{
  "best_global_step": 48000,
  "best_metric": 4.555622100830078,
  "best_model_checkpoint": "./model_save/checkpoint-48000",
  "epoch": 0.9829418630843897,
  "eval_steps": 2000,
  "global_step": 48000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00020477955480924785,
      "grad_norm": 6.629329204559326,
      "learning_rate": 7.2e-07,
      "loss": 11.4041,
      "step": 10
    },
    {
      "epoch": 0.0004095591096184957,
      "grad_norm": 6.498970985412598,
      "learning_rate": 1.52e-06,
      "loss": 11.3063,
      "step": 20
    },
    {
      "epoch": 0.0006143386644277435,
      "grad_norm": 6.277300834655762,
      "learning_rate": 2.32e-06,
      "loss": 11.0971,
      "step": 30
    },
    {
      "epoch": 0.0008191182192369914,
      "grad_norm": 5.17841911315918,
      "learning_rate": 3.12e-06,
      "loss": 10.8001,
      "step": 40
    },
    {
      "epoch": 0.0010238977740462392,
      "grad_norm": 3.385416269302368,
      "learning_rate": 3.92e-06,
      "loss": 10.4549,
      "step": 50
    },
    {
      "epoch": 0.001228677328855487,
      "grad_norm": 2.175173044204712,
      "learning_rate": 4.72e-06,
      "loss": 10.202,
      "step": 60
    },
    {
      "epoch": 0.001433456883664735,
      "grad_norm": 1.7192872762680054,
      "learning_rate": 5.5200000000000005e-06,
      "loss": 10.0135,
      "step": 70
    },
    {
      "epoch": 0.0016382364384739828,
      "grad_norm": 1.3520652055740356,
      "learning_rate": 6.320000000000001e-06,
      "loss": 9.8705,
      "step": 80
    },
    {
      "epoch": 0.0018430159932832306,
      "grad_norm": 1.2889796495437622,
      "learning_rate": 7.1200000000000004e-06,
      "loss": 9.7625,
      "step": 90
    },
    {
      "epoch": 0.0020477955480924783,
      "grad_norm": 1.2184574604034424,
      "learning_rate": 7.92e-06,
      "loss": 9.6682,
      "step": 100
    },
    {
      "epoch": 0.0022525751029017264,
      "grad_norm": 1.253600001335144,
      "learning_rate": 8.720000000000001e-06,
      "loss": 9.5655,
      "step": 110
    },
    {
      "epoch": 0.002457354657710974,
      "grad_norm": 1.5453048944473267,
      "learning_rate": 9.52e-06,
      "loss": 9.462,
      "step": 120
    },
    {
      "epoch": 0.002662134212520222,
      "grad_norm": 1.1592094898223877,
      "learning_rate": 1.0320000000000001e-05,
      "loss": 9.3696,
      "step": 130
    },
    {
      "epoch": 0.00286691376732947,
      "grad_norm": 1.1517513990402222,
      "learning_rate": 1.112e-05,
      "loss": 9.2833,
      "step": 140
    },
    {
      "epoch": 0.0030716933221387175,
      "grad_norm": 1.1195098161697388,
      "learning_rate": 1.1920000000000001e-05,
      "loss": 9.2143,
      "step": 150
    },
    {
      "epoch": 0.0032764728769479655,
      "grad_norm": 1.066959261894226,
      "learning_rate": 1.2720000000000002e-05,
      "loss": 9.1244,
      "step": 160
    },
    {
      "epoch": 0.0034812524317572135,
      "grad_norm": 1.2716888189315796,
      "learning_rate": 1.352e-05,
      "loss": 9.0943,
      "step": 170
    },
    {
      "epoch": 0.003686031986566461,
      "grad_norm": 1.0497322082519531,
      "learning_rate": 1.432e-05,
      "loss": 8.974,
      "step": 180
    },
    {
      "epoch": 0.003890811541375709,
      "grad_norm": 0.9636754393577576,
      "learning_rate": 1.5120000000000001e-05,
      "loss": 8.903,
      "step": 190
    },
    {
      "epoch": 0.004095591096184957,
      "grad_norm": 0.962215781211853,
      "learning_rate": 1.592e-05,
      "loss": 8.8231,
      "step": 200
    },
    {
      "epoch": 0.004300370650994205,
      "grad_norm": 1.1262612342834473,
      "learning_rate": 1.672e-05,
      "loss": 8.7727,
      "step": 210
    },
    {
      "epoch": 0.004505150205803453,
      "grad_norm": 0.8915610909461975,
      "learning_rate": 1.752e-05,
      "loss": 8.6879,
      "step": 220
    },
    {
      "epoch": 0.0047099297606127,
      "grad_norm": 0.9334535598754883,
      "learning_rate": 1.832e-05,
      "loss": 8.6236,
      "step": 230
    },
    {
      "epoch": 0.004914709315421948,
      "grad_norm": 1.2129932641983032,
      "learning_rate": 1.9120000000000003e-05,
      "loss": 8.572,
      "step": 240
    },
    {
      "epoch": 0.005119488870231196,
      "grad_norm": 2.270386219024658,
      "learning_rate": 1.992e-05,
      "loss": 8.4909,
      "step": 250
    },
    {
      "epoch": 0.005324268425040444,
      "grad_norm": 0.9250027537345886,
      "learning_rate": 2.072e-05,
      "loss": 8.4783,
      "step": 260
    },
    {
      "epoch": 0.0055290479798496914,
      "grad_norm": 0.9208921194076538,
      "learning_rate": 2.152e-05,
      "loss": 8.3651,
      "step": 270
    },
    {
      "epoch": 0.00573382753465894,
      "grad_norm": 0.8561158776283264,
      "learning_rate": 2.2320000000000003e-05,
      "loss": 8.2926,
      "step": 280
    },
    {
      "epoch": 0.0059386070894681875,
      "grad_norm": 0.9620749950408936,
      "learning_rate": 2.312e-05,
      "loss": 8.2353,
      "step": 290
    },
    {
      "epoch": 0.006143386644277435,
      "grad_norm": 0.834102213382721,
      "learning_rate": 2.392e-05,
      "loss": 8.1975,
      "step": 300
    },
    {
      "epoch": 0.0063481661990866835,
      "grad_norm": 0.7779641151428223,
      "learning_rate": 2.472e-05,
      "loss": 8.1244,
      "step": 310
    },
    {
      "epoch": 0.006552945753895931,
      "grad_norm": 0.8238368034362793,
      "learning_rate": 2.552e-05,
      "loss": 8.0652,
      "step": 320
    },
    {
      "epoch": 0.006757725308705179,
      "grad_norm": 0.8252228498458862,
      "learning_rate": 2.632e-05,
      "loss": 8.0452,
      "step": 330
    },
    {
      "epoch": 0.006962504863514427,
      "grad_norm": 0.7772361636161804,
      "learning_rate": 2.712e-05,
      "loss": 8.0057,
      "step": 340
    },
    {
      "epoch": 0.007167284418323675,
      "grad_norm": 0.9958386421203613,
      "learning_rate": 2.792e-05,
      "loss": 7.8924,
      "step": 350
    },
    {
      "epoch": 0.007372063973132922,
      "grad_norm": 0.7524231672286987,
      "learning_rate": 2.8720000000000003e-05,
      "loss": 7.8663,
      "step": 360
    },
    {
      "epoch": 0.007576843527942171,
      "grad_norm": 0.7005847096443176,
      "learning_rate": 2.9520000000000002e-05,
      "loss": 7.8309,
      "step": 370
    },
    {
      "epoch": 0.007781623082751418,
      "grad_norm": 0.844709038734436,
      "learning_rate": 3.0320000000000004e-05,
      "loss": 7.8128,
      "step": 380
    },
    {
      "epoch": 0.007986402637560666,
      "grad_norm": 0.6282876133918762,
      "learning_rate": 3.112e-05,
      "loss": 7.7429,
      "step": 390
    },
    {
      "epoch": 0.008191182192369913,
      "grad_norm": 2.5666661262512207,
      "learning_rate": 3.192e-05,
      "loss": 7.7099,
      "step": 400
    },
    {
      "epoch": 0.008395961747179161,
      "grad_norm": 0.791608989238739,
      "learning_rate": 3.272e-05,
      "loss": 7.7298,
      "step": 410
    },
    {
      "epoch": 0.00860074130198841,
      "grad_norm": 0.7391555905342102,
      "learning_rate": 3.3520000000000004e-05,
      "loss": 7.6316,
      "step": 420
    },
    {
      "epoch": 0.008805520856797658,
      "grad_norm": 0.7813675999641418,
      "learning_rate": 3.4320000000000003e-05,
      "loss": 7.6225,
      "step": 430
    },
    {
      "epoch": 0.009010300411606905,
      "grad_norm": 0.7512689828872681,
      "learning_rate": 3.512e-05,
      "loss": 7.59,
      "step": 440
    },
    {
      "epoch": 0.009215079966416153,
      "grad_norm": 0.7102991342544556,
      "learning_rate": 3.592e-05,
      "loss": 7.5673,
      "step": 450
    },
    {
      "epoch": 0.0094198595212254,
      "grad_norm": 0.8545809388160706,
      "learning_rate": 3.672000000000001e-05,
      "loss": 7.5217,
      "step": 460
    },
    {
      "epoch": 0.009624639076034648,
      "grad_norm": 0.9673335552215576,
      "learning_rate": 3.752e-05,
      "loss": 7.5043,
      "step": 470
    },
    {
      "epoch": 0.009829418630843896,
      "grad_norm": 1.040452003479004,
      "learning_rate": 3.832e-05,
      "loss": 7.5176,
      "step": 480
    },
    {
      "epoch": 0.010034198185653145,
      "grad_norm": 1.3076322078704834,
      "learning_rate": 3.912e-05,
      "loss": 7.4432,
      "step": 490
    },
    {
      "epoch": 0.010238977740462393,
      "grad_norm": 0.9396184682846069,
      "learning_rate": 3.9920000000000004e-05,
      "loss": 7.4431,
      "step": 500
    },
    {
      "epoch": 0.01044375729527164,
      "grad_norm": 0.7993614673614502,
      "learning_rate": 4.072e-05,
      "loss": 7.3762,
      "step": 510
    },
    {
      "epoch": 0.010648536850080888,
      "grad_norm": 1.016692876815796,
      "learning_rate": 4.152e-05,
      "loss": 7.3908,
      "step": 520
    },
    {
      "epoch": 0.010853316404890135,
      "grad_norm": 1.1701678037643433,
      "learning_rate": 4.232e-05,
      "loss": 7.3692,
      "step": 530
    },
    {
      "epoch": 0.011058095959699383,
      "grad_norm": 0.9889584183692932,
      "learning_rate": 4.312000000000001e-05,
      "loss": 7.3655,
      "step": 540
    },
    {
      "epoch": 0.011262875514508632,
      "grad_norm": 1.624703049659729,
      "learning_rate": 4.392e-05,
      "loss": 7.3559,
      "step": 550
    },
    {
      "epoch": 0.01146765506931788,
      "grad_norm": 1.0079149007797241,
      "learning_rate": 4.472e-05,
      "loss": 7.3154,
      "step": 560
    },
    {
      "epoch": 0.011672434624127127,
      "grad_norm": 0.9229303002357483,
      "learning_rate": 4.5520000000000005e-05,
      "loss": 7.2466,
      "step": 570
    },
    {
      "epoch": 0.011877214178936375,
      "grad_norm": 1.0020170211791992,
      "learning_rate": 4.6320000000000004e-05,
      "loss": 7.2919,
      "step": 580
    },
    {
      "epoch": 0.012081993733745622,
      "grad_norm": 1.0504316091537476,
      "learning_rate": 4.712e-05,
      "loss": 7.2288,
      "step": 590
    },
    {
      "epoch": 0.01228677328855487,
      "grad_norm": 1.2128161191940308,
      "learning_rate": 4.792e-05,
      "loss": 7.229,
      "step": 600
    },
    {
      "epoch": 0.01249155284336412,
      "grad_norm": 0.9260151386260986,
      "learning_rate": 4.872000000000001e-05,
      "loss": 7.1875,
      "step": 610
    },
    {
      "epoch": 0.012696332398173367,
      "grad_norm": 1.083197832107544,
      "learning_rate": 4.952e-05,
      "loss": 7.2263,
      "step": 620
    },
    {
      "epoch": 0.012901111952982615,
      "grad_norm": 1.748647689819336,
      "learning_rate": 5.032e-05,
      "loss": 7.1888,
      "step": 630
    },
    {
      "epoch": 0.013105891507791862,
      "grad_norm": 1.4586197137832642,
      "learning_rate": 5.112e-05,
      "loss": 7.1513,
      "step": 640
    },
    {
      "epoch": 0.01331067106260111,
      "grad_norm": 1.279341459274292,
      "learning_rate": 5.1920000000000004e-05,
      "loss": 7.1123,
      "step": 650
    },
    {
      "epoch": 0.013515450617410357,
      "grad_norm": 1.140655755996704,
      "learning_rate": 5.2720000000000003e-05,
      "loss": 7.0848,
      "step": 660
    },
    {
      "epoch": 0.013720230172219605,
      "grad_norm": 1.1404409408569336,
      "learning_rate": 5.352e-05,
      "loss": 7.1542,
      "step": 670
    },
    {
      "epoch": 0.013925009727028854,
      "grad_norm": 1.4052735567092896,
      "learning_rate": 5.432e-05,
      "loss": 7.0715,
      "step": 680
    },
    {
      "epoch": 0.014129789281838102,
      "grad_norm": 1.4490317106246948,
      "learning_rate": 5.512000000000001e-05,
      "loss": 7.1085,
      "step": 690
    },
    {
      "epoch": 0.01433456883664735,
      "grad_norm": 0.9407364130020142,
      "learning_rate": 5.592000000000001e-05,
      "loss": 7.0504,
      "step": 700
    },
    {
      "epoch": 0.014539348391456597,
      "grad_norm": 1.1838454008102417,
      "learning_rate": 5.6720000000000006e-05,
      "loss": 7.0176,
      "step": 710
    },
    {
      "epoch": 0.014744127946265844,
      "grad_norm": 1.223021149635315,
      "learning_rate": 5.7520000000000005e-05,
      "loss": 6.9734,
      "step": 720
    },
    {
      "epoch": 0.014948907501075092,
      "grad_norm": 1.185410737991333,
      "learning_rate": 5.832000000000001e-05,
      "loss": 6.9838,
      "step": 730
    },
    {
      "epoch": 0.015153687055884341,
      "grad_norm": 1.4732284545898438,
      "learning_rate": 5.9119999999999996e-05,
      "loss": 6.9732,
      "step": 740
    },
    {
      "epoch": 0.015358466610693589,
      "grad_norm": 1.0380061864852905,
      "learning_rate": 5.9919999999999996e-05,
      "loss": 7.0131,
      "step": 750
    },
    {
      "epoch": 0.015563246165502836,
      "grad_norm": 1.2068144083023071,
      "learning_rate": 6.072e-05,
      "loss": 6.9211,
      "step": 760
    },
    {
      "epoch": 0.015768025720312086,
      "grad_norm": 1.0961358547210693,
      "learning_rate": 6.152e-05,
      "loss": 6.9088,
      "step": 770
    },
    {
      "epoch": 0.01597280527512133,
      "grad_norm": 1.4672534465789795,
      "learning_rate": 6.232e-05,
      "loss": 6.9141,
      "step": 780
    },
    {
      "epoch": 0.01617758482993058,
      "grad_norm": 1.2274079322814941,
      "learning_rate": 6.312e-05,
      "loss": 6.888,
      "step": 790
    },
    {
      "epoch": 0.016382364384739827,
      "grad_norm": 1.1605645418167114,
      "learning_rate": 6.392e-05,
      "loss": 6.9103,
      "step": 800
    },
    {
      "epoch": 0.016587143939549076,
      "grad_norm": 1.0360841751098633,
      "learning_rate": 6.472e-05,
      "loss": 6.8925,
      "step": 810
    },
    {
      "epoch": 0.016791923494358322,
      "grad_norm": 1.1419285535812378,
      "learning_rate": 6.552000000000001e-05,
      "loss": 6.8396,
      "step": 820
    },
    {
      "epoch": 0.01699670304916757,
      "grad_norm": 1.2122325897216797,
      "learning_rate": 6.632e-05,
      "loss": 6.8401,
      "step": 830
    },
    {
      "epoch": 0.01720148260397682,
      "grad_norm": 1.0237847566604614,
      "learning_rate": 6.712000000000001e-05,
      "loss": 6.8111,
      "step": 840
    },
    {
      "epoch": 0.017406262158786066,
      "grad_norm": 1.3663208484649658,
      "learning_rate": 6.792e-05,
      "loss": 6.842,
      "step": 850
    },
    {
      "epoch": 0.017611041713595316,
      "grad_norm": 1.6236571073532104,
      "learning_rate": 6.872e-05,
      "loss": 6.8099,
      "step": 860
    },
    {
      "epoch": 0.01781582126840456,
      "grad_norm": 1.5716248750686646,
      "learning_rate": 6.952000000000001e-05,
      "loss": 6.8146,
      "step": 870
    },
    {
      "epoch": 0.01802060082321381,
      "grad_norm": 1.4074079990386963,
      "learning_rate": 7.032e-05,
      "loss": 6.7876,
      "step": 880
    },
    {
      "epoch": 0.018225380378023057,
      "grad_norm": 1.0738171339035034,
      "learning_rate": 7.112000000000001e-05,
      "loss": 6.8033,
      "step": 890
    },
    {
      "epoch": 0.018430159932832306,
      "grad_norm": 1.4551218748092651,
      "learning_rate": 7.192e-05,
      "loss": 6.7627,
      "step": 900
    },
    {
      "epoch": 0.018634939487641555,
      "grad_norm": 1.0792328119277954,
      "learning_rate": 7.272e-05,
      "loss": 6.7315,
      "step": 910
    },
    {
      "epoch": 0.0188397190424508,
      "grad_norm": 1.079581379890442,
      "learning_rate": 7.352e-05,
      "loss": 6.7201,
      "step": 920
    },
    {
      "epoch": 0.01904449859726005,
      "grad_norm": 1.7889859676361084,
      "learning_rate": 7.432e-05,
      "loss": 6.7276,
      "step": 930
    },
    {
      "epoch": 0.019249278152069296,
      "grad_norm": 1.1572691202163696,
      "learning_rate": 7.512e-05,
      "loss": 6.6996,
      "step": 940
    },
    {
      "epoch": 0.019454057706878546,
      "grad_norm": 0.9270385503768921,
      "learning_rate": 7.592e-05,
      "loss": 6.7369,
      "step": 950
    },
    {
      "epoch": 0.01965883726168779,
      "grad_norm": 1.431523084640503,
      "learning_rate": 7.672e-05,
      "loss": 6.6738,
      "step": 960
    },
    {
      "epoch": 0.01986361681649704,
      "grad_norm": 1.3019062280654907,
      "learning_rate": 7.752e-05,
      "loss": 6.7316,
      "step": 970
    },
    {
      "epoch": 0.02006839637130629,
      "grad_norm": 0.9801391363143921,
      "learning_rate": 7.832000000000001e-05,
      "loss": 6.6684,
      "step": 980
    },
    {
      "epoch": 0.020273175926115536,
      "grad_norm": 1.3378275632858276,
      "learning_rate": 7.912e-05,
      "loss": 6.6285,
      "step": 990
    },
    {
      "epoch": 0.020477955480924785,
      "grad_norm": 1.1381784677505493,
      "learning_rate": 7.992000000000001e-05,
      "loss": 6.6235,
      "step": 1000
    },
    {
      "epoch": 0.02068273503573403,
      "grad_norm": 1.0569130182266235,
      "learning_rate": 8.072000000000001e-05,
      "loss": 6.61,
      "step": 1010
    },
    {
      "epoch": 0.02088751459054328,
      "grad_norm": 1.4086412191390991,
      "learning_rate": 8.152e-05,
      "loss": 6.6142,
      "step": 1020
    },
    {
      "epoch": 0.02109229414535253,
      "grad_norm": 1.226737141609192,
      "learning_rate": 8.232000000000001e-05,
      "loss": 6.6019,
      "step": 1030
    },
    {
      "epoch": 0.021297073700161775,
      "grad_norm": 1.1184046268463135,
      "learning_rate": 8.312e-05,
      "loss": 6.5787,
      "step": 1040
    },
    {
      "epoch": 0.021501853254971025,
      "grad_norm": 1.263479232788086,
      "learning_rate": 8.392e-05,
      "loss": 6.6065,
      "step": 1050
    },
    {
      "epoch": 0.02170663280978027,
      "grad_norm": 1.3066644668579102,
      "learning_rate": 8.472e-05,
      "loss": 6.5692,
      "step": 1060
    },
    {
      "epoch": 0.02191141236458952,
      "grad_norm": 1.0745807886123657,
      "learning_rate": 8.552e-05,
      "loss": 6.5497,
      "step": 1070
    },
    {
      "epoch": 0.022116191919398766,
      "grad_norm": 1.042258381843567,
      "learning_rate": 8.632e-05,
      "loss": 6.5626,
      "step": 1080
    },
    {
      "epoch": 0.022320971474208015,
      "grad_norm": 1.2225663661956787,
      "learning_rate": 8.712e-05,
      "loss": 6.5448,
      "step": 1090
    },
    {
      "epoch": 0.022525751029017264,
      "grad_norm": 1.6050411462783813,
      "learning_rate": 8.792e-05,
      "loss": 6.5282,
      "step": 1100
    },
    {
      "epoch": 0.02273053058382651,
      "grad_norm": 1.11445951461792,
      "learning_rate": 8.872e-05,
      "loss": 6.5315,
      "step": 1110
    },
    {
      "epoch": 0.02293531013863576,
      "grad_norm": 1.0947785377502441,
      "learning_rate": 8.952000000000001e-05,
      "loss": 6.5531,
      "step": 1120
    },
    {
      "epoch": 0.023140089693445005,
      "grad_norm": 1.2313827276229858,
      "learning_rate": 9.032e-05,
      "loss": 6.4887,
      "step": 1130
    },
    {
      "epoch": 0.023344869248254255,
      "grad_norm": 1.2121127843856812,
      "learning_rate": 9.112000000000001e-05,
      "loss": 6.4987,
      "step": 1140
    },
    {
      "epoch": 0.0235496488030635,
      "grad_norm": 1.3728684186935425,
      "learning_rate": 9.192e-05,
      "loss": 6.4936,
      "step": 1150
    },
    {
      "epoch": 0.02375442835787275,
      "grad_norm": 1.3086248636245728,
      "learning_rate": 9.272e-05,
      "loss": 6.5013,
      "step": 1160
    },
    {
      "epoch": 0.023959207912682,
      "grad_norm": 1.3813849687576294,
      "learning_rate": 9.352000000000001e-05,
      "loss": 6.4949,
      "step": 1170
    },
    {
      "epoch": 0.024163987467491245,
      "grad_norm": 1.3366695642471313,
      "learning_rate": 9.432e-05,
      "loss": 6.4448,
      "step": 1180
    },
    {
      "epoch": 0.024368767022300494,
      "grad_norm": 1.1331437826156616,
      "learning_rate": 9.512000000000001e-05,
      "loss": 6.5141,
      "step": 1190
    },
    {
      "epoch": 0.02457354657710974,
      "grad_norm": 1.0496031045913696,
      "learning_rate": 9.592e-05,
      "loss": 6.4705,
      "step": 1200
    },
    {
      "epoch": 0.02477832613191899,
      "grad_norm": 1.246792197227478,
      "learning_rate": 9.672e-05,
      "loss": 6.4496,
      "step": 1210
    },
    {
      "epoch": 0.02498310568672824,
      "grad_norm": 1.4758825302124023,
      "learning_rate": 9.752e-05,
      "loss": 6.4878,
      "step": 1220
    },
    {
      "epoch": 0.025187885241537485,
      "grad_norm": 1.1107237339019775,
      "learning_rate": 9.832000000000001e-05,
      "loss": 6.419,
      "step": 1230
    },
    {
      "epoch": 0.025392664796346734,
      "grad_norm": 0.9504449367523193,
      "learning_rate": 9.912e-05,
      "loss": 6.4278,
      "step": 1240
    },
    {
      "epoch": 0.02559744435115598,
      "grad_norm": 1.2253371477127075,
      "learning_rate": 9.992e-05,
      "loss": 6.3827,
      "step": 1250
    },
    {
      "epoch": 0.02580222390596523,
      "grad_norm": 0.9804422855377197,
      "learning_rate": 0.00010072000000000001,
      "loss": 6.381,
      "step": 1260
    },
    {
      "epoch": 0.026007003460774475,
      "grad_norm": 2.4996328353881836,
      "learning_rate": 0.00010152000000000002,
      "loss": 6.3951,
      "step": 1270
    },
    {
      "epoch": 0.026211783015583724,
      "grad_norm": 1.1646678447723389,
      "learning_rate": 0.00010232000000000001,
      "loss": 6.4003,
      "step": 1280
    },
    {
      "epoch": 0.026416562570392974,
      "grad_norm": 1.1945327520370483,
      "learning_rate": 0.00010311999999999999,
      "loss": 6.3521,
      "step": 1290
    },
    {
      "epoch": 0.02662134212520222,
      "grad_norm": 1.2656117677688599,
      "learning_rate": 0.00010392,
      "loss": 6.3738,
      "step": 1300
    },
    {
      "epoch": 0.02682612168001147,
      "grad_norm": 1.1956391334533691,
      "learning_rate": 0.00010472,
      "loss": 6.3543,
      "step": 1310
    },
    {
      "epoch": 0.027030901234820714,
      "grad_norm": 1.0304642915725708,
      "learning_rate": 0.00010551999999999999,
      "loss": 6.3451,
      "step": 1320
    },
    {
      "epoch": 0.027235680789629964,
      "grad_norm": 1.1612216234207153,
      "learning_rate": 0.00010632,
      "loss": 6.3446,
      "step": 1330
    },
    {
      "epoch": 0.02744046034443921,
      "grad_norm": 1.0887846946716309,
      "learning_rate": 0.00010712,
      "loss": 6.3251,
      "step": 1340
    },
    {
      "epoch": 0.02764523989924846,
      "grad_norm": 1.2382864952087402,
      "learning_rate": 0.00010792,
      "loss": 6.3444,
      "step": 1350
    },
    {
      "epoch": 0.027850019454057708,
      "grad_norm": 1.3389307260513306,
      "learning_rate": 0.00010872,
      "loss": 6.3271,
      "step": 1360
    },
    {
      "epoch": 0.028054799008866954,
      "grad_norm": 0.9926040768623352,
      "learning_rate": 0.00010952,
      "loss": 6.2983,
      "step": 1370
    },
    {
      "epoch": 0.028259578563676203,
      "grad_norm": 1.138631820678711,
      "learning_rate": 0.00011032,
      "loss": 6.3132,
      "step": 1380
    },
    {
      "epoch": 0.02846435811848545,
      "grad_norm": 1.2368875741958618,
      "learning_rate": 0.00011112,
      "loss": 6.3244,
      "step": 1390
    },
    {
      "epoch": 0.0286691376732947,
      "grad_norm": 0.8522224426269531,
      "learning_rate": 0.00011192,
      "loss": 6.3035,
      "step": 1400
    },
    {
      "epoch": 0.028873917228103944,
      "grad_norm": 1.553653597831726,
      "learning_rate": 0.00011272,
      "loss": 6.2799,
      "step": 1410
    },
    {
      "epoch": 0.029078696782913194,
      "grad_norm": 1.0834499597549438,
      "learning_rate": 0.00011352000000000001,
      "loss": 6.3157,
      "step": 1420
    },
    {
      "epoch": 0.029283476337722443,
      "grad_norm": 0.9788590669631958,
      "learning_rate": 0.00011432,
      "loss": 6.2857,
      "step": 1430
    },
    {
      "epoch": 0.02948825589253169,
      "grad_norm": 1.1679108142852783,
      "learning_rate": 0.00011512000000000001,
      "loss": 6.2885,
      "step": 1440
    },
    {
      "epoch": 0.029693035447340938,
      "grad_norm": 1.140278697013855,
      "learning_rate": 0.00011592,
      "loss": 6.272,
      "step": 1450
    },
    {
      "epoch": 0.029897815002150184,
      "grad_norm": 1.0994911193847656,
      "learning_rate": 0.00011672000000000001,
      "loss": 6.2571,
      "step": 1460
    },
    {
      "epoch": 0.030102594556959433,
      "grad_norm": 1.0351468324661255,
      "learning_rate": 0.00011752000000000001,
      "loss": 6.2471,
      "step": 1470
    },
    {
      "epoch": 0.030307374111768683,
      "grad_norm": 1.0071357488632202,
      "learning_rate": 0.00011832,
      "loss": 6.2214,
      "step": 1480
    },
    {
      "epoch": 0.03051215366657793,
      "grad_norm": 0.8304218649864197,
      "learning_rate": 0.00011912000000000001,
      "loss": 6.2174,
      "step": 1490
    },
    {
      "epoch": 0.030716933221387178,
      "grad_norm": 0.8604931831359863,
      "learning_rate": 0.00011992,
      "loss": 6.2274,
      "step": 1500
    },
    {
      "epoch": 0.030921712776196424,
      "grad_norm": 0.8300955295562744,
      "learning_rate": 0.00012072000000000001,
      "loss": 6.2381,
      "step": 1510
    },
    {
      "epoch": 0.031126492331005673,
      "grad_norm": 1.0048444271087646,
      "learning_rate": 0.00012152000000000002,
      "loss": 6.1938,
      "step": 1520
    },
    {
      "epoch": 0.03133127188581492,
      "grad_norm": 0.9349138736724854,
      "learning_rate": 0.00012232,
      "loss": 6.1735,
      "step": 1530
    },
    {
      "epoch": 0.03153605144062417,
      "grad_norm": 0.8391649127006531,
      "learning_rate": 0.00012312,
      "loss": 6.1767,
      "step": 1540
    },
    {
      "epoch": 0.031740830995433414,
      "grad_norm": 0.8740177750587463,
      "learning_rate": 0.00012392000000000002,
      "loss": 6.1987,
      "step": 1550
    },
    {
      "epoch": 0.03194561055024266,
      "grad_norm": 1.0528550148010254,
      "learning_rate": 0.00012472,
      "loss": 6.185,
      "step": 1560
    },
    {
      "epoch": 0.03215039010505191,
      "grad_norm": 0.8692805767059326,
      "learning_rate": 0.00012552,
      "loss": 6.2316,
      "step": 1570
    },
    {
      "epoch": 0.03235516965986116,
      "grad_norm": 1.4903044700622559,
      "learning_rate": 0.00012632000000000002,
      "loss": 6.1422,
      "step": 1580
    },
    {
      "epoch": 0.032559949214670404,
      "grad_norm": 0.8822526335716248,
      "learning_rate": 0.00012712000000000002,
      "loss": 6.1607,
      "step": 1590
    },
    {
      "epoch": 0.032764728769479654,
      "grad_norm": 0.8137292265892029,
      "learning_rate": 0.00012792,
      "loss": 6.1814,
      "step": 1600
    },
    {
      "epoch": 0.0329695083242889,
      "grad_norm": 0.8272304534912109,
      "learning_rate": 0.00012872,
      "loss": 6.1417,
      "step": 1610
    },
    {
      "epoch": 0.03317428787909815,
      "grad_norm": 1.164596438407898,
      "learning_rate": 0.00012952,
      "loss": 6.1647,
      "step": 1620
    },
    {
      "epoch": 0.0333790674339074,
      "grad_norm": 3.4363532066345215,
      "learning_rate": 0.00013031999999999999,
      "loss": 6.1314,
      "step": 1630
    },
    {
      "epoch": 0.033583846988716644,
      "grad_norm": 0.964524507522583,
      "learning_rate": 0.00013112,
      "loss": 6.1275,
      "step": 1640
    },
    {
      "epoch": 0.03378862654352589,
      "grad_norm": 0.7767365574836731,
      "learning_rate": 0.00013192,
      "loss": 6.1244,
      "step": 1650
    },
    {
      "epoch": 0.03399340609833514,
      "grad_norm": 0.9796528220176697,
      "learning_rate": 0.00013272,
      "loss": 6.1024,
      "step": 1660
    },
    {
      "epoch": 0.03419818565314439,
      "grad_norm": 0.9104465842247009,
      "learning_rate": 0.00013352,
      "loss": 6.0983,
      "step": 1670
    },
    {
      "epoch": 0.03440296520795364,
      "grad_norm": 0.8891839981079102,
      "learning_rate": 0.00013432,
      "loss": 6.1659,
      "step": 1680
    },
    {
      "epoch": 0.03460774476276288,
      "grad_norm": 0.9721062183380127,
      "learning_rate": 0.00013512,
      "loss": 6.1095,
      "step": 1690
    },
    {
      "epoch": 0.03481252431757213,
      "grad_norm": 0.9213342070579529,
      "learning_rate": 0.00013592,
      "loss": 6.1123,
      "step": 1700
    },
    {
      "epoch": 0.03501730387238138,
      "grad_norm": 0.7819191813468933,
      "learning_rate": 0.00013672,
      "loss": 6.0874,
      "step": 1710
    },
    {
      "epoch": 0.03522208342719063,
      "grad_norm": 0.9591174721717834,
      "learning_rate": 0.00013752,
      "loss": 6.0865,
      "step": 1720
    },
    {
      "epoch": 0.035426862981999874,
      "grad_norm": 1.9827978610992432,
      "learning_rate": 0.00013832000000000002,
      "loss": 6.1134,
      "step": 1730
    },
    {
      "epoch": 0.03563164253680912,
      "grad_norm": 0.8233531713485718,
      "learning_rate": 0.00013912,
      "loss": 6.1046,
      "step": 1740
    },
    {
      "epoch": 0.03583642209161837,
      "grad_norm": 0.8631060123443604,
      "learning_rate": 0.00013992,
      "loss": 6.0755,
      "step": 1750
    },
    {
      "epoch": 0.03604120164642762,
      "grad_norm": 0.8472694754600525,
      "learning_rate": 0.00014072,
      "loss": 6.0528,
      "step": 1760
    },
    {
      "epoch": 0.03624598120123687,
      "grad_norm": 0.865189790725708,
      "learning_rate": 0.00014152000000000001,
      "loss": 6.0325,
      "step": 1770
    },
    {
      "epoch": 0.03645076075604611,
      "grad_norm": 0.7865936160087585,
      "learning_rate": 0.00014232,
      "loss": 6.0146,
      "step": 1780
    },
    {
      "epoch": 0.03665554031085536,
      "grad_norm": 0.7830580472946167,
      "learning_rate": 0.00014312,
      "loss": 6.0466,
      "step": 1790
    },
    {
      "epoch": 0.03686031986566461,
      "grad_norm": 0.7788213491439819,
      "learning_rate": 0.00014392000000000002,
      "loss": 6.0575,
      "step": 1800
    },
    {
      "epoch": 0.03706509942047386,
      "grad_norm": 0.7438129782676697,
      "learning_rate": 0.00014472,
      "loss": 6.0617,
      "step": 1810
    },
    {
      "epoch": 0.03726987897528311,
      "grad_norm": 0.8572232127189636,
      "learning_rate": 0.00014552,
      "loss": 6.005,
      "step": 1820
    },
    {
      "epoch": 0.03747465853009235,
      "grad_norm": 3.089932680130005,
      "learning_rate": 0.00014632000000000002,
      "loss": 6.0865,
      "step": 1830
    },
    {
      "epoch": 0.0376794380849016,
      "grad_norm": 0.8924036026000977,
      "learning_rate": 0.00014712000000000001,
      "loss": 6.0572,
      "step": 1840
    },
    {
      "epoch": 0.03788421763971085,
      "grad_norm": 0.9803346991539001,
      "learning_rate": 0.00014792,
      "loss": 6.0735,
      "step": 1850
    },
    {
      "epoch": 0.0380889971945201,
      "grad_norm": 0.7436922788619995,
      "learning_rate": 0.00014872000000000003,
      "loss": 6.001,
      "step": 1860
    },
    {
      "epoch": 0.03829377674932935,
      "grad_norm": 0.8128496408462524,
      "learning_rate": 0.00014952000000000002,
      "loss": 5.9787,
      "step": 1870
    },
    {
      "epoch": 0.03849855630413859,
      "grad_norm": 0.772544801235199,
      "learning_rate": 0.00015032,
      "loss": 5.9811,
      "step": 1880
    },
    {
      "epoch": 0.03870333585894784,
      "grad_norm": 0.7379671931266785,
      "learning_rate": 0.00015112000000000003,
      "loss": 6.0148,
      "step": 1890
    },
    {
      "epoch": 0.03890811541375709,
      "grad_norm": 0.6884296536445618,
      "learning_rate": 0.00015192000000000002,
      "loss": 5.9905,
      "step": 1900
    },
    {
      "epoch": 0.03911289496856634,
      "grad_norm": 0.774961531162262,
      "learning_rate": 0.00015272,
      "loss": 5.9748,
      "step": 1910
    },
    {
      "epoch": 0.03931767452337558,
      "grad_norm": 0.79966139793396,
      "learning_rate": 0.00015352,
      "loss": 6.0226,
      "step": 1920
    },
    {
      "epoch": 0.03952245407818483,
      "grad_norm": 0.783399224281311,
      "learning_rate": 0.00015432,
      "loss": 5.9568,
      "step": 1930
    },
    {
      "epoch": 0.03972723363299408,
      "grad_norm": 0.694803774356842,
      "learning_rate": 0.00015512,
      "loss": 5.9702,
      "step": 1940
    },
    {
      "epoch": 0.03993201318780333,
      "grad_norm": 0.7792879939079285,
      "learning_rate": 0.00015592,
      "loss": 5.9541,
      "step": 1950
    },
    {
      "epoch": 0.04013679274261258,
      "grad_norm": 0.7265322208404541,
      "learning_rate": 0.00015672,
      "loss": 5.9542,
      "step": 1960
    },
    {
      "epoch": 0.04034157229742182,
      "grad_norm": 0.8625994324684143,
      "learning_rate": 0.00015752,
      "loss": 5.9561,
      "step": 1970
    },
    {
      "epoch": 0.04054635185223107,
      "grad_norm": 0.7278620004653931,
      "learning_rate": 0.00015832,
      "loss": 5.9455,
      "step": 1980
    },
    {
      "epoch": 0.04075113140704032,
      "grad_norm": 0.793200671672821,
      "learning_rate": 0.00015912,
      "loss": 5.966,
      "step": 1990
    },
    {
      "epoch": 0.04095591096184957,
      "grad_norm": 1.2044384479522705,
      "learning_rate": 0.00015992,
      "loss": 5.9557,
      "step": 2000
    },
    {
      "epoch": 0.04095591096184957,
      "eval_loss": 5.943562984466553,
      "eval_runtime": 56.2969,
      "eval_samples_per_second": 20.712,
      "eval_steps_per_second": 2.593,
      "step": 2000
    },
    {
      "epoch": 0.04116069051665882,
      "grad_norm": 0.7351077795028687,
      "learning_rate": 0.00016072,
      "loss": 5.9482,
      "step": 2010
    },
    {
      "epoch": 0.04136547007146806,
      "grad_norm": 0.8293958902359009,
      "learning_rate": 0.00016152,
      "loss": 5.9143,
      "step": 2020
    },
    {
      "epoch": 0.04157024962627731,
      "grad_norm": 0.8600688576698303,
      "learning_rate": 0.00016232,
      "loss": 5.9089,
      "step": 2030
    },
    {
      "epoch": 0.04177502918108656,
      "grad_norm": 0.7429504990577698,
      "learning_rate": 0.00016312,
      "loss": 5.8794,
      "step": 2040
    },
    {
      "epoch": 0.04197980873589581,
      "grad_norm": 0.7266488671302795,
      "learning_rate": 0.00016392000000000002,
      "loss": 5.8786,
      "step": 2050
    },
    {
      "epoch": 0.04218458829070506,
      "grad_norm": 0.6896419525146484,
      "learning_rate": 0.00016472,
      "loss": 5.8846,
      "step": 2060
    },
    {
      "epoch": 0.0423893678455143,
      "grad_norm": 0.6630206108093262,
      "learning_rate": 0.00016552,
      "loss": 5.8895,
      "step": 2070
    },
    {
      "epoch": 0.04259414740032355,
      "grad_norm": 0.7611812949180603,
      "learning_rate": 0.00016632000000000002,
      "loss": 5.8851,
      "step": 2080
    },
    {
      "epoch": 0.0427989269551328,
      "grad_norm": 0.7763521671295166,
      "learning_rate": 0.00016712,
      "loss": 5.8954,
      "step": 2090
    },
    {
      "epoch": 0.04300370650994205,
      "grad_norm": 1.1050055027008057,
      "learning_rate": 0.00016792,
      "loss": 5.9114,
      "step": 2100
    },
    {
      "epoch": 0.04320848606475129,
      "grad_norm": 0.7318460941314697,
      "learning_rate": 0.00016872000000000002,
      "loss": 5.8902,
      "step": 2110
    },
    {
      "epoch": 0.04341326561956054,
      "grad_norm": 0.9724841713905334,
      "learning_rate": 0.00016952000000000002,
      "loss": 5.8692,
      "step": 2120
    },
    {
      "epoch": 0.04361804517436979,
      "grad_norm": 0.7182177901268005,
      "learning_rate": 0.00017032,
      "loss": 5.8422,
      "step": 2130
    },
    {
      "epoch": 0.04382282472917904,
      "grad_norm": 0.7816573977470398,
      "learning_rate": 0.00017112,
      "loss": 5.877,
      "step": 2140
    },
    {
      "epoch": 0.04402760428398829,
      "grad_norm": 0.6064931750297546,
      "learning_rate": 0.00017192000000000002,
      "loss": 5.8499,
      "step": 2150
    },
    {
      "epoch": 0.04423238383879753,
      "grad_norm": 0.7619910836219788,
      "learning_rate": 0.00017272,
      "loss": 5.8362,
      "step": 2160
    },
    {
      "epoch": 0.04443716339360678,
      "grad_norm": 0.7827998399734497,
      "learning_rate": 0.00017352,
      "loss": 5.8615,
      "step": 2170
    },
    {
      "epoch": 0.04464194294841603,
      "grad_norm": 0.8352086544036865,
      "learning_rate": 0.00017432000000000003,
      "loss": 5.8117,
      "step": 2180
    },
    {
      "epoch": 0.04484672250322528,
      "grad_norm": 0.7334675788879395,
      "learning_rate": 0.00017512000000000002,
      "loss": 5.8298,
      "step": 2190
    },
    {
      "epoch": 0.04505150205803453,
      "grad_norm": 0.7153773307800293,
      "learning_rate": 0.00017592,
      "loss": 5.8426,
      "step": 2200
    },
    {
      "epoch": 0.04525628161284377,
      "grad_norm": 0.6356229186058044,
      "learning_rate": 0.00017672000000000003,
      "loss": 5.802,
      "step": 2210
    },
    {
      "epoch": 0.04546106116765302,
      "grad_norm": 0.8019924759864807,
      "learning_rate": 0.00017752,
      "loss": 5.8023,
      "step": 2220
    },
    {
      "epoch": 0.04566584072246227,
      "grad_norm": 0.6777639389038086,
      "learning_rate": 0.00017832,
      "loss": 5.795,
      "step": 2230
    },
    {
      "epoch": 0.04587062027727152,
      "grad_norm": 0.7413334846496582,
      "learning_rate": 0.00017912,
      "loss": 5.8084,
      "step": 2240
    },
    {
      "epoch": 0.04607539983208077,
      "grad_norm": 0.6552919745445251,
      "learning_rate": 0.00017992,
      "loss": 5.7766,
      "step": 2250
    },
    {
      "epoch": 0.04628017938689001,
      "grad_norm": 0.713080108165741,
      "learning_rate": 0.00018072,
      "loss": 5.7853,
      "step": 2260
    },
    {
      "epoch": 0.04648495894169926,
      "grad_norm": 0.6190316081047058,
      "learning_rate": 0.00018152,
      "loss": 5.767,
      "step": 2270
    },
    {
      "epoch": 0.04668973849650851,
      "grad_norm": 1.1991640329360962,
      "learning_rate": 0.00018232,
      "loss": 5.7482,
      "step": 2280
    },
    {
      "epoch": 0.04689451805131776,
      "grad_norm": 0.7628166079521179,
      "learning_rate": 0.00018312,
      "loss": 5.7666,
      "step": 2290
    },
    {
      "epoch": 0.047099297606127,
      "grad_norm": 0.6202970743179321,
      "learning_rate": 0.00018392000000000001,
      "loss": 5.7852,
      "step": 2300
    },
    {
      "epoch": 0.04730407716093625,
      "grad_norm": 0.6502846479415894,
      "learning_rate": 0.00018472,
      "loss": 5.7699,
      "step": 2310
    },
    {
      "epoch": 0.0475088567157455,
      "grad_norm": 0.7047054171562195,
      "learning_rate": 0.00018552,
      "loss": 5.7713,
      "step": 2320
    },
    {
      "epoch": 0.04771363627055475,
      "grad_norm": 0.6129122972488403,
      "learning_rate": 0.00018632000000000002,
      "loss": 5.7456,
      "step": 2330
    },
    {
      "epoch": 0.047918415825364,
      "grad_norm": 0.6389017701148987,
      "learning_rate": 0.00018712,
      "loss": 5.7337,
      "step": 2340
    },
    {
      "epoch": 0.04812319538017324,
      "grad_norm": 0.7061182260513306,
      "learning_rate": 0.00018792,
      "loss": 5.7949,
      "step": 2350
    },
    {
      "epoch": 0.04832797493498249,
      "grad_norm": 0.6337869167327881,
      "learning_rate": 0.00018872,
      "loss": 5.762,
      "step": 2360
    },
    {
      "epoch": 0.04853275448979174,
      "grad_norm": 0.6508966684341431,
      "learning_rate": 0.00018952000000000002,
      "loss": 5.7238,
      "step": 2370
    },
    {
      "epoch": 0.04873753404460099,
      "grad_norm": 0.6663705706596375,
      "learning_rate": 0.00019032,
      "loss": 5.7431,
      "step": 2380
    },
    {
      "epoch": 0.04894231359941024,
      "grad_norm": 1.348676085472107,
      "learning_rate": 0.00019112,
      "loss": 5.6784,
      "step": 2390
    },
    {
      "epoch": 0.04914709315421948,
      "grad_norm": 0.9762350916862488,
      "learning_rate": 0.00019192000000000002,
      "loss": 5.6957,
      "step": 2400
    },
    {
      "epoch": 0.04935187270902873,
      "grad_norm": 0.8128533959388733,
      "learning_rate": 0.00019272,
      "loss": 5.7294,
      "step": 2410
    },
    {
      "epoch": 0.04955665226383798,
      "grad_norm": 0.6565982699394226,
      "learning_rate": 0.00019352,
      "loss": 5.7049,
      "step": 2420
    },
    {
      "epoch": 0.04976143181864723,
      "grad_norm": 0.680180549621582,
      "learning_rate": 0.00019432000000000002,
      "loss": 5.7156,
      "step": 2430
    },
    {
      "epoch": 0.04996621137345648,
      "grad_norm": 0.6522653698921204,
      "learning_rate": 0.00019512000000000002,
      "loss": 5.73,
      "step": 2440
    },
    {
      "epoch": 0.05017099092826572,
      "grad_norm": 0.6685560941696167,
      "learning_rate": 0.00019592,
      "loss": 5.7061,
      "step": 2450
    },
    {
      "epoch": 0.05037577048307497,
      "grad_norm": 0.5945119261741638,
      "learning_rate": 0.00019672000000000003,
      "loss": 5.7045,
      "step": 2460
    },
    {
      "epoch": 0.05058055003788422,
      "grad_norm": 0.645095944404602,
      "learning_rate": 0.00019752000000000002,
      "loss": 5.6842,
      "step": 2470
    },
    {
      "epoch": 0.05078532959269347,
      "grad_norm": 0.666318953037262,
      "learning_rate": 0.00019832,
      "loss": 5.7015,
      "step": 2480
    },
    {
      "epoch": 0.05099010914750271,
      "grad_norm": 0.5758560299873352,
      "learning_rate": 0.00019912,
      "loss": 5.6524,
      "step": 2490
    },
    {
      "epoch": 0.05119488870231196,
      "grad_norm": 0.5842757225036621,
      "learning_rate": 0.00019992000000000002,
      "loss": 5.664,
      "step": 2500
    },
    {
      "epoch": 0.05139966825712121,
      "grad_norm": 0.6742421388626099,
      "learning_rate": 0.00020072000000000002,
      "loss": 5.6619,
      "step": 2510
    },
    {
      "epoch": 0.05160444781193046,
      "grad_norm": 0.6421381235122681,
      "learning_rate": 0.00020152,
      "loss": 5.6534,
      "step": 2520
    },
    {
      "epoch": 0.05180922736673971,
      "grad_norm": 1.3708223104476929,
      "learning_rate": 0.00020232000000000003,
      "loss": 5.6575,
      "step": 2530
    },
    {
      "epoch": 0.05201400692154895,
      "grad_norm": 0.655500054359436,
      "learning_rate": 0.00020312000000000002,
      "loss": 5.6502,
      "step": 2540
    },
    {
      "epoch": 0.0522187864763582,
      "grad_norm": 0.6765331625938416,
      "learning_rate": 0.00020392,
      "loss": 5.6581,
      "step": 2550
    },
    {
      "epoch": 0.05242356603116745,
      "grad_norm": 0.6311305165290833,
      "learning_rate": 0.00020472000000000003,
      "loss": 5.6377,
      "step": 2560
    },
    {
      "epoch": 0.0526283455859767,
      "grad_norm": 0.5618401169776917,
      "learning_rate": 0.00020552000000000002,
      "loss": 5.6092,
      "step": 2570
    },
    {
      "epoch": 0.05283312514078595,
      "grad_norm": 0.5681129693984985,
      "learning_rate": 0.00020632000000000002,
      "loss": 5.6378,
      "step": 2580
    },
    {
      "epoch": 0.05303790469559519,
      "grad_norm": 0.6466996073722839,
      "learning_rate": 0.00020712000000000004,
      "loss": 5.6487,
      "step": 2590
    },
    {
      "epoch": 0.05324268425040444,
      "grad_norm": 0.5487588047981262,
      "learning_rate": 0.00020792000000000003,
      "loss": 5.6589,
      "step": 2600
    },
    {
      "epoch": 0.05344746380521369,
      "grad_norm": 2.6463348865509033,
      "learning_rate": 0.00020872000000000002,
      "loss": 5.5857,
      "step": 2610
    },
    {
      "epoch": 0.05365224336002294,
      "grad_norm": 0.6164767146110535,
      "learning_rate": 0.00020952000000000004,
      "loss": 5.6318,
      "step": 2620
    },
    {
      "epoch": 0.05385702291483219,
      "grad_norm": 0.5371426343917847,
      "learning_rate": 0.00021032000000000003,
      "loss": 5.6176,
      "step": 2630
    },
    {
      "epoch": 0.05406180246964143,
      "grad_norm": 0.8848806619644165,
      "learning_rate": 0.00021112000000000003,
      "loss": 5.643,
      "step": 2640
    },
    {
      "epoch": 0.05426658202445068,
      "grad_norm": 0.5556992888450623,
      "learning_rate": 0.00021192000000000002,
      "loss": 5.6202,
      "step": 2650
    },
    {
      "epoch": 0.05447136157925993,
      "grad_norm": 0.6171382069587708,
      "learning_rate": 0.00021272000000000004,
      "loss": 5.6159,
      "step": 2660
    },
    {
      "epoch": 0.05467614113406918,
      "grad_norm": 0.5811936259269714,
      "learning_rate": 0.00021352000000000003,
      "loss": 5.6064,
      "step": 2670
    },
    {
      "epoch": 0.05488092068887842,
      "grad_norm": 0.5863711833953857,
      "learning_rate": 0.00021432000000000002,
      "loss": 5.5979,
      "step": 2680
    },
    {
      "epoch": 0.05508570024368767,
      "grad_norm": 0.7278106808662415,
      "learning_rate": 0.00021512,
      "loss": 5.6075,
      "step": 2690
    },
    {
      "epoch": 0.05529047979849692,
      "grad_norm": 0.5531920790672302,
      "learning_rate": 0.00021591999999999998,
      "loss": 5.5884,
      "step": 2700
    },
    {
      "epoch": 0.05549525935330617,
      "grad_norm": 0.5820106863975525,
      "learning_rate": 0.00021672,
      "loss": 5.6176,
      "step": 2710
    },
    {
      "epoch": 0.055700038908115417,
      "grad_norm": 0.5418875217437744,
      "learning_rate": 0.00021752,
      "loss": 5.5852,
      "step": 2720
    },
    {
      "epoch": 0.05590481846292466,
      "grad_norm": 0.6495028734207153,
      "learning_rate": 0.00021831999999999998,
      "loss": 5.5635,
      "step": 2730
    },
    {
      "epoch": 0.05610959801773391,
      "grad_norm": 0.5675711035728455,
      "learning_rate": 0.00021912,
      "loss": 5.5842,
      "step": 2740
    },
    {
      "epoch": 0.05631437757254316,
      "grad_norm": 0.5400895476341248,
      "learning_rate": 0.00021992,
      "loss": 5.591,
      "step": 2750
    },
    {
      "epoch": 0.05651915712735241,
      "grad_norm": 0.5396080613136292,
      "learning_rate": 0.00022072,
      "loss": 5.5988,
      "step": 2760
    },
    {
      "epoch": 0.056723936682161656,
      "grad_norm": 0.5995794534683228,
      "learning_rate": 0.00022152,
      "loss": 5.5976,
      "step": 2770
    },
    {
      "epoch": 0.0569287162369709,
      "grad_norm": 0.7403063774108887,
      "learning_rate": 0.00022232,
      "loss": 5.6153,
      "step": 2780
    },
    {
      "epoch": 0.05713349579178015,
      "grad_norm": 0.4941953718662262,
      "learning_rate": 0.00022312,
      "loss": 5.5713,
      "step": 2790
    },
    {
      "epoch": 0.0573382753465894,
      "grad_norm": 0.5359123945236206,
      "learning_rate": 0.00022391999999999998,
      "loss": 5.5815,
      "step": 2800
    },
    {
      "epoch": 0.057543054901398646,
      "grad_norm": 0.5588081479072571,
      "learning_rate": 0.00022472,
      "loss": 5.5127,
      "step": 2810
    },
    {
      "epoch": 0.05774783445620789,
      "grad_norm": 0.5440400242805481,
      "learning_rate": 0.00022552,
      "loss": 5.5131,
      "step": 2820
    },
    {
      "epoch": 0.05795261401101714,
      "grad_norm": 0.5617867112159729,
      "learning_rate": 0.00022632,
      "loss": 5.559,
      "step": 2830
    },
    {
      "epoch": 0.05815739356582639,
      "grad_norm": 0.5289216637611389,
      "learning_rate": 0.00022712,
      "loss": 5.5476,
      "step": 2840
    },
    {
      "epoch": 0.05836217312063564,
      "grad_norm": 0.5092751383781433,
      "learning_rate": 0.00022792,
      "loss": 5.5619,
      "step": 2850
    },
    {
      "epoch": 0.058566952675444886,
      "grad_norm": 0.49928438663482666,
      "learning_rate": 0.00022872,
      "loss": 5.5425,
      "step": 2860
    },
    {
      "epoch": 0.05877173223025413,
      "grad_norm": 0.5320514440536499,
      "learning_rate": 0.00022952,
      "loss": 5.5353,
      "step": 2870
    },
    {
      "epoch": 0.05897651178506338,
      "grad_norm": 0.7999703288078308,
      "learning_rate": 0.00023032,
      "loss": 5.5285,
      "step": 2880
    },
    {
      "epoch": 0.05918129133987263,
      "grad_norm": 0.5325676798820496,
      "learning_rate": 0.00023112,
      "loss": 5.5579,
      "step": 2890
    },
    {
      "epoch": 0.059386070894681876,
      "grad_norm": 0.5199950337409973,
      "learning_rate": 0.00023192000000000002,
      "loss": 5.5152,
      "step": 2900
    },
    {
      "epoch": 0.059590850449491126,
      "grad_norm": 0.5351914763450623,
      "learning_rate": 0.00023272,
      "loss": 5.5453,
      "step": 2910
    },
    {
      "epoch": 0.05979563000430037,
      "grad_norm": 1.4484779834747314,
      "learning_rate": 0.00023352,
      "loss": 5.5814,
      "step": 2920
    },
    {
      "epoch": 0.06000040955910962,
      "grad_norm": 0.6165844202041626,
      "learning_rate": 0.00023432,
      "loss": 5.5068,
      "step": 2930
    },
    {
      "epoch": 0.06020518911391887,
      "grad_norm": 0.5223903656005859,
      "learning_rate": 0.00023512,
      "loss": 5.5456,
      "step": 2940
    },
    {
      "epoch": 0.060409968668728116,
      "grad_norm": 0.45100030303001404,
      "learning_rate": 0.00023592,
      "loss": 5.4977,
      "step": 2950
    },
    {
      "epoch": 0.060614748223537365,
      "grad_norm": 0.5048749446868896,
      "learning_rate": 0.00023672,
      "loss": 5.503,
      "step": 2960
    },
    {
      "epoch": 0.06081952777834661,
      "grad_norm": 0.47033455967903137,
      "learning_rate": 0.00023752000000000002,
      "loss": 5.5039,
      "step": 2970
    },
    {
      "epoch": 0.06102430733315586,
      "grad_norm": 0.5028645396232605,
      "learning_rate": 0.00023832,
      "loss": 5.4865,
      "step": 2980
    },
    {
      "epoch": 0.061229086887965106,
      "grad_norm": 0.516358494758606,
      "learning_rate": 0.00023912,
      "loss": 5.5025,
      "step": 2990
    },
    {
      "epoch": 0.061433866442774356,
      "grad_norm": 0.49913960695266724,
      "learning_rate": 0.00023992000000000002,
      "loss": 5.4662,
      "step": 3000
    },
    {
      "epoch": 0.0616386459975836,
      "grad_norm": 0.5260842442512512,
      "learning_rate": 0.00024072,
      "loss": 5.5041,
      "step": 3010
    },
    {
      "epoch": 0.06184342555239285,
      "grad_norm": 0.5125996470451355,
      "learning_rate": 0.00024152,
      "loss": 5.4991,
      "step": 3020
    },
    {
      "epoch": 0.0620482051072021,
      "grad_norm": 0.5200065970420837,
      "learning_rate": 0.00024232000000000002,
      "loss": 5.5116,
      "step": 3030
    },
    {
      "epoch": 0.062252984662011346,
      "grad_norm": 0.4517768919467926,
      "learning_rate": 0.00024312000000000002,
      "loss": 5.4968,
      "step": 3040
    },
    {
      "epoch": 0.062457764216820595,
      "grad_norm": 0.8140433430671692,
      "learning_rate": 0.00024392,
      "loss": 5.4922,
      "step": 3050
    },
    {
      "epoch": 0.06266254377162984,
      "grad_norm": 0.509969174861908,
      "learning_rate": 0.00024472000000000003,
      "loss": 5.4785,
      "step": 3060
    },
    {
      "epoch": 0.06286732332643909,
      "grad_norm": 0.4542042016983032,
      "learning_rate": 0.00024552,
      "loss": 5.5198,
      "step": 3070
    },
    {
      "epoch": 0.06307210288124834,
      "grad_norm": 0.46822962164878845,
      "learning_rate": 0.00024632,
      "loss": 5.4635,
      "step": 3080
    },
    {
      "epoch": 0.06327688243605759,
      "grad_norm": 0.47385135293006897,
      "learning_rate": 0.00024712000000000003,
      "loss": 5.4718,
      "step": 3090
    },
    {
      "epoch": 0.06348166199086683,
      "grad_norm": 0.5590242743492126,
      "learning_rate": 0.00024792,
      "loss": 5.4687,
      "step": 3100
    },
    {
      "epoch": 0.06368644154567608,
      "grad_norm": 0.579818844795227,
      "learning_rate": 0.00024872,
      "loss": 5.4881,
      "step": 3110
    },
    {
      "epoch": 0.06389122110048533,
      "grad_norm": 0.5298537611961365,
      "learning_rate": 0.00024952000000000004,
      "loss": 5.4568,
      "step": 3120
    },
    {
      "epoch": 0.06409600065529457,
      "grad_norm": 0.4780171513557434,
      "learning_rate": 0.00025032,
      "loss": 5.472,
      "step": 3130
    },
    {
      "epoch": 0.06430078021010383,
      "grad_norm": 1.5736125707626343,
      "learning_rate": 0.00025112,
      "loss": 5.5136,
      "step": 3140
    },
    {
      "epoch": 0.06450555976491307,
      "grad_norm": 0.531993567943573,
      "learning_rate": 0.00025192000000000004,
      "loss": 5.4582,
      "step": 3150
    },
    {
      "epoch": 0.06471033931972232,
      "grad_norm": 0.49365732073783875,
      "learning_rate": 0.00025272,
      "loss": 5.4815,
      "step": 3160
    },
    {
      "epoch": 0.06491511887453157,
      "grad_norm": 0.4865797162055969,
      "learning_rate": 0.00025352,
      "loss": 5.4585,
      "step": 3170
    },
    {
      "epoch": 0.06511989842934081,
      "grad_norm": 0.5386083722114563,
      "learning_rate": 0.00025432000000000005,
      "loss": 5.4386,
      "step": 3180
    },
    {
      "epoch": 0.06532467798415006,
      "grad_norm": 0.5513334274291992,
      "learning_rate": 0.00025512,
      "loss": 5.4632,
      "step": 3190
    },
    {
      "epoch": 0.06552945753895931,
      "grad_norm": 0.46924030780792236,
      "learning_rate": 0.00025592000000000003,
      "loss": 5.475,
      "step": 3200
    },
    {
      "epoch": 0.06573423709376856,
      "grad_norm": 0.49195021390914917,
      "learning_rate": 0.00025672000000000005,
      "loss": 5.458,
      "step": 3210
    },
    {
      "epoch": 0.0659390166485778,
      "grad_norm": 0.7491171956062317,
      "learning_rate": 0.00025752,
      "loss": 5.4291,
      "step": 3220
    },
    {
      "epoch": 0.06614379620338705,
      "grad_norm": 0.4588443636894226,
      "learning_rate": 0.00025832000000000003,
      "loss": 5.4352,
      "step": 3230
    },
    {
      "epoch": 0.0663485757581963,
      "grad_norm": 0.48641809821128845,
      "learning_rate": 0.00025912000000000005,
      "loss": 5.4437,
      "step": 3240
    },
    {
      "epoch": 0.06655335531300555,
      "grad_norm": 0.4803517460823059,
      "learning_rate": 0.00025992,
      "loss": 5.4641,
      "step": 3250
    },
    {
      "epoch": 0.0667581348678148,
      "grad_norm": 0.5579696297645569,
      "learning_rate": 0.00026072000000000004,
      "loss": 5.4292,
      "step": 3260
    },
    {
      "epoch": 0.06696291442262405,
      "grad_norm": 0.4361009895801544,
      "learning_rate": 0.00026152000000000006,
      "loss": 5.4202,
      "step": 3270
    },
    {
      "epoch": 0.06716769397743329,
      "grad_norm": 0.45630156993865967,
      "learning_rate": 0.00026232,
      "loss": 5.4358,
      "step": 3280
    },
    {
      "epoch": 0.06737247353224254,
      "grad_norm": 0.5145776271820068,
      "learning_rate": 0.00026312000000000004,
      "loss": 5.468,
      "step": 3290
    },
    {
      "epoch": 0.06757725308705179,
      "grad_norm": 0.457546204328537,
      "learning_rate": 0.00026392,
      "loss": 5.4354,
      "step": 3300
    },
    {
      "epoch": 0.06778203264186104,
      "grad_norm": 0.4389740526676178,
      "learning_rate": 0.00026472,
      "loss": 5.4383,
      "step": 3310
    },
    {
      "epoch": 0.06798681219667028,
      "grad_norm": 0.44789761304855347,
      "learning_rate": 0.00026552,
      "loss": 5.448,
      "step": 3320
    },
    {
      "epoch": 0.06819159175147953,
      "grad_norm": 0.45958659052848816,
      "learning_rate": 0.00026632,
      "loss": 5.4228,
      "step": 3330
    },
    {
      "epoch": 0.06839637130628878,
      "grad_norm": 0.6689921617507935,
      "learning_rate": 0.00026712,
      "loss": 5.4684,
      "step": 3340
    },
    {
      "epoch": 0.06860115086109803,
      "grad_norm": 0.49806907773017883,
      "learning_rate": 0.00026792,
      "loss": 5.4264,
      "step": 3350
    },
    {
      "epoch": 0.06880593041590728,
      "grad_norm": 0.524211049079895,
      "learning_rate": 0.00026872,
      "loss": 5.4247,
      "step": 3360
    },
    {
      "epoch": 0.06901070997071652,
      "grad_norm": 0.5470885038375854,
      "learning_rate": 0.00026952,
      "loss": 5.436,
      "step": 3370
    },
    {
      "epoch": 0.06921548952552577,
      "grad_norm": 0.4457398056983948,
      "learning_rate": 0.00027032,
      "loss": 5.4005,
      "step": 3380
    },
    {
      "epoch": 0.06942026908033502,
      "grad_norm": 0.4752626121044159,
      "learning_rate": 0.00027112,
      "loss": 5.4081,
      "step": 3390
    },
    {
      "epoch": 0.06962504863514427,
      "grad_norm": 0.4663235545158386,
      "learning_rate": 0.00027192,
      "loss": 5.3891,
      "step": 3400
    },
    {
      "epoch": 0.06982982818995352,
      "grad_norm": 0.4726623296737671,
      "learning_rate": 0.00027272,
      "loss": 5.4364,
      "step": 3410
    },
    {
      "epoch": 0.07003460774476276,
      "grad_norm": 0.4584830105304718,
      "learning_rate": 0.00027352,
      "loss": 5.3909,
      "step": 3420
    },
    {
      "epoch": 0.070239387299572,
      "grad_norm": 0.5055278539657593,
      "learning_rate": 0.00027432,
      "loss": 5.3835,
      "step": 3430
    },
    {
      "epoch": 0.07044416685438126,
      "grad_norm": 0.45341160893440247,
      "learning_rate": 0.00027512,
      "loss": 5.4318,
      "step": 3440
    },
    {
      "epoch": 0.0706489464091905,
      "grad_norm": 0.412478506565094,
      "learning_rate": 0.00027592,
      "loss": 5.3912,
      "step": 3450
    },
    {
      "epoch": 0.07085372596399975,
      "grad_norm": 0.7927995920181274,
      "learning_rate": 0.00027672,
      "loss": 5.4522,
      "step": 3460
    },
    {
      "epoch": 0.071058505518809,
      "grad_norm": 0.4442445635795593,
      "learning_rate": 0.00027752,
      "loss": 5.3683,
      "step": 3470
    },
    {
      "epoch": 0.07126328507361825,
      "grad_norm": 0.4521198570728302,
      "learning_rate": 0.00027832,
      "loss": 5.4093,
      "step": 3480
    },
    {
      "epoch": 0.0714680646284275,
      "grad_norm": 0.5341065526008606,
      "learning_rate": 0.00027912,
      "loss": 5.3989,
      "step": 3490
    },
    {
      "epoch": 0.07167284418323674,
      "grad_norm": 0.4413537383079529,
      "learning_rate": 0.00027992,
      "loss": 5.4228,
      "step": 3500
    },
    {
      "epoch": 0.07187762373804599,
      "grad_norm": 0.46327200531959534,
      "learning_rate": 0.00028072,
      "loss": 5.3654,
      "step": 3510
    },
    {
      "epoch": 0.07208240329285524,
      "grad_norm": 0.3957655131816864,
      "learning_rate": 0.00028152,
      "loss": 5.4269,
      "step": 3520
    },
    {
      "epoch": 0.07228718284766449,
      "grad_norm": 0.42554226517677307,
      "learning_rate": 0.00028232,
      "loss": 5.345,
      "step": 3530
    },
    {
      "epoch": 0.07249196240247374,
      "grad_norm": 0.506805419921875,
      "learning_rate": 0.00028312,
      "loss": 5.3619,
      "step": 3540
    },
    {
      "epoch": 0.07269674195728298,
      "grad_norm": 0.4607420265674591,
      "learning_rate": 0.00028392,
      "loss": 5.35,
      "step": 3550
    },
    {
      "epoch": 0.07290152151209223,
      "grad_norm": 0.47505366802215576,
      "learning_rate": 0.00028472,
      "loss": 5.399,
      "step": 3560
    },
    {
      "epoch": 0.07310630106690148,
      "grad_norm": 0.45166629552841187,
      "learning_rate": 0.00028552,
      "loss": 5.4003,
      "step": 3570
    },
    {
      "epoch": 0.07331108062171073,
      "grad_norm": 0.41713911294937134,
      "learning_rate": 0.00028632,
      "loss": 5.3659,
      "step": 3580
    },
    {
      "epoch": 0.07351586017651998,
      "grad_norm": 0.5334256291389465,
      "learning_rate": 0.00028712000000000003,
      "loss": 5.4177,
      "step": 3590
    },
    {
      "epoch": 0.07372063973132922,
      "grad_norm": 0.4191318154335022,
      "learning_rate": 0.00028792,
      "loss": 5.3886,
      "step": 3600
    },
    {
      "epoch": 0.07392541928613847,
      "grad_norm": 0.46192094683647156,
      "learning_rate": 0.00028872,
      "loss": 5.3861,
      "step": 3610
    },
    {
      "epoch": 0.07413019884094772,
      "grad_norm": 0.4636607766151428,
      "learning_rate": 0.00028952000000000003,
      "loss": 5.3859,
      "step": 3620
    },
    {
      "epoch": 0.07433497839575696,
      "grad_norm": 0.4274023175239563,
      "learning_rate": 0.00029032,
      "loss": 5.3749,
      "step": 3630
    },
    {
      "epoch": 0.07453975795056622,
      "grad_norm": 0.4159563481807709,
      "learning_rate": 0.00029112,
      "loss": 5.3433,
      "step": 3640
    },
    {
      "epoch": 0.07474453750537546,
      "grad_norm": 0.4604257047176361,
      "learning_rate": 0.00029192000000000004,
      "loss": 5.3618,
      "step": 3650
    },
    {
      "epoch": 0.0749493170601847,
      "grad_norm": 0.44258370995521545,
      "learning_rate": 0.00029272,
      "loss": 5.3936,
      "step": 3660
    },
    {
      "epoch": 0.07515409661499396,
      "grad_norm": 0.5245718359947205,
      "learning_rate": 0.00029352,
      "loss": 5.3848,
      "step": 3670
    },
    {
      "epoch": 0.0753588761698032,
      "grad_norm": 0.5492002964019775,
      "learning_rate": 0.00029432000000000004,
      "loss": 5.3355,
      "step": 3680
    },
    {
      "epoch": 0.07556365572461246,
      "grad_norm": 0.41918817162513733,
      "learning_rate": 0.00029512,
      "loss": 5.3354,
      "step": 3690
    },
    {
      "epoch": 0.0757684352794217,
      "grad_norm": 0.4255083501338959,
      "learning_rate": 0.00029592,
      "loss": 5.3575,
      "step": 3700
    },
    {
      "epoch": 0.07597321483423095,
      "grad_norm": 0.4720117449760437,
      "learning_rate": 0.00029672000000000005,
      "loss": 5.3405,
      "step": 3710
    },
    {
      "epoch": 0.0761779943890402,
      "grad_norm": 0.47471868991851807,
      "learning_rate": 0.00029752,
      "loss": 5.3518,
      "step": 3720
    },
    {
      "epoch": 0.07638277394384944,
      "grad_norm": 0.4269292652606964,
      "learning_rate": 0.00029832000000000003,
      "loss": 5.3825,
      "step": 3730
    },
    {
      "epoch": 0.0765875534986587,
      "grad_norm": 2.8523495197296143,
      "learning_rate": 0.00029912,
      "loss": 5.3626,
      "step": 3740
    },
    {
      "epoch": 0.07679233305346794,
      "grad_norm": 0.4189223647117615,
      "learning_rate": 0.00029992,
      "loss": 5.3676,
      "step": 3750
    },
    {
      "epoch": 0.07699711260827719,
      "grad_norm": 1.7700881958007812,
      "learning_rate": 0.00030072000000000003,
      "loss": 5.3279,
      "step": 3760
    },
    {
      "epoch": 0.07720189216308644,
      "grad_norm": 0.39521193504333496,
      "learning_rate": 0.00030152,
      "loss": 5.3587,
      "step": 3770
    },
    {
      "epoch": 0.07740667171789568,
      "grad_norm": 0.4292346239089966,
      "learning_rate": 0.00030232,
      "loss": 5.3058,
      "step": 3780
    },
    {
      "epoch": 0.07761145127270494,
      "grad_norm": 0.42175960540771484,
      "learning_rate": 0.00030312000000000004,
      "loss": 5.3168,
      "step": 3790
    },
    {
      "epoch": 0.07781623082751418,
      "grad_norm": 0.3920210599899292,
      "learning_rate": 0.00030392,
      "loss": 5.3741,
      "step": 3800
    },
    {
      "epoch": 0.07802101038232342,
      "grad_norm": 0.43849965929985046,
      "learning_rate": 0.00030472,
      "loss": 5.3406,
      "step": 3810
    },
    {
      "epoch": 0.07822578993713268,
      "grad_norm": 0.42803141474723816,
      "learning_rate": 0.00030552000000000004,
      "loss": 5.3347,
      "step": 3820
    },
    {
      "epoch": 0.07843056949194192,
      "grad_norm": 0.42235347628593445,
      "learning_rate": 0.00030632,
      "loss": 5.3508,
      "step": 3830
    },
    {
      "epoch": 0.07863534904675117,
      "grad_norm": 0.4283842146396637,
      "learning_rate": 0.00030712000000000003,
      "loss": 5.3453,
      "step": 3840
    },
    {
      "epoch": 0.07884012860156042,
      "grad_norm": 0.44694867730140686,
      "learning_rate": 0.00030792000000000005,
      "loss": 5.3724,
      "step": 3850
    },
    {
      "epoch": 0.07904490815636966,
      "grad_norm": 0.5036173462867737,
      "learning_rate": 0.00030872,
      "loss": 5.3517,
      "step": 3860
    },
    {
      "epoch": 0.07924968771117892,
      "grad_norm": 0.5203410387039185,
      "learning_rate": 0.00030952000000000003,
      "loss": 5.3277,
      "step": 3870
    },
    {
      "epoch": 0.07945446726598816,
      "grad_norm": 0.4823491871356964,
      "learning_rate": 0.00031032000000000005,
      "loss": 5.3418,
      "step": 3880
    },
    {
      "epoch": 0.0796592468207974,
      "grad_norm": 0.39805030822753906,
      "learning_rate": 0.00031112,
      "loss": 5.3395,
      "step": 3890
    },
    {
      "epoch": 0.07986402637560666,
      "grad_norm": 0.3759382367134094,
      "learning_rate": 0.00031192000000000004,
      "loss": 5.2811,
      "step": 3900
    },
    {
      "epoch": 0.0800688059304159,
      "grad_norm": 0.3856908977031708,
      "learning_rate": 0.00031272000000000006,
      "loss": 5.3364,
      "step": 3910
    },
    {
      "epoch": 0.08027358548522516,
      "grad_norm": 0.5101864337921143,
      "learning_rate": 0.00031352,
      "loss": 5.3166,
      "step": 3920
    },
    {
      "epoch": 0.0804783650400344,
      "grad_norm": 0.4429487884044647,
      "learning_rate": 0.00031432000000000004,
      "loss": 5.3429,
      "step": 3930
    },
    {
      "epoch": 0.08068314459484364,
      "grad_norm": 0.36321690678596497,
      "learning_rate": 0.00031512,
      "loss": 5.3095,
      "step": 3940
    },
    {
      "epoch": 0.0808879241496529,
      "grad_norm": 0.4004923701286316,
      "learning_rate": 0.00031591999999999997,
      "loss": 5.3063,
      "step": 3950
    },
    {
      "epoch": 0.08109270370446214,
      "grad_norm": 0.4149259030818939,
      "learning_rate": 0.00031672,
      "loss": 5.2982,
      "step": 3960
    },
    {
      "epoch": 0.0812974832592714,
      "grad_norm": 0.635617733001709,
      "learning_rate": 0.00031752,
      "loss": 5.3013,
      "step": 3970
    },
    {
      "epoch": 0.08150226281408064,
      "grad_norm": 0.3996580243110657,
      "learning_rate": 0.00031832,
      "loss": 5.3785,
      "step": 3980
    },
    {
      "epoch": 0.08170704236888988,
      "grad_norm": 0.37409737706184387,
      "learning_rate": 0.00031912,
      "loss": 5.3497,
      "step": 3990
    },
    {
      "epoch": 0.08191182192369914,
      "grad_norm": 0.3825286030769348,
      "learning_rate": 0.00031992,
      "loss": 5.3066,
      "step": 4000
    },
    {
      "epoch": 0.08191182192369914,
      "eval_loss": 5.318787097930908,
      "eval_runtime": 4.3875,
      "eval_samples_per_second": 265.753,
      "eval_steps_per_second": 33.276,
      "step": 4000
    },
    {
      "epoch": 0.08211660147850838,
      "grad_norm": 0.3755367398262024,
      "learning_rate": 0.00032072,
      "loss": 5.2894,
      "step": 4010
    },
    {
      "epoch": 0.08232138103331764,
      "grad_norm": 0.40809503197669983,
      "learning_rate": 0.00032152,
      "loss": 5.2955,
      "step": 4020
    },
    {
      "epoch": 0.08252616058812688,
      "grad_norm": 0.41797196865081787,
      "learning_rate": 0.00032232,
      "loss": 5.3023,
      "step": 4030
    },
    {
      "epoch": 0.08273094014293612,
      "grad_norm": 0.3694484531879425,
      "learning_rate": 0.00032312,
      "loss": 5.3087,
      "step": 4040
    },
    {
      "epoch": 0.08293571969774538,
      "grad_norm": 0.38349223136901855,
      "learning_rate": 0.00032392,
      "loss": 5.334,
      "step": 4050
    },
    {
      "epoch": 0.08314049925255462,
      "grad_norm": 0.44905778765678406,
      "learning_rate": 0.00032472,
      "loss": 5.3084,
      "step": 4060
    },
    {
      "epoch": 0.08334527880736388,
      "grad_norm": 0.37099722027778625,
      "learning_rate": 0.00032552,
      "loss": 5.2817,
      "step": 4070
    },
    {
      "epoch": 0.08355005836217312,
      "grad_norm": 0.41185104846954346,
      "learning_rate": 0.00032632,
      "loss": 5.2967,
      "step": 4080
    },
    {
      "epoch": 0.08375483791698236,
      "grad_norm": 0.3768998980522156,
      "learning_rate": 0.00032712,
      "loss": 5.3252,
      "step": 4090
    },
    {
      "epoch": 0.08395961747179162,
      "grad_norm": 0.49293774366378784,
      "learning_rate": 0.00032792,
      "loss": 5.2841,
      "step": 4100
    },
    {
      "epoch": 0.08416439702660086,
      "grad_norm": 0.3644670248031616,
      "learning_rate": 0.00032872,
      "loss": 5.2919,
      "step": 4110
    },
    {
      "epoch": 0.08436917658141012,
      "grad_norm": 0.4202214181423187,
      "learning_rate": 0.00032952000000000003,
      "loss": 5.3057,
      "step": 4120
    },
    {
      "epoch": 0.08457395613621936,
      "grad_norm": 0.39260733127593994,
      "learning_rate": 0.00033032,
      "loss": 5.3047,
      "step": 4130
    },
    {
      "epoch": 0.0847787356910286,
      "grad_norm": 0.3847532570362091,
      "learning_rate": 0.00033112,
      "loss": 5.2819,
      "step": 4140
    },
    {
      "epoch": 0.08498351524583786,
      "grad_norm": 0.41340094804763794,
      "learning_rate": 0.00033192000000000003,
      "loss": 5.2916,
      "step": 4150
    },
    {
      "epoch": 0.0851882948006471,
      "grad_norm": 0.3760600984096527,
      "learning_rate": 0.00033272,
      "loss": 5.2889,
      "step": 4160
    },
    {
      "epoch": 0.08539307435545636,
      "grad_norm": 0.3887149393558502,
      "learning_rate": 0.00033352,
      "loss": 5.2597,
      "step": 4170
    },
    {
      "epoch": 0.0855978539102656,
      "grad_norm": 0.4354073107242584,
      "learning_rate": 0.00033432,
      "loss": 5.2719,
      "step": 4180
    },
    {
      "epoch": 0.08580263346507484,
      "grad_norm": 0.4749308228492737,
      "learning_rate": 0.00033512,
      "loss": 5.2916,
      "step": 4190
    },
    {
      "epoch": 0.0860074130198841,
      "grad_norm": 0.41358011960983276,
      "learning_rate": 0.00033592,
      "loss": 5.2996,
      "step": 4200
    },
    {
      "epoch": 0.08621219257469334,
      "grad_norm": 0.3583889901638031,
      "learning_rate": 0.00033672,
      "loss": 5.2626,
      "step": 4210
    },
    {
      "epoch": 0.08641697212950258,
      "grad_norm": 0.4384559094905853,
      "learning_rate": 0.00033752,
      "loss": 5.2314,
      "step": 4220
    },
    {
      "epoch": 0.08662175168431184,
      "grad_norm": 0.37336528301239014,
      "learning_rate": 0.00033832,
      "loss": 5.2923,
      "step": 4230
    },
    {
      "epoch": 0.08682653123912108,
      "grad_norm": 0.39408349990844727,
      "learning_rate": 0.00033912,
      "loss": 5.259,
      "step": 4240
    },
    {
      "epoch": 0.08703131079393034,
      "grad_norm": 0.32332339882850647,
      "learning_rate": 0.00033992,
      "loss": 5.2699,
      "step": 4250
    },
    {
      "epoch": 0.08723609034873958,
      "grad_norm": 0.46973833441734314,
      "learning_rate": 0.00034072000000000003,
      "loss": 5.3307,
      "step": 4260
    },
    {
      "epoch": 0.08744086990354882,
      "grad_norm": 0.3726697564125061,
      "learning_rate": 0.00034152,
      "loss": 5.2893,
      "step": 4270
    },
    {
      "epoch": 0.08764564945835808,
      "grad_norm": 0.3420617878437042,
      "learning_rate": 0.00034232,
      "loss": 5.3053,
      "step": 4280
    },
    {
      "epoch": 0.08785042901316732,
      "grad_norm": 0.3470669984817505,
      "learning_rate": 0.00034312000000000004,
      "loss": 5.2869,
      "step": 4290
    },
    {
      "epoch": 0.08805520856797658,
      "grad_norm": 0.3572559952735901,
      "learning_rate": 0.00034392,
      "loss": 5.2383,
      "step": 4300
    },
    {
      "epoch": 0.08825998812278582,
      "grad_norm": 0.3921760320663452,
      "learning_rate": 0.00034472,
      "loss": 5.2638,
      "step": 4310
    },
    {
      "epoch": 0.08846476767759506,
      "grad_norm": 0.37090209126472473,
      "learning_rate": 0.00034552000000000004,
      "loss": 5.2768,
      "step": 4320
    },
    {
      "epoch": 0.08866954723240432,
      "grad_norm": 0.34295281767845154,
      "learning_rate": 0.00034632,
      "loss": 5.2567,
      "step": 4330
    },
    {
      "epoch": 0.08887432678721356,
      "grad_norm": 0.379252552986145,
      "learning_rate": 0.00034712,
      "loss": 5.2565,
      "step": 4340
    },
    {
      "epoch": 0.08907910634202282,
      "grad_norm": 0.35600078105926514,
      "learning_rate": 0.00034792000000000004,
      "loss": 5.2404,
      "step": 4350
    },
    {
      "epoch": 0.08928388589683206,
      "grad_norm": 0.37798258662223816,
      "learning_rate": 0.00034872,
      "loss": 5.2591,
      "step": 4360
    },
    {
      "epoch": 0.0894886654516413,
      "grad_norm": 0.4064588248729706,
      "learning_rate": 0.00034952000000000003,
      "loss": 5.2677,
      "step": 4370
    },
    {
      "epoch": 0.08969344500645056,
      "grad_norm": 0.4060820937156677,
      "learning_rate": 0.00035032000000000005,
      "loss": 5.2776,
      "step": 4380
    },
    {
      "epoch": 0.0898982245612598,
      "grad_norm": 0.41651928424835205,
      "learning_rate": 0.00035112,
      "loss": 5.2984,
      "step": 4390
    },
    {
      "epoch": 0.09010300411606906,
      "grad_norm": 0.3946080505847931,
      "learning_rate": 0.00035192000000000003,
      "loss": 5.2596,
      "step": 4400
    },
    {
      "epoch": 0.0903077836708783,
      "grad_norm": 0.3811205327510834,
      "learning_rate": 0.00035272000000000005,
      "loss": 5.2489,
      "step": 4410
    },
    {
      "epoch": 0.09051256322568754,
      "grad_norm": 0.36853381991386414,
      "learning_rate": 0.00035352,
      "loss": 5.2474,
      "step": 4420
    },
    {
      "epoch": 0.0907173427804968,
      "grad_norm": 0.41594240069389343,
      "learning_rate": 0.00035432000000000004,
      "loss": 5.2584,
      "step": 4430
    },
    {
      "epoch": 0.09092212233530604,
      "grad_norm": 0.35103175044059753,
      "learning_rate": 0.00035512000000000006,
      "loss": 5.2674,
      "step": 4440
    },
    {
      "epoch": 0.0911269018901153,
      "grad_norm": 0.3328525424003601,
      "learning_rate": 0.00035592,
      "loss": 5.2757,
      "step": 4450
    },
    {
      "epoch": 0.09133168144492454,
      "grad_norm": 0.36880525946617126,
      "learning_rate": 0.00035672000000000004,
      "loss": 5.2261,
      "step": 4460
    },
    {
      "epoch": 0.09153646099973378,
      "grad_norm": 0.35445812344551086,
      "learning_rate": 0.00035752,
      "loss": 5.2195,
      "step": 4470
    },
    {
      "epoch": 0.09174124055454304,
      "grad_norm": 0.36946800351142883,
      "learning_rate": 0.00035832,
      "loss": 5.2225,
      "step": 4480
    },
    {
      "epoch": 0.09194602010935228,
      "grad_norm": 0.375847727060318,
      "learning_rate": 0.00035912000000000004,
      "loss": 5.2446,
      "step": 4490
    },
    {
      "epoch": 0.09215079966416154,
      "grad_norm": 0.3652852773666382,
      "learning_rate": 0.00035992,
      "loss": 5.2209,
      "step": 4500
    },
    {
      "epoch": 0.09235557921897078,
      "grad_norm": 0.36818650364875793,
      "learning_rate": 0.00036072000000000003,
      "loss": 5.2382,
      "step": 4510
    },
    {
      "epoch": 0.09256035877378002,
      "grad_norm": 0.3429717421531677,
      "learning_rate": 0.00036152000000000005,
      "loss": 5.2163,
      "step": 4520
    },
    {
      "epoch": 0.09276513832858928,
      "grad_norm": 0.3466830551624298,
      "learning_rate": 0.00036232,
      "loss": 5.2521,
      "step": 4530
    },
    {
      "epoch": 0.09296991788339852,
      "grad_norm": 0.40972962975502014,
      "learning_rate": 0.00036312000000000003,
      "loss": 5.2236,
      "step": 4540
    },
    {
      "epoch": 0.09317469743820776,
      "grad_norm": 0.39865970611572266,
      "learning_rate": 0.00036392000000000005,
      "loss": 5.2733,
      "step": 4550
    },
    {
      "epoch": 0.09337947699301702,
      "grad_norm": 0.3470962643623352,
      "learning_rate": 0.00036472,
      "loss": 5.2474,
      "step": 4560
    },
    {
      "epoch": 0.09358425654782626,
      "grad_norm": 0.36036545038223267,
      "learning_rate": 0.00036552,
      "loss": 5.2573,
      "step": 4570
    },
    {
      "epoch": 0.09378903610263552,
      "grad_norm": 0.41582751274108887,
      "learning_rate": 0.00036632,
      "loss": 5.2342,
      "step": 4580
    },
    {
      "epoch": 0.09399381565744476,
      "grad_norm": 0.3146849274635315,
      "learning_rate": 0.00036712,
      "loss": 5.2397,
      "step": 4590
    },
    {
      "epoch": 0.094198595212254,
      "grad_norm": 0.3361073434352875,
      "learning_rate": 0.00036792,
      "loss": 5.2354,
      "step": 4600
    },
    {
      "epoch": 0.09440337476706326,
      "grad_norm": 0.36320430040359497,
      "learning_rate": 0.00036872,
      "loss": 5.2333,
      "step": 4610
    },
    {
      "epoch": 0.0946081543218725,
      "grad_norm": 0.3914796710014343,
      "learning_rate": 0.00036952,
      "loss": 5.2271,
      "step": 4620
    },
    {
      "epoch": 0.09481293387668176,
      "grad_norm": 0.38062742352485657,
      "learning_rate": 0.00037032,
      "loss": 5.2512,
      "step": 4630
    },
    {
      "epoch": 0.095017713431491,
      "grad_norm": 0.35448241233825684,
      "learning_rate": 0.00037112,
      "loss": 5.2494,
      "step": 4640
    },
    {
      "epoch": 0.09522249298630024,
      "grad_norm": 0.3220553994178772,
      "learning_rate": 0.00037192,
      "loss": 5.2291,
      "step": 4650
    },
    {
      "epoch": 0.0954272725411095,
      "grad_norm": 0.5660349726676941,
      "learning_rate": 0.00037272,
      "loss": 5.2063,
      "step": 4660
    },
    {
      "epoch": 0.09563205209591874,
      "grad_norm": 0.3554455041885376,
      "learning_rate": 0.00037352,
      "loss": 5.21,
      "step": 4670
    },
    {
      "epoch": 0.095836831650728,
      "grad_norm": 0.33699843287467957,
      "learning_rate": 0.00037432,
      "loss": 5.2057,
      "step": 4680
    },
    {
      "epoch": 0.09604161120553724,
      "grad_norm": 0.4169362485408783,
      "learning_rate": 0.00037512,
      "loss": 5.1735,
      "step": 4690
    },
    {
      "epoch": 0.09624639076034648,
      "grad_norm": 0.34874531626701355,
      "learning_rate": 0.00037592,
      "loss": 5.2358,
      "step": 4700
    },
    {
      "epoch": 0.09645117031515574,
      "grad_norm": 0.3854043483734131,
      "learning_rate": 0.00037672,
      "loss": 5.2211,
      "step": 4710
    },
    {
      "epoch": 0.09665594986996498,
      "grad_norm": 0.32648953795433044,
      "learning_rate": 0.00037752,
      "loss": 5.1953,
      "step": 4720
    },
    {
      "epoch": 0.09686072942477424,
      "grad_norm": 0.3376765847206116,
      "learning_rate": 0.00037832,
      "loss": 5.2278,
      "step": 4730
    },
    {
      "epoch": 0.09706550897958348,
      "grad_norm": 0.33131176233291626,
      "learning_rate": 0.00037912,
      "loss": 5.2227,
      "step": 4740
    },
    {
      "epoch": 0.09727028853439272,
      "grad_norm": 0.33954697847366333,
      "learning_rate": 0.00037992,
      "loss": 5.2328,
      "step": 4750
    },
    {
      "epoch": 0.09747506808920198,
      "grad_norm": 0.3093273937702179,
      "learning_rate": 0.00038072000000000003,
      "loss": 5.2087,
      "step": 4760
    },
    {
      "epoch": 0.09767984764401122,
      "grad_norm": 0.3387199640274048,
      "learning_rate": 0.00038152,
      "loss": 5.2066,
      "step": 4770
    },
    {
      "epoch": 0.09788462719882048,
      "grad_norm": 0.3928283452987671,
      "learning_rate": 0.00038232,
      "loss": 5.2122,
      "step": 4780
    },
    {
      "epoch": 0.09808940675362972,
      "grad_norm": 0.31002166867256165,
      "learning_rate": 0.00038312000000000003,
      "loss": 5.197,
      "step": 4790
    },
    {
      "epoch": 0.09829418630843896,
      "grad_norm": 0.3293517827987671,
      "learning_rate": 0.00038392,
      "loss": 5.1674,
      "step": 4800
    },
    {
      "epoch": 0.09849896586324822,
      "grad_norm": 0.3640994131565094,
      "learning_rate": 0.00038472,
      "loss": 5.2214,
      "step": 4810
    },
    {
      "epoch": 0.09870374541805746,
      "grad_norm": 0.33961260318756104,
      "learning_rate": 0.00038552000000000004,
      "loss": 5.2219,
      "step": 4820
    },
    {
      "epoch": 0.09890852497286672,
      "grad_norm": 0.4387868046760559,
      "learning_rate": 0.00038632,
      "loss": 5.2487,
      "step": 4830
    },
    {
      "epoch": 0.09911330452767596,
      "grad_norm": 0.3797737956047058,
      "learning_rate": 0.00038712,
      "loss": 5.2065,
      "step": 4840
    },
    {
      "epoch": 0.0993180840824852,
      "grad_norm": 0.39219486713409424,
      "learning_rate": 0.00038792000000000004,
      "loss": 5.2036,
      "step": 4850
    },
    {
      "epoch": 0.09952286363729446,
      "grad_norm": 0.3058601915836334,
      "learning_rate": 0.00038872,
      "loss": 5.2291,
      "step": 4860
    },
    {
      "epoch": 0.0997276431921037,
      "grad_norm": 0.35596054792404175,
      "learning_rate": 0.00038952,
      "loss": 5.1957,
      "step": 4870
    },
    {
      "epoch": 0.09993242274691296,
      "grad_norm": 0.3059404790401459,
      "learning_rate": 0.00039032000000000004,
      "loss": 5.1907,
      "step": 4880
    },
    {
      "epoch": 0.1001372023017222,
      "grad_norm": 0.3631693422794342,
      "learning_rate": 0.00039112,
      "loss": 5.1968,
      "step": 4890
    },
    {
      "epoch": 0.10034198185653144,
      "grad_norm": 0.34312257170677185,
      "learning_rate": 0.00039192000000000003,
      "loss": 5.2482,
      "step": 4900
    },
    {
      "epoch": 0.1005467614113407,
      "grad_norm": 0.3389482796192169,
      "learning_rate": 0.00039272000000000005,
      "loss": 5.1609,
      "step": 4910
    },
    {
      "epoch": 0.10075154096614994,
      "grad_norm": 0.3607906997203827,
      "learning_rate": 0.00039352,
      "loss": 5.2572,
      "step": 4920
    },
    {
      "epoch": 0.10095632052095918,
      "grad_norm": 0.36973777413368225,
      "learning_rate": 0.00039432000000000003,
      "loss": 5.2087,
      "step": 4930
    },
    {
      "epoch": 0.10116110007576844,
      "grad_norm": 0.35736119747161865,
      "learning_rate": 0.00039512,
      "loss": 5.2319,
      "step": 4940
    },
    {
      "epoch": 0.10136587963057768,
      "grad_norm": 0.33283448219299316,
      "learning_rate": 0.00039592,
      "loss": 5.1959,
      "step": 4950
    },
    {
      "epoch": 0.10157065918538694,
      "grad_norm": 0.329438716173172,
      "learning_rate": 0.00039672000000000004,
      "loss": 5.2262,
      "step": 4960
    },
    {
      "epoch": 0.10177543874019618,
      "grad_norm": 0.37530723214149475,
      "learning_rate": 0.00039752,
      "loss": 5.2094,
      "step": 4970
    },
    {
      "epoch": 0.10198021829500542,
      "grad_norm": 0.3056955337524414,
      "learning_rate": 0.00039832,
      "loss": 5.2204,
      "step": 4980
    },
    {
      "epoch": 0.10218499784981468,
      "grad_norm": 0.3704677224159241,
      "learning_rate": 0.00039912000000000004,
      "loss": 5.1631,
      "step": 4990
    },
    {
      "epoch": 0.10238977740462392,
      "grad_norm": 0.31214451789855957,
      "learning_rate": 0.00039992,
      "loss": 5.2125,
      "step": 5000
    },
    {
      "epoch": 0.10259455695943318,
      "grad_norm": 0.313644140958786,
      "learning_rate": 0.0004,
      "loss": 5.1836,
      "step": 5010
    },
    {
      "epoch": 0.10279933651424242,
      "grad_norm": 0.33864766359329224,
      "learning_rate": 0.0004,
      "loss": 5.1933,
      "step": 5020
    },
    {
      "epoch": 0.10300411606905166,
      "grad_norm": 0.3204938769340515,
      "learning_rate": 0.0004,
      "loss": 5.2015,
      "step": 5030
    },
    {
      "epoch": 0.10320889562386092,
      "grad_norm": 0.30994290113449097,
      "learning_rate": 0.0004,
      "loss": 5.184,
      "step": 5040
    },
    {
      "epoch": 0.10341367517867016,
      "grad_norm": 0.33178046345710754,
      "learning_rate": 0.0004,
      "loss": 5.1727,
      "step": 5050
    },
    {
      "epoch": 0.10361845473347941,
      "grad_norm": 0.32041501998901367,
      "learning_rate": 0.0004,
      "loss": 5.1822,
      "step": 5060
    },
    {
      "epoch": 0.10382323428828866,
      "grad_norm": 0.3314732313156128,
      "learning_rate": 0.0004,
      "loss": 5.1814,
      "step": 5070
    },
    {
      "epoch": 0.1040280138430979,
      "grad_norm": 0.39618968963623047,
      "learning_rate": 0.0004,
      "loss": 5.2003,
      "step": 5080
    },
    {
      "epoch": 0.10423279339790716,
      "grad_norm": 0.3051014542579651,
      "learning_rate": 0.0004,
      "loss": 5.1695,
      "step": 5090
    },
    {
      "epoch": 0.1044375729527164,
      "grad_norm": 0.34999150037765503,
      "learning_rate": 0.0004,
      "loss": 5.1811,
      "step": 5100
    },
    {
      "epoch": 0.10464235250752565,
      "grad_norm": 0.309879869222641,
      "learning_rate": 0.0004,
      "loss": 5.1595,
      "step": 5110
    },
    {
      "epoch": 0.1048471320623349,
      "grad_norm": 0.3067319989204407,
      "learning_rate": 0.0004,
      "loss": 5.2001,
      "step": 5120
    },
    {
      "epoch": 0.10505191161714414,
      "grad_norm": 0.32993996143341064,
      "learning_rate": 0.0004,
      "loss": 5.1486,
      "step": 5130
    },
    {
      "epoch": 0.1052566911719534,
      "grad_norm": 0.34084463119506836,
      "learning_rate": 0.0004,
      "loss": 5.1787,
      "step": 5140
    },
    {
      "epoch": 0.10546147072676264,
      "grad_norm": 0.3445000946521759,
      "learning_rate": 0.0004,
      "loss": 5.1714,
      "step": 5150
    },
    {
      "epoch": 0.1056662502815719,
      "grad_norm": 0.3222779631614685,
      "learning_rate": 0.0004,
      "loss": 5.2008,
      "step": 5160
    },
    {
      "epoch": 0.10587102983638114,
      "grad_norm": 0.3041796088218689,
      "learning_rate": 0.0004,
      "loss": 5.1863,
      "step": 5170
    },
    {
      "epoch": 0.10607580939119038,
      "grad_norm": 0.3209103047847748,
      "learning_rate": 0.0004,
      "loss": 5.1932,
      "step": 5180
    },
    {
      "epoch": 0.10628058894599964,
      "grad_norm": 0.3323422372341156,
      "learning_rate": 0.0004,
      "loss": 5.1952,
      "step": 5190
    },
    {
      "epoch": 0.10648536850080888,
      "grad_norm": 0.325749933719635,
      "learning_rate": 0.0004,
      "loss": 5.1658,
      "step": 5200
    },
    {
      "epoch": 0.10669014805561813,
      "grad_norm": 0.30505135655403137,
      "learning_rate": 0.0004,
      "loss": 5.1855,
      "step": 5210
    },
    {
      "epoch": 0.10689492761042738,
      "grad_norm": 0.306336909532547,
      "learning_rate": 0.0004,
      "loss": 5.1555,
      "step": 5220
    },
    {
      "epoch": 0.10709970716523662,
      "grad_norm": 0.3228214979171753,
      "learning_rate": 0.0004,
      "loss": 5.1525,
      "step": 5230
    },
    {
      "epoch": 0.10730448672004587,
      "grad_norm": 0.38324278593063354,
      "learning_rate": 0.0004,
      "loss": 5.166,
      "step": 5240
    },
    {
      "epoch": 0.10750926627485512,
      "grad_norm": 0.30399709939956665,
      "learning_rate": 0.0004,
      "loss": 5.1625,
      "step": 5250
    },
    {
      "epoch": 0.10771404582966437,
      "grad_norm": 0.2954542636871338,
      "learning_rate": 0.0004,
      "loss": 5.1716,
      "step": 5260
    },
    {
      "epoch": 0.10791882538447362,
      "grad_norm": 0.3116077780723572,
      "learning_rate": 0.0004,
      "loss": 5.1651,
      "step": 5270
    },
    {
      "epoch": 0.10812360493928286,
      "grad_norm": 0.30015772581100464,
      "learning_rate": 0.0004,
      "loss": 5.1664,
      "step": 5280
    },
    {
      "epoch": 0.10832838449409211,
      "grad_norm": 0.3169012665748596,
      "learning_rate": 0.0004,
      "loss": 5.1646,
      "step": 5290
    },
    {
      "epoch": 0.10853316404890136,
      "grad_norm": 0.3500293493270874,
      "learning_rate": 0.0004,
      "loss": 5.1535,
      "step": 5300
    },
    {
      "epoch": 0.1087379436037106,
      "grad_norm": 0.3506503701210022,
      "learning_rate": 0.0004,
      "loss": 5.1542,
      "step": 5310
    },
    {
      "epoch": 0.10894272315851986,
      "grad_norm": 0.32760336995124817,
      "learning_rate": 0.0004,
      "loss": 5.1302,
      "step": 5320
    },
    {
      "epoch": 0.1091475027133291,
      "grad_norm": 0.2881554663181305,
      "learning_rate": 0.0004,
      "loss": 5.1459,
      "step": 5330
    },
    {
      "epoch": 0.10935228226813835,
      "grad_norm": 0.30306053161621094,
      "learning_rate": 0.0004,
      "loss": 5.1472,
      "step": 5340
    },
    {
      "epoch": 0.1095570618229476,
      "grad_norm": 0.31296682357788086,
      "learning_rate": 0.0004,
      "loss": 5.1584,
      "step": 5350
    },
    {
      "epoch": 0.10976184137775684,
      "grad_norm": 0.29192468523979187,
      "learning_rate": 0.0004,
      "loss": 5.1401,
      "step": 5360
    },
    {
      "epoch": 0.1099666209325661,
      "grad_norm": 0.28675010800361633,
      "learning_rate": 0.0004,
      "loss": 5.1557,
      "step": 5370
    },
    {
      "epoch": 0.11017140048737534,
      "grad_norm": 0.27933311462402344,
      "learning_rate": 0.0004,
      "loss": 5.1469,
      "step": 5380
    },
    {
      "epoch": 0.1103761800421846,
      "grad_norm": 0.29722484946250916,
      "learning_rate": 0.0004,
      "loss": 5.1144,
      "step": 5390
    },
    {
      "epoch": 0.11058095959699384,
      "grad_norm": 0.2667056918144226,
      "learning_rate": 0.0004,
      "loss": 5.1626,
      "step": 5400
    },
    {
      "epoch": 0.11078573915180308,
      "grad_norm": 0.3032090961933136,
      "learning_rate": 0.0004,
      "loss": 5.1395,
      "step": 5410
    },
    {
      "epoch": 0.11099051870661233,
      "grad_norm": 0.309581995010376,
      "learning_rate": 0.0004,
      "loss": 5.1377,
      "step": 5420
    },
    {
      "epoch": 0.11119529826142158,
      "grad_norm": 0.32929155230522156,
      "learning_rate": 0.0004,
      "loss": 5.1539,
      "step": 5430
    },
    {
      "epoch": 0.11140007781623083,
      "grad_norm": 0.2987646758556366,
      "learning_rate": 0.0004,
      "loss": 5.1414,
      "step": 5440
    },
    {
      "epoch": 0.11160485737104008,
      "grad_norm": 0.30965328216552734,
      "learning_rate": 0.0004,
      "loss": 5.1647,
      "step": 5450
    },
    {
      "epoch": 0.11180963692584932,
      "grad_norm": 0.28466081619262695,
      "learning_rate": 0.0004,
      "loss": 5.1423,
      "step": 5460
    },
    {
      "epoch": 0.11201441648065857,
      "grad_norm": 0.30453965067863464,
      "learning_rate": 0.0004,
      "loss": 5.1451,
      "step": 5470
    },
    {
      "epoch": 0.11221919603546782,
      "grad_norm": 0.2841033339500427,
      "learning_rate": 0.0004,
      "loss": 5.1354,
      "step": 5480
    },
    {
      "epoch": 0.11242397559027707,
      "grad_norm": 0.2860296070575714,
      "learning_rate": 0.0004,
      "loss": 5.1409,
      "step": 5490
    },
    {
      "epoch": 0.11262875514508632,
      "grad_norm": 0.30634403228759766,
      "learning_rate": 0.0004,
      "loss": 5.1375,
      "step": 5500
    },
    {
      "epoch": 0.11283353469989556,
      "grad_norm": 0.32455432415008545,
      "learning_rate": 0.0004,
      "loss": 5.1184,
      "step": 5510
    },
    {
      "epoch": 0.11303831425470481,
      "grad_norm": 0.2667500078678131,
      "learning_rate": 0.0004,
      "loss": 5.1514,
      "step": 5520
    },
    {
      "epoch": 0.11324309380951406,
      "grad_norm": 0.3181164264678955,
      "learning_rate": 0.0004,
      "loss": 5.1479,
      "step": 5530
    },
    {
      "epoch": 0.11344787336432331,
      "grad_norm": 0.2994334399700165,
      "learning_rate": 0.0004,
      "loss": 5.1251,
      "step": 5540
    },
    {
      "epoch": 0.11365265291913255,
      "grad_norm": 0.3418193459510803,
      "learning_rate": 0.0004,
      "loss": 5.1177,
      "step": 5550
    },
    {
      "epoch": 0.1138574324739418,
      "grad_norm": 0.31572362780570984,
      "learning_rate": 0.0004,
      "loss": 5.1733,
      "step": 5560
    },
    {
      "epoch": 0.11406221202875105,
      "grad_norm": 0.2817554771900177,
      "learning_rate": 0.0004,
      "loss": 5.1609,
      "step": 5570
    },
    {
      "epoch": 0.1142669915835603,
      "grad_norm": 0.2846536338329315,
      "learning_rate": 0.0004,
      "loss": 5.1243,
      "step": 5580
    },
    {
      "epoch": 0.11447177113836955,
      "grad_norm": 0.28663066029548645,
      "learning_rate": 0.0004,
      "loss": 5.1329,
      "step": 5590
    },
    {
      "epoch": 0.1146765506931788,
      "grad_norm": 0.31789955496788025,
      "learning_rate": 0.0004,
      "loss": 5.1209,
      "step": 5600
    },
    {
      "epoch": 0.11488133024798804,
      "grad_norm": 0.29089292883872986,
      "learning_rate": 0.0004,
      "loss": 5.1083,
      "step": 5610
    },
    {
      "epoch": 0.11508610980279729,
      "grad_norm": 0.28004202246665955,
      "learning_rate": 0.0004,
      "loss": 5.1544,
      "step": 5620
    },
    {
      "epoch": 0.11529088935760654,
      "grad_norm": 0.32260385155677795,
      "learning_rate": 0.0004,
      "loss": 5.1685,
      "step": 5630
    },
    {
      "epoch": 0.11549566891241578,
      "grad_norm": 0.30765828490257263,
      "learning_rate": 0.0004,
      "loss": 5.1183,
      "step": 5640
    },
    {
      "epoch": 0.11570044846722503,
      "grad_norm": 0.2876403331756592,
      "learning_rate": 0.0004,
      "loss": 5.1507,
      "step": 5650
    },
    {
      "epoch": 0.11590522802203428,
      "grad_norm": 0.2831974923610687,
      "learning_rate": 0.0004,
      "loss": 5.1111,
      "step": 5660
    },
    {
      "epoch": 0.11611000757684353,
      "grad_norm": 0.3007749617099762,
      "learning_rate": 0.0004,
      "loss": 5.1042,
      "step": 5670
    },
    {
      "epoch": 0.11631478713165277,
      "grad_norm": 0.29860612750053406,
      "learning_rate": 0.0004,
      "loss": 5.1098,
      "step": 5680
    },
    {
      "epoch": 0.11651956668646202,
      "grad_norm": 0.2871202528476715,
      "learning_rate": 0.0004,
      "loss": 5.119,
      "step": 5690
    },
    {
      "epoch": 0.11672434624127127,
      "grad_norm": 0.3583846390247345,
      "learning_rate": 0.0004,
      "loss": 5.0853,
      "step": 5700
    },
    {
      "epoch": 0.11692912579608052,
      "grad_norm": 0.3196072578430176,
      "learning_rate": 0.0004,
      "loss": 5.1183,
      "step": 5710
    },
    {
      "epoch": 0.11713390535088977,
      "grad_norm": 0.3659673035144806,
      "learning_rate": 0.0004,
      "loss": 5.1341,
      "step": 5720
    },
    {
      "epoch": 0.11733868490569901,
      "grad_norm": 0.34671393036842346,
      "learning_rate": 0.0004,
      "loss": 5.1032,
      "step": 5730
    },
    {
      "epoch": 0.11754346446050826,
      "grad_norm": 0.6136424541473389,
      "learning_rate": 0.0004,
      "loss": 5.1141,
      "step": 5740
    },
    {
      "epoch": 0.11774824401531751,
      "grad_norm": 0.29831385612487793,
      "learning_rate": 0.0004,
      "loss": 5.1164,
      "step": 5750
    },
    {
      "epoch": 0.11795302357012676,
      "grad_norm": 0.2739589214324951,
      "learning_rate": 0.0004,
      "loss": 5.1285,
      "step": 5760
    },
    {
      "epoch": 0.11815780312493601,
      "grad_norm": 0.3025692105293274,
      "learning_rate": 0.0004,
      "loss": 5.0998,
      "step": 5770
    },
    {
      "epoch": 0.11836258267974525,
      "grad_norm": 0.30093103647232056,
      "learning_rate": 0.0004,
      "loss": 5.1416,
      "step": 5780
    },
    {
      "epoch": 0.1185673622345545,
      "grad_norm": 0.28240787982940674,
      "learning_rate": 0.0004,
      "loss": 5.0816,
      "step": 5790
    },
    {
      "epoch": 0.11877214178936375,
      "grad_norm": 0.3148246705532074,
      "learning_rate": 0.0004,
      "loss": 5.0887,
      "step": 5800
    },
    {
      "epoch": 0.118976921344173,
      "grad_norm": 0.2791157364845276,
      "learning_rate": 0.0004,
      "loss": 5.1228,
      "step": 5810
    },
    {
      "epoch": 0.11918170089898225,
      "grad_norm": 0.31769630312919617,
      "learning_rate": 0.0004,
      "loss": 5.1121,
      "step": 5820
    },
    {
      "epoch": 0.1193864804537915,
      "grad_norm": 3.882388114929199,
      "learning_rate": 0.0004,
      "loss": 5.1327,
      "step": 5830
    },
    {
      "epoch": 0.11959126000860074,
      "grad_norm": 0.27645838260650635,
      "learning_rate": 0.0004,
      "loss": 5.1025,
      "step": 5840
    },
    {
      "epoch": 0.11979603956340999,
      "grad_norm": 0.2904749810695648,
      "learning_rate": 0.0004,
      "loss": 5.1297,
      "step": 5850
    },
    {
      "epoch": 0.12000081911821923,
      "grad_norm": 0.32768234610557556,
      "learning_rate": 0.0004,
      "loss": 5.0841,
      "step": 5860
    },
    {
      "epoch": 0.12020559867302849,
      "grad_norm": 0.3125588595867157,
      "learning_rate": 0.0004,
      "loss": 5.1031,
      "step": 5870
    },
    {
      "epoch": 0.12041037822783773,
      "grad_norm": 0.29457178711891174,
      "learning_rate": 0.0004,
      "loss": 5.1234,
      "step": 5880
    },
    {
      "epoch": 0.12061515778264698,
      "grad_norm": 0.33318033814430237,
      "learning_rate": 0.0004,
      "loss": 5.0914,
      "step": 5890
    },
    {
      "epoch": 0.12081993733745623,
      "grad_norm": 0.32637280225753784,
      "learning_rate": 0.0004,
      "loss": 5.0767,
      "step": 5900
    },
    {
      "epoch": 0.12102471689226547,
      "grad_norm": 0.3051298260688782,
      "learning_rate": 0.0004,
      "loss": 5.0964,
      "step": 5910
    },
    {
      "epoch": 0.12122949644707473,
      "grad_norm": 0.31057214736938477,
      "learning_rate": 0.0004,
      "loss": 5.0921,
      "step": 5920
    },
    {
      "epoch": 0.12143427600188397,
      "grad_norm": 0.31014057993888855,
      "learning_rate": 0.0004,
      "loss": 5.0711,
      "step": 5930
    },
    {
      "epoch": 0.12163905555669322,
      "grad_norm": 0.28628087043762207,
      "learning_rate": 0.0004,
      "loss": 5.0847,
      "step": 5940
    },
    {
      "epoch": 0.12184383511150247,
      "grad_norm": 0.280165433883667,
      "learning_rate": 0.0004,
      "loss": 5.1159,
      "step": 5950
    },
    {
      "epoch": 0.12204861466631171,
      "grad_norm": 0.32177525758743286,
      "learning_rate": 0.0004,
      "loss": 5.0868,
      "step": 5960
    },
    {
      "epoch": 0.12225339422112097,
      "grad_norm": 0.2712959349155426,
      "learning_rate": 0.0004,
      "loss": 5.106,
      "step": 5970
    },
    {
      "epoch": 0.12245817377593021,
      "grad_norm": 0.3010253310203552,
      "learning_rate": 0.0004,
      "loss": 5.1076,
      "step": 5980
    },
    {
      "epoch": 0.12266295333073945,
      "grad_norm": 0.28395840525627136,
      "learning_rate": 0.0004,
      "loss": 5.0702,
      "step": 5990
    },
    {
      "epoch": 0.12286773288554871,
      "grad_norm": 0.3164820373058319,
      "learning_rate": 0.0004,
      "loss": 5.0433,
      "step": 6000
    },
    {
      "epoch": 0.12286773288554871,
      "eval_loss": 5.105718612670898,
      "eval_runtime": 4.4155,
      "eval_samples_per_second": 264.072,
      "eval_steps_per_second": 33.066,
      "step": 6000
    },
    {
      "epoch": 0.12307251244035795,
      "grad_norm": 0.2929603159427643,
      "learning_rate": 0.0004,
      "loss": 5.0877,
      "step": 6010
    },
    {
      "epoch": 0.1232772919951672,
      "grad_norm": 0.2534056603908539,
      "learning_rate": 0.0004,
      "loss": 5.0613,
      "step": 6020
    },
    {
      "epoch": 0.12348207154997645,
      "grad_norm": 0.2689981758594513,
      "learning_rate": 0.0004,
      "loss": 5.0937,
      "step": 6030
    },
    {
      "epoch": 0.1236868511047857,
      "grad_norm": 0.2826198935508728,
      "learning_rate": 0.0004,
      "loss": 5.0865,
      "step": 6040
    },
    {
      "epoch": 0.12389163065959495,
      "grad_norm": 0.3296550512313843,
      "learning_rate": 0.0004,
      "loss": 5.0676,
      "step": 6050
    },
    {
      "epoch": 0.1240964102144042,
      "grad_norm": 0.2820937931537628,
      "learning_rate": 0.0004,
      "loss": 5.059,
      "step": 6060
    },
    {
      "epoch": 0.12430118976921344,
      "grad_norm": 0.2708949148654938,
      "learning_rate": 0.0004,
      "loss": 5.1074,
      "step": 6070
    },
    {
      "epoch": 0.12450596932402269,
      "grad_norm": 0.28163817524909973,
      "learning_rate": 0.0004,
      "loss": 5.0812,
      "step": 6080
    },
    {
      "epoch": 0.12471074887883193,
      "grad_norm": 0.2787400782108307,
      "learning_rate": 0.0004,
      "loss": 5.0943,
      "step": 6090
    },
    {
      "epoch": 0.12491552843364119,
      "grad_norm": 0.29888108372688293,
      "learning_rate": 0.0004,
      "loss": 5.0569,
      "step": 6100
    },
    {
      "epoch": 0.12512030798845045,
      "grad_norm": 0.31003960967063904,
      "learning_rate": 0.0004,
      "loss": 5.0997,
      "step": 6110
    },
    {
      "epoch": 0.1253250875432597,
      "grad_norm": 0.3110891878604889,
      "learning_rate": 0.0004,
      "loss": 5.1029,
      "step": 6120
    },
    {
      "epoch": 0.12552986709806893,
      "grad_norm": 0.27945610880851746,
      "learning_rate": 0.0004,
      "loss": 5.0837,
      "step": 6130
    },
    {
      "epoch": 0.12573464665287817,
      "grad_norm": 0.3528691232204437,
      "learning_rate": 0.0004,
      "loss": 5.1196,
      "step": 6140
    },
    {
      "epoch": 0.12593942620768742,
      "grad_norm": 0.34919026494026184,
      "learning_rate": 0.0004,
      "loss": 5.0934,
      "step": 6150
    },
    {
      "epoch": 0.1261442057624967,
      "grad_norm": 0.3034595549106598,
      "learning_rate": 0.0004,
      "loss": 5.0804,
      "step": 6160
    },
    {
      "epoch": 0.12634898531730593,
      "grad_norm": 0.29646971821784973,
      "learning_rate": 0.0004,
      "loss": 5.1081,
      "step": 6170
    },
    {
      "epoch": 0.12655376487211517,
      "grad_norm": 0.3521294593811035,
      "learning_rate": 0.0004,
      "loss": 5.0633,
      "step": 6180
    },
    {
      "epoch": 0.1267585444269244,
      "grad_norm": 0.2630900740623474,
      "learning_rate": 0.0004,
      "loss": 5.0545,
      "step": 6190
    },
    {
      "epoch": 0.12696332398173366,
      "grad_norm": 0.27108538150787354,
      "learning_rate": 0.0004,
      "loss": 5.0702,
      "step": 6200
    },
    {
      "epoch": 0.1271681035365429,
      "grad_norm": 0.28313174843788147,
      "learning_rate": 0.0004,
      "loss": 5.0878,
      "step": 6210
    },
    {
      "epoch": 0.12737288309135217,
      "grad_norm": 0.2536354064941406,
      "learning_rate": 0.0004,
      "loss": 5.0829,
      "step": 6220
    },
    {
      "epoch": 0.1275776626461614,
      "grad_norm": 0.3007798194885254,
      "learning_rate": 0.0004,
      "loss": 5.1176,
      "step": 6230
    },
    {
      "epoch": 0.12778244220097065,
      "grad_norm": 0.290460467338562,
      "learning_rate": 0.0004,
      "loss": 5.0617,
      "step": 6240
    },
    {
      "epoch": 0.1279872217557799,
      "grad_norm": 0.27415731549263,
      "learning_rate": 0.0004,
      "loss": 5.0808,
      "step": 6250
    },
    {
      "epoch": 0.12819200131058914,
      "grad_norm": 0.28427478671073914,
      "learning_rate": 0.0004,
      "loss": 5.101,
      "step": 6260
    },
    {
      "epoch": 0.1283967808653984,
      "grad_norm": 0.7649832963943481,
      "learning_rate": 0.0004,
      "loss": 5.0595,
      "step": 6270
    },
    {
      "epoch": 0.12860156042020765,
      "grad_norm": 0.28299009799957275,
      "learning_rate": 0.0004,
      "loss": 5.0721,
      "step": 6280
    },
    {
      "epoch": 0.1288063399750169,
      "grad_norm": 0.27343279123306274,
      "learning_rate": 0.0004,
      "loss": 5.0611,
      "step": 6290
    },
    {
      "epoch": 0.12901111952982613,
      "grad_norm": 0.27850478887557983,
      "learning_rate": 0.0004,
      "loss": 5.059,
      "step": 6300
    },
    {
      "epoch": 0.12921589908463538,
      "grad_norm": 0.2900664806365967,
      "learning_rate": 0.0004,
      "loss": 5.0783,
      "step": 6310
    },
    {
      "epoch": 0.12942067863944465,
      "grad_norm": 0.2805745005607605,
      "learning_rate": 0.0004,
      "loss": 5.048,
      "step": 6320
    },
    {
      "epoch": 0.1296254581942539,
      "grad_norm": 0.2737388610839844,
      "learning_rate": 0.0004,
      "loss": 5.0706,
      "step": 6330
    },
    {
      "epoch": 0.12983023774906313,
      "grad_norm": 0.2718786895275116,
      "learning_rate": 0.0004,
      "loss": 5.0774,
      "step": 6340
    },
    {
      "epoch": 0.13003501730387237,
      "grad_norm": 0.304569810628891,
      "learning_rate": 0.0004,
      "loss": 5.0472,
      "step": 6350
    },
    {
      "epoch": 0.13023979685868162,
      "grad_norm": 0.2766701877117157,
      "learning_rate": 0.0004,
      "loss": 5.0896,
      "step": 6360
    },
    {
      "epoch": 0.1304445764134909,
      "grad_norm": 0.30049368739128113,
      "learning_rate": 0.0004,
      "loss": 5.055,
      "step": 6370
    },
    {
      "epoch": 0.13064935596830013,
      "grad_norm": 0.2772940993309021,
      "learning_rate": 0.0004,
      "loss": 5.0653,
      "step": 6380
    },
    {
      "epoch": 0.13085413552310937,
      "grad_norm": 0.26545464992523193,
      "learning_rate": 0.0004,
      "loss": 5.0381,
      "step": 6390
    },
    {
      "epoch": 0.13105891507791861,
      "grad_norm": 0.28183674812316895,
      "learning_rate": 0.0004,
      "loss": 5.0518,
      "step": 6400
    },
    {
      "epoch": 0.13126369463272786,
      "grad_norm": 0.2801971733570099,
      "learning_rate": 0.0004,
      "loss": 5.0005,
      "step": 6410
    },
    {
      "epoch": 0.13146847418753713,
      "grad_norm": 0.28805556893348694,
      "learning_rate": 0.0004,
      "loss": 5.0386,
      "step": 6420
    },
    {
      "epoch": 0.13167325374234637,
      "grad_norm": 0.32158318161964417,
      "learning_rate": 0.0004,
      "loss": 5.0674,
      "step": 6430
    },
    {
      "epoch": 0.1318780332971556,
      "grad_norm": 0.28521737456321716,
      "learning_rate": 0.0004,
      "loss": 5.0649,
      "step": 6440
    },
    {
      "epoch": 0.13208281285196485,
      "grad_norm": 0.34175628423690796,
      "learning_rate": 0.0004,
      "loss": 5.0622,
      "step": 6450
    },
    {
      "epoch": 0.1322875924067741,
      "grad_norm": 0.5137848258018494,
      "learning_rate": 0.0004,
      "loss": 5.0687,
      "step": 6460
    },
    {
      "epoch": 0.13249237196158337,
      "grad_norm": 0.2890058755874634,
      "learning_rate": 0.0004,
      "loss": 5.0341,
      "step": 6470
    },
    {
      "epoch": 0.1326971515163926,
      "grad_norm": 0.2562602162361145,
      "learning_rate": 0.0004,
      "loss": 5.0585,
      "step": 6480
    },
    {
      "epoch": 0.13290193107120185,
      "grad_norm": 0.2638840973377228,
      "learning_rate": 0.0004,
      "loss": 5.0344,
      "step": 6490
    },
    {
      "epoch": 0.1331067106260111,
      "grad_norm": 0.25415486097335815,
      "learning_rate": 0.0004,
      "loss": 5.0503,
      "step": 6500
    },
    {
      "epoch": 0.13331149018082034,
      "grad_norm": 0.27743104100227356,
      "learning_rate": 0.0004,
      "loss": 5.0592,
      "step": 6510
    },
    {
      "epoch": 0.1335162697356296,
      "grad_norm": 0.2603805363178253,
      "learning_rate": 0.0004,
      "loss": 5.0396,
      "step": 6520
    },
    {
      "epoch": 0.13372104929043885,
      "grad_norm": 0.28930333256721497,
      "learning_rate": 0.0004,
      "loss": 5.0449,
      "step": 6530
    },
    {
      "epoch": 0.1339258288452481,
      "grad_norm": 0.33630481362342834,
      "learning_rate": 0.0004,
      "loss": 5.1177,
      "step": 6540
    },
    {
      "epoch": 0.13413060840005733,
      "grad_norm": 0.334529846906662,
      "learning_rate": 0.0004,
      "loss": 5.0603,
      "step": 6550
    },
    {
      "epoch": 0.13433538795486658,
      "grad_norm": 0.27738118171691895,
      "learning_rate": 0.0004,
      "loss": 5.0817,
      "step": 6560
    },
    {
      "epoch": 0.13454016750967585,
      "grad_norm": 0.28533002734184265,
      "learning_rate": 0.0004,
      "loss": 5.0557,
      "step": 6570
    },
    {
      "epoch": 0.1347449470644851,
      "grad_norm": 0.28852105140686035,
      "learning_rate": 0.0004,
      "loss": 5.0284,
      "step": 6580
    },
    {
      "epoch": 0.13494972661929433,
      "grad_norm": 0.24459971487522125,
      "learning_rate": 0.0004,
      "loss": 5.0455,
      "step": 6590
    },
    {
      "epoch": 0.13515450617410357,
      "grad_norm": 0.25609290599823,
      "learning_rate": 0.0004,
      "loss": 5.0603,
      "step": 6600
    },
    {
      "epoch": 0.13535928572891281,
      "grad_norm": 0.3024841248989105,
      "learning_rate": 0.0004,
      "loss": 5.0508,
      "step": 6610
    },
    {
      "epoch": 0.13556406528372208,
      "grad_norm": 0.2652977705001831,
      "learning_rate": 0.0004,
      "loss": 5.0262,
      "step": 6620
    },
    {
      "epoch": 0.13576884483853133,
      "grad_norm": 0.28360605239868164,
      "learning_rate": 0.0004,
      "loss": 5.0439,
      "step": 6630
    },
    {
      "epoch": 0.13597362439334057,
      "grad_norm": 0.28990018367767334,
      "learning_rate": 0.0004,
      "loss": 5.0549,
      "step": 6640
    },
    {
      "epoch": 0.1361784039481498,
      "grad_norm": 0.2533547878265381,
      "learning_rate": 0.0004,
      "loss": 5.018,
      "step": 6650
    },
    {
      "epoch": 0.13638318350295905,
      "grad_norm": 0.2740333676338196,
      "learning_rate": 0.0004,
      "loss": 5.0678,
      "step": 6660
    },
    {
      "epoch": 0.13658796305776832,
      "grad_norm": 0.29817625880241394,
      "learning_rate": 0.0004,
      "loss": 5.0712,
      "step": 6670
    },
    {
      "epoch": 0.13679274261257757,
      "grad_norm": 0.31441426277160645,
      "learning_rate": 0.0004,
      "loss": 5.0656,
      "step": 6680
    },
    {
      "epoch": 0.1369975221673868,
      "grad_norm": 0.3231841027736664,
      "learning_rate": 0.0004,
      "loss": 5.0736,
      "step": 6690
    },
    {
      "epoch": 0.13720230172219605,
      "grad_norm": 0.30328667163848877,
      "learning_rate": 0.0004,
      "loss": 5.0546,
      "step": 6700
    },
    {
      "epoch": 0.1374070812770053,
      "grad_norm": 0.29146936535835266,
      "learning_rate": 0.0004,
      "loss": 5.0231,
      "step": 6710
    },
    {
      "epoch": 0.13761186083181456,
      "grad_norm": 0.26964816451072693,
      "learning_rate": 0.0004,
      "loss": 5.0198,
      "step": 6720
    },
    {
      "epoch": 0.1378166403866238,
      "grad_norm": 0.29934099316596985,
      "learning_rate": 0.0004,
      "loss": 5.0214,
      "step": 6730
    },
    {
      "epoch": 0.13802141994143305,
      "grad_norm": 0.31749799847602844,
      "learning_rate": 0.0004,
      "loss": 5.024,
      "step": 6740
    },
    {
      "epoch": 0.1382261994962423,
      "grad_norm": 0.27449607849121094,
      "learning_rate": 0.0004,
      "loss": 5.0448,
      "step": 6750
    },
    {
      "epoch": 0.13843097905105153,
      "grad_norm": 0.2636013627052307,
      "learning_rate": 0.0004,
      "loss": 5.0019,
      "step": 6760
    },
    {
      "epoch": 0.1386357586058608,
      "grad_norm": 0.2735285758972168,
      "learning_rate": 0.0004,
      "loss": 5.0317,
      "step": 6770
    },
    {
      "epoch": 0.13884053816067005,
      "grad_norm": 0.27161097526550293,
      "learning_rate": 0.0004,
      "loss": 5.0621,
      "step": 6780
    },
    {
      "epoch": 0.1390453177154793,
      "grad_norm": 0.2726326882839203,
      "learning_rate": 0.0004,
      "loss": 5.0115,
      "step": 6790
    },
    {
      "epoch": 0.13925009727028853,
      "grad_norm": 0.2701372802257538,
      "learning_rate": 0.0004,
      "loss": 5.0728,
      "step": 6800
    },
    {
      "epoch": 0.13945487682509777,
      "grad_norm": 0.28725504875183105,
      "learning_rate": 0.0004,
      "loss": 5.0328,
      "step": 6810
    },
    {
      "epoch": 0.13965965637990704,
      "grad_norm": 0.2574363946914673,
      "learning_rate": 0.0004,
      "loss": 5.0259,
      "step": 6820
    },
    {
      "epoch": 0.13986443593471629,
      "grad_norm": 0.3175942003726959,
      "learning_rate": 0.0004,
      "loss": 5.0614,
      "step": 6830
    },
    {
      "epoch": 0.14006921548952553,
      "grad_norm": 0.27931883931159973,
      "learning_rate": 0.0004,
      "loss": 5.0172,
      "step": 6840
    },
    {
      "epoch": 0.14027399504433477,
      "grad_norm": 0.2712056040763855,
      "learning_rate": 0.0004,
      "loss": 5.0233,
      "step": 6850
    },
    {
      "epoch": 0.140478774599144,
      "grad_norm": 0.2554517090320587,
      "learning_rate": 0.0004,
      "loss": 5.0289,
      "step": 6860
    },
    {
      "epoch": 0.14068355415395328,
      "grad_norm": 0.27673497796058655,
      "learning_rate": 0.0004,
      "loss": 5.0249,
      "step": 6870
    },
    {
      "epoch": 0.14088833370876253,
      "grad_norm": 0.28359177708625793,
      "learning_rate": 0.0004,
      "loss": 5.055,
      "step": 6880
    },
    {
      "epoch": 0.14109311326357177,
      "grad_norm": 0.9358149766921997,
      "learning_rate": 0.0004,
      "loss": 5.0159,
      "step": 6890
    },
    {
      "epoch": 0.141297892818381,
      "grad_norm": 0.29279693961143494,
      "learning_rate": 0.0004,
      "loss": 5.0301,
      "step": 6900
    },
    {
      "epoch": 0.14150267237319025,
      "grad_norm": 0.2918263077735901,
      "learning_rate": 0.0004,
      "loss": 5.03,
      "step": 6910
    },
    {
      "epoch": 0.1417074519279995,
      "grad_norm": 0.2591414749622345,
      "learning_rate": 0.0004,
      "loss": 4.9956,
      "step": 6920
    },
    {
      "epoch": 0.14191223148280876,
      "grad_norm": 0.2563314139842987,
      "learning_rate": 0.0004,
      "loss": 5.0245,
      "step": 6930
    },
    {
      "epoch": 0.142117011037618,
      "grad_norm": 0.29098746180534363,
      "learning_rate": 0.0004,
      "loss": 5.0174,
      "step": 6940
    },
    {
      "epoch": 0.14232179059242725,
      "grad_norm": 0.3085611164569855,
      "learning_rate": 0.0004,
      "loss": 5.0243,
      "step": 6950
    },
    {
      "epoch": 0.1425265701472365,
      "grad_norm": 0.26494288444519043,
      "learning_rate": 0.0004,
      "loss": 5.0631,
      "step": 6960
    },
    {
      "epoch": 0.14273134970204573,
      "grad_norm": 0.24861690402030945,
      "learning_rate": 0.0004,
      "loss": 5.036,
      "step": 6970
    },
    {
      "epoch": 0.142936129256855,
      "grad_norm": 0.2928735613822937,
      "learning_rate": 0.0004,
      "loss": 5.06,
      "step": 6980
    },
    {
      "epoch": 0.14314090881166425,
      "grad_norm": 0.3489916920661926,
      "learning_rate": 0.0004,
      "loss": 5.0146,
      "step": 6990
    },
    {
      "epoch": 0.1433456883664735,
      "grad_norm": 0.29149192571640015,
      "learning_rate": 0.0004,
      "loss": 4.9889,
      "step": 7000
    },
    {
      "epoch": 0.14355046792128273,
      "grad_norm": 0.26062437891960144,
      "learning_rate": 0.0004,
      "loss": 5.0303,
      "step": 7010
    },
    {
      "epoch": 0.14375524747609197,
      "grad_norm": 0.3030029237270355,
      "learning_rate": 0.0004,
      "loss": 4.9966,
      "step": 7020
    },
    {
      "epoch": 0.14396002703090124,
      "grad_norm": 0.2680526375770569,
      "learning_rate": 0.0004,
      "loss": 5.0185,
      "step": 7030
    },
    {
      "epoch": 0.1441648065857105,
      "grad_norm": 0.31157636642456055,
      "learning_rate": 0.0004,
      "loss": 5.0236,
      "step": 7040
    },
    {
      "epoch": 0.14436958614051973,
      "grad_norm": 0.2734978497028351,
      "learning_rate": 0.0004,
      "loss": 5.0232,
      "step": 7050
    },
    {
      "epoch": 0.14457436569532897,
      "grad_norm": 0.2717677354812622,
      "learning_rate": 0.0004,
      "loss": 4.9828,
      "step": 7060
    },
    {
      "epoch": 0.1447791452501382,
      "grad_norm": 0.2949996590614319,
      "learning_rate": 0.0004,
      "loss": 5.0299,
      "step": 7070
    },
    {
      "epoch": 0.14498392480494748,
      "grad_norm": 0.27881181240081787,
      "learning_rate": 0.0004,
      "loss": 5.0718,
      "step": 7080
    },
    {
      "epoch": 0.14518870435975673,
      "grad_norm": 0.2645169198513031,
      "learning_rate": 0.0004,
      "loss": 5.0287,
      "step": 7090
    },
    {
      "epoch": 0.14539348391456597,
      "grad_norm": 0.258129358291626,
      "learning_rate": 0.0004,
      "loss": 4.9836,
      "step": 7100
    },
    {
      "epoch": 0.1455982634693752,
      "grad_norm": 0.2882113754749298,
      "learning_rate": 0.0004,
      "loss": 5.0085,
      "step": 7110
    },
    {
      "epoch": 0.14580304302418445,
      "grad_norm": 0.27890050411224365,
      "learning_rate": 0.0004,
      "loss": 5.0173,
      "step": 7120
    },
    {
      "epoch": 0.14600782257899372,
      "grad_norm": 0.2726353704929352,
      "learning_rate": 0.0004,
      "loss": 4.9976,
      "step": 7130
    },
    {
      "epoch": 0.14621260213380297,
      "grad_norm": 0.26558926701545715,
      "learning_rate": 0.0004,
      "loss": 5.0114,
      "step": 7140
    },
    {
      "epoch": 0.1464173816886122,
      "grad_norm": 0.26955997943878174,
      "learning_rate": 0.0004,
      "loss": 5.0349,
      "step": 7150
    },
    {
      "epoch": 0.14662216124342145,
      "grad_norm": 0.31367459893226624,
      "learning_rate": 0.0004,
      "loss": 4.9823,
      "step": 7160
    },
    {
      "epoch": 0.1468269407982307,
      "grad_norm": 0.2597455084323883,
      "learning_rate": 0.0004,
      "loss": 5.034,
      "step": 7170
    },
    {
      "epoch": 0.14703172035303996,
      "grad_norm": 0.3116661012172699,
      "learning_rate": 0.0004,
      "loss": 5.0626,
      "step": 7180
    },
    {
      "epoch": 0.1472364999078492,
      "grad_norm": 0.28304263949394226,
      "learning_rate": 0.0004,
      "loss": 5.0162,
      "step": 7190
    },
    {
      "epoch": 0.14744127946265845,
      "grad_norm": 0.2740379571914673,
      "learning_rate": 0.0004,
      "loss": 4.9967,
      "step": 7200
    },
    {
      "epoch": 0.1476460590174677,
      "grad_norm": 0.297842413187027,
      "learning_rate": 0.0004,
      "loss": 5.0032,
      "step": 7210
    },
    {
      "epoch": 0.14785083857227693,
      "grad_norm": 0.29578113555908203,
      "learning_rate": 0.0004,
      "loss": 5.0335,
      "step": 7220
    },
    {
      "epoch": 0.1480556181270862,
      "grad_norm": 0.2504656910896301,
      "learning_rate": 0.0004,
      "loss": 5.0501,
      "step": 7230
    },
    {
      "epoch": 0.14826039768189544,
      "grad_norm": 0.2642020285129547,
      "learning_rate": 0.0004,
      "loss": 5.0067,
      "step": 7240
    },
    {
      "epoch": 0.1484651772367047,
      "grad_norm": 0.2889758348464966,
      "learning_rate": 0.0004,
      "loss": 5.0189,
      "step": 7250
    },
    {
      "epoch": 0.14866995679151393,
      "grad_norm": 0.25152787566185,
      "learning_rate": 0.0004,
      "loss": 5.0003,
      "step": 7260
    },
    {
      "epoch": 0.14887473634632317,
      "grad_norm": 0.26755887269973755,
      "learning_rate": 0.0004,
      "loss": 5.0276,
      "step": 7270
    },
    {
      "epoch": 0.14907951590113244,
      "grad_norm": 0.27270573377609253,
      "learning_rate": 0.0004,
      "loss": 5.0269,
      "step": 7280
    },
    {
      "epoch": 0.14928429545594168,
      "grad_norm": 0.28820136189460754,
      "learning_rate": 0.0004,
      "loss": 5.0243,
      "step": 7290
    },
    {
      "epoch": 0.14948907501075093,
      "grad_norm": 0.26135873794555664,
      "learning_rate": 0.0004,
      "loss": 5.0132,
      "step": 7300
    },
    {
      "epoch": 0.14969385456556017,
      "grad_norm": 0.2627226710319519,
      "learning_rate": 0.0004,
      "loss": 5.0009,
      "step": 7310
    },
    {
      "epoch": 0.1498986341203694,
      "grad_norm": 0.2626127004623413,
      "learning_rate": 0.0004,
      "loss": 5.0199,
      "step": 7320
    },
    {
      "epoch": 0.15010341367517868,
      "grad_norm": 0.2863093912601471,
      "learning_rate": 0.0004,
      "loss": 4.9827,
      "step": 7330
    },
    {
      "epoch": 0.15030819322998792,
      "grad_norm": 0.2922162711620331,
      "learning_rate": 0.0004,
      "loss": 4.9776,
      "step": 7340
    },
    {
      "epoch": 0.15051297278479717,
      "grad_norm": 0.2657029330730438,
      "learning_rate": 0.0004,
      "loss": 5.0001,
      "step": 7350
    },
    {
      "epoch": 0.1507177523396064,
      "grad_norm": 0.2926842272281647,
      "learning_rate": 0.0004,
      "loss": 4.9807,
      "step": 7360
    },
    {
      "epoch": 0.15092253189441565,
      "grad_norm": 0.2773357033729553,
      "learning_rate": 0.0004,
      "loss": 5.0257,
      "step": 7370
    },
    {
      "epoch": 0.15112731144922492,
      "grad_norm": 0.277850866317749,
      "learning_rate": 0.0004,
      "loss": 5.0355,
      "step": 7380
    },
    {
      "epoch": 0.15133209100403416,
      "grad_norm": 0.2935318648815155,
      "learning_rate": 0.0004,
      "loss": 5.0019,
      "step": 7390
    },
    {
      "epoch": 0.1515368705588434,
      "grad_norm": 0.25857090950012207,
      "learning_rate": 0.0004,
      "loss": 5.0039,
      "step": 7400
    },
    {
      "epoch": 0.15174165011365265,
      "grad_norm": 0.26904380321502686,
      "learning_rate": 0.0004,
      "loss": 4.9963,
      "step": 7410
    },
    {
      "epoch": 0.1519464296684619,
      "grad_norm": 0.29226887226104736,
      "learning_rate": 0.0004,
      "loss": 4.9976,
      "step": 7420
    },
    {
      "epoch": 0.15215120922327116,
      "grad_norm": 0.2924584448337555,
      "learning_rate": 0.0004,
      "loss": 5.0066,
      "step": 7430
    },
    {
      "epoch": 0.1523559887780804,
      "grad_norm": 0.3263062536716461,
      "learning_rate": 0.0004,
      "loss": 5.0121,
      "step": 7440
    },
    {
      "epoch": 0.15256076833288965,
      "grad_norm": 0.2633025348186493,
      "learning_rate": 0.0004,
      "loss": 5.0047,
      "step": 7450
    },
    {
      "epoch": 0.1527655478876989,
      "grad_norm": 0.2768848240375519,
      "learning_rate": 0.0004,
      "loss": 4.9915,
      "step": 7460
    },
    {
      "epoch": 0.15297032744250813,
      "grad_norm": 0.2501979172229767,
      "learning_rate": 0.0004,
      "loss": 5.0219,
      "step": 7470
    },
    {
      "epoch": 0.1531751069973174,
      "grad_norm": 0.28026828169822693,
      "learning_rate": 0.0004,
      "loss": 5.0037,
      "step": 7480
    },
    {
      "epoch": 0.15337988655212664,
      "grad_norm": 0.2698480784893036,
      "learning_rate": 0.0004,
      "loss": 4.9814,
      "step": 7490
    },
    {
      "epoch": 0.15358466610693589,
      "grad_norm": 0.239173024892807,
      "learning_rate": 0.0004,
      "loss": 5.0149,
      "step": 7500
    },
    {
      "epoch": 0.15378944566174513,
      "grad_norm": 0.27128395438194275,
      "learning_rate": 0.0004,
      "loss": 5.0123,
      "step": 7510
    },
    {
      "epoch": 0.15399422521655437,
      "grad_norm": 0.26015204191207886,
      "learning_rate": 0.0004,
      "loss": 5.0152,
      "step": 7520
    },
    {
      "epoch": 0.15419900477136364,
      "grad_norm": 0.27965638041496277,
      "learning_rate": 0.0004,
      "loss": 4.9905,
      "step": 7530
    },
    {
      "epoch": 0.15440378432617288,
      "grad_norm": 0.2805701494216919,
      "learning_rate": 0.0004,
      "loss": 4.978,
      "step": 7540
    },
    {
      "epoch": 0.15460856388098212,
      "grad_norm": 0.3172330856323242,
      "learning_rate": 0.0004,
      "loss": 4.996,
      "step": 7550
    },
    {
      "epoch": 0.15481334343579137,
      "grad_norm": 0.25200405716896057,
      "learning_rate": 0.0004,
      "loss": 4.9709,
      "step": 7560
    },
    {
      "epoch": 0.1550181229906006,
      "grad_norm": 0.2770288586616516,
      "learning_rate": 0.0004,
      "loss": 5.0222,
      "step": 7570
    },
    {
      "epoch": 0.15522290254540988,
      "grad_norm": 0.2640692889690399,
      "learning_rate": 0.0004,
      "loss": 5.007,
      "step": 7580
    },
    {
      "epoch": 0.15542768210021912,
      "grad_norm": 0.2653588354587555,
      "learning_rate": 0.0004,
      "loss": 4.9661,
      "step": 7590
    },
    {
      "epoch": 0.15563246165502836,
      "grad_norm": 0.2635646462440491,
      "learning_rate": 0.0004,
      "loss": 4.9884,
      "step": 7600
    },
    {
      "epoch": 0.1558372412098376,
      "grad_norm": 0.2831110954284668,
      "learning_rate": 0.0004,
      "loss": 4.9915,
      "step": 7610
    },
    {
      "epoch": 0.15604202076464685,
      "grad_norm": 0.2644828259944916,
      "learning_rate": 0.0004,
      "loss": 4.9793,
      "step": 7620
    },
    {
      "epoch": 0.1562468003194561,
      "grad_norm": 0.2570164203643799,
      "learning_rate": 0.0004,
      "loss": 4.9477,
      "step": 7630
    },
    {
      "epoch": 0.15645157987426536,
      "grad_norm": 0.27692949771881104,
      "learning_rate": 0.0004,
      "loss": 4.9855,
      "step": 7640
    },
    {
      "epoch": 0.1566563594290746,
      "grad_norm": 0.32910946011543274,
      "learning_rate": 0.0004,
      "loss": 5.003,
      "step": 7650
    },
    {
      "epoch": 0.15686113898388385,
      "grad_norm": 0.2637607753276825,
      "learning_rate": 0.0004,
      "loss": 4.971,
      "step": 7660
    },
    {
      "epoch": 0.1570659185386931,
      "grad_norm": 0.3097701966762543,
      "learning_rate": 0.0004,
      "loss": 5.0238,
      "step": 7670
    },
    {
      "epoch": 0.15727069809350233,
      "grad_norm": 0.30924656987190247,
      "learning_rate": 0.0004,
      "loss": 5.0226,
      "step": 7680
    },
    {
      "epoch": 0.1574754776483116,
      "grad_norm": 0.2889367938041687,
      "learning_rate": 0.0004,
      "loss": 5.023,
      "step": 7690
    },
    {
      "epoch": 0.15768025720312084,
      "grad_norm": 0.2582051455974579,
      "learning_rate": 0.0004,
      "loss": 4.9851,
      "step": 7700
    },
    {
      "epoch": 0.1578850367579301,
      "grad_norm": 0.4065515398979187,
      "learning_rate": 0.0004,
      "loss": 4.998,
      "step": 7710
    },
    {
      "epoch": 0.15808981631273933,
      "grad_norm": 0.2995768189430237,
      "learning_rate": 0.0004,
      "loss": 5.0239,
      "step": 7720
    },
    {
      "epoch": 0.15829459586754857,
      "grad_norm": 0.2907010614871979,
      "learning_rate": 0.0004,
      "loss": 5.0208,
      "step": 7730
    },
    {
      "epoch": 0.15849937542235784,
      "grad_norm": 0.2729869782924652,
      "learning_rate": 0.0004,
      "loss": 4.9835,
      "step": 7740
    },
    {
      "epoch": 0.15870415497716708,
      "grad_norm": 2.6075010299682617,
      "learning_rate": 0.0004,
      "loss": 4.985,
      "step": 7750
    },
    {
      "epoch": 0.15890893453197633,
      "grad_norm": 0.27993106842041016,
      "learning_rate": 0.0004,
      "loss": 5.0196,
      "step": 7760
    },
    {
      "epoch": 0.15911371408678557,
      "grad_norm": 0.27130773663520813,
      "learning_rate": 0.0004,
      "loss": 4.9265,
      "step": 7770
    },
    {
      "epoch": 0.1593184936415948,
      "grad_norm": 0.3074718713760376,
      "learning_rate": 0.0004,
      "loss": 4.9919,
      "step": 7780
    },
    {
      "epoch": 0.15952327319640408,
      "grad_norm": 0.3180055022239685,
      "learning_rate": 0.0004,
      "loss": 4.9662,
      "step": 7790
    },
    {
      "epoch": 0.15972805275121332,
      "grad_norm": 0.25885850191116333,
      "learning_rate": 0.0004,
      "loss": 4.9778,
      "step": 7800
    },
    {
      "epoch": 0.15993283230602257,
      "grad_norm": 0.2722429037094116,
      "learning_rate": 0.0004,
      "loss": 4.9624,
      "step": 7810
    },
    {
      "epoch": 0.1601376118608318,
      "grad_norm": 0.2777957618236542,
      "learning_rate": 0.0004,
      "loss": 4.9673,
      "step": 7820
    },
    {
      "epoch": 0.16034239141564105,
      "grad_norm": 0.24738921225070953,
      "learning_rate": 0.0004,
      "loss": 4.9967,
      "step": 7830
    },
    {
      "epoch": 0.16054717097045032,
      "grad_norm": 0.2691292464733124,
      "learning_rate": 0.0004,
      "loss": 4.9649,
      "step": 7840
    },
    {
      "epoch": 0.16075195052525956,
      "grad_norm": 0.25139743089675903,
      "learning_rate": 0.0004,
      "loss": 4.9643,
      "step": 7850
    },
    {
      "epoch": 0.1609567300800688,
      "grad_norm": 0.24456483125686646,
      "learning_rate": 0.0004,
      "loss": 4.9518,
      "step": 7860
    },
    {
      "epoch": 0.16116150963487805,
      "grad_norm": 0.2848648428916931,
      "learning_rate": 0.0004,
      "loss": 4.9543,
      "step": 7870
    },
    {
      "epoch": 0.1613662891896873,
      "grad_norm": 0.28265640139579773,
      "learning_rate": 0.0004,
      "loss": 4.9814,
      "step": 7880
    },
    {
      "epoch": 0.16157106874449656,
      "grad_norm": 0.25388023257255554,
      "learning_rate": 0.0004,
      "loss": 4.9623,
      "step": 7890
    },
    {
      "epoch": 0.1617758482993058,
      "grad_norm": 0.277907133102417,
      "learning_rate": 0.0004,
      "loss": 5.0045,
      "step": 7900
    },
    {
      "epoch": 0.16198062785411504,
      "grad_norm": 0.33783209323883057,
      "learning_rate": 0.0004,
      "loss": 5.014,
      "step": 7910
    },
    {
      "epoch": 0.1621854074089243,
      "grad_norm": 0.4126969277858734,
      "learning_rate": 0.0004,
      "loss": 4.9839,
      "step": 7920
    },
    {
      "epoch": 0.16239018696373353,
      "grad_norm": 0.2501499652862549,
      "learning_rate": 0.0004,
      "loss": 4.997,
      "step": 7930
    },
    {
      "epoch": 0.1625949665185428,
      "grad_norm": 0.31211671233177185,
      "learning_rate": 0.0004,
      "loss": 4.9882,
      "step": 7940
    },
    {
      "epoch": 0.16279974607335204,
      "grad_norm": 0.28681033849716187,
      "learning_rate": 0.0004,
      "loss": 4.9808,
      "step": 7950
    },
    {
      "epoch": 0.16300452562816128,
      "grad_norm": 0.30036285519599915,
      "learning_rate": 0.0004,
      "loss": 4.9682,
      "step": 7960
    },
    {
      "epoch": 0.16320930518297053,
      "grad_norm": 0.2959885001182556,
      "learning_rate": 0.0004,
      "loss": 4.9675,
      "step": 7970
    },
    {
      "epoch": 0.16341408473777977,
      "grad_norm": 0.2631414532661438,
      "learning_rate": 0.0004,
      "loss": 4.932,
      "step": 7980
    },
    {
      "epoch": 0.16361886429258904,
      "grad_norm": 0.2474392056465149,
      "learning_rate": 0.0004,
      "loss": 4.9942,
      "step": 7990
    },
    {
      "epoch": 0.16382364384739828,
      "grad_norm": 0.3146931827068329,
      "learning_rate": 0.0004,
      "loss": 4.9568,
      "step": 8000
    },
    {
      "epoch": 0.16382364384739828,
      "eval_loss": 4.985713005065918,
      "eval_runtime": 4.385,
      "eval_samples_per_second": 265.906,
      "eval_steps_per_second": 33.295,
      "step": 8000
    },
    {
      "epoch": 0.16402842340220752,
      "grad_norm": 0.27640268206596375,
      "learning_rate": 0.0004,
      "loss": 4.8895,
      "step": 8010
    },
    {
      "epoch": 0.16423320295701677,
      "grad_norm": 0.26192760467529297,
      "learning_rate": 0.0004,
      "loss": 4.9724,
      "step": 8020
    },
    {
      "epoch": 0.164437982511826,
      "grad_norm": 0.29010656476020813,
      "learning_rate": 0.0004,
      "loss": 4.9561,
      "step": 8030
    },
    {
      "epoch": 0.16464276206663528,
      "grad_norm": 0.2687608599662781,
      "learning_rate": 0.0004,
      "loss": 4.9903,
      "step": 8040
    },
    {
      "epoch": 0.16484754162144452,
      "grad_norm": 0.38182979822158813,
      "learning_rate": 0.0004,
      "loss": 4.9518,
      "step": 8050
    },
    {
      "epoch": 0.16505232117625376,
      "grad_norm": 0.253173291683197,
      "learning_rate": 0.0004,
      "loss": 4.9921,
      "step": 8060
    },
    {
      "epoch": 0.165257100731063,
      "grad_norm": 0.2626391053199768,
      "learning_rate": 0.0004,
      "loss": 4.9634,
      "step": 8070
    },
    {
      "epoch": 0.16546188028587225,
      "grad_norm": 0.28041955828666687,
      "learning_rate": 0.0004,
      "loss": 4.9432,
      "step": 8080
    },
    {
      "epoch": 0.16566665984068152,
      "grad_norm": 0.2790217399597168,
      "learning_rate": 0.0004,
      "loss": 4.9541,
      "step": 8090
    },
    {
      "epoch": 0.16587143939549076,
      "grad_norm": 0.2618565857410431,
      "learning_rate": 0.0004,
      "loss": 4.9578,
      "step": 8100
    },
    {
      "epoch": 0.1660762189503,
      "grad_norm": 0.316425085067749,
      "learning_rate": 0.0004,
      "loss": 4.9673,
      "step": 8110
    },
    {
      "epoch": 0.16628099850510925,
      "grad_norm": 0.30675795674324036,
      "learning_rate": 0.0004,
      "loss": 4.9583,
      "step": 8120
    },
    {
      "epoch": 0.1664857780599185,
      "grad_norm": 0.2543947696685791,
      "learning_rate": 0.0004,
      "loss": 4.9544,
      "step": 8130
    },
    {
      "epoch": 0.16669055761472776,
      "grad_norm": 0.2793639302253723,
      "learning_rate": 0.0004,
      "loss": 4.9822,
      "step": 8140
    },
    {
      "epoch": 0.166895337169537,
      "grad_norm": 0.25796473026275635,
      "learning_rate": 0.0004,
      "loss": 5.0105,
      "step": 8150
    },
    {
      "epoch": 0.16710011672434624,
      "grad_norm": 0.25799325108528137,
      "learning_rate": 0.0004,
      "loss": 4.9849,
      "step": 8160
    },
    {
      "epoch": 0.16730489627915549,
      "grad_norm": 0.27757272124290466,
      "learning_rate": 0.0004,
      "loss": 4.9951,
      "step": 8170
    },
    {
      "epoch": 0.16750967583396473,
      "grad_norm": 0.26644569635391235,
      "learning_rate": 0.0004,
      "loss": 4.9706,
      "step": 8180
    },
    {
      "epoch": 0.167714455388774,
      "grad_norm": 0.25647637248039246,
      "learning_rate": 0.0004,
      "loss": 4.9509,
      "step": 8190
    },
    {
      "epoch": 0.16791923494358324,
      "grad_norm": 0.28371259570121765,
      "learning_rate": 0.0004,
      "loss": 4.9445,
      "step": 8200
    },
    {
      "epoch": 0.16812401449839248,
      "grad_norm": 0.2952437698841095,
      "learning_rate": 0.0004,
      "loss": 4.9562,
      "step": 8210
    },
    {
      "epoch": 0.16832879405320172,
      "grad_norm": 0.39984285831451416,
      "learning_rate": 0.0004,
      "loss": 4.9866,
      "step": 8220
    },
    {
      "epoch": 0.16853357360801097,
      "grad_norm": 0.2723577320575714,
      "learning_rate": 0.0004,
      "loss": 4.9398,
      "step": 8230
    },
    {
      "epoch": 0.16873835316282024,
      "grad_norm": 0.300861656665802,
      "learning_rate": 0.0004,
      "loss": 4.9795,
      "step": 8240
    },
    {
      "epoch": 0.16894313271762948,
      "grad_norm": 0.2561473250389099,
      "learning_rate": 0.0004,
      "loss": 4.941,
      "step": 8250
    },
    {
      "epoch": 0.16914791227243872,
      "grad_norm": 0.2530801594257355,
      "learning_rate": 0.0004,
      "loss": 4.9801,
      "step": 8260
    },
    {
      "epoch": 0.16935269182724796,
      "grad_norm": 0.2439454346895218,
      "learning_rate": 0.0004,
      "loss": 4.942,
      "step": 8270
    },
    {
      "epoch": 0.1695574713820572,
      "grad_norm": 0.26756080985069275,
      "learning_rate": 0.0004,
      "loss": 4.9709,
      "step": 8280
    },
    {
      "epoch": 0.16976225093686648,
      "grad_norm": 0.26916906237602234,
      "learning_rate": 0.0004,
      "loss": 4.9696,
      "step": 8290
    },
    {
      "epoch": 0.16996703049167572,
      "grad_norm": 0.26618748903274536,
      "learning_rate": 0.0004,
      "loss": 4.9541,
      "step": 8300
    },
    {
      "epoch": 0.17017181004648496,
      "grad_norm": 0.25927209854125977,
      "learning_rate": 0.0004,
      "loss": 4.9438,
      "step": 8310
    },
    {
      "epoch": 0.1703765896012942,
      "grad_norm": 0.28418412804603577,
      "learning_rate": 0.0004,
      "loss": 4.956,
      "step": 8320
    },
    {
      "epoch": 0.17058136915610345,
      "grad_norm": 0.275880366563797,
      "learning_rate": 0.0004,
      "loss": 4.9916,
      "step": 8330
    },
    {
      "epoch": 0.17078614871091272,
      "grad_norm": 0.2647210955619812,
      "learning_rate": 0.0004,
      "loss": 4.9571,
      "step": 8340
    },
    {
      "epoch": 0.17099092826572196,
      "grad_norm": 0.2847307324409485,
      "learning_rate": 0.0004,
      "loss": 4.9439,
      "step": 8350
    },
    {
      "epoch": 0.1711957078205312,
      "grad_norm": 0.2591257691383362,
      "learning_rate": 0.0004,
      "loss": 4.9707,
      "step": 8360
    },
    {
      "epoch": 0.17140048737534044,
      "grad_norm": 0.27943095564842224,
      "learning_rate": 0.0004,
      "loss": 4.9864,
      "step": 8370
    },
    {
      "epoch": 0.17160526693014969,
      "grad_norm": 0.27202245593070984,
      "learning_rate": 0.0004,
      "loss": 4.9512,
      "step": 8380
    },
    {
      "epoch": 0.17181004648495893,
      "grad_norm": 0.2789130210876465,
      "learning_rate": 0.0004,
      "loss": 4.9389,
      "step": 8390
    },
    {
      "epoch": 0.1720148260397682,
      "grad_norm": 0.2560030519962311,
      "learning_rate": 0.0004,
      "loss": 4.9418,
      "step": 8400
    },
    {
      "epoch": 0.17221960559457744,
      "grad_norm": 0.2689448595046997,
      "learning_rate": 0.0004,
      "loss": 4.961,
      "step": 8410
    },
    {
      "epoch": 0.17242438514938668,
      "grad_norm": 0.265110582113266,
      "learning_rate": 0.0004,
      "loss": 4.9237,
      "step": 8420
    },
    {
      "epoch": 0.17262916470419593,
      "grad_norm": 0.259001761674881,
      "learning_rate": 0.0004,
      "loss": 4.9322,
      "step": 8430
    },
    {
      "epoch": 0.17283394425900517,
      "grad_norm": 0.2653099298477173,
      "learning_rate": 0.0004,
      "loss": 4.9607,
      "step": 8440
    },
    {
      "epoch": 0.17303872381381444,
      "grad_norm": 0.2665369510650635,
      "learning_rate": 0.0004,
      "loss": 4.9522,
      "step": 8450
    },
    {
      "epoch": 0.17324350336862368,
      "grad_norm": 0.2588624954223633,
      "learning_rate": 0.0004,
      "loss": 4.9419,
      "step": 8460
    },
    {
      "epoch": 0.17344828292343292,
      "grad_norm": 0.2519038915634155,
      "learning_rate": 0.0004,
      "loss": 4.946,
      "step": 8470
    },
    {
      "epoch": 0.17365306247824217,
      "grad_norm": 0.2395952045917511,
      "learning_rate": 0.0004,
      "loss": 4.915,
      "step": 8480
    },
    {
      "epoch": 0.1738578420330514,
      "grad_norm": 0.5889842510223389,
      "learning_rate": 0.0004,
      "loss": 4.9647,
      "step": 8490
    },
    {
      "epoch": 0.17406262158786068,
      "grad_norm": 0.27600687742233276,
      "learning_rate": 0.0004,
      "loss": 4.9693,
      "step": 8500
    },
    {
      "epoch": 0.17426740114266992,
      "grad_norm": 0.28736770153045654,
      "learning_rate": 0.0004,
      "loss": 4.9688,
      "step": 8510
    },
    {
      "epoch": 0.17447218069747916,
      "grad_norm": 0.2757735550403595,
      "learning_rate": 0.0004,
      "loss": 4.9578,
      "step": 8520
    },
    {
      "epoch": 0.1746769602522884,
      "grad_norm": 0.2744831442832947,
      "learning_rate": 0.0004,
      "loss": 4.9767,
      "step": 8530
    },
    {
      "epoch": 0.17488173980709765,
      "grad_norm": 0.33410757780075073,
      "learning_rate": 0.0004,
      "loss": 4.9444,
      "step": 8540
    },
    {
      "epoch": 0.17508651936190692,
      "grad_norm": 0.27433595061302185,
      "learning_rate": 0.0004,
      "loss": 4.9834,
      "step": 8550
    },
    {
      "epoch": 0.17529129891671616,
      "grad_norm": 0.27666985988616943,
      "learning_rate": 0.0004,
      "loss": 4.9854,
      "step": 8560
    },
    {
      "epoch": 0.1754960784715254,
      "grad_norm": 0.2830606997013092,
      "learning_rate": 0.0004,
      "loss": 4.9928,
      "step": 8570
    },
    {
      "epoch": 0.17570085802633464,
      "grad_norm": 0.7617194652557373,
      "learning_rate": 0.0004,
      "loss": 4.9465,
      "step": 8580
    },
    {
      "epoch": 0.1759056375811439,
      "grad_norm": 0.25556811690330505,
      "learning_rate": 0.0004,
      "loss": 4.9076,
      "step": 8590
    },
    {
      "epoch": 0.17611041713595316,
      "grad_norm": 0.2798091769218445,
      "learning_rate": 0.0004,
      "loss": 4.9809,
      "step": 8600
    },
    {
      "epoch": 0.1763151966907624,
      "grad_norm": 0.252435564994812,
      "learning_rate": 0.0004,
      "loss": 4.9489,
      "step": 8610
    },
    {
      "epoch": 0.17651997624557164,
      "grad_norm": 0.2565086781978607,
      "learning_rate": 0.0004,
      "loss": 4.9689,
      "step": 8620
    },
    {
      "epoch": 0.17672475580038088,
      "grad_norm": 0.25653916597366333,
      "learning_rate": 0.0004,
      "loss": 4.9555,
      "step": 8630
    },
    {
      "epoch": 0.17692953535519013,
      "grad_norm": 0.24862580001354218,
      "learning_rate": 0.0004,
      "loss": 4.9307,
      "step": 8640
    },
    {
      "epoch": 0.1771343149099994,
      "grad_norm": 0.26300904154777527,
      "learning_rate": 0.0004,
      "loss": 4.9638,
      "step": 8650
    },
    {
      "epoch": 0.17733909446480864,
      "grad_norm": 0.2595764696598053,
      "learning_rate": 0.0004,
      "loss": 4.9342,
      "step": 8660
    },
    {
      "epoch": 0.17754387401961788,
      "grad_norm": 0.2754005789756775,
      "learning_rate": 0.0004,
      "loss": 4.938,
      "step": 8670
    },
    {
      "epoch": 0.17774865357442712,
      "grad_norm": 0.2422814816236496,
      "learning_rate": 0.0004,
      "loss": 4.9392,
      "step": 8680
    },
    {
      "epoch": 0.17795343312923637,
      "grad_norm": 0.2505927085876465,
      "learning_rate": 0.0004,
      "loss": 4.9132,
      "step": 8690
    },
    {
      "epoch": 0.17815821268404564,
      "grad_norm": 0.3159048855304718,
      "learning_rate": 0.0004,
      "loss": 4.8876,
      "step": 8700
    },
    {
      "epoch": 0.17836299223885488,
      "grad_norm": 0.268097460269928,
      "learning_rate": 0.0004,
      "loss": 4.9008,
      "step": 8710
    },
    {
      "epoch": 0.17856777179366412,
      "grad_norm": 0.25193169713020325,
      "learning_rate": 0.0004,
      "loss": 4.9669,
      "step": 8720
    },
    {
      "epoch": 0.17877255134847336,
      "grad_norm": 0.2685261368751526,
      "learning_rate": 0.0004,
      "loss": 4.9527,
      "step": 8730
    },
    {
      "epoch": 0.1789773309032826,
      "grad_norm": 0.2474873661994934,
      "learning_rate": 0.0004,
      "loss": 4.9591,
      "step": 8740
    },
    {
      "epoch": 0.17918211045809188,
      "grad_norm": 0.2831394374370575,
      "learning_rate": 0.0004,
      "loss": 4.9445,
      "step": 8750
    },
    {
      "epoch": 0.17938689001290112,
      "grad_norm": 0.25589999556541443,
      "learning_rate": 0.0004,
      "loss": 4.9701,
      "step": 8760
    },
    {
      "epoch": 0.17959166956771036,
      "grad_norm": 0.2762419879436493,
      "learning_rate": 0.0004,
      "loss": 4.9525,
      "step": 8770
    },
    {
      "epoch": 0.1797964491225196,
      "grad_norm": 0.3691735565662384,
      "learning_rate": 0.0004,
      "loss": 4.9013,
      "step": 8780
    },
    {
      "epoch": 0.18000122867732885,
      "grad_norm": 0.27381083369255066,
      "learning_rate": 0.0004,
      "loss": 4.929,
      "step": 8790
    },
    {
      "epoch": 0.18020600823213812,
      "grad_norm": 0.2597827911376953,
      "learning_rate": 0.0004,
      "loss": 4.9431,
      "step": 8800
    },
    {
      "epoch": 0.18041078778694736,
      "grad_norm": 0.34209340810775757,
      "learning_rate": 0.0004,
      "loss": 4.9308,
      "step": 8810
    },
    {
      "epoch": 0.1806155673417566,
      "grad_norm": 0.26386550068855286,
      "learning_rate": 0.0004,
      "loss": 4.9484,
      "step": 8820
    },
    {
      "epoch": 0.18082034689656584,
      "grad_norm": 0.25239115953445435,
      "learning_rate": 0.0004,
      "loss": 4.9614,
      "step": 8830
    },
    {
      "epoch": 0.18102512645137508,
      "grad_norm": 0.2644425630569458,
      "learning_rate": 0.0004,
      "loss": 4.9233,
      "step": 8840
    },
    {
      "epoch": 0.18122990600618435,
      "grad_norm": 0.27113065123558044,
      "learning_rate": 0.0004,
      "loss": 4.9295,
      "step": 8850
    },
    {
      "epoch": 0.1814346855609936,
      "grad_norm": 0.24539048969745636,
      "learning_rate": 0.0004,
      "loss": 4.9393,
      "step": 8860
    },
    {
      "epoch": 0.18163946511580284,
      "grad_norm": 0.2639126181602478,
      "learning_rate": 0.0004,
      "loss": 4.9368,
      "step": 8870
    },
    {
      "epoch": 0.18184424467061208,
      "grad_norm": 0.2735327184200287,
      "learning_rate": 0.0004,
      "loss": 4.934,
      "step": 8880
    },
    {
      "epoch": 0.18204902422542132,
      "grad_norm": 0.27094408869743347,
      "learning_rate": 0.0004,
      "loss": 4.9357,
      "step": 8890
    },
    {
      "epoch": 0.1822538037802306,
      "grad_norm": 0.2365713119506836,
      "learning_rate": 0.0004,
      "loss": 4.9588,
      "step": 8900
    },
    {
      "epoch": 0.18245858333503984,
      "grad_norm": 0.24846386909484863,
      "learning_rate": 0.0004,
      "loss": 4.9726,
      "step": 8910
    },
    {
      "epoch": 0.18266336288984908,
      "grad_norm": 0.47635194659233093,
      "learning_rate": 0.0004,
      "loss": 4.9204,
      "step": 8920
    },
    {
      "epoch": 0.18286814244465832,
      "grad_norm": 0.23480106890201569,
      "learning_rate": 0.0004,
      "loss": 4.9501,
      "step": 8930
    },
    {
      "epoch": 0.18307292199946756,
      "grad_norm": 0.2801901698112488,
      "learning_rate": 0.0004,
      "loss": 4.9348,
      "step": 8940
    },
    {
      "epoch": 0.18327770155427683,
      "grad_norm": 0.25215595960617065,
      "learning_rate": 0.0004,
      "loss": 4.9502,
      "step": 8950
    },
    {
      "epoch": 0.18348248110908608,
      "grad_norm": 0.2644833028316498,
      "learning_rate": 0.0004,
      "loss": 4.9308,
      "step": 8960
    },
    {
      "epoch": 0.18368726066389532,
      "grad_norm": 0.27154913544654846,
      "learning_rate": 0.0004,
      "loss": 4.9412,
      "step": 8970
    },
    {
      "epoch": 0.18389204021870456,
      "grad_norm": 0.24921545386314392,
      "learning_rate": 0.0004,
      "loss": 4.9457,
      "step": 8980
    },
    {
      "epoch": 0.1840968197735138,
      "grad_norm": 0.2780075967311859,
      "learning_rate": 0.0004,
      "loss": 4.8964,
      "step": 8990
    },
    {
      "epoch": 0.18430159932832307,
      "grad_norm": 0.24482417106628418,
      "learning_rate": 0.0004,
      "loss": 4.9115,
      "step": 9000
    },
    {
      "epoch": 0.18450637888313232,
      "grad_norm": 0.2918320894241333,
      "learning_rate": 0.0004,
      "loss": 4.9544,
      "step": 9010
    },
    {
      "epoch": 0.18471115843794156,
      "grad_norm": 0.27354565262794495,
      "learning_rate": 0.0004,
      "loss": 4.9285,
      "step": 9020
    },
    {
      "epoch": 0.1849159379927508,
      "grad_norm": 0.2596583962440491,
      "learning_rate": 0.0004,
      "loss": 4.9105,
      "step": 9030
    },
    {
      "epoch": 0.18512071754756004,
      "grad_norm": 0.27231499552726746,
      "learning_rate": 0.0004,
      "loss": 4.923,
      "step": 9040
    },
    {
      "epoch": 0.1853254971023693,
      "grad_norm": 0.2770763039588928,
      "learning_rate": 0.0004,
      "loss": 4.9517,
      "step": 9050
    },
    {
      "epoch": 0.18553027665717856,
      "grad_norm": 0.2580932378768921,
      "learning_rate": 0.0004,
      "loss": 4.9304,
      "step": 9060
    },
    {
      "epoch": 0.1857350562119878,
      "grad_norm": 0.24377305805683136,
      "learning_rate": 0.0004,
      "loss": 4.93,
      "step": 9070
    },
    {
      "epoch": 0.18593983576679704,
      "grad_norm": 0.26057571172714233,
      "learning_rate": 0.0004,
      "loss": 4.9098,
      "step": 9080
    },
    {
      "epoch": 0.18614461532160628,
      "grad_norm": 0.27135148644447327,
      "learning_rate": 0.0004,
      "loss": 4.9042,
      "step": 9090
    },
    {
      "epoch": 0.18634939487641553,
      "grad_norm": 0.2588081359863281,
      "learning_rate": 0.0004,
      "loss": 4.906,
      "step": 9100
    },
    {
      "epoch": 0.1865541744312248,
      "grad_norm": 0.2858315110206604,
      "learning_rate": 0.0004,
      "loss": 4.9154,
      "step": 9110
    },
    {
      "epoch": 0.18675895398603404,
      "grad_norm": 0.26252084970474243,
      "learning_rate": 0.0004,
      "loss": 4.9141,
      "step": 9120
    },
    {
      "epoch": 0.18696373354084328,
      "grad_norm": 0.2764761745929718,
      "learning_rate": 0.0004,
      "loss": 4.8958,
      "step": 9130
    },
    {
      "epoch": 0.18716851309565252,
      "grad_norm": 0.29110220074653625,
      "learning_rate": 0.0004,
      "loss": 4.9123,
      "step": 9140
    },
    {
      "epoch": 0.18737329265046176,
      "grad_norm": 0.2575053870677948,
      "learning_rate": 0.0004,
      "loss": 4.9099,
      "step": 9150
    },
    {
      "epoch": 0.18757807220527103,
      "grad_norm": 0.29032933712005615,
      "learning_rate": 0.0004,
      "loss": 4.9262,
      "step": 9160
    },
    {
      "epoch": 0.18778285176008028,
      "grad_norm": 0.3194615840911865,
      "learning_rate": 0.0004,
      "loss": 4.9179,
      "step": 9170
    },
    {
      "epoch": 0.18798763131488952,
      "grad_norm": 0.26347652077674866,
      "learning_rate": 0.0004,
      "loss": 4.9459,
      "step": 9180
    },
    {
      "epoch": 0.18819241086969876,
      "grad_norm": 0.2674354314804077,
      "learning_rate": 0.0004,
      "loss": 4.9468,
      "step": 9190
    },
    {
      "epoch": 0.188397190424508,
      "grad_norm": 0.2521820068359375,
      "learning_rate": 0.0004,
      "loss": 4.9324,
      "step": 9200
    },
    {
      "epoch": 0.18860196997931727,
      "grad_norm": 0.24796158075332642,
      "learning_rate": 0.0004,
      "loss": 4.9299,
      "step": 9210
    },
    {
      "epoch": 0.18880674953412652,
      "grad_norm": 0.2503640949726105,
      "learning_rate": 0.0004,
      "loss": 4.9388,
      "step": 9220
    },
    {
      "epoch": 0.18901152908893576,
      "grad_norm": 0.2608177363872528,
      "learning_rate": 0.0004,
      "loss": 4.9164,
      "step": 9230
    },
    {
      "epoch": 0.189216308643745,
      "grad_norm": 0.2645958960056305,
      "learning_rate": 0.0004,
      "loss": 4.8984,
      "step": 9240
    },
    {
      "epoch": 0.18942108819855424,
      "grad_norm": 0.26529330015182495,
      "learning_rate": 0.0004,
      "loss": 4.9369,
      "step": 9250
    },
    {
      "epoch": 0.18962586775336351,
      "grad_norm": 0.25383853912353516,
      "learning_rate": 0.0004,
      "loss": 4.9377,
      "step": 9260
    },
    {
      "epoch": 0.18983064730817276,
      "grad_norm": 0.25946077704429626,
      "learning_rate": 0.0004,
      "loss": 4.9082,
      "step": 9270
    },
    {
      "epoch": 0.190035426862982,
      "grad_norm": 0.2694838345050812,
      "learning_rate": 0.0004,
      "loss": 4.8979,
      "step": 9280
    },
    {
      "epoch": 0.19024020641779124,
      "grad_norm": 0.24734584987163544,
      "learning_rate": 0.0004,
      "loss": 4.9369,
      "step": 9290
    },
    {
      "epoch": 0.19044498597260048,
      "grad_norm": 0.9209773540496826,
      "learning_rate": 0.0004,
      "loss": 4.929,
      "step": 9300
    },
    {
      "epoch": 0.19064976552740975,
      "grad_norm": 0.26379844546318054,
      "learning_rate": 0.0004,
      "loss": 4.9336,
      "step": 9310
    },
    {
      "epoch": 0.190854545082219,
      "grad_norm": 0.2513795793056488,
      "learning_rate": 0.0004,
      "loss": 4.9192,
      "step": 9320
    },
    {
      "epoch": 0.19105932463702824,
      "grad_norm": 0.2901628613471985,
      "learning_rate": 0.0004,
      "loss": 4.9422,
      "step": 9330
    },
    {
      "epoch": 0.19126410419183748,
      "grad_norm": 0.25602996349334717,
      "learning_rate": 0.0004,
      "loss": 4.8811,
      "step": 9340
    },
    {
      "epoch": 0.19146888374664672,
      "grad_norm": 0.2577371895313263,
      "learning_rate": 0.0004,
      "loss": 4.9184,
      "step": 9350
    },
    {
      "epoch": 0.191673663301456,
      "grad_norm": 0.2465871423482895,
      "learning_rate": 0.0004,
      "loss": 4.9836,
      "step": 9360
    },
    {
      "epoch": 0.19187844285626524,
      "grad_norm": 0.2691284120082855,
      "learning_rate": 0.0004,
      "loss": 4.9593,
      "step": 9370
    },
    {
      "epoch": 0.19208322241107448,
      "grad_norm": 0.28360679745674133,
      "learning_rate": 0.0004,
      "loss": 4.9323,
      "step": 9380
    },
    {
      "epoch": 0.19228800196588372,
      "grad_norm": 0.2639329135417938,
      "learning_rate": 0.0004,
      "loss": 4.8867,
      "step": 9390
    },
    {
      "epoch": 0.19249278152069296,
      "grad_norm": 0.26145270466804504,
      "learning_rate": 0.0004,
      "loss": 4.932,
      "step": 9400
    },
    {
      "epoch": 0.19269756107550223,
      "grad_norm": 0.24542520940303802,
      "learning_rate": 0.0004,
      "loss": 4.9079,
      "step": 9410
    },
    {
      "epoch": 0.19290234063031148,
      "grad_norm": 0.3000343441963196,
      "learning_rate": 0.0004,
      "loss": 4.9274,
      "step": 9420
    },
    {
      "epoch": 0.19310712018512072,
      "grad_norm": 0.2719687521457672,
      "learning_rate": 0.0004,
      "loss": 4.8972,
      "step": 9430
    },
    {
      "epoch": 0.19331189973992996,
      "grad_norm": 0.2948462665081024,
      "learning_rate": 0.0004,
      "loss": 4.9418,
      "step": 9440
    },
    {
      "epoch": 0.1935166792947392,
      "grad_norm": 0.266860693693161,
      "learning_rate": 0.0004,
      "loss": 4.9422,
      "step": 9450
    },
    {
      "epoch": 0.19372145884954847,
      "grad_norm": 0.2978024184703827,
      "learning_rate": 0.0004,
      "loss": 4.8979,
      "step": 9460
    },
    {
      "epoch": 0.19392623840435771,
      "grad_norm": 0.24917840957641602,
      "learning_rate": 0.0004,
      "loss": 4.9035,
      "step": 9470
    },
    {
      "epoch": 0.19413101795916696,
      "grad_norm": 0.2922961115837097,
      "learning_rate": 0.0004,
      "loss": 4.9164,
      "step": 9480
    },
    {
      "epoch": 0.1943357975139762,
      "grad_norm": 0.29190903902053833,
      "learning_rate": 0.0004,
      "loss": 4.8901,
      "step": 9490
    },
    {
      "epoch": 0.19454057706878544,
      "grad_norm": 0.24561089277267456,
      "learning_rate": 0.0004,
      "loss": 4.934,
      "step": 9500
    },
    {
      "epoch": 0.1947453566235947,
      "grad_norm": 0.2607162296772003,
      "learning_rate": 0.0004,
      "loss": 4.902,
      "step": 9510
    },
    {
      "epoch": 0.19495013617840395,
      "grad_norm": 0.26801997423171997,
      "learning_rate": 0.0004,
      "loss": 4.9062,
      "step": 9520
    },
    {
      "epoch": 0.1951549157332132,
      "grad_norm": 0.2481057345867157,
      "learning_rate": 0.0004,
      "loss": 4.9288,
      "step": 9530
    },
    {
      "epoch": 0.19535969528802244,
      "grad_norm": 0.31173771619796753,
      "learning_rate": 0.0004,
      "loss": 4.9154,
      "step": 9540
    },
    {
      "epoch": 0.19556447484283168,
      "grad_norm": 0.2514681816101074,
      "learning_rate": 0.0004,
      "loss": 4.9363,
      "step": 9550
    },
    {
      "epoch": 0.19576925439764095,
      "grad_norm": 0.2622108459472656,
      "learning_rate": 0.0004,
      "loss": 4.9236,
      "step": 9560
    },
    {
      "epoch": 0.1959740339524502,
      "grad_norm": 0.25495150685310364,
      "learning_rate": 0.0004,
      "loss": 4.908,
      "step": 9570
    },
    {
      "epoch": 0.19617881350725944,
      "grad_norm": 0.26596617698669434,
      "learning_rate": 0.0004,
      "loss": 4.8762,
      "step": 9580
    },
    {
      "epoch": 0.19638359306206868,
      "grad_norm": 0.2579491436481476,
      "learning_rate": 0.0004,
      "loss": 4.9194,
      "step": 9590
    },
    {
      "epoch": 0.19658837261687792,
      "grad_norm": 0.2557850480079651,
      "learning_rate": 0.0004,
      "loss": 4.9023,
      "step": 9600
    },
    {
      "epoch": 0.1967931521716872,
      "grad_norm": 0.24434374272823334,
      "learning_rate": 0.0004,
      "loss": 4.9162,
      "step": 9610
    },
    {
      "epoch": 0.19699793172649643,
      "grad_norm": 0.25301647186279297,
      "learning_rate": 0.0004,
      "loss": 4.8545,
      "step": 9620
    },
    {
      "epoch": 0.19720271128130568,
      "grad_norm": 0.2586268186569214,
      "learning_rate": 0.0004,
      "loss": 4.9286,
      "step": 9630
    },
    {
      "epoch": 0.19740749083611492,
      "grad_norm": 0.2630811631679535,
      "learning_rate": 0.0004,
      "loss": 4.9074,
      "step": 9640
    },
    {
      "epoch": 0.19761227039092416,
      "grad_norm": 0.28417789936065674,
      "learning_rate": 0.0004,
      "loss": 4.9275,
      "step": 9650
    },
    {
      "epoch": 0.19781704994573343,
      "grad_norm": 0.2516314685344696,
      "learning_rate": 0.0004,
      "loss": 4.9021,
      "step": 9660
    },
    {
      "epoch": 0.19802182950054267,
      "grad_norm": 0.2682010531425476,
      "learning_rate": 0.0004,
      "loss": 4.9054,
      "step": 9670
    },
    {
      "epoch": 0.19822660905535192,
      "grad_norm": 0.25043565034866333,
      "learning_rate": 0.0004,
      "loss": 4.8953,
      "step": 9680
    },
    {
      "epoch": 0.19843138861016116,
      "grad_norm": 0.2632971704006195,
      "learning_rate": 0.0004,
      "loss": 4.9153,
      "step": 9690
    },
    {
      "epoch": 0.1986361681649704,
      "grad_norm": 0.26531434059143066,
      "learning_rate": 0.0004,
      "loss": 4.8796,
      "step": 9700
    },
    {
      "epoch": 0.19884094771977967,
      "grad_norm": 0.26835283637046814,
      "learning_rate": 0.0004,
      "loss": 4.9055,
      "step": 9710
    },
    {
      "epoch": 0.1990457272745889,
      "grad_norm": 0.26542049646377563,
      "learning_rate": 0.0004,
      "loss": 4.9236,
      "step": 9720
    },
    {
      "epoch": 0.19925050682939816,
      "grad_norm": 0.28944799304008484,
      "learning_rate": 0.0004,
      "loss": 4.8907,
      "step": 9730
    },
    {
      "epoch": 0.1994552863842074,
      "grad_norm": 0.2508513629436493,
      "learning_rate": 0.0004,
      "loss": 4.9092,
      "step": 9740
    },
    {
      "epoch": 0.19966006593901664,
      "grad_norm": 0.24191100895404816,
      "learning_rate": 0.0004,
      "loss": 4.8827,
      "step": 9750
    },
    {
      "epoch": 0.1998648454938259,
      "grad_norm": 0.26619407534599304,
      "learning_rate": 0.0004,
      "loss": 4.8478,
      "step": 9760
    },
    {
      "epoch": 0.20006962504863515,
      "grad_norm": 0.2426200807094574,
      "learning_rate": 0.0004,
      "loss": 4.9073,
      "step": 9770
    },
    {
      "epoch": 0.2002744046034444,
      "grad_norm": 0.26734215021133423,
      "learning_rate": 0.0004,
      "loss": 4.895,
      "step": 9780
    },
    {
      "epoch": 0.20047918415825364,
      "grad_norm": 0.2878583073616028,
      "learning_rate": 0.0004,
      "loss": 4.9005,
      "step": 9790
    },
    {
      "epoch": 0.20068396371306288,
      "grad_norm": 0.23906348645687103,
      "learning_rate": 0.0004,
      "loss": 4.9572,
      "step": 9800
    },
    {
      "epoch": 0.20088874326787212,
      "grad_norm": 0.28374752402305603,
      "learning_rate": 0.0004,
      "loss": 4.916,
      "step": 9810
    },
    {
      "epoch": 0.2010935228226814,
      "grad_norm": 0.2827334403991699,
      "learning_rate": 0.0004,
      "loss": 4.9005,
      "step": 9820
    },
    {
      "epoch": 0.20129830237749063,
      "grad_norm": 0.3524470925331116,
      "learning_rate": 0.0004,
      "loss": 4.8969,
      "step": 9830
    },
    {
      "epoch": 0.20150308193229988,
      "grad_norm": 0.2859930992126465,
      "learning_rate": 0.0004,
      "loss": 4.9207,
      "step": 9840
    },
    {
      "epoch": 0.20170786148710912,
      "grad_norm": 0.2604099214076996,
      "learning_rate": 0.0004,
      "loss": 4.8868,
      "step": 9850
    },
    {
      "epoch": 0.20191264104191836,
      "grad_norm": 0.24747997522354126,
      "learning_rate": 0.0004,
      "loss": 4.9379,
      "step": 9860
    },
    {
      "epoch": 0.20211742059672763,
      "grad_norm": 0.27505552768707275,
      "learning_rate": 0.0004,
      "loss": 4.8898,
      "step": 9870
    },
    {
      "epoch": 0.20232220015153687,
      "grad_norm": 0.25134706497192383,
      "learning_rate": 0.0004,
      "loss": 4.9083,
      "step": 9880
    },
    {
      "epoch": 0.20252697970634612,
      "grad_norm": 0.2687472999095917,
      "learning_rate": 0.0004,
      "loss": 4.9199,
      "step": 9890
    },
    {
      "epoch": 0.20273175926115536,
      "grad_norm": 0.2637028992176056,
      "learning_rate": 0.0004,
      "loss": 4.9064,
      "step": 9900
    },
    {
      "epoch": 0.2029365388159646,
      "grad_norm": 0.34653931856155396,
      "learning_rate": 0.0004,
      "loss": 4.9107,
      "step": 9910
    },
    {
      "epoch": 0.20314131837077387,
      "grad_norm": 0.2523519992828369,
      "learning_rate": 0.0004,
      "loss": 4.9172,
      "step": 9920
    },
    {
      "epoch": 0.2033460979255831,
      "grad_norm": 0.2544276714324951,
      "learning_rate": 0.0004,
      "loss": 4.928,
      "step": 9930
    },
    {
      "epoch": 0.20355087748039236,
      "grad_norm": 0.2527393698692322,
      "learning_rate": 0.0004,
      "loss": 4.9027,
      "step": 9940
    },
    {
      "epoch": 0.2037556570352016,
      "grad_norm": 0.2704786956310272,
      "learning_rate": 0.0004,
      "loss": 4.8868,
      "step": 9950
    },
    {
      "epoch": 0.20396043659001084,
      "grad_norm": 0.24265450239181519,
      "learning_rate": 0.0004,
      "loss": 4.9287,
      "step": 9960
    },
    {
      "epoch": 0.2041652161448201,
      "grad_norm": 0.26238685846328735,
      "learning_rate": 0.0004,
      "loss": 4.882,
      "step": 9970
    },
    {
      "epoch": 0.20436999569962935,
      "grad_norm": 0.27310532331466675,
      "learning_rate": 0.0004,
      "loss": 4.8909,
      "step": 9980
    },
    {
      "epoch": 0.2045747752544386,
      "grad_norm": 0.2510082721710205,
      "learning_rate": 0.0004,
      "loss": 4.8839,
      "step": 9990
    },
    {
      "epoch": 0.20477955480924784,
      "grad_norm": 0.2608467638492584,
      "learning_rate": 0.0004,
      "loss": 4.9269,
      "step": 10000
    },
    {
      "epoch": 0.20477955480924784,
      "eval_loss": 4.915364742279053,
      "eval_runtime": 4.3939,
      "eval_samples_per_second": 265.37,
      "eval_steps_per_second": 33.228,
      "step": 10000
    },
    {
      "epoch": 0.20498433436405708,
      "grad_norm": 0.29385286569595337,
      "learning_rate": 0.0004,
      "loss": 4.8996,
      "step": 10010
    },
    {
      "epoch": 0.20518911391886635,
      "grad_norm": 0.26717716455459595,
      "learning_rate": 0.0004,
      "loss": 4.9464,
      "step": 10020
    },
    {
      "epoch": 0.2053938934736756,
      "grad_norm": 0.24872258305549622,
      "learning_rate": 0.0004,
      "loss": 4.927,
      "step": 10030
    },
    {
      "epoch": 0.20559867302848484,
      "grad_norm": 0.3139999508857727,
      "learning_rate": 0.0004,
      "loss": 4.9129,
      "step": 10040
    },
    {
      "epoch": 0.20580345258329408,
      "grad_norm": 0.2630196511745453,
      "learning_rate": 0.0004,
      "loss": 4.9052,
      "step": 10050
    },
    {
      "epoch": 0.20600823213810332,
      "grad_norm": 0.2669818699359894,
      "learning_rate": 0.0004,
      "loss": 4.9042,
      "step": 10060
    },
    {
      "epoch": 0.2062130116929126,
      "grad_norm": 0.23856590688228607,
      "learning_rate": 0.0004,
      "loss": 4.9258,
      "step": 10070
    },
    {
      "epoch": 0.20641779124772183,
      "grad_norm": 0.269837886095047,
      "learning_rate": 0.0004,
      "loss": 4.8969,
      "step": 10080
    },
    {
      "epoch": 0.20662257080253107,
      "grad_norm": 0.2798609733581543,
      "learning_rate": 0.0004,
      "loss": 4.9189,
      "step": 10090
    },
    {
      "epoch": 0.20682735035734032,
      "grad_norm": 0.2599700391292572,
      "learning_rate": 0.0004,
      "loss": 4.9153,
      "step": 10100
    },
    {
      "epoch": 0.20703212991214956,
      "grad_norm": 0.24514122307300568,
      "learning_rate": 0.0004,
      "loss": 4.9016,
      "step": 10110
    },
    {
      "epoch": 0.20723690946695883,
      "grad_norm": 0.25042402744293213,
      "learning_rate": 0.0004,
      "loss": 4.9002,
      "step": 10120
    },
    {
      "epoch": 0.20744168902176807,
      "grad_norm": 0.32178834080696106,
      "learning_rate": 0.0004,
      "loss": 4.8801,
      "step": 10130
    },
    {
      "epoch": 0.20764646857657731,
      "grad_norm": 0.2308470606803894,
      "learning_rate": 0.0004,
      "loss": 4.8786,
      "step": 10140
    },
    {
      "epoch": 0.20785124813138656,
      "grad_norm": 0.24936145544052124,
      "learning_rate": 0.0004,
      "loss": 4.8801,
      "step": 10150
    },
    {
      "epoch": 0.2080560276861958,
      "grad_norm": 0.26884016394615173,
      "learning_rate": 0.0004,
      "loss": 4.8996,
      "step": 10160
    },
    {
      "epoch": 0.20826080724100507,
      "grad_norm": 0.24865688383579254,
      "learning_rate": 0.0004,
      "loss": 4.9331,
      "step": 10170
    },
    {
      "epoch": 0.2084655867958143,
      "grad_norm": 0.23579901456832886,
      "learning_rate": 0.0004,
      "loss": 4.8948,
      "step": 10180
    },
    {
      "epoch": 0.20867036635062355,
      "grad_norm": 0.2458990514278412,
      "learning_rate": 0.0004,
      "loss": 4.8782,
      "step": 10190
    },
    {
      "epoch": 0.2088751459054328,
      "grad_norm": 0.2518019378185272,
      "learning_rate": 0.0004,
      "loss": 4.9196,
      "step": 10200
    },
    {
      "epoch": 0.20907992546024204,
      "grad_norm": 0.25939616560935974,
      "learning_rate": 0.0004,
      "loss": 4.9116,
      "step": 10210
    },
    {
      "epoch": 0.2092847050150513,
      "grad_norm": 0.31639620661735535,
      "learning_rate": 0.0004,
      "loss": 4.8841,
      "step": 10220
    },
    {
      "epoch": 0.20948948456986055,
      "grad_norm": 0.27148041129112244,
      "learning_rate": 0.0004,
      "loss": 4.8721,
      "step": 10230
    },
    {
      "epoch": 0.2096942641246698,
      "grad_norm": 0.23913012444972992,
      "learning_rate": 0.0004,
      "loss": 4.8817,
      "step": 10240
    },
    {
      "epoch": 0.20989904367947904,
      "grad_norm": 0.27129408717155457,
      "learning_rate": 0.0004,
      "loss": 4.8802,
      "step": 10250
    },
    {
      "epoch": 0.21010382323428828,
      "grad_norm": 0.26640304923057556,
      "learning_rate": 0.0004,
      "loss": 4.8998,
      "step": 10260
    },
    {
      "epoch": 0.21030860278909755,
      "grad_norm": 0.8138028979301453,
      "learning_rate": 0.0004,
      "loss": 4.9031,
      "step": 10270
    },
    {
      "epoch": 0.2105133823439068,
      "grad_norm": 0.26214420795440674,
      "learning_rate": 0.0004,
      "loss": 4.9146,
      "step": 10280
    },
    {
      "epoch": 0.21071816189871603,
      "grad_norm": 0.2588198184967041,
      "learning_rate": 0.0004,
      "loss": 4.8749,
      "step": 10290
    },
    {
      "epoch": 0.21092294145352528,
      "grad_norm": 0.26211073994636536,
      "learning_rate": 0.0004,
      "loss": 4.8954,
      "step": 10300
    },
    {
      "epoch": 0.21112772100833452,
      "grad_norm": 0.254951149225235,
      "learning_rate": 0.0004,
      "loss": 4.8531,
      "step": 10310
    },
    {
      "epoch": 0.2113325005631438,
      "grad_norm": 0.27760276198387146,
      "learning_rate": 0.0004,
      "loss": 4.8912,
      "step": 10320
    },
    {
      "epoch": 0.21153728011795303,
      "grad_norm": 0.27546319365501404,
      "learning_rate": 0.0004,
      "loss": 4.9153,
      "step": 10330
    },
    {
      "epoch": 0.21174205967276227,
      "grad_norm": 0.248246431350708,
      "learning_rate": 0.0004,
      "loss": 4.9059,
      "step": 10340
    },
    {
      "epoch": 0.21194683922757152,
      "grad_norm": 0.2779110372066498,
      "learning_rate": 0.0004,
      "loss": 4.8846,
      "step": 10350
    },
    {
      "epoch": 0.21215161878238076,
      "grad_norm": 0.25909432768821716,
      "learning_rate": 0.0004,
      "loss": 4.9028,
      "step": 10360
    },
    {
      "epoch": 0.21235639833719003,
      "grad_norm": 0.2711811661720276,
      "learning_rate": 0.0004,
      "loss": 4.903,
      "step": 10370
    },
    {
      "epoch": 0.21256117789199927,
      "grad_norm": 0.262154221534729,
      "learning_rate": 0.0004,
      "loss": 4.8841,
      "step": 10380
    },
    {
      "epoch": 0.2127659574468085,
      "grad_norm": 0.2557555139064789,
      "learning_rate": 0.0004,
      "loss": 4.9125,
      "step": 10390
    },
    {
      "epoch": 0.21297073700161775,
      "grad_norm": 0.3354072868824005,
      "learning_rate": 0.0004,
      "loss": 4.8713,
      "step": 10400
    },
    {
      "epoch": 0.213175516556427,
      "grad_norm": 0.26146626472473145,
      "learning_rate": 0.0004,
      "loss": 4.8713,
      "step": 10410
    },
    {
      "epoch": 0.21338029611123627,
      "grad_norm": 0.2696409523487091,
      "learning_rate": 0.0004,
      "loss": 4.8715,
      "step": 10420
    },
    {
      "epoch": 0.2135850756660455,
      "grad_norm": 0.29851871728897095,
      "learning_rate": 0.0004,
      "loss": 4.8633,
      "step": 10430
    },
    {
      "epoch": 0.21378985522085475,
      "grad_norm": 0.4111309051513672,
      "learning_rate": 0.0004,
      "loss": 4.8891,
      "step": 10440
    },
    {
      "epoch": 0.213994634775664,
      "grad_norm": 0.2570556700229645,
      "learning_rate": 0.0004,
      "loss": 4.8538,
      "step": 10450
    },
    {
      "epoch": 0.21419941433047324,
      "grad_norm": 0.27073532342910767,
      "learning_rate": 0.0004,
      "loss": 4.89,
      "step": 10460
    },
    {
      "epoch": 0.2144041938852825,
      "grad_norm": 0.2688610851764679,
      "learning_rate": 0.0004,
      "loss": 4.8796,
      "step": 10470
    },
    {
      "epoch": 0.21460897344009175,
      "grad_norm": 0.27269771695137024,
      "learning_rate": 0.0004,
      "loss": 4.8793,
      "step": 10480
    },
    {
      "epoch": 0.214813752994901,
      "grad_norm": 0.2495245784521103,
      "learning_rate": 0.0004,
      "loss": 4.8698,
      "step": 10490
    },
    {
      "epoch": 0.21501853254971023,
      "grad_norm": 0.2419559359550476,
      "learning_rate": 0.0004,
      "loss": 4.8997,
      "step": 10500
    },
    {
      "epoch": 0.21522331210451948,
      "grad_norm": 0.2923683226108551,
      "learning_rate": 0.0004,
      "loss": 4.8748,
      "step": 10510
    },
    {
      "epoch": 0.21542809165932875,
      "grad_norm": 0.2532537579536438,
      "learning_rate": 0.0004,
      "loss": 4.901,
      "step": 10520
    },
    {
      "epoch": 0.215632871214138,
      "grad_norm": 0.2619161605834961,
      "learning_rate": 0.0004,
      "loss": 4.8878,
      "step": 10530
    },
    {
      "epoch": 0.21583765076894723,
      "grad_norm": 0.2929726839065552,
      "learning_rate": 0.0004,
      "loss": 4.9189,
      "step": 10540
    },
    {
      "epoch": 0.21604243032375647,
      "grad_norm": 0.26524537801742554,
      "learning_rate": 0.0004,
      "loss": 4.895,
      "step": 10550
    },
    {
      "epoch": 0.21624720987856572,
      "grad_norm": 0.2544373571872711,
      "learning_rate": 0.0004,
      "loss": 4.8792,
      "step": 10560
    },
    {
      "epoch": 0.21645198943337496,
      "grad_norm": 0.25470906496047974,
      "learning_rate": 0.0004,
      "loss": 4.8639,
      "step": 10570
    },
    {
      "epoch": 0.21665676898818423,
      "grad_norm": 0.27853649854660034,
      "learning_rate": 0.0004,
      "loss": 4.9067,
      "step": 10580
    },
    {
      "epoch": 0.21686154854299347,
      "grad_norm": 0.26324573159217834,
      "learning_rate": 0.0004,
      "loss": 4.8631,
      "step": 10590
    },
    {
      "epoch": 0.2170663280978027,
      "grad_norm": 0.2501038908958435,
      "learning_rate": 0.0004,
      "loss": 4.9081,
      "step": 10600
    },
    {
      "epoch": 0.21727110765261196,
      "grad_norm": 0.25858011841773987,
      "learning_rate": 0.0004,
      "loss": 4.8712,
      "step": 10610
    },
    {
      "epoch": 0.2174758872074212,
      "grad_norm": 0.30494385957717896,
      "learning_rate": 0.0004,
      "loss": 4.896,
      "step": 10620
    },
    {
      "epoch": 0.21768066676223047,
      "grad_norm": 0.2731782793998718,
      "learning_rate": 0.0004,
      "loss": 4.8612,
      "step": 10630
    },
    {
      "epoch": 0.2178854463170397,
      "grad_norm": 0.27870678901672363,
      "learning_rate": 0.0004,
      "loss": 4.9169,
      "step": 10640
    },
    {
      "epoch": 0.21809022587184895,
      "grad_norm": 0.2358781397342682,
      "learning_rate": 0.0004,
      "loss": 4.8979,
      "step": 10650
    },
    {
      "epoch": 0.2182950054266582,
      "grad_norm": 0.27755194902420044,
      "learning_rate": 0.0004,
      "loss": 4.8713,
      "step": 10660
    },
    {
      "epoch": 0.21849978498146744,
      "grad_norm": 0.24838866293430328,
      "learning_rate": 0.0004,
      "loss": 4.8469,
      "step": 10670
    },
    {
      "epoch": 0.2187045645362767,
      "grad_norm": 0.46490806341171265,
      "learning_rate": 0.0004,
      "loss": 4.8521,
      "step": 10680
    },
    {
      "epoch": 0.21890934409108595,
      "grad_norm": 0.29437142610549927,
      "learning_rate": 0.0004,
      "loss": 4.867,
      "step": 10690
    },
    {
      "epoch": 0.2191141236458952,
      "grad_norm": 0.31970828771591187,
      "learning_rate": 0.0004,
      "loss": 4.8879,
      "step": 10700
    },
    {
      "epoch": 0.21931890320070443,
      "grad_norm": 0.25731003284454346,
      "learning_rate": 0.0004,
      "loss": 4.8562,
      "step": 10710
    },
    {
      "epoch": 0.21952368275551368,
      "grad_norm": 0.2647119462490082,
      "learning_rate": 0.0004,
      "loss": 4.8804,
      "step": 10720
    },
    {
      "epoch": 0.21972846231032295,
      "grad_norm": 0.3134158253669739,
      "learning_rate": 0.0004,
      "loss": 4.8631,
      "step": 10730
    },
    {
      "epoch": 0.2199332418651322,
      "grad_norm": 0.2509680986404419,
      "learning_rate": 0.0004,
      "loss": 4.8757,
      "step": 10740
    },
    {
      "epoch": 0.22013802141994143,
      "grad_norm": 0.26775500178337097,
      "learning_rate": 0.0004,
      "loss": 4.8928,
      "step": 10750
    },
    {
      "epoch": 0.22034280097475067,
      "grad_norm": 0.2678174674510956,
      "learning_rate": 0.0004,
      "loss": 4.844,
      "step": 10760
    },
    {
      "epoch": 0.22054758052955992,
      "grad_norm": 0.24718265235424042,
      "learning_rate": 0.0004,
      "loss": 4.884,
      "step": 10770
    },
    {
      "epoch": 0.2207523600843692,
      "grad_norm": 0.2549584209918976,
      "learning_rate": 0.0004,
      "loss": 4.8441,
      "step": 10780
    },
    {
      "epoch": 0.22095713963917843,
      "grad_norm": 0.2551555931568146,
      "learning_rate": 0.0004,
      "loss": 4.8953,
      "step": 10790
    },
    {
      "epoch": 0.22116191919398767,
      "grad_norm": 0.25722163915634155,
      "learning_rate": 0.0004,
      "loss": 4.872,
      "step": 10800
    },
    {
      "epoch": 0.22136669874879691,
      "grad_norm": 0.25247806310653687,
      "learning_rate": 0.0004,
      "loss": 4.8753,
      "step": 10810
    },
    {
      "epoch": 0.22157147830360616,
      "grad_norm": 0.2747890055179596,
      "learning_rate": 0.0004,
      "loss": 4.8829,
      "step": 10820
    },
    {
      "epoch": 0.22177625785841543,
      "grad_norm": 0.27008911967277527,
      "learning_rate": 0.0004,
      "loss": 4.8588,
      "step": 10830
    },
    {
      "epoch": 0.22198103741322467,
      "grad_norm": 0.302053302526474,
      "learning_rate": 0.0004,
      "loss": 4.8748,
      "step": 10840
    },
    {
      "epoch": 0.2221858169680339,
      "grad_norm": 0.26521286368370056,
      "learning_rate": 0.0004,
      "loss": 4.9043,
      "step": 10850
    },
    {
      "epoch": 0.22239059652284315,
      "grad_norm": 0.24146080017089844,
      "learning_rate": 0.0004,
      "loss": 4.901,
      "step": 10860
    },
    {
      "epoch": 0.2225953760776524,
      "grad_norm": 0.24695172905921936,
      "learning_rate": 0.0004,
      "loss": 4.8892,
      "step": 10870
    },
    {
      "epoch": 0.22280015563246167,
      "grad_norm": 0.2761094868183136,
      "learning_rate": 0.0004,
      "loss": 4.8787,
      "step": 10880
    },
    {
      "epoch": 0.2230049351872709,
      "grad_norm": 0.25380679965019226,
      "learning_rate": 0.0004,
      "loss": 4.866,
      "step": 10890
    },
    {
      "epoch": 0.22320971474208015,
      "grad_norm": 0.2534860372543335,
      "learning_rate": 0.0004,
      "loss": 4.8636,
      "step": 10900
    },
    {
      "epoch": 0.2234144942968894,
      "grad_norm": 0.26680171489715576,
      "learning_rate": 0.0004,
      "loss": 4.8631,
      "step": 10910
    },
    {
      "epoch": 0.22361927385169864,
      "grad_norm": 0.25146228075027466,
      "learning_rate": 0.0004,
      "loss": 4.8753,
      "step": 10920
    },
    {
      "epoch": 0.2238240534065079,
      "grad_norm": 0.2974937856197357,
      "learning_rate": 0.0004,
      "loss": 4.8866,
      "step": 10930
    },
    {
      "epoch": 0.22402883296131715,
      "grad_norm": 0.2717348337173462,
      "learning_rate": 0.0004,
      "loss": 4.8891,
      "step": 10940
    },
    {
      "epoch": 0.2242336125161264,
      "grad_norm": 0.2560259699821472,
      "learning_rate": 0.0004,
      "loss": 4.8772,
      "step": 10950
    },
    {
      "epoch": 0.22443839207093563,
      "grad_norm": 0.27203595638275146,
      "learning_rate": 0.0004,
      "loss": 4.8685,
      "step": 10960
    },
    {
      "epoch": 0.22464317162574488,
      "grad_norm": 0.26731157302856445,
      "learning_rate": 0.0004,
      "loss": 4.8656,
      "step": 10970
    },
    {
      "epoch": 0.22484795118055415,
      "grad_norm": 0.2695630192756653,
      "learning_rate": 0.0004,
      "loss": 4.8945,
      "step": 10980
    },
    {
      "epoch": 0.2250527307353634,
      "grad_norm": 0.28685614466667175,
      "learning_rate": 0.0004,
      "loss": 4.8796,
      "step": 10990
    },
    {
      "epoch": 0.22525751029017263,
      "grad_norm": 0.26128077507019043,
      "learning_rate": 0.0004,
      "loss": 4.8915,
      "step": 11000
    },
    {
      "epoch": 0.22546228984498187,
      "grad_norm": 0.24580059945583344,
      "learning_rate": 0.0004,
      "loss": 4.8352,
      "step": 11010
    },
    {
      "epoch": 0.22566706939979111,
      "grad_norm": 0.2452968955039978,
      "learning_rate": 0.0004,
      "loss": 4.9106,
      "step": 11020
    },
    {
      "epoch": 0.22587184895460038,
      "grad_norm": 0.26197394728660583,
      "learning_rate": 0.0004,
      "loss": 4.8995,
      "step": 11030
    },
    {
      "epoch": 0.22607662850940963,
      "grad_norm": 0.3471392095088959,
      "learning_rate": 0.0004,
      "loss": 4.8674,
      "step": 11040
    },
    {
      "epoch": 0.22628140806421887,
      "grad_norm": 0.37223684787750244,
      "learning_rate": 0.0004,
      "loss": 4.8858,
      "step": 11050
    },
    {
      "epoch": 0.2264861876190281,
      "grad_norm": 0.2760046124458313,
      "learning_rate": 0.0004,
      "loss": 4.8761,
      "step": 11060
    },
    {
      "epoch": 0.22669096717383735,
      "grad_norm": 0.25126567482948303,
      "learning_rate": 0.0004,
      "loss": 4.8696,
      "step": 11070
    },
    {
      "epoch": 0.22689574672864662,
      "grad_norm": 0.2638050317764282,
      "learning_rate": 0.0004,
      "loss": 4.8597,
      "step": 11080
    },
    {
      "epoch": 0.22710052628345587,
      "grad_norm": 0.24045559763908386,
      "learning_rate": 0.0004,
      "loss": 4.8482,
      "step": 11090
    },
    {
      "epoch": 0.2273053058382651,
      "grad_norm": 0.3191254734992981,
      "learning_rate": 0.0004,
      "loss": 4.8928,
      "step": 11100
    },
    {
      "epoch": 0.22751008539307435,
      "grad_norm": 0.2639184892177582,
      "learning_rate": 0.0004,
      "loss": 4.8888,
      "step": 11110
    },
    {
      "epoch": 0.2277148649478836,
      "grad_norm": 0.2678391635417938,
      "learning_rate": 0.0004,
      "loss": 4.8455,
      "step": 11120
    },
    {
      "epoch": 0.22791964450269286,
      "grad_norm": 0.2614695131778717,
      "learning_rate": 0.0004,
      "loss": 4.8596,
      "step": 11130
    },
    {
      "epoch": 0.2281244240575021,
      "grad_norm": 0.24117058515548706,
      "learning_rate": 0.0004,
      "loss": 4.8544,
      "step": 11140
    },
    {
      "epoch": 0.22832920361231135,
      "grad_norm": 0.27370381355285645,
      "learning_rate": 0.0004,
      "loss": 4.9183,
      "step": 11150
    },
    {
      "epoch": 0.2285339831671206,
      "grad_norm": 0.2558698356151581,
      "learning_rate": 0.0004,
      "loss": 4.8651,
      "step": 11160
    },
    {
      "epoch": 0.22873876272192983,
      "grad_norm": 0.24600282311439514,
      "learning_rate": 0.0004,
      "loss": 4.8372,
      "step": 11170
    },
    {
      "epoch": 0.2289435422767391,
      "grad_norm": 0.3030242323875427,
      "learning_rate": 0.0004,
      "loss": 4.8691,
      "step": 11180
    },
    {
      "epoch": 0.22914832183154835,
      "grad_norm": 0.24733074009418488,
      "learning_rate": 0.0004,
      "loss": 4.8805,
      "step": 11190
    },
    {
      "epoch": 0.2293531013863576,
      "grad_norm": 0.27617985010147095,
      "learning_rate": 0.0004,
      "loss": 4.8768,
      "step": 11200
    },
    {
      "epoch": 0.22955788094116683,
      "grad_norm": 0.2529560327529907,
      "learning_rate": 0.0004,
      "loss": 4.8199,
      "step": 11210
    },
    {
      "epoch": 0.22976266049597607,
      "grad_norm": 0.239194855093956,
      "learning_rate": 0.0004,
      "loss": 4.8392,
      "step": 11220
    },
    {
      "epoch": 0.22996744005078534,
      "grad_norm": 0.29008886218070984,
      "learning_rate": 0.0004,
      "loss": 4.8582,
      "step": 11230
    },
    {
      "epoch": 0.23017221960559459,
      "grad_norm": 0.2568531036376953,
      "learning_rate": 0.0004,
      "loss": 4.8622,
      "step": 11240
    },
    {
      "epoch": 0.23037699916040383,
      "grad_norm": 0.25861501693725586,
      "learning_rate": 0.0004,
      "loss": 4.8598,
      "step": 11250
    },
    {
      "epoch": 0.23058177871521307,
      "grad_norm": 0.280758798122406,
      "learning_rate": 0.0004,
      "loss": 4.8872,
      "step": 11260
    },
    {
      "epoch": 0.2307865582700223,
      "grad_norm": 0.2618058919906616,
      "learning_rate": 0.0004,
      "loss": 4.8539,
      "step": 11270
    },
    {
      "epoch": 0.23099133782483156,
      "grad_norm": 0.2842026650905609,
      "learning_rate": 0.0004,
      "loss": 4.8277,
      "step": 11280
    },
    {
      "epoch": 0.23119611737964083,
      "grad_norm": 0.25333431363105774,
      "learning_rate": 0.0004,
      "loss": 4.8711,
      "step": 11290
    },
    {
      "epoch": 0.23140089693445007,
      "grad_norm": 0.22926181554794312,
      "learning_rate": 0.0004,
      "loss": 4.8375,
      "step": 11300
    },
    {
      "epoch": 0.2316056764892593,
      "grad_norm": 0.27335697412490845,
      "learning_rate": 0.0004,
      "loss": 4.8778,
      "step": 11310
    },
    {
      "epoch": 0.23181045604406855,
      "grad_norm": 0.2511597275733948,
      "learning_rate": 0.0004,
      "loss": 4.891,
      "step": 11320
    },
    {
      "epoch": 0.2320152355988778,
      "grad_norm": 0.2607283592224121,
      "learning_rate": 0.0004,
      "loss": 4.8494,
      "step": 11330
    },
    {
      "epoch": 0.23222001515368706,
      "grad_norm": 0.24405378103256226,
      "learning_rate": 0.0004,
      "loss": 4.8579,
      "step": 11340
    },
    {
      "epoch": 0.2324247947084963,
      "grad_norm": 0.3528309166431427,
      "learning_rate": 0.0004,
      "loss": 4.8861,
      "step": 11350
    },
    {
      "epoch": 0.23262957426330555,
      "grad_norm": 0.30794113874435425,
      "learning_rate": 0.0004,
      "loss": 4.8548,
      "step": 11360
    },
    {
      "epoch": 0.2328343538181148,
      "grad_norm": 0.24918349087238312,
      "learning_rate": 0.0004,
      "loss": 4.8867,
      "step": 11370
    },
    {
      "epoch": 0.23303913337292403,
      "grad_norm": 0.28293827176094055,
      "learning_rate": 0.0004,
      "loss": 4.8692,
      "step": 11380
    },
    {
      "epoch": 0.2332439129277333,
      "grad_norm": 0.27161991596221924,
      "learning_rate": 0.0004,
      "loss": 4.8822,
      "step": 11390
    },
    {
      "epoch": 0.23344869248254255,
      "grad_norm": 0.2860964834690094,
      "learning_rate": 0.0004,
      "loss": 4.8204,
      "step": 11400
    },
    {
      "epoch": 0.2336534720373518,
      "grad_norm": 0.24801798164844513,
      "learning_rate": 0.0004,
      "loss": 4.8735,
      "step": 11410
    },
    {
      "epoch": 0.23385825159216103,
      "grad_norm": 0.26853352785110474,
      "learning_rate": 0.0004,
      "loss": 4.8847,
      "step": 11420
    },
    {
      "epoch": 0.23406303114697027,
      "grad_norm": 0.27251923084259033,
      "learning_rate": 0.0004,
      "loss": 4.8839,
      "step": 11430
    },
    {
      "epoch": 0.23426781070177954,
      "grad_norm": 0.2775683104991913,
      "learning_rate": 0.0004,
      "loss": 4.9025,
      "step": 11440
    },
    {
      "epoch": 0.2344725902565888,
      "grad_norm": 0.23779477179050446,
      "learning_rate": 0.0004,
      "loss": 4.855,
      "step": 11450
    },
    {
      "epoch": 0.23467736981139803,
      "grad_norm": 0.327796071767807,
      "learning_rate": 0.0004,
      "loss": 4.8983,
      "step": 11460
    },
    {
      "epoch": 0.23488214936620727,
      "grad_norm": 0.25727519392967224,
      "learning_rate": 0.0004,
      "loss": 4.9148,
      "step": 11470
    },
    {
      "epoch": 0.2350869289210165,
      "grad_norm": 0.2634661793708801,
      "learning_rate": 0.0004,
      "loss": 4.8479,
      "step": 11480
    },
    {
      "epoch": 0.23529170847582578,
      "grad_norm": 0.24151748418807983,
      "learning_rate": 0.0004,
      "loss": 4.8902,
      "step": 11490
    },
    {
      "epoch": 0.23549648803063503,
      "grad_norm": 0.27254611253738403,
      "learning_rate": 0.0004,
      "loss": 4.848,
      "step": 11500
    },
    {
      "epoch": 0.23570126758544427,
      "grad_norm": 0.3315676748752594,
      "learning_rate": 0.0004,
      "loss": 4.8379,
      "step": 11510
    },
    {
      "epoch": 0.2359060471402535,
      "grad_norm": 0.2895018756389618,
      "learning_rate": 0.0004,
      "loss": 4.9049,
      "step": 11520
    },
    {
      "epoch": 0.23611082669506275,
      "grad_norm": 0.28068599104881287,
      "learning_rate": 0.0004,
      "loss": 4.8904,
      "step": 11530
    },
    {
      "epoch": 0.23631560624987202,
      "grad_norm": 0.27318912744522095,
      "learning_rate": 0.0004,
      "loss": 4.8829,
      "step": 11540
    },
    {
      "epoch": 0.23652038580468127,
      "grad_norm": 0.2426539659500122,
      "learning_rate": 0.0004,
      "loss": 4.8632,
      "step": 11550
    },
    {
      "epoch": 0.2367251653594905,
      "grad_norm": 0.2518153488636017,
      "learning_rate": 0.0004,
      "loss": 4.8686,
      "step": 11560
    },
    {
      "epoch": 0.23692994491429975,
      "grad_norm": 0.2427929937839508,
      "learning_rate": 0.0004,
      "loss": 4.8466,
      "step": 11570
    },
    {
      "epoch": 0.237134724469109,
      "grad_norm": 0.252975195646286,
      "learning_rate": 0.0004,
      "loss": 4.8835,
      "step": 11580
    },
    {
      "epoch": 0.23733950402391826,
      "grad_norm": 0.24528580904006958,
      "learning_rate": 0.0004,
      "loss": 4.8544,
      "step": 11590
    },
    {
      "epoch": 0.2375442835787275,
      "grad_norm": 0.24127419292926788,
      "learning_rate": 0.0004,
      "loss": 4.8496,
      "step": 11600
    },
    {
      "epoch": 0.23774906313353675,
      "grad_norm": 0.2564906179904938,
      "learning_rate": 0.0004,
      "loss": 4.844,
      "step": 11610
    },
    {
      "epoch": 0.237953842688346,
      "grad_norm": 0.24771487712860107,
      "learning_rate": 0.0004,
      "loss": 4.8822,
      "step": 11620
    },
    {
      "epoch": 0.23815862224315523,
      "grad_norm": 0.2635604739189148,
      "learning_rate": 0.0004,
      "loss": 4.862,
      "step": 11630
    },
    {
      "epoch": 0.2383634017979645,
      "grad_norm": 0.2611459493637085,
      "learning_rate": 0.0004,
      "loss": 4.8637,
      "step": 11640
    },
    {
      "epoch": 0.23856818135277374,
      "grad_norm": 0.2843973636627197,
      "learning_rate": 0.0004,
      "loss": 4.881,
      "step": 11650
    },
    {
      "epoch": 0.238772960907583,
      "grad_norm": 0.2378414124250412,
      "learning_rate": 0.0004,
      "loss": 4.8555,
      "step": 11660
    },
    {
      "epoch": 0.23897774046239223,
      "grad_norm": 0.2471654862165451,
      "learning_rate": 0.0004,
      "loss": 4.8853,
      "step": 11670
    },
    {
      "epoch": 0.23918252001720147,
      "grad_norm": 0.23446987569332123,
      "learning_rate": 0.0004,
      "loss": 4.8558,
      "step": 11680
    },
    {
      "epoch": 0.23938729957201074,
      "grad_norm": 0.2632286250591278,
      "learning_rate": 0.0004,
      "loss": 4.8768,
      "step": 11690
    },
    {
      "epoch": 0.23959207912681998,
      "grad_norm": 0.23805192112922668,
      "learning_rate": 0.0004,
      "loss": 4.8559,
      "step": 11700
    },
    {
      "epoch": 0.23979685868162923,
      "grad_norm": 0.24758028984069824,
      "learning_rate": 0.0004,
      "loss": 4.8264,
      "step": 11710
    },
    {
      "epoch": 0.24000163823643847,
      "grad_norm": 0.25618085265159607,
      "learning_rate": 0.0004,
      "loss": 4.8964,
      "step": 11720
    },
    {
      "epoch": 0.2402064177912477,
      "grad_norm": 0.27674999833106995,
      "learning_rate": 0.0004,
      "loss": 4.8612,
      "step": 11730
    },
    {
      "epoch": 0.24041119734605698,
      "grad_norm": 0.27882149815559387,
      "learning_rate": 0.0004,
      "loss": 4.879,
      "step": 11740
    },
    {
      "epoch": 0.24061597690086622,
      "grad_norm": 0.2586473226547241,
      "learning_rate": 0.0004,
      "loss": 4.8955,
      "step": 11750
    },
    {
      "epoch": 0.24082075645567547,
      "grad_norm": 0.26056963205337524,
      "learning_rate": 0.0004,
      "loss": 4.8392,
      "step": 11760
    },
    {
      "epoch": 0.2410255360104847,
      "grad_norm": 0.32773515582084656,
      "learning_rate": 0.0004,
      "loss": 4.8796,
      "step": 11770
    },
    {
      "epoch": 0.24123031556529395,
      "grad_norm": 0.4262612760066986,
      "learning_rate": 0.0004,
      "loss": 4.8597,
      "step": 11780
    },
    {
      "epoch": 0.24143509512010322,
      "grad_norm": 0.2506944537162781,
      "learning_rate": 0.0004,
      "loss": 4.845,
      "step": 11790
    },
    {
      "epoch": 0.24163987467491246,
      "grad_norm": 0.23852330446243286,
      "learning_rate": 0.0004,
      "loss": 4.8575,
      "step": 11800
    },
    {
      "epoch": 0.2418446542297217,
      "grad_norm": 0.23398999869823456,
      "learning_rate": 0.0004,
      "loss": 4.8889,
      "step": 11810
    },
    {
      "epoch": 0.24204943378453095,
      "grad_norm": 1.147854208946228,
      "learning_rate": 0.0004,
      "loss": 4.8906,
      "step": 11820
    },
    {
      "epoch": 0.2422542133393402,
      "grad_norm": 0.2868060767650604,
      "learning_rate": 0.0004,
      "loss": 4.8874,
      "step": 11830
    },
    {
      "epoch": 0.24245899289414946,
      "grad_norm": 0.26699528098106384,
      "learning_rate": 0.0004,
      "loss": 4.8525,
      "step": 11840
    },
    {
      "epoch": 0.2426637724489587,
      "grad_norm": 0.2768281102180481,
      "learning_rate": 0.0004,
      "loss": 4.8584,
      "step": 11850
    },
    {
      "epoch": 0.24286855200376795,
      "grad_norm": 0.2719232141971588,
      "learning_rate": 0.0004,
      "loss": 4.8487,
      "step": 11860
    },
    {
      "epoch": 0.2430733315585772,
      "grad_norm": 0.2644749581813812,
      "learning_rate": 0.0004,
      "loss": 4.8635,
      "step": 11870
    },
    {
      "epoch": 0.24327811111338643,
      "grad_norm": 0.27209368348121643,
      "learning_rate": 0.0004,
      "loss": 4.8635,
      "step": 11880
    },
    {
      "epoch": 0.2434828906681957,
      "grad_norm": 0.248479425907135,
      "learning_rate": 0.0004,
      "loss": 4.8584,
      "step": 11890
    },
    {
      "epoch": 0.24368767022300494,
      "grad_norm": 0.3290988504886627,
      "learning_rate": 0.0004,
      "loss": 4.8623,
      "step": 11900
    },
    {
      "epoch": 0.24389244977781419,
      "grad_norm": 0.2829190492630005,
      "learning_rate": 0.0004,
      "loss": 4.8083,
      "step": 11910
    },
    {
      "epoch": 0.24409722933262343,
      "grad_norm": 0.2845098376274109,
      "learning_rate": 0.0004,
      "loss": 4.8614,
      "step": 11920
    },
    {
      "epoch": 0.24430200888743267,
      "grad_norm": 0.2518509328365326,
      "learning_rate": 0.0004,
      "loss": 4.8812,
      "step": 11930
    },
    {
      "epoch": 0.24450678844224194,
      "grad_norm": 0.24854156374931335,
      "learning_rate": 0.0004,
      "loss": 4.8624,
      "step": 11940
    },
    {
      "epoch": 0.24471156799705118,
      "grad_norm": 0.29206472635269165,
      "learning_rate": 0.0004,
      "loss": 4.8661,
      "step": 11950
    },
    {
      "epoch": 0.24491634755186042,
      "grad_norm": 0.24712790548801422,
      "learning_rate": 0.0004,
      "loss": 4.8687,
      "step": 11960
    },
    {
      "epoch": 0.24512112710666967,
      "grad_norm": 0.25610557198524475,
      "learning_rate": 0.0004,
      "loss": 4.8644,
      "step": 11970
    },
    {
      "epoch": 0.2453259066614789,
      "grad_norm": 0.24736113846302032,
      "learning_rate": 0.0004,
      "loss": 4.862,
      "step": 11980
    },
    {
      "epoch": 0.24553068621628815,
      "grad_norm": 0.2579999268054962,
      "learning_rate": 0.0004,
      "loss": 4.8854,
      "step": 11990
    },
    {
      "epoch": 0.24573546577109742,
      "grad_norm": 0.2651893198490143,
      "learning_rate": 0.0004,
      "loss": 4.8444,
      "step": 12000
    },
    {
      "epoch": 0.24573546577109742,
      "eval_loss": 4.867031574249268,
      "eval_runtime": 4.2851,
      "eval_samples_per_second": 272.103,
      "eval_steps_per_second": 34.071,
      "step": 12000
    },
    {
      "epoch": 0.24594024532590666,
      "grad_norm": 0.2691151797771454,
      "learning_rate": 0.0004,
      "loss": 4.8714,
      "step": 12010
    },
    {
      "epoch": 0.2461450248807159,
      "grad_norm": 0.3087450861930847,
      "learning_rate": 0.0004,
      "loss": 4.8911,
      "step": 12020
    },
    {
      "epoch": 0.24634980443552515,
      "grad_norm": 0.5508211851119995,
      "learning_rate": 0.0004,
      "loss": 4.829,
      "step": 12030
    },
    {
      "epoch": 0.2465545839903344,
      "grad_norm": 0.2314455211162567,
      "learning_rate": 0.0004,
      "loss": 4.8503,
      "step": 12040
    },
    {
      "epoch": 0.24675936354514366,
      "grad_norm": 0.23762647807598114,
      "learning_rate": 0.0004,
      "loss": 4.8723,
      "step": 12050
    },
    {
      "epoch": 0.2469641430999529,
      "grad_norm": 0.24536122381687164,
      "learning_rate": 0.0004,
      "loss": 4.8459,
      "step": 12060
    },
    {
      "epoch": 0.24716892265476215,
      "grad_norm": 0.24700681865215302,
      "learning_rate": 0.0004,
      "loss": 4.8566,
      "step": 12070
    },
    {
      "epoch": 0.2473737022095714,
      "grad_norm": 0.34846436977386475,
      "learning_rate": 0.0004,
      "loss": 4.8286,
      "step": 12080
    },
    {
      "epoch": 0.24757848176438063,
      "grad_norm": 0.3168438971042633,
      "learning_rate": 0.0004,
      "loss": 4.8503,
      "step": 12090
    },
    {
      "epoch": 0.2477832613191899,
      "grad_norm": 0.2919858396053314,
      "learning_rate": 0.0004,
      "loss": 4.8207,
      "step": 12100
    },
    {
      "epoch": 0.24798804087399914,
      "grad_norm": 0.3359951674938202,
      "learning_rate": 0.0004,
      "loss": 4.8341,
      "step": 12110
    },
    {
      "epoch": 0.2481928204288084,
      "grad_norm": 0.4131545126438141,
      "learning_rate": 0.0004,
      "loss": 4.8733,
      "step": 12120
    },
    {
      "epoch": 0.24839759998361763,
      "grad_norm": 0.2484077513217926,
      "learning_rate": 0.0004,
      "loss": 4.8611,
      "step": 12130
    },
    {
      "epoch": 0.24860237953842687,
      "grad_norm": 0.2616230845451355,
      "learning_rate": 0.0004,
      "loss": 4.8953,
      "step": 12140
    },
    {
      "epoch": 0.24880715909323614,
      "grad_norm": 0.2490583062171936,
      "learning_rate": 0.0004,
      "loss": 4.8732,
      "step": 12150
    },
    {
      "epoch": 0.24901193864804538,
      "grad_norm": 0.2691861093044281,
      "learning_rate": 0.0004,
      "loss": 4.8612,
      "step": 12160
    },
    {
      "epoch": 0.24921671820285463,
      "grad_norm": 0.2695310413837433,
      "learning_rate": 0.0004,
      "loss": 4.8501,
      "step": 12170
    },
    {
      "epoch": 0.24942149775766387,
      "grad_norm": 0.2886080741882324,
      "learning_rate": 0.0004,
      "loss": 4.8829,
      "step": 12180
    },
    {
      "epoch": 0.2496262773124731,
      "grad_norm": 0.2912190854549408,
      "learning_rate": 0.0004,
      "loss": 4.828,
      "step": 12190
    },
    {
      "epoch": 0.24983105686728238,
      "grad_norm": 0.27903711795806885,
      "learning_rate": 0.0004,
      "loss": 4.8104,
      "step": 12200
    },
    {
      "epoch": 0.2500358364220916,
      "grad_norm": 0.2646547853946686,
      "learning_rate": 0.0004,
      "loss": 4.838,
      "step": 12210
    },
    {
      "epoch": 0.2502406159769009,
      "grad_norm": 0.2762759029865265,
      "learning_rate": 0.0004,
      "loss": 4.8467,
      "step": 12220
    },
    {
      "epoch": 0.25044539553171014,
      "grad_norm": 0.23205457627773285,
      "learning_rate": 0.0004,
      "loss": 4.8558,
      "step": 12230
    },
    {
      "epoch": 0.2506501750865194,
      "grad_norm": 0.2654586732387543,
      "learning_rate": 0.0004,
      "loss": 4.8856,
      "step": 12240
    },
    {
      "epoch": 0.2508549546413286,
      "grad_norm": 0.26850640773773193,
      "learning_rate": 0.0004,
      "loss": 4.8427,
      "step": 12250
    },
    {
      "epoch": 0.25105973419613786,
      "grad_norm": 0.27965614199638367,
      "learning_rate": 0.0004,
      "loss": 4.8214,
      "step": 12260
    },
    {
      "epoch": 0.2512645137509471,
      "grad_norm": 0.24927881360054016,
      "learning_rate": 0.0004,
      "loss": 4.867,
      "step": 12270
    },
    {
      "epoch": 0.25146929330575635,
      "grad_norm": 0.2590121924877167,
      "learning_rate": 0.0004,
      "loss": 4.8543,
      "step": 12280
    },
    {
      "epoch": 0.2516740728605656,
      "grad_norm": 0.2736886739730835,
      "learning_rate": 0.0004,
      "loss": 4.8594,
      "step": 12290
    },
    {
      "epoch": 0.25187885241537483,
      "grad_norm": 0.27250829339027405,
      "learning_rate": 0.0004,
      "loss": 4.8308,
      "step": 12300
    },
    {
      "epoch": 0.2520836319701841,
      "grad_norm": 0.25367042422294617,
      "learning_rate": 0.0004,
      "loss": 4.8708,
      "step": 12310
    },
    {
      "epoch": 0.2522884115249934,
      "grad_norm": 0.2685873806476593,
      "learning_rate": 0.0004,
      "loss": 4.8263,
      "step": 12320
    },
    {
      "epoch": 0.2524931910798026,
      "grad_norm": 0.24380801618099213,
      "learning_rate": 0.0004,
      "loss": 4.8282,
      "step": 12330
    },
    {
      "epoch": 0.25269797063461186,
      "grad_norm": 0.25229397416114807,
      "learning_rate": 0.0004,
      "loss": 4.8737,
      "step": 12340
    },
    {
      "epoch": 0.2529027501894211,
      "grad_norm": 0.2535671293735504,
      "learning_rate": 0.0004,
      "loss": 4.8529,
      "step": 12350
    },
    {
      "epoch": 0.25310752974423034,
      "grad_norm": 0.25247514247894287,
      "learning_rate": 0.0004,
      "loss": 4.8421,
      "step": 12360
    },
    {
      "epoch": 0.2533123092990396,
      "grad_norm": 0.27063506841659546,
      "learning_rate": 0.0004,
      "loss": 4.8642,
      "step": 12370
    },
    {
      "epoch": 0.2535170888538488,
      "grad_norm": 0.2680909335613251,
      "learning_rate": 0.0004,
      "loss": 4.8485,
      "step": 12380
    },
    {
      "epoch": 0.25372186840865807,
      "grad_norm": 0.24620793759822845,
      "learning_rate": 0.0004,
      "loss": 4.8643,
      "step": 12390
    },
    {
      "epoch": 0.2539266479634673,
      "grad_norm": 0.47949567437171936,
      "learning_rate": 0.0004,
      "loss": 4.8472,
      "step": 12400
    },
    {
      "epoch": 0.25413142751827655,
      "grad_norm": 0.2958216071128845,
      "learning_rate": 0.0004,
      "loss": 4.8684,
      "step": 12410
    },
    {
      "epoch": 0.2543362070730858,
      "grad_norm": 0.2521269917488098,
      "learning_rate": 0.0004,
      "loss": 4.8361,
      "step": 12420
    },
    {
      "epoch": 0.2545409866278951,
      "grad_norm": 0.2519783079624176,
      "learning_rate": 0.0004,
      "loss": 4.8443,
      "step": 12430
    },
    {
      "epoch": 0.25474576618270434,
      "grad_norm": 0.2367222160100937,
      "learning_rate": 0.0004,
      "loss": 4.8228,
      "step": 12440
    },
    {
      "epoch": 0.2549505457375136,
      "grad_norm": 0.26444026827812195,
      "learning_rate": 0.0004,
      "loss": 4.8343,
      "step": 12450
    },
    {
      "epoch": 0.2551553252923228,
      "grad_norm": 0.27004826068878174,
      "learning_rate": 0.0004,
      "loss": 4.8437,
      "step": 12460
    },
    {
      "epoch": 0.25536010484713206,
      "grad_norm": 0.25829994678497314,
      "learning_rate": 0.0004,
      "loss": 4.8329,
      "step": 12470
    },
    {
      "epoch": 0.2555648844019413,
      "grad_norm": 0.24384234845638275,
      "learning_rate": 0.0004,
      "loss": 4.8512,
      "step": 12480
    },
    {
      "epoch": 0.25576966395675055,
      "grad_norm": 0.2575257420539856,
      "learning_rate": 0.0004,
      "loss": 4.8562,
      "step": 12490
    },
    {
      "epoch": 0.2559744435115598,
      "grad_norm": 0.26241886615753174,
      "learning_rate": 0.0004,
      "loss": 4.8509,
      "step": 12500
    },
    {
      "epoch": 0.25617922306636903,
      "grad_norm": 0.25610288977622986,
      "learning_rate": 0.0004,
      "loss": 4.8747,
      "step": 12510
    },
    {
      "epoch": 0.2563840026211783,
      "grad_norm": 0.26983582973480225,
      "learning_rate": 0.0004,
      "loss": 4.8247,
      "step": 12520
    },
    {
      "epoch": 0.2565887821759876,
      "grad_norm": 0.24773332476615906,
      "learning_rate": 0.0004,
      "loss": 4.8414,
      "step": 12530
    },
    {
      "epoch": 0.2567935617307968,
      "grad_norm": 0.3391345143318176,
      "learning_rate": 0.0004,
      "loss": 4.8386,
      "step": 12540
    },
    {
      "epoch": 0.25699834128560606,
      "grad_norm": 0.2731577754020691,
      "learning_rate": 0.0004,
      "loss": 4.8222,
      "step": 12550
    },
    {
      "epoch": 0.2572031208404153,
      "grad_norm": 0.24546635150909424,
      "learning_rate": 0.0004,
      "loss": 4.8283,
      "step": 12560
    },
    {
      "epoch": 0.25740790039522454,
      "grad_norm": 0.2784971296787262,
      "learning_rate": 0.0004,
      "loss": 4.8522,
      "step": 12570
    },
    {
      "epoch": 0.2576126799500338,
      "grad_norm": 0.25216221809387207,
      "learning_rate": 0.0004,
      "loss": 4.8627,
      "step": 12580
    },
    {
      "epoch": 0.257817459504843,
      "grad_norm": 0.24354113638401031,
      "learning_rate": 0.0004,
      "loss": 4.8624,
      "step": 12590
    },
    {
      "epoch": 0.25802223905965227,
      "grad_norm": 0.2683369815349579,
      "learning_rate": 0.0004,
      "loss": 4.8449,
      "step": 12600
    },
    {
      "epoch": 0.2582270186144615,
      "grad_norm": 0.24807274341583252,
      "learning_rate": 0.0004,
      "loss": 4.8605,
      "step": 12610
    },
    {
      "epoch": 0.25843179816927075,
      "grad_norm": 0.2839908301830292,
      "learning_rate": 0.0004,
      "loss": 4.8308,
      "step": 12620
    },
    {
      "epoch": 0.25863657772408005,
      "grad_norm": 0.27576714754104614,
      "learning_rate": 0.0004,
      "loss": 4.8367,
      "step": 12630
    },
    {
      "epoch": 0.2588413572788893,
      "grad_norm": 0.2834664583206177,
      "learning_rate": 0.0004,
      "loss": 4.8193,
      "step": 12640
    },
    {
      "epoch": 0.25904613683369854,
      "grad_norm": 0.26658859848976135,
      "learning_rate": 0.0004,
      "loss": 4.8781,
      "step": 12650
    },
    {
      "epoch": 0.2592509163885078,
      "grad_norm": 0.2482862025499344,
      "learning_rate": 0.0004,
      "loss": 4.8382,
      "step": 12660
    },
    {
      "epoch": 0.259455695943317,
      "grad_norm": 0.2739425599575043,
      "learning_rate": 0.0004,
      "loss": 4.8662,
      "step": 12670
    },
    {
      "epoch": 0.25966047549812626,
      "grad_norm": 0.2573910653591156,
      "learning_rate": 0.0004,
      "loss": 4.8293,
      "step": 12680
    },
    {
      "epoch": 0.2598652550529355,
      "grad_norm": 0.24086612462997437,
      "learning_rate": 0.0004,
      "loss": 4.8442,
      "step": 12690
    },
    {
      "epoch": 0.26007003460774475,
      "grad_norm": 0.28721368312835693,
      "learning_rate": 0.0004,
      "loss": 4.8473,
      "step": 12700
    },
    {
      "epoch": 0.260274814162554,
      "grad_norm": 0.28750866651535034,
      "learning_rate": 0.0004,
      "loss": 4.8133,
      "step": 12710
    },
    {
      "epoch": 0.26047959371736323,
      "grad_norm": 0.25055843591690063,
      "learning_rate": 0.0004,
      "loss": 4.8315,
      "step": 12720
    },
    {
      "epoch": 0.26068437327217253,
      "grad_norm": 0.2917591333389282,
      "learning_rate": 0.0004,
      "loss": 4.8668,
      "step": 12730
    },
    {
      "epoch": 0.2608891528269818,
      "grad_norm": 0.2694515287876129,
      "learning_rate": 0.0004,
      "loss": 4.8818,
      "step": 12740
    },
    {
      "epoch": 0.261093932381791,
      "grad_norm": 0.3116830289363861,
      "learning_rate": 0.0004,
      "loss": 4.8163,
      "step": 12750
    },
    {
      "epoch": 0.26129871193660026,
      "grad_norm": 0.28766903281211853,
      "learning_rate": 0.0004,
      "loss": 4.8063,
      "step": 12760
    },
    {
      "epoch": 0.2615034914914095,
      "grad_norm": 0.2460690587759018,
      "learning_rate": 0.0004,
      "loss": 4.8273,
      "step": 12770
    },
    {
      "epoch": 0.26170827104621874,
      "grad_norm": 0.2663492262363434,
      "learning_rate": 0.0004,
      "loss": 4.8314,
      "step": 12780
    },
    {
      "epoch": 0.261913050601028,
      "grad_norm": 0.32743126153945923,
      "learning_rate": 0.0004,
      "loss": 4.844,
      "step": 12790
    },
    {
      "epoch": 0.26211783015583723,
      "grad_norm": 0.26597467064857483,
      "learning_rate": 0.0004,
      "loss": 4.8344,
      "step": 12800
    },
    {
      "epoch": 0.26232260971064647,
      "grad_norm": 0.2584017515182495,
      "learning_rate": 0.0004,
      "loss": 4.7981,
      "step": 12810
    },
    {
      "epoch": 0.2625273892654557,
      "grad_norm": 0.26215118169784546,
      "learning_rate": 0.0004,
      "loss": 4.8678,
      "step": 12820
    },
    {
      "epoch": 0.262732168820265,
      "grad_norm": 0.3123892545700073,
      "learning_rate": 0.0004,
      "loss": 4.8201,
      "step": 12830
    },
    {
      "epoch": 0.26293694837507425,
      "grad_norm": 0.25252947211265564,
      "learning_rate": 0.0004,
      "loss": 4.8045,
      "step": 12840
    },
    {
      "epoch": 0.2631417279298835,
      "grad_norm": 0.3082652688026428,
      "learning_rate": 0.0004,
      "loss": 4.857,
      "step": 12850
    },
    {
      "epoch": 0.26334650748469274,
      "grad_norm": 0.2548525333404541,
      "learning_rate": 0.0004,
      "loss": 4.8549,
      "step": 12860
    },
    {
      "epoch": 0.263551287039502,
      "grad_norm": 0.3002852499485016,
      "learning_rate": 0.0004,
      "loss": 4.835,
      "step": 12870
    },
    {
      "epoch": 0.2637560665943112,
      "grad_norm": 0.2531322240829468,
      "learning_rate": 0.0004,
      "loss": 4.8445,
      "step": 12880
    },
    {
      "epoch": 0.26396084614912046,
      "grad_norm": 0.25382447242736816,
      "learning_rate": 0.0004,
      "loss": 4.8509,
      "step": 12890
    },
    {
      "epoch": 0.2641656257039297,
      "grad_norm": 0.24311015009880066,
      "learning_rate": 0.0004,
      "loss": 4.8653,
      "step": 12900
    },
    {
      "epoch": 0.26437040525873895,
      "grad_norm": 0.28109368681907654,
      "learning_rate": 0.0004,
      "loss": 4.8036,
      "step": 12910
    },
    {
      "epoch": 0.2645751848135482,
      "grad_norm": 0.3440668284893036,
      "learning_rate": 0.0004,
      "loss": 4.8436,
      "step": 12920
    },
    {
      "epoch": 0.2647799643683575,
      "grad_norm": 0.2777434289455414,
      "learning_rate": 0.0004,
      "loss": 4.8644,
      "step": 12930
    },
    {
      "epoch": 0.26498474392316673,
      "grad_norm": 0.2744606137275696,
      "learning_rate": 0.0004,
      "loss": 4.8402,
      "step": 12940
    },
    {
      "epoch": 0.265189523477976,
      "grad_norm": 0.27684512734413147,
      "learning_rate": 0.0004,
      "loss": 4.8427,
      "step": 12950
    },
    {
      "epoch": 0.2653943030327852,
      "grad_norm": 0.2579965889453888,
      "learning_rate": 0.0004,
      "loss": 4.8605,
      "step": 12960
    },
    {
      "epoch": 0.26559908258759446,
      "grad_norm": 0.2631833851337433,
      "learning_rate": 0.0004,
      "loss": 4.8222,
      "step": 12970
    },
    {
      "epoch": 0.2658038621424037,
      "grad_norm": 0.2552182674407959,
      "learning_rate": 0.0004,
      "loss": 4.8645,
      "step": 12980
    },
    {
      "epoch": 0.26600864169721294,
      "grad_norm": 0.29956743121147156,
      "learning_rate": 0.0004,
      "loss": 4.8358,
      "step": 12990
    },
    {
      "epoch": 0.2662134212520222,
      "grad_norm": 0.28119608759880066,
      "learning_rate": 0.0004,
      "loss": 4.8141,
      "step": 13000
    },
    {
      "epoch": 0.26641820080683143,
      "grad_norm": 0.26227396726608276,
      "learning_rate": 0.0004,
      "loss": 4.8162,
      "step": 13010
    },
    {
      "epoch": 0.26662298036164067,
      "grad_norm": 0.25072818994522095,
      "learning_rate": 0.0004,
      "loss": 4.8293,
      "step": 13020
    },
    {
      "epoch": 0.26682775991644997,
      "grad_norm": 0.2739667296409607,
      "learning_rate": 0.0004,
      "loss": 4.8323,
      "step": 13030
    },
    {
      "epoch": 0.2670325394712592,
      "grad_norm": 0.2423141598701477,
      "learning_rate": 0.0004,
      "loss": 4.8708,
      "step": 13040
    },
    {
      "epoch": 0.26723731902606845,
      "grad_norm": 0.2485387921333313,
      "learning_rate": 0.0004,
      "loss": 4.8406,
      "step": 13050
    },
    {
      "epoch": 0.2674420985808777,
      "grad_norm": 0.25066694617271423,
      "learning_rate": 0.0004,
      "loss": 4.8422,
      "step": 13060
    },
    {
      "epoch": 0.26764687813568694,
      "grad_norm": 0.3027220070362091,
      "learning_rate": 0.0004,
      "loss": 4.8095,
      "step": 13070
    },
    {
      "epoch": 0.2678516576904962,
      "grad_norm": 0.2532387673854828,
      "learning_rate": 0.0004,
      "loss": 4.8387,
      "step": 13080
    },
    {
      "epoch": 0.2680564372453054,
      "grad_norm": 0.2525840699672699,
      "learning_rate": 0.0004,
      "loss": 4.8176,
      "step": 13090
    },
    {
      "epoch": 0.26826121680011467,
      "grad_norm": 0.38074547052383423,
      "learning_rate": 0.0004,
      "loss": 4.8854,
      "step": 13100
    },
    {
      "epoch": 0.2684659963549239,
      "grad_norm": 0.2759318947792053,
      "learning_rate": 0.0004,
      "loss": 4.8315,
      "step": 13110
    },
    {
      "epoch": 0.26867077590973315,
      "grad_norm": 0.2519468069076538,
      "learning_rate": 0.0004,
      "loss": 4.8259,
      "step": 13120
    },
    {
      "epoch": 0.2688755554645424,
      "grad_norm": 0.269398033618927,
      "learning_rate": 0.0004,
      "loss": 4.8357,
      "step": 13130
    },
    {
      "epoch": 0.2690803350193517,
      "grad_norm": 0.2784993052482605,
      "learning_rate": 0.0004,
      "loss": 4.8182,
      "step": 13140
    },
    {
      "epoch": 0.26928511457416093,
      "grad_norm": 0.2523784935474396,
      "learning_rate": 0.0004,
      "loss": 4.8344,
      "step": 13150
    },
    {
      "epoch": 0.2694898941289702,
      "grad_norm": 0.2615765631198883,
      "learning_rate": 0.0004,
      "loss": 4.8556,
      "step": 13160
    },
    {
      "epoch": 0.2696946736837794,
      "grad_norm": 0.2687089443206787,
      "learning_rate": 0.0004,
      "loss": 4.8339,
      "step": 13170
    },
    {
      "epoch": 0.26989945323858866,
      "grad_norm": 0.2403348684310913,
      "learning_rate": 0.0004,
      "loss": 4.8512,
      "step": 13180
    },
    {
      "epoch": 0.2701042327933979,
      "grad_norm": 0.26277419924736023,
      "learning_rate": 0.0004,
      "loss": 4.845,
      "step": 13190
    },
    {
      "epoch": 0.27030901234820714,
      "grad_norm": 0.2699856162071228,
      "learning_rate": 0.0004,
      "loss": 4.8305,
      "step": 13200
    },
    {
      "epoch": 0.2705137919030164,
      "grad_norm": 0.2603802978992462,
      "learning_rate": 0.0004,
      "loss": 4.8595,
      "step": 13210
    },
    {
      "epoch": 0.27071857145782563,
      "grad_norm": 0.24295683205127716,
      "learning_rate": 0.0004,
      "loss": 4.8334,
      "step": 13220
    },
    {
      "epoch": 0.27092335101263487,
      "grad_norm": 0.26295214891433716,
      "learning_rate": 0.0004,
      "loss": 4.8032,
      "step": 13230
    },
    {
      "epoch": 0.27112813056744417,
      "grad_norm": 0.27498677372932434,
      "learning_rate": 0.0004,
      "loss": 4.8624,
      "step": 13240
    },
    {
      "epoch": 0.2713329101222534,
      "grad_norm": 0.27875176072120667,
      "learning_rate": 0.0004,
      "loss": 4.7901,
      "step": 13250
    },
    {
      "epoch": 0.27153768967706265,
      "grad_norm": 0.28385552763938904,
      "learning_rate": 0.0004,
      "loss": 4.8203,
      "step": 13260
    },
    {
      "epoch": 0.2717424692318719,
      "grad_norm": 0.2722179889678955,
      "learning_rate": 0.0004,
      "loss": 4.8618,
      "step": 13270
    },
    {
      "epoch": 0.27194724878668114,
      "grad_norm": 0.32632526755332947,
      "learning_rate": 0.0004,
      "loss": 4.8253,
      "step": 13280
    },
    {
      "epoch": 0.2721520283414904,
      "grad_norm": 0.33001449704170227,
      "learning_rate": 0.0004,
      "loss": 4.8079,
      "step": 13290
    },
    {
      "epoch": 0.2723568078962996,
      "grad_norm": 0.2521580755710602,
      "learning_rate": 0.0004,
      "loss": 4.8104,
      "step": 13300
    },
    {
      "epoch": 0.27256158745110887,
      "grad_norm": 0.25663989782333374,
      "learning_rate": 0.0004,
      "loss": 4.8222,
      "step": 13310
    },
    {
      "epoch": 0.2727663670059181,
      "grad_norm": 0.30672553181648254,
      "learning_rate": 0.0004,
      "loss": 4.8358,
      "step": 13320
    },
    {
      "epoch": 0.27297114656072735,
      "grad_norm": 0.2527655363082886,
      "learning_rate": 0.0004,
      "loss": 4.8398,
      "step": 13330
    },
    {
      "epoch": 0.27317592611553665,
      "grad_norm": 0.2610364258289337,
      "learning_rate": 0.0004,
      "loss": 4.8168,
      "step": 13340
    },
    {
      "epoch": 0.2733807056703459,
      "grad_norm": 0.26267045736312866,
      "learning_rate": 0.0004,
      "loss": 4.7884,
      "step": 13350
    },
    {
      "epoch": 0.27358548522515513,
      "grad_norm": 0.29419398307800293,
      "learning_rate": 0.0004,
      "loss": 4.8012,
      "step": 13360
    },
    {
      "epoch": 0.2737902647799644,
      "grad_norm": 0.30731770396232605,
      "learning_rate": 0.0004,
      "loss": 4.8163,
      "step": 13370
    },
    {
      "epoch": 0.2739950443347736,
      "grad_norm": 0.29380232095718384,
      "learning_rate": 0.0004,
      "loss": 4.8384,
      "step": 13380
    },
    {
      "epoch": 0.27419982388958286,
      "grad_norm": 0.2595800459384918,
      "learning_rate": 0.0004,
      "loss": 4.8695,
      "step": 13390
    },
    {
      "epoch": 0.2744046034443921,
      "grad_norm": 0.2919449508190155,
      "learning_rate": 0.0004,
      "loss": 4.8487,
      "step": 13400
    },
    {
      "epoch": 0.27460938299920135,
      "grad_norm": 0.34679651260375977,
      "learning_rate": 0.0004,
      "loss": 4.866,
      "step": 13410
    },
    {
      "epoch": 0.2748141625540106,
      "grad_norm": 0.26811766624450684,
      "learning_rate": 0.0004,
      "loss": 4.7992,
      "step": 13420
    },
    {
      "epoch": 0.27501894210881983,
      "grad_norm": 0.2668153941631317,
      "learning_rate": 0.0004,
      "loss": 4.795,
      "step": 13430
    },
    {
      "epoch": 0.27522372166362913,
      "grad_norm": 0.2654266953468323,
      "learning_rate": 0.0004,
      "loss": 4.7881,
      "step": 13440
    },
    {
      "epoch": 0.27542850121843837,
      "grad_norm": 0.27033060789108276,
      "learning_rate": 0.0004,
      "loss": 4.848,
      "step": 13450
    },
    {
      "epoch": 0.2756332807732476,
      "grad_norm": 0.27064186334609985,
      "learning_rate": 0.0004,
      "loss": 4.8293,
      "step": 13460
    },
    {
      "epoch": 0.27583806032805686,
      "grad_norm": 0.2640422582626343,
      "learning_rate": 0.0004,
      "loss": 4.8109,
      "step": 13470
    },
    {
      "epoch": 0.2760428398828661,
      "grad_norm": 0.2696080505847931,
      "learning_rate": 0.0004,
      "loss": 4.8407,
      "step": 13480
    },
    {
      "epoch": 0.27624761943767534,
      "grad_norm": 0.27713286876678467,
      "learning_rate": 0.0004,
      "loss": 4.8246,
      "step": 13490
    },
    {
      "epoch": 0.2764523989924846,
      "grad_norm": 0.27920326590538025,
      "learning_rate": 0.0004,
      "loss": 4.8669,
      "step": 13500
    },
    {
      "epoch": 0.2766571785472938,
      "grad_norm": 0.26664379239082336,
      "learning_rate": 0.0004,
      "loss": 4.7622,
      "step": 13510
    },
    {
      "epoch": 0.27686195810210307,
      "grad_norm": 0.25868895649909973,
      "learning_rate": 0.0004,
      "loss": 4.8356,
      "step": 13520
    },
    {
      "epoch": 0.2770667376569123,
      "grad_norm": 0.28665560483932495,
      "learning_rate": 0.0004,
      "loss": 4.8663,
      "step": 13530
    },
    {
      "epoch": 0.2772715172117216,
      "grad_norm": 0.24480880796909332,
      "learning_rate": 0.0004,
      "loss": 4.8462,
      "step": 13540
    },
    {
      "epoch": 0.27747629676653085,
      "grad_norm": 0.26968759298324585,
      "learning_rate": 0.0004,
      "loss": 4.8339,
      "step": 13550
    },
    {
      "epoch": 0.2776810763213401,
      "grad_norm": 0.24893125891685486,
      "learning_rate": 0.0004,
      "loss": 4.8065,
      "step": 13560
    },
    {
      "epoch": 0.27788585587614933,
      "grad_norm": 0.3047983646392822,
      "learning_rate": 0.0004,
      "loss": 4.7798,
      "step": 13570
    },
    {
      "epoch": 0.2780906354309586,
      "grad_norm": 0.2820693254470825,
      "learning_rate": 0.0004,
      "loss": 4.8267,
      "step": 13580
    },
    {
      "epoch": 0.2782954149857678,
      "grad_norm": 0.26072901487350464,
      "learning_rate": 0.0004,
      "loss": 4.8529,
      "step": 13590
    },
    {
      "epoch": 0.27850019454057706,
      "grad_norm": 0.4209215044975281,
      "learning_rate": 0.0004,
      "loss": 4.8032,
      "step": 13600
    },
    {
      "epoch": 0.2787049740953863,
      "grad_norm": 0.26476553082466125,
      "learning_rate": 0.0004,
      "loss": 4.8548,
      "step": 13610
    },
    {
      "epoch": 0.27890975365019555,
      "grad_norm": 0.28009968996047974,
      "learning_rate": 0.0004,
      "loss": 4.8097,
      "step": 13620
    },
    {
      "epoch": 0.2791145332050048,
      "grad_norm": 0.2644807994365692,
      "learning_rate": 0.0004,
      "loss": 4.8619,
      "step": 13630
    },
    {
      "epoch": 0.2793193127598141,
      "grad_norm": 0.2874789535999298,
      "learning_rate": 0.0004,
      "loss": 4.8178,
      "step": 13640
    },
    {
      "epoch": 0.27952409231462333,
      "grad_norm": 0.29766467213630676,
      "learning_rate": 0.0004,
      "loss": 4.8629,
      "step": 13650
    },
    {
      "epoch": 0.27972887186943257,
      "grad_norm": 0.2470841407775879,
      "learning_rate": 0.0004,
      "loss": 4.8008,
      "step": 13660
    },
    {
      "epoch": 0.2799336514242418,
      "grad_norm": 0.29466772079467773,
      "learning_rate": 0.0004,
      "loss": 4.8343,
      "step": 13670
    },
    {
      "epoch": 0.28013843097905106,
      "grad_norm": 0.24835602939128876,
      "learning_rate": 0.0004,
      "loss": 4.7946,
      "step": 13680
    },
    {
      "epoch": 0.2803432105338603,
      "grad_norm": 0.24725662171840668,
      "learning_rate": 0.0004,
      "loss": 4.8112,
      "step": 13690
    },
    {
      "epoch": 0.28054799008866954,
      "grad_norm": 0.34971049427986145,
      "learning_rate": 0.0004,
      "loss": 4.807,
      "step": 13700
    },
    {
      "epoch": 0.2807527696434788,
      "grad_norm": 0.2836594581604004,
      "learning_rate": 0.0004,
      "loss": 4.8198,
      "step": 13710
    },
    {
      "epoch": 0.280957549198288,
      "grad_norm": 0.25764623284339905,
      "learning_rate": 0.0004,
      "loss": 4.8185,
      "step": 13720
    },
    {
      "epoch": 0.28116232875309727,
      "grad_norm": 0.2625957131385803,
      "learning_rate": 0.0004,
      "loss": 4.8503,
      "step": 13730
    },
    {
      "epoch": 0.28136710830790657,
      "grad_norm": 0.27676263451576233,
      "learning_rate": 0.0004,
      "loss": 4.8225,
      "step": 13740
    },
    {
      "epoch": 0.2815718878627158,
      "grad_norm": 0.26145434379577637,
      "learning_rate": 0.0004,
      "loss": 4.8447,
      "step": 13750
    },
    {
      "epoch": 0.28177666741752505,
      "grad_norm": 0.269905686378479,
      "learning_rate": 0.0004,
      "loss": 4.8192,
      "step": 13760
    },
    {
      "epoch": 0.2819814469723343,
      "grad_norm": 0.27324187755584717,
      "learning_rate": 0.0004,
      "loss": 4.8376,
      "step": 13770
    },
    {
      "epoch": 0.28218622652714354,
      "grad_norm": 0.24123696982860565,
      "learning_rate": 0.0004,
      "loss": 4.8201,
      "step": 13780
    },
    {
      "epoch": 0.2823910060819528,
      "grad_norm": 0.29068803787231445,
      "learning_rate": 0.0004,
      "loss": 4.8178,
      "step": 13790
    },
    {
      "epoch": 0.282595785636762,
      "grad_norm": 0.31595665216445923,
      "learning_rate": 0.0004,
      "loss": 4.8576,
      "step": 13800
    },
    {
      "epoch": 0.28280056519157126,
      "grad_norm": 0.2802872955799103,
      "learning_rate": 0.0004,
      "loss": 4.8264,
      "step": 13810
    },
    {
      "epoch": 0.2830053447463805,
      "grad_norm": 0.26241886615753174,
      "learning_rate": 0.0004,
      "loss": 4.827,
      "step": 13820
    },
    {
      "epoch": 0.28321012430118975,
      "grad_norm": 0.2513846158981323,
      "learning_rate": 0.0004,
      "loss": 4.8288,
      "step": 13830
    },
    {
      "epoch": 0.283414903855999,
      "grad_norm": 0.28127583861351013,
      "learning_rate": 0.0004,
      "loss": 4.8314,
      "step": 13840
    },
    {
      "epoch": 0.2836196834108083,
      "grad_norm": 0.3425098657608032,
      "learning_rate": 0.0004,
      "loss": 4.8583,
      "step": 13850
    },
    {
      "epoch": 0.28382446296561753,
      "grad_norm": 0.2766761779785156,
      "learning_rate": 0.0004,
      "loss": 4.8232,
      "step": 13860
    },
    {
      "epoch": 0.2840292425204268,
      "grad_norm": 0.252049058675766,
      "learning_rate": 0.0004,
      "loss": 4.8189,
      "step": 13870
    },
    {
      "epoch": 0.284234022075236,
      "grad_norm": 0.26199719309806824,
      "learning_rate": 0.0004,
      "loss": 4.8369,
      "step": 13880
    },
    {
      "epoch": 0.28443880163004526,
      "grad_norm": 0.2577882707118988,
      "learning_rate": 0.0004,
      "loss": 4.8116,
      "step": 13890
    },
    {
      "epoch": 0.2846435811848545,
      "grad_norm": 0.2483375519514084,
      "learning_rate": 0.0004,
      "loss": 4.8511,
      "step": 13900
    },
    {
      "epoch": 0.28484836073966374,
      "grad_norm": 0.2529613673686981,
      "learning_rate": 0.0004,
      "loss": 4.8162,
      "step": 13910
    },
    {
      "epoch": 0.285053140294473,
      "grad_norm": 0.26325222849845886,
      "learning_rate": 0.0004,
      "loss": 4.8128,
      "step": 13920
    },
    {
      "epoch": 0.2852579198492822,
      "grad_norm": 0.25958263874053955,
      "learning_rate": 0.0004,
      "loss": 4.8054,
      "step": 13930
    },
    {
      "epoch": 0.28546269940409147,
      "grad_norm": 0.2949371933937073,
      "learning_rate": 0.0004,
      "loss": 4.8334,
      "step": 13940
    },
    {
      "epoch": 0.28566747895890077,
      "grad_norm": 0.27521225810050964,
      "learning_rate": 0.0004,
      "loss": 4.819,
      "step": 13950
    },
    {
      "epoch": 0.28587225851371,
      "grad_norm": 0.2950633764266968,
      "learning_rate": 0.0004,
      "loss": 4.8464,
      "step": 13960
    },
    {
      "epoch": 0.28607703806851925,
      "grad_norm": 0.29058393836021423,
      "learning_rate": 0.0004,
      "loss": 4.8675,
      "step": 13970
    },
    {
      "epoch": 0.2862818176233285,
      "grad_norm": 0.24924442172050476,
      "learning_rate": 0.0004,
      "loss": 4.8347,
      "step": 13980
    },
    {
      "epoch": 0.28648659717813774,
      "grad_norm": 0.25537240505218506,
      "learning_rate": 0.0004,
      "loss": 4.8463,
      "step": 13990
    },
    {
      "epoch": 0.286691376732947,
      "grad_norm": 0.2523159980773926,
      "learning_rate": 0.0004,
      "loss": 4.7991,
      "step": 14000
    },
    {
      "epoch": 0.286691376732947,
      "eval_loss": 4.831753253936768,
      "eval_runtime": 4.2725,
      "eval_samples_per_second": 272.908,
      "eval_steps_per_second": 34.172,
      "step": 14000
    },
    {
      "epoch": 0.2868961562877562,
      "grad_norm": 0.23810714483261108,
      "learning_rate": 0.0004,
      "loss": 4.8013,
      "step": 14010
    },
    {
      "epoch": 0.28710093584256546,
      "grad_norm": 0.2689630389213562,
      "learning_rate": 0.0004,
      "loss": 4.8474,
      "step": 14020
    },
    {
      "epoch": 0.2873057153973747,
      "grad_norm": 0.2767544686794281,
      "learning_rate": 0.0004,
      "loss": 4.8345,
      "step": 14030
    },
    {
      "epoch": 0.28751049495218395,
      "grad_norm": 0.687874972820282,
      "learning_rate": 0.0004,
      "loss": 4.8139,
      "step": 14040
    },
    {
      "epoch": 0.28771527450699325,
      "grad_norm": 0.24040289223194122,
      "learning_rate": 0.0004,
      "loss": 4.8437,
      "step": 14050
    },
    {
      "epoch": 0.2879200540618025,
      "grad_norm": 0.250606507062912,
      "learning_rate": 0.0004,
      "loss": 4.8557,
      "step": 14060
    },
    {
      "epoch": 0.28812483361661173,
      "grad_norm": 0.29520824551582336,
      "learning_rate": 0.0004,
      "loss": 4.8032,
      "step": 14070
    },
    {
      "epoch": 0.288329613171421,
      "grad_norm": 0.27100345492362976,
      "learning_rate": 0.0004,
      "loss": 4.8219,
      "step": 14080
    },
    {
      "epoch": 0.2885343927262302,
      "grad_norm": 0.2497226595878601,
      "learning_rate": 0.0004,
      "loss": 4.802,
      "step": 14090
    },
    {
      "epoch": 0.28873917228103946,
      "grad_norm": 0.2438780963420868,
      "learning_rate": 0.0004,
      "loss": 4.8118,
      "step": 14100
    },
    {
      "epoch": 0.2889439518358487,
      "grad_norm": 0.2748144268989563,
      "learning_rate": 0.0004,
      "loss": 4.8313,
      "step": 14110
    },
    {
      "epoch": 0.28914873139065794,
      "grad_norm": 0.2574600279331207,
      "learning_rate": 0.0004,
      "loss": 4.8441,
      "step": 14120
    },
    {
      "epoch": 0.2893535109454672,
      "grad_norm": 0.27852386236190796,
      "learning_rate": 0.0004,
      "loss": 4.8072,
      "step": 14130
    },
    {
      "epoch": 0.2895582905002764,
      "grad_norm": 0.26286157965660095,
      "learning_rate": 0.0004,
      "loss": 4.806,
      "step": 14140
    },
    {
      "epoch": 0.2897630700550857,
      "grad_norm": 0.26902690529823303,
      "learning_rate": 0.0004,
      "loss": 4.8018,
      "step": 14150
    },
    {
      "epoch": 0.28996784960989497,
      "grad_norm": 0.2873010039329529,
      "learning_rate": 0.0004,
      "loss": 4.829,
      "step": 14160
    },
    {
      "epoch": 0.2901726291647042,
      "grad_norm": 0.3127599358558655,
      "learning_rate": 0.0004,
      "loss": 4.8351,
      "step": 14170
    },
    {
      "epoch": 0.29037740871951345,
      "grad_norm": 0.2716214060783386,
      "learning_rate": 0.0004,
      "loss": 4.8234,
      "step": 14180
    },
    {
      "epoch": 0.2905821882743227,
      "grad_norm": 0.24700114130973816,
      "learning_rate": 0.0004,
      "loss": 4.8161,
      "step": 14190
    },
    {
      "epoch": 0.29078696782913194,
      "grad_norm": 0.23971880972385406,
      "learning_rate": 0.0004,
      "loss": 4.815,
      "step": 14200
    },
    {
      "epoch": 0.2909917473839412,
      "grad_norm": 0.25367897748947144,
      "learning_rate": 0.0004,
      "loss": 4.8097,
      "step": 14210
    },
    {
      "epoch": 0.2911965269387504,
      "grad_norm": 0.282354474067688,
      "learning_rate": 0.0004,
      "loss": 4.8113,
      "step": 14220
    },
    {
      "epoch": 0.29140130649355966,
      "grad_norm": 0.2775696814060211,
      "learning_rate": 0.0004,
      "loss": 4.7921,
      "step": 14230
    },
    {
      "epoch": 0.2916060860483689,
      "grad_norm": 0.26994824409484863,
      "learning_rate": 0.0004,
      "loss": 4.8218,
      "step": 14240
    },
    {
      "epoch": 0.2918108656031782,
      "grad_norm": 0.2710299491882324,
      "learning_rate": 0.0004,
      "loss": 4.8288,
      "step": 14250
    },
    {
      "epoch": 0.29201564515798745,
      "grad_norm": 0.27665871381759644,
      "learning_rate": 0.0004,
      "loss": 4.7902,
      "step": 14260
    },
    {
      "epoch": 0.2922204247127967,
      "grad_norm": 0.2653272747993469,
      "learning_rate": 0.0004,
      "loss": 4.8285,
      "step": 14270
    },
    {
      "epoch": 0.29242520426760593,
      "grad_norm": 0.2883239686489105,
      "learning_rate": 0.0004,
      "loss": 4.8096,
      "step": 14280
    },
    {
      "epoch": 0.2926299838224152,
      "grad_norm": 0.2962378263473511,
      "learning_rate": 0.0004,
      "loss": 4.8441,
      "step": 14290
    },
    {
      "epoch": 0.2928347633772244,
      "grad_norm": 0.7176985144615173,
      "learning_rate": 0.0004,
      "loss": 4.839,
      "step": 14300
    },
    {
      "epoch": 0.29303954293203366,
      "grad_norm": 0.2909333407878876,
      "learning_rate": 0.0004,
      "loss": 4.7834,
      "step": 14310
    },
    {
      "epoch": 0.2932443224868429,
      "grad_norm": 0.27297136187553406,
      "learning_rate": 0.0004,
      "loss": 4.8204,
      "step": 14320
    },
    {
      "epoch": 0.29344910204165214,
      "grad_norm": 0.2654514014720917,
      "learning_rate": 0.0004,
      "loss": 4.8241,
      "step": 14330
    },
    {
      "epoch": 0.2936538815964614,
      "grad_norm": 0.26377516984939575,
      "learning_rate": 0.0004,
      "loss": 4.8032,
      "step": 14340
    },
    {
      "epoch": 0.2938586611512707,
      "grad_norm": 0.2605004608631134,
      "learning_rate": 0.0004,
      "loss": 4.8014,
      "step": 14350
    },
    {
      "epoch": 0.2940634407060799,
      "grad_norm": 0.2723621129989624,
      "learning_rate": 0.0004,
      "loss": 4.7801,
      "step": 14360
    },
    {
      "epoch": 0.29426822026088917,
      "grad_norm": 0.29298287630081177,
      "learning_rate": 0.0004,
      "loss": 4.7942,
      "step": 14370
    },
    {
      "epoch": 0.2944729998156984,
      "grad_norm": 0.286304771900177,
      "learning_rate": 0.0004,
      "loss": 4.848,
      "step": 14380
    },
    {
      "epoch": 0.29467777937050765,
      "grad_norm": 0.2744176387786865,
      "learning_rate": 0.0004,
      "loss": 4.8144,
      "step": 14390
    },
    {
      "epoch": 0.2948825589253169,
      "grad_norm": 0.2628437876701355,
      "learning_rate": 0.0004,
      "loss": 4.7881,
      "step": 14400
    },
    {
      "epoch": 0.29508733848012614,
      "grad_norm": 0.24791161715984344,
      "learning_rate": 0.0004,
      "loss": 4.7821,
      "step": 14410
    },
    {
      "epoch": 0.2952921180349354,
      "grad_norm": 0.277873694896698,
      "learning_rate": 0.0004,
      "loss": 4.82,
      "step": 14420
    },
    {
      "epoch": 0.2954968975897446,
      "grad_norm": 0.378300279378891,
      "learning_rate": 0.0004,
      "loss": 4.8421,
      "step": 14430
    },
    {
      "epoch": 0.29570167714455386,
      "grad_norm": 0.2474977821111679,
      "learning_rate": 0.0004,
      "loss": 4.8351,
      "step": 14440
    },
    {
      "epoch": 0.29590645669936316,
      "grad_norm": 0.263503760099411,
      "learning_rate": 0.0004,
      "loss": 4.8063,
      "step": 14450
    },
    {
      "epoch": 0.2961112362541724,
      "grad_norm": 0.24829213321208954,
      "learning_rate": 0.0004,
      "loss": 4.798,
      "step": 14460
    },
    {
      "epoch": 0.29631601580898165,
      "grad_norm": 0.24435283243656158,
      "learning_rate": 0.0004,
      "loss": 4.8187,
      "step": 14470
    },
    {
      "epoch": 0.2965207953637909,
      "grad_norm": 0.25660616159439087,
      "learning_rate": 0.0004,
      "loss": 4.7885,
      "step": 14480
    },
    {
      "epoch": 0.29672557491860013,
      "grad_norm": 0.25041988492012024,
      "learning_rate": 0.0004,
      "loss": 4.8173,
      "step": 14490
    },
    {
      "epoch": 0.2969303544734094,
      "grad_norm": 0.26193779706954956,
      "learning_rate": 0.0004,
      "loss": 4.7869,
      "step": 14500
    },
    {
      "epoch": 0.2971351340282186,
      "grad_norm": 0.2585957646369934,
      "learning_rate": 0.0004,
      "loss": 4.805,
      "step": 14510
    },
    {
      "epoch": 0.29733991358302786,
      "grad_norm": 0.2570123076438904,
      "learning_rate": 0.0004,
      "loss": 4.8001,
      "step": 14520
    },
    {
      "epoch": 0.2975446931378371,
      "grad_norm": 0.2526560425758362,
      "learning_rate": 0.0004,
      "loss": 4.8152,
      "step": 14530
    },
    {
      "epoch": 0.29774947269264634,
      "grad_norm": 0.27037957310676575,
      "learning_rate": 0.0004,
      "loss": 4.8123,
      "step": 14540
    },
    {
      "epoch": 0.2979542522474556,
      "grad_norm": 0.2811831533908844,
      "learning_rate": 0.0004,
      "loss": 4.8307,
      "step": 14550
    },
    {
      "epoch": 0.2981590318022649,
      "grad_norm": 0.3060413599014282,
      "learning_rate": 0.0004,
      "loss": 4.8277,
      "step": 14560
    },
    {
      "epoch": 0.2983638113570741,
      "grad_norm": 0.25229233503341675,
      "learning_rate": 0.0004,
      "loss": 4.8506,
      "step": 14570
    },
    {
      "epoch": 0.29856859091188337,
      "grad_norm": 0.25974372029304504,
      "learning_rate": 0.0004,
      "loss": 4.7903,
      "step": 14580
    },
    {
      "epoch": 0.2987733704666926,
      "grad_norm": 0.30049508810043335,
      "learning_rate": 0.0004,
      "loss": 4.7895,
      "step": 14590
    },
    {
      "epoch": 0.29897815002150185,
      "grad_norm": 0.26262393593788147,
      "learning_rate": 0.0004,
      "loss": 4.7998,
      "step": 14600
    },
    {
      "epoch": 0.2991829295763111,
      "grad_norm": 0.27901288866996765,
      "learning_rate": 0.0004,
      "loss": 4.828,
      "step": 14610
    },
    {
      "epoch": 0.29938770913112034,
      "grad_norm": 0.2929536998271942,
      "learning_rate": 0.0004,
      "loss": 4.8018,
      "step": 14620
    },
    {
      "epoch": 0.2995924886859296,
      "grad_norm": 0.2543736696243286,
      "learning_rate": 0.0004,
      "loss": 4.808,
      "step": 14630
    },
    {
      "epoch": 0.2997972682407388,
      "grad_norm": 0.2771528959274292,
      "learning_rate": 0.0004,
      "loss": 4.8139,
      "step": 14640
    },
    {
      "epoch": 0.30000204779554807,
      "grad_norm": 0.26029521226882935,
      "learning_rate": 0.0004,
      "loss": 4.8371,
      "step": 14650
    },
    {
      "epoch": 0.30020682735035736,
      "grad_norm": 0.2559875547885895,
      "learning_rate": 0.0004,
      "loss": 4.8341,
      "step": 14660
    },
    {
      "epoch": 0.3004116069051666,
      "grad_norm": 0.2704298496246338,
      "learning_rate": 0.0004,
      "loss": 4.826,
      "step": 14670
    },
    {
      "epoch": 0.30061638645997585,
      "grad_norm": 0.2680201530456543,
      "learning_rate": 0.0004,
      "loss": 4.796,
      "step": 14680
    },
    {
      "epoch": 0.3008211660147851,
      "grad_norm": 0.3763655424118042,
      "learning_rate": 0.0004,
      "loss": 4.7792,
      "step": 14690
    },
    {
      "epoch": 0.30102594556959433,
      "grad_norm": 0.2631003260612488,
      "learning_rate": 0.0004,
      "loss": 4.8051,
      "step": 14700
    },
    {
      "epoch": 0.3012307251244036,
      "grad_norm": 0.27197256684303284,
      "learning_rate": 0.0004,
      "loss": 4.8303,
      "step": 14710
    },
    {
      "epoch": 0.3014355046792128,
      "grad_norm": 0.2803315818309784,
      "learning_rate": 0.0004,
      "loss": 4.7848,
      "step": 14720
    },
    {
      "epoch": 0.30164028423402206,
      "grad_norm": 0.25601300597190857,
      "learning_rate": 0.0004,
      "loss": 4.818,
      "step": 14730
    },
    {
      "epoch": 0.3018450637888313,
      "grad_norm": 0.2942025065422058,
      "learning_rate": 0.0004,
      "loss": 4.8216,
      "step": 14740
    },
    {
      "epoch": 0.30204984334364054,
      "grad_norm": 0.2761382758617401,
      "learning_rate": 0.0004,
      "loss": 4.8019,
      "step": 14750
    },
    {
      "epoch": 0.30225462289844984,
      "grad_norm": 0.25395074486732483,
      "learning_rate": 0.0004,
      "loss": 4.8356,
      "step": 14760
    },
    {
      "epoch": 0.3024594024532591,
      "grad_norm": 0.2864772379398346,
      "learning_rate": 0.0003999999923462465,
      "loss": 4.8108,
      "step": 14770
    },
    {
      "epoch": 0.3026641820080683,
      "grad_norm": 0.25308045744895935,
      "learning_rate": 0.00039999985627953394,
      "loss": 4.7768,
      "step": 14780
    },
    {
      "epoch": 0.30286896156287757,
      "grad_norm": 0.269564688205719,
      "learning_rate": 0.0003999995501295434,
      "loss": 4.8086,
      "step": 14790
    },
    {
      "epoch": 0.3030737411176868,
      "grad_norm": 0.25448814034461975,
      "learning_rate": 0.0003999990738965353,
      "loss": 4.7895,
      "step": 14800
    },
    {
      "epoch": 0.30327852067249605,
      "grad_norm": 0.33111077547073364,
      "learning_rate": 0.0003999984275809148,
      "loss": 4.8142,
      "step": 14810
    },
    {
      "epoch": 0.3034833002273053,
      "grad_norm": 0.357445627450943,
      "learning_rate": 0.00039999761118323117,
      "loss": 4.8057,
      "step": 14820
    },
    {
      "epoch": 0.30368807978211454,
      "grad_norm": 0.26140981912612915,
      "learning_rate": 0.000399996624704179,
      "loss": 4.8357,
      "step": 14830
    },
    {
      "epoch": 0.3038928593369238,
      "grad_norm": 0.2446521371603012,
      "learning_rate": 0.0003999954681445971,
      "loss": 4.7919,
      "step": 14840
    },
    {
      "epoch": 0.304097638891733,
      "grad_norm": 0.2549498379230499,
      "learning_rate": 0.000399994141505469,
      "loss": 4.7776,
      "step": 14850
    },
    {
      "epoch": 0.3043024184465423,
      "grad_norm": 0.29111766815185547,
      "learning_rate": 0.0003999926447879229,
      "loss": 4.8121,
      "step": 14860
    },
    {
      "epoch": 0.30450719800135156,
      "grad_norm": 0.2740670442581177,
      "learning_rate": 0.0003999909779932317,
      "loss": 4.7997,
      "step": 14870
    },
    {
      "epoch": 0.3047119775561608,
      "grad_norm": 0.2761414647102356,
      "learning_rate": 0.0003999891411228129,
      "loss": 4.8195,
      "step": 14880
    },
    {
      "epoch": 0.30491675711097005,
      "grad_norm": 0.4468514025211334,
      "learning_rate": 0.0003999871341782284,
      "loss": 4.796,
      "step": 14890
    },
    {
      "epoch": 0.3051215366657793,
      "grad_norm": 0.28377726674079895,
      "learning_rate": 0.00039998495716118515,
      "loss": 4.7818,
      "step": 14900
    },
    {
      "epoch": 0.30532631622058853,
      "grad_norm": 0.27489548921585083,
      "learning_rate": 0.00039998261007353445,
      "loss": 4.7957,
      "step": 14910
    },
    {
      "epoch": 0.3055310957753978,
      "grad_norm": 0.24948351085186005,
      "learning_rate": 0.0003999800929172723,
      "loss": 4.8064,
      "step": 14920
    },
    {
      "epoch": 0.305735875330207,
      "grad_norm": 0.3283112049102783,
      "learning_rate": 0.00039997740569453936,
      "loss": 4.7847,
      "step": 14930
    },
    {
      "epoch": 0.30594065488501626,
      "grad_norm": 0.28321975469589233,
      "learning_rate": 0.00039997454840762085,
      "loss": 4.7876,
      "step": 14940
    },
    {
      "epoch": 0.3061454344398255,
      "grad_norm": 0.2677721679210663,
      "learning_rate": 0.00039997152105894664,
      "loss": 4.8247,
      "step": 14950
    },
    {
      "epoch": 0.3063502139946348,
      "grad_norm": 0.26569730043411255,
      "learning_rate": 0.00039996832365109123,
      "loss": 4.8238,
      "step": 14960
    },
    {
      "epoch": 0.30655499354944404,
      "grad_norm": 0.25424227118492126,
      "learning_rate": 0.0003999649561867739,
      "loss": 4.8004,
      "step": 14970
    },
    {
      "epoch": 0.3067597731042533,
      "grad_norm": 0.25932741165161133,
      "learning_rate": 0.0003999614186688582,
      "loss": 4.8324,
      "step": 14980
    },
    {
      "epoch": 0.30696455265906253,
      "grad_norm": 0.2988094091415405,
      "learning_rate": 0.00039995771110035264,
      "loss": 4.796,
      "step": 14990
    },
    {
      "epoch": 0.30716933221387177,
      "grad_norm": 0.28092488646507263,
      "learning_rate": 0.0003999538334844101,
      "loss": 4.7976,
      "step": 15000
    },
    {
      "epoch": 0.307374111768681,
      "grad_norm": 0.24174724519252777,
      "learning_rate": 0.00039994978582432827,
      "loss": 4.8073,
      "step": 15010
    },
    {
      "epoch": 0.30757889132349026,
      "grad_norm": 0.26243990659713745,
      "learning_rate": 0.00039994556812354926,
      "loss": 4.8095,
      "step": 15020
    },
    {
      "epoch": 0.3077836708782995,
      "grad_norm": 0.2617025375366211,
      "learning_rate": 0.00039994118038565987,
      "loss": 4.8197,
      "step": 15030
    },
    {
      "epoch": 0.30798845043310874,
      "grad_norm": 0.28900280594825745,
      "learning_rate": 0.0003999366226143916,
      "loss": 4.7884,
      "step": 15040
    },
    {
      "epoch": 0.308193229987918,
      "grad_norm": 0.3000846207141876,
      "learning_rate": 0.0003999318948136204,
      "loss": 4.8049,
      "step": 15050
    },
    {
      "epoch": 0.3083980095427273,
      "grad_norm": 0.2622714340686798,
      "learning_rate": 0.00039992699698736684,
      "loss": 4.7619,
      "step": 15060
    },
    {
      "epoch": 0.3086027890975365,
      "grad_norm": 0.2860952913761139,
      "learning_rate": 0.0003999219291397961,
      "loss": 4.7925,
      "step": 15070
    },
    {
      "epoch": 0.30880756865234577,
      "grad_norm": 0.28790974617004395,
      "learning_rate": 0.0003999166912752181,
      "loss": 4.7417,
      "step": 15080
    },
    {
      "epoch": 0.309012348207155,
      "grad_norm": 0.25071680545806885,
      "learning_rate": 0.0003999112833980871,
      "loss": 4.8159,
      "step": 15090
    },
    {
      "epoch": 0.30921712776196425,
      "grad_norm": 0.27083414793014526,
      "learning_rate": 0.00039990570551300206,
      "loss": 4.7773,
      "step": 15100
    },
    {
      "epoch": 0.3094219073167735,
      "grad_norm": 0.29703348875045776,
      "learning_rate": 0.0003998999576247065,
      "loss": 4.812,
      "step": 15110
    },
    {
      "epoch": 0.30962668687158273,
      "grad_norm": 0.3592798411846161,
      "learning_rate": 0.0003998940397380886,
      "loss": 4.7812,
      "step": 15120
    },
    {
      "epoch": 0.309831466426392,
      "grad_norm": 0.267671138048172,
      "learning_rate": 0.0003998879518581809,
      "loss": 4.8375,
      "step": 15130
    },
    {
      "epoch": 0.3100362459812012,
      "grad_norm": 0.25858592987060547,
      "learning_rate": 0.00039988169399016074,
      "loss": 4.85,
      "step": 15140
    },
    {
      "epoch": 0.31024102553601046,
      "grad_norm": 0.2685125470161438,
      "learning_rate": 0.00039987526613934987,
      "loss": 4.8146,
      "step": 15150
    },
    {
      "epoch": 0.31044580509081976,
      "grad_norm": 0.47396326065063477,
      "learning_rate": 0.0003998686683112147,
      "loss": 4.7731,
      "step": 15160
    },
    {
      "epoch": 0.310650584645629,
      "grad_norm": 0.26775190234184265,
      "learning_rate": 0.00039986190051136605,
      "loss": 4.8311,
      "step": 15170
    },
    {
      "epoch": 0.31085536420043824,
      "grad_norm": 0.24957019090652466,
      "learning_rate": 0.0003998549627455594,
      "loss": 4.8156,
      "step": 15180
    },
    {
      "epoch": 0.3110601437552475,
      "grad_norm": 0.2548575699329376,
      "learning_rate": 0.0003998478550196948,
      "loss": 4.7477,
      "step": 15190
    },
    {
      "epoch": 0.31126492331005673,
      "grad_norm": 0.26530712842941284,
      "learning_rate": 0.00039984057733981674,
      "loss": 4.7858,
      "step": 15200
    },
    {
      "epoch": 0.31146970286486597,
      "grad_norm": 0.2542381286621094,
      "learning_rate": 0.0003998331297121143,
      "loss": 4.7727,
      "step": 15210
    },
    {
      "epoch": 0.3116744824196752,
      "grad_norm": 0.26388663053512573,
      "learning_rate": 0.00039982551214292104,
      "loss": 4.82,
      "step": 15220
    },
    {
      "epoch": 0.31187926197448446,
      "grad_norm": 0.2462894469499588,
      "learning_rate": 0.00039981772463871504,
      "loss": 4.8186,
      "step": 15230
    },
    {
      "epoch": 0.3120840415292937,
      "grad_norm": 0.28837886452674866,
      "learning_rate": 0.000399809767206119,
      "loss": 4.8143,
      "step": 15240
    },
    {
      "epoch": 0.31228882108410294,
      "grad_norm": 0.26220229268074036,
      "learning_rate": 0.00039980163985190006,
      "loss": 4.8026,
      "step": 15250
    },
    {
      "epoch": 0.3124936006389122,
      "grad_norm": 0.25457707047462463,
      "learning_rate": 0.00039979334258296976,
      "loss": 4.8067,
      "step": 15260
    },
    {
      "epoch": 0.3126983801937215,
      "grad_norm": 0.2709813117980957,
      "learning_rate": 0.0003997848754063844,
      "loss": 4.8274,
      "step": 15270
    },
    {
      "epoch": 0.3129031597485307,
      "grad_norm": 0.2797374427318573,
      "learning_rate": 0.0003997762383293444,
      "loss": 4.8108,
      "step": 15280
    },
    {
      "epoch": 0.31310793930333997,
      "grad_norm": 0.2793888449668884,
      "learning_rate": 0.0003997674313591951,
      "loss": 4.823,
      "step": 15290
    },
    {
      "epoch": 0.3133127188581492,
      "grad_norm": 0.2516084611415863,
      "learning_rate": 0.0003997584545034259,
      "loss": 4.7992,
      "step": 15300
    },
    {
      "epoch": 0.31351749841295845,
      "grad_norm": 0.25258857011795044,
      "learning_rate": 0.000399749307769671,
      "loss": 4.8223,
      "step": 15310
    },
    {
      "epoch": 0.3137222779677677,
      "grad_norm": 0.34943556785583496,
      "learning_rate": 0.00039973999116570895,
      "loss": 4.808,
      "step": 15320
    },
    {
      "epoch": 0.31392705752257694,
      "grad_norm": 0.25606784224510193,
      "learning_rate": 0.0003997305046994626,
      "loss": 4.7744,
      "step": 15330
    },
    {
      "epoch": 0.3141318370773862,
      "grad_norm": 0.27452796697616577,
      "learning_rate": 0.00039972084837899954,
      "loss": 4.7913,
      "step": 15340
    },
    {
      "epoch": 0.3143366166321954,
      "grad_norm": 0.2905919849872589,
      "learning_rate": 0.00039971102221253164,
      "loss": 4.7796,
      "step": 15350
    },
    {
      "epoch": 0.31454139618700466,
      "grad_norm": 0.2494192123413086,
      "learning_rate": 0.00039970102620841524,
      "loss": 4.7946,
      "step": 15360
    },
    {
      "epoch": 0.31474617574181396,
      "grad_norm": 0.25385144352912903,
      "learning_rate": 0.0003996908603751511,
      "loss": 4.8419,
      "step": 15370
    },
    {
      "epoch": 0.3149509552966232,
      "grad_norm": 0.2682260274887085,
      "learning_rate": 0.00039968052472138437,
      "loss": 4.7929,
      "step": 15380
    },
    {
      "epoch": 0.31515573485143245,
      "grad_norm": 0.9018281102180481,
      "learning_rate": 0.0003996700192559047,
      "loss": 4.8231,
      "step": 15390
    },
    {
      "epoch": 0.3153605144062417,
      "grad_norm": 0.27546435594558716,
      "learning_rate": 0.00039965934398764616,
      "loss": 4.8068,
      "step": 15400
    },
    {
      "epoch": 0.31556529396105093,
      "grad_norm": 0.27842995524406433,
      "learning_rate": 0.0003996484989256871,
      "loss": 4.836,
      "step": 15410
    },
    {
      "epoch": 0.3157700735158602,
      "grad_norm": 0.2818785309791565,
      "learning_rate": 0.0003996374840792505,
      "loss": 4.8011,
      "step": 15420
    },
    {
      "epoch": 0.3159748530706694,
      "grad_norm": 0.27837562561035156,
      "learning_rate": 0.0003996262994577034,
      "loss": 4.7473,
      "step": 15430
    },
    {
      "epoch": 0.31617963262547866,
      "grad_norm": 0.29244813323020935,
      "learning_rate": 0.00039961494507055747,
      "loss": 4.8349,
      "step": 15440
    },
    {
      "epoch": 0.3163844121802879,
      "grad_norm": 0.27803468704223633,
      "learning_rate": 0.00039960342092746866,
      "loss": 4.7763,
      "step": 15450
    },
    {
      "epoch": 0.31658919173509714,
      "grad_norm": 0.27822449803352356,
      "learning_rate": 0.0003995917270382373,
      "loss": 4.8026,
      "step": 15460
    },
    {
      "epoch": 0.31679397128990644,
      "grad_norm": 0.25251296162605286,
      "learning_rate": 0.00039957986341280805,
      "loss": 4.7876,
      "step": 15470
    },
    {
      "epoch": 0.3169987508447157,
      "grad_norm": 0.28681665658950806,
      "learning_rate": 0.00039956783006127,
      "loss": 4.7721,
      "step": 15480
    },
    {
      "epoch": 0.3172035303995249,
      "grad_norm": 0.27361565828323364,
      "learning_rate": 0.00039955562699385647,
      "loss": 4.7894,
      "step": 15490
    },
    {
      "epoch": 0.31740830995433417,
      "grad_norm": 0.264676570892334,
      "learning_rate": 0.00039954325422094516,
      "loss": 4.8182,
      "step": 15500
    },
    {
      "epoch": 0.3176130895091434,
      "grad_norm": 0.26645535230636597,
      "learning_rate": 0.00039953071175305807,
      "loss": 4.7915,
      "step": 15510
    },
    {
      "epoch": 0.31781786906395265,
      "grad_norm": 0.2505214810371399,
      "learning_rate": 0.00039951799960086156,
      "loss": 4.8,
      "step": 15520
    },
    {
      "epoch": 0.3180226486187619,
      "grad_norm": 0.2482193261384964,
      "learning_rate": 0.0003995051177751662,
      "loss": 4.8003,
      "step": 15530
    },
    {
      "epoch": 0.31822742817357114,
      "grad_norm": 0.2835414707660675,
      "learning_rate": 0.00039949206628692705,
      "loss": 4.813,
      "step": 15540
    },
    {
      "epoch": 0.3184322077283804,
      "grad_norm": 0.2924768030643463,
      "learning_rate": 0.0003994788451472432,
      "loss": 4.7809,
      "step": 15550
    },
    {
      "epoch": 0.3186369872831896,
      "grad_norm": 0.25978416204452515,
      "learning_rate": 0.0003994654543673581,
      "loss": 4.8088,
      "step": 15560
    },
    {
      "epoch": 0.3188417668379989,
      "grad_norm": 0.28737881779670715,
      "learning_rate": 0.0003994518939586596,
      "loss": 4.7877,
      "step": 15570
    },
    {
      "epoch": 0.31904654639280816,
      "grad_norm": 0.27220314741134644,
      "learning_rate": 0.00039943816393267967,
      "loss": 4.8257,
      "step": 15580
    },
    {
      "epoch": 0.3192513259476174,
      "grad_norm": 0.262319952249527,
      "learning_rate": 0.0003994242643010945,
      "loss": 4.7999,
      "step": 15590
    },
    {
      "epoch": 0.31945610550242665,
      "grad_norm": 0.24199426174163818,
      "learning_rate": 0.0003994101950757247,
      "loss": 4.7977,
      "step": 15600
    },
    {
      "epoch": 0.3196608850572359,
      "grad_norm": 0.32901597023010254,
      "learning_rate": 0.0003993959562685349,
      "loss": 4.8114,
      "step": 15610
    },
    {
      "epoch": 0.31986566461204513,
      "grad_norm": 0.24459275603294373,
      "learning_rate": 0.00039938154789163395,
      "loss": 4.7623,
      "step": 15620
    },
    {
      "epoch": 0.3200704441668544,
      "grad_norm": 0.26827767491340637,
      "learning_rate": 0.00039936696995727505,
      "loss": 4.7552,
      "step": 15630
    },
    {
      "epoch": 0.3202752237216636,
      "grad_norm": 0.254894882440567,
      "learning_rate": 0.00039935222247785554,
      "loss": 4.7585,
      "step": 15640
    },
    {
      "epoch": 0.32048000327647286,
      "grad_norm": 0.46504759788513184,
      "learning_rate": 0.0003993373054659169,
      "loss": 4.8143,
      "step": 15650
    },
    {
      "epoch": 0.3206847828312821,
      "grad_norm": 0.28007370233535767,
      "learning_rate": 0.00039932221893414487,
      "loss": 4.7931,
      "step": 15660
    },
    {
      "epoch": 0.3208895623860914,
      "grad_norm": 0.24671123921871185,
      "learning_rate": 0.0003993069628953692,
      "loss": 4.7946,
      "step": 15670
    },
    {
      "epoch": 0.32109434194090064,
      "grad_norm": 0.28686684370040894,
      "learning_rate": 0.000399291537362564,
      "loss": 4.84,
      "step": 15680
    },
    {
      "epoch": 0.3212991214957099,
      "grad_norm": 0.2491459995508194,
      "learning_rate": 0.0003992759423488472,
      "loss": 4.7923,
      "step": 15690
    },
    {
      "epoch": 0.3215039010505191,
      "grad_norm": 0.24470731616020203,
      "learning_rate": 0.00039926017786748135,
      "loss": 4.8088,
      "step": 15700
    },
    {
      "epoch": 0.32170868060532837,
      "grad_norm": 0.2530973553657532,
      "learning_rate": 0.00039924424393187263,
      "loss": 4.8138,
      "step": 15710
    },
    {
      "epoch": 0.3219134601601376,
      "grad_norm": 0.2529483437538147,
      "learning_rate": 0.0003992281405555716,
      "loss": 4.7946,
      "step": 15720
    },
    {
      "epoch": 0.32211823971494685,
      "grad_norm": 0.25579866766929626,
      "learning_rate": 0.00039921186775227277,
      "loss": 4.7957,
      "step": 15730
    },
    {
      "epoch": 0.3223230192697561,
      "grad_norm": 0.26418569684028625,
      "learning_rate": 0.0003991954255358149,
      "loss": 4.7991,
      "step": 15740
    },
    {
      "epoch": 0.32252779882456534,
      "grad_norm": 0.2771753966808319,
      "learning_rate": 0.00039917881392018064,
      "loss": 4.8127,
      "step": 15750
    },
    {
      "epoch": 0.3227325783793746,
      "grad_norm": 0.25980281829833984,
      "learning_rate": 0.0003991620329194969,
      "loss": 4.816,
      "step": 15760
    },
    {
      "epoch": 0.3229373579341839,
      "grad_norm": 0.27328717708587646,
      "learning_rate": 0.0003991450825480345,
      "loss": 4.8168,
      "step": 15770
    },
    {
      "epoch": 0.3231421374889931,
      "grad_norm": 0.2446042001247406,
      "learning_rate": 0.0003991279628202082,
      "loss": 4.7819,
      "step": 15780
    },
    {
      "epoch": 0.32334691704380236,
      "grad_norm": 0.3084605634212494,
      "learning_rate": 0.00039911067375057713,
      "loss": 4.8135,
      "step": 15790
    },
    {
      "epoch": 0.3235516965986116,
      "grad_norm": 0.2617594599723816,
      "learning_rate": 0.00039909321535384397,
      "loss": 4.8365,
      "step": 15800
    },
    {
      "epoch": 0.32375647615342085,
      "grad_norm": 0.24815472960472107,
      "learning_rate": 0.00039907558764485583,
      "loss": 4.7804,
      "step": 15810
    },
    {
      "epoch": 0.3239612557082301,
      "grad_norm": 0.2749323546886444,
      "learning_rate": 0.0003990577906386036,
      "loss": 4.8013,
      "step": 15820
    },
    {
      "epoch": 0.32416603526303933,
      "grad_norm": 0.2664775252342224,
      "learning_rate": 0.00039903982435022194,
      "loss": 4.8169,
      "step": 15830
    },
    {
      "epoch": 0.3243708148178486,
      "grad_norm": 0.28417569398880005,
      "learning_rate": 0.00039902168879498996,
      "loss": 4.777,
      "step": 15840
    },
    {
      "epoch": 0.3245755943726578,
      "grad_norm": 0.26321133971214294,
      "learning_rate": 0.00039900338398833027,
      "loss": 4.8145,
      "step": 15850
    },
    {
      "epoch": 0.32478037392746706,
      "grad_norm": 0.2757444977760315,
      "learning_rate": 0.00039898490994580963,
      "loss": 4.7887,
      "step": 15860
    },
    {
      "epoch": 0.32498515348227636,
      "grad_norm": 0.2671668231487274,
      "learning_rate": 0.00039896626668313874,
      "loss": 4.762,
      "step": 15870
    },
    {
      "epoch": 0.3251899330370856,
      "grad_norm": 0.279275506734848,
      "learning_rate": 0.0003989474542161721,
      "loss": 4.8086,
      "step": 15880
    },
    {
      "epoch": 0.32539471259189484,
      "grad_norm": 0.2854190170764923,
      "learning_rate": 0.0003989284725609081,
      "loss": 4.7784,
      "step": 15890
    },
    {
      "epoch": 0.3255994921467041,
      "grad_norm": 0.24825824797153473,
      "learning_rate": 0.0003989093217334891,
      "loss": 4.7646,
      "step": 15900
    },
    {
      "epoch": 0.3258042717015133,
      "grad_norm": 0.2487155646085739,
      "learning_rate": 0.0003988900017502013,
      "loss": 4.8313,
      "step": 15910
    },
    {
      "epoch": 0.32600905125632257,
      "grad_norm": 0.49503177404403687,
      "learning_rate": 0.0003988705126274749,
      "loss": 4.7791,
      "step": 15920
    },
    {
      "epoch": 0.3262138308111318,
      "grad_norm": 0.2913387715816498,
      "learning_rate": 0.0003988508543818835,
      "loss": 4.804,
      "step": 15930
    },
    {
      "epoch": 0.32641861036594105,
      "grad_norm": 0.2623058259487152,
      "learning_rate": 0.00039883102703014493,
      "loss": 4.7857,
      "step": 15940
    },
    {
      "epoch": 0.3266233899207503,
      "grad_norm": 0.25395411252975464,
      "learning_rate": 0.00039881103058912076,
      "loss": 4.7938,
      "step": 15950
    },
    {
      "epoch": 0.32682816947555954,
      "grad_norm": 0.2600069046020508,
      "learning_rate": 0.00039879086507581615,
      "loss": 4.7651,
      "step": 15960
    },
    {
      "epoch": 0.3270329490303688,
      "grad_norm": 0.287539005279541,
      "learning_rate": 0.0003987705305073804,
      "loss": 4.7746,
      "step": 15970
    },
    {
      "epoch": 0.3272377285851781,
      "grad_norm": 0.2536559998989105,
      "learning_rate": 0.00039875002690110625,
      "loss": 4.8204,
      "step": 15980
    },
    {
      "epoch": 0.3274425081399873,
      "grad_norm": 0.2836781144142151,
      "learning_rate": 0.00039872935427443033,
      "loss": 4.8302,
      "step": 15990
    },
    {
      "epoch": 0.32764728769479656,
      "grad_norm": 0.2694326341152191,
      "learning_rate": 0.00039870851264493297,
      "loss": 4.8092,
      "step": 16000
    },
    {
      "epoch": 0.32764728769479656,
      "eval_loss": 4.804255485534668,
      "eval_runtime": 4.3883,
      "eval_samples_per_second": 265.704,
      "eval_steps_per_second": 33.27,
      "step": 16000
    },
    {
      "epoch": 0.3278520672496058,
      "grad_norm": 0.2952719032764435,
      "learning_rate": 0.00039868750203033836,
      "loss": 4.7977,
      "step": 16010
    },
    {
      "epoch": 0.32805684680441505,
      "grad_norm": 0.2660292685031891,
      "learning_rate": 0.0003986663224485142,
      "loss": 4.8145,
      "step": 16020
    },
    {
      "epoch": 0.3282616263592243,
      "grad_norm": 0.2490932047367096,
      "learning_rate": 0.0003986449739174718,
      "loss": 4.8062,
      "step": 16030
    },
    {
      "epoch": 0.32846640591403353,
      "grad_norm": 0.2510133385658264,
      "learning_rate": 0.0003986234564553667,
      "loss": 4.7892,
      "step": 16040
    },
    {
      "epoch": 0.3286711854688428,
      "grad_norm": 0.2521056532859802,
      "learning_rate": 0.00039860177008049736,
      "loss": 4.8038,
      "step": 16050
    },
    {
      "epoch": 0.328875965023652,
      "grad_norm": 0.2704649567604065,
      "learning_rate": 0.00039857991481130644,
      "loss": 4.7656,
      "step": 16060
    },
    {
      "epoch": 0.32908074457846126,
      "grad_norm": 0.25223037600517273,
      "learning_rate": 0.0003985578906663799,
      "loss": 4.7934,
      "step": 16070
    },
    {
      "epoch": 0.32928552413327056,
      "grad_norm": 0.2978556752204895,
      "learning_rate": 0.0003985356976644476,
      "loss": 4.8232,
      "step": 16080
    },
    {
      "epoch": 0.3294903036880798,
      "grad_norm": 0.31600844860076904,
      "learning_rate": 0.0003985133358243827,
      "loss": 4.7498,
      "step": 16090
    },
    {
      "epoch": 0.32969508324288904,
      "grad_norm": 0.3390003442764282,
      "learning_rate": 0.0003984908051652022,
      "loss": 4.7813,
      "step": 16100
    },
    {
      "epoch": 0.3298998627976983,
      "grad_norm": 0.2710893154144287,
      "learning_rate": 0.0003984681057060665,
      "loss": 4.788,
      "step": 16110
    },
    {
      "epoch": 0.3301046423525075,
      "grad_norm": 0.27168989181518555,
      "learning_rate": 0.0003984452374662796,
      "loss": 4.804,
      "step": 16120
    },
    {
      "epoch": 0.33030942190731677,
      "grad_norm": 0.27508488297462463,
      "learning_rate": 0.00039842220046528906,
      "loss": 4.8179,
      "step": 16130
    },
    {
      "epoch": 0.330514201462126,
      "grad_norm": 0.3193860352039337,
      "learning_rate": 0.0003983989947226859,
      "loss": 4.8144,
      "step": 16140
    },
    {
      "epoch": 0.33071898101693525,
      "grad_norm": 0.24990639090538025,
      "learning_rate": 0.00039837562025820474,
      "loss": 4.7533,
      "step": 16150
    },
    {
      "epoch": 0.3309237605717445,
      "grad_norm": 0.25338494777679443,
      "learning_rate": 0.0003983520770917236,
      "loss": 4.7864,
      "step": 16160
    },
    {
      "epoch": 0.33112854012655374,
      "grad_norm": 0.27676770091056824,
      "learning_rate": 0.000398328365243264,
      "loss": 4.7785,
      "step": 16170
    },
    {
      "epoch": 0.33133331968136304,
      "grad_norm": 0.31176072359085083,
      "learning_rate": 0.0003983044847329908,
      "loss": 4.7863,
      "step": 16180
    },
    {
      "epoch": 0.3315380992361723,
      "grad_norm": 0.37212496995925903,
      "learning_rate": 0.0003982804355812125,
      "loss": 4.802,
      "step": 16190
    },
    {
      "epoch": 0.3317428787909815,
      "grad_norm": 0.2535504996776581,
      "learning_rate": 0.0003982562178083809,
      "loss": 4.8146,
      "step": 16200
    },
    {
      "epoch": 0.33194765834579076,
      "grad_norm": 0.26328185200691223,
      "learning_rate": 0.0003982318314350912,
      "loss": 4.758,
      "step": 16210
    },
    {
      "epoch": 0.3321524379006,
      "grad_norm": 0.26491978764533997,
      "learning_rate": 0.000398207276482082,
      "loss": 4.7593,
      "step": 16220
    },
    {
      "epoch": 0.33235721745540925,
      "grad_norm": 0.2541959285736084,
      "learning_rate": 0.0003981825529702351,
      "loss": 4.7874,
      "step": 16230
    },
    {
      "epoch": 0.3325619970102185,
      "grad_norm": 0.2653116285800934,
      "learning_rate": 0.00039815766092057603,
      "loss": 4.7881,
      "step": 16240
    },
    {
      "epoch": 0.33276677656502773,
      "grad_norm": 0.2704290449619293,
      "learning_rate": 0.0003981326003542733,
      "loss": 4.8138,
      "step": 16250
    },
    {
      "epoch": 0.332971556119837,
      "grad_norm": 0.25799617171287537,
      "learning_rate": 0.00039810737129263875,
      "loss": 4.7731,
      "step": 16260
    },
    {
      "epoch": 0.3331763356746462,
      "grad_norm": 0.28069770336151123,
      "learning_rate": 0.0003980819737571278,
      "loss": 4.7964,
      "step": 16270
    },
    {
      "epoch": 0.3333811152294555,
      "grad_norm": 0.32808852195739746,
      "learning_rate": 0.0003980564077693388,
      "loss": 4.7703,
      "step": 16280
    },
    {
      "epoch": 0.33358589478426476,
      "grad_norm": 0.28964656591415405,
      "learning_rate": 0.00039803067335101347,
      "loss": 4.799,
      "step": 16290
    },
    {
      "epoch": 0.333790674339074,
      "grad_norm": 0.2691846787929535,
      "learning_rate": 0.0003980047705240369,
      "loss": 4.7887,
      "step": 16300
    },
    {
      "epoch": 0.33399545389388324,
      "grad_norm": 0.2500579059123993,
      "learning_rate": 0.0003979786993104373,
      "loss": 4.7978,
      "step": 16310
    },
    {
      "epoch": 0.3342002334486925,
      "grad_norm": 0.25928881764411926,
      "learning_rate": 0.00039795245973238603,
      "loss": 4.7758,
      "step": 16320
    },
    {
      "epoch": 0.3344050130035017,
      "grad_norm": 0.28237369656562805,
      "learning_rate": 0.00039792605181219765,
      "loss": 4.7988,
      "step": 16330
    },
    {
      "epoch": 0.33460979255831097,
      "grad_norm": 0.2945230305194855,
      "learning_rate": 0.00039789947557232995,
      "loss": 4.7976,
      "step": 16340
    },
    {
      "epoch": 0.3348145721131202,
      "grad_norm": 0.24285341799259186,
      "learning_rate": 0.00039787273103538377,
      "loss": 4.8141,
      "step": 16350
    },
    {
      "epoch": 0.33501935166792945,
      "grad_norm": 0.3118240535259247,
      "learning_rate": 0.0003978458182241032,
      "loss": 4.7789,
      "step": 16360
    },
    {
      "epoch": 0.3352241312227387,
      "grad_norm": 0.285800963640213,
      "learning_rate": 0.0003978187371613752,
      "loss": 4.7804,
      "step": 16370
    },
    {
      "epoch": 0.335428910777548,
      "grad_norm": 0.25658172369003296,
      "learning_rate": 0.00039779148787023015,
      "loss": 4.7692,
      "step": 16380
    },
    {
      "epoch": 0.33563369033235724,
      "grad_norm": 0.2750941812992096,
      "learning_rate": 0.0003977640703738412,
      "loss": 4.78,
      "step": 16390
    },
    {
      "epoch": 0.3358384698871665,
      "grad_norm": 0.26658308506011963,
      "learning_rate": 0.00039773648469552464,
      "loss": 4.7697,
      "step": 16400
    },
    {
      "epoch": 0.3360432494419757,
      "grad_norm": 0.36863094568252563,
      "learning_rate": 0.00039770873085873994,
      "loss": 4.7782,
      "step": 16410
    },
    {
      "epoch": 0.33624802899678496,
      "grad_norm": 0.2577030062675476,
      "learning_rate": 0.00039768080888708923,
      "loss": 4.8157,
      "step": 16420
    },
    {
      "epoch": 0.3364528085515942,
      "grad_norm": 0.2842894494533539,
      "learning_rate": 0.000397652718804318,
      "loss": 4.785,
      "step": 16430
    },
    {
      "epoch": 0.33665758810640345,
      "grad_norm": 0.2669421434402466,
      "learning_rate": 0.0003976244606343144,
      "loss": 4.7567,
      "step": 16440
    },
    {
      "epoch": 0.3368623676612127,
      "grad_norm": 0.282748281955719,
      "learning_rate": 0.00039759603440110984,
      "loss": 4.777,
      "step": 16450
    },
    {
      "epoch": 0.33706714721602193,
      "grad_norm": 0.26537439227104187,
      "learning_rate": 0.00039756744012887826,
      "loss": 4.7889,
      "step": 16460
    },
    {
      "epoch": 0.3372719267708312,
      "grad_norm": 0.2713695168495178,
      "learning_rate": 0.00039753867784193684,
      "loss": 4.7887,
      "step": 16470
    },
    {
      "epoch": 0.3374767063256405,
      "grad_norm": 0.2605438530445099,
      "learning_rate": 0.0003975097475647455,
      "loss": 4.8369,
      "step": 16480
    },
    {
      "epoch": 0.3376814858804497,
      "grad_norm": 0.26819923520088196,
      "learning_rate": 0.00039748064932190703,
      "loss": 4.7778,
      "step": 16490
    },
    {
      "epoch": 0.33788626543525896,
      "grad_norm": 0.28676170110702515,
      "learning_rate": 0.000397451383138167,
      "loss": 4.8162,
      "step": 16500
    },
    {
      "epoch": 0.3380910449900682,
      "grad_norm": 0.2624913156032562,
      "learning_rate": 0.00039742194903841406,
      "loss": 4.809,
      "step": 16510
    },
    {
      "epoch": 0.33829582454487744,
      "grad_norm": 0.2781206965446472,
      "learning_rate": 0.00039739234704767927,
      "loss": 4.7931,
      "step": 16520
    },
    {
      "epoch": 0.3385006040996867,
      "grad_norm": 0.2613312005996704,
      "learning_rate": 0.0003973625771911368,
      "loss": 4.8285,
      "step": 16530
    },
    {
      "epoch": 0.33870538365449593,
      "grad_norm": 0.2740297317504883,
      "learning_rate": 0.0003973326394941034,
      "loss": 4.7699,
      "step": 16540
    },
    {
      "epoch": 0.33891016320930517,
      "grad_norm": 0.2648509740829468,
      "learning_rate": 0.00039730253398203853,
      "loss": 4.7953,
      "step": 16550
    },
    {
      "epoch": 0.3391149427641144,
      "grad_norm": 0.2660149931907654,
      "learning_rate": 0.0003972722606805445,
      "loss": 4.8125,
      "step": 16560
    },
    {
      "epoch": 0.33931972231892366,
      "grad_norm": 0.25253239274024963,
      "learning_rate": 0.0003972418196153662,
      "loss": 4.7355,
      "step": 16570
    },
    {
      "epoch": 0.33952450187373295,
      "grad_norm": 0.2715434730052948,
      "learning_rate": 0.00039721121081239127,
      "loss": 4.7798,
      "step": 16580
    },
    {
      "epoch": 0.3397292814285422,
      "grad_norm": 0.2549162209033966,
      "learning_rate": 0.00039718043429764994,
      "loss": 4.781,
      "step": 16590
    },
    {
      "epoch": 0.33993406098335144,
      "grad_norm": 0.38355880975723267,
      "learning_rate": 0.0003971494900973151,
      "loss": 4.8202,
      "step": 16600
    },
    {
      "epoch": 0.3401388405381607,
      "grad_norm": 0.28966614603996277,
      "learning_rate": 0.0003971183782377022,
      "loss": 4.8154,
      "step": 16610
    },
    {
      "epoch": 0.3403436200929699,
      "grad_norm": 0.2820665240287781,
      "learning_rate": 0.0003970870987452693,
      "loss": 4.7849,
      "step": 16620
    },
    {
      "epoch": 0.34054839964777917,
      "grad_norm": 0.2683079242706299,
      "learning_rate": 0.00039705565164661704,
      "loss": 4.7598,
      "step": 16630
    },
    {
      "epoch": 0.3407531792025884,
      "grad_norm": 0.2876993417739868,
      "learning_rate": 0.0003970240369684885,
      "loss": 4.7977,
      "step": 16640
    },
    {
      "epoch": 0.34095795875739765,
      "grad_norm": 0.2624385952949524,
      "learning_rate": 0.0003969922547377694,
      "loss": 4.7746,
      "step": 16650
    },
    {
      "epoch": 0.3411627383122069,
      "grad_norm": 0.28317347168922424,
      "learning_rate": 0.00039696030498148784,
      "loss": 4.7449,
      "step": 16660
    },
    {
      "epoch": 0.34136751786701613,
      "grad_norm": 0.27427518367767334,
      "learning_rate": 0.0003969281877268145,
      "loss": 4.7897,
      "step": 16670
    },
    {
      "epoch": 0.34157229742182543,
      "grad_norm": 0.26816168427467346,
      "learning_rate": 0.00039689590300106235,
      "loss": 4.7957,
      "step": 16680
    },
    {
      "epoch": 0.3417770769766347,
      "grad_norm": 0.27925974130630493,
      "learning_rate": 0.00039686345083168694,
      "loss": 4.77,
      "step": 16690
    },
    {
      "epoch": 0.3419818565314439,
      "grad_norm": 0.26638373732566833,
      "learning_rate": 0.00039683083124628614,
      "loss": 4.7762,
      "step": 16700
    },
    {
      "epoch": 0.34218663608625316,
      "grad_norm": 1.233842372894287,
      "learning_rate": 0.00039679804427260025,
      "loss": 4.8333,
      "step": 16710
    },
    {
      "epoch": 0.3423914156410624,
      "grad_norm": 0.2703392505645752,
      "learning_rate": 0.00039676508993851176,
      "loss": 4.7451,
      "step": 16720
    },
    {
      "epoch": 0.34259619519587164,
      "grad_norm": 0.34831905364990234,
      "learning_rate": 0.0003967319682720456,
      "loss": 4.7567,
      "step": 16730
    },
    {
      "epoch": 0.3428009747506809,
      "grad_norm": 0.28535258769989014,
      "learning_rate": 0.0003966986793013691,
      "loss": 4.7805,
      "step": 16740
    },
    {
      "epoch": 0.34300575430549013,
      "grad_norm": 0.3032117784023285,
      "learning_rate": 0.0003966652230547917,
      "loss": 4.7826,
      "step": 16750
    },
    {
      "epoch": 0.34321053386029937,
      "grad_norm": 0.30235326290130615,
      "learning_rate": 0.0003966315995607651,
      "loss": 4.815,
      "step": 16760
    },
    {
      "epoch": 0.3434153134151086,
      "grad_norm": 0.25587034225463867,
      "learning_rate": 0.00039659780884788347,
      "loss": 4.8103,
      "step": 16770
    },
    {
      "epoch": 0.34362009296991786,
      "grad_norm": 0.29329124093055725,
      "learning_rate": 0.00039656385094488287,
      "loss": 4.8138,
      "step": 16780
    },
    {
      "epoch": 0.34382487252472715,
      "grad_norm": 0.26573923230171204,
      "learning_rate": 0.0003965297258806417,
      "loss": 4.7919,
      "step": 16790
    },
    {
      "epoch": 0.3440296520795364,
      "grad_norm": 0.2600584924221039,
      "learning_rate": 0.0003964954336841805,
      "loss": 4.79,
      "step": 16800
    },
    {
      "epoch": 0.34423443163434564,
      "grad_norm": 0.2669816017150879,
      "learning_rate": 0.00039646097438466194,
      "loss": 4.775,
      "step": 16810
    },
    {
      "epoch": 0.3444392111891549,
      "grad_norm": 0.24775204062461853,
      "learning_rate": 0.00039642634801139077,
      "loss": 4.7953,
      "step": 16820
    },
    {
      "epoch": 0.3446439907439641,
      "grad_norm": 0.25428342819213867,
      "learning_rate": 0.00039639155459381397,
      "loss": 4.7679,
      "step": 16830
    },
    {
      "epoch": 0.34484877029877337,
      "grad_norm": 0.26264557242393494,
      "learning_rate": 0.0003963565941615202,
      "loss": 4.794,
      "step": 16840
    },
    {
      "epoch": 0.3450535498535826,
      "grad_norm": 0.26654911041259766,
      "learning_rate": 0.00039632146674424063,
      "loss": 4.7953,
      "step": 16850
    },
    {
      "epoch": 0.34525832940839185,
      "grad_norm": 0.3195827901363373,
      "learning_rate": 0.00039628617237184817,
      "loss": 4.758,
      "step": 16860
    },
    {
      "epoch": 0.3454631089632011,
      "grad_norm": 0.24894453585147858,
      "learning_rate": 0.00039625071107435766,
      "loss": 4.7523,
      "step": 16870
    },
    {
      "epoch": 0.34566788851801034,
      "grad_norm": 0.29483887553215027,
      "learning_rate": 0.00039621508288192615,
      "loss": 4.7372,
      "step": 16880
    },
    {
      "epoch": 0.34587266807281963,
      "grad_norm": 0.2702201008796692,
      "learning_rate": 0.0003961792878248523,
      "loss": 4.7731,
      "step": 16890
    },
    {
      "epoch": 0.3460774476276289,
      "grad_norm": 0.2855082154273987,
      "learning_rate": 0.0003961433259335769,
      "loss": 4.7888,
      "step": 16900
    },
    {
      "epoch": 0.3462822271824381,
      "grad_norm": 0.3111724257469177,
      "learning_rate": 0.0003961071972386826,
      "loss": 4.7917,
      "step": 16910
    },
    {
      "epoch": 0.34648700673724736,
      "grad_norm": 0.27766406536102295,
      "learning_rate": 0.00039607090177089374,
      "loss": 4.7401,
      "step": 16920
    },
    {
      "epoch": 0.3466917862920566,
      "grad_norm": 0.2639821171760559,
      "learning_rate": 0.00039603443956107674,
      "loss": 4.7449,
      "step": 16930
    },
    {
      "epoch": 0.34689656584686585,
      "grad_norm": 0.264573872089386,
      "learning_rate": 0.0003959978106402396,
      "loss": 4.7777,
      "step": 16940
    },
    {
      "epoch": 0.3471013454016751,
      "grad_norm": 0.25831058621406555,
      "learning_rate": 0.0003959610150395322,
      "loss": 4.7678,
      "step": 16950
    },
    {
      "epoch": 0.34730612495648433,
      "grad_norm": 0.27648264169692993,
      "learning_rate": 0.0003959240527902461,
      "loss": 4.7753,
      "step": 16960
    },
    {
      "epoch": 0.3475109045112936,
      "grad_norm": 0.23673586547374725,
      "learning_rate": 0.0003958869239238147,
      "loss": 4.7935,
      "step": 16970
    },
    {
      "epoch": 0.3477156840661028,
      "grad_norm": 0.27137354016304016,
      "learning_rate": 0.00039584962847181293,
      "loss": 4.768,
      "step": 16980
    },
    {
      "epoch": 0.3479204636209121,
      "grad_norm": 0.30974748730659485,
      "learning_rate": 0.0003958121664659576,
      "loss": 4.7705,
      "step": 16990
    },
    {
      "epoch": 0.34812524317572136,
      "grad_norm": 0.29479289054870605,
      "learning_rate": 0.000395774537938107,
      "loss": 4.7628,
      "step": 17000
    },
    {
      "epoch": 0.3483300227305306,
      "grad_norm": 0.2980476915836334,
      "learning_rate": 0.00039573674292026095,
      "loss": 4.7691,
      "step": 17010
    },
    {
      "epoch": 0.34853480228533984,
      "grad_norm": 0.26816049218177795,
      "learning_rate": 0.0003956987814445611,
      "loss": 4.7873,
      "step": 17020
    },
    {
      "epoch": 0.3487395818401491,
      "grad_norm": 0.29303935170173645,
      "learning_rate": 0.0003956606535432905,
      "loss": 4.7955,
      "step": 17030
    },
    {
      "epoch": 0.3489443613949583,
      "grad_norm": 0.27338966727256775,
      "learning_rate": 0.0003956223592488738,
      "loss": 4.7842,
      "step": 17040
    },
    {
      "epoch": 0.34914914094976757,
      "grad_norm": 0.2604229748249054,
      "learning_rate": 0.000395583898593877,
      "loss": 4.7882,
      "step": 17050
    },
    {
      "epoch": 0.3493539205045768,
      "grad_norm": 0.2727995216846466,
      "learning_rate": 0.00039554527161100785,
      "loss": 4.801,
      "step": 17060
    },
    {
      "epoch": 0.34955870005938605,
      "grad_norm": 0.26669710874557495,
      "learning_rate": 0.0003955064783331152,
      "loss": 4.7601,
      "step": 17070
    },
    {
      "epoch": 0.3497634796141953,
      "grad_norm": 0.2862171530723572,
      "learning_rate": 0.00039546751879318976,
      "loss": 4.8028,
      "step": 17080
    },
    {
      "epoch": 0.3499682591690046,
      "grad_norm": 0.25458213686943054,
      "learning_rate": 0.0003954283930243632,
      "loss": 4.7981,
      "step": 17090
    },
    {
      "epoch": 0.35017303872381383,
      "grad_norm": 0.2625376582145691,
      "learning_rate": 0.0003953891010599088,
      "loss": 4.7393,
      "step": 17100
    },
    {
      "epoch": 0.3503778182786231,
      "grad_norm": 0.25130653381347656,
      "learning_rate": 0.0003953496429332411,
      "loss": 4.8091,
      "step": 17110
    },
    {
      "epoch": 0.3505825978334323,
      "grad_norm": 0.3048926591873169,
      "learning_rate": 0.00039531001867791605,
      "loss": 4.7784,
      "step": 17120
    },
    {
      "epoch": 0.35078737738824156,
      "grad_norm": 0.2529113292694092,
      "learning_rate": 0.0003952702283276306,
      "loss": 4.764,
      "step": 17130
    },
    {
      "epoch": 0.3509921569430508,
      "grad_norm": 0.3067050576210022,
      "learning_rate": 0.00039523027191622337,
      "loss": 4.7556,
      "step": 17140
    },
    {
      "epoch": 0.35119693649786005,
      "grad_norm": 0.26279208064079285,
      "learning_rate": 0.00039519014947767374,
      "loss": 4.7482,
      "step": 17150
    },
    {
      "epoch": 0.3514017160526693,
      "grad_norm": 0.26440808176994324,
      "learning_rate": 0.00039514986104610266,
      "loss": 4.7705,
      "step": 17160
    },
    {
      "epoch": 0.35160649560747853,
      "grad_norm": 0.2666628360748291,
      "learning_rate": 0.00039510940665577214,
      "loss": 4.7793,
      "step": 17170
    },
    {
      "epoch": 0.3518112751622878,
      "grad_norm": 0.256744384765625,
      "learning_rate": 0.0003950687863410851,
      "loss": 4.7938,
      "step": 17180
    },
    {
      "epoch": 0.35201605471709707,
      "grad_norm": 0.2572229504585266,
      "learning_rate": 0.00039502800013658596,
      "loss": 4.7568,
      "step": 17190
    },
    {
      "epoch": 0.3522208342719063,
      "grad_norm": 0.28228095173835754,
      "learning_rate": 0.00039498704807695986,
      "loss": 4.7961,
      "step": 17200
    },
    {
      "epoch": 0.35242561382671556,
      "grad_norm": 0.3019843101501465,
      "learning_rate": 0.00039494593019703307,
      "loss": 4.7828,
      "step": 17210
    },
    {
      "epoch": 0.3526303933815248,
      "grad_norm": 0.2625148296356201,
      "learning_rate": 0.0003949046465317731,
      "loss": 4.756,
      "step": 17220
    },
    {
      "epoch": 0.35283517293633404,
      "grad_norm": 0.2715732455253601,
      "learning_rate": 0.00039486319711628826,
      "loss": 4.7641,
      "step": 17230
    },
    {
      "epoch": 0.3530399524911433,
      "grad_norm": 0.26373767852783203,
      "learning_rate": 0.0003948215819858277,
      "loss": 4.7837,
      "step": 17240
    },
    {
      "epoch": 0.3532447320459525,
      "grad_norm": 0.26473498344421387,
      "learning_rate": 0.00039477980117578175,
      "loss": 4.7717,
      "step": 17250
    },
    {
      "epoch": 0.35344951160076177,
      "grad_norm": 0.30615073442459106,
      "learning_rate": 0.0003947378547216815,
      "loss": 4.7593,
      "step": 17260
    },
    {
      "epoch": 0.353654291155571,
      "grad_norm": 0.2714005410671234,
      "learning_rate": 0.0003946957426591989,
      "loss": 4.7721,
      "step": 17270
    },
    {
      "epoch": 0.35385907071038025,
      "grad_norm": 0.2753409743309021,
      "learning_rate": 0.0003946534650241468,
      "loss": 4.7746,
      "step": 17280
    },
    {
      "epoch": 0.35406385026518955,
      "grad_norm": 0.280316025018692,
      "learning_rate": 0.00039461102185247876,
      "loss": 4.7368,
      "step": 17290
    },
    {
      "epoch": 0.3542686298199988,
      "grad_norm": 0.2765102684497833,
      "learning_rate": 0.00039456841318028933,
      "loss": 4.7986,
      "step": 17300
    },
    {
      "epoch": 0.35447340937480804,
      "grad_norm": 0.2677052617073059,
      "learning_rate": 0.0003945256390438134,
      "loss": 4.7974,
      "step": 17310
    },
    {
      "epoch": 0.3546781889296173,
      "grad_norm": 0.262820839881897,
      "learning_rate": 0.000394482699479427,
      "loss": 4.7682,
      "step": 17320
    },
    {
      "epoch": 0.3548829684844265,
      "grad_norm": 0.25526392459869385,
      "learning_rate": 0.00039443959452364674,
      "loss": 4.7675,
      "step": 17330
    },
    {
      "epoch": 0.35508774803923576,
      "grad_norm": 0.30862829089164734,
      "learning_rate": 0.0003943963242131296,
      "loss": 4.7875,
      "step": 17340
    },
    {
      "epoch": 0.355292527594045,
      "grad_norm": 0.2543276846408844,
      "learning_rate": 0.0003943528885846734,
      "loss": 4.7652,
      "step": 17350
    },
    {
      "epoch": 0.35549730714885425,
      "grad_norm": 0.24953249096870422,
      "learning_rate": 0.00039430928767521673,
      "loss": 4.7657,
      "step": 17360
    },
    {
      "epoch": 0.3557020867036635,
      "grad_norm": 0.26887795329093933,
      "learning_rate": 0.00039426552152183834,
      "loss": 4.7724,
      "step": 17370
    },
    {
      "epoch": 0.35590686625847273,
      "grad_norm": 0.27610909938812256,
      "learning_rate": 0.0003942215901617579,
      "loss": 4.7987,
      "step": 17380
    },
    {
      "epoch": 0.35611164581328203,
      "grad_norm": 0.26154568791389465,
      "learning_rate": 0.0003941774936323352,
      "loss": 4.7506,
      "step": 17390
    },
    {
      "epoch": 0.35631642536809127,
      "grad_norm": 0.456100732088089,
      "learning_rate": 0.0003941332319710708,
      "loss": 4.7871,
      "step": 17400
    },
    {
      "epoch": 0.3565212049229005,
      "grad_norm": 0.24634867906570435,
      "learning_rate": 0.0003940888052156056,
      "loss": 4.7905,
      "step": 17410
    },
    {
      "epoch": 0.35672598447770976,
      "grad_norm": 0.29051482677459717,
      "learning_rate": 0.00039404421340372066,
      "loss": 4.7968,
      "step": 17420
    },
    {
      "epoch": 0.356930764032519,
      "grad_norm": 0.2819485068321228,
      "learning_rate": 0.0003939994565733379,
      "loss": 4.7491,
      "step": 17430
    },
    {
      "epoch": 0.35713554358732824,
      "grad_norm": 0.26910969614982605,
      "learning_rate": 0.00039395453476251906,
      "loss": 4.7611,
      "step": 17440
    },
    {
      "epoch": 0.3573403231421375,
      "grad_norm": 0.2533196806907654,
      "learning_rate": 0.00039390944800946656,
      "loss": 4.7859,
      "step": 17450
    },
    {
      "epoch": 0.3575451026969467,
      "grad_norm": 0.3193216323852539,
      "learning_rate": 0.00039386419635252285,
      "loss": 4.7623,
      "step": 17460
    },
    {
      "epoch": 0.35774988225175597,
      "grad_norm": 0.2615881562232971,
      "learning_rate": 0.0003938187798301707,
      "loss": 4.7673,
      "step": 17470
    },
    {
      "epoch": 0.3579546618065652,
      "grad_norm": 0.26223790645599365,
      "learning_rate": 0.00039377319848103314,
      "loss": 4.769,
      "step": 17480
    },
    {
      "epoch": 0.35815944136137445,
      "grad_norm": 0.2690642476081848,
      "learning_rate": 0.0003937274523438733,
      "loss": 4.7683,
      "step": 17490
    },
    {
      "epoch": 0.35836422091618375,
      "grad_norm": 0.47632238268852234,
      "learning_rate": 0.00039368154145759453,
      "loss": 4.7411,
      "step": 17500
    },
    {
      "epoch": 0.358569000470993,
      "grad_norm": 0.2614399492740631,
      "learning_rate": 0.00039363546586124024,
      "loss": 4.7941,
      "step": 17510
    },
    {
      "epoch": 0.35877378002580224,
      "grad_norm": 0.28558123111724854,
      "learning_rate": 0.0003935892255939938,
      "loss": 4.764,
      "step": 17520
    },
    {
      "epoch": 0.3589785595806115,
      "grad_norm": 0.25479230284690857,
      "learning_rate": 0.00039354282069517873,
      "loss": 4.7787,
      "step": 17530
    },
    {
      "epoch": 0.3591833391354207,
      "grad_norm": 0.270384281873703,
      "learning_rate": 0.0003934962512042586,
      "loss": 4.7613,
      "step": 17540
    },
    {
      "epoch": 0.35938811869022996,
      "grad_norm": 0.29647931456565857,
      "learning_rate": 0.00039344951716083687,
      "loss": 4.7855,
      "step": 17550
    },
    {
      "epoch": 0.3595928982450392,
      "grad_norm": 0.29567044973373413,
      "learning_rate": 0.00039340261860465697,
      "loss": 4.7874,
      "step": 17560
    },
    {
      "epoch": 0.35979767779984845,
      "grad_norm": 0.2841157913208008,
      "learning_rate": 0.0003933555555756023,
      "loss": 4.7343,
      "step": 17570
    },
    {
      "epoch": 0.3600024573546577,
      "grad_norm": 0.292279988527298,
      "learning_rate": 0.0003933083281136959,
      "loss": 4.7657,
      "step": 17580
    },
    {
      "epoch": 0.36020723690946693,
      "grad_norm": 0.2682152986526489,
      "learning_rate": 0.00039326093625910094,
      "loss": 4.8004,
      "step": 17590
    },
    {
      "epoch": 0.36041201646427623,
      "grad_norm": 0.261972576379776,
      "learning_rate": 0.0003932133800521202,
      "loss": 4.7783,
      "step": 17600
    },
    {
      "epoch": 0.3606167960190855,
      "grad_norm": 0.26742562651634216,
      "learning_rate": 0.00039316565953319635,
      "loss": 4.725,
      "step": 17610
    },
    {
      "epoch": 0.3608215755738947,
      "grad_norm": 0.26306724548339844,
      "learning_rate": 0.0003931177747429117,
      "loss": 4.7564,
      "step": 17620
    },
    {
      "epoch": 0.36102635512870396,
      "grad_norm": 0.28656601905822754,
      "learning_rate": 0.0003930697257219882,
      "loss": 4.7925,
      "step": 17630
    },
    {
      "epoch": 0.3612311346835132,
      "grad_norm": 0.27165529131889343,
      "learning_rate": 0.0003930215125112876,
      "loss": 4.7735,
      "step": 17640
    },
    {
      "epoch": 0.36143591423832244,
      "grad_norm": 0.3619728684425354,
      "learning_rate": 0.00039297313515181137,
      "loss": 4.8043,
      "step": 17650
    },
    {
      "epoch": 0.3616406937931317,
      "grad_norm": 0.2687563896179199,
      "learning_rate": 0.00039292459368470026,
      "loss": 4.7906,
      "step": 17660
    },
    {
      "epoch": 0.3618454733479409,
      "grad_norm": 0.26408448815345764,
      "learning_rate": 0.0003928758881512349,
      "loss": 4.7839,
      "step": 17670
    },
    {
      "epoch": 0.36205025290275017,
      "grad_norm": 0.257796972990036,
      "learning_rate": 0.00039282701859283517,
      "loss": 4.8255,
      "step": 17680
    },
    {
      "epoch": 0.3622550324575594,
      "grad_norm": 0.24361667037010193,
      "learning_rate": 0.00039277798505106066,
      "loss": 4.7925,
      "step": 17690
    },
    {
      "epoch": 0.3624598120123687,
      "grad_norm": 0.27500542998313904,
      "learning_rate": 0.0003927287875676103,
      "loss": 4.8244,
      "step": 17700
    },
    {
      "epoch": 0.36266459156717795,
      "grad_norm": 0.27108463644981384,
      "learning_rate": 0.0003926794261843225,
      "loss": 4.793,
      "step": 17710
    },
    {
      "epoch": 0.3628693711219872,
      "grad_norm": 0.2628352642059326,
      "learning_rate": 0.00039262990094317496,
      "loss": 4.776,
      "step": 17720
    },
    {
      "epoch": 0.36307415067679644,
      "grad_norm": 0.26020681858062744,
      "learning_rate": 0.00039258021188628484,
      "loss": 4.7599,
      "step": 17730
    },
    {
      "epoch": 0.3632789302316057,
      "grad_norm": 0.2727605700492859,
      "learning_rate": 0.00039253035905590847,
      "loss": 4.8071,
      "step": 17740
    },
    {
      "epoch": 0.3634837097864149,
      "grad_norm": 0.28571629524230957,
      "learning_rate": 0.0003924803424944417,
      "loss": 4.7751,
      "step": 17750
    },
    {
      "epoch": 0.36368848934122416,
      "grad_norm": 0.27115678787231445,
      "learning_rate": 0.0003924301622444194,
      "loss": 4.769,
      "step": 17760
    },
    {
      "epoch": 0.3638932688960334,
      "grad_norm": 0.2659580707550049,
      "learning_rate": 0.00039237981834851557,
      "loss": 4.7609,
      "step": 17770
    },
    {
      "epoch": 0.36409804845084265,
      "grad_norm": 0.30356279015541077,
      "learning_rate": 0.0003923293108495437,
      "loss": 4.758,
      "step": 17780
    },
    {
      "epoch": 0.3643028280056519,
      "grad_norm": 0.28485971689224243,
      "learning_rate": 0.00039227863979045616,
      "loss": 4.7592,
      "step": 17790
    },
    {
      "epoch": 0.3645076075604612,
      "grad_norm": 0.25537344813346863,
      "learning_rate": 0.0003922278052143444,
      "loss": 4.7998,
      "step": 17800
    },
    {
      "epoch": 0.36471238711527043,
      "grad_norm": 0.2637980878353119,
      "learning_rate": 0.0003921768071644391,
      "loss": 4.774,
      "step": 17810
    },
    {
      "epoch": 0.3649171666700797,
      "grad_norm": 0.2536860704421997,
      "learning_rate": 0.0003921256456841099,
      "loss": 4.7985,
      "step": 17820
    },
    {
      "epoch": 0.3651219462248889,
      "grad_norm": 0.2698429226875305,
      "learning_rate": 0.00039207432081686533,
      "loss": 4.7772,
      "step": 17830
    },
    {
      "epoch": 0.36532672577969816,
      "grad_norm": 0.2777520716190338,
      "learning_rate": 0.000392022832606353,
      "loss": 4.7503,
      "step": 17840
    },
    {
      "epoch": 0.3655315053345074,
      "grad_norm": 0.2618495225906372,
      "learning_rate": 0.0003919711810963592,
      "loss": 4.8051,
      "step": 17850
    },
    {
      "epoch": 0.36573628488931664,
      "grad_norm": 0.26113712787628174,
      "learning_rate": 0.0003919193663308094,
      "loss": 4.7546,
      "step": 17860
    },
    {
      "epoch": 0.3659410644441259,
      "grad_norm": 0.2648015022277832,
      "learning_rate": 0.0003918673883537677,
      "loss": 4.7571,
      "step": 17870
    },
    {
      "epoch": 0.3661458439989351,
      "grad_norm": 0.2523532509803772,
      "learning_rate": 0.0003918152472094371,
      "loss": 4.7755,
      "step": 17880
    },
    {
      "epoch": 0.36635062355374437,
      "grad_norm": 0.26697415113449097,
      "learning_rate": 0.00039176294294215923,
      "loss": 4.7593,
      "step": 17890
    },
    {
      "epoch": 0.36655540310855367,
      "grad_norm": 0.32178786396980286,
      "learning_rate": 0.0003917104755964146,
      "loss": 4.7459,
      "step": 17900
    },
    {
      "epoch": 0.3667601826633629,
      "grad_norm": 0.2983376085758209,
      "learning_rate": 0.0003916578452168223,
      "loss": 4.7748,
      "step": 17910
    },
    {
      "epoch": 0.36696496221817215,
      "grad_norm": 0.2601611912250519,
      "learning_rate": 0.0003916050518481402,
      "loss": 4.7667,
      "step": 17920
    },
    {
      "epoch": 0.3671697417729814,
      "grad_norm": 0.29818862676620483,
      "learning_rate": 0.0003915520955352645,
      "loss": 4.8242,
      "step": 17930
    },
    {
      "epoch": 0.36737452132779064,
      "grad_norm": 0.2593207359313965,
      "learning_rate": 0.00039149897632323027,
      "loss": 4.7699,
      "step": 17940
    },
    {
      "epoch": 0.3675793008825999,
      "grad_norm": 0.29024702310562134,
      "learning_rate": 0.000391445694257211,
      "loss": 4.8051,
      "step": 17950
    },
    {
      "epoch": 0.3677840804374091,
      "grad_norm": 0.2815151810646057,
      "learning_rate": 0.00039139224938251855,
      "loss": 4.822,
      "step": 17960
    },
    {
      "epoch": 0.36798885999221836,
      "grad_norm": 0.2534072697162628,
      "learning_rate": 0.0003913386417446035,
      "loss": 4.7491,
      "step": 17970
    },
    {
      "epoch": 0.3681936395470276,
      "grad_norm": 0.2836243808269501,
      "learning_rate": 0.0003912848713890546,
      "loss": 4.7683,
      "step": 17980
    },
    {
      "epoch": 0.36839841910183685,
      "grad_norm": 0.2671874463558197,
      "learning_rate": 0.0003912309383615991,
      "loss": 4.7356,
      "step": 17990
    },
    {
      "epoch": 0.36860319865664615,
      "grad_norm": 0.2524053752422333,
      "learning_rate": 0.00039117684270810254,
      "loss": 4.7682,
      "step": 18000
    },
    {
      "epoch": 0.36860319865664615,
      "eval_loss": 4.784202575683594,
      "eval_runtime": 4.4019,
      "eval_samples_per_second": 264.887,
      "eval_steps_per_second": 33.168,
      "step": 18000
    },
    {
      "epoch": 0.3688079782114554,
      "grad_norm": 0.2831714153289795,
      "learning_rate": 0.0003911225844745688,
      "loss": 4.7654,
      "step": 18010
    },
    {
      "epoch": 0.36901275776626463,
      "grad_norm": 0.2763456404209137,
      "learning_rate": 0.00039106816370714,
      "loss": 4.7599,
      "step": 18020
    },
    {
      "epoch": 0.3692175373210739,
      "grad_norm": 0.2494124472141266,
      "learning_rate": 0.0003910135804520965,
      "loss": 4.7865,
      "step": 18030
    },
    {
      "epoch": 0.3694223168758831,
      "grad_norm": 0.2747681438922882,
      "learning_rate": 0.00039095883475585684,
      "loss": 4.757,
      "step": 18040
    },
    {
      "epoch": 0.36962709643069236,
      "grad_norm": 0.2882513403892517,
      "learning_rate": 0.0003909039266649777,
      "loss": 4.7679,
      "step": 18050
    },
    {
      "epoch": 0.3698318759855016,
      "grad_norm": 0.269535630941391,
      "learning_rate": 0.0003908488562261538,
      "loss": 4.7528,
      "step": 18060
    },
    {
      "epoch": 0.37003665554031084,
      "grad_norm": 0.2854122221469879,
      "learning_rate": 0.000390793623486218,
      "loss": 4.7805,
      "step": 18070
    },
    {
      "epoch": 0.3702414350951201,
      "grad_norm": 0.2871333658695221,
      "learning_rate": 0.0003907382284921412,
      "loss": 4.799,
      "step": 18080
    },
    {
      "epoch": 0.37044621464992933,
      "grad_norm": 0.2664596736431122,
      "learning_rate": 0.00039068267129103215,
      "loss": 4.7784,
      "step": 18090
    },
    {
      "epoch": 0.3706509942047386,
      "grad_norm": 0.29560941457748413,
      "learning_rate": 0.0003906269519301377,
      "loss": 4.7425,
      "step": 18100
    },
    {
      "epoch": 0.37085577375954787,
      "grad_norm": 0.27781563997268677,
      "learning_rate": 0.0003905710704568426,
      "loss": 4.7756,
      "step": 18110
    },
    {
      "epoch": 0.3710605533143571,
      "grad_norm": 0.25890418887138367,
      "learning_rate": 0.0003905150269186693,
      "loss": 4.7836,
      "step": 18120
    },
    {
      "epoch": 0.37126533286916635,
      "grad_norm": 0.24676479399204254,
      "learning_rate": 0.0003904588213632782,
      "loss": 4.7798,
      "step": 18130
    },
    {
      "epoch": 0.3714701124239756,
      "grad_norm": 0.2712087631225586,
      "learning_rate": 0.00039040245383846755,
      "loss": 4.7721,
      "step": 18140
    },
    {
      "epoch": 0.37167489197878484,
      "grad_norm": 0.2562553584575653,
      "learning_rate": 0.00039034592439217316,
      "loss": 4.7077,
      "step": 18150
    },
    {
      "epoch": 0.3718796715335941,
      "grad_norm": 0.25950464606285095,
      "learning_rate": 0.00039028923307246865,
      "loss": 4.7422,
      "step": 18160
    },
    {
      "epoch": 0.3720844510884033,
      "grad_norm": 0.30557316541671753,
      "learning_rate": 0.0003902323799275653,
      "loss": 4.7718,
      "step": 18170
    },
    {
      "epoch": 0.37228923064321257,
      "grad_norm": 0.26297929883003235,
      "learning_rate": 0.00039017536500581196,
      "loss": 4.7664,
      "step": 18180
    },
    {
      "epoch": 0.3724940101980218,
      "grad_norm": 0.3164907991886139,
      "learning_rate": 0.0003901181883556951,
      "loss": 4.8207,
      "step": 18190
    },
    {
      "epoch": 0.37269878975283105,
      "grad_norm": 0.269145131111145,
      "learning_rate": 0.00039006085002583874,
      "loss": 4.7828,
      "step": 18200
    },
    {
      "epoch": 0.37290356930764035,
      "grad_norm": 0.33246415853500366,
      "learning_rate": 0.00039000335006500437,
      "loss": 4.7621,
      "step": 18210
    },
    {
      "epoch": 0.3731083488624496,
      "grad_norm": 0.2733037769794464,
      "learning_rate": 0.0003899456885220909,
      "loss": 4.7872,
      "step": 18220
    },
    {
      "epoch": 0.37331312841725883,
      "grad_norm": 0.3253006935119629,
      "learning_rate": 0.0003898878654461347,
      "loss": 4.7675,
      "step": 18230
    },
    {
      "epoch": 0.3735179079720681,
      "grad_norm": 0.26410481333732605,
      "learning_rate": 0.00038982988088630954,
      "loss": 4.7431,
      "step": 18240
    },
    {
      "epoch": 0.3737226875268773,
      "grad_norm": 0.2555406987667084,
      "learning_rate": 0.00038977173489192636,
      "loss": 4.7816,
      "step": 18250
    },
    {
      "epoch": 0.37392746708168656,
      "grad_norm": 0.2523527443408966,
      "learning_rate": 0.00038971342751243363,
      "loss": 4.768,
      "step": 18260
    },
    {
      "epoch": 0.3741322466364958,
      "grad_norm": 0.30072832107543945,
      "learning_rate": 0.0003896549587974169,
      "loss": 4.7998,
      "step": 18270
    },
    {
      "epoch": 0.37433702619130504,
      "grad_norm": 0.28505706787109375,
      "learning_rate": 0.0003895963287965989,
      "loss": 4.8048,
      "step": 18280
    },
    {
      "epoch": 0.3745418057461143,
      "grad_norm": 0.3019676208496094,
      "learning_rate": 0.00038953753755983965,
      "loss": 4.8093,
      "step": 18290
    },
    {
      "epoch": 0.37474658530092353,
      "grad_norm": 0.28492453694343567,
      "learning_rate": 0.0003894785851371362,
      "loss": 4.7957,
      "step": 18300
    },
    {
      "epoch": 0.3749513648557328,
      "grad_norm": 0.2514193654060364,
      "learning_rate": 0.0003894194715786227,
      "loss": 4.7738,
      "step": 18310
    },
    {
      "epoch": 0.37515614441054207,
      "grad_norm": 0.2678600251674652,
      "learning_rate": 0.00038936019693457027,
      "loss": 4.7761,
      "step": 18320
    },
    {
      "epoch": 0.3753609239653513,
      "grad_norm": 0.25014519691467285,
      "learning_rate": 0.0003893007612553871,
      "loss": 4.7812,
      "step": 18330
    },
    {
      "epoch": 0.37556570352016055,
      "grad_norm": 0.27169695496559143,
      "learning_rate": 0.0003892411645916184,
      "loss": 4.7652,
      "step": 18340
    },
    {
      "epoch": 0.3757704830749698,
      "grad_norm": 0.25945594906806946,
      "learning_rate": 0.00038918140699394603,
      "loss": 4.7358,
      "step": 18350
    },
    {
      "epoch": 0.37597526262977904,
      "grad_norm": 0.28693246841430664,
      "learning_rate": 0.00038912148851318897,
      "loss": 4.8069,
      "step": 18360
    },
    {
      "epoch": 0.3761800421845883,
      "grad_norm": 0.255780428647995,
      "learning_rate": 0.0003890614092003029,
      "loss": 4.7505,
      "step": 18370
    },
    {
      "epoch": 0.3763848217393975,
      "grad_norm": 0.2798100411891937,
      "learning_rate": 0.00038900116910638025,
      "loss": 4.7633,
      "step": 18380
    },
    {
      "epoch": 0.37658960129420677,
      "grad_norm": 0.26866233348846436,
      "learning_rate": 0.00038894076828265017,
      "loss": 4.7489,
      "step": 18390
    },
    {
      "epoch": 0.376794380849016,
      "grad_norm": 0.24337950348854065,
      "learning_rate": 0.0003888802067804787,
      "loss": 4.7778,
      "step": 18400
    },
    {
      "epoch": 0.3769991604038253,
      "grad_norm": 0.2849843204021454,
      "learning_rate": 0.0003888194846513683,
      "loss": 4.7988,
      "step": 18410
    },
    {
      "epoch": 0.37720393995863455,
      "grad_norm": 0.25371941924095154,
      "learning_rate": 0.00038875860194695805,
      "loss": 4.7348,
      "step": 18420
    },
    {
      "epoch": 0.3774087195134438,
      "grad_norm": 0.27701646089553833,
      "learning_rate": 0.0003886975587190237,
      "loss": 4.7911,
      "step": 18430
    },
    {
      "epoch": 0.37761349906825303,
      "grad_norm": 0.2591417729854584,
      "learning_rate": 0.00038863635501947745,
      "loss": 4.7805,
      "step": 18440
    },
    {
      "epoch": 0.3778182786230623,
      "grad_norm": 0.2680375277996063,
      "learning_rate": 0.0003885749909003679,
      "loss": 4.7724,
      "step": 18450
    },
    {
      "epoch": 0.3780230581778715,
      "grad_norm": 0.2742789089679718,
      "learning_rate": 0.0003885134664138802,
      "loss": 4.7793,
      "step": 18460
    },
    {
      "epoch": 0.37822783773268076,
      "grad_norm": 0.25720226764678955,
      "learning_rate": 0.00038845178161233583,
      "loss": 4.7727,
      "step": 18470
    },
    {
      "epoch": 0.37843261728749,
      "grad_norm": 0.2753466069698334,
      "learning_rate": 0.0003883899365481925,
      "loss": 4.7827,
      "step": 18480
    },
    {
      "epoch": 0.37863739684229925,
      "grad_norm": 0.2798489034175873,
      "learning_rate": 0.00038832793127404453,
      "loss": 4.7856,
      "step": 18490
    },
    {
      "epoch": 0.3788421763971085,
      "grad_norm": 0.2578984498977661,
      "learning_rate": 0.00038826576584262195,
      "loss": 4.7821,
      "step": 18500
    },
    {
      "epoch": 0.3790469559519178,
      "grad_norm": 0.28740787506103516,
      "learning_rate": 0.0003882034403067915,
      "loss": 4.8192,
      "step": 18510
    },
    {
      "epoch": 0.37925173550672703,
      "grad_norm": 0.2980821430683136,
      "learning_rate": 0.0003881409547195558,
      "loss": 4.7659,
      "step": 18520
    },
    {
      "epoch": 0.37945651506153627,
      "grad_norm": 0.259356826543808,
      "learning_rate": 0.00038807830913405373,
      "loss": 4.7364,
      "step": 18530
    },
    {
      "epoch": 0.3796612946163455,
      "grad_norm": 0.2616250514984131,
      "learning_rate": 0.0003880155036035601,
      "loss": 4.7614,
      "step": 18540
    },
    {
      "epoch": 0.37986607417115476,
      "grad_norm": 0.45178496837615967,
      "learning_rate": 0.0003879525381814858,
      "loss": 4.7945,
      "step": 18550
    },
    {
      "epoch": 0.380070853725964,
      "grad_norm": 0.2830975651741028,
      "learning_rate": 0.0003878894129213777,
      "loss": 4.7457,
      "step": 18560
    },
    {
      "epoch": 0.38027563328077324,
      "grad_norm": 0.27111199498176575,
      "learning_rate": 0.00038782612787691866,
      "loss": 4.7598,
      "step": 18570
    },
    {
      "epoch": 0.3804804128355825,
      "grad_norm": 0.28992077708244324,
      "learning_rate": 0.00038776268310192723,
      "loss": 4.7677,
      "step": 18580
    },
    {
      "epoch": 0.3806851923903917,
      "grad_norm": 0.2773837447166443,
      "learning_rate": 0.000387699078650358,
      "loss": 4.7403,
      "step": 18590
    },
    {
      "epoch": 0.38088997194520097,
      "grad_norm": 0.23663872480392456,
      "learning_rate": 0.0003876353145763013,
      "loss": 4.7548,
      "step": 18600
    },
    {
      "epoch": 0.38109475150001026,
      "grad_norm": 0.2761427164077759,
      "learning_rate": 0.0003875713909339831,
      "loss": 4.7508,
      "step": 18610
    },
    {
      "epoch": 0.3812995310548195,
      "grad_norm": 0.2748549282550812,
      "learning_rate": 0.00038750730777776534,
      "loss": 4.7658,
      "step": 18620
    },
    {
      "epoch": 0.38150431060962875,
      "grad_norm": 0.2602144181728363,
      "learning_rate": 0.00038744306516214517,
      "loss": 4.7802,
      "step": 18630
    },
    {
      "epoch": 0.381709090164438,
      "grad_norm": 0.2753234803676605,
      "learning_rate": 0.00038737866314175577,
      "loss": 4.7783,
      "step": 18640
    },
    {
      "epoch": 0.38191386971924723,
      "grad_norm": 0.25762417912483215,
      "learning_rate": 0.0003873141017713657,
      "loss": 4.7518,
      "step": 18650
    },
    {
      "epoch": 0.3821186492740565,
      "grad_norm": 0.2581968903541565,
      "learning_rate": 0.00038724938110587894,
      "loss": 4.7841,
      "step": 18660
    },
    {
      "epoch": 0.3823234288288657,
      "grad_norm": 0.4594026505947113,
      "learning_rate": 0.00038718450120033525,
      "loss": 4.7624,
      "step": 18670
    },
    {
      "epoch": 0.38252820838367496,
      "grad_norm": 0.2587875425815582,
      "learning_rate": 0.0003871194621099093,
      "loss": 4.762,
      "step": 18680
    },
    {
      "epoch": 0.3827329879384842,
      "grad_norm": 0.31391558051109314,
      "learning_rate": 0.0003870542638899118,
      "loss": 4.7996,
      "step": 18690
    },
    {
      "epoch": 0.38293776749329345,
      "grad_norm": 0.26744288206100464,
      "learning_rate": 0.0003869889065957881,
      "loss": 4.7354,
      "step": 18700
    },
    {
      "epoch": 0.38314254704810274,
      "grad_norm": 0.26079440116882324,
      "learning_rate": 0.00038692339028311936,
      "loss": 4.7514,
      "step": 18710
    },
    {
      "epoch": 0.383347326602912,
      "grad_norm": 0.27135249972343445,
      "learning_rate": 0.0003868577150076217,
      "loss": 4.7754,
      "step": 18720
    },
    {
      "epoch": 0.38355210615772123,
      "grad_norm": 0.2575429379940033,
      "learning_rate": 0.0003867918808251464,
      "loss": 4.754,
      "step": 18730
    },
    {
      "epoch": 0.38375688571253047,
      "grad_norm": 0.2666241526603699,
      "learning_rate": 0.0003867258877916801,
      "loss": 4.7383,
      "step": 18740
    },
    {
      "epoch": 0.3839616652673397,
      "grad_norm": 0.25891196727752686,
      "learning_rate": 0.00038665973596334444,
      "loss": 4.758,
      "step": 18750
    },
    {
      "epoch": 0.38416644482214896,
      "grad_norm": 0.28033098578453064,
      "learning_rate": 0.00038659342539639586,
      "loss": 4.7603,
      "step": 18760
    },
    {
      "epoch": 0.3843712243769582,
      "grad_norm": 0.26347917318344116,
      "learning_rate": 0.00038652695614722616,
      "loss": 4.7399,
      "step": 18770
    },
    {
      "epoch": 0.38457600393176744,
      "grad_norm": 0.32026243209838867,
      "learning_rate": 0.00038646032827236186,
      "loss": 4.7516,
      "step": 18780
    },
    {
      "epoch": 0.3847807834865767,
      "grad_norm": 0.2726016044616699,
      "learning_rate": 0.0003863935418284645,
      "loss": 4.7387,
      "step": 18790
    },
    {
      "epoch": 0.3849855630413859,
      "grad_norm": 0.2759679853916168,
      "learning_rate": 0.0003863265968723303,
      "loss": 4.7852,
      "step": 18800
    },
    {
      "epoch": 0.3851903425961952,
      "grad_norm": 0.4238550662994385,
      "learning_rate": 0.0003862594934608904,
      "loss": 4.7714,
      "step": 18810
    },
    {
      "epoch": 0.38539512215100447,
      "grad_norm": 0.27191659808158875,
      "learning_rate": 0.0003861922316512108,
      "loss": 4.7935,
      "step": 18820
    },
    {
      "epoch": 0.3855999017058137,
      "grad_norm": 0.27841800451278687,
      "learning_rate": 0.0003861248115004919,
      "loss": 4.8057,
      "step": 18830
    },
    {
      "epoch": 0.38580468126062295,
      "grad_norm": 0.26735278964042664,
      "learning_rate": 0.00038605723306606914,
      "loss": 4.7604,
      "step": 18840
    },
    {
      "epoch": 0.3860094608154322,
      "grad_norm": 0.2815970182418823,
      "learning_rate": 0.00038598949640541227,
      "loss": 4.7796,
      "step": 18850
    },
    {
      "epoch": 0.38621424037024144,
      "grad_norm": 0.28123724460601807,
      "learning_rate": 0.0003859216015761257,
      "loss": 4.7221,
      "step": 18860
    },
    {
      "epoch": 0.3864190199250507,
      "grad_norm": 0.2720465660095215,
      "learning_rate": 0.0003858535486359484,
      "loss": 4.7925,
      "step": 18870
    },
    {
      "epoch": 0.3866237994798599,
      "grad_norm": 0.30233508348464966,
      "learning_rate": 0.00038578533764275363,
      "loss": 4.7629,
      "step": 18880
    },
    {
      "epoch": 0.38682857903466916,
      "grad_norm": 0.2735285460948944,
      "learning_rate": 0.0003857169686545493,
      "loss": 4.7696,
      "step": 18890
    },
    {
      "epoch": 0.3870333585894784,
      "grad_norm": 0.27527323365211487,
      "learning_rate": 0.0003856484417294775,
      "loss": 4.757,
      "step": 18900
    },
    {
      "epoch": 0.38723813814428765,
      "grad_norm": 0.2703647315502167,
      "learning_rate": 0.00038557975692581466,
      "loss": 4.776,
      "step": 18910
    },
    {
      "epoch": 0.38744291769909694,
      "grad_norm": 0.2896738052368164,
      "learning_rate": 0.0003855109143019716,
      "loss": 4.7545,
      "step": 18920
    },
    {
      "epoch": 0.3876476972539062,
      "grad_norm": 0.26229390501976013,
      "learning_rate": 0.00038544191391649325,
      "loss": 4.7466,
      "step": 18930
    },
    {
      "epoch": 0.38785247680871543,
      "grad_norm": 0.3293417990207672,
      "learning_rate": 0.00038537275582805854,
      "loss": 4.7214,
      "step": 18940
    },
    {
      "epoch": 0.38805725636352467,
      "grad_norm": 0.26354289054870605,
      "learning_rate": 0.0003853034400954809,
      "loss": 4.7472,
      "step": 18950
    },
    {
      "epoch": 0.3882620359183339,
      "grad_norm": 0.28598079085350037,
      "learning_rate": 0.0003852339667777075,
      "loss": 4.7531,
      "step": 18960
    },
    {
      "epoch": 0.38846681547314316,
      "grad_norm": 0.33052363991737366,
      "learning_rate": 0.0003851643359338196,
      "loss": 4.8054,
      "step": 18970
    },
    {
      "epoch": 0.3886715950279524,
      "grad_norm": 0.2778172194957733,
      "learning_rate": 0.0003850945476230326,
      "loss": 4.7523,
      "step": 18980
    },
    {
      "epoch": 0.38887637458276164,
      "grad_norm": 0.3006640076637268,
      "learning_rate": 0.0003850246019046955,
      "loss": 4.7239,
      "step": 18990
    },
    {
      "epoch": 0.3890811541375709,
      "grad_norm": 0.29719680547714233,
      "learning_rate": 0.00038495449883829143,
      "loss": 4.7954,
      "step": 19000
    },
    {
      "epoch": 0.3892859336923801,
      "grad_norm": 0.27522799372673035,
      "learning_rate": 0.00038488423848343715,
      "loss": 4.7821,
      "step": 19010
    },
    {
      "epoch": 0.3894907132471894,
      "grad_norm": 0.35163021087646484,
      "learning_rate": 0.0003848138208998833,
      "loss": 4.7532,
      "step": 19020
    },
    {
      "epoch": 0.38969549280199867,
      "grad_norm": 0.34216707944869995,
      "learning_rate": 0.0003847432461475142,
      "loss": 4.7659,
      "step": 19030
    },
    {
      "epoch": 0.3899002723568079,
      "grad_norm": 0.27476340532302856,
      "learning_rate": 0.00038467251428634785,
      "loss": 4.7632,
      "step": 19040
    },
    {
      "epoch": 0.39010505191161715,
      "grad_norm": 0.2805178165435791,
      "learning_rate": 0.00038460162537653586,
      "loss": 4.7819,
      "step": 19050
    },
    {
      "epoch": 0.3903098314664264,
      "grad_norm": 0.28047510981559753,
      "learning_rate": 0.0003845305794783632,
      "loss": 4.7759,
      "step": 19060
    },
    {
      "epoch": 0.39051461102123564,
      "grad_norm": 0.3435409963130951,
      "learning_rate": 0.00038445937665224864,
      "loss": 4.7387,
      "step": 19070
    },
    {
      "epoch": 0.3907193905760449,
      "grad_norm": 0.26504743099212646,
      "learning_rate": 0.0003843880169587442,
      "loss": 4.7575,
      "step": 19080
    },
    {
      "epoch": 0.3909241701308541,
      "grad_norm": 0.2697354853153229,
      "learning_rate": 0.00038431650045853545,
      "loss": 4.7719,
      "step": 19090
    },
    {
      "epoch": 0.39112894968566336,
      "grad_norm": 0.2649177610874176,
      "learning_rate": 0.00038424482721244123,
      "loss": 4.7562,
      "step": 19100
    },
    {
      "epoch": 0.3913337292404726,
      "grad_norm": 0.2683025300502777,
      "learning_rate": 0.0003841729972814136,
      "loss": 4.7737,
      "step": 19110
    },
    {
      "epoch": 0.3915385087952819,
      "grad_norm": 0.2645135819911957,
      "learning_rate": 0.00038410101072653807,
      "loss": 4.7911,
      "step": 19120
    },
    {
      "epoch": 0.39174328835009115,
      "grad_norm": 0.26734185218811035,
      "learning_rate": 0.00038402886760903314,
      "loss": 4.7697,
      "step": 19130
    },
    {
      "epoch": 0.3919480679049004,
      "grad_norm": 0.25954359769821167,
      "learning_rate": 0.00038395656799025056,
      "loss": 4.7931,
      "step": 19140
    },
    {
      "epoch": 0.39215284745970963,
      "grad_norm": 0.2955479621887207,
      "learning_rate": 0.00038388411193167516,
      "loss": 4.7503,
      "step": 19150
    },
    {
      "epoch": 0.3923576270145189,
      "grad_norm": 0.36487194895744324,
      "learning_rate": 0.0003838114994949249,
      "loss": 4.7783,
      "step": 19160
    },
    {
      "epoch": 0.3925624065693281,
      "grad_norm": 0.2788164019584656,
      "learning_rate": 0.00038373873074175053,
      "loss": 4.7736,
      "step": 19170
    },
    {
      "epoch": 0.39276718612413736,
      "grad_norm": 0.2730484902858734,
      "learning_rate": 0.0003836658057340358,
      "loss": 4.7615,
      "step": 19180
    },
    {
      "epoch": 0.3929719656789466,
      "grad_norm": 0.28349214792251587,
      "learning_rate": 0.00038359272453379747,
      "loss": 4.7403,
      "step": 19190
    },
    {
      "epoch": 0.39317674523375584,
      "grad_norm": 0.26683154702186584,
      "learning_rate": 0.000383519487203185,
      "loss": 4.7825,
      "step": 19200
    },
    {
      "epoch": 0.3933815247885651,
      "grad_norm": 0.2710789740085602,
      "learning_rate": 0.00038344609380448064,
      "loss": 4.7478,
      "step": 19210
    },
    {
      "epoch": 0.3935863043433744,
      "grad_norm": 0.26695185899734497,
      "learning_rate": 0.00038337254440009937,
      "loss": 4.7665,
      "step": 19220
    },
    {
      "epoch": 0.3937910838981836,
      "grad_norm": 0.273545503616333,
      "learning_rate": 0.0003832988390525889,
      "loss": 4.7497,
      "step": 19230
    },
    {
      "epoch": 0.39399586345299287,
      "grad_norm": 0.30231332778930664,
      "learning_rate": 0.0003832249778246295,
      "loss": 4.7653,
      "step": 19240
    },
    {
      "epoch": 0.3942006430078021,
      "grad_norm": 0.28207749128341675,
      "learning_rate": 0.00038315096077903404,
      "loss": 4.7185,
      "step": 19250
    },
    {
      "epoch": 0.39440542256261135,
      "grad_norm": 0.31809091567993164,
      "learning_rate": 0.00038307678797874783,
      "loss": 4.7526,
      "step": 19260
    },
    {
      "epoch": 0.3946102021174206,
      "grad_norm": 0.295795738697052,
      "learning_rate": 0.0003830024594868487,
      "loss": 4.7796,
      "step": 19270
    },
    {
      "epoch": 0.39481498167222984,
      "grad_norm": 0.2881593108177185,
      "learning_rate": 0.00038292797536654684,
      "loss": 4.7723,
      "step": 19280
    },
    {
      "epoch": 0.3950197612270391,
      "grad_norm": 0.286245733499527,
      "learning_rate": 0.0003828533356811848,
      "loss": 4.7548,
      "step": 19290
    },
    {
      "epoch": 0.3952245407818483,
      "grad_norm": 0.297598272562027,
      "learning_rate": 0.00038277854049423754,
      "loss": 4.7891,
      "step": 19300
    },
    {
      "epoch": 0.39542932033665756,
      "grad_norm": 0.2738468647003174,
      "learning_rate": 0.000382703589869312,
      "loss": 4.7706,
      "step": 19310
    },
    {
      "epoch": 0.39563409989146686,
      "grad_norm": 0.29014673829078674,
      "learning_rate": 0.0003826284838701476,
      "loss": 4.7256,
      "step": 19320
    },
    {
      "epoch": 0.3958388794462761,
      "grad_norm": 0.25152087211608887,
      "learning_rate": 0.0003825532225606157,
      "loss": 4.782,
      "step": 19330
    },
    {
      "epoch": 0.39604365900108535,
      "grad_norm": 0.2573102116584778,
      "learning_rate": 0.00038247780600471977,
      "loss": 4.7479,
      "step": 19340
    },
    {
      "epoch": 0.3962484385558946,
      "grad_norm": 0.29279622435569763,
      "learning_rate": 0.00038240223426659544,
      "loss": 4.747,
      "step": 19350
    },
    {
      "epoch": 0.39645321811070383,
      "grad_norm": 0.2536981403827667,
      "learning_rate": 0.00038232650741051004,
      "loss": 4.7579,
      "step": 19360
    },
    {
      "epoch": 0.3966579976655131,
      "grad_norm": 0.3073466718196869,
      "learning_rate": 0.0003822506255008631,
      "loss": 4.7325,
      "step": 19370
    },
    {
      "epoch": 0.3968627772203223,
      "grad_norm": 0.28557318449020386,
      "learning_rate": 0.00038217458860218586,
      "loss": 4.7271,
      "step": 19380
    },
    {
      "epoch": 0.39706755677513156,
      "grad_norm": 0.2675096392631531,
      "learning_rate": 0.00038209839677914143,
      "loss": 4.7627,
      "step": 19390
    },
    {
      "epoch": 0.3972723363299408,
      "grad_norm": 0.25349166989326477,
      "learning_rate": 0.0003820220500965245,
      "loss": 4.7664,
      "step": 19400
    },
    {
      "epoch": 0.39747711588475004,
      "grad_norm": 0.27463775873184204,
      "learning_rate": 0.0003819455486192617,
      "loss": 4.7561,
      "step": 19410
    },
    {
      "epoch": 0.39768189543955934,
      "grad_norm": 0.26222658157348633,
      "learning_rate": 0.00038186889241241116,
      "loss": 4.7602,
      "step": 19420
    },
    {
      "epoch": 0.3978866749943686,
      "grad_norm": 0.2814188301563263,
      "learning_rate": 0.00038179208154116265,
      "loss": 4.7413,
      "step": 19430
    },
    {
      "epoch": 0.3980914545491778,
      "grad_norm": 0.25720828771591187,
      "learning_rate": 0.0003817151160708373,
      "loss": 4.7399,
      "step": 19440
    },
    {
      "epoch": 0.39829623410398707,
      "grad_norm": 0.2466021180152893,
      "learning_rate": 0.00038163799606688803,
      "loss": 4.7245,
      "step": 19450
    },
    {
      "epoch": 0.3985010136587963,
      "grad_norm": 0.2728370428085327,
      "learning_rate": 0.00038156072159489893,
      "loss": 4.7553,
      "step": 19460
    },
    {
      "epoch": 0.39870579321360555,
      "grad_norm": 0.26696377992630005,
      "learning_rate": 0.00038148329272058557,
      "loss": 4.789,
      "step": 19470
    },
    {
      "epoch": 0.3989105727684148,
      "grad_norm": 0.25925275683403015,
      "learning_rate": 0.0003814057095097947,
      "loss": 4.7449,
      "step": 19480
    },
    {
      "epoch": 0.39911535232322404,
      "grad_norm": 0.3272514045238495,
      "learning_rate": 0.0003813279720285045,
      "loss": 4.7465,
      "step": 19490
    },
    {
      "epoch": 0.3993201318780333,
      "grad_norm": 0.2468366175889969,
      "learning_rate": 0.0003812500803428241,
      "loss": 4.7569,
      "step": 19500
    },
    {
      "epoch": 0.3995249114328425,
      "grad_norm": 0.26700812578201294,
      "learning_rate": 0.000381172034518994,
      "loss": 4.7355,
      "step": 19510
    },
    {
      "epoch": 0.3997296909876518,
      "grad_norm": 0.2623588442802429,
      "learning_rate": 0.0003810938346233858,
      "loss": 4.7481,
      "step": 19520
    },
    {
      "epoch": 0.39993447054246106,
      "grad_norm": 0.2693066895008087,
      "learning_rate": 0.00038101548072250185,
      "loss": 4.7676,
      "step": 19530
    },
    {
      "epoch": 0.4001392500972703,
      "grad_norm": 0.267498254776001,
      "learning_rate": 0.0003809369728829757,
      "loss": 4.7471,
      "step": 19540
    },
    {
      "epoch": 0.40034402965207955,
      "grad_norm": 0.2677249312400818,
      "learning_rate": 0.0003808583111715717,
      "loss": 4.7504,
      "step": 19550
    },
    {
      "epoch": 0.4005488092068888,
      "grad_norm": 0.27942562103271484,
      "learning_rate": 0.0003807794956551853,
      "loss": 4.7371,
      "step": 19560
    },
    {
      "epoch": 0.40075358876169803,
      "grad_norm": 0.3115752637386322,
      "learning_rate": 0.00038070052640084233,
      "loss": 4.7533,
      "step": 19570
    },
    {
      "epoch": 0.4009583683165073,
      "grad_norm": 0.32602185010910034,
      "learning_rate": 0.0003806214034756997,
      "loss": 4.7557,
      "step": 19580
    },
    {
      "epoch": 0.4011631478713165,
      "grad_norm": 0.2551416754722595,
      "learning_rate": 0.0003805421269470449,
      "loss": 4.7517,
      "step": 19590
    },
    {
      "epoch": 0.40136792742612576,
      "grad_norm": 0.29488468170166016,
      "learning_rate": 0.000380462696882296,
      "loss": 4.7618,
      "step": 19600
    },
    {
      "epoch": 0.401572706980935,
      "grad_norm": 0.2805251181125641,
      "learning_rate": 0.0003803831133490016,
      "loss": 4.7674,
      "step": 19610
    },
    {
      "epoch": 0.40177748653574424,
      "grad_norm": 0.25679463148117065,
      "learning_rate": 0.00038030337641484105,
      "loss": 4.7475,
      "step": 19620
    },
    {
      "epoch": 0.40198226609055354,
      "grad_norm": 0.2735605239868164,
      "learning_rate": 0.0003802234861476239,
      "loss": 4.7615,
      "step": 19630
    },
    {
      "epoch": 0.4021870456453628,
      "grad_norm": 0.2695820927619934,
      "learning_rate": 0.0003801434426152902,
      "loss": 4.7394,
      "step": 19640
    },
    {
      "epoch": 0.402391825200172,
      "grad_norm": 0.2904840409755707,
      "learning_rate": 0.0003800632458859103,
      "loss": 4.7193,
      "step": 19650
    },
    {
      "epoch": 0.40259660475498127,
      "grad_norm": 0.2951556146144867,
      "learning_rate": 0.000379982896027685,
      "loss": 4.7579,
      "step": 19660
    },
    {
      "epoch": 0.4028013843097905,
      "grad_norm": 0.27108797430992126,
      "learning_rate": 0.000379902393108945,
      "loss": 4.7922,
      "step": 19670
    },
    {
      "epoch": 0.40300616386459975,
      "grad_norm": 0.32287734746932983,
      "learning_rate": 0.00037982173719815145,
      "loss": 4.7191,
      "step": 19680
    },
    {
      "epoch": 0.403210943419409,
      "grad_norm": 0.2531939744949341,
      "learning_rate": 0.0003797409283638955,
      "loss": 4.7648,
      "step": 19690
    },
    {
      "epoch": 0.40341572297421824,
      "grad_norm": 0.25351694226264954,
      "learning_rate": 0.0003796599666748984,
      "loss": 4.7455,
      "step": 19700
    },
    {
      "epoch": 0.4036205025290275,
      "grad_norm": 0.27533668279647827,
      "learning_rate": 0.00037957885220001123,
      "loss": 4.7252,
      "step": 19710
    },
    {
      "epoch": 0.4038252820838367,
      "grad_norm": 0.2773134410381317,
      "learning_rate": 0.0003794975850082152,
      "loss": 4.7253,
      "step": 19720
    },
    {
      "epoch": 0.404030061638646,
      "grad_norm": 0.27424198389053345,
      "learning_rate": 0.00037941616516862133,
      "loss": 4.7617,
      "step": 19730
    },
    {
      "epoch": 0.40423484119345526,
      "grad_norm": 0.28016701340675354,
      "learning_rate": 0.0003793345927504704,
      "loss": 4.7635,
      "step": 19740
    },
    {
      "epoch": 0.4044396207482645,
      "grad_norm": 0.2677452862262726,
      "learning_rate": 0.00037925286782313297,
      "loss": 4.7704,
      "step": 19750
    },
    {
      "epoch": 0.40464440030307375,
      "grad_norm": 0.28092268109321594,
      "learning_rate": 0.00037917099045610934,
      "loss": 4.7673,
      "step": 19760
    },
    {
      "epoch": 0.404849179857883,
      "grad_norm": 0.25641924142837524,
      "learning_rate": 0.00037908896071902937,
      "loss": 4.7884,
      "step": 19770
    },
    {
      "epoch": 0.40505395941269223,
      "grad_norm": 0.29772433638572693,
      "learning_rate": 0.00037900677868165256,
      "loss": 4.6987,
      "step": 19780
    },
    {
      "epoch": 0.4052587389675015,
      "grad_norm": 0.2560875713825226,
      "learning_rate": 0.000378924444413868,
      "loss": 4.7711,
      "step": 19790
    },
    {
      "epoch": 0.4054635185223107,
      "grad_norm": 0.30855676531791687,
      "learning_rate": 0.00037884195798569396,
      "loss": 4.749,
      "step": 19800
    },
    {
      "epoch": 0.40566829807711996,
      "grad_norm": 0.28452619910240173,
      "learning_rate": 0.0003787593194672785,
      "loss": 4.7516,
      "step": 19810
    },
    {
      "epoch": 0.4058730776319292,
      "grad_norm": 0.2876434028148651,
      "learning_rate": 0.00037867652892889866,
      "loss": 4.732,
      "step": 19820
    },
    {
      "epoch": 0.4060778571867385,
      "grad_norm": 0.27363520860671997,
      "learning_rate": 0.00037859358644096104,
      "loss": 4.7889,
      "step": 19830
    },
    {
      "epoch": 0.40628263674154774,
      "grad_norm": 0.35006874799728394,
      "learning_rate": 0.00037851049207400126,
      "loss": 4.7257,
      "step": 19840
    },
    {
      "epoch": 0.406487416296357,
      "grad_norm": 0.2801511585712433,
      "learning_rate": 0.0003784272458986843,
      "loss": 4.7663,
      "step": 19850
    },
    {
      "epoch": 0.4066921958511662,
      "grad_norm": 0.2624078691005707,
      "learning_rate": 0.000378343847985804,
      "loss": 4.7674,
      "step": 19860
    },
    {
      "epoch": 0.40689697540597547,
      "grad_norm": 0.2904002070426941,
      "learning_rate": 0.0003782602984062834,
      "loss": 4.7124,
      "step": 19870
    },
    {
      "epoch": 0.4071017549607847,
      "grad_norm": 0.2745887041091919,
      "learning_rate": 0.0003781765972311745,
      "loss": 4.7743,
      "step": 19880
    },
    {
      "epoch": 0.40730653451559395,
      "grad_norm": 0.2452814280986786,
      "learning_rate": 0.00037809274453165823,
      "loss": 4.707,
      "step": 19890
    },
    {
      "epoch": 0.4075113140704032,
      "grad_norm": 0.32110917568206787,
      "learning_rate": 0.0003780087403790442,
      "loss": 4.7431,
      "step": 19900
    },
    {
      "epoch": 0.40771609362521244,
      "grad_norm": 0.24623283743858337,
      "learning_rate": 0.00037792458484477116,
      "loss": 4.7738,
      "step": 19910
    },
    {
      "epoch": 0.4079208731800217,
      "grad_norm": 0.26000604033470154,
      "learning_rate": 0.00037784027800040633,
      "loss": 4.7656,
      "step": 19920
    },
    {
      "epoch": 0.408125652734831,
      "grad_norm": 0.2891079783439636,
      "learning_rate": 0.0003777558199176457,
      "loss": 4.7705,
      "step": 19930
    },
    {
      "epoch": 0.4083304322896402,
      "grad_norm": 0.28778308629989624,
      "learning_rate": 0.0003776712106683138,
      "loss": 4.7482,
      "step": 19940
    },
    {
      "epoch": 0.40853521184444946,
      "grad_norm": 0.2763589322566986,
      "learning_rate": 0.00037758645032436386,
      "loss": 4.7377,
      "step": 19950
    },
    {
      "epoch": 0.4087399913992587,
      "grad_norm": 0.26450416445732117,
      "learning_rate": 0.00037750153895787744,
      "loss": 4.7238,
      "step": 19960
    },
    {
      "epoch": 0.40894477095406795,
      "grad_norm": 0.27174514532089233,
      "learning_rate": 0.0003774164766410646,
      "loss": 4.7706,
      "step": 19970
    },
    {
      "epoch": 0.4091495505088772,
      "grad_norm": 0.26067888736724854,
      "learning_rate": 0.0003773312634462639,
      "loss": 4.7752,
      "step": 19980
    },
    {
      "epoch": 0.40935433006368643,
      "grad_norm": 0.2728261649608612,
      "learning_rate": 0.000377245899445942,
      "loss": 4.7563,
      "step": 19990
    },
    {
      "epoch": 0.4095591096184957,
      "grad_norm": 0.3048381209373474,
      "learning_rate": 0.00037716038471269386,
      "loss": 4.7654,
      "step": 20000
    },
    {
      "epoch": 0.4095591096184957,
      "eval_loss": 4.762139797210693,
      "eval_runtime": 4.4093,
      "eval_samples_per_second": 264.44,
      "eval_steps_per_second": 33.112,
      "step": 20000
    },
    {
      "epoch": 0.4097638891733049,
      "grad_norm": 0.31729310750961304,
      "learning_rate": 0.00037707471931924276,
      "loss": 4.7428,
      "step": 20010
    },
    {
      "epoch": 0.40996866872811416,
      "grad_norm": 0.25164294242858887,
      "learning_rate": 0.00037698890333843994,
      "loss": 4.7363,
      "step": 20020
    },
    {
      "epoch": 0.41017344828292346,
      "grad_norm": 0.256254643201828,
      "learning_rate": 0.0003769029368432648,
      "loss": 4.7656,
      "step": 20030
    },
    {
      "epoch": 0.4103782278377327,
      "grad_norm": 0.27743199467658997,
      "learning_rate": 0.00037681681990682465,
      "loss": 4.7731,
      "step": 20040
    },
    {
      "epoch": 0.41058300739254194,
      "grad_norm": 0.2577210068702698,
      "learning_rate": 0.0003767305526023549,
      "loss": 4.7563,
      "step": 20050
    },
    {
      "epoch": 0.4107877869473512,
      "grad_norm": 0.28428512811660767,
      "learning_rate": 0.0003766441350032187,
      "loss": 4.7429,
      "step": 20060
    },
    {
      "epoch": 0.41099256650216043,
      "grad_norm": 0.2780125141143799,
      "learning_rate": 0.000376557567182907,
      "loss": 4.7418,
      "step": 20070
    },
    {
      "epoch": 0.41119734605696967,
      "grad_norm": 0.2464691400527954,
      "learning_rate": 0.0003764708492150385,
      "loss": 4.7388,
      "step": 20080
    },
    {
      "epoch": 0.4114021256117789,
      "grad_norm": 0.27678778767585754,
      "learning_rate": 0.0003763839811733598,
      "loss": 4.7563,
      "step": 20090
    },
    {
      "epoch": 0.41160690516658816,
      "grad_norm": 0.2662733495235443,
      "learning_rate": 0.00037629696313174485,
      "loss": 4.7278,
      "step": 20100
    },
    {
      "epoch": 0.4118116847213974,
      "grad_norm": 0.33604541420936584,
      "learning_rate": 0.0003762097951641952,
      "loss": 4.6944,
      "step": 20110
    },
    {
      "epoch": 0.41201646427620664,
      "grad_norm": 0.3161959946155548,
      "learning_rate": 0.0003761224773448401,
      "loss": 4.7477,
      "step": 20120
    },
    {
      "epoch": 0.41222124383101594,
      "grad_norm": 0.3201931416988373,
      "learning_rate": 0.0003760350097479361,
      "loss": 4.7126,
      "step": 20130
    },
    {
      "epoch": 0.4124260233858252,
      "grad_norm": 0.27246227860450745,
      "learning_rate": 0.0003759473924478671,
      "loss": 4.722,
      "step": 20140
    },
    {
      "epoch": 0.4126308029406344,
      "grad_norm": 0.2522353231906891,
      "learning_rate": 0.00037585962551914433,
      "loss": 4.7478,
      "step": 20150
    },
    {
      "epoch": 0.41283558249544366,
      "grad_norm": 0.29608771204948425,
      "learning_rate": 0.0003757717090364062,
      "loss": 4.7393,
      "step": 20160
    },
    {
      "epoch": 0.4130403620502529,
      "grad_norm": 0.2863655686378479,
      "learning_rate": 0.0003756836430744185,
      "loss": 4.7444,
      "step": 20170
    },
    {
      "epoch": 0.41324514160506215,
      "grad_norm": 0.28394028544425964,
      "learning_rate": 0.0003755954277080741,
      "loss": 4.7254,
      "step": 20180
    },
    {
      "epoch": 0.4134499211598714,
      "grad_norm": 0.33804798126220703,
      "learning_rate": 0.0003755070630123926,
      "loss": 4.7376,
      "step": 20190
    },
    {
      "epoch": 0.41365470071468063,
      "grad_norm": 0.3279488682746887,
      "learning_rate": 0.00037541854906252106,
      "loss": 4.7475,
      "step": 20200
    },
    {
      "epoch": 0.4138594802694899,
      "grad_norm": 0.29185786843299866,
      "learning_rate": 0.0003753298859337331,
      "loss": 4.7063,
      "step": 20210
    },
    {
      "epoch": 0.4140642598242991,
      "grad_norm": 0.3181179463863373,
      "learning_rate": 0.00037524107370142946,
      "loss": 4.6854,
      "step": 20220
    },
    {
      "epoch": 0.4142690393791084,
      "grad_norm": 0.2748674750328064,
      "learning_rate": 0.0003751521124411375,
      "loss": 4.7508,
      "step": 20230
    },
    {
      "epoch": 0.41447381893391766,
      "grad_norm": 0.26090162992477417,
      "learning_rate": 0.00037506300222851145,
      "loss": 4.7094,
      "step": 20240
    },
    {
      "epoch": 0.4146785984887269,
      "grad_norm": 0.2858532965183258,
      "learning_rate": 0.00037497374313933204,
      "loss": 4.7219,
      "step": 20250
    },
    {
      "epoch": 0.41488337804353614,
      "grad_norm": 0.27587875723838806,
      "learning_rate": 0.00037488433524950684,
      "loss": 4.7465,
      "step": 20260
    },
    {
      "epoch": 0.4150881575983454,
      "grad_norm": 0.2696737051010132,
      "learning_rate": 0.0003747947786350698,
      "loss": 4.7344,
      "step": 20270
    },
    {
      "epoch": 0.41529293715315463,
      "grad_norm": 0.2946103513240814,
      "learning_rate": 0.0003747050733721814,
      "loss": 4.7315,
      "step": 20280
    },
    {
      "epoch": 0.41549771670796387,
      "grad_norm": 0.2917758524417877,
      "learning_rate": 0.0003746152195371284,
      "loss": 4.7215,
      "step": 20290
    },
    {
      "epoch": 0.4157024962627731,
      "grad_norm": 0.2587776184082031,
      "learning_rate": 0.0003745252172063242,
      "loss": 4.7572,
      "step": 20300
    },
    {
      "epoch": 0.41590727581758236,
      "grad_norm": 0.257627010345459,
      "learning_rate": 0.0003744350664563083,
      "loss": 4.7421,
      "step": 20310
    },
    {
      "epoch": 0.4161120553723916,
      "grad_norm": 0.24898594617843628,
      "learning_rate": 0.00037434476736374626,
      "loss": 4.7361,
      "step": 20320
    },
    {
      "epoch": 0.41631683492720084,
      "grad_norm": 0.3044833838939667,
      "learning_rate": 0.0003742543200054301,
      "loss": 4.7529,
      "step": 20330
    },
    {
      "epoch": 0.41652161448201014,
      "grad_norm": 0.26270875334739685,
      "learning_rate": 0.0003741637244582778,
      "loss": 4.7394,
      "step": 20340
    },
    {
      "epoch": 0.4167263940368194,
      "grad_norm": 0.2890133559703827,
      "learning_rate": 0.00037407298079933333,
      "loss": 4.7101,
      "step": 20350
    },
    {
      "epoch": 0.4169311735916286,
      "grad_norm": 0.2605116069316864,
      "learning_rate": 0.0003739820891057666,
      "loss": 4.7135,
      "step": 20360
    },
    {
      "epoch": 0.41713595314643787,
      "grad_norm": 0.2764221131801605,
      "learning_rate": 0.00037389104945487347,
      "loss": 4.7378,
      "step": 20370
    },
    {
      "epoch": 0.4173407327012471,
      "grad_norm": 0.3196144700050354,
      "learning_rate": 0.00037379986192407564,
      "loss": 4.7522,
      "step": 20380
    },
    {
      "epoch": 0.41754551225605635,
      "grad_norm": 0.2779579758644104,
      "learning_rate": 0.0003737085265909205,
      "loss": 4.7347,
      "step": 20390
    },
    {
      "epoch": 0.4177502918108656,
      "grad_norm": 0.27223050594329834,
      "learning_rate": 0.0003736170435330812,
      "loss": 4.7316,
      "step": 20400
    },
    {
      "epoch": 0.41795507136567484,
      "grad_norm": 0.2609656751155853,
      "learning_rate": 0.00037352541282835645,
      "loss": 4.744,
      "step": 20410
    },
    {
      "epoch": 0.4181598509204841,
      "grad_norm": 0.27855393290519714,
      "learning_rate": 0.0003734336345546706,
      "loss": 4.7867,
      "step": 20420
    },
    {
      "epoch": 0.4183646304752933,
      "grad_norm": 0.30180808901786804,
      "learning_rate": 0.0003733417087900734,
      "loss": 4.7543,
      "step": 20430
    },
    {
      "epoch": 0.4185694100301026,
      "grad_norm": 0.2689778506755829,
      "learning_rate": 0.00037324963561274015,
      "loss": 4.7679,
      "step": 20440
    },
    {
      "epoch": 0.41877418958491186,
      "grad_norm": 0.2800348699092865,
      "learning_rate": 0.00037315741510097144,
      "loss": 4.7627,
      "step": 20450
    },
    {
      "epoch": 0.4189789691397211,
      "grad_norm": 0.2967687249183655,
      "learning_rate": 0.0003730650473331931,
      "loss": 4.7548,
      "step": 20460
    },
    {
      "epoch": 0.41918374869453034,
      "grad_norm": 0.27397188544273376,
      "learning_rate": 0.00037297253238795626,
      "loss": 4.7223,
      "step": 20470
    },
    {
      "epoch": 0.4193885282493396,
      "grad_norm": 0.26289302110671997,
      "learning_rate": 0.0003728798703439373,
      "loss": 4.7291,
      "step": 20480
    },
    {
      "epoch": 0.41959330780414883,
      "grad_norm": 0.24988263845443726,
      "learning_rate": 0.0003727870612799374,
      "loss": 4.7163,
      "step": 20490
    },
    {
      "epoch": 0.41979808735895807,
      "grad_norm": 0.2630736529827118,
      "learning_rate": 0.0003726941052748832,
      "loss": 4.7577,
      "step": 20500
    },
    {
      "epoch": 0.4200028669137673,
      "grad_norm": 0.2840738892555237,
      "learning_rate": 0.0003726010024078259,
      "loss": 4.7431,
      "step": 20510
    },
    {
      "epoch": 0.42020764646857656,
      "grad_norm": 0.36802077293395996,
      "learning_rate": 0.0003725077527579418,
      "loss": 4.7876,
      "step": 20520
    },
    {
      "epoch": 0.4204124260233858,
      "grad_norm": 0.2600573003292084,
      "learning_rate": 0.0003724143564045321,
      "loss": 4.8069,
      "step": 20530
    },
    {
      "epoch": 0.4206172055781951,
      "grad_norm": 0.2773413062095642,
      "learning_rate": 0.0003723208134270225,
      "loss": 4.7437,
      "step": 20540
    },
    {
      "epoch": 0.42082198513300434,
      "grad_norm": 0.2827748954296112,
      "learning_rate": 0.00037222712390496353,
      "loss": 4.7335,
      "step": 20550
    },
    {
      "epoch": 0.4210267646878136,
      "grad_norm": 0.2878071963787079,
      "learning_rate": 0.0003721332879180305,
      "loss": 4.7648,
      "step": 20560
    },
    {
      "epoch": 0.4212315442426228,
      "grad_norm": 0.2613888680934906,
      "learning_rate": 0.000372039305546023,
      "loss": 4.7907,
      "step": 20570
    },
    {
      "epoch": 0.42143632379743207,
      "grad_norm": 0.2486087679862976,
      "learning_rate": 0.0003719451768688653,
      "loss": 4.7224,
      "step": 20580
    },
    {
      "epoch": 0.4216411033522413,
      "grad_norm": 0.3080863356590271,
      "learning_rate": 0.000371850901966606,
      "loss": 4.7483,
      "step": 20590
    },
    {
      "epoch": 0.42184588290705055,
      "grad_norm": 0.29384005069732666,
      "learning_rate": 0.00037175648091941805,
      "loss": 4.7485,
      "step": 20600
    },
    {
      "epoch": 0.4220506624618598,
      "grad_norm": 0.2579139471054077,
      "learning_rate": 0.0003716619138075988,
      "loss": 4.753,
      "step": 20610
    },
    {
      "epoch": 0.42225544201666904,
      "grad_norm": 0.3039598762989044,
      "learning_rate": 0.00037156720071156965,
      "loss": 4.7608,
      "step": 20620
    },
    {
      "epoch": 0.4224602215714783,
      "grad_norm": 0.28961753845214844,
      "learning_rate": 0.00037147234171187627,
      "loss": 4.7226,
      "step": 20630
    },
    {
      "epoch": 0.4226650011262876,
      "grad_norm": 0.2628253996372223,
      "learning_rate": 0.0003713773368891883,
      "loss": 4.7403,
      "step": 20640
    },
    {
      "epoch": 0.4228697806810968,
      "grad_norm": 0.26308774948120117,
      "learning_rate": 0.0003712821863242996,
      "loss": 4.7542,
      "step": 20650
    },
    {
      "epoch": 0.42307456023590606,
      "grad_norm": 0.27785658836364746,
      "learning_rate": 0.00037118689009812766,
      "loss": 4.7289,
      "step": 20660
    },
    {
      "epoch": 0.4232793397907153,
      "grad_norm": 0.31840765476226807,
      "learning_rate": 0.0003710914482917141,
      "loss": 4.7255,
      "step": 20670
    },
    {
      "epoch": 0.42348411934552455,
      "grad_norm": 0.26973164081573486,
      "learning_rate": 0.0003709958609862243,
      "loss": 4.7273,
      "step": 20680
    },
    {
      "epoch": 0.4236888989003338,
      "grad_norm": 0.2843843698501587,
      "learning_rate": 0.0003709001282629472,
      "loss": 4.7286,
      "step": 20690
    },
    {
      "epoch": 0.42389367845514303,
      "grad_norm": 0.2571497857570648,
      "learning_rate": 0.00037080425020329566,
      "loss": 4.7225,
      "step": 20700
    },
    {
      "epoch": 0.4240984580099523,
      "grad_norm": 0.27932247519493103,
      "learning_rate": 0.00037070822688880595,
      "loss": 4.751,
      "step": 20710
    },
    {
      "epoch": 0.4243032375647615,
      "grad_norm": 0.30829671025276184,
      "learning_rate": 0.0003706120584011379,
      "loss": 4.7353,
      "step": 20720
    },
    {
      "epoch": 0.42450801711957076,
      "grad_norm": 0.2808021903038025,
      "learning_rate": 0.00037051574482207486,
      "loss": 4.7434,
      "step": 20730
    },
    {
      "epoch": 0.42471279667438006,
      "grad_norm": 0.30690625309944153,
      "learning_rate": 0.00037041928623352357,
      "loss": 4.7524,
      "step": 20740
    },
    {
      "epoch": 0.4249175762291893,
      "grad_norm": 0.25853148102760315,
      "learning_rate": 0.000370322682717514,
      "loss": 4.7344,
      "step": 20750
    },
    {
      "epoch": 0.42512235578399854,
      "grad_norm": 0.2632678151130676,
      "learning_rate": 0.00037022593435619947,
      "loss": 4.7313,
      "step": 20760
    },
    {
      "epoch": 0.4253271353388078,
      "grad_norm": 0.26891714334487915,
      "learning_rate": 0.00037012904123185644,
      "loss": 4.7425,
      "step": 20770
    },
    {
      "epoch": 0.425531914893617,
      "grad_norm": 0.2848103642463684,
      "learning_rate": 0.00037003200342688433,
      "loss": 4.7262,
      "step": 20780
    },
    {
      "epoch": 0.42573669444842627,
      "grad_norm": 0.26982730627059937,
      "learning_rate": 0.00036993482102380595,
      "loss": 4.7739,
      "step": 20790
    },
    {
      "epoch": 0.4259414740032355,
      "grad_norm": 0.2797873914241791,
      "learning_rate": 0.0003698374941052667,
      "loss": 4.7305,
      "step": 20800
    },
    {
      "epoch": 0.42614625355804475,
      "grad_norm": 0.24877497553825378,
      "learning_rate": 0.0003697400227540351,
      "loss": 4.7422,
      "step": 20810
    },
    {
      "epoch": 0.426351033112854,
      "grad_norm": 0.2713570296764374,
      "learning_rate": 0.0003696424070530025,
      "loss": 4.7698,
      "step": 20820
    },
    {
      "epoch": 0.42655581266766324,
      "grad_norm": 0.2632727324962616,
      "learning_rate": 0.00036954464708518293,
      "loss": 4.7068,
      "step": 20830
    },
    {
      "epoch": 0.42676059222247253,
      "grad_norm": 0.26218199729919434,
      "learning_rate": 0.0003694467429337131,
      "loss": 4.7559,
      "step": 20840
    },
    {
      "epoch": 0.4269653717772818,
      "grad_norm": 0.260280042886734,
      "learning_rate": 0.0003693486946818523,
      "loss": 4.7461,
      "step": 20850
    },
    {
      "epoch": 0.427170151332091,
      "grad_norm": 0.2665586471557617,
      "learning_rate": 0.00036925050241298253,
      "loss": 4.7795,
      "step": 20860
    },
    {
      "epoch": 0.42737493088690026,
      "grad_norm": 0.2628210186958313,
      "learning_rate": 0.00036915216621060813,
      "loss": 4.745,
      "step": 20870
    },
    {
      "epoch": 0.4275797104417095,
      "grad_norm": 0.29638341069221497,
      "learning_rate": 0.00036905368615835593,
      "loss": 4.7348,
      "step": 20880
    },
    {
      "epoch": 0.42778448999651875,
      "grad_norm": 0.35531070828437805,
      "learning_rate": 0.00036895506233997495,
      "loss": 4.7324,
      "step": 20890
    },
    {
      "epoch": 0.427989269551328,
      "grad_norm": 0.2501230537891388,
      "learning_rate": 0.0003688562948393366,
      "loss": 4.7334,
      "step": 20900
    },
    {
      "epoch": 0.42819404910613723,
      "grad_norm": 0.24845455586910248,
      "learning_rate": 0.0003687573837404345,
      "loss": 4.7743,
      "step": 20910
    },
    {
      "epoch": 0.4283988286609465,
      "grad_norm": 0.2809300720691681,
      "learning_rate": 0.0003686583291273842,
      "loss": 4.7217,
      "step": 20920
    },
    {
      "epoch": 0.4286036082157557,
      "grad_norm": 0.28208544850349426,
      "learning_rate": 0.0003685591310844235,
      "loss": 4.7296,
      "step": 20930
    },
    {
      "epoch": 0.428808387770565,
      "grad_norm": 0.2754552364349365,
      "learning_rate": 0.00036845978969591214,
      "loss": 4.723,
      "step": 20940
    },
    {
      "epoch": 0.42901316732537426,
      "grad_norm": 0.26814135909080505,
      "learning_rate": 0.0003683603050463317,
      "loss": 4.7164,
      "step": 20950
    },
    {
      "epoch": 0.4292179468801835,
      "grad_norm": 0.9360597133636475,
      "learning_rate": 0.00036826067722028563,
      "loss": 4.6658,
      "step": 20960
    },
    {
      "epoch": 0.42942272643499274,
      "grad_norm": 0.2701377272605896,
      "learning_rate": 0.00036816090630249906,
      "loss": 4.7398,
      "step": 20970
    },
    {
      "epoch": 0.429627505989802,
      "grad_norm": 0.2618829905986786,
      "learning_rate": 0.000368060992377819,
      "loss": 4.7182,
      "step": 20980
    },
    {
      "epoch": 0.4298322855446112,
      "grad_norm": 0.3635317087173462,
      "learning_rate": 0.0003679609355312137,
      "loss": 4.7688,
      "step": 20990
    },
    {
      "epoch": 0.43003706509942047,
      "grad_norm": 0.26796573400497437,
      "learning_rate": 0.0003678607358477735,
      "loss": 4.7562,
      "step": 21000
    },
    {
      "epoch": 0.4302418446542297,
      "grad_norm": 0.2521805763244629,
      "learning_rate": 0.0003677603934127098,
      "loss": 4.6947,
      "step": 21010
    },
    {
      "epoch": 0.43044662420903895,
      "grad_norm": 0.26770153641700745,
      "learning_rate": 0.0003676599083113555,
      "loss": 4.7798,
      "step": 21020
    },
    {
      "epoch": 0.4306514037638482,
      "grad_norm": 0.28818097710609436,
      "learning_rate": 0.00036755928062916476,
      "loss": 4.7426,
      "step": 21030
    },
    {
      "epoch": 0.4308561833186575,
      "grad_norm": 0.3475240170955658,
      "learning_rate": 0.00036745851045171326,
      "loss": 4.7895,
      "step": 21040
    },
    {
      "epoch": 0.43106096287346674,
      "grad_norm": 0.3224576711654663,
      "learning_rate": 0.0003673575978646975,
      "loss": 4.7569,
      "step": 21050
    },
    {
      "epoch": 0.431265742428276,
      "grad_norm": 0.2977074682712555,
      "learning_rate": 0.00036725654295393533,
      "loss": 4.7342,
      "step": 21060
    },
    {
      "epoch": 0.4314705219830852,
      "grad_norm": 0.24770216643810272,
      "learning_rate": 0.0003671553458053655,
      "loss": 4.7489,
      "step": 21070
    },
    {
      "epoch": 0.43167530153789446,
      "grad_norm": 0.3409057557582855,
      "learning_rate": 0.00036705400650504787,
      "loss": 4.7374,
      "step": 21080
    },
    {
      "epoch": 0.4318800810927037,
      "grad_norm": 0.26433828473091125,
      "learning_rate": 0.00036695252513916314,
      "loss": 4.7437,
      "step": 21090
    },
    {
      "epoch": 0.43208486064751295,
      "grad_norm": 0.2731941044330597,
      "learning_rate": 0.00036685090179401273,
      "loss": 4.6672,
      "step": 21100
    },
    {
      "epoch": 0.4322896402023222,
      "grad_norm": 0.28759586811065674,
      "learning_rate": 0.00036674913655601875,
      "loss": 4.7389,
      "step": 21110
    },
    {
      "epoch": 0.43249441975713143,
      "grad_norm": 0.2854158580303192,
      "learning_rate": 0.00036664722951172424,
      "loss": 4.762,
      "step": 21120
    },
    {
      "epoch": 0.4326991993119407,
      "grad_norm": 0.2920982241630554,
      "learning_rate": 0.0003665451807477927,
      "loss": 4.7418,
      "step": 21130
    },
    {
      "epoch": 0.4329039788667499,
      "grad_norm": 0.2870182693004608,
      "learning_rate": 0.00036644299035100804,
      "loss": 4.7353,
      "step": 21140
    },
    {
      "epoch": 0.4331087584215592,
      "grad_norm": 0.2995765507221222,
      "learning_rate": 0.00036634065840827467,
      "loss": 4.7341,
      "step": 21150
    },
    {
      "epoch": 0.43331353797636846,
      "grad_norm": 0.27756834030151367,
      "learning_rate": 0.00036623818500661757,
      "loss": 4.7674,
      "step": 21160
    },
    {
      "epoch": 0.4335183175311777,
      "grad_norm": 0.28837379813194275,
      "learning_rate": 0.0003661355702331817,
      "loss": 4.6978,
      "step": 21170
    },
    {
      "epoch": 0.43372309708598694,
      "grad_norm": 0.3934195041656494,
      "learning_rate": 0.00036603281417523257,
      "loss": 4.7227,
      "step": 21180
    },
    {
      "epoch": 0.4339278766407962,
      "grad_norm": 0.26383545994758606,
      "learning_rate": 0.0003659299169201556,
      "loss": 4.7412,
      "step": 21190
    },
    {
      "epoch": 0.4341326561956054,
      "grad_norm": 0.3552657663822174,
      "learning_rate": 0.0003658268785554563,
      "loss": 4.7382,
      "step": 21200
    },
    {
      "epoch": 0.43433743575041467,
      "grad_norm": 0.2834928035736084,
      "learning_rate": 0.0003657236991687603,
      "loss": 4.7276,
      "step": 21210
    },
    {
      "epoch": 0.4345422153052239,
      "grad_norm": 0.28129100799560547,
      "learning_rate": 0.00036562037884781313,
      "loss": 4.7123,
      "step": 21220
    },
    {
      "epoch": 0.43474699486003315,
      "grad_norm": 0.28226780891418457,
      "learning_rate": 0.00036551691768048014,
      "loss": 4.748,
      "step": 21230
    },
    {
      "epoch": 0.4349517744148424,
      "grad_norm": 0.2716659605503082,
      "learning_rate": 0.00036541331575474644,
      "loss": 4.7273,
      "step": 21240
    },
    {
      "epoch": 0.4351565539696517,
      "grad_norm": 0.28415346145629883,
      "learning_rate": 0.0003653095731587169,
      "loss": 4.7302,
      "step": 21250
    },
    {
      "epoch": 0.43536133352446094,
      "grad_norm": 0.30821385979652405,
      "learning_rate": 0.00036520568998061595,
      "loss": 4.741,
      "step": 21260
    },
    {
      "epoch": 0.4355661130792702,
      "grad_norm": 0.32026225328445435,
      "learning_rate": 0.00036510166630878766,
      "loss": 4.7348,
      "step": 21270
    },
    {
      "epoch": 0.4357708926340794,
      "grad_norm": 0.2765481173992157,
      "learning_rate": 0.0003649975022316955,
      "loss": 4.7263,
      "step": 21280
    },
    {
      "epoch": 0.43597567218888866,
      "grad_norm": 0.2862047851085663,
      "learning_rate": 0.00036489319783792235,
      "loss": 4.717,
      "step": 21290
    },
    {
      "epoch": 0.4361804517436979,
      "grad_norm": 0.25941938161849976,
      "learning_rate": 0.00036478875321617043,
      "loss": 4.7675,
      "step": 21300
    },
    {
      "epoch": 0.43638523129850715,
      "grad_norm": 0.2648298740386963,
      "learning_rate": 0.00036468416845526134,
      "loss": 4.7403,
      "step": 21310
    },
    {
      "epoch": 0.4365900108533164,
      "grad_norm": 0.27693983912467957,
      "learning_rate": 0.00036457944364413555,
      "loss": 4.7214,
      "step": 21320
    },
    {
      "epoch": 0.43679479040812563,
      "grad_norm": 0.31461483240127563,
      "learning_rate": 0.000364474578871853,
      "loss": 4.7298,
      "step": 21330
    },
    {
      "epoch": 0.4369995699629349,
      "grad_norm": 0.2823515832424164,
      "learning_rate": 0.00036436957422759236,
      "loss": 4.7274,
      "step": 21340
    },
    {
      "epoch": 0.4372043495177442,
      "grad_norm": 0.2560424506664276,
      "learning_rate": 0.00036426442980065144,
      "loss": 4.7195,
      "step": 21350
    },
    {
      "epoch": 0.4374091290725534,
      "grad_norm": 0.2791268825531006,
      "learning_rate": 0.0003641591456804468,
      "loss": 4.7352,
      "step": 21360
    },
    {
      "epoch": 0.43761390862736266,
      "grad_norm": 0.26592713594436646,
      "learning_rate": 0.0003640537219565138,
      "loss": 4.737,
      "step": 21370
    },
    {
      "epoch": 0.4378186881821719,
      "grad_norm": 0.2827947437763214,
      "learning_rate": 0.0003639481587185067,
      "loss": 4.7368,
      "step": 21380
    },
    {
      "epoch": 0.43802346773698114,
      "grad_norm": 0.5054929852485657,
      "learning_rate": 0.00036384245605619814,
      "loss": 4.6831,
      "step": 21390
    },
    {
      "epoch": 0.4382282472917904,
      "grad_norm": 0.25543633103370667,
      "learning_rate": 0.00036373661405947956,
      "loss": 4.7222,
      "step": 21400
    },
    {
      "epoch": 0.4384330268465996,
      "grad_norm": 0.2819623351097107,
      "learning_rate": 0.0003636306328183607,
      "loss": 4.7212,
      "step": 21410
    },
    {
      "epoch": 0.43863780640140887,
      "grad_norm": 0.29091677069664,
      "learning_rate": 0.0003635245124229699,
      "loss": 4.7207,
      "step": 21420
    },
    {
      "epoch": 0.4388425859562181,
      "grad_norm": 0.27804920077323914,
      "learning_rate": 0.00036341825296355365,
      "loss": 4.7396,
      "step": 21430
    },
    {
      "epoch": 0.43904736551102735,
      "grad_norm": 0.27245569229125977,
      "learning_rate": 0.00036331185453047693,
      "loss": 4.7117,
      "step": 21440
    },
    {
      "epoch": 0.43925214506583665,
      "grad_norm": 0.25816354155540466,
      "learning_rate": 0.0003632053172142227,
      "loss": 4.7511,
      "step": 21450
    },
    {
      "epoch": 0.4394569246206459,
      "grad_norm": 0.2858197093009949,
      "learning_rate": 0.00036309864110539204,
      "loss": 4.7622,
      "step": 21460
    },
    {
      "epoch": 0.43966170417545514,
      "grad_norm": 0.37263616919517517,
      "learning_rate": 0.00036299182629470427,
      "loss": 4.7448,
      "step": 21470
    },
    {
      "epoch": 0.4398664837302644,
      "grad_norm": 0.26793110370635986,
      "learning_rate": 0.00036288487287299643,
      "loss": 4.7803,
      "step": 21480
    },
    {
      "epoch": 0.4400712632850736,
      "grad_norm": 0.27125048637390137,
      "learning_rate": 0.0003627777809312235,
      "loss": 4.7227,
      "step": 21490
    },
    {
      "epoch": 0.44027604283988286,
      "grad_norm": 0.2748330533504486,
      "learning_rate": 0.00036267055056045833,
      "loss": 4.7303,
      "step": 21500
    },
    {
      "epoch": 0.4404808223946921,
      "grad_norm": 0.2754065990447998,
      "learning_rate": 0.00036256318185189156,
      "loss": 4.7009,
      "step": 21510
    },
    {
      "epoch": 0.44068560194950135,
      "grad_norm": 0.27515119314193726,
      "learning_rate": 0.00036245567489683115,
      "loss": 4.7158,
      "step": 21520
    },
    {
      "epoch": 0.4408903815043106,
      "grad_norm": 0.296744167804718,
      "learning_rate": 0.00036234802978670296,
      "loss": 4.7453,
      "step": 21530
    },
    {
      "epoch": 0.44109516105911983,
      "grad_norm": 0.26276278495788574,
      "learning_rate": 0.00036224024661305024,
      "loss": 4.7214,
      "step": 21540
    },
    {
      "epoch": 0.44129994061392913,
      "grad_norm": 0.46705666184425354,
      "learning_rate": 0.00036213232546753366,
      "loss": 4.714,
      "step": 21550
    },
    {
      "epoch": 0.4415047201687384,
      "grad_norm": 0.3351914882659912,
      "learning_rate": 0.00036202426644193103,
      "loss": 4.7077,
      "step": 21560
    },
    {
      "epoch": 0.4417094997235476,
      "grad_norm": 0.268264502286911,
      "learning_rate": 0.0003619160696281378,
      "loss": 4.7474,
      "step": 21570
    },
    {
      "epoch": 0.44191427927835686,
      "grad_norm": 0.5669612884521484,
      "learning_rate": 0.0003618077351181662,
      "loss": 4.7052,
      "step": 21580
    },
    {
      "epoch": 0.4421190588331661,
      "grad_norm": 0.30829110741615295,
      "learning_rate": 0.0003616992630041458,
      "loss": 4.7338,
      "step": 21590
    },
    {
      "epoch": 0.44232383838797534,
      "grad_norm": 0.2763216197490692,
      "learning_rate": 0.0003615906533783232,
      "loss": 4.7038,
      "step": 21600
    },
    {
      "epoch": 0.4425286179427846,
      "grad_norm": 0.2838597595691681,
      "learning_rate": 0.0003614819063330618,
      "loss": 4.7492,
      "step": 21610
    },
    {
      "epoch": 0.44273339749759383,
      "grad_norm": 0.2547735571861267,
      "learning_rate": 0.000361373021960842,
      "loss": 4.7308,
      "step": 21620
    },
    {
      "epoch": 0.44293817705240307,
      "grad_norm": 0.26135656237602234,
      "learning_rate": 0.0003612640003542608,
      "loss": 4.7146,
      "step": 21630
    },
    {
      "epoch": 0.4431429566072123,
      "grad_norm": 0.259290874004364,
      "learning_rate": 0.00036115484160603213,
      "loss": 4.7242,
      "step": 21640
    },
    {
      "epoch": 0.4433477361620216,
      "grad_norm": 0.2724955677986145,
      "learning_rate": 0.0003610455458089864,
      "loss": 4.7854,
      "step": 21650
    },
    {
      "epoch": 0.44355251571683085,
      "grad_norm": 0.2639175355434418,
      "learning_rate": 0.0003609361130560706,
      "loss": 4.7226,
      "step": 21660
    },
    {
      "epoch": 0.4437572952716401,
      "grad_norm": 0.2689358592033386,
      "learning_rate": 0.0003608265434403483,
      "loss": 4.7075,
      "step": 21670
    },
    {
      "epoch": 0.44396207482644934,
      "grad_norm": 0.2829224765300751,
      "learning_rate": 0.0003607168370549993,
      "loss": 4.7617,
      "step": 21680
    },
    {
      "epoch": 0.4441668543812586,
      "grad_norm": 0.25918588042259216,
      "learning_rate": 0.00036060699399331975,
      "loss": 4.7101,
      "step": 21690
    },
    {
      "epoch": 0.4443716339360678,
      "grad_norm": 0.2661386728286743,
      "learning_rate": 0.000360497014348722,
      "loss": 4.7684,
      "step": 21700
    },
    {
      "epoch": 0.44457641349087706,
      "grad_norm": 0.2601064145565033,
      "learning_rate": 0.00036038689821473477,
      "loss": 4.7369,
      "step": 21710
    },
    {
      "epoch": 0.4447811930456863,
      "grad_norm": 0.27942773699760437,
      "learning_rate": 0.0003602766456850026,
      "loss": 4.6933,
      "step": 21720
    },
    {
      "epoch": 0.44498597260049555,
      "grad_norm": 0.2897523045539856,
      "learning_rate": 0.0003601662568532862,
      "loss": 4.7398,
      "step": 21730
    },
    {
      "epoch": 0.4451907521553048,
      "grad_norm": 0.28540870547294617,
      "learning_rate": 0.00036005573181346196,
      "loss": 4.7736,
      "step": 21740
    },
    {
      "epoch": 0.4453955317101141,
      "grad_norm": 0.2802298069000244,
      "learning_rate": 0.00035994507065952237,
      "loss": 4.7264,
      "step": 21750
    },
    {
      "epoch": 0.44560031126492333,
      "grad_norm": 0.266023188829422,
      "learning_rate": 0.0003598342734855755,
      "loss": 4.7555,
      "step": 21760
    },
    {
      "epoch": 0.4458050908197326,
      "grad_norm": 0.29166194796562195,
      "learning_rate": 0.0003597233403858451,
      "loss": 4.7499,
      "step": 21770
    },
    {
      "epoch": 0.4460098703745418,
      "grad_norm": 0.27979788184165955,
      "learning_rate": 0.0003596122714546707,
      "loss": 4.6946,
      "step": 21780
    },
    {
      "epoch": 0.44621464992935106,
      "grad_norm": 0.28484687209129333,
      "learning_rate": 0.00035950106678650707,
      "loss": 4.7541,
      "step": 21790
    },
    {
      "epoch": 0.4464194294841603,
      "grad_norm": 0.3059164583683014,
      "learning_rate": 0.00035938972647592466,
      "loss": 4.6932,
      "step": 21800
    },
    {
      "epoch": 0.44662420903896954,
      "grad_norm": 0.28869423270225525,
      "learning_rate": 0.0003592782506176091,
      "loss": 4.7477,
      "step": 21810
    },
    {
      "epoch": 0.4468289885937788,
      "grad_norm": 0.29801151156425476,
      "learning_rate": 0.0003591666393063615,
      "loss": 4.7742,
      "step": 21820
    },
    {
      "epoch": 0.44703376814858803,
      "grad_norm": 0.2738795876502991,
      "learning_rate": 0.00035905489263709774,
      "loss": 4.7315,
      "step": 21830
    },
    {
      "epoch": 0.44723854770339727,
      "grad_norm": 0.2763841450214386,
      "learning_rate": 0.0003589430107048493,
      "loss": 4.751,
      "step": 21840
    },
    {
      "epoch": 0.4474433272582065,
      "grad_norm": 0.2792346477508545,
      "learning_rate": 0.0003588309936047624,
      "loss": 4.7352,
      "step": 21850
    },
    {
      "epoch": 0.4476481068130158,
      "grad_norm": 0.2677178978919983,
      "learning_rate": 0.0003587188414320983,
      "loss": 4.701,
      "step": 21860
    },
    {
      "epoch": 0.44785288636782505,
      "grad_norm": 0.26957011222839355,
      "learning_rate": 0.00035860655428223315,
      "loss": 4.7266,
      "step": 21870
    },
    {
      "epoch": 0.4480576659226343,
      "grad_norm": 0.2695551812648773,
      "learning_rate": 0.0003584941322506578,
      "loss": 4.7265,
      "step": 21880
    },
    {
      "epoch": 0.44826244547744354,
      "grad_norm": 0.28194916248321533,
      "learning_rate": 0.00035838157543297786,
      "loss": 4.741,
      "step": 21890
    },
    {
      "epoch": 0.4484672250322528,
      "grad_norm": 0.26357901096343994,
      "learning_rate": 0.00035826888392491364,
      "loss": 4.7307,
      "step": 21900
    },
    {
      "epoch": 0.448672004587062,
      "grad_norm": 0.2969251275062561,
      "learning_rate": 0.00035815605782229984,
      "loss": 4.7095,
      "step": 21910
    },
    {
      "epoch": 0.44887678414187127,
      "grad_norm": 0.2593156397342682,
      "learning_rate": 0.0003580430972210857,
      "loss": 4.7314,
      "step": 21920
    },
    {
      "epoch": 0.4490815636966805,
      "grad_norm": 0.29631832242012024,
      "learning_rate": 0.0003579300022173348,
      "loss": 4.7677,
      "step": 21930
    },
    {
      "epoch": 0.44928634325148975,
      "grad_norm": 0.34863293170928955,
      "learning_rate": 0.00035781677290722504,
      "loss": 4.7401,
      "step": 21940
    },
    {
      "epoch": 0.449491122806299,
      "grad_norm": 0.2681001126766205,
      "learning_rate": 0.0003577034093870487,
      "loss": 4.7624,
      "step": 21950
    },
    {
      "epoch": 0.4496959023611083,
      "grad_norm": 0.26574641466140747,
      "learning_rate": 0.0003575899117532119,
      "loss": 4.723,
      "step": 21960
    },
    {
      "epoch": 0.44990068191591753,
      "grad_norm": 0.2752978801727295,
      "learning_rate": 0.00035747628010223504,
      "loss": 4.7108,
      "step": 21970
    },
    {
      "epoch": 0.4501054614707268,
      "grad_norm": 0.31212377548217773,
      "learning_rate": 0.00035736251453075243,
      "loss": 4.738,
      "step": 21980
    },
    {
      "epoch": 0.450310241025536,
      "grad_norm": 0.3273192048072815,
      "learning_rate": 0.0003572486151355121,
      "loss": 4.7489,
      "step": 21990
    },
    {
      "epoch": 0.45051502058034526,
      "grad_norm": 0.3560446798801422,
      "learning_rate": 0.0003571345820133762,
      "loss": 4.6892,
      "step": 22000
    },
    {
      "epoch": 0.45051502058034526,
      "eval_loss": 4.739803791046143,
      "eval_runtime": 4.3827,
      "eval_samples_per_second": 266.044,
      "eval_steps_per_second": 33.313,
      "step": 22000
    },
    {
      "epoch": 0.4507198001351545,
      "grad_norm": 0.290144145488739,
      "learning_rate": 0.00035702041526132033,
      "loss": 4.7475,
      "step": 22010
    },
    {
      "epoch": 0.45092457968996374,
      "grad_norm": 0.28808197379112244,
      "learning_rate": 0.00035690611497643385,
      "loss": 4.7419,
      "step": 22020
    },
    {
      "epoch": 0.451129359244773,
      "grad_norm": 0.27849912643432617,
      "learning_rate": 0.0003567916812559197,
      "loss": 4.7365,
      "step": 22030
    },
    {
      "epoch": 0.45133413879958223,
      "grad_norm": 0.27138206362724304,
      "learning_rate": 0.00035667711419709423,
      "loss": 4.742,
      "step": 22040
    },
    {
      "epoch": 0.45153891835439147,
      "grad_norm": 0.26201629638671875,
      "learning_rate": 0.0003565624138973872,
      "loss": 4.7362,
      "step": 22050
    },
    {
      "epoch": 0.45174369790920077,
      "grad_norm": 0.2977285087108612,
      "learning_rate": 0.0003564475804543418,
      "loss": 4.7327,
      "step": 22060
    },
    {
      "epoch": 0.45194847746401,
      "grad_norm": 0.27661874890327454,
      "learning_rate": 0.00035633261396561425,
      "loss": 4.7272,
      "step": 22070
    },
    {
      "epoch": 0.45215325701881925,
      "grad_norm": 0.29062989354133606,
      "learning_rate": 0.000356217514528974,
      "loss": 4.7189,
      "step": 22080
    },
    {
      "epoch": 0.4523580365736285,
      "grad_norm": 0.2621004581451416,
      "learning_rate": 0.0003561022822423036,
      "loss": 4.6906,
      "step": 22090
    },
    {
      "epoch": 0.45256281612843774,
      "grad_norm": 0.27067258954048157,
      "learning_rate": 0.00035598691720359853,
      "loss": 4.7545,
      "step": 22100
    },
    {
      "epoch": 0.452767595683247,
      "grad_norm": 0.25577986240386963,
      "learning_rate": 0.0003558714195109672,
      "loss": 4.7186,
      "step": 22110
    },
    {
      "epoch": 0.4529723752380562,
      "grad_norm": 0.2554454803466797,
      "learning_rate": 0.00035575578926263075,
      "loss": 4.7174,
      "step": 22120
    },
    {
      "epoch": 0.45317715479286547,
      "grad_norm": 0.28933700919151306,
      "learning_rate": 0.00035564002655692317,
      "loss": 4.7488,
      "step": 22130
    },
    {
      "epoch": 0.4533819343476747,
      "grad_norm": 0.268405944108963,
      "learning_rate": 0.0003555241314922911,
      "loss": 4.7641,
      "step": 22140
    },
    {
      "epoch": 0.45358671390248395,
      "grad_norm": 0.2583339512348175,
      "learning_rate": 0.0003554081041672935,
      "loss": 4.7023,
      "step": 22150
    },
    {
      "epoch": 0.45379149345729325,
      "grad_norm": 0.2833841145038605,
      "learning_rate": 0.0003552919446806021,
      "loss": 4.7593,
      "step": 22160
    },
    {
      "epoch": 0.4539962730121025,
      "grad_norm": 0.250685453414917,
      "learning_rate": 0.0003551756531310009,
      "loss": 4.7128,
      "step": 22170
    },
    {
      "epoch": 0.45420105256691173,
      "grad_norm": 0.28385424613952637,
      "learning_rate": 0.00035505922961738616,
      "loss": 4.7244,
      "step": 22180
    },
    {
      "epoch": 0.454405832121721,
      "grad_norm": 0.27036428451538086,
      "learning_rate": 0.00035494267423876645,
      "loss": 4.7176,
      "step": 22190
    },
    {
      "epoch": 0.4546106116765302,
      "grad_norm": 0.2948550581932068,
      "learning_rate": 0.0003548259870942624,
      "loss": 4.6934,
      "step": 22200
    },
    {
      "epoch": 0.45481539123133946,
      "grad_norm": 0.25978729128837585,
      "learning_rate": 0.00035470916828310683,
      "loss": 4.7652,
      "step": 22210
    },
    {
      "epoch": 0.4550201707861487,
      "grad_norm": 0.28499385714530945,
      "learning_rate": 0.00035459221790464434,
      "loss": 4.7189,
      "step": 22220
    },
    {
      "epoch": 0.45522495034095795,
      "grad_norm": 0.3245640993118286,
      "learning_rate": 0.0003544751360583316,
      "loss": 4.7064,
      "step": 22230
    },
    {
      "epoch": 0.4554297298957672,
      "grad_norm": 0.2962793707847595,
      "learning_rate": 0.000354357922843737,
      "loss": 4.7432,
      "step": 22240
    },
    {
      "epoch": 0.45563450945057643,
      "grad_norm": 0.27355194091796875,
      "learning_rate": 0.00035424057836054064,
      "loss": 4.7244,
      "step": 22250
    },
    {
      "epoch": 0.45583928900538573,
      "grad_norm": 0.29868021607398987,
      "learning_rate": 0.0003541231027085342,
      "loss": 4.7054,
      "step": 22260
    },
    {
      "epoch": 0.45604406856019497,
      "grad_norm": 0.26717624068260193,
      "learning_rate": 0.000354005495987621,
      "loss": 4.6939,
      "step": 22270
    },
    {
      "epoch": 0.4562488481150042,
      "grad_norm": 0.2547793686389923,
      "learning_rate": 0.00035388775829781586,
      "loss": 4.7223,
      "step": 22280
    },
    {
      "epoch": 0.45645362766981346,
      "grad_norm": 0.27395254373550415,
      "learning_rate": 0.0003537698897392448,
      "loss": 4.7214,
      "step": 22290
    },
    {
      "epoch": 0.4566584072246227,
      "grad_norm": 0.38744744658470154,
      "learning_rate": 0.00035365189041214535,
      "loss": 4.7222,
      "step": 22300
    },
    {
      "epoch": 0.45686318677943194,
      "grad_norm": 0.2687004506587982,
      "learning_rate": 0.00035353376041686614,
      "loss": 4.6706,
      "step": 22310
    },
    {
      "epoch": 0.4570679663342412,
      "grad_norm": 0.3844720423221588,
      "learning_rate": 0.0003534154998538669,
      "loss": 4.7387,
      "step": 22320
    },
    {
      "epoch": 0.4572727458890504,
      "grad_norm": 0.33876246213912964,
      "learning_rate": 0.0003532971088237184,
      "loss": 4.7352,
      "step": 22330
    },
    {
      "epoch": 0.45747752544385967,
      "grad_norm": 0.3038104176521301,
      "learning_rate": 0.0003531785874271023,
      "loss": 4.7192,
      "step": 22340
    },
    {
      "epoch": 0.4576823049986689,
      "grad_norm": 0.29718440771102905,
      "learning_rate": 0.0003530599357648114,
      "loss": 4.7036,
      "step": 22350
    },
    {
      "epoch": 0.4578870845534782,
      "grad_norm": 0.2830037772655487,
      "learning_rate": 0.000352941153937749,
      "loss": 4.7056,
      "step": 22360
    },
    {
      "epoch": 0.45809186410828745,
      "grad_norm": 0.26650917530059814,
      "learning_rate": 0.0003528222420469292,
      "loss": 4.7275,
      "step": 22370
    },
    {
      "epoch": 0.4582966436630967,
      "grad_norm": 0.3146975040435791,
      "learning_rate": 0.0003527032001934767,
      "loss": 4.7309,
      "step": 22380
    },
    {
      "epoch": 0.45850142321790593,
      "grad_norm": 0.26475873589515686,
      "learning_rate": 0.00035258402847862665,
      "loss": 4.7211,
      "step": 22390
    },
    {
      "epoch": 0.4587062027727152,
      "grad_norm": 0.2605159282684326,
      "learning_rate": 0.0003524647270037248,
      "loss": 4.7134,
      "step": 22400
    },
    {
      "epoch": 0.4589109823275244,
      "grad_norm": 0.29321005940437317,
      "learning_rate": 0.00035234529587022703,
      "loss": 4.7272,
      "step": 22410
    },
    {
      "epoch": 0.45911576188233366,
      "grad_norm": 0.27110812067985535,
      "learning_rate": 0.0003522257351796997,
      "loss": 4.6885,
      "step": 22420
    },
    {
      "epoch": 0.4593205414371429,
      "grad_norm": 0.2615066468715668,
      "learning_rate": 0.00035210604503381926,
      "loss": 4.6825,
      "step": 22430
    },
    {
      "epoch": 0.45952532099195215,
      "grad_norm": 0.2677325904369354,
      "learning_rate": 0.0003519862255343722,
      "loss": 4.7029,
      "step": 22440
    },
    {
      "epoch": 0.4597301005467614,
      "grad_norm": 0.29238784313201904,
      "learning_rate": 0.00035186627678325506,
      "loss": 4.7393,
      "step": 22450
    },
    {
      "epoch": 0.4599348801015707,
      "grad_norm": 0.28893929719924927,
      "learning_rate": 0.00035174619888247435,
      "loss": 4.7201,
      "step": 22460
    },
    {
      "epoch": 0.46013965965637993,
      "grad_norm": 0.2545911371707916,
      "learning_rate": 0.0003516259919341463,
      "loss": 4.76,
      "step": 22470
    },
    {
      "epoch": 0.46034443921118917,
      "grad_norm": 0.28869879245758057,
      "learning_rate": 0.00035150565604049694,
      "loss": 4.7196,
      "step": 22480
    },
    {
      "epoch": 0.4605492187659984,
      "grad_norm": 0.27274906635284424,
      "learning_rate": 0.00035138519130386203,
      "loss": 4.7187,
      "step": 22490
    },
    {
      "epoch": 0.46075399832080766,
      "grad_norm": 0.27680766582489014,
      "learning_rate": 0.00035126459782668676,
      "loss": 4.7345,
      "step": 22500
    },
    {
      "epoch": 0.4609587778756169,
      "grad_norm": 0.2799220085144043,
      "learning_rate": 0.00035114387571152586,
      "loss": 4.7529,
      "step": 22510
    },
    {
      "epoch": 0.46116355743042614,
      "grad_norm": 0.2812877595424652,
      "learning_rate": 0.00035102302506104363,
      "loss": 4.7344,
      "step": 22520
    },
    {
      "epoch": 0.4613683369852354,
      "grad_norm": 0.2624521553516388,
      "learning_rate": 0.0003509020459780133,
      "loss": 4.7282,
      "step": 22530
    },
    {
      "epoch": 0.4615731165400446,
      "grad_norm": 0.28584733605384827,
      "learning_rate": 0.0003507809385653176,
      "loss": 4.7569,
      "step": 22540
    },
    {
      "epoch": 0.46177789609485387,
      "grad_norm": 0.29557710886001587,
      "learning_rate": 0.0003506597029259484,
      "loss": 4.7186,
      "step": 22550
    },
    {
      "epoch": 0.4619826756496631,
      "grad_norm": 0.33304429054260254,
      "learning_rate": 0.00035053833916300656,
      "loss": 4.7658,
      "step": 22560
    },
    {
      "epoch": 0.4621874552044724,
      "grad_norm": 0.27438461780548096,
      "learning_rate": 0.0003504168473797018,
      "loss": 4.7322,
      "step": 22570
    },
    {
      "epoch": 0.46239223475928165,
      "grad_norm": 0.27427247166633606,
      "learning_rate": 0.00035029522767935284,
      "loss": 4.6952,
      "step": 22580
    },
    {
      "epoch": 0.4625970143140909,
      "grad_norm": 0.3031047284603119,
      "learning_rate": 0.0003501734801653871,
      "loss": 4.694,
      "step": 22590
    },
    {
      "epoch": 0.46280179386890014,
      "grad_norm": 0.32042402029037476,
      "learning_rate": 0.0003500516049413409,
      "loss": 4.7333,
      "step": 22600
    },
    {
      "epoch": 0.4630065734237094,
      "grad_norm": 0.2565493881702423,
      "learning_rate": 0.00034992960211085875,
      "loss": 4.6752,
      "step": 22610
    },
    {
      "epoch": 0.4632113529785186,
      "grad_norm": 0.3041118383407593,
      "learning_rate": 0.0003498074717776941,
      "loss": 4.7079,
      "step": 22620
    },
    {
      "epoch": 0.46341613253332786,
      "grad_norm": 0.2607264518737793,
      "learning_rate": 0.00034968521404570864,
      "loss": 4.7254,
      "step": 22630
    },
    {
      "epoch": 0.4636209120881371,
      "grad_norm": 0.3120455741882324,
      "learning_rate": 0.00034956282901887236,
      "loss": 4.6809,
      "step": 22640
    },
    {
      "epoch": 0.46382569164294635,
      "grad_norm": 0.2700790464878082,
      "learning_rate": 0.0003494403168012637,
      "loss": 4.7507,
      "step": 22650
    },
    {
      "epoch": 0.4640304711977556,
      "grad_norm": 0.8952503800392151,
      "learning_rate": 0.00034931767749706895,
      "loss": 4.7224,
      "step": 22660
    },
    {
      "epoch": 0.4642352507525649,
      "grad_norm": 0.26338472962379456,
      "learning_rate": 0.00034919491121058284,
      "loss": 4.7215,
      "step": 22670
    },
    {
      "epoch": 0.46444003030737413,
      "grad_norm": 0.2886572480201721,
      "learning_rate": 0.0003490720180462078,
      "loss": 4.7211,
      "step": 22680
    },
    {
      "epoch": 0.46464480986218337,
      "grad_norm": 0.26196008920669556,
      "learning_rate": 0.00034894899810845425,
      "loss": 4.6938,
      "step": 22690
    },
    {
      "epoch": 0.4648495894169926,
      "grad_norm": 0.2621656656265259,
      "learning_rate": 0.00034882585150194045,
      "loss": 4.7492,
      "step": 22700
    },
    {
      "epoch": 0.46505436897180186,
      "grad_norm": 0.27103763818740845,
      "learning_rate": 0.0003487025783313925,
      "loss": 4.7302,
      "step": 22710
    },
    {
      "epoch": 0.4652591485266111,
      "grad_norm": 0.2672967314720154,
      "learning_rate": 0.00034857917870164377,
      "loss": 4.7376,
      "step": 22720
    },
    {
      "epoch": 0.46546392808142034,
      "grad_norm": 0.2624387741088867,
      "learning_rate": 0.00034845565271763554,
      "loss": 4.7141,
      "step": 22730
    },
    {
      "epoch": 0.4656687076362296,
      "grad_norm": 0.26433417201042175,
      "learning_rate": 0.0003483320004844164,
      "loss": 4.7191,
      "step": 22740
    },
    {
      "epoch": 0.4658734871910388,
      "grad_norm": 0.35606059432029724,
      "learning_rate": 0.0003482082221071424,
      "loss": 4.7023,
      "step": 22750
    },
    {
      "epoch": 0.46607826674584807,
      "grad_norm": 0.34465745091438293,
      "learning_rate": 0.00034808431769107665,
      "loss": 4.7355,
      "step": 22760
    },
    {
      "epoch": 0.46628304630065737,
      "grad_norm": 0.27950605750083923,
      "learning_rate": 0.0003479602873415895,
      "loss": 4.7628,
      "step": 22770
    },
    {
      "epoch": 0.4664878258554666,
      "grad_norm": 0.25964662432670593,
      "learning_rate": 0.00034783613116415866,
      "loss": 4.6999,
      "step": 22780
    },
    {
      "epoch": 0.46669260541027585,
      "grad_norm": 0.28749728202819824,
      "learning_rate": 0.0003477118492643685,
      "loss": 4.7659,
      "step": 22790
    },
    {
      "epoch": 0.4668973849650851,
      "grad_norm": 0.2610335648059845,
      "learning_rate": 0.00034758744174791055,
      "loss": 4.6715,
      "step": 22800
    },
    {
      "epoch": 0.46710216451989434,
      "grad_norm": 0.3397755026817322,
      "learning_rate": 0.0003474629087205831,
      "loss": 4.7363,
      "step": 22810
    },
    {
      "epoch": 0.4673069440747036,
      "grad_norm": 0.2962636947631836,
      "learning_rate": 0.00034733825028829106,
      "loss": 4.7283,
      "step": 22820
    },
    {
      "epoch": 0.4675117236295128,
      "grad_norm": 0.2690482437610626,
      "learning_rate": 0.00034721346655704617,
      "loss": 4.7064,
      "step": 22830
    },
    {
      "epoch": 0.46771650318432206,
      "grad_norm": 0.3107163906097412,
      "learning_rate": 0.0003470885576329665,
      "loss": 4.7619,
      "step": 22840
    },
    {
      "epoch": 0.4679212827391313,
      "grad_norm": 0.27058449387550354,
      "learning_rate": 0.0003469635236222768,
      "loss": 4.7168,
      "step": 22850
    },
    {
      "epoch": 0.46812606229394055,
      "grad_norm": 0.3666093051433563,
      "learning_rate": 0.00034683836463130825,
      "loss": 4.7055,
      "step": 22860
    },
    {
      "epoch": 0.46833084184874985,
      "grad_norm": 0.27504703402519226,
      "learning_rate": 0.000346713080766498,
      "loss": 4.7398,
      "step": 22870
    },
    {
      "epoch": 0.4685356214035591,
      "grad_norm": 0.3063828647136688,
      "learning_rate": 0.00034658767213438963,
      "loss": 4.76,
      "step": 22880
    },
    {
      "epoch": 0.46874040095836833,
      "grad_norm": 0.3170681595802307,
      "learning_rate": 0.0003464621388416328,
      "loss": 4.6956,
      "step": 22890
    },
    {
      "epoch": 0.4689451805131776,
      "grad_norm": 0.2757911682128906,
      "learning_rate": 0.0003463364809949832,
      "loss": 4.7313,
      "step": 22900
    },
    {
      "epoch": 0.4691499600679868,
      "grad_norm": 0.2873567044734955,
      "learning_rate": 0.0003462106987013023,
      "loss": 4.7163,
      "step": 22910
    },
    {
      "epoch": 0.46935473962279606,
      "grad_norm": 0.2824573516845703,
      "learning_rate": 0.0003460847920675575,
      "loss": 4.692,
      "step": 22920
    },
    {
      "epoch": 0.4695595191776053,
      "grad_norm": 0.2875484824180603,
      "learning_rate": 0.00034595876120082204,
      "loss": 4.716,
      "step": 22930
    },
    {
      "epoch": 0.46976429873241454,
      "grad_norm": 0.27463841438293457,
      "learning_rate": 0.0003458326062082746,
      "loss": 4.731,
      "step": 22940
    },
    {
      "epoch": 0.4699690782872238,
      "grad_norm": 0.2988135516643524,
      "learning_rate": 0.00034570632719719966,
      "loss": 4.7196,
      "step": 22950
    },
    {
      "epoch": 0.470173857842033,
      "grad_norm": 0.28052037954330444,
      "learning_rate": 0.0003455799242749869,
      "loss": 4.7489,
      "step": 22960
    },
    {
      "epoch": 0.4703786373968423,
      "grad_norm": 0.29746437072753906,
      "learning_rate": 0.0003454533975491316,
      "loss": 4.7202,
      "step": 22970
    },
    {
      "epoch": 0.47058341695165157,
      "grad_norm": 0.26427146792411804,
      "learning_rate": 0.00034532674712723423,
      "loss": 4.74,
      "step": 22980
    },
    {
      "epoch": 0.4707881965064608,
      "grad_norm": 0.263955295085907,
      "learning_rate": 0.00034519997311700047,
      "loss": 4.6972,
      "step": 22990
    },
    {
      "epoch": 0.47099297606127005,
      "grad_norm": 0.2544487416744232,
      "learning_rate": 0.00034507307562624105,
      "loss": 4.7433,
      "step": 23000
    },
    {
      "epoch": 0.4711977556160793,
      "grad_norm": 0.3022330701351166,
      "learning_rate": 0.00034494605476287176,
      "loss": 4.7164,
      "step": 23010
    },
    {
      "epoch": 0.47140253517088854,
      "grad_norm": 0.24907058477401733,
      "learning_rate": 0.00034481891063491333,
      "loss": 4.6942,
      "step": 23020
    },
    {
      "epoch": 0.4716073147256978,
      "grad_norm": 0.2630559504032135,
      "learning_rate": 0.0003446916433504913,
      "loss": 4.7131,
      "step": 23030
    },
    {
      "epoch": 0.471812094280507,
      "grad_norm": 0.27668577432632446,
      "learning_rate": 0.00034456425301783585,
      "loss": 4.7348,
      "step": 23040
    },
    {
      "epoch": 0.47201687383531626,
      "grad_norm": 0.28554463386535645,
      "learning_rate": 0.000344436739745282,
      "loss": 4.6918,
      "step": 23050
    },
    {
      "epoch": 0.4722216533901255,
      "grad_norm": 0.27102744579315186,
      "learning_rate": 0.00034430910364126907,
      "loss": 4.7203,
      "step": 23060
    },
    {
      "epoch": 0.4724264329449348,
      "grad_norm": 0.2832464873790741,
      "learning_rate": 0.00034418134481434113,
      "loss": 4.7284,
      "step": 23070
    },
    {
      "epoch": 0.47263121249974405,
      "grad_norm": 0.3161238729953766,
      "learning_rate": 0.00034405346337314637,
      "loss": 4.728,
      "step": 23080
    },
    {
      "epoch": 0.4728359920545533,
      "grad_norm": 0.3083503246307373,
      "learning_rate": 0.0003439254594264373,
      "loss": 4.7062,
      "step": 23090
    },
    {
      "epoch": 0.47304077160936253,
      "grad_norm": 0.2700280249118805,
      "learning_rate": 0.00034379733308307074,
      "loss": 4.6864,
      "step": 23100
    },
    {
      "epoch": 0.4732455511641718,
      "grad_norm": 0.3294050991535187,
      "learning_rate": 0.0003436690844520075,
      "loss": 4.7536,
      "step": 23110
    },
    {
      "epoch": 0.473450330718981,
      "grad_norm": 0.2681991159915924,
      "learning_rate": 0.0003435407136423123,
      "loss": 4.7242,
      "step": 23120
    },
    {
      "epoch": 0.47365511027379026,
      "grad_norm": 0.2928425371646881,
      "learning_rate": 0.000343412220763154,
      "loss": 4.7083,
      "step": 23130
    },
    {
      "epoch": 0.4738598898285995,
      "grad_norm": 0.271829217672348,
      "learning_rate": 0.000343283605923805,
      "loss": 4.7175,
      "step": 23140
    },
    {
      "epoch": 0.47406466938340874,
      "grad_norm": 0.2743125855922699,
      "learning_rate": 0.00034315486923364164,
      "loss": 4.6993,
      "step": 23150
    },
    {
      "epoch": 0.474269448938218,
      "grad_norm": 0.2834107577800751,
      "learning_rate": 0.00034302601080214376,
      "loss": 4.72,
      "step": 23160
    },
    {
      "epoch": 0.4744742284930273,
      "grad_norm": 0.2990763783454895,
      "learning_rate": 0.0003428970307388947,
      "loss": 4.7021,
      "step": 23170
    },
    {
      "epoch": 0.4746790080478365,
      "grad_norm": 0.28749608993530273,
      "learning_rate": 0.00034276792915358145,
      "loss": 4.724,
      "step": 23180
    },
    {
      "epoch": 0.47488378760264577,
      "grad_norm": 0.2540006935596466,
      "learning_rate": 0.000342638706155994,
      "loss": 4.7423,
      "step": 23190
    },
    {
      "epoch": 0.475088567157455,
      "grad_norm": 0.5250611305236816,
      "learning_rate": 0.0003425093618560259,
      "loss": 4.755,
      "step": 23200
    },
    {
      "epoch": 0.47529334671226425,
      "grad_norm": 0.2630171477794647,
      "learning_rate": 0.00034237989636367374,
      "loss": 4.7299,
      "step": 23210
    },
    {
      "epoch": 0.4754981262670735,
      "grad_norm": 0.28354302048683167,
      "learning_rate": 0.00034225030978903714,
      "loss": 4.7504,
      "step": 23220
    },
    {
      "epoch": 0.47570290582188274,
      "grad_norm": 0.3067748248577118,
      "learning_rate": 0.00034212060224231874,
      "loss": 4.7438,
      "step": 23230
    },
    {
      "epoch": 0.475907685376692,
      "grad_norm": 0.26697567105293274,
      "learning_rate": 0.0003419907738338241,
      "loss": 4.7372,
      "step": 23240
    },
    {
      "epoch": 0.4761124649315012,
      "grad_norm": 0.30045849084854126,
      "learning_rate": 0.0003418608246739614,
      "loss": 4.7013,
      "step": 23250
    },
    {
      "epoch": 0.47631724448631046,
      "grad_norm": 0.5347376465797424,
      "learning_rate": 0.0003417307548732417,
      "loss": 4.6699,
      "step": 23260
    },
    {
      "epoch": 0.4765220240411197,
      "grad_norm": 0.2729143500328064,
      "learning_rate": 0.0003416005645422785,
      "loss": 4.7144,
      "step": 23270
    },
    {
      "epoch": 0.476726803595929,
      "grad_norm": 0.2777980864048004,
      "learning_rate": 0.00034147025379178795,
      "loss": 4.7186,
      "step": 23280
    },
    {
      "epoch": 0.47693158315073825,
      "grad_norm": 0.2925539016723633,
      "learning_rate": 0.00034133982273258846,
      "loss": 4.7271,
      "step": 23290
    },
    {
      "epoch": 0.4771363627055475,
      "grad_norm": 0.2643592655658722,
      "learning_rate": 0.0003412092714756008,
      "loss": 4.7118,
      "step": 23300
    },
    {
      "epoch": 0.47734114226035673,
      "grad_norm": 0.2659567594528198,
      "learning_rate": 0.0003410786001318481,
      "loss": 4.7294,
      "step": 23310
    },
    {
      "epoch": 0.477545921815166,
      "grad_norm": 0.2851400673389435,
      "learning_rate": 0.0003409478088124554,
      "loss": 4.723,
      "step": 23320
    },
    {
      "epoch": 0.4777507013699752,
      "grad_norm": 0.2710046172142029,
      "learning_rate": 0.0003408168976286498,
      "loss": 4.7037,
      "step": 23330
    },
    {
      "epoch": 0.47795548092478446,
      "grad_norm": 0.4545775055885315,
      "learning_rate": 0.00034068586669176054,
      "loss": 4.7028,
      "step": 23340
    },
    {
      "epoch": 0.4781602604795937,
      "grad_norm": 0.2931750416755676,
      "learning_rate": 0.00034055471611321855,
      "loss": 4.7206,
      "step": 23350
    },
    {
      "epoch": 0.47836504003440294,
      "grad_norm": 0.2837902307510376,
      "learning_rate": 0.00034042344600455633,
      "loss": 4.6986,
      "step": 23360
    },
    {
      "epoch": 0.4785698195892122,
      "grad_norm": 0.2740366756916046,
      "learning_rate": 0.00034029205647740843,
      "loss": 4.7222,
      "step": 23370
    },
    {
      "epoch": 0.4787745991440215,
      "grad_norm": 0.271064430475235,
      "learning_rate": 0.00034016054764351056,
      "loss": 4.7188,
      "step": 23380
    },
    {
      "epoch": 0.4789793786988307,
      "grad_norm": 0.26214951276779175,
      "learning_rate": 0.0003400289196147002,
      "loss": 4.7356,
      "step": 23390
    },
    {
      "epoch": 0.47918415825363997,
      "grad_norm": 0.30367031693458557,
      "learning_rate": 0.000339897172502916,
      "loss": 4.7153,
      "step": 23400
    },
    {
      "epoch": 0.4793889378084492,
      "grad_norm": 0.3222086727619171,
      "learning_rate": 0.00033976530642019793,
      "loss": 4.7567,
      "step": 23410
    },
    {
      "epoch": 0.47959371736325845,
      "grad_norm": 0.2993817627429962,
      "learning_rate": 0.00033963332147868725,
      "loss": 4.7273,
      "step": 23420
    },
    {
      "epoch": 0.4797984969180677,
      "grad_norm": 0.2629406154155731,
      "learning_rate": 0.0003395012177906261,
      "loss": 4.7384,
      "step": 23430
    },
    {
      "epoch": 0.48000327647287694,
      "grad_norm": 0.2722770869731903,
      "learning_rate": 0.00033936899546835764,
      "loss": 4.6497,
      "step": 23440
    },
    {
      "epoch": 0.4802080560276862,
      "grad_norm": 0.28254806995391846,
      "learning_rate": 0.00033923665462432613,
      "loss": 4.6993,
      "step": 23450
    },
    {
      "epoch": 0.4804128355824954,
      "grad_norm": 0.25743168592453003,
      "learning_rate": 0.0003391041953710764,
      "loss": 4.722,
      "step": 23460
    },
    {
      "epoch": 0.48061761513730467,
      "grad_norm": 0.2827824354171753,
      "learning_rate": 0.0003389716178212541,
      "loss": 4.7193,
      "step": 23470
    },
    {
      "epoch": 0.48082239469211396,
      "grad_norm": 0.27589377760887146,
      "learning_rate": 0.0003388389220876053,
      "loss": 4.7486,
      "step": 23480
    },
    {
      "epoch": 0.4810271742469232,
      "grad_norm": 0.31261277198791504,
      "learning_rate": 0.0003387061082829769,
      "loss": 4.7422,
      "step": 23490
    },
    {
      "epoch": 0.48123195380173245,
      "grad_norm": 0.28861501812934875,
      "learning_rate": 0.0003385731765203159,
      "loss": 4.6575,
      "step": 23500
    },
    {
      "epoch": 0.4814367333565417,
      "grad_norm": 0.3081093728542328,
      "learning_rate": 0.0003384401269126697,
      "loss": 4.7299,
      "step": 23510
    },
    {
      "epoch": 0.48164151291135093,
      "grad_norm": 0.3001258373260498,
      "learning_rate": 0.000338306959573186,
      "loss": 4.7162,
      "step": 23520
    },
    {
      "epoch": 0.4818462924661602,
      "grad_norm": 0.3464072644710541,
      "learning_rate": 0.0003381736746151126,
      "loss": 4.6819,
      "step": 23530
    },
    {
      "epoch": 0.4820510720209694,
      "grad_norm": 0.2706254720687866,
      "learning_rate": 0.00033804027215179726,
      "loss": 4.7218,
      "step": 23540
    },
    {
      "epoch": 0.48225585157577866,
      "grad_norm": 0.28934282064437866,
      "learning_rate": 0.0003379067522966877,
      "loss": 4.715,
      "step": 23550
    },
    {
      "epoch": 0.4824606311305879,
      "grad_norm": 0.29465875029563904,
      "learning_rate": 0.0003377731151633315,
      "loss": 4.6931,
      "step": 23560
    },
    {
      "epoch": 0.48266541068539714,
      "grad_norm": 0.3002893626689911,
      "learning_rate": 0.00033763936086537587,
      "loss": 4.7003,
      "step": 23570
    },
    {
      "epoch": 0.48287019024020644,
      "grad_norm": 0.28075167536735535,
      "learning_rate": 0.00033750548951656784,
      "loss": 4.6993,
      "step": 23580
    },
    {
      "epoch": 0.4830749697950157,
      "grad_norm": 0.30318784713745117,
      "learning_rate": 0.0003373715012307538,
      "loss": 4.7241,
      "step": 23590
    },
    {
      "epoch": 0.4832797493498249,
      "grad_norm": 0.29993346333503723,
      "learning_rate": 0.0003372373961218797,
      "loss": 4.691,
      "step": 23600
    },
    {
      "epoch": 0.48348452890463417,
      "grad_norm": 0.282902330160141,
      "learning_rate": 0.00033710317430399084,
      "loss": 4.727,
      "step": 23610
    },
    {
      "epoch": 0.4836893084594434,
      "grad_norm": 0.27338647842407227,
      "learning_rate": 0.0003369688358912317,
      "loss": 4.7511,
      "step": 23620
    },
    {
      "epoch": 0.48389408801425265,
      "grad_norm": 0.2635759115219116,
      "learning_rate": 0.000336834380997846,
      "loss": 4.6827,
      "step": 23630
    },
    {
      "epoch": 0.4840988675690619,
      "grad_norm": 0.26382386684417725,
      "learning_rate": 0.00033669980973817634,
      "loss": 4.712,
      "step": 23640
    },
    {
      "epoch": 0.48430364712387114,
      "grad_norm": 0.26698169112205505,
      "learning_rate": 0.0003365651222266645,
      "loss": 4.7189,
      "step": 23650
    },
    {
      "epoch": 0.4845084266786804,
      "grad_norm": 0.2726365029811859,
      "learning_rate": 0.00033643031857785103,
      "loss": 4.6874,
      "step": 23660
    },
    {
      "epoch": 0.4847132062334896,
      "grad_norm": 0.26503750681877136,
      "learning_rate": 0.0003362953989063753,
      "loss": 4.6864,
      "step": 23670
    },
    {
      "epoch": 0.4849179857882989,
      "grad_norm": 0.29617050290107727,
      "learning_rate": 0.0003361603633269752,
      "loss": 4.707,
      "step": 23680
    },
    {
      "epoch": 0.48512276534310816,
      "grad_norm": 0.29559576511383057,
      "learning_rate": 0.0003360252119544872,
      "loss": 4.7303,
      "step": 23690
    },
    {
      "epoch": 0.4853275448979174,
      "grad_norm": 0.27144870162010193,
      "learning_rate": 0.00033588994490384654,
      "loss": 4.7226,
      "step": 23700
    },
    {
      "epoch": 0.48553232445272665,
      "grad_norm": 0.2561703026294708,
      "learning_rate": 0.00033575456229008647,
      "loss": 4.7255,
      "step": 23710
    },
    {
      "epoch": 0.4857371040075359,
      "grad_norm": 0.26624596118927,
      "learning_rate": 0.00033561906422833874,
      "loss": 4.7074,
      "step": 23720
    },
    {
      "epoch": 0.48594188356234513,
      "grad_norm": 0.28305935859680176,
      "learning_rate": 0.00033548345083383313,
      "loss": 4.7052,
      "step": 23730
    },
    {
      "epoch": 0.4861466631171544,
      "grad_norm": 0.2771523594856262,
      "learning_rate": 0.00033534772222189763,
      "loss": 4.717,
      "step": 23740
    },
    {
      "epoch": 0.4863514426719636,
      "grad_norm": 0.2788209915161133,
      "learning_rate": 0.0003352118785079582,
      "loss": 4.7351,
      "step": 23750
    },
    {
      "epoch": 0.48655622222677286,
      "grad_norm": 0.2830262780189514,
      "learning_rate": 0.00033507591980753857,
      "loss": 4.6921,
      "step": 23760
    },
    {
      "epoch": 0.4867610017815821,
      "grad_norm": 0.27070164680480957,
      "learning_rate": 0.00033493984623626035,
      "loss": 4.711,
      "step": 23770
    },
    {
      "epoch": 0.4869657813363914,
      "grad_norm": 0.2940160036087036,
      "learning_rate": 0.0003348036579098428,
      "loss": 4.6986,
      "step": 23780
    },
    {
      "epoch": 0.48717056089120064,
      "grad_norm": 0.2605251669883728,
      "learning_rate": 0.0003346673549441028,
      "loss": 4.7118,
      "step": 23790
    },
    {
      "epoch": 0.4873753404460099,
      "grad_norm": 1.1082611083984375,
      "learning_rate": 0.00033453093745495474,
      "loss": 4.7271,
      "step": 23800
    },
    {
      "epoch": 0.48758012000081913,
      "grad_norm": 0.2787920832633972,
      "learning_rate": 0.0003343944055584104,
      "loss": 4.7058,
      "step": 23810
    },
    {
      "epoch": 0.48778489955562837,
      "grad_norm": 0.26609036326408386,
      "learning_rate": 0.0003342577593705787,
      "loss": 4.7453,
      "step": 23820
    },
    {
      "epoch": 0.4879896791104376,
      "grad_norm": 0.2606726288795471,
      "learning_rate": 0.000334120999007666,
      "loss": 4.7264,
      "step": 23830
    },
    {
      "epoch": 0.48819445866524686,
      "grad_norm": 0.2958138585090637,
      "learning_rate": 0.00033398412458597555,
      "loss": 4.7058,
      "step": 23840
    },
    {
      "epoch": 0.4883992382200561,
      "grad_norm": 0.28008124232292175,
      "learning_rate": 0.0003338471362219078,
      "loss": 4.7463,
      "step": 23850
    },
    {
      "epoch": 0.48860401777486534,
      "grad_norm": 0.3037973642349243,
      "learning_rate": 0.0003337100340319599,
      "loss": 4.6901,
      "step": 23860
    },
    {
      "epoch": 0.4888087973296746,
      "grad_norm": 0.2755992114543915,
      "learning_rate": 0.0003335728181327258,
      "loss": 4.7143,
      "step": 23870
    },
    {
      "epoch": 0.4890135768844839,
      "grad_norm": 0.2757311165332794,
      "learning_rate": 0.00033343548864089644,
      "loss": 4.7267,
      "step": 23880
    },
    {
      "epoch": 0.4892183564392931,
      "grad_norm": 0.2758731544017792,
      "learning_rate": 0.000333298045673259,
      "loss": 4.7356,
      "step": 23890
    },
    {
      "epoch": 0.48942313599410237,
      "grad_norm": 0.2638200521469116,
      "learning_rate": 0.0003331604893466974,
      "loss": 4.727,
      "step": 23900
    },
    {
      "epoch": 0.4896279155489116,
      "grad_norm": 0.2907838225364685,
      "learning_rate": 0.0003330228197781918,
      "loss": 4.7076,
      "step": 23910
    },
    {
      "epoch": 0.48983269510372085,
      "grad_norm": 0.27843689918518066,
      "learning_rate": 0.00033288503708481877,
      "loss": 4.722,
      "step": 23920
    },
    {
      "epoch": 0.4900374746585301,
      "grad_norm": 0.33251845836639404,
      "learning_rate": 0.0003327471413837511,
      "loss": 4.7263,
      "step": 23930
    },
    {
      "epoch": 0.49024225421333933,
      "grad_norm": 0.2933971583843231,
      "learning_rate": 0.0003326091327922576,
      "loss": 4.698,
      "step": 23940
    },
    {
      "epoch": 0.4904470337681486,
      "grad_norm": 0.29155203700065613,
      "learning_rate": 0.0003324710114277031,
      "loss": 4.7129,
      "step": 23950
    },
    {
      "epoch": 0.4906518133229578,
      "grad_norm": 0.2822931408882141,
      "learning_rate": 0.00033233277740754845,
      "loss": 4.7123,
      "step": 23960
    },
    {
      "epoch": 0.49085659287776706,
      "grad_norm": 0.2747707664966583,
      "learning_rate": 0.00033219443084935,
      "loss": 4.6819,
      "step": 23970
    },
    {
      "epoch": 0.4910613724325763,
      "grad_norm": 0.2611922323703766,
      "learning_rate": 0.00033205597187076035,
      "loss": 4.7257,
      "step": 23980
    },
    {
      "epoch": 0.4912661519873856,
      "grad_norm": 0.28688105940818787,
      "learning_rate": 0.0003319174005895271,
      "loss": 4.7097,
      "step": 23990
    },
    {
      "epoch": 0.49147093154219484,
      "grad_norm": 0.28427520394325256,
      "learning_rate": 0.0003317787171234937,
      "loss": 4.745,
      "step": 24000
    },
    {
      "epoch": 0.49147093154219484,
      "eval_loss": 4.720487117767334,
      "eval_runtime": 4.3928,
      "eval_samples_per_second": 265.433,
      "eval_steps_per_second": 33.236,
      "step": 24000
    },
    {
      "epoch": 0.4916757110970041,
      "grad_norm": 0.2942088544368744,
      "learning_rate": 0.00033163992159059887,
      "loss": 4.724,
      "step": 24010
    },
    {
      "epoch": 0.49188049065181333,
      "grad_norm": 0.3350241184234619,
      "learning_rate": 0.0003315010141088768,
      "loss": 4.7125,
      "step": 24020
    },
    {
      "epoch": 0.49208527020662257,
      "grad_norm": 0.2688390016555786,
      "learning_rate": 0.0003313619947964568,
      "loss": 4.6787,
      "step": 24030
    },
    {
      "epoch": 0.4922900497614318,
      "grad_norm": 0.39939308166503906,
      "learning_rate": 0.00033122286377156306,
      "loss": 4.7109,
      "step": 24040
    },
    {
      "epoch": 0.49249482931624106,
      "grad_norm": 0.260379433631897,
      "learning_rate": 0.0003310836211525152,
      "loss": 4.7032,
      "step": 24050
    },
    {
      "epoch": 0.4926996088710503,
      "grad_norm": 0.28268808126449585,
      "learning_rate": 0.00033094426705772734,
      "loss": 4.7005,
      "step": 24060
    },
    {
      "epoch": 0.49290438842585954,
      "grad_norm": 0.2926914393901825,
      "learning_rate": 0.00033080480160570865,
      "loss": 4.697,
      "step": 24070
    },
    {
      "epoch": 0.4931091679806688,
      "grad_norm": 0.3147825002670288,
      "learning_rate": 0.00033066522491506287,
      "loss": 4.6947,
      "step": 24080
    },
    {
      "epoch": 0.4933139475354781,
      "grad_norm": 0.27262210845947266,
      "learning_rate": 0.00033052553710448846,
      "loss": 4.7289,
      "step": 24090
    },
    {
      "epoch": 0.4935187270902873,
      "grad_norm": 0.2598920464515686,
      "learning_rate": 0.0003303857382927783,
      "loss": 4.7219,
      "step": 24100
    },
    {
      "epoch": 0.49372350664509657,
      "grad_norm": 0.3277185261249542,
      "learning_rate": 0.00033024582859881956,
      "loss": 4.7224,
      "step": 24110
    },
    {
      "epoch": 0.4939282861999058,
      "grad_norm": 0.29700154066085815,
      "learning_rate": 0.000330105808141594,
      "loss": 4.7177,
      "step": 24120
    },
    {
      "epoch": 0.49413306575471505,
      "grad_norm": 0.2686825096607208,
      "learning_rate": 0.0003299656770401772,
      "loss": 4.7232,
      "step": 24130
    },
    {
      "epoch": 0.4943378453095243,
      "grad_norm": 0.2853672504425049,
      "learning_rate": 0.00032982543541373925,
      "loss": 4.6688,
      "step": 24140
    },
    {
      "epoch": 0.49454262486433354,
      "grad_norm": 0.30086320638656616,
      "learning_rate": 0.00032968508338154383,
      "loss": 4.7038,
      "step": 24150
    },
    {
      "epoch": 0.4947474044191428,
      "grad_norm": 0.32596907019615173,
      "learning_rate": 0.0003295446210629488,
      "loss": 4.7365,
      "step": 24160
    },
    {
      "epoch": 0.494952183973952,
      "grad_norm": 0.266591876745224,
      "learning_rate": 0.0003294040485774057,
      "loss": 4.7252,
      "step": 24170
    },
    {
      "epoch": 0.49515696352876126,
      "grad_norm": 0.42513614892959595,
      "learning_rate": 0.0003292633660444597,
      "loss": 4.7194,
      "step": 24180
    },
    {
      "epoch": 0.49536174308357056,
      "grad_norm": 0.25697004795074463,
      "learning_rate": 0.00032912257358374967,
      "loss": 4.7353,
      "step": 24190
    },
    {
      "epoch": 0.4955665226383798,
      "grad_norm": 0.2750478684902191,
      "learning_rate": 0.00032898167131500787,
      "loss": 4.706,
      "step": 24200
    },
    {
      "epoch": 0.49577130219318905,
      "grad_norm": 0.2928435504436493,
      "learning_rate": 0.00032884065935806,
      "loss": 4.7039,
      "step": 24210
    },
    {
      "epoch": 0.4959760817479983,
      "grad_norm": 0.26540809869766235,
      "learning_rate": 0.0003286995378328251,
      "loss": 4.7086,
      "step": 24220
    },
    {
      "epoch": 0.49618086130280753,
      "grad_norm": 0.27740904688835144,
      "learning_rate": 0.00032855830685931515,
      "loss": 4.7378,
      "step": 24230
    },
    {
      "epoch": 0.4963856408576168,
      "grad_norm": 0.27485188841819763,
      "learning_rate": 0.0003284169665576355,
      "loss": 4.7151,
      "step": 24240
    },
    {
      "epoch": 0.496590420412426,
      "grad_norm": 0.37828922271728516,
      "learning_rate": 0.00032827551704798434,
      "loss": 4.6945,
      "step": 24250
    },
    {
      "epoch": 0.49679519996723526,
      "grad_norm": 0.28058096766471863,
      "learning_rate": 0.0003281339584506527,
      "loss": 4.721,
      "step": 24260
    },
    {
      "epoch": 0.4969999795220445,
      "grad_norm": 0.25854894518852234,
      "learning_rate": 0.0003279922908860244,
      "loss": 4.6989,
      "step": 24270
    },
    {
      "epoch": 0.49720475907685374,
      "grad_norm": 0.26653802394866943,
      "learning_rate": 0.00032785051447457603,
      "loss": 4.7013,
      "step": 24280
    },
    {
      "epoch": 0.49740953863166304,
      "grad_norm": 0.28719645738601685,
      "learning_rate": 0.00032770862933687657,
      "loss": 4.7329,
      "step": 24290
    },
    {
      "epoch": 0.4976143181864723,
      "grad_norm": 0.28432992100715637,
      "learning_rate": 0.00032756663559358764,
      "loss": 4.6853,
      "step": 24300
    },
    {
      "epoch": 0.4978190977412815,
      "grad_norm": 0.3140254318714142,
      "learning_rate": 0.000327424533365463,
      "loss": 4.7235,
      "step": 24310
    },
    {
      "epoch": 0.49802387729609077,
      "grad_norm": 0.28151458501815796,
      "learning_rate": 0.0003272823227733489,
      "loss": 4.6886,
      "step": 24320
    },
    {
      "epoch": 0.4982286568509,
      "grad_norm": 0.2725624740123749,
      "learning_rate": 0.00032714000393818367,
      "loss": 4.7217,
      "step": 24330
    },
    {
      "epoch": 0.49843343640570925,
      "grad_norm": 0.33774906396865845,
      "learning_rate": 0.0003269975769809976,
      "loss": 4.719,
      "step": 24340
    },
    {
      "epoch": 0.4986382159605185,
      "grad_norm": 0.28247374296188354,
      "learning_rate": 0.000326855042022913,
      "loss": 4.645,
      "step": 24350
    },
    {
      "epoch": 0.49884299551532774,
      "grad_norm": 0.2772592306137085,
      "learning_rate": 0.0003267123991851441,
      "loss": 4.6848,
      "step": 24360
    },
    {
      "epoch": 0.499047775070137,
      "grad_norm": 0.291071355342865,
      "learning_rate": 0.00032656964858899666,
      "loss": 4.7023,
      "step": 24370
    },
    {
      "epoch": 0.4992525546249462,
      "grad_norm": 0.2632277309894562,
      "learning_rate": 0.00032642679035586837,
      "loss": 4.7196,
      "step": 24380
    },
    {
      "epoch": 0.4994573341797555,
      "grad_norm": 0.27937376499176025,
      "learning_rate": 0.0003262838246072481,
      "loss": 4.722,
      "step": 24390
    },
    {
      "epoch": 0.49966211373456476,
      "grad_norm": 0.27597543597221375,
      "learning_rate": 0.0003261407514647166,
      "loss": 4.698,
      "step": 24400
    },
    {
      "epoch": 0.499866893289374,
      "grad_norm": 0.2707703709602356,
      "learning_rate": 0.00032599757104994547,
      "loss": 4.7336,
      "step": 24410
    },
    {
      "epoch": 0.5000716728441832,
      "grad_norm": 0.2789210081100464,
      "learning_rate": 0.00032585428348469794,
      "loss": 4.6859,
      "step": 24420
    },
    {
      "epoch": 0.5002764523989924,
      "grad_norm": 0.32933616638183594,
      "learning_rate": 0.00032571088889082813,
      "loss": 4.6719,
      "step": 24430
    },
    {
      "epoch": 0.5004812319538018,
      "grad_norm": 0.28977784514427185,
      "learning_rate": 0.0003255673873902812,
      "loss": 4.6751,
      "step": 24440
    },
    {
      "epoch": 0.500686011508611,
      "grad_norm": 0.30091533064842224,
      "learning_rate": 0.00032542377910509335,
      "loss": 4.6725,
      "step": 24450
    },
    {
      "epoch": 0.5008907910634203,
      "grad_norm": 0.29741963744163513,
      "learning_rate": 0.0003252800641573914,
      "loss": 4.7524,
      "step": 24460
    },
    {
      "epoch": 0.5010955706182295,
      "grad_norm": 0.2790796458721161,
      "learning_rate": 0.0003251362426693931,
      "loss": 4.6917,
      "step": 24470
    },
    {
      "epoch": 0.5013003501730388,
      "grad_norm": 0.2658741772174835,
      "learning_rate": 0.00032499231476340666,
      "loss": 4.6869,
      "step": 24480
    },
    {
      "epoch": 0.501505129727848,
      "grad_norm": 0.28098657727241516,
      "learning_rate": 0.0003248482805618307,
      "loss": 4.7246,
      "step": 24490
    },
    {
      "epoch": 0.5017099092826572,
      "grad_norm": 0.28023362159729004,
      "learning_rate": 0.00032470414018715446,
      "loss": 4.7324,
      "step": 24500
    },
    {
      "epoch": 0.5019146888374665,
      "grad_norm": 0.2921772599220276,
      "learning_rate": 0.00032455989376195744,
      "loss": 4.7123,
      "step": 24510
    },
    {
      "epoch": 0.5021194683922757,
      "grad_norm": 0.2592458724975586,
      "learning_rate": 0.000324415541408909,
      "loss": 4.71,
      "step": 24520
    },
    {
      "epoch": 0.502324247947085,
      "grad_norm": 0.2661096751689911,
      "learning_rate": 0.0003242710832507692,
      "loss": 4.7153,
      "step": 24530
    },
    {
      "epoch": 0.5025290275018942,
      "grad_norm": 0.2882750928401947,
      "learning_rate": 0.00032412651941038736,
      "loss": 4.7062,
      "step": 24540
    },
    {
      "epoch": 0.5027338070567035,
      "grad_norm": 0.29600468277931213,
      "learning_rate": 0.0003239818500107032,
      "loss": 4.6724,
      "step": 24550
    },
    {
      "epoch": 0.5029385866115127,
      "grad_norm": 0.25009870529174805,
      "learning_rate": 0.00032383707517474604,
      "loss": 4.7344,
      "step": 24560
    },
    {
      "epoch": 0.5031433661663219,
      "grad_norm": 0.2783048748970032,
      "learning_rate": 0.0003236921950256348,
      "loss": 4.7215,
      "step": 24570
    },
    {
      "epoch": 0.5033481457211312,
      "grad_norm": 0.27534037828445435,
      "learning_rate": 0.00032354720968657814,
      "loss": 4.7264,
      "step": 24580
    },
    {
      "epoch": 0.5035529252759404,
      "grad_norm": 0.2851855754852295,
      "learning_rate": 0.0003234021192808739,
      "loss": 4.7154,
      "step": 24590
    },
    {
      "epoch": 0.5037577048307497,
      "grad_norm": 0.2746172249317169,
      "learning_rate": 0.0003232569239319095,
      "loss": 4.7098,
      "step": 24600
    },
    {
      "epoch": 0.5039624843855589,
      "grad_norm": 0.27588728070259094,
      "learning_rate": 0.0003231116237631616,
      "loss": 4.7102,
      "step": 24610
    },
    {
      "epoch": 0.5041672639403681,
      "grad_norm": 0.3069898784160614,
      "learning_rate": 0.00032296621889819585,
      "loss": 4.6668,
      "step": 24620
    },
    {
      "epoch": 0.5043720434951774,
      "grad_norm": 0.28509026765823364,
      "learning_rate": 0.00032282070946066706,
      "loss": 4.7222,
      "step": 24630
    },
    {
      "epoch": 0.5045768230499867,
      "grad_norm": 0.25804194808006287,
      "learning_rate": 0.00032267509557431883,
      "loss": 4.6905,
      "step": 24640
    },
    {
      "epoch": 0.504781602604796,
      "grad_norm": 0.27046477794647217,
      "learning_rate": 0.00032252937736298385,
      "loss": 4.6901,
      "step": 24650
    },
    {
      "epoch": 0.5049863821596052,
      "grad_norm": 0.2739366888999939,
      "learning_rate": 0.00032238355495058324,
      "loss": 4.7051,
      "step": 24660
    },
    {
      "epoch": 0.5051911617144145,
      "grad_norm": 0.2906496226787567,
      "learning_rate": 0.00032223762846112685,
      "loss": 4.7007,
      "step": 24670
    },
    {
      "epoch": 0.5053959412692237,
      "grad_norm": 0.2622486352920532,
      "learning_rate": 0.0003220915980187132,
      "loss": 4.7171,
      "step": 24680
    },
    {
      "epoch": 0.505600720824033,
      "grad_norm": 0.30467307567596436,
      "learning_rate": 0.00032194546374752884,
      "loss": 4.6864,
      "step": 24690
    },
    {
      "epoch": 0.5058055003788422,
      "grad_norm": 0.2761250436306,
      "learning_rate": 0.000321799225771849,
      "loss": 4.708,
      "step": 24700
    },
    {
      "epoch": 0.5060102799336514,
      "grad_norm": 0.2742674946784973,
      "learning_rate": 0.0003216528842160369,
      "loss": 4.6926,
      "step": 24710
    },
    {
      "epoch": 0.5062150594884607,
      "grad_norm": 0.3191431760787964,
      "learning_rate": 0.0003215064392045439,
      "loss": 4.7402,
      "step": 24720
    },
    {
      "epoch": 0.5064198390432699,
      "grad_norm": 0.27406784892082214,
      "learning_rate": 0.00032135989086190926,
      "loss": 4.7149,
      "step": 24730
    },
    {
      "epoch": 0.5066246185980792,
      "grad_norm": 0.29739177227020264,
      "learning_rate": 0.00032121323931276024,
      "loss": 4.6829,
      "step": 24740
    },
    {
      "epoch": 0.5068293981528884,
      "grad_norm": 0.2993490993976593,
      "learning_rate": 0.00032106648468181183,
      "loss": 4.7107,
      "step": 24750
    },
    {
      "epoch": 0.5070341777076977,
      "grad_norm": 0.2782713770866394,
      "learning_rate": 0.00032091962709386666,
      "loss": 4.694,
      "step": 24760
    },
    {
      "epoch": 0.5072389572625069,
      "grad_norm": 0.2802233099937439,
      "learning_rate": 0.0003207726666738148,
      "loss": 4.6925,
      "step": 24770
    },
    {
      "epoch": 0.5074437368173161,
      "grad_norm": 0.27374786138534546,
      "learning_rate": 0.0003206256035466341,
      "loss": 4.7394,
      "step": 24780
    },
    {
      "epoch": 0.5076485163721254,
      "grad_norm": 0.2811621129512787,
      "learning_rate": 0.00032047843783738935,
      "loss": 4.7339,
      "step": 24790
    },
    {
      "epoch": 0.5078532959269346,
      "grad_norm": 0.28678399324417114,
      "learning_rate": 0.0003203311696712328,
      "loss": 4.7298,
      "step": 24800
    },
    {
      "epoch": 0.5080580754817439,
      "grad_norm": 0.26202142238616943,
      "learning_rate": 0.0003201837991734039,
      "loss": 4.7034,
      "step": 24810
    },
    {
      "epoch": 0.5082628550365531,
      "grad_norm": 0.2799505889415741,
      "learning_rate": 0.0003200363264692291,
      "loss": 4.7083,
      "step": 24820
    },
    {
      "epoch": 0.5084676345913623,
      "grad_norm": 0.29472848773002625,
      "learning_rate": 0.0003198887516841214,
      "loss": 4.7215,
      "step": 24830
    },
    {
      "epoch": 0.5086724141461716,
      "grad_norm": 0.2675095796585083,
      "learning_rate": 0.0003197410749435811,
      "loss": 4.715,
      "step": 24840
    },
    {
      "epoch": 0.508877193700981,
      "grad_norm": 0.2702561318874359,
      "learning_rate": 0.00031959329637319504,
      "loss": 4.6935,
      "step": 24850
    },
    {
      "epoch": 0.5090819732557902,
      "grad_norm": 0.30072882771492004,
      "learning_rate": 0.00031944541609863663,
      "loss": 4.7057,
      "step": 24860
    },
    {
      "epoch": 0.5092867528105994,
      "grad_norm": 0.2939676344394684,
      "learning_rate": 0.0003192974342456656,
      "loss": 4.7177,
      "step": 24870
    },
    {
      "epoch": 0.5094915323654087,
      "grad_norm": 0.2770099639892578,
      "learning_rate": 0.0003191493509401284,
      "loss": 4.7156,
      "step": 24880
    },
    {
      "epoch": 0.5096963119202179,
      "grad_norm": 0.2761791944503784,
      "learning_rate": 0.00031900116630795755,
      "loss": 4.716,
      "step": 24890
    },
    {
      "epoch": 0.5099010914750272,
      "grad_norm": 0.2724962532520294,
      "learning_rate": 0.0003188528804751718,
      "loss": 4.6724,
      "step": 24900
    },
    {
      "epoch": 0.5101058710298364,
      "grad_norm": 0.26774343848228455,
      "learning_rate": 0.0003187044935678758,
      "loss": 4.6928,
      "step": 24910
    },
    {
      "epoch": 0.5103106505846456,
      "grad_norm": 0.3066468834877014,
      "learning_rate": 0.00031855600571226053,
      "loss": 4.7164,
      "step": 24920
    },
    {
      "epoch": 0.5105154301394549,
      "grad_norm": 0.26715534925460815,
      "learning_rate": 0.00031840741703460237,
      "loss": 4.6918,
      "step": 24930
    },
    {
      "epoch": 0.5107202096942641,
      "grad_norm": 0.2691773772239685,
      "learning_rate": 0.00031825872766126385,
      "loss": 4.6799,
      "step": 24940
    },
    {
      "epoch": 0.5109249892490734,
      "grad_norm": 0.3044761121273041,
      "learning_rate": 0.00031810993771869275,
      "loss": 4.7003,
      "step": 24950
    },
    {
      "epoch": 0.5111297688038826,
      "grad_norm": 0.2579299807548523,
      "learning_rate": 0.0003179610473334228,
      "loss": 4.7094,
      "step": 24960
    },
    {
      "epoch": 0.5113345483586919,
      "grad_norm": 0.27381426095962524,
      "learning_rate": 0.0003178120566320727,
      "loss": 4.73,
      "step": 24970
    },
    {
      "epoch": 0.5115393279135011,
      "grad_norm": 0.2825131118297577,
      "learning_rate": 0.0003176629657413467,
      "loss": 4.6995,
      "step": 24980
    },
    {
      "epoch": 0.5117441074683103,
      "grad_norm": 0.29649683833122253,
      "learning_rate": 0.0003175137747880344,
      "loss": 4.6955,
      "step": 24990
    },
    {
      "epoch": 0.5119488870231196,
      "grad_norm": 0.2876654267311096,
      "learning_rate": 0.0003173644838990101,
      "loss": 4.6683,
      "step": 25000
    },
    {
      "epoch": 0.5121536665779288,
      "grad_norm": 0.3204991817474365,
      "learning_rate": 0.0003172150932012335,
      "loss": 4.6938,
      "step": 25010
    },
    {
      "epoch": 0.5123584461327381,
      "grad_norm": 0.2765672206878662,
      "learning_rate": 0.00031706560282174885,
      "loss": 4.7108,
      "step": 25020
    },
    {
      "epoch": 0.5125632256875473,
      "grad_norm": 0.358657568693161,
      "learning_rate": 0.0003169160128876854,
      "loss": 4.7241,
      "step": 25030
    },
    {
      "epoch": 0.5127680052423566,
      "grad_norm": 0.30940568447113037,
      "learning_rate": 0.0003167663235262569,
      "loss": 4.6864,
      "step": 25040
    },
    {
      "epoch": 0.5129727847971659,
      "grad_norm": 0.28657031059265137,
      "learning_rate": 0.0003166165348647618,
      "loss": 4.6723,
      "step": 25050
    },
    {
      "epoch": 0.5131775643519751,
      "grad_norm": 0.2804965674877167,
      "learning_rate": 0.00031646664703058287,
      "loss": 4.6655,
      "step": 25060
    },
    {
      "epoch": 0.5133823439067844,
      "grad_norm": 0.29186683893203735,
      "learning_rate": 0.00031631666015118727,
      "loss": 4.6773,
      "step": 25070
    },
    {
      "epoch": 0.5135871234615936,
      "grad_norm": 0.3029853403568268,
      "learning_rate": 0.00031616657435412643,
      "loss": 4.7079,
      "step": 25080
    },
    {
      "epoch": 0.5137919030164029,
      "grad_norm": 0.273480087518692,
      "learning_rate": 0.0003160163897670359,
      "loss": 4.6923,
      "step": 25090
    },
    {
      "epoch": 0.5139966825712121,
      "grad_norm": 0.2673560678958893,
      "learning_rate": 0.0003158661065176351,
      "loss": 4.7063,
      "step": 25100
    },
    {
      "epoch": 0.5142014621260214,
      "grad_norm": 0.29640883207321167,
      "learning_rate": 0.0003157157247337276,
      "loss": 4.6679,
      "step": 25110
    },
    {
      "epoch": 0.5144062416808306,
      "grad_norm": 0.2878381907939911,
      "learning_rate": 0.00031556524454320043,
      "loss": 4.6889,
      "step": 25120
    },
    {
      "epoch": 0.5146110212356398,
      "grad_norm": 0.2771507799625397,
      "learning_rate": 0.00031541466607402463,
      "loss": 4.7019,
      "step": 25130
    },
    {
      "epoch": 0.5148158007904491,
      "grad_norm": 0.28070971369743347,
      "learning_rate": 0.00031526398945425473,
      "loss": 4.7209,
      "step": 25140
    },
    {
      "epoch": 0.5150205803452583,
      "grad_norm": 0.34403181076049805,
      "learning_rate": 0.0003151132148120286,
      "loss": 4.6986,
      "step": 25150
    },
    {
      "epoch": 0.5152253599000676,
      "grad_norm": 0.31226861476898193,
      "learning_rate": 0.0003149623422755676,
      "loss": 4.6746,
      "step": 25160
    },
    {
      "epoch": 0.5154301394548768,
      "grad_norm": 0.48295944929122925,
      "learning_rate": 0.0003148113719731763,
      "loss": 4.7339,
      "step": 25170
    },
    {
      "epoch": 0.515634919009686,
      "grad_norm": 0.29848966002464294,
      "learning_rate": 0.00031466030403324235,
      "loss": 4.7034,
      "step": 25180
    },
    {
      "epoch": 0.5158396985644953,
      "grad_norm": 0.2727344036102295,
      "learning_rate": 0.0003145091385842366,
      "loss": 4.7082,
      "step": 25190
    },
    {
      "epoch": 0.5160444781193045,
      "grad_norm": 0.2666777968406677,
      "learning_rate": 0.00031435787575471264,
      "loss": 4.6865,
      "step": 25200
    },
    {
      "epoch": 0.5162492576741138,
      "grad_norm": 0.2528291940689087,
      "learning_rate": 0.000314206515673307,
      "loss": 4.7122,
      "step": 25210
    },
    {
      "epoch": 0.516454037228923,
      "grad_norm": 0.27205219864845276,
      "learning_rate": 0.0003140550584687388,
      "loss": 4.6817,
      "step": 25220
    },
    {
      "epoch": 0.5166588167837323,
      "grad_norm": 0.283069908618927,
      "learning_rate": 0.00031390350426980986,
      "loss": 4.6998,
      "step": 25230
    },
    {
      "epoch": 0.5168635963385415,
      "grad_norm": 0.2868604063987732,
      "learning_rate": 0.0003137518532054045,
      "loss": 4.7136,
      "step": 25240
    },
    {
      "epoch": 0.5170683758933509,
      "grad_norm": 0.2935883402824402,
      "learning_rate": 0.0003136001054044893,
      "loss": 4.7054,
      "step": 25250
    },
    {
      "epoch": 0.5172731554481601,
      "grad_norm": 0.28250062465667725,
      "learning_rate": 0.00031344826099611315,
      "loss": 4.7147,
      "step": 25260
    },
    {
      "epoch": 0.5174779350029693,
      "grad_norm": 0.2601027488708496,
      "learning_rate": 0.0003132963201094072,
      "loss": 4.7112,
      "step": 25270
    },
    {
      "epoch": 0.5176827145577786,
      "grad_norm": 0.2703922986984253,
      "learning_rate": 0.00031314428287358454,
      "loss": 4.6906,
      "step": 25280
    },
    {
      "epoch": 0.5178874941125878,
      "grad_norm": 0.2875763475894928,
      "learning_rate": 0.0003129921494179401,
      "loss": 4.6944,
      "step": 25290
    },
    {
      "epoch": 0.5180922736673971,
      "grad_norm": 0.3172433078289032,
      "learning_rate": 0.0003128399198718509,
      "loss": 4.6961,
      "step": 25300
    },
    {
      "epoch": 0.5182970532222063,
      "grad_norm": 0.28155481815338135,
      "learning_rate": 0.0003126875943647755,
      "loss": 4.7292,
      "step": 25310
    },
    {
      "epoch": 0.5185018327770156,
      "grad_norm": 0.30049318075180054,
      "learning_rate": 0.0003125351730262541,
      "loss": 4.7018,
      "step": 25320
    },
    {
      "epoch": 0.5187066123318248,
      "grad_norm": 0.2589489221572876,
      "learning_rate": 0.0003123826559859083,
      "loss": 4.6716,
      "step": 25330
    },
    {
      "epoch": 0.518911391886634,
      "grad_norm": 0.28580641746520996,
      "learning_rate": 0.0003122300433734413,
      "loss": 4.7012,
      "step": 25340
    },
    {
      "epoch": 0.5191161714414433,
      "grad_norm": 0.2864598035812378,
      "learning_rate": 0.0003120773353186375,
      "loss": 4.7456,
      "step": 25350
    },
    {
      "epoch": 0.5193209509962525,
      "grad_norm": 0.2721398174762726,
      "learning_rate": 0.0003119245319513623,
      "loss": 4.6883,
      "step": 25360
    },
    {
      "epoch": 0.5195257305510618,
      "grad_norm": 0.28199484944343567,
      "learning_rate": 0.0003117716334015624,
      "loss": 4.7168,
      "step": 25370
    },
    {
      "epoch": 0.519730510105871,
      "grad_norm": 0.2768019437789917,
      "learning_rate": 0.0003116186397992652,
      "loss": 4.7534,
      "step": 25380
    },
    {
      "epoch": 0.5199352896606803,
      "grad_norm": 0.27264705300331116,
      "learning_rate": 0.0003114655512745791,
      "loss": 4.6925,
      "step": 25390
    },
    {
      "epoch": 0.5201400692154895,
      "grad_norm": 0.2672605514526367,
      "learning_rate": 0.00031131236795769324,
      "loss": 4.6965,
      "step": 25400
    },
    {
      "epoch": 0.5203448487702987,
      "grad_norm": 0.29574844241142273,
      "learning_rate": 0.00031115908997887725,
      "loss": 4.6935,
      "step": 25410
    },
    {
      "epoch": 0.520549628325108,
      "grad_norm": 0.2922261655330658,
      "learning_rate": 0.00031100571746848143,
      "loss": 4.7158,
      "step": 25420
    },
    {
      "epoch": 0.5207544078799172,
      "grad_norm": 0.26851898431777954,
      "learning_rate": 0.0003108522505569362,
      "loss": 4.7006,
      "step": 25430
    },
    {
      "epoch": 0.5209591874347265,
      "grad_norm": 0.33217042684555054,
      "learning_rate": 0.0003106986893747526,
      "loss": 4.6862,
      "step": 25440
    },
    {
      "epoch": 0.5211639669895357,
      "grad_norm": 0.31240522861480713,
      "learning_rate": 0.00031054503405252167,
      "loss": 4.693,
      "step": 25450
    },
    {
      "epoch": 0.5213687465443451,
      "grad_norm": 0.2913259267807007,
      "learning_rate": 0.0003103912847209143,
      "loss": 4.6595,
      "step": 25460
    },
    {
      "epoch": 0.5215735260991543,
      "grad_norm": 0.2755216658115387,
      "learning_rate": 0.0003102374415106818,
      "loss": 4.7245,
      "step": 25470
    },
    {
      "epoch": 0.5217783056539635,
      "grad_norm": 0.30526289343833923,
      "learning_rate": 0.0003100835045526548,
      "loss": 4.6517,
      "step": 25480
    },
    {
      "epoch": 0.5219830852087728,
      "grad_norm": 0.27394744753837585,
      "learning_rate": 0.0003099294739777441,
      "loss": 4.7576,
      "step": 25490
    },
    {
      "epoch": 0.522187864763582,
      "grad_norm": 0.29140064120292664,
      "learning_rate": 0.00030977534991693986,
      "loss": 4.7065,
      "step": 25500
    },
    {
      "epoch": 0.5223926443183913,
      "grad_norm": 0.27797731757164,
      "learning_rate": 0.00030962113250131184,
      "loss": 4.6969,
      "step": 25510
    },
    {
      "epoch": 0.5225974238732005,
      "grad_norm": 0.2931949198246002,
      "learning_rate": 0.00030946682186200905,
      "loss": 4.7124,
      "step": 25520
    },
    {
      "epoch": 0.5228022034280098,
      "grad_norm": 0.2820892035961151,
      "learning_rate": 0.0003093124181302599,
      "loss": 4.7162,
      "step": 25530
    },
    {
      "epoch": 0.523006982982819,
      "grad_norm": 0.2782002389431,
      "learning_rate": 0.00030915792143737205,
      "loss": 4.7207,
      "step": 25540
    },
    {
      "epoch": 0.5232117625376282,
      "grad_norm": 0.31584373116493225,
      "learning_rate": 0.00030900333191473205,
      "loss": 4.7256,
      "step": 25550
    },
    {
      "epoch": 0.5234165420924375,
      "grad_norm": 0.28503715991973877,
      "learning_rate": 0.0003088486496938054,
      "loss": 4.6784,
      "step": 25560
    },
    {
      "epoch": 0.5236213216472467,
      "grad_norm": 0.32100310921669006,
      "learning_rate": 0.0003086938749061366,
      "loss": 4.6991,
      "step": 25570
    },
    {
      "epoch": 0.523826101202056,
      "grad_norm": 0.25826606154441833,
      "learning_rate": 0.0003085390076833487,
      "loss": 4.6634,
      "step": 25580
    },
    {
      "epoch": 0.5240308807568652,
      "grad_norm": 0.2814335823059082,
      "learning_rate": 0.00030838404815714345,
      "loss": 4.7095,
      "step": 25590
    },
    {
      "epoch": 0.5242356603116745,
      "grad_norm": 0.2817402780056,
      "learning_rate": 0.00030822899645930105,
      "loss": 4.6766,
      "step": 25600
    },
    {
      "epoch": 0.5244404398664837,
      "grad_norm": 0.2802383005619049,
      "learning_rate": 0.00030807385272168,
      "loss": 4.6997,
      "step": 25610
    },
    {
      "epoch": 0.5246452194212929,
      "grad_norm": 0.3037165701389313,
      "learning_rate": 0.00030791861707621734,
      "loss": 4.7262,
      "step": 25620
    },
    {
      "epoch": 0.5248499989761022,
      "grad_norm": 0.3113444447517395,
      "learning_rate": 0.00030776328965492804,
      "loss": 4.6834,
      "step": 25630
    },
    {
      "epoch": 0.5250547785309114,
      "grad_norm": 0.3027448356151581,
      "learning_rate": 0.0003076078705899051,
      "loss": 4.6792,
      "step": 25640
    },
    {
      "epoch": 0.5252595580857207,
      "grad_norm": 0.2720443606376648,
      "learning_rate": 0.0003074523600133197,
      "loss": 4.6895,
      "step": 25650
    },
    {
      "epoch": 0.52546433764053,
      "grad_norm": 0.30578285455703735,
      "learning_rate": 0.0003072967580574206,
      "loss": 4.6978,
      "step": 25660
    },
    {
      "epoch": 0.5256691171953393,
      "grad_norm": 0.2918396294116974,
      "learning_rate": 0.0003071410648545342,
      "loss": 4.6637,
      "step": 25670
    },
    {
      "epoch": 0.5258738967501485,
      "grad_norm": 0.2783021330833435,
      "learning_rate": 0.00030698528053706484,
      "loss": 4.6948,
      "step": 25680
    },
    {
      "epoch": 0.5260786763049577,
      "grad_norm": 0.2891443371772766,
      "learning_rate": 0.00030682940523749406,
      "loss": 4.7025,
      "step": 25690
    },
    {
      "epoch": 0.526283455859767,
      "grad_norm": 0.2706538736820221,
      "learning_rate": 0.00030667343908838103,
      "loss": 4.6635,
      "step": 25700
    },
    {
      "epoch": 0.5264882354145762,
      "grad_norm": 0.2786348760128021,
      "learning_rate": 0.0003065173822223617,
      "loss": 4.7087,
      "step": 25710
    },
    {
      "epoch": 0.5266930149693855,
      "grad_norm": 0.2681735157966614,
      "learning_rate": 0.0003063612347721497,
      "loss": 4.7101,
      "step": 25720
    },
    {
      "epoch": 0.5268977945241947,
      "grad_norm": 0.3078657388687134,
      "learning_rate": 0.0003062049968705355,
      "loss": 4.6899,
      "step": 25730
    },
    {
      "epoch": 0.527102574079004,
      "grad_norm": 0.2678840160369873,
      "learning_rate": 0.00030604866865038637,
      "loss": 4.6822,
      "step": 25740
    },
    {
      "epoch": 0.5273073536338132,
      "grad_norm": 0.30802249908447266,
      "learning_rate": 0.00030589225024464656,
      "loss": 4.6478,
      "step": 25750
    },
    {
      "epoch": 0.5275121331886224,
      "grad_norm": 0.2710878849029541,
      "learning_rate": 0.0003057357417863368,
      "loss": 4.6963,
      "step": 25760
    },
    {
      "epoch": 0.5277169127434317,
      "grad_norm": 0.2686658501625061,
      "learning_rate": 0.0003055791434085547,
      "loss": 4.7031,
      "step": 25770
    },
    {
      "epoch": 0.5279216922982409,
      "grad_norm": 0.2729041576385498,
      "learning_rate": 0.0003054224552444742,
      "loss": 4.7009,
      "step": 25780
    },
    {
      "epoch": 0.5281264718530502,
      "grad_norm": 0.27089405059814453,
      "learning_rate": 0.00030526567742734543,
      "loss": 4.6705,
      "step": 25790
    },
    {
      "epoch": 0.5283312514078594,
      "grad_norm": 0.2662869095802307,
      "learning_rate": 0.000305108810090495,
      "loss": 4.6951,
      "step": 25800
    },
    {
      "epoch": 0.5285360309626687,
      "grad_norm": 0.27419552206993103,
      "learning_rate": 0.00030495185336732556,
      "loss": 4.6835,
      "step": 25810
    },
    {
      "epoch": 0.5287408105174779,
      "grad_norm": 0.2641861140727997,
      "learning_rate": 0.00030479480739131575,
      "loss": 4.6933,
      "step": 25820
    },
    {
      "epoch": 0.5289455900722871,
      "grad_norm": 0.29191505908966064,
      "learning_rate": 0.0003046376722960201,
      "loss": 4.69,
      "step": 25830
    },
    {
      "epoch": 0.5291503696270964,
      "grad_norm": 0.2895081043243408,
      "learning_rate": 0.0003044804482150691,
      "loss": 4.6974,
      "step": 25840
    },
    {
      "epoch": 0.5293551491819056,
      "grad_norm": 0.28528445959091187,
      "learning_rate": 0.00030432313528216867,
      "loss": 4.7084,
      "step": 25850
    },
    {
      "epoch": 0.529559928736715,
      "grad_norm": 0.2768077850341797,
      "learning_rate": 0.0003041657336311004,
      "loss": 4.6845,
      "step": 25860
    },
    {
      "epoch": 0.5297647082915242,
      "grad_norm": 0.33595651388168335,
      "learning_rate": 0.0003040082433957214,
      "loss": 4.6987,
      "step": 25870
    },
    {
      "epoch": 0.5299694878463335,
      "grad_norm": 0.27888068556785583,
      "learning_rate": 0.00030385066470996403,
      "loss": 4.6871,
      "step": 25880
    },
    {
      "epoch": 0.5301742674011427,
      "grad_norm": 0.270297110080719,
      "learning_rate": 0.00030369299770783583,
      "loss": 4.689,
      "step": 25890
    },
    {
      "epoch": 0.530379046955952,
      "grad_norm": 0.27477559447288513,
      "learning_rate": 0.0003035352425234195,
      "loss": 4.7047,
      "step": 25900
    },
    {
      "epoch": 0.5305838265107612,
      "grad_norm": 0.3273790776729584,
      "learning_rate": 0.0003033773992908728,
      "loss": 4.7141,
      "step": 25910
    },
    {
      "epoch": 0.5307886060655704,
      "grad_norm": 0.31001225113868713,
      "learning_rate": 0.0003032194681444282,
      "loss": 4.7561,
      "step": 25920
    },
    {
      "epoch": 0.5309933856203797,
      "grad_norm": 0.2798680365085602,
      "learning_rate": 0.0003030614492183931,
      "loss": 4.7198,
      "step": 25930
    },
    {
      "epoch": 0.5311981651751889,
      "grad_norm": 0.27496957778930664,
      "learning_rate": 0.00030290334264714957,
      "loss": 4.7175,
      "step": 25940
    },
    {
      "epoch": 0.5314029447299982,
      "grad_norm": 0.2571429908275604,
      "learning_rate": 0.000302745148565154,
      "loss": 4.7221,
      "step": 25950
    },
    {
      "epoch": 0.5316077242848074,
      "grad_norm": 0.2705236077308655,
      "learning_rate": 0.0003025868671069373,
      "loss": 4.6772,
      "step": 25960
    },
    {
      "epoch": 0.5318125038396166,
      "grad_norm": 0.2997802495956421,
      "learning_rate": 0.0003024284984071047,
      "loss": 4.7076,
      "step": 25970
    },
    {
      "epoch": 0.5320172833944259,
      "grad_norm": 0.3122205436229706,
      "learning_rate": 0.00030227004260033574,
      "loss": 4.6439,
      "step": 25980
    },
    {
      "epoch": 0.5322220629492351,
      "grad_norm": 0.3267391622066498,
      "learning_rate": 0.00030211149982138397,
      "loss": 4.7256,
      "step": 25990
    },
    {
      "epoch": 0.5324268425040444,
      "grad_norm": 0.2765333950519562,
      "learning_rate": 0.0003019528702050767,
      "loss": 4.7069,
      "step": 26000
    },
    {
      "epoch": 0.5324268425040444,
      "eval_loss": 4.6982269287109375,
      "eval_runtime": 4.2785,
      "eval_samples_per_second": 272.527,
      "eval_steps_per_second": 34.124,
      "step": 26000
    },
    {
      "epoch": 0.5326316220588536,
      "grad_norm": 0.3126332461833954,
      "learning_rate": 0.00030179415388631545,
      "loss": 4.6768,
      "step": 26010
    },
    {
      "epoch": 0.5328364016136629,
      "grad_norm": 0.279086709022522,
      "learning_rate": 0.00030163535100007506,
      "loss": 4.7024,
      "step": 26020
    },
    {
      "epoch": 0.5330411811684721,
      "grad_norm": 0.29606544971466064,
      "learning_rate": 0.00030147646168140433,
      "loss": 4.6972,
      "step": 26030
    },
    {
      "epoch": 0.5332459607232813,
      "grad_norm": 0.2799825072288513,
      "learning_rate": 0.00030131748606542543,
      "loss": 4.7069,
      "step": 26040
    },
    {
      "epoch": 0.5334507402780906,
      "grad_norm": 0.2848356068134308,
      "learning_rate": 0.00030115842428733394,
      "loss": 4.6925,
      "step": 26050
    },
    {
      "epoch": 0.5336555198328999,
      "grad_norm": 0.284119188785553,
      "learning_rate": 0.00030099927648239864,
      "loss": 4.6824,
      "step": 26060
    },
    {
      "epoch": 0.5338602993877092,
      "grad_norm": 0.2721329629421234,
      "learning_rate": 0.00030084004278596154,
      "loss": 4.7047,
      "step": 26070
    },
    {
      "epoch": 0.5340650789425184,
      "grad_norm": 0.29070404171943665,
      "learning_rate": 0.0003006807233334378,
      "loss": 4.7209,
      "step": 26080
    },
    {
      "epoch": 0.5342698584973277,
      "grad_norm": 0.2773502469062805,
      "learning_rate": 0.00030052131826031523,
      "loss": 4.693,
      "step": 26090
    },
    {
      "epoch": 0.5344746380521369,
      "grad_norm": 0.2657264471054077,
      "learning_rate": 0.00030036182770215467,
      "loss": 4.7017,
      "step": 26100
    },
    {
      "epoch": 0.5346794176069462,
      "grad_norm": 0.2874513864517212,
      "learning_rate": 0.00030020225179458964,
      "loss": 4.6929,
      "step": 26110
    },
    {
      "epoch": 0.5348841971617554,
      "grad_norm": 0.2968789041042328,
      "learning_rate": 0.00030004259067332615,
      "loss": 4.7381,
      "step": 26120
    },
    {
      "epoch": 0.5350889767165646,
      "grad_norm": 0.3313628137111664,
      "learning_rate": 0.0002998828444741428,
      "loss": 4.6787,
      "step": 26130
    },
    {
      "epoch": 0.5352937562713739,
      "grad_norm": 0.46929121017456055,
      "learning_rate": 0.00029972301333289035,
      "loss": 4.7263,
      "step": 26140
    },
    {
      "epoch": 0.5354985358261831,
      "grad_norm": 0.27975574135780334,
      "learning_rate": 0.0002995630973854921,
      "loss": 4.6692,
      "step": 26150
    },
    {
      "epoch": 0.5357033153809924,
      "grad_norm": 0.2827834486961365,
      "learning_rate": 0.0002994030967679431,
      "loss": 4.685,
      "step": 26160
    },
    {
      "epoch": 0.5359080949358016,
      "grad_norm": 0.2770453095436096,
      "learning_rate": 0.00029924301161631076,
      "loss": 4.6679,
      "step": 26170
    },
    {
      "epoch": 0.5361128744906108,
      "grad_norm": 0.2641393542289734,
      "learning_rate": 0.00029908284206673415,
      "loss": 4.7198,
      "step": 26180
    },
    {
      "epoch": 0.5363176540454201,
      "grad_norm": 0.2863255441188812,
      "learning_rate": 0.00029892258825542416,
      "loss": 4.6987,
      "step": 26190
    },
    {
      "epoch": 0.5365224336002293,
      "grad_norm": 0.2789994776248932,
      "learning_rate": 0.0002987622503186633,
      "loss": 4.722,
      "step": 26200
    },
    {
      "epoch": 0.5367272131550386,
      "grad_norm": 0.2993677854537964,
      "learning_rate": 0.00029860182839280575,
      "loss": 4.6807,
      "step": 26210
    },
    {
      "epoch": 0.5369319927098478,
      "grad_norm": 0.3469982147216797,
      "learning_rate": 0.00029844132261427713,
      "loss": 4.6821,
      "step": 26220
    },
    {
      "epoch": 0.5371367722646571,
      "grad_norm": 0.2888258397579193,
      "learning_rate": 0.00029828073311957416,
      "loss": 4.7409,
      "step": 26230
    },
    {
      "epoch": 0.5373415518194663,
      "grad_norm": 0.2735867202281952,
      "learning_rate": 0.0002981200600452649,
      "loss": 4.6525,
      "step": 26240
    },
    {
      "epoch": 0.5375463313742755,
      "grad_norm": 0.29521840810775757,
      "learning_rate": 0.00029795930352798837,
      "loss": 4.681,
      "step": 26250
    },
    {
      "epoch": 0.5377511109290848,
      "grad_norm": 0.3799820840358734,
      "learning_rate": 0.0002977984637044549,
      "loss": 4.7131,
      "step": 26260
    },
    {
      "epoch": 0.5379558904838941,
      "grad_norm": 0.30397945642471313,
      "learning_rate": 0.00029763754071144514,
      "loss": 4.7276,
      "step": 26270
    },
    {
      "epoch": 0.5381606700387034,
      "grad_norm": 0.2679298520088196,
      "learning_rate": 0.0002974765346858109,
      "loss": 4.6958,
      "step": 26280
    },
    {
      "epoch": 0.5383654495935126,
      "grad_norm": 0.2828942835330963,
      "learning_rate": 0.00029731544576447447,
      "loss": 4.6866,
      "step": 26290
    },
    {
      "epoch": 0.5385702291483219,
      "grad_norm": 0.29038602113723755,
      "learning_rate": 0.00029715427408442843,
      "loss": 4.6415,
      "step": 26300
    },
    {
      "epoch": 0.5387750087031311,
      "grad_norm": 0.29893389344215393,
      "learning_rate": 0.000296993019782736,
      "loss": 4.6895,
      "step": 26310
    },
    {
      "epoch": 0.5389797882579404,
      "grad_norm": 0.29326602816581726,
      "learning_rate": 0.00029683168299653065,
      "loss": 4.6806,
      "step": 26320
    },
    {
      "epoch": 0.5391845678127496,
      "grad_norm": 0.2700636684894562,
      "learning_rate": 0.0002966702638630159,
      "loss": 4.7156,
      "step": 26330
    },
    {
      "epoch": 0.5393893473675588,
      "grad_norm": 0.28488901257514954,
      "learning_rate": 0.0002965087625194653,
      "loss": 4.6573,
      "step": 26340
    },
    {
      "epoch": 0.5395941269223681,
      "grad_norm": 0.2597222328186035,
      "learning_rate": 0.00029634717910322224,
      "loss": 4.6872,
      "step": 26350
    },
    {
      "epoch": 0.5397989064771773,
      "grad_norm": 0.295816570520401,
      "learning_rate": 0.0002961855137517002,
      "loss": 4.6942,
      "step": 26360
    },
    {
      "epoch": 0.5400036860319866,
      "grad_norm": 0.2972418963909149,
      "learning_rate": 0.0002960237666023821,
      "loss": 4.6899,
      "step": 26370
    },
    {
      "epoch": 0.5402084655867958,
      "grad_norm": 0.3499514162540436,
      "learning_rate": 0.00029586193779282025,
      "loss": 4.6558,
      "step": 26380
    },
    {
      "epoch": 0.540413245141605,
      "grad_norm": 0.28130412101745605,
      "learning_rate": 0.0002957000274606369,
      "loss": 4.6368,
      "step": 26390
    },
    {
      "epoch": 0.5406180246964143,
      "grad_norm": 0.3309668004512787,
      "learning_rate": 0.0002955380357435232,
      "loss": 4.7212,
      "step": 26400
    },
    {
      "epoch": 0.5408228042512235,
      "grad_norm": 0.2834167778491974,
      "learning_rate": 0.00029537596277923964,
      "loss": 4.7359,
      "step": 26410
    },
    {
      "epoch": 0.5410275838060328,
      "grad_norm": 0.2899566888809204,
      "learning_rate": 0.00029521380870561594,
      "loss": 4.6559,
      "step": 26420
    },
    {
      "epoch": 0.541232363360842,
      "grad_norm": 0.2890217900276184,
      "learning_rate": 0.00029505157366055065,
      "loss": 4.687,
      "step": 26430
    },
    {
      "epoch": 0.5414371429156513,
      "grad_norm": 0.2887726128101349,
      "learning_rate": 0.00029488925778201107,
      "loss": 4.7041,
      "step": 26440
    },
    {
      "epoch": 0.5416419224704605,
      "grad_norm": 0.2790029048919678,
      "learning_rate": 0.0002947268612080335,
      "loss": 4.7045,
      "step": 26450
    },
    {
      "epoch": 0.5418467020252697,
      "grad_norm": 0.27893850207328796,
      "learning_rate": 0.0002945643840767228,
      "loss": 4.699,
      "step": 26460
    },
    {
      "epoch": 0.5420514815800791,
      "grad_norm": 0.28486934304237366,
      "learning_rate": 0.00029440182652625224,
      "loss": 4.6918,
      "step": 26470
    },
    {
      "epoch": 0.5422562611348883,
      "grad_norm": 0.2898862957954407,
      "learning_rate": 0.0002942391886948635,
      "loss": 4.6883,
      "step": 26480
    },
    {
      "epoch": 0.5424610406896976,
      "grad_norm": 0.30720099806785583,
      "learning_rate": 0.00029407647072086647,
      "loss": 4.6698,
      "step": 26490
    },
    {
      "epoch": 0.5426658202445068,
      "grad_norm": 0.31864264607429504,
      "learning_rate": 0.00029391367274263946,
      "loss": 4.7138,
      "step": 26500
    },
    {
      "epoch": 0.5428705997993161,
      "grad_norm": 0.2910747528076172,
      "learning_rate": 0.00029375079489862854,
      "loss": 4.7209,
      "step": 26510
    },
    {
      "epoch": 0.5430753793541253,
      "grad_norm": 0.35545462369918823,
      "learning_rate": 0.0002935878373273478,
      "loss": 4.7109,
      "step": 26520
    },
    {
      "epoch": 0.5432801589089346,
      "grad_norm": 0.31063807010650635,
      "learning_rate": 0.00029342480016737916,
      "loss": 4.7099,
      "step": 26530
    },
    {
      "epoch": 0.5434849384637438,
      "grad_norm": 0.31367921829223633,
      "learning_rate": 0.0002932616835573722,
      "loss": 4.6925,
      "step": 26540
    },
    {
      "epoch": 0.543689718018553,
      "grad_norm": 0.28063562512397766,
      "learning_rate": 0.00029309848763604395,
      "loss": 4.6912,
      "step": 26550
    },
    {
      "epoch": 0.5438944975733623,
      "grad_norm": 0.27479425072669983,
      "learning_rate": 0.00029293521254217914,
      "loss": 4.6824,
      "step": 26560
    },
    {
      "epoch": 0.5440992771281715,
      "grad_norm": 0.2851077914237976,
      "learning_rate": 0.0002927718584146296,
      "loss": 4.6928,
      "step": 26570
    },
    {
      "epoch": 0.5443040566829808,
      "grad_norm": 0.27119210362434387,
      "learning_rate": 0.0002926084253923145,
      "loss": 4.7131,
      "step": 26580
    },
    {
      "epoch": 0.54450883623779,
      "grad_norm": 0.28119131922721863,
      "learning_rate": 0.00029244491361421997,
      "loss": 4.6774,
      "step": 26590
    },
    {
      "epoch": 0.5447136157925992,
      "grad_norm": 0.27679044008255005,
      "learning_rate": 0.0002922813232193993,
      "loss": 4.6574,
      "step": 26600
    },
    {
      "epoch": 0.5449183953474085,
      "grad_norm": 0.27654358744621277,
      "learning_rate": 0.0002921176543469726,
      "loss": 4.6822,
      "step": 26610
    },
    {
      "epoch": 0.5451231749022177,
      "grad_norm": 0.2731030583381653,
      "learning_rate": 0.0002919539071361265,
      "loss": 4.6278,
      "step": 26620
    },
    {
      "epoch": 0.545327954457027,
      "grad_norm": 0.27976152300834656,
      "learning_rate": 0.0002917900817261145,
      "loss": 4.7067,
      "step": 26630
    },
    {
      "epoch": 0.5455327340118362,
      "grad_norm": 0.2779357135295868,
      "learning_rate": 0.0002916261782562565,
      "loss": 4.6693,
      "step": 26640
    },
    {
      "epoch": 0.5457375135666455,
      "grad_norm": 0.2789212167263031,
      "learning_rate": 0.0002914621968659389,
      "loss": 4.6428,
      "step": 26650
    },
    {
      "epoch": 0.5459422931214547,
      "grad_norm": 0.2688707113265991,
      "learning_rate": 0.00029129813769461407,
      "loss": 4.686,
      "step": 26660
    },
    {
      "epoch": 0.5461470726762641,
      "grad_norm": 0.2800160348415375,
      "learning_rate": 0.00029113400088180084,
      "loss": 4.6864,
      "step": 26670
    },
    {
      "epoch": 0.5463518522310733,
      "grad_norm": 0.28964245319366455,
      "learning_rate": 0.0002909697865670839,
      "loss": 4.664,
      "step": 26680
    },
    {
      "epoch": 0.5465566317858825,
      "grad_norm": 0.2826152443885803,
      "learning_rate": 0.0002908054948901139,
      "loss": 4.6698,
      "step": 26690
    },
    {
      "epoch": 0.5467614113406918,
      "grad_norm": 0.2769240140914917,
      "learning_rate": 0.00029064112599060734,
      "loss": 4.6627,
      "step": 26700
    },
    {
      "epoch": 0.546966190895501,
      "grad_norm": 0.27277007699012756,
      "learning_rate": 0.00029047668000834625,
      "loss": 4.6499,
      "step": 26710
    },
    {
      "epoch": 0.5471709704503103,
      "grad_norm": 0.29175493121147156,
      "learning_rate": 0.0002903121570831783,
      "loss": 4.685,
      "step": 26720
    },
    {
      "epoch": 0.5473757500051195,
      "grad_norm": 0.2729916274547577,
      "learning_rate": 0.0002901475573550166,
      "loss": 4.6603,
      "step": 26730
    },
    {
      "epoch": 0.5475805295599288,
      "grad_norm": 0.2936178147792816,
      "learning_rate": 0.0002899828809638396,
      "loss": 4.6712,
      "step": 26740
    },
    {
      "epoch": 0.547785309114738,
      "grad_norm": 0.28730419278144836,
      "learning_rate": 0.00028981812804969087,
      "loss": 4.6672,
      "step": 26750
    },
    {
      "epoch": 0.5479900886695472,
      "grad_norm": 0.28215673565864563,
      "learning_rate": 0.000289653298752679,
      "loss": 4.6573,
      "step": 26760
    },
    {
      "epoch": 0.5481948682243565,
      "grad_norm": 0.3288278877735138,
      "learning_rate": 0.00028948839321297777,
      "loss": 4.6569,
      "step": 26770
    },
    {
      "epoch": 0.5483996477791657,
      "grad_norm": 0.30455297231674194,
      "learning_rate": 0.0002893234115708255,
      "loss": 4.6549,
      "step": 26780
    },
    {
      "epoch": 0.548604427333975,
      "grad_norm": 0.2895998954772949,
      "learning_rate": 0.0002891583539665256,
      "loss": 4.7075,
      "step": 26790
    },
    {
      "epoch": 0.5488092068887842,
      "grad_norm": 0.3010871112346649,
      "learning_rate": 0.0002889932205404457,
      "loss": 4.719,
      "step": 26800
    },
    {
      "epoch": 0.5490139864435934,
      "grad_norm": 0.2826119661331177,
      "learning_rate": 0.00028882801143301803,
      "loss": 4.6746,
      "step": 26810
    },
    {
      "epoch": 0.5492187659984027,
      "grad_norm": 0.27935516834259033,
      "learning_rate": 0.0002886627267847394,
      "loss": 4.7028,
      "step": 26820
    },
    {
      "epoch": 0.5494235455532119,
      "grad_norm": 0.29263758659362793,
      "learning_rate": 0.0002884973667361705,
      "loss": 4.7084,
      "step": 26830
    },
    {
      "epoch": 0.5496283251080212,
      "grad_norm": 0.2951434552669525,
      "learning_rate": 0.0002883319314279364,
      "loss": 4.6304,
      "step": 26840
    },
    {
      "epoch": 0.5498331046628304,
      "grad_norm": 0.2866668701171875,
      "learning_rate": 0.00028816642100072616,
      "loss": 4.685,
      "step": 26850
    },
    {
      "epoch": 0.5500378842176397,
      "grad_norm": 0.28918665647506714,
      "learning_rate": 0.0002880008355952927,
      "loss": 4.6649,
      "step": 26860
    },
    {
      "epoch": 0.5502426637724489,
      "grad_norm": 0.27269914746284485,
      "learning_rate": 0.0002878351753524524,
      "loss": 4.6594,
      "step": 26870
    },
    {
      "epoch": 0.5504474433272583,
      "grad_norm": 0.2877494990825653,
      "learning_rate": 0.0002876694404130858,
      "loss": 4.6656,
      "step": 26880
    },
    {
      "epoch": 0.5506522228820675,
      "grad_norm": 0.27912789583206177,
      "learning_rate": 0.0002875036309181367,
      "loss": 4.6644,
      "step": 26890
    },
    {
      "epoch": 0.5508570024368767,
      "grad_norm": 0.26564154028892517,
      "learning_rate": 0.0002873377470086121,
      "loss": 4.653,
      "step": 26900
    },
    {
      "epoch": 0.551061781991686,
      "grad_norm": 0.2571597397327423,
      "learning_rate": 0.0002871717888255828,
      "loss": 4.6591,
      "step": 26910
    },
    {
      "epoch": 0.5512665615464952,
      "grad_norm": 0.25915297865867615,
      "learning_rate": 0.00028700575651018214,
      "loss": 4.6818,
      "step": 26920
    },
    {
      "epoch": 0.5514713411013045,
      "grad_norm": 0.2806718647480011,
      "learning_rate": 0.0002868396502036071,
      "loss": 4.6864,
      "step": 26930
    },
    {
      "epoch": 0.5516761206561137,
      "grad_norm": 0.29026949405670166,
      "learning_rate": 0.00028667347004711715,
      "loss": 4.646,
      "step": 26940
    },
    {
      "epoch": 0.551880900210923,
      "grad_norm": 0.28536713123321533,
      "learning_rate": 0.0002865072161820348,
      "loss": 4.6575,
      "step": 26950
    },
    {
      "epoch": 0.5520856797657322,
      "grad_norm": 0.3099000155925751,
      "learning_rate": 0.00028634088874974513,
      "loss": 4.6635,
      "step": 26960
    },
    {
      "epoch": 0.5522904593205414,
      "grad_norm": 0.2998659014701843,
      "learning_rate": 0.0002861744878916958,
      "loss": 4.7011,
      "step": 26970
    },
    {
      "epoch": 0.5524952388753507,
      "grad_norm": 0.2926270067691803,
      "learning_rate": 0.00028600801374939694,
      "loss": 4.6806,
      "step": 26980
    },
    {
      "epoch": 0.5527000184301599,
      "grad_norm": 0.26582396030426025,
      "learning_rate": 0.00028584146646442097,
      "loss": 4.7549,
      "step": 26990
    },
    {
      "epoch": 0.5529047979849692,
      "grad_norm": 0.2821336090564728,
      "learning_rate": 0.0002856748461784026,
      "loss": 4.6877,
      "step": 27000
    },
    {
      "epoch": 0.5531095775397784,
      "grad_norm": 0.3004828989505768,
      "learning_rate": 0.00028550815303303846,
      "loss": 4.6975,
      "step": 27010
    },
    {
      "epoch": 0.5533143570945876,
      "grad_norm": 0.3053824007511139,
      "learning_rate": 0.00028534138717008727,
      "loss": 4.6588,
      "step": 27020
    },
    {
      "epoch": 0.5535191366493969,
      "grad_norm": 0.2883572280406952,
      "learning_rate": 0.00028517454873136966,
      "loss": 4.6864,
      "step": 27030
    },
    {
      "epoch": 0.5537239162042061,
      "grad_norm": 0.2585112750530243,
      "learning_rate": 0.00028500763785876764,
      "loss": 4.7093,
      "step": 27040
    },
    {
      "epoch": 0.5539286957590154,
      "grad_norm": 0.28137773275375366,
      "learning_rate": 0.00028484065469422534,
      "loss": 4.6768,
      "step": 27050
    },
    {
      "epoch": 0.5541334753138246,
      "grad_norm": 0.27896785736083984,
      "learning_rate": 0.00028467359937974786,
      "loss": 4.6621,
      "step": 27060
    },
    {
      "epoch": 0.5543382548686339,
      "grad_norm": 0.3065791726112366,
      "learning_rate": 0.000284506472057402,
      "loss": 4.6868,
      "step": 27070
    },
    {
      "epoch": 0.5545430344234432,
      "grad_norm": 0.34164634346961975,
      "learning_rate": 0.00028433927286931557,
      "loss": 4.6731,
      "step": 27080
    },
    {
      "epoch": 0.5547478139782525,
      "grad_norm": 0.2731534540653229,
      "learning_rate": 0.00028417200195767774,
      "loss": 4.6487,
      "step": 27090
    },
    {
      "epoch": 0.5549525935330617,
      "grad_norm": 0.3001730144023895,
      "learning_rate": 0.0002840046594647385,
      "loss": 4.7111,
      "step": 27100
    },
    {
      "epoch": 0.5551573730878709,
      "grad_norm": 0.3216342628002167,
      "learning_rate": 0.0002838372455328086,
      "loss": 4.6878,
      "step": 27110
    },
    {
      "epoch": 0.5553621526426802,
      "grad_norm": 0.28366807103157043,
      "learning_rate": 0.00028366976030425983,
      "loss": 4.7044,
      "step": 27120
    },
    {
      "epoch": 0.5555669321974894,
      "grad_norm": 0.5463992953300476,
      "learning_rate": 0.0002835022039215245,
      "loss": 4.6733,
      "step": 27130
    },
    {
      "epoch": 0.5557717117522987,
      "grad_norm": 0.26576077938079834,
      "learning_rate": 0.00028333457652709537,
      "loss": 4.658,
      "step": 27140
    },
    {
      "epoch": 0.5559764913071079,
      "grad_norm": 0.2999189496040344,
      "learning_rate": 0.0002831668782635256,
      "loss": 4.628,
      "step": 27150
    },
    {
      "epoch": 0.5561812708619172,
      "grad_norm": 0.2850971817970276,
      "learning_rate": 0.0002829991092734286,
      "loss": 4.6603,
      "step": 27160
    },
    {
      "epoch": 0.5563860504167264,
      "grad_norm": 0.2797244191169739,
      "learning_rate": 0.0002828312696994781,
      "loss": 4.6823,
      "step": 27170
    },
    {
      "epoch": 0.5565908299715356,
      "grad_norm": 0.28097304701805115,
      "learning_rate": 0.0002826633596844077,
      "loss": 4.7016,
      "step": 27180
    },
    {
      "epoch": 0.5567956095263449,
      "grad_norm": 0.28281205892562866,
      "learning_rate": 0.0002824953793710108,
      "loss": 4.7176,
      "step": 27190
    },
    {
      "epoch": 0.5570003890811541,
      "grad_norm": 0.29616957902908325,
      "learning_rate": 0.00028232732890214084,
      "loss": 4.6836,
      "step": 27200
    },
    {
      "epoch": 0.5572051686359634,
      "grad_norm": 0.26958492398262024,
      "learning_rate": 0.00028215920842071074,
      "loss": 4.6967,
      "step": 27210
    },
    {
      "epoch": 0.5574099481907726,
      "grad_norm": 0.33172526955604553,
      "learning_rate": 0.0002819910180696931,
      "loss": 4.6903,
      "step": 27220
    },
    {
      "epoch": 0.5576147277455819,
      "grad_norm": 0.2655125558376312,
      "learning_rate": 0.00028182275799211974,
      "loss": 4.6666,
      "step": 27230
    },
    {
      "epoch": 0.5578195073003911,
      "grad_norm": 0.29412057995796204,
      "learning_rate": 0.00028165442833108197,
      "loss": 4.6836,
      "step": 27240
    },
    {
      "epoch": 0.5580242868552003,
      "grad_norm": 0.29321762919425964,
      "learning_rate": 0.0002814860292297301,
      "loss": 4.6663,
      "step": 27250
    },
    {
      "epoch": 0.5582290664100096,
      "grad_norm": 0.2720133364200592,
      "learning_rate": 0.0002813175608312737,
      "loss": 4.6888,
      "step": 27260
    },
    {
      "epoch": 0.5584338459648188,
      "grad_norm": 0.27242472767829895,
      "learning_rate": 0.0002811490232789811,
      "loss": 4.6799,
      "step": 27270
    },
    {
      "epoch": 0.5586386255196282,
      "grad_norm": 0.30829906463623047,
      "learning_rate": 0.00028098041671617953,
      "loss": 4.6187,
      "step": 27280
    },
    {
      "epoch": 0.5588434050744374,
      "grad_norm": 0.286232054233551,
      "learning_rate": 0.0002808117412862549,
      "loss": 4.6596,
      "step": 27290
    },
    {
      "epoch": 0.5590481846292467,
      "grad_norm": 0.26213908195495605,
      "learning_rate": 0.00028064299713265157,
      "loss": 4.6717,
      "step": 27300
    },
    {
      "epoch": 0.5592529641840559,
      "grad_norm": 0.3313126862049103,
      "learning_rate": 0.00028047418439887254,
      "loss": 4.666,
      "step": 27310
    },
    {
      "epoch": 0.5594577437388651,
      "grad_norm": 0.3153354227542877,
      "learning_rate": 0.00028030530322847895,
      "loss": 4.6635,
      "step": 27320
    },
    {
      "epoch": 0.5596625232936744,
      "grad_norm": 0.2815357446670532,
      "learning_rate": 0.00028013635376509025,
      "loss": 4.694,
      "step": 27330
    },
    {
      "epoch": 0.5598673028484836,
      "grad_norm": 0.2900370657444,
      "learning_rate": 0.00027996733615238387,
      "loss": 4.7114,
      "step": 27340
    },
    {
      "epoch": 0.5600720824032929,
      "grad_norm": 0.28624969720840454,
      "learning_rate": 0.0002797982505340954,
      "loss": 4.7158,
      "step": 27350
    },
    {
      "epoch": 0.5602768619581021,
      "grad_norm": 0.2943468689918518,
      "learning_rate": 0.0002796290970540181,
      "loss": 4.6346,
      "step": 27360
    },
    {
      "epoch": 0.5604816415129114,
      "grad_norm": 0.2638726234436035,
      "learning_rate": 0.0002794598758560028,
      "loss": 4.6508,
      "step": 27370
    },
    {
      "epoch": 0.5606864210677206,
      "grad_norm": 0.2831834554672241,
      "learning_rate": 0.00027929058708395834,
      "loss": 4.6682,
      "step": 27380
    },
    {
      "epoch": 0.5608912006225298,
      "grad_norm": 0.29787468910217285,
      "learning_rate": 0.00027912123088185054,
      "loss": 4.6922,
      "step": 27390
    },
    {
      "epoch": 0.5610959801773391,
      "grad_norm": 0.2765185236930847,
      "learning_rate": 0.0002789518073937029,
      "loss": 4.6695,
      "step": 27400
    },
    {
      "epoch": 0.5613007597321483,
      "grad_norm": 0.28119799494743347,
      "learning_rate": 0.000278782316763596,
      "loss": 4.7068,
      "step": 27410
    },
    {
      "epoch": 0.5615055392869576,
      "grad_norm": 0.2748514413833618,
      "learning_rate": 0.0002786127591356676,
      "loss": 4.7453,
      "step": 27420
    },
    {
      "epoch": 0.5617103188417668,
      "grad_norm": 0.2974112033843994,
      "learning_rate": 0.0002784431346541123,
      "loss": 4.678,
      "step": 27430
    },
    {
      "epoch": 0.561915098396576,
      "grad_norm": 0.4376599192619324,
      "learning_rate": 0.00027827344346318175,
      "loss": 4.6976,
      "step": 27440
    },
    {
      "epoch": 0.5621198779513853,
      "grad_norm": 0.285873681306839,
      "learning_rate": 0.0002781036857071842,
      "loss": 4.6554,
      "step": 27450
    },
    {
      "epoch": 0.5623246575061945,
      "grad_norm": 0.3002917170524597,
      "learning_rate": 0.00027793386153048444,
      "loss": 4.7362,
      "step": 27460
    },
    {
      "epoch": 0.5625294370610038,
      "grad_norm": 0.2769624590873718,
      "learning_rate": 0.000277763971077504,
      "loss": 4.6749,
      "step": 27470
    },
    {
      "epoch": 0.5627342166158131,
      "grad_norm": 0.36543720960617065,
      "learning_rate": 0.00027759401449272043,
      "loss": 4.6384,
      "step": 27480
    },
    {
      "epoch": 0.5629389961706224,
      "grad_norm": 0.2904614806175232,
      "learning_rate": 0.00027742399192066784,
      "loss": 4.6709,
      "step": 27490
    },
    {
      "epoch": 0.5631437757254316,
      "grad_norm": 0.26392224431037903,
      "learning_rate": 0.00027725390350593623,
      "loss": 4.6671,
      "step": 27500
    },
    {
      "epoch": 0.5633485552802409,
      "grad_norm": 0.3175850510597229,
      "learning_rate": 0.00027708374939317167,
      "loss": 4.6972,
      "step": 27510
    },
    {
      "epoch": 0.5635533348350501,
      "grad_norm": 0.3197113871574402,
      "learning_rate": 0.0002769135297270762,
      "loss": 4.6675,
      "step": 27520
    },
    {
      "epoch": 0.5637581143898593,
      "grad_norm": 0.30047979950904846,
      "learning_rate": 0.0002767432446524075,
      "loss": 4.6708,
      "step": 27530
    },
    {
      "epoch": 0.5639628939446686,
      "grad_norm": 0.5802862644195557,
      "learning_rate": 0.0002765728943139788,
      "loss": 4.6787,
      "step": 27540
    },
    {
      "epoch": 0.5641676734994778,
      "grad_norm": 0.29684823751449585,
      "learning_rate": 0.000276402478856659,
      "loss": 4.7068,
      "step": 27550
    },
    {
      "epoch": 0.5643724530542871,
      "grad_norm": 0.27814826369285583,
      "learning_rate": 0.0002762319984253723,
      "loss": 4.7162,
      "step": 27560
    },
    {
      "epoch": 0.5645772326090963,
      "grad_norm": 0.2693076431751251,
      "learning_rate": 0.0002760614531650982,
      "loss": 4.6658,
      "step": 27570
    },
    {
      "epoch": 0.5647820121639056,
      "grad_norm": 0.3031403124332428,
      "learning_rate": 0.00027589084322087115,
      "loss": 4.6555,
      "step": 27580
    },
    {
      "epoch": 0.5649867917187148,
      "grad_norm": 0.3396504521369934,
      "learning_rate": 0.0002757201687377809,
      "loss": 4.696,
      "step": 27590
    },
    {
      "epoch": 0.565191571273524,
      "grad_norm": 0.33264583349227905,
      "learning_rate": 0.0002755494298609718,
      "loss": 4.6536,
      "step": 27600
    },
    {
      "epoch": 0.5653963508283333,
      "grad_norm": 0.43447214365005493,
      "learning_rate": 0.0002753786267356431,
      "loss": 4.6716,
      "step": 27610
    },
    {
      "epoch": 0.5656011303831425,
      "grad_norm": 0.27637383341789246,
      "learning_rate": 0.0002752077595070488,
      "loss": 4.6651,
      "step": 27620
    },
    {
      "epoch": 0.5658059099379518,
      "grad_norm": 0.28410792350769043,
      "learning_rate": 0.0002750368283204973,
      "loss": 4.6952,
      "step": 27630
    },
    {
      "epoch": 0.566010689492761,
      "grad_norm": 0.27863118052482605,
      "learning_rate": 0.00027486583332135116,
      "loss": 4.6437,
      "step": 27640
    },
    {
      "epoch": 0.5662154690475703,
      "grad_norm": 0.2986036539077759,
      "learning_rate": 0.00027469477465502776,
      "loss": 4.6924,
      "step": 27650
    },
    {
      "epoch": 0.5664202486023795,
      "grad_norm": 0.2995145916938782,
      "learning_rate": 0.000274523652466998,
      "loss": 4.6502,
      "step": 27660
    },
    {
      "epoch": 0.5666250281571887,
      "grad_norm": 0.2952560484409332,
      "learning_rate": 0.0002743524669027873,
      "loss": 4.6527,
      "step": 27670
    },
    {
      "epoch": 0.566829807711998,
      "grad_norm": 0.273057222366333,
      "learning_rate": 0.00027418121810797474,
      "loss": 4.6607,
      "step": 27680
    },
    {
      "epoch": 0.5670345872668073,
      "grad_norm": 0.612025260925293,
      "learning_rate": 0.00027400990622819304,
      "loss": 4.6507,
      "step": 27690
    },
    {
      "epoch": 0.5672393668216166,
      "grad_norm": 0.2794952690601349,
      "learning_rate": 0.000273838531409129,
      "loss": 4.6476,
      "step": 27700
    },
    {
      "epoch": 0.5674441463764258,
      "grad_norm": 0.30170899629592896,
      "learning_rate": 0.0002736670937965224,
      "loss": 4.6531,
      "step": 27710
    },
    {
      "epoch": 0.5676489259312351,
      "grad_norm": 0.2786957323551178,
      "learning_rate": 0.000273495593536167,
      "loss": 4.679,
      "step": 27720
    },
    {
      "epoch": 0.5678537054860443,
      "grad_norm": 0.3750832974910736,
      "learning_rate": 0.00027332403077390933,
      "loss": 4.6541,
      "step": 27730
    },
    {
      "epoch": 0.5680584850408535,
      "grad_norm": 0.2858746647834778,
      "learning_rate": 0.00027315240565564935,
      "loss": 4.6986,
      "step": 27740
    },
    {
      "epoch": 0.5682632645956628,
      "grad_norm": 0.28511595726013184,
      "learning_rate": 0.00027298071832733986,
      "loss": 4.6615,
      "step": 27750
    },
    {
      "epoch": 0.568468044150472,
      "grad_norm": 0.2617349326610565,
      "learning_rate": 0.00027280896893498686,
      "loss": 4.7171,
      "step": 27760
    },
    {
      "epoch": 0.5686728237052813,
      "grad_norm": 0.2863673269748688,
      "learning_rate": 0.0002726371576246489,
      "loss": 4.6974,
      "step": 27770
    },
    {
      "epoch": 0.5688776032600905,
      "grad_norm": 0.29414480924606323,
      "learning_rate": 0.00027246528454243714,
      "loss": 4.6836,
      "step": 27780
    },
    {
      "epoch": 0.5690823828148998,
      "grad_norm": 0.26643306016921997,
      "learning_rate": 0.00027229334983451547,
      "loss": 4.68,
      "step": 27790
    },
    {
      "epoch": 0.569287162369709,
      "grad_norm": 0.38181358575820923,
      "learning_rate": 0.0002721213536471001,
      "loss": 4.6501,
      "step": 27800
    },
    {
      "epoch": 0.5694919419245182,
      "grad_norm": 0.3115479052066803,
      "learning_rate": 0.0002719492961264595,
      "loss": 4.6587,
      "step": 27810
    },
    {
      "epoch": 0.5696967214793275,
      "grad_norm": 0.2777344286441803,
      "learning_rate": 0.0002717771774189142,
      "loss": 4.6839,
      "step": 27820
    },
    {
      "epoch": 0.5699015010341367,
      "grad_norm": 0.30036577582359314,
      "learning_rate": 0.000271604997670837,
      "loss": 4.6428,
      "step": 27830
    },
    {
      "epoch": 0.570106280588946,
      "grad_norm": 0.2738266885280609,
      "learning_rate": 0.0002714327570286525,
      "loss": 4.6306,
      "step": 27840
    },
    {
      "epoch": 0.5703110601437552,
      "grad_norm": 0.2899834215641022,
      "learning_rate": 0.0002712604556388369,
      "loss": 4.6794,
      "step": 27850
    },
    {
      "epoch": 0.5705158396985645,
      "grad_norm": 0.2756260931491852,
      "learning_rate": 0.00027108809364791847,
      "loss": 4.6225,
      "step": 27860
    },
    {
      "epoch": 0.5707206192533737,
      "grad_norm": 0.28180938959121704,
      "learning_rate": 0.0002709156712024766,
      "loss": 4.6943,
      "step": 27870
    },
    {
      "epoch": 0.5709253988081829,
      "grad_norm": 0.2718702554702759,
      "learning_rate": 0.0002707431884491423,
      "loss": 4.6975,
      "step": 27880
    },
    {
      "epoch": 0.5711301783629923,
      "grad_norm": 0.2886705696582794,
      "learning_rate": 0.00027057064553459787,
      "loss": 4.663,
      "step": 27890
    },
    {
      "epoch": 0.5713349579178015,
      "grad_norm": 0.41911399364471436,
      "learning_rate": 0.00027039804260557676,
      "loss": 4.6453,
      "step": 27900
    },
    {
      "epoch": 0.5715397374726108,
      "grad_norm": 0.28789734840393066,
      "learning_rate": 0.0002702253798088634,
      "loss": 4.6751,
      "step": 27910
    },
    {
      "epoch": 0.57174451702742,
      "grad_norm": 0.29850611090660095,
      "learning_rate": 0.0002700526572912931,
      "loss": 4.6786,
      "step": 27920
    },
    {
      "epoch": 0.5719492965822293,
      "grad_norm": 0.2884978652000427,
      "learning_rate": 0.0002698798751997522,
      "loss": 4.6775,
      "step": 27930
    },
    {
      "epoch": 0.5721540761370385,
      "grad_norm": 0.30702218413352966,
      "learning_rate": 0.00026970703368117737,
      "loss": 4.6594,
      "step": 27940
    },
    {
      "epoch": 0.5723588556918477,
      "grad_norm": 0.28914186358451843,
      "learning_rate": 0.000269534132882556,
      "loss": 4.6714,
      "step": 27950
    },
    {
      "epoch": 0.572563635246657,
      "grad_norm": 0.2664184272289276,
      "learning_rate": 0.000269361172950926,
      "loss": 4.6827,
      "step": 27960
    },
    {
      "epoch": 0.5727684148014662,
      "grad_norm": 0.27715256810188293,
      "learning_rate": 0.00026918815403337526,
      "loss": 4.6838,
      "step": 27970
    },
    {
      "epoch": 0.5729731943562755,
      "grad_norm": 0.37566494941711426,
      "learning_rate": 0.00026901507627704214,
      "loss": 4.6353,
      "step": 27980
    },
    {
      "epoch": 0.5731779739110847,
      "grad_norm": 0.27423736453056335,
      "learning_rate": 0.00026884193982911473,
      "loss": 4.6614,
      "step": 27990
    },
    {
      "epoch": 0.573382753465894,
      "grad_norm": 0.2827259600162506,
      "learning_rate": 0.0002686687448368315,
      "loss": 4.6946,
      "step": 28000
    },
    {
      "epoch": 0.573382753465894,
      "eval_loss": 4.683334827423096,
      "eval_runtime": 4.3861,
      "eval_samples_per_second": 265.841,
      "eval_steps_per_second": 33.287,
      "step": 28000
    },
    {
      "epoch": 0.5735875330207032,
      "grad_norm": 0.3005334138870239,
      "learning_rate": 0.0002684954914474801,
      "loss": 4.6974,
      "step": 28010
    },
    {
      "epoch": 0.5737923125755124,
      "grad_norm": 0.27823692560195923,
      "learning_rate": 0.0002683221798083984,
      "loss": 4.649,
      "step": 28020
    },
    {
      "epoch": 0.5739970921303217,
      "grad_norm": 0.300249308347702,
      "learning_rate": 0.0002681488100669734,
      "loss": 4.6579,
      "step": 28030
    },
    {
      "epoch": 0.5742018716851309,
      "grad_norm": 0.26772481203079224,
      "learning_rate": 0.0002679753823706417,
      "loss": 4.6904,
      "step": 28040
    },
    {
      "epoch": 0.5744066512399402,
      "grad_norm": 0.29746416211128235,
      "learning_rate": 0.00026780189686688937,
      "loss": 4.6793,
      "step": 28050
    },
    {
      "epoch": 0.5746114307947494,
      "grad_norm": 0.31912198662757874,
      "learning_rate": 0.00026762835370325116,
      "loss": 4.676,
      "step": 28060
    },
    {
      "epoch": 0.5748162103495587,
      "grad_norm": 0.2853984236717224,
      "learning_rate": 0.0002674547530273112,
      "loss": 4.6521,
      "step": 28070
    },
    {
      "epoch": 0.5750209899043679,
      "grad_norm": 0.2779002785682678,
      "learning_rate": 0.00026728109498670255,
      "loss": 4.681,
      "step": 28080
    },
    {
      "epoch": 0.5752257694591772,
      "grad_norm": 0.29643452167510986,
      "learning_rate": 0.00026710737972910687,
      "loss": 4.6636,
      "step": 28090
    },
    {
      "epoch": 0.5754305490139865,
      "grad_norm": 0.2981843948364258,
      "learning_rate": 0.00026693360740225465,
      "loss": 4.6901,
      "step": 28100
    },
    {
      "epoch": 0.5756353285687957,
      "grad_norm": 0.27709734439849854,
      "learning_rate": 0.0002667597781539247,
      "loss": 4.6693,
      "step": 28110
    },
    {
      "epoch": 0.575840108123605,
      "grad_norm": 0.27088525891304016,
      "learning_rate": 0.0002665858921319445,
      "loss": 4.6942,
      "step": 28120
    },
    {
      "epoch": 0.5760448876784142,
      "grad_norm": 0.2808799147605896,
      "learning_rate": 0.0002664119494841896,
      "loss": 4.6597,
      "step": 28130
    },
    {
      "epoch": 0.5762496672332235,
      "grad_norm": 0.29753655195236206,
      "learning_rate": 0.00026623795035858377,
      "loss": 4.696,
      "step": 28140
    },
    {
      "epoch": 0.5764544467880327,
      "grad_norm": 0.5138372182846069,
      "learning_rate": 0.000266063894903099,
      "loss": 4.6547,
      "step": 28150
    },
    {
      "epoch": 0.576659226342842,
      "grad_norm": 0.28008517622947693,
      "learning_rate": 0.0002658897832657548,
      "loss": 4.6697,
      "step": 28160
    },
    {
      "epoch": 0.5768640058976512,
      "grad_norm": 0.2961454689502716,
      "learning_rate": 0.00026571561559461866,
      "loss": 4.6935,
      "step": 28170
    },
    {
      "epoch": 0.5770687854524604,
      "grad_norm": 0.2739742398262024,
      "learning_rate": 0.00026554139203780586,
      "loss": 4.6588,
      "step": 28180
    },
    {
      "epoch": 0.5772735650072697,
      "grad_norm": 0.28070080280303955,
      "learning_rate": 0.000265367112743479,
      "loss": 4.7207,
      "step": 28190
    },
    {
      "epoch": 0.5774783445620789,
      "grad_norm": 0.36495405435562134,
      "learning_rate": 0.00026519277785984823,
      "loss": 4.6554,
      "step": 28200
    },
    {
      "epoch": 0.5776831241168882,
      "grad_norm": 0.31353211402893066,
      "learning_rate": 0.00026501838753517084,
      "loss": 4.6916,
      "step": 28210
    },
    {
      "epoch": 0.5778879036716974,
      "grad_norm": 0.2827132046222687,
      "learning_rate": 0.00026484394191775133,
      "loss": 4.675,
      "step": 28220
    },
    {
      "epoch": 0.5780926832265066,
      "grad_norm": 0.28942957520484924,
      "learning_rate": 0.0002646694411559412,
      "loss": 4.6867,
      "step": 28230
    },
    {
      "epoch": 0.5782974627813159,
      "grad_norm": 0.28087642788887024,
      "learning_rate": 0.0002644948853981389,
      "loss": 4.6868,
      "step": 28240
    },
    {
      "epoch": 0.5785022423361251,
      "grad_norm": 0.297259122133255,
      "learning_rate": 0.0002643202747927896,
      "loss": 4.6863,
      "step": 28250
    },
    {
      "epoch": 0.5787070218909344,
      "grad_norm": 0.28089985251426697,
      "learning_rate": 0.0002641456094883852,
      "loss": 4.682,
      "step": 28260
    },
    {
      "epoch": 0.5789118014457436,
      "grad_norm": 0.27846214175224304,
      "learning_rate": 0.00026397088963346383,
      "loss": 4.6831,
      "step": 28270
    },
    {
      "epoch": 0.5791165810005529,
      "grad_norm": 0.2802942991256714,
      "learning_rate": 0.0002637961153766104,
      "loss": 4.647,
      "step": 28280
    },
    {
      "epoch": 0.5793213605553622,
      "grad_norm": 0.29859694838523865,
      "learning_rate": 0.00026362128686645595,
      "loss": 4.6354,
      "step": 28290
    },
    {
      "epoch": 0.5795261401101715,
      "grad_norm": 0.2947337329387665,
      "learning_rate": 0.00026344640425167757,
      "loss": 4.662,
      "step": 28300
    },
    {
      "epoch": 0.5797309196649807,
      "grad_norm": 0.30136755108833313,
      "learning_rate": 0.00026327146768099824,
      "loss": 4.6889,
      "step": 28310
    },
    {
      "epoch": 0.5799356992197899,
      "grad_norm": 0.2920510768890381,
      "learning_rate": 0.0002630964773031872,
      "loss": 4.6877,
      "step": 28320
    },
    {
      "epoch": 0.5801404787745992,
      "grad_norm": 0.2797960340976715,
      "learning_rate": 0.0002629214332670592,
      "loss": 4.6452,
      "step": 28330
    },
    {
      "epoch": 0.5803452583294084,
      "grad_norm": 0.2966301739215851,
      "learning_rate": 0.0002627463357214747,
      "loss": 4.6694,
      "step": 28340
    },
    {
      "epoch": 0.5805500378842177,
      "grad_norm": 0.2787582576274872,
      "learning_rate": 0.0002625711848153395,
      "loss": 4.6764,
      "step": 28350
    },
    {
      "epoch": 0.5807548174390269,
      "grad_norm": 0.2939908802509308,
      "learning_rate": 0.000262395980697605,
      "loss": 4.6838,
      "step": 28360
    },
    {
      "epoch": 0.5809595969938361,
      "grad_norm": 0.2859007716178894,
      "learning_rate": 0.0002622207235172677,
      "loss": 4.6842,
      "step": 28370
    },
    {
      "epoch": 0.5811643765486454,
      "grad_norm": 0.27991899847984314,
      "learning_rate": 0.0002620454134233693,
      "loss": 4.6537,
      "step": 28380
    },
    {
      "epoch": 0.5813691561034546,
      "grad_norm": 0.2813095450401306,
      "learning_rate": 0.0002618700505649966,
      "loss": 4.6625,
      "step": 28390
    },
    {
      "epoch": 0.5815739356582639,
      "grad_norm": 0.2621093690395355,
      "learning_rate": 0.00026169463509128103,
      "loss": 4.6845,
      "step": 28400
    },
    {
      "epoch": 0.5817787152130731,
      "grad_norm": 0.4915701448917389,
      "learning_rate": 0.000261519167151399,
      "loss": 4.6664,
      "step": 28410
    },
    {
      "epoch": 0.5819834947678824,
      "grad_norm": 0.2945091724395752,
      "learning_rate": 0.0002613436468945712,
      "loss": 4.6899,
      "step": 28420
    },
    {
      "epoch": 0.5821882743226916,
      "grad_norm": 0.29733526706695557,
      "learning_rate": 0.0002611680744700633,
      "loss": 4.6791,
      "step": 28430
    },
    {
      "epoch": 0.5823930538775008,
      "grad_norm": 0.2509337365627289,
      "learning_rate": 0.00026099245002718504,
      "loss": 4.6735,
      "step": 28440
    },
    {
      "epoch": 0.5825978334323101,
      "grad_norm": 0.28541046380996704,
      "learning_rate": 0.0002608167737152902,
      "loss": 4.6413,
      "step": 28450
    },
    {
      "epoch": 0.5828026129871193,
      "grad_norm": 0.42691245675086975,
      "learning_rate": 0.00026064104568377717,
      "loss": 4.6507,
      "step": 28460
    },
    {
      "epoch": 0.5830073925419286,
      "grad_norm": 0.3065917193889618,
      "learning_rate": 0.00026046526608208793,
      "loss": 4.6493,
      "step": 28470
    },
    {
      "epoch": 0.5832121720967378,
      "grad_norm": 0.2934316396713257,
      "learning_rate": 0.0002602894350597085,
      "loss": 4.6747,
      "step": 28480
    },
    {
      "epoch": 0.583416951651547,
      "grad_norm": 0.2911780774593353,
      "learning_rate": 0.00026011355276616847,
      "loss": 4.7028,
      "step": 28490
    },
    {
      "epoch": 0.5836217312063564,
      "grad_norm": 0.2835933566093445,
      "learning_rate": 0.0002599376193510412,
      "loss": 4.6543,
      "step": 28500
    },
    {
      "epoch": 0.5838265107611657,
      "grad_norm": 0.2826501131057739,
      "learning_rate": 0.0002597616349639434,
      "loss": 4.7099,
      "step": 28510
    },
    {
      "epoch": 0.5840312903159749,
      "grad_norm": 0.3032357394695282,
      "learning_rate": 0.0002595855997545353,
      "loss": 4.6852,
      "step": 28520
    },
    {
      "epoch": 0.5842360698707841,
      "grad_norm": 0.29295429587364197,
      "learning_rate": 0.0002594095138725202,
      "loss": 4.6671,
      "step": 28530
    },
    {
      "epoch": 0.5844408494255934,
      "grad_norm": 0.2917841672897339,
      "learning_rate": 0.0002592333774676446,
      "loss": 4.7393,
      "step": 28540
    },
    {
      "epoch": 0.5846456289804026,
      "grad_norm": 0.28423553705215454,
      "learning_rate": 0.00025905719068969763,
      "loss": 4.7046,
      "step": 28550
    },
    {
      "epoch": 0.5848504085352119,
      "grad_norm": 0.3000679016113281,
      "learning_rate": 0.0002588809536885117,
      "loss": 4.6685,
      "step": 28560
    },
    {
      "epoch": 0.5850551880900211,
      "grad_norm": 0.2691282033920288,
      "learning_rate": 0.0002587046666139618,
      "loss": 4.6771,
      "step": 28570
    },
    {
      "epoch": 0.5852599676448303,
      "grad_norm": 0.304921954870224,
      "learning_rate": 0.0002585283296159654,
      "loss": 4.6754,
      "step": 28580
    },
    {
      "epoch": 0.5854647471996396,
      "grad_norm": 0.28128403425216675,
      "learning_rate": 0.00025835194284448237,
      "loss": 4.6541,
      "step": 28590
    },
    {
      "epoch": 0.5856695267544488,
      "grad_norm": 0.296207994222641,
      "learning_rate": 0.00025817550644951517,
      "loss": 4.6693,
      "step": 28600
    },
    {
      "epoch": 0.5858743063092581,
      "grad_norm": 0.27633747458457947,
      "learning_rate": 0.0002579990205811083,
      "loss": 4.6618,
      "step": 28610
    },
    {
      "epoch": 0.5860790858640673,
      "grad_norm": 0.272427499294281,
      "learning_rate": 0.0002578224853893483,
      "loss": 4.6575,
      "step": 28620
    },
    {
      "epoch": 0.5862838654188766,
      "grad_norm": 0.39729827642440796,
      "learning_rate": 0.0002576459010243637,
      "loss": 4.6775,
      "step": 28630
    },
    {
      "epoch": 0.5864886449736858,
      "grad_norm": 0.30692756175994873,
      "learning_rate": 0.00025746926763632486,
      "loss": 4.6986,
      "step": 28640
    },
    {
      "epoch": 0.586693424528495,
      "grad_norm": 0.46451130509376526,
      "learning_rate": 0.00025729258537544376,
      "loss": 4.6689,
      "step": 28650
    },
    {
      "epoch": 0.5868982040833043,
      "grad_norm": 0.2835638225078583,
      "learning_rate": 0.00025711585439197403,
      "loss": 4.6172,
      "step": 28660
    },
    {
      "epoch": 0.5871029836381135,
      "grad_norm": 0.27264466881752014,
      "learning_rate": 0.00025693907483621077,
      "loss": 4.6735,
      "step": 28670
    },
    {
      "epoch": 0.5873077631929228,
      "grad_norm": 0.2823847234249115,
      "learning_rate": 0.00025676224685849025,
      "loss": 4.6562,
      "step": 28680
    },
    {
      "epoch": 0.587512542747732,
      "grad_norm": 0.28739845752716064,
      "learning_rate": 0.00025658537060918997,
      "loss": 4.6648,
      "step": 28690
    },
    {
      "epoch": 0.5877173223025414,
      "grad_norm": 0.2688329815864563,
      "learning_rate": 0.00025640844623872855,
      "loss": 4.6809,
      "step": 28700
    },
    {
      "epoch": 0.5879221018573506,
      "grad_norm": 0.2778913080692291,
      "learning_rate": 0.0002562314738975655,
      "loss": 4.6522,
      "step": 28710
    },
    {
      "epoch": 0.5881268814121599,
      "grad_norm": 0.2616167664527893,
      "learning_rate": 0.0002560544537362011,
      "loss": 4.6539,
      "step": 28720
    },
    {
      "epoch": 0.5883316609669691,
      "grad_norm": 0.29110246896743774,
      "learning_rate": 0.0002558773859051762,
      "loss": 4.6622,
      "step": 28730
    },
    {
      "epoch": 0.5885364405217783,
      "grad_norm": 0.2961254119873047,
      "learning_rate": 0.00025570027055507247,
      "loss": 4.6437,
      "step": 28740
    },
    {
      "epoch": 0.5887412200765876,
      "grad_norm": 0.2791956663131714,
      "learning_rate": 0.00025552310783651165,
      "loss": 4.6672,
      "step": 28750
    },
    {
      "epoch": 0.5889459996313968,
      "grad_norm": 0.28366756439208984,
      "learning_rate": 0.00025534589790015595,
      "loss": 4.682,
      "step": 28760
    },
    {
      "epoch": 0.5891507791862061,
      "grad_norm": 0.28468218445777893,
      "learning_rate": 0.00025516864089670784,
      "loss": 4.6409,
      "step": 28770
    },
    {
      "epoch": 0.5893555587410153,
      "grad_norm": 0.2833494544029236,
      "learning_rate": 0.0002549913369769096,
      "loss": 4.6494,
      "step": 28780
    },
    {
      "epoch": 0.5895603382958245,
      "grad_norm": 0.27835631370544434,
      "learning_rate": 0.00025481398629154345,
      "loss": 4.6835,
      "step": 28790
    },
    {
      "epoch": 0.5897651178506338,
      "grad_norm": 0.26863643527030945,
      "learning_rate": 0.0002546365889914315,
      "loss": 4.7076,
      "step": 28800
    },
    {
      "epoch": 0.589969897405443,
      "grad_norm": 0.2861613929271698,
      "learning_rate": 0.0002544591452274354,
      "loss": 4.6564,
      "step": 28810
    },
    {
      "epoch": 0.5901746769602523,
      "grad_norm": 0.3046806752681732,
      "learning_rate": 0.00025428165515045643,
      "loss": 4.6513,
      "step": 28820
    },
    {
      "epoch": 0.5903794565150615,
      "grad_norm": 0.284868448972702,
      "learning_rate": 0.00025410411891143507,
      "loss": 4.6808,
      "step": 28830
    },
    {
      "epoch": 0.5905842360698708,
      "grad_norm": 0.26851439476013184,
      "learning_rate": 0.0002539265366613511,
      "loss": 4.6494,
      "step": 28840
    },
    {
      "epoch": 0.59078901562468,
      "grad_norm": 0.2809792757034302,
      "learning_rate": 0.00025374890855122366,
      "loss": 4.6212,
      "step": 28850
    },
    {
      "epoch": 0.5909937951794892,
      "grad_norm": 0.29327404499053955,
      "learning_rate": 0.00025357123473211055,
      "loss": 4.6527,
      "step": 28860
    },
    {
      "epoch": 0.5911985747342985,
      "grad_norm": 0.29546239972114563,
      "learning_rate": 0.0002533935153551087,
      "loss": 4.6834,
      "step": 28870
    },
    {
      "epoch": 0.5914033542891077,
      "grad_norm": 0.30554676055908203,
      "learning_rate": 0.00025321575057135365,
      "loss": 4.6672,
      "step": 28880
    },
    {
      "epoch": 0.591608133843917,
      "grad_norm": 0.2724545896053314,
      "learning_rate": 0.0002530379405320197,
      "loss": 4.6319,
      "step": 28890
    },
    {
      "epoch": 0.5918129133987263,
      "grad_norm": 0.29854893684387207,
      "learning_rate": 0.00025286008538831934,
      "loss": 4.6888,
      "step": 28900
    },
    {
      "epoch": 0.5920176929535356,
      "grad_norm": 0.29128697514533997,
      "learning_rate": 0.00025268218529150377,
      "loss": 4.6923,
      "step": 28910
    },
    {
      "epoch": 0.5922224725083448,
      "grad_norm": 0.30360209941864014,
      "learning_rate": 0.0002525042403928622,
      "loss": 4.6774,
      "step": 28920
    },
    {
      "epoch": 0.592427252063154,
      "grad_norm": 0.30676624178886414,
      "learning_rate": 0.00025232625084372197,
      "loss": 4.6821,
      "step": 28930
    },
    {
      "epoch": 0.5926320316179633,
      "grad_norm": 0.33180487155914307,
      "learning_rate": 0.0002521482167954484,
      "loss": 4.6972,
      "step": 28940
    },
    {
      "epoch": 0.5928368111727725,
      "grad_norm": 0.30034247040748596,
      "learning_rate": 0.0002519701383994447,
      "loss": 4.6403,
      "step": 28950
    },
    {
      "epoch": 0.5930415907275818,
      "grad_norm": 0.31630831956863403,
      "learning_rate": 0.00025179201580715184,
      "loss": 4.6675,
      "step": 28960
    },
    {
      "epoch": 0.593246370282391,
      "grad_norm": 0.305792897939682,
      "learning_rate": 0.0002516138491700482,
      "loss": 4.655,
      "step": 28970
    },
    {
      "epoch": 0.5934511498372003,
      "grad_norm": 0.32938066124916077,
      "learning_rate": 0.00025143563863964984,
      "loss": 4.687,
      "step": 28980
    },
    {
      "epoch": 0.5936559293920095,
      "grad_norm": 0.29817572236061096,
      "learning_rate": 0.00025125738436750984,
      "loss": 4.6642,
      "step": 28990
    },
    {
      "epoch": 0.5938607089468187,
      "grad_norm": 0.28466013073921204,
      "learning_rate": 0.00025107908650521876,
      "loss": 4.6554,
      "step": 29000
    },
    {
      "epoch": 0.594065488501628,
      "grad_norm": 0.29398080706596375,
      "learning_rate": 0.00025090074520440423,
      "loss": 4.6416,
      "step": 29010
    },
    {
      "epoch": 0.5942702680564372,
      "grad_norm": 0.29575416445732117,
      "learning_rate": 0.00025072236061673053,
      "loss": 4.685,
      "step": 29020
    },
    {
      "epoch": 0.5944750476112465,
      "grad_norm": 0.2865046560764313,
      "learning_rate": 0.0002505439328938991,
      "loss": 4.6354,
      "step": 29030
    },
    {
      "epoch": 0.5946798271660557,
      "grad_norm": 0.29523858428001404,
      "learning_rate": 0.00025036546218764786,
      "loss": 4.6587,
      "step": 29040
    },
    {
      "epoch": 0.594884606720865,
      "grad_norm": 0.2861136794090271,
      "learning_rate": 0.00025018694864975126,
      "loss": 4.667,
      "step": 29050
    },
    {
      "epoch": 0.5950893862756742,
      "grad_norm": 0.2811325788497925,
      "learning_rate": 0.00025000839243202044,
      "loss": 4.7131,
      "step": 29060
    },
    {
      "epoch": 0.5952941658304834,
      "grad_norm": 0.27377763390541077,
      "learning_rate": 0.00024982979368630247,
      "loss": 4.6591,
      "step": 29070
    },
    {
      "epoch": 0.5954989453852927,
      "grad_norm": 0.2821992337703705,
      "learning_rate": 0.0002496511525644808,
      "loss": 4.6562,
      "step": 29080
    },
    {
      "epoch": 0.5957037249401019,
      "grad_norm": 0.2800423204898834,
      "learning_rate": 0.0002494724692184748,
      "loss": 4.6791,
      "step": 29090
    },
    {
      "epoch": 0.5959085044949112,
      "grad_norm": 0.29585859179496765,
      "learning_rate": 0.00024929374380024003,
      "loss": 4.6861,
      "step": 29100
    },
    {
      "epoch": 0.5961132840497205,
      "grad_norm": 0.30531808733940125,
      "learning_rate": 0.00024911497646176744,
      "loss": 4.6698,
      "step": 29110
    },
    {
      "epoch": 0.5963180636045298,
      "grad_norm": 0.26787129044532776,
      "learning_rate": 0.0002489361673550839,
      "loss": 4.699,
      "step": 29120
    },
    {
      "epoch": 0.596522843159339,
      "grad_norm": 0.2912231981754303,
      "learning_rate": 0.0002487573166322517,
      "loss": 4.6863,
      "step": 29130
    },
    {
      "epoch": 0.5967276227141483,
      "grad_norm": 0.28360515832901,
      "learning_rate": 0.0002485784244453685,
      "loss": 4.661,
      "step": 29140
    },
    {
      "epoch": 0.5969324022689575,
      "grad_norm": 0.3958587646484375,
      "learning_rate": 0.0002483994909465673,
      "loss": 4.6813,
      "step": 29150
    },
    {
      "epoch": 0.5971371818237667,
      "grad_norm": 0.2817666828632355,
      "learning_rate": 0.00024822051628801625,
      "loss": 4.644,
      "step": 29160
    },
    {
      "epoch": 0.597341961378576,
      "grad_norm": 0.2872312366962433,
      "learning_rate": 0.0002480415006219183,
      "loss": 4.649,
      "step": 29170
    },
    {
      "epoch": 0.5975467409333852,
      "grad_norm": 0.30035293102264404,
      "learning_rate": 0.00024786244410051147,
      "loss": 4.6831,
      "step": 29180
    },
    {
      "epoch": 0.5977515204881945,
      "grad_norm": 0.29184991121292114,
      "learning_rate": 0.00024768334687606846,
      "loss": 4.6233,
      "step": 29190
    },
    {
      "epoch": 0.5979563000430037,
      "grad_norm": 0.2948676347732544,
      "learning_rate": 0.0002475042091008967,
      "loss": 4.6672,
      "step": 29200
    },
    {
      "epoch": 0.598161079597813,
      "grad_norm": 0.29934364557266235,
      "learning_rate": 0.0002473250309273379,
      "loss": 4.6515,
      "step": 29210
    },
    {
      "epoch": 0.5983658591526222,
      "grad_norm": 0.2856956124305725,
      "learning_rate": 0.00024714581250776814,
      "loss": 4.6733,
      "step": 29220
    },
    {
      "epoch": 0.5985706387074314,
      "grad_norm": 0.27090075612068176,
      "learning_rate": 0.00024696655399459796,
      "loss": 4.6661,
      "step": 29230
    },
    {
      "epoch": 0.5987754182622407,
      "grad_norm": 0.29719436168670654,
      "learning_rate": 0.0002467872555402718,
      "loss": 4.668,
      "step": 29240
    },
    {
      "epoch": 0.5989801978170499,
      "grad_norm": 0.2906520962715149,
      "learning_rate": 0.00024660791729726823,
      "loss": 4.6676,
      "step": 29250
    },
    {
      "epoch": 0.5991849773718592,
      "grad_norm": 0.29557523131370544,
      "learning_rate": 0.00024642853941809925,
      "loss": 4.6871,
      "step": 29260
    },
    {
      "epoch": 0.5993897569266684,
      "grad_norm": 0.2694222629070282,
      "learning_rate": 0.0002462491220553112,
      "loss": 4.6781,
      "step": 29270
    },
    {
      "epoch": 0.5995945364814776,
      "grad_norm": 0.3162340819835663,
      "learning_rate": 0.0002460696653614833,
      "loss": 4.662,
      "step": 29280
    },
    {
      "epoch": 0.5997993160362869,
      "grad_norm": 0.2700228989124298,
      "learning_rate": 0.0002458901694892289,
      "loss": 4.6583,
      "step": 29290
    },
    {
      "epoch": 0.6000040955910961,
      "grad_norm": 0.2790652811527252,
      "learning_rate": 0.0002457106345911942,
      "loss": 4.6658,
      "step": 29300
    },
    {
      "epoch": 0.6002088751459055,
      "grad_norm": 0.28301629424095154,
      "learning_rate": 0.0002455310608200587,
      "loss": 4.6695,
      "step": 29310
    },
    {
      "epoch": 0.6004136547007147,
      "grad_norm": 0.9047539234161377,
      "learning_rate": 0.000245351448328535,
      "loss": 4.6894,
      "step": 29320
    },
    {
      "epoch": 0.600618434255524,
      "grad_norm": 0.2879190146923065,
      "learning_rate": 0.0002451717972693686,
      "loss": 4.6464,
      "step": 29330
    },
    {
      "epoch": 0.6008232138103332,
      "grad_norm": 0.2830010950565338,
      "learning_rate": 0.0002449921077953379,
      "loss": 4.6697,
      "step": 29340
    },
    {
      "epoch": 0.6010279933651425,
      "grad_norm": 0.3032263219356537,
      "learning_rate": 0.0002448123800592539,
      "loss": 4.6552,
      "step": 29350
    },
    {
      "epoch": 0.6012327729199517,
      "grad_norm": 0.301294207572937,
      "learning_rate": 0.0002446326142139599,
      "loss": 4.6622,
      "step": 29360
    },
    {
      "epoch": 0.6014375524747609,
      "grad_norm": 0.28562310338020325,
      "learning_rate": 0.0002444528104123321,
      "loss": 4.6945,
      "step": 29370
    },
    {
      "epoch": 0.6016423320295702,
      "grad_norm": 0.2849845588207245,
      "learning_rate": 0.00024427296880727856,
      "loss": 4.6569,
      "step": 29380
    },
    {
      "epoch": 0.6018471115843794,
      "grad_norm": 0.29991352558135986,
      "learning_rate": 0.00024409308955173972,
      "loss": 4.6809,
      "step": 29390
    },
    {
      "epoch": 0.6020518911391887,
      "grad_norm": 0.3000006079673767,
      "learning_rate": 0.0002439131727986879,
      "loss": 4.6621,
      "step": 29400
    },
    {
      "epoch": 0.6022566706939979,
      "grad_norm": 0.298909455537796,
      "learning_rate": 0.00024373321870112738,
      "loss": 4.6476,
      "step": 29410
    },
    {
      "epoch": 0.6024614502488072,
      "grad_norm": 0.2760293185710907,
      "learning_rate": 0.00024355322741209415,
      "loss": 4.6741,
      "step": 29420
    },
    {
      "epoch": 0.6026662298036164,
      "grad_norm": 0.2685551345348358,
      "learning_rate": 0.0002433731990846559,
      "loss": 4.6493,
      "step": 29430
    },
    {
      "epoch": 0.6028710093584256,
      "grad_norm": 0.3154749274253845,
      "learning_rate": 0.0002431931338719118,
      "loss": 4.6618,
      "step": 29440
    },
    {
      "epoch": 0.6030757889132349,
      "grad_norm": 0.3553539514541626,
      "learning_rate": 0.0002430130319269923,
      "loss": 4.6515,
      "step": 29450
    },
    {
      "epoch": 0.6032805684680441,
      "grad_norm": 0.28778523206710815,
      "learning_rate": 0.00024283289340305922,
      "loss": 4.6685,
      "step": 29460
    },
    {
      "epoch": 0.6034853480228534,
      "grad_norm": 0.39844146370887756,
      "learning_rate": 0.0002426527184533054,
      "loss": 4.6413,
      "step": 29470
    },
    {
      "epoch": 0.6036901275776626,
      "grad_norm": 0.2765115201473236,
      "learning_rate": 0.0002424725072309547,
      "loss": 4.6781,
      "step": 29480
    },
    {
      "epoch": 0.6038949071324718,
      "grad_norm": 0.29303398728370667,
      "learning_rate": 0.00024229225988926182,
      "loss": 4.6654,
      "step": 29490
    },
    {
      "epoch": 0.6040996866872811,
      "grad_norm": 0.30903899669647217,
      "learning_rate": 0.00024211197658151207,
      "loss": 4.6571,
      "step": 29500
    },
    {
      "epoch": 0.6043044662420904,
      "grad_norm": 0.284907728433609,
      "learning_rate": 0.0002419316574610215,
      "loss": 4.6826,
      "step": 29510
    },
    {
      "epoch": 0.6045092457968997,
      "grad_norm": 0.29485589265823364,
      "learning_rate": 0.00024175130268113662,
      "loss": 4.6339,
      "step": 29520
    },
    {
      "epoch": 0.6047140253517089,
      "grad_norm": 0.2721925675868988,
      "learning_rate": 0.00024157091239523405,
      "loss": 4.6884,
      "step": 29530
    },
    {
      "epoch": 0.6049188049065182,
      "grad_norm": 0.2897637188434601,
      "learning_rate": 0.00024139048675672094,
      "loss": 4.6537,
      "step": 29540
    },
    {
      "epoch": 0.6051235844613274,
      "grad_norm": 0.29843610525131226,
      "learning_rate": 0.0002412100259190342,
      "loss": 4.694,
      "step": 29550
    },
    {
      "epoch": 0.6053283640161367,
      "grad_norm": 0.3276248872280121,
      "learning_rate": 0.00024102953003564076,
      "loss": 4.671,
      "step": 29560
    },
    {
      "epoch": 0.6055331435709459,
      "grad_norm": 0.3016703128814697,
      "learning_rate": 0.00024084899926003748,
      "loss": 4.6504,
      "step": 29570
    },
    {
      "epoch": 0.6057379231257551,
      "grad_norm": 0.2773389220237732,
      "learning_rate": 0.00024066843374575078,
      "loss": 4.6773,
      "step": 29580
    },
    {
      "epoch": 0.6059427026805644,
      "grad_norm": 0.2965388000011444,
      "learning_rate": 0.00024048783364633667,
      "loss": 4.6447,
      "step": 29590
    },
    {
      "epoch": 0.6061474822353736,
      "grad_norm": 0.27457672357559204,
      "learning_rate": 0.00024030719911538042,
      "loss": 4.6403,
      "step": 29600
    },
    {
      "epoch": 0.6063522617901829,
      "grad_norm": 0.3018428087234497,
      "learning_rate": 0.00024012653030649683,
      "loss": 4.6588,
      "step": 29610
    },
    {
      "epoch": 0.6065570413449921,
      "grad_norm": 0.2927612364292145,
      "learning_rate": 0.00023994582737332966,
      "loss": 4.6432,
      "step": 29620
    },
    {
      "epoch": 0.6067618208998014,
      "grad_norm": 0.29504019021987915,
      "learning_rate": 0.00023976509046955187,
      "loss": 4.6328,
      "step": 29630
    },
    {
      "epoch": 0.6069666004546106,
      "grad_norm": 0.3035086691379547,
      "learning_rate": 0.00023958431974886507,
      "loss": 4.613,
      "step": 29640
    },
    {
      "epoch": 0.6071713800094198,
      "grad_norm": 0.2993575632572174,
      "learning_rate": 0.00023940351536499974,
      "loss": 4.6828,
      "step": 29650
    },
    {
      "epoch": 0.6073761595642291,
      "grad_norm": 0.3321745991706848,
      "learning_rate": 0.00023922267747171513,
      "loss": 4.7281,
      "step": 29660
    },
    {
      "epoch": 0.6075809391190383,
      "grad_norm": 0.3499547243118286,
      "learning_rate": 0.00023904180622279876,
      "loss": 4.6341,
      "step": 29670
    },
    {
      "epoch": 0.6077857186738476,
      "grad_norm": 0.28242287039756775,
      "learning_rate": 0.00023886090177206666,
      "loss": 4.6274,
      "step": 29680
    },
    {
      "epoch": 0.6079904982286568,
      "grad_norm": 0.27522993087768555,
      "learning_rate": 0.00023867996427336298,
      "loss": 4.6453,
      "step": 29690
    },
    {
      "epoch": 0.608195277783466,
      "grad_norm": 0.2626899182796478,
      "learning_rate": 0.00023849899388056017,
      "loss": 4.6578,
      "step": 29700
    },
    {
      "epoch": 0.6084000573382754,
      "grad_norm": 0.3001188039779663,
      "learning_rate": 0.0002383179907475584,
      "loss": 4.6493,
      "step": 29710
    },
    {
      "epoch": 0.6086048368930846,
      "grad_norm": 0.2864462435245514,
      "learning_rate": 0.00023813695502828582,
      "loss": 4.6365,
      "step": 29720
    },
    {
      "epoch": 0.6088096164478939,
      "grad_norm": 0.30062445998191833,
      "learning_rate": 0.00023795588687669844,
      "loss": 4.6756,
      "step": 29730
    },
    {
      "epoch": 0.6090143960027031,
      "grad_norm": 0.29453182220458984,
      "learning_rate": 0.00023777478644677956,
      "loss": 4.6742,
      "step": 29740
    },
    {
      "epoch": 0.6092191755575124,
      "grad_norm": 0.3395213782787323,
      "learning_rate": 0.00023759365389254002,
      "loss": 4.696,
      "step": 29750
    },
    {
      "epoch": 0.6094239551123216,
      "grad_norm": 0.28882744908332825,
      "learning_rate": 0.0002374124893680181,
      "loss": 4.6679,
      "step": 29760
    },
    {
      "epoch": 0.6096287346671309,
      "grad_norm": 0.28007251024246216,
      "learning_rate": 0.00023723129302727927,
      "loss": 4.655,
      "step": 29770
    },
    {
      "epoch": 0.6098335142219401,
      "grad_norm": 0.2873763144016266,
      "learning_rate": 0.00023705006502441585,
      "loss": 4.6593,
      "step": 29780
    },
    {
      "epoch": 0.6100382937767493,
      "grad_norm": 0.2954859137535095,
      "learning_rate": 0.00023686880551354724,
      "loss": 4.6623,
      "step": 29790
    },
    {
      "epoch": 0.6102430733315586,
      "grad_norm": 0.3011667728424072,
      "learning_rate": 0.00023668751464881967,
      "loss": 4.6686,
      "step": 29800
    },
    {
      "epoch": 0.6104478528863678,
      "grad_norm": 0.28950291872024536,
      "learning_rate": 0.00023650619258440588,
      "loss": 4.6395,
      "step": 29810
    },
    {
      "epoch": 0.6106526324411771,
      "grad_norm": 0.2699313461780548,
      "learning_rate": 0.00023632483947450523,
      "loss": 4.6728,
      "step": 29820
    },
    {
      "epoch": 0.6108574119959863,
      "grad_norm": 0.8729771375656128,
      "learning_rate": 0.00023614345547334366,
      "loss": 4.6861,
      "step": 29830
    },
    {
      "epoch": 0.6110621915507956,
      "grad_norm": 0.3100840449333191,
      "learning_rate": 0.00023596204073517306,
      "loss": 4.6585,
      "step": 29840
    },
    {
      "epoch": 0.6112669711056048,
      "grad_norm": 0.2908143103122711,
      "learning_rate": 0.0002357805954142716,
      "loss": 4.6209,
      "step": 29850
    },
    {
      "epoch": 0.611471750660414,
      "grad_norm": 0.2910616397857666,
      "learning_rate": 0.0002355991196649435,
      "loss": 4.6485,
      "step": 29860
    },
    {
      "epoch": 0.6116765302152233,
      "grad_norm": 0.2798650860786438,
      "learning_rate": 0.0002354176136415189,
      "loss": 4.6634,
      "step": 29870
    },
    {
      "epoch": 0.6118813097700325,
      "grad_norm": 0.2711614668369293,
      "learning_rate": 0.00023523607749835351,
      "loss": 4.6278,
      "step": 29880
    },
    {
      "epoch": 0.6120860893248418,
      "grad_norm": 0.2735280990600586,
      "learning_rate": 0.00023505451138982873,
      "loss": 4.6552,
      "step": 29890
    },
    {
      "epoch": 0.612290868879651,
      "grad_norm": 0.31333765387535095,
      "learning_rate": 0.00023487291547035155,
      "loss": 4.6217,
      "step": 29900
    },
    {
      "epoch": 0.6124956484344602,
      "grad_norm": 0.27783825993537903,
      "learning_rate": 0.00023469128989435422,
      "loss": 4.6715,
      "step": 29910
    },
    {
      "epoch": 0.6127004279892696,
      "grad_norm": 0.34906432032585144,
      "learning_rate": 0.00023450963481629416,
      "loss": 4.6359,
      "step": 29920
    },
    {
      "epoch": 0.6129052075440788,
      "grad_norm": 0.27939939498901367,
      "learning_rate": 0.00023432795039065392,
      "loss": 4.6543,
      "step": 29930
    },
    {
      "epoch": 0.6131099870988881,
      "grad_norm": 0.2987031936645508,
      "learning_rate": 0.00023414623677194113,
      "loss": 4.6372,
      "step": 29940
    },
    {
      "epoch": 0.6133147666536973,
      "grad_norm": 0.27738839387893677,
      "learning_rate": 0.00023396449411468805,
      "loss": 4.618,
      "step": 29950
    },
    {
      "epoch": 0.6135195462085066,
      "grad_norm": 0.3081669807434082,
      "learning_rate": 0.00023378272257345173,
      "loss": 4.6476,
      "step": 29960
    },
    {
      "epoch": 0.6137243257633158,
      "grad_norm": 0.2913212776184082,
      "learning_rate": 0.00023360092230281386,
      "loss": 4.648,
      "step": 29970
    },
    {
      "epoch": 0.6139291053181251,
      "grad_norm": 0.2736920118331909,
      "learning_rate": 0.00023341909345738043,
      "loss": 4.6325,
      "step": 29980
    },
    {
      "epoch": 0.6141338848729343,
      "grad_norm": 0.27339574694633484,
      "learning_rate": 0.00023323723619178175,
      "loss": 4.67,
      "step": 29990
    },
    {
      "epoch": 0.6143386644277435,
      "grad_norm": 0.32638460397720337,
      "learning_rate": 0.0002330553506606723,
      "loss": 4.6309,
      "step": 30000
    },
    {
      "epoch": 0.6143386644277435,
      "eval_loss": 4.663181304931641,
      "eval_runtime": 4.3853,
      "eval_samples_per_second": 265.891,
      "eval_steps_per_second": 33.293,
      "step": 30000
    },
    {
      "epoch": 0.6145434439825528,
      "grad_norm": 0.2861756682395935,
      "learning_rate": 0.00023287343701873078,
      "loss": 4.6396,
      "step": 30010
    },
    {
      "epoch": 0.614748223537362,
      "grad_norm": 0.2900616228580475,
      "learning_rate": 0.0002326914954206595,
      "loss": 4.6473,
      "step": 30020
    },
    {
      "epoch": 0.6149530030921713,
      "grad_norm": 0.2840682864189148,
      "learning_rate": 0.00023250952602118475,
      "loss": 4.6181,
      "step": 30030
    },
    {
      "epoch": 0.6151577826469805,
      "grad_norm": 0.2963438928127289,
      "learning_rate": 0.00023232752897505638,
      "loss": 4.6336,
      "step": 30040
    },
    {
      "epoch": 0.6153625622017898,
      "grad_norm": 0.3062077462673187,
      "learning_rate": 0.00023214550443704778,
      "loss": 4.6366,
      "step": 30050
    },
    {
      "epoch": 0.615567341756599,
      "grad_norm": 0.29020142555236816,
      "learning_rate": 0.0002319634525619557,
      "loss": 4.6699,
      "step": 30060
    },
    {
      "epoch": 0.6157721213114082,
      "grad_norm": 0.28839603066444397,
      "learning_rate": 0.0002317813735046002,
      "loss": 4.6326,
      "step": 30070
    },
    {
      "epoch": 0.6159769008662175,
      "grad_norm": 0.3556162416934967,
      "learning_rate": 0.00023159926741982435,
      "loss": 4.6607,
      "step": 30080
    },
    {
      "epoch": 0.6161816804210267,
      "grad_norm": 0.320284366607666,
      "learning_rate": 0.00023141713446249424,
      "loss": 4.661,
      "step": 30090
    },
    {
      "epoch": 0.616386459975836,
      "grad_norm": 0.2787932753562927,
      "learning_rate": 0.0002312349747874989,
      "loss": 4.6549,
      "step": 30100
    },
    {
      "epoch": 0.6165912395306452,
      "grad_norm": 0.3046487271785736,
      "learning_rate": 0.00023105278854974998,
      "loss": 4.5999,
      "step": 30110
    },
    {
      "epoch": 0.6167960190854546,
      "grad_norm": 0.31222203373908997,
      "learning_rate": 0.0002308705759041818,
      "loss": 4.6837,
      "step": 30120
    },
    {
      "epoch": 0.6170007986402638,
      "grad_norm": 0.28878629207611084,
      "learning_rate": 0.00023068833700575102,
      "loss": 4.6751,
      "step": 30130
    },
    {
      "epoch": 0.617205578195073,
      "grad_norm": 0.28158316016197205,
      "learning_rate": 0.00023050607200943667,
      "loss": 4.6209,
      "step": 30140
    },
    {
      "epoch": 0.6174103577498823,
      "grad_norm": 0.27760252356529236,
      "learning_rate": 0.00023032378107024014,
      "loss": 4.6572,
      "step": 30150
    },
    {
      "epoch": 0.6176151373046915,
      "grad_norm": 0.2882702350616455,
      "learning_rate": 0.0002301414643431846,
      "loss": 4.6716,
      "step": 30160
    },
    {
      "epoch": 0.6178199168595008,
      "grad_norm": 0.3012171685695648,
      "learning_rate": 0.00022995912198331544,
      "loss": 4.6543,
      "step": 30170
    },
    {
      "epoch": 0.61802469641431,
      "grad_norm": 0.2790381908416748,
      "learning_rate": 0.00022977675414569953,
      "loss": 4.6313,
      "step": 30180
    },
    {
      "epoch": 0.6182294759691193,
      "grad_norm": 0.3032726049423218,
      "learning_rate": 0.00022959436098542573,
      "loss": 4.6493,
      "step": 30190
    },
    {
      "epoch": 0.6184342555239285,
      "grad_norm": 0.29249200224876404,
      "learning_rate": 0.00022941194265760419,
      "loss": 4.6433,
      "step": 30200
    },
    {
      "epoch": 0.6186390350787377,
      "grad_norm": 0.296486496925354,
      "learning_rate": 0.00022922949931736663,
      "loss": 4.6352,
      "step": 30210
    },
    {
      "epoch": 0.618843814633547,
      "grad_norm": 0.27457305788993835,
      "learning_rate": 0.0002290470311198659,
      "loss": 4.645,
      "step": 30220
    },
    {
      "epoch": 0.6190485941883562,
      "grad_norm": 0.2674858570098877,
      "learning_rate": 0.00022886453822027608,
      "loss": 4.6826,
      "step": 30230
    },
    {
      "epoch": 0.6192533737431655,
      "grad_norm": 0.27665963768959045,
      "learning_rate": 0.0002286820207737922,
      "loss": 4.6881,
      "step": 30240
    },
    {
      "epoch": 0.6194581532979747,
      "grad_norm": 0.2948571443557739,
      "learning_rate": 0.00022849947893563029,
      "loss": 4.6533,
      "step": 30250
    },
    {
      "epoch": 0.619662932852784,
      "grad_norm": 0.3000718355178833,
      "learning_rate": 0.00022831691286102706,
      "loss": 4.645,
      "step": 30260
    },
    {
      "epoch": 0.6198677124075932,
      "grad_norm": 0.2924703061580658,
      "learning_rate": 0.00022813432270523963,
      "loss": 4.6611,
      "step": 30270
    },
    {
      "epoch": 0.6200724919624024,
      "grad_norm": 0.281405508518219,
      "learning_rate": 0.00022795170862354586,
      "loss": 4.6348,
      "step": 30280
    },
    {
      "epoch": 0.6202772715172117,
      "grad_norm": 0.2816663086414337,
      "learning_rate": 0.0002277690707712439,
      "loss": 4.6586,
      "step": 30290
    },
    {
      "epoch": 0.6204820510720209,
      "grad_norm": 0.3089105188846588,
      "learning_rate": 0.00022758640930365206,
      "loss": 4.6627,
      "step": 30300
    },
    {
      "epoch": 0.6206868306268302,
      "grad_norm": 0.3414783775806427,
      "learning_rate": 0.0002274037243761088,
      "loss": 4.6521,
      "step": 30310
    },
    {
      "epoch": 0.6208916101816395,
      "grad_norm": 0.31398820877075195,
      "learning_rate": 0.00022722101614397241,
      "loss": 4.6378,
      "step": 30320
    },
    {
      "epoch": 0.6210963897364488,
      "grad_norm": 0.2950122058391571,
      "learning_rate": 0.00022703828476262113,
      "loss": 4.6725,
      "step": 30330
    },
    {
      "epoch": 0.621301169291258,
      "grad_norm": 0.3036250174045563,
      "learning_rate": 0.0002268555303874528,
      "loss": 4.6697,
      "step": 30340
    },
    {
      "epoch": 0.6215059488460672,
      "grad_norm": 0.29848480224609375,
      "learning_rate": 0.0002266727531738849,
      "loss": 4.6406,
      "step": 30350
    },
    {
      "epoch": 0.6217107284008765,
      "grad_norm": 0.3059265613555908,
      "learning_rate": 0.00022648995327735424,
      "loss": 4.6607,
      "step": 30360
    },
    {
      "epoch": 0.6219155079556857,
      "grad_norm": 0.27917057275772095,
      "learning_rate": 0.00022630713085331695,
      "loss": 4.6418,
      "step": 30370
    },
    {
      "epoch": 0.622120287510495,
      "grad_norm": 0.2982562482357025,
      "learning_rate": 0.0002261242860572483,
      "loss": 4.6318,
      "step": 30380
    },
    {
      "epoch": 0.6223250670653042,
      "grad_norm": 0.2788049280643463,
      "learning_rate": 0.00022594141904464266,
      "loss": 4.6585,
      "step": 30390
    },
    {
      "epoch": 0.6225298466201135,
      "grad_norm": 0.3022199273109436,
      "learning_rate": 0.00022575852997101334,
      "loss": 4.6664,
      "step": 30400
    },
    {
      "epoch": 0.6227346261749227,
      "grad_norm": 0.32720696926116943,
      "learning_rate": 0.00022557561899189213,
      "loss": 4.6251,
      "step": 30410
    },
    {
      "epoch": 0.6229394057297319,
      "grad_norm": 0.2773551344871521,
      "learning_rate": 0.00022539268626282978,
      "loss": 4.6615,
      "step": 30420
    },
    {
      "epoch": 0.6231441852845412,
      "grad_norm": 0.3055232763290405,
      "learning_rate": 0.00022520973193939533,
      "loss": 4.651,
      "step": 30430
    },
    {
      "epoch": 0.6233489648393504,
      "grad_norm": 0.3110123872756958,
      "learning_rate": 0.0002250267561771762,
      "loss": 4.6476,
      "step": 30440
    },
    {
      "epoch": 0.6235537443941597,
      "grad_norm": 0.2774081230163574,
      "learning_rate": 0.00022484375913177823,
      "loss": 4.6644,
      "step": 30450
    },
    {
      "epoch": 0.6237585239489689,
      "grad_norm": 0.291970819234848,
      "learning_rate": 0.00022466074095882507,
      "loss": 4.6617,
      "step": 30460
    },
    {
      "epoch": 0.6239633035037782,
      "grad_norm": 0.2835977077484131,
      "learning_rate": 0.0002244777018139586,
      "loss": 4.6475,
      "step": 30470
    },
    {
      "epoch": 0.6241680830585874,
      "grad_norm": 0.33001837134361267,
      "learning_rate": 0.0002242946418528383,
      "loss": 4.6724,
      "step": 30480
    },
    {
      "epoch": 0.6243728626133966,
      "grad_norm": 0.3122660219669342,
      "learning_rate": 0.00022411156123114157,
      "loss": 4.6418,
      "step": 30490
    },
    {
      "epoch": 0.6245776421682059,
      "grad_norm": 0.28016114234924316,
      "learning_rate": 0.00022392846010456334,
      "loss": 4.6693,
      "step": 30500
    },
    {
      "epoch": 0.6247824217230151,
      "grad_norm": 0.30062586069107056,
      "learning_rate": 0.00022374533862881574,
      "loss": 4.6688,
      "step": 30510
    },
    {
      "epoch": 0.6249872012778244,
      "grad_norm": 0.3291855752468109,
      "learning_rate": 0.00022356219695962852,
      "loss": 4.6277,
      "step": 30520
    },
    {
      "epoch": 0.6251919808326337,
      "grad_norm": 0.2959108054637909,
      "learning_rate": 0.00022337903525274837,
      "loss": 4.642,
      "step": 30530
    },
    {
      "epoch": 0.625396760387443,
      "grad_norm": 0.28458911180496216,
      "learning_rate": 0.00022319585366393927,
      "loss": 4.6513,
      "step": 30540
    },
    {
      "epoch": 0.6256015399422522,
      "grad_norm": 0.3079531192779541,
      "learning_rate": 0.00022301265234898185,
      "loss": 4.6694,
      "step": 30550
    },
    {
      "epoch": 0.6258063194970614,
      "grad_norm": 0.30680134892463684,
      "learning_rate": 0.0002228294314636736,
      "loss": 4.6353,
      "step": 30560
    },
    {
      "epoch": 0.6260110990518707,
      "grad_norm": 0.29334500432014465,
      "learning_rate": 0.00022264619116382875,
      "loss": 4.6928,
      "step": 30570
    },
    {
      "epoch": 0.6262158786066799,
      "grad_norm": 0.2923773527145386,
      "learning_rate": 0.00022246293160527793,
      "loss": 4.6535,
      "step": 30580
    },
    {
      "epoch": 0.6264206581614892,
      "grad_norm": 0.31806227564811707,
      "learning_rate": 0.00022227965294386822,
      "loss": 4.685,
      "step": 30590
    },
    {
      "epoch": 0.6266254377162984,
      "grad_norm": 0.285145103931427,
      "learning_rate": 0.0002220963553354628,
      "loss": 4.6419,
      "step": 30600
    },
    {
      "epoch": 0.6268302172711077,
      "grad_norm": 0.30672797560691833,
      "learning_rate": 0.0002219130389359412,
      "loss": 4.6372,
      "step": 30610
    },
    {
      "epoch": 0.6270349968259169,
      "grad_norm": 0.30848565697669983,
      "learning_rate": 0.00022172970390119874,
      "loss": 4.6253,
      "step": 30620
    },
    {
      "epoch": 0.6272397763807261,
      "grad_norm": 0.2922016680240631,
      "learning_rate": 0.0002215463503871466,
      "loss": 4.6615,
      "step": 30630
    },
    {
      "epoch": 0.6274445559355354,
      "grad_norm": 0.2957032322883606,
      "learning_rate": 0.0002213629785497119,
      "loss": 4.6535,
      "step": 30640
    },
    {
      "epoch": 0.6276493354903446,
      "grad_norm": 0.28136834502220154,
      "learning_rate": 0.00022117958854483703,
      "loss": 4.6324,
      "step": 30650
    },
    {
      "epoch": 0.6278541150451539,
      "grad_norm": 0.3000693917274475,
      "learning_rate": 0.00022099618052847997,
      "loss": 4.6512,
      "step": 30660
    },
    {
      "epoch": 0.6280588945999631,
      "grad_norm": 0.2786167562007904,
      "learning_rate": 0.00022081275465661403,
      "loss": 4.6648,
      "step": 30670
    },
    {
      "epoch": 0.6282636741547724,
      "grad_norm": 0.2739909291267395,
      "learning_rate": 0.00022062931108522772,
      "loss": 4.6412,
      "step": 30680
    },
    {
      "epoch": 0.6284684537095816,
      "grad_norm": 0.2978883683681488,
      "learning_rate": 0.00022044584997032453,
      "loss": 4.6781,
      "step": 30690
    },
    {
      "epoch": 0.6286732332643908,
      "grad_norm": 0.3022817373275757,
      "learning_rate": 0.00022026237146792292,
      "loss": 4.669,
      "step": 30700
    },
    {
      "epoch": 0.6288780128192001,
      "grad_norm": 0.33219870924949646,
      "learning_rate": 0.00022007887573405622,
      "loss": 4.6393,
      "step": 30710
    },
    {
      "epoch": 0.6290827923740093,
      "grad_norm": 0.29787546396255493,
      "learning_rate": 0.00021989536292477217,
      "loss": 4.6452,
      "step": 30720
    },
    {
      "epoch": 0.6292875719288187,
      "grad_norm": 0.3121150732040405,
      "learning_rate": 0.00021971183319613325,
      "loss": 4.6268,
      "step": 30730
    },
    {
      "epoch": 0.6294923514836279,
      "grad_norm": 0.30353349447250366,
      "learning_rate": 0.00021952828670421628,
      "loss": 4.6538,
      "step": 30740
    },
    {
      "epoch": 0.6296971310384372,
      "grad_norm": 0.28227660059928894,
      "learning_rate": 0.0002193447236051123,
      "loss": 4.6357,
      "step": 30750
    },
    {
      "epoch": 0.6299019105932464,
      "grad_norm": 0.31946688890457153,
      "learning_rate": 0.00021916114405492651,
      "loss": 4.6547,
      "step": 30760
    },
    {
      "epoch": 0.6301066901480556,
      "grad_norm": 0.2845208942890167,
      "learning_rate": 0.00021897754820977806,
      "loss": 4.6753,
      "step": 30770
    },
    {
      "epoch": 0.6303114697028649,
      "grad_norm": 0.2979859709739685,
      "learning_rate": 0.0002187939362258,
      "loss": 4.6488,
      "step": 30780
    },
    {
      "epoch": 0.6305162492576741,
      "grad_norm": 0.27406734228134155,
      "learning_rate": 0.00021861030825913905,
      "loss": 4.6294,
      "step": 30790
    },
    {
      "epoch": 0.6307210288124834,
      "grad_norm": 0.28868621587753296,
      "learning_rate": 0.0002184266644659555,
      "loss": 4.6798,
      "step": 30800
    },
    {
      "epoch": 0.6309258083672926,
      "grad_norm": 0.2830265164375305,
      "learning_rate": 0.00021824300500242324,
      "loss": 4.6774,
      "step": 30810
    },
    {
      "epoch": 0.6311305879221019,
      "grad_norm": 0.2919405400753021,
      "learning_rate": 0.00021805933002472946,
      "loss": 4.6297,
      "step": 30820
    },
    {
      "epoch": 0.6313353674769111,
      "grad_norm": 0.29010897874832153,
      "learning_rate": 0.00021787563968907433,
      "loss": 4.6662,
      "step": 30830
    },
    {
      "epoch": 0.6315401470317203,
      "grad_norm": 0.27563849091529846,
      "learning_rate": 0.0002176919341516714,
      "loss": 4.6353,
      "step": 30840
    },
    {
      "epoch": 0.6317449265865296,
      "grad_norm": 0.2839495539665222,
      "learning_rate": 0.00021750821356874685,
      "loss": 4.6538,
      "step": 30850
    },
    {
      "epoch": 0.6319497061413388,
      "grad_norm": 0.2820706069469452,
      "learning_rate": 0.0002173244780965398,
      "loss": 4.6616,
      "step": 30860
    },
    {
      "epoch": 0.6321544856961481,
      "grad_norm": 0.30122271180152893,
      "learning_rate": 0.00021714072789130205,
      "loss": 4.6481,
      "step": 30870
    },
    {
      "epoch": 0.6323592652509573,
      "grad_norm": 0.2987443208694458,
      "learning_rate": 0.00021695696310929794,
      "loss": 4.671,
      "step": 30880
    },
    {
      "epoch": 0.6325640448057666,
      "grad_norm": 0.32780030369758606,
      "learning_rate": 0.0002167731839068041,
      "loss": 4.6852,
      "step": 30890
    },
    {
      "epoch": 0.6327688243605758,
      "grad_norm": 0.3136660158634186,
      "learning_rate": 0.00021658939044010952,
      "loss": 4.6578,
      "step": 30900
    },
    {
      "epoch": 0.632973603915385,
      "grad_norm": 0.3008350431919098,
      "learning_rate": 0.0002164055828655152,
      "loss": 4.648,
      "step": 30910
    },
    {
      "epoch": 0.6331783834701943,
      "grad_norm": 0.2733580470085144,
      "learning_rate": 0.00021622176133933443,
      "loss": 4.684,
      "step": 30920
    },
    {
      "epoch": 0.6333831630250036,
      "grad_norm": 0.297954797744751,
      "learning_rate": 0.00021603792601789196,
      "loss": 4.6215,
      "step": 30930
    },
    {
      "epoch": 0.6335879425798129,
      "grad_norm": 0.3264220952987671,
      "learning_rate": 0.00021585407705752458,
      "loss": 4.6458,
      "step": 30940
    },
    {
      "epoch": 0.6337927221346221,
      "grad_norm": 0.3224235773086548,
      "learning_rate": 0.0002156702146145805,
      "loss": 4.6816,
      "step": 30950
    },
    {
      "epoch": 0.6339975016894314,
      "grad_norm": 0.356788694858551,
      "learning_rate": 0.00021548633884541953,
      "loss": 4.6502,
      "step": 30960
    },
    {
      "epoch": 0.6342022812442406,
      "grad_norm": 0.32108941674232483,
      "learning_rate": 0.00021530244990641272,
      "loss": 4.6607,
      "step": 30970
    },
    {
      "epoch": 0.6344070607990498,
      "grad_norm": 0.27904924750328064,
      "learning_rate": 0.00021511854795394243,
      "loss": 4.6723,
      "step": 30980
    },
    {
      "epoch": 0.6346118403538591,
      "grad_norm": 0.28624746203422546,
      "learning_rate": 0.00021493463314440186,
      "loss": 4.6719,
      "step": 30990
    },
    {
      "epoch": 0.6348166199086683,
      "grad_norm": 0.2936543822288513,
      "learning_rate": 0.00021475070563419535,
      "loss": 4.6705,
      "step": 31000
    },
    {
      "epoch": 0.6350213994634776,
      "grad_norm": 0.3128805458545685,
      "learning_rate": 0.000214566765579738,
      "loss": 4.6363,
      "step": 31010
    },
    {
      "epoch": 0.6352261790182868,
      "grad_norm": 0.2871984839439392,
      "learning_rate": 0.00021438281313745564,
      "loss": 4.6357,
      "step": 31020
    },
    {
      "epoch": 0.6354309585730961,
      "grad_norm": 0.30281564593315125,
      "learning_rate": 0.00021419884846378442,
      "loss": 4.6751,
      "step": 31030
    },
    {
      "epoch": 0.6356357381279053,
      "grad_norm": 0.310698926448822,
      "learning_rate": 0.00021401487171517105,
      "loss": 4.651,
      "step": 31040
    },
    {
      "epoch": 0.6358405176827145,
      "grad_norm": 0.28934794664382935,
      "learning_rate": 0.00021383088304807247,
      "loss": 4.6618,
      "step": 31050
    },
    {
      "epoch": 0.6360452972375238,
      "grad_norm": 0.30262723565101624,
      "learning_rate": 0.0002136468826189559,
      "loss": 4.6018,
      "step": 31060
    },
    {
      "epoch": 0.636250076792333,
      "grad_norm": 0.32639217376708984,
      "learning_rate": 0.0002134628705842983,
      "loss": 4.6411,
      "step": 31070
    },
    {
      "epoch": 0.6364548563471423,
      "grad_norm": 0.3025152087211609,
      "learning_rate": 0.0002132788471005867,
      "loss": 4.6209,
      "step": 31080
    },
    {
      "epoch": 0.6366596359019515,
      "grad_norm": 0.28025031089782715,
      "learning_rate": 0.00021309481232431777,
      "loss": 4.629,
      "step": 31090
    },
    {
      "epoch": 0.6368644154567608,
      "grad_norm": 0.5899404883384705,
      "learning_rate": 0.0002129107664119978,
      "loss": 4.6398,
      "step": 31100
    },
    {
      "epoch": 0.63706919501157,
      "grad_norm": 0.280439555644989,
      "learning_rate": 0.0002127267095201426,
      "loss": 4.6465,
      "step": 31110
    },
    {
      "epoch": 0.6372739745663792,
      "grad_norm": 0.2808097004890442,
      "learning_rate": 0.0002125426418052773,
      "loss": 4.6526,
      "step": 31120
    },
    {
      "epoch": 0.6374787541211886,
      "grad_norm": 0.32698628306388855,
      "learning_rate": 0.00021235856342393614,
      "loss": 4.6345,
      "step": 31130
    },
    {
      "epoch": 0.6376835336759978,
      "grad_norm": 0.28672513365745544,
      "learning_rate": 0.0002121744745326625,
      "loss": 4.6502,
      "step": 31140
    },
    {
      "epoch": 0.6378883132308071,
      "grad_norm": 0.2829684913158417,
      "learning_rate": 0.00021199037528800877,
      "loss": 4.6454,
      "step": 31150
    },
    {
      "epoch": 0.6380930927856163,
      "grad_norm": 0.2866877615451813,
      "learning_rate": 0.000211806265846536,
      "loss": 4.6324,
      "step": 31160
    },
    {
      "epoch": 0.6382978723404256,
      "grad_norm": 0.2783838212490082,
      "learning_rate": 0.00021162214636481422,
      "loss": 4.6448,
      "step": 31170
    },
    {
      "epoch": 0.6385026518952348,
      "grad_norm": 0.26938238739967346,
      "learning_rate": 0.00021143801699942142,
      "loss": 4.6041,
      "step": 31180
    },
    {
      "epoch": 0.638707431450044,
      "grad_norm": 0.35899850726127625,
      "learning_rate": 0.00021125387790694452,
      "loss": 4.6596,
      "step": 31190
    },
    {
      "epoch": 0.6389122110048533,
      "grad_norm": 0.3072209358215332,
      "learning_rate": 0.00021106972924397858,
      "loss": 4.6259,
      "step": 31200
    },
    {
      "epoch": 0.6391169905596625,
      "grad_norm": 0.3054019808769226,
      "learning_rate": 0.00021088557116712665,
      "loss": 4.6186,
      "step": 31210
    },
    {
      "epoch": 0.6393217701144718,
      "grad_norm": 0.30291202664375305,
      "learning_rate": 0.000210701403833,
      "loss": 4.638,
      "step": 31220
    },
    {
      "epoch": 0.639526549669281,
      "grad_norm": 0.31307119131088257,
      "learning_rate": 0.00021051722739821754,
      "loss": 4.6226,
      "step": 31230
    },
    {
      "epoch": 0.6397313292240903,
      "grad_norm": 0.2849477231502533,
      "learning_rate": 0.00021033304201940615,
      "loss": 4.6709,
      "step": 31240
    },
    {
      "epoch": 0.6399361087788995,
      "grad_norm": 0.29145699739456177,
      "learning_rate": 0.0002101488478532001,
      "loss": 4.6387,
      "step": 31250
    },
    {
      "epoch": 0.6401408883337087,
      "grad_norm": 0.5871053338050842,
      "learning_rate": 0.00020996464505624132,
      "loss": 4.614,
      "step": 31260
    },
    {
      "epoch": 0.640345667888518,
      "grad_norm": 0.28802430629730225,
      "learning_rate": 0.00020978043378517907,
      "loss": 4.6842,
      "step": 31270
    },
    {
      "epoch": 0.6405504474433272,
      "grad_norm": 0.3322353661060333,
      "learning_rate": 0.00020959621419666956,
      "loss": 4.6597,
      "step": 31280
    },
    {
      "epoch": 0.6407552269981365,
      "grad_norm": 0.2919965088367462,
      "learning_rate": 0.0002094119864473763,
      "loss": 4.6283,
      "step": 31290
    },
    {
      "epoch": 0.6409600065529457,
      "grad_norm": 0.29532623291015625,
      "learning_rate": 0.0002092277506939698,
      "loss": 4.6599,
      "step": 31300
    },
    {
      "epoch": 0.641164786107755,
      "grad_norm": 0.2982107698917389,
      "learning_rate": 0.00020904350709312727,
      "loss": 4.6808,
      "step": 31310
    },
    {
      "epoch": 0.6413695656625642,
      "grad_norm": 0.3330329954624176,
      "learning_rate": 0.00020885925580153259,
      "loss": 4.6049,
      "step": 31320
    },
    {
      "epoch": 0.6415743452173734,
      "grad_norm": 0.2956121265888214,
      "learning_rate": 0.00020867499697587612,
      "loss": 4.6494,
      "step": 31330
    },
    {
      "epoch": 0.6417791247721828,
      "grad_norm": 0.2796705961227417,
      "learning_rate": 0.00020849073077285475,
      "loss": 4.6404,
      "step": 31340
    },
    {
      "epoch": 0.641983904326992,
      "grad_norm": 0.2776346802711487,
      "learning_rate": 0.00020830645734917154,
      "loss": 4.6322,
      "step": 31350
    },
    {
      "epoch": 0.6421886838818013,
      "grad_norm": 0.2897033989429474,
      "learning_rate": 0.00020812217686153584,
      "loss": 4.6275,
      "step": 31360
    },
    {
      "epoch": 0.6423934634366105,
      "grad_norm": 0.29280951619148254,
      "learning_rate": 0.0002079378894666628,
      "loss": 4.6103,
      "step": 31370
    },
    {
      "epoch": 0.6425982429914198,
      "grad_norm": 0.2698272466659546,
      "learning_rate": 0.0002077535953212737,
      "loss": 4.6664,
      "step": 31380
    },
    {
      "epoch": 0.642803022546229,
      "grad_norm": 0.27823346853256226,
      "learning_rate": 0.00020756929458209524,
      "loss": 4.6506,
      "step": 31390
    },
    {
      "epoch": 0.6430078021010383,
      "grad_norm": 0.29688760638237,
      "learning_rate": 0.00020738498740585998,
      "loss": 4.691,
      "step": 31400
    },
    {
      "epoch": 0.6432125816558475,
      "grad_norm": 0.3246428668498993,
      "learning_rate": 0.00020720067394930594,
      "loss": 4.6097,
      "step": 31410
    },
    {
      "epoch": 0.6434173612106567,
      "grad_norm": 0.2977891266345978,
      "learning_rate": 0.00020701635436917633,
      "loss": 4.6249,
      "step": 31420
    },
    {
      "epoch": 0.643622140765466,
      "grad_norm": 0.28034883737564087,
      "learning_rate": 0.00020683202882221965,
      "loss": 4.629,
      "step": 31430
    },
    {
      "epoch": 0.6438269203202752,
      "grad_norm": 0.2818579375743866,
      "learning_rate": 0.00020664769746518951,
      "loss": 4.6321,
      "step": 31440
    },
    {
      "epoch": 0.6440316998750845,
      "grad_norm": 0.30722710490226746,
      "learning_rate": 0.00020646336045484449,
      "loss": 4.6416,
      "step": 31450
    },
    {
      "epoch": 0.6442364794298937,
      "grad_norm": 0.6170060634613037,
      "learning_rate": 0.0002062790179479478,
      "loss": 4.6672,
      "step": 31460
    },
    {
      "epoch": 0.644441258984703,
      "grad_norm": 0.3114047646522522,
      "learning_rate": 0.00020609467010126756,
      "loss": 4.6055,
      "step": 31470
    },
    {
      "epoch": 0.6446460385395122,
      "grad_norm": 0.32174360752105713,
      "learning_rate": 0.00020591031707157626,
      "loss": 4.63,
      "step": 31480
    },
    {
      "epoch": 0.6448508180943214,
      "grad_norm": 0.2874440848827362,
      "learning_rate": 0.00020572595901565075,
      "loss": 4.6365,
      "step": 31490
    },
    {
      "epoch": 0.6450555976491307,
      "grad_norm": 0.28202199935913086,
      "learning_rate": 0.00020554159609027236,
      "loss": 4.6228,
      "step": 31500
    },
    {
      "epoch": 0.6452603772039399,
      "grad_norm": 0.29981857538223267,
      "learning_rate": 0.0002053572284522265,
      "loss": 4.6578,
      "step": 31510
    },
    {
      "epoch": 0.6454651567587492,
      "grad_norm": 0.2835997939109802,
      "learning_rate": 0.00020517285625830248,
      "loss": 4.6091,
      "step": 31520
    },
    {
      "epoch": 0.6456699363135584,
      "grad_norm": 0.26887619495391846,
      "learning_rate": 0.00020498847966529354,
      "loss": 4.6203,
      "step": 31530
    },
    {
      "epoch": 0.6458747158683678,
      "grad_norm": 0.39226752519607544,
      "learning_rate": 0.00020480409882999668,
      "loss": 4.6366,
      "step": 31540
    },
    {
      "epoch": 0.646079495423177,
      "grad_norm": 0.28772279620170593,
      "learning_rate": 0.00020461971390921252,
      "loss": 4.6073,
      "step": 31550
    },
    {
      "epoch": 0.6462842749779862,
      "grad_norm": 0.2829737067222595,
      "learning_rate": 0.00020443532505974514,
      "loss": 4.6557,
      "step": 31560
    },
    {
      "epoch": 0.6464890545327955,
      "grad_norm": 0.28044554591178894,
      "learning_rate": 0.00020425093243840189,
      "loss": 4.6202,
      "step": 31570
    },
    {
      "epoch": 0.6466938340876047,
      "grad_norm": 0.2747333347797394,
      "learning_rate": 0.00020406653620199338,
      "loss": 4.6379,
      "step": 31580
    },
    {
      "epoch": 0.646898613642414,
      "grad_norm": 0.30595648288726807,
      "learning_rate": 0.00020388213650733344,
      "loss": 4.6443,
      "step": 31590
    },
    {
      "epoch": 0.6471033931972232,
      "grad_norm": 0.27070680260658264,
      "learning_rate": 0.00020369773351123856,
      "loss": 4.6596,
      "step": 31600
    },
    {
      "epoch": 0.6473081727520325,
      "grad_norm": 0.2805611789226532,
      "learning_rate": 0.00020351332737052823,
      "loss": 4.6697,
      "step": 31610
    },
    {
      "epoch": 0.6475129523068417,
      "grad_norm": 0.3111336827278137,
      "learning_rate": 0.00020332891824202461,
      "loss": 4.6382,
      "step": 31620
    },
    {
      "epoch": 0.6477177318616509,
      "grad_norm": 0.28219151496887207,
      "learning_rate": 0.00020314450628255228,
      "loss": 4.6541,
      "step": 31630
    },
    {
      "epoch": 0.6479225114164602,
      "grad_norm": 0.28346508741378784,
      "learning_rate": 0.00020296009164893838,
      "loss": 4.5858,
      "step": 31640
    },
    {
      "epoch": 0.6481272909712694,
      "grad_norm": 0.28758737444877625,
      "learning_rate": 0.00020277567449801222,
      "loss": 4.6349,
      "step": 31650
    },
    {
      "epoch": 0.6483320705260787,
      "grad_norm": 0.3404221534729004,
      "learning_rate": 0.00020259125498660533,
      "loss": 4.648,
      "step": 31660
    },
    {
      "epoch": 0.6485368500808879,
      "grad_norm": 0.28170305490493774,
      "learning_rate": 0.00020240683327155104,
      "loss": 4.6316,
      "step": 31670
    },
    {
      "epoch": 0.6487416296356971,
      "grad_norm": 0.28445863723754883,
      "learning_rate": 0.00020222240950968482,
      "loss": 4.6254,
      "step": 31680
    },
    {
      "epoch": 0.6489464091905064,
      "grad_norm": 0.27748405933380127,
      "learning_rate": 0.00020203798385784384,
      "loss": 4.6671,
      "step": 31690
    },
    {
      "epoch": 0.6491511887453156,
      "grad_norm": 0.3047666847705841,
      "learning_rate": 0.00020185355647286678,
      "loss": 4.6289,
      "step": 31700
    },
    {
      "epoch": 0.6493559683001249,
      "grad_norm": 0.29950475692749023,
      "learning_rate": 0.0002016691275115937,
      "loss": 4.681,
      "step": 31710
    },
    {
      "epoch": 0.6495607478549341,
      "grad_norm": 0.3117060959339142,
      "learning_rate": 0.0002014846971308662,
      "loss": 4.6548,
      "step": 31720
    },
    {
      "epoch": 0.6497655274097434,
      "grad_norm": 0.28721022605895996,
      "learning_rate": 0.00020130026548752708,
      "loss": 4.6387,
      "step": 31730
    },
    {
      "epoch": 0.6499703069645527,
      "grad_norm": 0.28561025857925415,
      "learning_rate": 0.00020111583273842002,
      "loss": 4.5913,
      "step": 31740
    },
    {
      "epoch": 0.650175086519362,
      "grad_norm": 0.2865810990333557,
      "learning_rate": 0.00020093139904038987,
      "loss": 4.6432,
      "step": 31750
    },
    {
      "epoch": 0.6503798660741712,
      "grad_norm": 0.27878129482269287,
      "learning_rate": 0.00020074696455028214,
      "loss": 4.6465,
      "step": 31760
    },
    {
      "epoch": 0.6505846456289804,
      "grad_norm": 0.28597012162208557,
      "learning_rate": 0.00020056252942494304,
      "loss": 4.6175,
      "step": 31770
    },
    {
      "epoch": 0.6507894251837897,
      "grad_norm": 0.29955506324768066,
      "learning_rate": 0.00020037809382121937,
      "loss": 4.6587,
      "step": 31780
    },
    {
      "epoch": 0.6509942047385989,
      "grad_norm": 0.2863275110721588,
      "learning_rate": 0.00020019365789595826,
      "loss": 4.6116,
      "step": 31790
    },
    {
      "epoch": 0.6511989842934082,
      "grad_norm": 0.2816847860813141,
      "learning_rate": 0.00020000922180600728,
      "loss": 4.6215,
      "step": 31800
    },
    {
      "epoch": 0.6514037638482174,
      "grad_norm": 0.3599647581577301,
      "learning_rate": 0.00019982478570821383,
      "loss": 4.6497,
      "step": 31810
    },
    {
      "epoch": 0.6516085434030267,
      "grad_norm": 0.34417176246643066,
      "learning_rate": 0.00019964034975942556,
      "loss": 4.6141,
      "step": 31820
    },
    {
      "epoch": 0.6518133229578359,
      "grad_norm": 0.28384891152381897,
      "learning_rate": 0.00019945591411649,
      "loss": 4.6682,
      "step": 31830
    },
    {
      "epoch": 0.6520181025126451,
      "grad_norm": 0.2900743782520294,
      "learning_rate": 0.00019927147893625434,
      "loss": 4.6286,
      "step": 31840
    },
    {
      "epoch": 0.6522228820674544,
      "grad_norm": 0.2789555490016937,
      "learning_rate": 0.00019908704437556548,
      "loss": 4.6383,
      "step": 31850
    },
    {
      "epoch": 0.6524276616222636,
      "grad_norm": 0.27704811096191406,
      "learning_rate": 0.00019890261059126945,
      "loss": 4.6133,
      "step": 31860
    },
    {
      "epoch": 0.6526324411770729,
      "grad_norm": 0.31395819783210754,
      "learning_rate": 0.00019871817774021205,
      "loss": 4.6613,
      "step": 31870
    },
    {
      "epoch": 0.6528372207318821,
      "grad_norm": 0.27373069524765015,
      "learning_rate": 0.00019853374597923813,
      "loss": 4.6638,
      "step": 31880
    },
    {
      "epoch": 0.6530420002866913,
      "grad_norm": 0.32742395997047424,
      "learning_rate": 0.00019834931546519155,
      "loss": 4.6096,
      "step": 31890
    },
    {
      "epoch": 0.6532467798415006,
      "grad_norm": 0.33728116750717163,
      "learning_rate": 0.00019816488635491516,
      "loss": 4.63,
      "step": 31900
    },
    {
      "epoch": 0.6534515593963098,
      "grad_norm": 0.32113081216812134,
      "learning_rate": 0.00019798045880525057,
      "loss": 4.6654,
      "step": 31910
    },
    {
      "epoch": 0.6536563389511191,
      "grad_norm": 0.4503423571586609,
      "learning_rate": 0.00019779603297303826,
      "loss": 4.6327,
      "step": 31920
    },
    {
      "epoch": 0.6538611185059283,
      "grad_norm": 0.2936616837978363,
      "learning_rate": 0.0001976116090151169,
      "loss": 4.6077,
      "step": 31930
    },
    {
      "epoch": 0.6540658980607376,
      "grad_norm": 0.3016274869441986,
      "learning_rate": 0.00019742718708832386,
      "loss": 4.5875,
      "step": 31940
    },
    {
      "epoch": 0.6542706776155469,
      "grad_norm": 0.2945941388607025,
      "learning_rate": 0.00019724276734949463,
      "loss": 4.6379,
      "step": 31950
    },
    {
      "epoch": 0.6544754571703562,
      "grad_norm": 0.2721746861934662,
      "learning_rate": 0.00019705834995546292,
      "loss": 4.6521,
      "step": 31960
    },
    {
      "epoch": 0.6546802367251654,
      "grad_norm": 0.2935369610786438,
      "learning_rate": 0.0001968739350630604,
      "loss": 4.6509,
      "step": 31970
    },
    {
      "epoch": 0.6548850162799746,
      "grad_norm": 0.31681233644485474,
      "learning_rate": 0.0001966895228291167,
      "loss": 4.6515,
      "step": 31980
    },
    {
      "epoch": 0.6550897958347839,
      "grad_norm": 0.31349703669548035,
      "learning_rate": 0.0001965051134104591,
      "loss": 4.6517,
      "step": 31990
    },
    {
      "epoch": 0.6552945753895931,
      "grad_norm": 0.2852419912815094,
      "learning_rate": 0.0001963207069639124,
      "loss": 4.6436,
      "step": 32000
    },
    {
      "epoch": 0.6552945753895931,
      "eval_loss": 4.642320156097412,
      "eval_runtime": 4.3846,
      "eval_samples_per_second": 265.933,
      "eval_steps_per_second": 33.299,
      "step": 32000
    },
    {
      "epoch": 0.6554993549444024,
      "grad_norm": 0.3128270208835602,
      "learning_rate": 0.00019613630364629907,
      "loss": 4.629,
      "step": 32010
    },
    {
      "epoch": 0.6557041344992116,
      "grad_norm": 0.30352678894996643,
      "learning_rate": 0.00019595190361443876,
      "loss": 4.6378,
      "step": 32020
    },
    {
      "epoch": 0.6559089140540209,
      "grad_norm": 0.290731281042099,
      "learning_rate": 0.0001957675070251485,
      "loss": 4.6158,
      "step": 32030
    },
    {
      "epoch": 0.6561136936088301,
      "grad_norm": 0.3104674518108368,
      "learning_rate": 0.0001955831140352422,
      "loss": 4.6631,
      "step": 32040
    },
    {
      "epoch": 0.6563184731636393,
      "grad_norm": 0.2891213893890381,
      "learning_rate": 0.00019539872480153078,
      "loss": 4.6223,
      "step": 32050
    },
    {
      "epoch": 0.6565232527184486,
      "grad_norm": 0.2789028286933899,
      "learning_rate": 0.0001952143394808221,
      "loss": 4.6225,
      "step": 32060
    },
    {
      "epoch": 0.6567280322732578,
      "grad_norm": 0.2898847758769989,
      "learning_rate": 0.00019502995822992043,
      "loss": 4.6441,
      "step": 32070
    },
    {
      "epoch": 0.6569328118280671,
      "grad_norm": 0.2967241406440735,
      "learning_rate": 0.00019484558120562675,
      "loss": 4.6284,
      "step": 32080
    },
    {
      "epoch": 0.6571375913828763,
      "grad_norm": 0.28619837760925293,
      "learning_rate": 0.00019466120856473848,
      "loss": 4.6193,
      "step": 32090
    },
    {
      "epoch": 0.6573423709376855,
      "grad_norm": 0.29414695501327515,
      "learning_rate": 0.0001944768404640492,
      "loss": 4.6234,
      "step": 32100
    },
    {
      "epoch": 0.6575471504924948,
      "grad_norm": 0.28031256794929504,
      "learning_rate": 0.00019429247706034865,
      "loss": 4.6213,
      "step": 32110
    },
    {
      "epoch": 0.657751930047304,
      "grad_norm": 0.3011209964752197,
      "learning_rate": 0.00019410811851042269,
      "loss": 4.6298,
      "step": 32120
    },
    {
      "epoch": 0.6579567096021133,
      "grad_norm": 0.3096432387828827,
      "learning_rate": 0.00019392376497105293,
      "loss": 4.6453,
      "step": 32130
    },
    {
      "epoch": 0.6581614891569225,
      "grad_norm": 0.28561797738075256,
      "learning_rate": 0.00019373941659901674,
      "loss": 4.6063,
      "step": 32140
    },
    {
      "epoch": 0.6583662687117319,
      "grad_norm": 0.3114117383956909,
      "learning_rate": 0.00019355507355108705,
      "loss": 4.5976,
      "step": 32150
    },
    {
      "epoch": 0.6585710482665411,
      "grad_norm": 0.2873578667640686,
      "learning_rate": 0.0001933707359840324,
      "loss": 4.6026,
      "step": 32160
    },
    {
      "epoch": 0.6587758278213504,
      "grad_norm": 0.2891336679458618,
      "learning_rate": 0.00019318640405461654,
      "loss": 4.6773,
      "step": 32170
    },
    {
      "epoch": 0.6589806073761596,
      "grad_norm": 0.3008500933647156,
      "learning_rate": 0.00019300207791959857,
      "loss": 4.6259,
      "step": 32180
    },
    {
      "epoch": 0.6591853869309688,
      "grad_norm": 0.3336765468120575,
      "learning_rate": 0.00019281775773573247,
      "loss": 4.6412,
      "step": 32190
    },
    {
      "epoch": 0.6593901664857781,
      "grad_norm": 0.2977740466594696,
      "learning_rate": 0.00019263344365976737,
      "loss": 4.6599,
      "step": 32200
    },
    {
      "epoch": 0.6595949460405873,
      "grad_norm": 0.30444103479385376,
      "learning_rate": 0.00019244913584844694,
      "loss": 4.652,
      "step": 32210
    },
    {
      "epoch": 0.6597997255953966,
      "grad_norm": 0.34079089760780334,
      "learning_rate": 0.00019226483445850974,
      "loss": 4.656,
      "step": 32220
    },
    {
      "epoch": 0.6600045051502058,
      "grad_norm": 0.28978464007377625,
      "learning_rate": 0.00019208053964668887,
      "loss": 4.6458,
      "step": 32230
    },
    {
      "epoch": 0.660209284705015,
      "grad_norm": 0.28713274002075195,
      "learning_rate": 0.0001918962515697117,
      "loss": 4.6382,
      "step": 32240
    },
    {
      "epoch": 0.6604140642598243,
      "grad_norm": 0.3080507516860962,
      "learning_rate": 0.00019171197038429994,
      "loss": 4.6429,
      "step": 32250
    },
    {
      "epoch": 0.6606188438146335,
      "grad_norm": 0.3299945592880249,
      "learning_rate": 0.00019152769624716947,
      "loss": 4.6411,
      "step": 32260
    },
    {
      "epoch": 0.6608236233694428,
      "grad_norm": 0.3015105426311493,
      "learning_rate": 0.0001913434293150302,
      "loss": 4.6008,
      "step": 32270
    },
    {
      "epoch": 0.661028402924252,
      "grad_norm": 0.28456807136535645,
      "learning_rate": 0.00019115916974458584,
      "loss": 4.6208,
      "step": 32280
    },
    {
      "epoch": 0.6612331824790613,
      "grad_norm": 0.4723943769931793,
      "learning_rate": 0.00019097491769253373,
      "loss": 4.6645,
      "step": 32290
    },
    {
      "epoch": 0.6614379620338705,
      "grad_norm": 0.2876240611076355,
      "learning_rate": 0.00019079067331556507,
      "loss": 4.6347,
      "step": 32300
    },
    {
      "epoch": 0.6616427415886798,
      "grad_norm": 0.40209028124809265,
      "learning_rate": 0.00019060643677036435,
      "loss": 4.6263,
      "step": 32310
    },
    {
      "epoch": 0.661847521143489,
      "grad_norm": 0.2876284122467041,
      "learning_rate": 0.0001904222082136096,
      "loss": 4.6409,
      "step": 32320
    },
    {
      "epoch": 0.6620523006982982,
      "grad_norm": 0.3027927875518799,
      "learning_rate": 0.00019023798780197173,
      "loss": 4.6446,
      "step": 32330
    },
    {
      "epoch": 0.6622570802531075,
      "grad_norm": 0.2995139956474304,
      "learning_rate": 0.0001900537756921151,
      "loss": 4.698,
      "step": 32340
    },
    {
      "epoch": 0.6624618598079168,
      "grad_norm": 0.28222161531448364,
      "learning_rate": 0.00018986957204069658,
      "loss": 4.6691,
      "step": 32350
    },
    {
      "epoch": 0.6626666393627261,
      "grad_norm": 0.2969187796115875,
      "learning_rate": 0.0001896853770043663,
      "loss": 4.6232,
      "step": 32360
    },
    {
      "epoch": 0.6628714189175353,
      "grad_norm": 0.2937006950378418,
      "learning_rate": 0.00018950119073976674,
      "loss": 4.6295,
      "step": 32370
    },
    {
      "epoch": 0.6630761984723446,
      "grad_norm": 0.3069112300872803,
      "learning_rate": 0.00018931701340353307,
      "loss": 4.6248,
      "step": 32380
    },
    {
      "epoch": 0.6632809780271538,
      "grad_norm": 0.2813648581504822,
      "learning_rate": 0.0001891328451522928,
      "loss": 4.6259,
      "step": 32390
    },
    {
      "epoch": 0.663485757581963,
      "grad_norm": 0.31179484724998474,
      "learning_rate": 0.00018894868614266578,
      "loss": 4.6409,
      "step": 32400
    },
    {
      "epoch": 0.6636905371367723,
      "grad_norm": 0.2886295020580292,
      "learning_rate": 0.000188764536531264,
      "loss": 4.6497,
      "step": 32410
    },
    {
      "epoch": 0.6638953166915815,
      "grad_norm": 0.28318530321121216,
      "learning_rate": 0.0001885803964746913,
      "loss": 4.6297,
      "step": 32420
    },
    {
      "epoch": 0.6641000962463908,
      "grad_norm": 0.28957971930503845,
      "learning_rate": 0.0001883962661295436,
      "loss": 4.6455,
      "step": 32430
    },
    {
      "epoch": 0.6643048758012,
      "grad_norm": 0.32726556062698364,
      "learning_rate": 0.00018821214565240842,
      "loss": 4.6106,
      "step": 32440
    },
    {
      "epoch": 0.6645096553560093,
      "grad_norm": 0.281751424074173,
      "learning_rate": 0.000188028035199865,
      "loss": 4.6541,
      "step": 32450
    },
    {
      "epoch": 0.6647144349108185,
      "grad_norm": 0.30658701062202454,
      "learning_rate": 0.000187843934928484,
      "loss": 4.659,
      "step": 32460
    },
    {
      "epoch": 0.6649192144656277,
      "grad_norm": 0.3314231038093567,
      "learning_rate": 0.00018765984499482735,
      "loss": 4.6528,
      "step": 32470
    },
    {
      "epoch": 0.665123994020437,
      "grad_norm": 0.2996695041656494,
      "learning_rate": 0.00018747576555544842,
      "loss": 4.6306,
      "step": 32480
    },
    {
      "epoch": 0.6653287735752462,
      "grad_norm": 0.30856266617774963,
      "learning_rate": 0.00018729169676689127,
      "loss": 4.6478,
      "step": 32490
    },
    {
      "epoch": 0.6655335531300555,
      "grad_norm": 0.2853170931339264,
      "learning_rate": 0.00018710763878569128,
      "loss": 4.6279,
      "step": 32500
    },
    {
      "epoch": 0.6657383326848647,
      "grad_norm": 0.31832218170166016,
      "learning_rate": 0.0001869235917683744,
      "loss": 4.659,
      "step": 32510
    },
    {
      "epoch": 0.665943112239674,
      "grad_norm": 0.3200172185897827,
      "learning_rate": 0.00018673955587145743,
      "loss": 4.6466,
      "step": 32520
    },
    {
      "epoch": 0.6661478917944832,
      "grad_norm": 0.32725638151168823,
      "learning_rate": 0.00018655553125144757,
      "loss": 4.5589,
      "step": 32530
    },
    {
      "epoch": 0.6663526713492924,
      "grad_norm": 0.32515549659729004,
      "learning_rate": 0.00018637151806484247,
      "loss": 4.6269,
      "step": 32540
    },
    {
      "epoch": 0.6665574509041018,
      "grad_norm": 0.3151293396949768,
      "learning_rate": 0.0001861875164681302,
      "loss": 4.6074,
      "step": 32550
    },
    {
      "epoch": 0.666762230458911,
      "grad_norm": 0.28322288393974304,
      "learning_rate": 0.00018600352661778867,
      "loss": 4.6232,
      "step": 32560
    },
    {
      "epoch": 0.6669670100137203,
      "grad_norm": 0.30027103424072266,
      "learning_rate": 0.00018581954867028612,
      "loss": 4.6656,
      "step": 32570
    },
    {
      "epoch": 0.6671717895685295,
      "grad_norm": 0.35422566533088684,
      "learning_rate": 0.00018563558278208035,
      "loss": 4.629,
      "step": 32580
    },
    {
      "epoch": 0.6673765691233388,
      "grad_norm": 0.30949604511260986,
      "learning_rate": 0.00018545162910961924,
      "loss": 4.6407,
      "step": 32590
    },
    {
      "epoch": 0.667581348678148,
      "grad_norm": 0.29101812839508057,
      "learning_rate": 0.00018526768780934005,
      "loss": 4.6443,
      "step": 32600
    },
    {
      "epoch": 0.6677861282329572,
      "grad_norm": 0.3094857931137085,
      "learning_rate": 0.00018508375903766963,
      "loss": 4.6171,
      "step": 32610
    },
    {
      "epoch": 0.6679909077877665,
      "grad_norm": 0.2886817157268524,
      "learning_rate": 0.00018489984295102417,
      "loss": 4.6539,
      "step": 32620
    },
    {
      "epoch": 0.6681956873425757,
      "grad_norm": 0.29238563776016235,
      "learning_rate": 0.00018471593970580892,
      "loss": 4.6522,
      "step": 32630
    },
    {
      "epoch": 0.668400466897385,
      "grad_norm": 0.307925283908844,
      "learning_rate": 0.00018453204945841835,
      "loss": 4.6576,
      "step": 32640
    },
    {
      "epoch": 0.6686052464521942,
      "grad_norm": 0.3067364990711212,
      "learning_rate": 0.00018434817236523589,
      "loss": 4.677,
      "step": 32650
    },
    {
      "epoch": 0.6688100260070035,
      "grad_norm": 0.3166733384132385,
      "learning_rate": 0.0001841643085826338,
      "loss": 4.6513,
      "step": 32660
    },
    {
      "epoch": 0.6690148055618127,
      "grad_norm": 0.30757683515548706,
      "learning_rate": 0.00018398045826697286,
      "loss": 4.6589,
      "step": 32670
    },
    {
      "epoch": 0.6692195851166219,
      "grad_norm": 0.3488108515739441,
      "learning_rate": 0.00018379662157460254,
      "loss": 4.5935,
      "step": 32680
    },
    {
      "epoch": 0.6694243646714312,
      "grad_norm": 0.29875683784484863,
      "learning_rate": 0.00018361279866186078,
      "loss": 4.6348,
      "step": 32690
    },
    {
      "epoch": 0.6696291442262404,
      "grad_norm": 0.310260146856308,
      "learning_rate": 0.00018342898968507355,
      "loss": 4.6554,
      "step": 32700
    },
    {
      "epoch": 0.6698339237810497,
      "grad_norm": 0.3225691318511963,
      "learning_rate": 0.00018324519480055528,
      "loss": 4.6241,
      "step": 32710
    },
    {
      "epoch": 0.6700387033358589,
      "grad_norm": 0.31346482038497925,
      "learning_rate": 0.00018306141416460814,
      "loss": 4.6227,
      "step": 32720
    },
    {
      "epoch": 0.6702434828906682,
      "grad_norm": 0.34079501032829285,
      "learning_rate": 0.00018287764793352236,
      "loss": 4.6266,
      "step": 32730
    },
    {
      "epoch": 0.6704482624454774,
      "grad_norm": 0.2941148579120636,
      "learning_rate": 0.00018269389626357588,
      "loss": 4.6512,
      "step": 32740
    },
    {
      "epoch": 0.6706530420002866,
      "grad_norm": 0.32868993282318115,
      "learning_rate": 0.00018251015931103427,
      "loss": 4.6312,
      "step": 32750
    },
    {
      "epoch": 0.670857821555096,
      "grad_norm": 0.2995958924293518,
      "learning_rate": 0.00018232643723215048,
      "loss": 4.6151,
      "step": 32760
    },
    {
      "epoch": 0.6710626011099052,
      "grad_norm": 0.3182849586009979,
      "learning_rate": 0.00018214273018316498,
      "loss": 4.652,
      "step": 32770
    },
    {
      "epoch": 0.6712673806647145,
      "grad_norm": 0.28864336013793945,
      "learning_rate": 0.00018195903832030525,
      "loss": 4.6374,
      "step": 32780
    },
    {
      "epoch": 0.6714721602195237,
      "grad_norm": 0.2957040071487427,
      "learning_rate": 0.00018177536179978608,
      "loss": 4.6219,
      "step": 32790
    },
    {
      "epoch": 0.671676939774333,
      "grad_norm": 0.31795555353164673,
      "learning_rate": 0.00018159170077780904,
      "loss": 4.624,
      "step": 32800
    },
    {
      "epoch": 0.6718817193291422,
      "grad_norm": 0.29311829805374146,
      "learning_rate": 0.00018140805541056262,
      "loss": 4.6721,
      "step": 32810
    },
    {
      "epoch": 0.6720864988839514,
      "grad_norm": 0.28791841864585876,
      "learning_rate": 0.00018122442585422193,
      "loss": 4.6213,
      "step": 32820
    },
    {
      "epoch": 0.6722912784387607,
      "grad_norm": 0.3204101026058197,
      "learning_rate": 0.0001810408122649488,
      "loss": 4.6312,
      "step": 32830
    },
    {
      "epoch": 0.6724960579935699,
      "grad_norm": 0.3177260756492615,
      "learning_rate": 0.00018085721479889114,
      "loss": 4.599,
      "step": 32840
    },
    {
      "epoch": 0.6727008375483792,
      "grad_norm": 0.3134481906890869,
      "learning_rate": 0.00018067363361218348,
      "loss": 4.6367,
      "step": 32850
    },
    {
      "epoch": 0.6729056171031884,
      "grad_norm": 0.31451141834259033,
      "learning_rate": 0.00018049006886094638,
      "loss": 4.6799,
      "step": 32860
    },
    {
      "epoch": 0.6731103966579977,
      "grad_norm": 0.31027689576148987,
      "learning_rate": 0.00018030652070128635,
      "loss": 4.666,
      "step": 32870
    },
    {
      "epoch": 0.6733151762128069,
      "grad_norm": 0.279913991689682,
      "learning_rate": 0.00018012298928929594,
      "loss": 4.6256,
      "step": 32880
    },
    {
      "epoch": 0.6735199557676161,
      "grad_norm": 0.2795368731021881,
      "learning_rate": 0.0001799394747810534,
      "loss": 4.6507,
      "step": 32890
    },
    {
      "epoch": 0.6737247353224254,
      "grad_norm": 0.3519022464752197,
      "learning_rate": 0.00017975597733262252,
      "loss": 4.6403,
      "step": 32900
    },
    {
      "epoch": 0.6739295148772346,
      "grad_norm": 0.3112083971500397,
      "learning_rate": 0.0001795724971000527,
      "loss": 4.6478,
      "step": 32910
    },
    {
      "epoch": 0.6741342944320439,
      "grad_norm": 0.281672865152359,
      "learning_rate": 0.0001793890342393786,
      "loss": 4.671,
      "step": 32920
    },
    {
      "epoch": 0.6743390739868531,
      "grad_norm": 0.3238334357738495,
      "learning_rate": 0.00017920558890662015,
      "loss": 4.6135,
      "step": 32930
    },
    {
      "epoch": 0.6745438535416624,
      "grad_norm": 0.29171717166900635,
      "learning_rate": 0.00017902216125778244,
      "loss": 4.6022,
      "step": 32940
    },
    {
      "epoch": 0.6747486330964716,
      "grad_norm": 0.31077757477760315,
      "learning_rate": 0.00017883875144885546,
      "loss": 4.6145,
      "step": 32950
    },
    {
      "epoch": 0.674953412651281,
      "grad_norm": 0.2973254323005676,
      "learning_rate": 0.000178655359635814,
      "loss": 4.6681,
      "step": 32960
    },
    {
      "epoch": 0.6751581922060902,
      "grad_norm": 0.292275607585907,
      "learning_rate": 0.0001784719859746176,
      "loss": 4.6316,
      "step": 32970
    },
    {
      "epoch": 0.6753629717608994,
      "grad_norm": 0.3061160147190094,
      "learning_rate": 0.00017828863062121028,
      "loss": 4.6434,
      "step": 32980
    },
    {
      "epoch": 0.6755677513157087,
      "grad_norm": 0.31540408730506897,
      "learning_rate": 0.00017810529373152057,
      "loss": 4.63,
      "step": 32990
    },
    {
      "epoch": 0.6757725308705179,
      "grad_norm": 0.31090953946113586,
      "learning_rate": 0.00017792197546146135,
      "loss": 4.6172,
      "step": 33000
    },
    {
      "epoch": 0.6759773104253272,
      "grad_norm": 0.28982412815093994,
      "learning_rate": 0.00017773867596692947,
      "loss": 4.6441,
      "step": 33010
    },
    {
      "epoch": 0.6761820899801364,
      "grad_norm": 0.2939661145210266,
      "learning_rate": 0.00017755539540380603,
      "loss": 4.6386,
      "step": 33020
    },
    {
      "epoch": 0.6763868695349456,
      "grad_norm": 0.28937020897865295,
      "learning_rate": 0.0001773721339279559,
      "loss": 4.6325,
      "step": 33030
    },
    {
      "epoch": 0.6765916490897549,
      "grad_norm": 0.2975783944129944,
      "learning_rate": 0.00017718889169522788,
      "loss": 4.6562,
      "step": 33040
    },
    {
      "epoch": 0.6767964286445641,
      "grad_norm": 0.294565349817276,
      "learning_rate": 0.0001770056688614541,
      "loss": 4.6427,
      "step": 33050
    },
    {
      "epoch": 0.6770012081993734,
      "grad_norm": 0.29006120562553406,
      "learning_rate": 0.0001768224655824504,
      "loss": 4.6094,
      "step": 33060
    },
    {
      "epoch": 0.6772059877541826,
      "grad_norm": 0.2915450930595398,
      "learning_rate": 0.00017663928201401598,
      "loss": 4.6345,
      "step": 33070
    },
    {
      "epoch": 0.6774107673089919,
      "grad_norm": 0.38605785369873047,
      "learning_rate": 0.00017645611831193332,
      "loss": 4.6323,
      "step": 33080
    },
    {
      "epoch": 0.6776155468638011,
      "grad_norm": 0.30939891934394836,
      "learning_rate": 0.00017627297463196793,
      "loss": 4.6566,
      "step": 33090
    },
    {
      "epoch": 0.6778203264186103,
      "grad_norm": 0.3041756749153137,
      "learning_rate": 0.00017608985112986823,
      "loss": 4.625,
      "step": 33100
    },
    {
      "epoch": 0.6780251059734196,
      "grad_norm": 0.31380847096443176,
      "learning_rate": 0.00017590674796136576,
      "loss": 4.6472,
      "step": 33110
    },
    {
      "epoch": 0.6782298855282288,
      "grad_norm": 0.309291809797287,
      "learning_rate": 0.0001757236652821743,
      "loss": 4.6401,
      "step": 33120
    },
    {
      "epoch": 0.6784346650830381,
      "grad_norm": 0.30570313334465027,
      "learning_rate": 0.00017554060324799059,
      "loss": 4.6379,
      "step": 33130
    },
    {
      "epoch": 0.6786394446378473,
      "grad_norm": 0.6258738040924072,
      "learning_rate": 0.0001753575620144938,
      "loss": 4.6244,
      "step": 33140
    },
    {
      "epoch": 0.6788442241926566,
      "grad_norm": 0.2956477105617523,
      "learning_rate": 0.00017517454173734511,
      "loss": 4.604,
      "step": 33150
    },
    {
      "epoch": 0.6790490037474659,
      "grad_norm": 0.28767263889312744,
      "learning_rate": 0.0001749915425721882,
      "loss": 4.6152,
      "step": 33160
    },
    {
      "epoch": 0.6792537833022751,
      "grad_norm": 0.3020152151584625,
      "learning_rate": 0.00017480856467464863,
      "loss": 4.625,
      "step": 33170
    },
    {
      "epoch": 0.6794585628570844,
      "grad_norm": 0.3043646216392517,
      "learning_rate": 0.00017462560820033404,
      "loss": 4.6266,
      "step": 33180
    },
    {
      "epoch": 0.6796633424118936,
      "grad_norm": 0.32564735412597656,
      "learning_rate": 0.00017444267330483357,
      "loss": 4.6177,
      "step": 33190
    },
    {
      "epoch": 0.6798681219667029,
      "grad_norm": 0.3014850914478302,
      "learning_rate": 0.00017425976014371814,
      "loss": 4.6599,
      "step": 33200
    },
    {
      "epoch": 0.6800729015215121,
      "grad_norm": 0.3051404654979706,
      "learning_rate": 0.00017407686887254033,
      "loss": 4.6245,
      "step": 33210
    },
    {
      "epoch": 0.6802776810763214,
      "grad_norm": 0.2853024899959564,
      "learning_rate": 0.00017389399964683392,
      "loss": 4.6068,
      "step": 33220
    },
    {
      "epoch": 0.6804824606311306,
      "grad_norm": 0.28191468119621277,
      "learning_rate": 0.0001737111526221141,
      "loss": 4.616,
      "step": 33230
    },
    {
      "epoch": 0.6806872401859398,
      "grad_norm": 0.2870134115219116,
      "learning_rate": 0.00017352832795387694,
      "loss": 4.6341,
      "step": 33240
    },
    {
      "epoch": 0.6808920197407491,
      "grad_norm": 0.3020907938480377,
      "learning_rate": 0.0001733455257975998,
      "loss": 4.5707,
      "step": 33250
    },
    {
      "epoch": 0.6810967992955583,
      "grad_norm": 0.31421855092048645,
      "learning_rate": 0.00017316274630874064,
      "loss": 4.6089,
      "step": 33260
    },
    {
      "epoch": 0.6813015788503676,
      "grad_norm": 0.30290868878364563,
      "learning_rate": 0.0001729799896427382,
      "loss": 4.6041,
      "step": 33270
    },
    {
      "epoch": 0.6815063584051768,
      "grad_norm": 0.28204119205474854,
      "learning_rate": 0.00017279725595501202,
      "loss": 4.6333,
      "step": 33280
    },
    {
      "epoch": 0.6817111379599861,
      "grad_norm": 0.2897128462791443,
      "learning_rate": 0.00017261454540096178,
      "loss": 4.6101,
      "step": 33290
    },
    {
      "epoch": 0.6819159175147953,
      "grad_norm": 0.3139142096042633,
      "learning_rate": 0.0001724318581359677,
      "loss": 4.6587,
      "step": 33300
    },
    {
      "epoch": 0.6821206970696045,
      "grad_norm": 0.2964023947715759,
      "learning_rate": 0.00017224919431539014,
      "loss": 4.6474,
      "step": 33310
    },
    {
      "epoch": 0.6823254766244138,
      "grad_norm": 0.31991204619407654,
      "learning_rate": 0.0001720665540945696,
      "loss": 4.6123,
      "step": 33320
    },
    {
      "epoch": 0.682530256179223,
      "grad_norm": 0.30030396580696106,
      "learning_rate": 0.0001718839376288263,
      "loss": 4.6197,
      "step": 33330
    },
    {
      "epoch": 0.6827350357340323,
      "grad_norm": 0.2920781672000885,
      "learning_rate": 0.0001717013450734604,
      "loss": 4.6595,
      "step": 33340
    },
    {
      "epoch": 0.6829398152888415,
      "grad_norm": 0.2850472927093506,
      "learning_rate": 0.00017151877658375177,
      "loss": 4.6699,
      "step": 33350
    },
    {
      "epoch": 0.6831445948436509,
      "grad_norm": 0.30785560607910156,
      "learning_rate": 0.0001713362323149597,
      "loss": 4.6184,
      "step": 33360
    },
    {
      "epoch": 0.6833493743984601,
      "grad_norm": 0.2960036098957062,
      "learning_rate": 0.00017115371242232302,
      "loss": 4.6207,
      "step": 33370
    },
    {
      "epoch": 0.6835541539532694,
      "grad_norm": 0.32618141174316406,
      "learning_rate": 0.00017097121706105965,
      "loss": 4.6288,
      "step": 33380
    },
    {
      "epoch": 0.6837589335080786,
      "grad_norm": 0.3154570460319519,
      "learning_rate": 0.00017078874638636692,
      "loss": 4.5996,
      "step": 33390
    },
    {
      "epoch": 0.6839637130628878,
      "grad_norm": 0.29100146889686584,
      "learning_rate": 0.00017060630055342076,
      "loss": 4.6553,
      "step": 33400
    },
    {
      "epoch": 0.6841684926176971,
      "grad_norm": 0.29402703046798706,
      "learning_rate": 0.00017042387971737632,
      "loss": 4.6209,
      "step": 33410
    },
    {
      "epoch": 0.6843732721725063,
      "grad_norm": 0.2883686423301697,
      "learning_rate": 0.00017024148403336738,
      "loss": 4.6541,
      "step": 33420
    },
    {
      "epoch": 0.6845780517273156,
      "grad_norm": 0.3109899163246155,
      "learning_rate": 0.00017005911365650643,
      "loss": 4.6319,
      "step": 33430
    },
    {
      "epoch": 0.6847828312821248,
      "grad_norm": 0.28234413266181946,
      "learning_rate": 0.00016987676874188415,
      "loss": 4.6673,
      "step": 33440
    },
    {
      "epoch": 0.684987610836934,
      "grad_norm": 0.32163006067276,
      "learning_rate": 0.00016969444944456985,
      "loss": 4.5921,
      "step": 33450
    },
    {
      "epoch": 0.6851923903917433,
      "grad_norm": 0.28209173679351807,
      "learning_rate": 0.00016951215591961104,
      "loss": 4.5846,
      "step": 33460
    },
    {
      "epoch": 0.6853971699465525,
      "grad_norm": 0.2900984585285187,
      "learning_rate": 0.00016932988832203306,
      "loss": 4.6055,
      "step": 33470
    },
    {
      "epoch": 0.6856019495013618,
      "grad_norm": 0.32391712069511414,
      "learning_rate": 0.00016914764680683953,
      "loss": 4.63,
      "step": 33480
    },
    {
      "epoch": 0.685806729056171,
      "grad_norm": 0.28276559710502625,
      "learning_rate": 0.0001689654315290116,
      "loss": 4.6479,
      "step": 33490
    },
    {
      "epoch": 0.6860115086109803,
      "grad_norm": 0.31503188610076904,
      "learning_rate": 0.00016878324264350833,
      "loss": 4.6333,
      "step": 33500
    },
    {
      "epoch": 0.6862162881657895,
      "grad_norm": 0.2737639248371124,
      "learning_rate": 0.00016860108030526623,
      "loss": 4.6192,
      "step": 33510
    },
    {
      "epoch": 0.6864210677205987,
      "grad_norm": 0.3093930184841156,
      "learning_rate": 0.00016841894466919924,
      "loss": 4.6159,
      "step": 33520
    },
    {
      "epoch": 0.686625847275408,
      "grad_norm": 0.2998504936695099,
      "learning_rate": 0.00016823683589019869,
      "loss": 4.6206,
      "step": 33530
    },
    {
      "epoch": 0.6868306268302172,
      "grad_norm": 0.30359306931495667,
      "learning_rate": 0.00016805475412313278,
      "loss": 4.629,
      "step": 33540
    },
    {
      "epoch": 0.6870354063850265,
      "grad_norm": 0.28318941593170166,
      "learning_rate": 0.00016787269952284704,
      "loss": 4.5992,
      "step": 33550
    },
    {
      "epoch": 0.6872401859398357,
      "grad_norm": 0.29930534958839417,
      "learning_rate": 0.00016769067224416386,
      "loss": 4.626,
      "step": 33560
    },
    {
      "epoch": 0.6874449654946451,
      "grad_norm": 0.28812313079833984,
      "learning_rate": 0.0001675086724418823,
      "loss": 4.6257,
      "step": 33570
    },
    {
      "epoch": 0.6876497450494543,
      "grad_norm": 0.2946549952030182,
      "learning_rate": 0.00016732670027077803,
      "loss": 4.6182,
      "step": 33580
    },
    {
      "epoch": 0.6878545246042636,
      "grad_norm": 0.28934773802757263,
      "learning_rate": 0.00016714475588560333,
      "loss": 4.6001,
      "step": 33590
    },
    {
      "epoch": 0.6880593041590728,
      "grad_norm": 0.2975788712501526,
      "learning_rate": 0.00016696283944108684,
      "loss": 4.6209,
      "step": 33600
    },
    {
      "epoch": 0.688264083713882,
      "grad_norm": 0.30068328976631165,
      "learning_rate": 0.0001667809510919333,
      "loss": 4.6335,
      "step": 33610
    },
    {
      "epoch": 0.6884688632686913,
      "grad_norm": 0.2877258360385895,
      "learning_rate": 0.0001665990909928237,
      "loss": 4.6544,
      "step": 33620
    },
    {
      "epoch": 0.6886736428235005,
      "grad_norm": 0.2939065098762512,
      "learning_rate": 0.00016641725929841494,
      "loss": 4.5978,
      "step": 33630
    },
    {
      "epoch": 0.6888784223783098,
      "grad_norm": 0.2837385833263397,
      "learning_rate": 0.0001662354561633398,
      "loss": 4.6174,
      "step": 33640
    },
    {
      "epoch": 0.689083201933119,
      "grad_norm": 0.31246641278266907,
      "learning_rate": 0.00016605368174220676,
      "loss": 4.6092,
      "step": 33650
    },
    {
      "epoch": 0.6892879814879282,
      "grad_norm": 0.33488237857818604,
      "learning_rate": 0.0001658719361895999,
      "loss": 4.6392,
      "step": 33660
    },
    {
      "epoch": 0.6894927610427375,
      "grad_norm": 0.30829524993896484,
      "learning_rate": 0.00016569021966007875,
      "loss": 4.6542,
      "step": 33670
    },
    {
      "epoch": 0.6896975405975467,
      "grad_norm": 0.3291963040828705,
      "learning_rate": 0.00016550853230817803,
      "loss": 4.636,
      "step": 33680
    },
    {
      "epoch": 0.689902320152356,
      "grad_norm": 0.29185518622398376,
      "learning_rate": 0.00016532687428840782,
      "loss": 4.6201,
      "step": 33690
    },
    {
      "epoch": 0.6901070997071652,
      "grad_norm": 0.27664265036582947,
      "learning_rate": 0.00016514524575525315,
      "loss": 4.6101,
      "step": 33700
    },
    {
      "epoch": 0.6903118792619745,
      "grad_norm": 0.2888222932815552,
      "learning_rate": 0.00016496364686317412,
      "loss": 4.6383,
      "step": 33710
    },
    {
      "epoch": 0.6905166588167837,
      "grad_norm": 0.29128146171569824,
      "learning_rate": 0.00016478207776660537,
      "loss": 4.6394,
      "step": 33720
    },
    {
      "epoch": 0.6907214383715929,
      "grad_norm": 0.299492746591568,
      "learning_rate": 0.0001646005386199564,
      "loss": 4.5757,
      "step": 33730
    },
    {
      "epoch": 0.6909262179264022,
      "grad_norm": 0.29915815591812134,
      "learning_rate": 0.00016441902957761131,
      "loss": 4.6178,
      "step": 33740
    },
    {
      "epoch": 0.6911309974812114,
      "grad_norm": 0.35317280888557434,
      "learning_rate": 0.0001642375507939283,
      "loss": 4.6439,
      "step": 33750
    },
    {
      "epoch": 0.6913357770360207,
      "grad_norm": 0.2961081266403198,
      "learning_rate": 0.00016405610242324004,
      "loss": 4.6273,
      "step": 33760
    },
    {
      "epoch": 0.69154055659083,
      "grad_norm": 0.2951876223087311,
      "learning_rate": 0.00016387468461985343,
      "loss": 4.5987,
      "step": 33770
    },
    {
      "epoch": 0.6917453361456393,
      "grad_norm": 0.33545276522636414,
      "learning_rate": 0.00016369329753804915,
      "loss": 4.6142,
      "step": 33780
    },
    {
      "epoch": 0.6919501157004485,
      "grad_norm": 0.45325595140457153,
      "learning_rate": 0.00016351194133208184,
      "loss": 4.6249,
      "step": 33790
    },
    {
      "epoch": 0.6921548952552578,
      "grad_norm": 0.29584935307502747,
      "learning_rate": 0.00016333061615617998,
      "loss": 4.617,
      "step": 33800
    },
    {
      "epoch": 0.692359674810067,
      "grad_norm": 0.28129342198371887,
      "learning_rate": 0.00016314932216454558,
      "loss": 4.6244,
      "step": 33810
    },
    {
      "epoch": 0.6925644543648762,
      "grad_norm": 0.2985042631626129,
      "learning_rate": 0.00016296805951135409,
      "loss": 4.6105,
      "step": 33820
    },
    {
      "epoch": 0.6927692339196855,
      "grad_norm": 0.2735280692577362,
      "learning_rate": 0.00016278682835075428,
      "loss": 4.6201,
      "step": 33830
    },
    {
      "epoch": 0.6929740134744947,
      "grad_norm": 0.3145197629928589,
      "learning_rate": 0.0001626056288368683,
      "loss": 4.6109,
      "step": 33840
    },
    {
      "epoch": 0.693178793029304,
      "grad_norm": 0.2885589599609375,
      "learning_rate": 0.00016242446112379132,
      "loss": 4.6203,
      "step": 33850
    },
    {
      "epoch": 0.6933835725841132,
      "grad_norm": 0.3050481975078583,
      "learning_rate": 0.00016224332536559144,
      "loss": 4.606,
      "step": 33860
    },
    {
      "epoch": 0.6935883521389224,
      "grad_norm": 0.3099835515022278,
      "learning_rate": 0.00016206222171630952,
      "loss": 4.6058,
      "step": 33870
    },
    {
      "epoch": 0.6937931316937317,
      "grad_norm": 0.2893453538417816,
      "learning_rate": 0.0001618811503299593,
      "loss": 4.6087,
      "step": 33880
    },
    {
      "epoch": 0.6939979112485409,
      "grad_norm": 0.2874526381492615,
      "learning_rate": 0.00016170011136052678,
      "loss": 4.6221,
      "step": 33890
    },
    {
      "epoch": 0.6942026908033502,
      "grad_norm": 0.28769010305404663,
      "learning_rate": 0.00016151910496197072,
      "loss": 4.6597,
      "step": 33900
    },
    {
      "epoch": 0.6944074703581594,
      "grad_norm": 0.3656673729419708,
      "learning_rate": 0.000161338131288222,
      "loss": 4.6209,
      "step": 33910
    },
    {
      "epoch": 0.6946122499129687,
      "grad_norm": 0.29951661825180054,
      "learning_rate": 0.00016115719049318372,
      "loss": 4.6536,
      "step": 33920
    },
    {
      "epoch": 0.6948170294677779,
      "grad_norm": 0.2999086081981659,
      "learning_rate": 0.00016097628273073096,
      "loss": 4.6137,
      "step": 33930
    },
    {
      "epoch": 0.6950218090225871,
      "grad_norm": 0.32161378860473633,
      "learning_rate": 0.0001607954081547108,
      "loss": 4.6369,
      "step": 33940
    },
    {
      "epoch": 0.6952265885773964,
      "grad_norm": 0.29670730233192444,
      "learning_rate": 0.00016061456691894213,
      "loss": 4.625,
      "step": 33950
    },
    {
      "epoch": 0.6954313681322056,
      "grad_norm": 0.32408106327056885,
      "learning_rate": 0.00016043375917721531,
      "loss": 4.6082,
      "step": 33960
    },
    {
      "epoch": 0.695636147687015,
      "grad_norm": 0.30162012577056885,
      "learning_rate": 0.0001602529850832923,
      "loss": 4.6108,
      "step": 33970
    },
    {
      "epoch": 0.6958409272418242,
      "grad_norm": 0.2911863327026367,
      "learning_rate": 0.0001600722447909065,
      "loss": 4.6358,
      "step": 33980
    },
    {
      "epoch": 0.6960457067966335,
      "grad_norm": 0.29991620779037476,
      "learning_rate": 0.0001598915384537625,
      "loss": 4.6072,
      "step": 33990
    },
    {
      "epoch": 0.6962504863514427,
      "grad_norm": 0.30318012833595276,
      "learning_rate": 0.00015971086622553615,
      "loss": 4.587,
      "step": 34000
    },
    {
      "epoch": 0.6962504863514427,
      "eval_loss": 4.624014377593994,
      "eval_runtime": 4.3958,
      "eval_samples_per_second": 265.255,
      "eval_steps_per_second": 33.214,
      "step": 34000
    },
    {
      "epoch": 0.696455265906252,
      "grad_norm": 0.2930164039134979,
      "learning_rate": 0.00015953022825987401,
      "loss": 4.6127,
      "step": 34010
    },
    {
      "epoch": 0.6966600454610612,
      "grad_norm": 0.31689298152923584,
      "learning_rate": 0.0001593496247103939,
      "loss": 4.604,
      "step": 34020
    },
    {
      "epoch": 0.6968648250158704,
      "grad_norm": 0.30203554034233093,
      "learning_rate": 0.00015916905573068395,
      "loss": 4.6311,
      "step": 34030
    },
    {
      "epoch": 0.6970696045706797,
      "grad_norm": 0.33920392394065857,
      "learning_rate": 0.00015898852147430313,
      "loss": 4.6343,
      "step": 34040
    },
    {
      "epoch": 0.6972743841254889,
      "grad_norm": 0.29707950353622437,
      "learning_rate": 0.00015880802209478092,
      "loss": 4.6333,
      "step": 34050
    },
    {
      "epoch": 0.6974791636802982,
      "grad_norm": 0.29657280445098877,
      "learning_rate": 0.00015862755774561697,
      "loss": 4.6022,
      "step": 34060
    },
    {
      "epoch": 0.6976839432351074,
      "grad_norm": 0.297585129737854,
      "learning_rate": 0.00015844712858028128,
      "loss": 4.5858,
      "step": 34070
    },
    {
      "epoch": 0.6978887227899166,
      "grad_norm": 0.30895403027534485,
      "learning_rate": 0.00015826673475221386,
      "loss": 4.6116,
      "step": 34080
    },
    {
      "epoch": 0.6980935023447259,
      "grad_norm": 0.29268407821655273,
      "learning_rate": 0.00015808637641482478,
      "loss": 4.5968,
      "step": 34090
    },
    {
      "epoch": 0.6982982818995351,
      "grad_norm": 0.2873637080192566,
      "learning_rate": 0.0001579060537214937,
      "loss": 4.6052,
      "step": 34100
    },
    {
      "epoch": 0.6985030614543444,
      "grad_norm": 0.2870800793170929,
      "learning_rate": 0.0001577257668255702,
      "loss": 4.642,
      "step": 34110
    },
    {
      "epoch": 0.6987078410091536,
      "grad_norm": 0.29790377616882324,
      "learning_rate": 0.0001575455158803732,
      "loss": 4.6406,
      "step": 34120
    },
    {
      "epoch": 0.6989126205639629,
      "grad_norm": 0.2890074551105499,
      "learning_rate": 0.0001573653010391913,
      "loss": 4.6389,
      "step": 34130
    },
    {
      "epoch": 0.6991174001187721,
      "grad_norm": 0.30178284645080566,
      "learning_rate": 0.0001571851224552822,
      "loss": 4.616,
      "step": 34140
    },
    {
      "epoch": 0.6993221796735813,
      "grad_norm": 0.32394957542419434,
      "learning_rate": 0.00015700498028187285,
      "loss": 4.6141,
      "step": 34150
    },
    {
      "epoch": 0.6995269592283906,
      "grad_norm": 0.32160764932632446,
      "learning_rate": 0.0001568248746721593,
      "loss": 4.6395,
      "step": 34160
    },
    {
      "epoch": 0.6997317387831998,
      "grad_norm": 0.3074822723865509,
      "learning_rate": 0.00015664480577930623,
      "loss": 4.6031,
      "step": 34170
    },
    {
      "epoch": 0.6999365183380092,
      "grad_norm": 0.3199803829193115,
      "learning_rate": 0.00015646477375644738,
      "loss": 4.6103,
      "step": 34180
    },
    {
      "epoch": 0.7001412978928184,
      "grad_norm": 0.30673491954803467,
      "learning_rate": 0.0001562847787566851,
      "loss": 4.5934,
      "step": 34190
    },
    {
      "epoch": 0.7003460774476277,
      "grad_norm": 0.6456322073936462,
      "learning_rate": 0.00015610482093309015,
      "loss": 4.6557,
      "step": 34200
    },
    {
      "epoch": 0.7005508570024369,
      "grad_norm": 0.3115524351596832,
      "learning_rate": 0.0001559249004387017,
      "loss": 4.6471,
      "step": 34210
    },
    {
      "epoch": 0.7007556365572462,
      "grad_norm": 0.32813143730163574,
      "learning_rate": 0.00015574501742652726,
      "loss": 4.611,
      "step": 34220
    },
    {
      "epoch": 0.7009604161120554,
      "grad_norm": 0.33972465991973877,
      "learning_rate": 0.0001555651720495424,
      "loss": 4.6277,
      "step": 34230
    },
    {
      "epoch": 0.7011651956668646,
      "grad_norm": 0.3158153295516968,
      "learning_rate": 0.0001553853644606906,
      "loss": 4.6047,
      "step": 34240
    },
    {
      "epoch": 0.7013699752216739,
      "grad_norm": 0.2914464771747589,
      "learning_rate": 0.00015520559481288336,
      "loss": 4.6304,
      "step": 34250
    },
    {
      "epoch": 0.7015747547764831,
      "grad_norm": 0.29459428787231445,
      "learning_rate": 0.00015502586325899978,
      "loss": 4.6396,
      "step": 34260
    },
    {
      "epoch": 0.7017795343312924,
      "grad_norm": 0.298574835062027,
      "learning_rate": 0.0001548461699518867,
      "loss": 4.6511,
      "step": 34270
    },
    {
      "epoch": 0.7019843138861016,
      "grad_norm": 0.3120969831943512,
      "learning_rate": 0.00015466651504435834,
      "loss": 4.6389,
      "step": 34280
    },
    {
      "epoch": 0.7021890934409108,
      "grad_norm": 0.30967774987220764,
      "learning_rate": 0.00015448689868919631,
      "loss": 4.6313,
      "step": 34290
    },
    {
      "epoch": 0.7023938729957201,
      "grad_norm": 0.2917247414588928,
      "learning_rate": 0.00015430732103914943,
      "loss": 4.6116,
      "step": 34300
    },
    {
      "epoch": 0.7025986525505293,
      "grad_norm": 0.287887305021286,
      "learning_rate": 0.00015412778224693348,
      "loss": 4.5931,
      "step": 34310
    },
    {
      "epoch": 0.7028034321053386,
      "grad_norm": 0.2971671223640442,
      "learning_rate": 0.00015394828246523135,
      "loss": 4.6077,
      "step": 34320
    },
    {
      "epoch": 0.7030082116601478,
      "grad_norm": 0.30850762128829956,
      "learning_rate": 0.00015376882184669273,
      "loss": 4.6117,
      "step": 34330
    },
    {
      "epoch": 0.7032129912149571,
      "grad_norm": 0.28424036502838135,
      "learning_rate": 0.000153589400543934,
      "loss": 4.6151,
      "step": 34340
    },
    {
      "epoch": 0.7034177707697663,
      "grad_norm": 0.3723028004169464,
      "learning_rate": 0.00015341001870953805,
      "loss": 4.6219,
      "step": 34350
    },
    {
      "epoch": 0.7036225503245755,
      "grad_norm": 0.29439976811408997,
      "learning_rate": 0.00015323067649605424,
      "loss": 4.6086,
      "step": 34360
    },
    {
      "epoch": 0.7038273298793848,
      "grad_norm": 0.3045384883880615,
      "learning_rate": 0.00015305137405599822,
      "loss": 4.601,
      "step": 34370
    },
    {
      "epoch": 0.7040321094341941,
      "grad_norm": 0.3009319007396698,
      "learning_rate": 0.00015287211154185198,
      "loss": 4.6008,
      "step": 34380
    },
    {
      "epoch": 0.7042368889890034,
      "grad_norm": 0.30488184094429016,
      "learning_rate": 0.00015269288910606322,
      "loss": 4.6439,
      "step": 34390
    },
    {
      "epoch": 0.7044416685438126,
      "grad_norm": 0.31904223561286926,
      "learning_rate": 0.0001525137069010458,
      "loss": 4.6291,
      "step": 34400
    },
    {
      "epoch": 0.7046464480986219,
      "grad_norm": 0.3596601188182831,
      "learning_rate": 0.0001523345650791793,
      "loss": 4.6407,
      "step": 34410
    },
    {
      "epoch": 0.7048512276534311,
      "grad_norm": 0.29149162769317627,
      "learning_rate": 0.00015215546379280902,
      "loss": 4.6207,
      "step": 34420
    },
    {
      "epoch": 0.7050560072082404,
      "grad_norm": 0.3004820644855499,
      "learning_rate": 0.00015197640319424577,
      "loss": 4.6233,
      "step": 34430
    },
    {
      "epoch": 0.7052607867630496,
      "grad_norm": 0.3024444878101349,
      "learning_rate": 0.00015179738343576565,
      "loss": 4.5967,
      "step": 34440
    },
    {
      "epoch": 0.7054655663178588,
      "grad_norm": 0.3134540319442749,
      "learning_rate": 0.00015161840466961017,
      "loss": 4.617,
      "step": 34450
    },
    {
      "epoch": 0.7056703458726681,
      "grad_norm": 0.2759465277194977,
      "learning_rate": 0.00015143946704798583,
      "loss": 4.6282,
      "step": 34460
    },
    {
      "epoch": 0.7058751254274773,
      "grad_norm": 0.2962445318698883,
      "learning_rate": 0.00015126057072306425,
      "loss": 4.5925,
      "step": 34470
    },
    {
      "epoch": 0.7060799049822866,
      "grad_norm": 0.2982873022556305,
      "learning_rate": 0.00015108171584698202,
      "loss": 4.6238,
      "step": 34480
    },
    {
      "epoch": 0.7062846845370958,
      "grad_norm": 0.29654350876808167,
      "learning_rate": 0.0001509029025718402,
      "loss": 4.5964,
      "step": 34490
    },
    {
      "epoch": 0.706489464091905,
      "grad_norm": 0.31302574276924133,
      "learning_rate": 0.0001507241310497047,
      "loss": 4.6509,
      "step": 34500
    },
    {
      "epoch": 0.7066942436467143,
      "grad_norm": 0.9012963175773621,
      "learning_rate": 0.00015054540143260588,
      "loss": 4.5972,
      "step": 34510
    },
    {
      "epoch": 0.7068990232015235,
      "grad_norm": 0.3148398995399475,
      "learning_rate": 0.0001503667138725386,
      "loss": 4.6054,
      "step": 34520
    },
    {
      "epoch": 0.7071038027563328,
      "grad_norm": 0.29970496892929077,
      "learning_rate": 0.00015018806852146147,
      "loss": 4.6303,
      "step": 34530
    },
    {
      "epoch": 0.707308582311142,
      "grad_norm": 0.2997286915779114,
      "learning_rate": 0.0001500094655312978,
      "loss": 4.6397,
      "step": 34540
    },
    {
      "epoch": 0.7075133618659513,
      "grad_norm": 0.5666271448135376,
      "learning_rate": 0.00014983090505393443,
      "loss": 4.6496,
      "step": 34550
    },
    {
      "epoch": 0.7077181414207605,
      "grad_norm": 0.31896817684173584,
      "learning_rate": 0.00014965238724122236,
      "loss": 4.6252,
      "step": 34560
    },
    {
      "epoch": 0.7079229209755697,
      "grad_norm": 0.3208785653114319,
      "learning_rate": 0.00014947391224497616,
      "loss": 4.6231,
      "step": 34570
    },
    {
      "epoch": 0.7081277005303791,
      "grad_norm": 0.2764456868171692,
      "learning_rate": 0.00014929548021697388,
      "loss": 4.6422,
      "step": 34580
    },
    {
      "epoch": 0.7083324800851883,
      "grad_norm": 0.47031036019325256,
      "learning_rate": 0.00014911709130895737,
      "loss": 4.5437,
      "step": 34590
    },
    {
      "epoch": 0.7085372596399976,
      "grad_norm": 0.3066767156124115,
      "learning_rate": 0.00014893874567263132,
      "loss": 4.6301,
      "step": 34600
    },
    {
      "epoch": 0.7087420391948068,
      "grad_norm": 0.2902284860610962,
      "learning_rate": 0.00014876044345966407,
      "loss": 4.5781,
      "step": 34610
    },
    {
      "epoch": 0.7089468187496161,
      "grad_norm": 0.29947999119758606,
      "learning_rate": 0.00014858218482168679,
      "loss": 4.6241,
      "step": 34620
    },
    {
      "epoch": 0.7091515983044253,
      "grad_norm": 0.28963664174079895,
      "learning_rate": 0.0001484039699102937,
      "loss": 4.6062,
      "step": 34630
    },
    {
      "epoch": 0.7093563778592346,
      "grad_norm": 0.2964884340763092,
      "learning_rate": 0.00014822579887704176,
      "loss": 4.6447,
      "step": 34640
    },
    {
      "epoch": 0.7095611574140438,
      "grad_norm": 0.2854631543159485,
      "learning_rate": 0.00014804767187345062,
      "loss": 4.632,
      "step": 34650
    },
    {
      "epoch": 0.709765936968853,
      "grad_norm": 0.30145594477653503,
      "learning_rate": 0.00014786958905100265,
      "loss": 4.6122,
      "step": 34660
    },
    {
      "epoch": 0.7099707165236623,
      "grad_norm": 0.30469653010368347,
      "learning_rate": 0.0001476915505611423,
      "loss": 4.6022,
      "step": 34670
    },
    {
      "epoch": 0.7101754960784715,
      "grad_norm": 0.29701927304267883,
      "learning_rate": 0.00014751355655527671,
      "loss": 4.5936,
      "step": 34680
    },
    {
      "epoch": 0.7103802756332808,
      "grad_norm": 0.27827855944633484,
      "learning_rate": 0.00014733560718477486,
      "loss": 4.6606,
      "step": 34690
    },
    {
      "epoch": 0.71058505518809,
      "grad_norm": 0.2819177210330963,
      "learning_rate": 0.00014715770260096802,
      "loss": 4.6358,
      "step": 34700
    },
    {
      "epoch": 0.7107898347428993,
      "grad_norm": 0.3499496579170227,
      "learning_rate": 0.0001469798429551492,
      "loss": 4.6445,
      "step": 34710
    },
    {
      "epoch": 0.7109946142977085,
      "grad_norm": 0.3143846094608307,
      "learning_rate": 0.00014680202839857336,
      "loss": 4.6115,
      "step": 34720
    },
    {
      "epoch": 0.7111993938525177,
      "grad_norm": 0.3078066408634186,
      "learning_rate": 0.00014662425908245705,
      "loss": 4.6021,
      "step": 34730
    },
    {
      "epoch": 0.711404173407327,
      "grad_norm": 0.2983221113681793,
      "learning_rate": 0.0001464465351579781,
      "loss": 4.6188,
      "step": 34740
    },
    {
      "epoch": 0.7116089529621362,
      "grad_norm": 0.3281814754009247,
      "learning_rate": 0.00014626885677627614,
      "loss": 4.5792,
      "step": 34750
    },
    {
      "epoch": 0.7118137325169455,
      "grad_norm": 0.3432719111442566,
      "learning_rate": 0.00014609122408845184,
      "loss": 4.6304,
      "step": 34760
    },
    {
      "epoch": 0.7120185120717547,
      "grad_norm": 0.28590285778045654,
      "learning_rate": 0.0001459136372455671,
      "loss": 4.5996,
      "step": 34770
    },
    {
      "epoch": 0.7122232916265641,
      "grad_norm": 0.3112219572067261,
      "learning_rate": 0.0001457360963986447,
      "loss": 4.6033,
      "step": 34780
    },
    {
      "epoch": 0.7124280711813733,
      "grad_norm": 0.30003437399864197,
      "learning_rate": 0.00014555860169866843,
      "loss": 4.6144,
      "step": 34790
    },
    {
      "epoch": 0.7126328507361825,
      "grad_norm": 0.2984277904033661,
      "learning_rate": 0.00014538115329658293,
      "loss": 4.6164,
      "step": 34800
    },
    {
      "epoch": 0.7128376302909918,
      "grad_norm": 0.3116559088230133,
      "learning_rate": 0.00014520375134329314,
      "loss": 4.6242,
      "step": 34810
    },
    {
      "epoch": 0.713042409845801,
      "grad_norm": 0.3221040666103363,
      "learning_rate": 0.0001450263959896648,
      "loss": 4.5388,
      "step": 34820
    },
    {
      "epoch": 0.7132471894006103,
      "grad_norm": 0.2988126873970032,
      "learning_rate": 0.0001448490873865239,
      "loss": 4.5904,
      "step": 34830
    },
    {
      "epoch": 0.7134519689554195,
      "grad_norm": 0.3226899206638336,
      "learning_rate": 0.00014467182568465666,
      "loss": 4.6182,
      "step": 34840
    },
    {
      "epoch": 0.7136567485102288,
      "grad_norm": 0.3137524425983429,
      "learning_rate": 0.00014449461103480945,
      "loss": 4.624,
      "step": 34850
    },
    {
      "epoch": 0.713861528065038,
      "grad_norm": 0.3023771643638611,
      "learning_rate": 0.0001443174435876887,
      "loss": 4.5885,
      "step": 34860
    },
    {
      "epoch": 0.7140663076198472,
      "grad_norm": 0.3101402819156647,
      "learning_rate": 0.00014414032349396058,
      "loss": 4.5458,
      "step": 34870
    },
    {
      "epoch": 0.7142710871746565,
      "grad_norm": 0.29617607593536377,
      "learning_rate": 0.00014396325090425103,
      "loss": 4.5918,
      "step": 34880
    },
    {
      "epoch": 0.7144758667294657,
      "grad_norm": 0.3024255931377411,
      "learning_rate": 0.00014378622596914552,
      "loss": 4.5991,
      "step": 34890
    },
    {
      "epoch": 0.714680646284275,
      "grad_norm": 0.3139723837375641,
      "learning_rate": 0.00014360924883918908,
      "loss": 4.6125,
      "step": 34900
    },
    {
      "epoch": 0.7148854258390842,
      "grad_norm": 0.3257853388786316,
      "learning_rate": 0.00014343231966488623,
      "loss": 4.6163,
      "step": 34910
    },
    {
      "epoch": 0.7150902053938935,
      "grad_norm": 0.30789676308631897,
      "learning_rate": 0.0001432554385967003,
      "loss": 4.623,
      "step": 34920
    },
    {
      "epoch": 0.7152949849487027,
      "grad_norm": 0.3053744435310364,
      "learning_rate": 0.00014307860578505413,
      "loss": 4.6242,
      "step": 34930
    },
    {
      "epoch": 0.7154997645035119,
      "grad_norm": 0.3180652856826782,
      "learning_rate": 0.00014290182138032936,
      "loss": 4.6371,
      "step": 34940
    },
    {
      "epoch": 0.7157045440583212,
      "grad_norm": 0.3113649785518646,
      "learning_rate": 0.00014272508553286634,
      "loss": 4.6102,
      "step": 34950
    },
    {
      "epoch": 0.7159093236131304,
      "grad_norm": 0.30352339148521423,
      "learning_rate": 0.00014254839839296428,
      "loss": 4.6238,
      "step": 34960
    },
    {
      "epoch": 0.7161141031679397,
      "grad_norm": 0.3152178227901459,
      "learning_rate": 0.00014237176011088097,
      "loss": 4.605,
      "step": 34970
    },
    {
      "epoch": 0.7163188827227489,
      "grad_norm": 0.2903915345668793,
      "learning_rate": 0.00014219517083683257,
      "loss": 4.5949,
      "step": 34980
    },
    {
      "epoch": 0.7165236622775583,
      "grad_norm": 0.30046382546424866,
      "learning_rate": 0.0001420186307209936,
      "loss": 4.6819,
      "step": 34990
    },
    {
      "epoch": 0.7167284418323675,
      "grad_norm": 0.5529968738555908,
      "learning_rate": 0.00014184213991349682,
      "loss": 4.5912,
      "step": 35000
    },
    {
      "epoch": 0.7169332213871767,
      "grad_norm": 0.3269819915294647,
      "learning_rate": 0.000141665698564433,
      "loss": 4.6077,
      "step": 35010
    },
    {
      "epoch": 0.717138000941986,
      "grad_norm": 0.2949140667915344,
      "learning_rate": 0.00014148930682385085,
      "loss": 4.5954,
      "step": 35020
    },
    {
      "epoch": 0.7173427804967952,
      "grad_norm": 0.30816182494163513,
      "learning_rate": 0.00014131296484175683,
      "loss": 4.667,
      "step": 35030
    },
    {
      "epoch": 0.7175475600516045,
      "grad_norm": 0.27781638503074646,
      "learning_rate": 0.00014113667276811524,
      "loss": 4.6247,
      "step": 35040
    },
    {
      "epoch": 0.7177523396064137,
      "grad_norm": 0.30939093232154846,
      "learning_rate": 0.00014096043075284784,
      "loss": 4.6372,
      "step": 35050
    },
    {
      "epoch": 0.717957119161223,
      "grad_norm": 0.2976728677749634,
      "learning_rate": 0.00014078423894583388,
      "loss": 4.5985,
      "step": 35060
    },
    {
      "epoch": 0.7181618987160322,
      "grad_norm": 0.30731210112571716,
      "learning_rate": 0.0001406080974969098,
      "loss": 4.6399,
      "step": 35070
    },
    {
      "epoch": 0.7183666782708414,
      "grad_norm": 0.3040066063404083,
      "learning_rate": 0.00014043200655586939,
      "loss": 4.6362,
      "step": 35080
    },
    {
      "epoch": 0.7185714578256507,
      "grad_norm": 0.3073359429836273,
      "learning_rate": 0.0001402559662724632,
      "loss": 4.624,
      "step": 35090
    },
    {
      "epoch": 0.7187762373804599,
      "grad_norm": 0.28751540184020996,
      "learning_rate": 0.00014007997679639899,
      "loss": 4.637,
      "step": 35100
    },
    {
      "epoch": 0.7189810169352692,
      "grad_norm": 0.3024277687072754,
      "learning_rate": 0.00013990403827734125,
      "loss": 4.6077,
      "step": 35110
    },
    {
      "epoch": 0.7191857964900784,
      "grad_norm": 0.32687389850616455,
      "learning_rate": 0.00013972815086491096,
      "loss": 4.5683,
      "step": 35120
    },
    {
      "epoch": 0.7193905760448877,
      "grad_norm": 0.2875973582267761,
      "learning_rate": 0.00013955231470868588,
      "loss": 4.6181,
      "step": 35130
    },
    {
      "epoch": 0.7195953555996969,
      "grad_norm": 0.307827889919281,
      "learning_rate": 0.0001393765299582,
      "loss": 4.6064,
      "step": 35140
    },
    {
      "epoch": 0.7198001351545061,
      "grad_norm": 0.2840040922164917,
      "learning_rate": 0.00013920079676294378,
      "loss": 4.6192,
      "step": 35150
    },
    {
      "epoch": 0.7200049147093154,
      "grad_norm": 0.42402467131614685,
      "learning_rate": 0.00013902511527236357,
      "loss": 4.5974,
      "step": 35160
    },
    {
      "epoch": 0.7202096942641246,
      "grad_norm": 0.34591928124427795,
      "learning_rate": 0.0001388494856358619,
      "loss": 4.6187,
      "step": 35170
    },
    {
      "epoch": 0.7204144738189339,
      "grad_norm": 0.32590940594673157,
      "learning_rate": 0.00013867390800279724,
      "loss": 4.5899,
      "step": 35180
    },
    {
      "epoch": 0.7206192533737432,
      "grad_norm": 0.28604620695114136,
      "learning_rate": 0.0001384983825224838,
      "loss": 4.5828,
      "step": 35190
    },
    {
      "epoch": 0.7208240329285525,
      "grad_norm": 0.3120392858982086,
      "learning_rate": 0.0001383229093441915,
      "loss": 4.6411,
      "step": 35200
    },
    {
      "epoch": 0.7210288124833617,
      "grad_norm": 0.3106038570404053,
      "learning_rate": 0.00013814748861714558,
      "loss": 4.5886,
      "step": 35210
    },
    {
      "epoch": 0.721233592038171,
      "grad_norm": 0.2892000675201416,
      "learning_rate": 0.00013797212049052697,
      "loss": 4.6308,
      "step": 35220
    },
    {
      "epoch": 0.7214383715929802,
      "grad_norm": 0.29461976885795593,
      "learning_rate": 0.00013779680511347153,
      "loss": 4.604,
      "step": 35230
    },
    {
      "epoch": 0.7216431511477894,
      "grad_norm": 0.31296923756599426,
      "learning_rate": 0.0001376215426350705,
      "loss": 4.6196,
      "step": 35240
    },
    {
      "epoch": 0.7218479307025987,
      "grad_norm": 0.28251272439956665,
      "learning_rate": 0.00013744633320437015,
      "loss": 4.6191,
      "step": 35250
    },
    {
      "epoch": 0.7220527102574079,
      "grad_norm": 0.3122941553592682,
      "learning_rate": 0.00013727117697037144,
      "loss": 4.6331,
      "step": 35260
    },
    {
      "epoch": 0.7222574898122172,
      "grad_norm": 0.32094359397888184,
      "learning_rate": 0.00013709607408203027,
      "loss": 4.6171,
      "step": 35270
    },
    {
      "epoch": 0.7224622693670264,
      "grad_norm": 0.2993016541004181,
      "learning_rate": 0.0001369210246882571,
      "loss": 4.6386,
      "step": 35280
    },
    {
      "epoch": 0.7226670489218356,
      "grad_norm": 0.30319884419441223,
      "learning_rate": 0.00013674602893791704,
      "loss": 4.5797,
      "step": 35290
    },
    {
      "epoch": 0.7228718284766449,
      "grad_norm": 0.48357534408569336,
      "learning_rate": 0.00013657108697982928,
      "loss": 4.5882,
      "step": 35300
    },
    {
      "epoch": 0.7230766080314541,
      "grad_norm": 0.3391391932964325,
      "learning_rate": 0.00013639619896276743,
      "loss": 4.6366,
      "step": 35310
    },
    {
      "epoch": 0.7232813875862634,
      "grad_norm": 0.32843708992004395,
      "learning_rate": 0.0001362213650354593,
      "loss": 4.5935,
      "step": 35320
    },
    {
      "epoch": 0.7234861671410726,
      "grad_norm": 0.2876911461353302,
      "learning_rate": 0.0001360465853465866,
      "loss": 4.6425,
      "step": 35330
    },
    {
      "epoch": 0.7236909466958819,
      "grad_norm": 0.2941429615020752,
      "learning_rate": 0.00013587186004478504,
      "loss": 4.619,
      "step": 35340
    },
    {
      "epoch": 0.7238957262506911,
      "grad_norm": 0.32524776458740234,
      "learning_rate": 0.00013569718927864386,
      "loss": 4.619,
      "step": 35350
    },
    {
      "epoch": 0.7241005058055003,
      "grad_norm": 0.3023153245449066,
      "learning_rate": 0.00013552257319670623,
      "loss": 4.6155,
      "step": 35360
    },
    {
      "epoch": 0.7243052853603096,
      "grad_norm": 0.30180665850639343,
      "learning_rate": 0.0001353480119474684,
      "loss": 4.6609,
      "step": 35370
    },
    {
      "epoch": 0.7245100649151188,
      "grad_norm": 0.3072219491004944,
      "learning_rate": 0.00013517350567938033,
      "loss": 4.5982,
      "step": 35380
    },
    {
      "epoch": 0.7247148444699282,
      "grad_norm": 0.3067971169948578,
      "learning_rate": 0.0001349990545408452,
      "loss": 4.5951,
      "step": 35390
    },
    {
      "epoch": 0.7249196240247374,
      "grad_norm": 0.31510570645332336,
      "learning_rate": 0.0001348246586802191,
      "loss": 4.617,
      "step": 35400
    },
    {
      "epoch": 0.7251244035795467,
      "grad_norm": 0.32711508870124817,
      "learning_rate": 0.00013465031824581123,
      "loss": 4.5909,
      "step": 35410
    },
    {
      "epoch": 0.7253291831343559,
      "grad_norm": 0.42030251026153564,
      "learning_rate": 0.00013447603338588378,
      "loss": 4.5875,
      "step": 35420
    },
    {
      "epoch": 0.7255339626891651,
      "grad_norm": 0.30846327543258667,
      "learning_rate": 0.00013430180424865156,
      "loss": 4.5955,
      "step": 35430
    },
    {
      "epoch": 0.7257387422439744,
      "grad_norm": 0.31136104464530945,
      "learning_rate": 0.0001341276309822818,
      "loss": 4.5894,
      "step": 35440
    },
    {
      "epoch": 0.7259435217987836,
      "grad_norm": 0.288819283246994,
      "learning_rate": 0.00013395351373489463,
      "loss": 4.6092,
      "step": 35450
    },
    {
      "epoch": 0.7261483013535929,
      "grad_norm": 0.30843454599380493,
      "learning_rate": 0.0001337794526545621,
      "loss": 4.5506,
      "step": 35460
    },
    {
      "epoch": 0.7263530809084021,
      "grad_norm": 0.32559868693351746,
      "learning_rate": 0.00013360544788930886,
      "loss": 4.6309,
      "step": 35470
    },
    {
      "epoch": 0.7265578604632114,
      "grad_norm": 0.34923264384269714,
      "learning_rate": 0.00013343149958711155,
      "loss": 4.6171,
      "step": 35480
    },
    {
      "epoch": 0.7267626400180206,
      "grad_norm": 0.296554833650589,
      "learning_rate": 0.00013325760789589865,
      "loss": 4.6257,
      "step": 35490
    },
    {
      "epoch": 0.7269674195728298,
      "grad_norm": 0.3782421350479126,
      "learning_rate": 0.00013308377296355074,
      "loss": 4.6031,
      "step": 35500
    },
    {
      "epoch": 0.7271721991276391,
      "grad_norm": 0.29011818766593933,
      "learning_rate": 0.00013290999493789985,
      "loss": 4.5951,
      "step": 35510
    },
    {
      "epoch": 0.7273769786824483,
      "grad_norm": 0.30199551582336426,
      "learning_rate": 0.0001327362739667298,
      "loss": 4.5943,
      "step": 35520
    },
    {
      "epoch": 0.7275817582372576,
      "grad_norm": 0.30701789259910583,
      "learning_rate": 0.00013256261019777593,
      "loss": 4.6092,
      "step": 35530
    },
    {
      "epoch": 0.7277865377920668,
      "grad_norm": 0.3072527050971985,
      "learning_rate": 0.00013238900377872487,
      "loss": 4.6298,
      "step": 35540
    },
    {
      "epoch": 0.727991317346876,
      "grad_norm": 0.3173460066318512,
      "learning_rate": 0.0001322154548572144,
      "loss": 4.6172,
      "step": 35550
    },
    {
      "epoch": 0.7281960969016853,
      "grad_norm": 0.3216269612312317,
      "learning_rate": 0.00013204196358083348,
      "loss": 4.6169,
      "step": 35560
    },
    {
      "epoch": 0.7284008764564945,
      "grad_norm": 0.30654504895210266,
      "learning_rate": 0.00013186853009712217,
      "loss": 4.6202,
      "step": 35570
    },
    {
      "epoch": 0.7286056560113038,
      "grad_norm": 0.31015145778656006,
      "learning_rate": 0.00013169515455357104,
      "loss": 4.5773,
      "step": 35580
    },
    {
      "epoch": 0.728810435566113,
      "grad_norm": 0.30494025349617004,
      "learning_rate": 0.00013152183709762175,
      "loss": 4.5958,
      "step": 35590
    },
    {
      "epoch": 0.7290152151209224,
      "grad_norm": 0.32751893997192383,
      "learning_rate": 0.00013134857787666632,
      "loss": 4.6148,
      "step": 35600
    },
    {
      "epoch": 0.7292199946757316,
      "grad_norm": 0.2925833761692047,
      "learning_rate": 0.0001311753770380474,
      "loss": 4.6189,
      "step": 35610
    },
    {
      "epoch": 0.7294247742305409,
      "grad_norm": 0.3233659565448761,
      "learning_rate": 0.00013100223472905793,
      "loss": 4.6007,
      "step": 35620
    },
    {
      "epoch": 0.7296295537853501,
      "grad_norm": 0.2875606417655945,
      "learning_rate": 0.00013082915109694106,
      "loss": 4.5747,
      "step": 35630
    },
    {
      "epoch": 0.7298343333401593,
      "grad_norm": 0.29883286356925964,
      "learning_rate": 0.00013065612628889012,
      "loss": 4.6044,
      "step": 35640
    },
    {
      "epoch": 0.7300391128949686,
      "grad_norm": 0.3009844720363617,
      "learning_rate": 0.0001304831604520482,
      "loss": 4.6102,
      "step": 35650
    },
    {
      "epoch": 0.7302438924497778,
      "grad_norm": 0.2949434518814087,
      "learning_rate": 0.00013031025373350845,
      "loss": 4.6147,
      "step": 35660
    },
    {
      "epoch": 0.7304486720045871,
      "grad_norm": 0.3077467679977417,
      "learning_rate": 0.00013013740628031368,
      "loss": 4.5675,
      "step": 35670
    },
    {
      "epoch": 0.7306534515593963,
      "grad_norm": 0.48693525791168213,
      "learning_rate": 0.0001299646182394564,
      "loss": 4.5524,
      "step": 35680
    },
    {
      "epoch": 0.7308582311142056,
      "grad_norm": 0.3349159061908722,
      "learning_rate": 0.00012979188975787833,
      "loss": 4.6117,
      "step": 35690
    },
    {
      "epoch": 0.7310630106690148,
      "grad_norm": 0.34192952513694763,
      "learning_rate": 0.00012961922098247083,
      "loss": 4.6365,
      "step": 35700
    },
    {
      "epoch": 0.731267790223824,
      "grad_norm": 0.29073071479797363,
      "learning_rate": 0.0001294466120600744,
      "loss": 4.6343,
      "step": 35710
    },
    {
      "epoch": 0.7314725697786333,
      "grad_norm": 0.32743415236473083,
      "learning_rate": 0.00012927406313747842,
      "loss": 4.5932,
      "step": 35720
    },
    {
      "epoch": 0.7316773493334425,
      "grad_norm": 0.29877787828445435,
      "learning_rate": 0.00012910157436142163,
      "loss": 4.6212,
      "step": 35730
    },
    {
      "epoch": 0.7318821288882518,
      "grad_norm": 0.3163726329803467,
      "learning_rate": 0.00012892914587859126,
      "loss": 4.597,
      "step": 35740
    },
    {
      "epoch": 0.732086908443061,
      "grad_norm": 0.3142485022544861,
      "learning_rate": 0.00012875677783562346,
      "loss": 4.5922,
      "step": 35750
    },
    {
      "epoch": 0.7322916879978703,
      "grad_norm": 0.2876586616039276,
      "learning_rate": 0.00012858447037910305,
      "loss": 4.6066,
      "step": 35760
    },
    {
      "epoch": 0.7324964675526795,
      "grad_norm": 0.308336079120636,
      "learning_rate": 0.00012841222365556312,
      "loss": 4.5955,
      "step": 35770
    },
    {
      "epoch": 0.7327012471074887,
      "grad_norm": 0.32495421171188354,
      "learning_rate": 0.00012824003781148523,
      "loss": 4.5893,
      "step": 35780
    },
    {
      "epoch": 0.732906026662298,
      "grad_norm": 0.29860562086105347,
      "learning_rate": 0.00012806791299329917,
      "loss": 4.5805,
      "step": 35790
    },
    {
      "epoch": 0.7331108062171073,
      "grad_norm": 0.3136485815048218,
      "learning_rate": 0.00012789584934738276,
      "loss": 4.6073,
      "step": 35800
    },
    {
      "epoch": 0.7333155857719166,
      "grad_norm": 0.29984745383262634,
      "learning_rate": 0.0001277238470200619,
      "loss": 4.621,
      "step": 35810
    },
    {
      "epoch": 0.7335203653267258,
      "grad_norm": 0.2900507152080536,
      "learning_rate": 0.0001275519061576103,
      "loss": 4.6068,
      "step": 35820
    },
    {
      "epoch": 0.7337251448815351,
      "grad_norm": 0.29403650760650635,
      "learning_rate": 0.0001273800269062493,
      "loss": 4.5958,
      "step": 35830
    },
    {
      "epoch": 0.7339299244363443,
      "grad_norm": 0.2939351201057434,
      "learning_rate": 0.00012720820941214804,
      "loss": 4.6091,
      "step": 35840
    },
    {
      "epoch": 0.7341347039911535,
      "grad_norm": 0.3304155170917511,
      "learning_rate": 0.0001270364538214231,
      "loss": 4.613,
      "step": 35850
    },
    {
      "epoch": 0.7343394835459628,
      "grad_norm": 0.5047217011451721,
      "learning_rate": 0.00012686476028013813,
      "loss": 4.6207,
      "step": 35860
    },
    {
      "epoch": 0.734544263100772,
      "grad_norm": 0.3045598566532135,
      "learning_rate": 0.00012669312893430444,
      "loss": 4.5792,
      "step": 35870
    },
    {
      "epoch": 0.7347490426555813,
      "grad_norm": 0.30130526423454285,
      "learning_rate": 0.00012652155992988015,
      "loss": 4.5705,
      "step": 35880
    },
    {
      "epoch": 0.7349538222103905,
      "grad_norm": 0.30585214495658875,
      "learning_rate": 0.00012635005341277044,
      "loss": 4.6215,
      "step": 35890
    },
    {
      "epoch": 0.7351586017651998,
      "grad_norm": 0.2994743287563324,
      "learning_rate": 0.00012617860952882746,
      "loss": 4.5876,
      "step": 35900
    },
    {
      "epoch": 0.735363381320009,
      "grad_norm": 0.3443034887313843,
      "learning_rate": 0.00012600722842384996,
      "loss": 4.6038,
      "step": 35910
    },
    {
      "epoch": 0.7355681608748182,
      "grad_norm": 0.30167892575263977,
      "learning_rate": 0.0001258359102435833,
      "loss": 4.635,
      "step": 35920
    },
    {
      "epoch": 0.7357729404296275,
      "grad_norm": 0.2863532304763794,
      "learning_rate": 0.00012566465513371946,
      "loss": 4.6039,
      "step": 35930
    },
    {
      "epoch": 0.7359777199844367,
      "grad_norm": 0.3017323911190033,
      "learning_rate": 0.00012549346323989657,
      "loss": 4.6143,
      "step": 35940
    },
    {
      "epoch": 0.736182499539246,
      "grad_norm": 0.29638591408729553,
      "learning_rate": 0.00012532233470769922,
      "loss": 4.628,
      "step": 35950
    },
    {
      "epoch": 0.7363872790940552,
      "grad_norm": 0.3080417811870575,
      "learning_rate": 0.000125151269682658,
      "loss": 4.6092,
      "step": 35960
    },
    {
      "epoch": 0.7365920586488645,
      "grad_norm": 0.3476355969905853,
      "learning_rate": 0.00012498026831024957,
      "loss": 4.5768,
      "step": 35970
    },
    {
      "epoch": 0.7367968382036737,
      "grad_norm": 0.3063572347164154,
      "learning_rate": 0.0001248093307358963,
      "loss": 4.6202,
      "step": 35980
    },
    {
      "epoch": 0.7370016177584829,
      "grad_norm": 0.30983617901802063,
      "learning_rate": 0.00012463845710496658,
      "loss": 4.6741,
      "step": 35990
    },
    {
      "epoch": 0.7372063973132923,
      "grad_norm": 0.3034866154193878,
      "learning_rate": 0.00012446764756277406,
      "loss": 4.5873,
      "step": 36000
    },
    {
      "epoch": 0.7372063973132923,
      "eval_loss": 4.60663366317749,
      "eval_runtime": 4.399,
      "eval_samples_per_second": 265.063,
      "eval_steps_per_second": 33.19,
      "step": 36000
    },
    {
      "epoch": 0.7374111768681015,
      "grad_norm": 0.28170254826545715,
      "learning_rate": 0.0001242969022545781,
      "loss": 4.6308,
      "step": 36010
    },
    {
      "epoch": 0.7376159564229108,
      "grad_norm": 0.3137228190898895,
      "learning_rate": 0.00012412622132558355,
      "loss": 4.5872,
      "step": 36020
    },
    {
      "epoch": 0.73782073597772,
      "grad_norm": 0.300894558429718,
      "learning_rate": 0.00012395560492094022,
      "loss": 4.6159,
      "step": 36030
    },
    {
      "epoch": 0.7380255155325293,
      "grad_norm": 0.3135361969470978,
      "learning_rate": 0.0001237850531857433,
      "loss": 4.5941,
      "step": 36040
    },
    {
      "epoch": 0.7382302950873385,
      "grad_norm": 0.2942076623439789,
      "learning_rate": 0.00012361456626503286,
      "loss": 4.6294,
      "step": 36050
    },
    {
      "epoch": 0.7384350746421477,
      "grad_norm": 0.3123762607574463,
      "learning_rate": 0.0001234441443037939,
      "loss": 4.5859,
      "step": 36060
    },
    {
      "epoch": 0.738639854196957,
      "grad_norm": 0.29585447907447815,
      "learning_rate": 0.00012327378744695613,
      "loss": 4.5828,
      "step": 36070
    },
    {
      "epoch": 0.7388446337517662,
      "grad_norm": 0.2900315523147583,
      "learning_rate": 0.00012310349583939388,
      "loss": 4.6177,
      "step": 36080
    },
    {
      "epoch": 0.7390494133065755,
      "grad_norm": 0.3082260489463806,
      "learning_rate": 0.00012293326962592603,
      "loss": 4.6403,
      "step": 36090
    },
    {
      "epoch": 0.7392541928613847,
      "grad_norm": 0.30478453636169434,
      "learning_rate": 0.0001227631089513159,
      "loss": 4.6214,
      "step": 36100
    },
    {
      "epoch": 0.739458972416194,
      "grad_norm": 0.2973614037036896,
      "learning_rate": 0.00012259301396027103,
      "loss": 4.6199,
      "step": 36110
    },
    {
      "epoch": 0.7396637519710032,
      "grad_norm": 0.2853838801383972,
      "learning_rate": 0.00012242298479744302,
      "loss": 4.5925,
      "step": 36120
    },
    {
      "epoch": 0.7398685315258124,
      "grad_norm": 0.3152982294559479,
      "learning_rate": 0.0001222530216074277,
      "loss": 4.5864,
      "step": 36130
    },
    {
      "epoch": 0.7400733110806217,
      "grad_norm": 0.29252418875694275,
      "learning_rate": 0.00012208312453476443,
      "loss": 4.6193,
      "step": 36140
    },
    {
      "epoch": 0.7402780906354309,
      "grad_norm": 0.325703889131546,
      "learning_rate": 0.00012191329372393674,
      "loss": 4.6008,
      "step": 36150
    },
    {
      "epoch": 0.7404828701902402,
      "grad_norm": 0.31771841645240784,
      "learning_rate": 0.00012174352931937161,
      "loss": 4.5972,
      "step": 36160
    },
    {
      "epoch": 0.7406876497450494,
      "grad_norm": 0.31104570627212524,
      "learning_rate": 0.00012157383146543955,
      "loss": 4.6389,
      "step": 36170
    },
    {
      "epoch": 0.7408924292998587,
      "grad_norm": 0.33067429065704346,
      "learning_rate": 0.00012140420030645448,
      "loss": 4.6244,
      "step": 36180
    },
    {
      "epoch": 0.7410972088546679,
      "grad_norm": 0.31437385082244873,
      "learning_rate": 0.00012123463598667366,
      "loss": 4.6347,
      "step": 36190
    },
    {
      "epoch": 0.7413019884094773,
      "grad_norm": 0.3336164355278015,
      "learning_rate": 0.00012106513865029754,
      "loss": 4.5844,
      "step": 36200
    },
    {
      "epoch": 0.7415067679642865,
      "grad_norm": 0.3269065320491791,
      "learning_rate": 0.00012089570844146937,
      "loss": 4.6002,
      "step": 36210
    },
    {
      "epoch": 0.7417115475190957,
      "grad_norm": 0.3056379556655884,
      "learning_rate": 0.00012072634550427562,
      "loss": 4.607,
      "step": 36220
    },
    {
      "epoch": 0.741916327073905,
      "grad_norm": 0.3031263053417206,
      "learning_rate": 0.00012055704998274531,
      "loss": 4.5836,
      "step": 36230
    },
    {
      "epoch": 0.7421211066287142,
      "grad_norm": 0.309474915266037,
      "learning_rate": 0.0001203878220208503,
      "loss": 4.5703,
      "step": 36240
    },
    {
      "epoch": 0.7423258861835235,
      "grad_norm": 0.29727813601493835,
      "learning_rate": 0.00012021866176250491,
      "loss": 4.5984,
      "step": 36250
    },
    {
      "epoch": 0.7425306657383327,
      "grad_norm": 0.28859925270080566,
      "learning_rate": 0.00012004956935156588,
      "loss": 4.632,
      "step": 36260
    },
    {
      "epoch": 0.742735445293142,
      "grad_norm": 0.30916011333465576,
      "learning_rate": 0.00011988054493183236,
      "loss": 4.5911,
      "step": 36270
    },
    {
      "epoch": 0.7429402248479512,
      "grad_norm": 0.3086493909358978,
      "learning_rate": 0.00011971158864704546,
      "loss": 4.6114,
      "step": 36280
    },
    {
      "epoch": 0.7431450044027604,
      "grad_norm": 0.2961837351322174,
      "learning_rate": 0.0001195427006408885,
      "loss": 4.5906,
      "step": 36290
    },
    {
      "epoch": 0.7433497839575697,
      "grad_norm": 0.32759809494018555,
      "learning_rate": 0.00011937388105698675,
      "loss": 4.6136,
      "step": 36300
    },
    {
      "epoch": 0.7435545635123789,
      "grad_norm": 0.32991155982017517,
      "learning_rate": 0.0001192051300389073,
      "loss": 4.6276,
      "step": 36310
    },
    {
      "epoch": 0.7437593430671882,
      "grad_norm": 0.30752769112586975,
      "learning_rate": 0.00011903644773015878,
      "loss": 4.5985,
      "step": 36320
    },
    {
      "epoch": 0.7439641226219974,
      "grad_norm": 0.2945796549320221,
      "learning_rate": 0.00011886783427419153,
      "loss": 4.6031,
      "step": 36330
    },
    {
      "epoch": 0.7441689021768066,
      "grad_norm": 0.29451462626457214,
      "learning_rate": 0.00011869928981439741,
      "loss": 4.5902,
      "step": 36340
    },
    {
      "epoch": 0.7443736817316159,
      "grad_norm": 0.2927614748477936,
      "learning_rate": 0.00011853081449410934,
      "loss": 4.6031,
      "step": 36350
    },
    {
      "epoch": 0.7445784612864251,
      "grad_norm": 0.31154003739356995,
      "learning_rate": 0.00011836240845660169,
      "loss": 4.6021,
      "step": 36360
    },
    {
      "epoch": 0.7447832408412344,
      "grad_norm": 0.3381774425506592,
      "learning_rate": 0.0001181940718450897,
      "loss": 4.6047,
      "step": 36370
    },
    {
      "epoch": 0.7449880203960436,
      "grad_norm": 0.30159881711006165,
      "learning_rate": 0.00011802580480272983,
      "loss": 4.5748,
      "step": 36380
    },
    {
      "epoch": 0.7451927999508529,
      "grad_norm": 0.3149918019771576,
      "learning_rate": 0.00011785760747261913,
      "loss": 4.5956,
      "step": 36390
    },
    {
      "epoch": 0.7453975795056621,
      "grad_norm": 0.3004690706729889,
      "learning_rate": 0.00011768947999779558,
      "loss": 4.5967,
      "step": 36400
    },
    {
      "epoch": 0.7456023590604715,
      "grad_norm": 0.47824764251708984,
      "learning_rate": 0.00011752142252123759,
      "loss": 4.6055,
      "step": 36410
    },
    {
      "epoch": 0.7458071386152807,
      "grad_norm": 0.3609195649623871,
      "learning_rate": 0.00011735343518586402,
      "loss": 4.5984,
      "step": 36420
    },
    {
      "epoch": 0.7460119181700899,
      "grad_norm": 0.30279165506362915,
      "learning_rate": 0.0001171855181345342,
      "loss": 4.6286,
      "step": 36430
    },
    {
      "epoch": 0.7462166977248992,
      "grad_norm": 0.32302072644233704,
      "learning_rate": 0.00011701767151004768,
      "loss": 4.668,
      "step": 36440
    },
    {
      "epoch": 0.7464214772797084,
      "grad_norm": 0.3194332420825958,
      "learning_rate": 0.00011684989545514408,
      "loss": 4.6015,
      "step": 36450
    },
    {
      "epoch": 0.7466262568345177,
      "grad_norm": 0.29721352458000183,
      "learning_rate": 0.00011668219011250299,
      "loss": 4.5754,
      "step": 36460
    },
    {
      "epoch": 0.7468310363893269,
      "grad_norm": 0.31784340739250183,
      "learning_rate": 0.00011651455562474385,
      "loss": 4.6197,
      "step": 36470
    },
    {
      "epoch": 0.7470358159441362,
      "grad_norm": 0.291959673166275,
      "learning_rate": 0.00011634699213442602,
      "loss": 4.5766,
      "step": 36480
    },
    {
      "epoch": 0.7472405954989454,
      "grad_norm": 0.30183809995651245,
      "learning_rate": 0.00011617949978404814,
      "loss": 4.5665,
      "step": 36490
    },
    {
      "epoch": 0.7474453750537546,
      "grad_norm": 0.2972424328327179,
      "learning_rate": 0.00011601207871604864,
      "loss": 4.5871,
      "step": 36500
    },
    {
      "epoch": 0.7476501546085639,
      "grad_norm": 0.28370097279548645,
      "learning_rate": 0.00011584472907280529,
      "loss": 4.6079,
      "step": 36510
    },
    {
      "epoch": 0.7478549341633731,
      "grad_norm": 0.3022497892379761,
      "learning_rate": 0.00011567745099663507,
      "loss": 4.5993,
      "step": 36520
    },
    {
      "epoch": 0.7480597137181824,
      "grad_norm": 0.3323211967945099,
      "learning_rate": 0.00011551024462979403,
      "loss": 4.6169,
      "step": 36530
    },
    {
      "epoch": 0.7482644932729916,
      "grad_norm": 0.29171788692474365,
      "learning_rate": 0.00011534311011447729,
      "loss": 4.6293,
      "step": 36540
    },
    {
      "epoch": 0.7484692728278008,
      "grad_norm": 0.2999872863292694,
      "learning_rate": 0.00011517604759281906,
      "loss": 4.5968,
      "step": 36550
    },
    {
      "epoch": 0.7486740523826101,
      "grad_norm": 0.3147103488445282,
      "learning_rate": 0.0001150090572068919,
      "loss": 4.555,
      "step": 36560
    },
    {
      "epoch": 0.7488788319374193,
      "grad_norm": 0.30209001898765564,
      "learning_rate": 0.00011484213909870743,
      "loss": 4.5927,
      "step": 36570
    },
    {
      "epoch": 0.7490836114922286,
      "grad_norm": 0.29491347074508667,
      "learning_rate": 0.0001146752934102156,
      "loss": 4.6015,
      "step": 36580
    },
    {
      "epoch": 0.7492883910470378,
      "grad_norm": 0.3398585021495819,
      "learning_rate": 0.00011450852028330481,
      "loss": 4.6068,
      "step": 36590
    },
    {
      "epoch": 0.7494931706018471,
      "grad_norm": 0.30051180720329285,
      "learning_rate": 0.00011434181985980182,
      "loss": 4.6119,
      "step": 36600
    },
    {
      "epoch": 0.7496979501566564,
      "grad_norm": 0.3194768726825714,
      "learning_rate": 0.00011417519228147153,
      "loss": 4.6278,
      "step": 36610
    },
    {
      "epoch": 0.7499027297114657,
      "grad_norm": 0.306112676858902,
      "learning_rate": 0.00011400863769001673,
      "loss": 4.5769,
      "step": 36620
    },
    {
      "epoch": 0.7501075092662749,
      "grad_norm": 0.30987900495529175,
      "learning_rate": 0.00011384215622707843,
      "loss": 4.5989,
      "step": 36630
    },
    {
      "epoch": 0.7503122888210841,
      "grad_norm": 0.3256874680519104,
      "learning_rate": 0.00011367574803423512,
      "loss": 4.5877,
      "step": 36640
    },
    {
      "epoch": 0.7505170683758934,
      "grad_norm": 0.2956451177597046,
      "learning_rate": 0.00011350941325300328,
      "loss": 4.6411,
      "step": 36650
    },
    {
      "epoch": 0.7507218479307026,
      "grad_norm": 0.32243770360946655,
      "learning_rate": 0.00011334315202483679,
      "loss": 4.6034,
      "step": 36660
    },
    {
      "epoch": 0.7509266274855119,
      "grad_norm": 0.3013778626918793,
      "learning_rate": 0.00011317696449112705,
      "loss": 4.6063,
      "step": 36670
    },
    {
      "epoch": 0.7511314070403211,
      "grad_norm": 0.307576447725296,
      "learning_rate": 0.00011301085079320275,
      "loss": 4.5892,
      "step": 36680
    },
    {
      "epoch": 0.7513361865951304,
      "grad_norm": 0.3025556802749634,
      "learning_rate": 0.00011284481107232991,
      "loss": 4.5921,
      "step": 36690
    },
    {
      "epoch": 0.7515409661499396,
      "grad_norm": 0.30304038524627686,
      "learning_rate": 0.00011267884546971133,
      "loss": 4.5788,
      "step": 36700
    },
    {
      "epoch": 0.7517457457047488,
      "grad_norm": 0.3039272427558899,
      "learning_rate": 0.0001125129541264871,
      "loss": 4.5963,
      "step": 36710
    },
    {
      "epoch": 0.7519505252595581,
      "grad_norm": 0.2991809546947479,
      "learning_rate": 0.00011234713718373409,
      "loss": 4.6294,
      "step": 36720
    },
    {
      "epoch": 0.7521553048143673,
      "grad_norm": 0.2960982322692871,
      "learning_rate": 0.0001121813947824657,
      "loss": 4.5752,
      "step": 36730
    },
    {
      "epoch": 0.7523600843691766,
      "grad_norm": 0.30794692039489746,
      "learning_rate": 0.00011201572706363215,
      "loss": 4.5912,
      "step": 36740
    },
    {
      "epoch": 0.7525648639239858,
      "grad_norm": 0.3139549791812897,
      "learning_rate": 0.00011185013416812012,
      "loss": 4.6058,
      "step": 36750
    },
    {
      "epoch": 0.752769643478795,
      "grad_norm": 0.3105529844760895,
      "learning_rate": 0.00011168461623675266,
      "loss": 4.5728,
      "step": 36760
    },
    {
      "epoch": 0.7529744230336043,
      "grad_norm": 0.3028867840766907,
      "learning_rate": 0.0001115191734102889,
      "loss": 4.5847,
      "step": 36770
    },
    {
      "epoch": 0.7531792025884135,
      "grad_norm": 0.2984515130519867,
      "learning_rate": 0.00011135380582942432,
      "loss": 4.6105,
      "step": 36780
    },
    {
      "epoch": 0.7533839821432228,
      "grad_norm": 0.3259239196777344,
      "learning_rate": 0.00011118851363479028,
      "loss": 4.5859,
      "step": 36790
    },
    {
      "epoch": 0.753588761698032,
      "grad_norm": 0.33328720927238464,
      "learning_rate": 0.00011102329696695414,
      "loss": 4.5849,
      "step": 36800
    },
    {
      "epoch": 0.7537935412528414,
      "grad_norm": 0.4639306366443634,
      "learning_rate": 0.00011085815596641898,
      "loss": 4.6112,
      "step": 36810
    },
    {
      "epoch": 0.7539983208076506,
      "grad_norm": 0.30320459604263306,
      "learning_rate": 0.00011069309077362338,
      "loss": 4.588,
      "step": 36820
    },
    {
      "epoch": 0.7542031003624599,
      "grad_norm": 0.3142356872558594,
      "learning_rate": 0.00011052810152894176,
      "loss": 4.6379,
      "step": 36830
    },
    {
      "epoch": 0.7544078799172691,
      "grad_norm": 0.30572620034217834,
      "learning_rate": 0.00011036318837268358,
      "loss": 4.5899,
      "step": 36840
    },
    {
      "epoch": 0.7546126594720783,
      "grad_norm": 0.3086054027080536,
      "learning_rate": 0.0001101983514450939,
      "loss": 4.6065,
      "step": 36850
    },
    {
      "epoch": 0.7548174390268876,
      "grad_norm": 0.29797255992889404,
      "learning_rate": 0.00011003359088635284,
      "loss": 4.6234,
      "step": 36860
    },
    {
      "epoch": 0.7550222185816968,
      "grad_norm": 0.2883145809173584,
      "learning_rate": 0.00010986890683657558,
      "loss": 4.5927,
      "step": 36870
    },
    {
      "epoch": 0.7552269981365061,
      "grad_norm": 0.2922157347202301,
      "learning_rate": 0.00010970429943581226,
      "loss": 4.6015,
      "step": 36880
    },
    {
      "epoch": 0.7554317776913153,
      "grad_norm": 0.31169426441192627,
      "learning_rate": 0.00010953976882404777,
      "loss": 4.5758,
      "step": 36890
    },
    {
      "epoch": 0.7556365572461246,
      "grad_norm": 0.3007355332374573,
      "learning_rate": 0.0001093753151412018,
      "loss": 4.6062,
      "step": 36900
    },
    {
      "epoch": 0.7558413368009338,
      "grad_norm": 0.4668189287185669,
      "learning_rate": 0.00010921093852712858,
      "loss": 4.5949,
      "step": 36910
    },
    {
      "epoch": 0.756046116355743,
      "grad_norm": 0.3131042420864105,
      "learning_rate": 0.00010904663912161659,
      "loss": 4.6125,
      "step": 36920
    },
    {
      "epoch": 0.7562508959105523,
      "grad_norm": 0.3142430782318115,
      "learning_rate": 0.000108882417064389,
      "loss": 4.6042,
      "step": 36930
    },
    {
      "epoch": 0.7564556754653615,
      "grad_norm": 0.2970556318759918,
      "learning_rate": 0.00010871827249510296,
      "loss": 4.5867,
      "step": 36940
    },
    {
      "epoch": 0.7566604550201708,
      "grad_norm": 0.28931474685668945,
      "learning_rate": 0.00010855420555334986,
      "loss": 4.6236,
      "step": 36950
    },
    {
      "epoch": 0.75686523457498,
      "grad_norm": 0.31609848141670227,
      "learning_rate": 0.00010839021637865498,
      "loss": 4.5771,
      "step": 36960
    },
    {
      "epoch": 0.7570700141297892,
      "grad_norm": 0.3069298565387726,
      "learning_rate": 0.00010822630511047762,
      "loss": 4.5832,
      "step": 36970
    },
    {
      "epoch": 0.7572747936845985,
      "grad_norm": 0.30308017134666443,
      "learning_rate": 0.00010806247188821052,
      "loss": 4.5756,
      "step": 36980
    },
    {
      "epoch": 0.7574795732394077,
      "grad_norm": 0.3147658705711365,
      "learning_rate": 0.00010789871685118033,
      "loss": 4.605,
      "step": 36990
    },
    {
      "epoch": 0.757684352794217,
      "grad_norm": 0.2971055805683136,
      "learning_rate": 0.00010773504013864722,
      "loss": 4.5754,
      "step": 37000
    },
    {
      "epoch": 0.7578891323490262,
      "grad_norm": 0.29778382182121277,
      "learning_rate": 0.00010757144188980441,
      "loss": 4.5987,
      "step": 37010
    },
    {
      "epoch": 0.7580939119038356,
      "grad_norm": 0.29821091890335083,
      "learning_rate": 0.0001074079222437788,
      "loss": 4.6008,
      "step": 37020
    },
    {
      "epoch": 0.7582986914586448,
      "grad_norm": 0.30124008655548096,
      "learning_rate": 0.00010724448133963023,
      "loss": 4.5957,
      "step": 37030
    },
    {
      "epoch": 0.7585034710134541,
      "grad_norm": 0.32034429907798767,
      "learning_rate": 0.00010708111931635172,
      "loss": 4.6259,
      "step": 37040
    },
    {
      "epoch": 0.7587082505682633,
      "grad_norm": 0.3446129262447357,
      "learning_rate": 0.00010691783631286895,
      "loss": 4.5889,
      "step": 37050
    },
    {
      "epoch": 0.7589130301230725,
      "grad_norm": 0.3032030463218689,
      "learning_rate": 0.00010675463246804066,
      "loss": 4.5788,
      "step": 37060
    },
    {
      "epoch": 0.7591178096778818,
      "grad_norm": 0.2998063266277313,
      "learning_rate": 0.00010659150792065813,
      "loss": 4.6174,
      "step": 37070
    },
    {
      "epoch": 0.759322589232691,
      "grad_norm": 0.31689971685409546,
      "learning_rate": 0.00010642846280944527,
      "loss": 4.5681,
      "step": 37080
    },
    {
      "epoch": 0.7595273687875003,
      "grad_norm": 0.6803306341171265,
      "learning_rate": 0.0001062654972730585,
      "loss": 4.6225,
      "step": 37090
    },
    {
      "epoch": 0.7597321483423095,
      "grad_norm": 0.3014300763607025,
      "learning_rate": 0.0001061026114500863,
      "loss": 4.6133,
      "step": 37100
    },
    {
      "epoch": 0.7599369278971188,
      "grad_norm": 0.30094754695892334,
      "learning_rate": 0.00010593980547904978,
      "loss": 4.5921,
      "step": 37110
    },
    {
      "epoch": 0.760141707451928,
      "grad_norm": 0.30403319001197815,
      "learning_rate": 0.00010577707949840165,
      "loss": 4.6003,
      "step": 37120
    },
    {
      "epoch": 0.7603464870067372,
      "grad_norm": 0.30160754919052124,
      "learning_rate": 0.00010561443364652697,
      "loss": 4.6264,
      "step": 37130
    },
    {
      "epoch": 0.7605512665615465,
      "grad_norm": 0.28557589650154114,
      "learning_rate": 0.00010545186806174254,
      "loss": 4.5817,
      "step": 37140
    },
    {
      "epoch": 0.7607560461163557,
      "grad_norm": 0.3045676052570343,
      "learning_rate": 0.00010528938288229686,
      "loss": 4.6053,
      "step": 37150
    },
    {
      "epoch": 0.760960825671165,
      "grad_norm": 0.2812917232513428,
      "learning_rate": 0.00010512697824637016,
      "loss": 4.6218,
      "step": 37160
    },
    {
      "epoch": 0.7611656052259742,
      "grad_norm": 0.30035391449928284,
      "learning_rate": 0.00010496465429207407,
      "loss": 4.5886,
      "step": 37170
    },
    {
      "epoch": 0.7613703847807834,
      "grad_norm": 0.36716723442077637,
      "learning_rate": 0.0001048024111574517,
      "loss": 4.5839,
      "step": 37180
    },
    {
      "epoch": 0.7615751643355927,
      "grad_norm": 0.3121277689933777,
      "learning_rate": 0.00010464024898047724,
      "loss": 4.5836,
      "step": 37190
    },
    {
      "epoch": 0.7617799438904019,
      "grad_norm": 0.3098382353782654,
      "learning_rate": 0.00010447816789905628,
      "loss": 4.5854,
      "step": 37200
    },
    {
      "epoch": 0.7619847234452112,
      "grad_norm": 0.295818030834198,
      "learning_rate": 0.00010431616805102523,
      "loss": 4.6093,
      "step": 37210
    },
    {
      "epoch": 0.7621895030000205,
      "grad_norm": 0.32165735960006714,
      "learning_rate": 0.00010415424957415151,
      "loss": 4.6063,
      "step": 37220
    },
    {
      "epoch": 0.7623942825548298,
      "grad_norm": 0.3486019968986511,
      "learning_rate": 0.00010399241260613344,
      "loss": 4.5722,
      "step": 37230
    },
    {
      "epoch": 0.762599062109639,
      "grad_norm": 0.29253706336021423,
      "learning_rate": 0.00010383065728459986,
      "loss": 4.6153,
      "step": 37240
    },
    {
      "epoch": 0.7628038416644483,
      "grad_norm": 0.31030285358428955,
      "learning_rate": 0.00010366898374711035,
      "loss": 4.5734,
      "step": 37250
    },
    {
      "epoch": 0.7630086212192575,
      "grad_norm": 0.30715376138687134,
      "learning_rate": 0.0001035073921311547,
      "loss": 4.5758,
      "step": 37260
    },
    {
      "epoch": 0.7632134007740667,
      "grad_norm": 0.302558571100235,
      "learning_rate": 0.00010334588257415321,
      "loss": 4.6307,
      "step": 37270
    },
    {
      "epoch": 0.763418180328876,
      "grad_norm": 0.2951318919658661,
      "learning_rate": 0.00010318445521345636,
      "loss": 4.6049,
      "step": 37280
    },
    {
      "epoch": 0.7636229598836852,
      "grad_norm": 0.29280537366867065,
      "learning_rate": 0.00010302311018634481,
      "loss": 4.566,
      "step": 37290
    },
    {
      "epoch": 0.7638277394384945,
      "grad_norm": 0.2905118763446808,
      "learning_rate": 0.00010286184763002893,
      "loss": 4.5883,
      "step": 37300
    },
    {
      "epoch": 0.7640325189933037,
      "grad_norm": 0.3478008210659027,
      "learning_rate": 0.00010270066768164927,
      "loss": 4.5553,
      "step": 37310
    },
    {
      "epoch": 0.764237298548113,
      "grad_norm": 0.34253203868865967,
      "learning_rate": 0.00010253957047827603,
      "loss": 4.5991,
      "step": 37320
    },
    {
      "epoch": 0.7644420781029222,
      "grad_norm": 0.30275651812553406,
      "learning_rate": 0.00010237855615690887,
      "loss": 4.5748,
      "step": 37330
    },
    {
      "epoch": 0.7646468576577314,
      "grad_norm": 0.30308985710144043,
      "learning_rate": 0.00010221762485447719,
      "loss": 4.5899,
      "step": 37340
    },
    {
      "epoch": 0.7648516372125407,
      "grad_norm": 0.28643912076950073,
      "learning_rate": 0.00010205677670783971,
      "loss": 4.5908,
      "step": 37350
    },
    {
      "epoch": 0.7650564167673499,
      "grad_norm": 0.33123040199279785,
      "learning_rate": 0.00010189601185378444,
      "loss": 4.6209,
      "step": 37360
    },
    {
      "epoch": 0.7652611963221592,
      "grad_norm": 0.32299116253852844,
      "learning_rate": 0.00010173533042902847,
      "loss": 4.5494,
      "step": 37370
    },
    {
      "epoch": 0.7654659758769684,
      "grad_norm": 0.3314318358898163,
      "learning_rate": 0.00010157473257021819,
      "loss": 4.578,
      "step": 37380
    },
    {
      "epoch": 0.7656707554317776,
      "grad_norm": 0.3351702094078064,
      "learning_rate": 0.00010141421841392853,
      "loss": 4.6043,
      "step": 37390
    },
    {
      "epoch": 0.7658755349865869,
      "grad_norm": 0.3096257150173187,
      "learning_rate": 0.00010125378809666361,
      "loss": 4.593,
      "step": 37400
    },
    {
      "epoch": 0.7660803145413961,
      "grad_norm": 0.33093225955963135,
      "learning_rate": 0.00010109344175485602,
      "loss": 4.5737,
      "step": 37410
    },
    {
      "epoch": 0.7662850940962055,
      "grad_norm": 0.2949813902378082,
      "learning_rate": 0.00010093317952486701,
      "loss": 4.5586,
      "step": 37420
    },
    {
      "epoch": 0.7664898736510147,
      "grad_norm": 0.306266725063324,
      "learning_rate": 0.00010077300154298634,
      "loss": 4.594,
      "step": 37430
    },
    {
      "epoch": 0.766694653205824,
      "grad_norm": 0.3152048587799072,
      "learning_rate": 0.0001006129079454321,
      "loss": 4.5772,
      "step": 37440
    },
    {
      "epoch": 0.7668994327606332,
      "grad_norm": 0.32708024978637695,
      "learning_rate": 0.00010045289886835058,
      "loss": 4.6239,
      "step": 37450
    },
    {
      "epoch": 0.7671042123154425,
      "grad_norm": 0.30479657649993896,
      "learning_rate": 0.00010029297444781631,
      "loss": 4.6122,
      "step": 37460
    },
    {
      "epoch": 0.7673089918702517,
      "grad_norm": 0.30367597937583923,
      "learning_rate": 0.00010013313481983155,
      "loss": 4.5983,
      "step": 37470
    },
    {
      "epoch": 0.7675137714250609,
      "grad_norm": 0.3227415382862091,
      "learning_rate": 9.997338012032678e-05,
      "loss": 4.5872,
      "step": 37480
    },
    {
      "epoch": 0.7677185509798702,
      "grad_norm": 0.3020555078983307,
      "learning_rate": 9.981371048516015e-05,
      "loss": 4.5939,
      "step": 37490
    },
    {
      "epoch": 0.7679233305346794,
      "grad_norm": 0.31975632905960083,
      "learning_rate": 9.965412605011728e-05,
      "loss": 4.6055,
      "step": 37500
    },
    {
      "epoch": 0.7681281100894887,
      "grad_norm": 0.3118675649166107,
      "learning_rate": 9.949462695091156e-05,
      "loss": 4.5831,
      "step": 37510
    },
    {
      "epoch": 0.7683328896442979,
      "grad_norm": 0.30988624691963196,
      "learning_rate": 9.933521332318374e-05,
      "loss": 4.6069,
      "step": 37520
    },
    {
      "epoch": 0.7685376691991072,
      "grad_norm": 0.3117412328720093,
      "learning_rate": 9.917588530250195e-05,
      "loss": 4.5924,
      "step": 37530
    },
    {
      "epoch": 0.7687424487539164,
      "grad_norm": 0.30757349729537964,
      "learning_rate": 9.901664302436131e-05,
      "loss": 4.5811,
      "step": 37540
    },
    {
      "epoch": 0.7689472283087256,
      "grad_norm": 0.397675096988678,
      "learning_rate": 9.885748662418423e-05,
      "loss": 4.6202,
      "step": 37550
    },
    {
      "epoch": 0.7691520078635349,
      "grad_norm": 0.37244829535484314,
      "learning_rate": 9.869841623732001e-05,
      "loss": 4.6222,
      "step": 37560
    },
    {
      "epoch": 0.7693567874183441,
      "grad_norm": 0.2990577518939972,
      "learning_rate": 9.853943199904483e-05,
      "loss": 4.5844,
      "step": 37570
    },
    {
      "epoch": 0.7695615669731534,
      "grad_norm": 0.30558040738105774,
      "learning_rate": 9.838053404456163e-05,
      "loss": 4.5972,
      "step": 37580
    },
    {
      "epoch": 0.7697663465279626,
      "grad_norm": 0.30110323429107666,
      "learning_rate": 9.822172250899982e-05,
      "loss": 4.6029,
      "step": 37590
    },
    {
      "epoch": 0.7699711260827719,
      "grad_norm": 0.3397175371646881,
      "learning_rate": 9.806299752741557e-05,
      "loss": 4.5866,
      "step": 37600
    },
    {
      "epoch": 0.7701759056375811,
      "grad_norm": 0.31180909276008606,
      "learning_rate": 9.790435923479113e-05,
      "loss": 4.5937,
      "step": 37610
    },
    {
      "epoch": 0.7703806851923904,
      "grad_norm": 0.3076137900352478,
      "learning_rate": 9.77458077660353e-05,
      "loss": 4.5775,
      "step": 37620
    },
    {
      "epoch": 0.7705854647471997,
      "grad_norm": 0.3023734986782074,
      "learning_rate": 9.758734325598294e-05,
      "loss": 4.6057,
      "step": 37630
    },
    {
      "epoch": 0.7707902443020089,
      "grad_norm": 0.3215528428554535,
      "learning_rate": 9.742896583939494e-05,
      "loss": 4.5711,
      "step": 37640
    },
    {
      "epoch": 0.7709950238568182,
      "grad_norm": 0.3696591854095459,
      "learning_rate": 9.72706756509582e-05,
      "loss": 4.5684,
      "step": 37650
    },
    {
      "epoch": 0.7711998034116274,
      "grad_norm": 0.3098335862159729,
      "learning_rate": 9.711247282528533e-05,
      "loss": 4.5918,
      "step": 37660
    },
    {
      "epoch": 0.7714045829664367,
      "grad_norm": 0.2871721088886261,
      "learning_rate": 9.695435749691481e-05,
      "loss": 4.6285,
      "step": 37670
    },
    {
      "epoch": 0.7716093625212459,
      "grad_norm": 0.3242889940738678,
      "learning_rate": 9.679632980031053e-05,
      "loss": 4.5914,
      "step": 37680
    },
    {
      "epoch": 0.7718141420760551,
      "grad_norm": 0.2907813787460327,
      "learning_rate": 9.663838986986182e-05,
      "loss": 4.6088,
      "step": 37690
    },
    {
      "epoch": 0.7720189216308644,
      "grad_norm": 0.3036256432533264,
      "learning_rate": 9.64805378398836e-05,
      "loss": 4.5727,
      "step": 37700
    },
    {
      "epoch": 0.7722237011856736,
      "grad_norm": 0.3210693597793579,
      "learning_rate": 9.63227738446159e-05,
      "loss": 4.5851,
      "step": 37710
    },
    {
      "epoch": 0.7724284807404829,
      "grad_norm": 0.2903750538825989,
      "learning_rate": 9.616509801822391e-05,
      "loss": 4.6,
      "step": 37720
    },
    {
      "epoch": 0.7726332602952921,
      "grad_norm": 0.32857513427734375,
      "learning_rate": 9.600751049479783e-05,
      "loss": 4.5819,
      "step": 37730
    },
    {
      "epoch": 0.7728380398501014,
      "grad_norm": 0.3713500499725342,
      "learning_rate": 9.585001140835283e-05,
      "loss": 4.5911,
      "step": 37740
    },
    {
      "epoch": 0.7730428194049106,
      "grad_norm": 0.29100435972213745,
      "learning_rate": 9.56926008928287e-05,
      "loss": 4.6329,
      "step": 37750
    },
    {
      "epoch": 0.7732475989597198,
      "grad_norm": 0.3058205544948578,
      "learning_rate": 9.553527908209005e-05,
      "loss": 4.5956,
      "step": 37760
    },
    {
      "epoch": 0.7734523785145291,
      "grad_norm": 0.3001843988895416,
      "learning_rate": 9.53780461099261e-05,
      "loss": 4.5968,
      "step": 37770
    },
    {
      "epoch": 0.7736571580693383,
      "grad_norm": 0.36424750089645386,
      "learning_rate": 9.522090211005034e-05,
      "loss": 4.5813,
      "step": 37780
    },
    {
      "epoch": 0.7738619376241476,
      "grad_norm": 0.30522090196609497,
      "learning_rate": 9.506384721610074e-05,
      "loss": 4.57,
      "step": 37790
    },
    {
      "epoch": 0.7740667171789568,
      "grad_norm": 0.3051900863647461,
      "learning_rate": 9.490688156163943e-05,
      "loss": 4.601,
      "step": 37800
    },
    {
      "epoch": 0.774271496733766,
      "grad_norm": 0.29674139618873596,
      "learning_rate": 9.475000528015279e-05,
      "loss": 4.6044,
      "step": 37810
    },
    {
      "epoch": 0.7744762762885753,
      "grad_norm": 0.32481276988983154,
      "learning_rate": 9.459321850505088e-05,
      "loss": 4.6168,
      "step": 37820
    },
    {
      "epoch": 0.7746810558433846,
      "grad_norm": 0.32346010208129883,
      "learning_rate": 9.443652136966797e-05,
      "loss": 4.6014,
      "step": 37830
    },
    {
      "epoch": 0.7748858353981939,
      "grad_norm": 0.30109137296676636,
      "learning_rate": 9.427991400726191e-05,
      "loss": 4.5801,
      "step": 37840
    },
    {
      "epoch": 0.7750906149530031,
      "grad_norm": 0.2935716509819031,
      "learning_rate": 9.412339655101432e-05,
      "loss": 4.5461,
      "step": 37850
    },
    {
      "epoch": 0.7752953945078124,
      "grad_norm": 0.34225064516067505,
      "learning_rate": 9.396696913403034e-05,
      "loss": 4.5706,
      "step": 37860
    },
    {
      "epoch": 0.7755001740626216,
      "grad_norm": 0.3051619529724121,
      "learning_rate": 9.381063188933836e-05,
      "loss": 4.6069,
      "step": 37870
    },
    {
      "epoch": 0.7757049536174309,
      "grad_norm": 0.30239740014076233,
      "learning_rate": 9.365438494989039e-05,
      "loss": 4.5534,
      "step": 37880
    },
    {
      "epoch": 0.7759097331722401,
      "grad_norm": 0.31871289014816284,
      "learning_rate": 9.349822844856133e-05,
      "loss": 4.5807,
      "step": 37890
    },
    {
      "epoch": 0.7761145127270493,
      "grad_norm": 0.31296980381011963,
      "learning_rate": 9.334216251814943e-05,
      "loss": 4.5859,
      "step": 37900
    },
    {
      "epoch": 0.7763192922818586,
      "grad_norm": 0.295104444026947,
      "learning_rate": 9.318618729137574e-05,
      "loss": 4.6118,
      "step": 37910
    },
    {
      "epoch": 0.7765240718366678,
      "grad_norm": 0.2946823835372925,
      "learning_rate": 9.30303029008843e-05,
      "loss": 4.5727,
      "step": 37920
    },
    {
      "epoch": 0.7767288513914771,
      "grad_norm": 0.42669346928596497,
      "learning_rate": 9.287450947924183e-05,
      "loss": 4.6105,
      "step": 37930
    },
    {
      "epoch": 0.7769336309462863,
      "grad_norm": 0.3017657399177551,
      "learning_rate": 9.27188071589377e-05,
      "loss": 4.5842,
      "step": 37940
    },
    {
      "epoch": 0.7771384105010956,
      "grad_norm": 0.3124864399433136,
      "learning_rate": 9.25631960723839e-05,
      "loss": 4.5936,
      "step": 37950
    },
    {
      "epoch": 0.7773431900559048,
      "grad_norm": 0.30705711245536804,
      "learning_rate": 9.240767635191459e-05,
      "loss": 4.585,
      "step": 37960
    },
    {
      "epoch": 0.777547969610714,
      "grad_norm": 0.3201311230659485,
      "learning_rate": 9.225224812978652e-05,
      "loss": 4.5776,
      "step": 37970
    },
    {
      "epoch": 0.7777527491655233,
      "grad_norm": 0.31811508536338806,
      "learning_rate": 9.209691153817835e-05,
      "loss": 4.591,
      "step": 37980
    },
    {
      "epoch": 0.7779575287203325,
      "grad_norm": 0.30487823486328125,
      "learning_rate": 9.194166670919105e-05,
      "loss": 4.6108,
      "step": 37990
    },
    {
      "epoch": 0.7781623082751418,
      "grad_norm": 0.29707804322242737,
      "learning_rate": 9.178651377484742e-05,
      "loss": 4.591,
      "step": 38000
    },
    {
      "epoch": 0.7781623082751418,
      "eval_loss": 4.590142726898193,
      "eval_runtime": 4.4005,
      "eval_samples_per_second": 264.969,
      "eval_steps_per_second": 33.178,
      "step": 38000
    },
    {
      "epoch": 0.778367087829951,
      "grad_norm": 0.3073808252811432,
      "learning_rate": 9.163145286709218e-05,
      "loss": 4.5674,
      "step": 38010
    },
    {
      "epoch": 0.7785718673847603,
      "grad_norm": 0.28783494234085083,
      "learning_rate": 9.147648411779186e-05,
      "loss": 4.6029,
      "step": 38020
    },
    {
      "epoch": 0.7787766469395696,
      "grad_norm": 0.3142697215080261,
      "learning_rate": 9.132160765873433e-05,
      "loss": 4.5866,
      "step": 38030
    },
    {
      "epoch": 0.7789814264943788,
      "grad_norm": 0.31494787335395813,
      "learning_rate": 9.116682362162925e-05,
      "loss": 4.5598,
      "step": 38040
    },
    {
      "epoch": 0.7791862060491881,
      "grad_norm": 0.6348517537117004,
      "learning_rate": 9.10121321381076e-05,
      "loss": 4.5746,
      "step": 38050
    },
    {
      "epoch": 0.7793909856039973,
      "grad_norm": 0.3252737522125244,
      "learning_rate": 9.085753333972174e-05,
      "loss": 4.5977,
      "step": 38060
    },
    {
      "epoch": 0.7795957651588066,
      "grad_norm": 0.31998464465141296,
      "learning_rate": 9.070302735794493e-05,
      "loss": 4.5725,
      "step": 38070
    },
    {
      "epoch": 0.7798005447136158,
      "grad_norm": 0.2928812801837921,
      "learning_rate": 9.054861432417179e-05,
      "loss": 4.5628,
      "step": 38080
    },
    {
      "epoch": 0.7800053242684251,
      "grad_norm": 0.29339835047721863,
      "learning_rate": 9.039429436971783e-05,
      "loss": 4.5714,
      "step": 38090
    },
    {
      "epoch": 0.7802101038232343,
      "grad_norm": 0.3693881630897522,
      "learning_rate": 9.024006762581924e-05,
      "loss": 4.6113,
      "step": 38100
    },
    {
      "epoch": 0.7804148833780435,
      "grad_norm": 0.32775619626045227,
      "learning_rate": 9.008593422363316e-05,
      "loss": 4.537,
      "step": 38110
    },
    {
      "epoch": 0.7806196629328528,
      "grad_norm": 0.3084632456302643,
      "learning_rate": 8.993189429423723e-05,
      "loss": 4.5917,
      "step": 38120
    },
    {
      "epoch": 0.780824442487662,
      "grad_norm": 0.29846644401550293,
      "learning_rate": 8.97779479686296e-05,
      "loss": 4.563,
      "step": 38130
    },
    {
      "epoch": 0.7810292220424713,
      "grad_norm": 0.30271339416503906,
      "learning_rate": 8.962409537772886e-05,
      "loss": 4.5815,
      "step": 38140
    },
    {
      "epoch": 0.7812340015972805,
      "grad_norm": 0.345956027507782,
      "learning_rate": 8.947033665237398e-05,
      "loss": 4.5569,
      "step": 38150
    },
    {
      "epoch": 0.7814387811520898,
      "grad_norm": 0.3943561613559723,
      "learning_rate": 8.931667192332389e-05,
      "loss": 4.5851,
      "step": 38160
    },
    {
      "epoch": 0.781643560706899,
      "grad_norm": 0.3255510628223419,
      "learning_rate": 8.91631013212576e-05,
      "loss": 4.587,
      "step": 38170
    },
    {
      "epoch": 0.7818483402617082,
      "grad_norm": 0.34418126940727234,
      "learning_rate": 8.900962497677426e-05,
      "loss": 4.588,
      "step": 38180
    },
    {
      "epoch": 0.7820531198165175,
      "grad_norm": 0.35830941796302795,
      "learning_rate": 8.885624302039275e-05,
      "loss": 4.6346,
      "step": 38190
    },
    {
      "epoch": 0.7822578993713267,
      "grad_norm": 0.4159878194332123,
      "learning_rate": 8.870295558255168e-05,
      "loss": 4.6069,
      "step": 38200
    },
    {
      "epoch": 0.782462678926136,
      "grad_norm": 0.3376203179359436,
      "learning_rate": 8.854976279360932e-05,
      "loss": 4.5778,
      "step": 38210
    },
    {
      "epoch": 0.7826674584809452,
      "grad_norm": 0.3103373944759369,
      "learning_rate": 8.839666478384341e-05,
      "loss": 4.5563,
      "step": 38220
    },
    {
      "epoch": 0.7828722380357546,
      "grad_norm": 0.2928243577480316,
      "learning_rate": 8.824366168345117e-05,
      "loss": 4.5848,
      "step": 38230
    },
    {
      "epoch": 0.7830770175905638,
      "grad_norm": 0.31541907787323,
      "learning_rate": 8.80907536225489e-05,
      "loss": 4.6161,
      "step": 38240
    },
    {
      "epoch": 0.783281797145373,
      "grad_norm": 0.2942469120025635,
      "learning_rate": 8.793794073117235e-05,
      "loss": 4.5592,
      "step": 38250
    },
    {
      "epoch": 0.7834865767001823,
      "grad_norm": 0.643306314945221,
      "learning_rate": 8.778522313927606e-05,
      "loss": 4.5243,
      "step": 38260
    },
    {
      "epoch": 0.7836913562549915,
      "grad_norm": 0.3006947636604309,
      "learning_rate": 8.763260097673374e-05,
      "loss": 4.5895,
      "step": 38270
    },
    {
      "epoch": 0.7838961358098008,
      "grad_norm": 0.3057100176811218,
      "learning_rate": 8.748007437333787e-05,
      "loss": 4.5599,
      "step": 38280
    },
    {
      "epoch": 0.78410091536461,
      "grad_norm": 0.3289993703365326,
      "learning_rate": 8.73276434587997e-05,
      "loss": 4.5967,
      "step": 38290
    },
    {
      "epoch": 0.7843056949194193,
      "grad_norm": 0.30016857385635376,
      "learning_rate": 8.717530836274908e-05,
      "loss": 4.5848,
      "step": 38300
    },
    {
      "epoch": 0.7845104744742285,
      "grad_norm": 0.3119611144065857,
      "learning_rate": 8.702306921473424e-05,
      "loss": 4.6063,
      "step": 38310
    },
    {
      "epoch": 0.7847152540290377,
      "grad_norm": 0.3316038250923157,
      "learning_rate": 8.687092614422205e-05,
      "loss": 4.5819,
      "step": 38320
    },
    {
      "epoch": 0.784920033583847,
      "grad_norm": 0.3957536518573761,
      "learning_rate": 8.671887928059754e-05,
      "loss": 4.608,
      "step": 38330
    },
    {
      "epoch": 0.7851248131386562,
      "grad_norm": 0.322087824344635,
      "learning_rate": 8.656692875316401e-05,
      "loss": 4.6024,
      "step": 38340
    },
    {
      "epoch": 0.7853295926934655,
      "grad_norm": 0.33519017696380615,
      "learning_rate": 8.641507469114263e-05,
      "loss": 4.588,
      "step": 38350
    },
    {
      "epoch": 0.7855343722482747,
      "grad_norm": 0.3098636865615845,
      "learning_rate": 8.626331722367276e-05,
      "loss": 4.58,
      "step": 38360
    },
    {
      "epoch": 0.785739151803084,
      "grad_norm": 0.2896534204483032,
      "learning_rate": 8.611165647981161e-05,
      "loss": 4.5414,
      "step": 38370
    },
    {
      "epoch": 0.7859439313578932,
      "grad_norm": 0.28859734535217285,
      "learning_rate": 8.596009258853388e-05,
      "loss": 4.5305,
      "step": 38380
    },
    {
      "epoch": 0.7861487109127024,
      "grad_norm": 0.3139727711677551,
      "learning_rate": 8.580862567873216e-05,
      "loss": 4.5886,
      "step": 38390
    },
    {
      "epoch": 0.7863534904675117,
      "grad_norm": 0.30453288555145264,
      "learning_rate": 8.565725587921651e-05,
      "loss": 4.5885,
      "step": 38400
    },
    {
      "epoch": 0.7865582700223209,
      "grad_norm": 0.3050139844417572,
      "learning_rate": 8.550598331871435e-05,
      "loss": 4.5737,
      "step": 38410
    },
    {
      "epoch": 0.7867630495771302,
      "grad_norm": 0.31867069005966187,
      "learning_rate": 8.535480812587047e-05,
      "loss": 4.5477,
      "step": 38420
    },
    {
      "epoch": 0.7869678291319395,
      "grad_norm": 0.3078056573867798,
      "learning_rate": 8.520373042924688e-05,
      "loss": 4.5869,
      "step": 38430
    },
    {
      "epoch": 0.7871726086867488,
      "grad_norm": 0.31189337372779846,
      "learning_rate": 8.505275035732246e-05,
      "loss": 4.5967,
      "step": 38440
    },
    {
      "epoch": 0.787377388241558,
      "grad_norm": 0.30547335743904114,
      "learning_rate": 8.49018680384934e-05,
      "loss": 4.6224,
      "step": 38450
    },
    {
      "epoch": 0.7875821677963672,
      "grad_norm": 0.31830891966819763,
      "learning_rate": 8.475108360107242e-05,
      "loss": 4.5486,
      "step": 38460
    },
    {
      "epoch": 0.7877869473511765,
      "grad_norm": 0.3292132318019867,
      "learning_rate": 8.460039717328928e-05,
      "loss": 4.5746,
      "step": 38470
    },
    {
      "epoch": 0.7879917269059857,
      "grad_norm": 0.30216705799102783,
      "learning_rate": 8.444980888329024e-05,
      "loss": 4.5959,
      "step": 38480
    },
    {
      "epoch": 0.788196506460795,
      "grad_norm": 0.31861424446105957,
      "learning_rate": 8.429931885913815e-05,
      "loss": 4.5709,
      "step": 38490
    },
    {
      "epoch": 0.7884012860156042,
      "grad_norm": 0.29862529039382935,
      "learning_rate": 8.414892722881231e-05,
      "loss": 4.5941,
      "step": 38500
    },
    {
      "epoch": 0.7886060655704135,
      "grad_norm": 0.33931490778923035,
      "learning_rate": 8.399863412020836e-05,
      "loss": 4.5432,
      "step": 38510
    },
    {
      "epoch": 0.7888108451252227,
      "grad_norm": 0.33243846893310547,
      "learning_rate": 8.3848439661138e-05,
      "loss": 4.5719,
      "step": 38520
    },
    {
      "epoch": 0.789015624680032,
      "grad_norm": 0.3161660134792328,
      "learning_rate": 8.369834397932919e-05,
      "loss": 4.6025,
      "step": 38530
    },
    {
      "epoch": 0.7892204042348412,
      "grad_norm": 0.3273294270038605,
      "learning_rate": 8.354834720242594e-05,
      "loss": 4.5742,
      "step": 38540
    },
    {
      "epoch": 0.7894251837896504,
      "grad_norm": 0.34032896161079407,
      "learning_rate": 8.339844945798793e-05,
      "loss": 4.5872,
      "step": 38550
    },
    {
      "epoch": 0.7896299633444597,
      "grad_norm": 0.30530282855033875,
      "learning_rate": 8.32486508734908e-05,
      "loss": 4.5377,
      "step": 38560
    },
    {
      "epoch": 0.7898347428992689,
      "grad_norm": 0.30698108673095703,
      "learning_rate": 8.309895157632582e-05,
      "loss": 4.5683,
      "step": 38570
    },
    {
      "epoch": 0.7900395224540782,
      "grad_norm": 0.3153245151042938,
      "learning_rate": 8.294935169379989e-05,
      "loss": 4.5721,
      "step": 38580
    },
    {
      "epoch": 0.7902443020088874,
      "grad_norm": 0.3361703157424927,
      "learning_rate": 8.279985135313514e-05,
      "loss": 4.5559,
      "step": 38590
    },
    {
      "epoch": 0.7904490815636966,
      "grad_norm": 0.37249207496643066,
      "learning_rate": 8.26504506814693e-05,
      "loss": 4.5314,
      "step": 38600
    },
    {
      "epoch": 0.7906538611185059,
      "grad_norm": 0.32031577825546265,
      "learning_rate": 8.250114980585521e-05,
      "loss": 4.5006,
      "step": 38610
    },
    {
      "epoch": 0.7908586406733151,
      "grad_norm": 0.3019481897354126,
      "learning_rate": 8.235194885326088e-05,
      "loss": 4.5888,
      "step": 38620
    },
    {
      "epoch": 0.7910634202281244,
      "grad_norm": 0.30363279581069946,
      "learning_rate": 8.220284795056943e-05,
      "loss": 4.5844,
      "step": 38630
    },
    {
      "epoch": 0.7912681997829337,
      "grad_norm": 0.317623496055603,
      "learning_rate": 8.205384722457863e-05,
      "loss": 4.596,
      "step": 38640
    },
    {
      "epoch": 0.791472979337743,
      "grad_norm": 0.31147992610931396,
      "learning_rate": 8.190494680200138e-05,
      "loss": 4.5856,
      "step": 38650
    },
    {
      "epoch": 0.7916777588925522,
      "grad_norm": 0.3023722171783447,
      "learning_rate": 8.175614680946502e-05,
      "loss": 4.5863,
      "step": 38660
    },
    {
      "epoch": 0.7918825384473615,
      "grad_norm": 0.3793337345123291,
      "learning_rate": 8.160744737351163e-05,
      "loss": 4.57,
      "step": 38670
    },
    {
      "epoch": 0.7920873180021707,
      "grad_norm": 0.3383292555809021,
      "learning_rate": 8.145884862059773e-05,
      "loss": 4.5943,
      "step": 38680
    },
    {
      "epoch": 0.7922920975569799,
      "grad_norm": 0.35156553983688354,
      "learning_rate": 8.131035067709425e-05,
      "loss": 4.5936,
      "step": 38690
    },
    {
      "epoch": 0.7924968771117892,
      "grad_norm": 0.32815688848495483,
      "learning_rate": 8.116195366928637e-05,
      "loss": 4.6,
      "step": 38700
    },
    {
      "epoch": 0.7927016566665984,
      "grad_norm": 0.32850661873817444,
      "learning_rate": 8.101365772337341e-05,
      "loss": 4.6056,
      "step": 38710
    },
    {
      "epoch": 0.7929064362214077,
      "grad_norm": 0.31346115469932556,
      "learning_rate": 8.086546296546884e-05,
      "loss": 4.5885,
      "step": 38720
    },
    {
      "epoch": 0.7931112157762169,
      "grad_norm": 0.327250212430954,
      "learning_rate": 8.071736952159988e-05,
      "loss": 4.6033,
      "step": 38730
    },
    {
      "epoch": 0.7933159953310261,
      "grad_norm": 0.31105944514274597,
      "learning_rate": 8.056937751770782e-05,
      "loss": 4.5954,
      "step": 38740
    },
    {
      "epoch": 0.7935207748858354,
      "grad_norm": 0.384827196598053,
      "learning_rate": 8.042148707964746e-05,
      "loss": 4.5761,
      "step": 38750
    },
    {
      "epoch": 0.7937255544406446,
      "grad_norm": 0.44899365305900574,
      "learning_rate": 8.027369833318739e-05,
      "loss": 4.6025,
      "step": 38760
    },
    {
      "epoch": 0.7939303339954539,
      "grad_norm": 0.30384767055511475,
      "learning_rate": 8.012601140400968e-05,
      "loss": 4.589,
      "step": 38770
    },
    {
      "epoch": 0.7941351135502631,
      "grad_norm": 0.29325801134109497,
      "learning_rate": 7.997842641770983e-05,
      "loss": 4.5984,
      "step": 38780
    },
    {
      "epoch": 0.7943398931050724,
      "grad_norm": 0.2913464903831482,
      "learning_rate": 7.983094349979666e-05,
      "loss": 4.5825,
      "step": 38790
    },
    {
      "epoch": 0.7945446726598816,
      "grad_norm": 0.3160167634487152,
      "learning_rate": 7.968356277569202e-05,
      "loss": 4.5716,
      "step": 38800
    },
    {
      "epoch": 0.7947494522146908,
      "grad_norm": 0.3500344157218933,
      "learning_rate": 7.953628437073108e-05,
      "loss": 4.589,
      "step": 38810
    },
    {
      "epoch": 0.7949542317695001,
      "grad_norm": 0.30142685770988464,
      "learning_rate": 7.938910841016185e-05,
      "loss": 4.6059,
      "step": 38820
    },
    {
      "epoch": 0.7951590113243093,
      "grad_norm": 0.33031365275382996,
      "learning_rate": 7.92420350191454e-05,
      "loss": 4.5984,
      "step": 38830
    },
    {
      "epoch": 0.7953637908791187,
      "grad_norm": 0.3248433470726013,
      "learning_rate": 7.909506432275524e-05,
      "loss": 4.5974,
      "step": 38840
    },
    {
      "epoch": 0.7955685704339279,
      "grad_norm": 0.30029308795928955,
      "learning_rate": 7.894819644597784e-05,
      "loss": 4.6273,
      "step": 38850
    },
    {
      "epoch": 0.7957733499887372,
      "grad_norm": 0.3154873549938202,
      "learning_rate": 7.880143151371225e-05,
      "loss": 4.577,
      "step": 38860
    },
    {
      "epoch": 0.7959781295435464,
      "grad_norm": 0.38484513759613037,
      "learning_rate": 7.865476965076965e-05,
      "loss": 4.5998,
      "step": 38870
    },
    {
      "epoch": 0.7961829090983557,
      "grad_norm": 0.3259485363960266,
      "learning_rate": 7.850821098187391e-05,
      "loss": 4.5566,
      "step": 38880
    },
    {
      "epoch": 0.7963876886531649,
      "grad_norm": 0.3180842995643616,
      "learning_rate": 7.836175563166099e-05,
      "loss": 4.5578,
      "step": 38890
    },
    {
      "epoch": 0.7965924682079741,
      "grad_norm": 0.30802080035209656,
      "learning_rate": 7.821540372467901e-05,
      "loss": 4.5905,
      "step": 38900
    },
    {
      "epoch": 0.7967972477627834,
      "grad_norm": 0.30659064650535583,
      "learning_rate": 7.806915538538812e-05,
      "loss": 4.5798,
      "step": 38910
    },
    {
      "epoch": 0.7970020273175926,
      "grad_norm": 0.30987635254859924,
      "learning_rate": 7.792301073816049e-05,
      "loss": 4.5848,
      "step": 38920
    },
    {
      "epoch": 0.7972068068724019,
      "grad_norm": 0.30157986283302307,
      "learning_rate": 7.777696990727992e-05,
      "loss": 4.5751,
      "step": 38930
    },
    {
      "epoch": 0.7974115864272111,
      "grad_norm": 0.3000452518463135,
      "learning_rate": 7.763103301694193e-05,
      "loss": 4.6118,
      "step": 38940
    },
    {
      "epoch": 0.7976163659820203,
      "grad_norm": 0.3065103590488434,
      "learning_rate": 7.74852001912538e-05,
      "loss": 4.5769,
      "step": 38950
    },
    {
      "epoch": 0.7978211455368296,
      "grad_norm": 0.31863853335380554,
      "learning_rate": 7.733947155423429e-05,
      "loss": 4.6016,
      "step": 38960
    },
    {
      "epoch": 0.7980259250916388,
      "grad_norm": 0.321843683719635,
      "learning_rate": 7.719384722981349e-05,
      "loss": 4.5946,
      "step": 38970
    },
    {
      "epoch": 0.7982307046464481,
      "grad_norm": 0.33706068992614746,
      "learning_rate": 7.704832734183276e-05,
      "loss": 4.5924,
      "step": 38980
    },
    {
      "epoch": 0.7984354842012573,
      "grad_norm": 0.32501912117004395,
      "learning_rate": 7.690291201404476e-05,
      "loss": 4.559,
      "step": 38990
    },
    {
      "epoch": 0.7986402637560666,
      "grad_norm": 0.3753548860549927,
      "learning_rate": 7.675760137011317e-05,
      "loss": 4.5857,
      "step": 39000
    },
    {
      "epoch": 0.7988450433108758,
      "grad_norm": 0.36469289660453796,
      "learning_rate": 7.66123955336125e-05,
      "loss": 4.5677,
      "step": 39010
    },
    {
      "epoch": 0.799049822865685,
      "grad_norm": 0.31120210886001587,
      "learning_rate": 7.646729462802842e-05,
      "loss": 4.5682,
      "step": 39020
    },
    {
      "epoch": 0.7992546024204943,
      "grad_norm": 0.320067822933197,
      "learning_rate": 7.632229877675706e-05,
      "loss": 4.5813,
      "step": 39030
    },
    {
      "epoch": 0.7994593819753036,
      "grad_norm": 0.3053686320781708,
      "learning_rate": 7.617740810310543e-05,
      "loss": 4.6089,
      "step": 39040
    },
    {
      "epoch": 0.7996641615301129,
      "grad_norm": 0.31952452659606934,
      "learning_rate": 7.603262273029099e-05,
      "loss": 4.5673,
      "step": 39050
    },
    {
      "epoch": 0.7998689410849221,
      "grad_norm": 0.8099822998046875,
      "learning_rate": 7.588794278144173e-05,
      "loss": 4.5772,
      "step": 39060
    },
    {
      "epoch": 0.8000737206397314,
      "grad_norm": 0.3223879635334015,
      "learning_rate": 7.574336837959602e-05,
      "loss": 4.6052,
      "step": 39070
    },
    {
      "epoch": 0.8002785001945406,
      "grad_norm": 0.30738234519958496,
      "learning_rate": 7.559889964770216e-05,
      "loss": 4.5682,
      "step": 39080
    },
    {
      "epoch": 0.8004832797493499,
      "grad_norm": 0.30835387110710144,
      "learning_rate": 7.545453670861897e-05,
      "loss": 4.5833,
      "step": 39090
    },
    {
      "epoch": 0.8006880593041591,
      "grad_norm": 0.29173022508621216,
      "learning_rate": 7.531027968511512e-05,
      "loss": 4.633,
      "step": 39100
    },
    {
      "epoch": 0.8008928388589683,
      "grad_norm": 0.30273181200027466,
      "learning_rate": 7.516612869986932e-05,
      "loss": 4.5468,
      "step": 39110
    },
    {
      "epoch": 0.8010976184137776,
      "grad_norm": 0.3193420171737671,
      "learning_rate": 7.502208387546986e-05,
      "loss": 4.5844,
      "step": 39120
    },
    {
      "epoch": 0.8013023979685868,
      "grad_norm": 0.29910337924957275,
      "learning_rate": 7.487814533441498e-05,
      "loss": 4.5595,
      "step": 39130
    },
    {
      "epoch": 0.8015071775233961,
      "grad_norm": 0.2982931435108185,
      "learning_rate": 7.47343131991125e-05,
      "loss": 4.5478,
      "step": 39140
    },
    {
      "epoch": 0.8017119570782053,
      "grad_norm": 0.3038698732852936,
      "learning_rate": 7.459058759187964e-05,
      "loss": 4.5984,
      "step": 39150
    },
    {
      "epoch": 0.8019167366330145,
      "grad_norm": 0.3017473816871643,
      "learning_rate": 7.444696863494314e-05,
      "loss": 4.5882,
      "step": 39160
    },
    {
      "epoch": 0.8021215161878238,
      "grad_norm": 0.39330267906188965,
      "learning_rate": 7.430345645043899e-05,
      "loss": 4.6182,
      "step": 39170
    },
    {
      "epoch": 0.802326295742633,
      "grad_norm": 0.305624783039093,
      "learning_rate": 7.416005116041241e-05,
      "loss": 4.6245,
      "step": 39180
    },
    {
      "epoch": 0.8025310752974423,
      "grad_norm": 0.33081772923469543,
      "learning_rate": 7.401675288681769e-05,
      "loss": 4.5942,
      "step": 39190
    },
    {
      "epoch": 0.8027358548522515,
      "grad_norm": 0.29383283853530884,
      "learning_rate": 7.387356175151816e-05,
      "loss": 4.595,
      "step": 39200
    },
    {
      "epoch": 0.8029406344070608,
      "grad_norm": 0.2955620288848877,
      "learning_rate": 7.37304778762859e-05,
      "loss": 4.5823,
      "step": 39210
    },
    {
      "epoch": 0.80314541396187,
      "grad_norm": 0.33134540915489197,
      "learning_rate": 7.358750138280199e-05,
      "loss": 4.568,
      "step": 39220
    },
    {
      "epoch": 0.8033501935166792,
      "grad_norm": 0.31039419770240784,
      "learning_rate": 7.344463239265595e-05,
      "loss": 4.6131,
      "step": 39230
    },
    {
      "epoch": 0.8035549730714885,
      "grad_norm": 0.3270411789417267,
      "learning_rate": 7.330187102734603e-05,
      "loss": 4.5861,
      "step": 39240
    },
    {
      "epoch": 0.8037597526262978,
      "grad_norm": 0.3229215443134308,
      "learning_rate": 7.315921740827896e-05,
      "loss": 4.5928,
      "step": 39250
    },
    {
      "epoch": 0.8039645321811071,
      "grad_norm": 0.33217817544937134,
      "learning_rate": 7.301667165676977e-05,
      "loss": 4.5854,
      "step": 39260
    },
    {
      "epoch": 0.8041693117359163,
      "grad_norm": 0.3282391130924225,
      "learning_rate": 7.28742338940418e-05,
      "loss": 4.5349,
      "step": 39270
    },
    {
      "epoch": 0.8043740912907256,
      "grad_norm": 0.2931309938430786,
      "learning_rate": 7.273190424122665e-05,
      "loss": 4.5938,
      "step": 39280
    },
    {
      "epoch": 0.8045788708455348,
      "grad_norm": 0.3457520008087158,
      "learning_rate": 7.258968281936367e-05,
      "loss": 4.5426,
      "step": 39290
    },
    {
      "epoch": 0.804783650400344,
      "grad_norm": 0.5809755921363831,
      "learning_rate": 7.24475697494005e-05,
      "loss": 4.5743,
      "step": 39300
    },
    {
      "epoch": 0.8049884299551533,
      "grad_norm": 0.30670252442359924,
      "learning_rate": 7.230556515219257e-05,
      "loss": 4.5955,
      "step": 39310
    },
    {
      "epoch": 0.8051932095099625,
      "grad_norm": 0.30223187804222107,
      "learning_rate": 7.216366914850285e-05,
      "loss": 4.5722,
      "step": 39320
    },
    {
      "epoch": 0.8053979890647718,
      "grad_norm": 0.29483914375305176,
      "learning_rate": 7.202188185900221e-05,
      "loss": 4.5918,
      "step": 39330
    },
    {
      "epoch": 0.805602768619581,
      "grad_norm": 0.2921166718006134,
      "learning_rate": 7.188020340426893e-05,
      "loss": 4.5968,
      "step": 39340
    },
    {
      "epoch": 0.8058075481743903,
      "grad_norm": 0.3002696633338928,
      "learning_rate": 7.173863390478887e-05,
      "loss": 4.571,
      "step": 39350
    },
    {
      "epoch": 0.8060123277291995,
      "grad_norm": 0.29883912205696106,
      "learning_rate": 7.159717348095501e-05,
      "loss": 4.5579,
      "step": 39360
    },
    {
      "epoch": 0.8062171072840087,
      "grad_norm": 0.29541438817977905,
      "learning_rate": 7.145582225306774e-05,
      "loss": 4.5117,
      "step": 39370
    },
    {
      "epoch": 0.806421886838818,
      "grad_norm": 0.32466834783554077,
      "learning_rate": 7.131458034133457e-05,
      "loss": 4.602,
      "step": 39380
    },
    {
      "epoch": 0.8066266663936272,
      "grad_norm": 0.2947414219379425,
      "learning_rate": 7.117344786587003e-05,
      "loss": 4.5947,
      "step": 39390
    },
    {
      "epoch": 0.8068314459484365,
      "grad_norm": 0.3411257565021515,
      "learning_rate": 7.103242494669562e-05,
      "loss": 4.5912,
      "step": 39400
    },
    {
      "epoch": 0.8070362255032457,
      "grad_norm": 0.31953033804893494,
      "learning_rate": 7.089151170373952e-05,
      "loss": 4.5708,
      "step": 39410
    },
    {
      "epoch": 0.807241005058055,
      "grad_norm": 0.32291150093078613,
      "learning_rate": 7.075070825683684e-05,
      "loss": 4.5931,
      "step": 39420
    },
    {
      "epoch": 0.8074457846128642,
      "grad_norm": 0.31212931871414185,
      "learning_rate": 7.061001472572914e-05,
      "loss": 4.5684,
      "step": 39430
    },
    {
      "epoch": 0.8076505641676734,
      "grad_norm": 0.30312198400497437,
      "learning_rate": 7.046943123006464e-05,
      "loss": 4.6063,
      "step": 39440
    },
    {
      "epoch": 0.8078553437224828,
      "grad_norm": 0.2988637685775757,
      "learning_rate": 7.032895788939791e-05,
      "loss": 4.562,
      "step": 39450
    },
    {
      "epoch": 0.808060123277292,
      "grad_norm": 0.3108832538127899,
      "learning_rate": 7.018859482318989e-05,
      "loss": 4.5668,
      "step": 39460
    },
    {
      "epoch": 0.8082649028321013,
      "grad_norm": 0.30393844842910767,
      "learning_rate": 7.004834215080773e-05,
      "loss": 4.59,
      "step": 39470
    },
    {
      "epoch": 0.8084696823869105,
      "grad_norm": 0.3082975745201111,
      "learning_rate": 6.990819999152467e-05,
      "loss": 4.5609,
      "step": 39480
    },
    {
      "epoch": 0.8086744619417198,
      "grad_norm": 0.34625446796417236,
      "learning_rate": 6.976816846452003e-05,
      "loss": 4.5768,
      "step": 39490
    },
    {
      "epoch": 0.808879241496529,
      "grad_norm": 0.2989501953125,
      "learning_rate": 6.96282476888789e-05,
      "loss": 4.5678,
      "step": 39500
    },
    {
      "epoch": 0.8090840210513383,
      "grad_norm": 0.3937240540981293,
      "learning_rate": 6.948843778359241e-05,
      "loss": 4.567,
      "step": 39510
    },
    {
      "epoch": 0.8092888006061475,
      "grad_norm": 0.32027167081832886,
      "learning_rate": 6.934873886755717e-05,
      "loss": 4.5707,
      "step": 39520
    },
    {
      "epoch": 0.8094935801609567,
      "grad_norm": 0.3013392984867096,
      "learning_rate": 6.920915105957555e-05,
      "loss": 4.5628,
      "step": 39530
    },
    {
      "epoch": 0.809698359715766,
      "grad_norm": 0.3077249825000763,
      "learning_rate": 6.90696744783554e-05,
      "loss": 4.5951,
      "step": 39540
    },
    {
      "epoch": 0.8099031392705752,
      "grad_norm": 0.3117411434650421,
      "learning_rate": 6.893030924250996e-05,
      "loss": 4.56,
      "step": 39550
    },
    {
      "epoch": 0.8101079188253845,
      "grad_norm": 0.29708433151245117,
      "learning_rate": 6.879105547055789e-05,
      "loss": 4.5894,
      "step": 39560
    },
    {
      "epoch": 0.8103126983801937,
      "grad_norm": 0.29647096991539,
      "learning_rate": 6.865191328092282e-05,
      "loss": 4.5692,
      "step": 39570
    },
    {
      "epoch": 0.810517477935003,
      "grad_norm": 0.2984810471534729,
      "learning_rate": 6.851288279193371e-05,
      "loss": 4.5446,
      "step": 39580
    },
    {
      "epoch": 0.8107222574898122,
      "grad_norm": 0.30566325783729553,
      "learning_rate": 6.837396412182444e-05,
      "loss": 4.5553,
      "step": 39590
    },
    {
      "epoch": 0.8109270370446214,
      "grad_norm": 0.31332987546920776,
      "learning_rate": 6.82351573887339e-05,
      "loss": 4.5493,
      "step": 39600
    },
    {
      "epoch": 0.8111318165994307,
      "grad_norm": 0.34441831707954407,
      "learning_rate": 6.80964627107055e-05,
      "loss": 4.5772,
      "step": 39610
    },
    {
      "epoch": 0.8113365961542399,
      "grad_norm": 0.29685527086257935,
      "learning_rate": 6.79578802056877e-05,
      "loss": 4.628,
      "step": 39620
    },
    {
      "epoch": 0.8115413757090492,
      "grad_norm": 0.30922380089759827,
      "learning_rate": 6.781940999153345e-05,
      "loss": 4.615,
      "step": 39630
    },
    {
      "epoch": 0.8117461552638584,
      "grad_norm": 0.3546513617038727,
      "learning_rate": 6.7681052186e-05,
      "loss": 4.5814,
      "step": 39640
    },
    {
      "epoch": 0.8119509348186678,
      "grad_norm": 0.3195394277572632,
      "learning_rate": 6.75428069067493e-05,
      "loss": 4.56,
      "step": 39650
    },
    {
      "epoch": 0.812155714373477,
      "grad_norm": 0.3146194815635681,
      "learning_rate": 6.740467427134746e-05,
      "loss": 4.5716,
      "step": 39660
    },
    {
      "epoch": 0.8123604939282862,
      "grad_norm": 0.5386914610862732,
      "learning_rate": 6.726665439726485e-05,
      "loss": 4.5808,
      "step": 39670
    },
    {
      "epoch": 0.8125652734830955,
      "grad_norm": 0.31036967039108276,
      "learning_rate": 6.712874740187589e-05,
      "loss": 4.5571,
      "step": 39680
    },
    {
      "epoch": 0.8127700530379047,
      "grad_norm": 0.3137417435646057,
      "learning_rate": 6.69909534024591e-05,
      "loss": 4.5759,
      "step": 39690
    },
    {
      "epoch": 0.812974832592714,
      "grad_norm": 0.3191232979297638,
      "learning_rate": 6.68532725161968e-05,
      "loss": 4.6141,
      "step": 39700
    },
    {
      "epoch": 0.8131796121475232,
      "grad_norm": 0.30594417452812195,
      "learning_rate": 6.671570486017507e-05,
      "loss": 4.5617,
      "step": 39710
    },
    {
      "epoch": 0.8133843917023325,
      "grad_norm": 0.31159308552742004,
      "learning_rate": 6.657825055138382e-05,
      "loss": 4.5892,
      "step": 39720
    },
    {
      "epoch": 0.8135891712571417,
      "grad_norm": 0.37162676453590393,
      "learning_rate": 6.64409097067166e-05,
      "loss": 4.5883,
      "step": 39730
    },
    {
      "epoch": 0.8137939508119509,
      "grad_norm": 0.2946584224700928,
      "learning_rate": 6.630368244297034e-05,
      "loss": 4.5841,
      "step": 39740
    },
    {
      "epoch": 0.8139987303667602,
      "grad_norm": 0.3434107303619385,
      "learning_rate": 6.616656887684547e-05,
      "loss": 4.5739,
      "step": 39750
    },
    {
      "epoch": 0.8142035099215694,
      "grad_norm": 0.3184031844139099,
      "learning_rate": 6.602956912494567e-05,
      "loss": 4.5956,
      "step": 39760
    },
    {
      "epoch": 0.8144082894763787,
      "grad_norm": 0.2952013909816742,
      "learning_rate": 6.589268330377794e-05,
      "loss": 4.5757,
      "step": 39770
    },
    {
      "epoch": 0.8146130690311879,
      "grad_norm": 0.31298646330833435,
      "learning_rate": 6.57559115297522e-05,
      "loss": 4.5678,
      "step": 39780
    },
    {
      "epoch": 0.8148178485859972,
      "grad_norm": 0.3092459440231323,
      "learning_rate": 6.561925391918158e-05,
      "loss": 4.6025,
      "step": 39790
    },
    {
      "epoch": 0.8150226281408064,
      "grad_norm": 0.3376334607601166,
      "learning_rate": 6.548271058828195e-05,
      "loss": 4.5802,
      "step": 39800
    },
    {
      "epoch": 0.8152274076956156,
      "grad_norm": 0.5934453010559082,
      "learning_rate": 6.534628165317209e-05,
      "loss": 4.4972,
      "step": 39810
    },
    {
      "epoch": 0.8154321872504249,
      "grad_norm": 0.31874537467956543,
      "learning_rate": 6.520996722987354e-05,
      "loss": 4.5606,
      "step": 39820
    },
    {
      "epoch": 0.8156369668052341,
      "grad_norm": 0.3155401945114136,
      "learning_rate": 6.507376743431037e-05,
      "loss": 4.5875,
      "step": 39830
    },
    {
      "epoch": 0.8158417463600434,
      "grad_norm": 0.327118843793869,
      "learning_rate": 6.493768238230929e-05,
      "loss": 4.576,
      "step": 39840
    },
    {
      "epoch": 0.8160465259148527,
      "grad_norm": 0.3194453716278076,
      "learning_rate": 6.480171218959918e-05,
      "loss": 4.5893,
      "step": 39850
    },
    {
      "epoch": 0.816251305469662,
      "grad_norm": 0.3144104480743408,
      "learning_rate": 6.466585697181148e-05,
      "loss": 4.5746,
      "step": 39860
    },
    {
      "epoch": 0.8164560850244712,
      "grad_norm": 0.31391122937202454,
      "learning_rate": 6.453011684447983e-05,
      "loss": 4.5492,
      "step": 39870
    },
    {
      "epoch": 0.8166608645792804,
      "grad_norm": 0.3237147927284241,
      "learning_rate": 6.439449192303992e-05,
      "loss": 4.5762,
      "step": 39880
    },
    {
      "epoch": 0.8168656441340897,
      "grad_norm": 0.3276256322860718,
      "learning_rate": 6.425898232282941e-05,
      "loss": 4.5694,
      "step": 39890
    },
    {
      "epoch": 0.8170704236888989,
      "grad_norm": 0.30594319105148315,
      "learning_rate": 6.412358815908801e-05,
      "loss": 4.5978,
      "step": 39900
    },
    {
      "epoch": 0.8172752032437082,
      "grad_norm": 0.29324057698249817,
      "learning_rate": 6.398830954695732e-05,
      "loss": 4.5764,
      "step": 39910
    },
    {
      "epoch": 0.8174799827985174,
      "grad_norm": 0.319190114736557,
      "learning_rate": 6.385314660148041e-05,
      "loss": 4.564,
      "step": 39920
    },
    {
      "epoch": 0.8176847623533267,
      "grad_norm": 0.31613823771476746,
      "learning_rate": 6.371809943760221e-05,
      "loss": 4.5646,
      "step": 39930
    },
    {
      "epoch": 0.8178895419081359,
      "grad_norm": 0.3119392991065979,
      "learning_rate": 6.358316817016913e-05,
      "loss": 4.5953,
      "step": 39940
    },
    {
      "epoch": 0.8180943214629451,
      "grad_norm": 0.3238719403743744,
      "learning_rate": 6.344835291392903e-05,
      "loss": 4.6055,
      "step": 39950
    },
    {
      "epoch": 0.8182991010177544,
      "grad_norm": 0.3044256567955017,
      "learning_rate": 6.331365378353109e-05,
      "loss": 4.5751,
      "step": 39960
    },
    {
      "epoch": 0.8185038805725636,
      "grad_norm": 0.3040337860584259,
      "learning_rate": 6.317907089352579e-05,
      "loss": 4.5915,
      "step": 39970
    },
    {
      "epoch": 0.8187086601273729,
      "grad_norm": 0.33757469058036804,
      "learning_rate": 6.304460435836458e-05,
      "loss": 4.5909,
      "step": 39980
    },
    {
      "epoch": 0.8189134396821821,
      "grad_norm": 0.3282129764556885,
      "learning_rate": 6.291025429240023e-05,
      "loss": 4.5882,
      "step": 39990
    },
    {
      "epoch": 0.8191182192369914,
      "grad_norm": 0.3238188326358795,
      "learning_rate": 6.277602080988619e-05,
      "loss": 4.5796,
      "step": 40000
    },
    {
      "epoch": 0.8191182192369914,
      "eval_loss": 4.575884819030762,
      "eval_runtime": 4.2712,
      "eval_samples_per_second": 272.991,
      "eval_steps_per_second": 34.182,
      "step": 40000
    },
    {
      "epoch": 0.8193229987918006,
      "grad_norm": 0.3096143901348114,
      "learning_rate": 6.264190402497693e-05,
      "loss": 4.5938,
      "step": 40010
    },
    {
      "epoch": 0.8195277783466098,
      "grad_norm": 0.3325948417186737,
      "learning_rate": 6.25079040517277e-05,
      "loss": 4.5276,
      "step": 40020
    },
    {
      "epoch": 0.8197325579014191,
      "grad_norm": 0.34610995650291443,
      "learning_rate": 6.237402100409431e-05,
      "loss": 4.5836,
      "step": 40030
    },
    {
      "epoch": 0.8199373374562283,
      "grad_norm": 0.30577176809310913,
      "learning_rate": 6.224025499593318e-05,
      "loss": 4.611,
      "step": 40040
    },
    {
      "epoch": 0.8201421170110376,
      "grad_norm": 0.3057403266429901,
      "learning_rate": 6.21066061410013e-05,
      "loss": 4.5841,
      "step": 40050
    },
    {
      "epoch": 0.8203468965658469,
      "grad_norm": 0.31216299533843994,
      "learning_rate": 6.197307455295575e-05,
      "loss": 4.5334,
      "step": 40060
    },
    {
      "epoch": 0.8205516761206562,
      "grad_norm": 0.3139328062534332,
      "learning_rate": 6.183966034535418e-05,
      "loss": 4.5958,
      "step": 40070
    },
    {
      "epoch": 0.8207564556754654,
      "grad_norm": 0.2837695777416229,
      "learning_rate": 6.170636363165432e-05,
      "loss": 4.5757,
      "step": 40080
    },
    {
      "epoch": 0.8209612352302746,
      "grad_norm": 0.30150434374809265,
      "learning_rate": 6.157318452521385e-05,
      "loss": 4.5938,
      "step": 40090
    },
    {
      "epoch": 0.8211660147850839,
      "grad_norm": 0.3371080458164215,
      "learning_rate": 6.14401231392906e-05,
      "loss": 4.562,
      "step": 40100
    },
    {
      "epoch": 0.8213707943398931,
      "grad_norm": 0.3148494064807892,
      "learning_rate": 6.130717958704226e-05,
      "loss": 4.5811,
      "step": 40110
    },
    {
      "epoch": 0.8215755738947024,
      "grad_norm": 0.3045748770236969,
      "learning_rate": 6.117435398152634e-05,
      "loss": 4.5936,
      "step": 40120
    },
    {
      "epoch": 0.8217803534495116,
      "grad_norm": 0.2946961522102356,
      "learning_rate": 6.104164643569985e-05,
      "loss": 4.5752,
      "step": 40130
    },
    {
      "epoch": 0.8219851330043209,
      "grad_norm": 0.30119648575782776,
      "learning_rate": 6.0909057062419604e-05,
      "loss": 4.5784,
      "step": 40140
    },
    {
      "epoch": 0.8221899125591301,
      "grad_norm": 0.32062435150146484,
      "learning_rate": 6.077658597444189e-05,
      "loss": 4.6356,
      "step": 40150
    },
    {
      "epoch": 0.8223946921139393,
      "grad_norm": 0.34336739778518677,
      "learning_rate": 6.064423328442237e-05,
      "loss": 4.5859,
      "step": 40160
    },
    {
      "epoch": 0.8225994716687486,
      "grad_norm": 0.3080267906188965,
      "learning_rate": 6.051199910491605e-05,
      "loss": 4.5982,
      "step": 40170
    },
    {
      "epoch": 0.8228042512235578,
      "grad_norm": 0.29620522260665894,
      "learning_rate": 6.0379883548377026e-05,
      "loss": 4.6064,
      "step": 40180
    },
    {
      "epoch": 0.8230090307783671,
      "grad_norm": 0.29631346464157104,
      "learning_rate": 6.024788672715873e-05,
      "loss": 4.5781,
      "step": 40190
    },
    {
      "epoch": 0.8232138103331763,
      "grad_norm": 0.3337494134902954,
      "learning_rate": 6.0116008753513375e-05,
      "loss": 4.5816,
      "step": 40200
    },
    {
      "epoch": 0.8234185898879856,
      "grad_norm": 0.3190304934978485,
      "learning_rate": 5.9984249739592315e-05,
      "loss": 4.6157,
      "step": 40210
    },
    {
      "epoch": 0.8236233694427948,
      "grad_norm": 0.31650981307029724,
      "learning_rate": 5.985260979744563e-05,
      "loss": 4.5511,
      "step": 40220
    },
    {
      "epoch": 0.823828148997604,
      "grad_norm": 0.30655616521835327,
      "learning_rate": 5.972108903902218e-05,
      "loss": 4.5511,
      "step": 40230
    },
    {
      "epoch": 0.8240329285524133,
      "grad_norm": 0.31079205870628357,
      "learning_rate": 5.958968757616943e-05,
      "loss": 4.5771,
      "step": 40240
    },
    {
      "epoch": 0.8242377081072225,
      "grad_norm": 0.31540369987487793,
      "learning_rate": 5.945840552063346e-05,
      "loss": 4.5963,
      "step": 40250
    },
    {
      "epoch": 0.8244424876620319,
      "grad_norm": 0.31257107853889465,
      "learning_rate": 5.932724298405879e-05,
      "loss": 4.5456,
      "step": 40260
    },
    {
      "epoch": 0.8246472672168411,
      "grad_norm": 0.32176050543785095,
      "learning_rate": 5.919620007798823e-05,
      "loss": 4.6102,
      "step": 40270
    },
    {
      "epoch": 0.8248520467716504,
      "grad_norm": 0.2959055006504059,
      "learning_rate": 5.906527691386279e-05,
      "loss": 4.5573,
      "step": 40280
    },
    {
      "epoch": 0.8250568263264596,
      "grad_norm": 0.30281054973602295,
      "learning_rate": 5.893447360302189e-05,
      "loss": 4.5637,
      "step": 40290
    },
    {
      "epoch": 0.8252616058812688,
      "grad_norm": 0.2997758984565735,
      "learning_rate": 5.8803790256702815e-05,
      "loss": 4.5415,
      "step": 40300
    },
    {
      "epoch": 0.8254663854360781,
      "grad_norm": 0.3132002651691437,
      "learning_rate": 5.8673226986040944e-05,
      "loss": 4.6224,
      "step": 40310
    },
    {
      "epoch": 0.8256711649908873,
      "grad_norm": 0.3072299063205719,
      "learning_rate": 5.854278390206951e-05,
      "loss": 4.564,
      "step": 40320
    },
    {
      "epoch": 0.8258759445456966,
      "grad_norm": 0.30027976632118225,
      "learning_rate": 5.8412461115719565e-05,
      "loss": 4.5962,
      "step": 40330
    },
    {
      "epoch": 0.8260807241005058,
      "grad_norm": 0.29778626561164856,
      "learning_rate": 5.828225873781972e-05,
      "loss": 4.5576,
      "step": 40340
    },
    {
      "epoch": 0.8262855036553151,
      "grad_norm": 0.30664029717445374,
      "learning_rate": 5.815217687909633e-05,
      "loss": 4.4757,
      "step": 40350
    },
    {
      "epoch": 0.8264902832101243,
      "grad_norm": 0.3251498341560364,
      "learning_rate": 5.80222156501733e-05,
      "loss": 4.5965,
      "step": 40360
    },
    {
      "epoch": 0.8266950627649335,
      "grad_norm": 0.3085460960865021,
      "learning_rate": 5.789237516157173e-05,
      "loss": 4.5657,
      "step": 40370
    },
    {
      "epoch": 0.8268998423197428,
      "grad_norm": 0.2986665666103363,
      "learning_rate": 5.776265552371027e-05,
      "loss": 4.5658,
      "step": 40380
    },
    {
      "epoch": 0.827104621874552,
      "grad_norm": 0.3298749327659607,
      "learning_rate": 5.7633056846904674e-05,
      "loss": 4.5702,
      "step": 40390
    },
    {
      "epoch": 0.8273094014293613,
      "grad_norm": 0.29386669397354126,
      "learning_rate": 5.750357924136793e-05,
      "loss": 4.5646,
      "step": 40400
    },
    {
      "epoch": 0.8275141809841705,
      "grad_norm": 0.3200646936893463,
      "learning_rate": 5.7374222817209877e-05,
      "loss": 4.539,
      "step": 40410
    },
    {
      "epoch": 0.8277189605389798,
      "grad_norm": 0.32497406005859375,
      "learning_rate": 5.724498768443747e-05,
      "loss": 4.5672,
      "step": 40420
    },
    {
      "epoch": 0.827923740093789,
      "grad_norm": 0.3958125114440918,
      "learning_rate": 5.711587395295446e-05,
      "loss": 4.6014,
      "step": 40430
    },
    {
      "epoch": 0.8281285196485982,
      "grad_norm": 0.30142369866371155,
      "learning_rate": 5.6986881732561394e-05,
      "loss": 4.5771,
      "step": 40440
    },
    {
      "epoch": 0.8283332992034075,
      "grad_norm": 0.3324086666107178,
      "learning_rate": 5.6858011132955455e-05,
      "loss": 4.5214,
      "step": 40450
    },
    {
      "epoch": 0.8285380787582168,
      "grad_norm": 0.3738360106945038,
      "learning_rate": 5.6729262263730324e-05,
      "loss": 4.5905,
      "step": 40460
    },
    {
      "epoch": 0.8287428583130261,
      "grad_norm": 0.3204920291900635,
      "learning_rate": 5.660063523437631e-05,
      "loss": 4.5946,
      "step": 40470
    },
    {
      "epoch": 0.8289476378678353,
      "grad_norm": 0.3194018602371216,
      "learning_rate": 5.647213015427992e-05,
      "loss": 4.5686,
      "step": 40480
    },
    {
      "epoch": 0.8291524174226446,
      "grad_norm": 0.3234045207500458,
      "learning_rate": 5.634374713272412e-05,
      "loss": 4.6237,
      "step": 40490
    },
    {
      "epoch": 0.8293571969774538,
      "grad_norm": 0.3342304229736328,
      "learning_rate": 5.6215486278888016e-05,
      "loss": 4.5502,
      "step": 40500
    },
    {
      "epoch": 0.829561976532263,
      "grad_norm": 0.31595751643180847,
      "learning_rate": 5.608734770184681e-05,
      "loss": 4.5497,
      "step": 40510
    },
    {
      "epoch": 0.8297667560870723,
      "grad_norm": 0.3042425215244293,
      "learning_rate": 5.595933151057173e-05,
      "loss": 4.5548,
      "step": 40520
    },
    {
      "epoch": 0.8299715356418815,
      "grad_norm": 0.3285609185695648,
      "learning_rate": 5.583143781392994e-05,
      "loss": 4.545,
      "step": 40530
    },
    {
      "epoch": 0.8301763151966908,
      "grad_norm": 0.30174845457077026,
      "learning_rate": 5.570366672068443e-05,
      "loss": 4.5705,
      "step": 40540
    },
    {
      "epoch": 0.8303810947515,
      "grad_norm": 0.3143061697483063,
      "learning_rate": 5.557601833949384e-05,
      "loss": 4.5473,
      "step": 40550
    },
    {
      "epoch": 0.8305858743063093,
      "grad_norm": 0.3141131103038788,
      "learning_rate": 5.544849277891262e-05,
      "loss": 4.5631,
      "step": 40560
    },
    {
      "epoch": 0.8307906538611185,
      "grad_norm": 0.29421284794807434,
      "learning_rate": 5.532109014739053e-05,
      "loss": 4.5634,
      "step": 40570
    },
    {
      "epoch": 0.8309954334159277,
      "grad_norm": 0.3055809438228607,
      "learning_rate": 5.519381055327304e-05,
      "loss": 4.5718,
      "step": 40580
    },
    {
      "epoch": 0.831200212970737,
      "grad_norm": 0.3031426668167114,
      "learning_rate": 5.506665410480085e-05,
      "loss": 4.5831,
      "step": 40590
    },
    {
      "epoch": 0.8314049925255462,
      "grad_norm": 0.29634007811546326,
      "learning_rate": 5.493962091010998e-05,
      "loss": 4.5598,
      "step": 40600
    },
    {
      "epoch": 0.8316097720803555,
      "grad_norm": 0.3001534342765808,
      "learning_rate": 5.481271107723167e-05,
      "loss": 4.5927,
      "step": 40610
    },
    {
      "epoch": 0.8318145516351647,
      "grad_norm": 0.3114277422428131,
      "learning_rate": 5.4685924714092105e-05,
      "loss": 4.5577,
      "step": 40620
    },
    {
      "epoch": 0.832019331189974,
      "grad_norm": 0.33328965306282043,
      "learning_rate": 5.4559261928512575e-05,
      "loss": 4.5839,
      "step": 40630
    },
    {
      "epoch": 0.8322241107447832,
      "grad_norm": 0.30717888474464417,
      "learning_rate": 5.4432722828209305e-05,
      "loss": 4.5541,
      "step": 40640
    },
    {
      "epoch": 0.8324288902995924,
      "grad_norm": 0.2950066328048706,
      "learning_rate": 5.430630752079333e-05,
      "loss": 4.5491,
      "step": 40650
    },
    {
      "epoch": 0.8326336698544017,
      "grad_norm": 0.32168129086494446,
      "learning_rate": 5.4180016113770284e-05,
      "loss": 4.5931,
      "step": 40660
    },
    {
      "epoch": 0.832838449409211,
      "grad_norm": 0.36206507682800293,
      "learning_rate": 5.405384871454058e-05,
      "loss": 4.5691,
      "step": 40670
    },
    {
      "epoch": 0.8330432289640203,
      "grad_norm": 0.298918753862381,
      "learning_rate": 5.39278054303991e-05,
      "loss": 4.6016,
      "step": 40680
    },
    {
      "epoch": 0.8332480085188295,
      "grad_norm": 0.2934432327747345,
      "learning_rate": 5.380188636853527e-05,
      "loss": 4.5834,
      "step": 40690
    },
    {
      "epoch": 0.8334527880736388,
      "grad_norm": 0.30298522114753723,
      "learning_rate": 5.367609163603269e-05,
      "loss": 4.562,
      "step": 40700
    },
    {
      "epoch": 0.833657567628448,
      "grad_norm": 0.3066384792327881,
      "learning_rate": 5.355042133986938e-05,
      "loss": 4.5785,
      "step": 40710
    },
    {
      "epoch": 0.8338623471832572,
      "grad_norm": 0.31653040647506714,
      "learning_rate": 5.3424875586917514e-05,
      "loss": 4.5656,
      "step": 40720
    },
    {
      "epoch": 0.8340671267380665,
      "grad_norm": 0.32101601362228394,
      "learning_rate": 5.329945448394331e-05,
      "loss": 4.5812,
      "step": 40730
    },
    {
      "epoch": 0.8342719062928757,
      "grad_norm": 0.3386152386665344,
      "learning_rate": 5.317415813760709e-05,
      "loss": 4.5628,
      "step": 40740
    },
    {
      "epoch": 0.834476685847685,
      "grad_norm": 0.3049459159374237,
      "learning_rate": 5.3048986654462875e-05,
      "loss": 4.6111,
      "step": 40750
    },
    {
      "epoch": 0.8346814654024942,
      "grad_norm": 0.29170680046081543,
      "learning_rate": 5.2923940140958716e-05,
      "loss": 4.6017,
      "step": 40760
    },
    {
      "epoch": 0.8348862449573035,
      "grad_norm": 0.3085028827190399,
      "learning_rate": 5.279901870343622e-05,
      "loss": 4.5733,
      "step": 40770
    },
    {
      "epoch": 0.8350910245121127,
      "grad_norm": 0.2996695935726166,
      "learning_rate": 5.2674222448130714e-05,
      "loss": 4.6044,
      "step": 40780
    },
    {
      "epoch": 0.8352958040669219,
      "grad_norm": 0.3270297646522522,
      "learning_rate": 5.254955148117109e-05,
      "loss": 4.6011,
      "step": 40790
    },
    {
      "epoch": 0.8355005836217312,
      "grad_norm": 0.303653359413147,
      "learning_rate": 5.2425005908579636e-05,
      "loss": 4.5859,
      "step": 40800
    },
    {
      "epoch": 0.8357053631765404,
      "grad_norm": 0.31010833382606506,
      "learning_rate": 5.230058583627204e-05,
      "loss": 4.5363,
      "step": 40810
    },
    {
      "epoch": 0.8359101427313497,
      "grad_norm": 0.32453277707099915,
      "learning_rate": 5.217629137005726e-05,
      "loss": 4.5699,
      "step": 40820
    },
    {
      "epoch": 0.8361149222861589,
      "grad_norm": 0.3030214309692383,
      "learning_rate": 5.2052122615637433e-05,
      "loss": 4.5729,
      "step": 40830
    },
    {
      "epoch": 0.8363197018409682,
      "grad_norm": 0.30498307943344116,
      "learning_rate": 5.192807967860769e-05,
      "loss": 4.6002,
      "step": 40840
    },
    {
      "epoch": 0.8365244813957774,
      "grad_norm": 0.2962140142917633,
      "learning_rate": 5.180416266445638e-05,
      "loss": 4.5837,
      "step": 40850
    },
    {
      "epoch": 0.8367292609505866,
      "grad_norm": 0.2984965145587921,
      "learning_rate": 5.168037167856452e-05,
      "loss": 4.5627,
      "step": 40860
    },
    {
      "epoch": 0.836934040505396,
      "grad_norm": 0.32174327969551086,
      "learning_rate": 5.1556706826206105e-05,
      "loss": 4.5712,
      "step": 40870
    },
    {
      "epoch": 0.8371388200602052,
      "grad_norm": 0.3211826980113983,
      "learning_rate": 5.143316821254784e-05,
      "loss": 4.587,
      "step": 40880
    },
    {
      "epoch": 0.8373435996150145,
      "grad_norm": 0.3006008267402649,
      "learning_rate": 5.130975594264909e-05,
      "loss": 4.5925,
      "step": 40890
    },
    {
      "epoch": 0.8375483791698237,
      "grad_norm": 0.3336687386035919,
      "learning_rate": 5.118647012146176e-05,
      "loss": 4.5807,
      "step": 40900
    },
    {
      "epoch": 0.837753158724633,
      "grad_norm": 0.2978768050670624,
      "learning_rate": 5.106331085383009e-05,
      "loss": 4.5507,
      "step": 40910
    },
    {
      "epoch": 0.8379579382794422,
      "grad_norm": 0.30832746624946594,
      "learning_rate": 5.094027824449092e-05,
      "loss": 4.5604,
      "step": 40920
    },
    {
      "epoch": 0.8381627178342514,
      "grad_norm": 0.3506377339363098,
      "learning_rate": 5.0817372398073246e-05,
      "loss": 4.5421,
      "step": 40930
    },
    {
      "epoch": 0.8383674973890607,
      "grad_norm": 0.30899617075920105,
      "learning_rate": 5.069459341909837e-05,
      "loss": 4.5737,
      "step": 40940
    },
    {
      "epoch": 0.8385722769438699,
      "grad_norm": 0.3059753179550171,
      "learning_rate": 5.057194141197945e-05,
      "loss": 4.5215,
      "step": 40950
    },
    {
      "epoch": 0.8387770564986792,
      "grad_norm": 0.40165573358535767,
      "learning_rate": 5.044941648102195e-05,
      "loss": 4.5892,
      "step": 40960
    },
    {
      "epoch": 0.8389818360534884,
      "grad_norm": 0.3088058829307556,
      "learning_rate": 5.0327018730423206e-05,
      "loss": 4.5868,
      "step": 40970
    },
    {
      "epoch": 0.8391866156082977,
      "grad_norm": 0.31243231892585754,
      "learning_rate": 5.0204748264272194e-05,
      "loss": 4.5605,
      "step": 40980
    },
    {
      "epoch": 0.8393913951631069,
      "grad_norm": 0.309691458940506,
      "learning_rate": 5.008260518654988e-05,
      "loss": 4.578,
      "step": 40990
    },
    {
      "epoch": 0.8395961747179161,
      "grad_norm": 0.34093743562698364,
      "learning_rate": 4.996058960112884e-05,
      "loss": 4.5843,
      "step": 41000
    },
    {
      "epoch": 0.8398009542727254,
      "grad_norm": 0.31313571333885193,
      "learning_rate": 4.9838701611773155e-05,
      "loss": 4.5593,
      "step": 41010
    },
    {
      "epoch": 0.8400057338275346,
      "grad_norm": 0.35509222745895386,
      "learning_rate": 4.9716941322138465e-05,
      "loss": 4.5816,
      "step": 41020
    },
    {
      "epoch": 0.8402105133823439,
      "grad_norm": 0.32133156061172485,
      "learning_rate": 4.959530883577184e-05,
      "loss": 4.5616,
      "step": 41030
    },
    {
      "epoch": 0.8404152929371531,
      "grad_norm": 0.31465378403663635,
      "learning_rate": 4.9473804256111563e-05,
      "loss": 4.5598,
      "step": 41040
    },
    {
      "epoch": 0.8406200724919624,
      "grad_norm": 0.32268306612968445,
      "learning_rate": 4.9352427686487155e-05,
      "loss": 4.6018,
      "step": 41050
    },
    {
      "epoch": 0.8408248520467716,
      "grad_norm": 0.3060460686683655,
      "learning_rate": 4.923117923011935e-05,
      "loss": 4.53,
      "step": 41060
    },
    {
      "epoch": 0.841029631601581,
      "grad_norm": 0.3377113938331604,
      "learning_rate": 4.9110058990119886e-05,
      "loss": 4.5621,
      "step": 41070
    },
    {
      "epoch": 0.8412344111563902,
      "grad_norm": 0.30276891589164734,
      "learning_rate": 4.898906706949151e-05,
      "loss": 4.5804,
      "step": 41080
    },
    {
      "epoch": 0.8414391907111994,
      "grad_norm": 0.3168056905269623,
      "learning_rate": 4.8868203571127756e-05,
      "loss": 4.5622,
      "step": 41090
    },
    {
      "epoch": 0.8416439702660087,
      "grad_norm": 0.3039287030696869,
      "learning_rate": 4.874746859781305e-05,
      "loss": 4.5951,
      "step": 41100
    },
    {
      "epoch": 0.8418487498208179,
      "grad_norm": 0.3187820613384247,
      "learning_rate": 4.862686225222251e-05,
      "loss": 4.541,
      "step": 41110
    },
    {
      "epoch": 0.8420535293756272,
      "grad_norm": 0.35254228115081787,
      "learning_rate": 4.85063846369217e-05,
      "loss": 4.5532,
      "step": 41120
    },
    {
      "epoch": 0.8422583089304364,
      "grad_norm": 0.3221297264099121,
      "learning_rate": 4.838603585436692e-05,
      "loss": 4.6066,
      "step": 41130
    },
    {
      "epoch": 0.8424630884852456,
      "grad_norm": 0.3298701345920563,
      "learning_rate": 4.826581600690476e-05,
      "loss": 4.5563,
      "step": 41140
    },
    {
      "epoch": 0.8426678680400549,
      "grad_norm": 0.30806881189346313,
      "learning_rate": 4.814572519677223e-05,
      "loss": 4.591,
      "step": 41150
    },
    {
      "epoch": 0.8428726475948641,
      "grad_norm": 0.3149655759334564,
      "learning_rate": 4.8025763526096625e-05,
      "loss": 4.5419,
      "step": 41160
    },
    {
      "epoch": 0.8430774271496734,
      "grad_norm": 0.30739855766296387,
      "learning_rate": 4.790593109689541e-05,
      "loss": 4.598,
      "step": 41170
    },
    {
      "epoch": 0.8432822067044826,
      "grad_norm": 0.30234190821647644,
      "learning_rate": 4.7786228011076104e-05,
      "loss": 4.5808,
      "step": 41180
    },
    {
      "epoch": 0.8434869862592919,
      "grad_norm": 0.3341679573059082,
      "learning_rate": 4.76666543704362e-05,
      "loss": 4.5671,
      "step": 41190
    },
    {
      "epoch": 0.8436917658141011,
      "grad_norm": 0.33971861004829407,
      "learning_rate": 4.75472102766632e-05,
      "loss": 4.5599,
      "step": 41200
    },
    {
      "epoch": 0.8438965453689103,
      "grad_norm": 0.3058462142944336,
      "learning_rate": 4.742789583133438e-05,
      "loss": 4.5596,
      "step": 41210
    },
    {
      "epoch": 0.8441013249237196,
      "grad_norm": 0.32935091853141785,
      "learning_rate": 4.730871113591682e-05,
      "loss": 4.5711,
      "step": 41220
    },
    {
      "epoch": 0.8443061044785288,
      "grad_norm": 0.30706486105918884,
      "learning_rate": 4.7189656291767125e-05,
      "loss": 4.604,
      "step": 41230
    },
    {
      "epoch": 0.8445108840333381,
      "grad_norm": 0.32395413517951965,
      "learning_rate": 4.707073140013158e-05,
      "loss": 4.5554,
      "step": 41240
    },
    {
      "epoch": 0.8447156635881473,
      "grad_norm": 0.3302842080593109,
      "learning_rate": 4.6951936562146025e-05,
      "loss": 4.5623,
      "step": 41250
    },
    {
      "epoch": 0.8449204431429566,
      "grad_norm": 0.3194911777973175,
      "learning_rate": 4.683327187883548e-05,
      "loss": 4.5365,
      "step": 41260
    },
    {
      "epoch": 0.8451252226977659,
      "grad_norm": 0.6366714835166931,
      "learning_rate": 4.671473745111446e-05,
      "loss": 4.4751,
      "step": 41270
    },
    {
      "epoch": 0.8453300022525752,
      "grad_norm": 0.3164162337779999,
      "learning_rate": 4.659633337978668e-05,
      "loss": 4.5657,
      "step": 41280
    },
    {
      "epoch": 0.8455347818073844,
      "grad_norm": 0.30572304129600525,
      "learning_rate": 4.6478059765544955e-05,
      "loss": 4.5952,
      "step": 41290
    },
    {
      "epoch": 0.8457395613621936,
      "grad_norm": 0.5788365602493286,
      "learning_rate": 4.635991670897117e-05,
      "loss": 4.5821,
      "step": 41300
    },
    {
      "epoch": 0.8459443409170029,
      "grad_norm": 0.31250980496406555,
      "learning_rate": 4.624190431053628e-05,
      "loss": 4.5627,
      "step": 41310
    },
    {
      "epoch": 0.8461491204718121,
      "grad_norm": 0.35064229369163513,
      "learning_rate": 4.6124022670599875e-05,
      "loss": 4.5678,
      "step": 41320
    },
    {
      "epoch": 0.8463539000266214,
      "grad_norm": 0.31303027272224426,
      "learning_rate": 4.600627188941067e-05,
      "loss": 4.5762,
      "step": 41330
    },
    {
      "epoch": 0.8465586795814306,
      "grad_norm": 0.33292683959007263,
      "learning_rate": 4.5888652067105774e-05,
      "loss": 4.5662,
      "step": 41340
    },
    {
      "epoch": 0.8467634591362398,
      "grad_norm": 0.34634777903556824,
      "learning_rate": 4.577116330371116e-05,
      "loss": 4.5643,
      "step": 41350
    },
    {
      "epoch": 0.8469682386910491,
      "grad_norm": 0.2940186858177185,
      "learning_rate": 4.5653805699141286e-05,
      "loss": 4.5593,
      "step": 41360
    },
    {
      "epoch": 0.8471730182458583,
      "grad_norm": 0.3219406008720398,
      "learning_rate": 4.553657935319901e-05,
      "loss": 4.558,
      "step": 41370
    },
    {
      "epoch": 0.8473777978006676,
      "grad_norm": 0.2876724898815155,
      "learning_rate": 4.5419484365575636e-05,
      "loss": 4.5379,
      "step": 41380
    },
    {
      "epoch": 0.8475825773554768,
      "grad_norm": 0.3168316185474396,
      "learning_rate": 4.53025208358508e-05,
      "loss": 4.5693,
      "step": 41390
    },
    {
      "epoch": 0.8477873569102861,
      "grad_norm": 0.3215070962905884,
      "learning_rate": 4.518568886349217e-05,
      "loss": 4.5863,
      "step": 41400
    },
    {
      "epoch": 0.8479921364650953,
      "grad_norm": 0.3174405097961426,
      "learning_rate": 4.506898854785564e-05,
      "loss": 4.5721,
      "step": 41410
    },
    {
      "epoch": 0.8481969160199045,
      "grad_norm": 0.3024792969226837,
      "learning_rate": 4.495241998818527e-05,
      "loss": 4.5722,
      "step": 41420
    },
    {
      "epoch": 0.8484016955747138,
      "grad_norm": 0.2967088222503662,
      "learning_rate": 4.483598328361278e-05,
      "loss": 4.5923,
      "step": 41430
    },
    {
      "epoch": 0.848606475129523,
      "grad_norm": 0.31650328636169434,
      "learning_rate": 4.4719678533158015e-05,
      "loss": 4.5574,
      "step": 41440
    },
    {
      "epoch": 0.8488112546843323,
      "grad_norm": 0.30982720851898193,
      "learning_rate": 4.46035058357285e-05,
      "loss": 4.5649,
      "step": 41450
    },
    {
      "epoch": 0.8490160342391415,
      "grad_norm": 0.4292488992214203,
      "learning_rate": 4.448746529011951e-05,
      "loss": 4.5609,
      "step": 41460
    },
    {
      "epoch": 0.8492208137939508,
      "grad_norm": 0.3487224578857422,
      "learning_rate": 4.4371556995013805e-05,
      "loss": 4.48,
      "step": 41470
    },
    {
      "epoch": 0.8494255933487601,
      "grad_norm": 0.2968166172504425,
      "learning_rate": 4.425578104898181e-05,
      "loss": 4.5607,
      "step": 41480
    },
    {
      "epoch": 0.8496303729035694,
      "grad_norm": 0.3533106744289398,
      "learning_rate": 4.414013755048139e-05,
      "loss": 4.5658,
      "step": 41490
    },
    {
      "epoch": 0.8498351524583786,
      "grad_norm": 0.31141340732574463,
      "learning_rate": 4.402462659785771e-05,
      "loss": 4.5568,
      "step": 41500
    },
    {
      "epoch": 0.8500399320131878,
      "grad_norm": 0.32396769523620605,
      "learning_rate": 4.3909248289343305e-05,
      "loss": 4.5341,
      "step": 41510
    },
    {
      "epoch": 0.8502447115679971,
      "grad_norm": 0.32039234042167664,
      "learning_rate": 4.3794002723057756e-05,
      "loss": 4.6012,
      "step": 41520
    },
    {
      "epoch": 0.8504494911228063,
      "grad_norm": 0.30332982540130615,
      "learning_rate": 4.367888999700795e-05,
      "loss": 4.5529,
      "step": 41530
    },
    {
      "epoch": 0.8506542706776156,
      "grad_norm": 0.30191323161125183,
      "learning_rate": 4.356391020908761e-05,
      "loss": 4.5609,
      "step": 41540
    },
    {
      "epoch": 0.8508590502324248,
      "grad_norm": 0.32889968156814575,
      "learning_rate": 4.3449063457077555e-05,
      "loss": 4.5905,
      "step": 41550
    },
    {
      "epoch": 0.851063829787234,
      "grad_norm": 0.31554514169692993,
      "learning_rate": 4.3334349838645393e-05,
      "loss": 4.5765,
      "step": 41560
    },
    {
      "epoch": 0.8512686093420433,
      "grad_norm": 0.29832085967063904,
      "learning_rate": 4.321976945134556e-05,
      "loss": 4.602,
      "step": 41570
    },
    {
      "epoch": 0.8514733888968525,
      "grad_norm": 0.32858118414878845,
      "learning_rate": 4.3105322392619154e-05,
      "loss": 4.5643,
      "step": 41580
    },
    {
      "epoch": 0.8516781684516618,
      "grad_norm": 0.3178858757019043,
      "learning_rate": 4.299100875979392e-05,
      "loss": 4.6036,
      "step": 41590
    },
    {
      "epoch": 0.851882948006471,
      "grad_norm": 0.5597391724586487,
      "learning_rate": 4.287682865008413e-05,
      "loss": 4.5431,
      "step": 41600
    },
    {
      "epoch": 0.8520877275612803,
      "grad_norm": 0.30292996764183044,
      "learning_rate": 4.27627821605904e-05,
      "loss": 4.5701,
      "step": 41610
    },
    {
      "epoch": 0.8522925071160895,
      "grad_norm": 0.3429625630378723,
      "learning_rate": 4.2648869388299926e-05,
      "loss": 4.5742,
      "step": 41620
    },
    {
      "epoch": 0.8524972866708987,
      "grad_norm": 0.308368444442749,
      "learning_rate": 4.2535090430085944e-05,
      "loss": 4.5558,
      "step": 41630
    },
    {
      "epoch": 0.852702066225708,
      "grad_norm": 0.3081679344177246,
      "learning_rate": 4.242144538270807e-05,
      "loss": 4.5283,
      "step": 41640
    },
    {
      "epoch": 0.8529068457805172,
      "grad_norm": 0.2995325028896332,
      "learning_rate": 4.2307934342811976e-05,
      "loss": 4.5819,
      "step": 41650
    },
    {
      "epoch": 0.8531116253353265,
      "grad_norm": 0.30443987250328064,
      "learning_rate": 4.219455740692939e-05,
      "loss": 4.5793,
      "step": 41660
    },
    {
      "epoch": 0.8533164048901357,
      "grad_norm": 0.30008235573768616,
      "learning_rate": 4.208131467147804e-05,
      "loss": 4.5818,
      "step": 41670
    },
    {
      "epoch": 0.8535211844449451,
      "grad_norm": 0.3058653473854065,
      "learning_rate": 4.1968206232761387e-05,
      "loss": 4.5616,
      "step": 41680
    },
    {
      "epoch": 0.8537259639997543,
      "grad_norm": 0.2992249131202698,
      "learning_rate": 4.185523218696879e-05,
      "loss": 4.5618,
      "step": 41690
    },
    {
      "epoch": 0.8539307435545636,
      "grad_norm": 0.29264143109321594,
      "learning_rate": 4.1742392630175294e-05,
      "loss": 4.5486,
      "step": 41700
    },
    {
      "epoch": 0.8541355231093728,
      "grad_norm": 0.3090275526046753,
      "learning_rate": 4.162968765834166e-05,
      "loss": 4.6067,
      "step": 41710
    },
    {
      "epoch": 0.854340302664182,
      "grad_norm": 0.3257957696914673,
      "learning_rate": 4.1517117367313983e-05,
      "loss": 4.5606,
      "step": 41720
    },
    {
      "epoch": 0.8545450822189913,
      "grad_norm": 0.2962327003479004,
      "learning_rate": 4.1404681852824025e-05,
      "loss": 4.5485,
      "step": 41730
    },
    {
      "epoch": 0.8547498617738005,
      "grad_norm": 0.3346448540687561,
      "learning_rate": 4.1292381210488906e-05,
      "loss": 4.5218,
      "step": 41740
    },
    {
      "epoch": 0.8549546413286098,
      "grad_norm": 0.2993049919605255,
      "learning_rate": 4.118021553581086e-05,
      "loss": 4.5781,
      "step": 41750
    },
    {
      "epoch": 0.855159420883419,
      "grad_norm": 0.2944900393486023,
      "learning_rate": 4.106818492417759e-05,
      "loss": 4.5567,
      "step": 41760
    },
    {
      "epoch": 0.8553642004382283,
      "grad_norm": 0.36183977127075195,
      "learning_rate": 4.095628947086181e-05,
      "loss": 4.5426,
      "step": 41770
    },
    {
      "epoch": 0.8555689799930375,
      "grad_norm": 0.3020537495613098,
      "learning_rate": 4.0844529271021315e-05,
      "loss": 4.5681,
      "step": 41780
    },
    {
      "epoch": 0.8557737595478467,
      "grad_norm": 0.30587658286094666,
      "learning_rate": 4.073290441969888e-05,
      "loss": 4.5776,
      "step": 41790
    },
    {
      "epoch": 0.855978539102656,
      "grad_norm": 0.30620551109313965,
      "learning_rate": 4.062141501182222e-05,
      "loss": 4.6001,
      "step": 41800
    },
    {
      "epoch": 0.8561833186574652,
      "grad_norm": 0.32741957902908325,
      "learning_rate": 4.051006114220379e-05,
      "loss": 4.5872,
      "step": 41810
    },
    {
      "epoch": 0.8563880982122745,
      "grad_norm": 0.3100329637527466,
      "learning_rate": 4.0398842905540744e-05,
      "loss": 4.5438,
      "step": 41820
    },
    {
      "epoch": 0.8565928777670837,
      "grad_norm": 0.315884530544281,
      "learning_rate": 4.028776039641502e-05,
      "loss": 4.5589,
      "step": 41830
    },
    {
      "epoch": 0.856797657321893,
      "grad_norm": 0.30159467458724976,
      "learning_rate": 4.017681370929307e-05,
      "loss": 4.5447,
      "step": 41840
    },
    {
      "epoch": 0.8570024368767022,
      "grad_norm": 0.30200254917144775,
      "learning_rate": 4.0066002938525846e-05,
      "loss": 4.5295,
      "step": 41850
    },
    {
      "epoch": 0.8572072164315114,
      "grad_norm": 0.33325955271720886,
      "learning_rate": 4e-05,
      "loss": 4.578,
      "step": 41860
    },
    {
      "epoch": 0.8574119959863207,
      "grad_norm": 0.3209024965763092,
      "learning_rate": 4e-05,
      "loss": 4.5784,
      "step": 41870
    },
    {
      "epoch": 0.85761677554113,
      "grad_norm": 0.3402951955795288,
      "learning_rate": 4e-05,
      "loss": 4.5534,
      "step": 41880
    },
    {
      "epoch": 0.8578215550959393,
      "grad_norm": 0.292206346988678,
      "learning_rate": 4e-05,
      "loss": 4.5532,
      "step": 41890
    },
    {
      "epoch": 0.8580263346507485,
      "grad_norm": 0.3007727563381195,
      "learning_rate": 4e-05,
      "loss": 4.543,
      "step": 41900
    },
    {
      "epoch": 0.8582311142055578,
      "grad_norm": 0.31559112668037415,
      "learning_rate": 4e-05,
      "loss": 4.5518,
      "step": 41910
    },
    {
      "epoch": 0.858435893760367,
      "grad_norm": 0.30795371532440186,
      "learning_rate": 4e-05,
      "loss": 4.5488,
      "step": 41920
    },
    {
      "epoch": 0.8586406733151762,
      "grad_norm": 0.3247368037700653,
      "learning_rate": 4e-05,
      "loss": 4.5674,
      "step": 41930
    },
    {
      "epoch": 0.8588454528699855,
      "grad_norm": 0.32226502895355225,
      "learning_rate": 4e-05,
      "loss": 4.5727,
      "step": 41940
    },
    {
      "epoch": 0.8590502324247947,
      "grad_norm": 0.31973397731781006,
      "learning_rate": 4e-05,
      "loss": 4.5588,
      "step": 41950
    },
    {
      "epoch": 0.859255011979604,
      "grad_norm": 0.31184062361717224,
      "learning_rate": 4e-05,
      "loss": 4.5854,
      "step": 41960
    },
    {
      "epoch": 0.8594597915344132,
      "grad_norm": 0.30193692445755005,
      "learning_rate": 4e-05,
      "loss": 4.5686,
      "step": 41970
    },
    {
      "epoch": 0.8596645710892225,
      "grad_norm": 0.3234449028968811,
      "learning_rate": 4e-05,
      "loss": 4.5888,
      "step": 41980
    },
    {
      "epoch": 0.8598693506440317,
      "grad_norm": 0.3161271810531616,
      "learning_rate": 4e-05,
      "loss": 4.5499,
      "step": 41990
    },
    {
      "epoch": 0.8600741301988409,
      "grad_norm": 0.32918792963027954,
      "learning_rate": 4e-05,
      "loss": 4.5268,
      "step": 42000
    },
    {
      "epoch": 0.8600741301988409,
      "eval_loss": 4.56494665145874,
      "eval_runtime": 4.3997,
      "eval_samples_per_second": 265.017,
      "eval_steps_per_second": 33.184,
      "step": 42000
    },
    {
      "epoch": 0.8602789097536502,
      "grad_norm": 0.30065762996673584,
      "learning_rate": 4e-05,
      "loss": 4.5516,
      "step": 42010
    },
    {
      "epoch": 0.8604836893084594,
      "grad_norm": 0.30190443992614746,
      "learning_rate": 4e-05,
      "loss": 4.5792,
      "step": 42020
    },
    {
      "epoch": 0.8606884688632687,
      "grad_norm": 0.3021637797355652,
      "learning_rate": 4e-05,
      "loss": 4.5673,
      "step": 42030
    },
    {
      "epoch": 0.8608932484180779,
      "grad_norm": 0.3549637496471405,
      "learning_rate": 4e-05,
      "loss": 4.5634,
      "step": 42040
    },
    {
      "epoch": 0.8610980279728871,
      "grad_norm": 0.3007580637931824,
      "learning_rate": 4e-05,
      "loss": 4.5767,
      "step": 42050
    },
    {
      "epoch": 0.8613028075276964,
      "grad_norm": 0.37332987785339355,
      "learning_rate": 4e-05,
      "loss": 4.531,
      "step": 42060
    },
    {
      "epoch": 0.8615075870825056,
      "grad_norm": 0.32356762886047363,
      "learning_rate": 4e-05,
      "loss": 4.5386,
      "step": 42070
    },
    {
      "epoch": 0.861712366637315,
      "grad_norm": 0.30995801091194153,
      "learning_rate": 4e-05,
      "loss": 4.5428,
      "step": 42080
    },
    {
      "epoch": 0.8619171461921242,
      "grad_norm": 0.30891141295433044,
      "learning_rate": 4e-05,
      "loss": 4.5678,
      "step": 42090
    },
    {
      "epoch": 0.8621219257469335,
      "grad_norm": 0.3204641044139862,
      "learning_rate": 4e-05,
      "loss": 4.5559,
      "step": 42100
    },
    {
      "epoch": 0.8623267053017427,
      "grad_norm": 0.30229073762893677,
      "learning_rate": 4e-05,
      "loss": 4.5764,
      "step": 42110
    },
    {
      "epoch": 0.862531484856552,
      "grad_norm": 0.30723774433135986,
      "learning_rate": 4e-05,
      "loss": 4.5702,
      "step": 42120
    },
    {
      "epoch": 0.8627362644113612,
      "grad_norm": 0.3078003227710724,
      "learning_rate": 4e-05,
      "loss": 4.5797,
      "step": 42130
    },
    {
      "epoch": 0.8629410439661704,
      "grad_norm": 0.29912465810775757,
      "learning_rate": 4e-05,
      "loss": 4.5567,
      "step": 42140
    },
    {
      "epoch": 0.8631458235209797,
      "grad_norm": 0.3218294680118561,
      "learning_rate": 4e-05,
      "loss": 4.5727,
      "step": 42150
    },
    {
      "epoch": 0.8633506030757889,
      "grad_norm": 0.29612016677856445,
      "learning_rate": 4e-05,
      "loss": 4.5948,
      "step": 42160
    },
    {
      "epoch": 0.8635553826305982,
      "grad_norm": 0.299450546503067,
      "learning_rate": 4e-05,
      "loss": 4.5824,
      "step": 42170
    },
    {
      "epoch": 0.8637601621854074,
      "grad_norm": 0.32398706674575806,
      "learning_rate": 4e-05,
      "loss": 4.6054,
      "step": 42180
    },
    {
      "epoch": 0.8639649417402167,
      "grad_norm": 0.3253730237483978,
      "learning_rate": 4e-05,
      "loss": 4.5552,
      "step": 42190
    },
    {
      "epoch": 0.8641697212950259,
      "grad_norm": 0.311602920293808,
      "learning_rate": 4e-05,
      "loss": 4.5308,
      "step": 42200
    },
    {
      "epoch": 0.8643745008498351,
      "grad_norm": 0.3064606189727783,
      "learning_rate": 4e-05,
      "loss": 4.5182,
      "step": 42210
    },
    {
      "epoch": 0.8645792804046444,
      "grad_norm": 0.3285084068775177,
      "learning_rate": 4e-05,
      "loss": 4.5551,
      "step": 42220
    },
    {
      "epoch": 0.8647840599594536,
      "grad_norm": 0.2966082990169525,
      "learning_rate": 4e-05,
      "loss": 4.5678,
      "step": 42230
    },
    {
      "epoch": 0.8649888395142629,
      "grad_norm": 0.3242189288139343,
      "learning_rate": 4e-05,
      "loss": 4.5797,
      "step": 42240
    },
    {
      "epoch": 0.8651936190690721,
      "grad_norm": 0.30055972933769226,
      "learning_rate": 4e-05,
      "loss": 4.5374,
      "step": 42250
    },
    {
      "epoch": 0.8653983986238813,
      "grad_norm": 0.3144788444042206,
      "learning_rate": 4e-05,
      "loss": 4.5678,
      "step": 42260
    },
    {
      "epoch": 0.8656031781786906,
      "grad_norm": 0.33898335695266724,
      "learning_rate": 4e-05,
      "loss": 4.5581,
      "step": 42270
    },
    {
      "epoch": 0.8658079577334998,
      "grad_norm": 0.41513872146606445,
      "learning_rate": 4e-05,
      "loss": 4.5611,
      "step": 42280
    },
    {
      "epoch": 0.8660127372883092,
      "grad_norm": 0.3530650734901428,
      "learning_rate": 4e-05,
      "loss": 4.5378,
      "step": 42290
    },
    {
      "epoch": 0.8662175168431184,
      "grad_norm": 0.3262985646724701,
      "learning_rate": 4e-05,
      "loss": 4.5524,
      "step": 42300
    },
    {
      "epoch": 0.8664222963979277,
      "grad_norm": 0.35833224654197693,
      "learning_rate": 4e-05,
      "loss": 4.5554,
      "step": 42310
    },
    {
      "epoch": 0.8666270759527369,
      "grad_norm": 0.33766356110572815,
      "learning_rate": 4e-05,
      "loss": 4.5149,
      "step": 42320
    },
    {
      "epoch": 0.8668318555075462,
      "grad_norm": 0.30668026208877563,
      "learning_rate": 4e-05,
      "loss": 4.5318,
      "step": 42330
    },
    {
      "epoch": 0.8670366350623554,
      "grad_norm": 0.2991671562194824,
      "learning_rate": 4e-05,
      "loss": 4.6324,
      "step": 42340
    },
    {
      "epoch": 0.8672414146171646,
      "grad_norm": 0.3278384208679199,
      "learning_rate": 4e-05,
      "loss": 4.5374,
      "step": 42350
    },
    {
      "epoch": 0.8674461941719739,
      "grad_norm": 0.3133397400379181,
      "learning_rate": 4e-05,
      "loss": 4.5671,
      "step": 42360
    },
    {
      "epoch": 0.8676509737267831,
      "grad_norm": 0.3142462968826294,
      "learning_rate": 4e-05,
      "loss": 4.5164,
      "step": 42370
    },
    {
      "epoch": 0.8678557532815924,
      "grad_norm": 0.3174171447753906,
      "learning_rate": 4e-05,
      "loss": 4.572,
      "step": 42380
    },
    {
      "epoch": 0.8680605328364016,
      "grad_norm": 0.3513506352901459,
      "learning_rate": 4e-05,
      "loss": 4.5723,
      "step": 42390
    },
    {
      "epoch": 0.8682653123912109,
      "grad_norm": 0.300601065158844,
      "learning_rate": 4e-05,
      "loss": 4.5689,
      "step": 42400
    },
    {
      "epoch": 0.8684700919460201,
      "grad_norm": 0.313957154750824,
      "learning_rate": 4e-05,
      "loss": 4.5398,
      "step": 42410
    },
    {
      "epoch": 0.8686748715008293,
      "grad_norm": 0.31928592920303345,
      "learning_rate": 4e-05,
      "loss": 4.5752,
      "step": 42420
    },
    {
      "epoch": 0.8688796510556386,
      "grad_norm": 0.32813411951065063,
      "learning_rate": 4e-05,
      "loss": 4.5802,
      "step": 42430
    },
    {
      "epoch": 0.8690844306104478,
      "grad_norm": 0.34862348437309265,
      "learning_rate": 4e-05,
      "loss": 4.583,
      "step": 42440
    },
    {
      "epoch": 0.8692892101652571,
      "grad_norm": 0.3211762607097626,
      "learning_rate": 4e-05,
      "loss": 4.5685,
      "step": 42450
    },
    {
      "epoch": 0.8694939897200663,
      "grad_norm": 0.3033827245235443,
      "learning_rate": 4e-05,
      "loss": 4.536,
      "step": 42460
    },
    {
      "epoch": 0.8696987692748755,
      "grad_norm": 0.3189534544944763,
      "learning_rate": 4e-05,
      "loss": 4.5762,
      "step": 42470
    },
    {
      "epoch": 0.8699035488296848,
      "grad_norm": 0.3091621398925781,
      "learning_rate": 4e-05,
      "loss": 4.5487,
      "step": 42480
    },
    {
      "epoch": 0.8701083283844941,
      "grad_norm": 0.3051370680332184,
      "learning_rate": 4e-05,
      "loss": 4.5098,
      "step": 42490
    },
    {
      "epoch": 0.8703131079393034,
      "grad_norm": 0.30030304193496704,
      "learning_rate": 4e-05,
      "loss": 4.5589,
      "step": 42500
    },
    {
      "epoch": 0.8705178874941126,
      "grad_norm": 0.3978952169418335,
      "learning_rate": 4e-05,
      "loss": 4.5803,
      "step": 42510
    },
    {
      "epoch": 0.8707226670489219,
      "grad_norm": 0.3171425759792328,
      "learning_rate": 4e-05,
      "loss": 4.5616,
      "step": 42520
    },
    {
      "epoch": 0.8709274466037311,
      "grad_norm": 0.31384173035621643,
      "learning_rate": 4e-05,
      "loss": 4.559,
      "step": 42530
    },
    {
      "epoch": 0.8711322261585404,
      "grad_norm": 0.3421635627746582,
      "learning_rate": 4e-05,
      "loss": 4.5509,
      "step": 42540
    },
    {
      "epoch": 0.8713370057133496,
      "grad_norm": 0.3201238214969635,
      "learning_rate": 4e-05,
      "loss": 4.5384,
      "step": 42550
    },
    {
      "epoch": 0.8715417852681588,
      "grad_norm": 0.30009725689888,
      "learning_rate": 4e-05,
      "loss": 4.5758,
      "step": 42560
    },
    {
      "epoch": 0.8717465648229681,
      "grad_norm": 0.3192896544933319,
      "learning_rate": 4e-05,
      "loss": 4.5587,
      "step": 42570
    },
    {
      "epoch": 0.8719513443777773,
      "grad_norm": 0.3128572702407837,
      "learning_rate": 4e-05,
      "loss": 4.5677,
      "step": 42580
    },
    {
      "epoch": 0.8721561239325866,
      "grad_norm": 0.33970656991004944,
      "learning_rate": 4e-05,
      "loss": 4.5495,
      "step": 42590
    },
    {
      "epoch": 0.8723609034873958,
      "grad_norm": 0.3380085825920105,
      "learning_rate": 4e-05,
      "loss": 4.529,
      "step": 42600
    },
    {
      "epoch": 0.872565683042205,
      "grad_norm": 0.3085273504257202,
      "learning_rate": 4e-05,
      "loss": 4.5476,
      "step": 42610
    },
    {
      "epoch": 0.8727704625970143,
      "grad_norm": 0.3030671775341034,
      "learning_rate": 4e-05,
      "loss": 4.561,
      "step": 42620
    },
    {
      "epoch": 0.8729752421518235,
      "grad_norm": 0.30912870168685913,
      "learning_rate": 4e-05,
      "loss": 4.5538,
      "step": 42630
    },
    {
      "epoch": 0.8731800217066328,
      "grad_norm": 0.30605942010879517,
      "learning_rate": 4e-05,
      "loss": 4.5321,
      "step": 42640
    },
    {
      "epoch": 0.873384801261442,
      "grad_norm": 0.43395280838012695,
      "learning_rate": 4e-05,
      "loss": 4.5088,
      "step": 42650
    },
    {
      "epoch": 0.8735895808162513,
      "grad_norm": 0.3205019235610962,
      "learning_rate": 4e-05,
      "loss": 4.5411,
      "step": 42660
    },
    {
      "epoch": 0.8737943603710605,
      "grad_norm": 0.3275338113307953,
      "learning_rate": 4e-05,
      "loss": 4.5494,
      "step": 42670
    },
    {
      "epoch": 0.8739991399258698,
      "grad_norm": 0.30848512053489685,
      "learning_rate": 4e-05,
      "loss": 4.5023,
      "step": 42680
    },
    {
      "epoch": 0.8742039194806791,
      "grad_norm": 0.311075896024704,
      "learning_rate": 4e-05,
      "loss": 4.5639,
      "step": 42690
    },
    {
      "epoch": 0.8744086990354883,
      "grad_norm": 0.3298903703689575,
      "learning_rate": 4e-05,
      "loss": 4.6063,
      "step": 42700
    },
    {
      "epoch": 0.8746134785902976,
      "grad_norm": 0.3591431677341461,
      "learning_rate": 4e-05,
      "loss": 4.5294,
      "step": 42710
    },
    {
      "epoch": 0.8748182581451068,
      "grad_norm": 0.3300228714942932,
      "learning_rate": 4e-05,
      "loss": 4.5411,
      "step": 42720
    },
    {
      "epoch": 0.8750230376999161,
      "grad_norm": 0.31195318698883057,
      "learning_rate": 4e-05,
      "loss": 4.5967,
      "step": 42730
    },
    {
      "epoch": 0.8752278172547253,
      "grad_norm": 0.3583310842514038,
      "learning_rate": 4e-05,
      "loss": 4.5543,
      "step": 42740
    },
    {
      "epoch": 0.8754325968095346,
      "grad_norm": 0.302247017621994,
      "learning_rate": 4e-05,
      "loss": 4.6008,
      "step": 42750
    },
    {
      "epoch": 0.8756373763643438,
      "grad_norm": 0.31350117921829224,
      "learning_rate": 4e-05,
      "loss": 4.5515,
      "step": 42760
    },
    {
      "epoch": 0.875842155919153,
      "grad_norm": 0.32353147864341736,
      "learning_rate": 4e-05,
      "loss": 4.5739,
      "step": 42770
    },
    {
      "epoch": 0.8760469354739623,
      "grad_norm": 0.29817453026771545,
      "learning_rate": 4e-05,
      "loss": 4.5479,
      "step": 42780
    },
    {
      "epoch": 0.8762517150287715,
      "grad_norm": 0.30565959215164185,
      "learning_rate": 4e-05,
      "loss": 4.5859,
      "step": 42790
    },
    {
      "epoch": 0.8764564945835808,
      "grad_norm": 0.31741973757743835,
      "learning_rate": 4e-05,
      "loss": 4.5799,
      "step": 42800
    },
    {
      "epoch": 0.87666127413839,
      "grad_norm": 0.3071531653404236,
      "learning_rate": 4e-05,
      "loss": 4.5868,
      "step": 42810
    },
    {
      "epoch": 0.8768660536931993,
      "grad_norm": 0.31487077474594116,
      "learning_rate": 4e-05,
      "loss": 4.5586,
      "step": 42820
    },
    {
      "epoch": 0.8770708332480085,
      "grad_norm": 0.309831827878952,
      "learning_rate": 4e-05,
      "loss": 4.591,
      "step": 42830
    },
    {
      "epoch": 0.8772756128028177,
      "grad_norm": 0.30910763144493103,
      "learning_rate": 4e-05,
      "loss": 4.5528,
      "step": 42840
    },
    {
      "epoch": 0.877480392357627,
      "grad_norm": 0.33074498176574707,
      "learning_rate": 4e-05,
      "loss": 4.5742,
      "step": 42850
    },
    {
      "epoch": 0.8776851719124362,
      "grad_norm": 0.30193716287612915,
      "learning_rate": 4e-05,
      "loss": 4.54,
      "step": 42860
    },
    {
      "epoch": 0.8778899514672455,
      "grad_norm": 0.31014299392700195,
      "learning_rate": 4e-05,
      "loss": 4.5476,
      "step": 42870
    },
    {
      "epoch": 0.8780947310220547,
      "grad_norm": 0.3074629008769989,
      "learning_rate": 4e-05,
      "loss": 4.5636,
      "step": 42880
    },
    {
      "epoch": 0.878299510576864,
      "grad_norm": 0.3182469606399536,
      "learning_rate": 4e-05,
      "loss": 4.5752,
      "step": 42890
    },
    {
      "epoch": 0.8785042901316733,
      "grad_norm": 0.3700508177280426,
      "learning_rate": 4e-05,
      "loss": 4.5677,
      "step": 42900
    },
    {
      "epoch": 0.8787090696864825,
      "grad_norm": 0.31222066283226013,
      "learning_rate": 4e-05,
      "loss": 4.5701,
      "step": 42910
    },
    {
      "epoch": 0.8789138492412918,
      "grad_norm": 0.32930290699005127,
      "learning_rate": 4e-05,
      "loss": 4.5347,
      "step": 42920
    },
    {
      "epoch": 0.879118628796101,
      "grad_norm": 0.3188677132129669,
      "learning_rate": 4e-05,
      "loss": 4.5241,
      "step": 42930
    },
    {
      "epoch": 0.8793234083509103,
      "grad_norm": 0.3106885552406311,
      "learning_rate": 4e-05,
      "loss": 4.5822,
      "step": 42940
    },
    {
      "epoch": 0.8795281879057195,
      "grad_norm": 0.31368157267570496,
      "learning_rate": 4e-05,
      "loss": 4.5754,
      "step": 42950
    },
    {
      "epoch": 0.8797329674605288,
      "grad_norm": 0.30185890197753906,
      "learning_rate": 4e-05,
      "loss": 4.5788,
      "step": 42960
    },
    {
      "epoch": 0.879937747015338,
      "grad_norm": 0.3285743296146393,
      "learning_rate": 4e-05,
      "loss": 4.5545,
      "step": 42970
    },
    {
      "epoch": 0.8801425265701472,
      "grad_norm": 0.29756179451942444,
      "learning_rate": 4e-05,
      "loss": 4.5765,
      "step": 42980
    },
    {
      "epoch": 0.8803473061249565,
      "grad_norm": 0.3096250891685486,
      "learning_rate": 4e-05,
      "loss": 4.5244,
      "step": 42990
    },
    {
      "epoch": 0.8805520856797657,
      "grad_norm": 0.31590718030929565,
      "learning_rate": 4e-05,
      "loss": 4.554,
      "step": 43000
    },
    {
      "epoch": 0.880756865234575,
      "grad_norm": 0.3368012011051178,
      "learning_rate": 4e-05,
      "loss": 4.5296,
      "step": 43010
    },
    {
      "epoch": 0.8809616447893842,
      "grad_norm": 0.3366091251373291,
      "learning_rate": 4e-05,
      "loss": 4.5244,
      "step": 43020
    },
    {
      "epoch": 0.8811664243441935,
      "grad_norm": 0.30473312735557556,
      "learning_rate": 4e-05,
      "loss": 4.5565,
      "step": 43030
    },
    {
      "epoch": 0.8813712038990027,
      "grad_norm": 0.3344012498855591,
      "learning_rate": 4e-05,
      "loss": 4.5351,
      "step": 43040
    },
    {
      "epoch": 0.8815759834538119,
      "grad_norm": 0.34798553586006165,
      "learning_rate": 4e-05,
      "loss": 4.5809,
      "step": 43050
    },
    {
      "epoch": 0.8817807630086212,
      "grad_norm": 0.32808583974838257,
      "learning_rate": 4e-05,
      "loss": 4.5515,
      "step": 43060
    },
    {
      "epoch": 0.8819855425634304,
      "grad_norm": 0.3296571373939514,
      "learning_rate": 4e-05,
      "loss": 4.5609,
      "step": 43070
    },
    {
      "epoch": 0.8821903221182397,
      "grad_norm": 0.3262360692024231,
      "learning_rate": 4e-05,
      "loss": 4.5828,
      "step": 43080
    },
    {
      "epoch": 0.8823951016730489,
      "grad_norm": 0.31047987937927246,
      "learning_rate": 4e-05,
      "loss": 4.5527,
      "step": 43090
    },
    {
      "epoch": 0.8825998812278583,
      "grad_norm": 0.30903884768486023,
      "learning_rate": 4e-05,
      "loss": 4.5466,
      "step": 43100
    },
    {
      "epoch": 0.8828046607826675,
      "grad_norm": 0.3132905066013336,
      "learning_rate": 4e-05,
      "loss": 4.5694,
      "step": 43110
    },
    {
      "epoch": 0.8830094403374767,
      "grad_norm": 0.3017004132270813,
      "learning_rate": 4e-05,
      "loss": 4.5603,
      "step": 43120
    },
    {
      "epoch": 0.883214219892286,
      "grad_norm": 0.3122783601284027,
      "learning_rate": 4e-05,
      "loss": 4.5507,
      "step": 43130
    },
    {
      "epoch": 0.8834189994470952,
      "grad_norm": 0.3365567922592163,
      "learning_rate": 4e-05,
      "loss": 4.603,
      "step": 43140
    },
    {
      "epoch": 0.8836237790019045,
      "grad_norm": 0.30653059482574463,
      "learning_rate": 4e-05,
      "loss": 4.5499,
      "step": 43150
    },
    {
      "epoch": 0.8838285585567137,
      "grad_norm": 0.3418235778808594,
      "learning_rate": 4e-05,
      "loss": 4.5295,
      "step": 43160
    },
    {
      "epoch": 0.884033338111523,
      "grad_norm": 0.30453822016716003,
      "learning_rate": 4e-05,
      "loss": 4.626,
      "step": 43170
    },
    {
      "epoch": 0.8842381176663322,
      "grad_norm": 0.31867340207099915,
      "learning_rate": 4e-05,
      "loss": 4.5563,
      "step": 43180
    },
    {
      "epoch": 0.8844428972211414,
      "grad_norm": 0.29632946848869324,
      "learning_rate": 4e-05,
      "loss": 4.5367,
      "step": 43190
    },
    {
      "epoch": 0.8846476767759507,
      "grad_norm": 0.30943238735198975,
      "learning_rate": 4e-05,
      "loss": 4.5314,
      "step": 43200
    },
    {
      "epoch": 0.8848524563307599,
      "grad_norm": 0.3099985122680664,
      "learning_rate": 4e-05,
      "loss": 4.543,
      "step": 43210
    },
    {
      "epoch": 0.8850572358855692,
      "grad_norm": 0.35353514552116394,
      "learning_rate": 4e-05,
      "loss": 4.579,
      "step": 43220
    },
    {
      "epoch": 0.8852620154403784,
      "grad_norm": 0.3314644992351532,
      "learning_rate": 4e-05,
      "loss": 4.5508,
      "step": 43230
    },
    {
      "epoch": 0.8854667949951877,
      "grad_norm": 0.3235672414302826,
      "learning_rate": 4e-05,
      "loss": 4.5293,
      "step": 43240
    },
    {
      "epoch": 0.8856715745499969,
      "grad_norm": 0.306649386882782,
      "learning_rate": 4e-05,
      "loss": 4.5802,
      "step": 43250
    },
    {
      "epoch": 0.8858763541048061,
      "grad_norm": 0.3142267167568207,
      "learning_rate": 4e-05,
      "loss": 4.5543,
      "step": 43260
    },
    {
      "epoch": 0.8860811336596154,
      "grad_norm": 0.3082157075405121,
      "learning_rate": 4e-05,
      "loss": 4.5236,
      "step": 43270
    },
    {
      "epoch": 0.8862859132144246,
      "grad_norm": 0.31786486506462097,
      "learning_rate": 4e-05,
      "loss": 4.5831,
      "step": 43280
    },
    {
      "epoch": 0.8864906927692339,
      "grad_norm": 0.31692785024642944,
      "learning_rate": 4e-05,
      "loss": 4.5494,
      "step": 43290
    },
    {
      "epoch": 0.8866954723240432,
      "grad_norm": 0.333349347114563,
      "learning_rate": 4e-05,
      "loss": 4.5663,
      "step": 43300
    },
    {
      "epoch": 0.8869002518788525,
      "grad_norm": 0.39977821707725525,
      "learning_rate": 4e-05,
      "loss": 4.5384,
      "step": 43310
    },
    {
      "epoch": 0.8871050314336617,
      "grad_norm": 0.31643402576446533,
      "learning_rate": 4e-05,
      "loss": 4.5889,
      "step": 43320
    },
    {
      "epoch": 0.887309810988471,
      "grad_norm": 0.3167843818664551,
      "learning_rate": 4e-05,
      "loss": 4.5776,
      "step": 43330
    },
    {
      "epoch": 0.8875145905432802,
      "grad_norm": 0.32853609323501587,
      "learning_rate": 4e-05,
      "loss": 4.5614,
      "step": 43340
    },
    {
      "epoch": 0.8877193700980894,
      "grad_norm": 0.3533945679664612,
      "learning_rate": 4e-05,
      "loss": 4.563,
      "step": 43350
    },
    {
      "epoch": 0.8879241496528987,
      "grad_norm": 0.3062794506549835,
      "learning_rate": 4e-05,
      "loss": 4.5707,
      "step": 43360
    },
    {
      "epoch": 0.8881289292077079,
      "grad_norm": 0.3259894847869873,
      "learning_rate": 4e-05,
      "loss": 4.5285,
      "step": 43370
    },
    {
      "epoch": 0.8883337087625172,
      "grad_norm": 0.3182832598686218,
      "learning_rate": 4e-05,
      "loss": 4.5414,
      "step": 43380
    },
    {
      "epoch": 0.8885384883173264,
      "grad_norm": 0.29748913645744324,
      "learning_rate": 4e-05,
      "loss": 4.5454,
      "step": 43390
    },
    {
      "epoch": 0.8887432678721356,
      "grad_norm": 0.32087647914886475,
      "learning_rate": 4e-05,
      "loss": 4.5723,
      "step": 43400
    },
    {
      "epoch": 0.8889480474269449,
      "grad_norm": 0.317262202501297,
      "learning_rate": 4e-05,
      "loss": 4.5367,
      "step": 43410
    },
    {
      "epoch": 0.8891528269817541,
      "grad_norm": 0.3280443847179413,
      "learning_rate": 4e-05,
      "loss": 4.5594,
      "step": 43420
    },
    {
      "epoch": 0.8893576065365634,
      "grad_norm": 0.3304807245731354,
      "learning_rate": 4e-05,
      "loss": 4.5985,
      "step": 43430
    },
    {
      "epoch": 0.8895623860913726,
      "grad_norm": 0.6725199818611145,
      "learning_rate": 4e-05,
      "loss": 4.5865,
      "step": 43440
    },
    {
      "epoch": 0.8897671656461819,
      "grad_norm": 0.33667856454849243,
      "learning_rate": 4e-05,
      "loss": 4.5777,
      "step": 43450
    },
    {
      "epoch": 0.8899719452009911,
      "grad_norm": 0.31030434370040894,
      "learning_rate": 4e-05,
      "loss": 4.5264,
      "step": 43460
    },
    {
      "epoch": 0.8901767247558003,
      "grad_norm": 0.31836238503456116,
      "learning_rate": 4e-05,
      "loss": 4.5815,
      "step": 43470
    },
    {
      "epoch": 0.8903815043106096,
      "grad_norm": 0.3213574290275574,
      "learning_rate": 4e-05,
      "loss": 4.5866,
      "step": 43480
    },
    {
      "epoch": 0.8905862838654188,
      "grad_norm": 0.37443673610687256,
      "learning_rate": 4e-05,
      "loss": 4.5857,
      "step": 43490
    },
    {
      "epoch": 0.8907910634202282,
      "grad_norm": 0.3113357126712799,
      "learning_rate": 4e-05,
      "loss": 4.5476,
      "step": 43500
    },
    {
      "epoch": 0.8909958429750374,
      "grad_norm": 0.45056983828544617,
      "learning_rate": 4e-05,
      "loss": 4.5835,
      "step": 43510
    },
    {
      "epoch": 0.8912006225298467,
      "grad_norm": 0.40995118021965027,
      "learning_rate": 4e-05,
      "loss": 4.5072,
      "step": 43520
    },
    {
      "epoch": 0.8914054020846559,
      "grad_norm": 0.31137964129447937,
      "learning_rate": 4e-05,
      "loss": 4.567,
      "step": 43530
    },
    {
      "epoch": 0.8916101816394651,
      "grad_norm": 0.327394962310791,
      "learning_rate": 4e-05,
      "loss": 4.5762,
      "step": 43540
    },
    {
      "epoch": 0.8918149611942744,
      "grad_norm": 0.31518206000328064,
      "learning_rate": 4e-05,
      "loss": 4.5505,
      "step": 43550
    },
    {
      "epoch": 0.8920197407490836,
      "grad_norm": 0.3340022563934326,
      "learning_rate": 4e-05,
      "loss": 4.5584,
      "step": 43560
    },
    {
      "epoch": 0.8922245203038929,
      "grad_norm": 0.3266768753528595,
      "learning_rate": 4e-05,
      "loss": 4.5653,
      "step": 43570
    },
    {
      "epoch": 0.8924292998587021,
      "grad_norm": 0.3193603456020355,
      "learning_rate": 4e-05,
      "loss": 4.569,
      "step": 43580
    },
    {
      "epoch": 0.8926340794135114,
      "grad_norm": 0.35704925656318665,
      "learning_rate": 4e-05,
      "loss": 4.5703,
      "step": 43590
    },
    {
      "epoch": 0.8928388589683206,
      "grad_norm": 0.30354368686676025,
      "learning_rate": 4e-05,
      "loss": 4.549,
      "step": 43600
    },
    {
      "epoch": 0.8930436385231298,
      "grad_norm": 0.344433069229126,
      "learning_rate": 4e-05,
      "loss": 4.5633,
      "step": 43610
    },
    {
      "epoch": 0.8932484180779391,
      "grad_norm": 0.32204896211624146,
      "learning_rate": 4e-05,
      "loss": 4.557,
      "step": 43620
    },
    {
      "epoch": 0.8934531976327483,
      "grad_norm": 0.3304206430912018,
      "learning_rate": 4e-05,
      "loss": 4.5733,
      "step": 43630
    },
    {
      "epoch": 0.8936579771875576,
      "grad_norm": 0.30991101264953613,
      "learning_rate": 4e-05,
      "loss": 4.5768,
      "step": 43640
    },
    {
      "epoch": 0.8938627567423668,
      "grad_norm": 0.3452933728694916,
      "learning_rate": 4e-05,
      "loss": 4.5519,
      "step": 43650
    },
    {
      "epoch": 0.8940675362971761,
      "grad_norm": 0.36112919449806213,
      "learning_rate": 4e-05,
      "loss": 4.5493,
      "step": 43660
    },
    {
      "epoch": 0.8942723158519853,
      "grad_norm": 0.30103543400764465,
      "learning_rate": 4e-05,
      "loss": 4.5507,
      "step": 43670
    },
    {
      "epoch": 0.8944770954067945,
      "grad_norm": 0.32138338685035706,
      "learning_rate": 4e-05,
      "loss": 4.5913,
      "step": 43680
    },
    {
      "epoch": 0.8946818749616038,
      "grad_norm": 0.30917757749557495,
      "learning_rate": 4e-05,
      "loss": 4.5479,
      "step": 43690
    },
    {
      "epoch": 0.894886654516413,
      "grad_norm": 0.3278537392616272,
      "learning_rate": 4e-05,
      "loss": 4.5493,
      "step": 43700
    },
    {
      "epoch": 0.8950914340712224,
      "grad_norm": 0.3118800222873688,
      "learning_rate": 4e-05,
      "loss": 4.5413,
      "step": 43710
    },
    {
      "epoch": 0.8952962136260316,
      "grad_norm": 0.4983890950679779,
      "learning_rate": 4e-05,
      "loss": 4.5461,
      "step": 43720
    },
    {
      "epoch": 0.8955009931808409,
      "grad_norm": 0.322793185710907,
      "learning_rate": 4e-05,
      "loss": 4.5156,
      "step": 43730
    },
    {
      "epoch": 0.8957057727356501,
      "grad_norm": 0.34943801164627075,
      "learning_rate": 4e-05,
      "loss": 4.5893,
      "step": 43740
    },
    {
      "epoch": 0.8959105522904593,
      "grad_norm": 0.546573281288147,
      "learning_rate": 4e-05,
      "loss": 4.5652,
      "step": 43750
    },
    {
      "epoch": 0.8961153318452686,
      "grad_norm": 0.4451698362827301,
      "learning_rate": 4e-05,
      "loss": 4.5489,
      "step": 43760
    },
    {
      "epoch": 0.8963201114000778,
      "grad_norm": 0.32471200823783875,
      "learning_rate": 4e-05,
      "loss": 4.5438,
      "step": 43770
    },
    {
      "epoch": 0.8965248909548871,
      "grad_norm": 0.3037568926811218,
      "learning_rate": 4e-05,
      "loss": 4.5591,
      "step": 43780
    },
    {
      "epoch": 0.8967296705096963,
      "grad_norm": 0.3210413157939911,
      "learning_rate": 4e-05,
      "loss": 4.5663,
      "step": 43790
    },
    {
      "epoch": 0.8969344500645056,
      "grad_norm": 0.33913564682006836,
      "learning_rate": 4e-05,
      "loss": 4.5989,
      "step": 43800
    },
    {
      "epoch": 0.8971392296193148,
      "grad_norm": 0.3082035481929779,
      "learning_rate": 4e-05,
      "loss": 4.5827,
      "step": 43810
    },
    {
      "epoch": 0.897344009174124,
      "grad_norm": 0.34691470861434937,
      "learning_rate": 4e-05,
      "loss": 4.5176,
      "step": 43820
    },
    {
      "epoch": 0.8975487887289333,
      "grad_norm": 0.3143591284751892,
      "learning_rate": 4e-05,
      "loss": 4.5473,
      "step": 43830
    },
    {
      "epoch": 0.8977535682837425,
      "grad_norm": 0.2929944396018982,
      "learning_rate": 4e-05,
      "loss": 4.5722,
      "step": 43840
    },
    {
      "epoch": 0.8979583478385518,
      "grad_norm": 0.3059447705745697,
      "learning_rate": 4e-05,
      "loss": 4.5399,
      "step": 43850
    },
    {
      "epoch": 0.898163127393361,
      "grad_norm": 0.3196268081665039,
      "learning_rate": 4e-05,
      "loss": 4.5565,
      "step": 43860
    },
    {
      "epoch": 0.8983679069481703,
      "grad_norm": 0.3168346881866455,
      "learning_rate": 4e-05,
      "loss": 4.5745,
      "step": 43870
    },
    {
      "epoch": 0.8985726865029795,
      "grad_norm": 0.31572356820106506,
      "learning_rate": 4e-05,
      "loss": 4.5789,
      "step": 43880
    },
    {
      "epoch": 0.8987774660577887,
      "grad_norm": 0.30310994386672974,
      "learning_rate": 4e-05,
      "loss": 4.5478,
      "step": 43890
    },
    {
      "epoch": 0.898982245612598,
      "grad_norm": 0.3090900480747223,
      "learning_rate": 4e-05,
      "loss": 4.5711,
      "step": 43900
    },
    {
      "epoch": 0.8991870251674073,
      "grad_norm": 0.31497764587402344,
      "learning_rate": 4e-05,
      "loss": 4.5667,
      "step": 43910
    },
    {
      "epoch": 0.8993918047222166,
      "grad_norm": 0.3240913152694702,
      "learning_rate": 4e-05,
      "loss": 4.5514,
      "step": 43920
    },
    {
      "epoch": 0.8995965842770258,
      "grad_norm": 0.3159175515174866,
      "learning_rate": 4e-05,
      "loss": 4.5659,
      "step": 43930
    },
    {
      "epoch": 0.8998013638318351,
      "grad_norm": 0.30631309747695923,
      "learning_rate": 4e-05,
      "loss": 4.5506,
      "step": 43940
    },
    {
      "epoch": 0.9000061433866443,
      "grad_norm": 0.3201664984226227,
      "learning_rate": 4e-05,
      "loss": 4.535,
      "step": 43950
    },
    {
      "epoch": 0.9002109229414536,
      "grad_norm": 0.32394763827323914,
      "learning_rate": 4e-05,
      "loss": 4.5698,
      "step": 43960
    },
    {
      "epoch": 0.9004157024962628,
      "grad_norm": 0.3251607120037079,
      "learning_rate": 4e-05,
      "loss": 4.5394,
      "step": 43970
    },
    {
      "epoch": 0.900620482051072,
      "grad_norm": 0.3281831443309784,
      "learning_rate": 4e-05,
      "loss": 4.5592,
      "step": 43980
    },
    {
      "epoch": 0.9008252616058813,
      "grad_norm": 0.32417479157447815,
      "learning_rate": 4e-05,
      "loss": 4.5466,
      "step": 43990
    },
    {
      "epoch": 0.9010300411606905,
      "grad_norm": 0.3111242651939392,
      "learning_rate": 4e-05,
      "loss": 4.5398,
      "step": 44000
    },
    {
      "epoch": 0.9010300411606905,
      "eval_loss": 4.560854434967041,
      "eval_runtime": 4.4095,
      "eval_samples_per_second": 264.43,
      "eval_steps_per_second": 33.11,
      "step": 44000
    },
    {
      "epoch": 0.9012348207154998,
      "grad_norm": 0.3045864999294281,
      "learning_rate": 4e-05,
      "loss": 4.5549,
      "step": 44010
    },
    {
      "epoch": 0.901439600270309,
      "grad_norm": 0.30493706464767456,
      "learning_rate": 4e-05,
      "loss": 4.5632,
      "step": 44020
    },
    {
      "epoch": 0.9016443798251182,
      "grad_norm": 0.338446706533432,
      "learning_rate": 4e-05,
      "loss": 4.5729,
      "step": 44030
    },
    {
      "epoch": 0.9018491593799275,
      "grad_norm": 0.3336330056190491,
      "learning_rate": 4e-05,
      "loss": 4.533,
      "step": 44040
    },
    {
      "epoch": 0.9020539389347367,
      "grad_norm": 0.33428311347961426,
      "learning_rate": 4e-05,
      "loss": 4.5455,
      "step": 44050
    },
    {
      "epoch": 0.902258718489546,
      "grad_norm": 0.3272670805454254,
      "learning_rate": 4e-05,
      "loss": 4.5272,
      "step": 44060
    },
    {
      "epoch": 0.9024634980443552,
      "grad_norm": 0.33472102880477905,
      "learning_rate": 4e-05,
      "loss": 4.5763,
      "step": 44070
    },
    {
      "epoch": 0.9026682775991645,
      "grad_norm": 0.3110867440700531,
      "learning_rate": 4e-05,
      "loss": 4.5516,
      "step": 44080
    },
    {
      "epoch": 0.9028730571539737,
      "grad_norm": 0.3802002966403961,
      "learning_rate": 4e-05,
      "loss": 4.5858,
      "step": 44090
    },
    {
      "epoch": 0.9030778367087829,
      "grad_norm": 0.3172958493232727,
      "learning_rate": 4e-05,
      "loss": 4.5838,
      "step": 44100
    },
    {
      "epoch": 0.9032826162635923,
      "grad_norm": 0.32478705048561096,
      "learning_rate": 4e-05,
      "loss": 4.5695,
      "step": 44110
    },
    {
      "epoch": 0.9034873958184015,
      "grad_norm": 0.32184964418411255,
      "learning_rate": 4e-05,
      "loss": 4.597,
      "step": 44120
    },
    {
      "epoch": 0.9036921753732108,
      "grad_norm": 0.3221733272075653,
      "learning_rate": 4e-05,
      "loss": 4.5722,
      "step": 44130
    },
    {
      "epoch": 0.90389695492802,
      "grad_norm": 0.31805217266082764,
      "learning_rate": 4e-05,
      "loss": 4.6067,
      "step": 44140
    },
    {
      "epoch": 0.9041017344828293,
      "grad_norm": 0.3431871235370636,
      "learning_rate": 4e-05,
      "loss": 4.5375,
      "step": 44150
    },
    {
      "epoch": 0.9043065140376385,
      "grad_norm": 0.302630752325058,
      "learning_rate": 4e-05,
      "loss": 4.5408,
      "step": 44160
    },
    {
      "epoch": 0.9045112935924478,
      "grad_norm": 0.31336262822151184,
      "learning_rate": 4e-05,
      "loss": 4.5731,
      "step": 44170
    },
    {
      "epoch": 0.904716073147257,
      "grad_norm": 0.3739093244075775,
      "learning_rate": 4e-05,
      "loss": 4.5485,
      "step": 44180
    },
    {
      "epoch": 0.9049208527020662,
      "grad_norm": 0.31083807349205017,
      "learning_rate": 4e-05,
      "loss": 4.556,
      "step": 44190
    },
    {
      "epoch": 0.9051256322568755,
      "grad_norm": 0.31648892164230347,
      "learning_rate": 4e-05,
      "loss": 4.5896,
      "step": 44200
    },
    {
      "epoch": 0.9053304118116847,
      "grad_norm": 0.3103238344192505,
      "learning_rate": 4e-05,
      "loss": 4.5782,
      "step": 44210
    },
    {
      "epoch": 0.905535191366494,
      "grad_norm": 0.3150661289691925,
      "learning_rate": 4e-05,
      "loss": 4.5722,
      "step": 44220
    },
    {
      "epoch": 0.9057399709213032,
      "grad_norm": 0.30225515365600586,
      "learning_rate": 4e-05,
      "loss": 4.5595,
      "step": 44230
    },
    {
      "epoch": 0.9059447504761124,
      "grad_norm": 0.31010523438453674,
      "learning_rate": 4e-05,
      "loss": 4.5365,
      "step": 44240
    },
    {
      "epoch": 0.9061495300309217,
      "grad_norm": 0.3108595013618469,
      "learning_rate": 4e-05,
      "loss": 4.5376,
      "step": 44250
    },
    {
      "epoch": 0.9063543095857309,
      "grad_norm": 0.31890687346458435,
      "learning_rate": 4e-05,
      "loss": 4.5588,
      "step": 44260
    },
    {
      "epoch": 0.9065590891405402,
      "grad_norm": 0.3646460473537445,
      "learning_rate": 4e-05,
      "loss": 4.5203,
      "step": 44270
    },
    {
      "epoch": 0.9067638686953494,
      "grad_norm": 0.33907946944236755,
      "learning_rate": 4e-05,
      "loss": 4.571,
      "step": 44280
    },
    {
      "epoch": 0.9069686482501587,
      "grad_norm": 0.34278184175491333,
      "learning_rate": 4e-05,
      "loss": 4.5299,
      "step": 44290
    },
    {
      "epoch": 0.9071734278049679,
      "grad_norm": 0.30765822529792786,
      "learning_rate": 4e-05,
      "loss": 4.5202,
      "step": 44300
    },
    {
      "epoch": 0.9073782073597771,
      "grad_norm": 0.32606104016304016,
      "learning_rate": 4e-05,
      "loss": 4.576,
      "step": 44310
    },
    {
      "epoch": 0.9075829869145865,
      "grad_norm": 0.3165709376335144,
      "learning_rate": 4e-05,
      "loss": 4.5648,
      "step": 44320
    },
    {
      "epoch": 0.9077877664693957,
      "grad_norm": 0.31286153197288513,
      "learning_rate": 4e-05,
      "loss": 4.5698,
      "step": 44330
    },
    {
      "epoch": 0.907992546024205,
      "grad_norm": 0.3222188353538513,
      "learning_rate": 4e-05,
      "loss": 4.5682,
      "step": 44340
    },
    {
      "epoch": 0.9081973255790142,
      "grad_norm": 0.332038015127182,
      "learning_rate": 4e-05,
      "loss": 4.5676,
      "step": 44350
    },
    {
      "epoch": 0.9084021051338235,
      "grad_norm": 0.32790395617485046,
      "learning_rate": 4e-05,
      "loss": 4.585,
      "step": 44360
    },
    {
      "epoch": 0.9086068846886327,
      "grad_norm": 0.3362048864364624,
      "learning_rate": 4e-05,
      "loss": 4.5528,
      "step": 44370
    },
    {
      "epoch": 0.908811664243442,
      "grad_norm": 0.32595202326774597,
      "learning_rate": 4e-05,
      "loss": 4.563,
      "step": 44380
    },
    {
      "epoch": 0.9090164437982512,
      "grad_norm": 0.3537389636039734,
      "learning_rate": 4e-05,
      "loss": 4.5868,
      "step": 44390
    },
    {
      "epoch": 0.9092212233530604,
      "grad_norm": 0.3152723014354706,
      "learning_rate": 4e-05,
      "loss": 4.567,
      "step": 44400
    },
    {
      "epoch": 0.9094260029078697,
      "grad_norm": 0.310278058052063,
      "learning_rate": 4e-05,
      "loss": 4.5368,
      "step": 44410
    },
    {
      "epoch": 0.9096307824626789,
      "grad_norm": 0.3246089518070221,
      "learning_rate": 4e-05,
      "loss": 4.5708,
      "step": 44420
    },
    {
      "epoch": 0.9098355620174882,
      "grad_norm": 0.3013988435268402,
      "learning_rate": 4e-05,
      "loss": 4.5438,
      "step": 44430
    },
    {
      "epoch": 0.9100403415722974,
      "grad_norm": 0.33126717805862427,
      "learning_rate": 4e-05,
      "loss": 4.5729,
      "step": 44440
    },
    {
      "epoch": 0.9102451211271066,
      "grad_norm": 0.31058260798454285,
      "learning_rate": 4e-05,
      "loss": 4.5681,
      "step": 44450
    },
    {
      "epoch": 0.9104499006819159,
      "grad_norm": 0.5047004818916321,
      "learning_rate": 4e-05,
      "loss": 4.5414,
      "step": 44460
    },
    {
      "epoch": 0.9106546802367251,
      "grad_norm": 0.3140256702899933,
      "learning_rate": 4e-05,
      "loss": 4.5565,
      "step": 44470
    },
    {
      "epoch": 0.9108594597915344,
      "grad_norm": 0.42531779408454895,
      "learning_rate": 4e-05,
      "loss": 4.5674,
      "step": 44480
    },
    {
      "epoch": 0.9110642393463436,
      "grad_norm": 0.3069171905517578,
      "learning_rate": 4e-05,
      "loss": 4.388,
      "step": 44490
    },
    {
      "epoch": 0.9112690189011529,
      "grad_norm": 0.3190549314022064,
      "learning_rate": 4e-05,
      "loss": 4.5512,
      "step": 44500
    },
    {
      "epoch": 0.9114737984559621,
      "grad_norm": 0.31640002131462097,
      "learning_rate": 4e-05,
      "loss": 4.5761,
      "step": 44510
    },
    {
      "epoch": 0.9116785780107715,
      "grad_norm": 0.31353873014450073,
      "learning_rate": 4e-05,
      "loss": 4.5616,
      "step": 44520
    },
    {
      "epoch": 0.9118833575655807,
      "grad_norm": 0.3220072090625763,
      "learning_rate": 4e-05,
      "loss": 4.5697,
      "step": 44530
    },
    {
      "epoch": 0.9120881371203899,
      "grad_norm": 0.35407111048698425,
      "learning_rate": 4e-05,
      "loss": 4.5294,
      "step": 44540
    },
    {
      "epoch": 0.9122929166751992,
      "grad_norm": 0.3476737439632416,
      "learning_rate": 4e-05,
      "loss": 4.5531,
      "step": 44550
    },
    {
      "epoch": 0.9124976962300084,
      "grad_norm": 0.32941582798957825,
      "learning_rate": 4e-05,
      "loss": 4.5444,
      "step": 44560
    },
    {
      "epoch": 0.9127024757848177,
      "grad_norm": 0.31432151794433594,
      "learning_rate": 4e-05,
      "loss": 4.6001,
      "step": 44570
    },
    {
      "epoch": 0.9129072553396269,
      "grad_norm": 0.3961334228515625,
      "learning_rate": 4e-05,
      "loss": 4.578,
      "step": 44580
    },
    {
      "epoch": 0.9131120348944362,
      "grad_norm": 0.8139594793319702,
      "learning_rate": 4e-05,
      "loss": 4.5495,
      "step": 44590
    },
    {
      "epoch": 0.9133168144492454,
      "grad_norm": 0.3410652279853821,
      "learning_rate": 4e-05,
      "loss": 4.6008,
      "step": 44600
    },
    {
      "epoch": 0.9135215940040546,
      "grad_norm": 0.30633702874183655,
      "learning_rate": 4e-05,
      "loss": 4.5491,
      "step": 44610
    },
    {
      "epoch": 0.9137263735588639,
      "grad_norm": 0.3259240388870239,
      "learning_rate": 4e-05,
      "loss": 4.556,
      "step": 44620
    },
    {
      "epoch": 0.9139311531136731,
      "grad_norm": 0.33315950632095337,
      "learning_rate": 4e-05,
      "loss": 4.5631,
      "step": 44630
    },
    {
      "epoch": 0.9141359326684824,
      "grad_norm": 0.3183877170085907,
      "learning_rate": 4e-05,
      "loss": 4.5838,
      "step": 44640
    },
    {
      "epoch": 0.9143407122232916,
      "grad_norm": 0.3292442858219147,
      "learning_rate": 4e-05,
      "loss": 4.5895,
      "step": 44650
    },
    {
      "epoch": 0.9145454917781008,
      "grad_norm": 0.33461958169937134,
      "learning_rate": 4e-05,
      "loss": 4.5707,
      "step": 44660
    },
    {
      "epoch": 0.9147502713329101,
      "grad_norm": 0.5829816460609436,
      "learning_rate": 4e-05,
      "loss": 4.5615,
      "step": 44670
    },
    {
      "epoch": 0.9149550508877193,
      "grad_norm": 0.31403544545173645,
      "learning_rate": 4e-05,
      "loss": 4.5727,
      "step": 44680
    },
    {
      "epoch": 0.9151598304425286,
      "grad_norm": 0.3185792863368988,
      "learning_rate": 4e-05,
      "loss": 4.5475,
      "step": 44690
    },
    {
      "epoch": 0.9153646099973378,
      "grad_norm": 0.34621044993400574,
      "learning_rate": 4e-05,
      "loss": 4.5746,
      "step": 44700
    },
    {
      "epoch": 0.9155693895521471,
      "grad_norm": 0.29865196347236633,
      "learning_rate": 4e-05,
      "loss": 4.5576,
      "step": 44710
    },
    {
      "epoch": 0.9157741691069564,
      "grad_norm": 0.3081812560558319,
      "learning_rate": 4e-05,
      "loss": 4.5532,
      "step": 44720
    },
    {
      "epoch": 0.9159789486617657,
      "grad_norm": 0.32022973895072937,
      "learning_rate": 4e-05,
      "loss": 4.5551,
      "step": 44730
    },
    {
      "epoch": 0.9161837282165749,
      "grad_norm": 0.318764865398407,
      "learning_rate": 4e-05,
      "loss": 4.524,
      "step": 44740
    },
    {
      "epoch": 0.9163885077713841,
      "grad_norm": 0.31628209352493286,
      "learning_rate": 4e-05,
      "loss": 4.5506,
      "step": 44750
    },
    {
      "epoch": 0.9165932873261934,
      "grad_norm": 0.3148849308490753,
      "learning_rate": 4e-05,
      "loss": 4.533,
      "step": 44760
    },
    {
      "epoch": 0.9167980668810026,
      "grad_norm": 0.3610849380493164,
      "learning_rate": 4e-05,
      "loss": 4.5294,
      "step": 44770
    },
    {
      "epoch": 0.9170028464358119,
      "grad_norm": 0.31631144881248474,
      "learning_rate": 4e-05,
      "loss": 4.5483,
      "step": 44780
    },
    {
      "epoch": 0.9172076259906211,
      "grad_norm": 0.3129100203514099,
      "learning_rate": 4e-05,
      "loss": 4.5571,
      "step": 44790
    },
    {
      "epoch": 0.9174124055454304,
      "grad_norm": 0.4124535918235779,
      "learning_rate": 4e-05,
      "loss": 4.5468,
      "step": 44800
    },
    {
      "epoch": 0.9176171851002396,
      "grad_norm": 0.3188845217227936,
      "learning_rate": 4e-05,
      "loss": 4.5311,
      "step": 44810
    },
    {
      "epoch": 0.9178219646550488,
      "grad_norm": 0.32192277908325195,
      "learning_rate": 4e-05,
      "loss": 4.5175,
      "step": 44820
    },
    {
      "epoch": 0.9180267442098581,
      "grad_norm": 0.31410878896713257,
      "learning_rate": 4e-05,
      "loss": 4.5588,
      "step": 44830
    },
    {
      "epoch": 0.9182315237646673,
      "grad_norm": 0.32310882210731506,
      "learning_rate": 4e-05,
      "loss": 4.5471,
      "step": 44840
    },
    {
      "epoch": 0.9184363033194766,
      "grad_norm": 0.3387526273727417,
      "learning_rate": 4e-05,
      "loss": 4.5345,
      "step": 44850
    },
    {
      "epoch": 0.9186410828742858,
      "grad_norm": 0.3155510425567627,
      "learning_rate": 4e-05,
      "loss": 4.6002,
      "step": 44860
    },
    {
      "epoch": 0.918845862429095,
      "grad_norm": 0.3108542859554291,
      "learning_rate": 4e-05,
      "loss": 4.5653,
      "step": 44870
    },
    {
      "epoch": 0.9190506419839043,
      "grad_norm": 0.30360421538352966,
      "learning_rate": 4e-05,
      "loss": 4.565,
      "step": 44880
    },
    {
      "epoch": 0.9192554215387135,
      "grad_norm": 0.29860028624534607,
      "learning_rate": 4e-05,
      "loss": 4.5713,
      "step": 44890
    },
    {
      "epoch": 0.9194602010935228,
      "grad_norm": 0.32581397891044617,
      "learning_rate": 4e-05,
      "loss": 4.5627,
      "step": 44900
    },
    {
      "epoch": 0.919664980648332,
      "grad_norm": 0.32004186511039734,
      "learning_rate": 4e-05,
      "loss": 4.5541,
      "step": 44910
    },
    {
      "epoch": 0.9198697602031414,
      "grad_norm": 0.3210557699203491,
      "learning_rate": 4e-05,
      "loss": 4.5672,
      "step": 44920
    },
    {
      "epoch": 0.9200745397579506,
      "grad_norm": 0.34228792786598206,
      "learning_rate": 4e-05,
      "loss": 4.5653,
      "step": 44930
    },
    {
      "epoch": 0.9202793193127599,
      "grad_norm": 0.3095720410346985,
      "learning_rate": 4e-05,
      "loss": 4.5762,
      "step": 44940
    },
    {
      "epoch": 0.9204840988675691,
      "grad_norm": 0.34053659439086914,
      "learning_rate": 4e-05,
      "loss": 4.5799,
      "step": 44950
    },
    {
      "epoch": 0.9206888784223783,
      "grad_norm": 0.37338224053382874,
      "learning_rate": 4e-05,
      "loss": 4.5936,
      "step": 44960
    },
    {
      "epoch": 0.9208936579771876,
      "grad_norm": 0.3042196035385132,
      "learning_rate": 4e-05,
      "loss": 4.5614,
      "step": 44970
    },
    {
      "epoch": 0.9210984375319968,
      "grad_norm": 0.31008726358413696,
      "learning_rate": 4e-05,
      "loss": 4.5336,
      "step": 44980
    },
    {
      "epoch": 0.9213032170868061,
      "grad_norm": 0.3075076937675476,
      "learning_rate": 4e-05,
      "loss": 4.5432,
      "step": 44990
    },
    {
      "epoch": 0.9215079966416153,
      "grad_norm": 0.4271358549594879,
      "learning_rate": 4e-05,
      "loss": 4.5865,
      "step": 45000
    },
    {
      "epoch": 0.9217127761964246,
      "grad_norm": 0.3291766941547394,
      "learning_rate": 4e-05,
      "loss": 4.5299,
      "step": 45010
    },
    {
      "epoch": 0.9219175557512338,
      "grad_norm": 0.34627512097358704,
      "learning_rate": 4e-05,
      "loss": 4.5393,
      "step": 45020
    },
    {
      "epoch": 0.922122335306043,
      "grad_norm": 0.31433627009391785,
      "learning_rate": 4e-05,
      "loss": 4.569,
      "step": 45030
    },
    {
      "epoch": 0.9223271148608523,
      "grad_norm": 0.33536338806152344,
      "learning_rate": 4e-05,
      "loss": 4.5606,
      "step": 45040
    },
    {
      "epoch": 0.9225318944156615,
      "grad_norm": 0.31146395206451416,
      "learning_rate": 4e-05,
      "loss": 4.5485,
      "step": 45050
    },
    {
      "epoch": 0.9227366739704708,
      "grad_norm": 0.35025808215141296,
      "learning_rate": 4e-05,
      "loss": 4.5272,
      "step": 45060
    },
    {
      "epoch": 0.92294145352528,
      "grad_norm": 0.318724125623703,
      "learning_rate": 4e-05,
      "loss": 4.6059,
      "step": 45070
    },
    {
      "epoch": 0.9231462330800893,
      "grad_norm": 0.31548213958740234,
      "learning_rate": 4e-05,
      "loss": 4.5931,
      "step": 45080
    },
    {
      "epoch": 0.9233510126348985,
      "grad_norm": 0.35086187720298767,
      "learning_rate": 4e-05,
      "loss": 4.5536,
      "step": 45090
    },
    {
      "epoch": 0.9235557921897077,
      "grad_norm": 0.44874176383018494,
      "learning_rate": 4e-05,
      "loss": 4.5411,
      "step": 45100
    },
    {
      "epoch": 0.923760571744517,
      "grad_norm": 0.3447568118572235,
      "learning_rate": 4e-05,
      "loss": 4.541,
      "step": 45110
    },
    {
      "epoch": 0.9239653512993262,
      "grad_norm": 0.34852394461631775,
      "learning_rate": 4e-05,
      "loss": 4.5474,
      "step": 45120
    },
    {
      "epoch": 0.9241701308541356,
      "grad_norm": 0.3143792152404785,
      "learning_rate": 4e-05,
      "loss": 4.5602,
      "step": 45130
    },
    {
      "epoch": 0.9243749104089448,
      "grad_norm": 0.3151521682739258,
      "learning_rate": 4e-05,
      "loss": 4.5705,
      "step": 45140
    },
    {
      "epoch": 0.9245796899637541,
      "grad_norm": 0.34277409315109253,
      "learning_rate": 4e-05,
      "loss": 4.5778,
      "step": 45150
    },
    {
      "epoch": 0.9247844695185633,
      "grad_norm": 0.3213508427143097,
      "learning_rate": 4e-05,
      "loss": 4.562,
      "step": 45160
    },
    {
      "epoch": 0.9249892490733725,
      "grad_norm": 0.30300119519233704,
      "learning_rate": 4e-05,
      "loss": 4.5725,
      "step": 45170
    },
    {
      "epoch": 0.9251940286281818,
      "grad_norm": 0.3328953981399536,
      "learning_rate": 4e-05,
      "loss": 4.5863,
      "step": 45180
    },
    {
      "epoch": 0.925398808182991,
      "grad_norm": 0.342203289270401,
      "learning_rate": 4e-05,
      "loss": 4.5504,
      "step": 45190
    },
    {
      "epoch": 0.9256035877378003,
      "grad_norm": 0.3172464966773987,
      "learning_rate": 4e-05,
      "loss": 4.586,
      "step": 45200
    },
    {
      "epoch": 0.9258083672926095,
      "grad_norm": 0.3222777247428894,
      "learning_rate": 4e-05,
      "loss": 4.5512,
      "step": 45210
    },
    {
      "epoch": 0.9260131468474188,
      "grad_norm": 0.31430578231811523,
      "learning_rate": 4e-05,
      "loss": 4.5716,
      "step": 45220
    },
    {
      "epoch": 0.926217926402228,
      "grad_norm": 0.3673795759677887,
      "learning_rate": 4e-05,
      "loss": 4.5336,
      "step": 45230
    },
    {
      "epoch": 0.9264227059570372,
      "grad_norm": 0.35292544960975647,
      "learning_rate": 4e-05,
      "loss": 4.5614,
      "step": 45240
    },
    {
      "epoch": 0.9266274855118465,
      "grad_norm": 0.31124240159988403,
      "learning_rate": 4e-05,
      "loss": 4.5489,
      "step": 45250
    },
    {
      "epoch": 0.9268322650666557,
      "grad_norm": 0.2988051772117615,
      "learning_rate": 4e-05,
      "loss": 4.5485,
      "step": 45260
    },
    {
      "epoch": 0.927037044621465,
      "grad_norm": 0.314090371131897,
      "learning_rate": 4e-05,
      "loss": 4.5719,
      "step": 45270
    },
    {
      "epoch": 0.9272418241762742,
      "grad_norm": 0.32014816999435425,
      "learning_rate": 4e-05,
      "loss": 4.601,
      "step": 45280
    },
    {
      "epoch": 0.9274466037310835,
      "grad_norm": 0.32191765308380127,
      "learning_rate": 4e-05,
      "loss": 4.5615,
      "step": 45290
    },
    {
      "epoch": 0.9276513832858927,
      "grad_norm": 0.33087530732154846,
      "learning_rate": 4e-05,
      "loss": 4.6015,
      "step": 45300
    },
    {
      "epoch": 0.9278561628407019,
      "grad_norm": 0.3054031729698181,
      "learning_rate": 4e-05,
      "loss": 4.5461,
      "step": 45310
    },
    {
      "epoch": 0.9280609423955112,
      "grad_norm": 0.31847915053367615,
      "learning_rate": 4e-05,
      "loss": 4.5764,
      "step": 45320
    },
    {
      "epoch": 0.9282657219503205,
      "grad_norm": 0.31255143880844116,
      "learning_rate": 4e-05,
      "loss": 4.5722,
      "step": 45330
    },
    {
      "epoch": 0.9284705015051298,
      "grad_norm": 0.3137277364730835,
      "learning_rate": 4e-05,
      "loss": 4.5806,
      "step": 45340
    },
    {
      "epoch": 0.928675281059939,
      "grad_norm": 0.30868300795555115,
      "learning_rate": 4e-05,
      "loss": 4.561,
      "step": 45350
    },
    {
      "epoch": 0.9288800606147483,
      "grad_norm": 0.3062467873096466,
      "learning_rate": 4e-05,
      "loss": 4.5745,
      "step": 45360
    },
    {
      "epoch": 0.9290848401695575,
      "grad_norm": 0.3123590648174286,
      "learning_rate": 4e-05,
      "loss": 4.5523,
      "step": 45370
    },
    {
      "epoch": 0.9292896197243667,
      "grad_norm": 0.34719258546829224,
      "learning_rate": 4e-05,
      "loss": 4.5775,
      "step": 45380
    },
    {
      "epoch": 0.929494399279176,
      "grad_norm": 0.33141425251960754,
      "learning_rate": 4e-05,
      "loss": 4.5424,
      "step": 45390
    },
    {
      "epoch": 0.9296991788339852,
      "grad_norm": 0.35615402460098267,
      "learning_rate": 4e-05,
      "loss": 4.5491,
      "step": 45400
    },
    {
      "epoch": 0.9299039583887945,
      "grad_norm": 0.31258466839790344,
      "learning_rate": 4e-05,
      "loss": 4.5409,
      "step": 45410
    },
    {
      "epoch": 0.9301087379436037,
      "grad_norm": 0.3252570629119873,
      "learning_rate": 4e-05,
      "loss": 4.6003,
      "step": 45420
    },
    {
      "epoch": 0.930313517498413,
      "grad_norm": 0.3250385820865631,
      "learning_rate": 4e-05,
      "loss": 4.5763,
      "step": 45430
    },
    {
      "epoch": 0.9305182970532222,
      "grad_norm": 0.32301104068756104,
      "learning_rate": 4e-05,
      "loss": 4.5527,
      "step": 45440
    },
    {
      "epoch": 0.9307230766080314,
      "grad_norm": 0.33571791648864746,
      "learning_rate": 4e-05,
      "loss": 4.5366,
      "step": 45450
    },
    {
      "epoch": 0.9309278561628407,
      "grad_norm": 0.3223508894443512,
      "learning_rate": 4e-05,
      "loss": 4.5766,
      "step": 45460
    },
    {
      "epoch": 0.9311326357176499,
      "grad_norm": 0.38597506284713745,
      "learning_rate": 4e-05,
      "loss": 4.5291,
      "step": 45470
    },
    {
      "epoch": 0.9313374152724592,
      "grad_norm": 0.31094756722450256,
      "learning_rate": 4e-05,
      "loss": 4.5959,
      "step": 45480
    },
    {
      "epoch": 0.9315421948272684,
      "grad_norm": 0.3371400237083435,
      "learning_rate": 4e-05,
      "loss": 4.5605,
      "step": 45490
    },
    {
      "epoch": 0.9317469743820777,
      "grad_norm": 0.3200722932815552,
      "learning_rate": 4e-05,
      "loss": 4.5638,
      "step": 45500
    },
    {
      "epoch": 0.9319517539368869,
      "grad_norm": 0.3067943751811981,
      "learning_rate": 4e-05,
      "loss": 4.5505,
      "step": 45510
    },
    {
      "epoch": 0.9321565334916961,
      "grad_norm": 0.3259231746196747,
      "learning_rate": 4e-05,
      "loss": 4.5644,
      "step": 45520
    },
    {
      "epoch": 0.9323613130465055,
      "grad_norm": 0.3075989782810211,
      "learning_rate": 4e-05,
      "loss": 4.5436,
      "step": 45530
    },
    {
      "epoch": 0.9325660926013147,
      "grad_norm": 0.32256752252578735,
      "learning_rate": 4e-05,
      "loss": 4.5671,
      "step": 45540
    },
    {
      "epoch": 0.932770872156124,
      "grad_norm": 0.3286307156085968,
      "learning_rate": 4e-05,
      "loss": 4.5673,
      "step": 45550
    },
    {
      "epoch": 0.9329756517109332,
      "grad_norm": 0.3804124891757965,
      "learning_rate": 4e-05,
      "loss": 4.5413,
      "step": 45560
    },
    {
      "epoch": 0.9331804312657425,
      "grad_norm": 0.3336116373538971,
      "learning_rate": 4e-05,
      "loss": 4.4751,
      "step": 45570
    },
    {
      "epoch": 0.9333852108205517,
      "grad_norm": 0.31247478723526,
      "learning_rate": 4e-05,
      "loss": 4.6037,
      "step": 45580
    },
    {
      "epoch": 0.933589990375361,
      "grad_norm": 0.32099565863609314,
      "learning_rate": 4e-05,
      "loss": 4.5669,
      "step": 45590
    },
    {
      "epoch": 0.9337947699301702,
      "grad_norm": 0.316138356924057,
      "learning_rate": 4e-05,
      "loss": 4.5341,
      "step": 45600
    },
    {
      "epoch": 0.9339995494849794,
      "grad_norm": 0.3377699851989746,
      "learning_rate": 4e-05,
      "loss": 4.56,
      "step": 45610
    },
    {
      "epoch": 0.9342043290397887,
      "grad_norm": 0.3313315510749817,
      "learning_rate": 4e-05,
      "loss": 4.5437,
      "step": 45620
    },
    {
      "epoch": 0.9344091085945979,
      "grad_norm": 0.318208783864975,
      "learning_rate": 4e-05,
      "loss": 4.5828,
      "step": 45630
    },
    {
      "epoch": 0.9346138881494072,
      "grad_norm": 0.33342334628105164,
      "learning_rate": 4e-05,
      "loss": 4.5715,
      "step": 45640
    },
    {
      "epoch": 0.9348186677042164,
      "grad_norm": 0.3693804442882538,
      "learning_rate": 4e-05,
      "loss": 4.5503,
      "step": 45650
    },
    {
      "epoch": 0.9350234472590256,
      "grad_norm": 0.3329488933086395,
      "learning_rate": 4e-05,
      "loss": 4.5663,
      "step": 45660
    },
    {
      "epoch": 0.9352282268138349,
      "grad_norm": 0.3339942693710327,
      "learning_rate": 4e-05,
      "loss": 4.5704,
      "step": 45670
    },
    {
      "epoch": 0.9354330063686441,
      "grad_norm": 0.3084401488304138,
      "learning_rate": 4e-05,
      "loss": 4.5613,
      "step": 45680
    },
    {
      "epoch": 0.9356377859234534,
      "grad_norm": 0.32649314403533936,
      "learning_rate": 4e-05,
      "loss": 4.5802,
      "step": 45690
    },
    {
      "epoch": 0.9358425654782626,
      "grad_norm": 0.39799973368644714,
      "learning_rate": 4e-05,
      "loss": 4.5817,
      "step": 45700
    },
    {
      "epoch": 0.9360473450330719,
      "grad_norm": 0.34029585123062134,
      "learning_rate": 4e-05,
      "loss": 4.5592,
      "step": 45710
    },
    {
      "epoch": 0.9362521245878811,
      "grad_norm": 0.3444403111934662,
      "learning_rate": 4e-05,
      "loss": 4.5635,
      "step": 45720
    },
    {
      "epoch": 0.9364569041426903,
      "grad_norm": 0.34990790486335754,
      "learning_rate": 4e-05,
      "loss": 4.5199,
      "step": 45730
    },
    {
      "epoch": 0.9366616836974997,
      "grad_norm": 0.32193541526794434,
      "learning_rate": 4e-05,
      "loss": 4.5397,
      "step": 45740
    },
    {
      "epoch": 0.9368664632523089,
      "grad_norm": 0.321878045797348,
      "learning_rate": 4e-05,
      "loss": 4.5521,
      "step": 45750
    },
    {
      "epoch": 0.9370712428071182,
      "grad_norm": 0.32384225726127625,
      "learning_rate": 4e-05,
      "loss": 4.5573,
      "step": 45760
    },
    {
      "epoch": 0.9372760223619274,
      "grad_norm": 0.30843082070350647,
      "learning_rate": 4e-05,
      "loss": 4.5788,
      "step": 45770
    },
    {
      "epoch": 0.9374808019167367,
      "grad_norm": 0.3273121416568756,
      "learning_rate": 4e-05,
      "loss": 4.5212,
      "step": 45780
    },
    {
      "epoch": 0.9376855814715459,
      "grad_norm": 0.3097444772720337,
      "learning_rate": 4e-05,
      "loss": 4.5484,
      "step": 45790
    },
    {
      "epoch": 0.9378903610263551,
      "grad_norm": 0.32343387603759766,
      "learning_rate": 4e-05,
      "loss": 4.5447,
      "step": 45800
    },
    {
      "epoch": 0.9380951405811644,
      "grad_norm": 0.30633851885795593,
      "learning_rate": 4e-05,
      "loss": 4.5429,
      "step": 45810
    },
    {
      "epoch": 0.9382999201359736,
      "grad_norm": 0.32246527075767517,
      "learning_rate": 4e-05,
      "loss": 4.5258,
      "step": 45820
    },
    {
      "epoch": 0.9385046996907829,
      "grad_norm": 0.34284520149230957,
      "learning_rate": 4e-05,
      "loss": 4.583,
      "step": 45830
    },
    {
      "epoch": 0.9387094792455921,
      "grad_norm": 0.32740962505340576,
      "learning_rate": 4e-05,
      "loss": 4.5577,
      "step": 45840
    },
    {
      "epoch": 0.9389142588004014,
      "grad_norm": 0.35072508454322815,
      "learning_rate": 4e-05,
      "loss": 4.5987,
      "step": 45850
    },
    {
      "epoch": 0.9391190383552106,
      "grad_norm": 0.30853354930877686,
      "learning_rate": 4e-05,
      "loss": 4.535,
      "step": 45860
    },
    {
      "epoch": 0.9393238179100198,
      "grad_norm": 0.3089509606361389,
      "learning_rate": 4e-05,
      "loss": 4.5692,
      "step": 45870
    },
    {
      "epoch": 0.9395285974648291,
      "grad_norm": 0.3152104914188385,
      "learning_rate": 4e-05,
      "loss": 4.5415,
      "step": 45880
    },
    {
      "epoch": 0.9397333770196383,
      "grad_norm": 0.33188074827194214,
      "learning_rate": 4e-05,
      "loss": 4.5635,
      "step": 45890
    },
    {
      "epoch": 0.9399381565744476,
      "grad_norm": 0.31641748547554016,
      "learning_rate": 4e-05,
      "loss": 4.5238,
      "step": 45900
    },
    {
      "epoch": 0.9401429361292568,
      "grad_norm": 0.3385186791419983,
      "learning_rate": 4e-05,
      "loss": 4.5406,
      "step": 45910
    },
    {
      "epoch": 0.940347715684066,
      "grad_norm": 0.33622637391090393,
      "learning_rate": 4e-05,
      "loss": 4.5564,
      "step": 45920
    },
    {
      "epoch": 0.9405524952388753,
      "grad_norm": 0.32635486125946045,
      "learning_rate": 4e-05,
      "loss": 4.5768,
      "step": 45930
    },
    {
      "epoch": 0.9407572747936847,
      "grad_norm": 0.318810373544693,
      "learning_rate": 4e-05,
      "loss": 4.5554,
      "step": 45940
    },
    {
      "epoch": 0.9409620543484939,
      "grad_norm": 0.3088144063949585,
      "learning_rate": 4e-05,
      "loss": 4.5607,
      "step": 45950
    },
    {
      "epoch": 0.9411668339033031,
      "grad_norm": 0.36967018246650696,
      "learning_rate": 4e-05,
      "loss": 4.5944,
      "step": 45960
    },
    {
      "epoch": 0.9413716134581124,
      "grad_norm": 0.3272745907306671,
      "learning_rate": 4e-05,
      "loss": 4.5561,
      "step": 45970
    },
    {
      "epoch": 0.9415763930129216,
      "grad_norm": 0.30740636587142944,
      "learning_rate": 4e-05,
      "loss": 4.5619,
      "step": 45980
    },
    {
      "epoch": 0.9417811725677309,
      "grad_norm": 0.3262521028518677,
      "learning_rate": 4e-05,
      "loss": 4.5829,
      "step": 45990
    },
    {
      "epoch": 0.9419859521225401,
      "grad_norm": 0.3408719599246979,
      "learning_rate": 4e-05,
      "loss": 4.5521,
      "step": 46000
    },
    {
      "epoch": 0.9419859521225401,
      "eval_loss": 4.558419704437256,
      "eval_runtime": 4.3817,
      "eval_samples_per_second": 266.108,
      "eval_steps_per_second": 33.321,
      "step": 46000
    },
    {
      "epoch": 0.9421907316773493,
      "grad_norm": 0.3294796049594879,
      "learning_rate": 4e-05,
      "loss": 4.5504,
      "step": 46010
    },
    {
      "epoch": 0.9423955112321586,
      "grad_norm": 0.31579163670539856,
      "learning_rate": 4e-05,
      "loss": 4.5573,
      "step": 46020
    },
    {
      "epoch": 0.9426002907869678,
      "grad_norm": 0.34491610527038574,
      "learning_rate": 4e-05,
      "loss": 4.5627,
      "step": 46030
    },
    {
      "epoch": 0.9428050703417771,
      "grad_norm": 0.31735047698020935,
      "learning_rate": 4e-05,
      "loss": 4.535,
      "step": 46040
    },
    {
      "epoch": 0.9430098498965863,
      "grad_norm": 0.3627663850784302,
      "learning_rate": 4e-05,
      "loss": 4.5832,
      "step": 46050
    },
    {
      "epoch": 0.9432146294513956,
      "grad_norm": 0.3657461106777191,
      "learning_rate": 4e-05,
      "loss": 4.5651,
      "step": 46060
    },
    {
      "epoch": 0.9434194090062048,
      "grad_norm": 0.31911537051200867,
      "learning_rate": 4e-05,
      "loss": 4.5475,
      "step": 46070
    },
    {
      "epoch": 0.943624188561014,
      "grad_norm": 0.38343313336372375,
      "learning_rate": 4e-05,
      "loss": 4.5401,
      "step": 46080
    },
    {
      "epoch": 0.9438289681158233,
      "grad_norm": 0.3523477613925934,
      "learning_rate": 4e-05,
      "loss": 4.5553,
      "step": 46090
    },
    {
      "epoch": 0.9440337476706325,
      "grad_norm": 0.31751102209091187,
      "learning_rate": 4e-05,
      "loss": 4.571,
      "step": 46100
    },
    {
      "epoch": 0.9442385272254418,
      "grad_norm": 0.3188442289829254,
      "learning_rate": 4e-05,
      "loss": 4.5349,
      "step": 46110
    },
    {
      "epoch": 0.944443306780251,
      "grad_norm": 0.36513665318489075,
      "learning_rate": 4e-05,
      "loss": 4.5515,
      "step": 46120
    },
    {
      "epoch": 0.9446480863350603,
      "grad_norm": 0.31229493021965027,
      "learning_rate": 4e-05,
      "loss": 4.5726,
      "step": 46130
    },
    {
      "epoch": 0.9448528658898696,
      "grad_norm": 0.31164535880088806,
      "learning_rate": 4e-05,
      "loss": 4.561,
      "step": 46140
    },
    {
      "epoch": 0.9450576454446789,
      "grad_norm": 0.32203739881515503,
      "learning_rate": 4e-05,
      "loss": 4.545,
      "step": 46150
    },
    {
      "epoch": 0.9452624249994881,
      "grad_norm": 0.3296814262866974,
      "learning_rate": 4e-05,
      "loss": 4.4901,
      "step": 46160
    },
    {
      "epoch": 0.9454672045542973,
      "grad_norm": 0.3368031084537506,
      "learning_rate": 4e-05,
      "loss": 4.5014,
      "step": 46170
    },
    {
      "epoch": 0.9456719841091066,
      "grad_norm": 0.3548653721809387,
      "learning_rate": 4e-05,
      "loss": 4.562,
      "step": 46180
    },
    {
      "epoch": 0.9458767636639158,
      "grad_norm": 0.3123794198036194,
      "learning_rate": 4e-05,
      "loss": 4.5671,
      "step": 46190
    },
    {
      "epoch": 0.9460815432187251,
      "grad_norm": 0.3127918243408203,
      "learning_rate": 4e-05,
      "loss": 4.5618,
      "step": 46200
    },
    {
      "epoch": 0.9462863227735343,
      "grad_norm": 0.3214389979839325,
      "learning_rate": 4e-05,
      "loss": 4.5453,
      "step": 46210
    },
    {
      "epoch": 0.9464911023283435,
      "grad_norm": 0.3183940649032593,
      "learning_rate": 4e-05,
      "loss": 4.5651,
      "step": 46220
    },
    {
      "epoch": 0.9466958818831528,
      "grad_norm": 0.30345025658607483,
      "learning_rate": 4e-05,
      "loss": 4.5724,
      "step": 46230
    },
    {
      "epoch": 0.946900661437962,
      "grad_norm": 0.30393528938293457,
      "learning_rate": 4e-05,
      "loss": 4.589,
      "step": 46240
    },
    {
      "epoch": 0.9471054409927713,
      "grad_norm": 0.35439199209213257,
      "learning_rate": 4e-05,
      "loss": 4.5866,
      "step": 46250
    },
    {
      "epoch": 0.9473102205475805,
      "grad_norm": 0.32344692945480347,
      "learning_rate": 4e-05,
      "loss": 4.5418,
      "step": 46260
    },
    {
      "epoch": 0.9475150001023898,
      "grad_norm": 0.34853124618530273,
      "learning_rate": 4e-05,
      "loss": 4.5437,
      "step": 46270
    },
    {
      "epoch": 0.947719779657199,
      "grad_norm": 0.33046939969062805,
      "learning_rate": 4e-05,
      "loss": 4.5398,
      "step": 46280
    },
    {
      "epoch": 0.9479245592120082,
      "grad_norm": 0.5163120627403259,
      "learning_rate": 4e-05,
      "loss": 4.5591,
      "step": 46290
    },
    {
      "epoch": 0.9481293387668175,
      "grad_norm": 0.3317198157310486,
      "learning_rate": 4e-05,
      "loss": 4.4819,
      "step": 46300
    },
    {
      "epoch": 0.9483341183216267,
      "grad_norm": 0.32500961422920227,
      "learning_rate": 4e-05,
      "loss": 4.5252,
      "step": 46310
    },
    {
      "epoch": 0.948538897876436,
      "grad_norm": 0.34918829798698425,
      "learning_rate": 4e-05,
      "loss": 4.5657,
      "step": 46320
    },
    {
      "epoch": 0.9487436774312452,
      "grad_norm": 0.32156091928482056,
      "learning_rate": 4e-05,
      "loss": 4.5388,
      "step": 46330
    },
    {
      "epoch": 0.9489484569860546,
      "grad_norm": 0.3192023038864136,
      "learning_rate": 4e-05,
      "loss": 4.5612,
      "step": 46340
    },
    {
      "epoch": 0.9491532365408638,
      "grad_norm": 0.32930466532707214,
      "learning_rate": 4e-05,
      "loss": 4.5581,
      "step": 46350
    },
    {
      "epoch": 0.949358016095673,
      "grad_norm": 0.30861541628837585,
      "learning_rate": 4e-05,
      "loss": 4.5325,
      "step": 46360
    },
    {
      "epoch": 0.9495627956504823,
      "grad_norm": 0.30958154797554016,
      "learning_rate": 4e-05,
      "loss": 4.5478,
      "step": 46370
    },
    {
      "epoch": 0.9497675752052915,
      "grad_norm": 0.3103031516075134,
      "learning_rate": 4e-05,
      "loss": 4.5132,
      "step": 46380
    },
    {
      "epoch": 0.9499723547601008,
      "grad_norm": 0.31730809807777405,
      "learning_rate": 4e-05,
      "loss": 4.5647,
      "step": 46390
    },
    {
      "epoch": 0.95017713431491,
      "grad_norm": 0.33080166578292847,
      "learning_rate": 4e-05,
      "loss": 4.5706,
      "step": 46400
    },
    {
      "epoch": 0.9503819138697193,
      "grad_norm": 0.33791664242744446,
      "learning_rate": 4e-05,
      "loss": 4.5262,
      "step": 46410
    },
    {
      "epoch": 0.9505866934245285,
      "grad_norm": 0.3538973033428192,
      "learning_rate": 4e-05,
      "loss": 4.5165,
      "step": 46420
    },
    {
      "epoch": 0.9507914729793377,
      "grad_norm": 0.3252312242984772,
      "learning_rate": 4e-05,
      "loss": 4.5591,
      "step": 46430
    },
    {
      "epoch": 0.950996252534147,
      "grad_norm": 0.3147037923336029,
      "learning_rate": 4e-05,
      "loss": 4.584,
      "step": 46440
    },
    {
      "epoch": 0.9512010320889562,
      "grad_norm": 0.3096960484981537,
      "learning_rate": 4e-05,
      "loss": 4.5081,
      "step": 46450
    },
    {
      "epoch": 0.9514058116437655,
      "grad_norm": 0.32310470938682556,
      "learning_rate": 4e-05,
      "loss": 4.5431,
      "step": 46460
    },
    {
      "epoch": 0.9516105911985747,
      "grad_norm": 0.3410964012145996,
      "learning_rate": 4e-05,
      "loss": 4.5303,
      "step": 46470
    },
    {
      "epoch": 0.951815370753384,
      "grad_norm": 0.32467493414878845,
      "learning_rate": 4e-05,
      "loss": 4.608,
      "step": 46480
    },
    {
      "epoch": 0.9520201503081932,
      "grad_norm": 0.317932665348053,
      "learning_rate": 4e-05,
      "loss": 4.5558,
      "step": 46490
    },
    {
      "epoch": 0.9522249298630024,
      "grad_norm": 0.3132171332836151,
      "learning_rate": 4e-05,
      "loss": 4.5643,
      "step": 46500
    },
    {
      "epoch": 0.9524297094178117,
      "grad_norm": 0.3279651701450348,
      "learning_rate": 4e-05,
      "loss": 4.5692,
      "step": 46510
    },
    {
      "epoch": 0.9526344889726209,
      "grad_norm": 0.3203917443752289,
      "learning_rate": 4e-05,
      "loss": 4.557,
      "step": 46520
    },
    {
      "epoch": 0.9528392685274302,
      "grad_norm": 0.32041600346565247,
      "learning_rate": 4e-05,
      "loss": 4.5214,
      "step": 46530
    },
    {
      "epoch": 0.9530440480822394,
      "grad_norm": 0.326524943113327,
      "learning_rate": 4e-05,
      "loss": 4.5483,
      "step": 46540
    },
    {
      "epoch": 0.9532488276370488,
      "grad_norm": 0.337062269449234,
      "learning_rate": 4e-05,
      "loss": 4.5515,
      "step": 46550
    },
    {
      "epoch": 0.953453607191858,
      "grad_norm": 0.3208051919937134,
      "learning_rate": 4e-05,
      "loss": 4.5726,
      "step": 46560
    },
    {
      "epoch": 0.9536583867466673,
      "grad_norm": 0.3239714801311493,
      "learning_rate": 4e-05,
      "loss": 4.5697,
      "step": 46570
    },
    {
      "epoch": 0.9538631663014765,
      "grad_norm": 0.36318278312683105,
      "learning_rate": 4e-05,
      "loss": 4.5534,
      "step": 46580
    },
    {
      "epoch": 0.9540679458562857,
      "grad_norm": 0.31034743785858154,
      "learning_rate": 4e-05,
      "loss": 4.5486,
      "step": 46590
    },
    {
      "epoch": 0.954272725411095,
      "grad_norm": 0.31020817160606384,
      "learning_rate": 4e-05,
      "loss": 4.5603,
      "step": 46600
    },
    {
      "epoch": 0.9544775049659042,
      "grad_norm": 0.38229674100875854,
      "learning_rate": 4e-05,
      "loss": 4.5629,
      "step": 46610
    },
    {
      "epoch": 0.9546822845207135,
      "grad_norm": 0.32211023569107056,
      "learning_rate": 4e-05,
      "loss": 4.5421,
      "step": 46620
    },
    {
      "epoch": 0.9548870640755227,
      "grad_norm": 0.3195001780986786,
      "learning_rate": 4e-05,
      "loss": 4.5701,
      "step": 46630
    },
    {
      "epoch": 0.955091843630332,
      "grad_norm": 0.3793138861656189,
      "learning_rate": 4e-05,
      "loss": 4.5458,
      "step": 46640
    },
    {
      "epoch": 0.9552966231851412,
      "grad_norm": 0.3213236927986145,
      "learning_rate": 4e-05,
      "loss": 4.565,
      "step": 46650
    },
    {
      "epoch": 0.9555014027399504,
      "grad_norm": 0.304331511259079,
      "learning_rate": 4e-05,
      "loss": 4.5342,
      "step": 46660
    },
    {
      "epoch": 0.9557061822947597,
      "grad_norm": 0.3650117516517639,
      "learning_rate": 4e-05,
      "loss": 4.518,
      "step": 46670
    },
    {
      "epoch": 0.9559109618495689,
      "grad_norm": 0.3532241880893707,
      "learning_rate": 4e-05,
      "loss": 4.5568,
      "step": 46680
    },
    {
      "epoch": 0.9561157414043782,
      "grad_norm": 0.3322131335735321,
      "learning_rate": 4e-05,
      "loss": 4.584,
      "step": 46690
    },
    {
      "epoch": 0.9563205209591874,
      "grad_norm": 0.3379342257976532,
      "learning_rate": 4e-05,
      "loss": 4.5572,
      "step": 46700
    },
    {
      "epoch": 0.9565253005139966,
      "grad_norm": 0.31786176562309265,
      "learning_rate": 4e-05,
      "loss": 4.5634,
      "step": 46710
    },
    {
      "epoch": 0.9567300800688059,
      "grad_norm": 0.31300413608551025,
      "learning_rate": 4e-05,
      "loss": 4.5617,
      "step": 46720
    },
    {
      "epoch": 0.9569348596236151,
      "grad_norm": 0.3562588691711426,
      "learning_rate": 4e-05,
      "loss": 4.5802,
      "step": 46730
    },
    {
      "epoch": 0.9571396391784244,
      "grad_norm": 0.3157178461551666,
      "learning_rate": 4e-05,
      "loss": 4.5796,
      "step": 46740
    },
    {
      "epoch": 0.9573444187332337,
      "grad_norm": 0.34579920768737793,
      "learning_rate": 4e-05,
      "loss": 4.5561,
      "step": 46750
    },
    {
      "epoch": 0.957549198288043,
      "grad_norm": 0.3953477442264557,
      "learning_rate": 4e-05,
      "loss": 4.5164,
      "step": 46760
    },
    {
      "epoch": 0.9577539778428522,
      "grad_norm": 0.3326374590396881,
      "learning_rate": 4e-05,
      "loss": 4.556,
      "step": 46770
    },
    {
      "epoch": 0.9579587573976615,
      "grad_norm": 0.31724971532821655,
      "learning_rate": 4e-05,
      "loss": 4.606,
      "step": 46780
    },
    {
      "epoch": 0.9581635369524707,
      "grad_norm": 0.32476869225502014,
      "learning_rate": 4e-05,
      "loss": 4.5487,
      "step": 46790
    },
    {
      "epoch": 0.9583683165072799,
      "grad_norm": 0.33330246806144714,
      "learning_rate": 4e-05,
      "loss": 4.5204,
      "step": 46800
    },
    {
      "epoch": 0.9585730960620892,
      "grad_norm": 0.3505427837371826,
      "learning_rate": 4e-05,
      "loss": 4.5862,
      "step": 46810
    },
    {
      "epoch": 0.9587778756168984,
      "grad_norm": 0.3296929597854614,
      "learning_rate": 4e-05,
      "loss": 4.5904,
      "step": 46820
    },
    {
      "epoch": 0.9589826551717077,
      "grad_norm": 0.31201407313346863,
      "learning_rate": 4e-05,
      "loss": 4.5867,
      "step": 46830
    },
    {
      "epoch": 0.9591874347265169,
      "grad_norm": 0.31656521558761597,
      "learning_rate": 4e-05,
      "loss": 4.5303,
      "step": 46840
    },
    {
      "epoch": 0.9593922142813261,
      "grad_norm": 0.3408479392528534,
      "learning_rate": 4e-05,
      "loss": 4.5728,
      "step": 46850
    },
    {
      "epoch": 0.9595969938361354,
      "grad_norm": 0.3489954471588135,
      "learning_rate": 4e-05,
      "loss": 4.508,
      "step": 46860
    },
    {
      "epoch": 0.9598017733909446,
      "grad_norm": 0.32541826367378235,
      "learning_rate": 4e-05,
      "loss": 4.5661,
      "step": 46870
    },
    {
      "epoch": 0.9600065529457539,
      "grad_norm": 0.7616615295410156,
      "learning_rate": 4e-05,
      "loss": 4.5414,
      "step": 46880
    },
    {
      "epoch": 0.9602113325005631,
      "grad_norm": 0.3301410377025604,
      "learning_rate": 4e-05,
      "loss": 4.5427,
      "step": 46890
    },
    {
      "epoch": 0.9604161120553724,
      "grad_norm": 0.31184107065200806,
      "learning_rate": 4e-05,
      "loss": 4.5606,
      "step": 46900
    },
    {
      "epoch": 0.9606208916101816,
      "grad_norm": 0.3217566907405853,
      "learning_rate": 4e-05,
      "loss": 4.5715,
      "step": 46910
    },
    {
      "epoch": 0.9608256711649908,
      "grad_norm": 0.3093664348125458,
      "learning_rate": 4e-05,
      "loss": 4.519,
      "step": 46920
    },
    {
      "epoch": 0.9610304507198001,
      "grad_norm": 0.31126877665519714,
      "learning_rate": 4e-05,
      "loss": 4.5201,
      "step": 46930
    },
    {
      "epoch": 0.9612352302746093,
      "grad_norm": 0.3557184636592865,
      "learning_rate": 4e-05,
      "loss": 4.5645,
      "step": 46940
    },
    {
      "epoch": 0.9614400098294187,
      "grad_norm": 0.3358568847179413,
      "learning_rate": 4e-05,
      "loss": 4.5646,
      "step": 46950
    },
    {
      "epoch": 0.9616447893842279,
      "grad_norm": 0.34344351291656494,
      "learning_rate": 4e-05,
      "loss": 4.5857,
      "step": 46960
    },
    {
      "epoch": 0.9618495689390372,
      "grad_norm": 0.3141220510005951,
      "learning_rate": 4e-05,
      "loss": 4.5409,
      "step": 46970
    },
    {
      "epoch": 0.9620543484938464,
      "grad_norm": 0.32317131757736206,
      "learning_rate": 4e-05,
      "loss": 4.5726,
      "step": 46980
    },
    {
      "epoch": 0.9622591280486557,
      "grad_norm": 0.3247339427471161,
      "learning_rate": 4e-05,
      "loss": 4.5388,
      "step": 46990
    },
    {
      "epoch": 0.9624639076034649,
      "grad_norm": 0.3239203989505768,
      "learning_rate": 4e-05,
      "loss": 4.5892,
      "step": 47000
    },
    {
      "epoch": 0.9626686871582741,
      "grad_norm": 0.35522451996803284,
      "learning_rate": 4e-05,
      "loss": 4.5013,
      "step": 47010
    },
    {
      "epoch": 0.9628734667130834,
      "grad_norm": 0.32539790868759155,
      "learning_rate": 4e-05,
      "loss": 4.569,
      "step": 47020
    },
    {
      "epoch": 0.9630782462678926,
      "grad_norm": 0.41466546058654785,
      "learning_rate": 4e-05,
      "loss": 4.5335,
      "step": 47030
    },
    {
      "epoch": 0.9632830258227019,
      "grad_norm": 0.33411848545074463,
      "learning_rate": 4e-05,
      "loss": 4.5566,
      "step": 47040
    },
    {
      "epoch": 0.9634878053775111,
      "grad_norm": 0.3331572711467743,
      "learning_rate": 4e-05,
      "loss": 4.5539,
      "step": 47050
    },
    {
      "epoch": 0.9636925849323204,
      "grad_norm": 0.38816383481025696,
      "learning_rate": 4e-05,
      "loss": 4.5683,
      "step": 47060
    },
    {
      "epoch": 0.9638973644871296,
      "grad_norm": 0.3279906213283539,
      "learning_rate": 4e-05,
      "loss": 4.5379,
      "step": 47070
    },
    {
      "epoch": 0.9641021440419388,
      "grad_norm": 0.3425002992153168,
      "learning_rate": 4e-05,
      "loss": 4.5698,
      "step": 47080
    },
    {
      "epoch": 0.9643069235967481,
      "grad_norm": 0.320560097694397,
      "learning_rate": 4e-05,
      "loss": 4.5611,
      "step": 47090
    },
    {
      "epoch": 0.9645117031515573,
      "grad_norm": 0.3480735719203949,
      "learning_rate": 4e-05,
      "loss": 4.5903,
      "step": 47100
    },
    {
      "epoch": 0.9647164827063666,
      "grad_norm": 0.6516593098640442,
      "learning_rate": 4e-05,
      "loss": 4.5654,
      "step": 47110
    },
    {
      "epoch": 0.9649212622611758,
      "grad_norm": 0.32140108942985535,
      "learning_rate": 4e-05,
      "loss": 4.5859,
      "step": 47120
    },
    {
      "epoch": 0.965126041815985,
      "grad_norm": 0.34245118498802185,
      "learning_rate": 4e-05,
      "loss": 4.5617,
      "step": 47130
    },
    {
      "epoch": 0.9653308213707943,
      "grad_norm": 0.32911524176597595,
      "learning_rate": 4e-05,
      "loss": 4.5324,
      "step": 47140
    },
    {
      "epoch": 0.9655356009256036,
      "grad_norm": 0.3488873839378357,
      "learning_rate": 4e-05,
      "loss": 4.5632,
      "step": 47150
    },
    {
      "epoch": 0.9657403804804129,
      "grad_norm": 0.3380149304866791,
      "learning_rate": 4e-05,
      "loss": 4.5397,
      "step": 47160
    },
    {
      "epoch": 0.9659451600352221,
      "grad_norm": 0.3188472092151642,
      "learning_rate": 4e-05,
      "loss": 4.573,
      "step": 47170
    },
    {
      "epoch": 0.9661499395900314,
      "grad_norm": 0.3182220458984375,
      "learning_rate": 4e-05,
      "loss": 4.5646,
      "step": 47180
    },
    {
      "epoch": 0.9663547191448406,
      "grad_norm": 0.3335462808609009,
      "learning_rate": 4e-05,
      "loss": 4.5637,
      "step": 47190
    },
    {
      "epoch": 0.9665594986996499,
      "grad_norm": 0.32872721552848816,
      "learning_rate": 4e-05,
      "loss": 4.545,
      "step": 47200
    },
    {
      "epoch": 0.9667642782544591,
      "grad_norm": 0.31972619891166687,
      "learning_rate": 4e-05,
      "loss": 4.5491,
      "step": 47210
    },
    {
      "epoch": 0.9669690578092683,
      "grad_norm": 0.3711688816547394,
      "learning_rate": 4e-05,
      "loss": 4.5732,
      "step": 47220
    },
    {
      "epoch": 0.9671738373640776,
      "grad_norm": 0.3483497202396393,
      "learning_rate": 4e-05,
      "loss": 4.5873,
      "step": 47230
    },
    {
      "epoch": 0.9673786169188868,
      "grad_norm": 0.3899175822734833,
      "learning_rate": 4e-05,
      "loss": 4.5106,
      "step": 47240
    },
    {
      "epoch": 0.9675833964736961,
      "grad_norm": 0.3247251808643341,
      "learning_rate": 4e-05,
      "loss": 4.5453,
      "step": 47250
    },
    {
      "epoch": 0.9677881760285053,
      "grad_norm": 0.32415032386779785,
      "learning_rate": 4e-05,
      "loss": 4.4923,
      "step": 47260
    },
    {
      "epoch": 0.9679929555833146,
      "grad_norm": 0.32311588525772095,
      "learning_rate": 4e-05,
      "loss": 4.583,
      "step": 47270
    },
    {
      "epoch": 0.9681977351381238,
      "grad_norm": 0.3264346122741699,
      "learning_rate": 4e-05,
      "loss": 4.5331,
      "step": 47280
    },
    {
      "epoch": 0.968402514692933,
      "grad_norm": 0.34599217772483826,
      "learning_rate": 4e-05,
      "loss": 4.5525,
      "step": 47290
    },
    {
      "epoch": 0.9686072942477423,
      "grad_norm": 0.31252530217170715,
      "learning_rate": 4e-05,
      "loss": 4.5669,
      "step": 47300
    },
    {
      "epoch": 0.9688120738025515,
      "grad_norm": 0.32626235485076904,
      "learning_rate": 4e-05,
      "loss": 4.6154,
      "step": 47310
    },
    {
      "epoch": 0.9690168533573608,
      "grad_norm": 0.3365563750267029,
      "learning_rate": 4e-05,
      "loss": 4.5359,
      "step": 47320
    },
    {
      "epoch": 0.96922163291217,
      "grad_norm": 0.3094619810581207,
      "learning_rate": 4e-05,
      "loss": 4.571,
      "step": 47330
    },
    {
      "epoch": 0.9694264124669792,
      "grad_norm": 0.3282161355018616,
      "learning_rate": 4e-05,
      "loss": 4.5689,
      "step": 47340
    },
    {
      "epoch": 0.9696311920217885,
      "grad_norm": 0.33691951632499695,
      "learning_rate": 4e-05,
      "loss": 4.575,
      "step": 47350
    },
    {
      "epoch": 0.9698359715765978,
      "grad_norm": 0.3157297968864441,
      "learning_rate": 4e-05,
      "loss": 4.5732,
      "step": 47360
    },
    {
      "epoch": 0.9700407511314071,
      "grad_norm": 0.3412298858165741,
      "learning_rate": 4e-05,
      "loss": 4.558,
      "step": 47370
    },
    {
      "epoch": 0.9702455306862163,
      "grad_norm": 0.41333726048469543,
      "learning_rate": 4e-05,
      "loss": 4.5732,
      "step": 47380
    },
    {
      "epoch": 0.9704503102410256,
      "grad_norm": 0.3142140209674835,
      "learning_rate": 4e-05,
      "loss": 4.5775,
      "step": 47390
    },
    {
      "epoch": 0.9706550897958348,
      "grad_norm": 0.36438649892807007,
      "learning_rate": 4e-05,
      "loss": 4.5636,
      "step": 47400
    },
    {
      "epoch": 0.9708598693506441,
      "grad_norm": 0.4539795219898224,
      "learning_rate": 4e-05,
      "loss": 4.5388,
      "step": 47410
    },
    {
      "epoch": 0.9710646489054533,
      "grad_norm": 0.32517075538635254,
      "learning_rate": 4e-05,
      "loss": 4.5736,
      "step": 47420
    },
    {
      "epoch": 0.9712694284602625,
      "grad_norm": 0.34917378425598145,
      "learning_rate": 4e-05,
      "loss": 4.5442,
      "step": 47430
    },
    {
      "epoch": 0.9714742080150718,
      "grad_norm": 0.320481538772583,
      "learning_rate": 4e-05,
      "loss": 4.5308,
      "step": 47440
    },
    {
      "epoch": 0.971678987569881,
      "grad_norm": 0.32384875416755676,
      "learning_rate": 4e-05,
      "loss": 4.5529,
      "step": 47450
    },
    {
      "epoch": 0.9718837671246903,
      "grad_norm": 0.32067564129829407,
      "learning_rate": 4e-05,
      "loss": 4.5583,
      "step": 47460
    },
    {
      "epoch": 0.9720885466794995,
      "grad_norm": 0.33786696195602417,
      "learning_rate": 4e-05,
      "loss": 4.565,
      "step": 47470
    },
    {
      "epoch": 0.9722933262343088,
      "grad_norm": 0.303669273853302,
      "learning_rate": 4e-05,
      "loss": 4.5522,
      "step": 47480
    },
    {
      "epoch": 0.972498105789118,
      "grad_norm": 0.3249942660331726,
      "learning_rate": 4e-05,
      "loss": 4.5413,
      "step": 47490
    },
    {
      "epoch": 0.9727028853439272,
      "grad_norm": 0.3326072096824646,
      "learning_rate": 4e-05,
      "loss": 4.5572,
      "step": 47500
    },
    {
      "epoch": 0.9729076648987365,
      "grad_norm": 0.3195289969444275,
      "learning_rate": 4e-05,
      "loss": 4.5653,
      "step": 47510
    },
    {
      "epoch": 0.9731124444535457,
      "grad_norm": 0.3481857478618622,
      "learning_rate": 4e-05,
      "loss": 4.5637,
      "step": 47520
    },
    {
      "epoch": 0.973317224008355,
      "grad_norm": 0.4060842990875244,
      "learning_rate": 4e-05,
      "loss": 4.5625,
      "step": 47530
    },
    {
      "epoch": 0.9735220035631642,
      "grad_norm": 0.3097035586833954,
      "learning_rate": 4e-05,
      "loss": 4.5666,
      "step": 47540
    },
    {
      "epoch": 0.9737267831179734,
      "grad_norm": 0.3601599633693695,
      "learning_rate": 4e-05,
      "loss": 4.6152,
      "step": 47550
    },
    {
      "epoch": 0.9739315626727828,
      "grad_norm": 0.3242459297180176,
      "learning_rate": 4e-05,
      "loss": 4.5512,
      "step": 47560
    },
    {
      "epoch": 0.974136342227592,
      "grad_norm": 0.3202042877674103,
      "learning_rate": 4e-05,
      "loss": 4.5766,
      "step": 47570
    },
    {
      "epoch": 0.9743411217824013,
      "grad_norm": 0.33238908648490906,
      "learning_rate": 4e-05,
      "loss": 4.5425,
      "step": 47580
    },
    {
      "epoch": 0.9745459013372105,
      "grad_norm": 0.3203936815261841,
      "learning_rate": 4e-05,
      "loss": 4.5736,
      "step": 47590
    },
    {
      "epoch": 0.9747506808920198,
      "grad_norm": 0.32392263412475586,
      "learning_rate": 4e-05,
      "loss": 4.5763,
      "step": 47600
    },
    {
      "epoch": 0.974955460446829,
      "grad_norm": 0.3200019598007202,
      "learning_rate": 4e-05,
      "loss": 4.5398,
      "step": 47610
    },
    {
      "epoch": 0.9751602400016383,
      "grad_norm": 0.3280363976955414,
      "learning_rate": 4e-05,
      "loss": 4.568,
      "step": 47620
    },
    {
      "epoch": 0.9753650195564475,
      "grad_norm": 0.3076491057872772,
      "learning_rate": 4e-05,
      "loss": 4.5231,
      "step": 47630
    },
    {
      "epoch": 0.9755697991112567,
      "grad_norm": 0.32782208919525146,
      "learning_rate": 4e-05,
      "loss": 4.5419,
      "step": 47640
    },
    {
      "epoch": 0.975774578666066,
      "grad_norm": 0.33048853278160095,
      "learning_rate": 4e-05,
      "loss": 4.516,
      "step": 47650
    },
    {
      "epoch": 0.9759793582208752,
      "grad_norm": 0.3323243260383606,
      "learning_rate": 4e-05,
      "loss": 4.5358,
      "step": 47660
    },
    {
      "epoch": 0.9761841377756845,
      "grad_norm": 0.34078171849250793,
      "learning_rate": 4e-05,
      "loss": 4.5525,
      "step": 47670
    },
    {
      "epoch": 0.9763889173304937,
      "grad_norm": 0.3658129870891571,
      "learning_rate": 4e-05,
      "loss": 4.5469,
      "step": 47680
    },
    {
      "epoch": 0.976593696885303,
      "grad_norm": 0.31622937321662903,
      "learning_rate": 4e-05,
      "loss": 4.5849,
      "step": 47690
    },
    {
      "epoch": 0.9767984764401122,
      "grad_norm": 0.3148673474788666,
      "learning_rate": 4e-05,
      "loss": 4.5298,
      "step": 47700
    },
    {
      "epoch": 0.9770032559949214,
      "grad_norm": 0.3100682199001312,
      "learning_rate": 4e-05,
      "loss": 4.5349,
      "step": 47710
    },
    {
      "epoch": 0.9772080355497307,
      "grad_norm": 0.38230493664741516,
      "learning_rate": 4e-05,
      "loss": 4.5278,
      "step": 47720
    },
    {
      "epoch": 0.9774128151045399,
      "grad_norm": 0.339199960231781,
      "learning_rate": 4e-05,
      "loss": 4.5906,
      "step": 47730
    },
    {
      "epoch": 0.9776175946593492,
      "grad_norm": 0.3349709212779999,
      "learning_rate": 4e-05,
      "loss": 4.53,
      "step": 47740
    },
    {
      "epoch": 0.9778223742141584,
      "grad_norm": 0.3122599422931671,
      "learning_rate": 4e-05,
      "loss": 4.5659,
      "step": 47750
    },
    {
      "epoch": 0.9780271537689678,
      "grad_norm": 0.3261248469352722,
      "learning_rate": 4e-05,
      "loss": 4.5501,
      "step": 47760
    },
    {
      "epoch": 0.978231933323777,
      "grad_norm": 0.31458988785743713,
      "learning_rate": 4e-05,
      "loss": 4.531,
      "step": 47770
    },
    {
      "epoch": 0.9784367128785862,
      "grad_norm": 0.313297837972641,
      "learning_rate": 4e-05,
      "loss": 4.5518,
      "step": 47780
    },
    {
      "epoch": 0.9786414924333955,
      "grad_norm": 0.33505669236183167,
      "learning_rate": 4e-05,
      "loss": 4.5619,
      "step": 47790
    },
    {
      "epoch": 0.9788462719882047,
      "grad_norm": 0.32072460651397705,
      "learning_rate": 4e-05,
      "loss": 4.5512,
      "step": 47800
    },
    {
      "epoch": 0.979051051543014,
      "grad_norm": 0.3592844605445862,
      "learning_rate": 4e-05,
      "loss": 4.5553,
      "step": 47810
    },
    {
      "epoch": 0.9792558310978232,
      "grad_norm": 0.35318103432655334,
      "learning_rate": 4e-05,
      "loss": 4.516,
      "step": 47820
    },
    {
      "epoch": 0.9794606106526325,
      "grad_norm": 0.3392002284526825,
      "learning_rate": 4e-05,
      "loss": 4.5302,
      "step": 47830
    },
    {
      "epoch": 0.9796653902074417,
      "grad_norm": 0.3325815200805664,
      "learning_rate": 4e-05,
      "loss": 4.5727,
      "step": 47840
    },
    {
      "epoch": 0.9798701697622509,
      "grad_norm": 0.3270987272262573,
      "learning_rate": 4e-05,
      "loss": 4.4671,
      "step": 47850
    },
    {
      "epoch": 0.9800749493170602,
      "grad_norm": 0.32717788219451904,
      "learning_rate": 4e-05,
      "loss": 4.5396,
      "step": 47860
    },
    {
      "epoch": 0.9802797288718694,
      "grad_norm": 0.32797834277153015,
      "learning_rate": 4e-05,
      "loss": 4.57,
      "step": 47870
    },
    {
      "epoch": 0.9804845084266787,
      "grad_norm": 0.35029080510139465,
      "learning_rate": 4e-05,
      "loss": 4.5932,
      "step": 47880
    },
    {
      "epoch": 0.9806892879814879,
      "grad_norm": 0.3288595676422119,
      "learning_rate": 4e-05,
      "loss": 4.5366,
      "step": 47890
    },
    {
      "epoch": 0.9808940675362972,
      "grad_norm": 0.33771997690200806,
      "learning_rate": 4e-05,
      "loss": 4.5549,
      "step": 47900
    },
    {
      "epoch": 0.9810988470911064,
      "grad_norm": 0.32012638449668884,
      "learning_rate": 4e-05,
      "loss": 4.5551,
      "step": 47910
    },
    {
      "epoch": 0.9813036266459156,
      "grad_norm": 0.30867964029312134,
      "learning_rate": 4e-05,
      "loss": 4.5521,
      "step": 47920
    },
    {
      "epoch": 0.9815084062007249,
      "grad_norm": 0.3192601799964905,
      "learning_rate": 4e-05,
      "loss": 4.5501,
      "step": 47930
    },
    {
      "epoch": 0.9817131857555341,
      "grad_norm": 0.3232973515987396,
      "learning_rate": 4e-05,
      "loss": 4.5697,
      "step": 47940
    },
    {
      "epoch": 0.9819179653103434,
      "grad_norm": 0.3176027238368988,
      "learning_rate": 4e-05,
      "loss": 4.5854,
      "step": 47950
    },
    {
      "epoch": 0.9821227448651526,
      "grad_norm": 0.359607070684433,
      "learning_rate": 4e-05,
      "loss": 4.5621,
      "step": 47960
    },
    {
      "epoch": 0.982327524419962,
      "grad_norm": 0.41966933012008667,
      "learning_rate": 4e-05,
      "loss": 4.5726,
      "step": 47970
    },
    {
      "epoch": 0.9825323039747712,
      "grad_norm": 0.32640331983566284,
      "learning_rate": 4e-05,
      "loss": 4.5528,
      "step": 47980
    },
    {
      "epoch": 0.9827370835295804,
      "grad_norm": 0.3352060317993164,
      "learning_rate": 4e-05,
      "loss": 4.5369,
      "step": 47990
    },
    {
      "epoch": 0.9829418630843897,
      "grad_norm": 0.3422849774360657,
      "learning_rate": 4e-05,
      "loss": 4.5835,
      "step": 48000
    },
    {
      "epoch": 0.9829418630843897,
      "eval_loss": 4.555622100830078,
      "eval_runtime": 4.3696,
      "eval_samples_per_second": 266.845,
      "eval_steps_per_second": 33.413,
      "step": 48000
    }
  ],
  "logging_steps": 10,
  "max_steps": 48833,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}