    {
      "epoch": 0.3852664665457322,
      "grad_norm": 1661.3873291015625,
      "learning_rate": 3.110037318683012e-05,
      "loss": 43.2306,
      "step": 95360
    },
    {
      "epoch": 0.38530686781109985,
      "grad_norm": 529.8228149414062,
      "learning_rate": 3.109805014900914e-05,
      "loss": 72.6942,
      "step": 95370
    },
    {
      "epoch": 0.3853472690764675,
      "grad_norm": 951.1519165039062,
      "learning_rate": 3.109572689482595e-05,
      "loss": 48.7285,
      "step": 95380
    },
    {
      "epoch": 0.38538767034183513,
      "grad_norm": 720.6952514648438,
      "learning_rate": 3.1093403424325855e-05,
      "loss": 60.9023,
      "step": 95390
    },
    {
      "epoch": 0.3854280716072027,
      "grad_norm": 672.1053466796875,
      "learning_rate": 3.109107973755415e-05,
      "loss": 48.6235,
      "step": 95400
    },
    {
      "epoch": 0.38546847287257036,
      "grad_norm": 1297.8304443359375,
      "learning_rate": 3.1088755834556135e-05,
      "loss": 60.5914,
      "step": 95410
    },
    {
      "epoch": 0.385508874137938,
      "grad_norm": 1021.5657348632812,
      "learning_rate": 3.108643171537711e-05,
      "loss": 63.9048,
      "step": 95420
    },
    {
      "epoch": 0.38554927540330564,
      "grad_norm": 987.0369262695312,
      "learning_rate": 3.10841073800624e-05,
      "loss": 55.4489,
      "step": 95430
    },
    {
      "epoch": 0.3855896766686733,
      "grad_norm": 723.7199096679688,
      "learning_rate": 3.10817828286573e-05,
      "loss": 69.069,
      "step": 95440
    },
    {
      "epoch": 0.3856300779340409,
      "grad_norm": 919.791015625,
      "learning_rate": 3.107945806120715e-05,
      "loss": 55.2413,
      "step": 95450
    },
    {
      "epoch": 0.3856704791994085,
      "grad_norm": 731.2409057617188,
      "learning_rate": 3.1077133077757257e-05,
      "loss": 77.2168,
      "step": 95460
    },
    {
      "epoch": 0.38571088046477614,
      "grad_norm": 551.4536743164062,
      "learning_rate": 3.107480787835295e-05,
      "loss": 69.0535,
      "step": 95470
    },
    {
      "epoch": 0.3857512817301438,
      "grad_norm": 505.9342041015625,
      "learning_rate": 3.107248246303956e-05,
      "loss": 51.6537,
      "step": 95480
    },
    {
      "epoch": 0.3857916829955114,
      "grad_norm": 969.6237182617188,
      "learning_rate": 3.1070156831862424e-05,
      "loss": 51.4906,
      "step": 95490
    },
    {
      "epoch": 0.38583208426087906,
      "grad_norm": 1107.892822265625,
      "learning_rate": 3.106783098486688e-05,
      "loss": 45.553,
      "step": 95500
    },
    {
      "epoch": 0.3858724855262467,
      "grad_norm": 786.083251953125,
      "learning_rate": 3.106550492209828e-05,
      "loss": 53.2214,
      "step": 95510
    },
    {
      "epoch": 0.38591288679161434,
      "grad_norm": 904.0626831054688,
      "learning_rate": 3.1063178643601957e-05,
      "loss": 66.1209,
      "step": 95520
    },
    {
      "epoch": 0.3859532880569819,
      "grad_norm": 1052.266845703125,
      "learning_rate": 3.106085214942328e-05,
      "loss": 54.094,
      "step": 95530
    },
    {
      "epoch": 0.38599368932234956,
      "grad_norm": 478.57110595703125,
      "learning_rate": 3.105852543960759e-05,
      "loss": 66.3925,
      "step": 95540
    },
    {
      "epoch": 0.3860340905877172,
      "grad_norm": 403.1227722167969,
      "learning_rate": 3.105619851420025e-05,
      "loss": 81.0497,
      "step": 95550
    },
    {
      "epoch": 0.38607449185308484,
      "grad_norm": 477.69342041015625,
      "learning_rate": 3.105387137324663e-05,
      "loss": 85.4535,
      "step": 95560
    },
    {
      "epoch": 0.3861148931184525,
      "grad_norm": 773.2382202148438,
      "learning_rate": 3.1051544016792096e-05,
      "loss": 60.2238,
      "step": 95570
    },
    {
      "epoch": 0.3861552943838201,
      "grad_norm": 445.6334533691406,
      "learning_rate": 3.1049216444882024e-05,
      "loss": 53.3764,
      "step": 95580
    },
    {
      "epoch": 0.3861956956491877,
      "grad_norm": 531.4219360351562,
      "learning_rate": 3.104688865756179e-05,
      "loss": 73.8038,
      "step": 95590
    },
    {
      "epoch": 0.38623609691455535,
      "grad_norm": 833.8771362304688,
      "learning_rate": 3.1044560654876775e-05,
      "loss": 80.2652,
      "step": 95600
    },
    {
      "epoch": 0.386276498179923,
      "grad_norm": 525.5472412109375,
      "learning_rate": 3.104223243687236e-05,
      "loss": 46.4954,
      "step": 95610
    },
    {
      "epoch": 0.38631689944529063,
      "grad_norm": 1135.0772705078125,
      "learning_rate": 3.1039904003593946e-05,
      "loss": 68.1692,
      "step": 95620
    },
    {
      "epoch": 0.38635730071065827,
      "grad_norm": 527.236572265625,
      "learning_rate": 3.103757535508691e-05,
      "loss": 47.8166,
      "step": 95630
    },
    {
      "epoch": 0.3863977019760259,
      "grad_norm": 1182.5828857421875,
      "learning_rate": 3.103524649139667e-05,
      "loss": 66.4084,
      "step": 95640
    },
    {
      "epoch": 0.3864381032413935,
      "grad_norm": 556.0507202148438,
      "learning_rate": 3.103291741256861e-05,
      "loss": 61.376,
      "step": 95650
    },
    {
      "epoch": 0.38647850450676113,
      "grad_norm": 0.0,
      "learning_rate": 3.103058811864815e-05,
      "loss": 57.1049,
      "step": 95660
    },
    {
      "epoch": 0.3865189057721288,
      "grad_norm": 1166.5506591796875,
      "learning_rate": 3.102825860968069e-05,
      "loss": 71.8964,
      "step": 95670
    },
    {
      "epoch": 0.3865593070374964,
      "grad_norm": 438.2193603515625,
      "learning_rate": 3.102592888571166e-05,
      "loss": 311.3151,
      "step": 95680
    },
    {
      "epoch": 0.38659970830286405,
      "grad_norm": 1997.8970947265625,
      "learning_rate": 3.102359894678645e-05,
      "loss": 74.6881,
      "step": 95690
    },
    {
      "epoch": 0.3866401095682317,
      "grad_norm": 487.5499267578125,
      "learning_rate": 3.1021268792950524e-05,
      "loss": 74.8267,
      "step": 95700
    },
    {
      "epoch": 0.38668051083359933,
      "grad_norm": 360.4342346191406,
      "learning_rate": 3.101893842424928e-05,
      "loss": 69.5386,
      "step": 95710
    },
    {
      "epoch": 0.3867209120989669,
      "grad_norm": 716.3482666015625,
      "learning_rate": 3.101660784072816e-05,
      "loss": 78.6839,
      "step": 95720
    },
    {
      "epoch": 0.38676131336433456,
      "grad_norm": 293.6958312988281,
      "learning_rate": 3.10142770424326e-05,
      "loss": 63.8406,
      "step": 95730
    },
    {
      "epoch": 0.3868017146297022,
      "grad_norm": 427.47882080078125,
      "learning_rate": 3.1011946029408035e-05,
      "loss": 71.2774,
      "step": 95740
    },
    {
      "epoch": 0.38684211589506984,
      "grad_norm": 781.7965087890625,
      "learning_rate": 3.1009614801699914e-05,
      "loss": 59.1214,
      "step": 95750
    },
    {
      "epoch": 0.3868825171604375,
      "grad_norm": 654.1803588867188,
      "learning_rate": 3.100728335935369e-05,
      "loss": 84.9144,
      "step": 95760
    },
    {
      "epoch": 0.3869229184258051,
      "grad_norm": 693.10595703125,
      "learning_rate": 3.10049517024148e-05,
      "loss": 56.505,
      "step": 95770
    },
    {
      "epoch": 0.3869633196911727,
      "grad_norm": 450.68280029296875,
      "learning_rate": 3.1002619830928715e-05,
      "loss": 40.1912,
      "step": 95780
    },
    {
      "epoch": 0.38700372095654034,
      "grad_norm": 664.6299438476562,
      "learning_rate": 3.100028774494089e-05,
      "loss": 60.4155,
      "step": 95790
    },
    {
      "epoch": 0.387044122221908,
      "grad_norm": 1185.4459228515625,
      "learning_rate": 3.09979554444968e-05,
      "loss": 45.9476,
      "step": 95800
    },
    {
      "epoch": 0.3870845234872756,
      "grad_norm": 973.5237426757812,
      "learning_rate": 3.09956229296419e-05,
      "loss": 75.4969,
      "step": 95810
    },
    {
      "epoch": 0.38712492475264326,
      "grad_norm": 921.3255004882812,
      "learning_rate": 3.0993290200421676e-05,
      "loss": 58.8776,
      "step": 95820
    },
    {
      "epoch": 0.3871653260180109,
      "grad_norm": 965.7318115234375,
      "learning_rate": 3.099095725688159e-05,
      "loss": 55.5938,
      "step": 95830
    },
    {
      "epoch": 0.38720572728337854,
      "grad_norm": 1331.021240234375,
      "learning_rate": 3.098862409906714e-05,
      "loss": 79.6023,
      "step": 95840
    },
    {
      "epoch": 0.3872461285487461,
      "grad_norm": 560.6642456054688,
      "learning_rate": 3.09862907270238e-05,
      "loss": 57.35,
      "step": 95850
    },
    {
      "epoch": 0.38728652981411377,
      "grad_norm": 604.1090087890625,
      "learning_rate": 3.098395714079708e-05,
      "loss": 70.6695,
      "step": 95860
    },
    {
      "epoch": 0.3873269310794814,
      "grad_norm": 310.1190185546875,
      "learning_rate": 3.0981623340432446e-05,
      "loss": 61.8486,
      "step": 95870
    },
    {
      "epoch": 0.38736733234484905,
      "grad_norm": 540.4014282226562,
      "learning_rate": 3.097928932597542e-05,
      "loss": 50.6746,
      "step": 95880
    },
    {
      "epoch": 0.3874077336102167,
      "grad_norm": 940.4336547851562,
      "learning_rate": 3.097695509747149e-05,
      "loss": 77.3138,
      "step": 95890
    },
    {
      "epoch": 0.3874481348755843,
      "grad_norm": 624.2680053710938,
      "learning_rate": 3.0974620654966175e-05,
      "loss": 63.9111,
      "step": 95900
    },
    {
      "epoch": 0.3874885361409519,
      "grad_norm": 513.7975463867188,
      "learning_rate": 3.097228599850498e-05,
      "loss": 71.1971,
      "step": 95910
    },
    {
      "epoch": 0.38752893740631955,
      "grad_norm": 398.7431335449219,
      "learning_rate": 3.096995112813341e-05,
      "loss": 63.1596,
      "step": 95920
    },
    {
      "epoch": 0.3875693386716872,
      "grad_norm": 432.8017272949219,
      "learning_rate": 3.0967616043897e-05,
      "loss": 46.5326,
      "step": 95930
    },
    {
      "epoch": 0.38760973993705483,
      "grad_norm": 445.6565856933594,
      "learning_rate": 3.096528074584128e-05,
      "loss": 58.431,
      "step": 95940
    },
    {
      "epoch": 0.38765014120242247,
      "grad_norm": 456.55230712890625,
      "learning_rate": 3.0962945234011755e-05,
      "loss": 59.9908,
      "step": 95950
    },
    {
      "epoch": 0.3876905424677901,
      "grad_norm": 472.28472900390625,
      "learning_rate": 3.096060950845397e-05,
      "loss": 48.5587,
      "step": 95960
    },
    {
      "epoch": 0.3877309437331577,
      "grad_norm": 1048.744873046875,
      "learning_rate": 3.0958273569213456e-05,
      "loss": 67.3828,
      "step": 95970
    },
    {
      "epoch": 0.38777134499852534,
      "grad_norm": 1082.98876953125,
      "learning_rate": 3.095593741633577e-05,
      "loss": 82.9235,
      "step": 95980
    },
    {
      "epoch": 0.387811746263893,
      "grad_norm": 530.9251708984375,
      "learning_rate": 3.095360104986643e-05,
      "loss": 64.6616,
      "step": 95990
    },
    {
      "epoch": 0.3878521475292606,
      "grad_norm": 1165.728759765625,
      "learning_rate": 3.095126446985101e-05,
      "loss": 79.6478,
      "step": 96000
    },
    {
      "epoch": 0.38789254879462826,
      "grad_norm": 1305.37158203125,
      "learning_rate": 3.0948927676335044e-05,
      "loss": 83.1801,
      "step": 96010
    },
    {
      "epoch": 0.3879329500599959,
      "grad_norm": 529.7338256835938,
      "learning_rate": 3.09465906693641e-05,
      "loss": 76.9614,
      "step": 96020
    },
    {
      "epoch": 0.38797335132536354,
      "grad_norm": 837.6686401367188,
      "learning_rate": 3.0944253448983725e-05,
      "loss": 91.7809,
      "step": 96030
    },
    {
      "epoch": 0.3880137525907311,
      "grad_norm": 457.90679931640625,
      "learning_rate": 3.0941916015239505e-05,
      "loss": 40.1575,
      "step": 96040
    },
    {
      "epoch": 0.38805415385609876,
      "grad_norm": 667.5697021484375,
      "learning_rate": 3.0939578368176997e-05,
      "loss": 51.0978,
      "step": 96050
    },
    {
      "epoch": 0.3880945551214664,
      "grad_norm": 684.8994750976562,
      "learning_rate": 3.093724050784178e-05,
      "loss": 61.2667,
      "step": 96060
    },
    {
      "epoch": 0.38813495638683404,
      "grad_norm": 405.88916015625,
      "learning_rate": 3.093490243427942e-05,
      "loss": 49.8711,
      "step": 96070
    },
    {
      "epoch": 0.3881753576522017,
      "grad_norm": 485.23602294921875,
      "learning_rate": 3.093256414753551e-05,
      "loss": 40.8211,
      "step": 96080
    },
    {
      "epoch": 0.3882157589175693,
      "grad_norm": 1215.4119873046875,
      "learning_rate": 3.093022564765564e-05,
      "loss": 81.3737,
      "step": 96090
    },
    {
      "epoch": 0.3882561601829369,
      "grad_norm": 568.2235717773438,
      "learning_rate": 3.092788693468539e-05,
      "loss": 55.6435,
      "step": 96100
    },
    {
      "epoch": 0.38829656144830454,
      "grad_norm": 774.2294311523438,
      "learning_rate": 3.092554800867035e-05,
      "loss": 42.8646,
      "step": 96110
    },
    {
      "epoch": 0.3883369627136722,
      "grad_norm": 765.2290649414062,
      "learning_rate": 3.092320886965614e-05,
      "loss": 67.741,
      "step": 96120
    },
    {
      "epoch": 0.3883773639790398,
      "grad_norm": 740.162841796875,
      "learning_rate": 3.0920869517688336e-05,
      "loss": 100.8398,
      "step": 96130
    },
    {
      "epoch": 0.38841776524440746,
      "grad_norm": 1602.5484619140625,
      "learning_rate": 3.091852995281257e-05,
      "loss": 82.2206,
      "step": 96140
    },
    {
      "epoch": 0.3884581665097751,
      "grad_norm": 579.366943359375,
      "learning_rate": 3.091619017507443e-05,
      "loss": 56.4242,
      "step": 96150
    },
    {
      "epoch": 0.38849856777514274,
      "grad_norm": 873.070556640625,
      "learning_rate": 3.091385018451955e-05,
      "loss": 84.0062,
      "step": 96160
    },
    {
      "epoch": 0.38853896904051033,
      "grad_norm": 472.2259521484375,
      "learning_rate": 3.091150998119354e-05,
      "loss": 66.6976,
      "step": 96170
    },
    {
      "epoch": 0.38857937030587797,
      "grad_norm": 2321.2109375,
      "learning_rate": 3.090916956514203e-05,
      "loss": 95.4598,
      "step": 96180
    },
    {
      "epoch": 0.3886197715712456,
      "grad_norm": 929.7415161132812,
      "learning_rate": 3.090682893641064e-05,
      "loss": 70.1647,
      "step": 96190
    },
    {
      "epoch": 0.38866017283661325,
      "grad_norm": 642.71630859375,
      "learning_rate": 3.090448809504501e-05,
      "loss": 56.4267,
      "step": 96200
    },
    {
      "epoch": 0.3887005741019809,
      "grad_norm": 531.6314697265625,
      "learning_rate": 3.090214704109076e-05,
      "loss": 41.6155,
      "step": 96210
    },
    {
      "epoch": 0.38874097536734853,
      "grad_norm": 1281.093505859375,
      "learning_rate": 3.089980577459354e-05,
      "loss": 76.9379,
      "step": 96220
    },
    {
      "epoch": 0.3887813766327161,
      "grad_norm": 184.43612670898438,
      "learning_rate": 3.0897464295599e-05,
      "loss": 61.2117,
      "step": 96230
    },
    {
      "epoch": 0.38882177789808375,
      "grad_norm": 752.0773315429688,
      "learning_rate": 3.0895122604152784e-05,
      "loss": 69.0763,
      "step": 96240
    },
    {
      "epoch": 0.3888621791634514,
      "grad_norm": 1775.0201416015625,
      "learning_rate": 3.0892780700300544e-05,
      "loss": 57.2486,
      "step": 96250
    },
    {
      "epoch": 0.38890258042881903,
      "grad_norm": 431.5108642578125,
      "learning_rate": 3.0890438584087936e-05,
      "loss": 42.2329,
      "step": 96260
    },
    {
      "epoch": 0.3889429816941867,
      "grad_norm": 1414.23046875,
      "learning_rate": 3.088809625556062e-05,
      "loss": 43.4289,
      "step": 96270
    },
    {
      "epoch": 0.3889833829595543,
      "grad_norm": 717.4227905273438,
      "learning_rate": 3.088575371476426e-05,
      "loss": 52.6565,
      "step": 96280
    },
    {
      "epoch": 0.3890237842249219,
      "grad_norm": 547.7136840820312,
      "learning_rate": 3.0883410961744536e-05,
      "loss": 43.894,
      "step": 96290
    },
    {
      "epoch": 0.38906418549028954,
      "grad_norm": 648.9277954101562,
      "learning_rate": 3.08810679965471e-05,
      "loss": 75.1155,
      "step": 96300
    },
    {
      "epoch": 0.3891045867556572,
      "grad_norm": 1267.9052734375,
      "learning_rate": 3.087872481921765e-05,
      "loss": 105.556,
      "step": 96310
    },
    {
      "epoch": 0.3891449880210248,
      "grad_norm": 1205.15234375,
      "learning_rate": 3.087638142980185e-05,
      "loss": 64.842,
      "step": 96320
    },
    {
      "epoch": 0.38918538928639246,
      "grad_norm": 394.9377746582031,
      "learning_rate": 3.0874037828345406e-05,
      "loss": 61.5966,
      "step": 96330
    },
    {
      "epoch": 0.3892257905517601,
      "grad_norm": 1072.0274658203125,
      "learning_rate": 3.0871694014893985e-05,
      "loss": 63.1424,
      "step": 96340
    },
    {
      "epoch": 0.38926619181712774,
      "grad_norm": 830.2392578125,
      "learning_rate": 3.08693499894933e-05,
      "loss": 47.6457,
      "step": 96350
    },
    {
      "epoch": 0.3893065930824953,
      "grad_norm": 832.5903930664062,
      "learning_rate": 3.0867005752189037e-05,
      "loss": 68.6419,
      "step": 96360
    },
    {
      "epoch": 0.38934699434786296,
      "grad_norm": 576.5301513671875,
      "learning_rate": 3.086466130302691e-05,
      "loss": 85.2657,
      "step": 96370
    },
    {
      "epoch": 0.3893873956132306,
      "grad_norm": 896.1205444335938,
      "learning_rate": 3.086231664205261e-05,
      "loss": 65.5966,
      "step": 96380
    },
    {
      "epoch": 0.38942779687859824,
      "grad_norm": 477.4507751464844,
      "learning_rate": 3.085997176931185e-05,
      "loss": 52.3738,
      "step": 96390
    },
    {
      "epoch": 0.3894681981439659,
      "grad_norm": 638.5934448242188,
      "learning_rate": 3.0857626684850355e-05,
      "loss": 88.827,
      "step": 96400
    },
    {
      "epoch": 0.3895085994093335,
      "grad_norm": 1253.0484619140625,
      "learning_rate": 3.085528138871384e-05,
      "loss": 44.4382,
      "step": 96410
    },
    {
      "epoch": 0.3895490006747011,
      "grad_norm": 620.13623046875,
      "learning_rate": 3.085293588094802e-05,
      "loss": 55.8784,
      "step": 96420
    },
    {
      "epoch": 0.38958940194006875,
      "grad_norm": 458.2687072753906,
      "learning_rate": 3.085059016159863e-05,
      "loss": 44.469,
      "step": 96430
    },
    {
      "epoch": 0.3896298032054364,
      "grad_norm": 1276.606689453125,
      "learning_rate": 3.08482442307114e-05,
      "loss": 50.4212,
      "step": 96440
    },
    {
      "epoch": 0.389670204470804,
      "grad_norm": 1068.017578125,
      "learning_rate": 3.084589808833206e-05,
      "loss": 58.046,
      "step": 96450
    },
    {
      "epoch": 0.38971060573617167,
      "grad_norm": 1498.3258056640625,
      "learning_rate": 3.084355173450636e-05,
      "loss": 45.1507,
      "step": 96460
    },
    {
      "epoch": 0.3897510070015393,
      "grad_norm": 1129.8160400390625,
      "learning_rate": 3.084120516928003e-05,
      "loss": 49.5181,
      "step": 96470
    },
    {
      "epoch": 0.38979140826690695,
      "grad_norm": 323.3406982421875,
      "learning_rate": 3.0838858392698825e-05,
      "loss": 33.0907,
      "step": 96480
    },
    {
      "epoch": 0.38983180953227453,
      "grad_norm": 1923.0677490234375,
      "learning_rate": 3.08365114048085e-05,
      "loss": 69.1391,
      "step": 96490
    },
    {
      "epoch": 0.38987221079764217,
      "grad_norm": 1291.444580078125,
      "learning_rate": 3.08341642056548e-05,
      "loss": 63.5927,
      "step": 96500
    },
    {
      "epoch": 0.3899126120630098,
      "grad_norm": 527.068359375,
      "learning_rate": 3.083181679528349e-05,
      "loss": 48.1131,
      "step": 96510
    },
    {
      "epoch": 0.38995301332837745,
      "grad_norm": 575.4950561523438,
      "learning_rate": 3.0829469173740344e-05,
      "loss": 60.9173,
      "step": 96520
    },
    {
      "epoch": 0.3899934145937451,
      "grad_norm": 1124.62744140625,
      "learning_rate": 3.082712134107111e-05,
      "loss": 63.2891,
      "step": 96530
    },
    {
      "epoch": 0.39003381585911273,
      "grad_norm": 2542.8984375,
      "learning_rate": 3.0824773297321585e-05,
      "loss": 85.4169,
      "step": 96540
    },
    {
      "epoch": 0.3900742171244803,
      "grad_norm": 523.2858276367188,
      "learning_rate": 3.082242504253752e-05,
      "loss": 39.4402,
      "step": 96550
    },
    {
      "epoch": 0.39011461838984796,
      "grad_norm": 902.9794311523438,
      "learning_rate": 3.082007657676471e-05,
      "loss": 52.0898,
      "step": 96560
    },
    {
      "epoch": 0.3901550196552156,
      "grad_norm": 1041.2705078125,
      "learning_rate": 3.081772790004894e-05,
      "loss": 53.8274,
      "step": 96570
    },
    {
      "epoch": 0.39019542092058324,
      "grad_norm": 767.1126098632812,
      "learning_rate": 3.081537901243599e-05,
      "loss": 59.9269,
      "step": 96580
    },
    {
      "epoch": 0.3902358221859509,
      "grad_norm": 1110.6146240234375,
      "learning_rate": 3.081302991397166e-05,
      "loss": 56.435,
      "step": 96590
    },
    {
      "epoch": 0.3902762234513185,
      "grad_norm": 398.89532470703125,
      "learning_rate": 3.081068060470174e-05,
      "loss": 66.5738,
      "step": 96600
    },
    {
      "epoch": 0.3903166247166861,
      "grad_norm": 1773.4658203125,
      "learning_rate": 3.080833108467204e-05,
      "loss": 90.1683,
      "step": 96610
    },
    {
      "epoch": 0.39035702598205374,
      "grad_norm": 671.03271484375,
      "learning_rate": 3.080598135392837e-05,
      "loss": 75.9407,
      "step": 96620
    },
    {
      "epoch": 0.3903974272474214,
      "grad_norm": 1317.63623046875,
      "learning_rate": 3.080363141251652e-05,
      "loss": 70.7445,
      "step": 96630
    },
    {
      "epoch": 0.390437828512789,
      "grad_norm": 946.417724609375,
      "learning_rate": 3.080128126048232e-05,
      "loss": 89.1544,
      "step": 96640
    },
    {
      "epoch": 0.39047822977815666,
      "grad_norm": 1541.2022705078125,
      "learning_rate": 3.0798930897871576e-05,
      "loss": 72.9328,
      "step": 96650
    },
    {
      "epoch": 0.3905186310435243,
      "grad_norm": 589.8878173828125,
      "learning_rate": 3.079658032473011e-05,
      "loss": 45.5723,
      "step": 96660
    },
    {
      "epoch": 0.39055903230889194,
      "grad_norm": 699.7723999023438,
      "learning_rate": 3.079422954110376e-05,
      "loss": 65.5379,
      "step": 96670
    },
    {
      "epoch": 0.3905994335742595,
      "grad_norm": 956.462646484375,
      "learning_rate": 3.0791878547038345e-05,
      "loss": 106.3264,
      "step": 96680
    },
    {
      "epoch": 0.39063983483962716,
      "grad_norm": 381.62884521484375,
      "learning_rate": 3.0789527342579705e-05,
      "loss": 60.4416,
      "step": 96690
    },
    {
      "epoch": 0.3906802361049948,
      "grad_norm": 853.3964233398438,
      "learning_rate": 3.078717592777367e-05,
      "loss": 56.7422,
      "step": 96700
    },
    {
      "epoch": 0.39072063737036244,
      "grad_norm": 980.0914916992188,
      "learning_rate": 3.078482430266609e-05,
      "loss": 70.416,
      "step": 96710
    },
    {
      "epoch": 0.3907610386357301,
      "grad_norm": 391.09124755859375,
      "learning_rate": 3.07824724673028e-05,
      "loss": 67.9484,
      "step": 96720
    },
    {
      "epoch": 0.3908014399010977,
      "grad_norm": 548.7298583984375,
      "learning_rate": 3.078012042172967e-05,
      "loss": 79.7029,
      "step": 96730
    },
    {
      "epoch": 0.3908418411664653,
      "grad_norm": 487.2161865234375,
      "learning_rate": 3.077776816599253e-05,
      "loss": 44.6873,
      "step": 96740
    },
    {
      "epoch": 0.39088224243183295,
      "grad_norm": 810.355224609375,
      "learning_rate": 3.0775415700137266e-05,
      "loss": 81.676,
      "step": 96750
    },
    {
      "epoch": 0.3909226436972006,
      "grad_norm": 484.19561767578125,
      "learning_rate": 3.077306302420971e-05,
      "loss": 47.9547,
      "step": 96760
    },
    {
      "epoch": 0.39096304496256823,
      "grad_norm": 1085.0274658203125,
      "learning_rate": 3.0770710138255754e-05,
      "loss": 59.3709,
      "step": 96770
    },
    {
      "epoch": 0.39100344622793587,
      "grad_norm": 805.6069946289062,
      "learning_rate": 3.076835704232126e-05,
      "loss": 57.6012,
      "step": 96780
    },
    {
      "epoch": 0.3910438474933035,
      "grad_norm": 517.8592529296875,
      "learning_rate": 3.076600373645209e-05,
      "loss": 58.5938,
      "step": 96790
    },
    {
      "epoch": 0.39108424875867115,
      "grad_norm": 1210.808349609375,
      "learning_rate": 3.0763650220694145e-05,
      "loss": 82.4248,
      "step": 96800
    },
    {
      "epoch": 0.39112465002403873,
      "grad_norm": 493.9080505371094,
      "learning_rate": 3.0761296495093294e-05,
      "loss": 82.6033,
      "step": 96810
    },
    {
      "epoch": 0.3911650512894064,
      "grad_norm": 649.4231567382812,
      "learning_rate": 3.075894255969543e-05,
      "loss": 45.7087,
      "step": 96820
    },
    {
      "epoch": 0.391205452554774,
      "grad_norm": 1165.13671875,
      "learning_rate": 3.075658841454644e-05,
      "loss": 46.0291,
      "step": 96830
    },
    {
      "epoch": 0.39124585382014165,
      "grad_norm": 1765.5262451171875,
      "learning_rate": 3.075423405969222e-05,
      "loss": 49.481,
      "step": 96840
    },
    {
      "epoch": 0.3912862550855093,
      "grad_norm": 1000.0485229492188,
      "learning_rate": 3.075187949517867e-05,
      "loss": 69.4488,
      "step": 96850
    },
    {
      "epoch": 0.39132665635087693,
      "grad_norm": 1381.7174072265625,
      "learning_rate": 3.07495247210517e-05,
      "loss": 90.2943,
      "step": 96860
    },
    {
      "epoch": 0.3913670576162445,
      "grad_norm": 1334.7869873046875,
      "learning_rate": 3.07471697373572e-05,
      "loss": 59.9629,
      "step": 96870
    },
    {
      "epoch": 0.39140745888161216,
      "grad_norm": 530.7562255859375,
      "learning_rate": 3.0744814544141096e-05,
      "loss": 80.5092,
      "step": 96880
    },
    {
      "epoch": 0.3914478601469798,
      "grad_norm": 398.388671875,
      "learning_rate": 3.07424591414493e-05,
      "loss": 88.7593,
      "step": 96890
    },
    {
      "epoch": 0.39148826141234744,
      "grad_norm": 830.0609741210938,
      "learning_rate": 3.0740103529327744e-05,
      "loss": 79.1814,
      "step": 96900
    },
    {
      "epoch": 0.3915286626777151,
      "grad_norm": 0.0,
      "learning_rate": 3.0737747707822334e-05,
      "loss": 45.4471,
      "step": 96910
    },
    {
      "epoch": 0.3915690639430827,
      "grad_norm": 781.1683349609375,
      "learning_rate": 3.0735391676978993e-05,
      "loss": 49.7049,
      "step": 96920
    },
    {
      "epoch": 0.3916094652084503,
      "grad_norm": 841.4095458984375,
      "learning_rate": 3.0733035436843684e-05,
      "loss": 74.8256,
      "step": 96930
    },
    {
      "epoch": 0.39164986647381794,
      "grad_norm": 909.210693359375,
      "learning_rate": 3.073067898746231e-05,
      "loss": 70.243,
      "step": 96940
    },
    {
      "epoch": 0.3916902677391856,
      "grad_norm": 695.3448486328125,
      "learning_rate": 3.0728322328880825e-05,
      "loss": 42.8311,
      "step": 96950
    },
    {
      "epoch": 0.3917306690045532,
      "grad_norm": 480.36798095703125,
      "learning_rate": 3.0725965461145175e-05,
      "loss": 64.1772,
      "step": 96960
    },
    {
      "epoch": 0.39177107026992086,
      "grad_norm": 376.0433349609375,
      "learning_rate": 3.072360838430131e-05,
      "loss": 57.2788,
      "step": 96970
    },
    {
      "epoch": 0.3918114715352885,
      "grad_norm": 926.780029296875,
      "learning_rate": 3.0721251098395175e-05,
      "loss": 62.5882,
      "step": 96980
    },
    {
      "epoch": 0.39185187280065614,
      "grad_norm": 625.6805419921875,
      "learning_rate": 3.0718893603472725e-05,
      "loss": 53.4292,
      "step": 96990
    },
    {
      "epoch": 0.3918922740660237,
      "grad_norm": 733.6526489257812,
      "learning_rate": 3.0716535899579936e-05,
      "loss": 48.0489,
      "step": 97000
    },
    {
      "epoch": 0.39193267533139137,
      "grad_norm": 882.0390625,
      "learning_rate": 3.071417798676276e-05,
      "loss": 60.4882,
      "step": 97010
    },
    {
      "epoch": 0.391973076596759,
      "grad_norm": 218.1538543701172,
      "learning_rate": 3.0711819865067165e-05,
      "loss": 66.0629,
      "step": 97020
    },
    {
      "epoch": 0.39201347786212665,
      "grad_norm": 232.99609375,
      "learning_rate": 3.070946153453913e-05,
      "loss": 50.7751,
      "step": 97030
    },
    {
      "epoch": 0.3920538791274943,
      "grad_norm": 958.5475463867188,
      "learning_rate": 3.070710299522462e-05,
      "loss": 48.1551,
      "step": 97040
    },
    {
      "epoch": 0.3920942803928619,
      "grad_norm": 574.308837890625,
      "learning_rate": 3.070474424716963e-05,
      "loss": 63.7208,
      "step": 97050
    },
    {
      "epoch": 0.3921346816582295,
      "grad_norm": 977.0123901367188,
      "learning_rate": 3.0702385290420145e-05,
      "loss": 67.0578,
      "step": 97060
    },
    {
      "epoch": 0.39217508292359715,
      "grad_norm": 2389.80859375,
      "learning_rate": 3.070002612502215e-05,
      "loss": 61.1409,
      "step": 97070
    },
    {
      "epoch": 0.3922154841889648,
      "grad_norm": 498.7288818359375,
      "learning_rate": 3.069766675102162e-05,
      "loss": 48.6175,
      "step": 97080
    },
    {
      "epoch": 0.39225588545433243,
      "grad_norm": 1029.843994140625,
      "learning_rate": 3.0695307168464584e-05,
      "loss": 69.121,
      "step": 97090
    },
    {
      "epoch": 0.39229628671970007,
      "grad_norm": 618.8443603515625,
      "learning_rate": 3.069294737739703e-05,
      "loss": 76.3978,
      "step": 97100
    },
    {
      "epoch": 0.3923366879850677,
      "grad_norm": 963.9727172851562,
      "learning_rate": 3.069058737786496e-05,
      "loss": 49.647,
      "step": 97110
    },
    {
      "epoch": 0.39237708925043535,
      "grad_norm": 422.73382568359375,
      "learning_rate": 3.068822716991438e-05,
      "loss": 51.2614,
      "step": 97120
    },
    {
      "epoch": 0.39241749051580294,
      "grad_norm": 1414.1259765625,
      "learning_rate": 3.068586675359131e-05,
      "loss": 74.2478,
      "step": 97130
    },
    {
      "epoch": 0.3924578917811706,
      "grad_norm": 1767.93017578125,
      "learning_rate": 3.0683506128941774e-05,
      "loss": 59.4232,
      "step": 97140
    },
    {
      "epoch": 0.3924982930465382,
      "grad_norm": 662.4166870117188,
      "learning_rate": 3.0681145296011774e-05,
      "loss": 78.8385,
      "step": 97150
    },
    {
      "epoch": 0.39253869431190586,
      "grad_norm": 504.7801208496094,
      "learning_rate": 3.067878425484735e-05,
      "loss": 52.6057,
      "step": 97160
    },
    {
      "epoch": 0.3925790955772735,
      "grad_norm": 617.1817016601562,
      "learning_rate": 3.067642300549453e-05,
      "loss": 48.8813,
      "step": 97170
    },
    {
      "epoch": 0.39261949684264114,
      "grad_norm": 898.6566772460938,
      "learning_rate": 3.067406154799935e-05,
      "loss": 56.1757,
      "step": 97180
    },
    {
      "epoch": 0.3926598981080087,
      "grad_norm": 545.7286987304688,
      "learning_rate": 3.0671699882407845e-05,
      "loss": 77.3499,
      "step": 97190
    },
    {
      "epoch": 0.39270029937337636,
      "grad_norm": 533.9075927734375,
      "learning_rate": 3.066933800876606e-05,
      "loss": 60.7204,
      "step": 97200
    },
    {
      "epoch": 0.392740700638744,
      "grad_norm": 599.5408325195312,
      "learning_rate": 3.066697592712003e-05,
      "loss": 52.177,
      "step": 97210
    },
    {
      "epoch": 0.39278110190411164,
      "grad_norm": 304.1415710449219,
      "learning_rate": 3.066461363751582e-05,
      "loss": 43.3299,
      "step": 97220
    },
    {
      "epoch": 0.3928215031694793,
      "grad_norm": 899.7566528320312,
      "learning_rate": 3.066225113999946e-05,
      "loss": 63.157,
      "step": 97230
    },
    {
      "epoch": 0.3928619044348469,
      "grad_norm": 364.7635803222656,
      "learning_rate": 3.0659888434617045e-05,
      "loss": 58.8745,
      "step": 97240
    },
    {
      "epoch": 0.3929023057002145,
      "grad_norm": 661.5354614257812,
      "learning_rate": 3.06575255214146e-05,
      "loss": 43.469,
      "step": 97250
    },
    {
      "epoch": 0.39294270696558214,
      "grad_norm": 199.37904357910156,
      "learning_rate": 3.0655162400438214e-05,
      "loss": 70.2049,
      "step": 97260
    },
    {
      "epoch": 0.3929831082309498,
      "grad_norm": 2784.505615234375,
      "learning_rate": 3.0652799071733956e-05,
      "loss": 69.4415,
      "step": 97270
    },
    {
      "epoch": 0.3930235094963174,
      "grad_norm": 2461.99951171875,
      "learning_rate": 3.06504355353479e-05,
      "loss": 87.7411,
      "step": 97280
    },
    {
      "epoch": 0.39306391076168506,
      "grad_norm": 1862.85400390625,
      "learning_rate": 3.0648071791326114e-05,
      "loss": 56.1075,
      "step": 97290
    },
    {
      "epoch": 0.3931043120270527,
      "grad_norm": 2298.992431640625,
      "learning_rate": 3.064570783971468e-05,
      "loss": 87.6112,
      "step": 97300
    },
    {
      "epoch": 0.39314471329242034,
      "grad_norm": 1390.7154541015625,
      "learning_rate": 3.0643343680559704e-05,
      "loss": 77.7914,
      "step": 97310
    },
    {
      "epoch": 0.39318511455778793,
      "grad_norm": 696.4583740234375,
      "learning_rate": 3.064097931390725e-05,
      "loss": 55.1994,
      "step": 97320
    },
    {
      "epoch": 0.39322551582315557,
      "grad_norm": 442.60455322265625,
      "learning_rate": 3.0638614739803435e-05,
      "loss": 54.2116,
      "step": 97330
    },
    {
      "epoch": 0.3932659170885232,
      "grad_norm": 1173.42431640625,
      "learning_rate": 3.063624995829434e-05,
      "loss": 57.2077,
      "step": 97340
    },
    {
      "epoch": 0.39330631835389085,
      "grad_norm": 678.0792236328125,
      "learning_rate": 3.0633884969426085e-05,
      "loss": 39.9277,
      "step": 97350
    },
    {
      "epoch": 0.3933467196192585,
      "grad_norm": 648.6259765625,
      "learning_rate": 3.063151977324476e-05,
      "loss": 83.9109,
      "step": 97360
    },
    {
      "epoch": 0.39338712088462613,
      "grad_norm": 984.8261108398438,
      "learning_rate": 3.0629154369796494e-05,
      "loss": 61.3448,
      "step": 97370
    },
    {
      "epoch": 0.3934275221499937,
      "grad_norm": 1044.814208984375,
      "learning_rate": 3.0626788759127384e-05,
      "loss": 60.0398,
      "step": 97380
    },
    {
      "epoch": 0.39346792341536135,
      "grad_norm": 487.1873779296875,
      "learning_rate": 3.062442294128356e-05,
      "loss": 61.3103,
      "step": 97390
    },
    {
      "epoch": 0.393508324680729,
      "grad_norm": 865.4752807617188,
      "learning_rate": 3.0622056916311134e-05,
      "loss": 90.9223,
      "step": 97400
    },
    {
      "epoch": 0.39354872594609663,
      "grad_norm": 667.01123046875,
      "learning_rate": 3.061969068425625e-05,
      "loss": 65.7805,
      "step": 97410
    },
    {
      "epoch": 0.3935891272114643,
      "grad_norm": 314.469482421875,
      "learning_rate": 3.0617324245165014e-05,
      "loss": 67.8273,
      "step": 97420
    },
    {
      "epoch": 0.3936295284768319,
      "grad_norm": 1064.36181640625,
      "learning_rate": 3.061495759908359e-05,
      "loss": 74.8667,
      "step": 97430
    },
    {
      "epoch": 0.3936699297421995,
      "grad_norm": 2008.50537109375,
      "learning_rate": 3.0612590746058094e-05,
      "loss": 96.3463,
      "step": 97440
    },
    {
      "epoch": 0.39371033100756714,
      "grad_norm": 670.7850952148438,
      "learning_rate": 3.0610223686134685e-05,
      "loss": 74.3781,
      "step": 97450
    },
    {
      "epoch": 0.3937507322729348,
      "grad_norm": 371.1806640625,
      "learning_rate": 3.06078564193595e-05,
      "loss": 58.2147,
      "step": 97460
    },
    {
      "epoch": 0.3937911335383024,
      "grad_norm": 210.1017608642578,
      "learning_rate": 3.060548894577869e-05,
      "loss": 50.8013,
      "step": 97470
    },
    {
      "epoch": 0.39383153480367006,
      "grad_norm": 1518.7855224609375,
      "learning_rate": 3.060312126543842e-05,
      "loss": 62.9751,
      "step": 97480
    },
    {
      "epoch": 0.3938719360690377,
      "grad_norm": 422.8009033203125,
      "learning_rate": 3.060075337838483e-05,
      "loss": 41.03,
      "step": 97490
    },
    {
      "epoch": 0.39391233733440534,
      "grad_norm": 1671.2236328125,
      "learning_rate": 3.05983852846641e-05,
      "loss": 60.0833,
      "step": 97500
    },
    {
      "epoch": 0.3939527385997729,
      "grad_norm": 1641.0908203125,
      "learning_rate": 3.05960169843224e-05,
      "loss": 67.803,
      "step": 97510
    },
    {
      "epoch": 0.39399313986514056,
      "grad_norm": 810.8970947265625,
      "learning_rate": 3.0593648477405885e-05,
      "loss": 76.5214,
      "step": 97520
    },
    {
      "epoch": 0.3940335411305082,
      "grad_norm": 1572.5703125,
      "learning_rate": 3.0591279763960735e-05,
      "loss": 68.239,
      "step": 97530
    },
    {
      "epoch": 0.39407394239587584,
      "grad_norm": 717.9578857421875,
      "learning_rate": 3.058891084403315e-05,
      "loss": 83.6105,
      "step": 97540
    },
    {
      "epoch": 0.3941143436612435,
      "grad_norm": 220.7050018310547,
      "learning_rate": 3.0586541717669286e-05,
      "loss": 51.4329,
      "step": 97550
    },
    {
      "epoch": 0.3941547449266111,
      "grad_norm": 1055.88671875,
      "learning_rate": 3.058417238491534e-05,
      "loss": 53.919,
      "step": 97560
    },
    {
      "epoch": 0.3941951461919787,
      "grad_norm": 330.67095947265625,
      "learning_rate": 3.058180284581751e-05,
      "loss": 52.7184,
      "step": 97570
    },
    {
      "epoch": 0.39423554745734635,
      "grad_norm": 620.1004638671875,
      "learning_rate": 3.0579433100421974e-05,
      "loss": 52.7724,
      "step": 97580
    },
    {
      "epoch": 0.394275948722714,
      "grad_norm": 1500.8912353515625,
      "learning_rate": 3.0577063148774955e-05,
      "loss": 52.9456,
      "step": 97590
    },
    {
      "epoch": 0.3943163499880816,
      "grad_norm": 718.37060546875,
      "learning_rate": 3.057469299092264e-05,
      "loss": 89.2843,
      "step": 97600
    },
    {
      "epoch": 0.39435675125344927,
      "grad_norm": 690.8677368164062,
      "learning_rate": 3.0572322626911235e-05,
      "loss": 58.6851,
      "step": 97610
    },
    {
      "epoch": 0.3943971525188169,
      "grad_norm": 586.640380859375,
      "learning_rate": 3.056995205678697e-05,
      "loss": 49.8112,
      "step": 97620
    },
    {
      "epoch": 0.39443755378418455,
      "grad_norm": 488.9980163574219,
      "learning_rate": 3.056758128059603e-05,
      "loss": 68.7142,
      "step": 97630
    },
    {
      "epoch": 0.39447795504955213,
      "grad_norm": 1710.490966796875,
      "learning_rate": 3.056521029838467e-05,
      "loss": 63.958,
      "step": 97640
    },
    {
      "epoch": 0.39451835631491977,
      "grad_norm": 1692.4881591796875,
      "learning_rate": 3.056283911019909e-05,
      "loss": 98.5401,
      "step": 97650
    },
    {
      "epoch": 0.3945587575802874,
      "grad_norm": 545.5447387695312,
      "learning_rate": 3.056046771608552e-05,
      "loss": 49.8526,
      "step": 97660
    },
    {
      "epoch": 0.39459915884565505,
      "grad_norm": 1038.90478515625,
      "learning_rate": 3.0558096116090206e-05,
      "loss": 74.6395,
      "step": 97670
    },
    {
      "epoch": 0.3946395601110227,
      "grad_norm": 1187.12109375,
      "learning_rate": 3.055572431025936e-05,
      "loss": 64.7024,
      "step": 97680
    },
    {
      "epoch": 0.39467996137639033,
      "grad_norm": 892.6353759765625,
      "learning_rate": 3.055335229863924e-05,
      "loss": 50.8247,
      "step": 97690
    },
    {
      "epoch": 0.3947203626417579,
      "grad_norm": 588.580322265625,
      "learning_rate": 3.0550980081276075e-05,
      "loss": 44.0533,
      "step": 97700
    },
    {
      "epoch": 0.39476076390712556,
      "grad_norm": 586.703857421875,
      "learning_rate": 3.054860765821613e-05,
      "loss": 65.8129,
      "step": 97710
    },
    {
      "epoch": 0.3948011651724932,
      "grad_norm": 501.59954833984375,
      "learning_rate": 3.054623502950565e-05,
      "loss": 54.969,
      "step": 97720
    },
    {
      "epoch": 0.39484156643786084,
      "grad_norm": 2821.66357421875,
      "learning_rate": 3.054386219519088e-05,
      "loss": 90.8097,
      "step": 97730
    },
    {
      "epoch": 0.3948819677032285,
      "grad_norm": 957.9890747070312,
      "learning_rate": 3.0541489155318096e-05,
      "loss": 73.9644,
      "step": 97740
    },
    {
      "epoch": 0.3949223689685961,
      "grad_norm": 2029.6026611328125,
      "learning_rate": 3.0539115909933554e-05,
      "loss": 70.6401,
      "step": 97750
    },
    {
      "epoch": 0.3949627702339637,
      "grad_norm": 469.2944030761719,
      "learning_rate": 3.053674245908352e-05,
      "loss": 71.6168,
      "step": 97760
    },
    {
      "epoch": 0.39500317149933134,
      "grad_norm": 521.1866455078125,
      "learning_rate": 3.0534368802814266e-05,
      "loss": 60.3303,
      "step": 97770
    },
    {
      "epoch": 0.395043572764699,
      "grad_norm": 829.1405639648438,
      "learning_rate": 3.053199494117206e-05,
      "loss": 59.322,
      "step": 97780
    },
    {
      "epoch": 0.3950839740300666,
      "grad_norm": 768.0505981445312,
      "learning_rate": 3.0529620874203204e-05,
      "loss": 80.8334,
      "step": 97790
    },
    {
      "epoch": 0.39512437529543426,
      "grad_norm": 376.0151672363281,
      "learning_rate": 3.052724660195396e-05,
      "loss": 40.3237,
      "step": 97800
    },
    {
      "epoch": 0.3951647765608019,
      "grad_norm": 0.0,
      "learning_rate": 3.052487212447063e-05,
      "loss": 76.3937,
      "step": 97810
    },
    {
      "epoch": 0.39520517782616954,
      "grad_norm": 284.3887023925781,
      "learning_rate": 3.05224974417995e-05,
      "loss": 48.1156,
      "step": 97820
    },
    {
      "epoch": 0.3952455790915371,
      "grad_norm": 613.3410034179688,
      "learning_rate": 3.0520122553986855e-05,
      "loss": 58.0234,
      "step": 97830
    },
    {
      "epoch": 0.39528598035690476,
      "grad_norm": 1629.0135498046875,
      "learning_rate": 3.051774746107901e-05,
      "loss": 82.4246,
      "step": 97840
    },
    {
      "epoch": 0.3953263816222724,
      "grad_norm": 217.6327667236328,
      "learning_rate": 3.051537216312227e-05,
      "loss": 55.1513,
      "step": 97850
    },
    {
      "epoch": 0.39536678288764004,
      "grad_norm": 827.3215942382812,
      "learning_rate": 3.0512996660162923e-05,
      "loss": 49.9513,
      "step": 97860
    },
    {
      "epoch": 0.3954071841530077,
      "grad_norm": 958.5142822265625,
      "learning_rate": 3.05106209522473e-05,
      "loss": 64.4401,
      "step": 97870
    },
    {
      "epoch": 0.3954475854183753,
      "grad_norm": 453.5467834472656,
      "learning_rate": 3.0508245039421713e-05,
      "loss": 39.7421,
      "step": 97880
    },
    {
      "epoch": 0.3954879866837429,
      "grad_norm": 733.8382568359375,
      "learning_rate": 3.050586892173248e-05,
      "loss": 61.8497,
      "step": 97890
    },
    {
      "epoch": 0.39552838794911055,
      "grad_norm": 1093.428955078125,
      "learning_rate": 3.0503492599225918e-05,
      "loss": 101.909,
      "step": 97900
    },
    {
      "epoch": 0.3955687892144782,
      "grad_norm": 477.01220703125,
      "learning_rate": 3.0501116071948363e-05,
      "loss": 63.0303,
      "step": 97910
    },
    {
      "epoch": 0.39560919047984583,
      "grad_norm": 235.736572265625,
      "learning_rate": 3.0498739339946143e-05,
      "loss": 43.2636,
      "step": 97920
    },
    {
      "epoch": 0.39564959174521347,
      "grad_norm": 570.0729370117188,
      "learning_rate": 3.0496362403265594e-05,
      "loss": 61.491,
      "step": 97930
    },
    {
      "epoch": 0.3956899930105811,
      "grad_norm": 1310.9716796875,
      "learning_rate": 3.0493985261953056e-05,
      "loss": 111.6765,
      "step": 97940
    },
    {
      "epoch": 0.39573039427594875,
      "grad_norm": 589.3435668945312,
      "learning_rate": 3.0491607916054867e-05,
      "loss": 42.1561,
      "step": 97950
    },
    {
      "epoch": 0.39577079554131633,
      "grad_norm": 1587.877685546875,
      "learning_rate": 3.048923036561739e-05,
      "loss": 58.7071,
      "step": 97960
    },
    {
      "epoch": 0.395811196806684,
      "grad_norm": 567.8264770507812,
      "learning_rate": 3.0486852610686967e-05,
      "loss": 64.5263,
      "step": 97970
    },
    {
      "epoch": 0.3958515980720516,
      "grad_norm": 651.44287109375,
      "learning_rate": 3.0484474651309947e-05,
      "loss": 54.3802,
      "step": 97980
    },
    {
      "epoch": 0.39589199933741925,
      "grad_norm": 1693.81494140625,
      "learning_rate": 3.04820964875327e-05,
      "loss": 48.9196,
      "step": 97990
    },
    {
      "epoch": 0.3959324006027869,
      "grad_norm": 304.3753967285156,
      "learning_rate": 3.0479718119401588e-05,
      "loss": 68.3677,
      "step": 98000
    },
    {
      "epoch": 0.39597280186815453,
      "grad_norm": 679.2041015625,
      "learning_rate": 3.0477339546962974e-05,
      "loss": 40.7732,
      "step": 98010
    },
    {
      "epoch": 0.3960132031335221,
      "grad_norm": 432.08453369140625,
      "learning_rate": 3.0474960770263223e-05,
      "loss": 65.5436,
      "step": 98020
    },
    {
      "epoch": 0.39605360439888976,
      "grad_norm": 646.598876953125,
      "learning_rate": 3.0472581789348728e-05,
      "loss": 85.996,
      "step": 98030
    },
    {
      "epoch": 0.3960940056642574,
      "grad_norm": 449.9706726074219,
      "learning_rate": 3.0470202604265858e-05,
      "loss": 49.8928,
      "step": 98040
    },
    {
      "epoch": 0.39613440692962504,
      "grad_norm": 1422.8955078125,
      "learning_rate": 3.0467823215060997e-05,
      "loss": 52.3219,
      "step": 98050
    },
    {
      "epoch": 0.3961748081949927,
      "grad_norm": 512.1029052734375,
      "learning_rate": 3.046544362178053e-05,
      "loss": 55.1474,
      "step": 98060
    },
    {
      "epoch": 0.3962152094603603,
      "grad_norm": 421.8543395996094,
      "learning_rate": 3.0463063824470854e-05,
      "loss": 64.9397,
      "step": 98070
    },
    {
      "epoch": 0.3962556107257279,
      "grad_norm": 1446.7802734375,
      "learning_rate": 3.0460683823178366e-05,
      "loss": 85.9275,
      "step": 98080
    },
    {
      "epoch": 0.39629601199109554,
      "grad_norm": 751.712158203125,
      "learning_rate": 3.045830361794946e-05,
      "loss": 53.0268,
      "step": 98090
    },
    {
      "epoch": 0.3963364132564632,
      "grad_norm": 984.9591064453125,
      "learning_rate": 3.0455923208830532e-05,
      "loss": 66.3646,
      "step": 98100
    },
    {
      "epoch": 0.3963768145218308,
      "grad_norm": 952.146240234375,
      "learning_rate": 3.0453542595868008e-05,
      "loss": 45.042,
      "step": 98110
    },
    {
      "epoch": 0.39641721578719846,
      "grad_norm": 1196.6591796875,
      "learning_rate": 3.045116177910829e-05,
      "loss": 65.4604,
      "step": 98120
    },
    {
      "epoch": 0.3964576170525661,
      "grad_norm": 808.1182861328125,
      "learning_rate": 3.044878075859778e-05,
      "loss": 72.1062,
      "step": 98130
    },
    {
      "epoch": 0.39649801831793374,
      "grad_norm": 558.8911743164062,
      "learning_rate": 3.0446399534382923e-05,
      "loss": 42.1433,
      "step": 98140
    },
    {
      "epoch": 0.3965384195833013,
      "grad_norm": 771.634521484375,
      "learning_rate": 3.0444018106510122e-05,
      "loss": 54.9019,
      "step": 98150
    },
    {
      "epoch": 0.39657882084866897,
      "grad_norm": 369.722412109375,
      "learning_rate": 3.0441636475025816e-05,
      "loss": 56.4451,
      "step": 98160
    },
    {
      "epoch": 0.3966192221140366,
      "grad_norm": 693.5859375,
      "learning_rate": 3.0439254639976427e-05,
      "loss": 71.9817,
      "step": 98170
    },
    {
      "epoch": 0.39665962337940425,
      "grad_norm": 650.7984619140625,
      "learning_rate": 3.0436872601408392e-05,
      "loss": 63.2401,
      "step": 98180
    },
    {
      "epoch": 0.3967000246447719,
      "grad_norm": 570.8035888671875,
      "learning_rate": 3.0434490359368155e-05,
      "loss": 51.4647,
      "step": 98190
    },
    {
      "epoch": 0.3967404259101395,
      "grad_norm": 1090.2740478515625,
      "learning_rate": 3.0432107913902162e-05,
      "loss": 66.7061,
      "step": 98200
    },
    {
      "epoch": 0.3967808271755071,
      "grad_norm": 1143.412353515625,
      "learning_rate": 3.0429725265056843e-05,
      "loss": 48.8705,
      "step": 98210
    },
    {
      "epoch": 0.39682122844087475,
      "grad_norm": 892.337158203125,
      "learning_rate": 3.0427342412878666e-05,
      "loss": 39.7136,
      "step": 98220
    },
    {
      "epoch": 0.3968616297062424,
      "grad_norm": 858.3418579101562,
      "learning_rate": 3.042495935741409e-05,
      "loss": 52.7396,
      "step": 98230
    },
    {
      "epoch": 0.39690203097161003,
      "grad_norm": 1290.9305419921875,
      "learning_rate": 3.0422576098709554e-05,
      "loss": 61.2084,
      "step": 98240
    },
    {
      "epoch": 0.39694243223697767,
      "grad_norm": 867.2406005859375,
      "learning_rate": 3.0420192636811535e-05,
      "loss": 60.7522,
      "step": 98250
    },
    {
      "epoch": 0.3969828335023453,
      "grad_norm": 968.0877075195312,
      "learning_rate": 3.0417808971766488e-05,
      "loss": 66.9356,
      "step": 98260
    },
    {
      "epoch": 0.39702323476771295,
      "grad_norm": 539.1856079101562,
      "learning_rate": 3.0415425103620893e-05,
      "loss": 52.8227,
      "step": 98270
    },
    {
      "epoch": 0.39706363603308054,
      "grad_norm": 1524.3682861328125,
      "learning_rate": 3.041304103242123e-05,
      "loss": 65.6602,
      "step": 98280
    },
    {
      "epoch": 0.3971040372984482,
      "grad_norm": 585.87158203125,
      "learning_rate": 3.041065675821397e-05,
      "loss": 60.9574,
      "step": 98290
    },
    {
      "epoch": 0.3971444385638158,
      "grad_norm": 426.2937927246094,
      "learning_rate": 3.0408272281045598e-05,
      "loss": 52.4128,
      "step": 98300
    },
    {
      "epoch": 0.39718483982918346,
      "grad_norm": 905.8636474609375,
      "learning_rate": 3.0405887600962592e-05,
      "loss": 62.0832,
      "step": 98310
    },
    {
      "epoch": 0.3972252410945511,
      "grad_norm": 684.369873046875,
      "learning_rate": 3.0403502718011456e-05,
      "loss": 54.1064,
      "step": 98320
    },
    {
      "epoch": 0.39726564235991874,
      "grad_norm": 713.6025390625,
      "learning_rate": 3.040111763223868e-05,
      "loss": 46.3603,
      "step": 98330
    },
    {
      "epoch": 0.3973060436252863,
      "grad_norm": 483.4386901855469,
      "learning_rate": 3.0398732343690755e-05,
      "loss": 55.9319,
      "step": 98340
    },
    {
      "epoch": 0.39734644489065396,
      "grad_norm": 411.5086669921875,
      "learning_rate": 3.039634685241419e-05,
      "loss": 48.6646,
      "step": 98350
    },
    {
      "epoch": 0.3973868461560216,
      "grad_norm": 785.1961669921875,
      "learning_rate": 3.0393961158455494e-05,
      "loss": 74.4425,
      "step": 98360
    },
    {
      "epoch": 0.39742724742138924,
      "grad_norm": 511.2066650390625,
      "learning_rate": 3.0391575261861175e-05,
      "loss": 58.8918,
      "step": 98370
    },
    {
      "epoch": 0.3974676486867569,
      "grad_norm": 531.2205810546875,
      "learning_rate": 3.0389189162677746e-05,
      "loss": 50.1892,
      "step": 98380
    },
    {
      "epoch": 0.3975080499521245,
      "grad_norm": 1097.158203125,
      "learning_rate": 3.0386802860951722e-05,
      "loss": 71.8221,
      "step": 98390
    },
    {
      "epoch": 0.3975484512174921,
      "grad_norm": 1047.31982421875,
      "learning_rate": 3.038441635672963e-05,
      "loss": 68.5992,
      "step": 98400
    },
    {
      "epoch": 0.39758885248285974,
      "grad_norm": 752.1648559570312,
      "learning_rate": 3.0382029650058e-05,
      "loss": 61.804,
      "step": 98410
    },
    {
      "epoch": 0.3976292537482274,
      "grad_norm": 765.8179931640625,
      "learning_rate": 3.037964274098335e-05,
      "loss": 53.9927,
      "step": 98420
    },
    {
      "epoch": 0.397669655013595,
      "grad_norm": 1045.579345703125,
      "learning_rate": 3.0377255629552222e-05,
      "loss": 61.845,
      "step": 98430
    },
    {
      "epoch": 0.39771005627896266,
      "grad_norm": 386.06671142578125,
      "learning_rate": 3.037486831581115e-05,
      "loss": 58.5219,
      "step": 98440
    },
    {
      "epoch": 0.3977504575443303,
      "grad_norm": 904.6885986328125,
      "learning_rate": 3.0372480799806686e-05,
      "loss": 73.4442,
      "step": 98450
    },
    {
      "epoch": 0.39779085880969794,
      "grad_norm": 1277.4661865234375,
      "learning_rate": 3.0370093081585366e-05,
      "loss": 110.8596,
      "step": 98460
    },
    {
      "epoch": 0.39783126007506553,
      "grad_norm": 544.1910400390625,
      "learning_rate": 3.036770516119374e-05,
      "loss": 116.7728,
      "step": 98470
    },
    {
      "epoch": 0.39787166134043317,
      "grad_norm": 1025.6640625,
      "learning_rate": 3.0365317038678362e-05,
      "loss": 53.2361,
      "step": 98480
    },
    {
      "epoch": 0.3979120626058008,
      "grad_norm": 1097.7906494140625,
      "learning_rate": 3.0362928714085804e-05,
      "loss": 46.7724,
      "step": 98490
    },
    {
      "epoch": 0.39795246387116845,
      "grad_norm": 776.1931762695312,
      "learning_rate": 3.036054018746261e-05,
      "loss": 55.6301,
      "step": 98500
    },
    {
      "epoch": 0.3979928651365361,
      "grad_norm": 755.2182006835938,
      "learning_rate": 3.0358151458855345e-05,
      "loss": 61.8012,
      "step": 98510
    },
    {
      "epoch": 0.39803326640190373,
      "grad_norm": 938.1300048828125,
      "learning_rate": 3.0355762528310588e-05,
      "loss": 44.2787,
      "step": 98520
    },
    {
      "epoch": 0.3980736676672713,
      "grad_norm": 2723.06005859375,
      "learning_rate": 3.035337339587491e-05,
      "loss": 58.1217,
      "step": 98530
    },
    {
      "epoch": 0.39811406893263895,
      "grad_norm": 1637.096923828125,
      "learning_rate": 3.035098406159489e-05,
      "loss": 88.4366,
      "step": 98540
    },
    {
      "epoch": 0.3981544701980066,
      "grad_norm": 576.7115478515625,
      "learning_rate": 3.0348594525517102e-05,
      "loss": 53.7875,
      "step": 98550
    },
    {
      "epoch": 0.39819487146337423,
      "grad_norm": 355.59912109375,
      "learning_rate": 3.0346204787688137e-05,
      "loss": 64.696,
      "step": 98560
    },
    {
      "epoch": 0.3982352727287419,
      "grad_norm": 449.38360595703125,
      "learning_rate": 3.0343814848154586e-05,
      "loss": 32.5586,
      "step": 98570
    },
    {
      "epoch": 0.3982756739941095,
      "grad_norm": 826.9142456054688,
      "learning_rate": 3.0341424706963035e-05,
      "loss": 69.185,
      "step": 98580
    },
    {
      "epoch": 0.39831607525947715,
      "grad_norm": 781.5284423828125,
      "learning_rate": 3.033903436416009e-05,
      "loss": 61.5975,
      "step": 98590
    },
    {
      "epoch": 0.39835647652484474,
      "grad_norm": 741.2518920898438,
      "learning_rate": 3.0336643819792342e-05,
      "loss": 68.1245,
      "step": 98600
    },
    {
      "epoch": 0.3983968777902124,
      "grad_norm": 894.6944580078125,
      "learning_rate": 3.0334253073906406e-05,
      "loss": 79.9229,
      "step": 98610
    },
    {
      "epoch": 0.39843727905558,
      "grad_norm": 473.85565185546875,
      "learning_rate": 3.0331862126548883e-05,
      "loss": 61.3682,
      "step": 98620
    },
    {
      "epoch": 0.39847768032094766,
      "grad_norm": 603.6492919921875,
      "learning_rate": 3.0329470977766387e-05,
      "loss": 66.0297,
      "step": 98630
    },
    {
      "epoch": 0.3985180815863153,
      "grad_norm": 889.0967407226562,
      "learning_rate": 3.0327079627605534e-05,
      "loss": 51.923,
      "step": 98640
    },
    {
      "epoch": 0.39855848285168294,
      "grad_norm": 610.2051391601562,
      "learning_rate": 3.0324688076112953e-05,
      "loss": 45.7892,
      "step": 98650
    },
    {
      "epoch": 0.3985988841170505,
      "grad_norm": 423.3575744628906,
      "learning_rate": 3.0322296323335263e-05,
      "loss": 77.9606,
      "step": 98660
    },
    {
      "epoch": 0.39863928538241816,
      "grad_norm": 492.7673645019531,
      "learning_rate": 3.031990436931909e-05,
      "loss": 45.4414,
      "step": 98670
    },
    {
      "epoch": 0.3986796866477858,
      "grad_norm": 400.84423828125,
      "learning_rate": 3.0317512214111066e-05,
      "loss": 55.4323,
      "step": 98680
    },
    {
      "epoch": 0.39872008791315344,
      "grad_norm": 1153.99560546875,
      "learning_rate": 3.031511985775783e-05,
      "loss": 68.0575,
      "step": 98690
    },
    {
      "epoch": 0.3987604891785211,
      "grad_norm": 528.0145263671875,
      "learning_rate": 3.0312727300306024e-05,
      "loss": 53.2187,
      "step": 98700
    },
    {
      "epoch": 0.3988008904438887,
      "grad_norm": 570.9860229492188,
      "learning_rate": 3.0310334541802287e-05,
      "loss": 86.0171,
      "step": 98710
    },
    {
      "epoch": 0.3988412917092563,
      "grad_norm": 479.5833740234375,
      "learning_rate": 3.030794158229327e-05,
      "loss": 80.2573,
      "step": 98720
    },
    {
      "epoch": 0.39888169297462395,
      "grad_norm": 2084.32666015625,
      "learning_rate": 3.030554842182563e-05,
      "loss": 64.0403,
      "step": 98730
    },
    {
      "epoch": 0.3989220942399916,
      "grad_norm": 534.7925415039062,
      "learning_rate": 3.0303155060446014e-05,
      "loss": 50.2246,
      "step": 98740
    },
    {
      "epoch": 0.3989624955053592,
      "grad_norm": 519.5526123046875,
      "learning_rate": 3.030076149820109e-05,
      "loss": 58.6772,
      "step": 98750
    },
    {
      "epoch": 0.39900289677072687,
      "grad_norm": 689.9923706054688,
      "learning_rate": 3.029836773513751e-05,
      "loss": 85.5789,
      "step": 98760
    },
    {
      "epoch": 0.3990432980360945,
      "grad_norm": 457.3630676269531,
      "learning_rate": 3.0295973771301956e-05,
      "loss": 43.8749,
      "step": 98770
    },
    {
      "epoch": 0.39908369930146215,
      "grad_norm": 532.2931518554688,
      "learning_rate": 3.0293579606741084e-05,
      "loss": 43.1888,
      "step": 98780
    },
    {
      "epoch": 0.39912410056682973,
      "grad_norm": 720.2774658203125,
      "learning_rate": 3.0291185241501587e-05,
      "loss": 82.4107,
      "step": 98790
    },
    {
      "epoch": 0.39916450183219737,
      "grad_norm": 822.1246948242188,
      "learning_rate": 3.028879067563013e-05,
      "loss": 43.3162,
      "step": 98800
    },
    {
      "epoch": 0.399204903097565,
      "grad_norm": 633.0422973632812,
      "learning_rate": 3.0286395909173403e-05,
      "loss": 56.5374,
      "step": 98810
    },
    {
      "epoch": 0.39924530436293265,
      "grad_norm": 1895.790283203125,
      "learning_rate": 3.0284000942178095e-05,
      "loss": 63.1824,
      "step": 98820
    },
    {
      "epoch": 0.3992857056283003,
      "grad_norm": 363.9815979003906,
      "learning_rate": 3.028160577469089e-05,
      "loss": 76.7197,
      "step": 98830
    },
    {
      "epoch": 0.39932610689366793,
      "grad_norm": 1479.4224853515625,
      "learning_rate": 3.0279210406758493e-05,
      "loss": 98.4598,
      "step": 98840
    },
    {
      "epoch": 0.3993665081590355,
      "grad_norm": 1820.2047119140625,
      "learning_rate": 3.0276814838427593e-05,
      "loss": 73.8971,
      "step": 98850
    },
    {
      "epoch": 0.39940690942440316,
      "grad_norm": 579.8154296875,
      "learning_rate": 3.02744190697449e-05,
      "loss": 60.1406,
      "step": 98860
    },
    {
      "epoch": 0.3994473106897708,
      "grad_norm": 701.7965087890625,
      "learning_rate": 3.027202310075711e-05,
      "loss": 67.9611,
      "step": 98870
    },
    {
      "epoch": 0.39948771195513844,
      "grad_norm": 425.21319580078125,
      "learning_rate": 3.026962693151094e-05,
      "loss": 50.4619,
      "step": 98880
    },
    {
      "epoch": 0.3995281132205061,
      "grad_norm": 491.5628356933594,
      "learning_rate": 3.0267230562053113e-05,
      "loss": 41.3457,
      "step": 98890
    },
    {
      "epoch": 0.3995685144858737,
      "grad_norm": 484.04644775390625,
      "learning_rate": 3.0264833992430343e-05,
      "loss": 84.0886,
      "step": 98900
    },
    {
      "epoch": 0.39960891575124136,
      "grad_norm": 869.12158203125,
      "learning_rate": 3.0262437222689344e-05,
      "loss": 63.6839,
      "step": 98910
    },
    {
      "epoch": 0.39964931701660894,
      "grad_norm": 781.686767578125,
      "learning_rate": 3.0260040252876856e-05,
      "loss": 70.227,
      "step": 98920
    },
    {
      "epoch": 0.3996897182819766,
      "grad_norm": 722.117919921875,
      "learning_rate": 3.0257643083039587e-05,
      "loss": 73.3941,
      "step": 98930
    },
    {
      "epoch": 0.3997301195473442,
      "grad_norm": 1093.8751220703125,
      "learning_rate": 3.0255245713224294e-05,
      "loss": 70.9664,
      "step": 98940
    },
    {
      "epoch": 0.39977052081271186,
      "grad_norm": 495.45074462890625,
      "learning_rate": 3.0252848143477706e-05,
      "loss": 46.6267,
      "step": 98950
    },
    {
      "epoch": 0.3998109220780795,
      "grad_norm": 625.2247314453125,
      "learning_rate": 3.025045037384656e-05,
      "loss": 67.4223,
      "step": 98960
    },
    {
      "epoch": 0.39985132334344714,
      "grad_norm": 591.3223876953125,
      "learning_rate": 3.0248052404377613e-05,
      "loss": 47.861,
      "step": 98970
    },
    {
      "epoch": 0.3998917246088147,
      "grad_norm": 2261.043212890625,
      "learning_rate": 3.0245654235117605e-05,
      "loss": 109.4352,
      "step": 98980
    },
    {
      "epoch": 0.39993212587418236,
      "grad_norm": 539.4677124023438,
      "learning_rate": 3.0243255866113292e-05,
      "loss": 58.3937,
      "step": 98990
    },
    {
      "epoch": 0.39997252713955,
      "grad_norm": 3327.10107421875,
      "learning_rate": 3.024085729741143e-05,
      "loss": 74.1484,
      "step": 99000
    },
    {
      "epoch": 0.40001292840491764,
      "grad_norm": 791.87890625,
      "learning_rate": 3.0238458529058792e-05,
      "loss": 73.247,
      "step": 99010
    },
    {
      "epoch": 0.4000533296702853,
      "grad_norm": 1245.586669921875,
      "learning_rate": 3.023605956110213e-05,
      "loss": 74.173,
      "step": 99020
    },
    {
      "epoch": 0.4000937309356529,
      "grad_norm": 538.5578002929688,
      "learning_rate": 3.0233660393588217e-05,
      "loss": 69.8499,
      "step": 99030
    },
    {
      "epoch": 0.4001341322010205,
      "grad_norm": 1586.10302734375,
      "learning_rate": 3.0231261026563817e-05,
      "loss": 58.608,
      "step": 99040
    },
    {
      "epoch": 0.40017453346638815,
      "grad_norm": 1514.328369140625,
      "learning_rate": 3.022886146007572e-05,
      "loss": 71.7001,
      "step": 99050
    },
    {
      "epoch": 0.4002149347317558,
      "grad_norm": 884.135986328125,
      "learning_rate": 3.0226461694170706e-05,
      "loss": 53.863,
      "step": 99060
    },
    {
      "epoch": 0.40025533599712343,
      "grad_norm": 366.4808654785156,
      "learning_rate": 3.022406172889556e-05,
      "loss": 58.081,
      "step": 99070
    },
    {
      "epoch": 0.40029573726249107,
      "grad_norm": 778.1707763671875,
      "learning_rate": 3.0221661564297062e-05,
      "loss": 53.0686,
      "step": 99080
    },
    {
      "epoch": 0.4003361385278587,
      "grad_norm": 570.4125366210938,
      "learning_rate": 3.021926120042201e-05,
      "loss": 67.3994,
      "step": 99090
    },
    {
      "epoch": 0.40037653979322635,
      "grad_norm": 640.3318481445312,
      "learning_rate": 3.02168606373172e-05,
      "loss": 56.3023,
      "step": 99100
    },
    {
      "epoch": 0.40041694105859393,
      "grad_norm": 757.1697998046875,
      "learning_rate": 3.0214459875029437e-05,
      "loss": 62.845,
      "step": 99110
    },
    {
      "epoch": 0.4004573423239616,
      "grad_norm": 313.5112609863281,
      "learning_rate": 3.0212058913605512e-05,
      "loss": 65.6793,
      "step": 99120
    },
    {
      "epoch": 0.4004977435893292,
      "grad_norm": 529.2091064453125,
      "learning_rate": 3.0209657753092242e-05,
      "loss": 58.1838,
      "step": 99130
    },
    {
      "epoch": 0.40053814485469685,
      "grad_norm": 1389.11083984375,
      "learning_rate": 3.0207256393536438e-05,
      "loss": 68.3362,
      "step": 99140
    },
    {
      "epoch": 0.4005785461200645,
      "grad_norm": 977.894287109375,
      "learning_rate": 3.020485483498492e-05,
      "loss": 56.8673,
      "step": 99150
    },
    {
      "epoch": 0.40061894738543213,
      "grad_norm": 1881.879150390625,
      "learning_rate": 3.0202453077484496e-05,
      "loss": 60.6721,
      "step": 99160
    },
    {
      "epoch": 0.4006593486507997,
      "grad_norm": 1947.5126953125,
      "learning_rate": 3.0200051121081996e-05,
      "loss": 67.9177,
      "step": 99170
    },
    {
      "epoch": 0.40069974991616736,
      "grad_norm": 1017.6248168945312,
      "learning_rate": 3.0197648965824258e-05,
      "loss": 57.1318,
      "step": 99180
    },
    {
      "epoch": 0.400740151181535,
      "grad_norm": 485.7748107910156,
      "learning_rate": 3.01952466117581e-05,
      "loss": 54.8188,
      "step": 99190
    },
    {
      "epoch": 0.40078055244690264,
      "grad_norm": 944.8651733398438,
      "learning_rate": 3.0192844058930356e-05,
      "loss": 60.584,
      "step": 99200
    },
    {
      "epoch": 0.4008209537122703,
      "grad_norm": 964.3688354492188,
      "learning_rate": 3.019044130738787e-05,
      "loss": 94.7983,
      "step": 99210
    },
    {
      "epoch": 0.4008613549776379,
      "grad_norm": 1407.83056640625,
      "learning_rate": 3.0188038357177497e-05,
      "loss": 48.4079,
      "step": 99220
    },
    {
      "epoch": 0.40090175624300556,
      "grad_norm": 583.0308837890625,
      "learning_rate": 3.0185635208346053e-05,
      "loss": 41.2452,
      "step": 99230
    },
    {
      "epoch": 0.40094215750837314,
      "grad_norm": 591.8720092773438,
      "learning_rate": 3.0183231860940412e-05,
      "loss": 55.6352,
      "step": 99240
    },
    {
      "epoch": 0.4009825587737408,
      "grad_norm": 1591.0367431640625,
      "learning_rate": 3.018082831500743e-05,
      "loss": 54.9869,
      "step": 99250
    },
    {
      "epoch": 0.4010229600391084,
      "grad_norm": 947.4282836914062,
      "learning_rate": 3.0178424570593954e-05,
      "loss": 60.0285,
      "step": 99260
    },
    {
      "epoch": 0.40106336130447606,
      "grad_norm": 726.7391967773438,
      "learning_rate": 3.0176020627746853e-05,
      "loss": 99.4484,
      "step": 99270
    },
    {
      "epoch": 0.4011037625698437,
      "grad_norm": 487.5047302246094,
      "learning_rate": 3.0173616486512983e-05,
      "loss": 83.2084,
      "step": 99280
    },
    {
      "epoch": 0.40114416383521134,
      "grad_norm": 2766.857177734375,
      "learning_rate": 3.017121214693923e-05,
      "loss": 72.9982,
      "step": 99290
    },
    {
      "epoch": 0.4011845651005789,
      "grad_norm": 781.8902587890625,
      "learning_rate": 3.016880760907246e-05,
      "loss": 47.2423,
      "step": 99300
    },
    {
      "epoch": 0.40122496636594657,
      "grad_norm": 644.5406494140625,
      "learning_rate": 3.0166402872959547e-05,
      "loss": 34.2714,
      "step": 99310
    },
    {
      "epoch": 0.4012653676313142,
      "grad_norm": 536.601806640625,
      "learning_rate": 3.0163997938647377e-05,
      "loss": 66.7018,
      "step": 99320
    },
    {
      "epoch": 0.40130576889668185,
      "grad_norm": 428.3992614746094,
      "learning_rate": 3.0161592806182826e-05,
      "loss": 62.3801,
      "step": 99330
    },
    {
      "epoch": 0.4013461701620495,
      "grad_norm": 1168.3619384765625,
      "learning_rate": 3.01591874756128e-05,
      "loss": 56.47,
      "step": 99340
    },
    {
      "epoch": 0.4013865714274171,
      "grad_norm": 704.8736572265625,
      "learning_rate": 3.0156781946984187e-05,
      "loss": 51.4939,
      "step": 99350
    },
    {
      "epoch": 0.4014269726927847,
      "grad_norm": 1718.3294677734375,
      "learning_rate": 3.0154376220343883e-05,
      "loss": 46.5367,
      "step": 99360
    },
    {
      "epoch": 0.40146737395815235,
      "grad_norm": 679.1923828125,
      "learning_rate": 3.0151970295738775e-05,
      "loss": 51.8275,
      "step": 99370
    },
    {
      "epoch": 0.40150777522352,
      "grad_norm": 598.9464721679688,
      "learning_rate": 3.0149564173215786e-05,
      "loss": 47.5122,
      "step": 99380
    },
    {
      "epoch": 0.40154817648888763,
      "grad_norm": 979.5762329101562,
      "learning_rate": 3.014715785282182e-05,
      "loss": 66.2203,
      "step": 99390
    },
    {
      "epoch": 0.40158857775425527,
      "grad_norm": 330.1836242675781,
      "learning_rate": 3.0144751334603787e-05,
      "loss": 57.3458,
      "step": 99400
    },
    {
      "epoch": 0.4016289790196229,
      "grad_norm": 316.8360290527344,
      "learning_rate": 3.01423446186086e-05,
      "loss": 98.0352,
      "step": 99410
    },
    {
      "epoch": 0.40166938028499055,
      "grad_norm": 561.7307739257812,
      "learning_rate": 3.013993770488318e-05,
      "loss": 58.3772,
      "step": 99420
    },
    {
      "epoch": 0.40170978155035814,
      "grad_norm": 954.3633422851562,
      "learning_rate": 3.0137530593474467e-05,
      "loss": 55.0279,
      "step": 99430
    },
    {
      "epoch": 0.4017501828157258,
      "grad_norm": 1188.335693359375,
      "learning_rate": 3.0135123284429366e-05,
      "loss": 70.7138,
      "step": 99440
    },
    {
      "epoch": 0.4017905840810934,
      "grad_norm": 947.3251953125,
      "learning_rate": 3.013271577779482e-05,
      "loss": 57.347,
      "step": 99450
    },
    {
      "epoch": 0.40183098534646106,
      "grad_norm": 872.726318359375,
      "learning_rate": 3.0130308073617765e-05,
      "loss": 59.0678,
      "step": 99460
    },
    {
      "epoch": 0.4018713866118287,
      "grad_norm": 1036.0936279296875,
      "learning_rate": 3.012790017194514e-05,
      "loss": 70.4954,
      "step": 99470
    },
    {
      "epoch": 0.40191178787719634,
      "grad_norm": 752.7295532226562,
      "learning_rate": 3.0125492072823884e-05,
      "loss": 35.581,
      "step": 99480
    },
    {
      "epoch": 0.4019521891425639,
      "grad_norm": 340.7840881347656,
      "learning_rate": 3.0123083776300946e-05,
      "loss": 63.2269,
      "step": 99490
    },
    {
      "epoch": 0.40199259040793156,
      "grad_norm": 818.9947509765625,
      "learning_rate": 3.0120675282423274e-05,
      "loss": 64.8064,
      "step": 99500
    },
    {
      "epoch": 0.4020329916732992,
      "grad_norm": 644.2744750976562,
      "learning_rate": 3.011826659123784e-05,
      "loss": 40.1466,
      "step": 99510
    },
    {
      "epoch": 0.40207339293866684,
      "grad_norm": 890.6390380859375,
      "learning_rate": 3.011585770279158e-05,
      "loss": 65.8333,
      "step": 99520
    },
    {
      "epoch": 0.4021137942040345,
      "grad_norm": 690.338623046875,
      "learning_rate": 3.011344861713147e-05,
      "loss": 69.3904,
      "step": 99530
    },
    {
      "epoch": 0.4021541954694021,
      "grad_norm": 902.1892700195312,
      "learning_rate": 3.0111039334304474e-05,
      "loss": 63.0215,
      "step": 99540
    },
    {
      "epoch": 0.40219459673476976,
      "grad_norm": 741.2757568359375,
      "learning_rate": 3.0108629854357557e-05,
      "loss": 68.3149,
      "step": 99550
    },
    {
      "epoch": 0.40223499800013734,
      "grad_norm": 670.7085571289062,
      "learning_rate": 3.0106220177337696e-05,
      "loss": 39.7456,
      "step": 99560
    },
    {
      "epoch": 0.402275399265505,
      "grad_norm": 1225.3446044921875,
      "learning_rate": 3.010381030329187e-05,
      "loss": 71.8467,
      "step": 99570
    },
    {
      "epoch": 0.4023158005308726,
      "grad_norm": 931.3955078125,
      "learning_rate": 3.010140023226706e-05,
      "loss": 48.1199,
      "step": 99580
    },
    {
      "epoch": 0.40235620179624026,
      "grad_norm": 948.512451171875,
      "learning_rate": 3.0098989964310254e-05,
      "loss": 54.7482,
      "step": 99590
    },
    {
      "epoch": 0.4023966030616079,
      "grad_norm": 883.859619140625,
      "learning_rate": 3.009657949946844e-05,
      "loss": 64.0414,
      "step": 99600
    },
    {
      "epoch": 0.40243700432697554,
      "grad_norm": 647.8932495117188,
      "learning_rate": 3.009416883778861e-05,
      "loss": 41.6642,
      "step": 99610
    },
    {
      "epoch": 0.40247740559234313,
      "grad_norm": 890.1572875976562,
      "learning_rate": 3.009175797931776e-05,
      "loss": 57.7219,
      "step": 99620
    },
    {
      "epoch": 0.40251780685771077,
      "grad_norm": 1720.4071044921875,
      "learning_rate": 3.0089346924102892e-05,
      "loss": 76.2947,
      "step": 99630
    },
    {
      "epoch": 0.4025582081230784,
      "grad_norm": 1014.928466796875,
      "learning_rate": 3.0086935672191012e-05,
      "loss": 60.2255,
      "step": 99640
    },
    {
      "epoch": 0.40259860938844605,
      "grad_norm": 1435.76611328125,
      "learning_rate": 3.008452422362913e-05,
      "loss": 53.2729,
      "step": 99650
    },
    {
      "epoch": 0.4026390106538137,
      "grad_norm": 1477.9378662109375,
      "learning_rate": 3.0082112578464252e-05,
      "loss": 62.3835,
      "step": 99660
    },
    {
      "epoch": 0.40267941191918133,
      "grad_norm": 1133.9884033203125,
      "learning_rate": 3.0079700736743406e-05,
      "loss": 64.5264,
      "step": 99670
    },
    {
      "epoch": 0.4027198131845489,
      "grad_norm": 529.5723876953125,
      "learning_rate": 3.0077288698513595e-05,
      "loss": 61.7479,
      "step": 99680
    },
    {
      "epoch": 0.40276021444991655,
      "grad_norm": 437.4736328125,
      "learning_rate": 3.0074876463821855e-05,
      "loss": 44.3355,
      "step": 99690
    },
    {
      "epoch": 0.4028006157152842,
      "grad_norm": 440.5912780761719,
      "learning_rate": 3.007246403271522e-05,
      "loss": 60.6817,
      "step": 99700
    },
    {
      "epoch": 0.40284101698065183,
      "grad_norm": 632.7706298828125,
      "learning_rate": 3.0070051405240712e-05,
      "loss": 50.7286,
      "step": 99710
    },
    {
      "epoch": 0.4028814182460195,
      "grad_norm": 480.6388244628906,
      "learning_rate": 3.006763858144536e-05,
      "loss": 41.7258,
      "step": 99720
    },
    {
      "epoch": 0.4029218195113871,
      "grad_norm": 668.8065795898438,
      "learning_rate": 3.006522556137621e-05,
      "loss": 89.2431,
      "step": 99730
    },
    {
      "epoch": 0.40296222077675475,
      "grad_norm": 831.2247924804688,
      "learning_rate": 3.006281234508031e-05,
      "loss": 64.0738,
      "step": 99740
    },
    {
      "epoch": 0.40300262204212234,
      "grad_norm": 710.2486572265625,
      "learning_rate": 3.0060398932604707e-05,
      "loss": 52.5265,
      "step": 99750
    },
    {
      "epoch": 0.40304302330749,
      "grad_norm": 569.1033325195312,
      "learning_rate": 3.0057985323996443e-05,
      "loss": 77.308,
      "step": 99760
    },
    {
      "epoch": 0.4030834245728576,
      "grad_norm": 1512.23828125,
      "learning_rate": 3.0055571519302572e-05,
      "loss": 46.1707,
      "step": 99770
    },
    {
      "epoch": 0.40312382583822526,
      "grad_norm": 615.7180786132812,
      "learning_rate": 3.0053157518570163e-05,
      "loss": 46.0806,
      "step": 99780
    },
    {
      "epoch": 0.4031642271035929,
      "grad_norm": 826.94580078125,
      "learning_rate": 3.0050743321846273e-05,
      "loss": 78.669,
      "step": 99790
    },
    {
      "epoch": 0.40320462836896054,
      "grad_norm": 1966.9554443359375,
      "learning_rate": 3.0048328929177963e-05,
      "loss": 89.0282,
      "step": 99800
    },
    {
      "epoch": 0.4032450296343281,
      "grad_norm": 836.5111694335938,
      "learning_rate": 3.0045914340612312e-05,
      "loss": 68.0784,
      "step": 99810
    },
    {
      "epoch": 0.40328543089969576,
      "grad_norm": 1213.034912109375,
      "learning_rate": 3.0043499556196384e-05,
      "loss": 51.5873,
      "step": 99820
    },
    {
      "epoch": 0.4033258321650634,
      "grad_norm": 1511.6014404296875,
      "learning_rate": 3.0041084575977267e-05,
      "loss": 54.6265,
      "step": 99830
    },
    {
      "epoch": 0.40336623343043104,
      "grad_norm": 698.375,
      "learning_rate": 3.0038669400002035e-05,
      "loss": 52.8981,
      "step": 99840
    },
    {
      "epoch": 0.4034066346957987,
      "grad_norm": 552.4383544921875,
      "learning_rate": 3.003625402831777e-05,
      "loss": 52.009,
      "step": 99850
    },
    {
      "epoch": 0.4034470359611663,
      "grad_norm": 708.4588623046875,
      "learning_rate": 3.003383846097157e-05,
      "loss": 56.2367,
      "step": 99860
    },
    {
      "epoch": 0.40348743722653396,
      "grad_norm": 606.9962158203125,
      "learning_rate": 3.0031422698010523e-05,
      "loss": 66.7383,
      "step": 99870
    },
    {
      "epoch": 0.40352783849190155,
      "grad_norm": 607.9722900390625,
      "learning_rate": 3.002900673948173e-05,
      "loss": 46.2622,
      "step": 99880
    },
    {
      "epoch": 0.4035682397572692,
      "grad_norm": 1990.047607421875,
      "learning_rate": 3.0026590585432277e-05,
      "loss": 67.8025,
      "step": 99890
    },
    {
      "epoch": 0.4036086410226368,
      "grad_norm": 401.2357482910156,
      "learning_rate": 3.0024174235909286e-05,
      "loss": 68.7798,
      "step": 99900
    },
    {
      "epoch": 0.40364904228800447,
      "grad_norm": 1142.0714111328125,
      "learning_rate": 3.0021757690959856e-05,
      "loss": 59.7364,
      "step": 99910
    },
    {
      "epoch": 0.4036894435533721,
      "grad_norm": 627.4801025390625,
      "learning_rate": 3.0019340950631103e-05,
      "loss": 95.0898,
      "step": 99920
    },
    {
      "epoch": 0.40372984481873975,
      "grad_norm": 6133.18408203125,
      "learning_rate": 3.0016924014970138e-05,
      "loss": 97.9595,
      "step": 99930
    },
    {
      "epoch": 0.40377024608410733,
      "grad_norm": 0.0,
      "learning_rate": 3.001450688402408e-05,
      "loss": 60.2756,
      "step": 99940
    },
    {
      "epoch": 0.40381064734947497,
      "grad_norm": 460.2475891113281,
      "learning_rate": 3.0012089557840056e-05,
      "loss": 56.6995,
      "step": 99950
    },
    {
      "epoch": 0.4038510486148426,
      "grad_norm": 1136.620361328125,
      "learning_rate": 3.000967203646519e-05,
      "loss": 58.4774,
      "step": 99960
    },
    {
      "epoch": 0.40389144988021025,
      "grad_norm": 561.9410400390625,
      "learning_rate": 3.000725431994662e-05,
      "loss": 56.8739,
      "step": 99970
    },
    {
      "epoch": 0.4039318511455779,
      "grad_norm": 1254.2598876953125,
      "learning_rate": 3.000483640833147e-05,
      "loss": 47.4478,
      "step": 99980
    },
    {
      "epoch": 0.40397225241094553,
      "grad_norm": 540.2157592773438,
      "learning_rate": 3.0002418301666886e-05,
      "loss": 56.2391,
      "step": 99990
    },
    {
      "epoch": 0.4040126536763131,
      "grad_norm": 1104.1842041015625,
      "learning_rate": 3.0000000000000004e-05,
      "loss": 74.0873,
      "step": 100000
    }
  ],
  "logging_steps": 10,
  "max_steps": 250000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 4000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}