{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.999990383783212,
  "eval_steps": 500,
  "global_step": 51995,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.9232433575982536e-05,
      "grad_norm": 94608.6885681578,
      "learning_rate": 3.846153846153847e-08,
      "loss": 12643.8086,
      "step": 1
    },
    {
      "epoch": 0.00019232433575982536,
      "grad_norm": 100578.99176905143,
      "learning_rate": 3.846153846153847e-07,
      "loss": 12553.6059,
      "step": 10
    },
    {
      "epoch": 0.0003846486715196507,
      "grad_norm": 88772.19286704408,
      "learning_rate": 7.692307692307694e-07,
      "loss": 11972.1172,
      "step": 20
    },
    {
      "epoch": 0.0005769730072794761,
      "grad_norm": 59360.15295432608,
      "learning_rate": 1.153846153846154e-06,
      "loss": 9967.2266,
      "step": 30
    },
    {
      "epoch": 0.0007692973430393014,
      "grad_norm": 28494.91588680842,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 7446.3711,
      "step": 40
    },
    {
      "epoch": 0.0009616216787991269,
      "grad_norm": 16366.768222931018,
      "learning_rate": 1.9230769230769234e-06,
      "loss": 5403.1641,
      "step": 50
    },
    {
      "epoch": 0.0011539460145589522,
      "grad_norm": 8839.161009449497,
      "learning_rate": 2.307692307692308e-06,
      "loss": 3814.0988,
      "step": 60
    },
    {
      "epoch": 0.0013462703503187776,
      "grad_norm": 5383.497342018133,
      "learning_rate": 2.6923076923076923e-06,
      "loss": 2743.5736,
      "step": 70
    },
    {
      "epoch": 0.001538594686078603,
      "grad_norm": 3725.296652871175,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 2219.7875,
      "step": 80
    },
    {
      "epoch": 0.0017309190218384282,
      "grad_norm": 3599.0985769624353,
      "learning_rate": 3.4615384615384617e-06,
      "loss": 1872.5768,
      "step": 90
    },
    {
      "epoch": 0.0019232433575982538,
      "grad_norm": 3094.252836745076,
      "learning_rate": 3.846153846153847e-06,
      "loss": 1735.4064,
      "step": 100
    },
    {
      "epoch": 0.002115567693358079,
      "grad_norm": 3300.1362331979553,
      "learning_rate": 4.230769230769231e-06,
      "loss": 1558.9357,
      "step": 110
    },
    {
      "epoch": 0.0023078920291179045,
      "grad_norm": 2104.1956388095073,
      "learning_rate": 4.615384615384616e-06,
      "loss": 1429.3346,
      "step": 120
    },
    {
      "epoch": 0.00250021636487773,
      "grad_norm": 2513.861867038951,
      "learning_rate": 5e-06,
      "loss": 1316.1369,
      "step": 130
    },
    {
      "epoch": 0.002692540700637555,
      "grad_norm": 3141.9987854701485,
      "learning_rate": 5.384615384615385e-06,
      "loss": 1260.3841,
      "step": 140
    },
    {
      "epoch": 0.0028848650363973807,
      "grad_norm": 1951.151761018987,
      "learning_rate": 5.769230769230769e-06,
      "loss": 1205.7934,
      "step": 150
    },
    {
      "epoch": 0.003077189372157206,
      "grad_norm": 1503.5840998796243,
      "learning_rate": 6.153846153846155e-06,
      "loss": 1149.7426,
      "step": 160
    },
    {
      "epoch": 0.0032695137079170314,
      "grad_norm": 1550.680002852119,
      "learning_rate": 6.538461538461539e-06,
      "loss": 1118.103,
      "step": 170
    },
    {
      "epoch": 0.0034618380436768565,
      "grad_norm": 1560.5468989439112,
      "learning_rate": 6.923076923076923e-06,
      "loss": 1092.9717,
      "step": 180
    },
    {
      "epoch": 0.003654162379436682,
      "grad_norm": 1425.4218406177279,
      "learning_rate": 7.307692307692308e-06,
      "loss": 1052.8212,
      "step": 190
    },
    {
      "epoch": 0.0038464867151965076,
      "grad_norm": 1516.6197391542155,
      "learning_rate": 7.692307692307694e-06,
      "loss": 1052.4535,
      "step": 200
    },
    {
      "epoch": 0.004038811050956333,
      "grad_norm": 1780.8347820865515,
      "learning_rate": 8.076923076923077e-06,
      "loss": 999.0068,
      "step": 210
    },
    {
      "epoch": 0.004231135386716158,
      "grad_norm": 1318.6042934766288,
      "learning_rate": 8.461538461538462e-06,
      "loss": 976.9977,
      "step": 220
    },
    {
      "epoch": 0.004423459722475984,
      "grad_norm": 1107.1960966004538,
      "learning_rate": 8.846153846153847e-06,
      "loss": 960.0191,
      "step": 230
    },
    {
      "epoch": 0.004615784058235809,
      "grad_norm": 1243.931843838465,
      "learning_rate": 9.230769230769232e-06,
      "loss": 950.4475,
      "step": 240
    },
    {
      "epoch": 0.004808108393995634,
      "grad_norm": 1603.148311812429,
      "learning_rate": 9.615384615384616e-06,
      "loss": 950.857,
      "step": 250
    },
    {
      "epoch": 0.00500043272975546,
      "grad_norm": 1234.584566075074,
      "learning_rate": 1e-05,
      "loss": 954.7617,
      "step": 260
    },
    {
      "epoch": 0.005192757065515285,
      "grad_norm": 1457.858865452839,
      "learning_rate": 1.0384615384615386e-05,
      "loss": 932.0896,
      "step": 270
    },
    {
      "epoch": 0.00538508140127511,
      "grad_norm": 1626.39266826436,
      "learning_rate": 1.076923076923077e-05,
      "loss": 900.2836,
      "step": 280
    },
    {
      "epoch": 0.005577405737034935,
      "grad_norm": 1341.1192889501265,
      "learning_rate": 1.1153846153846154e-05,
      "loss": 905.4902,
      "step": 290
    },
    {
      "epoch": 0.005769730072794761,
      "grad_norm": 2004.7690948696184,
      "learning_rate": 1.1538461538461538e-05,
      "loss": 872.6172,
      "step": 300
    },
    {
      "epoch": 0.0059620544085545865,
      "grad_norm": 1513.3944788118977,
      "learning_rate": 1.1923076923076925e-05,
      "loss": 869.792,
      "step": 310
    },
    {
      "epoch": 0.006154378744314412,
      "grad_norm": 1113.0910903993315,
      "learning_rate": 1.230769230769231e-05,
      "loss": 874.9273,
      "step": 320
    },
    {
      "epoch": 0.006346703080074238,
      "grad_norm": 3193.214792895934,
      "learning_rate": 1.2692307692307693e-05,
      "loss": 858.4352,
      "step": 330
    },
    {
      "epoch": 0.006539027415834063,
      "grad_norm": 1117.0052091046994,
      "learning_rate": 1.3076923076923078e-05,
      "loss": 843.8179,
      "step": 340
    },
    {
      "epoch": 0.006731351751593888,
      "grad_norm": 1068.3254606697517,
      "learning_rate": 1.3461538461538463e-05,
      "loss": 811.4118,
      "step": 350
    },
    {
      "epoch": 0.006923676087353713,
      "grad_norm": 1131.4811824271526,
      "learning_rate": 1.3846153846153847e-05,
      "loss": 823.7328,
      "step": 360
    },
    {
      "epoch": 0.007116000423113539,
      "grad_norm": 1245.6413691901007,
      "learning_rate": 1.4230769230769232e-05,
      "loss": 838.5344,
      "step": 370
    },
    {
      "epoch": 0.007308324758873364,
      "grad_norm": 1188.0536582603752,
      "learning_rate": 1.4615384615384615e-05,
      "loss": 833.5658,
      "step": 380
    },
    {
      "epoch": 0.007500649094633189,
      "grad_norm": 1033.2554609126928,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 816.3561,
      "step": 390
    },
    {
      "epoch": 0.007692973430393015,
      "grad_norm": 1106.619330781247,
      "learning_rate": 1.5384615384615387e-05,
      "loss": 799.7768,
      "step": 400
    },
    {
      "epoch": 0.00788529776615284,
      "grad_norm": 965.6234161186003,
      "learning_rate": 1.576923076923077e-05,
      "loss": 783.5753,
      "step": 410
    },
    {
      "epoch": 0.008077622101912665,
      "grad_norm": 1063.0371172029031,
      "learning_rate": 1.6153846153846154e-05,
      "loss": 807.2202,
      "step": 420
    },
    {
      "epoch": 0.00826994643767249,
      "grad_norm": 1401.141549219412,
      "learning_rate": 1.653846153846154e-05,
      "loss": 782.2932,
      "step": 430
    },
    {
      "epoch": 0.008462270773432316,
      "grad_norm": 961.9169300295785,
      "learning_rate": 1.6923076923076924e-05,
      "loss": 783.0447,
      "step": 440
    },
    {
      "epoch": 0.008654595109192142,
      "grad_norm": 1113.6035912659463,
      "learning_rate": 1.730769230769231e-05,
      "loss": 817.2932,
      "step": 450
    },
    {
      "epoch": 0.008846919444951968,
      "grad_norm": 1018.1114190982277,
      "learning_rate": 1.7692307692307694e-05,
      "loss": 768.6484,
      "step": 460
    },
    {
      "epoch": 0.009039243780711793,
      "grad_norm": 994.1764027298699,
      "learning_rate": 1.807692307692308e-05,
      "loss": 767.3088,
      "step": 470
    },
    {
      "epoch": 0.009231568116471618,
      "grad_norm": 1079.9442609599444,
      "learning_rate": 1.8461538461538465e-05,
      "loss": 746.3848,
      "step": 480
    },
    {
      "epoch": 0.009423892452231443,
      "grad_norm": 1132.6673906167728,
      "learning_rate": 1.8846153846153846e-05,
      "loss": 767.0798,
      "step": 490
    },
    {
      "epoch": 0.009616216787991268,
      "grad_norm": 972.2311975595254,
      "learning_rate": 1.923076923076923e-05,
      "loss": 750.1532,
      "step": 500
    },
    {
      "epoch": 0.009808541123751093,
      "grad_norm": 953.2034517617658,
      "learning_rate": 1.9615384615384617e-05,
      "loss": 765.4885,
      "step": 510
    },
    {
      "epoch": 0.01000086545951092,
      "grad_norm": 1080.8681332093806,
      "learning_rate": 2e-05,
      "loss": 762.509,
      "step": 520
    },
    {
      "epoch": 0.010193189795270745,
      "grad_norm": 1388.4857033156295,
      "learning_rate": 1.999999813758258e-05,
      "loss": 732.4688,
      "step": 530
    },
    {
      "epoch": 0.01038551413103057,
      "grad_norm": 1106.684533714725,
      "learning_rate": 1.9999992550331007e-05,
      "loss": 740.4309,
      "step": 540
    },
    {
      "epoch": 0.010577838466790395,
      "grad_norm": 1090.5369324499081,
      "learning_rate": 1.999998323824736e-05,
      "loss": 739.5119,
      "step": 550
    },
    {
      "epoch": 0.01077016280255022,
      "grad_norm": 1129.680526403887,
      "learning_rate": 1.999997020133512e-05,
      "loss": 732.7231,
      "step": 560
    },
    {
      "epoch": 0.010962487138310046,
      "grad_norm": 930.7364821135466,
      "learning_rate": 1.9999953439599132e-05,
      "loss": 733.2015,
      "step": 570
    },
    {
      "epoch": 0.01115481147406987,
      "grad_norm": 933.4120343343305,
      "learning_rate": 1.9999932953045638e-05,
      "loss": 737.304,
      "step": 580
    },
    {
      "epoch": 0.011347135809829698,
      "grad_norm": 915.0633771476615,
      "learning_rate": 1.999990874168228e-05,
      "loss": 724.2898,
      "step": 590
    },
    {
      "epoch": 0.011539460145589523,
      "grad_norm": 1128.361341996108,
      "learning_rate": 1.9999880805518067e-05,
      "loss": 735.0713,
      "step": 600
    },
    {
      "epoch": 0.011731784481349348,
      "grad_norm": 849.0208141614002,
      "learning_rate": 1.9999849144563406e-05,
      "loss": 720.8895,
      "step": 610
    },
    {
      "epoch": 0.011924108817109173,
      "grad_norm": 1015.5962484886528,
      "learning_rate": 1.9999813758830092e-05,
      "loss": 713.2033,
      "step": 620
    },
    {
      "epoch": 0.012116433152868998,
      "grad_norm": 1026.2741464178252,
      "learning_rate": 1.9999774648331307e-05,
      "loss": 716.7091,
      "step": 630
    },
    {
      "epoch": 0.012308757488628823,
      "grad_norm": 969.4774400659321,
      "learning_rate": 1.9999731813081616e-05,
      "loss": 718.2322,
      "step": 640
    },
    {
      "epoch": 0.012501081824388648,
      "grad_norm": 967.3757615945976,
      "learning_rate": 1.9999685253096975e-05,
      "loss": 710.2178,
      "step": 650
    },
    {
      "epoch": 0.012693406160148475,
      "grad_norm": 871.670618965462,
      "learning_rate": 1.999963496839473e-05,
      "loss": 688.6732,
      "step": 660
    },
    {
      "epoch": 0.0128857304959083,
      "grad_norm": 918.4170497435479,
      "learning_rate": 1.9999580958993606e-05,
      "loss": 692.1547,
      "step": 670
    },
    {
      "epoch": 0.013078054831668125,
      "grad_norm": 957.1468754682305,
      "learning_rate": 1.9999523224913722e-05,
      "loss": 714.7746,
      "step": 680
    },
    {
      "epoch": 0.01327037916742795,
      "grad_norm": 865.7781073300224,
      "learning_rate": 1.999946176617659e-05,
      "loss": 705.0376,
      "step": 690
    },
    {
      "epoch": 0.013462703503187776,
      "grad_norm": 936.9998829159986,
      "learning_rate": 1.999939658280509e-05,
      "loss": 687.2993,
      "step": 700
    },
    {
      "epoch": 0.0136550278389476,
      "grad_norm": 906.7755440766291,
      "learning_rate": 1.9999327674823513e-05,
      "loss": 706.7441,
      "step": 710
    },
    {
      "epoch": 0.013847352174707426,
      "grad_norm": 856.9815132697244,
      "learning_rate": 1.9999255042257522e-05,
      "loss": 697.0999,
      "step": 720
    },
    {
      "epoch": 0.014039676510467253,
      "grad_norm": 927.2050883683178,
      "learning_rate": 1.999917868513417e-05,
      "loss": 681.3204,
      "step": 730
    },
    {
      "epoch": 0.014232000846227078,
      "grad_norm": 922.0296694917098,
      "learning_rate": 1.9999098603481895e-05,
      "loss": 662.6776,
      "step": 740
    },
    {
      "epoch": 0.014424325181986903,
      "grad_norm": 959.1175136904378,
      "learning_rate": 1.9999014797330536e-05,
      "loss": 681.5562,
      "step": 750
    },
    {
      "epoch": 0.014616649517746728,
      "grad_norm": 816.3936907296245,
      "learning_rate": 1.99989272667113e-05,
      "loss": 685.389,
      "step": 760
    },
    {
      "epoch": 0.014808973853506553,
      "grad_norm": 874.4326400369638,
      "learning_rate": 1.99988360116568e-05,
      "loss": 678.6024,
      "step": 770
    },
    {
      "epoch": 0.015001298189266378,
      "grad_norm": 798.9969239141585,
      "learning_rate": 1.999874103220102e-05,
      "loss": 669.6026,
      "step": 780
    },
    {
      "epoch": 0.015193622525026203,
      "grad_norm": 904.5693607601238,
      "learning_rate": 1.999864232837934e-05,
      "loss": 676.1861,
      "step": 790
    },
    {
      "epoch": 0.01538594686078603,
      "grad_norm": 1098.9356397046763,
      "learning_rate": 1.9998539900228526e-05,
      "loss": 677.7401,
      "step": 800
    },
    {
      "epoch": 0.015578271196545855,
      "grad_norm": 838.3649701940624,
      "learning_rate": 1.999843374778673e-05,
      "loss": 675.1596,
      "step": 810
    },
    {
      "epoch": 0.01577059553230568,
      "grad_norm": 910.3491848511147,
      "learning_rate": 1.999832387109349e-05,
      "loss": 661.2108,
      "step": 820
    },
    {
      "epoch": 0.015962919868065507,
      "grad_norm": 839.7251367189293,
      "learning_rate": 1.9998210270189736e-05,
      "loss": 653.5074,
      "step": 830
    },
    {
      "epoch": 0.01615524420382533,
      "grad_norm": 922.4346859381994,
      "learning_rate": 1.9998092945117786e-05,
      "loss": 652.9832,
      "step": 840
    },
    {
      "epoch": 0.016347568539585158,
      "grad_norm": 797.9334887512913,
      "learning_rate": 1.999797189592134e-05,
      "loss": 658.8486,
      "step": 850
    },
    {
      "epoch": 0.01653989287534498,
      "grad_norm": 26733.46203732987,
      "learning_rate": 1.999784712264548e-05,
      "loss": 698.5595,
      "step": 860
    },
    {
      "epoch": 0.016732217211104808,
      "grad_norm": 828.643285218162,
      "learning_rate": 1.9997718625336686e-05,
      "loss": 651.8198,
      "step": 870
    },
    {
      "epoch": 0.01692454154686463,
      "grad_norm": 898.1396587421336,
      "learning_rate": 1.9997586404042825e-05,
      "loss": 671.2599,
      "step": 880
    },
    {
      "epoch": 0.017116865882624458,
      "grad_norm": 802.1362445755046,
      "learning_rate": 1.9997450458813142e-05,
      "loss": 645.942,
      "step": 890
    },
    {
      "epoch": 0.017309190218384285,
      "grad_norm": 1016.8188200039426,
      "learning_rate": 1.9997310789698276e-05,
      "loss": 639.8075,
      "step": 900
    },
    {
      "epoch": 0.01750151455414411,
      "grad_norm": 821.0411601718398,
      "learning_rate": 1.999716739675025e-05,
      "loss": 655.8576,
      "step": 910
    },
    {
      "epoch": 0.017693838889903935,
      "grad_norm": 1936.8704555826098,
      "learning_rate": 1.9997020280022482e-05,
      "loss": 647.9312,
      "step": 920
    },
    {
      "epoch": 0.01788616322566376,
      "grad_norm": 1018.8993638712327,
      "learning_rate": 1.9996869439569767e-05,
      "loss": 638.7674,
      "step": 930
    },
    {
      "epoch": 0.018078487561423585,
      "grad_norm": 904.9237018635373,
      "learning_rate": 1.9996714875448286e-05,
      "loss": 641.0137,
      "step": 940
    },
    {
      "epoch": 0.01827081189718341,
      "grad_norm": 1003.3093857172769,
      "learning_rate": 1.9996556587715617e-05,
      "loss": 649.2616,
      "step": 950
    },
    {
      "epoch": 0.018463136232943236,
      "grad_norm": 960.2339889734326,
      "learning_rate": 1.9996394576430716e-05,
      "loss": 645.8987,
      "step": 960
    },
    {
      "epoch": 0.018655460568703063,
      "grad_norm": 836.5678203183564,
      "learning_rate": 1.9996228841653932e-05,
      "loss": 647.2,
      "step": 970
    },
    {
      "epoch": 0.018847784904462886,
      "grad_norm": 798.1541783785667,
      "learning_rate": 1.9996059383447e-05,
      "loss": 632.6011,
      "step": 980
    },
    {
      "epoch": 0.019040109240222713,
      "grad_norm": 1001.9244344517015,
      "learning_rate": 1.999588620187303e-05,
      "loss": 637.499,
      "step": 990
    },
    {
      "epoch": 0.019232433575982536,
      "grad_norm": 927.656831492326,
      "learning_rate": 1.9995709296996545e-05,
      "loss": 644.8208,
      "step": 1000
    },
    {
      "epoch": 0.019424757911742363,
      "grad_norm": 767.8154780334495,
      "learning_rate": 1.999552866888343e-05,
      "loss": 642.4208,
      "step": 1010
    },
    {
      "epoch": 0.019617082247502186,
      "grad_norm": 761.3185760521654,
      "learning_rate": 1.9995344317600965e-05,
      "loss": 652.1784,
      "step": 1020
    },
    {
      "epoch": 0.019809406583262013,
      "grad_norm": 1340.2077546097548,
      "learning_rate": 1.9995156243217824e-05,
      "loss": 641.2602,
      "step": 1030
    },
    {
      "epoch": 0.02000173091902184,
      "grad_norm": 846.3841401165058,
      "learning_rate": 1.9994964445804054e-05,
      "loss": 656.1286,
      "step": 1040
    },
    {
      "epoch": 0.020194055254781663,
      "grad_norm": 806.7606160265955,
      "learning_rate": 1.9994768925431104e-05,
      "loss": 634.8761,
      "step": 1050
    },
    {
      "epoch": 0.02038637959054149,
      "grad_norm": 707.4507239896416,
      "learning_rate": 1.9994569682171796e-05,
      "loss": 628.0639,
      "step": 1060
    },
    {
      "epoch": 0.020578703926301314,
      "grad_norm": 756.2349192400593,
      "learning_rate": 1.9994366716100346e-05,
      "loss": 631.8284,
      "step": 1070
    },
    {
      "epoch": 0.02077102826206114,
      "grad_norm": 891.7453215514997,
      "learning_rate": 1.999416002729236e-05,
      "loss": 629.0493,
      "step": 1080
    },
    {
      "epoch": 0.020963352597820964,
      "grad_norm": 908.0499458662863,
      "learning_rate": 1.999394961582482e-05,
      "loss": 645.8078,
      "step": 1090
    },
    {
      "epoch": 0.02115567693358079,
      "grad_norm": 858.8921391277137,
      "learning_rate": 1.9993735481776105e-05,
      "loss": 613.4529,
      "step": 1100
    },
    {
      "epoch": 0.021348001269340618,
      "grad_norm": 1634.7873260446481,
      "learning_rate": 1.9993517625225976e-05,
      "loss": 617.2008,
      "step": 1110
    },
    {
      "epoch": 0.02154032560510044,
      "grad_norm": 820.0560949206612,
      "learning_rate": 1.9993296046255578e-05,
      "loss": 608.7404,
      "step": 1120
    },
    {
      "epoch": 0.021732649940860268,
      "grad_norm": 804.3695313812165,
      "learning_rate": 1.9993070744947447e-05,
      "loss": 626.9961,
      "step": 1130
    },
    {
      "epoch": 0.02192497427662009,
      "grad_norm": 1114.8561548281725,
      "learning_rate": 1.9992841721385508e-05,
      "loss": 611.4933,
      "step": 1140
    },
    {
      "epoch": 0.022117298612379918,
      "grad_norm": 3419.5258436900294,
      "learning_rate": 1.999260897565506e-05,
      "loss": 614.8227,
      "step": 1150
    },
    {
      "epoch": 0.02230962294813974,
      "grad_norm": 1076.6006001682858,
      "learning_rate": 1.9992372507842807e-05,
      "loss": 627.309,
      "step": 1160
    },
    {
      "epoch": 0.02250194728389957,
      "grad_norm": 732.8498874304387,
      "learning_rate": 1.9992132318036825e-05,
      "loss": 615.4989,
      "step": 1170
    },
    {
      "epoch": 0.022694271619659395,
      "grad_norm": 995.8140259128272,
      "learning_rate": 1.9991888406326575e-05,
      "loss": 630.0663,
      "step": 1180
    },
    {
      "epoch": 0.02288659595541922,
      "grad_norm": 763.5901984809835,
      "learning_rate": 1.9991640772802916e-05,
      "loss": 606.8447,
      "step": 1190
    },
    {
      "epoch": 0.023078920291179045,
      "grad_norm": 727.5466510360628,
      "learning_rate": 1.9991389417558088e-05,
      "loss": 612.981,
      "step": 1200
    },
    {
      "epoch": 0.02327124462693887,
      "grad_norm": 821.0411142467469,
      "learning_rate": 1.9991134340685713e-05,
      "loss": 602.0013,
      "step": 1210
    },
    {
      "epoch": 0.023463568962698696,
      "grad_norm": 840.802859921454,
      "learning_rate": 1.999087554228081e-05,
      "loss": 598.6377,
      "step": 1220
    },
    {
      "epoch": 0.02365589329845852,
      "grad_norm": 756.3177316860919,
      "learning_rate": 1.999061302243977e-05,
      "loss": 612.0126,
      "step": 1230
    },
    {
      "epoch": 0.023848217634218346,
      "grad_norm": 1195.0447257910407,
      "learning_rate": 1.9990346781260378e-05,
      "loss": 616.7188,
      "step": 1240
    },
    {
      "epoch": 0.024040541969978173,
      "grad_norm": 922.9388146082537,
      "learning_rate": 1.9990076818841805e-05,
      "loss": 596.7434,
      "step": 1250
    },
    {
      "epoch": 0.024232866305737996,
      "grad_norm": 700.6416809095787,
      "learning_rate": 1.998980313528461e-05,
      "loss": 609.2294,
      "step": 1260
    },
    {
      "epoch": 0.024425190641497823,
      "grad_norm": 1257.6949867941905,
      "learning_rate": 1.9989525730690736e-05,
      "loss": 609.9669,
      "step": 1270
    },
    {
      "epoch": 0.024617514977257646,
      "grad_norm": 903.549613148072,
      "learning_rate": 1.998924460516351e-05,
      "loss": 620.5952,
      "step": 1280
    },
    {
      "epoch": 0.024809839313017473,
      "grad_norm": 764.6321620873817,
      "learning_rate": 1.9988959758807645e-05,
      "loss": 614.1899,
      "step": 1290
    },
    {
      "epoch": 0.025002163648777297,
      "grad_norm": 778.3259603306221,
      "learning_rate": 1.9988671191729243e-05,
      "loss": 624.8979,
      "step": 1300
    },
    {
      "epoch": 0.025194487984537123,
      "grad_norm": 1005.8223323049509,
      "learning_rate": 1.998837890403579e-05,
      "loss": 615.0917,
      "step": 1310
    },
    {
      "epoch": 0.02538681232029695,
      "grad_norm": 1440.4206048755198,
      "learning_rate": 1.998808289583616e-05,
      "loss": 604.339,
      "step": 1320
    },
    {
      "epoch": 0.025579136656056774,
      "grad_norm": 999.4877466798821,
      "learning_rate": 1.998778316724061e-05,
      "loss": 595.0511,
      "step": 1330
    },
    {
      "epoch": 0.0257714609918166,
      "grad_norm": 856.5499567850246,
      "learning_rate": 1.998747971836078e-05,
      "loss": 582.4183,
      "step": 1340
    },
    {
      "epoch": 0.025963785327576424,
      "grad_norm": 771.1349971770352,
      "learning_rate": 1.9987172549309707e-05,
      "loss": 608.028,
      "step": 1350
    },
    {
      "epoch": 0.02615610966333625,
      "grad_norm": 855.138500276869,
      "learning_rate": 1.9986861660201802e-05,
      "loss": 596.8151,
      "step": 1360
    },
    {
      "epoch": 0.026348433999096074,
      "grad_norm": 792.8188188026194,
      "learning_rate": 1.998654705115287e-05,
      "loss": 623.3355,
      "step": 1370
    },
    {
      "epoch": 0.0265407583348559,
      "grad_norm": 717.0052410633494,
      "learning_rate": 1.9986228722280093e-05,
      "loss": 615.0966,
      "step": 1380
    },
    {
      "epoch": 0.026733082670615728,
      "grad_norm": 743.0989252905181,
      "learning_rate": 1.998590667370204e-05,
      "loss": 595.9803,
      "step": 1390
    },
    {
      "epoch": 0.02692540700637555,
      "grad_norm": 760.5612612863387,
      "learning_rate": 1.998558090553868e-05,
      "loss": 609.7504,
      "step": 1400
    },
    {
      "epoch": 0.027117731342135378,
      "grad_norm": 775.1687306616708,
      "learning_rate": 1.9985251417911347e-05,
      "loss": 593.0434,
      "step": 1410
    },
    {
      "epoch": 0.0273100556778952,
      "grad_norm": 741.3442683045452,
      "learning_rate": 1.9984918210942776e-05,
      "loss": 585.3288,
      "step": 1420
    },
    {
      "epoch": 0.02750238001365503,
      "grad_norm": 730.328436539225,
      "learning_rate": 1.9984581284757074e-05,
      "loss": 596.1948,
      "step": 1430
    },
    {
      "epoch": 0.027694704349414852,
      "grad_norm": 698.1410628688395,
      "learning_rate": 1.9984240639479745e-05,
      "loss": 589.9968,
      "step": 1440
    },
    {
      "epoch": 0.02788702868517468,
      "grad_norm": 886.6719463044703,
      "learning_rate": 1.9983896275237677e-05,
      "loss": 586.9773,
      "step": 1450
    },
    {
      "epoch": 0.028079353020934505,
      "grad_norm": 848.7504027618456,
      "learning_rate": 1.9983548192159132e-05,
      "loss": 587.161,
      "step": 1460
    },
    {
      "epoch": 0.02827167735669433,
      "grad_norm": 796.0128772973629,
      "learning_rate": 1.998319639037377e-05,
      "loss": 598.3176,
      "step": 1470
    },
    {
      "epoch": 0.028464001692454156,
      "grad_norm": 932.687167624479,
      "learning_rate": 1.9982840870012626e-05,
      "loss": 589.4393,
      "step": 1480
    },
    {
      "epoch": 0.02865632602821398,
      "grad_norm": 769.0697081505059,
      "learning_rate": 1.9982481631208138e-05,
      "loss": 582.1562,
      "step": 1490
    },
    {
      "epoch": 0.028848650363973806,
      "grad_norm": 927.1602634645985,
      "learning_rate": 1.9982118674094104e-05,
      "loss": 587.0485,
      "step": 1500
    },
    {
      "epoch": 0.02904097469973363,
      "grad_norm": 827.0889935686532,
      "learning_rate": 1.9981751998805725e-05,
      "loss": 581.6353,
      "step": 1510
    },
    {
      "epoch": 0.029233299035493456,
      "grad_norm": 996.163060843575,
      "learning_rate": 1.998138160547958e-05,
      "loss": 590.505,
      "step": 1520
    },
    {
      "epoch": 0.029425623371253283,
      "grad_norm": 914.838305631208,
      "learning_rate": 1.9981007494253638e-05,
      "loss": 588.7489,
      "step": 1530
    },
    {
      "epoch": 0.029617947707013106,
      "grad_norm": 911.7105459419249,
      "learning_rate": 1.998062966526724e-05,
      "loss": 594.3789,
      "step": 1540
    },
    {
      "epoch": 0.029810272042772933,
      "grad_norm": 705.4340084272872,
      "learning_rate": 1.998024811866113e-05,
      "loss": 593.7086,
      "step": 1550
    },
    {
      "epoch": 0.030002596378532757,
      "grad_norm": 861.251850051679,
      "learning_rate": 1.9979862854577427e-05,
      "loss": 597.0841,
      "step": 1560
    },
    {
      "epoch": 0.030194920714292584,
      "grad_norm": 750.188392481234,
      "learning_rate": 1.9979473873159635e-05,
      "loss": 584.6696,
      "step": 1570
    },
    {
      "epoch": 0.030387245050052407,
      "grad_norm": 757.1086820583647,
      "learning_rate": 1.9979081174552638e-05,
      "loss": 594.2866,
      "step": 1580
    },
    {
      "epoch": 0.030579569385812234,
      "grad_norm": 760.443045425971,
      "learning_rate": 1.997868475890271e-05,
      "loss": 589.7756,
      "step": 1590
    },
    {
      "epoch": 0.03077189372157206,
      "grad_norm": 760.888452583621,
      "learning_rate": 1.997828462635752e-05,
      "loss": 575.5263,
      "step": 1600
    },
    {
      "epoch": 0.030964218057331884,
      "grad_norm": 939.2634714082963,
      "learning_rate": 1.99778807770661e-05,
      "loss": 578.2944,
      "step": 1610
    },
    {
      "epoch": 0.03115654239309171,
      "grad_norm": 817.6277925232175,
      "learning_rate": 1.9977473211178882e-05,
      "loss": 593.5157,
      "step": 1620
    },
    {
      "epoch": 0.03134886672885154,
      "grad_norm": 736.0775460201876,
      "learning_rate": 1.9977061928847676e-05,
      "loss": 586.7161,
      "step": 1630
    },
    {
      "epoch": 0.03154119106461136,
      "grad_norm": 711.1986925625198,
      "learning_rate": 1.9976646930225678e-05,
      "loss": 592.29,
      "step": 1640
    },
    {
      "epoch": 0.031733515400371184,
      "grad_norm": 876.663942507758,
      "learning_rate": 1.997622821546747e-05,
      "loss": 580.1834,
      "step": 1650
    },
    {
      "epoch": 0.031925839736131015,
      "grad_norm": 754.3849455814952,
      "learning_rate": 1.9975805784729008e-05,
      "loss": 581.797,
      "step": 1660
    },
    {
      "epoch": 0.03211816407189084,
      "grad_norm": 781.673019472033,
      "learning_rate": 1.9975379638167654e-05,
      "loss": 620.744,
      "step": 1670
    },
    {
      "epoch": 0.03231048840765066,
      "grad_norm": 692.1175529132593,
      "learning_rate": 1.9974949775942134e-05,
      "loss": 564.507,
      "step": 1680
    },
    {
      "epoch": 0.032502812743410485,
      "grad_norm": 731.9105990335349,
      "learning_rate": 1.997451619821256e-05,
      "loss": 589.0906,
      "step": 1690
    },
    {
      "epoch": 0.032695137079170315,
      "grad_norm": 856.315165121257,
      "learning_rate": 1.997407890514044e-05,
      "loss": 576.5989,
      "step": 1700
    },
    {
      "epoch": 0.03288746141493014,
      "grad_norm": 797.0996376202884,
      "learning_rate": 1.9973637896888652e-05,
      "loss": 564.9523,
      "step": 1710
    },
    {
      "epoch": 0.03307978575068996,
      "grad_norm": 682.1407433412485,
      "learning_rate": 1.997319317362147e-05,
      "loss": 571.1717,
      "step": 1720
    },
    {
      "epoch": 0.03327211008644979,
      "grad_norm": 722.5978016546433,
      "learning_rate": 1.9972744735504542e-05,
      "loss": 557.932,
      "step": 1730
    },
    {
      "epoch": 0.033464434422209616,
      "grad_norm": 1066.4776783343032,
      "learning_rate": 1.9972292582704905e-05,
      "loss": 583.6356,
      "step": 1740
    },
    {
      "epoch": 0.03365675875796944,
      "grad_norm": 1091.442768853797,
      "learning_rate": 1.997183671539098e-05,
      "loss": 575.6317,
      "step": 1750
    },
    {
      "epoch": 0.03384908309372926,
      "grad_norm": 690.316815081678,
      "learning_rate": 1.9971377133732567e-05,
      "loss": 570.6772,
      "step": 1760
    },
    {
      "epoch": 0.03404140742948909,
      "grad_norm": 671.2236932546693,
      "learning_rate": 1.997091383790086e-05,
      "loss": 582.4128,
      "step": 1770
    },
    {
      "epoch": 0.034233731765248916,
      "grad_norm": 757.1943605183349,
      "learning_rate": 1.997044682806842e-05,
      "loss": 558.5421,
      "step": 1780
    },
    {
      "epoch": 0.03442605610100874,
      "grad_norm": 753.0373151865076,
      "learning_rate": 1.9969976104409202e-05,
      "loss": 576.9115,
      "step": 1790
    },
    {
      "epoch": 0.03461838043676857,
      "grad_norm": 815.0380471542311,
      "learning_rate": 1.9969501667098547e-05,
      "loss": 582.4467,
      "step": 1800
    },
    {
      "epoch": 0.03481070477252839,
      "grad_norm": 829.7918441384254,
      "learning_rate": 1.996902351631317e-05,
      "loss": 561.6938,
      "step": 1810
    },
    {
      "epoch": 0.03500302910828822,
      "grad_norm": 963.6197521916137,
      "learning_rate": 1.996854165223118e-05,
      "loss": 560.3726,
      "step": 1820
    },
    {
      "epoch": 0.03519535344404804,
      "grad_norm": 796.0395944595921,
      "learning_rate": 1.9968056075032058e-05,
      "loss": 560.0353,
      "step": 1830
    },
    {
      "epoch": 0.03538767777980787,
      "grad_norm": 694.7128512741667,
      "learning_rate": 1.9967566784896676e-05,
      "loss": 557.3366,
      "step": 1840
    },
    {
      "epoch": 0.035580002115567694,
      "grad_norm": 692.2461914066811,
      "learning_rate": 1.996707378200729e-05,
      "loss": 547.8145,
      "step": 1850
    },
    {
      "epoch": 0.03577232645132752,
      "grad_norm": 1022.0757647180731,
      "learning_rate": 1.9966577066547526e-05,
      "loss": 556.1324,
      "step": 1860
    },
    {
      "epoch": 0.03596465078708735,
      "grad_norm": 754.5291255951288,
      "learning_rate": 1.9966076638702412e-05,
      "loss": 567.7352,
      "step": 1870
    },
    {
      "epoch": 0.03615697512284717,
      "grad_norm": 658.4566590087529,
      "learning_rate": 1.9965572498658346e-05,
      "loss": 564.1984,
      "step": 1880
    },
    {
      "epoch": 0.036349299458606994,
      "grad_norm": 703.9357386627887,
      "learning_rate": 1.996506464660311e-05,
      "loss": 562.5533,
      "step": 1890
    },
    {
      "epoch": 0.03654162379436682,
      "grad_norm": 1354.1007956418598,
      "learning_rate": 1.9964553082725873e-05,
      "loss": 550.8262,
      "step": 1900
    },
    {
      "epoch": 0.03673394813012665,
      "grad_norm": 965.1647425113485,
      "learning_rate": 1.996403780721718e-05,
      "loss": 557.0399,
      "step": 1910
    },
    {
      "epoch": 0.03692627246588647,
      "grad_norm": 856.3895064371682,
      "learning_rate": 1.9963518820268968e-05,
      "loss": 568.2071,
      "step": 1920
    },
    {
      "epoch": 0.037118596801646295,
      "grad_norm": 726.6624727792246,
      "learning_rate": 1.996299612207455e-05,
      "loss": 552.3866,
      "step": 1930
    },
    {
      "epoch": 0.037310921137406125,
      "grad_norm": 833.5772157518286,
      "learning_rate": 1.9962469712828613e-05,
      "loss": 571.4761,
      "step": 1940
    },
    {
      "epoch": 0.03750324547316595,
      "grad_norm": 767.736432325189,
      "learning_rate": 1.996193959272725e-05,
      "loss": 543.1805,
      "step": 1950
    },
    {
      "epoch": 0.03769556980892577,
      "grad_norm": 716.1463033104436,
      "learning_rate": 1.9961405761967914e-05,
      "loss": 558.2827,
      "step": 1960
    },
    {
      "epoch": 0.037887894144685595,
      "grad_norm": 760.6145737591669,
      "learning_rate": 1.996086822074945e-05,
      "loss": 551.1155,
      "step": 1970
    },
    {
      "epoch": 0.038080218480445426,
      "grad_norm": 780.5334606807841,
      "learning_rate": 1.996032696927208e-05,
      "loss": 556.977,
      "step": 1980
    },
    {
      "epoch": 0.03827254281620525,
      "grad_norm": 1030.3170317696865,
      "learning_rate": 1.9959782007737418e-05,
      "loss": 556.1245,
      "step": 1990
    },
    {
      "epoch": 0.03846486715196507,
      "grad_norm": 826.3580099606195,
      "learning_rate": 1.9959233336348452e-05,
      "loss": 541.4277,
      "step": 2000
    },
    {
      "epoch": 0.0386571914877249,
      "grad_norm": 693.1981882140224,
      "learning_rate": 1.9958680955309546e-05,
      "loss": 560.1344,
      "step": 2010
    },
    {
      "epoch": 0.038849515823484726,
      "grad_norm": 1055.8601947595325,
      "learning_rate": 1.9958124864826457e-05,
      "loss": 558.6848,
      "step": 2020
    },
    {
      "epoch": 0.03904184015924455,
      "grad_norm": 748.4590657616682,
      "learning_rate": 1.9957565065106318e-05,
      "loss": 542.2576,
      "step": 2030
    },
    {
      "epoch": 0.03923416449500437,
      "grad_norm": 668.2717036729244,
      "learning_rate": 1.9957001556357652e-05,
      "loss": 547.7366,
      "step": 2040
    },
    {
      "epoch": 0.0394264888307642,
      "grad_norm": 771.5865978323359,
      "learning_rate": 1.995643433879035e-05,
      "loss": 538.1941,
      "step": 2050
    },
    {
      "epoch": 0.039618813166524026,
      "grad_norm": 677.3405758876926,
      "learning_rate": 1.9955863412615693e-05,
      "loss": 550.9306,
      "step": 2060
    },
    {
      "epoch": 0.03981113750228385,
      "grad_norm": 743.9922405719423,
      "learning_rate": 1.9955288778046338e-05,
      "loss": 555.3232,
      "step": 2070
    },
    {
      "epoch": 0.04000346183804368,
      "grad_norm": 698.9378102470318,
      "learning_rate": 1.995471043529633e-05,
      "loss": 550.2594,
      "step": 2080
    },
    {
      "epoch": 0.040195786173803504,
      "grad_norm": 941.1099281364698,
      "learning_rate": 1.99541283845811e-05,
      "loss": 551.2195,
      "step": 2090
    },
    {
      "epoch": 0.04038811050956333,
      "grad_norm": 758.8590943012098,
      "learning_rate": 1.9953542626117437e-05,
      "loss": 553.01,
      "step": 2100
    },
    {
      "epoch": 0.04058043484532315,
      "grad_norm": 728.6791740588354,
      "learning_rate": 1.9952953160123537e-05,
      "loss": 560.2909,
      "step": 2110
    },
    {
      "epoch": 0.04077275918108298,
      "grad_norm": 704.9364163187533,
      "learning_rate": 1.995235998681896e-05,
      "loss": 551.6747,
      "step": 2120
    },
    {
      "epoch": 0.040965083516842804,
      "grad_norm": 867.618758450042,
      "learning_rate": 1.9951763106424658e-05,
      "loss": 552.1828,
      "step": 2130
    },
    {
      "epoch": 0.04115740785260263,
      "grad_norm": 671.4482081969627,
      "learning_rate": 1.9951162519162962e-05,
      "loss": 546.2693,
      "step": 2140
    },
    {
      "epoch": 0.04134973218836246,
      "grad_norm": 681.6085942169542,
      "learning_rate": 1.9950558225257574e-05,
      "loss": 544.3579,
      "step": 2150
    },
    {
      "epoch": 0.04154205652412228,
      "grad_norm": 659.0744185297452,
      "learning_rate": 1.9949950224933584e-05,
      "loss": 545.4658,
      "step": 2160
    },
    {
      "epoch": 0.041734380859882104,
      "grad_norm": 669.992502636515,
      "learning_rate": 1.994933851841747e-05,
      "loss": 545.2745,
      "step": 2170
    },
    {
      "epoch": 0.04192670519564193,
      "grad_norm": 738.3814003175634,
      "learning_rate": 1.994872310593707e-05,
      "loss": 547.8611,
      "step": 2180
    },
    {
      "epoch": 0.04211902953140176,
      "grad_norm": 771.670011949177,
      "learning_rate": 1.994810398772162e-05,
      "loss": 549.3855,
      "step": 2190
    },
    {
      "epoch": 0.04231135386716158,
      "grad_norm": 739.3839708115137,
      "learning_rate": 1.994748116400174e-05,
      "loss": 539.5586,
      "step": 2200
    },
    {
      "epoch": 0.042503678202921405,
      "grad_norm": 756.7255259576679,
      "learning_rate": 1.994685463500941e-05,
      "loss": 548.5997,
      "step": 2210
    },
    {
      "epoch": 0.042696002538681235,
      "grad_norm": 797.4589123090307,
      "learning_rate": 1.9946224400978006e-05,
      "loss": 540.3652,
      "step": 2220
    },
    {
      "epoch": 0.04288832687444106,
      "grad_norm": 1007.0318957752902,
      "learning_rate": 1.994559046214228e-05,
      "loss": 541.5651,
      "step": 2230
    },
    {
      "epoch": 0.04308065121020088,
      "grad_norm": 754.2131677595725,
      "learning_rate": 1.9944952818738366e-05,
      "loss": 555.0396,
      "step": 2240
    },
    {
      "epoch": 0.043272975545960705,
      "grad_norm": 842.0956363485972,
      "learning_rate": 1.9944311471003775e-05,
      "loss": 555.0413,
      "step": 2250
    },
    {
      "epoch": 0.043465299881720536,
      "grad_norm": 662.8958046412165,
      "learning_rate": 1.9943666419177392e-05,
      "loss": 525.6874,
      "step": 2260
    },
    {
      "epoch": 0.04365762421748036,
      "grad_norm": 786.6938179819454,
      "learning_rate": 1.9943017663499497e-05,
      "loss": 541.0315,
      "step": 2270
    },
    {
      "epoch": 0.04384994855324018,
      "grad_norm": 851.9566168187908,
      "learning_rate": 1.9942365204211734e-05,
      "loss": 538.0033,
      "step": 2280
    },
    {
      "epoch": 0.04404227288900001,
      "grad_norm": 1014.0267740996094,
      "learning_rate": 1.9941709041557134e-05,
      "loss": 541.5473,
      "step": 2290
    },
    {
      "epoch": 0.044234597224759836,
      "grad_norm": 880.1497073582835,
      "learning_rate": 1.994104917578011e-05,
      "loss": 558.7345,
      "step": 2300
    },
    {
      "epoch": 0.04442692156051966,
      "grad_norm": 857.6755267738223,
      "learning_rate": 1.9940385607126456e-05,
      "loss": 535.1024,
      "step": 2310
    },
    {
      "epoch": 0.04461924589627948,
      "grad_norm": 708.2875686175457,
      "learning_rate": 1.9939718335843326e-05,
      "loss": 548.1394,
      "step": 2320
    },
    {
      "epoch": 0.04481157023203931,
      "grad_norm": 707.9581612505445,
      "learning_rate": 1.9939047362179283e-05,
      "loss": 545.4577,
      "step": 2330
    },
    {
      "epoch": 0.04500389456779914,
      "grad_norm": 1014.2719202407706,
      "learning_rate": 1.9938372686384244e-05,
      "loss": 528.5873,
      "step": 2340
    },
    {
      "epoch": 0.04519621890355896,
      "grad_norm": 646.5182432340446,
      "learning_rate": 1.9937694308709514e-05,
      "loss": 536.911,
      "step": 2350
    },
    {
      "epoch": 0.04538854323931879,
      "grad_norm": 728.5652322845784,
      "learning_rate": 1.9937012229407783e-05,
      "loss": 545.6943,
      "step": 2360
    },
    {
      "epoch": 0.045580867575078614,
      "grad_norm": 705.5261108419323,
      "learning_rate": 1.9936326448733107e-05,
      "loss": 543.6517,
      "step": 2370
    },
    {
      "epoch": 0.04577319191083844,
      "grad_norm": 641.42054684661,
      "learning_rate": 1.9935636966940938e-05,
      "loss": 530.3073,
      "step": 2380
    },
    {
      "epoch": 0.04596551624659826,
      "grad_norm": 890.057060664522,
      "learning_rate": 1.993494378428809e-05,
      "loss": 539.9008,
      "step": 2390
    },
    {
      "epoch": 0.04615784058235809,
      "grad_norm": 976.3084239308178,
      "learning_rate": 1.9934246901032764e-05,
      "loss": 525.3227,
      "step": 2400
    },
    {
      "epoch": 0.046350164918117914,
      "grad_norm": 1298.319451345551,
      "learning_rate": 1.9933546317434536e-05,
      "loss": 538.6964,
      "step": 2410
    },
    {
      "epoch": 0.04654248925387774,
      "grad_norm": 847.2600433963399,
      "learning_rate": 1.9932842033754363e-05,
      "loss": 542.708,
      "step": 2420
    },
    {
      "epoch": 0.04673481358963757,
      "grad_norm": 750.3045876851812,
      "learning_rate": 1.9932134050254578e-05,
      "loss": 537.4875,
      "step": 2430
    },
    {
      "epoch": 0.04692713792539739,
      "grad_norm": 761.5212981614922,
      "learning_rate": 1.9931422367198893e-05,
      "loss": 529.205,
      "step": 2440
    },
    {
      "epoch": 0.047119462261157215,
      "grad_norm": 729.1939858107567,
      "learning_rate": 1.9930706984852403e-05,
      "loss": 533.5258,
      "step": 2450
    },
    {
      "epoch": 0.04731178659691704,
      "grad_norm": 693.8060340474619,
      "learning_rate": 1.992998790348157e-05,
      "loss": 533.9992,
      "step": 2460
    },
    {
      "epoch": 0.04750411093267687,
      "grad_norm": 685.9913702085655,
      "learning_rate": 1.992926512335424e-05,
      "loss": 548.7127,
      "step": 2470
    },
    {
      "epoch": 0.04769643526843669,
      "grad_norm": 659.4068670772733,
      "learning_rate": 1.992853864473964e-05,
      "loss": 532.6922,
      "step": 2480
    },
    {
      "epoch": 0.047888759604196515,
      "grad_norm": 744.2674592726822,
      "learning_rate": 1.992780846790837e-05,
      "loss": 537.2677,
      "step": 2490
    },
    {
      "epoch": 0.048081083939956346,
      "grad_norm": 742.7395697062876,
      "learning_rate": 1.9927074593132412e-05,
      "loss": 534.2826,
      "step": 2500
    },
    {
      "epoch": 0.04827340827571617,
      "grad_norm": 765.7297010871453,
      "learning_rate": 1.9926337020685114e-05,
      "loss": 543.3615,
      "step": 2510
    },
    {
      "epoch": 0.04846573261147599,
      "grad_norm": 705.1322359295946,
      "learning_rate": 1.992559575084122e-05,
      "loss": 541.1059,
      "step": 2520
    },
    {
      "epoch": 0.048658056947235816,
      "grad_norm": 729.6861253272316,
      "learning_rate": 1.9924850783876832e-05,
      "loss": 532.5268,
      "step": 2530
    },
    {
      "epoch": 0.048850381282995646,
      "grad_norm": 990.1228190611076,
      "learning_rate": 1.992410212006944e-05,
      "loss": 535.9039,
      "step": 2540
    },
    {
      "epoch": 0.04904270561875547,
      "grad_norm": 724.5479997891864,
      "learning_rate": 1.9923349759697915e-05,
      "loss": 526.4707,
      "step": 2550
    },
    {
      "epoch": 0.04923502995451529,
      "grad_norm": 835.7920156059688,
      "learning_rate": 1.992259370304249e-05,
      "loss": 523.1281,
      "step": 2560
    },
    {
      "epoch": 0.04942735429027512,
      "grad_norm": 750.6988564118371,
      "learning_rate": 1.992183395038479e-05,
      "loss": 535.6861,
      "step": 2570
    },
    {
      "epoch": 0.049619678626034947,
      "grad_norm": 649.3047660881562,
      "learning_rate": 1.9921070502007808e-05,
      "loss": 528.5413,
      "step": 2580
    },
    {
      "epoch": 0.04981200296179477,
      "grad_norm": 635.5945736581797,
      "learning_rate": 1.9920303358195916e-05,
      "loss": 524.0945,
      "step": 2590
    },
    {
      "epoch": 0.05000432729755459,
      "grad_norm": 741.3078339093645,
      "learning_rate": 1.9919532519234858e-05,
      "loss": 533.6659,
      "step": 2600
    },
    {
      "epoch": 0.050196651633314424,
      "grad_norm": 665.3001146406066,
      "learning_rate": 1.9918757985411767e-05,
      "loss": 522.319,
      "step": 2610
    },
    {
      "epoch": 0.05038897596907425,
      "grad_norm": 763.3581388836408,
      "learning_rate": 1.991797975701514e-05,
      "loss": 533.2575,
      "step": 2620
    },
    {
      "epoch": 0.05058130030483407,
      "grad_norm": 1598.675239624877,
      "learning_rate": 1.9917197834334858e-05,
      "loss": 522.5906,
      "step": 2630
    },
    {
      "epoch": 0.0507736246405939,
      "grad_norm": 862.7382749409518,
      "learning_rate": 1.9916412217662162e-05,
      "loss": 520.266,
      "step": 2640
    },
    {
      "epoch": 0.050965948976353724,
      "grad_norm": 656.1481470342628,
      "learning_rate": 1.9915622907289695e-05,
      "loss": 522.355,
      "step": 2650
    },
    {
      "epoch": 0.05115827331211355,
      "grad_norm": 733.1657818999223,
      "learning_rate": 1.9914829903511458e-05,
      "loss": 519.6336,
      "step": 2660
    },
    {
      "epoch": 0.05135059764787337,
      "grad_norm": 811.9638977805398,
      "learning_rate": 1.9914033206622828e-05,
      "loss": 535.8986,
      "step": 2670
    },
    {
      "epoch": 0.0515429219836332,
      "grad_norm": 718.5237045758081,
      "learning_rate": 1.9913232816920565e-05,
      "loss": 529.5738,
      "step": 2680
    },
    {
      "epoch": 0.051735246319393025,
      "grad_norm": 709.1296709887417,
      "learning_rate": 1.99124287347028e-05,
      "loss": 533.9658,
      "step": 2690
    },
    {
      "epoch": 0.05192757065515285,
      "grad_norm": 681.1701492583577,
      "learning_rate": 1.991162096026904e-05,
      "loss": 533.8906,
      "step": 2700
    },
    {
      "epoch": 0.05211989499091268,
      "grad_norm": 665.1251504988176,
      "learning_rate": 1.9910809493920172e-05,
      "loss": 525.5392,
      "step": 2710
    },
    {
      "epoch": 0.0523122193266725,
      "grad_norm": 676.1904534285137,
      "learning_rate": 1.990999433595845e-05,
      "loss": 529.293,
      "step": 2720
    },
    {
      "epoch": 0.052504543662432325,
      "grad_norm": 888.0866247953736,
      "learning_rate": 1.99091754866875e-05,
      "loss": 527.442,
      "step": 2730
    },
    {
      "epoch": 0.05269686799819215,
      "grad_norm": 734.1583158081239,
      "learning_rate": 1.990835294641234e-05,
      "loss": 521.3226,
      "step": 2740
    },
    {
      "epoch": 0.05288919233395198,
      "grad_norm": 611.4985369452567,
      "learning_rate": 1.990752671543935e-05,
      "loss": 525.7815,
      "step": 2750
    },
    {
      "epoch": 0.0530815166697118,
      "grad_norm": 678.5443627783844,
      "learning_rate": 1.9906696794076284e-05,
      "loss": 527.4831,
      "step": 2760
    },
    {
      "epoch": 0.053273841005471625,
      "grad_norm": 621.1472476181906,
      "learning_rate": 1.9905863182632285e-05,
      "loss": 530.1809,
      "step": 2770
    },
    {
      "epoch": 0.053466165341231456,
      "grad_norm": 613.6255766300031,
      "learning_rate": 1.990502588141784e-05,
      "loss": 519.0969,
      "step": 2780
    },
    {
      "epoch": 0.05365848967699128,
      "grad_norm": 631.3053582863905,
      "learning_rate": 1.990418489074485e-05,
      "loss": 509.2854,
      "step": 2790
    },
    {
      "epoch": 0.0538508140127511,
      "grad_norm": 753.1234591908322,
      "learning_rate": 1.9903340210926555e-05,
      "loss": 510.0524,
      "step": 2800
    },
    {
      "epoch": 0.054043138348510926,
      "grad_norm": 745.726058901604,
      "learning_rate": 1.9902491842277592e-05,
      "loss": 531.1826,
      "step": 2810
    },
    {
      "epoch": 0.054235462684270756,
      "grad_norm": 786.1870834276812,
      "learning_rate": 1.9901639785113967e-05,
      "loss": 526.6046,
      "step": 2820
    },
    {
      "epoch": 0.05442778702003058,
      "grad_norm": 816.9446739495014,
      "learning_rate": 1.990078403975305e-05,
      "loss": 537.0502,
      "step": 2830
    },
    {
      "epoch": 0.0546201113557904,
      "grad_norm": 662.5454116835482,
      "learning_rate": 1.9899924606513593e-05,
      "loss": 521.8435,
      "step": 2840
    },
    {
      "epoch": 0.05481243569155023,
      "grad_norm": 697.6644935840952,
      "learning_rate": 1.9899061485715726e-05,
      "loss": 512.287,
      "step": 2850
    },
    {
      "epoch": 0.05500476002731006,
      "grad_norm": 657.042049649758,
      "learning_rate": 1.9898194677680943e-05,
      "loss": 522.9424,
      "step": 2860
    },
    {
      "epoch": 0.05519708436306988,
      "grad_norm": 704.66905540707,
      "learning_rate": 1.9897324182732118e-05,
      "loss": 515.9715,
      "step": 2870
    },
    {
      "epoch": 0.055389408698829704,
      "grad_norm": 653.9461352074152,
      "learning_rate": 1.989645000119349e-05,
      "loss": 513.6467,
      "step": 2880
    },
    {
      "epoch": 0.055581733034589534,
      "grad_norm": 648.845462458859,
      "learning_rate": 1.9895572133390687e-05,
      "loss": 517.3306,
      "step": 2890
    },
    {
      "epoch": 0.05577405737034936,
      "grad_norm": 617.4711130107623,
      "learning_rate": 1.9894690579650694e-05,
      "loss": 518.482,
      "step": 2900
    },
    {
      "epoch": 0.05596638170610918,
      "grad_norm": 737.4042827552958,
      "learning_rate": 1.9893805340301876e-05,
      "loss": 521.9151,
      "step": 2910
    },
    {
      "epoch": 0.05615870604186901,
      "grad_norm": 665.3499871063476,
      "learning_rate": 1.989291641567397e-05,
      "loss": 516.3638,
      "step": 2920
    },
    {
      "epoch": 0.056351030377628834,
      "grad_norm": 634.0834194937779,
      "learning_rate": 1.9892023806098083e-05,
      "loss": 502.445,
      "step": 2930
    },
    {
      "epoch": 0.05654335471338866,
      "grad_norm": 626.6913492134656,
      "learning_rate": 1.9891127511906703e-05,
      "loss": 511.0661,
      "step": 2940
    },
    {
      "epoch": 0.05673567904914848,
      "grad_norm": 637.3891412661674,
      "learning_rate": 1.9890227533433685e-05,
      "loss": 515.9606,
      "step": 2950
    },
    {
      "epoch": 0.05692800338490831,
      "grad_norm": 774.722354100866,
      "learning_rate": 1.988932387101425e-05,
      "loss": 530.3114,
      "step": 2960
    },
    {
      "epoch": 0.057120327720668135,
      "grad_norm": 661.0222702671855,
      "learning_rate": 1.9888416524985e-05,
      "loss": 512.457,
      "step": 2970
    },
    {
      "epoch": 0.05731265205642796,
      "grad_norm": 596.821145071028,
      "learning_rate": 1.988750549568391e-05,
      "loss": 523.493,
      "step": 2980
    },
    {
      "epoch": 0.05750497639218779,
      "grad_norm": 684.7345599870473,
      "learning_rate": 1.9886590783450317e-05,
      "loss": 501.1971,
      "step": 2990
    },
    {
      "epoch": 0.05769730072794761,
      "grad_norm": 626.9730550631467,
      "learning_rate": 1.9885672388624942e-05,
      "loss": 510.5386,
      "step": 3000
    },
    {
      "epoch": 0.057889625063707435,
      "grad_norm": 1118.3852260251188,
      "learning_rate": 1.9884750311549868e-05,
      "loss": 522.3505,
      "step": 3010
    },
    {
      "epoch": 0.05808194939946726,
      "grad_norm": 647.0029106792531,
      "learning_rate": 1.9883824552568557e-05,
      "loss": 520.7158,
      "step": 3020
    },
    {
      "epoch": 0.05827427373522709,
      "grad_norm": 662.8229603194394,
      "learning_rate": 1.9882895112025835e-05,
      "loss": 526.3525,
      "step": 3030
    },
    {
      "epoch": 0.05846659807098691,
      "grad_norm": 704.9460673982611,
      "learning_rate": 1.9881961990267906e-05,
      "loss": 500.5047,
      "step": 3040
    },
    {
      "epoch": 0.058658922406746736,
      "grad_norm": 1698.837077403239,
      "learning_rate": 1.988102518764234e-05,
      "loss": 521.0469,
      "step": 3050
    },
    {
      "epoch": 0.058851246742506566,
      "grad_norm": 597.8717905084069,
      "learning_rate": 1.9880084704498084e-05,
      "loss": 519.1069,
      "step": 3060
    },
    {
      "epoch": 0.05904357107826639,
      "grad_norm": 712.4482798190649,
      "learning_rate": 1.987914054118545e-05,
      "loss": 532.2404,
      "step": 3070
    },
    {
      "epoch": 0.05923589541402621,
      "grad_norm": 871.7664288960812,
      "learning_rate": 1.9878192698056125e-05,
      "loss": 516.484,
      "step": 3080
    },
    {
      "epoch": 0.059428219749786036,
      "grad_norm": 621.4967780730051,
      "learning_rate": 1.9877241175463165e-05,
      "loss": 516.0212,
      "step": 3090
    },
    {
      "epoch": 0.05962054408554587,
      "grad_norm": 622.6022269382079,
      "learning_rate": 1.9876285973760993e-05,
      "loss": 523.7819,
      "step": 3100
    },
    {
      "epoch": 0.05981286842130569,
      "grad_norm": 654.5456179019733,
      "learning_rate": 1.9875327093305405e-05,
      "loss": 517.6488,
      "step": 3110
    },
    {
      "epoch": 0.06000519275706551,
      "grad_norm": 674.4448361971492,
      "learning_rate": 1.9874364534453577e-05,
      "loss": 514.8555,
      "step": 3120
    },
    {
      "epoch": 0.060197517092825344,
      "grad_norm": 682.141457904393,
      "learning_rate": 1.9873398297564036e-05,
      "loss": 517.7271,
      "step": 3130
    },
    {
      "epoch": 0.06038984142858517,
      "grad_norm": 550.8065749052461,
      "learning_rate": 1.9872428382996697e-05,
      "loss": 521.9763,
      "step": 3140
    },
    {
      "epoch": 0.06058216576434499,
      "grad_norm": 937.2468781077604,
      "learning_rate": 1.987145479111283e-05,
      "loss": 503.1943,
      "step": 3150
    },
    {
      "epoch": 0.060774490100104814,
      "grad_norm": 763.5125743270676,
      "learning_rate": 1.987047752227509e-05,
      "loss": 513.4222,
      "step": 3160
    },
    {
      "epoch": 0.060966814435864644,
      "grad_norm": 633.7946765874968,
      "learning_rate": 1.9869496576847488e-05,
      "loss": 499.4493,
      "step": 3170
    },
    {
      "epoch": 0.06115913877162447,
      "grad_norm": 629.8401231851526,
      "learning_rate": 1.9868511955195407e-05,
      "loss": 514.617,
      "step": 3180
    },
    {
      "epoch": 0.06135146310738429,
      "grad_norm": 722.274363839671,
      "learning_rate": 1.986752365768561e-05,
      "loss": 491.6552,
      "step": 3190
    },
    {
      "epoch": 0.06154378744314412,
      "grad_norm": 698.004629411705,
      "learning_rate": 1.986653168468622e-05,
      "loss": 519.5537,
      "step": 3200
    },
    {
      "epoch": 0.061736111778903945,
      "grad_norm": 778.116353487612,
      "learning_rate": 1.9865536036566727e-05,
      "loss": 511.8583,
      "step": 3210
    },
    {
      "epoch": 0.06192843611466377,
      "grad_norm": 670.6659366755904,
      "learning_rate": 1.9864536713697992e-05,
      "loss": 521.8036,
      "step": 3220
    },
    {
      "epoch": 0.06212076045042359,
      "grad_norm": 613.7695500085715,
      "learning_rate": 1.986353371645225e-05,
      "loss": 523.6647,
      "step": 3230
    },
    {
      "epoch": 0.06231308478618342,
      "grad_norm": 652.4045245318075,
      "learning_rate": 1.9862527045203105e-05,
      "loss": 515.5668,
      "step": 3240
    },
    {
      "epoch": 0.06250540912194325,
      "grad_norm": 640.2665108728766,
      "learning_rate": 1.986151670032552e-05,
      "loss": 510.1916,
      "step": 3250
    },
    {
      "epoch": 0.06269773345770308,
      "grad_norm": 672.8530359257265,
      "learning_rate": 1.986050268219583e-05,
      "loss": 500.603,
      "step": 3260
    },
    {
      "epoch": 0.06289005779346289,
      "grad_norm": 643.0303715876427,
      "learning_rate": 1.9859484991191742e-05,
      "loss": 518.3387,
      "step": 3270
    },
    {
      "epoch": 0.06308238212922272,
      "grad_norm": 667.8351982269853,
      "learning_rate": 1.985846362769233e-05,
      "loss": 522.2421,
      "step": 3280
    },
    {
      "epoch": 0.06327470646498255,
      "grad_norm": 706.1733963549115,
      "learning_rate": 1.9857438592078034e-05,
      "loss": 509.3566,
      "step": 3290
    },
    {
      "epoch": 0.06346703080074237,
      "grad_norm": 823.7219002972914,
      "learning_rate": 1.9856409884730667e-05,
      "loss": 512.2886,
      "step": 3300
    },
    {
      "epoch": 0.0636593551365022,
      "grad_norm": 604.2720020574882,
      "learning_rate": 1.98553775060334e-05,
      "loss": 502.6998,
      "step": 3310
    },
    {
      "epoch": 0.06385167947226203,
      "grad_norm": 626.5921201793383,
      "learning_rate": 1.9854341456370777e-05,
      "loss": 493.7466,
      "step": 3320
    },
    {
      "epoch": 0.06404400380802185,
      "grad_norm": 674.7114276650651,
      "learning_rate": 1.9853301736128712e-05,
      "loss": 510.9708,
      "step": 3330
    },
    {
      "epoch": 0.06423632814378168,
      "grad_norm": 727.7131791678172,
      "learning_rate": 1.9852258345694486e-05,
      "loss": 521.3875,
      "step": 3340
    },
    {
      "epoch": 0.06442865247954149,
      "grad_norm": 708.5336543391381,
      "learning_rate": 1.9851211285456738e-05,
      "loss": 507.2278,
      "step": 3350
    },
    {
      "epoch": 0.06462097681530132,
      "grad_norm": 699.332774977859,
      "learning_rate": 1.9850160555805485e-05,
      "loss": 518.902,
      "step": 3360
    },
    {
      "epoch": 0.06481330115106115,
      "grad_norm": 754.9990775585326,
      "learning_rate": 1.9849106157132105e-05,
      "loss": 521.9002,
      "step": 3370
    },
    {
      "epoch": 0.06500562548682097,
      "grad_norm": 628.1105872878063,
      "learning_rate": 1.9848048089829347e-05,
      "loss": 513.4633,
      "step": 3380
    },
    {
      "epoch": 0.0651979498225808,
      "grad_norm": 607.6423577389,
      "learning_rate": 1.9846986354291324e-05,
      "loss": 501.1044,
      "step": 3390
    },
    {
      "epoch": 0.06539027415834063,
      "grad_norm": 685.3186577512565,
      "learning_rate": 1.9845920950913506e-05,
      "loss": 516.6853,
      "step": 3400
    },
    {
      "epoch": 0.06558259849410045,
      "grad_norm": 663.6144566856839,
      "learning_rate": 1.9844851880092748e-05,
      "loss": 495.0083,
      "step": 3410
    },
    {
      "epoch": 0.06577492282986028,
      "grad_norm": 655.7296760262857,
      "learning_rate": 1.9843779142227258e-05,
      "loss": 503.3423,
      "step": 3420
    },
    {
      "epoch": 0.06596724716562011,
      "grad_norm": 619.243560898213,
      "learning_rate": 1.984270273771661e-05,
      "loss": 504.1719,
      "step": 3430
    },
    {
      "epoch": 0.06615957150137992,
      "grad_norm": 694.8542065917636,
      "learning_rate": 1.9841622666961756e-05,
      "loss": 503.6424,
      "step": 3440
    },
    {
      "epoch": 0.06635189583713975,
      "grad_norm": 673.2736125358978,
      "learning_rate": 1.9840538930364992e-05,
      "loss": 514.9401,
      "step": 3450
    },
    {
      "epoch": 0.06654422017289958,
      "grad_norm": 628.7796022492911,
      "learning_rate": 1.983945152833e-05,
      "loss": 494.9331,
      "step": 3460
    },
    {
      "epoch": 0.0667365445086594,
      "grad_norm": 696.7574145593314,
      "learning_rate": 1.9838360461261817e-05,
      "loss": 512.15,
      "step": 3470
    },
    {
      "epoch": 0.06692886884441923,
      "grad_norm": 637.5080529680322,
      "learning_rate": 1.9837265729566853e-05,
      "loss": 496.1206,
      "step": 3480
    },
    {
      "epoch": 0.06712119318017905,
      "grad_norm": 699.5220534995685,
      "learning_rate": 1.9836167333652866e-05,
      "loss": 514.8845,
      "step": 3490
    },
    {
      "epoch": 0.06731351751593888,
      "grad_norm": 639.8540894236521,
      "learning_rate": 1.9835065273929002e-05,
      "loss": 501.3714,
      "step": 3500
    },
    {
      "epoch": 0.06750584185169871,
      "grad_norm": 633.729983004065,
      "learning_rate": 1.9833959550805754e-05,
      "loss": 501.0371,
      "step": 3510
    },
    {
      "epoch": 0.06769816618745852,
      "grad_norm": 802.133374494615,
      "learning_rate": 1.9832850164694983e-05,
      "loss": 500.5452,
      "step": 3520
    },
    {
      "epoch": 0.06789049052321836,
      "grad_norm": 830.7796341589134,
      "learning_rate": 1.9831737116009924e-05,
      "loss": 497.9683,
      "step": 3530
    },
    {
      "epoch": 0.06808281485897819,
      "grad_norm": 630.399254399982,
      "learning_rate": 1.9830620405165164e-05,
      "loss": 493.7095,
      "step": 3540
    },
    {
      "epoch": 0.068275139194738,
      "grad_norm": 644.411241438413,
      "learning_rate": 1.982950003257666e-05,
      "loss": 517.4553,
      "step": 3550
    },
    {
      "epoch": 0.06846746353049783,
      "grad_norm": 588.2682798659015,
      "learning_rate": 1.9828375998661738e-05,
      "loss": 507.9328,
      "step": 3560
    },
    {
      "epoch": 0.06865978786625766,
      "grad_norm": 617.9867513432937,
      "learning_rate": 1.9827248303839073e-05,
      "loss": 478.5115,
      "step": 3570
    },
    {
      "epoch": 0.06885211220201748,
      "grad_norm": 650.826612173906,
      "learning_rate": 1.982611694852872e-05,
      "loss": 500.3652,
      "step": 3580
    },
    {
      "epoch": 0.06904443653777731,
      "grad_norm": 649.8580989326786,
      "learning_rate": 1.9824981933152087e-05,
      "loss": 489.6465,
      "step": 3590
    },
    {
      "epoch": 0.06923676087353714,
      "grad_norm": 621.0773120313047,
      "learning_rate": 1.9823843258131945e-05,
      "loss": 500.1181,
      "step": 3600
    },
    {
      "epoch": 0.06942908520929696,
      "grad_norm": 703.0441909093954,
      "learning_rate": 1.9822700923892438e-05,
      "loss": 517.5681,
      "step": 3610
    },
    {
      "epoch": 0.06962140954505679,
      "grad_norm": 655.7881223496453,
      "learning_rate": 1.9821554930859066e-05,
      "loss": 502.2795,
      "step": 3620
    },
    {
      "epoch": 0.0698137338808166,
      "grad_norm": 717.8903663067391,
      "learning_rate": 1.982040527945869e-05,
      "loss": 506.7368,
      "step": 3630
    },
    {
      "epoch": 0.07000605821657643,
      "grad_norm": 573.5901477923879,
      "learning_rate": 1.9819251970119534e-05,
      "loss": 511.8817,
      "step": 3640
    },
    {
      "epoch": 0.07019838255233626,
      "grad_norm": 687.4724535149858,
      "learning_rate": 1.9818095003271193e-05,
      "loss": 496.4292,
      "step": 3650
    },
    {
      "epoch": 0.07039070688809608,
      "grad_norm": 639.0458926308955,
      "learning_rate": 1.9816934379344613e-05,
      "loss": 501.0932,
      "step": 3660
    },
    {
      "epoch": 0.07058303122385591,
      "grad_norm": 680.6213480239904,
      "learning_rate": 1.9815770098772108e-05,
      "loss": 499.6543,
      "step": 3670
    },
    {
      "epoch": 0.07077535555961574,
      "grad_norm": 705.8244315480342,
      "learning_rate": 1.9814602161987354e-05,
      "loss": 490.7324,
      "step": 3680
    },
    {
      "epoch": 0.07096767989537556,
      "grad_norm": 583.136369592737,
      "learning_rate": 1.981343056942539e-05,
      "loss": 490.0691,
      "step": 3690
    },
    {
      "epoch": 0.07116000423113539,
      "grad_norm": 715.6044089791625,
      "learning_rate": 1.9812255321522614e-05,
      "loss": 515.5474,
      "step": 3700
    },
    {
      "epoch": 0.07135232856689522,
      "grad_norm": 581.4536845406911,
      "learning_rate": 1.981107641871678e-05,
      "loss": 496.5484,
      "step": 3710
    },
    {
      "epoch": 0.07154465290265503,
      "grad_norm": 657.1353485262981,
      "learning_rate": 1.980989386144702e-05,
      "loss": 494.1127,
      "step": 3720
    },
    {
      "epoch": 0.07173697723841486,
      "grad_norm": 614.2914353568027,
      "learning_rate": 1.980870765015381e-05,
      "loss": 502.2589,
      "step": 3730
    },
    {
      "epoch": 0.0719293015741747,
      "grad_norm": 698.3911986377062,
      "learning_rate": 1.9807517785278997e-05,
      "loss": 501.1257,
      "step": 3740
    },
    {
      "epoch": 0.07212162590993451,
      "grad_norm": 630.4361176425793,
      "learning_rate": 1.9806324267265786e-05,
      "loss": 498.7661,
      "step": 3750
    },
    {
      "epoch": 0.07231395024569434,
      "grad_norm": 615.211225033671,
      "learning_rate": 1.9805127096558742e-05,
      "loss": 498.0036,
      "step": 3760
    },
    {
      "epoch": 0.07250627458145416,
      "grad_norm": 719.8962573967688,
      "learning_rate": 1.980392627360379e-05,
      "loss": 501.0617,
      "step": 3770
    },
    {
      "epoch": 0.07269859891721399,
      "grad_norm": 741.5868184209745,
      "learning_rate": 1.9802721798848225e-05,
      "loss": 509.3568,
      "step": 3780
    },
    {
      "epoch": 0.07289092325297382,
      "grad_norm": 665.4832813423726,
      "learning_rate": 1.980151367274068e-05,
      "loss": 488.9466,
      "step": 3790
    },
    {
      "epoch": 0.07308324758873364,
      "grad_norm": 603.8607554267116,
      "learning_rate": 1.9800301895731172e-05,
      "loss": 502.7106,
      "step": 3800
    },
    {
      "epoch": 0.07327557192449347,
      "grad_norm": 627.9793823986435,
      "learning_rate": 1.9799086468271065e-05,
      "loss": 497.9342,
      "step": 3810
    },
    {
      "epoch": 0.0734678962602533,
      "grad_norm": 636.4435313966623,
      "learning_rate": 1.9797867390813086e-05,
      "loss": 490.8221,
      "step": 3820
    },
    {
      "epoch": 0.07366022059601311,
      "grad_norm": 632.9088111605367,
      "learning_rate": 1.9796644663811318e-05,
      "loss": 497.4042,
      "step": 3830
    },
    {
      "epoch": 0.07385254493177294,
      "grad_norm": 718.9900162397643,
      "learning_rate": 1.9795418287721215e-05,
      "loss": 492.4667,
      "step": 3840
    },
    {
      "epoch": 0.07404486926753277,
      "grad_norm": 687.1497841147152,
      "learning_rate": 1.9794188262999574e-05,
      "loss": 504.8531,
      "step": 3850
    },
    {
      "epoch": 0.07423719360329259,
      "grad_norm": 642.1877408136563,
      "learning_rate": 1.979295459010456e-05,
      "loss": 495.9551,
      "step": 3860
    },
    {
      "epoch": 0.07442951793905242,
      "grad_norm": 642.986179627092,
      "learning_rate": 1.9791717269495698e-05,
      "loss": 486.4746,
      "step": 3870
    },
    {
      "epoch": 0.07462184227481225,
      "grad_norm": 722.7348743238974,
      "learning_rate": 1.979047630163387e-05,
      "loss": 489.8139,
      "step": 3880
    },
    {
      "epoch": 0.07481416661057207,
      "grad_norm": 713.3383230851642,
      "learning_rate": 1.9789231686981313e-05,
      "loss": 507.3725,
      "step": 3890
    },
    {
      "epoch": 0.0750064909463319,
      "grad_norm": 584.1448448604131,
      "learning_rate": 1.978798342600163e-05,
      "loss": 489.6483,
      "step": 3900
    },
    {
      "epoch": 0.07519881528209171,
      "grad_norm": 642.4517334947999,
      "learning_rate": 1.978673151915977e-05,
      "loss": 499.9061,
      "step": 3910
    },
    {
      "epoch": 0.07539113961785154,
      "grad_norm": 587.61012014815,
      "learning_rate": 1.9785475966922055e-05,
      "loss": 497.4966,
      "step": 3920
    },
    {
      "epoch": 0.07558346395361137,
      "grad_norm": 647.6673123961527,
      "learning_rate": 1.9784216769756156e-05,
      "loss": 489.6694,
      "step": 3930
    },
    {
      "epoch": 0.07577578828937119,
      "grad_norm": 767.1562758524656,
      "learning_rate": 1.97829539281311e-05,
      "loss": 486.8688,
      "step": 3940
    },
    {
      "epoch": 0.07596811262513102,
      "grad_norm": 622.1411124647373,
      "learning_rate": 1.9781687442517278e-05,
      "loss": 495.8756,
      "step": 3950
    },
    {
      "epoch": 0.07616043696089085,
      "grad_norm": 611.8275224683359,
      "learning_rate": 1.9780417313386433e-05,
      "loss": 503.6595,
      "step": 3960
    },
    {
      "epoch": 0.07635276129665067,
      "grad_norm": 635.9795739979108,
      "learning_rate": 1.9779143541211664e-05,
      "loss": 492.0267,
      "step": 3970
    },
    {
      "epoch": 0.0765450856324105,
      "grad_norm": 605.1944045484206,
      "learning_rate": 1.9777866126467436e-05,
      "loss": 474.1518,
      "step": 3980
    },
    {
      "epoch": 0.07673740996817033,
      "grad_norm": 688.9065962615937,
      "learning_rate": 1.9776585069629566e-05,
      "loss": 488.8311,
      "step": 3990
    },
    {
      "epoch": 0.07692973430393014,
      "grad_norm": 616.907512994056,
      "learning_rate": 1.9775300371175225e-05,
      "loss": 495.4689,
      "step": 4000
    },
    {
      "epoch": 0.07712205863968997,
      "grad_norm": 869.3229600963,
      "learning_rate": 1.9774012031582935e-05,
      "loss": 484.9483,
      "step": 4010
    },
    {
      "epoch": 0.0773143829754498,
      "grad_norm": 621.4725522442566,
      "learning_rate": 1.9772720051332585e-05,
      "loss": 492.811,
      "step": 4020
    },
    {
      "epoch": 0.07750670731120962,
      "grad_norm": 883.3922953575607,
      "learning_rate": 1.977142443090542e-05,
      "loss": 490.3932,
      "step": 4030
    },
    {
      "epoch": 0.07769903164696945,
      "grad_norm": 624.9347849676961,
      "learning_rate": 1.9770125170784035e-05,
      "loss": 491.0477,
      "step": 4040
    },
    {
      "epoch": 0.07789135598272927,
      "grad_norm": 646.4092159406114,
      "learning_rate": 1.9768822271452385e-05,
      "loss": 501.6062,
      "step": 4050
    },
    {
      "epoch": 0.0780836803184891,
      "grad_norm": 627.6555519603643,
      "learning_rate": 1.9767515733395774e-05,
      "loss": 491.7407,
      "step": 4060
    },
    {
      "epoch": 0.07827600465424893,
      "grad_norm": 1351.6676370674604,
      "learning_rate": 1.976620555710087e-05,
      "loss": 491.9886,
      "step": 4070
    },
    {
      "epoch": 0.07846832899000875,
      "grad_norm": 629.1803078798862,
      "learning_rate": 1.976489174305569e-05,
      "loss": 494.6177,
      "step": 4080
    },
    {
      "epoch": 0.07866065332576858,
      "grad_norm": 578.7649810787432,
      "learning_rate": 1.9763574291749603e-05,
      "loss": 479.5029,
      "step": 4090
    },
    {
      "epoch": 0.0788529776615284,
      "grad_norm": 540.8769728899618,
      "learning_rate": 1.9762253203673348e-05,
      "loss": 483.3324,
      "step": 4100
    },
    {
      "epoch": 0.07904530199728822,
      "grad_norm": 656.8305147234112,
      "learning_rate": 1.9760928479319003e-05,
      "loss": 494.2564,
      "step": 4110
    },
    {
      "epoch": 0.07923762633304805,
      "grad_norm": 571.891691206646,
      "learning_rate": 1.9759600119180005e-05,
      "loss": 490.0733,
      "step": 4120
    },
    {
      "epoch": 0.07942995066880788,
      "grad_norm": 568.0849472261627,
      "learning_rate": 1.975826812375115e-05,
      "loss": 483.9812,
      "step": 4130
    },
    {
      "epoch": 0.0796222750045677,
      "grad_norm": 620.4971891151242,
      "learning_rate": 1.9756932493528583e-05,
      "loss": 468.2408,
      "step": 4140
    },
    {
      "epoch": 0.07981459934032753,
      "grad_norm": 690.5422084610557,
      "learning_rate": 1.97555932290098e-05,
      "loss": 516.2057,
      "step": 4150
    },
    {
      "epoch": 0.08000692367608736,
      "grad_norm": 631.7672672240832,
      "learning_rate": 1.9754250330693658e-05,
      "loss": 485.5635,
      "step": 4160
    },
    {
      "epoch": 0.08019924801184718,
      "grad_norm": 548.6785879389789,
      "learning_rate": 1.9752903799080366e-05,
      "loss": 499.1404,
      "step": 4170
    },
    {
      "epoch": 0.08039157234760701,
      "grad_norm": 580.5544406665524,
      "learning_rate": 1.9751553634671485e-05,
      "loss": 474.7078,
      "step": 4180
    },
    {
      "epoch": 0.08058389668336682,
      "grad_norm": 707.3800613133382,
      "learning_rate": 1.9750199837969922e-05,
      "loss": 479.0868,
      "step": 4190
    },
    {
      "epoch": 0.08077622101912665,
      "grad_norm": 607.9172149397061,
      "learning_rate": 1.9748842409479953e-05,
      "loss": 487.9297,
      "step": 4200
    },
    {
      "epoch": 0.08096854535488648,
      "grad_norm": 686.0145755679489,
      "learning_rate": 1.9747481349707197e-05,
      "loss": 497.5591,
      "step": 4210
    },
    {
      "epoch": 0.0811608696906463,
      "grad_norm": 669.0150916061837,
      "learning_rate": 1.9746116659158618e-05,
      "loss": 488.224,
      "step": 4220
    },
    {
      "epoch": 0.08135319402640613,
      "grad_norm": 597.9819691648493,
      "learning_rate": 1.9744748338342546e-05,
      "loss": 481.8612,
      "step": 4230
    },
    {
      "epoch": 0.08154551836216596,
      "grad_norm": 654.7117910540604,
      "learning_rate": 1.974337638776866e-05,
      "loss": 486.0175,
      "step": 4240
    },
    {
      "epoch": 0.08173784269792578,
      "grad_norm": 594.8033843059904,
      "learning_rate": 1.9742000807947986e-05,
      "loss": 476.3571,
      "step": 4250
    },
    {
      "epoch": 0.08193016703368561,
      "grad_norm": 641.1753752875821,
      "learning_rate": 1.9740621599392907e-05,
      "loss": 479.6258,
      "step": 4260
    },
    {
      "epoch": 0.08212249136944544,
      "grad_norm": 639.0411963835763,
      "learning_rate": 1.9739238762617155e-05,
      "loss": 479.2322,
      "step": 4270
    },
    {
      "epoch": 0.08231481570520525,
      "grad_norm": 609.621309326222,
      "learning_rate": 1.973785229813581e-05,
      "loss": 489.9551,
      "step": 4280
    },
    {
      "epoch": 0.08250714004096509,
      "grad_norm": 595.0031502326282,
      "learning_rate": 1.973646220646531e-05,
      "loss": 492.7053,
      "step": 4290
    },
    {
      "epoch": 0.08269946437672492,
      "grad_norm": 605.150045864125,
      "learning_rate": 1.973506848812344e-05,
      "loss": 482.1862,
      "step": 4300
    },
    {
      "epoch": 0.08289178871248473,
      "grad_norm": 643.5863400964225,
      "learning_rate": 1.9733671143629342e-05,
      "loss": 501.3153,
      "step": 4310
    },
    {
      "epoch": 0.08308411304824456,
      "grad_norm": 590.4906240524667,
      "learning_rate": 1.9732270173503493e-05,
      "loss": 483.9846,
      "step": 4320
    },
    {
      "epoch": 0.08327643738400438,
      "grad_norm": 583.8559636423248,
      "learning_rate": 1.9730865578267745e-05,
      "loss": 476.9609,
      "step": 4330
    },
    {
      "epoch": 0.08346876171976421,
      "grad_norm": 579.2100154212275,
      "learning_rate": 1.972945735844528e-05,
      "loss": 482.5675,
      "step": 4340
    },
    {
      "epoch": 0.08366108605552404,
      "grad_norm": 628.1190919574348,
      "learning_rate": 1.972804551456063e-05,
      "loss": 488.1154,
      "step": 4350
    },
    {
      "epoch": 0.08385341039128386,
      "grad_norm": 635.6027094674298,
      "learning_rate": 1.9726630047139695e-05,
      "loss": 490.5554,
      "step": 4360
    },
    {
      "epoch": 0.08404573472704369,
      "grad_norm": 563.4844302115933,
      "learning_rate": 1.9725210956709707e-05,
      "loss": 474.7476,
      "step": 4370
    },
    {
      "epoch": 0.08423805906280352,
      "grad_norm": 663.2736077910538,
      "learning_rate": 1.9723788243799253e-05,
      "loss": 494.2354,
      "step": 4380
    },
    {
      "epoch": 0.08443038339856333,
      "grad_norm": 577.906945078492,
      "learning_rate": 1.972236190893827e-05,
      "loss": 473.9004,
      "step": 4390
    },
    {
      "epoch": 0.08462270773432316,
      "grad_norm": 819.4941920482539,
      "learning_rate": 1.972093195265805e-05,
      "loss": 502.3352,
      "step": 4400
    },
    {
      "epoch": 0.084815032070083,
      "grad_norm": 715.8667133204852,
      "learning_rate": 1.9719498375491224e-05,
      "loss": 476.0669,
      "step": 4410
    },
    {
      "epoch": 0.08500735640584281,
      "grad_norm": 639.8410343934174,
      "learning_rate": 1.9718061177971777e-05,
      "loss": 477.4385,
      "step": 4420
    },
    {
      "epoch": 0.08519968074160264,
      "grad_norm": 568.7116382152404,
      "learning_rate": 1.9716620360635036e-05,
      "loss": 486.7625,
      "step": 4430
    },
    {
      "epoch": 0.08539200507736247,
      "grad_norm": 731.286113852741,
      "learning_rate": 1.971517592401769e-05,
      "loss": 497.6304,
      "step": 4440
    },
    {
      "epoch": 0.08558432941312229,
      "grad_norm": 615.9179155234693,
      "learning_rate": 1.9713727868657764e-05,
      "loss": 476.1781,
      "step": 4450
    },
    {
      "epoch": 0.08577665374888212,
      "grad_norm": 650.3429361883706,
      "learning_rate": 1.971227619509463e-05,
      "loss": 472.3176,
      "step": 4460
    },
    {
      "epoch": 0.08596897808464193,
      "grad_norm": 551.4978595475194,
      "learning_rate": 1.971082090386902e-05,
      "loss": 493.2781,
      "step": 4470
    },
    {
      "epoch": 0.08616130242040176,
      "grad_norm": 4711.236231799215,
      "learning_rate": 1.9709361995523e-05,
      "loss": 482.4984,
      "step": 4480
    },
    {
      "epoch": 0.0863536267561616,
      "grad_norm": 715.8205450069757,
      "learning_rate": 1.9707899470599998e-05,
      "loss": 497.31,
      "step": 4490
    },
    {
      "epoch": 0.08654595109192141,
      "grad_norm": 665.0607681375076,
      "learning_rate": 1.970643332964477e-05,
      "loss": 478.2403,
      "step": 4500
    },
    {
      "epoch": 0.08673827542768124,
      "grad_norm": 613.021352941912,
      "learning_rate": 1.9704963573203435e-05,
      "loss": 474.9517,
      "step": 4510
    },
    {
      "epoch": 0.08693059976344107,
      "grad_norm": 637.2150147876924,
      "learning_rate": 1.970349020182345e-05,
      "loss": 489.9517,
      "step": 4520
    },
    {
      "epoch": 0.08712292409920089,
      "grad_norm": 599.3458803137219,
      "learning_rate": 1.9702013216053623e-05,
      "loss": 469.1975,
      "step": 4530
    },
    {
      "epoch": 0.08731524843496072,
      "grad_norm": 610.6760803314122,
      "learning_rate": 1.9700532616444114e-05,
      "loss": 485.5857,
      "step": 4540
    },
    {
      "epoch": 0.08750757277072055,
      "grad_norm": 601.7390395855866,
      "learning_rate": 1.969904840354641e-05,
      "loss": 479.4551,
      "step": 4550
    },
    {
      "epoch": 0.08769989710648037,
      "grad_norm": 599.5265816559986,
      "learning_rate": 1.9697560577913358e-05,
      "loss": 493.1255,
      "step": 4560
    },
    {
      "epoch": 0.0878922214422402,
      "grad_norm": 596.6153359263633,
      "learning_rate": 1.9696069140099152e-05,
      "loss": 476.3837,
      "step": 4570
    },
    {
      "epoch": 0.08808454577800003,
      "grad_norm": 600.991320286704,
      "learning_rate": 1.969457409065933e-05,
      "loss": 478.4364,
      "step": 4580
    },
    {
      "epoch": 0.08827687011375984,
      "grad_norm": 559.9529040626103,
      "learning_rate": 1.969307543015077e-05,
      "loss": 477.7218,
      "step": 4590
    },
    {
      "epoch": 0.08846919444951967,
      "grad_norm": 589.8268999503209,
      "learning_rate": 1.9691573159131696e-05,
      "loss": 481.6495,
      "step": 4600
    },
    {
      "epoch": 0.08866151878527949,
      "grad_norm": 597.181130058833,
      "learning_rate": 1.9690067278161686e-05,
      "loss": 482.775,
      "step": 4610
    },
    {
      "epoch": 0.08885384312103932,
      "grad_norm": 635.0713423449403,
      "learning_rate": 1.9688557787801647e-05,
      "loss": 477.3972,
      "step": 4620
    },
    {
      "epoch": 0.08904616745679915,
      "grad_norm": 583.9772768431457,
      "learning_rate": 1.968704468861385e-05,
      "loss": 484.344,
      "step": 4630
    },
    {
      "epoch": 0.08923849179255897,
      "grad_norm": 556.6993701208112,
      "learning_rate": 1.968552798116189e-05,
      "loss": 465.836,
      "step": 4640
    },
    {
      "epoch": 0.0894308161283188,
      "grad_norm": 659.3678227798055,
      "learning_rate": 1.9684007666010716e-05,
      "loss": 471.6841,
      "step": 4650
    },
    {
      "epoch": 0.08962314046407863,
      "grad_norm": 577.0937028794929,
      "learning_rate": 1.9682483743726624e-05,
      "loss": 485.9341,
      "step": 4660
    },
    {
      "epoch": 0.08981546479983844,
      "grad_norm": 593.5870471604508,
      "learning_rate": 1.968095621487725e-05,
      "loss": 480.7809,
      "step": 4670
    },
    {
      "epoch": 0.09000778913559827,
      "grad_norm": 569.4212941663662,
      "learning_rate": 1.9679425080031574e-05,
      "loss": 474.5443,
      "step": 4680
    },
    {
      "epoch": 0.0902001134713581,
      "grad_norm": 620.810725310614,
      "learning_rate": 1.9677890339759914e-05,
      "loss": 494.733,
      "step": 4690
    },
    {
      "epoch": 0.09039243780711792,
      "grad_norm": 641.0236578652243,
      "learning_rate": 1.967635199463394e-05,
      "loss": 472.8692,
      "step": 4700
    },
    {
      "epoch": 0.09058476214287775,
      "grad_norm": 588.2953424814439,
      "learning_rate": 1.9674810045226658e-05,
      "loss": 478.5095,
      "step": 4710
    },
    {
      "epoch": 0.09077708647863758,
      "grad_norm": 606.7000221463143,
      "learning_rate": 1.967326449211242e-05,
      "loss": 473.909,
      "step": 4720
    },
    {
      "epoch": 0.0909694108143974,
      "grad_norm": 615.4246537147368,
      "learning_rate": 1.9671715335866915e-05,
      "loss": 472.5508,
      "step": 4730
    },
    {
      "epoch": 0.09116173515015723,
      "grad_norm": 641.6407347507667,
      "learning_rate": 1.9670162577067182e-05,
      "loss": 494.7355,
      "step": 4740
    },
    {
      "epoch": 0.09135405948591704,
      "grad_norm": 641.7626247280142,
      "learning_rate": 1.9668606216291598e-05,
      "loss": 474.9429,
      "step": 4750
    },
    {
      "epoch": 0.09154638382167687,
      "grad_norm": 661.5624953738348,
      "learning_rate": 1.9667046254119878e-05,
      "loss": 468.8927,
      "step": 4760
    },
    {
      "epoch": 0.0917387081574367,
      "grad_norm": 547.2739020006313,
      "learning_rate": 1.966548269113309e-05,
      "loss": 478.5688,
      "step": 4770
    },
    {
      "epoch": 0.09193103249319652,
      "grad_norm": 588.1387559099504,
      "learning_rate": 1.9663915527913628e-05,
      "loss": 480.806,
      "step": 4780
    },
    {
      "epoch": 0.09212335682895635,
      "grad_norm": 622.5556008004595,
      "learning_rate": 1.9662344765045237e-05,
      "loss": 486.7343,
      "step": 4790
    },
    {
      "epoch": 0.09231568116471618,
      "grad_norm": 588.7538340142726,
      "learning_rate": 1.9660770403112996e-05,
      "loss": 477.6282,
      "step": 4800
    },
    {
      "epoch": 0.092508005500476,
      "grad_norm": 599.5534553495175,
      "learning_rate": 1.9659192442703336e-05,
      "loss": 480.692,
      "step": 4810
    },
    {
      "epoch": 0.09270032983623583,
      "grad_norm": 629.8478171353657,
      "learning_rate": 1.965761088440402e-05,
      "loss": 482.3751,
      "step": 4820
    },
    {
      "epoch": 0.09289265417199566,
      "grad_norm": 574.2939691622477,
      "learning_rate": 1.9656025728804147e-05,
      "loss": 479.2723,
      "step": 4830
    },
    {
      "epoch": 0.09308497850775548,
      "grad_norm": 571.5501890537847,
      "learning_rate": 1.9654436976494165e-05,
      "loss": 472.0582,
      "step": 4840
    },
    {
      "epoch": 0.0932773028435153,
      "grad_norm": 581.3352572393229,
      "learning_rate": 1.9652844628065857e-05,
      "loss": 465.4456,
      "step": 4850
    },
    {
      "epoch": 0.09346962717927514,
      "grad_norm": 600.1076490653732,
      "learning_rate": 1.965124868411235e-05,
      "loss": 484.4775,
      "step": 4860
    },
    {
      "epoch": 0.09366195151503495,
      "grad_norm": 632.8925638683768,
      "learning_rate": 1.96496491452281e-05,
      "loss": 468.282,
      "step": 4870
    },
    {
      "epoch": 0.09385427585079478,
      "grad_norm": 623.0519412152863,
      "learning_rate": 1.9648046012008916e-05,
      "loss": 466.2744,
      "step": 4880
    },
    {
      "epoch": 0.0940466001865546,
      "grad_norm": 588.051955241533,
      "learning_rate": 1.9646439285051936e-05,
      "loss": 477.8211,
      "step": 4890
    },
    {
      "epoch": 0.09423892452231443,
      "grad_norm": 1277.4530072499047,
      "learning_rate": 1.9644828964955633e-05,
      "loss": 481.9776,
      "step": 4900
    },
    {
      "epoch": 0.09443124885807426,
      "grad_norm": 634.1287597467085,
      "learning_rate": 1.9643215052319836e-05,
      "loss": 486.0713,
      "step": 4910
    },
    {
      "epoch": 0.09462357319383408,
      "grad_norm": 643.2370525932306,
      "learning_rate": 1.9641597547745694e-05,
      "loss": 489.6987,
      "step": 4920
    },
    {
      "epoch": 0.0948158975295939,
      "grad_norm": 577.7070532458126,
      "learning_rate": 1.9639976451835698e-05,
      "loss": 488.959,
      "step": 4930
    },
    {
      "epoch": 0.09500822186535374,
      "grad_norm": 602.5186322884324,
      "learning_rate": 1.9638351765193685e-05,
      "loss": 482.8837,
      "step": 4940
    },
    {
      "epoch": 0.09520054620111355,
      "grad_norm": 581.9110060326702,
      "learning_rate": 1.9636723488424823e-05,
      "loss": 459.0135,
      "step": 4950
    },
    {
      "epoch": 0.09539287053687338,
      "grad_norm": 598.9069945196771,
      "learning_rate": 1.9635091622135616e-05,
      "loss": 478.5167,
      "step": 4960
    },
    {
      "epoch": 0.09558519487263321,
      "grad_norm": 554.4119378962503,
      "learning_rate": 1.963345616693391e-05,
      "loss": 471.1859,
      "step": 4970
    },
    {
      "epoch": 0.09577751920839303,
      "grad_norm": 633.1868778820232,
      "learning_rate": 1.9631817123428883e-05,
      "loss": 478.766,
      "step": 4980
    },
    {
      "epoch": 0.09596984354415286,
      "grad_norm": 671.3344848959622,
      "learning_rate": 1.9630174492231052e-05,
      "loss": 470.221,
      "step": 4990
    },
    {
      "epoch": 0.09616216787991269,
      "grad_norm": 589.7150538203424,
      "learning_rate": 1.962852827395227e-05,
      "loss": 491.5578,
      "step": 5000
    },
    {
      "epoch": 0.09635449221567251,
      "grad_norm": 587.107949743048,
      "learning_rate": 1.962687846920573e-05,
      "loss": 472.3505,
      "step": 5010
    },
    {
      "epoch": 0.09654681655143234,
      "grad_norm": 666.7828972123168,
      "learning_rate": 1.9625225078605946e-05,
      "loss": 468.2737,
      "step": 5020
    },
    {
      "epoch": 0.09673914088719215,
      "grad_norm": 555.3826730095677,
      "learning_rate": 1.9623568102768792e-05,
      "loss": 473.1111,
      "step": 5030
    },
    {
      "epoch": 0.09693146522295198,
      "grad_norm": 612.9316558466332,
      "learning_rate": 1.9621907542311457e-05,
      "loss": 473.7984,
      "step": 5040
    },
    {
      "epoch": 0.09712378955871181,
      "grad_norm": 631.319563204153,
      "learning_rate": 1.9620243397852473e-05,
      "loss": 457.7611,
      "step": 5050
    },
    {
      "epoch": 0.09731611389447163,
      "grad_norm": 585.1780378158704,
      "learning_rate": 1.9618575670011705e-05,
      "loss": 457.3799,
      "step": 5060
    },
    {
      "epoch": 0.09750843823023146,
      "grad_norm": 602.3842116350421,
      "learning_rate": 1.9616904359410357e-05,
      "loss": 465.7362,
      "step": 5070
    },
    {
      "epoch": 0.09770076256599129,
      "grad_norm": 532.7673408628551,
      "learning_rate": 1.9615229466670963e-05,
      "loss": 470.901,
      "step": 5080
    },
    {
      "epoch": 0.09789308690175111,
      "grad_norm": 559.1806780182956,
      "learning_rate": 1.9613550992417396e-05,
      "loss": 470.3552,
      "step": 5090
    },
    {
      "epoch": 0.09808541123751094,
      "grad_norm": 543.1499297463367,
      "learning_rate": 1.961186893727486e-05,
      "loss": 468.5514,
      "step": 5100
    },
    {
      "epoch": 0.09827773557327077,
      "grad_norm": 654.4171332285497,
      "learning_rate": 1.9610183301869882e-05,
      "loss": 455.5978,
      "step": 5110
    },
    {
      "epoch": 0.09847005990903059,
      "grad_norm": 619.8780298423139,
      "learning_rate": 1.9608494086830348e-05,
      "loss": 462.9577,
      "step": 5120
    },
    {
      "epoch": 0.09866238424479042,
      "grad_norm": 594.9394519211523,
      "learning_rate": 1.9606801292785452e-05,
      "loss": 471.1731,
      "step": 5130
    },
    {
      "epoch": 0.09885470858055025,
      "grad_norm": 564.264842204434,
      "learning_rate": 1.960510492036574e-05,
      "loss": 471.615,
      "step": 5140
    },
    {
      "epoch": 0.09904703291631006,
      "grad_norm": 600.5482761333604,
      "learning_rate": 1.9603404970203078e-05,
      "loss": 475.2653,
      "step": 5150
    },
    {
      "epoch": 0.09923935725206989,
      "grad_norm": 585.9806699820225,
      "learning_rate": 1.9601701442930667e-05,
      "loss": 482.664,
      "step": 5160
    },
    {
      "epoch": 0.09943168158782971,
      "grad_norm": 633.2626147438331,
      "learning_rate": 1.9599994339183047e-05,
      "loss": 475.1929,
      "step": 5170
    },
    {
      "epoch": 0.09962400592358954,
      "grad_norm": 663.6061801987727,
      "learning_rate": 1.9598283659596084e-05,
      "loss": 461.7257,
      "step": 5180
    },
    {
      "epoch": 0.09981633025934937,
      "grad_norm": 705.174353073422,
      "learning_rate": 1.9596569404806983e-05,
      "loss": 462.4165,
      "step": 5190
    },
    {
      "epoch": 0.10000865459510919,
      "grad_norm": 604.372733532367,
      "learning_rate": 1.9594851575454266e-05,
      "loss": 460.5077,
      "step": 5200
    },
    {
      "epoch": 0.10020097893086902,
      "grad_norm": 539.4772132662268,
      "learning_rate": 1.9593130172177806e-05,
      "loss": 480.9955,
      "step": 5210
    },
    {
      "epoch": 0.10039330326662885,
      "grad_norm": 644.937214270631,
      "learning_rate": 1.959140519561879e-05,
      "loss": 471.4614,
      "step": 5220
    },
    {
      "epoch": 0.10058562760238866,
      "grad_norm": 619.1647634857868,
      "learning_rate": 1.9589676646419744e-05,
      "loss": 473.1965,
      "step": 5230
    },
    {
      "epoch": 0.1007779519381485,
      "grad_norm": 552.5511189129061,
      "learning_rate": 1.958794452522453e-05,
      "loss": 477.8996,
      "step": 5240
    },
    {
      "epoch": 0.10097027627390832,
      "grad_norm": 614.2803353963909,
      "learning_rate": 1.9586208832678328e-05,
      "loss": 464.7093,
      "step": 5250
    },
    {
      "epoch": 0.10116260060966814,
      "grad_norm": 598.2840360964544,
      "learning_rate": 1.958446956942766e-05,
      "loss": 471.2755,
      "step": 5260
    },
    {
      "epoch": 0.10135492494542797,
      "grad_norm": 554.1458141527467,
      "learning_rate": 1.9582726736120365e-05,
      "loss": 464.1896,
      "step": 5270
    },
    {
      "epoch": 0.1015472492811878,
      "grad_norm": 583.5294180046756,
      "learning_rate": 1.958098033340563e-05,
      "loss": 468.7423,
      "step": 5280
    },
    {
      "epoch": 0.10173957361694762,
      "grad_norm": 614.4101624241066,
      "learning_rate": 1.9579230361933952e-05,
      "loss": 483.4132,
      "step": 5290
    },
    {
      "epoch": 0.10193189795270745,
      "grad_norm": 629.9302868086264,
      "learning_rate": 1.9577476822357174e-05,
      "loss": 458.8688,
      "step": 5300
    },
    {
      "epoch": 0.10212422228846726,
      "grad_norm": 552.3628393502764,
      "learning_rate": 1.9575719715328457e-05,
      "loss": 480.5301,
      "step": 5310
    },
    {
      "epoch": 0.1023165466242271,
      "grad_norm": 618.3053732075059,
      "learning_rate": 1.957395904150229e-05,
      "loss": 458.3162,
      "step": 5320
    },
    {
      "epoch": 0.10250887095998693,
      "grad_norm": 628.7291530324881,
      "learning_rate": 1.9572194801534504e-05,
      "loss": 470.3111,
      "step": 5330
    },
    {
      "epoch": 0.10270119529574674,
      "grad_norm": 585.4955338752067,
      "learning_rate": 1.9570426996082238e-05,
      "loss": 469.1822,
      "step": 5340
    },
    {
      "epoch": 0.10289351963150657,
      "grad_norm": 594.2863616162359,
      "learning_rate": 1.9568655625803982e-05,
      "loss": 458.7862,
      "step": 5350
    },
    {
      "epoch": 0.1030858439672664,
      "grad_norm": 553.7806659575447,
      "learning_rate": 1.956688069135954e-05,
      "loss": 470.0884,
      "step": 5360
    },
    {
      "epoch": 0.10327816830302622,
      "grad_norm": 557.6974868851648,
      "learning_rate": 1.9565102193410035e-05,
      "loss": 462.1694,
      "step": 5370
    },
    {
      "epoch": 0.10347049263878605,
      "grad_norm": 569.6451341412975,
      "learning_rate": 1.956332013261794e-05,
      "loss": 465.1062,
      "step": 5380
    },
    {
      "epoch": 0.10366281697454588,
      "grad_norm": 615.4239535877139,
      "learning_rate": 1.9561534509647038e-05,
      "loss": 452.582,
      "step": 5390
    },
    {
      "epoch": 0.1038551413103057,
      "grad_norm": 664.4226117018029,
      "learning_rate": 1.9559745325162445e-05,
      "loss": 477.0128,
      "step": 5400
    },
    {
      "epoch": 0.10404746564606553,
      "grad_norm": 569.4112852430824,
      "learning_rate": 1.9557952579830604e-05,
      "loss": 468.2923,
      "step": 5410
    },
    {
      "epoch": 0.10423978998182536,
      "grad_norm": 504.5785090520457,
      "learning_rate": 1.955615627431928e-05,
      "loss": 468.452,
      "step": 5420
    },
    {
      "epoch": 0.10443211431758517,
      "grad_norm": 577.922135541517,
      "learning_rate": 1.955435640929757e-05,
      "loss": 452.7184,
      "step": 5430
    },
    {
      "epoch": 0.104624438653345,
      "grad_norm": 573.4704796404985,
      "learning_rate": 1.9552552985435893e-05,
      "loss": 454.6495,
      "step": 5440
    },
    {
      "epoch": 0.10481676298910482,
      "grad_norm": 591.1003705233481,
      "learning_rate": 1.9550746003405996e-05,
      "loss": 460.658,
      "step": 5450
    },
    {
      "epoch": 0.10500908732486465,
      "grad_norm": 559.711833742097,
      "learning_rate": 1.9548935463880945e-05,
      "loss": 472.0356,
      "step": 5460
    },
    {
      "epoch": 0.10520141166062448,
      "grad_norm": 731.4096514399223,
      "learning_rate": 1.9547121367535143e-05,
      "loss": 475.5825,
      "step": 5470
    },
    {
      "epoch": 0.1053937359963843,
      "grad_norm": 675.1186885970267,
      "learning_rate": 1.9545303715044305e-05,
      "loss": 472.9401,
      "step": 5480
    },
    {
      "epoch": 0.10558606033214413,
      "grad_norm": 621.5446954509481,
      "learning_rate": 1.9543482507085484e-05,
      "loss": 470.6092,
      "step": 5490
    },
    {
      "epoch": 0.10577838466790396,
      "grad_norm": 564.8005480639721,
      "learning_rate": 1.9541657744337038e-05,
      "loss": 457.4794,
      "step": 5500
    },
    {
      "epoch": 0.10597070900366377,
      "grad_norm": 557.3165313706415,
      "learning_rate": 1.9539829427478675e-05,
      "loss": 472.0952,
      "step": 5510
    },
    {
      "epoch": 0.1061630333394236,
      "grad_norm": 596.9725020294912,
      "learning_rate": 1.95379975571914e-05,
      "loss": 477.8515,
      "step": 5520
    },
    {
      "epoch": 0.10635535767518343,
      "grad_norm": 610.5795461524461,
      "learning_rate": 1.953616213415756e-05,
      "loss": 465.1081,
      "step": 5530
    },
    {
      "epoch": 0.10654768201094325,
      "grad_norm": 532.9527100251266,
      "learning_rate": 1.9534323159060824e-05,
      "loss": 477.3551,
      "step": 5540
    },
    {
      "epoch": 0.10674000634670308,
      "grad_norm": 618.7761228781006,
      "learning_rate": 1.9532480632586175e-05,
      "loss": 479.9234,
      "step": 5550
    },
    {
      "epoch": 0.10693233068246291,
      "grad_norm": 586.1059966469652,
      "learning_rate": 1.953063455541992e-05,
      "loss": 476.143,
      "step": 5560
    },
    {
      "epoch": 0.10712465501822273,
      "grad_norm": 644.2358708785018,
      "learning_rate": 1.9528784928249703e-05,
      "loss": 460.9408,
      "step": 5570
    },
    {
      "epoch": 0.10731697935398256,
      "grad_norm": 561.5106647817859,
      "learning_rate": 1.9526931751764467e-05,
      "loss": 450.0845,
      "step": 5580
    },
    {
      "epoch": 0.10750930368974237,
      "grad_norm": 586.2308262734489,
      "learning_rate": 1.95250750266545e-05,
      "loss": 464.2511,
      "step": 5590
    },
    {
      "epoch": 0.1077016280255022,
      "grad_norm": 713.7751058815495,
      "learning_rate": 1.9523214753611398e-05,
      "loss": 450.9627,
      "step": 5600
    },
    {
      "epoch": 0.10789395236126204,
      "grad_norm": 563.3784121067714,
      "learning_rate": 1.952135093332808e-05,
      "loss": 474.7403,
      "step": 5610
    },
    {
      "epoch": 0.10808627669702185,
      "grad_norm": 541.5234523770434,
      "learning_rate": 1.9519483566498788e-05,
      "loss": 459.177,
      "step": 5620
    },
    {
      "epoch": 0.10827860103278168,
      "grad_norm": 533.0503444636863,
      "learning_rate": 1.9517612653819088e-05,
      "loss": 462.7783,
      "step": 5630
    },
    {
      "epoch": 0.10847092536854151,
      "grad_norm": 601.3699334313936,
      "learning_rate": 1.9515738195985868e-05,
      "loss": 479.1894,
      "step": 5640
    },
    {
      "epoch": 0.10866324970430133,
      "grad_norm": 565.7627426136539,
      "learning_rate": 1.951386019369732e-05,
      "loss": 458.9444,
      "step": 5650
    },
    {
      "epoch": 0.10885557404006116,
      "grad_norm": 601.6013229962927,
      "learning_rate": 1.9511978647652984e-05,
      "loss": 453.5309,
      "step": 5660
    },
    {
      "epoch": 0.10904789837582099,
      "grad_norm": 572.5979504101963,
      "learning_rate": 1.9510093558553687e-05,
      "loss": 460.8308,
      "step": 5670
    },
    {
      "epoch": 0.1092402227115808,
      "grad_norm": 555.5591057204285,
      "learning_rate": 1.950820492710161e-05,
      "loss": 457.2931,
      "step": 5680
    },
    {
      "epoch": 0.10943254704734064,
      "grad_norm": 627.0639773158829,
      "learning_rate": 1.9506312754000235e-05,
      "loss": 460.4723,
      "step": 5690
    },
    {
      "epoch": 0.10962487138310047,
      "grad_norm": 567.7280179595467,
      "learning_rate": 1.9504417039954357e-05,
      "loss": 454.7062,
      "step": 5700
    },
    {
      "epoch": 0.10981719571886028,
      "grad_norm": 589.7552420184576,
      "learning_rate": 1.9502517785670098e-05,
      "loss": 472.0097,
      "step": 5710
    },
    {
      "epoch": 0.11000952005462011,
      "grad_norm": 609.0864005083679,
      "learning_rate": 1.950061499185491e-05,
      "loss": 472.8547,
      "step": 5720
    },
    {
      "epoch": 0.11020184439037993,
      "grad_norm": 574.8393854788927,
      "learning_rate": 1.9498708659217542e-05,
      "loss": 458.5378,
      "step": 5730
    },
    {
      "epoch": 0.11039416872613976,
      "grad_norm": 593.3147818039952,
      "learning_rate": 1.9496798788468077e-05,
      "loss": 451.2261,
      "step": 5740
    },
    {
      "epoch": 0.11058649306189959,
      "grad_norm": 591.8585009842711,
      "learning_rate": 1.9494885380317906e-05,
      "loss": 453.768,
      "step": 5750
    },
    {
      "epoch": 0.11077881739765941,
      "grad_norm": 556.8757756377022,
      "learning_rate": 1.9492968435479744e-05,
      "loss": 450.8655,
      "step": 5760
    },
    {
      "epoch": 0.11097114173341924,
      "grad_norm": 567.0838774827757,
      "learning_rate": 1.949104795466762e-05,
      "loss": 453.8706,
      "step": 5770
    },
    {
      "epoch": 0.11116346606917907,
      "grad_norm": 572.239523238843,
      "learning_rate": 1.9489123938596886e-05,
      "loss": 444.5173,
      "step": 5780
    },
    {
      "epoch": 0.11135579040493888,
      "grad_norm": 572.460747506397,
      "learning_rate": 1.94871963879842e-05,
      "loss": 460.8298,
      "step": 5790
    },
    {
      "epoch": 0.11154811474069871,
      "grad_norm": 547.762611217189,
      "learning_rate": 1.9485265303547547e-05,
      "loss": 456.2129,
      "step": 5800
    },
    {
      "epoch": 0.11174043907645854,
      "grad_norm": 681.0375694733043,
      "learning_rate": 1.9483330686006223e-05,
      "loss": 463.6338,
      "step": 5810
    },
    {
      "epoch": 0.11193276341221836,
      "grad_norm": 685.4994674000491,
      "learning_rate": 1.948139253608084e-05,
      "loss": 486.1659,
      "step": 5820
    },
    {
      "epoch": 0.11212508774797819,
      "grad_norm": 598.1425119667529,
      "learning_rate": 1.9479450854493327e-05,
      "loss": 456.0472,
      "step": 5830
    },
    {
      "epoch": 0.11231741208373802,
      "grad_norm": 570.9363313117783,
      "learning_rate": 1.9477505641966933e-05,
      "loss": 451.9121,
      "step": 5840
    },
    {
      "epoch": 0.11250973641949784,
      "grad_norm": 577.1784111188559,
      "learning_rate": 1.9475556899226213e-05,
      "loss": 460.049,
      "step": 5850
    },
    {
      "epoch": 0.11270206075525767,
      "grad_norm": 607.1651708353087,
      "learning_rate": 1.9473604626997037e-05,
      "loss": 451.1603,
      "step": 5860
    },
    {
      "epoch": 0.11289438509101749,
      "grad_norm": 575.331530518834,
      "learning_rate": 1.94716488260066e-05,
      "loss": 458.459,
      "step": 5870
    },
    {
      "epoch": 0.11308670942677732,
      "grad_norm": 608.6725118986951,
      "learning_rate": 1.946968949698341e-05,
      "loss": 456.1864,
      "step": 5880
    },
    {
      "epoch": 0.11327903376253715,
      "grad_norm": 645.1061651579687,
      "learning_rate": 1.9467726640657277e-05,
      "loss": 450.295,
      "step": 5890
    },
    {
      "epoch": 0.11347135809829696,
      "grad_norm": 586.8906624550228,
      "learning_rate": 1.9465760257759336e-05,
      "loss": 455.3526,
      "step": 5900
    },
    {
      "epoch": 0.11366368243405679,
      "grad_norm": 612.8186262513243,
      "learning_rate": 1.9463790349022027e-05,
      "loss": 446.5723,
      "step": 5910
    },
    {
      "epoch": 0.11385600676981662,
      "grad_norm": 526.1600416399643,
      "learning_rate": 1.9461816915179117e-05,
      "loss": 456.4721,
      "step": 5920
    },
    {
      "epoch": 0.11404833110557644,
      "grad_norm": 624.0233025646951,
      "learning_rate": 1.945983995696567e-05,
      "loss": 462.4404,
      "step": 5930
    },
    {
      "epoch": 0.11424065544133627,
      "grad_norm": 597.7485044490996,
      "learning_rate": 1.9457859475118077e-05,
      "loss": 454.947,
      "step": 5940
    },
    {
      "epoch": 0.1144329797770961,
      "grad_norm": 582.2865791588138,
      "learning_rate": 1.9455875470374027e-05,
      "loss": 469.912,
      "step": 5950
    },
    {
      "epoch": 0.11462530411285592,
      "grad_norm": 527.2170495374962,
      "learning_rate": 1.9453887943472532e-05,
      "loss": 440.7697,
      "step": 5960
    },
    {
      "epoch": 0.11481762844861575,
      "grad_norm": 609.9048837986057,
      "learning_rate": 1.945189689515392e-05,
      "loss": 467.5322,
      "step": 5970
    },
    {
      "epoch": 0.11500995278437558,
      "grad_norm": 517.0803414211166,
      "learning_rate": 1.9449902326159815e-05,
      "loss": 443.7057,
      "step": 5980
    },
    {
      "epoch": 0.1152022771201354,
      "grad_norm": 559.1504196549349,
      "learning_rate": 1.9447904237233164e-05,
      "loss": 457.1309,
      "step": 5990
    },
    {
      "epoch": 0.11539460145589522,
      "grad_norm": 717.0088099356094,
      "learning_rate": 1.9445902629118223e-05,
      "loss": 451.0554,
      "step": 6000
    },
    {
      "epoch": 0.11558692579165504,
      "grad_norm": 553.1601859777684,
      "learning_rate": 1.9443897502560555e-05,
      "loss": 465.8052,
      "step": 6010
    },
    {
      "epoch": 0.11577925012741487,
      "grad_norm": 556.5726901982019,
      "learning_rate": 1.9441888858307042e-05,
      "loss": 476.5284,
      "step": 6020
    },
    {
      "epoch": 0.1159715744631747,
      "grad_norm": 529.381321358997,
      "learning_rate": 1.943987669710586e-05,
      "loss": 444.8907,
      "step": 6030
    },
    {
      "epoch": 0.11616389879893452,
      "grad_norm": 571.4391456040348,
      "learning_rate": 1.9437861019706522e-05,
      "loss": 445.6038,
      "step": 6040
    },
    {
      "epoch": 0.11635622313469435,
      "grad_norm": 609.125673479725,
      "learning_rate": 1.943584182685982e-05,
      "loss": 463.3053,
      "step": 6050
    },
    {
      "epoch": 0.11654854747045418,
      "grad_norm": 865.9092772804071,
      "learning_rate": 1.9433819119317878e-05,
      "loss": 456.327,
      "step": 6060
    },
    {
      "epoch": 0.116740871806214,
      "grad_norm": 609.1420721393692,
      "learning_rate": 1.9431792897834115e-05,
      "loss": 449.2634,
      "step": 6070
    },
    {
      "epoch": 0.11693319614197382,
      "grad_norm": 602.5476476123913,
      "learning_rate": 1.9429763163163273e-05,
      "loss": 467.8189,
      "step": 6080
    },
    {
      "epoch": 0.11712552047773366,
      "grad_norm": 545.4484795961454,
      "learning_rate": 1.942772991606139e-05,
      "loss": 441.9267,
      "step": 6090
    },
    {
      "epoch": 0.11731784481349347,
      "grad_norm": 736.771299779919,
      "learning_rate": 1.9425693157285816e-05,
      "loss": 454.4545,
      "step": 6100
    },
    {
      "epoch": 0.1175101691492533,
      "grad_norm": 592.854521574643,
      "learning_rate": 1.942365288759521e-05,
      "loss": 443.1416,
      "step": 6110
    },
    {
      "epoch": 0.11770249348501313,
      "grad_norm": 556.0528704358943,
      "learning_rate": 1.9421609107749542e-05,
      "loss": 464.8642,
      "step": 6120
    },
    {
      "epoch": 0.11789481782077295,
      "grad_norm": 538.044132231871,
      "learning_rate": 1.9419561818510085e-05,
      "loss": 454.9487,
      "step": 6130
    },
    {
      "epoch": 0.11808714215653278,
      "grad_norm": 579.3182249998647,
      "learning_rate": 1.9417511020639416e-05,
      "loss": 455.0798,
      "step": 6140
    },
    {
      "epoch": 0.1182794664922926,
      "grad_norm": 560.7693432823665,
      "learning_rate": 1.9415456714901432e-05,
      "loss": 460.8459,
      "step": 6150
    },
    {
      "epoch": 0.11847179082805243,
      "grad_norm": 548.6833992249758,
      "learning_rate": 1.941339890206132e-05,
      "loss": 459.5579,
      "step": 6160
    },
    {
      "epoch": 0.11866411516381226,
      "grad_norm": 546.2777243774797,
      "learning_rate": 1.9411337582885587e-05,
      "loss": 451.0988,
      "step": 6170
    },
    {
      "epoch": 0.11885643949957207,
      "grad_norm": 569.7658880448489,
      "learning_rate": 1.9409272758142034e-05,
      "loss": 442.799,
      "step": 6180
    },
    {
      "epoch": 0.1190487638353319,
      "grad_norm": 598.5469988105174,
      "learning_rate": 1.940720442859978e-05,
      "loss": 454.6981,
      "step": 6190
    },
    {
      "epoch": 0.11924108817109173,
      "grad_norm": 520.1388083901044,
      "learning_rate": 1.940513259502924e-05,
      "loss": 455.6266,
      "step": 6200
    },
    {
      "epoch": 0.11943341250685155,
      "grad_norm": 592.2642859558092,
      "learning_rate": 1.9403057258202144e-05,
      "loss": 452.3042,
      "step": 6210
    },
    {
      "epoch": 0.11962573684261138,
      "grad_norm": 538.2159895835466,
      "learning_rate": 1.940097841889151e-05,
      "loss": 449.2261,
      "step": 6220
    },
    {
      "epoch": 0.11981806117837121,
      "grad_norm": 555.7028374629331,
      "learning_rate": 1.939889607787168e-05,
      "loss": 461.0196,
      "step": 6230
    },
    {
      "epoch": 0.12001038551413103,
      "grad_norm": 581.2120867911883,
      "learning_rate": 1.9396810235918287e-05,
      "loss": 456.7599,
      "step": 6240
    },
    {
      "epoch": 0.12020270984989086,
      "grad_norm": 585.3890168566849,
      "learning_rate": 1.939472089380828e-05,
      "loss": 450.5199,
      "step": 6250
    },
    {
      "epoch": 0.12039503418565069,
      "grad_norm": 537.6026806644555,
      "learning_rate": 1.9392628052319895e-05,
      "loss": 457.9451,
      "step": 6260
    },
    {
      "epoch": 0.1205873585214105,
      "grad_norm": 582.1901000546167,
      "learning_rate": 1.9390531712232687e-05,
      "loss": 454.9115,
      "step": 6270
    },
    {
      "epoch": 0.12077968285717033,
      "grad_norm": 556.1801994889596,
      "learning_rate": 1.9388431874327504e-05,
      "loss": 468.9773,
      "step": 6280
    },
    {
      "epoch": 0.12097200719293016,
      "grad_norm": 655.9631280020292,
      "learning_rate": 1.9386328539386502e-05,
      "loss": 458.9917,
      "step": 6290
    },
    {
      "epoch": 0.12116433152868998,
      "grad_norm": 536.1963488882964,
      "learning_rate": 1.938422170819314e-05,
      "loss": 458.6655,
      "step": 6300
    },
    {
      "epoch": 0.12135665586444981,
      "grad_norm": 566.4790735212034,
      "learning_rate": 1.938211138153218e-05,
      "loss": 444.7368,
      "step": 6310
    },
    {
      "epoch": 0.12154898020020963,
      "grad_norm": 540.2566765307556,
      "learning_rate": 1.9379997560189677e-05,
      "loss": 450.3039,
      "step": 6320
    },
    {
      "epoch": 0.12174130453596946,
      "grad_norm": 588.4333892241063,
      "learning_rate": 1.9377880244953e-05,
      "loss": 447.3329,
      "step": 6330
    },
    {
      "epoch": 0.12193362887172929,
      "grad_norm": 559.2034383246234,
      "learning_rate": 1.9375759436610813e-05,
      "loss": 456.1573,
      "step": 6340
    },
    {
      "epoch": 0.1221259532074891,
      "grad_norm": 675.705887382901,
      "learning_rate": 1.937363513595308e-05,
      "loss": 456.2759,
      "step": 6350
    },
    {
      "epoch": 0.12231827754324893,
      "grad_norm": 557.1691693464365,
      "learning_rate": 1.937150734377107e-05,
      "loss": 439.1136,
      "step": 6360
    },
    {
      "epoch": 0.12251060187900877,
      "grad_norm": 529.5951367319399,
      "learning_rate": 1.9369376060857354e-05,
      "loss": 441.9863,
      "step": 6370
    },
    {
      "epoch": 0.12270292621476858,
      "grad_norm": 574.2605059738312,
      "learning_rate": 1.936724128800579e-05,
      "loss": 460.2209,
      "step": 6380
    },
    {
      "epoch": 0.12289525055052841,
      "grad_norm": 553.2665262790448,
      "learning_rate": 1.9365103026011555e-05,
      "loss": 453.9132,
      "step": 6390
    },
    {
      "epoch": 0.12308757488628824,
      "grad_norm": 605.4233027827128,
      "learning_rate": 1.9362961275671112e-05,
      "loss": 457.6689,
      "step": 6400
    },
    {
      "epoch": 0.12327989922204806,
      "grad_norm": 586.7659932436243,
      "learning_rate": 1.936081603778223e-05,
      "loss": 467.9555,
      "step": 6410
    },
    {
      "epoch": 0.12347222355780789,
      "grad_norm": 579.1158928957905,
      "learning_rate": 1.9358667313143972e-05,
      "loss": 455.3643,
      "step": 6420
    },
    {
      "epoch": 0.12366454789356772,
      "grad_norm": 599.4161547485973,
      "learning_rate": 1.93565151025567e-05,
      "loss": 454.3246,
      "step": 6430
    },
    {
      "epoch": 0.12385687222932754,
      "grad_norm": 590.9679448435286,
      "learning_rate": 1.9354359406822084e-05,
      "loss": 459.4577,
      "step": 6440
    },
    {
      "epoch": 0.12404919656508737,
      "grad_norm": 570.0842259149974,
      "learning_rate": 1.935220022674308e-05,
      "loss": 445.058,
      "step": 6450
    },
    {
      "epoch": 0.12424152090084718,
      "grad_norm": 574.0282928884872,
      "learning_rate": 1.9350037563123947e-05,
      "loss": 437.3688,
      "step": 6460
    },
    {
      "epoch": 0.12443384523660701,
      "grad_norm": 567.5621462821342,
      "learning_rate": 1.9347871416770245e-05,
      "loss": 457.1672,
      "step": 6470
    },
    {
      "epoch": 0.12462616957236684,
      "grad_norm": 627.125907188958,
      "learning_rate": 1.9345701788488825e-05,
      "loss": 443.653,
      "step": 6480
    },
    {
      "epoch": 0.12481849390812666,
      "grad_norm": 528.493118446666,
      "learning_rate": 1.9343528679087837e-05,
      "loss": 452.8219,
      "step": 6490
    },
    {
      "epoch": 0.1250108182438865,
      "grad_norm": 589.1564062016865,
      "learning_rate": 1.934135208937673e-05,
      "loss": 451.268,
      "step": 6500
    },
    {
      "epoch": 0.12520314257964632,
      "grad_norm": 588.1787184927141,
      "learning_rate": 1.9339172020166245e-05,
      "loss": 460.5093,
      "step": 6510
    },
    {
      "epoch": 0.12539546691540615,
      "grad_norm": 559.0172907781034,
      "learning_rate": 1.933698847226843e-05,
      "loss": 454.6871,
      "step": 6520
    },
    {
      "epoch": 0.12558779125116595,
      "grad_norm": 522.4693165831601,
      "learning_rate": 1.9334801446496606e-05,
      "loss": 445.7012,
      "step": 6530
    },
    {
      "epoch": 0.12578011558692578,
      "grad_norm": 530.8457545579569,
      "learning_rate": 1.933261094366542e-05,
      "loss": 451.3096,
      "step": 6540
    },
    {
      "epoch": 0.12597243992268561,
      "grad_norm": 562.1488463433628,
      "learning_rate": 1.9330416964590787e-05,
      "loss": 436.5796,
      "step": 6550
    },
    {
      "epoch": 0.12616476425844544,
      "grad_norm": 615.5752112743173,
      "learning_rate": 1.932821951008993e-05,
      "loss": 461.9782,
      "step": 6560
    },
    {
      "epoch": 0.12635708859420527,
      "grad_norm": 610.040926518066,
      "learning_rate": 1.9326018580981373e-05,
      "loss": 447.5911,
      "step": 6570
    },
    {
      "epoch": 0.1265494129299651,
      "grad_norm": 550.5832465583508,
      "learning_rate": 1.9323814178084914e-05,
      "loss": 440.8309,
      "step": 6580
    },
    {
      "epoch": 0.1267417372657249,
      "grad_norm": 591.049276600134,
      "learning_rate": 1.9321606302221662e-05,
      "loss": 454.5273,
      "step": 6590
    },
    {
      "epoch": 0.12693406160148474,
      "grad_norm": 593.8169916602016,
      "learning_rate": 1.9319394954214013e-05,
      "loss": 455.2549,
      "step": 6600
    },
    {
      "epoch": 0.12712638593724457,
      "grad_norm": 620.7210887126705,
      "learning_rate": 1.931718013488566e-05,
      "loss": 451.1783,
      "step": 6610
    },
    {
      "epoch": 0.1273187102730044,
      "grad_norm": 631.7601020832637,
      "learning_rate": 1.9314961845061584e-05,
      "loss": 455.9168,
      "step": 6620
    },
    {
      "epoch": 0.12751103460876423,
      "grad_norm": 790.7017724612514,
      "learning_rate": 1.9312740085568063e-05,
      "loss": 463.5684,
      "step": 6630
    },
    {
      "epoch": 0.12770335894452406,
      "grad_norm": 534.8783070848594,
      "learning_rate": 1.9310514857232666e-05,
      "loss": 440.6818,
      "step": 6640
    },
    {
      "epoch": 0.12789568328028386,
      "grad_norm": 535.2064559744528,
      "learning_rate": 1.930828616088425e-05,
      "loss": 434.7803,
      "step": 6650
    },
    {
      "epoch": 0.1280880076160437,
      "grad_norm": 545.1513186662329,
      "learning_rate": 1.9306053997352973e-05,
      "loss": 442.1026,
      "step": 6660
    },
    {
      "epoch": 0.12828033195180352,
      "grad_norm": 555.7306761069738,
      "learning_rate": 1.9303818367470274e-05,
      "loss": 444.6883,
      "step": 6670
    },
    {
      "epoch": 0.12847265628756335,
      "grad_norm": 541.9976527210671,
      "learning_rate": 1.9301579272068894e-05,
      "loss": 448.924,
      "step": 6680
    },
    {
      "epoch": 0.12866498062332318,
      "grad_norm": 585.6049150429797,
      "learning_rate": 1.9299336711982853e-05,
      "loss": 449.9302,
      "step": 6690
    },
    {
      "epoch": 0.12885730495908299,
      "grad_norm": 575.5148590127932,
      "learning_rate": 1.9297090688047473e-05,
      "loss": 450.3919,
      "step": 6700
    },
    {
      "epoch": 0.12904962929484282,
      "grad_norm": 563.0153338400012,
      "learning_rate": 1.929484120109936e-05,
      "loss": 444.159,
      "step": 6710
    },
    {
      "epoch": 0.12924195363060265,
      "grad_norm": 561.6015685341068,
      "learning_rate": 1.9292588251976404e-05,
      "loss": 439.5263,
      "step": 6720
    },
    {
      "epoch": 0.12943427796636248,
      "grad_norm": 618.0408070492041,
      "learning_rate": 1.92903318415178e-05,
      "loss": 448.6772,
      "step": 6730
    },
    {
      "epoch": 0.1296266023021223,
      "grad_norm": 560.1286072818397,
      "learning_rate": 1.9288071970564015e-05,
      "loss": 460.3758,
      "step": 6740
    },
    {
      "epoch": 0.12981892663788214,
      "grad_norm": 579.6981835780392,
      "learning_rate": 1.9285808639956823e-05,
      "loss": 450.6625,
      "step": 6750
    },
    {
      "epoch": 0.13001125097364194,
      "grad_norm": 569.7116518232411,
      "learning_rate": 1.9283541850539272e-05,
      "loss": 446.0654,
      "step": 6760
    },
    {
      "epoch": 0.13020357530940177,
      "grad_norm": 546.358091447522,
      "learning_rate": 1.92812716031557e-05,
      "loss": 440.4798,
      "step": 6770
    },
    {
      "epoch": 0.1303958996451616,
      "grad_norm": 574.5179906842806,
      "learning_rate": 1.9278997898651746e-05,
      "loss": 451.1808,
      "step": 6780
    },
    {
      "epoch": 0.13058822398092143,
      "grad_norm": 549.0151048075636,
      "learning_rate": 1.9276720737874327e-05,
      "loss": 448.2003,
      "step": 6790
    },
    {
      "epoch": 0.13078054831668126,
      "grad_norm": 572.9689293748356,
      "learning_rate": 1.9274440121671637e-05,
      "loss": 437.8925,
      "step": 6800
    },
    {
      "epoch": 0.13097287265244106,
      "grad_norm": 562.673525311016,
      "learning_rate": 1.9272156050893173e-05,
      "loss": 446.0264,
      "step": 6810
    },
    {
      "epoch": 0.1311651969882009,
      "grad_norm": 658.2692168239895,
      "learning_rate": 1.926986852638972e-05,
      "loss": 447.0198,
      "step": 6820
    },
    {
      "epoch": 0.13135752132396072,
      "grad_norm": 677.0630203293408,
      "learning_rate": 1.926757754901333e-05,
      "loss": 439.7201,
      "step": 6830
    },
    {
      "epoch": 0.13154984565972055,
      "grad_norm": 558.8686078058347,
      "learning_rate": 1.926528311961737e-05,
      "loss": 445.9571,
      "step": 6840
    },
    {
      "epoch": 0.13174216999548038,
      "grad_norm": 554.1681325389345,
      "learning_rate": 1.9262985239056463e-05,
      "loss": 453.8511,
      "step": 6850
    },
    {
      "epoch": 0.13193449433124022,
      "grad_norm": 602.1042896629361,
      "learning_rate": 1.9260683908186544e-05,
      "loss": 442.8998,
      "step": 6860
    },
    {
      "epoch": 0.13212681866700002,
      "grad_norm": 539.0167513606403,
      "learning_rate": 1.9258379127864808e-05,
      "loss": 438.5836,
      "step": 6870
    },
    {
      "epoch": 0.13231914300275985,
      "grad_norm": 612.1029205950629,
      "learning_rate": 1.925607089894976e-05,
      "loss": 440.011,
      "step": 6880
    },
    {
      "epoch": 0.13251146733851968,
      "grad_norm": 545.441470772547,
      "learning_rate": 1.9253759222301168e-05,
      "loss": 445.2561,
      "step": 6890
    },
    {
      "epoch": 0.1327037916742795,
      "grad_norm": 547.8940711851035,
      "learning_rate": 1.9251444098780095e-05,
      "loss": 444.2211,
      "step": 6900
    },
    {
      "epoch": 0.13289611601003934,
      "grad_norm": 567.2072916810628,
      "learning_rate": 1.924912552924889e-05,
      "loss": 444.5792,
      "step": 6910
    },
    {
      "epoch": 0.13308844034579917,
      "grad_norm": 690.1238130681841,
      "learning_rate": 1.924680351457118e-05,
      "loss": 449.0135,
      "step": 6920
    },
    {
      "epoch": 0.13328076468155897,
      "grad_norm": 625.3567629918222,
      "learning_rate": 1.9244478055611875e-05,
      "loss": 435.281,
      "step": 6930
    },
    {
      "epoch": 0.1334730890173188,
      "grad_norm": 562.8921699418266,
      "learning_rate": 1.9242149153237175e-05,
      "loss": 452.8662,
      "step": 6940
    },
    {
      "epoch": 0.13366541335307863,
      "grad_norm": 566.8539377652648,
      "learning_rate": 1.923981680831455e-05,
      "loss": 434.2891,
      "step": 6950
    },
    {
      "epoch": 0.13385773768883846,
      "grad_norm": 529.1648959684138,
      "learning_rate": 1.923748102171277e-05,
      "loss": 449.7102,
      "step": 6960
    },
    {
      "epoch": 0.1340500620245983,
      "grad_norm": 622.7985948956308,
      "learning_rate": 1.9235141794301867e-05,
      "loss": 443.65,
      "step": 6970
    },
    {
      "epoch": 0.1342423863603581,
      "grad_norm": 653.9551519168286,
      "learning_rate": 1.9232799126953173e-05,
      "loss": 446.7727,
      "step": 6980
    },
    {
      "epoch": 0.13443471069611793,
      "grad_norm": 580.8320821440503,
      "learning_rate": 1.9230453020539285e-05,
      "loss": 447.424,
      "step": 6990
    },
    {
      "epoch": 0.13462703503187776,
      "grad_norm": 613.0775728874669,
      "learning_rate": 1.9228103475934096e-05,
      "loss": 434.6862,
      "step": 7000
    },
    {
      "epoch": 0.1348193593676376,
      "grad_norm": 544.3346854360603,
      "learning_rate": 1.9225750494012767e-05,
      "loss": 452.8522,
      "step": 7010
    },
    {
      "epoch": 0.13501168370339742,
      "grad_norm": 589.083557180598,
      "learning_rate": 1.9223394075651748e-05,
      "loss": 449.5531,
      "step": 7020
    },
    {
      "epoch": 0.13520400803915725,
      "grad_norm": 609.831405804266,
      "learning_rate": 1.9221034221728764e-05,
      "loss": 441.9063,
      "step": 7030
    },
    {
      "epoch": 0.13539633237491705,
      "grad_norm": 610.3204446716359,
      "learning_rate": 1.9218670933122826e-05,
      "loss": 446.5802,
      "step": 7040
    },
    {
      "epoch": 0.13558865671067688,
      "grad_norm": 575.5031137801874,
      "learning_rate": 1.9216304210714213e-05,
      "loss": 435.0925,
      "step": 7050
    },
    {
      "epoch": 0.1357809810464367,
      "grad_norm": 568.071511098741,
      "learning_rate": 1.9213934055384498e-05,
      "loss": 463.8728,
      "step": 7060
    },
    {
      "epoch": 0.13597330538219654,
      "grad_norm": 564.9895317015021,
      "learning_rate": 1.9211560468016516e-05,
      "loss": 452.2563,
      "step": 7070
    },
    {
      "epoch": 0.13616562971795637,
      "grad_norm": 601.5690005736805,
      "learning_rate": 1.9209183449494397e-05,
      "loss": 437.316,
      "step": 7080
    },
    {
      "epoch": 0.13635795405371617,
      "grad_norm": 612.0170495369792,
      "learning_rate": 1.9206803000703534e-05,
      "loss": 447.462,
      "step": 7090
    },
    {
      "epoch": 0.136550278389476,
      "grad_norm": 536.0231868247605,
      "learning_rate": 1.9204419122530614e-05,
      "loss": 442.7653,
      "step": 7100
    },
    {
      "epoch": 0.13674260272523583,
      "grad_norm": 569.45332421543,
      "learning_rate": 1.9202031815863583e-05,
      "loss": 443.6591,
      "step": 7110
    },
    {
      "epoch": 0.13693492706099566,
      "grad_norm": 533.1756418302409,
      "learning_rate": 1.919964108159168e-05,
      "loss": 445.0802,
      "step": 7120
    },
    {
      "epoch": 0.1371272513967555,
      "grad_norm": 526.6507440457178,
      "learning_rate": 1.9197246920605408e-05,
      "loss": 432.5921,
      "step": 7130
    },
    {
      "epoch": 0.13731957573251533,
      "grad_norm": 545.0966495594668,
      "learning_rate": 1.9194849333796557e-05,
      "loss": 440.4764,
      "step": 7140
    },
    {
      "epoch": 0.13751190006827513,
      "grad_norm": 565.1981793829112,
      "learning_rate": 1.9192448322058187e-05,
      "loss": 449.5943,
      "step": 7150
    },
    {
      "epoch": 0.13770422440403496,
      "grad_norm": 516.7266678204488,
      "learning_rate": 1.9190043886284635e-05,
      "loss": 443.0106,
      "step": 7160
    },
    {
      "epoch": 0.1378965487397948,
      "grad_norm": 595.3751727839116,
      "learning_rate": 1.9187636027371518e-05,
      "loss": 433.9862,
      "step": 7170
    },
    {
      "epoch": 0.13808887307555462,
      "grad_norm": 658.1726007929548,
      "learning_rate": 1.9185224746215714e-05,
      "loss": 457.4265,
      "step": 7180
    },
    {
      "epoch": 0.13828119741131445,
      "grad_norm": 602.2515559857055,
      "learning_rate": 1.9182810043715394e-05,
      "loss": 438.1577,
      "step": 7190
    },
    {
      "epoch": 0.13847352174707428,
      "grad_norm": 523.6108092055384,
      "learning_rate": 1.9180391920769993e-05,
      "loss": 438.2888,
      "step": 7200
    },
    {
      "epoch": 0.13866584608283408,
      "grad_norm": 577.8699183900699,
      "learning_rate": 1.9177970378280215e-05,
      "loss": 441.081,
      "step": 7210
    },
    {
      "epoch": 0.1388581704185939,
      "grad_norm": 555.7984157792681,
      "learning_rate": 1.9175545417148056e-05,
      "loss": 449.9366,
      "step": 7220
    },
    {
      "epoch": 0.13905049475435374,
      "grad_norm": 601.0036850989364,
      "learning_rate": 1.9173117038276766e-05,
      "loss": 445.1274,
      "step": 7230
    },
    {
      "epoch": 0.13924281909011357,
      "grad_norm": 649.1687776871014,
      "learning_rate": 1.9170685242570878e-05,
      "loss": 450.8615,
      "step": 7240
    },
    {
      "epoch": 0.1394351434258734,
      "grad_norm": 527.2314933455276,
      "learning_rate": 1.9168250030936195e-05,
      "loss": 432.5535,
      "step": 7250
    },
    {
      "epoch": 0.1396274677616332,
      "grad_norm": 532.4805675864951,
      "learning_rate": 1.91658114042798e-05,
      "loss": 443.7549,
      "step": 7260
    },
    {
      "epoch": 0.13981979209739304,
      "grad_norm": 580.159318741437,
      "learning_rate": 1.9163369363510026e-05,
      "loss": 435.2532,
      "step": 7270
    },
    {
      "epoch": 0.14001211643315287,
      "grad_norm": 578.2713835246826,
      "learning_rate": 1.916092390953651e-05,
      "loss": 441.5688,
      "step": 7280
    },
    {
      "epoch": 0.1402044407689127,
      "grad_norm": 558.372779805351,
      "learning_rate": 1.915847504327013e-05,
      "loss": 439.2564,
      "step": 7290
    },
    {
      "epoch": 0.14039676510467253,
      "grad_norm": 529.1449548373271,
      "learning_rate": 1.9156022765623057e-05,
      "loss": 439.4631,
      "step": 7300
    },
    {
      "epoch": 0.14058908944043236,
      "grad_norm": 545.7670719788337,
      "learning_rate": 1.9153567077508718e-05,
      "loss": 432.0012,
      "step": 7310
    },
    {
      "epoch": 0.14078141377619216,
      "grad_norm": 595.2851457029889,
      "learning_rate": 1.9151107979841824e-05,
      "loss": 436.38,
      "step": 7320
    },
    {
      "epoch": 0.140973738111952,
      "grad_norm": 538.6557840713295,
      "learning_rate": 1.9148645473538338e-05,
      "loss": 430.5198,
      "step": 7330
    },
    {
      "epoch": 0.14116606244771182,
      "grad_norm": 546.0718350433435,
      "learning_rate": 1.9146179559515507e-05,
      "loss": 434.9198,
      "step": 7340
    },
    {
      "epoch": 0.14135838678347165,
      "grad_norm": 604.7659032605735,
      "learning_rate": 1.9143710238691847e-05,
      "loss": 435.8765,
      "step": 7350
    },
    {
      "epoch": 0.14155071111923148,
      "grad_norm": 544.1282262601212,
      "learning_rate": 1.9141237511987137e-05,
      "loss": 427.6289,
      "step": 7360
    },
    {
      "epoch": 0.14174303545499128,
      "grad_norm": 557.6461067823977,
      "learning_rate": 1.9138761380322425e-05,
      "loss": 448.1304,
      "step": 7370
    },
    {
      "epoch": 0.14193535979075111,
      "grad_norm": 609.9076211931574,
      "learning_rate": 1.913628184462003e-05,
      "loss": 436.6113,
      "step": 7380
    },
    {
      "epoch": 0.14212768412651094,
      "grad_norm": 705.8008060059544,
      "learning_rate": 1.913379890580354e-05,
      "loss": 441.1925,
      "step": 7390
    },
    {
      "epoch": 0.14232000846227078,
      "grad_norm": 534.110582723359,
      "learning_rate": 1.9131312564797805e-05,
      "loss": 439.0979,
      "step": 7400
    },
    {
      "epoch": 0.1425123327980306,
      "grad_norm": 636.9791987731925,
      "learning_rate": 1.912882282252895e-05,
      "loss": 425.6083,
      "step": 7410
    },
    {
      "epoch": 0.14270465713379044,
      "grad_norm": 556.9531193170636,
      "learning_rate": 1.9126329679924364e-05,
      "loss": 443.8383,
      "step": 7420
    },
    {
      "epoch": 0.14289698146955024,
      "grad_norm": 544.4647112172504,
      "learning_rate": 1.9123833137912693e-05,
      "loss": 440.5781,
      "step": 7430
    },
    {
      "epoch": 0.14308930580531007,
      "grad_norm": 527.4625931596618,
      "learning_rate": 1.9121333197423867e-05,
      "loss": 458.2057,
      "step": 7440
    },
    {
      "epoch": 0.1432816301410699,
      "grad_norm": 2117.5634089858195,
      "learning_rate": 1.9118829859389067e-05,
      "loss": 451.6985,
      "step": 7450
    },
    {
      "epoch": 0.14347395447682973,
      "grad_norm": 598.4948095429457,
      "learning_rate": 1.9116323124740748e-05,
      "loss": 436.7958,
      "step": 7460
    },
    {
      "epoch": 0.14366627881258956,
      "grad_norm": 570.4295972789612,
      "learning_rate": 1.9113812994412627e-05,
      "loss": 440.0646,
      "step": 7470
    },
    {
      "epoch": 0.1438586031483494,
      "grad_norm": 576.0305848119193,
      "learning_rate": 1.911129946933968e-05,
      "loss": 440.1847,
      "step": 7480
    },
    {
      "epoch": 0.1440509274841092,
      "grad_norm": 542.6346419853453,
      "learning_rate": 1.9108782550458164e-05,
      "loss": 453.2081,
      "step": 7490
    },
    {
      "epoch": 0.14424325181986902,
      "grad_norm": 575.3669732122477,
      "learning_rate": 1.9106262238705583e-05,
      "loss": 449.5358,
      "step": 7500
    },
    {
      "epoch": 0.14443557615562885,
      "grad_norm": 537.6084302704626,
      "learning_rate": 1.9103738535020713e-05,
      "loss": 446.0702,
      "step": 7510
    },
    {
      "epoch": 0.14462790049138868,
      "grad_norm": 543.6680320093975,
      "learning_rate": 1.910121144034359e-05,
      "loss": 428.9582,
      "step": 7520
    },
    {
      "epoch": 0.1448202248271485,
      "grad_norm": 542.7679299190743,
      "learning_rate": 1.909868095561552e-05,
      "loss": 436.3199,
      "step": 7530
    },
    {
      "epoch": 0.14501254916290832,
      "grad_norm": 524.253006819328,
      "learning_rate": 1.9096147081779063e-05,
      "loss": 445.5708,
      "step": 7540
    },
    {
      "epoch": 0.14520487349866815,
      "grad_norm": 554.829815846205,
      "learning_rate": 1.9093609819778044e-05,
      "loss": 437.1036,
      "step": 7550
    },
    {
      "epoch": 0.14539719783442798,
      "grad_norm": 501.9055615810935,
      "learning_rate": 1.9091069170557554e-05,
      "loss": 443.9232,
      "step": 7560
    },
    {
      "epoch": 0.1455895221701878,
      "grad_norm": 559.9313886576359,
      "learning_rate": 1.9088525135063944e-05,
      "loss": 437.4566,
      "step": 7570
    },
    {
      "epoch": 0.14578184650594764,
      "grad_norm": 517.4741249400627,
      "learning_rate": 1.9085977714244822e-05,
      "loss": 435.0832,
      "step": 7580
    },
    {
      "epoch": 0.14597417084170747,
      "grad_norm": 550.9501878286719,
      "learning_rate": 1.908342690904906e-05,
      "loss": 443.4718,
      "step": 7590
    },
    {
      "epoch": 0.14616649517746727,
      "grad_norm": 603.8445489928295,
      "learning_rate": 1.9080872720426793e-05,
      "loss": 440.3922,
      "step": 7600
    },
    {
      "epoch": 0.1463588195132271,
      "grad_norm": 561.9610965814381,
      "learning_rate": 1.9078315149329413e-05,
      "loss": 420.2626,
      "step": 7610
    },
    {
      "epoch": 0.14655114384898693,
      "grad_norm": 511.9348084977807,
      "learning_rate": 1.9075754196709574e-05,
      "loss": 446.088,
      "step": 7620
    },
    {
      "epoch": 0.14674346818474676,
      "grad_norm": 514.9396544232271,
      "learning_rate": 1.9073189863521184e-05,
      "loss": 433.6025,
      "step": 7630
    },
    {
      "epoch": 0.1469357925205066,
      "grad_norm": 629.7431275383268,
      "learning_rate": 1.9070622150719423e-05,
      "loss": 439.0813,
      "step": 7640
    },
    {
      "epoch": 0.1471281168562664,
      "grad_norm": 608.7647068301741,
      "learning_rate": 1.9068051059260716e-05,
      "loss": 437.1541,
      "step": 7650
    },
    {
      "epoch": 0.14732044119202622,
      "grad_norm": 591.4084677037018,
      "learning_rate": 1.9065476590102752e-05,
      "loss": 443.5418,
      "step": 7660
    },
    {
      "epoch": 0.14751276552778605,
      "grad_norm": 670.2283456290832,
      "learning_rate": 1.906289874420448e-05,
      "loss": 451.7118,
      "step": 7670
    },
    {
      "epoch": 0.14770508986354589,
      "grad_norm": 530.2510227646507,
      "learning_rate": 1.9060317522526105e-05,
      "loss": 430.0094,
      "step": 7680
    },
    {
      "epoch": 0.14789741419930572,
      "grad_norm": 582.8801334902874,
      "learning_rate": 1.905773292602909e-05,
      "loss": 442.6443,
      "step": 7690
    },
    {
      "epoch": 0.14808973853506555,
      "grad_norm": 548.2699834322779,
      "learning_rate": 1.905514495567615e-05,
      "loss": 433.7478,
      "step": 7700
    },
    {
      "epoch": 0.14828206287082535,
      "grad_norm": 546.8084915195556,
      "learning_rate": 1.9052553612431268e-05,
      "loss": 453.4112,
      "step": 7710
    },
    {
      "epoch": 0.14847438720658518,
      "grad_norm": 593.3055536179546,
      "learning_rate": 1.9049958897259674e-05,
      "loss": 434.2642,
      "step": 7720
    },
    {
      "epoch": 0.148666711542345,
      "grad_norm": 508.2996949859877,
      "learning_rate": 1.904736081112785e-05,
      "loss": 439.8122,
      "step": 7730
    },
    {
      "epoch": 0.14885903587810484,
      "grad_norm": 589.5306799327531,
      "learning_rate": 1.9044759355003552e-05,
      "loss": 441.0922,
      "step": 7740
    },
    {
      "epoch": 0.14905136021386467,
      "grad_norm": 585.7134046183394,
      "learning_rate": 1.904215452985577e-05,
      "loss": 433.4066,
      "step": 7750
    },
    {
      "epoch": 0.1492436845496245,
      "grad_norm": 691.8373481101901,
      "learning_rate": 1.9039546336654765e-05,
      "loss": 440.9248,
      "step": 7760
    },
    {
      "epoch": 0.1494360088853843,
      "grad_norm": 627.92254723191,
      "learning_rate": 1.903693477637204e-05,
      "loss": 431.1621,
      "step": 7770
    },
    {
      "epoch": 0.14962833322114413,
      "grad_norm": 611.342956564641,
      "learning_rate": 1.903431984998036e-05,
      "loss": 436.4596,
      "step": 7780
    },
    {
      "epoch": 0.14982065755690396,
      "grad_norm": 580.9808788783724,
      "learning_rate": 1.9031701558453747e-05,
      "loss": 453.6431,
      "step": 7790
    },
    {
      "epoch": 0.1500129818926638,
      "grad_norm": 508.2592836367386,
      "learning_rate": 1.902907990276746e-05,
      "loss": 418.0719,
      "step": 7800
    },
    {
      "epoch": 0.15020530622842362,
      "grad_norm": 566.2762269851959,
      "learning_rate": 1.9026454883898036e-05,
      "loss": 441.5839,
      "step": 7810
    },
    {
      "epoch": 0.15039763056418343,
      "grad_norm": 553.876784327105,
      "learning_rate": 1.902382650282324e-05,
      "loss": 428.4753,
      "step": 7820
    },
    {
      "epoch": 0.15058995489994326,
      "grad_norm": 509.25067367581033,
      "learning_rate": 1.902119476052211e-05,
      "loss": 424.0172,
      "step": 7830
    },
    {
      "epoch": 0.1507822792357031,
      "grad_norm": 573.8549114711305,
      "learning_rate": 1.9018559657974918e-05,
      "loss": 436.3267,
      "step": 7840
    },
    {
      "epoch": 0.15097460357146292,
      "grad_norm": 578.8517407293792,
      "learning_rate": 1.90159211961632e-05,
      "loss": 476.6084,
      "step": 7850
    },
    {
      "epoch": 0.15116692790722275,
      "grad_norm": 604.5224267412457,
      "learning_rate": 1.901327937606974e-05,
      "loss": 436.9109,
      "step": 7860
    },
    {
      "epoch": 0.15135925224298258,
      "grad_norm": 556.5233939076458,
      "learning_rate": 1.901063419867857e-05,
      "loss": 444.9464,
      "step": 7870
    },
    {
      "epoch": 0.15155157657874238,
      "grad_norm": 566.4432173386184,
      "learning_rate": 1.900798566497498e-05,
      "loss": 437.7201,
      "step": 7880
    },
    {
      "epoch": 0.1517439009145022,
      "grad_norm": 523.0909791332912,
      "learning_rate": 1.9005333775945496e-05,
      "loss": 439.1833,
      "step": 7890
    },
    {
      "epoch": 0.15193622525026204,
      "grad_norm": 590.1169918399377,
      "learning_rate": 1.9002678532577915e-05,
      "loss": 430.2732,
      "step": 7900
    },
    {
      "epoch": 0.15212854958602187,
      "grad_norm": 538.7629382249178,
      "learning_rate": 1.900001993586126e-05,
      "loss": 441.559,
      "step": 7910
    },
    {
      "epoch": 0.1523208739217817,
      "grad_norm": 556.1148531125073,
      "learning_rate": 1.8997357986785822e-05,
      "loss": 446.4016,
      "step": 7920
    },
    {
      "epoch": 0.1525131982575415,
      "grad_norm": 643.330176953753,
      "learning_rate": 1.899469268634313e-05,
      "loss": 442.2081,
      "step": 7930
    },
    {
      "epoch": 0.15270552259330133,
      "grad_norm": 502.7658311238469,
      "learning_rate": 1.8992024035525964e-05,
      "loss": 426.8118,
      "step": 7940
    },
    {
      "epoch": 0.15289784692906117,
      "grad_norm": 563.7536501867388,
      "learning_rate": 1.8989352035328352e-05,
      "loss": 438.9939,
      "step": 7950
    },
    {
      "epoch": 0.153090171264821,
      "grad_norm": 595.729334769332,
      "learning_rate": 1.8986676686745572e-05,
      "loss": 431.4548,
      "step": 7960
    },
    {
      "epoch": 0.15328249560058083,
      "grad_norm": 541.6423282934109,
      "learning_rate": 1.8983997990774145e-05,
      "loss": 434.93,
      "step": 7970
    },
    {
      "epoch": 0.15347481993634066,
      "grad_norm": 584.8342988997091,
      "learning_rate": 1.8981315948411842e-05,
      "loss": 443.9966,
      "step": 7980
    },
    {
      "epoch": 0.15366714427210046,
      "grad_norm": 620.1016078368535,
      "learning_rate": 1.897863056065768e-05,
      "loss": 444.0354,
      "step": 7990
    },
    {
      "epoch": 0.1538594686078603,
      "grad_norm": 543.2149365694071,
      "learning_rate": 1.8975941828511923e-05,
      "loss": 434.4141,
      "step": 8000
    },
    {
      "epoch": 0.15405179294362012,
      "grad_norm": 549.2536788675371,
      "learning_rate": 1.8973249752976075e-05,
      "loss": 433.059,
      "step": 8010
    },
    {
      "epoch": 0.15424411727937995,
      "grad_norm": 566.1975761435316,
      "learning_rate": 1.8970554335052897e-05,
      "loss": 428.917,
      "step": 8020
    },
    {
      "epoch": 0.15443644161513978,
      "grad_norm": 601.2707463310079,
      "learning_rate": 1.8967855575746375e-05,
      "loss": 434.3125,
      "step": 8030
    },
    {
      "epoch": 0.1546287659508996,
      "grad_norm": 572.9370704510053,
      "learning_rate": 1.8965153476061763e-05,
      "loss": 441.1062,
      "step": 8040
    },
    {
      "epoch": 0.1548210902866594,
      "grad_norm": 544.8283044070936,
      "learning_rate": 1.896244803700555e-05,
      "loss": 434.5202,
      "step": 8050
    },
    {
      "epoch": 0.15501341462241924,
      "grad_norm": 551.435718815261,
      "learning_rate": 1.8959739259585458e-05,
      "loss": 435.1253,
      "step": 8060
    },
    {
      "epoch": 0.15520573895817907,
      "grad_norm": 580.6890855690795,
      "learning_rate": 1.895702714481047e-05,
      "loss": 421.8108,
      "step": 8070
    },
    {
      "epoch": 0.1553980632939389,
      "grad_norm": 519.697868206756,
      "learning_rate": 1.8954311693690798e-05,
      "loss": 417.7399,
      "step": 8080
    },
    {
      "epoch": 0.15559038762969873,
      "grad_norm": 533.1929671672967,
      "learning_rate": 1.8951592907237906e-05,
      "loss": 432.3487,
      "step": 8090
    },
    {
      "epoch": 0.15578271196545854,
      "grad_norm": 690.4581820900596,
      "learning_rate": 1.8948870786464496e-05,
      "loss": 444.6823,
      "step": 8100
    },
    {
      "epoch": 0.15597503630121837,
      "grad_norm": 592.2038275260893,
      "learning_rate": 1.8946145332384515e-05,
      "loss": 423.3146,
      "step": 8110
    },
    {
      "epoch": 0.1561673606369782,
      "grad_norm": 629.1982185777041,
      "learning_rate": 1.8943416546013148e-05,
      "loss": 441.5614,
      "step": 8120
    },
    {
      "epoch": 0.15635968497273803,
      "grad_norm": 500.79431278963153,
      "learning_rate": 1.894068442836682e-05,
      "loss": 432.5976,
      "step": 8130
    },
    {
      "epoch": 0.15655200930849786,
      "grad_norm": 575.801820386397,
      "learning_rate": 1.8937948980463207e-05,
      "loss": 428.3051,
      "step": 8140
    },
    {
      "epoch": 0.1567443336442577,
      "grad_norm": 585.8597750759566,
      "learning_rate": 1.893521020332121e-05,
      "loss": 445.5108,
      "step": 8150
    },
    {
      "epoch": 0.1569366579800175,
      "grad_norm": 581.9024039306552,
      "learning_rate": 1.8932468097960988e-05,
      "loss": 429.6226,
      "step": 8160
    },
    {
      "epoch": 0.15712898231577732,
      "grad_norm": 548.2506755004573,
      "learning_rate": 1.892972266540392e-05,
      "loss": 420.4457,
      "step": 8170
    },
    {
      "epoch": 0.15732130665153715,
      "grad_norm": 518.0948250716367,
      "learning_rate": 1.8926973906672635e-05,
      "loss": 422.4104,
      "step": 8180
    },
    {
      "epoch": 0.15751363098729698,
      "grad_norm": 539.6317603547405,
      "learning_rate": 1.892422182279101e-05,
      "loss": 438.4853,
      "step": 8190
    },
    {
      "epoch": 0.1577059553230568,
      "grad_norm": 560.3636254848747,
      "learning_rate": 1.892146641478414e-05,
      "loss": 431.9881,
      "step": 8200
    },
    {
      "epoch": 0.15789827965881661,
      "grad_norm": 581.6132816498555,
      "learning_rate": 1.8918707683678376e-05,
      "loss": 421.6939,
      "step": 8210
    },
    {
      "epoch": 0.15809060399457645,
      "grad_norm": 608.7072010580962,
      "learning_rate": 1.8915945630501296e-05,
      "loss": 431.0424,
      "step": 8220
    },
    {
      "epoch": 0.15828292833033628,
      "grad_norm": 562.8788592397115,
      "learning_rate": 1.8913180256281723e-05,
      "loss": 433.2843,
      "step": 8230
    },
    {
      "epoch": 0.1584752526660961,
      "grad_norm": 519.4687530271713,
      "learning_rate": 1.8910411562049706e-05,
      "loss": 423.8484,
      "step": 8240
    },
    {
      "epoch": 0.15866757700185594,
      "grad_norm": 540.379523632361,
      "learning_rate": 1.8907639548836548e-05,
      "loss": 429.4329,
      "step": 8250
    },
    {
      "epoch": 0.15885990133761577,
      "grad_norm": 584.1014209106212,
      "learning_rate": 1.8904864217674766e-05,
      "loss": 447.2186,
      "step": 8260
    },
    {
      "epoch": 0.15905222567337557,
      "grad_norm": 611.6965536357849,
      "learning_rate": 1.8902085569598136e-05,
      "loss": 421.9906,
      "step": 8270
    },
    {
      "epoch": 0.1592445500091354,
      "grad_norm": 549.0467627488447,
      "learning_rate": 1.889930360564165e-05,
      "loss": 425.6736,
      "step": 8280
    },
    {
      "epoch": 0.15943687434489523,
      "grad_norm": 553.4719219076825,
      "learning_rate": 1.8896518326841554e-05,
      "loss": 419.4191,
      "step": 8290
    },
    {
      "epoch": 0.15962919868065506,
      "grad_norm": 539.6735805099422,
      "learning_rate": 1.889372973423531e-05,
      "loss": 430.6573,
      "step": 8300
    },
    {
      "epoch": 0.1598215230164149,
      "grad_norm": 501.766952099889,
      "learning_rate": 1.889093782886162e-05,
      "loss": 428.0232,
      "step": 8310
    },
    {
      "epoch": 0.16001384735217472,
      "grad_norm": 551.0356240631205,
      "learning_rate": 1.8888142611760433e-05,
      "loss": 426.1552,
      "step": 8320
    },
    {
      "epoch": 0.16020617168793452,
      "grad_norm": 562.525599816841,
      "learning_rate": 1.8885344083972912e-05,
      "loss": 423.6793,
      "step": 8330
    },
    {
      "epoch": 0.16039849602369435,
      "grad_norm": 583.1779236079669,
      "learning_rate": 1.8882542246541468e-05,
      "loss": 445.9307,
      "step": 8340
    },
    {
      "epoch": 0.16059082035945418,
      "grad_norm": 615.1845129614437,
      "learning_rate": 1.887973710050974e-05,
      "loss": 443.9851,
      "step": 8350
    },
    {
      "epoch": 0.16078314469521401,
      "grad_norm": 584.8857841386294,
      "learning_rate": 1.887692864692259e-05,
      "loss": 423.9522,
      "step": 8360
    },
    {
      "epoch": 0.16097546903097384,
      "grad_norm": 553.4133223514797,
      "learning_rate": 1.887411688682613e-05,
      "loss": 438.2559,
      "step": 8370
    },
    {
      "epoch": 0.16116779336673365,
      "grad_norm": 570.8960463191577,
      "learning_rate": 1.887130182126769e-05,
      "loss": 441.2844,
      "step": 8380
    },
    {
      "epoch": 0.16136011770249348,
      "grad_norm": 667.5638120400549,
      "learning_rate": 1.8868483451295835e-05,
      "loss": 433.5899,
      "step": 8390
    },
    {
      "epoch": 0.1615524420382533,
      "grad_norm": 569.025894881582,
      "learning_rate": 1.8865661777960366e-05,
      "loss": 427.1694,
      "step": 8400
    },
    {
      "epoch": 0.16174476637401314,
      "grad_norm": 513.2618461942593,
      "learning_rate": 1.88628368023123e-05,
      "loss": 432.4041,
      "step": 8410
    },
    {
      "epoch": 0.16193709070977297,
      "grad_norm": 620.9651312400831,
      "learning_rate": 1.8860008525403903e-05,
      "loss": 449.5799,
      "step": 8420
    },
    {
      "epoch": 0.1621294150455328,
      "grad_norm": 523.06292503676,
      "learning_rate": 1.885717694828866e-05,
      "loss": 431.8335,
      "step": 8430
    },
    {
      "epoch": 0.1623217393812926,
      "grad_norm": 653.042768179501,
      "learning_rate": 1.8854342072021282e-05,
      "loss": 418.552,
      "step": 8440
    },
    {
      "epoch": 0.16251406371705243,
      "grad_norm": 511.6522064427705,
      "learning_rate": 1.8851503897657717e-05,
      "loss": 434.6689,
      "step": 8450
    },
    {
      "epoch": 0.16270638805281226,
      "grad_norm": 521.8439186722719,
      "learning_rate": 1.8848662426255135e-05,
      "loss": 440.1868,
      "step": 8460
    },
    {
      "epoch": 0.1628987123885721,
      "grad_norm": 532.5330484085582,
      "learning_rate": 1.8845817658871942e-05,
      "loss": 434.4043,
      "step": 8470
    },
    {
      "epoch": 0.16309103672433192,
      "grad_norm": 546.976816839853,
      "learning_rate": 1.8842969596567765e-05,
      "loss": 441.8241,
      "step": 8480
    },
    {
      "epoch": 0.16328336106009173,
      "grad_norm": 576.69187127638,
      "learning_rate": 1.884011824040346e-05,
      "loss": 449.426,
      "step": 8490
    },
    {
      "epoch": 0.16347568539585156,
      "grad_norm": 621.7037817843438,
      "learning_rate": 1.883726359144111e-05,
      "loss": 434.0773,
      "step": 8500
    },
    {
      "epoch": 0.16366800973161139,
      "grad_norm": 504.5021465667742,
      "learning_rate": 1.8834405650744023e-05,
      "loss": 426.3316,
      "step": 8510
    },
    {
      "epoch": 0.16386033406737122,
      "grad_norm": 628.6596615396477,
      "learning_rate": 1.883154441937674e-05,
      "loss": 437.8812,
      "step": 8520
    },
    {
      "epoch": 0.16405265840313105,
      "grad_norm": 542.6726097126638,
      "learning_rate": 1.8828679898405015e-05,
      "loss": 416.4292,
      "step": 8530
    },
    {
      "epoch": 0.16424498273889088,
      "grad_norm": 554.6754415804415,
      "learning_rate": 1.8825812088895835e-05,
      "loss": 434.5456,
      "step": 8540
    },
    {
      "epoch": 0.16443730707465068,
      "grad_norm": 529.5274527416682,
      "learning_rate": 1.882294099191742e-05,
      "loss": 435.6156,
      "step": 8550
    },
    {
      "epoch": 0.1646296314104105,
      "grad_norm": 545.3900996071341,
      "learning_rate": 1.88200666085392e-05,
      "loss": 426.0449,
      "step": 8560
    },
    {
      "epoch": 0.16482195574617034,
      "grad_norm": 561.5017344100828,
      "learning_rate": 1.8817188939831838e-05,
      "loss": 434.16,
      "step": 8570
    },
    {
      "epoch": 0.16501428008193017,
      "grad_norm": 524.1371948724865,
      "learning_rate": 1.8814307986867214e-05,
      "loss": 415.4602,
      "step": 8580
    },
    {
      "epoch": 0.16520660441769,
      "grad_norm": 538.5403613520467,
      "learning_rate": 1.881142375071844e-05,
      "loss": 408.1654,
      "step": 8590
    },
    {
      "epoch": 0.16539892875344983,
      "grad_norm": 660.09730136131,
      "learning_rate": 1.8808536232459844e-05,
      "loss": 438.1817,
      "step": 8600
    },
    {
      "epoch": 0.16559125308920963,
      "grad_norm": 509.92483346796786,
      "learning_rate": 1.8805645433166976e-05,
      "loss": 441.5352,
      "step": 8610
    },
    {
      "epoch": 0.16578357742496946,
      "grad_norm": 560.7575613908119,
      "learning_rate": 1.8802751353916618e-05,
      "loss": 437.8606,
      "step": 8620
    },
    {
      "epoch": 0.1659759017607293,
      "grad_norm": 525.5924896458009,
      "learning_rate": 1.8799853995786763e-05,
      "loss": 430.4469,
      "step": 8630
    },
    {
      "epoch": 0.16616822609648912,
      "grad_norm": 563.1080871985357,
      "learning_rate": 1.8796953359856626e-05,
      "loss": 409.7498,
      "step": 8640
    },
    {
      "epoch": 0.16636055043224895,
      "grad_norm": 582.690469897271,
      "learning_rate": 1.879404944720665e-05,
      "loss": 427.2426,
      "step": 8650
    },
    {
      "epoch": 0.16655287476800876,
      "grad_norm": 560.4089513612888,
      "learning_rate": 1.8791142258918496e-05,
      "loss": 429.3333,
      "step": 8660
    },
    {
      "epoch": 0.1667451991037686,
      "grad_norm": 573.2907731727554,
      "learning_rate": 1.8788231796075037e-05,
      "loss": 438.3569,
      "step": 8670
    },
    {
      "epoch": 0.16693752343952842,
      "grad_norm": 616.8718364832039,
      "learning_rate": 1.8785318059760384e-05,
      "loss": 418.9777,
      "step": 8680
    },
    {
      "epoch": 0.16712984777528825,
      "grad_norm": 536.2940676573769,
      "learning_rate": 1.8782401051059838e-05,
      "loss": 424.638,
      "step": 8690
    },
    {
      "epoch": 0.16732217211104808,
      "grad_norm": 517.2732316134967,
      "learning_rate": 1.8779480771059954e-05,
      "loss": 413.4565,
      "step": 8700
    },
    {
      "epoch": 0.1675144964468079,
      "grad_norm": 548.6441593095126,
      "learning_rate": 1.8776557220848477e-05,
      "loss": 424.9253,
      "step": 8710
    },
    {
      "epoch": 0.1677068207825677,
      "grad_norm": 704.5101201740893,
      "learning_rate": 1.8773630401514388e-05,
      "loss": 425.7483,
      "step": 8720
    },
    {
      "epoch": 0.16789914511832754,
      "grad_norm": 570.644947874167,
      "learning_rate": 1.877070031414787e-05,
      "loss": 430.1442,
      "step": 8730
    },
    {
      "epoch": 0.16809146945408737,
      "grad_norm": 753.9641535372008,
      "learning_rate": 1.876776695984034e-05,
      "loss": 423.4344,
      "step": 8740
    },
    {
      "epoch": 0.1682837937898472,
      "grad_norm": 601.3750020968382,
      "learning_rate": 1.8764830339684426e-05,
      "loss": 419.546,
      "step": 8750
    },
    {
      "epoch": 0.16847611812560703,
      "grad_norm": 555.995162734277,
      "learning_rate": 1.8761890454773965e-05,
      "loss": 424.8769,
      "step": 8760
    },
    {
      "epoch": 0.16866844246136684,
      "grad_norm": 555.6905106933057,
      "learning_rate": 1.8758947306204012e-05,
      "loss": 434.1687,
      "step": 8770
    },
    {
      "epoch": 0.16886076679712667,
      "grad_norm": 563.0678086087067,
      "learning_rate": 1.8756000895070854e-05,
      "loss": 410.9748,
      "step": 8780
    },
    {
      "epoch": 0.1690530911328865,
      "grad_norm": 525.1484063246721,
      "learning_rate": 1.8753051222471968e-05,
      "loss": 433.5676,
      "step": 8790
    },
    {
      "epoch": 0.16924541546864633,
      "grad_norm": 582.5866661832113,
      "learning_rate": 1.8750098289506066e-05,
      "loss": 421.106,
      "step": 8800
    },
    {
      "epoch": 0.16943773980440616,
      "grad_norm": 582.2388981661788,
      "learning_rate": 1.8747142097273057e-05,
      "loss": 417.332,
      "step": 8810
    },
    {
      "epoch": 0.169630064140166,
      "grad_norm": 515.6525576635153,
      "learning_rate": 1.8744182646874085e-05,
      "loss": 425.4491,
      "step": 8820
    },
    {
      "epoch": 0.1698223884759258,
      "grad_norm": 518.7076773187805,
      "learning_rate": 1.8741219939411494e-05,
      "loss": 439.8529,
      "step": 8830
    },
    {
      "epoch": 0.17001471281168562,
      "grad_norm": 567.5260696340699,
      "learning_rate": 1.873825397598884e-05,
      "loss": 434.1176,
      "step": 8840
    },
    {
      "epoch": 0.17020703714744545,
      "grad_norm": 509.8572598415664,
      "learning_rate": 1.8735284757710897e-05,
      "loss": 428.2947,
      "step": 8850
    },
    {
      "epoch": 0.17039936148320528,
      "grad_norm": 578.9582231928556,
      "learning_rate": 1.873231228568365e-05,
      "loss": 433.9714,
      "step": 8860
    },
    {
      "epoch": 0.1705916858189651,
      "grad_norm": 542.4515139172046,
      "learning_rate": 1.8729336561014294e-05,
      "loss": 434.2283,
      "step": 8870
    },
    {
      "epoch": 0.17078401015472494,
      "grad_norm": 543.8338730974002,
      "learning_rate": 1.8726357584811242e-05,
      "loss": 420.8061,
      "step": 8880
    },
    {
      "epoch": 0.17097633449048474,
      "grad_norm": 532.7654305582791,
      "learning_rate": 1.8723375358184107e-05,
      "loss": 433.0151,
      "step": 8890
    },
    {
      "epoch": 0.17116865882624457,
      "grad_norm": 512.5359719309298,
      "learning_rate": 1.8720389882243722e-05,
      "loss": 420.664,
      "step": 8900
    },
    {
      "epoch": 0.1713609831620044,
      "grad_norm": 534.9811641267368,
      "learning_rate": 1.871740115810213e-05,
      "loss": 438.8542,
      "step": 8910
    },
    {
      "epoch": 0.17155330749776423,
      "grad_norm": 527.4542376506438,
      "learning_rate": 1.8714409186872578e-05,
      "loss": 433.1957,
      "step": 8920
    },
    {
      "epoch": 0.17174563183352407,
      "grad_norm": 534.2467297715082,
      "learning_rate": 1.871141396966953e-05,
      "loss": 434.9628,
      "step": 8930
    },
    {
      "epoch": 0.17193795616928387,
      "grad_norm": 548.1164226003325,
      "learning_rate": 1.8708415507608647e-05,
      "loss": 420.644,
      "step": 8940
    },
    {
      "epoch": 0.1721302805050437,
      "grad_norm": 492.71640548491524,
      "learning_rate": 1.8705413801806817e-05,
      "loss": 427.233,
      "step": 8950
    },
    {
      "epoch": 0.17232260484080353,
      "grad_norm": 547.2007064477028,
      "learning_rate": 1.8702408853382114e-05,
      "loss": 420.3109,
      "step": 8960
    },
    {
      "epoch": 0.17251492917656336,
      "grad_norm": 523.5479372406793,
      "learning_rate": 1.8699400663453842e-05,
      "loss": 422.3723,
      "step": 8970
    },
    {
      "epoch": 0.1727072535123232,
      "grad_norm": 519.4993779419578,
      "learning_rate": 1.8696389233142498e-05,
      "loss": 419.6555,
      "step": 8980
    },
    {
      "epoch": 0.17289957784808302,
      "grad_norm": 502.62735831219584,
      "learning_rate": 1.8693374563569792e-05,
      "loss": 420.121,
      "step": 8990
    },
    {
      "epoch": 0.17309190218384282,
      "grad_norm": 566.7886844877982,
      "learning_rate": 1.8690356655858634e-05,
      "loss": 424.814,
      "step": 9000
    },
    {
      "epoch": 0.17328422651960265,
      "grad_norm": 591.3701541543867,
      "learning_rate": 1.868733551113315e-05,
      "loss": 433.5979,
      "step": 9010
    },
    {
      "epoch": 0.17347655085536248,
      "grad_norm": 553.2749494894153,
      "learning_rate": 1.8684311130518663e-05,
      "loss": 420.8928,
      "step": 9020
    },
    {
      "epoch": 0.1736688751911223,
      "grad_norm": 613.200549157228,
      "learning_rate": 1.8681283515141705e-05,
      "loss": 421.6006,
      "step": 9030
    },
    {
      "epoch": 0.17386119952688214,
      "grad_norm": 516.7305214545012,
      "learning_rate": 1.8678252666130016e-05,
      "loss": 424.7881,
      "step": 9040
    },
    {
      "epoch": 0.17405352386264195,
      "grad_norm": 566.6093792031394,
      "learning_rate": 1.8675218584612534e-05,
      "loss": 421.7676,
      "step": 9050
    },
    {
      "epoch": 0.17424584819840178,
      "grad_norm": 511.2119317871291,
      "learning_rate": 1.8672181271719406e-05,
      "loss": 427.5226,
      "step": 9060
    },
    {
      "epoch": 0.1744381725341616,
      "grad_norm": 596.8432639819043,
      "learning_rate": 1.866914072858198e-05,
      "loss": 435.9604,
      "step": 9070
    },
    {
      "epoch": 0.17463049686992144,
      "grad_norm": 558.3075578096414,
      "learning_rate": 1.8666096956332805e-05,
      "loss": 426.1812,
      "step": 9080
    },
    {
      "epoch": 0.17482282120568127,
      "grad_norm": 511.4292529690994,
      "learning_rate": 1.8663049956105642e-05,
      "loss": 422.1414,
      "step": 9090
    },
    {
      "epoch": 0.1750151455414411,
      "grad_norm": 522.7046412746399,
      "learning_rate": 1.8659999729035445e-05,
      "loss": 420.2742,
      "step": 9100
    },
    {
      "epoch": 0.1752074698772009,
      "grad_norm": 493.81231795962515,
      "learning_rate": 1.8656946276258373e-05,
      "loss": 420.0853,
      "step": 9110
    },
    {
      "epoch": 0.17539979421296073,
      "grad_norm": 571.5146890274276,
      "learning_rate": 1.8653889598911787e-05,
      "loss": 431.4177,
      "step": 9120
    },
    {
      "epoch": 0.17559211854872056,
      "grad_norm": 521.4456421773015,
      "learning_rate": 1.865082969813425e-05,
      "loss": 418.0671,
      "step": 9130
    },
    {
      "epoch": 0.1757844428844804,
      "grad_norm": 564.0059254649086,
      "learning_rate": 1.8647766575065523e-05,
      "loss": 425.494,
      "step": 9140
    },
    {
      "epoch": 0.17597676722024022,
      "grad_norm": 577.5935009146527,
      "learning_rate": 1.864470023084657e-05,
      "loss": 444.1621,
      "step": 9150
    },
    {
      "epoch": 0.17616909155600005,
      "grad_norm": 518.2614350062311,
      "learning_rate": 1.864163066661955e-05,
      "loss": 427.1668,
      "step": 9160
    },
    {
      "epoch": 0.17636141589175985,
      "grad_norm": 602.2692324132776,
      "learning_rate": 1.8638557883527833e-05,
      "loss": 419.9159,
      "step": 9170
    },
    {
      "epoch": 0.17655374022751968,
      "grad_norm": 591.646699534163,
      "learning_rate": 1.8635481882715975e-05,
      "loss": 413.4255,
      "step": 9180
    },
    {
      "epoch": 0.17674606456327951,
      "grad_norm": 489.06767888855086,
      "learning_rate": 1.863240266532973e-05,
      "loss": 414.8031,
      "step": 9190
    },
    {
      "epoch": 0.17693838889903934,
      "grad_norm": 575.6543170412791,
      "learning_rate": 1.8629320232516063e-05,
      "loss": 427.3708,
      "step": 9200
    },
    {
      "epoch": 0.17713071323479918,
      "grad_norm": 549.4387331358224,
      "learning_rate": 1.862623458542313e-05,
      "loss": 422.5452,
      "step": 9210
    },
    {
      "epoch": 0.17732303757055898,
      "grad_norm": 524.0464185455141,
      "learning_rate": 1.862314572520028e-05,
      "loss": 419.5607,
      "step": 9220
    },
    {
      "epoch": 0.1775153619063188,
      "grad_norm": 595.4925303858677,
      "learning_rate": 1.862005365299806e-05,
      "loss": 414.3376,
      "step": 9230
    },
    {
      "epoch": 0.17770768624207864,
      "grad_norm": 532.4377562848898,
      "learning_rate": 1.8616958369968223e-05,
      "loss": 441.0964,
      "step": 9240
    },
    {
      "epoch": 0.17790001057783847,
      "grad_norm": 503.47459913660003,
      "learning_rate": 1.8613859877263708e-05,
      "loss": 422.7668,
      "step": 9250
    },
    {
      "epoch": 0.1780923349135983,
      "grad_norm": 505.177979755498,
      "learning_rate": 1.8610758176038647e-05,
      "loss": 410.8366,
      "step": 9260
    },
    {
      "epoch": 0.17828465924935813,
      "grad_norm": 580.1178516448416,
      "learning_rate": 1.860765326744838e-05,
      "loss": 416.7083,
      "step": 9270
    },
    {
      "epoch": 0.17847698358511793,
      "grad_norm": 531.6274491521162,
      "learning_rate": 1.8604545152649426e-05,
      "loss": 425.9408,
      "step": 9280
    },
    {
      "epoch": 0.17866930792087776,
      "grad_norm": 511.7090105790692,
      "learning_rate": 1.860143383279952e-05,
      "loss": 430.9702,
      "step": 9290
    },
    {
      "epoch": 0.1788616322566376,
      "grad_norm": 578.6974413585342,
      "learning_rate": 1.859831930905756e-05,
      "loss": 415.5254,
      "step": 9300
    },
    {
      "epoch": 0.17905395659239742,
      "grad_norm": 552.8420613889978,
      "learning_rate": 1.8595201582583668e-05,
      "loss": 423.8861,
      "step": 9310
    },
    {
      "epoch": 0.17924628092815725,
      "grad_norm": 519.5427538259271,
      "learning_rate": 1.859208065453914e-05,
      "loss": 416.2268,
      "step": 9320
    },
    {
      "epoch": 0.17943860526391706,
      "grad_norm": 553.5123808009505,
      "learning_rate": 1.8588956526086472e-05,
      "loss": 433.5628,
      "step": 9330
    },
    {
      "epoch": 0.1796309295996769,
      "grad_norm": 549.8761157194938,
      "learning_rate": 1.8585829198389347e-05,
      "loss": 416.0442,
      "step": 9340
    },
    {
      "epoch": 0.17982325393543672,
      "grad_norm": 562.8631278908741,
      "learning_rate": 1.8582698672612646e-05,
      "loss": 417.3363,
      "step": 9350
    },
    {
      "epoch": 0.18001557827119655,
      "grad_norm": 586.7017995530338,
      "learning_rate": 1.8579564949922438e-05,
      "loss": 417.5483,
      "step": 9360
    },
    {
      "epoch": 0.18020790260695638,
      "grad_norm": 528.6126212847728,
      "learning_rate": 1.8576428031485984e-05,
      "loss": 409.244,
      "step": 9370
    },
    {
      "epoch": 0.1804002269427162,
      "grad_norm": 530.6686096377931,
      "learning_rate": 1.8573287918471728e-05,
      "loss": 421.8057,
      "step": 9380
    },
    {
      "epoch": 0.180592551278476,
      "grad_norm": 585.0771653810731,
      "learning_rate": 1.8570144612049322e-05,
      "loss": 417.3723,
      "step": 9390
    },
    {
      "epoch": 0.18078487561423584,
      "grad_norm": 559.0368429444012,
      "learning_rate": 1.856699811338958e-05,
      "loss": 419.4453,
      "step": 9400
    },
    {
      "epoch": 0.18097719994999567,
      "grad_norm": 654.8321479458963,
      "learning_rate": 1.8563848423664536e-05,
      "loss": 430.0406,
      "step": 9410
    },
    {
      "epoch": 0.1811695242857555,
      "grad_norm": 495.766456885269,
      "learning_rate": 1.8560695544047388e-05,
      "loss": 436.637,
      "step": 9420
    },
    {
      "epoch": 0.18136184862151533,
      "grad_norm": 578.2104264562734,
      "learning_rate": 1.8557539475712538e-05,
      "loss": 446.099,
      "step": 9430
    },
    {
      "epoch": 0.18155417295727516,
      "grad_norm": 578.2861321575405,
      "learning_rate": 1.855438021983556e-05,
      "loss": 420.0583,
      "step": 9440
    },
    {
      "epoch": 0.18174649729303496,
      "grad_norm": 659.5385427960081,
      "learning_rate": 1.8551217777593233e-05,
      "loss": 420.5492,
      "step": 9450
    },
    {
      "epoch": 0.1819388216287948,
      "grad_norm": 534.7839812928478,
      "learning_rate": 1.8548052150163514e-05,
      "loss": 416.5834,
      "step": 9460
    },
    {
      "epoch": 0.18213114596455462,
      "grad_norm": 522.8373626061169,
      "learning_rate": 1.8544883338725544e-05,
      "loss": 404.8246,
      "step": 9470
    },
    {
      "epoch": 0.18232347030031446,
      "grad_norm": 520.1017128871034,
      "learning_rate": 1.8541711344459652e-05,
      "loss": 418.9685,
      "step": 9480
    },
    {
      "epoch": 0.18251579463607429,
      "grad_norm": 560.0962448637667,
      "learning_rate": 1.8538536168547353e-05,
      "loss": 428.3481,
      "step": 9490
    },
    {
      "epoch": 0.1827081189718341,
      "grad_norm": 622.9568825824865,
      "learning_rate": 1.8535357812171356e-05,
      "loss": 434.1384,
      "step": 9500
    },
    {
      "epoch": 0.18290044330759392,
      "grad_norm": 555.6372958533534,
      "learning_rate": 1.8532176276515538e-05,
      "loss": 421.5643,
      "step": 9510
    },
    {
      "epoch": 0.18309276764335375,
      "grad_norm": 522.7995245990104,
      "learning_rate": 1.8528991562764967e-05,
      "loss": 420.3086,
      "step": 9520
    },
    {
      "epoch": 0.18328509197911358,
      "grad_norm": 521.8334563410268,
      "learning_rate": 1.85258036721059e-05,
      "loss": 416.3376,
      "step": 9530
    },
    {
      "epoch": 0.1834774163148734,
      "grad_norm": 499.68224234408467,
      "learning_rate": 1.8522612605725777e-05,
      "loss": 426.5406,
      "step": 9540
    },
    {
      "epoch": 0.18366974065063324,
      "grad_norm": 578.4621884185445,
      "learning_rate": 1.8519418364813215e-05,
      "loss": 419.4123,
      "step": 9550
    },
    {
      "epoch": 0.18386206498639304,
      "grad_norm": 522.0543225762189,
      "learning_rate": 1.851622095055801e-05,
      "loss": 434.0053,
      "step": 9560
    },
    {
      "epoch": 0.18405438932215287,
      "grad_norm": 523.6593173386211,
      "learning_rate": 1.8513020364151155e-05,
      "loss": 419.1064,
      "step": 9570
    },
    {
      "epoch": 0.1842467136579127,
      "grad_norm": 662.5500637187648,
      "learning_rate": 1.850981660678481e-05,
      "loss": 418.0609,
      "step": 9580
    },
    {
      "epoch": 0.18443903799367253,
      "grad_norm": 529.6694191680093,
      "learning_rate": 1.8506609679652323e-05,
      "loss": 415.039,
      "step": 9590
    },
    {
      "epoch": 0.18463136232943236,
      "grad_norm": 572.3020842816497,
      "learning_rate": 1.8503399583948224e-05,
      "loss": 429.1895,
      "step": 9600
    },
    {
      "epoch": 0.1848236866651922,
      "grad_norm": 496.5845821703022,
      "learning_rate": 1.8500186320868215e-05,
      "loss": 408.0225,
      "step": 9610
    },
    {
      "epoch": 0.185016011000952,
      "grad_norm": 531.7513028622932,
      "learning_rate": 1.8496969891609186e-05,
      "loss": 434.5869,
      "step": 9620
    },
    {
      "epoch": 0.18520833533671183,
      "grad_norm": 589.8906599286073,
      "learning_rate": 1.8493750297369208e-05,
      "loss": 430.4186,
      "step": 9630
    },
    {
      "epoch": 0.18540065967247166,
      "grad_norm": 539.6848176097469,
      "learning_rate": 1.849052753934752e-05,
      "loss": 430.6536,
      "step": 9640
    },
    {
      "epoch": 0.1855929840082315,
      "grad_norm": 520.5519171859622,
      "learning_rate": 1.8487301618744552e-05,
      "loss": 431.9636,
      "step": 9650
    },
    {
      "epoch": 0.18578530834399132,
      "grad_norm": 526.9089236180882,
      "learning_rate": 1.84840725367619e-05,
      "loss": 409.1337,
      "step": 9660
    },
    {
      "epoch": 0.18597763267975112,
      "grad_norm": 549.8399229515857,
      "learning_rate": 1.8480840294602352e-05,
      "loss": 414.6718,
      "step": 9670
    },
    {
      "epoch": 0.18616995701551095,
      "grad_norm": 514.7280426238358,
      "learning_rate": 1.8477604893469857e-05,
      "loss": 429.3853,
      "step": 9680
    },
    {
      "epoch": 0.18636228135127078,
      "grad_norm": 558.315549484784,
      "learning_rate": 1.847436633456955e-05,
      "loss": 426.2568,
      "step": 9690
    },
    {
      "epoch": 0.1865546056870306,
      "grad_norm": 528.9386090871537,
      "learning_rate": 1.8471124619107744e-05,
      "loss": 410.7179,
      "step": 9700
    },
    {
      "epoch": 0.18674693002279044,
      "grad_norm": 497.6646727906393,
      "learning_rate": 1.846787974829192e-05,
      "loss": 413.8442,
      "step": 9710
    },
    {
      "epoch": 0.18693925435855027,
      "grad_norm": 569.0517040183468,
      "learning_rate": 1.8464631723330745e-05,
      "loss": 423.8359,
      "step": 9720
    },
    {
      "epoch": 0.18713157869431007,
      "grad_norm": 544.2586133501704,
      "learning_rate": 1.8461380545434054e-05,
      "loss": 406.8935,
      "step": 9730
    },
    {
      "epoch": 0.1873239030300699,
      "grad_norm": 554.8309689483496,
      "learning_rate": 1.8458126215812848e-05,
      "loss": 408.8487,
      "step": 9740
    },
    {
      "epoch": 0.18751622736582974,
      "grad_norm": 690.3802857522378,
      "learning_rate": 1.845486873567932e-05,
      "loss": 415.6544,
      "step": 9750
    },
    {
      "epoch": 0.18770855170158957,
      "grad_norm": 487.27608875891974,
      "learning_rate": 1.8451608106246822e-05,
      "loss": 408.5747,
      "step": 9760
    },
    {
      "epoch": 0.1879008760373494,
      "grad_norm": 1665.5502816082571,
      "learning_rate": 1.8448344328729893e-05,
      "loss": 424.0524,
      "step": 9770
    },
    {
      "epoch": 0.1880932003731092,
      "grad_norm": 523.0179811206775,
      "learning_rate": 1.8445077404344226e-05,
      "loss": 408.2082,
      "step": 9780
    },
    {
      "epoch": 0.18828552470886903,
      "grad_norm": 550.6570585819196,
      "learning_rate": 1.8441807334306702e-05,
      "loss": 415.0986,
      "step": 9790
    },
    {
      "epoch": 0.18847784904462886,
      "grad_norm": 519.3988019514969,
      "learning_rate": 1.8438534119835365e-05,
      "loss": 407.2675,
      "step": 9800
    },
    {
      "epoch": 0.1886701733803887,
      "grad_norm": 542.7050796549003,
      "learning_rate": 1.8435257762149436e-05,
      "loss": 420.9396,
      "step": 9810
    },
    {
      "epoch": 0.18886249771614852,
      "grad_norm": 500.71488346322286,
      "learning_rate": 1.8431978262469305e-05,
      "loss": 427.5086,
      "step": 9820
    },
    {
      "epoch": 0.18905482205190835,
      "grad_norm": 599.7988511294999,
      "learning_rate": 1.8428695622016532e-05,
      "loss": 410.2052,
      "step": 9830
    },
    {
      "epoch": 0.18924714638766815,
      "grad_norm": 529.0811037911369,
      "learning_rate": 1.8425409842013843e-05,
      "loss": 424.4892,
      "step": 9840
    },
    {
      "epoch": 0.18943947072342798,
      "grad_norm": 494.85889023281277,
      "learning_rate": 1.8422120923685135e-05,
      "loss": 406.3137,
      "step": 9850
    },
    {
      "epoch": 0.1896317950591878,
      "grad_norm": 528.8127401570667,
      "learning_rate": 1.8418828868255484e-05,
      "loss": 412.9516,
      "step": 9860
    },
    {
      "epoch": 0.18982411939494764,
      "grad_norm": 537.0267192602489,
      "learning_rate": 1.8415533676951117e-05,
      "loss": 421.6755,
      "step": 9870
    },
    {
      "epoch": 0.19001644373070747,
      "grad_norm": 513.3969136082288,
      "learning_rate": 1.8412235350999444e-05,
      "loss": 425.0687,
      "step": 9880
    },
    {
      "epoch": 0.1902087680664673,
      "grad_norm": 527.09258917363,
      "learning_rate": 1.840893389162903e-05,
      "loss": 445.8872,
      "step": 9890
    },
    {
      "epoch": 0.1904010924022271,
      "grad_norm": 500.25795635803814,
      "learning_rate": 1.8405629300069626e-05,
      "loss": 421.5463,
      "step": 9900
    },
    {
      "epoch": 0.19059341673798694,
      "grad_norm": 545.3130687777858,
      "learning_rate": 1.840232157755213e-05,
      "loss": 423.7159,
      "step": 9910
    },
    {
      "epoch": 0.19078574107374677,
      "grad_norm": 608.618059123915,
      "learning_rate": 1.8399010725308616e-05,
      "loss": 411.9484,
      "step": 9920
    },
    {
      "epoch": 0.1909780654095066,
      "grad_norm": 545.1218608516824,
      "learning_rate": 1.839569674457232e-05,
      "loss": 421.4794,
      "step": 9930
    },
    {
      "epoch": 0.19117038974526643,
      "grad_norm": 504.94712412278,
      "learning_rate": 1.8392379636577647e-05,
      "loss": 432.3682,
      "step": 9940
    },
    {
      "epoch": 0.19136271408102623,
      "grad_norm": 546.9329024442096,
      "learning_rate": 1.8389059402560165e-05,
      "loss": 423.1993,
      "step": 9950
    },
    {
      "epoch": 0.19155503841678606,
      "grad_norm": 502.32540284404547,
      "learning_rate": 1.8385736043756605e-05,
      "loss": 416.5917,
      "step": 9960
    },
    {
      "epoch": 0.1917473627525459,
      "grad_norm": 509.708538396133,
      "learning_rate": 1.838240956140486e-05,
      "loss": 421.7241,
      "step": 9970
    },
    {
      "epoch": 0.19193968708830572,
      "grad_norm": 525.1293879716845,
      "learning_rate": 1.8379079956743996e-05,
      "loss": 433.5008,
      "step": 9980
    },
    {
      "epoch": 0.19213201142406555,
      "grad_norm": 497.61637306609146,
      "learning_rate": 1.8375747231014233e-05,
      "loss": 418.8824,
      "step": 9990
    },
    {
      "epoch": 0.19232433575982538,
      "grad_norm": 572.406381802282,
      "learning_rate": 1.8372411385456956e-05,
      "loss": 420.2644,
      "step": 10000
    },
    {
      "epoch": 0.19251666009558518,
      "grad_norm": 536.8888580901031,
      "learning_rate": 1.8369072421314717e-05,
      "loss": 416.1535,
      "step": 10010
    },
    {
      "epoch": 0.19270898443134502,
      "grad_norm": 539.5311818166617,
      "learning_rate": 1.8365730339831212e-05,
      "loss": 410.9119,
      "step": 10020
    },
    {
      "epoch": 0.19290130876710485,
      "grad_norm": 523.245654348139,
      "learning_rate": 1.8362385142251328e-05,
      "loss": 414.0204,
      "step": 10030
    },
    {
      "epoch": 0.19309363310286468,
      "grad_norm": 536.474516444986,
      "learning_rate": 1.8359036829821085e-05,
      "loss": 410.5013,
      "step": 10040
    },
    {
      "epoch": 0.1932859574386245,
      "grad_norm": 515.0024761825331,
      "learning_rate": 1.8355685403787677e-05,
      "loss": 407.9228,
      "step": 10050
    },
    {
      "epoch": 0.1934782817743843,
      "grad_norm": 527.230463860649,
      "learning_rate": 1.8352330865399457e-05,
      "loss": 407.3625,
      "step": 10060
    },
    {
      "epoch": 0.19367060611014414,
      "grad_norm": 642.3091996399853,
      "learning_rate": 1.834897321590593e-05,
      "loss": 412.4108,
      "step": 10070
    },
    {
      "epoch": 0.19386293044590397,
      "grad_norm": 540.0323798323193,
      "learning_rate": 1.8345612456557767e-05,
      "loss": 403.2625,
      "step": 10080
    },
    {
      "epoch": 0.1940552547816638,
      "grad_norm": 575.5953061828302,
      "learning_rate": 1.83422485886068e-05,
      "loss": 406.2676,
      "step": 10090
    },
    {
      "epoch": 0.19424757911742363,
      "grad_norm": 656.8468315467643,
      "learning_rate": 1.833888161330601e-05,
      "loss": 428.1458,
      "step": 10100
    },
    {
      "epoch": 0.19443990345318346,
      "grad_norm": 576.5050128483706,
      "learning_rate": 1.833551153190954e-05,
      "loss": 411.1276,
      "step": 10110
    },
    {
      "epoch": 0.19463222778894326,
      "grad_norm": 566.5844914913063,
      "learning_rate": 1.8332138345672686e-05,
      "loss": 407.6996,
      "step": 10120
    },
    {
      "epoch": 0.1948245521247031,
      "grad_norm": 544.3805983252685,
      "learning_rate": 1.832876205585191e-05,
      "loss": 418.7956,
      "step": 10130
    },
    {
      "epoch": 0.19501687646046292,
      "grad_norm": 501.0720341320704,
      "learning_rate": 1.8325382663704826e-05,
      "loss": 412.451,
      "step": 10140
    },
    {
      "epoch": 0.19520920079622275,
      "grad_norm": 574.3306194291852,
      "learning_rate": 1.8322000170490194e-05,
      "loss": 404.8666,
      "step": 10150
    },
    {
      "epoch": 0.19540152513198258,
      "grad_norm": 544.7437684730154,
      "learning_rate": 1.831861457746794e-05,
      "loss": 415.9554,
      "step": 10160
    },
    {
      "epoch": 0.19559384946774241,
      "grad_norm": 529.9547074984308,
      "learning_rate": 1.8315225885899144e-05,
      "loss": 417.0008,
      "step": 10170
    },
    {
      "epoch": 0.19578617380350222,
      "grad_norm": 528.9189770261082,
      "learning_rate": 1.8311834097046038e-05,
      "loss": 406.9135,
      "step": 10180
    },
    {
      "epoch": 0.19597849813926205,
      "grad_norm": 548.3876544489895,
      "learning_rate": 1.8308439212172e-05,
      "loss": 417.8647,
      "step": 10190
    },
    {
      "epoch": 0.19617082247502188,
      "grad_norm": 557.4102901046656,
      "learning_rate": 1.830504123254158e-05,
      "loss": 425.4412,
      "step": 10200
    },
    {
      "epoch": 0.1963631468107817,
      "grad_norm": 521.4352191559834,
      "learning_rate": 1.830164015942046e-05,
      "loss": 417.0783,
      "step": 10210
    },
    {
      "epoch": 0.19655547114654154,
      "grad_norm": 1354.3653106645966,
      "learning_rate": 1.8298235994075488e-05,
      "loss": 408.7513,
      "step": 10220
    },
    {
      "epoch": 0.19674779548230134,
      "grad_norm": 502.4758460590438,
      "learning_rate": 1.829482873777466e-05,
      "loss": 423.2251,
      "step": 10230
    },
    {
      "epoch": 0.19694011981806117,
      "grad_norm": 1317.3278761813658,
      "learning_rate": 1.8291418391787116e-05,
      "loss": 426.1462,
      "step": 10240
    },
    {
      "epoch": 0.197132444153821,
      "grad_norm": 500.45854920060214,
      "learning_rate": 1.8288004957383162e-05,
      "loss": 414.126,
      "step": 10250
    },
    {
      "epoch": 0.19732476848958083,
      "grad_norm": 530.5593261671512,
      "learning_rate": 1.8284588435834242e-05,
      "loss": 415.1306,
      "step": 10260
    },
    {
      "epoch": 0.19751709282534066,
      "grad_norm": 519.9739315938061,
      "learning_rate": 1.828116882841295e-05,
      "loss": 416.7035,
      "step": 10270
    },
    {
      "epoch": 0.1977094171611005,
      "grad_norm": 524.8884270767219,
      "learning_rate": 1.8277746136393042e-05,
      "loss": 426.8112,
      "step": 10280
    },
    {
      "epoch": 0.1979017414968603,
      "grad_norm": 524.3284840384099,
      "learning_rate": 1.827432036104941e-05,
      "loss": 412.6839,
      "step": 10290
    },
    {
      "epoch": 0.19809406583262013,
      "grad_norm": 503.33586436995387,
      "learning_rate": 1.8270891503658096e-05,
      "loss": 412.2217,
      "step": 10300
    },
    {
      "epoch": 0.19828639016837996,
      "grad_norm": 546.0982813263964,
      "learning_rate": 1.8267459565496298e-05,
      "loss": 419.6713,
      "step": 10310
    },
    {
      "epoch": 0.19847871450413979,
      "grad_norm": 511.52602813136303,
      "learning_rate": 1.8264024547842346e-05,
      "loss": 413.6944,
      "step": 10320
    },
    {
      "epoch": 0.19867103883989962,
      "grad_norm": 518.0320879974444,
      "learning_rate": 1.8260586451975745e-05,
      "loss": 409.5191,
      "step": 10330
    },
    {
      "epoch": 0.19886336317565942,
      "grad_norm": 554.6757703597846,
      "learning_rate": 1.825714527917711e-05,
      "loss": 424.7147,
      "step": 10340
    },
    {
      "epoch": 0.19905568751141925,
      "grad_norm": 623.6783765698033,
      "learning_rate": 1.8253701030728235e-05,
      "loss": 418.4444,
      "step": 10350
    },
    {
      "epoch": 0.19924801184717908,
      "grad_norm": 515.3750985319824,
      "learning_rate": 1.8250253707912036e-05,
      "loss": 420.7123,
      "step": 10360
    },
    {
      "epoch": 0.1994403361829389,
      "grad_norm": 515.2598514122711,
      "learning_rate": 1.8246803312012593e-05,
      "loss": 401.7974,
      "step": 10370
    },
    {
      "epoch": 0.19963266051869874,
      "grad_norm": 519.8969995253459,
      "learning_rate": 1.8243349844315116e-05,
      "loss": 404.8949,
      "step": 10380
    },
    {
      "epoch": 0.19982498485445857,
      "grad_norm": 515.4214128174964,
      "learning_rate": 1.8239893306105966e-05,
      "loss": 406.2662,
      "step": 10390
    },
    {
      "epoch": 0.20001730919021837,
      "grad_norm": 486.61132653951995,
      "learning_rate": 1.823643369867264e-05,
      "loss": 407.4249,
      "step": 10400
    },
    {
      "epoch": 0.2002096335259782,
      "grad_norm": 474.6687855552454,
      "learning_rate": 1.8232971023303798e-05,
      "loss": 400.1477,
      "step": 10410
    },
    {
      "epoch": 0.20040195786173803,
      "grad_norm": 516.3606444909349,
      "learning_rate": 1.8229505281289216e-05,
      "loss": 421.7604,
      "step": 10420
    },
    {
      "epoch": 0.20059428219749786,
      "grad_norm": 492.3888246658609,
      "learning_rate": 1.8226036473919836e-05,
      "loss": 401.3217,
      "step": 10430
    },
    {
      "epoch": 0.2007866065332577,
      "grad_norm": 495.7204526571697,
      "learning_rate": 1.8222564602487724e-05,
      "loss": 412.7046,
      "step": 10440
    },
    {
      "epoch": 0.20097893086901752,
      "grad_norm": 510.8913180339026,
      "learning_rate": 1.82190896682861e-05,
      "loss": 410.9804,
      "step": 10450
    },
    {
      "epoch": 0.20117125520477733,
      "grad_norm": 523.1719450732343,
      "learning_rate": 1.8215611672609316e-05,
      "loss": 397.5099,
      "step": 10460
    },
    {
      "epoch": 0.20136357954053716,
      "grad_norm": 523.0320537596587,
      "learning_rate": 1.821213061675287e-05,
      "loss": 409.6579,
      "step": 10470
    },
    {
      "epoch": 0.201555903876297,
      "grad_norm": 503.92571864409314,
      "learning_rate": 1.8208646502013395e-05,
      "loss": 410.3252,
      "step": 10480
    },
    {
      "epoch": 0.20174822821205682,
      "grad_norm": 515.077737494529,
      "learning_rate": 1.820515932968867e-05,
      "loss": 410.6644,
      "step": 10490
    },
    {
      "epoch": 0.20194055254781665,
      "grad_norm": 531.9273618650429,
      "learning_rate": 1.8201669101077608e-05,
      "loss": 412.3847,
      "step": 10500
    },
    {
      "epoch": 0.20213287688357645,
      "grad_norm": 551.2360124120286,
      "learning_rate": 1.819817581748026e-05,
      "loss": 428.8473,
      "step": 10510
    },
    {
      "epoch": 0.20232520121933628,
      "grad_norm": 511.39960278942294,
      "learning_rate": 1.8194679480197817e-05,
      "loss": 409.6806,
      "step": 10520
    },
    {
      "epoch": 0.2025175255550961,
      "grad_norm": 534.2559471885919,
      "learning_rate": 1.8191180090532608e-05,
      "loss": 417.9859,
      "step": 10530
    },
    {
      "epoch": 0.20270984989085594,
      "grad_norm": 501.39575407441913,
      "learning_rate": 1.8187677649788097e-05,
      "loss": 406.1442,
      "step": 10540
    },
    {
      "epoch": 0.20290217422661577,
      "grad_norm": 528.6779947784553,
      "learning_rate": 1.8184172159268884e-05,
      "loss": 417.8737,
      "step": 10550
    },
    {
      "epoch": 0.2030944985623756,
      "grad_norm": 524.4025728183602,
      "learning_rate": 1.818066362028071e-05,
      "loss": 411.8988,
      "step": 10560
    },
    {
      "epoch": 0.2032868228981354,
      "grad_norm": 595.632420522306,
      "learning_rate": 1.8177152034130442e-05,
      "loss": 426.585,
      "step": 10570
    },
    {
      "epoch": 0.20347914723389524,
      "grad_norm": 510.89813590025693,
      "learning_rate": 1.8173637402126093e-05,
      "loss": 425.3784,
      "step": 10580
    },
    {
      "epoch": 0.20367147156965507,
      "grad_norm": 513.174119758378,
      "learning_rate": 1.8170119725576808e-05,
      "loss": 415.6281,
      "step": 10590
    },
    {
      "epoch": 0.2038637959054149,
      "grad_norm": 505.80519222164645,
      "learning_rate": 1.8166599005792852e-05,
      "loss": 410.2682,
      "step": 10600
    },
    {
      "epoch": 0.20405612024117473,
      "grad_norm": 544.6945237539323,
      "learning_rate": 1.816307524408565e-05,
      "loss": 404.582,
      "step": 10610
    },
    {
      "epoch": 0.20424844457693453,
      "grad_norm": 520.879276910382,
      "learning_rate": 1.8159548441767732e-05,
      "loss": 408.4701,
      "step": 10620
    },
    {
      "epoch": 0.20444076891269436,
      "grad_norm": 477.1172774106002,
      "learning_rate": 1.8156018600152777e-05,
      "loss": 412.6134,
      "step": 10630
    },
    {
      "epoch": 0.2046330932484542,
      "grad_norm": 604.3311480874454,
      "learning_rate": 1.81524857205556e-05,
      "loss": 416.6437,
      "step": 10640
    },
    {
      "epoch": 0.20482541758421402,
      "grad_norm": 554.351834134022,
      "learning_rate": 1.814894980429213e-05,
      "loss": 430.0104,
      "step": 10650
    },
    {
      "epoch": 0.20501774191997385,
      "grad_norm": 513.705109326578,
      "learning_rate": 1.8145410852679447e-05,
      "loss": 411.1,
      "step": 10660
    },
    {
      "epoch": 0.20521006625573368,
      "grad_norm": 512.0090545380464,
      "learning_rate": 1.8141868867035745e-05,
      "loss": 407.035,
      "step": 10670
    },
    {
      "epoch": 0.20540239059149348,
      "grad_norm": 535.4145618185576,
      "learning_rate": 1.8138323848680354e-05,
      "loss": 405.9104,
      "step": 10680
    },
    {
      "epoch": 0.2055947149272533,
      "grad_norm": 500.0198126110811,
      "learning_rate": 1.813477579893374e-05,
      "loss": 409.5342,
      "step": 10690
    },
    {
      "epoch": 0.20578703926301314,
      "grad_norm": 542.989652958391,
      "learning_rate": 1.8131224719117497e-05,
      "loss": 409.1924,
      "step": 10700
    },
    {
      "epoch": 0.20597936359877297,
      "grad_norm": 539.8265827173909,
      "learning_rate": 1.8127670610554332e-05,
      "loss": 412.5566,
      "step": 10710
    },
    {
      "epoch": 0.2061716879345328,
      "grad_norm": 516.3627271813201,
      "learning_rate": 1.81241134745681e-05,
      "loss": 405.9106,
      "step": 10720
    },
    {
      "epoch": 0.20636401227029263,
      "grad_norm": 537.7886605094616,
      "learning_rate": 1.812055331248377e-05,
      "loss": 418.2211,
      "step": 10730
    },
    {
      "epoch": 0.20655633660605244,
      "grad_norm": 531.4364475833352,
      "learning_rate": 1.811699012562745e-05,
      "loss": 414.2586,
      "step": 10740
    },
    {
      "epoch": 0.20674866094181227,
      "grad_norm": 538.2785709537496,
      "learning_rate": 1.8113423915326362e-05,
      "loss": 422.3838,
      "step": 10750
    },
    {
      "epoch": 0.2069409852775721,
      "grad_norm": 537.5690068034183,
      "learning_rate": 1.8109854682908864e-05,
      "loss": 421.5552,
      "step": 10760
    },
    {
      "epoch": 0.20713330961333193,
      "grad_norm": 561.8138977513785,
      "learning_rate": 1.8106282429704436e-05,
      "loss": 413.7626,
      "step": 10770
    },
    {
      "epoch": 0.20732563394909176,
      "grad_norm": 530.3932512048439,
      "learning_rate": 1.810270715704368e-05,
      "loss": 421.6035,
      "step": 10780
    },
    {
      "epoch": 0.20751795828485156,
      "grad_norm": 528.0991394333488,
      "learning_rate": 1.809912886625833e-05,
      "loss": 410.8647,
      "step": 10790
    },
    {
      "epoch": 0.2077102826206114,
      "grad_norm": 484.9974792580638,
      "learning_rate": 1.8095547558681243e-05,
      "loss": 426.7599,
      "step": 10800
    },
    {
      "epoch": 0.20790260695637122,
      "grad_norm": 542.5439490570113,
      "learning_rate": 1.809196323564639e-05,
      "loss": 423.1881,
      "step": 10810
    },
    {
      "epoch": 0.20809493129213105,
      "grad_norm": 498.19894917262286,
      "learning_rate": 1.8088375898488873e-05,
      "loss": 405.2069,
      "step": 10820
    },
    {
      "epoch": 0.20828725562789088,
      "grad_norm": 503.4427791105701,
      "learning_rate": 1.808478554854492e-05,
      "loss": 416.0496,
      "step": 10830
    },
    {
      "epoch": 0.2084795799636507,
      "grad_norm": 504.0248423442197,
      "learning_rate": 1.8081192187151873e-05,
      "loss": 409.1388,
      "step": 10840
    },
    {
      "epoch": 0.20867190429941052,
      "grad_norm": 507.40350465388576,
      "learning_rate": 1.8077595815648202e-05,
      "loss": 404.2461,
      "step": 10850
    },
    {
      "epoch": 0.20886422863517035,
      "grad_norm": 490.9277981461252,
      "learning_rate": 1.8073996435373494e-05,
      "loss": 413.7467,
      "step": 10860
    },
    {
      "epoch": 0.20905655297093018,
      "grad_norm": 531.0572375677967,
      "learning_rate": 1.8070394047668466e-05,
      "loss": 410.2942,
      "step": 10870
    },
    {
      "epoch": 0.20924887730669,
      "grad_norm": 502.7509631142382,
      "learning_rate": 1.8066788653874936e-05,
      "loss": 399.2792,
      "step": 10880
    },
    {
      "epoch": 0.20944120164244984,
      "grad_norm": 501.4967718909897,
      "learning_rate": 1.806318025533586e-05,
      "loss": 411.6022,
      "step": 10890
    },
    {
      "epoch": 0.20963352597820964,
      "grad_norm": 541.2113298892342,
      "learning_rate": 1.805956885339531e-05,
      "loss": 421.027,
      "step": 10900
    },
    {
      "epoch": 0.20982585031396947,
      "grad_norm": 554.7804161009711,
      "learning_rate": 1.8055954449398472e-05,
      "loss": 423.1687,
      "step": 10910
    },
    {
      "epoch": 0.2100181746497293,
      "grad_norm": 577.628870388658,
      "learning_rate": 1.8052337044691648e-05,
      "loss": 407.1125,
      "step": 10920
    },
    {
      "epoch": 0.21021049898548913,
      "grad_norm": 481.78927926055235,
      "learning_rate": 1.8048716640622262e-05,
      "loss": 415.8896,
      "step": 10930
    },
    {
      "epoch": 0.21040282332124896,
      "grad_norm": 516.7605948800522,
      "learning_rate": 1.8045093238538856e-05,
      "loss": 422.0905,
      "step": 10940
    },
    {
      "epoch": 0.2105951476570088,
      "grad_norm": 516.2539525082283,
      "learning_rate": 1.8041466839791087e-05,
      "loss": 410.3786,
      "step": 10950
    },
    {
      "epoch": 0.2107874719927686,
      "grad_norm": 493.1980218771841,
      "learning_rate": 1.8037837445729733e-05,
      "loss": 410.5156,
      "step": 10960
    },
    {
      "epoch": 0.21097979632852842,
      "grad_norm": 542.3396341658416,
      "learning_rate": 1.803420505770668e-05,
      "loss": 405.1088,
      "step": 10970
    },
    {
      "epoch": 0.21117212066428825,
      "grad_norm": 517.8850704321621,
      "learning_rate": 1.803056967707493e-05,
      "loss": 405.2751,
      "step": 10980
    },
    {
      "epoch": 0.21136444500004808,
      "grad_norm": 482.43393376057685,
      "learning_rate": 1.8026931305188603e-05,
      "loss": 416.6449,
      "step": 10990
    },
    {
      "epoch": 0.21155676933580791,
      "grad_norm": 497.11512978520756,
      "learning_rate": 1.802328994340294e-05,
      "loss": 415.077,
      "step": 11000
    },
    {
      "epoch": 0.21174909367156775,
      "grad_norm": 582.0300690596135,
      "learning_rate": 1.8019645593074275e-05,
      "loss": 409.7937,
      "step": 11010
    },
    {
      "epoch": 0.21194141800732755,
      "grad_norm": 504.7725007010852,
      "learning_rate": 1.8015998255560082e-05,
      "loss": 394.9882,
      "step": 11020
    },
    {
      "epoch": 0.21213374234308738,
      "grad_norm": 528.387777493482,
      "learning_rate": 1.801234793221892e-05,
      "loss": 400.3989,
      "step": 11030
    },
    {
      "epoch": 0.2123260666788472,
      "grad_norm": 467.06890144295636,
      "learning_rate": 1.800869462441049e-05,
      "loss": 386.9421,
      "step": 11040
    },
    {
      "epoch": 0.21251839101460704,
      "grad_norm": 510.04313871139175,
      "learning_rate": 1.8005038333495572e-05,
      "loss": 409.0464,
      "step": 11050
    },
    {
      "epoch": 0.21271071535036687,
      "grad_norm": 564.1915961993303,
      "learning_rate": 1.8001379060836088e-05,
      "loss": 410.9805,
      "step": 11060
    },
    {
      "epoch": 0.21290303968612667,
      "grad_norm": 495.65682272039265,
      "learning_rate": 1.7997716807795046e-05,
      "loss": 418.5792,
      "step": 11070
    },
    {
      "epoch": 0.2130953640218865,
      "grad_norm": 532.2225276470311,
      "learning_rate": 1.7994051575736585e-05,
      "loss": 402.7211,
      "step": 11080
    },
    {
      "epoch": 0.21328768835764633,
      "grad_norm": 460.54738570783405,
      "learning_rate": 1.7990383366025935e-05,
      "loss": 413.6778,
      "step": 11090
    },
    {
      "epoch": 0.21348001269340616,
      "grad_norm": 563.9381616777556,
      "learning_rate": 1.7986712180029448e-05,
      "loss": 402.2324,
      "step": 11100
    },
    {
      "epoch": 0.213672337029166,
      "grad_norm": 557.7393630089101,
      "learning_rate": 1.798303801911458e-05,
      "loss": 409.0407,
      "step": 11110
    },
    {
      "epoch": 0.21386466136492582,
      "grad_norm": 464.7240432785937,
      "learning_rate": 1.7979360884649894e-05,
      "loss": 406.1634,
      "step": 11120
    },
    {
      "epoch": 0.21405698570068563,
      "grad_norm": 519.6839412427632,
      "learning_rate": 1.7975680778005058e-05,
      "loss": 403.6662,
      "step": 11130
    },
    {
      "epoch": 0.21424931003644546,
      "grad_norm": 520.6865750830053,
      "learning_rate": 1.7971997700550856e-05,
      "loss": 412.701,
      "step": 11140
    },
    {
      "epoch": 0.2144416343722053,
      "grad_norm": 532.444937859847,
      "learning_rate": 1.7968311653659177e-05,
      "loss": 413.2475,
      "step": 11150
    },
    {
      "epoch": 0.21463395870796512,
      "grad_norm": 513.7990626397482,
      "learning_rate": 1.7964622638703003e-05,
      "loss": 414.746,
      "step": 11160
    },
    {
      "epoch": 0.21482628304372495,
      "grad_norm": 572.7112169973356,
      "learning_rate": 1.796093065705644e-05,
      "loss": 401.9031,
      "step": 11170
    },
    {
      "epoch": 0.21501860737948475,
      "grad_norm": 539.3257332781393,
      "learning_rate": 1.7957235710094686e-05,
      "loss": 394.6945,
      "step": 11180
    },
    {
      "epoch": 0.21521093171524458,
      "grad_norm": 497.9512755554518,
      "learning_rate": 1.7953537799194042e-05,
      "loss": 411.4992,
      "step": 11190
    },
    {
      "epoch": 0.2154032560510044,
      "grad_norm": 514.0453246141834,
      "learning_rate": 1.7949836925731934e-05,
      "loss": 421.7444,
      "step": 11200
    },
    {
      "epoch": 0.21559558038676424,
      "grad_norm": 511.29439032055143,
      "learning_rate": 1.7946133091086858e-05,
      "loss": 414.8357,
      "step": 11210
    },
    {
      "epoch": 0.21578790472252407,
      "grad_norm": 529.0137973053264,
      "learning_rate": 1.7942426296638447e-05,
      "loss": 411.2592,
      "step": 11220
    },
    {
      "epoch": 0.2159802290582839,
      "grad_norm": 486.62988226461977,
      "learning_rate": 1.7938716543767412e-05,
      "loss": 405.5414,
      "step": 11230
    },
    {
      "epoch": 0.2161725533940437,
      "grad_norm": 562.0868523718697,
      "learning_rate": 1.7935003833855576e-05,
      "loss": 416.3538,
      "step": 11240
    },
    {
      "epoch": 0.21636487772980353,
      "grad_norm": 536.2486147099664,
      "learning_rate": 1.7931288168285863e-05,
      "loss": 408.2126,
      "step": 11250
    },
    {
      "epoch": 0.21655720206556336,
      "grad_norm": 523.3446221690112,
      "learning_rate": 1.79275695484423e-05,
      "loss": 403.6002,
      "step": 11260
    },
    {
      "epoch": 0.2167495264013232,
      "grad_norm": 604.8486062384638,
      "learning_rate": 1.7923847975710003e-05,
      "loss": 410.7343,
      "step": 11270
    },
    {
      "epoch": 0.21694185073708303,
      "grad_norm": 524.8062386188335,
      "learning_rate": 1.7920123451475203e-05,
      "loss": 421.6781,
      "step": 11280
    },
    {
      "epoch": 0.21713417507284286,
      "grad_norm": 501.58986967916536,
      "learning_rate": 1.7916395977125227e-05,
      "loss": 402.3847,
      "step": 11290
    },
    {
      "epoch": 0.21732649940860266,
      "grad_norm": 497.1449132876496,
      "learning_rate": 1.7912665554048486e-05,
      "loss": 413.3773,
      "step": 11300
    },
    {
      "epoch": 0.2175188237443625,
      "grad_norm": 492.83356620960893,
      "learning_rate": 1.7908932183634515e-05,
      "loss": 407.8787,
      "step": 11310
    },
    {
      "epoch": 0.21771114808012232,
      "grad_norm": 493.8232105420126,
      "learning_rate": 1.790519586727392e-05,
      "loss": 413.0008,
      "step": 11320
    },
    {
      "epoch": 0.21790347241588215,
      "grad_norm": 511.52334341120667,
      "learning_rate": 1.790145660635843e-05,
      "loss": 409.4102,
      "step": 11330
    },
    {
      "epoch": 0.21809579675164198,
      "grad_norm": 485.98400907009767,
      "learning_rate": 1.7897714402280844e-05,
      "loss": 412.8913,
      "step": 11340
    },
    {
      "epoch": 0.21828812108740178,
      "grad_norm": 529.7334552106086,
      "learning_rate": 1.789396925643508e-05,
      "loss": 426.9078,
      "step": 11350
    },
    {
      "epoch": 0.2184804454231616,
      "grad_norm": 511.9475347544775,
      "learning_rate": 1.7890221170216144e-05,
      "loss": 404.4276,
      "step": 11360
    },
    {
      "epoch": 0.21867276975892144,
      "grad_norm": 493.1437768338328,
      "learning_rate": 1.788647014502013e-05,
      "loss": 401.8671,
      "step": 11370
    },
    {
      "epoch": 0.21886509409468127,
      "grad_norm": 528.4241937381482,
      "learning_rate": 1.7882716182244242e-05,
      "loss": 400.7938,
      "step": 11380
    },
    {
      "epoch": 0.2190574184304411,
      "grad_norm": 514.5513086963199,
      "learning_rate": 1.7878959283286758e-05,
      "loss": 404.7781,
      "step": 11390
    },
    {
      "epoch": 0.21924974276620093,
      "grad_norm": 572.177159998548,
      "learning_rate": 1.787519944954707e-05,
      "loss": 403.674,
      "step": 11400
    },
    {
      "epoch": 0.21944206710196074,
      "grad_norm": 587.3937210816053,
      "learning_rate": 1.7871436682425645e-05,
      "loss": 405.5933,
      "step": 11410
    },
    {
      "epoch": 0.21963439143772057,
      "grad_norm": 1165.6479538826209,
      "learning_rate": 1.786767098332406e-05,
      "loss": 413.9093,
      "step": 11420
    },
    {
      "epoch": 0.2198267157734804,
      "grad_norm": 478.1020254282917,
      "learning_rate": 1.7863902353644972e-05,
      "loss": 393.565,
      "step": 11430
    },
    {
      "epoch": 0.22001904010924023,
      "grad_norm": 492.4046983344083,
      "learning_rate": 1.7860130794792137e-05,
      "loss": 404.3131,
      "step": 11440
    },
    {
      "epoch": 0.22021136444500006,
      "grad_norm": 551.1213544422876,
      "learning_rate": 1.7856356308170394e-05,
      "loss": 413.0437,
      "step": 11450
    },
    {
      "epoch": 0.22040368878075986,
      "grad_norm": 578.7197861461077,
      "learning_rate": 1.7852578895185675e-05,
      "loss": 421.9933,
      "step": 11460
    },
    {
      "epoch": 0.2205960131165197,
      "grad_norm": 647.3976116167437,
      "learning_rate": 1.7848798557245008e-05,
      "loss": 413.7521,
      "step": 11470
    },
    {
      "epoch": 0.22078833745227952,
      "grad_norm": 535.4830253335278,
      "learning_rate": 1.7845015295756506e-05,
      "loss": 417.5281,
      "step": 11480
    },
    {
      "epoch": 0.22098066178803935,
      "grad_norm": 486.6803793001602,
      "learning_rate": 1.784122911212937e-05,
      "loss": 406.7004,
      "step": 11490
    },
    {
      "epoch": 0.22117298612379918,
      "grad_norm": 473.37143544185335,
      "learning_rate": 1.7837440007773895e-05,
      "loss": 401.1356,
      "step": 11500
    },
    {
      "epoch": 0.221365310459559,
      "grad_norm": 537.4054390682203,
      "learning_rate": 1.783364798410146e-05,
      "loss": 403.5251,
      "step": 11510
    },
    {
      "epoch": 0.22155763479531881,
      "grad_norm": 545.1805428427086,
      "learning_rate": 1.782985304252452e-05,
      "loss": 404.6236,
      "step": 11520
    },
    {
      "epoch": 0.22174995913107864,
      "grad_norm": 519.245572912035,
      "learning_rate": 1.7826055184456643e-05,
      "loss": 414.8062,
      "step": 11530
    },
    {
      "epoch": 0.22194228346683847,
      "grad_norm": 658.6581715472673,
      "learning_rate": 1.7822254411312455e-05,
      "loss": 411.0006,
      "step": 11540
    },
    {
      "epoch": 0.2221346078025983,
      "grad_norm": 505.36060857534454,
      "learning_rate": 1.781845072450769e-05,
      "loss": 399.6282,
      "step": 11550
    },
    {
      "epoch": 0.22232693213835814,
      "grad_norm": 628.4760229429728,
      "learning_rate": 1.7814644125459157e-05,
      "loss": 412.5159,
      "step": 11560
    },
    {
      "epoch": 0.22251925647411797,
      "grad_norm": 570.8506741190848,
      "learning_rate": 1.781083461558475e-05,
      "loss": 402.9419,
      "step": 11570
    },
    {
      "epoch": 0.22271158080987777,
      "grad_norm": 513.782891668989,
      "learning_rate": 1.7807022196303447e-05,
      "loss": 408.1303,
      "step": 11580
    },
    {
      "epoch": 0.2229039051456376,
      "grad_norm": 501.2028742366539,
      "learning_rate": 1.7803206869035318e-05,
      "loss": 403.4169,
      "step": 11590
    },
    {
      "epoch": 0.22309622948139743,
      "grad_norm": 519.7860185158811,
      "learning_rate": 1.7799388635201498e-05,
      "loss": 410.635,
      "step": 11600
    },
    {
      "epoch": 0.22328855381715726,
      "grad_norm": 471.10481959750166,
      "learning_rate": 1.7795567496224226e-05,
      "loss": 404.7676,
      "step": 11610
    },
    {
      "epoch": 0.2234808781529171,
      "grad_norm": 525.033983104948,
      "learning_rate": 1.7791743453526812e-05,
      "loss": 411.6758,
      "step": 11620
    },
    {
      "epoch": 0.2236732024886769,
      "grad_norm": 506.5371381669754,
      "learning_rate": 1.778791650853364e-05,
      "loss": 405.416,
      "step": 11630
    },
    {
      "epoch": 0.22386552682443672,
      "grad_norm": 542.2771407282709,
      "learning_rate": 1.77840866626702e-05,
      "loss": 402.695,
      "step": 11640
    },
    {
      "epoch": 0.22405785116019655,
      "grad_norm": 489.49733714748487,
      "learning_rate": 1.7780253917363026e-05,
      "loss": 407.5623,
      "step": 11650
    },
    {
      "epoch": 0.22425017549595638,
      "grad_norm": 537.2319933491199,
      "learning_rate": 1.7776418274039767e-05,
      "loss": 404.8456,
      "step": 11660
    },
    {
      "epoch": 0.2244424998317162,
      "grad_norm": 598.2556782268568,
      "learning_rate": 1.7772579734129136e-05,
      "loss": 411.4863,
      "step": 11670
    },
    {
      "epoch": 0.22463482416747604,
      "grad_norm": 475.40109318967745,
      "learning_rate": 1.776873829906092e-05,
      "loss": 405.2609,
      "step": 11680
    },
    {
      "epoch": 0.22482714850323585,
      "grad_norm": 580.8312358329363,
      "learning_rate": 1.776489397026599e-05,
      "loss": 422.9873,
      "step": 11690
    },
    {
      "epoch": 0.22501947283899568,
      "grad_norm": 541.0152834448922,
      "learning_rate": 1.7761046749176302e-05,
      "loss": 399.0734,
      "step": 11700
    },
    {
      "epoch": 0.2252117971747555,
      "grad_norm": 540.985137145699,
      "learning_rate": 1.7757196637224874e-05,
      "loss": 409.2649,
      "step": 11710
    },
    {
      "epoch": 0.22540412151051534,
      "grad_norm": 497.6752914536555,
      "learning_rate": 1.7753343635845817e-05,
      "loss": 402.5908,
      "step": 11720
    },
    {
      "epoch": 0.22559644584627517,
      "grad_norm": 490.4517266066059,
      "learning_rate": 1.7749487746474305e-05,
      "loss": 402.6892,
      "step": 11730
    },
    {
      "epoch": 0.22578877018203497,
      "grad_norm": 508.15890225317537,
      "learning_rate": 1.7745628970546592e-05,
      "loss": 408.5553,
      "step": 11740
    },
    {
      "epoch": 0.2259810945177948,
      "grad_norm": 496.42361433271526,
      "learning_rate": 1.774176730950001e-05,
      "loss": 388.7166,
      "step": 11750
    },
    {
      "epoch": 0.22617341885355463,
      "grad_norm": 500.959891259976,
      "learning_rate": 1.7737902764772967e-05,
      "loss": 417.2532,
      "step": 11760
    },
    {
      "epoch": 0.22636574318931446,
      "grad_norm": 626.9393137696261,
      "learning_rate": 1.773403533780494e-05,
      "loss": 402.5101,
      "step": 11770
    },
    {
      "epoch": 0.2265580675250743,
      "grad_norm": 491.34585223492417,
      "learning_rate": 1.7730165030036482e-05,
      "loss": 401.105,
      "step": 11780
    },
    {
      "epoch": 0.22675039186083412,
      "grad_norm": 529.215403286338,
      "learning_rate": 1.772629184290922e-05,
      "loss": 402.1561,
      "step": 11790
    },
    {
      "epoch": 0.22694271619659392,
      "grad_norm": 508.9339115369465,
      "learning_rate": 1.7722415777865845e-05,
      "loss": 395.756,
      "step": 11800
    },
    {
      "epoch": 0.22713504053235375,
      "grad_norm": 520.80420386598,
      "learning_rate": 1.771853683635014e-05,
      "loss": 402.6091,
      "step": 11810
    },
    {
      "epoch": 0.22732736486811359,
      "grad_norm": 476.04083255495846,
      "learning_rate": 1.7714655019806932e-05,
      "loss": 403.5805,
      "step": 11820
    },
    {
      "epoch": 0.22751968920387342,
      "grad_norm": 550.7913044470721,
      "learning_rate": 1.7710770329682145e-05,
      "loss": 416.7787,
      "step": 11830
    },
    {
      "epoch": 0.22771201353963325,
      "grad_norm": 505.5853126590449,
      "learning_rate": 1.770688276742276e-05,
      "loss": 409.7005,
      "step": 11840
    },
    {
      "epoch": 0.22790433787539308,
      "grad_norm": 515.014467614758,
      "learning_rate": 1.770299233447682e-05,
      "loss": 404.2626,
      "step": 11850
    },
    {
      "epoch": 0.22809666221115288,
      "grad_norm": 473.6618477989999,
      "learning_rate": 1.769909903229346e-05,
      "loss": 398.4137,
      "step": 11860
    },
    {
      "epoch": 0.2282889865469127,
      "grad_norm": 689.5768301005184,
      "learning_rate": 1.7695202862322863e-05,
      "loss": 403.3624,
      "step": 11870
    },
    {
      "epoch": 0.22848131088267254,
      "grad_norm": 513.5343481211617,
      "learning_rate": 1.769130382601629e-05,
      "loss": 410.9995,
      "step": 11880
    },
    {
      "epoch": 0.22867363521843237,
      "grad_norm": 537.7866624459691,
      "learning_rate": 1.768740192482607e-05,
      "loss": 408.0978,
      "step": 11890
    },
    {
      "epoch": 0.2288659595541922,
      "grad_norm": 519.0439904283717,
      "learning_rate": 1.7683497160205595e-05,
      "loss": 408.7737,
      "step": 11900
    },
    {
      "epoch": 0.229058283889952,
      "grad_norm": 520.000781685986,
      "learning_rate": 1.7679589533609323e-05,
      "loss": 412.1072,
      "step": 11910
    },
    {
      "epoch": 0.22925060822571183,
      "grad_norm": 504.07478428993613,
      "learning_rate": 1.767567904649278e-05,
      "loss": 418.4932,
      "step": 11920
    },
    {
      "epoch": 0.22944293256147166,
      "grad_norm": 548.4436125907348,
      "learning_rate": 1.7671765700312567e-05,
      "loss": 412.4563,
      "step": 11930
    },
    {
      "epoch": 0.2296352568972315,
      "grad_norm": 516.7824892769144,
      "learning_rate": 1.7667849496526327e-05,
      "loss": 399.2371,
      "step": 11940
    },
    {
      "epoch": 0.22982758123299132,
      "grad_norm": 548.6939758778841,
      "learning_rate": 1.7663930436592793e-05,
      "loss": 409.4398,
      "step": 11950
    },
    {
      "epoch": 0.23001990556875115,
      "grad_norm": 490.7281621447608,
      "learning_rate": 1.7660008521971744e-05,
      "loss": 405.1733,
      "step": 11960
    },
    {
      "epoch": 0.23021222990451096,
      "grad_norm": 511.16713007607234,
      "learning_rate": 1.765608375412403e-05,
      "loss": 403.498,
      "step": 11970
    },
    {
      "epoch": 0.2304045542402708,
      "grad_norm": 515.4528928202382,
      "learning_rate": 1.7652156134511563e-05,
      "loss": 410.0643,
      "step": 11980
    },
    {
      "epoch": 0.23059687857603062,
      "grad_norm": 479.16690893280617,
      "learning_rate": 1.7648225664597314e-05,
      "loss": 412.6704,
      "step": 11990
    },
    {
      "epoch": 0.23078920291179045,
      "grad_norm": 494.1166178415702,
      "learning_rate": 1.764429234584532e-05,
      "loss": 406.721,
      "step": 12000
    },
    {
      "epoch": 0.23098152724755028,
      "grad_norm": 482.23576240889537,
      "learning_rate": 1.7640356179720674e-05,
      "loss": 386.7119,
      "step": 12010
    },
    {
      "epoch": 0.23117385158331008,
      "grad_norm": 509.9384623017886,
      "learning_rate": 1.7636417167689538e-05,
      "loss": 411.914,
      "step": 12020
    },
    {
      "epoch": 0.2313661759190699,
      "grad_norm": 524.1668014236113,
      "learning_rate": 1.7632475311219125e-05,
      "loss": 394.6435,
      "step": 12030
    },
    {
      "epoch": 0.23155850025482974,
      "grad_norm": 518.9214490502344,
      "learning_rate": 1.7628530611777716e-05,
      "loss": 405.5806,
      "step": 12040
    },
    {
      "epoch": 0.23175082459058957,
      "grad_norm": 525.9921426863668,
      "learning_rate": 1.7624583070834646e-05,
      "loss": 410.241,
      "step": 12050
    },
    {
      "epoch": 0.2319431489263494,
      "grad_norm": 586.9930140832246,
      "learning_rate": 1.7620632689860298e-05,
      "loss": 410.2238,
      "step": 12060
    },
    {
      "epoch": 0.23213547326210923,
      "grad_norm": 535.9314428181268,
      "learning_rate": 1.761667947032614e-05,
      "loss": 414.5618,
      "step": 12070
    },
    {
      "epoch": 0.23232779759786903,
      "grad_norm": 493.13645952804507,
      "learning_rate": 1.761272341370467e-05,
      "loss": 400.5868,
      "step": 12080
    },
    {
      "epoch": 0.23252012193362886,
      "grad_norm": 502.0074543058723,
      "learning_rate": 1.7608764521469456e-05,
      "loss": 394.2096,
      "step": 12090
    },
    {
      "epoch": 0.2327124462693887,
      "grad_norm": 447.8403171622099,
      "learning_rate": 1.760480279509512e-05,
      "loss": 400.9692,
      "step": 12100
    },
    {
      "epoch": 0.23290477060514853,
      "grad_norm": 487.7660518837125,
      "learning_rate": 1.7600838236057342e-05,
      "loss": 408.061,
      "step": 12110
    },
    {
      "epoch": 0.23309709494090836,
      "grad_norm": 504.32252923341,
      "learning_rate": 1.759687084583285e-05,
      "loss": 400.5435,
      "step": 12120
    },
    {
      "epoch": 0.2332894192766682,
      "grad_norm": 521.4183192579749,
      "learning_rate": 1.7592900625899437e-05,
      "loss": 410.5279,
      "step": 12130
    },
    {
      "epoch": 0.233481743612428,
      "grad_norm": 520.2086183398725,
      "learning_rate": 1.7588927577735943e-05,
      "loss": 396.2346,
      "step": 12140
    },
    {
      "epoch": 0.23367406794818782,
      "grad_norm": 541.2095691095466,
      "learning_rate": 1.758495170282226e-05,
      "loss": 402.8773,
      "step": 12150
    },
    {
      "epoch": 0.23386639228394765,
      "grad_norm": 497.34174095448043,
      "learning_rate": 1.7580973002639337e-05,
      "loss": 409.8557,
      "step": 12160
    },
    {
      "epoch": 0.23405871661970748,
      "grad_norm": 515.004309869215,
      "learning_rate": 1.7576991478669174e-05,
      "loss": 403.8827,
      "step": 12170
    },
    {
      "epoch": 0.2342510409554673,
      "grad_norm": 490.6503805558587,
      "learning_rate": 1.7573007132394823e-05,
      "loss": 403.6582,
      "step": 12180
    },
    {
      "epoch": 0.2344433652912271,
      "grad_norm": 486.47122799114857,
      "learning_rate": 1.756901996530039e-05,
      "loss": 406.2761,
      "step": 12190
    },
    {
      "epoch": 0.23463568962698694,
      "grad_norm": 510.7373931463473,
      "learning_rate": 1.7565029978871025e-05,
      "loss": 409.7734,
      "step": 12200
    },
    {
      "epoch": 0.23482801396274677,
      "grad_norm": 503.7204131351527,
      "learning_rate": 1.7561037174592933e-05,
      "loss": 402.4177,
      "step": 12210
    },
    {
      "epoch": 0.2350203382985066,
      "grad_norm": 613.9987338779152,
      "learning_rate": 1.7557041553953368e-05,
      "loss": 396.2206,
      "step": 12220
    },
    {
      "epoch": 0.23521266263426643,
      "grad_norm": 497.95913507578257,
      "learning_rate": 1.7553043118440634e-05,
      "loss": 400.1966,
      "step": 12230
    },
    {
      "epoch": 0.23540498697002626,
      "grad_norm": 515.1885704157239,
      "learning_rate": 1.7549041869544077e-05,
      "loss": 401.2695,
      "step": 12240
    },
    {
      "epoch": 0.23559731130578607,
      "grad_norm": 526.6068757020573,
      "learning_rate": 1.7545037808754105e-05,
      "loss": 394.7766,
      "step": 12250
    },
    {
      "epoch": 0.2357896356415459,
      "grad_norm": 499.4230551838952,
      "learning_rate": 1.754103093756216e-05,
      "loss": 387.9822,
      "step": 12260
    },
    {
      "epoch": 0.23598195997730573,
      "grad_norm": 463.07631244890723,
      "learning_rate": 1.7537021257460732e-05,
      "loss": 398.1867,
      "step": 12270
    },
    {
      "epoch": 0.23617428431306556,
      "grad_norm": 490.5667042577999,
      "learning_rate": 1.7533008769943366e-05,
      "loss": 394.0945,
      "step": 12280
    },
    {
      "epoch": 0.2363666086488254,
      "grad_norm": 486.8045600759694,
      "learning_rate": 1.7528993476504644e-05,
      "loss": 398.946,
      "step": 12290
    },
    {
      "epoch": 0.2365589329845852,
      "grad_norm": 477.7223469608313,
      "learning_rate": 1.7524975378640198e-05,
      "loss": 399.5925,
      "step": 12300
    },
    {
      "epoch": 0.23675125732034502,
      "grad_norm": 490.00348840587196,
      "learning_rate": 1.75209544778467e-05,
      "loss": 400.93,
      "step": 12310
    },
    {
      "epoch": 0.23694358165610485,
      "grad_norm": 526.6182594372053,
      "learning_rate": 1.7516930775621873e-05,
      "loss": 417.465,
      "step": 12320
    },
    {
      "epoch": 0.23713590599186468,
      "grad_norm": 511.61066304254655,
      "learning_rate": 1.751290427346448e-05,
      "loss": 400.8351,
      "step": 12330
    },
    {
      "epoch": 0.2373282303276245,
      "grad_norm": 494.9780163791452,
      "learning_rate": 1.7508874972874325e-05,
      "loss": 408.4519,
      "step": 12340
    },
    {
      "epoch": 0.23752055466338434,
      "grad_norm": 546.2624513285988,
      "learning_rate": 1.7504842875352254e-05,
      "loss": 403.3538,
      "step": 12350
    },
    {
      "epoch": 0.23771287899914414,
      "grad_norm": 499.7649936159823,
      "learning_rate": 1.7500807982400155e-05,
      "loss": 400.1042,
      "step": 12360
    },
    {
      "epoch": 0.23790520333490398,
      "grad_norm": 483.472968093783,
      "learning_rate": 1.749677029552097e-05,
      "loss": 397.2355,
      "step": 12370
    },
    {
      "epoch": 0.2380975276706638,
      "grad_norm": 490.58246762273086,
      "learning_rate": 1.7492729816218656e-05,
      "loss": 404.0796,
      "step": 12380
    },
    {
      "epoch": 0.23828985200642364,
      "grad_norm": 512.0959230228034,
      "learning_rate": 1.7488686545998237e-05,
      "loss": 392.8472,
      "step": 12390
    },
    {
      "epoch": 0.23848217634218347,
      "grad_norm": 577.4452246167682,
      "learning_rate": 1.7484640486365757e-05,
      "loss": 401.7335,
      "step": 12400
    },
    {
      "epoch": 0.2386745006779433,
      "grad_norm": 558.3170516246616,
      "learning_rate": 1.748059163882831e-05,
      "loss": 406.5239,
      "step": 12410
    },
    {
      "epoch": 0.2388668250137031,
      "grad_norm": 505.0730131020004,
      "learning_rate": 1.747654000489402e-05,
      "loss": 399.255,
      "step": 12420
    },
    {
      "epoch": 0.23905914934946293,
      "grad_norm": 482.447998986777,
      "learning_rate": 1.7472485586072062e-05,
      "loss": 397.8921,
      "step": 12430
    },
    {
      "epoch": 0.23925147368522276,
      "grad_norm": 527.1088462627331,
      "learning_rate": 1.746842838387264e-05,
      "loss": 393.3306,
      "step": 12440
    },
    {
      "epoch": 0.2394437980209826,
      "grad_norm": 545.7977664145235,
      "learning_rate": 1.746436839980698e-05,
      "loss": 402.0523,
      "step": 12450
    },
    {
      "epoch": 0.23963612235674242,
      "grad_norm": 476.520326166508,
      "learning_rate": 1.7460305635387372e-05,
      "loss": 407.1924,
      "step": 12460
    },
    {
      "epoch": 0.23982844669250222,
      "grad_norm": 535.8142212509558,
      "learning_rate": 1.745624009212713e-05,
      "loss": 402.2665,
      "step": 12470
    },
    {
      "epoch": 0.24002077102826205,
      "grad_norm": 523.9774108562103,
      "learning_rate": 1.7452171771540593e-05,
      "loss": 406.2526,
      "step": 12480
    },
    {
      "epoch": 0.24021309536402188,
      "grad_norm": 518.210626682501,
      "learning_rate": 1.744810067514315e-05,
      "loss": 404.5391,
      "step": 12490
    },
    {
      "epoch": 0.24040541969978171,
      "grad_norm": 464.7851437183551,
      "learning_rate": 1.7444026804451214e-05,
      "loss": 401.0928,
      "step": 12500
    },
    {
      "epoch": 0.24059774403554154,
      "grad_norm": 511.1698751849852,
      "learning_rate": 1.7439950160982236e-05,
      "loss": 398.2425,
      "step": 12510
    },
    {
      "epoch": 0.24079006837130137,
      "grad_norm": 494.84867844615843,
      "learning_rate": 1.7435870746254697e-05,
      "loss": 399.8055,
      "step": 12520
    },
    {
      "epoch": 0.24098239270706118,
      "grad_norm": 489.06260400465635,
      "learning_rate": 1.7431788561788116e-05,
      "loss": 397.8201,
      "step": 12530
    },
    {
      "epoch": 0.241174717042821,
      "grad_norm": 460.25957431996915,
      "learning_rate": 1.742770360910303e-05,
      "loss": 409.0396,
      "step": 12540
    },
    {
      "epoch": 0.24136704137858084,
      "grad_norm": 471.6208125880307,
      "learning_rate": 1.7423615889721027e-05,
      "loss": 395.8972,
      "step": 12550
    },
    {
      "epoch": 0.24155936571434067,
      "grad_norm": 485.304085396627,
      "learning_rate": 1.7419525405164705e-05,
      "loss": 411.7617,
      "step": 12560
    },
    {
      "epoch": 0.2417516900501005,
      "grad_norm": 478.4964002415521,
      "learning_rate": 1.7415432156957713e-05,
      "loss": 403.1268,
      "step": 12570
    },
    {
      "epoch": 0.24194401438586033,
      "grad_norm": 542.8550894111927,
      "learning_rate": 1.741133614662471e-05,
      "loss": 403.8331,
      "step": 12580
    },
    {
      "epoch": 0.24213633872162013,
      "grad_norm": 523.1138850783602,
      "learning_rate": 1.7407237375691394e-05,
      "loss": 414.3643,
      "step": 12590
    },
    {
      "epoch": 0.24232866305737996,
      "grad_norm": 524.4502760690234,
      "learning_rate": 1.740313584568449e-05,
      "loss": 395.4663,
      "step": 12600
    },
    {
      "epoch": 0.2425209873931398,
      "grad_norm": 478.83937448034806,
      "learning_rate": 1.739903155813175e-05,
      "loss": 409.1269,
      "step": 12610
    },
    {
      "epoch": 0.24271331172889962,
      "grad_norm": 506.720439151921,
      "learning_rate": 1.7394924514561955e-05,
      "loss": 403.15,
      "step": 12620
    },
    {
      "epoch": 0.24290563606465945,
      "grad_norm": 453.23440418633027,
      "learning_rate": 1.739081471650491e-05,
      "loss": 394.0672,
      "step": 12630
    },
    {
      "epoch": 0.24309796040041926,
      "grad_norm": 489.07783280922706,
      "learning_rate": 1.7386702165491443e-05,
      "loss": 409.9712,
      "step": 12640
    },
    {
      "epoch": 0.24329028473617909,
      "grad_norm": 520.0588970362297,
      "learning_rate": 1.738258686305342e-05,
      "loss": 396.5729,
      "step": 12650
    },
    {
      "epoch": 0.24348260907193892,
      "grad_norm": 549.5339785495672,
      "learning_rate": 1.7378468810723713e-05,
      "loss": 407.8526,
      "step": 12660
    },
    {
      "epoch": 0.24367493340769875,
      "grad_norm": 500.9440479359089,
      "learning_rate": 1.7374348010036235e-05,
      "loss": 405.7856,
      "step": 12670
    },
    {
      "epoch": 0.24386725774345858,
      "grad_norm": 493.83052897723354,
      "learning_rate": 1.7370224462525916e-05,
      "loss": 401.2475,
      "step": 12680
    },
    {
      "epoch": 0.2440595820792184,
      "grad_norm": 550.5896558027431,
      "learning_rate": 1.736609816972871e-05,
      "loss": 404.3938,
      "step": 12690
    },
    {
      "epoch": 0.2442519064149782,
      "grad_norm": 493.96843272790596,
      "learning_rate": 1.7361969133181585e-05,
      "loss": 385.9142,
      "step": 12700
    },
    {
      "epoch": 0.24444423075073804,
      "grad_norm": 540.9977871999351,
      "learning_rate": 1.735783735442255e-05,
      "loss": 403.5269,
      "step": 12710
    },
    {
      "epoch": 0.24463655508649787,
      "grad_norm": 510.8629280468882,
      "learning_rate": 1.7353702834990617e-05,
      "loss": 425.3035,
      "step": 12720
    },
    {
      "epoch": 0.2448288794222577,
      "grad_norm": 544.1964570733115,
      "learning_rate": 1.7349565576425828e-05,
      "loss": 386.6944,
      "step": 12730
    },
    {
      "epoch": 0.24502120375801753,
      "grad_norm": 516.9311598070396,
      "learning_rate": 1.7345425580269245e-05,
      "loss": 403.9177,
      "step": 12740
    },
    {
      "epoch": 0.24521352809377733,
      "grad_norm": 539.7679779782113,
      "learning_rate": 1.734128284806295e-05,
      "loss": 397.8656,
      "step": 12750
    },
    {
      "epoch": 0.24540585242953716,
      "grad_norm": 514.0789548166919,
      "learning_rate": 1.7337137381350033e-05,
      "loss": 397.7061,
      "step": 12760
    },
    {
      "epoch": 0.245598176765297,
      "grad_norm": 503.9537545664038,
      "learning_rate": 1.7332989181674623e-05,
      "loss": 386.2186,
      "step": 12770
    },
    {
      "epoch": 0.24579050110105682,
      "grad_norm": 517.8654270094286,
      "learning_rate": 1.7328838250581846e-05,
      "loss": 395.1542,
      "step": 12780
    },
    {
      "epoch": 0.24598282543681665,
      "grad_norm": 512.873321453811,
      "learning_rate": 1.7324684589617862e-05,
      "loss": 394.3579,
      "step": 12790
    },
    {
      "epoch": 0.24617514977257648,
      "grad_norm": 582.1269460735923,
      "learning_rate": 1.7320528200329846e-05,
      "loss": 403.6785,
      "step": 12800
    },
    {
      "epoch": 0.2463674741083363,
      "grad_norm": 548.9962404996943,
      "learning_rate": 1.7316369084265973e-05,
      "loss": 390.926,
      "step": 12810
    },
    {
      "epoch": 0.24655979844409612,
      "grad_norm": 555.163049282594,
      "learning_rate": 1.731220724297545e-05,
      "loss": 386.4555,
      "step": 12820
    },
    {
      "epoch": 0.24675212277985595,
      "grad_norm": 631.6775856350101,
      "learning_rate": 1.730804267800849e-05,
      "loss": 401.4694,
      "step": 12830
    },
    {
      "epoch": 0.24694444711561578,
      "grad_norm": 602.7890327694288,
      "learning_rate": 1.7303875390916338e-05,
      "loss": 388.7352,
      "step": 12840
    },
    {
      "epoch": 0.2471367714513756,
      "grad_norm": 526.4574971592747,
      "learning_rate": 1.729970538325122e-05,
      "loss": 389.842,
      "step": 12850
    },
    {
      "epoch": 0.24732909578713544,
      "grad_norm": 529.7133168402188,
      "learning_rate": 1.7295532656566413e-05,
      "loss": 405.9639,
      "step": 12860
    },
    {
      "epoch": 0.24752142012289524,
      "grad_norm": 507.10701570307333,
      "learning_rate": 1.729135721241618e-05,
      "loss": 392.3189,
      "step": 12870
    },
    {
      "epoch": 0.24771374445865507,
      "grad_norm": 519.7929294037986,
      "learning_rate": 1.7287179052355803e-05,
      "loss": 407.6865,
      "step": 12880
    },
    {
      "epoch": 0.2479060687944149,
      "grad_norm": 545.1179594167253,
      "learning_rate": 1.7282998177941586e-05,
      "loss": 410.6187,
      "step": 12890
    },
    {
      "epoch": 0.24809839313017473,
      "grad_norm": 479.04999404581645,
      "learning_rate": 1.7278814590730826e-05,
      "loss": 393.8491,
      "step": 12900
    },
    {
      "epoch": 0.24829071746593456,
      "grad_norm": 488.16137893157065,
      "learning_rate": 1.7274628292281846e-05,
      "loss": 392.4036,
      "step": 12910
    },
    {
      "epoch": 0.24848304180169437,
      "grad_norm": 491.38506954197476,
      "learning_rate": 1.727043928415397e-05,
      "loss": 396.4322,
      "step": 12920
    },
    {
      "epoch": 0.2486753661374542,
      "grad_norm": 500.6580624908905,
      "learning_rate": 1.726624756790754e-05,
      "loss": 395.8262,
      "step": 12930
    },
    {
      "epoch": 0.24886769047321403,
      "grad_norm": 481.581111220438,
      "learning_rate": 1.7262053145103893e-05,
      "loss": 405.2584,
      "step": 12940
    },
    {
      "epoch": 0.24906001480897386,
      "grad_norm": 559.6000491084448,
      "learning_rate": 1.7257856017305387e-05,
      "loss": 402.569,
      "step": 12950
    },
    {
      "epoch": 0.2492523391447337,
      "grad_norm": 523.9419639673177,
      "learning_rate": 1.725365618607538e-05,
      "loss": 393.6709,
      "step": 12960
    },
    {
      "epoch": 0.24944466348049352,
      "grad_norm": 484.91979811737974,
      "learning_rate": 1.7249453652978244e-05,
      "loss": 394.9398,
      "step": 12970
    },
    {
      "epoch": 0.24963698781625332,
      "grad_norm": 458.8732700503137,
      "learning_rate": 1.7245248419579353e-05,
      "loss": 404.7233,
      "step": 12980
    },
    {
      "epoch": 0.24982931215201315,
      "grad_norm": 577.4922009839557,
      "learning_rate": 1.7241040487445082e-05,
      "loss": 392.0792,
      "step": 12990
    },
    {
      "epoch": 0.250021636487773,
      "grad_norm": 478.1826052259454,
      "learning_rate": 1.723682985814282e-05,
      "loss": 394.2405,
      "step": 13000
    },
    {
      "epoch": 0.2502139608235328,
      "grad_norm": 531.693456444788,
      "learning_rate": 1.7232616533240958e-05,
      "loss": 399.8074,
      "step": 13010
    },
    {
      "epoch": 0.25040628515929264,
      "grad_norm": 522.370861953215,
      "learning_rate": 1.7228400514308884e-05,
      "loss": 404.5146,
      "step": 13020
    },
    {
      "epoch": 0.25059860949505247,
      "grad_norm": 538.5987587828238,
      "learning_rate": 1.7224181802917003e-05,
      "loss": 393.8064,
      "step": 13030
    },
    {
      "epoch": 0.2507909338308123,
      "grad_norm": 488.79228764564897,
      "learning_rate": 1.721996040063671e-05,
      "loss": 401.6155,
      "step": 13040
    },
    {
      "epoch": 0.25098325816657213,
      "grad_norm": 575.877792016825,
      "learning_rate": 1.7215736309040408e-05,
      "loss": 394.1593,
      "step": 13050
    },
    {
      "epoch": 0.2511755825023319,
      "grad_norm": 573.6376108302794,
      "learning_rate": 1.7211509529701507e-05,
      "loss": 394.0724,
      "step": 13060
    },
    {
      "epoch": 0.25136790683809174,
      "grad_norm": 502.46104757318847,
      "learning_rate": 1.7207280064194403e-05,
      "loss": 400.841,
      "step": 13070
    },
    {
      "epoch": 0.25156023117385157,
      "grad_norm": 593.6212569720309,
      "learning_rate": 1.7203047914094514e-05,
      "loss": 392.9266,
      "step": 13080
    },
    {
      "epoch": 0.2517525555096114,
      "grad_norm": 534.8493134328581,
      "learning_rate": 1.7198813080978235e-05,
      "loss": 398.4978,
      "step": 13090
    },
    {
      "epoch": 0.25194487984537123,
      "grad_norm": 508.50000822439955,
      "learning_rate": 1.7194575566422975e-05,
      "loss": 402.9541,
      "step": 13100
    },
    {
      "epoch": 0.25213720418113106,
      "grad_norm": 503.46795981620033,
      "learning_rate": 1.719033537200714e-05,
      "loss": 387.8704,
      "step": 13110
    },
    {
      "epoch": 0.2523295285168909,
      "grad_norm": 552.8418782386735,
      "learning_rate": 1.7186092499310133e-05,
      "loss": 400.2994,
      "step": 13120
    },
    {
      "epoch": 0.2525218528526507,
      "grad_norm": 490.59823005418565,
      "learning_rate": 1.7181846949912347e-05,
      "loss": 394.2668,
      "step": 13130
    },
    {
      "epoch": 0.25271417718841055,
      "grad_norm": 456.0405912709396,
      "learning_rate": 1.717759872539519e-05,
      "loss": 387.0123,
      "step": 13140
    },
    {
      "epoch": 0.2529065015241704,
      "grad_norm": 527.3645399737716,
      "learning_rate": 1.7173347827341046e-05,
      "loss": 391.4127,
      "step": 13150
    },
    {
      "epoch": 0.2530988258599302,
      "grad_norm": 500.5945889224094,
      "learning_rate": 1.7169094257333307e-05,
      "loss": 407.2625,
      "step": 13160
    },
    {
      "epoch": 0.25329115019569,
      "grad_norm": 560.8620724587252,
      "learning_rate": 1.716483801695636e-05,
      "loss": 402.7937,
      "step": 13170
    },
    {
      "epoch": 0.2534834745314498,
      "grad_norm": 515.6376667044018,
      "learning_rate": 1.7160579107795587e-05,
      "loss": 402.5097,
      "step": 13180
    },
    {
      "epoch": 0.25367579886720965,
      "grad_norm": 495.8000810614592,
      "learning_rate": 1.715631753143735e-05,
      "loss": 390.1797,
      "step": 13190
    },
    {
      "epoch": 0.2538681232029695,
      "grad_norm": 506.2588397467559,
      "learning_rate": 1.715205328946903e-05,
      "loss": 406.007,
      "step": 13200
    },
    {
      "epoch": 0.2540604475387293,
      "grad_norm": 486.92849925236,
      "learning_rate": 1.7147786383478978e-05,
      "loss": 393.3736,
      "step": 13210
    },
    {
      "epoch": 0.25425277187448914,
      "grad_norm": 472.11951109854493,
      "learning_rate": 1.7143516815056545e-05,
      "loss": 405.0434,
      "step": 13220
    },
    {
      "epoch": 0.25444509621024897,
      "grad_norm": 485.4528015488731,
      "learning_rate": 1.713924458579208e-05,
      "loss": 413.6899,
      "step": 13230
    },
    {
      "epoch": 0.2546374205460088,
      "grad_norm": 464.97778065736014,
      "learning_rate": 1.7134969697276912e-05,
      "loss": 391.6427,
      "step": 13240
    },
    {
      "epoch": 0.2548297448817686,
      "grad_norm": 518.6565357349871,
      "learning_rate": 1.7130692151103373e-05,
      "loss": 391.0698,
      "step": 13250
    },
    {
      "epoch": 0.25502206921752846,
      "grad_norm": 520.9885861519131,
      "learning_rate": 1.7126411948864776e-05,
      "loss": 394.4865,
      "step": 13260
    },
    {
      "epoch": 0.2552143935532883,
      "grad_norm": 538.6008653690043,
      "learning_rate": 1.7122129092155422e-05,
      "loss": 409.9459,
      "step": 13270
    },
    {
      "epoch": 0.2554067178890481,
      "grad_norm": 523.6484876908094,
      "learning_rate": 1.7117843582570608e-05,
      "loss": 394.9764,
      "step": 13280
    },
    {
      "epoch": 0.2555990422248079,
      "grad_norm": 500.8952287163879,
      "learning_rate": 1.711355542170661e-05,
      "loss": 382.9939,
      "step": 13290
    },
    {
      "epoch": 0.2557913665605677,
      "grad_norm": 506.43507480449324,
      "learning_rate": 1.710926461116071e-05,
      "loss": 385.2951,
      "step": 13300
    },
    {
      "epoch": 0.25598369089632755,
      "grad_norm": 465.0683375799653,
      "learning_rate": 1.710497115253115e-05,
      "loss": 397.325,
      "step": 13310
    },
    {
      "epoch": 0.2561760152320874,
      "grad_norm": 479.49249798039165,
      "learning_rate": 1.7100675047417178e-05,
      "loss": 394.8781,
      "step": 13320
    },
    {
      "epoch": 0.2563683395678472,
      "grad_norm": 491.94065035437836,
      "learning_rate": 1.7096376297419027e-05,
      "loss": 395.2518,
      "step": 13330
    },
    {
      "epoch": 0.25656066390360704,
      "grad_norm": 482.79102308649334,
      "learning_rate": 1.70920749041379e-05,
      "loss": 399.4447,
      "step": 13340
    },
    {
      "epoch": 0.2567529882393669,
      "grad_norm": 501.13073800526644,
      "learning_rate": 1.7087770869176005e-05,
      "loss": 393.9273,
      "step": 13350
    },
    {
      "epoch": 0.2569453125751267,
      "grad_norm": 487.66926214288037,
      "learning_rate": 1.7083464194136517e-05,
      "loss": 394.5412,
      "step": 13360
    },
    {
      "epoch": 0.25713763691088654,
      "grad_norm": 501.87717967736893,
      "learning_rate": 1.707915488062361e-05,
      "loss": 397.7266,
      "step": 13370
    },
    {
      "epoch": 0.25732996124664637,
      "grad_norm": 476.5338352256334,
      "learning_rate": 1.7074842930242418e-05,
      "loss": 391.353,
      "step": 13380
    },
    {
      "epoch": 0.2575222855824062,
      "grad_norm": 498.2845587910708,
      "learning_rate": 1.7070528344599083e-05,
      "loss": 396.7885,
      "step": 13390
    },
    {
      "epoch": 0.25771460991816597,
      "grad_norm": 461.0064445443026,
      "learning_rate": 1.7066211125300713e-05,
      "loss": 386.897,
      "step": 13400
    },
    {
      "epoch": 0.2579069342539258,
      "grad_norm": 560.8637582656743,
      "learning_rate": 1.70618912739554e-05,
      "loss": 386.8501,
      "step": 13410
    },
    {
      "epoch": 0.25809925858968563,
      "grad_norm": 469.4646679721641,
      "learning_rate": 1.705756879217222e-05,
      "loss": 393.7109,
      "step": 13420
    },
    {
      "epoch": 0.25829158292544546,
      "grad_norm": 511.92778143924806,
      "learning_rate": 1.7053243681561225e-05,
      "loss": 410.817,
      "step": 13430
    },
    {
      "epoch": 0.2584839072612053,
      "grad_norm": 528.1499790472774,
      "learning_rate": 1.7048915943733444e-05,
      "loss": 398.5653,
      "step": 13440
    },
    {
      "epoch": 0.2586762315969651,
      "grad_norm": 462.0393087196226,
      "learning_rate": 1.704458558030089e-05,
      "loss": 391.2169,
      "step": 13450
    },
    {
      "epoch": 0.25886855593272495,
      "grad_norm": 523.8683383069764,
      "learning_rate": 1.704025259287656e-05,
      "loss": 381.1235,
      "step": 13460
    },
    {
      "epoch": 0.2590608802684848,
      "grad_norm": 513.565424708402,
      "learning_rate": 1.7035916983074405e-05,
      "loss": 396.7118,
      "step": 13470
    },
    {
      "epoch": 0.2592532046042446,
      "grad_norm": 501.4891746489902,
      "learning_rate": 1.7031578752509377e-05,
      "loss": 408.4267,
      "step": 13480
    },
    {
      "epoch": 0.25944552894000444,
      "grad_norm": 487.54437565653166,
      "learning_rate": 1.70272379027974e-05,
      "loss": 384.9325,
      "step": 13490
    },
    {
      "epoch": 0.2596378532757643,
      "grad_norm": 564.2934710791556,
      "learning_rate": 1.7022894435555356e-05,
      "loss": 394.9,
      "step": 13500
    },
    {
      "epoch": 0.25983017761152405,
      "grad_norm": 539.3788348620219,
      "learning_rate": 1.7018548352401123e-05,
      "loss": 393.6208,
      "step": 13510
    },
    {
      "epoch": 0.2600225019472839,
      "grad_norm": 473.30599910217467,
      "learning_rate": 1.7014199654953543e-05,
      "loss": 389.3117,
      "step": 13520
    },
    {
      "epoch": 0.2602148262830437,
      "grad_norm": 515.5798223507278,
      "learning_rate": 1.700984834483244e-05,
      "loss": 392.133,
      "step": 13530
    },
    {
      "epoch": 0.26040715061880354,
      "grad_norm": 504.6545423291964,
      "learning_rate": 1.7005494423658598e-05,
      "loss": 395.9388,
      "step": 13540
    },
    {
      "epoch": 0.26059947495456337,
      "grad_norm": 483.3783504290054,
      "learning_rate": 1.7001137893053782e-05,
      "loss": 385.5046,
      "step": 13550
    },
    {
      "epoch": 0.2607917992903232,
      "grad_norm": 490.27407806206344,
      "learning_rate": 1.6996778754640727e-05,
      "loss": 389.0065,
      "step": 13560
    },
    {
      "epoch": 0.26098412362608303,
      "grad_norm": 589.0909621649263,
      "learning_rate": 1.6992417010043144e-05,
      "loss": 392.4468,
      "step": 13570
    },
    {
      "epoch": 0.26117644796184286,
      "grad_norm": 520.0976561948931,
      "learning_rate": 1.6988052660885707e-05,
      "loss": 390.2126,
      "step": 13580
    },
    {
      "epoch": 0.2613687722976027,
      "grad_norm": 463.3302057527136,
      "learning_rate": 1.6983685708794064e-05,
      "loss": 386.2181,
      "step": 13590
    },
    {
      "epoch": 0.2615610966333625,
      "grad_norm": 477.1543438658674,
      "learning_rate": 1.6979316155394834e-05,
      "loss": 393.59,
      "step": 13600
    },
    {
      "epoch": 0.26175342096912235,
      "grad_norm": 496.3409429170343,
      "learning_rate": 1.6974944002315605e-05,
      "loss": 388.0618,
      "step": 13610
    },
    {
      "epoch": 0.2619457453048821,
      "grad_norm": 497.3053983864858,
      "learning_rate": 1.697056925118493e-05,
      "loss": 390.2847,
      "step": 13620
    },
    {
      "epoch": 0.26213806964064196,
      "grad_norm": 461.43407013193405,
      "learning_rate": 1.696619190363233e-05,
      "loss": 404.1869,
      "step": 13630
    },
    {
      "epoch": 0.2623303939764018,
      "grad_norm": 506.0156931045527,
      "learning_rate": 1.69618119612883e-05,
      "loss": 386.771,
      "step": 13640
    },
    {
      "epoch": 0.2625227183121616,
      "grad_norm": 487.07559184360855,
      "learning_rate": 1.695742942578429e-05,
      "loss": 393.2927,
      "step": 13650
    },
    {
      "epoch": 0.26271504264792145,
      "grad_norm": 469.9499839843871,
      "learning_rate": 1.6953044298752724e-05,
      "loss": 389.4216,
      "step": 13660
    },
    {
      "epoch": 0.2629073669836813,
      "grad_norm": 538.5546733998156,
      "learning_rate": 1.694865658182699e-05,
      "loss": 410.8956,
      "step": 13670
    },
    {
      "epoch": 0.2630996913194411,
      "grad_norm": 476.89414366245006,
      "learning_rate": 1.6944266276641442e-05,
      "loss": 393.5947,
      "step": 13680
    },
    {
      "epoch": 0.26329201565520094,
      "grad_norm": 501.89278804852927,
      "learning_rate": 1.6939873384831394e-05,
      "loss": 385.8564,
      "step": 13690
    },
    {
      "epoch": 0.26348433999096077,
      "grad_norm": 456.58366158960615,
      "learning_rate": 1.6935477908033124e-05,
      "loss": 397.3424,
      "step": 13700
    },
    {
      "epoch": 0.2636766643267206,
      "grad_norm": 485.9892859826436,
      "learning_rate": 1.6931079847883877e-05,
      "loss": 395.0168,
      "step": 13710
    },
    {
      "epoch": 0.26386898866248043,
      "grad_norm": 515.9973931537564,
      "learning_rate": 1.692667920602186e-05,
      "loss": 395.985,
      "step": 13720
    },
    {
      "epoch": 0.2640613129982402,
      "grad_norm": 497.03522776962814,
      "learning_rate": 1.6922275984086233e-05,
      "loss": 402.1692,
      "step": 13730
    },
    {
      "epoch": 0.26425363733400004,
      "grad_norm": 564.9704882150211,
      "learning_rate": 1.691787018371713e-05,
      "loss": 391.1026,
      "step": 13740
    },
    {
      "epoch": 0.26444596166975987,
      "grad_norm": 454.8470592230021,
      "learning_rate": 1.6913461806555635e-05,
      "loss": 401.463,
      "step": 13750
    },
    {
      "epoch": 0.2646382860055197,
      "grad_norm": 477.6611023031933,
      "learning_rate": 1.6909050854243797e-05,
      "loss": 395.7901,
      "step": 13760
    },
    {
      "epoch": 0.2648306103412795,
      "grad_norm": 486.2583210183533,
      "learning_rate": 1.690463732842462e-05,
      "loss": 392.4859,
      "step": 13770
    },
    {
      "epoch": 0.26502293467703936,
      "grad_norm": 482.2059275168781,
      "learning_rate": 1.6900221230742073e-05,
      "loss": 397.4946,
      "step": 13780
    },
    {
      "epoch": 0.2652152590127992,
      "grad_norm": 515.6381207804257,
      "learning_rate": 1.689580256284108e-05,
      "loss": 406.7184,
      "step": 13790
    },
    {
      "epoch": 0.265407583348559,
      "grad_norm": 492.64407558052153,
      "learning_rate": 1.689138132636752e-05,
      "loss": 391.2247,
      "step": 13800
    },
    {
      "epoch": 0.26559990768431885,
      "grad_norm": 482.2207179184786,
      "learning_rate": 1.688695752296823e-05,
      "loss": 387.1333,
      "step": 13810
    },
    {
      "epoch": 0.2657922320200787,
      "grad_norm": 496.6269979700899,
      "learning_rate": 1.6882531154291007e-05,
      "loss": 391.7797,
      "step": 13820
    },
    {
      "epoch": 0.2659845563558385,
      "grad_norm": 494.5378802620604,
      "learning_rate": 1.6878102221984593e-05,
      "loss": 402.8014,
      "step": 13830
    },
    {
      "epoch": 0.26617688069159834,
      "grad_norm": 518.8220787544024,
      "learning_rate": 1.6873670727698702e-05,
      "loss": 405.8197,
      "step": 13840
    },
    {
      "epoch": 0.2663692050273581,
      "grad_norm": 437.1255653537561,
      "learning_rate": 1.686923667308398e-05,
      "loss": 396.0104,
      "step": 13850
    },
    {
      "epoch": 0.26656152936311794,
      "grad_norm": 489.23596809080226,
      "learning_rate": 1.6864800059792057e-05,
      "loss": 393.2639,
      "step": 13860
    },
    {
      "epoch": 0.2667538536988778,
      "grad_norm": 479.96032692242295,
      "learning_rate": 1.686036088947548e-05,
      "loss": 415.1752,
      "step": 13870
    },
    {
      "epoch": 0.2669461780346376,
      "grad_norm": 531.1733812480209,
      "learning_rate": 1.6855919163787777e-05,
      "loss": 401.2976,
      "step": 13880
    },
    {
      "epoch": 0.26713850237039743,
      "grad_norm": 475.35689357547085,
      "learning_rate": 1.6851474884383416e-05,
      "loss": 387.0333,
      "step": 13890
    },
    {
      "epoch": 0.26733082670615727,
      "grad_norm": 491.80161008614084,
      "learning_rate": 1.6847028052917814e-05,
      "loss": 393.2367,
      "step": 13900
    },
    {
      "epoch": 0.2675231510419171,
      "grad_norm": 573.1466331439958,
      "learning_rate": 1.6842578671047345e-05,
      "loss": 386.1845,
      "step": 13910
    },
    {
      "epoch": 0.2677154753776769,
      "grad_norm": 493.13460555719877,
      "learning_rate": 1.683812674042933e-05,
      "loss": 379.9144,
      "step": 13920
    },
    {
      "epoch": 0.26790779971343676,
      "grad_norm": 462.4759753129118,
      "learning_rate": 1.683367226272204e-05,
      "loss": 390.1884,
      "step": 13930
    },
    {
      "epoch": 0.2681001240491966,
      "grad_norm": 466.18931766774114,
      "learning_rate": 1.6829215239584695e-05,
      "loss": 393.8679,
      "step": 13940
    },
    {
      "epoch": 0.2682924483849564,
      "grad_norm": 491.7159484394516,
      "learning_rate": 1.6824755672677458e-05,
      "loss": 393.7705,
      "step": 13950
    },
    {
      "epoch": 0.2684847727207162,
      "grad_norm": 588.9248694701199,
      "learning_rate": 1.682029356366145e-05,
      "loss": 393.9963,
      "step": 13960
    },
    {
      "epoch": 0.268677097056476,
      "grad_norm": 501.10526433530686,
      "learning_rate": 1.6815828914198732e-05,
      "loss": 377.7531,
      "step": 13970
    },
    {
      "epoch": 0.26886942139223585,
      "grad_norm": 523.2026065633,
      "learning_rate": 1.6811361725952308e-05,
      "loss": 405.5974,
      "step": 13980
    },
    {
      "epoch": 0.2690617457279957,
      "grad_norm": 506.67394583214536,
      "learning_rate": 1.6806892000586135e-05,
      "loss": 390.8128,
      "step": 13990
    },
    {
      "epoch": 0.2692540700637555,
      "grad_norm": 544.1896223915375,
      "learning_rate": 1.6802419739765114e-05,
      "loss": 389.8078,
      "step": 14000
    },
    {
      "epoch": 0.26944639439951534,
      "grad_norm": 491.96316076754937,
      "learning_rate": 1.679794494515508e-05,
      "loss": 391.9018,
      "step": 14010
    },
    {
      "epoch": 0.2696387187352752,
      "grad_norm": 542.9530459510884,
      "learning_rate": 1.6793467618422828e-05,
      "loss": 385.7537,
      "step": 14020
    },
    {
      "epoch": 0.269831043071035,
      "grad_norm": 535.404844202274,
      "learning_rate": 1.6788987761236088e-05,
      "loss": 395.4638,
      "step": 14030
    },
    {
      "epoch": 0.27002336740679483,
      "grad_norm": 489.38268937887545,
      "learning_rate": 1.6784505375263533e-05,
      "loss": 387.0627,
      "step": 14040
    },
    {
      "epoch": 0.27021569174255466,
      "grad_norm": 493.74504632673415,
      "learning_rate": 1.678002046217477e-05,
      "loss": 390.6436,
      "step": 14050
    },
    {
      "epoch": 0.2704080160783145,
      "grad_norm": 599.7337319808317,
      "learning_rate": 1.6775533023640363e-05,
      "loss": 426.1557,
      "step": 14060
    },
    {
      "epoch": 0.27060034041407427,
      "grad_norm": 623.8127113724317,
      "learning_rate": 1.6771043061331806e-05,
      "loss": 415.4726,
      "step": 14070
    },
    {
      "epoch": 0.2707926647498341,
      "grad_norm": 495.45378039437037,
      "learning_rate": 1.6766550576921533e-05,
      "loss": 387.8605,
      "step": 14080
    },
    {
      "epoch": 0.27098498908559393,
      "grad_norm": 504.15813492014405,
      "learning_rate": 1.676205557208293e-05,
      "loss": 396.13,
      "step": 14090
    },
    {
      "epoch": 0.27117731342135376,
      "grad_norm": 547.2105305436389,
      "learning_rate": 1.67575580484903e-05,
      "loss": 377.4808,
      "step": 14100
    },
    {
      "epoch": 0.2713696377571136,
      "grad_norm": 476.53585613180485,
      "learning_rate": 1.6753058007818906e-05,
      "loss": 395.6949,
      "step": 14110
    },
    {
      "epoch": 0.2715619620928734,
      "grad_norm": 512.0339523348306,
      "learning_rate": 1.674855545174493e-05,
      "loss": 385.6241,
      "step": 14120
    },
    {
      "epoch": 0.27175428642863325,
      "grad_norm": 486.4501433186579,
      "learning_rate": 1.6744050381945507e-05,
      "loss": 398.1865,
      "step": 14130
    },
    {
      "epoch": 0.2719466107643931,
      "grad_norm": 481.17986956039204,
      "learning_rate": 1.67395428000987e-05,
      "loss": 386.8289,
      "step": 14140
    },
    {
      "epoch": 0.2721389351001529,
      "grad_norm": 498.87643368094456,
      "learning_rate": 1.6735032707883502e-05,
      "loss": 391.9912,
      "step": 14150
    },
    {
      "epoch": 0.27233125943591274,
      "grad_norm": 487.406389720137,
      "learning_rate": 1.6730520106979855e-05,
      "loss": 379.5662,
      "step": 14160
    },
    {
      "epoch": 0.2725235837716726,
      "grad_norm": 514.7910188117253,
      "learning_rate": 1.672600499906863e-05,
      "loss": 392.7333,
      "step": 14170
    },
    {
      "epoch": 0.27271590810743235,
      "grad_norm": 511.7755234491555,
      "learning_rate": 1.6721487385831622e-05,
      "loss": 395.4198,
      "step": 14180
    },
    {
      "epoch": 0.2729082324431922,
      "grad_norm": 487.05765967267655,
      "learning_rate": 1.6716967268951574e-05,
      "loss": 393.7244,
      "step": 14190
    },
    {
      "epoch": 0.273100556778952,
      "grad_norm": 486.21869972819553,
      "learning_rate": 1.6712444650112152e-05,
      "loss": 398.4978,
      "step": 14200
    },
    {
      "epoch": 0.27329288111471184,
      "grad_norm": 559.8126928923966,
      "learning_rate": 1.6707919530997956e-05,
      "loss": 401.5352,
      "step": 14210
    },
    {
      "epoch": 0.27348520545047167,
      "grad_norm": 525.4416841159325,
      "learning_rate": 1.6703391913294524e-05,
      "loss": 387.6162,
      "step": 14220
    },
    {
      "epoch": 0.2736775297862315,
      "grad_norm": 518.055564266573,
      "learning_rate": 1.6698861798688312e-05,
      "loss": 395.3866,
      "step": 14230
    },
    {
      "epoch": 0.27386985412199133,
      "grad_norm": 504.06500478413204,
      "learning_rate": 1.6694329188866717e-05,
      "loss": 394.1962,
      "step": 14240
    },
    {
      "epoch": 0.27406217845775116,
      "grad_norm": 476.36847485542984,
      "learning_rate": 1.6689794085518057e-05,
      "loss": 382.5779,
      "step": 14250
    },
    {
      "epoch": 0.274254502793511,
      "grad_norm": 524.5001467805347,
      "learning_rate": 1.668525649033159e-05,
      "loss": 401.2712,
      "step": 14260
    },
    {
      "epoch": 0.2744468271292708,
      "grad_norm": 511.7771635917971,
      "learning_rate": 1.6680716404997482e-05,
      "loss": 390.3644,
      "step": 14270
    },
    {
      "epoch": 0.27463915146503065,
      "grad_norm": 495.52370130659904,
      "learning_rate": 1.667617383120686e-05,
      "loss": 396.2208,
      "step": 14280
    },
    {
      "epoch": 0.2748314758007904,
      "grad_norm": 474.7706071353199,
      "learning_rate": 1.667162877065174e-05,
      "loss": 390.4807,
      "step": 14290
    },
    {
      "epoch": 0.27502380013655026,
      "grad_norm": 479.8887676170485,
      "learning_rate": 1.6667081225025087e-05,
      "loss": 387.1438,
      "step": 14300
    },
    {
      "epoch": 0.2752161244723101,
      "grad_norm": 515.0957857214537,
      "learning_rate": 1.666253119602079e-05,
      "loss": 394.5681,
      "step": 14310
    },
    {
      "epoch": 0.2754084488080699,
      "grad_norm": 503.76659880261764,
      "learning_rate": 1.665797868533366e-05,
      "loss": 382.0825,
      "step": 14320
    },
    {
      "epoch": 0.27560077314382975,
      "grad_norm": 536.9037136753394,
      "learning_rate": 1.6653423694659433e-05,
      "loss": 404.3942,
      "step": 14330
    },
    {
      "epoch": 0.2757930974795896,
      "grad_norm": 558.0341787521504,
      "learning_rate": 1.6648866225694757e-05,
      "loss": 394.4466,
      "step": 14340
    },
    {
      "epoch": 0.2759854218153494,
      "grad_norm": 487.9536840470224,
      "learning_rate": 1.6644306280137227e-05,
      "loss": 388.3453,
      "step": 14350
    },
    {
      "epoch": 0.27617774615110924,
      "grad_norm": 549.9656942677718,
      "learning_rate": 1.6639743859685336e-05,
      "loss": 397.5311,
      "step": 14360
    },
    {
      "epoch": 0.27637007048686907,
      "grad_norm": 467.65786759439186,
      "learning_rate": 1.663517896603852e-05,
      "loss": 391.5482,
      "step": 14370
    },
    {
      "epoch": 0.2765623948226289,
      "grad_norm": 520.5947403263569,
      "learning_rate": 1.6630611600897126e-05,
      "loss": 392.006,
      "step": 14380
    },
    {
      "epoch": 0.27675471915838873,
      "grad_norm": 491.3663480775731,
      "learning_rate": 1.6626041765962413e-05,
      "loss": 403.8781,
      "step": 14390
    },
    {
      "epoch": 0.27694704349414856,
      "grad_norm": 465.5328672239248,
      "learning_rate": 1.662146946293658e-05,
      "loss": 391.9352,
      "step": 14400
    },
    {
      "epoch": 0.27713936782990833,
      "grad_norm": 493.89827320644065,
      "learning_rate": 1.6616894693522727e-05,
      "loss": 385.1233,
      "step": 14410
    },
    {
      "epoch": 0.27733169216566816,
      "grad_norm": 499.00496418938826,
      "learning_rate": 1.6612317459424884e-05,
      "loss": 383.6975,
      "step": 14420
    },
    {
      "epoch": 0.277524016501428,
      "grad_norm": 499.0241251917486,
      "learning_rate": 1.6607737762347987e-05,
      "loss": 391.0595,
      "step": 14430
    },
    {
      "epoch": 0.2777163408371878,
      "grad_norm": 488.43702927538936,
      "learning_rate": 1.6603155603997908e-05,
      "loss": 391.3056,
      "step": 14440
    },
    {
      "epoch": 0.27790866517294766,
      "grad_norm": 513.7568596869821,
      "learning_rate": 1.6598570986081424e-05,
      "loss": 389.4551,
      "step": 14450
    },
    {
      "epoch": 0.2781009895087075,
      "grad_norm": 488.8984804680765,
      "learning_rate": 1.6593983910306225e-05,
      "loss": 386.0653,
      "step": 14460
    },
    {
      "epoch": 0.2782933138444673,
      "grad_norm": 475.5486903350814,
      "learning_rate": 1.658939437838092e-05,
      "loss": 381.9182,
      "step": 14470
    },
    {
      "epoch": 0.27848563818022715,
      "grad_norm": 483.7020740351836,
      "learning_rate": 1.658480239201504e-05,
      "loss": 388.15,
      "step": 14480
    },
    {
      "epoch": 0.278677962515987,
      "grad_norm": 474.2297926591958,
      "learning_rate": 1.6580207952919018e-05,
      "loss": 380.9073,
      "step": 14490
    },
    {
      "epoch": 0.2788702868517468,
      "grad_norm": 532.3122715536738,
      "learning_rate": 1.657561106280421e-05,
      "loss": 386.3968,
      "step": 14500
    },
    {
      "epoch": 0.27906261118750664,
      "grad_norm": 487.3535681886864,
      "learning_rate": 1.6571011723382882e-05,
      "loss": 387.3301,
      "step": 14510
    },
    {
      "epoch": 0.2792549355232664,
      "grad_norm": 488.6986066079575,
      "learning_rate": 1.6566409936368207e-05,
      "loss": 393.3085,
      "step": 14520
    },
    {
      "epoch": 0.27944725985902624,
      "grad_norm": 510.40771262185467,
      "learning_rate": 1.6561805703474285e-05,
      "loss": 380.7416,
      "step": 14530
    },
    {
      "epoch": 0.2796395841947861,
      "grad_norm": 492.0546940217477,
      "learning_rate": 1.655719902641611e-05,
      "loss": 390.49,
      "step": 14540
    },
    {
      "epoch": 0.2798319085305459,
      "grad_norm": 474.56344068259085,
      "learning_rate": 1.6552589906909586e-05,
      "loss": 383.2617,
      "step": 14550
    },
    {
      "epoch": 0.28002423286630573,
      "grad_norm": 536.2365934811677,
      "learning_rate": 1.654797834667155e-05,
      "loss": 374.7162,
      "step": 14560
    },
    {
      "epoch": 0.28021655720206556,
      "grad_norm": 535.0234100697963,
      "learning_rate": 1.6543364347419714e-05,
      "loss": 395.434,
      "step": 14570
    },
    {
      "epoch": 0.2804088815378254,
      "grad_norm": 534.4528084084313,
      "learning_rate": 1.6538747910872733e-05,
      "loss": 390.3084,
      "step": 14580
    },
    {
      "epoch": 0.2806012058735852,
      "grad_norm": 518.1222162094313,
      "learning_rate": 1.6534129038750145e-05,
      "loss": 386.5567,
      "step": 14590
    },
    {
      "epoch": 0.28079353020934505,
      "grad_norm": 632.019829050338,
      "learning_rate": 1.65295077327724e-05,
      "loss": 387.525,
      "step": 14600
    },
    {
      "epoch": 0.2809858545451049,
      "grad_norm": 484.1064687775313,
      "learning_rate": 1.652488399466087e-05,
      "loss": 377.9122,
      "step": 14610
    },
    {
      "epoch": 0.2811781788808647,
      "grad_norm": 469.8652315484747,
      "learning_rate": 1.6520257826137807e-05,
      "loss": 392.4902,
      "step": 14620
    },
    {
      "epoch": 0.2813705032166245,
      "grad_norm": 540.3384561437886,
      "learning_rate": 1.6515629228926396e-05,
      "loss": 396.9217,
      "step": 14630
    },
    {
      "epoch": 0.2815628275523843,
      "grad_norm": 476.7879068094032,
      "learning_rate": 1.6510998204750702e-05,
      "loss": 402.6011,
      "step": 14640
    },
    {
      "epoch": 0.28175515188814415,
      "grad_norm": 458.3851532035752,
      "learning_rate": 1.650636475533571e-05,
      "loss": 386.3834,
      "step": 14650
    },
    {
      "epoch": 0.281947476223904,
      "grad_norm": 488.3971521177812,
      "learning_rate": 1.6501728882407305e-05,
      "loss": 391.1553,
      "step": 14660
    },
    {
      "epoch": 0.2821398005596638,
      "grad_norm": 519.911671875725,
      "learning_rate": 1.649709058769227e-05,
      "loss": 378.8972,
      "step": 14670
    },
    {
      "epoch": 0.28233212489542364,
      "grad_norm": 517.2565307701216,
      "learning_rate": 1.6492449872918293e-05,
      "loss": 378.7213,
      "step": 14680
    },
    {
      "epoch": 0.28252444923118347,
      "grad_norm": 463.86330751564213,
      "learning_rate": 1.6487806739813966e-05,
      "loss": 382.2467,
      "step": 14690
    },
    {
      "epoch": 0.2827167735669433,
      "grad_norm": 491.7713185776623,
      "learning_rate": 1.6483161190108778e-05,
      "loss": 385.834,
      "step": 14700
    },
    {
      "epoch": 0.28290909790270313,
      "grad_norm": 492.04990272489835,
      "learning_rate": 1.6478513225533117e-05,
      "loss": 393.276,
      "step": 14710
    },
    {
      "epoch": 0.28310142223846296,
      "grad_norm": 473.92846724740224,
      "learning_rate": 1.647386284781828e-05,
      "loss": 392.6294,
      "step": 14720
    },
    {
      "epoch": 0.2832937465742228,
      "grad_norm": 473.51366958046833,
      "learning_rate": 1.6469210058696448e-05,
      "loss": 396.0201,
      "step": 14730
    },
    {
      "epoch": 0.28348607090998257,
      "grad_norm": 513.5508599611892,
      "learning_rate": 1.646455485990071e-05,
      "loss": 383.8247,
      "step": 14740
    },
    {
      "epoch": 0.2836783952457424,
      "grad_norm": 497.50634405561897,
      "learning_rate": 1.645989725316506e-05,
      "loss": 384.7214,
      "step": 14750
    },
    {
      "epoch": 0.28387071958150223,
      "grad_norm": 471.5952807884002,
      "learning_rate": 1.6455237240224364e-05,
      "loss": 379.8508,
      "step": 14760
    },
    {
      "epoch": 0.28406304391726206,
      "grad_norm": 497.51785648599713,
      "learning_rate": 1.6450574822814412e-05,
      "loss": 387.6232,
      "step": 14770
    },
    {
      "epoch": 0.2842553682530219,
      "grad_norm": 462.968041044798,
      "learning_rate": 1.6445910002671872e-05,
      "loss": 387.0854,
      "step": 14780
    },
    {
      "epoch": 0.2844476925887817,
      "grad_norm": 511.1203351303572,
      "learning_rate": 1.644124278153431e-05,
      "loss": 389.8439,
      "step": 14790
    },
    {
      "epoch": 0.28464001692454155,
      "grad_norm": 543.5307978614018,
      "learning_rate": 1.64365731611402e-05,
      "loss": 390.1835,
      "step": 14800
    },
    {
      "epoch": 0.2848323412603014,
      "grad_norm": 468.3616680586934,
      "learning_rate": 1.6431901143228888e-05,
      "loss": 378.7895,
      "step": 14810
    },
    {
      "epoch": 0.2850246655960612,
      "grad_norm": 488.52103847851134,
      "learning_rate": 1.6427226729540623e-05,
      "loss": 380.1911,
      "step": 14820
    },
    {
      "epoch": 0.28521698993182104,
      "grad_norm": 487.61978203376117,
      "learning_rate": 1.6422549921816556e-05,
      "loss": 393.6369,
      "step": 14830
    },
    {
      "epoch": 0.28540931426758087,
      "grad_norm": 472.1358548853999,
      "learning_rate": 1.641787072179871e-05,
      "loss": 381.1461,
      "step": 14840
    },
    {
      "epoch": 0.28560163860334065,
      "grad_norm": 520.1675409127897,
      "learning_rate": 1.6413189131230022e-05,
      "loss": 380.0909,
      "step": 14850
    },
    {
      "epoch": 0.2857939629391005,
      "grad_norm": 485.978698749814,
      "learning_rate": 1.6408505151854292e-05,
      "loss": 384.8645,
      "step": 14860
    },
    {
      "epoch": 0.2859862872748603,
      "grad_norm": 495.28185861111274,
      "learning_rate": 1.6403818785416236e-05,
      "loss": 380.8901,
      "step": 14870
    },
    {
      "epoch": 0.28617861161062014,
      "grad_norm": 458.01453173995077,
      "learning_rate": 1.6399130033661444e-05,
      "loss": 397.4296,
      "step": 14880
    },
    {
      "epoch": 0.28637093594637997,
      "grad_norm": 476.64320326477275,
      "learning_rate": 1.6394438898336402e-05,
      "loss": 385.6506,
      "step": 14890
    },
    {
      "epoch": 0.2865632602821398,
      "grad_norm": 541.8504396274184,
      "learning_rate": 1.6389745381188475e-05,
      "loss": 396.2868,
      "step": 14900
    },
    {
      "epoch": 0.28675558461789963,
      "grad_norm": 476.24906697976616,
      "learning_rate": 1.6385049483965926e-05,
      "loss": 380.1042,
      "step": 14910
    },
    {
      "epoch": 0.28694790895365946,
      "grad_norm": 512.7942479475595,
      "learning_rate": 1.6380351208417897e-05,
      "loss": 385.5607,
      "step": 14920
    },
    {
      "epoch": 0.2871402332894193,
      "grad_norm": 471.00853555682755,
      "learning_rate": 1.6375650556294417e-05,
      "loss": 382.3335,
      "step": 14930
    },
    {
      "epoch": 0.2873325576251791,
      "grad_norm": 499.52034733524675,
      "learning_rate": 1.6370947529346404e-05,
      "loss": 385.2976,
      "step": 14940
    },
    {
      "epoch": 0.28752488196093895,
      "grad_norm": 486.82498278013855,
      "learning_rate": 1.6366242129325652e-05,
      "loss": 386.699,
      "step": 14950
    },
    {
      "epoch": 0.2877172062966988,
      "grad_norm": 461.49359895748785,
      "learning_rate": 1.636153435798485e-05,
      "loss": 391.78,
      "step": 14960
    },
    {
      "epoch": 0.28790953063245855,
      "grad_norm": 493.8995605913665,
      "learning_rate": 1.6356824217077564e-05,
      "loss": 373.2573,
      "step": 14970
    },
    {
      "epoch": 0.2881018549682184,
      "grad_norm": 521.052487777547,
      "learning_rate": 1.6352111708358243e-05,
      "loss": 384.3046,
      "step": 14980
    },
    {
      "epoch": 0.2882941793039782,
      "grad_norm": 487.92242776300156,
      "learning_rate": 1.6347396833582224e-05,
      "loss": 389.7221,
      "step": 14990
    },
    {
      "epoch": 0.28848650363973805,
      "grad_norm": 463.67397745910336,
      "learning_rate": 1.634267959450571e-05,
      "loss": 380.6968,
      "step": 15000
    },
    {
      "epoch": 0.2886788279754979,
      "grad_norm": 481.26091086966454,
      "learning_rate": 1.63379599928858e-05,
      "loss": 396.7292,
      "step": 15010
    },
    {
      "epoch": 0.2888711523112577,
      "grad_norm": 434.5124362503905,
      "learning_rate": 1.6333238030480473e-05,
      "loss": 387.3863,
      "step": 15020
    },
    {
      "epoch": 0.28906347664701754,
      "grad_norm": 491.90650335544296,
      "learning_rate": 1.6328513709048573e-05,
      "loss": 378.2793,
      "step": 15030
    },
    {
      "epoch": 0.28925580098277737,
      "grad_norm": 546.9311713653154,
      "learning_rate": 1.6323787030349833e-05,
      "loss": 384.7709,
      "step": 15040
    },
    {
      "epoch": 0.2894481253185372,
      "grad_norm": 530.2226410803686,
      "learning_rate": 1.6319057996144868e-05,
      "loss": 389.6169,
      "step": 15050
    },
    {
      "epoch": 0.289640449654297,
      "grad_norm": 495.49817364137556,
      "learning_rate": 1.631432660819516e-05,
      "loss": 376.3631,
      "step": 15060
    },
    {
      "epoch": 0.28983277399005686,
      "grad_norm": 517.459453193237,
      "learning_rate": 1.6309592868263075e-05,
      "loss": 386.113,
      "step": 15070
    },
    {
      "epoch": 0.29002509832581663,
      "grad_norm": 520.8782391961886,
      "learning_rate": 1.630485677811185e-05,
      "loss": 390.316,
      "step": 15080
    },
    {
      "epoch": 0.29021742266157646,
      "grad_norm": 494.6786563660572,
      "learning_rate": 1.630011833950561e-05,
      "loss": 385.1296,
      "step": 15090
    },
    {
      "epoch": 0.2904097469973363,
      "grad_norm": 496.32032175509653,
      "learning_rate": 1.6295377554209338e-05,
      "loss": 390.7008,
      "step": 15100
    },
    {
      "epoch": 0.2906020713330961,
      "grad_norm": 524.6278616686512,
      "learning_rate": 1.6290634423988897e-05,
      "loss": 388.9809,
      "step": 15110
    },
    {
      "epoch": 0.29079439566885595,
      "grad_norm": 490.09439006565714,
      "learning_rate": 1.6285888950611023e-05,
      "loss": 391.4503,
      "step": 15120
    },
    {
      "epoch": 0.2909867200046158,
      "grad_norm": 486.966021625674,
      "learning_rate": 1.6281141135843334e-05,
      "loss": 383.3823,
      "step": 15130
    },
    {
      "epoch": 0.2911790443403756,
      "grad_norm": 488.781315739911,
      "learning_rate": 1.6276390981454306e-05,
      "loss": 386.7401,
      "step": 15140
    },
    {
      "epoch": 0.29137136867613544,
      "grad_norm": 503.57567011536383,
      "learning_rate": 1.6271638489213297e-05,
      "loss": 397.332,
      "step": 15150
    },
    {
      "epoch": 0.2915636930118953,
      "grad_norm": 673.8851475203368,
      "learning_rate": 1.6266883660890527e-05,
      "loss": 398.2306,
      "step": 15160
    },
    {
      "epoch": 0.2917560173476551,
      "grad_norm": 496.3247441244141,
      "learning_rate": 1.6262126498257098e-05,
      "loss": 384.9945,
      "step": 15170
    },
    {
      "epoch": 0.29194834168341494,
      "grad_norm": 476.35917625149597,
      "learning_rate": 1.625736700308497e-05,
      "loss": 387.6532,
      "step": 15180
    },
    {
      "epoch": 0.2921406660191747,
      "grad_norm": 473.77252573428257,
      "learning_rate": 1.6252605177146978e-05,
      "loss": 383.0787,
      "step": 15190
    },
    {
      "epoch": 0.29233299035493454,
      "grad_norm": 481.34409260225357,
      "learning_rate": 1.624784102221682e-05,
      "loss": 392.2435,
      "step": 15200
    },
    {
      "epoch": 0.29252531469069437,
      "grad_norm": 475.5771470039311,
      "learning_rate": 1.6243074540069067e-05,
      "loss": 382.1738,
      "step": 15210
    },
    {
      "epoch": 0.2927176390264542,
      "grad_norm": 517.829452811324,
      "learning_rate": 1.6238305732479158e-05,
      "loss": 392.2777,
      "step": 15220
    },
    {
      "epoch": 0.29290996336221403,
      "grad_norm": 526.4621373465495,
      "learning_rate": 1.6233534601223396e-05,
      "loss": 387.8406,
      "step": 15230
    },
    {
      "epoch": 0.29310228769797386,
      "grad_norm": 478.25607538973367,
      "learning_rate": 1.6228761148078943e-05,
      "loss": 391.3679,
      "step": 15240
    },
    {
      "epoch": 0.2932946120337337,
      "grad_norm": 475.47471105435926,
      "learning_rate": 1.622398537482383e-05,
      "loss": 393.5634,
      "step": 15250
    },
    {
      "epoch": 0.2934869363694935,
      "grad_norm": 560.4740962824944,
      "learning_rate": 1.621920728323696e-05,
      "loss": 392.568,
      "step": 15260
    },
    {
      "epoch": 0.29367926070525335,
      "grad_norm": 494.87613487730755,
      "learning_rate": 1.621442687509809e-05,
      "loss": 385.4871,
      "step": 15270
    },
    {
      "epoch": 0.2938715850410132,
      "grad_norm": 512.9864101092595,
      "learning_rate": 1.6209644152187848e-05,
      "loss": 392.8923,
      "step": 15280
    },
    {
      "epoch": 0.294063909376773,
      "grad_norm": 486.19639377330276,
      "learning_rate": 1.620485911628771e-05,
      "loss": 394.0132,
      "step": 15290
    },
    {
      "epoch": 0.2942562337125328,
      "grad_norm": 470.3526868614509,
      "learning_rate": 1.6200071769180026e-05,
      "loss": 368.9018,
      "step": 15300
    },
    {
      "epoch": 0.2944485580482926,
      "grad_norm": 445.00269907400354,
      "learning_rate": 1.6195282112648007e-05,
      "loss": 386.3872,
      "step": 15310
    },
    {
      "epoch": 0.29464088238405245,
      "grad_norm": 506.9645941907079,
      "learning_rate": 1.6190490148475724e-05,
      "loss": 381.0153,
      "step": 15320
    },
    {
      "epoch": 0.2948332067198123,
      "grad_norm": 455.66778163181453,
      "learning_rate": 1.6185695878448094e-05,
      "loss": 389.7398,
      "step": 15330
    },
    {
      "epoch": 0.2950255310555721,
      "grad_norm": 434.4250529789861,
      "learning_rate": 1.6180899304350915e-05,
      "loss": 372.087,
      "step": 15340
    },
    {
      "epoch": 0.29521785539133194,
      "grad_norm": 500.35196090291123,
      "learning_rate": 1.6176100427970826e-05,
      "loss": 380.515,
      "step": 15350
    },
    {
      "epoch": 0.29541017972709177,
      "grad_norm": 484.0534434395125,
      "learning_rate": 1.6171299251095324e-05,
      "loss": 384.4088,
      "step": 15360
    },
    {
      "epoch": 0.2956025040628516,
      "grad_norm": 496.5740929991856,
      "learning_rate": 1.6166495775512777e-05,
      "loss": 396.1473,
      "step": 15370
    },
    {
      "epoch": 0.29579482839861143,
      "grad_norm": 517.0120237674353,
      "learning_rate": 1.6161690003012392e-05,
      "loss": 390.5824,
      "step": 15380
    },
    {
      "epoch": 0.29598715273437126,
      "grad_norm": 494.9591564232326,
      "learning_rate": 1.615688193538425e-05,
      "loss": 378.619,
      "step": 15390
    },
    {
      "epoch": 0.2961794770701311,
      "grad_norm": 571.1103475973266,
      "learning_rate": 1.615207157441927e-05,
      "loss": 389.4476,
      "step": 15400
    },
    {
      "epoch": 0.29637180140589087,
      "grad_norm": 474.7299271462728,
      "learning_rate": 1.6147258921909236e-05,
      "loss": 380.0867,
      "step": 15410
    },
    {
      "epoch": 0.2965641257416507,
      "grad_norm": 560.9972230006571,
      "learning_rate": 1.6142443979646774e-05,
      "loss": 398.957,
      "step": 15420
    },
    {
      "epoch": 0.2967564500774105,
      "grad_norm": 448.23823648255143,
      "learning_rate": 1.6137626749425377e-05,
      "loss": 377.3599,
      "step": 15430
    },
    {
      "epoch": 0.29694877441317036,
      "grad_norm": 479.3145501516596,
      "learning_rate": 1.6132807233039382e-05,
      "loss": 386.1266,
      "step": 15440
    },
    {
      "epoch": 0.2971410987489302,
      "grad_norm": 528.0677323746904,
      "learning_rate": 1.612798543228398e-05,
      "loss": 383.0138,
      "step": 15450
    },
    {
      "epoch": 0.29733342308469,
      "grad_norm": 490.62521823390904,
      "learning_rate": 1.612316134895521e-05,
      "loss": 387.7146,
      "step": 15460
    },
    {
      "epoch": 0.29752574742044985,
      "grad_norm": 455.12196157480594,
      "learning_rate": 1.611833498484997e-05,
      "loss": 387.1293,
      "step": 15470
    },
    {
      "epoch": 0.2977180717562097,
      "grad_norm": 540.2640954521449,
      "learning_rate": 1.611350634176599e-05,
      "loss": 389.78,
      "step": 15480
    },
    {
      "epoch": 0.2979103960919695,
      "grad_norm": 500.9857814104484,
      "learning_rate": 1.6108675421501865e-05,
      "loss": 391.0004,
      "step": 15490
    },
    {
      "epoch": 0.29810272042772934,
      "grad_norm": 491.1924764973054,
      "learning_rate": 1.610384222585704e-05,
      "loss": 388.4165,
      "step": 15500
    },
    {
      "epoch": 0.29829504476348917,
      "grad_norm": 473.6500171513281,
      "learning_rate": 1.609900675663179e-05,
      "loss": 376.6896,
      "step": 15510
    },
    {
      "epoch": 0.298487369099249,
      "grad_norm": 477.5587209017475,
      "learning_rate": 1.609416901562725e-05,
      "loss": 371.4006,
      "step": 15520
    },
    {
      "epoch": 0.2986796934350088,
      "grad_norm": 507.91221115477657,
      "learning_rate": 1.60893290046454e-05,
      "loss": 384.6357,
      "step": 15530
    },
    {
      "epoch": 0.2988720177707686,
      "grad_norm": 493.17435750231357,
      "learning_rate": 1.608448672548907e-05,
      "loss": 389.1049,
      "step": 15540
    },
    {
      "epoch": 0.29906434210652844,
      "grad_norm": 530.1929565826192,
      "learning_rate": 1.6079642179961917e-05,
      "loss": 372.4196,
      "step": 15550
    },
    {
      "epoch": 0.29925666644228827,
      "grad_norm": 513.9989479149148,
      "learning_rate": 1.6074795369868463e-05,
      "loss": 388.418,
      "step": 15560
    },
    {
      "epoch": 0.2994489907780481,
      "grad_norm": 469.28213013333186,
      "learning_rate": 1.6069946297014064e-05,
      "loss": 382.0257,
      "step": 15570
    },
    {
      "epoch": 0.2996413151138079,
      "grad_norm": 519.1702966244364,
      "learning_rate": 1.6065094963204915e-05,
      "loss": 380.3084,
      "step": 15580
    },
    {
      "epoch": 0.29983363944956776,
      "grad_norm": 458.70827536114274,
      "learning_rate": 1.6060241370248064e-05,
      "loss": 386.5981,
      "step": 15590
    },
    {
      "epoch": 0.3000259637853276,
      "grad_norm": 494.09100236320444,
      "learning_rate": 1.605538551995139e-05,
      "loss": 389.0104,
      "step": 15600
    },
    {
      "epoch": 0.3002182881210874,
      "grad_norm": 499.8257326714157,
      "learning_rate": 1.6050527414123616e-05,
      "loss": 381.525,
      "step": 15610
    },
    {
      "epoch": 0.30041061245684725,
      "grad_norm": 514.2821900561872,
      "learning_rate": 1.604566705457431e-05,
      "loss": 370.4945,
      "step": 15620
    },
    {
      "epoch": 0.3006029367926071,
      "grad_norm": 495.99600409824876,
      "learning_rate": 1.604080444311387e-05,
      "loss": 376.4167,
      "step": 15630
    },
    {
      "epoch": 0.30079526112836685,
      "grad_norm": 496.12042045174064,
      "learning_rate": 1.6035939581553543e-05,
      "loss": 389.8231,
      "step": 15640
    },
    {
      "epoch": 0.3009875854641267,
      "grad_norm": 500.4805466121026,
      "learning_rate": 1.603107247170541e-05,
      "loss": 394.658,
      "step": 15650
    },
    {
      "epoch": 0.3011799097998865,
      "grad_norm": 488.2017398651639,
      "learning_rate": 1.6026203115382392e-05,
      "loss": 394.7382,
      "step": 15660
    },
    {
      "epoch": 0.30137223413564634,
      "grad_norm": 505.8285315529847,
      "learning_rate": 1.6021331514398233e-05,
      "loss": 376.4537,
      "step": 15670
    },
    {
      "epoch": 0.3015645584714062,
      "grad_norm": 538.4909365883506,
      "learning_rate": 1.6016457670567535e-05,
      "loss": 396.45,
      "step": 15680
    },
    {
      "epoch": 0.301756882807166,
      "grad_norm": 520.0627945628461,
      "learning_rate": 1.6011581585705715e-05,
      "loss": 384.5404,
      "step": 15690
    },
    {
      "epoch": 0.30194920714292584,
      "grad_norm": 533.1833578871967,
      "learning_rate": 1.600670326162904e-05,
      "loss": 391.3917,
      "step": 15700
    },
    {
      "epoch": 0.30214153147868567,
      "grad_norm": 496.59726655207214,
      "learning_rate": 1.600182270015461e-05,
      "loss": 376.9667,
      "step": 15710
    },
    {
      "epoch": 0.3023338558144455,
      "grad_norm": 470.33865795550236,
      "learning_rate": 1.5996939903100338e-05,
      "loss": 373.5374,
      "step": 15720
    },
    {
      "epoch": 0.3025261801502053,
      "grad_norm": 494.7224862314734,
      "learning_rate": 1.5992054872285005e-05,
      "loss": 377.8496,
      "step": 15730
    },
    {
      "epoch": 0.30271850448596516,
      "grad_norm": 462.42162046846744,
      "learning_rate": 1.5987167609528187e-05,
      "loss": 386.7111,
      "step": 15740
    },
    {
      "epoch": 0.30291082882172493,
      "grad_norm": 549.5813713322203,
      "learning_rate": 1.598227811665032e-05,
      "loss": 394.5676,
      "step": 15750
    },
    {
      "epoch": 0.30310315315748476,
      "grad_norm": 479.453865277826,
      "learning_rate": 1.597738639547265e-05,
      "loss": 372.977,
      "step": 15760
    },
    {
      "epoch": 0.3032954774932446,
      "grad_norm": 437.4432263127666,
      "learning_rate": 1.597249244781727e-05,
      "loss": 387.8498,
      "step": 15770
    },
    {
      "epoch": 0.3034878018290044,
      "grad_norm": 480.8793637734585,
      "learning_rate": 1.5967596275507094e-05,
      "loss": 391.69,
      "step": 15780
    },
    {
      "epoch": 0.30368012616476425,
      "grad_norm": 480.18538183593864,
      "learning_rate": 1.5962697880365863e-05,
      "loss": 374.785,
      "step": 15790
    },
    {
      "epoch": 0.3038724505005241,
      "grad_norm": 485.7658136927617,
      "learning_rate": 1.5957797264218145e-05,
      "loss": 372.742,
      "step": 15800
    },
    {
      "epoch": 0.3040647748362839,
      "grad_norm": 526.1846228185616,
      "learning_rate": 1.5952894428889347e-05,
      "loss": 402.5581,
      "step": 15810
    },
    {
      "epoch": 0.30425709917204374,
      "grad_norm": 474.2349139837495,
      "learning_rate": 1.594798937620569e-05,
      "loss": 374.9354,
      "step": 15820
    },
    {
      "epoch": 0.3044494235078036,
      "grad_norm": 487.1475350282878,
      "learning_rate": 1.594308210799422e-05,
      "loss": 381.5031,
      "step": 15830
    },
    {
      "epoch": 0.3046417478435634,
      "grad_norm": 526.0957708674093,
      "learning_rate": 1.5938172626082823e-05,
      "loss": 389.1943,
      "step": 15840
    },
    {
      "epoch": 0.30483407217932323,
      "grad_norm": 484.2232035646498,
      "learning_rate": 1.5933260932300192e-05,
      "loss": 382.6517,
      "step": 15850
    },
    {
      "epoch": 0.305026396515083,
      "grad_norm": 481.9646853442586,
      "learning_rate": 1.5928347028475855e-05,
      "loss": 385.7438,
      "step": 15860
    },
    {
      "epoch": 0.30521872085084284,
      "grad_norm": 506.3687257691799,
      "learning_rate": 1.592343091644016e-05,
      "loss": 379.9382,
      "step": 15870
    },
    {
      "epoch": 0.30541104518660267,
      "grad_norm": 550.512446852848,
      "learning_rate": 1.5918512598024275e-05,
      "loss": 368.3928,
      "step": 15880
    },
    {
      "epoch": 0.3056033695223625,
      "grad_norm": 516.7421443143606,
      "learning_rate": 1.59135920750602e-05,
      "loss": 394.3213,
      "step": 15890
    },
    {
      "epoch": 0.30579569385812233,
      "grad_norm": 479.36253297341716,
      "learning_rate": 1.590866934938074e-05,
      "loss": 386.138,
      "step": 15900
    },
    {
      "epoch": 0.30598801819388216,
      "grad_norm": 457.4125480024917,
      "learning_rate": 1.590374442281953e-05,
      "loss": 375.0515,
      "step": 15910
    },
    {
      "epoch": 0.306180342529642,
      "grad_norm": 507.5572160106832,
      "learning_rate": 1.5898817297211028e-05,
      "loss": 377.9083,
      "step": 15920
    },
    {
      "epoch": 0.3063726668654018,
      "grad_norm": 513.6180166698787,
      "learning_rate": 1.589388797439051e-05,
      "loss": 387.941,
      "step": 15930
    },
    {
      "epoch": 0.30656499120116165,
      "grad_norm": 540.7423722592335,
      "learning_rate": 1.5888956456194056e-05,
      "loss": 379.9804,
      "step": 15940
    },
    {
      "epoch": 0.3067573155369215,
      "grad_norm": 494.6165946657413,
      "learning_rate": 1.588402274445858e-05,
      "loss": 397.443,
      "step": 15950
    },
    {
      "epoch": 0.3069496398726813,
      "grad_norm": 485.2879526777606,
      "learning_rate": 1.5879086841021815e-05,
      "loss": 387.4972,
      "step": 15960
    },
    {
      "epoch": 0.30714196420844114,
      "grad_norm": 483.77146964448315,
      "learning_rate": 1.5874148747722294e-05,
      "loss": 385.6518,
      "step": 15970
    },
    {
      "epoch": 0.3073342885442009,
      "grad_norm": 500.09250591109884,
      "learning_rate": 1.5869208466399382e-05,
      "loss": 381.8277,
      "step": 15980
    },
    {
      "epoch": 0.30752661287996075,
      "grad_norm": 519.7212693848686,
      "learning_rate": 1.586426599889325e-05,
      "loss": 388.2954,
      "step": 15990
    },
    {
      "epoch": 0.3077189372157206,
      "grad_norm": 519.4154904155673,
      "learning_rate": 1.5859321347044882e-05,
      "loss": 393.8029,
      "step": 16000
    },
    {
      "epoch": 0.3079112615514804,
      "grad_norm": 475.0755911633883,
      "learning_rate": 1.5854374512696084e-05,
      "loss": 388.3107,
      "step": 16010
    },
    {
      "epoch": 0.30810358588724024,
      "grad_norm": 515.056518059897,
      "learning_rate": 1.584942549768947e-05,
      "loss": 383.1705,
      "step": 16020
    },
    {
      "epoch": 0.30829591022300007,
      "grad_norm": 489.79656202666257,
      "learning_rate": 1.584447430386846e-05,
      "loss": 395.4364,
      "step": 16030
    },
    {
      "epoch": 0.3084882345587599,
      "grad_norm": 471.0922792299047,
      "learning_rate": 1.58395209330773e-05,
      "loss": 368.5746,
      "step": 16040
    },
    {
      "epoch": 0.30868055889451973,
      "grad_norm": 507.9029922206225,
      "learning_rate": 1.5834565387161034e-05,
      "loss": 389.2412,
      "step": 16050
    },
    {
      "epoch": 0.30887288323027956,
      "grad_norm": 505.36307091146045,
      "learning_rate": 1.5829607667965524e-05,
      "loss": 379.7645,
      "step": 16060
    },
    {
      "epoch": 0.3090652075660394,
      "grad_norm": 533.0401365312647,
      "learning_rate": 1.5824647777337433e-05,
      "loss": 402.0754,
      "step": 16070
    },
    {
      "epoch": 0.3092575319017992,
      "grad_norm": 477.822109995146,
      "learning_rate": 1.5819685717124245e-05,
      "loss": 378.4888,
      "step": 16080
    },
    {
      "epoch": 0.309449856237559,
      "grad_norm": 491.97574484715693,
      "learning_rate": 1.5814721489174246e-05,
      "loss": 378.9121,
      "step": 16090
    },
    {
      "epoch": 0.3096421805733188,
      "grad_norm": 504.15390175929986,
      "learning_rate": 1.580975509533652e-05,
      "loss": 386.5091,
      "step": 16100
    },
    {
      "epoch": 0.30983450490907866,
      "grad_norm": 472.79917790946797,
      "learning_rate": 1.5804786537460972e-05,
      "loss": 374.8191,
      "step": 16110
    },
    {
      "epoch": 0.3100268292448385,
      "grad_norm": 527.9536707655863,
      "learning_rate": 1.5799815817398312e-05,
      "loss": 380.7867,
      "step": 16120
    },
    {
      "epoch": 0.3102191535805983,
      "grad_norm": 540.2626487985623,
      "learning_rate": 1.579484293700004e-05,
      "loss": 385.7923,
      "step": 16130
    },
    {
      "epoch": 0.31041147791635815,
      "grad_norm": 487.14746385446807,
      "learning_rate": 1.578986789811849e-05,
      "loss": 393.2775,
      "step": 16140
    },
    {
      "epoch": 0.310603802252118,
      "grad_norm": 442.3817324977554,
      "learning_rate": 1.5784890702606763e-05,
      "loss": 376.5007,
      "step": 16150
    },
    {
      "epoch": 0.3107961265878778,
      "grad_norm": 530.164851806514,
      "learning_rate": 1.5779911352318792e-05,
      "loss": 381.8134,
      "step": 16160
    },
    {
      "epoch": 0.31098845092363764,
      "grad_norm": 562.094692156776,
      "learning_rate": 1.5774929849109303e-05,
      "loss": 380.7581,
      "step": 16170
    },
    {
      "epoch": 0.31118077525939747,
      "grad_norm": 472.9884759106849,
      "learning_rate": 1.5769946194833816e-05,
      "loss": 395.6762,
      "step": 16180
    },
    {
      "epoch": 0.3113730995951573,
      "grad_norm": 653.2759011964929,
      "learning_rate": 1.5764960391348666e-05,
      "loss": 391.3308,
      "step": 16190
    },
    {
      "epoch": 0.3115654239309171,
      "grad_norm": 453.7480143451823,
      "learning_rate": 1.5759972440510985e-05,
      "loss": 370.0301,
      "step": 16200
    },
    {
      "epoch": 0.3117577482666769,
      "grad_norm": 462.127878654082,
      "learning_rate": 1.5754982344178697e-05,
      "loss": 372.8345,
      "step": 16210
    },
    {
      "epoch": 0.31195007260243673,
      "grad_norm": 458.6353890517486,
      "learning_rate": 1.5749990104210534e-05,
      "loss": 385.0438,
      "step": 16220
    },
    {
      "epoch": 0.31214239693819656,
      "grad_norm": 505.3959018430225,
      "learning_rate": 1.574499572246602e-05,
      "loss": 377.8,
      "step": 16230
    },
    {
      "epoch": 0.3123347212739564,
      "grad_norm": 530.6902398817294,
      "learning_rate": 1.5739999200805483e-05,
      "loss": 396.8385,
      "step": 16240
    },
    {
      "epoch": 0.3125270456097162,
      "grad_norm": 469.0159835432282,
      "learning_rate": 1.573500054109004e-05,
      "loss": 382.3637,
      "step": 16250
    },
    {
      "epoch": 0.31271936994547606,
      "grad_norm": 468.0069546163065,
      "learning_rate": 1.5729999745181617e-05,
      "loss": 380.0171,
      "step": 16260
    },
    {
      "epoch": 0.3129116942812359,
      "grad_norm": 477.51891179615933,
      "learning_rate": 1.572499681494292e-05,
      "loss": 379.426,
      "step": 16270
    },
    {
      "epoch": 0.3131040186169957,
      "grad_norm": 480.5057157160723,
      "learning_rate": 1.571999175223746e-05,
      "loss": 376.0035,
      "step": 16280
    },
    {
      "epoch": 0.31329634295275555,
      "grad_norm": 499.78679071245256,
      "learning_rate": 1.571498455892954e-05,
      "loss": 387.6434,
      "step": 16290
    },
    {
      "epoch": 0.3134886672885154,
      "grad_norm": 476.4294157500738,
      "learning_rate": 1.570997523688426e-05,
      "loss": 388.008,
      "step": 16300
    },
    {
      "epoch": 0.31368099162427515,
      "grad_norm": 447.5674210996581,
      "learning_rate": 1.570496378796751e-05,
      "loss": 377.4202,
      "step": 16310
    },
    {
      "epoch": 0.313873315960035,
      "grad_norm": 467.2936933672317,
      "learning_rate": 1.5699950214045966e-05,
      "loss": 390.1317,
      "step": 16320
    },
    {
      "epoch": 0.3140656402957948,
      "grad_norm": 592.210511601536,
      "learning_rate": 1.5694934516987102e-05,
      "loss": 387.3247,
      "step": 16330
    },
    {
      "epoch": 0.31425796463155464,
      "grad_norm": 509.6399695265995,
      "learning_rate": 1.5689916698659193e-05,
      "loss": 391.6255,
      "step": 16340
    },
    {
      "epoch": 0.3144502889673145,
      "grad_norm": 475.9107510277038,
      "learning_rate": 1.568489676093128e-05,
      "loss": 370.1673,
      "step": 16350
    },
    {
      "epoch": 0.3146426133030743,
      "grad_norm": 479.4800658769051,
      "learning_rate": 1.5679874705673215e-05,
      "loss": 387.8395,
      "step": 16360
    },
    {
      "epoch": 0.31483493763883413,
      "grad_norm": 471.42216324652395,
      "learning_rate": 1.5674850534755628e-05,
      "loss": 387.203,
      "step": 16370
    },
    {
      "epoch": 0.31502726197459396,
      "grad_norm": 469.9233379455086,
      "learning_rate": 1.566982425004994e-05,
      "loss": 384.5128,
      "step": 16380
    },
    {
      "epoch": 0.3152195863103538,
      "grad_norm": 477.30929266097127,
      "learning_rate": 1.5664795853428357e-05,
      "loss": 380.8335,
      "step": 16390
    },
    {
      "epoch": 0.3154119106461136,
      "grad_norm": 450.40161426989624,
      "learning_rate": 1.565976534676388e-05,
      "loss": 375.3493,
      "step": 16400
    },
    {
      "epoch": 0.31560423498187345,
      "grad_norm": 508.10142288422304,
      "learning_rate": 1.5654732731930286e-05,
      "loss": 387.4695,
      "step": 16410
    },
    {
      "epoch": 0.31579655931763323,
      "grad_norm": 520.6726798846985,
      "learning_rate": 1.5649698010802138e-05,
      "loss": 388.4852,
      "step": 16420
    },
    {
      "epoch": 0.31598888365339306,
      "grad_norm": 505.42801805197837,
      "learning_rate": 1.564466118525479e-05,
      "loss": 376.6891,
      "step": 16430
    },
    {
      "epoch": 0.3161812079891529,
      "grad_norm": 477.95172533009156,
      "learning_rate": 1.5639622257164372e-05,
      "loss": 381.4712,
      "step": 16440
    },
    {
      "epoch": 0.3163735323249127,
      "grad_norm": 496.54564262131134,
      "learning_rate": 1.5634581228407807e-05,
      "loss": 369.9816,
      "step": 16450
    },
    {
      "epoch": 0.31656585666067255,
      "grad_norm": 486.65491609609694,
      "learning_rate": 1.562953810086279e-05,
      "loss": 380.1699,
      "step": 16460
    },
    {
      "epoch": 0.3167581809964324,
      "grad_norm": 508.2947239889114,
      "learning_rate": 1.562449287640781e-05,
      "loss": 374.3872,
      "step": 16470
    },
    {
      "epoch": 0.3169505053321922,
      "grad_norm": 475.81346313671924,
      "learning_rate": 1.5619445556922118e-05,
      "loss": 382.2745,
      "step": 16480
    },
    {
      "epoch": 0.31714282966795204,
      "grad_norm": 470.4893587917357,
      "learning_rate": 1.561439614428577e-05,
      "loss": 385.8443,
      "step": 16490
    },
    {
      "epoch": 0.3173351540037119,
      "grad_norm": 485.9223921477143,
      "learning_rate": 1.5609344640379585e-05,
      "loss": 384.2544,
      "step": 16500
    },
    {
      "epoch": 0.3175274783394717,
      "grad_norm": 521.2054049523408,
      "learning_rate": 1.560429104708516e-05,
      "loss": 373.9025,
      "step": 16510
    },
    {
      "epoch": 0.31771980267523153,
      "grad_norm": 485.2264953291109,
      "learning_rate": 1.5599235366284874e-05,
      "loss": 376.6931,
      "step": 16520
    },
    {
      "epoch": 0.31791212701099136,
      "grad_norm": 480.9287832495446,
      "learning_rate": 1.5594177599861894e-05,
      "loss": 385.223,
      "step": 16530
    },
    {
      "epoch": 0.31810445134675114,
      "grad_norm": 489.4742675039022,
      "learning_rate": 1.5589117749700147e-05,
      "loss": 381.2006,
      "step": 16540
    },
    {
      "epoch": 0.31829677568251097,
      "grad_norm": 458.57385493405224,
      "learning_rate": 1.5584055817684346e-05,
      "loss": 372.0469,
      "step": 16550
    },
    {
      "epoch": 0.3184891000182708,
      "grad_norm": 517.4634174509583,
      "learning_rate": 1.5578991805699975e-05,
      "loss": 382.9156,
      "step": 16560
    },
    {
      "epoch": 0.31868142435403063,
      "grad_norm": 496.50548891733985,
      "learning_rate": 1.5573925715633297e-05,
      "loss": 378.2142,
      "step": 16570
    },
    {
      "epoch": 0.31887374868979046,
      "grad_norm": 485.7748332980264,
      "learning_rate": 1.5568857549371348e-05,
      "loss": 385.6914,
      "step": 16580
    },
    {
      "epoch": 0.3190660730255503,
      "grad_norm": 514.8659900741009,
      "learning_rate": 1.5563787308801934e-05,
      "loss": 378.1995,
      "step": 16590
    },
    {
      "epoch": 0.3192583973613101,
      "grad_norm": 465.27092153134146,
      "learning_rate": 1.5558714995813636e-05,
      "loss": 378.0243,
      "step": 16600
    },
    {
      "epoch": 0.31945072169706995,
      "grad_norm": 480.4739220381383,
      "learning_rate": 1.5553640612295807e-05,
      "loss": 377.963,
      "step": 16610
    },
    {
      "epoch": 0.3196430460328298,
      "grad_norm": 468.7087140912121,
      "learning_rate": 1.5548564160138572e-05,
      "loss": 388.8835,
      "step": 16620
    },
    {
      "epoch": 0.3198353703685896,
      "grad_norm": 481.73616248602275,
      "learning_rate": 1.5543485641232825e-05,
      "loss": 374.0541,
      "step": 16630
    },
    {
      "epoch": 0.32002769470434944,
      "grad_norm": 489.5761439883091,
      "learning_rate": 1.553840505747023e-05,
      "loss": 371.5378,
      "step": 16640
    },
    {
      "epoch": 0.3202200190401092,
      "grad_norm": 452.1163008133125,
      "learning_rate": 1.5533322410743223e-05,
      "loss": 380.3839,
      "step": 16650
    },
    {
      "epoch": 0.32041234337586905,
      "grad_norm": 474.70436798658284,
      "learning_rate": 1.5528237702945e-05,
      "loss": 378.9041,
      "step": 16660
    },
    {
      "epoch": 0.3206046677116289,
      "grad_norm": 466.62285022413806,
      "learning_rate": 1.5523150935969534e-05,
      "loss": 373.3779,
      "step": 16670
    },
    {
      "epoch": 0.3207969920473887,
      "grad_norm": 474.8741849334241,
      "learning_rate": 1.5518062111711566e-05,
      "loss": 375.5115,
      "step": 16680
    },
    {
      "epoch": 0.32098931638314854,
      "grad_norm": 518.483571008647,
      "learning_rate": 1.5512971232066593e-05,
      "loss": 375.157,
      "step": 16690
    },
    {
      "epoch": 0.32118164071890837,
      "grad_norm": 484.72655755235917,
      "learning_rate": 1.5507878298930888e-05,
      "loss": 375.746,
      "step": 16700
    },
    {
      "epoch": 0.3213739650546682,
      "grad_norm": 498.67486744118185,
      "learning_rate": 1.5502783314201478e-05,
      "loss": 379.8609,
      "step": 16710
    },
    {
      "epoch": 0.32156628939042803,
      "grad_norm": 477.88687550637076,
      "learning_rate": 1.549768627977617e-05,
      "loss": 382.8419,
      "step": 16720
    },
    {
      "epoch": 0.32175861372618786,
      "grad_norm": 484.6069756423416,
      "learning_rate": 1.5492587197553517e-05,
      "loss": 388.9117,
      "step": 16730
    },
    {
      "epoch": 0.3219509380619477,
      "grad_norm": 505.736213411871,
      "learning_rate": 1.5487486069432848e-05,
      "loss": 378.6362,
      "step": 16740
    },
    {
      "epoch": 0.3221432623977075,
      "grad_norm": 497.96417750337025,
      "learning_rate": 1.5482382897314243e-05,
      "loss": 376.4035,
      "step": 16750
    },
    {
      "epoch": 0.3223355867334673,
      "grad_norm": 494.82135744094677,
      "learning_rate": 1.5477277683098555e-05,
      "loss": 392.0643,
      "step": 16760
    },
    {
      "epoch": 0.3225279110692271,
      "grad_norm": 484.427335221297,
      "learning_rate": 1.547217042868739e-05,
      "loss": 370.8638,
      "step": 16770
    },
    {
      "epoch": 0.32272023540498695,
      "grad_norm": 462.99282168935787,
      "learning_rate": 1.546706113598312e-05,
      "loss": 371.024,
      "step": 16780
    },
    {
      "epoch": 0.3229125597407468,
      "grad_norm": 491.05421379025603,
      "learning_rate": 1.5461949806888867e-05,
      "loss": 371.8982,
      "step": 16790
    },
    {
      "epoch": 0.3231048840765066,
      "grad_norm": 508.7786608188533,
      "learning_rate": 1.5456836443308512e-05,
      "loss": 379.871,
      "step": 16800
    },
    {
      "epoch": 0.32329720841226645,
      "grad_norm": 520.4386966025636,
      "learning_rate": 1.545172104714671e-05,
      "loss": 378.7275,
      "step": 16810
    },
    {
      "epoch": 0.3234895327480263,
      "grad_norm": 521.6918398886307,
      "learning_rate": 1.544660362030886e-05,
      "loss": 375.9082,
      "step": 16820
    },
    {
      "epoch": 0.3236818570837861,
      "grad_norm": 512.1836726299016,
      "learning_rate": 1.544148416470111e-05,
      "loss": 383.3218,
      "step": 16830
    },
    {
      "epoch": 0.32387418141954594,
      "grad_norm": 489.12754225502795,
      "learning_rate": 1.5436362682230378e-05,
      "loss": 386.0422,
      "step": 16840
    },
    {
      "epoch": 0.32406650575530577,
      "grad_norm": 466.65374940666953,
      "learning_rate": 1.543123917480433e-05,
      "loss": 381.8087,
      "step": 16850
    },
    {
      "epoch": 0.3242588300910656,
      "grad_norm": 461.40002697156746,
      "learning_rate": 1.542611364433139e-05,
      "loss": 385.4735,
      "step": 16860
    },
    {
      "epoch": 0.32445115442682537,
      "grad_norm": 459.4202788763098,
      "learning_rate": 1.5420986092720735e-05,
      "loss": 371.1076,
      "step": 16870
    },
    {
      "epoch": 0.3246434787625852,
      "grad_norm": 449.98582200463227,
      "learning_rate": 1.541585652188229e-05,
      "loss": 373.7883,
      "step": 16880
    },
    {
      "epoch": 0.32483580309834503,
      "grad_norm": 457.0824188517946,
      "learning_rate": 1.5410724933726732e-05,
      "loss": 385.3052,
      "step": 16890
    },
    {
      "epoch": 0.32502812743410486,
      "grad_norm": 470.0828548991895,
      "learning_rate": 1.5405591330165503e-05,
      "loss": 367.4301,
      "step": 16900
    },
    {
      "epoch": 0.3252204517698647,
      "grad_norm": 494.3550963053494,
      "learning_rate": 1.5400455713110777e-05,
      "loss": 381.5691,
      "step": 16910
    },
    {
      "epoch": 0.3254127761056245,
      "grad_norm": 495.14418149445703,
      "learning_rate": 1.539531808447549e-05,
      "loss": 385.7987,
      "step": 16920
    },
    {
      "epoch": 0.32560510044138435,
      "grad_norm": 472.9642363089013,
      "learning_rate": 1.5390178446173325e-05,
      "loss": 383.8926,
      "step": 16930
    },
    {
      "epoch": 0.3257974247771442,
      "grad_norm": 517.9876336877029,
      "learning_rate": 1.538503680011871e-05,
      "loss": 376.7862,
      "step": 16940
    },
    {
      "epoch": 0.325989749112904,
      "grad_norm": 505.3528534634227,
      "learning_rate": 1.537989314822682e-05,
      "loss": 376.2068,
      "step": 16950
    },
    {
      "epoch": 0.32618207344866385,
      "grad_norm": 474.1088352158935,
      "learning_rate": 1.5374747492413587e-05,
      "loss": 381.843,
      "step": 16960
    },
    {
      "epoch": 0.3263743977844237,
      "grad_norm": 509.77581047387105,
      "learning_rate": 1.536959983459568e-05,
      "loss": 381.9839,
      "step": 16970
    },
    {
      "epoch": 0.32656672212018345,
      "grad_norm": 466.4720994420121,
      "learning_rate": 1.536445017669052e-05,
      "loss": 373.449,
      "step": 16980
    },
    {
      "epoch": 0.3267590464559433,
      "grad_norm": 499.65117401835397,
      "learning_rate": 1.535929852061626e-05,
      "loss": 370.986,
      "step": 16990
    },
    {
      "epoch": 0.3269513707917031,
      "grad_norm": 502.879039502589,
      "learning_rate": 1.5354144868291817e-05,
      "loss": 387.5408,
      "step": 17000
    },
    {
      "epoch": 0.32714369512746294,
      "grad_norm": 509.3116178854893,
      "learning_rate": 1.5348989221636835e-05,
      "loss": 375.3288,
      "step": 17010
    },
    {
      "epoch": 0.32733601946322277,
      "grad_norm": 495.2204101094012,
      "learning_rate": 1.5343831582571706e-05,
      "loss": 395.3084,
      "step": 17020
    },
    {
      "epoch": 0.3275283437989826,
      "grad_norm": 458.64683776833766,
      "learning_rate": 1.5338671953017576e-05,
      "loss": 373.3518,
      "step": 17030
    },
    {
      "epoch": 0.32772066813474243,
      "grad_norm": 446.723133129233,
      "learning_rate": 1.5333510334896308e-05,
      "loss": 375.0342,
      "step": 17040
    },
    {
      "epoch": 0.32791299247050226,
      "grad_norm": 71497.58265802731,
      "learning_rate": 1.532834673013053e-05,
      "loss": 393.9046,
      "step": 17050
    },
    {
      "epoch": 0.3281053168062621,
      "grad_norm": 465.9825861908541,
      "learning_rate": 1.5323181140643598e-05,
      "loss": 373.7633,
      "step": 17060
    },
    {
      "epoch": 0.3282976411420219,
      "grad_norm": 511.23713354141057,
      "learning_rate": 1.5318013568359603e-05,
      "loss": 375.508,
      "step": 17070
    },
    {
      "epoch": 0.32848996547778175,
      "grad_norm": 468.334192498827,
      "learning_rate": 1.531284401520338e-05,
      "loss": 371.5967,
      "step": 17080
    },
    {
      "epoch": 0.3286822898135416,
      "grad_norm": 467.54562368668144,
      "learning_rate": 1.530767248310051e-05,
      "loss": 377.1894,
      "step": 17090
    },
    {
      "epoch": 0.32887461414930136,
      "grad_norm": 460.4204135496014,
      "learning_rate": 1.53024989739773e-05,
      "loss": 377.0717,
      "step": 17100
    },
    {
      "epoch": 0.3290669384850612,
      "grad_norm": 482.75785645727285,
      "learning_rate": 1.5297323489760792e-05,
      "loss": 374.5172,
      "step": 17110
    },
    {
      "epoch": 0.329259262820821,
      "grad_norm": 459.16051444697376,
      "learning_rate": 1.5292146032378778e-05,
      "loss": 383.0635,
      "step": 17120
    },
    {
      "epoch": 0.32945158715658085,
      "grad_norm": 488.2350542751421,
      "learning_rate": 1.5286966603759767e-05,
      "loss": 378.1121,
      "step": 17130
    },
    {
      "epoch": 0.3296439114923407,
      "grad_norm": 529.2036957718933,
      "learning_rate": 1.5281785205833013e-05,
      "loss": 373.3984,
      "step": 17140
    },
    {
      "epoch": 0.3298362358281005,
      "grad_norm": 495.58232268012483,
      "learning_rate": 1.52766018405285e-05,
      "loss": 374.0483,
      "step": 17150
    },
    {
      "epoch": 0.33002856016386034,
      "grad_norm": 462.7853111302842,
      "learning_rate": 1.5271416509776948e-05,
      "loss": 371.5238,
      "step": 17160
    },
    {
      "epoch": 0.33022088449962017,
      "grad_norm": 455.4394390356612,
      "learning_rate": 1.5266229215509806e-05,
      "loss": 375.2193,
      "step": 17170
    },
    {
      "epoch": 0.33041320883538,
      "grad_norm": 546.3913112434052,
      "learning_rate": 1.5261039959659257e-05,
      "loss": 382.3061,
      "step": 17180
    },
    {
      "epoch": 0.33060553317113983,
      "grad_norm": 516.4477945404207,
      "learning_rate": 1.5255848744158214e-05,
      "loss": 361.9489,
      "step": 17190
    },
    {
      "epoch": 0.33079785750689966,
      "grad_norm": 449.4434195607374,
      "learning_rate": 1.5250655570940317e-05,
      "loss": 390.1684,
      "step": 17200
    },
    {
      "epoch": 0.33099018184265944,
      "grad_norm": 490.190130280452,
      "learning_rate": 1.5245460441939934e-05,
      "loss": 382.9938,
      "step": 17210
    },
    {
      "epoch": 0.33118250617841927,
      "grad_norm": 527.3113368564473,
      "learning_rate": 1.5240263359092167e-05,
      "loss": 380.4742,
      "step": 17220
    },
    {
      "epoch": 0.3313748305141791,
      "grad_norm": 470.0570538586981,
      "learning_rate": 1.5235064324332846e-05,
      "loss": 386.2793,
      "step": 17230
    },
    {
      "epoch": 0.3315671548499389,
      "grad_norm": 486.9989422305508,
      "learning_rate": 1.5229863339598528e-05,
      "loss": 371.9208,
      "step": 17240
    },
    {
      "epoch": 0.33175947918569876,
      "grad_norm": 508.8063668870747,
      "learning_rate": 1.5224660406826486e-05,
      "loss": 368.9817,
      "step": 17250
    },
    {
      "epoch": 0.3319518035214586,
      "grad_norm": 504.0257942047159,
      "learning_rate": 1.5219455527954732e-05,
      "loss": 388.4633,
      "step": 17260
    },
    {
      "epoch": 0.3321441278572184,
      "grad_norm": 472.301557222717,
      "learning_rate": 1.5214248704921995e-05,
      "loss": 373.2826,
      "step": 17270
    },
    {
      "epoch": 0.33233645219297825,
      "grad_norm": 517.4597406367698,
      "learning_rate": 1.5209039939667731e-05,
      "loss": 376.0201,
      "step": 17280
    },
    {
      "epoch": 0.3325287765287381,
      "grad_norm": 545.273822728609,
      "learning_rate": 1.5203829234132118e-05,
      "loss": 375.969,
      "step": 17290
    },
    {
      "epoch": 0.3327211008644979,
      "grad_norm": 562.1083999001257,
      "learning_rate": 1.5198616590256064e-05,
      "loss": 382.0835,
      "step": 17300
    },
    {
      "epoch": 0.33291342520025774,
      "grad_norm": 495.6276533090367,
      "learning_rate": 1.5193402009981187e-05,
      "loss": 361.7445,
      "step": 17310
    },
    {
      "epoch": 0.3331057495360175,
      "grad_norm": 476.47835534030764,
      "learning_rate": 1.5188185495249832e-05,
      "loss": 381.098,
      "step": 17320
    },
    {
      "epoch": 0.33329807387177735,
      "grad_norm": 444.11661287010764,
      "learning_rate": 1.5182967048005069e-05,
      "loss": 376.8885,
      "step": 17330
    },
    {
      "epoch": 0.3334903982075372,
      "grad_norm": 470.2516077821844,
      "learning_rate": 1.5177746670190674e-05,
      "loss": 370.4099,
      "step": 17340
    },
    {
      "epoch": 0.333682722543297,
      "grad_norm": 482.9856806878494,
      "learning_rate": 1.5172524363751162e-05,
      "loss": 369.4453,
      "step": 17350
    },
    {
      "epoch": 0.33387504687905684,
      "grad_norm": 532.1990923733988,
      "learning_rate": 1.5167300130631748e-05,
      "loss": 376.5783,
      "step": 17360
    },
    {
      "epoch": 0.33406737121481667,
      "grad_norm": 485.45210052577,
      "learning_rate": 1.516207397277838e-05,
      "loss": 373.4351,
      "step": 17370
    },
    {
      "epoch": 0.3342596955505765,
      "grad_norm": 469.7862255954336,
      "learning_rate": 1.5156845892137711e-05,
      "loss": 361.6263,
      "step": 17380
    },
    {
      "epoch": 0.3344520198863363,
      "grad_norm": 478.2530516753436,
      "learning_rate": 1.5151615890657113e-05,
      "loss": 380.667,
      "step": 17390
    },
    {
      "epoch": 0.33464434422209616,
      "grad_norm": 469.38863379886374,
      "learning_rate": 1.5146383970284679e-05,
      "loss": 375.3877,
      "step": 17400
    },
    {
      "epoch": 0.334836668557856,
      "grad_norm": 417.1455417074075,
      "learning_rate": 1.514115013296921e-05,
      "loss": 372.6206,
      "step": 17410
    },
    {
      "epoch": 0.3350289928936158,
      "grad_norm": 494.0037941883318,
      "learning_rate": 1.513591438066023e-05,
      "loss": 374.6683,
      "step": 17420
    },
    {
      "epoch": 0.3352213172293756,
      "grad_norm": 505.1937804210555,
      "learning_rate": 1.5130676715307962e-05,
      "loss": 391.8111,
      "step": 17430
    },
    {
      "epoch": 0.3354136415651354,
      "grad_norm": 480.34724741981717,
      "learning_rate": 1.5125437138863353e-05,
      "loss": 390.2885,
      "step": 17440
    },
    {
      "epoch": 0.33560596590089525,
      "grad_norm": 478.2841053518114,
      "learning_rate": 1.512019565327806e-05,
      "loss": 379.6284,
      "step": 17450
    },
    {
      "epoch": 0.3357982902366551,
      "grad_norm": 483.98494783902385,
      "learning_rate": 1.5114952260504448e-05,
      "loss": 388.126,
      "step": 17460
    },
    {
      "epoch": 0.3359906145724149,
      "grad_norm": 465.02478443190154,
      "learning_rate": 1.5109706962495596e-05,
      "loss": 378.6843,
      "step": 17470
    },
    {
      "epoch": 0.33618293890817474,
      "grad_norm": 498.44092870110927,
      "learning_rate": 1.510445976120529e-05,
      "loss": 373.266,
      "step": 17480
    },
    {
      "epoch": 0.3363752632439346,
      "grad_norm": 474.0369123843142,
      "learning_rate": 1.5099210658588029e-05,
      "loss": 376.8539,
      "step": 17490
    },
    {
      "epoch": 0.3365675875796944,
      "grad_norm": 486.4450465086325,
      "learning_rate": 1.5093959656599008e-05,
      "loss": 380.5545,
      "step": 17500
    },
    {
      "epoch": 0.33675991191545424,
      "grad_norm": 484.1369924739329,
      "learning_rate": 1.5088706757194147e-05,
      "loss": 368.5146,
      "step": 17510
    },
    {
      "epoch": 0.33695223625121407,
      "grad_norm": 475.1897734956671,
      "learning_rate": 1.5083451962330062e-05,
      "loss": 370.8672,
      "step": 17520
    },
    {
      "epoch": 0.3371445605869739,
      "grad_norm": 504.2752531643805,
      "learning_rate": 1.5078195273964081e-05,
      "loss": 369.6901,
      "step": 17530
    },
    {
      "epoch": 0.33733688492273367,
      "grad_norm": 446.3898974181883,
      "learning_rate": 1.5072936694054222e-05,
      "loss": 370.656,
      "step": 17540
    },
    {
      "epoch": 0.3375292092584935,
      "grad_norm": 487.0597556124582,
      "learning_rate": 1.5067676224559231e-05,
      "loss": 377.0279,
      "step": 17550
    },
    {
      "epoch": 0.33772153359425333,
      "grad_norm": 479.5423938975111,
      "learning_rate": 1.506241386743854e-05,
      "loss": 376.4824,
      "step": 17560
    },
    {
      "epoch": 0.33791385793001316,
      "grad_norm": 477.43146565089535,
      "learning_rate": 1.5057149624652297e-05,
      "loss": 363.8776,
      "step": 17570
    },
    {
      "epoch": 0.338106182265773,
      "grad_norm": 541.270034106954,
      "learning_rate": 1.5051883498161334e-05,
      "loss": 369.7278,
      "step": 17580
    },
    {
      "epoch": 0.3382985066015328,
      "grad_norm": 520.5892848187016,
      "learning_rate": 1.5046615489927206e-05,
      "loss": 369.1312,
      "step": 17590
    },
    {
      "epoch": 0.33849083093729265,
      "grad_norm": 475.4938536630045,
      "learning_rate": 1.5041345601912154e-05,
      "loss": 371.8933,
      "step": 17600
    },
    {
      "epoch": 0.3386831552730525,
      "grad_norm": 498.1856271685703,
      "learning_rate": 1.5036073836079128e-05,
      "loss": 379.6265,
      "step": 17610
    },
    {
      "epoch": 0.3388754796088123,
      "grad_norm": 487.4943707001513,
      "learning_rate": 1.5030800194391773e-05,
      "loss": 366.008,
      "step": 17620
    },
    {
      "epoch": 0.33906780394457214,
      "grad_norm": 489.1970871708965,
      "learning_rate": 1.5025524678814428e-05,
      "loss": 378.8491,
      "step": 17630
    },
    {
      "epoch": 0.339260128280332,
      "grad_norm": 461.82368666616,
      "learning_rate": 1.5020247291312138e-05,
      "loss": 378.9849,
      "step": 17640
    },
    {
      "epoch": 0.3394524526160918,
      "grad_norm": 489.880742186845,
      "learning_rate": 1.5014968033850647e-05,
      "loss": 392.6186,
      "step": 17650
    },
    {
      "epoch": 0.3396447769518516,
      "grad_norm": 587.6277073218204,
      "learning_rate": 1.5009686908396388e-05,
      "loss": 378.6439,
      "step": 17660
    },
    {
      "epoch": 0.3398371012876114,
      "grad_norm": 448.2616848898422,
      "learning_rate": 1.5004403916916494e-05,
      "loss": 377.4304,
      "step": 17670
    },
    {
      "epoch": 0.34002942562337124,
      "grad_norm": 460.38681861251325,
      "learning_rate": 1.4999119061378791e-05,
      "loss": 379.6803,
      "step": 17680
    },
    {
      "epoch": 0.34022174995913107,
      "grad_norm": 475.82737567797994,
      "learning_rate": 1.4993832343751802e-05,
      "loss": 390.8007,
      "step": 17690
    },
    {
      "epoch": 0.3404140742948909,
      "grad_norm": 450.60313886665847,
      "learning_rate": 1.4988543766004737e-05,
      "loss": 369.8139,
      "step": 17700
    },
    {
      "epoch": 0.34060639863065073,
      "grad_norm": 485.42499080953183,
      "learning_rate": 1.498325333010751e-05,
      "loss": 362.2229,
      "step": 17710
    },
    {
      "epoch": 0.34079872296641056,
      "grad_norm": 466.7144395419084,
      "learning_rate": 1.497796103803072e-05,
      "loss": 378.5834,
      "step": 17720
    },
    {
      "epoch": 0.3409910473021704,
      "grad_norm": 491.45758109377846,
      "learning_rate": 1.4972666891745655e-05,
      "loss": 376.6357,
      "step": 17730
    },
    {
      "epoch": 0.3411833716379302,
      "grad_norm": 479.1086596452101,
      "learning_rate": 1.49673708932243e-05,
      "loss": 361.8653,
      "step": 17740
    },
    {
      "epoch": 0.34137569597369005,
      "grad_norm": 469.8843122325225,
      "learning_rate": 1.4962073044439328e-05,
      "loss": 379.0124,
      "step": 17750
    },
    {
      "epoch": 0.3415680203094499,
      "grad_norm": 462.44513637880374,
      "learning_rate": 1.4956773347364095e-05,
      "loss": 380.2196,
      "step": 17760
    },
    {
      "epoch": 0.34176034464520966,
      "grad_norm": 518.7330123514714,
      "learning_rate": 1.4951471803972657e-05,
      "loss": 375.6815,
      "step": 17770
    },
    {
      "epoch": 0.3419526689809695,
      "grad_norm": 530.270547201222,
      "learning_rate": 1.4946168416239746e-05,
      "loss": 372.9629,
      "step": 17780
    },
    {
      "epoch": 0.3421449933167293,
      "grad_norm": 482.13253350069255,
      "learning_rate": 1.4940863186140788e-05,
      "loss": 385.9331,
      "step": 17790
    },
    {
      "epoch": 0.34233731765248915,
      "grad_norm": 483.2864946447858,
      "learning_rate": 1.4935556115651898e-05,
      "loss": 364.5144,
      "step": 17800
    },
    {
      "epoch": 0.342529641988249,
      "grad_norm": 461.10352016301266,
      "learning_rate": 1.4930247206749863e-05,
      "loss": 367.5356,
      "step": 17810
    },
    {
      "epoch": 0.3427219663240088,
      "grad_norm": 500.35232414817864,
      "learning_rate": 1.4924936461412172e-05,
      "loss": 377.9878,
      "step": 17820
    },
    {
      "epoch": 0.34291429065976864,
      "grad_norm": 490.674885762262,
      "learning_rate": 1.4919623881616989e-05,
      "loss": 367.6417,
      "step": 17830
    },
    {
      "epoch": 0.34310661499552847,
      "grad_norm": 467.642760018096,
      "learning_rate": 1.4914309469343158e-05,
      "loss": 361.5473,
      "step": 17840
    },
    {
      "epoch": 0.3432989393312883,
      "grad_norm": 478.12902654557894,
      "learning_rate": 1.4908993226570214e-05,
      "loss": 389.5556,
      "step": 17850
    },
    {
      "epoch": 0.34349126366704813,
      "grad_norm": 486.2603294498519,
      "learning_rate": 1.4903675155278365e-05,
      "loss": 369.2395,
      "step": 17860
    },
    {
      "epoch": 0.34368358800280796,
      "grad_norm": 466.3609027262824,
      "learning_rate": 1.4898355257448508e-05,
      "loss": 378.8496,
      "step": 17870
    },
    {
      "epoch": 0.34387591233856774,
      "grad_norm": 446.2215810772876,
      "learning_rate": 1.4893033535062219e-05,
      "loss": 369.5298,
      "step": 17880
    },
    {
      "epoch": 0.34406823667432757,
      "grad_norm": 465.8005000615126,
      "learning_rate": 1.4887709990101748e-05,
      "loss": 374.6228,
      "step": 17890
    },
    {
      "epoch": 0.3442605610100874,
      "grad_norm": 487.0060744397368,
      "learning_rate": 1.4882384624550028e-05,
      "loss": 380.0092,
      "step": 17900
    },
    {
      "epoch": 0.3444528853458472,
      "grad_norm": 484.50067612771096,
      "learning_rate": 1.4877057440390672e-05,
      "loss": 364.349,
      "step": 17910
    },
    {
      "epoch": 0.34464520968160706,
      "grad_norm": 484.7299552912734,
      "learning_rate": 1.4871728439607967e-05,
      "loss": 371.8014,
      "step": 17920
    },
    {
      "epoch": 0.3448375340173669,
      "grad_norm": 459.16385695759345,
      "learning_rate": 1.4866397624186877e-05,
      "loss": 370.6781,
      "step": 17930
    },
    {
      "epoch": 0.3450298583531267,
      "grad_norm": 443.9478373940645,
      "learning_rate": 1.4861064996113042e-05,
      "loss": 367.4106,
      "step": 17940
    },
    {
      "epoch": 0.34522218268888655,
      "grad_norm": 495.3128492277379,
      "learning_rate": 1.485573055737278e-05,
      "loss": 379.6042,
      "step": 17950
    },
    {
      "epoch": 0.3454145070246464,
      "grad_norm": 511.57493450094523,
      "learning_rate": 1.485039430995308e-05,
      "loss": 374.9716,
      "step": 17960
    },
    {
      "epoch": 0.3456068313604062,
      "grad_norm": 509.76636955661064,
      "learning_rate": 1.4845056255841608e-05,
      "loss": 378.0369,
      "step": 17970
    },
    {
      "epoch": 0.34579915569616604,
      "grad_norm": 518.4135318649124,
      "learning_rate": 1.48397163970267e-05,
      "loss": 386.0346,
      "step": 17980
    },
    {
      "epoch": 0.3459914800319258,
      "grad_norm": 477.46111326271,
      "learning_rate": 1.4834374735497362e-05,
      "loss": 373.7529,
      "step": 17990
    },
    {
      "epoch": 0.34618380436768564,
      "grad_norm": 517.0975109192298,
      "learning_rate": 1.4829031273243277e-05,
      "loss": 382.4908,
      "step": 18000
    },
    {
      "epoch": 0.3463761287034455,
      "grad_norm": 481.90150309383307,
      "learning_rate": 1.4823686012254798e-05,
      "loss": 368.5236,
      "step": 18010
    },
    {
      "epoch": 0.3465684530392053,
      "grad_norm": 432.72358211087396,
      "learning_rate": 1.4818338954522943e-05,
      "loss": 363.4754,
      "step": 18020
    },
    {
      "epoch": 0.34676077737496513,
      "grad_norm": 475.8098001656779,
      "learning_rate": 1.4812990102039411e-05,
      "loss": 368.5819,
      "step": 18030
    },
    {
      "epoch": 0.34695310171072496,
      "grad_norm": 477.2507632238008,
      "learning_rate": 1.480763945679655e-05,
      "loss": 367.3182,
      "step": 18040
    },
    {
      "epoch": 0.3471454260464848,
      "grad_norm": 481.42520559266967,
      "learning_rate": 1.4802287020787396e-05,
      "loss": 367.6229,
      "step": 18050
    },
    {
      "epoch": 0.3473377503822446,
      "grad_norm": 521.5074361785358,
      "learning_rate": 1.4796932796005634e-05,
      "loss": 376.0522,
      "step": 18060
    },
    {
      "epoch": 0.34753007471800446,
      "grad_norm": 474.8537616330755,
      "learning_rate": 1.4791576784445632e-05,
      "loss": 366.5921,
      "step": 18070
    },
    {
      "epoch": 0.3477223990537643,
      "grad_norm": 473.9143272349859,
      "learning_rate": 1.4786218988102414e-05,
      "loss": 367.4316,
      "step": 18080
    },
    {
      "epoch": 0.3479147233895241,
      "grad_norm": 477.1176968146836,
      "learning_rate": 1.4780859408971668e-05,
      "loss": 363.1174,
      "step": 18090
    },
    {
      "epoch": 0.3481070477252839,
      "grad_norm": 482.7478512382826,
      "learning_rate": 1.4775498049049754e-05,
      "loss": 375.75,
      "step": 18100
    },
    {
      "epoch": 0.3482993720610437,
      "grad_norm": 468.48901283900074,
      "learning_rate": 1.4770134910333684e-05,
      "loss": 378.7698,
      "step": 18110
    },
    {
      "epoch": 0.34849169639680355,
      "grad_norm": 470.13430588110907,
      "learning_rate": 1.4764769994821145e-05,
      "loss": 381.0392,
      "step": 18120
    },
    {
      "epoch": 0.3486840207325634,
      "grad_norm": 521.5453057145618,
      "learning_rate": 1.4759403304510472e-05,
      "loss": 380.8982,
      "step": 18130
    },
    {
      "epoch": 0.3488763450683232,
      "grad_norm": 489.6819104975119,
      "learning_rate": 1.475403484140067e-05,
      "loss": 382.0617,
      "step": 18140
    },
    {
      "epoch": 0.34906866940408304,
      "grad_norm": 493.11120681882915,
      "learning_rate": 1.4748664607491408e-05,
      "loss": 373.9825,
      "step": 18150
    },
    {
      "epoch": 0.3492609937398429,
      "grad_norm": 531.8998048889769,
      "learning_rate": 1.4743292604783008e-05,
      "loss": 378.1855,
      "step": 18160
    },
    {
      "epoch": 0.3494533180756027,
      "grad_norm": 460.6150979845959,
      "learning_rate": 1.4737918835276451e-05,
      "loss": 369.2993,
      "step": 18170
    },
    {
      "epoch": 0.34964564241136253,
      "grad_norm": 566.5690486177965,
      "learning_rate": 1.4732543300973374e-05,
      "loss": 373.2924,
      "step": 18180
    },
    {
      "epoch": 0.34983796674712236,
      "grad_norm": 507.62894627917535,
      "learning_rate": 1.472716600387608e-05,
      "loss": 370.3426,
      "step": 18190
    },
    {
      "epoch": 0.3500302910828822,
      "grad_norm": 454.14600624306524,
      "learning_rate": 1.4721786945987519e-05,
      "loss": 378.0637,
      "step": 18200
    },
    {
      "epoch": 0.350222615418642,
      "grad_norm": 487.7970833153401,
      "learning_rate": 1.4716406129311307e-05,
      "loss": 375.8229,
      "step": 18210
    },
    {
      "epoch": 0.3504149397544018,
      "grad_norm": 557.1772760818087,
      "learning_rate": 1.4711023555851702e-05,
      "loss": 379.5749,
      "step": 18220
    },
    {
      "epoch": 0.35060726409016163,
      "grad_norm": 472.7505135280188,
      "learning_rate": 1.470563922761363e-05,
      "loss": 358.2809,
      "step": 18230
    },
    {
      "epoch": 0.35079958842592146,
      "grad_norm": 447.35925071647546,
      "learning_rate": 1.470025314660266e-05,
      "loss": 371.3621,
      "step": 18240
    },
    {
      "epoch": 0.3509919127616813,
      "grad_norm": 472.2543227029913,
      "learning_rate": 1.4694865314825024e-05,
      "loss": 378.2429,
      "step": 18250
    },
    {
      "epoch": 0.3511842370974411,
      "grad_norm": 513.1253676462142,
      "learning_rate": 1.4689475734287596e-05,
      "loss": 372.8578,
      "step": 18260
    },
    {
      "epoch": 0.35137656143320095,
      "grad_norm": 460.6382226840667,
      "learning_rate": 1.4684084406997903e-05,
      "loss": 376.7589,
      "step": 18270
    },
    {
      "epoch": 0.3515688857689608,
      "grad_norm": 493.2486285313152,
      "learning_rate": 1.467869133496413e-05,
      "loss": 372.8126,
      "step": 18280
    },
    {
      "epoch": 0.3517612101047206,
      "grad_norm": 460.3758687516171,
      "learning_rate": 1.4673296520195105e-05,
      "loss": 366.9916,
      "step": 18290
    },
    {
      "epoch": 0.35195353444048044,
      "grad_norm": 424.9863611582297,
      "learning_rate": 1.4667899964700309e-05,
      "loss": 369.9615,
      "step": 18300
    },
    {
      "epoch": 0.3521458587762403,
      "grad_norm": 471.53786531469933,
      "learning_rate": 1.466250167048987e-05,
      "loss": 371.1262,
      "step": 18310
    },
    {
      "epoch": 0.3523381831120001,
      "grad_norm": 477.33761854951507,
      "learning_rate": 1.4657101639574563e-05,
      "loss": 380.0143,
      "step": 18320
    },
    {
      "epoch": 0.3525305074477599,
      "grad_norm": 472.84647606185706,
      "learning_rate": 1.4651699873965808e-05,
      "loss": 373.6507,
      "step": 18330
    },
    {
      "epoch": 0.3527228317835197,
      "grad_norm": 492.7427337478879,
      "learning_rate": 1.4646296375675676e-05,
      "loss": 381.0266,
      "step": 18340
    },
    {
      "epoch": 0.35291515611927954,
      "grad_norm": 489.60803894214706,
      "learning_rate": 1.464089114671688e-05,
      "loss": 384.5405,
      "step": 18350
    },
    {
      "epoch": 0.35310748045503937,
      "grad_norm": 537.1395429440232,
      "learning_rate": 1.4635484189102776e-05,
      "loss": 369.3183,
      "step": 18360
    },
    {
      "epoch": 0.3532998047907992,
      "grad_norm": 507.2652867755757,
      "learning_rate": 1.4630075504847373e-05,
      "loss": 372.9186,
      "step": 18370
    },
    {
      "epoch": 0.35349212912655903,
      "grad_norm": 490.8104944284974,
      "learning_rate": 1.4624665095965311e-05,
      "loss": 359.4041,
      "step": 18380
    },
    {
      "epoch": 0.35368445346231886,
      "grad_norm": 500.95964086887125,
      "learning_rate": 1.4619252964471881e-05,
      "loss": 366.1107,
      "step": 18390
    },
    {
      "epoch": 0.3538767777980787,
      "grad_norm": 486.0674668984741,
      "learning_rate": 1.461383911238301e-05,
      "loss": 382.9969,
      "step": 18400
    },
    {
      "epoch": 0.3540691021338385,
      "grad_norm": 527.2809508699592,
      "learning_rate": 1.4608423541715273e-05,
      "loss": 368.5316,
      "step": 18410
    },
    {
      "epoch": 0.35426142646959835,
      "grad_norm": 501.4414508551843,
      "learning_rate": 1.4603006254485874e-05,
      "loss": 373.755,
      "step": 18420
    },
    {
      "epoch": 0.3544537508053582,
      "grad_norm": 466.86780574911705,
      "learning_rate": 1.4597587252712666e-05,
      "loss": 371.6094,
      "step": 18430
    },
    {
      "epoch": 0.35464607514111796,
      "grad_norm": 483.7910960149139,
      "learning_rate": 1.4592166538414136e-05,
      "loss": 360.8165,
      "step": 18440
    },
    {
      "epoch": 0.3548383994768778,
      "grad_norm": 473.212601935416,
      "learning_rate": 1.4586744113609416e-05,
      "loss": 374.5418,
      "step": 18450
    },
    {
      "epoch": 0.3550307238126376,
      "grad_norm": 457.5055478376314,
      "learning_rate": 1.4581319980318266e-05,
      "loss": 377.9521,
      "step": 18460
    },
    {
      "epoch": 0.35522304814839745,
      "grad_norm": 496.8836948958783,
      "learning_rate": 1.4575894140561086e-05,
      "loss": 361.657,
      "step": 18470
    },
    {
      "epoch": 0.3554153724841573,
      "grad_norm": 556.6375691072625,
      "learning_rate": 1.4570466596358914e-05,
      "loss": 381.3114,
      "step": 18480
    },
    {
      "epoch": 0.3556076968199171,
      "grad_norm": 466.10931472007314,
      "learning_rate": 1.4565037349733415e-05,
      "loss": 367.4475,
      "step": 18490
    },
    {
      "epoch": 0.35580002115567694,
      "grad_norm": 468.4653706677871,
      "learning_rate": 1.45596064027069e-05,
      "loss": 371.8457,
      "step": 18500
    },
    {
      "epoch": 0.35599234549143677,
      "grad_norm": 478.9980049459254,
      "learning_rate": 1.4554173757302303e-05,
      "loss": 361.1702,
      "step": 18510
    },
    {
      "epoch": 0.3561846698271966,
      "grad_norm": 514.8019688614664,
      "learning_rate": 1.4548739415543197e-05,
      "loss": 363.7348,
      "step": 18520
    },
    {
      "epoch": 0.35637699416295643,
      "grad_norm": 460.8756793552097,
      "learning_rate": 1.454330337945378e-05,
      "loss": 364.9218,
      "step": 18530
    },
    {
      "epoch": 0.35656931849871626,
      "grad_norm": 451.6966638431431,
      "learning_rate": 1.4537865651058893e-05,
      "loss": 364.4594,
      "step": 18540
    },
    {
      "epoch": 0.35676164283447603,
      "grad_norm": 483.5397751016099,
      "learning_rate": 1.4532426232383998e-05,
      "loss": 368.4219,
      "step": 18550
    },
    {
      "epoch": 0.35695396717023586,
      "grad_norm": 479.499384908411,
      "learning_rate": 1.4526985125455184e-05,
      "loss": 371.799,
      "step": 18560
    },
    {
      "epoch": 0.3571462915059957,
      "grad_norm": 480.66528347190183,
      "learning_rate": 1.4521542332299177e-05,
      "loss": 376.2643,
      "step": 18570
    },
    {
      "epoch": 0.3573386158417555,
      "grad_norm": 440.27272620204917,
      "learning_rate": 1.4516097854943325e-05,
      "loss": 364.4861,
      "step": 18580
    },
    {
      "epoch": 0.35753094017751536,
      "grad_norm": 470.423216406799,
      "learning_rate": 1.4510651695415612e-05,
      "loss": 358.5396,
      "step": 18590
    },
    {
      "epoch": 0.3577232645132752,
      "grad_norm": 544.9613137094686,
      "learning_rate": 1.4505203855744637e-05,
      "loss": 372.2926,
      "step": 18600
    },
    {
      "epoch": 0.357915588849035,
      "grad_norm": 469.33719869134734,
      "learning_rate": 1.4499754337959628e-05,
      "loss": 373.9996,
      "step": 18610
    },
    {
      "epoch": 0.35810791318479485,
      "grad_norm": 466.49979759304574,
      "learning_rate": 1.449430314409045e-05,
      "loss": 364.2124,
      "step": 18620
    },
    {
      "epoch": 0.3583002375205547,
      "grad_norm": 463.99293359458255,
      "learning_rate": 1.4488850276167572e-05,
      "loss": 370.5665,
      "step": 18630
    },
    {
      "epoch": 0.3584925618563145,
      "grad_norm": 470.6263087023456,
      "learning_rate": 1.44833957362221e-05,
      "loss": 377.4652,
      "step": 18640
    },
    {
      "epoch": 0.35868488619207434,
      "grad_norm": 499.339775205238,
      "learning_rate": 1.4477939526285767e-05,
      "loss": 379.4685,
      "step": 18650
    },
    {
      "epoch": 0.3588772105278341,
      "grad_norm": 483.2300276479665,
      "learning_rate": 1.4472481648390914e-05,
      "loss": 383.0591,
      "step": 18660
    },
    {
      "epoch": 0.35906953486359394,
      "grad_norm": 521.9206208099488,
      "learning_rate": 1.4467022104570514e-05,
      "loss": 376.079,
      "step": 18670
    },
    {
      "epoch": 0.3592618591993538,
      "grad_norm": 531.1641275059612,
      "learning_rate": 1.4461560896858156e-05,
      "loss": 370.4005,
      "step": 18680
    },
    {
      "epoch": 0.3594541835351136,
      "grad_norm": 467.6617049989319,
      "learning_rate": 1.4456098027288046e-05,
      "loss": 366.0533,
      "step": 18690
    },
    {
      "epoch": 0.35964650787087343,
      "grad_norm": 498.8822631824095,
      "learning_rate": 1.4450633497895017e-05,
      "loss": 368.0153,
      "step": 18700
    },
    {
      "epoch": 0.35983883220663326,
      "grad_norm": 461.3429916911708,
      "learning_rate": 1.4445167310714514e-05,
      "loss": 379.8898,
      "step": 18710
    },
    {
      "epoch": 0.3600311565423931,
      "grad_norm": 489.02969225920174,
      "learning_rate": 1.4439699467782602e-05,
      "loss": 365.7172,
      "step": 18720
    },
    {
      "epoch": 0.3602234808781529,
      "grad_norm": 483.3793464633447,
      "learning_rate": 1.4434229971135965e-05,
      "loss": 373.2573,
      "step": 18730
    },
    {
      "epoch": 0.36041580521391275,
      "grad_norm": 472.3147493799003,
      "learning_rate": 1.4428758822811894e-05,
      "loss": 366.5052,
      "step": 18740
    },
    {
      "epoch": 0.3606081295496726,
      "grad_norm": 537.1914996117569,
      "learning_rate": 1.442328602484831e-05,
      "loss": 378.5692,
      "step": 18750
    },
    {
      "epoch": 0.3608004538854324,
      "grad_norm": 469.97997058596286,
      "learning_rate": 1.441781157928373e-05,
      "loss": 363.9328,
      "step": 18760
    },
    {
      "epoch": 0.36099277822119225,
      "grad_norm": 463.8028779875606,
      "learning_rate": 1.44123354881573e-05,
      "loss": 359.4484,
      "step": 18770
    },
    {
      "epoch": 0.361185102556952,
      "grad_norm": 495.25411281555085,
      "learning_rate": 1.4406857753508772e-05,
      "loss": 375.4433,
      "step": 18780
    },
    {
      "epoch": 0.36137742689271185,
      "grad_norm": 500.1093255159945,
      "learning_rate": 1.4401378377378512e-05,
      "loss": 372.3941,
      "step": 18790
    },
    {
      "epoch": 0.3615697512284717,
      "grad_norm": 500.5009044830819,
      "learning_rate": 1.43958973618075e-05,
      "loss": 360.7825,
      "step": 18800
    },
    {
      "epoch": 0.3617620755642315,
      "grad_norm": 436.95753915479196,
      "learning_rate": 1.4390414708837322e-05,
      "loss": 362.9499,
      "step": 18810
    },
    {
      "epoch": 0.36195439989999134,
      "grad_norm": 481.0786028952087,
      "learning_rate": 1.4384930420510173e-05,
      "loss": 364.6544,
      "step": 18820
    },
    {
      "epoch": 0.36214672423575117,
      "grad_norm": 687.8924801110628,
      "learning_rate": 1.4379444498868864e-05,
      "loss": 362.6777,
      "step": 18830
    },
    {
      "epoch": 0.362339048571511,
      "grad_norm": 507.4525976413074,
      "learning_rate": 1.4373956945956807e-05,
      "loss": 382.8648,
      "step": 18840
    },
    {
      "epoch": 0.36253137290727083,
      "grad_norm": 482.6485776818181,
      "learning_rate": 1.4368467763818026e-05,
      "loss": 362.4334,
      "step": 18850
    },
    {
      "epoch": 0.36272369724303066,
      "grad_norm": 478.8925713377844,
      "learning_rate": 1.436297695449715e-05,
      "loss": 376.4112,
      "step": 18860
    },
    {
      "epoch": 0.3629160215787905,
      "grad_norm": 492.0712608783987,
      "learning_rate": 1.4357484520039412e-05,
      "loss": 364.4529,
      "step": 18870
    },
    {
      "epoch": 0.3631083459145503,
      "grad_norm": 549.5561926360407,
      "learning_rate": 1.4351990462490662e-05,
      "loss": 368.4865,
      "step": 18880
    },
    {
      "epoch": 0.3633006702503101,
      "grad_norm": 435.3191540823031,
      "learning_rate": 1.434649478389734e-05,
      "loss": 361.5154,
      "step": 18890
    },
    {
      "epoch": 0.36349299458606993,
      "grad_norm": 454.04073118695345,
      "learning_rate": 1.4340997486306491e-05,
      "loss": 376.9093,
      "step": 18900
    },
    {
      "epoch": 0.36368531892182976,
      "grad_norm": 478.90101693098524,
      "learning_rate": 1.4335498571765777e-05,
      "loss": 377.3971,
      "step": 18910
    },
    {
      "epoch": 0.3638776432575896,
      "grad_norm": 489.3196143523086,
      "learning_rate": 1.4329998042323447e-05,
      "loss": 357.2565,
      "step": 18920
    },
    {
      "epoch": 0.3640699675933494,
      "grad_norm": 487.7085603866977,
      "learning_rate": 1.4324495900028358e-05,
      "loss": 376.0046,
      "step": 18930
    },
    {
      "epoch": 0.36426229192910925,
      "grad_norm": 517.9821557927096,
      "learning_rate": 1.4318992146929967e-05,
      "loss": 367.6477,
      "step": 18940
    },
    {
      "epoch": 0.3644546162648691,
      "grad_norm": 465.237196888301,
      "learning_rate": 1.4313486785078335e-05,
      "loss": 365.7833,
      "step": 18950
    },
    {
      "epoch": 0.3646469406006289,
      "grad_norm": 494.7187196327382,
      "learning_rate": 1.4307979816524111e-05,
      "loss": 371.6348,
      "step": 18960
    },
    {
      "epoch": 0.36483926493638874,
      "grad_norm": 475.7723785490395,
      "learning_rate": 1.4302471243318554e-05,
      "loss": 374.1505,
      "step": 18970
    },
    {
      "epoch": 0.36503158927214857,
      "grad_norm": 446.2522602947378,
      "learning_rate": 1.4296961067513519e-05,
      "loss": 368.5986,
      "step": 18980
    },
    {
      "epoch": 0.3652239136079084,
      "grad_norm": 483.7849072662519,
      "learning_rate": 1.4291449291161452e-05,
      "loss": 365.7878,
      "step": 18990
    },
    {
      "epoch": 0.3654162379436682,
      "grad_norm": 447.9650747489853,
      "learning_rate": 1.4285935916315401e-05,
      "loss": 365.5911,
      "step": 19000
    },
    {
      "epoch": 0.365608562279428,
      "grad_norm": 540.3015209984369,
      "learning_rate": 1.4280420945029004e-05,
      "loss": 379.3237,
      "step": 19010
    },
    {
      "epoch": 0.36580088661518784,
      "grad_norm": 453.9456310980776,
      "learning_rate": 1.4274904379356498e-05,
      "loss": 370.3269,
      "step": 19020
    },
    {
      "epoch": 0.36599321095094767,
      "grad_norm": 478.07166752877004,
      "learning_rate": 1.4269386221352714e-05,
      "loss": 367.498,
      "step": 19030
    },
    {
      "epoch": 0.3661855352867075,
      "grad_norm": 474.19457759997107,
      "learning_rate": 1.4263866473073076e-05,
      "loss": 363.9463,
      "step": 19040
    },
    {
      "epoch": 0.36637785962246733,
      "grad_norm": 475.50977595489707,
      "learning_rate": 1.4258345136573595e-05,
      "loss": 360.9163,
      "step": 19050
    },
    {
      "epoch": 0.36657018395822716,
      "grad_norm": 490.7470611620901,
      "learning_rate": 1.4252822213910878e-05,
      "loss": 369.506,
      "step": 19060
    },
    {
      "epoch": 0.366762508293987,
      "grad_norm": 515.9097384940071,
      "learning_rate": 1.4247297707142126e-05,
      "loss": 365.0083,
      "step": 19070
    },
    {
      "epoch": 0.3669548326297468,
      "grad_norm": 463.39629709100745,
      "learning_rate": 1.4241771618325123e-05,
      "loss": 366.3742,
      "step": 19080
    },
    {
      "epoch": 0.36714715696550665,
      "grad_norm": 469.1828279196171,
      "learning_rate": 1.4236243949518249e-05,
      "loss": 384.597,
      "step": 19090
    },
    {
      "epoch": 0.3673394813012665,
      "grad_norm": 455.8680547902369,
      "learning_rate": 1.4230714702780466e-05,
      "loss": 367.4729,
      "step": 19100
    },
    {
      "epoch": 0.36753180563702625,
      "grad_norm": 467.31672538611986,
      "learning_rate": 1.422518388017133e-05,
      "loss": 374.8073,
      "step": 19110
    },
    {
      "epoch": 0.3677241299727861,
      "grad_norm": 457.39484363915466,
      "learning_rate": 1.4219651483750978e-05,
      "loss": 363.8252,
      "step": 19120
    },
    {
      "epoch": 0.3679164543085459,
      "grad_norm": 450.0351127175283,
      "learning_rate": 1.4214117515580139e-05,
      "loss": 370.6834,
      "step": 19130
    },
    {
      "epoch": 0.36810877864430575,
      "grad_norm": 480.7195428253656,
      "learning_rate": 1.4208581977720124e-05,
      "loss": 374.3641,
      "step": 19140
    },
    {
      "epoch": 0.3683011029800656,
      "grad_norm": 455.6731196219641,
      "learning_rate": 1.420304487223283e-05,
      "loss": 361.2098,
      "step": 19150
    },
    {
      "epoch": 0.3684934273158254,
      "grad_norm": 466.3299826478185,
      "learning_rate": 1.4197506201180737e-05,
      "loss": 367.955,
      "step": 19160
    },
    {
      "epoch": 0.36868575165158524,
      "grad_norm": 498.94977562167395,
      "learning_rate": 1.4191965966626908e-05,
      "loss": 357.7075,
      "step": 19170
    },
    {
      "epoch": 0.36887807598734507,
      "grad_norm": 461.95530452123086,
      "learning_rate": 1.418642417063499e-05,
      "loss": 360.8351,
      "step": 19180
    },
    {
      "epoch": 0.3690704003231049,
      "grad_norm": 492.1237723966069,
      "learning_rate": 1.4180880815269207e-05,
      "loss": 356.3411,
      "step": 19190
    },
    {
      "epoch": 0.3692627246588647,
      "grad_norm": 492.5773755139487,
      "learning_rate": 1.4175335902594372e-05,
      "loss": 365.5533,
      "step": 19200
    },
    {
      "epoch": 0.36945504899462456,
      "grad_norm": 478.34406730903527,
      "learning_rate": 1.416978943467587e-05,
      "loss": 370.8607,
      "step": 19210
    },
    {
      "epoch": 0.3696473733303844,
      "grad_norm": 490.27236538774264,
      "learning_rate": 1.4164241413579669e-05,
      "loss": 372.9625,
      "step": 19220
    },
    {
      "epoch": 0.36983969766614416,
      "grad_norm": 469.43304628567404,
      "learning_rate": 1.4158691841372318e-05,
      "loss": 372.9711,
      "step": 19230
    },
    {
      "epoch": 0.370032022001904,
      "grad_norm": 470.9117717020862,
      "learning_rate": 1.4153140720120936e-05,
      "loss": 363.0882,
      "step": 19240
    },
    {
      "epoch": 0.3702243463376638,
      "grad_norm": 456.1484694341779,
      "learning_rate": 1.4147588051893233e-05,
      "loss": 363.6552,
      "step": 19250
    },
    {
      "epoch": 0.37041667067342365,
      "grad_norm": 455.73373772994705,
      "learning_rate": 1.4142033838757476e-05,
      "loss": 364.5586,
      "step": 19260
    },
    {
      "epoch": 0.3706089950091835,
      "grad_norm": 442.4048520737741,
      "learning_rate": 1.4136478082782525e-05,
      "loss": 365.4893,
      "step": 19270
    },
    {
      "epoch": 0.3708013193449433,
      "grad_norm": 450.78640381123984,
      "learning_rate": 1.4130920786037798e-05,
      "loss": 366.0341,
      "step": 19280
    },
    {
      "epoch": 0.37099364368070314,
      "grad_norm": 465.85437142306637,
      "learning_rate": 1.412536195059331e-05,
      "loss": 373.8014,
      "step": 19290
    },
    {
      "epoch": 0.371185968016463,
      "grad_norm": 450.2548230321448,
      "learning_rate": 1.4119801578519625e-05,
      "loss": 375.3096,
      "step": 19300
    },
    {
      "epoch": 0.3713782923522228,
      "grad_norm": 467.73712802617746,
      "learning_rate": 1.4114239671887892e-05,
      "loss": 375.294,
      "step": 19310
    },
    {
      "epoch": 0.37157061668798264,
      "grad_norm": 519.3381325164414,
      "learning_rate": 1.4108676232769831e-05,
      "loss": 367.66,
      "step": 19320
    },
    {
      "epoch": 0.37176294102374247,
      "grad_norm": 491.61357238814594,
      "learning_rate": 1.410311126323773e-05,
      "loss": 369.1357,
      "step": 19330
    },
    {
      "epoch": 0.37195526535950224,
      "grad_norm": 520.1743038584326,
      "learning_rate": 1.409754476536445e-05,
      "loss": 359.8314,
      "step": 19340
    },
    {
      "epoch": 0.37214758969526207,
      "grad_norm": 455.0012210234154,
      "learning_rate": 1.4091976741223414e-05,
      "loss": 360.3283,
      "step": 19350
    },
    {
      "epoch": 0.3723399140310219,
      "grad_norm": 529.730473525027,
      "learning_rate": 1.408640719288863e-05,
      "loss": 382.2609,
      "step": 19360
    },
    {
      "epoch": 0.37253223836678173,
      "grad_norm": 463.0720758737707,
      "learning_rate": 1.408083612243465e-05,
      "loss": 375.0657,
      "step": 19370
    },
    {
      "epoch": 0.37272456270254156,
      "grad_norm": 458.8636394086308,
      "learning_rate": 1.4075263531936614e-05,
      "loss": 371.4612,
      "step": 19380
    },
    {
      "epoch": 0.3729168870383014,
      "grad_norm": 475.31286439351413,
      "learning_rate": 1.4069689423470219e-05,
      "loss": 360.5068,
      "step": 19390
    },
    {
      "epoch": 0.3731092113740612,
      "grad_norm": 475.28123605968545,
      "learning_rate": 1.4064113799111725e-05,
      "loss": 369.6281,
      "step": 19400
    },
    {
      "epoch": 0.37330153570982105,
      "grad_norm": 525.3843914109004,
      "learning_rate": 1.405853666093796e-05,
      "loss": 363.4456,
      "step": 19410
    },
    {
      "epoch": 0.3734938600455809,
      "grad_norm": 527.4287327678679,
      "learning_rate": 1.405295801102632e-05,
      "loss": 366.8147,
      "step": 19420
    },
    {
      "epoch": 0.3736861843813407,
      "grad_norm": 486.28243353481656,
      "learning_rate": 1.4047377851454758e-05,
      "loss": 374.8594,
      "step": 19430
    },
    {
      "epoch": 0.37387850871710054,
      "grad_norm": 459.48720811620757,
      "learning_rate": 1.4041796184301788e-05,
      "loss": 357.8949,
      "step": 19440
    },
    {
      "epoch": 0.3740708330528603,
      "grad_norm": 496.1689285256519,
      "learning_rate": 1.4036213011646496e-05,
      "loss": 375.414,
      "step": 19450
    },
    {
      "epoch": 0.37426315738862015,
      "grad_norm": 504.21145921952154,
      "learning_rate": 1.4030628335568515e-05,
      "loss": 376.2479,
      "step": 19460
    },
    {
      "epoch": 0.37445548172438,
      "grad_norm": 487.32683024066483,
      "learning_rate": 1.4025042158148048e-05,
      "loss": 365.3398,
      "step": 19470
    },
    {
      "epoch": 0.3746478060601398,
      "grad_norm": 500.42549569346295,
      "learning_rate": 1.4019454481465853e-05,
      "loss": 368.8787,
      "step": 19480
    },
    {
      "epoch": 0.37484013039589964,
      "grad_norm": 487.6545501117345,
      "learning_rate": 1.4013865307603248e-05,
      "loss": 360.3625,
      "step": 19490
    },
    {
      "epoch": 0.37503245473165947,
      "grad_norm": 485.5966714022845,
      "learning_rate": 1.4008274638642103e-05,
      "loss": 372.803,
      "step": 19500
    },
    {
      "epoch": 0.3752247790674193,
      "grad_norm": 464.27557752870007,
      "learning_rate": 1.4002682476664857e-05,
      "loss": 361.566,
      "step": 19510
    },
    {
      "epoch": 0.37541710340317913,
      "grad_norm": 468.2358679428454,
      "learning_rate": 1.3997088823754494e-05,
      "loss": 357.5025,
      "step": 19520
    },
    {
      "epoch": 0.37560942773893896,
      "grad_norm": 548.9838545754756,
      "learning_rate": 1.399149368199456e-05,
      "loss": 372.736,
      "step": 19530
    },
    {
      "epoch": 0.3758017520746988,
      "grad_norm": 470.50530292960065,
      "learning_rate": 1.398589705346915e-05,
      "loss": 362.6622,
      "step": 19540
    },
    {
      "epoch": 0.3759940764104586,
      "grad_norm": 472.32480058256823,
      "learning_rate": 1.3980298940262918e-05,
      "loss": 364.1768,
      "step": 19550
    },
    {
      "epoch": 0.3761864007462184,
      "grad_norm": 454.8190138741512,
      "learning_rate": 1.3974699344461065e-05,
      "loss": 363.9456,
      "step": 19560
    },
    {
      "epoch": 0.3763787250819782,
      "grad_norm": 462.7899615553797,
      "learning_rate": 1.396909826814935e-05,
      "loss": 361.9878,
      "step": 19570
    },
    {
      "epoch": 0.37657104941773806,
      "grad_norm": 496.5285422127198,
      "learning_rate": 1.3963495713414085e-05,
      "loss": 365.7559,
      "step": 19580
    },
    {
      "epoch": 0.3767633737534979,
      "grad_norm": 463.41592531727036,
      "learning_rate": 1.3957891682342127e-05,
      "loss": 369.7199,
      "step": 19590
    },
    {
      "epoch": 0.3769556980892577,
      "grad_norm": 513.3141011068556,
      "learning_rate": 1.3952286177020879e-05,
      "loss": 368.0453,
      "step": 19600
    },
    {
      "epoch": 0.37714802242501755,
      "grad_norm": 493.3014176705202,
      "learning_rate": 1.3946679199538308e-05,
      "loss": 359.7146,
      "step": 19610
    },
    {
      "epoch": 0.3773403467607774,
      "grad_norm": 457.4156258324006,
      "learning_rate": 1.3941070751982917e-05,
      "loss": 359.958,
      "step": 19620
    },
    {
      "epoch": 0.3775326710965372,
      "grad_norm": 458.2063917914075,
      "learning_rate": 1.3935460836443758e-05,
      "loss": 364.2169,
      "step": 19630
    },
    {
      "epoch": 0.37772499543229704,
      "grad_norm": 465.3891272534213,
      "learning_rate": 1.3929849455010433e-05,
      "loss": 371.9292,
      "step": 19640
    },
    {
      "epoch": 0.37791731976805687,
      "grad_norm": 473.1018145388463,
      "learning_rate": 1.3924236609773094e-05,
      "loss": 366.6291,
      "step": 19650
    },
    {
      "epoch": 0.3781096441038167,
      "grad_norm": 452.29374844932244,
      "learning_rate": 1.3918622302822425e-05,
      "loss": 364.0584,
      "step": 19660
    },
    {
      "epoch": 0.3783019684395765,
      "grad_norm": 452.7003382353477,
      "learning_rate": 1.391300653624967e-05,
      "loss": 373.7889,
      "step": 19670
    },
    {
      "epoch": 0.3784942927753363,
      "grad_norm": 431.7075819021544,
      "learning_rate": 1.39073893121466e-05,
      "loss": 367.3789,
      "step": 19680
    },
    {
      "epoch": 0.37868661711109614,
      "grad_norm": 474.0787559246814,
      "learning_rate": 1.3901770632605546e-05,
      "loss": 364.3479,
      "step": 19690
    },
    {
      "epoch": 0.37887894144685597,
      "grad_norm": 456.01172132628227,
      "learning_rate": 1.3896150499719372e-05,
      "loss": 359.5419,
      "step": 19700
    },
    {
      "epoch": 0.3790712657826158,
      "grad_norm": 487.28696942430304,
      "learning_rate": 1.3890528915581482e-05,
      "loss": 367.0523,
      "step": 19710
    },
    {
      "epoch": 0.3792635901183756,
      "grad_norm": 495.7553666359008,
      "learning_rate": 1.3884905882285829e-05,
      "loss": 367.023,
      "step": 19720
    },
    {
      "epoch": 0.37945591445413546,
      "grad_norm": 431.17508580800506,
      "learning_rate": 1.3879281401926894e-05,
      "loss": 360.7892,
      "step": 19730
    },
    {
      "epoch": 0.3796482387898953,
      "grad_norm": 477.5925671809703,
      "learning_rate": 1.3873655476599707e-05,
      "loss": 358.9177,
      "step": 19740
    },
    {
      "epoch": 0.3798405631256551,
      "grad_norm": 445.276611087243,
      "learning_rate": 1.3868028108399829e-05,
      "loss": 362.4861,
      "step": 19750
    },
    {
      "epoch": 0.38003288746141495,
      "grad_norm": 446.24543483448787,
      "learning_rate": 1.3862399299423364e-05,
      "loss": 363.028,
      "step": 19760
    },
    {
      "epoch": 0.3802252117971748,
      "grad_norm": 475.1323465293365,
      "learning_rate": 1.3856769051766947e-05,
      "loss": 366.61,
      "step": 19770
    },
    {
      "epoch": 0.3804175361329346,
      "grad_norm": 453.86170588270767,
      "learning_rate": 1.3851137367527757e-05,
      "loss": 352.2456,
      "step": 19780
    },
    {
      "epoch": 0.3806098604686944,
      "grad_norm": 457.86424782971244,
      "learning_rate": 1.3845504248803501e-05,
      "loss": 366.6129,
      "step": 19790
    },
    {
      "epoch": 0.3808021848044542,
      "grad_norm": 471.2643524137598,
      "learning_rate": 1.383986969769242e-05,
      "loss": 377.4821,
      "step": 19800
    },
    {
      "epoch": 0.38099450914021404,
      "grad_norm": 482.84436395394437,
      "learning_rate": 1.38342337162933e-05,
      "loss": 368.8804,
      "step": 19810
    },
    {
      "epoch": 0.3811868334759739,
      "grad_norm": 457.05015934093933,
      "learning_rate": 1.3828596306705442e-05,
      "loss": 360.2316,
      "step": 19820
    },
    {
      "epoch": 0.3813791578117337,
      "grad_norm": 452.45954436302424,
      "learning_rate": 1.3822957471028693e-05,
      "loss": 354.8089,
      "step": 19830
    },
    {
      "epoch": 0.38157148214749353,
      "grad_norm": 493.8366998821656,
      "learning_rate": 1.3817317211363422e-05,
      "loss": 355.6104,
      "step": 19840
    },
    {
      "epoch": 0.38176380648325337,
      "grad_norm": 533.0477094840109,
      "learning_rate": 1.3811675529810535e-05,
      "loss": 362.2145,
      "step": 19850
    },
    {
      "epoch": 0.3819561308190132,
      "grad_norm": 501.3865763672517,
      "learning_rate": 1.3806032428471463e-05,
      "loss": 365.4072,
      "step": 19860
    },
    {
      "epoch": 0.382148455154773,
      "grad_norm": 477.5726462219742,
      "learning_rate": 1.3800387909448171e-05,
      "loss": 366.6195,
      "step": 19870
    },
    {
      "epoch": 0.38234077949053286,
      "grad_norm": 483.90916861451507,
      "learning_rate": 1.3794741974843154e-05,
      "loss": 372.125,
      "step": 19880
    },
    {
      "epoch": 0.3825331038262927,
      "grad_norm": 466.06497337132845,
      "learning_rate": 1.3789094626759419e-05,
      "loss": 358.4106,
      "step": 19890
    },
    {
      "epoch": 0.38272542816205246,
      "grad_norm": 439.4312828146732,
      "learning_rate": 1.3783445867300515e-05,
      "loss": 368.8052,
      "step": 19900
    },
    {
      "epoch": 0.3829177524978123,
      "grad_norm": 448.6623636390141,
      "learning_rate": 1.3777795698570511e-05,
      "loss": 371.1609,
      "step": 19910
    },
    {
      "epoch": 0.3831100768335721,
      "grad_norm": 534.8038140927481,
      "learning_rate": 1.3772144122674e-05,
      "loss": 356.932,
      "step": 19920
    },
    {
      "epoch": 0.38330240116933195,
      "grad_norm": 445.9865402663116,
      "learning_rate": 1.3766491141716103e-05,
      "loss": 360.1602,
      "step": 19930
    },
    {
      "epoch": 0.3834947255050918,
      "grad_norm": 471.0283681984944,
      "learning_rate": 1.3760836757802462e-05,
      "loss": 366.6469,
      "step": 19940
    },
    {
      "epoch": 0.3836870498408516,
      "grad_norm": 468.1572778667453,
      "learning_rate": 1.3755180973039241e-05,
      "loss": 366.2488,
      "step": 19950
    },
    {
      "epoch": 0.38387937417661144,
      "grad_norm": 506.7616728459948,
      "learning_rate": 1.3749523789533128e-05,
      "loss": 362.2985,
      "step": 19960
    },
    {
      "epoch": 0.3840716985123713,
      "grad_norm": 472.9821945188994,
      "learning_rate": 1.374386520939133e-05,
      "loss": 358.6702,
      "step": 19970
    },
    {
      "epoch": 0.3842640228481311,
      "grad_norm": 509.3010479427292,
      "learning_rate": 1.3738205234721568e-05,
      "loss": 379.9848,
      "step": 19980
    },
    {
      "epoch": 0.38445634718389093,
      "grad_norm": 485.03072891581843,
      "learning_rate": 1.3732543867632098e-05,
      "loss": 362.7552,
      "step": 19990
    },
    {
      "epoch": 0.38464867151965076,
      "grad_norm": 453.2582155133797,
      "learning_rate": 1.3726881110231682e-05,
      "loss": 362.8308,
      "step": 20000
    },
    {
      "epoch": 0.38484099585541054,
      "grad_norm": 460.1577311549998,
      "learning_rate": 1.3721216964629605e-05,
      "loss": 366.6201,
      "step": 20010
    },
    {
      "epoch": 0.38503332019117037,
      "grad_norm": 461.28718521613234,
      "learning_rate": 1.3715551432935664e-05,
      "loss": 377.0017,
      "step": 20020
    },
    {
      "epoch": 0.3852256445269302,
      "grad_norm": 472.8270687580648,
      "learning_rate": 1.3709884517260178e-05,
      "loss": 378.4079,
      "step": 20030
    },
    {
      "epoch": 0.38541796886269003,
      "grad_norm": 479.17719695105575,
      "learning_rate": 1.3704216219713984e-05,
      "loss": 356.01,
      "step": 20040
    },
    {
      "epoch": 0.38561029319844986,
      "grad_norm": 491.09679332711613,
      "learning_rate": 1.3698546542408424e-05,
      "loss": 366.6224,
      "step": 20050
    },
    {
      "epoch": 0.3858026175342097,
      "grad_norm": 472.94637955620817,
      "learning_rate": 1.369287548745536e-05,
      "loss": 358.1708,
      "step": 20060
    },
    {
      "epoch": 0.3859949418699695,
      "grad_norm": 633.3068023514486,
      "learning_rate": 1.3687203056967165e-05,
      "loss": 367.8477,
      "step": 20070
    },
    {
      "epoch": 0.38618726620572935,
      "grad_norm": 493.0946979828676,
      "learning_rate": 1.368152925305673e-05,
      "loss": 370.3541,
      "step": 20080
    },
    {
      "epoch": 0.3863795905414892,
      "grad_norm": 465.67294376250896,
      "learning_rate": 1.3675854077837449e-05,
      "loss": 365.836,
      "step": 20090
    },
    {
      "epoch": 0.386571914877249,
      "grad_norm": 438.3645445878314,
      "learning_rate": 1.3670177533423234e-05,
      "loss": 354.1514,
      "step": 20100
    },
    {
      "epoch": 0.38676423921300884,
      "grad_norm": 458.8383294964573,
      "learning_rate": 1.3664499621928502e-05,
      "loss": 361.6879,
      "step": 20110
    },
    {
      "epoch": 0.3869565635487686,
      "grad_norm": 479.9508481813655,
      "learning_rate": 1.3658820345468183e-05,
      "loss": 367.6514,
      "step": 20120
    },
    {
      "epoch": 0.38714888788452845,
      "grad_norm": 450.4661863666756,
      "learning_rate": 1.365313970615771e-05,
      "loss": 357.3286,
      "step": 20130
    },
    {
      "epoch": 0.3873412122202883,
      "grad_norm": 476.57308222670775,
      "learning_rate": 1.3647457706113031e-05,
      "loss": 368.3551,
      "step": 20140
    },
    {
      "epoch": 0.3875335365560481,
      "grad_norm": 478.1509107793025,
      "learning_rate": 1.36417743474506e-05,
      "loss": 369.8673,
      "step": 20150
    },
    {
      "epoch": 0.38772586089180794,
      "grad_norm": 450.1315484878896,
      "learning_rate": 1.3636089632287369e-05,
      "loss": 376.2891,
      "step": 20160
    },
    {
      "epoch": 0.38791818522756777,
      "grad_norm": 466.403720146349,
      "learning_rate": 1.36304035627408e-05,
      "loss": 368.5667,
      "step": 20170
    },
    {
      "epoch": 0.3881105095633276,
      "grad_norm": 438.24305150782135,
      "learning_rate": 1.3624716140928861e-05,
      "loss": 376.8687,
      "step": 20180
    },
    {
      "epoch": 0.38830283389908743,
      "grad_norm": 448.70357037877056,
      "learning_rate": 1.3619027368970025e-05,
      "loss": 369.4069,
      "step": 20190
    },
    {
      "epoch": 0.38849515823484726,
      "grad_norm": 488.48898350293024,
      "learning_rate": 1.3613337248983265e-05,
      "loss": 360.4751,
      "step": 20200
    },
    {
      "epoch": 0.3886874825706071,
      "grad_norm": 478.67847656757516,
      "learning_rate": 1.3607645783088055e-05,
      "loss": 366.1508,
      "step": 20210
    },
    {
      "epoch": 0.3888798069063669,
      "grad_norm": 460.859899181671,
      "learning_rate": 1.3601952973404376e-05,
      "loss": 373.7587,
      "step": 20220
    },
    {
      "epoch": 0.3890721312421267,
      "grad_norm": 427.59912678933176,
      "learning_rate": 1.35962588220527e-05,
      "loss": 376.6293,
      "step": 20230
    },
    {
      "epoch": 0.3892644555778865,
      "grad_norm": 483.46941360647116,
      "learning_rate": 1.3590563331154008e-05,
      "loss": 376.1973,
      "step": 20240
    },
    {
      "epoch": 0.38945677991364636,
      "grad_norm": 462.9376483176115,
      "learning_rate": 1.3584866502829774e-05,
      "loss": 368.2831,
      "step": 20250
    },
    {
      "epoch": 0.3896491042494062,
      "grad_norm": 478.95172047171945,
      "learning_rate": 1.3579168339201975e-05,
      "loss": 366.7303,
      "step": 20260
    },
    {
      "epoch": 0.389841428585166,
      "grad_norm": 463.70949404051413,
      "learning_rate": 1.3573468842393077e-05,
      "loss": 361.8117,
      "step": 20270
    },
    {
      "epoch": 0.39003375292092585,
      "grad_norm": 458.8731170959038,
      "learning_rate": 1.356776801452606e-05,
      "loss": 366.3602,
      "step": 20280
    },
    {
      "epoch": 0.3902260772566857,
      "grad_norm": 446.06714177503085,
      "learning_rate": 1.3562065857724378e-05,
      "loss": 366.7551,
      "step": 20290
    },
    {
      "epoch": 0.3904184015924455,
      "grad_norm": 518.5084723170153,
      "learning_rate": 1.3556362374111993e-05,
      "loss": 362.1282,
      "step": 20300
    },
    {
      "epoch": 0.39061072592820534,
      "grad_norm": 508.5864425234927,
      "learning_rate": 1.3550657565813362e-05,
      "loss": 369.1649,
      "step": 20310
    },
    {
      "epoch": 0.39080305026396517,
      "grad_norm": 470.47778414380946,
      "learning_rate": 1.3544951434953423e-05,
      "loss": 358.6583,
      "step": 20320
    },
    {
      "epoch": 0.390995374599725,
      "grad_norm": 468.5432529014539,
      "learning_rate": 1.3539243983657627e-05,
      "loss": 363.1038,
      "step": 20330
    },
    {
      "epoch": 0.39118769893548483,
      "grad_norm": 453.1641573954629,
      "learning_rate": 1.3533535214051896e-05,
      "loss": 363.9854,
      "step": 20340
    },
    {
      "epoch": 0.3913800232712446,
      "grad_norm": 484.96424082685286,
      "learning_rate": 1.3527825128262656e-05,
      "loss": 367.4173,
      "step": 20350
    },
    {
      "epoch": 0.39157234760700443,
      "grad_norm": 448.83586770584895,
      "learning_rate": 1.3522113728416821e-05,
      "loss": 372.012,
      "step": 20360
    },
    {
      "epoch": 0.39176467194276426,
      "grad_norm": 458.8468604306312,
      "learning_rate": 1.3516401016641793e-05,
      "loss": 359.7743,
      "step": 20370
    },
    {
      "epoch": 0.3919569962785241,
      "grad_norm": 494.97774992361997,
      "learning_rate": 1.351068699506546e-05,
      "loss": 365.4681,
      "step": 20380
    },
    {
      "epoch": 0.3921493206142839,
      "grad_norm": 436.9973681365952,
      "learning_rate": 1.3504971665816202e-05,
      "loss": 360.1445,
      "step": 20390
    },
    {
      "epoch": 0.39234164495004376,
      "grad_norm": 467.59108084396087,
      "learning_rate": 1.3499255031022887e-05,
      "loss": 353.143,
      "step": 20400
    },
    {
      "epoch": 0.3925339692858036,
      "grad_norm": 455.5612168175912,
      "learning_rate": 1.3493537092814863e-05,
      "loss": 366.5097,
      "step": 20410
    },
    {
      "epoch": 0.3927262936215634,
      "grad_norm": 525.4407351856285,
      "learning_rate": 1.348781785332197e-05,
      "loss": 363.0196,
      "step": 20420
    },
    {
      "epoch": 0.39291861795732325,
      "grad_norm": 479.77735869209687,
      "learning_rate": 1.3482097314674526e-05,
      "loss": 359.4147,
      "step": 20430
    },
    {
      "epoch": 0.3931109422930831,
      "grad_norm": 501.05363325645897,
      "learning_rate": 1.3476375479003347e-05,
      "loss": 357.4058,
      "step": 20440
    },
    {
      "epoch": 0.3933032666288429,
      "grad_norm": 499.8402290218398,
      "learning_rate": 1.3470652348439715e-05,
      "loss": 361.6666,
      "step": 20450
    },
    {
      "epoch": 0.3934955909646027,
      "grad_norm": 778.1187248257576,
      "learning_rate": 1.3464927925115405e-05,
      "loss": 379.1409,
      "step": 20460
    },
    {
      "epoch": 0.3936879153003625,
      "grad_norm": 480.1673190600996,
      "learning_rate": 1.3459202211162663e-05,
      "loss": 359.0808,
      "step": 20470
    },
    {
      "epoch": 0.39388023963612234,
      "grad_norm": 524.8709136005673,
      "learning_rate": 1.345347520871423e-05,
      "loss": 360.9238,
      "step": 20480
    },
    {
      "epoch": 0.3940725639718822,
      "grad_norm": 478.43321389043695,
      "learning_rate": 1.3447746919903318e-05,
      "loss": 356.7554,
      "step": 20490
    },
    {
      "epoch": 0.394264888307642,
      "grad_norm": 435.3551589654242,
      "learning_rate": 1.3442017346863618e-05,
      "loss": 366.2917,
      "step": 20500
    },
    {
      "epoch": 0.39445721264340183,
      "grad_norm": 468.6913303477722,
      "learning_rate": 1.3436286491729306e-05,
      "loss": 372.9864,
      "step": 20510
    },
    {
      "epoch": 0.39464953697916166,
      "grad_norm": 466.00802438493156,
      "learning_rate": 1.3430554356635029e-05,
      "loss": 367.0901,
      "step": 20520
    },
    {
      "epoch": 0.3948418613149215,
      "grad_norm": 452.08959574790794,
      "learning_rate": 1.342482094371591e-05,
      "loss": 359.0619,
      "step": 20530
    },
    {
      "epoch": 0.3950341856506813,
      "grad_norm": 453.30372623543707,
      "learning_rate": 1.341908625510755e-05,
      "loss": 372.5314,
      "step": 20540
    },
    {
      "epoch": 0.39522650998644115,
      "grad_norm": 434.6113949594394,
      "learning_rate": 1.341335029294603e-05,
      "loss": 357.4971,
      "step": 20550
    },
    {
      "epoch": 0.395418834322201,
      "grad_norm": 441.5489979908519,
      "learning_rate": 1.3407613059367898e-05,
      "loss": 360.6357,
      "step": 20560
    },
    {
      "epoch": 0.39561115865796076,
      "grad_norm": 498.03942572073885,
      "learning_rate": 1.3401874556510182e-05,
      "loss": 361.7844,
      "step": 20570
    },
    {
      "epoch": 0.3958034829937206,
      "grad_norm": 477.9780355552977,
      "learning_rate": 1.3396134786510375e-05,
      "loss": 368.7838,
      "step": 20580
    },
    {
      "epoch": 0.3959958073294804,
      "grad_norm": 534.6909859502371,
      "learning_rate": 1.3390393751506452e-05,
      "loss": 365.4386,
      "step": 20590
    },
    {
      "epoch": 0.39618813166524025,
      "grad_norm": 459.42743548275524,
      "learning_rate": 1.338465145363685e-05,
      "loss": 354.5065,
      "step": 20600
    },
    {
      "epoch": 0.3963804560010001,
      "grad_norm": 472.3562074849193,
      "learning_rate": 1.3378907895040478e-05,
      "loss": 367.1442,
      "step": 20610
    },
    {
      "epoch": 0.3965727803367599,
      "grad_norm": 467.66253604461303,
      "learning_rate": 1.3373163077856724e-05,
      "loss": 359.797,
      "step": 20620
    },
    {
      "epoch": 0.39676510467251974,
      "grad_norm": 461.52725967280736,
      "learning_rate": 1.3367417004225429e-05,
      "loss": 367.3525,
      "step": 20630
    },
    {
      "epoch": 0.39695742900827957,
      "grad_norm": 451.3392313190913,
      "learning_rate": 1.3361669676286919e-05,
      "loss": 366.7544,
      "step": 20640
    },
    {
      "epoch": 0.3971497533440394,
      "grad_norm": 459.12627050804315,
      "learning_rate": 1.335592109618197e-05,
      "loss": 349.8331,
      "step": 20650
    },
    {
      "epoch": 0.39734207767979923,
      "grad_norm": 472.8396109907538,
      "learning_rate": 1.335017126605184e-05,
      "loss": 367.9225,
      "step": 20660
    },
    {
      "epoch": 0.39753440201555906,
      "grad_norm": 446.27922333169556,
      "learning_rate": 1.3344420188038243e-05,
      "loss": 356.1516,
      "step": 20670
    },
    {
      "epoch": 0.39772672635131884,
      "grad_norm": 453.5543158766576,
      "learning_rate": 1.333866786428336e-05,
      "loss": 365.2807,
      "step": 20680
    },
    {
      "epoch": 0.39791905068707867,
      "grad_norm": 480.12907925444244,
      "learning_rate": 1.3332914296929838e-05,
      "loss": 370.1983,
      "step": 20690
    },
    {
      "epoch": 0.3981113750228385,
      "grad_norm": 440.4908319470761,
      "learning_rate": 1.3327159488120784e-05,
      "loss": 360.8958,
      "step": 20700
    },
    {
      "epoch": 0.39830369935859833,
      "grad_norm": 478.82279115675357,
      "learning_rate": 1.3321403439999775e-05,
      "loss": 363.128,
      "step": 20710
    },
    {
      "epoch": 0.39849602369435816,
      "grad_norm": 462.46524586081466,
      "learning_rate": 1.3315646154710835e-05,
      "loss": 372.2664,
      "step": 20720
    },
    {
      "epoch": 0.398688348030118,
      "grad_norm": 483.8009497849133,
      "learning_rate": 1.3309887634398466e-05,
      "loss": 372.2889,
      "step": 20730
    },
    {
      "epoch": 0.3988806723658778,
      "grad_norm": 477.1379085372006,
      "learning_rate": 1.3304127881207614e-05,
      "loss": 357.0821,
      "step": 20740
    },
    {
      "epoch": 0.39907299670163765,
      "grad_norm": 481.4974602358291,
      "learning_rate": 1.3298366897283697e-05,
      "loss": 369.5656,
      "step": 20750
    },
    {
      "epoch": 0.3992653210373975,
      "grad_norm": 435.3978580088629,
      "learning_rate": 1.3292604684772585e-05,
      "loss": 361.2189,
      "step": 20760
    },
    {
      "epoch": 0.3994576453731573,
      "grad_norm": 440.91582015318886,
      "learning_rate": 1.3286841245820605e-05,
      "loss": 359.8629,
      "step": 20770
    },
    {
      "epoch": 0.39964996970891714,
      "grad_norm": 451.1196756325273,
      "learning_rate": 1.3281076582574548e-05,
      "loss": 359.707,
      "step": 20780
    },
    {
      "epoch": 0.3998422940446769,
      "grad_norm": 452.4582348651019,
      "learning_rate": 1.3275310697181652e-05,
      "loss": 357.2123,
      "step": 20790
    },
    {
      "epoch": 0.40003461838043675,
      "grad_norm": 452.9152479344711,
      "learning_rate": 1.3269543591789616e-05,
      "loss": 354.137,
      "step": 20800
    },
    {
      "epoch": 0.4002269427161966,
      "grad_norm": 466.0739270843239,
      "learning_rate": 1.3263775268546588e-05,
      "loss": 363.9303,
      "step": 20810
    },
    {
      "epoch": 0.4004192670519564,
      "grad_norm": 448.51023925319134,
      "learning_rate": 1.3258005729601178e-05,
      "loss": 365.6434,
      "step": 20820
    },
    {
      "epoch": 0.40061159138771624,
      "grad_norm": 476.91273009734255,
      "learning_rate": 1.325223497710244e-05,
      "loss": 369.1298,
      "step": 20830
    },
    {
      "epoch": 0.40080391572347607,
      "grad_norm": 483.803283035268,
      "learning_rate": 1.3246463013199882e-05,
      "loss": 369.1609,
      "step": 20840
    },
    {
      "epoch": 0.4009962400592359,
      "grad_norm": 576.2222722747586,
      "learning_rate": 1.3240689840043475e-05,
      "loss": 365.8833,
      "step": 20850
    },
    {
      "epoch": 0.40118856439499573,
      "grad_norm": 496.65959556817023,
      "learning_rate": 1.323491545978362e-05,
      "loss": 358.0746,
      "step": 20860
    },
    {
      "epoch": 0.40138088873075556,
      "grad_norm": 456.8910471692591,
      "learning_rate": 1.3229139874571186e-05,
      "loss": 359.431,
      "step": 20870
    },
    {
      "epoch": 0.4015732130665154,
      "grad_norm": 460.0008152653636,
      "learning_rate": 1.3223363086557477e-05,
      "loss": 355.9718,
      "step": 20880
    },
    {
      "epoch": 0.4017655374022752,
      "grad_norm": 466.65609178345323,
      "learning_rate": 1.3217585097894255e-05,
      "loss": 349.2668,
      "step": 20890
    },
    {
      "epoch": 0.40195786173803505,
      "grad_norm": 477.95348563997453,
      "learning_rate": 1.3211805910733724e-05,
      "loss": 367.3587,
      "step": 20900
    },
    {
      "epoch": 0.4021501860737948,
      "grad_norm": 524.414526216332,
      "learning_rate": 1.3206025527228535e-05,
      "loss": 359.227,
      "step": 20910
    },
    {
      "epoch": 0.40234251040955465,
      "grad_norm": 469.5417929189521,
      "learning_rate": 1.3200243949531788e-05,
      "loss": 368.1941,
      "step": 20920
    },
    {
      "epoch": 0.4025348347453145,
      "grad_norm": 470.9943123772341,
      "learning_rate": 1.3194461179797023e-05,
      "loss": 358.7273,
      "step": 20930
    },
    {
      "epoch": 0.4027271590810743,
      "grad_norm": 452.1509868117934,
      "learning_rate": 1.3188677220178225e-05,
      "loss": 345.5576,
      "step": 20940
    },
    {
      "epoch": 0.40291948341683415,
      "grad_norm": 480.1860625748449,
      "learning_rate": 1.3182892072829828e-05,
      "loss": 347.6384,
      "step": 20950
    },
    {
      "epoch": 0.403111807752594,
      "grad_norm": 529.9280626935827,
      "learning_rate": 1.3177105739906702e-05,
      "loss": 349.8616,
      "step": 20960
    },
    {
      "epoch": 0.4033041320883538,
      "grad_norm": 472.1405012673907,
      "learning_rate": 1.3171318223564156e-05,
      "loss": 361.0386,
      "step": 20970
    },
    {
      "epoch": 0.40349645642411364,
      "grad_norm": 438.29909332627955,
      "learning_rate": 1.3165529525957947e-05,
      "loss": 363.7229,
      "step": 20980
    },
    {
      "epoch": 0.40368878075987347,
      "grad_norm": 493.9356324770707,
      "learning_rate": 1.3159739649244271e-05,
      "loss": 361.6345,
      "step": 20990
    },
    {
      "epoch": 0.4038811050956333,
      "grad_norm": 484.5949545828873,
      "learning_rate": 1.3153948595579764e-05,
      "loss": 364.8096,
      "step": 21000
    },
    {
      "epoch": 0.4040734294313931,
      "grad_norm": 418.17653626818446,
      "learning_rate": 1.3148156367121491e-05,
      "loss": 368.4321,
      "step": 21010
    },
    {
      "epoch": 0.4042657537671529,
      "grad_norm": 499.708926652198,
      "learning_rate": 1.3142362966026967e-05,
      "loss": 359.458,
      "step": 21020
    },
    {
      "epoch": 0.40445807810291273,
      "grad_norm": 459.6739275932414,
      "learning_rate": 1.3136568394454136e-05,
      "loss": 362.5559,
      "step": 21030
    },
    {
      "epoch": 0.40465040243867256,
      "grad_norm": 441.3454725459896,
      "learning_rate": 1.313077265456138e-05,
      "loss": 357.4751,
      "step": 21040
    },
    {
      "epoch": 0.4048427267744324,
      "grad_norm": 482.74303749486506,
      "learning_rate": 1.3124975748507514e-05,
      "loss": 365.2877,
      "step": 21050
    },
    {
      "epoch": 0.4050350511101922,
      "grad_norm": 454.77702380243915,
      "learning_rate": 1.3119177678451793e-05,
      "loss": 371.1625,
      "step": 21060
    },
    {
      "epoch": 0.40522737544595205,
      "grad_norm": 447.1521105136589,
      "learning_rate": 1.3113378446553903e-05,
      "loss": 369.1298,
      "step": 21070
    },
    {
      "epoch": 0.4054196997817119,
      "grad_norm": 470.1238170890203,
      "learning_rate": 1.3107578054973962e-05,
      "loss": 363.8412,
      "step": 21080
    },
    {
      "epoch": 0.4056120241174717,
      "grad_norm": 449.4914150454394,
      "learning_rate": 1.3101776505872516e-05,
      "loss": 363.123,
      "step": 21090
    },
    {
      "epoch": 0.40580434845323154,
      "grad_norm": 501.1557750397434,
      "learning_rate": 1.309597380141055e-05,
      "loss": 359.8808,
      "step": 21100
    },
    {
      "epoch": 0.4059966727889914,
      "grad_norm": 491.25296606898326,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 357.7032,
      "step": 21110
    },
    {
      "epoch": 0.4061889971247512,
      "grad_norm": 464.6999900958375,
      "learning_rate": 1.3084364935051132e-05,
      "loss": 359.5759,
      "step": 21120
    },
    {
      "epoch": 0.406381321460511,
      "grad_norm": 465.8542145995899,
      "learning_rate": 1.307855877747779e-05,
      "loss": 367.1791,
      "step": 21130
    },
    {
      "epoch": 0.4065736457962708,
      "grad_norm": 467.3494052218952,
      "learning_rate": 1.307275147319215e-05,
      "loss": 350.4204,
      "step": 21140
    },
    {
      "epoch": 0.40676597013203064,
      "grad_norm": 477.22362338579916,
      "learning_rate": 1.3066943024357333e-05,
      "loss": 351.9511,
      "step": 21150
    },
    {
      "epoch": 0.40695829446779047,
      "grad_norm": 505.4582191018218,
      "learning_rate": 1.306113343313689e-05,
      "loss": 362.5992,
      "step": 21160
    },
    {
      "epoch": 0.4071506188035503,
      "grad_norm": 473.4444719359127,
      "learning_rate": 1.3055322701694801e-05,
      "loss": 369.0544,
      "step": 21170
    },
    {
      "epoch": 0.40734294313931013,
      "grad_norm": 516.8069004273458,
      "learning_rate": 1.3049510832195466e-05,
      "loss": 364.5951,
      "step": 21180
    },
    {
      "epoch": 0.40753526747506996,
      "grad_norm": 494.8109604879005,
      "learning_rate": 1.3043697826803707e-05,
      "loss": 355.1831,
      "step": 21190
    },
    {
      "epoch": 0.4077275918108298,
      "grad_norm": 442.1788140648132,
      "learning_rate": 1.303788368768478e-05,
      "loss": 347.0145,
      "step": 21200
    },
    {
      "epoch": 0.4079199161465896,
      "grad_norm": 483.60524242773596,
      "learning_rate": 1.3032068417004351e-05,
      "loss": 366.5362,
      "step": 21210
    },
    {
      "epoch": 0.40811224048234945,
      "grad_norm": 483.5920979964635,
      "learning_rate": 1.302625201692851e-05,
      "loss": 360.8435,
      "step": 21220
    },
    {
      "epoch": 0.4083045648181093,
      "grad_norm": 486.00452919303007,
      "learning_rate": 1.302043448962378e-05,
      "loss": 353.2764,
      "step": 21230
    },
    {
      "epoch": 0.40849688915386906,
      "grad_norm": 556.2820255780632,
      "learning_rate": 1.301461583725708e-05,
      "loss": 365.7417,
      "step": 21240
    },
    {
      "epoch": 0.4086892134896289,
      "grad_norm": 557.8909420988612,
      "learning_rate": 1.3008796061995772e-05,
      "loss": 363.3791,
      "step": 21250
    },
    {
      "epoch": 0.4088815378253887,
      "grad_norm": 446.95090091587434,
      "learning_rate": 1.3002975166007618e-05,
      "loss": 355.1656,
      "step": 21260
    },
    {
      "epoch": 0.40907386216114855,
      "grad_norm": 483.7531629355145,
      "learning_rate": 1.2997153151460814e-05,
      "loss": 360.441,
      "step": 21270
    },
    {
      "epoch": 0.4092661864969084,
      "grad_norm": 449.98429188400655,
      "learning_rate": 1.299133002052396e-05,
      "loss": 356.2514,
      "step": 21280
    },
    {
      "epoch": 0.4094585108326682,
      "grad_norm": 519.9671908061794,
      "learning_rate": 1.2985505775366079e-05,
      "loss": 364.107,
      "step": 21290
    },
    {
      "epoch": 0.40965083516842804,
      "grad_norm": 475.6833797538685,
      "learning_rate": 1.2979680418156604e-05,
      "loss": 370.397,
      "step": 21300
    },
    {
      "epoch": 0.40984315950418787,
      "grad_norm": 505.61185361114025,
      "learning_rate": 1.297385395106538e-05,
      "loss": 362.5435,
      "step": 21310
    },
    {
      "epoch": 0.4100354838399477,
      "grad_norm": 467.9278210641104,
      "learning_rate": 1.2968026376262679e-05,
      "loss": 345.4082,
      "step": 21320
    },
    {
      "epoch": 0.41022780817570753,
      "grad_norm": 474.7007701764091,
      "learning_rate": 1.2962197695919167e-05,
      "loss": 366.5094,
      "step": 21330
    },
    {
      "epoch": 0.41042013251146736,
      "grad_norm": 440.36632850954436,
      "learning_rate": 1.2956367912205939e-05,
      "loss": 364.4144,
      "step": 21340
    },
    {
      "epoch": 0.41061245684722714,
      "grad_norm": 439.9674708133569,
      "learning_rate": 1.2950537027294487e-05,
      "loss": 355.7068,
      "step": 21350
    },
    {
      "epoch": 0.41080478118298697,
      "grad_norm": 460.83750225291146,
      "learning_rate": 1.2944705043356722e-05,
      "loss": 359.8593,
      "step": 21360
    },
    {
      "epoch": 0.4109971055187468,
      "grad_norm": 470.44102474487124,
      "learning_rate": 1.2938871962564965e-05,
      "loss": 352.2021,
      "step": 21370
    },
    {
      "epoch": 0.4111894298545066,
      "grad_norm": 490.9972094918669,
      "learning_rate": 1.2933037787091935e-05,
      "loss": 368.0638,
      "step": 21380
    },
    {
      "epoch": 0.41138175419026646,
      "grad_norm": 488.8015397753577,
      "learning_rate": 1.2927202519110775e-05,
      "loss": 372.5835,
      "step": 21390
    },
    {
      "epoch": 0.4115740785260263,
      "grad_norm": 458.14919799764624,
      "learning_rate": 1.2921366160795017e-05,
      "loss": 358.4192,
      "step": 21400
    },
    {
      "epoch": 0.4117664028617861,
      "grad_norm": 432.9748994515109,
      "learning_rate": 1.2915528714318612e-05,
      "loss": 368.0081,
      "step": 21410
    },
    {
      "epoch": 0.41195872719754595,
      "grad_norm": 460.7771613286394,
      "learning_rate": 1.2909690181855914e-05,
      "loss": 356.3427,
      "step": 21420
    },
    {
      "epoch": 0.4121510515333058,
      "grad_norm": 505.9727688541971,
      "learning_rate": 1.290385056558168e-05,
      "loss": 360.1893,
      "step": 21430
    },
    {
      "epoch": 0.4123433758690656,
      "grad_norm": 477.4148390171022,
      "learning_rate": 1.2898009867671066e-05,
      "loss": 357.5101,
      "step": 21440
    },
    {
      "epoch": 0.41253570020482544,
      "grad_norm": 496.8967425078629,
      "learning_rate": 1.2892168090299639e-05,
      "loss": 352.5651,
      "step": 21450
    },
    {
      "epoch": 0.41272802454058527,
      "grad_norm": 475.16826605810854,
      "learning_rate": 1.2886325235643367e-05,
      "loss": 353.5881,
      "step": 21460
    },
    {
      "epoch": 0.41292034887634504,
      "grad_norm": 441.48714774515406,
      "learning_rate": 1.288048130587861e-05,
      "loss": 352.4681,
      "step": 21470
    },
    {
      "epoch": 0.4131126732121049,
      "grad_norm": 463.2352084650711,
      "learning_rate": 1.287463630318214e-05,
      "loss": 358.8107,
      "step": 21480
    },
    {
      "epoch": 0.4133049975478647,
      "grad_norm": 485.4626674830089,
      "learning_rate": 1.2868790229731123e-05,
      "loss": 359.6656,
      "step": 21490
    },
    {
      "epoch": 0.41349732188362454,
      "grad_norm": 468.374109728343,
      "learning_rate": 1.2862943087703127e-05,
      "loss": 355.3887,
      "step": 21500
    },
    {
      "epoch": 0.41368964621938437,
      "grad_norm": 451.3774990309563,
      "learning_rate": 1.2857094879276115e-05,
      "loss": 363.3622,
      "step": 21510
    },
    {
      "epoch": 0.4138819705551442,
      "grad_norm": 491.3516656098727,
      "learning_rate": 1.2851245606628447e-05,
      "loss": 359.746,
      "step": 21520
    },
    {
      "epoch": 0.414074294890904,
      "grad_norm": 473.53688745001983,
      "learning_rate": 1.2845395271938876e-05,
      "loss": 357.2325,
      "step": 21530
    },
    {
      "epoch": 0.41426661922666386,
      "grad_norm": 511.927983187308,
      "learning_rate": 1.2839543877386562e-05,
      "loss": 351.3271,
      "step": 21540
    },
    {
      "epoch": 0.4144589435624237,
      "grad_norm": 449.0944658901166,
      "learning_rate": 1.283369142515105e-05,
      "loss": 361.6248,
      "step": 21550
    },
    {
      "epoch": 0.4146512678981835,
      "grad_norm": 465.4977112841231,
      "learning_rate": 1.282783791741228e-05,
      "loss": 370.761,
      "step": 21560
    },
    {
      "epoch": 0.41484359223394335,
      "grad_norm": 5484.992293667798,
      "learning_rate": 1.2821983356350593e-05,
      "loss": 380.6152,
      "step": 21570
    },
    {
      "epoch": 0.4150359165697031,
      "grad_norm": 471.3213825403217,
      "learning_rate": 1.2816127744146711e-05,
      "loss": 359.9935,
      "step": 21580
    },
    {
      "epoch": 0.41522824090546295,
      "grad_norm": 447.56332457228893,
      "learning_rate": 1.2810271082981755e-05,
      "loss": 359.3304,
      "step": 21590
    },
    {
      "epoch": 0.4154205652412228,
      "grad_norm": 498.61162266238523,
      "learning_rate": 1.2804413375037232e-05,
      "loss": 351.4054,
      "step": 21600
    },
    {
      "epoch": 0.4156128895769826,
      "grad_norm": 490.62680887650953,
      "learning_rate": 1.2798554622495042e-05,
      "loss": 357.0392,
      "step": 21610
    },
    {
      "epoch": 0.41580521391274244,
      "grad_norm": 455.8535222343372,
      "learning_rate": 1.2792694827537477e-05,
      "loss": 362.5506,
      "step": 21620
    },
    {
      "epoch": 0.4159975382485023,
      "grad_norm": 486.23131546873833,
      "learning_rate": 1.2786833992347212e-05,
      "loss": 356.4791,
      "step": 21630
    },
    {
      "epoch": 0.4161898625842621,
      "grad_norm": 432.2308869162765,
      "learning_rate": 1.2780972119107312e-05,
      "loss": 361.85,
      "step": 21640
    },
    {
      "epoch": 0.41638218692002193,
      "grad_norm": 436.94642001588204,
      "learning_rate": 1.2775109210001225e-05,
      "loss": 349.1665,
      "step": 21650
    },
    {
      "epoch": 0.41657451125578177,
      "grad_norm": 516.9578667716966,
      "learning_rate": 1.2769245267212789e-05,
      "loss": 357.1654,
      "step": 21660
    },
    {
      "epoch": 0.4167668355915416,
      "grad_norm": 483.2205202416933,
      "learning_rate": 1.2763380292926227e-05,
      "loss": 356.3035,
      "step": 21670
    },
    {
      "epoch": 0.4169591599273014,
      "grad_norm": 451.76594064335154,
      "learning_rate": 1.2757514289326146e-05,
      "loss": 353.5011,
      "step": 21680
    },
    {
      "epoch": 0.4171514842630612,
      "grad_norm": 461.48383791289245,
      "learning_rate": 1.2751647258597531e-05,
      "loss": 358.6389,
      "step": 21690
    },
    {
      "epoch": 0.41734380859882103,
      "grad_norm": 481.68683753036305,
      "learning_rate": 1.274577920292576e-05,
      "loss": 346.4288,
      "step": 21700
    },
    {
      "epoch": 0.41753613293458086,
      "grad_norm": 440.34213962332115,
      "learning_rate": 1.2739910124496585e-05,
      "loss": 361.8271,
      "step": 21710
    },
    {
      "epoch": 0.4177284572703407,
      "grad_norm": 464.86968818352574,
      "learning_rate": 1.2734040025496137e-05,
      "loss": 352.9531,
      "step": 21720
    },
    {
      "epoch": 0.4179207816061005,
      "grad_norm": 462.0077211343192,
      "learning_rate": 1.2728168908110937e-05,
      "loss": 355.3349,
      "step": 21730
    },
    {
      "epoch": 0.41811310594186035,
      "grad_norm": 436.0814640607625,
      "learning_rate": 1.2722296774527871e-05,
      "loss": 347.6623,
      "step": 21740
    },
    {
      "epoch": 0.4183054302776202,
      "grad_norm": 432.97812381628853,
      "learning_rate": 1.271642362693422e-05,
      "loss": 367.2855,
      "step": 21750
    },
    {
      "epoch": 0.41849775461338,
      "grad_norm": 538.6227091185007,
      "learning_rate": 1.2710549467517628e-05,
      "loss": 353.0426,
      "step": 21760
    },
    {
      "epoch": 0.41869007894913984,
      "grad_norm": 443.31428722031995,
      "learning_rate": 1.270467429846613e-05,
      "loss": 340.3564,
      "step": 21770
    },
    {
      "epoch": 0.4188824032848997,
      "grad_norm": 431.40700391780126,
      "learning_rate": 1.2698798121968122e-05,
      "loss": 352.4626,
      "step": 21780
    },
    {
      "epoch": 0.4190747276206595,
      "grad_norm": 457.55028204982324,
      "learning_rate": 1.2692920940212387e-05,
      "loss": 366.8255,
      "step": 21790
    },
    {
      "epoch": 0.4192670519564193,
      "grad_norm": 445.569704287574,
      "learning_rate": 1.2687042755388077e-05,
      "loss": 347.9427,
      "step": 21800
    },
    {
      "epoch": 0.4194593762921791,
      "grad_norm": 428.7648916948125,
      "learning_rate": 1.2681163569684718e-05,
      "loss": 359.4237,
      "step": 21810
    },
    {
      "epoch": 0.41965170062793894,
      "grad_norm": 437.35057059980613,
      "learning_rate": 1.2675283385292212e-05,
      "loss": 360.2542,
      "step": 21820
    },
    {
      "epoch": 0.41984402496369877,
      "grad_norm": 488.98922795721256,
      "learning_rate": 1.2669402204400825e-05,
      "loss": 350.4583,
      "step": 21830
    },
    {
      "epoch": 0.4200363492994586,
      "grad_norm": 495.8385665928512,
      "learning_rate": 1.2663520029201206e-05,
      "loss": 347.4213,
      "step": 21840
    },
    {
      "epoch": 0.42022867363521843,
      "grad_norm": 551.2231368701516,
      "learning_rate": 1.2657636861884363e-05,
      "loss": 352.0523,
      "step": 21850
    },
    {
      "epoch": 0.42042099797097826,
      "grad_norm": 475.12387201222555,
      "learning_rate": 1.2651752704641686e-05,
      "loss": 356.7493,
      "step": 21860
    },
    {
      "epoch": 0.4206133223067381,
      "grad_norm": 483.72055523352617,
      "learning_rate": 1.2645867559664918e-05,
      "loss": 358.9258,
      "step": 21870
    },
    {
      "epoch": 0.4208056466424979,
      "grad_norm": 422.2589474800384,
      "learning_rate": 1.2639981429146184e-05,
      "loss": 346.4987,
      "step": 21880
    },
    {
      "epoch": 0.42099797097825775,
      "grad_norm": 532.9596887838914,
      "learning_rate": 1.2634094315277967e-05,
      "loss": 352.5222,
      "step": 21890
    },
    {
      "epoch": 0.4211902953140176,
      "grad_norm": 482.06855643493924,
      "learning_rate": 1.262820622025312e-05,
      "loss": 363.8119,
      "step": 21900
    },
    {
      "epoch": 0.42138261964977736,
      "grad_norm": 521.8497325382895,
      "learning_rate": 1.2622317146264864e-05,
      "loss": 361.689,
      "step": 21910
    },
    {
      "epoch": 0.4215749439855372,
      "grad_norm": 490.2268848417426,
      "learning_rate": 1.2616427095506779e-05,
      "loss": 358.514,
      "step": 21920
    },
    {
      "epoch": 0.421767268321297,
      "grad_norm": 422.1126332182719,
      "learning_rate": 1.2610536070172815e-05,
      "loss": 359.9763,
      "step": 21930
    },
    {
      "epoch": 0.42195959265705685,
      "grad_norm": 483.16607311328096,
      "learning_rate": 1.2604644072457277e-05,
      "loss": 351.9336,
      "step": 21940
    },
    {
      "epoch": 0.4221519169928167,
      "grad_norm": 486.5645570245076,
      "learning_rate": 1.259875110455484e-05,
      "loss": 357.5309,
      "step": 21950
    },
    {
      "epoch": 0.4223442413285765,
      "grad_norm": 434.6644044157051,
      "learning_rate": 1.2592857168660535e-05,
      "loss": 356.7706,
      "step": 21960
    },
    {
      "epoch": 0.42253656566433634,
      "grad_norm": 475.64407494901366,
      "learning_rate": 1.2586962266969758e-05,
      "loss": 359.2397,
      "step": 21970
    },
    {
      "epoch": 0.42272889000009617,
      "grad_norm": 544.7920515048304,
      "learning_rate": 1.2581066401678261e-05,
      "loss": 371.0822,
      "step": 21980
    },
    {
      "epoch": 0.422921214335856,
      "grad_norm": 456.6653580156985,
      "learning_rate": 1.2575169574982158e-05,
      "loss": 353.69,
      "step": 21990
    },
    {
      "epoch": 0.42311353867161583,
      "grad_norm": 413.937897057499,
      "learning_rate": 1.2569271789077919e-05,
      "loss": 358.3366,
      "step": 22000
    },
    {
      "epoch": 0.42330586300737566,
      "grad_norm": 464.0643570938417,
      "learning_rate": 1.256337304616237e-05,
      "loss": 348.7437,
      "step": 22010
    },
    {
      "epoch": 0.4234981873431355,
      "grad_norm": 483.55449784929607,
      "learning_rate": 1.25574733484327e-05,
      "loss": 352.3562,
      "step": 22020
    },
    {
      "epoch": 0.42369051167889527,
      "grad_norm": 471.6017174578337,
      "learning_rate": 1.2551572698086446e-05,
      "loss": 361.4498,
      "step": 22030
    },
    {
      "epoch": 0.4238828360146551,
      "grad_norm": 452.1224267963627,
      "learning_rate": 1.2545671097321497e-05,
      "loss": 355.5773,
      "step": 22040
    },
    {
      "epoch": 0.4240751603504149,
      "grad_norm": 459.1790605332593,
      "learning_rate": 1.2539768548336112e-05,
      "loss": 356.8583,
      "step": 22050
    },
    {
      "epoch": 0.42426748468617476,
      "grad_norm": 510.35860960140946,
      "learning_rate": 1.2533865053328886e-05,
      "loss": 354.9296,
      "step": 22060
    },
    {
      "epoch": 0.4244598090219346,
      "grad_norm": 461.7441282465687,
      "learning_rate": 1.2527960614498778e-05,
      "loss": 361.6722,
      "step": 22070
    },
    {
      "epoch": 0.4246521333576944,
      "grad_norm": 474.9706255978627,
      "learning_rate": 1.252205523404509e-05,
      "loss": 362.0148,
      "step": 22080
    },
    {
      "epoch": 0.42484445769345425,
      "grad_norm": 454.50327474860177,
      "learning_rate": 1.2516148914167481e-05,
      "loss": 361.5063,
      "step": 22090
    },
    {
      "epoch": 0.4250367820292141,
      "grad_norm": 470.22588490796045,
      "learning_rate": 1.2510241657065958e-05,
      "loss": 346.1036,
      "step": 22100
    },
    {
      "epoch": 0.4252291063649739,
      "grad_norm": 450.4961956923022,
      "learning_rate": 1.2504333464940874e-05,
      "loss": 347.454,
      "step": 22110
    },
    {
      "epoch": 0.42542143070073374,
      "grad_norm": 1166.5237137262181,
      "learning_rate": 1.2498424339992934e-05,
      "loss": 353.2138,
      "step": 22120
    },
    {
      "epoch": 0.42561375503649357,
      "grad_norm": 489.17746331812305,
      "learning_rate": 1.2492514284423194e-05,
      "loss": 356.392,
      "step": 22130
    },
    {
      "epoch": 0.42580607937225334,
      "grad_norm": 477.7647586631091,
      "learning_rate": 1.2486603300433045e-05,
      "loss": 353.9992,
      "step": 22140
    },
    {
      "epoch": 0.4259984037080132,
      "grad_norm": 517.421499289549,
      "learning_rate": 1.2480691390224238e-05,
      "loss": 360.8063,
      "step": 22150
    },
    {
      "epoch": 0.426190728043773,
      "grad_norm": 459.9200542352227,
      "learning_rate": 1.2474778555998855e-05,
      "loss": 341.3725,
      "step": 22160
    },
    {
      "epoch": 0.42638305237953283,
      "grad_norm": 465.6041733316549,
      "learning_rate": 1.246886479995933e-05,
      "loss": 362.1571,
      "step": 22170
    },
    {
      "epoch": 0.42657537671529266,
      "grad_norm": 491.530544993616,
      "learning_rate": 1.2462950124308444e-05,
      "loss": 358.6839,
      "step": 22180
    },
    {
      "epoch": 0.4267677010510525,
      "grad_norm": 511.6491291065285,
      "learning_rate": 1.2457034531249313e-05,
      "loss": 352.3591,
      "step": 22190
    },
    {
      "epoch": 0.4269600253868123,
      "grad_norm": 464.55501962277793,
      "learning_rate": 1.2451118022985402e-05,
      "loss": 353.4329,
      "step": 22200
    },
    {
      "epoch": 0.42715234972257216,
      "grad_norm": 464.2570395726874,
      "learning_rate": 1.2445200601720504e-05,
      "loss": 346.6129,
      "step": 22210
    },
    {
      "epoch": 0.427344674058332,
      "grad_norm": 451.3595060238119,
      "learning_rate": 1.243928226965877e-05,
      "loss": 353.174,
      "step": 22220
    },
    {
      "epoch": 0.4275369983940918,
      "grad_norm": 428.8214651667687,
      "learning_rate": 1.2433363029004675e-05,
      "loss": 353.4309,
      "step": 22230
    },
    {
      "epoch": 0.42772932272985165,
      "grad_norm": 447.660059888906,
      "learning_rate": 1.2427442881963042e-05,
      "loss": 353.6233,
      "step": 22240
    },
    {
      "epoch": 0.4279216470656114,
      "grad_norm": 428.85548521373573,
      "learning_rate": 1.242152183073902e-05,
      "loss": 357.1601,
      "step": 22250
    },
    {
      "epoch": 0.42811397140137125,
      "grad_norm": 450.36372940478407,
      "learning_rate": 1.2415599877538111e-05,
      "loss": 345.0228,
      "step": 22260
    },
    {
      "epoch": 0.4283062957371311,
      "grad_norm": 458.84300925126934,
      "learning_rate": 1.2409677024566145e-05,
      "loss": 365.5294,
      "step": 22270
    },
    {
      "epoch": 0.4284986200728909,
      "grad_norm": 487.5794962862781,
      "learning_rate": 1.2403753274029281e-05,
      "loss": 359.2313,
      "step": 22280
    },
    {
      "epoch": 0.42869094440865074,
      "grad_norm": 434.0123600246192,
      "learning_rate": 1.2397828628134028e-05,
      "loss": 345.9444,
      "step": 22290
    },
    {
      "epoch": 0.4288832687444106,
      "grad_norm": 455.26191386780675,
      "learning_rate": 1.2391903089087208e-05,
      "loss": 353.5667,
      "step": 22300
    },
    {
      "epoch": 0.4290755930801704,
      "grad_norm": 478.61224251356435,
      "learning_rate": 1.2385976659095993e-05,
      "loss": 353.1015,
      "step": 22310
    },
    {
      "epoch": 0.42926791741593023,
      "grad_norm": 442.75037662501694,
      "learning_rate": 1.2380049340367876e-05,
      "loss": 347.4218,
      "step": 22320
    },
    {
      "epoch": 0.42946024175169006,
      "grad_norm": 478.25701038187225,
      "learning_rate": 1.2374121135110688e-05,
      "loss": 367.8834,
      "step": 22330
    },
    {
      "epoch": 0.4296525660874499,
      "grad_norm": 431.9551326904518,
      "learning_rate": 1.2368192045532586e-05,
      "loss": 354.8841,
      "step": 22340
    },
    {
      "epoch": 0.4298448904232097,
      "grad_norm": 433.5347825432444,
      "learning_rate": 1.236226207384206e-05,
      "loss": 351.6588,
      "step": 22350
    },
    {
      "epoch": 0.4300372147589695,
      "grad_norm": 474.89242289033746,
      "learning_rate": 1.2356331222247929e-05,
      "loss": 359.1791,
      "step": 22360
    },
    {
      "epoch": 0.43022953909472933,
      "grad_norm": 467.42679677909086,
      "learning_rate": 1.2350399492959328e-05,
      "loss": 362.0955,
      "step": 22370
    },
    {
      "epoch": 0.43042186343048916,
      "grad_norm": 469.85750400246616,
      "learning_rate": 1.234446688818574e-05,
      "loss": 356.8853,
      "step": 22380
    },
    {
      "epoch": 0.430614187766249,
      "grad_norm": 461.90255826053743,
      "learning_rate": 1.233853341013695e-05,
      "loss": 352.9148,
      "step": 22390
    },
    {
      "epoch": 0.4308065121020088,
      "grad_norm": 449.86036415667627,
      "learning_rate": 1.233259906102309e-05,
      "loss": 352.1118,
      "step": 22400
    },
    {
      "epoch": 0.43099883643776865,
      "grad_norm": 441.29320721249115,
      "learning_rate": 1.2326663843054603e-05,
      "loss": 350.8514,
      "step": 22410
    },
    {
      "epoch": 0.4311911607735285,
      "grad_norm": 443.4054153774575,
      "learning_rate": 1.2320727758442264e-05,
      "loss": 358.9528,
      "step": 22420
    },
    {
      "epoch": 0.4313834851092883,
      "grad_norm": 464.48785147118207,
      "learning_rate": 1.231479080939716e-05,
      "loss": 364.3797,
      "step": 22430
    },
    {
      "epoch": 0.43157580944504814,
      "grad_norm": 412.516768195798,
      "learning_rate": 1.230885299813071e-05,
      "loss": 363.6567,
      "step": 22440
    },
    {
      "epoch": 0.43176813378080797,
      "grad_norm": 495.99355667153515,
      "learning_rate": 1.2302914326854651e-05,
      "loss": 356.0047,
      "step": 22450
    },
    {
      "epoch": 0.4319604581165678,
      "grad_norm": 473.1165962262813,
      "learning_rate": 1.2296974797781036e-05,
      "loss": 364.2631,
      "step": 22460
    },
    {
      "epoch": 0.43215278245232763,
      "grad_norm": 470.9665452868922,
      "learning_rate": 1.2291034413122247e-05,
      "loss": 368.466,
      "step": 22470
    },
    {
      "epoch": 0.4323451067880874,
      "grad_norm": 477.7097465613152,
      "learning_rate": 1.2285093175090976e-05,
      "loss": 354.1781,
      "step": 22480
    },
    {
      "epoch": 0.43253743112384724,
      "grad_norm": 487.1634400935877,
      "learning_rate": 1.227915108590024e-05,
      "loss": 348.4954,
      "step": 22490
    },
    {
      "epoch": 0.43272975545960707,
      "grad_norm": 438.2995394209405,
      "learning_rate": 1.2273208147763363e-05,
      "loss": 342.7937,
      "step": 22500
    },
    {
      "epoch": 0.4329220797953669,
      "grad_norm": 483.84122120624323,
      "learning_rate": 1.2267264362893995e-05,
      "loss": 356.5782,
      "step": 22510
    },
    {
      "epoch": 0.43311440413112673,
      "grad_norm": 452.2892054335234,
      "learning_rate": 1.2261319733506096e-05,
      "loss": 355.2324,
      "step": 22520
    },
    {
      "epoch": 0.43330672846688656,
      "grad_norm": 470.97288871365726,
      "learning_rate": 1.2255374261813944e-05,
      "loss": 362.2226,
      "step": 22530
    },
    {
      "epoch": 0.4334990528026464,
      "grad_norm": 451.9701225379797,
      "learning_rate": 1.2249427950032127e-05,
      "loss": 358.8834,
      "step": 22540
    },
    {
      "epoch": 0.4336913771384062,
      "grad_norm": 527.6292859766575,
      "learning_rate": 1.224348080037555e-05,
      "loss": 367.1118,
      "step": 22550
    },
    {
      "epoch": 0.43388370147416605,
      "grad_norm": 481.7096752559559,
      "learning_rate": 1.2237532815059427e-05,
      "loss": 351.9961,
      "step": 22560
    },
    {
      "epoch": 0.4340760258099259,
      "grad_norm": 440.6525927364171,
      "learning_rate": 1.2231583996299285e-05,
      "loss": 351.348,
      "step": 22570
    },
    {
      "epoch": 0.4342683501456857,
      "grad_norm": 459.4464599866437,
      "learning_rate": 1.2225634346310962e-05,
      "loss": 349.3989,
      "step": 22580
    },
    {
      "epoch": 0.4344606744814455,
      "grad_norm": 434.8497012594701,
      "learning_rate": 1.22196838673106e-05,
      "loss": 359.4571,
      "step": 22590
    },
    {
      "epoch": 0.4346529988172053,
      "grad_norm": 473.9689727384777,
      "learning_rate": 1.2213732561514657e-05,
      "loss": 361.0863,
      "step": 22600
    },
    {
      "epoch": 0.43484532315296515,
      "grad_norm": 429.20142883517275,
      "learning_rate": 1.2207780431139894e-05,
      "loss": 358.7701,
      "step": 22610
    },
    {
      "epoch": 0.435037647488725,
      "grad_norm": 496.32285337930824,
      "learning_rate": 1.2201827478403385e-05,
      "loss": 352.4951,
      "step": 22620
    },
    {
      "epoch": 0.4352299718244848,
      "grad_norm": 409.46571043219234,
      "learning_rate": 1.2195873705522508e-05,
      "loss": 339.3425,
      "step": 22630
    },
    {
      "epoch": 0.43542229616024464,
      "grad_norm": 444.3838341019518,
      "learning_rate": 1.2189919114714936e-05,
      "loss": 341.6629,
      "step": 22640
    },
    {
      "epoch": 0.43561462049600447,
      "grad_norm": 464.2398396802652,
      "learning_rate": 1.2183963708198668e-05,
      "loss": 365.7007,
      "step": 22650
    },
    {
      "epoch": 0.4358069448317643,
      "grad_norm": 499.83234232424223,
      "learning_rate": 1.2178007488191983e-05,
      "loss": 350.4901,
      "step": 22660
    },
    {
      "epoch": 0.43599926916752413,
      "grad_norm": 442.55899736757885,
      "learning_rate": 1.2172050456913482e-05,
      "loss": 358.6078,
      "step": 22670
    },
    {
      "epoch": 0.43619159350328396,
      "grad_norm": 462.1902943855575,
      "learning_rate": 1.2166092616582055e-05,
      "loss": 342.5886,
      "step": 22680
    },
    {
      "epoch": 0.4363839178390438,
      "grad_norm": 549.4000100993889,
      "learning_rate": 1.2160133969416903e-05,
      "loss": 364.8587,
      "step": 22690
    },
    {
      "epoch": 0.43657624217480356,
      "grad_norm": 432.36243197849393,
      "learning_rate": 1.2154174517637526e-05,
      "loss": 350.1362,
      "step": 22700
    },
    {
      "epoch": 0.4367685665105634,
      "grad_norm": 458.9124051394236,
      "learning_rate": 1.2148214263463718e-05,
      "loss": 343.0685,
      "step": 22710
    },
    {
      "epoch": 0.4369608908463232,
      "grad_norm": 448.1825617633941,
      "learning_rate": 1.2142253209115577e-05,
      "loss": 352.7216,
      "step": 22720
    },
    {
      "epoch": 0.43715321518208305,
      "grad_norm": 439.7398588746751,
      "learning_rate": 1.2136291356813494e-05,
      "loss": 364.9622,
      "step": 22730
    },
    {
      "epoch": 0.4373455395178429,
      "grad_norm": 519.7291884971663,
      "learning_rate": 1.2130328708778162e-05,
      "loss": 352.5206,
      "step": 22740
    },
    {
      "epoch": 0.4375378638536027,
      "grad_norm": 444.6701436437316,
      "learning_rate": 1.2124365267230571e-05,
      "loss": 357.2035,
      "step": 22750
    },
    {
      "epoch": 0.43773018818936255,
      "grad_norm": 480.83404938359894,
      "learning_rate": 1.2118401034392003e-05,
      "loss": 357.8444,
      "step": 22760
    },
    {
      "epoch": 0.4379225125251224,
      "grad_norm": 431.38970168151576,
      "learning_rate": 1.2112436012484035e-05,
      "loss": 347.8432,
      "step": 22770
    },
    {
      "epoch": 0.4381148368608822,
      "grad_norm": 450.4498394827926,
      "learning_rate": 1.210647020372854e-05,
      "loss": 357.4706,
      "step": 22780
    },
    {
      "epoch": 0.43830716119664204,
      "grad_norm": 454.58896351006905,
      "learning_rate": 1.2100503610347686e-05,
      "loss": 351.4655,
      "step": 22790
    },
    {
      "epoch": 0.43849948553240187,
      "grad_norm": 464.49865169827825,
      "learning_rate": 1.2094536234563927e-05,
      "loss": 357.5448,
      "step": 22800
    },
    {
      "epoch": 0.43869180986816164,
      "grad_norm": 454.9531691770624,
      "learning_rate": 1.2088568078600013e-05,
      "loss": 353.4185,
      "step": 22810
    },
    {
      "epoch": 0.43888413420392147,
      "grad_norm": 491.9449169806258,
      "learning_rate": 1.2082599144678983e-05,
      "loss": 361.997,
      "step": 22820
    },
    {
      "epoch": 0.4390764585396813,
      "grad_norm": 459.65266714544276,
      "learning_rate": 1.2076629435024168e-05,
      "loss": 357.8603,
      "step": 22830
    },
    {
      "epoch": 0.43926878287544113,
      "grad_norm": 434.1950049394593,
      "learning_rate": 1.2070658951859183e-05,
      "loss": 367.5184,
      "step": 22840
    },
    {
      "epoch": 0.43946110721120096,
      "grad_norm": 444.2003240155268,
      "learning_rate": 1.2064687697407939e-05,
      "loss": 368.1171,
      "step": 22850
    },
    {
      "epoch": 0.4396534315469608,
      "grad_norm": 462.3118097966811,
      "learning_rate": 1.2058715673894625e-05,
      "loss": 347.1098,
      "step": 22860
    },
    {
      "epoch": 0.4398457558827206,
      "grad_norm": 479.1257367580937,
      "learning_rate": 1.2052742883543724e-05,
      "loss": 369.2424,
      "step": 22870
    },
    {
      "epoch": 0.44003808021848045,
      "grad_norm": 475.7591165528743,
      "learning_rate": 1.2046769328580004e-05,
      "loss": 346.5411,
      "step": 22880
    },
    {
      "epoch": 0.4402304045542403,
      "grad_norm": 439.72942995054444,
      "learning_rate": 1.204079501122851e-05,
      "loss": 354.0866,
      "step": 22890
    },
    {
      "epoch": 0.4404227288900001,
      "grad_norm": 464.5375736387472,
      "learning_rate": 1.2034819933714576e-05,
      "loss": 351.7987,
      "step": 22900
    },
    {
      "epoch": 0.44061505322575995,
      "grad_norm": 476.52778047454865,
      "learning_rate": 1.2028844098263827e-05,
      "loss": 357.7463,
      "step": 22910
    },
    {
      "epoch": 0.4408073775615197,
      "grad_norm": 490.0229149954796,
      "learning_rate": 1.2022867507102159e-05,
      "loss": 360.2738,
      "step": 22920
    },
    {
      "epoch": 0.44099970189727955,
      "grad_norm": 436.4063805346974,
      "learning_rate": 1.2016890162455752e-05,
      "loss": 352.7715,
      "step": 22930
    },
    {
      "epoch": 0.4411920262330394,
      "grad_norm": 435.8970063266725,
      "learning_rate": 1.2010912066551072e-05,
      "loss": 348.4483,
      "step": 22940
    },
    {
      "epoch": 0.4413843505687992,
      "grad_norm": 435.64221107343036,
      "learning_rate": 1.2004933221614854e-05,
      "loss": 346.5264,
      "step": 22950
    },
    {
      "epoch": 0.44157667490455904,
      "grad_norm": 516.5056519694289,
      "learning_rate": 1.1998953629874126e-05,
      "loss": 347.5064,
      "step": 22960
    },
    {
      "epoch": 0.44176899924031887,
      "grad_norm": 461.9243961115514,
      "learning_rate": 1.1992973293556184e-05,
      "loss": 363.5012,
      "step": 22970
    },
    {
      "epoch": 0.4419613235760787,
      "grad_norm": 473.5388989045999,
      "learning_rate": 1.1986992214888607e-05,
      "loss": 361.0567,
      "step": 22980
    },
    {
      "epoch": 0.44215364791183853,
      "grad_norm": 466.29663227564714,
      "learning_rate": 1.1981010396099244e-05,
      "loss": 353.8068,
      "step": 22990
    },
    {
      "epoch": 0.44234597224759836,
      "grad_norm": 505.03568454371253,
      "learning_rate": 1.1975027839416227e-05,
      "loss": 354.4532,
      "step": 23000
    },
    {
      "epoch": 0.4425382965833582,
      "grad_norm": 459.8080569670777,
      "learning_rate": 1.1969044547067961e-05,
      "loss": 350.0083,
      "step": 23010
    },
    {
      "epoch": 0.442730620919118,
      "grad_norm": 455.4227014655408,
      "learning_rate": 1.1963060521283117e-05,
      "loss": 346.3249,
      "step": 23020
    },
    {
      "epoch": 0.44292294525487785,
      "grad_norm": 474.5281833041822,
      "learning_rate": 1.1957075764290652e-05,
      "loss": 350.3306,
      "step": 23030
    },
    {
      "epoch": 0.44311526959063763,
      "grad_norm": 422.75569934678424,
      "learning_rate": 1.1951090278319784e-05,
      "loss": 363.0249,
      "step": 23040
    },
    {
      "epoch": 0.44330759392639746,
      "grad_norm": 443.50751408249164,
      "learning_rate": 1.1945104065600013e-05,
      "loss": 345.0779,
      "step": 23050
    },
    {
      "epoch": 0.4434999182621573,
      "grad_norm": 445.4630900307036,
      "learning_rate": 1.1939117128361101e-05,
      "loss": 352.458,
      "step": 23060
    },
    {
      "epoch": 0.4436922425979171,
      "grad_norm": 1081.5978111395164,
      "learning_rate": 1.1933129468833087e-05,
      "loss": 345.2083,
      "step": 23070
    },
    {
      "epoch": 0.44388456693367695,
      "grad_norm": 447.04742239637517,
      "learning_rate": 1.192714108924627e-05,
      "loss": 347.1263,
      "step": 23080
    },
    {
      "epoch": 0.4440768912694368,
      "grad_norm": 510.0560859588336,
      "learning_rate": 1.1921151991831225e-05,
      "loss": 356.1168,
      "step": 23090
    },
    {
      "epoch": 0.4442692156051966,
      "grad_norm": 449.24248333120875,
      "learning_rate": 1.1915162178818793e-05,
      "loss": 361.5072,
      "step": 23100
    },
    {
      "epoch": 0.44446153994095644,
      "grad_norm": 461.5912139116939,
      "learning_rate": 1.1909171652440079e-05,
      "loss": 355.7509,
      "step": 23110
    },
    {
      "epoch": 0.44465386427671627,
      "grad_norm": 448.03856233600294,
      "learning_rate": 1.1903180414926457e-05,
      "loss": 352.4296,
      "step": 23120
    },
    {
      "epoch": 0.4448461886124761,
      "grad_norm": 452.9595065196257,
      "learning_rate": 1.189718846850956e-05,
      "loss": 346.3238,
      "step": 23130
    },
    {
      "epoch": 0.44503851294823593,
      "grad_norm": 477.2244147339225,
      "learning_rate": 1.1891195815421293e-05,
      "loss": 336.5658,
      "step": 23140
    },
    {
      "epoch": 0.4452308372839957,
      "grad_norm": 478.1351204430414,
      "learning_rate": 1.1885202457893819e-05,
      "loss": 353.1867,
      "step": 23150
    },
    {
      "epoch": 0.44542316161975554,
      "grad_norm": 458.6464189651247,
      "learning_rate": 1.1879208398159563e-05,
      "loss": 354.6019,
      "step": 23160
    },
    {
      "epoch": 0.44561548595551537,
      "grad_norm": 421.5239837489341,
      "learning_rate": 1.1873213638451215e-05,
      "loss": 341.7047,
      "step": 23170
    },
    {
      "epoch": 0.4458078102912752,
      "grad_norm": 452.2073834229528,
      "learning_rate": 1.1867218181001725e-05,
      "loss": 355.6551,
      "step": 23180
    },
    {
      "epoch": 0.446000134627035,
      "grad_norm": 570.7736551151988,
      "learning_rate": 1.1861222028044301e-05,
      "loss": 362.1008,
      "step": 23190
    },
    {
      "epoch": 0.44619245896279486,
      "grad_norm": 458.37915504573186,
      "learning_rate": 1.1855225181812408e-05,
      "loss": 345.1014,
      "step": 23200
    },
    {
      "epoch": 0.4463847832985547,
      "grad_norm": 436.68572876331586,
      "learning_rate": 1.1849227644539781e-05,
      "loss": 346.1532,
      "step": 23210
    },
    {
      "epoch": 0.4465771076343145,
      "grad_norm": 440.556803127086,
      "learning_rate": 1.1843229418460391e-05,
      "loss": 345.0815,
      "step": 23220
    },
    {
      "epoch": 0.44676943197007435,
      "grad_norm": 432.7037947552592,
      "learning_rate": 1.1837230505808485e-05,
      "loss": 344.6313,
      "step": 23230
    },
    {
      "epoch": 0.4469617563058342,
      "grad_norm": 504.46737785188816,
      "learning_rate": 1.1831230908818563e-05,
      "loss": 345.6509,
      "step": 23240
    },
    {
      "epoch": 0.447154080641594,
      "grad_norm": 496.82318608976516,
      "learning_rate": 1.1825230629725366e-05,
      "loss": 358.9157,
      "step": 23250
    },
    {
      "epoch": 0.4473464049773538,
      "grad_norm": 438.07573626612026,
      "learning_rate": 1.1819229670763908e-05,
      "loss": 355.3396,
      "step": 23260
    },
    {
      "epoch": 0.4475387293131136,
      "grad_norm": 526.8128568267338,
      "learning_rate": 1.1813228034169442e-05,
      "loss": 363.1398,
      "step": 23270
    },
    {
      "epoch": 0.44773105364887344,
      "grad_norm": 438.2688938757383,
      "learning_rate": 1.180722572217748e-05,
      "loss": 354.8272,
      "step": 23280
    },
    {
      "epoch": 0.4479233779846333,
      "grad_norm": 446.2383070355474,
      "learning_rate": 1.1801222737023783e-05,
      "loss": 355.2811,
      "step": 23290
    },
    {
      "epoch": 0.4481157023203931,
      "grad_norm": 432.95901976838326,
      "learning_rate": 1.1795219080944366e-05,
      "loss": 351.8697,
      "step": 23300
    },
    {
      "epoch": 0.44830802665615294,
      "grad_norm": 426.69914460236123,
      "learning_rate": 1.178921475617549e-05,
      "loss": 352.2932,
      "step": 23310
    },
    {
      "epoch": 0.44850035099191277,
      "grad_norm": 542.3162635281931,
      "learning_rate": 1.1783209764953666e-05,
      "loss": 357.4332,
      "step": 23320
    },
    {
      "epoch": 0.4486926753276726,
      "grad_norm": 482.6538928496427,
      "learning_rate": 1.1777204109515653e-05,
      "loss": 362.0119,
      "step": 23330
    },
    {
      "epoch": 0.4488849996634324,
      "grad_norm": 474.0828419038547,
      "learning_rate": 1.1771197792098465e-05,
      "loss": 352.2589,
      "step": 23340
    },
    {
      "epoch": 0.44907732399919226,
      "grad_norm": 419.41728830651,
      "learning_rate": 1.1765190814939351e-05,
      "loss": 351.6807,
      "step": 23350
    },
    {
      "epoch": 0.4492696483349521,
      "grad_norm": 519.4215891522969,
      "learning_rate": 1.175918318027581e-05,
      "loss": 344.8265,
      "step": 23360
    },
    {
      "epoch": 0.44946197267071186,
      "grad_norm": 468.4725018251048,
      "learning_rate": 1.1753174890345591e-05,
      "loss": 352.2972,
      "step": 23370
    },
    {
      "epoch": 0.4496542970064717,
      "grad_norm": 479.12714410744263,
      "learning_rate": 1.174716594738668e-05,
      "loss": 355.5,
      "step": 23380
    },
    {
      "epoch": 0.4498466213422315,
      "grad_norm": 491.0914008434106,
      "learning_rate": 1.1741156353637304e-05,
      "loss": 343.2382,
      "step": 23390
    },
    {
      "epoch": 0.45003894567799135,
      "grad_norm": 429.42297383282795,
      "learning_rate": 1.1735146111335945e-05,
      "loss": 348.132,
      "step": 23400
    },
    {
      "epoch": 0.4502312700137512,
      "grad_norm": 421.7864502342012,
      "learning_rate": 1.1729135222721315e-05,
      "loss": 347.8242,
      "step": 23410
    },
    {
      "epoch": 0.450423594349511,
      "grad_norm": 447.68249484641063,
      "learning_rate": 1.1723123690032376e-05,
      "loss": 336.8539,
      "step": 23420
    },
    {
      "epoch": 0.45061591868527084,
      "grad_norm": 485.7670008683627,
      "learning_rate": 1.171711151550832e-05,
      "loss": 363.7877,
      "step": 23430
    },
    {
      "epoch": 0.4508082430210307,
      "grad_norm": 449.00491653264834,
      "learning_rate": 1.1711098701388581e-05,
      "loss": 350.8491,
      "step": 23440
    },
    {
      "epoch": 0.4510005673567905,
      "grad_norm": 454.2703523796391,
      "learning_rate": 1.1705085249912837e-05,
      "loss": 347.3355,
      "step": 23450
    },
    {
      "epoch": 0.45119289169255034,
      "grad_norm": 411.9916057270429,
      "learning_rate": 1.1699071163320997e-05,
      "loss": 353.2647,
      "step": 23460
    },
    {
      "epoch": 0.45138521602831017,
      "grad_norm": 452.8958070873488,
      "learning_rate": 1.169305644385321e-05,
      "loss": 345.1969,
      "step": 23470
    },
    {
      "epoch": 0.45157754036406994,
      "grad_norm": 494.2100717146746,
      "learning_rate": 1.168704109374986e-05,
      "loss": 353.7135,
      "step": 23480
    },
    {
      "epoch": 0.45176986469982977,
      "grad_norm": 449.68152600840335,
      "learning_rate": 1.1681025115251566e-05,
      "loss": 355.7168,
      "step": 23490
    },
    {
      "epoch": 0.4519621890355896,
      "grad_norm": 674.964144990722,
      "learning_rate": 1.1675008510599176e-05,
      "loss": 343.0748,
      "step": 23500
    },
    {
      "epoch": 0.45215451337134943,
      "grad_norm": 490.0261202930183,
      "learning_rate": 1.1668991282033784e-05,
      "loss": 348.6971,
      "step": 23510
    },
    {
      "epoch": 0.45234683770710926,
      "grad_norm": 455.8273639377138,
      "learning_rate": 1.16629734317967e-05,
      "loss": 358.1657,
      "step": 23520
    },
    {
      "epoch": 0.4525391620428691,
      "grad_norm": 419.0916901909572,
      "learning_rate": 1.165695496212948e-05,
      "loss": 362.4954,
      "step": 23530
    },
    {
      "epoch": 0.4527314863786289,
      "grad_norm": 454.6441146595929,
      "learning_rate": 1.1650935875273901e-05,
      "loss": 354.3914,
      "step": 23540
    },
    {
      "epoch": 0.45292381071438875,
      "grad_norm": 419.06496780810573,
      "learning_rate": 1.1644916173471976e-05,
      "loss": 349.879,
      "step": 23550
    },
    {
      "epoch": 0.4531161350501486,
      "grad_norm": 478.5405629393259,
      "learning_rate": 1.1638895858965942e-05,
      "loss": 349.3788,
      "step": 23560
    },
    {
      "epoch": 0.4533084593859084,
      "grad_norm": 410.9707440162878,
      "learning_rate": 1.1632874933998268e-05,
      "loss": 347.2258,
      "step": 23570
    },
    {
      "epoch": 0.45350078372166824,
      "grad_norm": 455.2224512740443,
      "learning_rate": 1.1626853400811649e-05,
      "loss": 352.5157,
      "step": 23580
    },
    {
      "epoch": 0.4536931080574281,
      "grad_norm": 385.77117263918154,
      "learning_rate": 1.1620831261649003e-05,
      "loss": 344.0658,
      "step": 23590
    },
    {
      "epoch": 0.45388543239318785,
      "grad_norm": 440.1833825299997,
      "learning_rate": 1.1614808518753485e-05,
      "loss": 350.3631,
      "step": 23600
    },
    {
      "epoch": 0.4540777567289477,
      "grad_norm": 480.10720954439904,
      "learning_rate": 1.1608785174368461e-05,
      "loss": 352.6639,
      "step": 23610
    },
    {
      "epoch": 0.4542700810647075,
      "grad_norm": 441.5500416075034,
      "learning_rate": 1.1602761230737531e-05,
      "loss": 366.6013,
      "step": 23620
    },
    {
      "epoch": 0.45446240540046734,
      "grad_norm": 482.1848272942076,
      "learning_rate": 1.1596736690104514e-05,
      "loss": 367.0901,
      "step": 23630
    },
    {
      "epoch": 0.45465472973622717,
      "grad_norm": 460.45210208397236,
      "learning_rate": 1.1590711554713452e-05,
      "loss": 346.6765,
      "step": 23640
    },
    {
      "epoch": 0.454847054071987,
      "grad_norm": 456.1787475203777,
      "learning_rate": 1.1584685826808604e-05,
      "loss": 363.1558,
      "step": 23650
    },
    {
      "epoch": 0.45503937840774683,
      "grad_norm": 499.47085897902195,
      "learning_rate": 1.157865950863446e-05,
      "loss": 353.6442,
      "step": 23660
    },
    {
      "epoch": 0.45523170274350666,
      "grad_norm": 536.0662737737993,
      "learning_rate": 1.1572632602435717e-05,
      "loss": 361.2149,
      "step": 23670
    },
    {
      "epoch": 0.4554240270792665,
      "grad_norm": 436.0792700804202,
      "learning_rate": 1.1566605110457305e-05,
      "loss": 347.6642,
      "step": 23680
    },
    {
      "epoch": 0.4556163514150263,
      "grad_norm": 505.564823658509,
      "learning_rate": 1.1560577034944364e-05,
      "loss": 353.9503,
      "step": 23690
    },
    {
      "epoch": 0.45580867575078615,
      "grad_norm": 465.59527106465043,
      "learning_rate": 1.1554548378142249e-05,
      "loss": 355.2777,
      "step": 23700
    },
    {
      "epoch": 0.4560010000865459,
      "grad_norm": 506.2045142510194,
      "learning_rate": 1.1548519142296541e-05,
      "loss": 353.3511,
      "step": 23710
    },
    {
      "epoch": 0.45619332442230576,
      "grad_norm": 440.91948793154995,
      "learning_rate": 1.1542489329653024e-05,
      "loss": 358.9648,
      "step": 23720
    },
    {
      "epoch": 0.4563856487580656,
      "grad_norm": 452.31877685288396,
      "learning_rate": 1.153645894245771e-05,
      "loss": 340.7169,
      "step": 23730
    },
    {
      "epoch": 0.4565779730938254,
      "grad_norm": 461.49590591788785,
      "learning_rate": 1.1530427982956813e-05,
      "loss": 360.2601,
      "step": 23740
    },
    {
      "epoch": 0.45677029742958525,
      "grad_norm": 453.1214623453998,
      "learning_rate": 1.1524396453396767e-05,
      "loss": 348.997,
      "step": 23750
    },
    {
      "epoch": 0.4569626217653451,
      "grad_norm": 479.4461904394601,
      "learning_rate": 1.1518364356024219e-05,
      "loss": 357.3673,
      "step": 23760
    },
    {
      "epoch": 0.4571549461011049,
      "grad_norm": 460.89689096615183,
      "learning_rate": 1.1512331693086025e-05,
      "loss": 342.6755,
      "step": 23770
    },
    {
      "epoch": 0.45734727043686474,
      "grad_norm": 462.8234758102437,
      "learning_rate": 1.1506298466829256e-05,
      "loss": 358.9926,
      "step": 23780
    },
    {
      "epoch": 0.45753959477262457,
      "grad_norm": 448.5863191433053,
      "learning_rate": 1.1500264679501181e-05,
      "loss": 348.5272,
      "step": 23790
    },
    {
      "epoch": 0.4577319191083844,
      "grad_norm": 460.53409831470395,
      "learning_rate": 1.1494230333349292e-05,
      "loss": 355.6848,
      "step": 23800
    },
    {
      "epoch": 0.45792424344414423,
      "grad_norm": 509.920675443593,
      "learning_rate": 1.1488195430621284e-05,
      "loss": 350.3394,
      "step": 23810
    },
    {
      "epoch": 0.458116567779904,
      "grad_norm": 439.49786410227006,
      "learning_rate": 1.1482159973565051e-05,
      "loss": 359.6057,
      "step": 23820
    },
    {
      "epoch": 0.45830889211566384,
      "grad_norm": 491.9157748010782,
      "learning_rate": 1.147612396442871e-05,
      "loss": 349.4035,
      "step": 23830
    },
    {
      "epoch": 0.45850121645142367,
      "grad_norm": 451.0924836074746,
      "learning_rate": 1.1470087405460572e-05,
      "loss": 343.7919,
      "step": 23840
    },
    {
      "epoch": 0.4586935407871835,
      "grad_norm": 457.8414577037548,
      "learning_rate": 1.1464050298909153e-05,
      "loss": 341.696,
      "step": 23850
    },
    {
      "epoch": 0.4588858651229433,
      "grad_norm": 412.019896624018,
      "learning_rate": 1.1458012647023178e-05,
      "loss": 331.7996,
      "step": 23860
    },
    {
      "epoch": 0.45907818945870316,
      "grad_norm": 482.92689908053995,
      "learning_rate": 1.1451974452051572e-05,
      "loss": 354.1885,
      "step": 23870
    },
    {
      "epoch": 0.459270513794463,
      "grad_norm": 495.59841852125925,
      "learning_rate": 1.1445935716243463e-05,
      "loss": 350.7379,
      "step": 23880
    },
    {
      "epoch": 0.4594628381302228,
      "grad_norm": 454.90738787300114,
      "learning_rate": 1.143989644184818e-05,
      "loss": 350.8042,
      "step": 23890
    },
    {
      "epoch": 0.45965516246598265,
      "grad_norm": 460.26627063019345,
      "learning_rate": 1.1433856631115252e-05,
      "loss": 360.6821,
      "step": 23900
    },
    {
      "epoch": 0.4598474868017425,
      "grad_norm": 402.56429958468243,
      "learning_rate": 1.142781628629441e-05,
      "loss": 345.0776,
      "step": 23910
    },
    {
      "epoch": 0.4600398111375023,
      "grad_norm": 440.5609777778486,
      "learning_rate": 1.1421775409635585e-05,
      "loss": 346.5627,
      "step": 23920
    },
    {
      "epoch": 0.4602321354732621,
      "grad_norm": 578.4568510348026,
      "learning_rate": 1.1415734003388899e-05,
      "loss": 357.2716,
      "step": 23930
    },
    {
      "epoch": 0.4604244598090219,
      "grad_norm": 462.71039126999233,
      "learning_rate": 1.1409692069804678e-05,
      "loss": 356.5922,
      "step": 23940
    },
    {
      "epoch": 0.46061678414478174,
      "grad_norm": 413.1532488354179,
      "learning_rate": 1.1403649611133444e-05,
      "loss": 339.4667,
      "step": 23950
    },
    {
      "epoch": 0.4608091084805416,
      "grad_norm": 488.3438516601459,
      "learning_rate": 1.1397606629625913e-05,
      "loss": 358.0177,
      "step": 23960
    },
    {
      "epoch": 0.4610014328163014,
      "grad_norm": 437.8413528757877,
      "learning_rate": 1.1391563127532992e-05,
      "loss": 347.4919,
      "step": 23970
    },
    {
      "epoch": 0.46119375715206123,
      "grad_norm": 431.40359132969706,
      "learning_rate": 1.1385519107105791e-05,
      "loss": 350.7815,
      "step": 23980
    },
    {
      "epoch": 0.46138608148782106,
      "grad_norm": 473.6656634299338,
      "learning_rate": 1.1379474570595604e-05,
      "loss": 352.9971,
      "step": 23990
    },
    {
      "epoch": 0.4615784058235809,
      "grad_norm": 449.7846065380691,
      "learning_rate": 1.1373429520253922e-05,
      "loss": 341.5583,
      "step": 24000
    },
    {
      "epoch": 0.4617707301593407,
      "grad_norm": 471.5925685796337,
      "learning_rate": 1.1367383958332427e-05,
      "loss": 343.8156,
      "step": 24010
    },
    {
      "epoch": 0.46196305449510056,
      "grad_norm": 473.61604947104604,
      "learning_rate": 1.1361337887082991e-05,
      "loss": 350.4561,
      "step": 24020
    },
    {
      "epoch": 0.4621553788308604,
      "grad_norm": 452.7277506604807,
      "learning_rate": 1.1355291308757672e-05,
      "loss": 352.6881,
      "step": 24030
    },
    {
      "epoch": 0.46234770316662016,
      "grad_norm": 433.32400679299263,
      "learning_rate": 1.1349244225608727e-05,
      "loss": 350.1903,
      "step": 24040
    },
    {
      "epoch": 0.46254002750238,
      "grad_norm": 418.122197251669,
      "learning_rate": 1.1343196639888591e-05,
      "loss": 354.7574,
      "step": 24050
    },
    {
      "epoch": 0.4627323518381398,
      "grad_norm": 446.06130360221,
      "learning_rate": 1.133714855384989e-05,
      "loss": 357.2964,
      "step": 24060
    },
    {
      "epoch": 0.46292467617389965,
      "grad_norm": 446.2298532602672,
      "learning_rate": 1.1331099969745439e-05,
      "loss": 349.6846,
      "step": 24070
    },
    {
      "epoch": 0.4631170005096595,
      "grad_norm": 517.8338606966761,
      "learning_rate": 1.132505088982823e-05,
      "loss": 365.5651,
      "step": 24080
    },
    {
      "epoch": 0.4633093248454193,
      "grad_norm": 435.9096581886154,
      "learning_rate": 1.131900131635145e-05,
      "loss": 344.9095,
      "step": 24090
    },
    {
      "epoch": 0.46350164918117914,
      "grad_norm": 502.3136728521285,
      "learning_rate": 1.1312951251568461e-05,
      "loss": 352.8324,
      "step": 24100
    },
    {
      "epoch": 0.463693973516939,
      "grad_norm": 444.2673845684395,
      "learning_rate": 1.1306900697732816e-05,
      "loss": 343.4255,
      "step": 24110
    },
    {
      "epoch": 0.4638862978526988,
      "grad_norm": 453.9107994548356,
      "learning_rate": 1.1300849657098248e-05,
      "loss": 350.1235,
      "step": 24120
    },
    {
      "epoch": 0.46407862218845863,
      "grad_norm": 474.68764861178875,
      "learning_rate": 1.1294798131918665e-05,
      "loss": 344.1036,
      "step": 24130
    },
    {
      "epoch": 0.46427094652421846,
      "grad_norm": 429.4393498052354,
      "learning_rate": 1.1288746124448164e-05,
      "loss": 346.141,
      "step": 24140
    },
    {
      "epoch": 0.4644632708599783,
      "grad_norm": 462.7185174091672,
      "learning_rate": 1.1282693636941013e-05,
      "loss": 340.0787,
      "step": 24150
    },
    {
      "epoch": 0.46465559519573807,
      "grad_norm": 416.5424380174432,
      "learning_rate": 1.1276640671651671e-05,
      "loss": 358.5521,
      "step": 24160
    },
    {
      "epoch": 0.4648479195314979,
      "grad_norm": 468.1693665282551,
      "learning_rate": 1.1270587230834757e-05,
      "loss": 355.7998,
      "step": 24170
    },
    {
      "epoch": 0.46504024386725773,
      "grad_norm": 440.98128068223815,
      "learning_rate": 1.1264533316745088e-05,
      "loss": 348.2278,
      "step": 24180
    },
    {
      "epoch": 0.46523256820301756,
      "grad_norm": 426.1283064895602,
      "learning_rate": 1.1258478931637641e-05,
      "loss": 346.072,
      "step": 24190
    },
    {
      "epoch": 0.4654248925387774,
      "grad_norm": 439.37164269656154,
      "learning_rate": 1.1252424077767577e-05,
      "loss": 332.2517,
      "step": 24200
    },
    {
      "epoch": 0.4656172168745372,
      "grad_norm": 579.5531226976567,
      "learning_rate": 1.1246368757390231e-05,
      "loss": 342.8532,
      "step": 24210
    },
    {
      "epoch": 0.46580954121029705,
      "grad_norm": 536.2588621443251,
      "learning_rate": 1.1240312972761105e-05,
      "loss": 342.9382,
      "step": 24220
    },
    {
      "epoch": 0.4660018655460569,
      "grad_norm": 469.8016300904265,
      "learning_rate": 1.1234256726135882e-05,
      "loss": 348.8735,
      "step": 24230
    },
    {
      "epoch": 0.4661941898818167,
      "grad_norm": 458.4639804768228,
      "learning_rate": 1.1228200019770412e-05,
      "loss": 333.2857,
      "step": 24240
    },
    {
      "epoch": 0.46638651421757654,
      "grad_norm": 539.0390267204716,
      "learning_rate": 1.1222142855920719e-05,
      "loss": 353.6003,
      "step": 24250
    },
    {
      "epoch": 0.4665788385533364,
      "grad_norm": 451.7144947091559,
      "learning_rate": 1.1216085236842997e-05,
      "loss": 344.9879,
      "step": 24260
    },
    {
      "epoch": 0.46677116288909615,
      "grad_norm": 462.0655486019301,
      "learning_rate": 1.1210027164793609e-05,
      "loss": 356.0899,
      "step": 24270
    },
    {
      "epoch": 0.466963487224856,
      "grad_norm": 430.84130865051253,
      "learning_rate": 1.1203968642029086e-05,
      "loss": 354.1064,
      "step": 24280
    },
    {
      "epoch": 0.4671558115606158,
      "grad_norm": 464.77943367322086,
      "learning_rate": 1.1197909670806126e-05,
      "loss": 347.4563,
      "step": 24290
    },
    {
      "epoch": 0.46734813589637564,
      "grad_norm": 461.17496182656316,
      "learning_rate": 1.1191850253381602e-05,
      "loss": 360.7592,
      "step": 24300
    },
    {
      "epoch": 0.46754046023213547,
      "grad_norm": 414.7179900777237,
      "learning_rate": 1.1185790392012538e-05,
      "loss": 348.1267,
      "step": 24310
    },
    {
      "epoch": 0.4677327845678953,
      "grad_norm": 514.2895881477375,
      "learning_rate": 1.117973008895614e-05,
      "loss": 348.7906,
      "step": 24320
    },
    {
      "epoch": 0.46792510890365513,
      "grad_norm": 437.8789944275789,
      "learning_rate": 1.1173669346469767e-05,
      "loss": 357.3282,
      "step": 24330
    },
    {
      "epoch": 0.46811743323941496,
      "grad_norm": 449.6214027720906,
      "learning_rate": 1.1167608166810948e-05,
      "loss": 345.5132,
      "step": 24340
    },
    {
      "epoch": 0.4683097575751748,
      "grad_norm": 456.3285175844381,
      "learning_rate": 1.1161546552237368e-05,
      "loss": 363.5194,
      "step": 24350
    },
    {
      "epoch": 0.4685020819109346,
      "grad_norm": 467.80311061048957,
      "learning_rate": 1.1155484505006884e-05,
      "loss": 349.5496,
      "step": 24360
    },
    {
      "epoch": 0.46869440624669445,
      "grad_norm": 466.4334589354807,
      "learning_rate": 1.1149422027377501e-05,
      "loss": 349.8633,
      "step": 24370
    },
    {
      "epoch": 0.4688867305824542,
      "grad_norm": 449.03343841662524,
      "learning_rate": 1.1143359121607397e-05,
      "loss": 346.7382,
      "step": 24380
    },
    {
      "epoch": 0.46907905491821406,
      "grad_norm": 427.20341280497115,
      "learning_rate": 1.1137295789954904e-05,
      "loss": 341.3352,
      "step": 24390
    },
    {
      "epoch": 0.4692713792539739,
      "grad_norm": 447.96260766478883,
      "learning_rate": 1.1131232034678513e-05,
      "loss": 345.6423,
      "step": 24400
    },
    {
      "epoch": 0.4694637035897337,
      "grad_norm": 482.5977086710106,
      "learning_rate": 1.1125167858036874e-05,
      "loss": 349.5774,
      "step": 24410
    },
    {
      "epoch": 0.46965602792549355,
      "grad_norm": 537.1234688920891,
      "learning_rate": 1.1119103262288788e-05,
      "loss": 344.8953,
      "step": 24420
    },
    {
      "epoch": 0.4698483522612534,
      "grad_norm": 481.41717467916874,
      "learning_rate": 1.1113038249693221e-05,
      "loss": 343.588,
      "step": 24430
    },
    {
      "epoch": 0.4700406765970132,
      "grad_norm": 438.699428985415,
      "learning_rate": 1.1106972822509287e-05,
      "loss": 347.0128,
      "step": 24440
    },
    {
      "epoch": 0.47023300093277304,
      "grad_norm": 561.2251274783999,
      "learning_rate": 1.1100906982996257e-05,
      "loss": 357.1581,
      "step": 24450
    },
    {
      "epoch": 0.47042532526853287,
      "grad_norm": 418.10743233194864,
      "learning_rate": 1.109484073341356e-05,
      "loss": 347.4563,
      "step": 24460
    },
    {
      "epoch": 0.4706176496042927,
      "grad_norm": 457.9113594480013,
      "learning_rate": 1.1088774076020772e-05,
      "loss": 335.8791,
      "step": 24470
    },
    {
      "epoch": 0.47080997394005253,
      "grad_norm": 448.84675583006396,
      "learning_rate": 1.1082707013077625e-05,
      "loss": 343.1197,
      "step": 24480
    },
    {
      "epoch": 0.4710022982758123,
      "grad_norm": 486.6496154504289,
      "learning_rate": 1.1076639546843993e-05,
      "loss": 350.1951,
      "step": 24490
    },
    {
      "epoch": 0.47119462261157213,
      "grad_norm": 434.43273951451687,
      "learning_rate": 1.1070571679579913e-05,
      "loss": 350.046,
      "step": 24500
    },
    {
      "epoch": 0.47138694694733196,
      "grad_norm": 424.2807672383882,
      "learning_rate": 1.1064503413545565e-05,
      "loss": 354.8194,
      "step": 24510
    },
    {
      "epoch": 0.4715792712830918,
      "grad_norm": 464.69086297043356,
      "learning_rate": 1.1058434751001272e-05,
      "loss": 349.3823,
      "step": 24520
    },
    {
      "epoch": 0.4717715956188516,
      "grad_norm": 546.2101556802386,
      "learning_rate": 1.1052365694207516e-05,
      "loss": 345.8188,
      "step": 24530
    },
    {
      "epoch": 0.47196391995461145,
      "grad_norm": 442.90774448197226,
      "learning_rate": 1.1046296245424922e-05,
      "loss": 340.8246,
      "step": 24540
    },
    {
      "epoch": 0.4721562442903713,
      "grad_norm": 460.72259924303825,
      "learning_rate": 1.1040226406914254e-05,
      "loss": 343.2918,
      "step": 24550
    },
    {
      "epoch": 0.4723485686261311,
      "grad_norm": 443.35966540667556,
      "learning_rate": 1.103415618093643e-05,
      "loss": 358.9477,
      "step": 24560
    },
    {
      "epoch": 0.47254089296189095,
      "grad_norm": 492.01720815670456,
      "learning_rate": 1.1028085569752512e-05,
      "loss": 345.4642,
      "step": 24570
    },
    {
      "epoch": 0.4727332172976508,
      "grad_norm": 412.1990961267601,
      "learning_rate": 1.1022014575623695e-05,
      "loss": 353.017,
      "step": 24580
    },
    {
      "epoch": 0.4729255416334106,
      "grad_norm": 540.5796744301151,
      "learning_rate": 1.1015943200811325e-05,
      "loss": 345.3408,
      "step": 24590
    },
    {
      "epoch": 0.4731178659691704,
      "grad_norm": 422.4793078979871,
      "learning_rate": 1.1009871447576894e-05,
      "loss": 338.4367,
      "step": 24600
    },
    {
      "epoch": 0.4733101903049302,
      "grad_norm": 451.29858440959424,
      "learning_rate": 1.100379931818203e-05,
      "loss": 350.746,
      "step": 24610
    },
    {
      "epoch": 0.47350251464069004,
      "grad_norm": 485.95574517008765,
      "learning_rate": 1.0997726814888497e-05,
      "loss": 364.0745,
      "step": 24620
    },
    {
      "epoch": 0.4736948389764499,
      "grad_norm": 418.0621755984242,
      "learning_rate": 1.0991653939958203e-05,
      "loss": 337.7672,
      "step": 24630
    },
    {
      "epoch": 0.4738871633122097,
      "grad_norm": 455.56264805557186,
      "learning_rate": 1.0985580695653193e-05,
      "loss": 353.3663,
      "step": 24640
    },
    {
      "epoch": 0.47407948764796953,
      "grad_norm": 418.28588400058135,
      "learning_rate": 1.0979507084235653e-05,
      "loss": 346.353,
      "step": 24650
    },
    {
      "epoch": 0.47427181198372936,
      "grad_norm": 409.83371591037053,
      "learning_rate": 1.0973433107967901e-05,
      "loss": 344.1408,
      "step": 24660
    },
    {
      "epoch": 0.4744641363194892,
      "grad_norm": 490.30897601421594,
      "learning_rate": 1.0967358769112389e-05,
      "loss": 350.0065,
      "step": 24670
    },
    {
      "epoch": 0.474656460655249,
      "grad_norm": 484.6349494526896,
      "learning_rate": 1.0961284069931717e-05,
      "loss": 341.357,
      "step": 24680
    },
    {
      "epoch": 0.47484878499100885,
      "grad_norm": 465.423951443859,
      "learning_rate": 1.0955209012688602e-05,
      "loss": 347.5928,
      "step": 24690
    },
    {
      "epoch": 0.4750411093267687,
      "grad_norm": 409.1567254486132,
      "learning_rate": 1.094913359964591e-05,
      "loss": 341.2007,
      "step": 24700
    },
    {
      "epoch": 0.4752334336625285,
      "grad_norm": 445.6050355718344,
      "learning_rate": 1.0943057833066622e-05,
      "loss": 350.1684,
      "step": 24710
    },
    {
      "epoch": 0.4754257579982883,
      "grad_norm": 458.31830053364575,
      "learning_rate": 1.093698171521387e-05,
      "loss": 347.2435,
      "step": 24720
    },
    {
      "epoch": 0.4756180823340481,
      "grad_norm": 442.15865168854026,
      "learning_rate": 1.0930905248350903e-05,
      "loss": 336.4149,
      "step": 24730
    },
    {
      "epoch": 0.47581040666980795,
      "grad_norm": 446.323305378914,
      "learning_rate": 1.0924828434741101e-05,
      "loss": 349.5124,
      "step": 24740
    },
    {
      "epoch": 0.4760027310055678,
      "grad_norm": 438.08495807593556,
      "learning_rate": 1.0918751276647988e-05,
      "loss": 350.5676,
      "step": 24750
    },
    {
      "epoch": 0.4761950553413276,
      "grad_norm": 445.1443369993769,
      "learning_rate": 1.0912673776335194e-05,
      "loss": 344.2197,
      "step": 24760
    },
    {
      "epoch": 0.47638737967708744,
      "grad_norm": 443.9768021761547,
      "learning_rate": 1.0906595936066496e-05,
      "loss": 347.4774,
      "step": 24770
    },
    {
      "epoch": 0.47657970401284727,
      "grad_norm": 425.05006112185043,
      "learning_rate": 1.090051775810578e-05,
      "loss": 345.046,
      "step": 24780
    },
    {
      "epoch": 0.4767720283486071,
      "grad_norm": 478.3464622247465,
      "learning_rate": 1.0894439244717075e-05,
      "loss": 354.5189,
      "step": 24790
    },
    {
      "epoch": 0.47696435268436693,
      "grad_norm": 443.74489667286645,
      "learning_rate": 1.0888360398164521e-05,
      "loss": 361.6882,
      "step": 24800
    },
    {
      "epoch": 0.47715667702012676,
      "grad_norm": 439.81802183197306,
      "learning_rate": 1.088228122071239e-05,
      "loss": 351.7037,
      "step": 24810
    },
    {
      "epoch": 0.4773490013558866,
      "grad_norm": 431.5731288746098,
      "learning_rate": 1.0876201714625076e-05,
      "loss": 349.9723,
      "step": 24820
    },
    {
      "epoch": 0.47754132569164637,
      "grad_norm": 462.3459775062199,
      "learning_rate": 1.0870121882167095e-05,
      "loss": 345.8595,
      "step": 24830
    },
    {
      "epoch": 0.4777336500274062,
      "grad_norm": 436.5308037709048,
      "learning_rate": 1.0864041725603085e-05,
      "loss": 344.7583,
      "step": 24840
    },
    {
      "epoch": 0.47792597436316603,
      "grad_norm": 460.26871532608396,
      "learning_rate": 1.08579612471978e-05,
      "loss": 339.4395,
      "step": 24850
    },
    {
      "epoch": 0.47811829869892586,
      "grad_norm": 431.9694158221801,
      "learning_rate": 1.0851880449216123e-05,
      "loss": 339.2742,
      "step": 24860
    },
    {
      "epoch": 0.4783106230346857,
      "grad_norm": 484.2571614779925,
      "learning_rate": 1.0845799333923045e-05,
      "loss": 351.7642,
      "step": 24870
    },
    {
      "epoch": 0.4785029473704455,
      "grad_norm": 444.5009495762413,
      "learning_rate": 1.0839717903583684e-05,
      "loss": 344.0675,
      "step": 24880
    },
    {
      "epoch": 0.47869527170620535,
      "grad_norm": 466.60153754361727,
      "learning_rate": 1.0833636160463273e-05,
      "loss": 345.1038,
      "step": 24890
    },
    {
      "epoch": 0.4788875960419652,
      "grad_norm": 454.12358645908995,
      "learning_rate": 1.0827554106827162e-05,
      "loss": 349.2369,
      "step": 24900
    },
    {
      "epoch": 0.479079920377725,
      "grad_norm": 458.23607739890093,
      "learning_rate": 1.0821471744940812e-05,
      "loss": 341.8702,
      "step": 24910
    },
    {
      "epoch": 0.47927224471348484,
      "grad_norm": 439.01173869618395,
      "learning_rate": 1.0815389077069805e-05,
      "loss": 340.1212,
      "step": 24920
    },
    {
      "epoch": 0.47946456904924467,
      "grad_norm": 506.9600893811272,
      "learning_rate": 1.0809306105479834e-05,
      "loss": 350.3729,
      "step": 24930
    },
    {
      "epoch": 0.47965689338500445,
      "grad_norm": 439.36878697300597,
      "learning_rate": 1.0803222832436701e-05,
      "loss": 338.7325,
      "step": 24940
    },
    {
      "epoch": 0.4798492177207643,
      "grad_norm": 494.8668702701581,
      "learning_rate": 1.0797139260206331e-05,
      "loss": 347.2341,
      "step": 24950
    },
    {
      "epoch": 0.4800415420565241,
      "grad_norm": 459.6139563612464,
      "learning_rate": 1.079105539105475e-05,
      "loss": 350.5526,
      "step": 24960
    },
    {
      "epoch": 0.48023386639228394,
      "grad_norm": 569.5012562851917,
      "learning_rate": 1.0784971227248104e-05,
      "loss": 352.9748,
      "step": 24970
    },
    {
      "epoch": 0.48042619072804377,
      "grad_norm": 424.89197083814355,
      "learning_rate": 1.077888677105264e-05,
      "loss": 352.4325,
      "step": 24980
    },
    {
      "epoch": 0.4806185150638036,
      "grad_norm": 446.49260866953017,
      "learning_rate": 1.0772802024734716e-05,
      "loss": 343.1698,
      "step": 24990
    },
    {
      "epoch": 0.48081083939956343,
      "grad_norm": 477.87822001902447,
      "learning_rate": 1.07667169905608e-05,
      "loss": 352.7179,
      "step": 25000
    },
    {
      "epoch": 0.48100316373532326,
      "grad_norm": 455.3227866533238,
      "learning_rate": 1.0760631670797468e-05,
      "loss": 343.0752,
      "step": 25010
    },
    {
      "epoch": 0.4811954880710831,
      "grad_norm": 494.27166505087615,
      "learning_rate": 1.07545460677114e-05,
      "loss": 337.4043,
      "step": 25020
    },
    {
      "epoch": 0.4813878124068429,
      "grad_norm": 503.07761611765466,
      "learning_rate": 1.0748460183569385e-05,
      "loss": 350.7558,
      "step": 25030
    },
    {
      "epoch": 0.48158013674260275,
      "grad_norm": 475.1194980159393,
      "learning_rate": 1.0742374020638315e-05,
      "loss": 346.2251,
      "step": 25040
    },
    {
      "epoch": 0.4817724610783625,
      "grad_norm": 438.784613352954,
      "learning_rate": 1.073628758118518e-05,
      "loss": 357.0386,
      "step": 25050
    },
    {
      "epoch": 0.48196478541412235,
      "grad_norm": 478.48266691770704,
      "learning_rate": 1.0730200867477083e-05,
      "loss": 342.8381,
      "step": 25060
    },
    {
      "epoch": 0.4821571097498822,
      "grad_norm": 471.0869564054891,
      "learning_rate": 1.0724113881781222e-05,
      "loss": 346.0839,
      "step": 25070
    },
    {
      "epoch": 0.482349434085642,
      "grad_norm": 513.9448779597403,
      "learning_rate": 1.0718026626364902e-05,
      "loss": 357.383,
      "step": 25080
    },
    {
      "epoch": 0.48254175842140185,
      "grad_norm": 431.4311270140063,
      "learning_rate": 1.0711939103495515e-05,
      "loss": 344.1433,
      "step": 25090
    },
    {
      "epoch": 0.4827340827571617,
      "grad_norm": 498.95380197289,
      "learning_rate": 1.0705851315440575e-05,
      "loss": 345.7734,
      "step": 25100
    },
    {
      "epoch": 0.4829264070929215,
      "grad_norm": 438.87532980026094,
      "learning_rate": 1.0699763264467675e-05,
      "loss": 345.4024,
      "step": 25110
    },
    {
      "epoch": 0.48311873142868134,
      "grad_norm": 472.51704784160995,
      "learning_rate": 1.069367495284452e-05,
      "loss": 343.9813,
      "step": 25120
    },
    {
      "epoch": 0.48331105576444117,
      "grad_norm": 422.5549587358019,
      "learning_rate": 1.06875863828389e-05,
      "loss": 334.9597,
      "step": 25130
    },
    {
      "epoch": 0.483503380100201,
      "grad_norm": 490.89726557182894,
      "learning_rate": 1.0681497556718706e-05,
      "loss": 331.7853,
      "step": 25140
    },
    {
      "epoch": 0.4836957044359608,
      "grad_norm": 475.2300081889788,
      "learning_rate": 1.067540847675193e-05,
      "loss": 348.329,
      "step": 25150
    },
    {
      "epoch": 0.48388802877172066,
      "grad_norm": 463.3089969042881,
      "learning_rate": 1.0669319145206646e-05,
      "loss": 349.3792,
      "step": 25160
    },
    {
      "epoch": 0.48408035310748043,
      "grad_norm": 455.0400266361233,
      "learning_rate": 1.066322956435104e-05,
      "loss": 348.8611,
      "step": 25170
    },
    {
      "epoch": 0.48427267744324026,
      "grad_norm": 458.31353774106776,
      "learning_rate": 1.0657139736453375e-05,
      "loss": 354.1627,
      "step": 25180
    },
    {
      "epoch": 0.4844650017790001,
      "grad_norm": 467.24138652429815,
      "learning_rate": 1.0651049663782007e-05,
      "loss": 356.1149,
      "step": 25190
    },
    {
      "epoch": 0.4846573261147599,
      "grad_norm": 449.78103138295876,
      "learning_rate": 1.0644959348605397e-05,
      "loss": 345.8042,
      "step": 25200
    },
    {
      "epoch": 0.48484965045051975,
      "grad_norm": 443.87809744652037,
      "learning_rate": 1.0638868793192079e-05,
      "loss": 340.828,
      "step": 25210
    },
    {
      "epoch": 0.4850419747862796,
      "grad_norm": 452.6505135387983,
      "learning_rate": 1.0632777999810685e-05,
      "loss": 338.9143,
      "step": 25220
    },
    {
      "epoch": 0.4852342991220394,
      "grad_norm": 439.7766405435959,
      "learning_rate": 1.062668697072994e-05,
      "loss": 345.1837,
      "step": 25230
    },
    {
      "epoch": 0.48542662345779924,
      "grad_norm": 448.4700508249548,
      "learning_rate": 1.0620595708218646e-05,
      "loss": 341.9236,
      "step": 25240
    },
    {
      "epoch": 0.4856189477935591,
      "grad_norm": 475.4474428490237,
      "learning_rate": 1.06145042145457e-05,
      "loss": 352.3563,
      "step": 25250
    },
    {
      "epoch": 0.4858112721293189,
      "grad_norm": 419.36365099043456,
      "learning_rate": 1.0608412491980084e-05,
      "loss": 336.5221,
      "step": 25260
    },
    {
      "epoch": 0.48600359646507874,
      "grad_norm": 444.23904760051914,
      "learning_rate": 1.0602320542790866e-05,
      "loss": 345.2479,
      "step": 25270
    },
    {
      "epoch": 0.4861959208008385,
      "grad_norm": 509.203731905521,
      "learning_rate": 1.0596228369247188e-05,
      "loss": 340.8519,
      "step": 25280
    },
    {
      "epoch": 0.48638824513659834,
      "grad_norm": 461.5633830418135,
      "learning_rate": 1.0590135973618294e-05,
      "loss": 341.8667,
      "step": 25290
    },
    {
      "epoch": 0.48658056947235817,
      "grad_norm": 492.74475147986857,
      "learning_rate": 1.0584043358173492e-05,
      "loss": 344.5896,
      "step": 25300
    },
    {
      "epoch": 0.486772893808118,
      "grad_norm": 441.82248779718276,
      "learning_rate": 1.0577950525182189e-05,
      "loss": 345.3792,
      "step": 25310
    },
    {
      "epoch": 0.48696521814387783,
      "grad_norm": 496.70645990489965,
      "learning_rate": 1.0571857476913856e-05,
      "loss": 334.5547,
      "step": 25320
    },
    {
      "epoch": 0.48715754247963766,
      "grad_norm": 438.4203165582315,
      "learning_rate": 1.056576421563806e-05,
      "loss": 347.8875,
      "step": 25330
    },
    {
      "epoch": 0.4873498668153975,
      "grad_norm": 485.23372980210036,
      "learning_rate": 1.0559670743624439e-05,
      "loss": 333.4783,
      "step": 25340
    },
    {
      "epoch": 0.4875421911511573,
      "grad_norm": 520.4504717812571,
      "learning_rate": 1.0553577063142705e-05,
      "loss": 355.2171,
      "step": 25350
    },
    {
      "epoch": 0.48773451548691715,
      "grad_norm": 456.3826035251919,
      "learning_rate": 1.0547483176462662e-05,
      "loss": 350.5201,
      "step": 25360
    },
    {
      "epoch": 0.487926839822677,
      "grad_norm": 604.3539239542757,
      "learning_rate": 1.0541389085854177e-05,
      "loss": 348.1062,
      "step": 25370
    },
    {
      "epoch": 0.4881191641584368,
      "grad_norm": 462.9639866093167,
      "learning_rate": 1.0535294793587197e-05,
      "loss": 347.1798,
      "step": 25380
    },
    {
      "epoch": 0.4883114884941966,
      "grad_norm": 459.1180620545904,
      "learning_rate": 1.0529200301931747e-05,
      "loss": 334.194,
      "step": 25390
    },
    {
      "epoch": 0.4885038128299564,
      "grad_norm": 466.0139643714818,
      "learning_rate": 1.052310561315793e-05,
      "loss": 344.5897,
      "step": 25400
    },
    {
      "epoch": 0.48869613716571625,
      "grad_norm": 411.7056966811325,
      "learning_rate": 1.0517010729535903e-05,
      "loss": 351.3513,
      "step": 25410
    },
    {
      "epoch": 0.4888884615014761,
      "grad_norm": 447.7577930113551,
      "learning_rate": 1.0510915653335925e-05,
      "loss": 339.0789,
      "step": 25420
    },
    {
      "epoch": 0.4890807858372359,
      "grad_norm": 462.4783046700875,
      "learning_rate": 1.05048203868283e-05,
      "loss": 354.5434,
      "step": 25430
    },
    {
      "epoch": 0.48927311017299574,
      "grad_norm": 484.2055523382258,
      "learning_rate": 1.0498724932283419e-05,
      "loss": 344.0051,
      "step": 25440
    },
    {
      "epoch": 0.48946543450875557,
      "grad_norm": 462.1476748192149,
      "learning_rate": 1.0492629291971738e-05,
      "loss": 349.671,
      "step": 25450
    },
    {
      "epoch": 0.4896577588445154,
      "grad_norm": 433.1666272164113,
      "learning_rate": 1.0486533468163782e-05,
      "loss": 341.0747,
      "step": 25460
    },
    {
      "epoch": 0.48985008318027523,
      "grad_norm": 427.28080531477684,
      "learning_rate": 1.0480437463130145e-05,
      "loss": 338.3895,
      "step": 25470
    },
    {
      "epoch": 0.49004240751603506,
      "grad_norm": 426.31975680146144,
      "learning_rate": 1.0474341279141486e-05,
      "loss": 353.6146,
      "step": 25480
    },
    {
      "epoch": 0.4902347318517949,
      "grad_norm": 443.26554421686126,
      "learning_rate": 1.0468244918468538e-05,
      "loss": 348.0944,
      "step": 25490
    },
    {
      "epoch": 0.49042705618755467,
      "grad_norm": 488.49539952721847,
      "learning_rate": 1.0462148383382086e-05,
      "loss": 341.5997,
      "step": 25500
    },
    {
      "epoch": 0.4906193805233145,
      "grad_norm": 475.67857010537426,
      "learning_rate": 1.0456051676152996e-05,
      "loss": 338.9557,
      "step": 25510
    },
    {
      "epoch": 0.4908117048590743,
      "grad_norm": 606.8907150405572,
      "learning_rate": 1.0449954799052189e-05,
      "loss": 349.8166,
      "step": 25520
    },
    {
      "epoch": 0.49100402919483416,
      "grad_norm": 504.9694016136632,
      "learning_rate": 1.044385775435065e-05,
      "loss": 352.5424,
      "step": 25530
    },
    {
      "epoch": 0.491196353530594,
      "grad_norm": 441.06303350524996,
      "learning_rate": 1.043776054431943e-05,
      "loss": 344.1763,
      "step": 25540
    },
    {
      "epoch": 0.4913886778663538,
      "grad_norm": 428.6514306100753,
      "learning_rate": 1.0431663171229636e-05,
      "loss": 344.2802,
      "step": 25550
    },
    {
      "epoch": 0.49158100220211365,
      "grad_norm": 465.9579446728845,
      "learning_rate": 1.0425565637352441e-05,
      "loss": 348.4694,
      "step": 25560
    },
    {
      "epoch": 0.4917733265378735,
      "grad_norm": 454.88875844656116,
      "learning_rate": 1.0419467944959072e-05,
      "loss": 349.0906,
      "step": 25570
    },
    {
      "epoch": 0.4919656508736333,
      "grad_norm": 446.72372455965024,
      "learning_rate": 1.0413370096320823e-05,
      "loss": 334.4174,
      "step": 25580
    },
    {
      "epoch": 0.49215797520939314,
      "grad_norm": 436.1817000916842,
      "learning_rate": 1.0407272093709038e-05,
      "loss": 344.6982,
      "step": 25590
    },
    {
      "epoch": 0.49235029954515297,
      "grad_norm": 454.19384015769054,
      "learning_rate": 1.0401173939395128e-05,
      "loss": 342.2402,
      "step": 25600
    },
    {
      "epoch": 0.49254262388091274,
      "grad_norm": 465.78700065151713,
      "learning_rate": 1.0395075635650549e-05,
      "loss": 339.2602,
      "step": 25610
    },
    {
      "epoch": 0.4927349482166726,
      "grad_norm": 423.61803979987485,
      "learning_rate": 1.038897718474682e-05,
      "loss": 341.5205,
      "step": 25620
    },
    {
      "epoch": 0.4929272725524324,
      "grad_norm": 446.3128603090774,
      "learning_rate": 1.0382878588955517e-05,
      "loss": 350.9064,
      "step": 25630
    },
    {
      "epoch": 0.49311959688819224,
      "grad_norm": 437.4944675601813,
      "learning_rate": 1.0376779850548257e-05,
      "loss": 345.0944,
      "step": 25640
    },
    {
      "epoch": 0.49331192122395207,
      "grad_norm": 434.73650695158506,
      "learning_rate": 1.0370680971796732e-05,
      "loss": 336.7031,
      "step": 25650
    },
    {
      "epoch": 0.4935042455597119,
      "grad_norm": 455.07444197146157,
      "learning_rate": 1.0364581954972662e-05,
      "loss": 337.445,
      "step": 25660
    },
    {
      "epoch": 0.4936965698954717,
      "grad_norm": 470.63817511194935,
      "learning_rate": 1.0358482802347838e-05,
      "loss": 347.9157,
      "step": 25670
    },
    {
      "epoch": 0.49388889423123156,
      "grad_norm": 615.3420191970537,
      "learning_rate": 1.0352383516194088e-05,
      "loss": 343.8957,
      "step": 25680
    },
    {
      "epoch": 0.4940812185669914,
      "grad_norm": 449.8024243375401,
      "learning_rate": 1.0346284098783304e-05,
      "loss": 344.9775,
      "step": 25690
    },
    {
      "epoch": 0.4942735429027512,
      "grad_norm": 439.79371865907626,
      "learning_rate": 1.0340184552387406e-05,
      "loss": 344.3167,
      "step": 25700
    },
    {
      "epoch": 0.49446586723851105,
      "grad_norm": 510.9504219105106,
      "learning_rate": 1.0334084879278381e-05,
      "loss": 344.9335,
      "step": 25710
    },
    {
      "epoch": 0.4946581915742709,
      "grad_norm": 453.71646428694226,
      "learning_rate": 1.032798508172826e-05,
      "loss": 345.6492,
      "step": 25720
    },
    {
      "epoch": 0.49485051591003065,
      "grad_norm": 417.9122891919555,
      "learning_rate": 1.0321885162009111e-05,
      "loss": 334.8714,
      "step": 25730
    },
    {
      "epoch": 0.4950428402457905,
      "grad_norm": 426.3977137430743,
      "learning_rate": 1.0315785122393053e-05,
      "loss": 339.017,
      "step": 25740
    },
    {
      "epoch": 0.4952351645815503,
      "grad_norm": 455.42934401603463,
      "learning_rate": 1.0309684965152254e-05,
      "loss": 346.5886,
      "step": 25750
    },
    {
      "epoch": 0.49542748891731014,
      "grad_norm": 427.4702055917018,
      "learning_rate": 1.030358469255892e-05,
      "loss": 349.8492,
      "step": 25760
    },
    {
      "epoch": 0.49561981325307,
      "grad_norm": 503.1026610452885,
      "learning_rate": 1.0297484306885304e-05,
      "loss": 342.484,
      "step": 25770
    },
    {
      "epoch": 0.4958121375888298,
      "grad_norm": 422.15219926086354,
      "learning_rate": 1.0291383810403697e-05,
      "loss": 343.0117,
      "step": 25780
    },
    {
      "epoch": 0.49600446192458963,
      "grad_norm": 426.2328515088724,
      "learning_rate": 1.028528320538643e-05,
      "loss": 345.1555,
      "step": 25790
    },
    {
      "epoch": 0.49619678626034947,
      "grad_norm": 420.70888404290093,
      "learning_rate": 1.0279182494105879e-05,
      "loss": 337.1838,
      "step": 25800
    },
    {
      "epoch": 0.4963891105961093,
      "grad_norm": 430.721565898764,
      "learning_rate": 1.0273081678834462e-05,
      "loss": 337.5997,
      "step": 25810
    },
    {
      "epoch": 0.4965814349318691,
      "grad_norm": 429.10717872706454,
      "learning_rate": 1.026698076184463e-05,
      "loss": 335.7756,
      "step": 25820
    },
    {
      "epoch": 0.49677375926762896,
      "grad_norm": 445.63623859890174,
      "learning_rate": 1.0260879745408876e-05,
      "loss": 344.804,
      "step": 25830
    },
    {
      "epoch": 0.49696608360338873,
      "grad_norm": 515.0421940610776,
      "learning_rate": 1.0254778631799722e-05,
      "loss": 332.7388,
      "step": 25840
    },
    {
      "epoch": 0.49715840793914856,
      "grad_norm": 547.9305516906084,
      "learning_rate": 1.024867742328974e-05,
      "loss": 349.1427,
      "step": 25850
    },
    {
      "epoch": 0.4973507322749084,
      "grad_norm": 457.29395189890954,
      "learning_rate": 1.024257612215152e-05,
      "loss": 347.0169,
      "step": 25860
    },
    {
      "epoch": 0.4975430566106682,
      "grad_norm": 436.5744785798286,
      "learning_rate": 1.02364747306577e-05,
      "loss": 347.5799,
      "step": 25870
    },
    {
      "epoch": 0.49773538094642805,
      "grad_norm": 437.3957172964602,
      "learning_rate": 1.023037325108095e-05,
      "loss": 344.1047,
      "step": 25880
    },
    {
      "epoch": 0.4979277052821879,
      "grad_norm": 451.2434663139427,
      "learning_rate": 1.022427168569397e-05,
      "loss": 336.6407,
      "step": 25890
    },
    {
      "epoch": 0.4981200296179477,
      "grad_norm": 429.12410838179204,
      "learning_rate": 1.021817003676949e-05,
      "loss": 346.0296,
      "step": 25900
    },
    {
      "epoch": 0.49831235395370754,
      "grad_norm": 540.7591793592441,
      "learning_rate": 1.0212068306580274e-05,
      "loss": 350.6603,
      "step": 25910
    },
    {
      "epoch": 0.4985046782894674,
      "grad_norm": 432.4797951328987,
      "learning_rate": 1.0205966497399118e-05,
      "loss": 331.3958,
      "step": 25920
    },
    {
      "epoch": 0.4986970026252272,
      "grad_norm": 475.8132467748236,
      "learning_rate": 1.0199864611498841e-05,
      "loss": 351.4599,
      "step": 25930
    },
    {
      "epoch": 0.49888932696098703,
      "grad_norm": 430.0734617528874,
      "learning_rate": 1.0193762651152299e-05,
      "loss": 351.4419,
      "step": 25940
    },
    {
      "epoch": 0.4990816512967468,
      "grad_norm": 472.1029956897937,
      "learning_rate": 1.018766061863237e-05,
      "loss": 340.3128,
      "step": 25950
    },
    {
      "epoch": 0.49927397563250664,
      "grad_norm": 465.6110802717735,
      "learning_rate": 1.018155851621196e-05,
      "loss": 343.5428,
      "step": 25960
    },
    {
      "epoch": 0.49946629996826647,
      "grad_norm": 430.92541766427337,
      "learning_rate": 1.0175456346164e-05,
      "loss": 339.8401,
      "step": 25970
    },
    {
      "epoch": 0.4996586243040263,
      "grad_norm": 422.5119090674598,
      "learning_rate": 1.0169354110761447e-05,
      "loss": 342.0407,
      "step": 25980
    },
    {
      "epoch": 0.49985094863978613,
      "grad_norm": 435.6800807681979,
      "learning_rate": 1.0163251812277289e-05,
      "loss": 338.9048,
      "step": 25990
    },
    {
      "epoch": 0.500043272975546,
      "grad_norm": 473.092265132417,
      "learning_rate": 1.0157149452984523e-05,
      "loss": 341.6524,
      "step": 26000
    },
    {
      "epoch": 0.5002355973113057,
      "grad_norm": 427.8281945237504,
      "learning_rate": 1.0151047035156182e-05,
      "loss": 343.6659,
      "step": 26010
    },
    {
      "epoch": 0.5004279216470656,
      "grad_norm": 478.5189418354036,
      "learning_rate": 1.0144944561065316e-05,
      "loss": 339.8255,
      "step": 26020
    },
    {
      "epoch": 0.5006202459828254,
      "grad_norm": 435.2745113248596,
      "learning_rate": 1.0138842032984996e-05,
      "loss": 356.8999,
      "step": 26030
    },
    {
      "epoch": 0.5008125703185853,
      "grad_norm": 405.26899659127605,
      "learning_rate": 1.013273945318831e-05,
      "loss": 349.762,
      "step": 26040
    },
    {
      "epoch": 0.5010048946543451,
      "grad_norm": 428.87900165739535,
      "learning_rate": 1.0126636823948373e-05,
      "loss": 338.0337,
      "step": 26050
    },
    {
      "epoch": 0.5011972189901049,
      "grad_norm": 455.91267921591265,
      "learning_rate": 1.0120534147538305e-05,
      "loss": 345.6226,
      "step": 26060
    },
    {
      "epoch": 0.5013895433258647,
      "grad_norm": 443.5850119841078,
      "learning_rate": 1.011443142623126e-05,
      "loss": 351.0039,
      "step": 26070
    },
    {
      "epoch": 0.5015818676616246,
      "grad_norm": 436.0291216771967,
      "learning_rate": 1.0108328662300399e-05,
      "loss": 340.9716,
      "step": 26080
    },
    {
      "epoch": 0.5017741919973844,
      "grad_norm": 446.3422677503715,
      "learning_rate": 1.0102225858018902e-05,
      "loss": 336.164,
      "step": 26090
    },
    {
      "epoch": 0.5019665163331443,
      "grad_norm": 438.47251662104185,
      "learning_rate": 1.009612301565996e-05,
      "loss": 342.6302,
      "step": 26100
    },
    {
      "epoch": 0.502158840668904,
      "grad_norm": 433.6151797711422,
      "learning_rate": 1.0090020137496783e-05,
      "loss": 338.885,
      "step": 26110
    },
    {
      "epoch": 0.5023511650046638,
      "grad_norm": 499.6250064975691,
      "learning_rate": 1.008391722580259e-05,
      "loss": 354.9817,
      "step": 26120
    },
    {
      "epoch": 0.5025434893404237,
      "grad_norm": 468.84803857442233,
      "learning_rate": 1.0077814282850617e-05,
      "loss": 358.8431,
      "step": 26130
    },
    {
      "epoch": 0.5027358136761835,
      "grad_norm": 439.65479580682444,
      "learning_rate": 1.0071711310914111e-05,
      "loss": 342.4324,
      "step": 26140
    },
    {
      "epoch": 0.5029281380119434,
      "grad_norm": 438.4205862312812,
      "learning_rate": 1.0065608312266324e-05,
      "loss": 346.871,
      "step": 26150
    },
    {
      "epoch": 0.5031204623477031,
      "grad_norm": 467.38154208423816,
      "learning_rate": 1.005950528918052e-05,
      "loss": 344.1982,
      "step": 26160
    },
    {
      "epoch": 0.503312786683463,
      "grad_norm": 434.153971787788,
      "learning_rate": 1.0053402243929986e-05,
      "loss": 348.1347,
      "step": 26170
    },
    {
      "epoch": 0.5035051110192228,
      "grad_norm": 464.53257993576017,
      "learning_rate": 1.0047299178787993e-05,
      "loss": 343.7879,
      "step": 26180
    },
    {
      "epoch": 0.5036974353549827,
      "grad_norm": 440.69561714345843,
      "learning_rate": 1.0041196096027841e-05,
      "loss": 340.0599,
      "step": 26190
    },
    {
      "epoch": 0.5038897596907425,
      "grad_norm": 431.1380537992603,
      "learning_rate": 1.003509299792282e-05,
      "loss": 340.2102,
      "step": 26200
    },
    {
      "epoch": 0.5040820840265023,
      "grad_norm": 530.5361816157434,
      "learning_rate": 1.0028989886746241e-05,
      "loss": 329.6086,
      "step": 26210
    },
    {
      "epoch": 0.5042744083622621,
      "grad_norm": 420.87720352454573,
      "learning_rate": 1.0022886764771405e-05,
      "loss": 339.4763,
      "step": 26220
    },
    {
      "epoch": 0.5044667326980219,
      "grad_norm": 451.4880905735666,
      "learning_rate": 1.0016783634271626e-05,
      "loss": 338.5879,
      "step": 26230
    },
    {
      "epoch": 0.5046590570337818,
      "grad_norm": 480.27448745078203,
      "learning_rate": 1.001068049752022e-05,
      "loss": 342.3657,
      "step": 26240
    },
    {
      "epoch": 0.5048513813695416,
      "grad_norm": 437.82798984123554,
      "learning_rate": 1.0004577356790506e-05,
      "loss": 334.7288,
      "step": 26250
    },
    {
      "epoch": 0.5050437057053014,
      "grad_norm": 489.6626013709203,
      "learning_rate": 9.998474214355805e-06,
      "loss": 341.5562,
      "step": 26260
    },
    {
      "epoch": 0.5052360300410612,
      "grad_norm": 465.29240718561823,
      "learning_rate": 9.992371072489434e-06,
      "loss": 351.4869,
      "step": 26270
    },
    {
      "epoch": 0.5054283543768211,
      "grad_norm": 442.5453968598927,
      "learning_rate": 9.986267933464707e-06,
      "loss": 354.0186,
      "step": 26280
    },
    {
      "epoch": 0.5056206787125809,
      "grad_norm": 449.96081343231003,
      "learning_rate": 9.980164799554953e-06,
      "loss": 347.4886,
      "step": 26290
    },
    {
      "epoch": 0.5058130030483408,
      "grad_norm": 465.16494406161587,
      "learning_rate": 9.97406167303348e-06,
      "loss": 343.1912,
      "step": 26300
    },
    {
      "epoch": 0.5060053273841005,
      "grad_norm": 536.5672845204716,
      "learning_rate": 9.967958556173612e-06,
      "loss": 345.5935,
      "step": 26310
    },
    {
      "epoch": 0.5061976517198604,
      "grad_norm": 484.91161652683945,
      "learning_rate": 9.961855451248645e-06,
      "loss": 336.4124,
      "step": 26320
    },
    {
      "epoch": 0.5063899760556202,
      "grad_norm": 414.40591062278037,
      "learning_rate": 9.955752360531896e-06,
      "loss": 351.0156,
      "step": 26330
    },
    {
      "epoch": 0.50658230039138,
      "grad_norm": 518.9984050344088,
      "learning_rate": 9.949649286296663e-06,
      "loss": 341.3864,
      "step": 26340
    },
    {
      "epoch": 0.5067746247271399,
      "grad_norm": 515.5651496992565,
      "learning_rate": 9.943546230816236e-06,
      "loss": 343.5138,
      "step": 26350
    },
    {
      "epoch": 0.5069669490628996,
      "grad_norm": 450.28204301907545,
      "learning_rate": 9.937443196363908e-06,
      "loss": 350.0704,
      "step": 26360
    },
    {
      "epoch": 0.5071592733986595,
      "grad_norm": 453.02053857871084,
      "learning_rate": 9.931340185212955e-06,
      "loss": 336.2348,
      "step": 26370
    },
    {
      "epoch": 0.5073515977344193,
      "grad_norm": 491.1923399841011,
      "learning_rate": 9.925237199636652e-06,
      "loss": 334.6663,
      "step": 26380
    },
    {
      "epoch": 0.5075439220701792,
      "grad_norm": 513.6348863780322,
      "learning_rate": 9.919134241908252e-06,
      "loss": 339.5493,
      "step": 26390
    },
    {
      "epoch": 0.507736246405939,
      "grad_norm": 468.4940037945204,
      "learning_rate": 9.913031314301016e-06,
      "loss": 351.3495,
      "step": 26400
    },
    {
      "epoch": 0.5079285707416988,
      "grad_norm": 468.49295621127857,
      "learning_rate": 9.906928419088178e-06,
      "loss": 336.9728,
      "step": 26410
    },
    {
      "epoch": 0.5081208950774586,
      "grad_norm": 462.0725195256504,
      "learning_rate": 9.900825558542965e-06,
      "loss": 347.8383,
      "step": 26420
    },
    {
      "epoch": 0.5083132194132185,
      "grad_norm": 459.97071544613885,
      "learning_rate": 9.894722734938595e-06,
      "loss": 352.6653,
      "step": 26430
    },
    {
      "epoch": 0.5085055437489783,
      "grad_norm": 445.8809364873831,
      "learning_rate": 9.888619950548267e-06,
      "loss": 338.8954,
      "step": 26440
    },
    {
      "epoch": 0.508697868084738,
      "grad_norm": 406.24969992272116,
      "learning_rate": 9.88251720764517e-06,
      "loss": 333.8568,
      "step": 26450
    },
    {
      "epoch": 0.5088901924204979,
      "grad_norm": 434.58611560041817,
      "learning_rate": 9.876414508502468e-06,
      "loss": 340.9845,
      "step": 26460
    },
    {
      "epoch": 0.5090825167562577,
      "grad_norm": 453.8723914764098,
      "learning_rate": 9.870311855393324e-06,
      "loss": 346.499,
      "step": 26470
    },
    {
      "epoch": 0.5092748410920176,
      "grad_norm": 449.97169361927297,
      "learning_rate": 9.864209250590875e-06,
      "loss": 332.3278,
      "step": 26480
    },
    {
      "epoch": 0.5094671654277774,
      "grad_norm": 401.3479610629499,
      "learning_rate": 9.858106696368235e-06,
      "loss": 345.8816,
      "step": 26490
    },
    {
      "epoch": 0.5096594897635373,
      "grad_norm": 438.93567781557203,
      "learning_rate": 9.852004194998503e-06,
      "loss": 338.5684,
      "step": 26500
    },
    {
      "epoch": 0.509851814099297,
      "grad_norm": 440.5472231891131,
      "learning_rate": 9.845901748754767e-06,
      "loss": 338.7455,
      "step": 26510
    },
    {
      "epoch": 0.5100441384350569,
      "grad_norm": 415.21322748978696,
      "learning_rate": 9.839799359910087e-06,
      "loss": 344.3596,
      "step": 26520
    },
    {
      "epoch": 0.5102364627708167,
      "grad_norm": 458.1084664626786,
      "learning_rate": 9.833697030737495e-06,
      "loss": 354.6294,
      "step": 26530
    },
    {
      "epoch": 0.5104287871065766,
      "grad_norm": 436.6787253675564,
      "learning_rate": 9.827594763510016e-06,
      "loss": 346.794,
      "step": 26540
    },
    {
      "epoch": 0.5106211114423364,
      "grad_norm": 448.1170740412977,
      "learning_rate": 9.821492560500641e-06,
      "loss": 337.7722,
      "step": 26550
    },
    {
      "epoch": 0.5108134357780962,
      "grad_norm": 396.6410997935971,
      "learning_rate": 9.815390423982339e-06,
      "loss": 341.1862,
      "step": 26560
    },
    {
      "epoch": 0.511005760113856,
      "grad_norm": 470.0484816188522,
      "learning_rate": 9.809288356228051e-06,
      "loss": 338.5166,
      "step": 26570
    },
    {
      "epoch": 0.5111980844496158,
      "grad_norm": 447.3056913642524,
      "learning_rate": 9.803186359510701e-06,
      "loss": 339.6327,
      "step": 26580
    },
    {
      "epoch": 0.5113904087853757,
      "grad_norm": 459.000569588546,
      "learning_rate": 9.797084436103186e-06,
      "loss": 342.616,
      "step": 26590
    },
    {
      "epoch": 0.5115827331211354,
      "grad_norm": 435.33280173405103,
      "learning_rate": 9.790982588278362e-06,
      "loss": 345.4626,
      "step": 26600
    },
    {
      "epoch": 0.5117750574568953,
      "grad_norm": 466.38512320023113,
      "learning_rate": 9.784880818309076e-06,
      "loss": 338.4864,
      "step": 26610
    },
    {
      "epoch": 0.5119673817926551,
      "grad_norm": 465.4942602488181,
      "learning_rate": 9.778779128468133e-06,
      "loss": 361.0277,
      "step": 26620
    },
    {
      "epoch": 0.512159706128415,
      "grad_norm": 455.13323019582765,
      "learning_rate": 9.77267752102831e-06,
      "loss": 341.2256,
      "step": 26630
    },
    {
      "epoch": 0.5123520304641748,
      "grad_norm": 471.91920957852267,
      "learning_rate": 9.766575998262353e-06,
      "loss": 337.2599,
      "step": 26640
    },
    {
      "epoch": 0.5125443547999347,
      "grad_norm": 449.80734038194635,
      "learning_rate": 9.760474562442984e-06,
      "loss": 358.0246,
      "step": 26650
    },
    {
      "epoch": 0.5127366791356944,
      "grad_norm": 521.8266663715146,
      "learning_rate": 9.754373215842884e-06,
      "loss": 343.768,
      "step": 26660
    },
    {
      "epoch": 0.5129290034714543,
      "grad_norm": 492.9214205865808,
      "learning_rate": 9.748271960734708e-06,
      "loss": 351.5626,
      "step": 26670
    },
    {
      "epoch": 0.5131213278072141,
      "grad_norm": 426.94099297025866,
      "learning_rate": 9.742170799391063e-06,
      "loss": 331.4579,
      "step": 26680
    },
    {
      "epoch": 0.5133136521429739,
      "grad_norm": 471.2429557142538,
      "learning_rate": 9.736069734084541e-06,
      "loss": 329.301,
      "step": 26690
    },
    {
      "epoch": 0.5135059764787337,
      "grad_norm": 428.84147913856174,
      "learning_rate": 9.729968767087685e-06,
      "loss": 338.2121,
      "step": 26700
    },
    {
      "epoch": 0.5136983008144935,
      "grad_norm": 479.6254456239305,
      "learning_rate": 9.723867900673e-06,
      "loss": 331.2304,
      "step": 26710
    },
    {
      "epoch": 0.5138906251502534,
      "grad_norm": 417.61116967195187,
      "learning_rate": 9.717767137112964e-06,
      "loss": 341.0746,
      "step": 26720
    },
    {
      "epoch": 0.5140829494860132,
      "grad_norm": 473.1536989765579,
      "learning_rate": 9.711666478680007e-06,
      "loss": 347.7481,
      "step": 26730
    },
    {
      "epoch": 0.5142752738217731,
      "grad_norm": 456.8626738321443,
      "learning_rate": 9.705565927646526e-06,
      "loss": 339.454,
      "step": 26740
    },
    {
      "epoch": 0.5144675981575328,
      "grad_norm": 481.5129464094531,
      "learning_rate": 9.699465486284871e-06,
      "loss": 339.195,
      "step": 26750
    },
    {
      "epoch": 0.5146599224932927,
      "grad_norm": 452.08185532759103,
      "learning_rate": 9.693365156867363e-06,
      "loss": 334.3235,
      "step": 26760
    },
    {
      "epoch": 0.5148522468290525,
      "grad_norm": 479.66554519802463,
      "learning_rate": 9.68726494166627e-06,
      "loss": 333.8176,
      "step": 26770
    },
    {
      "epoch": 0.5150445711648124,
      "grad_norm": 454.6945929340308,
      "learning_rate": 9.681164842953816e-06,
      "loss": 340.9728,
      "step": 26780
    },
    {
      "epoch": 0.5152368955005722,
      "grad_norm": 476.6845113981557,
      "learning_rate": 9.675064863002196e-06,
      "loss": 340.5218,
      "step": 26790
    },
    {
      "epoch": 0.5154292198363319,
      "grad_norm": 455.3659689305939,
      "learning_rate": 9.668965004083549e-06,
      "loss": 346.2917,
      "step": 26800
    },
    {
      "epoch": 0.5156215441720918,
      "grad_norm": 493.80965921803494,
      "learning_rate": 9.66286526846997e-06,
      "loss": 342.8252,
      "step": 26810
    },
    {
      "epoch": 0.5158138685078516,
      "grad_norm": 504.6513010726089,
      "learning_rate": 9.656765658433507e-06,
      "loss": 334.8217,
      "step": 26820
    },
    {
      "epoch": 0.5160061928436115,
      "grad_norm": 467.309410542043,
      "learning_rate": 9.650666176246171e-06,
      "loss": 339.3484,
      "step": 26830
    },
    {
      "epoch": 0.5161985171793713,
      "grad_norm": 476.0467276490168,
      "learning_rate": 9.644566824179916e-06,
      "loss": 331.1993,
      "step": 26840
    },
    {
      "epoch": 0.5163908415151311,
      "grad_norm": 441.6495571161134,
      "learning_rate": 9.638467604506648e-06,
      "loss": 343.9717,
      "step": 26850
    },
    {
      "epoch": 0.5165831658508909,
      "grad_norm": 417.00699745682414,
      "learning_rate": 9.632368519498224e-06,
      "loss": 337.1766,
      "step": 26860
    },
    {
      "epoch": 0.5167754901866508,
      "grad_norm": 437.5703591096981,
      "learning_rate": 9.626269571426456e-06,
      "loss": 348.3298,
      "step": 26870
    },
    {
      "epoch": 0.5169678145224106,
      "grad_norm": 456.85078227257094,
      "learning_rate": 9.620170762563103e-06,
      "loss": 332.2039,
      "step": 26880
    },
    {
      "epoch": 0.5171601388581705,
      "grad_norm": 429.5608359887313,
      "learning_rate": 9.614072095179862e-06,
      "loss": 338.3338,
      "step": 26890
    },
    {
      "epoch": 0.5173524631939302,
      "grad_norm": 463.06188230953273,
      "learning_rate": 9.607973571548396e-06,
      "loss": 348.9774,
      "step": 26900
    },
    {
      "epoch": 0.51754478752969,
      "grad_norm": 428.1768626436941,
      "learning_rate": 9.601875193940301e-06,
      "loss": 330.0414,
      "step": 26910
    },
    {
      "epoch": 0.5177371118654499,
      "grad_norm": 438.71866643416524,
      "learning_rate": 9.595776964627119e-06,
      "loss": 335.5818,
      "step": 26920
    },
    {
      "epoch": 0.5179294362012097,
      "grad_norm": 446.2781621492192,
      "learning_rate": 9.589678885880342e-06,
      "loss": 341.9717,
      "step": 26930
    },
    {
      "epoch": 0.5181217605369696,
      "grad_norm": 463.88599220541624,
      "learning_rate": 9.583580959971402e-06,
      "loss": 339.0089,
      "step": 26940
    },
    {
      "epoch": 0.5183140848727293,
      "grad_norm": 443.0918244425555,
      "learning_rate": 9.577483189171681e-06,
      "loss": 345.7707,
      "step": 26950
    },
    {
      "epoch": 0.5185064092084892,
      "grad_norm": 433.1425729845799,
      "learning_rate": 9.571385575752487e-06,
      "loss": 335.6523,
      "step": 26960
    },
    {
      "epoch": 0.518698733544249,
      "grad_norm": 498.50752877677576,
      "learning_rate": 9.565288121985093e-06,
      "loss": 348.2144,
      "step": 26970
    },
    {
      "epoch": 0.5188910578800089,
      "grad_norm": 443.5120851101149,
      "learning_rate": 9.559190830140695e-06,
      "loss": 345.7477,
      "step": 26980
    },
    {
      "epoch": 0.5190833822157687,
      "grad_norm": 446.2507898899773,
      "learning_rate": 9.553093702490433e-06,
      "loss": 334.2978,
      "step": 26990
    },
    {
      "epoch": 0.5192757065515285,
      "grad_norm": 437.6071370750127,
      "learning_rate": 9.54699674130538e-06,
      "loss": 338.2994,
      "step": 27000
    },
    {
      "epoch": 0.5194680308872883,
      "grad_norm": 422.12308663996794,
      "learning_rate": 9.540899948856561e-06,
      "loss": 328.964,
      "step": 27010
    },
    {
      "epoch": 0.5196603552230481,
      "grad_norm": 427.6474875637078,
      "learning_rate": 9.534803327414931e-06,
      "loss": 345.2805,
      "step": 27020
    },
    {
      "epoch": 0.519852679558808,
      "grad_norm": 444.04750253279036,
      "learning_rate": 9.52870687925138e-06,
      "loss": 341.8681,
      "step": 27030
    },
    {
      "epoch": 0.5200450038945678,
      "grad_norm": 465.5837361454199,
      "learning_rate": 9.522610606636728e-06,
      "loss": 340.9822,
      "step": 27040
    },
    {
      "epoch": 0.5202373282303276,
      "grad_norm": 440.35732478272877,
      "learning_rate": 9.516514511841745e-06,
      "loss": 343.4004,
      "step": 27050
    },
    {
      "epoch": 0.5204296525660874,
      "grad_norm": 426.34681471216334,
      "learning_rate": 9.51041859713712e-06,
      "loss": 329.5743,
      "step": 27060
    },
    {
      "epoch": 0.5206219769018473,
      "grad_norm": 433.8876551556778,
      "learning_rate": 9.50432286479348e-06,
      "loss": 337.3153,
      "step": 27070
    },
    {
      "epoch": 0.5208143012376071,
      "grad_norm": 440.3975265547607,
      "learning_rate": 9.498227317081387e-06,
      "loss": 341.7763,
      "step": 27080
    },
    {
      "epoch": 0.521006625573367,
      "grad_norm": 491.4696973561834,
      "learning_rate": 9.492131956271334e-06,
      "loss": 338.346,
      "step": 27090
    },
    {
      "epoch": 0.5211989499091267,
      "grad_norm": 478.4658942725379,
      "learning_rate": 9.486036784633738e-06,
      "loss": 341.305,
      "step": 27100
    },
    {
      "epoch": 0.5213912742448866,
      "grad_norm": 452.927882810684,
      "learning_rate": 9.47994180443895e-06,
      "loss": 332.9051,
      "step": 27110
    },
    {
      "epoch": 0.5215835985806464,
      "grad_norm": 444.6085391917634,
      "learning_rate": 9.473847017957254e-06,
      "loss": 339.4169,
      "step": 27120
    },
    {
      "epoch": 0.5217759229164062,
      "grad_norm": 450.68569710805474,
      "learning_rate": 9.467752427458851e-06,
      "loss": 342.388,
      "step": 27130
    },
    {
      "epoch": 0.5219682472521661,
      "grad_norm": 451.06512983722774,
      "learning_rate": 9.461658035213878e-06,
      "loss": 337.2664,
      "step": 27140
    },
    {
      "epoch": 0.5221605715879258,
      "grad_norm": 428.33853337240384,
      "learning_rate": 9.455563843492397e-06,
      "loss": 339.9756,
      "step": 27150
    },
    {
      "epoch": 0.5223528959236857,
      "grad_norm": 459.54159229164367,
      "learning_rate": 9.449469854564393e-06,
      "loss": 332.8475,
      "step": 27160
    },
    {
      "epoch": 0.5225452202594455,
      "grad_norm": 414.48512219442904,
      "learning_rate": 9.443376070699778e-06,
      "loss": 340.6676,
      "step": 27170
    },
    {
      "epoch": 0.5227375445952054,
      "grad_norm": 467.4786499837992,
      "learning_rate": 9.437282494168379e-06,
      "loss": 334.2827,
      "step": 27180
    },
    {
      "epoch": 0.5229298689309652,
      "grad_norm": 452.12718048029274,
      "learning_rate": 9.431189127239962e-06,
      "loss": 348.6354,
      "step": 27190
    },
    {
      "epoch": 0.523122193266725,
      "grad_norm": 429.63075900356364,
      "learning_rate": 9.4250959721842e-06,
      "loss": 334.5864,
      "step": 27200
    },
    {
      "epoch": 0.5233145176024848,
      "grad_norm": 481.2536545754605,
      "learning_rate": 9.419003031270692e-06,
      "loss": 339.6702,
      "step": 27210
    },
    {
      "epoch": 0.5235068419382447,
      "grad_norm": 403.6776440665063,
      "learning_rate": 9.412910306768959e-06,
      "loss": 331.6599,
      "step": 27220
    },
    {
      "epoch": 0.5236991662740045,
      "grad_norm": 433.9658409070433,
      "learning_rate": 9.40681780094844e-06,
      "loss": 329.3873,
      "step": 27230
    },
    {
      "epoch": 0.5238914906097643,
      "grad_norm": 453.95931032368793,
      "learning_rate": 9.400725516078496e-06,
      "loss": 336.5731,
      "step": 27240
    },
    {
      "epoch": 0.5240838149455241,
      "grad_norm": 442.5289560235294,
      "learning_rate": 9.394633454428396e-06,
      "loss": 345.3696,
      "step": 27250
    },
    {
      "epoch": 0.5242761392812839,
      "grad_norm": 392.6976284047434,
      "learning_rate": 9.388541618267341e-06,
      "loss": 340.6594,
      "step": 27260
    },
    {
      "epoch": 0.5244684636170438,
      "grad_norm": 424.63016958522695,
      "learning_rate": 9.382450009864434e-06,
      "loss": 346.652,
      "step": 27270
    },
    {
      "epoch": 0.5246607879528036,
      "grad_norm": 431.39515318511565,
      "learning_rate": 9.376358631488697e-06,
      "loss": 325.4798,
      "step": 27280
    },
    {
      "epoch": 0.5248531122885635,
      "grad_norm": 450.44302218789613,
      "learning_rate": 9.37026748540907e-06,
      "loss": 342.5619,
      "step": 27290
    },
    {
      "epoch": 0.5250454366243232,
      "grad_norm": 480.7471023848163,
      "learning_rate": 9.364176573894404e-06,
      "loss": 331.1925,
      "step": 27300
    },
    {
      "epoch": 0.5252377609600831,
      "grad_norm": 496.0676672748972,
      "learning_rate": 9.358085899213467e-06,
      "loss": 337.8682,
      "step": 27310
    },
    {
      "epoch": 0.5254300852958429,
      "grad_norm": 445.49214195866523,
      "learning_rate": 9.351995463634925e-06,
      "loss": 334.9769,
      "step": 27320
    },
    {
      "epoch": 0.5256224096316028,
      "grad_norm": 437.53513656220525,
      "learning_rate": 9.345905269427374e-06,
      "loss": 334.3453,
      "step": 27330
    },
    {
      "epoch": 0.5258147339673626,
      "grad_norm": 396.73516754663257,
      "learning_rate": 9.33981531885931e-06,
      "loss": 340.4432,
      "step": 27340
    },
    {
      "epoch": 0.5260070583031223,
      "grad_norm": 464.1043517284434,
      "learning_rate": 9.333725614199132e-06,
      "loss": 341.525,
      "step": 27350
    },
    {
      "epoch": 0.5261993826388822,
      "grad_norm": 472.77488696354254,
      "learning_rate": 9.32763615771516e-06,
      "loss": 341.5732,
      "step": 27360
    },
    {
      "epoch": 0.526391706974642,
      "grad_norm": 602.5340527676817,
      "learning_rate": 9.321546951675616e-06,
      "loss": 341.2068,
      "step": 27370
    },
    {
      "epoch": 0.5265840313104019,
      "grad_norm": 430.6197362546596,
      "learning_rate": 9.315457998348627e-06,
      "loss": 334.2109,
      "step": 27380
    },
    {
      "epoch": 0.5267763556461617,
      "grad_norm": 419.1736102297784,
      "learning_rate": 9.309369300002224e-06,
      "loss": 349.3417,
      "step": 27390
    },
    {
      "epoch": 0.5269686799819215,
      "grad_norm": 441.6157714767051,
      "learning_rate": 9.303280858904356e-06,
      "loss": 334.7883,
      "step": 27400
    },
    {
      "epoch": 0.5271610043176813,
      "grad_norm": 464.59223017060003,
      "learning_rate": 9.297192677322862e-06,
      "loss": 345.1679,
      "step": 27410
    },
    {
      "epoch": 0.5273533286534412,
      "grad_norm": 446.44145284009414,
      "learning_rate": 9.291104757525486e-06,
      "loss": 345.3021,
      "step": 27420
    },
    {
      "epoch": 0.527545652989201,
      "grad_norm": 468.80816541459,
      "learning_rate": 9.28501710177988e-06,
      "loss": 328.7764,
      "step": 27430
    },
    {
      "epoch": 0.5277379773249609,
      "grad_norm": 444.29357835953067,
      "learning_rate": 9.278929712353595e-06,
      "loss": 346.5714,
      "step": 27440
    },
    {
      "epoch": 0.5279303016607206,
      "grad_norm": 469.3491646014543,
      "learning_rate": 9.272842591514083e-06,
      "loss": 358.628,
      "step": 27450
    },
    {
      "epoch": 0.5281226259964804,
      "grad_norm": 428.40195476739024,
      "learning_rate": 9.2667557415287e-06,
      "loss": 342.0147,
      "step": 27460
    },
    {
      "epoch": 0.5283149503322403,
      "grad_norm": 476.3116740634766,
      "learning_rate": 9.260669164664687e-06,
      "loss": 344.0848,
      "step": 27470
    },
    {
      "epoch": 0.5285072746680001,
      "grad_norm": 446.56240586600103,
      "learning_rate": 9.254582863189205e-06,
      "loss": 347.999,
      "step": 27480
    },
    {
      "epoch": 0.52869959900376,
      "grad_norm": 504.0513679274186,
      "learning_rate": 9.248496839369293e-06,
      "loss": 331.0626,
      "step": 27490
    },
    {
      "epoch": 0.5288919233395197,
      "grad_norm": 436.8948134547532,
      "learning_rate": 9.242411095471897e-06,
      "loss": 360.6836,
      "step": 27500
    },
    {
      "epoch": 0.5290842476752796,
      "grad_norm": 424.060867543618,
      "learning_rate": 9.236325633763856e-06,
      "loss": 335.3242,
      "step": 27510
    },
    {
      "epoch": 0.5292765720110394,
      "grad_norm": 413.5356040210607,
      "learning_rate": 9.230240456511905e-06,
      "loss": 351.8245,
      "step": 27520
    },
    {
      "epoch": 0.5294688963467993,
      "grad_norm": 474.9002354952727,
      "learning_rate": 9.224155565982673e-06,
      "loss": 332.4841,
      "step": 27530
    },
    {
      "epoch": 0.529661220682559,
      "grad_norm": 448.1461839060164,
      "learning_rate": 9.218070964442673e-06,
      "loss": 339.3383,
      "step": 27540
    },
    {
      "epoch": 0.5298535450183189,
      "grad_norm": 412.1948335911245,
      "learning_rate": 9.21198665415833e-06,
      "loss": 338.0674,
      "step": 27550
    },
    {
      "epoch": 0.5300458693540787,
      "grad_norm": 438.62678980169545,
      "learning_rate": 9.205902637395943e-06,
      "loss": 341.4978,
      "step": 27560
    },
    {
      "epoch": 0.5302381936898385,
      "grad_norm": 473.17742750787113,
      "learning_rate": 9.199818916421706e-06,
      "loss": 332.2519,
      "step": 27570
    },
    {
      "epoch": 0.5304305180255984,
      "grad_norm": 446.67672469028355,
      "learning_rate": 9.193735493501707e-06,
      "loss": 338.8791,
      "step": 27580
    },
    {
      "epoch": 0.5306228423613581,
      "grad_norm": 435.28041816545954,
      "learning_rate": 9.187652370901925e-06,
      "loss": 345.6248,
      "step": 27590
    },
    {
      "epoch": 0.530815166697118,
      "grad_norm": 418.6282966369881,
      "learning_rate": 9.181569550888217e-06,
      "loss": 340.5824,
      "step": 27600
    },
    {
      "epoch": 0.5310074910328778,
      "grad_norm": 502.5156746744452,
      "learning_rate": 9.175487035726332e-06,
      "loss": 335.7792,
      "step": 27610
    },
    {
      "epoch": 0.5311998153686377,
      "grad_norm": 469.1641062599217,
      "learning_rate": 9.169404827681912e-06,
      "loss": 331.6646,
      "step": 27620
    },
    {
      "epoch": 0.5313921397043975,
      "grad_norm": 427.13753895045767,
      "learning_rate": 9.163322929020476e-06,
      "loss": 331.2961,
      "step": 27630
    },
    {
      "epoch": 0.5315844640401574,
      "grad_norm": 424.02002431219785,
      "learning_rate": 9.157241342007428e-06,
      "loss": 337.6597,
      "step": 27640
    },
    {
      "epoch": 0.5317767883759171,
      "grad_norm": 425.14340527355495,
      "learning_rate": 9.15116006890806e-06,
      "loss": 341.9425,
      "step": 27650
    },
    {
      "epoch": 0.531969112711677,
      "grad_norm": 470.85583914878305,
      "learning_rate": 9.145079111987552e-06,
      "loss": 341.8651,
      "step": 27660
    },
    {
      "epoch": 0.5321614370474368,
      "grad_norm": 425.60658910311287,
      "learning_rate": 9.138998473510953e-06,
      "loss": 334.6015,
      "step": 27670
    },
    {
      "epoch": 0.5323537613831967,
      "grad_norm": 423.49096563894926,
      "learning_rate": 9.1329181557432e-06,
      "loss": 341.0399,
      "step": 27680
    },
    {
      "epoch": 0.5325460857189565,
      "grad_norm": 438.7966182790001,
      "learning_rate": 9.126838160949119e-06,
      "loss": 338.1275,
      "step": 27690
    },
    {
      "epoch": 0.5327384100547162,
      "grad_norm": 436.2738776572443,
      "learning_rate": 9.120758491393402e-06,
      "loss": 345.8831,
      "step": 27700
    },
    {
      "epoch": 0.5329307343904761,
      "grad_norm": 490.6268343034542,
      "learning_rate": 9.114679149340623e-06,
      "loss": 347.0099,
      "step": 27710
    },
    {
      "epoch": 0.5331230587262359,
      "grad_norm": 441.4693682034415,
      "learning_rate": 9.10860013705524e-06,
      "loss": 337.0759,
      "step": 27720
    },
    {
      "epoch": 0.5333153830619958,
      "grad_norm": 414.15192782864995,
      "learning_rate": 9.102521456801582e-06,
      "loss": 330.4536,
      "step": 27730
    },
    {
      "epoch": 0.5335077073977555,
      "grad_norm": 434.794170602652,
      "learning_rate": 9.096443110843864e-06,
      "loss": 335.0601,
      "step": 27740
    },
    {
      "epoch": 0.5337000317335154,
      "grad_norm": 447.43551757004855,
      "learning_rate": 9.090365101446157e-06,
      "loss": 337.0183,
      "step": 27750
    },
    {
      "epoch": 0.5338923560692752,
      "grad_norm": 446.93597862565804,
      "learning_rate": 9.084287430872435e-06,
      "loss": 338.7575,
      "step": 27760
    },
    {
      "epoch": 0.5340846804050351,
      "grad_norm": 481.3634539599402,
      "learning_rate": 9.078210101386518e-06,
      "loss": 333.1981,
      "step": 27770
    },
    {
      "epoch": 0.5342770047407949,
      "grad_norm": 447.3149994843473,
      "learning_rate": 9.072133115252113e-06,
      "loss": 336.9669,
      "step": 27780
    },
    {
      "epoch": 0.5344693290765548,
      "grad_norm": 429.3612659469217,
      "learning_rate": 9.066056474732798e-06,
      "loss": 336.0402,
      "step": 27790
    },
    {
      "epoch": 0.5346616534123145,
      "grad_norm": 414.5918973974524,
      "learning_rate": 9.059980182092022e-06,
      "loss": 331.5065,
      "step": 27800
    },
    {
      "epoch": 0.5348539777480743,
      "grad_norm": 417.84237302182044,
      "learning_rate": 9.053904239593106e-06,
      "loss": 333.6915,
      "step": 27810
    },
    {
      "epoch": 0.5350463020838342,
      "grad_norm": 446.77193543813934,
      "learning_rate": 9.047828649499236e-06,
      "loss": 338.6586,
      "step": 27820
    },
    {
      "epoch": 0.535238626419594,
      "grad_norm": 444.4986007671738,
      "learning_rate": 9.041753414073463e-06,
      "loss": 334.6368,
      "step": 27830
    },
    {
      "epoch": 0.5354309507553539,
      "grad_norm": 485.7434297371959,
      "learning_rate": 9.035678535578723e-06,
      "loss": 334.8524,
      "step": 27840
    },
    {
      "epoch": 0.5356232750911136,
      "grad_norm": 500.19237962256324,
      "learning_rate": 9.029604016277798e-06,
      "loss": 338.1684,
      "step": 27850
    },
    {
      "epoch": 0.5358155994268735,
      "grad_norm": 461.7583356155207,
      "learning_rate": 9.02352985843335e-06,
      "loss": 345.0043,
      "step": 27860
    },
    {
      "epoch": 0.5360079237626333,
      "grad_norm": 437.61452419759644,
      "learning_rate": 9.017456064307904e-06,
      "loss": 338.4087,
      "step": 27870
    },
    {
      "epoch": 0.5362002480983932,
      "grad_norm": 438.8446878519873,
      "learning_rate": 9.01138263616385e-06,
      "loss": 339.3034,
      "step": 27880
    },
    {
      "epoch": 0.536392572434153,
      "grad_norm": 461.63840676842267,
      "learning_rate": 9.005309576263436e-06,
      "loss": 334.8043,
      "step": 27890
    },
    {
      "epoch": 0.5365848967699128,
      "grad_norm": 461.7534268183935,
      "learning_rate": 8.999236886868772e-06,
      "loss": 329.6729,
      "step": 27900
    },
    {
      "epoch": 0.5367772211056726,
      "grad_norm": 407.3645047384184,
      "learning_rate": 8.993164570241844e-06,
      "loss": 332.8074,
      "step": 27910
    },
    {
      "epoch": 0.5369695454414324,
      "grad_norm": 425.52965102653695,
      "learning_rate": 8.987092628644483e-06,
      "loss": 335.9697,
      "step": 27920
    },
    {
      "epoch": 0.5371618697771923,
      "grad_norm": 418.56227043833223,
      "learning_rate": 8.981021064338388e-06,
      "loss": 336.1161,
      "step": 27930
    },
    {
      "epoch": 0.537354194112952,
      "grad_norm": 461.23536225491625,
      "learning_rate": 8.974949879585118e-06,
      "loss": 339.2944,
      "step": 27940
    },
    {
      "epoch": 0.5375465184487119,
      "grad_norm": 485.3516253607373,
      "learning_rate": 8.968879076646093e-06,
      "loss": 334.7745,
      "step": 27950
    },
    {
      "epoch": 0.5377388427844717,
      "grad_norm": 421.53071117785976,
      "learning_rate": 8.96280865778258e-06,
      "loss": 335.2337,
      "step": 27960
    },
    {
      "epoch": 0.5379311671202316,
      "grad_norm": 430.1855333776133,
      "learning_rate": 8.956738625255709e-06,
      "loss": 342.5138,
      "step": 27970
    },
    {
      "epoch": 0.5381234914559914,
      "grad_norm": 475.16221216607164,
      "learning_rate": 8.950668981326473e-06,
      "loss": 333.5263,
      "step": 27980
    },
    {
      "epoch": 0.5383158157917513,
      "grad_norm": 511.645706997364,
      "learning_rate": 8.94459972825571e-06,
      "loss": 335.4495,
      "step": 27990
    },
    {
      "epoch": 0.538508140127511,
      "grad_norm": 410.51887273755403,
      "learning_rate": 8.938530868304121e-06,
      "loss": 332.1829,
      "step": 28000
    },
    {
      "epoch": 0.5387004644632709,
      "grad_norm": 437.6134356795408,
      "learning_rate": 8.932462403732248e-06,
      "loss": 334.7665,
      "step": 28010
    },
    {
      "epoch": 0.5388927887990307,
      "grad_norm": 428.4209467792851,
      "learning_rate": 8.926394336800502e-06,
      "loss": 343.598,
      "step": 28020
    },
    {
      "epoch": 0.5390851131347905,
      "grad_norm": 451.84650321673274,
      "learning_rate": 8.920326669769134e-06,
      "loss": 343.8959,
      "step": 28030
    },
    {
      "epoch": 0.5392774374705503,
      "grad_norm": 518.4843242374923,
      "learning_rate": 8.914259404898247e-06,
      "loss": 337.2625,
      "step": 28040
    },
    {
      "epoch": 0.5394697618063101,
      "grad_norm": 416.23432604720193,
      "learning_rate": 8.908192544447803e-06,
      "loss": 335.3577,
      "step": 28050
    },
    {
      "epoch": 0.53966208614207,
      "grad_norm": 426.30592660350817,
      "learning_rate": 8.902126090677605e-06,
      "loss": 324.2099,
      "step": 28060
    },
    {
      "epoch": 0.5398544104778298,
      "grad_norm": 402.30371872416225,
      "learning_rate": 8.896060045847305e-06,
      "loss": 337.4351,
      "step": 28070
    },
    {
      "epoch": 0.5400467348135897,
      "grad_norm": 456.36624915387966,
      "learning_rate": 8.889994412216403e-06,
      "loss": 333.0561,
      "step": 28080
    },
    {
      "epoch": 0.5402390591493494,
      "grad_norm": 412.24727874851567,
      "learning_rate": 8.883929192044254e-06,
      "loss": 331.1373,
      "step": 28090
    },
    {
      "epoch": 0.5404313834851093,
      "grad_norm": 380.62541716075714,
      "learning_rate": 8.877864387590049e-06,
      "loss": 340.2437,
      "step": 28100
    },
    {
      "epoch": 0.5406237078208691,
      "grad_norm": 440.1228835952414,
      "learning_rate": 8.871800001112822e-06,
      "loss": 338.534,
      "step": 28110
    },
    {
      "epoch": 0.540816032156629,
      "grad_norm": 480.08817067322906,
      "learning_rate": 8.865736034871468e-06,
      "loss": 345.7047,
      "step": 28120
    },
    {
      "epoch": 0.5410083564923888,
      "grad_norm": 471.61473402532386,
      "learning_rate": 8.859672491124706e-06,
      "loss": 338.4267,
      "step": 28130
    },
    {
      "epoch": 0.5412006808281485,
      "grad_norm": 432.3467408308856,
      "learning_rate": 8.853609372131105e-06,
      "loss": 341.381,
      "step": 28140
    },
    {
      "epoch": 0.5413930051639084,
      "grad_norm": 443.80785402538953,
      "learning_rate": 8.84754668014908e-06,
      "loss": 335.8323,
      "step": 28150
    },
    {
      "epoch": 0.5415853294996682,
      "grad_norm": 439.0447745842612,
      "learning_rate": 8.841484417436886e-06,
      "loss": 334.076,
      "step": 28160
    },
    {
      "epoch": 0.5417776538354281,
      "grad_norm": 422.7650893979843,
      "learning_rate": 8.835422586252613e-06,
      "loss": 347.1306,
      "step": 28170
    },
    {
      "epoch": 0.5419699781711879,
      "grad_norm": 421.6508770064547,
      "learning_rate": 8.829361188854194e-06,
      "loss": 321.6093,
      "step": 28180
    },
    {
      "epoch": 0.5421623025069477,
      "grad_norm": 427.31250226236494,
      "learning_rate": 8.823300227499393e-06,
      "loss": 333.0436,
      "step": 28190
    },
    {
      "epoch": 0.5423546268427075,
      "grad_norm": 468.52047744918286,
      "learning_rate": 8.817239704445827e-06,
      "loss": 340.8722,
      "step": 28200
    },
    {
      "epoch": 0.5425469511784674,
      "grad_norm": 504.0240027744351,
      "learning_rate": 8.811179621950937e-06,
      "loss": 330.1855,
      "step": 28210
    },
    {
      "epoch": 0.5427392755142272,
      "grad_norm": 478.89209220649013,
      "learning_rate": 8.805119982272001e-06,
      "loss": 337.9279,
      "step": 28220
    },
    {
      "epoch": 0.5429315998499871,
      "grad_norm": 423.3389409426329,
      "learning_rate": 8.799060787666142e-06,
      "loss": 340.6609,
      "step": 28230
    },
    {
      "epoch": 0.5431239241857468,
      "grad_norm": 393.1420448334663,
      "learning_rate": 8.793002040390304e-06,
      "loss": 339.8627,
      "step": 28240
    },
    {
      "epoch": 0.5433162485215066,
      "grad_norm": 414.86516448883225,
      "learning_rate": 8.786943742701273e-06,
      "loss": 334.7134,
      "step": 28250
    },
    {
      "epoch": 0.5435085728572665,
      "grad_norm": 412.89993580484315,
      "learning_rate": 8.780885896855659e-06,
      "loss": 329.9108,
      "step": 28260
    },
    {
      "epoch": 0.5437008971930263,
      "grad_norm": 481.13998132736816,
      "learning_rate": 8.774828505109918e-06,
      "loss": 345.4943,
      "step": 28270
    },
    {
      "epoch": 0.5438932215287862,
      "grad_norm": 500.99070497146204,
      "learning_rate": 8.768771569720324e-06,
      "loss": 336.8532,
      "step": 28280
    },
    {
      "epoch": 0.5440855458645459,
      "grad_norm": 453.980302741379,
      "learning_rate": 8.762715092942983e-06,
      "loss": 347.41,
      "step": 28290
    },
    {
      "epoch": 0.5442778702003058,
      "grad_norm": 409.65322457432677,
      "learning_rate": 8.756659077033838e-06,
      "loss": 335.5657,
      "step": 28300
    },
    {
      "epoch": 0.5444701945360656,
      "grad_norm": 414.8686596089426,
      "learning_rate": 8.750603524248653e-06,
      "loss": 347.6419,
      "step": 28310
    },
    {
      "epoch": 0.5446625188718255,
      "grad_norm": 421.86740367267794,
      "learning_rate": 8.744548436843021e-06,
      "loss": 340.6217,
      "step": 28320
    },
    {
      "epoch": 0.5448548432075853,
      "grad_norm": 482.2323568937171,
      "learning_rate": 8.738493817072359e-06,
      "loss": 338.5933,
      "step": 28330
    },
    {
      "epoch": 0.5450471675433451,
      "grad_norm": 467.3871180532147,
      "learning_rate": 8.73243966719192e-06,
      "loss": 339.4357,
      "step": 28340
    },
    {
      "epoch": 0.5452394918791049,
      "grad_norm": 401.4908893834571,
      "learning_rate": 8.726385989456764e-06,
      "loss": 334.9473,
      "step": 28350
    },
    {
      "epoch": 0.5454318162148647,
      "grad_norm": 433.4794396308252,
      "learning_rate": 8.7203327861218e-06,
      "loss": 339.971,
      "step": 28360
    },
    {
      "epoch": 0.5456241405506246,
      "grad_norm": 441.0387351790818,
      "learning_rate": 8.71428005944173e-06,
      "loss": 337.72,
      "step": 28370
    },
    {
      "epoch": 0.5458164648863844,
      "grad_norm": 428.13136384157383,
      "learning_rate": 8.708227811671112e-06,
      "loss": 333.3676,
      "step": 28380
    },
    {
      "epoch": 0.5460087892221442,
      "grad_norm": 442.5879824465687,
      "learning_rate": 8.702176045064296e-06,
      "loss": 336.1903,
      "step": 28390
    },
    {
      "epoch": 0.546201113557904,
      "grad_norm": 422.39172087490203,
      "learning_rate": 8.696124761875467e-06,
      "loss": 339.8283,
      "step": 28400
    },
    {
      "epoch": 0.5463934378936639,
      "grad_norm": 434.0395528927452,
      "learning_rate": 8.690073964358635e-06,
      "loss": 343.2911,
      "step": 28410
    },
    {
      "epoch": 0.5465857622294237,
      "grad_norm": 463.67618687858186,
      "learning_rate": 8.684023654767613e-06,
      "loss": 337.2746,
      "step": 28420
    },
    {
      "epoch": 0.5467780865651836,
      "grad_norm": 419.11976645542535,
      "learning_rate": 8.677973835356048e-06,
      "loss": 329.8752,
      "step": 28430
    },
    {
      "epoch": 0.5469704109009433,
      "grad_norm": 412.4573191019436,
      "learning_rate": 8.671924508377392e-06,
      "loss": 330.0221,
      "step": 28440
    },
    {
      "epoch": 0.5471627352367032,
      "grad_norm": 421.7986683894783,
      "learning_rate": 8.665875676084927e-06,
      "loss": 333.1989,
      "step": 28450
    },
    {
      "epoch": 0.547355059572463,
      "grad_norm": 454.27721239662213,
      "learning_rate": 8.659827340731738e-06,
      "loss": 340.4138,
      "step": 28460
    },
    {
      "epoch": 0.5475473839082228,
      "grad_norm": 448.56329087880783,
      "learning_rate": 8.653779504570728e-06,
      "loss": 337.8382,
      "step": 28470
    },
    {
      "epoch": 0.5477397082439827,
      "grad_norm": 455.7554211130853,
      "learning_rate": 8.647732169854622e-06,
      "loss": 338.9848,
      "step": 28480
    },
    {
      "epoch": 0.5479320325797424,
      "grad_norm": 425.8428844021126,
      "learning_rate": 8.641685338835947e-06,
      "loss": 335.3184,
      "step": 28490
    },
    {
      "epoch": 0.5481243569155023,
      "grad_norm": 418.6209147011269,
      "learning_rate": 8.635639013767053e-06,
      "loss": 338.6773,
      "step": 28500
    },
    {
      "epoch": 0.5483166812512621,
      "grad_norm": 454.39227518125404,
      "learning_rate": 8.629593196900088e-06,
      "loss": 334.4415,
      "step": 28510
    },
    {
      "epoch": 0.548509005587022,
      "grad_norm": 443.6295486310688,
      "learning_rate": 8.62354789048703e-06,
      "loss": 335.5656,
      "step": 28520
    },
    {
      "epoch": 0.5487013299227818,
      "grad_norm": 455.3809299234011,
      "learning_rate": 8.617503096779648e-06,
      "loss": 344.2566,
      "step": 28530
    },
    {
      "epoch": 0.5488936542585416,
      "grad_norm": 432.61163492997537,
      "learning_rate": 8.61145881802953e-06,
      "loss": 338.4444,
      "step": 28540
    },
    {
      "epoch": 0.5490859785943014,
      "grad_norm": 416.04342025552677,
      "learning_rate": 8.605415056488067e-06,
      "loss": 337.1958,
      "step": 28550
    },
    {
      "epoch": 0.5492783029300613,
      "grad_norm": 433.7150243152945,
      "learning_rate": 8.599371814406465e-06,
      "loss": 328.8556,
      "step": 28560
    },
    {
      "epoch": 0.5494706272658211,
      "grad_norm": 430.8442735669157,
      "learning_rate": 8.59332909403573e-06,
      "loss": 362.964,
      "step": 28570
    },
    {
      "epoch": 0.5496629516015809,
      "grad_norm": 465.53509833606705,
      "learning_rate": 8.587286897626672e-06,
      "loss": 335.1742,
      "step": 28580
    },
    {
      "epoch": 0.5498552759373407,
      "grad_norm": 427.74232433655965,
      "learning_rate": 8.581245227429918e-06,
      "loss": 340.6384,
      "step": 28590
    },
    {
      "epoch": 0.5500476002731005,
      "grad_norm": 436.3727526506871,
      "learning_rate": 8.575204085695887e-06,
      "loss": 327.1928,
      "step": 28600
    },
    {
      "epoch": 0.5502399246088604,
      "grad_norm": 423.15984783366486,
      "learning_rate": 8.5691634746748e-06,
      "loss": 341.7284,
      "step": 28610
    },
    {
      "epoch": 0.5504322489446202,
      "grad_norm": 418.5620886294004,
      "learning_rate": 8.563123396616683e-06,
      "loss": 328.5208,
      "step": 28620
    },
    {
      "epoch": 0.5506245732803801,
      "grad_norm": 434.9170721520363,
      "learning_rate": 8.557083853771377e-06,
      "loss": 333.2414,
      "step": 28630
    },
    {
      "epoch": 0.5508168976161398,
      "grad_norm": 470.2420065683713,
      "learning_rate": 8.551044848388502e-06,
      "loss": 338.3116,
      "step": 28640
    },
    {
      "epoch": 0.5510092219518997,
      "grad_norm": 431.9839568196808,
      "learning_rate": 8.545006382717487e-06,
      "loss": 336.1055,
      "step": 28650
    },
    {
      "epoch": 0.5512015462876595,
      "grad_norm": 439.70073126859603,
      "learning_rate": 8.538968459007569e-06,
      "loss": 346.576,
      "step": 28660
    },
    {
      "epoch": 0.5513938706234194,
      "grad_norm": 407.1652728810677,
      "learning_rate": 8.532931079507772e-06,
      "loss": 335.7871,
      "step": 28670
    },
    {
      "epoch": 0.5515861949591792,
      "grad_norm": 430.1923176470468,
      "learning_rate": 8.526894246466916e-06,
      "loss": 333.8787,
      "step": 28680
    },
    {
      "epoch": 0.551778519294939,
      "grad_norm": 405.54655513701874,
      "learning_rate": 8.520857962133623e-06,
      "loss": 332.1585,
      "step": 28690
    },
    {
      "epoch": 0.5519708436306988,
      "grad_norm": 448.57622451269,
      "learning_rate": 8.514822228756311e-06,
      "loss": 340.4178,
      "step": 28700
    },
    {
      "epoch": 0.5521631679664586,
      "grad_norm": 412.2554033897965,
      "learning_rate": 8.508787048583191e-06,
      "loss": 335.7281,
      "step": 28710
    },
    {
      "epoch": 0.5523554923022185,
      "grad_norm": 436.5841673392677,
      "learning_rate": 8.502752423862264e-06,
      "loss": 340.6044,
      "step": 28720
    },
    {
      "epoch": 0.5525478166379783,
      "grad_norm": 417.799958675462,
      "learning_rate": 8.496718356841335e-06,
      "loss": 329.6618,
      "step": 28730
    },
    {
      "epoch": 0.5527401409737381,
      "grad_norm": 418.7008634354597,
      "learning_rate": 8.49068484976799e-06,
      "loss": 336.898,
      "step": 28740
    },
    {
      "epoch": 0.5529324653094979,
      "grad_norm": 424.53243014044205,
      "learning_rate": 8.484651904889614e-06,
      "loss": 331.572,
      "step": 28750
    },
    {
      "epoch": 0.5531247896452578,
      "grad_norm": 470.03338812860176,
      "learning_rate": 8.478619524453369e-06,
      "loss": 332.2833,
      "step": 28760
    },
    {
      "epoch": 0.5533171139810176,
      "grad_norm": 446.2105367755079,
      "learning_rate": 8.472587710706232e-06,
      "loss": 329.3469,
      "step": 28770
    },
    {
      "epoch": 0.5535094383167775,
      "grad_norm": 417.8468134694729,
      "learning_rate": 8.466556465894942e-06,
      "loss": 337.4566,
      "step": 28780
    },
    {
      "epoch": 0.5537017626525372,
      "grad_norm": 454.18677070536796,
      "learning_rate": 8.460525792266046e-06,
      "loss": 342.5244,
      "step": 28790
    },
    {
      "epoch": 0.5538940869882971,
      "grad_norm": 433.468099090712,
      "learning_rate": 8.454495692065862e-06,
      "loss": 346.3313,
      "step": 28800
    },
    {
      "epoch": 0.5540864113240569,
      "grad_norm": 440.8969137114263,
      "learning_rate": 8.448466167540514e-06,
      "loss": 330.4774,
      "step": 28810
    },
    {
      "epoch": 0.5542787356598167,
      "grad_norm": 430.8910401230213,
      "learning_rate": 8.442437220935893e-06,
      "loss": 346.404,
      "step": 28820
    },
    {
      "epoch": 0.5544710599955766,
      "grad_norm": 420.68848721767137,
      "learning_rate": 8.436408854497679e-06,
      "loss": 338.6633,
      "step": 28830
    },
    {
      "epoch": 0.5546633843313363,
      "grad_norm": 447.5909695442307,
      "learning_rate": 8.430381070471348e-06,
      "loss": 339.4675,
      "step": 28840
    },
    {
      "epoch": 0.5548557086670962,
      "grad_norm": 458.24278042564714,
      "learning_rate": 8.424353871102144e-06,
      "loss": 330.7136,
      "step": 28850
    },
    {
      "epoch": 0.555048033002856,
      "grad_norm": 463.9529781575977,
      "learning_rate": 8.4183272586351e-06,
      "loss": 337.8066,
      "step": 28860
    },
    {
      "epoch": 0.5552403573386159,
      "grad_norm": 451.15439965981165,
      "learning_rate": 8.412301235315026e-06,
      "loss": 347.0402,
      "step": 28870
    },
    {
      "epoch": 0.5554326816743757,
      "grad_norm": 468.96241894049217,
      "learning_rate": 8.406275803386525e-06,
      "loss": 338.5047,
      "step": 28880
    },
    {
      "epoch": 0.5556250060101355,
      "grad_norm": 432.69849751099804,
      "learning_rate": 8.400250965093968e-06,
      "loss": 337.7101,
      "step": 28890
    },
    {
      "epoch": 0.5558173303458953,
      "grad_norm": 464.24731570275924,
      "learning_rate": 8.394226722681498e-06,
      "loss": 341.856,
      "step": 28900
    },
    {
      "epoch": 0.5560096546816552,
      "grad_norm": 474.9783446731229,
      "learning_rate": 8.38820307839306e-06,
      "loss": 334.5956,
      "step": 28910
    },
    {
      "epoch": 0.556201979017415,
      "grad_norm": 502.81647467693927,
      "learning_rate": 8.382180034472353e-06,
      "loss": 325.1009,
      "step": 28920
    },
    {
      "epoch": 0.5563943033531747,
      "grad_norm": 414.34753733509274,
      "learning_rate": 8.376157593162867e-06,
      "loss": 329.5131,
      "step": 28930
    },
    {
      "epoch": 0.5565866276889346,
      "grad_norm": 462.21592865856906,
      "learning_rate": 8.370135756707853e-06,
      "loss": 336.8699,
      "step": 28940
    },
    {
      "epoch": 0.5567789520246944,
      "grad_norm": 502.0915206721188,
      "learning_rate": 8.364114527350357e-06,
      "loss": 333.491,
      "step": 28950
    },
    {
      "epoch": 0.5569712763604543,
      "grad_norm": 434.6518119218972,
      "learning_rate": 8.358093907333182e-06,
      "loss": 329.48,
      "step": 28960
    },
    {
      "epoch": 0.5571636006962141,
      "grad_norm": 462.289336722964,
      "learning_rate": 8.35207389889891e-06,
      "loss": 339.1186,
      "step": 28970
    },
    {
      "epoch": 0.557355925031974,
      "grad_norm": 455.06554300250497,
      "learning_rate": 8.346054504289888e-06,
      "loss": 339.4001,
      "step": 28980
    },
    {
      "epoch": 0.5575482493677337,
      "grad_norm": 440.7847408197368,
      "learning_rate": 8.34003572574825e-06,
      "loss": 331.0541,
      "step": 28990
    },
    {
      "epoch": 0.5577405737034936,
      "grad_norm": 419.28611684688883,
      "learning_rate": 8.334017565515892e-06,
      "loss": 336.4699,
      "step": 29000
    },
    {
      "epoch": 0.5579328980392534,
      "grad_norm": 440.0149818815394,
      "learning_rate": 8.328000025834472e-06,
      "loss": 323.6228,
      "step": 29010
    },
    {
      "epoch": 0.5581252223750133,
      "grad_norm": 463.7826318548719,
      "learning_rate": 8.321983108945431e-06,
      "loss": 335.2554,
      "step": 29020
    },
    {
      "epoch": 0.558317546710773,
      "grad_norm": 467.4520094011459,
      "learning_rate": 8.315966817089972e-06,
      "loss": 344.2589,
      "step": 29030
    },
    {
      "epoch": 0.5585098710465328,
      "grad_norm": 442.478723335083,
      "learning_rate": 8.309951152509057e-06,
      "loss": 334.4404,
      "step": 29040
    },
    {
      "epoch": 0.5587021953822927,
      "grad_norm": 405.7022535682226,
      "learning_rate": 8.303936117443422e-06,
      "loss": 343.2435,
      "step": 29050
    },
    {
      "epoch": 0.5588945197180525,
      "grad_norm": 407.3883669865574,
      "learning_rate": 8.297921714133576e-06,
      "loss": 338.1101,
      "step": 29060
    },
    {
      "epoch": 0.5590868440538124,
      "grad_norm": 470.5200609320266,
      "learning_rate": 8.291907944819782e-06,
      "loss": 330.1531,
      "step": 29070
    },
    {
      "epoch": 0.5592791683895721,
      "grad_norm": 506.1059406661888,
      "learning_rate": 8.285894811742065e-06,
      "loss": 322.5621,
      "step": 29080
    },
    {
      "epoch": 0.559471492725332,
      "grad_norm": 431.05327168838556,
      "learning_rate": 8.279882317140224e-06,
      "loss": 328.4875,
      "step": 29090
    },
    {
      "epoch": 0.5596638170610918,
      "grad_norm": 424.87231380658955,
      "learning_rate": 8.273870463253813e-06,
      "loss": 335.0859,
      "step": 29100
    },
    {
      "epoch": 0.5598561413968517,
      "grad_norm": 459.6517685777888,
      "learning_rate": 8.267859252322144e-06,
      "loss": 339.4594,
      "step": 29110
    },
    {
      "epoch": 0.5600484657326115,
      "grad_norm": 462.0607954601287,
      "learning_rate": 8.261848686584293e-06,
      "loss": 329.8024,
      "step": 29120
    },
    {
      "epoch": 0.5602407900683714,
      "grad_norm": 452.9882794501342,
      "learning_rate": 8.255838768279106e-06,
      "loss": 332.184,
      "step": 29130
    },
    {
      "epoch": 0.5604331144041311,
      "grad_norm": 447.47040112793513,
      "learning_rate": 8.249829499645167e-06,
      "loss": 334.8988,
      "step": 29140
    },
    {
      "epoch": 0.5606254387398909,
      "grad_norm": 458.17452602336243,
      "learning_rate": 8.243820882920837e-06,
      "loss": 327.148,
      "step": 29150
    },
    {
      "epoch": 0.5608177630756508,
      "grad_norm": 487.41883874216603,
      "learning_rate": 8.23781292034422e-06,
      "loss": 333.2591,
      "step": 29160
    },
    {
      "epoch": 0.5610100874114106,
      "grad_norm": 423.82459507322125,
      "learning_rate": 8.231805614153192e-06,
      "loss": 331.0956,
      "step": 29170
    },
    {
      "epoch": 0.5612024117471704,
      "grad_norm": 442.5060143198106,
      "learning_rate": 8.22579896658537e-06,
      "loss": 330.2205,
      "step": 29180
    },
    {
      "epoch": 0.5613947360829302,
      "grad_norm": 426.49906324196724,
      "learning_rate": 8.219792979878126e-06,
      "loss": 327.4737,
      "step": 29190
    },
    {
      "epoch": 0.5615870604186901,
      "grad_norm": 422.81930245385394,
      "learning_rate": 8.213787656268599e-06,
      "loss": 338.3242,
      "step": 29200
    },
    {
      "epoch": 0.5617793847544499,
      "grad_norm": 462.3170358537627,
      "learning_rate": 8.20778299799367e-06,
      "loss": 334.1542,
      "step": 29210
    },
    {
      "epoch": 0.5619717090902098,
      "grad_norm": 437.1471185988838,
      "learning_rate": 8.201779007289975e-06,
      "loss": 331.3995,
      "step": 29220
    },
    {
      "epoch": 0.5621640334259695,
      "grad_norm": 487.7957088093334,
      "learning_rate": 8.195775686393898e-06,
      "loss": 344.2795,
      "step": 29230
    },
    {
      "epoch": 0.5623563577617294,
      "grad_norm": 478.33005089591705,
      "learning_rate": 8.189773037541585e-06,
      "loss": 336.6861,
      "step": 29240
    },
    {
      "epoch": 0.5625486820974892,
      "grad_norm": 422.10092133872394,
      "learning_rate": 8.183771062968917e-06,
      "loss": 340.3007,
      "step": 29250
    },
    {
      "epoch": 0.562741006433249,
      "grad_norm": 474.0324858154835,
      "learning_rate": 8.177769764911528e-06,
      "loss": 336.0106,
      "step": 29260
    },
    {
      "epoch": 0.5629333307690089,
      "grad_norm": 434.3888681914373,
      "learning_rate": 8.171769145604812e-06,
      "loss": 335.4993,
      "step": 29270
    },
    {
      "epoch": 0.5631256551047686,
      "grad_norm": 408.68330738599025,
      "learning_rate": 8.165769207283891e-06,
      "loss": 329.405,
      "step": 29280
    },
    {
      "epoch": 0.5633179794405285,
      "grad_norm": 467.13611665112876,
      "learning_rate": 8.15976995218365e-06,
      "loss": 334.0891,
      "step": 29290
    },
    {
      "epoch": 0.5635103037762883,
      "grad_norm": 409.14244844029383,
      "learning_rate": 8.153771382538706e-06,
      "loss": 324.47,
      "step": 29300
    },
    {
      "epoch": 0.5637026281120482,
      "grad_norm": 432.7894556210294,
      "learning_rate": 8.147773500583434e-06,
      "loss": 335.9282,
      "step": 29310
    },
    {
      "epoch": 0.563894952447808,
      "grad_norm": 417.8412144481969,
      "learning_rate": 8.141776308551942e-06,
      "loss": 332.9616,
      "step": 29320
    },
    {
      "epoch": 0.5640872767835678,
      "grad_norm": 454.71876988862135,
      "learning_rate": 8.135779808678084e-06,
      "loss": 337.533,
      "step": 29330
    },
    {
      "epoch": 0.5642796011193276,
      "grad_norm": 423.2581077297414,
      "learning_rate": 8.129784003195458e-06,
      "loss": 324.7285,
      "step": 29340
    },
    {
      "epoch": 0.5644719254550875,
      "grad_norm": 441.9834783817504,
      "learning_rate": 8.123788894337405e-06,
      "loss": 332.1972,
      "step": 29350
    },
    {
      "epoch": 0.5646642497908473,
      "grad_norm": 434.08298623716075,
      "learning_rate": 8.117794484337003e-06,
      "loss": 332.6274,
      "step": 29360
    },
    {
      "epoch": 0.5648565741266071,
      "grad_norm": 453.8837857720592,
      "learning_rate": 8.111800775427066e-06,
      "loss": 332.9875,
      "step": 29370
    },
    {
      "epoch": 0.5650488984623669,
      "grad_norm": 443.85682117737525,
      "learning_rate": 8.10580776984016e-06,
      "loss": 332.123,
      "step": 29380
    },
    {
      "epoch": 0.5652412227981267,
      "grad_norm": 400.4770174997833,
      "learning_rate": 8.099815469808573e-06,
      "loss": 332.0791,
      "step": 29390
    },
    {
      "epoch": 0.5654335471338866,
      "grad_norm": 421.0863326202375,
      "learning_rate": 8.093823877564343e-06,
      "loss": 332.9803,
      "step": 29400
    },
    {
      "epoch": 0.5656258714696464,
      "grad_norm": 460.39039016403785,
      "learning_rate": 8.087832995339236e-06,
      "loss": 331.3364,
      "step": 29410
    },
    {
      "epoch": 0.5658181958054063,
      "grad_norm": 436.94490182003267,
      "learning_rate": 8.081842825364756e-06,
      "loss": 325.6589,
      "step": 29420
    },
    {
      "epoch": 0.566010520141166,
      "grad_norm": 435.7743717292608,
      "learning_rate": 8.075853369872149e-06,
      "loss": 332.4927,
      "step": 29430
    },
    {
      "epoch": 0.5662028444769259,
      "grad_norm": 424.79688028392485,
      "learning_rate": 8.069864631092377e-06,
      "loss": 332.5448,
      "step": 29440
    },
    {
      "epoch": 0.5663951688126857,
      "grad_norm": 411.1753534629818,
      "learning_rate": 8.063876611256158e-06,
      "loss": 335.4941,
      "step": 29450
    },
    {
      "epoch": 0.5665874931484456,
      "grad_norm": 401.227188464867,
      "learning_rate": 8.057889312593924e-06,
      "loss": 340.2273,
      "step": 29460
    },
    {
      "epoch": 0.5667798174842054,
      "grad_norm": 3810.649817924996,
      "learning_rate": 8.051902737335847e-06,
      "loss": 347.7162,
      "step": 29470
    },
    {
      "epoch": 0.5669721418199651,
      "grad_norm": 438.02312995201396,
      "learning_rate": 8.045916887711822e-06,
      "loss": 337.5929,
      "step": 29480
    },
    {
      "epoch": 0.567164466155725,
      "grad_norm": 448.4780252859747,
      "learning_rate": 8.039931765951485e-06,
      "loss": 345.2213,
      "step": 29490
    },
    {
      "epoch": 0.5673567904914848,
      "grad_norm": 571.3342952328112,
      "learning_rate": 8.033947374284196e-06,
      "loss": 329.6985,
      "step": 29500
    },
    {
      "epoch": 0.5675491148272447,
      "grad_norm": 439.33405526116576,
      "learning_rate": 8.02796371493904e-06,
      "loss": 338.3913,
      "step": 29510
    },
    {
      "epoch": 0.5677414391630045,
      "grad_norm": 455.9108263313518,
      "learning_rate": 8.021980790144828e-06,
      "loss": 324.158,
      "step": 29520
    },
    {
      "epoch": 0.5679337634987643,
      "grad_norm": 412.031586350393,
      "learning_rate": 8.015998602130107e-06,
      "loss": 329.6155,
      "step": 29530
    },
    {
      "epoch": 0.5681260878345241,
      "grad_norm": 441.49365493333977,
      "learning_rate": 8.01001715312314e-06,
      "loss": 326.0366,
      "step": 29540
    },
    {
      "epoch": 0.568318412170284,
      "grad_norm": 420.2265928528216,
      "learning_rate": 8.00403644535191e-06,
      "loss": 331.1729,
      "step": 29550
    },
    {
      "epoch": 0.5685107365060438,
      "grad_norm": 427.4948133929633,
      "learning_rate": 7.998056481044146e-06,
      "loss": 336.6901,
      "step": 29560
    },
    {
      "epoch": 0.5687030608418037,
      "grad_norm": 506.5082940779469,
      "learning_rate": 7.992077262427282e-06,
      "loss": 343.8321,
      "step": 29570
    },
    {
      "epoch": 0.5688953851775634,
      "grad_norm": 434.22758575124624,
      "learning_rate": 7.986098791728475e-06,
      "loss": 330.4786,
      "step": 29580
    },
    {
      "epoch": 0.5690877095133232,
      "grad_norm": 421.47017215140454,
      "learning_rate": 7.9801210711746e-06,
      "loss": 337.844,
      "step": 29590
    },
    {
      "epoch": 0.5692800338490831,
      "grad_norm": 418.32667221583915,
      "learning_rate": 7.974144102992273e-06,
      "loss": 330.2382,
      "step": 29600
    },
    {
      "epoch": 0.5694723581848429,
      "grad_norm": 512.6053370121507,
      "learning_rate": 7.968167889407813e-06,
      "loss": 328.4573,
      "step": 29610
    },
    {
      "epoch": 0.5696646825206028,
      "grad_norm": 438.324798620883,
      "learning_rate": 7.96219243264725e-06,
      "loss": 335.7818,
      "step": 29620
    },
    {
      "epoch": 0.5698570068563625,
      "grad_norm": 455.7128143073738,
      "learning_rate": 7.956217734936353e-06,
      "loss": 330.2597,
      "step": 29630
    },
    {
      "epoch": 0.5700493311921224,
      "grad_norm": 430.46145668708493,
      "learning_rate": 7.950243798500593e-06,
      "loss": 326.9545,
      "step": 29640
    },
    {
      "epoch": 0.5702416555278822,
      "grad_norm": 441.55785457803626,
      "learning_rate": 7.94427062556517e-06,
      "loss": 336.7078,
      "step": 29650
    },
    {
      "epoch": 0.5704339798636421,
      "grad_norm": 407.74719322133365,
      "learning_rate": 7.938298218354985e-06,
      "loss": 329.5589,
      "step": 29660
    },
    {
      "epoch": 0.5706263041994019,
      "grad_norm": 456.31055397247604,
      "learning_rate": 7.932326579094665e-06,
      "loss": 336.0979,
      "step": 29670
    },
    {
      "epoch": 0.5708186285351617,
      "grad_norm": 450.69316985112437,
      "learning_rate": 7.926355710008545e-06,
      "loss": 332.5213,
      "step": 29680
    },
    {
      "epoch": 0.5710109528709215,
      "grad_norm": 422.9822101382185,
      "learning_rate": 7.920385613320675e-06,
      "loss": 326.7693,
      "step": 29690
    },
    {
      "epoch": 0.5712032772066813,
      "grad_norm": 472.8179066379729,
      "learning_rate": 7.914416291254817e-06,
      "loss": 322.3947,
      "step": 29700
    },
    {
      "epoch": 0.5713956015424412,
      "grad_norm": 420.4229825097595,
      "learning_rate": 7.908447746034447e-06,
      "loss": 325.7185,
      "step": 29710
    },
    {
      "epoch": 0.571587925878201,
      "grad_norm": 449.5150610677103,
      "learning_rate": 7.902479979882749e-06,
      "loss": 341.1461,
      "step": 29720
    },
    {
      "epoch": 0.5717802502139608,
      "grad_norm": 499.1224914926725,
      "learning_rate": 7.896512995022614e-06,
      "loss": 334.9708,
      "step": 29730
    },
    {
      "epoch": 0.5719725745497206,
      "grad_norm": 590.0282859279401,
      "learning_rate": 7.890546793676652e-06,
      "loss": 330.1338,
      "step": 29740
    },
    {
      "epoch": 0.5721648988854805,
      "grad_norm": 454.7010902208668,
      "learning_rate": 7.884581378067171e-06,
      "loss": 334.7914,
      "step": 29750
    },
    {
      "epoch": 0.5723572232212403,
      "grad_norm": 418.45804727090473,
      "learning_rate": 7.878616750416186e-06,
      "loss": 330.687,
      "step": 29760
    },
    {
      "epoch": 0.5725495475570002,
      "grad_norm": 458.1393066528645,
      "learning_rate": 7.872652912945426e-06,
      "loss": 330.6302,
      "step": 29770
    },
    {
      "epoch": 0.5727418718927599,
      "grad_norm": 495.0496315737,
      "learning_rate": 7.866689867876323e-06,
      "loss": 334.8875,
      "step": 29780
    },
    {
      "epoch": 0.5729341962285198,
      "grad_norm": 438.2565273464387,
      "learning_rate": 7.860727617430013e-06,
      "loss": 330.4333,
      "step": 29790
    },
    {
      "epoch": 0.5731265205642796,
      "grad_norm": 412.9805257332131,
      "learning_rate": 7.85476616382733e-06,
      "loss": 320.4414,
      "step": 29800
    },
    {
      "epoch": 0.5733188449000395,
      "grad_norm": 483.069705121499,
      "learning_rate": 7.848805509288824e-06,
      "loss": 341.8112,
      "step": 29810
    },
    {
      "epoch": 0.5735111692357993,
      "grad_norm": 398.53307287294416,
      "learning_rate": 7.84284565603474e-06,
      "loss": 328.7659,
      "step": 29820
    },
    {
      "epoch": 0.573703493571559,
      "grad_norm": 455.5459483468834,
      "learning_rate": 7.83688660628502e-06,
      "loss": 332.784,
      "step": 29830
    },
    {
      "epoch": 0.5738958179073189,
      "grad_norm": 490.62416658254017,
      "learning_rate": 7.830928362259313e-06,
      "loss": 332.3923,
      "step": 29840
    },
    {
      "epoch": 0.5740881422430787,
      "grad_norm": 429.79524937179906,
      "learning_rate": 7.824970926176967e-06,
      "loss": 334.61,
      "step": 29850
    },
    {
      "epoch": 0.5742804665788386,
      "grad_norm": 449.1893593883112,
      "learning_rate": 7.819014300257033e-06,
      "loss": 329.0212,
      "step": 29860
    },
    {
      "epoch": 0.5744727909145984,
      "grad_norm": 414.33619573207307,
      "learning_rate": 7.813058486718252e-06,
      "loss": 336.6774,
      "step": 29870
    },
    {
      "epoch": 0.5746651152503582,
      "grad_norm": 439.29259653225347,
      "learning_rate": 7.80710348777906e-06,
      "loss": 333.8371,
      "step": 29880
    },
    {
      "epoch": 0.574857439586118,
      "grad_norm": 428.49964835785215,
      "learning_rate": 7.801149305657609e-06,
      "loss": 331.8615,
      "step": 29890
    },
    {
      "epoch": 0.5750497639218779,
      "grad_norm": 427.2398552866976,
      "learning_rate": 7.795195942571722e-06,
      "loss": 342.7597,
      "step": 29900
    },
    {
      "epoch": 0.5752420882576377,
      "grad_norm": 425.80215562660857,
      "learning_rate": 7.789243400738934e-06,
      "loss": 330.5996,
      "step": 29910
    },
    {
      "epoch": 0.5754344125933976,
      "grad_norm": 406.52357920872674,
      "learning_rate": 7.783291682376465e-06,
      "loss": 339.7319,
      "step": 29920
    },
    {
      "epoch": 0.5756267369291573,
      "grad_norm": 463.2569794056569,
      "learning_rate": 7.77734078970124e-06,
      "loss": 340.5203,
      "step": 29930
    },
    {
      "epoch": 0.5758190612649171,
      "grad_norm": 444.6726578722199,
      "learning_rate": 7.77139072492986e-06,
      "loss": 330.1363,
      "step": 29940
    },
    {
      "epoch": 0.576011385600677,
      "grad_norm": 442.0133493623101,
      "learning_rate": 7.765441490278625e-06,
      "loss": 331.1776,
      "step": 29950
    },
    {
      "epoch": 0.5762037099364368,
      "grad_norm": 447.5851833126177,
      "learning_rate": 7.759493087963535e-06,
      "loss": 335.1232,
      "step": 29960
    },
    {
      "epoch": 0.5763960342721967,
      "grad_norm": 475.53567280781186,
      "learning_rate": 7.753545520200264e-06,
      "loss": 335.765,
      "step": 29970
    },
    {
      "epoch": 0.5765883586079564,
      "grad_norm": 421.54015641946467,
      "learning_rate": 7.747598789204183e-06,
      "loss": 329.2269,
      "step": 29980
    },
    {
      "epoch": 0.5767806829437163,
      "grad_norm": 426.1310802483252,
      "learning_rate": 7.741652897190355e-06,
      "loss": 336.0074,
      "step": 29990
    },
    {
      "epoch": 0.5769730072794761,
      "grad_norm": 427.86259920965443,
      "learning_rate": 7.735707846373527e-06,
      "loss": 328.8445,
      "step": 30000
    },
    {
      "epoch": 0.577165331615236,
      "grad_norm": 428.41995362601057,
      "learning_rate": 7.72976363896813e-06,
      "loss": 333.8077,
      "step": 30010
    },
    {
      "epoch": 0.5773576559509958,
      "grad_norm": 430.2802609969973,
      "learning_rate": 7.723820277188278e-06,
      "loss": 336.9656,
      "step": 30020
    },
    {
      "epoch": 0.5775499802867556,
      "grad_norm": 440.6500255127248,
      "learning_rate": 7.717877763247787e-06,
      "loss": 330.8104,
      "step": 30030
    },
    {
      "epoch": 0.5777423046225154,
      "grad_norm": 472.56819220412467,
      "learning_rate": 7.711936099360138e-06,
      "loss": 337.4611,
      "step": 30040
    },
    {
      "epoch": 0.5779346289582752,
      "grad_norm": 496.4691959227554,
      "learning_rate": 7.705995287738498e-06,
      "loss": 330.234,
      "step": 30050
    },
    {
      "epoch": 0.5781269532940351,
      "grad_norm": 461.17763073437783,
      "learning_rate": 7.70005533059573e-06,
      "loss": 334.2353,
      "step": 30060
    },
    {
      "epoch": 0.5783192776297948,
      "grad_norm": 449.07961400479104,
      "learning_rate": 7.694116230144368e-06,
      "loss": 330.7339,
      "step": 30070
    },
    {
      "epoch": 0.5785116019655547,
      "grad_norm": 442.34392009174167,
      "learning_rate": 7.688177988596628e-06,
      "loss": 333.8402,
      "step": 30080
    },
    {
      "epoch": 0.5787039263013145,
      "grad_norm": 461.90023247640244,
      "learning_rate": 7.682240608164401e-06,
      "loss": 324.8395,
      "step": 30090
    },
    {
      "epoch": 0.5788962506370744,
      "grad_norm": 401.98428729685395,
      "learning_rate": 7.676304091059273e-06,
      "loss": 329.8856,
      "step": 30100
    },
    {
      "epoch": 0.5790885749728342,
      "grad_norm": 442.8307087132358,
      "learning_rate": 7.670368439492495e-06,
      "loss": 334.4132,
      "step": 30110
    },
    {
      "epoch": 0.579280899308594,
      "grad_norm": 438.18728197992647,
      "learning_rate": 7.664433655674995e-06,
      "loss": 313.7942,
      "step": 30120
    },
    {
      "epoch": 0.5794732236443538,
      "grad_norm": 442.9409576515072,
      "learning_rate": 7.658499741817383e-06,
      "loss": 340.8581,
      "step": 30130
    },
    {
      "epoch": 0.5796655479801137,
      "grad_norm": 429.8861541271232,
      "learning_rate": 7.65256670012995e-06,
      "loss": 326.3264,
      "step": 30140
    },
    {
      "epoch": 0.5798578723158735,
      "grad_norm": 467.8462077252699,
      "learning_rate": 7.646634532822652e-06,
      "loss": 333.515,
      "step": 30150
    },
    {
      "epoch": 0.5800501966516333,
      "grad_norm": 459.2481688825721,
      "learning_rate": 7.64070324210512e-06,
      "loss": 336.135,
      "step": 30160
    },
    {
      "epoch": 0.5802425209873932,
      "grad_norm": 430.40511731683125,
      "learning_rate": 7.634772830186668e-06,
      "loss": 324.1047,
      "step": 30170
    },
    {
      "epoch": 0.5804348453231529,
      "grad_norm": 468.6415184680704,
      "learning_rate": 7.628843299276276e-06,
      "loss": 339.1803,
      "step": 30180
    },
    {
      "epoch": 0.5806271696589128,
      "grad_norm": 401.7519481676133,
      "learning_rate": 7.62291465158259e-06,
      "loss": 325.6971,
      "step": 30190
    },
    {
      "epoch": 0.5808194939946726,
      "grad_norm": 415.3591764446833,
      "learning_rate": 7.616986889313939e-06,
      "loss": 323.2868,
      "step": 30200
    },
    {
      "epoch": 0.5810118183304325,
      "grad_norm": 446.6236767599377,
      "learning_rate": 7.611060014678313e-06,
      "loss": 329.1341,
      "step": 30210
    },
    {
      "epoch": 0.5812041426661922,
      "grad_norm": 459.8590181303199,
      "learning_rate": 7.605134029883381e-06,
      "loss": 329.813,
      "step": 30220
    },
    {
      "epoch": 0.5813964670019521,
      "grad_norm": 445.0618593825332,
      "learning_rate": 7.599208937136465e-06,
      "loss": 329.2379,
      "step": 30230
    },
    {
      "epoch": 0.5815887913377119,
      "grad_norm": 485.4991220989977,
      "learning_rate": 7.593284738644574e-06,
      "loss": 338.6362,
      "step": 30240
    },
    {
      "epoch": 0.5817811156734718,
      "grad_norm": 408.2764466326972,
      "learning_rate": 7.58736143661437e-06,
      "loss": 330.9625,
      "step": 30250
    },
    {
      "epoch": 0.5819734400092316,
      "grad_norm": 411.6901003315401,
      "learning_rate": 7.5814390332521824e-06,
      "loss": 331.4792,
      "step": 30260
    },
    {
      "epoch": 0.5821657643449913,
      "grad_norm": 427.6664496008229,
      "learning_rate": 7.575517530764011e-06,
      "loss": 336.1827,
      "step": 30270
    },
    {
      "epoch": 0.5823580886807512,
      "grad_norm": 409.60128877955236,
      "learning_rate": 7.569596931355517e-06,
      "loss": 330.2883,
      "step": 30280
    },
    {
      "epoch": 0.582550413016511,
      "grad_norm": 411.2913629827442,
      "learning_rate": 7.56367723723203e-06,
      "loss": 328.3345,
      "step": 30290
    },
    {
      "epoch": 0.5827427373522709,
      "grad_norm": 441.0386184492986,
      "learning_rate": 7.557758450598534e-06,
      "loss": 336.6578,
      "step": 30300
    },
    {
      "epoch": 0.5829350616880307,
      "grad_norm": 444.7976225529548,
      "learning_rate": 7.551840573659677e-06,
      "loss": 333.3413,
      "step": 30310
    },
    {
      "epoch": 0.5831273860237906,
      "grad_norm": 434.18355514212993,
      "learning_rate": 7.5459236086197775e-06,
      "loss": 334.238,
      "step": 30320
    },
    {
      "epoch": 0.5833197103595503,
      "grad_norm": 412.83184305261153,
      "learning_rate": 7.5400075576828e-06,
      "loss": 329.6643,
      "step": 30330
    },
    {
      "epoch": 0.5835120346953102,
      "grad_norm": 436.73047040176095,
      "learning_rate": 7.534092423052382e-06,
      "loss": 329.2323,
      "step": 30340
    },
    {
      "epoch": 0.58370435903107,
      "grad_norm": 421.70849754147355,
      "learning_rate": 7.5281782069318075e-06,
      "loss": 335.3367,
      "step": 30350
    },
    {
      "epoch": 0.5838966833668299,
      "grad_norm": 417.6416592533216,
      "learning_rate": 7.522264911524031e-06,
      "loss": 325.445,
      "step": 30360
    },
    {
      "epoch": 0.5840890077025896,
      "grad_norm": 466.31279118258874,
      "learning_rate": 7.516352539031654e-06,
      "loss": 330.2999,
      "step": 30370
    },
    {
      "epoch": 0.5842813320383494,
      "grad_norm": 479.3594140106477,
      "learning_rate": 7.510441091656933e-06,
      "loss": 330.6483,
      "step": 30380
    },
    {
      "epoch": 0.5844736563741093,
      "grad_norm": 430.07724722480975,
      "learning_rate": 7.504530571601792e-06,
      "loss": 331.6849,
      "step": 30390
    },
    {
      "epoch": 0.5846659807098691,
      "grad_norm": 498.1720790483102,
      "learning_rate": 7.498620981067799e-06,
      "loss": 329.8774,
      "step": 30400
    },
    {
      "epoch": 0.584858305045629,
      "grad_norm": 461.2205110539114,
      "learning_rate": 7.492712322256177e-06,
      "loss": 336.4995,
      "step": 30410
    },
    {
      "epoch": 0.5850506293813887,
      "grad_norm": 414.40799409005166,
      "learning_rate": 7.486804597367807e-06,
      "loss": 326.0727,
      "step": 30420
    },
    {
      "epoch": 0.5852429537171486,
      "grad_norm": 439.59222443430093,
      "learning_rate": 7.480897808603219e-06,
      "loss": 325.2597,
      "step": 30430
    },
    {
      "epoch": 0.5854352780529084,
      "grad_norm": 419.77136939642673,
      "learning_rate": 7.474991958162594e-06,
      "loss": 328.0964,
      "step": 30440
    },
    {
      "epoch": 0.5856276023886683,
      "grad_norm": 410.85890223519806,
      "learning_rate": 7.469087048245758e-06,
      "loss": 326.8105,
      "step": 30450
    },
    {
      "epoch": 0.5858199267244281,
      "grad_norm": 479.17213907929494,
      "learning_rate": 7.463183081052201e-06,
      "loss": 340.601,
      "step": 30460
    },
    {
      "epoch": 0.586012251060188,
      "grad_norm": 475.3548038233099,
      "learning_rate": 7.457280058781049e-06,
      "loss": 335.7268,
      "step": 30470
    },
    {
      "epoch": 0.5862045753959477,
      "grad_norm": 420.529168022175,
      "learning_rate": 7.451377983631078e-06,
      "loss": 340.9229,
      "step": 30480
    },
    {
      "epoch": 0.5863968997317075,
      "grad_norm": 463.02557577749565,
      "learning_rate": 7.445476857800717e-06,
      "loss": 329.5683,
      "step": 30490
    },
    {
      "epoch": 0.5865892240674674,
      "grad_norm": 430.83495633025575,
      "learning_rate": 7.439576683488039e-06,
      "loss": 336.5015,
      "step": 30500
    },
    {
      "epoch": 0.5867815484032272,
      "grad_norm": 453.3661164215135,
      "learning_rate": 7.4336774628907604e-06,
      "loss": 334.8364,
      "step": 30510
    },
    {
      "epoch": 0.586973872738987,
      "grad_norm": 397.206072454155,
      "learning_rate": 7.427779198206238e-06,
      "loss": 324.6285,
      "step": 30520
    },
    {
      "epoch": 0.5871661970747468,
      "grad_norm": 428.51349209302424,
      "learning_rate": 7.421881891631487e-06,
      "loss": 331.0998,
      "step": 30530
    },
    {
      "epoch": 0.5873585214105067,
      "grad_norm": 428.6108243354417,
      "learning_rate": 7.415985545363152e-06,
      "loss": 325.7343,
      "step": 30540
    },
    {
      "epoch": 0.5875508457462665,
      "grad_norm": 421.9440933291155,
      "learning_rate": 7.410090161597523e-06,
      "loss": 328.0761,
      "step": 30550
    },
    {
      "epoch": 0.5877431700820264,
      "grad_norm": 446.0539710711119,
      "learning_rate": 7.404195742530533e-06,
      "loss": 326.7984,
      "step": 30560
    },
    {
      "epoch": 0.5879354944177861,
      "grad_norm": 464.76931857488046,
      "learning_rate": 7.398302290357763e-06,
      "loss": 328.7076,
      "step": 30570
    },
    {
      "epoch": 0.588127818753546,
      "grad_norm": 464.6798010441902,
      "learning_rate": 7.392409807274421e-06,
      "loss": 325.0724,
      "step": 30580
    },
    {
      "epoch": 0.5883201430893058,
      "grad_norm": 418.15569085958595,
      "learning_rate": 7.386518295475355e-06,
      "loss": 331.5551,
      "step": 30590
    },
    {
      "epoch": 0.5885124674250656,
      "grad_norm": 428.93737309888206,
      "learning_rate": 7.380627757155065e-06,
      "loss": 333.5645,
      "step": 30600
    },
    {
      "epoch": 0.5887047917608255,
      "grad_norm": 443.75068237274616,
      "learning_rate": 7.374738194507675e-06,
      "loss": 326.6464,
      "step": 30610
    },
    {
      "epoch": 0.5888971160965852,
      "grad_norm": 444.4822789658591,
      "learning_rate": 7.3688496097269494e-06,
      "loss": 328.3985,
      "step": 30620
    },
    {
      "epoch": 0.5890894404323451,
      "grad_norm": 484.85193210810803,
      "learning_rate": 7.362962005006286e-06,
      "loss": 342.7203,
      "step": 30630
    },
    {
      "epoch": 0.5892817647681049,
      "grad_norm": 452.9430165062033,
      "learning_rate": 7.3570753825387275e-06,
      "loss": 321.6017,
      "step": 30640
    },
    {
      "epoch": 0.5894740891038648,
      "grad_norm": 417.8218990676746,
      "learning_rate": 7.35118974451694e-06,
      "loss": 321.0145,
      "step": 30650
    },
    {
      "epoch": 0.5896664134396246,
      "grad_norm": 416.3957295283941,
      "learning_rate": 7.345305093133226e-06,
      "loss": 325.3449,
      "step": 30660
    },
    {
      "epoch": 0.5898587377753844,
      "grad_norm": 413.5927335537093,
      "learning_rate": 7.3394214305795175e-06,
      "loss": 332.4665,
      "step": 30670
    },
    {
      "epoch": 0.5900510621111442,
      "grad_norm": 433.5886633728642,
      "learning_rate": 7.33353875904739e-06,
      "loss": 327.536,
      "step": 30680
    },
    {
      "epoch": 0.5902433864469041,
      "grad_norm": 439.71164942072016,
      "learning_rate": 7.327657080728032e-06,
      "loss": 324.6934,
      "step": 30690
    },
    {
      "epoch": 0.5904357107826639,
      "grad_norm": 435.7450282510498,
      "learning_rate": 7.321776397812279e-06,
      "loss": 326.868,
      "step": 30700
    },
    {
      "epoch": 0.5906280351184237,
      "grad_norm": 452.5244934922471,
      "learning_rate": 7.315896712490584e-06,
      "loss": 332.5115,
      "step": 30710
    },
    {
      "epoch": 0.5908203594541835,
      "grad_norm": 447.61515435948786,
      "learning_rate": 7.310018026953036e-06,
      "loss": 331.0866,
      "step": 30720
    },
    {
      "epoch": 0.5910126837899433,
      "grad_norm": 430.70014246441565,
      "learning_rate": 7.304140343389348e-06,
      "loss": 329.881,
      "step": 30730
    },
    {
      "epoch": 0.5912050081257032,
      "grad_norm": 429.9615360622204,
      "learning_rate": 7.298263663988853e-06,
      "loss": 336.6074,
      "step": 30740
    },
    {
      "epoch": 0.591397332461463,
      "grad_norm": 481.5978794481741,
      "learning_rate": 7.292387990940526e-06,
      "loss": 318.0565,
      "step": 30750
    },
    {
      "epoch": 0.5915896567972229,
      "grad_norm": 415.670224469492,
      "learning_rate": 7.286513326432953e-06,
      "loss": 328.3108,
      "step": 30760
    },
    {
      "epoch": 0.5917819811329826,
      "grad_norm": 422.412987484641,
      "learning_rate": 7.2806396726543526e-06,
      "loss": 341.2282,
      "step": 30770
    },
    {
      "epoch": 0.5919743054687425,
      "grad_norm": 465.98049976748166,
      "learning_rate": 7.2747670317925625e-06,
      "loss": 333.1395,
      "step": 30780
    },
    {
      "epoch": 0.5921666298045023,
      "grad_norm": 477.49473669294923,
      "learning_rate": 7.268895406035046e-06,
      "loss": 326.3879,
      "step": 30790
    },
    {
      "epoch": 0.5923589541402622,
      "grad_norm": 456.04511489001106,
      "learning_rate": 7.263024797568884e-06,
      "loss": 337.6323,
      "step": 30800
    },
    {
      "epoch": 0.592551278476022,
      "grad_norm": 400.7959342469601,
      "learning_rate": 7.257155208580778e-06,
      "loss": 328.681,
      "step": 30810
    },
    {
      "epoch": 0.5927436028117817,
      "grad_norm": 419.09008314305464,
      "learning_rate": 7.251286641257062e-06,
      "loss": 333.4711,
      "step": 30820
    },
    {
      "epoch": 0.5929359271475416,
      "grad_norm": 431.7195446667333,
      "learning_rate": 7.245419097783674e-06,
      "loss": 339.5082,
      "step": 30830
    },
    {
      "epoch": 0.5931282514833014,
      "grad_norm": 420.27025602517847,
      "learning_rate": 7.239552580346181e-06,
      "loss": 319.8995,
      "step": 30840
    },
    {
      "epoch": 0.5933205758190613,
      "grad_norm": 440.9464286333631,
      "learning_rate": 7.233687091129757e-06,
      "loss": 344.2386,
      "step": 30850
    },
    {
      "epoch": 0.593512900154821,
      "grad_norm": 483.89962623035797,
      "learning_rate": 7.227822632319208e-06,
      "loss": 343.4308,
      "step": 30860
    },
    {
      "epoch": 0.5937052244905809,
      "grad_norm": 481.7308087127284,
      "learning_rate": 7.221959206098945e-06,
      "loss": 347.1303,
      "step": 30870
    },
    {
      "epoch": 0.5938975488263407,
      "grad_norm": 448.4888737988553,
      "learning_rate": 7.216096814652992e-06,
      "loss": 346.2199,
      "step": 30880
    },
    {
      "epoch": 0.5940898731621006,
      "grad_norm": 418.0307840420486,
      "learning_rate": 7.210235460165002e-06,
      "loss": 331.0121,
      "step": 30890
    },
    {
      "epoch": 0.5942821974978604,
      "grad_norm": 416.2280929286208,
      "learning_rate": 7.2043751448182275e-06,
      "loss": 328.2097,
      "step": 30900
    },
    {
      "epoch": 0.5944745218336203,
      "grad_norm": 396.2929183727845,
      "learning_rate": 7.198515870795542e-06,
      "loss": 326.6847,
      "step": 30910
    },
    {
      "epoch": 0.59466684616938,
      "grad_norm": 414.9848902286814,
      "learning_rate": 7.192657640279421e-06,
      "loss": 335.425,
      "step": 30920
    },
    {
      "epoch": 0.5948591705051399,
      "grad_norm": 457.7861496970338,
      "learning_rate": 7.18680045545197e-06,
      "loss": 335.1131,
      "step": 30930
    },
    {
      "epoch": 0.5950514948408997,
      "grad_norm": 494.03918143631216,
      "learning_rate": 7.180944318494888e-06,
      "loss": 334.7091,
      "step": 30940
    },
    {
      "epoch": 0.5952438191766595,
      "grad_norm": 455.3685004409763,
      "learning_rate": 7.175089231589485e-06,
      "loss": 334.6876,
      "step": 30950
    },
    {
      "epoch": 0.5954361435124194,
      "grad_norm": 439.6711012667096,
      "learning_rate": 7.1692351969166905e-06,
      "loss": 347.1738,
      "step": 30960
    },
    {
      "epoch": 0.5956284678481791,
      "grad_norm": 433.3878199910252,
      "learning_rate": 7.163382216657033e-06,
      "loss": 326.3027,
      "step": 30970
    },
    {
      "epoch": 0.595820792183939,
      "grad_norm": 409.99590740432757,
      "learning_rate": 7.157530292990654e-06,
      "loss": 329.2366,
      "step": 30980
    },
    {
      "epoch": 0.5960131165196988,
      "grad_norm": 431.52405336260904,
      "learning_rate": 7.151679428097291e-06,
      "loss": 319.1754,
      "step": 30990
    },
    {
      "epoch": 0.5962054408554587,
      "grad_norm": 455.1807812575659,
      "learning_rate": 7.145829624156304e-06,
      "loss": 333.7459,
      "step": 31000
    },
    {
      "epoch": 0.5963977651912185,
      "grad_norm": 493.3453544067815,
      "learning_rate": 7.1399808833466445e-06,
      "loss": 328.3524,
      "step": 31010
    },
    {
      "epoch": 0.5965900895269783,
      "grad_norm": 483.4519052459808,
      "learning_rate": 7.134133207846869e-06,
      "loss": 331.9075,
      "step": 31020
    },
    {
      "epoch": 0.5967824138627381,
      "grad_norm": 460.9509047632126,
      "learning_rate": 7.128286599835139e-06,
      "loss": 337.3895,
      "step": 31030
    },
    {
      "epoch": 0.596974738198498,
      "grad_norm": 424.24732218577697,
      "learning_rate": 7.122441061489228e-06,
      "loss": 318.9668,
      "step": 31040
    },
    {
      "epoch": 0.5971670625342578,
      "grad_norm": 457.798484478798,
      "learning_rate": 7.1165965949864934e-06,
      "loss": 330.7322,
      "step": 31050
    },
    {
      "epoch": 0.5973593868700176,
      "grad_norm": 468.46423289349167,
      "learning_rate": 7.110753202503906e-06,
      "loss": 329.8272,
      "step": 31060
    },
    {
      "epoch": 0.5975517112057774,
      "grad_norm": 452.47958392396356,
      "learning_rate": 7.104910886218036e-06,
      "loss": 332.745,
      "step": 31070
    },
    {
      "epoch": 0.5977440355415372,
      "grad_norm": 442.86853921388683,
      "learning_rate": 7.0990696483050466e-06,
      "loss": 330.2383,
      "step": 31080
    },
    {
      "epoch": 0.5979363598772971,
      "grad_norm": 418.626475384388,
      "learning_rate": 7.093229490940704e-06,
      "loss": 331.1363,
      "step": 31090
    },
    {
      "epoch": 0.5981286842130569,
      "grad_norm": 406.5862014101304,
      "learning_rate": 7.087390416300364e-06,
      "loss": 321.2443,
      "step": 31100
    },
    {
      "epoch": 0.5983210085488168,
      "grad_norm": 439.4052352894715,
      "learning_rate": 7.081552426558995e-06,
      "loss": 339.0512,
      "step": 31110
    },
    {
      "epoch": 0.5985133328845765,
      "grad_norm": 424.5968584083964,
      "learning_rate": 7.075715523891146e-06,
      "loss": 324.1058,
      "step": 31120
    },
    {
      "epoch": 0.5987056572203364,
      "grad_norm": 448.70573702347997,
      "learning_rate": 7.069879710470965e-06,
      "loss": 324.4255,
      "step": 31130
    },
    {
      "epoch": 0.5988979815560962,
      "grad_norm": 435.67281719470986,
      "learning_rate": 7.064044988472204e-06,
      "loss": 320.2699,
      "step": 31140
    },
    {
      "epoch": 0.5990903058918561,
      "grad_norm": 461.8817151359626,
      "learning_rate": 7.058211360068196e-06,
      "loss": 324.3273,
      "step": 31150
    },
    {
      "epoch": 0.5992826302276159,
      "grad_norm": 452.24181285786625,
      "learning_rate": 7.052378827431871e-06,
      "loss": 330.5285,
      "step": 31160
    },
    {
      "epoch": 0.5994749545633756,
      "grad_norm": 475.31707579050004,
      "learning_rate": 7.046547392735747e-06,
      "loss": 322.7347,
      "step": 31170
    },
    {
      "epoch": 0.5996672788991355,
      "grad_norm": 495.86005226998174,
      "learning_rate": 7.040717058151945e-06,
      "loss": 324.2601,
      "step": 31180
    },
    {
      "epoch": 0.5998596032348953,
      "grad_norm": 415.33022435390995,
      "learning_rate": 7.034887825852164e-06,
      "loss": 329.2483,
      "step": 31190
    },
    {
      "epoch": 0.6000519275706552,
      "grad_norm": 436.19998092436464,
      "learning_rate": 7.029059698007699e-06,
      "loss": 320.7651,
      "step": 31200
    },
    {
      "epoch": 0.600244251906415,
      "grad_norm": 471.59220586959736,
      "learning_rate": 7.023232676789424e-06,
      "loss": 341.8868,
      "step": 31210
    },
    {
      "epoch": 0.6004365762421748,
      "grad_norm": 476.8202834855007,
      "learning_rate": 7.01740676436782e-06,
      "loss": 328.4182,
      "step": 31220
    },
    {
      "epoch": 0.6006289005779346,
      "grad_norm": 428.9616607929938,
      "learning_rate": 7.011581962912936e-06,
      "loss": 332.7413,
      "step": 31230
    },
    {
      "epoch": 0.6008212249136945,
      "grad_norm": 463.69521784344965,
      "learning_rate": 7.005758274594412e-06,
      "loss": 331.1642,
      "step": 31240
    },
    {
      "epoch": 0.6010135492494543,
      "grad_norm": 423.72747756075296,
      "learning_rate": 6.999935701581482e-06,
      "loss": 327.5476,
      "step": 31250
    },
    {
      "epoch": 0.6012058735852142,
      "grad_norm": 423.9380522363376,
      "learning_rate": 6.9941142460429555e-06,
      "loss": 326.322,
      "step": 31260
    },
    {
      "epoch": 0.6013981979209739,
      "grad_norm": 430.2591285280863,
      "learning_rate": 6.988293910147229e-06,
      "loss": 328.2301,
      "step": 31270
    },
    {
      "epoch": 0.6015905222567337,
      "grad_norm": 419.2111748804502,
      "learning_rate": 6.982474696062278e-06,
      "loss": 331.0351,
      "step": 31280
    },
    {
      "epoch": 0.6017828465924936,
      "grad_norm": 420.13574881257034,
      "learning_rate": 6.97665660595567e-06,
      "loss": 329.3996,
      "step": 31290
    },
    {
      "epoch": 0.6019751709282534,
      "grad_norm": 423.05504045628146,
      "learning_rate": 6.970839641994545e-06,
      "loss": 325.3394,
      "step": 31300
    },
    {
      "epoch": 0.6021674952640133,
      "grad_norm": 409.03972788519604,
      "learning_rate": 6.965023806345619e-06,
      "loss": 326.1781,
      "step": 31310
    },
    {
      "epoch": 0.602359819599773,
      "grad_norm": 399.1468065778292,
      "learning_rate": 6.959209101175206e-06,
      "loss": 323.4126,
      "step": 31320
    },
    {
      "epoch": 0.6025521439355329,
      "grad_norm": 443.7106738892151,
      "learning_rate": 6.9533955286491805e-06,
      "loss": 333.6226,
      "step": 31330
    },
    {
      "epoch": 0.6027444682712927,
      "grad_norm": 390.7697000691092,
      "learning_rate": 6.947583090933008e-06,
      "loss": 320.8648,
      "step": 31340
    },
    {
      "epoch": 0.6029367926070526,
      "grad_norm": 406.5422917215672,
      "learning_rate": 6.941771790191716e-06,
      "loss": 322.8538,
      "step": 31350
    },
    {
      "epoch": 0.6031291169428123,
      "grad_norm": 418.194064783171,
      "learning_rate": 6.9359616285899266e-06,
      "loss": 328.5125,
      "step": 31360
    },
    {
      "epoch": 0.6033214412785722,
      "grad_norm": 445.7989038206677,
      "learning_rate": 6.930152608291829e-06,
      "loss": 337.901,
      "step": 31370
    },
    {
      "epoch": 0.603513765614332,
      "grad_norm": 423.173245012053,
      "learning_rate": 6.924344731461179e-06,
      "loss": 319.9932,
      "step": 31380
    },
    {
      "epoch": 0.6037060899500918,
      "grad_norm": 411.5581267282828,
      "learning_rate": 6.918538000261325e-06,
      "loss": 324.1152,
      "step": 31390
    },
    {
      "epoch": 0.6038984142858517,
      "grad_norm": 472.8986853644844,
      "learning_rate": 6.912732416855171e-06,
      "loss": 335.1744,
      "step": 31400
    },
    {
      "epoch": 0.6040907386216114,
      "grad_norm": 476.3325024728093,
      "learning_rate": 6.906927983405207e-06,
      "loss": 336.0696,
      "step": 31410
    },
    {
      "epoch": 0.6042830629573713,
      "grad_norm": 408.8631082548548,
      "learning_rate": 6.901124702073481e-06,
      "loss": 325.908,
      "step": 31420
    },
    {
      "epoch": 0.6044753872931311,
      "grad_norm": 430.6616833450127,
      "learning_rate": 6.895322575021628e-06,
      "loss": 333.273,
      "step": 31430
    },
    {
      "epoch": 0.604667711628891,
      "grad_norm": 434.6259514020785,
      "learning_rate": 6.88952160441084e-06,
      "loss": 334.1016,
      "step": 31440
    },
    {
      "epoch": 0.6048600359646508,
      "grad_norm": 436.98857614808065,
      "learning_rate": 6.8837217924018825e-06,
      "loss": 327.0335,
      "step": 31450
    },
    {
      "epoch": 0.6050523603004107,
      "grad_norm": 434.1359932462219,
      "learning_rate": 6.877923141155087e-06,
      "loss": 336.8191,
      "step": 31460
    },
    {
      "epoch": 0.6052446846361704,
      "grad_norm": 420.9197192491584,
      "learning_rate": 6.87212565283036e-06,
      "loss": 322.0309,
      "step": 31470
    },
    {
      "epoch": 0.6054370089719303,
      "grad_norm": 459.55648391653125,
      "learning_rate": 6.8663293295871715e-06,
      "loss": 322.854,
      "step": 31480
    },
    {
      "epoch": 0.6056293333076901,
      "grad_norm": 409.7728528591732,
      "learning_rate": 6.86053417358455e-06,
      "loss": 333.7287,
      "step": 31490
    },
    {
      "epoch": 0.6058216576434499,
      "grad_norm": 424.42545183647627,
      "learning_rate": 6.854740186981102e-06,
      "loss": 327.6027,
      "step": 31500
    },
    {
      "epoch": 0.6060139819792097,
      "grad_norm": 431.8624804371158,
      "learning_rate": 6.848947371934989e-06,
      "loss": 333.8613,
      "step": 31510
    },
    {
      "epoch": 0.6062063063149695,
      "grad_norm": 440.882114932241,
      "learning_rate": 6.843155730603939e-06,
      "loss": 335.2275,
      "step": 31520
    },
    {
      "epoch": 0.6063986306507294,
      "grad_norm": 410.87712071861705,
      "learning_rate": 6.837365265145237e-06,
      "loss": 323.1684,
      "step": 31530
    },
    {
      "epoch": 0.6065909549864892,
      "grad_norm": 441.7371065461392,
      "learning_rate": 6.831575977715745e-06,
      "loss": 332.0648,
      "step": 31540
    },
    {
      "epoch": 0.6067832793222491,
      "grad_norm": 418.2044710315922,
      "learning_rate": 6.825787870471873e-06,
      "loss": 325.2839,
      "step": 31550
    },
    {
      "epoch": 0.6069756036580088,
      "grad_norm": 477.9637089706895,
      "learning_rate": 6.820000945569592e-06,
      "loss": 325.3218,
      "step": 31560
    },
    {
      "epoch": 0.6071679279937687,
      "grad_norm": 478.63470381819985,
      "learning_rate": 6.814215205164444e-06,
      "loss": 337.3299,
      "step": 31570
    },
    {
      "epoch": 0.6073602523295285,
      "grad_norm": 431.2063649604528,
      "learning_rate": 6.808430651411518e-06,
      "loss": 326.4344,
      "step": 31580
    },
    {
      "epoch": 0.6075525766652884,
      "grad_norm": 437.5634034043336,
      "learning_rate": 6.802647286465461e-06,
      "loss": 328.3208,
      "step": 31590
    },
    {
      "epoch": 0.6077449010010482,
      "grad_norm": 443.68659553090095,
      "learning_rate": 6.796865112480482e-06,
      "loss": 336.2546,
      "step": 31600
    },
    {
      "epoch": 0.6079372253368079,
      "grad_norm": 410.68802734590344,
      "learning_rate": 6.79108413161035e-06,
      "loss": 331.4933,
      "step": 31610
    },
    {
      "epoch": 0.6081295496725678,
      "grad_norm": 450.82071981640553,
      "learning_rate": 6.785304346008381e-06,
      "loss": 331.3573,
      "step": 31620
    },
    {
      "epoch": 0.6083218740083276,
      "grad_norm": 413.8835795348315,
      "learning_rate": 6.779525757827452e-06,
      "loss": 321.4032,
      "step": 31630
    },
    {
      "epoch": 0.6085141983440875,
      "grad_norm": 407.97395421733364,
      "learning_rate": 6.773748369219986e-06,
      "loss": 331.3999,
      "step": 31640
    },
    {
      "epoch": 0.6087065226798473,
      "grad_norm": 446.3660351781709,
      "learning_rate": 6.767972182337974e-06,
      "loss": 338.6244,
      "step": 31650
    },
    {
      "epoch": 0.6088988470156071,
      "grad_norm": 412.42102182963106,
      "learning_rate": 6.762197199332945e-06,
      "loss": 326.3635,
      "step": 31660
    },
    {
      "epoch": 0.6090911713513669,
      "grad_norm": 417.5113755716868,
      "learning_rate": 6.756423422355981e-06,
      "loss": 320.6295,
      "step": 31670
    },
    {
      "epoch": 0.6092834956871268,
      "grad_norm": 396.98151134937456,
      "learning_rate": 6.750650853557728e-06,
      "loss": 321.2423,
      "step": 31680
    },
    {
      "epoch": 0.6094758200228866,
      "grad_norm": 426.87495631595505,
      "learning_rate": 6.744879495088364e-06,
      "loss": 331.0351,
      "step": 31690
    },
    {
      "epoch": 0.6096681443586465,
      "grad_norm": 429.4301663260041,
      "learning_rate": 6.7391093490976285e-06,
      "loss": 317.8627,
      "step": 31700
    },
    {
      "epoch": 0.6098604686944062,
      "grad_norm": 438.9367309224993,
      "learning_rate": 6.7333404177348036e-06,
      "loss": 318.1759,
      "step": 31710
    },
    {
      "epoch": 0.610052793030166,
      "grad_norm": 421.5808481922939,
      "learning_rate": 6.727572703148726e-06,
      "loss": 321.6035,
      "step": 31720
    },
    {
      "epoch": 0.6102451173659259,
      "grad_norm": 421.2256198499856,
      "learning_rate": 6.721806207487769e-06,
      "loss": 330.6125,
      "step": 31730
    },
    {
      "epoch": 0.6104374417016857,
      "grad_norm": 426.228325026148,
      "learning_rate": 6.716040932899857e-06,
      "loss": 337.3397,
      "step": 31740
    },
    {
      "epoch": 0.6106297660374456,
      "grad_norm": 436.38767737451286,
      "learning_rate": 6.710276881532463e-06,
      "loss": 328.5183,
      "step": 31750
    },
    {
      "epoch": 0.6108220903732053,
      "grad_norm": 410.4774005376651,
      "learning_rate": 6.704514055532597e-06,
      "loss": 331.1836,
      "step": 31760
    },
    {
      "epoch": 0.6110144147089652,
      "grad_norm": 448.99529407431874,
      "learning_rate": 6.698752457046822e-06,
      "loss": 320.2948,
      "step": 31770
    },
    {
      "epoch": 0.611206739044725,
      "grad_norm": 453.78535420099576,
      "learning_rate": 6.692992088221231e-06,
      "loss": 336.2243,
      "step": 31780
    },
    {
      "epoch": 0.6113990633804849,
      "grad_norm": 432.39546381511616,
      "learning_rate": 6.687232951201473e-06,
      "loss": 321.0264,
      "step": 31790
    },
    {
      "epoch": 0.6115913877162447,
      "grad_norm": 431.60681963290085,
      "learning_rate": 6.681475048132729e-06,
      "loss": 337.4963,
      "step": 31800
    },
    {
      "epoch": 0.6117837120520045,
      "grad_norm": 412.00998093735575,
      "learning_rate": 6.675718381159719e-06,
      "loss": 325.3441,
      "step": 31810
    },
    {
      "epoch": 0.6119760363877643,
      "grad_norm": 414.27886611671755,
      "learning_rate": 6.6699629524267114e-06,
      "loss": 321.3202,
      "step": 31820
    },
    {
      "epoch": 0.6121683607235241,
      "grad_norm": 429.11378557268085,
      "learning_rate": 6.664208764077507e-06,
      "loss": 335.3874,
      "step": 31830
    },
    {
      "epoch": 0.612360685059284,
      "grad_norm": 424.95879573324083,
      "learning_rate": 6.658455818255445e-06,
      "loss": 319.7482,
      "step": 31840
    },
    {
      "epoch": 0.6125530093950438,
      "grad_norm": 409.34126518998613,
      "learning_rate": 6.652704117103401e-06,
      "loss": 321.8787,
      "step": 31850
    },
    {
      "epoch": 0.6127453337308036,
      "grad_norm": 413.338977258228,
      "learning_rate": 6.646953662763796e-06,
      "loss": 338.9022,
      "step": 31860
    },
    {
      "epoch": 0.6129376580665634,
      "grad_norm": 432.3472601520748,
      "learning_rate": 6.6412044573785725e-06,
      "loss": 328.2723,
      "step": 31870
    },
    {
      "epoch": 0.6131299824023233,
      "grad_norm": 416.94271118525955,
      "learning_rate": 6.635456503089217e-06,
      "loss": 326.2529,
      "step": 31880
    },
    {
      "epoch": 0.6133223067380831,
      "grad_norm": 520.2741569123933,
      "learning_rate": 6.6297098020367435e-06,
      "loss": 344.4937,
      "step": 31890
    },
    {
      "epoch": 0.613514631073843,
      "grad_norm": 414.0172107168599,
      "learning_rate": 6.623964356361707e-06,
      "loss": 342.6946,
      "step": 31900
    },
    {
      "epoch": 0.6137069554096027,
      "grad_norm": 447.82016238858694,
      "learning_rate": 6.618220168204193e-06,
      "loss": 325.3615,
      "step": 31910
    },
    {
      "epoch": 0.6138992797453626,
      "grad_norm": 406.40458600547805,
      "learning_rate": 6.6124772397038115e-06,
      "loss": 328.425,
      "step": 31920
    },
    {
      "epoch": 0.6140916040811224,
      "grad_norm": 428.2966682996343,
      "learning_rate": 6.606735572999714e-06,
      "loss": 336.6029,
      "step": 31930
    },
    {
      "epoch": 0.6142839284168823,
      "grad_norm": 404.30106573083583,
      "learning_rate": 6.600995170230575e-06,
      "loss": 316.1977,
      "step": 31940
    },
    {
      "epoch": 0.6144762527526421,
      "grad_norm": 440.7107669034454,
      "learning_rate": 6.595256033534598e-06,
      "loss": 322.2435,
      "step": 31950
    },
    {
      "epoch": 0.6146685770884018,
      "grad_norm": 450.99281026713794,
      "learning_rate": 6.589518165049514e-06,
      "loss": 323.5361,
      "step": 31960
    },
    {
      "epoch": 0.6148609014241617,
      "grad_norm": 434.4686367799548,
      "learning_rate": 6.5837815669125906e-06,
      "loss": 327.8862,
      "step": 31970
    },
    {
      "epoch": 0.6150532257599215,
      "grad_norm": 406.74916648632023,
      "learning_rate": 6.5780462412606124e-06,
      "loss": 313.7675,
      "step": 31980
    },
    {
      "epoch": 0.6152455500956814,
      "grad_norm": 413.2782229599621,
      "learning_rate": 6.572312190229895e-06,
      "loss": 324.2452,
      "step": 31990
    },
    {
      "epoch": 0.6154378744314412,
      "grad_norm": 428.5324791770067,
      "learning_rate": 6.5665794159562734e-06,
      "loss": 322.2049,
      "step": 32000
    },
    {
      "epoch": 0.615630198767201,
      "grad_norm": 435.7705616014478,
      "learning_rate": 6.560847920575118e-06,
      "loss": 318.6313,
      "step": 32010
    },
    {
      "epoch": 0.6158225231029608,
      "grad_norm": 413.6048996748444,
      "learning_rate": 6.5551177062213126e-06,
      "loss": 333.2631,
      "step": 32020
    },
    {
      "epoch": 0.6160148474387207,
      "grad_norm": 410.4890441406439,
      "learning_rate": 6.5493887750292616e-06,
      "loss": 324.2066,
      "step": 32030
    },
    {
      "epoch": 0.6162071717744805,
      "grad_norm": 401.06790112767646,
      "learning_rate": 6.54366112913291e-06,
      "loss": 337.9793,
      "step": 32040
    },
    {
      "epoch": 0.6163994961102404,
      "grad_norm": 456.3108819641096,
      "learning_rate": 6.537934770665701e-06,
      "loss": 334.1323,
      "step": 32050
    },
    {
      "epoch": 0.6165918204460001,
      "grad_norm": 421.1389596570581,
      "learning_rate": 6.532209701760615e-06,
      "loss": 340.9781,
      "step": 32060
    },
    {
      "epoch": 0.6167841447817599,
      "grad_norm": 426.95528614319517,
      "learning_rate": 6.526485924550138e-06,
      "loss": 332.3481,
      "step": 32070
    },
    {
      "epoch": 0.6169764691175198,
      "grad_norm": 573.6889626483983,
      "learning_rate": 6.520763441166291e-06,
      "loss": 331.3861,
      "step": 32080
    },
    {
      "epoch": 0.6171687934532796,
      "grad_norm": 440.45618569058104,
      "learning_rate": 6.515042253740601e-06,
      "loss": 312.4993,
      "step": 32090
    },
    {
      "epoch": 0.6173611177890395,
      "grad_norm": 447.4479672089497,
      "learning_rate": 6.509322364404112e-06,
      "loss": 331.0521,
      "step": 32100
    },
    {
      "epoch": 0.6175534421247992,
      "grad_norm": 425.50743641148125,
      "learning_rate": 6.503603775287395e-06,
      "loss": 327.4585,
      "step": 32110
    },
    {
      "epoch": 0.6177457664605591,
      "grad_norm": 428.49746655448627,
      "learning_rate": 6.497886488520524e-06,
      "loss": 323.4842,
      "step": 32120
    },
    {
      "epoch": 0.6179380907963189,
      "grad_norm": 457.8434213004478,
      "learning_rate": 6.4921705062331e-06,
      "loss": 335.378,
      "step": 32130
    },
    {
      "epoch": 0.6181304151320788,
      "grad_norm": 442.70575999445475,
      "learning_rate": 6.486455830554224e-06,
      "loss": 320.6056,
      "step": 32140
    },
    {
      "epoch": 0.6183227394678386,
      "grad_norm": 594.1806569317398,
      "learning_rate": 6.4807424636125285e-06,
      "loss": 326.3652,
      "step": 32150
    },
    {
      "epoch": 0.6185150638035984,
      "grad_norm": 431.1665881155747,
      "learning_rate": 6.475030407536141e-06,
      "loss": 328.5355,
      "step": 32160
    },
    {
      "epoch": 0.6187073881393582,
      "grad_norm": 421.0350598890005,
      "learning_rate": 6.469319664452709e-06,
      "loss": 326.7443,
      "step": 32170
    },
    {
      "epoch": 0.618899712475118,
      "grad_norm": 409.23853737120083,
      "learning_rate": 6.463610236489391e-06,
      "loss": 319.5542,
      "step": 32180
    },
    {
      "epoch": 0.6190920368108779,
      "grad_norm": 456.83870606542223,
      "learning_rate": 6.457902125772854e-06,
      "loss": 327.7158,
      "step": 32190
    },
    {
      "epoch": 0.6192843611466377,
      "grad_norm": 453.99404556395046,
      "learning_rate": 6.452195334429277e-06,
      "loss": 329.5449,
      "step": 32200
    },
    {
      "epoch": 0.6194766854823975,
      "grad_norm": 475.0510378981282,
      "learning_rate": 6.446489864584341e-06,
      "loss": 322.3272,
      "step": 32210
    },
    {
      "epoch": 0.6196690098181573,
      "grad_norm": 425.11870277903165,
      "learning_rate": 6.440785718363245e-06,
      "loss": 321.6365,
      "step": 32220
    },
    {
      "epoch": 0.6198613341539172,
      "grad_norm": 433.46603621909895,
      "learning_rate": 6.435082897890688e-06,
      "loss": 324.7147,
      "step": 32230
    },
    {
      "epoch": 0.620053658489677,
      "grad_norm": 414.7375458168306,
      "learning_rate": 6.429381405290873e-06,
      "loss": 323.154,
      "step": 32240
    },
    {
      "epoch": 0.6202459828254369,
      "grad_norm": 452.5861336513786,
      "learning_rate": 6.4236812426875124e-06,
      "loss": 327.5643,
      "step": 32250
    },
    {
      "epoch": 0.6204383071611966,
      "grad_norm": 469.30850682635673,
      "learning_rate": 6.4179824122038244e-06,
      "loss": 327.8253,
      "step": 32260
    },
    {
      "epoch": 0.6206306314969565,
      "grad_norm": 451.42769607913056,
      "learning_rate": 6.412284915962532e-06,
      "loss": 330.3799,
      "step": 32270
    },
    {
      "epoch": 0.6208229558327163,
      "grad_norm": 410.43366971821615,
      "learning_rate": 6.406588756085849e-06,
      "loss": 320.7363,
      "step": 32280
    },
    {
      "epoch": 0.6210152801684761,
      "grad_norm": 418.3653168917956,
      "learning_rate": 6.400893934695514e-06,
      "loss": 323.9497,
      "step": 32290
    },
    {
      "epoch": 0.621207604504236,
      "grad_norm": 455.3732134667118,
      "learning_rate": 6.395200453912747e-06,
      "loss": 330.9363,
      "step": 32300
    },
    {
      "epoch": 0.6213999288399957,
      "grad_norm": 414.77459704481186,
      "learning_rate": 6.389508315858272e-06,
      "loss": 328.9256,
      "step": 32310
    },
    {
      "epoch": 0.6215922531757556,
      "grad_norm": 444.7560679248882,
      "learning_rate": 6.38381752265232e-06,
      "loss": 325.5793,
      "step": 32320
    },
    {
      "epoch": 0.6217845775115154,
      "grad_norm": 465.0531572072124,
      "learning_rate": 6.378128076414619e-06,
      "loss": 330.5696,
      "step": 32330
    },
    {
      "epoch": 0.6219769018472753,
      "grad_norm": 468.5882004724719,
      "learning_rate": 6.372439979264393e-06,
      "loss": 332.7551,
      "step": 32340
    },
    {
      "epoch": 0.622169226183035,
      "grad_norm": 409.0557983184617,
      "learning_rate": 6.3667532333203655e-06,
      "loss": 319.2977,
      "step": 32350
    },
    {
      "epoch": 0.6223615505187949,
      "grad_norm": 432.3491555331506,
      "learning_rate": 6.361067840700747e-06,
      "loss": 322.4629,
      "step": 32360
    },
    {
      "epoch": 0.6225538748545547,
      "grad_norm": 423.5721209957189,
      "learning_rate": 6.355383803523265e-06,
      "loss": 324.1534,
      "step": 32370
    },
    {
      "epoch": 0.6227461991903146,
      "grad_norm": 441.7958215177036,
      "learning_rate": 6.349701123905123e-06,
      "loss": 326.1752,
      "step": 32380
    },
    {
      "epoch": 0.6229385235260744,
      "grad_norm": 418.3116862715124,
      "learning_rate": 6.344019803963021e-06,
      "loss": 328.9536,
      "step": 32390
    },
    {
      "epoch": 0.6231308478618341,
      "grad_norm": 418.21228146984464,
      "learning_rate": 6.338339845813164e-06,
      "loss": 324.4175,
      "step": 32400
    },
    {
      "epoch": 0.623323172197594,
      "grad_norm": 424.9867330009795,
      "learning_rate": 6.332661251571241e-06,
      "loss": 321.1011,
      "step": 32410
    },
    {
      "epoch": 0.6235154965333538,
      "grad_norm": 429.0430428008001,
      "learning_rate": 6.326984023352435e-06,
      "loss": 330.4229,
      "step": 32420
    },
    {
      "epoch": 0.6237078208691137,
      "grad_norm": 393.81594969739876,
      "learning_rate": 6.321308163271413e-06,
      "loss": 324.0193,
      "step": 32430
    },
    {
      "epoch": 0.6239001452048735,
      "grad_norm": 456.6985511449763,
      "learning_rate": 6.315633673442349e-06,
      "loss": 324.1999,
      "step": 32440
    },
    {
      "epoch": 0.6240924695406334,
      "grad_norm": 419.92969770799095,
      "learning_rate": 6.309960555978894e-06,
      "loss": 320.1714,
      "step": 32450
    },
    {
      "epoch": 0.6242847938763931,
      "grad_norm": 417.7106530617901,
      "learning_rate": 6.304288812994183e-06,
      "loss": 321.9302,
      "step": 32460
    },
    {
      "epoch": 0.624477118212153,
      "grad_norm": 407.2339064814913,
      "learning_rate": 6.298618446600856e-06,
      "loss": 334.5645,
      "step": 32470
    },
    {
      "epoch": 0.6246694425479128,
      "grad_norm": 422.27652461561996,
      "learning_rate": 6.292949458911029e-06,
      "loss": 326.5539,
      "step": 32480
    },
    {
      "epoch": 0.6248617668836727,
      "grad_norm": 432.9597248352087,
      "learning_rate": 6.287281852036304e-06,
      "loss": 318.7616,
      "step": 32490
    },
    {
      "epoch": 0.6250540912194325,
      "grad_norm": 416.9766790941515,
      "learning_rate": 6.2816156280877675e-06,
      "loss": 322.7292,
      "step": 32500
    },
    {
      "epoch": 0.6252464155551922,
      "grad_norm": 408.8695050778491,
      "learning_rate": 6.275950789176002e-06,
      "loss": 328.5652,
      "step": 32510
    },
    {
      "epoch": 0.6254387398909521,
      "grad_norm": 406.750794698313,
      "learning_rate": 6.270287337411064e-06,
      "loss": 321.6231,
      "step": 32520
    },
    {
      "epoch": 0.6256310642267119,
      "grad_norm": 434.35189119278624,
      "learning_rate": 6.264625274902492e-06,
      "loss": 328.9104,
      "step": 32530
    },
    {
      "epoch": 0.6258233885624718,
      "grad_norm": 441.217565698764,
      "learning_rate": 6.258964603759311e-06,
      "loss": 332.9562,
      "step": 32540
    },
    {
      "epoch": 0.6260157128982315,
      "grad_norm": 426.77229867822825,
      "learning_rate": 6.2533053260900345e-06,
      "loss": 322.5945,
      "step": 32550
    },
    {
      "epoch": 0.6262080372339914,
      "grad_norm": 483.1264069683457,
      "learning_rate": 6.247647444002644e-06,
      "loss": 327.2905,
      "step": 32560
    },
    {
      "epoch": 0.6264003615697512,
      "grad_norm": 436.6549684532777,
      "learning_rate": 6.241990959604607e-06,
      "loss": 334.4606,
      "step": 32570
    },
    {
      "epoch": 0.6265926859055111,
      "grad_norm": 401.2974328382389,
      "learning_rate": 6.2363358750028745e-06,
      "loss": 325.6229,
      "step": 32580
    },
    {
      "epoch": 0.6267850102412709,
      "grad_norm": 406.44672587720225,
      "learning_rate": 6.23068219230387e-06,
      "loss": 318.5489,
      "step": 32590
    },
    {
      "epoch": 0.6269773345770308,
      "grad_norm": 455.20375356590705,
      "learning_rate": 6.2250299136134925e-06,
      "loss": 319.4099,
      "step": 32600
    },
    {
      "epoch": 0.6271696589127905,
      "grad_norm": 418.82334411779806,
      "learning_rate": 6.219379041037128e-06,
      "loss": 326.8413,
      "step": 32610
    },
    {
      "epoch": 0.6273619832485503,
      "grad_norm": 439.19329369503697,
      "learning_rate": 6.213729576679632e-06,
      "loss": 323.3758,
      "step": 32620
    },
    {
      "epoch": 0.6275543075843102,
      "grad_norm": 416.2625185128422,
      "learning_rate": 6.208081522645339e-06,
      "loss": 318.5643,
      "step": 32630
    },
    {
      "epoch": 0.62774663192007,
      "grad_norm": 445.4966848300152,
      "learning_rate": 6.202434881038048e-06,
      "loss": 323.1161,
      "step": 32640
    },
    {
      "epoch": 0.6279389562558298,
      "grad_norm": 403.7901714338289,
      "learning_rate": 6.196789653961048e-06,
      "loss": 321.4052,
      "step": 32650
    },
    {
      "epoch": 0.6281312805915896,
      "grad_norm": 402.4884676658384,
      "learning_rate": 6.191145843517093e-06,
      "loss": 322.4316,
      "step": 32660
    },
    {
      "epoch": 0.6283236049273495,
      "grad_norm": 471.4715054154106,
      "learning_rate": 6.185503451808401e-06,
      "loss": 325.7773,
      "step": 32670
    },
    {
      "epoch": 0.6285159292631093,
      "grad_norm": 424.61665334536804,
      "learning_rate": 6.1798624809366755e-06,
      "loss": 333.7152,
      "step": 32680
    },
    {
      "epoch": 0.6287082535988692,
      "grad_norm": 463.05504565767166,
      "learning_rate": 6.174222933003084e-06,
      "loss": 326.8595,
      "step": 32690
    },
    {
      "epoch": 0.628900577934629,
      "grad_norm": 447.63783470891326,
      "learning_rate": 6.168584810108269e-06,
      "loss": 328.4714,
      "step": 32700
    },
    {
      "epoch": 0.6290929022703888,
      "grad_norm": 473.41771899779656,
      "learning_rate": 6.162948114352328e-06,
      "loss": 335.0917,
      "step": 32710
    },
    {
      "epoch": 0.6292852266061486,
      "grad_norm": 465.20344187668286,
      "learning_rate": 6.157312847834848e-06,
      "loss": 324.3855,
      "step": 32720
    },
    {
      "epoch": 0.6294775509419084,
      "grad_norm": 419.9382018746983,
      "learning_rate": 6.1516790126548695e-06,
      "loss": 321.8709,
      "step": 32730
    },
    {
      "epoch": 0.6296698752776683,
      "grad_norm": 432.3435457974304,
      "learning_rate": 6.146046610910899e-06,
      "loss": 326.0685,
      "step": 32740
    },
    {
      "epoch": 0.629862199613428,
      "grad_norm": 410.18715744874953,
      "learning_rate": 6.140415644700915e-06,
      "loss": 339.1354,
      "step": 32750
    },
    {
      "epoch": 0.6300545239491879,
      "grad_norm": 440.33709101015756,
      "learning_rate": 6.13478611612236e-06,
      "loss": 328.9889,
      "step": 32760
    },
    {
      "epoch": 0.6302468482849477,
      "grad_norm": 402.57351663000355,
      "learning_rate": 6.129158027272144e-06,
      "loss": 323.6507,
      "step": 32770
    },
    {
      "epoch": 0.6304391726207076,
      "grad_norm": 452.5979517759669,
      "learning_rate": 6.123531380246632e-06,
      "loss": 320.9756,
      "step": 32780
    },
    {
      "epoch": 0.6306314969564674,
      "grad_norm": 404.4887640766371,
      "learning_rate": 6.117906177141657e-06,
      "loss": 328.8006,
      "step": 32790
    },
    {
      "epoch": 0.6308238212922272,
      "grad_norm": 462.76630626964396,
      "learning_rate": 6.112282420052518e-06,
      "loss": 321.5611,
      "step": 32800
    },
    {
      "epoch": 0.631016145627987,
      "grad_norm": 419.70725828797083,
      "learning_rate": 6.106660111073969e-06,
      "loss": 327.3932,
      "step": 32810
    },
    {
      "epoch": 0.6312084699637469,
      "grad_norm": 406.57270690324384,
      "learning_rate": 6.101039252300227e-06,
      "loss": 328.7421,
      "step": 32820
    },
    {
      "epoch": 0.6314007942995067,
      "grad_norm": 424.11730292183296,
      "learning_rate": 6.095419845824971e-06,
      "loss": 323.4531,
      "step": 32830
    },
    {
      "epoch": 0.6315931186352665,
      "grad_norm": 435.44789791149435,
      "learning_rate": 6.089801893741338e-06,
      "loss": 325.521,
      "step": 32840
    },
    {
      "epoch": 0.6317854429710263,
      "grad_norm": 406.92073548289306,
      "learning_rate": 6.084185398141921e-06,
      "loss": 324.5324,
      "step": 32850
    },
    {
      "epoch": 0.6319777673067861,
      "grad_norm": 445.37035709504715,
      "learning_rate": 6.078570361118768e-06,
      "loss": 328.6068,
      "step": 32860
    },
    {
      "epoch": 0.632170091642546,
      "grad_norm": 436.58632830997254,
      "learning_rate": 6.072956784763393e-06,
      "loss": 324.0492,
      "step": 32870
    },
    {
      "epoch": 0.6323624159783058,
      "grad_norm": 395.89775800273503,
      "learning_rate": 6.067344671166757e-06,
      "loss": 327.2833,
      "step": 32880
    },
    {
      "epoch": 0.6325547403140657,
      "grad_norm": 437.50506006442527,
      "learning_rate": 6.061734022419281e-06,
      "loss": 326.9905,
      "step": 32890
    },
    {
      "epoch": 0.6327470646498254,
      "grad_norm": 416.31786384346094,
      "learning_rate": 6.056124840610839e-06,
      "loss": 313.8626,
      "step": 32900
    },
    {
      "epoch": 0.6329393889855853,
      "grad_norm": 463.5627340464092,
      "learning_rate": 6.050517127830761e-06,
      "loss": 326.8129,
      "step": 32910
    },
    {
      "epoch": 0.6331317133213451,
      "grad_norm": 416.54406993323664,
      "learning_rate": 6.044910886167825e-06,
      "loss": 336.6158,
      "step": 32920
    },
    {
      "epoch": 0.633324037657105,
      "grad_norm": 439.75612499363956,
      "learning_rate": 6.03930611771026e-06,
      "loss": 321.1817,
      "step": 32930
    },
    {
      "epoch": 0.6335163619928648,
      "grad_norm": 462.1824301255701,
      "learning_rate": 6.033702824545755e-06,
      "loss": 316.3526,
      "step": 32940
    },
    {
      "epoch": 0.6337086863286245,
      "grad_norm": 437.57200648004044,
      "learning_rate": 6.028101008761445e-06,
      "loss": 312.8964,
      "step": 32950
    },
    {
      "epoch": 0.6339010106643844,
      "grad_norm": 417.85734741965825,
      "learning_rate": 6.022500672443907e-06,
      "loss": 321.7556,
      "step": 32960
    },
    {
      "epoch": 0.6340933350001442,
      "grad_norm": 458.4178721966177,
      "learning_rate": 6.016901817679177e-06,
      "loss": 331.59,
      "step": 32970
    },
    {
      "epoch": 0.6342856593359041,
      "grad_norm": 410.8474244278934,
      "learning_rate": 6.011304446552741e-06,
      "loss": 318.5654,
      "step": 32980
    },
    {
      "epoch": 0.6344779836716639,
      "grad_norm": 434.78701503041884,
      "learning_rate": 6.005708561149523e-06,
      "loss": 324.2937,
      "step": 32990
    },
    {
      "epoch": 0.6346703080074237,
      "grad_norm": 448.34652220499385,
      "learning_rate": 6.000114163553894e-06,
      "loss": 328.6474,
      "step": 33000
    },
    {
      "epoch": 0.6348626323431835,
      "grad_norm": 434.05873206618304,
      "learning_rate": 5.994521255849684e-06,
      "loss": 331.5147,
      "step": 33010
    },
    {
      "epoch": 0.6350549566789434,
      "grad_norm": 411.50099490053105,
      "learning_rate": 5.988929840120151e-06,
      "loss": 328.2983,
      "step": 33020
    },
    {
      "epoch": 0.6352472810147032,
      "grad_norm": 445.10534357339293,
      "learning_rate": 5.983339918448008e-06,
      "loss": 324.5035,
      "step": 33030
    },
    {
      "epoch": 0.6354396053504631,
      "grad_norm": 445.902211599106,
      "learning_rate": 5.977751492915404e-06,
      "loss": 318.4406,
      "step": 33040
    },
    {
      "epoch": 0.6356319296862228,
      "grad_norm": 414.1709800694302,
      "learning_rate": 5.972164565603944e-06,
      "loss": 324.7382,
      "step": 33050
    },
    {
      "epoch": 0.6358242540219827,
      "grad_norm": 407.91460852423523,
      "learning_rate": 5.966579138594661e-06,
      "loss": 333.8346,
      "step": 33060
    },
    {
      "epoch": 0.6360165783577425,
      "grad_norm": 425.83599914913617,
      "learning_rate": 5.960995213968033e-06,
      "loss": 326.7666,
      "step": 33070
    },
    {
      "epoch": 0.6362089026935023,
      "grad_norm": 429.2653214507619,
      "learning_rate": 5.955412793803983e-06,
      "loss": 318.4811,
      "step": 33080
    },
    {
      "epoch": 0.6364012270292622,
      "grad_norm": 514.921101360999,
      "learning_rate": 5.949831880181869e-06,
      "loss": 339.2025,
      "step": 33090
    },
    {
      "epoch": 0.6365935513650219,
      "grad_norm": 439.47238772651775,
      "learning_rate": 5.944252475180487e-06,
      "loss": 336.8574,
      "step": 33100
    },
    {
      "epoch": 0.6367858757007818,
      "grad_norm": 408.60479581193187,
      "learning_rate": 5.938674580878077e-06,
      "loss": 324.96,
      "step": 33110
    },
    {
      "epoch": 0.6369782000365416,
      "grad_norm": 440.49902995174443,
      "learning_rate": 5.933098199352307e-06,
      "loss": 329.8435,
      "step": 33120
    },
    {
      "epoch": 0.6371705243723015,
      "grad_norm": 430.0941986481869,
      "learning_rate": 5.927523332680296e-06,
      "loss": 330.0749,
      "step": 33130
    },
    {
      "epoch": 0.6373628487080613,
      "grad_norm": 426.120737751968,
      "learning_rate": 5.921949982938583e-06,
      "loss": 324.2136,
      "step": 33140
    },
    {
      "epoch": 0.6375551730438211,
      "grad_norm": 446.29464797735096,
      "learning_rate": 5.916378152203146e-06,
      "loss": 330.048,
      "step": 33150
    },
    {
      "epoch": 0.6377474973795809,
      "grad_norm": 389.33807674876357,
      "learning_rate": 5.910807842549409e-06,
      "loss": 323.6343,
      "step": 33160
    },
    {
      "epoch": 0.6379398217153408,
      "grad_norm": 439.3824082828931,
      "learning_rate": 5.905239056052212e-06,
      "loss": 321.4603,
      "step": 33170
    },
    {
      "epoch": 0.6381321460511006,
      "grad_norm": 420.43187384435345,
      "learning_rate": 5.8996717947858395e-06,
      "loss": 324.7336,
      "step": 33180
    },
    {
      "epoch": 0.6383244703868604,
      "grad_norm": 413.1979360709277,
      "learning_rate": 5.894106060824005e-06,
      "loss": 324.3086,
      "step": 33190
    },
    {
      "epoch": 0.6385167947226202,
      "grad_norm": 402.21970248440675,
      "learning_rate": 5.888541856239854e-06,
      "loss": 318.3952,
      "step": 33200
    },
    {
      "epoch": 0.63870911905838,
      "grad_norm": 416.30673265789903,
      "learning_rate": 5.882979183105959e-06,
      "loss": 323.8405,
      "step": 33210
    },
    {
      "epoch": 0.6389014433941399,
      "grad_norm": 388.146357708491,
      "learning_rate": 5.8774180434943184e-06,
      "loss": 320.0798,
      "step": 33220
    },
    {
      "epoch": 0.6390937677298997,
      "grad_norm": 441.4225023402244,
      "learning_rate": 5.871858439476374e-06,
      "loss": 323.9578,
      "step": 33230
    },
    {
      "epoch": 0.6392860920656596,
      "grad_norm": 478.43504781601206,
      "learning_rate": 5.86630037312298e-06,
      "loss": 321.9073,
      "step": 33240
    },
    {
      "epoch": 0.6394784164014193,
      "grad_norm": 483.573101210037,
      "learning_rate": 5.860743846504429e-06,
      "loss": 328.7776,
      "step": 33250
    },
    {
      "epoch": 0.6396707407371792,
      "grad_norm": 418.2961299977733,
      "learning_rate": 5.8551888616904305e-06,
      "loss": 325.4701,
      "step": 33260
    },
    {
      "epoch": 0.639863065072939,
      "grad_norm": 420.5245189112849,
      "learning_rate": 5.849635420750131e-06,
      "loss": 335.8387,
      "step": 33270
    },
    {
      "epoch": 0.6400553894086989,
      "grad_norm": 426.4050479463985,
      "learning_rate": 5.84408352575209e-06,
      "loss": 321.9759,
      "step": 33280
    },
    {
      "epoch": 0.6402477137444587,
      "grad_norm": 423.98399844741635,
      "learning_rate": 5.8385331787642956e-06,
      "loss": 315.5348,
      "step": 33290
    },
    {
      "epoch": 0.6404400380802184,
      "grad_norm": 435.67662918847105,
      "learning_rate": 5.8329843818541665e-06,
      "loss": 326.7342,
      "step": 33300
    },
    {
      "epoch": 0.6406323624159783,
      "grad_norm": 463.14679690394445,
      "learning_rate": 5.827437137088535e-06,
      "loss": 312.5115,
      "step": 33310
    },
    {
      "epoch": 0.6408246867517381,
      "grad_norm": 444.29042742403766,
      "learning_rate": 5.8218914465336585e-06,
      "loss": 321.6006,
      "step": 33320
    },
    {
      "epoch": 0.641017011087498,
      "grad_norm": 430.5522148589923,
      "learning_rate": 5.816347312255209e-06,
      "loss": 321.4361,
      "step": 33330
    },
    {
      "epoch": 0.6412093354232578,
      "grad_norm": 433.57878399989,
      "learning_rate": 5.810804736318295e-06,
      "loss": 315.8478,
      "step": 33340
    },
    {
      "epoch": 0.6414016597590176,
      "grad_norm": 410.5329670801972,
      "learning_rate": 5.805263720787426e-06,
      "loss": 313.4332,
      "step": 33350
    },
    {
      "epoch": 0.6415939840947774,
      "grad_norm": 413.8264513873521,
      "learning_rate": 5.799724267726547e-06,
      "loss": 331.2219,
      "step": 33360
    },
    {
      "epoch": 0.6417863084305373,
      "grad_norm": 434.498285955691,
      "learning_rate": 5.794186379199004e-06,
      "loss": 324.3416,
      "step": 33370
    },
    {
      "epoch": 0.6419786327662971,
      "grad_norm": 415.7685317419322,
      "learning_rate": 5.7886500572675774e-06,
      "loss": 331.7816,
      "step": 33380
    },
    {
      "epoch": 0.642170957102057,
      "grad_norm": 411.44326253029175,
      "learning_rate": 5.783115303994451e-06,
      "loss": 331.2956,
      "step": 33390
    },
    {
      "epoch": 0.6423632814378167,
      "grad_norm": 480.5991771259947,
      "learning_rate": 5.777582121441227e-06,
      "loss": 329.7617,
      "step": 33400
    },
    {
      "epoch": 0.6425556057735765,
      "grad_norm": 469.1966076991373,
      "learning_rate": 5.772050511668931e-06,
      "loss": 325.405,
      "step": 33410
    },
    {
      "epoch": 0.6427479301093364,
      "grad_norm": 430.2266558638055,
      "learning_rate": 5.766520476737993e-06,
      "loss": 329.6934,
      "step": 33420
    },
    {
      "epoch": 0.6429402544450962,
      "grad_norm": 417.9938475095403,
      "learning_rate": 5.760992018708253e-06,
      "loss": 321.3423,
      "step": 33430
    },
    {
      "epoch": 0.6431325787808561,
      "grad_norm": 445.7673609332607,
      "learning_rate": 5.755465139638983e-06,
      "loss": 308.9907,
      "step": 33440
    },
    {
      "epoch": 0.6433249031166158,
      "grad_norm": 432.2937425308705,
      "learning_rate": 5.749939841588846e-06,
      "loss": 333.224,
      "step": 33450
    },
    {
      "epoch": 0.6435172274523757,
      "grad_norm": 396.1072917640877,
      "learning_rate": 5.744416126615926e-06,
      "loss": 322.1843,
      "step": 33460
    },
    {
      "epoch": 0.6437095517881355,
      "grad_norm": 448.37514029819243,
      "learning_rate": 5.738893996777713e-06,
      "loss": 330.3824,
      "step": 33470
    },
    {
      "epoch": 0.6439018761238954,
      "grad_norm": 443.17029410102015,
      "learning_rate": 5.7333734541311144e-06,
      "loss": 319.5919,
      "step": 33480
    },
    {
      "epoch": 0.6440942004596552,
      "grad_norm": 612.602437334116,
      "learning_rate": 5.727854500732435e-06,
      "loss": 329.0085,
      "step": 33490
    },
    {
      "epoch": 0.644286524795415,
      "grad_norm": 395.1838202480761,
      "learning_rate": 5.722337138637402e-06,
      "loss": 317.8712,
      "step": 33500
    },
    {
      "epoch": 0.6444788491311748,
      "grad_norm": 408.5772804821038,
      "learning_rate": 5.716821369901131e-06,
      "loss": 322.3591,
      "step": 33510
    },
    {
      "epoch": 0.6446711734669346,
      "grad_norm": 437.4398299274796,
      "learning_rate": 5.711307196578166e-06,
      "loss": 318.0684,
      "step": 33520
    },
    {
      "epoch": 0.6448634978026945,
      "grad_norm": 400.2229432446763,
      "learning_rate": 5.70579462072244e-06,
      "loss": 319.5992,
      "step": 33530
    },
    {
      "epoch": 0.6450558221384542,
      "grad_norm": 427.599522981999,
      "learning_rate": 5.700283644387292e-06,
      "loss": 323.8386,
      "step": 33540
    },
    {
      "epoch": 0.6452481464742141,
      "grad_norm": 430.965032771086,
      "learning_rate": 5.6947742696254794e-06,
      "loss": 329.6379,
      "step": 33550
    },
    {
      "epoch": 0.6454404708099739,
      "grad_norm": 393.64273338958515,
      "learning_rate": 5.689266498489148e-06,
      "loss": 324.8614,
      "step": 33560
    },
    {
      "epoch": 0.6456327951457338,
      "grad_norm": 423.86460278823523,
      "learning_rate": 5.683760333029851e-06,
      "loss": 327.9987,
      "step": 33570
    },
    {
      "epoch": 0.6458251194814936,
      "grad_norm": 426.0800268488462,
      "learning_rate": 5.678255775298542e-06,
      "loss": 318.0062,
      "step": 33580
    },
    {
      "epoch": 0.6460174438172535,
      "grad_norm": 473.4428249994104,
      "learning_rate": 5.672752827345584e-06,
      "loss": 318.4216,
      "step": 33590
    },
    {
      "epoch": 0.6462097681530132,
      "grad_norm": 425.11221321286064,
      "learning_rate": 5.667251491220731e-06,
      "loss": 314.7529,
      "step": 33600
    },
    {
      "epoch": 0.6464020924887731,
      "grad_norm": 397.038250473068,
      "learning_rate": 5.661751768973136e-06,
      "loss": 318.8346,
      "step": 33610
    },
    {
      "epoch": 0.6465944168245329,
      "grad_norm": 470.6579011756659,
      "learning_rate": 5.656253662651362e-06,
      "loss": 328.6394,
      "step": 33620
    },
    {
      "epoch": 0.6467867411602927,
      "grad_norm": 434.3707844144886,
      "learning_rate": 5.650757174303356e-06,
      "loss": 313.9123,
      "step": 33630
    },
    {
      "epoch": 0.6469790654960526,
      "grad_norm": 454.15912114352625,
      "learning_rate": 5.645262305976476e-06,
      "loss": 325.0698,
      "step": 33640
    },
    {
      "epoch": 0.6471713898318123,
      "grad_norm": 394.343867723725,
      "learning_rate": 5.639769059717462e-06,
      "loss": 317.2573,
      "step": 33650
    },
    {
      "epoch": 0.6473637141675722,
      "grad_norm": 443.84290679732754,
      "learning_rate": 5.634277437572466e-06,
      "loss": 322.7038,
      "step": 33660
    },
    {
      "epoch": 0.647556038503332,
      "grad_norm": 426.55076482747074,
      "learning_rate": 5.6287874415870225e-06,
      "loss": 322.7959,
      "step": 33670
    },
    {
      "epoch": 0.6477483628390919,
      "grad_norm": 438.08260777866343,
      "learning_rate": 5.623299073806063e-06,
      "loss": 317.4958,
      "step": 33680
    },
    {
      "epoch": 0.6479406871748516,
      "grad_norm": 402.14329703250354,
      "learning_rate": 5.617812336273912e-06,
      "loss": 323.1414,
      "step": 33690
    },
    {
      "epoch": 0.6481330115106115,
      "grad_norm": 453.62004875708766,
      "learning_rate": 5.612327231034296e-06,
      "loss": 317.6323,
      "step": 33700
    },
    {
      "epoch": 0.6483253358463713,
      "grad_norm": 411.53354932360133,
      "learning_rate": 5.606843760130321e-06,
      "loss": 318.9939,
      "step": 33710
    },
    {
      "epoch": 0.6485176601821312,
      "grad_norm": 459.62900713929673,
      "learning_rate": 5.601361925604485e-06,
      "loss": 322.6615,
      "step": 33720
    },
    {
      "epoch": 0.648709984517891,
      "grad_norm": 429.6514691001418,
      "learning_rate": 5.595881729498691e-06,
      "loss": 316.6743,
      "step": 33730
    },
    {
      "epoch": 0.6489023088536507,
      "grad_norm": 392.2215857929651,
      "learning_rate": 5.590403173854215e-06,
      "loss": 326.778,
      "step": 33740
    },
    {
      "epoch": 0.6490946331894106,
      "grad_norm": 391.2446133210011,
      "learning_rate": 5.584926260711732e-06,
      "loss": 324.8535,
      "step": 33750
    },
    {
      "epoch": 0.6492869575251704,
      "grad_norm": 413.7997329191996,
      "learning_rate": 5.579450992111294e-06,
      "loss": 325.6451,
      "step": 33760
    },
    {
      "epoch": 0.6494792818609303,
      "grad_norm": 388.137565885144,
      "learning_rate": 5.573977370092358e-06,
      "loss": 314.3849,
      "step": 33770
    },
    {
      "epoch": 0.6496716061966901,
      "grad_norm": 442.3418880343542,
      "learning_rate": 5.568505396693749e-06,
      "loss": 325.2625,
      "step": 33780
    },
    {
      "epoch": 0.64986393053245,
      "grad_norm": 444.99577517429634,
      "learning_rate": 5.563035073953691e-06,
      "loss": 331.6391,
      "step": 33790
    },
    {
      "epoch": 0.6500562548682097,
      "grad_norm": 406.7391152014691,
      "learning_rate": 5.557566403909794e-06,
      "loss": 315.0927,
      "step": 33800
    },
    {
      "epoch": 0.6502485792039696,
      "grad_norm": 416.18513318889,
      "learning_rate": 5.552099388599042e-06,
      "loss": 327.7045,
      "step": 33810
    },
    {
      "epoch": 0.6504409035397294,
      "grad_norm": 435.6835734121358,
      "learning_rate": 5.5466340300578095e-06,
      "loss": 316.6872,
      "step": 33820
    },
    {
      "epoch": 0.6506332278754893,
      "grad_norm": 425.301982879258,
      "learning_rate": 5.541170330321845e-06,
      "loss": 321.9507,
      "step": 33830
    },
    {
      "epoch": 0.650825552211249,
      "grad_norm": 388.1779247762942,
      "learning_rate": 5.535708291426297e-06,
      "loss": 322.6335,
      "step": 33840
    },
    {
      "epoch": 0.6510178765470088,
      "grad_norm": 417.2170676311644,
      "learning_rate": 5.53024791540568e-06,
      "loss": 318.0769,
      "step": 33850
    },
    {
      "epoch": 0.6512102008827687,
      "grad_norm": 411.30815786731057,
      "learning_rate": 5.5247892042938944e-06,
      "loss": 314.6812,
      "step": 33860
    },
    {
      "epoch": 0.6514025252185285,
      "grad_norm": 426.40200860632825,
      "learning_rate": 5.519332160124215e-06,
      "loss": 321.7491,
      "step": 33870
    },
    {
      "epoch": 0.6515948495542884,
      "grad_norm": 461.55399752605183,
      "learning_rate": 5.513876784929311e-06,
      "loss": 314.5473,
      "step": 33880
    },
    {
      "epoch": 0.6517871738900481,
      "grad_norm": 412.7237247104228,
      "learning_rate": 5.5084230807412135e-06,
      "loss": 313.2229,
      "step": 33890
    },
    {
      "epoch": 0.651979498225808,
      "grad_norm": 471.67806684335244,
      "learning_rate": 5.502971049591332e-06,
      "loss": 328.9412,
      "step": 33900
    },
    {
      "epoch": 0.6521718225615678,
      "grad_norm": 407.6270918724697,
      "learning_rate": 5.497520693510469e-06,
      "loss": 321.9584,
      "step": 33910
    },
    {
      "epoch": 0.6523641468973277,
      "grad_norm": 391.1070096585722,
      "learning_rate": 5.492072014528783e-06,
      "loss": 326.1006,
      "step": 33920
    },
    {
      "epoch": 0.6525564712330875,
      "grad_norm": 422.2585049304015,
      "learning_rate": 5.4866250146758235e-06,
      "loss": 323.3586,
      "step": 33930
    },
    {
      "epoch": 0.6527487955688474,
      "grad_norm": 420.1783076752337,
      "learning_rate": 5.481179695980503e-06,
      "loss": 321.3275,
      "step": 33940
    },
    {
      "epoch": 0.6529411199046071,
      "grad_norm": 391.98385017473686,
      "learning_rate": 5.475736060471117e-06,
      "loss": 327.6176,
      "step": 33950
    },
    {
      "epoch": 0.6531334442403669,
      "grad_norm": 408.94302995629175,
      "learning_rate": 5.470294110175329e-06,
      "loss": 321.7637,
      "step": 33960
    },
    {
      "epoch": 0.6533257685761268,
      "grad_norm": 406.17111579506235,
      "learning_rate": 5.464853847120169e-06,
      "loss": 315.7161,
      "step": 33970
    },
    {
      "epoch": 0.6535180929118866,
      "grad_norm": 444.5009568534264,
      "learning_rate": 5.459415273332056e-06,
      "loss": 325.0126,
      "step": 33980
    },
    {
      "epoch": 0.6537104172476464,
      "grad_norm": 410.50299705956877,
      "learning_rate": 5.453978390836763e-06,
      "loss": 313.3642,
      "step": 33990
    },
    {
      "epoch": 0.6539027415834062,
      "grad_norm": 422.1007201084567,
      "learning_rate": 5.44854320165944e-06,
      "loss": 326.4321,
      "step": 34000
    },
    {
      "epoch": 0.6540950659191661,
      "grad_norm": 434.0990973610923,
      "learning_rate": 5.443109707824599e-06,
      "loss": 330.3254,
      "step": 34010
    },
    {
      "epoch": 0.6542873902549259,
      "grad_norm": 432.81951475895295,
      "learning_rate": 5.437677911356137e-06,
      "loss": 321.7455,
      "step": 34020
    },
    {
      "epoch": 0.6544797145906858,
      "grad_norm": 432.9774423501474,
      "learning_rate": 5.432247814277305e-06,
      "loss": 314.3346,
      "step": 34030
    },
    {
      "epoch": 0.6546720389264455,
      "grad_norm": 464.89339637834706,
      "learning_rate": 5.426819418610718e-06,
      "loss": 328.4793,
      "step": 34040
    },
    {
      "epoch": 0.6548643632622054,
      "grad_norm": 422.37295282356837,
      "learning_rate": 5.4213927263783725e-06,
      "loss": 313.1352,
      "step": 34050
    },
    {
      "epoch": 0.6550566875979652,
      "grad_norm": 406.7230782663906,
      "learning_rate": 5.415967739601616e-06,
      "loss": 312.4417,
      "step": 34060
    },
    {
      "epoch": 0.655249011933725,
      "grad_norm": 430.46783667465684,
      "learning_rate": 5.41054446030117e-06,
      "loss": 321.0382,
      "step": 34070
    },
    {
      "epoch": 0.6554413362694849,
      "grad_norm": 466.0152177198623,
      "learning_rate": 5.405122890497114e-06,
      "loss": 331.1604,
      "step": 34080
    },
    {
      "epoch": 0.6556336606052446,
      "grad_norm": 404.7552280434815,
      "learning_rate": 5.399703032208896e-06,
      "loss": 321.9568,
      "step": 34090
    },
    {
      "epoch": 0.6558259849410045,
      "grad_norm": 384.31666367720516,
      "learning_rate": 5.3942848874553235e-06,
      "loss": 319.4767,
      "step": 34100
    },
    {
      "epoch": 0.6560183092767643,
      "grad_norm": 403.78580996538915,
      "learning_rate": 5.388868458254565e-06,
      "loss": 310.9621,
      "step": 34110
    },
    {
      "epoch": 0.6562106336125242,
      "grad_norm": 412.74071685789914,
      "learning_rate": 5.3834537466241455e-06,
      "loss": 316.402,
      "step": 34120
    },
    {
      "epoch": 0.656402957948284,
      "grad_norm": 394.62577335859555,
      "learning_rate": 5.378040754580964e-06,
      "loss": 312.2696,
      "step": 34130
    },
    {
      "epoch": 0.6565952822840438,
      "grad_norm": 484.3278084268684,
      "learning_rate": 5.37262948414127e-06,
      "loss": 323.8389,
      "step": 34140
    },
    {
      "epoch": 0.6567876066198036,
      "grad_norm": 454.9515756714042,
      "learning_rate": 5.367219937320663e-06,
      "loss": 315.6518,
      "step": 34150
    },
    {
      "epoch": 0.6569799309555635,
      "grad_norm": 429.8524095947134,
      "learning_rate": 5.361812116134122e-06,
      "loss": 319.3386,
      "step": 34160
    },
    {
      "epoch": 0.6571722552913233,
      "grad_norm": 462.0170696979145,
      "learning_rate": 5.356406022595963e-06,
      "loss": 324.6877,
      "step": 34170
    },
    {
      "epoch": 0.6573645796270832,
      "grad_norm": 434.2684872638781,
      "learning_rate": 5.351001658719872e-06,
      "loss": 323.9905,
      "step": 34180
    },
    {
      "epoch": 0.6575569039628429,
      "grad_norm": 384.55361116744785,
      "learning_rate": 5.345599026518877e-06,
      "loss": 322.8377,
      "step": 34190
    },
    {
      "epoch": 0.6577492282986027,
      "grad_norm": 408.24448614574965,
      "learning_rate": 5.3401981280053745e-06,
      "loss": 329.8235,
      "step": 34200
    },
    {
      "epoch": 0.6579415526343626,
      "grad_norm": 412.46697449906543,
      "learning_rate": 5.334798965191115e-06,
      "loss": 319.9216,
      "step": 34210
    },
    {
      "epoch": 0.6581338769701224,
      "grad_norm": 453.2313210783639,
      "learning_rate": 5.329401540087188e-06,
      "loss": 316.3315,
      "step": 34220
    },
    {
      "epoch": 0.6583262013058823,
      "grad_norm": 410.4965679772034,
      "learning_rate": 5.3240058547040525e-06,
      "loss": 311.8159,
      "step": 34230
    },
    {
      "epoch": 0.658518525641642,
      "grad_norm": 418.2916279810732,
      "learning_rate": 5.318611911051512e-06,
      "loss": 317.0869,
      "step": 34240
    },
    {
      "epoch": 0.6587108499774019,
      "grad_norm": 398.0746003532323,
      "learning_rate": 5.313219711138717e-06,
      "loss": 321.4545,
      "step": 34250
    },
    {
      "epoch": 0.6589031743131617,
      "grad_norm": 435.32959353558465,
      "learning_rate": 5.307829256974173e-06,
      "loss": 314.1013,
      "step": 34260
    },
    {
      "epoch": 0.6590954986489216,
      "grad_norm": 412.71587127992666,
      "learning_rate": 5.302440550565739e-06,
      "loss": 307.7437,
      "step": 34270
    },
    {
      "epoch": 0.6592878229846814,
      "grad_norm": 417.5824587680408,
      "learning_rate": 5.297053593920618e-06,
      "loss": 310.0909,
      "step": 34280
    },
    {
      "epoch": 0.6594801473204412,
      "grad_norm": 427.7726785260718,
      "learning_rate": 5.291668389045362e-06,
      "loss": 318.7255,
      "step": 34290
    },
    {
      "epoch": 0.659672471656201,
      "grad_norm": 404.77648442538276,
      "learning_rate": 5.286284937945866e-06,
      "loss": 322.9675,
      "step": 34300
    },
    {
      "epoch": 0.6598647959919608,
      "grad_norm": 419.15125676750876,
      "learning_rate": 5.280903242627384e-06,
      "loss": 324.9051,
      "step": 34310
    },
    {
      "epoch": 0.6600571203277207,
      "grad_norm": 700.4601942372726,
      "learning_rate": 5.2755233050945076e-06,
      "loss": 323.9069,
      "step": 34320
    },
    {
      "epoch": 0.6602494446634805,
      "grad_norm": 440.5554254683184,
      "learning_rate": 5.270145127351168e-06,
      "loss": 329.7043,
      "step": 34330
    },
    {
      "epoch": 0.6604417689992403,
      "grad_norm": 452.4434197686861,
      "learning_rate": 5.264768711400656e-06,
      "loss": 319.6636,
      "step": 34340
    },
    {
      "epoch": 0.6606340933350001,
      "grad_norm": 439.113167807011,
      "learning_rate": 5.25939405924559e-06,
      "loss": 321.309,
      "step": 34350
    },
    {
      "epoch": 0.66082641767076,
      "grad_norm": 419.5085140287832,
      "learning_rate": 5.254021172887947e-06,
      "loss": 328.4319,
      "step": 34360
    },
    {
      "epoch": 0.6610187420065198,
      "grad_norm": 420.4206765441334,
      "learning_rate": 5.248650054329032e-06,
      "loss": 327.2183,
      "step": 34370
    },
    {
      "epoch": 0.6612110663422797,
      "grad_norm": 406.7490039045254,
      "learning_rate": 5.2432807055695035e-06,
      "loss": 328.0459,
      "step": 34380
    },
    {
      "epoch": 0.6614033906780394,
      "grad_norm": 424.36551611046895,
      "learning_rate": 5.237913128609352e-06,
      "loss": 323.0216,
      "step": 34390
    },
    {
      "epoch": 0.6615957150137993,
      "grad_norm": 437.52935730821406,
      "learning_rate": 5.232547325447908e-06,
      "loss": 325.0063,
      "step": 34400
    },
    {
      "epoch": 0.6617880393495591,
      "grad_norm": 437.01287618092044,
      "learning_rate": 5.227183298083854e-06,
      "loss": 318.581,
      "step": 34410
    },
    {
      "epoch": 0.6619803636853189,
      "grad_norm": 448.95864817232075,
      "learning_rate": 5.221821048515193e-06,
      "loss": 321.3659,
      "step": 34420
    },
    {
      "epoch": 0.6621726880210788,
      "grad_norm": 423.5284873839197,
      "learning_rate": 5.216460578739278e-06,
      "loss": 315.2833,
      "step": 34430
    },
    {
      "epoch": 0.6623650123568385,
      "grad_norm": 409.9628990423758,
      "learning_rate": 5.211101890752792e-06,
      "loss": 314.3752,
      "step": 34440
    },
    {
      "epoch": 0.6625573366925984,
      "grad_norm": 440.23287487525386,
      "learning_rate": 5.205744986551763e-06,
      "loss": 317.4021,
      "step": 34450
    },
    {
      "epoch": 0.6627496610283582,
      "grad_norm": 419.76651942099187,
      "learning_rate": 5.200389868131547e-06,
      "loss": 330.2907,
      "step": 34460
    },
    {
      "epoch": 0.6629419853641181,
      "grad_norm": 422.6388629850351,
      "learning_rate": 5.195036537486833e-06,
      "loss": 328.2027,
      "step": 34470
    },
    {
      "epoch": 0.6631343096998779,
      "grad_norm": 401.3370957544165,
      "learning_rate": 5.189684996611657e-06,
      "loss": 328.9475,
      "step": 34480
    },
    {
      "epoch": 0.6633266340356377,
      "grad_norm": 389.84335812195116,
      "learning_rate": 5.18433524749937e-06,
      "loss": 313.9279,
      "step": 34490
    },
    {
      "epoch": 0.6635189583713975,
      "grad_norm": 447.09148903429923,
      "learning_rate": 5.178987292142674e-06,
      "loss": 334.1997,
      "step": 34500
    },
    {
      "epoch": 0.6637112827071574,
      "grad_norm": 408.0765166723503,
      "learning_rate": 5.173641132533586e-06,
      "loss": 320.0878,
      "step": 34510
    },
    {
      "epoch": 0.6639036070429172,
      "grad_norm": 451.49801340447186,
      "learning_rate": 5.16829677066347e-06,
      "loss": 329.4619,
      "step": 34520
    },
    {
      "epoch": 0.664095931378677,
      "grad_norm": 463.31000007246547,
      "learning_rate": 5.16295420852301e-06,
      "loss": 317.359,
      "step": 34530
    },
    {
      "epoch": 0.6642882557144368,
      "grad_norm": 425.00419191410685,
      "learning_rate": 5.157613448102219e-06,
      "loss": 322.7488,
      "step": 34540
    },
    {
      "epoch": 0.6644805800501966,
      "grad_norm": 403.49033485417215,
      "learning_rate": 5.152274491390441e-06,
      "loss": 319.9063,
      "step": 34550
    },
    {
      "epoch": 0.6646729043859565,
      "grad_norm": 458.3740647040998,
      "learning_rate": 5.1469373403763555e-06,
      "loss": 326.2286,
      "step": 34560
    },
    {
      "epoch": 0.6648652287217163,
      "grad_norm": 406.11874601058656,
      "learning_rate": 5.14160199704796e-06,
      "loss": 320.0305,
      "step": 34570
    },
    {
      "epoch": 0.6650575530574762,
      "grad_norm": 448.023791067687,
      "learning_rate": 5.136268463392578e-06,
      "loss": 316.2592,
      "step": 34580
    },
    {
      "epoch": 0.6652498773932359,
      "grad_norm": 461.41336024953245,
      "learning_rate": 5.13093674139687e-06,
      "loss": 319.7057,
      "step": 34590
    },
    {
      "epoch": 0.6654422017289958,
      "grad_norm": 409.72632238289253,
      "learning_rate": 5.12560683304681e-06,
      "loss": 313.9694,
      "step": 34600
    },
    {
      "epoch": 0.6656345260647556,
      "grad_norm": 407.0339279205974,
      "learning_rate": 5.120278740327702e-06,
      "loss": 316.8297,
      "step": 34610
    },
    {
      "epoch": 0.6658268504005155,
      "grad_norm": 443.27103645300116,
      "learning_rate": 5.114952465224168e-06,
      "loss": 319.7591,
      "step": 34620
    },
    {
      "epoch": 0.6660191747362753,
      "grad_norm": 434.1902683370954,
      "learning_rate": 5.109628009720162e-06,
      "loss": 330.3795,
      "step": 34630
    },
    {
      "epoch": 0.666211499072035,
      "grad_norm": 457.3037224180103,
      "learning_rate": 5.104305375798958e-06,
      "loss": 323.4473,
      "step": 34640
    },
    {
      "epoch": 0.6664038234077949,
      "grad_norm": 434.7740603021931,
      "learning_rate": 5.098984565443148e-06,
      "loss": 327.71,
      "step": 34650
    },
    {
      "epoch": 0.6665961477435547,
      "grad_norm": 445.9017906778416,
      "learning_rate": 5.093665580634639e-06,
      "loss": 313.9246,
      "step": 34660
    },
    {
      "epoch": 0.6667884720793146,
      "grad_norm": 442.4596365556121,
      "learning_rate": 5.088348423354674e-06,
      "loss": 314.1189,
      "step": 34670
    },
    {
      "epoch": 0.6669807964150744,
      "grad_norm": 426.5848077492119,
      "learning_rate": 5.0830330955838045e-06,
      "loss": 324.6211,
      "step": 34680
    },
    {
      "epoch": 0.6671731207508342,
      "grad_norm": 414.6957175063593,
      "learning_rate": 5.077719599301895e-06,
      "loss": 320.5044,
      "step": 34690
    },
    {
      "epoch": 0.667365445086594,
      "grad_norm": 450.4426821588187,
      "learning_rate": 5.072407936488145e-06,
      "loss": 339.2428,
      "step": 34700
    },
    {
      "epoch": 0.6675577694223539,
      "grad_norm": 429.1245757343828,
      "learning_rate": 5.067098109121058e-06,
      "loss": 319.8875,
      "step": 34710
    },
    {
      "epoch": 0.6677500937581137,
      "grad_norm": 434.6340471275704,
      "learning_rate": 5.0617901191784536e-06,
      "loss": 321.0468,
      "step": 34720
    },
    {
      "epoch": 0.6679424180938736,
      "grad_norm": 414.75166287316534,
      "learning_rate": 5.05648396863747e-06,
      "loss": 318.6448,
      "step": 34730
    },
    {
      "epoch": 0.6681347424296333,
      "grad_norm": 441.0143680065906,
      "learning_rate": 5.051179659474568e-06,
      "loss": 322.9612,
      "step": 34740
    },
    {
      "epoch": 0.6683270667653931,
      "grad_norm": 422.7741616526519,
      "learning_rate": 5.045877193665508e-06,
      "loss": 321.4521,
      "step": 34750
    },
    {
      "epoch": 0.668519391101153,
      "grad_norm": 433.93211677484453,
      "learning_rate": 5.040576573185372e-06,
      "loss": 320.7336,
      "step": 34760
    },
    {
      "epoch": 0.6687117154369128,
      "grad_norm": 447.1348325928919,
      "learning_rate": 5.0352778000085565e-06,
      "loss": 319.1764,
      "step": 34770
    },
    {
      "epoch": 0.6689040397726727,
      "grad_norm": 420.30364596003307,
      "learning_rate": 5.029980876108762e-06,
      "loss": 325.4064,
      "step": 34780
    },
    {
      "epoch": 0.6690963641084324,
      "grad_norm": 379.8542432665582,
      "learning_rate": 5.024685803459013e-06,
      "loss": 314.0862,
      "step": 34790
    },
    {
      "epoch": 0.6692886884441923,
      "grad_norm": 399.0285316648089,
      "learning_rate": 5.019392584031628e-06,
      "loss": 310.9487,
      "step": 34800
    },
    {
      "epoch": 0.6694810127799521,
      "grad_norm": 447.18871457515337,
      "learning_rate": 5.0141012197982534e-06,
      "loss": 329.5792,
      "step": 34810
    },
    {
      "epoch": 0.669673337115712,
      "grad_norm": 429.254181904457,
      "learning_rate": 5.0088117127298285e-06,
      "loss": 317.2112,
      "step": 34820
    },
    {
      "epoch": 0.6698656614514717,
      "grad_norm": 412.40775690696927,
      "learning_rate": 5.003524064796608e-06,
      "loss": 321.0009,
      "step": 34830
    },
    {
      "epoch": 0.6700579857872316,
      "grad_norm": 428.9910498813792,
      "learning_rate": 4.998238277968149e-06,
      "loss": 322.7637,
      "step": 34840
    },
    {
      "epoch": 0.6702503101229914,
      "grad_norm": 443.3119990918627,
      "learning_rate": 4.992954354213331e-06,
      "loss": 314.5496,
      "step": 34850
    },
    {
      "epoch": 0.6704426344587512,
      "grad_norm": 497.6257895821623,
      "learning_rate": 4.98767229550032e-06,
      "loss": 327.0348,
      "step": 34860
    },
    {
      "epoch": 0.6706349587945111,
      "grad_norm": 436.69055454832454,
      "learning_rate": 4.982392103796595e-06,
      "loss": 315.5988,
      "step": 34870
    },
    {
      "epoch": 0.6708272831302708,
      "grad_norm": 427.79002311571855,
      "learning_rate": 4.977113781068945e-06,
      "loss": 326.8617,
      "step": 34880
    },
    {
      "epoch": 0.6710196074660307,
      "grad_norm": 405.54540379520216,
      "learning_rate": 4.971837329283458e-06,
      "loss": 323.7293,
      "step": 34890
    },
    {
      "epoch": 0.6712119318017905,
      "grad_norm": 415.8029035353139,
      "learning_rate": 4.966562750405517e-06,
      "loss": 330.4927,
      "step": 34900
    },
    {
      "epoch": 0.6714042561375504,
      "grad_norm": 461.77870002641004,
      "learning_rate": 4.9612900463998274e-06,
      "loss": 318.3586,
      "step": 34910
    },
    {
      "epoch": 0.6715965804733102,
      "grad_norm": 447.17270875950743,
      "learning_rate": 4.9560192192303735e-06,
      "loss": 330.4827,
      "step": 34920
    },
    {
      "epoch": 0.67178890480907,
      "grad_norm": 436.8488392769194,
      "learning_rate": 4.95075027086046e-06,
      "loss": 316.7468,
      "step": 34930
    },
    {
      "epoch": 0.6719812291448298,
      "grad_norm": 434.0188965839647,
      "learning_rate": 4.9454832032526755e-06,
      "loss": 328.8788,
      "step": 34940
    },
    {
      "epoch": 0.6721735534805897,
      "grad_norm": 415.7735931947023,
      "learning_rate": 4.940218018368924e-06,
      "loss": 323.4168,
      "step": 34950
    },
    {
      "epoch": 0.6723658778163495,
      "grad_norm": 413.5492003926558,
      "learning_rate": 4.934954718170396e-06,
      "loss": 318.2195,
      "step": 34960
    },
    {
      "epoch": 0.6725582021521093,
      "grad_norm": 418.6289685293497,
      "learning_rate": 4.9296933046175834e-06,
      "loss": 312.7955,
      "step": 34970
    },
    {
      "epoch": 0.6727505264878691,
      "grad_norm": 435.4569139195881,
      "learning_rate": 4.924433779670271e-06,
      "loss": 311.7696,
      "step": 34980
    },
    {
      "epoch": 0.6729428508236289,
      "grad_norm": 437.9520875898177,
      "learning_rate": 4.9191761452875554e-06,
      "loss": 328.6258,
      "step": 34990
    },
    {
      "epoch": 0.6731351751593888,
      "grad_norm": 433.1435276509337,
      "learning_rate": 4.913920403427812e-06,
      "loss": 317.9559,
      "step": 35000
    },
    {
      "epoch": 0.6733274994951486,
      "grad_norm": 445.6820626253493,
      "learning_rate": 4.908666556048719e-06,
      "loss": 314.5841,
      "step": 35010
    },
    {
      "epoch": 0.6735198238309085,
      "grad_norm": 384.0392196330658,
      "learning_rate": 4.903414605107244e-06,
      "loss": 317.4141,
      "step": 35020
    },
    {
      "epoch": 0.6737121481666682,
      "grad_norm": 416.9583123576296,
      "learning_rate": 4.89816455255966e-06,
      "loss": 317.8913,
      "step": 35030
    },
    {
      "epoch": 0.6739044725024281,
      "grad_norm": 430.47849034999916,
      "learning_rate": 4.892916400361516e-06,
      "loss": 321.4291,
      "step": 35040
    },
    {
      "epoch": 0.6740967968381879,
      "grad_norm": 422.74257258075477,
      "learning_rate": 4.887670150467671e-06,
      "loss": 320.9816,
      "step": 35050
    },
    {
      "epoch": 0.6742891211739478,
      "grad_norm": 415.55334900158067,
      "learning_rate": 4.882425804832258e-06,
      "loss": 320.2086,
      "step": 35060
    },
    {
      "epoch": 0.6744814455097076,
      "grad_norm": 1750.8438591972854,
      "learning_rate": 4.8771833654087165e-06,
      "loss": 318.8852,
      "step": 35070
    },
    {
      "epoch": 0.6746737698454673,
      "grad_norm": 452.511906008904,
      "learning_rate": 4.8719428341497665e-06,
      "loss": 320.3105,
      "step": 35080
    },
    {
      "epoch": 0.6748660941812272,
      "grad_norm": 414.2766381834859,
      "learning_rate": 4.866704213007413e-06,
      "loss": 324.9075,
      "step": 35090
    },
    {
      "epoch": 0.675058418516987,
      "grad_norm": 440.11892878794276,
      "learning_rate": 4.861467503932965e-06,
      "loss": 323.142,
      "step": 35100
    },
    {
      "epoch": 0.6752507428527469,
      "grad_norm": 386.32848106126545,
      "learning_rate": 4.856232708877007e-06,
      "loss": 312.7934,
      "step": 35110
    },
    {
      "epoch": 0.6754430671885067,
      "grad_norm": 463.09324815980216,
      "learning_rate": 4.8509998297894075e-06,
      "loss": 325.0859,
      "step": 35120
    },
    {
      "epoch": 0.6756353915242665,
      "grad_norm": 454.8062208909201,
      "learning_rate": 4.845768868619338e-06,
      "loss": 318.9287,
      "step": 35130
    },
    {
      "epoch": 0.6758277158600263,
      "grad_norm": 453.26999025394167,
      "learning_rate": 4.8405398273152405e-06,
      "loss": 327.3516,
      "step": 35140
    },
    {
      "epoch": 0.6760200401957862,
      "grad_norm": 434.3070018409594,
      "learning_rate": 4.8353127078248455e-06,
      "loss": 334.8855,
      "step": 35150
    },
    {
      "epoch": 0.676212364531546,
      "grad_norm": 427.8667220911036,
      "learning_rate": 4.830087512095164e-06,
      "loss": 315.9448,
      "step": 35160
    },
    {
      "epoch": 0.6764046888673059,
      "grad_norm": 424.11467993643754,
      "learning_rate": 4.824864242072506e-06,
      "loss": 317.7206,
      "step": 35170
    },
    {
      "epoch": 0.6765970132030656,
      "grad_norm": 468.14699644869796,
      "learning_rate": 4.819642899702447e-06,
      "loss": 320.1482,
      "step": 35180
    },
    {
      "epoch": 0.6767893375388255,
      "grad_norm": 416.84003049414696,
      "learning_rate": 4.814423486929846e-06,
      "loss": 318.9567,
      "step": 35190
    },
    {
      "epoch": 0.6769816618745853,
      "grad_norm": 450.1386803380564,
      "learning_rate": 4.809206005698856e-06,
      "loss": 315.3346,
      "step": 35200
    },
    {
      "epoch": 0.6771739862103451,
      "grad_norm": 408.23576990355565,
      "learning_rate": 4.803990457952903e-06,
      "loss": 332.8654,
      "step": 35210
    },
    {
      "epoch": 0.677366310546105,
      "grad_norm": 426.6508150301355,
      "learning_rate": 4.798776845634688e-06,
      "loss": 315.4395,
      "step": 35220
    },
    {
      "epoch": 0.6775586348818647,
      "grad_norm": 427.3549892373833,
      "learning_rate": 4.793565170686196e-06,
      "loss": 324.9042,
      "step": 35230
    },
    {
      "epoch": 0.6777509592176246,
      "grad_norm": 471.2052981869491,
      "learning_rate": 4.7883554350486925e-06,
      "loss": 327.6963,
      "step": 35240
    },
    {
      "epoch": 0.6779432835533844,
      "grad_norm": 408.1241504944268,
      "learning_rate": 4.783147640662716e-06,
      "loss": 313.6297,
      "step": 35250
    },
    {
      "epoch": 0.6781356078891443,
      "grad_norm": 484.14036270031346,
      "learning_rate": 4.777941789468086e-06,
      "loss": 318.3194,
      "step": 35260
    },
    {
      "epoch": 0.6783279322249041,
      "grad_norm": 436.8865340266188,
      "learning_rate": 4.772737883403889e-06,
      "loss": 329.4075,
      "step": 35270
    },
    {
      "epoch": 0.678520256560664,
      "grad_norm": 452.4959936775624,
      "learning_rate": 4.767535924408504e-06,
      "loss": 323.2735,
      "step": 35280
    },
    {
      "epoch": 0.6787125808964237,
      "grad_norm": 413.2911630874319,
      "learning_rate": 4.76233591441957e-06,
      "loss": 322.1622,
      "step": 35290
    },
    {
      "epoch": 0.6789049052321836,
      "grad_norm": 455.08512498410465,
      "learning_rate": 4.757137855374001e-06,
      "loss": 315.3964,
      "step": 35300
    },
    {
      "epoch": 0.6790972295679434,
      "grad_norm": 438.72567379585894,
      "learning_rate": 4.751941749207996e-06,
      "loss": 324.6634,
      "step": 35310
    },
    {
      "epoch": 0.6792895539037032,
      "grad_norm": 389.61972316237666,
      "learning_rate": 4.746747597857014e-06,
      "loss": 318.8967,
      "step": 35320
    },
    {
      "epoch": 0.679481878239463,
      "grad_norm": 439.1397239993643,
      "learning_rate": 4.74155540325579e-06,
      "loss": 321.9838,
      "step": 35330
    },
    {
      "epoch": 0.6796742025752228,
      "grad_norm": 459.8977558007983,
      "learning_rate": 4.736365167338333e-06,
      "loss": 331.2183,
      "step": 35340
    },
    {
      "epoch": 0.6798665269109827,
      "grad_norm": 426.0949228352632,
      "learning_rate": 4.731176892037918e-06,
      "loss": 315.3601,
      "step": 35350
    },
    {
      "epoch": 0.6800588512467425,
      "grad_norm": 390.0340840659755,
      "learning_rate": 4.725990579287097e-06,
      "loss": 313.4447,
      "step": 35360
    },
    {
      "epoch": 0.6802511755825024,
      "grad_norm": 456.2995428175703,
      "learning_rate": 4.720806231017676e-06,
      "loss": 322.5771,
      "step": 35370
    },
    {
      "epoch": 0.6804434999182621,
      "grad_norm": 441.44460948213367,
      "learning_rate": 4.7156238491607506e-06,
      "loss": 321.9458,
      "step": 35380
    },
    {
      "epoch": 0.680635824254022,
      "grad_norm": 430.9441742966411,
      "learning_rate": 4.710443435646666e-06,
      "loss": 316.4179,
      "step": 35390
    },
    {
      "epoch": 0.6808281485897818,
      "grad_norm": 435.88259915838825,
      "learning_rate": 4.705264992405043e-06,
      "loss": 316.0891,
      "step": 35400
    },
    {
      "epoch": 0.6810204729255417,
      "grad_norm": 416.05257552492986,
      "learning_rate": 4.700088521364761e-06,
      "loss": 316.9074,
      "step": 35410
    },
    {
      "epoch": 0.6812127972613015,
      "grad_norm": 428.28329795980795,
      "learning_rate": 4.694914024453977e-06,
      "loss": 322.7706,
      "step": 35420
    },
    {
      "epoch": 0.6814051215970612,
      "grad_norm": 459.55013655554416,
      "learning_rate": 4.689741503600103e-06,
      "loss": 321.8156,
      "step": 35430
    },
    {
      "epoch": 0.6815974459328211,
      "grad_norm": 479.46912568338354,
      "learning_rate": 4.684570960729818e-06,
      "loss": 320.3851,
      "step": 35440
    },
    {
      "epoch": 0.6817897702685809,
      "grad_norm": 422.8128373242532,
      "learning_rate": 4.679402397769057e-06,
      "loss": 314.1304,
      "step": 35450
    },
    {
      "epoch": 0.6819820946043408,
      "grad_norm": 441.5462664805623,
      "learning_rate": 4.674235816643035e-06,
      "loss": 332.4933,
      "step": 35460
    },
    {
      "epoch": 0.6821744189401006,
      "grad_norm": 384.86895803155187,
      "learning_rate": 4.66907121927621e-06,
      "loss": 317.1574,
      "step": 35470
    },
    {
      "epoch": 0.6823667432758604,
      "grad_norm": 410.05636297872826,
      "learning_rate": 4.663908607592316e-06,
      "loss": 314.3527,
      "step": 35480
    },
    {
      "epoch": 0.6825590676116202,
      "grad_norm": 425.6598640479102,
      "learning_rate": 4.658747983514334e-06,
      "loss": 308.4277,
      "step": 35490
    },
    {
      "epoch": 0.6827513919473801,
      "grad_norm": 452.20897191454713,
      "learning_rate": 4.653589348964517e-06,
      "loss": 318.5503,
      "step": 35500
    },
    {
      "epoch": 0.6829437162831399,
      "grad_norm": 427.8196403112476,
      "learning_rate": 4.648432705864369e-06,
      "loss": 342.3499,
      "step": 35510
    },
    {
      "epoch": 0.6831360406188998,
      "grad_norm": 442.14298183885467,
      "learning_rate": 4.64327805613465e-06,
      "loss": 328.952,
      "step": 35520
    },
    {
      "epoch": 0.6833283649546595,
      "grad_norm": 447.62615924007883,
      "learning_rate": 4.638125401695391e-06,
      "loss": 316.3195,
      "step": 35530
    },
    {
      "epoch": 0.6835206892904193,
      "grad_norm": 442.65206444983187,
      "learning_rate": 4.632974744465865e-06,
      "loss": 325.5538,
      "step": 35540
    },
    {
      "epoch": 0.6837130136261792,
      "grad_norm": 423.9748888459649,
      "learning_rate": 4.627826086364603e-06,
      "loss": 328.1231,
      "step": 35550
    },
    {
      "epoch": 0.683905337961939,
      "grad_norm": 448.42457965304294,
      "learning_rate": 4.622679429309404e-06,
      "loss": 321.4165,
      "step": 35560
    },
    {
      "epoch": 0.6840976622976989,
      "grad_norm": 410.47197474996636,
      "learning_rate": 4.617534775217307e-06,
      "loss": 309.5952,
      "step": 35570
    },
    {
      "epoch": 0.6842899866334586,
      "grad_norm": 413.0516274306436,
      "learning_rate": 4.6123921260046135e-06,
      "loss": 315.1094,
      "step": 35580
    },
    {
      "epoch": 0.6844823109692185,
      "grad_norm": 411.0255269052447,
      "learning_rate": 4.607251483586869e-06,
      "loss": 326.479,
      "step": 35590
    },
    {
      "epoch": 0.6846746353049783,
      "grad_norm": 392.1154093571456,
      "learning_rate": 4.6021128498788855e-06,
      "loss": 311.27,
      "step": 35600
    },
    {
      "epoch": 0.6848669596407382,
      "grad_norm": 394.2181861863902,
      "learning_rate": 4.596976226794718e-06,
      "loss": 312.2333,
      "step": 35610
    },
    {
      "epoch": 0.685059283976498,
      "grad_norm": 407.61710030899235,
      "learning_rate": 4.591841616247669e-06,
      "loss": 326.9352,
      "step": 35620
    },
    {
      "epoch": 0.6852516083122578,
      "grad_norm": 437.2303815803078,
      "learning_rate": 4.586709020150297e-06,
      "loss": 317.9281,
      "step": 35630
    },
    {
      "epoch": 0.6854439326480176,
      "grad_norm": 432.15993690773496,
      "learning_rate": 4.581578440414417e-06,
      "loss": 314.4608,
      "step": 35640
    },
    {
      "epoch": 0.6856362569837774,
      "grad_norm": 417.4586990853319,
      "learning_rate": 4.576449878951079e-06,
      "loss": 323.0189,
      "step": 35650
    },
    {
      "epoch": 0.6858285813195373,
      "grad_norm": 468.54385730553463,
      "learning_rate": 4.571323337670584e-06,
      "loss": 318.4184,
      "step": 35660
    },
    {
      "epoch": 0.686020905655297,
      "grad_norm": 382.41200637957206,
      "learning_rate": 4.566198818482494e-06,
      "loss": 322.2061,
      "step": 35670
    },
    {
      "epoch": 0.6862132299910569,
      "grad_norm": 400.2019477047444,
      "learning_rate": 4.5610763232956e-06,
      "loss": 322.0258,
      "step": 35680
    },
    {
      "epoch": 0.6864055543268167,
      "grad_norm": 441.2491479190202,
      "learning_rate": 4.55595585401795e-06,
      "loss": 330.1082,
      "step": 35690
    },
    {
      "epoch": 0.6865978786625766,
      "grad_norm": 470.1692026415292,
      "learning_rate": 4.550837412556828e-06,
      "loss": 332.463,
      "step": 35700
    },
    {
      "epoch": 0.6867902029983364,
      "grad_norm": 438.80469802375,
      "learning_rate": 4.545721000818778e-06,
      "loss": 330.1236,
      "step": 35710
    },
    {
      "epoch": 0.6869825273340963,
      "grad_norm": 421.0891993596618,
      "learning_rate": 4.540606620709575e-06,
      "loss": 336.9304,
      "step": 35720
    },
    {
      "epoch": 0.687174851669856,
      "grad_norm": 416.42238642172526,
      "learning_rate": 4.535494274134236e-06,
      "loss": 329.0043,
      "step": 35730
    },
    {
      "epoch": 0.6873671760056159,
      "grad_norm": 424.04608668078015,
      "learning_rate": 4.530383962997036e-06,
      "loss": 323.8193,
      "step": 35740
    },
    {
      "epoch": 0.6875595003413757,
      "grad_norm": 437.76803575672636,
      "learning_rate": 4.525275689201476e-06,
      "loss": 333.0717,
      "step": 35750
    },
    {
      "epoch": 0.6877518246771355,
      "grad_norm": 462.1188378558478,
      "learning_rate": 4.520169454650299e-06,
      "loss": 325.3063,
      "step": 35760
    },
    {
      "epoch": 0.6879441490128954,
      "grad_norm": 461.56490783765344,
      "learning_rate": 4.515065261245498e-06,
      "loss": 326.2405,
      "step": 35770
    },
    {
      "epoch": 0.6881364733486551,
      "grad_norm": 432.38445745970904,
      "learning_rate": 4.509963110888305e-06,
      "loss": 320.8102,
      "step": 35780
    },
    {
      "epoch": 0.688328797684415,
      "grad_norm": 422.6874764980561,
      "learning_rate": 4.504863005479182e-06,
      "loss": 312.7877,
      "step": 35790
    },
    {
      "epoch": 0.6885211220201748,
      "grad_norm": 456.0143171697234,
      "learning_rate": 4.499764946917834e-06,
      "loss": 325.3387,
      "step": 35800
    },
    {
      "epoch": 0.6887134463559347,
      "grad_norm": 417.593090752741,
      "learning_rate": 4.494668937103201e-06,
      "loss": 322.0784,
      "step": 35810
    },
    {
      "epoch": 0.6889057706916945,
      "grad_norm": 410.56786109087756,
      "learning_rate": 4.489574977933467e-06,
      "loss": 316.2785,
      "step": 35820
    },
    {
      "epoch": 0.6890980950274543,
      "grad_norm": 422.9725731446824,
      "learning_rate": 4.484483071306048e-06,
      "loss": 329.7869,
      "step": 35830
    },
    {
      "epoch": 0.6892904193632141,
      "grad_norm": 390.9590335764464,
      "learning_rate": 4.479393219117588e-06,
      "loss": 315.1229,
      "step": 35840
    },
    {
      "epoch": 0.689482743698974,
      "grad_norm": 389.16371732470805,
      "learning_rate": 4.474305423263984e-06,
      "loss": 318.954,
      "step": 35850
    },
    {
      "epoch": 0.6896750680347338,
      "grad_norm": 469.5618475205626,
      "learning_rate": 4.469219685640348e-06,
      "loss": 312.1851,
      "step": 35860
    },
    {
      "epoch": 0.6898673923704935,
      "grad_norm": 434.07403473881715,
      "learning_rate": 4.4641360081410356e-06,
      "loss": 321.7083,
      "step": 35870
    },
    {
      "epoch": 0.6900597167062534,
      "grad_norm": 452.0394084384201,
      "learning_rate": 4.4590543926596285e-06,
      "loss": 327.3819,
      "step": 35880
    },
    {
      "epoch": 0.6902520410420132,
      "grad_norm": 420.01793573626543,
      "learning_rate": 4.453974841088953e-06,
      "loss": 326.828,
      "step": 35890
    },
    {
      "epoch": 0.6904443653777731,
      "grad_norm": 432.57851679897203,
      "learning_rate": 4.448897355321049e-06,
      "loss": 320.2067,
      "step": 35900
    },
    {
      "epoch": 0.6906366897135329,
      "grad_norm": 400.5118038017992,
      "learning_rate": 4.443821937247205e-06,
      "loss": 319.4693,
      "step": 35910
    },
    {
      "epoch": 0.6908290140492928,
      "grad_norm": 453.0344500732784,
      "learning_rate": 4.438748588757922e-06,
      "loss": 320.6646,
      "step": 35920
    },
    {
      "epoch": 0.6910213383850525,
      "grad_norm": 402.8498358206257,
      "learning_rate": 4.433677311742945e-06,
      "loss": 317.5354,
      "step": 35930
    },
    {
      "epoch": 0.6912136627208124,
      "grad_norm": 421.0047258099239,
      "learning_rate": 4.428608108091241e-06,
      "loss": 316.8761,
      "step": 35940
    },
    {
      "epoch": 0.6914059870565722,
      "grad_norm": 434.6702319251514,
      "learning_rate": 4.4235409796909965e-06,
      "loss": 318.32,
      "step": 35950
    },
    {
      "epoch": 0.6915983113923321,
      "grad_norm": 453.69007174373684,
      "learning_rate": 4.418475928429644e-06,
      "loss": 318.7631,
      "step": 35960
    },
    {
      "epoch": 0.6917906357280919,
      "grad_norm": 427.19543539884825,
      "learning_rate": 4.413412956193826e-06,
      "loss": 331.0233,
      "step": 35970
    },
    {
      "epoch": 0.6919829600638516,
      "grad_norm": 462.50477600824416,
      "learning_rate": 4.4083520648694156e-06,
      "loss": 309.2859,
      "step": 35980
    },
    {
      "epoch": 0.6921752843996115,
      "grad_norm": 407.72338468544166,
      "learning_rate": 4.4032932563415075e-06,
      "loss": 317.1252,
      "step": 35990
    },
    {
      "epoch": 0.6923676087353713,
      "grad_norm": 410.47854713205356,
      "learning_rate": 4.398236532494434e-06,
      "loss": 318.1686,
      "step": 36000
    },
    {
      "epoch": 0.6925599330711312,
      "grad_norm": 426.3098330289822,
      "learning_rate": 4.393181895211735e-06,
      "loss": 330.8465,
      "step": 36010
    },
    {
      "epoch": 0.692752257406891,
      "grad_norm": 432.56987064272624,
      "learning_rate": 4.388129346376177e-06,
      "loss": 319.1718,
      "step": 36020
    },
    {
      "epoch": 0.6929445817426508,
      "grad_norm": 422.49937614367343,
      "learning_rate": 4.383078887869759e-06,
      "loss": 314.139,
      "step": 36030
    },
    {
      "epoch": 0.6931369060784106,
      "grad_norm": 393.68416366476873,
      "learning_rate": 4.378030521573683e-06,
      "loss": 312.853,
      "step": 36040
    },
    {
      "epoch": 0.6933292304141705,
      "grad_norm": 433.5787346913957,
      "learning_rate": 4.372984249368393e-06,
      "loss": 319.1204,
      "step": 36050
    },
    {
      "epoch": 0.6935215547499303,
      "grad_norm": 404.5171984953886,
      "learning_rate": 4.367940073133533e-06,
      "loss": 316.0725,
      "step": 36060
    },
    {
      "epoch": 0.6937138790856902,
      "grad_norm": 406.25670450214375,
      "learning_rate": 4.362897994747982e-06,
      "loss": 318.8176,
      "step": 36070
    },
    {
      "epoch": 0.6939062034214499,
      "grad_norm": 452.1534131155957,
      "learning_rate": 4.35785801608983e-06,
      "loss": 310.5954,
      "step": 36080
    },
    {
      "epoch": 0.6940985277572097,
      "grad_norm": 421.65515297007295,
      "learning_rate": 4.352820139036379e-06,
      "loss": 318.6872,
      "step": 36090
    },
    {
      "epoch": 0.6942908520929696,
      "grad_norm": 421.3804530036144,
      "learning_rate": 4.347784365464163e-06,
      "loss": 316.7017,
      "step": 36100
    },
    {
      "epoch": 0.6944831764287294,
      "grad_norm": 397.0646841525965,
      "learning_rate": 4.342750697248922e-06,
      "loss": 321.6771,
      "step": 36110
    },
    {
      "epoch": 0.6946755007644893,
      "grad_norm": 422.6723280240146,
      "learning_rate": 4.337719136265614e-06,
      "loss": 309.0128,
      "step": 36120
    },
    {
      "epoch": 0.694867825100249,
      "grad_norm": 415.55311502447444,
      "learning_rate": 4.332689684388408e-06,
      "loss": 313.8722,
      "step": 36130
    },
    {
      "epoch": 0.6950601494360089,
      "grad_norm": 423.4773718166429,
      "learning_rate": 4.327662343490701e-06,
      "loss": 317.5374,
      "step": 36140
    },
    {
      "epoch": 0.6952524737717687,
      "grad_norm": 452.1517974212016,
      "learning_rate": 4.322637115445088e-06,
      "loss": 328.0838,
      "step": 36150
    },
    {
      "epoch": 0.6954447981075286,
      "grad_norm": 426.1388858511226,
      "learning_rate": 4.3176140021233845e-06,
      "loss": 317.6973,
      "step": 36160
    },
    {
      "epoch": 0.6956371224432883,
      "grad_norm": 429.45168713059724,
      "learning_rate": 4.312593005396615e-06,
      "loss": 320.0352,
      "step": 36170
    },
    {
      "epoch": 0.6958294467790482,
      "grad_norm": 403.3143830262678,
      "learning_rate": 4.307574127135022e-06,
      "loss": 313.7953,
      "step": 36180
    },
    {
      "epoch": 0.696021771114808,
      "grad_norm": 436.50226966201575,
      "learning_rate": 4.3025573692080516e-06,
      "loss": 320.5324,
      "step": 36190
    },
    {
      "epoch": 0.6962140954505678,
      "grad_norm": 453.66195104880705,
      "learning_rate": 4.297542733484364e-06,
      "loss": 314.4829,
      "step": 36200
    },
    {
      "epoch": 0.6964064197863277,
      "grad_norm": 405.5417394794248,
      "learning_rate": 4.292530221831832e-06,
      "loss": 321.7942,
      "step": 36210
    },
    {
      "epoch": 0.6965987441220874,
      "grad_norm": 404.73616107383924,
      "learning_rate": 4.2875198361175305e-06,
      "loss": 318.7937,
      "step": 36220
    },
    {
      "epoch": 0.6967910684578473,
      "grad_norm": 399.56937900877875,
      "learning_rate": 4.282511578207746e-06,
      "loss": 314.5185,
      "step": 36230
    },
    {
      "epoch": 0.6969833927936071,
      "grad_norm": 406.6588756610703,
      "learning_rate": 4.277505449967967e-06,
      "loss": 322.8663,
      "step": 36240
    },
    {
      "epoch": 0.697175717129367,
      "grad_norm": 402.17404522228526,
      "learning_rate": 4.2725014532629015e-06,
      "loss": 319.3661,
      "step": 36250
    },
    {
      "epoch": 0.6973680414651268,
      "grad_norm": 452.1472078977409,
      "learning_rate": 4.267499589956453e-06,
      "loss": 319.5087,
      "step": 36260
    },
    {
      "epoch": 0.6975603658008867,
      "grad_norm": 403.30188694757953,
      "learning_rate": 4.262499861911727e-06,
      "loss": 326.616,
      "step": 36270
    },
    {
      "epoch": 0.6977526901366464,
      "grad_norm": 419.7092005627183,
      "learning_rate": 4.257502270991048e-06,
      "loss": 319.4938,
      "step": 36280
    },
    {
      "epoch": 0.6979450144724063,
      "grad_norm": 419.52135988075264,
      "learning_rate": 4.252506819055934e-06,
      "loss": 323.4875,
      "step": 36290
    },
    {
      "epoch": 0.6981373388081661,
      "grad_norm": 427.88695466954175,
      "learning_rate": 4.2475135079671045e-06,
      "loss": 319.5693,
      "step": 36300
    },
    {
      "epoch": 0.698329663143926,
      "grad_norm": 393.3490937110835,
      "learning_rate": 4.242522339584486e-06,
      "loss": 313.5424,
      "step": 36310
    },
    {
      "epoch": 0.6985219874796857,
      "grad_norm": 409.02350478239816,
      "learning_rate": 4.2375333157672114e-06,
      "loss": 319.2571,
      "step": 36320
    },
    {
      "epoch": 0.6987143118154455,
      "grad_norm": 423.795398104954,
      "learning_rate": 4.232546438373604e-06,
      "loss": 318.6126,
      "step": 36330
    },
    {
      "epoch": 0.6989066361512054,
      "grad_norm": 435.46200281753823,
      "learning_rate": 4.227561709261198e-06,
      "loss": 311.0569,
      "step": 36340
    },
    {
      "epoch": 0.6990989604869652,
      "grad_norm": 397.6911741704681,
      "learning_rate": 4.222579130286716e-06,
      "loss": 322.0474,
      "step": 36350
    },
    {
      "epoch": 0.6992912848227251,
      "grad_norm": 409.05316196009346,
      "learning_rate": 4.217598703306095e-06,
      "loss": 325.3665,
      "step": 36360
    },
    {
      "epoch": 0.6994836091584848,
      "grad_norm": 473.30051214422394,
      "learning_rate": 4.212620430174457e-06,
      "loss": 325.2845,
      "step": 36370
    },
    {
      "epoch": 0.6996759334942447,
      "grad_norm": 435.49511057850685,
      "learning_rate": 4.207644312746124e-06,
      "loss": 317.5285,
      "step": 36380
    },
    {
      "epoch": 0.6998682578300045,
      "grad_norm": 399.91260472400734,
      "learning_rate": 4.202670352874625e-06,
      "loss": 315.4795,
      "step": 36390
    },
    {
      "epoch": 0.7000605821657644,
      "grad_norm": 423.6605439545644,
      "learning_rate": 4.197698552412672e-06,
      "loss": 329.5323,
      "step": 36400
    },
    {
      "epoch": 0.7002529065015242,
      "grad_norm": 475.38245952731415,
      "learning_rate": 4.192728913212181e-06,
      "loss": 322.5399,
      "step": 36410
    },
    {
      "epoch": 0.700445230837284,
      "grad_norm": 390.3484468111378,
      "learning_rate": 4.187761437124256e-06,
      "loss": 309.5572,
      "step": 36420
    },
    {
      "epoch": 0.7006375551730438,
      "grad_norm": 475.8194702864543,
      "learning_rate": 4.182796125999207e-06,
      "loss": 327.1869,
      "step": 36430
    },
    {
      "epoch": 0.7008298795088036,
      "grad_norm": 509.108401621469,
      "learning_rate": 4.177832981686526e-06,
      "loss": 323.9206,
      "step": 36440
    },
    {
      "epoch": 0.7010222038445635,
      "grad_norm": 405.4920872724414,
      "learning_rate": 4.172872006034899e-06,
      "loss": 319.7336,
      "step": 36450
    },
    {
      "epoch": 0.7012145281803233,
      "grad_norm": 458.9281640547833,
      "learning_rate": 4.167913200892217e-06,
      "loss": 313.4159,
      "step": 36460
    },
    {
      "epoch": 0.7014068525160831,
      "grad_norm": 398.40214948296756,
      "learning_rate": 4.162956568105543e-06,
      "loss": 330.2787,
      "step": 36470
    },
    {
      "epoch": 0.7015991768518429,
      "grad_norm": 417.856516036245,
      "learning_rate": 4.158002109521149e-06,
      "loss": 314.8625,
      "step": 36480
    },
    {
      "epoch": 0.7017915011876028,
      "grad_norm": 445.42659397270586,
      "learning_rate": 4.153049826984482e-06,
      "loss": 319.1083,
      "step": 36490
    },
    {
      "epoch": 0.7019838255233626,
      "grad_norm": 418.46396241992085,
      "learning_rate": 4.148099722340192e-06,
      "loss": 330.4729,
      "step": 36500
    },
    {
      "epoch": 0.7021761498591225,
      "grad_norm": 424.9163536276126,
      "learning_rate": 4.143151797432109e-06,
      "loss": 316.9279,
      "step": 36510
    },
    {
      "epoch": 0.7023684741948822,
      "grad_norm": 411.56526316672574,
      "learning_rate": 4.1382060541032505e-06,
      "loss": 319.7038,
      "step": 36520
    },
    {
      "epoch": 0.7025607985306421,
      "grad_norm": 390.34747349856855,
      "learning_rate": 4.133262494195824e-06,
      "loss": 320.4265,
      "step": 36530
    },
    {
      "epoch": 0.7027531228664019,
      "grad_norm": 402.78935819454125,
      "learning_rate": 4.12832111955123e-06,
      "loss": 308.1256,
      "step": 36540
    },
    {
      "epoch": 0.7029454472021617,
      "grad_norm": 426.8640507669044,
      "learning_rate": 4.123381932010044e-06,
      "loss": 323.2139,
      "step": 36550
    },
    {
      "epoch": 0.7031377715379216,
      "grad_norm": 407.3328436129975,
      "learning_rate": 4.11844493341203e-06,
      "loss": 319.6881,
      "step": 36560
    },
    {
      "epoch": 0.7033300958736813,
      "grad_norm": 398.9993406988549,
      "learning_rate": 4.113510125596145e-06,
      "loss": 323.4684,
      "step": 36570
    },
    {
      "epoch": 0.7035224202094412,
      "grad_norm": 439.52176042845457,
      "learning_rate": 4.1085775104005186e-06,
      "loss": 308.9553,
      "step": 36580
    },
    {
      "epoch": 0.703714744545201,
      "grad_norm": 429.7479719974011,
      "learning_rate": 4.10364708966247e-06,
      "loss": 321.9735,
      "step": 36590
    },
    {
      "epoch": 0.7039070688809609,
      "grad_norm": 401.8833282337944,
      "learning_rate": 4.098718865218496e-06,
      "loss": 309.0076,
      "step": 36600
    },
    {
      "epoch": 0.7040993932167207,
      "grad_norm": 411.4078023684322,
      "learning_rate": 4.0937928389042815e-06,
      "loss": 327.422,
      "step": 36610
    },
    {
      "epoch": 0.7042917175524805,
      "grad_norm": 396.5296601982585,
      "learning_rate": 4.088869012554694e-06,
      "loss": 312.1202,
      "step": 36620
    },
    {
      "epoch": 0.7044840418882403,
      "grad_norm": 424.8644766208361,
      "learning_rate": 4.08394738800377e-06,
      "loss": 319.6057,
      "step": 36630
    },
    {
      "epoch": 0.7046763662240002,
      "grad_norm": 405.32483089107455,
      "learning_rate": 4.07902796708474e-06,
      "loss": 311.9721,
      "step": 36640
    },
    {
      "epoch": 0.70486869055976,
      "grad_norm": 435.80359522449277,
      "learning_rate": 4.074110751630005e-06,
      "loss": 326.5005,
      "step": 36650
    },
    {
      "epoch": 0.7050610148955198,
      "grad_norm": 424.77444273655794,
      "learning_rate": 4.0691957434711446e-06,
      "loss": 315.0795,
      "step": 36660
    },
    {
      "epoch": 0.7052533392312796,
      "grad_norm": 403.6574748576981,
      "learning_rate": 4.0642829444389165e-06,
      "loss": 315.7512,
      "step": 36670
    },
    {
      "epoch": 0.7054456635670394,
      "grad_norm": 390.903316881906,
      "learning_rate": 4.059372356363263e-06,
      "loss": 328.6013,
      "step": 36680
    },
    {
      "epoch": 0.7056379879027993,
      "grad_norm": 374.0867880608783,
      "learning_rate": 4.054463981073296e-06,
      "loss": 317.724,
      "step": 36690
    },
    {
      "epoch": 0.7058303122385591,
      "grad_norm": 407.68246192039663,
      "learning_rate": 4.049557820397297e-06,
      "loss": 322.0198,
      "step": 36700
    },
    {
      "epoch": 0.706022636574319,
      "grad_norm": 434.1481626400727,
      "learning_rate": 4.044653876162738e-06,
      "loss": 309.617,
      "step": 36710
    },
    {
      "epoch": 0.7062149609100787,
      "grad_norm": 412.56129671744196,
      "learning_rate": 4.039752150196257e-06,
      "loss": 327.7791,
      "step": 36720
    },
    {
      "epoch": 0.7064072852458386,
      "grad_norm": 381.78473739539356,
      "learning_rate": 4.034852644323661e-06,
      "loss": 320.3679,
      "step": 36730
    },
    {
      "epoch": 0.7065996095815984,
      "grad_norm": 392.7972631087244,
      "learning_rate": 4.029955360369935e-06,
      "loss": 312.0091,
      "step": 36740
    },
    {
      "epoch": 0.7067919339173583,
      "grad_norm": 432.1985367463212,
      "learning_rate": 4.0250603001592416e-06,
      "loss": 307.5055,
      "step": 36750
    },
    {
      "epoch": 0.7069842582531181,
      "grad_norm": 382.8019494597333,
      "learning_rate": 4.020167465514903e-06,
      "loss": 313.0305,
      "step": 36760
    },
    {
      "epoch": 0.7071765825888778,
      "grad_norm": 503.78838334506656,
      "learning_rate": 4.015276858259427e-06,
      "loss": 318.0343,
      "step": 36770
    },
    {
      "epoch": 0.7073689069246377,
      "grad_norm": 437.9965086349302,
      "learning_rate": 4.0103884802144775e-06,
      "loss": 320.7152,
      "step": 36780
    },
    {
      "epoch": 0.7075612312603975,
      "grad_norm": 408.0095816573821,
      "learning_rate": 4.0055023332009e-06,
      "loss": 319.4751,
      "step": 36790
    },
    {
      "epoch": 0.7077535555961574,
      "grad_norm": 396.7262535962733,
      "learning_rate": 4.000618419038702e-06,
      "loss": 302.0847,
      "step": 36800
    },
    {
      "epoch": 0.7079458799319172,
      "grad_norm": 407.658502782889,
      "learning_rate": 3.9957367395470555e-06,
      "loss": 323.6847,
      "step": 36810
    },
    {
      "epoch": 0.708138204267677,
      "grad_norm": 389.20227287430475,
      "learning_rate": 3.990857296544315e-06,
      "loss": 308.0567,
      "step": 36820
    },
    {
      "epoch": 0.7083305286034368,
      "grad_norm": 417.6877878225282,
      "learning_rate": 3.985980091847985e-06,
      "loss": 314.1982,
      "step": 36830
    },
    {
      "epoch": 0.7085228529391967,
      "grad_norm": 431.57888892217744,
      "learning_rate": 3.981105127274748e-06,
      "loss": 320.3472,
      "step": 36840
    },
    {
      "epoch": 0.7087151772749565,
      "grad_norm": 469.0253328628941,
      "learning_rate": 3.976232404640441e-06,
      "loss": 307.5793,
      "step": 36850
    },
    {
      "epoch": 0.7089075016107164,
      "grad_norm": 407.1340414933974,
      "learning_rate": 3.971361925760081e-06,
      "loss": 318.378,
      "step": 36860
    },
    {
      "epoch": 0.7090998259464761,
      "grad_norm": 418.751691105859,
      "learning_rate": 3.966493692447838e-06,
      "loss": 319.5382,
      "step": 36870
    },
    {
      "epoch": 0.7092921502822359,
      "grad_norm": 402.356303614616,
      "learning_rate": 3.961627706517044e-06,
      "loss": 314.2855,
      "step": 36880
    },
    {
      "epoch": 0.7094844746179958,
      "grad_norm": 425.57654010596576,
      "learning_rate": 3.956763969780206e-06,
      "loss": 314.7584,
      "step": 36890
    },
    {
      "epoch": 0.7096767989537556,
      "grad_norm": 401.22077890427266,
      "learning_rate": 3.951902484048978e-06,
      "loss": 311.5525,
      "step": 36900
    },
    {
      "epoch": 0.7098691232895155,
      "grad_norm": 469.4901116631315,
      "learning_rate": 3.94704325113419e-06,
      "loss": 314.7782,
      "step": 36910
    },
    {
      "epoch": 0.7100614476252752,
      "grad_norm": 392.1512097168587,
      "learning_rate": 3.942186272845821e-06,
      "loss": 313.5791,
      "step": 36920
    },
    {
      "epoch": 0.7102537719610351,
      "grad_norm": 399.68716512586286,
      "learning_rate": 3.937331550993021e-06,
      "loss": 316.7366,
      "step": 36930
    },
    {
      "epoch": 0.7104460962967949,
      "grad_norm": 462.0706652871905,
      "learning_rate": 3.932479087384089e-06,
      "loss": 323.077,
      "step": 36940
    },
    {
      "epoch": 0.7106384206325548,
      "grad_norm": 420.04291996058214,
      "learning_rate": 3.927628883826488e-06,
      "loss": 316.5559,
      "step": 36950
    },
    {
      "epoch": 0.7108307449683146,
      "grad_norm": 407.3918509059511,
      "learning_rate": 3.922780942126837e-06,
      "loss": 316.4484,
      "step": 36960
    },
    {
      "epoch": 0.7110230693040744,
      "grad_norm": 384.23940502453195,
      "learning_rate": 3.91793526409092e-06,
      "loss": 317.5919,
      "step": 36970
    },
    {
      "epoch": 0.7112153936398342,
      "grad_norm": 431.2796373970221,
      "learning_rate": 3.913091851523667e-06,
      "loss": 318.7198,
      "step": 36980
    },
    {
      "epoch": 0.711407717975594,
      "grad_norm": 404.95734120667794,
      "learning_rate": 3.908250706229168e-06,
      "loss": 310.8542,
      "step": 36990
    },
    {
      "epoch": 0.7116000423113539,
      "grad_norm": 391.65027125668917,
      "learning_rate": 3.903411830010676e-06,
      "loss": 326.9259,
      "step": 37000
    },
    {
      "epoch": 0.7117923666471137,
      "grad_norm": 409.17589512948126,
      "learning_rate": 3.8985752246705885e-06,
      "loss": 317.0762,
      "step": 37010
    },
    {
      "epoch": 0.7119846909828735,
      "grad_norm": 420.86180758733366,
      "learning_rate": 3.893740892010463e-06,
      "loss": 317.103,
      "step": 37020
    },
    {
      "epoch": 0.7121770153186333,
      "grad_norm": 404.1603006743833,
      "learning_rate": 3.888908833831002e-06,
      "loss": 320.6646,
      "step": 37030
    },
    {
      "epoch": 0.7123693396543932,
      "grad_norm": 407.91243268882334,
      "learning_rate": 3.884079051932073e-06,
      "loss": 318.9525,
      "step": 37040
    },
    {
      "epoch": 0.712561663990153,
      "grad_norm": 403.0385762888256,
      "learning_rate": 3.879251548112692e-06,
      "loss": 323.5288,
      "step": 37050
    },
    {
      "epoch": 0.7127539883259129,
      "grad_norm": 447.7552976398021,
      "learning_rate": 3.874426324171019e-06,
      "loss": 332.6827,
      "step": 37060
    },
    {
      "epoch": 0.7129463126616726,
      "grad_norm": 413.7088950370587,
      "learning_rate": 3.869603381904377e-06,
      "loss": 319.8625,
      "step": 37070
    },
    {
      "epoch": 0.7131386369974325,
      "grad_norm": 411.6214524011452,
      "learning_rate": 3.864782723109227e-06,
      "loss": 311.0306,
      "step": 37080
    },
    {
      "epoch": 0.7133309613331923,
      "grad_norm": 448.99238479862066,
      "learning_rate": 3.859964349581187e-06,
      "loss": 316.294,
      "step": 37090
    },
    {
      "epoch": 0.7135232856689521,
      "grad_norm": 428.94616260505717,
      "learning_rate": 3.855148263115017e-06,
      "loss": 316.1106,
      "step": 37100
    },
    {
      "epoch": 0.713715610004712,
      "grad_norm": 423.9711590581681,
      "learning_rate": 3.850334465504637e-06,
      "loss": 320.5208,
      "step": 37110
    },
    {
      "epoch": 0.7139079343404717,
      "grad_norm": 443.3688503581884,
      "learning_rate": 3.845522958543104e-06,
      "loss": 326.4777,
      "step": 37120
    },
    {
      "epoch": 0.7141002586762316,
      "grad_norm": 459.4295602855482,
      "learning_rate": 3.840713744022624e-06,
      "loss": 318.4702,
      "step": 37130
    },
    {
      "epoch": 0.7142925830119914,
      "grad_norm": 448.60210578785166,
      "learning_rate": 3.835906823734548e-06,
      "loss": 322.9629,
      "step": 37140
    },
    {
      "epoch": 0.7144849073477513,
      "grad_norm": 414.71344020079107,
      "learning_rate": 3.831102199469379e-06,
      "loss": 335.3739,
      "step": 37150
    },
    {
      "epoch": 0.714677231683511,
      "grad_norm": 408.4548346224963,
      "learning_rate": 3.826299873016758e-06,
      "loss": 320.6636,
      "step": 37160
    },
    {
      "epoch": 0.7148695560192709,
      "grad_norm": 396.12786440636427,
      "learning_rate": 3.821499846165468e-06,
      "loss": 307.3479,
      "step": 37170
    },
    {
      "epoch": 0.7150618803550307,
      "grad_norm": 440.68683484793706,
      "learning_rate": 3.816702120703449e-06,
      "loss": 317.8695,
      "step": 37180
    },
    {
      "epoch": 0.7152542046907906,
      "grad_norm": 400.4405466537782,
      "learning_rate": 3.8119066984177654e-06,
      "loss": 312.7973,
      "step": 37190
    },
    {
      "epoch": 0.7154465290265504,
      "grad_norm": 453.7587678612687,
      "learning_rate": 3.8071135810946415e-06,
      "loss": 318.936,
      "step": 37200
    },
    {
      "epoch": 0.7156388533623101,
      "grad_norm": 414.11514208287883,
      "learning_rate": 3.802322770519424e-06,
      "loss": 310.4042,
      "step": 37210
    },
    {
      "epoch": 0.71583117769807,
      "grad_norm": 402.54813440693385,
      "learning_rate": 3.7975342684766215e-06,
      "loss": 310.9606,
      "step": 37220
    },
    {
      "epoch": 0.7160235020338298,
      "grad_norm": 436.92638637071775,
      "learning_rate": 3.792748076749867e-06,
      "loss": 319.4578,
      "step": 37230
    },
    {
      "epoch": 0.7162158263695897,
      "grad_norm": 441.1669366932506,
      "learning_rate": 3.787964197121934e-06,
      "loss": 318.59,
      "step": 37240
    },
    {
      "epoch": 0.7164081507053495,
      "grad_norm": 407.8454849081635,
      "learning_rate": 3.7831826313747454e-06,
      "loss": 309.5386,
      "step": 37250
    },
    {
      "epoch": 0.7166004750411094,
      "grad_norm": 442.3110762810688,
      "learning_rate": 3.778403381289353e-06,
      "loss": 309.9025,
      "step": 37260
    },
    {
      "epoch": 0.7167927993768691,
      "grad_norm": 390.04546748849305,
      "learning_rate": 3.7736264486459486e-06,
      "loss": 312.0679,
      "step": 37270
    },
    {
      "epoch": 0.716985123712629,
      "grad_norm": 409.9691365520833,
      "learning_rate": 3.7688518352238555e-06,
      "loss": 313.9818,
      "step": 37280
    },
    {
      "epoch": 0.7171774480483888,
      "grad_norm": 427.13058434146336,
      "learning_rate": 3.7640795428015462e-06,
      "loss": 319.6796,
      "step": 37290
    },
    {
      "epoch": 0.7173697723841487,
      "grad_norm": 424.5965196432928,
      "learning_rate": 3.7593095731566186e-06,
      "loss": 314.6164,
      "step": 37300
    },
    {
      "epoch": 0.7175620967199084,
      "grad_norm": 425.14250060424666,
      "learning_rate": 3.7545419280658025e-06,
      "loss": 318.4672,
      "step": 37310
    },
    {
      "epoch": 0.7177544210556682,
      "grad_norm": 431.1156015438004,
      "learning_rate": 3.749776609304975e-06,
      "loss": 316.3071,
      "step": 37320
    },
    {
      "epoch": 0.7179467453914281,
      "grad_norm": 445.2991048054992,
      "learning_rate": 3.7450136186491315e-06,
      "loss": 320.987,
      "step": 37330
    },
    {
      "epoch": 0.7181390697271879,
      "grad_norm": 423.6614317551579,
      "learning_rate": 3.7402529578724134e-06,
      "loss": 315.6607,
      "step": 37340
    },
    {
      "epoch": 0.7183313940629478,
      "grad_norm": 403.65496163124425,
      "learning_rate": 3.735494628748082e-06,
      "loss": 315.3974,
      "step": 37350
    },
    {
      "epoch": 0.7185237183987075,
      "grad_norm": 417.0197634954238,
      "learning_rate": 3.730738633048543e-06,
      "loss": 312.5922,
      "step": 37360
    },
    {
      "epoch": 0.7187160427344674,
      "grad_norm": 401.36956774855327,
      "learning_rate": 3.7259849725453225e-06,
      "loss": 322.5538,
      "step": 37370
    },
    {
      "epoch": 0.7189083670702272,
      "grad_norm": 395.92819477555054,
      "learning_rate": 3.7212336490090815e-06,
      "loss": 314.0357,
      "step": 37380
    },
    {
      "epoch": 0.7191006914059871,
      "grad_norm": 416.44554550550464,
      "learning_rate": 3.7164846642096053e-06,
      "loss": 317.2763,
      "step": 37390
    },
    {
      "epoch": 0.7192930157417469,
      "grad_norm": 405.8560779783164,
      "learning_rate": 3.7117380199158204e-06,
      "loss": 324.1765,
      "step": 37400
    },
    {
      "epoch": 0.7194853400775068,
      "grad_norm": 405.44566321724454,
      "learning_rate": 3.706993717895768e-06,
      "loss": 312.6861,
      "step": 37410
    },
    {
      "epoch": 0.7196776644132665,
      "grad_norm": 443.2028923968177,
      "learning_rate": 3.70225175991662e-06,
      "loss": 317.0176,
      "step": 37420
    },
    {
      "epoch": 0.7198699887490264,
      "grad_norm": 447.2753215806813,
      "learning_rate": 3.697512147744684e-06,
      "loss": 323.2874,
      "step": 37430
    },
    {
      "epoch": 0.7200623130847862,
      "grad_norm": 427.65607213351524,
      "learning_rate": 3.6927748831453835e-06,
      "loss": 310.19,
      "step": 37440
    },
    {
      "epoch": 0.720254637420546,
      "grad_norm": 428.7363170292159,
      "learning_rate": 3.688039967883269e-06,
      "loss": 316.7893,
      "step": 37450
    },
    {
      "epoch": 0.7204469617563058,
      "grad_norm": 411.783774523239,
      "learning_rate": 3.683307403722025e-06,
      "loss": 312.1469,
      "step": 37460
    },
    {
      "epoch": 0.7206392860920656,
      "grad_norm": 406.5757628224676,
      "learning_rate": 3.678577192424445e-06,
      "loss": 312.4352,
      "step": 37470
    },
    {
      "epoch": 0.7208316104278255,
      "grad_norm": 469.7002525236319,
      "learning_rate": 3.6738493357524628e-06,
      "loss": 313.0454,
      "step": 37480
    },
    {
      "epoch": 0.7210239347635853,
      "grad_norm": 405.9544252843232,
      "learning_rate": 3.6691238354671233e-06,
      "loss": 317.7413,
      "step": 37490
    },
    {
      "epoch": 0.7212162590993452,
      "grad_norm": 450.3838190527834,
      "learning_rate": 3.664400693328595e-06,
      "loss": 317.8098,
      "step": 37500
    },
    {
      "epoch": 0.7214085834351049,
      "grad_norm": 397.37207343356437,
      "learning_rate": 3.6596799110961746e-06,
      "loss": 310.4092,
      "step": 37510
    },
    {
      "epoch": 0.7216009077708648,
      "grad_norm": 425.1938275623798,
      "learning_rate": 3.6549614905282724e-06,
      "loss": 322.1988,
      "step": 37520
    },
    {
      "epoch": 0.7217932321066246,
      "grad_norm": 409.6033057323231,
      "learning_rate": 3.6502454333824224e-06,
      "loss": 319.1281,
      "step": 37530
    },
    {
      "epoch": 0.7219855564423845,
      "grad_norm": 432.7548608601217,
      "learning_rate": 3.6455317414152803e-06,
      "loss": 317.6438,
      "step": 37540
    },
    {
      "epoch": 0.7221778807781443,
      "grad_norm": 429.2340631888676,
      "learning_rate": 3.640820416382618e-06,
      "loss": 318.313,
      "step": 37550
    },
    {
      "epoch": 0.722370205113904,
      "grad_norm": 388.91906563645216,
      "learning_rate": 3.6361114600393242e-06,
      "loss": 305.6224,
      "step": 37560
    },
    {
      "epoch": 0.7225625294496639,
      "grad_norm": 409.6265087580773,
      "learning_rate": 3.6314048741394057e-06,
      "loss": 322.2113,
      "step": 37570
    },
    {
      "epoch": 0.7227548537854237,
      "grad_norm": 405.2649128830983,
      "learning_rate": 3.6267006604359943e-06,
      "loss": 318.1316,
      "step": 37580
    },
    {
      "epoch": 0.7229471781211836,
      "grad_norm": 437.09071311299414,
      "learning_rate": 3.6219988206813285e-06,
      "loss": 325.6359,
      "step": 37590
    },
    {
      "epoch": 0.7231395024569434,
      "grad_norm": 401.24198030662205,
      "learning_rate": 3.6172993566267623e-06,
      "loss": 311.0217,
      "step": 37600
    },
    {
      "epoch": 0.7233318267927032,
      "grad_norm": 438.65354274660905,
      "learning_rate": 3.6126022700227715e-06,
      "loss": 316.6846,
      "step": 37610
    },
    {
      "epoch": 0.723524151128463,
      "grad_norm": 454.8033362099199,
      "learning_rate": 3.6079075626189476e-06,
      "loss": 319.8325,
      "step": 37620
    },
    {
      "epoch": 0.7237164754642229,
      "grad_norm": 412.55436477767176,
      "learning_rate": 3.603215236163987e-06,
      "loss": 309.7259,
      "step": 37630
    },
    {
      "epoch": 0.7239087997999827,
      "grad_norm": 407.03193529433855,
      "learning_rate": 3.5985252924057023e-06,
      "loss": 312.1112,
      "step": 37640
    },
    {
      "epoch": 0.7241011241357426,
      "grad_norm": 394.0981915137137,
      "learning_rate": 3.5938377330910245e-06,
      "loss": 315.1933,
      "step": 37650
    },
    {
      "epoch": 0.7242934484715023,
      "grad_norm": 422.2728382752941,
      "learning_rate": 3.5891525599659905e-06,
      "loss": 314.2316,
      "step": 37660
    },
    {
      "epoch": 0.7244857728072621,
      "grad_norm": 406.6467869133794,
      "learning_rate": 3.5844697747757496e-06,
      "loss": 310.8174,
      "step": 37670
    },
    {
      "epoch": 0.724678097143022,
      "grad_norm": 416.15640247974386,
      "learning_rate": 3.5797893792645577e-06,
      "loss": 316.6512,
      "step": 37680
    },
    {
      "epoch": 0.7248704214787818,
      "grad_norm": 468.2073568637038,
      "learning_rate": 3.5751113751757925e-06,
      "loss": 319.5047,
      "step": 37690
    },
    {
      "epoch": 0.7250627458145417,
      "grad_norm": 399.7030211088204,
      "learning_rate": 3.5704357642519295e-06,
      "loss": 311.1336,
      "step": 37700
    },
    {
      "epoch": 0.7252550701503014,
      "grad_norm": 466.3737316535616,
      "learning_rate": 3.5657625482345526e-06,
      "loss": 312.7548,
      "step": 37710
    },
    {
      "epoch": 0.7254473944860613,
      "grad_norm": 453.07729264993145,
      "learning_rate": 3.5610917288643655e-06,
      "loss": 313.9828,
      "step": 37720
    },
    {
      "epoch": 0.7256397188218211,
      "grad_norm": 442.5665134675431,
      "learning_rate": 3.556423307881167e-06,
      "loss": 318.8357,
      "step": 37730
    },
    {
      "epoch": 0.725832043157581,
      "grad_norm": 443.4626834334787,
      "learning_rate": 3.551757287023865e-06,
      "loss": 318.1125,
      "step": 37740
    },
    {
      "epoch": 0.7260243674933408,
      "grad_norm": 397.9692858351469,
      "learning_rate": 3.547093668030479e-06,
      "loss": 319.7514,
      "step": 37750
    },
    {
      "epoch": 0.7262166918291006,
      "grad_norm": 412.3175209156679,
      "learning_rate": 3.542432452638126e-06,
      "loss": 316.257,
      "step": 37760
    },
    {
      "epoch": 0.7264090161648604,
      "grad_norm": 416.65854698004955,
      "learning_rate": 3.5377736425830366e-06,
      "loss": 322.3939,
      "step": 37770
    },
    {
      "epoch": 0.7266013405006202,
      "grad_norm": 495.64094641424106,
      "learning_rate": 3.5331172396005354e-06,
      "loss": 310.4271,
      "step": 37780
    },
    {
      "epoch": 0.7267936648363801,
      "grad_norm": 414.5428819124079,
      "learning_rate": 3.528463245425062e-06,
      "loss": 322.1516,
      "step": 37790
    },
    {
      "epoch": 0.7269859891721399,
      "grad_norm": 405.47334307072526,
      "learning_rate": 3.5238116617901486e-06,
      "loss": 313.9452,
      "step": 37800
    },
    {
      "epoch": 0.7271783135078997,
      "grad_norm": 429.37216642645814,
      "learning_rate": 3.519162490428433e-06,
      "loss": 324.7638,
      "step": 37810
    },
    {
      "epoch": 0.7273706378436595,
      "grad_norm": 414.80287021366075,
      "learning_rate": 3.5145157330716516e-06,
      "loss": 310.1896,
      "step": 37820
    },
    {
      "epoch": 0.7275629621794194,
      "grad_norm": 404.3157837780895,
      "learning_rate": 3.509871391450652e-06,
      "loss": 324.8781,
      "step": 37830
    },
    {
      "epoch": 0.7277552865151792,
      "grad_norm": 388.671118795425,
      "learning_rate": 3.505229467295371e-06,
      "loss": 313.4352,
      "step": 37840
    },
    {
      "epoch": 0.7279476108509391,
      "grad_norm": 415.43047737031253,
      "learning_rate": 3.5005899623348493e-06,
      "loss": 327.8867,
      "step": 37850
    },
    {
      "epoch": 0.7281399351866988,
      "grad_norm": 403.89400308963826,
      "learning_rate": 3.495952878297221e-06,
      "loss": 319.3212,
      "step": 37860
    },
    {
      "epoch": 0.7283322595224587,
      "grad_norm": 403.2433499585081,
      "learning_rate": 3.4913182169097315e-06,
      "loss": 311.9184,
      "step": 37870
    },
    {
      "epoch": 0.7285245838582185,
      "grad_norm": 398.9078501818294,
      "learning_rate": 3.4866859798987084e-06,
      "loss": 314.8927,
      "step": 37880
    },
    {
      "epoch": 0.7287169081939783,
      "grad_norm": 443.4683761769683,
      "learning_rate": 3.4820561689895906e-06,
      "loss": 316.0644,
      "step": 37890
    },
    {
      "epoch": 0.7289092325297382,
      "grad_norm": 449.0231561169924,
      "learning_rate": 3.4774287859068988e-06,
      "loss": 310.4906,
      "step": 37900
    },
    {
      "epoch": 0.7291015568654979,
      "grad_norm": 452.6404643791422,
      "learning_rate": 3.472803832374263e-06,
      "loss": 306.8345,
      "step": 37910
    },
    {
      "epoch": 0.7292938812012578,
      "grad_norm": 406.7836696763572,
      "learning_rate": 3.4681813101144e-06,
      "loss": 318.8201,
      "step": 37920
    },
    {
      "epoch": 0.7294862055370176,
      "grad_norm": 389.4193236787394,
      "learning_rate": 3.4635612208491197e-06,
      "loss": 319.6306,
      "step": 37930
    },
    {
      "epoch": 0.7296785298727775,
      "grad_norm": 389.113814219266,
      "learning_rate": 3.458943566299334e-06,
      "loss": 304.6019,
      "step": 37940
    },
    {
      "epoch": 0.7298708542085373,
      "grad_norm": 451.71800765339873,
      "learning_rate": 3.454328348185042e-06,
      "loss": 304.9947,
      "step": 37950
    },
    {
      "epoch": 0.7300631785442971,
      "grad_norm": 406.58528268947873,
      "learning_rate": 3.4497155682253314e-06,
      "loss": 322.682,
      "step": 37960
    },
    {
      "epoch": 0.7302555028800569,
      "grad_norm": 387.11664039755044,
      "learning_rate": 3.4451052281383922e-06,
      "loss": 309.2688,
      "step": 37970
    },
    {
      "epoch": 0.7304478272158168,
      "grad_norm": 438.7083412236591,
      "learning_rate": 3.440497329641499e-06,
      "loss": 321.3788,
      "step": 37980
    },
    {
      "epoch": 0.7306401515515766,
      "grad_norm": 436.585959123442,
      "learning_rate": 3.435891874451017e-06,
      "loss": 312.4924,
      "step": 37990
    },
    {
      "epoch": 0.7308324758873364,
      "grad_norm": 408.98789655135624,
      "learning_rate": 3.431288864282398e-06,
      "loss": 312.9037,
      "step": 38000
    },
    {
      "epoch": 0.7310248002230962,
      "grad_norm": 417.8799339967779,
      "learning_rate": 3.4266883008501937e-06,
      "loss": 308.0152,
      "step": 38010
    },
    {
      "epoch": 0.731217124558856,
      "grad_norm": 443.45688469634183,
      "learning_rate": 3.4220901858680365e-06,
      "loss": 321.1792,
      "step": 38020
    },
    {
      "epoch": 0.7314094488946159,
      "grad_norm": 424.7660227997121,
      "learning_rate": 3.4174945210486445e-06,
      "loss": 319.3943,
      "step": 38030
    },
    {
      "epoch": 0.7316017732303757,
      "grad_norm": 384.3072078283791,
      "learning_rate": 3.4129013081038285e-06,
      "loss": 319.874,
      "step": 38040
    },
    {
      "epoch": 0.7317940975661356,
      "grad_norm": 412.9184162702637,
      "learning_rate": 3.40831054874449e-06,
      "loss": 311.6062,
      "step": 38050
    },
    {
      "epoch": 0.7319864219018953,
      "grad_norm": 406.7459799692194,
      "learning_rate": 3.403722244680606e-06,
      "loss": 311.9208,
      "step": 38060
    },
    {
      "epoch": 0.7321787462376552,
      "grad_norm": 434.1742568733442,
      "learning_rate": 3.3991363976212423e-06,
      "loss": 324.9531,
      "step": 38070
    },
    {
      "epoch": 0.732371070573415,
      "grad_norm": 425.61658690525115,
      "learning_rate": 3.394553009274556e-06,
      "loss": 310.1825,
      "step": 38080
    },
    {
      "epoch": 0.7325633949091749,
      "grad_norm": 405.2715800197643,
      "learning_rate": 3.389972081347782e-06,
      "loss": 311.7358,
      "step": 38090
    },
    {
      "epoch": 0.7327557192449347,
      "grad_norm": 430.6477516855534,
      "learning_rate": 3.385393615547239e-06,
      "loss": 323.1989,
      "step": 38100
    },
    {
      "epoch": 0.7329480435806944,
      "grad_norm": 417.5917856276371,
      "learning_rate": 3.3808176135783276e-06,
      "loss": 329.5747,
      "step": 38110
    },
    {
      "epoch": 0.7331403679164543,
      "grad_norm": 420.3486848807001,
      "learning_rate": 3.3762440771455386e-06,
      "loss": 320.8793,
      "step": 38120
    },
    {
      "epoch": 0.7333326922522141,
      "grad_norm": 402.3293222437787,
      "learning_rate": 3.371673007952435e-06,
      "loss": 313.1816,
      "step": 38130
    },
    {
      "epoch": 0.733525016587974,
      "grad_norm": 448.2393352823683,
      "learning_rate": 3.3671044077016634e-06,
      "loss": 328.3257,
      "step": 38140
    },
    {
      "epoch": 0.7337173409237338,
      "grad_norm": 385.64988460853056,
      "learning_rate": 3.3625382780949576e-06,
      "loss": 320.212,
      "step": 38150
    },
    {
      "epoch": 0.7339096652594936,
      "grad_norm": 413.66587294721336,
      "learning_rate": 3.357974620833121e-06,
      "loss": 319.9798,
      "step": 38160
    },
    {
      "epoch": 0.7341019895952534,
      "grad_norm": 438.22631973845125,
      "learning_rate": 3.353413437616039e-06,
      "loss": 320.5506,
      "step": 38170
    },
    {
      "epoch": 0.7342943139310133,
      "grad_norm": 448.269953081504,
      "learning_rate": 3.3488547301426786e-06,
      "loss": 301.8495,
      "step": 38180
    },
    {
      "epoch": 0.7344866382667731,
      "grad_norm": 405.8556961922498,
      "learning_rate": 3.344298500111087e-06,
      "loss": 314.2083,
      "step": 38190
    },
    {
      "epoch": 0.734678962602533,
      "grad_norm": 409.08674419414484,
      "learning_rate": 3.3397447492183833e-06,
      "loss": 315.1931,
      "step": 38200
    },
    {
      "epoch": 0.7348712869382927,
      "grad_norm": 427.04741052235056,
      "learning_rate": 3.3351934791607576e-06,
      "loss": 317.8736,
      "step": 38210
    },
    {
      "epoch": 0.7350636112740525,
      "grad_norm": 397.0244738739313,
      "learning_rate": 3.330644691633492e-06,
      "loss": 315.4427,
      "step": 38220
    },
    {
      "epoch": 0.7352559356098124,
      "grad_norm": 427.23375430807096,
      "learning_rate": 3.3260983883309306e-06,
      "loss": 324.4806,
      "step": 38230
    },
    {
      "epoch": 0.7354482599455722,
      "grad_norm": 403.6796958049927,
      "learning_rate": 3.321554570946497e-06,
      "loss": 310.0045,
      "step": 38240
    },
    {
      "epoch": 0.735640584281332,
      "grad_norm": 411.0996078047051,
      "learning_rate": 3.317013241172684e-06,
      "loss": 316.0688,
      "step": 38250
    },
    {
      "epoch": 0.7358329086170918,
      "grad_norm": 403.78661228846784,
      "learning_rate": 3.3124744007010688e-06,
      "loss": 312.0583,
      "step": 38260
    },
    {
      "epoch": 0.7360252329528517,
      "grad_norm": 430.2940958442948,
      "learning_rate": 3.3079380512222904e-06,
      "loss": 320.3145,
      "step": 38270
    },
    {
      "epoch": 0.7362175572886115,
      "grad_norm": 409.9006288150114,
      "learning_rate": 3.3034041944260654e-06,
      "loss": 318.8557,
      "step": 38280
    },
    {
      "epoch": 0.7364098816243714,
      "grad_norm": 424.47316890104634,
      "learning_rate": 3.2988728320011774e-06,
      "loss": 314.2572,
      "step": 38290
    },
    {
      "epoch": 0.7366022059601312,
      "grad_norm": 419.28056990895914,
      "learning_rate": 3.294343965635489e-06,
      "loss": 314.8191,
      "step": 38300
    },
    {
      "epoch": 0.736794530295891,
      "grad_norm": 373.9067611329959,
      "learning_rate": 3.289817597015923e-06,
      "loss": 305.4609,
      "step": 38310
    },
    {
      "epoch": 0.7369868546316508,
      "grad_norm": 422.6967809774257,
      "learning_rate": 3.2852937278284837e-06,
      "loss": 318.4763,
      "step": 38320
    },
    {
      "epoch": 0.7371791789674106,
      "grad_norm": 406.8098272778151,
      "learning_rate": 3.28077235975823e-06,
      "loss": 313.6993,
      "step": 38330
    },
    {
      "epoch": 0.7373715033031705,
      "grad_norm": 411.6673975567329,
      "learning_rate": 3.2762534944893033e-06,
      "loss": 319.8252,
      "step": 38340
    },
    {
      "epoch": 0.7375638276389302,
      "grad_norm": 395.8632164995816,
      "learning_rate": 3.271737133704904e-06,
      "loss": 308.1031,
      "step": 38350
    },
    {
      "epoch": 0.7377561519746901,
      "grad_norm": 398.6223668239311,
      "learning_rate": 3.2672232790872983e-06,
      "loss": 316.9474,
      "step": 38360
    },
    {
      "epoch": 0.7379484763104499,
      "grad_norm": 407.15574223548026,
      "learning_rate": 3.262711932317828e-06,
      "loss": 319.3968,
      "step": 38370
    },
    {
      "epoch": 0.7381408006462098,
      "grad_norm": 421.81286585835966,
      "learning_rate": 3.258203095076894e-06,
      "loss": 313.7264,
      "step": 38380
    },
    {
      "epoch": 0.7383331249819696,
      "grad_norm": 436.11465659176275,
      "learning_rate": 3.2536967690439592e-06,
      "loss": 326.133,
      "step": 38390
    },
    {
      "epoch": 0.7385254493177295,
      "grad_norm": 385.3559491294718,
      "learning_rate": 3.249192955897562e-06,
      "loss": 313.9984,
      "step": 38400
    },
    {
      "epoch": 0.7387177736534892,
      "grad_norm": 394.3691437351689,
      "learning_rate": 3.2446916573152955e-06,
      "loss": 318.277,
      "step": 38410
    },
    {
      "epoch": 0.7389100979892491,
      "grad_norm": 392.2305322079415,
      "learning_rate": 3.24019287497382e-06,
      "loss": 309.2865,
      "step": 38420
    },
    {
      "epoch": 0.7391024223250089,
      "grad_norm": 406.1065949033608,
      "learning_rate": 3.235696610548852e-06,
      "loss": 305.6187,
      "step": 38430
    },
    {
      "epoch": 0.7392947466607688,
      "grad_norm": 436.93456010581,
      "learning_rate": 3.231202865715184e-06,
      "loss": 317.0469,
      "step": 38440
    },
    {
      "epoch": 0.7394870709965286,
      "grad_norm": 380.5858387775878,
      "learning_rate": 3.226711642146655e-06,
      "loss": 325.8808,
      "step": 38450
    },
    {
      "epoch": 0.7396793953322883,
      "grad_norm": 412.50051100575945,
      "learning_rate": 3.222222941516179e-06,
      "loss": 313.3394,
      "step": 38460
    },
    {
      "epoch": 0.7398717196680482,
      "grad_norm": 379.09514848672194,
      "learning_rate": 3.2177367654957137e-06,
      "loss": 305.7161,
      "step": 38470
    },
    {
      "epoch": 0.740064044003808,
      "grad_norm": 444.2549819893377,
      "learning_rate": 3.213253115756295e-06,
      "loss": 315.6344,
      "step": 38480
    },
    {
      "epoch": 0.7402563683395679,
      "grad_norm": 428.0221820660699,
      "learning_rate": 3.208771993968003e-06,
      "loss": 315.8494,
      "step": 38490
    },
    {
      "epoch": 0.7404486926753276,
      "grad_norm": 472.9267878368491,
      "learning_rate": 3.2042934017999795e-06,
      "loss": 311.2093,
      "step": 38500
    },
    {
      "epoch": 0.7406410170110875,
      "grad_norm": 400.60202215807976,
      "learning_rate": 3.1998173409204326e-06,
      "loss": 313.0489,
      "step": 38510
    },
    {
      "epoch": 0.7408333413468473,
      "grad_norm": 413.8431248478503,
      "learning_rate": 3.1953438129966175e-06,
      "loss": 324.425,
      "step": 38520
    },
    {
      "epoch": 0.7410256656826072,
      "grad_norm": 430.22678825788074,
      "learning_rate": 3.190872819694849e-06,
      "loss": 323.9846,
      "step": 38530
    },
    {
      "epoch": 0.741217990018367,
      "grad_norm": 384.82680452873507,
      "learning_rate": 3.1864043626804953e-06,
      "loss": 307.6771,
      "step": 38540
    },
    {
      "epoch": 0.7414103143541269,
      "grad_norm": 394.34326680230373,
      "learning_rate": 3.18193844361799e-06,
      "loss": 305.6898,
      "step": 38550
    },
    {
      "epoch": 0.7416026386898866,
      "grad_norm": 398.7812262812955,
      "learning_rate": 3.1774750641708095e-06,
      "loss": 308.9979,
      "step": 38560
    },
    {
      "epoch": 0.7417949630256464,
      "grad_norm": 404.1360343295936,
      "learning_rate": 3.1730142260014875e-06,
      "loss": 315.4173,
      "step": 38570
    },
    {
      "epoch": 0.7419872873614063,
      "grad_norm": 436.04929807229973,
      "learning_rate": 3.1685559307716187e-06,
      "loss": 312.3365,
      "step": 38580
    },
    {
      "epoch": 0.7421796116971661,
      "grad_norm": 455.0903400495209,
      "learning_rate": 3.164100180141839e-06,
      "loss": 313.875,
      "step": 38590
    },
    {
      "epoch": 0.742371936032926,
      "grad_norm": 425.8210960824381,
      "learning_rate": 3.159646975771842e-06,
      "loss": 318.1196,
      "step": 38600
    },
    {
      "epoch": 0.7425642603686857,
      "grad_norm": 461.5249003242236,
      "learning_rate": 3.155196319320374e-06,
      "loss": 320.8121,
      "step": 38610
    },
    {
      "epoch": 0.7427565847044456,
      "grad_norm": 409.07309283229864,
      "learning_rate": 3.1507482124452337e-06,
      "loss": 319.5608,
      "step": 38620
    },
    {
      "epoch": 0.7429489090402054,
      "grad_norm": 400.91881172283905,
      "learning_rate": 3.146302656803266e-06,
      "loss": 311.4764,
      "step": 38630
    },
    {
      "epoch": 0.7431412333759653,
      "grad_norm": 410.3369833827823,
      "learning_rate": 3.1418596540503653e-06,
      "loss": 311.0298,
      "step": 38640
    },
    {
      "epoch": 0.743333557711725,
      "grad_norm": 415.79677935980015,
      "learning_rate": 3.1374192058414755e-06,
      "loss": 325.3141,
      "step": 38650
    },
    {
      "epoch": 0.7435258820474849,
      "grad_norm": 408.23200671095384,
      "learning_rate": 3.1329813138305944e-06,
      "loss": 319.9275,
      "step": 38660
    },
    {
      "epoch": 0.7437182063832447,
      "grad_norm": 406.28556278001923,
      "learning_rate": 3.128545979670762e-06,
      "loss": 305.9547,
      "step": 38670
    },
    {
      "epoch": 0.7439105307190045,
      "grad_norm": 392.1745443789781,
      "learning_rate": 3.124113205014063e-06,
      "loss": 309.163,
      "step": 38680
    },
    {
      "epoch": 0.7441028550547644,
      "grad_norm": 415.78221120051256,
      "learning_rate": 3.119682991511639e-06,
      "loss": 316.8025,
      "step": 38690
    },
    {
      "epoch": 0.7442951793905241,
      "grad_norm": 405.75088787988915,
      "learning_rate": 3.1152553408136686e-06,
      "loss": 311.4661,
      "step": 38700
    },
    {
      "epoch": 0.744487503726284,
      "grad_norm": 421.81033576565244,
      "learning_rate": 3.110830254569378e-06,
      "loss": 315.2426,
      "step": 38710
    },
    {
      "epoch": 0.7446798280620438,
      "grad_norm": 402.1520592952528,
      "learning_rate": 3.106407734427037e-06,
      "loss": 317.4716,
      "step": 38720
    },
    {
      "epoch": 0.7448721523978037,
      "grad_norm": 437.1015581421016,
      "learning_rate": 3.101987782033966e-06,
      "loss": 316.8594,
      "step": 38730
    },
    {
      "epoch": 0.7450644767335635,
      "grad_norm": 408.8132878412377,
      "learning_rate": 3.097570399036519e-06,
      "loss": 316.3783,
      "step": 38740
    },
    {
      "epoch": 0.7452568010693233,
      "grad_norm": 409.239612133476,
      "learning_rate": 3.0931555870801033e-06,
      "loss": 307.0697,
      "step": 38750
    },
    {
      "epoch": 0.7454491254050831,
      "grad_norm": 381.41884243184774,
      "learning_rate": 3.0887433478091587e-06,
      "loss": 311.733,
      "step": 38760
    },
    {
      "epoch": 0.745641449740843,
      "grad_norm": 391.3036778784672,
      "learning_rate": 3.0843336828671765e-06,
      "loss": 314.4483,
      "step": 38770
    },
    {
      "epoch": 0.7458337740766028,
      "grad_norm": 428.2361792481648,
      "learning_rate": 3.079926593896683e-06,
      "loss": 313.0411,
      "step": 38780
    },
    {
      "epoch": 0.7460260984123626,
      "grad_norm": 417.96610594513123,
      "learning_rate": 3.0755220825392397e-06,
      "loss": 309.2721,
      "step": 38790
    },
    {
      "epoch": 0.7462184227481224,
      "grad_norm": 429.42513687118833,
      "learning_rate": 3.0711201504354628e-06,
      "loss": 314.5482,
      "step": 38800
    },
    {
      "epoch": 0.7464107470838822,
      "grad_norm": 384.40194923734737,
      "learning_rate": 3.0667207992249948e-06,
      "loss": 317.0427,
      "step": 38810
    },
    {
      "epoch": 0.7466030714196421,
      "grad_norm": 415.2178957596277,
      "learning_rate": 3.062324030546523e-06,
      "loss": 316.397,
      "step": 38820
    },
    {
      "epoch": 0.7467953957554019,
      "grad_norm": 412.9864857331533,
      "learning_rate": 3.057929846037767e-06,
      "loss": 309.3145,
      "step": 38830
    },
    {
      "epoch": 0.7469877200911618,
      "grad_norm": 480.89559260098406,
      "learning_rate": 3.0535382473354945e-06,
      "loss": 317.5547,
      "step": 38840
    },
    {
      "epoch": 0.7471800444269215,
      "grad_norm": 436.69637992833316,
      "learning_rate": 3.0491492360755003e-06,
      "loss": 317.3891,
      "step": 38850
    },
    {
      "epoch": 0.7473723687626814,
      "grad_norm": 450.4614629079015,
      "learning_rate": 3.0447628138926153e-06,
      "loss": 305.826,
      "step": 38860
    },
    {
      "epoch": 0.7475646930984412,
      "grad_norm": 423.0453978926065,
      "learning_rate": 3.0403789824207165e-06,
      "loss": 305.9098,
      "step": 38870
    },
    {
      "epoch": 0.7477570174342011,
      "grad_norm": 406.5975703764876,
      "learning_rate": 3.0359977432927013e-06,
      "loss": 310.9963,
      "step": 38880
    },
    {
      "epoch": 0.7479493417699609,
      "grad_norm": 394.79896699951405,
      "learning_rate": 3.0316190981405147e-06,
      "loss": 309.3581,
      "step": 38890
    },
    {
      "epoch": 0.7481416661057206,
      "grad_norm": 400.8230855064271,
      "learning_rate": 3.0272430485951244e-06,
      "loss": 307.8479,
      "step": 38900
    },
    {
      "epoch": 0.7483339904414805,
      "grad_norm": 393.4205622519115,
      "learning_rate": 3.0228695962865438e-06,
      "loss": 316.2204,
      "step": 38910
    },
    {
      "epoch": 0.7485263147772403,
      "grad_norm": 398.46960597663616,
      "learning_rate": 3.018498742843806e-06,
      "loss": 318.3917,
      "step": 38920
    },
    {
      "epoch": 0.7487186391130002,
      "grad_norm": 425.43287271154475,
      "learning_rate": 3.014130489894982e-06,
      "loss": 314.5338,
      "step": 38930
    },
    {
      "epoch": 0.74891096344876,
      "grad_norm": 417.4891674980009,
      "learning_rate": 3.0097648390671765e-06,
      "loss": 306.0104,
      "step": 38940
    },
    {
      "epoch": 0.7491032877845198,
      "grad_norm": 413.84242900851444,
      "learning_rate": 3.005401791986522e-06,
      "loss": 322.1133,
      "step": 38950
    },
    {
      "epoch": 0.7492956121202796,
      "grad_norm": 411.02152390458764,
      "learning_rate": 3.00104135027818e-06,
      "loss": 321.4357,
      "step": 38960
    },
    {
      "epoch": 0.7494879364560395,
      "grad_norm": 415.3366665583693,
      "learning_rate": 2.99668351556634e-06,
      "loss": 316.4434,
      "step": 38970
    },
    {
      "epoch": 0.7496802607917993,
      "grad_norm": 409.21065674768164,
      "learning_rate": 2.99232828947423e-06,
      "loss": 317.5683,
      "step": 38980
    },
    {
      "epoch": 0.7498725851275592,
      "grad_norm": 428.6979813797801,
      "learning_rate": 2.987975673624096e-06,
      "loss": 307.8015,
      "step": 38990
    },
    {
      "epoch": 0.7500649094633189,
      "grad_norm": 420.5654456185048,
      "learning_rate": 2.9836256696372178e-06,
      "loss": 321.1392,
      "step": 39000
    },
    {
      "epoch": 0.7502572337990787,
      "grad_norm": 423.43887924544737,
      "learning_rate": 2.9792782791338936e-06,
      "loss": 316.6789,
      "step": 39010
    },
    {
      "epoch": 0.7504495581348386,
      "grad_norm": 403.8965121103728,
      "learning_rate": 2.9749335037334604e-06,
      "loss": 318.7557,
      "step": 39020
    },
    {
      "epoch": 0.7506418824705984,
      "grad_norm": 407.46914141929426,
      "learning_rate": 2.9705913450542777e-06,
      "loss": 308.4499,
      "step": 39030
    },
    {
      "epoch": 0.7508342068063583,
      "grad_norm": 400.33160146764527,
      "learning_rate": 2.9662518047137214e-06,
      "loss": 310.8097,
      "step": 39040
    },
    {
      "epoch": 0.751026531142118,
      "grad_norm": 427.741101676746,
      "learning_rate": 2.961914884328203e-06,
      "loss": 315.2042,
      "step": 39050
    },
    {
      "epoch": 0.7512188554778779,
      "grad_norm": 414.16283904309296,
      "learning_rate": 2.9575805855131546e-06,
      "loss": 313.6494,
      "step": 39060
    },
    {
      "epoch": 0.7514111798136377,
      "grad_norm": 397.01997779053977,
      "learning_rate": 2.9532489098830274e-06,
      "loss": 303.9551,
      "step": 39070
    },
    {
      "epoch": 0.7516035041493976,
      "grad_norm": 418.6923368309888,
      "learning_rate": 2.9489198590512967e-06,
      "loss": 312.293,
      "step": 39080
    },
    {
      "epoch": 0.7517958284851574,
      "grad_norm": 436.20305301883366,
      "learning_rate": 2.9445934346304706e-06,
      "loss": 315.7977,
      "step": 39090
    },
    {
      "epoch": 0.7519881528209172,
      "grad_norm": 399.2215920290596,
      "learning_rate": 2.940269638232065e-06,
      "loss": 313.1067,
      "step": 39100
    },
    {
      "epoch": 0.752180477156677,
      "grad_norm": 410.72525240529393,
      "learning_rate": 2.935948471466622e-06,
      "loss": 310.0512,
      "step": 39110
    },
    {
      "epoch": 0.7523728014924368,
      "grad_norm": 401.15164377495205,
      "learning_rate": 2.9316299359437085e-06,
      "loss": 311.5574,
      "step": 39120
    },
    {
      "epoch": 0.7525651258281967,
      "grad_norm": 435.2378524559211,
      "learning_rate": 2.9273140332719064e-06,
      "loss": 309.722,
      "step": 39130
    },
    {
      "epoch": 0.7527574501639565,
      "grad_norm": 447.7548916501179,
      "learning_rate": 2.923000765058818e-06,
      "loss": 312.8755,
      "step": 39140
    },
    {
      "epoch": 0.7529497744997163,
      "grad_norm": 425.27265018010866,
      "learning_rate": 2.9186901329110605e-06,
      "loss": 305.6466,
      "step": 39150
    },
    {
      "epoch": 0.7531420988354761,
      "grad_norm": 434.84847325270107,
      "learning_rate": 2.9143821384342808e-06,
      "loss": 308.9021,
      "step": 39160
    },
    {
      "epoch": 0.753334423171236,
      "grad_norm": 442.1234647641681,
      "learning_rate": 2.9100767832331277e-06,
      "loss": 310.0948,
      "step": 39170
    },
    {
      "epoch": 0.7535267475069958,
      "grad_norm": 408.522686489624,
      "learning_rate": 2.9057740689112822e-06,
      "loss": 319.8424,
      "step": 39180
    },
    {
      "epoch": 0.7537190718427557,
      "grad_norm": 412.93660270428484,
      "learning_rate": 2.901473997071428e-06,
      "loss": 312.1669,
      "step": 39190
    },
    {
      "epoch": 0.7539113961785154,
      "grad_norm": 398.20484252143274,
      "learning_rate": 2.8971765693152767e-06,
      "loss": 311.1424,
      "step": 39200
    },
    {
      "epoch": 0.7541037205142753,
      "grad_norm": 380.1655257161292,
      "learning_rate": 2.8928817872435465e-06,
      "loss": 308.9361,
      "step": 39210
    },
    {
      "epoch": 0.7542960448500351,
      "grad_norm": 393.81366191113125,
      "learning_rate": 2.8885896524559696e-06,
      "loss": 316.0979,
      "step": 39220
    },
    {
      "epoch": 0.7544883691857949,
      "grad_norm": 428.9532058511396,
      "learning_rate": 2.8843001665513016e-06,
      "loss": 312.6022,
      "step": 39230
    },
    {
      "epoch": 0.7546806935215548,
      "grad_norm": 403.03062248401363,
      "learning_rate": 2.8800133311273016e-06,
      "loss": 319.5132,
      "step": 39240
    },
    {
      "epoch": 0.7548730178573145,
      "grad_norm": 434.6596352874526,
      "learning_rate": 2.875729147780745e-06,
      "loss": 316.7485,
      "step": 39250
    },
    {
      "epoch": 0.7550653421930744,
      "grad_norm": 433.0055960376108,
      "learning_rate": 2.871447618107417e-06,
      "loss": 312.0756,
      "step": 39260
    },
    {
      "epoch": 0.7552576665288342,
      "grad_norm": 403.7441360507335,
      "learning_rate": 2.867168743702122e-06,
      "loss": 305.5635,
      "step": 39270
    },
    {
      "epoch": 0.7554499908645941,
      "grad_norm": 442.02211654745304,
      "learning_rate": 2.8628925261586683e-06,
      "loss": 317.6886,
      "step": 39280
    },
    {
      "epoch": 0.7556423152003539,
      "grad_norm": 452.1194678872526,
      "learning_rate": 2.8586189670698717e-06,
      "loss": 313.3257,
      "step": 39290
    },
    {
      "epoch": 0.7558346395361137,
      "grad_norm": 389.67226983157343,
      "learning_rate": 2.854348068027568e-06,
      "loss": 311.0332,
      "step": 39300
    },
    {
      "epoch": 0.7560269638718735,
      "grad_norm": 423.33268277708675,
      "learning_rate": 2.850079830622593e-06,
      "loss": 319.8657,
      "step": 39310
    },
    {
      "epoch": 0.7562192882076334,
      "grad_norm": 442.3294562210493,
      "learning_rate": 2.845814256444799e-06,
      "loss": 326.6072,
      "step": 39320
    },
    {
      "epoch": 0.7564116125433932,
      "grad_norm": 395.6122928340354,
      "learning_rate": 2.8415513470830357e-06,
      "loss": 320.7574,
      "step": 39330
    },
    {
      "epoch": 0.756603936879153,
      "grad_norm": 431.5769886032502,
      "learning_rate": 2.837291104125174e-06,
      "loss": 319.5021,
      "step": 39340
    },
    {
      "epoch": 0.7567962612149128,
      "grad_norm": 403.0776949615088,
      "learning_rate": 2.833033529158079e-06,
      "loss": 310.1233,
      "step": 39350
    },
    {
      "epoch": 0.7569885855506726,
      "grad_norm": 420.767776531208,
      "learning_rate": 2.8287786237676253e-06,
      "loss": 322.8885,
      "step": 39360
    },
    {
      "epoch": 0.7571809098864325,
      "grad_norm": 435.1093309648183,
      "learning_rate": 2.824526389538701e-06,
      "loss": 322.0231,
      "step": 39370
    },
    {
      "epoch": 0.7573732342221923,
      "grad_norm": 428.29271802338576,
      "learning_rate": 2.8202768280551894e-06,
      "loss": 308.8249,
      "step": 39380
    },
    {
      "epoch": 0.7575655585579522,
      "grad_norm": 420.9540628043602,
      "learning_rate": 2.8160299408999827e-06,
      "loss": 322.9439,
      "step": 39390
    },
    {
      "epoch": 0.7577578828937119,
      "grad_norm": 407.3357297867067,
      "learning_rate": 2.811785729654972e-06,
      "loss": 315.7543,
      "step": 39400
    },
    {
      "epoch": 0.7579502072294718,
      "grad_norm": 405.290836739935,
      "learning_rate": 2.8075441959010628e-06,
      "loss": 309.8894,
      "step": 39410
    },
    {
      "epoch": 0.7581425315652316,
      "grad_norm": 401.76609331373015,
      "learning_rate": 2.8033053412181543e-06,
      "loss": 311.5659,
      "step": 39420
    },
    {
      "epoch": 0.7583348559009915,
      "grad_norm": 375.5352116966792,
      "learning_rate": 2.799069167185148e-06,
      "loss": 297.6623,
      "step": 39430
    },
    {
      "epoch": 0.7585271802367513,
      "grad_norm": 412.79498904241206,
      "learning_rate": 2.7948356753799466e-06,
      "loss": 322.5181,
      "step": 39440
    },
    {
      "epoch": 0.758719504572511,
      "grad_norm": 407.22493665973343,
      "learning_rate": 2.7906048673794593e-06,
      "loss": 310.8787,
      "step": 39450
    },
    {
      "epoch": 0.7589118289082709,
      "grad_norm": 412.6630847413725,
      "learning_rate": 2.7863767447595946e-06,
      "loss": 318.2519,
      "step": 39460
    },
    {
      "epoch": 0.7591041532440307,
      "grad_norm": 384.441955634724,
      "learning_rate": 2.7821513090952523e-06,
      "loss": 308.0261,
      "step": 39470
    },
    {
      "epoch": 0.7592964775797906,
      "grad_norm": 406.67638619866375,
      "learning_rate": 2.7779285619603446e-06,
      "loss": 313.2911,
      "step": 39480
    },
    {
      "epoch": 0.7594888019155503,
      "grad_norm": 381.2990944588106,
      "learning_rate": 2.77370850492777e-06,
      "loss": 325.8067,
      "step": 39490
    },
    {
      "epoch": 0.7596811262513102,
      "grad_norm": 445.0230904938098,
      "learning_rate": 2.7694911395694324e-06,
      "loss": 314.7334,
      "step": 39500
    },
    {
      "epoch": 0.75987345058707,
      "grad_norm": 403.1905932745522,
      "learning_rate": 2.765276467456225e-06,
      "loss": 318.6229,
      "step": 39510
    },
    {
      "epoch": 0.7600657749228299,
      "grad_norm": 458.2073033713597,
      "learning_rate": 2.761064490158052e-06,
      "loss": 316.019,
      "step": 39520
    },
    {
      "epoch": 0.7602580992585897,
      "grad_norm": 408.56165535659346,
      "learning_rate": 2.7568552092438018e-06,
      "loss": 312.4274,
      "step": 39530
    },
    {
      "epoch": 0.7604504235943496,
      "grad_norm": 385.1376898888618,
      "learning_rate": 2.7526486262813578e-06,
      "loss": 311.3018,
      "step": 39540
    },
    {
      "epoch": 0.7606427479301093,
      "grad_norm": 440.24580985049084,
      "learning_rate": 2.7484447428376094e-06,
      "loss": 304.7419,
      "step": 39550
    },
    {
      "epoch": 0.7608350722658692,
      "grad_norm": 432.4582625779512,
      "learning_rate": 2.7442435604784313e-06,
      "loss": 316.1508,
      "step": 39560
    },
    {
      "epoch": 0.761027396601629,
      "grad_norm": 448.45357357743336,
      "learning_rate": 2.740045080768694e-06,
      "loss": 313.9462,
      "step": 39570
    },
    {
      "epoch": 0.7612197209373888,
      "grad_norm": 380.86378975493,
      "learning_rate": 2.7358493052722603e-06,
      "loss": 310.8535,
      "step": 39580
    },
    {
      "epoch": 0.7614120452731487,
      "grad_norm": 445.4903320462483,
      "learning_rate": 2.7316562355519904e-06,
      "loss": 311.0319,
      "step": 39590
    },
    {
      "epoch": 0.7616043696089084,
      "grad_norm": 445.3212157028574,
      "learning_rate": 2.727465873169729e-06,
      "loss": 313.6938,
      "step": 39600
    },
    {
      "epoch": 0.7617966939446683,
      "grad_norm": 404.25712049662417,
      "learning_rate": 2.723278219686324e-06,
      "loss": 313.5266,
      "step": 39610
    },
    {
      "epoch": 0.7619890182804281,
      "grad_norm": 421.3557108061471,
      "learning_rate": 2.7190932766615998e-06,
      "loss": 314.9394,
      "step": 39620
    },
    {
      "epoch": 0.762181342616188,
      "grad_norm": 386.97066855709954,
      "learning_rate": 2.714911045654385e-06,
      "loss": 306.5084,
      "step": 39630
    },
    {
      "epoch": 0.7623736669519477,
      "grad_norm": 460.82167527419216,
      "learning_rate": 2.7107315282224878e-06,
      "loss": 312.0479,
      "step": 39640
    },
    {
      "epoch": 0.7625659912877076,
      "grad_norm": 406.8182848455515,
      "learning_rate": 2.7065547259227078e-06,
      "loss": 316.5486,
      "step": 39650
    },
    {
      "epoch": 0.7627583156234674,
      "grad_norm": 413.99856626749084,
      "learning_rate": 2.7023806403108397e-06,
      "loss": 319.0753,
      "step": 39660
    },
    {
      "epoch": 0.7629506399592273,
      "grad_norm": 411.90051587927303,
      "learning_rate": 2.698209272941659e-06,
      "loss": 306.7832,
      "step": 39670
    },
    {
      "epoch": 0.7631429642949871,
      "grad_norm": 457.43580844036217,
      "learning_rate": 2.694040625368931e-06,
      "loss": 326.498,
      "step": 39680
    },
    {
      "epoch": 0.7633352886307468,
      "grad_norm": 411.30369848863705,
      "learning_rate": 2.689874699145405e-06,
      "loss": 308.9756,
      "step": 39690
    },
    {
      "epoch": 0.7635276129665067,
      "grad_norm": 398.3373121598249,
      "learning_rate": 2.685711495822827e-06,
      "loss": 311.1085,
      "step": 39700
    },
    {
      "epoch": 0.7637199373022665,
      "grad_norm": 428.94310525011616,
      "learning_rate": 2.6815510169519164e-06,
      "loss": 308.3243,
      "step": 39710
    },
    {
      "epoch": 0.7639122616380264,
      "grad_norm": 396.94977533954454,
      "learning_rate": 2.677393264082381e-06,
      "loss": 323.8346,
      "step": 39720
    },
    {
      "epoch": 0.7641045859737862,
      "grad_norm": 385.67101920801804,
      "learning_rate": 2.673238238762921e-06,
      "loss": 315.482,
      "step": 39730
    },
    {
      "epoch": 0.764296910309546,
      "grad_norm": 436.1851237957672,
      "learning_rate": 2.6690859425412075e-06,
      "loss": 315.1936,
      "step": 39740
    },
    {
      "epoch": 0.7644892346453058,
      "grad_norm": 448.8862210291068,
      "learning_rate": 2.6649363769639103e-06,
      "loss": 326.1787,
      "step": 39750
    },
    {
      "epoch": 0.7646815589810657,
      "grad_norm": 423.4744383861163,
      "learning_rate": 2.660789543576667e-06,
      "loss": 313.9057,
      "step": 39760
    },
    {
      "epoch": 0.7648738833168255,
      "grad_norm": 393.9053991402812,
      "learning_rate": 2.6566454439241107e-06,
      "loss": 317.3791,
      "step": 39770
    },
    {
      "epoch": 0.7650662076525854,
      "grad_norm": 428.47595962990806,
      "learning_rate": 2.652504079549848e-06,
      "loss": 323.639,
      "step": 39780
    },
    {
      "epoch": 0.7652585319883451,
      "grad_norm": 410.35898028314455,
      "learning_rate": 2.648365451996466e-06,
      "loss": 315.4431,
      "step": 39790
    },
    {
      "epoch": 0.7654508563241049,
      "grad_norm": 373.75404609373265,
      "learning_rate": 2.6442295628055346e-06,
      "loss": 305.6955,
      "step": 39800
    },
    {
      "epoch": 0.7656431806598648,
      "grad_norm": 417.54418082322036,
      "learning_rate": 2.64009641351761e-06,
      "loss": 310.0194,
      "step": 39810
    },
    {
      "epoch": 0.7658355049956246,
      "grad_norm": 385.0059852460938,
      "learning_rate": 2.635966005672218e-06,
      "loss": 314.499,
      "step": 39820
    },
    {
      "epoch": 0.7660278293313845,
      "grad_norm": 415.00045271537886,
      "learning_rate": 2.631838340807865e-06,
      "loss": 316.8466,
      "step": 39830
    },
    {
      "epoch": 0.7662201536671442,
      "grad_norm": 387.6144328066203,
      "learning_rate": 2.6277134204620436e-06,
      "loss": 312.4498,
      "step": 39840
    },
    {
      "epoch": 0.7664124780029041,
      "grad_norm": 447.39456456731784,
      "learning_rate": 2.6235912461712167e-06,
      "loss": 323.8219,
      "step": 39850
    },
    {
      "epoch": 0.7666048023386639,
      "grad_norm": 440.41725177903106,
      "learning_rate": 2.619471819470821e-06,
      "loss": 310.6018,
      "step": 39860
    },
    {
      "epoch": 0.7667971266744238,
      "grad_norm": 394.7354704432082,
      "learning_rate": 2.6153551418952827e-06,
      "loss": 304.8577,
      "step": 39870
    },
    {
      "epoch": 0.7669894510101836,
      "grad_norm": 403.09031460703915,
      "learning_rate": 2.6112412149779888e-06,
      "loss": 314.8679,
      "step": 39880
    },
    {
      "epoch": 0.7671817753459435,
      "grad_norm": 417.9209442539809,
      "learning_rate": 2.6071300402513165e-06,
      "loss": 315.9194,
      "step": 39890
    },
    {
      "epoch": 0.7673740996817032,
      "grad_norm": 420.64007850047153,
      "learning_rate": 2.603021619246604e-06,
      "loss": 318.3601,
      "step": 39900
    },
    {
      "epoch": 0.767566424017463,
      "grad_norm": 396.55820321757284,
      "learning_rate": 2.5989159534941768e-06,
      "loss": 309.9709,
      "step": 39910
    },
    {
      "epoch": 0.7677587483532229,
      "grad_norm": 367.8801136751504,
      "learning_rate": 2.5948130445233232e-06,
      "loss": 312.2183,
      "step": 39920
    },
    {
      "epoch": 0.7679510726889827,
      "grad_norm": 414.36640348418854,
      "learning_rate": 2.5907128938623093e-06,
      "loss": 313.2414,
      "step": 39930
    },
    {
      "epoch": 0.7681433970247425,
      "grad_norm": 430.1949195026494,
      "learning_rate": 2.5866155030383722e-06,
      "loss": 321.1011,
      "step": 39940
    },
    {
      "epoch": 0.7683357213605023,
      "grad_norm": 429.312936849423,
      "learning_rate": 2.582520873577726e-06,
      "loss": 307.8642,
      "step": 39950
    },
    {
      "epoch": 0.7685280456962622,
      "grad_norm": 393.2588040831892,
      "learning_rate": 2.578429007005552e-06,
      "loss": 311.7097,
      "step": 39960
    },
    {
      "epoch": 0.768720370032022,
      "grad_norm": 416.42721342845925,
      "learning_rate": 2.5743399048460004e-06,
      "loss": 317.4759,
      "step": 39970
    },
    {
      "epoch": 0.7689126943677819,
      "grad_norm": 400.9684695727884,
      "learning_rate": 2.570253568622193e-06,
      "loss": 314.3595,
      "step": 39980
    },
    {
      "epoch": 0.7691050187035416,
      "grad_norm": 427.113626606421,
      "learning_rate": 2.5661699998562286e-06,
      "loss": 315.2968,
      "step": 39990
    },
    {
      "epoch": 0.7692973430393015,
      "grad_norm": 423.94129070785107,
      "learning_rate": 2.5620892000691643e-06,
      "loss": 327.664,
      "step": 40000
    },
    {
      "epoch": 0.7694896673750613,
      "grad_norm": 401.10430676856475,
      "learning_rate": 2.5580111707810296e-06,
      "loss": 305.0914,
      "step": 40010
    },
    {
      "epoch": 0.7696819917108211,
      "grad_norm": 468.06844511648166,
      "learning_rate": 2.5539359135108244e-06,
      "loss": 315.7456,
      "step": 40020
    },
    {
      "epoch": 0.769874316046581,
      "grad_norm": 386.7072814924101,
      "learning_rate": 2.549863429776519e-06,
      "loss": 306.9337,
      "step": 40030
    },
    {
      "epoch": 0.7700666403823407,
      "grad_norm": 401.47087065572555,
      "learning_rate": 2.5457937210950433e-06,
      "loss": 313.9831,
      "step": 40040
    },
    {
      "epoch": 0.7702589647181006,
      "grad_norm": 431.8558423452183,
      "learning_rate": 2.541726788982294e-06,
      "loss": 312.0791,
      "step": 40050
    },
    {
      "epoch": 0.7704512890538604,
      "grad_norm": 390.3045247412946,
      "learning_rate": 2.5376626349531395e-06,
      "loss": 312.8002,
      "step": 40060
    },
    {
      "epoch": 0.7706436133896203,
      "grad_norm": 423.7457867952994,
      "learning_rate": 2.53360126052141e-06,
      "loss": 319.1218,
      "step": 40070
    },
    {
      "epoch": 0.7708359377253801,
      "grad_norm": 427.5206262482864,
      "learning_rate": 2.529542667199896e-06,
      "loss": 320.6042,
      "step": 40080
    },
    {
      "epoch": 0.77102826206114,
      "grad_norm": 401.83857228784666,
      "learning_rate": 2.525486856500363e-06,
      "loss": 309.5803,
      "step": 40090
    },
    {
      "epoch": 0.7712205863968997,
      "grad_norm": 371.0653425091557,
      "learning_rate": 2.5214338299335306e-06,
      "loss": 314.1878,
      "step": 40100
    },
    {
      "epoch": 0.7714129107326596,
      "grad_norm": 403.61409540579876,
      "learning_rate": 2.5173835890090826e-06,
      "loss": 315.9101,
      "step": 40110
    },
    {
      "epoch": 0.7716052350684194,
      "grad_norm": 432.4538903693824,
      "learning_rate": 2.5133361352356666e-06,
      "loss": 311.8193,
      "step": 40120
    },
    {
      "epoch": 0.7717975594041792,
      "grad_norm": 397.61404430746956,
      "learning_rate": 2.5092914701208958e-06,
      "loss": 302.5074,
      "step": 40130
    },
    {
      "epoch": 0.771989883739939,
      "grad_norm": 380.3919254318495,
      "learning_rate": 2.5052495951713406e-06,
      "loss": 313.205,
      "step": 40140
    },
    {
      "epoch": 0.7721822080756988,
      "grad_norm": 391.2171480059374,
      "learning_rate": 2.5012105118925267e-06,
      "loss": 309.1393,
      "step": 40150
    },
    {
      "epoch": 0.7723745324114587,
      "grad_norm": 437.88153051460057,
      "learning_rate": 2.497174221788955e-06,
      "loss": 309.5633,
      "step": 40160
    },
    {
      "epoch": 0.7725668567472185,
      "grad_norm": 417.6253970950156,
      "learning_rate": 2.4931407263640683e-06,
      "loss": 313.5437,
      "step": 40170
    },
    {
      "epoch": 0.7727591810829784,
      "grad_norm": 413.36382260639436,
      "learning_rate": 2.489110027120285e-06,
      "loss": 315.954,
      "step": 40180
    },
    {
      "epoch": 0.7729515054187381,
      "grad_norm": 411.60598692056806,
      "learning_rate": 2.4850821255589664e-06,
      "loss": 309.8239,
      "step": 40190
    },
    {
      "epoch": 0.773143829754498,
      "grad_norm": 400.3319595781661,
      "learning_rate": 2.4810570231804463e-06,
      "loss": 320.8611,
      "step": 40200
    },
    {
      "epoch": 0.7733361540902578,
      "grad_norm": 393.2527964700215,
      "learning_rate": 2.4770347214840063e-06,
      "loss": 309.9603,
      "step": 40210
    },
    {
      "epoch": 0.7735284784260177,
      "grad_norm": 378.5322457103529,
      "learning_rate": 2.473015221967886e-06,
      "loss": 309.0794,
      "step": 40220
    },
    {
      "epoch": 0.7737208027617775,
      "grad_norm": 443.02013964119135,
      "learning_rate": 2.4689985261292805e-06,
      "loss": 307.6548,
      "step": 40230
    },
    {
      "epoch": 0.7739131270975372,
      "grad_norm": 416.5870913965198,
      "learning_rate": 2.464984635464348e-06,
      "loss": 311.1075,
      "step": 40240
    },
    {
      "epoch": 0.7741054514332971,
      "grad_norm": 373.9705085984144,
      "learning_rate": 2.460973551468194e-06,
      "loss": 296.5506,
      "step": 40250
    },
    {
      "epoch": 0.7742977757690569,
      "grad_norm": 421.02398762512104,
      "learning_rate": 2.456965275634878e-06,
      "loss": 318.1376,
      "step": 40260
    },
    {
      "epoch": 0.7744901001048168,
      "grad_norm": 420.4784677538794,
      "learning_rate": 2.4529598094574226e-06,
      "loss": 317.219,
      "step": 40270
    },
    {
      "epoch": 0.7746824244405766,
      "grad_norm": 417.6769893268246,
      "learning_rate": 2.4489571544277944e-06,
      "loss": 309.7965,
      "step": 40280
    },
    {
      "epoch": 0.7748747487763364,
      "grad_norm": 391.3772824446553,
      "learning_rate": 2.444957312036914e-06,
      "loss": 319.464,
      "step": 40290
    },
    {
      "epoch": 0.7750670731120962,
      "grad_norm": 401.08784043456615,
      "learning_rate": 2.4409602837746625e-06,
      "loss": 306.3166,
      "step": 40300
    },
    {
      "epoch": 0.7752593974478561,
      "grad_norm": 381.8706334187539,
      "learning_rate": 2.4369660711298603e-06,
      "loss": 307.2619,
      "step": 40310
    },
    {
      "epoch": 0.7754517217836159,
      "grad_norm": 420.02826108132604,
      "learning_rate": 2.4329746755902917e-06,
      "loss": 320.832,
      "step": 40320
    },
    {
      "epoch": 0.7756440461193758,
      "grad_norm": 431.8368821581145,
      "learning_rate": 2.428986098642684e-06,
      "loss": 309.1702,
      "step": 40330
    },
    {
      "epoch": 0.7758363704551355,
      "grad_norm": 403.6901944052936,
      "learning_rate": 2.425000341772711e-06,
      "loss": 312.0167,
      "step": 40340
    },
    {
      "epoch": 0.7760286947908953,
      "grad_norm": 428.7012456511869,
      "learning_rate": 2.4210174064650084e-06,
      "loss": 302.6539,
      "step": 40350
    },
    {
      "epoch": 0.7762210191266552,
      "grad_norm": 437.6464942151356,
      "learning_rate": 2.4170372942031506e-06,
      "loss": 310.3356,
      "step": 40360
    },
    {
      "epoch": 0.776413343462415,
      "grad_norm": 442.77140645883435,
      "learning_rate": 2.4130600064696618e-06,
      "loss": 311.8165,
      "step": 40370
    },
    {
      "epoch": 0.7766056677981749,
      "grad_norm": 419.62815482563906,
      "learning_rate": 2.4090855447460205e-06,
      "loss": 316.7124,
      "step": 40380
    },
    {
      "epoch": 0.7767979921339346,
      "grad_norm": 423.4392566345319,
      "learning_rate": 2.4051139105126463e-06,
      "loss": 314.872,
      "step": 40390
    },
    {
      "epoch": 0.7769903164696945,
      "grad_norm": 448.4811597860652,
      "learning_rate": 2.4011451052489064e-06,
      "loss": 319.0694,
      "step": 40400
    },
    {
      "epoch": 0.7771826408054543,
      "grad_norm": 426.027252547883,
      "learning_rate": 2.3971791304331125e-06,
      "loss": 306.6793,
      "step": 40410
    },
    {
      "epoch": 0.7773749651412142,
      "grad_norm": 402.44634297955156,
      "learning_rate": 2.393215987542531e-06,
      "loss": 312.2662,
      "step": 40420
    },
    {
      "epoch": 0.777567289476974,
      "grad_norm": 410.22711626699464,
      "learning_rate": 2.3892556780533606e-06,
      "loss": 328.041,
      "step": 40430
    },
    {
      "epoch": 0.7777596138127338,
      "grad_norm": 423.5711571934931,
      "learning_rate": 2.385298203440758e-06,
      "loss": 309.2229,
      "step": 40440
    },
    {
      "epoch": 0.7779519381484936,
      "grad_norm": 405.83359144268087,
      "learning_rate": 2.3813435651788107e-06,
      "loss": 312.5401,
      "step": 40450
    },
    {
      "epoch": 0.7781442624842534,
      "grad_norm": 380.9433366245735,
      "learning_rate": 2.377391764740562e-06,
      "loss": 306.0173,
      "step": 40460
    },
    {
      "epoch": 0.7783365868200133,
      "grad_norm": 400.21931784841496,
      "learning_rate": 2.3734428035979883e-06,
      "loss": 301.3098,
      "step": 40470
    },
    {
      "epoch": 0.778528911155773,
      "grad_norm": 391.2951260508504,
      "learning_rate": 2.3694966832220123e-06,
      "loss": 312.3013,
      "step": 40480
    },
    {
      "epoch": 0.7787212354915329,
      "grad_norm": 422.0511609043098,
      "learning_rate": 2.365553405082501e-06,
      "loss": 315.2574,
      "step": 40490
    },
    {
      "epoch": 0.7789135598272927,
      "grad_norm": 381.0920268929168,
      "learning_rate": 2.3616129706482604e-06,
      "loss": 314.7387,
      "step": 40500
    },
    {
      "epoch": 0.7791058841630526,
      "grad_norm": 421.1473939104334,
      "learning_rate": 2.357675381387036e-06,
      "loss": 312.7013,
      "step": 40510
    },
    {
      "epoch": 0.7792982084988124,
      "grad_norm": 408.1076473581412,
      "learning_rate": 2.3537406387655114e-06,
      "loss": 308.5221,
      "step": 40520
    },
    {
      "epoch": 0.7794905328345723,
      "grad_norm": 401.4157039030911,
      "learning_rate": 2.349808744249321e-06,
      "loss": 312.1836,
      "step": 40530
    },
    {
      "epoch": 0.779682857170332,
      "grad_norm": 389.9347954939756,
      "learning_rate": 2.345879699303025e-06,
      "loss": 311.4137,
      "step": 40540
    },
    {
      "epoch": 0.7798751815060919,
      "grad_norm": 423.9304598170786,
      "learning_rate": 2.3419535053901264e-06,
      "loss": 310.1337,
      "step": 40550
    },
    {
      "epoch": 0.7800675058418517,
      "grad_norm": 477.8369375150172,
      "learning_rate": 2.338030163973073e-06,
      "loss": 327.3379,
      "step": 40560
    },
    {
      "epoch": 0.7802598301776115,
      "grad_norm": 466.7243912405653,
      "learning_rate": 2.334109676513242e-06,
      "loss": 312.5931,
      "step": 40570
    },
    {
      "epoch": 0.7804521545133714,
      "grad_norm": 394.05278841500524,
      "learning_rate": 2.330192044470948e-06,
      "loss": 308.7047,
      "step": 40580
    },
    {
      "epoch": 0.7806444788491311,
      "grad_norm": 408.1179151660319,
      "learning_rate": 2.3262772693054457e-06,
      "loss": 320.7805,
      "step": 40590
    },
    {
      "epoch": 0.780836803184891,
      "grad_norm": 435.99017207023576,
      "learning_rate": 2.322365352474928e-06,
      "loss": 310.3925,
      "step": 40600
    },
    {
      "epoch": 0.7810291275206508,
      "grad_norm": 430.64363398922904,
      "learning_rate": 2.3184562954365153e-06,
      "loss": 320.0543,
      "step": 40610
    },
    {
      "epoch": 0.7812214518564107,
      "grad_norm": 428.5405584811241,
      "learning_rate": 2.3145500996462656e-06,
      "loss": 318.8713,
      "step": 40620
    },
    {
      "epoch": 0.7814137761921705,
      "grad_norm": 442.9949011554765,
      "learning_rate": 2.310646766559177e-06,
      "loss": 321.6667,
      "step": 40630
    },
    {
      "epoch": 0.7816061005279303,
      "grad_norm": 394.32070730531797,
      "learning_rate": 2.3067462976291744e-06,
      "loss": 324.4715,
      "step": 40640
    },
    {
      "epoch": 0.7817984248636901,
      "grad_norm": 395.5026398048101,
      "learning_rate": 2.302848694309118e-06,
      "loss": 317.8809,
      "step": 40650
    },
    {
      "epoch": 0.78199074919945,
      "grad_norm": 411.3453331322445,
      "learning_rate": 2.2989539580507957e-06,
      "loss": 315.451,
      "step": 40660
    },
    {
      "epoch": 0.7821830735352098,
      "grad_norm": 390.90484889462147,
      "learning_rate": 2.2950620903049414e-06,
      "loss": 316.9897,
      "step": 40670
    },
    {
      "epoch": 0.7823753978709697,
      "grad_norm": 421.70828080504424,
      "learning_rate": 2.2911730925212073e-06,
      "loss": 316.1592,
      "step": 40680
    },
    {
      "epoch": 0.7825677222067294,
      "grad_norm": 451.22643446392846,
      "learning_rate": 2.2872869661481766e-06,
      "loss": 305.51,
      "step": 40690
    },
    {
      "epoch": 0.7827600465424892,
      "grad_norm": 414.62841862885125,
      "learning_rate": 2.283403712633375e-06,
      "loss": 306.5684,
      "step": 40700
    },
    {
      "epoch": 0.7829523708782491,
      "grad_norm": 430.6780913238953,
      "learning_rate": 2.279523333423247e-06,
      "loss": 322.1448,
      "step": 40710
    },
    {
      "epoch": 0.7831446952140089,
      "grad_norm": 409.51508280048375,
      "learning_rate": 2.2756458299631667e-06,
      "loss": 302.9692,
      "step": 40720
    },
    {
      "epoch": 0.7833370195497688,
      "grad_norm": 437.0953657669929,
      "learning_rate": 2.271771203697445e-06,
      "loss": 311.2765,
      "step": 40730
    },
    {
      "epoch": 0.7835293438855285,
      "grad_norm": 424.05630715867176,
      "learning_rate": 2.267899456069311e-06,
      "loss": 313.9041,
      "step": 40740
    },
    {
      "epoch": 0.7837216682212884,
      "grad_norm": 420.6614023581561,
      "learning_rate": 2.2640305885209336e-06,
      "loss": 320.6284,
      "step": 40750
    },
    {
      "epoch": 0.7839139925570482,
      "grad_norm": 415.27999943813995,
      "learning_rate": 2.2601646024933976e-06,
      "loss": 321.8764,
      "step": 40760
    },
    {
      "epoch": 0.7841063168928081,
      "grad_norm": 396.26756268775796,
      "learning_rate": 2.256301499426716e-06,
      "loss": 310.1289,
      "step": 40770
    },
    {
      "epoch": 0.7842986412285678,
      "grad_norm": 417.7440705776239,
      "learning_rate": 2.252441280759838e-06,
      "loss": 311.4468,
      "step": 40780
    },
    {
      "epoch": 0.7844909655643277,
      "grad_norm": 422.10711189813924,
      "learning_rate": 2.248583947930628e-06,
      "loss": 303.672,
      "step": 40790
    },
    {
      "epoch": 0.7846832899000875,
      "grad_norm": 392.77485211293487,
      "learning_rate": 2.2447295023758755e-06,
      "loss": 310.353,
      "step": 40800
    },
    {
      "epoch": 0.7848756142358473,
      "grad_norm": 398.8194065818253,
      "learning_rate": 2.2408779455313035e-06,
      "loss": 312.5769,
      "step": 40810
    },
    {
      "epoch": 0.7850679385716072,
      "grad_norm": 398.37553887579645,
      "learning_rate": 2.2370292788315505e-06,
      "loss": 314.5786,
      "step": 40820
    },
    {
      "epoch": 0.785260262907367,
      "grad_norm": 406.3424975986708,
      "learning_rate": 2.2331835037101825e-06,
      "loss": 304.1552,
      "step": 40830
    },
    {
      "epoch": 0.7854525872431268,
      "grad_norm": 408.10032093031566,
      "learning_rate": 2.2293406215996814e-06,
      "loss": 309.1903,
      "step": 40840
    },
    {
      "epoch": 0.7856449115788866,
      "grad_norm": 383.6693692367047,
      "learning_rate": 2.2255006339314667e-06,
      "loss": 312.4329,
      "step": 40850
    },
    {
      "epoch": 0.7858372359146465,
      "grad_norm": 413.7728525325508,
      "learning_rate": 2.2216635421358623e-06,
      "loss": 304.8318,
      "step": 40860
    },
    {
      "epoch": 0.7860295602504063,
      "grad_norm": 408.1206672763752,
      "learning_rate": 2.2178293476421276e-06,
      "loss": 307.9965,
      "step": 40870
    },
    {
      "epoch": 0.7862218845861662,
      "grad_norm": 513.5604920056612,
      "learning_rate": 2.213998051878431e-06,
      "loss": 309.4396,
      "step": 40880
    },
    {
      "epoch": 0.7864142089219259,
      "grad_norm": 429.4090592087335,
      "learning_rate": 2.2101696562718735e-06,
      "loss": 304.6626,
      "step": 40890
    },
    {
      "epoch": 0.7866065332576858,
      "grad_norm": 437.87832050621137,
      "learning_rate": 2.206344162248466e-06,
      "loss": 320.112,
      "step": 40900
    },
    {
      "epoch": 0.7867988575934456,
      "grad_norm": 397.4649874463456,
      "learning_rate": 2.2025215712331383e-06,
      "loss": 311.9292,
      "step": 40910
    },
    {
      "epoch": 0.7869911819292054,
      "grad_norm": 436.76947708713067,
      "learning_rate": 2.1987018846497487e-06,
      "loss": 307.7339,
      "step": 40920
    },
    {
      "epoch": 0.7871835062649652,
      "grad_norm": 394.1935927330882,
      "learning_rate": 2.194885103921064e-06,
      "loss": 312.8891,
      "step": 40930
    },
    {
      "epoch": 0.787375830600725,
      "grad_norm": 424.1501999635146,
      "learning_rate": 2.191071230468772e-06,
      "loss": 315.917,
      "step": 40940
    },
    {
      "epoch": 0.7875681549364849,
      "grad_norm": 431.6847064803192,
      "learning_rate": 2.1872602657134757e-06,
      "loss": 316.4769,
      "step": 40950
    },
    {
      "epoch": 0.7877604792722447,
      "grad_norm": 408.5040175877471,
      "learning_rate": 2.1834522110747014e-06,
      "loss": 312.625,
      "step": 40960
    },
    {
      "epoch": 0.7879528036080046,
      "grad_norm": 403.65461434670306,
      "learning_rate": 2.179647067970885e-06,
      "loss": 309.2533,
      "step": 40970
    },
    {
      "epoch": 0.7881451279437643,
      "grad_norm": 463.5459539500682,
      "learning_rate": 2.1758448378193743e-06,
      "loss": 318.881,
      "step": 40980
    },
    {
      "epoch": 0.7883374522795242,
      "grad_norm": 409.3855141503293,
      "learning_rate": 2.1720455220364443e-06,
      "loss": 310.3153,
      "step": 40990
    },
    {
      "epoch": 0.788529776615284,
      "grad_norm": 418.3573944071805,
      "learning_rate": 2.168249122037275e-06,
      "loss": 307.567,
      "step": 41000
    },
    {
      "epoch": 0.7887221009510439,
      "grad_norm": 413.41596894292144,
      "learning_rate": 2.1644556392359583e-06,
      "loss": 309.2228,
      "step": 41010
    },
    {
      "epoch": 0.7889144252868037,
      "grad_norm": 396.264812271915,
      "learning_rate": 2.160665075045508e-06,
      "loss": 300.119,
      "step": 41020
    },
    {
      "epoch": 0.7891067496225634,
      "grad_norm": 384.9750396955149,
      "learning_rate": 2.1568774308778494e-06,
      "loss": 314.1672,
      "step": 41030
    },
    {
      "epoch": 0.7892990739583233,
      "grad_norm": 432.08155574589415,
      "learning_rate": 2.1530927081438148e-06,
      "loss": 324.8616,
      "step": 41040
    },
    {
      "epoch": 0.7894913982940831,
      "grad_norm": 445.69000324077496,
      "learning_rate": 2.1493109082531473e-06,
      "loss": 300.0533,
      "step": 41050
    },
    {
      "epoch": 0.789683722629843,
      "grad_norm": 433.16143918617195,
      "learning_rate": 2.1455320326145103e-06,
      "loss": 304.4936,
      "step": 41060
    },
    {
      "epoch": 0.7898760469656028,
      "grad_norm": 402.76524694665744,
      "learning_rate": 2.141756082635471e-06,
      "loss": 308.7866,
      "step": 41070
    },
    {
      "epoch": 0.7900683713013626,
      "grad_norm": 419.62007628768663,
      "learning_rate": 2.137983059722507e-06,
      "loss": 301.3151,
      "step": 41080
    },
    {
      "epoch": 0.7902606956371224,
      "grad_norm": 413.6466504810756,
      "learning_rate": 2.1342129652810063e-06,
      "loss": 308.5127,
      "step": 41090
    },
    {
      "epoch": 0.7904530199728823,
      "grad_norm": 417.4033953263661,
      "learning_rate": 2.1304458007152694e-06,
      "loss": 305.7939,
      "step": 41100
    },
    {
      "epoch": 0.7906453443086421,
      "grad_norm": 420.63154716783026,
      "learning_rate": 2.1266815674285026e-06,
      "loss": 319.9178,
      "step": 41110
    },
    {
      "epoch": 0.790837668644402,
      "grad_norm": 426.74679328452436,
      "learning_rate": 2.1229202668228197e-06,
      "loss": 325.0511,
      "step": 41120
    },
    {
      "epoch": 0.7910299929801617,
      "grad_norm": 424.0738356287959,
      "learning_rate": 2.1191619002992405e-06,
      "loss": 300.2503,
      "step": 41130
    },
    {
      "epoch": 0.7912223173159215,
      "grad_norm": 460.0166816375219,
      "learning_rate": 2.1154064692577e-06,
      "loss": 308.9102,
      "step": 41140
    },
    {
      "epoch": 0.7914146416516814,
      "grad_norm": 401.33104977383886,
      "learning_rate": 2.111653975097029e-06,
      "loss": 306.6995,
      "step": 41150
    },
    {
      "epoch": 0.7916069659874412,
      "grad_norm": 426.6695098261237,
      "learning_rate": 2.1079044192149713e-06,
      "loss": 311.5517,
      "step": 41160
    },
    {
      "epoch": 0.7917992903232011,
      "grad_norm": 402.4697929884768,
      "learning_rate": 2.1041578030081777e-06,
      "loss": 309.8065,
      "step": 41170
    },
    {
      "epoch": 0.7919916146589608,
      "grad_norm": 398.7737487701014,
      "learning_rate": 2.100414127872198e-06,
      "loss": 303.9714,
      "step": 41180
    },
    {
      "epoch": 0.7921839389947207,
      "grad_norm": 399.56181410402843,
      "learning_rate": 2.0966733952014904e-06,
      "loss": 321.0781,
      "step": 41190
    },
    {
      "epoch": 0.7923762633304805,
      "grad_norm": 370.74344280725614,
      "learning_rate": 2.0929356063894125e-06,
      "loss": 310.3235,
      "step": 41200
    },
    {
      "epoch": 0.7925685876662404,
      "grad_norm": 433.22970589251315,
      "learning_rate": 2.089200762828234e-06,
      "loss": 312.2444,
      "step": 41210
    },
    {
      "epoch": 0.7927609120020002,
      "grad_norm": 388.8603080085678,
      "learning_rate": 2.0854688659091203e-06,
      "loss": 313.439,
      "step": 41220
    },
    {
      "epoch": 0.79295323633776,
      "grad_norm": 395.7810147279743,
      "learning_rate": 2.08173991702214e-06,
      "loss": 302.2224,
      "step": 41230
    },
    {
      "epoch": 0.7931455606735198,
      "grad_norm": 413.84484823595136,
      "learning_rate": 2.0780139175562675e-06,
      "loss": 311.3913,
      "step": 41240
    },
    {
      "epoch": 0.7933378850092796,
      "grad_norm": 414.3545464302663,
      "learning_rate": 2.0742908688993746e-06,
      "loss": 306.7947,
      "step": 41250
    },
    {
      "epoch": 0.7935302093450395,
      "grad_norm": 408.7211595559714,
      "learning_rate": 2.070570772438236e-06,
      "loss": 304.9645,
      "step": 41260
    },
    {
      "epoch": 0.7937225336807993,
      "grad_norm": 391.41070124901245,
      "learning_rate": 2.066853629558524e-06,
      "loss": 312.0633,
      "step": 41270
    },
    {
      "epoch": 0.7939148580165591,
      "grad_norm": 392.0298225013226,
      "learning_rate": 2.0631394416448157e-06,
      "loss": 300.5851,
      "step": 41280
    },
    {
      "epoch": 0.7941071823523189,
      "grad_norm": 394.33972644436676,
      "learning_rate": 2.059428210080583e-06,
      "loss": 302.7163,
      "step": 41290
    },
    {
      "epoch": 0.7942995066880788,
      "grad_norm": 464.0277523095237,
      "learning_rate": 2.0557199362482005e-06,
      "loss": 314.2422,
      "step": 41300
    },
    {
      "epoch": 0.7944918310238386,
      "grad_norm": 408.88500787597485,
      "learning_rate": 2.052014621528935e-06,
      "loss": 308.871,
      "step": 41310
    },
    {
      "epoch": 0.7946841553595985,
      "grad_norm": 393.6820137702138,
      "learning_rate": 2.048312267302961e-06,
      "loss": 318.9189,
      "step": 41320
    },
    {
      "epoch": 0.7948764796953582,
      "grad_norm": 389.13671333433956,
      "learning_rate": 2.044612874949341e-06,
      "loss": 304.0426,
      "step": 41330
    },
    {
      "epoch": 0.7950688040311181,
      "grad_norm": 402.4689379174277,
      "learning_rate": 2.040916445846034e-06,
      "loss": 318.7442,
      "step": 41340
    },
    {
      "epoch": 0.7952611283668779,
      "grad_norm": 534.4974650455013,
      "learning_rate": 2.037222981369905e-06,
      "loss": 305.6872,
      "step": 41350
    },
    {
      "epoch": 0.7954534527026377,
      "grad_norm": 403.443000767358,
      "learning_rate": 2.033532482896707e-06,
      "loss": 307.3848,
      "step": 41360
    },
    {
      "epoch": 0.7956457770383976,
      "grad_norm": 445.155693695892,
      "learning_rate": 2.0298449518010875e-06,
      "loss": 308.726,
      "step": 41370
    },
    {
      "epoch": 0.7958381013741573,
      "grad_norm": 427.4768329158586,
      "learning_rate": 2.0261603894565897e-06,
      "loss": 309.1523,
      "step": 41380
    },
    {
      "epoch": 0.7960304257099172,
      "grad_norm": 411.55110561014817,
      "learning_rate": 2.0224787972356574e-06,
      "loss": 312.8725,
      "step": 41390
    },
    {
      "epoch": 0.796222750045677,
      "grad_norm": 400.2907779344978,
      "learning_rate": 2.0188001765096198e-06,
      "loss": 313.9252,
      "step": 41400
    },
    {
      "epoch": 0.7964150743814369,
      "grad_norm": 412.6303180111913,
      "learning_rate": 2.0151245286486998e-06,
      "loss": 310.0146,
      "step": 41410
    },
    {
      "epoch": 0.7966073987171967,
      "grad_norm": 400.0735837232573,
      "learning_rate": 2.011451855022021e-06,
      "loss": 313.221,
      "step": 41420
    },
    {
      "epoch": 0.7967997230529565,
      "grad_norm": 404.70054361888714,
      "learning_rate": 2.0077821569975885e-06,
      "loss": 311.697,
      "step": 41430
    },
    {
      "epoch": 0.7969920473887163,
      "grad_norm": 430.3644867725602,
      "learning_rate": 2.0041154359423087e-06,
      "loss": 317.9353,
      "step": 41440
    },
    {
      "epoch": 0.7971843717244762,
      "grad_norm": 428.0505087686813,
      "learning_rate": 2.000451693221971e-06,
      "loss": 317.9721,
      "step": 41450
    },
    {
      "epoch": 0.797376696060236,
      "grad_norm": 410.3627598673023,
      "learning_rate": 1.9967909302012635e-06,
      "loss": 308.9829,
      "step": 41460
    },
    {
      "epoch": 0.7975690203959958,
      "grad_norm": 406.2366473799651,
      "learning_rate": 1.9931331482437553e-06,
      "loss": 309.653,
      "step": 41470
    },
    {
      "epoch": 0.7977613447317556,
      "grad_norm": 385.7246857268013,
      "learning_rate": 1.989478348711913e-06,
      "loss": 307.2674,
      "step": 41480
    },
    {
      "epoch": 0.7979536690675154,
      "grad_norm": 404.9555301471717,
      "learning_rate": 1.9858265329670844e-06,
      "loss": 307.7042,
      "step": 41490
    },
    {
      "epoch": 0.7981459934032753,
      "grad_norm": 409.2744055140272,
      "learning_rate": 1.9821777023695178e-06,
      "loss": 308.5514,
      "step": 41500
    },
    {
      "epoch": 0.7983383177390351,
      "grad_norm": 411.3852387421349,
      "learning_rate": 1.9785318582783375e-06,
      "loss": 308.0267,
      "step": 41510
    },
    {
      "epoch": 0.798530642074795,
      "grad_norm": 384.05684103370396,
      "learning_rate": 1.9748890020515577e-06,
      "loss": 312.3036,
      "step": 41520
    },
    {
      "epoch": 0.7987229664105547,
      "grad_norm": 385.8838582318483,
      "learning_rate": 1.9712491350460895e-06,
      "loss": 304.0441,
      "step": 41530
    },
    {
      "epoch": 0.7989152907463146,
      "grad_norm": 394.6691368185795,
      "learning_rate": 1.967612258617718e-06,
      "loss": 304.6756,
      "step": 41540
    },
    {
      "epoch": 0.7991076150820744,
      "grad_norm": 429.2317074748059,
      "learning_rate": 1.9639783741211218e-06,
      "loss": 311.8391,
      "step": 41550
    },
    {
      "epoch": 0.7992999394178343,
      "grad_norm": 397.7645364719865,
      "learning_rate": 1.960347482909859e-06,
      "loss": 309.7134,
      "step": 41560
    },
    {
      "epoch": 0.7994922637535941,
      "grad_norm": 452.2229297317847,
      "learning_rate": 1.956719586336382e-06,
      "loss": 316.8328,
      "step": 41570
    },
    {
      "epoch": 0.7996845880893538,
      "grad_norm": 410.56945513535027,
      "learning_rate": 1.953094685752017e-06,
      "loss": 310.9083,
      "step": 41580
    },
    {
      "epoch": 0.7998769124251137,
      "grad_norm": 378.70035488693077,
      "learning_rate": 1.949472782506984e-06,
      "loss": 313.304,
      "step": 41590
    },
    {
      "epoch": 0.8000692367608735,
      "grad_norm": 413.5076669215031,
      "learning_rate": 1.945853877950382e-06,
      "loss": 325.3368,
      "step": 41600
    },
    {
      "epoch": 0.8002615610966334,
      "grad_norm": 431.09325237040275,
      "learning_rate": 1.942237973430192e-06,
      "loss": 317.6258,
      "step": 41610
    },
    {
      "epoch": 0.8004538854323932,
      "grad_norm": 438.3780318282062,
      "learning_rate": 1.9386250702932784e-06,
      "loss": 314.4511,
      "step": 41620
    },
    {
      "epoch": 0.800646209768153,
      "grad_norm": 407.703498976155,
      "learning_rate": 1.9350151698853857e-06,
      "loss": 314.5751,
      "step": 41630
    },
    {
      "epoch": 0.8008385341039128,
      "grad_norm": 402.9200771757179,
      "learning_rate": 1.9314082735511475e-06,
      "loss": 309.926,
      "step": 41640
    },
    {
      "epoch": 0.8010308584396727,
      "grad_norm": 419.98269993257713,
      "learning_rate": 1.92780438263407e-06,
      "loss": 305.309,
      "step": 41650
    },
    {
      "epoch": 0.8012231827754325,
      "grad_norm": 387.0709198264562,
      "learning_rate": 1.9242034984765436e-06,
      "loss": 312.4188,
      "step": 41660
    },
    {
      "epoch": 0.8014155071111924,
      "grad_norm": 395.1476476816383,
      "learning_rate": 1.9206056224198346e-06,
      "loss": 305.0891,
      "step": 41670
    },
    {
      "epoch": 0.8016078314469521,
      "grad_norm": 431.28337069401505,
      "learning_rate": 1.9170107558040983e-06,
      "loss": 315.3144,
      "step": 41680
    },
    {
      "epoch": 0.801800155782712,
      "grad_norm": 408.65066544976474,
      "learning_rate": 1.9134188999683613e-06,
      "loss": 302.9139,
      "step": 41690
    },
    {
      "epoch": 0.8019924801184718,
      "grad_norm": 415.65819937197,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 310.0961,
      "step": 41700
    },
    {
      "epoch": 0.8021848044542316,
      "grad_norm": 395.66579858118723,
      "learning_rate": 1.9062442259873847e-06,
      "loss": 313.4762,
      "step": 41710
    },
    {
      "epoch": 0.8023771287899915,
      "grad_norm": 421.56141913034077,
      "learning_rate": 1.9026614105145935e-06,
      "loss": 321.0154,
      "step": 41720
    },
    {
      "epoch": 0.8025694531257512,
      "grad_norm": 405.29192833926294,
      "learning_rate": 1.8990816111666976e-06,
      "loss": 309.3742,
      "step": 41730
    },
    {
      "epoch": 0.8027617774615111,
      "grad_norm": 416.2764926481287,
      "learning_rate": 1.8955048292771083e-06,
      "loss": 307.1205,
      "step": 41740
    },
    {
      "epoch": 0.8029541017972709,
      "grad_norm": 408.4451145932948,
      "learning_rate": 1.891931066178122e-06,
      "loss": 299.2861,
      "step": 41750
    },
    {
      "epoch": 0.8031464261330308,
      "grad_norm": 385.6296985539054,
      "learning_rate": 1.888360323200904e-06,
      "loss": 298.5853,
      "step": 41760
    },
    {
      "epoch": 0.8033387504687906,
      "grad_norm": 390.6209917945643,
      "learning_rate": 1.8847926016754947e-06,
      "loss": 307.6281,
      "step": 41770
    },
    {
      "epoch": 0.8035310748045504,
      "grad_norm": 414.7057062991525,
      "learning_rate": 1.8812279029308177e-06,
      "loss": 307.143,
      "step": 41780
    },
    {
      "epoch": 0.8037233991403102,
      "grad_norm": 412.90339155146097,
      "learning_rate": 1.87766622829466e-06,
      "loss": 311.2535,
      "step": 41790
    },
    {
      "epoch": 0.8039157234760701,
      "grad_norm": 413.0134697210266,
      "learning_rate": 1.874107579093688e-06,
      "loss": 324.2331,
      "step": 41800
    },
    {
      "epoch": 0.8041080478118299,
      "grad_norm": 410.15688941487105,
      "learning_rate": 1.870551956653437e-06,
      "loss": 317.0837,
      "step": 41810
    },
    {
      "epoch": 0.8043003721475896,
      "grad_norm": 405.4574073762015,
      "learning_rate": 1.8669993622983217e-06,
      "loss": 316.0981,
      "step": 41820
    },
    {
      "epoch": 0.8044926964833495,
      "grad_norm": 404.6597887685546,
      "learning_rate": 1.863449797351624e-06,
      "loss": 309.9964,
      "step": 41830
    },
    {
      "epoch": 0.8046850208191093,
      "grad_norm": 403.83109312868817,
      "learning_rate": 1.8599032631354963e-06,
      "loss": 313.5171,
      "step": 41840
    },
    {
      "epoch": 0.8048773451548692,
      "grad_norm": 439.15257366140855,
      "learning_rate": 1.8563597609709626e-06,
      "loss": 321.2541,
      "step": 41850
    },
    {
      "epoch": 0.805069669490629,
      "grad_norm": 408.9617193632844,
      "learning_rate": 1.852819292177922e-06,
      "loss": 312.5056,
      "step": 41860
    },
    {
      "epoch": 0.8052619938263889,
      "grad_norm": 444.4408849629124,
      "learning_rate": 1.8492818580751414e-06,
      "loss": 323.8703,
      "step": 41870
    },
    {
      "epoch": 0.8054543181621486,
      "grad_norm": 428.96391093624396,
      "learning_rate": 1.8457474599802527e-06,
      "loss": 304.4847,
      "step": 41880
    },
    {
      "epoch": 0.8056466424979085,
      "grad_norm": 421.5004817056809,
      "learning_rate": 1.842216099209767e-06,
      "loss": 316.7431,
      "step": 41890
    },
    {
      "epoch": 0.8058389668336683,
      "grad_norm": 388.40606039820835,
      "learning_rate": 1.8386877770790524e-06,
      "loss": 307.68,
      "step": 41900
    },
    {
      "epoch": 0.8060312911694282,
      "grad_norm": 453.65573103567783,
      "learning_rate": 1.8351624949023539e-06,
      "loss": 319.7826,
      "step": 41910
    },
    {
      "epoch": 0.806223615505188,
      "grad_norm": 413.60801458361624,
      "learning_rate": 1.8316402539927757e-06,
      "loss": 312.6104,
      "step": 41920
    },
    {
      "epoch": 0.8064159398409477,
      "grad_norm": 419.5876485086959,
      "learning_rate": 1.8281210556623007e-06,
      "loss": 304.2615,
      "step": 41930
    },
    {
      "epoch": 0.8066082641767076,
      "grad_norm": 413.97681046550025,
      "learning_rate": 1.8246049012217693e-06,
      "loss": 305.5498,
      "step": 41940
    },
    {
      "epoch": 0.8068005885124674,
      "grad_norm": 391.72215458436597,
      "learning_rate": 1.8210917919808891e-06,
      "loss": 315.7686,
      "step": 41950
    },
    {
      "epoch": 0.8069929128482273,
      "grad_norm": 401.94078560143623,
      "learning_rate": 1.817581729248239e-06,
      "loss": 309.178,
      "step": 41960
    },
    {
      "epoch": 0.807185237183987,
      "grad_norm": 463.05005440890915,
      "learning_rate": 1.8140747143312588e-06,
      "loss": 309.883,
      "step": 41970
    },
    {
      "epoch": 0.8073775615197469,
      "grad_norm": 433.2230208778572,
      "learning_rate": 1.8105707485362511e-06,
      "loss": 302.4812,
      "step": 41980
    },
    {
      "epoch": 0.8075698858555067,
      "grad_norm": 383.7629326672162,
      "learning_rate": 1.8070698331683844e-06,
      "loss": 308.8602,
      "step": 41990
    },
    {
      "epoch": 0.8077622101912666,
      "grad_norm": 401.09806979534505,
      "learning_rate": 1.8035719695316955e-06,
      "loss": 319.9707,
      "step": 42000
    },
    {
      "epoch": 0.8079545345270264,
      "grad_norm": 394.22092803620967,
      "learning_rate": 1.800077158929081e-06,
      "loss": 311.2034,
      "step": 42010
    },
    {
      "epoch": 0.8081468588627863,
      "grad_norm": 409.7361859692149,
      "learning_rate": 1.7965854026622953e-06,
      "loss": 306.2973,
      "step": 42020
    },
    {
      "epoch": 0.808339183198546,
      "grad_norm": 439.4300326694273,
      "learning_rate": 1.7930967020319667e-06,
      "loss": 314.2226,
      "step": 42030
    },
    {
      "epoch": 0.8085315075343058,
      "grad_norm": 402.80174441963084,
      "learning_rate": 1.7896110583375747e-06,
      "loss": 299.4835,
      "step": 42040
    },
    {
      "epoch": 0.8087238318700657,
      "grad_norm": 363.44072180640245,
      "learning_rate": 1.7861284728774652e-06,
      "loss": 309.756,
      "step": 42050
    },
    {
      "epoch": 0.8089161562058255,
      "grad_norm": 407.1417539220159,
      "learning_rate": 1.7826489469488395e-06,
      "loss": 305.5184,
      "step": 42060
    },
    {
      "epoch": 0.8091084805415854,
      "grad_norm": 423.7589204811408,
      "learning_rate": 1.7791724818477708e-06,
      "loss": 307.9455,
      "step": 42070
    },
    {
      "epoch": 0.8093008048773451,
      "grad_norm": 414.5921398814254,
      "learning_rate": 1.7756990788691797e-06,
      "loss": 303.6039,
      "step": 42080
    },
    {
      "epoch": 0.809493129213105,
      "grad_norm": 388.42061646518516,
      "learning_rate": 1.772228739306854e-06,
      "loss": 309.3741,
      "step": 42090
    },
    {
      "epoch": 0.8096854535488648,
      "grad_norm": 404.0016850680053,
      "learning_rate": 1.7687614644534333e-06,
      "loss": 304.8437,
      "step": 42100
    },
    {
      "epoch": 0.8098777778846247,
      "grad_norm": 442.6996275222591,
      "learning_rate": 1.7652972556004267e-06,
      "loss": 317.3699,
      "step": 42110
    },
    {
      "epoch": 0.8100701022203844,
      "grad_norm": 441.3560619649883,
      "learning_rate": 1.7618361140381922e-06,
      "loss": 322.6206,
      "step": 42120
    },
    {
      "epoch": 0.8102624265561443,
      "grad_norm": 420.88822753152056,
      "learning_rate": 1.7583780410559449e-06,
      "loss": 312.3599,
      "step": 42130
    },
    {
      "epoch": 0.8104547508919041,
      "grad_norm": 404.81850041085477,
      "learning_rate": 1.7549230379417636e-06,
      "loss": 312.039,
      "step": 42140
    },
    {
      "epoch": 0.8106470752276639,
      "grad_norm": 404.90281054148977,
      "learning_rate": 1.7514711059825773e-06,
      "loss": 304.1775,
      "step": 42150
    },
    {
      "epoch": 0.8108393995634238,
      "grad_norm": 396.2542202053286,
      "learning_rate": 1.7480222464641783e-06,
      "loss": 323.5015,
      "step": 42160
    },
    {
      "epoch": 0.8110317238991835,
      "grad_norm": 423.1607763548634,
      "learning_rate": 1.7445764606712024e-06,
      "loss": 305.1296,
      "step": 42170
    },
    {
      "epoch": 0.8112240482349434,
      "grad_norm": 418.2332118298525,
      "learning_rate": 1.7411337498871561e-06,
      "loss": 307.3521,
      "step": 42180
    },
    {
      "epoch": 0.8114163725707032,
      "grad_norm": 428.055348548745,
      "learning_rate": 1.737694115394387e-06,
      "loss": 314.606,
      "step": 42190
    },
    {
      "epoch": 0.8116086969064631,
      "grad_norm": 410.03672431689245,
      "learning_rate": 1.7342575584741018e-06,
      "loss": 308.3802,
      "step": 42200
    },
    {
      "epoch": 0.8118010212422229,
      "grad_norm": 426.633963182364,
      "learning_rate": 1.7308240804063648e-06,
      "loss": 308.3395,
      "step": 42210
    },
    {
      "epoch": 0.8119933455779828,
      "grad_norm": 407.0921867161722,
      "learning_rate": 1.7273936824700888e-06,
      "loss": 315.7191,
      "step": 42220
    },
    {
      "epoch": 0.8121856699137425,
      "grad_norm": 374.81437917432567,
      "learning_rate": 1.7239663659430384e-06,
      "loss": 310.0117,
      "step": 42230
    },
    {
      "epoch": 0.8123779942495024,
      "grad_norm": 409.10388015131576,
      "learning_rate": 1.7205421321018312e-06,
      "loss": 317.0021,
      "step": 42240
    },
    {
      "epoch": 0.8125703185852622,
      "grad_norm": 382.54398085650195,
      "learning_rate": 1.7171209822219427e-06,
      "loss": 319.6945,
      "step": 42250
    },
    {
      "epoch": 0.812762642921022,
      "grad_norm": 392.3959586105822,
      "learning_rate": 1.713702917577692e-06,
      "loss": 312.9114,
      "step": 42260
    },
    {
      "epoch": 0.8129549672567818,
      "grad_norm": 406.31131552509754,
      "learning_rate": 1.71028793944225e-06,
      "loss": 307.5117,
      "step": 42270
    },
    {
      "epoch": 0.8131472915925416,
      "grad_norm": 401.0170245517488,
      "learning_rate": 1.7068760490876425e-06,
      "loss": 308.5841,
      "step": 42280
    },
    {
      "epoch": 0.8133396159283015,
      "grad_norm": 394.7075119154168,
      "learning_rate": 1.7034672477847402e-06,
      "loss": 310.3857,
      "step": 42290
    },
    {
      "epoch": 0.8135319402640613,
      "grad_norm": 385.3946486335663,
      "learning_rate": 1.700061536803268e-06,
      "loss": 306.9722,
      "step": 42300
    },
    {
      "epoch": 0.8137242645998212,
      "grad_norm": 448.8436897946402,
      "learning_rate": 1.696658917411793e-06,
      "loss": 307.0883,
      "step": 42310
    },
    {
      "epoch": 0.8139165889355809,
      "grad_norm": 402.0297537593429,
      "learning_rate": 1.6932593908777394e-06,
      "loss": 312.4774,
      "step": 42320
    },
    {
      "epoch": 0.8141089132713408,
      "grad_norm": 401.9973192021659,
      "learning_rate": 1.689862958467372e-06,
      "loss": 306.6643,
      "step": 42330
    },
    {
      "epoch": 0.8143012376071006,
      "grad_norm": 412.21038921291057,
      "learning_rate": 1.6864696214458065e-06,
      "loss": 309.7551,
      "step": 42340
    },
    {
      "epoch": 0.8144935619428605,
      "grad_norm": 392.7780980269767,
      "learning_rate": 1.683079381077003e-06,
      "loss": 314.4754,
      "step": 42350
    },
    {
      "epoch": 0.8146858862786203,
      "grad_norm": 409.5128937394212,
      "learning_rate": 1.6796922386237724e-06,
      "loss": 318.146,
      "step": 42360
    },
    {
      "epoch": 0.81487821061438,
      "grad_norm": 412.30147847210947,
      "learning_rate": 1.67630819534777e-06,
      "loss": 313.1921,
      "step": 42370
    },
    {
      "epoch": 0.8150705349501399,
      "grad_norm": 440.11693125471027,
      "learning_rate": 1.6729272525094908e-06,
      "loss": 311.6013,
      "step": 42380
    },
    {
      "epoch": 0.8152628592858997,
      "grad_norm": 431.98961622993335,
      "learning_rate": 1.6695494113682874e-06,
      "loss": 314.7857,
      "step": 42390
    },
    {
      "epoch": 0.8154551836216596,
      "grad_norm": 411.42898088462056,
      "learning_rate": 1.6661746731823458e-06,
      "loss": 306.7567,
      "step": 42400
    },
    {
      "epoch": 0.8156475079574194,
      "grad_norm": 410.6316817635866,
      "learning_rate": 1.6628030392087001e-06,
      "loss": 311.9825,
      "step": 42410
    },
    {
      "epoch": 0.8158398322931792,
      "grad_norm": 409.6194112839262,
      "learning_rate": 1.6594345107032273e-06,
      "loss": 317.3002,
      "step": 42420
    },
    {
      "epoch": 0.816032156628939,
      "grad_norm": 400.9852784179425,
      "learning_rate": 1.6560690889206499e-06,
      "loss": 313.499,
      "step": 42430
    },
    {
      "epoch": 0.8162244809646989,
      "grad_norm": 401.8736664051403,
      "learning_rate": 1.6527067751145354e-06,
      "loss": 301.383,
      "step": 42440
    },
    {
      "epoch": 0.8164168053004587,
      "grad_norm": 398.26230356371184,
      "learning_rate": 1.6493475705372863e-06,
      "loss": 309.8659,
      "step": 42450
    },
    {
      "epoch": 0.8166091296362186,
      "grad_norm": 415.14565715216116,
      "learning_rate": 1.6459914764401497e-06,
      "loss": 317.2495,
      "step": 42460
    },
    {
      "epoch": 0.8168014539719783,
      "grad_norm": 432.73069664734874,
      "learning_rate": 1.64263849407322e-06,
      "loss": 310.5572,
      "step": 42470
    },
    {
      "epoch": 0.8169937783077381,
      "grad_norm": 379.21529022894157,
      "learning_rate": 1.6392886246854234e-06,
      "loss": 296.3163,
      "step": 42480
    },
    {
      "epoch": 0.817186102643498,
      "grad_norm": 416.17721523460403,
      "learning_rate": 1.6359418695245311e-06,
      "loss": 305.976,
      "step": 42490
    },
    {
      "epoch": 0.8173784269792578,
      "grad_norm": 418.29091803976485,
      "learning_rate": 1.632598229837158e-06,
      "loss": 310.422,
      "step": 42500
    },
    {
      "epoch": 0.8175707513150177,
      "grad_norm": 399.3470917060732,
      "learning_rate": 1.629257706868751e-06,
      "loss": 307.9481,
      "step": 42510
    },
    {
      "epoch": 0.8177630756507774,
      "grad_norm": 379.01930220379717,
      "learning_rate": 1.6259203018636016e-06,
      "loss": 306.1668,
      "step": 42520
    },
    {
      "epoch": 0.8179553999865373,
      "grad_norm": 425.2605228905251,
      "learning_rate": 1.6225860160648343e-06,
      "loss": 318.8865,
      "step": 42530
    },
    {
      "epoch": 0.8181477243222971,
      "grad_norm": 396.26121774724527,
      "learning_rate": 1.6192548507144213e-06,
      "loss": 313.6861,
      "step": 42540
    },
    {
      "epoch": 0.818340048658057,
      "grad_norm": 388.96172309999923,
      "learning_rate": 1.6159268070531642e-06,
      "loss": 303.8297,
      "step": 42550
    },
    {
      "epoch": 0.8185323729938168,
      "grad_norm": 445.15160080702293,
      "learning_rate": 1.6126018863207005e-06,
      "loss": 318.8677,
      "step": 42560
    },
    {
      "epoch": 0.8187246973295766,
      "grad_norm": 409.65713462434377,
      "learning_rate": 1.609280089755515e-06,
      "loss": 312.0928,
      "step": 42570
    },
    {
      "epoch": 0.8189170216653364,
      "grad_norm": 442.47644572618,
      "learning_rate": 1.6059614185949157e-06,
      "loss": 312.5717,
      "step": 42580
    },
    {
      "epoch": 0.8191093460010962,
      "grad_norm": 394.5766013685885,
      "learning_rate": 1.6026458740750584e-06,
      "loss": 297.5948,
      "step": 42590
    },
    {
      "epoch": 0.8193016703368561,
      "grad_norm": 387.29266393085965,
      "learning_rate": 1.5993334574309238e-06,
      "loss": 314.3256,
      "step": 42600
    },
    {
      "epoch": 0.8194939946726159,
      "grad_norm": 398.6022759364853,
      "learning_rate": 1.5960241698963374e-06,
      "loss": 302.7601,
      "step": 42610
    },
    {
      "epoch": 0.8196863190083757,
      "grad_norm": 412.5073786107496,
      "learning_rate": 1.592718012703951e-06,
      "loss": 309.0457,
      "step": 42620
    },
    {
      "epoch": 0.8198786433441355,
      "grad_norm": 402.668192180866,
      "learning_rate": 1.589414987085255e-06,
      "loss": 298.2543,
      "step": 42630
    },
    {
      "epoch": 0.8200709676798954,
      "grad_norm": 405.7025109899609,
      "learning_rate": 1.5861150942705672e-06,
      "loss": 306.7637,
      "step": 42640
    },
    {
      "epoch": 0.8202632920156552,
      "grad_norm": 382.3442127885861,
      "learning_rate": 1.5828183354890504e-06,
      "loss": 299.5104,
      "step": 42650
    },
    {
      "epoch": 0.8204556163514151,
      "grad_norm": 379.2150502137799,
      "learning_rate": 1.5795247119686885e-06,
      "loss": 299.6946,
      "step": 42660
    },
    {
      "epoch": 0.8206479406871748,
      "grad_norm": 435.7394531715828,
      "learning_rate": 1.5762342249363006e-06,
      "loss": 316.0116,
      "step": 42670
    },
    {
      "epoch": 0.8208402650229347,
      "grad_norm": 385.3948038543622,
      "learning_rate": 1.5729468756175426e-06,
      "loss": 299.1303,
      "step": 42680
    },
    {
      "epoch": 0.8210325893586945,
      "grad_norm": 411.6885845379898,
      "learning_rate": 1.5696626652368973e-06,
      "loss": 305.3687,
      "step": 42690
    },
    {
      "epoch": 0.8212249136944543,
      "grad_norm": 440.36084463891024,
      "learning_rate": 1.5663815950176742e-06,
      "loss": 311.0056,
      "step": 42700
    },
    {
      "epoch": 0.8214172380302142,
      "grad_norm": 410.38556544557576,
      "learning_rate": 1.5631036661820232e-06,
      "loss": 316.9325,
      "step": 42710
    },
    {
      "epoch": 0.8216095623659739,
      "grad_norm": 393.16886637814446,
      "learning_rate": 1.5598288799509153e-06,
      "loss": 304.1306,
      "step": 42720
    },
    {
      "epoch": 0.8218018867017338,
      "grad_norm": 420.4057562437857,
      "learning_rate": 1.5565572375441573e-06,
      "loss": 308.2313,
      "step": 42730
    },
    {
      "epoch": 0.8219942110374936,
      "grad_norm": 431.61374927043823,
      "learning_rate": 1.5532887401803787e-06,
      "loss": 309.4908,
      "step": 42740
    },
    {
      "epoch": 0.8221865353732535,
      "grad_norm": 405.95297521991796,
      "learning_rate": 1.5500233890770434e-06,
      "loss": 309.826,
      "step": 42750
    },
    {
      "epoch": 0.8223788597090133,
      "grad_norm": 404.76047493259057,
      "learning_rate": 1.5467611854504406e-06,
      "loss": 301.0147,
      "step": 42760
    },
    {
      "epoch": 0.8225711840447731,
      "grad_norm": 396.45738573903355,
      "learning_rate": 1.5435021305156862e-06,
      "loss": 309.4137,
      "step": 42770
    },
    {
      "epoch": 0.8227635083805329,
      "grad_norm": 401.13026869766855,
      "learning_rate": 1.5402462254867222e-06,
      "loss": 316.6085,
      "step": 42780
    },
    {
      "epoch": 0.8229558327162928,
      "grad_norm": 434.444044410929,
      "learning_rate": 1.5369934715763235e-06,
      "loss": 309.6957,
      "step": 42790
    },
    {
      "epoch": 0.8231481570520526,
      "grad_norm": 406.8286170554677,
      "learning_rate": 1.533743869996086e-06,
      "loss": 304.6467,
      "step": 42800
    },
    {
      "epoch": 0.8233404813878125,
      "grad_norm": 413.56825052349353,
      "learning_rate": 1.5304974219564318e-06,
      "loss": 309.2877,
      "step": 42810
    },
    {
      "epoch": 0.8235328057235722,
      "grad_norm": 411.52541558689006,
      "learning_rate": 1.5272541286666075e-06,
      "loss": 303.7095,
      "step": 42820
    },
    {
      "epoch": 0.823725130059332,
      "grad_norm": 428.39787018305475,
      "learning_rate": 1.5240139913346906e-06,
      "loss": 310.0144,
      "step": 42830
    },
    {
      "epoch": 0.8239174543950919,
      "grad_norm": 441.49099139823363,
      "learning_rate": 1.5207770111675735e-06,
      "loss": 315.4356,
      "step": 42840
    },
    {
      "epoch": 0.8241097787308517,
      "grad_norm": 371.96768192793553,
      "learning_rate": 1.5175431893709836e-06,
      "loss": 310.1807,
      "step": 42850
    },
    {
      "epoch": 0.8243021030666116,
      "grad_norm": 407.47598290847037,
      "learning_rate": 1.5143125271494607e-06,
      "loss": 304.953,
      "step": 42860
    },
    {
      "epoch": 0.8244944274023713,
      "grad_norm": 401.4889777858116,
      "learning_rate": 1.511085025706378e-06,
      "loss": 306.8126,
      "step": 42870
    },
    {
      "epoch": 0.8246867517381312,
      "grad_norm": 384.51772526809947,
      "learning_rate": 1.5078606862439248e-06,
      "loss": 304.6681,
      "step": 42880
    },
    {
      "epoch": 0.824879076073891,
      "grad_norm": 414.39931039407116,
      "learning_rate": 1.5046395099631106e-06,
      "loss": 309.4831,
      "step": 42890
    },
    {
      "epoch": 0.8250714004096509,
      "grad_norm": 432.4809190805487,
      "learning_rate": 1.5014214980637754e-06,
      "loss": 307.1895,
      "step": 42900
    },
    {
      "epoch": 0.8252637247454107,
      "grad_norm": 421.95882028972665,
      "learning_rate": 1.4982066517445748e-06,
      "loss": 304.3493,
      "step": 42910
    },
    {
      "epoch": 0.8254560490811705,
      "grad_norm": 432.7778176649635,
      "learning_rate": 1.4949949722029811e-06,
      "loss": 307.8679,
      "step": 42920
    },
    {
      "epoch": 0.8256483734169303,
      "grad_norm": 435.7686026497326,
      "learning_rate": 1.4917864606352983e-06,
      "loss": 321.3161,
      "step": 42930
    },
    {
      "epoch": 0.8258406977526901,
      "grad_norm": 403.7324293283603,
      "learning_rate": 1.4885811182366406e-06,
      "loss": 315.8647,
      "step": 42940
    },
    {
      "epoch": 0.82603302208845,
      "grad_norm": 392.8614771109709,
      "learning_rate": 1.485378946200946e-06,
      "loss": 308.2161,
      "step": 42950
    },
    {
      "epoch": 0.8262253464242098,
      "grad_norm": 377.5813441608248,
      "learning_rate": 1.4821799457209684e-06,
      "loss": 303.9158,
      "step": 42960
    },
    {
      "epoch": 0.8264176707599696,
      "grad_norm": 398.5752891718774,
      "learning_rate": 1.478984117988287e-06,
      "loss": 309.6031,
      "step": 42970
    },
    {
      "epoch": 0.8266099950957294,
      "grad_norm": 408.2175427744864,
      "learning_rate": 1.4757914641932924e-06,
      "loss": 304.9219,
      "step": 42980
    },
    {
      "epoch": 0.8268023194314893,
      "grad_norm": 377.3102620453641,
      "learning_rate": 1.4726019855251928e-06,
      "loss": 310.5686,
      "step": 42990
    },
    {
      "epoch": 0.8269946437672491,
      "grad_norm": 372.7674117475207,
      "learning_rate": 1.4694156831720185e-06,
      "loss": 307.1472,
      "step": 43000
    },
    {
      "epoch": 0.827186968103009,
      "grad_norm": 406.9858561650125,
      "learning_rate": 1.4662325583206172e-06,
      "loss": 315.3097,
      "step": 43010
    },
    {
      "epoch": 0.8273792924387687,
      "grad_norm": 407.5473076985246,
      "learning_rate": 1.463052612156649e-06,
      "loss": 314.7602,
      "step": 43020
    },
    {
      "epoch": 0.8275716167745286,
      "grad_norm": 460.82731087343296,
      "learning_rate": 1.4598758458645878e-06,
      "loss": 313.4908,
      "step": 43030
    },
    {
      "epoch": 0.8277639411102884,
      "grad_norm": 424.11726780910897,
      "learning_rate": 1.4567022606277314e-06,
      "loss": 308.193,
      "step": 43040
    },
    {
      "epoch": 0.8279562654460482,
      "grad_norm": 408.77580804971666,
      "learning_rate": 1.4535318576281854e-06,
      "loss": 302.9278,
      "step": 43050
    },
    {
      "epoch": 0.828148589781808,
      "grad_norm": 405.97204339467817,
      "learning_rate": 1.4503646380468729e-06,
      "loss": 302.8712,
      "step": 43060
    },
    {
      "epoch": 0.8283409141175678,
      "grad_norm": 446.90529183618355,
      "learning_rate": 1.4472006030635288e-06,
      "loss": 311.4265,
      "step": 43070
    },
    {
      "epoch": 0.8285332384533277,
      "grad_norm": 435.4659847610761,
      "learning_rate": 1.4440397538567086e-06,
      "loss": 313.404,
      "step": 43080
    },
    {
      "epoch": 0.8287255627890875,
      "grad_norm": 379.9167230649679,
      "learning_rate": 1.4408820916037735e-06,
      "loss": 306.1127,
      "step": 43090
    },
    {
      "epoch": 0.8289178871248474,
      "grad_norm": 435.5973177012765,
      "learning_rate": 1.4377276174808984e-06,
      "loss": 304.7473,
      "step": 43100
    },
    {
      "epoch": 0.8291102114606071,
      "grad_norm": 438.0031909485522,
      "learning_rate": 1.4345763326630768e-06,
      "loss": 308.3987,
      "step": 43110
    },
    {
      "epoch": 0.829302535796367,
      "grad_norm": 404.40719707800486,
      "learning_rate": 1.4314282383241097e-06,
      "loss": 304.1587,
      "step": 43120
    },
    {
      "epoch": 0.8294948601321268,
      "grad_norm": 403.00195421340226,
      "learning_rate": 1.4282833356366066e-06,
      "loss": 306.0333,
      "step": 43130
    },
    {
      "epoch": 0.8296871844678867,
      "grad_norm": 381.43647340158674,
      "learning_rate": 1.4251416257719962e-06,
      "loss": 303.9909,
      "step": 43140
    },
    {
      "epoch": 0.8298795088036465,
      "grad_norm": 441.02615502985316,
      "learning_rate": 1.4220031099005094e-06,
      "loss": 308.5014,
      "step": 43150
    },
    {
      "epoch": 0.8300718331394062,
      "grad_norm": 379.29968437602855,
      "learning_rate": 1.4188677891911961e-06,
      "loss": 297.6056,
      "step": 43160
    },
    {
      "epoch": 0.8302641574751661,
      "grad_norm": 398.65431769775125,
      "learning_rate": 1.4157356648119103e-06,
      "loss": 303.8602,
      "step": 43170
    },
    {
      "epoch": 0.8304564818109259,
      "grad_norm": 419.8705696419292,
      "learning_rate": 1.412606737929313e-06,
      "loss": 311.4084,
      "step": 43180
    },
    {
      "epoch": 0.8306488061466858,
      "grad_norm": 387.25796585656343,
      "learning_rate": 1.4094810097088817e-06,
      "loss": 309.5396,
      "step": 43190
    },
    {
      "epoch": 0.8308411304824456,
      "grad_norm": 395.52609984033694,
      "learning_rate": 1.4063584813148979e-06,
      "loss": 305.6008,
      "step": 43200
    },
    {
      "epoch": 0.8310334548182055,
      "grad_norm": 420.2690005893665,
      "learning_rate": 1.4032391539104484e-06,
      "loss": 312.0884,
      "step": 43210
    },
    {
      "epoch": 0.8312257791539652,
      "grad_norm": 388.7972106008941,
      "learning_rate": 1.4001230286574363e-06,
      "loss": 310.4407,
      "step": 43220
    },
    {
      "epoch": 0.8314181034897251,
      "grad_norm": 420.6261331747331,
      "learning_rate": 1.3970101067165642e-06,
      "loss": 306.8434,
      "step": 43230
    },
    {
      "epoch": 0.8316104278254849,
      "grad_norm": 399.7898065970122,
      "learning_rate": 1.3939003892473446e-06,
      "loss": 305.6393,
      "step": 43240
    },
    {
      "epoch": 0.8318027521612448,
      "grad_norm": 465.89894772511826,
      "learning_rate": 1.390793877408093e-06,
      "loss": 310.9626,
      "step": 43250
    },
    {
      "epoch": 0.8319950764970045,
      "grad_norm": 419.9633912750672,
      "learning_rate": 1.3876905723559397e-06,
      "loss": 306.179,
      "step": 43260
    },
    {
      "epoch": 0.8321874008327643,
      "grad_norm": 382.302731153885,
      "learning_rate": 1.3845904752468075e-06,
      "loss": 311.5814,
      "step": 43270
    },
    {
      "epoch": 0.8323797251685242,
      "grad_norm": 464.7376973882011,
      "learning_rate": 1.3814935872354385e-06,
      "loss": 303.6585,
      "step": 43280
    },
    {
      "epoch": 0.832572049504284,
      "grad_norm": 400.69595239377145,
      "learning_rate": 1.3783999094753653e-06,
      "loss": 307.3913,
      "step": 43290
    },
    {
      "epoch": 0.8327643738400439,
      "grad_norm": 420.1239247443546,
      "learning_rate": 1.3753094431189385e-06,
      "loss": 312.9189,
      "step": 43300
    },
    {
      "epoch": 0.8329566981758036,
      "grad_norm": 387.0049299290587,
      "learning_rate": 1.3722221893173027e-06,
      "loss": 300.6439,
      "step": 43310
    },
    {
      "epoch": 0.8331490225115635,
      "grad_norm": 389.1267103728879,
      "learning_rate": 1.369138149220407e-06,
      "loss": 307.6452,
      "step": 43320
    },
    {
      "epoch": 0.8333413468473233,
      "grad_norm": 428.0639024149446,
      "learning_rate": 1.3660573239770091e-06,
      "loss": 313.8387,
      "step": 43330
    },
    {
      "epoch": 0.8335336711830832,
      "grad_norm": 425.0378746736501,
      "learning_rate": 1.3629797147346635e-06,
      "loss": 317.9529,
      "step": 43340
    },
    {
      "epoch": 0.833725995518843,
      "grad_norm": 425.08552944771554,
      "learning_rate": 1.3599053226397275e-06,
      "loss": 312.3346,
      "step": 43350
    },
    {
      "epoch": 0.8339183198546029,
      "grad_norm": 457.2770452154439,
      "learning_rate": 1.3568341488373637e-06,
      "loss": 311.3716,
      "step": 43360
    },
    {
      "epoch": 0.8341106441903626,
      "grad_norm": 383.3616072731839,
      "learning_rate": 1.3537661944715342e-06,
      "loss": 305.7466,
      "step": 43370
    },
    {
      "epoch": 0.8343029685261224,
      "grad_norm": 378.19750986663985,
      "learning_rate": 1.350701460685e-06,
      "loss": 312.2435,
      "step": 43380
    },
    {
      "epoch": 0.8344952928618823,
      "grad_norm": 406.6171330880953,
      "learning_rate": 1.3476399486193214e-06,
      "loss": 304.5825,
      "step": 43390
    },
    {
      "epoch": 0.8346876171976421,
      "grad_norm": 434.06189110066236,
      "learning_rate": 1.3445816594148654e-06,
      "loss": 320.4958,
      "step": 43400
    },
    {
      "epoch": 0.834879941533402,
      "grad_norm": 439.5465007523655,
      "learning_rate": 1.3415265942107925e-06,
      "loss": 316.4296,
      "step": 43410
    },
    {
      "epoch": 0.8350722658691617,
      "grad_norm": 392.45425060422684,
      "learning_rate": 1.3384747541450615e-06,
      "loss": 300.5675,
      "step": 43420
    },
    {
      "epoch": 0.8352645902049216,
      "grad_norm": 436.9471421328807,
      "learning_rate": 1.3354261403544345e-06,
      "loss": 309.4496,
      "step": 43430
    },
    {
      "epoch": 0.8354569145406814,
      "grad_norm": 428.5568221156891,
      "learning_rate": 1.3323807539744726e-06,
      "loss": 320.4364,
      "step": 43440
    },
    {
      "epoch": 0.8356492388764413,
      "grad_norm": 406.89079248718434,
      "learning_rate": 1.32933859613953e-06,
      "loss": 311.7307,
      "step": 43450
    },
    {
      "epoch": 0.835841563212201,
      "grad_norm": 432.8153667171676,
      "learning_rate": 1.3262996679827567e-06,
      "loss": 310.3885,
      "step": 43460
    },
    {
      "epoch": 0.8360338875479609,
      "grad_norm": 426.9602538393154,
      "learning_rate": 1.3232639706361083e-06,
      "loss": 308.9982,
      "step": 43470
    },
    {
      "epoch": 0.8362262118837207,
      "grad_norm": 420.303895961567,
      "learning_rate": 1.3202315052303304e-06,
      "loss": 296.2161,
      "step": 43480
    },
    {
      "epoch": 0.8364185362194805,
      "grad_norm": 391.9618038875555,
      "learning_rate": 1.3172022728949651e-06,
      "loss": 307.7306,
      "step": 43490
    },
    {
      "epoch": 0.8366108605552404,
      "grad_norm": 418.06481889559166,
      "learning_rate": 1.3141762747583498e-06,
      "loss": 300.9838,
      "step": 43500
    },
    {
      "epoch": 0.8368031848910001,
      "grad_norm": 401.4725497286522,
      "learning_rate": 1.3111535119476237e-06,
      "loss": 309.4073,
      "step": 43510
    },
    {
      "epoch": 0.83699550922676,
      "grad_norm": 424.73035958519137,
      "learning_rate": 1.3081339855887133e-06,
      "loss": 316.1311,
      "step": 43520
    },
    {
      "epoch": 0.8371878335625198,
      "grad_norm": 420.0330412783158,
      "learning_rate": 1.3051176968063407e-06,
      "loss": 313.0947,
      "step": 43530
    },
    {
      "epoch": 0.8373801578982797,
      "grad_norm": 429.4214631178607,
      "learning_rate": 1.302104646724026e-06,
      "loss": 307.8752,
      "step": 43540
    },
    {
      "epoch": 0.8375724822340395,
      "grad_norm": 412.1896170880573,
      "learning_rate": 1.299094836464081e-06,
      "loss": 312.5322,
      "step": 43550
    },
    {
      "epoch": 0.8377648065697993,
      "grad_norm": 421.7241075889308,
      "learning_rate": 1.2960882671476062e-06,
      "loss": 304.3012,
      "step": 43560
    },
    {
      "epoch": 0.8379571309055591,
      "grad_norm": 363.007332446123,
      "learning_rate": 1.2930849398945033e-06,
      "loss": 303.4901,
      "step": 43570
    },
    {
      "epoch": 0.838149455241319,
      "grad_norm": 405.42988986184895,
      "learning_rate": 1.2900848558234625e-06,
      "loss": 309.2977,
      "step": 43580
    },
    {
      "epoch": 0.8383417795770788,
      "grad_norm": 403.78264280695635,
      "learning_rate": 1.2870880160519628e-06,
      "loss": 311.6537,
      "step": 43590
    },
    {
      "epoch": 0.8385341039128386,
      "grad_norm": 408.8079662781415,
      "learning_rate": 1.2840944216962802e-06,
      "loss": 311.8632,
      "step": 43600
    },
    {
      "epoch": 0.8387264282485984,
      "grad_norm": 405.3287525107537,
      "learning_rate": 1.2811040738714742e-06,
      "loss": 305.9968,
      "step": 43610
    },
    {
      "epoch": 0.8389187525843582,
      "grad_norm": 396.6957967160807,
      "learning_rate": 1.2781169736914067e-06,
      "loss": 303.5648,
      "step": 43620
    },
    {
      "epoch": 0.8391110769201181,
      "grad_norm": 401.18494693017317,
      "learning_rate": 1.275133122268719e-06,
      "loss": 300.0335,
      "step": 43630
    },
    {
      "epoch": 0.8393034012558779,
      "grad_norm": 402.0169973967491,
      "learning_rate": 1.2721525207148456e-06,
      "loss": 307.8517,
      "step": 43640
    },
    {
      "epoch": 0.8394957255916378,
      "grad_norm": 441.2071924621578,
      "learning_rate": 1.2691751701400145e-06,
      "loss": 309.8471,
      "step": 43650
    },
    {
      "epoch": 0.8396880499273975,
      "grad_norm": 416.63996687023854,
      "learning_rate": 1.2662010716532392e-06,
      "loss": 309.8868,
      "step": 43660
    },
    {
      "epoch": 0.8398803742631574,
      "grad_norm": 423.3957295128237,
      "learning_rate": 1.2632302263623198e-06,
      "loss": 300.3247,
      "step": 43670
    },
    {
      "epoch": 0.8400726985989172,
      "grad_norm": 385.76173705849806,
      "learning_rate": 1.2602626353738479e-06,
      "loss": 310.8522,
      "step": 43680
    },
    {
      "epoch": 0.8402650229346771,
      "grad_norm": 419.35014208195366,
      "learning_rate": 1.2572982997932037e-06,
      "loss": 314.829,
      "step": 43690
    },
    {
      "epoch": 0.8404573472704369,
      "grad_norm": 437.044226891431,
      "learning_rate": 1.2543372207245508e-06,
      "loss": 308.5821,
      "step": 43700
    },
    {
      "epoch": 0.8406496716061966,
      "grad_norm": 392.24831579143245,
      "learning_rate": 1.2513793992708467e-06,
      "loss": 314.5268,
      "step": 43710
    },
    {
      "epoch": 0.8408419959419565,
      "grad_norm": 415.8970967651331,
      "learning_rate": 1.2484248365338248e-06,
      "loss": 308.462,
      "step": 43720
    },
    {
      "epoch": 0.8410343202777163,
      "grad_norm": 396.3238155580709,
      "learning_rate": 1.2454735336140167e-06,
      "loss": 309.7371,
      "step": 43730
    },
    {
      "epoch": 0.8412266446134762,
      "grad_norm": 392.4687504825427,
      "learning_rate": 1.2425254916107321e-06,
      "loss": 308.669,
      "step": 43740
    },
    {
      "epoch": 0.841418968949236,
      "grad_norm": 399.34987357090506,
      "learning_rate": 1.2395807116220648e-06,
      "loss": 312.6412,
      "step": 43750
    },
    {
      "epoch": 0.8416112932849958,
      "grad_norm": 417.38855913189497,
      "learning_rate": 1.236639194744902e-06,
      "loss": 315.5179,
      "step": 43760
    },
    {
      "epoch": 0.8418036176207556,
      "grad_norm": 429.66975190352093,
      "learning_rate": 1.233700942074907e-06,
      "loss": 307.1796,
      "step": 43770
    },
    {
      "epoch": 0.8419959419565155,
      "grad_norm": 382.2906613329996,
      "learning_rate": 1.230765954706531e-06,
      "loss": 303.7333,
      "step": 43780
    },
    {
      "epoch": 0.8421882662922753,
      "grad_norm": 448.6587954719122,
      "learning_rate": 1.227834233733005e-06,
      "loss": 310.9136,
      "step": 43790
    },
    {
      "epoch": 0.8423805906280352,
      "grad_norm": 398.58347546090255,
      "learning_rate": 1.2249057802463527e-06,
      "loss": 309.7333,
      "step": 43800
    },
    {
      "epoch": 0.8425729149637949,
      "grad_norm": 382.80139291560556,
      "learning_rate": 1.221980595337372e-06,
      "loss": 305.871,
      "step": 43810
    },
    {
      "epoch": 0.8427652392995547,
      "grad_norm": 384.28099373040374,
      "learning_rate": 1.219058680095644e-06,
      "loss": 310.9953,
      "step": 43820
    },
    {
      "epoch": 0.8429575636353146,
      "grad_norm": 390.50239126599206,
      "learning_rate": 1.2161400356095376e-06,
      "loss": 300.5285,
      "step": 43830
    },
    {
      "epoch": 0.8431498879710744,
      "grad_norm": 394.7135930797511,
      "learning_rate": 1.2132246629661948e-06,
      "loss": 307.0489,
      "step": 43840
    },
    {
      "epoch": 0.8433422123068343,
      "grad_norm": 389.18594248152215,
      "learning_rate": 1.21031256325155e-06,
      "loss": 309.5433,
      "step": 43850
    },
    {
      "epoch": 0.843534536642594,
      "grad_norm": 410.64966859334305,
      "learning_rate": 1.2074037375503056e-06,
      "loss": 298.6748,
      "step": 43860
    },
    {
      "epoch": 0.8437268609783539,
      "grad_norm": 384.6362019726913,
      "learning_rate": 1.2044981869459571e-06,
      "loss": 313.5294,
      "step": 43870
    },
    {
      "epoch": 0.8439191853141137,
      "grad_norm": 404.3040948416852,
      "learning_rate": 1.201595912520771e-06,
      "loss": 310.4116,
      "step": 43880
    },
    {
      "epoch": 0.8441115096498736,
      "grad_norm": 399.8936341399085,
      "learning_rate": 1.198696915355796e-06,
      "loss": 321.4883,
      "step": 43890
    },
    {
      "epoch": 0.8443038339856334,
      "grad_norm": 414.61600652742237,
      "learning_rate": 1.1958011965308624e-06,
      "loss": 316.2538,
      "step": 43900
    },
    {
      "epoch": 0.8444961583213932,
      "grad_norm": 413.8168917707275,
      "learning_rate": 1.192908757124578e-06,
      "loss": 310.5925,
      "step": 43910
    },
    {
      "epoch": 0.844688482657153,
      "grad_norm": 381.0727520182692,
      "learning_rate": 1.190019598214327e-06,
      "loss": 311.7601,
      "step": 43920
    },
    {
      "epoch": 0.8448808069929129,
      "grad_norm": 434.9484055456739,
      "learning_rate": 1.1871337208762723e-06,
      "loss": 303.6549,
      "step": 43930
    },
    {
      "epoch": 0.8450731313286727,
      "grad_norm": 460.2362992462162,
      "learning_rate": 1.1842511261853596e-06,
      "loss": 316.0865,
      "step": 43940
    },
    {
      "epoch": 0.8452654556644325,
      "grad_norm": 386.9826462964263,
      "learning_rate": 1.1813718152153054e-06,
      "loss": 303.8425,
      "step": 43950
    },
    {
      "epoch": 0.8454577800001923,
      "grad_norm": 424.40337115020503,
      "learning_rate": 1.1784957890386051e-06,
      "loss": 305.3876,
      "step": 43960
    },
    {
      "epoch": 0.8456501043359521,
      "grad_norm": 404.1909632525495,
      "learning_rate": 1.1756230487265296e-06,
      "loss": 305.2239,
      "step": 43970
    },
    {
      "epoch": 0.845842428671712,
      "grad_norm": 398.4495033039118,
      "learning_rate": 1.1727535953491308e-06,
      "loss": 308.1323,
      "step": 43980
    },
    {
      "epoch": 0.8460347530074718,
      "grad_norm": 389.2480479244216,
      "learning_rate": 1.1698874299752293e-06,
      "loss": 299.6056,
      "step": 43990
    },
    {
      "epoch": 0.8462270773432317,
      "grad_norm": 394.36355835421574,
      "learning_rate": 1.1670245536724267e-06,
      "loss": 309.4105,
      "step": 44000
    },
    {
      "epoch": 0.8464194016789914,
      "grad_norm": 397.68585784214036,
      "learning_rate": 1.1641649675070975e-06,
      "loss": 312.691,
      "step": 44010
    },
    {
      "epoch": 0.8466117260147513,
      "grad_norm": 410.6147010479267,
      "learning_rate": 1.161308672544389e-06,
      "loss": 326.4685,
      "step": 44020
    },
    {
      "epoch": 0.8468040503505111,
      "grad_norm": 410.55330677161714,
      "learning_rate": 1.1584556698482252e-06,
      "loss": 305.0545,
      "step": 44030
    },
    {
      "epoch": 0.846996374686271,
      "grad_norm": 431.9844551976268,
      "learning_rate": 1.1556059604812985e-06,
      "loss": 306.52,
      "step": 44040
    },
    {
      "epoch": 0.8471886990220308,
      "grad_norm": 405.437695786694,
      "learning_rate": 1.1527595455050844e-06,
      "loss": 309.3423,
      "step": 44050
    },
    {
      "epoch": 0.8473810233577905,
      "grad_norm": 459.3400718099756,
      "learning_rate": 1.1499164259798223e-06,
      "loss": 310.3186,
      "step": 44060
    },
    {
      "epoch": 0.8475733476935504,
      "grad_norm": 385.3061173062442,
      "learning_rate": 1.1470766029645253e-06,
      "loss": 315.4323,
      "step": 44070
    },
    {
      "epoch": 0.8477656720293102,
      "grad_norm": 430.51500251146587,
      "learning_rate": 1.1442400775169849e-06,
      "loss": 310.8122,
      "step": 44080
    },
    {
      "epoch": 0.8479579963650701,
      "grad_norm": 382.0245139719309,
      "learning_rate": 1.141406850693757e-06,
      "loss": 299.5895,
      "step": 44090
    },
    {
      "epoch": 0.8481503207008299,
      "grad_norm": 393.55300023450496,
      "learning_rate": 1.1385769235501742e-06,
      "loss": 311.6865,
      "step": 44100
    },
    {
      "epoch": 0.8483426450365897,
      "grad_norm": 439.4190745537439,
      "learning_rate": 1.1357502971403335e-06,
      "loss": 313.2368,
      "step": 44110
    },
    {
      "epoch": 0.8485349693723495,
      "grad_norm": 409.10469866554496,
      "learning_rate": 1.132926972517111e-06,
      "loss": 306.4891,
      "step": 44120
    },
    {
      "epoch": 0.8487272937081094,
      "grad_norm": 424.9147398193525,
      "learning_rate": 1.130106950732145e-06,
      "loss": 302.083,
      "step": 44130
    },
    {
      "epoch": 0.8489196180438692,
      "grad_norm": 425.26982812103034,
      "learning_rate": 1.1272902328358514e-06,
      "loss": 309.5698,
      "step": 44140
    },
    {
      "epoch": 0.8491119423796291,
      "grad_norm": 474.2980147483282,
      "learning_rate": 1.1244768198774047e-06,
      "loss": 317.164,
      "step": 44150
    },
    {
      "epoch": 0.8493042667153888,
      "grad_norm": 443.4806721760526,
      "learning_rate": 1.121666712904762e-06,
      "loss": 329.2048,
      "step": 44160
    },
    {
      "epoch": 0.8494965910511486,
      "grad_norm": 376.2163407206971,
      "learning_rate": 1.1188599129646382e-06,
      "loss": 307.9291,
      "step": 44170
    },
    {
      "epoch": 0.8496889153869085,
      "grad_norm": 441.9483483619356,
      "learning_rate": 1.116056421102517e-06,
      "loss": 320.0675,
      "step": 44180
    },
    {
      "epoch": 0.8498812397226683,
      "grad_norm": 401.7978514444669,
      "learning_rate": 1.113256238362659e-06,
      "loss": 313.4776,
      "step": 44190
    },
    {
      "epoch": 0.8500735640584282,
      "grad_norm": 437.6590866201023,
      "learning_rate": 1.1104593657880812e-06,
      "loss": 306.1304,
      "step": 44200
    },
    {
      "epoch": 0.8502658883941879,
      "grad_norm": 417.4508570371858,
      "learning_rate": 1.1076658044205746e-06,
      "loss": 316.2343,
      "step": 44210
    },
    {
      "epoch": 0.8504582127299478,
      "grad_norm": 397.3498340990389,
      "learning_rate": 1.1048755553006928e-06,
      "loss": 313.7024,
      "step": 44220
    },
    {
      "epoch": 0.8506505370657076,
      "grad_norm": 413.2392266095835,
      "learning_rate": 1.1020886194677605e-06,
      "loss": 302.59,
      "step": 44230
    },
    {
      "epoch": 0.8508428614014675,
      "grad_norm": 400.93040562028807,
      "learning_rate": 1.0993049979598635e-06,
      "loss": 309.4014,
      "step": 44240
    },
    {
      "epoch": 0.8510351857372273,
      "grad_norm": 404.9059784641159,
      "learning_rate": 1.0965246918138529e-06,
      "loss": 317.5931,
      "step": 44250
    },
    {
      "epoch": 0.8512275100729871,
      "grad_norm": 397.77934194055143,
      "learning_rate": 1.093747702065351e-06,
      "loss": 306.4594,
      "step": 44260
    },
    {
      "epoch": 0.8514198344087469,
      "grad_norm": 400.8242068851222,
      "learning_rate": 1.090974029748736e-06,
      "loss": 312.0326,
      "step": 44270
    },
    {
      "epoch": 0.8516121587445067,
      "grad_norm": 399.29201610333854,
      "learning_rate": 1.0882036758971592e-06,
      "loss": 307.1808,
      "step": 44280
    },
    {
      "epoch": 0.8518044830802666,
      "grad_norm": 394.45936479744114,
      "learning_rate": 1.0854366415425289e-06,
      "loss": 319.9591,
      "step": 44290
    },
    {
      "epoch": 0.8519968074160263,
      "grad_norm": 366.56628147727963,
      "learning_rate": 1.0826729277155224e-06,
      "loss": 306.8888,
      "step": 44300
    },
    {
      "epoch": 0.8521891317517862,
      "grad_norm": 433.5542684729299,
      "learning_rate": 1.0799125354455752e-06,
      "loss": 319.3518,
      "step": 44310
    },
    {
      "epoch": 0.852381456087546,
      "grad_norm": 408.8071207439392,
      "learning_rate": 1.0771554657608896e-06,
      "loss": 308.4293,
      "step": 44320
    },
    {
      "epoch": 0.8525737804233059,
      "grad_norm": 411.1904044875333,
      "learning_rate": 1.0744017196884248e-06,
      "loss": 306.739,
      "step": 44330
    },
    {
      "epoch": 0.8527661047590657,
      "grad_norm": 403.2756905542512,
      "learning_rate": 1.0716512982539106e-06,
      "loss": 304.3856,
      "step": 44340
    },
    {
      "epoch": 0.8529584290948256,
      "grad_norm": 398.3733658713629,
      "learning_rate": 1.0689042024818307e-06,
      "loss": 308.7248,
      "step": 44350
    },
    {
      "epoch": 0.8531507534305853,
      "grad_norm": 456.1789084319189,
      "learning_rate": 1.0661604333954312e-06,
      "loss": 316.1972,
      "step": 44360
    },
    {
      "epoch": 0.8533430777663452,
      "grad_norm": 374.37026480962015,
      "learning_rate": 1.0634199920167255e-06,
      "loss": 303.3029,
      "step": 44370
    },
    {
      "epoch": 0.853535402102105,
      "grad_norm": 402.3953841877796,
      "learning_rate": 1.0606828793664804e-06,
      "loss": 307.8187,
      "step": 44380
    },
    {
      "epoch": 0.8537277264378648,
      "grad_norm": 416.9853259567134,
      "learning_rate": 1.0579490964642247e-06,
      "loss": 312.1465,
      "step": 44390
    },
    {
      "epoch": 0.8539200507736247,
      "grad_norm": 436.07586502455825,
      "learning_rate": 1.0552186443282464e-06,
      "loss": 310.3381,
      "step": 44400
    },
    {
      "epoch": 0.8541123751093844,
      "grad_norm": 385.18047741672416,
      "learning_rate": 1.0524915239755939e-06,
      "loss": 302.7083,
      "step": 44410
    },
    {
      "epoch": 0.8543046994451443,
      "grad_norm": 435.882890877593,
      "learning_rate": 1.0497677364220792e-06,
      "loss": 308.524,
      "step": 44420
    },
    {
      "epoch": 0.8544970237809041,
      "grad_norm": 381.2653958431397,
      "learning_rate": 1.047047282682262e-06,
      "loss": 313.531,
      "step": 44430
    },
    {
      "epoch": 0.854689348116664,
      "grad_norm": 412.582012402211,
      "learning_rate": 1.0443301637694713e-06,
      "loss": 317.9873,
      "step": 44440
    },
    {
      "epoch": 0.8548816724524237,
      "grad_norm": 416.6160209458399,
      "learning_rate": 1.0416163806957857e-06,
      "loss": 296.6295,
      "step": 44450
    },
    {
      "epoch": 0.8550739967881836,
      "grad_norm": 413.6593353856753,
      "learning_rate": 1.0389059344720475e-06,
      "loss": 307.4951,
      "step": 44460
    },
    {
      "epoch": 0.8552663211239434,
      "grad_norm": 379.9253461959537,
      "learning_rate": 1.0361988261078482e-06,
      "loss": 302.6177,
      "step": 44470
    },
    {
      "epoch": 0.8554586454597033,
      "grad_norm": 396.7851390931577,
      "learning_rate": 1.0334950566115466e-06,
      "loss": 299.1118,
      "step": 44480
    },
    {
      "epoch": 0.8556509697954631,
      "grad_norm": 409.273740356896,
      "learning_rate": 1.0307946269902492e-06,
      "loss": 310.5079,
      "step": 44490
    },
    {
      "epoch": 0.8558432941312228,
      "grad_norm": 380.53345495027446,
      "learning_rate": 1.0280975382498225e-06,
      "loss": 309.5644,
      "step": 44500
    },
    {
      "epoch": 0.8560356184669827,
      "grad_norm": 401.0114511136919,
      "learning_rate": 1.0254037913948845e-06,
      "loss": 295.9165,
      "step": 44510
    },
    {
      "epoch": 0.8562279428027425,
      "grad_norm": 417.74737228382287,
      "learning_rate": 1.0227133874288152e-06,
      "loss": 314.8125,
      "step": 44520
    },
    {
      "epoch": 0.8564202671385024,
      "grad_norm": 389.81975755023103,
      "learning_rate": 1.0200263273537458e-06,
      "loss": 307.6695,
      "step": 44530
    },
    {
      "epoch": 0.8566125914742622,
      "grad_norm": 407.9799030203983,
      "learning_rate": 1.0173426121705577e-06,
      "loss": 314.1563,
      "step": 44540
    },
    {
      "epoch": 0.856804915810022,
      "grad_norm": 413.7191571147601,
      "learning_rate": 1.0146622428788943e-06,
      "loss": 309.1433,
      "step": 44550
    },
    {
      "epoch": 0.8569972401457818,
      "grad_norm": 403.3886196539769,
      "learning_rate": 1.0119852204771463e-06,
      "loss": 317.8424,
      "step": 44560
    },
    {
      "epoch": 0.8571895644815417,
      "grad_norm": 393.5342442825214,
      "learning_rate": 1.0093115459624637e-06,
      "loss": 305.5662,
      "step": 44570
    },
    {
      "epoch": 0.8573818888173015,
      "grad_norm": 387.4852282953983,
      "learning_rate": 1.0066412203307419e-06,
      "loss": 314.7758,
      "step": 44580
    },
    {
      "epoch": 0.8575742131530614,
      "grad_norm": 409.5449582952906,
      "learning_rate": 1.0039742445766376e-06,
      "loss": 306.6252,
      "step": 44590
    },
    {
      "epoch": 0.8577665374888211,
      "grad_norm": 417.5166939829627,
      "learning_rate": 1.0013106196935528e-06,
      "loss": 313.7952,
      "step": 44600
    },
    {
      "epoch": 0.8579588618245809,
      "grad_norm": 423.1998638370606,
      "learning_rate": 9.986503466736419e-07,
      "loss": 303.5087,
      "step": 44610
    },
    {
      "epoch": 0.8581511861603408,
      "grad_norm": 421.73776360246177,
      "learning_rate": 9.959934265078176e-07,
      "loss": 317.2739,
      "step": 44620
    },
    {
      "epoch": 0.8583435104961006,
      "grad_norm": 433.1108423808941,
      "learning_rate": 9.933398601857347e-07,
      "loss": 304.4954,
      "step": 44630
    },
    {
      "epoch": 0.8585358348318605,
      "grad_norm": 396.3648053219482,
      "learning_rate": 9.90689648695804e-07,
      "loss": 305.1951,
      "step": 44640
    },
    {
      "epoch": 0.8587281591676202,
      "grad_norm": 407.0273369340253,
      "learning_rate": 9.880427930251834e-07,
      "loss": 305.0792,
      "step": 44650
    },
    {
      "epoch": 0.8589204835033801,
      "grad_norm": 385.9401191788157,
      "learning_rate": 9.853992941597878e-07,
      "loss": 305.2848,
      "step": 44660
    },
    {
      "epoch": 0.8591128078391399,
      "grad_norm": 384.6131563985859,
      "learning_rate": 9.827591530842729e-07,
      "loss": 302.1396,
      "step": 44670
    },
    {
      "epoch": 0.8593051321748998,
      "grad_norm": 379.82487906748423,
      "learning_rate": 9.801223707820484e-07,
      "loss": 305.8356,
      "step": 44680
    },
    {
      "epoch": 0.8594974565106596,
      "grad_norm": 386.42129594011146,
      "learning_rate": 9.774889482352735e-07,
      "loss": 300.0386,
      "step": 44690
    },
    {
      "epoch": 0.8596897808464194,
      "grad_norm": 376.1601710340815,
      "learning_rate": 9.74858886424852e-07,
      "loss": 303.9091,
      "step": 44700
    },
    {
      "epoch": 0.8598821051821792,
      "grad_norm": 403.6189853606279,
      "learning_rate": 9.722321863304418e-07,
      "loss": 311.3838,
      "step": 44710
    },
    {
      "epoch": 0.860074429517939,
      "grad_norm": 419.844702330754,
      "learning_rate": 9.696088489304412e-07,
      "loss": 312.1175,
      "step": 44720
    },
    {
      "epoch": 0.8602667538536989,
      "grad_norm": 405.32264576387394,
      "learning_rate": 9.669888752020061e-07,
      "loss": 306.867,
      "step": 44730
    },
    {
      "epoch": 0.8604590781894587,
      "grad_norm": 392.7020945862724,
      "learning_rate": 9.643722661210285e-07,
      "loss": 297.6295,
      "step": 44740
    },
    {
      "epoch": 0.8606514025252185,
      "grad_norm": 385.5000714377669,
      "learning_rate": 9.617590226621543e-07,
      "loss": 303.4644,
      "step": 44750
    },
    {
      "epoch": 0.8608437268609783,
      "grad_norm": 395.86705638238874,
      "learning_rate": 9.5914914579877e-07,
      "loss": 303.246,
      "step": 44760
    },
    {
      "epoch": 0.8610360511967382,
      "grad_norm": 384.34447705197493,
      "learning_rate": 9.565426365030172e-07,
      "loss": 309.0579,
      "step": 44770
    },
    {
      "epoch": 0.861228375532498,
      "grad_norm": 389.97989738581896,
      "learning_rate": 9.539394957457737e-07,
      "loss": 316.2567,
      "step": 44780
    },
    {
      "epoch": 0.8614206998682579,
      "grad_norm": 379.5702998136998,
      "learning_rate": 9.51339724496666e-07,
      "loss": 305.6684,
      "step": 44790
    },
    {
      "epoch": 0.8616130242040176,
      "grad_norm": 380.28903954875796,
      "learning_rate": 9.487433237240695e-07,
      "loss": 301.4612,
      "step": 44800
    },
    {
      "epoch": 0.8618053485397775,
      "grad_norm": 408.0930043897811,
      "learning_rate": 9.461502943950973e-07,
      "loss": 316.251,
      "step": 44810
    },
    {
      "epoch": 0.8619976728755373,
      "grad_norm": 359.93829401593445,
      "learning_rate": 9.435606374756123e-07,
      "loss": 296.3609,
      "step": 44820
    },
    {
      "epoch": 0.8621899972112971,
      "grad_norm": 393.5957222170627,
      "learning_rate": 9.409743539302152e-07,
      "loss": 298.2376,
      "step": 44830
    },
    {
      "epoch": 0.862382321547057,
      "grad_norm": 416.6577577424038,
      "learning_rate": 9.383914447222576e-07,
      "loss": 308.6023,
      "step": 44840
    },
    {
      "epoch": 0.8625746458828167,
      "grad_norm": 417.04512185974056,
      "learning_rate": 9.358119108138309e-07,
      "loss": 311.7073,
      "step": 44850
    },
    {
      "epoch": 0.8627669702185766,
      "grad_norm": 381.7221785189243,
      "learning_rate": 9.332357531657644e-07,
      "loss": 296.6584,
      "step": 44860
    },
    {
      "epoch": 0.8629592945543364,
      "grad_norm": 381.2964103584031,
      "learning_rate": 9.306629727376404e-07,
      "loss": 304.0799,
      "step": 44870
    },
    {
      "epoch": 0.8631516188900963,
      "grad_norm": 418.31063300944476,
      "learning_rate": 9.280935704877736e-07,
      "loss": 302.2871,
      "step": 44880
    },
    {
      "epoch": 0.8633439432258561,
      "grad_norm": 440.54529215087496,
      "learning_rate": 9.255275473732239e-07,
      "loss": 311.563,
      "step": 44890
    },
    {
      "epoch": 0.8635362675616159,
      "grad_norm": 414.4605870446404,
      "learning_rate": 9.229649043497924e-07,
      "loss": 309.905,
      "step": 44900
    },
    {
      "epoch": 0.8637285918973757,
      "grad_norm": 429.4819911621607,
      "learning_rate": 9.20405642372022e-07,
      "loss": 313.3387,
      "step": 44910
    },
    {
      "epoch": 0.8639209162331356,
      "grad_norm": 469.97345671395095,
      "learning_rate": 9.178497623931959e-07,
      "loss": 309.0582,
      "step": 44920
    },
    {
      "epoch": 0.8641132405688954,
      "grad_norm": 401.4636649157551,
      "learning_rate": 9.152972653653369e-07,
      "loss": 303.3662,
      "step": 44930
    },
    {
      "epoch": 0.8643055649046553,
      "grad_norm": 412.3965830265295,
      "learning_rate": 9.127481522392068e-07,
      "loss": 308.8958,
      "step": 44940
    },
    {
      "epoch": 0.864497889240415,
      "grad_norm": 386.2214286793937,
      "learning_rate": 9.102024239643092e-07,
      "loss": 307.6219,
      "step": 44950
    },
    {
      "epoch": 0.8646902135761748,
      "grad_norm": 441.73853293879324,
      "learning_rate": 9.076600814888869e-07,
      "loss": 308.5775,
      "step": 44960
    },
    {
      "epoch": 0.8648825379119347,
      "grad_norm": 389.30577518183196,
      "learning_rate": 9.051211257599169e-07,
      "loss": 310.0769,
      "step": 44970
    },
    {
      "epoch": 0.8650748622476945,
      "grad_norm": 368.22314029030315,
      "learning_rate": 9.025855577231224e-07,
      "loss": 310.5161,
      "step": 44980
    },
    {
      "epoch": 0.8652671865834544,
      "grad_norm": 462.5940260468949,
      "learning_rate": 9.000533783229581e-07,
      "loss": 309.9495,
      "step": 44990
    },
    {
      "epoch": 0.8654595109192141,
      "grad_norm": 383.42489132058324,
      "learning_rate": 8.975245885026207e-07,
      "loss": 314.3997,
      "step": 45000
    },
    {
      "epoch": 0.865651835254974,
      "grad_norm": 459.18740096829055,
      "learning_rate": 8.949991892040399e-07,
      "loss": 309.851,
      "step": 45010
    },
    {
      "epoch": 0.8658441595907338,
      "grad_norm": 431.3174954620526,
      "learning_rate": 8.92477181367889e-07,
      "loss": 303.2337,
      "step": 45020
    },
    {
      "epoch": 0.8660364839264937,
      "grad_norm": 399.99411197624624,
      "learning_rate": 8.899585659335719e-07,
      "loss": 305.1389,
      "step": 45030
    },
    {
      "epoch": 0.8662288082622535,
      "grad_norm": 421.8670178565545,
      "learning_rate": 8.874433438392305e-07,
      "loss": 328.1207,
      "step": 45040
    },
    {
      "epoch": 0.8664211325980133,
      "grad_norm": 405.9968790015555,
      "learning_rate": 8.849315160217465e-07,
      "loss": 308.3939,
      "step": 45050
    },
    {
      "epoch": 0.8666134569337731,
      "grad_norm": 422.5961644624331,
      "learning_rate": 8.824230834167325e-07,
      "loss": 306.0928,
      "step": 45060
    },
    {
      "epoch": 0.8668057812695329,
      "grad_norm": 390.32379695472224,
      "learning_rate": 8.799180469585378e-07,
      "loss": 322.4722,
      "step": 45070
    },
    {
      "epoch": 0.8669981056052928,
      "grad_norm": 399.9720048675275,
      "learning_rate": 8.77416407580246e-07,
      "loss": 307.3033,
      "step": 45080
    },
    {
      "epoch": 0.8671904299410526,
      "grad_norm": 414.4201419526888,
      "learning_rate": 8.749181662136785e-07,
      "loss": 316.3143,
      "step": 45090
    },
    {
      "epoch": 0.8673827542768124,
      "grad_norm": 427.3499862849529,
      "learning_rate": 8.724233237893897e-07,
      "loss": 306.0224,
      "step": 45100
    },
    {
      "epoch": 0.8675750786125722,
      "grad_norm": 402.6087166065143,
      "learning_rate": 8.699318812366641e-07,
      "loss": 309.4246,
      "step": 45110
    },
    {
      "epoch": 0.8677674029483321,
      "grad_norm": 393.55856381573716,
      "learning_rate": 8.67443839483526e-07,
      "loss": 313.1001,
      "step": 45120
    },
    {
      "epoch": 0.8679597272840919,
      "grad_norm": 409.6104681860515,
      "learning_rate": 8.649591994567275e-07,
      "loss": 305.6595,
      "step": 45130
    },
    {
      "epoch": 0.8681520516198518,
      "grad_norm": 376.10437758020055,
      "learning_rate": 8.62477962081758e-07,
      "loss": 307.8717,
      "step": 45140
    },
    {
      "epoch": 0.8683443759556115,
      "grad_norm": 434.1553607527366,
      "learning_rate": 8.600001282828341e-07,
      "loss": 316.7945,
      "step": 45150
    },
    {
      "epoch": 0.8685367002913714,
      "grad_norm": 436.2063378999651,
      "learning_rate": 8.57525698982914e-07,
      "loss": 311.1892,
      "step": 45160
    },
    {
      "epoch": 0.8687290246271312,
      "grad_norm": 398.01392891360763,
      "learning_rate": 8.550546751036759e-07,
      "loss": 315.2501,
      "step": 45170
    },
    {
      "epoch": 0.868921348962891,
      "grad_norm": 419.1874741231257,
      "learning_rate": 8.525870575655393e-07,
      "loss": 307.8161,
      "step": 45180
    },
    {
      "epoch": 0.8691136732986509,
      "grad_norm": 405.6690941889232,
      "learning_rate": 8.501228472876466e-07,
      "loss": 311.2014,
      "step": 45190
    },
    {
      "epoch": 0.8693059976344106,
      "grad_norm": 403.5564878829972,
      "learning_rate": 8.476620451878803e-07,
      "loss": 303.7543,
      "step": 45200
    },
    {
      "epoch": 0.8694983219701705,
      "grad_norm": 390.73081632872675,
      "learning_rate": 8.45204652182846e-07,
      "loss": 310.5041,
      "step": 45210
    },
    {
      "epoch": 0.8696906463059303,
      "grad_norm": 397.2903063208496,
      "learning_rate": 8.427506691878806e-07,
      "loss": 311.0627,
      "step": 45220
    },
    {
      "epoch": 0.8698829706416902,
      "grad_norm": 490.0805569838791,
      "learning_rate": 8.403000971170561e-07,
      "loss": 303.6106,
      "step": 45230
    },
    {
      "epoch": 0.87007529497745,
      "grad_norm": 400.1254759972835,
      "learning_rate": 8.378529368831667e-07,
      "loss": 307.972,
      "step": 45240
    },
    {
      "epoch": 0.8702676193132098,
      "grad_norm": 417.7594179144115,
      "learning_rate": 8.354091893977401e-07,
      "loss": 305.1977,
      "step": 45250
    },
    {
      "epoch": 0.8704599436489696,
      "grad_norm": 365.87429074246427,
      "learning_rate": 8.329688555710336e-07,
      "loss": 306.2153,
      "step": 45260
    },
    {
      "epoch": 0.8706522679847295,
      "grad_norm": 389.4945732877371,
      "learning_rate": 8.305319363120279e-07,
      "loss": 314.2629,
      "step": 45270
    },
    {
      "epoch": 0.8708445923204893,
      "grad_norm": 395.0934662531337,
      "learning_rate": 8.280984325284392e-07,
      "loss": 306.9449,
      "step": 45280
    },
    {
      "epoch": 0.871036916656249,
      "grad_norm": 395.28329275841406,
      "learning_rate": 8.256683451267044e-07,
      "loss": 303.7966,
      "step": 45290
    },
    {
      "epoch": 0.8712292409920089,
      "grad_norm": 424.28038815872304,
      "learning_rate": 8.232416750119921e-07,
      "loss": 307.4502,
      "step": 45300
    },
    {
      "epoch": 0.8714215653277687,
      "grad_norm": 416.23726587443855,
      "learning_rate": 8.208184230881966e-07,
      "loss": 307.9138,
      "step": 45310
    },
    {
      "epoch": 0.8716138896635286,
      "grad_norm": 420.02821998004293,
      "learning_rate": 8.183985902579405e-07,
      "loss": 303.6252,
      "step": 45320
    },
    {
      "epoch": 0.8718062139992884,
      "grad_norm": 381.04014653126745,
      "learning_rate": 8.159821774225685e-07,
      "loss": 297.8478,
      "step": 45330
    },
    {
      "epoch": 0.8719985383350483,
      "grad_norm": 414.5565269988919,
      "learning_rate": 8.13569185482157e-07,
      "loss": 302.6021,
      "step": 45340
    },
    {
      "epoch": 0.872190862670808,
      "grad_norm": 383.19898362444815,
      "learning_rate": 8.111596153355061e-07,
      "loss": 299.3064,
      "step": 45350
    },
    {
      "epoch": 0.8723831870065679,
      "grad_norm": 389.4528379518342,
      "learning_rate": 8.08753467880139e-07,
      "loss": 307.9255,
      "step": 45360
    },
    {
      "epoch": 0.8725755113423277,
      "grad_norm": 401.9641240797891,
      "learning_rate": 8.063507440123052e-07,
      "loss": 300.6217,
      "step": 45370
    },
    {
      "epoch": 0.8727678356780876,
      "grad_norm": 446.3358967244388,
      "learning_rate": 8.039514446269836e-07,
      "loss": 313.8597,
      "step": 45380
    },
    {
      "epoch": 0.8729601600138474,
      "grad_norm": 404.7477785466647,
      "learning_rate": 8.015555706178702e-07,
      "loss": 307.3693,
      "step": 45390
    },
    {
      "epoch": 0.8731524843496071,
      "grad_norm": 422.3906376953504,
      "learning_rate": 7.991631228773889e-07,
      "loss": 311.8219,
      "step": 45400
    },
    {
      "epoch": 0.873344808685367,
      "grad_norm": 391.46905166129295,
      "learning_rate": 7.967741022966857e-07,
      "loss": 300.5323,
      "step": 45410
    },
    {
      "epoch": 0.8735371330211268,
      "grad_norm": 402.91008021356276,
      "learning_rate": 7.943885097656356e-07,
      "loss": 297.8026,
      "step": 45420
    },
    {
      "epoch": 0.8737294573568867,
      "grad_norm": 391.83827296192266,
      "learning_rate": 7.920063461728311e-07,
      "loss": 303.6322,
      "step": 45430
    },
    {
      "epoch": 0.8739217816926464,
      "grad_norm": 401.53853187885346,
      "learning_rate": 7.896276124055846e-07,
      "loss": 306.0083,
      "step": 45440
    },
    {
      "epoch": 0.8741141060284063,
      "grad_norm": 404.838459447113,
      "learning_rate": 7.872523093499396e-07,
      "loss": 311.1368,
      "step": 45450
    },
    {
      "epoch": 0.8743064303641661,
      "grad_norm": 409.24741584776075,
      "learning_rate": 7.848804378906561e-07,
      "loss": 303.0447,
      "step": 45460
    },
    {
      "epoch": 0.874498754699926,
      "grad_norm": 365.37622364170477,
      "learning_rate": 7.825119989112173e-07,
      "loss": 297.1866,
      "step": 45470
    },
    {
      "epoch": 0.8746910790356858,
      "grad_norm": 405.4105845532776,
      "learning_rate": 7.801469932938255e-07,
      "loss": 307.8844,
      "step": 45480
    },
    {
      "epoch": 0.8748834033714457,
      "grad_norm": 374.1113914829126,
      "learning_rate": 7.777854219194092e-07,
      "loss": 301.2761,
      "step": 45490
    },
    {
      "epoch": 0.8750757277072054,
      "grad_norm": 396.8369103400041,
      "learning_rate": 7.754272856676126e-07,
      "loss": 311.6533,
      "step": 45500
    },
    {
      "epoch": 0.8752680520429652,
      "grad_norm": 436.8521525047514,
      "learning_rate": 7.73072585416802e-07,
      "loss": 313.9562,
      "step": 45510
    },
    {
      "epoch": 0.8754603763787251,
      "grad_norm": 405.09612251943463,
      "learning_rate": 7.707213220440679e-07,
      "loss": 297.5888,
      "step": 45520
    },
    {
      "epoch": 0.8756527007144849,
      "grad_norm": 391.916365289241,
      "learning_rate": 7.683734964252143e-07,
      "loss": 305.2545,
      "step": 45530
    },
    {
      "epoch": 0.8758450250502448,
      "grad_norm": 425.0453364247075,
      "learning_rate": 7.66029109434766e-07,
      "loss": 312.3987,
      "step": 45540
    },
    {
      "epoch": 0.8760373493860045,
      "grad_norm": 394.3628989612224,
      "learning_rate": 7.636881619459724e-07,
      "loss": 307.1669,
      "step": 45550
    },
    {
      "epoch": 0.8762296737217644,
      "grad_norm": 435.05202901875793,
      "learning_rate": 7.613506548307936e-07,
      "loss": 308.8373,
      "step": 45560
    },
    {
      "epoch": 0.8764219980575242,
      "grad_norm": 415.56905710505737,
      "learning_rate": 7.590165889599166e-07,
      "loss": 305.002,
      "step": 45570
    },
    {
      "epoch": 0.8766143223932841,
      "grad_norm": 376.6664542530555,
      "learning_rate": 7.566859652027381e-07,
      "loss": 302.9801,
      "step": 45580
    },
    {
      "epoch": 0.8768066467290438,
      "grad_norm": 448.8970396402597,
      "learning_rate": 7.543587844273814e-07,
      "loss": 306.5511,
      "step": 45590
    },
    {
      "epoch": 0.8769989710648037,
      "grad_norm": 405.98444269970327,
      "learning_rate": 7.52035047500681e-07,
      "loss": 300.8533,
      "step": 45600
    },
    {
      "epoch": 0.8771912954005635,
      "grad_norm": 432.5742276359327,
      "learning_rate": 7.497147552881901e-07,
      "loss": 308.6317,
      "step": 45610
    },
    {
      "epoch": 0.8773836197363233,
      "grad_norm": 410.1458948491779,
      "learning_rate": 7.473979086541772e-07,
      "loss": 308.2932,
      "step": 45620
    },
    {
      "epoch": 0.8775759440720832,
      "grad_norm": 408.73801461737315,
      "learning_rate": 7.450845084616332e-07,
      "loss": 303.7726,
      "step": 45630
    },
    {
      "epoch": 0.8777682684078429,
      "grad_norm": 390.61162966833723,
      "learning_rate": 7.427745555722598e-07,
      "loss": 315.3282,
      "step": 45640
    },
    {
      "epoch": 0.8779605927436028,
      "grad_norm": 410.695603424447,
      "learning_rate": 7.404680508464767e-07,
      "loss": 310.4774,
      "step": 45650
    },
    {
      "epoch": 0.8781529170793626,
      "grad_norm": 442.99437144817625,
      "learning_rate": 7.381649951434167e-07,
      "loss": 309.927,
      "step": 45660
    },
    {
      "epoch": 0.8783452414151225,
      "grad_norm": 406.4934380191147,
      "learning_rate": 7.358653893209333e-07,
      "loss": 316.9576,
      "step": 45670
    },
    {
      "epoch": 0.8785375657508823,
      "grad_norm": 376.22778044279886,
      "learning_rate": 7.335692342355882e-07,
      "loss": 306.7312,
      "step": 45680
    },
    {
      "epoch": 0.8787298900866422,
      "grad_norm": 376.9568684998429,
      "learning_rate": 7.312765307426662e-07,
      "loss": 300.716,
      "step": 45690
    },
    {
      "epoch": 0.8789222144224019,
      "grad_norm": 407.9861560067712,
      "learning_rate": 7.28987279696155e-07,
      "loss": 309.2125,
      "step": 45700
    },
    {
      "epoch": 0.8791145387581618,
      "grad_norm": 400.3655920357576,
      "learning_rate": 7.267014819487695e-07,
      "loss": 315.2592,
      "step": 45710
    },
    {
      "epoch": 0.8793068630939216,
      "grad_norm": 390.18770948619374,
      "learning_rate": 7.244191383519272e-07,
      "loss": 304.3402,
      "step": 45720
    },
    {
      "epoch": 0.8794991874296814,
      "grad_norm": 438.375235517891,
      "learning_rate": 7.221402497557629e-07,
      "loss": 301.9918,
      "step": 45730
    },
    {
      "epoch": 0.8796915117654412,
      "grad_norm": 396.5705019696801,
      "learning_rate": 7.198648170091294e-07,
      "loss": 299.4299,
      "step": 45740
    },
    {
      "epoch": 0.879883836101201,
      "grad_norm": 419.8434919393224,
      "learning_rate": 7.175928409595844e-07,
      "loss": 307.1673,
      "step": 45750
    },
    {
      "epoch": 0.8800761604369609,
      "grad_norm": 390.2152470925275,
      "learning_rate": 7.153243224534001e-07,
      "loss": 311.8675,
      "step": 45760
    },
    {
      "epoch": 0.8802684847727207,
      "grad_norm": 422.84121762723163,
      "learning_rate": 7.130592623355659e-07,
      "loss": 303.4092,
      "step": 45770
    },
    {
      "epoch": 0.8804608091084806,
      "grad_norm": 377.63453160700215,
      "learning_rate": 7.10797661449778e-07,
      "loss": 306.8756,
      "step": 45780
    },
    {
      "epoch": 0.8806531334442403,
      "grad_norm": 414.79211569357,
      "learning_rate": 7.085395206384449e-07,
      "loss": 307.6578,
      "step": 45790
    },
    {
      "epoch": 0.8808454577800002,
      "grad_norm": 381.620530340443,
      "learning_rate": 7.062848407426859e-07,
      "loss": 302.2135,
      "step": 45800
    },
    {
      "epoch": 0.88103778211576,
      "grad_norm": 397.56799132021905,
      "learning_rate": 7.040336226023336e-07,
      "loss": 299.7477,
      "step": 45810
    },
    {
      "epoch": 0.8812301064515199,
      "grad_norm": 385.3582867290881,
      "learning_rate": 7.017858670559274e-07,
      "loss": 309.8197,
      "step": 45820
    },
    {
      "epoch": 0.8814224307872797,
      "grad_norm": 422.29913694888336,
      "learning_rate": 6.99541574940722e-07,
      "loss": 313.8146,
      "step": 45830
    },
    {
      "epoch": 0.8816147551230394,
      "grad_norm": 395.110243612897,
      "learning_rate": 6.973007470926774e-07,
      "loss": 306.8336,
      "step": 45840
    },
    {
      "epoch": 0.8818070794587993,
      "grad_norm": 393.68255603138243,
      "learning_rate": 6.95063384346466e-07,
      "loss": 308.8002,
      "step": 45850
    },
    {
      "epoch": 0.8819994037945591,
      "grad_norm": 394.6390426831646,
      "learning_rate": 6.92829487535468e-07,
      "loss": 311.4544,
      "step": 45860
    },
    {
      "epoch": 0.882191728130319,
      "grad_norm": 404.18403133658063,
      "learning_rate": 6.905990574917709e-07,
      "loss": 308.3257,
      "step": 45870
    },
    {
      "epoch": 0.8823840524660788,
      "grad_norm": 382.3695178134656,
      "learning_rate": 6.88372095046177e-07,
      "loss": 300.5742,
      "step": 45880
    },
    {
      "epoch": 0.8825763768018386,
      "grad_norm": 403.9944582127581,
      "learning_rate": 6.861486010281915e-07,
      "loss": 314.6948,
      "step": 45890
    },
    {
      "epoch": 0.8827687011375984,
      "grad_norm": 391.0733677084212,
      "learning_rate": 6.839285762660275e-07,
      "loss": 297.6921,
      "step": 45900
    },
    {
      "epoch": 0.8829610254733583,
      "grad_norm": 397.42788057613114,
      "learning_rate": 6.81712021586608e-07,
      "loss": 306.5648,
      "step": 45910
    },
    {
      "epoch": 0.8831533498091181,
      "grad_norm": 382.44111398233173,
      "learning_rate": 6.794989378155659e-07,
      "loss": 309.9393,
      "step": 45920
    },
    {
      "epoch": 0.883345674144878,
      "grad_norm": 396.18882027417163,
      "learning_rate": 6.772893257772361e-07,
      "loss": 313.6461,
      "step": 45930
    },
    {
      "epoch": 0.8835379984806377,
      "grad_norm": 410.79810345289565,
      "learning_rate": 6.750831862946605e-07,
      "loss": 319.9223,
      "step": 45940
    },
    {
      "epoch": 0.8837303228163975,
      "grad_norm": 413.3058841324921,
      "learning_rate": 6.728805201895949e-07,
      "loss": 313.6198,
      "step": 45950
    },
    {
      "epoch": 0.8839226471521574,
      "grad_norm": 393.6352528130019,
      "learning_rate": 6.70681328282492e-07,
      "loss": 290.7339,
      "step": 45960
    },
    {
      "epoch": 0.8841149714879172,
      "grad_norm": 415.7383568075715,
      "learning_rate": 6.684856113925143e-07,
      "loss": 308.1027,
      "step": 45970
    },
    {
      "epoch": 0.8843072958236771,
      "grad_norm": 416.42752137420587,
      "learning_rate": 6.662933703375307e-07,
      "loss": 316.0137,
      "step": 45980
    },
    {
      "epoch": 0.8844996201594368,
      "grad_norm": 405.45697836834063,
      "learning_rate": 6.641046059341171e-07,
      "loss": 297.9852,
      "step": 45990
    },
    {
      "epoch": 0.8846919444951967,
      "grad_norm": 416.18297552140035,
      "learning_rate": 6.619193189975515e-07,
      "loss": 309.7994,
      "step": 46000
    },
    {
      "epoch": 0.8848842688309565,
      "grad_norm": 393.6140314846193,
      "learning_rate": 6.597375103418135e-07,
      "loss": 304.9628,
      "step": 46010
    },
    {
      "epoch": 0.8850765931667164,
      "grad_norm": 380.7352856643751,
      "learning_rate": 6.575591807795944e-07,
      "loss": 294.9117,
      "step": 46020
    },
    {
      "epoch": 0.8852689175024762,
      "grad_norm": 405.5867603469064,
      "learning_rate": 6.553843311222863e-07,
      "loss": 301.155,
      "step": 46030
    },
    {
      "epoch": 0.885461241838236,
      "grad_norm": 382.85821435645323,
      "learning_rate": 6.532129621799832e-07,
      "loss": 307.5864,
      "step": 46040
    },
    {
      "epoch": 0.8856535661739958,
      "grad_norm": 428.7914636275491,
      "learning_rate": 6.510450747614816e-07,
      "loss": 314.8727,
      "step": 46050
    },
    {
      "epoch": 0.8858458905097557,
      "grad_norm": 389.7594958058858,
      "learning_rate": 6.488806696742889e-07,
      "loss": 317.6137,
      "step": 46060
    },
    {
      "epoch": 0.8860382148455155,
      "grad_norm": 394.5978663407577,
      "learning_rate": 6.46719747724609e-07,
      "loss": 322.8763,
      "step": 46070
    },
    {
      "epoch": 0.8862305391812753,
      "grad_norm": 420.1356553267791,
      "learning_rate": 6.44562309717347e-07,
      "loss": 316.5685,
      "step": 46080
    },
    {
      "epoch": 0.8864228635170351,
      "grad_norm": 427.54016973685555,
      "learning_rate": 6.424083564561134e-07,
      "loss": 309.695,
      "step": 46090
    },
    {
      "epoch": 0.8866151878527949,
      "grad_norm": 392.8559528694308,
      "learning_rate": 6.402578887432232e-07,
      "loss": 301.731,
      "step": 46100
    },
    {
      "epoch": 0.8868075121885548,
      "grad_norm": 427.0669161675345,
      "learning_rate": 6.381109073796865e-07,
      "loss": 309.626,
      "step": 46110
    },
    {
      "epoch": 0.8869998365243146,
      "grad_norm": 391.30889067923704,
      "learning_rate": 6.359674131652204e-07,
      "loss": 299.6323,
      "step": 46120
    },
    {
      "epoch": 0.8871921608600745,
      "grad_norm": 442.34950434563586,
      "learning_rate": 6.338274068982408e-07,
      "loss": 309.3837,
      "step": 46130
    },
    {
      "epoch": 0.8873844851958342,
      "grad_norm": 407.20357961255,
      "learning_rate": 6.316908893758656e-07,
      "loss": 305.2207,
      "step": 46140
    },
    {
      "epoch": 0.8875768095315941,
      "grad_norm": 396.3075894298535,
      "learning_rate": 6.295578613939113e-07,
      "loss": 308.2404,
      "step": 46150
    },
    {
      "epoch": 0.8877691338673539,
      "grad_norm": 418.4500811248668,
      "learning_rate": 6.274283237468948e-07,
      "loss": 298.5642,
      "step": 46160
    },
    {
      "epoch": 0.8879614582031138,
      "grad_norm": 408.9619948815748,
      "learning_rate": 6.25302277228036e-07,
      "loss": 311.3323,
      "step": 46170
    },
    {
      "epoch": 0.8881537825388736,
      "grad_norm": 418.6031689027424,
      "learning_rate": 6.231797226292502e-07,
      "loss": 312.8917,
      "step": 46180
    },
    {
      "epoch": 0.8883461068746333,
      "grad_norm": 390.6458889010803,
      "learning_rate": 6.210606607411529e-07,
      "loss": 300.4923,
      "step": 46190
    },
    {
      "epoch": 0.8885384312103932,
      "grad_norm": 396.7457081422147,
      "learning_rate": 6.189450923530627e-07,
      "loss": 300.7349,
      "step": 46200
    },
    {
      "epoch": 0.888730755546153,
      "grad_norm": 395.79390788646,
      "learning_rate": 6.168330182529924e-07,
      "loss": 313.5702,
      "step": 46210
    },
    {
      "epoch": 0.8889230798819129,
      "grad_norm": 381.346933784914,
      "learning_rate": 6.147244392276541e-07,
      "loss": 308.3087,
      "step": 46220
    },
    {
      "epoch": 0.8891154042176727,
      "grad_norm": 374.7443409959245,
      "learning_rate": 6.126193560624583e-07,
      "loss": 311.5375,
      "step": 46230
    },
    {
      "epoch": 0.8893077285534325,
      "grad_norm": 391.1134842833984,
      "learning_rate": 6.105177695415165e-07,
      "loss": 304.2555,
      "step": 46240
    },
    {
      "epoch": 0.8895000528891923,
      "grad_norm": 399.0237920108718,
      "learning_rate": 6.084196804476317e-07,
      "loss": 305.1379,
      "step": 46250
    },
    {
      "epoch": 0.8896923772249522,
      "grad_norm": 387.0806488803156,
      "learning_rate": 6.063250895623096e-07,
      "loss": 302.4917,
      "step": 46260
    },
    {
      "epoch": 0.889884701560712,
      "grad_norm": 459.1373913912829,
      "learning_rate": 6.042339976657486e-07,
      "loss": 316.0654,
      "step": 46270
    },
    {
      "epoch": 0.8900770258964719,
      "grad_norm": 413.1743800559773,
      "learning_rate": 6.021464055368498e-07,
      "loss": 308.2002,
      "step": 46280
    },
    {
      "epoch": 0.8902693502322316,
      "grad_norm": 383.7871833378915,
      "learning_rate": 6.000623139532036e-07,
      "loss": 316.0735,
      "step": 46290
    },
    {
      "epoch": 0.8904616745679914,
      "grad_norm": 385.48950766825516,
      "learning_rate": 5.979817236910979e-07,
      "loss": 307.3239,
      "step": 46300
    },
    {
      "epoch": 0.8906539989037513,
      "grad_norm": 392.92931098911106,
      "learning_rate": 5.959046355255238e-07,
      "loss": 320.3373,
      "step": 46310
    },
    {
      "epoch": 0.8908463232395111,
      "grad_norm": 392.2716459066827,
      "learning_rate": 5.93831050230158e-07,
      "loss": 307.411,
      "step": 46320
    },
    {
      "epoch": 0.891038647575271,
      "grad_norm": 396.56704491173207,
      "learning_rate": 5.917609685773784e-07,
      "loss": 304.5664,
      "step": 46330
    },
    {
      "epoch": 0.8912309719110307,
      "grad_norm": 404.43538476368377,
      "learning_rate": 5.896943913382547e-07,
      "loss": 297.9583,
      "step": 46340
    },
    {
      "epoch": 0.8914232962467906,
      "grad_norm": 363.7866243581948,
      "learning_rate": 5.876313192825544e-07,
      "loss": 300.0678,
      "step": 46350
    },
    {
      "epoch": 0.8916156205825504,
      "grad_norm": 403.3344072931486,
      "learning_rate": 5.855717531787375e-07,
      "loss": 313.7008,
      "step": 46360
    },
    {
      "epoch": 0.8918079449183103,
      "grad_norm": 419.4885340202976,
      "learning_rate": 5.835156937939568e-07,
      "loss": 313.4582,
      "step": 46370
    },
    {
      "epoch": 0.89200026925407,
      "grad_norm": 411.7132493909285,
      "learning_rate": 5.814631418940641e-07,
      "loss": 304.3654,
      "step": 46380
    },
    {
      "epoch": 0.8921925935898299,
      "grad_norm": 394.67088583255133,
      "learning_rate": 5.794140982435981e-07,
      "loss": 305.937,
      "step": 46390
    },
    {
      "epoch": 0.8923849179255897,
      "grad_norm": 394.4623270911905,
      "learning_rate": 5.773685636057924e-07,
      "loss": 308.0931,
      "step": 46400
    },
    {
      "epoch": 0.8925772422613495,
      "grad_norm": 391.743733333567,
      "learning_rate": 5.753265387425777e-07,
      "loss": 318.6562,
      "step": 46410
    },
    {
      "epoch": 0.8927695665971094,
      "grad_norm": 439.9736607301654,
      "learning_rate": 5.732880244145744e-07,
      "loss": 301.2937,
      "step": 46420
    },
    {
      "epoch": 0.8929618909328692,
      "grad_norm": 373.2950884327214,
      "learning_rate": 5.712530213810951e-07,
      "loss": 308.4244,
      "step": 46430
    },
    {
      "epoch": 0.893154215268629,
      "grad_norm": 391.7507865793555,
      "learning_rate": 5.692215304001447e-07,
      "loss": 302.032,
      "step": 46440
    },
    {
      "epoch": 0.8933465396043888,
      "grad_norm": 389.7173195728798,
      "learning_rate": 5.671935522284177e-07,
      "loss": 312.5521,
      "step": 46450
    },
    {
      "epoch": 0.8935388639401487,
      "grad_norm": 412.1229719256818,
      "learning_rate": 5.651690876213067e-07,
      "loss": 314.3553,
      "step": 46460
    },
    {
      "epoch": 0.8937311882759085,
      "grad_norm": 415.673015676758,
      "learning_rate": 5.631481373328895e-07,
      "loss": 317.5739,
      "step": 46470
    },
    {
      "epoch": 0.8939235126116684,
      "grad_norm": 383.5959857472058,
      "learning_rate": 5.61130702115934e-07,
      "loss": 310.8006,
      "step": 46480
    },
    {
      "epoch": 0.8941158369474281,
      "grad_norm": 424.48620059147385,
      "learning_rate": 5.591167827219057e-07,
      "loss": 306.3784,
      "step": 46490
    },
    {
      "epoch": 0.894308161283188,
      "grad_norm": 432.72999307470707,
      "learning_rate": 5.571063799009546e-07,
      "loss": 310.2934,
      "step": 46500
    },
    {
      "epoch": 0.8945004856189478,
      "grad_norm": 385.8921818135607,
      "learning_rate": 5.550994944019216e-07,
      "loss": 309.3458,
      "step": 46510
    },
    {
      "epoch": 0.8946928099547076,
      "grad_norm": 393.38009021685053,
      "learning_rate": 5.53096126972339e-07,
      "loss": 310.8305,
      "step": 46520
    },
    {
      "epoch": 0.8948851342904675,
      "grad_norm": 403.06746748671895,
      "learning_rate": 5.510962783584295e-07,
      "loss": 309.2699,
      "step": 46530
    },
    {
      "epoch": 0.8950774586262272,
      "grad_norm": 418.46543849267914,
      "learning_rate": 5.490999493051008e-07,
      "loss": 304.8145,
      "step": 46540
    },
    {
      "epoch": 0.8952697829619871,
      "grad_norm": 402.55515222979574,
      "learning_rate": 5.471071405559547e-07,
      "loss": 307.2068,
      "step": 46550
    },
    {
      "epoch": 0.8954621072977469,
      "grad_norm": 399.46919542924513,
      "learning_rate": 5.451178528532786e-07,
      "loss": 309.3905,
      "step": 46560
    },
    {
      "epoch": 0.8956544316335068,
      "grad_norm": 393.45511438942214,
      "learning_rate": 5.431320869380519e-07,
      "loss": 295.8229,
      "step": 46570
    },
    {
      "epoch": 0.8958467559692666,
      "grad_norm": 385.21869754053773,
      "learning_rate": 5.411498435499363e-07,
      "loss": 314.3867,
      "step": 46580
    },
    {
      "epoch": 0.8960390803050264,
      "grad_norm": 401.35334405474435,
      "learning_rate": 5.391711234272856e-07,
      "loss": 302.6217,
      "step": 46590
    },
    {
      "epoch": 0.8962314046407862,
      "grad_norm": 371.52936120465677,
      "learning_rate": 5.371959273071414e-07,
      "loss": 309.0508,
      "step": 46600
    },
    {
      "epoch": 0.8964237289765461,
      "grad_norm": 368.6968738196107,
      "learning_rate": 5.352242559252308e-07,
      "loss": 296.9199,
      "step": 46610
    },
    {
      "epoch": 0.8966160533123059,
      "grad_norm": 391.51795468470993,
      "learning_rate": 5.332561100159683e-07,
      "loss": 310.5974,
      "step": 46620
    },
    {
      "epoch": 0.8968083776480656,
      "grad_norm": 404.8906658797306,
      "learning_rate": 5.312914903124566e-07,
      "loss": 303.6626,
      "step": 46630
    },
    {
      "epoch": 0.8970007019838255,
      "grad_norm": 398.56791648787976,
      "learning_rate": 5.293303975464836e-07,
      "loss": 306.0725,
      "step": 46640
    },
    {
      "epoch": 0.8971930263195853,
      "grad_norm": 409.44681130094625,
      "learning_rate": 5.273728324485261e-07,
      "loss": 309.3985,
      "step": 46650
    },
    {
      "epoch": 0.8973853506553452,
      "grad_norm": 404.49078265445223,
      "learning_rate": 5.254187957477397e-07,
      "loss": 321.8518,
      "step": 46660
    },
    {
      "epoch": 0.897577674991105,
      "grad_norm": 399.3909204247682,
      "learning_rate": 5.234682881719766e-07,
      "loss": 310.4104,
      "step": 46670
    },
    {
      "epoch": 0.8977699993268649,
      "grad_norm": 386.29112737800665,
      "learning_rate": 5.215213104477645e-07,
      "loss": 303.7211,
      "step": 46680
    },
    {
      "epoch": 0.8979623236626246,
      "grad_norm": 395.3838043275145,
      "learning_rate": 5.195778633003223e-07,
      "loss": 300.6631,
      "step": 46690
    },
    {
      "epoch": 0.8981546479983845,
      "grad_norm": 383.18928443229163,
      "learning_rate": 5.176379474535509e-07,
      "loss": 301.8354,
      "step": 46700
    },
    {
      "epoch": 0.8983469723341443,
      "grad_norm": 414.57286653272627,
      "learning_rate": 5.1570156363004e-07,
      "loss": 311.391,
      "step": 46710
    },
    {
      "epoch": 0.8985392966699042,
      "grad_norm": 388.16176822708604,
      "learning_rate": 5.13768712551057e-07,
      "loss": 315.8273,
      "step": 46720
    },
    {
      "epoch": 0.898731621005664,
      "grad_norm": 383.8670608578188,
      "learning_rate": 5.118393949365574e-07,
      "loss": 305.7346,
      "step": 46730
    },
    {
      "epoch": 0.8989239453414237,
      "grad_norm": 455.2550830240003,
      "learning_rate": 5.099136115051829e-07,
      "loss": 306.9304,
      "step": 46740
    },
    {
      "epoch": 0.8991162696771836,
      "grad_norm": 401.90538581807203,
      "learning_rate": 5.079913629742539e-07,
      "loss": 308.4706,
      "step": 46750
    },
    {
      "epoch": 0.8993085940129434,
      "grad_norm": 434.08363082390514,
      "learning_rate": 5.060726500597768e-07,
      "loss": 317.7397,
      "step": 46760
    },
    {
      "epoch": 0.8995009183487033,
      "grad_norm": 401.18885328075567,
      "learning_rate": 5.041574734764376e-07,
      "loss": 305.9172,
      "step": 46770
    },
    {
      "epoch": 0.899693242684463,
      "grad_norm": 408.8802718021376,
      "learning_rate": 5.022458339376124e-07,
      "loss": 304.0198,
      "step": 46780
    },
    {
      "epoch": 0.8998855670202229,
      "grad_norm": 409.6269404814617,
      "learning_rate": 5.003377321553538e-07,
      "loss": 312.0058,
      "step": 46790
    },
    {
      "epoch": 0.9000778913559827,
      "grad_norm": 388.8489158257935,
      "learning_rate": 4.984331688403976e-07,
      "loss": 312.4344,
      "step": 46800
    },
    {
      "epoch": 0.9002702156917426,
      "grad_norm": 407.2707005391624,
      "learning_rate": 4.96532144702162e-07,
      "loss": 304.1944,
      "step": 46810
    },
    {
      "epoch": 0.9004625400275024,
      "grad_norm": 397.98932769010725,
      "learning_rate": 4.946346604487462e-07,
      "loss": 321.0849,
      "step": 46820
    },
    {
      "epoch": 0.9006548643632623,
      "grad_norm": 403.93556181436435,
      "learning_rate": 4.927407167869346e-07,
      "loss": 305.4765,
      "step": 46830
    },
    {
      "epoch": 0.900847188699022,
      "grad_norm": 414.8852899985486,
      "learning_rate": 4.908503144221877e-07,
      "loss": 307.014,
      "step": 46840
    },
    {
      "epoch": 0.9010395130347818,
      "grad_norm": 393.6270191712424,
      "learning_rate": 4.889634540586518e-07,
      "loss": 307.8562,
      "step": 46850
    },
    {
      "epoch": 0.9012318373705417,
      "grad_norm": 411.5943234646785,
      "learning_rate": 4.870801363991484e-07,
      "loss": 316.6229,
      "step": 46860
    },
    {
      "epoch": 0.9014241617063015,
      "grad_norm": 392.1474518141912,
      "learning_rate": 4.852003621451829e-07,
      "loss": 301.0697,
      "step": 46870
    },
    {
      "epoch": 0.9016164860420613,
      "grad_norm": 409.64889906283616,
      "learning_rate": 4.833241319969395e-07,
      "loss": 299.2146,
      "step": 46880
    },
    {
      "epoch": 0.9018088103778211,
      "grad_norm": 389.6474963585249,
      "learning_rate": 4.814514466532849e-07,
      "loss": 307.1873,
      "step": 46890
    },
    {
      "epoch": 0.902001134713581,
      "grad_norm": 424.1156366381604,
      "learning_rate": 4.795823068117622e-07,
      "loss": 313.1375,
      "step": 46900
    },
    {
      "epoch": 0.9021934590493408,
      "grad_norm": 403.40640177961683,
      "learning_rate": 4.777167131685945e-07,
      "loss": 300.0229,
      "step": 46910
    },
    {
      "epoch": 0.9023857833851007,
      "grad_norm": 381.20234247532073,
      "learning_rate": 4.7585466641868696e-07,
      "loss": 315.0416,
      "step": 46920
    },
    {
      "epoch": 0.9025781077208604,
      "grad_norm": 388.3331273384083,
      "learning_rate": 4.7399616725561925e-07,
      "loss": 302.3271,
      "step": 46930
    },
    {
      "epoch": 0.9027704320566203,
      "grad_norm": 427.0967529725314,
      "learning_rate": 4.721412163716521e-07,
      "loss": 306.9506,
      "step": 46940
    },
    {
      "epoch": 0.9029627563923801,
      "grad_norm": 422.53308076831155,
      "learning_rate": 4.702898144577228e-07,
      "loss": 320.1698,
      "step": 46950
    },
    {
      "epoch": 0.9031550807281399,
      "grad_norm": 396.06198883312834,
      "learning_rate": 4.6844196220345086e-07,
      "loss": 301.159,
      "step": 46960
    },
    {
      "epoch": 0.9033474050638998,
      "grad_norm": 409.804926935526,
      "learning_rate": 4.665976602971278e-07,
      "loss": 313.4145,
      "step": 46970
    },
    {
      "epoch": 0.9035397293996595,
      "grad_norm": 387.72794018017765,
      "learning_rate": 4.647569094257276e-07,
      "loss": 315.9578,
      "step": 46980
    },
    {
      "epoch": 0.9037320537354194,
      "grad_norm": 383.42527174430387,
      "learning_rate": 4.629197102748984e-07,
      "loss": 299.4948,
      "step": 46990
    },
    {
      "epoch": 0.9039243780711792,
      "grad_norm": 407.20465709677194,
      "learning_rate": 4.610860635289671e-07,
      "loss": 308.1644,
      "step": 47000
    },
    {
      "epoch": 0.9041167024069391,
      "grad_norm": 392.6398786002841,
      "learning_rate": 4.592559698709387e-07,
      "loss": 306.0993,
      "step": 47010
    },
    {
      "epoch": 0.9043090267426989,
      "grad_norm": 386.98735841037086,
      "learning_rate": 4.5742942998248774e-07,
      "loss": 305.9105,
      "step": 47020
    },
    {
      "epoch": 0.9045013510784587,
      "grad_norm": 401.9132128681427,
      "learning_rate": 4.5560644454397563e-07,
      "loss": 307.4984,
      "step": 47030
    },
    {
      "epoch": 0.9046936754142185,
      "grad_norm": 390.8832774728099,
      "learning_rate": 4.537870142344314e-07,
      "loss": 308.8956,
      "step": 47040
    },
    {
      "epoch": 0.9048859997499784,
      "grad_norm": 388.3048246783133,
      "learning_rate": 4.5197113973156403e-07,
      "loss": 304.6942,
      "step": 47050
    },
    {
      "epoch": 0.9050783240857382,
      "grad_norm": 412.44373128855847,
      "learning_rate": 4.5015882171175476e-07,
      "loss": 313.6644,
      "step": 47060
    },
    {
      "epoch": 0.905270648421498,
      "grad_norm": 464.0814530708157,
      "learning_rate": 4.483500608500657e-07,
      "loss": 304.5468,
      "step": 47070
    },
    {
      "epoch": 0.9054629727572578,
      "grad_norm": 424.29406018528863,
      "learning_rate": 4.4654485782022697e-07,
      "loss": 306.5049,
      "step": 47080
    },
    {
      "epoch": 0.9056552970930176,
      "grad_norm": 413.4845756821773,
      "learning_rate": 4.447432132946472e-07,
      "loss": 313.5461,
      "step": 47090
    },
    {
      "epoch": 0.9058476214287775,
      "grad_norm": 429.2563096228966,
      "learning_rate": 4.429451279444119e-07,
      "loss": 308.5096,
      "step": 47100
    },
    {
      "epoch": 0.9060399457645373,
      "grad_norm": 396.6540074936668,
      "learning_rate": 4.411506024392753e-07,
      "loss": 312.3194,
      "step": 47110
    },
    {
      "epoch": 0.9062322701002972,
      "grad_norm": 445.98581476751997,
      "learning_rate": 4.393596374476705e-07,
      "loss": 299.0574,
      "step": 47120
    },
    {
      "epoch": 0.9064245944360569,
      "grad_norm": 377.0372439405971,
      "learning_rate": 4.3757223363670055e-07,
      "loss": 297.6503,
      "step": 47130
    },
    {
      "epoch": 0.9066169187718168,
      "grad_norm": 382.8075243290351,
      "learning_rate": 4.3578839167214505e-07,
      "loss": 312.4722,
      "step": 47140
    },
    {
      "epoch": 0.9068092431075766,
      "grad_norm": 372.54105592565537,
      "learning_rate": 4.3400811221845693e-07,
      "loss": 311.4449,
      "step": 47150
    },
    {
      "epoch": 0.9070015674433365,
      "grad_norm": 380.14881384319904,
      "learning_rate": 4.322313959387592e-07,
      "loss": 306.3151,
      "step": 47160
    },
    {
      "epoch": 0.9071938917790963,
      "grad_norm": 377.9214585066291,
      "learning_rate": 4.304582434948479e-07,
      "loss": 306.5595,
      "step": 47170
    },
    {
      "epoch": 0.9073862161148561,
      "grad_norm": 363.7041959721657,
      "learning_rate": 4.2868865554719583e-07,
      "loss": 308.2623,
      "step": 47180
    },
    {
      "epoch": 0.9075785404506159,
      "grad_norm": 359.8307204599658,
      "learning_rate": 4.269226327549447e-07,
      "loss": 310.797,
      "step": 47190
    },
    {
      "epoch": 0.9077708647863757,
      "grad_norm": 392.42271178670643,
      "learning_rate": 4.251601757759061e-07,
      "loss": 297.5345,
      "step": 47200
    },
    {
      "epoch": 0.9079631891221356,
      "grad_norm": 410.47760106090544,
      "learning_rate": 4.234012852665703e-07,
      "loss": 314.4087,
      "step": 47210
    },
    {
      "epoch": 0.9081555134578954,
      "grad_norm": 412.60923116712473,
      "learning_rate": 4.2164596188209226e-07,
      "loss": 312.397,
      "step": 47220
    },
    {
      "epoch": 0.9083478377936552,
      "grad_norm": 441.0582246059681,
      "learning_rate": 4.198942062763023e-07,
      "loss": 311.3601,
      "step": 47230
    },
    {
      "epoch": 0.908540162129415,
      "grad_norm": 412.8262065674468,
      "learning_rate": 4.181460191016984e-07,
      "loss": 303.7134,
      "step": 47240
    },
    {
      "epoch": 0.9087324864651749,
      "grad_norm": 385.8560553277431,
      "learning_rate": 4.1640140100945304e-07,
      "loss": 313.7107,
      "step": 47250
    },
    {
      "epoch": 0.9089248108009347,
      "grad_norm": 434.44396051005725,
      "learning_rate": 4.146603526494086e-07,
      "loss": 305.7453,
      "step": 47260
    },
    {
      "epoch": 0.9091171351366946,
      "grad_norm": 433.1057345524686,
      "learning_rate": 4.129228746700742e-07,
      "loss": 313.6865,
      "step": 47270
    },
    {
      "epoch": 0.9093094594724543,
      "grad_norm": 456.50581322726777,
      "learning_rate": 4.111889677186354e-07,
      "loss": 306.2014,
      "step": 47280
    },
    {
      "epoch": 0.9095017838082142,
      "grad_norm": 414.5498867449801,
      "learning_rate": 4.094586324409411e-07,
      "loss": 303.7974,
      "step": 47290
    },
    {
      "epoch": 0.909694108143974,
      "grad_norm": 377.9031480509657,
      "learning_rate": 4.0773186948151246e-07,
      "loss": 311.4135,
      "step": 47300
    },
    {
      "epoch": 0.9098864324797338,
      "grad_norm": 454.26882143084265,
      "learning_rate": 4.060086794835405e-07,
      "loss": 303.403,
      "step": 47310
    },
    {
      "epoch": 0.9100787568154937,
      "grad_norm": 391.93242191769303,
      "learning_rate": 4.042890630888863e-07,
      "loss": 304.2959,
      "step": 47320
    },
    {
      "epoch": 0.9102710811512534,
      "grad_norm": 372.6859946321374,
      "learning_rate": 4.025730209380774e-07,
      "loss": 296.9763,
      "step": 47330
    },
    {
      "epoch": 0.9104634054870133,
      "grad_norm": 416.9777548055224,
      "learning_rate": 4.0086055367031027e-07,
      "loss": 312.5712,
      "step": 47340
    },
    {
      "epoch": 0.9106557298227731,
      "grad_norm": 373.3456540045296,
      "learning_rate": 3.9915166192345365e-07,
      "loss": 306.0556,
      "step": 47350
    },
    {
      "epoch": 0.910848054158533,
      "grad_norm": 400.7407150167145,
      "learning_rate": 3.9744634633403944e-07,
      "loss": 303.4901,
      "step": 47360
    },
    {
      "epoch": 0.9110403784942928,
      "grad_norm": 394.11532067744247,
      "learning_rate": 3.957446075372706e-07,
      "loss": 298.7103,
      "step": 47370
    },
    {
      "epoch": 0.9112327028300526,
      "grad_norm": 404.2754907300138,
      "learning_rate": 3.940464461670135e-07,
      "loss": 312.5205,
      "step": 47380
    },
    {
      "epoch": 0.9114250271658124,
      "grad_norm": 411.70067068935276,
      "learning_rate": 3.923518628558087e-07,
      "loss": 304.3337,
      "step": 47390
    },
    {
      "epoch": 0.9116173515015723,
      "grad_norm": 385.19714376097517,
      "learning_rate": 3.9066085823485923e-07,
      "loss": 308.4251,
      "step": 47400
    },
    {
      "epoch": 0.9118096758373321,
      "grad_norm": 391.45339507284046,
      "learning_rate": 3.8897343293403777e-07,
      "loss": 314.9632,
      "step": 47410
    },
    {
      "epoch": 0.9120020001730919,
      "grad_norm": 394.73197519988577,
      "learning_rate": 3.872895875818794e-07,
      "loss": 305.2058,
      "step": 47420
    },
    {
      "epoch": 0.9121943245088517,
      "grad_norm": 419.10308006123546,
      "learning_rate": 3.856093228055924e-07,
      "loss": 302.65,
      "step": 47430
    },
    {
      "epoch": 0.9123866488446115,
      "grad_norm": 408.8303017265764,
      "learning_rate": 3.83932639231045e-07,
      "loss": 310.5522,
      "step": 47440
    },
    {
      "epoch": 0.9125789731803714,
      "grad_norm": 416.2099356486548,
      "learning_rate": 3.822595374827742e-07,
      "loss": 303.4325,
      "step": 47450
    },
    {
      "epoch": 0.9127712975161312,
      "grad_norm": 391.5307407301064,
      "learning_rate": 3.805900181839839e-07,
      "loss": 310.8053,
      "step": 47460
    },
    {
      "epoch": 0.9129636218518911,
      "grad_norm": 391.2676107436919,
      "learning_rate": 3.789240819565432e-07,
      "loss": 302.8504,
      "step": 47470
    },
    {
      "epoch": 0.9131559461876508,
      "grad_norm": 385.40043254337735,
      "learning_rate": 3.772617294209835e-07,
      "loss": 306.8966,
      "step": 47480
    },
    {
      "epoch": 0.9133482705234107,
      "grad_norm": 397.9634200225289,
      "learning_rate": 3.7560296119650396e-07,
      "loss": 307.583,
      "step": 47490
    },
    {
      "epoch": 0.9135405948591705,
      "grad_norm": 427.62903183076037,
      "learning_rate": 3.739477779009704e-07,
      "loss": 319.3813,
      "step": 47500
    },
    {
      "epoch": 0.9137329191949304,
      "grad_norm": 395.64708827553017,
      "learning_rate": 3.7229618015091065e-07,
      "loss": 311.9443,
      "step": 47510
    },
    {
      "epoch": 0.9139252435306902,
      "grad_norm": 369.4565438294151,
      "learning_rate": 3.7064816856151484e-07,
      "loss": 303.8605,
      "step": 47520
    },
    {
      "epoch": 0.9141175678664499,
      "grad_norm": 454.6926552513965,
      "learning_rate": 3.6900374374664425e-07,
      "loss": 314.5326,
      "step": 47530
    },
    {
      "epoch": 0.9143098922022098,
      "grad_norm": 391.4264203251334,
      "learning_rate": 3.6736290631881667e-07,
      "loss": 310.9554,
      "step": 47540
    },
    {
      "epoch": 0.9145022165379696,
      "grad_norm": 387.13696803554063,
      "learning_rate": 3.657256568892187e-07,
      "loss": 312.8435,
      "step": 47550
    },
    {
      "epoch": 0.9146945408737295,
      "grad_norm": 420.61641793334394,
      "learning_rate": 3.6409199606769806e-07,
      "loss": 297.8118,
      "step": 47560
    },
    {
      "epoch": 0.9148868652094893,
      "grad_norm": 386.45301945373654,
      "learning_rate": 3.6246192446276694e-07,
      "loss": 302.9943,
      "step": 47570
    },
    {
      "epoch": 0.9150791895452491,
      "grad_norm": 427.13290626235886,
      "learning_rate": 3.6083544268160077e-07,
      "loss": 310.9614,
      "step": 47580
    },
    {
      "epoch": 0.9152715138810089,
      "grad_norm": 397.3602625014417,
      "learning_rate": 3.5921255133003483e-07,
      "loss": 305.3927,
      "step": 47590
    },
    {
      "epoch": 0.9154638382167688,
      "grad_norm": 413.5541572634133,
      "learning_rate": 3.5759325101257013e-07,
      "loss": 309.5832,
      "step": 47600
    },
    {
      "epoch": 0.9156561625525286,
      "grad_norm": 417.4621322747239,
      "learning_rate": 3.559775423323708e-07,
      "loss": 305.1015,
      "step": 47610
    },
    {
      "epoch": 0.9158484868882885,
      "grad_norm": 405.6204122728192,
      "learning_rate": 3.54365425891261e-07,
      "loss": 300.9579,
      "step": 47620
    },
    {
      "epoch": 0.9160408112240482,
      "grad_norm": 391.38108303414936,
      "learning_rate": 3.527569022897259e-07,
      "loss": 316.3043,
      "step": 47630
    },
    {
      "epoch": 0.916233135559808,
      "grad_norm": 395.2986918578219,
      "learning_rate": 3.511519721269163e-07,
      "loss": 310.4572,
      "step": 47640
    },
    {
      "epoch": 0.9164254598955679,
      "grad_norm": 404.9160792814242,
      "learning_rate": 3.4955063600064177e-07,
      "loss": 300.7953,
      "step": 47650
    },
    {
      "epoch": 0.9166177842313277,
      "grad_norm": 397.29818292525414,
      "learning_rate": 3.479528945073707e-07,
      "loss": 305.4276,
      "step": 47660
    },
    {
      "epoch": 0.9168101085670876,
      "grad_norm": 410.26718070099497,
      "learning_rate": 3.4635874824223924e-07,
      "loss": 309.4091,
      "step": 47670
    },
    {
      "epoch": 0.9170024329028473,
      "grad_norm": 478.90730364080486,
      "learning_rate": 3.4476819779903694e-07,
      "loss": 305.3964,
      "step": 47680
    },
    {
      "epoch": 0.9171947572386072,
      "grad_norm": 434.3395458451395,
      "learning_rate": 3.43181243770222e-07,
      "loss": 303.0155,
      "step": 47690
    },
    {
      "epoch": 0.917387081574367,
      "grad_norm": 426.04525326116374,
      "learning_rate": 3.4159788674690386e-07,
      "loss": 302.1621,
      "step": 47700
    },
    {
      "epoch": 0.9175794059101269,
      "grad_norm": 385.0049710114499,
      "learning_rate": 3.4001812731886077e-07,
      "loss": 297.9396,
      "step": 47710
    },
    {
      "epoch": 0.9177717302458867,
      "grad_norm": 376.6403131061554,
      "learning_rate": 3.384419660745253e-07,
      "loss": 300.3971,
      "step": 47720
    },
    {
      "epoch": 0.9179640545816465,
      "grad_norm": 420.8032008218724,
      "learning_rate": 3.368694036009923e-07,
      "loss": 311.8968,
      "step": 47730
    },
    {
      "epoch": 0.9181563789174063,
      "grad_norm": 399.704339037201,
      "learning_rate": 3.353004404840121e-07,
      "loss": 296.0103,
      "step": 47740
    },
    {
      "epoch": 0.9183487032531661,
      "grad_norm": 427.13907898273413,
      "learning_rate": 3.3373507730800167e-07,
      "loss": 319.8219,
      "step": 47750
    },
    {
      "epoch": 0.918541027588926,
      "grad_norm": 434.9184272947517,
      "learning_rate": 3.321733146560324e-07,
      "loss": 306.2606,
      "step": 47760
    },
    {
      "epoch": 0.9187333519246857,
      "grad_norm": 427.44417930091885,
      "learning_rate": 3.306151531098323e-07,
      "loss": 312.6792,
      "step": 47770
    },
    {
      "epoch": 0.9189256762604456,
      "grad_norm": 416.03131125671723,
      "learning_rate": 3.2906059324979255e-07,
      "loss": 306.7176,
      "step": 47780
    },
    {
      "epoch": 0.9191180005962054,
      "grad_norm": 459.21913196223795,
      "learning_rate": 3.275096356549612e-07,
      "loss": 313.7354,
      "step": 47790
    },
    {
      "epoch": 0.9193103249319653,
      "grad_norm": 388.8252851182349,
      "learning_rate": 3.2596228090304496e-07,
      "loss": 296.0648,
      "step": 47800
    },
    {
      "epoch": 0.9195026492677251,
      "grad_norm": 389.0709254200922,
      "learning_rate": 3.2441852957040607e-07,
      "loss": 306.8889,
      "step": 47810
    },
    {
      "epoch": 0.919694973603485,
      "grad_norm": 363.8965786606022,
      "learning_rate": 3.228783822320669e-07,
      "loss": 306.5515,
      "step": 47820
    },
    {
      "epoch": 0.9198872979392447,
      "grad_norm": 373.08081727882825,
      "learning_rate": 3.213418394617085e-07,
      "loss": 307.475,
      "step": 47830
    },
    {
      "epoch": 0.9200796222750046,
      "grad_norm": 404.33700687831754,
      "learning_rate": 3.1980890183166633e-07,
      "loss": 305.8507,
      "step": 47840
    },
    {
      "epoch": 0.9202719466107644,
      "grad_norm": 399.776064460427,
      "learning_rate": 3.1827956991293374e-07,
      "loss": 307.8798,
      "step": 47850
    },
    {
      "epoch": 0.9204642709465242,
      "grad_norm": 390.0528466674345,
      "learning_rate": 3.167538442751639e-07,
      "loss": 309.2005,
      "step": 47860
    },
    {
      "epoch": 0.920656595282284,
      "grad_norm": 393.19543182039837,
      "learning_rate": 3.1523172548666215e-07,
      "loss": 304.9871,
      "step": 47870
    },
    {
      "epoch": 0.9208489196180438,
      "grad_norm": 397.9231074449624,
      "learning_rate": 3.1371321411439284e-07,
      "loss": 311.978,
      "step": 47880
    },
    {
      "epoch": 0.9210412439538037,
      "grad_norm": 399.7454165452534,
      "learning_rate": 3.1219831072397787e-07,
      "loss": 308.1011,
      "step": 47890
    },
    {
      "epoch": 0.9212335682895635,
      "grad_norm": 366.566005147306,
      "learning_rate": 3.1068701587969375e-07,
      "loss": 313.1039,
      "step": 47900
    },
    {
      "epoch": 0.9214258926253234,
      "grad_norm": 426.1127708948236,
      "learning_rate": 3.091793301444701e-07,
      "loss": 308.1743,
      "step": 47910
    },
    {
      "epoch": 0.9216182169610831,
      "grad_norm": 410.33449062514165,
      "learning_rate": 3.076752540798977e-07,
      "loss": 309.3588,
      "step": 47920
    },
    {
      "epoch": 0.921810541296843,
      "grad_norm": 395.76997745563017,
      "learning_rate": 3.061747882462185e-07,
      "loss": 310.8947,
      "step": 47930
    },
    {
      "epoch": 0.9220028656326028,
      "grad_norm": 384.98688699766063,
      "learning_rate": 3.0467793320233306e-07,
      "loss": 307.1477,
      "step": 47940
    },
    {
      "epoch": 0.9221951899683627,
      "grad_norm": 374.34915834955206,
      "learning_rate": 3.031846895057922e-07,
      "loss": 311.2257,
      "step": 47950
    },
    {
      "epoch": 0.9223875143041225,
      "grad_norm": 407.39256496700875,
      "learning_rate": 3.0169505771280747e-07,
      "loss": 306.5492,
      "step": 47960
    },
    {
      "epoch": 0.9225798386398822,
      "grad_norm": 392.1264654575426,
      "learning_rate": 3.002090383782408e-07,
      "loss": 302.2202,
      "step": 47970
    },
    {
      "epoch": 0.9227721629756421,
      "grad_norm": 421.66941636906466,
      "learning_rate": 2.9872663205561035e-07,
      "loss": 303.7292,
      "step": 47980
    },
    {
      "epoch": 0.9229644873114019,
      "grad_norm": 433.9303991403661,
      "learning_rate": 2.972478392970857e-07,
      "loss": 312.4748,
      "step": 47990
    },
    {
      "epoch": 0.9231568116471618,
      "grad_norm": 427.746277049765,
      "learning_rate": 2.9577266065349716e-07,
      "loss": 304.2794,
      "step": 48000
    },
    {
      "epoch": 0.9233491359829216,
      "grad_norm": 391.11532197073797,
      "learning_rate": 2.9430109667432096e-07,
      "loss": 305.4531,
      "step": 48010
    },
    {
      "epoch": 0.9235414603186815,
      "grad_norm": 434.51988344651664,
      "learning_rate": 2.9283314790769177e-07,
      "loss": 314.5872,
      "step": 48020
    },
    {
      "epoch": 0.9237337846544412,
      "grad_norm": 393.49782163556256,
      "learning_rate": 2.913688149003946e-07,
      "loss": 301.1704,
      "step": 48030
    },
    {
      "epoch": 0.9239261089902011,
      "grad_norm": 404.44809273378036,
      "learning_rate": 2.899080981978719e-07,
      "loss": 312.3442,
      "step": 48040
    },
    {
      "epoch": 0.9241184333259609,
      "grad_norm": 387.5157200282338,
      "learning_rate": 2.8845099834421517e-07,
      "loss": 301.7772,
      "step": 48050
    },
    {
      "epoch": 0.9243107576617208,
      "grad_norm": 411.80442132545346,
      "learning_rate": 2.869975158821681e-07,
      "loss": 304.2401,
      "step": 48060
    },
    {
      "epoch": 0.9245030819974805,
      "grad_norm": 418.6088207425718,
      "learning_rate": 2.8554765135313303e-07,
      "loss": 314.1443,
      "step": 48070
    },
    {
      "epoch": 0.9246954063332403,
      "grad_norm": 373.7045028644747,
      "learning_rate": 2.8410140529715803e-07,
      "loss": 298.7632,
      "step": 48080
    },
    {
      "epoch": 0.9248877306690002,
      "grad_norm": 419.7575110753599,
      "learning_rate": 2.826587782529444e-07,
      "loss": 296.541,
      "step": 48090
    },
    {
      "epoch": 0.92508005500476,
      "grad_norm": 415.29859382412235,
      "learning_rate": 2.812197707578501e-07,
      "loss": 311.2337,
      "step": 48100
    },
    {
      "epoch": 0.9252723793405199,
      "grad_norm": 418.7754859089795,
      "learning_rate": 2.797843833478797e-07,
      "loss": 306.2839,
      "step": 48110
    },
    {
      "epoch": 0.9254647036762796,
      "grad_norm": 416.093994573165,
      "learning_rate": 2.7835261655769217e-07,
      "loss": 300.157,
      "step": 48120
    },
    {
      "epoch": 0.9256570280120395,
      "grad_norm": 407.7696217736816,
      "learning_rate": 2.769244709205976e-07,
      "loss": 305.1632,
      "step": 48130
    },
    {
      "epoch": 0.9258493523477993,
      "grad_norm": 386.8632437854869,
      "learning_rate": 2.7549994696855376e-07,
      "loss": 299.6427,
      "step": 48140
    },
    {
      "epoch": 0.9260416766835592,
      "grad_norm": 388.3744927833977,
      "learning_rate": 2.740790452321751e-07,
      "loss": 298.3499,
      "step": 48150
    },
    {
      "epoch": 0.926234001019319,
      "grad_norm": 380.1212958145323,
      "learning_rate": 2.726617662407238e-07,
      "loss": 299.283,
      "step": 48160
    },
    {
      "epoch": 0.9264263253550789,
      "grad_norm": 404.928903643814,
      "learning_rate": 2.7124811052211097e-07,
      "loss": 311.721,
      "step": 48170
    },
    {
      "epoch": 0.9266186496908386,
      "grad_norm": 408.78550540664503,
      "learning_rate": 2.698380786029031e-07,
      "loss": 303.4837,
      "step": 48180
    },
    {
      "epoch": 0.9268109740265985,
      "grad_norm": 416.6285787389231,
      "learning_rate": 2.6843167100831125e-07,
      "loss": 306.3239,
      "step": 48190
    },
    {
      "epoch": 0.9270032983623583,
      "grad_norm": 399.5713125741659,
      "learning_rate": 2.6702888826219965e-07,
      "loss": 310.2678,
      "step": 48200
    },
    {
      "epoch": 0.9271956226981181,
      "grad_norm": 404.0405697824927,
      "learning_rate": 2.6562973088708146e-07,
      "loss": 303.4187,
      "step": 48210
    },
    {
      "epoch": 0.927387947033878,
      "grad_norm": 395.0537069860546,
      "learning_rate": 2.6423419940412086e-07,
      "loss": 301.2331,
      "step": 48220
    },
    {
      "epoch": 0.9275802713696377,
      "grad_norm": 398.3143279777804,
      "learning_rate": 2.628422943331288e-07,
      "loss": 310.1888,
      "step": 48230
    },
    {
      "epoch": 0.9277725957053976,
      "grad_norm": 382.9889266888244,
      "learning_rate": 2.614540161925683e-07,
      "loss": 312.7567,
      "step": 48240
    },
    {
      "epoch": 0.9279649200411574,
      "grad_norm": 394.12458619812935,
      "learning_rate": 2.6006936549954784e-07,
      "loss": 304.4813,
      "step": 48250
    },
    {
      "epoch": 0.9281572443769173,
      "grad_norm": 372.88057147235764,
      "learning_rate": 2.5868834276983057e-07,
      "loss": 310.6033,
      "step": 48260
    },
    {
      "epoch": 0.928349568712677,
      "grad_norm": 379.9207204739021,
      "learning_rate": 2.573109485178216e-07,
      "loss": 314.7656,
      "step": 48270
    },
    {
      "epoch": 0.9285418930484369,
      "grad_norm": 387.6038332399289,
      "learning_rate": 2.5593718325657713e-07,
      "loss": 313.9663,
      "step": 48280
    },
    {
      "epoch": 0.9287342173841967,
      "grad_norm": 432.3752052009221,
      "learning_rate": 2.545670474978057e-07,
      "loss": 298.0133,
      "step": 48290
    },
    {
      "epoch": 0.9289265417199566,
      "grad_norm": 400.46269006345216,
      "learning_rate": 2.532005417518568e-07,
      "loss": 301.4644,
      "step": 48300
    },
    {
      "epoch": 0.9291188660557164,
      "grad_norm": 396.8081183846114,
      "learning_rate": 2.5183766652773336e-07,
      "loss": 301.6045,
      "step": 48310
    },
    {
      "epoch": 0.9293111903914761,
      "grad_norm": 398.644412032208,
      "learning_rate": 2.504784223330814e-07,
      "loss": 303.1359,
      "step": 48320
    },
    {
      "epoch": 0.929503514727236,
      "grad_norm": 425.3370328785918,
      "learning_rate": 2.4912280967419934e-07,
      "loss": 309.1978,
      "step": 48330
    },
    {
      "epoch": 0.9296958390629958,
      "grad_norm": 408.3348327436733,
      "learning_rate": 2.477708290560299e-07,
      "loss": 317.0573,
      "step": 48340
    },
    {
      "epoch": 0.9298881633987557,
      "grad_norm": 390.7232295085053,
      "learning_rate": 2.464224809821614e-07,
      "loss": 317.6602,
      "step": 48350
    },
    {
      "epoch": 0.9300804877345155,
      "grad_norm": 400.2972888701405,
      "learning_rate": 2.450777659548353e-07,
      "loss": 310.7288,
      "step": 48360
    },
    {
      "epoch": 0.9302728120702753,
      "grad_norm": 417.79385169680745,
      "learning_rate": 2.4373668447493225e-07,
      "loss": 311.584,
      "step": 48370
    },
    {
      "epoch": 0.9304651364060351,
      "grad_norm": 359.57601381210014,
      "learning_rate": 2.4239923704198476e-07,
      "loss": 303.2301,
      "step": 48380
    },
    {
      "epoch": 0.930657460741795,
      "grad_norm": 385.84907036685,
      "learning_rate": 2.410654241541688e-07,
      "loss": 302.2753,
      "step": 48390
    },
    {
      "epoch": 0.9308497850775548,
      "grad_norm": 398.40108298436763,
      "learning_rate": 2.3973524630830804e-07,
      "loss": 306.8883,
      "step": 48400
    },
    {
      "epoch": 0.9310421094133147,
      "grad_norm": 401.04926996355346,
      "learning_rate": 2.3840870399987283e-07,
      "loss": 311.6254,
      "step": 48410
    },
    {
      "epoch": 0.9312344337490744,
      "grad_norm": 375.93244778795986,
      "learning_rate": 2.370857977229768e-07,
      "loss": 311.1448,
      "step": 48420
    },
    {
      "epoch": 0.9314267580848342,
      "grad_norm": 384.1908223565569,
      "learning_rate": 2.3576652797038247e-07,
      "loss": 309.7895,
      "step": 48430
    },
    {
      "epoch": 0.9316190824205941,
      "grad_norm": 392.4629389107022,
      "learning_rate": 2.344508952334934e-07,
      "loss": 304.0092,
      "step": 48440
    },
    {
      "epoch": 0.9318114067563539,
      "grad_norm": 391.3649359115568,
      "learning_rate": 2.3313890000236316e-07,
      "loss": 299.921,
      "step": 48450
    },
    {
      "epoch": 0.9320037310921138,
      "grad_norm": 388.21067740670526,
      "learning_rate": 2.3183054276568752e-07,
      "loss": 305.5152,
      "step": 48460
    },
    {
      "epoch": 0.9321960554278735,
      "grad_norm": 393.65427404561126,
      "learning_rate": 2.305258240108077e-07,
      "loss": 301.6203,
      "step": 48470
    },
    {
      "epoch": 0.9323883797636334,
      "grad_norm": 405.66105792409974,
      "learning_rate": 2.2922474422371166e-07,
      "loss": 306.842,
      "step": 48480
    },
    {
      "epoch": 0.9325807040993932,
      "grad_norm": 410.0883644699099,
      "learning_rate": 2.279273038890273e-07,
      "loss": 310.1019,
      "step": 48490
    },
    {
      "epoch": 0.9327730284351531,
      "grad_norm": 406.1702012289618,
      "learning_rate": 2.2663350349003134e-07,
      "loss": 306.3602,
      "step": 48500
    },
    {
      "epoch": 0.9329653527709129,
      "grad_norm": 386.66094171385373,
      "learning_rate": 2.2534334350864274e-07,
      "loss": 298.1951,
      "step": 48510
    },
    {
      "epoch": 0.9331576771066727,
      "grad_norm": 709.1092518689194,
      "learning_rate": 2.2405682442542487e-07,
      "loss": 308.5916,
      "step": 48520
    },
    {
      "epoch": 0.9333500014424325,
      "grad_norm": 372.2512476546447,
      "learning_rate": 2.2277394671958442e-07,
      "loss": 307.1227,
      "step": 48530
    },
    {
      "epoch": 0.9335423257781923,
      "grad_norm": 397.79894090076215,
      "learning_rate": 2.2149471086897355e-07,
      "loss": 306.5278,
      "step": 48540
    },
    {
      "epoch": 0.9337346501139522,
      "grad_norm": 459.51028813612584,
      "learning_rate": 2.202191173500845e-07,
      "loss": 312.0135,
      "step": 48550
    },
    {
      "epoch": 0.933926974449712,
      "grad_norm": 417.684765613212,
      "learning_rate": 2.1894716663805716e-07,
      "loss": 310.5669,
      "step": 48560
    },
    {
      "epoch": 0.9341192987854718,
      "grad_norm": 401.319909761339,
      "learning_rate": 2.176788592066692e-07,
      "loss": 319.8216,
      "step": 48570
    },
    {
      "epoch": 0.9343116231212316,
      "grad_norm": 412.5158792404211,
      "learning_rate": 2.164141955283472e-07,
      "loss": 300.1668,
      "step": 48580
    },
    {
      "epoch": 0.9345039474569915,
      "grad_norm": 384.07131377349486,
      "learning_rate": 2.1515317607415654e-07,
      "loss": 305.6861,
      "step": 48590
    },
    {
      "epoch": 0.9346962717927513,
      "grad_norm": 395.1357619209503,
      "learning_rate": 2.1389580131380373e-07,
      "loss": 326.2871,
      "step": 48600
    },
    {
      "epoch": 0.9348885961285112,
      "grad_norm": 424.3619104491749,
      "learning_rate": 2.126420717156441e-07,
      "loss": 320.5873,
      "step": 48610
    },
    {
      "epoch": 0.9350809204642709,
      "grad_norm": 416.9206505964984,
      "learning_rate": 2.113919877466686e-07,
      "loss": 301.9237,
      "step": 48620
    },
    {
      "epoch": 0.9352732448000308,
      "grad_norm": 451.18930335852923,
      "learning_rate": 2.1014554987251356e-07,
      "loss": 314.8972,
      "step": 48630
    },
    {
      "epoch": 0.9354655691357906,
      "grad_norm": 389.0993931771178,
      "learning_rate": 2.0890275855745546e-07,
      "loss": 304.6365,
      "step": 48640
    },
    {
      "epoch": 0.9356578934715504,
      "grad_norm": 394.0023278736881,
      "learning_rate": 2.0766361426441505e-07,
      "loss": 302.1834,
      "step": 48650
    },
    {
      "epoch": 0.9358502178073103,
      "grad_norm": 374.53596992770554,
      "learning_rate": 2.0642811745495206e-07,
      "loss": 314.5554,
      "step": 48660
    },
    {
      "epoch": 0.93604254214307,
      "grad_norm": 418.4275970323863,
      "learning_rate": 2.0519626858926944e-07,
      "loss": 306.7219,
      "step": 48670
    },
    {
      "epoch": 0.9362348664788299,
      "grad_norm": 382.5377869789409,
      "learning_rate": 2.0396806812621018e-07,
      "loss": 305.8214,
      "step": 48680
    },
    {
      "epoch": 0.9364271908145897,
      "grad_norm": 411.2371343496123,
      "learning_rate": 2.0274351652325942e-07,
      "loss": 315.2089,
      "step": 48690
    },
    {
      "epoch": 0.9366195151503496,
      "grad_norm": 403.2280829162947,
      "learning_rate": 2.0152261423654118e-07,
      "loss": 312.0557,
      "step": 48700
    },
    {
      "epoch": 0.9368118394861094,
      "grad_norm": 380.12067956481604,
      "learning_rate": 2.003053617208217e-07,
      "loss": 303.6725,
      "step": 48710
    },
    {
      "epoch": 0.9370041638218692,
      "grad_norm": 412.14216115147207,
      "learning_rate": 1.9909175942950832e-07,
      "loss": 318.0202,
      "step": 48720
    },
    {
      "epoch": 0.937196488157629,
      "grad_norm": 447.68326884775234,
      "learning_rate": 1.9788180781464716e-07,
      "loss": 310.3637,
      "step": 48730
    },
    {
      "epoch": 0.9373888124933889,
      "grad_norm": 409.3970588100632,
      "learning_rate": 1.9667550732692554e-07,
      "loss": 309.9213,
      "step": 48740
    },
    {
      "epoch": 0.9375811368291487,
      "grad_norm": 370.2104350869597,
      "learning_rate": 1.9547285841566843e-07,
      "loss": 300.6118,
      "step": 48750
    },
    {
      "epoch": 0.9377734611649085,
      "grad_norm": 398.5902324477153,
      "learning_rate": 1.9427386152884643e-07,
      "loss": 306.0105,
      "step": 48760
    },
    {
      "epoch": 0.9379657855006683,
      "grad_norm": 416.05418949302583,
      "learning_rate": 1.9307851711306336e-07,
      "loss": 308.0399,
      "step": 48770
    },
    {
      "epoch": 0.9381581098364281,
      "grad_norm": 390.8036209363838,
      "learning_rate": 1.918868256135653e-07,
      "loss": 313.6243,
      "step": 48780
    },
    {
      "epoch": 0.938350434172188,
      "grad_norm": 381.1551355579358,
      "learning_rate": 1.9069878747423943e-07,
      "loss": 306.6806,
      "step": 48790
    },
    {
      "epoch": 0.9385427585079478,
      "grad_norm": 438.6893297102449,
      "learning_rate": 1.8951440313760838e-07,
      "loss": 308.8067,
      "step": 48800
    },
    {
      "epoch": 0.9387350828437077,
      "grad_norm": 384.22215021989064,
      "learning_rate": 1.8833367304483708e-07,
      "loss": 310.653,
      "step": 48810
    },
    {
      "epoch": 0.9389274071794674,
      "grad_norm": 397.3750748611407,
      "learning_rate": 1.8715659763572703e-07,
      "loss": 307.2456,
      "step": 48820
    },
    {
      "epoch": 0.9391197315152273,
      "grad_norm": 396.56710352266924,
      "learning_rate": 1.8598317734872086e-07,
      "loss": 305.1906,
      "step": 48830
    },
    {
      "epoch": 0.9393120558509871,
      "grad_norm": 402.34455709251995,
      "learning_rate": 1.8481341262089668e-07,
      "loss": 309.2093,
      "step": 48840
    },
    {
      "epoch": 0.939504380186747,
      "grad_norm": 436.5600980760861,
      "learning_rate": 1.836473038879727e-07,
      "loss": 303.6096,
      "step": 48850
    },
    {
      "epoch": 0.9396967045225068,
      "grad_norm": 441.5939688861406,
      "learning_rate": 1.8248485158430696e-07,
      "loss": 305.9887,
      "step": 48860
    },
    {
      "epoch": 0.9398890288582665,
      "grad_norm": 416.38813922428216,
      "learning_rate": 1.813260561428909e-07,
      "loss": 305.2926,
      "step": 48870
    },
    {
      "epoch": 0.9400813531940264,
      "grad_norm": 412.8071228424195,
      "learning_rate": 1.8017091799535924e-07,
      "loss": 299.6569,
      "step": 48880
    },
    {
      "epoch": 0.9402736775297862,
      "grad_norm": 382.5826220256806,
      "learning_rate": 1.7901943757198003e-07,
      "loss": 300.0388,
      "step": 48890
    },
    {
      "epoch": 0.9404660018655461,
      "grad_norm": 384.9660267445075,
      "learning_rate": 1.7787161530166242e-07,
      "loss": 305.3296,
      "step": 48900
    },
    {
      "epoch": 0.9406583262013058,
      "grad_norm": 414.76878037652773,
      "learning_rate": 1.7672745161194992e-07,
      "loss": 309.6388,
      "step": 48910
    },
    {
      "epoch": 0.9408506505370657,
      "grad_norm": 394.44532063860356,
      "learning_rate": 1.75586946929025e-07,
      "loss": 305.7398,
      "step": 48920
    },
    {
      "epoch": 0.9410429748728255,
      "grad_norm": 394.61710030903174,
      "learning_rate": 1.7445010167770672e-07,
      "loss": 311.1875,
      "step": 48930
    },
    {
      "epoch": 0.9412352992085854,
      "grad_norm": 409.11691607984994,
      "learning_rate": 1.7331691628145076e-07,
      "loss": 310.648,
      "step": 48940
    },
    {
      "epoch": 0.9414276235443452,
      "grad_norm": 390.60901755971804,
      "learning_rate": 1.7218739116235061e-07,
      "loss": 306.5787,
      "step": 48950
    },
    {
      "epoch": 0.9416199478801051,
      "grad_norm": 388.3207986203266,
      "learning_rate": 1.710615267411353e-07,
      "loss": 299.6847,
      "step": 48960
    },
    {
      "epoch": 0.9418122722158648,
      "grad_norm": 430.40217576926017,
      "learning_rate": 1.6993932343717158e-07,
      "loss": 307.952,
      "step": 48970
    },
    {
      "epoch": 0.9420045965516246,
      "grad_norm": 389.1449077798511,
      "learning_rate": 1.6882078166846173e-07,
      "loss": 301.6381,
      "step": 48980
    },
    {
      "epoch": 0.9421969208873845,
      "grad_norm": 382.38945738082765,
      "learning_rate": 1.677059018516425e-07,
      "loss": 299.7711,
      "step": 48990
    },
    {
      "epoch": 0.9423892452231443,
      "grad_norm": 414.03371972323305,
      "learning_rate": 1.6659468440198835e-07,
      "loss": 310.2514,
      "step": 49000
    },
    {
      "epoch": 0.9425815695589042,
      "grad_norm": 367.14080485365145,
      "learning_rate": 1.6548712973341152e-07,
      "loss": 306.2677,
      "step": 49010
    },
    {
      "epoch": 0.9427738938946639,
      "grad_norm": 387.68516701771307,
      "learning_rate": 1.6438323825845647e-07,
      "loss": 301.7731,
      "step": 49020
    },
    {
      "epoch": 0.9429662182304238,
      "grad_norm": 396.13741559680955,
      "learning_rate": 1.6328301038830429e-07,
      "loss": 308.4436,
      "step": 49030
    },
    {
      "epoch": 0.9431585425661836,
      "grad_norm": 421.965799381177,
      "learning_rate": 1.621864465327716e-07,
      "loss": 297.3285,
      "step": 49040
    },
    {
      "epoch": 0.9433508669019435,
      "grad_norm": 421.77266319393664,
      "learning_rate": 1.6109354710031167e-07,
      "loss": 313.5459,
      "step": 49050
    },
    {
      "epoch": 0.9435431912377032,
      "grad_norm": 378.15834874652194,
      "learning_rate": 1.6000431249800995e-07,
      "loss": 293.139,
      "step": 49060
    },
    {
      "epoch": 0.9437355155734631,
      "grad_norm": 373.7622191531093,
      "learning_rate": 1.5891874313158862e-07,
      "loss": 294.0775,
      "step": 49070
    },
    {
      "epoch": 0.9439278399092229,
      "grad_norm": 403.74011461644466,
      "learning_rate": 1.5783683940540528e-07,
      "loss": 301.0671,
      "step": 49080
    },
    {
      "epoch": 0.9441201642449827,
      "grad_norm": 376.2667831523362,
      "learning_rate": 1.5675860172244982e-07,
      "loss": 299.9229,
      "step": 49090
    },
    {
      "epoch": 0.9443124885807426,
      "grad_norm": 401.3229630982269,
      "learning_rate": 1.5568403048434877e-07,
      "loss": 300.5737,
      "step": 49100
    },
    {
      "epoch": 0.9445048129165023,
      "grad_norm": 373.0953195404252,
      "learning_rate": 1.5461312609136192e-07,
      "loss": 302.8909,
      "step": 49110
    },
    {
      "epoch": 0.9446971372522622,
      "grad_norm": 405.59863465885144,
      "learning_rate": 1.5354588894238465e-07,
      "loss": 300.2997,
      "step": 49120
    },
    {
      "epoch": 0.944889461588022,
      "grad_norm": 379.78001345837527,
      "learning_rate": 1.524823194349434e-07,
      "loss": 308.7511,
      "step": 49130
    },
    {
      "epoch": 0.9450817859237819,
      "grad_norm": 398.4280543255053,
      "learning_rate": 1.514224179652013e-07,
      "loss": 311.1649,
      "step": 49140
    },
    {
      "epoch": 0.9452741102595417,
      "grad_norm": 422.2010323876551,
      "learning_rate": 1.5036618492795473e-07,
      "loss": 308.9265,
      "step": 49150
    },
    {
      "epoch": 0.9454664345953016,
      "grad_norm": 385.51112533019807,
      "learning_rate": 1.4931362071663125e-07,
      "loss": 305.2235,
      "step": 49160
    },
    {
      "epoch": 0.9456587589310613,
      "grad_norm": 375.55332556015094,
      "learning_rate": 1.4826472572329498e-07,
      "loss": 304.5089,
      "step": 49170
    },
    {
      "epoch": 0.9458510832668212,
      "grad_norm": 413.1389251388203,
      "learning_rate": 1.4721950033864118e-07,
      "loss": 298.3317,
      "step": 49180
    },
    {
      "epoch": 0.946043407602581,
      "grad_norm": 388.9181132545055,
      "learning_rate": 1.4617794495199956e-07,
      "loss": 308.9312,
      "step": 49190
    },
    {
      "epoch": 0.9462357319383408,
      "grad_norm": 425.9297379286984,
      "learning_rate": 1.4514005995133197e-07,
      "loss": 307.9106,
      "step": 49200
    },
    {
      "epoch": 0.9464280562741006,
      "grad_norm": 397.88801482406404,
      "learning_rate": 1.441058457232336e-07,
      "loss": 301.4593,
      "step": 49210
    },
    {
      "epoch": 0.9466203806098604,
      "grad_norm": 402.3691909098564,
      "learning_rate": 1.43075302652933e-07,
      "loss": 315.4089,
      "step": 49220
    },
    {
      "epoch": 0.9468127049456203,
      "grad_norm": 406.7789718816777,
      "learning_rate": 1.4204843112428867e-07,
      "loss": 306.0235,
      "step": 49230
    },
    {
      "epoch": 0.9470050292813801,
      "grad_norm": 392.7534964178116,
      "learning_rate": 1.4102523151979574e-07,
      "loss": 306.0012,
      "step": 49240
    },
    {
      "epoch": 0.94719735361714,
      "grad_norm": 427.2808460701671,
      "learning_rate": 1.40005704220576e-07,
      "loss": 310.0522,
      "step": 49250
    },
    {
      "epoch": 0.9473896779528997,
      "grad_norm": 392.41231814618135,
      "learning_rate": 1.3898984960638907e-07,
      "loss": 308.7167,
      "step": 49260
    },
    {
      "epoch": 0.9475820022886596,
      "grad_norm": 430.16492428588924,
      "learning_rate": 1.3797766805562328e-07,
      "loss": 312.7875,
      "step": 49270
    },
    {
      "epoch": 0.9477743266244194,
      "grad_norm": 449.19329225948957,
      "learning_rate": 1.3696915994530048e-07,
      "loss": 306.6118,
      "step": 49280
    },
    {
      "epoch": 0.9479666509601793,
      "grad_norm": 384.21682787348857,
      "learning_rate": 1.359643256510701e-07,
      "loss": 301.349,
      "step": 49290
    },
    {
      "epoch": 0.9481589752959391,
      "grad_norm": 384.98434269197816,
      "learning_rate": 1.3496316554722056e-07,
      "loss": 303.1915,
      "step": 49300
    },
    {
      "epoch": 0.948351299631699,
      "grad_norm": 417.5960335747955,
      "learning_rate": 1.3396568000666467e-07,
      "loss": 307.0324,
      "step": 49310
    },
    {
      "epoch": 0.9485436239674587,
      "grad_norm": 391.2034856262081,
      "learning_rate": 1.3297186940094853e-07,
      "loss": 311.3499,
      "step": 49320
    },
    {
      "epoch": 0.9487359483032185,
      "grad_norm": 434.08483463007167,
      "learning_rate": 1.3198173410025383e-07,
      "loss": 303.7476,
      "step": 49330
    },
    {
      "epoch": 0.9489282726389784,
      "grad_norm": 406.11912099371114,
      "learning_rate": 1.3099527447338668e-07,
      "loss": 310.9488,
      "step": 49340
    },
    {
      "epoch": 0.9491205969747382,
      "grad_norm": 399.76251666063945,
      "learning_rate": 1.3001249088778756e-07,
      "loss": 310.5564,
      "step": 49350
    },
    {
      "epoch": 0.949312921310498,
      "grad_norm": 447.8622510640455,
      "learning_rate": 1.29033383709527e-07,
      "loss": 310.721,
      "step": 49360
    },
    {
      "epoch": 0.9495052456462578,
      "grad_norm": 396.2974496831271,
      "learning_rate": 1.2805795330330774e-07,
      "loss": 311.9839,
      "step": 49370
    },
    {
      "epoch": 0.9496975699820177,
      "grad_norm": 385.20160765765115,
      "learning_rate": 1.2708620003245908e-07,
      "loss": 314.5979,
      "step": 49380
    },
    {
      "epoch": 0.9498898943177775,
      "grad_norm": 387.96535125530716,
      "learning_rate": 1.2611812425894487e-07,
      "loss": 296.1735,
      "step": 49390
    },
    {
      "epoch": 0.9500822186535374,
      "grad_norm": 398.5841305986151,
      "learning_rate": 1.2515372634335666e-07,
      "loss": 300.194,
      "step": 49400
    },
    {
      "epoch": 0.9502745429892971,
      "grad_norm": 403.5354337373987,
      "learning_rate": 1.2419300664491707e-07,
      "loss": 322.5583,
      "step": 49410
    },
    {
      "epoch": 0.950466867325057,
      "grad_norm": 361.2817104121931,
      "learning_rate": 1.2323596552147876e-07,
      "loss": 307.058,
      "step": 49420
    },
    {
      "epoch": 0.9506591916608168,
      "grad_norm": 400.2264206297339,
      "learning_rate": 1.2228260332952213e-07,
      "loss": 307.1072,
      "step": 49430
    },
    {
      "epoch": 0.9508515159965766,
      "grad_norm": 428.97362552393,
      "learning_rate": 1.213329204241609e-07,
      "loss": 301.4423,
      "step": 49440
    },
    {
      "epoch": 0.9510438403323365,
      "grad_norm": 404.18163144765833,
      "learning_rate": 1.203869171591343e-07,
      "loss": 300.4965,
      "step": 49450
    },
    {
      "epoch": 0.9512361646680962,
      "grad_norm": 430.2789604869439,
      "learning_rate": 1.1944459388681496e-07,
      "loss": 316.7147,
      "step": 49460
    },
    {
      "epoch": 0.9514284890038561,
      "grad_norm": 421.6363352017169,
      "learning_rate": 1.1850595095820095e-07,
      "loss": 310.342,
      "step": 49470
    },
    {
      "epoch": 0.9516208133396159,
      "grad_norm": 426.77375756749086,
      "learning_rate": 1.175709887229215e-07,
      "loss": 300.3431,
      "step": 49480
    },
    {
      "epoch": 0.9518131376753758,
      "grad_norm": 406.0751528898561,
      "learning_rate": 1.1663970752923581e-07,
      "loss": 305.0877,
      "step": 49490
    },
    {
      "epoch": 0.9520054620111356,
      "grad_norm": 367.432229710688,
      "learning_rate": 1.1571210772402975e-07,
      "loss": 304.2503,
      "step": 49500
    },
    {
      "epoch": 0.9521977863468954,
      "grad_norm": 403.0718464674492,
      "learning_rate": 1.1478818965281912e-07,
      "loss": 303.6261,
      "step": 49510
    },
    {
      "epoch": 0.9523901106826552,
      "grad_norm": 420.67460824295307,
      "learning_rate": 1.1386795365974757e-07,
      "loss": 312.8814,
      "step": 49520
    },
    {
      "epoch": 0.9525824350184151,
      "grad_norm": 423.5096050607596,
      "learning_rate": 1.1295140008758864e-07,
      "loss": 305.2243,
      "step": 49530
    },
    {
      "epoch": 0.9527747593541749,
      "grad_norm": 425.2835309463311,
      "learning_rate": 1.1203852927774372e-07,
      "loss": 308.5336,
      "step": 49540
    },
    {
      "epoch": 0.9529670836899347,
      "grad_norm": 423.538469079853,
      "learning_rate": 1.111293415702408e-07,
      "loss": 304.1133,
      "step": 49550
    },
    {
      "epoch": 0.9531594080256945,
      "grad_norm": 402.5345550285309,
      "learning_rate": 1.1022383730373897e-07,
      "loss": 311.5206,
      "step": 49560
    },
    {
      "epoch": 0.9533517323614543,
      "grad_norm": 381.8244980876525,
      "learning_rate": 1.093220168155218e-07,
      "loss": 305.0832,
      "step": 49570
    },
    {
      "epoch": 0.9535440566972142,
      "grad_norm": 404.30082496260746,
      "learning_rate": 1.0842388044150387e-07,
      "loss": 307.0992,
      "step": 49580
    },
    {
      "epoch": 0.953736381032974,
      "grad_norm": 414.38757151108206,
      "learning_rate": 1.0752942851622649e-07,
      "loss": 304.3737,
      "step": 49590
    },
    {
      "epoch": 0.9539287053687339,
      "grad_norm": 401.888699612354,
      "learning_rate": 1.066386613728565e-07,
      "loss": 303.2939,
      "step": 49600
    },
    {
      "epoch": 0.9541210297044936,
      "grad_norm": 410.2954464534775,
      "learning_rate": 1.0575157934319069e-07,
      "loss": 312.0517,
      "step": 49610
    },
    {
      "epoch": 0.9543133540402535,
      "grad_norm": 400.4149457331645,
      "learning_rate": 1.0486818275765364e-07,
      "loss": 305.2278,
      "step": 49620
    },
    {
      "epoch": 0.9545056783760133,
      "grad_norm": 404.71467144260447,
      "learning_rate": 1.0398847194529437e-07,
      "loss": 309.8971,
      "step": 49630
    },
    {
      "epoch": 0.9546980027117732,
      "grad_norm": 413.4985580261763,
      "learning_rate": 1.0311244723379188e-07,
      "loss": 305.9196,
      "step": 49640
    },
    {
      "epoch": 0.954890327047533,
      "grad_norm": 395.9696575498284,
      "learning_rate": 1.0224010894944958e-07,
      "loss": 313.5794,
      "step": 49650
    },
    {
      "epoch": 0.9550826513832927,
      "grad_norm": 382.596699608233,
      "learning_rate": 1.0137145741719867e-07,
      "loss": 322.5343,
      "step": 49660
    },
    {
      "epoch": 0.9552749757190526,
      "grad_norm": 419.6672520534782,
      "learning_rate": 1.0050649296060033e-07,
      "loss": 323.1996,
      "step": 49670
    },
    {
      "epoch": 0.9554673000548124,
      "grad_norm": 426.67888625184986,
      "learning_rate": 9.964521590183684e-08,
      "loss": 312.8907,
      "step": 49680
    },
    {
      "epoch": 0.9556596243905723,
      "grad_norm": 419.05212244028445,
      "learning_rate": 9.878762656172159e-08,
      "loss": 308.5332,
      "step": 49690
    },
    {
      "epoch": 0.9558519487263321,
      "grad_norm": 402.4653081142015,
      "learning_rate": 9.793372525969125e-08,
      "loss": 318.4444,
      "step": 49700
    },
    {
      "epoch": 0.9560442730620919,
      "grad_norm": 424.62233924339193,
      "learning_rate": 9.70835123138103e-08,
      "loss": 309.7108,
      "step": 49710
    },
    {
      "epoch": 0.9562365973978517,
      "grad_norm": 413.8927788632199,
      "learning_rate": 9.623698804076875e-08,
      "loss": 310.468,
      "step": 49720
    },
    {
      "epoch": 0.9564289217336116,
      "grad_norm": 415.85893275566184,
      "learning_rate": 9.539415275588326e-08,
      "loss": 318.2136,
      "step": 49730
    },
    {
      "epoch": 0.9566212460693714,
      "grad_norm": 393.99944606521944,
      "learning_rate": 9.455500677309603e-08,
      "loss": 328.1828,
      "step": 49740
    },
    {
      "epoch": 0.9568135704051313,
      "grad_norm": 408.8276224792121,
      "learning_rate": 9.371955040497371e-08,
      "loss": 304.5924,
      "step": 49750
    },
    {
      "epoch": 0.957005894740891,
      "grad_norm": 400.3012703666059,
      "learning_rate": 9.288778396271292e-08,
      "loss": 316.8154,
      "step": 49760
    },
    {
      "epoch": 0.9571982190766508,
      "grad_norm": 417.31839820868305,
      "learning_rate": 9.205970775613027e-08,
      "loss": 307.1784,
      "step": 49770
    },
    {
      "epoch": 0.9573905434124107,
      "grad_norm": 477.06723231406625,
      "learning_rate": 9.123532209367237e-08,
      "loss": 316.4184,
      "step": 49780
    },
    {
      "epoch": 0.9575828677481705,
      "grad_norm": 372.45949209567954,
      "learning_rate": 9.04146272824069e-08,
      "loss": 294.5549,
      "step": 49790
    },
    {
      "epoch": 0.9577751920839304,
      "grad_norm": 375.24122129543525,
      "learning_rate": 8.959762362803159e-08,
      "loss": 303.8367,
      "step": 49800
    },
    {
      "epoch": 0.9579675164196901,
      "grad_norm": 377.53739854846873,
      "learning_rate": 8.87843114348652e-08,
      "loss": 304.0116,
      "step": 49810
    },
    {
      "epoch": 0.95815984075545,
      "grad_norm": 406.4118666624434,
      "learning_rate": 8.797469100585432e-08,
      "loss": 312.4471,
      "step": 49820
    },
    {
      "epoch": 0.9583521650912098,
      "grad_norm": 459.15009359201656,
      "learning_rate": 8.716876264256768e-08,
      "loss": 321.9091,
      "step": 49830
    },
    {
      "epoch": 0.9585444894269697,
      "grad_norm": 408.4950976532906,
      "learning_rate": 8.636652664520184e-08,
      "loss": 308.0166,
      "step": 49840
    },
    {
      "epoch": 0.9587368137627295,
      "grad_norm": 395.5132455886173,
      "learning_rate": 8.556798331257555e-08,
      "loss": 310.2185,
      "step": 49850
    },
    {
      "epoch": 0.9589291380984893,
      "grad_norm": 415.90346541180594,
      "learning_rate": 8.477313294213307e-08,
      "loss": 307.0948,
      "step": 49860
    },
    {
      "epoch": 0.9591214624342491,
      "grad_norm": 391.7613206470682,
      "learning_rate": 8.398197582994316e-08,
      "loss": 297.7784,
      "step": 49870
    },
    {
      "epoch": 0.9593137867700089,
      "grad_norm": 397.5111267836956,
      "learning_rate": 8.319451227069897e-08,
      "loss": 307.3765,
      "step": 49880
    },
    {
      "epoch": 0.9595061111057688,
      "grad_norm": 385.8547203788916,
      "learning_rate": 8.241074255771808e-08,
      "loss": 301.4813,
      "step": 49890
    },
    {
      "epoch": 0.9596984354415286,
      "grad_norm": 382.45215401884656,
      "learning_rate": 8.163066698294031e-08,
      "loss": 306.9535,
      "step": 49900
    },
    {
      "epoch": 0.9598907597772884,
      "grad_norm": 388.8994936404778,
      "learning_rate": 8.085428583693211e-08,
      "loss": 304.2015,
      "step": 49910
    },
    {
      "epoch": 0.9600830841130482,
      "grad_norm": 392.0252979421027,
      "learning_rate": 8.008159940888216e-08,
      "loss": 305.9475,
      "step": 49920
    },
    {
      "epoch": 0.9602754084488081,
      "grad_norm": 392.7824997345464,
      "learning_rate": 7.931260798660356e-08,
      "loss": 310.9074,
      "step": 49930
    },
    {
      "epoch": 0.9604677327845679,
      "grad_norm": 377.2899378856775,
      "learning_rate": 7.854731185653386e-08,
      "loss": 300.7823,
      "step": 49940
    },
    {
      "epoch": 0.9606600571203278,
      "grad_norm": 399.7146138685817,
      "learning_rate": 7.778571130373059e-08,
      "loss": 305.756,
      "step": 49950
    },
    {
      "epoch": 0.9608523814560875,
      "grad_norm": 393.1386161906627,
      "learning_rate": 7.702780661188014e-08,
      "loss": 310.4812,
      "step": 49960
    },
    {
      "epoch": 0.9610447057918474,
      "grad_norm": 411.01353957222665,
      "learning_rate": 7.627359806328782e-08,
      "loss": 308.5993,
      "step": 49970
    },
    {
      "epoch": 0.9612370301276072,
      "grad_norm": 404.54551564524786,
      "learning_rate": 7.552308593888558e-08,
      "loss": 308.0479,
      "step": 49980
    },
    {
      "epoch": 0.961429354463367,
      "grad_norm": 397.9958138875089,
      "learning_rate": 7.477627051822534e-08,
      "loss": 312.099,
      "step": 49990
    },
    {
      "epoch": 0.9616216787991269,
      "grad_norm": 413.00497974495903,
      "learning_rate": 7.403315207948236e-08,
      "loss": 319.8621,
      "step": 50000
    },
    {
      "epoch": 0.9618140031348866,
      "grad_norm": 422.56634417496,
      "learning_rate": 7.329373089945968e-08,
      "loss": 299.7168,
      "step": 50010
    },
    {
      "epoch": 0.9620063274706465,
      "grad_norm": 394.99070959861695,
      "learning_rate": 7.255800725357586e-08,
      "loss": 310.4783,
      "step": 50020
    },
    {
      "epoch": 0.9621986518064063,
      "grad_norm": 407.5660028983763,
      "learning_rate": 7.182598141587838e-08,
      "loss": 309.7587,
      "step": 50030
    },
    {
      "epoch": 0.9623909761421662,
      "grad_norm": 395.4002094215585,
      "learning_rate": 7.109765365903243e-08,
      "loss": 312.1669,
      "step": 50040
    },
    {
      "epoch": 0.962583300477926,
      "grad_norm": 393.15639269285674,
      "learning_rate": 7.03730242543299e-08,
      "loss": 309.2928,
      "step": 50050
    },
    {
      "epoch": 0.9627756248136858,
      "grad_norm": 376.6052912892147,
      "learning_rate": 6.965209347168156e-08,
      "loss": 302.5852,
      "step": 50060
    },
    {
      "epoch": 0.9629679491494456,
      "grad_norm": 393.38766705849474,
      "learning_rate": 6.89348615796237e-08,
      "loss": 300.5664,
      "step": 50070
    },
    {
      "epoch": 0.9631602734852055,
      "grad_norm": 386.8942351982451,
      "learning_rate": 6.822132884531373e-08,
      "loss": 304.1994,
      "step": 50080
    },
    {
      "epoch": 0.9633525978209653,
      "grad_norm": 441.2890245819458,
      "learning_rate": 6.751149553452907e-08,
      "loss": 313.6467,
      "step": 50090
    },
    {
      "epoch": 0.963544922156725,
      "grad_norm": 375.4967465461503,
      "learning_rate": 6.680536191167263e-08,
      "loss": 302.0366,
      "step": 50100
    },
    {
      "epoch": 0.9637372464924849,
      "grad_norm": 390.45468352443703,
      "learning_rate": 6.610292823976628e-08,
      "loss": 301.4835,
      "step": 50110
    },
    {
      "epoch": 0.9639295708282447,
      "grad_norm": 385.14826788375,
      "learning_rate": 6.540419478045623e-08,
      "loss": 311.1793,
      "step": 50120
    },
    {
      "epoch": 0.9641218951640046,
      "grad_norm": 406.9672248852778,
      "learning_rate": 6.470916179400765e-08,
      "loss": 300.1742,
      "step": 50130
    },
    {
      "epoch": 0.9643142194997644,
      "grad_norm": 407.471481756683,
      "learning_rate": 6.401782953931013e-08,
      "loss": 302.3209,
      "step": 50140
    },
    {
      "epoch": 0.9645065438355243,
      "grad_norm": 415.3177685416555,
      "learning_rate": 6.33301982738721e-08,
      "loss": 304.1317,
      "step": 50150
    },
    {
      "epoch": 0.964698868171284,
      "grad_norm": 381.7151504954698,
      "learning_rate": 6.264626825382647e-08,
      "loss": 314.0569,
      "step": 50160
    },
    {
      "epoch": 0.9648911925070439,
      "grad_norm": 391.9590881332063,
      "learning_rate": 6.196603973392501e-08,
      "loss": 311.9383,
      "step": 50170
    },
    {
      "epoch": 0.9650835168428037,
      "grad_norm": 386.02426289811626,
      "learning_rate": 6.12895129675406e-08,
      "loss": 299.6151,
      "step": 50180
    },
    {
      "epoch": 0.9652758411785636,
      "grad_norm": 419.27527632559145,
      "learning_rate": 6.061668820667055e-08,
      "loss": 311.5686,
      "step": 50190
    },
    {
      "epoch": 0.9654681655143234,
      "grad_norm": 400.5806593390106,
      "learning_rate": 5.994756570192994e-08,
      "loss": 315.774,
      "step": 50200
    },
    {
      "epoch": 0.9656604898500831,
      "grad_norm": 389.998173460643,
      "learning_rate": 5.928214570255497e-08,
      "loss": 295.7583,
      "step": 50210
    },
    {
      "epoch": 0.965852814185843,
      "grad_norm": 410.2645306611366,
      "learning_rate": 5.862042845640403e-08,
      "loss": 303.524,
      "step": 50220
    },
    {
      "epoch": 0.9660451385216028,
      "grad_norm": 387.06745229011955,
      "learning_rate": 5.796241420995663e-08,
      "loss": 295.1914,
      "step": 50230
    },
    {
      "epoch": 0.9662374628573627,
      "grad_norm": 380.5509252063031,
      "learning_rate": 5.730810320831226e-08,
      "loss": 306.6095,
      "step": 50240
    },
    {
      "epoch": 0.9664297871931224,
      "grad_norm": 386.61970623577,
      "learning_rate": 5.66574956951893e-08,
      "loss": 309.1897,
      "step": 50250
    },
    {
      "epoch": 0.9666221115288823,
      "grad_norm": 422.5208012942953,
      "learning_rate": 5.6010591912930565e-08,
      "loss": 315.5407,
      "step": 50260
    },
    {
      "epoch": 0.9668144358646421,
      "grad_norm": 398.18764631578574,
      "learning_rate": 5.5367392102495534e-08,
      "loss": 307.9312,
      "step": 50270
    },
    {
      "epoch": 0.967006760200402,
      "grad_norm": 407.6783027921625,
      "learning_rate": 5.472789650346588e-08,
      "loss": 311.0954,
      "step": 50280
    },
    {
      "epoch": 0.9671990845361618,
      "grad_norm": 398.9438511540597,
      "learning_rate": 5.4092105354043304e-08,
      "loss": 309.1076,
      "step": 50290
    },
    {
      "epoch": 0.9673914088719217,
      "grad_norm": 394.09759904739724,
      "learning_rate": 5.346001889104946e-08,
      "loss": 297.5935,
      "step": 50300
    },
    {
      "epoch": 0.9675837332076814,
      "grad_norm": 430.35533188076937,
      "learning_rate": 5.2831637349926026e-08,
      "loss": 304.3507,
      "step": 50310
    },
    {
      "epoch": 0.9677760575434413,
      "grad_norm": 401.7287500696597,
      "learning_rate": 5.220696096473465e-08,
      "loss": 302.7906,
      "step": 50320
    },
    {
      "epoch": 0.9679683818792011,
      "grad_norm": 448.3982342545897,
      "learning_rate": 5.1585989968157e-08,
      "loss": 322.4165,
      "step": 50330
    },
    {
      "epoch": 0.9681607062149609,
      "grad_norm": 364.5716205433589,
      "learning_rate": 5.096872459149471e-08,
      "loss": 304.3308,
      "step": 50340
    },
    {
      "epoch": 0.9683530305507208,
      "grad_norm": 415.3120502310781,
      "learning_rate": 5.035516506466942e-08,
      "loss": 319.0656,
      "step": 50350
    },
    {
      "epoch": 0.9685453548864805,
      "grad_norm": 469.2289410725958,
      "learning_rate": 4.9745311616220535e-08,
      "loss": 320.0059,
      "step": 50360
    },
    {
      "epoch": 0.9687376792222404,
      "grad_norm": 375.84391029795586,
      "learning_rate": 4.913916447330858e-08,
      "loss": 306.8135,
      "step": 50370
    },
    {
      "epoch": 0.9689300035580002,
      "grad_norm": 394.38854920005514,
      "learning_rate": 4.853672386171515e-08,
      "loss": 312.7281,
      "step": 50380
    },
    {
      "epoch": 0.9691223278937601,
      "grad_norm": 393.5334830174472,
      "learning_rate": 4.793799000583743e-08,
      "loss": 304.467,
      "step": 50390
    },
    {
      "epoch": 0.9693146522295198,
      "grad_norm": 424.7752850602039,
      "learning_rate": 4.73429631286948e-08,
      "loss": 304.6041,
      "step": 50400
    },
    {
      "epoch": 0.9695069765652797,
      "grad_norm": 404.8066271402032,
      "learning_rate": 4.675164345192551e-08,
      "loss": 295.7238,
      "step": 50410
    },
    {
      "epoch": 0.9696993009010395,
      "grad_norm": 409.1511846227577,
      "learning_rate": 4.6164031195785606e-08,
      "loss": 307.2115,
      "step": 50420
    },
    {
      "epoch": 0.9698916252367994,
      "grad_norm": 374.41955508403254,
      "learning_rate": 4.558012657915112e-08,
      "loss": 308.1001,
      "step": 50430
    },
    {
      "epoch": 0.9700839495725592,
      "grad_norm": 396.90723459075224,
      "learning_rate": 4.4999929819515844e-08,
      "loss": 314.8178,
      "step": 50440
    },
    {
      "epoch": 0.9702762739083189,
      "grad_norm": 422.1570173600825,
      "learning_rate": 4.442344113299579e-08,
      "loss": 307.5798,
      "step": 50450
    },
    {
      "epoch": 0.9704685982440788,
      "grad_norm": 397.710503498998,
      "learning_rate": 4.385066073432143e-08,
      "loss": 299.9531,
      "step": 50460
    },
    {
      "epoch": 0.9706609225798386,
      "grad_norm": 429.1507150623487,
      "learning_rate": 4.3281588836844303e-08,
      "loss": 306.3626,
      "step": 50470
    },
    {
      "epoch": 0.9708532469155985,
      "grad_norm": 390.79850266021,
      "learning_rate": 4.271622565253486e-08,
      "loss": 298.7711,
      "step": 50480
    },
    {
      "epoch": 0.9710455712513583,
      "grad_norm": 399.88431374344094,
      "learning_rate": 4.2154571391982425e-08,
      "loss": 304.1253,
      "step": 50490
    },
    {
      "epoch": 0.9712378955871181,
      "grad_norm": 410.7327727315363,
      "learning_rate": 4.159662626439187e-08,
      "loss": 310.6277,
      "step": 50500
    },
    {
      "epoch": 0.9714302199228779,
      "grad_norm": 420.84145442808045,
      "learning_rate": 4.104239047758918e-08,
      "loss": 311.6168,
      "step": 50510
    },
    {
      "epoch": 0.9716225442586378,
      "grad_norm": 393.2143837689044,
      "learning_rate": 4.049186423801921e-08,
      "loss": 297.8025,
      "step": 50520
    },
    {
      "epoch": 0.9718148685943976,
      "grad_norm": 415.3830146922677,
      "learning_rate": 3.994504775074237e-08,
      "loss": 299.5151,
      "step": 50530
    },
    {
      "epoch": 0.9720071929301575,
      "grad_norm": 406.4663555662246,
      "learning_rate": 3.9401941219440186e-08,
      "loss": 312.5655,
      "step": 50540
    },
    {
      "epoch": 0.9721995172659172,
      "grad_norm": 401.13190759834015,
      "learning_rate": 3.8862544846409725e-08,
      "loss": 303.4033,
      "step": 50550
    },
    {
      "epoch": 0.972391841601677,
      "grad_norm": 429.3602747938986,
      "learning_rate": 3.832685883256915e-08,
      "loss": 309.5042,
      "step": 50560
    },
    {
      "epoch": 0.9725841659374369,
      "grad_norm": 393.8213768007156,
      "learning_rate": 3.7794883377449966e-08,
      "loss": 302.9365,
      "step": 50570
    },
    {
      "epoch": 0.9727764902731967,
      "grad_norm": 391.2815581564051,
      "learning_rate": 3.726661867920478e-08,
      "loss": 305.5504,
      "step": 50580
    },
    {
      "epoch": 0.9729688146089566,
      "grad_norm": 393.73702787501804,
      "learning_rate": 3.674206493460508e-08,
      "loss": 304.4101,
      "step": 50590
    },
    {
      "epoch": 0.9731611389447163,
      "grad_norm": 387.53807641553425,
      "learning_rate": 3.622122233903791e-08,
      "loss": 302.7742,
      "step": 50600
    },
    {
      "epoch": 0.9733534632804762,
      "grad_norm": 437.84801072834466,
      "learning_rate": 3.5704091086508076e-08,
      "loss": 311.2802,
      "step": 50610
    },
    {
      "epoch": 0.973545787616236,
      "grad_norm": 432.73723220634884,
      "learning_rate": 3.519067136963705e-08,
      "loss": 304.799,
      "step": 50620
    },
    {
      "epoch": 0.9737381119519959,
      "grad_norm": 421.13571721828066,
      "learning_rate": 3.468096337966853e-08,
      "loss": 307.158,
      "step": 50630
    },
    {
      "epoch": 0.9739304362877557,
      "grad_norm": 428.0337305552007,
      "learning_rate": 3.417496730645731e-08,
      "loss": 307.0303,
      "step": 50640
    },
    {
      "epoch": 0.9741227606235155,
      "grad_norm": 407.82962995120494,
      "learning_rate": 3.3672683338480396e-08,
      "loss": 309.2827,
      "step": 50650
    },
    {
      "epoch": 0.9743150849592753,
      "grad_norm": 406.9420647927319,
      "learning_rate": 3.317411166282813e-08,
      "loss": 304.6213,
      "step": 50660
    },
    {
      "epoch": 0.9745074092950351,
      "grad_norm": 375.8363907832895,
      "learning_rate": 3.2679252465213085e-08,
      "loss": 306.5873,
      "step": 50670
    },
    {
      "epoch": 0.974699733630795,
      "grad_norm": 421.29060346936956,
      "learning_rate": 3.218810592996113e-08,
      "loss": 310.1658,
      "step": 50680
    },
    {
      "epoch": 0.9748920579665548,
      "grad_norm": 455.6126356492348,
      "learning_rate": 3.170067224001483e-08,
      "loss": 321.2601,
      "step": 50690
    },
    {
      "epoch": 0.9750843823023146,
      "grad_norm": 378.29428729541024,
      "learning_rate": 3.1216951576936714e-08,
      "loss": 293.8388,
      "step": 50700
    },
    {
      "epoch": 0.9752767066380744,
      "grad_norm": 397.17722750018015,
      "learning_rate": 3.073694412090489e-08,
      "loss": 304.5924,
      "step": 50710
    },
    {
      "epoch": 0.9754690309738343,
      "grad_norm": 409.93933600121204,
      "learning_rate": 3.026065005071188e-08,
      "loss": 308.2416,
      "step": 50720
    },
    {
      "epoch": 0.9756613553095941,
      "grad_norm": 398.2808171212146,
      "learning_rate": 2.9788069543772445e-08,
      "loss": 313.6742,
      "step": 50730
    },
    {
      "epoch": 0.975853679645354,
      "grad_norm": 377.4287067372057,
      "learning_rate": 2.9319202776113553e-08,
      "loss": 301.8081,
      "step": 50740
    },
    {
      "epoch": 0.9760460039811137,
      "grad_norm": 389.89803132758874,
      "learning_rate": 2.8854049922379946e-08,
      "loss": 301.6007,
      "step": 50750
    },
    {
      "epoch": 0.9762383283168736,
      "grad_norm": 417.8896116356527,
      "learning_rate": 2.839261115583303e-08,
      "loss": 309.4142,
      "step": 50760
    },
    {
      "epoch": 0.9764306526526334,
      "grad_norm": 470.34866542704964,
      "learning_rate": 2.793488664835309e-08,
      "loss": 306.002,
      "step": 50770
    },
    {
      "epoch": 0.9766229769883932,
      "grad_norm": 407.1327735245423,
      "learning_rate": 2.7480876570433746e-08,
      "loss": 300.7665,
      "step": 50780
    },
    {
      "epoch": 0.9768153013241531,
      "grad_norm": 400.87714116694724,
      "learning_rate": 2.7030581091186393e-08,
      "loss": 314.4448,
      "step": 50790
    },
    {
      "epoch": 0.9770076256599128,
      "grad_norm": 410.89369144701084,
      "learning_rate": 2.658400037833686e-08,
      "loss": 302.8533,
      "step": 50800
    },
    {
      "epoch": 0.9771999499956727,
      "grad_norm": 382.7188830757354,
      "learning_rate": 2.6141134598233197e-08,
      "loss": 303.5628,
      "step": 50810
    },
    {
      "epoch": 0.9773922743314325,
      "grad_norm": 433.4141542239082,
      "learning_rate": 2.5701983915831232e-08,
      "loss": 310.3563,
      "step": 50820
    },
    {
      "epoch": 0.9775845986671924,
      "grad_norm": 373.79385501027184,
      "learning_rate": 2.5266548494710108e-08,
      "loss": 306.4014,
      "step": 50830
    },
    {
      "epoch": 0.9777769230029522,
      "grad_norm": 377.9372646502148,
      "learning_rate": 2.4834828497062315e-08,
      "loss": 300.3727,
      "step": 50840
    },
    {
      "epoch": 0.977969247338712,
      "grad_norm": 485.5862600261297,
      "learning_rate": 2.4406824083694768e-08,
      "loss": 312.0458,
      "step": 50850
    },
    {
      "epoch": 0.9781615716744718,
      "grad_norm": 394.00691307111975,
      "learning_rate": 2.398253541403217e-08,
      "loss": 317.0317,
      "step": 50860
    },
    {
      "epoch": 0.9783538960102317,
      "grad_norm": 386.2322868237349,
      "learning_rate": 2.3561962646116988e-08,
      "loss": 305.768,
      "step": 50870
    },
    {
      "epoch": 0.9785462203459915,
      "grad_norm": 399.529279335095,
      "learning_rate": 2.3145105936603906e-08,
      "loss": 296.2705,
      "step": 50880
    },
    {
      "epoch": 0.9787385446817513,
      "grad_norm": 425.15762829141664,
      "learning_rate": 2.273196544076539e-08,
      "loss": 314.122,
      "step": 50890
    },
    {
      "epoch": 0.9789308690175111,
      "grad_norm": 380.179433289847,
      "learning_rate": 2.2322541312490565e-08,
      "loss": 306.0842,
      "step": 50900
    },
    {
      "epoch": 0.9791231933532709,
      "grad_norm": 402.3194928844894,
      "learning_rate": 2.1916833704281882e-08,
      "loss": 319.9796,
      "step": 50910
    },
    {
      "epoch": 0.9793155176890308,
      "grad_norm": 416.90339187460955,
      "learning_rate": 2.1514842767258458e-08,
      "loss": 298.4591,
      "step": 50920
    },
    {
      "epoch": 0.9795078420247906,
      "grad_norm": 392.84490320889864,
      "learning_rate": 2.1116568651156076e-08,
      "loss": 312.3883,
      "step": 50930
    },
    {
      "epoch": 0.9797001663605505,
      "grad_norm": 393.7744491721409,
      "learning_rate": 2.0722011504326066e-08,
      "loss": 304.6168,
      "step": 50940
    },
    {
      "epoch": 0.9798924906963102,
      "grad_norm": 382.6497005103055,
      "learning_rate": 2.033117147373309e-08,
      "loss": 301.5205,
      "step": 50950
    },
    {
      "epoch": 0.9800848150320701,
      "grad_norm": 379.180241426101,
      "learning_rate": 1.9944048704959583e-08,
      "loss": 305.3807,
      "step": 50960
    },
    {
      "epoch": 0.9802771393678299,
      "grad_norm": 368.2175644298272,
      "learning_rate": 1.956064334220131e-08,
      "loss": 299.8416,
      "step": 50970
    },
    {
      "epoch": 0.9804694637035898,
      "grad_norm": 385.60592737707225,
      "learning_rate": 1.9180955528270706e-08,
      "loss": 312.7126,
      "step": 50980
    },
    {
      "epoch": 0.9806617880393496,
      "grad_norm": 427.2375132529489,
      "learning_rate": 1.8804985404595743e-08,
      "loss": 313.5155,
      "step": 50990
    },
    {
      "epoch": 0.9808541123751093,
      "grad_norm": 387.3286278280876,
      "learning_rate": 1.843273311121885e-08,
      "loss": 300.3423,
      "step": 51000
    },
    {
      "epoch": 0.9810464367108692,
      "grad_norm": 433.04233986949123,
      "learning_rate": 1.806419878679799e-08,
      "loss": 311.5273,
      "step": 51010
    },
    {
      "epoch": 0.981238761046629,
      "grad_norm": 397.1104184447403,
      "learning_rate": 1.7699382568605595e-08,
      "loss": 301.9248,
      "step": 51020
    },
    {
      "epoch": 0.9814310853823889,
      "grad_norm": 382.9905869992039,
      "learning_rate": 1.733828459253073e-08,
      "loss": 310.2416,
      "step": 51030
    },
    {
      "epoch": 0.9816234097181487,
      "grad_norm": 405.24995014246383,
      "learning_rate": 1.6980904993075808e-08,
      "loss": 304.8208,
      "step": 51040
    },
    {
      "epoch": 0.9818157340539085,
      "grad_norm": 388.39876378459087,
      "learning_rate": 1.6627243903357682e-08,
      "loss": 325.4466,
      "step": 51050
    },
    {
      "epoch": 0.9820080583896683,
      "grad_norm": 371.9757594796885,
      "learning_rate": 1.6277301455110972e-08,
      "loss": 306.0827,
      "step": 51060
    },
    {
      "epoch": 0.9822003827254282,
      "grad_norm": 404.00571243814244,
      "learning_rate": 1.5931077778682524e-08,
      "loss": 309.9252,
      "step": 51070
    },
    {
      "epoch": 0.982392707061188,
      "grad_norm": 406.902630266166,
      "learning_rate": 1.5588573003035847e-08,
      "loss": 309.5406,
      "step": 51080
    },
    {
      "epoch": 0.9825850313969479,
      "grad_norm": 468.74044916363795,
      "learning_rate": 1.5249787255747774e-08,
      "loss": 309.0986,
      "step": 51090
    },
    {
      "epoch": 0.9827773557327076,
      "grad_norm": 379.65768166783414,
      "learning_rate": 1.4914720663009585e-08,
      "loss": 302.6833,
      "step": 51100
    },
    {
      "epoch": 0.9829696800684674,
      "grad_norm": 413.8310771702398,
      "learning_rate": 1.4583373349629226e-08,
      "loss": 304.697,
      "step": 51110
    },
    {
      "epoch": 0.9831620044042273,
      "grad_norm": 389.0675171929064,
      "learning_rate": 1.4255745439027968e-08,
      "loss": 300.6031,
      "step": 51120
    },
    {
      "epoch": 0.9833543287399871,
      "grad_norm": 406.32247314575346,
      "learning_rate": 1.3931837053241526e-08,
      "loss": 311.4055,
      "step": 51130
    },
    {
      "epoch": 0.983546653075747,
      "grad_norm": 402.2565677137849,
      "learning_rate": 1.3611648312920057e-08,
      "loss": 312.1779,
      "step": 51140
    },
    {
      "epoch": 0.9837389774115067,
      "grad_norm": 399.8158105292107,
      "learning_rate": 1.3295179337329267e-08,
      "loss": 305.4676,
      "step": 51150
    },
    {
      "epoch": 0.9839313017472666,
      "grad_norm": 400.69074590236204,
      "learning_rate": 1.2982430244347088e-08,
      "loss": 315.5982,
      "step": 51160
    },
    {
      "epoch": 0.9841236260830264,
      "grad_norm": 419.02739384459056,
      "learning_rate": 1.2673401150470333e-08,
      "loss": 314.1231,
      "step": 51170
    },
    {
      "epoch": 0.9843159504187863,
      "grad_norm": 413.87051591035225,
      "learning_rate": 1.2368092170804702e-08,
      "loss": 308.4232,
      "step": 51180
    },
    {
      "epoch": 0.984508274754546,
      "grad_norm": 451.87349675268644,
      "learning_rate": 1.2066503419073672e-08,
      "loss": 307.2372,
      "step": 51190
    },
    {
      "epoch": 0.9847005990903059,
      "grad_norm": 381.44144011771385,
      "learning_rate": 1.1768635007614049e-08,
      "loss": 306.4786,
      "step": 51200
    },
    {
      "epoch": 0.9848929234260657,
      "grad_norm": 375.4641172892268,
      "learning_rate": 1.1474487047375971e-08,
      "loss": 306.0885,
      "step": 51210
    },
    {
      "epoch": 0.9850852477618255,
      "grad_norm": 466.5386189686091,
      "learning_rate": 1.1184059647926238e-08,
      "loss": 310.9882,
      "step": 51220
    },
    {
      "epoch": 0.9852775720975854,
      "grad_norm": 417.79839407018136,
      "learning_rate": 1.0897352917443871e-08,
      "loss": 313.2638,
      "step": 51230
    },
    {
      "epoch": 0.9854698964333451,
      "grad_norm": 404.86815546221015,
      "learning_rate": 1.0614366962721223e-08,
      "loss": 309.3784,
      "step": 51240
    },
    {
      "epoch": 0.985662220769105,
      "grad_norm": 357.5197060227325,
      "learning_rate": 1.033510188916731e-08,
      "loss": 294.0667,
      "step": 51250
    },
    {
      "epoch": 0.9858545451048648,
      "grad_norm": 406.77961151528575,
      "learning_rate": 1.0059557800802256e-08,
      "loss": 305.8481,
      "step": 51260
    },
    {
      "epoch": 0.9860468694406247,
      "grad_norm": 384.74437075020825,
      "learning_rate": 9.78773480026396e-09,
      "loss": 299.6159,
      "step": 51270
    },
    {
      "epoch": 0.9862391937763845,
      "grad_norm": 406.31745349212474,
      "learning_rate": 9.519632988800321e-09,
      "loss": 311.7801,
      "step": 51280
    },
    {
      "epoch": 0.9864315181121444,
      "grad_norm": 380.0166883704215,
      "learning_rate": 9.25525246627479e-09,
      "loss": 311.1568,
      "step": 51290
    },
    {
      "epoch": 0.9866238424479041,
      "grad_norm": 409.1457322928211,
      "learning_rate": 8.994593331165257e-09,
      "loss": 312.1888,
      "step": 51300
    },
    {
      "epoch": 0.986816166783664,
      "grad_norm": 386.033341229226,
      "learning_rate": 8.737655680562951e-09,
      "loss": 297.5578,
      "step": 51310
    },
    {
      "epoch": 0.9870084911194238,
      "grad_norm": 420.5638003223459,
      "learning_rate": 8.484439610172424e-09,
      "loss": 305.1758,
      "step": 51320
    },
    {
      "epoch": 0.9872008154551836,
      "grad_norm": 398.77872534682115,
      "learning_rate": 8.234945214312673e-09,
      "loss": 309.3612,
      "step": 51330
    },
    {
      "epoch": 0.9873931397909435,
      "grad_norm": 374.43751095335097,
      "learning_rate": 7.989172585917138e-09,
      "loss": 307.6127,
      "step": 51340
    },
    {
      "epoch": 0.9875854641267032,
      "grad_norm": 403.3645140835388,
      "learning_rate": 7.747121816530368e-09,
      "loss": 306.856,
      "step": 51350
    },
    {
      "epoch": 0.9877777884624631,
      "grad_norm": 389.4955670725102,
      "learning_rate": 7.508792996313573e-09,
      "loss": 310.2944,
      "step": 51360
    },
    {
      "epoch": 0.9879701127982229,
      "grad_norm": 385.192448654295,
      "learning_rate": 7.274186214040191e-09,
      "loss": 302.005,
      "step": 51370
    },
    {
      "epoch": 0.9881624371339828,
      "grad_norm": 403.6882028588385,
      "learning_rate": 7.043301557096982e-09,
      "loss": 303.7258,
      "step": 51380
    },
    {
      "epoch": 0.9883547614697425,
      "grad_norm": 400.42846357418983,
      "learning_rate": 6.816139111484044e-09,
      "loss": 303.64,
      "step": 51390
    },
    {
      "epoch": 0.9885470858055024,
      "grad_norm": 374.76557117762826,
      "learning_rate": 6.592698961818134e-09,
      "loss": 308.6927,
      "step": 51400
    },
    {
      "epoch": 0.9887394101412622,
      "grad_norm": 394.6435981372079,
      "learning_rate": 6.37298119132379e-09,
      "loss": 298.0467,
      "step": 51410
    },
    {
      "epoch": 0.9889317344770221,
      "grad_norm": 391.99315736533697,
      "learning_rate": 6.156985881844435e-09,
      "loss": 302.1709,
      "step": 51420
    },
    {
      "epoch": 0.9891240588127819,
      "grad_norm": 389.3439157971508,
      "learning_rate": 5.944713113833489e-09,
      "loss": 305.8769,
      "step": 51430
    },
    {
      "epoch": 0.9893163831485418,
      "grad_norm": 381.8496468000146,
      "learning_rate": 5.736162966359926e-09,
      "loss": 310.9098,
      "step": 51440
    },
    {
      "epoch": 0.9895087074843015,
      "grad_norm": 393.8951535453929,
      "learning_rate": 5.531335517104941e-09,
      "loss": 303.4278,
      "step": 51450
    },
    {
      "epoch": 0.9897010318200613,
      "grad_norm": 409.8679255772124,
      "learning_rate": 5.3302308423641704e-09,
      "loss": 305.5253,
      "step": 51460
    },
    {
      "epoch": 0.9898933561558212,
      "grad_norm": 406.33180432288617,
      "learning_rate": 5.132849017044361e-09,
      "loss": 311.3281,
      "step": 51470
    },
    {
      "epoch": 0.990085680491581,
      "grad_norm": 430.82599841616536,
      "learning_rate": 4.939190114666703e-09,
      "loss": 317.1013,
      "step": 51480
    },
    {
      "epoch": 0.9902780048273409,
      "grad_norm": 385.7016786372432,
      "learning_rate": 4.749254207367937e-09,
      "loss": 301.6066,
      "step": 51490
    },
    {
      "epoch": 0.9904703291631006,
      "grad_norm": 367.50235603668244,
      "learning_rate": 4.563041365894805e-09,
      "loss": 307.9614,
      "step": 51500
    },
    {
      "epoch": 0.9906626534988605,
      "grad_norm": 437.3992355238702,
      "learning_rate": 4.380551659608489e-09,
      "loss": 308.0566,
      "step": 51510
    },
    {
      "epoch": 0.9908549778346203,
      "grad_norm": 395.9166235781837,
      "learning_rate": 4.2017851564835065e-09,
      "loss": 305.9135,
      "step": 51520
    },
    {
      "epoch": 0.9910473021703802,
      "grad_norm": 390.4537718561425,
      "learning_rate": 4.026741923107702e-09,
      "loss": 300.276,
      "step": 51530
    },
    {
      "epoch": 0.99123962650614,
      "grad_norm": 445.79770036122716,
      "learning_rate": 3.855422024681144e-09,
      "loss": 312.3877,
      "step": 51540
    },
    {
      "epoch": 0.9914319508418998,
      "grad_norm": 361.8853476592868,
      "learning_rate": 3.6878255250183406e-09,
      "loss": 295.5108,
      "step": 51550
    },
    {
      "epoch": 0.9916242751776596,
      "grad_norm": 365.2106398601837,
      "learning_rate": 3.5239524865460227e-09,
      "loss": 301.0301,
      "step": 51560
    },
    {
      "epoch": 0.9918165995134194,
      "grad_norm": 412.5798797133672,
      "learning_rate": 3.363802970304253e-09,
      "loss": 316.894,
      "step": 51570
    },
    {
      "epoch": 0.9920089238491793,
      "grad_norm": 389.77594090198704,
      "learning_rate": 3.207377035946424e-09,
      "loss": 308.6997,
      "step": 51580
    },
    {
      "epoch": 0.992201248184939,
      "grad_norm": 384.3906314377048,
      "learning_rate": 3.05467474173704e-09,
      "loss": 308.2544,
      "step": 51590
    },
    {
      "epoch": 0.9923935725206989,
      "grad_norm": 416.19656893957654,
      "learning_rate": 2.9056961445572686e-09,
      "loss": 306.6549,
      "step": 51600
    },
    {
      "epoch": 0.9925858968564587,
      "grad_norm": 398.3854214826971,
      "learning_rate": 2.7604412998982754e-09,
      "loss": 307.95,
      "step": 51610
    },
    {
      "epoch": 0.9927782211922186,
      "grad_norm": 410.56161503258375,
      "learning_rate": 2.61891026186456e-09,
      "loss": 303.2625,
      "step": 51620
    },
    {
      "epoch": 0.9929705455279784,
      "grad_norm": 406.7332818404458,
      "learning_rate": 2.4811030831739525e-09,
      "loss": 318.5634,
      "step": 51630
    },
    {
      "epoch": 0.9931628698637383,
      "grad_norm": 415.37002368973316,
      "learning_rate": 2.347019815158724e-09,
      "loss": 312.0445,
      "step": 51640
    },
    {
      "epoch": 0.993355194199498,
      "grad_norm": 434.88001844916295,
      "learning_rate": 2.216660507762258e-09,
      "loss": 309.0632,
      "step": 51650
    },
    {
      "epoch": 0.9935475185352579,
      "grad_norm": 401.1180103953426,
      "learning_rate": 2.0900252095401583e-09,
      "loss": 302.0537,
      "step": 51660
    },
    {
      "epoch": 0.9937398428710177,
      "grad_norm": 424.51657310923054,
      "learning_rate": 1.9671139676624707e-09,
      "loss": 308.1321,
      "step": 51670
    },
    {
      "epoch": 0.9939321672067775,
      "grad_norm": 399.2239784555877,
      "learning_rate": 1.8479268279125717e-09,
      "loss": 308.0895,
      "step": 51680
    },
    {
      "epoch": 0.9941244915425373,
      "grad_norm": 398.4795944942946,
      "learning_rate": 1.73246383468495e-09,
      "loss": 300.3031,
      "step": 51690
    },
    {
      "epoch": 0.9943168158782971,
      "grad_norm": 377.56698551165994,
      "learning_rate": 1.6207250309874246e-09,
      "loss": 306.1907,
      "step": 51700
    },
    {
      "epoch": 0.994509140214057,
      "grad_norm": 448.83962938024837,
      "learning_rate": 1.512710458442257e-09,
      "loss": 312.7012,
      "step": 51710
    },
    {
      "epoch": 0.9947014645498168,
      "grad_norm": 400.69775770345666,
      "learning_rate": 1.408420157280599e-09,
      "loss": 315.2197,
      "step": 51720
    },
    {
      "epoch": 0.9948937888855767,
      "grad_norm": 376.03710073961975,
      "learning_rate": 1.3078541663502642e-09,
      "loss": 309.4352,
      "step": 51730
    },
    {
      "epoch": 0.9950861132213364,
      "grad_norm": 396.9308645386203,
      "learning_rate": 1.2110125231112879e-09,
      "loss": 311.7786,
      "step": 51740
    },
    {
      "epoch": 0.9952784375570963,
      "grad_norm": 415.3260137729683,
      "learning_rate": 1.117895263633706e-09,
      "loss": 307.5083,
      "step": 51750
    },
    {
      "epoch": 0.9954707618928561,
      "grad_norm": 408.5010771689888,
      "learning_rate": 1.0285024226042162e-09,
      "loss": 308.7961,
      "step": 51760
    },
    {
      "epoch": 0.995663086228616,
      "grad_norm": 376.73186518642893,
      "learning_rate": 9.428340333184072e-10,
      "loss": 300.4339,
      "step": 51770
    },
    {
      "epoch": 0.9958554105643758,
      "grad_norm": 407.29523918210987,
      "learning_rate": 8.608901276874193e-10,
      "loss": 320.3509,
      "step": 51780
    },
    {
      "epoch": 0.9960477349001355,
      "grad_norm": 415.36622235217936,
      "learning_rate": 7.826707362335039e-10,
      "loss": 314.0036,
      "step": 51790
    },
    {
      "epoch": 0.9962400592358954,
      "grad_norm": 399.2257955451147,
      "learning_rate": 7.081758880911338e-10,
      "loss": 328.2634,
      "step": 51800
    },
    {
      "epoch": 0.9964323835716552,
      "grad_norm": 393.6085898693909,
      "learning_rate": 6.374056110103332e-10,
      "loss": 299.7243,
      "step": 51810
    },
    {
      "epoch": 0.9966247079074151,
      "grad_norm": 405.0272867713518,
      "learning_rate": 5.703599313511277e-10,
      "loss": 305.4929,
      "step": 51820
    },
    {
      "epoch": 0.9968170322431749,
      "grad_norm": 397.6661898612281,
      "learning_rate": 5.070388740868737e-10,
      "loss": 308.362,
      "step": 51830
    },
    {
      "epoch": 0.9970093565789347,
      "grad_norm": 404.80826520029643,
      "learning_rate": 4.474424628031493e-10,
      "loss": 308.8702,
      "step": 51840
    },
    {
      "epoch": 0.9972016809146945,
      "grad_norm": 387.30946528310625,
      "learning_rate": 3.9157071969997407e-10,
      "loss": 297.5715,
      "step": 51850
    },
    {
      "epoch": 0.9973940052504544,
      "grad_norm": 428.2876429970752,
      "learning_rate": 3.394236655873684e-10,
      "loss": 310.9495,
      "step": 51860
    },
    {
      "epoch": 0.9975863295862142,
      "grad_norm": 379.36230039227706,
      "learning_rate": 2.910013198886841e-10,
      "loss": 308.731,
      "step": 51870
    },
    {
      "epoch": 0.9977786539219741,
      "grad_norm": 419.4352909245947,
      "learning_rate": 2.4630370064171463e-10,
      "loss": 311.907,
      "step": 51880
    },
    {
      "epoch": 0.9979709782577338,
      "grad_norm": 383.4827735798926,
      "learning_rate": 2.0533082449647467e-10,
      "loss": 305.1834,
      "step": 51890
    },
    {
      "epoch": 0.9981633025934936,
      "grad_norm": 393.88449878209974,
      "learning_rate": 1.6808270671186954e-10,
      "loss": 315.9007,
      "step": 51900
    },
    {
      "epoch": 0.9983556269292535,
      "grad_norm": 443.98924404030754,
      "learning_rate": 1.345593611645768e-10,
      "loss": 303.1187,
      "step": 51910
    },
    {
      "epoch": 0.9985479512650133,
      "grad_norm": 408.95890434584715,
      "learning_rate": 1.0476080034016456e-10,
      "loss": 316.9108,
      "step": 51920
    },
    {
      "epoch": 0.9987402756007732,
      "grad_norm": 400.75925737361706,
      "learning_rate": 7.868703533864264e-11,
      "loss": 307.9794,
      "step": 51930
    },
    {
      "epoch": 0.9989325999365329,
      "grad_norm": 426.5420963544196,
      "learning_rate": 5.633807587224205e-11,
      "loss": 311.6262,
      "step": 51940
    },
    {
      "epoch": 0.9991249242722928,
      "grad_norm": 407.1191060452394,
      "learning_rate": 3.771393026541503e-11,
      "loss": 304.6595,
      "step": 51950
    },
    {
      "epoch": 0.9993172486080526,
      "grad_norm": 391.9785877798069,
      "learning_rate": 2.281460545594527e-11,
      "loss": 301.8431,
      "step": 51960
    },
    {
      "epoch": 0.9995095729438125,
      "grad_norm": 400.3879591656393,
      "learning_rate": 1.164010699272744e-11,
      "loss": 305.2578,
      "step": 51970
    },
    {
      "epoch": 0.9997018972795723,
      "grad_norm": 381.6965270229846,
      "learning_rate": 4.190439037987659e-12,
      "loss": 299.185,
      "step": 51980
    },
    {
      "epoch": 0.9998942216153321,
      "grad_norm": 393.96873763468045,
      "learning_rate": 4.656043661732668e-13,
      "loss": 329.7069,
      "step": 51990
    },
    {
      "epoch": 0.999990383783212,
      "eval_loss": 375.7167663574219,
      "eval_runtime": 60.9941,
      "eval_samples_per_second": 11.411,
      "eval_steps_per_second": 0.361,
      "step": 51995
    },
    {
      "epoch": 0.999990383783212,
      "step": 51995,
      "total_flos": 4.760752445875814e+16,
      "train_loss": 199.8116063359082,
      "train_runtime": 218479.6594,
      "train_samples_per_second": 15.231,
      "train_steps_per_second": 0.238
    }
  ],
  "logging_steps": 10,
  "max_steps": 51995,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.760752445875814e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}