{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.45263025128968226,
  "eval_steps": 500,
  "global_step": 229376,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.314596139644004e-05,
      "grad_norm": 10.52633285522461,
      "learning_rate": 1.5625e-06,
      "loss": 11.3101,
      "step": 32
    },
    {
      "epoch": 0.0001262919227928801,
      "grad_norm": 10.210962295532227,
      "learning_rate": 3.125e-06,
      "loss": 11.2799,
      "step": 64
    },
    {
      "epoch": 0.00018943788418932013,
      "grad_norm": 10.53814697265625,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 11.2291,
      "step": 96
    },
    {
      "epoch": 0.0002525838455857602,
      "grad_norm": 10.390519142150879,
      "learning_rate": 6.25e-06,
      "loss": 11.1376,
      "step": 128
    },
    {
      "epoch": 0.00031572980698220025,
      "grad_norm": 10.355998992919922,
      "learning_rate": 7.8125e-06,
      "loss": 10.9842,
      "step": 160
    },
    {
      "epoch": 0.00037887576837864026,
      "grad_norm": 10.46985912322998,
      "learning_rate": 9.375000000000001e-06,
      "loss": 10.5984,
      "step": 192
    },
    {
      "epoch": 0.00044202172977508033,
      "grad_norm": 10.829180717468262,
      "learning_rate": 1.09375e-05,
      "loss": 9.3536,
      "step": 224
    },
    {
      "epoch": 0.0005051676911715204,
      "grad_norm": 6.283628463745117,
      "learning_rate": 1.25e-05,
      "loss": 6.5904,
      "step": 256
    },
    {
      "epoch": 0.0005683136525679605,
      "grad_norm": 0.3195990324020386,
      "learning_rate": 1.4062500000000001e-05,
      "loss": 4.9049,
      "step": 288
    },
    {
      "epoch": 0.0006314596139644005,
      "grad_norm": 0.2334459125995636,
      "learning_rate": 1.5625e-05,
      "loss": 4.7834,
      "step": 320
    },
    {
      "epoch": 0.0006946055753608405,
      "grad_norm": 0.8244943022727966,
      "learning_rate": 1.71875e-05,
      "loss": 4.6017,
      "step": 352
    },
    {
      "epoch": 0.0007577515367572805,
      "grad_norm": 1.0876139402389526,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 4.1688,
      "step": 384
    },
    {
      "epoch": 0.0008208974981537207,
      "grad_norm": 0.6743049025535583,
      "learning_rate": 2.0312500000000002e-05,
      "loss": 3.7257,
      "step": 416
    },
    {
      "epoch": 0.0008840434595501607,
      "grad_norm": 0.3528505265712738,
      "learning_rate": 2.1875e-05,
      "loss": 3.4752,
      "step": 448
    },
    {
      "epoch": 0.0009471894209466007,
      "grad_norm": 0.21036797761917114,
      "learning_rate": 2.34375e-05,
      "loss": 3.4045,
      "step": 480
    },
    {
      "epoch": 0.0010103353823430407,
      "grad_norm": 0.16363564133644104,
      "learning_rate": 2.5e-05,
      "loss": 3.3964,
      "step": 512
    },
    {
      "epoch": 0.0010734813437394807,
      "grad_norm": 0.12469228357076645,
      "learning_rate": 2.6562500000000002e-05,
      "loss": 3.3912,
      "step": 544
    },
    {
      "epoch": 0.001136627305135921,
      "grad_norm": 0.11611668020486832,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 3.3176,
      "step": 576
    },
    {
      "epoch": 0.001199773266532361,
      "grad_norm": 0.1084858626127243,
      "learning_rate": 2.96875e-05,
      "loss": 3.296,
      "step": 608
    },
    {
      "epoch": 0.001262919227928801,
      "grad_norm": 0.11642524600028992,
      "learning_rate": 3.125e-05,
      "loss": 3.3348,
      "step": 640
    },
    {
      "epoch": 0.001326065189325241,
      "grad_norm": 0.09771055728197098,
      "learning_rate": 3.2812500000000005e-05,
      "loss": 3.2892,
      "step": 672
    },
    {
      "epoch": 0.001389211150721681,
      "grad_norm": 0.08836262673139572,
      "learning_rate": 3.4375e-05,
      "loss": 3.2728,
      "step": 704
    },
    {
      "epoch": 0.001452357112118121,
      "grad_norm": 0.07660011202096939,
      "learning_rate": 3.59375e-05,
      "loss": 3.2759,
      "step": 736
    },
    {
      "epoch": 0.001515503073514561,
      "grad_norm": 0.0760306715965271,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 3.2631,
      "step": 768
    },
    {
      "epoch": 0.0015786490349110013,
      "grad_norm": 0.07291432470083237,
      "learning_rate": 3.90625e-05,
      "loss": 3.2215,
      "step": 800
    },
    {
      "epoch": 0.0016417949963074413,
      "grad_norm": 0.07679640501737595,
      "learning_rate": 4.0625000000000005e-05,
      "loss": 3.2085,
      "step": 832
    },
    {
      "epoch": 0.0017049409577038813,
      "grad_norm": 0.07241988927125931,
      "learning_rate": 4.21875e-05,
      "loss": 3.1709,
      "step": 864
    },
    {
      "epoch": 0.0017680869191003213,
      "grad_norm": 0.06889630109071732,
      "learning_rate": 4.375e-05,
      "loss": 3.1494,
      "step": 896
    },
    {
      "epoch": 0.0018312328804967614,
      "grad_norm": 0.06774609535932541,
      "learning_rate": 4.5312500000000004e-05,
      "loss": 3.1271,
      "step": 928
    },
    {
      "epoch": 0.0018943788418932014,
      "grad_norm": 0.05798167362809181,
      "learning_rate": 4.6875e-05,
      "loss": 3.1373,
      "step": 960
    },
    {
      "epoch": 0.0019575248032896414,
      "grad_norm": 0.062139470130205154,
      "learning_rate": 4.8437500000000005e-05,
      "loss": 3.1317,
      "step": 992
    },
    {
      "epoch": 0.0020206707646860814,
      "grad_norm": 0.058672286570072174,
      "learning_rate": 5e-05,
      "loss": 3.096,
      "step": 1024
    },
    {
      "epoch": 0.0020838167260825214,
      "grad_norm": 0.06083087623119354,
      "learning_rate": 5.15625e-05,
      "loss": 3.0899,
      "step": 1056
    },
    {
      "epoch": 0.0021469626874789614,
      "grad_norm": 0.06108880415558815,
      "learning_rate": 5.3125000000000004e-05,
      "loss": 3.0704,
      "step": 1088
    },
    {
      "epoch": 0.002210108648875402,
      "grad_norm": 0.05641736090183258,
      "learning_rate": 5.46875e-05,
      "loss": 3.0791,
      "step": 1120
    },
    {
      "epoch": 0.002273254610271842,
      "grad_norm": 0.05347690358757973,
      "learning_rate": 5.6250000000000005e-05,
      "loss": 3.0401,
      "step": 1152
    },
    {
      "epoch": 0.002336400571668282,
      "grad_norm": 0.07868298143148422,
      "learning_rate": 5.78125e-05,
      "loss": 3.0288,
      "step": 1184
    },
    {
      "epoch": 0.002399546533064722,
      "grad_norm": 0.11979499459266663,
      "learning_rate": 5.9375e-05,
      "loss": 3.0046,
      "step": 1216
    },
    {
      "epoch": 0.002462692494461162,
      "grad_norm": 0.06157733127474785,
      "learning_rate": 6.0937500000000004e-05,
      "loss": 3.0156,
      "step": 1248
    },
    {
      "epoch": 0.002525838455857602,
      "grad_norm": 0.07994844764471054,
      "learning_rate": 6.25e-05,
      "loss": 3.0079,
      "step": 1280
    },
    {
      "epoch": 0.002588984417254042,
      "grad_norm": 0.05458033084869385,
      "learning_rate": 6.40625e-05,
      "loss": 3.0012,
      "step": 1312
    },
    {
      "epoch": 0.002652130378650482,
      "grad_norm": 0.05429359897971153,
      "learning_rate": 6.562500000000001e-05,
      "loss": 2.9843,
      "step": 1344
    },
    {
      "epoch": 0.002715276340046922,
      "grad_norm": 0.05572785809636116,
      "learning_rate": 6.71875e-05,
      "loss": 2.967,
      "step": 1376
    },
    {
      "epoch": 0.002778422301443362,
      "grad_norm": 0.05930449441075325,
      "learning_rate": 6.875e-05,
      "loss": 2.9612,
      "step": 1408
    },
    {
      "epoch": 0.002841568262839802,
      "grad_norm": 0.05627207085490227,
      "learning_rate": 7.031250000000001e-05,
      "loss": 2.9641,
      "step": 1440
    },
    {
      "epoch": 0.002904714224236242,
      "grad_norm": 0.05794452130794525,
      "learning_rate": 7.1875e-05,
      "loss": 2.9627,
      "step": 1472
    },
    {
      "epoch": 0.002967860185632682,
      "grad_norm": 0.07208114117383957,
      "learning_rate": 7.34375e-05,
      "loss": 2.9441,
      "step": 1504
    },
    {
      "epoch": 0.003031006147029122,
      "grad_norm": 0.07248008996248245,
      "learning_rate": 7.500000000000001e-05,
      "loss": 2.9143,
      "step": 1536
    },
    {
      "epoch": 0.0030941521084255626,
      "grad_norm": 0.06737381219863892,
      "learning_rate": 7.65625e-05,
      "loss": 2.94,
      "step": 1568
    },
    {
      "epoch": 0.0031572980698220026,
      "grad_norm": 0.05484765022993088,
      "learning_rate": 7.8125e-05,
      "loss": 2.9153,
      "step": 1600
    },
    {
      "epoch": 0.0032204440312184426,
      "grad_norm": 0.07437755912542343,
      "learning_rate": 7.96875e-05,
      "loss": 2.8954,
      "step": 1632
    },
    {
      "epoch": 0.0032835899926148826,
      "grad_norm": 0.06876343488693237,
      "learning_rate": 8.125000000000001e-05,
      "loss": 2.8888,
      "step": 1664
    },
    {
      "epoch": 0.0033467359540113226,
      "grad_norm": 0.08410373330116272,
      "learning_rate": 8.28125e-05,
      "loss": 2.8979,
      "step": 1696
    },
    {
      "epoch": 0.0034098819154077626,
      "grad_norm": 0.09201439470052719,
      "learning_rate": 8.4375e-05,
      "loss": 2.9048,
      "step": 1728
    },
    {
      "epoch": 0.0034730278768042027,
      "grad_norm": 0.07328721880912781,
      "learning_rate": 8.593750000000001e-05,
      "loss": 2.8579,
      "step": 1760
    },
    {
      "epoch": 0.0035361738382006427,
      "grad_norm": 0.06861106306314468,
      "learning_rate": 8.75e-05,
      "loss": 2.8691,
      "step": 1792
    },
    {
      "epoch": 0.0035993197995970827,
      "grad_norm": 0.060703031718730927,
      "learning_rate": 8.90625e-05,
      "loss": 2.8603,
      "step": 1824
    },
    {
      "epoch": 0.0036624657609935227,
      "grad_norm": 0.06833851337432861,
      "learning_rate": 9.062500000000001e-05,
      "loss": 2.8635,
      "step": 1856
    },
    {
      "epoch": 0.0037256117223899627,
      "grad_norm": 0.06204499304294586,
      "learning_rate": 9.21875e-05,
      "loss": 2.8366,
      "step": 1888
    },
    {
      "epoch": 0.0037887576837864027,
      "grad_norm": 0.06592980027198792,
      "learning_rate": 9.375e-05,
      "loss": 2.8378,
      "step": 1920
    },
    {
      "epoch": 0.0038519036451828428,
      "grad_norm": 0.06360165029764175,
      "learning_rate": 9.53125e-05,
      "loss": 2.8363,
      "step": 1952
    },
    {
      "epoch": 0.003915049606579283,
      "grad_norm": 0.08830870687961578,
      "learning_rate": 9.687500000000001e-05,
      "loss": 2.8353,
      "step": 1984
    },
    {
      "epoch": 0.003978195567975723,
      "grad_norm": 0.056621383875608444,
      "learning_rate": 9.84375e-05,
      "loss": 2.8128,
      "step": 2016
    },
    {
      "epoch": 0.004041341529372163,
      "grad_norm": 0.13722540438175201,
      "learning_rate": 0.0001,
      "loss": 2.8067,
      "step": 2048
    },
    {
      "epoch": 0.004104487490768603,
      "grad_norm": 0.09581632167100906,
      "learning_rate": 9.999365977563532e-05,
      "loss": 2.8154,
      "step": 2080
    },
    {
      "epoch": 0.004167633452165043,
      "grad_norm": 0.08235198259353638,
      "learning_rate": 9.998731955127063e-05,
      "loss": 2.7958,
      "step": 2112
    },
    {
      "epoch": 0.004230779413561483,
      "grad_norm": 0.08382923156023026,
      "learning_rate": 9.998097932690593e-05,
      "loss": 2.7944,
      "step": 2144
    },
    {
      "epoch": 0.004293925374957923,
      "grad_norm": 0.06371873617172241,
      "learning_rate": 9.997463910254125e-05,
      "loss": 2.802,
      "step": 2176
    },
    {
      "epoch": 0.004357071336354363,
      "grad_norm": 0.08306098729372025,
      "learning_rate": 9.996829887817656e-05,
      "loss": 2.7686,
      "step": 2208
    },
    {
      "epoch": 0.004420217297750804,
      "grad_norm": 0.08489057421684265,
      "learning_rate": 9.996195865381187e-05,
      "loss": 2.7685,
      "step": 2240
    },
    {
      "epoch": 0.004483363259147243,
      "grad_norm": 0.12560348212718964,
      "learning_rate": 9.995561842944718e-05,
      "loss": 2.7651,
      "step": 2272
    },
    {
      "epoch": 0.004546509220543684,
      "grad_norm": 0.11862937361001968,
      "learning_rate": 9.994927820508248e-05,
      "loss": 2.7598,
      "step": 2304
    },
    {
      "epoch": 0.004609655181940123,
      "grad_norm": 0.08164895325899124,
      "learning_rate": 9.99429379807178e-05,
      "loss": 2.7607,
      "step": 2336
    },
    {
      "epoch": 0.004672801143336564,
      "grad_norm": 0.10594496130943298,
      "learning_rate": 9.993659775635311e-05,
      "loss": 2.723,
      "step": 2368
    },
    {
      "epoch": 0.004735947104733003,
      "grad_norm": 0.10034490376710892,
      "learning_rate": 9.993025753198841e-05,
      "loss": 2.7459,
      "step": 2400
    },
    {
      "epoch": 0.004799093066129444,
      "grad_norm": 0.08003844320774078,
      "learning_rate": 9.992391730762373e-05,
      "loss": 2.7368,
      "step": 2432
    },
    {
      "epoch": 0.0048622390275258835,
      "grad_norm": 0.06444230675697327,
      "learning_rate": 9.991757708325904e-05,
      "loss": 2.76,
      "step": 2464
    },
    {
      "epoch": 0.004925384988922324,
      "grad_norm": 0.1225021705031395,
      "learning_rate": 9.991123685889435e-05,
      "loss": 2.7376,
      "step": 2496
    },
    {
      "epoch": 0.0049885309503187635,
      "grad_norm": 0.15236003696918488,
      "learning_rate": 9.990489663452966e-05,
      "loss": 2.726,
      "step": 2528
    },
    {
      "epoch": 0.005051676911715204,
      "grad_norm": 0.11898784339427948,
      "learning_rate": 9.989855641016497e-05,
      "loss": 2.7456,
      "step": 2560
    },
    {
      "epoch": 0.0051148228731116435,
      "grad_norm": 0.0697527602314949,
      "learning_rate": 9.989221618580028e-05,
      "loss": 2.7145,
      "step": 2592
    },
    {
      "epoch": 0.005177968834508084,
      "grad_norm": 0.15741455554962158,
      "learning_rate": 9.98858759614356e-05,
      "loss": 2.7207,
      "step": 2624
    },
    {
      "epoch": 0.0052411147959045236,
      "grad_norm": 0.11154522001743317,
      "learning_rate": 9.98795357370709e-05,
      "loss": 2.6988,
      "step": 2656
    },
    {
      "epoch": 0.005304260757300964,
      "grad_norm": 0.14753562211990356,
      "learning_rate": 9.987319551270621e-05,
      "loss": 2.7388,
      "step": 2688
    },
    {
      "epoch": 0.0053674067186974045,
      "grad_norm": 0.1524190604686737,
      "learning_rate": 9.986685528834152e-05,
      "loss": 2.707,
      "step": 2720
    },
    {
      "epoch": 0.005430552680093844,
      "grad_norm": 0.09351467341184616,
      "learning_rate": 9.986051506397683e-05,
      "loss": 2.7009,
      "step": 2752
    },
    {
      "epoch": 0.0054936986414902845,
      "grad_norm": 0.14326535165309906,
      "learning_rate": 9.985417483961214e-05,
      "loss": 2.7022,
      "step": 2784
    },
    {
      "epoch": 0.005556844602886724,
      "grad_norm": 0.10329147428274155,
      "learning_rate": 9.984783461524745e-05,
      "loss": 2.6885,
      "step": 2816
    },
    {
      "epoch": 0.0056199905642831645,
      "grad_norm": 0.12978807091712952,
      "learning_rate": 9.984149439088276e-05,
      "loss": 2.6897,
      "step": 2848
    },
    {
      "epoch": 0.005683136525679604,
      "grad_norm": 0.13619905710220337,
      "learning_rate": 9.983515416651807e-05,
      "loss": 2.6886,
      "step": 2880
    },
    {
      "epoch": 0.0057462824870760446,
      "grad_norm": 0.11991044133901596,
      "learning_rate": 9.982881394215339e-05,
      "loss": 2.6807,
      "step": 2912
    },
    {
      "epoch": 0.005809428448472484,
      "grad_norm": 0.11966120451688766,
      "learning_rate": 9.982247371778869e-05,
      "loss": 2.6532,
      "step": 2944
    },
    {
      "epoch": 0.005872574409868925,
      "grad_norm": 0.0946977287530899,
      "learning_rate": 9.9816133493424e-05,
      "loss": 2.6688,
      "step": 2976
    },
    {
      "epoch": 0.005935720371265364,
      "grad_norm": 0.10586100816726685,
      "learning_rate": 9.980979326905932e-05,
      "loss": 2.6861,
      "step": 3008
    },
    {
      "epoch": 0.005998866332661805,
      "grad_norm": 0.10303099453449249,
      "learning_rate": 9.980345304469463e-05,
      "loss": 2.6866,
      "step": 3040
    },
    {
      "epoch": 0.006062012294058244,
      "grad_norm": 0.09335581958293915,
      "learning_rate": 9.979711282032994e-05,
      "loss": 2.6949,
      "step": 3072
    },
    {
      "epoch": 0.006125158255454685,
      "grad_norm": 0.1507342904806137,
      "learning_rate": 9.979077259596524e-05,
      "loss": 2.6616,
      "step": 3104
    },
    {
      "epoch": 0.006188304216851125,
      "grad_norm": 0.15320587158203125,
      "learning_rate": 9.978443237160055e-05,
      "loss": 2.6616,
      "step": 3136
    },
    {
      "epoch": 0.006251450178247565,
      "grad_norm": 0.12013068795204163,
      "learning_rate": 9.977809214723587e-05,
      "loss": 2.6482,
      "step": 3168
    },
    {
      "epoch": 0.006314596139644005,
      "grad_norm": 0.1357378363609314,
      "learning_rate": 9.977175192287117e-05,
      "loss": 2.6615,
      "step": 3200
    },
    {
      "epoch": 0.006377742101040445,
      "grad_norm": 0.12570512294769287,
      "learning_rate": 9.976541169850648e-05,
      "loss": 2.6419,
      "step": 3232
    },
    {
      "epoch": 0.006440888062436885,
      "grad_norm": 0.13948673009872437,
      "learning_rate": 9.97590714741418e-05,
      "loss": 2.6446,
      "step": 3264
    },
    {
      "epoch": 0.006504034023833325,
      "grad_norm": 0.08918385952711105,
      "learning_rate": 9.975273124977711e-05,
      "loss": 2.6327,
      "step": 3296
    },
    {
      "epoch": 0.006567179985229765,
      "grad_norm": 0.14465107023715973,
      "learning_rate": 9.974639102541241e-05,
      "loss": 2.6517,
      "step": 3328
    },
    {
      "epoch": 0.006630325946626205,
      "grad_norm": 0.10744307190179825,
      "learning_rate": 9.974005080104773e-05,
      "loss": 2.6524,
      "step": 3360
    },
    {
      "epoch": 0.006693471908022645,
      "grad_norm": 0.1016412228345871,
      "learning_rate": 9.973371057668304e-05,
      "loss": 2.6375,
      "step": 3392
    },
    {
      "epoch": 0.006756617869419085,
      "grad_norm": 0.1091398224234581,
      "learning_rate": 9.972737035231835e-05,
      "loss": 2.6379,
      "step": 3424
    },
    {
      "epoch": 0.006819763830815525,
      "grad_norm": 0.0891808420419693,
      "learning_rate": 9.972103012795367e-05,
      "loss": 2.6314,
      "step": 3456
    },
    {
      "epoch": 0.006882909792211965,
      "grad_norm": 0.11935264617204666,
      "learning_rate": 9.971468990358897e-05,
      "loss": 2.6265,
      "step": 3488
    },
    {
      "epoch": 0.006946055753608405,
      "grad_norm": 0.10122443735599518,
      "learning_rate": 9.970834967922427e-05,
      "loss": 2.6441,
      "step": 3520
    },
    {
      "epoch": 0.007009201715004845,
      "grad_norm": 0.12963150441646576,
      "learning_rate": 9.970200945485959e-05,
      "loss": 2.6326,
      "step": 3552
    },
    {
      "epoch": 0.007072347676401285,
      "grad_norm": 0.2340211272239685,
      "learning_rate": 9.96956692304949e-05,
      "loss": 2.6125,
      "step": 3584
    },
    {
      "epoch": 0.007135493637797726,
      "grad_norm": 0.1019572764635086,
      "learning_rate": 9.96893290061302e-05,
      "loss": 2.6032,
      "step": 3616
    },
    {
      "epoch": 0.007198639599194165,
      "grad_norm": 0.18724122643470764,
      "learning_rate": 9.968298878176552e-05,
      "loss": 2.6258,
      "step": 3648
    },
    {
      "epoch": 0.007261785560590606,
      "grad_norm": 0.14609533548355103,
      "learning_rate": 9.967664855740083e-05,
      "loss": 2.5863,
      "step": 3680
    },
    {
      "epoch": 0.007324931521987045,
      "grad_norm": 0.11218056827783585,
      "learning_rate": 9.967030833303615e-05,
      "loss": 2.6391,
      "step": 3712
    },
    {
      "epoch": 0.007388077483383486,
      "grad_norm": 0.09486357867717743,
      "learning_rate": 9.966396810867145e-05,
      "loss": 2.5997,
      "step": 3744
    },
    {
      "epoch": 0.0074512234447799254,
      "grad_norm": 0.1694619506597519,
      "learning_rate": 9.965762788430676e-05,
      "loss": 2.5852,
      "step": 3776
    },
    {
      "epoch": 0.007514369406176366,
      "grad_norm": 0.13817843794822693,
      "learning_rate": 9.965128765994208e-05,
      "loss": 2.5845,
      "step": 3808
    },
    {
      "epoch": 0.0075775153675728055,
      "grad_norm": 0.16682597994804382,
      "learning_rate": 9.964494743557739e-05,
      "loss": 2.6041,
      "step": 3840
    },
    {
      "epoch": 0.007640661328969246,
      "grad_norm": 0.08626808971166611,
      "learning_rate": 9.963860721121269e-05,
      "loss": 2.6296,
      "step": 3872
    },
    {
      "epoch": 0.0077038072903656855,
      "grad_norm": 0.11467722803354263,
      "learning_rate": 9.9632266986848e-05,
      "loss": 2.5933,
      "step": 3904
    },
    {
      "epoch": 0.007766953251762126,
      "grad_norm": 0.11957038938999176,
      "learning_rate": 9.96259267624833e-05,
      "loss": 2.5891,
      "step": 3936
    },
    {
      "epoch": 0.007830099213158566,
      "grad_norm": 0.10939373075962067,
      "learning_rate": 9.961958653811862e-05,
      "loss": 2.5775,
      "step": 3968
    },
    {
      "epoch": 0.007893245174555006,
      "grad_norm": 0.13572102785110474,
      "learning_rate": 9.961324631375392e-05,
      "loss": 2.5927,
      "step": 4000
    },
    {
      "epoch": 0.007956391135951446,
      "grad_norm": 0.10408812016248703,
      "learning_rate": 9.960690608938924e-05,
      "loss": 2.5642,
      "step": 4032
    },
    {
      "epoch": 0.008019537097347887,
      "grad_norm": 0.10480830818414688,
      "learning_rate": 9.960056586502455e-05,
      "loss": 2.5907,
      "step": 4064
    },
    {
      "epoch": 0.008082683058744326,
      "grad_norm": 0.13999754190444946,
      "learning_rate": 9.959422564065987e-05,
      "loss": 2.5944,
      "step": 4096
    },
    {
      "epoch": 0.008145829020140766,
      "grad_norm": 0.12907759845256805,
      "learning_rate": 9.958788541629518e-05,
      "loss": 2.5821,
      "step": 4128
    },
    {
      "epoch": 0.008208974981537207,
      "grad_norm": 0.12192639708518982,
      "learning_rate": 9.958154519193048e-05,
      "loss": 2.5634,
      "step": 4160
    },
    {
      "epoch": 0.008272120942933647,
      "grad_norm": 0.09974056482315063,
      "learning_rate": 9.95752049675658e-05,
      "loss": 2.5784,
      "step": 4192
    },
    {
      "epoch": 0.008335266904330086,
      "grad_norm": 0.12721870839595795,
      "learning_rate": 9.956886474320111e-05,
      "loss": 2.5651,
      "step": 4224
    },
    {
      "epoch": 0.008398412865726526,
      "grad_norm": 0.11077609658241272,
      "learning_rate": 9.956252451883643e-05,
      "loss": 2.5719,
      "step": 4256
    },
    {
      "epoch": 0.008461558827122967,
      "grad_norm": 0.16812993586063385,
      "learning_rate": 9.955618429447173e-05,
      "loss": 2.5628,
      "step": 4288
    },
    {
      "epoch": 0.008524704788519407,
      "grad_norm": 0.12394724786281586,
      "learning_rate": 9.954984407010704e-05,
      "loss": 2.5557,
      "step": 4320
    },
    {
      "epoch": 0.008587850749915846,
      "grad_norm": 0.09741286188364029,
      "learning_rate": 9.954350384574234e-05,
      "loss": 2.5337,
      "step": 4352
    },
    {
      "epoch": 0.008650996711312286,
      "grad_norm": 0.15195594727993011,
      "learning_rate": 9.953716362137766e-05,
      "loss": 2.5611,
      "step": 4384
    },
    {
      "epoch": 0.008714142672708727,
      "grad_norm": 0.12760357558727264,
      "learning_rate": 9.953082339701296e-05,
      "loss": 2.5581,
      "step": 4416
    },
    {
      "epoch": 0.008777288634105167,
      "grad_norm": 0.12614698708057404,
      "learning_rate": 9.952448317264827e-05,
      "loss": 2.5575,
      "step": 4448
    },
    {
      "epoch": 0.008840434595501608,
      "grad_norm": 0.1622260957956314,
      "learning_rate": 9.951814294828359e-05,
      "loss": 2.5658,
      "step": 4480
    },
    {
      "epoch": 0.008903580556898046,
      "grad_norm": 0.14965364336967468,
      "learning_rate": 9.95118027239189e-05,
      "loss": 2.5405,
      "step": 4512
    },
    {
      "epoch": 0.008966726518294487,
      "grad_norm": 0.12377427518367767,
      "learning_rate": 9.95054624995542e-05,
      "loss": 2.5316,
      "step": 4544
    },
    {
      "epoch": 0.009029872479690927,
      "grad_norm": 0.12055251747369766,
      "learning_rate": 9.949912227518952e-05,
      "loss": 2.5432,
      "step": 4576
    },
    {
      "epoch": 0.009093018441087368,
      "grad_norm": 0.12006771564483643,
      "learning_rate": 9.949278205082483e-05,
      "loss": 2.548,
      "step": 4608
    },
    {
      "epoch": 0.009156164402483806,
      "grad_norm": 0.12881246209144592,
      "learning_rate": 9.948644182646015e-05,
      "loss": 2.5318,
      "step": 4640
    },
    {
      "epoch": 0.009219310363880247,
      "grad_norm": 0.19436371326446533,
      "learning_rate": 9.948010160209545e-05,
      "loss": 2.5281,
      "step": 4672
    },
    {
      "epoch": 0.009282456325276687,
      "grad_norm": 0.16575217247009277,
      "learning_rate": 9.947376137773076e-05,
      "loss": 2.5385,
      "step": 4704
    },
    {
      "epoch": 0.009345602286673128,
      "grad_norm": 0.14120186865329742,
      "learning_rate": 9.946742115336608e-05,
      "loss": 2.5213,
      "step": 4736
    },
    {
      "epoch": 0.009408748248069566,
      "grad_norm": 0.1341506838798523,
      "learning_rate": 9.946108092900138e-05,
      "loss": 2.5289,
      "step": 4768
    },
    {
      "epoch": 0.009471894209466007,
      "grad_norm": 0.08819320797920227,
      "learning_rate": 9.945474070463669e-05,
      "loss": 2.5144,
      "step": 4800
    },
    {
      "epoch": 0.009535040170862447,
      "grad_norm": 0.10903503000736237,
      "learning_rate": 9.944840048027199e-05,
      "loss": 2.5452,
      "step": 4832
    },
    {
      "epoch": 0.009598186132258888,
      "grad_norm": 0.1760348528623581,
      "learning_rate": 9.944206025590731e-05,
      "loss": 2.5241,
      "step": 4864
    },
    {
      "epoch": 0.009661332093655326,
      "grad_norm": 0.1575450599193573,
      "learning_rate": 9.943572003154262e-05,
      "loss": 2.5206,
      "step": 4896
    },
    {
      "epoch": 0.009724478055051767,
      "grad_norm": 0.14991284906864166,
      "learning_rate": 9.942937980717794e-05,
      "loss": 2.514,
      "step": 4928
    },
    {
      "epoch": 0.009787624016448207,
      "grad_norm": 0.13919000327587128,
      "learning_rate": 9.942303958281324e-05,
      "loss": 2.5254,
      "step": 4960
    },
    {
      "epoch": 0.009850769977844648,
      "grad_norm": 0.14055348932743073,
      "learning_rate": 9.941669935844855e-05,
      "loss": 2.5022,
      "step": 4992
    },
    {
      "epoch": 0.009913915939241088,
      "grad_norm": 0.13361525535583496,
      "learning_rate": 9.941035913408387e-05,
      "loss": 2.5265,
      "step": 5024
    },
    {
      "epoch": 0.009977061900637527,
      "grad_norm": 0.09767939150333405,
      "learning_rate": 9.940401890971918e-05,
      "loss": 2.5184,
      "step": 5056
    },
    {
      "epoch": 0.010040207862033967,
      "grad_norm": 0.10839032381772995,
      "learning_rate": 9.939767868535448e-05,
      "loss": 2.5075,
      "step": 5088
    },
    {
      "epoch": 0.010103353823430408,
      "grad_norm": 0.17004302144050598,
      "learning_rate": 9.93913384609898e-05,
      "loss": 2.4998,
      "step": 5120
    },
    {
      "epoch": 0.010166499784826848,
      "grad_norm": 0.15774109959602356,
      "learning_rate": 9.938499823662511e-05,
      "loss": 2.4954,
      "step": 5152
    },
    {
      "epoch": 0.010229645746223287,
      "grad_norm": 0.11758945882320404,
      "learning_rate": 9.937865801226041e-05,
      "loss": 2.5128,
      "step": 5184
    },
    {
      "epoch": 0.010292791707619727,
      "grad_norm": 0.12542724609375,
      "learning_rate": 9.937231778789571e-05,
      "loss": 2.4798,
      "step": 5216
    },
    {
      "epoch": 0.010355937669016168,
      "grad_norm": 0.14986108243465424,
      "learning_rate": 9.936597756353103e-05,
      "loss": 2.4902,
      "step": 5248
    },
    {
      "epoch": 0.010419083630412608,
      "grad_norm": 0.14334240555763245,
      "learning_rate": 9.935963733916634e-05,
      "loss": 2.5063,
      "step": 5280
    },
    {
      "epoch": 0.010482229591809047,
      "grad_norm": 0.15978749096393585,
      "learning_rate": 9.935329711480166e-05,
      "loss": 2.4669,
      "step": 5312
    },
    {
      "epoch": 0.010545375553205488,
      "grad_norm": 0.14394745230674744,
      "learning_rate": 9.934695689043696e-05,
      "loss": 2.4786,
      "step": 5344
    },
    {
      "epoch": 0.010608521514601928,
      "grad_norm": 0.11807334423065186,
      "learning_rate": 9.934061666607227e-05,
      "loss": 2.4714,
      "step": 5376
    },
    {
      "epoch": 0.010671667475998368,
      "grad_norm": 0.11549796909093857,
      "learning_rate": 9.933427644170759e-05,
      "loss": 2.4937,
      "step": 5408
    },
    {
      "epoch": 0.010734813437394809,
      "grad_norm": 0.09807826578617096,
      "learning_rate": 9.93279362173429e-05,
      "loss": 2.5044,
      "step": 5440
    },
    {
      "epoch": 0.010797959398791248,
      "grad_norm": 0.15426675975322723,
      "learning_rate": 9.932159599297822e-05,
      "loss": 2.4747,
      "step": 5472
    },
    {
      "epoch": 0.010861105360187688,
      "grad_norm": 0.11509829014539719,
      "learning_rate": 9.931525576861352e-05,
      "loss": 2.4819,
      "step": 5504
    },
    {
      "epoch": 0.010924251321584129,
      "grad_norm": 0.11389163881540298,
      "learning_rate": 9.930891554424883e-05,
      "loss": 2.4693,
      "step": 5536
    },
    {
      "epoch": 0.010987397282980569,
      "grad_norm": 0.15173538029193878,
      "learning_rate": 9.930257531988413e-05,
      "loss": 2.4741,
      "step": 5568
    },
    {
      "epoch": 0.011050543244377008,
      "grad_norm": 0.08514861762523651,
      "learning_rate": 9.929623509551945e-05,
      "loss": 2.4691,
      "step": 5600
    },
    {
      "epoch": 0.011113689205773448,
      "grad_norm": 0.14991246163845062,
      "learning_rate": 9.928989487115475e-05,
      "loss": 2.4765,
      "step": 5632
    },
    {
      "epoch": 0.011176835167169889,
      "grad_norm": 0.18143261969089508,
      "learning_rate": 9.928355464679006e-05,
      "loss": 2.4821,
      "step": 5664
    },
    {
      "epoch": 0.011239981128566329,
      "grad_norm": 0.09394911676645279,
      "learning_rate": 9.927721442242538e-05,
      "loss": 2.4814,
      "step": 5696
    },
    {
      "epoch": 0.011303127089962768,
      "grad_norm": 0.14808283746242523,
      "learning_rate": 9.927087419806069e-05,
      "loss": 2.4528,
      "step": 5728
    },
    {
      "epoch": 0.011366273051359208,
      "grad_norm": 0.15319721400737762,
      "learning_rate": 9.926453397369599e-05,
      "loss": 2.4617,
      "step": 5760
    },
    {
      "epoch": 0.011429419012755649,
      "grad_norm": 0.1394672840833664,
      "learning_rate": 9.925819374933131e-05,
      "loss": 2.4676,
      "step": 5792
    },
    {
      "epoch": 0.011492564974152089,
      "grad_norm": 0.12285158038139343,
      "learning_rate": 9.925185352496662e-05,
      "loss": 2.4441,
      "step": 5824
    },
    {
      "epoch": 0.01155571093554853,
      "grad_norm": 0.1122804805636406,
      "learning_rate": 9.924551330060194e-05,
      "loss": 2.4419,
      "step": 5856
    },
    {
      "epoch": 0.011618856896944968,
      "grad_norm": 0.15353912115097046,
      "learning_rate": 9.923917307623724e-05,
      "loss": 2.4574,
      "step": 5888
    },
    {
      "epoch": 0.011682002858341409,
      "grad_norm": 0.11223366111516953,
      "learning_rate": 9.923283285187255e-05,
      "loss": 2.4401,
      "step": 5920
    },
    {
      "epoch": 0.01174514881973785,
      "grad_norm": 0.144102081656456,
      "learning_rate": 9.922649262750787e-05,
      "loss": 2.4421,
      "step": 5952
    },
    {
      "epoch": 0.01180829478113429,
      "grad_norm": 0.11691421270370483,
      "learning_rate": 9.922015240314317e-05,
      "loss": 2.4482,
      "step": 5984
    },
    {
      "epoch": 0.011871440742530728,
      "grad_norm": 0.10664689540863037,
      "learning_rate": 9.921381217877848e-05,
      "loss": 2.4675,
      "step": 6016
    },
    {
      "epoch": 0.011934586703927169,
      "grad_norm": 0.17872704565525055,
      "learning_rate": 9.920747195441378e-05,
      "loss": 2.4587,
      "step": 6048
    },
    {
      "epoch": 0.01199773266532361,
      "grad_norm": 0.18183505535125732,
      "learning_rate": 9.92011317300491e-05,
      "loss": 2.456,
      "step": 6080
    },
    {
      "epoch": 0.01206087862672005,
      "grad_norm": 0.1468466818332672,
      "learning_rate": 9.919479150568441e-05,
      "loss": 2.4525,
      "step": 6112
    },
    {
      "epoch": 0.012124024588116488,
      "grad_norm": 0.1823159009218216,
      "learning_rate": 9.918845128131973e-05,
      "loss": 2.4499,
      "step": 6144
    },
    {
      "epoch": 0.012187170549512929,
      "grad_norm": 0.1419377326965332,
      "learning_rate": 9.918211105695503e-05,
      "loss": 2.4478,
      "step": 6176
    },
    {
      "epoch": 0.01225031651090937,
      "grad_norm": 0.1857587844133377,
      "learning_rate": 9.917577083259034e-05,
      "loss": 2.4584,
      "step": 6208
    },
    {
      "epoch": 0.01231346247230581,
      "grad_norm": 0.11730950325727463,
      "learning_rate": 9.916943060822566e-05,
      "loss": 2.4331,
      "step": 6240
    },
    {
      "epoch": 0.01237660843370225,
      "grad_norm": 0.13131648302078247,
      "learning_rate": 9.916309038386097e-05,
      "loss": 2.4243,
      "step": 6272
    },
    {
      "epoch": 0.012439754395098689,
      "grad_norm": 0.1440175622701645,
      "learning_rate": 9.915675015949627e-05,
      "loss": 2.4368,
      "step": 6304
    },
    {
      "epoch": 0.01250290035649513,
      "grad_norm": 0.13897374272346497,
      "learning_rate": 9.915040993513159e-05,
      "loss": 2.4205,
      "step": 6336
    },
    {
      "epoch": 0.01256604631789157,
      "grad_norm": 0.11762618273496628,
      "learning_rate": 9.91440697107669e-05,
      "loss": 2.4145,
      "step": 6368
    },
    {
      "epoch": 0.01262919227928801,
      "grad_norm": 0.13914981484413147,
      "learning_rate": 9.91377294864022e-05,
      "loss": 2.4289,
      "step": 6400
    },
    {
      "epoch": 0.012692338240684449,
      "grad_norm": 0.13673797249794006,
      "learning_rate": 9.913138926203752e-05,
      "loss": 2.4316,
      "step": 6432
    },
    {
      "epoch": 0.01275548420208089,
      "grad_norm": 0.13205789029598236,
      "learning_rate": 9.912504903767282e-05,
      "loss": 2.4166,
      "step": 6464
    },
    {
      "epoch": 0.01281863016347733,
      "grad_norm": 0.1587177962064743,
      "learning_rate": 9.911870881330813e-05,
      "loss": 2.4257,
      "step": 6496
    },
    {
      "epoch": 0.01288177612487377,
      "grad_norm": 0.15956047177314758,
      "learning_rate": 9.911236858894345e-05,
      "loss": 2.4288,
      "step": 6528
    },
    {
      "epoch": 0.012944922086270209,
      "grad_norm": 0.14064407348632812,
      "learning_rate": 9.910602836457875e-05,
      "loss": 2.394,
      "step": 6560
    },
    {
      "epoch": 0.01300806804766665,
      "grad_norm": 0.13206064701080322,
      "learning_rate": 9.909968814021406e-05,
      "loss": 2.4063,
      "step": 6592
    },
    {
      "epoch": 0.01307121400906309,
      "grad_norm": 0.13762450218200684,
      "learning_rate": 9.909334791584938e-05,
      "loss": 2.4285,
      "step": 6624
    },
    {
      "epoch": 0.01313435997045953,
      "grad_norm": 0.12068444490432739,
      "learning_rate": 9.908700769148469e-05,
      "loss": 2.3994,
      "step": 6656
    },
    {
      "epoch": 0.013197505931855969,
      "grad_norm": 0.1713280975818634,
      "learning_rate": 9.908066746711999e-05,
      "loss": 2.4185,
      "step": 6688
    },
    {
      "epoch": 0.01326065189325241,
      "grad_norm": 0.15375898778438568,
      "learning_rate": 9.907432724275531e-05,
      "loss": 2.4047,
      "step": 6720
    },
    {
      "epoch": 0.01332379785464885,
      "grad_norm": 0.13617244362831116,
      "learning_rate": 9.906798701839062e-05,
      "loss": 2.4041,
      "step": 6752
    },
    {
      "epoch": 0.01338694381604529,
      "grad_norm": 0.10807432234287262,
      "learning_rate": 9.906164679402594e-05,
      "loss": 2.3981,
      "step": 6784
    },
    {
      "epoch": 0.013450089777441731,
      "grad_norm": 0.12593461573123932,
      "learning_rate": 9.905530656966124e-05,
      "loss": 2.4024,
      "step": 6816
    },
    {
      "epoch": 0.01351323573883817,
      "grad_norm": 0.1572636216878891,
      "learning_rate": 9.904896634529655e-05,
      "loss": 2.3851,
      "step": 6848
    },
    {
      "epoch": 0.01357638170023461,
      "grad_norm": 0.12052024155855179,
      "learning_rate": 9.904262612093185e-05,
      "loss": 2.4108,
      "step": 6880
    },
    {
      "epoch": 0.01363952766163105,
      "grad_norm": 0.12341807782649994,
      "learning_rate": 9.903628589656717e-05,
      "loss": 2.378,
      "step": 6912
    },
    {
      "epoch": 0.013702673623027491,
      "grad_norm": 0.11022727936506271,
      "learning_rate": 9.902994567220248e-05,
      "loss": 2.3936,
      "step": 6944
    },
    {
      "epoch": 0.01376581958442393,
      "grad_norm": 0.1105339452624321,
      "learning_rate": 9.902360544783778e-05,
      "loss": 2.3723,
      "step": 6976
    },
    {
      "epoch": 0.01382896554582037,
      "grad_norm": 0.10231024026870728,
      "learning_rate": 9.90172652234731e-05,
      "loss": 2.3974,
      "step": 7008
    },
    {
      "epoch": 0.01389211150721681,
      "grad_norm": 0.1630864292383194,
      "learning_rate": 9.901092499910841e-05,
      "loss": 2.3695,
      "step": 7040
    },
    {
      "epoch": 0.013955257468613251,
      "grad_norm": 0.13933351635932922,
      "learning_rate": 9.900458477474373e-05,
      "loss": 2.4029,
      "step": 7072
    },
    {
      "epoch": 0.01401840343000969,
      "grad_norm": 0.12549299001693726,
      "learning_rate": 9.899824455037903e-05,
      "loss": 2.402,
      "step": 7104
    },
    {
      "epoch": 0.01408154939140613,
      "grad_norm": 0.16632123291492462,
      "learning_rate": 9.899190432601434e-05,
      "loss": 2.3822,
      "step": 7136
    },
    {
      "epoch": 0.01414469535280257,
      "grad_norm": 0.0944233387708664,
      "learning_rate": 9.898556410164966e-05,
      "loss": 2.3816,
      "step": 7168
    },
    {
      "epoch": 0.014207841314199011,
      "grad_norm": 0.1514974981546402,
      "learning_rate": 9.897922387728497e-05,
      "loss": 2.3894,
      "step": 7200
    },
    {
      "epoch": 0.014270987275595452,
      "grad_norm": 0.10387679934501648,
      "learning_rate": 9.897288365292027e-05,
      "loss": 2.3789,
      "step": 7232
    },
    {
      "epoch": 0.01433413323699189,
      "grad_norm": 0.12096551805734634,
      "learning_rate": 9.896654342855557e-05,
      "loss": 2.3716,
      "step": 7264
    },
    {
      "epoch": 0.01439727919838833,
      "grad_norm": 0.12126098573207855,
      "learning_rate": 9.896020320419089e-05,
      "loss": 2.371,
      "step": 7296
    },
    {
      "epoch": 0.014460425159784771,
      "grad_norm": 0.1568283885717392,
      "learning_rate": 9.89538629798262e-05,
      "loss": 2.3794,
      "step": 7328
    },
    {
      "epoch": 0.014523571121181212,
      "grad_norm": 0.14460304379463196,
      "learning_rate": 9.894752275546152e-05,
      "loss": 2.3773,
      "step": 7360
    },
    {
      "epoch": 0.01458671708257765,
      "grad_norm": 0.11914809048175812,
      "learning_rate": 9.894118253109682e-05,
      "loss": 2.3522,
      "step": 7392
    },
    {
      "epoch": 0.01464986304397409,
      "grad_norm": 0.11162387579679489,
      "learning_rate": 9.893484230673213e-05,
      "loss": 2.3808,
      "step": 7424
    },
    {
      "epoch": 0.014713009005370531,
      "grad_norm": 0.170000359416008,
      "learning_rate": 9.892850208236745e-05,
      "loss": 2.3767,
      "step": 7456
    },
    {
      "epoch": 0.014776154966766972,
      "grad_norm": 0.1726778894662857,
      "learning_rate": 9.892216185800276e-05,
      "loss": 2.3874,
      "step": 7488
    },
    {
      "epoch": 0.01483930092816341,
      "grad_norm": 0.15378707647323608,
      "learning_rate": 9.891582163363806e-05,
      "loss": 2.3731,
      "step": 7520
    },
    {
      "epoch": 0.014902446889559851,
      "grad_norm": 0.12425557523965836,
      "learning_rate": 9.890948140927338e-05,
      "loss": 2.366,
      "step": 7552
    },
    {
      "epoch": 0.014965592850956291,
      "grad_norm": 0.12605996429920197,
      "learning_rate": 9.890314118490869e-05,
      "loss": 2.3739,
      "step": 7584
    },
    {
      "epoch": 0.015028738812352732,
      "grad_norm": 0.12717851996421814,
      "learning_rate": 9.889680096054401e-05,
      "loss": 2.3655,
      "step": 7616
    },
    {
      "epoch": 0.015091884773749172,
      "grad_norm": 0.16373498737812042,
      "learning_rate": 9.889046073617931e-05,
      "loss": 2.3521,
      "step": 7648
    },
    {
      "epoch": 0.015155030735145611,
      "grad_norm": 0.1463109701871872,
      "learning_rate": 9.888412051181461e-05,
      "loss": 2.3661,
      "step": 7680
    },
    {
      "epoch": 0.015218176696542051,
      "grad_norm": 0.13224859535694122,
      "learning_rate": 9.887778028744992e-05,
      "loss": 2.3706,
      "step": 7712
    },
    {
      "epoch": 0.015281322657938492,
      "grad_norm": 0.11841218918561935,
      "learning_rate": 9.887144006308524e-05,
      "loss": 2.3845,
      "step": 7744
    },
    {
      "epoch": 0.015344468619334932,
      "grad_norm": 0.13370342552661896,
      "learning_rate": 9.886509983872054e-05,
      "loss": 2.3745,
      "step": 7776
    },
    {
      "epoch": 0.015407614580731371,
      "grad_norm": 0.12113891541957855,
      "learning_rate": 9.885875961435585e-05,
      "loss": 2.3523,
      "step": 7808
    },
    {
      "epoch": 0.015470760542127811,
      "grad_norm": 0.1323300004005432,
      "learning_rate": 9.885241938999117e-05,
      "loss": 2.3777,
      "step": 7840
    },
    {
      "epoch": 0.015533906503524252,
      "grad_norm": 0.12009269744157791,
      "learning_rate": 9.884607916562648e-05,
      "loss": 2.3426,
      "step": 7872
    },
    {
      "epoch": 0.015597052464920692,
      "grad_norm": 0.11964011192321777,
      "learning_rate": 9.883973894126178e-05,
      "loss": 2.3754,
      "step": 7904
    },
    {
      "epoch": 0.01566019842631713,
      "grad_norm": 0.1414196938276291,
      "learning_rate": 9.88333987168971e-05,
      "loss": 2.3454,
      "step": 7936
    },
    {
      "epoch": 0.015723344387713573,
      "grad_norm": 0.11974673718214035,
      "learning_rate": 9.882705849253241e-05,
      "loss": 2.3345,
      "step": 7968
    },
    {
      "epoch": 0.015786490349110012,
      "grad_norm": 0.12504935264587402,
      "learning_rate": 9.882071826816773e-05,
      "loss": 2.3336,
      "step": 8000
    },
    {
      "epoch": 0.01584963631050645,
      "grad_norm": 0.13010753691196442,
      "learning_rate": 9.881437804380304e-05,
      "loss": 2.3503,
      "step": 8032
    },
    {
      "epoch": 0.015912782271902893,
      "grad_norm": 0.10797622054815292,
      "learning_rate": 9.880803781943834e-05,
      "loss": 2.3447,
      "step": 8064
    },
    {
      "epoch": 0.01597592823329933,
      "grad_norm": 0.1536203771829605,
      "learning_rate": 9.880169759507364e-05,
      "loss": 2.332,
      "step": 8096
    },
    {
      "epoch": 0.016039074194695774,
      "grad_norm": 0.1467064470052719,
      "learning_rate": 9.879535737070896e-05,
      "loss": 2.3462,
      "step": 8128
    },
    {
      "epoch": 0.016102220156092212,
      "grad_norm": 0.1106255054473877,
      "learning_rate": 9.878901714634427e-05,
      "loss": 2.3298,
      "step": 8160
    },
    {
      "epoch": 0.01616536611748865,
      "grad_norm": 0.12443240731954575,
      "learning_rate": 9.878267692197957e-05,
      "loss": 2.3472,
      "step": 8192
    },
    {
      "epoch": 0.016228512078885093,
      "grad_norm": 0.12979218363761902,
      "learning_rate": 9.877633669761489e-05,
      "loss": 2.3486,
      "step": 8224
    },
    {
      "epoch": 0.016291658040281532,
      "grad_norm": 0.10633663088083267,
      "learning_rate": 9.87699964732502e-05,
      "loss": 2.3483,
      "step": 8256
    },
    {
      "epoch": 0.01635480400167797,
      "grad_norm": 0.13850173354148865,
      "learning_rate": 9.876365624888552e-05,
      "loss": 2.3359,
      "step": 8288
    },
    {
      "epoch": 0.016417949963074413,
      "grad_norm": 0.16345906257629395,
      "learning_rate": 9.875731602452082e-05,
      "loss": 2.332,
      "step": 8320
    },
    {
      "epoch": 0.01648109592447085,
      "grad_norm": 0.1388985812664032,
      "learning_rate": 9.875097580015613e-05,
      "loss": 2.3383,
      "step": 8352
    },
    {
      "epoch": 0.016544241885867294,
      "grad_norm": 0.10737188905477524,
      "learning_rate": 9.874463557579145e-05,
      "loss": 2.3233,
      "step": 8384
    },
    {
      "epoch": 0.016607387847263733,
      "grad_norm": 0.15225957334041595,
      "learning_rate": 9.873829535142676e-05,
      "loss": 2.319,
      "step": 8416
    },
    {
      "epoch": 0.01667053380866017,
      "grad_norm": 0.15039414167404175,
      "learning_rate": 9.873195512706206e-05,
      "loss": 2.3375,
      "step": 8448
    },
    {
      "epoch": 0.016733679770056614,
      "grad_norm": 0.15000516176223755,
      "learning_rate": 9.872561490269738e-05,
      "loss": 2.3286,
      "step": 8480
    },
    {
      "epoch": 0.016796825731453052,
      "grad_norm": 0.12463720142841339,
      "learning_rate": 9.871927467833268e-05,
      "loss": 2.3407,
      "step": 8512
    },
    {
      "epoch": 0.016859971692849494,
      "grad_norm": 0.1400393396615982,
      "learning_rate": 9.8712934453968e-05,
      "loss": 2.3171,
      "step": 8544
    },
    {
      "epoch": 0.016923117654245933,
      "grad_norm": 0.15409910678863525,
      "learning_rate": 9.87065942296033e-05,
      "loss": 2.3156,
      "step": 8576
    },
    {
      "epoch": 0.016986263615642372,
      "grad_norm": 0.13042177259922028,
      "learning_rate": 9.870025400523861e-05,
      "loss": 2.3182,
      "step": 8608
    },
    {
      "epoch": 0.017049409577038814,
      "grad_norm": 0.1081709936261177,
      "learning_rate": 9.869391378087392e-05,
      "loss": 2.3263,
      "step": 8640
    },
    {
      "epoch": 0.017112555538435253,
      "grad_norm": 0.12457891553640366,
      "learning_rate": 9.868757355650924e-05,
      "loss": 2.325,
      "step": 8672
    },
    {
      "epoch": 0.01717570149983169,
      "grad_norm": 0.17015835642814636,
      "learning_rate": 9.868123333214455e-05,
      "loss": 2.3214,
      "step": 8704
    },
    {
      "epoch": 0.017238847461228134,
      "grad_norm": 0.13807617127895355,
      "learning_rate": 9.867489310777985e-05,
      "loss": 2.315,
      "step": 8736
    },
    {
      "epoch": 0.017301993422624572,
      "grad_norm": 0.11597169935703278,
      "learning_rate": 9.866855288341517e-05,
      "loss": 2.3442,
      "step": 8768
    },
    {
      "epoch": 0.017365139384021015,
      "grad_norm": 0.13555516302585602,
      "learning_rate": 9.866221265905048e-05,
      "loss": 2.3178,
      "step": 8800
    },
    {
      "epoch": 0.017428285345417453,
      "grad_norm": 0.14685258269309998,
      "learning_rate": 9.86558724346858e-05,
      "loss": 2.3154,
      "step": 8832
    },
    {
      "epoch": 0.017491431306813892,
      "grad_norm": 0.1215466782450676,
      "learning_rate": 9.86495322103211e-05,
      "loss": 2.3259,
      "step": 8864
    },
    {
      "epoch": 0.017554577268210334,
      "grad_norm": 0.13843005895614624,
      "learning_rate": 9.864319198595641e-05,
      "loss": 2.3255,
      "step": 8896
    },
    {
      "epoch": 0.017617723229606773,
      "grad_norm": 0.1204628273844719,
      "learning_rate": 9.863685176159171e-05,
      "loss": 2.32,
      "step": 8928
    },
    {
      "epoch": 0.017680869191003215,
      "grad_norm": 0.13078300654888153,
      "learning_rate": 9.863051153722703e-05,
      "loss": 2.3022,
      "step": 8960
    },
    {
      "epoch": 0.017744015152399654,
      "grad_norm": 0.09985719621181488,
      "learning_rate": 9.862417131286233e-05,
      "loss": 2.3046,
      "step": 8992
    },
    {
      "epoch": 0.017807161113796093,
      "grad_norm": 0.16064776480197906,
      "learning_rate": 9.861783108849764e-05,
      "loss": 2.3086,
      "step": 9024
    },
    {
      "epoch": 0.017870307075192535,
      "grad_norm": 0.15166877210140228,
      "learning_rate": 9.861149086413296e-05,
      "loss": 2.319,
      "step": 9056
    },
    {
      "epoch": 0.017933453036588973,
      "grad_norm": 0.11681035161018372,
      "learning_rate": 9.860515063976827e-05,
      "loss": 2.3085,
      "step": 9088
    },
    {
      "epoch": 0.017996598997985412,
      "grad_norm": 0.16018840670585632,
      "learning_rate": 9.859881041540357e-05,
      "loss": 2.3146,
      "step": 9120
    },
    {
      "epoch": 0.018059744959381854,
      "grad_norm": 0.12750431895256042,
      "learning_rate": 9.859247019103889e-05,
      "loss": 2.3054,
      "step": 9152
    },
    {
      "epoch": 0.018122890920778293,
      "grad_norm": 0.1396716833114624,
      "learning_rate": 9.85861299666742e-05,
      "loss": 2.3153,
      "step": 9184
    },
    {
      "epoch": 0.018186036882174735,
      "grad_norm": 0.1469569206237793,
      "learning_rate": 9.857978974230952e-05,
      "loss": 2.2942,
      "step": 9216
    },
    {
      "epoch": 0.018249182843571174,
      "grad_norm": 0.16730424761772156,
      "learning_rate": 9.857344951794482e-05,
      "loss": 2.2763,
      "step": 9248
    },
    {
      "epoch": 0.018312328804967613,
      "grad_norm": 0.11629340797662735,
      "learning_rate": 9.856710929358013e-05,
      "loss": 2.3194,
      "step": 9280
    },
    {
      "epoch": 0.018375474766364055,
      "grad_norm": 0.14931993186473846,
      "learning_rate": 9.856076906921545e-05,
      "loss": 2.2907,
      "step": 9312
    },
    {
      "epoch": 0.018438620727760494,
      "grad_norm": 0.14245635271072388,
      "learning_rate": 9.855442884485075e-05,
      "loss": 2.3068,
      "step": 9344
    },
    {
      "epoch": 0.018501766689156932,
      "grad_norm": 0.12499593198299408,
      "learning_rate": 9.854808862048606e-05,
      "loss": 2.2925,
      "step": 9376
    },
    {
      "epoch": 0.018564912650553374,
      "grad_norm": 0.11900224536657333,
      "learning_rate": 9.854174839612136e-05,
      "loss": 2.2877,
      "step": 9408
    },
    {
      "epoch": 0.018628058611949813,
      "grad_norm": 0.11563645303249359,
      "learning_rate": 9.853540817175668e-05,
      "loss": 2.284,
      "step": 9440
    },
    {
      "epoch": 0.018691204573346255,
      "grad_norm": 0.1066991463303566,
      "learning_rate": 9.8529067947392e-05,
      "loss": 2.2858,
      "step": 9472
    },
    {
      "epoch": 0.018754350534742694,
      "grad_norm": 0.1500285118818283,
      "learning_rate": 9.852272772302731e-05,
      "loss": 2.2971,
      "step": 9504
    },
    {
      "epoch": 0.018817496496139133,
      "grad_norm": 0.14508545398712158,
      "learning_rate": 9.851638749866261e-05,
      "loss": 2.2873,
      "step": 9536
    },
    {
      "epoch": 0.018880642457535575,
      "grad_norm": 0.12327894568443298,
      "learning_rate": 9.851004727429792e-05,
      "loss": 2.2817,
      "step": 9568
    },
    {
      "epoch": 0.018943788418932014,
      "grad_norm": 0.1686916947364807,
      "learning_rate": 9.850370704993324e-05,
      "loss": 2.2881,
      "step": 9600
    },
    {
      "epoch": 0.019006934380328456,
      "grad_norm": 0.11902410537004471,
      "learning_rate": 9.849736682556855e-05,
      "loss": 2.2848,
      "step": 9632
    },
    {
      "epoch": 0.019070080341724895,
      "grad_norm": 0.1380203515291214,
      "learning_rate": 9.849102660120385e-05,
      "loss": 2.285,
      "step": 9664
    },
    {
      "epoch": 0.019133226303121333,
      "grad_norm": 0.12518124282360077,
      "learning_rate": 9.848468637683917e-05,
      "loss": 2.2955,
      "step": 9696
    },
    {
      "epoch": 0.019196372264517775,
      "grad_norm": 0.11999049782752991,
      "learning_rate": 9.847834615247448e-05,
      "loss": 2.2709,
      "step": 9728
    },
    {
      "epoch": 0.019259518225914214,
      "grad_norm": 0.1295236349105835,
      "learning_rate": 9.847200592810978e-05,
      "loss": 2.2726,
      "step": 9760
    },
    {
      "epoch": 0.019322664187310653,
      "grad_norm": 0.12159834057092667,
      "learning_rate": 9.846566570374509e-05,
      "loss": 2.2713,
      "step": 9792
    },
    {
      "epoch": 0.019385810148707095,
      "grad_norm": 0.12995825707912445,
      "learning_rate": 9.84593254793804e-05,
      "loss": 2.2846,
      "step": 9824
    },
    {
      "epoch": 0.019448956110103534,
      "grad_norm": 0.13964006304740906,
      "learning_rate": 9.845298525501571e-05,
      "loss": 2.2957,
      "step": 9856
    },
    {
      "epoch": 0.019512102071499976,
      "grad_norm": 0.1186000406742096,
      "learning_rate": 9.844664503065103e-05,
      "loss": 2.2771,
      "step": 9888
    },
    {
      "epoch": 0.019575248032896415,
      "grad_norm": 0.13176684081554413,
      "learning_rate": 9.844030480628633e-05,
      "loss": 2.2811,
      "step": 9920
    },
    {
      "epoch": 0.019638393994292853,
      "grad_norm": 0.1463262289762497,
      "learning_rate": 9.843396458192164e-05,
      "loss": 2.2699,
      "step": 9952
    },
    {
      "epoch": 0.019701539955689296,
      "grad_norm": 0.13183870911598206,
      "learning_rate": 9.842762435755696e-05,
      "loss": 2.2847,
      "step": 9984
    },
    {
      "epoch": 0.019764685917085734,
      "grad_norm": 0.179069384932518,
      "learning_rate": 9.842128413319227e-05,
      "loss": 2.2534,
      "step": 10016
    },
    {
      "epoch": 0.019827831878482177,
      "grad_norm": 0.10940461605787277,
      "learning_rate": 9.841494390882759e-05,
      "loss": 2.2843,
      "step": 10048
    },
    {
      "epoch": 0.019890977839878615,
      "grad_norm": 0.13837820291519165,
      "learning_rate": 9.840860368446289e-05,
      "loss": 2.2703,
      "step": 10080
    },
    {
      "epoch": 0.019954123801275054,
      "grad_norm": 0.16267909109592438,
      "learning_rate": 9.84022634600982e-05,
      "loss": 2.2649,
      "step": 10112
    },
    {
      "epoch": 0.020017269762671496,
      "grad_norm": 0.1511298567056656,
      "learning_rate": 9.83959232357335e-05,
      "loss": 2.2737,
      "step": 10144
    },
    {
      "epoch": 0.020080415724067935,
      "grad_norm": 0.10784769058227539,
      "learning_rate": 9.838958301136882e-05,
      "loss": 2.2954,
      "step": 10176
    },
    {
      "epoch": 0.020143561685464374,
      "grad_norm": 0.11363289505243301,
      "learning_rate": 9.838324278700412e-05,
      "loss": 2.2527,
      "step": 10208
    },
    {
      "epoch": 0.020206707646860816,
      "grad_norm": 0.12205290794372559,
      "learning_rate": 9.837690256263944e-05,
      "loss": 2.2546,
      "step": 10240
    },
    {
      "epoch": 0.020269853608257254,
      "grad_norm": 0.11637623608112335,
      "learning_rate": 9.837056233827475e-05,
      "loss": 2.2757,
      "step": 10272
    },
    {
      "epoch": 0.020332999569653697,
      "grad_norm": 0.12465595453977585,
      "learning_rate": 9.836422211391006e-05,
      "loss": 2.2612,
      "step": 10304
    },
    {
      "epoch": 0.020396145531050135,
      "grad_norm": 0.13645561039447784,
      "learning_rate": 9.835788188954537e-05,
      "loss": 2.2678,
      "step": 10336
    },
    {
      "epoch": 0.020459291492446574,
      "grad_norm": 0.15595458447933197,
      "learning_rate": 9.835154166518068e-05,
      "loss": 2.2642,
      "step": 10368
    },
    {
      "epoch": 0.020522437453843016,
      "grad_norm": 0.11267196387052536,
      "learning_rate": 9.8345201440816e-05,
      "loss": 2.2501,
      "step": 10400
    },
    {
      "epoch": 0.020585583415239455,
      "grad_norm": 0.16104404628276825,
      "learning_rate": 9.833886121645131e-05,
      "loss": 2.2697,
      "step": 10432
    },
    {
      "epoch": 0.020648729376635897,
      "grad_norm": 0.13484841585159302,
      "learning_rate": 9.833252099208661e-05,
      "loss": 2.2534,
      "step": 10464
    },
    {
      "epoch": 0.020711875338032336,
      "grad_norm": 0.15047989785671234,
      "learning_rate": 9.832618076772192e-05,
      "loss": 2.2551,
      "step": 10496
    },
    {
      "epoch": 0.020775021299428775,
      "grad_norm": 0.12250261753797531,
      "learning_rate": 9.831984054335724e-05,
      "loss": 2.2522,
      "step": 10528
    },
    {
      "epoch": 0.020838167260825217,
      "grad_norm": 0.12256615608930588,
      "learning_rate": 9.831350031899254e-05,
      "loss": 2.2388,
      "step": 10560
    },
    {
      "epoch": 0.020901313222221655,
      "grad_norm": 0.13912375271320343,
      "learning_rate": 9.830716009462785e-05,
      "loss": 2.2516,
      "step": 10592
    },
    {
      "epoch": 0.020964459183618094,
      "grad_norm": 0.16619546711444855,
      "learning_rate": 9.830081987026316e-05,
      "loss": 2.2593,
      "step": 10624
    },
    {
      "epoch": 0.021027605145014536,
      "grad_norm": 0.12345771491527557,
      "learning_rate": 9.829447964589847e-05,
      "loss": 2.2534,
      "step": 10656
    },
    {
      "epoch": 0.021090751106410975,
      "grad_norm": 0.14183910191059113,
      "learning_rate": 9.828813942153378e-05,
      "loss": 2.2348,
      "step": 10688
    },
    {
      "epoch": 0.021153897067807417,
      "grad_norm": 0.14614734053611755,
      "learning_rate": 9.82817991971691e-05,
      "loss": 2.2442,
      "step": 10720
    },
    {
      "epoch": 0.021217043029203856,
      "grad_norm": 0.15900371968746185,
      "learning_rate": 9.82754589728044e-05,
      "loss": 2.2624,
      "step": 10752
    },
    {
      "epoch": 0.021280188990600295,
      "grad_norm": 0.12799856066703796,
      "learning_rate": 9.826911874843971e-05,
      "loss": 2.2443,
      "step": 10784
    },
    {
      "epoch": 0.021343334951996737,
      "grad_norm": 0.12659910321235657,
      "learning_rate": 9.826277852407503e-05,
      "loss": 2.2653,
      "step": 10816
    },
    {
      "epoch": 0.021406480913393176,
      "grad_norm": 0.1434067338705063,
      "learning_rate": 9.825643829971034e-05,
      "loss": 2.2754,
      "step": 10848
    },
    {
      "epoch": 0.021469626874789618,
      "grad_norm": 0.1296793818473816,
      "learning_rate": 9.825009807534564e-05,
      "loss": 2.2522,
      "step": 10880
    },
    {
      "epoch": 0.021532772836186057,
      "grad_norm": 0.12483330070972443,
      "learning_rate": 9.824375785098096e-05,
      "loss": 2.2547,
      "step": 10912
    },
    {
      "epoch": 0.021595918797582495,
      "grad_norm": 0.11052534729242325,
      "learning_rate": 9.823741762661627e-05,
      "loss": 2.2392,
      "step": 10944
    },
    {
      "epoch": 0.021659064758978937,
      "grad_norm": 0.1412501484155655,
      "learning_rate": 9.823107740225158e-05,
      "loss": 2.2684,
      "step": 10976
    },
    {
      "epoch": 0.021722210720375376,
      "grad_norm": 0.12963959574699402,
      "learning_rate": 9.822473717788689e-05,
      "loss": 2.24,
      "step": 11008
    },
    {
      "epoch": 0.021785356681771815,
      "grad_norm": 0.12713031470775604,
      "learning_rate": 9.821839695352219e-05,
      "loss": 2.2349,
      "step": 11040
    },
    {
      "epoch": 0.021848502643168257,
      "grad_norm": 0.1355070173740387,
      "learning_rate": 9.82120567291575e-05,
      "loss": 2.2297,
      "step": 11072
    },
    {
      "epoch": 0.021911648604564696,
      "grad_norm": 0.11268522590398788,
      "learning_rate": 9.820571650479282e-05,
      "loss": 2.2379,
      "step": 11104
    },
    {
      "epoch": 0.021974794565961138,
      "grad_norm": 0.11818798631429672,
      "learning_rate": 9.819937628042812e-05,
      "loss": 2.225,
      "step": 11136
    },
    {
      "epoch": 0.022037940527357577,
      "grad_norm": 0.10650172084569931,
      "learning_rate": 9.819303605606344e-05,
      "loss": 2.246,
      "step": 11168
    },
    {
      "epoch": 0.022101086488754015,
      "grad_norm": 0.14348286390304565,
      "learning_rate": 9.818669583169875e-05,
      "loss": 2.2465,
      "step": 11200
    },
    {
      "epoch": 0.022164232450150458,
      "grad_norm": 0.13717545568943024,
      "learning_rate": 9.818035560733406e-05,
      "loss": 2.2242,
      "step": 11232
    },
    {
      "epoch": 0.022227378411546896,
      "grad_norm": 0.12716437876224518,
      "learning_rate": 9.817401538296937e-05,
      "loss": 2.1999,
      "step": 11264
    },
    {
      "epoch": 0.02229052437294334,
      "grad_norm": 0.14523211121559143,
      "learning_rate": 9.816767515860468e-05,
      "loss": 2.2169,
      "step": 11296
    },
    {
      "epoch": 0.022353670334339777,
      "grad_norm": 0.13594989478588104,
      "learning_rate": 9.816133493424e-05,
      "loss": 2.2351,
      "step": 11328
    },
    {
      "epoch": 0.022416816295736216,
      "grad_norm": 0.10901647061109543,
      "learning_rate": 9.815499470987531e-05,
      "loss": 2.2271,
      "step": 11360
    },
    {
      "epoch": 0.022479962257132658,
      "grad_norm": 0.12008926272392273,
      "learning_rate": 9.814865448551061e-05,
      "loss": 2.2311,
      "step": 11392
    },
    {
      "epoch": 0.022543108218529097,
      "grad_norm": 0.11971398442983627,
      "learning_rate": 9.814231426114591e-05,
      "loss": 2.2217,
      "step": 11424
    },
    {
      "epoch": 0.022606254179925536,
      "grad_norm": 0.11262718588113785,
      "learning_rate": 9.813597403678123e-05,
      "loss": 2.2355,
      "step": 11456
    },
    {
      "epoch": 0.022669400141321978,
      "grad_norm": 0.11999446898698807,
      "learning_rate": 9.812963381241654e-05,
      "loss": 2.2373,
      "step": 11488
    },
    {
      "epoch": 0.022732546102718416,
      "grad_norm": 0.11067375540733337,
      "learning_rate": 9.812329358805185e-05,
      "loss": 2.2084,
      "step": 11520
    },
    {
      "epoch": 0.02279569206411486,
      "grad_norm": 0.12921836972236633,
      "learning_rate": 9.811695336368716e-05,
      "loss": 2.2285,
      "step": 11552
    },
    {
      "epoch": 0.022858838025511297,
      "grad_norm": 0.10505735874176025,
      "learning_rate": 9.811061313932247e-05,
      "loss": 2.2215,
      "step": 11584
    },
    {
      "epoch": 0.022921983986907736,
      "grad_norm": 0.10381817072629929,
      "learning_rate": 9.810427291495779e-05,
      "loss": 2.2292,
      "step": 11616
    },
    {
      "epoch": 0.022985129948304178,
      "grad_norm": 0.11980079114437103,
      "learning_rate": 9.80979326905931e-05,
      "loss": 2.2335,
      "step": 11648
    },
    {
      "epoch": 0.023048275909700617,
      "grad_norm": 0.13260555267333984,
      "learning_rate": 9.80915924662284e-05,
      "loss": 2.2082,
      "step": 11680
    },
    {
      "epoch": 0.02311142187109706,
      "grad_norm": 0.11776828020811081,
      "learning_rate": 9.808525224186372e-05,
      "loss": 2.2405,
      "step": 11712
    },
    {
      "epoch": 0.023174567832493498,
      "grad_norm": 0.12389204651117325,
      "learning_rate": 9.807891201749903e-05,
      "loss": 2.2277,
      "step": 11744
    },
    {
      "epoch": 0.023237713793889937,
      "grad_norm": 0.11607711762189865,
      "learning_rate": 9.807257179313434e-05,
      "loss": 2.2196,
      "step": 11776
    },
    {
      "epoch": 0.02330085975528638,
      "grad_norm": 0.1326410174369812,
      "learning_rate": 9.806623156876965e-05,
      "loss": 2.2098,
      "step": 11808
    },
    {
      "epoch": 0.023364005716682817,
      "grad_norm": 0.11321288347244263,
      "learning_rate": 9.805989134440495e-05,
      "loss": 2.2129,
      "step": 11840
    },
    {
      "epoch": 0.023427151678079256,
      "grad_norm": 0.13293352723121643,
      "learning_rate": 9.805355112004026e-05,
      "loss": 2.2336,
      "step": 11872
    },
    {
      "epoch": 0.0234902976394757,
      "grad_norm": 0.11404547095298767,
      "learning_rate": 9.804721089567558e-05,
      "loss": 2.2018,
      "step": 11904
    },
    {
      "epoch": 0.023553443600872137,
      "grad_norm": 0.12156685441732407,
      "learning_rate": 9.804087067131088e-05,
      "loss": 2.213,
      "step": 11936
    },
    {
      "epoch": 0.02361658956226858,
      "grad_norm": 0.12596964836120605,
      "learning_rate": 9.803453044694619e-05,
      "loss": 2.2218,
      "step": 11968
    },
    {
      "epoch": 0.023679735523665018,
      "grad_norm": 0.15181002020835876,
      "learning_rate": 9.80281902225815e-05,
      "loss": 2.202,
      "step": 12000
    },
    {
      "epoch": 0.023742881485061457,
      "grad_norm": 0.1088121235370636,
      "learning_rate": 9.802184999821682e-05,
      "loss": 2.206,
      "step": 12032
    },
    {
      "epoch": 0.0238060274464579,
      "grad_norm": 0.12130860239267349,
      "learning_rate": 9.801550977385213e-05,
      "loss": 2.1916,
      "step": 12064
    },
    {
      "epoch": 0.023869173407854338,
      "grad_norm": 0.10830409079790115,
      "learning_rate": 9.800916954948744e-05,
      "loss": 2.2073,
      "step": 12096
    },
    {
      "epoch": 0.02393231936925078,
      "grad_norm": 0.1181134432554245,
      "learning_rate": 9.800282932512275e-05,
      "loss": 2.2139,
      "step": 12128
    },
    {
      "epoch": 0.02399546533064722,
      "grad_norm": 0.10429253429174423,
      "learning_rate": 9.799648910075806e-05,
      "loss": 2.1931,
      "step": 12160
    },
    {
      "epoch": 0.024058611292043657,
      "grad_norm": 0.12049050629138947,
      "learning_rate": 9.799014887639338e-05,
      "loss": 2.2107,
      "step": 12192
    },
    {
      "epoch": 0.0241217572534401,
      "grad_norm": 0.15965713560581207,
      "learning_rate": 9.798380865202868e-05,
      "loss": 2.1845,
      "step": 12224
    },
    {
      "epoch": 0.024184903214836538,
      "grad_norm": 0.14251761138439178,
      "learning_rate": 9.797746842766398e-05,
      "loss": 2.1937,
      "step": 12256
    },
    {
      "epoch": 0.024248049176232977,
      "grad_norm": 0.14183920621871948,
      "learning_rate": 9.79711282032993e-05,
      "loss": 2.1973,
      "step": 12288
    },
    {
      "epoch": 0.02431119513762942,
      "grad_norm": 0.11488441377878189,
      "learning_rate": 9.796478797893461e-05,
      "loss": 2.2127,
      "step": 12320
    },
    {
      "epoch": 0.024374341099025858,
      "grad_norm": 0.1270715445280075,
      "learning_rate": 9.795844775456991e-05,
      "loss": 2.2019,
      "step": 12352
    },
    {
      "epoch": 0.0244374870604223,
      "grad_norm": 0.11428068578243256,
      "learning_rate": 9.795210753020523e-05,
      "loss": 2.1863,
      "step": 12384
    },
    {
      "epoch": 0.02450063302181874,
      "grad_norm": 0.12566788494586945,
      "learning_rate": 9.794576730584054e-05,
      "loss": 2.1871,
      "step": 12416
    },
    {
      "epoch": 0.024563778983215177,
      "grad_norm": 0.12264042347669601,
      "learning_rate": 9.793942708147586e-05,
      "loss": 2.1987,
      "step": 12448
    },
    {
      "epoch": 0.02462692494461162,
      "grad_norm": 0.11069290339946747,
      "learning_rate": 9.793308685711116e-05,
      "loss": 2.1985,
      "step": 12480
    },
    {
      "epoch": 0.024690070906008058,
      "grad_norm": 0.11952099204063416,
      "learning_rate": 9.792674663274647e-05,
      "loss": 2.1909,
      "step": 12512
    },
    {
      "epoch": 0.0247532168674045,
      "grad_norm": 0.14786210656166077,
      "learning_rate": 9.792040640838179e-05,
      "loss": 2.1795,
      "step": 12544
    },
    {
      "epoch": 0.02481636282880094,
      "grad_norm": 0.13370029628276825,
      "learning_rate": 9.79140661840171e-05,
      "loss": 2.1859,
      "step": 12576
    },
    {
      "epoch": 0.024879508790197378,
      "grad_norm": 0.14006869494915009,
      "learning_rate": 9.79077259596524e-05,
      "loss": 2.1762,
      "step": 12608
    },
    {
      "epoch": 0.02494265475159382,
      "grad_norm": 0.12157852202653885,
      "learning_rate": 9.790138573528772e-05,
      "loss": 2.1924,
      "step": 12640
    },
    {
      "epoch": 0.02500580071299026,
      "grad_norm": 0.11815094202756882,
      "learning_rate": 9.789504551092302e-05,
      "loss": 2.179,
      "step": 12672
    },
    {
      "epoch": 0.025068946674386697,
      "grad_norm": 0.14207015931606293,
      "learning_rate": 9.788870528655833e-05,
      "loss": 2.1935,
      "step": 12704
    },
    {
      "epoch": 0.02513209263578314,
      "grad_norm": 0.12679685652256012,
      "learning_rate": 9.788236506219365e-05,
      "loss": 2.1804,
      "step": 12736
    },
    {
      "epoch": 0.02519523859717958,
      "grad_norm": 0.13686706125736237,
      "learning_rate": 9.787602483782895e-05,
      "loss": 2.1689,
      "step": 12768
    },
    {
      "epoch": 0.02525838455857602,
      "grad_norm": 0.12256750464439392,
      "learning_rate": 9.786968461346426e-05,
      "loss": 2.1587,
      "step": 12800
    },
    {
      "epoch": 0.02532153051997246,
      "grad_norm": 0.15027447044849396,
      "learning_rate": 9.786334438909958e-05,
      "loss": 2.1729,
      "step": 12832
    },
    {
      "epoch": 0.025384676481368898,
      "grad_norm": 0.132175475358963,
      "learning_rate": 9.785700416473489e-05,
      "loss": 2.1735,
      "step": 12864
    },
    {
      "epoch": 0.02544782244276534,
      "grad_norm": 0.12547612190246582,
      "learning_rate": 9.785066394037019e-05,
      "loss": 2.1533,
      "step": 12896
    },
    {
      "epoch": 0.02551096840416178,
      "grad_norm": 0.13197678327560425,
      "learning_rate": 9.78443237160055e-05,
      "loss": 2.1683,
      "step": 12928
    },
    {
      "epoch": 0.02557411436555822,
      "grad_norm": 0.11401775479316711,
      "learning_rate": 9.783798349164082e-05,
      "loss": 2.1549,
      "step": 12960
    },
    {
      "epoch": 0.02563726032695466,
      "grad_norm": 0.1391398012638092,
      "learning_rate": 9.783164326727614e-05,
      "loss": 2.1663,
      "step": 12992
    },
    {
      "epoch": 0.0257004062883511,
      "grad_norm": 0.13919059932231903,
      "learning_rate": 9.782530304291144e-05,
      "loss": 2.1685,
      "step": 13024
    },
    {
      "epoch": 0.02576355224974754,
      "grad_norm": 0.12351135164499283,
      "learning_rate": 9.781896281854675e-05,
      "loss": 2.1502,
      "step": 13056
    },
    {
      "epoch": 0.02582669821114398,
      "grad_norm": 0.14834736287593842,
      "learning_rate": 9.781262259418205e-05,
      "loss": 2.1729,
      "step": 13088
    },
    {
      "epoch": 0.025889844172540418,
      "grad_norm": 0.128239244222641,
      "learning_rate": 9.780628236981737e-05,
      "loss": 2.1727,
      "step": 13120
    },
    {
      "epoch": 0.02595299013393686,
      "grad_norm": 0.11607243865728378,
      "learning_rate": 9.779994214545267e-05,
      "loss": 2.1596,
      "step": 13152
    },
    {
      "epoch": 0.0260161360953333,
      "grad_norm": 0.13490678369998932,
      "learning_rate": 9.779360192108798e-05,
      "loss": 2.1572,
      "step": 13184
    },
    {
      "epoch": 0.02607928205672974,
      "grad_norm": 0.12663912773132324,
      "learning_rate": 9.77872616967233e-05,
      "loss": 2.1667,
      "step": 13216
    },
    {
      "epoch": 0.02614242801812618,
      "grad_norm": 0.12302709370851517,
      "learning_rate": 9.778092147235861e-05,
      "loss": 2.1672,
      "step": 13248
    },
    {
      "epoch": 0.02620557397952262,
      "grad_norm": 0.11665955930948257,
      "learning_rate": 9.777458124799391e-05,
      "loss": 2.1497,
      "step": 13280
    },
    {
      "epoch": 0.02626871994091906,
      "grad_norm": 0.1355448067188263,
      "learning_rate": 9.776824102362923e-05,
      "loss": 2.1525,
      "step": 13312
    },
    {
      "epoch": 0.0263318659023155,
      "grad_norm": 0.13190394639968872,
      "learning_rate": 9.776190079926454e-05,
      "loss": 2.1861,
      "step": 13344
    },
    {
      "epoch": 0.026395011863711938,
      "grad_norm": 0.11398578435182571,
      "learning_rate": 9.775556057489986e-05,
      "loss": 2.1596,
      "step": 13376
    },
    {
      "epoch": 0.02645815782510838,
      "grad_norm": 0.1408715397119522,
      "learning_rate": 9.774922035053517e-05,
      "loss": 2.168,
      "step": 13408
    },
    {
      "epoch": 0.02652130378650482,
      "grad_norm": 0.11405431479215622,
      "learning_rate": 9.774288012617047e-05,
      "loss": 2.1462,
      "step": 13440
    },
    {
      "epoch": 0.02658444974790126,
      "grad_norm": 0.11981717497110367,
      "learning_rate": 9.773653990180579e-05,
      "loss": 2.1588,
      "step": 13472
    },
    {
      "epoch": 0.0266475957092977,
      "grad_norm": 0.12904761731624603,
      "learning_rate": 9.773019967744109e-05,
      "loss": 2.1454,
      "step": 13504
    },
    {
      "epoch": 0.02671074167069414,
      "grad_norm": 0.1340661495923996,
      "learning_rate": 9.77238594530764e-05,
      "loss": 2.1666,
      "step": 13536
    },
    {
      "epoch": 0.02677388763209058,
      "grad_norm": 0.12820249795913696,
      "learning_rate": 9.77175192287117e-05,
      "loss": 2.1488,
      "step": 13568
    },
    {
      "epoch": 0.02683703359348702,
      "grad_norm": 0.1250142604112625,
      "learning_rate": 9.771117900434702e-05,
      "loss": 2.144,
      "step": 13600
    },
    {
      "epoch": 0.026900179554883462,
      "grad_norm": 0.1425981968641281,
      "learning_rate": 9.770483877998233e-05,
      "loss": 2.1547,
      "step": 13632
    },
    {
      "epoch": 0.0269633255162799,
      "grad_norm": 0.15214186906814575,
      "learning_rate": 9.769849855561765e-05,
      "loss": 2.1559,
      "step": 13664
    },
    {
      "epoch": 0.02702647147767634,
      "grad_norm": 0.13166294991970062,
      "learning_rate": 9.769215833125295e-05,
      "loss": 2.141,
      "step": 13696
    },
    {
      "epoch": 0.02708961743907278,
      "grad_norm": 0.12783321738243103,
      "learning_rate": 9.768581810688826e-05,
      "loss": 2.1352,
      "step": 13728
    },
    {
      "epoch": 0.02715276340046922,
      "grad_norm": 0.1286151260137558,
      "learning_rate": 9.767947788252358e-05,
      "loss": 2.1382,
      "step": 13760
    },
    {
      "epoch": 0.02721590936186566,
      "grad_norm": 0.1499750167131424,
      "learning_rate": 9.767313765815889e-05,
      "loss": 2.1579,
      "step": 13792
    },
    {
      "epoch": 0.0272790553232621,
      "grad_norm": 0.11294298619031906,
      "learning_rate": 9.766679743379419e-05,
      "loss": 2.1374,
      "step": 13824
    },
    {
      "epoch": 0.02734220128465854,
      "grad_norm": 0.13401901721954346,
      "learning_rate": 9.76604572094295e-05,
      "loss": 2.1487,
      "step": 13856
    },
    {
      "epoch": 0.027405347246054982,
      "grad_norm": 0.1361786127090454,
      "learning_rate": 9.765411698506482e-05,
      "loss": 2.1131,
      "step": 13888
    },
    {
      "epoch": 0.02746849320745142,
      "grad_norm": 0.13850685954093933,
      "learning_rate": 9.764777676070012e-05,
      "loss": 2.1621,
      "step": 13920
    },
    {
      "epoch": 0.02753163916884786,
      "grad_norm": 0.12793563306331635,
      "learning_rate": 9.764143653633542e-05,
      "loss": 2.1394,
      "step": 13952
    },
    {
      "epoch": 0.0275947851302443,
      "grad_norm": 0.13138705492019653,
      "learning_rate": 9.763509631197074e-05,
      "loss": 2.1425,
      "step": 13984
    },
    {
      "epoch": 0.02765793109164074,
      "grad_norm": 0.11367518454790115,
      "learning_rate": 9.762875608760605e-05,
      "loss": 2.1313,
      "step": 14016
    },
    {
      "epoch": 0.027721077053037183,
      "grad_norm": 0.12217991054058075,
      "learning_rate": 9.762241586324137e-05,
      "loss": 2.1252,
      "step": 14048
    },
    {
      "epoch": 0.02778422301443362,
      "grad_norm": 0.12337450683116913,
      "learning_rate": 9.761607563887668e-05,
      "loss": 2.1275,
      "step": 14080
    },
    {
      "epoch": 0.02784736897583006,
      "grad_norm": 0.13017290830612183,
      "learning_rate": 9.760973541451198e-05,
      "loss": 2.1174,
      "step": 14112
    },
    {
      "epoch": 0.027910514937226502,
      "grad_norm": 0.12573756277561188,
      "learning_rate": 9.76033951901473e-05,
      "loss": 2.1314,
      "step": 14144
    },
    {
      "epoch": 0.02797366089862294,
      "grad_norm": 0.13559311628341675,
      "learning_rate": 9.759705496578261e-05,
      "loss": 2.1253,
      "step": 14176
    },
    {
      "epoch": 0.02803680686001938,
      "grad_norm": 0.1147247925400734,
      "learning_rate": 9.759071474141793e-05,
      "loss": 2.1322,
      "step": 14208
    },
    {
      "epoch": 0.028099952821415822,
      "grad_norm": 0.14051298797130585,
      "learning_rate": 9.758437451705323e-05,
      "loss": 2.1349,
      "step": 14240
    },
    {
      "epoch": 0.02816309878281226,
      "grad_norm": 0.12124649435281754,
      "learning_rate": 9.757803429268854e-05,
      "loss": 2.1138,
      "step": 14272
    },
    {
      "epoch": 0.028226244744208703,
      "grad_norm": 0.13273552060127258,
      "learning_rate": 9.757169406832384e-05,
      "loss": 2.1174,
      "step": 14304
    },
    {
      "epoch": 0.02828939070560514,
      "grad_norm": 0.13307848572731018,
      "learning_rate": 9.756535384395916e-05,
      "loss": 2.1194,
      "step": 14336
    },
    {
      "epoch": 0.02835253666700158,
      "grad_norm": 0.12077753245830536,
      "learning_rate": 9.755901361959446e-05,
      "loss": 2.1104,
      "step": 14368
    },
    {
      "epoch": 0.028415682628398022,
      "grad_norm": 0.11729800701141357,
      "learning_rate": 9.755267339522977e-05,
      "loss": 2.1161,
      "step": 14400
    },
    {
      "epoch": 0.02847882858979446,
      "grad_norm": 0.1477431207895279,
      "learning_rate": 9.754633317086509e-05,
      "loss": 2.1252,
      "step": 14432
    },
    {
      "epoch": 0.028541974551190903,
      "grad_norm": 0.12713754177093506,
      "learning_rate": 9.75399929465004e-05,
      "loss": 2.1255,
      "step": 14464
    },
    {
      "epoch": 0.028605120512587342,
      "grad_norm": 0.11540602892637253,
      "learning_rate": 9.75336527221357e-05,
      "loss": 2.1277,
      "step": 14496
    },
    {
      "epoch": 0.02866826647398378,
      "grad_norm": 0.14172641932964325,
      "learning_rate": 9.752731249777102e-05,
      "loss": 2.1122,
      "step": 14528
    },
    {
      "epoch": 0.028731412435380223,
      "grad_norm": 0.10793290287256241,
      "learning_rate": 9.752097227340633e-05,
      "loss": 2.1159,
      "step": 14560
    },
    {
      "epoch": 0.02879455839677666,
      "grad_norm": 0.11252595484256744,
      "learning_rate": 9.751463204904165e-05,
      "loss": 2.1054,
      "step": 14592
    },
    {
      "epoch": 0.0288577043581731,
      "grad_norm": 0.1248583123087883,
      "learning_rate": 9.750829182467695e-05,
      "loss": 2.1166,
      "step": 14624
    },
    {
      "epoch": 0.028920850319569542,
      "grad_norm": 0.11714232712984085,
      "learning_rate": 9.750195160031226e-05,
      "loss": 2.0992,
      "step": 14656
    },
    {
      "epoch": 0.02898399628096598,
      "grad_norm": 0.11430927366018295,
      "learning_rate": 9.749561137594758e-05,
      "loss": 2.0934,
      "step": 14688
    },
    {
      "epoch": 0.029047142242362423,
      "grad_norm": 0.11717204004526138,
      "learning_rate": 9.748927115158288e-05,
      "loss": 2.1056,
      "step": 14720
    },
    {
      "epoch": 0.029110288203758862,
      "grad_norm": 0.1542053520679474,
      "learning_rate": 9.748293092721819e-05,
      "loss": 2.1077,
      "step": 14752
    },
    {
      "epoch": 0.0291734341651553,
      "grad_norm": 0.1308835744857788,
      "learning_rate": 9.747659070285349e-05,
      "loss": 2.1103,
      "step": 14784
    },
    {
      "epoch": 0.029236580126551743,
      "grad_norm": 0.10842951387166977,
      "learning_rate": 9.747025047848881e-05,
      "loss": 2.109,
      "step": 14816
    },
    {
      "epoch": 0.02929972608794818,
      "grad_norm": 0.12598951160907745,
      "learning_rate": 9.746391025412412e-05,
      "loss": 2.1072,
      "step": 14848
    },
    {
      "epoch": 0.029362872049344624,
      "grad_norm": 0.10338089615106583,
      "learning_rate": 9.745757002975944e-05,
      "loss": 2.1041,
      "step": 14880
    },
    {
      "epoch": 0.029426018010741063,
      "grad_norm": 0.11851800978183746,
      "learning_rate": 9.745122980539474e-05,
      "loss": 2.0945,
      "step": 14912
    },
    {
      "epoch": 0.0294891639721375,
      "grad_norm": 0.11729994416236877,
      "learning_rate": 9.744488958103005e-05,
      "loss": 2.112,
      "step": 14944
    },
    {
      "epoch": 0.029552309933533943,
      "grad_norm": 0.13360081613063812,
      "learning_rate": 9.743854935666537e-05,
      "loss": 2.1067,
      "step": 14976
    },
    {
      "epoch": 0.029615455894930382,
      "grad_norm": 0.13326188921928406,
      "learning_rate": 9.743220913230068e-05,
      "loss": 2.102,
      "step": 15008
    },
    {
      "epoch": 0.02967860185632682,
      "grad_norm": 0.11109422147274017,
      "learning_rate": 9.742586890793598e-05,
      "loss": 2.0932,
      "step": 15040
    },
    {
      "epoch": 0.029741747817723263,
      "grad_norm": 0.14367780089378357,
      "learning_rate": 9.74195286835713e-05,
      "loss": 2.1007,
      "step": 15072
    },
    {
      "epoch": 0.029804893779119702,
      "grad_norm": 0.12622703611850739,
      "learning_rate": 9.741318845920661e-05,
      "loss": 2.1006,
      "step": 15104
    },
    {
      "epoch": 0.029868039740516144,
      "grad_norm": 0.1098259836435318,
      "learning_rate": 9.740684823484191e-05,
      "loss": 2.1098,
      "step": 15136
    },
    {
      "epoch": 0.029931185701912583,
      "grad_norm": 0.12503789365291595,
      "learning_rate": 9.740050801047723e-05,
      "loss": 2.1026,
      "step": 15168
    },
    {
      "epoch": 0.02999433166330902,
      "grad_norm": 0.12734921276569366,
      "learning_rate": 9.739416778611253e-05,
      "loss": 2.0925,
      "step": 15200
    },
    {
      "epoch": 0.030057477624705464,
      "grad_norm": 0.12283790111541748,
      "learning_rate": 9.738782756174784e-05,
      "loss": 2.0946,
      "step": 15232
    },
    {
      "epoch": 0.030120623586101902,
      "grad_norm": 0.12231511622667313,
      "learning_rate": 9.738148733738316e-05,
      "loss": 2.0867,
      "step": 15264
    },
    {
      "epoch": 0.030183769547498344,
      "grad_norm": 0.1260044127702713,
      "learning_rate": 9.737514711301846e-05,
      "loss": 2.0838,
      "step": 15296
    },
    {
      "epoch": 0.030246915508894783,
      "grad_norm": 0.1352124810218811,
      "learning_rate": 9.736880688865377e-05,
      "loss": 2.1022,
      "step": 15328
    },
    {
      "epoch": 0.030310061470291222,
      "grad_norm": 0.13228322565555573,
      "learning_rate": 9.736246666428909e-05,
      "loss": 2.0997,
      "step": 15360
    },
    {
      "epoch": 0.030373207431687664,
      "grad_norm": 0.11998327076435089,
      "learning_rate": 9.73561264399244e-05,
      "loss": 2.0854,
      "step": 15392
    },
    {
      "epoch": 0.030436353393084103,
      "grad_norm": 0.12817572057247162,
      "learning_rate": 9.734978621555972e-05,
      "loss": 2.0972,
      "step": 15424
    },
    {
      "epoch": 0.03049949935448054,
      "grad_norm": 0.1324385553598404,
      "learning_rate": 9.734344599119502e-05,
      "loss": 2.0763,
      "step": 15456
    },
    {
      "epoch": 0.030562645315876984,
      "grad_norm": 0.13350187242031097,
      "learning_rate": 9.733710576683033e-05,
      "loss": 2.1011,
      "step": 15488
    },
    {
      "epoch": 0.030625791277273422,
      "grad_norm": 0.1624910533428192,
      "learning_rate": 9.733076554246565e-05,
      "loss": 2.0783,
      "step": 15520
    },
    {
      "epoch": 0.030688937238669865,
      "grad_norm": 0.11206740140914917,
      "learning_rate": 9.732442531810095e-05,
      "loss": 2.0922,
      "step": 15552
    },
    {
      "epoch": 0.030752083200066303,
      "grad_norm": 0.14127673208713531,
      "learning_rate": 9.731808509373625e-05,
      "loss": 2.0792,
      "step": 15584
    },
    {
      "epoch": 0.030815229161462742,
      "grad_norm": 0.13760198652744293,
      "learning_rate": 9.731174486937156e-05,
      "loss": 2.0921,
      "step": 15616
    },
    {
      "epoch": 0.030878375122859184,
      "grad_norm": 0.14275386929512024,
      "learning_rate": 9.730540464500688e-05,
      "loss": 2.0867,
      "step": 15648
    },
    {
      "epoch": 0.030941521084255623,
      "grad_norm": 0.12568217515945435,
      "learning_rate": 9.729906442064219e-05,
      "loss": 2.1021,
      "step": 15680
    },
    {
      "epoch": 0.031004667045652065,
      "grad_norm": 0.10878398269414902,
      "learning_rate": 9.72927241962775e-05,
      "loss": 2.0805,
      "step": 15712
    },
    {
      "epoch": 0.031067813007048504,
      "grad_norm": 0.11523103713989258,
      "learning_rate": 9.728638397191281e-05,
      "loss": 2.0612,
      "step": 15744
    },
    {
      "epoch": 0.031130958968444943,
      "grad_norm": 0.12170586735010147,
      "learning_rate": 9.728004374754812e-05,
      "loss": 2.0731,
      "step": 15776
    },
    {
      "epoch": 0.031194104929841385,
      "grad_norm": 0.12763361632823944,
      "learning_rate": 9.727370352318344e-05,
      "loss": 2.0939,
      "step": 15808
    },
    {
      "epoch": 0.03125725089123783,
      "grad_norm": 0.13741229474544525,
      "learning_rate": 9.726736329881874e-05,
      "loss": 2.0912,
      "step": 15840
    },
    {
      "epoch": 0.03132039685263426,
      "grad_norm": 0.12540291249752045,
      "learning_rate": 9.726102307445405e-05,
      "loss": 2.0721,
      "step": 15872
    },
    {
      "epoch": 0.031383542814030704,
      "grad_norm": 0.11759001761674881,
      "learning_rate": 9.725468285008937e-05,
      "loss": 2.0875,
      "step": 15904
    },
    {
      "epoch": 0.031446688775427147,
      "grad_norm": 0.1085490733385086,
      "learning_rate": 9.724834262572468e-05,
      "loss": 2.0747,
      "step": 15936
    },
    {
      "epoch": 0.03150983473682358,
      "grad_norm": 0.11623985320329666,
      "learning_rate": 9.724200240135998e-05,
      "loss": 2.078,
      "step": 15968
    },
    {
      "epoch": 0.031572980698220024,
      "grad_norm": 0.11528050899505615,
      "learning_rate": 9.723566217699528e-05,
      "loss": 2.0874,
      "step": 16000
    },
    {
      "epoch": 0.031636126659616466,
      "grad_norm": 0.12489989399909973,
      "learning_rate": 9.72293219526306e-05,
      "loss": 2.0722,
      "step": 16032
    },
    {
      "epoch": 0.0316992726210129,
      "grad_norm": 0.11752578616142273,
      "learning_rate": 9.722298172826591e-05,
      "loss": 2.0719,
      "step": 16064
    },
    {
      "epoch": 0.031762418582409344,
      "grad_norm": 0.14901821315288544,
      "learning_rate": 9.721664150390123e-05,
      "loss": 2.0599,
      "step": 16096
    },
    {
      "epoch": 0.031825564543805786,
      "grad_norm": 0.11683712154626846,
      "learning_rate": 9.721030127953653e-05,
      "loss": 2.0638,
      "step": 16128
    },
    {
      "epoch": 0.03188871050520222,
      "grad_norm": 0.13710802793502808,
      "learning_rate": 9.720396105517184e-05,
      "loss": 2.0547,
      "step": 16160
    },
    {
      "epoch": 0.03195185646659866,
      "grad_norm": 0.12909361720085144,
      "learning_rate": 9.719762083080716e-05,
      "loss": 2.0644,
      "step": 16192
    },
    {
      "epoch": 0.032015002427995105,
      "grad_norm": 0.12128318846225739,
      "learning_rate": 9.719128060644247e-05,
      "loss": 2.0633,
      "step": 16224
    },
    {
      "epoch": 0.03207814838939155,
      "grad_norm": 0.1090121939778328,
      "learning_rate": 9.718494038207777e-05,
      "loss": 2.0816,
      "step": 16256
    },
    {
      "epoch": 0.03214129435078798,
      "grad_norm": 0.11851469427347183,
      "learning_rate": 9.717860015771309e-05,
      "loss": 2.0671,
      "step": 16288
    },
    {
      "epoch": 0.032204440312184425,
      "grad_norm": 0.11412264406681061,
      "learning_rate": 9.71722599333484e-05,
      "loss": 2.0629,
      "step": 16320
    },
    {
      "epoch": 0.03226758627358087,
      "grad_norm": 0.11426562815904617,
      "learning_rate": 9.716591970898372e-05,
      "loss": 2.0619,
      "step": 16352
    },
    {
      "epoch": 0.0323307322349773,
      "grad_norm": 0.1323927342891693,
      "learning_rate": 9.715957948461902e-05,
      "loss": 2.0627,
      "step": 16384
    },
    {
      "epoch": 0.032393878196373745,
      "grad_norm": 0.12511026859283447,
      "learning_rate": 9.715323926025432e-05,
      "loss": 2.0607,
      "step": 16416
    },
    {
      "epoch": 0.03245702415777019,
      "grad_norm": 0.14868119359016418,
      "learning_rate": 9.714689903588963e-05,
      "loss": 2.0559,
      "step": 16448
    },
    {
      "epoch": 0.03252017011916662,
      "grad_norm": 0.14075885713100433,
      "learning_rate": 9.714055881152495e-05,
      "loss": 2.0768,
      "step": 16480
    },
    {
      "epoch": 0.032583316080563064,
      "grad_norm": 0.13771525025367737,
      "learning_rate": 9.713421858716025e-05,
      "loss": 2.0716,
      "step": 16512
    },
    {
      "epoch": 0.032646462041959506,
      "grad_norm": 0.11929049342870712,
      "learning_rate": 9.712787836279556e-05,
      "loss": 2.0606,
      "step": 16544
    },
    {
      "epoch": 0.03270960800335594,
      "grad_norm": 0.11700630187988281,
      "learning_rate": 9.712153813843088e-05,
      "loss": 2.0737,
      "step": 16576
    },
    {
      "epoch": 0.032772753964752384,
      "grad_norm": 0.12586750090122223,
      "learning_rate": 9.711519791406619e-05,
      "loss": 2.0771,
      "step": 16608
    },
    {
      "epoch": 0.032835899926148826,
      "grad_norm": 0.10553145408630371,
      "learning_rate": 9.71088576897015e-05,
      "loss": 2.0423,
      "step": 16640
    },
    {
      "epoch": 0.03289904588754527,
      "grad_norm": 0.1436128169298172,
      "learning_rate": 9.710251746533681e-05,
      "loss": 2.0559,
      "step": 16672
    },
    {
      "epoch": 0.0329621918489417,
      "grad_norm": 0.1168614998459816,
      "learning_rate": 9.709617724097212e-05,
      "loss": 2.0443,
      "step": 16704
    },
    {
      "epoch": 0.033025337810338146,
      "grad_norm": 0.12289905548095703,
      "learning_rate": 9.708983701660744e-05,
      "loss": 2.056,
      "step": 16736
    },
    {
      "epoch": 0.03308848377173459,
      "grad_norm": 0.1273569017648697,
      "learning_rate": 9.708349679224274e-05,
      "loss": 2.0509,
      "step": 16768
    },
    {
      "epoch": 0.03315162973313102,
      "grad_norm": 0.11731128394603729,
      "learning_rate": 9.707715656787805e-05,
      "loss": 2.0505,
      "step": 16800
    },
    {
      "epoch": 0.033214775694527465,
      "grad_norm": 0.11746280640363693,
      "learning_rate": 9.707081634351335e-05,
      "loss": 2.0482,
      "step": 16832
    },
    {
      "epoch": 0.03327792165592391,
      "grad_norm": 0.12205459177494049,
      "learning_rate": 9.706447611914867e-05,
      "loss": 2.0475,
      "step": 16864
    },
    {
      "epoch": 0.03334106761732034,
      "grad_norm": 0.12475898116827011,
      "learning_rate": 9.705813589478398e-05,
      "loss": 2.0579,
      "step": 16896
    },
    {
      "epoch": 0.033404213578716785,
      "grad_norm": 0.11365223675966263,
      "learning_rate": 9.705179567041928e-05,
      "loss": 2.056,
      "step": 16928
    },
    {
      "epoch": 0.03346735954011323,
      "grad_norm": 0.11358797550201416,
      "learning_rate": 9.70454554460546e-05,
      "loss": 2.0318,
      "step": 16960
    },
    {
      "epoch": 0.03353050550150966,
      "grad_norm": 0.1325814425945282,
      "learning_rate": 9.703911522168991e-05,
      "loss": 2.0621,
      "step": 16992
    },
    {
      "epoch": 0.033593651462906104,
      "grad_norm": 0.1368623822927475,
      "learning_rate": 9.703277499732523e-05,
      "loss": 2.0488,
      "step": 17024
    },
    {
      "epoch": 0.03365679742430255,
      "grad_norm": 0.1259676218032837,
      "learning_rate": 9.702643477296053e-05,
      "loss": 2.0363,
      "step": 17056
    },
    {
      "epoch": 0.03371994338569899,
      "grad_norm": 0.11567375808954239,
      "learning_rate": 9.702009454859584e-05,
      "loss": 2.0385,
      "step": 17088
    },
    {
      "epoch": 0.033783089347095424,
      "grad_norm": 0.12076389789581299,
      "learning_rate": 9.701375432423116e-05,
      "loss": 2.0652,
      "step": 17120
    },
    {
      "epoch": 0.033846235308491866,
      "grad_norm": 0.11648164689540863,
      "learning_rate": 9.700741409986647e-05,
      "loss": 2.0465,
      "step": 17152
    },
    {
      "epoch": 0.03390938126988831,
      "grad_norm": 0.12384187430143356,
      "learning_rate": 9.700107387550177e-05,
      "loss": 2.0479,
      "step": 17184
    },
    {
      "epoch": 0.033972527231284744,
      "grad_norm": 0.11063065379858017,
      "learning_rate": 9.699473365113709e-05,
      "loss": 2.0426,
      "step": 17216
    },
    {
      "epoch": 0.034035673192681186,
      "grad_norm": 0.11873957514762878,
      "learning_rate": 9.698839342677239e-05,
      "loss": 2.0488,
      "step": 17248
    },
    {
      "epoch": 0.03409881915407763,
      "grad_norm": 0.12810015678405762,
      "learning_rate": 9.69820532024077e-05,
      "loss": 2.0444,
      "step": 17280
    },
    {
      "epoch": 0.03416196511547406,
      "grad_norm": 0.11562523245811462,
      "learning_rate": 9.697571297804302e-05,
      "loss": 2.0455,
      "step": 17312
    },
    {
      "epoch": 0.034225111076870506,
      "grad_norm": 0.12042046338319778,
      "learning_rate": 9.696937275367832e-05,
      "loss": 2.0547,
      "step": 17344
    },
    {
      "epoch": 0.03428825703826695,
      "grad_norm": 0.10553345084190369,
      "learning_rate": 9.696303252931363e-05,
      "loss": 2.0426,
      "step": 17376
    },
    {
      "epoch": 0.03435140299966338,
      "grad_norm": 0.12902826070785522,
      "learning_rate": 9.695669230494895e-05,
      "loss": 2.0638,
      "step": 17408
    },
    {
      "epoch": 0.034414548961059825,
      "grad_norm": 0.11707167327404022,
      "learning_rate": 9.695035208058426e-05,
      "loss": 2.0423,
      "step": 17440
    },
    {
      "epoch": 0.03447769492245627,
      "grad_norm": 0.13379773497581482,
      "learning_rate": 9.694401185621956e-05,
      "loss": 2.0388,
      "step": 17472
    },
    {
      "epoch": 0.03454084088385271,
      "grad_norm": 0.1197654977440834,
      "learning_rate": 9.693767163185488e-05,
      "loss": 2.028,
      "step": 17504
    },
    {
      "epoch": 0.034603986845249145,
      "grad_norm": 0.11672238260507584,
      "learning_rate": 9.693133140749019e-05,
      "loss": 2.023,
      "step": 17536
    },
    {
      "epoch": 0.03466713280664559,
      "grad_norm": 0.11840718239545822,
      "learning_rate": 9.692499118312551e-05,
      "loss": 2.0377,
      "step": 17568
    },
    {
      "epoch": 0.03473027876804203,
      "grad_norm": 0.1268356293439865,
      "learning_rate": 9.691865095876081e-05,
      "loss": 2.0193,
      "step": 17600
    },
    {
      "epoch": 0.034793424729438464,
      "grad_norm": 0.1247047707438469,
      "learning_rate": 9.691231073439612e-05,
      "loss": 2.0511,
      "step": 17632
    },
    {
      "epoch": 0.03485657069083491,
      "grad_norm": 0.13573023676872253,
      "learning_rate": 9.690597051003142e-05,
      "loss": 2.0541,
      "step": 17664
    },
    {
      "epoch": 0.03491971665223135,
      "grad_norm": 0.12114041298627853,
      "learning_rate": 9.689963028566674e-05,
      "loss": 2.0323,
      "step": 17696
    },
    {
      "epoch": 0.034982862613627784,
      "grad_norm": 0.12819577753543854,
      "learning_rate": 9.689329006130204e-05,
      "loss": 2.0265,
      "step": 17728
    },
    {
      "epoch": 0.035046008575024226,
      "grad_norm": 0.11567926406860352,
      "learning_rate": 9.688694983693735e-05,
      "loss": 2.0435,
      "step": 17760
    },
    {
      "epoch": 0.03510915453642067,
      "grad_norm": 0.13010470569133759,
      "learning_rate": 9.688060961257267e-05,
      "loss": 2.0298,
      "step": 17792
    },
    {
      "epoch": 0.035172300497817104,
      "grad_norm": 0.14122366905212402,
      "learning_rate": 9.687426938820798e-05,
      "loss": 2.0302,
      "step": 17824
    },
    {
      "epoch": 0.035235446459213546,
      "grad_norm": 0.11040329188108444,
      "learning_rate": 9.686792916384328e-05,
      "loss": 2.0237,
      "step": 17856
    },
    {
      "epoch": 0.03529859242060999,
      "grad_norm": 0.1371491402387619,
      "learning_rate": 9.68615889394786e-05,
      "loss": 2.0594,
      "step": 17888
    },
    {
      "epoch": 0.03536173838200643,
      "grad_norm": 0.11435120552778244,
      "learning_rate": 9.685524871511391e-05,
      "loss": 2.0307,
      "step": 17920
    },
    {
      "epoch": 0.035424884343402865,
      "grad_norm": 0.11154137551784515,
      "learning_rate": 9.684890849074923e-05,
      "loss": 2.03,
      "step": 17952
    },
    {
      "epoch": 0.03548803030479931,
      "grad_norm": 0.13120387494564056,
      "learning_rate": 9.684256826638454e-05,
      "loss": 2.0287,
      "step": 17984
    },
    {
      "epoch": 0.03555117626619575,
      "grad_norm": 0.11339680105447769,
      "learning_rate": 9.683622804201984e-05,
      "loss": 2.0382,
      "step": 18016
    },
    {
      "epoch": 0.035614322227592185,
      "grad_norm": 0.10982314497232437,
      "learning_rate": 9.682988781765516e-05,
      "loss": 2.0297,
      "step": 18048
    },
    {
      "epoch": 0.03567746818898863,
      "grad_norm": 0.12217368185520172,
      "learning_rate": 9.682354759329046e-05,
      "loss": 2.0192,
      "step": 18080
    },
    {
      "epoch": 0.03574061415038507,
      "grad_norm": 0.11636436730623245,
      "learning_rate": 9.681720736892577e-05,
      "loss": 2.0114,
      "step": 18112
    },
    {
      "epoch": 0.035803760111781505,
      "grad_norm": 0.11972784250974655,
      "learning_rate": 9.681086714456107e-05,
      "loss": 2.0184,
      "step": 18144
    },
    {
      "epoch": 0.03586690607317795,
      "grad_norm": 0.13137874007225037,
      "learning_rate": 9.680452692019639e-05,
      "loss": 2.0184,
      "step": 18176
    },
    {
      "epoch": 0.03593005203457439,
      "grad_norm": 0.11658748239278793,
      "learning_rate": 9.67981866958317e-05,
      "loss": 2.0203,
      "step": 18208
    },
    {
      "epoch": 0.035993197995970824,
      "grad_norm": 0.1266217827796936,
      "learning_rate": 9.679184647146702e-05,
      "loss": 2.0241,
      "step": 18240
    },
    {
      "epoch": 0.036056343957367266,
      "grad_norm": 0.12376082688570023,
      "learning_rate": 9.678550624710232e-05,
      "loss": 2.0128,
      "step": 18272
    },
    {
      "epoch": 0.03611948991876371,
      "grad_norm": 0.12112032622098923,
      "learning_rate": 9.677916602273763e-05,
      "loss": 2.0316,
      "step": 18304
    },
    {
      "epoch": 0.03618263588016015,
      "grad_norm": 0.11909458041191101,
      "learning_rate": 9.677282579837295e-05,
      "loss": 2.0278,
      "step": 18336
    },
    {
      "epoch": 0.036245781841556586,
      "grad_norm": 0.12725473940372467,
      "learning_rate": 9.676648557400826e-05,
      "loss": 2.0218,
      "step": 18368
    },
    {
      "epoch": 0.03630892780295303,
      "grad_norm": 0.11885615438222885,
      "learning_rate": 9.676014534964356e-05,
      "loss": 2.0056,
      "step": 18400
    },
    {
      "epoch": 0.03637207376434947,
      "grad_norm": 0.11043411493301392,
      "learning_rate": 9.675380512527888e-05,
      "loss": 2.0259,
      "step": 18432
    },
    {
      "epoch": 0.036435219725745906,
      "grad_norm": 0.12948209047317505,
      "learning_rate": 9.674746490091418e-05,
      "loss": 2.0014,
      "step": 18464
    },
    {
      "epoch": 0.03649836568714235,
      "grad_norm": 0.12479124963283539,
      "learning_rate": 9.67411246765495e-05,
      "loss": 2.0164,
      "step": 18496
    },
    {
      "epoch": 0.03656151164853879,
      "grad_norm": 0.11851473152637482,
      "learning_rate": 9.67347844521848e-05,
      "loss": 2.0192,
      "step": 18528
    },
    {
      "epoch": 0.036624657609935225,
      "grad_norm": 0.12374136596918106,
      "learning_rate": 9.672844422782011e-05,
      "loss": 2.0157,
      "step": 18560
    },
    {
      "epoch": 0.03668780357133167,
      "grad_norm": 0.11049731075763702,
      "learning_rate": 9.672210400345542e-05,
      "loss": 1.9978,
      "step": 18592
    },
    {
      "epoch": 0.03675094953272811,
      "grad_norm": 0.12011417001485825,
      "learning_rate": 9.671576377909074e-05,
      "loss": 2.019,
      "step": 18624
    },
    {
      "epoch": 0.036814095494124545,
      "grad_norm": 0.14559400081634521,
      "learning_rate": 9.670942355472605e-05,
      "loss": 2.027,
      "step": 18656
    },
    {
      "epoch": 0.03687724145552099,
      "grad_norm": 0.13493861258029938,
      "learning_rate": 9.670308333036135e-05,
      "loss": 2.0104,
      "step": 18688
    },
    {
      "epoch": 0.03694038741691743,
      "grad_norm": 0.11162148416042328,
      "learning_rate": 9.669674310599667e-05,
      "loss": 2.0195,
      "step": 18720
    },
    {
      "epoch": 0.037003533378313865,
      "grad_norm": 0.1158028393983841,
      "learning_rate": 9.669040288163198e-05,
      "loss": 2.0086,
      "step": 18752
    },
    {
      "epoch": 0.03706667933971031,
      "grad_norm": 0.10774479806423187,
      "learning_rate": 9.66840626572673e-05,
      "loss": 2.0136,
      "step": 18784
    },
    {
      "epoch": 0.03712982530110675,
      "grad_norm": 0.11519818007946014,
      "learning_rate": 9.66777224329026e-05,
      "loss": 2.0052,
      "step": 18816
    },
    {
      "epoch": 0.03719297126250319,
      "grad_norm": 0.12130032479763031,
      "learning_rate": 9.667138220853791e-05,
      "loss": 2.0088,
      "step": 18848
    },
    {
      "epoch": 0.037256117223899626,
      "grad_norm": 0.12328891456127167,
      "learning_rate": 9.666504198417321e-05,
      "loss": 2.0259,
      "step": 18880
    },
    {
      "epoch": 0.03731926318529607,
      "grad_norm": 0.12909002602100372,
      "learning_rate": 9.665870175980853e-05,
      "loss": 2.0058,
      "step": 18912
    },
    {
      "epoch": 0.03738240914669251,
      "grad_norm": 0.11103764921426773,
      "learning_rate": 9.665236153544383e-05,
      "loss": 2.0082,
      "step": 18944
    },
    {
      "epoch": 0.037445555108088946,
      "grad_norm": 0.13746172189712524,
      "learning_rate": 9.664602131107915e-05,
      "loss": 2.0122,
      "step": 18976
    },
    {
      "epoch": 0.03750870106948539,
      "grad_norm": 0.13963700830936432,
      "learning_rate": 9.663968108671446e-05,
      "loss": 2.0104,
      "step": 19008
    },
    {
      "epoch": 0.03757184703088183,
      "grad_norm": 0.11920302361249924,
      "learning_rate": 9.663334086234977e-05,
      "loss": 2.029,
      "step": 19040
    },
    {
      "epoch": 0.037634992992278266,
      "grad_norm": 0.11948902159929276,
      "learning_rate": 9.662700063798508e-05,
      "loss": 2.0033,
      "step": 19072
    },
    {
      "epoch": 0.03769813895367471,
      "grad_norm": 0.11687491834163666,
      "learning_rate": 9.662066041362039e-05,
      "loss": 1.998,
      "step": 19104
    },
    {
      "epoch": 0.03776128491507115,
      "grad_norm": 0.12691032886505127,
      "learning_rate": 9.66143201892557e-05,
      "loss": 2.0032,
      "step": 19136
    },
    {
      "epoch": 0.037824430876467585,
      "grad_norm": 0.11522775888442993,
      "learning_rate": 9.660797996489102e-05,
      "loss": 2.0181,
      "step": 19168
    },
    {
      "epoch": 0.03788757683786403,
      "grad_norm": 0.1483718752861023,
      "learning_rate": 9.660163974052632e-05,
      "loss": 2.0104,
      "step": 19200
    },
    {
      "epoch": 0.03795072279926047,
      "grad_norm": 0.12110782414674759,
      "learning_rate": 9.659529951616163e-05,
      "loss": 2.0086,
      "step": 19232
    },
    {
      "epoch": 0.03801386876065691,
      "grad_norm": 0.1225944310426712,
      "learning_rate": 9.658895929179695e-05,
      "loss": 2.0026,
      "step": 19264
    },
    {
      "epoch": 0.03807701472205335,
      "grad_norm": 0.11542484164237976,
      "learning_rate": 9.658261906743225e-05,
      "loss": 1.9814,
      "step": 19296
    },
    {
      "epoch": 0.03814016068344979,
      "grad_norm": 0.1180296316742897,
      "learning_rate": 9.657627884306756e-05,
      "loss": 1.9988,
      "step": 19328
    },
    {
      "epoch": 0.03820330664484623,
      "grad_norm": 0.11563392728567123,
      "learning_rate": 9.656993861870287e-05,
      "loss": 1.9985,
      "step": 19360
    },
    {
      "epoch": 0.03826645260624267,
      "grad_norm": 0.1254437267780304,
      "learning_rate": 9.656359839433818e-05,
      "loss": 2.0047,
      "step": 19392
    },
    {
      "epoch": 0.03832959856763911,
      "grad_norm": 0.11752956360578537,
      "learning_rate": 9.65572581699735e-05,
      "loss": 1.9895,
      "step": 19424
    },
    {
      "epoch": 0.03839274452903555,
      "grad_norm": 0.14189307391643524,
      "learning_rate": 9.655091794560881e-05,
      "loss": 1.9955,
      "step": 19456
    },
    {
      "epoch": 0.038455890490431986,
      "grad_norm": 0.12572942674160004,
      "learning_rate": 9.654457772124411e-05,
      "loss": 2.0009,
      "step": 19488
    },
    {
      "epoch": 0.03851903645182843,
      "grad_norm": 0.12132708728313446,
      "learning_rate": 9.653823749687942e-05,
      "loss": 1.9865,
      "step": 19520
    },
    {
      "epoch": 0.03858218241322487,
      "grad_norm": 0.12943652272224426,
      "learning_rate": 9.653189727251474e-05,
      "loss": 2.0015,
      "step": 19552
    },
    {
      "epoch": 0.038645328374621306,
      "grad_norm": 0.12172172963619232,
      "learning_rate": 9.652555704815005e-05,
      "loss": 1.9924,
      "step": 19584
    },
    {
      "epoch": 0.03870847433601775,
      "grad_norm": 0.10879118740558624,
      "learning_rate": 9.651921682378536e-05,
      "loss": 1.9977,
      "step": 19616
    },
    {
      "epoch": 0.03877162029741419,
      "grad_norm": 0.1369899958372116,
      "learning_rate": 9.651287659942067e-05,
      "loss": 1.996,
      "step": 19648
    },
    {
      "epoch": 0.03883476625881063,
      "grad_norm": 0.12542742490768433,
      "learning_rate": 9.650653637505598e-05,
      "loss": 1.9921,
      "step": 19680
    },
    {
      "epoch": 0.03889791222020707,
      "grad_norm": 0.13500872254371643,
      "learning_rate": 9.650019615069129e-05,
      "loss": 1.9998,
      "step": 19712
    },
    {
      "epoch": 0.03896105818160351,
      "grad_norm": 0.11500067263841629,
      "learning_rate": 9.649385592632659e-05,
      "loss": 1.9864,
      "step": 19744
    },
    {
      "epoch": 0.03902420414299995,
      "grad_norm": 0.14147840440273285,
      "learning_rate": 9.64875157019619e-05,
      "loss": 1.9939,
      "step": 19776
    },
    {
      "epoch": 0.03908735010439639,
      "grad_norm": 0.11108619719743729,
      "learning_rate": 9.648117547759722e-05,
      "loss": 2.0023,
      "step": 19808
    },
    {
      "epoch": 0.03915049606579283,
      "grad_norm": 0.1156499832868576,
      "learning_rate": 9.647483525323253e-05,
      "loss": 2.0081,
      "step": 19840
    },
    {
      "epoch": 0.03921364202718927,
      "grad_norm": 0.11189933866262436,
      "learning_rate": 9.646849502886783e-05,
      "loss": 1.9745,
      "step": 19872
    },
    {
      "epoch": 0.03927678798858571,
      "grad_norm": 0.13415177166461945,
      "learning_rate": 9.646215480450315e-05,
      "loss": 1.9928,
      "step": 19904
    },
    {
      "epoch": 0.03933993394998215,
      "grad_norm": 0.11870824545621872,
      "learning_rate": 9.645581458013846e-05,
      "loss": 2.0,
      "step": 19936
    },
    {
      "epoch": 0.03940307991137859,
      "grad_norm": 0.11301769316196442,
      "learning_rate": 9.644947435577377e-05,
      "loss": 1.971,
      "step": 19968
    },
    {
      "epoch": 0.039466225872775026,
      "grad_norm": 0.11457088589668274,
      "learning_rate": 9.644313413140909e-05,
      "loss": 1.986,
      "step": 20000
    },
    {
      "epoch": 0.03952937183417147,
      "grad_norm": 0.12496864050626755,
      "learning_rate": 9.643679390704439e-05,
      "loss": 2.0052,
      "step": 20032
    },
    {
      "epoch": 0.03959251779556791,
      "grad_norm": 0.10965562611818314,
      "learning_rate": 9.64304536826797e-05,
      "loss": 1.9727,
      "step": 20064
    },
    {
      "epoch": 0.03965566375696435,
      "grad_norm": 0.1069643646478653,
      "learning_rate": 9.642411345831502e-05,
      "loss": 1.9937,
      "step": 20096
    },
    {
      "epoch": 0.03971880971836079,
      "grad_norm": 0.11410249024629593,
      "learning_rate": 9.641777323395032e-05,
      "loss": 1.9865,
      "step": 20128
    },
    {
      "epoch": 0.03978195567975723,
      "grad_norm": 0.12830950319766998,
      "learning_rate": 9.641143300958562e-05,
      "loss": 1.9844,
      "step": 20160
    },
    {
      "epoch": 0.03984510164115367,
      "grad_norm": 0.12240003794431686,
      "learning_rate": 9.640509278522094e-05,
      "loss": 1.9719,
      "step": 20192
    },
    {
      "epoch": 0.03990824760255011,
      "grad_norm": 0.1246711015701294,
      "learning_rate": 9.639875256085625e-05,
      "loss": 1.9975,
      "step": 20224
    },
    {
      "epoch": 0.03997139356394655,
      "grad_norm": 0.10353072732686996,
      "learning_rate": 9.639241233649156e-05,
      "loss": 1.9852,
      "step": 20256
    },
    {
      "epoch": 0.04003453952534299,
      "grad_norm": 0.10986755043268204,
      "learning_rate": 9.638607211212687e-05,
      "loss": 1.9754,
      "step": 20288
    },
    {
      "epoch": 0.04009768548673943,
      "grad_norm": 0.12473009526729584,
      "learning_rate": 9.637973188776218e-05,
      "loss": 1.9844,
      "step": 20320
    },
    {
      "epoch": 0.04016083144813587,
      "grad_norm": 0.12376633286476135,
      "learning_rate": 9.63733916633975e-05,
      "loss": 1.9733,
      "step": 20352
    },
    {
      "epoch": 0.04022397740953231,
      "grad_norm": 0.12706255912780762,
      "learning_rate": 9.636705143903281e-05,
      "loss": 1.9858,
      "step": 20384
    },
    {
      "epoch": 0.04028712337092875,
      "grad_norm": 0.1247047558426857,
      "learning_rate": 9.636071121466811e-05,
      "loss": 1.9652,
      "step": 20416
    },
    {
      "epoch": 0.04035026933232519,
      "grad_norm": 0.13165326416492462,
      "learning_rate": 9.635437099030343e-05,
      "loss": 1.9796,
      "step": 20448
    },
    {
      "epoch": 0.04041341529372163,
      "grad_norm": 0.12569890916347504,
      "learning_rate": 9.634803076593874e-05,
      "loss": 1.977,
      "step": 20480
    },
    {
      "epoch": 0.040476561255118074,
      "grad_norm": 0.10226519405841827,
      "learning_rate": 9.634169054157405e-05,
      "loss": 1.9805,
      "step": 20512
    },
    {
      "epoch": 0.04053970721651451,
      "grad_norm": 0.11636253446340561,
      "learning_rate": 9.633535031720936e-05,
      "loss": 2.0009,
      "step": 20544
    },
    {
      "epoch": 0.04060285317791095,
      "grad_norm": 0.12742649018764496,
      "learning_rate": 9.632901009284466e-05,
      "loss": 1.9942,
      "step": 20576
    },
    {
      "epoch": 0.04066599913930739,
      "grad_norm": 0.13119980692863464,
      "learning_rate": 9.632266986847997e-05,
      "loss": 1.981,
      "step": 20608
    },
    {
      "epoch": 0.04072914510070383,
      "grad_norm": 0.11325479298830032,
      "learning_rate": 9.631632964411529e-05,
      "loss": 1.9585,
      "step": 20640
    },
    {
      "epoch": 0.04079229106210027,
      "grad_norm": 0.10867073386907578,
      "learning_rate": 9.63099894197506e-05,
      "loss": 1.9908,
      "step": 20672
    },
    {
      "epoch": 0.04085543702349671,
      "grad_norm": 0.12300091981887817,
      "learning_rate": 9.63036491953859e-05,
      "loss": 1.9766,
      "step": 20704
    },
    {
      "epoch": 0.04091858298489315,
      "grad_norm": 0.10913708806037903,
      "learning_rate": 9.629730897102122e-05,
      "loss": 1.986,
      "step": 20736
    },
    {
      "epoch": 0.04098172894628959,
      "grad_norm": 0.12194447964429855,
      "learning_rate": 9.629096874665653e-05,
      "loss": 1.9827,
      "step": 20768
    },
    {
      "epoch": 0.04104487490768603,
      "grad_norm": 0.11287178844213486,
      "learning_rate": 9.628462852229184e-05,
      "loss": 1.9894,
      "step": 20800
    },
    {
      "epoch": 0.04110802086908247,
      "grad_norm": 0.1337476372718811,
      "learning_rate": 9.627828829792715e-05,
      "loss": 1.9593,
      "step": 20832
    },
    {
      "epoch": 0.04117116683047891,
      "grad_norm": 0.1178978905081749,
      "learning_rate": 9.627194807356246e-05,
      "loss": 1.9828,
      "step": 20864
    },
    {
      "epoch": 0.04123431279187535,
      "grad_norm": 0.11720005422830582,
      "learning_rate": 9.626560784919777e-05,
      "loss": 1.9865,
      "step": 20896
    },
    {
      "epoch": 0.041297458753271794,
      "grad_norm": 0.1083127111196518,
      "learning_rate": 9.625926762483308e-05,
      "loss": 1.9826,
      "step": 20928
    },
    {
      "epoch": 0.04136060471466823,
      "grad_norm": 0.10887587070465088,
      "learning_rate": 9.625292740046839e-05,
      "loss": 1.9741,
      "step": 20960
    },
    {
      "epoch": 0.04142375067606467,
      "grad_norm": 0.12325512617826462,
      "learning_rate": 9.624658717610369e-05,
      "loss": 1.9762,
      "step": 20992
    },
    {
      "epoch": 0.041486896637461114,
      "grad_norm": 0.11215290427207947,
      "learning_rate": 9.6240246951739e-05,
      "loss": 1.9804,
      "step": 21024
    },
    {
      "epoch": 0.04155004259885755,
      "grad_norm": 0.12300876528024673,
      "learning_rate": 9.623390672737432e-05,
      "loss": 1.9713,
      "step": 21056
    },
    {
      "epoch": 0.04161318856025399,
      "grad_norm": 0.10765711218118668,
      "learning_rate": 9.622756650300962e-05,
      "loss": 1.9738,
      "step": 21088
    },
    {
      "epoch": 0.041676334521650434,
      "grad_norm": 0.1282142698764801,
      "learning_rate": 9.622122627864494e-05,
      "loss": 1.9588,
      "step": 21120
    },
    {
      "epoch": 0.04173948048304687,
      "grad_norm": 0.12258365750312805,
      "learning_rate": 9.621488605428025e-05,
      "loss": 1.9814,
      "step": 21152
    },
    {
      "epoch": 0.04180262644444331,
      "grad_norm": 0.12109114229679108,
      "learning_rate": 9.620854582991557e-05,
      "loss": 1.9573,
      "step": 21184
    },
    {
      "epoch": 0.04186577240583975,
      "grad_norm": 0.11835681647062302,
      "learning_rate": 9.620220560555087e-05,
      "loss": 1.9777,
      "step": 21216
    },
    {
      "epoch": 0.04192891836723619,
      "grad_norm": 0.11301271617412567,
      "learning_rate": 9.619586538118618e-05,
      "loss": 1.9748,
      "step": 21248
    },
    {
      "epoch": 0.04199206432863263,
      "grad_norm": 0.11349284648895264,
      "learning_rate": 9.61895251568215e-05,
      "loss": 1.9781,
      "step": 21280
    },
    {
      "epoch": 0.04205521029002907,
      "grad_norm": 0.11842041462659836,
      "learning_rate": 9.618318493245681e-05,
      "loss": 1.9841,
      "step": 21312
    },
    {
      "epoch": 0.042118356251425515,
      "grad_norm": 0.11143847554922104,
      "learning_rate": 9.617684470809211e-05,
      "loss": 1.9714,
      "step": 21344
    },
    {
      "epoch": 0.04218150221282195,
      "grad_norm": 0.10669038444757462,
      "learning_rate": 9.617050448372743e-05,
      "loss": 1.974,
      "step": 21376
    },
    {
      "epoch": 0.04224464817421839,
      "grad_norm": 0.103308767080307,
      "learning_rate": 9.616416425936273e-05,
      "loss": 1.9596,
      "step": 21408
    },
    {
      "epoch": 0.042307794135614835,
      "grad_norm": 0.10760737955570221,
      "learning_rate": 9.615782403499804e-05,
      "loss": 1.9776,
      "step": 21440
    },
    {
      "epoch": 0.04237094009701127,
      "grad_norm": 0.11961367726325989,
      "learning_rate": 9.615148381063336e-05,
      "loss": 1.9557,
      "step": 21472
    },
    {
      "epoch": 0.04243408605840771,
      "grad_norm": 0.12137002497911453,
      "learning_rate": 9.614514358626866e-05,
      "loss": 1.9654,
      "step": 21504
    },
    {
      "epoch": 0.042497232019804154,
      "grad_norm": 0.10979246348142624,
      "learning_rate": 9.613880336190397e-05,
      "loss": 1.9636,
      "step": 21536
    },
    {
      "epoch": 0.04256037798120059,
      "grad_norm": 0.12187566608190536,
      "learning_rate": 9.613246313753929e-05,
      "loss": 1.9586,
      "step": 21568
    },
    {
      "epoch": 0.04262352394259703,
      "grad_norm": 0.11190672963857651,
      "learning_rate": 9.61261229131746e-05,
      "loss": 1.9561,
      "step": 21600
    },
    {
      "epoch": 0.042686669903993474,
      "grad_norm": 0.12588316202163696,
      "learning_rate": 9.61197826888099e-05,
      "loss": 1.959,
      "step": 21632
    },
    {
      "epoch": 0.04274981586538991,
      "grad_norm": 0.11184147745370865,
      "learning_rate": 9.611344246444522e-05,
      "loss": 1.9643,
      "step": 21664
    },
    {
      "epoch": 0.04281296182678635,
      "grad_norm": 0.1143326535820961,
      "learning_rate": 9.610710224008053e-05,
      "loss": 1.9755,
      "step": 21696
    },
    {
      "epoch": 0.04287610778818279,
      "grad_norm": 0.1150774285197258,
      "learning_rate": 9.610076201571585e-05,
      "loss": 1.967,
      "step": 21728
    },
    {
      "epoch": 0.042939253749579236,
      "grad_norm": 0.11972521990537643,
      "learning_rate": 9.609442179135115e-05,
      "loss": 1.9678,
      "step": 21760
    },
    {
      "epoch": 0.04300239971097567,
      "grad_norm": 0.12995310127735138,
      "learning_rate": 9.608808156698646e-05,
      "loss": 1.958,
      "step": 21792
    },
    {
      "epoch": 0.04306554567237211,
      "grad_norm": 0.13119225203990936,
      "learning_rate": 9.608174134262176e-05,
      "loss": 1.9775,
      "step": 21824
    },
    {
      "epoch": 0.043128691633768555,
      "grad_norm": 0.11164345592260361,
      "learning_rate": 9.607540111825708e-05,
      "loss": 1.9784,
      "step": 21856
    },
    {
      "epoch": 0.04319183759516499,
      "grad_norm": 0.1257980614900589,
      "learning_rate": 9.606906089389238e-05,
      "loss": 1.9688,
      "step": 21888
    },
    {
      "epoch": 0.04325498355656143,
      "grad_norm": 0.11268804967403412,
      "learning_rate": 9.606272066952769e-05,
      "loss": 1.967,
      "step": 21920
    },
    {
      "epoch": 0.043318129517957875,
      "grad_norm": 0.1409463733434677,
      "learning_rate": 9.6056380445163e-05,
      "loss": 1.9557,
      "step": 21952
    },
    {
      "epoch": 0.04338127547935431,
      "grad_norm": 0.126288503408432,
      "learning_rate": 9.605004022079832e-05,
      "loss": 1.9611,
      "step": 21984
    },
    {
      "epoch": 0.04344442144075075,
      "grad_norm": 0.10051508992910385,
      "learning_rate": 9.604369999643364e-05,
      "loss": 1.9593,
      "step": 22016
    },
    {
      "epoch": 0.043507567402147194,
      "grad_norm": 0.12255176156759262,
      "learning_rate": 9.603735977206894e-05,
      "loss": 1.9688,
      "step": 22048
    },
    {
      "epoch": 0.04357071336354363,
      "grad_norm": 0.11303877085447311,
      "learning_rate": 9.603101954770425e-05,
      "loss": 1.9406,
      "step": 22080
    },
    {
      "epoch": 0.04363385932494007,
      "grad_norm": 0.11862237006425858,
      "learning_rate": 9.602467932333957e-05,
      "loss": 1.9634,
      "step": 22112
    },
    {
      "epoch": 0.043697005286336514,
      "grad_norm": 0.10821133852005005,
      "learning_rate": 9.601833909897488e-05,
      "loss": 1.9418,
      "step": 22144
    },
    {
      "epoch": 0.043760151247732956,
      "grad_norm": 0.1072334423661232,
      "learning_rate": 9.601199887461018e-05,
      "loss": 1.9793,
      "step": 22176
    },
    {
      "epoch": 0.04382329720912939,
      "grad_norm": 0.10535736382007599,
      "learning_rate": 9.60056586502455e-05,
      "loss": 1.9575,
      "step": 22208
    },
    {
      "epoch": 0.043886443170525834,
      "grad_norm": 0.12884929776191711,
      "learning_rate": 9.59993184258808e-05,
      "loss": 1.9629,
      "step": 22240
    },
    {
      "epoch": 0.043949589131922276,
      "grad_norm": 0.11832892894744873,
      "learning_rate": 9.599297820151611e-05,
      "loss": 1.9699,
      "step": 22272
    },
    {
      "epoch": 0.04401273509331871,
      "grad_norm": 0.14499302208423615,
      "learning_rate": 9.598663797715141e-05,
      "loss": 1.9592,
      "step": 22304
    },
    {
      "epoch": 0.04407588105471515,
      "grad_norm": 0.11090567708015442,
      "learning_rate": 9.598029775278673e-05,
      "loss": 1.9468,
      "step": 22336
    },
    {
      "epoch": 0.044139027016111596,
      "grad_norm": 0.11581122875213623,
      "learning_rate": 9.597395752842204e-05,
      "loss": 1.9616,
      "step": 22368
    },
    {
      "epoch": 0.04420217297750803,
      "grad_norm": 0.11382453143596649,
      "learning_rate": 9.596761730405736e-05,
      "loss": 1.9456,
      "step": 22400
    },
    {
      "epoch": 0.04426531893890447,
      "grad_norm": 0.12293533235788345,
      "learning_rate": 9.596127707969266e-05,
      "loss": 1.9452,
      "step": 22432
    },
    {
      "epoch": 0.044328464900300915,
      "grad_norm": 0.13024349510669708,
      "learning_rate": 9.595493685532797e-05,
      "loss": 1.9593,
      "step": 22464
    },
    {
      "epoch": 0.04439161086169735,
      "grad_norm": 0.14722859859466553,
      "learning_rate": 9.594859663096329e-05,
      "loss": 1.9746,
      "step": 22496
    },
    {
      "epoch": 0.04445475682309379,
      "grad_norm": 0.10557448118925095,
      "learning_rate": 9.59422564065986e-05,
      "loss": 1.9641,
      "step": 22528
    },
    {
      "epoch": 0.044517902784490235,
      "grad_norm": 0.11203256249427795,
      "learning_rate": 9.59359161822339e-05,
      "loss": 1.9505,
      "step": 22560
    },
    {
      "epoch": 0.04458104874588668,
      "grad_norm": 0.11080003529787064,
      "learning_rate": 9.592957595786922e-05,
      "loss": 1.966,
      "step": 22592
    },
    {
      "epoch": 0.04464419470728311,
      "grad_norm": 0.10656057298183441,
      "learning_rate": 9.592323573350452e-05,
      "loss": 1.9584,
      "step": 22624
    },
    {
      "epoch": 0.044707340668679554,
      "grad_norm": 0.12340083718299866,
      "learning_rate": 9.591689550913983e-05,
      "loss": 1.9526,
      "step": 22656
    },
    {
      "epoch": 0.044770486630076,
      "grad_norm": 0.1045255959033966,
      "learning_rate": 9.591055528477515e-05,
      "loss": 1.9646,
      "step": 22688
    },
    {
      "epoch": 0.04483363259147243,
      "grad_norm": 0.09781979024410248,
      "learning_rate": 9.590421506041045e-05,
      "loss": 1.9531,
      "step": 22720
    },
    {
      "epoch": 0.044896778552868874,
      "grad_norm": 0.11920382082462311,
      "learning_rate": 9.589787483604576e-05,
      "loss": 1.9566,
      "step": 22752
    },
    {
      "epoch": 0.044959924514265316,
      "grad_norm": 0.10923431068658829,
      "learning_rate": 9.589153461168108e-05,
      "loss": 1.9396,
      "step": 22784
    },
    {
      "epoch": 0.04502307047566175,
      "grad_norm": 0.1006297841668129,
      "learning_rate": 9.588519438731639e-05,
      "loss": 1.9499,
      "step": 22816
    },
    {
      "epoch": 0.045086216437058194,
      "grad_norm": 0.11739280074834824,
      "learning_rate": 9.587885416295169e-05,
      "loss": 1.9578,
      "step": 22848
    },
    {
      "epoch": 0.045149362398454636,
      "grad_norm": 0.12821663916110992,
      "learning_rate": 9.5872513938587e-05,
      "loss": 1.9512,
      "step": 22880
    },
    {
      "epoch": 0.04521250835985107,
      "grad_norm": 0.13252002000808716,
      "learning_rate": 9.586617371422232e-05,
      "loss": 1.9356,
      "step": 22912
    },
    {
      "epoch": 0.04527565432124751,
      "grad_norm": 0.12858210504055023,
      "learning_rate": 9.585983348985764e-05,
      "loss": 1.9307,
      "step": 22944
    },
    {
      "epoch": 0.045338800282643955,
      "grad_norm": 0.10175346583127975,
      "learning_rate": 9.585349326549294e-05,
      "loss": 1.9372,
      "step": 22976
    },
    {
      "epoch": 0.0454019462440404,
      "grad_norm": 0.1078830286860466,
      "learning_rate": 9.584715304112825e-05,
      "loss": 1.9599,
      "step": 23008
    },
    {
      "epoch": 0.04546509220543683,
      "grad_norm": 0.10003285855054855,
      "learning_rate": 9.584081281676355e-05,
      "loss": 1.9467,
      "step": 23040
    },
    {
      "epoch": 0.045528238166833275,
      "grad_norm": 0.10893139243125916,
      "learning_rate": 9.583447259239887e-05,
      "loss": 1.9467,
      "step": 23072
    },
    {
      "epoch": 0.04559138412822972,
      "grad_norm": 0.10260356217622757,
      "learning_rate": 9.582813236803417e-05,
      "loss": 1.9394,
      "step": 23104
    },
    {
      "epoch": 0.04565453008962615,
      "grad_norm": 0.11526516079902649,
      "learning_rate": 9.582179214366948e-05,
      "loss": 1.9511,
      "step": 23136
    },
    {
      "epoch": 0.045717676051022595,
      "grad_norm": 0.12039726227521896,
      "learning_rate": 9.58154519193048e-05,
      "loss": 1.9502,
      "step": 23168
    },
    {
      "epoch": 0.04578082201241904,
      "grad_norm": 0.10691642761230469,
      "learning_rate": 9.580911169494011e-05,
      "loss": 1.937,
      "step": 23200
    },
    {
      "epoch": 0.04584396797381547,
      "grad_norm": 0.11702313274145126,
      "learning_rate": 9.580277147057541e-05,
      "loss": 1.9532,
      "step": 23232
    },
    {
      "epoch": 0.045907113935211914,
      "grad_norm": 0.11025272309780121,
      "learning_rate": 9.579643124621073e-05,
      "loss": 1.9378,
      "step": 23264
    },
    {
      "epoch": 0.045970259896608356,
      "grad_norm": 0.10566860437393188,
      "learning_rate": 9.579009102184604e-05,
      "loss": 1.9355,
      "step": 23296
    },
    {
      "epoch": 0.04603340585800479,
      "grad_norm": 0.1200597733259201,
      "learning_rate": 9.578375079748136e-05,
      "loss": 1.9406,
      "step": 23328
    },
    {
      "epoch": 0.046096551819401234,
      "grad_norm": 0.10709843784570694,
      "learning_rate": 9.577741057311667e-05,
      "loss": 1.9403,
      "step": 23360
    },
    {
      "epoch": 0.046159697780797676,
      "grad_norm": 0.11283406615257263,
      "learning_rate": 9.577107034875197e-05,
      "loss": 1.9302,
      "step": 23392
    },
    {
      "epoch": 0.04622284374219412,
      "grad_norm": 0.1264004111289978,
      "learning_rate": 9.576473012438729e-05,
      "loss": 1.9477,
      "step": 23424
    },
    {
      "epoch": 0.046285989703590553,
      "grad_norm": 0.09890617430210114,
      "learning_rate": 9.575838990002259e-05,
      "loss": 1.929,
      "step": 23456
    },
    {
      "epoch": 0.046349135664986996,
      "grad_norm": 0.11972219496965408,
      "learning_rate": 9.57520496756579e-05,
      "loss": 1.9466,
      "step": 23488
    },
    {
      "epoch": 0.04641228162638344,
      "grad_norm": 0.11952702701091766,
      "learning_rate": 9.57457094512932e-05,
      "loss": 1.9369,
      "step": 23520
    },
    {
      "epoch": 0.04647542758777987,
      "grad_norm": 0.12452511489391327,
      "learning_rate": 9.573936922692852e-05,
      "loss": 1.9306,
      "step": 23552
    },
    {
      "epoch": 0.046538573549176315,
      "grad_norm": 0.14350657165050507,
      "learning_rate": 9.573302900256383e-05,
      "loss": 1.925,
      "step": 23584
    },
    {
      "epoch": 0.04660171951057276,
      "grad_norm": 0.10996013134717941,
      "learning_rate": 9.572668877819915e-05,
      "loss": 1.9475,
      "step": 23616
    },
    {
      "epoch": 0.04666486547196919,
      "grad_norm": 0.12992681562900543,
      "learning_rate": 9.572034855383445e-05,
      "loss": 1.9342,
      "step": 23648
    },
    {
      "epoch": 0.046728011433365635,
      "grad_norm": 0.10513056814670563,
      "learning_rate": 9.571400832946976e-05,
      "loss": 1.9313,
      "step": 23680
    },
    {
      "epoch": 0.04679115739476208,
      "grad_norm": 0.11533654481172562,
      "learning_rate": 9.570766810510508e-05,
      "loss": 1.9481,
      "step": 23712
    },
    {
      "epoch": 0.04685430335615851,
      "grad_norm": 0.10715834051370621,
      "learning_rate": 9.570132788074039e-05,
      "loss": 1.9474,
      "step": 23744
    },
    {
      "epoch": 0.046917449317554955,
      "grad_norm": 0.11807551980018616,
      "learning_rate": 9.569498765637569e-05,
      "loss": 1.9461,
      "step": 23776
    },
    {
      "epoch": 0.0469805952789514,
      "grad_norm": 0.10314300656318665,
      "learning_rate": 9.568864743201101e-05,
      "loss": 1.9291,
      "step": 23808
    },
    {
      "epoch": 0.04704374124034784,
      "grad_norm": 0.1296030431985855,
      "learning_rate": 9.568230720764632e-05,
      "loss": 1.9255,
      "step": 23840
    },
    {
      "epoch": 0.047106887201744274,
      "grad_norm": 0.1086842343211174,
      "learning_rate": 9.567596698328162e-05,
      "loss": 1.9336,
      "step": 23872
    },
    {
      "epoch": 0.047170033163140716,
      "grad_norm": 0.1167302206158638,
      "learning_rate": 9.566962675891692e-05,
      "loss": 1.9315,
      "step": 23904
    },
    {
      "epoch": 0.04723317912453716,
      "grad_norm": 0.11264488846063614,
      "learning_rate": 9.566328653455224e-05,
      "loss": 1.9334,
      "step": 23936
    },
    {
      "epoch": 0.047296325085933594,
      "grad_norm": 0.10670772939920425,
      "learning_rate": 9.565694631018755e-05,
      "loss": 1.927,
      "step": 23968
    },
    {
      "epoch": 0.047359471047330036,
      "grad_norm": 0.11302254348993301,
      "learning_rate": 9.565060608582287e-05,
      "loss": 1.9355,
      "step": 24000
    },
    {
      "epoch": 0.04742261700872648,
      "grad_norm": 0.1167798787355423,
      "learning_rate": 9.564426586145818e-05,
      "loss": 1.9291,
      "step": 24032
    },
    {
      "epoch": 0.04748576297012291,
      "grad_norm": 0.11112770438194275,
      "learning_rate": 9.563792563709348e-05,
      "loss": 1.9323,
      "step": 24064
    },
    {
      "epoch": 0.047548908931519356,
      "grad_norm": 0.10099230706691742,
      "learning_rate": 9.56315854127288e-05,
      "loss": 1.9368,
      "step": 24096
    },
    {
      "epoch": 0.0476120548929158,
      "grad_norm": 0.11324863880872726,
      "learning_rate": 9.562524518836411e-05,
      "loss": 1.9417,
      "step": 24128
    },
    {
      "epoch": 0.04767520085431223,
      "grad_norm": 0.1250937134027481,
      "learning_rate": 9.561890496399943e-05,
      "loss": 1.9231,
      "step": 24160
    },
    {
      "epoch": 0.047738346815708675,
      "grad_norm": 0.13605114817619324,
      "learning_rate": 9.561256473963473e-05,
      "loss": 1.9371,
      "step": 24192
    },
    {
      "epoch": 0.04780149277710512,
      "grad_norm": 0.10294576734304428,
      "learning_rate": 9.560622451527004e-05,
      "loss": 1.9365,
      "step": 24224
    },
    {
      "epoch": 0.04786463873850156,
      "grad_norm": 0.10928324609994888,
      "learning_rate": 9.559988429090536e-05,
      "loss": 1.9221,
      "step": 24256
    },
    {
      "epoch": 0.047927784699897995,
      "grad_norm": 0.11262323707342148,
      "learning_rate": 9.559354406654066e-05,
      "loss": 1.9337,
      "step": 24288
    },
    {
      "epoch": 0.04799093066129444,
      "grad_norm": 0.10129617899656296,
      "learning_rate": 9.558720384217596e-05,
      "loss": 1.9378,
      "step": 24320
    },
    {
      "epoch": 0.04805407662269088,
      "grad_norm": 0.1110493540763855,
      "learning_rate": 9.558086361781127e-05,
      "loss": 1.9252,
      "step": 24352
    },
    {
      "epoch": 0.048117222584087314,
      "grad_norm": 0.11660345643758774,
      "learning_rate": 9.557452339344659e-05,
      "loss": 1.9259,
      "step": 24384
    },
    {
      "epoch": 0.04818036854548376,
      "grad_norm": 0.11428438872098923,
      "learning_rate": 9.55681831690819e-05,
      "loss": 1.9261,
      "step": 24416
    },
    {
      "epoch": 0.0482435145068802,
      "grad_norm": 0.12056509405374527,
      "learning_rate": 9.55618429447172e-05,
      "loss": 1.9343,
      "step": 24448
    },
    {
      "epoch": 0.048306660468276634,
      "grad_norm": 0.14168789982795715,
      "learning_rate": 9.555550272035252e-05,
      "loss": 1.9282,
      "step": 24480
    },
    {
      "epoch": 0.048369806429673076,
      "grad_norm": 0.11102975159883499,
      "learning_rate": 9.554916249598783e-05,
      "loss": 1.9352,
      "step": 24512
    },
    {
      "epoch": 0.04843295239106952,
      "grad_norm": 0.11062134057283401,
      "learning_rate": 9.554282227162315e-05,
      "loss": 1.9253,
      "step": 24544
    },
    {
      "epoch": 0.048496098352465954,
      "grad_norm": 0.10571175068616867,
      "learning_rate": 9.553648204725845e-05,
      "loss": 1.9398,
      "step": 24576
    },
    {
      "epoch": 0.048559244313862396,
      "grad_norm": 0.11620599776506424,
      "learning_rate": 9.553014182289376e-05,
      "loss": 1.9344,
      "step": 24608
    },
    {
      "epoch": 0.04862239027525884,
      "grad_norm": 0.11041609942913055,
      "learning_rate": 9.552380159852908e-05,
      "loss": 1.9211,
      "step": 24640
    },
    {
      "epoch": 0.04868553623665528,
      "grad_norm": 0.1092856228351593,
      "learning_rate": 9.551746137416439e-05,
      "loss": 1.9293,
      "step": 24672
    },
    {
      "epoch": 0.048748682198051715,
      "grad_norm": 0.12742461264133453,
      "learning_rate": 9.551112114979969e-05,
      "loss": 1.9154,
      "step": 24704
    },
    {
      "epoch": 0.04881182815944816,
      "grad_norm": 0.11318569630384445,
      "learning_rate": 9.5504780925435e-05,
      "loss": 1.9275,
      "step": 24736
    },
    {
      "epoch": 0.0488749741208446,
      "grad_norm": 0.10278608649969101,
      "learning_rate": 9.549844070107031e-05,
      "loss": 1.9182,
      "step": 24768
    },
    {
      "epoch": 0.048938120082241035,
      "grad_norm": 0.10825817286968231,
      "learning_rate": 9.549210047670562e-05,
      "loss": 1.924,
      "step": 24800
    },
    {
      "epoch": 0.04900126604363748,
      "grad_norm": 0.13008153438568115,
      "learning_rate": 9.548576025234094e-05,
      "loss": 1.9356,
      "step": 24832
    },
    {
      "epoch": 0.04906441200503392,
      "grad_norm": 0.0987807959318161,
      "learning_rate": 9.547942002797624e-05,
      "loss": 1.9309,
      "step": 24864
    },
    {
      "epoch": 0.049127557966430355,
      "grad_norm": 0.12102729082107544,
      "learning_rate": 9.547307980361155e-05,
      "loss": 1.9162,
      "step": 24896
    },
    {
      "epoch": 0.0491907039278268,
      "grad_norm": 0.10271728038787842,
      "learning_rate": 9.546673957924687e-05,
      "loss": 1.9239,
      "step": 24928
    },
    {
      "epoch": 0.04925384988922324,
      "grad_norm": 0.11824354529380798,
      "learning_rate": 9.546039935488218e-05,
      "loss": 1.9357,
      "step": 24960
    },
    {
      "epoch": 0.049316995850619674,
      "grad_norm": 0.10709060728549957,
      "learning_rate": 9.545405913051748e-05,
      "loss": 1.9299,
      "step": 24992
    },
    {
      "epoch": 0.049380141812016116,
      "grad_norm": 0.10140976309776306,
      "learning_rate": 9.54477189061528e-05,
      "loss": 1.9323,
      "step": 25024
    },
    {
      "epoch": 0.04944328777341256,
      "grad_norm": 0.11804775148630142,
      "learning_rate": 9.544137868178811e-05,
      "loss": 1.9181,
      "step": 25056
    },
    {
      "epoch": 0.049506433734809,
      "grad_norm": 0.11727722734212875,
      "learning_rate": 9.543503845742343e-05,
      "loss": 1.9206,
      "step": 25088
    },
    {
      "epoch": 0.049569579696205436,
      "grad_norm": 0.10891515016555786,
      "learning_rate": 9.542869823305873e-05,
      "loss": 1.9103,
      "step": 25120
    },
    {
      "epoch": 0.04963272565760188,
      "grad_norm": 0.11480802297592163,
      "learning_rate": 9.542235800869403e-05,
      "loss": 1.916,
      "step": 25152
    },
    {
      "epoch": 0.04969587161899832,
      "grad_norm": 0.1161581426858902,
      "learning_rate": 9.541601778432934e-05,
      "loss": 1.9142,
      "step": 25184
    },
    {
      "epoch": 0.049759017580394756,
      "grad_norm": 0.11386236548423767,
      "learning_rate": 9.540967755996466e-05,
      "loss": 1.9098,
      "step": 25216
    },
    {
      "epoch": 0.0498221635417912,
      "grad_norm": 0.1235729232430458,
      "learning_rate": 9.540333733559996e-05,
      "loss": 1.9319,
      "step": 25248
    },
    {
      "epoch": 0.04988530950318764,
      "grad_norm": 0.1056947186589241,
      "learning_rate": 9.539699711123527e-05,
      "loss": 1.9405,
      "step": 25280
    },
    {
      "epoch": 0.049948455464584075,
      "grad_norm": 0.11457525193691254,
      "learning_rate": 9.539065688687059e-05,
      "loss": 1.9017,
      "step": 25312
    },
    {
      "epoch": 0.05001160142598052,
      "grad_norm": 0.11316017061471939,
      "learning_rate": 9.53843166625059e-05,
      "loss": 1.9172,
      "step": 25344
    },
    {
      "epoch": 0.05007474738737696,
      "grad_norm": 0.10579028725624084,
      "learning_rate": 9.537797643814122e-05,
      "loss": 1.9146,
      "step": 25376
    },
    {
      "epoch": 0.050137893348773395,
      "grad_norm": 0.09902004897594452,
      "learning_rate": 9.537163621377652e-05,
      "loss": 1.9151,
      "step": 25408
    },
    {
      "epoch": 0.05020103931016984,
      "grad_norm": 0.11914989352226257,
      "learning_rate": 9.536529598941183e-05,
      "loss": 1.922,
      "step": 25440
    },
    {
      "epoch": 0.05026418527156628,
      "grad_norm": 0.10532984137535095,
      "learning_rate": 9.535895576504715e-05,
      "loss": 1.915,
      "step": 25472
    },
    {
      "epoch": 0.05032733123296272,
      "grad_norm": 0.11634562164545059,
      "learning_rate": 9.535261554068245e-05,
      "loss": 1.9147,
      "step": 25504
    },
    {
      "epoch": 0.05039047719435916,
      "grad_norm": 0.11276385188102722,
      "learning_rate": 9.534627531631776e-05,
      "loss": 1.9116,
      "step": 25536
    },
    {
      "epoch": 0.0504536231557556,
      "grad_norm": 0.10931231826543808,
      "learning_rate": 9.533993509195306e-05,
      "loss": 1.8955,
      "step": 25568
    },
    {
      "epoch": 0.05051676911715204,
      "grad_norm": 0.10813896358013153,
      "learning_rate": 9.533359486758838e-05,
      "loss": 1.9097,
      "step": 25600
    },
    {
      "epoch": 0.050579915078548476,
      "grad_norm": 0.11851253360509872,
      "learning_rate": 9.532725464322369e-05,
      "loss": 1.9201,
      "step": 25632
    },
    {
      "epoch": 0.05064306103994492,
      "grad_norm": 0.09754260629415512,
      "learning_rate": 9.5320914418859e-05,
      "loss": 1.9143,
      "step": 25664
    },
    {
      "epoch": 0.05070620700134136,
      "grad_norm": 0.09885375201702118,
      "learning_rate": 9.531457419449431e-05,
      "loss": 1.9091,
      "step": 25696
    },
    {
      "epoch": 0.050769352962737796,
      "grad_norm": 0.10297214984893799,
      "learning_rate": 9.530823397012962e-05,
      "loss": 1.9143,
      "step": 25728
    },
    {
      "epoch": 0.05083249892413424,
      "grad_norm": 0.10813931375741959,
      "learning_rate": 9.530189374576494e-05,
      "loss": 1.9143,
      "step": 25760
    },
    {
      "epoch": 0.05089564488553068,
      "grad_norm": 0.10659068822860718,
      "learning_rate": 9.529555352140024e-05,
      "loss": 1.91,
      "step": 25792
    },
    {
      "epoch": 0.050958790846927116,
      "grad_norm": 0.12919165194034576,
      "learning_rate": 9.528921329703555e-05,
      "loss": 1.9225,
      "step": 25824
    },
    {
      "epoch": 0.05102193680832356,
      "grad_norm": 0.10619667917490005,
      "learning_rate": 9.528287307267087e-05,
      "loss": 1.9228,
      "step": 25856
    },
    {
      "epoch": 0.05108508276972,
      "grad_norm": 0.10115159302949905,
      "learning_rate": 9.527653284830618e-05,
      "loss": 1.9056,
      "step": 25888
    },
    {
      "epoch": 0.05114822873111644,
      "grad_norm": 0.11068226397037506,
      "learning_rate": 9.527019262394148e-05,
      "loss": 1.918,
      "step": 25920
    },
    {
      "epoch": 0.05121137469251288,
      "grad_norm": 0.11327261477708817,
      "learning_rate": 9.52638523995768e-05,
      "loss": 1.9259,
      "step": 25952
    },
    {
      "epoch": 0.05127452065390932,
      "grad_norm": 0.1283046156167984,
      "learning_rate": 9.52575121752121e-05,
      "loss": 1.9046,
      "step": 25984
    },
    {
      "epoch": 0.05133766661530576,
      "grad_norm": 0.10593913495540619,
      "learning_rate": 9.525117195084741e-05,
      "loss": 1.8981,
      "step": 26016
    },
    {
      "epoch": 0.0514008125767022,
      "grad_norm": 0.11951139569282532,
      "learning_rate": 9.524483172648273e-05,
      "loss": 1.9145,
      "step": 26048
    },
    {
      "epoch": 0.05146395853809864,
      "grad_norm": 0.10877706110477448,
      "learning_rate": 9.523849150211803e-05,
      "loss": 1.9183,
      "step": 26080
    },
    {
      "epoch": 0.05152710449949508,
      "grad_norm": 0.10846201330423355,
      "learning_rate": 9.523215127775334e-05,
      "loss": 1.9038,
      "step": 26112
    },
    {
      "epoch": 0.05159025046089152,
      "grad_norm": 0.10390051454305649,
      "learning_rate": 9.522581105338866e-05,
      "loss": 1.9048,
      "step": 26144
    },
    {
      "epoch": 0.05165339642228796,
      "grad_norm": 0.12031883001327515,
      "learning_rate": 9.521947082902397e-05,
      "loss": 1.9125,
      "step": 26176
    },
    {
      "epoch": 0.0517165423836844,
      "grad_norm": 0.10653280466794968,
      "learning_rate": 9.521313060465927e-05,
      "loss": 1.908,
      "step": 26208
    },
    {
      "epoch": 0.051779688345080836,
      "grad_norm": 0.12093240022659302,
      "learning_rate": 9.520679038029459e-05,
      "loss": 1.9114,
      "step": 26240
    },
    {
      "epoch": 0.05184283430647728,
      "grad_norm": 0.1351175308227539,
      "learning_rate": 9.52004501559299e-05,
      "loss": 1.8928,
      "step": 26272
    },
    {
      "epoch": 0.05190598026787372,
      "grad_norm": 0.11259331554174423,
      "learning_rate": 9.519410993156522e-05,
      "loss": 1.9057,
      "step": 26304
    },
    {
      "epoch": 0.051969126229270156,
      "grad_norm": 0.11353620886802673,
      "learning_rate": 9.518776970720052e-05,
      "loss": 1.9012,
      "step": 26336
    },
    {
      "epoch": 0.0520322721906666,
      "grad_norm": 0.0989617332816124,
      "learning_rate": 9.518142948283583e-05,
      "loss": 1.9019,
      "step": 26368
    },
    {
      "epoch": 0.05209541815206304,
      "grad_norm": 0.10656723380088806,
      "learning_rate": 9.517508925847113e-05,
      "loss": 1.9125,
      "step": 26400
    },
    {
      "epoch": 0.05215856411345948,
      "grad_norm": 0.1056458055973053,
      "learning_rate": 9.516874903410645e-05,
      "loss": 1.9088,
      "step": 26432
    },
    {
      "epoch": 0.05222171007485592,
      "grad_norm": 0.11590812355279922,
      "learning_rate": 9.516240880974175e-05,
      "loss": 1.8987,
      "step": 26464
    },
    {
      "epoch": 0.05228485603625236,
      "grad_norm": 0.11630557477474213,
      "learning_rate": 9.515606858537706e-05,
      "loss": 1.9037,
      "step": 26496
    },
    {
      "epoch": 0.0523480019976488,
      "grad_norm": 0.1062152162194252,
      "learning_rate": 9.514972836101238e-05,
      "loss": 1.9047,
      "step": 26528
    },
    {
      "epoch": 0.05241114795904524,
      "grad_norm": 0.11292682588100433,
      "learning_rate": 9.51433881366477e-05,
      "loss": 1.9205,
      "step": 26560
    },
    {
      "epoch": 0.05247429392044168,
      "grad_norm": 0.11707049608230591,
      "learning_rate": 9.5137047912283e-05,
      "loss": 1.9085,
      "step": 26592
    },
    {
      "epoch": 0.05253743988183812,
      "grad_norm": 0.10360373556613922,
      "learning_rate": 9.513070768791831e-05,
      "loss": 1.8961,
      "step": 26624
    },
    {
      "epoch": 0.05260058584323456,
      "grad_norm": 0.10200996696949005,
      "learning_rate": 9.512436746355362e-05,
      "loss": 1.8976,
      "step": 26656
    },
    {
      "epoch": 0.052663731804631,
      "grad_norm": 0.09945542365312576,
      "learning_rate": 9.511802723918894e-05,
      "loss": 1.9122,
      "step": 26688
    },
    {
      "epoch": 0.05272687776602744,
      "grad_norm": 0.10186757892370224,
      "learning_rate": 9.511168701482425e-05,
      "loss": 1.9053,
      "step": 26720
    },
    {
      "epoch": 0.052790023727423876,
      "grad_norm": 0.11728771030902863,
      "learning_rate": 9.510534679045955e-05,
      "loss": 1.901,
      "step": 26752
    },
    {
      "epoch": 0.05285316968882032,
      "grad_norm": 0.10647526383399963,
      "learning_rate": 9.509900656609485e-05,
      "loss": 1.8989,
      "step": 26784
    },
    {
      "epoch": 0.05291631565021676,
      "grad_norm": 0.11470351368188858,
      "learning_rate": 9.509266634173017e-05,
      "loss": 1.8961,
      "step": 26816
    },
    {
      "epoch": 0.0529794616116132,
      "grad_norm": 0.11885487288236618,
      "learning_rate": 9.508632611736548e-05,
      "loss": 1.9105,
      "step": 26848
    },
    {
      "epoch": 0.05304260757300964,
      "grad_norm": 0.10806205868721008,
      "learning_rate": 9.507998589300078e-05,
      "loss": 1.8858,
      "step": 26880
    },
    {
      "epoch": 0.05310575353440608,
      "grad_norm": 0.10871472209692001,
      "learning_rate": 9.50736456686361e-05,
      "loss": 1.8853,
      "step": 26912
    },
    {
      "epoch": 0.05316889949580252,
      "grad_norm": 0.10162289440631866,
      "learning_rate": 9.506730544427141e-05,
      "loss": 1.8977,
      "step": 26944
    },
    {
      "epoch": 0.05323204545719896,
      "grad_norm": 0.11214316636323929,
      "learning_rate": 9.506096521990673e-05,
      "loss": 1.8947,
      "step": 26976
    },
    {
      "epoch": 0.0532951914185954,
      "grad_norm": 0.09722857177257538,
      "learning_rate": 9.505462499554203e-05,
      "loss": 1.8828,
      "step": 27008
    },
    {
      "epoch": 0.05335833737999184,
      "grad_norm": 0.10414082556962967,
      "learning_rate": 9.504828477117734e-05,
      "loss": 1.8992,
      "step": 27040
    },
    {
      "epoch": 0.05342148334138828,
      "grad_norm": 0.11235885322093964,
      "learning_rate": 9.504194454681266e-05,
      "loss": 1.8978,
      "step": 27072
    },
    {
      "epoch": 0.05348462930278472,
      "grad_norm": 0.10151505470275879,
      "learning_rate": 9.503560432244797e-05,
      "loss": 1.8798,
      "step": 27104
    },
    {
      "epoch": 0.05354777526418116,
      "grad_norm": 0.09740497916936874,
      "learning_rate": 9.502926409808327e-05,
      "loss": 1.8949,
      "step": 27136
    },
    {
      "epoch": 0.0536109212255776,
      "grad_norm": 0.10710079967975616,
      "learning_rate": 9.502292387371859e-05,
      "loss": 1.9016,
      "step": 27168
    },
    {
      "epoch": 0.05367406718697404,
      "grad_norm": 0.12225481122732162,
      "learning_rate": 9.501658364935389e-05,
      "loss": 1.8886,
      "step": 27200
    },
    {
      "epoch": 0.05373721314837048,
      "grad_norm": 0.10491971671581268,
      "learning_rate": 9.50102434249892e-05,
      "loss": 1.8994,
      "step": 27232
    },
    {
      "epoch": 0.053800359109766924,
      "grad_norm": 0.11765529215335846,
      "learning_rate": 9.50039032006245e-05,
      "loss": 1.9054,
      "step": 27264
    },
    {
      "epoch": 0.05386350507116336,
      "grad_norm": 0.10463310033082962,
      "learning_rate": 9.499756297625982e-05,
      "loss": 1.895,
      "step": 27296
    },
    {
      "epoch": 0.0539266510325598,
      "grad_norm": 0.10823880881071091,
      "learning_rate": 9.499122275189513e-05,
      "loss": 1.9042,
      "step": 27328
    },
    {
      "epoch": 0.05398979699395624,
      "grad_norm": 0.09078743308782578,
      "learning_rate": 9.498488252753045e-05,
      "loss": 1.8862,
      "step": 27360
    },
    {
      "epoch": 0.05405294295535268,
      "grad_norm": 0.11342493444681168,
      "learning_rate": 9.497854230316576e-05,
      "loss": 1.9163,
      "step": 27392
    },
    {
      "epoch": 0.05411608891674912,
      "grad_norm": 0.1100945919752121,
      "learning_rate": 9.497220207880106e-05,
      "loss": 1.9048,
      "step": 27424
    },
    {
      "epoch": 0.05417923487814556,
      "grad_norm": 0.11827767640352249,
      "learning_rate": 9.496586185443638e-05,
      "loss": 1.8797,
      "step": 27456
    },
    {
      "epoch": 0.054242380839542,
      "grad_norm": 0.11936810612678528,
      "learning_rate": 9.49595216300717e-05,
      "loss": 1.908,
      "step": 27488
    },
    {
      "epoch": 0.05430552680093844,
      "grad_norm": 0.10007096827030182,
      "learning_rate": 9.495318140570701e-05,
      "loss": 1.8897,
      "step": 27520
    },
    {
      "epoch": 0.05436867276233488,
      "grad_norm": 0.09597699344158173,
      "learning_rate": 9.494684118134231e-05,
      "loss": 1.8686,
      "step": 27552
    },
    {
      "epoch": 0.05443181872373132,
      "grad_norm": 0.11114552617073059,
      "learning_rate": 9.494050095697762e-05,
      "loss": 1.8969,
      "step": 27584
    },
    {
      "epoch": 0.05449496468512776,
      "grad_norm": 0.09870576858520508,
      "learning_rate": 9.493416073261293e-05,
      "loss": 1.9002,
      "step": 27616
    },
    {
      "epoch": 0.0545581106465242,
      "grad_norm": 0.10398736596107483,
      "learning_rate": 9.492782050824824e-05,
      "loss": 1.8967,
      "step": 27648
    },
    {
      "epoch": 0.054621256607920644,
      "grad_norm": 0.10817091166973114,
      "learning_rate": 9.492148028388354e-05,
      "loss": 1.9013,
      "step": 27680
    },
    {
      "epoch": 0.05468440256931708,
      "grad_norm": 0.11018930375576019,
      "learning_rate": 9.491514005951886e-05,
      "loss": 1.891,
      "step": 27712
    },
    {
      "epoch": 0.05474754853071352,
      "grad_norm": 0.11235178261995316,
      "learning_rate": 9.490879983515417e-05,
      "loss": 1.8873,
      "step": 27744
    },
    {
      "epoch": 0.054810694492109964,
      "grad_norm": 0.12674926221370697,
      "learning_rate": 9.490245961078948e-05,
      "loss": 1.8918,
      "step": 27776
    },
    {
      "epoch": 0.0548738404535064,
      "grad_norm": 0.10456321388483047,
      "learning_rate": 9.489611938642479e-05,
      "loss": 1.8827,
      "step": 27808
    },
    {
      "epoch": 0.05493698641490284,
      "grad_norm": 0.10634982585906982,
      "learning_rate": 9.48897791620601e-05,
      "loss": 1.9104,
      "step": 27840
    },
    {
      "epoch": 0.055000132376299284,
      "grad_norm": 0.10933510214090347,
      "learning_rate": 9.488343893769541e-05,
      "loss": 1.8867,
      "step": 27872
    },
    {
      "epoch": 0.05506327833769572,
      "grad_norm": 0.10465336591005325,
      "learning_rate": 9.487709871333073e-05,
      "loss": 1.8801,
      "step": 27904
    },
    {
      "epoch": 0.05512642429909216,
      "grad_norm": 0.11155618727207184,
      "learning_rate": 9.487075848896604e-05,
      "loss": 1.8933,
      "step": 27936
    },
    {
      "epoch": 0.0551895702604886,
      "grad_norm": 0.10860616713762283,
      "learning_rate": 9.486441826460134e-05,
      "loss": 1.9008,
      "step": 27968
    },
    {
      "epoch": 0.05525271622188504,
      "grad_norm": 0.11804484575986862,
      "learning_rate": 9.485807804023666e-05,
      "loss": 1.9105,
      "step": 28000
    },
    {
      "epoch": 0.05531586218328148,
      "grad_norm": 0.09635041654109955,
      "learning_rate": 9.485173781587196e-05,
      "loss": 1.8917,
      "step": 28032
    },
    {
      "epoch": 0.05537900814467792,
      "grad_norm": 0.10530688613653183,
      "learning_rate": 9.484539759150727e-05,
      "loss": 1.9062,
      "step": 28064
    },
    {
      "epoch": 0.055442154106074365,
      "grad_norm": 0.11746163666248322,
      "learning_rate": 9.483905736714258e-05,
      "loss": 1.8866,
      "step": 28096
    },
    {
      "epoch": 0.0555053000674708,
      "grad_norm": 0.11865773051977158,
      "learning_rate": 9.483271714277789e-05,
      "loss": 1.8951,
      "step": 28128
    },
    {
      "epoch": 0.05556844602886724,
      "grad_norm": 0.10170378535985947,
      "learning_rate": 9.48263769184132e-05,
      "loss": 1.8927,
      "step": 28160
    },
    {
      "epoch": 0.055631591990263685,
      "grad_norm": 0.10263235121965408,
      "learning_rate": 9.482003669404852e-05,
      "loss": 1.8792,
      "step": 28192
    },
    {
      "epoch": 0.05569473795166012,
      "grad_norm": 0.10327958315610886,
      "learning_rate": 9.481369646968382e-05,
      "loss": 1.8872,
      "step": 28224
    },
    {
      "epoch": 0.05575788391305656,
      "grad_norm": 0.10997182130813599,
      "learning_rate": 9.480735624531913e-05,
      "loss": 1.8944,
      "step": 28256
    },
    {
      "epoch": 0.055821029874453004,
      "grad_norm": 0.11584657430648804,
      "learning_rate": 9.480101602095445e-05,
      "loss": 1.8791,
      "step": 28288
    },
    {
      "epoch": 0.05588417583584944,
      "grad_norm": 0.10891609638929367,
      "learning_rate": 9.479467579658976e-05,
      "loss": 1.8887,
      "step": 28320
    },
    {
      "epoch": 0.05594732179724588,
      "grad_norm": 0.1165160983800888,
      "learning_rate": 9.478833557222507e-05,
      "loss": 1.8915,
      "step": 28352
    },
    {
      "epoch": 0.056010467758642324,
      "grad_norm": 0.09597717225551605,
      "learning_rate": 9.478199534786038e-05,
      "loss": 1.8957,
      "step": 28384
    },
    {
      "epoch": 0.05607361372003876,
      "grad_norm": 0.11800549924373627,
      "learning_rate": 9.47756551234957e-05,
      "loss": 1.8891,
      "step": 28416
    },
    {
      "epoch": 0.0561367596814352,
      "grad_norm": 0.11111108958721161,
      "learning_rate": 9.4769314899131e-05,
      "loss": 1.8804,
      "step": 28448
    },
    {
      "epoch": 0.056199905642831643,
      "grad_norm": 0.10995693504810333,
      "learning_rate": 9.47629746747663e-05,
      "loss": 1.8816,
      "step": 28480
    },
    {
      "epoch": 0.056263051604228086,
      "grad_norm": 0.11199373006820679,
      "learning_rate": 9.475663445040161e-05,
      "loss": 1.8914,
      "step": 28512
    },
    {
      "epoch": 0.05632619756562452,
      "grad_norm": 0.10174044966697693,
      "learning_rate": 9.475029422603693e-05,
      "loss": 1.8792,
      "step": 28544
    },
    {
      "epoch": 0.05638934352702096,
      "grad_norm": 0.11624758690595627,
      "learning_rate": 9.474395400167224e-05,
      "loss": 1.8874,
      "step": 28576
    },
    {
      "epoch": 0.056452489488417405,
      "grad_norm": 0.11394298076629639,
      "learning_rate": 9.473761377730755e-05,
      "loss": 1.8796,
      "step": 28608
    },
    {
      "epoch": 0.05651563544981384,
      "grad_norm": 0.11033052206039429,
      "learning_rate": 9.473127355294286e-05,
      "loss": 1.8838,
      "step": 28640
    },
    {
      "epoch": 0.05657878141121028,
      "grad_norm": 0.10413506627082825,
      "learning_rate": 9.472493332857817e-05,
      "loss": 1.8771,
      "step": 28672
    },
    {
      "epoch": 0.056641927372606725,
      "grad_norm": 0.09666626900434494,
      "learning_rate": 9.471859310421348e-05,
      "loss": 1.8729,
      "step": 28704
    },
    {
      "epoch": 0.05670507333400316,
      "grad_norm": 0.11766234785318375,
      "learning_rate": 9.47122528798488e-05,
      "loss": 1.8831,
      "step": 28736
    },
    {
      "epoch": 0.0567682192953996,
      "grad_norm": 0.11246274411678314,
      "learning_rate": 9.47059126554841e-05,
      "loss": 1.8915,
      "step": 28768
    },
    {
      "epoch": 0.056831365256796045,
      "grad_norm": 0.10445088893175125,
      "learning_rate": 9.469957243111941e-05,
      "loss": 1.8941,
      "step": 28800
    },
    {
      "epoch": 0.05689451121819248,
      "grad_norm": 0.10135059803724289,
      "learning_rate": 9.469323220675473e-05,
      "loss": 1.8897,
      "step": 28832
    },
    {
      "epoch": 0.05695765717958892,
      "grad_norm": 0.10946794599294662,
      "learning_rate": 9.468689198239003e-05,
      "loss": 1.8911,
      "step": 28864
    },
    {
      "epoch": 0.057020803140985364,
      "grad_norm": 0.11228682845830917,
      "learning_rate": 9.468055175802533e-05,
      "loss": 1.8852,
      "step": 28896
    },
    {
      "epoch": 0.057083949102381806,
      "grad_norm": 0.11112036556005478,
      "learning_rate": 9.467421153366065e-05,
      "loss": 1.9012,
      "step": 28928
    },
    {
      "epoch": 0.05714709506377824,
      "grad_norm": 0.10750123113393784,
      "learning_rate": 9.466787130929596e-05,
      "loss": 1.8819,
      "step": 28960
    },
    {
      "epoch": 0.057210241025174684,
      "grad_norm": 0.1083349660038948,
      "learning_rate": 9.466153108493128e-05,
      "loss": 1.8892,
      "step": 28992
    },
    {
      "epoch": 0.057273386986571126,
      "grad_norm": 0.09778472036123276,
      "learning_rate": 9.465519086056658e-05,
      "loss": 1.8827,
      "step": 29024
    },
    {
      "epoch": 0.05733653294796756,
      "grad_norm": 0.10571292042732239,
      "learning_rate": 9.464885063620189e-05,
      "loss": 1.8831,
      "step": 29056
    },
    {
      "epoch": 0.057399678909364,
      "grad_norm": 0.10290931165218353,
      "learning_rate": 9.46425104118372e-05,
      "loss": 1.8693,
      "step": 29088
    },
    {
      "epoch": 0.057462824870760446,
      "grad_norm": 0.11763163655996323,
      "learning_rate": 9.463617018747252e-05,
      "loss": 1.8697,
      "step": 29120
    },
    {
      "epoch": 0.05752597083215688,
      "grad_norm": 0.10329300910234451,
      "learning_rate": 9.462982996310782e-05,
      "loss": 1.8864,
      "step": 29152
    },
    {
      "epoch": 0.05758911679355332,
      "grad_norm": 0.11135107278823853,
      "learning_rate": 9.462348973874314e-05,
      "loss": 1.8804,
      "step": 29184
    },
    {
      "epoch": 0.057652262754949765,
      "grad_norm": 0.10401959717273712,
      "learning_rate": 9.461714951437845e-05,
      "loss": 1.892,
      "step": 29216
    },
    {
      "epoch": 0.0577154087163462,
      "grad_norm": 0.10142754018306732,
      "learning_rate": 9.461080929001376e-05,
      "loss": 1.884,
      "step": 29248
    },
    {
      "epoch": 0.05777855467774264,
      "grad_norm": 0.10389091074466705,
      "learning_rate": 9.460446906564907e-05,
      "loss": 1.8801,
      "step": 29280
    },
    {
      "epoch": 0.057841700639139085,
      "grad_norm": 0.10583070665597916,
      "learning_rate": 9.459812884128437e-05,
      "loss": 1.8735,
      "step": 29312
    },
    {
      "epoch": 0.05790484660053553,
      "grad_norm": 0.10557013750076294,
      "learning_rate": 9.459178861691968e-05,
      "loss": 1.8728,
      "step": 29344
    },
    {
      "epoch": 0.05796799256193196,
      "grad_norm": 0.10017932206392288,
      "learning_rate": 9.4585448392555e-05,
      "loss": 1.8893,
      "step": 29376
    },
    {
      "epoch": 0.058031138523328404,
      "grad_norm": 0.10164113342761993,
      "learning_rate": 9.457910816819031e-05,
      "loss": 1.8814,
      "step": 29408
    },
    {
      "epoch": 0.05809428448472485,
      "grad_norm": 0.10548080503940582,
      "learning_rate": 9.457276794382561e-05,
      "loss": 1.8938,
      "step": 29440
    },
    {
      "epoch": 0.05815743044612128,
      "grad_norm": 0.10909568518400192,
      "learning_rate": 9.456642771946093e-05,
      "loss": 1.8881,
      "step": 29472
    },
    {
      "epoch": 0.058220576407517724,
      "grad_norm": 0.1213250458240509,
      "learning_rate": 9.456008749509624e-05,
      "loss": 1.893,
      "step": 29504
    },
    {
      "epoch": 0.058283722368914166,
      "grad_norm": 0.10932382196187973,
      "learning_rate": 9.455374727073155e-05,
      "loss": 1.8733,
      "step": 29536
    },
    {
      "epoch": 0.0583468683303106,
      "grad_norm": 0.10439607501029968,
      "learning_rate": 9.454740704636686e-05,
      "loss": 1.8755,
      "step": 29568
    },
    {
      "epoch": 0.058410014291707044,
      "grad_norm": 0.09857314079999924,
      "learning_rate": 9.454106682200217e-05,
      "loss": 1.8853,
      "step": 29600
    },
    {
      "epoch": 0.058473160253103486,
      "grad_norm": 0.1132400706410408,
      "learning_rate": 9.453472659763748e-05,
      "loss": 1.8798,
      "step": 29632
    },
    {
      "epoch": 0.05853630621449992,
      "grad_norm": 0.100809745490551,
      "learning_rate": 9.452838637327279e-05,
      "loss": 1.8763,
      "step": 29664
    },
    {
      "epoch": 0.05859945217589636,
      "grad_norm": 0.10080935060977936,
      "learning_rate": 9.45220461489081e-05,
      "loss": 1.8827,
      "step": 29696
    },
    {
      "epoch": 0.058662598137292805,
      "grad_norm": 0.1064523234963417,
      "learning_rate": 9.45157059245434e-05,
      "loss": 1.8887,
      "step": 29728
    },
    {
      "epoch": 0.05872574409868925,
      "grad_norm": 0.11582209914922714,
      "learning_rate": 9.450936570017872e-05,
      "loss": 1.8871,
      "step": 29760
    },
    {
      "epoch": 0.05878889006008568,
      "grad_norm": 0.11003582924604416,
      "learning_rate": 9.450302547581403e-05,
      "loss": 1.8731,
      "step": 29792
    },
    {
      "epoch": 0.058852036021482125,
      "grad_norm": 0.10873007774353027,
      "learning_rate": 9.449668525144933e-05,
      "loss": 1.8671,
      "step": 29824
    },
    {
      "epoch": 0.05891518198287857,
      "grad_norm": 0.09862037003040314,
      "learning_rate": 9.449034502708465e-05,
      "loss": 1.8787,
      "step": 29856
    },
    {
      "epoch": 0.058978327944275,
      "grad_norm": 0.12812422215938568,
      "learning_rate": 9.448400480271996e-05,
      "loss": 1.8892,
      "step": 29888
    },
    {
      "epoch": 0.059041473905671445,
      "grad_norm": 0.10228711366653442,
      "learning_rate": 9.447766457835528e-05,
      "loss": 1.8613,
      "step": 29920
    },
    {
      "epoch": 0.05910461986706789,
      "grad_norm": 0.10565419495105743,
      "learning_rate": 9.447132435399059e-05,
      "loss": 1.8811,
      "step": 29952
    },
    {
      "epoch": 0.05916776582846432,
      "grad_norm": 0.11540845036506653,
      "learning_rate": 9.446498412962589e-05,
      "loss": 1.8816,
      "step": 29984
    },
    {
      "epoch": 0.059230911789860764,
      "grad_norm": 0.10832008719444275,
      "learning_rate": 9.44586439052612e-05,
      "loss": 1.881,
      "step": 30016
    },
    {
      "epoch": 0.059294057751257206,
      "grad_norm": 0.09858869016170502,
      "learning_rate": 9.445230368089652e-05,
      "loss": 1.8787,
      "step": 30048
    },
    {
      "epoch": 0.05935720371265364,
      "grad_norm": 0.10506797581911087,
      "learning_rate": 9.444596345653182e-05,
      "loss": 1.8671,
      "step": 30080
    },
    {
      "epoch": 0.059420349674050084,
      "grad_norm": 0.11254420876502991,
      "learning_rate": 9.443962323216714e-05,
      "loss": 1.8822,
      "step": 30112
    },
    {
      "epoch": 0.059483495635446526,
      "grad_norm": 0.10230351984500885,
      "learning_rate": 9.443328300780244e-05,
      "loss": 1.8785,
      "step": 30144
    },
    {
      "epoch": 0.05954664159684297,
      "grad_norm": 0.10950108617544174,
      "learning_rate": 9.442694278343775e-05,
      "loss": 1.8559,
      "step": 30176
    },
    {
      "epoch": 0.059609787558239404,
      "grad_norm": 0.1156979352235794,
      "learning_rate": 9.442060255907307e-05,
      "loss": 1.8648,
      "step": 30208
    },
    {
      "epoch": 0.059672933519635846,
      "grad_norm": 0.1161760613322258,
      "learning_rate": 9.441426233470837e-05,
      "loss": 1.8852,
      "step": 30240
    },
    {
      "epoch": 0.05973607948103229,
      "grad_norm": 0.11749791353940964,
      "learning_rate": 9.440792211034368e-05,
      "loss": 1.8618,
      "step": 30272
    },
    {
      "epoch": 0.05979922544242872,
      "grad_norm": 0.10605373978614807,
      "learning_rate": 9.4401581885979e-05,
      "loss": 1.8673,
      "step": 30304
    },
    {
      "epoch": 0.059862371403825165,
      "grad_norm": 0.0999060571193695,
      "learning_rate": 9.439524166161431e-05,
      "loss": 1.8675,
      "step": 30336
    },
    {
      "epoch": 0.05992551736522161,
      "grad_norm": 0.10648031532764435,
      "learning_rate": 9.438890143724961e-05,
      "loss": 1.8563,
      "step": 30368
    },
    {
      "epoch": 0.05998866332661804,
      "grad_norm": 0.11411280930042267,
      "learning_rate": 9.438256121288493e-05,
      "loss": 1.8631,
      "step": 30400
    },
    {
      "epoch": 0.060051809288014485,
      "grad_norm": 0.1032540425658226,
      "learning_rate": 9.437622098852024e-05,
      "loss": 1.8676,
      "step": 30432
    },
    {
      "epoch": 0.06011495524941093,
      "grad_norm": 0.1066264808177948,
      "learning_rate": 9.436988076415556e-05,
      "loss": 1.8756,
      "step": 30464
    },
    {
      "epoch": 0.06017810121080736,
      "grad_norm": 0.10596653074026108,
      "learning_rate": 9.436354053979086e-05,
      "loss": 1.8663,
      "step": 30496
    },
    {
      "epoch": 0.060241247172203805,
      "grad_norm": 0.1073792353272438,
      "learning_rate": 9.435720031542617e-05,
      "loss": 1.8723,
      "step": 30528
    },
    {
      "epoch": 0.06030439313360025,
      "grad_norm": 0.10586169362068176,
      "learning_rate": 9.435086009106147e-05,
      "loss": 1.8722,
      "step": 30560
    },
    {
      "epoch": 0.06036753909499669,
      "grad_norm": 0.10382824391126633,
      "learning_rate": 9.434451986669679e-05,
      "loss": 1.8712,
      "step": 30592
    },
    {
      "epoch": 0.060430685056393124,
      "grad_norm": 0.09873615950345993,
      "learning_rate": 9.43381796423321e-05,
      "loss": 1.8701,
      "step": 30624
    },
    {
      "epoch": 0.060493831017789566,
      "grad_norm": 0.11185143142938614,
      "learning_rate": 9.43318394179674e-05,
      "loss": 1.8588,
      "step": 30656
    },
    {
      "epoch": 0.06055697697918601,
      "grad_norm": 0.10457509011030197,
      "learning_rate": 9.432549919360272e-05,
      "loss": 1.874,
      "step": 30688
    },
    {
      "epoch": 0.060620122940582444,
      "grad_norm": 0.09991231560707092,
      "learning_rate": 9.431915896923803e-05,
      "loss": 1.8771,
      "step": 30720
    },
    {
      "epoch": 0.060683268901978886,
      "grad_norm": 0.10770970582962036,
      "learning_rate": 9.431281874487335e-05,
      "loss": 1.8782,
      "step": 30752
    },
    {
      "epoch": 0.06074641486337533,
      "grad_norm": 0.11045695841312408,
      "learning_rate": 9.430647852050865e-05,
      "loss": 1.8703,
      "step": 30784
    },
    {
      "epoch": 0.06080956082477176,
      "grad_norm": 0.10054302215576172,
      "learning_rate": 9.430013829614396e-05,
      "loss": 1.8633,
      "step": 30816
    },
    {
      "epoch": 0.060872706786168206,
      "grad_norm": 0.10831281542778015,
      "learning_rate": 9.429379807177928e-05,
      "loss": 1.8725,
      "step": 30848
    },
    {
      "epoch": 0.06093585274756465,
      "grad_norm": 0.10674191266298294,
      "learning_rate": 9.428745784741459e-05,
      "loss": 1.8616,
      "step": 30880
    },
    {
      "epoch": 0.06099899870896108,
      "grad_norm": 0.12082844227552414,
      "learning_rate": 9.428111762304989e-05,
      "loss": 1.8689,
      "step": 30912
    },
    {
      "epoch": 0.061062144670357525,
      "grad_norm": 0.10098271071910858,
      "learning_rate": 9.427477739868519e-05,
      "loss": 1.8662,
      "step": 30944
    },
    {
      "epoch": 0.06112529063175397,
      "grad_norm": 0.11558236181735992,
      "learning_rate": 9.426843717432051e-05,
      "loss": 1.8718,
      "step": 30976
    },
    {
      "epoch": 0.06118843659315041,
      "grad_norm": 0.09774626046419144,
      "learning_rate": 9.426209694995582e-05,
      "loss": 1.8606,
      "step": 31008
    },
    {
      "epoch": 0.061251582554546845,
      "grad_norm": 0.10142559558153152,
      "learning_rate": 9.425575672559112e-05,
      "loss": 1.8713,
      "step": 31040
    },
    {
      "epoch": 0.06131472851594329,
      "grad_norm": 0.12105502188205719,
      "learning_rate": 9.424941650122644e-05,
      "loss": 1.8643,
      "step": 31072
    },
    {
      "epoch": 0.06137787447733973,
      "grad_norm": 0.11434998363256454,
      "learning_rate": 9.424307627686175e-05,
      "loss": 1.8754,
      "step": 31104
    },
    {
      "epoch": 0.061441020438736164,
      "grad_norm": 0.106901615858078,
      "learning_rate": 9.423673605249707e-05,
      "loss": 1.8629,
      "step": 31136
    },
    {
      "epoch": 0.06150416640013261,
      "grad_norm": 0.10285452753305435,
      "learning_rate": 9.423039582813237e-05,
      "loss": 1.8641,
      "step": 31168
    },
    {
      "epoch": 0.06156731236152905,
      "grad_norm": 0.11215592175722122,
      "learning_rate": 9.422405560376768e-05,
      "loss": 1.8532,
      "step": 31200
    },
    {
      "epoch": 0.061630458322925484,
      "grad_norm": 0.10553357750177383,
      "learning_rate": 9.4217715379403e-05,
      "loss": 1.862,
      "step": 31232
    },
    {
      "epoch": 0.061693604284321926,
      "grad_norm": 0.10379686951637268,
      "learning_rate": 9.421137515503831e-05,
      "loss": 1.8578,
      "step": 31264
    },
    {
      "epoch": 0.06175675024571837,
      "grad_norm": 0.1035488024353981,
      "learning_rate": 9.420503493067363e-05,
      "loss": 1.8621,
      "step": 31296
    },
    {
      "epoch": 0.061819896207114804,
      "grad_norm": 0.10524856299161911,
      "learning_rate": 9.419869470630893e-05,
      "loss": 1.8707,
      "step": 31328
    },
    {
      "epoch": 0.061883042168511246,
      "grad_norm": 0.1163095161318779,
      "learning_rate": 9.419235448194423e-05,
      "loss": 1.8554,
      "step": 31360
    },
    {
      "epoch": 0.06194618812990769,
      "grad_norm": 0.1016009971499443,
      "learning_rate": 9.418601425757954e-05,
      "loss": 1.8507,
      "step": 31392
    },
    {
      "epoch": 0.06200933409130413,
      "grad_norm": 0.11709136515855789,
      "learning_rate": 9.417967403321486e-05,
      "loss": 1.8746,
      "step": 31424
    },
    {
      "epoch": 0.062072480052700565,
      "grad_norm": 0.11425372958183289,
      "learning_rate": 9.417333380885016e-05,
      "loss": 1.8665,
      "step": 31456
    },
    {
      "epoch": 0.06213562601409701,
      "grad_norm": 0.11852588504552841,
      "learning_rate": 9.416699358448547e-05,
      "loss": 1.8692,
      "step": 31488
    },
    {
      "epoch": 0.06219877197549345,
      "grad_norm": 0.09975966066122055,
      "learning_rate": 9.416065336012079e-05,
      "loss": 1.841,
      "step": 31520
    },
    {
      "epoch": 0.062261917936889885,
      "grad_norm": 0.10193809121847153,
      "learning_rate": 9.41543131357561e-05,
      "loss": 1.8695,
      "step": 31552
    },
    {
      "epoch": 0.06232506389828633,
      "grad_norm": 0.10371892154216766,
      "learning_rate": 9.41479729113914e-05,
      "loss": 1.8671,
      "step": 31584
    },
    {
      "epoch": 0.06238820985968277,
      "grad_norm": 0.1013907864689827,
      "learning_rate": 9.414163268702672e-05,
      "loss": 1.856,
      "step": 31616
    },
    {
      "epoch": 0.062451355821079205,
      "grad_norm": 0.11418340355157852,
      "learning_rate": 9.413529246266203e-05,
      "loss": 1.845,
      "step": 31648
    },
    {
      "epoch": 0.06251450178247565,
      "grad_norm": 0.10536934435367584,
      "learning_rate": 9.412895223829735e-05,
      "loss": 1.849,
      "step": 31680
    },
    {
      "epoch": 0.06257764774387209,
      "grad_norm": 0.11544335633516312,
      "learning_rate": 9.412261201393265e-05,
      "loss": 1.8558,
      "step": 31712
    },
    {
      "epoch": 0.06264079370526852,
      "grad_norm": 0.10208461433649063,
      "learning_rate": 9.411627178956796e-05,
      "loss": 1.8505,
      "step": 31744
    },
    {
      "epoch": 0.06270393966666497,
      "grad_norm": 0.11636324226856232,
      "learning_rate": 9.410993156520326e-05,
      "loss": 1.8476,
      "step": 31776
    },
    {
      "epoch": 0.06276708562806141,
      "grad_norm": 0.11715982109308243,
      "learning_rate": 9.410359134083858e-05,
      "loss": 1.8606,
      "step": 31808
    },
    {
      "epoch": 0.06283023158945784,
      "grad_norm": 0.10133142024278641,
      "learning_rate": 9.409725111647388e-05,
      "loss": 1.854,
      "step": 31840
    },
    {
      "epoch": 0.06289337755085429,
      "grad_norm": 0.10231012105941772,
      "learning_rate": 9.409091089210919e-05,
      "loss": 1.849,
      "step": 31872
    },
    {
      "epoch": 0.06295652351225073,
      "grad_norm": 0.10346190631389618,
      "learning_rate": 9.408457066774451e-05,
      "loss": 1.8464,
      "step": 31904
    },
    {
      "epoch": 0.06301966947364716,
      "grad_norm": 0.11396406590938568,
      "learning_rate": 9.407823044337982e-05,
      "loss": 1.8518,
      "step": 31936
    },
    {
      "epoch": 0.06308281543504361,
      "grad_norm": 0.10464853793382645,
      "learning_rate": 9.407189021901514e-05,
      "loss": 1.8655,
      "step": 31968
    },
    {
      "epoch": 0.06314596139644005,
      "grad_norm": 0.0973316878080368,
      "learning_rate": 9.406554999465044e-05,
      "loss": 1.8616,
      "step": 32000
    },
    {
      "epoch": 0.06320910735783648,
      "grad_norm": 0.10807619988918304,
      "learning_rate": 9.405920977028575e-05,
      "loss": 1.8431,
      "step": 32032
    },
    {
      "epoch": 0.06327225331923293,
      "grad_norm": 0.10671120136976242,
      "learning_rate": 9.405286954592107e-05,
      "loss": 1.8544,
      "step": 32064
    },
    {
      "epoch": 0.06333539928062937,
      "grad_norm": 0.12205708026885986,
      "learning_rate": 9.404652932155638e-05,
      "loss": 1.8684,
      "step": 32096
    },
    {
      "epoch": 0.0633985452420258,
      "grad_norm": 0.1130564957857132,
      "learning_rate": 9.404018909719168e-05,
      "loss": 1.8424,
      "step": 32128
    },
    {
      "epoch": 0.06346169120342225,
      "grad_norm": 0.09570758044719696,
      "learning_rate": 9.4033848872827e-05,
      "loss": 1.8601,
      "step": 32160
    },
    {
      "epoch": 0.06352483716481869,
      "grad_norm": 0.11315812915563583,
      "learning_rate": 9.40275086484623e-05,
      "loss": 1.8599,
      "step": 32192
    },
    {
      "epoch": 0.06358798312621512,
      "grad_norm": 0.0973125696182251,
      "learning_rate": 9.402116842409761e-05,
      "loss": 1.8553,
      "step": 32224
    },
    {
      "epoch": 0.06365112908761157,
      "grad_norm": 0.10498806834220886,
      "learning_rate": 9.401482819973291e-05,
      "loss": 1.8643,
      "step": 32256
    },
    {
      "epoch": 0.063714275049008,
      "grad_norm": 0.10018099844455719,
      "learning_rate": 9.400848797536823e-05,
      "loss": 1.823,
      "step": 32288
    },
    {
      "epoch": 0.06377742101040444,
      "grad_norm": 0.10291516780853271,
      "learning_rate": 9.400214775100354e-05,
      "loss": 1.8516,
      "step": 32320
    },
    {
      "epoch": 0.06384056697180089,
      "grad_norm": 0.10548707097768784,
      "learning_rate": 9.399580752663886e-05,
      "loss": 1.8485,
      "step": 32352
    },
    {
      "epoch": 0.06390371293319733,
      "grad_norm": 0.11257361620664597,
      "learning_rate": 9.398946730227416e-05,
      "loss": 1.8574,
      "step": 32384
    },
    {
      "epoch": 0.06396685889459376,
      "grad_norm": 0.10809215158224106,
      "learning_rate": 9.398312707790947e-05,
      "loss": 1.8471,
      "step": 32416
    },
    {
      "epoch": 0.06403000485599021,
      "grad_norm": 0.10717517882585526,
      "learning_rate": 9.397678685354479e-05,
      "loss": 1.8476,
      "step": 32448
    },
    {
      "epoch": 0.06409315081738665,
      "grad_norm": 0.10977809876203537,
      "learning_rate": 9.39704466291801e-05,
      "loss": 1.857,
      "step": 32480
    },
    {
      "epoch": 0.0641562967787831,
      "grad_norm": 0.1052245944738388,
      "learning_rate": 9.39641064048154e-05,
      "loss": 1.8523,
      "step": 32512
    },
    {
      "epoch": 0.06421944274017953,
      "grad_norm": 0.10304667800664902,
      "learning_rate": 9.395776618045072e-05,
      "loss": 1.8453,
      "step": 32544
    },
    {
      "epoch": 0.06428258870157597,
      "grad_norm": 0.10384541004896164,
      "learning_rate": 9.395142595608603e-05,
      "loss": 1.87,
      "step": 32576
    },
    {
      "epoch": 0.06434573466297241,
      "grad_norm": 0.1081276535987854,
      "learning_rate": 9.394508573172133e-05,
      "loss": 1.8541,
      "step": 32608
    },
    {
      "epoch": 0.06440888062436885,
      "grad_norm": 0.11084219068288803,
      "learning_rate": 9.393874550735665e-05,
      "loss": 1.8466,
      "step": 32640
    },
    {
      "epoch": 0.06447202658576529,
      "grad_norm": 0.09971606731414795,
      "learning_rate": 9.393240528299195e-05,
      "loss": 1.8521,
      "step": 32672
    },
    {
      "epoch": 0.06453517254716173,
      "grad_norm": 0.09828321635723114,
      "learning_rate": 9.392606505862726e-05,
      "loss": 1.8525,
      "step": 32704
    },
    {
      "epoch": 0.06459831850855817,
      "grad_norm": 0.10448254644870758,
      "learning_rate": 9.391972483426258e-05,
      "loss": 1.8488,
      "step": 32736
    },
    {
      "epoch": 0.0646614644699546,
      "grad_norm": 0.11569347232580185,
      "learning_rate": 9.391338460989789e-05,
      "loss": 1.8632,
      "step": 32768
    },
    {
      "epoch": 0.06472461043135105,
      "grad_norm": 0.11592109501361847,
      "learning_rate": 9.390704438553319e-05,
      "loss": 1.8493,
      "step": 32800
    },
    {
      "epoch": 0.06478775639274749,
      "grad_norm": 0.10566107928752899,
      "learning_rate": 9.390070416116851e-05,
      "loss": 1.8402,
      "step": 32832
    },
    {
      "epoch": 0.06485090235414392,
      "grad_norm": 0.11042334884405136,
      "learning_rate": 9.389436393680382e-05,
      "loss": 1.8508,
      "step": 32864
    },
    {
      "epoch": 0.06491404831554037,
      "grad_norm": 0.10361830145120621,
      "learning_rate": 9.388802371243914e-05,
      "loss": 1.859,
      "step": 32896
    },
    {
      "epoch": 0.06497719427693681,
      "grad_norm": 0.10170776396989822,
      "learning_rate": 9.388168348807444e-05,
      "loss": 1.8428,
      "step": 32928
    },
    {
      "epoch": 0.06504034023833324,
      "grad_norm": 0.10321789234876633,
      "learning_rate": 9.387534326370975e-05,
      "loss": 1.8575,
      "step": 32960
    },
    {
      "epoch": 0.0651034861997297,
      "grad_norm": 0.10872825235128403,
      "learning_rate": 9.386900303934507e-05,
      "loss": 1.8482,
      "step": 32992
    },
    {
      "epoch": 0.06516663216112613,
      "grad_norm": 0.09724610298871994,
      "learning_rate": 9.386266281498037e-05,
      "loss": 1.837,
      "step": 33024
    },
    {
      "epoch": 0.06522977812252256,
      "grad_norm": 0.11104436218738556,
      "learning_rate": 9.385632259061567e-05,
      "loss": 1.8588,
      "step": 33056
    },
    {
      "epoch": 0.06529292408391901,
      "grad_norm": 0.10021993517875671,
      "learning_rate": 9.384998236625098e-05,
      "loss": 1.8586,
      "step": 33088
    },
    {
      "epoch": 0.06535607004531545,
      "grad_norm": 0.10966669023036957,
      "learning_rate": 9.38436421418863e-05,
      "loss": 1.8523,
      "step": 33120
    },
    {
      "epoch": 0.06541921600671188,
      "grad_norm": 0.11023969948291779,
      "learning_rate": 9.383730191752161e-05,
      "loss": 1.8482,
      "step": 33152
    },
    {
      "epoch": 0.06548236196810833,
      "grad_norm": 0.09609818458557129,
      "learning_rate": 9.383096169315691e-05,
      "loss": 1.848,
      "step": 33184
    },
    {
      "epoch": 0.06554550792950477,
      "grad_norm": 0.10640505701303482,
      "learning_rate": 9.382462146879223e-05,
      "loss": 1.8451,
      "step": 33216
    },
    {
      "epoch": 0.0656086538909012,
      "grad_norm": 0.1084061861038208,
      "learning_rate": 9.381828124442754e-05,
      "loss": 1.8529,
      "step": 33248
    },
    {
      "epoch": 0.06567179985229765,
      "grad_norm": 0.10528600960969925,
      "learning_rate": 9.381194102006286e-05,
      "loss": 1.8577,
      "step": 33280
    },
    {
      "epoch": 0.06573494581369409,
      "grad_norm": 0.10225199162960052,
      "learning_rate": 9.380560079569817e-05,
      "loss": 1.8445,
      "step": 33312
    },
    {
      "epoch": 0.06579809177509054,
      "grad_norm": 0.11053970456123352,
      "learning_rate": 9.379926057133347e-05,
      "loss": 1.8526,
      "step": 33344
    },
    {
      "epoch": 0.06586123773648697,
      "grad_norm": 0.10372111201286316,
      "learning_rate": 9.379292034696879e-05,
      "loss": 1.8407,
      "step": 33376
    },
    {
      "epoch": 0.0659243836978834,
      "grad_norm": 0.09763038903474808,
      "learning_rate": 9.37865801226041e-05,
      "loss": 1.841,
      "step": 33408
    },
    {
      "epoch": 0.06598752965927986,
      "grad_norm": 0.11436961591243744,
      "learning_rate": 9.37802398982394e-05,
      "loss": 1.8472,
      "step": 33440
    },
    {
      "epoch": 0.06605067562067629,
      "grad_norm": 0.10392463952302933,
      "learning_rate": 9.37738996738747e-05,
      "loss": 1.8603,
      "step": 33472
    },
    {
      "epoch": 0.06611382158207273,
      "grad_norm": 0.10825751721858978,
      "learning_rate": 9.376755944951002e-05,
      "loss": 1.8329,
      "step": 33504
    },
    {
      "epoch": 0.06617696754346918,
      "grad_norm": 0.1062132716178894,
      "learning_rate": 9.376121922514533e-05,
      "loss": 1.8494,
      "step": 33536
    },
    {
      "epoch": 0.06624011350486561,
      "grad_norm": 0.12203536182641983,
      "learning_rate": 9.375487900078065e-05,
      "loss": 1.834,
      "step": 33568
    },
    {
      "epoch": 0.06630325946626205,
      "grad_norm": 0.11732408404350281,
      "learning_rate": 9.374853877641595e-05,
      "loss": 1.835,
      "step": 33600
    },
    {
      "epoch": 0.0663664054276585,
      "grad_norm": 0.12333063036203384,
      "learning_rate": 9.374219855205126e-05,
      "loss": 1.8399,
      "step": 33632
    },
    {
      "epoch": 0.06642955138905493,
      "grad_norm": 0.1118704080581665,
      "learning_rate": 9.373585832768658e-05,
      "loss": 1.8354,
      "step": 33664
    },
    {
      "epoch": 0.06649269735045137,
      "grad_norm": 0.09639861434698105,
      "learning_rate": 9.372951810332189e-05,
      "loss": 1.8518,
      "step": 33696
    },
    {
      "epoch": 0.06655584331184781,
      "grad_norm": 0.1065998300909996,
      "learning_rate": 9.37231778789572e-05,
      "loss": 1.8421,
      "step": 33728
    },
    {
      "epoch": 0.06661898927324425,
      "grad_norm": 0.103479765355587,
      "learning_rate": 9.371683765459251e-05,
      "loss": 1.8475,
      "step": 33760
    },
    {
      "epoch": 0.06668213523464069,
      "grad_norm": 0.1073637381196022,
      "learning_rate": 9.371049743022782e-05,
      "loss": 1.8495,
      "step": 33792
    },
    {
      "epoch": 0.06674528119603713,
      "grad_norm": 0.10078073292970657,
      "learning_rate": 9.370415720586312e-05,
      "loss": 1.8184,
      "step": 33824
    },
    {
      "epoch": 0.06680842715743357,
      "grad_norm": 0.11597784608602524,
      "learning_rate": 9.369781698149844e-05,
      "loss": 1.8486,
      "step": 33856
    },
    {
      "epoch": 0.06687157311883,
      "grad_norm": 0.10159990191459656,
      "learning_rate": 9.369147675713374e-05,
      "loss": 1.8371,
      "step": 33888
    },
    {
      "epoch": 0.06693471908022645,
      "grad_norm": 0.10640428215265274,
      "learning_rate": 9.368513653276905e-05,
      "loss": 1.8417,
      "step": 33920
    },
    {
      "epoch": 0.06699786504162289,
      "grad_norm": 0.09588663280010223,
      "learning_rate": 9.367879630840437e-05,
      "loss": 1.8506,
      "step": 33952
    },
    {
      "epoch": 0.06706101100301932,
      "grad_norm": 0.10144439339637756,
      "learning_rate": 9.367245608403968e-05,
      "loss": 1.8284,
      "step": 33984
    },
    {
      "epoch": 0.06712415696441577,
      "grad_norm": 0.09015002846717834,
      "learning_rate": 9.366611585967498e-05,
      "loss": 1.8413,
      "step": 34016
    },
    {
      "epoch": 0.06718730292581221,
      "grad_norm": 0.10386817902326584,
      "learning_rate": 9.36597756353103e-05,
      "loss": 1.8349,
      "step": 34048
    },
    {
      "epoch": 0.06725044888720864,
      "grad_norm": 0.10589347034692764,
      "learning_rate": 9.365343541094561e-05,
      "loss": 1.8433,
      "step": 34080
    },
    {
      "epoch": 0.0673135948486051,
      "grad_norm": 0.1025700718164444,
      "learning_rate": 9.364709518658093e-05,
      "loss": 1.846,
      "step": 34112
    },
    {
      "epoch": 0.06737674081000153,
      "grad_norm": 0.09602469950914383,
      "learning_rate": 9.364075496221623e-05,
      "loss": 1.8336,
      "step": 34144
    },
    {
      "epoch": 0.06743988677139798,
      "grad_norm": 0.10800101608037949,
      "learning_rate": 9.363441473785154e-05,
      "loss": 1.8537,
      "step": 34176
    },
    {
      "epoch": 0.06750303273279441,
      "grad_norm": 0.09845973551273346,
      "learning_rate": 9.362807451348686e-05,
      "loss": 1.8374,
      "step": 34208
    },
    {
      "epoch": 0.06756617869419085,
      "grad_norm": 0.10679402947425842,
      "learning_rate": 9.362173428912216e-05,
      "loss": 1.8411,
      "step": 34240
    },
    {
      "epoch": 0.0676293246555873,
      "grad_norm": 0.10255902260541916,
      "learning_rate": 9.361539406475747e-05,
      "loss": 1.853,
      "step": 34272
    },
    {
      "epoch": 0.06769247061698373,
      "grad_norm": 0.09579630196094513,
      "learning_rate": 9.360905384039277e-05,
      "loss": 1.8328,
      "step": 34304
    },
    {
      "epoch": 0.06775561657838017,
      "grad_norm": 0.10283277183771133,
      "learning_rate": 9.360271361602809e-05,
      "loss": 1.8425,
      "step": 34336
    },
    {
      "epoch": 0.06781876253977662,
      "grad_norm": 0.11232533305883408,
      "learning_rate": 9.35963733916634e-05,
      "loss": 1.8424,
      "step": 34368
    },
    {
      "epoch": 0.06788190850117305,
      "grad_norm": 0.10461856424808502,
      "learning_rate": 9.35900331672987e-05,
      "loss": 1.8263,
      "step": 34400
    },
    {
      "epoch": 0.06794505446256949,
      "grad_norm": 0.10416267812252045,
      "learning_rate": 9.358369294293402e-05,
      "loss": 1.828,
      "step": 34432
    },
    {
      "epoch": 0.06800820042396594,
      "grad_norm": 0.11975955218076706,
      "learning_rate": 9.357735271856933e-05,
      "loss": 1.8534,
      "step": 34464
    },
    {
      "epoch": 0.06807134638536237,
      "grad_norm": 0.10487686097621918,
      "learning_rate": 9.357101249420465e-05,
      "loss": 1.8388,
      "step": 34496
    },
    {
      "epoch": 0.06813449234675881,
      "grad_norm": 0.11208721995353699,
      "learning_rate": 9.356467226983995e-05,
      "loss": 1.8204,
      "step": 34528
    },
    {
      "epoch": 0.06819763830815526,
      "grad_norm": 0.11162339150905609,
      "learning_rate": 9.355833204547526e-05,
      "loss": 1.8441,
      "step": 34560
    },
    {
      "epoch": 0.06826078426955169,
      "grad_norm": 0.10376334935426712,
      "learning_rate": 9.355199182111058e-05,
      "loss": 1.8387,
      "step": 34592
    },
    {
      "epoch": 0.06832393023094813,
      "grad_norm": 0.09627466648817062,
      "learning_rate": 9.354565159674589e-05,
      "loss": 1.8387,
      "step": 34624
    },
    {
      "epoch": 0.06838707619234458,
      "grad_norm": 0.10894633084535599,
      "learning_rate": 9.35393113723812e-05,
      "loss": 1.8272,
      "step": 34656
    },
    {
      "epoch": 0.06845022215374101,
      "grad_norm": 0.09806417673826218,
      "learning_rate": 9.353297114801651e-05,
      "loss": 1.8304,
      "step": 34688
    },
    {
      "epoch": 0.06851336811513745,
      "grad_norm": 0.10127557814121246,
      "learning_rate": 9.352663092365181e-05,
      "loss": 1.8213,
      "step": 34720
    },
    {
      "epoch": 0.0685765140765339,
      "grad_norm": 0.13220052421092987,
      "learning_rate": 9.352029069928712e-05,
      "loss": 1.8507,
      "step": 34752
    },
    {
      "epoch": 0.06863966003793033,
      "grad_norm": 0.11304617673158646,
      "learning_rate": 9.351395047492244e-05,
      "loss": 1.8452,
      "step": 34784
    },
    {
      "epoch": 0.06870280599932677,
      "grad_norm": 0.10286716371774673,
      "learning_rate": 9.350761025055774e-05,
      "loss": 1.826,
      "step": 34816
    },
    {
      "epoch": 0.06876595196072322,
      "grad_norm": 0.1059626117348671,
      "learning_rate": 9.350127002619305e-05,
      "loss": 1.8255,
      "step": 34848
    },
    {
      "epoch": 0.06882909792211965,
      "grad_norm": 0.11121994256973267,
      "learning_rate": 9.349492980182837e-05,
      "loss": 1.8272,
      "step": 34880
    },
    {
      "epoch": 0.06889224388351609,
      "grad_norm": 0.09956642240285873,
      "learning_rate": 9.348858957746368e-05,
      "loss": 1.8269,
      "step": 34912
    },
    {
      "epoch": 0.06895538984491253,
      "grad_norm": 0.10039377212524414,
      "learning_rate": 9.348224935309898e-05,
      "loss": 1.841,
      "step": 34944
    },
    {
      "epoch": 0.06901853580630897,
      "grad_norm": 0.11147228628396988,
      "learning_rate": 9.34759091287343e-05,
      "loss": 1.8451,
      "step": 34976
    },
    {
      "epoch": 0.06908168176770542,
      "grad_norm": 0.1100914403796196,
      "learning_rate": 9.346956890436961e-05,
      "loss": 1.8242,
      "step": 35008
    },
    {
      "epoch": 0.06914482772910185,
      "grad_norm": 0.10677902400493622,
      "learning_rate": 9.346322868000493e-05,
      "loss": 1.8255,
      "step": 35040
    },
    {
      "epoch": 0.06920797369049829,
      "grad_norm": 0.10005491971969604,
      "learning_rate": 9.345688845564023e-05,
      "loss": 1.8378,
      "step": 35072
    },
    {
      "epoch": 0.06927111965189474,
      "grad_norm": 0.10176543146371841,
      "learning_rate": 9.345054823127553e-05,
      "loss": 1.8412,
      "step": 35104
    },
    {
      "epoch": 0.06933426561329117,
      "grad_norm": 0.11025370657444,
      "learning_rate": 9.344420800691084e-05,
      "loss": 1.832,
      "step": 35136
    },
    {
      "epoch": 0.06939741157468761,
      "grad_norm": 0.09753858298063278,
      "learning_rate": 9.343786778254616e-05,
      "loss": 1.8441,
      "step": 35168
    },
    {
      "epoch": 0.06946055753608406,
      "grad_norm": 0.13318485021591187,
      "learning_rate": 9.343152755818146e-05,
      "loss": 1.8396,
      "step": 35200
    },
    {
      "epoch": 0.0695237034974805,
      "grad_norm": 0.10949849337339401,
      "learning_rate": 9.342518733381677e-05,
      "loss": 1.8351,
      "step": 35232
    },
    {
      "epoch": 0.06958684945887693,
      "grad_norm": 0.10659553110599518,
      "learning_rate": 9.341884710945209e-05,
      "loss": 1.8205,
      "step": 35264
    },
    {
      "epoch": 0.06964999542027338,
      "grad_norm": 0.11543211340904236,
      "learning_rate": 9.34125068850874e-05,
      "loss": 1.8368,
      "step": 35296
    },
    {
      "epoch": 0.06971314138166981,
      "grad_norm": 0.10537153482437134,
      "learning_rate": 9.340616666072272e-05,
      "loss": 1.8328,
      "step": 35328
    },
    {
      "epoch": 0.06977628734306625,
      "grad_norm": 0.10511999577283859,
      "learning_rate": 9.339982643635802e-05,
      "loss": 1.8268,
      "step": 35360
    },
    {
      "epoch": 0.0698394333044627,
      "grad_norm": 0.10195047408342361,
      "learning_rate": 9.339348621199333e-05,
      "loss": 1.8243,
      "step": 35392
    },
    {
      "epoch": 0.06990257926585913,
      "grad_norm": 0.10528641939163208,
      "learning_rate": 9.338714598762865e-05,
      "loss": 1.8472,
      "step": 35424
    },
    {
      "epoch": 0.06996572522725557,
      "grad_norm": 0.10591515153646469,
      "learning_rate": 9.338080576326396e-05,
      "loss": 1.823,
      "step": 35456
    },
    {
      "epoch": 0.07002887118865202,
      "grad_norm": 0.10703007131814957,
      "learning_rate": 9.337446553889926e-05,
      "loss": 1.8035,
      "step": 35488
    },
    {
      "epoch": 0.07009201715004845,
      "grad_norm": 0.10312335938215256,
      "learning_rate": 9.336812531453456e-05,
      "loss": 1.8261,
      "step": 35520
    },
    {
      "epoch": 0.07015516311144489,
      "grad_norm": 0.13176143169403076,
      "learning_rate": 9.336178509016988e-05,
      "loss": 1.8245,
      "step": 35552
    },
    {
      "epoch": 0.07021830907284134,
      "grad_norm": 0.09669128805398941,
      "learning_rate": 9.33554448658052e-05,
      "loss": 1.8196,
      "step": 35584
    },
    {
      "epoch": 0.07028145503423777,
      "grad_norm": 0.09845513850450516,
      "learning_rate": 9.33491046414405e-05,
      "loss": 1.8271,
      "step": 35616
    },
    {
      "epoch": 0.07034460099563421,
      "grad_norm": 0.10833404213190079,
      "learning_rate": 9.334276441707581e-05,
      "loss": 1.8224,
      "step": 35648
    },
    {
      "epoch": 0.07040774695703066,
      "grad_norm": 0.11137785017490387,
      "learning_rate": 9.333642419271112e-05,
      "loss": 1.8191,
      "step": 35680
    },
    {
      "epoch": 0.07047089291842709,
      "grad_norm": 0.11151645332574844,
      "learning_rate": 9.333008396834644e-05,
      "loss": 1.8361,
      "step": 35712
    },
    {
      "epoch": 0.07053403887982353,
      "grad_norm": 0.09397818893194199,
      "learning_rate": 9.332374374398174e-05,
      "loss": 1.8441,
      "step": 35744
    },
    {
      "epoch": 0.07059718484121998,
      "grad_norm": 0.11775724589824677,
      "learning_rate": 9.331740351961705e-05,
      "loss": 1.8303,
      "step": 35776
    },
    {
      "epoch": 0.07066033080261641,
      "grad_norm": 0.11066251248121262,
      "learning_rate": 9.331106329525237e-05,
      "loss": 1.8267,
      "step": 35808
    },
    {
      "epoch": 0.07072347676401286,
      "grad_norm": 0.11496420949697495,
      "learning_rate": 9.330472307088768e-05,
      "loss": 1.8324,
      "step": 35840
    },
    {
      "epoch": 0.0707866227254093,
      "grad_norm": 0.10406922549009323,
      "learning_rate": 9.329838284652298e-05,
      "loss": 1.8273,
      "step": 35872
    },
    {
      "epoch": 0.07084976868680573,
      "grad_norm": 0.09711103141307831,
      "learning_rate": 9.32920426221583e-05,
      "loss": 1.8346,
      "step": 35904
    },
    {
      "epoch": 0.07091291464820218,
      "grad_norm": 0.11311649531126022,
      "learning_rate": 9.32857023977936e-05,
      "loss": 1.8224,
      "step": 35936
    },
    {
      "epoch": 0.07097606060959862,
      "grad_norm": 0.10964278876781464,
      "learning_rate": 9.327936217342891e-05,
      "loss": 1.8288,
      "step": 35968
    },
    {
      "epoch": 0.07103920657099505,
      "grad_norm": 0.09969433397054672,
      "learning_rate": 9.327302194906423e-05,
      "loss": 1.8556,
      "step": 36000
    },
    {
      "epoch": 0.0711023525323915,
      "grad_norm": 0.10349023342132568,
      "learning_rate": 9.326668172469953e-05,
      "loss": 1.8193,
      "step": 36032
    },
    {
      "epoch": 0.07116549849378793,
      "grad_norm": 0.10405611246824265,
      "learning_rate": 9.326034150033484e-05,
      "loss": 1.8229,
      "step": 36064
    },
    {
      "epoch": 0.07122864445518437,
      "grad_norm": 0.09794848412275314,
      "learning_rate": 9.325400127597016e-05,
      "loss": 1.8301,
      "step": 36096
    },
    {
      "epoch": 0.07129179041658082,
      "grad_norm": 0.09608778357505798,
      "learning_rate": 9.324766105160547e-05,
      "loss": 1.823,
      "step": 36128
    },
    {
      "epoch": 0.07135493637797725,
      "grad_norm": 0.10224819928407669,
      "learning_rate": 9.324132082724077e-05,
      "loss": 1.8375,
      "step": 36160
    },
    {
      "epoch": 0.07141808233937369,
      "grad_norm": 0.10872190445661545,
      "learning_rate": 9.323498060287609e-05,
      "loss": 1.8158,
      "step": 36192
    },
    {
      "epoch": 0.07148122830077014,
      "grad_norm": 0.11490211635828018,
      "learning_rate": 9.32286403785114e-05,
      "loss": 1.8213,
      "step": 36224
    },
    {
      "epoch": 0.07154437426216657,
      "grad_norm": 0.09908118844032288,
      "learning_rate": 9.322230015414672e-05,
      "loss": 1.8214,
      "step": 36256
    },
    {
      "epoch": 0.07160752022356301,
      "grad_norm": 0.10633298754692078,
      "learning_rate": 9.321595992978202e-05,
      "loss": 1.8294,
      "step": 36288
    },
    {
      "epoch": 0.07167066618495946,
      "grad_norm": 0.1104525774717331,
      "learning_rate": 9.320961970541733e-05,
      "loss": 1.834,
      "step": 36320
    },
    {
      "epoch": 0.0717338121463559,
      "grad_norm": 0.09994344413280487,
      "learning_rate": 9.320327948105264e-05,
      "loss": 1.8192,
      "step": 36352
    },
    {
      "epoch": 0.07179695810775233,
      "grad_norm": 0.10000725835561752,
      "learning_rate": 9.319693925668795e-05,
      "loss": 1.8291,
      "step": 36384
    },
    {
      "epoch": 0.07186010406914878,
      "grad_norm": 0.10379309952259064,
      "learning_rate": 9.319059903232325e-05,
      "loss": 1.8159,
      "step": 36416
    },
    {
      "epoch": 0.07192325003054521,
      "grad_norm": 0.10898219794034958,
      "learning_rate": 9.318425880795857e-05,
      "loss": 1.8293,
      "step": 36448
    },
    {
      "epoch": 0.07198639599194165,
      "grad_norm": 0.11423615366220474,
      "learning_rate": 9.317791858359388e-05,
      "loss": 1.8243,
      "step": 36480
    },
    {
      "epoch": 0.0720495419533381,
      "grad_norm": 0.10108506679534912,
      "learning_rate": 9.31715783592292e-05,
      "loss": 1.8247,
      "step": 36512
    },
    {
      "epoch": 0.07211268791473453,
      "grad_norm": 0.12401416152715683,
      "learning_rate": 9.31652381348645e-05,
      "loss": 1.8166,
      "step": 36544
    },
    {
      "epoch": 0.07217583387613097,
      "grad_norm": 0.10530426353216171,
      "learning_rate": 9.315889791049981e-05,
      "loss": 1.8176,
      "step": 36576
    },
    {
      "epoch": 0.07223897983752742,
      "grad_norm": 0.09537271410226822,
      "learning_rate": 9.315255768613512e-05,
      "loss": 1.8203,
      "step": 36608
    },
    {
      "epoch": 0.07230212579892385,
      "grad_norm": 0.10115291178226471,
      "learning_rate": 9.314621746177044e-05,
      "loss": 1.8296,
      "step": 36640
    },
    {
      "epoch": 0.0723652717603203,
      "grad_norm": 0.11395847052335739,
      "learning_rate": 9.313987723740575e-05,
      "loss": 1.8294,
      "step": 36672
    },
    {
      "epoch": 0.07242841772171674,
      "grad_norm": 0.10678514093160629,
      "learning_rate": 9.313353701304105e-05,
      "loss": 1.8013,
      "step": 36704
    },
    {
      "epoch": 0.07249156368311317,
      "grad_norm": 0.10725942254066467,
      "learning_rate": 9.312719678867637e-05,
      "loss": 1.8217,
      "step": 36736
    },
    {
      "epoch": 0.07255470964450962,
      "grad_norm": 0.1011452004313469,
      "learning_rate": 9.312085656431167e-05,
      "loss": 1.8101,
      "step": 36768
    },
    {
      "epoch": 0.07261785560590606,
      "grad_norm": 0.12700913846492767,
      "learning_rate": 9.311451633994698e-05,
      "loss": 1.8211,
      "step": 36800
    },
    {
      "epoch": 0.07268100156730249,
      "grad_norm": 0.1157853752374649,
      "learning_rate": 9.310817611558229e-05,
      "loss": 1.8159,
      "step": 36832
    },
    {
      "epoch": 0.07274414752869894,
      "grad_norm": 0.10147330909967422,
      "learning_rate": 9.31018358912176e-05,
      "loss": 1.8103,
      "step": 36864
    },
    {
      "epoch": 0.07280729349009538,
      "grad_norm": 0.1215939000248909,
      "learning_rate": 9.309549566685291e-05,
      "loss": 1.8265,
      "step": 36896
    },
    {
      "epoch": 0.07287043945149181,
      "grad_norm": 0.11663418263196945,
      "learning_rate": 9.308915544248823e-05,
      "loss": 1.8169,
      "step": 36928
    },
    {
      "epoch": 0.07293358541288826,
      "grad_norm": 0.10314902663230896,
      "learning_rate": 9.308281521812353e-05,
      "loss": 1.8128,
      "step": 36960
    },
    {
      "epoch": 0.0729967313742847,
      "grad_norm": 0.09695911407470703,
      "learning_rate": 9.307647499375885e-05,
      "loss": 1.8461,
      "step": 36992
    },
    {
      "epoch": 0.07305987733568113,
      "grad_norm": 0.09900636225938797,
      "learning_rate": 9.307013476939416e-05,
      "loss": 1.8228,
      "step": 37024
    },
    {
      "epoch": 0.07312302329707758,
      "grad_norm": 0.10411260277032852,
      "learning_rate": 9.306379454502947e-05,
      "loss": 1.8257,
      "step": 37056
    },
    {
      "epoch": 0.07318616925847402,
      "grad_norm": 0.1080189198255539,
      "learning_rate": 9.305745432066478e-05,
      "loss": 1.8154,
      "step": 37088
    },
    {
      "epoch": 0.07324931521987045,
      "grad_norm": 0.09551020711660385,
      "learning_rate": 9.305111409630009e-05,
      "loss": 1.84,
      "step": 37120
    },
    {
      "epoch": 0.0733124611812669,
      "grad_norm": 0.11926567554473877,
      "learning_rate": 9.30447738719354e-05,
      "loss": 1.8262,
      "step": 37152
    },
    {
      "epoch": 0.07337560714266333,
      "grad_norm": 0.1030915156006813,
      "learning_rate": 9.30384336475707e-05,
      "loss": 1.8131,
      "step": 37184
    },
    {
      "epoch": 0.07343875310405977,
      "grad_norm": 0.1016959697008133,
      "learning_rate": 9.3032093423206e-05,
      "loss": 1.8185,
      "step": 37216
    },
    {
      "epoch": 0.07350189906545622,
      "grad_norm": 0.10035569965839386,
      "learning_rate": 9.302575319884132e-05,
      "loss": 1.8153,
      "step": 37248
    },
    {
      "epoch": 0.07356504502685265,
      "grad_norm": 0.10914278775453568,
      "learning_rate": 9.301941297447664e-05,
      "loss": 1.813,
      "step": 37280
    },
    {
      "epoch": 0.07362819098824909,
      "grad_norm": 0.11313147842884064,
      "learning_rate": 9.301307275011195e-05,
      "loss": 1.8114,
      "step": 37312
    },
    {
      "epoch": 0.07369133694964554,
      "grad_norm": 0.11023863404989243,
      "learning_rate": 9.300673252574726e-05,
      "loss": 1.8167,
      "step": 37344
    },
    {
      "epoch": 0.07375448291104197,
      "grad_norm": 0.09830230474472046,
      "learning_rate": 9.300039230138257e-05,
      "loss": 1.804,
      "step": 37376
    },
    {
      "epoch": 0.07381762887243841,
      "grad_norm": 0.10366123914718628,
      "learning_rate": 9.299405207701788e-05,
      "loss": 1.8213,
      "step": 37408
    },
    {
      "epoch": 0.07388077483383486,
      "grad_norm": 0.11017373204231262,
      "learning_rate": 9.29877118526532e-05,
      "loss": 1.8149,
      "step": 37440
    },
    {
      "epoch": 0.0739439207952313,
      "grad_norm": 0.09416994452476501,
      "learning_rate": 9.298137162828851e-05,
      "loss": 1.8254,
      "step": 37472
    },
    {
      "epoch": 0.07400706675662773,
      "grad_norm": 0.10092901438474655,
      "learning_rate": 9.297503140392381e-05,
      "loss": 1.809,
      "step": 37504
    },
    {
      "epoch": 0.07407021271802418,
      "grad_norm": 0.10300584882497787,
      "learning_rate": 9.296869117955912e-05,
      "loss": 1.8281,
      "step": 37536
    },
    {
      "epoch": 0.07413335867942061,
      "grad_norm": 0.09270688146352768,
      "learning_rate": 9.296235095519444e-05,
      "loss": 1.8268,
      "step": 37568
    },
    {
      "epoch": 0.07419650464081706,
      "grad_norm": 0.11394724994897842,
      "learning_rate": 9.295601073082974e-05,
      "loss": 1.8228,
      "step": 37600
    },
    {
      "epoch": 0.0742596506022135,
      "grad_norm": 0.09486842900514603,
      "learning_rate": 9.294967050646504e-05,
      "loss": 1.8305,
      "step": 37632
    },
    {
      "epoch": 0.07432279656360993,
      "grad_norm": 0.0975983738899231,
      "learning_rate": 9.294333028210036e-05,
      "loss": 1.8077,
      "step": 37664
    },
    {
      "epoch": 0.07438594252500638,
      "grad_norm": 0.11098145693540573,
      "learning_rate": 9.293699005773567e-05,
      "loss": 1.8202,
      "step": 37696
    },
    {
      "epoch": 0.07444908848640282,
      "grad_norm": 0.1053488478064537,
      "learning_rate": 9.293064983337099e-05,
      "loss": 1.8218,
      "step": 37728
    },
    {
      "epoch": 0.07451223444779925,
      "grad_norm": 0.10563009232282639,
      "learning_rate": 9.292430960900629e-05,
      "loss": 1.8237,
      "step": 37760
    },
    {
      "epoch": 0.0745753804091957,
      "grad_norm": 0.11022141575813293,
      "learning_rate": 9.29179693846416e-05,
      "loss": 1.8189,
      "step": 37792
    },
    {
      "epoch": 0.07463852637059214,
      "grad_norm": 0.10405189543962479,
      "learning_rate": 9.291162916027692e-05,
      "loss": 1.8158,
      "step": 37824
    },
    {
      "epoch": 0.07470167233198857,
      "grad_norm": 0.10142597556114197,
      "learning_rate": 9.290528893591223e-05,
      "loss": 1.8115,
      "step": 37856
    },
    {
      "epoch": 0.07476481829338502,
      "grad_norm": 0.11152772605419159,
      "learning_rate": 9.289894871154754e-05,
      "loss": 1.8083,
      "step": 37888
    },
    {
      "epoch": 0.07482796425478146,
      "grad_norm": 0.10542437434196472,
      "learning_rate": 9.289260848718285e-05,
      "loss": 1.8219,
      "step": 37920
    },
    {
      "epoch": 0.07489111021617789,
      "grad_norm": 0.10162851959466934,
      "learning_rate": 9.288626826281816e-05,
      "loss": 1.8184,
      "step": 37952
    },
    {
      "epoch": 0.07495425617757434,
      "grad_norm": 0.0991579219698906,
      "learning_rate": 9.287992803845346e-05,
      "loss": 1.8167,
      "step": 37984
    },
    {
      "epoch": 0.07501740213897078,
      "grad_norm": 0.1067013218998909,
      "learning_rate": 9.287358781408878e-05,
      "loss": 1.8097,
      "step": 38016
    },
    {
      "epoch": 0.07508054810036721,
      "grad_norm": 0.10220654308795929,
      "learning_rate": 9.286724758972408e-05,
      "loss": 1.8114,
      "step": 38048
    },
    {
      "epoch": 0.07514369406176366,
      "grad_norm": 0.12089502066373825,
      "learning_rate": 9.286090736535939e-05,
      "loss": 1.8146,
      "step": 38080
    },
    {
      "epoch": 0.0752068400231601,
      "grad_norm": 0.09971319884061813,
      "learning_rate": 9.28545671409947e-05,
      "loss": 1.8044,
      "step": 38112
    },
    {
      "epoch": 0.07526998598455653,
      "grad_norm": 0.11873181164264679,
      "learning_rate": 9.284822691663002e-05,
      "loss": 1.7931,
      "step": 38144
    },
    {
      "epoch": 0.07533313194595298,
      "grad_norm": 0.10177662968635559,
      "learning_rate": 9.284188669226532e-05,
      "loss": 1.8277,
      "step": 38176
    },
    {
      "epoch": 0.07539627790734942,
      "grad_norm": 0.0983022078871727,
      "learning_rate": 9.283554646790064e-05,
      "loss": 1.8167,
      "step": 38208
    },
    {
      "epoch": 0.07545942386874585,
      "grad_norm": 0.09422893077135086,
      "learning_rate": 9.282920624353595e-05,
      "loss": 1.8133,
      "step": 38240
    },
    {
      "epoch": 0.0755225698301423,
      "grad_norm": 0.10076016187667847,
      "learning_rate": 9.282286601917126e-05,
      "loss": 1.808,
      "step": 38272
    },
    {
      "epoch": 0.07558571579153874,
      "grad_norm": 0.10187894105911255,
      "learning_rate": 9.281652579480657e-05,
      "loss": 1.8169,
      "step": 38304
    },
    {
      "epoch": 0.07564886175293517,
      "grad_norm": 0.1030297726392746,
      "learning_rate": 9.281018557044188e-05,
      "loss": 1.8038,
      "step": 38336
    },
    {
      "epoch": 0.07571200771433162,
      "grad_norm": 0.11084078997373581,
      "learning_rate": 9.28038453460772e-05,
      "loss": 1.8102,
      "step": 38368
    },
    {
      "epoch": 0.07577515367572805,
      "grad_norm": 0.10150840878486633,
      "learning_rate": 9.27975051217125e-05,
      "loss": 1.8206,
      "step": 38400
    },
    {
      "epoch": 0.0758382996371245,
      "grad_norm": 0.10247491300106049,
      "learning_rate": 9.279116489734781e-05,
      "loss": 1.8095,
      "step": 38432
    },
    {
      "epoch": 0.07590144559852094,
      "grad_norm": 0.10542050004005432,
      "learning_rate": 9.278482467298311e-05,
      "loss": 1.8101,
      "step": 38464
    },
    {
      "epoch": 0.07596459155991737,
      "grad_norm": 0.10321768373250961,
      "learning_rate": 9.277848444861843e-05,
      "loss": 1.814,
      "step": 38496
    },
    {
      "epoch": 0.07602773752131382,
      "grad_norm": 0.11027068644762039,
      "learning_rate": 9.277214422425374e-05,
      "loss": 1.801,
      "step": 38528
    },
    {
      "epoch": 0.07609088348271026,
      "grad_norm": 0.10503915697336197,
      "learning_rate": 9.276580399988906e-05,
      "loss": 1.8123,
      "step": 38560
    },
    {
      "epoch": 0.0761540294441067,
      "grad_norm": 0.09527040272951126,
      "learning_rate": 9.275946377552436e-05,
      "loss": 1.818,
      "step": 38592
    },
    {
      "epoch": 0.07621717540550314,
      "grad_norm": 0.0944819524884224,
      "learning_rate": 9.275312355115967e-05,
      "loss": 1.8021,
      "step": 38624
    },
    {
      "epoch": 0.07628032136689958,
      "grad_norm": 0.1021362841129303,
      "learning_rate": 9.274678332679499e-05,
      "loss": 1.8008,
      "step": 38656
    },
    {
      "epoch": 0.07634346732829601,
      "grad_norm": 0.11262723058462143,
      "learning_rate": 9.27404431024303e-05,
      "loss": 1.8125,
      "step": 38688
    },
    {
      "epoch": 0.07640661328969246,
      "grad_norm": 0.12389929592609406,
      "learning_rate": 9.27341028780656e-05,
      "loss": 1.8122,
      "step": 38720
    },
    {
      "epoch": 0.0764697592510889,
      "grad_norm": 0.12069406360387802,
      "learning_rate": 9.272776265370092e-05,
      "loss": 1.8153,
      "step": 38752
    },
    {
      "epoch": 0.07653290521248533,
      "grad_norm": 0.1107429787516594,
      "learning_rate": 9.272142242933623e-05,
      "loss": 1.8074,
      "step": 38784
    },
    {
      "epoch": 0.07659605117388178,
      "grad_norm": 0.10065750032663345,
      "learning_rate": 9.271508220497153e-05,
      "loss": 1.7971,
      "step": 38816
    },
    {
      "epoch": 0.07665919713527822,
      "grad_norm": 0.1054379791021347,
      "learning_rate": 9.270874198060685e-05,
      "loss": 1.8116,
      "step": 38848
    },
    {
      "epoch": 0.07672234309667465,
      "grad_norm": 0.10749877244234085,
      "learning_rate": 9.270240175624215e-05,
      "loss": 1.8134,
      "step": 38880
    },
    {
      "epoch": 0.0767854890580711,
      "grad_norm": 0.10346371680498123,
      "learning_rate": 9.269606153187746e-05,
      "loss": 1.7979,
      "step": 38912
    },
    {
      "epoch": 0.07684863501946754,
      "grad_norm": 0.09499409049749374,
      "learning_rate": 9.268972130751278e-05,
      "loss": 1.787,
      "step": 38944
    },
    {
      "epoch": 0.07691178098086397,
      "grad_norm": 0.10238825529813766,
      "learning_rate": 9.268338108314808e-05,
      "loss": 1.8043,
      "step": 38976
    },
    {
      "epoch": 0.07697492694226042,
      "grad_norm": 0.10353446006774902,
      "learning_rate": 9.267704085878339e-05,
      "loss": 1.8027,
      "step": 39008
    },
    {
      "epoch": 0.07703807290365686,
      "grad_norm": 0.10532581061124802,
      "learning_rate": 9.26707006344187e-05,
      "loss": 1.8136,
      "step": 39040
    },
    {
      "epoch": 0.07710121886505329,
      "grad_norm": 0.10245820134878159,
      "learning_rate": 9.266436041005402e-05,
      "loss": 1.7957,
      "step": 39072
    },
    {
      "epoch": 0.07716436482644974,
      "grad_norm": 0.12150374799966812,
      "learning_rate": 9.265802018568932e-05,
      "loss": 1.8121,
      "step": 39104
    },
    {
      "epoch": 0.07722751078784618,
      "grad_norm": 0.09707668423652649,
      "learning_rate": 9.265167996132464e-05,
      "loss": 1.8083,
      "step": 39136
    },
    {
      "epoch": 0.07729065674924261,
      "grad_norm": 0.10562973469495773,
      "learning_rate": 9.264533973695995e-05,
      "loss": 1.8108,
      "step": 39168
    },
    {
      "epoch": 0.07735380271063906,
      "grad_norm": 0.1110835075378418,
      "learning_rate": 9.263899951259527e-05,
      "loss": 1.8058,
      "step": 39200
    },
    {
      "epoch": 0.0774169486720355,
      "grad_norm": 0.12374687939882278,
      "learning_rate": 9.263265928823057e-05,
      "loss": 1.8182,
      "step": 39232
    },
    {
      "epoch": 0.07748009463343195,
      "grad_norm": 0.1003134623169899,
      "learning_rate": 9.262631906386587e-05,
      "loss": 1.8034,
      "step": 39264
    },
    {
      "epoch": 0.07754324059482838,
      "grad_norm": 0.10557498782873154,
      "learning_rate": 9.261997883950118e-05,
      "loss": 1.8188,
      "step": 39296
    },
    {
      "epoch": 0.07760638655622482,
      "grad_norm": 0.11196437478065491,
      "learning_rate": 9.26136386151365e-05,
      "loss": 1.8039,
      "step": 39328
    },
    {
      "epoch": 0.07766953251762126,
      "grad_norm": 0.11173975467681885,
      "learning_rate": 9.260729839077181e-05,
      "loss": 1.813,
      "step": 39360
    },
    {
      "epoch": 0.0777326784790177,
      "grad_norm": 0.11359525471925735,
      "learning_rate": 9.260095816640711e-05,
      "loss": 1.8039,
      "step": 39392
    },
    {
      "epoch": 0.07779582444041414,
      "grad_norm": 0.10052287578582764,
      "learning_rate": 9.259461794204243e-05,
      "loss": 1.7947,
      "step": 39424
    },
    {
      "epoch": 0.07785897040181058,
      "grad_norm": 0.10982044041156769,
      "learning_rate": 9.258827771767774e-05,
      "loss": 1.7962,
      "step": 39456
    },
    {
      "epoch": 0.07792211636320702,
      "grad_norm": 0.1478874832391739,
      "learning_rate": 9.258193749331306e-05,
      "loss": 1.7988,
      "step": 39488
    },
    {
      "epoch": 0.07798526232460345,
      "grad_norm": 0.11171736568212509,
      "learning_rate": 9.257559726894836e-05,
      "loss": 1.792,
      "step": 39520
    },
    {
      "epoch": 0.0780484082859999,
      "grad_norm": 0.11384129524230957,
      "learning_rate": 9.256925704458367e-05,
      "loss": 1.7974,
      "step": 39552
    },
    {
      "epoch": 0.07811155424739634,
      "grad_norm": 0.09803447127342224,
      "learning_rate": 9.256291682021899e-05,
      "loss": 1.7955,
      "step": 39584
    },
    {
      "epoch": 0.07817470020879277,
      "grad_norm": 0.11528968065977097,
      "learning_rate": 9.25565765958543e-05,
      "loss": 1.7994,
      "step": 39616
    },
    {
      "epoch": 0.07823784617018922,
      "grad_norm": 0.10794825106859207,
      "learning_rate": 9.25502363714896e-05,
      "loss": 1.8061,
      "step": 39648
    },
    {
      "epoch": 0.07830099213158566,
      "grad_norm": 0.10372241586446762,
      "learning_rate": 9.25438961471249e-05,
      "loss": 1.8079,
      "step": 39680
    },
    {
      "epoch": 0.0783641380929821,
      "grad_norm": 0.1086694598197937,
      "learning_rate": 9.253755592276022e-05,
      "loss": 1.7964,
      "step": 39712
    },
    {
      "epoch": 0.07842728405437854,
      "grad_norm": 0.10435189306735992,
      "learning_rate": 9.253121569839553e-05,
      "loss": 1.8014,
      "step": 39744
    },
    {
      "epoch": 0.07849043001577498,
      "grad_norm": 0.09914956986904144,
      "learning_rate": 9.252487547403083e-05,
      "loss": 1.8016,
      "step": 39776
    },
    {
      "epoch": 0.07855357597717141,
      "grad_norm": 0.10343772917985916,
      "learning_rate": 9.251853524966615e-05,
      "loss": 1.7982,
      "step": 39808
    },
    {
      "epoch": 0.07861672193856786,
      "grad_norm": 0.10450094193220139,
      "learning_rate": 9.251219502530146e-05,
      "loss": 1.8064,
      "step": 39840
    },
    {
      "epoch": 0.0786798678999643,
      "grad_norm": 0.10399092733860016,
      "learning_rate": 9.250585480093678e-05,
      "loss": 1.8072,
      "step": 39872
    },
    {
      "epoch": 0.07874301386136073,
      "grad_norm": 0.0940062552690506,
      "learning_rate": 9.249951457657209e-05,
      "loss": 1.81,
      "step": 39904
    },
    {
      "epoch": 0.07880615982275718,
      "grad_norm": 0.10817860811948776,
      "learning_rate": 9.249317435220739e-05,
      "loss": 1.8054,
      "step": 39936
    },
    {
      "epoch": 0.07886930578415362,
      "grad_norm": 0.10647379606962204,
      "learning_rate": 9.24868341278427e-05,
      "loss": 1.8002,
      "step": 39968
    },
    {
      "epoch": 0.07893245174555005,
      "grad_norm": 0.10552795976400375,
      "learning_rate": 9.248049390347802e-05,
      "loss": 1.8109,
      "step": 40000
    },
    {
      "epoch": 0.0789955977069465,
      "grad_norm": 0.09389588981866837,
      "learning_rate": 9.247415367911334e-05,
      "loss": 1.7971,
      "step": 40032
    },
    {
      "epoch": 0.07905874366834294,
      "grad_norm": 0.09789952635765076,
      "learning_rate": 9.246781345474864e-05,
      "loss": 1.7939,
      "step": 40064
    },
    {
      "epoch": 0.07912188962973939,
      "grad_norm": 0.10113061219453812,
      "learning_rate": 9.246147323038394e-05,
      "loss": 1.8024,
      "step": 40096
    },
    {
      "epoch": 0.07918503559113582,
      "grad_norm": 0.11077635735273361,
      "learning_rate": 9.245513300601925e-05,
      "loss": 1.7936,
      "step": 40128
    },
    {
      "epoch": 0.07924818155253226,
      "grad_norm": 0.10890121757984161,
      "learning_rate": 9.244879278165457e-05,
      "loss": 1.8042,
      "step": 40160
    },
    {
      "epoch": 0.0793113275139287,
      "grad_norm": 0.09608139842748642,
      "learning_rate": 9.244245255728987e-05,
      "loss": 1.7896,
      "step": 40192
    },
    {
      "epoch": 0.07937447347532514,
      "grad_norm": 0.11083830893039703,
      "learning_rate": 9.243611233292518e-05,
      "loss": 1.7935,
      "step": 40224
    },
    {
      "epoch": 0.07943761943672158,
      "grad_norm": 0.10516346246004105,
      "learning_rate": 9.24297721085605e-05,
      "loss": 1.8001,
      "step": 40256
    },
    {
      "epoch": 0.07950076539811803,
      "grad_norm": 0.09649717062711716,
      "learning_rate": 9.242343188419581e-05,
      "loss": 1.7929,
      "step": 40288
    },
    {
      "epoch": 0.07956391135951446,
      "grad_norm": 0.1006387397646904,
      "learning_rate": 9.241709165983111e-05,
      "loss": 1.7992,
      "step": 40320
    },
    {
      "epoch": 0.0796270573209109,
      "grad_norm": 0.10673325508832932,
      "learning_rate": 9.241075143546643e-05,
      "loss": 1.795,
      "step": 40352
    },
    {
      "epoch": 0.07969020328230735,
      "grad_norm": 0.09951280057430267,
      "learning_rate": 9.240441121110174e-05,
      "loss": 1.7951,
      "step": 40384
    },
    {
      "epoch": 0.07975334924370378,
      "grad_norm": 0.10043288767337799,
      "learning_rate": 9.239807098673706e-05,
      "loss": 1.7918,
      "step": 40416
    },
    {
      "epoch": 0.07981649520510022,
      "grad_norm": 0.099413201212883,
      "learning_rate": 9.239173076237236e-05,
      "loss": 1.8007,
      "step": 40448
    },
    {
      "epoch": 0.07987964116649666,
      "grad_norm": 0.10252483189105988,
      "learning_rate": 9.238539053800767e-05,
      "loss": 1.8108,
      "step": 40480
    },
    {
      "epoch": 0.0799427871278931,
      "grad_norm": 0.11060453951358795,
      "learning_rate": 9.237905031364297e-05,
      "loss": 1.8088,
      "step": 40512
    },
    {
      "epoch": 0.08000593308928954,
      "grad_norm": 0.11427866667509079,
      "learning_rate": 9.237271008927829e-05,
      "loss": 1.8075,
      "step": 40544
    },
    {
      "epoch": 0.08006907905068598,
      "grad_norm": 0.11861047893762589,
      "learning_rate": 9.23663698649136e-05,
      "loss": 1.8043,
      "step": 40576
    },
    {
      "epoch": 0.08013222501208242,
      "grad_norm": 0.10218507796525955,
      "learning_rate": 9.23600296405489e-05,
      "loss": 1.786,
      "step": 40608
    },
    {
      "epoch": 0.08019537097347886,
      "grad_norm": 0.09573070704936981,
      "learning_rate": 9.235368941618422e-05,
      "loss": 1.8028,
      "step": 40640
    },
    {
      "epoch": 0.0802585169348753,
      "grad_norm": 0.10090518742799759,
      "learning_rate": 9.234734919181953e-05,
      "loss": 1.7961,
      "step": 40672
    },
    {
      "epoch": 0.08032166289627174,
      "grad_norm": 0.10573742538690567,
      "learning_rate": 9.234100896745485e-05,
      "loss": 1.7983,
      "step": 40704
    },
    {
      "epoch": 0.08038480885766817,
      "grad_norm": 0.10804630070924759,
      "learning_rate": 9.233466874309015e-05,
      "loss": 1.8034,
      "step": 40736
    },
    {
      "epoch": 0.08044795481906462,
      "grad_norm": 0.09635918587446213,
      "learning_rate": 9.232832851872546e-05,
      "loss": 1.7872,
      "step": 40768
    },
    {
      "epoch": 0.08051110078046106,
      "grad_norm": 0.10776662826538086,
      "learning_rate": 9.232198829436078e-05,
      "loss": 1.7856,
      "step": 40800
    },
    {
      "epoch": 0.0805742467418575,
      "grad_norm": 0.09461846947669983,
      "learning_rate": 9.231564806999609e-05,
      "loss": 1.7823,
      "step": 40832
    },
    {
      "epoch": 0.08063739270325394,
      "grad_norm": 0.11045592278242111,
      "learning_rate": 9.230930784563139e-05,
      "loss": 1.7996,
      "step": 40864
    },
    {
      "epoch": 0.08070053866465038,
      "grad_norm": 0.11291345208883286,
      "learning_rate": 9.23029676212667e-05,
      "loss": 1.7871,
      "step": 40896
    },
    {
      "epoch": 0.08076368462604683,
      "grad_norm": 0.09787199646234512,
      "learning_rate": 9.229662739690201e-05,
      "loss": 1.7935,
      "step": 40928
    },
    {
      "epoch": 0.08082683058744326,
      "grad_norm": 0.1011669784784317,
      "learning_rate": 9.229028717253732e-05,
      "loss": 1.7938,
      "step": 40960
    },
    {
      "epoch": 0.0808899765488397,
      "grad_norm": 0.11360353231430054,
      "learning_rate": 9.228394694817262e-05,
      "loss": 1.7995,
      "step": 40992
    },
    {
      "epoch": 0.08095312251023615,
      "grad_norm": 0.1012316346168518,
      "learning_rate": 9.227760672380794e-05,
      "loss": 1.7962,
      "step": 41024
    },
    {
      "epoch": 0.08101626847163258,
      "grad_norm": 0.092242531478405,
      "learning_rate": 9.227126649944325e-05,
      "loss": 1.8005,
      "step": 41056
    },
    {
      "epoch": 0.08107941443302902,
      "grad_norm": 0.09925538301467896,
      "learning_rate": 9.226492627507857e-05,
      "loss": 1.8026,
      "step": 41088
    },
    {
      "epoch": 0.08114256039442547,
      "grad_norm": 0.10542775690555573,
      "learning_rate": 9.225858605071387e-05,
      "loss": 1.805,
      "step": 41120
    },
    {
      "epoch": 0.0812057063558219,
      "grad_norm": 0.0951443612575531,
      "learning_rate": 9.225224582634918e-05,
      "loss": 1.7847,
      "step": 41152
    },
    {
      "epoch": 0.08126885231721834,
      "grad_norm": 0.09704306721687317,
      "learning_rate": 9.22459056019845e-05,
      "loss": 1.8043,
      "step": 41184
    },
    {
      "epoch": 0.08133199827861479,
      "grad_norm": 0.10940796136856079,
      "learning_rate": 9.223956537761981e-05,
      "loss": 1.7811,
      "step": 41216
    },
    {
      "epoch": 0.08139514424001122,
      "grad_norm": 0.10112254321575165,
      "learning_rate": 9.223322515325513e-05,
      "loss": 1.8039,
      "step": 41248
    },
    {
      "epoch": 0.08145829020140766,
      "grad_norm": 0.1044624075293541,
      "learning_rate": 9.222688492889043e-05,
      "loss": 1.8049,
      "step": 41280
    },
    {
      "epoch": 0.0815214361628041,
      "grad_norm": 0.0949002057313919,
      "learning_rate": 9.222054470452574e-05,
      "loss": 1.7953,
      "step": 41312
    },
    {
      "epoch": 0.08158458212420054,
      "grad_norm": 0.0980764627456665,
      "learning_rate": 9.221420448016104e-05,
      "loss": 1.8033,
      "step": 41344
    },
    {
      "epoch": 0.08164772808559698,
      "grad_norm": 0.09994692355394363,
      "learning_rate": 9.220786425579636e-05,
      "loss": 1.7902,
      "step": 41376
    },
    {
      "epoch": 0.08171087404699343,
      "grad_norm": 0.09915197640657425,
      "learning_rate": 9.220152403143166e-05,
      "loss": 1.8045,
      "step": 41408
    },
    {
      "epoch": 0.08177402000838986,
      "grad_norm": 0.10263440757989883,
      "learning_rate": 9.219518380706697e-05,
      "loss": 1.7937,
      "step": 41440
    },
    {
      "epoch": 0.0818371659697863,
      "grad_norm": 0.10403347760438919,
      "learning_rate": 9.218884358270229e-05,
      "loss": 1.7966,
      "step": 41472
    },
    {
      "epoch": 0.08190031193118275,
      "grad_norm": 0.10131601244211197,
      "learning_rate": 9.21825033583376e-05,
      "loss": 1.7821,
      "step": 41504
    },
    {
      "epoch": 0.08196345789257918,
      "grad_norm": 0.11343494802713394,
      "learning_rate": 9.21761631339729e-05,
      "loss": 1.7958,
      "step": 41536
    },
    {
      "epoch": 0.08202660385397562,
      "grad_norm": 0.11137432605028152,
      "learning_rate": 9.216982290960822e-05,
      "loss": 1.7931,
      "step": 41568
    },
    {
      "epoch": 0.08208974981537207,
      "grad_norm": 0.10910090059041977,
      "learning_rate": 9.216348268524353e-05,
      "loss": 1.784,
      "step": 41600
    },
    {
      "epoch": 0.0821528957767685,
      "grad_norm": 0.10436250269412994,
      "learning_rate": 9.215714246087885e-05,
      "loss": 1.8033,
      "step": 41632
    },
    {
      "epoch": 0.08221604173816494,
      "grad_norm": 0.09831598401069641,
      "learning_rate": 9.215080223651415e-05,
      "loss": 1.7731,
      "step": 41664
    },
    {
      "epoch": 0.08227918769956138,
      "grad_norm": 0.1071930006146431,
      "learning_rate": 9.214446201214946e-05,
      "loss": 1.7773,
      "step": 41696
    },
    {
      "epoch": 0.08234233366095782,
      "grad_norm": 0.10158498585224152,
      "learning_rate": 9.213812178778478e-05,
      "loss": 1.8027,
      "step": 41728
    },
    {
      "epoch": 0.08240547962235427,
      "grad_norm": 0.09204748272895813,
      "learning_rate": 9.213178156342008e-05,
      "loss": 1.7938,
      "step": 41760
    },
    {
      "epoch": 0.0824686255837507,
      "grad_norm": 0.10261339694261551,
      "learning_rate": 9.212544133905538e-05,
      "loss": 1.7923,
      "step": 41792
    },
    {
      "epoch": 0.08253177154514714,
      "grad_norm": 0.1180913895368576,
      "learning_rate": 9.21191011146907e-05,
      "loss": 1.7916,
      "step": 41824
    },
    {
      "epoch": 0.08259491750654359,
      "grad_norm": 0.106290303170681,
      "learning_rate": 9.211276089032601e-05,
      "loss": 1.7908,
      "step": 41856
    },
    {
      "epoch": 0.08265806346794002,
      "grad_norm": 0.10435096174478531,
      "learning_rate": 9.210642066596132e-05,
      "loss": 1.7848,
      "step": 41888
    },
    {
      "epoch": 0.08272120942933646,
      "grad_norm": 0.10200298577547073,
      "learning_rate": 9.210008044159664e-05,
      "loss": 1.7808,
      "step": 41920
    },
    {
      "epoch": 0.08278435539073291,
      "grad_norm": 0.09501753002405167,
      "learning_rate": 9.209374021723194e-05,
      "loss": 1.8,
      "step": 41952
    },
    {
      "epoch": 0.08284750135212934,
      "grad_norm": 0.10118541121482849,
      "learning_rate": 9.208739999286725e-05,
      "loss": 1.7862,
      "step": 41984
    },
    {
      "epoch": 0.08291064731352578,
      "grad_norm": 0.11437226831912994,
      "learning_rate": 9.208105976850257e-05,
      "loss": 1.7996,
      "step": 42016
    },
    {
      "epoch": 0.08297379327492223,
      "grad_norm": 0.10282813757658005,
      "learning_rate": 9.207471954413788e-05,
      "loss": 1.7932,
      "step": 42048
    },
    {
      "epoch": 0.08303693923631866,
      "grad_norm": 0.11247044801712036,
      "learning_rate": 9.206837931977318e-05,
      "loss": 1.795,
      "step": 42080
    },
    {
      "epoch": 0.0831000851977151,
      "grad_norm": 0.10876993089914322,
      "learning_rate": 9.20620390954085e-05,
      "loss": 1.7984,
      "step": 42112
    },
    {
      "epoch": 0.08316323115911155,
      "grad_norm": 0.10570403188467026,
      "learning_rate": 9.20556988710438e-05,
      "loss": 1.7801,
      "step": 42144
    },
    {
      "epoch": 0.08322637712050798,
      "grad_norm": 0.09808826446533203,
      "learning_rate": 9.204935864667911e-05,
      "loss": 1.7848,
      "step": 42176
    },
    {
      "epoch": 0.08328952308190442,
      "grad_norm": 0.09764552861452103,
      "learning_rate": 9.204301842231441e-05,
      "loss": 1.7892,
      "step": 42208
    },
    {
      "epoch": 0.08335266904330087,
      "grad_norm": 0.10342146456241608,
      "learning_rate": 9.203667819794973e-05,
      "loss": 1.787,
      "step": 42240
    },
    {
      "epoch": 0.0834158150046973,
      "grad_norm": 0.10156437009572983,
      "learning_rate": 9.203033797358504e-05,
      "loss": 1.7839,
      "step": 42272
    },
    {
      "epoch": 0.08347896096609374,
      "grad_norm": 0.10073494911193848,
      "learning_rate": 9.202399774922036e-05,
      "loss": 1.7848,
      "step": 42304
    },
    {
      "epoch": 0.08354210692749019,
      "grad_norm": 0.09972520917654037,
      "learning_rate": 9.201765752485566e-05,
      "loss": 1.7969,
      "step": 42336
    },
    {
      "epoch": 0.08360525288888662,
      "grad_norm": 0.10750122368335724,
      "learning_rate": 9.201131730049097e-05,
      "loss": 1.7742,
      "step": 42368
    },
    {
      "epoch": 0.08366839885028306,
      "grad_norm": 0.10817664116621017,
      "learning_rate": 9.200497707612629e-05,
      "loss": 1.7913,
      "step": 42400
    },
    {
      "epoch": 0.0837315448116795,
      "grad_norm": 0.0960138589143753,
      "learning_rate": 9.19986368517616e-05,
      "loss": 1.7763,
      "step": 42432
    },
    {
      "epoch": 0.08379469077307594,
      "grad_norm": 0.1208823025226593,
      "learning_rate": 9.19922966273969e-05,
      "loss": 1.7922,
      "step": 42464
    },
    {
      "epoch": 0.08385783673447238,
      "grad_norm": 0.09739633649587631,
      "learning_rate": 9.198595640303222e-05,
      "loss": 1.7814,
      "step": 42496
    },
    {
      "epoch": 0.08392098269586883,
      "grad_norm": 0.09989912062883377,
      "learning_rate": 9.197961617866753e-05,
      "loss": 1.7784,
      "step": 42528
    },
    {
      "epoch": 0.08398412865726526,
      "grad_norm": 0.09694484621286392,
      "learning_rate": 9.197327595430283e-05,
      "loss": 1.7898,
      "step": 42560
    },
    {
      "epoch": 0.08404727461866171,
      "grad_norm": 0.10541020333766937,
      "learning_rate": 9.196693572993815e-05,
      "loss": 1.7813,
      "step": 42592
    },
    {
      "epoch": 0.08411042058005815,
      "grad_norm": 0.09711425006389618,
      "learning_rate": 9.196059550557345e-05,
      "loss": 1.7851,
      "step": 42624
    },
    {
      "epoch": 0.08417356654145458,
      "grad_norm": 0.10375643521547318,
      "learning_rate": 9.195425528120876e-05,
      "loss": 1.7821,
      "step": 42656
    },
    {
      "epoch": 0.08423671250285103,
      "grad_norm": 0.10384511947631836,
      "learning_rate": 9.194791505684408e-05,
      "loss": 1.7672,
      "step": 42688
    },
    {
      "epoch": 0.08429985846424747,
      "grad_norm": 0.10180196911096573,
      "learning_rate": 9.194157483247939e-05,
      "loss": 1.8067,
      "step": 42720
    },
    {
      "epoch": 0.0843630044256439,
      "grad_norm": 0.10466025769710541,
      "learning_rate": 9.19352346081147e-05,
      "loss": 1.7788,
      "step": 42752
    },
    {
      "epoch": 0.08442615038704035,
      "grad_norm": 0.101494699716568,
      "learning_rate": 9.192889438375001e-05,
      "loss": 1.77,
      "step": 42784
    },
    {
      "epoch": 0.08448929634843678,
      "grad_norm": 0.09314579516649246,
      "learning_rate": 9.192255415938532e-05,
      "loss": 1.7882,
      "step": 42816
    },
    {
      "epoch": 0.08455244230983322,
      "grad_norm": 0.09800131618976593,
      "learning_rate": 9.191621393502064e-05,
      "loss": 1.7885,
      "step": 42848
    },
    {
      "epoch": 0.08461558827122967,
      "grad_norm": 0.1176493763923645,
      "learning_rate": 9.190987371065594e-05,
      "loss": 1.7811,
      "step": 42880
    },
    {
      "epoch": 0.0846787342326261,
      "grad_norm": 0.09414546191692352,
      "learning_rate": 9.190353348629125e-05,
      "loss": 1.7718,
      "step": 42912
    },
    {
      "epoch": 0.08474188019402254,
      "grad_norm": 0.10884156078100204,
      "learning_rate": 9.189719326192657e-05,
      "loss": 1.7932,
      "step": 42944
    },
    {
      "epoch": 0.08480502615541899,
      "grad_norm": 0.10244859009981155,
      "learning_rate": 9.189085303756187e-05,
      "loss": 1.789,
      "step": 42976
    },
    {
      "epoch": 0.08486817211681542,
      "grad_norm": 0.1141287088394165,
      "learning_rate": 9.188451281319718e-05,
      "loss": 1.7932,
      "step": 43008
    },
    {
      "epoch": 0.08493131807821186,
      "grad_norm": 0.10819052904844284,
      "learning_rate": 9.187817258883248e-05,
      "loss": 1.7814,
      "step": 43040
    },
    {
      "epoch": 0.08499446403960831,
      "grad_norm": 0.1084299311041832,
      "learning_rate": 9.18718323644678e-05,
      "loss": 1.7944,
      "step": 43072
    },
    {
      "epoch": 0.08505761000100474,
      "grad_norm": 0.0994194969534874,
      "learning_rate": 9.186549214010311e-05,
      "loss": 1.7865,
      "step": 43104
    },
    {
      "epoch": 0.08512075596240118,
      "grad_norm": 0.0987580195069313,
      "learning_rate": 9.185915191573841e-05,
      "loss": 1.7902,
      "step": 43136
    },
    {
      "epoch": 0.08518390192379763,
      "grad_norm": 0.10071149468421936,
      "learning_rate": 9.185281169137373e-05,
      "loss": 1.7826,
      "step": 43168
    },
    {
      "epoch": 0.08524704788519406,
      "grad_norm": 0.09697140753269196,
      "learning_rate": 9.184647146700904e-05,
      "loss": 1.7816,
      "step": 43200
    },
    {
      "epoch": 0.0853101938465905,
      "grad_norm": 0.10094501823186874,
      "learning_rate": 9.184013124264436e-05,
      "loss": 1.7721,
      "step": 43232
    },
    {
      "epoch": 0.08537333980798695,
      "grad_norm": 0.12242378294467926,
      "learning_rate": 9.183379101827967e-05,
      "loss": 1.7691,
      "step": 43264
    },
    {
      "epoch": 0.08543648576938338,
      "grad_norm": 0.09958004951477051,
      "learning_rate": 9.182745079391497e-05,
      "loss": 1.7728,
      "step": 43296
    },
    {
      "epoch": 0.08549963173077982,
      "grad_norm": 0.09500090032815933,
      "learning_rate": 9.182111056955029e-05,
      "loss": 1.788,
      "step": 43328
    },
    {
      "epoch": 0.08556277769217627,
      "grad_norm": 0.10032177716493607,
      "learning_rate": 9.18147703451856e-05,
      "loss": 1.779,
      "step": 43360
    },
    {
      "epoch": 0.0856259236535727,
      "grad_norm": 0.11116743832826614,
      "learning_rate": 9.18084301208209e-05,
      "loss": 1.7763,
      "step": 43392
    },
    {
      "epoch": 0.08568906961496915,
      "grad_norm": 0.09952928870916367,
      "learning_rate": 9.18020898964562e-05,
      "loss": 1.7928,
      "step": 43424
    },
    {
      "epoch": 0.08575221557636559,
      "grad_norm": 0.09488236159086227,
      "learning_rate": 9.179574967209152e-05,
      "loss": 1.7815,
      "step": 43456
    },
    {
      "epoch": 0.08581536153776202,
      "grad_norm": 0.0935351774096489,
      "learning_rate": 9.178940944772683e-05,
      "loss": 1.7812,
      "step": 43488
    },
    {
      "epoch": 0.08587850749915847,
      "grad_norm": 0.10840978473424911,
      "learning_rate": 9.178306922336215e-05,
      "loss": 1.784,
      "step": 43520
    },
    {
      "epoch": 0.0859416534605549,
      "grad_norm": 0.09159401804208755,
      "learning_rate": 9.177672899899745e-05,
      "loss": 1.7635,
      "step": 43552
    },
    {
      "epoch": 0.08600479942195134,
      "grad_norm": 0.1033141240477562,
      "learning_rate": 9.177038877463276e-05,
      "loss": 1.7757,
      "step": 43584
    },
    {
      "epoch": 0.08606794538334779,
      "grad_norm": 0.11672907322645187,
      "learning_rate": 9.176404855026808e-05,
      "loss": 1.7754,
      "step": 43616
    },
    {
      "epoch": 0.08613109134474423,
      "grad_norm": 0.09875050187110901,
      "learning_rate": 9.175770832590339e-05,
      "loss": 1.7823,
      "step": 43648
    },
    {
      "epoch": 0.08619423730614066,
      "grad_norm": 0.1074860617518425,
      "learning_rate": 9.17513681015387e-05,
      "loss": 1.7694,
      "step": 43680
    },
    {
      "epoch": 0.08625738326753711,
      "grad_norm": 0.09928867220878601,
      "learning_rate": 9.174502787717401e-05,
      "loss": 1.784,
      "step": 43712
    },
    {
      "epoch": 0.08632052922893355,
      "grad_norm": 0.09989845752716064,
      "learning_rate": 9.173868765280932e-05,
      "loss": 1.7733,
      "step": 43744
    },
    {
      "epoch": 0.08638367519032998,
      "grad_norm": 0.10041379928588867,
      "learning_rate": 9.173234742844464e-05,
      "loss": 1.7821,
      "step": 43776
    },
    {
      "epoch": 0.08644682115172643,
      "grad_norm": 0.10040847957134247,
      "learning_rate": 9.172600720407994e-05,
      "loss": 1.7918,
      "step": 43808
    },
    {
      "epoch": 0.08650996711312287,
      "grad_norm": 0.1039457842707634,
      "learning_rate": 9.171966697971524e-05,
      "loss": 1.7796,
      "step": 43840
    },
    {
      "epoch": 0.0865731130745193,
      "grad_norm": 0.1016286239027977,
      "learning_rate": 9.171332675535055e-05,
      "loss": 1.798,
      "step": 43872
    },
    {
      "epoch": 0.08663625903591575,
      "grad_norm": 0.10243168473243713,
      "learning_rate": 9.170698653098587e-05,
      "loss": 1.7655,
      "step": 43904
    },
    {
      "epoch": 0.08669940499731218,
      "grad_norm": 0.10807812958955765,
      "learning_rate": 9.170064630662118e-05,
      "loss": 1.7855,
      "step": 43936
    },
    {
      "epoch": 0.08676255095870862,
      "grad_norm": 0.09867890924215317,
      "learning_rate": 9.169430608225648e-05,
      "loss": 1.784,
      "step": 43968
    },
    {
      "epoch": 0.08682569692010507,
      "grad_norm": 0.10520311444997787,
      "learning_rate": 9.16879658578918e-05,
      "loss": 1.7829,
      "step": 44000
    },
    {
      "epoch": 0.0868888428815015,
      "grad_norm": 0.10657200217247009,
      "learning_rate": 9.168162563352711e-05,
      "loss": 1.7779,
      "step": 44032
    },
    {
      "epoch": 0.08695198884289794,
      "grad_norm": 0.10489603132009506,
      "learning_rate": 9.167528540916243e-05,
      "loss": 1.7712,
      "step": 44064
    },
    {
      "epoch": 0.08701513480429439,
      "grad_norm": 0.11926059424877167,
      "learning_rate": 9.166894518479773e-05,
      "loss": 1.7596,
      "step": 44096
    },
    {
      "epoch": 0.08707828076569082,
      "grad_norm": 0.10963926464319229,
      "learning_rate": 9.166260496043304e-05,
      "loss": 1.7784,
      "step": 44128
    },
    {
      "epoch": 0.08714142672708726,
      "grad_norm": 0.09565390646457672,
      "learning_rate": 9.165626473606836e-05,
      "loss": 1.7787,
      "step": 44160
    },
    {
      "epoch": 0.08720457268848371,
      "grad_norm": 0.09736917912960052,
      "learning_rate": 9.164992451170367e-05,
      "loss": 1.7737,
      "step": 44192
    },
    {
      "epoch": 0.08726771864988014,
      "grad_norm": 0.09433715790510178,
      "learning_rate": 9.164358428733897e-05,
      "loss": 1.7831,
      "step": 44224
    },
    {
      "epoch": 0.08733086461127659,
      "grad_norm": 0.10643383860588074,
      "learning_rate": 9.163724406297427e-05,
      "loss": 1.7752,
      "step": 44256
    },
    {
      "epoch": 0.08739401057267303,
      "grad_norm": 0.10283531993627548,
      "learning_rate": 9.163090383860959e-05,
      "loss": 1.7671,
      "step": 44288
    },
    {
      "epoch": 0.08745715653406946,
      "grad_norm": 0.1017676293849945,
      "learning_rate": 9.16245636142449e-05,
      "loss": 1.7745,
      "step": 44320
    },
    {
      "epoch": 0.08752030249546591,
      "grad_norm": 0.09568081796169281,
      "learning_rate": 9.16182233898802e-05,
      "loss": 1.764,
      "step": 44352
    },
    {
      "epoch": 0.08758344845686235,
      "grad_norm": 0.09435157477855682,
      "learning_rate": 9.161188316551552e-05,
      "loss": 1.7754,
      "step": 44384
    },
    {
      "epoch": 0.08764659441825878,
      "grad_norm": 0.09382575750350952,
      "learning_rate": 9.160554294115083e-05,
      "loss": 1.7806,
      "step": 44416
    },
    {
      "epoch": 0.08770974037965523,
      "grad_norm": 0.11132270097732544,
      "learning_rate": 9.159920271678615e-05,
      "loss": 1.7952,
      "step": 44448
    },
    {
      "epoch": 0.08777288634105167,
      "grad_norm": 0.10555444657802582,
      "learning_rate": 9.159286249242145e-05,
      "loss": 1.7851,
      "step": 44480
    },
    {
      "epoch": 0.0878360323024481,
      "grad_norm": 0.12005648761987686,
      "learning_rate": 9.158652226805676e-05,
      "loss": 1.7731,
      "step": 44512
    },
    {
      "epoch": 0.08789917826384455,
      "grad_norm": 0.10580897331237793,
      "learning_rate": 9.158018204369208e-05,
      "loss": 1.7764,
      "step": 44544
    },
    {
      "epoch": 0.08796232422524099,
      "grad_norm": 0.1080353707075119,
      "learning_rate": 9.15738418193274e-05,
      "loss": 1.7906,
      "step": 44576
    },
    {
      "epoch": 0.08802547018663742,
      "grad_norm": 0.09693998098373413,
      "learning_rate": 9.156750159496271e-05,
      "loss": 1.7681,
      "step": 44608
    },
    {
      "epoch": 0.08808861614803387,
      "grad_norm": 0.10377060621976852,
      "learning_rate": 9.156116137059801e-05,
      "loss": 1.7765,
      "step": 44640
    },
    {
      "epoch": 0.0881517621094303,
      "grad_norm": 0.09804272651672363,
      "learning_rate": 9.155482114623331e-05,
      "loss": 1.7676,
      "step": 44672
    },
    {
      "epoch": 0.08821490807082674,
      "grad_norm": 0.09506504982709885,
      "learning_rate": 9.154848092186862e-05,
      "loss": 1.7788,
      "step": 44704
    },
    {
      "epoch": 0.08827805403222319,
      "grad_norm": 0.10393403470516205,
      "learning_rate": 9.154214069750394e-05,
      "loss": 1.7785,
      "step": 44736
    },
    {
      "epoch": 0.08834119999361963,
      "grad_norm": 0.10242267698049545,
      "learning_rate": 9.153580047313924e-05,
      "loss": 1.763,
      "step": 44768
    },
    {
      "epoch": 0.08840434595501606,
      "grad_norm": 0.1032901257276535,
      "learning_rate": 9.152946024877455e-05,
      "loss": 1.7592,
      "step": 44800
    },
    {
      "epoch": 0.08846749191641251,
      "grad_norm": 0.10288886725902557,
      "learning_rate": 9.152312002440987e-05,
      "loss": 1.7737,
      "step": 44832
    },
    {
      "epoch": 0.08853063787780895,
      "grad_norm": 0.11493588984012604,
      "learning_rate": 9.151677980004518e-05,
      "loss": 1.7862,
      "step": 44864
    },
    {
      "epoch": 0.08859378383920538,
      "grad_norm": 0.10273680835962296,
      "learning_rate": 9.151043957568048e-05,
      "loss": 1.7551,
      "step": 44896
    },
    {
      "epoch": 0.08865692980060183,
      "grad_norm": 0.09941155463457108,
      "learning_rate": 9.15040993513158e-05,
      "loss": 1.7706,
      "step": 44928
    },
    {
      "epoch": 0.08872007576199827,
      "grad_norm": 0.09910742193460464,
      "learning_rate": 9.149775912695111e-05,
      "loss": 1.7847,
      "step": 44960
    },
    {
      "epoch": 0.0887832217233947,
      "grad_norm": 0.09402486681938171,
      "learning_rate": 9.149141890258643e-05,
      "loss": 1.7863,
      "step": 44992
    },
    {
      "epoch": 0.08884636768479115,
      "grad_norm": 0.1073371097445488,
      "learning_rate": 9.148507867822173e-05,
      "loss": 1.7717,
      "step": 45024
    },
    {
      "epoch": 0.08890951364618759,
      "grad_norm": 0.0946664810180664,
      "learning_rate": 9.147873845385704e-05,
      "loss": 1.7773,
      "step": 45056
    },
    {
      "epoch": 0.08897265960758402,
      "grad_norm": 0.09917966276407242,
      "learning_rate": 9.147239822949235e-05,
      "loss": 1.7713,
      "step": 45088
    },
    {
      "epoch": 0.08903580556898047,
      "grad_norm": 0.097694993019104,
      "learning_rate": 9.146605800512766e-05,
      "loss": 1.7715,
      "step": 45120
    },
    {
      "epoch": 0.0890989515303769,
      "grad_norm": 0.09713596850633621,
      "learning_rate": 9.145971778076296e-05,
      "loss": 1.7701,
      "step": 45152
    },
    {
      "epoch": 0.08916209749177335,
      "grad_norm": 0.10814554244279861,
      "learning_rate": 9.145337755639828e-05,
      "loss": 1.7682,
      "step": 45184
    },
    {
      "epoch": 0.08922524345316979,
      "grad_norm": 0.09634155035018921,
      "learning_rate": 9.144703733203359e-05,
      "loss": 1.77,
      "step": 45216
    },
    {
      "epoch": 0.08928838941456622,
      "grad_norm": 0.10277990996837616,
      "learning_rate": 9.14406971076689e-05,
      "loss": 1.7702,
      "step": 45248
    },
    {
      "epoch": 0.08935153537596267,
      "grad_norm": 0.10231760144233704,
      "learning_rate": 9.143435688330422e-05,
      "loss": 1.7677,
      "step": 45280
    },
    {
      "epoch": 0.08941468133735911,
      "grad_norm": 0.09473563730716705,
      "learning_rate": 9.142801665893952e-05,
      "loss": 1.7676,
      "step": 45312
    },
    {
      "epoch": 0.08947782729875554,
      "grad_norm": 0.0933060497045517,
      "learning_rate": 9.142167643457483e-05,
      "loss": 1.7573,
      "step": 45344
    },
    {
      "epoch": 0.089540973260152,
      "grad_norm": 0.11009597033262253,
      "learning_rate": 9.141533621021015e-05,
      "loss": 1.7727,
      "step": 45376
    },
    {
      "epoch": 0.08960411922154843,
      "grad_norm": 0.10508358478546143,
      "learning_rate": 9.140899598584546e-05,
      "loss": 1.7667,
      "step": 45408
    },
    {
      "epoch": 0.08966726518294486,
      "grad_norm": 0.11706570535898209,
      "learning_rate": 9.140265576148076e-05,
      "loss": 1.7728,
      "step": 45440
    },
    {
      "epoch": 0.08973041114434131,
      "grad_norm": 0.1036912128329277,
      "learning_rate": 9.139631553711608e-05,
      "loss": 1.7708,
      "step": 45472
    },
    {
      "epoch": 0.08979355710573775,
      "grad_norm": 0.10347336530685425,
      "learning_rate": 9.138997531275138e-05,
      "loss": 1.782,
      "step": 45504
    },
    {
      "epoch": 0.08985670306713418,
      "grad_norm": 0.12362384796142578,
      "learning_rate": 9.13836350883867e-05,
      "loss": 1.7812,
      "step": 45536
    },
    {
      "epoch": 0.08991984902853063,
      "grad_norm": 0.09904732555150986,
      "learning_rate": 9.1377294864022e-05,
      "loss": 1.765,
      "step": 45568
    },
    {
      "epoch": 0.08998299498992707,
      "grad_norm": 0.12025132030248642,
      "learning_rate": 9.137095463965731e-05,
      "loss": 1.7708,
      "step": 45600
    },
    {
      "epoch": 0.0900461409513235,
      "grad_norm": 0.0950314849615097,
      "learning_rate": 9.136461441529262e-05,
      "loss": 1.7588,
      "step": 45632
    },
    {
      "epoch": 0.09010928691271995,
      "grad_norm": 0.0949680283665657,
      "learning_rate": 9.135827419092794e-05,
      "loss": 1.7692,
      "step": 45664
    },
    {
      "epoch": 0.09017243287411639,
      "grad_norm": 0.09779734164476395,
      "learning_rate": 9.135193396656324e-05,
      "loss": 1.7705,
      "step": 45696
    },
    {
      "epoch": 0.09023557883551282,
      "grad_norm": 0.09569840878248215,
      "learning_rate": 9.134559374219856e-05,
      "loss": 1.7637,
      "step": 45728
    },
    {
      "epoch": 0.09029872479690927,
      "grad_norm": 0.11018886417150497,
      "learning_rate": 9.133925351783387e-05,
      "loss": 1.7847,
      "step": 45760
    },
    {
      "epoch": 0.0903618707583057,
      "grad_norm": 0.13354440033435822,
      "learning_rate": 9.133291329346918e-05,
      "loss": 1.7658,
      "step": 45792
    },
    {
      "epoch": 0.09042501671970214,
      "grad_norm": 0.09451562166213989,
      "learning_rate": 9.132657306910449e-05,
      "loss": 1.7802,
      "step": 45824
    },
    {
      "epoch": 0.09048816268109859,
      "grad_norm": 0.10078594833612442,
      "learning_rate": 9.13202328447398e-05,
      "loss": 1.781,
      "step": 45856
    },
    {
      "epoch": 0.09055130864249503,
      "grad_norm": 0.10404020547866821,
      "learning_rate": 9.131389262037511e-05,
      "loss": 1.7625,
      "step": 45888
    },
    {
      "epoch": 0.09061445460389146,
      "grad_norm": 0.104709692299366,
      "learning_rate": 9.130755239601042e-05,
      "loss": 1.7666,
      "step": 45920
    },
    {
      "epoch": 0.09067760056528791,
      "grad_norm": 0.09853670001029968,
      "learning_rate": 9.130121217164573e-05,
      "loss": 1.7837,
      "step": 45952
    },
    {
      "epoch": 0.09074074652668435,
      "grad_norm": 0.09726610779762268,
      "learning_rate": 9.129487194728103e-05,
      "loss": 1.766,
      "step": 45984
    },
    {
      "epoch": 0.0908038924880808,
      "grad_norm": 0.10068164020776749,
      "learning_rate": 9.128853172291635e-05,
      "loss": 1.7705,
      "step": 46016
    },
    {
      "epoch": 0.09086703844947723,
      "grad_norm": 0.09436113387346268,
      "learning_rate": 9.128219149855166e-05,
      "loss": 1.7621,
      "step": 46048
    },
    {
      "epoch": 0.09093018441087367,
      "grad_norm": 0.09907054156064987,
      "learning_rate": 9.127585127418697e-05,
      "loss": 1.7715,
      "step": 46080
    },
    {
      "epoch": 0.09099333037227011,
      "grad_norm": 0.10897371917963028,
      "learning_rate": 9.126951104982228e-05,
      "loss": 1.7669,
      "step": 46112
    },
    {
      "epoch": 0.09105647633366655,
      "grad_norm": 0.10993722081184387,
      "learning_rate": 9.126317082545759e-05,
      "loss": 1.7759,
      "step": 46144
    },
    {
      "epoch": 0.09111962229506299,
      "grad_norm": 0.09244080632925034,
      "learning_rate": 9.12568306010929e-05,
      "loss": 1.7744,
      "step": 46176
    },
    {
      "epoch": 0.09118276825645943,
      "grad_norm": 0.1071285679936409,
      "learning_rate": 9.125049037672822e-05,
      "loss": 1.7786,
      "step": 46208
    },
    {
      "epoch": 0.09124591421785587,
      "grad_norm": 0.09893108159303665,
      "learning_rate": 9.124415015236352e-05,
      "loss": 1.7712,
      "step": 46240
    },
    {
      "epoch": 0.0913090601792523,
      "grad_norm": 0.10827232152223587,
      "learning_rate": 9.123780992799883e-05,
      "loss": 1.7685,
      "step": 46272
    },
    {
      "epoch": 0.09137220614064875,
      "grad_norm": 0.09942732006311417,
      "learning_rate": 9.123146970363414e-05,
      "loss": 1.7526,
      "step": 46304
    },
    {
      "epoch": 0.09143535210204519,
      "grad_norm": 0.10993362218141556,
      "learning_rate": 9.122512947926945e-05,
      "loss": 1.7748,
      "step": 46336
    },
    {
      "epoch": 0.09149849806344162,
      "grad_norm": 0.09984655678272247,
      "learning_rate": 9.121878925490475e-05,
      "loss": 1.773,
      "step": 46368
    },
    {
      "epoch": 0.09156164402483807,
      "grad_norm": 0.09968504309654236,
      "learning_rate": 9.121244903054007e-05,
      "loss": 1.7652,
      "step": 46400
    },
    {
      "epoch": 0.09162478998623451,
      "grad_norm": 0.0972420945763588,
      "learning_rate": 9.120610880617538e-05,
      "loss": 1.771,
      "step": 46432
    },
    {
      "epoch": 0.09168793594763094,
      "grad_norm": 0.09219255298376083,
      "learning_rate": 9.11997685818107e-05,
      "loss": 1.7695,
      "step": 46464
    },
    {
      "epoch": 0.0917510819090274,
      "grad_norm": 0.09459584206342697,
      "learning_rate": 9.1193428357446e-05,
      "loss": 1.7714,
      "step": 46496
    },
    {
      "epoch": 0.09181422787042383,
      "grad_norm": 0.09561120718717575,
      "learning_rate": 9.118708813308131e-05,
      "loss": 1.7632,
      "step": 46528
    },
    {
      "epoch": 0.09187737383182026,
      "grad_norm": 0.10604512691497803,
      "learning_rate": 9.118074790871663e-05,
      "loss": 1.771,
      "step": 46560
    },
    {
      "epoch": 0.09194051979321671,
      "grad_norm": 0.11895725131034851,
      "learning_rate": 9.117440768435194e-05,
      "loss": 1.768,
      "step": 46592
    },
    {
      "epoch": 0.09200366575461315,
      "grad_norm": 0.12577363848686218,
      "learning_rate": 9.116806745998725e-05,
      "loss": 1.776,
      "step": 46624
    },
    {
      "epoch": 0.09206681171600958,
      "grad_norm": 0.10248183459043503,
      "learning_rate": 9.116172723562256e-05,
      "loss": 1.769,
      "step": 46656
    },
    {
      "epoch": 0.09212995767740603,
      "grad_norm": 0.09341636300086975,
      "learning_rate": 9.115538701125787e-05,
      "loss": 1.7673,
      "step": 46688
    },
    {
      "epoch": 0.09219310363880247,
      "grad_norm": 0.10240786522626877,
      "learning_rate": 9.114904678689317e-05,
      "loss": 1.7577,
      "step": 46720
    },
    {
      "epoch": 0.0922562496001989,
      "grad_norm": 0.10875356197357178,
      "learning_rate": 9.114270656252849e-05,
      "loss": 1.7614,
      "step": 46752
    },
    {
      "epoch": 0.09231939556159535,
      "grad_norm": 0.10090479254722595,
      "learning_rate": 9.113636633816379e-05,
      "loss": 1.7592,
      "step": 46784
    },
    {
      "epoch": 0.09238254152299179,
      "grad_norm": 0.10226140171289444,
      "learning_rate": 9.11300261137991e-05,
      "loss": 1.7618,
      "step": 46816
    },
    {
      "epoch": 0.09244568748438824,
      "grad_norm": 0.10055588185787201,
      "learning_rate": 9.112368588943442e-05,
      "loss": 1.7646,
      "step": 46848
    },
    {
      "epoch": 0.09250883344578467,
      "grad_norm": 0.10336578637361526,
      "learning_rate": 9.111734566506973e-05,
      "loss": 1.7587,
      "step": 46880
    },
    {
      "epoch": 0.09257197940718111,
      "grad_norm": 0.09818453341722488,
      "learning_rate": 9.111100544070503e-05,
      "loss": 1.7655,
      "step": 46912
    },
    {
      "epoch": 0.09263512536857756,
      "grad_norm": 0.1010284349322319,
      "learning_rate": 9.110466521634035e-05,
      "loss": 1.7573,
      "step": 46944
    },
    {
      "epoch": 0.09269827132997399,
      "grad_norm": 0.09641031175851822,
      "learning_rate": 9.109832499197566e-05,
      "loss": 1.7644,
      "step": 46976
    },
    {
      "epoch": 0.09276141729137043,
      "grad_norm": 0.1004476398229599,
      "learning_rate": 9.109198476761097e-05,
      "loss": 1.7531,
      "step": 47008
    },
    {
      "epoch": 0.09282456325276688,
      "grad_norm": 0.09911256283521652,
      "learning_rate": 9.108564454324628e-05,
      "loss": 1.7625,
      "step": 47040
    },
    {
      "epoch": 0.09288770921416331,
      "grad_norm": 0.0979299247264862,
      "learning_rate": 9.107930431888159e-05,
      "loss": 1.7647,
      "step": 47072
    },
    {
      "epoch": 0.09295085517555975,
      "grad_norm": 0.10227309912443161,
      "learning_rate": 9.10729640945169e-05,
      "loss": 1.7736,
      "step": 47104
    },
    {
      "epoch": 0.0930140011369562,
      "grad_norm": 0.10089951753616333,
      "learning_rate": 9.10666238701522e-05,
      "loss": 1.7699,
      "step": 47136
    },
    {
      "epoch": 0.09307714709835263,
      "grad_norm": 0.09610005468130112,
      "learning_rate": 9.106028364578752e-05,
      "loss": 1.7673,
      "step": 47168
    },
    {
      "epoch": 0.09314029305974907,
      "grad_norm": 0.12601642310619354,
      "learning_rate": 9.105394342142282e-05,
      "loss": 1.7654,
      "step": 47200
    },
    {
      "epoch": 0.09320343902114551,
      "grad_norm": 0.09972970187664032,
      "learning_rate": 9.104760319705814e-05,
      "loss": 1.7547,
      "step": 47232
    },
    {
      "epoch": 0.09326658498254195,
      "grad_norm": 0.1147465631365776,
      "learning_rate": 9.104126297269345e-05,
      "loss": 1.7547,
      "step": 47264
    },
    {
      "epoch": 0.09332973094393839,
      "grad_norm": 0.1079910472035408,
      "learning_rate": 9.103492274832877e-05,
      "loss": 1.7565,
      "step": 47296
    },
    {
      "epoch": 0.09339287690533483,
      "grad_norm": 0.11165102571249008,
      "learning_rate": 9.102858252396407e-05,
      "loss": 1.7594,
      "step": 47328
    },
    {
      "epoch": 0.09345602286673127,
      "grad_norm": 0.11357971280813217,
      "learning_rate": 9.102224229959938e-05,
      "loss": 1.7637,
      "step": 47360
    },
    {
      "epoch": 0.0935191688281277,
      "grad_norm": 0.1016388013958931,
      "learning_rate": 9.10159020752347e-05,
      "loss": 1.7649,
      "step": 47392
    },
    {
      "epoch": 0.09358231478952415,
      "grad_norm": 0.09839114546775818,
      "learning_rate": 9.100956185087001e-05,
      "loss": 1.7532,
      "step": 47424
    },
    {
      "epoch": 0.09364546075092059,
      "grad_norm": 0.09849175065755844,
      "learning_rate": 9.100322162650531e-05,
      "loss": 1.7457,
      "step": 47456
    },
    {
      "epoch": 0.09370860671231702,
      "grad_norm": 0.11595937609672546,
      "learning_rate": 9.099688140214063e-05,
      "loss": 1.76,
      "step": 47488
    },
    {
      "epoch": 0.09377175267371347,
      "grad_norm": 0.1018570140004158,
      "learning_rate": 9.099054117777594e-05,
      "loss": 1.7662,
      "step": 47520
    },
    {
      "epoch": 0.09383489863510991,
      "grad_norm": 0.10275154560804367,
      "learning_rate": 9.098420095341124e-05,
      "loss": 1.7672,
      "step": 47552
    },
    {
      "epoch": 0.09389804459650634,
      "grad_norm": 0.09677479416131973,
      "learning_rate": 9.097786072904654e-05,
      "loss": 1.7593,
      "step": 47584
    },
    {
      "epoch": 0.0939611905579028,
      "grad_norm": 0.09905097633600235,
      "learning_rate": 9.097152050468186e-05,
      "loss": 1.7521,
      "step": 47616
    },
    {
      "epoch": 0.09402433651929923,
      "grad_norm": 0.0992780551314354,
      "learning_rate": 9.096518028031717e-05,
      "loss": 1.7695,
      "step": 47648
    },
    {
      "epoch": 0.09408748248069568,
      "grad_norm": 0.10553256422281265,
      "learning_rate": 9.095884005595249e-05,
      "loss": 1.773,
      "step": 47680
    },
    {
      "epoch": 0.09415062844209211,
      "grad_norm": 0.09222009032964706,
      "learning_rate": 9.095249983158779e-05,
      "loss": 1.7375,
      "step": 47712
    },
    {
      "epoch": 0.09421377440348855,
      "grad_norm": 0.09913988411426544,
      "learning_rate": 9.09461596072231e-05,
      "loss": 1.7556,
      "step": 47744
    },
    {
      "epoch": 0.094276920364885,
      "grad_norm": 0.10213114321231842,
      "learning_rate": 9.093981938285842e-05,
      "loss": 1.7566,
      "step": 47776
    },
    {
      "epoch": 0.09434006632628143,
      "grad_norm": 0.09298853576183319,
      "learning_rate": 9.093347915849373e-05,
      "loss": 1.7702,
      "step": 47808
    },
    {
      "epoch": 0.09440321228767787,
      "grad_norm": 0.11833615601062775,
      "learning_rate": 9.092713893412905e-05,
      "loss": 1.7736,
      "step": 47840
    },
    {
      "epoch": 0.09446635824907432,
      "grad_norm": 0.10365059971809387,
      "learning_rate": 9.092079870976435e-05,
      "loss": 1.742,
      "step": 47872
    },
    {
      "epoch": 0.09452950421047075,
      "grad_norm": 0.09631436318159103,
      "learning_rate": 9.091445848539966e-05,
      "loss": 1.7562,
      "step": 47904
    },
    {
      "epoch": 0.09459265017186719,
      "grad_norm": 0.09646230190992355,
      "learning_rate": 9.090811826103498e-05,
      "loss": 1.7812,
      "step": 47936
    },
    {
      "epoch": 0.09465579613326364,
      "grad_norm": 0.12156777083873749,
      "learning_rate": 9.090177803667028e-05,
      "loss": 1.7538,
      "step": 47968
    },
    {
      "epoch": 0.09471894209466007,
      "grad_norm": 0.10942567884922028,
      "learning_rate": 9.089543781230558e-05,
      "loss": 1.7459,
      "step": 48000
    },
    {
      "epoch": 0.09478208805605651,
      "grad_norm": 0.09691857546567917,
      "learning_rate": 9.088909758794089e-05,
      "loss": 1.7617,
      "step": 48032
    },
    {
      "epoch": 0.09484523401745296,
      "grad_norm": 0.09761829674243927,
      "learning_rate": 9.08827573635762e-05,
      "loss": 1.766,
      "step": 48064
    },
    {
      "epoch": 0.09490837997884939,
      "grad_norm": 0.09544821828603745,
      "learning_rate": 9.087641713921152e-05,
      "loss": 1.7778,
      "step": 48096
    },
    {
      "epoch": 0.09497152594024583,
      "grad_norm": 0.09921257942914963,
      "learning_rate": 9.087007691484682e-05,
      "loss": 1.747,
      "step": 48128
    },
    {
      "epoch": 0.09503467190164228,
      "grad_norm": 0.0977461040019989,
      "learning_rate": 9.086373669048214e-05,
      "loss": 1.7597,
      "step": 48160
    },
    {
      "epoch": 0.09509781786303871,
      "grad_norm": 0.10993138700723648,
      "learning_rate": 9.085739646611745e-05,
      "loss": 1.7586,
      "step": 48192
    },
    {
      "epoch": 0.09516096382443515,
      "grad_norm": 0.10029158741235733,
      "learning_rate": 9.085105624175277e-05,
      "loss": 1.7691,
      "step": 48224
    },
    {
      "epoch": 0.0952241097858316,
      "grad_norm": 0.09948234260082245,
      "learning_rate": 9.084471601738807e-05,
      "loss": 1.7622,
      "step": 48256
    },
    {
      "epoch": 0.09528725574722803,
      "grad_norm": 0.10209999978542328,
      "learning_rate": 9.083837579302338e-05,
      "loss": 1.7573,
      "step": 48288
    },
    {
      "epoch": 0.09535040170862447,
      "grad_norm": 0.11497338861227036,
      "learning_rate": 9.08320355686587e-05,
      "loss": 1.7633,
      "step": 48320
    },
    {
      "epoch": 0.09541354767002092,
      "grad_norm": 0.09922916442155838,
      "learning_rate": 9.082569534429401e-05,
      "loss": 1.7697,
      "step": 48352
    },
    {
      "epoch": 0.09547669363141735,
      "grad_norm": 0.09739507734775543,
      "learning_rate": 9.081935511992931e-05,
      "loss": 1.7589,
      "step": 48384
    },
    {
      "epoch": 0.09553983959281379,
      "grad_norm": 0.09743262827396393,
      "learning_rate": 9.081301489556461e-05,
      "loss": 1.7704,
      "step": 48416
    },
    {
      "epoch": 0.09560298555421023,
      "grad_norm": 0.10023558139801025,
      "learning_rate": 9.080667467119993e-05,
      "loss": 1.7497,
      "step": 48448
    },
    {
      "epoch": 0.09566613151560667,
      "grad_norm": 0.10142900794744492,
      "learning_rate": 9.080033444683524e-05,
      "loss": 1.7444,
      "step": 48480
    },
    {
      "epoch": 0.09572927747700312,
      "grad_norm": 0.09515956044197083,
      "learning_rate": 9.079399422247056e-05,
      "loss": 1.7384,
      "step": 48512
    },
    {
      "epoch": 0.09579242343839955,
      "grad_norm": 0.10556641221046448,
      "learning_rate": 9.078765399810586e-05,
      "loss": 1.7564,
      "step": 48544
    },
    {
      "epoch": 0.09585556939979599,
      "grad_norm": 0.09588541090488434,
      "learning_rate": 9.078131377374117e-05,
      "loss": 1.7479,
      "step": 48576
    },
    {
      "epoch": 0.09591871536119244,
      "grad_norm": 0.09864659607410431,
      "learning_rate": 9.077497354937649e-05,
      "loss": 1.7538,
      "step": 48608
    },
    {
      "epoch": 0.09598186132258887,
      "grad_norm": 0.10459697991609573,
      "learning_rate": 9.07686333250118e-05,
      "loss": 1.7524,
      "step": 48640
    },
    {
      "epoch": 0.09604500728398531,
      "grad_norm": 0.09505860507488251,
      "learning_rate": 9.07622931006471e-05,
      "loss": 1.7566,
      "step": 48672
    },
    {
      "epoch": 0.09610815324538176,
      "grad_norm": 0.09968584775924683,
      "learning_rate": 9.075595287628242e-05,
      "loss": 1.7651,
      "step": 48704
    },
    {
      "epoch": 0.0961712992067782,
      "grad_norm": 0.1054338738322258,
      "learning_rate": 9.074961265191773e-05,
      "loss": 1.7603,
      "step": 48736
    },
    {
      "epoch": 0.09623444516817463,
      "grad_norm": 0.10174793750047684,
      "learning_rate": 9.074327242755305e-05,
      "loss": 1.7536,
      "step": 48768
    },
    {
      "epoch": 0.09629759112957108,
      "grad_norm": 0.10760358721017838,
      "learning_rate": 9.073693220318835e-05,
      "loss": 1.7307,
      "step": 48800
    },
    {
      "epoch": 0.09636073709096751,
      "grad_norm": 0.0917457640171051,
      "learning_rate": 9.073059197882365e-05,
      "loss": 1.7466,
      "step": 48832
    },
    {
      "epoch": 0.09642388305236395,
      "grad_norm": 0.09406810998916626,
      "learning_rate": 9.072425175445896e-05,
      "loss": 1.7496,
      "step": 48864
    },
    {
      "epoch": 0.0964870290137604,
      "grad_norm": 0.09314677119255066,
      "learning_rate": 9.071791153009428e-05,
      "loss": 1.7507,
      "step": 48896
    },
    {
      "epoch": 0.09655017497515683,
      "grad_norm": 0.0993109866976738,
      "learning_rate": 9.071157130572958e-05,
      "loss": 1.7444,
      "step": 48928
    },
    {
      "epoch": 0.09661332093655327,
      "grad_norm": 0.09146977216005325,
      "learning_rate": 9.070523108136489e-05,
      "loss": 1.7481,
      "step": 48960
    },
    {
      "epoch": 0.09667646689794972,
      "grad_norm": 0.09343001991510391,
      "learning_rate": 9.069889085700021e-05,
      "loss": 1.7532,
      "step": 48992
    },
    {
      "epoch": 0.09673961285934615,
      "grad_norm": 0.09770987927913666,
      "learning_rate": 9.069255063263552e-05,
      "loss": 1.7599,
      "step": 49024
    },
    {
      "epoch": 0.09680275882074259,
      "grad_norm": 0.10346570611000061,
      "learning_rate": 9.068621040827082e-05,
      "loss": 1.7743,
      "step": 49056
    },
    {
      "epoch": 0.09686590478213904,
      "grad_norm": 0.11342669278383255,
      "learning_rate": 9.067987018390614e-05,
      "loss": 1.7583,
      "step": 49088
    },
    {
      "epoch": 0.09692905074353547,
      "grad_norm": 0.09998240321874619,
      "learning_rate": 9.067352995954145e-05,
      "loss": 1.7566,
      "step": 49120
    },
    {
      "epoch": 0.09699219670493191,
      "grad_norm": 0.10686476528644562,
      "learning_rate": 9.066718973517677e-05,
      "loss": 1.7471,
      "step": 49152
    },
    {
      "epoch": 0.09705534266632836,
      "grad_norm": 0.09399770945310593,
      "learning_rate": 9.066084951081207e-05,
      "loss": 1.7476,
      "step": 49184
    },
    {
      "epoch": 0.09711848862772479,
      "grad_norm": 0.1129143163561821,
      "learning_rate": 9.065450928644738e-05,
      "loss": 1.7607,
      "step": 49216
    },
    {
      "epoch": 0.09718163458912123,
      "grad_norm": 0.10864224284887314,
      "learning_rate": 9.064816906208268e-05,
      "loss": 1.7519,
      "step": 49248
    },
    {
      "epoch": 0.09724478055051768,
      "grad_norm": 0.10117150098085403,
      "learning_rate": 9.0641828837718e-05,
      "loss": 1.7462,
      "step": 49280
    },
    {
      "epoch": 0.09730792651191411,
      "grad_norm": 0.1072075292468071,
      "learning_rate": 9.063548861335331e-05,
      "loss": 1.7569,
      "step": 49312
    },
    {
      "epoch": 0.09737107247331056,
      "grad_norm": 0.10172448307275772,
      "learning_rate": 9.062914838898861e-05,
      "loss": 1.7662,
      "step": 49344
    },
    {
      "epoch": 0.097434218434707,
      "grad_norm": 0.10960230231285095,
      "learning_rate": 9.062280816462393e-05,
      "loss": 1.7557,
      "step": 49376
    },
    {
      "epoch": 0.09749736439610343,
      "grad_norm": 0.10693272203207016,
      "learning_rate": 9.061646794025924e-05,
      "loss": 1.7604,
      "step": 49408
    },
    {
      "epoch": 0.09756051035749988,
      "grad_norm": 0.1062304675579071,
      "learning_rate": 9.061012771589456e-05,
      "loss": 1.7713,
      "step": 49440
    },
    {
      "epoch": 0.09762365631889632,
      "grad_norm": 0.11206618696451187,
      "learning_rate": 9.060378749152986e-05,
      "loss": 1.7575,
      "step": 49472
    },
    {
      "epoch": 0.09768680228029275,
      "grad_norm": 0.09645897150039673,
      "learning_rate": 9.059744726716517e-05,
      "loss": 1.7553,
      "step": 49504
    },
    {
      "epoch": 0.0977499482416892,
      "grad_norm": 0.10010252147912979,
      "learning_rate": 9.059110704280049e-05,
      "loss": 1.7536,
      "step": 49536
    },
    {
      "epoch": 0.09781309420308563,
      "grad_norm": 0.09469565749168396,
      "learning_rate": 9.05847668184358e-05,
      "loss": 1.7515,
      "step": 49568
    },
    {
      "epoch": 0.09787624016448207,
      "grad_norm": 0.09600904583930969,
      "learning_rate": 9.05784265940711e-05,
      "loss": 1.7476,
      "step": 49600
    },
    {
      "epoch": 0.09793938612587852,
      "grad_norm": 0.10906678438186646,
      "learning_rate": 9.057208636970642e-05,
      "loss": 1.767,
      "step": 49632
    },
    {
      "epoch": 0.09800253208727495,
      "grad_norm": 0.09677731245756149,
      "learning_rate": 9.056574614534172e-05,
      "loss": 1.7581,
      "step": 49664
    },
    {
      "epoch": 0.09806567804867139,
      "grad_norm": 0.10066740214824677,
      "learning_rate": 9.055940592097703e-05,
      "loss": 1.7547,
      "step": 49696
    },
    {
      "epoch": 0.09812882401006784,
      "grad_norm": 0.10452299565076828,
      "learning_rate": 9.055306569661233e-05,
      "loss": 1.7583,
      "step": 49728
    },
    {
      "epoch": 0.09819196997146427,
      "grad_norm": 0.09318119287490845,
      "learning_rate": 9.054672547224765e-05,
      "loss": 1.7407,
      "step": 49760
    },
    {
      "epoch": 0.09825511593286071,
      "grad_norm": 0.11290091276168823,
      "learning_rate": 9.054038524788296e-05,
      "loss": 1.7462,
      "step": 49792
    },
    {
      "epoch": 0.09831826189425716,
      "grad_norm": 0.1076420322060585,
      "learning_rate": 9.053404502351828e-05,
      "loss": 1.7718,
      "step": 49824
    },
    {
      "epoch": 0.0983814078556536,
      "grad_norm": 0.10183405876159668,
      "learning_rate": 9.052770479915359e-05,
      "loss": 1.7366,
      "step": 49856
    },
    {
      "epoch": 0.09844455381705003,
      "grad_norm": 0.1181720644235611,
      "learning_rate": 9.052136457478889e-05,
      "loss": 1.7686,
      "step": 49888
    },
    {
      "epoch": 0.09850769977844648,
      "grad_norm": 0.09957777708768845,
      "learning_rate": 9.051502435042421e-05,
      "loss": 1.769,
      "step": 49920
    },
    {
      "epoch": 0.09857084573984291,
      "grad_norm": 0.09811808913946152,
      "learning_rate": 9.050868412605952e-05,
      "loss": 1.7455,
      "step": 49952
    },
    {
      "epoch": 0.09863399170123935,
      "grad_norm": 0.11086859554052353,
      "learning_rate": 9.050234390169484e-05,
      "loss": 1.7642,
      "step": 49984
    },
    {
      "epoch": 0.0986971376626358,
      "grad_norm": 0.10157608240842819,
      "learning_rate": 9.049600367733014e-05,
      "loss": 1.759,
      "step": 50016
    },
    {
      "epoch": 0.09876028362403223,
      "grad_norm": 0.0975349023938179,
      "learning_rate": 9.048966345296545e-05,
      "loss": 1.7359,
      "step": 50048
    },
    {
      "epoch": 0.09882342958542867,
      "grad_norm": 0.09794504195451736,
      "learning_rate": 9.048332322860075e-05,
      "loss": 1.7671,
      "step": 50080
    },
    {
      "epoch": 0.09888657554682512,
      "grad_norm": 0.10445244610309601,
      "learning_rate": 9.047698300423607e-05,
      "loss": 1.7574,
      "step": 50112
    },
    {
      "epoch": 0.09894972150822155,
      "grad_norm": 0.11278077960014343,
      "learning_rate": 9.047064277987137e-05,
      "loss": 1.7454,
      "step": 50144
    },
    {
      "epoch": 0.099012867469618,
      "grad_norm": 0.10108131170272827,
      "learning_rate": 9.046430255550668e-05,
      "loss": 1.7542,
      "step": 50176
    },
    {
      "epoch": 0.09907601343101444,
      "grad_norm": 0.0978446826338768,
      "learning_rate": 9.0457962331142e-05,
      "loss": 1.7593,
      "step": 50208
    },
    {
      "epoch": 0.09913915939241087,
      "grad_norm": 0.10433097183704376,
      "learning_rate": 9.045162210677731e-05,
      "loss": 1.7427,
      "step": 50240
    },
    {
      "epoch": 0.09920230535380732,
      "grad_norm": 0.10728266835212708,
      "learning_rate": 9.044528188241261e-05,
      "loss": 1.7432,
      "step": 50272
    },
    {
      "epoch": 0.09926545131520376,
      "grad_norm": 0.10774000734090805,
      "learning_rate": 9.043894165804793e-05,
      "loss": 1.7526,
      "step": 50304
    },
    {
      "epoch": 0.09932859727660019,
      "grad_norm": 0.1002371534705162,
      "learning_rate": 9.043260143368324e-05,
      "loss": 1.7552,
      "step": 50336
    },
    {
      "epoch": 0.09939174323799664,
      "grad_norm": 0.09350822120904922,
      "learning_rate": 9.042626120931856e-05,
      "loss": 1.7462,
      "step": 50368
    },
    {
      "epoch": 0.09945488919939308,
      "grad_norm": 0.09911242127418518,
      "learning_rate": 9.041992098495386e-05,
      "loss": 1.763,
      "step": 50400
    },
    {
      "epoch": 0.09951803516078951,
      "grad_norm": 0.09527602791786194,
      "learning_rate": 9.041358076058917e-05,
      "loss": 1.7546,
      "step": 50432
    },
    {
      "epoch": 0.09958118112218596,
      "grad_norm": 0.11248945444822311,
      "learning_rate": 9.040724053622447e-05,
      "loss": 1.7441,
      "step": 50464
    },
    {
      "epoch": 0.0996443270835824,
      "grad_norm": 0.0988423228263855,
      "learning_rate": 9.040090031185979e-05,
      "loss": 1.7575,
      "step": 50496
    },
    {
      "epoch": 0.09970747304497883,
      "grad_norm": 0.09794781357049942,
      "learning_rate": 9.03945600874951e-05,
      "loss": 1.7613,
      "step": 50528
    },
    {
      "epoch": 0.09977061900637528,
      "grad_norm": 0.09636078774929047,
      "learning_rate": 9.03882198631304e-05,
      "loss": 1.734,
      "step": 50560
    },
    {
      "epoch": 0.09983376496777172,
      "grad_norm": 0.10354897379875183,
      "learning_rate": 9.038187963876572e-05,
      "loss": 1.7527,
      "step": 50592
    },
    {
      "epoch": 0.09989691092916815,
      "grad_norm": 0.10228206217288971,
      "learning_rate": 9.037553941440103e-05,
      "loss": 1.757,
      "step": 50624
    },
    {
      "epoch": 0.0999600568905646,
      "grad_norm": 0.09956642240285873,
      "learning_rate": 9.036919919003635e-05,
      "loss": 1.7482,
      "step": 50656
    },
    {
      "epoch": 0.10002320285196103,
      "grad_norm": 0.09959997236728668,
      "learning_rate": 9.036285896567165e-05,
      "loss": 1.7565,
      "step": 50688
    },
    {
      "epoch": 0.10008634881335747,
      "grad_norm": 0.09238331764936447,
      "learning_rate": 9.035651874130696e-05,
      "loss": 1.7331,
      "step": 50720
    },
    {
      "epoch": 0.10014949477475392,
      "grad_norm": 0.1037338376045227,
      "learning_rate": 9.035017851694228e-05,
      "loss": 1.752,
      "step": 50752
    },
    {
      "epoch": 0.10021264073615035,
      "grad_norm": 0.09862665832042694,
      "learning_rate": 9.034383829257759e-05,
      "loss": 1.7587,
      "step": 50784
    },
    {
      "epoch": 0.10027578669754679,
      "grad_norm": 0.10530706495046616,
      "learning_rate": 9.033749806821289e-05,
      "loss": 1.7563,
      "step": 50816
    },
    {
      "epoch": 0.10033893265894324,
      "grad_norm": 0.11070514470338821,
      "learning_rate": 9.033115784384821e-05,
      "loss": 1.7535,
      "step": 50848
    },
    {
      "epoch": 0.10040207862033967,
      "grad_norm": 0.1009884625673294,
      "learning_rate": 9.032481761948351e-05,
      "loss": 1.7684,
      "step": 50880
    },
    {
      "epoch": 0.10046522458173611,
      "grad_norm": 0.09597666561603546,
      "learning_rate": 9.031847739511882e-05,
      "loss": 1.7538,
      "step": 50912
    },
    {
      "epoch": 0.10052837054313256,
      "grad_norm": 0.09755109995603561,
      "learning_rate": 9.031213717075412e-05,
      "loss": 1.7545,
      "step": 50944
    },
    {
      "epoch": 0.100591516504529,
      "grad_norm": 0.12461982667446136,
      "learning_rate": 9.030579694638944e-05,
      "loss": 1.7583,
      "step": 50976
    },
    {
      "epoch": 0.10065466246592544,
      "grad_norm": 0.09848106652498245,
      "learning_rate": 9.029945672202475e-05,
      "loss": 1.7576,
      "step": 51008
    },
    {
      "epoch": 0.10071780842732188,
      "grad_norm": 0.09583540260791779,
      "learning_rate": 9.029311649766007e-05,
      "loss": 1.7342,
      "step": 51040
    },
    {
      "epoch": 0.10078095438871831,
      "grad_norm": 0.09553193300962448,
      "learning_rate": 9.028677627329537e-05,
      "loss": 1.7515,
      "step": 51072
    },
    {
      "epoch": 0.10084410035011476,
      "grad_norm": 0.10564205050468445,
      "learning_rate": 9.028043604893068e-05,
      "loss": 1.7594,
      "step": 51104
    },
    {
      "epoch": 0.1009072463115112,
      "grad_norm": 0.11469171941280365,
      "learning_rate": 9.0274095824566e-05,
      "loss": 1.7479,
      "step": 51136
    },
    {
      "epoch": 0.10097039227290763,
      "grad_norm": 0.09900109469890594,
      "learning_rate": 9.026775560020131e-05,
      "loss": 1.7549,
      "step": 51168
    },
    {
      "epoch": 0.10103353823430408,
      "grad_norm": 0.10750380903482437,
      "learning_rate": 9.026141537583663e-05,
      "loss": 1.7587,
      "step": 51200
    },
    {
      "epoch": 0.10109668419570052,
      "grad_norm": 0.09829618036746979,
      "learning_rate": 9.025507515147193e-05,
      "loss": 1.7375,
      "step": 51232
    },
    {
      "epoch": 0.10115983015709695,
      "grad_norm": 0.09377654641866684,
      "learning_rate": 9.024873492710724e-05,
      "loss": 1.7531,
      "step": 51264
    },
    {
      "epoch": 0.1012229761184934,
      "grad_norm": 0.09642499685287476,
      "learning_rate": 9.024239470274254e-05,
      "loss": 1.7497,
      "step": 51296
    },
    {
      "epoch": 0.10128612207988984,
      "grad_norm": 0.09268879145383835,
      "learning_rate": 9.023605447837786e-05,
      "loss": 1.7363,
      "step": 51328
    },
    {
      "epoch": 0.10134926804128627,
      "grad_norm": 0.10005810856819153,
      "learning_rate": 9.022971425401316e-05,
      "loss": 1.7644,
      "step": 51360
    },
    {
      "epoch": 0.10141241400268272,
      "grad_norm": 0.1035057082772255,
      "learning_rate": 9.022337402964847e-05,
      "loss": 1.7439,
      "step": 51392
    },
    {
      "epoch": 0.10147555996407916,
      "grad_norm": 0.09726722538471222,
      "learning_rate": 9.021703380528379e-05,
      "loss": 1.7447,
      "step": 51424
    },
    {
      "epoch": 0.10153870592547559,
      "grad_norm": 0.11155097931623459,
      "learning_rate": 9.02106935809191e-05,
      "loss": 1.7487,
      "step": 51456
    },
    {
      "epoch": 0.10160185188687204,
      "grad_norm": 0.09736192971467972,
      "learning_rate": 9.02043533565544e-05,
      "loss": 1.7633,
      "step": 51488
    },
    {
      "epoch": 0.10166499784826848,
      "grad_norm": 0.09542526304721832,
      "learning_rate": 9.019801313218972e-05,
      "loss": 1.7509,
      "step": 51520
    },
    {
      "epoch": 0.10172814380966491,
      "grad_norm": 0.10050925612449646,
      "learning_rate": 9.019167290782503e-05,
      "loss": 1.7541,
      "step": 51552
    },
    {
      "epoch": 0.10179128977106136,
      "grad_norm": 0.09684551507234573,
      "learning_rate": 9.018533268346035e-05,
      "loss": 1.7556,
      "step": 51584
    },
    {
      "epoch": 0.1018544357324578,
      "grad_norm": 0.09515704214572906,
      "learning_rate": 9.017899245909565e-05,
      "loss": 1.7416,
      "step": 51616
    },
    {
      "epoch": 0.10191758169385423,
      "grad_norm": 0.10582724958658218,
      "learning_rate": 9.017265223473096e-05,
      "loss": 1.7554,
      "step": 51648
    },
    {
      "epoch": 0.10198072765525068,
      "grad_norm": 0.131036639213562,
      "learning_rate": 9.016631201036628e-05,
      "loss": 1.7368,
      "step": 51680
    },
    {
      "epoch": 0.10204387361664712,
      "grad_norm": 0.09360115230083466,
      "learning_rate": 9.015997178600158e-05,
      "loss": 1.7453,
      "step": 51712
    },
    {
      "epoch": 0.10210701957804355,
      "grad_norm": 0.10363040864467621,
      "learning_rate": 9.01536315616369e-05,
      "loss": 1.7382,
      "step": 51744
    },
    {
      "epoch": 0.10217016553944,
      "grad_norm": 0.10199641436338425,
      "learning_rate": 9.01472913372722e-05,
      "loss": 1.7367,
      "step": 51776
    },
    {
      "epoch": 0.10223331150083644,
      "grad_norm": 0.09684382379055023,
      "learning_rate": 9.014095111290751e-05,
      "loss": 1.7474,
      "step": 51808
    },
    {
      "epoch": 0.10229645746223288,
      "grad_norm": 0.0927148312330246,
      "learning_rate": 9.013461088854282e-05,
      "loss": 1.7359,
      "step": 51840
    },
    {
      "epoch": 0.10235960342362932,
      "grad_norm": 0.0999477207660675,
      "learning_rate": 9.012827066417814e-05,
      "loss": 1.7423,
      "step": 51872
    },
    {
      "epoch": 0.10242274938502575,
      "grad_norm": 0.11755125969648361,
      "learning_rate": 9.012193043981344e-05,
      "loss": 1.728,
      "step": 51904
    },
    {
      "epoch": 0.1024858953464222,
      "grad_norm": 0.10312087833881378,
      "learning_rate": 9.011559021544875e-05,
      "loss": 1.7523,
      "step": 51936
    },
    {
      "epoch": 0.10254904130781864,
      "grad_norm": 0.10827986896038055,
      "learning_rate": 9.010924999108407e-05,
      "loss": 1.7375,
      "step": 51968
    },
    {
      "epoch": 0.10261218726921507,
      "grad_norm": 0.09504279494285583,
      "learning_rate": 9.010290976671938e-05,
      "loss": 1.7525,
      "step": 52000
    },
    {
      "epoch": 0.10267533323061152,
      "grad_norm": 0.10902871936559677,
      "learning_rate": 9.009656954235468e-05,
      "loss": 1.7462,
      "step": 52032
    },
    {
      "epoch": 0.10273847919200796,
      "grad_norm": 0.09501804411411285,
      "learning_rate": 9.009022931799e-05,
      "loss": 1.7481,
      "step": 52064
    },
    {
      "epoch": 0.1028016251534044,
      "grad_norm": 0.10616398602724075,
      "learning_rate": 9.008388909362531e-05,
      "loss": 1.734,
      "step": 52096
    },
    {
      "epoch": 0.10286477111480084,
      "grad_norm": 0.10708317905664444,
      "learning_rate": 9.007754886926061e-05,
      "loss": 1.756,
      "step": 52128
    },
    {
      "epoch": 0.10292791707619728,
      "grad_norm": 0.09232480823993683,
      "learning_rate": 9.007120864489591e-05,
      "loss": 1.7443,
      "step": 52160
    },
    {
      "epoch": 0.10299106303759371,
      "grad_norm": 0.0943201407790184,
      "learning_rate": 9.006486842053123e-05,
      "loss": 1.7603,
      "step": 52192
    },
    {
      "epoch": 0.10305420899899016,
      "grad_norm": 0.10176809132099152,
      "learning_rate": 9.005852819616654e-05,
      "loss": 1.7408,
      "step": 52224
    },
    {
      "epoch": 0.1031173549603866,
      "grad_norm": 0.10179544240236282,
      "learning_rate": 9.005218797180186e-05,
      "loss": 1.7465,
      "step": 52256
    },
    {
      "epoch": 0.10318050092178303,
      "grad_norm": 0.09759853035211563,
      "learning_rate": 9.004584774743716e-05,
      "loss": 1.7493,
      "step": 52288
    },
    {
      "epoch": 0.10324364688317948,
      "grad_norm": 0.10105500370264053,
      "learning_rate": 9.003950752307247e-05,
      "loss": 1.7492,
      "step": 52320
    },
    {
      "epoch": 0.10330679284457592,
      "grad_norm": 0.10510502755641937,
      "learning_rate": 9.003316729870779e-05,
      "loss": 1.7401,
      "step": 52352
    },
    {
      "epoch": 0.10336993880597235,
      "grad_norm": 0.09681029617786407,
      "learning_rate": 9.00268270743431e-05,
      "loss": 1.7497,
      "step": 52384
    },
    {
      "epoch": 0.1034330847673688,
      "grad_norm": 0.09967440366744995,
      "learning_rate": 9.00204868499784e-05,
      "loss": 1.7477,
      "step": 52416
    },
    {
      "epoch": 0.10349623072876524,
      "grad_norm": 0.0955488458275795,
      "learning_rate": 9.001414662561372e-05,
      "loss": 1.7481,
      "step": 52448
    },
    {
      "epoch": 0.10355937669016167,
      "grad_norm": 0.10663046687841415,
      "learning_rate": 9.000780640124903e-05,
      "loss": 1.741,
      "step": 52480
    },
    {
      "epoch": 0.10362252265155812,
      "grad_norm": 0.09638260304927826,
      "learning_rate": 9.000146617688435e-05,
      "loss": 1.7371,
      "step": 52512
    },
    {
      "epoch": 0.10368566861295456,
      "grad_norm": 0.11361704021692276,
      "learning_rate": 8.999512595251965e-05,
      "loss": 1.7477,
      "step": 52544
    },
    {
      "epoch": 0.10374881457435099,
      "grad_norm": 0.09574690461158752,
      "learning_rate": 8.998878572815495e-05,
      "loss": 1.7409,
      "step": 52576
    },
    {
      "epoch": 0.10381196053574744,
      "grad_norm": 0.1048581600189209,
      "learning_rate": 8.998244550379026e-05,
      "loss": 1.742,
      "step": 52608
    },
    {
      "epoch": 0.10387510649714388,
      "grad_norm": 0.09773201495409012,
      "learning_rate": 8.997610527942558e-05,
      "loss": 1.7472,
      "step": 52640
    },
    {
      "epoch": 0.10393825245854031,
      "grad_norm": 0.09244991838932037,
      "learning_rate": 8.99697650550609e-05,
      "loss": 1.7266,
      "step": 52672
    },
    {
      "epoch": 0.10400139841993676,
      "grad_norm": 0.10884491354227066,
      "learning_rate": 8.99634248306962e-05,
      "loss": 1.7568,
      "step": 52704
    },
    {
      "epoch": 0.1040645443813332,
      "grad_norm": 0.09903937578201294,
      "learning_rate": 8.995708460633151e-05,
      "loss": 1.7393,
      "step": 52736
    },
    {
      "epoch": 0.10412769034272965,
      "grad_norm": 0.10828418284654617,
      "learning_rate": 8.995074438196682e-05,
      "loss": 1.7463,
      "step": 52768
    },
    {
      "epoch": 0.10419083630412608,
      "grad_norm": 0.10341514647006989,
      "learning_rate": 8.994440415760214e-05,
      "loss": 1.7403,
      "step": 52800
    },
    {
      "epoch": 0.10425398226552252,
      "grad_norm": 0.09311387687921524,
      "learning_rate": 8.993806393323744e-05,
      "loss": 1.7469,
      "step": 52832
    },
    {
      "epoch": 0.10431712822691896,
      "grad_norm": 0.1073962152004242,
      "learning_rate": 8.993172370887275e-05,
      "loss": 1.7437,
      "step": 52864
    },
    {
      "epoch": 0.1043802741883154,
      "grad_norm": 0.10534194856882095,
      "learning_rate": 8.992538348450807e-05,
      "loss": 1.7476,
      "step": 52896
    },
    {
      "epoch": 0.10444342014971184,
      "grad_norm": 0.11144720017910004,
      "learning_rate": 8.991904326014338e-05,
      "loss": 1.7348,
      "step": 52928
    },
    {
      "epoch": 0.10450656611110828,
      "grad_norm": 0.09449729323387146,
      "learning_rate": 8.991270303577868e-05,
      "loss": 1.7415,
      "step": 52960
    },
    {
      "epoch": 0.10456971207250472,
      "grad_norm": 0.1120440736413002,
      "learning_rate": 8.990636281141399e-05,
      "loss": 1.7415,
      "step": 52992
    },
    {
      "epoch": 0.10463285803390115,
      "grad_norm": 0.1097162514925003,
      "learning_rate": 8.99000225870493e-05,
      "loss": 1.7453,
      "step": 53024
    },
    {
      "epoch": 0.1046960039952976,
      "grad_norm": 0.09272979199886322,
      "learning_rate": 8.989368236268461e-05,
      "loss": 1.7514,
      "step": 53056
    },
    {
      "epoch": 0.10475914995669404,
      "grad_norm": 0.10094426572322845,
      "learning_rate": 8.988734213831992e-05,
      "loss": 1.7378,
      "step": 53088
    },
    {
      "epoch": 0.10482229591809047,
      "grad_norm": 0.11835931241512299,
      "learning_rate": 8.988100191395523e-05,
      "loss": 1.742,
      "step": 53120
    },
    {
      "epoch": 0.10488544187948692,
      "grad_norm": 0.09697704017162323,
      "learning_rate": 8.987466168959054e-05,
      "loss": 1.7373,
      "step": 53152
    },
    {
      "epoch": 0.10494858784088336,
      "grad_norm": 0.1052129715681076,
      "learning_rate": 8.986832146522586e-05,
      "loss": 1.732,
      "step": 53184
    },
    {
      "epoch": 0.1050117338022798,
      "grad_norm": 0.09859216213226318,
      "learning_rate": 8.986198124086117e-05,
      "loss": 1.7252,
      "step": 53216
    },
    {
      "epoch": 0.10507487976367624,
      "grad_norm": 0.09212696552276611,
      "learning_rate": 8.985564101649647e-05,
      "loss": 1.7544,
      "step": 53248
    },
    {
      "epoch": 0.10513802572507268,
      "grad_norm": 0.11304720491170883,
      "learning_rate": 8.984930079213179e-05,
      "loss": 1.7486,
      "step": 53280
    },
    {
      "epoch": 0.10520117168646911,
      "grad_norm": 0.09030237048864365,
      "learning_rate": 8.98429605677671e-05,
      "loss": 1.7276,
      "step": 53312
    },
    {
      "epoch": 0.10526431764786556,
      "grad_norm": 0.09863579273223877,
      "learning_rate": 8.98366203434024e-05,
      "loss": 1.7398,
      "step": 53344
    },
    {
      "epoch": 0.105327463609262,
      "grad_norm": 0.1069789007306099,
      "learning_rate": 8.983028011903772e-05,
      "loss": 1.7422,
      "step": 53376
    },
    {
      "epoch": 0.10539060957065843,
      "grad_norm": 0.11715696007013321,
      "learning_rate": 8.982393989467302e-05,
      "loss": 1.7258,
      "step": 53408
    },
    {
      "epoch": 0.10545375553205488,
      "grad_norm": 0.10883591324090958,
      "learning_rate": 8.981759967030833e-05,
      "loss": 1.7326,
      "step": 53440
    },
    {
      "epoch": 0.10551690149345132,
      "grad_norm": 0.09361957013607025,
      "learning_rate": 8.981125944594365e-05,
      "loss": 1.7437,
      "step": 53472
    },
    {
      "epoch": 0.10558004745484775,
      "grad_norm": 0.09470156580209732,
      "learning_rate": 8.980491922157895e-05,
      "loss": 1.7413,
      "step": 53504
    },
    {
      "epoch": 0.1056431934162442,
      "grad_norm": 0.09954388439655304,
      "learning_rate": 8.979857899721426e-05,
      "loss": 1.744,
      "step": 53536
    },
    {
      "epoch": 0.10570633937764064,
      "grad_norm": 0.09837543964385986,
      "learning_rate": 8.979223877284958e-05,
      "loss": 1.7267,
      "step": 53568
    },
    {
      "epoch": 0.10576948533903709,
      "grad_norm": 0.10575970262289047,
      "learning_rate": 8.97858985484849e-05,
      "loss": 1.7402,
      "step": 53600
    },
    {
      "epoch": 0.10583263130043352,
      "grad_norm": 0.09170810878276825,
      "learning_rate": 8.97795583241202e-05,
      "loss": 1.7413,
      "step": 53632
    },
    {
      "epoch": 0.10589577726182996,
      "grad_norm": 0.09788578003644943,
      "learning_rate": 8.977321809975551e-05,
      "loss": 1.744,
      "step": 53664
    },
    {
      "epoch": 0.1059589232232264,
      "grad_norm": 0.09518428891897202,
      "learning_rate": 8.976687787539082e-05,
      "loss": 1.7404,
      "step": 53696
    },
    {
      "epoch": 0.10602206918462284,
      "grad_norm": 0.09772198647260666,
      "learning_rate": 8.976053765102614e-05,
      "loss": 1.7378,
      "step": 53728
    },
    {
      "epoch": 0.10608521514601928,
      "grad_norm": 0.09868532419204712,
      "learning_rate": 8.975419742666144e-05,
      "loss": 1.7313,
      "step": 53760
    },
    {
      "epoch": 0.10614836110741573,
      "grad_norm": 0.09947213530540466,
      "learning_rate": 8.974785720229675e-05,
      "loss": 1.7266,
      "step": 53792
    },
    {
      "epoch": 0.10621150706881216,
      "grad_norm": 0.09919548779726028,
      "learning_rate": 8.974151697793206e-05,
      "loss": 1.7165,
      "step": 53824
    },
    {
      "epoch": 0.1062746530302086,
      "grad_norm": 0.09903539717197418,
      "learning_rate": 8.973517675356737e-05,
      "loss": 1.7326,
      "step": 53856
    },
    {
      "epoch": 0.10633779899160505,
      "grad_norm": 0.1100149005651474,
      "learning_rate": 8.972883652920268e-05,
      "loss": 1.7391,
      "step": 53888
    },
    {
      "epoch": 0.10640094495300148,
      "grad_norm": 0.10203427821397781,
      "learning_rate": 8.972249630483799e-05,
      "loss": 1.7243,
      "step": 53920
    },
    {
      "epoch": 0.10646409091439792,
      "grad_norm": 0.10099450498819351,
      "learning_rate": 8.97161560804733e-05,
      "loss": 1.7305,
      "step": 53952
    },
    {
      "epoch": 0.10652723687579436,
      "grad_norm": 0.0974942222237587,
      "learning_rate": 8.970981585610861e-05,
      "loss": 1.7459,
      "step": 53984
    },
    {
      "epoch": 0.1065903828371908,
      "grad_norm": 0.1016789898276329,
      "learning_rate": 8.970347563174393e-05,
      "loss": 1.7316,
      "step": 54016
    },
    {
      "epoch": 0.10665352879858724,
      "grad_norm": 0.10867282003164291,
      "learning_rate": 8.969713540737923e-05,
      "loss": 1.7457,
      "step": 54048
    },
    {
      "epoch": 0.10671667475998368,
      "grad_norm": 0.10432814061641693,
      "learning_rate": 8.969079518301454e-05,
      "loss": 1.739,
      "step": 54080
    },
    {
      "epoch": 0.10677982072138012,
      "grad_norm": 0.10358621925115585,
      "learning_rate": 8.968445495864986e-05,
      "loss": 1.7187,
      "step": 54112
    },
    {
      "epoch": 0.10684296668277656,
      "grad_norm": 0.11256913095712662,
      "learning_rate": 8.967811473428517e-05,
      "loss": 1.7448,
      "step": 54144
    },
    {
      "epoch": 0.106906112644173,
      "grad_norm": 0.09977933764457703,
      "learning_rate": 8.967177450992047e-05,
      "loss": 1.7367,
      "step": 54176
    },
    {
      "epoch": 0.10696925860556944,
      "grad_norm": 0.09639386832714081,
      "learning_rate": 8.966543428555579e-05,
      "loss": 1.7493,
      "step": 54208
    },
    {
      "epoch": 0.10703240456696587,
      "grad_norm": 0.11345695704221725,
      "learning_rate": 8.965909406119109e-05,
      "loss": 1.7533,
      "step": 54240
    },
    {
      "epoch": 0.10709555052836232,
      "grad_norm": 0.09637809544801712,
      "learning_rate": 8.96527538368264e-05,
      "loss": 1.7382,
      "step": 54272
    },
    {
      "epoch": 0.10715869648975876,
      "grad_norm": 0.1110880970954895,
      "learning_rate": 8.96464136124617e-05,
      "loss": 1.7507,
      "step": 54304
    },
    {
      "epoch": 0.1072218424511552,
      "grad_norm": 0.09914442896842957,
      "learning_rate": 8.964007338809702e-05,
      "loss": 1.7505,
      "step": 54336
    },
    {
      "epoch": 0.10728498841255164,
      "grad_norm": 0.09976788610219955,
      "learning_rate": 8.963373316373234e-05,
      "loss": 1.7446,
      "step": 54368
    },
    {
      "epoch": 0.10734813437394808,
      "grad_norm": 0.1007564514875412,
      "learning_rate": 8.962739293936765e-05,
      "loss": 1.7352,
      "step": 54400
    },
    {
      "epoch": 0.10741128033534453,
      "grad_norm": 0.09931594133377075,
      "learning_rate": 8.962105271500295e-05,
      "loss": 1.7415,
      "step": 54432
    },
    {
      "epoch": 0.10747442629674096,
      "grad_norm": 0.10601203143596649,
      "learning_rate": 8.961471249063827e-05,
      "loss": 1.7268,
      "step": 54464
    },
    {
      "epoch": 0.1075375722581374,
      "grad_norm": 0.11378839612007141,
      "learning_rate": 8.960837226627358e-05,
      "loss": 1.7416,
      "step": 54496
    },
    {
      "epoch": 0.10760071821953385,
      "grad_norm": 0.09632743895053864,
      "learning_rate": 8.96020320419089e-05,
      "loss": 1.7297,
      "step": 54528
    },
    {
      "epoch": 0.10766386418093028,
      "grad_norm": 0.09724630415439606,
      "learning_rate": 8.959569181754421e-05,
      "loss": 1.7426,
      "step": 54560
    },
    {
      "epoch": 0.10772701014232672,
      "grad_norm": 0.09703235328197479,
      "learning_rate": 8.958935159317951e-05,
      "loss": 1.7295,
      "step": 54592
    },
    {
      "epoch": 0.10779015610372317,
      "grad_norm": 0.11917123943567276,
      "learning_rate": 8.958301136881481e-05,
      "loss": 1.7309,
      "step": 54624
    },
    {
      "epoch": 0.1078533020651196,
      "grad_norm": 0.09530433267354965,
      "learning_rate": 8.957667114445013e-05,
      "loss": 1.7289,
      "step": 54656
    },
    {
      "epoch": 0.10791644802651604,
      "grad_norm": 0.10603784769773483,
      "learning_rate": 8.957033092008544e-05,
      "loss": 1.7499,
      "step": 54688
    },
    {
      "epoch": 0.10797959398791249,
      "grad_norm": 0.0967014953494072,
      "learning_rate": 8.956399069572074e-05,
      "loss": 1.7316,
      "step": 54720
    },
    {
      "epoch": 0.10804273994930892,
      "grad_norm": 0.09749648720026016,
      "learning_rate": 8.955765047135606e-05,
      "loss": 1.7405,
      "step": 54752
    },
    {
      "epoch": 0.10810588591070536,
      "grad_norm": 0.0986323282122612,
      "learning_rate": 8.955131024699137e-05,
      "loss": 1.7389,
      "step": 54784
    },
    {
      "epoch": 0.1081690318721018,
      "grad_norm": 0.09602555632591248,
      "learning_rate": 8.954497002262668e-05,
      "loss": 1.7367,
      "step": 54816
    },
    {
      "epoch": 0.10823217783349824,
      "grad_norm": 0.09541743993759155,
      "learning_rate": 8.953862979826199e-05,
      "loss": 1.7159,
      "step": 54848
    },
    {
      "epoch": 0.10829532379489468,
      "grad_norm": 0.12124884873628616,
      "learning_rate": 8.95322895738973e-05,
      "loss": 1.728,
      "step": 54880
    },
    {
      "epoch": 0.10835846975629113,
      "grad_norm": 0.09567489475011826,
      "learning_rate": 8.952594934953261e-05,
      "loss": 1.7364,
      "step": 54912
    },
    {
      "epoch": 0.10842161571768756,
      "grad_norm": 0.09893244504928589,
      "learning_rate": 8.951960912516793e-05,
      "loss": 1.7361,
      "step": 54944
    },
    {
      "epoch": 0.108484761679084,
      "grad_norm": 0.0997154638171196,
      "learning_rate": 8.951326890080323e-05,
      "loss": 1.7301,
      "step": 54976
    },
    {
      "epoch": 0.10854790764048045,
      "grad_norm": 0.09740784764289856,
      "learning_rate": 8.950692867643854e-05,
      "loss": 1.7363,
      "step": 55008
    },
    {
      "epoch": 0.10861105360187688,
      "grad_norm": 0.09866963326931,
      "learning_rate": 8.950058845207385e-05,
      "loss": 1.7195,
      "step": 55040
    },
    {
      "epoch": 0.10867419956327332,
      "grad_norm": 0.1122496947646141,
      "learning_rate": 8.949424822770916e-05,
      "loss": 1.7322,
      "step": 55072
    },
    {
      "epoch": 0.10873734552466977,
      "grad_norm": 0.11134584248065948,
      "learning_rate": 8.948790800334446e-05,
      "loss": 1.7324,
      "step": 55104
    },
    {
      "epoch": 0.1088004914860662,
      "grad_norm": 0.09876733273267746,
      "learning_rate": 8.948156777897978e-05,
      "loss": 1.7384,
      "step": 55136
    },
    {
      "epoch": 0.10886363744746264,
      "grad_norm": 0.10825956612825394,
      "learning_rate": 8.947522755461509e-05,
      "loss": 1.7386,
      "step": 55168
    },
    {
      "epoch": 0.10892678340885908,
      "grad_norm": 0.09949720650911331,
      "learning_rate": 8.94688873302504e-05,
      "loss": 1.7369,
      "step": 55200
    },
    {
      "epoch": 0.10898992937025552,
      "grad_norm": 0.1249699518084526,
      "learning_rate": 8.946254710588572e-05,
      "loss": 1.7318,
      "step": 55232
    },
    {
      "epoch": 0.10905307533165197,
      "grad_norm": 0.10329774022102356,
      "learning_rate": 8.945620688152102e-05,
      "loss": 1.7334,
      "step": 55264
    },
    {
      "epoch": 0.1091162212930484,
      "grad_norm": 0.09686914086341858,
      "learning_rate": 8.944986665715634e-05,
      "loss": 1.7248,
      "step": 55296
    },
    {
      "epoch": 0.10917936725444484,
      "grad_norm": 0.10577890276908875,
      "learning_rate": 8.944352643279165e-05,
      "loss": 1.7601,
      "step": 55328
    },
    {
      "epoch": 0.10924251321584129,
      "grad_norm": 0.10029914230108261,
      "learning_rate": 8.943718620842696e-05,
      "loss": 1.73,
      "step": 55360
    },
    {
      "epoch": 0.10930565917723772,
      "grad_norm": 0.10715453326702118,
      "learning_rate": 8.943084598406227e-05,
      "loss": 1.7327,
      "step": 55392
    },
    {
      "epoch": 0.10936880513863416,
      "grad_norm": 0.10437731444835663,
      "learning_rate": 8.942450575969758e-05,
      "loss": 1.7187,
      "step": 55424
    },
    {
      "epoch": 0.10943195110003061,
      "grad_norm": 0.09671631455421448,
      "learning_rate": 8.941816553533288e-05,
      "loss": 1.7194,
      "step": 55456
    },
    {
      "epoch": 0.10949509706142704,
      "grad_norm": 0.10065539181232452,
      "learning_rate": 8.94118253109682e-05,
      "loss": 1.7306,
      "step": 55488
    },
    {
      "epoch": 0.10955824302282348,
      "grad_norm": 0.09755487740039825,
      "learning_rate": 8.94054850866035e-05,
      "loss": 1.7344,
      "step": 55520
    },
    {
      "epoch": 0.10962138898421993,
      "grad_norm": 0.09347401559352875,
      "learning_rate": 8.939914486223881e-05,
      "loss": 1.7286,
      "step": 55552
    },
    {
      "epoch": 0.10968453494561636,
      "grad_norm": 0.09633270651102066,
      "learning_rate": 8.939280463787413e-05,
      "loss": 1.7531,
      "step": 55584
    },
    {
      "epoch": 0.1097476809070128,
      "grad_norm": 0.0984179675579071,
      "learning_rate": 8.938646441350944e-05,
      "loss": 1.7392,
      "step": 55616
    },
    {
      "epoch": 0.10981082686840925,
      "grad_norm": 0.09896954894065857,
      "learning_rate": 8.938012418914474e-05,
      "loss": 1.7257,
      "step": 55648
    },
    {
      "epoch": 0.10987397282980568,
      "grad_norm": 0.0924028754234314,
      "learning_rate": 8.937378396478006e-05,
      "loss": 1.7216,
      "step": 55680
    },
    {
      "epoch": 0.10993711879120212,
      "grad_norm": 0.09533971548080444,
      "learning_rate": 8.936744374041537e-05,
      "loss": 1.7245,
      "step": 55712
    },
    {
      "epoch": 0.11000026475259857,
      "grad_norm": 0.0952165424823761,
      "learning_rate": 8.936110351605069e-05,
      "loss": 1.7227,
      "step": 55744
    },
    {
      "epoch": 0.110063410713995,
      "grad_norm": 0.1044548973441124,
      "learning_rate": 8.935476329168599e-05,
      "loss": 1.7389,
      "step": 55776
    },
    {
      "epoch": 0.11012655667539144,
      "grad_norm": 0.100387804210186,
      "learning_rate": 8.93484230673213e-05,
      "loss": 1.7347,
      "step": 55808
    },
    {
      "epoch": 0.11018970263678789,
      "grad_norm": 0.09685352444648743,
      "learning_rate": 8.934208284295662e-05,
      "loss": 1.7543,
      "step": 55840
    },
    {
      "epoch": 0.11025284859818432,
      "grad_norm": 0.10507971793413162,
      "learning_rate": 8.933574261859192e-05,
      "loss": 1.7258,
      "step": 55872
    },
    {
      "epoch": 0.11031599455958076,
      "grad_norm": 0.09932276606559753,
      "learning_rate": 8.932940239422723e-05,
      "loss": 1.7273,
      "step": 55904
    },
    {
      "epoch": 0.1103791405209772,
      "grad_norm": 0.1030394434928894,
      "learning_rate": 8.932306216986253e-05,
      "loss": 1.737,
      "step": 55936
    },
    {
      "epoch": 0.11044228648237364,
      "grad_norm": 0.11110737919807434,
      "learning_rate": 8.931672194549785e-05,
      "loss": 1.7277,
      "step": 55968
    },
    {
      "epoch": 0.11050543244377008,
      "grad_norm": 0.1007714718580246,
      "learning_rate": 8.931038172113316e-05,
      "loss": 1.7252,
      "step": 56000
    },
    {
      "epoch": 0.11056857840516653,
      "grad_norm": 0.10777892917394638,
      "learning_rate": 8.930404149676848e-05,
      "loss": 1.7268,
      "step": 56032
    },
    {
      "epoch": 0.11063172436656296,
      "grad_norm": 0.09986016899347305,
      "learning_rate": 8.929770127240378e-05,
      "loss": 1.7196,
      "step": 56064
    },
    {
      "epoch": 0.11069487032795941,
      "grad_norm": 0.09925538301467896,
      "learning_rate": 8.929136104803909e-05,
      "loss": 1.7371,
      "step": 56096
    },
    {
      "epoch": 0.11075801628935585,
      "grad_norm": 0.10937350243330002,
      "learning_rate": 8.92850208236744e-05,
      "loss": 1.7441,
      "step": 56128
    },
    {
      "epoch": 0.11082116225075228,
      "grad_norm": 0.0906435176730156,
      "learning_rate": 8.927868059930972e-05,
      "loss": 1.7282,
      "step": 56160
    },
    {
      "epoch": 0.11088430821214873,
      "grad_norm": 0.10186203569173813,
      "learning_rate": 8.927234037494502e-05,
      "loss": 1.7303,
      "step": 56192
    },
    {
      "epoch": 0.11094745417354517,
      "grad_norm": 0.10142674297094345,
      "learning_rate": 8.926600015058034e-05,
      "loss": 1.7121,
      "step": 56224
    },
    {
      "epoch": 0.1110106001349416,
      "grad_norm": 0.09603843837976456,
      "learning_rate": 8.925965992621565e-05,
      "loss": 1.725,
      "step": 56256
    },
    {
      "epoch": 0.11107374609633805,
      "grad_norm": 0.09735945612192154,
      "learning_rate": 8.925331970185095e-05,
      "loss": 1.7226,
      "step": 56288
    },
    {
      "epoch": 0.11113689205773448,
      "grad_norm": 0.101629838347435,
      "learning_rate": 8.924697947748625e-05,
      "loss": 1.7249,
      "step": 56320
    },
    {
      "epoch": 0.11120003801913092,
      "grad_norm": 0.09802871197462082,
      "learning_rate": 8.924063925312157e-05,
      "loss": 1.7285,
      "step": 56352
    },
    {
      "epoch": 0.11126318398052737,
      "grad_norm": 0.09986495226621628,
      "learning_rate": 8.923429902875688e-05,
      "loss": 1.7227,
      "step": 56384
    },
    {
      "epoch": 0.1113263299419238,
      "grad_norm": 0.10159213840961456,
      "learning_rate": 8.92279588043922e-05,
      "loss": 1.722,
      "step": 56416
    },
    {
      "epoch": 0.11138947590332024,
      "grad_norm": 0.09450211375951767,
      "learning_rate": 8.92216185800275e-05,
      "loss": 1.7415,
      "step": 56448
    },
    {
      "epoch": 0.11145262186471669,
      "grad_norm": 0.10010956227779388,
      "learning_rate": 8.921527835566281e-05,
      "loss": 1.7214,
      "step": 56480
    },
    {
      "epoch": 0.11151576782611312,
      "grad_norm": 0.10517623275518417,
      "learning_rate": 8.920893813129813e-05,
      "loss": 1.7443,
      "step": 56512
    },
    {
      "epoch": 0.11157891378750956,
      "grad_norm": 0.10186073929071426,
      "learning_rate": 8.920259790693344e-05,
      "loss": 1.7316,
      "step": 56544
    },
    {
      "epoch": 0.11164205974890601,
      "grad_norm": 0.10859499126672745,
      "learning_rate": 8.919625768256876e-05,
      "loss": 1.7185,
      "step": 56576
    },
    {
      "epoch": 0.11170520571030244,
      "grad_norm": 0.09792271256446838,
      "learning_rate": 8.918991745820406e-05,
      "loss": 1.7227,
      "step": 56608
    },
    {
      "epoch": 0.11176835167169888,
      "grad_norm": 0.10137863457202911,
      "learning_rate": 8.918357723383937e-05,
      "loss": 1.7333,
      "step": 56640
    },
    {
      "epoch": 0.11183149763309533,
      "grad_norm": 0.0956532284617424,
      "learning_rate": 8.917723700947469e-05,
      "loss": 1.7403,
      "step": 56672
    },
    {
      "epoch": 0.11189464359449176,
      "grad_norm": 0.09691862016916275,
      "learning_rate": 8.917089678510999e-05,
      "loss": 1.7317,
      "step": 56704
    },
    {
      "epoch": 0.1119577895558882,
      "grad_norm": 0.1034805104136467,
      "learning_rate": 8.916455656074529e-05,
      "loss": 1.7369,
      "step": 56736
    },
    {
      "epoch": 0.11202093551728465,
      "grad_norm": 0.0991164967417717,
      "learning_rate": 8.91582163363806e-05,
      "loss": 1.7297,
      "step": 56768
    },
    {
      "epoch": 0.11208408147868108,
      "grad_norm": 0.10262954980134964,
      "learning_rate": 8.915187611201592e-05,
      "loss": 1.7233,
      "step": 56800
    },
    {
      "epoch": 0.11214722744007752,
      "grad_norm": 0.10207094252109528,
      "learning_rate": 8.914553588765123e-05,
      "loss": 1.7227,
      "step": 56832
    },
    {
      "epoch": 0.11221037340147397,
      "grad_norm": 0.10577253252267838,
      "learning_rate": 8.913919566328653e-05,
      "loss": 1.7331,
      "step": 56864
    },
    {
      "epoch": 0.1122735193628704,
      "grad_norm": 0.11363503336906433,
      "learning_rate": 8.913285543892185e-05,
      "loss": 1.7172,
      "step": 56896
    },
    {
      "epoch": 0.11233666532426685,
      "grad_norm": 0.09643087536096573,
      "learning_rate": 8.912651521455716e-05,
      "loss": 1.7321,
      "step": 56928
    },
    {
      "epoch": 0.11239981128566329,
      "grad_norm": 0.11154823005199432,
      "learning_rate": 8.912017499019248e-05,
      "loss": 1.726,
      "step": 56960
    },
    {
      "epoch": 0.11246295724705972,
      "grad_norm": 0.10231855511665344,
      "learning_rate": 8.911383476582778e-05,
      "loss": 1.7308,
      "step": 56992
    },
    {
      "epoch": 0.11252610320845617,
      "grad_norm": 0.0947440043091774,
      "learning_rate": 8.910749454146309e-05,
      "loss": 1.7315,
      "step": 57024
    },
    {
      "epoch": 0.1125892491698526,
      "grad_norm": 0.10054609179496765,
      "learning_rate": 8.91011543170984e-05,
      "loss": 1.727,
      "step": 57056
    },
    {
      "epoch": 0.11265239513124904,
      "grad_norm": 0.09855375438928604,
      "learning_rate": 8.909481409273372e-05,
      "loss": 1.7351,
      "step": 57088
    },
    {
      "epoch": 0.11271554109264549,
      "grad_norm": 0.09844506531953812,
      "learning_rate": 8.908847386836902e-05,
      "loss": 1.7132,
      "step": 57120
    },
    {
      "epoch": 0.11277868705404193,
      "grad_norm": 0.10131185501813889,
      "learning_rate": 8.908213364400432e-05,
      "loss": 1.7026,
      "step": 57152
    },
    {
      "epoch": 0.11284183301543836,
      "grad_norm": 0.11000480502843857,
      "learning_rate": 8.907579341963964e-05,
      "loss": 1.719,
      "step": 57184
    },
    {
      "epoch": 0.11290497897683481,
      "grad_norm": 0.10741524398326874,
      "learning_rate": 8.906945319527495e-05,
      "loss": 1.7258,
      "step": 57216
    },
    {
      "epoch": 0.11296812493823125,
      "grad_norm": 0.10290786623954773,
      "learning_rate": 8.906311297091027e-05,
      "loss": 1.7338,
      "step": 57248
    },
    {
      "epoch": 0.11303127089962768,
      "grad_norm": 0.09821883589029312,
      "learning_rate": 8.905677274654557e-05,
      "loss": 1.7201,
      "step": 57280
    },
    {
      "epoch": 0.11309441686102413,
      "grad_norm": 0.09282093495130539,
      "learning_rate": 8.905043252218088e-05,
      "loss": 1.7238,
      "step": 57312
    },
    {
      "epoch": 0.11315756282242057,
      "grad_norm": 0.1000959724187851,
      "learning_rate": 8.90440922978162e-05,
      "loss": 1.7278,
      "step": 57344
    },
    {
      "epoch": 0.113220708783817,
      "grad_norm": 0.10023311525583267,
      "learning_rate": 8.903775207345151e-05,
      "loss": 1.7389,
      "step": 57376
    },
    {
      "epoch": 0.11328385474521345,
      "grad_norm": 0.11677708476781845,
      "learning_rate": 8.903141184908681e-05,
      "loss": 1.726,
      "step": 57408
    },
    {
      "epoch": 0.11334700070660989,
      "grad_norm": 0.10847847908735275,
      "learning_rate": 8.902507162472213e-05,
      "loss": 1.7173,
      "step": 57440
    },
    {
      "epoch": 0.11341014666800632,
      "grad_norm": 0.09579665958881378,
      "learning_rate": 8.901873140035744e-05,
      "loss": 1.714,
      "step": 57472
    },
    {
      "epoch": 0.11347329262940277,
      "grad_norm": 0.09655775129795074,
      "learning_rate": 8.901239117599274e-05,
      "loss": 1.7304,
      "step": 57504
    },
    {
      "epoch": 0.1135364385907992,
      "grad_norm": 0.1029743105173111,
      "learning_rate": 8.900605095162806e-05,
      "loss": 1.7349,
      "step": 57536
    },
    {
      "epoch": 0.11359958455219564,
      "grad_norm": 0.09406114369630814,
      "learning_rate": 8.899971072726336e-05,
      "loss": 1.7313,
      "step": 57568
    },
    {
      "epoch": 0.11366273051359209,
      "grad_norm": 0.09342000633478165,
      "learning_rate": 8.899337050289867e-05,
      "loss": 1.725,
      "step": 57600
    },
    {
      "epoch": 0.11372587647498852,
      "grad_norm": 0.09975459426641464,
      "learning_rate": 8.898703027853399e-05,
      "loss": 1.7249,
      "step": 57632
    },
    {
      "epoch": 0.11378902243638496,
      "grad_norm": 0.10720018297433853,
      "learning_rate": 8.898069005416929e-05,
      "loss": 1.731,
      "step": 57664
    },
    {
      "epoch": 0.11385216839778141,
      "grad_norm": 0.09851442277431488,
      "learning_rate": 8.89743498298046e-05,
      "loss": 1.7413,
      "step": 57696
    },
    {
      "epoch": 0.11391531435917784,
      "grad_norm": 0.09014596790075302,
      "learning_rate": 8.896800960543992e-05,
      "loss": 1.7276,
      "step": 57728
    },
    {
      "epoch": 0.11397846032057429,
      "grad_norm": 0.09720245003700256,
      "learning_rate": 8.896166938107523e-05,
      "loss": 1.7305,
      "step": 57760
    },
    {
      "epoch": 0.11404160628197073,
      "grad_norm": 0.09888971596956253,
      "learning_rate": 8.895532915671055e-05,
      "loss": 1.7381,
      "step": 57792
    },
    {
      "epoch": 0.11410475224336716,
      "grad_norm": 0.10641719400882721,
      "learning_rate": 8.894898893234585e-05,
      "loss": 1.7299,
      "step": 57824
    },
    {
      "epoch": 0.11416789820476361,
      "grad_norm": 0.10379938781261444,
      "learning_rate": 8.894264870798116e-05,
      "loss": 1.7114,
      "step": 57856
    },
    {
      "epoch": 0.11423104416616005,
      "grad_norm": 0.10532380640506744,
      "learning_rate": 8.893630848361648e-05,
      "loss": 1.7444,
      "step": 57888
    },
    {
      "epoch": 0.11429419012755648,
      "grad_norm": 0.11497674137353897,
      "learning_rate": 8.892996825925178e-05,
      "loss": 1.721,
      "step": 57920
    },
    {
      "epoch": 0.11435733608895293,
      "grad_norm": 0.09847824275493622,
      "learning_rate": 8.892362803488709e-05,
      "loss": 1.7108,
      "step": 57952
    },
    {
      "epoch": 0.11442048205034937,
      "grad_norm": 0.1033286601305008,
      "learning_rate": 8.891728781052239e-05,
      "loss": 1.7205,
      "step": 57984
    },
    {
      "epoch": 0.1144836280117458,
      "grad_norm": 0.10216425359249115,
      "learning_rate": 8.891094758615771e-05,
      "loss": 1.7115,
      "step": 58016
    },
    {
      "epoch": 0.11454677397314225,
      "grad_norm": 0.11663391441106796,
      "learning_rate": 8.890460736179302e-05,
      "loss": 1.7167,
      "step": 58048
    },
    {
      "epoch": 0.11460991993453869,
      "grad_norm": 0.10209348797798157,
      "learning_rate": 8.889826713742832e-05,
      "loss": 1.7327,
      "step": 58080
    },
    {
      "epoch": 0.11467306589593512,
      "grad_norm": 0.10272000730037689,
      "learning_rate": 8.889192691306364e-05,
      "loss": 1.716,
      "step": 58112
    },
    {
      "epoch": 0.11473621185733157,
      "grad_norm": 0.11681805551052094,
      "learning_rate": 8.888558668869895e-05,
      "loss": 1.7172,
      "step": 58144
    },
    {
      "epoch": 0.114799357818728,
      "grad_norm": 0.10092071443796158,
      "learning_rate": 8.887924646433427e-05,
      "loss": 1.7253,
      "step": 58176
    },
    {
      "epoch": 0.11486250378012444,
      "grad_norm": 0.09716968238353729,
      "learning_rate": 8.887290623996957e-05,
      "loss": 1.7247,
      "step": 58208
    },
    {
      "epoch": 0.11492564974152089,
      "grad_norm": 0.09291841089725494,
      "learning_rate": 8.886656601560488e-05,
      "loss": 1.7265,
      "step": 58240
    },
    {
      "epoch": 0.11498879570291733,
      "grad_norm": 0.10266035795211792,
      "learning_rate": 8.88602257912402e-05,
      "loss": 1.7138,
      "step": 58272
    },
    {
      "epoch": 0.11505194166431376,
      "grad_norm": 0.10050240904092789,
      "learning_rate": 8.885388556687551e-05,
      "loss": 1.7305,
      "step": 58304
    },
    {
      "epoch": 0.11511508762571021,
      "grad_norm": 0.10351265966892242,
      "learning_rate": 8.884754534251081e-05,
      "loss": 1.7236,
      "step": 58336
    },
    {
      "epoch": 0.11517823358710665,
      "grad_norm": 0.10937555134296417,
      "learning_rate": 8.884120511814613e-05,
      "loss": 1.731,
      "step": 58368
    },
    {
      "epoch": 0.11524137954850308,
      "grad_norm": 0.09581410884857178,
      "learning_rate": 8.883486489378143e-05,
      "loss": 1.7404,
      "step": 58400
    },
    {
      "epoch": 0.11530452550989953,
      "grad_norm": 0.09766798466444016,
      "learning_rate": 8.882852466941674e-05,
      "loss": 1.7239,
      "step": 58432
    },
    {
      "epoch": 0.11536767147129597,
      "grad_norm": 0.10437649488449097,
      "learning_rate": 8.882218444505206e-05,
      "loss": 1.7229,
      "step": 58464
    },
    {
      "epoch": 0.1154308174326924,
      "grad_norm": 0.10612741857767105,
      "learning_rate": 8.881584422068736e-05,
      "loss": 1.7104,
      "step": 58496
    },
    {
      "epoch": 0.11549396339408885,
      "grad_norm": 0.09655754268169403,
      "learning_rate": 8.880950399632267e-05,
      "loss": 1.7354,
      "step": 58528
    },
    {
      "epoch": 0.11555710935548529,
      "grad_norm": 0.09518884867429733,
      "learning_rate": 8.880316377195799e-05,
      "loss": 1.7071,
      "step": 58560
    },
    {
      "epoch": 0.11562025531688173,
      "grad_norm": 0.11409979313611984,
      "learning_rate": 8.87968235475933e-05,
      "loss": 1.7165,
      "step": 58592
    },
    {
      "epoch": 0.11568340127827817,
      "grad_norm": 0.10633133351802826,
      "learning_rate": 8.87904833232286e-05,
      "loss": 1.7251,
      "step": 58624
    },
    {
      "epoch": 0.1157465472396746,
      "grad_norm": 0.10412392020225525,
      "learning_rate": 8.878414309886392e-05,
      "loss": 1.7237,
      "step": 58656
    },
    {
      "epoch": 0.11580969320107105,
      "grad_norm": 0.10202760994434357,
      "learning_rate": 8.877780287449923e-05,
      "loss": 1.7154,
      "step": 58688
    },
    {
      "epoch": 0.11587283916246749,
      "grad_norm": 0.09684973955154419,
      "learning_rate": 8.877146265013455e-05,
      "loss": 1.7328,
      "step": 58720
    },
    {
      "epoch": 0.11593598512386392,
      "grad_norm": 0.09888684004545212,
      "learning_rate": 8.876512242576985e-05,
      "loss": 1.7259,
      "step": 58752
    },
    {
      "epoch": 0.11599913108526037,
      "grad_norm": 0.09473881870508194,
      "learning_rate": 8.875878220140515e-05,
      "loss": 1.7188,
      "step": 58784
    },
    {
      "epoch": 0.11606227704665681,
      "grad_norm": 0.10119607299566269,
      "learning_rate": 8.875244197704046e-05,
      "loss": 1.7373,
      "step": 58816
    },
    {
      "epoch": 0.11612542300805324,
      "grad_norm": 0.11298197507858276,
      "learning_rate": 8.874610175267578e-05,
      "loss": 1.7256,
      "step": 58848
    },
    {
      "epoch": 0.1161885689694497,
      "grad_norm": 0.10415863990783691,
      "learning_rate": 8.873976152831108e-05,
      "loss": 1.7198,
      "step": 58880
    },
    {
      "epoch": 0.11625171493084613,
      "grad_norm": 0.09710899740457535,
      "learning_rate": 8.873342130394639e-05,
      "loss": 1.7213,
      "step": 58912
    },
    {
      "epoch": 0.11631486089224256,
      "grad_norm": 0.09486311674118042,
      "learning_rate": 8.872708107958171e-05,
      "loss": 1.7239,
      "step": 58944
    },
    {
      "epoch": 0.11637800685363901,
      "grad_norm": 0.10320895165205002,
      "learning_rate": 8.872074085521702e-05,
      "loss": 1.7173,
      "step": 58976
    },
    {
      "epoch": 0.11644115281503545,
      "grad_norm": 0.10717189311981201,
      "learning_rate": 8.871440063085232e-05,
      "loss": 1.7186,
      "step": 59008
    },
    {
      "epoch": 0.11650429877643188,
      "grad_norm": 0.09883598238229752,
      "learning_rate": 8.870806040648764e-05,
      "loss": 1.7141,
      "step": 59040
    },
    {
      "epoch": 0.11656744473782833,
      "grad_norm": 0.10612888634204865,
      "learning_rate": 8.870172018212295e-05,
      "loss": 1.7167,
      "step": 59072
    },
    {
      "epoch": 0.11663059069922477,
      "grad_norm": 0.10070229321718216,
      "learning_rate": 8.869537995775827e-05,
      "loss": 1.7282,
      "step": 59104
    },
    {
      "epoch": 0.1166937366606212,
      "grad_norm": 0.09305015951395035,
      "learning_rate": 8.868903973339358e-05,
      "loss": 1.723,
      "step": 59136
    },
    {
      "epoch": 0.11675688262201765,
      "grad_norm": 0.10157781839370728,
      "learning_rate": 8.868269950902888e-05,
      "loss": 1.7131,
      "step": 59168
    },
    {
      "epoch": 0.11682002858341409,
      "grad_norm": 0.10942184925079346,
      "learning_rate": 8.867635928466418e-05,
      "loss": 1.7332,
      "step": 59200
    },
    {
      "epoch": 0.11688317454481052,
      "grad_norm": 0.09981514513492584,
      "learning_rate": 8.86700190602995e-05,
      "loss": 1.7079,
      "step": 59232
    },
    {
      "epoch": 0.11694632050620697,
      "grad_norm": 0.09638652205467224,
      "learning_rate": 8.866367883593481e-05,
      "loss": 1.7142,
      "step": 59264
    },
    {
      "epoch": 0.1170094664676034,
      "grad_norm": 0.09642908722162247,
      "learning_rate": 8.865733861157011e-05,
      "loss": 1.7273,
      "step": 59296
    },
    {
      "epoch": 0.11707261242899984,
      "grad_norm": 0.09426500648260117,
      "learning_rate": 8.865099838720543e-05,
      "loss": 1.7278,
      "step": 59328
    },
    {
      "epoch": 0.11713575839039629,
      "grad_norm": 0.09257742762565613,
      "learning_rate": 8.864465816284074e-05,
      "loss": 1.7116,
      "step": 59360
    },
    {
      "epoch": 0.11719890435179273,
      "grad_norm": 0.09522250294685364,
      "learning_rate": 8.863831793847606e-05,
      "loss": 1.7336,
      "step": 59392
    },
    {
      "epoch": 0.11726205031318918,
      "grad_norm": 0.10382465273141861,
      "learning_rate": 8.863197771411136e-05,
      "loss": 1.7237,
      "step": 59424
    },
    {
      "epoch": 0.11732519627458561,
      "grad_norm": 0.09743598848581314,
      "learning_rate": 8.862563748974667e-05,
      "loss": 1.7137,
      "step": 59456
    },
    {
      "epoch": 0.11738834223598205,
      "grad_norm": 0.0995405837893486,
      "learning_rate": 8.861929726538199e-05,
      "loss": 1.7205,
      "step": 59488
    },
    {
      "epoch": 0.1174514881973785,
      "grad_norm": 0.10108558088541031,
      "learning_rate": 8.86129570410173e-05,
      "loss": 1.7195,
      "step": 59520
    },
    {
      "epoch": 0.11751463415877493,
      "grad_norm": 0.10459664463996887,
      "learning_rate": 8.86066168166526e-05,
      "loss": 1.7146,
      "step": 59552
    },
    {
      "epoch": 0.11757778012017137,
      "grad_norm": 0.09926500171422958,
      "learning_rate": 8.860027659228792e-05,
      "loss": 1.7228,
      "step": 59584
    },
    {
      "epoch": 0.11764092608156781,
      "grad_norm": 0.1011614203453064,
      "learning_rate": 8.859393636792322e-05,
      "loss": 1.7235,
      "step": 59616
    },
    {
      "epoch": 0.11770407204296425,
      "grad_norm": 0.09708654880523682,
      "learning_rate": 8.858759614355853e-05,
      "loss": 1.7182,
      "step": 59648
    },
    {
      "epoch": 0.11776721800436069,
      "grad_norm": 0.1018308699131012,
      "learning_rate": 8.858125591919383e-05,
      "loss": 1.7181,
      "step": 59680
    },
    {
      "epoch": 0.11783036396575713,
      "grad_norm": 0.09900863468647003,
      "learning_rate": 8.857491569482915e-05,
      "loss": 1.7293,
      "step": 59712
    },
    {
      "epoch": 0.11789350992715357,
      "grad_norm": 0.09655536711215973,
      "learning_rate": 8.856857547046446e-05,
      "loss": 1.7174,
      "step": 59744
    },
    {
      "epoch": 0.11795665588855,
      "grad_norm": 0.09684063494205475,
      "learning_rate": 8.856223524609978e-05,
      "loss": 1.7208,
      "step": 59776
    },
    {
      "epoch": 0.11801980184994645,
      "grad_norm": 0.10317421704530716,
      "learning_rate": 8.855589502173509e-05,
      "loss": 1.7114,
      "step": 59808
    },
    {
      "epoch": 0.11808294781134289,
      "grad_norm": 0.1004984974861145,
      "learning_rate": 8.85495547973704e-05,
      "loss": 1.7205,
      "step": 59840
    },
    {
      "epoch": 0.11814609377273932,
      "grad_norm": 0.09612350910902023,
      "learning_rate": 8.854321457300571e-05,
      "loss": 1.7275,
      "step": 59872
    },
    {
      "epoch": 0.11820923973413577,
      "grad_norm": 0.09707285463809967,
      "learning_rate": 8.853687434864102e-05,
      "loss": 1.7085,
      "step": 59904
    },
    {
      "epoch": 0.11827238569553221,
      "grad_norm": 0.10384999960660934,
      "learning_rate": 8.853053412427634e-05,
      "loss": 1.7203,
      "step": 59936
    },
    {
      "epoch": 0.11833553165692864,
      "grad_norm": 0.10721999406814575,
      "learning_rate": 8.852419389991164e-05,
      "loss": 1.7226,
      "step": 59968
    },
    {
      "epoch": 0.1183986776183251,
      "grad_norm": 0.09263718873262405,
      "learning_rate": 8.851785367554695e-05,
      "loss": 1.7085,
      "step": 60000
    },
    {
      "epoch": 0.11846182357972153,
      "grad_norm": 0.10204324871301651,
      "learning_rate": 8.851151345118225e-05,
      "loss": 1.73,
      "step": 60032
    },
    {
      "epoch": 0.11852496954111796,
      "grad_norm": 0.0958547443151474,
      "learning_rate": 8.850517322681757e-05,
      "loss": 1.7162,
      "step": 60064
    },
    {
      "epoch": 0.11858811550251441,
      "grad_norm": 0.09493812173604965,
      "learning_rate": 8.849883300245287e-05,
      "loss": 1.7146,
      "step": 60096
    },
    {
      "epoch": 0.11865126146391085,
      "grad_norm": 0.10567105561494827,
      "learning_rate": 8.849249277808818e-05,
      "loss": 1.722,
      "step": 60128
    },
    {
      "epoch": 0.11871440742530728,
      "grad_norm": 0.10352890193462372,
      "learning_rate": 8.84861525537235e-05,
      "loss": 1.6995,
      "step": 60160
    },
    {
      "epoch": 0.11877755338670373,
      "grad_norm": 0.10077771544456482,
      "learning_rate": 8.847981232935881e-05,
      "loss": 1.7096,
      "step": 60192
    },
    {
      "epoch": 0.11884069934810017,
      "grad_norm": 0.09664720296859741,
      "learning_rate": 8.847347210499411e-05,
      "loss": 1.7085,
      "step": 60224
    },
    {
      "epoch": 0.1189038453094966,
      "grad_norm": 0.09669395536184311,
      "learning_rate": 8.846713188062943e-05,
      "loss": 1.723,
      "step": 60256
    },
    {
      "epoch": 0.11896699127089305,
      "grad_norm": 0.10533349215984344,
      "learning_rate": 8.846079165626474e-05,
      "loss": 1.7222,
      "step": 60288
    },
    {
      "epoch": 0.11903013723228949,
      "grad_norm": 0.09265675395727158,
      "learning_rate": 8.845445143190006e-05,
      "loss": 1.7205,
      "step": 60320
    },
    {
      "epoch": 0.11909328319368594,
      "grad_norm": 0.09601378440856934,
      "learning_rate": 8.844811120753536e-05,
      "loss": 1.7231,
      "step": 60352
    },
    {
      "epoch": 0.11915642915508237,
      "grad_norm": 0.10059671103954315,
      "learning_rate": 8.844177098317067e-05,
      "loss": 1.7213,
      "step": 60384
    },
    {
      "epoch": 0.11921957511647881,
      "grad_norm": 0.10184638202190399,
      "learning_rate": 8.843543075880599e-05,
      "loss": 1.7136,
      "step": 60416
    },
    {
      "epoch": 0.11928272107787526,
      "grad_norm": 0.09648539870977402,
      "learning_rate": 8.842909053444129e-05,
      "loss": 1.7194,
      "step": 60448
    },
    {
      "epoch": 0.11934586703927169,
      "grad_norm": 0.10598673671483994,
      "learning_rate": 8.84227503100766e-05,
      "loss": 1.7254,
      "step": 60480
    },
    {
      "epoch": 0.11940901300066813,
      "grad_norm": 0.10552442073822021,
      "learning_rate": 8.84164100857119e-05,
      "loss": 1.7104,
      "step": 60512
    },
    {
      "epoch": 0.11947215896206458,
      "grad_norm": 0.0955764427781105,
      "learning_rate": 8.841006986134722e-05,
      "loss": 1.6987,
      "step": 60544
    },
    {
      "epoch": 0.11953530492346101,
      "grad_norm": 0.11034785956144333,
      "learning_rate": 8.840372963698253e-05,
      "loss": 1.7107,
      "step": 60576
    },
    {
      "epoch": 0.11959845088485745,
      "grad_norm": 0.09071777760982513,
      "learning_rate": 8.839738941261785e-05,
      "loss": 1.7111,
      "step": 60608
    },
    {
      "epoch": 0.1196615968462539,
      "grad_norm": 0.11077333241701126,
      "learning_rate": 8.839104918825315e-05,
      "loss": 1.7147,
      "step": 60640
    },
    {
      "epoch": 0.11972474280765033,
      "grad_norm": 0.09640567749738693,
      "learning_rate": 8.838470896388846e-05,
      "loss": 1.7043,
      "step": 60672
    },
    {
      "epoch": 0.11978788876904677,
      "grad_norm": 0.09565507620573044,
      "learning_rate": 8.837836873952378e-05,
      "loss": 1.718,
      "step": 60704
    },
    {
      "epoch": 0.11985103473044321,
      "grad_norm": 0.11187933385372162,
      "learning_rate": 8.837202851515909e-05,
      "loss": 1.7183,
      "step": 60736
    },
    {
      "epoch": 0.11991418069183965,
      "grad_norm": 0.09896146506071091,
      "learning_rate": 8.83656882907944e-05,
      "loss": 1.7038,
      "step": 60768
    },
    {
      "epoch": 0.11997732665323609,
      "grad_norm": 0.09657255560159683,
      "learning_rate": 8.835934806642971e-05,
      "loss": 1.7216,
      "step": 60800
    },
    {
      "epoch": 0.12004047261463253,
      "grad_norm": 0.10551982372999191,
      "learning_rate": 8.835300784206502e-05,
      "loss": 1.7189,
      "step": 60832
    },
    {
      "epoch": 0.12010361857602897,
      "grad_norm": 0.10490927845239639,
      "learning_rate": 8.834666761770032e-05,
      "loss": 1.7255,
      "step": 60864
    },
    {
      "epoch": 0.1201667645374254,
      "grad_norm": 0.1005062609910965,
      "learning_rate": 8.834032739333562e-05,
      "loss": 1.7058,
      "step": 60896
    },
    {
      "epoch": 0.12022991049882185,
      "grad_norm": 0.09513265639543533,
      "learning_rate": 8.833398716897094e-05,
      "loss": 1.7279,
      "step": 60928
    },
    {
      "epoch": 0.12029305646021829,
      "grad_norm": 0.10015387833118439,
      "learning_rate": 8.832764694460625e-05,
      "loss": 1.7229,
      "step": 60960
    },
    {
      "epoch": 0.12035620242161472,
      "grad_norm": 0.09876395016908646,
      "learning_rate": 8.832130672024157e-05,
      "loss": 1.7146,
      "step": 60992
    },
    {
      "epoch": 0.12041934838301117,
      "grad_norm": 0.10086855292320251,
      "learning_rate": 8.831496649587687e-05,
      "loss": 1.7053,
      "step": 61024
    },
    {
      "epoch": 0.12048249434440761,
      "grad_norm": 0.10953648388385773,
      "learning_rate": 8.830862627151218e-05,
      "loss": 1.7104,
      "step": 61056
    },
    {
      "epoch": 0.12054564030580404,
      "grad_norm": 0.10189200192689896,
      "learning_rate": 8.83022860471475e-05,
      "loss": 1.7271,
      "step": 61088
    },
    {
      "epoch": 0.1206087862672005,
      "grad_norm": 0.09535510838031769,
      "learning_rate": 8.829594582278281e-05,
      "loss": 1.6988,
      "step": 61120
    },
    {
      "epoch": 0.12067193222859693,
      "grad_norm": 0.10282225161790848,
      "learning_rate": 8.828960559841813e-05,
      "loss": 1.7099,
      "step": 61152
    },
    {
      "epoch": 0.12073507818999338,
      "grad_norm": 0.09956984221935272,
      "learning_rate": 8.828326537405343e-05,
      "loss": 1.7108,
      "step": 61184
    },
    {
      "epoch": 0.12079822415138981,
      "grad_norm": 0.11226121336221695,
      "learning_rate": 8.827692514968874e-05,
      "loss": 1.7123,
      "step": 61216
    },
    {
      "epoch": 0.12086137011278625,
      "grad_norm": 0.0932110920548439,
      "learning_rate": 8.827058492532406e-05,
      "loss": 1.7199,
      "step": 61248
    },
    {
      "epoch": 0.1209245160741827,
      "grad_norm": 0.10740521550178528,
      "learning_rate": 8.826424470095936e-05,
      "loss": 1.7231,
      "step": 61280
    },
    {
      "epoch": 0.12098766203557913,
      "grad_norm": 0.09979068487882614,
      "learning_rate": 8.825790447659466e-05,
      "loss": 1.7151,
      "step": 61312
    },
    {
      "epoch": 0.12105080799697557,
      "grad_norm": 0.10132623463869095,
      "learning_rate": 8.825156425222997e-05,
      "loss": 1.7129,
      "step": 61344
    },
    {
      "epoch": 0.12111395395837202,
      "grad_norm": 0.12113574147224426,
      "learning_rate": 8.824522402786529e-05,
      "loss": 1.7142,
      "step": 61376
    },
    {
      "epoch": 0.12117709991976845,
      "grad_norm": 0.09612365812063217,
      "learning_rate": 8.82388838035006e-05,
      "loss": 1.704,
      "step": 61408
    },
    {
      "epoch": 0.12124024588116489,
      "grad_norm": 0.0958484411239624,
      "learning_rate": 8.82325435791359e-05,
      "loss": 1.7213,
      "step": 61440
    },
    {
      "epoch": 0.12130339184256134,
      "grad_norm": 0.0972626656293869,
      "learning_rate": 8.822620335477122e-05,
      "loss": 1.7246,
      "step": 61472
    },
    {
      "epoch": 0.12136653780395777,
      "grad_norm": 0.10045930743217468,
      "learning_rate": 8.821986313040653e-05,
      "loss": 1.7039,
      "step": 61504
    },
    {
      "epoch": 0.12142968376535421,
      "grad_norm": 0.10228867828845978,
      "learning_rate": 8.821352290604185e-05,
      "loss": 1.7062,
      "step": 61536
    },
    {
      "epoch": 0.12149282972675066,
      "grad_norm": 0.09936395287513733,
      "learning_rate": 8.820718268167715e-05,
      "loss": 1.7161,
      "step": 61568
    },
    {
      "epoch": 0.12155597568814709,
      "grad_norm": 0.10981740802526474,
      "learning_rate": 8.820084245731246e-05,
      "loss": 1.7156,
      "step": 61600
    },
    {
      "epoch": 0.12161912164954353,
      "grad_norm": 0.09943246096372604,
      "learning_rate": 8.819450223294778e-05,
      "loss": 1.7159,
      "step": 61632
    },
    {
      "epoch": 0.12168226761093998,
      "grad_norm": 0.10377268493175507,
      "learning_rate": 8.818816200858308e-05,
      "loss": 1.7165,
      "step": 61664
    },
    {
      "epoch": 0.12174541357233641,
      "grad_norm": 0.09910544008016586,
      "learning_rate": 8.81818217842184e-05,
      "loss": 1.7191,
      "step": 61696
    },
    {
      "epoch": 0.12180855953373285,
      "grad_norm": 0.10430502146482468,
      "learning_rate": 8.81754815598537e-05,
      "loss": 1.7162,
      "step": 61728
    },
    {
      "epoch": 0.1218717054951293,
      "grad_norm": 0.10514651238918304,
      "learning_rate": 8.816914133548901e-05,
      "loss": 1.7163,
      "step": 61760
    },
    {
      "epoch": 0.12193485145652573,
      "grad_norm": 0.09861918538808823,
      "learning_rate": 8.816280111112432e-05,
      "loss": 1.7119,
      "step": 61792
    },
    {
      "epoch": 0.12199799741792217,
      "grad_norm": 0.09835562109947205,
      "learning_rate": 8.815646088675964e-05,
      "loss": 1.6996,
      "step": 61824
    },
    {
      "epoch": 0.12206114337931862,
      "grad_norm": 0.11428391188383102,
      "learning_rate": 8.815012066239494e-05,
      "loss": 1.7223,
      "step": 61856
    },
    {
      "epoch": 0.12212428934071505,
      "grad_norm": 0.09933076798915863,
      "learning_rate": 8.814378043803025e-05,
      "loss": 1.7192,
      "step": 61888
    },
    {
      "epoch": 0.12218743530211149,
      "grad_norm": 0.1008901372551918,
      "learning_rate": 8.813744021366557e-05,
      "loss": 1.7213,
      "step": 61920
    },
    {
      "epoch": 0.12225058126350793,
      "grad_norm": 0.10245493054389954,
      "learning_rate": 8.813109998930088e-05,
      "loss": 1.7096,
      "step": 61952
    },
    {
      "epoch": 0.12231372722490437,
      "grad_norm": 0.09385039657354355,
      "learning_rate": 8.812475976493618e-05,
      "loss": 1.7096,
      "step": 61984
    },
    {
      "epoch": 0.12237687318630082,
      "grad_norm": 0.09650552272796631,
      "learning_rate": 8.81184195405715e-05,
      "loss": 1.7089,
      "step": 62016
    },
    {
      "epoch": 0.12244001914769725,
      "grad_norm": 0.1181628555059433,
      "learning_rate": 8.811207931620681e-05,
      "loss": 1.7232,
      "step": 62048
    },
    {
      "epoch": 0.12250316510909369,
      "grad_norm": 0.11358977854251862,
      "learning_rate": 8.810573909184211e-05,
      "loss": 1.7075,
      "step": 62080
    },
    {
      "epoch": 0.12256631107049014,
      "grad_norm": 0.0945497378706932,
      "learning_rate": 8.809939886747743e-05,
      "loss": 1.7116,
      "step": 62112
    },
    {
      "epoch": 0.12262945703188657,
      "grad_norm": 0.09654837846755981,
      "learning_rate": 8.809305864311273e-05,
      "loss": 1.7164,
      "step": 62144
    },
    {
      "epoch": 0.12269260299328301,
      "grad_norm": 0.1133982315659523,
      "learning_rate": 8.808671841874804e-05,
      "loss": 1.7247,
      "step": 62176
    },
    {
      "epoch": 0.12275574895467946,
      "grad_norm": 0.1014733836054802,
      "learning_rate": 8.808037819438336e-05,
      "loss": 1.723,
      "step": 62208
    },
    {
      "epoch": 0.1228188949160759,
      "grad_norm": 0.09367592632770538,
      "learning_rate": 8.807403797001866e-05,
      "loss": 1.7207,
      "step": 62240
    },
    {
      "epoch": 0.12288204087747233,
      "grad_norm": 0.10122451186180115,
      "learning_rate": 8.806769774565397e-05,
      "loss": 1.712,
      "step": 62272
    },
    {
      "epoch": 0.12294518683886878,
      "grad_norm": 0.10430467128753662,
      "learning_rate": 8.806135752128929e-05,
      "loss": 1.7166,
      "step": 62304
    },
    {
      "epoch": 0.12300833280026521,
      "grad_norm": 0.10268769413232803,
      "learning_rate": 8.80550172969246e-05,
      "loss": 1.6995,
      "step": 62336
    },
    {
      "epoch": 0.12307147876166165,
      "grad_norm": 0.10345062613487244,
      "learning_rate": 8.80486770725599e-05,
      "loss": 1.7149,
      "step": 62368
    },
    {
      "epoch": 0.1231346247230581,
      "grad_norm": 0.10112529247999191,
      "learning_rate": 8.804233684819522e-05,
      "loss": 1.6989,
      "step": 62400
    },
    {
      "epoch": 0.12319777068445453,
      "grad_norm": 0.10453299432992935,
      "learning_rate": 8.803599662383053e-05,
      "loss": 1.7199,
      "step": 62432
    },
    {
      "epoch": 0.12326091664585097,
      "grad_norm": 0.09306968748569489,
      "learning_rate": 8.802965639946585e-05,
      "loss": 1.7138,
      "step": 62464
    },
    {
      "epoch": 0.12332406260724742,
      "grad_norm": 0.09521358460187912,
      "learning_rate": 8.802331617510115e-05,
      "loss": 1.7107,
      "step": 62496
    },
    {
      "epoch": 0.12338720856864385,
      "grad_norm": 0.10357385128736496,
      "learning_rate": 8.801697595073646e-05,
      "loss": 1.7197,
      "step": 62528
    },
    {
      "epoch": 0.12345035453004029,
      "grad_norm": 0.0984930694103241,
      "learning_rate": 8.801063572637177e-05,
      "loss": 1.717,
      "step": 62560
    },
    {
      "epoch": 0.12351350049143674,
      "grad_norm": 0.09655478596687317,
      "learning_rate": 8.800429550200708e-05,
      "loss": 1.7079,
      "step": 62592
    },
    {
      "epoch": 0.12357664645283317,
      "grad_norm": 0.10092873871326447,
      "learning_rate": 8.79979552776424e-05,
      "loss": 1.696,
      "step": 62624
    },
    {
      "epoch": 0.12363979241422961,
      "grad_norm": 0.0903581902384758,
      "learning_rate": 8.79916150532777e-05,
      "loss": 1.7151,
      "step": 62656
    },
    {
      "epoch": 0.12370293837562606,
      "grad_norm": 0.09429716318845749,
      "learning_rate": 8.798527482891301e-05,
      "loss": 1.706,
      "step": 62688
    },
    {
      "epoch": 0.12376608433702249,
      "grad_norm": 0.10034441202878952,
      "learning_rate": 8.797893460454832e-05,
      "loss": 1.7179,
      "step": 62720
    },
    {
      "epoch": 0.12382923029841893,
      "grad_norm": 0.10464702546596527,
      "learning_rate": 8.797259438018364e-05,
      "loss": 1.7106,
      "step": 62752
    },
    {
      "epoch": 0.12389237625981538,
      "grad_norm": 0.09326034784317017,
      "learning_rate": 8.796625415581894e-05,
      "loss": 1.7161,
      "step": 62784
    },
    {
      "epoch": 0.12395552222121181,
      "grad_norm": 0.09914770722389221,
      "learning_rate": 8.795991393145425e-05,
      "loss": 1.7073,
      "step": 62816
    },
    {
      "epoch": 0.12401866818260826,
      "grad_norm": 0.11192018538713455,
      "learning_rate": 8.795357370708957e-05,
      "loss": 1.7125,
      "step": 62848
    },
    {
      "epoch": 0.1240818141440047,
      "grad_norm": 0.10720349103212357,
      "learning_rate": 8.794723348272488e-05,
      "loss": 1.7086,
      "step": 62880
    },
    {
      "epoch": 0.12414496010540113,
      "grad_norm": 0.10631818324327469,
      "learning_rate": 8.794089325836018e-05,
      "loss": 1.7089,
      "step": 62912
    },
    {
      "epoch": 0.12420810606679758,
      "grad_norm": 0.10009995102882385,
      "learning_rate": 8.793455303399549e-05,
      "loss": 1.7149,
      "step": 62944
    },
    {
      "epoch": 0.12427125202819402,
      "grad_norm": 0.09808431565761566,
      "learning_rate": 8.79282128096308e-05,
      "loss": 1.7085,
      "step": 62976
    },
    {
      "epoch": 0.12433439798959045,
      "grad_norm": 0.10020089894533157,
      "learning_rate": 8.792187258526611e-05,
      "loss": 1.7159,
      "step": 63008
    },
    {
      "epoch": 0.1243975439509869,
      "grad_norm": 0.09870421141386032,
      "learning_rate": 8.791553236090142e-05,
      "loss": 1.7046,
      "step": 63040
    },
    {
      "epoch": 0.12446068991238333,
      "grad_norm": 0.10279548913240433,
      "learning_rate": 8.790919213653673e-05,
      "loss": 1.6907,
      "step": 63072
    },
    {
      "epoch": 0.12452383587377977,
      "grad_norm": 0.12537583708763123,
      "learning_rate": 8.790285191217205e-05,
      "loss": 1.7072,
      "step": 63104
    },
    {
      "epoch": 0.12458698183517622,
      "grad_norm": 0.09676754474639893,
      "learning_rate": 8.789651168780736e-05,
      "loss": 1.7032,
      "step": 63136
    },
    {
      "epoch": 0.12465012779657265,
      "grad_norm": 0.09899728000164032,
      "learning_rate": 8.789017146344267e-05,
      "loss": 1.7181,
      "step": 63168
    },
    {
      "epoch": 0.12471327375796909,
      "grad_norm": 0.10402088612318039,
      "learning_rate": 8.788383123907798e-05,
      "loss": 1.7157,
      "step": 63200
    },
    {
      "epoch": 0.12477641971936554,
      "grad_norm": 0.1084921583533287,
      "learning_rate": 8.787749101471329e-05,
      "loss": 1.7142,
      "step": 63232
    },
    {
      "epoch": 0.12483956568076197,
      "grad_norm": 0.09379089623689651,
      "learning_rate": 8.78711507903486e-05,
      "loss": 1.7125,
      "step": 63264
    },
    {
      "epoch": 0.12490271164215841,
      "grad_norm": 0.0964265912771225,
      "learning_rate": 8.786481056598392e-05,
      "loss": 1.6972,
      "step": 63296
    },
    {
      "epoch": 0.12496585760355486,
      "grad_norm": 0.09247568249702454,
      "learning_rate": 8.785847034161922e-05,
      "loss": 1.7035,
      "step": 63328
    },
    {
      "epoch": 0.1250290035649513,
      "grad_norm": 0.09406764805316925,
      "learning_rate": 8.785213011725452e-05,
      "loss": 1.7115,
      "step": 63360
    },
    {
      "epoch": 0.12509214952634773,
      "grad_norm": 0.09933844208717346,
      "learning_rate": 8.784578989288984e-05,
      "loss": 1.7162,
      "step": 63392
    },
    {
      "epoch": 0.12515529548774418,
      "grad_norm": 0.10537005960941315,
      "learning_rate": 8.783944966852515e-05,
      "loss": 1.7138,
      "step": 63424
    },
    {
      "epoch": 0.12521844144914063,
      "grad_norm": 0.10083386301994324,
      "learning_rate": 8.783310944416045e-05,
      "loss": 1.717,
      "step": 63456
    },
    {
      "epoch": 0.12528158741053705,
      "grad_norm": 0.10026167333126068,
      "learning_rate": 8.782676921979577e-05,
      "loss": 1.7303,
      "step": 63488
    },
    {
      "epoch": 0.1253447333719335,
      "grad_norm": 0.09965796023607254,
      "learning_rate": 8.782042899543108e-05,
      "loss": 1.703,
      "step": 63520
    },
    {
      "epoch": 0.12540787933332995,
      "grad_norm": 0.09772130101919174,
      "learning_rate": 8.78140887710664e-05,
      "loss": 1.7146,
      "step": 63552
    },
    {
      "epoch": 0.12547102529472637,
      "grad_norm": 0.10148956626653671,
      "learning_rate": 8.78077485467017e-05,
      "loss": 1.7183,
      "step": 63584
    },
    {
      "epoch": 0.12553417125612282,
      "grad_norm": 0.10039090365171432,
      "learning_rate": 8.780140832233701e-05,
      "loss": 1.7228,
      "step": 63616
    },
    {
      "epoch": 0.12559731721751927,
      "grad_norm": 0.10283941775560379,
      "learning_rate": 8.779506809797232e-05,
      "loss": 1.6946,
      "step": 63648
    },
    {
      "epoch": 0.1256604631789157,
      "grad_norm": 0.09719688445329666,
      "learning_rate": 8.778872787360764e-05,
      "loss": 1.7038,
      "step": 63680
    },
    {
      "epoch": 0.12572360914031214,
      "grad_norm": 0.094937764108181,
      "learning_rate": 8.778238764924294e-05,
      "loss": 1.7007,
      "step": 63712
    },
    {
      "epoch": 0.12578675510170859,
      "grad_norm": 0.11529593169689178,
      "learning_rate": 8.777604742487826e-05,
      "loss": 1.7021,
      "step": 63744
    },
    {
      "epoch": 0.125849901063105,
      "grad_norm": 0.1040341705083847,
      "learning_rate": 8.776970720051356e-05,
      "loss": 1.7168,
      "step": 63776
    },
    {
      "epoch": 0.12591304702450146,
      "grad_norm": 0.12705889344215393,
      "learning_rate": 8.776336697614887e-05,
      "loss": 1.7126,
      "step": 63808
    },
    {
      "epoch": 0.1259761929858979,
      "grad_norm": 0.1022736057639122,
      "learning_rate": 8.775702675178419e-05,
      "loss": 1.6892,
      "step": 63840
    },
    {
      "epoch": 0.12603933894729433,
      "grad_norm": 0.09732268750667572,
      "learning_rate": 8.775068652741949e-05,
      "loss": 1.7202,
      "step": 63872
    },
    {
      "epoch": 0.12610248490869078,
      "grad_norm": 0.11430452764034271,
      "learning_rate": 8.77443463030548e-05,
      "loss": 1.7009,
      "step": 63904
    },
    {
      "epoch": 0.12616563087008723,
      "grad_norm": 0.10175000131130219,
      "learning_rate": 8.773800607869012e-05,
      "loss": 1.7026,
      "step": 63936
    },
    {
      "epoch": 0.12622877683148365,
      "grad_norm": 0.1036100834608078,
      "learning_rate": 8.773166585432543e-05,
      "loss": 1.7036,
      "step": 63968
    },
    {
      "epoch": 0.1262919227928801,
      "grad_norm": 0.09711123257875443,
      "learning_rate": 8.772532562996073e-05,
      "loss": 1.7073,
      "step": 64000
    },
    {
      "epoch": 0.12635506875427654,
      "grad_norm": 0.10777747631072998,
      "learning_rate": 8.771898540559605e-05,
      "loss": 1.7026,
      "step": 64032
    },
    {
      "epoch": 0.12641821471567297,
      "grad_norm": 0.10371820628643036,
      "learning_rate": 8.771264518123136e-05,
      "loss": 1.7074,
      "step": 64064
    },
    {
      "epoch": 0.12648136067706942,
      "grad_norm": 0.10325604677200317,
      "learning_rate": 8.770630495686667e-05,
      "loss": 1.7056,
      "step": 64096
    },
    {
      "epoch": 0.12654450663846586,
      "grad_norm": 0.10680054873228073,
      "learning_rate": 8.769996473250198e-05,
      "loss": 1.7121,
      "step": 64128
    },
    {
      "epoch": 0.12660765259986229,
      "grad_norm": 0.1007707491517067,
      "learning_rate": 8.769362450813729e-05,
      "loss": 1.7146,
      "step": 64160
    },
    {
      "epoch": 0.12667079856125874,
      "grad_norm": 0.11234920471906662,
      "learning_rate": 8.768728428377259e-05,
      "loss": 1.7141,
      "step": 64192
    },
    {
      "epoch": 0.12673394452265518,
      "grad_norm": 0.10464319586753845,
      "learning_rate": 8.76809440594079e-05,
      "loss": 1.7175,
      "step": 64224
    },
    {
      "epoch": 0.1267970904840516,
      "grad_norm": 0.09534218162298203,
      "learning_rate": 8.76746038350432e-05,
      "loss": 1.6995,
      "step": 64256
    },
    {
      "epoch": 0.12686023644544805,
      "grad_norm": 0.10238602757453918,
      "learning_rate": 8.766826361067852e-05,
      "loss": 1.7134,
      "step": 64288
    },
    {
      "epoch": 0.1269233824068445,
      "grad_norm": 0.09826625883579254,
      "learning_rate": 8.766192338631384e-05,
      "loss": 1.708,
      "step": 64320
    },
    {
      "epoch": 0.12698652836824093,
      "grad_norm": 0.10708026587963104,
      "learning_rate": 8.765558316194915e-05,
      "loss": 1.7147,
      "step": 64352
    },
    {
      "epoch": 0.12704967432963737,
      "grad_norm": 0.0951545387506485,
      "learning_rate": 8.764924293758445e-05,
      "loss": 1.7102,
      "step": 64384
    },
    {
      "epoch": 0.12711282029103382,
      "grad_norm": 0.09865587949752808,
      "learning_rate": 8.764290271321977e-05,
      "loss": 1.713,
      "step": 64416
    },
    {
      "epoch": 0.12717596625243024,
      "grad_norm": 0.10516698658466339,
      "learning_rate": 8.763656248885508e-05,
      "loss": 1.7155,
      "step": 64448
    },
    {
      "epoch": 0.1272391122138267,
      "grad_norm": 0.0997084230184555,
      "learning_rate": 8.76302222644904e-05,
      "loss": 1.7251,
      "step": 64480
    },
    {
      "epoch": 0.12730225817522314,
      "grad_norm": 0.09966826438903809,
      "learning_rate": 8.762388204012571e-05,
      "loss": 1.7083,
      "step": 64512
    },
    {
      "epoch": 0.12736540413661956,
      "grad_norm": 0.10060398280620575,
      "learning_rate": 8.761754181576101e-05,
      "loss": 1.702,
      "step": 64544
    },
    {
      "epoch": 0.127428550098016,
      "grad_norm": 0.1012912318110466,
      "learning_rate": 8.761120159139633e-05,
      "loss": 1.6979,
      "step": 64576
    },
    {
      "epoch": 0.12749169605941246,
      "grad_norm": 0.10541724413633347,
      "learning_rate": 8.760486136703163e-05,
      "loss": 1.7123,
      "step": 64608
    },
    {
      "epoch": 0.12755484202080888,
      "grad_norm": 0.09584583342075348,
      "learning_rate": 8.759852114266694e-05,
      "loss": 1.6894,
      "step": 64640
    },
    {
      "epoch": 0.12761798798220533,
      "grad_norm": 0.09712526202201843,
      "learning_rate": 8.759218091830224e-05,
      "loss": 1.7176,
      "step": 64672
    },
    {
      "epoch": 0.12768113394360178,
      "grad_norm": 0.10084573924541473,
      "learning_rate": 8.758584069393756e-05,
      "loss": 1.7092,
      "step": 64704
    },
    {
      "epoch": 0.1277442799049982,
      "grad_norm": 0.09810017794370651,
      "learning_rate": 8.757950046957287e-05,
      "loss": 1.7095,
      "step": 64736
    },
    {
      "epoch": 0.12780742586639465,
      "grad_norm": 0.09851551055908203,
      "learning_rate": 8.757316024520819e-05,
      "loss": 1.7135,
      "step": 64768
    },
    {
      "epoch": 0.1278705718277911,
      "grad_norm": 0.11118162423372269,
      "learning_rate": 8.756682002084349e-05,
      "loss": 1.6993,
      "step": 64800
    },
    {
      "epoch": 0.12793371778918752,
      "grad_norm": 0.09544149041175842,
      "learning_rate": 8.75604797964788e-05,
      "loss": 1.7134,
      "step": 64832
    },
    {
      "epoch": 0.12799686375058397,
      "grad_norm": 0.10334372520446777,
      "learning_rate": 8.755413957211412e-05,
      "loss": 1.7134,
      "step": 64864
    },
    {
      "epoch": 0.12806000971198042,
      "grad_norm": 0.10433602333068848,
      "learning_rate": 8.754779934774943e-05,
      "loss": 1.6914,
      "step": 64896
    },
    {
      "epoch": 0.12812315567337687,
      "grad_norm": 0.0969606265425682,
      "learning_rate": 8.754145912338473e-05,
      "loss": 1.7003,
      "step": 64928
    },
    {
      "epoch": 0.1281863016347733,
      "grad_norm": 0.09980641305446625,
      "learning_rate": 8.753511889902005e-05,
      "loss": 1.7107,
      "step": 64960
    },
    {
      "epoch": 0.12824944759616974,
      "grad_norm": 0.10546495020389557,
      "learning_rate": 8.752877867465536e-05,
      "loss": 1.6991,
      "step": 64992
    },
    {
      "epoch": 0.1283125935575662,
      "grad_norm": 0.10080936551094055,
      "learning_rate": 8.752243845029066e-05,
      "loss": 1.7025,
      "step": 65024
    },
    {
      "epoch": 0.1283757395189626,
      "grad_norm": 0.10327662527561188,
      "learning_rate": 8.751609822592596e-05,
      "loss": 1.7167,
      "step": 65056
    },
    {
      "epoch": 0.12843888548035906,
      "grad_norm": 0.09645973145961761,
      "learning_rate": 8.750975800156128e-05,
      "loss": 1.7163,
      "step": 65088
    },
    {
      "epoch": 0.1285020314417555,
      "grad_norm": 0.09340842813253403,
      "learning_rate": 8.750341777719659e-05,
      "loss": 1.7169,
      "step": 65120
    },
    {
      "epoch": 0.12856517740315193,
      "grad_norm": 0.10235944390296936,
      "learning_rate": 8.74970775528319e-05,
      "loss": 1.6974,
      "step": 65152
    },
    {
      "epoch": 0.12862832336454838,
      "grad_norm": 0.11224786192178726,
      "learning_rate": 8.749073732846722e-05,
      "loss": 1.7222,
      "step": 65184
    },
    {
      "epoch": 0.12869146932594483,
      "grad_norm": 0.11072751879692078,
      "learning_rate": 8.748439710410252e-05,
      "loss": 1.6987,
      "step": 65216
    },
    {
      "epoch": 0.12875461528734125,
      "grad_norm": 0.09741399437189102,
      "learning_rate": 8.747805687973784e-05,
      "loss": 1.7056,
      "step": 65248
    },
    {
      "epoch": 0.1288177612487377,
      "grad_norm": 0.09321194142103195,
      "learning_rate": 8.747171665537315e-05,
      "loss": 1.6989,
      "step": 65280
    },
    {
      "epoch": 0.12888090721013415,
      "grad_norm": 0.10490649193525314,
      "learning_rate": 8.746537643100847e-05,
      "loss": 1.6959,
      "step": 65312
    },
    {
      "epoch": 0.12894405317153057,
      "grad_norm": 0.09913329780101776,
      "learning_rate": 8.745903620664377e-05,
      "loss": 1.7053,
      "step": 65344
    },
    {
      "epoch": 0.12900719913292702,
      "grad_norm": 0.10102079063653946,
      "learning_rate": 8.745269598227908e-05,
      "loss": 1.69,
      "step": 65376
    },
    {
      "epoch": 0.12907034509432347,
      "grad_norm": 0.09921082854270935,
      "learning_rate": 8.74463557579144e-05,
      "loss": 1.6981,
      "step": 65408
    },
    {
      "epoch": 0.1291334910557199,
      "grad_norm": 0.0991610437631607,
      "learning_rate": 8.74400155335497e-05,
      "loss": 1.7063,
      "step": 65440
    },
    {
      "epoch": 0.12919663701711634,
      "grad_norm": 0.10011253505945206,
      "learning_rate": 8.7433675309185e-05,
      "loss": 1.7143,
      "step": 65472
    },
    {
      "epoch": 0.1292597829785128,
      "grad_norm": 0.10191270709037781,
      "learning_rate": 8.742733508482031e-05,
      "loss": 1.704,
      "step": 65504
    },
    {
      "epoch": 0.1293229289399092,
      "grad_norm": 0.10369692742824554,
      "learning_rate": 8.742099486045563e-05,
      "loss": 1.694,
      "step": 65536
    },
    {
      "epoch": 0.12938607490130566,
      "grad_norm": 0.10494969040155411,
      "learning_rate": 8.741465463609094e-05,
      "loss": 1.6927,
      "step": 65568
    },
    {
      "epoch": 0.1294492208627021,
      "grad_norm": 0.10710588842630386,
      "learning_rate": 8.740831441172624e-05,
      "loss": 1.706,
      "step": 65600
    },
    {
      "epoch": 0.12951236682409853,
      "grad_norm": 0.12005143612623215,
      "learning_rate": 8.740197418736156e-05,
      "loss": 1.7084,
      "step": 65632
    },
    {
      "epoch": 0.12957551278549498,
      "grad_norm": 0.09254439175128937,
      "learning_rate": 8.739563396299687e-05,
      "loss": 1.6939,
      "step": 65664
    },
    {
      "epoch": 0.12963865874689143,
      "grad_norm": 0.10031227767467499,
      "learning_rate": 8.738929373863219e-05,
      "loss": 1.7105,
      "step": 65696
    },
    {
      "epoch": 0.12970180470828785,
      "grad_norm": 0.10078656673431396,
      "learning_rate": 8.738295351426749e-05,
      "loss": 1.6924,
      "step": 65728
    },
    {
      "epoch": 0.1297649506696843,
      "grad_norm": 0.10685756057500839,
      "learning_rate": 8.73766132899028e-05,
      "loss": 1.701,
      "step": 65760
    },
    {
      "epoch": 0.12982809663108075,
      "grad_norm": 0.09442702680826187,
      "learning_rate": 8.737027306553812e-05,
      "loss": 1.6992,
      "step": 65792
    },
    {
      "epoch": 0.12989124259247717,
      "grad_norm": 0.10017342865467072,
      "learning_rate": 8.736393284117342e-05,
      "loss": 1.7157,
      "step": 65824
    },
    {
      "epoch": 0.12995438855387362,
      "grad_norm": 0.09943683445453644,
      "learning_rate": 8.735759261680873e-05,
      "loss": 1.7059,
      "step": 65856
    },
    {
      "epoch": 0.13001753451527007,
      "grad_norm": 0.09687332063913345,
      "learning_rate": 8.735125239244403e-05,
      "loss": 1.7026,
      "step": 65888
    },
    {
      "epoch": 0.1300806804766665,
      "grad_norm": 0.09685111790895462,
      "learning_rate": 8.734491216807935e-05,
      "loss": 1.6921,
      "step": 65920
    },
    {
      "epoch": 0.13014382643806294,
      "grad_norm": 0.10358686745166779,
      "learning_rate": 8.733857194371466e-05,
      "loss": 1.6936,
      "step": 65952
    },
    {
      "epoch": 0.1302069723994594,
      "grad_norm": 0.09398703277111053,
      "learning_rate": 8.733223171934998e-05,
      "loss": 1.6947,
      "step": 65984
    },
    {
      "epoch": 0.1302701183608558,
      "grad_norm": 0.10856349021196365,
      "learning_rate": 8.732589149498528e-05,
      "loss": 1.6984,
      "step": 66016
    },
    {
      "epoch": 0.13033326432225226,
      "grad_norm": 0.10347195714712143,
      "learning_rate": 8.731955127062059e-05,
      "loss": 1.696,
      "step": 66048
    },
    {
      "epoch": 0.1303964102836487,
      "grad_norm": 0.10519301891326904,
      "learning_rate": 8.73132110462559e-05,
      "loss": 1.7046,
      "step": 66080
    },
    {
      "epoch": 0.13045955624504513,
      "grad_norm": 0.10505388677120209,
      "learning_rate": 8.730687082189122e-05,
      "loss": 1.6987,
      "step": 66112
    },
    {
      "epoch": 0.13052270220644158,
      "grad_norm": 0.10282491147518158,
      "learning_rate": 8.730053059752652e-05,
      "loss": 1.6933,
      "step": 66144
    },
    {
      "epoch": 0.13058584816783803,
      "grad_norm": 0.09779899567365646,
      "learning_rate": 8.729419037316184e-05,
      "loss": 1.6921,
      "step": 66176
    },
    {
      "epoch": 0.13064899412923445,
      "grad_norm": 0.10245571285486221,
      "learning_rate": 8.728785014879715e-05,
      "loss": 1.6907,
      "step": 66208
    },
    {
      "epoch": 0.1307121400906309,
      "grad_norm": 0.11757884919643402,
      "learning_rate": 8.728150992443245e-05,
      "loss": 1.7188,
      "step": 66240
    },
    {
      "epoch": 0.13077528605202735,
      "grad_norm": 0.09794843941926956,
      "learning_rate": 8.727516970006777e-05,
      "loss": 1.7053,
      "step": 66272
    },
    {
      "epoch": 0.13083843201342377,
      "grad_norm": 0.09846805781126022,
      "learning_rate": 8.726882947570307e-05,
      "loss": 1.6925,
      "step": 66304
    },
    {
      "epoch": 0.13090157797482022,
      "grad_norm": 0.10118546336889267,
      "learning_rate": 8.726248925133838e-05,
      "loss": 1.7072,
      "step": 66336
    },
    {
      "epoch": 0.13096472393621666,
      "grad_norm": 0.09692150354385376,
      "learning_rate": 8.72561490269737e-05,
      "loss": 1.7097,
      "step": 66368
    },
    {
      "epoch": 0.1310278698976131,
      "grad_norm": 0.09960059821605682,
      "learning_rate": 8.7249808802609e-05,
      "loss": 1.6942,
      "step": 66400
    },
    {
      "epoch": 0.13109101585900954,
      "grad_norm": 0.0997898280620575,
      "learning_rate": 8.724346857824431e-05,
      "loss": 1.7031,
      "step": 66432
    },
    {
      "epoch": 0.13115416182040598,
      "grad_norm": 0.10471518337726593,
      "learning_rate": 8.723712835387963e-05,
      "loss": 1.7006,
      "step": 66464
    },
    {
      "epoch": 0.1312173077818024,
      "grad_norm": 0.09863907098770142,
      "learning_rate": 8.723078812951494e-05,
      "loss": 1.7123,
      "step": 66496
    },
    {
      "epoch": 0.13128045374319885,
      "grad_norm": 0.10759855061769485,
      "learning_rate": 8.722444790515026e-05,
      "loss": 1.6977,
      "step": 66528
    },
    {
      "epoch": 0.1313435997045953,
      "grad_norm": 0.11002608388662338,
      "learning_rate": 8.721810768078556e-05,
      "loss": 1.6967,
      "step": 66560
    },
    {
      "epoch": 0.13140674566599175,
      "grad_norm": 0.12025301903486252,
      "learning_rate": 8.721176745642087e-05,
      "loss": 1.6869,
      "step": 66592
    },
    {
      "epoch": 0.13146989162738817,
      "grad_norm": 0.0912991538643837,
      "learning_rate": 8.720542723205619e-05,
      "loss": 1.72,
      "step": 66624
    },
    {
      "epoch": 0.13153303758878462,
      "grad_norm": 0.09873910993337631,
      "learning_rate": 8.719908700769149e-05,
      "loss": 1.7048,
      "step": 66656
    },
    {
      "epoch": 0.13159618355018107,
      "grad_norm": 0.10642040520906448,
      "learning_rate": 8.71927467833268e-05,
      "loss": 1.7016,
      "step": 66688
    },
    {
      "epoch": 0.1316593295115775,
      "grad_norm": 0.10963353514671326,
      "learning_rate": 8.71864065589621e-05,
      "loss": 1.6879,
      "step": 66720
    },
    {
      "epoch": 0.13172247547297394,
      "grad_norm": 0.10779455304145813,
      "learning_rate": 8.718006633459742e-05,
      "loss": 1.7122,
      "step": 66752
    },
    {
      "epoch": 0.1317856214343704,
      "grad_norm": 0.11267006397247314,
      "learning_rate": 8.717372611023273e-05,
      "loss": 1.7023,
      "step": 66784
    },
    {
      "epoch": 0.1318487673957668,
      "grad_norm": 0.10699082911014557,
      "learning_rate": 8.716738588586803e-05,
      "loss": 1.695,
      "step": 66816
    },
    {
      "epoch": 0.13191191335716326,
      "grad_norm": 0.10018612444400787,
      "learning_rate": 8.716104566150335e-05,
      "loss": 1.6949,
      "step": 66848
    },
    {
      "epoch": 0.1319750593185597,
      "grad_norm": 0.10509435087442398,
      "learning_rate": 8.715470543713866e-05,
      "loss": 1.7061,
      "step": 66880
    },
    {
      "epoch": 0.13203820527995613,
      "grad_norm": 0.09271152317523956,
      "learning_rate": 8.714836521277398e-05,
      "loss": 1.7038,
      "step": 66912
    },
    {
      "epoch": 0.13210135124135258,
      "grad_norm": 0.1133396327495575,
      "learning_rate": 8.714202498840928e-05,
      "loss": 1.6939,
      "step": 66944
    },
    {
      "epoch": 0.13216449720274903,
      "grad_norm": 0.0986180230975151,
      "learning_rate": 8.713568476404459e-05,
      "loss": 1.6971,
      "step": 66976
    },
    {
      "epoch": 0.13222764316414545,
      "grad_norm": 0.10401751846075058,
      "learning_rate": 8.71293445396799e-05,
      "loss": 1.6987,
      "step": 67008
    },
    {
      "epoch": 0.1322907891255419,
      "grad_norm": 0.10454605519771576,
      "learning_rate": 8.712300431531522e-05,
      "loss": 1.705,
      "step": 67040
    },
    {
      "epoch": 0.13235393508693835,
      "grad_norm": 0.09792706370353699,
      "learning_rate": 8.711666409095052e-05,
      "loss": 1.7028,
      "step": 67072
    },
    {
      "epoch": 0.13241708104833477,
      "grad_norm": 0.09791544824838638,
      "learning_rate": 8.711032386658584e-05,
      "loss": 1.6958,
      "step": 67104
    },
    {
      "epoch": 0.13248022700973122,
      "grad_norm": 0.1010831817984581,
      "learning_rate": 8.710398364222114e-05,
      "loss": 1.7015,
      "step": 67136
    },
    {
      "epoch": 0.13254337297112767,
      "grad_norm": 0.096627376973629,
      "learning_rate": 8.709764341785645e-05,
      "loss": 1.6955,
      "step": 67168
    },
    {
      "epoch": 0.1326065189325241,
      "grad_norm": 0.10321414470672607,
      "learning_rate": 8.709130319349177e-05,
      "loss": 1.7067,
      "step": 67200
    },
    {
      "epoch": 0.13266966489392054,
      "grad_norm": 0.09757894277572632,
      "learning_rate": 8.708496296912707e-05,
      "loss": 1.6998,
      "step": 67232
    },
    {
      "epoch": 0.132732810855317,
      "grad_norm": 0.10374078154563904,
      "learning_rate": 8.707862274476238e-05,
      "loss": 1.7074,
      "step": 67264
    },
    {
      "epoch": 0.1327959568167134,
      "grad_norm": 0.11785580217838287,
      "learning_rate": 8.70722825203977e-05,
      "loss": 1.7011,
      "step": 67296
    },
    {
      "epoch": 0.13285910277810986,
      "grad_norm": 0.09936001896858215,
      "learning_rate": 8.706594229603301e-05,
      "loss": 1.6901,
      "step": 67328
    },
    {
      "epoch": 0.1329222487395063,
      "grad_norm": 0.09808773547410965,
      "learning_rate": 8.705960207166831e-05,
      "loss": 1.7006,
      "step": 67360
    },
    {
      "epoch": 0.13298539470090273,
      "grad_norm": 0.10902006924152374,
      "learning_rate": 8.705326184730363e-05,
      "loss": 1.7021,
      "step": 67392
    },
    {
      "epoch": 0.13304854066229918,
      "grad_norm": 0.10547816008329391,
      "learning_rate": 8.704692162293894e-05,
      "loss": 1.6901,
      "step": 67424
    },
    {
      "epoch": 0.13311168662369563,
      "grad_norm": 0.09985819458961487,
      "learning_rate": 8.704058139857426e-05,
      "loss": 1.6839,
      "step": 67456
    },
    {
      "epoch": 0.13317483258509205,
      "grad_norm": 0.10067432373762131,
      "learning_rate": 8.703424117420956e-05,
      "loss": 1.6984,
      "step": 67488
    },
    {
      "epoch": 0.1332379785464885,
      "grad_norm": 0.10011441260576248,
      "learning_rate": 8.702790094984486e-05,
      "loss": 1.6826,
      "step": 67520
    },
    {
      "epoch": 0.13330112450788495,
      "grad_norm": 0.09623958170413971,
      "learning_rate": 8.702156072548017e-05,
      "loss": 1.7043,
      "step": 67552
    },
    {
      "epoch": 0.13336427046928137,
      "grad_norm": 0.10766778886318207,
      "learning_rate": 8.701522050111549e-05,
      "loss": 1.694,
      "step": 67584
    },
    {
      "epoch": 0.13342741643067782,
      "grad_norm": 0.10048247873783112,
      "learning_rate": 8.700888027675079e-05,
      "loss": 1.6976,
      "step": 67616
    },
    {
      "epoch": 0.13349056239207427,
      "grad_norm": 0.1025388091802597,
      "learning_rate": 8.70025400523861e-05,
      "loss": 1.7094,
      "step": 67648
    },
    {
      "epoch": 0.1335537083534707,
      "grad_norm": 0.0898028239607811,
      "learning_rate": 8.699619982802142e-05,
      "loss": 1.7003,
      "step": 67680
    },
    {
      "epoch": 0.13361685431486714,
      "grad_norm": 0.10830457508563995,
      "learning_rate": 8.698985960365673e-05,
      "loss": 1.6987,
      "step": 67712
    },
    {
      "epoch": 0.1336800002762636,
      "grad_norm": 0.0960337445139885,
      "learning_rate": 8.698351937929203e-05,
      "loss": 1.6911,
      "step": 67744
    },
    {
      "epoch": 0.13374314623766,
      "grad_norm": 0.10730069130659103,
      "learning_rate": 8.697717915492735e-05,
      "loss": 1.6852,
      "step": 67776
    },
    {
      "epoch": 0.13380629219905646,
      "grad_norm": 0.10252933204174042,
      "learning_rate": 8.697083893056266e-05,
      "loss": 1.7069,
      "step": 67808
    },
    {
      "epoch": 0.1338694381604529,
      "grad_norm": 0.09704408049583435,
      "learning_rate": 8.696449870619798e-05,
      "loss": 1.6924,
      "step": 67840
    },
    {
      "epoch": 0.13393258412184933,
      "grad_norm": 0.09831680357456207,
      "learning_rate": 8.695815848183329e-05,
      "loss": 1.7129,
      "step": 67872
    },
    {
      "epoch": 0.13399573008324578,
      "grad_norm": 0.09816844016313553,
      "learning_rate": 8.695181825746859e-05,
      "loss": 1.6999,
      "step": 67904
    },
    {
      "epoch": 0.13405887604464223,
      "grad_norm": 0.09870949387550354,
      "learning_rate": 8.69454780331039e-05,
      "loss": 1.6901,
      "step": 67936
    },
    {
      "epoch": 0.13412202200603865,
      "grad_norm": 0.0937604159116745,
      "learning_rate": 8.693913780873921e-05,
      "loss": 1.6785,
      "step": 67968
    },
    {
      "epoch": 0.1341851679674351,
      "grad_norm": 0.10146182775497437,
      "learning_rate": 8.693279758437452e-05,
      "loss": 1.6888,
      "step": 68000
    },
    {
      "epoch": 0.13424831392883155,
      "grad_norm": 0.09773188829421997,
      "learning_rate": 8.692645736000982e-05,
      "loss": 1.693,
      "step": 68032
    },
    {
      "epoch": 0.13431145989022797,
      "grad_norm": 0.10485115647315979,
      "learning_rate": 8.692011713564514e-05,
      "loss": 1.705,
      "step": 68064
    },
    {
      "epoch": 0.13437460585162442,
      "grad_norm": 0.09446601569652557,
      "learning_rate": 8.691377691128045e-05,
      "loss": 1.6872,
      "step": 68096
    },
    {
      "epoch": 0.13443775181302087,
      "grad_norm": 0.09718290716409683,
      "learning_rate": 8.690743668691577e-05,
      "loss": 1.7068,
      "step": 68128
    },
    {
      "epoch": 0.1345008977744173,
      "grad_norm": 0.10884904116392136,
      "learning_rate": 8.690109646255107e-05,
      "loss": 1.6895,
      "step": 68160
    },
    {
      "epoch": 0.13456404373581374,
      "grad_norm": 0.10205969959497452,
      "learning_rate": 8.689475623818638e-05,
      "loss": 1.6945,
      "step": 68192
    },
    {
      "epoch": 0.1346271896972102,
      "grad_norm": 0.09928876906633377,
      "learning_rate": 8.68884160138217e-05,
      "loss": 1.6973,
      "step": 68224
    },
    {
      "epoch": 0.1346903356586066,
      "grad_norm": 0.09771773219108582,
      "learning_rate": 8.688207578945701e-05,
      "loss": 1.6968,
      "step": 68256
    },
    {
      "epoch": 0.13475348162000306,
      "grad_norm": 0.10454849153757095,
      "learning_rate": 8.687573556509231e-05,
      "loss": 1.6856,
      "step": 68288
    },
    {
      "epoch": 0.1348166275813995,
      "grad_norm": 0.10128848254680634,
      "learning_rate": 8.686939534072763e-05,
      "loss": 1.6928,
      "step": 68320
    },
    {
      "epoch": 0.13487977354279596,
      "grad_norm": 0.10038738697767258,
      "learning_rate": 8.686305511636293e-05,
      "loss": 1.693,
      "step": 68352
    },
    {
      "epoch": 0.13494291950419238,
      "grad_norm": 0.09970016032457352,
      "learning_rate": 8.685671489199824e-05,
      "loss": 1.68,
      "step": 68384
    },
    {
      "epoch": 0.13500606546558883,
      "grad_norm": 0.10426627844572067,
      "learning_rate": 8.685037466763356e-05,
      "loss": 1.6847,
      "step": 68416
    },
    {
      "epoch": 0.13506921142698528,
      "grad_norm": 0.09571852535009384,
      "learning_rate": 8.684403444326886e-05,
      "loss": 1.6855,
      "step": 68448
    },
    {
      "epoch": 0.1351323573883817,
      "grad_norm": 0.10176054388284683,
      "learning_rate": 8.683769421890417e-05,
      "loss": 1.6867,
      "step": 68480
    },
    {
      "epoch": 0.13519550334977815,
      "grad_norm": 0.09997376054525375,
      "learning_rate": 8.683135399453949e-05,
      "loss": 1.6961,
      "step": 68512
    },
    {
      "epoch": 0.1352586493111746,
      "grad_norm": 0.09780901670455933,
      "learning_rate": 8.68250137701748e-05,
      "loss": 1.6979,
      "step": 68544
    },
    {
      "epoch": 0.13532179527257102,
      "grad_norm": 0.11168446391820908,
      "learning_rate": 8.68186735458101e-05,
      "loss": 1.7007,
      "step": 68576
    },
    {
      "epoch": 0.13538494123396747,
      "grad_norm": 0.10168793052434921,
      "learning_rate": 8.681233332144542e-05,
      "loss": 1.6982,
      "step": 68608
    },
    {
      "epoch": 0.13544808719536391,
      "grad_norm": 0.10820365697145462,
      "learning_rate": 8.680599309708073e-05,
      "loss": 1.695,
      "step": 68640
    },
    {
      "epoch": 0.13551123315676034,
      "grad_norm": 0.10211111605167389,
      "learning_rate": 8.679965287271605e-05,
      "loss": 1.6922,
      "step": 68672
    },
    {
      "epoch": 0.13557437911815678,
      "grad_norm": 0.10094490647315979,
      "learning_rate": 8.679331264835135e-05,
      "loss": 1.6898,
      "step": 68704
    },
    {
      "epoch": 0.13563752507955323,
      "grad_norm": 0.100893534719944,
      "learning_rate": 8.678697242398666e-05,
      "loss": 1.7035,
      "step": 68736
    },
    {
      "epoch": 0.13570067104094966,
      "grad_norm": 0.10502218455076218,
      "learning_rate": 8.678063219962196e-05,
      "loss": 1.7052,
      "step": 68768
    },
    {
      "epoch": 0.1357638170023461,
      "grad_norm": 0.10092008858919144,
      "learning_rate": 8.677429197525728e-05,
      "loss": 1.7052,
      "step": 68800
    },
    {
      "epoch": 0.13582696296374255,
      "grad_norm": 0.09909343719482422,
      "learning_rate": 8.676795175089258e-05,
      "loss": 1.696,
      "step": 68832
    },
    {
      "epoch": 0.13589010892513897,
      "grad_norm": 0.09472015500068665,
      "learning_rate": 8.67616115265279e-05,
      "loss": 1.7003,
      "step": 68864
    },
    {
      "epoch": 0.13595325488653542,
      "grad_norm": 0.09557487815618515,
      "learning_rate": 8.675527130216321e-05,
      "loss": 1.687,
      "step": 68896
    },
    {
      "epoch": 0.13601640084793187,
      "grad_norm": 0.10137154161930084,
      "learning_rate": 8.674893107779852e-05,
      "loss": 1.6939,
      "step": 68928
    },
    {
      "epoch": 0.1360795468093283,
      "grad_norm": 0.12496348470449448,
      "learning_rate": 8.674259085343382e-05,
      "loss": 1.6881,
      "step": 68960
    },
    {
      "epoch": 0.13614269277072474,
      "grad_norm": 0.10233995318412781,
      "learning_rate": 8.673625062906914e-05,
      "loss": 1.7001,
      "step": 68992
    },
    {
      "epoch": 0.1362058387321212,
      "grad_norm": 0.09763490408658981,
      "learning_rate": 8.672991040470445e-05,
      "loss": 1.6847,
      "step": 69024
    },
    {
      "epoch": 0.13626898469351761,
      "grad_norm": 0.0937042310833931,
      "learning_rate": 8.672357018033977e-05,
      "loss": 1.6921,
      "step": 69056
    },
    {
      "epoch": 0.13633213065491406,
      "grad_norm": 0.10340123623609543,
      "learning_rate": 8.671722995597508e-05,
      "loss": 1.6991,
      "step": 69088
    },
    {
      "epoch": 0.1363952766163105,
      "grad_norm": 0.09613881260156631,
      "learning_rate": 8.671088973161038e-05,
      "loss": 1.697,
      "step": 69120
    },
    {
      "epoch": 0.13645842257770693,
      "grad_norm": 0.10652626305818558,
      "learning_rate": 8.67045495072457e-05,
      "loss": 1.6987,
      "step": 69152
    },
    {
      "epoch": 0.13652156853910338,
      "grad_norm": 0.1023712083697319,
      "learning_rate": 8.6698209282881e-05,
      "loss": 1.6832,
      "step": 69184
    },
    {
      "epoch": 0.13658471450049983,
      "grad_norm": 0.09921175241470337,
      "learning_rate": 8.669186905851631e-05,
      "loss": 1.6974,
      "step": 69216
    },
    {
      "epoch": 0.13664786046189625,
      "grad_norm": 0.09992415457963943,
      "learning_rate": 8.668552883415161e-05,
      "loss": 1.6914,
      "step": 69248
    },
    {
      "epoch": 0.1367110064232927,
      "grad_norm": 0.09259074181318283,
      "learning_rate": 8.667918860978693e-05,
      "loss": 1.6866,
      "step": 69280
    },
    {
      "epoch": 0.13677415238468915,
      "grad_norm": 0.10414275527000427,
      "learning_rate": 8.667284838542224e-05,
      "loss": 1.716,
      "step": 69312
    },
    {
      "epoch": 0.13683729834608557,
      "grad_norm": 0.10414236783981323,
      "learning_rate": 8.666650816105756e-05,
      "loss": 1.6923,
      "step": 69344
    },
    {
      "epoch": 0.13690044430748202,
      "grad_norm": 0.10356683284044266,
      "learning_rate": 8.666016793669286e-05,
      "loss": 1.6898,
      "step": 69376
    },
    {
      "epoch": 0.13696359026887847,
      "grad_norm": 0.09934966266155243,
      "learning_rate": 8.665382771232817e-05,
      "loss": 1.6875,
      "step": 69408
    },
    {
      "epoch": 0.1370267362302749,
      "grad_norm": 0.1002814769744873,
      "learning_rate": 8.664748748796349e-05,
      "loss": 1.6931,
      "step": 69440
    },
    {
      "epoch": 0.13708988219167134,
      "grad_norm": 0.10130573064088821,
      "learning_rate": 8.66411472635988e-05,
      "loss": 1.7054,
      "step": 69472
    },
    {
      "epoch": 0.1371530281530678,
      "grad_norm": 0.1078111007809639,
      "learning_rate": 8.66348070392341e-05,
      "loss": 1.6925,
      "step": 69504
    },
    {
      "epoch": 0.1372161741144642,
      "grad_norm": 0.10099471360445023,
      "learning_rate": 8.662846681486942e-05,
      "loss": 1.6778,
      "step": 69536
    },
    {
      "epoch": 0.13727932007586066,
      "grad_norm": 0.09544076025485992,
      "learning_rate": 8.662212659050473e-05,
      "loss": 1.6901,
      "step": 69568
    },
    {
      "epoch": 0.1373424660372571,
      "grad_norm": 0.09985319525003433,
      "learning_rate": 8.661578636614003e-05,
      "loss": 1.6984,
      "step": 69600
    },
    {
      "epoch": 0.13740561199865353,
      "grad_norm": 0.09363614767789841,
      "learning_rate": 8.660944614177533e-05,
      "loss": 1.6837,
      "step": 69632
    },
    {
      "epoch": 0.13746875796004998,
      "grad_norm": 0.10166219621896744,
      "learning_rate": 8.660310591741065e-05,
      "loss": 1.6917,
      "step": 69664
    },
    {
      "epoch": 0.13753190392144643,
      "grad_norm": 0.10159972310066223,
      "learning_rate": 8.659676569304596e-05,
      "loss": 1.7055,
      "step": 69696
    },
    {
      "epoch": 0.13759504988284285,
      "grad_norm": 0.10600333660840988,
      "learning_rate": 8.659042546868128e-05,
      "loss": 1.6968,
      "step": 69728
    },
    {
      "epoch": 0.1376581958442393,
      "grad_norm": 0.10244552791118622,
      "learning_rate": 8.658408524431659e-05,
      "loss": 1.6881,
      "step": 69760
    },
    {
      "epoch": 0.13772134180563575,
      "grad_norm": 0.10649577528238297,
      "learning_rate": 8.65777450199519e-05,
      "loss": 1.6974,
      "step": 69792
    },
    {
      "epoch": 0.13778448776703217,
      "grad_norm": 0.10574222356081009,
      "learning_rate": 8.657140479558721e-05,
      "loss": 1.6966,
      "step": 69824
    },
    {
      "epoch": 0.13784763372842862,
      "grad_norm": 0.094143807888031,
      "learning_rate": 8.656506457122252e-05,
      "loss": 1.6726,
      "step": 69856
    },
    {
      "epoch": 0.13791077968982507,
      "grad_norm": 0.09454295784235,
      "learning_rate": 8.655872434685784e-05,
      "loss": 1.6967,
      "step": 69888
    },
    {
      "epoch": 0.1379739256512215,
      "grad_norm": 0.10063689202070236,
      "learning_rate": 8.655238412249314e-05,
      "loss": 1.6965,
      "step": 69920
    },
    {
      "epoch": 0.13803707161261794,
      "grad_norm": 0.10030588507652283,
      "learning_rate": 8.654604389812845e-05,
      "loss": 1.7071,
      "step": 69952
    },
    {
      "epoch": 0.1381002175740144,
      "grad_norm": 0.09860387444496155,
      "learning_rate": 8.653970367376375e-05,
      "loss": 1.6798,
      "step": 69984
    },
    {
      "epoch": 0.13816336353541084,
      "grad_norm": 0.09735734760761261,
      "learning_rate": 8.653336344939907e-05,
      "loss": 1.6821,
      "step": 70016
    },
    {
      "epoch": 0.13822650949680726,
      "grad_norm": 0.09824249148368835,
      "learning_rate": 8.652702322503437e-05,
      "loss": 1.6951,
      "step": 70048
    },
    {
      "epoch": 0.1382896554582037,
      "grad_norm": 0.09516239166259766,
      "learning_rate": 8.652068300066968e-05,
      "loss": 1.6866,
      "step": 70080
    },
    {
      "epoch": 0.13835280141960016,
      "grad_norm": 0.10811588168144226,
      "learning_rate": 8.6514342776305e-05,
      "loss": 1.7058,
      "step": 70112
    },
    {
      "epoch": 0.13841594738099658,
      "grad_norm": 0.09876111894845963,
      "learning_rate": 8.650800255194031e-05,
      "loss": 1.697,
      "step": 70144
    },
    {
      "epoch": 0.13847909334239303,
      "grad_norm": 0.09289198368787766,
      "learning_rate": 8.650166232757561e-05,
      "loss": 1.7014,
      "step": 70176
    },
    {
      "epoch": 0.13854223930378948,
      "grad_norm": 0.1037587970495224,
      "learning_rate": 8.649532210321093e-05,
      "loss": 1.6868,
      "step": 70208
    },
    {
      "epoch": 0.1386053852651859,
      "grad_norm": 0.09075438231229782,
      "learning_rate": 8.648898187884624e-05,
      "loss": 1.6971,
      "step": 70240
    },
    {
      "epoch": 0.13866853122658235,
      "grad_norm": 0.09826972335577011,
      "learning_rate": 8.648264165448156e-05,
      "loss": 1.696,
      "step": 70272
    },
    {
      "epoch": 0.1387316771879788,
      "grad_norm": 0.10796775668859482,
      "learning_rate": 8.647630143011686e-05,
      "loss": 1.691,
      "step": 70304
    },
    {
      "epoch": 0.13879482314937522,
      "grad_norm": 0.10104010999202728,
      "learning_rate": 8.646996120575217e-05,
      "loss": 1.704,
      "step": 70336
    },
    {
      "epoch": 0.13885796911077167,
      "grad_norm": 0.1002320870757103,
      "learning_rate": 8.646362098138749e-05,
      "loss": 1.7012,
      "step": 70368
    },
    {
      "epoch": 0.13892111507216812,
      "grad_norm": 0.10547274351119995,
      "learning_rate": 8.645728075702279e-05,
      "loss": 1.6937,
      "step": 70400
    },
    {
      "epoch": 0.13898426103356454,
      "grad_norm": 0.11008907854557037,
      "learning_rate": 8.64509405326581e-05,
      "loss": 1.7009,
      "step": 70432
    },
    {
      "epoch": 0.139047406994961,
      "grad_norm": 0.10129993408918381,
      "learning_rate": 8.64446003082934e-05,
      "loss": 1.7099,
      "step": 70464
    },
    {
      "epoch": 0.13911055295635744,
      "grad_norm": 0.09014508128166199,
      "learning_rate": 8.643826008392872e-05,
      "loss": 1.6737,
      "step": 70496
    },
    {
      "epoch": 0.13917369891775386,
      "grad_norm": 0.11317227780818939,
      "learning_rate": 8.643191985956403e-05,
      "loss": 1.6905,
      "step": 70528
    },
    {
      "epoch": 0.1392368448791503,
      "grad_norm": 0.10088500380516052,
      "learning_rate": 8.642557963519935e-05,
      "loss": 1.6769,
      "step": 70560
    },
    {
      "epoch": 0.13929999084054676,
      "grad_norm": 0.1000090166926384,
      "learning_rate": 8.641923941083465e-05,
      "loss": 1.6932,
      "step": 70592
    },
    {
      "epoch": 0.13936313680194318,
      "grad_norm": 0.09973660111427307,
      "learning_rate": 8.641289918646996e-05,
      "loss": 1.6857,
      "step": 70624
    },
    {
      "epoch": 0.13942628276333963,
      "grad_norm": 0.10063603520393372,
      "learning_rate": 8.640655896210528e-05,
      "loss": 1.6998,
      "step": 70656
    },
    {
      "epoch": 0.13948942872473608,
      "grad_norm": 0.10236351191997528,
      "learning_rate": 8.64002187377406e-05,
      "loss": 1.6902,
      "step": 70688
    },
    {
      "epoch": 0.1395525746861325,
      "grad_norm": 0.10239796340465546,
      "learning_rate": 8.63938785133759e-05,
      "loss": 1.6699,
      "step": 70720
    },
    {
      "epoch": 0.13961572064752895,
      "grad_norm": 0.10240877419710159,
      "learning_rate": 8.638753828901121e-05,
      "loss": 1.7011,
      "step": 70752
    },
    {
      "epoch": 0.1396788666089254,
      "grad_norm": 0.09900277853012085,
      "learning_rate": 8.638119806464652e-05,
      "loss": 1.6929,
      "step": 70784
    },
    {
      "epoch": 0.13974201257032182,
      "grad_norm": 0.0964280515909195,
      "learning_rate": 8.637485784028182e-05,
      "loss": 1.6951,
      "step": 70816
    },
    {
      "epoch": 0.13980515853171827,
      "grad_norm": 0.09807398170232773,
      "learning_rate": 8.636851761591714e-05,
      "loss": 1.6998,
      "step": 70848
    },
    {
      "epoch": 0.13986830449311471,
      "grad_norm": 0.10209651291370392,
      "learning_rate": 8.636217739155244e-05,
      "loss": 1.6926,
      "step": 70880
    },
    {
      "epoch": 0.13993145045451114,
      "grad_norm": 0.10472094267606735,
      "learning_rate": 8.635583716718775e-05,
      "loss": 1.6878,
      "step": 70912
    },
    {
      "epoch": 0.13999459641590759,
      "grad_norm": 0.09496672451496124,
      "learning_rate": 8.634949694282307e-05,
      "loss": 1.6821,
      "step": 70944
    },
    {
      "epoch": 0.14005774237730403,
      "grad_norm": 0.09268078207969666,
      "learning_rate": 8.634315671845837e-05,
      "loss": 1.6884,
      "step": 70976
    },
    {
      "epoch": 0.14012088833870046,
      "grad_norm": 0.13251662254333496,
      "learning_rate": 8.633681649409368e-05,
      "loss": 1.6901,
      "step": 71008
    },
    {
      "epoch": 0.1401840343000969,
      "grad_norm": 0.10127288103103638,
      "learning_rate": 8.6330476269729e-05,
      "loss": 1.6996,
      "step": 71040
    },
    {
      "epoch": 0.14024718026149335,
      "grad_norm": 0.11469605565071106,
      "learning_rate": 8.632413604536431e-05,
      "loss": 1.697,
      "step": 71072
    },
    {
      "epoch": 0.14031032622288978,
      "grad_norm": 0.10494356602430344,
      "learning_rate": 8.631779582099963e-05,
      "loss": 1.6868,
      "step": 71104
    },
    {
      "epoch": 0.14037347218428622,
      "grad_norm": 0.10769851505756378,
      "learning_rate": 8.631145559663493e-05,
      "loss": 1.6928,
      "step": 71136
    },
    {
      "epoch": 0.14043661814568267,
      "grad_norm": 0.09526403248310089,
      "learning_rate": 8.630511537227024e-05,
      "loss": 1.704,
      "step": 71168
    },
    {
      "epoch": 0.1404997641070791,
      "grad_norm": 0.10641741752624512,
      "learning_rate": 8.629877514790556e-05,
      "loss": 1.6911,
      "step": 71200
    },
    {
      "epoch": 0.14056291006847554,
      "grad_norm": 0.10984072834253311,
      "learning_rate": 8.629243492354086e-05,
      "loss": 1.6918,
      "step": 71232
    },
    {
      "epoch": 0.140626056029872,
      "grad_norm": 0.09944106638431549,
      "learning_rate": 8.628609469917617e-05,
      "loss": 1.6914,
      "step": 71264
    },
    {
      "epoch": 0.14068920199126841,
      "grad_norm": 0.10753072053194046,
      "learning_rate": 8.627975447481148e-05,
      "loss": 1.6877,
      "step": 71296
    },
    {
      "epoch": 0.14075234795266486,
      "grad_norm": 0.09593670070171356,
      "learning_rate": 8.627341425044679e-05,
      "loss": 1.6755,
      "step": 71328
    },
    {
      "epoch": 0.1408154939140613,
      "grad_norm": 0.10216441750526428,
      "learning_rate": 8.62670740260821e-05,
      "loss": 1.6989,
      "step": 71360
    },
    {
      "epoch": 0.14087863987545773,
      "grad_norm": 0.09993880987167358,
      "learning_rate": 8.62607338017174e-05,
      "loss": 1.6991,
      "step": 71392
    },
    {
      "epoch": 0.14094178583685418,
      "grad_norm": 0.10341391712427139,
      "learning_rate": 8.625439357735272e-05,
      "loss": 1.6955,
      "step": 71424
    },
    {
      "epoch": 0.14100493179825063,
      "grad_norm": 0.10363147407770157,
      "learning_rate": 8.624805335298803e-05,
      "loss": 1.6899,
      "step": 71456
    },
    {
      "epoch": 0.14106807775964705,
      "grad_norm": 0.09979966282844543,
      "learning_rate": 8.624171312862335e-05,
      "loss": 1.6814,
      "step": 71488
    },
    {
      "epoch": 0.1411312237210435,
      "grad_norm": 0.10351796448230743,
      "learning_rate": 8.623537290425865e-05,
      "loss": 1.6875,
      "step": 71520
    },
    {
      "epoch": 0.14119436968243995,
      "grad_norm": 0.11131934076547623,
      "learning_rate": 8.622903267989396e-05,
      "loss": 1.695,
      "step": 71552
    },
    {
      "epoch": 0.14125751564383637,
      "grad_norm": 0.11410072445869446,
      "learning_rate": 8.622269245552928e-05,
      "loss": 1.681,
      "step": 71584
    },
    {
      "epoch": 0.14132066160523282,
      "grad_norm": 0.10665030032396317,
      "learning_rate": 8.62163522311646e-05,
      "loss": 1.6993,
      "step": 71616
    },
    {
      "epoch": 0.14138380756662927,
      "grad_norm": 0.10028386116027832,
      "learning_rate": 8.62100120067999e-05,
      "loss": 1.6771,
      "step": 71648
    },
    {
      "epoch": 0.14144695352802572,
      "grad_norm": 0.10181780159473419,
      "learning_rate": 8.62036717824352e-05,
      "loss": 1.7073,
      "step": 71680
    },
    {
      "epoch": 0.14151009948942214,
      "grad_norm": 0.09605645388364792,
      "learning_rate": 8.619733155807051e-05,
      "loss": 1.6922,
      "step": 71712
    },
    {
      "epoch": 0.1415732454508186,
      "grad_norm": 0.09963902831077576,
      "learning_rate": 8.619099133370583e-05,
      "loss": 1.6964,
      "step": 71744
    },
    {
      "epoch": 0.14163639141221504,
      "grad_norm": 0.10210099816322327,
      "learning_rate": 8.618465110934114e-05,
      "loss": 1.7024,
      "step": 71776
    },
    {
      "epoch": 0.14169953737361146,
      "grad_norm": 0.10228384286165237,
      "learning_rate": 8.617831088497644e-05,
      "loss": 1.6848,
      "step": 71808
    },
    {
      "epoch": 0.1417626833350079,
      "grad_norm": 0.11028777807950974,
      "learning_rate": 8.617197066061176e-05,
      "loss": 1.6844,
      "step": 71840
    },
    {
      "epoch": 0.14182582929640436,
      "grad_norm": 0.09842659533023834,
      "learning_rate": 8.616563043624707e-05,
      "loss": 1.6947,
      "step": 71872
    },
    {
      "epoch": 0.14188897525780078,
      "grad_norm": 0.09887228161096573,
      "learning_rate": 8.615929021188238e-05,
      "loss": 1.679,
      "step": 71904
    },
    {
      "epoch": 0.14195212121919723,
      "grad_norm": 0.10034212470054626,
      "learning_rate": 8.615294998751769e-05,
      "loss": 1.6803,
      "step": 71936
    },
    {
      "epoch": 0.14201526718059368,
      "grad_norm": 0.10476851463317871,
      "learning_rate": 8.6146609763153e-05,
      "loss": 1.6906,
      "step": 71968
    },
    {
      "epoch": 0.1420784131419901,
      "grad_norm": 0.10309062153100967,
      "learning_rate": 8.614026953878831e-05,
      "loss": 1.6927,
      "step": 72000
    },
    {
      "epoch": 0.14214155910338655,
      "grad_norm": 0.10585368424654007,
      "learning_rate": 8.613392931442363e-05,
      "loss": 1.7076,
      "step": 72032
    },
    {
      "epoch": 0.142204705064783,
      "grad_norm": 0.10662491619586945,
      "learning_rate": 8.612758909005893e-05,
      "loss": 1.6831,
      "step": 72064
    },
    {
      "epoch": 0.14226785102617942,
      "grad_norm": 0.10366757959127426,
      "learning_rate": 8.612124886569423e-05,
      "loss": 1.6861,
      "step": 72096
    },
    {
      "epoch": 0.14233099698757587,
      "grad_norm": 0.09764306992292404,
      "learning_rate": 8.611490864132955e-05,
      "loss": 1.6837,
      "step": 72128
    },
    {
      "epoch": 0.14239414294897232,
      "grad_norm": 0.10423771291971207,
      "learning_rate": 8.610856841696486e-05,
      "loss": 1.6838,
      "step": 72160
    },
    {
      "epoch": 0.14245728891036874,
      "grad_norm": 0.10319622606039047,
      "learning_rate": 8.610222819260016e-05,
      "loss": 1.6836,
      "step": 72192
    },
    {
      "epoch": 0.1425204348717652,
      "grad_norm": 0.09485956281423569,
      "learning_rate": 8.609588796823548e-05,
      "loss": 1.6806,
      "step": 72224
    },
    {
      "epoch": 0.14258358083316164,
      "grad_norm": 0.09834043681621552,
      "learning_rate": 8.608954774387079e-05,
      "loss": 1.687,
      "step": 72256
    },
    {
      "epoch": 0.14264672679455806,
      "grad_norm": 0.10211475193500519,
      "learning_rate": 8.60832075195061e-05,
      "loss": 1.6771,
      "step": 72288
    },
    {
      "epoch": 0.1427098727559545,
      "grad_norm": 0.10114586353302002,
      "learning_rate": 8.60768672951414e-05,
      "loss": 1.6837,
      "step": 72320
    },
    {
      "epoch": 0.14277301871735096,
      "grad_norm": 0.09665479511022568,
      "learning_rate": 8.607052707077672e-05,
      "loss": 1.67,
      "step": 72352
    },
    {
      "epoch": 0.14283616467874738,
      "grad_norm": 0.10244018584489822,
      "learning_rate": 8.606418684641203e-05,
      "loss": 1.6902,
      "step": 72384
    },
    {
      "epoch": 0.14289931064014383,
      "grad_norm": 0.10259303450584412,
      "learning_rate": 8.605784662204735e-05,
      "loss": 1.6757,
      "step": 72416
    },
    {
      "epoch": 0.14296245660154028,
      "grad_norm": 0.09692277014255524,
      "learning_rate": 8.605150639768266e-05,
      "loss": 1.6925,
      "step": 72448
    },
    {
      "epoch": 0.1430256025629367,
      "grad_norm": 0.10549569129943848,
      "learning_rate": 8.604516617331797e-05,
      "loss": 1.6673,
      "step": 72480
    },
    {
      "epoch": 0.14308874852433315,
      "grad_norm": 0.09814869612455368,
      "learning_rate": 8.603882594895327e-05,
      "loss": 1.6909,
      "step": 72512
    },
    {
      "epoch": 0.1431518944857296,
      "grad_norm": 0.10366464406251907,
      "learning_rate": 8.603248572458858e-05,
      "loss": 1.6755,
      "step": 72544
    },
    {
      "epoch": 0.14321504044712602,
      "grad_norm": 0.10869229584932327,
      "learning_rate": 8.60261455002239e-05,
      "loss": 1.696,
      "step": 72576
    },
    {
      "epoch": 0.14327818640852247,
      "grad_norm": 0.10145995020866394,
      "learning_rate": 8.60198052758592e-05,
      "loss": 1.6836,
      "step": 72608
    },
    {
      "epoch": 0.14334133236991892,
      "grad_norm": 0.10098624974489212,
      "learning_rate": 8.601346505149451e-05,
      "loss": 1.694,
      "step": 72640
    },
    {
      "epoch": 0.14340447833131534,
      "grad_norm": 0.09813161194324493,
      "learning_rate": 8.600712482712983e-05,
      "loss": 1.7053,
      "step": 72672
    },
    {
      "epoch": 0.1434676242927118,
      "grad_norm": 0.09813837707042694,
      "learning_rate": 8.600078460276514e-05,
      "loss": 1.6982,
      "step": 72704
    },
    {
      "epoch": 0.14353077025410824,
      "grad_norm": 0.1082853302359581,
      "learning_rate": 8.599444437840044e-05,
      "loss": 1.6991,
      "step": 72736
    },
    {
      "epoch": 0.14359391621550466,
      "grad_norm": 0.1084531843662262,
      "learning_rate": 8.598810415403576e-05,
      "loss": 1.69,
      "step": 72768
    },
    {
      "epoch": 0.1436570621769011,
      "grad_norm": 0.11164296418428421,
      "learning_rate": 8.598176392967107e-05,
      "loss": 1.6979,
      "step": 72800
    },
    {
      "epoch": 0.14372020813829756,
      "grad_norm": 0.1038421243429184,
      "learning_rate": 8.597542370530638e-05,
      "loss": 1.6716,
      "step": 72832
    },
    {
      "epoch": 0.14378335409969398,
      "grad_norm": 0.10914447903633118,
      "learning_rate": 8.596908348094169e-05,
      "loss": 1.67,
      "step": 72864
    },
    {
      "epoch": 0.14384650006109043,
      "grad_norm": 0.09301458299160004,
      "learning_rate": 8.5962743256577e-05,
      "loss": 1.6815,
      "step": 72896
    },
    {
      "epoch": 0.14390964602248688,
      "grad_norm": 0.10775292664766312,
      "learning_rate": 8.59564030322123e-05,
      "loss": 1.6871,
      "step": 72928
    },
    {
      "epoch": 0.1439727919838833,
      "grad_norm": 0.10177384316921234,
      "learning_rate": 8.595006280784762e-05,
      "loss": 1.6841,
      "step": 72960
    },
    {
      "epoch": 0.14403593794527975,
      "grad_norm": 0.09420222043991089,
      "learning_rate": 8.594372258348292e-05,
      "loss": 1.6732,
      "step": 72992
    },
    {
      "epoch": 0.1440990839066762,
      "grad_norm": 0.1003427729010582,
      "learning_rate": 8.593738235911823e-05,
      "loss": 1.6898,
      "step": 73024
    },
    {
      "epoch": 0.14416222986807262,
      "grad_norm": 0.09997972846031189,
      "learning_rate": 8.593104213475355e-05,
      "loss": 1.6964,
      "step": 73056
    },
    {
      "epoch": 0.14422537582946907,
      "grad_norm": 0.1105961725115776,
      "learning_rate": 8.592470191038886e-05,
      "loss": 1.6845,
      "step": 73088
    },
    {
      "epoch": 0.14428852179086551,
      "grad_norm": 0.09686896204948425,
      "learning_rate": 8.591836168602418e-05,
      "loss": 1.6908,
      "step": 73120
    },
    {
      "epoch": 0.14435166775226194,
      "grad_norm": 0.09825839102268219,
      "learning_rate": 8.591202146165948e-05,
      "loss": 1.6752,
      "step": 73152
    },
    {
      "epoch": 0.14441481371365839,
      "grad_norm": 0.09540785104036331,
      "learning_rate": 8.590568123729479e-05,
      "loss": 1.6784,
      "step": 73184
    },
    {
      "epoch": 0.14447795967505483,
      "grad_norm": 0.1011510118842125,
      "learning_rate": 8.58993410129301e-05,
      "loss": 1.6986,
      "step": 73216
    },
    {
      "epoch": 0.14454110563645126,
      "grad_norm": 0.09987585246562958,
      "learning_rate": 8.589300078856542e-05,
      "loss": 1.6775,
      "step": 73248
    },
    {
      "epoch": 0.1446042515978477,
      "grad_norm": 0.09723809361457825,
      "learning_rate": 8.588666056420072e-05,
      "loss": 1.6746,
      "step": 73280
    },
    {
      "epoch": 0.14466739755924415,
      "grad_norm": 0.09988243877887726,
      "learning_rate": 8.588032033983604e-05,
      "loss": 1.6779,
      "step": 73312
    },
    {
      "epoch": 0.1447305435206406,
      "grad_norm": 0.10036462545394897,
      "learning_rate": 8.587398011547134e-05,
      "loss": 1.6814,
      "step": 73344
    },
    {
      "epoch": 0.14479368948203702,
      "grad_norm": 0.10327211767435074,
      "learning_rate": 8.586763989110665e-05,
      "loss": 1.6775,
      "step": 73376
    },
    {
      "epoch": 0.14485683544343347,
      "grad_norm": 0.09883878380060196,
      "learning_rate": 8.586129966674195e-05,
      "loss": 1.6837,
      "step": 73408
    },
    {
      "epoch": 0.14491998140482992,
      "grad_norm": 0.11086346954107285,
      "learning_rate": 8.585495944237727e-05,
      "loss": 1.6891,
      "step": 73440
    },
    {
      "epoch": 0.14498312736622634,
      "grad_norm": 0.09583024680614471,
      "learning_rate": 8.584861921801258e-05,
      "loss": 1.6845,
      "step": 73472
    },
    {
      "epoch": 0.1450462733276228,
      "grad_norm": 0.0922607034444809,
      "learning_rate": 8.58422789936479e-05,
      "loss": 1.6743,
      "step": 73504
    },
    {
      "epoch": 0.14510941928901924,
      "grad_norm": 0.10028504580259323,
      "learning_rate": 8.58359387692832e-05,
      "loss": 1.6673,
      "step": 73536
    },
    {
      "epoch": 0.14517256525041566,
      "grad_norm": 0.09664369374513626,
      "learning_rate": 8.582959854491851e-05,
      "loss": 1.6664,
      "step": 73568
    },
    {
      "epoch": 0.1452357112118121,
      "grad_norm": 0.09518463164567947,
      "learning_rate": 8.582325832055383e-05,
      "loss": 1.6921,
      "step": 73600
    },
    {
      "epoch": 0.14529885717320856,
      "grad_norm": 0.10064047574996948,
      "learning_rate": 8.581691809618914e-05,
      "loss": 1.6965,
      "step": 73632
    },
    {
      "epoch": 0.14536200313460498,
      "grad_norm": 0.10300011932849884,
      "learning_rate": 8.581057787182444e-05,
      "loss": 1.6768,
      "step": 73664
    },
    {
      "epoch": 0.14542514909600143,
      "grad_norm": 0.10181638598442078,
      "learning_rate": 8.580423764745976e-05,
      "loss": 1.6763,
      "step": 73696
    },
    {
      "epoch": 0.14548829505739788,
      "grad_norm": 0.10829749703407288,
      "learning_rate": 8.579789742309507e-05,
      "loss": 1.6851,
      "step": 73728
    },
    {
      "epoch": 0.1455514410187943,
      "grad_norm": 0.09572859853506088,
      "learning_rate": 8.579155719873037e-05,
      "loss": 1.694,
      "step": 73760
    },
    {
      "epoch": 0.14561458698019075,
      "grad_norm": 0.10126513987779617,
      "learning_rate": 8.578521697436569e-05,
      "loss": 1.6806,
      "step": 73792
    },
    {
      "epoch": 0.1456777329415872,
      "grad_norm": 0.09824902564287186,
      "learning_rate": 8.577887675000099e-05,
      "loss": 1.6818,
      "step": 73824
    },
    {
      "epoch": 0.14574087890298362,
      "grad_norm": 0.10116986930370331,
      "learning_rate": 8.57725365256363e-05,
      "loss": 1.6851,
      "step": 73856
    },
    {
      "epoch": 0.14580402486438007,
      "grad_norm": 0.10895473510026932,
      "learning_rate": 8.576619630127162e-05,
      "loss": 1.6653,
      "step": 73888
    },
    {
      "epoch": 0.14586717082577652,
      "grad_norm": 0.0989847332239151,
      "learning_rate": 8.575985607690693e-05,
      "loss": 1.6867,
      "step": 73920
    },
    {
      "epoch": 0.14593031678717294,
      "grad_norm": 0.10052268207073212,
      "learning_rate": 8.575351585254223e-05,
      "loss": 1.6774,
      "step": 73952
    },
    {
      "epoch": 0.1459934627485694,
      "grad_norm": 0.10881651192903519,
      "learning_rate": 8.574717562817755e-05,
      "loss": 1.6932,
      "step": 73984
    },
    {
      "epoch": 0.14605660870996584,
      "grad_norm": 0.09766258299350739,
      "learning_rate": 8.574083540381286e-05,
      "loss": 1.6763,
      "step": 74016
    },
    {
      "epoch": 0.14611975467136226,
      "grad_norm": 0.10215530544519424,
      "learning_rate": 8.573449517944818e-05,
      "loss": 1.6939,
      "step": 74048
    },
    {
      "epoch": 0.1461829006327587,
      "grad_norm": 0.09876430034637451,
      "learning_rate": 8.572815495508348e-05,
      "loss": 1.6628,
      "step": 74080
    },
    {
      "epoch": 0.14624604659415516,
      "grad_norm": 0.10266532748937607,
      "learning_rate": 8.572181473071879e-05,
      "loss": 1.6937,
      "step": 74112
    },
    {
      "epoch": 0.14630919255555158,
      "grad_norm": 0.10835704952478409,
      "learning_rate": 8.571547450635409e-05,
      "loss": 1.6927,
      "step": 74144
    },
    {
      "epoch": 0.14637233851694803,
      "grad_norm": 0.09848356992006302,
      "learning_rate": 8.57091342819894e-05,
      "loss": 1.6849,
      "step": 74176
    },
    {
      "epoch": 0.14643548447834448,
      "grad_norm": 0.11198996752500534,
      "learning_rate": 8.570279405762471e-05,
      "loss": 1.6871,
      "step": 74208
    },
    {
      "epoch": 0.1464986304397409,
      "grad_norm": 0.0901528149843216,
      "learning_rate": 8.569645383326002e-05,
      "loss": 1.6763,
      "step": 74240
    },
    {
      "epoch": 0.14656177640113735,
      "grad_norm": 0.10375019907951355,
      "learning_rate": 8.569011360889534e-05,
      "loss": 1.6819,
      "step": 74272
    },
    {
      "epoch": 0.1466249223625338,
      "grad_norm": 0.09366860240697861,
      "learning_rate": 8.568377338453065e-05,
      "loss": 1.6828,
      "step": 74304
    },
    {
      "epoch": 0.14668806832393022,
      "grad_norm": 0.1028008833527565,
      "learning_rate": 8.567743316016595e-05,
      "loss": 1.6846,
      "step": 74336
    },
    {
      "epoch": 0.14675121428532667,
      "grad_norm": 0.09658585488796234,
      "learning_rate": 8.567109293580127e-05,
      "loss": 1.6836,
      "step": 74368
    },
    {
      "epoch": 0.14681436024672312,
      "grad_norm": 0.10165378451347351,
      "learning_rate": 8.566475271143658e-05,
      "loss": 1.6827,
      "step": 74400
    },
    {
      "epoch": 0.14687750620811954,
      "grad_norm": 0.09922339767217636,
      "learning_rate": 8.56584124870719e-05,
      "loss": 1.6709,
      "step": 74432
    },
    {
      "epoch": 0.146940652169516,
      "grad_norm": 0.09646771848201752,
      "learning_rate": 8.565207226270721e-05,
      "loss": 1.6728,
      "step": 74464
    },
    {
      "epoch": 0.14700379813091244,
      "grad_norm": 0.09826529026031494,
      "learning_rate": 8.564573203834251e-05,
      "loss": 1.6709,
      "step": 74496
    },
    {
      "epoch": 0.14706694409230886,
      "grad_norm": 0.09823821485042572,
      "learning_rate": 8.563939181397783e-05,
      "loss": 1.6891,
      "step": 74528
    },
    {
      "epoch": 0.1471300900537053,
      "grad_norm": 0.10077358037233353,
      "learning_rate": 8.563305158961313e-05,
      "loss": 1.6731,
      "step": 74560
    },
    {
      "epoch": 0.14719323601510176,
      "grad_norm": 0.09781774878501892,
      "learning_rate": 8.562671136524844e-05,
      "loss": 1.6838,
      "step": 74592
    },
    {
      "epoch": 0.14725638197649818,
      "grad_norm": 0.1077883169054985,
      "learning_rate": 8.562037114088374e-05,
      "loss": 1.6772,
      "step": 74624
    },
    {
      "epoch": 0.14731952793789463,
      "grad_norm": 0.10241179168224335,
      "learning_rate": 8.561403091651906e-05,
      "loss": 1.6659,
      "step": 74656
    },
    {
      "epoch": 0.14738267389929108,
      "grad_norm": 0.10631418973207474,
      "learning_rate": 8.560769069215437e-05,
      "loss": 1.6809,
      "step": 74688
    },
    {
      "epoch": 0.1474458198606875,
      "grad_norm": 0.09572099149227142,
      "learning_rate": 8.560135046778969e-05,
      "loss": 1.6817,
      "step": 74720
    },
    {
      "epoch": 0.14750896582208395,
      "grad_norm": 0.09712091088294983,
      "learning_rate": 8.559501024342499e-05,
      "loss": 1.6787,
      "step": 74752
    },
    {
      "epoch": 0.1475721117834804,
      "grad_norm": 0.10180314630270004,
      "learning_rate": 8.55886700190603e-05,
      "loss": 1.6854,
      "step": 74784
    },
    {
      "epoch": 0.14763525774487682,
      "grad_norm": 0.10517259687185287,
      "learning_rate": 8.558232979469562e-05,
      "loss": 1.6719,
      "step": 74816
    },
    {
      "epoch": 0.14769840370627327,
      "grad_norm": 0.10625191032886505,
      "learning_rate": 8.557598957033093e-05,
      "loss": 1.6805,
      "step": 74848
    },
    {
      "epoch": 0.14776154966766972,
      "grad_norm": 0.11206270009279251,
      "learning_rate": 8.556964934596623e-05,
      "loss": 1.6888,
      "step": 74880
    },
    {
      "epoch": 0.14782469562906614,
      "grad_norm": 0.09769652783870697,
      "learning_rate": 8.556330912160155e-05,
      "loss": 1.6808,
      "step": 74912
    },
    {
      "epoch": 0.1478878415904626,
      "grad_norm": 0.0974825918674469,
      "learning_rate": 8.555696889723686e-05,
      "loss": 1.6759,
      "step": 74944
    },
    {
      "epoch": 0.14795098755185904,
      "grad_norm": 0.11091966181993484,
      "learning_rate": 8.555062867287216e-05,
      "loss": 1.6925,
      "step": 74976
    },
    {
      "epoch": 0.14801413351325546,
      "grad_norm": 0.09856279194355011,
      "learning_rate": 8.554428844850748e-05,
      "loss": 1.6884,
      "step": 75008
    },
    {
      "epoch": 0.1480772794746519,
      "grad_norm": 0.10277500003576279,
      "learning_rate": 8.553794822414278e-05,
      "loss": 1.6786,
      "step": 75040
    },
    {
      "epoch": 0.14814042543604836,
      "grad_norm": 0.10396499931812286,
      "learning_rate": 8.553160799977809e-05,
      "loss": 1.6783,
      "step": 75072
    },
    {
      "epoch": 0.1482035713974448,
      "grad_norm": 0.10312742739915848,
      "learning_rate": 8.552526777541341e-05,
      "loss": 1.6952,
      "step": 75104
    },
    {
      "epoch": 0.14826671735884123,
      "grad_norm": 0.09461307525634766,
      "learning_rate": 8.551892755104872e-05,
      "loss": 1.6655,
      "step": 75136
    },
    {
      "epoch": 0.14832986332023768,
      "grad_norm": 0.09683814644813538,
      "learning_rate": 8.551258732668402e-05,
      "loss": 1.6874,
      "step": 75168
    },
    {
      "epoch": 0.14839300928163413,
      "grad_norm": 0.09330712258815765,
      "learning_rate": 8.550624710231934e-05,
      "loss": 1.69,
      "step": 75200
    },
    {
      "epoch": 0.14845615524303055,
      "grad_norm": 0.1111738532781601,
      "learning_rate": 8.549990687795465e-05,
      "loss": 1.6786,
      "step": 75232
    },
    {
      "epoch": 0.148519301204427,
      "grad_norm": 0.09655512124300003,
      "learning_rate": 8.549356665358997e-05,
      "loss": 1.6778,
      "step": 75264
    },
    {
      "epoch": 0.14858244716582344,
      "grad_norm": 0.11547809094190598,
      "learning_rate": 8.548722642922527e-05,
      "loss": 1.6858,
      "step": 75296
    },
    {
      "epoch": 0.14864559312721987,
      "grad_norm": 0.0961538702249527,
      "learning_rate": 8.548088620486058e-05,
      "loss": 1.6843,
      "step": 75328
    },
    {
      "epoch": 0.14870873908861632,
      "grad_norm": 0.09542649984359741,
      "learning_rate": 8.54745459804959e-05,
      "loss": 1.685,
      "step": 75360
    },
    {
      "epoch": 0.14877188505001276,
      "grad_norm": 0.10553640127182007,
      "learning_rate": 8.54682057561312e-05,
      "loss": 1.6773,
      "step": 75392
    },
    {
      "epoch": 0.14883503101140919,
      "grad_norm": 0.09616170078516006,
      "learning_rate": 8.546186553176651e-05,
      "loss": 1.6752,
      "step": 75424
    },
    {
      "epoch": 0.14889817697280563,
      "grad_norm": 0.09604093432426453,
      "learning_rate": 8.545552530740181e-05,
      "loss": 1.683,
      "step": 75456
    },
    {
      "epoch": 0.14896132293420208,
      "grad_norm": 0.09946244955062866,
      "learning_rate": 8.544918508303713e-05,
      "loss": 1.6827,
      "step": 75488
    },
    {
      "epoch": 0.1490244688955985,
      "grad_norm": 0.1011558547616005,
      "learning_rate": 8.544284485867244e-05,
      "loss": 1.6661,
      "step": 75520
    },
    {
      "epoch": 0.14908761485699495,
      "grad_norm": 0.09891995042562485,
      "learning_rate": 8.543650463430774e-05,
      "loss": 1.6751,
      "step": 75552
    },
    {
      "epoch": 0.1491507608183914,
      "grad_norm": 0.10548239946365356,
      "learning_rate": 8.543016440994306e-05,
      "loss": 1.6824,
      "step": 75584
    },
    {
      "epoch": 0.14921390677978782,
      "grad_norm": 0.10210949182510376,
      "learning_rate": 8.542382418557837e-05,
      "loss": 1.6775,
      "step": 75616
    },
    {
      "epoch": 0.14927705274118427,
      "grad_norm": 0.0944737046957016,
      "learning_rate": 8.541748396121369e-05,
      "loss": 1.6696,
      "step": 75648
    },
    {
      "epoch": 0.14934019870258072,
      "grad_norm": 0.10408736765384674,
      "learning_rate": 8.541114373684899e-05,
      "loss": 1.6815,
      "step": 75680
    },
    {
      "epoch": 0.14940334466397714,
      "grad_norm": 0.10370989888906479,
      "learning_rate": 8.54048035124843e-05,
      "loss": 1.6828,
      "step": 75712
    },
    {
      "epoch": 0.1494664906253736,
      "grad_norm": 0.11968197673559189,
      "learning_rate": 8.539846328811962e-05,
      "loss": 1.6826,
      "step": 75744
    },
    {
      "epoch": 0.14952963658677004,
      "grad_norm": 0.10264799743890762,
      "learning_rate": 8.539212306375493e-05,
      "loss": 1.6844,
      "step": 75776
    },
    {
      "epoch": 0.14959278254816646,
      "grad_norm": 0.11237389594316483,
      "learning_rate": 8.538578283939023e-05,
      "loss": 1.7027,
      "step": 75808
    },
    {
      "epoch": 0.1496559285095629,
      "grad_norm": 0.10504571348428726,
      "learning_rate": 8.537944261502553e-05,
      "loss": 1.6617,
      "step": 75840
    },
    {
      "epoch": 0.14971907447095936,
      "grad_norm": 0.11485075205564499,
      "learning_rate": 8.537310239066085e-05,
      "loss": 1.6814,
      "step": 75872
    },
    {
      "epoch": 0.14978222043235578,
      "grad_norm": 0.09677886962890625,
      "learning_rate": 8.536676216629616e-05,
      "loss": 1.6867,
      "step": 75904
    },
    {
      "epoch": 0.14984536639375223,
      "grad_norm": 0.0998137816786766,
      "learning_rate": 8.536042194193148e-05,
      "loss": 1.6789,
      "step": 75936
    },
    {
      "epoch": 0.14990851235514868,
      "grad_norm": 0.10647508502006531,
      "learning_rate": 8.535408171756678e-05,
      "loss": 1.668,
      "step": 75968
    },
    {
      "epoch": 0.1499716583165451,
      "grad_norm": 0.09866531938314438,
      "learning_rate": 8.534774149320209e-05,
      "loss": 1.675,
      "step": 76000
    },
    {
      "epoch": 0.15003480427794155,
      "grad_norm": 0.10397178679704666,
      "learning_rate": 8.534140126883741e-05,
      "loss": 1.6773,
      "step": 76032
    },
    {
      "epoch": 0.150097950239338,
      "grad_norm": 0.10752294212579727,
      "learning_rate": 8.533506104447272e-05,
      "loss": 1.676,
      "step": 76064
    },
    {
      "epoch": 0.15016109620073442,
      "grad_norm": 0.10332654416561127,
      "learning_rate": 8.532872082010802e-05,
      "loss": 1.6925,
      "step": 76096
    },
    {
      "epoch": 0.15022424216213087,
      "grad_norm": 0.10872095078229904,
      "learning_rate": 8.532238059574334e-05,
      "loss": 1.6749,
      "step": 76128
    },
    {
      "epoch": 0.15028738812352732,
      "grad_norm": 0.1017443910241127,
      "learning_rate": 8.531604037137865e-05,
      "loss": 1.6711,
      "step": 76160
    },
    {
      "epoch": 0.15035053408492374,
      "grad_norm": 0.09586425870656967,
      "learning_rate": 8.530970014701397e-05,
      "loss": 1.6746,
      "step": 76192
    },
    {
      "epoch": 0.1504136800463202,
      "grad_norm": 0.10036095231771469,
      "learning_rate": 8.530335992264927e-05,
      "loss": 1.6751,
      "step": 76224
    },
    {
      "epoch": 0.15047682600771664,
      "grad_norm": 0.09590496122837067,
      "learning_rate": 8.529701969828457e-05,
      "loss": 1.6907,
      "step": 76256
    },
    {
      "epoch": 0.15053997196911306,
      "grad_norm": 0.10554616898298264,
      "learning_rate": 8.529067947391988e-05,
      "loss": 1.6753,
      "step": 76288
    },
    {
      "epoch": 0.1506031179305095,
      "grad_norm": 0.1041039377450943,
      "learning_rate": 8.52843392495552e-05,
      "loss": 1.6745,
      "step": 76320
    },
    {
      "epoch": 0.15066626389190596,
      "grad_norm": 0.10178229212760925,
      "learning_rate": 8.52779990251905e-05,
      "loss": 1.6791,
      "step": 76352
    },
    {
      "epoch": 0.15072940985330238,
      "grad_norm": 0.09858375042676926,
      "learning_rate": 8.527165880082581e-05,
      "loss": 1.681,
      "step": 76384
    },
    {
      "epoch": 0.15079255581469883,
      "grad_norm": 0.10965598374605179,
      "learning_rate": 8.526531857646113e-05,
      "loss": 1.6801,
      "step": 76416
    },
    {
      "epoch": 0.15085570177609528,
      "grad_norm": 0.11225444823503494,
      "learning_rate": 8.525897835209644e-05,
      "loss": 1.6729,
      "step": 76448
    },
    {
      "epoch": 0.1509188477374917,
      "grad_norm": 0.10393206775188446,
      "learning_rate": 8.525263812773176e-05,
      "loss": 1.6569,
      "step": 76480
    },
    {
      "epoch": 0.15098199369888815,
      "grad_norm": 0.10081935673952103,
      "learning_rate": 8.524629790336706e-05,
      "loss": 1.6748,
      "step": 76512
    },
    {
      "epoch": 0.1510451396602846,
      "grad_norm": 0.10383206605911255,
      "learning_rate": 8.523995767900237e-05,
      "loss": 1.6951,
      "step": 76544
    },
    {
      "epoch": 0.15110828562168102,
      "grad_norm": 0.09918275475502014,
      "learning_rate": 8.523361745463769e-05,
      "loss": 1.6797,
      "step": 76576
    },
    {
      "epoch": 0.15117143158307747,
      "grad_norm": 0.10046989470720291,
      "learning_rate": 8.5227277230273e-05,
      "loss": 1.6592,
      "step": 76608
    },
    {
      "epoch": 0.15123457754447392,
      "grad_norm": 0.1048884242773056,
      "learning_rate": 8.52209370059083e-05,
      "loss": 1.6787,
      "step": 76640
    },
    {
      "epoch": 0.15129772350587034,
      "grad_norm": 0.10680338740348816,
      "learning_rate": 8.52145967815436e-05,
      "loss": 1.6862,
      "step": 76672
    },
    {
      "epoch": 0.1513608694672668,
      "grad_norm": 0.09893324971199036,
      "learning_rate": 8.520825655717892e-05,
      "loss": 1.6653,
      "step": 76704
    },
    {
      "epoch": 0.15142401542866324,
      "grad_norm": 0.10264407098293304,
      "learning_rate": 8.520191633281423e-05,
      "loss": 1.7012,
      "step": 76736
    },
    {
      "epoch": 0.1514871613900597,
      "grad_norm": 0.1042848452925682,
      "learning_rate": 8.519557610844953e-05,
      "loss": 1.6781,
      "step": 76768
    },
    {
      "epoch": 0.1515503073514561,
      "grad_norm": 0.10645639896392822,
      "learning_rate": 8.518923588408485e-05,
      "loss": 1.6829,
      "step": 76800
    },
    {
      "epoch": 0.15161345331285256,
      "grad_norm": 0.10583385825157166,
      "learning_rate": 8.518289565972016e-05,
      "loss": 1.6711,
      "step": 76832
    },
    {
      "epoch": 0.151676599274249,
      "grad_norm": 0.10256838798522949,
      "learning_rate": 8.517655543535548e-05,
      "loss": 1.6751,
      "step": 76864
    },
    {
      "epoch": 0.15173974523564543,
      "grad_norm": 0.09623630344867706,
      "learning_rate": 8.517021521099078e-05,
      "loss": 1.6767,
      "step": 76896
    },
    {
      "epoch": 0.15180289119704188,
      "grad_norm": 0.09691829979419708,
      "learning_rate": 8.516387498662609e-05,
      "loss": 1.6617,
      "step": 76928
    },
    {
      "epoch": 0.15186603715843833,
      "grad_norm": 0.09584102779626846,
      "learning_rate": 8.515753476226141e-05,
      "loss": 1.6815,
      "step": 76960
    },
    {
      "epoch": 0.15192918311983475,
      "grad_norm": 0.10497911274433136,
      "learning_rate": 8.515119453789672e-05,
      "loss": 1.6784,
      "step": 76992
    },
    {
      "epoch": 0.1519923290812312,
      "grad_norm": 0.10860483348369598,
      "learning_rate": 8.514485431353202e-05,
      "loss": 1.675,
      "step": 77024
    },
    {
      "epoch": 0.15205547504262765,
      "grad_norm": 0.10359463095664978,
      "learning_rate": 8.513851408916734e-05,
      "loss": 1.6742,
      "step": 77056
    },
    {
      "epoch": 0.15211862100402407,
      "grad_norm": 0.1057921051979065,
      "learning_rate": 8.513217386480264e-05,
      "loss": 1.685,
      "step": 77088
    },
    {
      "epoch": 0.15218176696542052,
      "grad_norm": 0.09901981055736542,
      "learning_rate": 8.512583364043795e-05,
      "loss": 1.6793,
      "step": 77120
    },
    {
      "epoch": 0.15224491292681697,
      "grad_norm": 0.09833051264286041,
      "learning_rate": 8.511949341607327e-05,
      "loss": 1.6694,
      "step": 77152
    },
    {
      "epoch": 0.1523080588882134,
      "grad_norm": 0.099812813103199,
      "learning_rate": 8.511315319170857e-05,
      "loss": 1.6693,
      "step": 77184
    },
    {
      "epoch": 0.15237120484960984,
      "grad_norm": 0.09430838376283646,
      "learning_rate": 8.510681296734388e-05,
      "loss": 1.6628,
      "step": 77216
    },
    {
      "epoch": 0.1524343508110063,
      "grad_norm": 0.10515926033258438,
      "learning_rate": 8.51004727429792e-05,
      "loss": 1.6719,
      "step": 77248
    },
    {
      "epoch": 0.1524974967724027,
      "grad_norm": 0.10583945363759995,
      "learning_rate": 8.509413251861451e-05,
      "loss": 1.6751,
      "step": 77280
    },
    {
      "epoch": 0.15256064273379916,
      "grad_norm": 0.12096399068832397,
      "learning_rate": 8.508779229424981e-05,
      "loss": 1.6875,
      "step": 77312
    },
    {
      "epoch": 0.1526237886951956,
      "grad_norm": 0.10812398046255112,
      "learning_rate": 8.508145206988513e-05,
      "loss": 1.6746,
      "step": 77344
    },
    {
      "epoch": 0.15268693465659203,
      "grad_norm": 0.10630808025598526,
      "learning_rate": 8.507511184552044e-05,
      "loss": 1.6723,
      "step": 77376
    },
    {
      "epoch": 0.15275008061798848,
      "grad_norm": 0.1007491871714592,
      "learning_rate": 8.506877162115576e-05,
      "loss": 1.6876,
      "step": 77408
    },
    {
      "epoch": 0.15281322657938493,
      "grad_norm": 0.09950321167707443,
      "learning_rate": 8.506243139679106e-05,
      "loss": 1.6703,
      "step": 77440
    },
    {
      "epoch": 0.15287637254078135,
      "grad_norm": 0.11235208809375763,
      "learning_rate": 8.505609117242637e-05,
      "loss": 1.6802,
      "step": 77472
    },
    {
      "epoch": 0.1529395185021778,
      "grad_norm": 0.10004584491252899,
      "learning_rate": 8.504975094806167e-05,
      "loss": 1.6815,
      "step": 77504
    },
    {
      "epoch": 0.15300266446357424,
      "grad_norm": 0.12749536335468292,
      "learning_rate": 8.504341072369699e-05,
      "loss": 1.6744,
      "step": 77536
    },
    {
      "epoch": 0.15306581042497067,
      "grad_norm": 0.0963485911488533,
      "learning_rate": 8.503707049933229e-05,
      "loss": 1.661,
      "step": 77568
    },
    {
      "epoch": 0.15312895638636712,
      "grad_norm": 0.11256939172744751,
      "learning_rate": 8.50307302749676e-05,
      "loss": 1.6712,
      "step": 77600
    },
    {
      "epoch": 0.15319210234776356,
      "grad_norm": 0.10460944473743439,
      "learning_rate": 8.502439005060292e-05,
      "loss": 1.6841,
      "step": 77632
    },
    {
      "epoch": 0.15325524830915999,
      "grad_norm": 0.09804650396108627,
      "learning_rate": 8.501804982623823e-05,
      "loss": 1.6739,
      "step": 77664
    },
    {
      "epoch": 0.15331839427055644,
      "grad_norm": 0.09925612807273865,
      "learning_rate": 8.501170960187353e-05,
      "loss": 1.6678,
      "step": 77696
    },
    {
      "epoch": 0.15338154023195288,
      "grad_norm": 0.0983809307217598,
      "learning_rate": 8.500536937750885e-05,
      "loss": 1.675,
      "step": 77728
    },
    {
      "epoch": 0.1534446861933493,
      "grad_norm": 0.09494610130786896,
      "learning_rate": 8.499902915314416e-05,
      "loss": 1.6699,
      "step": 77760
    },
    {
      "epoch": 0.15350783215474575,
      "grad_norm": 0.10970768332481384,
      "learning_rate": 8.499268892877948e-05,
      "loss": 1.6911,
      "step": 77792
    },
    {
      "epoch": 0.1535709781161422,
      "grad_norm": 0.09740414470434189,
      "learning_rate": 8.498634870441479e-05,
      "loss": 1.6793,
      "step": 77824
    },
    {
      "epoch": 0.15363412407753863,
      "grad_norm": 0.10191775858402252,
      "learning_rate": 8.49800084800501e-05,
      "loss": 1.6697,
      "step": 77856
    },
    {
      "epoch": 0.15369727003893507,
      "grad_norm": 0.10068080574274063,
      "learning_rate": 8.497366825568541e-05,
      "loss": 1.6671,
      "step": 77888
    },
    {
      "epoch": 0.15376041600033152,
      "grad_norm": 0.10468345135450363,
      "learning_rate": 8.496732803132071e-05,
      "loss": 1.6843,
      "step": 77920
    },
    {
      "epoch": 0.15382356196172794,
      "grad_norm": 0.10640867799520493,
      "learning_rate": 8.496098780695602e-05,
      "loss": 1.6702,
      "step": 77952
    },
    {
      "epoch": 0.1538867079231244,
      "grad_norm": 0.09602980315685272,
      "learning_rate": 8.495464758259132e-05,
      "loss": 1.6641,
      "step": 77984
    },
    {
      "epoch": 0.15394985388452084,
      "grad_norm": 0.09832219779491425,
      "learning_rate": 8.494830735822664e-05,
      "loss": 1.6727,
      "step": 78016
    },
    {
      "epoch": 0.15401299984591726,
      "grad_norm": 0.09909886121749878,
      "learning_rate": 8.494196713386195e-05,
      "loss": 1.65,
      "step": 78048
    },
    {
      "epoch": 0.1540761458073137,
      "grad_norm": 0.10727988183498383,
      "learning_rate": 8.493562690949727e-05,
      "loss": 1.6878,
      "step": 78080
    },
    {
      "epoch": 0.15413929176871016,
      "grad_norm": 0.09718786180019379,
      "learning_rate": 8.492928668513257e-05,
      "loss": 1.686,
      "step": 78112
    },
    {
      "epoch": 0.15420243773010658,
      "grad_norm": 0.10660644620656967,
      "learning_rate": 8.492294646076788e-05,
      "loss": 1.6685,
      "step": 78144
    },
    {
      "epoch": 0.15426558369150303,
      "grad_norm": 0.09951602667570114,
      "learning_rate": 8.49166062364032e-05,
      "loss": 1.6774,
      "step": 78176
    },
    {
      "epoch": 0.15432872965289948,
      "grad_norm": 0.10155228525400162,
      "learning_rate": 8.491026601203851e-05,
      "loss": 1.6757,
      "step": 78208
    },
    {
      "epoch": 0.1543918756142959,
      "grad_norm": 0.099421925842762,
      "learning_rate": 8.490392578767381e-05,
      "loss": 1.673,
      "step": 78240
    },
    {
      "epoch": 0.15445502157569235,
      "grad_norm": 0.10386461764574051,
      "learning_rate": 8.489758556330913e-05,
      "loss": 1.6877,
      "step": 78272
    },
    {
      "epoch": 0.1545181675370888,
      "grad_norm": 0.09559035301208496,
      "learning_rate": 8.489124533894444e-05,
      "loss": 1.6781,
      "step": 78304
    },
    {
      "epoch": 0.15458131349848522,
      "grad_norm": 0.10044359415769577,
      "learning_rate": 8.488490511457974e-05,
      "loss": 1.6683,
      "step": 78336
    },
    {
      "epoch": 0.15464445945988167,
      "grad_norm": 0.09693678468465805,
      "learning_rate": 8.487856489021506e-05,
      "loss": 1.6837,
      "step": 78368
    },
    {
      "epoch": 0.15470760542127812,
      "grad_norm": 0.10030453652143478,
      "learning_rate": 8.487222466585036e-05,
      "loss": 1.6698,
      "step": 78400
    },
    {
      "epoch": 0.15477075138267457,
      "grad_norm": 0.10268690437078476,
      "learning_rate": 8.486588444148567e-05,
      "loss": 1.6572,
      "step": 78432
    },
    {
      "epoch": 0.154833897344071,
      "grad_norm": 0.09646078199148178,
      "learning_rate": 8.485954421712099e-05,
      "loss": 1.682,
      "step": 78464
    },
    {
      "epoch": 0.15489704330546744,
      "grad_norm": 0.10658205300569534,
      "learning_rate": 8.48532039927563e-05,
      "loss": 1.673,
      "step": 78496
    },
    {
      "epoch": 0.1549601892668639,
      "grad_norm": 0.09471545368432999,
      "learning_rate": 8.48468637683916e-05,
      "loss": 1.6803,
      "step": 78528
    },
    {
      "epoch": 0.1550233352282603,
      "grad_norm": 0.10045652091503143,
      "learning_rate": 8.484052354402692e-05,
      "loss": 1.6719,
      "step": 78560
    },
    {
      "epoch": 0.15508648118965676,
      "grad_norm": 0.09457363188266754,
      "learning_rate": 8.483418331966223e-05,
      "loss": 1.6612,
      "step": 78592
    },
    {
      "epoch": 0.1551496271510532,
      "grad_norm": 0.09976271539926529,
      "learning_rate": 8.482784309529755e-05,
      "loss": 1.6803,
      "step": 78624
    },
    {
      "epoch": 0.15521277311244963,
      "grad_norm": 0.09982792288064957,
      "learning_rate": 8.482150287093285e-05,
      "loss": 1.6812,
      "step": 78656
    },
    {
      "epoch": 0.15527591907384608,
      "grad_norm": 0.10151281207799911,
      "learning_rate": 8.481516264656816e-05,
      "loss": 1.6768,
      "step": 78688
    },
    {
      "epoch": 0.15533906503524253,
      "grad_norm": 0.10529918968677521,
      "learning_rate": 8.480882242220346e-05,
      "loss": 1.6673,
      "step": 78720
    },
    {
      "epoch": 0.15540221099663895,
      "grad_norm": 0.09730803221464157,
      "learning_rate": 8.480248219783878e-05,
      "loss": 1.6841,
      "step": 78752
    },
    {
      "epoch": 0.1554653569580354,
      "grad_norm": 0.10061978548765182,
      "learning_rate": 8.479614197347408e-05,
      "loss": 1.6829,
      "step": 78784
    },
    {
      "epoch": 0.15552850291943185,
      "grad_norm": 0.1007780134677887,
      "learning_rate": 8.47898017491094e-05,
      "loss": 1.6852,
      "step": 78816
    },
    {
      "epoch": 0.15559164888082827,
      "grad_norm": 0.10080109536647797,
      "learning_rate": 8.478346152474471e-05,
      "loss": 1.6831,
      "step": 78848
    },
    {
      "epoch": 0.15565479484222472,
      "grad_norm": 0.1095278188586235,
      "learning_rate": 8.477712130038002e-05,
      "loss": 1.6578,
      "step": 78880
    },
    {
      "epoch": 0.15571794080362117,
      "grad_norm": 0.11861991137266159,
      "learning_rate": 8.477078107601532e-05,
      "loss": 1.6692,
      "step": 78912
    },
    {
      "epoch": 0.1557810867650176,
      "grad_norm": 0.10310599207878113,
      "learning_rate": 8.476444085165064e-05,
      "loss": 1.6588,
      "step": 78944
    },
    {
      "epoch": 0.15584423272641404,
      "grad_norm": 0.11733296513557434,
      "learning_rate": 8.475810062728595e-05,
      "loss": 1.6691,
      "step": 78976
    },
    {
      "epoch": 0.1559073786878105,
      "grad_norm": 0.10434328764677048,
      "learning_rate": 8.475176040292127e-05,
      "loss": 1.6661,
      "step": 79008
    },
    {
      "epoch": 0.1559705246492069,
      "grad_norm": 0.10890328139066696,
      "learning_rate": 8.474542017855658e-05,
      "loss": 1.6771,
      "step": 79040
    },
    {
      "epoch": 0.15603367061060336,
      "grad_norm": 0.09577399492263794,
      "learning_rate": 8.473907995419188e-05,
      "loss": 1.6596,
      "step": 79072
    },
    {
      "epoch": 0.1560968165719998,
      "grad_norm": 0.10378000140190125,
      "learning_rate": 8.47327397298272e-05,
      "loss": 1.6549,
      "step": 79104
    },
    {
      "epoch": 0.15615996253339623,
      "grad_norm": 0.1202513799071312,
      "learning_rate": 8.47263995054625e-05,
      "loss": 1.669,
      "step": 79136
    },
    {
      "epoch": 0.15622310849479268,
      "grad_norm": 0.10733233392238617,
      "learning_rate": 8.472005928109781e-05,
      "loss": 1.6779,
      "step": 79168
    },
    {
      "epoch": 0.15628625445618913,
      "grad_norm": 0.10416159778833389,
      "learning_rate": 8.471371905673312e-05,
      "loss": 1.6851,
      "step": 79200
    },
    {
      "epoch": 0.15634940041758555,
      "grad_norm": 0.09278441220521927,
      "learning_rate": 8.470737883236843e-05,
      "loss": 1.6622,
      "step": 79232
    },
    {
      "epoch": 0.156412546378982,
      "grad_norm": 0.10320501774549484,
      "learning_rate": 8.470103860800374e-05,
      "loss": 1.6763,
      "step": 79264
    },
    {
      "epoch": 0.15647569234037845,
      "grad_norm": 0.11157459765672684,
      "learning_rate": 8.469469838363906e-05,
      "loss": 1.6868,
      "step": 79296
    },
    {
      "epoch": 0.15653883830177487,
      "grad_norm": 0.09679748862981796,
      "learning_rate": 8.468835815927436e-05,
      "loss": 1.6767,
      "step": 79328
    },
    {
      "epoch": 0.15660198426317132,
      "grad_norm": 0.10498563945293427,
      "learning_rate": 8.468201793490967e-05,
      "loss": 1.6637,
      "step": 79360
    },
    {
      "epoch": 0.15666513022456777,
      "grad_norm": 0.0998353511095047,
      "learning_rate": 8.467567771054499e-05,
      "loss": 1.6701,
      "step": 79392
    },
    {
      "epoch": 0.1567282761859642,
      "grad_norm": 0.10383282601833344,
      "learning_rate": 8.46693374861803e-05,
      "loss": 1.6921,
      "step": 79424
    },
    {
      "epoch": 0.15679142214736064,
      "grad_norm": 0.09842722117900848,
      "learning_rate": 8.46629972618156e-05,
      "loss": 1.6815,
      "step": 79456
    },
    {
      "epoch": 0.1568545681087571,
      "grad_norm": 0.11832182109355927,
      "learning_rate": 8.465665703745092e-05,
      "loss": 1.6766,
      "step": 79488
    },
    {
      "epoch": 0.1569177140701535,
      "grad_norm": 0.0950767919421196,
      "learning_rate": 8.465031681308623e-05,
      "loss": 1.6702,
      "step": 79520
    },
    {
      "epoch": 0.15698086003154996,
      "grad_norm": 0.09939881414175034,
      "learning_rate": 8.464397658872153e-05,
      "loss": 1.6637,
      "step": 79552
    },
    {
      "epoch": 0.1570440059929464,
      "grad_norm": 0.09540250152349472,
      "learning_rate": 8.463763636435685e-05,
      "loss": 1.6622,
      "step": 79584
    },
    {
      "epoch": 0.15710715195434283,
      "grad_norm": 0.11049431562423706,
      "learning_rate": 8.463129613999215e-05,
      "loss": 1.6683,
      "step": 79616
    },
    {
      "epoch": 0.15717029791573928,
      "grad_norm": 0.10123123973608017,
      "learning_rate": 8.462495591562746e-05,
      "loss": 1.6642,
      "step": 79648
    },
    {
      "epoch": 0.15723344387713573,
      "grad_norm": 0.09708897024393082,
      "learning_rate": 8.461861569126278e-05,
      "loss": 1.6746,
      "step": 79680
    },
    {
      "epoch": 0.15729658983853215,
      "grad_norm": 0.1010143831372261,
      "learning_rate": 8.46122754668981e-05,
      "loss": 1.6707,
      "step": 79712
    },
    {
      "epoch": 0.1573597357999286,
      "grad_norm": 0.10873035341501236,
      "learning_rate": 8.46059352425334e-05,
      "loss": 1.6878,
      "step": 79744
    },
    {
      "epoch": 0.15742288176132505,
      "grad_norm": 0.11764772236347198,
      "learning_rate": 8.459959501816871e-05,
      "loss": 1.6702,
      "step": 79776
    },
    {
      "epoch": 0.15748602772272147,
      "grad_norm": 0.10337798297405243,
      "learning_rate": 8.459325479380402e-05,
      "loss": 1.6721,
      "step": 79808
    },
    {
      "epoch": 0.15754917368411792,
      "grad_norm": 0.09046148508787155,
      "learning_rate": 8.458691456943934e-05,
      "loss": 1.6639,
      "step": 79840
    },
    {
      "epoch": 0.15761231964551436,
      "grad_norm": 0.10244788229465485,
      "learning_rate": 8.458057434507464e-05,
      "loss": 1.6601,
      "step": 79872
    },
    {
      "epoch": 0.1576754656069108,
      "grad_norm": 0.09827053546905518,
      "learning_rate": 8.457423412070995e-05,
      "loss": 1.6807,
      "step": 79904
    },
    {
      "epoch": 0.15773861156830724,
      "grad_norm": 0.10390309244394302,
      "learning_rate": 8.456789389634527e-05,
      "loss": 1.6795,
      "step": 79936
    },
    {
      "epoch": 0.15780175752970368,
      "grad_norm": 0.10280296206474304,
      "learning_rate": 8.456155367198057e-05,
      "loss": 1.6539,
      "step": 79968
    },
    {
      "epoch": 0.1578649034911001,
      "grad_norm": 0.09721670299768448,
      "learning_rate": 8.455521344761587e-05,
      "loss": 1.6642,
      "step": 80000
    },
    {
      "epoch": 0.15792804945249656,
      "grad_norm": 0.09536055475473404,
      "learning_rate": 8.454887322325119e-05,
      "loss": 1.6592,
      "step": 80032
    },
    {
      "epoch": 0.157991195413893,
      "grad_norm": 0.09576831758022308,
      "learning_rate": 8.45425329988865e-05,
      "loss": 1.6741,
      "step": 80064
    },
    {
      "epoch": 0.15805434137528945,
      "grad_norm": 0.10265585035085678,
      "learning_rate": 8.453619277452181e-05,
      "loss": 1.6604,
      "step": 80096
    },
    {
      "epoch": 0.15811748733668587,
      "grad_norm": 0.09348326921463013,
      "learning_rate": 8.452985255015712e-05,
      "loss": 1.6843,
      "step": 80128
    },
    {
      "epoch": 0.15818063329808232,
      "grad_norm": 0.09281941503286362,
      "learning_rate": 8.452351232579243e-05,
      "loss": 1.6637,
      "step": 80160
    },
    {
      "epoch": 0.15824377925947877,
      "grad_norm": 0.09963396936655045,
      "learning_rate": 8.451717210142774e-05,
      "loss": 1.6616,
      "step": 80192
    },
    {
      "epoch": 0.1583069252208752,
      "grad_norm": 0.10201126337051392,
      "learning_rate": 8.451083187706306e-05,
      "loss": 1.6703,
      "step": 80224
    },
    {
      "epoch": 0.15837007118227164,
      "grad_norm": 0.09980650991201401,
      "learning_rate": 8.450449165269836e-05,
      "loss": 1.6725,
      "step": 80256
    },
    {
      "epoch": 0.1584332171436681,
      "grad_norm": 0.11375914514064789,
      "learning_rate": 8.449815142833367e-05,
      "loss": 1.6625,
      "step": 80288
    },
    {
      "epoch": 0.1584963631050645,
      "grad_norm": 0.11290016770362854,
      "learning_rate": 8.449181120396899e-05,
      "loss": 1.6745,
      "step": 80320
    },
    {
      "epoch": 0.15855950906646096,
      "grad_norm": 0.0976036787033081,
      "learning_rate": 8.44854709796043e-05,
      "loss": 1.673,
      "step": 80352
    },
    {
      "epoch": 0.1586226550278574,
      "grad_norm": 0.10798491537570953,
      "learning_rate": 8.44791307552396e-05,
      "loss": 1.6888,
      "step": 80384
    },
    {
      "epoch": 0.15868580098925383,
      "grad_norm": 0.0982152670621872,
      "learning_rate": 8.44727905308749e-05,
      "loss": 1.6695,
      "step": 80416
    },
    {
      "epoch": 0.15874894695065028,
      "grad_norm": 0.1046973317861557,
      "learning_rate": 8.446645030651022e-05,
      "loss": 1.6588,
      "step": 80448
    },
    {
      "epoch": 0.15881209291204673,
      "grad_norm": 0.1027870625257492,
      "learning_rate": 8.446011008214554e-05,
      "loss": 1.6746,
      "step": 80480
    },
    {
      "epoch": 0.15887523887344315,
      "grad_norm": 0.09794238954782486,
      "learning_rate": 8.445376985778085e-05,
      "loss": 1.6704,
      "step": 80512
    },
    {
      "epoch": 0.1589383848348396,
      "grad_norm": 0.11949475109577179,
      "learning_rate": 8.444742963341615e-05,
      "loss": 1.6639,
      "step": 80544
    },
    {
      "epoch": 0.15900153079623605,
      "grad_norm": 0.11027876287698746,
      "learning_rate": 8.444108940905147e-05,
      "loss": 1.6652,
      "step": 80576
    },
    {
      "epoch": 0.15906467675763247,
      "grad_norm": 0.10230275243520737,
      "learning_rate": 8.443474918468678e-05,
      "loss": 1.6739,
      "step": 80608
    },
    {
      "epoch": 0.15912782271902892,
      "grad_norm": 0.09737065434455872,
      "learning_rate": 8.44284089603221e-05,
      "loss": 1.666,
      "step": 80640
    },
    {
      "epoch": 0.15919096868042537,
      "grad_norm": 0.10272245109081268,
      "learning_rate": 8.44220687359574e-05,
      "loss": 1.6733,
      "step": 80672
    },
    {
      "epoch": 0.1592541146418218,
      "grad_norm": 0.10037832707166672,
      "learning_rate": 8.441572851159271e-05,
      "loss": 1.6661,
      "step": 80704
    },
    {
      "epoch": 0.15931726060321824,
      "grad_norm": 0.1051454022526741,
      "learning_rate": 8.440938828722802e-05,
      "loss": 1.6788,
      "step": 80736
    },
    {
      "epoch": 0.1593804065646147,
      "grad_norm": 0.10609681904315948,
      "learning_rate": 8.440304806286334e-05,
      "loss": 1.6648,
      "step": 80768
    },
    {
      "epoch": 0.1594435525260111,
      "grad_norm": 0.09505396336317062,
      "learning_rate": 8.439670783849864e-05,
      "loss": 1.6644,
      "step": 80800
    },
    {
      "epoch": 0.15950669848740756,
      "grad_norm": 0.09724284708499908,
      "learning_rate": 8.439036761413394e-05,
      "loss": 1.6721,
      "step": 80832
    },
    {
      "epoch": 0.159569844448804,
      "grad_norm": 0.10502568632364273,
      "learning_rate": 8.438402738976926e-05,
      "loss": 1.6718,
      "step": 80864
    },
    {
      "epoch": 0.15963299041020043,
      "grad_norm": 0.10341182351112366,
      "learning_rate": 8.437768716540457e-05,
      "loss": 1.6497,
      "step": 80896
    },
    {
      "epoch": 0.15969613637159688,
      "grad_norm": 0.10201447457075119,
      "learning_rate": 8.437134694103987e-05,
      "loss": 1.6765,
      "step": 80928
    },
    {
      "epoch": 0.15975928233299333,
      "grad_norm": 0.1063053160905838,
      "learning_rate": 8.436500671667519e-05,
      "loss": 1.6829,
      "step": 80960
    },
    {
      "epoch": 0.15982242829438975,
      "grad_norm": 0.11010798811912537,
      "learning_rate": 8.43586664923105e-05,
      "loss": 1.6729,
      "step": 80992
    },
    {
      "epoch": 0.1598855742557862,
      "grad_norm": 0.1061517745256424,
      "learning_rate": 8.435232626794581e-05,
      "loss": 1.6761,
      "step": 81024
    },
    {
      "epoch": 0.15994872021718265,
      "grad_norm": 0.10358405858278275,
      "learning_rate": 8.434598604358113e-05,
      "loss": 1.6689,
      "step": 81056
    },
    {
      "epoch": 0.16001186617857907,
      "grad_norm": 0.09589546918869019,
      "learning_rate": 8.433964581921643e-05,
      "loss": 1.6787,
      "step": 81088
    },
    {
      "epoch": 0.16007501213997552,
      "grad_norm": 0.1017046868801117,
      "learning_rate": 8.433330559485175e-05,
      "loss": 1.6678,
      "step": 81120
    },
    {
      "epoch": 0.16013815810137197,
      "grad_norm": 0.0994737520813942,
      "learning_rate": 8.432696537048706e-05,
      "loss": 1.657,
      "step": 81152
    },
    {
      "epoch": 0.1602013040627684,
      "grad_norm": 0.10340292751789093,
      "learning_rate": 8.432062514612236e-05,
      "loss": 1.6601,
      "step": 81184
    },
    {
      "epoch": 0.16026445002416484,
      "grad_norm": 0.09561502188444138,
      "learning_rate": 8.431428492175768e-05,
      "loss": 1.6774,
      "step": 81216
    },
    {
      "epoch": 0.1603275959855613,
      "grad_norm": 0.10668741166591644,
      "learning_rate": 8.430794469739298e-05,
      "loss": 1.6667,
      "step": 81248
    },
    {
      "epoch": 0.1603907419469577,
      "grad_norm": 0.09333831816911697,
      "learning_rate": 8.430160447302829e-05,
      "loss": 1.6738,
      "step": 81280
    },
    {
      "epoch": 0.16045388790835416,
      "grad_norm": 0.10256980359554291,
      "learning_rate": 8.42952642486636e-05,
      "loss": 1.6556,
      "step": 81312
    },
    {
      "epoch": 0.1605170338697506,
      "grad_norm": 0.09653441607952118,
      "learning_rate": 8.42889240242989e-05,
      "loss": 1.6774,
      "step": 81344
    },
    {
      "epoch": 0.16058017983114703,
      "grad_norm": 0.10778200626373291,
      "learning_rate": 8.428258379993422e-05,
      "loss": 1.6588,
      "step": 81376
    },
    {
      "epoch": 0.16064332579254348,
      "grad_norm": 0.10164060443639755,
      "learning_rate": 8.427624357556954e-05,
      "loss": 1.6669,
      "step": 81408
    },
    {
      "epoch": 0.16070647175393993,
      "grad_norm": 0.10250438004732132,
      "learning_rate": 8.426990335120485e-05,
      "loss": 1.6779,
      "step": 81440
    },
    {
      "epoch": 0.16076961771533635,
      "grad_norm": 0.09784379601478577,
      "learning_rate": 8.426356312684015e-05,
      "loss": 1.6629,
      "step": 81472
    },
    {
      "epoch": 0.1608327636767328,
      "grad_norm": 0.10251688212156296,
      "learning_rate": 8.425722290247547e-05,
      "loss": 1.671,
      "step": 81504
    },
    {
      "epoch": 0.16089590963812925,
      "grad_norm": 0.09898275136947632,
      "learning_rate": 8.425088267811078e-05,
      "loss": 1.6728,
      "step": 81536
    },
    {
      "epoch": 0.16095905559952567,
      "grad_norm": 0.10107841342687607,
      "learning_rate": 8.42445424537461e-05,
      "loss": 1.6674,
      "step": 81568
    },
    {
      "epoch": 0.16102220156092212,
      "grad_norm": 0.11603739857673645,
      "learning_rate": 8.42382022293814e-05,
      "loss": 1.6776,
      "step": 81600
    },
    {
      "epoch": 0.16108534752231857,
      "grad_norm": 0.10540154576301575,
      "learning_rate": 8.423186200501671e-05,
      "loss": 1.6736,
      "step": 81632
    },
    {
      "epoch": 0.161148493483715,
      "grad_norm": 0.09439444541931152,
      "learning_rate": 8.422552178065201e-05,
      "loss": 1.6667,
      "step": 81664
    },
    {
      "epoch": 0.16121163944511144,
      "grad_norm": 0.10692646354436874,
      "learning_rate": 8.421918155628733e-05,
      "loss": 1.6667,
      "step": 81696
    },
    {
      "epoch": 0.1612747854065079,
      "grad_norm": 0.10389810055494308,
      "learning_rate": 8.421284133192264e-05,
      "loss": 1.6669,
      "step": 81728
    },
    {
      "epoch": 0.16133793136790434,
      "grad_norm": 0.10967563092708588,
      "learning_rate": 8.420650110755794e-05,
      "loss": 1.6598,
      "step": 81760
    },
    {
      "epoch": 0.16140107732930076,
      "grad_norm": 0.1001477912068367,
      "learning_rate": 8.420016088319326e-05,
      "loss": 1.6773,
      "step": 81792
    },
    {
      "epoch": 0.1614642232906972,
      "grad_norm": 0.1112506166100502,
      "learning_rate": 8.419382065882857e-05,
      "loss": 1.6747,
      "step": 81824
    },
    {
      "epoch": 0.16152736925209366,
      "grad_norm": 0.10486098378896713,
      "learning_rate": 8.418748043446389e-05,
      "loss": 1.6714,
      "step": 81856
    },
    {
      "epoch": 0.16159051521349008,
      "grad_norm": 0.10470446199178696,
      "learning_rate": 8.418114021009919e-05,
      "loss": 1.6786,
      "step": 81888
    },
    {
      "epoch": 0.16165366117488653,
      "grad_norm": 0.09796930849552155,
      "learning_rate": 8.41747999857345e-05,
      "loss": 1.6629,
      "step": 81920
    },
    {
      "epoch": 0.16171680713628298,
      "grad_norm": 0.10023059695959091,
      "learning_rate": 8.416845976136982e-05,
      "loss": 1.6488,
      "step": 81952
    },
    {
      "epoch": 0.1617799530976794,
      "grad_norm": 0.10906977206468582,
      "learning_rate": 8.416211953700513e-05,
      "loss": 1.6728,
      "step": 81984
    },
    {
      "epoch": 0.16184309905907585,
      "grad_norm": 0.09570545703172684,
      "learning_rate": 8.415577931264043e-05,
      "loss": 1.6658,
      "step": 82016
    },
    {
      "epoch": 0.1619062450204723,
      "grad_norm": 0.10845465958118439,
      "learning_rate": 8.414943908827575e-05,
      "loss": 1.6776,
      "step": 82048
    },
    {
      "epoch": 0.16196939098186872,
      "grad_norm": 0.10419274866580963,
      "learning_rate": 8.414309886391105e-05,
      "loss": 1.6665,
      "step": 82080
    },
    {
      "epoch": 0.16203253694326517,
      "grad_norm": 0.10235957056283951,
      "learning_rate": 8.413675863954636e-05,
      "loss": 1.6595,
      "step": 82112
    },
    {
      "epoch": 0.16209568290466161,
      "grad_norm": 0.10116744041442871,
      "learning_rate": 8.413041841518166e-05,
      "loss": 1.6545,
      "step": 82144
    },
    {
      "epoch": 0.16215882886605804,
      "grad_norm": 0.10140691697597504,
      "learning_rate": 8.412407819081698e-05,
      "loss": 1.6705,
      "step": 82176
    },
    {
      "epoch": 0.16222197482745448,
      "grad_norm": 0.1144220307469368,
      "learning_rate": 8.411773796645229e-05,
      "loss": 1.6673,
      "step": 82208
    },
    {
      "epoch": 0.16228512078885093,
      "grad_norm": 0.10034649074077606,
      "learning_rate": 8.41113977420876e-05,
      "loss": 1.6582,
      "step": 82240
    },
    {
      "epoch": 0.16234826675024736,
      "grad_norm": 0.10167563706636429,
      "learning_rate": 8.41050575177229e-05,
      "loss": 1.6585,
      "step": 82272
    },
    {
      "epoch": 0.1624114127116438,
      "grad_norm": 0.10308732092380524,
      "learning_rate": 8.409871729335822e-05,
      "loss": 1.6816,
      "step": 82304
    },
    {
      "epoch": 0.16247455867304025,
      "grad_norm": 0.10687581449747086,
      "learning_rate": 8.409237706899354e-05,
      "loss": 1.662,
      "step": 82336
    },
    {
      "epoch": 0.16253770463443667,
      "grad_norm": 0.09884197264909744,
      "learning_rate": 8.408603684462885e-05,
      "loss": 1.6751,
      "step": 82368
    },
    {
      "epoch": 0.16260085059583312,
      "grad_norm": 0.10333535820245743,
      "learning_rate": 8.407969662026416e-05,
      "loss": 1.6728,
      "step": 82400
    },
    {
      "epoch": 0.16266399655722957,
      "grad_norm": 0.10113651305437088,
      "learning_rate": 8.407335639589947e-05,
      "loss": 1.6579,
      "step": 82432
    },
    {
      "epoch": 0.162727142518626,
      "grad_norm": 0.10472423583269119,
      "learning_rate": 8.406701617153478e-05,
      "loss": 1.6762,
      "step": 82464
    },
    {
      "epoch": 0.16279028848002244,
      "grad_norm": 0.0948818027973175,
      "learning_rate": 8.406067594717008e-05,
      "loss": 1.6623,
      "step": 82496
    },
    {
      "epoch": 0.1628534344414189,
      "grad_norm": 0.10804177820682526,
      "learning_rate": 8.40543357228054e-05,
      "loss": 1.6655,
      "step": 82528
    },
    {
      "epoch": 0.16291658040281531,
      "grad_norm": 0.11333528161048889,
      "learning_rate": 8.40479954984407e-05,
      "loss": 1.6641,
      "step": 82560
    },
    {
      "epoch": 0.16297972636421176,
      "grad_norm": 0.10604579001665115,
      "learning_rate": 8.404165527407601e-05,
      "loss": 1.6605,
      "step": 82592
    },
    {
      "epoch": 0.1630428723256082,
      "grad_norm": 0.09933172911405563,
      "learning_rate": 8.403531504971133e-05,
      "loss": 1.694,
      "step": 82624
    },
    {
      "epoch": 0.16310601828700463,
      "grad_norm": 0.10318920016288757,
      "learning_rate": 8.402897482534664e-05,
      "loss": 1.6812,
      "step": 82656
    },
    {
      "epoch": 0.16316916424840108,
      "grad_norm": 0.09602219611406326,
      "learning_rate": 8.402263460098194e-05,
      "loss": 1.6704,
      "step": 82688
    },
    {
      "epoch": 0.16323231020979753,
      "grad_norm": 0.10676294565200806,
      "learning_rate": 8.401629437661726e-05,
      "loss": 1.6566,
      "step": 82720
    },
    {
      "epoch": 0.16329545617119395,
      "grad_norm": 0.10345187038183212,
      "learning_rate": 8.400995415225257e-05,
      "loss": 1.6609,
      "step": 82752
    },
    {
      "epoch": 0.1633586021325904,
      "grad_norm": 0.10429804027080536,
      "learning_rate": 8.400361392788789e-05,
      "loss": 1.6608,
      "step": 82784
    },
    {
      "epoch": 0.16342174809398685,
      "grad_norm": 0.10219980031251907,
      "learning_rate": 8.399727370352319e-05,
      "loss": 1.6531,
      "step": 82816
    },
    {
      "epoch": 0.16348489405538327,
      "grad_norm": 0.0946764349937439,
      "learning_rate": 8.39909334791585e-05,
      "loss": 1.6587,
      "step": 82848
    },
    {
      "epoch": 0.16354804001677972,
      "grad_norm": 0.10328608751296997,
      "learning_rate": 8.39845932547938e-05,
      "loss": 1.6666,
      "step": 82880
    },
    {
      "epoch": 0.16361118597817617,
      "grad_norm": 0.10191641002893448,
      "learning_rate": 8.397825303042912e-05,
      "loss": 1.657,
      "step": 82912
    },
    {
      "epoch": 0.1636743319395726,
      "grad_norm": 0.10231659561395645,
      "learning_rate": 8.397191280606442e-05,
      "loss": 1.6588,
      "step": 82944
    },
    {
      "epoch": 0.16373747790096904,
      "grad_norm": 0.10225328803062439,
      "learning_rate": 8.396557258169973e-05,
      "loss": 1.6701,
      "step": 82976
    },
    {
      "epoch": 0.1638006238623655,
      "grad_norm": 0.1010192334651947,
      "learning_rate": 8.395923235733505e-05,
      "loss": 1.6781,
      "step": 83008
    },
    {
      "epoch": 0.1638637698237619,
      "grad_norm": 0.09910804033279419,
      "learning_rate": 8.395289213297036e-05,
      "loss": 1.6547,
      "step": 83040
    },
    {
      "epoch": 0.16392691578515836,
      "grad_norm": 0.10391703993082047,
      "learning_rate": 8.394655190860568e-05,
      "loss": 1.6599,
      "step": 83072
    },
    {
      "epoch": 0.1639900617465548,
      "grad_norm": 0.10021750628948212,
      "learning_rate": 8.394021168424098e-05,
      "loss": 1.6714,
      "step": 83104
    },
    {
      "epoch": 0.16405320770795123,
      "grad_norm": 0.10518373548984528,
      "learning_rate": 8.393387145987629e-05,
      "loss": 1.6654,
      "step": 83136
    },
    {
      "epoch": 0.16411635366934768,
      "grad_norm": 0.09555603563785553,
      "learning_rate": 8.39275312355116e-05,
      "loss": 1.66,
      "step": 83168
    },
    {
      "epoch": 0.16417949963074413,
      "grad_norm": 0.10554922372102737,
      "learning_rate": 8.392119101114692e-05,
      "loss": 1.6524,
      "step": 83200
    },
    {
      "epoch": 0.16424264559214055,
      "grad_norm": 0.10174940526485443,
      "learning_rate": 8.391485078678222e-05,
      "loss": 1.6525,
      "step": 83232
    },
    {
      "epoch": 0.164305791553537,
      "grad_norm": 0.11020279675722122,
      "learning_rate": 8.390851056241754e-05,
      "loss": 1.66,
      "step": 83264
    },
    {
      "epoch": 0.16436893751493345,
      "grad_norm": 0.10660487413406372,
      "learning_rate": 8.390217033805284e-05,
      "loss": 1.6656,
      "step": 83296
    },
    {
      "epoch": 0.16443208347632987,
      "grad_norm": 0.11338844895362854,
      "learning_rate": 8.389583011368815e-05,
      "loss": 1.6686,
      "step": 83328
    },
    {
      "epoch": 0.16449522943772632,
      "grad_norm": 0.11645433306694031,
      "learning_rate": 8.388948988932345e-05,
      "loss": 1.6536,
      "step": 83360
    },
    {
      "epoch": 0.16455837539912277,
      "grad_norm": 0.1066615954041481,
      "learning_rate": 8.388314966495877e-05,
      "loss": 1.6455,
      "step": 83392
    },
    {
      "epoch": 0.1646215213605192,
      "grad_norm": 0.09756056219339371,
      "learning_rate": 8.387680944059408e-05,
      "loss": 1.6617,
      "step": 83424
    },
    {
      "epoch": 0.16468466732191564,
      "grad_norm": 0.10661061108112335,
      "learning_rate": 8.38704692162294e-05,
      "loss": 1.6584,
      "step": 83456
    },
    {
      "epoch": 0.1647478132833121,
      "grad_norm": 0.10062295198440552,
      "learning_rate": 8.38641289918647e-05,
      "loss": 1.6647,
      "step": 83488
    },
    {
      "epoch": 0.16481095924470854,
      "grad_norm": 0.10491617769002914,
      "learning_rate": 8.385778876750001e-05,
      "loss": 1.6738,
      "step": 83520
    },
    {
      "epoch": 0.16487410520610496,
      "grad_norm": 0.10141460597515106,
      "learning_rate": 8.385144854313533e-05,
      "loss": 1.6648,
      "step": 83552
    },
    {
      "epoch": 0.1649372511675014,
      "grad_norm": 0.09942572563886642,
      "learning_rate": 8.384510831877064e-05,
      "loss": 1.6558,
      "step": 83584
    },
    {
      "epoch": 0.16500039712889786,
      "grad_norm": 0.10586068034172058,
      "learning_rate": 8.383876809440594e-05,
      "loss": 1.6622,
      "step": 83616
    },
    {
      "epoch": 0.16506354309029428,
      "grad_norm": 0.10023793578147888,
      "learning_rate": 8.383242787004126e-05,
      "loss": 1.6782,
      "step": 83648
    },
    {
      "epoch": 0.16512668905169073,
      "grad_norm": 0.09988317638635635,
      "learning_rate": 8.382608764567657e-05,
      "loss": 1.6652,
      "step": 83680
    },
    {
      "epoch": 0.16518983501308718,
      "grad_norm": 0.09722260385751724,
      "learning_rate": 8.381974742131187e-05,
      "loss": 1.6531,
      "step": 83712
    },
    {
      "epoch": 0.1652529809744836,
      "grad_norm": 0.09963898360729218,
      "learning_rate": 8.381340719694719e-05,
      "loss": 1.6614,
      "step": 83744
    },
    {
      "epoch": 0.16531612693588005,
      "grad_norm": 0.10098837316036224,
      "learning_rate": 8.380706697258249e-05,
      "loss": 1.6596,
      "step": 83776
    },
    {
      "epoch": 0.1653792728972765,
      "grad_norm": 0.10564694553613663,
      "learning_rate": 8.38007267482178e-05,
      "loss": 1.6697,
      "step": 83808
    },
    {
      "epoch": 0.16544241885867292,
      "grad_norm": 0.10261000692844391,
      "learning_rate": 8.379438652385312e-05,
      "loss": 1.6735,
      "step": 83840
    },
    {
      "epoch": 0.16550556482006937,
      "grad_norm": 0.09658705443143845,
      "learning_rate": 8.378804629948843e-05,
      "loss": 1.6668,
      "step": 83872
    },
    {
      "epoch": 0.16556871078146582,
      "grad_norm": 0.09841986000537872,
      "learning_rate": 8.378170607512373e-05,
      "loss": 1.6716,
      "step": 83904
    },
    {
      "epoch": 0.16563185674286224,
      "grad_norm": 0.09860815852880478,
      "learning_rate": 8.377536585075905e-05,
      "loss": 1.6633,
      "step": 83936
    },
    {
      "epoch": 0.1656950027042587,
      "grad_norm": 0.1066829040646553,
      "learning_rate": 8.376902562639436e-05,
      "loss": 1.6799,
      "step": 83968
    },
    {
      "epoch": 0.16575814866565514,
      "grad_norm": 0.10175075381994247,
      "learning_rate": 8.376268540202968e-05,
      "loss": 1.662,
      "step": 84000
    },
    {
      "epoch": 0.16582129462705156,
      "grad_norm": 0.09894043207168579,
      "learning_rate": 8.375634517766498e-05,
      "loss": 1.6349,
      "step": 84032
    },
    {
      "epoch": 0.165884440588448,
      "grad_norm": 0.10034500062465668,
      "learning_rate": 8.375000495330029e-05,
      "loss": 1.66,
      "step": 84064
    },
    {
      "epoch": 0.16594758654984446,
      "grad_norm": 0.10211653262376785,
      "learning_rate": 8.37436647289356e-05,
      "loss": 1.6538,
      "step": 84096
    },
    {
      "epoch": 0.16601073251124088,
      "grad_norm": 0.0991155281662941,
      "learning_rate": 8.373732450457091e-05,
      "loss": 1.6637,
      "step": 84128
    },
    {
      "epoch": 0.16607387847263733,
      "grad_norm": 0.10942571610212326,
      "learning_rate": 8.373098428020621e-05,
      "loss": 1.6692,
      "step": 84160
    },
    {
      "epoch": 0.16613702443403378,
      "grad_norm": 0.11085399240255356,
      "learning_rate": 8.372464405584152e-05,
      "loss": 1.6748,
      "step": 84192
    },
    {
      "epoch": 0.1662001703954302,
      "grad_norm": 0.10118217766284943,
      "learning_rate": 8.371830383147684e-05,
      "loss": 1.6611,
      "step": 84224
    },
    {
      "epoch": 0.16626331635682665,
      "grad_norm": 0.09579329937696457,
      "learning_rate": 8.371196360711215e-05,
      "loss": 1.6676,
      "step": 84256
    },
    {
      "epoch": 0.1663264623182231,
      "grad_norm": 0.09783437103033066,
      "learning_rate": 8.370562338274745e-05,
      "loss": 1.6649,
      "step": 84288
    },
    {
      "epoch": 0.16638960827961952,
      "grad_norm": 0.10741367936134338,
      "learning_rate": 8.369928315838277e-05,
      "loss": 1.6542,
      "step": 84320
    },
    {
      "epoch": 0.16645275424101597,
      "grad_norm": 0.10241501778364182,
      "learning_rate": 8.369294293401808e-05,
      "loss": 1.6619,
      "step": 84352
    },
    {
      "epoch": 0.16651590020241241,
      "grad_norm": 0.10578735917806625,
      "learning_rate": 8.36866027096534e-05,
      "loss": 1.6591,
      "step": 84384
    },
    {
      "epoch": 0.16657904616380884,
      "grad_norm": 0.09896484762430191,
      "learning_rate": 8.368026248528871e-05,
      "loss": 1.646,
      "step": 84416
    },
    {
      "epoch": 0.16664219212520529,
      "grad_norm": 0.10287652909755707,
      "learning_rate": 8.367392226092401e-05,
      "loss": 1.6564,
      "step": 84448
    },
    {
      "epoch": 0.16670533808660173,
      "grad_norm": 0.10269158333539963,
      "learning_rate": 8.366758203655933e-05,
      "loss": 1.6732,
      "step": 84480
    },
    {
      "epoch": 0.16676848404799816,
      "grad_norm": 0.1048140749335289,
      "learning_rate": 8.366124181219464e-05,
      "loss": 1.6784,
      "step": 84512
    },
    {
      "epoch": 0.1668316300093946,
      "grad_norm": 0.09970742464065552,
      "learning_rate": 8.365490158782994e-05,
      "loss": 1.6595,
      "step": 84544
    },
    {
      "epoch": 0.16689477597079105,
      "grad_norm": 0.10886450111865997,
      "learning_rate": 8.364856136346524e-05,
      "loss": 1.653,
      "step": 84576
    },
    {
      "epoch": 0.16695792193218748,
      "grad_norm": 0.09977026283740997,
      "learning_rate": 8.364222113910056e-05,
      "loss": 1.6516,
      "step": 84608
    },
    {
      "epoch": 0.16702106789358392,
      "grad_norm": 0.09926028549671173,
      "learning_rate": 8.363588091473587e-05,
      "loss": 1.6632,
      "step": 84640
    },
    {
      "epoch": 0.16708421385498037,
      "grad_norm": 0.1021278128027916,
      "learning_rate": 8.362954069037119e-05,
      "loss": 1.6649,
      "step": 84672
    },
    {
      "epoch": 0.1671473598163768,
      "grad_norm": 0.1026868224143982,
      "learning_rate": 8.362320046600649e-05,
      "loss": 1.6732,
      "step": 84704
    },
    {
      "epoch": 0.16721050577777324,
      "grad_norm": 0.10374830663204193,
      "learning_rate": 8.36168602416418e-05,
      "loss": 1.6543,
      "step": 84736
    },
    {
      "epoch": 0.1672736517391697,
      "grad_norm": 0.10258857160806656,
      "learning_rate": 8.361052001727712e-05,
      "loss": 1.6708,
      "step": 84768
    },
    {
      "epoch": 0.16733679770056611,
      "grad_norm": 0.0977817177772522,
      "learning_rate": 8.360417979291243e-05,
      "loss": 1.6613,
      "step": 84800
    },
    {
      "epoch": 0.16739994366196256,
      "grad_norm": 0.10027941316366196,
      "learning_rate": 8.359783956854773e-05,
      "loss": 1.6462,
      "step": 84832
    },
    {
      "epoch": 0.167463089623359,
      "grad_norm": 0.10649436712265015,
      "learning_rate": 8.359149934418305e-05,
      "loss": 1.6552,
      "step": 84864
    },
    {
      "epoch": 0.16752623558475543,
      "grad_norm": 0.09952892363071442,
      "learning_rate": 8.358515911981836e-05,
      "loss": 1.6554,
      "step": 84896
    },
    {
      "epoch": 0.16758938154615188,
      "grad_norm": 0.1102275550365448,
      "learning_rate": 8.357881889545368e-05,
      "loss": 1.6701,
      "step": 84928
    },
    {
      "epoch": 0.16765252750754833,
      "grad_norm": 0.10210400074720383,
      "learning_rate": 8.357247867108898e-05,
      "loss": 1.6791,
      "step": 84960
    },
    {
      "epoch": 0.16771567346894475,
      "grad_norm": 0.09926417469978333,
      "learning_rate": 8.356613844672428e-05,
      "loss": 1.6625,
      "step": 84992
    },
    {
      "epoch": 0.1677788194303412,
      "grad_norm": 0.09948869794607162,
      "learning_rate": 8.355979822235959e-05,
      "loss": 1.6543,
      "step": 85024
    },
    {
      "epoch": 0.16784196539173765,
      "grad_norm": 0.10216159373521805,
      "learning_rate": 8.355345799799491e-05,
      "loss": 1.6558,
      "step": 85056
    },
    {
      "epoch": 0.16790511135313407,
      "grad_norm": 0.10084868222475052,
      "learning_rate": 8.354711777363022e-05,
      "loss": 1.6678,
      "step": 85088
    },
    {
      "epoch": 0.16796825731453052,
      "grad_norm": 0.10927193611860275,
      "learning_rate": 8.354077754926552e-05,
      "loss": 1.6493,
      "step": 85120
    },
    {
      "epoch": 0.16803140327592697,
      "grad_norm": 0.10268424451351166,
      "learning_rate": 8.353443732490084e-05,
      "loss": 1.6592,
      "step": 85152
    },
    {
      "epoch": 0.16809454923732342,
      "grad_norm": 0.09640819579362869,
      "learning_rate": 8.352809710053615e-05,
      "loss": 1.6541,
      "step": 85184
    },
    {
      "epoch": 0.16815769519871984,
      "grad_norm": 0.09388112276792526,
      "learning_rate": 8.352175687617147e-05,
      "loss": 1.6603,
      "step": 85216
    },
    {
      "epoch": 0.1682208411601163,
      "grad_norm": 0.09656675904989243,
      "learning_rate": 8.351541665180677e-05,
      "loss": 1.6556,
      "step": 85248
    },
    {
      "epoch": 0.16828398712151274,
      "grad_norm": 0.10099435597658157,
      "learning_rate": 8.350907642744208e-05,
      "loss": 1.6662,
      "step": 85280
    },
    {
      "epoch": 0.16834713308290916,
      "grad_norm": 0.10132823139429092,
      "learning_rate": 8.35027362030774e-05,
      "loss": 1.6738,
      "step": 85312
    },
    {
      "epoch": 0.1684102790443056,
      "grad_norm": 0.10148778557777405,
      "learning_rate": 8.34963959787127e-05,
      "loss": 1.6613,
      "step": 85344
    },
    {
      "epoch": 0.16847342500570206,
      "grad_norm": 0.10271482914686203,
      "learning_rate": 8.349005575434801e-05,
      "loss": 1.6661,
      "step": 85376
    },
    {
      "epoch": 0.16853657096709848,
      "grad_norm": 0.10997402667999268,
      "learning_rate": 8.348371552998331e-05,
      "loss": 1.6614,
      "step": 85408
    },
    {
      "epoch": 0.16859971692849493,
      "grad_norm": 0.0976165235042572,
      "learning_rate": 8.347737530561863e-05,
      "loss": 1.6626,
      "step": 85440
    },
    {
      "epoch": 0.16866286288989138,
      "grad_norm": 0.10330215841531754,
      "learning_rate": 8.347103508125394e-05,
      "loss": 1.6656,
      "step": 85472
    },
    {
      "epoch": 0.1687260088512878,
      "grad_norm": 0.10073370486497879,
      "learning_rate": 8.346469485688924e-05,
      "loss": 1.6602,
      "step": 85504
    },
    {
      "epoch": 0.16878915481268425,
      "grad_norm": 0.10443852841854095,
      "learning_rate": 8.345835463252456e-05,
      "loss": 1.6633,
      "step": 85536
    },
    {
      "epoch": 0.1688523007740807,
      "grad_norm": 0.10332836210727692,
      "learning_rate": 8.345201440815987e-05,
      "loss": 1.6622,
      "step": 85568
    },
    {
      "epoch": 0.16891544673547712,
      "grad_norm": 0.09791575372219086,
      "learning_rate": 8.344567418379519e-05,
      "loss": 1.6719,
      "step": 85600
    },
    {
      "epoch": 0.16897859269687357,
      "grad_norm": 0.09709091484546661,
      "learning_rate": 8.343933395943049e-05,
      "loss": 1.6531,
      "step": 85632
    },
    {
      "epoch": 0.16904173865827002,
      "grad_norm": 0.10924059897661209,
      "learning_rate": 8.34329937350658e-05,
      "loss": 1.6542,
      "step": 85664
    },
    {
      "epoch": 0.16910488461966644,
      "grad_norm": 0.10236849635839462,
      "learning_rate": 8.342665351070112e-05,
      "loss": 1.6637,
      "step": 85696
    },
    {
      "epoch": 0.1691680305810629,
      "grad_norm": 0.10873173177242279,
      "learning_rate": 8.342031328633643e-05,
      "loss": 1.6609,
      "step": 85728
    },
    {
      "epoch": 0.16923117654245934,
      "grad_norm": 0.10035929828882217,
      "learning_rate": 8.341397306197173e-05,
      "loss": 1.6609,
      "step": 85760
    },
    {
      "epoch": 0.16929432250385576,
      "grad_norm": 0.10346246510744095,
      "learning_rate": 8.340763283760705e-05,
      "loss": 1.6717,
      "step": 85792
    },
    {
      "epoch": 0.1693574684652522,
      "grad_norm": 0.10286826640367508,
      "learning_rate": 8.340129261324235e-05,
      "loss": 1.6653,
      "step": 85824
    },
    {
      "epoch": 0.16942061442664866,
      "grad_norm": 0.10834146291017532,
      "learning_rate": 8.339495238887766e-05,
      "loss": 1.6583,
      "step": 85856
    },
    {
      "epoch": 0.16948376038804508,
      "grad_norm": 0.10519353300333023,
      "learning_rate": 8.338861216451298e-05,
      "loss": 1.6515,
      "step": 85888
    },
    {
      "epoch": 0.16954690634944153,
      "grad_norm": 0.09746549278497696,
      "learning_rate": 8.338227194014828e-05,
      "loss": 1.6484,
      "step": 85920
    },
    {
      "epoch": 0.16961005231083798,
      "grad_norm": 0.10242948681116104,
      "learning_rate": 8.33759317157836e-05,
      "loss": 1.6576,
      "step": 85952
    },
    {
      "epoch": 0.1696731982722344,
      "grad_norm": 0.10463360697031021,
      "learning_rate": 8.336959149141891e-05,
      "loss": 1.6571,
      "step": 85984
    },
    {
      "epoch": 0.16973634423363085,
      "grad_norm": 0.09747793525457382,
      "learning_rate": 8.336325126705422e-05,
      "loss": 1.6421,
      "step": 86016
    },
    {
      "epoch": 0.1697994901950273,
      "grad_norm": 0.09792450070381165,
      "learning_rate": 8.335691104268952e-05,
      "loss": 1.6697,
      "step": 86048
    },
    {
      "epoch": 0.16986263615642372,
      "grad_norm": 0.10239899158477783,
      "learning_rate": 8.335057081832484e-05,
      "loss": 1.6641,
      "step": 86080
    },
    {
      "epoch": 0.16992578211782017,
      "grad_norm": 0.10340501368045807,
      "learning_rate": 8.334423059396015e-05,
      "loss": 1.6655,
      "step": 86112
    },
    {
      "epoch": 0.16998892807921662,
      "grad_norm": 0.10833476483821869,
      "learning_rate": 8.333789036959547e-05,
      "loss": 1.6686,
      "step": 86144
    },
    {
      "epoch": 0.17005207404061304,
      "grad_norm": 0.1196788027882576,
      "learning_rate": 8.333155014523077e-05,
      "loss": 1.6523,
      "step": 86176
    },
    {
      "epoch": 0.1701152200020095,
      "grad_norm": 0.1082020178437233,
      "learning_rate": 8.332520992086608e-05,
      "loss": 1.658,
      "step": 86208
    },
    {
      "epoch": 0.17017836596340594,
      "grad_norm": 0.10265053063631058,
      "learning_rate": 8.331886969650138e-05,
      "loss": 1.665,
      "step": 86240
    },
    {
      "epoch": 0.17024151192480236,
      "grad_norm": 0.10248216241598129,
      "learning_rate": 8.33125294721367e-05,
      "loss": 1.6588,
      "step": 86272
    },
    {
      "epoch": 0.1703046578861988,
      "grad_norm": 0.10473192483186722,
      "learning_rate": 8.3306189247772e-05,
      "loss": 1.6745,
      "step": 86304
    },
    {
      "epoch": 0.17036780384759526,
      "grad_norm": 0.10582578927278519,
      "learning_rate": 8.329984902340731e-05,
      "loss": 1.6676,
      "step": 86336
    },
    {
      "epoch": 0.17043094980899168,
      "grad_norm": 0.100953109562397,
      "learning_rate": 8.329350879904263e-05,
      "loss": 1.6838,
      "step": 86368
    },
    {
      "epoch": 0.17049409577038813,
      "grad_norm": 0.09963128715753555,
      "learning_rate": 8.328716857467794e-05,
      "loss": 1.6503,
      "step": 86400
    },
    {
      "epoch": 0.17055724173178458,
      "grad_norm": 0.10131825506687164,
      "learning_rate": 8.328082835031326e-05,
      "loss": 1.6704,
      "step": 86432
    },
    {
      "epoch": 0.170620387693181,
      "grad_norm": 0.10538109391927719,
      "learning_rate": 8.327448812594856e-05,
      "loss": 1.6662,
      "step": 86464
    },
    {
      "epoch": 0.17068353365457745,
      "grad_norm": 0.09289366006851196,
      "learning_rate": 8.326814790158387e-05,
      "loss": 1.6533,
      "step": 86496
    },
    {
      "epoch": 0.1707466796159739,
      "grad_norm": 0.09712038189172745,
      "learning_rate": 8.326180767721919e-05,
      "loss": 1.6626,
      "step": 86528
    },
    {
      "epoch": 0.17080982557737032,
      "grad_norm": 0.10773643851280212,
      "learning_rate": 8.32554674528545e-05,
      "loss": 1.6691,
      "step": 86560
    },
    {
      "epoch": 0.17087297153876677,
      "grad_norm": 0.11154456436634064,
      "learning_rate": 8.32491272284898e-05,
      "loss": 1.6543,
      "step": 86592
    },
    {
      "epoch": 0.17093611750016321,
      "grad_norm": 0.10421478003263474,
      "learning_rate": 8.324278700412512e-05,
      "loss": 1.6655,
      "step": 86624
    },
    {
      "epoch": 0.17099926346155964,
      "grad_norm": 0.09560660272836685,
      "learning_rate": 8.323644677976042e-05,
      "loss": 1.6557,
      "step": 86656
    },
    {
      "epoch": 0.17106240942295609,
      "grad_norm": 0.09936953336000443,
      "learning_rate": 8.323010655539573e-05,
      "loss": 1.6489,
      "step": 86688
    },
    {
      "epoch": 0.17112555538435253,
      "grad_norm": 0.11029977351427078,
      "learning_rate": 8.322376633103103e-05,
      "loss": 1.6391,
      "step": 86720
    },
    {
      "epoch": 0.17118870134574896,
      "grad_norm": 0.09954982995986938,
      "learning_rate": 8.321742610666635e-05,
      "loss": 1.6586,
      "step": 86752
    },
    {
      "epoch": 0.1712518473071454,
      "grad_norm": 0.10885919630527496,
      "learning_rate": 8.321108588230166e-05,
      "loss": 1.6642,
      "step": 86784
    },
    {
      "epoch": 0.17131499326854185,
      "grad_norm": 0.10321956872940063,
      "learning_rate": 8.320474565793698e-05,
      "loss": 1.6514,
      "step": 86816
    },
    {
      "epoch": 0.1713781392299383,
      "grad_norm": 0.09847011417150497,
      "learning_rate": 8.319840543357228e-05,
      "loss": 1.6589,
      "step": 86848
    },
    {
      "epoch": 0.17144128519133472,
      "grad_norm": 0.1070474162697792,
      "learning_rate": 8.31920652092076e-05,
      "loss": 1.6661,
      "step": 86880
    },
    {
      "epoch": 0.17150443115273117,
      "grad_norm": 0.10999678820371628,
      "learning_rate": 8.318572498484291e-05,
      "loss": 1.6527,
      "step": 86912
    },
    {
      "epoch": 0.17156757711412762,
      "grad_norm": 0.09914568066596985,
      "learning_rate": 8.317938476047822e-05,
      "loss": 1.6672,
      "step": 86944
    },
    {
      "epoch": 0.17163072307552404,
      "grad_norm": 0.10314370691776276,
      "learning_rate": 8.317304453611352e-05,
      "loss": 1.6613,
      "step": 86976
    },
    {
      "epoch": 0.1716938690369205,
      "grad_norm": 0.10294325649738312,
      "learning_rate": 8.316670431174884e-05,
      "loss": 1.6444,
      "step": 87008
    },
    {
      "epoch": 0.17175701499831694,
      "grad_norm": 0.10406816005706787,
      "learning_rate": 8.316036408738414e-05,
      "loss": 1.6573,
      "step": 87040
    },
    {
      "epoch": 0.17182016095971336,
      "grad_norm": 0.10395383089780807,
      "learning_rate": 8.315402386301945e-05,
      "loss": 1.6482,
      "step": 87072
    },
    {
      "epoch": 0.1718833069211098,
      "grad_norm": 0.10011985152959824,
      "learning_rate": 8.314768363865477e-05,
      "loss": 1.6689,
      "step": 87104
    },
    {
      "epoch": 0.17194645288250626,
      "grad_norm": 0.11181795597076416,
      "learning_rate": 8.314134341429007e-05,
      "loss": 1.6703,
      "step": 87136
    },
    {
      "epoch": 0.17200959884390268,
      "grad_norm": 0.09706149995326996,
      "learning_rate": 8.313500318992538e-05,
      "loss": 1.6728,
      "step": 87168
    },
    {
      "epoch": 0.17207274480529913,
      "grad_norm": 0.10702026635408401,
      "learning_rate": 8.31286629655607e-05,
      "loss": 1.6594,
      "step": 87200
    },
    {
      "epoch": 0.17213589076669558,
      "grad_norm": 0.10278090834617615,
      "learning_rate": 8.312232274119601e-05,
      "loss": 1.6631,
      "step": 87232
    },
    {
      "epoch": 0.172199036728092,
      "grad_norm": 0.10035611689090729,
      "learning_rate": 8.311598251683131e-05,
      "loss": 1.6727,
      "step": 87264
    },
    {
      "epoch": 0.17226218268948845,
      "grad_norm": 0.1062588095664978,
      "learning_rate": 8.310964229246663e-05,
      "loss": 1.6535,
      "step": 87296
    },
    {
      "epoch": 0.1723253286508849,
      "grad_norm": 0.1101216971874237,
      "learning_rate": 8.310330206810194e-05,
      "loss": 1.6557,
      "step": 87328
    },
    {
      "epoch": 0.17238847461228132,
      "grad_norm": 0.11037938296794891,
      "learning_rate": 8.309696184373726e-05,
      "loss": 1.6762,
      "step": 87360
    },
    {
      "epoch": 0.17245162057367777,
      "grad_norm": 0.0970001369714737,
      "learning_rate": 8.309062161937256e-05,
      "loss": 1.6585,
      "step": 87392
    },
    {
      "epoch": 0.17251476653507422,
      "grad_norm": 0.10286374390125275,
      "learning_rate": 8.308428139500787e-05,
      "loss": 1.6728,
      "step": 87424
    },
    {
      "epoch": 0.17257791249647064,
      "grad_norm": 0.09633680433034897,
      "learning_rate": 8.307794117064317e-05,
      "loss": 1.6608,
      "step": 87456
    },
    {
      "epoch": 0.1726410584578671,
      "grad_norm": 0.10094623267650604,
      "learning_rate": 8.307160094627849e-05,
      "loss": 1.6587,
      "step": 87488
    },
    {
      "epoch": 0.17270420441926354,
      "grad_norm": 0.09444501996040344,
      "learning_rate": 8.306526072191379e-05,
      "loss": 1.6594,
      "step": 87520
    },
    {
      "epoch": 0.17276735038065996,
      "grad_norm": 0.10603666305541992,
      "learning_rate": 8.30589204975491e-05,
      "loss": 1.6687,
      "step": 87552
    },
    {
      "epoch": 0.1728304963420564,
      "grad_norm": 0.10152776539325714,
      "learning_rate": 8.305258027318442e-05,
      "loss": 1.666,
      "step": 87584
    },
    {
      "epoch": 0.17289364230345286,
      "grad_norm": 0.1043393611907959,
      "learning_rate": 8.304624004881973e-05,
      "loss": 1.6549,
      "step": 87616
    },
    {
      "epoch": 0.17295678826484928,
      "grad_norm": 0.10143306851387024,
      "learning_rate": 8.303989982445503e-05,
      "loss": 1.6486,
      "step": 87648
    },
    {
      "epoch": 0.17301993422624573,
      "grad_norm": 0.09123295545578003,
      "learning_rate": 8.303355960009035e-05,
      "loss": 1.664,
      "step": 87680
    },
    {
      "epoch": 0.17308308018764218,
      "grad_norm": 0.10808151215314865,
      "learning_rate": 8.302721937572566e-05,
      "loss": 1.6567,
      "step": 87712
    },
    {
      "epoch": 0.1731462261490386,
      "grad_norm": 0.10542468726634979,
      "learning_rate": 8.302087915136098e-05,
      "loss": 1.6639,
      "step": 87744
    },
    {
      "epoch": 0.17320937211043505,
      "grad_norm": 0.10240886360406876,
      "learning_rate": 8.301453892699629e-05,
      "loss": 1.6557,
      "step": 87776
    },
    {
      "epoch": 0.1732725180718315,
      "grad_norm": 0.11001458019018173,
      "learning_rate": 8.30081987026316e-05,
      "loss": 1.6639,
      "step": 87808
    },
    {
      "epoch": 0.17333566403322792,
      "grad_norm": 0.09899230301380157,
      "learning_rate": 8.300185847826691e-05,
      "loss": 1.6707,
      "step": 87840
    },
    {
      "epoch": 0.17339880999462437,
      "grad_norm": 0.09759450703859329,
      "learning_rate": 8.299551825390221e-05,
      "loss": 1.651,
      "step": 87872
    },
    {
      "epoch": 0.17346195595602082,
      "grad_norm": 0.11124710738658905,
      "learning_rate": 8.298917802953752e-05,
      "loss": 1.6525,
      "step": 87904
    },
    {
      "epoch": 0.17352510191741724,
      "grad_norm": 0.10308229178190231,
      "learning_rate": 8.298283780517283e-05,
      "loss": 1.6621,
      "step": 87936
    },
    {
      "epoch": 0.1735882478788137,
      "grad_norm": 0.10963082313537598,
      "learning_rate": 8.297649758080814e-05,
      "loss": 1.6491,
      "step": 87968
    },
    {
      "epoch": 0.17365139384021014,
      "grad_norm": 0.0992642194032669,
      "learning_rate": 8.297015735644345e-05,
      "loss": 1.6526,
      "step": 88000
    },
    {
      "epoch": 0.17371453980160656,
      "grad_norm": 0.10322271287441254,
      "learning_rate": 8.296381713207877e-05,
      "loss": 1.6607,
      "step": 88032
    },
    {
      "epoch": 0.173777685763003,
      "grad_norm": 0.10885316878557205,
      "learning_rate": 8.295747690771407e-05,
      "loss": 1.6594,
      "step": 88064
    },
    {
      "epoch": 0.17384083172439946,
      "grad_norm": 0.10019826889038086,
      "learning_rate": 8.295113668334938e-05,
      "loss": 1.6461,
      "step": 88096
    },
    {
      "epoch": 0.17390397768579588,
      "grad_norm": 0.10750409960746765,
      "learning_rate": 8.29447964589847e-05,
      "loss": 1.6511,
      "step": 88128
    },
    {
      "epoch": 0.17396712364719233,
      "grad_norm": 0.11462727934122086,
      "learning_rate": 8.293845623462001e-05,
      "loss": 1.6696,
      "step": 88160
    },
    {
      "epoch": 0.17403026960858878,
      "grad_norm": 0.108080193400383,
      "learning_rate": 8.293211601025531e-05,
      "loss": 1.6668,
      "step": 88192
    },
    {
      "epoch": 0.1740934155699852,
      "grad_norm": 0.10029991716146469,
      "learning_rate": 8.292577578589063e-05,
      "loss": 1.655,
      "step": 88224
    },
    {
      "epoch": 0.17415656153138165,
      "grad_norm": 0.109747976064682,
      "learning_rate": 8.291943556152594e-05,
      "loss": 1.6587,
      "step": 88256
    },
    {
      "epoch": 0.1742197074927781,
      "grad_norm": 0.1071368008852005,
      "learning_rate": 8.291309533716124e-05,
      "loss": 1.6532,
      "step": 88288
    },
    {
      "epoch": 0.17428285345417452,
      "grad_norm": 0.09972008317708969,
      "learning_rate": 8.290675511279656e-05,
      "loss": 1.6526,
      "step": 88320
    },
    {
      "epoch": 0.17434599941557097,
      "grad_norm": 0.09849822521209717,
      "learning_rate": 8.290041488843186e-05,
      "loss": 1.655,
      "step": 88352
    },
    {
      "epoch": 0.17440914537696742,
      "grad_norm": 0.10606779903173447,
      "learning_rate": 8.289407466406717e-05,
      "loss": 1.653,
      "step": 88384
    },
    {
      "epoch": 0.17447229133836384,
      "grad_norm": 0.10402847826480865,
      "learning_rate": 8.288773443970249e-05,
      "loss": 1.6592,
      "step": 88416
    },
    {
      "epoch": 0.1745354372997603,
      "grad_norm": 0.10727513581514359,
      "learning_rate": 8.28813942153378e-05,
      "loss": 1.6612,
      "step": 88448
    },
    {
      "epoch": 0.17459858326115674,
      "grad_norm": 0.10152298957109451,
      "learning_rate": 8.28750539909731e-05,
      "loss": 1.6444,
      "step": 88480
    },
    {
      "epoch": 0.17466172922255319,
      "grad_norm": 0.10337173193693161,
      "learning_rate": 8.286871376660842e-05,
      "loss": 1.6538,
      "step": 88512
    },
    {
      "epoch": 0.1747248751839496,
      "grad_norm": 0.10739100724458694,
      "learning_rate": 8.286237354224373e-05,
      "loss": 1.6705,
      "step": 88544
    },
    {
      "epoch": 0.17478802114534606,
      "grad_norm": 0.10928995162248611,
      "learning_rate": 8.285603331787905e-05,
      "loss": 1.6643,
      "step": 88576
    },
    {
      "epoch": 0.1748511671067425,
      "grad_norm": 0.09856017678976059,
      "learning_rate": 8.284969309351435e-05,
      "loss": 1.6558,
      "step": 88608
    },
    {
      "epoch": 0.17491431306813893,
      "grad_norm": 0.1065458208322525,
      "learning_rate": 8.284335286914966e-05,
      "loss": 1.6535,
      "step": 88640
    },
    {
      "epoch": 0.17497745902953538,
      "grad_norm": 0.10433677583932877,
      "learning_rate": 8.283701264478498e-05,
      "loss": 1.6422,
      "step": 88672
    },
    {
      "epoch": 0.17504060499093183,
      "grad_norm": 0.11038319021463394,
      "learning_rate": 8.283067242042028e-05,
      "loss": 1.6613,
      "step": 88704
    },
    {
      "epoch": 0.17510375095232825,
      "grad_norm": 0.10992168635129929,
      "learning_rate": 8.282433219605558e-05,
      "loss": 1.6723,
      "step": 88736
    },
    {
      "epoch": 0.1751668969137247,
      "grad_norm": 0.10349271446466446,
      "learning_rate": 8.28179919716909e-05,
      "loss": 1.6745,
      "step": 88768
    },
    {
      "epoch": 0.17523004287512114,
      "grad_norm": 0.10459519922733307,
      "learning_rate": 8.281165174732621e-05,
      "loss": 1.6475,
      "step": 88800
    },
    {
      "epoch": 0.17529318883651757,
      "grad_norm": 0.10314363241195679,
      "learning_rate": 8.280531152296152e-05,
      "loss": 1.6568,
      "step": 88832
    },
    {
      "epoch": 0.17535633479791402,
      "grad_norm": 0.10256712138652802,
      "learning_rate": 8.279897129859683e-05,
      "loss": 1.6794,
      "step": 88864
    },
    {
      "epoch": 0.17541948075931046,
      "grad_norm": 0.09958847612142563,
      "learning_rate": 8.279263107423214e-05,
      "loss": 1.6492,
      "step": 88896
    },
    {
      "epoch": 0.17548262672070689,
      "grad_norm": 0.0999009907245636,
      "learning_rate": 8.278629084986745e-05,
      "loss": 1.6571,
      "step": 88928
    },
    {
      "epoch": 0.17554577268210333,
      "grad_norm": 0.09854394942522049,
      "learning_rate": 8.277995062550277e-05,
      "loss": 1.6457,
      "step": 88960
    },
    {
      "epoch": 0.17560891864349978,
      "grad_norm": 0.10845145583152771,
      "learning_rate": 8.277361040113808e-05,
      "loss": 1.6399,
      "step": 88992
    },
    {
      "epoch": 0.1756720646048962,
      "grad_norm": 0.1146244928240776,
      "learning_rate": 8.276727017677338e-05,
      "loss": 1.6602,
      "step": 89024
    },
    {
      "epoch": 0.17573521056629265,
      "grad_norm": 0.11173972487449646,
      "learning_rate": 8.27609299524087e-05,
      "loss": 1.654,
      "step": 89056
    },
    {
      "epoch": 0.1757983565276891,
      "grad_norm": 0.10688459128141403,
      "learning_rate": 8.275458972804401e-05,
      "loss": 1.6638,
      "step": 89088
    },
    {
      "epoch": 0.17586150248908552,
      "grad_norm": 0.09491217881441116,
      "learning_rate": 8.274824950367932e-05,
      "loss": 1.6491,
      "step": 89120
    },
    {
      "epoch": 0.17592464845048197,
      "grad_norm": 0.10553118586540222,
      "learning_rate": 8.274190927931462e-05,
      "loss": 1.6748,
      "step": 89152
    },
    {
      "epoch": 0.17598779441187842,
      "grad_norm": 0.10010967403650284,
      "learning_rate": 8.273556905494993e-05,
      "loss": 1.6575,
      "step": 89184
    },
    {
      "epoch": 0.17605094037327484,
      "grad_norm": 0.10946615785360336,
      "learning_rate": 8.272922883058525e-05,
      "loss": 1.6682,
      "step": 89216
    },
    {
      "epoch": 0.1761140863346713,
      "grad_norm": 0.09834275394678116,
      "learning_rate": 8.272288860622056e-05,
      "loss": 1.6632,
      "step": 89248
    },
    {
      "epoch": 0.17617723229606774,
      "grad_norm": 0.09829218685626984,
      "learning_rate": 8.271654838185586e-05,
      "loss": 1.6543,
      "step": 89280
    },
    {
      "epoch": 0.17624037825746416,
      "grad_norm": 0.10909516364336014,
      "learning_rate": 8.271020815749118e-05,
      "loss": 1.6566,
      "step": 89312
    },
    {
      "epoch": 0.1763035242188606,
      "grad_norm": 0.09997043758630753,
      "learning_rate": 8.270386793312649e-05,
      "loss": 1.6619,
      "step": 89344
    },
    {
      "epoch": 0.17636667018025706,
      "grad_norm": 0.09957852214574814,
      "learning_rate": 8.26975277087618e-05,
      "loss": 1.6583,
      "step": 89376
    },
    {
      "epoch": 0.17642981614165348,
      "grad_norm": 0.10366418957710266,
      "learning_rate": 8.26911874843971e-05,
      "loss": 1.6503,
      "step": 89408
    },
    {
      "epoch": 0.17649296210304993,
      "grad_norm": 0.10896145552396774,
      "learning_rate": 8.268484726003242e-05,
      "loss": 1.6603,
      "step": 89440
    },
    {
      "epoch": 0.17655610806444638,
      "grad_norm": 0.1021653488278389,
      "learning_rate": 8.267850703566773e-05,
      "loss": 1.671,
      "step": 89472
    },
    {
      "epoch": 0.1766192540258428,
      "grad_norm": 0.0990075170993805,
      "learning_rate": 8.267216681130304e-05,
      "loss": 1.6483,
      "step": 89504
    },
    {
      "epoch": 0.17668239998723925,
      "grad_norm": 0.09864112734794617,
      "learning_rate": 8.266582658693835e-05,
      "loss": 1.6608,
      "step": 89536
    },
    {
      "epoch": 0.1767455459486357,
      "grad_norm": 0.10186667740345001,
      "learning_rate": 8.265948636257365e-05,
      "loss": 1.6636,
      "step": 89568
    },
    {
      "epoch": 0.17680869191003212,
      "grad_norm": 0.10535477846860886,
      "learning_rate": 8.265314613820897e-05,
      "loss": 1.6502,
      "step": 89600
    },
    {
      "epoch": 0.17687183787142857,
      "grad_norm": 0.10260170698165894,
      "learning_rate": 8.264680591384428e-05,
      "loss": 1.6501,
      "step": 89632
    },
    {
      "epoch": 0.17693498383282502,
      "grad_norm": 0.10581181943416595,
      "learning_rate": 8.26404656894796e-05,
      "loss": 1.6527,
      "step": 89664
    },
    {
      "epoch": 0.17699812979422144,
      "grad_norm": 0.10850777477025986,
      "learning_rate": 8.26341254651149e-05,
      "loss": 1.64,
      "step": 89696
    },
    {
      "epoch": 0.1770612757556179,
      "grad_norm": 0.11619777232408524,
      "learning_rate": 8.262778524075021e-05,
      "loss": 1.6551,
      "step": 89728
    },
    {
      "epoch": 0.17712442171701434,
      "grad_norm": 0.09978790581226349,
      "learning_rate": 8.262144501638552e-05,
      "loss": 1.6633,
      "step": 89760
    },
    {
      "epoch": 0.17718756767841076,
      "grad_norm": 0.10603735595941544,
      "learning_rate": 8.261510479202084e-05,
      "loss": 1.6648,
      "step": 89792
    },
    {
      "epoch": 0.1772507136398072,
      "grad_norm": 0.10047462582588196,
      "learning_rate": 8.260876456765614e-05,
      "loss": 1.6599,
      "step": 89824
    },
    {
      "epoch": 0.17731385960120366,
      "grad_norm": 0.0955331027507782,
      "learning_rate": 8.260242434329146e-05,
      "loss": 1.6526,
      "step": 89856
    },
    {
      "epoch": 0.17737700556260008,
      "grad_norm": 0.10489872097969055,
      "learning_rate": 8.259608411892677e-05,
      "loss": 1.6587,
      "step": 89888
    },
    {
      "epoch": 0.17744015152399653,
      "grad_norm": 0.09369321912527084,
      "learning_rate": 8.258974389456207e-05,
      "loss": 1.6498,
      "step": 89920
    },
    {
      "epoch": 0.17750329748539298,
      "grad_norm": 0.10730299353599548,
      "learning_rate": 8.258340367019739e-05,
      "loss": 1.6617,
      "step": 89952
    },
    {
      "epoch": 0.1775664434467894,
      "grad_norm": 0.0987553671002388,
      "learning_rate": 8.257706344583269e-05,
      "loss": 1.6505,
      "step": 89984
    },
    {
      "epoch": 0.17762958940818585,
      "grad_norm": 0.09921061247587204,
      "learning_rate": 8.2570723221468e-05,
      "loss": 1.6546,
      "step": 90016
    },
    {
      "epoch": 0.1776927353695823,
      "grad_norm": 0.10513906925916672,
      "learning_rate": 8.256438299710332e-05,
      "loss": 1.672,
      "step": 90048
    },
    {
      "epoch": 0.17775588133097872,
      "grad_norm": 0.09985551983118057,
      "learning_rate": 8.255804277273862e-05,
      "loss": 1.6495,
      "step": 90080
    },
    {
      "epoch": 0.17781902729237517,
      "grad_norm": 0.10031275451183319,
      "learning_rate": 8.255170254837393e-05,
      "loss": 1.6528,
      "step": 90112
    },
    {
      "epoch": 0.17788217325377162,
      "grad_norm": 0.10402368754148483,
      "learning_rate": 8.254536232400925e-05,
      "loss": 1.6631,
      "step": 90144
    },
    {
      "epoch": 0.17794531921516804,
      "grad_norm": 0.10302762687206268,
      "learning_rate": 8.253902209964456e-05,
      "loss": 1.6448,
      "step": 90176
    },
    {
      "epoch": 0.1780084651765645,
      "grad_norm": 0.09971826523542404,
      "learning_rate": 8.253268187527986e-05,
      "loss": 1.6439,
      "step": 90208
    },
    {
      "epoch": 0.17807161113796094,
      "grad_norm": 0.10372737050056458,
      "learning_rate": 8.252634165091518e-05,
      "loss": 1.654,
      "step": 90240
    },
    {
      "epoch": 0.1781347570993574,
      "grad_norm": 0.10416607558727264,
      "learning_rate": 8.252000142655049e-05,
      "loss": 1.6666,
      "step": 90272
    },
    {
      "epoch": 0.1781979030607538,
      "grad_norm": 0.10966816544532776,
      "learning_rate": 8.25136612021858e-05,
      "loss": 1.6614,
      "step": 90304
    },
    {
      "epoch": 0.17826104902215026,
      "grad_norm": 0.10601796954870224,
      "learning_rate": 8.25073209778211e-05,
      "loss": 1.6616,
      "step": 90336
    },
    {
      "epoch": 0.1783241949835467,
      "grad_norm": 0.10499513894319534,
      "learning_rate": 8.250098075345642e-05,
      "loss": 1.647,
      "step": 90368
    },
    {
      "epoch": 0.17838734094494313,
      "grad_norm": 0.10670746117830276,
      "learning_rate": 8.249464052909172e-05,
      "loss": 1.6515,
      "step": 90400
    },
    {
      "epoch": 0.17845048690633958,
      "grad_norm": 0.10835777968168259,
      "learning_rate": 8.248830030472704e-05,
      "loss": 1.6452,
      "step": 90432
    },
    {
      "epoch": 0.17851363286773603,
      "grad_norm": 0.10320121049880981,
      "learning_rate": 8.248196008036235e-05,
      "loss": 1.6503,
      "step": 90464
    },
    {
      "epoch": 0.17857677882913245,
      "grad_norm": 0.09848032146692276,
      "learning_rate": 8.247561985599765e-05,
      "loss": 1.6581,
      "step": 90496
    },
    {
      "epoch": 0.1786399247905289,
      "grad_norm": 0.10776699334383011,
      "learning_rate": 8.246927963163297e-05,
      "loss": 1.6595,
      "step": 90528
    },
    {
      "epoch": 0.17870307075192535,
      "grad_norm": 0.10353954136371613,
      "learning_rate": 8.246293940726828e-05,
      "loss": 1.6547,
      "step": 90560
    },
    {
      "epoch": 0.17876621671332177,
      "grad_norm": 0.10759712755680084,
      "learning_rate": 8.24565991829036e-05,
      "loss": 1.6559,
      "step": 90592
    },
    {
      "epoch": 0.17882936267471822,
      "grad_norm": 0.09907542914152145,
      "learning_rate": 8.24502589585389e-05,
      "loss": 1.6543,
      "step": 90624
    },
    {
      "epoch": 0.17889250863611467,
      "grad_norm": 0.11011697351932526,
      "learning_rate": 8.244391873417421e-05,
      "loss": 1.6623,
      "step": 90656
    },
    {
      "epoch": 0.1789556545975111,
      "grad_norm": 0.09947836399078369,
      "learning_rate": 8.243757850980953e-05,
      "loss": 1.6707,
      "step": 90688
    },
    {
      "epoch": 0.17901880055890754,
      "grad_norm": 0.10960274189710617,
      "learning_rate": 8.243123828544484e-05,
      "loss": 1.6604,
      "step": 90720
    },
    {
      "epoch": 0.179081946520304,
      "grad_norm": 0.10730334371328354,
      "learning_rate": 8.242489806108014e-05,
      "loss": 1.6465,
      "step": 90752
    },
    {
      "epoch": 0.1791450924817004,
      "grad_norm": 0.0949479416012764,
      "learning_rate": 8.241855783671546e-05,
      "loss": 1.6613,
      "step": 90784
    },
    {
      "epoch": 0.17920823844309686,
      "grad_norm": 0.09434281289577484,
      "learning_rate": 8.241221761235076e-05,
      "loss": 1.6629,
      "step": 90816
    },
    {
      "epoch": 0.1792713844044933,
      "grad_norm": 0.10668272525072098,
      "learning_rate": 8.240587738798607e-05,
      "loss": 1.6484,
      "step": 90848
    },
    {
      "epoch": 0.17933453036588973,
      "grad_norm": 0.104609914124012,
      "learning_rate": 8.239953716362137e-05,
      "loss": 1.6483,
      "step": 90880
    },
    {
      "epoch": 0.17939767632728618,
      "grad_norm": 0.10034120082855225,
      "learning_rate": 8.239319693925669e-05,
      "loss": 1.6598,
      "step": 90912
    },
    {
      "epoch": 0.17946082228868263,
      "grad_norm": 0.10214868187904358,
      "learning_rate": 8.2386856714892e-05,
      "loss": 1.6564,
      "step": 90944
    },
    {
      "epoch": 0.17952396825007905,
      "grad_norm": 0.10459335148334503,
      "learning_rate": 8.238051649052732e-05,
      "loss": 1.6404,
      "step": 90976
    },
    {
      "epoch": 0.1795871142114755,
      "grad_norm": 0.09940114617347717,
      "learning_rate": 8.237417626616263e-05,
      "loss": 1.6498,
      "step": 91008
    },
    {
      "epoch": 0.17965026017287195,
      "grad_norm": 0.10567575693130493,
      "learning_rate": 8.236783604179793e-05,
      "loss": 1.6629,
      "step": 91040
    },
    {
      "epoch": 0.17971340613426837,
      "grad_norm": 0.10702252388000488,
      "learning_rate": 8.236149581743325e-05,
      "loss": 1.6512,
      "step": 91072
    },
    {
      "epoch": 0.17977655209566482,
      "grad_norm": 0.10372652113437653,
      "learning_rate": 8.235515559306856e-05,
      "loss": 1.6443,
      "step": 91104
    },
    {
      "epoch": 0.17983969805706126,
      "grad_norm": 0.10476701706647873,
      "learning_rate": 8.234881536870387e-05,
      "loss": 1.655,
      "step": 91136
    },
    {
      "epoch": 0.17990284401845769,
      "grad_norm": 0.10643497109413147,
      "learning_rate": 8.234247514433918e-05,
      "loss": 1.6516,
      "step": 91168
    },
    {
      "epoch": 0.17996598997985414,
      "grad_norm": 0.10193724185228348,
      "learning_rate": 8.233613491997448e-05,
      "loss": 1.6629,
      "step": 91200
    },
    {
      "epoch": 0.18002913594125058,
      "grad_norm": 0.10508453100919724,
      "learning_rate": 8.232979469560979e-05,
      "loss": 1.6526,
      "step": 91232
    },
    {
      "epoch": 0.180092281902647,
      "grad_norm": 0.09920556098222733,
      "learning_rate": 8.23234544712451e-05,
      "loss": 1.6447,
      "step": 91264
    },
    {
      "epoch": 0.18015542786404345,
      "grad_norm": 0.10268726199865341,
      "learning_rate": 8.231711424688041e-05,
      "loss": 1.651,
      "step": 91296
    },
    {
      "epoch": 0.1802185738254399,
      "grad_norm": 0.10527651011943817,
      "learning_rate": 8.231077402251572e-05,
      "loss": 1.6539,
      "step": 91328
    },
    {
      "epoch": 0.18028171978683633,
      "grad_norm": 0.10914020985364914,
      "learning_rate": 8.230443379815104e-05,
      "loss": 1.6589,
      "step": 91360
    },
    {
      "epoch": 0.18034486574823277,
      "grad_norm": 0.10070778429508209,
      "learning_rate": 8.229809357378635e-05,
      "loss": 1.6443,
      "step": 91392
    },
    {
      "epoch": 0.18040801170962922,
      "grad_norm": 0.09839003533124924,
      "learning_rate": 8.229175334942165e-05,
      "loss": 1.6563,
      "step": 91424
    },
    {
      "epoch": 0.18047115767102564,
      "grad_norm": 0.10623380541801453,
      "learning_rate": 8.228541312505697e-05,
      "loss": 1.6433,
      "step": 91456
    },
    {
      "epoch": 0.1805343036324221,
      "grad_norm": 0.1055963784456253,
      "learning_rate": 8.227907290069228e-05,
      "loss": 1.649,
      "step": 91488
    },
    {
      "epoch": 0.18059744959381854,
      "grad_norm": 0.10425406694412231,
      "learning_rate": 8.22727326763276e-05,
      "loss": 1.6502,
      "step": 91520
    },
    {
      "epoch": 0.18066059555521496,
      "grad_norm": 0.10616173595190048,
      "learning_rate": 8.22663924519629e-05,
      "loss": 1.6495,
      "step": 91552
    },
    {
      "epoch": 0.1807237415166114,
      "grad_norm": 0.09970536828041077,
      "learning_rate": 8.226005222759821e-05,
      "loss": 1.6535,
      "step": 91584
    },
    {
      "epoch": 0.18078688747800786,
      "grad_norm": 0.10234661400318146,
      "learning_rate": 8.225371200323351e-05,
      "loss": 1.6435,
      "step": 91616
    },
    {
      "epoch": 0.18085003343940428,
      "grad_norm": 0.10297203809022903,
      "learning_rate": 8.224737177886883e-05,
      "loss": 1.6538,
      "step": 91648
    },
    {
      "epoch": 0.18091317940080073,
      "grad_norm": 0.10357316583395004,
      "learning_rate": 8.224103155450414e-05,
      "loss": 1.6466,
      "step": 91680
    },
    {
      "epoch": 0.18097632536219718,
      "grad_norm": 0.10353853553533554,
      "learning_rate": 8.223469133013944e-05,
      "loss": 1.6504,
      "step": 91712
    },
    {
      "epoch": 0.1810394713235936,
      "grad_norm": 0.10125371068716049,
      "learning_rate": 8.222835110577476e-05,
      "loss": 1.6519,
      "step": 91744
    },
    {
      "epoch": 0.18110261728499005,
      "grad_norm": 0.1040688008069992,
      "learning_rate": 8.222201088141007e-05,
      "loss": 1.6606,
      "step": 91776
    },
    {
      "epoch": 0.1811657632463865,
      "grad_norm": 0.11518832296133041,
      "learning_rate": 8.221567065704539e-05,
      "loss": 1.6491,
      "step": 91808
    },
    {
      "epoch": 0.18122890920778292,
      "grad_norm": 0.09601999074220657,
      "learning_rate": 8.220933043268069e-05,
      "loss": 1.6473,
      "step": 91840
    },
    {
      "epoch": 0.18129205516917937,
      "grad_norm": 0.10341492295265198,
      "learning_rate": 8.2202990208316e-05,
      "loss": 1.6492,
      "step": 91872
    },
    {
      "epoch": 0.18135520113057582,
      "grad_norm": 0.10092576593160629,
      "learning_rate": 8.219664998395132e-05,
      "loss": 1.6522,
      "step": 91904
    },
    {
      "epoch": 0.18141834709197227,
      "grad_norm": 0.09618775546550751,
      "learning_rate": 8.219030975958663e-05,
      "loss": 1.6568,
      "step": 91936
    },
    {
      "epoch": 0.1814814930533687,
      "grad_norm": 0.10186668485403061,
      "learning_rate": 8.218396953522193e-05,
      "loss": 1.6578,
      "step": 91968
    },
    {
      "epoch": 0.18154463901476514,
      "grad_norm": 0.1145521029829979,
      "learning_rate": 8.217762931085725e-05,
      "loss": 1.6568,
      "step": 92000
    },
    {
      "epoch": 0.1816077849761616,
      "grad_norm": 0.09803149849176407,
      "learning_rate": 8.217128908649255e-05,
      "loss": 1.6666,
      "step": 92032
    },
    {
      "epoch": 0.181670930937558,
      "grad_norm": 0.09554032236337662,
      "learning_rate": 8.216494886212786e-05,
      "loss": 1.6577,
      "step": 92064
    },
    {
      "epoch": 0.18173407689895446,
      "grad_norm": 0.11648745089769363,
      "learning_rate": 8.215860863776316e-05,
      "loss": 1.6508,
      "step": 92096
    },
    {
      "epoch": 0.1817972228603509,
      "grad_norm": 0.10111765563488007,
      "learning_rate": 8.215226841339848e-05,
      "loss": 1.653,
      "step": 92128
    },
    {
      "epoch": 0.18186036882174733,
      "grad_norm": 0.10026899725198746,
      "learning_rate": 8.214592818903379e-05,
      "loss": 1.637,
      "step": 92160
    },
    {
      "epoch": 0.18192351478314378,
      "grad_norm": 0.10087400674819946,
      "learning_rate": 8.21395879646691e-05,
      "loss": 1.6597,
      "step": 92192
    },
    {
      "epoch": 0.18198666074454023,
      "grad_norm": 0.10220714658498764,
      "learning_rate": 8.213324774030441e-05,
      "loss": 1.6447,
      "step": 92224
    },
    {
      "epoch": 0.18204980670593665,
      "grad_norm": 0.10861620306968689,
      "learning_rate": 8.212690751593972e-05,
      "loss": 1.6596,
      "step": 92256
    },
    {
      "epoch": 0.1821129526673331,
      "grad_norm": 0.11353089660406113,
      "learning_rate": 8.212056729157504e-05,
      "loss": 1.6526,
      "step": 92288
    },
    {
      "epoch": 0.18217609862872955,
      "grad_norm": 0.1047775149345398,
      "learning_rate": 8.211422706721035e-05,
      "loss": 1.649,
      "step": 92320
    },
    {
      "epoch": 0.18223924459012597,
      "grad_norm": 0.10837597399950027,
      "learning_rate": 8.210788684284567e-05,
      "loss": 1.6404,
      "step": 92352
    },
    {
      "epoch": 0.18230239055152242,
      "grad_norm": 0.10726919025182724,
      "learning_rate": 8.210154661848097e-05,
      "loss": 1.6497,
      "step": 92384
    },
    {
      "epoch": 0.18236553651291887,
      "grad_norm": 0.09631171822547913,
      "learning_rate": 8.209520639411628e-05,
      "loss": 1.6521,
      "step": 92416
    },
    {
      "epoch": 0.1824286824743153,
      "grad_norm": 0.10086216777563095,
      "learning_rate": 8.208886616975158e-05,
      "loss": 1.6444,
      "step": 92448
    },
    {
      "epoch": 0.18249182843571174,
      "grad_norm": 0.10379354655742645,
      "learning_rate": 8.20825259453869e-05,
      "loss": 1.64,
      "step": 92480
    },
    {
      "epoch": 0.1825549743971082,
      "grad_norm": 0.1042555496096611,
      "learning_rate": 8.20761857210222e-05,
      "loss": 1.662,
      "step": 92512
    },
    {
      "epoch": 0.1826181203585046,
      "grad_norm": 0.09675726294517517,
      "learning_rate": 8.206984549665751e-05,
      "loss": 1.6483,
      "step": 92544
    },
    {
      "epoch": 0.18268126631990106,
      "grad_norm": 0.10146626085042953,
      "learning_rate": 8.206350527229283e-05,
      "loss": 1.6493,
      "step": 92576
    },
    {
      "epoch": 0.1827444122812975,
      "grad_norm": 0.10710596293210983,
      "learning_rate": 8.205716504792814e-05,
      "loss": 1.6514,
      "step": 92608
    },
    {
      "epoch": 0.18280755824269393,
      "grad_norm": 0.10241590440273285,
      "learning_rate": 8.205082482356344e-05,
      "loss": 1.6461,
      "step": 92640
    },
    {
      "epoch": 0.18287070420409038,
      "grad_norm": 0.10124043375253677,
      "learning_rate": 8.204448459919876e-05,
      "loss": 1.6521,
      "step": 92672
    },
    {
      "epoch": 0.18293385016548683,
      "grad_norm": 0.10410934686660767,
      "learning_rate": 8.203814437483407e-05,
      "loss": 1.655,
      "step": 92704
    },
    {
      "epoch": 0.18299699612688325,
      "grad_norm": 0.10063810646533966,
      "learning_rate": 8.203180415046939e-05,
      "loss": 1.6452,
      "step": 92736
    },
    {
      "epoch": 0.1830601420882797,
      "grad_norm": 0.10326386988162994,
      "learning_rate": 8.202546392610469e-05,
      "loss": 1.6671,
      "step": 92768
    },
    {
      "epoch": 0.18312328804967615,
      "grad_norm": 0.10978447645902634,
      "learning_rate": 8.201912370174e-05,
      "loss": 1.6435,
      "step": 92800
    },
    {
      "epoch": 0.18318643401107257,
      "grad_norm": 0.10418623685836792,
      "learning_rate": 8.201278347737532e-05,
      "loss": 1.6475,
      "step": 92832
    },
    {
      "epoch": 0.18324957997246902,
      "grad_norm": 0.10795113444328308,
      "learning_rate": 8.200644325301062e-05,
      "loss": 1.6503,
      "step": 92864
    },
    {
      "epoch": 0.18331272593386547,
      "grad_norm": 0.10588126629590988,
      "learning_rate": 8.200010302864592e-05,
      "loss": 1.6523,
      "step": 92896
    },
    {
      "epoch": 0.1833758718952619,
      "grad_norm": 0.1011577844619751,
      "learning_rate": 8.199376280428123e-05,
      "loss": 1.6487,
      "step": 92928
    },
    {
      "epoch": 0.18343901785665834,
      "grad_norm": 0.1028437688946724,
      "learning_rate": 8.198742257991655e-05,
      "loss": 1.6524,
      "step": 92960
    },
    {
      "epoch": 0.1835021638180548,
      "grad_norm": 0.1025761291384697,
      "learning_rate": 8.198108235555186e-05,
      "loss": 1.6411,
      "step": 92992
    },
    {
      "epoch": 0.1835653097794512,
      "grad_norm": 0.10055080056190491,
      "learning_rate": 8.197474213118718e-05,
      "loss": 1.6431,
      "step": 93024
    },
    {
      "epoch": 0.18362845574084766,
      "grad_norm": 0.1111232340335846,
      "learning_rate": 8.196840190682248e-05,
      "loss": 1.6582,
      "step": 93056
    },
    {
      "epoch": 0.1836916017022441,
      "grad_norm": 0.10614980012178421,
      "learning_rate": 8.196206168245779e-05,
      "loss": 1.6341,
      "step": 93088
    },
    {
      "epoch": 0.18375474766364053,
      "grad_norm": 0.09853687882423401,
      "learning_rate": 8.195572145809311e-05,
      "loss": 1.6498,
      "step": 93120
    },
    {
      "epoch": 0.18381789362503698,
      "grad_norm": 0.10568626970052719,
      "learning_rate": 8.194938123372842e-05,
      "loss": 1.6558,
      "step": 93152
    },
    {
      "epoch": 0.18388103958643343,
      "grad_norm": 0.1024804636836052,
      "learning_rate": 8.194304100936372e-05,
      "loss": 1.6557,
      "step": 93184
    },
    {
      "epoch": 0.18394418554782985,
      "grad_norm": 0.10066686570644379,
      "learning_rate": 8.193670078499904e-05,
      "loss": 1.6621,
      "step": 93216
    },
    {
      "epoch": 0.1840073315092263,
      "grad_norm": 0.10116623342037201,
      "learning_rate": 8.193036056063435e-05,
      "loss": 1.6483,
      "step": 93248
    },
    {
      "epoch": 0.18407047747062275,
      "grad_norm": 0.09657549113035202,
      "learning_rate": 8.192402033626965e-05,
      "loss": 1.6482,
      "step": 93280
    },
    {
      "epoch": 0.18413362343201917,
      "grad_norm": 0.09711366891860962,
      "learning_rate": 8.191768011190495e-05,
      "loss": 1.6507,
      "step": 93312
    },
    {
      "epoch": 0.18419676939341562,
      "grad_norm": 0.10164396464824677,
      "learning_rate": 8.191133988754027e-05,
      "loss": 1.6527,
      "step": 93344
    },
    {
      "epoch": 0.18425991535481206,
      "grad_norm": 0.10205285996198654,
      "learning_rate": 8.190499966317558e-05,
      "loss": 1.6543,
      "step": 93376
    },
    {
      "epoch": 0.1843230613162085,
      "grad_norm": 0.09781279414892197,
      "learning_rate": 8.18986594388109e-05,
      "loss": 1.6514,
      "step": 93408
    },
    {
      "epoch": 0.18438620727760494,
      "grad_norm": 0.10196979343891144,
      "learning_rate": 8.18923192144462e-05,
      "loss": 1.6664,
      "step": 93440
    },
    {
      "epoch": 0.18444935323900138,
      "grad_norm": 0.10308445990085602,
      "learning_rate": 8.188597899008151e-05,
      "loss": 1.6408,
      "step": 93472
    },
    {
      "epoch": 0.1845124992003978,
      "grad_norm": 0.11130218207836151,
      "learning_rate": 8.187963876571683e-05,
      "loss": 1.6604,
      "step": 93504
    },
    {
      "epoch": 0.18457564516179426,
      "grad_norm": 0.10474035143852234,
      "learning_rate": 8.187329854135214e-05,
      "loss": 1.6585,
      "step": 93536
    },
    {
      "epoch": 0.1846387911231907,
      "grad_norm": 0.09855648875236511,
      "learning_rate": 8.186695831698744e-05,
      "loss": 1.651,
      "step": 93568
    },
    {
      "epoch": 0.18470193708458715,
      "grad_norm": 0.1026657447218895,
      "learning_rate": 8.186061809262276e-05,
      "loss": 1.6564,
      "step": 93600
    },
    {
      "epoch": 0.18476508304598357,
      "grad_norm": 0.1057642474770546,
      "learning_rate": 8.185427786825807e-05,
      "loss": 1.6538,
      "step": 93632
    },
    {
      "epoch": 0.18482822900738002,
      "grad_norm": 0.09877024590969086,
      "learning_rate": 8.184793764389339e-05,
      "loss": 1.6652,
      "step": 93664
    },
    {
      "epoch": 0.18489137496877647,
      "grad_norm": 0.10119491070508957,
      "learning_rate": 8.184159741952869e-05,
      "loss": 1.656,
      "step": 93696
    },
    {
      "epoch": 0.1849545209301729,
      "grad_norm": 0.1023765280842781,
      "learning_rate": 8.183525719516399e-05,
      "loss": 1.6511,
      "step": 93728
    },
    {
      "epoch": 0.18501766689156934,
      "grad_norm": 0.10401959717273712,
      "learning_rate": 8.18289169707993e-05,
      "loss": 1.6505,
      "step": 93760
    },
    {
      "epoch": 0.1850808128529658,
      "grad_norm": 0.10199481248855591,
      "learning_rate": 8.182257674643462e-05,
      "loss": 1.6528,
      "step": 93792
    },
    {
      "epoch": 0.18514395881436221,
      "grad_norm": 0.10202162712812424,
      "learning_rate": 8.181623652206993e-05,
      "loss": 1.6555,
      "step": 93824
    },
    {
      "epoch": 0.18520710477575866,
      "grad_norm": 0.10206763446331024,
      "learning_rate": 8.180989629770523e-05,
      "loss": 1.6496,
      "step": 93856
    },
    {
      "epoch": 0.1852702507371551,
      "grad_norm": 0.11224686354398727,
      "learning_rate": 8.180355607334055e-05,
      "loss": 1.6544,
      "step": 93888
    },
    {
      "epoch": 0.18533339669855153,
      "grad_norm": 0.09769473224878311,
      "learning_rate": 8.179721584897586e-05,
      "loss": 1.6499,
      "step": 93920
    },
    {
      "epoch": 0.18539654265994798,
      "grad_norm": 0.10412193089723587,
      "learning_rate": 8.179087562461118e-05,
      "loss": 1.6426,
      "step": 93952
    },
    {
      "epoch": 0.18545968862134443,
      "grad_norm": 0.09697689861059189,
      "learning_rate": 8.178453540024648e-05,
      "loss": 1.6478,
      "step": 93984
    },
    {
      "epoch": 0.18552283458274085,
      "grad_norm": 0.10029570013284683,
      "learning_rate": 8.177819517588179e-05,
      "loss": 1.6487,
      "step": 94016
    },
    {
      "epoch": 0.1855859805441373,
      "grad_norm": 0.10300135612487793,
      "learning_rate": 8.177185495151711e-05,
      "loss": 1.6592,
      "step": 94048
    },
    {
      "epoch": 0.18564912650553375,
      "grad_norm": 0.09986173361539841,
      "learning_rate": 8.176551472715241e-05,
      "loss": 1.6507,
      "step": 94080
    },
    {
      "epoch": 0.18571227246693017,
      "grad_norm": 0.10211902111768723,
      "learning_rate": 8.175917450278772e-05,
      "loss": 1.6558,
      "step": 94112
    },
    {
      "epoch": 0.18577541842832662,
      "grad_norm": 0.1036137118935585,
      "learning_rate": 8.175283427842302e-05,
      "loss": 1.6543,
      "step": 94144
    },
    {
      "epoch": 0.18583856438972307,
      "grad_norm": 0.09529624879360199,
      "learning_rate": 8.174649405405834e-05,
      "loss": 1.6417,
      "step": 94176
    },
    {
      "epoch": 0.1859017103511195,
      "grad_norm": 0.10255629569292068,
      "learning_rate": 8.174015382969365e-05,
      "loss": 1.639,
      "step": 94208
    },
    {
      "epoch": 0.18596485631251594,
      "grad_norm": 0.10615867376327515,
      "learning_rate": 8.173381360532895e-05,
      "loss": 1.6408,
      "step": 94240
    },
    {
      "epoch": 0.1860280022739124,
      "grad_norm": 0.1062152087688446,
      "learning_rate": 8.172747338096427e-05,
      "loss": 1.6544,
      "step": 94272
    },
    {
      "epoch": 0.1860911482353088,
      "grad_norm": 0.0993291586637497,
      "learning_rate": 8.172113315659958e-05,
      "loss": 1.6549,
      "step": 94304
    },
    {
      "epoch": 0.18615429419670526,
      "grad_norm": 0.10131142288446426,
      "learning_rate": 8.17147929322349e-05,
      "loss": 1.656,
      "step": 94336
    },
    {
      "epoch": 0.1862174401581017,
      "grad_norm": 0.11048464477062225,
      "learning_rate": 8.170845270787021e-05,
      "loss": 1.6445,
      "step": 94368
    },
    {
      "epoch": 0.18628058611949813,
      "grad_norm": 0.10709990561008453,
      "learning_rate": 8.170211248350551e-05,
      "loss": 1.6411,
      "step": 94400
    },
    {
      "epoch": 0.18634373208089458,
      "grad_norm": 0.10426360368728638,
      "learning_rate": 8.169577225914083e-05,
      "loss": 1.6504,
      "step": 94432
    },
    {
      "epoch": 0.18640687804229103,
      "grad_norm": 0.10206352174282074,
      "learning_rate": 8.168943203477614e-05,
      "loss": 1.6369,
      "step": 94464
    },
    {
      "epoch": 0.18647002400368745,
      "grad_norm": 0.11030671745538712,
      "learning_rate": 8.168309181041144e-05,
      "loss": 1.6452,
      "step": 94496
    },
    {
      "epoch": 0.1865331699650839,
      "grad_norm": 0.100029356777668,
      "learning_rate": 8.167675158604676e-05,
      "loss": 1.6405,
      "step": 94528
    },
    {
      "epoch": 0.18659631592648035,
      "grad_norm": 0.10335873812437057,
      "learning_rate": 8.167041136168206e-05,
      "loss": 1.657,
      "step": 94560
    },
    {
      "epoch": 0.18665946188787677,
      "grad_norm": 0.10152248293161392,
      "learning_rate": 8.166407113731737e-05,
      "loss": 1.6485,
      "step": 94592
    },
    {
      "epoch": 0.18672260784927322,
      "grad_norm": 0.09640410542488098,
      "learning_rate": 8.165773091295269e-05,
      "loss": 1.6314,
      "step": 94624
    },
    {
      "epoch": 0.18678575381066967,
      "grad_norm": 0.10848820954561234,
      "learning_rate": 8.165139068858799e-05,
      "loss": 1.6466,
      "step": 94656
    },
    {
      "epoch": 0.1868488997720661,
      "grad_norm": 0.10458540171384811,
      "learning_rate": 8.16450504642233e-05,
      "loss": 1.6482,
      "step": 94688
    },
    {
      "epoch": 0.18691204573346254,
      "grad_norm": 0.10645361989736557,
      "learning_rate": 8.163871023985862e-05,
      "loss": 1.6547,
      "step": 94720
    },
    {
      "epoch": 0.186975191694859,
      "grad_norm": 0.10136350989341736,
      "learning_rate": 8.163237001549393e-05,
      "loss": 1.6428,
      "step": 94752
    },
    {
      "epoch": 0.1870383376562554,
      "grad_norm": 0.10477007925510406,
      "learning_rate": 8.162602979112923e-05,
      "loss": 1.6424,
      "step": 94784
    },
    {
      "epoch": 0.18710148361765186,
      "grad_norm": 0.11905611306428909,
      "learning_rate": 8.161968956676455e-05,
      "loss": 1.6503,
      "step": 94816
    },
    {
      "epoch": 0.1871646295790483,
      "grad_norm": 0.09771967679262161,
      "learning_rate": 8.161334934239986e-05,
      "loss": 1.6336,
      "step": 94848
    },
    {
      "epoch": 0.18722777554044473,
      "grad_norm": 0.10688287019729614,
      "learning_rate": 8.160700911803518e-05,
      "loss": 1.6517,
      "step": 94880
    },
    {
      "epoch": 0.18729092150184118,
      "grad_norm": 0.09580983966588974,
      "learning_rate": 8.160066889367048e-05,
      "loss": 1.6467,
      "step": 94912
    },
    {
      "epoch": 0.18735406746323763,
      "grad_norm": 0.09686318039894104,
      "learning_rate": 8.159432866930579e-05,
      "loss": 1.6563,
      "step": 94944
    },
    {
      "epoch": 0.18741721342463405,
      "grad_norm": 0.09595602750778198,
      "learning_rate": 8.15879884449411e-05,
      "loss": 1.6466,
      "step": 94976
    },
    {
      "epoch": 0.1874803593860305,
      "grad_norm": 0.1114514172077179,
      "learning_rate": 8.158164822057641e-05,
      "loss": 1.638,
      "step": 95008
    },
    {
      "epoch": 0.18754350534742695,
      "grad_norm": 0.10126730054616928,
      "learning_rate": 8.157530799621172e-05,
      "loss": 1.6645,
      "step": 95040
    },
    {
      "epoch": 0.18760665130882337,
      "grad_norm": 0.10387317091226578,
      "learning_rate": 8.156896777184702e-05,
      "loss": 1.636,
      "step": 95072
    },
    {
      "epoch": 0.18766979727021982,
      "grad_norm": 0.10348210483789444,
      "learning_rate": 8.156262754748234e-05,
      "loss": 1.651,
      "step": 95104
    },
    {
      "epoch": 0.18773294323161627,
      "grad_norm": 0.09826581925153732,
      "learning_rate": 8.155628732311765e-05,
      "loss": 1.6471,
      "step": 95136
    },
    {
      "epoch": 0.1877960891930127,
      "grad_norm": 0.10650411248207092,
      "learning_rate": 8.154994709875297e-05,
      "loss": 1.6481,
      "step": 95168
    },
    {
      "epoch": 0.18785923515440914,
      "grad_norm": 0.09995251148939133,
      "learning_rate": 8.154360687438827e-05,
      "loss": 1.6544,
      "step": 95200
    },
    {
      "epoch": 0.1879223811158056,
      "grad_norm": 0.10455673933029175,
      "learning_rate": 8.153726665002358e-05,
      "loss": 1.6416,
      "step": 95232
    },
    {
      "epoch": 0.18798552707720204,
      "grad_norm": 0.10279633849859238,
      "learning_rate": 8.15309264256589e-05,
      "loss": 1.6414,
      "step": 95264
    },
    {
      "epoch": 0.18804867303859846,
      "grad_norm": 0.10274940729141235,
      "learning_rate": 8.152458620129421e-05,
      "loss": 1.6594,
      "step": 95296
    },
    {
      "epoch": 0.1881118189999949,
      "grad_norm": 0.10336265712976456,
      "learning_rate": 8.151824597692951e-05,
      "loss": 1.6567,
      "step": 95328
    },
    {
      "epoch": 0.18817496496139136,
      "grad_norm": 0.10061810910701752,
      "learning_rate": 8.151190575256481e-05,
      "loss": 1.6391,
      "step": 95360
    },
    {
      "epoch": 0.18823811092278778,
      "grad_norm": 0.10806973278522491,
      "learning_rate": 8.150556552820013e-05,
      "loss": 1.6399,
      "step": 95392
    },
    {
      "epoch": 0.18830125688418423,
      "grad_norm": 0.10195987671613693,
      "learning_rate": 8.149922530383544e-05,
      "loss": 1.6512,
      "step": 95424
    },
    {
      "epoch": 0.18836440284558068,
      "grad_norm": 0.10850729048252106,
      "learning_rate": 8.149288507947074e-05,
      "loss": 1.6606,
      "step": 95456
    },
    {
      "epoch": 0.1884275488069771,
      "grad_norm": 0.10320477932691574,
      "learning_rate": 8.148654485510606e-05,
      "loss": 1.6406,
      "step": 95488
    },
    {
      "epoch": 0.18849069476837355,
      "grad_norm": 0.10265101492404938,
      "learning_rate": 8.148020463074137e-05,
      "loss": 1.6553,
      "step": 95520
    },
    {
      "epoch": 0.18855384072977,
      "grad_norm": 0.1066662147641182,
      "learning_rate": 8.147386440637669e-05,
      "loss": 1.654,
      "step": 95552
    },
    {
      "epoch": 0.18861698669116642,
      "grad_norm": 0.09863730520009995,
      "learning_rate": 8.146752418201199e-05,
      "loss": 1.6564,
      "step": 95584
    },
    {
      "epoch": 0.18868013265256287,
      "grad_norm": 0.10677376389503479,
      "learning_rate": 8.14611839576473e-05,
      "loss": 1.6508,
      "step": 95616
    },
    {
      "epoch": 0.18874327861395931,
      "grad_norm": 0.1141144335269928,
      "learning_rate": 8.145484373328262e-05,
      "loss": 1.6624,
      "step": 95648
    },
    {
      "epoch": 0.18880642457535574,
      "grad_norm": 0.10073503851890564,
      "learning_rate": 8.144850350891793e-05,
      "loss": 1.6435,
      "step": 95680
    },
    {
      "epoch": 0.18886957053675218,
      "grad_norm": 0.10807400941848755,
      "learning_rate": 8.144216328455325e-05,
      "loss": 1.6561,
      "step": 95712
    },
    {
      "epoch": 0.18893271649814863,
      "grad_norm": 0.10667506605386734,
      "learning_rate": 8.143582306018855e-05,
      "loss": 1.6476,
      "step": 95744
    },
    {
      "epoch": 0.18899586245954506,
      "grad_norm": 0.10062408447265625,
      "learning_rate": 8.142948283582385e-05,
      "loss": 1.6498,
      "step": 95776
    },
    {
      "epoch": 0.1890590084209415,
      "grad_norm": 0.09419935196638107,
      "learning_rate": 8.142314261145916e-05,
      "loss": 1.6497,
      "step": 95808
    },
    {
      "epoch": 0.18912215438233795,
      "grad_norm": 0.10771109163761139,
      "learning_rate": 8.141680238709448e-05,
      "loss": 1.6449,
      "step": 95840
    },
    {
      "epoch": 0.18918530034373437,
      "grad_norm": 0.1017397940158844,
      "learning_rate": 8.141046216272978e-05,
      "loss": 1.6556,
      "step": 95872
    },
    {
      "epoch": 0.18924844630513082,
      "grad_norm": 0.10358379781246185,
      "learning_rate": 8.14041219383651e-05,
      "loss": 1.6353,
      "step": 95904
    },
    {
      "epoch": 0.18931159226652727,
      "grad_norm": 0.10079692304134369,
      "learning_rate": 8.139778171400041e-05,
      "loss": 1.6432,
      "step": 95936
    },
    {
      "epoch": 0.1893747382279237,
      "grad_norm": 0.10643357038497925,
      "learning_rate": 8.139144148963572e-05,
      "loss": 1.6492,
      "step": 95968
    },
    {
      "epoch": 0.18943788418932014,
      "grad_norm": 0.10666496306657791,
      "learning_rate": 8.138510126527102e-05,
      "loss": 1.6451,
      "step": 96000
    },
    {
      "epoch": 0.1895010301507166,
      "grad_norm": 0.09939305484294891,
      "learning_rate": 8.137876104090634e-05,
      "loss": 1.6441,
      "step": 96032
    },
    {
      "epoch": 0.18956417611211301,
      "grad_norm": 0.10081891715526581,
      "learning_rate": 8.137242081654165e-05,
      "loss": 1.6482,
      "step": 96064
    },
    {
      "epoch": 0.18962732207350946,
      "grad_norm": 0.10111203044652939,
      "learning_rate": 8.136608059217697e-05,
      "loss": 1.6519,
      "step": 96096
    },
    {
      "epoch": 0.1896904680349059,
      "grad_norm": 0.10509985685348511,
      "learning_rate": 8.135974036781227e-05,
      "loss": 1.6562,
      "step": 96128
    },
    {
      "epoch": 0.18975361399630233,
      "grad_norm": 0.10449133068323135,
      "learning_rate": 8.135340014344758e-05,
      "loss": 1.6607,
      "step": 96160
    },
    {
      "epoch": 0.18981675995769878,
      "grad_norm": 0.10193048417568207,
      "learning_rate": 8.134705991908288e-05,
      "loss": 1.6527,
      "step": 96192
    },
    {
      "epoch": 0.18987990591909523,
      "grad_norm": 0.10266821831464767,
      "learning_rate": 8.13407196947182e-05,
      "loss": 1.6349,
      "step": 96224
    },
    {
      "epoch": 0.18994305188049165,
      "grad_norm": 0.10425172746181488,
      "learning_rate": 8.13343794703535e-05,
      "loss": 1.6547,
      "step": 96256
    },
    {
      "epoch": 0.1900061978418881,
      "grad_norm": 0.097136490046978,
      "learning_rate": 8.132803924598881e-05,
      "loss": 1.6369,
      "step": 96288
    },
    {
      "epoch": 0.19006934380328455,
      "grad_norm": 0.10471000522375107,
      "learning_rate": 8.132169902162413e-05,
      "loss": 1.6486,
      "step": 96320
    },
    {
      "epoch": 0.19013248976468097,
      "grad_norm": 0.10519730299711227,
      "learning_rate": 8.131535879725944e-05,
      "loss": 1.6568,
      "step": 96352
    },
    {
      "epoch": 0.19019563572607742,
      "grad_norm": 0.1161370649933815,
      "learning_rate": 8.130901857289476e-05,
      "loss": 1.6442,
      "step": 96384
    },
    {
      "epoch": 0.19025878168747387,
      "grad_norm": 0.10171066969633102,
      "learning_rate": 8.130267834853006e-05,
      "loss": 1.6534,
      "step": 96416
    },
    {
      "epoch": 0.1903219276488703,
      "grad_norm": 0.10538149625062943,
      "learning_rate": 8.129633812416537e-05,
      "loss": 1.645,
      "step": 96448
    },
    {
      "epoch": 0.19038507361026674,
      "grad_norm": 0.10869000107049942,
      "learning_rate": 8.128999789980069e-05,
      "loss": 1.6618,
      "step": 96480
    },
    {
      "epoch": 0.1904482195716632,
      "grad_norm": 0.10533742606639862,
      "learning_rate": 8.1283657675436e-05,
      "loss": 1.6469,
      "step": 96512
    },
    {
      "epoch": 0.1905113655330596,
      "grad_norm": 0.10774150490760803,
      "learning_rate": 8.12773174510713e-05,
      "loss": 1.6399,
      "step": 96544
    },
    {
      "epoch": 0.19057451149445606,
      "grad_norm": 0.09870371967554092,
      "learning_rate": 8.127097722670662e-05,
      "loss": 1.6704,
      "step": 96576
    },
    {
      "epoch": 0.1906376574558525,
      "grad_norm": 0.11601744592189789,
      "learning_rate": 8.126463700234192e-05,
      "loss": 1.6337,
      "step": 96608
    },
    {
      "epoch": 0.19070080341724893,
      "grad_norm": 0.09750094264745712,
      "learning_rate": 8.125829677797723e-05,
      "loss": 1.6564,
      "step": 96640
    },
    {
      "epoch": 0.19076394937864538,
      "grad_norm": 0.10632169246673584,
      "learning_rate": 8.125195655361254e-05,
      "loss": 1.6354,
      "step": 96672
    },
    {
      "epoch": 0.19082709534004183,
      "grad_norm": 0.10485920310020447,
      "learning_rate": 8.124561632924785e-05,
      "loss": 1.6384,
      "step": 96704
    },
    {
      "epoch": 0.19089024130143825,
      "grad_norm": 0.10848258435726166,
      "learning_rate": 8.123927610488316e-05,
      "loss": 1.6351,
      "step": 96736
    },
    {
      "epoch": 0.1909533872628347,
      "grad_norm": 0.09728173911571503,
      "learning_rate": 8.123293588051848e-05,
      "loss": 1.6388,
      "step": 96768
    },
    {
      "epoch": 0.19101653322423115,
      "grad_norm": 0.10131725668907166,
      "learning_rate": 8.122659565615378e-05,
      "loss": 1.6474,
      "step": 96800
    },
    {
      "epoch": 0.19107967918562757,
      "grad_norm": 0.10624504834413528,
      "learning_rate": 8.12202554317891e-05,
      "loss": 1.6605,
      "step": 96832
    },
    {
      "epoch": 0.19114282514702402,
      "grad_norm": 0.10106166452169418,
      "learning_rate": 8.121391520742441e-05,
      "loss": 1.6469,
      "step": 96864
    },
    {
      "epoch": 0.19120597110842047,
      "grad_norm": 0.10225090384483337,
      "learning_rate": 8.120757498305972e-05,
      "loss": 1.6605,
      "step": 96896
    },
    {
      "epoch": 0.19126911706981692,
      "grad_norm": 0.11132453382015228,
      "learning_rate": 8.120123475869502e-05,
      "loss": 1.6549,
      "step": 96928
    },
    {
      "epoch": 0.19133226303121334,
      "grad_norm": 0.10704923421144485,
      "learning_rate": 8.119489453433034e-05,
      "loss": 1.6479,
      "step": 96960
    },
    {
      "epoch": 0.1913954089926098,
      "grad_norm": 0.09917685389518738,
      "learning_rate": 8.118855430996565e-05,
      "loss": 1.6501,
      "step": 96992
    },
    {
      "epoch": 0.19145855495400624,
      "grad_norm": 0.1037558913230896,
      "learning_rate": 8.118221408560095e-05,
      "loss": 1.6623,
      "step": 97024
    },
    {
      "epoch": 0.19152170091540266,
      "grad_norm": 0.10285740345716476,
      "learning_rate": 8.117587386123627e-05,
      "loss": 1.652,
      "step": 97056
    },
    {
      "epoch": 0.1915848468767991,
      "grad_norm": 0.11165329068899155,
      "learning_rate": 8.116953363687157e-05,
      "loss": 1.6458,
      "step": 97088
    },
    {
      "epoch": 0.19164799283819556,
      "grad_norm": 0.09738541394472122,
      "learning_rate": 8.116319341250689e-05,
      "loss": 1.6244,
      "step": 97120
    },
    {
      "epoch": 0.19171113879959198,
      "grad_norm": 0.1026107519865036,
      "learning_rate": 8.11568531881422e-05,
      "loss": 1.6272,
      "step": 97152
    },
    {
      "epoch": 0.19177428476098843,
      "grad_norm": 0.09903274476528168,
      "learning_rate": 8.115051296377751e-05,
      "loss": 1.6434,
      "step": 97184
    },
    {
      "epoch": 0.19183743072238488,
      "grad_norm": 0.10325904935598373,
      "learning_rate": 8.114417273941282e-05,
      "loss": 1.6486,
      "step": 97216
    },
    {
      "epoch": 0.1919005766837813,
      "grad_norm": 0.11294377595186234,
      "learning_rate": 8.113783251504813e-05,
      "loss": 1.6392,
      "step": 97248
    },
    {
      "epoch": 0.19196372264517775,
      "grad_norm": 0.09898438304662704,
      "learning_rate": 8.113149229068344e-05,
      "loss": 1.6489,
      "step": 97280
    },
    {
      "epoch": 0.1920268686065742,
      "grad_norm": 0.1110379546880722,
      "learning_rate": 8.112515206631876e-05,
      "loss": 1.6338,
      "step": 97312
    },
    {
      "epoch": 0.19209001456797062,
      "grad_norm": 0.10507860779762268,
      "learning_rate": 8.111881184195406e-05,
      "loss": 1.6593,
      "step": 97344
    },
    {
      "epoch": 0.19215316052936707,
      "grad_norm": 0.1059076115489006,
      "learning_rate": 8.111247161758937e-05,
      "loss": 1.6513,
      "step": 97376
    },
    {
      "epoch": 0.19221630649076352,
      "grad_norm": 0.10829482972621918,
      "learning_rate": 8.110613139322469e-05,
      "loss": 1.6439,
      "step": 97408
    },
    {
      "epoch": 0.19227945245215994,
      "grad_norm": 0.10728301107883453,
      "learning_rate": 8.109979116885999e-05,
      "loss": 1.6492,
      "step": 97440
    },
    {
      "epoch": 0.1923425984135564,
      "grad_norm": 0.1041484922170639,
      "learning_rate": 8.109345094449529e-05,
      "loss": 1.6485,
      "step": 97472
    },
    {
      "epoch": 0.19240574437495284,
      "grad_norm": 0.11218147724866867,
      "learning_rate": 8.10871107201306e-05,
      "loss": 1.6417,
      "step": 97504
    },
    {
      "epoch": 0.19246889033634926,
      "grad_norm": 0.09830255061388016,
      "learning_rate": 8.108077049576592e-05,
      "loss": 1.6525,
      "step": 97536
    },
    {
      "epoch": 0.1925320362977457,
      "grad_norm": 0.10048898309469223,
      "learning_rate": 8.107443027140123e-05,
      "loss": 1.6614,
      "step": 97568
    },
    {
      "epoch": 0.19259518225914216,
      "grad_norm": 0.11010058224201202,
      "learning_rate": 8.106809004703654e-05,
      "loss": 1.6481,
      "step": 97600
    },
    {
      "epoch": 0.19265832822053858,
      "grad_norm": 0.09566009044647217,
      "learning_rate": 8.106174982267185e-05,
      "loss": 1.6518,
      "step": 97632
    },
    {
      "epoch": 0.19272147418193503,
      "grad_norm": 0.11182036995887756,
      "learning_rate": 8.105540959830716e-05,
      "loss": 1.6405,
      "step": 97664
    },
    {
      "epoch": 0.19278462014333148,
      "grad_norm": 0.10775816440582275,
      "learning_rate": 8.104906937394248e-05,
      "loss": 1.6487,
      "step": 97696
    },
    {
      "epoch": 0.1928477661047279,
      "grad_norm": 0.1047726571559906,
      "learning_rate": 8.10427291495778e-05,
      "loss": 1.6442,
      "step": 97728
    },
    {
      "epoch": 0.19291091206612435,
      "grad_norm": 0.09861469268798828,
      "learning_rate": 8.10363889252131e-05,
      "loss": 1.6592,
      "step": 97760
    },
    {
      "epoch": 0.1929740580275208,
      "grad_norm": 0.10144228488206863,
      "learning_rate": 8.103004870084841e-05,
      "loss": 1.6347,
      "step": 97792
    },
    {
      "epoch": 0.19303720398891722,
      "grad_norm": 0.10166819393634796,
      "learning_rate": 8.102370847648372e-05,
      "loss": 1.6351,
      "step": 97824
    },
    {
      "epoch": 0.19310034995031367,
      "grad_norm": 0.09624498337507248,
      "learning_rate": 8.101736825211903e-05,
      "loss": 1.6453,
      "step": 97856
    },
    {
      "epoch": 0.19316349591171011,
      "grad_norm": 0.10432087630033493,
      "learning_rate": 8.101102802775433e-05,
      "loss": 1.6373,
      "step": 97888
    },
    {
      "epoch": 0.19322664187310654,
      "grad_norm": 0.09928113222122192,
      "learning_rate": 8.100468780338964e-05,
      "loss": 1.6521,
      "step": 97920
    },
    {
      "epoch": 0.19328978783450299,
      "grad_norm": 0.10615003108978271,
      "learning_rate": 8.099834757902496e-05,
      "loss": 1.6345,
      "step": 97952
    },
    {
      "epoch": 0.19335293379589943,
      "grad_norm": 0.11123167723417282,
      "learning_rate": 8.099200735466027e-05,
      "loss": 1.6516,
      "step": 97984
    },
    {
      "epoch": 0.19341607975729586,
      "grad_norm": 0.10558531433343887,
      "learning_rate": 8.098566713029557e-05,
      "loss": 1.6458,
      "step": 98016
    },
    {
      "epoch": 0.1934792257186923,
      "grad_norm": 0.1033475399017334,
      "learning_rate": 8.097932690593089e-05,
      "loss": 1.648,
      "step": 98048
    },
    {
      "epoch": 0.19354237168008875,
      "grad_norm": 0.10164621472358704,
      "learning_rate": 8.09729866815662e-05,
      "loss": 1.634,
      "step": 98080
    },
    {
      "epoch": 0.19360551764148518,
      "grad_norm": 0.11083465069532394,
      "learning_rate": 8.096664645720151e-05,
      "loss": 1.6413,
      "step": 98112
    },
    {
      "epoch": 0.19366866360288162,
      "grad_norm": 0.10140354931354523,
      "learning_rate": 8.096030623283682e-05,
      "loss": 1.653,
      "step": 98144
    },
    {
      "epoch": 0.19373180956427807,
      "grad_norm": 0.10933768004179001,
      "learning_rate": 8.095396600847213e-05,
      "loss": 1.6367,
      "step": 98176
    },
    {
      "epoch": 0.1937949555256745,
      "grad_norm": 0.10203826427459717,
      "learning_rate": 8.094762578410744e-05,
      "loss": 1.6484,
      "step": 98208
    },
    {
      "epoch": 0.19385810148707094,
      "grad_norm": 0.11817627400159836,
      "learning_rate": 8.094128555974275e-05,
      "loss": 1.6633,
      "step": 98240
    },
    {
      "epoch": 0.1939212474484674,
      "grad_norm": 0.10960616916418076,
      "learning_rate": 8.093494533537806e-05,
      "loss": 1.642,
      "step": 98272
    },
    {
      "epoch": 0.19398439340986381,
      "grad_norm": 0.10867169499397278,
      "learning_rate": 8.092860511101336e-05,
      "loss": 1.6415,
      "step": 98304
    },
    {
      "epoch": 0.19404753937126026,
      "grad_norm": 0.10847428441047668,
      "learning_rate": 8.092226488664868e-05,
      "loss": 1.6447,
      "step": 98336
    },
    {
      "epoch": 0.1941106853326567,
      "grad_norm": 0.11210865527391434,
      "learning_rate": 8.091592466228399e-05,
      "loss": 1.6458,
      "step": 98368
    },
    {
      "epoch": 0.19417383129405313,
      "grad_norm": 0.0990021750330925,
      "learning_rate": 8.09095844379193e-05,
      "loss": 1.6528,
      "step": 98400
    },
    {
      "epoch": 0.19423697725544958,
      "grad_norm": 0.10736454278230667,
      "learning_rate": 8.09032442135546e-05,
      "loss": 1.6336,
      "step": 98432
    },
    {
      "epoch": 0.19430012321684603,
      "grad_norm": 0.11273426562547684,
      "learning_rate": 8.089690398918992e-05,
      "loss": 1.6453,
      "step": 98464
    },
    {
      "epoch": 0.19436326917824245,
      "grad_norm": 0.09996658563613892,
      "learning_rate": 8.089056376482524e-05,
      "loss": 1.6415,
      "step": 98496
    },
    {
      "epoch": 0.1944264151396389,
      "grad_norm": 0.10359546542167664,
      "learning_rate": 8.088422354046055e-05,
      "loss": 1.6481,
      "step": 98528
    },
    {
      "epoch": 0.19448956110103535,
      "grad_norm": 0.10674504190683365,
      "learning_rate": 8.087788331609585e-05,
      "loss": 1.6489,
      "step": 98560
    },
    {
      "epoch": 0.19455270706243177,
      "grad_norm": 0.11016551405191422,
      "learning_rate": 8.087154309173117e-05,
      "loss": 1.6447,
      "step": 98592
    },
    {
      "epoch": 0.19461585302382822,
      "grad_norm": 0.10759218037128448,
      "learning_rate": 8.086520286736648e-05,
      "loss": 1.6423,
      "step": 98624
    },
    {
      "epoch": 0.19467899898522467,
      "grad_norm": 0.10278622806072235,
      "learning_rate": 8.085886264300178e-05,
      "loss": 1.637,
      "step": 98656
    },
    {
      "epoch": 0.19474214494662112,
      "grad_norm": 0.10291390120983124,
      "learning_rate": 8.08525224186371e-05,
      "loss": 1.6439,
      "step": 98688
    },
    {
      "epoch": 0.19480529090801754,
      "grad_norm": 0.10670579224824905,
      "learning_rate": 8.08461821942724e-05,
      "loss": 1.6431,
      "step": 98720
    },
    {
      "epoch": 0.194868436869414,
      "grad_norm": 0.10281068831682205,
      "learning_rate": 8.083984196990771e-05,
      "loss": 1.652,
      "step": 98752
    },
    {
      "epoch": 0.19493158283081044,
      "grad_norm": 0.10091810673475266,
      "learning_rate": 8.083350174554303e-05,
      "loss": 1.6324,
      "step": 98784
    },
    {
      "epoch": 0.19499472879220686,
      "grad_norm": 0.10356218367815018,
      "learning_rate": 8.082716152117833e-05,
      "loss": 1.6405,
      "step": 98816
    },
    {
      "epoch": 0.1950578747536033,
      "grad_norm": 0.11054695397615433,
      "learning_rate": 8.082082129681364e-05,
      "loss": 1.6401,
      "step": 98848
    },
    {
      "epoch": 0.19512102071499976,
      "grad_norm": 0.10458224266767502,
      "learning_rate": 8.081448107244896e-05,
      "loss": 1.6454,
      "step": 98880
    },
    {
      "epoch": 0.19518416667639618,
      "grad_norm": 0.09834691882133484,
      "learning_rate": 8.080814084808427e-05,
      "loss": 1.637,
      "step": 98912
    },
    {
      "epoch": 0.19524731263779263,
      "grad_norm": 0.10757139325141907,
      "learning_rate": 8.080180062371958e-05,
      "loss": 1.6443,
      "step": 98944
    },
    {
      "epoch": 0.19531045859918908,
      "grad_norm": 0.10837163031101227,
      "learning_rate": 8.079546039935489e-05,
      "loss": 1.6451,
      "step": 98976
    },
    {
      "epoch": 0.1953736045605855,
      "grad_norm": 0.10756513476371765,
      "learning_rate": 8.07891201749902e-05,
      "loss": 1.6424,
      "step": 99008
    },
    {
      "epoch": 0.19543675052198195,
      "grad_norm": 0.11152730137109756,
      "learning_rate": 8.078277995062551e-05,
      "loss": 1.6454,
      "step": 99040
    },
    {
      "epoch": 0.1954998964833784,
      "grad_norm": 0.11061135679483414,
      "learning_rate": 8.077643972626082e-05,
      "loss": 1.633,
      "step": 99072
    },
    {
      "epoch": 0.19556304244477482,
      "grad_norm": 0.11073538661003113,
      "learning_rate": 8.077009950189613e-05,
      "loss": 1.6572,
      "step": 99104
    },
    {
      "epoch": 0.19562618840617127,
      "grad_norm": 0.10706045478582382,
      "learning_rate": 8.076375927753143e-05,
      "loss": 1.6454,
      "step": 99136
    },
    {
      "epoch": 0.19568933436756772,
      "grad_norm": 0.10636785626411438,
      "learning_rate": 8.075741905316675e-05,
      "loss": 1.6424,
      "step": 99168
    },
    {
      "epoch": 0.19575248032896414,
      "grad_norm": 0.10608632117509842,
      "learning_rate": 8.075107882880206e-05,
      "loss": 1.6418,
      "step": 99200
    },
    {
      "epoch": 0.1958156262903606,
      "grad_norm": 0.10915641486644745,
      "learning_rate": 8.074473860443736e-05,
      "loss": 1.6452,
      "step": 99232
    },
    {
      "epoch": 0.19587877225175704,
      "grad_norm": 0.11602920293807983,
      "learning_rate": 8.073839838007268e-05,
      "loss": 1.6507,
      "step": 99264
    },
    {
      "epoch": 0.19594191821315346,
      "grad_norm": 0.10733691602945328,
      "learning_rate": 8.073205815570799e-05,
      "loss": 1.6323,
      "step": 99296
    },
    {
      "epoch": 0.1960050641745499,
      "grad_norm": 0.10154245048761368,
      "learning_rate": 8.07257179313433e-05,
      "loss": 1.628,
      "step": 99328
    },
    {
      "epoch": 0.19606821013594636,
      "grad_norm": 0.1138070598244667,
      "learning_rate": 8.07193777069786e-05,
      "loss": 1.6535,
      "step": 99360
    },
    {
      "epoch": 0.19613135609734278,
      "grad_norm": 0.10772701352834702,
      "learning_rate": 8.071303748261392e-05,
      "loss": 1.6271,
      "step": 99392
    },
    {
      "epoch": 0.19619450205873923,
      "grad_norm": 0.1014992892742157,
      "learning_rate": 8.070669725824924e-05,
      "loss": 1.6415,
      "step": 99424
    },
    {
      "epoch": 0.19625764802013568,
      "grad_norm": 0.10349030047655106,
      "learning_rate": 8.070035703388455e-05,
      "loss": 1.6422,
      "step": 99456
    },
    {
      "epoch": 0.1963207939815321,
      "grad_norm": 0.10261155664920807,
      "learning_rate": 8.069401680951985e-05,
      "loss": 1.6474,
      "step": 99488
    },
    {
      "epoch": 0.19638393994292855,
      "grad_norm": 0.1023111417889595,
      "learning_rate": 8.068767658515515e-05,
      "loss": 1.644,
      "step": 99520
    },
    {
      "epoch": 0.196447085904325,
      "grad_norm": 0.10375040024518967,
      "learning_rate": 8.068133636079047e-05,
      "loss": 1.6466,
      "step": 99552
    },
    {
      "epoch": 0.19651023186572142,
      "grad_norm": 0.10253901034593582,
      "learning_rate": 8.067499613642578e-05,
      "loss": 1.6493,
      "step": 99584
    },
    {
      "epoch": 0.19657337782711787,
      "grad_norm": 0.09769316762685776,
      "learning_rate": 8.06686559120611e-05,
      "loss": 1.6492,
      "step": 99616
    },
    {
      "epoch": 0.19663652378851432,
      "grad_norm": 0.09973352402448654,
      "learning_rate": 8.06623156876964e-05,
      "loss": 1.6459,
      "step": 99648
    },
    {
      "epoch": 0.19669966974991074,
      "grad_norm": 0.09752755612134933,
      "learning_rate": 8.065597546333171e-05,
      "loss": 1.6495,
      "step": 99680
    },
    {
      "epoch": 0.1967628157113072,
      "grad_norm": 0.10267551243305206,
      "learning_rate": 8.064963523896703e-05,
      "loss": 1.6457,
      "step": 99712
    },
    {
      "epoch": 0.19682596167270364,
      "grad_norm": 0.10137288272380829,
      "learning_rate": 8.064329501460234e-05,
      "loss": 1.6338,
      "step": 99744
    },
    {
      "epoch": 0.19688910763410006,
      "grad_norm": 0.10639408975839615,
      "learning_rate": 8.063695479023764e-05,
      "loss": 1.6423,
      "step": 99776
    },
    {
      "epoch": 0.1969522535954965,
      "grad_norm": 0.10027466714382172,
      "learning_rate": 8.063061456587296e-05,
      "loss": 1.6394,
      "step": 99808
    },
    {
      "epoch": 0.19701539955689296,
      "grad_norm": 0.10378637909889221,
      "learning_rate": 8.062427434150827e-05,
      "loss": 1.6373,
      "step": 99840
    },
    {
      "epoch": 0.19707854551828938,
      "grad_norm": 0.10521262139081955,
      "learning_rate": 8.061793411714359e-05,
      "loss": 1.6414,
      "step": 99872
    },
    {
      "epoch": 0.19714169147968583,
      "grad_norm": 0.108073890209198,
      "learning_rate": 8.061159389277889e-05,
      "loss": 1.6466,
      "step": 99904
    },
    {
      "epoch": 0.19720483744108228,
      "grad_norm": 0.10325919836759567,
      "learning_rate": 8.060525366841419e-05,
      "loss": 1.6486,
      "step": 99936
    },
    {
      "epoch": 0.1972679834024787,
      "grad_norm": 0.10606803745031357,
      "learning_rate": 8.05989134440495e-05,
      "loss": 1.6402,
      "step": 99968
    },
    {
      "epoch": 0.19733112936387515,
      "grad_norm": 0.10333341360092163,
      "learning_rate": 8.059257321968482e-05,
      "loss": 1.649,
      "step": 100000
    },
    {
      "epoch": 0.1973942753252716,
      "grad_norm": 0.09961511939764023,
      "learning_rate": 8.058623299532012e-05,
      "loss": 1.6369,
      "step": 100032
    },
    {
      "epoch": 0.19745742128666802,
      "grad_norm": 0.10387607663869858,
      "learning_rate": 8.057989277095543e-05,
      "loss": 1.6453,
      "step": 100064
    },
    {
      "epoch": 0.19752056724806447,
      "grad_norm": 0.10061222314834595,
      "learning_rate": 8.057355254659075e-05,
      "loss": 1.638,
      "step": 100096
    },
    {
      "epoch": 0.19758371320946091,
      "grad_norm": 0.1025899276137352,
      "learning_rate": 8.056721232222606e-05,
      "loss": 1.6349,
      "step": 100128
    },
    {
      "epoch": 0.19764685917085734,
      "grad_norm": 0.10175150632858276,
      "learning_rate": 8.056087209786136e-05,
      "loss": 1.6466,
      "step": 100160
    },
    {
      "epoch": 0.19771000513225379,
      "grad_norm": 0.10213925689458847,
      "learning_rate": 8.055453187349668e-05,
      "loss": 1.6409,
      "step": 100192
    },
    {
      "epoch": 0.19777315109365023,
      "grad_norm": 0.10493280738592148,
      "learning_rate": 8.054819164913199e-05,
      "loss": 1.6476,
      "step": 100224
    },
    {
      "epoch": 0.19783629705504666,
      "grad_norm": 0.10523036867380142,
      "learning_rate": 8.05418514247673e-05,
      "loss": 1.65,
      "step": 100256
    },
    {
      "epoch": 0.1978994430164431,
      "grad_norm": 0.10567599534988403,
      "learning_rate": 8.053551120040262e-05,
      "loss": 1.631,
      "step": 100288
    },
    {
      "epoch": 0.19796258897783955,
      "grad_norm": 0.10249998420476913,
      "learning_rate": 8.052917097603792e-05,
      "loss": 1.6429,
      "step": 100320
    },
    {
      "epoch": 0.198025734939236,
      "grad_norm": 0.10519619286060333,
      "learning_rate": 8.052283075167322e-05,
      "loss": 1.6386,
      "step": 100352
    },
    {
      "epoch": 0.19808888090063242,
      "grad_norm": 0.1096009612083435,
      "learning_rate": 8.051649052730854e-05,
      "loss": 1.6612,
      "step": 100384
    },
    {
      "epoch": 0.19815202686202887,
      "grad_norm": 0.10052002221345901,
      "learning_rate": 8.051015030294385e-05,
      "loss": 1.6266,
      "step": 100416
    },
    {
      "epoch": 0.19821517282342532,
      "grad_norm": 0.10516545921564102,
      "learning_rate": 8.050381007857915e-05,
      "loss": 1.6359,
      "step": 100448
    },
    {
      "epoch": 0.19827831878482174,
      "grad_norm": 0.11198201030492783,
      "learning_rate": 8.049746985421447e-05,
      "loss": 1.6374,
      "step": 100480
    },
    {
      "epoch": 0.1983414647462182,
      "grad_norm": 0.10974428802728653,
      "learning_rate": 8.049112962984978e-05,
      "loss": 1.6302,
      "step": 100512
    },
    {
      "epoch": 0.19840461070761464,
      "grad_norm": 0.106208935379982,
      "learning_rate": 8.04847894054851e-05,
      "loss": 1.6484,
      "step": 100544
    },
    {
      "epoch": 0.19846775666901106,
      "grad_norm": 0.10340795665979385,
      "learning_rate": 8.04784491811204e-05,
      "loss": 1.6364,
      "step": 100576
    },
    {
      "epoch": 0.1985309026304075,
      "grad_norm": 0.10531339049339294,
      "learning_rate": 8.047210895675571e-05,
      "loss": 1.6321,
      "step": 100608
    },
    {
      "epoch": 0.19859404859180396,
      "grad_norm": 0.10440301895141602,
      "learning_rate": 8.046576873239103e-05,
      "loss": 1.6376,
      "step": 100640
    },
    {
      "epoch": 0.19865719455320038,
      "grad_norm": 0.10433050245046616,
      "learning_rate": 8.045942850802634e-05,
      "loss": 1.6453,
      "step": 100672
    },
    {
      "epoch": 0.19872034051459683,
      "grad_norm": 0.11418783664703369,
      "learning_rate": 8.045308828366164e-05,
      "loss": 1.6548,
      "step": 100704
    },
    {
      "epoch": 0.19878348647599328,
      "grad_norm": 0.10041820257902145,
      "learning_rate": 8.044674805929696e-05,
      "loss": 1.6455,
      "step": 100736
    },
    {
      "epoch": 0.1988466324373897,
      "grad_norm": 0.10225223749876022,
      "learning_rate": 8.044040783493226e-05,
      "loss": 1.6588,
      "step": 100768
    },
    {
      "epoch": 0.19890977839878615,
      "grad_norm": 0.10490626841783524,
      "learning_rate": 8.043406761056757e-05,
      "loss": 1.6358,
      "step": 100800
    },
    {
      "epoch": 0.1989729243601826,
      "grad_norm": 0.09783979505300522,
      "learning_rate": 8.042772738620287e-05,
      "loss": 1.6426,
      "step": 100832
    },
    {
      "epoch": 0.19903607032157902,
      "grad_norm": 0.1107398271560669,
      "learning_rate": 8.042138716183819e-05,
      "loss": 1.6348,
      "step": 100864
    },
    {
      "epoch": 0.19909921628297547,
      "grad_norm": 0.10240809619426727,
      "learning_rate": 8.04150469374735e-05,
      "loss": 1.637,
      "step": 100896
    },
    {
      "epoch": 0.19916236224437192,
      "grad_norm": 0.10322824865579605,
      "learning_rate": 8.040870671310882e-05,
      "loss": 1.636,
      "step": 100928
    },
    {
      "epoch": 0.19922550820576834,
      "grad_norm": 0.10999355465173721,
      "learning_rate": 8.040236648874413e-05,
      "loss": 1.6548,
      "step": 100960
    },
    {
      "epoch": 0.1992886541671648,
      "grad_norm": 0.11258241534233093,
      "learning_rate": 8.039602626437943e-05,
      "loss": 1.6374,
      "step": 100992
    },
    {
      "epoch": 0.19935180012856124,
      "grad_norm": 0.10544874519109726,
      "learning_rate": 8.038968604001475e-05,
      "loss": 1.6326,
      "step": 101024
    },
    {
      "epoch": 0.19941494608995766,
      "grad_norm": 0.10838599503040314,
      "learning_rate": 8.038334581565006e-05,
      "loss": 1.6444,
      "step": 101056
    },
    {
      "epoch": 0.1994780920513541,
      "grad_norm": 0.10652278363704681,
      "learning_rate": 8.037700559128538e-05,
      "loss": 1.634,
      "step": 101088
    },
    {
      "epoch": 0.19954123801275056,
      "grad_norm": 0.09899964183568954,
      "learning_rate": 8.037066536692068e-05,
      "loss": 1.6305,
      "step": 101120
    },
    {
      "epoch": 0.19960438397414698,
      "grad_norm": 0.09979450702667236,
      "learning_rate": 8.036432514255599e-05,
      "loss": 1.6409,
      "step": 101152
    },
    {
      "epoch": 0.19966752993554343,
      "grad_norm": 0.1164548322558403,
      "learning_rate": 8.035798491819129e-05,
      "loss": 1.6286,
      "step": 101184
    },
    {
      "epoch": 0.19973067589693988,
      "grad_norm": 0.10702712833881378,
      "learning_rate": 8.035164469382661e-05,
      "loss": 1.6503,
      "step": 101216
    },
    {
      "epoch": 0.1997938218583363,
      "grad_norm": 0.11386249959468842,
      "learning_rate": 8.034530446946191e-05,
      "loss": 1.6249,
      "step": 101248
    },
    {
      "epoch": 0.19985696781973275,
      "grad_norm": 0.11760639399290085,
      "learning_rate": 8.033896424509722e-05,
      "loss": 1.6389,
      "step": 101280
    },
    {
      "epoch": 0.1999201137811292,
      "grad_norm": 0.11131103336811066,
      "learning_rate": 8.033262402073254e-05,
      "loss": 1.6484,
      "step": 101312
    },
    {
      "epoch": 0.19998325974252562,
      "grad_norm": 0.10473434627056122,
      "learning_rate": 8.032628379636785e-05,
      "loss": 1.6389,
      "step": 101344
    },
    {
      "epoch": 0.20004640570392207,
      "grad_norm": 0.09926243871450424,
      "learning_rate": 8.031994357200315e-05,
      "loss": 1.6279,
      "step": 101376
    },
    {
      "epoch": 0.20010955166531852,
      "grad_norm": 0.10468669980764389,
      "learning_rate": 8.031360334763847e-05,
      "loss": 1.6538,
      "step": 101408
    },
    {
      "epoch": 0.20017269762671494,
      "grad_norm": 0.10861735045909882,
      "learning_rate": 8.030726312327378e-05,
      "loss": 1.6387,
      "step": 101440
    },
    {
      "epoch": 0.2002358435881114,
      "grad_norm": 0.10439712554216385,
      "learning_rate": 8.03009228989091e-05,
      "loss": 1.6422,
      "step": 101472
    },
    {
      "epoch": 0.20029898954950784,
      "grad_norm": 0.10245823860168457,
      "learning_rate": 8.02945826745444e-05,
      "loss": 1.6305,
      "step": 101504
    },
    {
      "epoch": 0.20036213551090426,
      "grad_norm": 0.10010068118572235,
      "learning_rate": 8.028824245017971e-05,
      "loss": 1.6429,
      "step": 101536
    },
    {
      "epoch": 0.2004252814723007,
      "grad_norm": 0.11026593297719955,
      "learning_rate": 8.028190222581503e-05,
      "loss": 1.6346,
      "step": 101568
    },
    {
      "epoch": 0.20048842743369716,
      "grad_norm": 0.09516335278749466,
      "learning_rate": 8.027556200145033e-05,
      "loss": 1.6437,
      "step": 101600
    },
    {
      "epoch": 0.20055157339509358,
      "grad_norm": 0.10414741933345795,
      "learning_rate": 8.026922177708564e-05,
      "loss": 1.6344,
      "step": 101632
    },
    {
      "epoch": 0.20061471935649003,
      "grad_norm": 0.09960043430328369,
      "learning_rate": 8.026288155272094e-05,
      "loss": 1.6429,
      "step": 101664
    },
    {
      "epoch": 0.20067786531788648,
      "grad_norm": 0.1058763787150383,
      "learning_rate": 8.025654132835626e-05,
      "loss": 1.6422,
      "step": 101696
    },
    {
      "epoch": 0.2007410112792829,
      "grad_norm": 0.10479961335659027,
      "learning_rate": 8.025020110399157e-05,
      "loss": 1.6385,
      "step": 101728
    },
    {
      "epoch": 0.20080415724067935,
      "grad_norm": 0.10158281028270721,
      "learning_rate": 8.024386087962689e-05,
      "loss": 1.6297,
      "step": 101760
    },
    {
      "epoch": 0.2008673032020758,
      "grad_norm": 0.10812821984291077,
      "learning_rate": 8.023752065526219e-05,
      "loss": 1.6454,
      "step": 101792
    },
    {
      "epoch": 0.20093044916347222,
      "grad_norm": 0.10637827217578888,
      "learning_rate": 8.02311804308975e-05,
      "loss": 1.6391,
      "step": 101824
    },
    {
      "epoch": 0.20099359512486867,
      "grad_norm": 0.09711770713329315,
      "learning_rate": 8.022484020653282e-05,
      "loss": 1.6454,
      "step": 101856
    },
    {
      "epoch": 0.20105674108626512,
      "grad_norm": 0.10218976438045502,
      "learning_rate": 8.021849998216813e-05,
      "loss": 1.6299,
      "step": 101888
    },
    {
      "epoch": 0.20111988704766154,
      "grad_norm": 0.10815039277076721,
      "learning_rate": 8.021215975780343e-05,
      "loss": 1.6494,
      "step": 101920
    },
    {
      "epoch": 0.201183033009058,
      "grad_norm": 0.1069837212562561,
      "learning_rate": 8.020581953343875e-05,
      "loss": 1.6388,
      "step": 101952
    },
    {
      "epoch": 0.20124617897045444,
      "grad_norm": 0.10718067735433578,
      "learning_rate": 8.019947930907406e-05,
      "loss": 1.6392,
      "step": 101984
    },
    {
      "epoch": 0.20130932493185089,
      "grad_norm": 0.10629121214151382,
      "learning_rate": 8.019313908470936e-05,
      "loss": 1.6318,
      "step": 102016
    },
    {
      "epoch": 0.2013724708932473,
      "grad_norm": 0.10247395187616348,
      "learning_rate": 8.018679886034466e-05,
      "loss": 1.6257,
      "step": 102048
    },
    {
      "epoch": 0.20143561685464376,
      "grad_norm": 0.10103968530893326,
      "learning_rate": 8.018045863597998e-05,
      "loss": 1.6292,
      "step": 102080
    },
    {
      "epoch": 0.2014987628160402,
      "grad_norm": 0.1039213314652443,
      "learning_rate": 8.017411841161529e-05,
      "loss": 1.6437,
      "step": 102112
    },
    {
      "epoch": 0.20156190877743663,
      "grad_norm": 0.10181664675474167,
      "learning_rate": 8.016777818725061e-05,
      "loss": 1.6263,
      "step": 102144
    },
    {
      "epoch": 0.20162505473883308,
      "grad_norm": 0.10486700385808945,
      "learning_rate": 8.016143796288591e-05,
      "loss": 1.6438,
      "step": 102176
    },
    {
      "epoch": 0.20168820070022953,
      "grad_norm": 0.10240813344717026,
      "learning_rate": 8.015509773852122e-05,
      "loss": 1.6496,
      "step": 102208
    },
    {
      "epoch": 0.20175134666162595,
      "grad_norm": 0.10481508076190948,
      "learning_rate": 8.014875751415654e-05,
      "loss": 1.6448,
      "step": 102240
    },
    {
      "epoch": 0.2018144926230224,
      "grad_norm": 0.10472988337278366,
      "learning_rate": 8.014241728979185e-05,
      "loss": 1.6536,
      "step": 102272
    },
    {
      "epoch": 0.20187763858441884,
      "grad_norm": 0.10133840888738632,
      "learning_rate": 8.013607706542717e-05,
      "loss": 1.6377,
      "step": 102304
    },
    {
      "epoch": 0.20194078454581527,
      "grad_norm": 0.10089903324842453,
      "learning_rate": 8.012973684106247e-05,
      "loss": 1.6298,
      "step": 102336
    },
    {
      "epoch": 0.20200393050721172,
      "grad_norm": 0.09654972702264786,
      "learning_rate": 8.012339661669778e-05,
      "loss": 1.6373,
      "step": 102368
    },
    {
      "epoch": 0.20206707646860816,
      "grad_norm": 0.11219391226768494,
      "learning_rate": 8.011705639233308e-05,
      "loss": 1.6376,
      "step": 102400
    },
    {
      "epoch": 0.20213022243000459,
      "grad_norm": 0.0997326597571373,
      "learning_rate": 8.01107161679684e-05,
      "loss": 1.6414,
      "step": 102432
    },
    {
      "epoch": 0.20219336839140103,
      "grad_norm": 0.10251238942146301,
      "learning_rate": 8.01043759436037e-05,
      "loss": 1.6398,
      "step": 102464
    },
    {
      "epoch": 0.20225651435279748,
      "grad_norm": 0.10804726928472519,
      "learning_rate": 8.009803571923901e-05,
      "loss": 1.6321,
      "step": 102496
    },
    {
      "epoch": 0.2023196603141939,
      "grad_norm": 0.10709121823310852,
      "learning_rate": 8.009169549487433e-05,
      "loss": 1.6401,
      "step": 102528
    },
    {
      "epoch": 0.20238280627559035,
      "grad_norm": 0.10653090476989746,
      "learning_rate": 8.008535527050964e-05,
      "loss": 1.6388,
      "step": 102560
    },
    {
      "epoch": 0.2024459522369868,
      "grad_norm": 0.10393144935369492,
      "learning_rate": 8.007901504614494e-05,
      "loss": 1.6322,
      "step": 102592
    },
    {
      "epoch": 0.20250909819838323,
      "grad_norm": 0.10259247571229935,
      "learning_rate": 8.007267482178026e-05,
      "loss": 1.6508,
      "step": 102624
    },
    {
      "epoch": 0.20257224415977967,
      "grad_norm": 0.09982217848300934,
      "learning_rate": 8.006633459741557e-05,
      "loss": 1.633,
      "step": 102656
    },
    {
      "epoch": 0.20263539012117612,
      "grad_norm": 0.10510411113500595,
      "learning_rate": 8.005999437305089e-05,
      "loss": 1.6486,
      "step": 102688
    },
    {
      "epoch": 0.20269853608257254,
      "grad_norm": 0.10338443517684937,
      "learning_rate": 8.005365414868619e-05,
      "loss": 1.6382,
      "step": 102720
    },
    {
      "epoch": 0.202761682043969,
      "grad_norm": 0.10875093191862106,
      "learning_rate": 8.00473139243215e-05,
      "loss": 1.6319,
      "step": 102752
    },
    {
      "epoch": 0.20282482800536544,
      "grad_norm": 0.10368429124355316,
      "learning_rate": 8.004097369995682e-05,
      "loss": 1.642,
      "step": 102784
    },
    {
      "epoch": 0.20288797396676186,
      "grad_norm": 0.10730479657649994,
      "learning_rate": 8.003463347559212e-05,
      "loss": 1.6423,
      "step": 102816
    },
    {
      "epoch": 0.2029511199281583,
      "grad_norm": 0.10140582174062729,
      "learning_rate": 8.002829325122743e-05,
      "loss": 1.6465,
      "step": 102848
    },
    {
      "epoch": 0.20301426588955476,
      "grad_norm": 0.10043517500162125,
      "learning_rate": 8.002195302686273e-05,
      "loss": 1.6381,
      "step": 102880
    },
    {
      "epoch": 0.20307741185095118,
      "grad_norm": 0.10014060139656067,
      "learning_rate": 8.001561280249805e-05,
      "loss": 1.6382,
      "step": 102912
    },
    {
      "epoch": 0.20314055781234763,
      "grad_norm": 0.10281931608915329,
      "learning_rate": 8.000927257813336e-05,
      "loss": 1.6392,
      "step": 102944
    },
    {
      "epoch": 0.20320370377374408,
      "grad_norm": 0.10233409702777863,
      "learning_rate": 8.000293235376868e-05,
      "loss": 1.6471,
      "step": 102976
    },
    {
      "epoch": 0.2032668497351405,
      "grad_norm": 0.10273443162441254,
      "learning_rate": 7.999659212940398e-05,
      "loss": 1.6392,
      "step": 103008
    },
    {
      "epoch": 0.20332999569653695,
      "grad_norm": 0.11665486544370651,
      "learning_rate": 7.999025190503929e-05,
      "loss": 1.6329,
      "step": 103040
    },
    {
      "epoch": 0.2033931416579334,
      "grad_norm": 0.10916301608085632,
      "learning_rate": 7.998391168067461e-05,
      "loss": 1.6349,
      "step": 103072
    },
    {
      "epoch": 0.20345628761932982,
      "grad_norm": 0.11598243564367294,
      "learning_rate": 7.997757145630992e-05,
      "loss": 1.6355,
      "step": 103104
    },
    {
      "epoch": 0.20351943358072627,
      "grad_norm": 0.09926288574934006,
      "learning_rate": 7.997123123194522e-05,
      "loss": 1.6362,
      "step": 103136
    },
    {
      "epoch": 0.20358257954212272,
      "grad_norm": 0.1062379777431488,
      "learning_rate": 7.996489100758054e-05,
      "loss": 1.6415,
      "step": 103168
    },
    {
      "epoch": 0.20364572550351914,
      "grad_norm": 0.10139545798301697,
      "learning_rate": 7.995855078321585e-05,
      "loss": 1.6482,
      "step": 103200
    },
    {
      "epoch": 0.2037088714649156,
      "grad_norm": 0.1033335030078888,
      "learning_rate": 7.995221055885115e-05,
      "loss": 1.6377,
      "step": 103232
    },
    {
      "epoch": 0.20377201742631204,
      "grad_norm": 0.09786981344223022,
      "learning_rate": 7.994587033448647e-05,
      "loss": 1.6401,
      "step": 103264
    },
    {
      "epoch": 0.20383516338770846,
      "grad_norm": 0.10496127605438232,
      "learning_rate": 7.993953011012177e-05,
      "loss": 1.6311,
      "step": 103296
    },
    {
      "epoch": 0.2038983093491049,
      "grad_norm": 0.10922686010599136,
      "learning_rate": 7.993318988575708e-05,
      "loss": 1.6566,
      "step": 103328
    },
    {
      "epoch": 0.20396145531050136,
      "grad_norm": 0.10109837353229523,
      "learning_rate": 7.99268496613924e-05,
      "loss": 1.6406,
      "step": 103360
    },
    {
      "epoch": 0.20402460127189778,
      "grad_norm": 0.10708053410053253,
      "learning_rate": 7.99205094370277e-05,
      "loss": 1.6286,
      "step": 103392
    },
    {
      "epoch": 0.20408774723329423,
      "grad_norm": 0.10658697038888931,
      "learning_rate": 7.991416921266301e-05,
      "loss": 1.6402,
      "step": 103424
    },
    {
      "epoch": 0.20415089319469068,
      "grad_norm": 0.1058441624045372,
      "learning_rate": 7.990782898829833e-05,
      "loss": 1.6323,
      "step": 103456
    },
    {
      "epoch": 0.2042140391560871,
      "grad_norm": 0.10291415452957153,
      "learning_rate": 7.990148876393364e-05,
      "loss": 1.652,
      "step": 103488
    },
    {
      "epoch": 0.20427718511748355,
      "grad_norm": 0.10548091679811478,
      "learning_rate": 7.989514853956894e-05,
      "loss": 1.6589,
      "step": 103520
    },
    {
      "epoch": 0.20434033107888,
      "grad_norm": 0.10049108415842056,
      "learning_rate": 7.988880831520426e-05,
      "loss": 1.6392,
      "step": 103552
    },
    {
      "epoch": 0.20440347704027642,
      "grad_norm": 0.10302495956420898,
      "learning_rate": 7.988246809083957e-05,
      "loss": 1.6507,
      "step": 103584
    },
    {
      "epoch": 0.20446662300167287,
      "grad_norm": 0.11602891236543655,
      "learning_rate": 7.987612786647489e-05,
      "loss": 1.6406,
      "step": 103616
    },
    {
      "epoch": 0.20452976896306932,
      "grad_norm": 0.11603625118732452,
      "learning_rate": 7.986978764211019e-05,
      "loss": 1.6338,
      "step": 103648
    },
    {
      "epoch": 0.20459291492446577,
      "grad_norm": 0.1035018116235733,
      "learning_rate": 7.986344741774549e-05,
      "loss": 1.6365,
      "step": 103680
    },
    {
      "epoch": 0.2046560608858622,
      "grad_norm": 0.10047438740730286,
      "learning_rate": 7.98571071933808e-05,
      "loss": 1.6358,
      "step": 103712
    },
    {
      "epoch": 0.20471920684725864,
      "grad_norm": 0.10895229876041412,
      "learning_rate": 7.985076696901612e-05,
      "loss": 1.6266,
      "step": 103744
    },
    {
      "epoch": 0.2047823528086551,
      "grad_norm": 0.10601717233657837,
      "learning_rate": 7.984442674465143e-05,
      "loss": 1.6484,
      "step": 103776
    },
    {
      "epoch": 0.2048454987700515,
      "grad_norm": 0.10165352374315262,
      "learning_rate": 7.983808652028673e-05,
      "loss": 1.6343,
      "step": 103808
    },
    {
      "epoch": 0.20490864473144796,
      "grad_norm": 0.10823352634906769,
      "learning_rate": 7.983174629592205e-05,
      "loss": 1.6422,
      "step": 103840
    },
    {
      "epoch": 0.2049717906928444,
      "grad_norm": 0.1027594804763794,
      "learning_rate": 7.982540607155736e-05,
      "loss": 1.6399,
      "step": 103872
    },
    {
      "epoch": 0.20503493665424083,
      "grad_norm": 0.10999977588653564,
      "learning_rate": 7.981906584719268e-05,
      "loss": 1.6346,
      "step": 103904
    },
    {
      "epoch": 0.20509808261563728,
      "grad_norm": 0.11509554833173752,
      "learning_rate": 7.981272562282798e-05,
      "loss": 1.6308,
      "step": 103936
    },
    {
      "epoch": 0.20516122857703373,
      "grad_norm": 0.1042904406785965,
      "learning_rate": 7.98063853984633e-05,
      "loss": 1.6328,
      "step": 103968
    },
    {
      "epoch": 0.20522437453843015,
      "grad_norm": 0.10567998886108398,
      "learning_rate": 7.980004517409861e-05,
      "loss": 1.642,
      "step": 104000
    },
    {
      "epoch": 0.2052875204998266,
      "grad_norm": 0.1036311537027359,
      "learning_rate": 7.979370494973392e-05,
      "loss": 1.6318,
      "step": 104032
    },
    {
      "epoch": 0.20535066646122305,
      "grad_norm": 0.11015824228525162,
      "learning_rate": 7.978736472536922e-05,
      "loss": 1.6422,
      "step": 104064
    },
    {
      "epoch": 0.20541381242261947,
      "grad_norm": 0.10671118646860123,
      "learning_rate": 7.978102450100452e-05,
      "loss": 1.6493,
      "step": 104096
    },
    {
      "epoch": 0.20547695838401592,
      "grad_norm": 0.10612434893846512,
      "learning_rate": 7.977468427663984e-05,
      "loss": 1.6368,
      "step": 104128
    },
    {
      "epoch": 0.20554010434541237,
      "grad_norm": 0.11015817523002625,
      "learning_rate": 7.976834405227515e-05,
      "loss": 1.6262,
      "step": 104160
    },
    {
      "epoch": 0.2056032503068088,
      "grad_norm": 0.10126221179962158,
      "learning_rate": 7.976200382791045e-05,
      "loss": 1.623,
      "step": 104192
    },
    {
      "epoch": 0.20566639626820524,
      "grad_norm": 0.10176030546426773,
      "learning_rate": 7.975566360354577e-05,
      "loss": 1.6464,
      "step": 104224
    },
    {
      "epoch": 0.2057295422296017,
      "grad_norm": 0.09650059789419174,
      "learning_rate": 7.974932337918108e-05,
      "loss": 1.654,
      "step": 104256
    },
    {
      "epoch": 0.2057926881909981,
      "grad_norm": 0.11535985767841339,
      "learning_rate": 7.97429831548164e-05,
      "loss": 1.629,
      "step": 104288
    },
    {
      "epoch": 0.20585583415239456,
      "grad_norm": 0.1117384284734726,
      "learning_rate": 7.973664293045171e-05,
      "loss": 1.638,
      "step": 104320
    },
    {
      "epoch": 0.205918980113791,
      "grad_norm": 0.10364539176225662,
      "learning_rate": 7.973030270608701e-05,
      "loss": 1.6406,
      "step": 104352
    },
    {
      "epoch": 0.20598212607518743,
      "grad_norm": 0.11449093371629715,
      "learning_rate": 7.972396248172233e-05,
      "loss": 1.6387,
      "step": 104384
    },
    {
      "epoch": 0.20604527203658388,
      "grad_norm": 0.10183209180831909,
      "learning_rate": 7.971762225735764e-05,
      "loss": 1.6382,
      "step": 104416
    },
    {
      "epoch": 0.20610841799798033,
      "grad_norm": 0.10846957564353943,
      "learning_rate": 7.971128203299296e-05,
      "loss": 1.642,
      "step": 104448
    },
    {
      "epoch": 0.20617156395937675,
      "grad_norm": 0.10982728749513626,
      "learning_rate": 7.970494180862826e-05,
      "loss": 1.6434,
      "step": 104480
    },
    {
      "epoch": 0.2062347099207732,
      "grad_norm": 0.10392747819423676,
      "learning_rate": 7.969860158426356e-05,
      "loss": 1.6378,
      "step": 104512
    },
    {
      "epoch": 0.20629785588216965,
      "grad_norm": 0.1094764694571495,
      "learning_rate": 7.969226135989887e-05,
      "loss": 1.6288,
      "step": 104544
    },
    {
      "epoch": 0.20636100184356607,
      "grad_norm": 0.10573791712522507,
      "learning_rate": 7.968592113553419e-05,
      "loss": 1.6465,
      "step": 104576
    },
    {
      "epoch": 0.20642414780496252,
      "grad_norm": 0.10767697542905807,
      "learning_rate": 7.967958091116949e-05,
      "loss": 1.6346,
      "step": 104608
    },
    {
      "epoch": 0.20648729376635896,
      "grad_norm": 0.10751378536224365,
      "learning_rate": 7.96732406868048e-05,
      "loss": 1.6503,
      "step": 104640
    },
    {
      "epoch": 0.20655043972775539,
      "grad_norm": 0.10429476201534271,
      "learning_rate": 7.966690046244012e-05,
      "loss": 1.628,
      "step": 104672
    },
    {
      "epoch": 0.20661358568915184,
      "grad_norm": 0.10951858758926392,
      "learning_rate": 7.966056023807543e-05,
      "loss": 1.6357,
      "step": 104704
    },
    {
      "epoch": 0.20667673165054828,
      "grad_norm": 0.11229857802391052,
      "learning_rate": 7.965422001371073e-05,
      "loss": 1.6276,
      "step": 104736
    },
    {
      "epoch": 0.2067398776119447,
      "grad_norm": 0.10861329734325409,
      "learning_rate": 7.964787978934605e-05,
      "loss": 1.6384,
      "step": 104768
    },
    {
      "epoch": 0.20680302357334115,
      "grad_norm": 0.11048752069473267,
      "learning_rate": 7.964153956498136e-05,
      "loss": 1.6302,
      "step": 104800
    },
    {
      "epoch": 0.2068661695347376,
      "grad_norm": 0.10989987105131149,
      "learning_rate": 7.963519934061668e-05,
      "loss": 1.6341,
      "step": 104832
    },
    {
      "epoch": 0.20692931549613403,
      "grad_norm": 0.11207330226898193,
      "learning_rate": 7.962885911625198e-05,
      "loss": 1.6355,
      "step": 104864
    },
    {
      "epoch": 0.20699246145753047,
      "grad_norm": 0.1130172461271286,
      "learning_rate": 7.96225188918873e-05,
      "loss": 1.6352,
      "step": 104896
    },
    {
      "epoch": 0.20705560741892692,
      "grad_norm": 0.10822995752096176,
      "learning_rate": 7.96161786675226e-05,
      "loss": 1.6302,
      "step": 104928
    },
    {
      "epoch": 0.20711875338032334,
      "grad_norm": 0.10452887415885925,
      "learning_rate": 7.960983844315791e-05,
      "loss": 1.6229,
      "step": 104960
    },
    {
      "epoch": 0.2071818993417198,
      "grad_norm": 0.10353318601846695,
      "learning_rate": 7.960349821879322e-05,
      "loss": 1.6546,
      "step": 104992
    },
    {
      "epoch": 0.20724504530311624,
      "grad_norm": 0.10682032257318497,
      "learning_rate": 7.959715799442852e-05,
      "loss": 1.6504,
      "step": 105024
    },
    {
      "epoch": 0.20730819126451266,
      "grad_norm": 0.1023780032992363,
      "learning_rate": 7.959081777006384e-05,
      "loss": 1.6447,
      "step": 105056
    },
    {
      "epoch": 0.2073713372259091,
      "grad_norm": 0.1099579930305481,
      "learning_rate": 7.958447754569915e-05,
      "loss": 1.6346,
      "step": 105088
    },
    {
      "epoch": 0.20743448318730556,
      "grad_norm": 0.10792119055986404,
      "learning_rate": 7.957813732133447e-05,
      "loss": 1.6371,
      "step": 105120
    },
    {
      "epoch": 0.20749762914870198,
      "grad_norm": 0.1058160662651062,
      "learning_rate": 7.957179709696977e-05,
      "loss": 1.6322,
      "step": 105152
    },
    {
      "epoch": 0.20756077511009843,
      "grad_norm": 0.10008879750967026,
      "learning_rate": 7.956545687260508e-05,
      "loss": 1.6285,
      "step": 105184
    },
    {
      "epoch": 0.20762392107149488,
      "grad_norm": 0.10394812375307083,
      "learning_rate": 7.95591166482404e-05,
      "loss": 1.6568,
      "step": 105216
    },
    {
      "epoch": 0.2076870670328913,
      "grad_norm": 0.103737473487854,
      "learning_rate": 7.955277642387571e-05,
      "loss": 1.6271,
      "step": 105248
    },
    {
      "epoch": 0.20775021299428775,
      "grad_norm": 0.09900253266096115,
      "learning_rate": 7.954643619951101e-05,
      "loss": 1.6324,
      "step": 105280
    },
    {
      "epoch": 0.2078133589556842,
      "grad_norm": 0.10779475420713425,
      "learning_rate": 7.954009597514633e-05,
      "loss": 1.6382,
      "step": 105312
    },
    {
      "epoch": 0.20787650491708062,
      "grad_norm": 0.10209625959396362,
      "learning_rate": 7.953375575078163e-05,
      "loss": 1.6425,
      "step": 105344
    },
    {
      "epoch": 0.20793965087847707,
      "grad_norm": 0.09763719141483307,
      "learning_rate": 7.952741552641694e-05,
      "loss": 1.6378,
      "step": 105376
    },
    {
      "epoch": 0.20800279683987352,
      "grad_norm": 0.10402107238769531,
      "learning_rate": 7.952107530205225e-05,
      "loss": 1.6357,
      "step": 105408
    },
    {
      "epoch": 0.20806594280126997,
      "grad_norm": 0.10512713342905045,
      "learning_rate": 7.951473507768756e-05,
      "loss": 1.6365,
      "step": 105440
    },
    {
      "epoch": 0.2081290887626664,
      "grad_norm": 0.09336667507886887,
      "learning_rate": 7.950839485332287e-05,
      "loss": 1.6368,
      "step": 105472
    },
    {
      "epoch": 0.20819223472406284,
      "grad_norm": 0.11189181357622147,
      "learning_rate": 7.950205462895819e-05,
      "loss": 1.6453,
      "step": 105504
    },
    {
      "epoch": 0.2082553806854593,
      "grad_norm": 0.10604702681303024,
      "learning_rate": 7.949571440459349e-05,
      "loss": 1.6456,
      "step": 105536
    },
    {
      "epoch": 0.2083185266468557,
      "grad_norm": 0.1117272675037384,
      "learning_rate": 7.94893741802288e-05,
      "loss": 1.647,
      "step": 105568
    },
    {
      "epoch": 0.20838167260825216,
      "grad_norm": 0.10491900891065598,
      "learning_rate": 7.948303395586412e-05,
      "loss": 1.6448,
      "step": 105600
    },
    {
      "epoch": 0.2084448185696486,
      "grad_norm": 0.10880821198225021,
      "learning_rate": 7.947669373149943e-05,
      "loss": 1.6403,
      "step": 105632
    },
    {
      "epoch": 0.20850796453104503,
      "grad_norm": 0.10090239346027374,
      "learning_rate": 7.947035350713475e-05,
      "loss": 1.6436,
      "step": 105664
    },
    {
      "epoch": 0.20857111049244148,
      "grad_norm": 0.1004345566034317,
      "learning_rate": 7.946401328277005e-05,
      "loss": 1.6449,
      "step": 105696
    },
    {
      "epoch": 0.20863425645383793,
      "grad_norm": 0.10114094614982605,
      "learning_rate": 7.945767305840536e-05,
      "loss": 1.6206,
      "step": 105728
    },
    {
      "epoch": 0.20869740241523435,
      "grad_norm": 0.1005459651350975,
      "learning_rate": 7.945133283404066e-05,
      "loss": 1.6432,
      "step": 105760
    },
    {
      "epoch": 0.2087605483766308,
      "grad_norm": 0.1044689491391182,
      "learning_rate": 7.944499260967598e-05,
      "loss": 1.6372,
      "step": 105792
    },
    {
      "epoch": 0.20882369433802725,
      "grad_norm": 0.10328377783298492,
      "learning_rate": 7.943865238531128e-05,
      "loss": 1.6316,
      "step": 105824
    },
    {
      "epoch": 0.20888684029942367,
      "grad_norm": 0.10217892378568649,
      "learning_rate": 7.94323121609466e-05,
      "loss": 1.6456,
      "step": 105856
    },
    {
      "epoch": 0.20894998626082012,
      "grad_norm": 0.10909678786993027,
      "learning_rate": 7.942597193658191e-05,
      "loss": 1.6392,
      "step": 105888
    },
    {
      "epoch": 0.20901313222221657,
      "grad_norm": 0.10893984884023666,
      "learning_rate": 7.941963171221722e-05,
      "loss": 1.6347,
      "step": 105920
    },
    {
      "epoch": 0.209076278183613,
      "grad_norm": 0.10438256710767746,
      "learning_rate": 7.941329148785253e-05,
      "loss": 1.6432,
      "step": 105952
    },
    {
      "epoch": 0.20913942414500944,
      "grad_norm": 0.10000976175069809,
      "learning_rate": 7.940695126348784e-05,
      "loss": 1.6315,
      "step": 105984
    },
    {
      "epoch": 0.2092025701064059,
      "grad_norm": 0.10432066768407822,
      "learning_rate": 7.940061103912315e-05,
      "loss": 1.6335,
      "step": 106016
    },
    {
      "epoch": 0.2092657160678023,
      "grad_norm": 0.10000544786453247,
      "learning_rate": 7.939427081475847e-05,
      "loss": 1.647,
      "step": 106048
    },
    {
      "epoch": 0.20932886202919876,
      "grad_norm": 0.10180745273828506,
      "learning_rate": 7.938793059039377e-05,
      "loss": 1.6445,
      "step": 106080
    },
    {
      "epoch": 0.2093920079905952,
      "grad_norm": 0.10191638767719269,
      "learning_rate": 7.938159036602908e-05,
      "loss": 1.633,
      "step": 106112
    },
    {
      "epoch": 0.20945515395199163,
      "grad_norm": 0.10588212311267853,
      "learning_rate": 7.93752501416644e-05,
      "loss": 1.633,
      "step": 106144
    },
    {
      "epoch": 0.20951829991338808,
      "grad_norm": 0.10806088894605637,
      "learning_rate": 7.93689099172997e-05,
      "loss": 1.6169,
      "step": 106176
    },
    {
      "epoch": 0.20958144587478453,
      "grad_norm": 0.10362087190151215,
      "learning_rate": 7.9362569692935e-05,
      "loss": 1.6265,
      "step": 106208
    },
    {
      "epoch": 0.20964459183618095,
      "grad_norm": 0.11316730082035065,
      "learning_rate": 7.935622946857032e-05,
      "loss": 1.6222,
      "step": 106240
    },
    {
      "epoch": 0.2097077377975774,
      "grad_norm": 0.09931106120347977,
      "learning_rate": 7.934988924420563e-05,
      "loss": 1.6397,
      "step": 106272
    },
    {
      "epoch": 0.20977088375897385,
      "grad_norm": 0.10171699523925781,
      "learning_rate": 7.934354901984094e-05,
      "loss": 1.6344,
      "step": 106304
    },
    {
      "epoch": 0.20983402972037027,
      "grad_norm": 0.09845674782991409,
      "learning_rate": 7.933720879547626e-05,
      "loss": 1.6288,
      "step": 106336
    },
    {
      "epoch": 0.20989717568176672,
      "grad_norm": 0.10100769996643066,
      "learning_rate": 7.933086857111156e-05,
      "loss": 1.6289,
      "step": 106368
    },
    {
      "epoch": 0.20996032164316317,
      "grad_norm": 0.10174597054719925,
      "learning_rate": 7.932452834674687e-05,
      "loss": 1.6276,
      "step": 106400
    },
    {
      "epoch": 0.2100234676045596,
      "grad_norm": 0.1099524050951004,
      "learning_rate": 7.931818812238219e-05,
      "loss": 1.6393,
      "step": 106432
    },
    {
      "epoch": 0.21008661356595604,
      "grad_norm": 0.10879639536142349,
      "learning_rate": 7.93118478980175e-05,
      "loss": 1.6305,
      "step": 106464
    },
    {
      "epoch": 0.2101497595273525,
      "grad_norm": 0.09787270426750183,
      "learning_rate": 7.93055076736528e-05,
      "loss": 1.6328,
      "step": 106496
    },
    {
      "epoch": 0.2102129054887489,
      "grad_norm": 0.11112847924232483,
      "learning_rate": 7.929916744928812e-05,
      "loss": 1.6346,
      "step": 106528
    },
    {
      "epoch": 0.21027605145014536,
      "grad_norm": 0.10521311312913895,
      "learning_rate": 7.929282722492342e-05,
      "loss": 1.6384,
      "step": 106560
    },
    {
      "epoch": 0.2103391974115418,
      "grad_norm": 0.10209573060274124,
      "learning_rate": 7.928648700055874e-05,
      "loss": 1.6377,
      "step": 106592
    },
    {
      "epoch": 0.21040234337293823,
      "grad_norm": 0.10402011126279831,
      "learning_rate": 7.928014677619404e-05,
      "loss": 1.624,
      "step": 106624
    },
    {
      "epoch": 0.21046548933433468,
      "grad_norm": 0.10467496514320374,
      "learning_rate": 7.927380655182935e-05,
      "loss": 1.6284,
      "step": 106656
    },
    {
      "epoch": 0.21052863529573113,
      "grad_norm": 0.10447464883327484,
      "learning_rate": 7.926746632746467e-05,
      "loss": 1.6357,
      "step": 106688
    },
    {
      "epoch": 0.21059178125712755,
      "grad_norm": 0.11272741109132767,
      "learning_rate": 7.926112610309998e-05,
      "loss": 1.6341,
      "step": 106720
    },
    {
      "epoch": 0.210654927218524,
      "grad_norm": 0.10509485751390457,
      "learning_rate": 7.925478587873528e-05,
      "loss": 1.6352,
      "step": 106752
    },
    {
      "epoch": 0.21071807317992045,
      "grad_norm": 0.10457901656627655,
      "learning_rate": 7.92484456543706e-05,
      "loss": 1.6272,
      "step": 106784
    },
    {
      "epoch": 0.21078121914131687,
      "grad_norm": 0.10728678107261658,
      "learning_rate": 7.924210543000591e-05,
      "loss": 1.6277,
      "step": 106816
    },
    {
      "epoch": 0.21084436510271332,
      "grad_norm": 0.09993508458137512,
      "learning_rate": 7.923576520564122e-05,
      "loss": 1.6368,
      "step": 106848
    },
    {
      "epoch": 0.21090751106410977,
      "grad_norm": 0.10373237729072571,
      "learning_rate": 7.922942498127653e-05,
      "loss": 1.631,
      "step": 106880
    },
    {
      "epoch": 0.2109706570255062,
      "grad_norm": 0.0996611937880516,
      "learning_rate": 7.922308475691184e-05,
      "loss": 1.649,
      "step": 106912
    },
    {
      "epoch": 0.21103380298690264,
      "grad_norm": 0.10735130310058594,
      "learning_rate": 7.921674453254715e-05,
      "loss": 1.6348,
      "step": 106944
    },
    {
      "epoch": 0.21109694894829908,
      "grad_norm": 0.10662022233009338,
      "learning_rate": 7.921040430818246e-05,
      "loss": 1.6241,
      "step": 106976
    },
    {
      "epoch": 0.2111600949096955,
      "grad_norm": 0.10203149914741516,
      "learning_rate": 7.920406408381777e-05,
      "loss": 1.6451,
      "step": 107008
    },
    {
      "epoch": 0.21122324087109196,
      "grad_norm": 0.10294393450021744,
      "learning_rate": 7.919772385945307e-05,
      "loss": 1.6302,
      "step": 107040
    },
    {
      "epoch": 0.2112863868324884,
      "grad_norm": 0.11116877943277359,
      "learning_rate": 7.919138363508839e-05,
      "loss": 1.6367,
      "step": 107072
    },
    {
      "epoch": 0.21134953279388485,
      "grad_norm": 0.10466013103723526,
      "learning_rate": 7.91850434107237e-05,
      "loss": 1.6338,
      "step": 107104
    },
    {
      "epoch": 0.21141267875528127,
      "grad_norm": 0.09852197021245956,
      "learning_rate": 7.917870318635901e-05,
      "loss": 1.6211,
      "step": 107136
    },
    {
      "epoch": 0.21147582471667772,
      "grad_norm": 0.10259540379047394,
      "learning_rate": 7.917236296199432e-05,
      "loss": 1.6187,
      "step": 107168
    },
    {
      "epoch": 0.21153897067807417,
      "grad_norm": 0.10774032771587372,
      "learning_rate": 7.916602273762963e-05,
      "loss": 1.6416,
      "step": 107200
    },
    {
      "epoch": 0.2116021166394706,
      "grad_norm": 0.1024177297949791,
      "learning_rate": 7.915968251326495e-05,
      "loss": 1.6322,
      "step": 107232
    },
    {
      "epoch": 0.21166526260086704,
      "grad_norm": 0.0955820307135582,
      "learning_rate": 7.915334228890026e-05,
      "loss": 1.6275,
      "step": 107264
    },
    {
      "epoch": 0.2117284085622635,
      "grad_norm": 0.10093463957309723,
      "learning_rate": 7.914700206453556e-05,
      "loss": 1.6226,
      "step": 107296
    },
    {
      "epoch": 0.21179155452365991,
      "grad_norm": 0.10453131794929504,
      "learning_rate": 7.914066184017088e-05,
      "loss": 1.6311,
      "step": 107328
    },
    {
      "epoch": 0.21185470048505636,
      "grad_norm": 0.11554372310638428,
      "learning_rate": 7.913432161580619e-05,
      "loss": 1.6197,
      "step": 107360
    },
    {
      "epoch": 0.2119178464464528,
      "grad_norm": 0.1025693267583847,
      "learning_rate": 7.912798139144149e-05,
      "loss": 1.6342,
      "step": 107392
    },
    {
      "epoch": 0.21198099240784923,
      "grad_norm": 0.11289840191602707,
      "learning_rate": 7.91216411670768e-05,
      "loss": 1.6404,
      "step": 107424
    },
    {
      "epoch": 0.21204413836924568,
      "grad_norm": 0.10061497986316681,
      "learning_rate": 7.91153009427121e-05,
      "loss": 1.6438,
      "step": 107456
    },
    {
      "epoch": 0.21210728433064213,
      "grad_norm": 0.09793902188539505,
      "learning_rate": 7.910896071834742e-05,
      "loss": 1.6358,
      "step": 107488
    },
    {
      "epoch": 0.21217043029203855,
      "grad_norm": 0.11443709582090378,
      "learning_rate": 7.910262049398274e-05,
      "loss": 1.6305,
      "step": 107520
    },
    {
      "epoch": 0.212233576253435,
      "grad_norm": 0.10491108149290085,
      "learning_rate": 7.909628026961804e-05,
      "loss": 1.6145,
      "step": 107552
    },
    {
      "epoch": 0.21229672221483145,
      "grad_norm": 0.09725751727819443,
      "learning_rate": 7.908994004525335e-05,
      "loss": 1.6182,
      "step": 107584
    },
    {
      "epoch": 0.21235986817622787,
      "grad_norm": 0.11422659456729889,
      "learning_rate": 7.908359982088867e-05,
      "loss": 1.6441,
      "step": 107616
    },
    {
      "epoch": 0.21242301413762432,
      "grad_norm": 0.1142466589808464,
      "learning_rate": 7.907725959652398e-05,
      "loss": 1.6341,
      "step": 107648
    },
    {
      "epoch": 0.21248616009902077,
      "grad_norm": 0.09683635085821152,
      "learning_rate": 7.90709193721593e-05,
      "loss": 1.6293,
      "step": 107680
    },
    {
      "epoch": 0.2125493060604172,
      "grad_norm": 0.10389380156993866,
      "learning_rate": 7.90645791477946e-05,
      "loss": 1.6246,
      "step": 107712
    },
    {
      "epoch": 0.21261245202181364,
      "grad_norm": 0.10858426988124847,
      "learning_rate": 7.905823892342991e-05,
      "loss": 1.6392,
      "step": 107744
    },
    {
      "epoch": 0.2126755979832101,
      "grad_norm": 0.10831128060817719,
      "learning_rate": 7.905189869906522e-05,
      "loss": 1.6263,
      "step": 107776
    },
    {
      "epoch": 0.2127387439446065,
      "grad_norm": 0.1085173562169075,
      "learning_rate": 7.904555847470053e-05,
      "loss": 1.6341,
      "step": 107808
    },
    {
      "epoch": 0.21280188990600296,
      "grad_norm": 0.11627600342035294,
      "learning_rate": 7.903921825033583e-05,
      "loss": 1.618,
      "step": 107840
    },
    {
      "epoch": 0.2128650358673994,
      "grad_norm": 0.10929540544748306,
      "learning_rate": 7.903287802597114e-05,
      "loss": 1.6351,
      "step": 107872
    },
    {
      "epoch": 0.21292818182879583,
      "grad_norm": 0.10462719947099686,
      "learning_rate": 7.902653780160646e-05,
      "loss": 1.656,
      "step": 107904
    },
    {
      "epoch": 0.21299132779019228,
      "grad_norm": 0.10402841120958328,
      "learning_rate": 7.902019757724177e-05,
      "loss": 1.6345,
      "step": 107936
    },
    {
      "epoch": 0.21305447375158873,
      "grad_norm": 0.0970005989074707,
      "learning_rate": 7.901385735287707e-05,
      "loss": 1.6419,
      "step": 107968
    },
    {
      "epoch": 0.21311761971298515,
      "grad_norm": 0.10343700647354126,
      "learning_rate": 7.900751712851239e-05,
      "loss": 1.6301,
      "step": 108000
    },
    {
      "epoch": 0.2131807656743816,
      "grad_norm": 0.10491510480642319,
      "learning_rate": 7.90011769041477e-05,
      "loss": 1.6357,
      "step": 108032
    },
    {
      "epoch": 0.21324391163577805,
      "grad_norm": 0.10545338690280914,
      "learning_rate": 7.899483667978302e-05,
      "loss": 1.6268,
      "step": 108064
    },
    {
      "epoch": 0.21330705759717447,
      "grad_norm": 0.10435418784618378,
      "learning_rate": 7.898849645541832e-05,
      "loss": 1.6279,
      "step": 108096
    },
    {
      "epoch": 0.21337020355857092,
      "grad_norm": 0.10306353867053986,
      "learning_rate": 7.898215623105363e-05,
      "loss": 1.6348,
      "step": 108128
    },
    {
      "epoch": 0.21343334951996737,
      "grad_norm": 0.1135459765791893,
      "learning_rate": 7.897581600668895e-05,
      "loss": 1.6312,
      "step": 108160
    },
    {
      "epoch": 0.2134964954813638,
      "grad_norm": 0.10249552875757217,
      "learning_rate": 7.896947578232426e-05,
      "loss": 1.6269,
      "step": 108192
    },
    {
      "epoch": 0.21355964144276024,
      "grad_norm": 0.1019309014081955,
      "learning_rate": 7.896313555795956e-05,
      "loss": 1.6337,
      "step": 108224
    },
    {
      "epoch": 0.2136227874041567,
      "grad_norm": 0.1040746346116066,
      "learning_rate": 7.895679533359486e-05,
      "loss": 1.6297,
      "step": 108256
    },
    {
      "epoch": 0.2136859333655531,
      "grad_norm": 0.10757636278867722,
      "learning_rate": 7.895045510923018e-05,
      "loss": 1.6263,
      "step": 108288
    },
    {
      "epoch": 0.21374907932694956,
      "grad_norm": 0.10473567992448807,
      "learning_rate": 7.894411488486549e-05,
      "loss": 1.6301,
      "step": 108320
    },
    {
      "epoch": 0.213812225288346,
      "grad_norm": 0.09779185801744461,
      "learning_rate": 7.89377746605008e-05,
      "loss": 1.6464,
      "step": 108352
    },
    {
      "epoch": 0.21387537124974243,
      "grad_norm": 0.10053712874650955,
      "learning_rate": 7.89314344361361e-05,
      "loss": 1.6334,
      "step": 108384
    },
    {
      "epoch": 0.21393851721113888,
      "grad_norm": 0.10482043027877808,
      "learning_rate": 7.892509421177142e-05,
      "loss": 1.6208,
      "step": 108416
    },
    {
      "epoch": 0.21400166317253533,
      "grad_norm": 0.10508330166339874,
      "learning_rate": 7.891875398740674e-05,
      "loss": 1.6293,
      "step": 108448
    },
    {
      "epoch": 0.21406480913393175,
      "grad_norm": 0.09850221127271652,
      "learning_rate": 7.891241376304205e-05,
      "loss": 1.6223,
      "step": 108480
    },
    {
      "epoch": 0.2141279550953282,
      "grad_norm": 0.098767951130867,
      "learning_rate": 7.890607353867735e-05,
      "loss": 1.6353,
      "step": 108512
    },
    {
      "epoch": 0.21419110105672465,
      "grad_norm": 0.10548055917024612,
      "learning_rate": 7.889973331431267e-05,
      "loss": 1.637,
      "step": 108544
    },
    {
      "epoch": 0.21425424701812107,
      "grad_norm": 0.102800153195858,
      "learning_rate": 7.889339308994798e-05,
      "loss": 1.6297,
      "step": 108576
    },
    {
      "epoch": 0.21431739297951752,
      "grad_norm": 0.1027311161160469,
      "learning_rate": 7.88870528655833e-05,
      "loss": 1.6205,
      "step": 108608
    },
    {
      "epoch": 0.21438053894091397,
      "grad_norm": 0.09942883998155594,
      "learning_rate": 7.88807126412186e-05,
      "loss": 1.6228,
      "step": 108640
    },
    {
      "epoch": 0.2144436849023104,
      "grad_norm": 0.11116909980773926,
      "learning_rate": 7.88743724168539e-05,
      "loss": 1.6206,
      "step": 108672
    },
    {
      "epoch": 0.21450683086370684,
      "grad_norm": 0.10296850651502609,
      "learning_rate": 7.886803219248921e-05,
      "loss": 1.6278,
      "step": 108704
    },
    {
      "epoch": 0.2145699768251033,
      "grad_norm": 0.10577276349067688,
      "learning_rate": 7.886169196812453e-05,
      "loss": 1.6245,
      "step": 108736
    },
    {
      "epoch": 0.21463312278649974,
      "grad_norm": 0.1063087061047554,
      "learning_rate": 7.885535174375983e-05,
      "loss": 1.623,
      "step": 108768
    },
    {
      "epoch": 0.21469626874789616,
      "grad_norm": 0.11270727962255478,
      "learning_rate": 7.884901151939514e-05,
      "loss": 1.6391,
      "step": 108800
    },
    {
      "epoch": 0.2147594147092926,
      "grad_norm": 0.1048850566148758,
      "learning_rate": 7.884267129503046e-05,
      "loss": 1.6455,
      "step": 108832
    },
    {
      "epoch": 0.21482256067068906,
      "grad_norm": 0.11248733103275299,
      "learning_rate": 7.883633107066577e-05,
      "loss": 1.6375,
      "step": 108864
    },
    {
      "epoch": 0.21488570663208548,
      "grad_norm": 0.10962484031915665,
      "learning_rate": 7.882999084630109e-05,
      "loss": 1.6284,
      "step": 108896
    },
    {
      "epoch": 0.21494885259348193,
      "grad_norm": 0.10210379213094711,
      "learning_rate": 7.882365062193639e-05,
      "loss": 1.6452,
      "step": 108928
    },
    {
      "epoch": 0.21501199855487838,
      "grad_norm": 0.1018432229757309,
      "learning_rate": 7.88173103975717e-05,
      "loss": 1.6334,
      "step": 108960
    },
    {
      "epoch": 0.2150751445162748,
      "grad_norm": 0.1035958006978035,
      "learning_rate": 7.881097017320702e-05,
      "loss": 1.6233,
      "step": 108992
    },
    {
      "epoch": 0.21513829047767125,
      "grad_norm": 0.10939712077379227,
      "learning_rate": 7.880462994884233e-05,
      "loss": 1.6263,
      "step": 109024
    },
    {
      "epoch": 0.2152014364390677,
      "grad_norm": 0.10484683513641357,
      "learning_rate": 7.879828972447763e-05,
      "loss": 1.6209,
      "step": 109056
    },
    {
      "epoch": 0.21526458240046412,
      "grad_norm": 0.10833864659070969,
      "learning_rate": 7.879194950011293e-05,
      "loss": 1.624,
      "step": 109088
    },
    {
      "epoch": 0.21532772836186057,
      "grad_norm": 0.1055784523487091,
      "learning_rate": 7.878560927574825e-05,
      "loss": 1.641,
      "step": 109120
    },
    {
      "epoch": 0.21539087432325701,
      "grad_norm": 0.11066529899835587,
      "learning_rate": 7.877926905138356e-05,
      "loss": 1.6356,
      "step": 109152
    },
    {
      "epoch": 0.21545402028465344,
      "grad_norm": 0.09954646229743958,
      "learning_rate": 7.877292882701886e-05,
      "loss": 1.6374,
      "step": 109184
    },
    {
      "epoch": 0.21551716624604988,
      "grad_norm": 0.09930533170700073,
      "learning_rate": 7.876658860265418e-05,
      "loss": 1.633,
      "step": 109216
    },
    {
      "epoch": 0.21558031220744633,
      "grad_norm": 0.11131240427494049,
      "learning_rate": 7.876024837828949e-05,
      "loss": 1.6155,
      "step": 109248
    },
    {
      "epoch": 0.21564345816884276,
      "grad_norm": 0.09750407934188843,
      "learning_rate": 7.87539081539248e-05,
      "loss": 1.6244,
      "step": 109280
    },
    {
      "epoch": 0.2157066041302392,
      "grad_norm": 0.10745207220315933,
      "learning_rate": 7.874756792956011e-05,
      "loss": 1.6204,
      "step": 109312
    },
    {
      "epoch": 0.21576975009163565,
      "grad_norm": 0.10814482718706131,
      "learning_rate": 7.874122770519542e-05,
      "loss": 1.6231,
      "step": 109344
    },
    {
      "epoch": 0.21583289605303208,
      "grad_norm": 0.1093771904706955,
      "learning_rate": 7.873488748083074e-05,
      "loss": 1.6354,
      "step": 109376
    },
    {
      "epoch": 0.21589604201442852,
      "grad_norm": 0.10462377965450287,
      "learning_rate": 7.872854725646605e-05,
      "loss": 1.6178,
      "step": 109408
    },
    {
      "epoch": 0.21595918797582497,
      "grad_norm": 0.1065559834241867,
      "learning_rate": 7.872220703210135e-05,
      "loss": 1.6292,
      "step": 109440
    },
    {
      "epoch": 0.2160223339372214,
      "grad_norm": 0.0998968705534935,
      "learning_rate": 7.871586680773667e-05,
      "loss": 1.6084,
      "step": 109472
    },
    {
      "epoch": 0.21608547989861784,
      "grad_norm": 0.10516436398029327,
      "learning_rate": 7.870952658337197e-05,
      "loss": 1.6281,
      "step": 109504
    },
    {
      "epoch": 0.2161486258600143,
      "grad_norm": 0.1136736273765564,
      "learning_rate": 7.870318635900728e-05,
      "loss": 1.6293,
      "step": 109536
    },
    {
      "epoch": 0.21621177182141071,
      "grad_norm": 0.10644941031932831,
      "learning_rate": 7.86968461346426e-05,
      "loss": 1.6367,
      "step": 109568
    },
    {
      "epoch": 0.21627491778280716,
      "grad_norm": 0.09942994266748428,
      "learning_rate": 7.86905059102779e-05,
      "loss": 1.6251,
      "step": 109600
    },
    {
      "epoch": 0.2163380637442036,
      "grad_norm": 0.10130082815885544,
      "learning_rate": 7.868416568591321e-05,
      "loss": 1.633,
      "step": 109632
    },
    {
      "epoch": 0.21640120970560003,
      "grad_norm": 0.11468770354986191,
      "learning_rate": 7.867782546154853e-05,
      "loss": 1.6193,
      "step": 109664
    },
    {
      "epoch": 0.21646435566699648,
      "grad_norm": 0.10606314241886139,
      "learning_rate": 7.867148523718384e-05,
      "loss": 1.6291,
      "step": 109696
    },
    {
      "epoch": 0.21652750162839293,
      "grad_norm": 0.10423869639635086,
      "learning_rate": 7.866514501281914e-05,
      "loss": 1.6407,
      "step": 109728
    },
    {
      "epoch": 0.21659064758978935,
      "grad_norm": 0.10519355535507202,
      "learning_rate": 7.865880478845446e-05,
      "loss": 1.6379,
      "step": 109760
    },
    {
      "epoch": 0.2166537935511858,
      "grad_norm": 0.10765162855386734,
      "learning_rate": 7.865246456408977e-05,
      "loss": 1.6475,
      "step": 109792
    },
    {
      "epoch": 0.21671693951258225,
      "grad_norm": 0.10810939222574234,
      "learning_rate": 7.864612433972509e-05,
      "loss": 1.6257,
      "step": 109824
    },
    {
      "epoch": 0.21678008547397867,
      "grad_norm": 0.10260454565286636,
      "learning_rate": 7.863978411536039e-05,
      "loss": 1.6276,
      "step": 109856
    },
    {
      "epoch": 0.21684323143537512,
      "grad_norm": 0.10651948302984238,
      "learning_rate": 7.86334438909957e-05,
      "loss": 1.6372,
      "step": 109888
    },
    {
      "epoch": 0.21690637739677157,
      "grad_norm": 0.11261123418807983,
      "learning_rate": 7.8627103666631e-05,
      "loss": 1.6366,
      "step": 109920
    },
    {
      "epoch": 0.216969523358168,
      "grad_norm": 0.09782645851373672,
      "learning_rate": 7.862076344226632e-05,
      "loss": 1.6174,
      "step": 109952
    },
    {
      "epoch": 0.21703266931956444,
      "grad_norm": 0.1075483039021492,
      "learning_rate": 7.861442321790162e-05,
      "loss": 1.6262,
      "step": 109984
    },
    {
      "epoch": 0.2170958152809609,
      "grad_norm": 0.11073613166809082,
      "learning_rate": 7.860808299353693e-05,
      "loss": 1.6423,
      "step": 110016
    },
    {
      "epoch": 0.2171589612423573,
      "grad_norm": 0.10925047099590302,
      "learning_rate": 7.860174276917225e-05,
      "loss": 1.6297,
      "step": 110048
    },
    {
      "epoch": 0.21722210720375376,
      "grad_norm": 0.10803259164094925,
      "learning_rate": 7.859540254480756e-05,
      "loss": 1.635,
      "step": 110080
    },
    {
      "epoch": 0.2172852531651502,
      "grad_norm": 0.10276775062084198,
      "learning_rate": 7.858906232044286e-05,
      "loss": 1.6267,
      "step": 110112
    },
    {
      "epoch": 0.21734839912654663,
      "grad_norm": 0.1135217472910881,
      "learning_rate": 7.858272209607818e-05,
      "loss": 1.6354,
      "step": 110144
    },
    {
      "epoch": 0.21741154508794308,
      "grad_norm": 0.10434950888156891,
      "learning_rate": 7.857638187171349e-05,
      "loss": 1.6338,
      "step": 110176
    },
    {
      "epoch": 0.21747469104933953,
      "grad_norm": 0.10705443471670151,
      "learning_rate": 7.85700416473488e-05,
      "loss": 1.6225,
      "step": 110208
    },
    {
      "epoch": 0.21753783701073595,
      "grad_norm": 0.10687093436717987,
      "learning_rate": 7.856370142298412e-05,
      "loss": 1.6357,
      "step": 110240
    },
    {
      "epoch": 0.2176009829721324,
      "grad_norm": 0.10969510674476624,
      "learning_rate": 7.855736119861942e-05,
      "loss": 1.6346,
      "step": 110272
    },
    {
      "epoch": 0.21766412893352885,
      "grad_norm": 0.10750909894704819,
      "learning_rate": 7.855102097425474e-05,
      "loss": 1.6319,
      "step": 110304
    },
    {
      "epoch": 0.21772727489492527,
      "grad_norm": 0.10937602072954178,
      "learning_rate": 7.854468074989004e-05,
      "loss": 1.6322,
      "step": 110336
    },
    {
      "epoch": 0.21779042085632172,
      "grad_norm": 0.10048750042915344,
      "learning_rate": 7.853834052552535e-05,
      "loss": 1.6354,
      "step": 110368
    },
    {
      "epoch": 0.21785356681771817,
      "grad_norm": 0.10500896722078323,
      "learning_rate": 7.853200030116065e-05,
      "loss": 1.6237,
      "step": 110400
    },
    {
      "epoch": 0.21791671277911462,
      "grad_norm": 0.10095417499542236,
      "learning_rate": 7.852566007679597e-05,
      "loss": 1.6419,
      "step": 110432
    },
    {
      "epoch": 0.21797985874051104,
      "grad_norm": 0.10962829738855362,
      "learning_rate": 7.851931985243128e-05,
      "loss": 1.6383,
      "step": 110464
    },
    {
      "epoch": 0.2180430047019075,
      "grad_norm": 0.10402845591306686,
      "learning_rate": 7.85129796280666e-05,
      "loss": 1.6309,
      "step": 110496
    },
    {
      "epoch": 0.21810615066330394,
      "grad_norm": 0.09595850855112076,
      "learning_rate": 7.85066394037019e-05,
      "loss": 1.639,
      "step": 110528
    },
    {
      "epoch": 0.21816929662470036,
      "grad_norm": 0.10197676718235016,
      "learning_rate": 7.850029917933721e-05,
      "loss": 1.6458,
      "step": 110560
    },
    {
      "epoch": 0.2182324425860968,
      "grad_norm": 0.10455411672592163,
      "learning_rate": 7.849395895497253e-05,
      "loss": 1.6337,
      "step": 110592
    },
    {
      "epoch": 0.21829558854749326,
      "grad_norm": 0.10910439491271973,
      "learning_rate": 7.848761873060784e-05,
      "loss": 1.637,
      "step": 110624
    },
    {
      "epoch": 0.21835873450888968,
      "grad_norm": 0.09921760857105255,
      "learning_rate": 7.848127850624314e-05,
      "loss": 1.6142,
      "step": 110656
    },
    {
      "epoch": 0.21842188047028613,
      "grad_norm": 0.11171793192625046,
      "learning_rate": 7.847493828187846e-05,
      "loss": 1.6172,
      "step": 110688
    },
    {
      "epoch": 0.21848502643168258,
      "grad_norm": 0.10553435236215591,
      "learning_rate": 7.846859805751376e-05,
      "loss": 1.6285,
      "step": 110720
    },
    {
      "epoch": 0.218548172393079,
      "grad_norm": 0.1026960238814354,
      "learning_rate": 7.846225783314907e-05,
      "loss": 1.6174,
      "step": 110752
    },
    {
      "epoch": 0.21861131835447545,
      "grad_norm": 0.11526110023260117,
      "learning_rate": 7.845591760878437e-05,
      "loss": 1.6267,
      "step": 110784
    },
    {
      "epoch": 0.2186744643158719,
      "grad_norm": 0.10280951112508774,
      "learning_rate": 7.844957738441969e-05,
      "loss": 1.6265,
      "step": 110816
    },
    {
      "epoch": 0.21873761027726832,
      "grad_norm": 0.10278615355491638,
      "learning_rate": 7.8443237160055e-05,
      "loss": 1.6355,
      "step": 110848
    },
    {
      "epoch": 0.21880075623866477,
      "grad_norm": 0.10710864514112473,
      "learning_rate": 7.843689693569032e-05,
      "loss": 1.6314,
      "step": 110880
    },
    {
      "epoch": 0.21886390220006122,
      "grad_norm": 0.10066132247447968,
      "learning_rate": 7.843055671132563e-05,
      "loss": 1.6355,
      "step": 110912
    },
    {
      "epoch": 0.21892704816145764,
      "grad_norm": 0.09862487763166428,
      "learning_rate": 7.842421648696093e-05,
      "loss": 1.6373,
      "step": 110944
    },
    {
      "epoch": 0.2189901941228541,
      "grad_norm": 0.11181669682264328,
      "learning_rate": 7.841787626259625e-05,
      "loss": 1.632,
      "step": 110976
    },
    {
      "epoch": 0.21905334008425054,
      "grad_norm": 0.10781589895486832,
      "learning_rate": 7.841153603823156e-05,
      "loss": 1.6292,
      "step": 111008
    },
    {
      "epoch": 0.21911648604564696,
      "grad_norm": 0.10374114662408829,
      "learning_rate": 7.840519581386688e-05,
      "loss": 1.6212,
      "step": 111040
    },
    {
      "epoch": 0.2191796320070434,
      "grad_norm": 0.10779701173305511,
      "learning_rate": 7.839885558950218e-05,
      "loss": 1.6298,
      "step": 111072
    },
    {
      "epoch": 0.21924277796843986,
      "grad_norm": 0.10612470656633377,
      "learning_rate": 7.839251536513749e-05,
      "loss": 1.6366,
      "step": 111104
    },
    {
      "epoch": 0.21930592392983628,
      "grad_norm": 0.1071191355586052,
      "learning_rate": 7.838617514077279e-05,
      "loss": 1.6401,
      "step": 111136
    },
    {
      "epoch": 0.21936906989123273,
      "grad_norm": 0.1048671081662178,
      "learning_rate": 7.837983491640811e-05,
      "loss": 1.6349,
      "step": 111168
    },
    {
      "epoch": 0.21943221585262918,
      "grad_norm": 0.10103890299797058,
      "learning_rate": 7.837349469204341e-05,
      "loss": 1.6188,
      "step": 111200
    },
    {
      "epoch": 0.2194953618140256,
      "grad_norm": 0.10815919935703278,
      "learning_rate": 7.836715446767872e-05,
      "loss": 1.6303,
      "step": 111232
    },
    {
      "epoch": 0.21955850777542205,
      "grad_norm": 0.10404631495475769,
      "learning_rate": 7.836081424331404e-05,
      "loss": 1.6282,
      "step": 111264
    },
    {
      "epoch": 0.2196216537368185,
      "grad_norm": 0.1171780377626419,
      "learning_rate": 7.835447401894935e-05,
      "loss": 1.6413,
      "step": 111296
    },
    {
      "epoch": 0.21968479969821492,
      "grad_norm": 0.10239430516958237,
      "learning_rate": 7.834813379458465e-05,
      "loss": 1.632,
      "step": 111328
    },
    {
      "epoch": 0.21974794565961137,
      "grad_norm": 0.10737922787666321,
      "learning_rate": 7.834179357021997e-05,
      "loss": 1.6254,
      "step": 111360
    },
    {
      "epoch": 0.21981109162100781,
      "grad_norm": 0.10107067227363586,
      "learning_rate": 7.833545334585528e-05,
      "loss": 1.6256,
      "step": 111392
    },
    {
      "epoch": 0.21987423758240424,
      "grad_norm": 0.1047099232673645,
      "learning_rate": 7.83291131214906e-05,
      "loss": 1.6514,
      "step": 111424
    },
    {
      "epoch": 0.21993738354380069,
      "grad_norm": 0.10276010632514954,
      "learning_rate": 7.83227728971259e-05,
      "loss": 1.6278,
      "step": 111456
    },
    {
      "epoch": 0.22000052950519713,
      "grad_norm": 0.10510295629501343,
      "learning_rate": 7.831643267276121e-05,
      "loss": 1.6251,
      "step": 111488
    },
    {
      "epoch": 0.22006367546659356,
      "grad_norm": 0.10550041496753693,
      "learning_rate": 7.831009244839653e-05,
      "loss": 1.6351,
      "step": 111520
    },
    {
      "epoch": 0.22012682142799,
      "grad_norm": 0.10613173246383667,
      "learning_rate": 7.830375222403183e-05,
      "loss": 1.6296,
      "step": 111552
    },
    {
      "epoch": 0.22018996738938645,
      "grad_norm": 0.12164752185344696,
      "learning_rate": 7.829741199966714e-05,
      "loss": 1.627,
      "step": 111584
    },
    {
      "epoch": 0.22025311335078288,
      "grad_norm": 0.10672342777252197,
      "learning_rate": 7.829107177530244e-05,
      "loss": 1.6196,
      "step": 111616
    },
    {
      "epoch": 0.22031625931217932,
      "grad_norm": 0.10044343769550323,
      "learning_rate": 7.828473155093776e-05,
      "loss": 1.6281,
      "step": 111648
    },
    {
      "epoch": 0.22037940527357577,
      "grad_norm": 0.11962555348873138,
      "learning_rate": 7.827839132657307e-05,
      "loss": 1.6275,
      "step": 111680
    },
    {
      "epoch": 0.2204425512349722,
      "grad_norm": 0.10664345324039459,
      "learning_rate": 7.827205110220839e-05,
      "loss": 1.6387,
      "step": 111712
    },
    {
      "epoch": 0.22050569719636864,
      "grad_norm": 0.10825929790735245,
      "learning_rate": 7.826571087784369e-05,
      "loss": 1.6404,
      "step": 111744
    },
    {
      "epoch": 0.2205688431577651,
      "grad_norm": 0.10334636270999908,
      "learning_rate": 7.8259370653479e-05,
      "loss": 1.6308,
      "step": 111776
    },
    {
      "epoch": 0.22063198911916151,
      "grad_norm": 0.11190655827522278,
      "learning_rate": 7.825303042911432e-05,
      "loss": 1.629,
      "step": 111808
    },
    {
      "epoch": 0.22069513508055796,
      "grad_norm": 0.1063610389828682,
      "learning_rate": 7.824669020474963e-05,
      "loss": 1.633,
      "step": 111840
    },
    {
      "epoch": 0.2207582810419544,
      "grad_norm": 0.1004890576004982,
      "learning_rate": 7.824034998038493e-05,
      "loss": 1.6394,
      "step": 111872
    },
    {
      "epoch": 0.22082142700335083,
      "grad_norm": 0.10761027038097382,
      "learning_rate": 7.823400975602025e-05,
      "loss": 1.6454,
      "step": 111904
    },
    {
      "epoch": 0.22088457296474728,
      "grad_norm": 0.12183315306901932,
      "learning_rate": 7.822766953165556e-05,
      "loss": 1.6354,
      "step": 111936
    },
    {
      "epoch": 0.22094771892614373,
      "grad_norm": 0.11122620850801468,
      "learning_rate": 7.822132930729086e-05,
      "loss": 1.6235,
      "step": 111968
    },
    {
      "epoch": 0.22101086488754015,
      "grad_norm": 0.10745202004909515,
      "learning_rate": 7.821498908292616e-05,
      "loss": 1.6342,
      "step": 112000
    },
    {
      "epoch": 0.2210740108489366,
      "grad_norm": 0.10377126932144165,
      "learning_rate": 7.820864885856148e-05,
      "loss": 1.6305,
      "step": 112032
    },
    {
      "epoch": 0.22113715681033305,
      "grad_norm": 0.10169932246208191,
      "learning_rate": 7.82023086341968e-05,
      "loss": 1.6444,
      "step": 112064
    },
    {
      "epoch": 0.2212003027717295,
      "grad_norm": 0.0997363030910492,
      "learning_rate": 7.819596840983211e-05,
      "loss": 1.6336,
      "step": 112096
    },
    {
      "epoch": 0.22126344873312592,
      "grad_norm": 0.10159129649400711,
      "learning_rate": 7.818962818546741e-05,
      "loss": 1.6231,
      "step": 112128
    },
    {
      "epoch": 0.22132659469452237,
      "grad_norm": 0.10715489089488983,
      "learning_rate": 7.818328796110272e-05,
      "loss": 1.6213,
      "step": 112160
    },
    {
      "epoch": 0.22138974065591882,
      "grad_norm": 0.10581538081169128,
      "learning_rate": 7.817694773673804e-05,
      "loss": 1.6162,
      "step": 112192
    },
    {
      "epoch": 0.22145288661731524,
      "grad_norm": 0.10206440836191177,
      "learning_rate": 7.817060751237335e-05,
      "loss": 1.6133,
      "step": 112224
    },
    {
      "epoch": 0.2215160325787117,
      "grad_norm": 0.1013316810131073,
      "learning_rate": 7.816426728800867e-05,
      "loss": 1.622,
      "step": 112256
    },
    {
      "epoch": 0.22157917854010814,
      "grad_norm": 0.11109724640846252,
      "learning_rate": 7.815792706364397e-05,
      "loss": 1.624,
      "step": 112288
    },
    {
      "epoch": 0.22164232450150456,
      "grad_norm": 0.10483244806528091,
      "learning_rate": 7.815158683927928e-05,
      "loss": 1.6306,
      "step": 112320
    },
    {
      "epoch": 0.221705470462901,
      "grad_norm": 0.11185155063867569,
      "learning_rate": 7.81452466149146e-05,
      "loss": 1.6205,
      "step": 112352
    },
    {
      "epoch": 0.22176861642429746,
      "grad_norm": 0.104886494576931,
      "learning_rate": 7.81389063905499e-05,
      "loss": 1.6258,
      "step": 112384
    },
    {
      "epoch": 0.22183176238569388,
      "grad_norm": 0.12286294251680374,
      "learning_rate": 7.81325661661852e-05,
      "loss": 1.632,
      "step": 112416
    },
    {
      "epoch": 0.22189490834709033,
      "grad_norm": 0.11867167800664902,
      "learning_rate": 7.812622594182051e-05,
      "loss": 1.6258,
      "step": 112448
    },
    {
      "epoch": 0.22195805430848678,
      "grad_norm": 0.10756305605173111,
      "learning_rate": 7.811988571745583e-05,
      "loss": 1.6264,
      "step": 112480
    },
    {
      "epoch": 0.2220212002698832,
      "grad_norm": 0.09912282973527908,
      "learning_rate": 7.811354549309114e-05,
      "loss": 1.6366,
      "step": 112512
    },
    {
      "epoch": 0.22208434623127965,
      "grad_norm": 0.10603366047143936,
      "learning_rate": 7.810720526872644e-05,
      "loss": 1.6315,
      "step": 112544
    },
    {
      "epoch": 0.2221474921926761,
      "grad_norm": 0.10342442989349365,
      "learning_rate": 7.810086504436176e-05,
      "loss": 1.6292,
      "step": 112576
    },
    {
      "epoch": 0.22221063815407252,
      "grad_norm": 0.10729510337114334,
      "learning_rate": 7.809452481999707e-05,
      "loss": 1.6243,
      "step": 112608
    },
    {
      "epoch": 0.22227378411546897,
      "grad_norm": 0.10366231203079224,
      "learning_rate": 7.808818459563239e-05,
      "loss": 1.627,
      "step": 112640
    },
    {
      "epoch": 0.22233693007686542,
      "grad_norm": 0.11111960560083389,
      "learning_rate": 7.808184437126769e-05,
      "loss": 1.6344,
      "step": 112672
    },
    {
      "epoch": 0.22240007603826184,
      "grad_norm": 0.10580338537693024,
      "learning_rate": 7.8075504146903e-05,
      "loss": 1.6203,
      "step": 112704
    },
    {
      "epoch": 0.2224632219996583,
      "grad_norm": 0.11580169200897217,
      "learning_rate": 7.806916392253832e-05,
      "loss": 1.6272,
      "step": 112736
    },
    {
      "epoch": 0.22252636796105474,
      "grad_norm": 0.10675030201673508,
      "learning_rate": 7.806282369817363e-05,
      "loss": 1.6122,
      "step": 112768
    },
    {
      "epoch": 0.22258951392245116,
      "grad_norm": 0.11238189786672592,
      "learning_rate": 7.805648347380893e-05,
      "loss": 1.6018,
      "step": 112800
    },
    {
      "epoch": 0.2226526598838476,
      "grad_norm": 0.09811877459287643,
      "learning_rate": 7.805014324944423e-05,
      "loss": 1.6356,
      "step": 112832
    },
    {
      "epoch": 0.22271580584524406,
      "grad_norm": 0.11010882258415222,
      "learning_rate": 7.804380302507955e-05,
      "loss": 1.6277,
      "step": 112864
    },
    {
      "epoch": 0.22277895180664048,
      "grad_norm": 0.10838333517313004,
      "learning_rate": 7.803746280071486e-05,
      "loss": 1.6393,
      "step": 112896
    },
    {
      "epoch": 0.22284209776803693,
      "grad_norm": 0.1051255315542221,
      "learning_rate": 7.803112257635018e-05,
      "loss": 1.6095,
      "step": 112928
    },
    {
      "epoch": 0.22290524372943338,
      "grad_norm": 0.103900246322155,
      "learning_rate": 7.802478235198548e-05,
      "loss": 1.6386,
      "step": 112960
    },
    {
      "epoch": 0.2229683896908298,
      "grad_norm": 0.10753915458917618,
      "learning_rate": 7.80184421276208e-05,
      "loss": 1.6295,
      "step": 112992
    },
    {
      "epoch": 0.22303153565222625,
      "grad_norm": 0.10351253300905228,
      "learning_rate": 7.801210190325611e-05,
      "loss": 1.629,
      "step": 113024
    },
    {
      "epoch": 0.2230946816136227,
      "grad_norm": 0.1029263287782669,
      "learning_rate": 7.800576167889142e-05,
      "loss": 1.6203,
      "step": 113056
    },
    {
      "epoch": 0.22315782757501912,
      "grad_norm": 0.1170961931347847,
      "learning_rate": 7.799942145452672e-05,
      "loss": 1.6383,
      "step": 113088
    },
    {
      "epoch": 0.22322097353641557,
      "grad_norm": 0.10815365612506866,
      "learning_rate": 7.799308123016204e-05,
      "loss": 1.6331,
      "step": 113120
    },
    {
      "epoch": 0.22328411949781202,
      "grad_norm": 0.10970927774906158,
      "learning_rate": 7.798674100579735e-05,
      "loss": 1.6354,
      "step": 113152
    },
    {
      "epoch": 0.22334726545920844,
      "grad_norm": 0.10320336371660233,
      "learning_rate": 7.798040078143267e-05,
      "loss": 1.6152,
      "step": 113184
    },
    {
      "epoch": 0.2234104114206049,
      "grad_norm": 0.1083802804350853,
      "learning_rate": 7.797406055706797e-05,
      "loss": 1.629,
      "step": 113216
    },
    {
      "epoch": 0.22347355738200134,
      "grad_norm": 0.11191002279520035,
      "learning_rate": 7.796772033270327e-05,
      "loss": 1.6214,
      "step": 113248
    },
    {
      "epoch": 0.22353670334339776,
      "grad_norm": 0.10184124857187271,
      "learning_rate": 7.796138010833858e-05,
      "loss": 1.6162,
      "step": 113280
    },
    {
      "epoch": 0.2235998493047942,
      "grad_norm": 0.10529950261116028,
      "learning_rate": 7.79550398839739e-05,
      "loss": 1.627,
      "step": 113312
    },
    {
      "epoch": 0.22366299526619066,
      "grad_norm": 0.10743270069360733,
      "learning_rate": 7.79486996596092e-05,
      "loss": 1.6362,
      "step": 113344
    },
    {
      "epoch": 0.22372614122758708,
      "grad_norm": 0.0962747111916542,
      "learning_rate": 7.794235943524451e-05,
      "loss": 1.6191,
      "step": 113376
    },
    {
      "epoch": 0.22378928718898353,
      "grad_norm": 0.11665450781583786,
      "learning_rate": 7.793601921087983e-05,
      "loss": 1.6256,
      "step": 113408
    },
    {
      "epoch": 0.22385243315037998,
      "grad_norm": 0.09606826305389404,
      "learning_rate": 7.792967898651514e-05,
      "loss": 1.6178,
      "step": 113440
    },
    {
      "epoch": 0.2239155791117764,
      "grad_norm": 0.10067195445299149,
      "learning_rate": 7.792333876215044e-05,
      "loss": 1.6176,
      "step": 113472
    },
    {
      "epoch": 0.22397872507317285,
      "grad_norm": 0.1075863465666771,
      "learning_rate": 7.791699853778576e-05,
      "loss": 1.629,
      "step": 113504
    },
    {
      "epoch": 0.2240418710345693,
      "grad_norm": 0.11087308079004288,
      "learning_rate": 7.791065831342107e-05,
      "loss": 1.6161,
      "step": 113536
    },
    {
      "epoch": 0.22410501699596572,
      "grad_norm": 0.10315633565187454,
      "learning_rate": 7.790431808905639e-05,
      "loss": 1.6276,
      "step": 113568
    },
    {
      "epoch": 0.22416816295736217,
      "grad_norm": 0.10783594846725464,
      "learning_rate": 7.789797786469169e-05,
      "loss": 1.6371,
      "step": 113600
    },
    {
      "epoch": 0.22423130891875862,
      "grad_norm": 0.110627181828022,
      "learning_rate": 7.7891637640327e-05,
      "loss": 1.6214,
      "step": 113632
    },
    {
      "epoch": 0.22429445488015504,
      "grad_norm": 0.11503355205059052,
      "learning_rate": 7.78852974159623e-05,
      "loss": 1.6195,
      "step": 113664
    },
    {
      "epoch": 0.22435760084155149,
      "grad_norm": 0.11128552258014679,
      "learning_rate": 7.787895719159762e-05,
      "loss": 1.6152,
      "step": 113696
    },
    {
      "epoch": 0.22442074680294793,
      "grad_norm": 0.10582909733057022,
      "learning_rate": 7.787261696723293e-05,
      "loss": 1.6315,
      "step": 113728
    },
    {
      "epoch": 0.22448389276434436,
      "grad_norm": 0.10193505883216858,
      "learning_rate": 7.786627674286823e-05,
      "loss": 1.6284,
      "step": 113760
    },
    {
      "epoch": 0.2245470387257408,
      "grad_norm": 0.10276355594396591,
      "learning_rate": 7.785993651850355e-05,
      "loss": 1.6116,
      "step": 113792
    },
    {
      "epoch": 0.22461018468713725,
      "grad_norm": 0.10170938074588776,
      "learning_rate": 7.785359629413886e-05,
      "loss": 1.6322,
      "step": 113824
    },
    {
      "epoch": 0.2246733306485337,
      "grad_norm": 0.10109400749206543,
      "learning_rate": 7.784725606977418e-05,
      "loss": 1.6193,
      "step": 113856
    },
    {
      "epoch": 0.22473647660993012,
      "grad_norm": 0.11129441112279892,
      "learning_rate": 7.784091584540948e-05,
      "loss": 1.6155,
      "step": 113888
    },
    {
      "epoch": 0.22479962257132657,
      "grad_norm": 0.10492438077926636,
      "learning_rate": 7.78345756210448e-05,
      "loss": 1.6357,
      "step": 113920
    },
    {
      "epoch": 0.22486276853272302,
      "grad_norm": 0.10089823603630066,
      "learning_rate": 7.782823539668011e-05,
      "loss": 1.6277,
      "step": 113952
    },
    {
      "epoch": 0.22492591449411944,
      "grad_norm": 0.10287986695766449,
      "learning_rate": 7.782189517231542e-05,
      "loss": 1.6342,
      "step": 113984
    },
    {
      "epoch": 0.2249890604555159,
      "grad_norm": 0.10186152905225754,
      "learning_rate": 7.781555494795072e-05,
      "loss": 1.6243,
      "step": 114016
    },
    {
      "epoch": 0.22505220641691234,
      "grad_norm": 0.1056632474064827,
      "learning_rate": 7.780921472358604e-05,
      "loss": 1.6346,
      "step": 114048
    },
    {
      "epoch": 0.22511535237830876,
      "grad_norm": 0.1158563420176506,
      "learning_rate": 7.780287449922134e-05,
      "loss": 1.6201,
      "step": 114080
    },
    {
      "epoch": 0.2251784983397052,
      "grad_norm": 0.11432074010372162,
      "learning_rate": 7.779653427485665e-05,
      "loss": 1.6165,
      "step": 114112
    },
    {
      "epoch": 0.22524164430110166,
      "grad_norm": 0.10810406506061554,
      "learning_rate": 7.779019405049196e-05,
      "loss": 1.6169,
      "step": 114144
    },
    {
      "epoch": 0.22530479026249808,
      "grad_norm": 0.10560592263936996,
      "learning_rate": 7.778385382612727e-05,
      "loss": 1.6383,
      "step": 114176
    },
    {
      "epoch": 0.22536793622389453,
      "grad_norm": 0.10545812547206879,
      "learning_rate": 7.777751360176258e-05,
      "loss": 1.631,
      "step": 114208
    },
    {
      "epoch": 0.22543108218529098,
      "grad_norm": 0.1102619618177414,
      "learning_rate": 7.77711733773979e-05,
      "loss": 1.6376,
      "step": 114240
    },
    {
      "epoch": 0.2254942281466874,
      "grad_norm": 0.10669015347957611,
      "learning_rate": 7.776483315303321e-05,
      "loss": 1.6094,
      "step": 114272
    },
    {
      "epoch": 0.22555737410808385,
      "grad_norm": 0.10579013079404831,
      "learning_rate": 7.775849292866851e-05,
      "loss": 1.615,
      "step": 114304
    },
    {
      "epoch": 0.2256205200694803,
      "grad_norm": 0.10540979355573654,
      "learning_rate": 7.775215270430383e-05,
      "loss": 1.623,
      "step": 114336
    },
    {
      "epoch": 0.22568366603087672,
      "grad_norm": 0.11036309599876404,
      "learning_rate": 7.774581247993914e-05,
      "loss": 1.6198,
      "step": 114368
    },
    {
      "epoch": 0.22574681199227317,
      "grad_norm": 0.11022590100765228,
      "learning_rate": 7.773947225557446e-05,
      "loss": 1.6219,
      "step": 114400
    },
    {
      "epoch": 0.22580995795366962,
      "grad_norm": 0.10671652853488922,
      "learning_rate": 7.773313203120976e-05,
      "loss": 1.6306,
      "step": 114432
    },
    {
      "epoch": 0.22587310391506604,
      "grad_norm": 0.10119830816984177,
      "learning_rate": 7.772679180684507e-05,
      "loss": 1.619,
      "step": 114464
    },
    {
      "epoch": 0.2259362498764625,
      "grad_norm": 0.11654257774353027,
      "learning_rate": 7.772045158248037e-05,
      "loss": 1.636,
      "step": 114496
    },
    {
      "epoch": 0.22599939583785894,
      "grad_norm": 0.10148677229881287,
      "learning_rate": 7.771411135811569e-05,
      "loss": 1.6217,
      "step": 114528
    },
    {
      "epoch": 0.22606254179925536,
      "grad_norm": 0.10720031708478928,
      "learning_rate": 7.770777113375099e-05,
      "loss": 1.6238,
      "step": 114560
    },
    {
      "epoch": 0.2261256877606518,
      "grad_norm": 0.10637203603982925,
      "learning_rate": 7.77014309093863e-05,
      "loss": 1.6256,
      "step": 114592
    },
    {
      "epoch": 0.22618883372204826,
      "grad_norm": 0.1099848672747612,
      "learning_rate": 7.769509068502162e-05,
      "loss": 1.644,
      "step": 114624
    },
    {
      "epoch": 0.22625197968344468,
      "grad_norm": 0.10704099386930466,
      "learning_rate": 7.768875046065693e-05,
      "loss": 1.619,
      "step": 114656
    },
    {
      "epoch": 0.22631512564484113,
      "grad_norm": 0.10641255229711533,
      "learning_rate": 7.768241023629224e-05,
      "loss": 1.6262,
      "step": 114688
    },
    {
      "epoch": 0.22637827160623758,
      "grad_norm": 0.11466799676418304,
      "learning_rate": 7.767607001192755e-05,
      "loss": 1.6193,
      "step": 114720
    },
    {
      "epoch": 0.226441417567634,
      "grad_norm": 0.11430511623620987,
      "learning_rate": 7.766972978756286e-05,
      "loss": 1.6166,
      "step": 114752
    },
    {
      "epoch": 0.22650456352903045,
      "grad_norm": 0.10588062554597855,
      "learning_rate": 7.766338956319818e-05,
      "loss": 1.6203,
      "step": 114784
    },
    {
      "epoch": 0.2265677094904269,
      "grad_norm": 0.1109849363565445,
      "learning_rate": 7.765704933883348e-05,
      "loss": 1.6167,
      "step": 114816
    },
    {
      "epoch": 0.22663085545182332,
      "grad_norm": 0.11077682673931122,
      "learning_rate": 7.76507091144688e-05,
      "loss": 1.636,
      "step": 114848
    },
    {
      "epoch": 0.22669400141321977,
      "grad_norm": 0.10223282128572464,
      "learning_rate": 7.76443688901041e-05,
      "loss": 1.6335,
      "step": 114880
    },
    {
      "epoch": 0.22675714737461622,
      "grad_norm": 0.11035159230232239,
      "learning_rate": 7.763802866573941e-05,
      "loss": 1.6218,
      "step": 114912
    },
    {
      "epoch": 0.22682029333601264,
      "grad_norm": 0.11320916563272476,
      "learning_rate": 7.763168844137472e-05,
      "loss": 1.6266,
      "step": 114944
    },
    {
      "epoch": 0.2268834392974091,
      "grad_norm": 0.1096683070063591,
      "learning_rate": 7.762534821701003e-05,
      "loss": 1.6222,
      "step": 114976
    },
    {
      "epoch": 0.22694658525880554,
      "grad_norm": 0.10047774016857147,
      "learning_rate": 7.761900799264534e-05,
      "loss": 1.6443,
      "step": 115008
    },
    {
      "epoch": 0.22700973122020196,
      "grad_norm": 0.11166009306907654,
      "learning_rate": 7.761266776828065e-05,
      "loss": 1.6109,
      "step": 115040
    },
    {
      "epoch": 0.2270728771815984,
      "grad_norm": 0.10214067250490189,
      "learning_rate": 7.760632754391597e-05,
      "loss": 1.6226,
      "step": 115072
    },
    {
      "epoch": 0.22713602314299486,
      "grad_norm": 0.1106981560587883,
      "learning_rate": 7.759998731955127e-05,
      "loss": 1.6232,
      "step": 115104
    },
    {
      "epoch": 0.22719916910439128,
      "grad_norm": 0.10561151057481766,
      "learning_rate": 7.759364709518658e-05,
      "loss": 1.6325,
      "step": 115136
    },
    {
      "epoch": 0.22726231506578773,
      "grad_norm": 0.10548792034387589,
      "learning_rate": 7.75873068708219e-05,
      "loss": 1.6204,
      "step": 115168
    },
    {
      "epoch": 0.22732546102718418,
      "grad_norm": 0.1109824851155281,
      "learning_rate": 7.758096664645721e-05,
      "loss": 1.6164,
      "step": 115200
    },
    {
      "epoch": 0.2273886069885806,
      "grad_norm": 0.1021735891699791,
      "learning_rate": 7.757462642209252e-05,
      "loss": 1.6226,
      "step": 115232
    },
    {
      "epoch": 0.22745175294997705,
      "grad_norm": 0.09949398040771484,
      "learning_rate": 7.756828619772783e-05,
      "loss": 1.6191,
      "step": 115264
    },
    {
      "epoch": 0.2275148989113735,
      "grad_norm": 0.10839954018592834,
      "learning_rate": 7.756194597336313e-05,
      "loss": 1.618,
      "step": 115296
    },
    {
      "epoch": 0.22757804487276992,
      "grad_norm": 0.10837170481681824,
      "learning_rate": 7.755560574899845e-05,
      "loss": 1.6243,
      "step": 115328
    },
    {
      "epoch": 0.22764119083416637,
      "grad_norm": 0.10774082690477371,
      "learning_rate": 7.754926552463375e-05,
      "loss": 1.6252,
      "step": 115360
    },
    {
      "epoch": 0.22770433679556282,
      "grad_norm": 0.12567077577114105,
      "learning_rate": 7.754292530026906e-05,
      "loss": 1.6376,
      "step": 115392
    },
    {
      "epoch": 0.22776748275695924,
      "grad_norm": 0.10726109892129898,
      "learning_rate": 7.753658507590438e-05,
      "loss": 1.6287,
      "step": 115424
    },
    {
      "epoch": 0.2278306287183557,
      "grad_norm": 0.10645340383052826,
      "learning_rate": 7.753024485153969e-05,
      "loss": 1.6353,
      "step": 115456
    },
    {
      "epoch": 0.22789377467975214,
      "grad_norm": 0.10601025074720383,
      "learning_rate": 7.752390462717499e-05,
      "loss": 1.6215,
      "step": 115488
    },
    {
      "epoch": 0.22795692064114859,
      "grad_norm": 0.10729113221168518,
      "learning_rate": 7.75175644028103e-05,
      "loss": 1.6231,
      "step": 115520
    },
    {
      "epoch": 0.228020066602545,
      "grad_norm": 0.11095377057790756,
      "learning_rate": 7.751122417844562e-05,
      "loss": 1.6328,
      "step": 115552
    },
    {
      "epoch": 0.22808321256394146,
      "grad_norm": 0.10204273462295532,
      "learning_rate": 7.750488395408093e-05,
      "loss": 1.6416,
      "step": 115584
    },
    {
      "epoch": 0.2281463585253379,
      "grad_norm": 0.1039181798696518,
      "learning_rate": 7.749854372971625e-05,
      "loss": 1.6196,
      "step": 115616
    },
    {
      "epoch": 0.22820950448673433,
      "grad_norm": 0.09892261773347855,
      "learning_rate": 7.749220350535155e-05,
      "loss": 1.6329,
      "step": 115648
    },
    {
      "epoch": 0.22827265044813078,
      "grad_norm": 0.10598491132259369,
      "learning_rate": 7.748586328098686e-05,
      "loss": 1.6269,
      "step": 115680
    },
    {
      "epoch": 0.22833579640952723,
      "grad_norm": 0.11424053460359573,
      "learning_rate": 7.747952305662217e-05,
      "loss": 1.6298,
      "step": 115712
    },
    {
      "epoch": 0.22839894237092365,
      "grad_norm": 0.1030338779091835,
      "learning_rate": 7.747318283225748e-05,
      "loss": 1.6294,
      "step": 115744
    },
    {
      "epoch": 0.2284620883323201,
      "grad_norm": 0.10474897176027298,
      "learning_rate": 7.746684260789278e-05,
      "loss": 1.6263,
      "step": 115776
    },
    {
      "epoch": 0.22852523429371654,
      "grad_norm": 0.10928239673376083,
      "learning_rate": 7.74605023835281e-05,
      "loss": 1.6134,
      "step": 115808
    },
    {
      "epoch": 0.22858838025511297,
      "grad_norm": 0.10650389641523361,
      "learning_rate": 7.745416215916341e-05,
      "loss": 1.6332,
      "step": 115840
    },
    {
      "epoch": 0.22865152621650942,
      "grad_norm": 0.10185643285512924,
      "learning_rate": 7.744782193479872e-05,
      "loss": 1.6155,
      "step": 115872
    },
    {
      "epoch": 0.22871467217790586,
      "grad_norm": 0.11756762862205505,
      "learning_rate": 7.744148171043403e-05,
      "loss": 1.6167,
      "step": 115904
    },
    {
      "epoch": 0.22877781813930229,
      "grad_norm": 0.10846003890037537,
      "learning_rate": 7.743514148606934e-05,
      "loss": 1.6309,
      "step": 115936
    },
    {
      "epoch": 0.22884096410069873,
      "grad_norm": 0.10317414999008179,
      "learning_rate": 7.742880126170466e-05,
      "loss": 1.6248,
      "step": 115968
    },
    {
      "epoch": 0.22890411006209518,
      "grad_norm": 0.10486005246639252,
      "learning_rate": 7.742246103733997e-05,
      "loss": 1.6118,
      "step": 116000
    },
    {
      "epoch": 0.2289672560234916,
      "grad_norm": 0.10296592116355896,
      "learning_rate": 7.741612081297527e-05,
      "loss": 1.639,
      "step": 116032
    },
    {
      "epoch": 0.22903040198488805,
      "grad_norm": 0.09826012700796127,
      "learning_rate": 7.740978058861059e-05,
      "loss": 1.6255,
      "step": 116064
    },
    {
      "epoch": 0.2290935479462845,
      "grad_norm": 0.11436039954423904,
      "learning_rate": 7.74034403642459e-05,
      "loss": 1.6207,
      "step": 116096
    },
    {
      "epoch": 0.22915669390768093,
      "grad_norm": 0.10209383815526962,
      "learning_rate": 7.73971001398812e-05,
      "loss": 1.6264,
      "step": 116128
    },
    {
      "epoch": 0.22921983986907737,
      "grad_norm": 0.10097664594650269,
      "learning_rate": 7.73907599155165e-05,
      "loss": 1.6191,
      "step": 116160
    },
    {
      "epoch": 0.22928298583047382,
      "grad_norm": 0.10899578779935837,
      "learning_rate": 7.738441969115182e-05,
      "loss": 1.6348,
      "step": 116192
    },
    {
      "epoch": 0.22934613179187024,
      "grad_norm": 0.10263419896364212,
      "learning_rate": 7.737807946678713e-05,
      "loss": 1.6136,
      "step": 116224
    },
    {
      "epoch": 0.2294092777532667,
      "grad_norm": 0.10021056234836578,
      "learning_rate": 7.737173924242245e-05,
      "loss": 1.6207,
      "step": 116256
    },
    {
      "epoch": 0.22947242371466314,
      "grad_norm": 0.11048859357833862,
      "learning_rate": 7.736539901805776e-05,
      "loss": 1.6295,
      "step": 116288
    },
    {
      "epoch": 0.22953556967605956,
      "grad_norm": 0.10607375204563141,
      "learning_rate": 7.735905879369306e-05,
      "loss": 1.6272,
      "step": 116320
    },
    {
      "epoch": 0.229598715637456,
      "grad_norm": 0.10816348344087601,
      "learning_rate": 7.735271856932838e-05,
      "loss": 1.6261,
      "step": 116352
    },
    {
      "epoch": 0.22966186159885246,
      "grad_norm": 0.1096382662653923,
      "learning_rate": 7.734637834496369e-05,
      "loss": 1.6177,
      "step": 116384
    },
    {
      "epoch": 0.22972500756024888,
      "grad_norm": 0.10628510266542435,
      "learning_rate": 7.7340038120599e-05,
      "loss": 1.6317,
      "step": 116416
    },
    {
      "epoch": 0.22978815352164533,
      "grad_norm": 0.10687847435474396,
      "learning_rate": 7.73336978962343e-05,
      "loss": 1.6353,
      "step": 116448
    },
    {
      "epoch": 0.22985129948304178,
      "grad_norm": 0.10397474467754364,
      "learning_rate": 7.732735767186962e-05,
      "loss": 1.6378,
      "step": 116480
    },
    {
      "epoch": 0.2299144454444382,
      "grad_norm": 0.10762056708335876,
      "learning_rate": 7.732101744750493e-05,
      "loss": 1.6124,
      "step": 116512
    },
    {
      "epoch": 0.22997759140583465,
      "grad_norm": 0.10956263542175293,
      "learning_rate": 7.731467722314024e-05,
      "loss": 1.624,
      "step": 116544
    },
    {
      "epoch": 0.2300407373672311,
      "grad_norm": 0.10909552872180939,
      "learning_rate": 7.730833699877554e-05,
      "loss": 1.6212,
      "step": 116576
    },
    {
      "epoch": 0.23010388332862752,
      "grad_norm": 0.11167658120393753,
      "learning_rate": 7.730199677441085e-05,
      "loss": 1.623,
      "step": 116608
    },
    {
      "epoch": 0.23016702929002397,
      "grad_norm": 0.10431547462940216,
      "learning_rate": 7.729565655004617e-05,
      "loss": 1.6174,
      "step": 116640
    },
    {
      "epoch": 0.23023017525142042,
      "grad_norm": 0.10333403944969177,
      "learning_rate": 7.728931632568148e-05,
      "loss": 1.6385,
      "step": 116672
    },
    {
      "epoch": 0.23029332121281684,
      "grad_norm": 0.10513138025999069,
      "learning_rate": 7.728297610131678e-05,
      "loss": 1.6286,
      "step": 116704
    },
    {
      "epoch": 0.2303564671742133,
      "grad_norm": 0.10103452205657959,
      "learning_rate": 7.72766358769521e-05,
      "loss": 1.6222,
      "step": 116736
    },
    {
      "epoch": 0.23041961313560974,
      "grad_norm": 0.10250106453895569,
      "learning_rate": 7.727029565258741e-05,
      "loss": 1.6283,
      "step": 116768
    },
    {
      "epoch": 0.23048275909700616,
      "grad_norm": 0.10892458260059357,
      "learning_rate": 7.726395542822273e-05,
      "loss": 1.6353,
      "step": 116800
    },
    {
      "epoch": 0.2305459050584026,
      "grad_norm": 0.09996946901082993,
      "learning_rate": 7.725761520385803e-05,
      "loss": 1.6132,
      "step": 116832
    },
    {
      "epoch": 0.23060905101979906,
      "grad_norm": 0.11390671879053116,
      "learning_rate": 7.725127497949334e-05,
      "loss": 1.635,
      "step": 116864
    },
    {
      "epoch": 0.23067219698119548,
      "grad_norm": 0.099349744617939,
      "learning_rate": 7.724493475512866e-05,
      "loss": 1.6217,
      "step": 116896
    },
    {
      "epoch": 0.23073534294259193,
      "grad_norm": 0.1030801460146904,
      "learning_rate": 7.723859453076397e-05,
      "loss": 1.6336,
      "step": 116928
    },
    {
      "epoch": 0.23079848890398838,
      "grad_norm": 0.10645098239183426,
      "learning_rate": 7.723225430639927e-05,
      "loss": 1.6161,
      "step": 116960
    },
    {
      "epoch": 0.2308616348653848,
      "grad_norm": 0.103032186627388,
      "learning_rate": 7.722591408203457e-05,
      "loss": 1.6203,
      "step": 116992
    },
    {
      "epoch": 0.23092478082678125,
      "grad_norm": 0.10170614719390869,
      "learning_rate": 7.721957385766989e-05,
      "loss": 1.6201,
      "step": 117024
    },
    {
      "epoch": 0.2309879267881777,
      "grad_norm": 0.10592218488454819,
      "learning_rate": 7.72132336333052e-05,
      "loss": 1.6107,
      "step": 117056
    },
    {
      "epoch": 0.23105107274957412,
      "grad_norm": 0.10595889389514923,
      "learning_rate": 7.720689340894052e-05,
      "loss": 1.6207,
      "step": 117088
    },
    {
      "epoch": 0.23111421871097057,
      "grad_norm": 0.10585210472345352,
      "learning_rate": 7.720055318457582e-05,
      "loss": 1.6333,
      "step": 117120
    },
    {
      "epoch": 0.23117736467236702,
      "grad_norm": 0.11004262417554855,
      "learning_rate": 7.719421296021113e-05,
      "loss": 1.6279,
      "step": 117152
    },
    {
      "epoch": 0.23124051063376347,
      "grad_norm": 0.10163483768701553,
      "learning_rate": 7.718787273584645e-05,
      "loss": 1.6141,
      "step": 117184
    },
    {
      "epoch": 0.2313036565951599,
      "grad_norm": 0.11740569025278091,
      "learning_rate": 7.718153251148176e-05,
      "loss": 1.6342,
      "step": 117216
    },
    {
      "epoch": 0.23136680255655634,
      "grad_norm": 0.10180219262838364,
      "learning_rate": 7.717519228711706e-05,
      "loss": 1.6163,
      "step": 117248
    },
    {
      "epoch": 0.2314299485179528,
      "grad_norm": 0.10720863193273544,
      "learning_rate": 7.716885206275238e-05,
      "loss": 1.6184,
      "step": 117280
    },
    {
      "epoch": 0.2314930944793492,
      "grad_norm": 0.11184300482273102,
      "learning_rate": 7.716251183838769e-05,
      "loss": 1.6138,
      "step": 117312
    },
    {
      "epoch": 0.23155624044074566,
      "grad_norm": 0.1033792644739151,
      "learning_rate": 7.7156171614023e-05,
      "loss": 1.6305,
      "step": 117344
    },
    {
      "epoch": 0.2316193864021421,
      "grad_norm": 0.10217909514904022,
      "learning_rate": 7.71498313896583e-05,
      "loss": 1.6233,
      "step": 117376
    },
    {
      "epoch": 0.23168253236353853,
      "grad_norm": 0.10086463391780853,
      "learning_rate": 7.714349116529361e-05,
      "loss": 1.6145,
      "step": 117408
    },
    {
      "epoch": 0.23174567832493498,
      "grad_norm": 0.10798216611146927,
      "learning_rate": 7.713715094092892e-05,
      "loss": 1.6145,
      "step": 117440
    },
    {
      "epoch": 0.23180882428633143,
      "grad_norm": 0.10482317954301834,
      "learning_rate": 7.713081071656424e-05,
      "loss": 1.6309,
      "step": 117472
    },
    {
      "epoch": 0.23187197024772785,
      "grad_norm": 0.09853389859199524,
      "learning_rate": 7.712447049219954e-05,
      "loss": 1.6266,
      "step": 117504
    },
    {
      "epoch": 0.2319351162091243,
      "grad_norm": 0.11287181079387665,
      "learning_rate": 7.711813026783485e-05,
      "loss": 1.6198,
      "step": 117536
    },
    {
      "epoch": 0.23199826217052075,
      "grad_norm": 0.10568402707576752,
      "learning_rate": 7.711179004347017e-05,
      "loss": 1.6265,
      "step": 117568
    },
    {
      "epoch": 0.23206140813191717,
      "grad_norm": 0.1097228080034256,
      "learning_rate": 7.710544981910548e-05,
      "loss": 1.6193,
      "step": 117600
    },
    {
      "epoch": 0.23212455409331362,
      "grad_norm": 0.10588759183883667,
      "learning_rate": 7.70991095947408e-05,
      "loss": 1.6299,
      "step": 117632
    },
    {
      "epoch": 0.23218770005471007,
      "grad_norm": 0.10230513662099838,
      "learning_rate": 7.70927693703761e-05,
      "loss": 1.6201,
      "step": 117664
    },
    {
      "epoch": 0.2322508460161065,
      "grad_norm": 0.10774840414524078,
      "learning_rate": 7.708642914601141e-05,
      "loss": 1.626,
      "step": 117696
    },
    {
      "epoch": 0.23231399197750294,
      "grad_norm": 0.10887067019939423,
      "learning_rate": 7.708008892164673e-05,
      "loss": 1.612,
      "step": 117728
    },
    {
      "epoch": 0.2323771379388994,
      "grad_norm": 0.10515390336513519,
      "learning_rate": 7.707374869728203e-05,
      "loss": 1.6159,
      "step": 117760
    },
    {
      "epoch": 0.2324402839002958,
      "grad_norm": 0.10412167757749557,
      "learning_rate": 7.706740847291734e-05,
      "loss": 1.616,
      "step": 117792
    },
    {
      "epoch": 0.23250342986169226,
      "grad_norm": 0.11328978836536407,
      "learning_rate": 7.706106824855264e-05,
      "loss": 1.6289,
      "step": 117824
    },
    {
      "epoch": 0.2325665758230887,
      "grad_norm": 0.10668544471263885,
      "learning_rate": 7.705472802418796e-05,
      "loss": 1.6341,
      "step": 117856
    },
    {
      "epoch": 0.23262972178448513,
      "grad_norm": 0.1112147718667984,
      "learning_rate": 7.704838779982327e-05,
      "loss": 1.6227,
      "step": 117888
    },
    {
      "epoch": 0.23269286774588158,
      "grad_norm": 0.10626817494630814,
      "learning_rate": 7.704204757545857e-05,
      "loss": 1.6235,
      "step": 117920
    },
    {
      "epoch": 0.23275601370727803,
      "grad_norm": 0.09796631336212158,
      "learning_rate": 7.703570735109389e-05,
      "loss": 1.6276,
      "step": 117952
    },
    {
      "epoch": 0.23281915966867445,
      "grad_norm": 0.10077675431966782,
      "learning_rate": 7.70293671267292e-05,
      "loss": 1.6252,
      "step": 117984
    },
    {
      "epoch": 0.2328823056300709,
      "grad_norm": 0.11095286905765533,
      "learning_rate": 7.702302690236452e-05,
      "loss": 1.6266,
      "step": 118016
    },
    {
      "epoch": 0.23294545159146735,
      "grad_norm": 0.09941935539245605,
      "learning_rate": 7.701668667799982e-05,
      "loss": 1.6307,
      "step": 118048
    },
    {
      "epoch": 0.23300859755286377,
      "grad_norm": 0.10696975886821747,
      "learning_rate": 7.701034645363513e-05,
      "loss": 1.6205,
      "step": 118080
    },
    {
      "epoch": 0.23307174351426022,
      "grad_norm": 0.10444318503141403,
      "learning_rate": 7.700400622927045e-05,
      "loss": 1.6321,
      "step": 118112
    },
    {
      "epoch": 0.23313488947565666,
      "grad_norm": 0.12032336741685867,
      "learning_rate": 7.699766600490576e-05,
      "loss": 1.628,
      "step": 118144
    },
    {
      "epoch": 0.2331980354370531,
      "grad_norm": 0.10192475467920303,
      "learning_rate": 7.699132578054106e-05,
      "loss": 1.6122,
      "step": 118176
    },
    {
      "epoch": 0.23326118139844954,
      "grad_norm": 0.10944230854511261,
      "learning_rate": 7.698498555617638e-05,
      "loss": 1.6268,
      "step": 118208
    },
    {
      "epoch": 0.23332432735984598,
      "grad_norm": 0.10616450756788254,
      "learning_rate": 7.697864533181168e-05,
      "loss": 1.6141,
      "step": 118240
    },
    {
      "epoch": 0.2333874733212424,
      "grad_norm": 0.10260586440563202,
      "learning_rate": 7.697230510744699e-05,
      "loss": 1.6184,
      "step": 118272
    },
    {
      "epoch": 0.23345061928263885,
      "grad_norm": 0.10613703727722168,
      "learning_rate": 7.69659648830823e-05,
      "loss": 1.6197,
      "step": 118304
    },
    {
      "epoch": 0.2335137652440353,
      "grad_norm": 0.10300738364458084,
      "learning_rate": 7.695962465871761e-05,
      "loss": 1.6113,
      "step": 118336
    },
    {
      "epoch": 0.23357691120543173,
      "grad_norm": 0.10916262120008469,
      "learning_rate": 7.695328443435292e-05,
      "loss": 1.6207,
      "step": 118368
    },
    {
      "epoch": 0.23364005716682817,
      "grad_norm": 0.10659836232662201,
      "learning_rate": 7.694694420998824e-05,
      "loss": 1.6212,
      "step": 118400
    },
    {
      "epoch": 0.23370320312822462,
      "grad_norm": 0.1065870001912117,
      "learning_rate": 7.694060398562355e-05,
      "loss": 1.6312,
      "step": 118432
    },
    {
      "epoch": 0.23376634908962104,
      "grad_norm": 0.11020667850971222,
      "learning_rate": 7.693426376125885e-05,
      "loss": 1.6279,
      "step": 118464
    },
    {
      "epoch": 0.2338294950510175,
      "grad_norm": 0.10344437509775162,
      "learning_rate": 7.692792353689417e-05,
      "loss": 1.6178,
      "step": 118496
    },
    {
      "epoch": 0.23389264101241394,
      "grad_norm": 0.11212826520204544,
      "learning_rate": 7.692158331252948e-05,
      "loss": 1.6237,
      "step": 118528
    },
    {
      "epoch": 0.23395578697381036,
      "grad_norm": 0.10625287145376205,
      "learning_rate": 7.69152430881648e-05,
      "loss": 1.6182,
      "step": 118560
    },
    {
      "epoch": 0.2340189329352068,
      "grad_norm": 0.10378042608499527,
      "learning_rate": 7.69089028638001e-05,
      "loss": 1.616,
      "step": 118592
    },
    {
      "epoch": 0.23408207889660326,
      "grad_norm": 0.10604571551084518,
      "learning_rate": 7.690256263943541e-05,
      "loss": 1.635,
      "step": 118624
    },
    {
      "epoch": 0.23414522485799968,
      "grad_norm": 0.10951421409845352,
      "learning_rate": 7.689622241507071e-05,
      "loss": 1.6033,
      "step": 118656
    },
    {
      "epoch": 0.23420837081939613,
      "grad_norm": 0.10516619682312012,
      "learning_rate": 7.688988219070603e-05,
      "loss": 1.6222,
      "step": 118688
    },
    {
      "epoch": 0.23427151678079258,
      "grad_norm": 0.10432136058807373,
      "learning_rate": 7.688354196634133e-05,
      "loss": 1.6281,
      "step": 118720
    },
    {
      "epoch": 0.234334662742189,
      "grad_norm": 0.10414012521505356,
      "learning_rate": 7.687720174197664e-05,
      "loss": 1.6334,
      "step": 118752
    },
    {
      "epoch": 0.23439780870358545,
      "grad_norm": 0.10413164645433426,
      "learning_rate": 7.687086151761196e-05,
      "loss": 1.6199,
      "step": 118784
    },
    {
      "epoch": 0.2344609546649819,
      "grad_norm": 0.11605276167392731,
      "learning_rate": 7.686452129324727e-05,
      "loss": 1.6143,
      "step": 118816
    },
    {
      "epoch": 0.23452410062637835,
      "grad_norm": 0.10303189605474472,
      "learning_rate": 7.685818106888259e-05,
      "loss": 1.6195,
      "step": 118848
    },
    {
      "epoch": 0.23458724658777477,
      "grad_norm": 0.10710170120000839,
      "learning_rate": 7.685184084451789e-05,
      "loss": 1.6238,
      "step": 118880
    },
    {
      "epoch": 0.23465039254917122,
      "grad_norm": 0.10300502181053162,
      "learning_rate": 7.68455006201532e-05,
      "loss": 1.6176,
      "step": 118912
    },
    {
      "epoch": 0.23471353851056767,
      "grad_norm": 0.11010536551475525,
      "learning_rate": 7.683916039578852e-05,
      "loss": 1.6148,
      "step": 118944
    },
    {
      "epoch": 0.2347766844719641,
      "grad_norm": 0.10205256938934326,
      "learning_rate": 7.683282017142383e-05,
      "loss": 1.6227,
      "step": 118976
    },
    {
      "epoch": 0.23483983043336054,
      "grad_norm": 0.10697031766176224,
      "learning_rate": 7.682647994705913e-05,
      "loss": 1.6195,
      "step": 119008
    },
    {
      "epoch": 0.234902976394757,
      "grad_norm": 0.09970673173666,
      "learning_rate": 7.682013972269443e-05,
      "loss": 1.6236,
      "step": 119040
    },
    {
      "epoch": 0.2349661223561534,
      "grad_norm": 0.10200951248407364,
      "learning_rate": 7.681379949832975e-05,
      "loss": 1.6167,
      "step": 119072
    },
    {
      "epoch": 0.23502926831754986,
      "grad_norm": 0.10827691853046417,
      "learning_rate": 7.680745927396506e-05,
      "loss": 1.6206,
      "step": 119104
    },
    {
      "epoch": 0.2350924142789463,
      "grad_norm": 0.1034129336476326,
      "learning_rate": 7.680111904960036e-05,
      "loss": 1.6108,
      "step": 119136
    },
    {
      "epoch": 0.23515556024034273,
      "grad_norm": 0.10959687829017639,
      "learning_rate": 7.679477882523568e-05,
      "loss": 1.6319,
      "step": 119168
    },
    {
      "epoch": 0.23521870620173918,
      "grad_norm": 0.11860638856887817,
      "learning_rate": 7.678843860087099e-05,
      "loss": 1.6121,
      "step": 119200
    },
    {
      "epoch": 0.23528185216313563,
      "grad_norm": 0.1124686673283577,
      "learning_rate": 7.678209837650631e-05,
      "loss": 1.6297,
      "step": 119232
    },
    {
      "epoch": 0.23534499812453205,
      "grad_norm": 0.10974572598934174,
      "learning_rate": 7.677575815214161e-05,
      "loss": 1.6156,
      "step": 119264
    },
    {
      "epoch": 0.2354081440859285,
      "grad_norm": 0.10951638966798782,
      "learning_rate": 7.676941792777692e-05,
      "loss": 1.6235,
      "step": 119296
    },
    {
      "epoch": 0.23547129004732495,
      "grad_norm": 0.11050344258546829,
      "learning_rate": 7.676307770341224e-05,
      "loss": 1.6151,
      "step": 119328
    },
    {
      "epoch": 0.23553443600872137,
      "grad_norm": 0.1017633005976677,
      "learning_rate": 7.675673747904755e-05,
      "loss": 1.6199,
      "step": 119360
    },
    {
      "epoch": 0.23559758197011782,
      "grad_norm": 0.109011210501194,
      "learning_rate": 7.675039725468285e-05,
      "loss": 1.6324,
      "step": 119392
    },
    {
      "epoch": 0.23566072793151427,
      "grad_norm": 0.11013071238994598,
      "learning_rate": 7.674405703031817e-05,
      "loss": 1.6052,
      "step": 119424
    },
    {
      "epoch": 0.2357238738929107,
      "grad_norm": 0.10388562083244324,
      "learning_rate": 7.673771680595347e-05,
      "loss": 1.6241,
      "step": 119456
    },
    {
      "epoch": 0.23578701985430714,
      "grad_norm": 0.1053258553147316,
      "learning_rate": 7.673137658158878e-05,
      "loss": 1.6252,
      "step": 119488
    },
    {
      "epoch": 0.2358501658157036,
      "grad_norm": 0.10670289397239685,
      "learning_rate": 7.67250363572241e-05,
      "loss": 1.6118,
      "step": 119520
    },
    {
      "epoch": 0.2359133117771,
      "grad_norm": 0.1083621084690094,
      "learning_rate": 7.67186961328594e-05,
      "loss": 1.6155,
      "step": 119552
    },
    {
      "epoch": 0.23597645773849646,
      "grad_norm": 0.10241899639368057,
      "learning_rate": 7.671235590849471e-05,
      "loss": 1.6313,
      "step": 119584
    },
    {
      "epoch": 0.2360396036998929,
      "grad_norm": 0.10821599513292313,
      "learning_rate": 7.670601568413003e-05,
      "loss": 1.6246,
      "step": 119616
    },
    {
      "epoch": 0.23610274966128933,
      "grad_norm": 0.10203195363283157,
      "learning_rate": 7.669967545976534e-05,
      "loss": 1.6148,
      "step": 119648
    },
    {
      "epoch": 0.23616589562268578,
      "grad_norm": 0.09916464984416962,
      "learning_rate": 7.669333523540064e-05,
      "loss": 1.6168,
      "step": 119680
    },
    {
      "epoch": 0.23622904158408223,
      "grad_norm": 0.10602273792028427,
      "learning_rate": 7.668699501103596e-05,
      "loss": 1.6138,
      "step": 119712
    },
    {
      "epoch": 0.23629218754547865,
      "grad_norm": 0.10990050435066223,
      "learning_rate": 7.668065478667127e-05,
      "loss": 1.607,
      "step": 119744
    },
    {
      "epoch": 0.2363553335068751,
      "grad_norm": 0.10659512877464294,
      "learning_rate": 7.667431456230659e-05,
      "loss": 1.6323,
      "step": 119776
    },
    {
      "epoch": 0.23641847946827155,
      "grad_norm": 0.10102102905511856,
      "learning_rate": 7.666797433794189e-05,
      "loss": 1.6218,
      "step": 119808
    },
    {
      "epoch": 0.23648162542966797,
      "grad_norm": 0.10293369740247726,
      "learning_rate": 7.66616341135772e-05,
      "loss": 1.6224,
      "step": 119840
    },
    {
      "epoch": 0.23654477139106442,
      "grad_norm": 0.11067891120910645,
      "learning_rate": 7.66552938892125e-05,
      "loss": 1.625,
      "step": 119872
    },
    {
      "epoch": 0.23660791735246087,
      "grad_norm": 0.1084989532828331,
      "learning_rate": 7.664895366484782e-05,
      "loss": 1.6255,
      "step": 119904
    },
    {
      "epoch": 0.2366710633138573,
      "grad_norm": 0.10479579865932465,
      "learning_rate": 7.664261344048312e-05,
      "loss": 1.6412,
      "step": 119936
    },
    {
      "epoch": 0.23673420927525374,
      "grad_norm": 0.10684318095445633,
      "learning_rate": 7.663627321611843e-05,
      "loss": 1.628,
      "step": 119968
    },
    {
      "epoch": 0.2367973552366502,
      "grad_norm": 0.10191134363412857,
      "learning_rate": 7.662993299175375e-05,
      "loss": 1.6146,
      "step": 120000
    },
    {
      "epoch": 0.2368605011980466,
      "grad_norm": 0.10358243435621262,
      "learning_rate": 7.662359276738906e-05,
      "loss": 1.6172,
      "step": 120032
    },
    {
      "epoch": 0.23692364715944306,
      "grad_norm": 0.1038525253534317,
      "learning_rate": 7.661725254302436e-05,
      "loss": 1.6192,
      "step": 120064
    },
    {
      "epoch": 0.2369867931208395,
      "grad_norm": 0.1099800243973732,
      "learning_rate": 7.661091231865968e-05,
      "loss": 1.6291,
      "step": 120096
    },
    {
      "epoch": 0.23704993908223593,
      "grad_norm": 0.10635606944561005,
      "learning_rate": 7.660457209429499e-05,
      "loss": 1.6295,
      "step": 120128
    },
    {
      "epoch": 0.23711308504363238,
      "grad_norm": 0.10405123233795166,
      "learning_rate": 7.659823186993031e-05,
      "loss": 1.632,
      "step": 120160
    },
    {
      "epoch": 0.23717623100502883,
      "grad_norm": 0.10176390409469604,
      "learning_rate": 7.659189164556562e-05,
      "loss": 1.6195,
      "step": 120192
    },
    {
      "epoch": 0.23723937696642525,
      "grad_norm": 0.11222195625305176,
      "learning_rate": 7.658555142120092e-05,
      "loss": 1.6242,
      "step": 120224
    },
    {
      "epoch": 0.2373025229278217,
      "grad_norm": 0.11476761847734451,
      "learning_rate": 7.657921119683624e-05,
      "loss": 1.6232,
      "step": 120256
    },
    {
      "epoch": 0.23736566888921815,
      "grad_norm": 0.10762138664722443,
      "learning_rate": 7.657287097247154e-05,
      "loss": 1.6143,
      "step": 120288
    },
    {
      "epoch": 0.23742881485061457,
      "grad_norm": 0.10720644146203995,
      "learning_rate": 7.656653074810685e-05,
      "loss": 1.6152,
      "step": 120320
    },
    {
      "epoch": 0.23749196081201102,
      "grad_norm": 0.1090712621808052,
      "learning_rate": 7.656019052374215e-05,
      "loss": 1.639,
      "step": 120352
    },
    {
      "epoch": 0.23755510677340747,
      "grad_norm": 0.10782972723245621,
      "learning_rate": 7.655385029937747e-05,
      "loss": 1.6135,
      "step": 120384
    },
    {
      "epoch": 0.2376182527348039,
      "grad_norm": 0.10414179414510727,
      "learning_rate": 7.654751007501278e-05,
      "loss": 1.6198,
      "step": 120416
    },
    {
      "epoch": 0.23768139869620034,
      "grad_norm": 0.10828063637018204,
      "learning_rate": 7.65411698506481e-05,
      "loss": 1.6209,
      "step": 120448
    },
    {
      "epoch": 0.23774454465759678,
      "grad_norm": 0.10708325356245041,
      "learning_rate": 7.65348296262834e-05,
      "loss": 1.6269,
      "step": 120480
    },
    {
      "epoch": 0.2378076906189932,
      "grad_norm": 0.10508102923631668,
      "learning_rate": 7.652848940191871e-05,
      "loss": 1.6306,
      "step": 120512
    },
    {
      "epoch": 0.23787083658038966,
      "grad_norm": 0.11044111102819443,
      "learning_rate": 7.652214917755403e-05,
      "loss": 1.6009,
      "step": 120544
    },
    {
      "epoch": 0.2379339825417861,
      "grad_norm": 0.12157759815454483,
      "learning_rate": 7.651580895318934e-05,
      "loss": 1.6292,
      "step": 120576
    },
    {
      "epoch": 0.23799712850318255,
      "grad_norm": 0.1084010899066925,
      "learning_rate": 7.650946872882464e-05,
      "loss": 1.6169,
      "step": 120608
    },
    {
      "epoch": 0.23806027446457897,
      "grad_norm": 0.10739398747682571,
      "learning_rate": 7.650312850445996e-05,
      "loss": 1.6183,
      "step": 120640
    },
    {
      "epoch": 0.23812342042597542,
      "grad_norm": 0.11482581496238708,
      "learning_rate": 7.649678828009527e-05,
      "loss": 1.6293,
      "step": 120672
    },
    {
      "epoch": 0.23818656638737187,
      "grad_norm": 0.10567828267812729,
      "learning_rate": 7.649044805573057e-05,
      "loss": 1.6363,
      "step": 120704
    },
    {
      "epoch": 0.2382497123487683,
      "grad_norm": 0.10498349368572235,
      "learning_rate": 7.648410783136587e-05,
      "loss": 1.6093,
      "step": 120736
    },
    {
      "epoch": 0.23831285831016474,
      "grad_norm": 0.10828934609889984,
      "learning_rate": 7.647776760700119e-05,
      "loss": 1.6326,
      "step": 120768
    },
    {
      "epoch": 0.2383760042715612,
      "grad_norm": 0.11244882643222809,
      "learning_rate": 7.64714273826365e-05,
      "loss": 1.6256,
      "step": 120800
    },
    {
      "epoch": 0.23843915023295761,
      "grad_norm": 0.12038005143404007,
      "learning_rate": 7.646508715827182e-05,
      "loss": 1.6235,
      "step": 120832
    },
    {
      "epoch": 0.23850229619435406,
      "grad_norm": 0.09573744982481003,
      "learning_rate": 7.645874693390713e-05,
      "loss": 1.6046,
      "step": 120864
    },
    {
      "epoch": 0.2385654421557505,
      "grad_norm": 0.10723771899938583,
      "learning_rate": 7.645240670954243e-05,
      "loss": 1.6188,
      "step": 120896
    },
    {
      "epoch": 0.23862858811714693,
      "grad_norm": 0.11568645387887955,
      "learning_rate": 7.644606648517775e-05,
      "loss": 1.6096,
      "step": 120928
    },
    {
      "epoch": 0.23869173407854338,
      "grad_norm": 0.1048082485795021,
      "learning_rate": 7.643972626081306e-05,
      "loss": 1.6276,
      "step": 120960
    },
    {
      "epoch": 0.23875488003993983,
      "grad_norm": 0.10420051217079163,
      "learning_rate": 7.643338603644838e-05,
      "loss": 1.627,
      "step": 120992
    },
    {
      "epoch": 0.23881802600133625,
      "grad_norm": 0.10620584338903427,
      "learning_rate": 7.642704581208368e-05,
      "loss": 1.6226,
      "step": 121024
    },
    {
      "epoch": 0.2388811719627327,
      "grad_norm": 0.10347410291433334,
      "learning_rate": 7.642070558771899e-05,
      "loss": 1.6274,
      "step": 121056
    },
    {
      "epoch": 0.23894431792412915,
      "grad_norm": 0.09775253385305405,
      "learning_rate": 7.641436536335431e-05,
      "loss": 1.6168,
      "step": 121088
    },
    {
      "epoch": 0.23900746388552557,
      "grad_norm": 0.11735594272613525,
      "learning_rate": 7.640802513898961e-05,
      "loss": 1.637,
      "step": 121120
    },
    {
      "epoch": 0.23907060984692202,
      "grad_norm": 0.10499858111143112,
      "learning_rate": 7.640168491462491e-05,
      "loss": 1.6231,
      "step": 121152
    },
    {
      "epoch": 0.23913375580831847,
      "grad_norm": 0.10351528227329254,
      "learning_rate": 7.639534469026022e-05,
      "loss": 1.6211,
      "step": 121184
    },
    {
      "epoch": 0.2391969017697149,
      "grad_norm": 0.10469400137662888,
      "learning_rate": 7.638900446589554e-05,
      "loss": 1.6216,
      "step": 121216
    },
    {
      "epoch": 0.23926004773111134,
      "grad_norm": 0.10248269885778427,
      "learning_rate": 7.638266424153085e-05,
      "loss": 1.6173,
      "step": 121248
    },
    {
      "epoch": 0.2393231936925078,
      "grad_norm": 0.10857024043798447,
      "learning_rate": 7.637632401716615e-05,
      "loss": 1.6248,
      "step": 121280
    },
    {
      "epoch": 0.2393863396539042,
      "grad_norm": 0.1071888655424118,
      "learning_rate": 7.636998379280147e-05,
      "loss": 1.6234,
      "step": 121312
    },
    {
      "epoch": 0.23944948561530066,
      "grad_norm": 0.11180950701236725,
      "learning_rate": 7.636364356843678e-05,
      "loss": 1.6264,
      "step": 121344
    },
    {
      "epoch": 0.2395126315766971,
      "grad_norm": 0.10684220492839813,
      "learning_rate": 7.63573033440721e-05,
      "loss": 1.6198,
      "step": 121376
    },
    {
      "epoch": 0.23957577753809353,
      "grad_norm": 0.11148609220981598,
      "learning_rate": 7.63509631197074e-05,
      "loss": 1.6185,
      "step": 121408
    },
    {
      "epoch": 0.23963892349948998,
      "grad_norm": 0.09844603389501572,
      "learning_rate": 7.634462289534271e-05,
      "loss": 1.6092,
      "step": 121440
    },
    {
      "epoch": 0.23970206946088643,
      "grad_norm": 0.10835153609514236,
      "learning_rate": 7.633828267097803e-05,
      "loss": 1.6345,
      "step": 121472
    },
    {
      "epoch": 0.23976521542228285,
      "grad_norm": 0.10451914370059967,
      "learning_rate": 7.633194244661334e-05,
      "loss": 1.6291,
      "step": 121504
    },
    {
      "epoch": 0.2398283613836793,
      "grad_norm": 0.11012529581785202,
      "learning_rate": 7.632560222224864e-05,
      "loss": 1.6177,
      "step": 121536
    },
    {
      "epoch": 0.23989150734507575,
      "grad_norm": 0.10615980625152588,
      "learning_rate": 7.631926199788394e-05,
      "loss": 1.6299,
      "step": 121568
    },
    {
      "epoch": 0.23995465330647217,
      "grad_norm": 0.10427898913621902,
      "learning_rate": 7.631292177351926e-05,
      "loss": 1.6097,
      "step": 121600
    },
    {
      "epoch": 0.24001779926786862,
      "grad_norm": 0.11751525104045868,
      "learning_rate": 7.630658154915457e-05,
      "loss": 1.6162,
      "step": 121632
    },
    {
      "epoch": 0.24008094522926507,
      "grad_norm": 0.10239632427692413,
      "learning_rate": 7.630024132478989e-05,
      "loss": 1.618,
      "step": 121664
    },
    {
      "epoch": 0.2401440911906615,
      "grad_norm": 0.10828424245119095,
      "learning_rate": 7.629390110042519e-05,
      "loss": 1.6073,
      "step": 121696
    },
    {
      "epoch": 0.24020723715205794,
      "grad_norm": 0.10856421291828156,
      "learning_rate": 7.62875608760605e-05,
      "loss": 1.6267,
      "step": 121728
    },
    {
      "epoch": 0.2402703831134544,
      "grad_norm": 0.11210472881793976,
      "learning_rate": 7.628122065169582e-05,
      "loss": 1.6184,
      "step": 121760
    },
    {
      "epoch": 0.2403335290748508,
      "grad_norm": 0.11003146320581436,
      "learning_rate": 7.627488042733113e-05,
      "loss": 1.6172,
      "step": 121792
    },
    {
      "epoch": 0.24039667503624726,
      "grad_norm": 0.10011272132396698,
      "learning_rate": 7.626854020296643e-05,
      "loss": 1.6215,
      "step": 121824
    },
    {
      "epoch": 0.2404598209976437,
      "grad_norm": 0.1020115315914154,
      "learning_rate": 7.626219997860175e-05,
      "loss": 1.6154,
      "step": 121856
    },
    {
      "epoch": 0.24052296695904013,
      "grad_norm": 0.11512121558189392,
      "learning_rate": 7.625585975423706e-05,
      "loss": 1.6212,
      "step": 121888
    },
    {
      "epoch": 0.24058611292043658,
      "grad_norm": 0.10439430177211761,
      "learning_rate": 7.624951952987236e-05,
      "loss": 1.6301,
      "step": 121920
    },
    {
      "epoch": 0.24064925888183303,
      "grad_norm": 0.10639136284589767,
      "learning_rate": 7.624317930550768e-05,
      "loss": 1.6083,
      "step": 121952
    },
    {
      "epoch": 0.24071240484322945,
      "grad_norm": 0.10879534482955933,
      "learning_rate": 7.623683908114298e-05,
      "loss": 1.6088,
      "step": 121984
    },
    {
      "epoch": 0.2407755508046259,
      "grad_norm": 0.10547618567943573,
      "learning_rate": 7.62304988567783e-05,
      "loss": 1.6125,
      "step": 122016
    },
    {
      "epoch": 0.24083869676602235,
      "grad_norm": 0.11240384727716446,
      "learning_rate": 7.622415863241361e-05,
      "loss": 1.6153,
      "step": 122048
    },
    {
      "epoch": 0.24090184272741877,
      "grad_norm": 0.10646174848079681,
      "learning_rate": 7.621781840804891e-05,
      "loss": 1.6095,
      "step": 122080
    },
    {
      "epoch": 0.24096498868881522,
      "grad_norm": 0.10142052918672562,
      "learning_rate": 7.621147818368422e-05,
      "loss": 1.6192,
      "step": 122112
    },
    {
      "epoch": 0.24102813465021167,
      "grad_norm": 0.11388655006885529,
      "learning_rate": 7.620513795931954e-05,
      "loss": 1.6336,
      "step": 122144
    },
    {
      "epoch": 0.2410912806116081,
      "grad_norm": 0.11353683471679688,
      "learning_rate": 7.619879773495485e-05,
      "loss": 1.6259,
      "step": 122176
    },
    {
      "epoch": 0.24115442657300454,
      "grad_norm": 0.10896036028862,
      "learning_rate": 7.619245751059017e-05,
      "loss": 1.6274,
      "step": 122208
    },
    {
      "epoch": 0.241217572534401,
      "grad_norm": 0.1057293489575386,
      "learning_rate": 7.618611728622547e-05,
      "loss": 1.6136,
      "step": 122240
    },
    {
      "epoch": 0.24128071849579744,
      "grad_norm": 0.10727423429489136,
      "learning_rate": 7.617977706186078e-05,
      "loss": 1.6189,
      "step": 122272
    },
    {
      "epoch": 0.24134386445719386,
      "grad_norm": 0.10312186926603317,
      "learning_rate": 7.61734368374961e-05,
      "loss": 1.6172,
      "step": 122304
    },
    {
      "epoch": 0.2414070104185903,
      "grad_norm": 0.10512793064117432,
      "learning_rate": 7.61670966131314e-05,
      "loss": 1.6157,
      "step": 122336
    },
    {
      "epoch": 0.24147015637998676,
      "grad_norm": 0.10570508241653442,
      "learning_rate": 7.616075638876671e-05,
      "loss": 1.6266,
      "step": 122368
    },
    {
      "epoch": 0.24153330234138318,
      "grad_norm": 0.1021624356508255,
      "learning_rate": 7.615441616440201e-05,
      "loss": 1.6125,
      "step": 122400
    },
    {
      "epoch": 0.24159644830277963,
      "grad_norm": 0.10386671870946884,
      "learning_rate": 7.614807594003733e-05,
      "loss": 1.6195,
      "step": 122432
    },
    {
      "epoch": 0.24165959426417608,
      "grad_norm": 0.10835690051317215,
      "learning_rate": 7.614173571567264e-05,
      "loss": 1.6215,
      "step": 122464
    },
    {
      "epoch": 0.2417227402255725,
      "grad_norm": 0.10276900976896286,
      "learning_rate": 7.613539549130794e-05,
      "loss": 1.6162,
      "step": 122496
    },
    {
      "epoch": 0.24178588618696895,
      "grad_norm": 0.10647771507501602,
      "learning_rate": 7.612905526694326e-05,
      "loss": 1.6083,
      "step": 122528
    },
    {
      "epoch": 0.2418490321483654,
      "grad_norm": 0.10218511521816254,
      "learning_rate": 7.612271504257857e-05,
      "loss": 1.6139,
      "step": 122560
    },
    {
      "epoch": 0.24191217810976182,
      "grad_norm": 0.11111932247877121,
      "learning_rate": 7.611637481821389e-05,
      "loss": 1.6239,
      "step": 122592
    },
    {
      "epoch": 0.24197532407115827,
      "grad_norm": 0.11231356114149094,
      "learning_rate": 7.611003459384919e-05,
      "loss": 1.6143,
      "step": 122624
    },
    {
      "epoch": 0.24203847003255471,
      "grad_norm": 0.10202392935752869,
      "learning_rate": 7.61036943694845e-05,
      "loss": 1.6149,
      "step": 122656
    },
    {
      "epoch": 0.24210161599395114,
      "grad_norm": 0.10742413997650146,
      "learning_rate": 7.609735414511982e-05,
      "loss": 1.6176,
      "step": 122688
    },
    {
      "epoch": 0.24216476195534758,
      "grad_norm": 0.10752253234386444,
      "learning_rate": 7.609101392075513e-05,
      "loss": 1.607,
      "step": 122720
    },
    {
      "epoch": 0.24222790791674403,
      "grad_norm": 0.10861130803823471,
      "learning_rate": 7.608467369639043e-05,
      "loss": 1.6263,
      "step": 122752
    },
    {
      "epoch": 0.24229105387814046,
      "grad_norm": 0.10310644656419754,
      "learning_rate": 7.607833347202575e-05,
      "loss": 1.6253,
      "step": 122784
    },
    {
      "epoch": 0.2423541998395369,
      "grad_norm": 0.10626522451639175,
      "learning_rate": 7.607199324766105e-05,
      "loss": 1.6209,
      "step": 122816
    },
    {
      "epoch": 0.24241734580093335,
      "grad_norm": 0.10614936053752899,
      "learning_rate": 7.606565302329636e-05,
      "loss": 1.6251,
      "step": 122848
    },
    {
      "epoch": 0.24248049176232978,
      "grad_norm": 0.10486698895692825,
      "learning_rate": 7.605931279893168e-05,
      "loss": 1.614,
      "step": 122880
    },
    {
      "epoch": 0.24254363772372622,
      "grad_norm": 0.10899936407804489,
      "learning_rate": 7.605297257456698e-05,
      "loss": 1.6314,
      "step": 122912
    },
    {
      "epoch": 0.24260678368512267,
      "grad_norm": 0.11536278575658798,
      "learning_rate": 7.60466323502023e-05,
      "loss": 1.6183,
      "step": 122944
    },
    {
      "epoch": 0.2426699296465191,
      "grad_norm": 0.10541034489870071,
      "learning_rate": 7.604029212583761e-05,
      "loss": 1.6166,
      "step": 122976
    },
    {
      "epoch": 0.24273307560791554,
      "grad_norm": 0.10871303826570511,
      "learning_rate": 7.603395190147292e-05,
      "loss": 1.623,
      "step": 123008
    },
    {
      "epoch": 0.242796221569312,
      "grad_norm": 0.10295688360929489,
      "learning_rate": 7.602761167710822e-05,
      "loss": 1.6119,
      "step": 123040
    },
    {
      "epoch": 0.24285936753070841,
      "grad_norm": 0.10653199255466461,
      "learning_rate": 7.602127145274354e-05,
      "loss": 1.6075,
      "step": 123072
    },
    {
      "epoch": 0.24292251349210486,
      "grad_norm": 0.10827381908893585,
      "learning_rate": 7.601493122837885e-05,
      "loss": 1.6183,
      "step": 123104
    },
    {
      "epoch": 0.2429856594535013,
      "grad_norm": 0.11906562000513077,
      "learning_rate": 7.600859100401417e-05,
      "loss": 1.6161,
      "step": 123136
    },
    {
      "epoch": 0.24304880541489773,
      "grad_norm": 0.1009439006447792,
      "learning_rate": 7.600225077964947e-05,
      "loss": 1.6167,
      "step": 123168
    },
    {
      "epoch": 0.24311195137629418,
      "grad_norm": 0.1109309047460556,
      "learning_rate": 7.599591055528477e-05,
      "loss": 1.6144,
      "step": 123200
    },
    {
      "epoch": 0.24317509733769063,
      "grad_norm": 0.10044911503791809,
      "learning_rate": 7.598957033092009e-05,
      "loss": 1.6068,
      "step": 123232
    },
    {
      "epoch": 0.24323824329908705,
      "grad_norm": 0.11914646625518799,
      "learning_rate": 7.59832301065554e-05,
      "loss": 1.6164,
      "step": 123264
    },
    {
      "epoch": 0.2433013892604835,
      "grad_norm": 0.09595836699008942,
      "learning_rate": 7.59768898821907e-05,
      "loss": 1.6242,
      "step": 123296
    },
    {
      "epoch": 0.24336453522187995,
      "grad_norm": 0.10583670437335968,
      "learning_rate": 7.597054965782602e-05,
      "loss": 1.6196,
      "step": 123328
    },
    {
      "epoch": 0.24342768118327637,
      "grad_norm": 0.10334964841604233,
      "learning_rate": 7.596420943346133e-05,
      "loss": 1.6191,
      "step": 123360
    },
    {
      "epoch": 0.24349082714467282,
      "grad_norm": 0.11159883439540863,
      "learning_rate": 7.595786920909664e-05,
      "loss": 1.6159,
      "step": 123392
    },
    {
      "epoch": 0.24355397310606927,
      "grad_norm": 0.1027933657169342,
      "learning_rate": 7.595152898473195e-05,
      "loss": 1.631,
      "step": 123424
    },
    {
      "epoch": 0.2436171190674657,
      "grad_norm": 0.11988910287618637,
      "learning_rate": 7.594518876036726e-05,
      "loss": 1.628,
      "step": 123456
    },
    {
      "epoch": 0.24368026502886214,
      "grad_norm": 0.10404536873102188,
      "learning_rate": 7.593884853600257e-05,
      "loss": 1.6239,
      "step": 123488
    },
    {
      "epoch": 0.2437434109902586,
      "grad_norm": 0.10335434973239899,
      "learning_rate": 7.593250831163789e-05,
      "loss": 1.6106,
      "step": 123520
    },
    {
      "epoch": 0.243806556951655,
      "grad_norm": 0.10373678803443909,
      "learning_rate": 7.59261680872732e-05,
      "loss": 1.6208,
      "step": 123552
    },
    {
      "epoch": 0.24386970291305146,
      "grad_norm": 0.10799238830804825,
      "learning_rate": 7.59198278629085e-05,
      "loss": 1.6094,
      "step": 123584
    },
    {
      "epoch": 0.2439328488744479,
      "grad_norm": 0.1039934903383255,
      "learning_rate": 7.59134876385438e-05,
      "loss": 1.6169,
      "step": 123616
    },
    {
      "epoch": 0.24399599483584433,
      "grad_norm": 0.09921015053987503,
      "learning_rate": 7.590714741417912e-05,
      "loss": 1.6056,
      "step": 123648
    },
    {
      "epoch": 0.24405914079724078,
      "grad_norm": 0.10938850045204163,
      "learning_rate": 7.590080718981443e-05,
      "loss": 1.6197,
      "step": 123680
    },
    {
      "epoch": 0.24412228675863723,
      "grad_norm": 0.10939442366361618,
      "learning_rate": 7.589446696544974e-05,
      "loss": 1.6115,
      "step": 123712
    },
    {
      "epoch": 0.24418543272003365,
      "grad_norm": 0.10520564764738083,
      "learning_rate": 7.588812674108505e-05,
      "loss": 1.6093,
      "step": 123744
    },
    {
      "epoch": 0.2442485786814301,
      "grad_norm": 0.10615213960409164,
      "learning_rate": 7.588178651672036e-05,
      "loss": 1.6184,
      "step": 123776
    },
    {
      "epoch": 0.24431172464282655,
      "grad_norm": 0.11066926270723343,
      "learning_rate": 7.587544629235568e-05,
      "loss": 1.6177,
      "step": 123808
    },
    {
      "epoch": 0.24437487060422297,
      "grad_norm": 0.1180395632982254,
      "learning_rate": 7.586910606799098e-05,
      "loss": 1.6358,
      "step": 123840
    },
    {
      "epoch": 0.24443801656561942,
      "grad_norm": 0.11051355302333832,
      "learning_rate": 7.58627658436263e-05,
      "loss": 1.6385,
      "step": 123872
    },
    {
      "epoch": 0.24450116252701587,
      "grad_norm": 0.12290903180837631,
      "learning_rate": 7.585642561926161e-05,
      "loss": 1.6114,
      "step": 123904
    },
    {
      "epoch": 0.24456430848841232,
      "grad_norm": 0.10706045478582382,
      "learning_rate": 7.585008539489692e-05,
      "loss": 1.624,
      "step": 123936
    },
    {
      "epoch": 0.24462745444980874,
      "grad_norm": 0.10720265656709671,
      "learning_rate": 7.584374517053223e-05,
      "loss": 1.6209,
      "step": 123968
    },
    {
      "epoch": 0.2446906004112052,
      "grad_norm": 0.10642603784799576,
      "learning_rate": 7.583740494616754e-05,
      "loss": 1.6215,
      "step": 124000
    },
    {
      "epoch": 0.24475374637260164,
      "grad_norm": 0.10877221822738647,
      "learning_rate": 7.583106472180284e-05,
      "loss": 1.6178,
      "step": 124032
    },
    {
      "epoch": 0.24481689233399806,
      "grad_norm": 0.11477097868919373,
      "learning_rate": 7.582472449743816e-05,
      "loss": 1.6281,
      "step": 124064
    },
    {
      "epoch": 0.2448800382953945,
      "grad_norm": 0.10735027492046356,
      "learning_rate": 7.581838427307346e-05,
      "loss": 1.6287,
      "step": 124096
    },
    {
      "epoch": 0.24494318425679096,
      "grad_norm": 0.11125703155994415,
      "learning_rate": 7.581204404870877e-05,
      "loss": 1.6065,
      "step": 124128
    },
    {
      "epoch": 0.24500633021818738,
      "grad_norm": 0.11185038089752197,
      "learning_rate": 7.580570382434409e-05,
      "loss": 1.6221,
      "step": 124160
    },
    {
      "epoch": 0.24506947617958383,
      "grad_norm": 0.10352326184511185,
      "learning_rate": 7.57993635999794e-05,
      "loss": 1.6224,
      "step": 124192
    },
    {
      "epoch": 0.24513262214098028,
      "grad_norm": 0.1192038357257843,
      "learning_rate": 7.579302337561471e-05,
      "loss": 1.6145,
      "step": 124224
    },
    {
      "epoch": 0.2451957681023767,
      "grad_norm": 0.11741012334823608,
      "learning_rate": 7.578668315125002e-05,
      "loss": 1.6174,
      "step": 124256
    },
    {
      "epoch": 0.24525891406377315,
      "grad_norm": 0.10417269915342331,
      "learning_rate": 7.578034292688533e-05,
      "loss": 1.6176,
      "step": 124288
    },
    {
      "epoch": 0.2453220600251696,
      "grad_norm": 0.10842493921518326,
      "learning_rate": 7.577400270252064e-05,
      "loss": 1.624,
      "step": 124320
    },
    {
      "epoch": 0.24538520598656602,
      "grad_norm": 0.10870698094367981,
      "learning_rate": 7.576766247815596e-05,
      "loss": 1.6218,
      "step": 124352
    },
    {
      "epoch": 0.24544835194796247,
      "grad_norm": 0.11051846295595169,
      "learning_rate": 7.576132225379126e-05,
      "loss": 1.6166,
      "step": 124384
    },
    {
      "epoch": 0.24551149790935892,
      "grad_norm": 0.1120854839682579,
      "learning_rate": 7.575498202942657e-05,
      "loss": 1.6255,
      "step": 124416
    },
    {
      "epoch": 0.24557464387075534,
      "grad_norm": 0.10920020937919617,
      "learning_rate": 7.574864180506188e-05,
      "loss": 1.6147,
      "step": 124448
    },
    {
      "epoch": 0.2456377898321518,
      "grad_norm": 0.10512637346982956,
      "learning_rate": 7.574230158069719e-05,
      "loss": 1.6308,
      "step": 124480
    },
    {
      "epoch": 0.24570093579354824,
      "grad_norm": 0.10023616254329681,
      "learning_rate": 7.573596135633249e-05,
      "loss": 1.6312,
      "step": 124512
    },
    {
      "epoch": 0.24576408175494466,
      "grad_norm": 0.1029682457447052,
      "learning_rate": 7.57296211319678e-05,
      "loss": 1.6175,
      "step": 124544
    },
    {
      "epoch": 0.2458272277163411,
      "grad_norm": 0.10144051909446716,
      "learning_rate": 7.572328090760312e-05,
      "loss": 1.6243,
      "step": 124576
    },
    {
      "epoch": 0.24589037367773756,
      "grad_norm": 0.10957455635070801,
      "learning_rate": 7.571694068323844e-05,
      "loss": 1.631,
      "step": 124608
    },
    {
      "epoch": 0.24595351963913398,
      "grad_norm": 0.1120717003941536,
      "learning_rate": 7.571060045887374e-05,
      "loss": 1.6076,
      "step": 124640
    },
    {
      "epoch": 0.24601666560053043,
      "grad_norm": 0.11887475103139877,
      "learning_rate": 7.570426023450905e-05,
      "loss": 1.6283,
      "step": 124672
    },
    {
      "epoch": 0.24607981156192688,
      "grad_norm": 0.10277330875396729,
      "learning_rate": 7.569792001014437e-05,
      "loss": 1.6106,
      "step": 124704
    },
    {
      "epoch": 0.2461429575233233,
      "grad_norm": 0.10494889318943024,
      "learning_rate": 7.569157978577968e-05,
      "loss": 1.626,
      "step": 124736
    },
    {
      "epoch": 0.24620610348471975,
      "grad_norm": 0.10465250164270401,
      "learning_rate": 7.568523956141498e-05,
      "loss": 1.6173,
      "step": 124768
    },
    {
      "epoch": 0.2462692494461162,
      "grad_norm": 0.10026372969150543,
      "learning_rate": 7.56788993370503e-05,
      "loss": 1.626,
      "step": 124800
    },
    {
      "epoch": 0.24633239540751262,
      "grad_norm": 0.11129790544509888,
      "learning_rate": 7.567255911268561e-05,
      "loss": 1.6253,
      "step": 124832
    },
    {
      "epoch": 0.24639554136890907,
      "grad_norm": 0.10483391582965851,
      "learning_rate": 7.566621888832091e-05,
      "loss": 1.6145,
      "step": 124864
    },
    {
      "epoch": 0.24645868733030551,
      "grad_norm": 0.10571115463972092,
      "learning_rate": 7.565987866395623e-05,
      "loss": 1.6224,
      "step": 124896
    },
    {
      "epoch": 0.24652183329170194,
      "grad_norm": 0.1026662290096283,
      "learning_rate": 7.565353843959153e-05,
      "loss": 1.6161,
      "step": 124928
    },
    {
      "epoch": 0.24658497925309839,
      "grad_norm": 0.10826900601387024,
      "learning_rate": 7.564719821522684e-05,
      "loss": 1.6068,
      "step": 124960
    },
    {
      "epoch": 0.24664812521449483,
      "grad_norm": 0.10200421512126923,
      "learning_rate": 7.564085799086216e-05,
      "loss": 1.6164,
      "step": 124992
    },
    {
      "epoch": 0.24671127117589126,
      "grad_norm": 0.10194798558950424,
      "learning_rate": 7.563451776649747e-05,
      "loss": 1.6234,
      "step": 125024
    },
    {
      "epoch": 0.2467744171372877,
      "grad_norm": 0.10575179755687714,
      "learning_rate": 7.562817754213277e-05,
      "loss": 1.6135,
      "step": 125056
    },
    {
      "epoch": 0.24683756309868415,
      "grad_norm": 0.11280045658349991,
      "learning_rate": 7.562183731776809e-05,
      "loss": 1.6241,
      "step": 125088
    },
    {
      "epoch": 0.24690070906008058,
      "grad_norm": 0.10755287855863571,
      "learning_rate": 7.56154970934034e-05,
      "loss": 1.6103,
      "step": 125120
    },
    {
      "epoch": 0.24696385502147702,
      "grad_norm": 0.10866934806108475,
      "learning_rate": 7.560915686903871e-05,
      "loss": 1.6105,
      "step": 125152
    },
    {
      "epoch": 0.24702700098287347,
      "grad_norm": 0.10039102286100388,
      "learning_rate": 7.560281664467402e-05,
      "loss": 1.6229,
      "step": 125184
    },
    {
      "epoch": 0.2470901469442699,
      "grad_norm": 0.10975603759288788,
      "learning_rate": 7.559647642030933e-05,
      "loss": 1.6343,
      "step": 125216
    },
    {
      "epoch": 0.24715329290566634,
      "grad_norm": 0.11331671476364136,
      "learning_rate": 7.559013619594464e-05,
      "loss": 1.617,
      "step": 125248
    },
    {
      "epoch": 0.2472164388670628,
      "grad_norm": 0.10674507170915604,
      "learning_rate": 7.558379597157995e-05,
      "loss": 1.6151,
      "step": 125280
    },
    {
      "epoch": 0.24727958482845921,
      "grad_norm": 0.10739534348249435,
      "learning_rate": 7.557745574721525e-05,
      "loss": 1.6082,
      "step": 125312
    },
    {
      "epoch": 0.24734273078985566,
      "grad_norm": 0.10259276628494263,
      "learning_rate": 7.557111552285056e-05,
      "loss": 1.6138,
      "step": 125344
    },
    {
      "epoch": 0.2474058767512521,
      "grad_norm": 0.1094990149140358,
      "learning_rate": 7.556477529848588e-05,
      "loss": 1.6166,
      "step": 125376
    },
    {
      "epoch": 0.24746902271264853,
      "grad_norm": 0.10647404193878174,
      "learning_rate": 7.555843507412119e-05,
      "loss": 1.6132,
      "step": 125408
    },
    {
      "epoch": 0.24753216867404498,
      "grad_norm": 0.10291962325572968,
      "learning_rate": 7.555209484975649e-05,
      "loss": 1.6062,
      "step": 125440
    },
    {
      "epoch": 0.24759531463544143,
      "grad_norm": 0.1110311895608902,
      "learning_rate": 7.55457546253918e-05,
      "loss": 1.6241,
      "step": 125472
    },
    {
      "epoch": 0.24765846059683785,
      "grad_norm": 0.1133258193731308,
      "learning_rate": 7.553941440102712e-05,
      "loss": 1.6159,
      "step": 125504
    },
    {
      "epoch": 0.2477216065582343,
      "grad_norm": 0.10049762576818466,
      "learning_rate": 7.553307417666244e-05,
      "loss": 1.5989,
      "step": 125536
    },
    {
      "epoch": 0.24778475251963075,
      "grad_norm": 0.10998736321926117,
      "learning_rate": 7.552673395229775e-05,
      "loss": 1.6183,
      "step": 125568
    },
    {
      "epoch": 0.2478478984810272,
      "grad_norm": 0.10934124886989594,
      "learning_rate": 7.552039372793305e-05,
      "loss": 1.6179,
      "step": 125600
    },
    {
      "epoch": 0.24791104444242362,
      "grad_norm": 0.10284886509180069,
      "learning_rate": 7.551405350356837e-05,
      "loss": 1.599,
      "step": 125632
    },
    {
      "epoch": 0.24797419040382007,
      "grad_norm": 0.104775071144104,
      "learning_rate": 7.550771327920368e-05,
      "loss": 1.6121,
      "step": 125664
    },
    {
      "epoch": 0.24803733636521652,
      "grad_norm": 0.10824383795261383,
      "learning_rate": 7.550137305483898e-05,
      "loss": 1.6299,
      "step": 125696
    },
    {
      "epoch": 0.24810048232661294,
      "grad_norm": 0.10084754228591919,
      "learning_rate": 7.549503283047428e-05,
      "loss": 1.6213,
      "step": 125728
    },
    {
      "epoch": 0.2481636282880094,
      "grad_norm": 0.11840032786130905,
      "learning_rate": 7.54886926061096e-05,
      "loss": 1.6227,
      "step": 125760
    },
    {
      "epoch": 0.24822677424940584,
      "grad_norm": 0.10971036553382874,
      "learning_rate": 7.548235238174491e-05,
      "loss": 1.6164,
      "step": 125792
    },
    {
      "epoch": 0.24828992021080226,
      "grad_norm": 0.10688149183988571,
      "learning_rate": 7.547601215738023e-05,
      "loss": 1.6259,
      "step": 125824
    },
    {
      "epoch": 0.2483530661721987,
      "grad_norm": 0.10770287364721298,
      "learning_rate": 7.546967193301553e-05,
      "loss": 1.6299,
      "step": 125856
    },
    {
      "epoch": 0.24841621213359516,
      "grad_norm": 0.1067945659160614,
      "learning_rate": 7.546333170865084e-05,
      "loss": 1.6107,
      "step": 125888
    },
    {
      "epoch": 0.24847935809499158,
      "grad_norm": 0.10258124768733978,
      "learning_rate": 7.545699148428616e-05,
      "loss": 1.6125,
      "step": 125920
    },
    {
      "epoch": 0.24854250405638803,
      "grad_norm": 0.11382295191287994,
      "learning_rate": 7.545065125992147e-05,
      "loss": 1.607,
      "step": 125952
    },
    {
      "epoch": 0.24860565001778448,
      "grad_norm": 0.10734063386917114,
      "learning_rate": 7.544431103555677e-05,
      "loss": 1.616,
      "step": 125984
    },
    {
      "epoch": 0.2486687959791809,
      "grad_norm": 0.10081788152456284,
      "learning_rate": 7.543797081119209e-05,
      "loss": 1.6169,
      "step": 126016
    },
    {
      "epoch": 0.24873194194057735,
      "grad_norm": 0.10751203447580338,
      "learning_rate": 7.54316305868274e-05,
      "loss": 1.6135,
      "step": 126048
    },
    {
      "epoch": 0.2487950879019738,
      "grad_norm": 0.10247769206762314,
      "learning_rate": 7.54252903624627e-05,
      "loss": 1.6079,
      "step": 126080
    },
    {
      "epoch": 0.24885823386337022,
      "grad_norm": 0.10922478139400482,
      "learning_rate": 7.541895013809802e-05,
      "loss": 1.6103,
      "step": 126112
    },
    {
      "epoch": 0.24892137982476667,
      "grad_norm": 0.10713105648756027,
      "learning_rate": 7.541260991373332e-05,
      "loss": 1.6267,
      "step": 126144
    },
    {
      "epoch": 0.24898452578616312,
      "grad_norm": 0.10570314526557922,
      "learning_rate": 7.540626968936863e-05,
      "loss": 1.6226,
      "step": 126176
    },
    {
      "epoch": 0.24904767174755954,
      "grad_norm": 0.11996707320213318,
      "learning_rate": 7.539992946500395e-05,
      "loss": 1.6128,
      "step": 126208
    },
    {
      "epoch": 0.249110817708956,
      "grad_norm": 0.10169323533773422,
      "learning_rate": 7.539358924063926e-05,
      "loss": 1.6204,
      "step": 126240
    },
    {
      "epoch": 0.24917396367035244,
      "grad_norm": 0.10233422368764877,
      "learning_rate": 7.538724901627456e-05,
      "loss": 1.6225,
      "step": 126272
    },
    {
      "epoch": 0.24923710963174886,
      "grad_norm": 0.10392436385154724,
      "learning_rate": 7.538090879190988e-05,
      "loss": 1.6096,
      "step": 126304
    },
    {
      "epoch": 0.2493002555931453,
      "grad_norm": 0.10452143847942352,
      "learning_rate": 7.537456856754519e-05,
      "loss": 1.6239,
      "step": 126336
    },
    {
      "epoch": 0.24936340155454176,
      "grad_norm": 0.10403753817081451,
      "learning_rate": 7.53682283431805e-05,
      "loss": 1.6102,
      "step": 126368
    },
    {
      "epoch": 0.24942654751593818,
      "grad_norm": 0.10682003945112228,
      "learning_rate": 7.53618881188158e-05,
      "loss": 1.6267,
      "step": 126400
    },
    {
      "epoch": 0.24948969347733463,
      "grad_norm": 0.11028899252414703,
      "learning_rate": 7.535554789445112e-05,
      "loss": 1.6113,
      "step": 126432
    },
    {
      "epoch": 0.24955283943873108,
      "grad_norm": 0.10885583609342575,
      "learning_rate": 7.534920767008644e-05,
      "loss": 1.6332,
      "step": 126464
    },
    {
      "epoch": 0.2496159854001275,
      "grad_norm": 0.11095141619443893,
      "learning_rate": 7.534286744572174e-05,
      "loss": 1.6187,
      "step": 126496
    },
    {
      "epoch": 0.24967913136152395,
      "grad_norm": 0.10756413638591766,
      "learning_rate": 7.533652722135705e-05,
      "loss": 1.6119,
      "step": 126528
    },
    {
      "epoch": 0.2497422773229204,
      "grad_norm": 0.11318307369947433,
      "learning_rate": 7.533018699699235e-05,
      "loss": 1.6234,
      "step": 126560
    },
    {
      "epoch": 0.24980542328431682,
      "grad_norm": 0.11176694184541702,
      "learning_rate": 7.532384677262767e-05,
      "loss": 1.615,
      "step": 126592
    },
    {
      "epoch": 0.24986856924571327,
      "grad_norm": 0.09958535432815552,
      "learning_rate": 7.531750654826298e-05,
      "loss": 1.6073,
      "step": 126624
    },
    {
      "epoch": 0.24993171520710972,
      "grad_norm": 0.09965774416923523,
      "learning_rate": 7.531116632389828e-05,
      "loss": 1.6066,
      "step": 126656
    },
    {
      "epoch": 0.24999486116850614,
      "grad_norm": 0.11404179781675339,
      "learning_rate": 7.53048260995336e-05,
      "loss": 1.6205,
      "step": 126688
    },
    {
      "epoch": 0.2500580071299026,
      "grad_norm": 0.10913736373186111,
      "learning_rate": 7.529848587516891e-05,
      "loss": 1.5992,
      "step": 126720
    },
    {
      "epoch": 0.250121153091299,
      "grad_norm": 0.10473071783781052,
      "learning_rate": 7.529214565080423e-05,
      "loss": 1.6016,
      "step": 126752
    },
    {
      "epoch": 0.25018429905269546,
      "grad_norm": 0.1100083440542221,
      "learning_rate": 7.528580542643953e-05,
      "loss": 1.6142,
      "step": 126784
    },
    {
      "epoch": 0.2502474450140919,
      "grad_norm": 0.11019501835107803,
      "learning_rate": 7.527946520207484e-05,
      "loss": 1.6168,
      "step": 126816
    },
    {
      "epoch": 0.25031059097548836,
      "grad_norm": 0.10891743749380112,
      "learning_rate": 7.527312497771016e-05,
      "loss": 1.6129,
      "step": 126848
    },
    {
      "epoch": 0.2503737369368848,
      "grad_norm": 0.10849962383508682,
      "learning_rate": 7.526678475334547e-05,
      "loss": 1.6029,
      "step": 126880
    },
    {
      "epoch": 0.25043688289828125,
      "grad_norm": 0.10497497767210007,
      "learning_rate": 7.526044452898077e-05,
      "loss": 1.6155,
      "step": 126912
    },
    {
      "epoch": 0.25050002885967765,
      "grad_norm": 0.10756964236497879,
      "learning_rate": 7.525410430461609e-05,
      "loss": 1.6181,
      "step": 126944
    },
    {
      "epoch": 0.2505631748210741,
      "grad_norm": 0.10947880893945694,
      "learning_rate": 7.524776408025139e-05,
      "loss": 1.6008,
      "step": 126976
    },
    {
      "epoch": 0.25062632078247055,
      "grad_norm": 0.11128494143486023,
      "learning_rate": 7.52414238558867e-05,
      "loss": 1.6238,
      "step": 127008
    },
    {
      "epoch": 0.250689466743867,
      "grad_norm": 0.10451777279376984,
      "learning_rate": 7.523508363152202e-05,
      "loss": 1.6196,
      "step": 127040
    },
    {
      "epoch": 0.25075261270526344,
      "grad_norm": 0.11082389205694199,
      "learning_rate": 7.522874340715732e-05,
      "loss": 1.6312,
      "step": 127072
    },
    {
      "epoch": 0.2508157586666599,
      "grad_norm": 0.10895415395498276,
      "learning_rate": 7.522240318279263e-05,
      "loss": 1.6101,
      "step": 127104
    },
    {
      "epoch": 0.2508789046280563,
      "grad_norm": 0.11142008006572723,
      "learning_rate": 7.521606295842795e-05,
      "loss": 1.6148,
      "step": 127136
    },
    {
      "epoch": 0.25094205058945274,
      "grad_norm": 0.10830048471689224,
      "learning_rate": 7.520972273406326e-05,
      "loss": 1.6213,
      "step": 127168
    },
    {
      "epoch": 0.2510051965508492,
      "grad_norm": 0.10632851719856262,
      "learning_rate": 7.520338250969856e-05,
      "loss": 1.6179,
      "step": 127200
    },
    {
      "epoch": 0.25106834251224563,
      "grad_norm": 0.1038040891289711,
      "learning_rate": 7.519704228533388e-05,
      "loss": 1.6056,
      "step": 127232
    },
    {
      "epoch": 0.2511314884736421,
      "grad_norm": 0.11971843987703323,
      "learning_rate": 7.519070206096919e-05,
      "loss": 1.6144,
      "step": 127264
    },
    {
      "epoch": 0.25119463443503853,
      "grad_norm": 0.10849187523126602,
      "learning_rate": 7.51843618366045e-05,
      "loss": 1.6181,
      "step": 127296
    },
    {
      "epoch": 0.2512577803964349,
      "grad_norm": 0.1097431480884552,
      "learning_rate": 7.517802161223981e-05,
      "loss": 1.5983,
      "step": 127328
    },
    {
      "epoch": 0.2513209263578314,
      "grad_norm": 0.11160177737474442,
      "learning_rate": 7.517168138787511e-05,
      "loss": 1.6177,
      "step": 127360
    },
    {
      "epoch": 0.2513840723192278,
      "grad_norm": 0.10210192203521729,
      "learning_rate": 7.516534116351042e-05,
      "loss": 1.624,
      "step": 127392
    },
    {
      "epoch": 0.2514472182806243,
      "grad_norm": 0.10520271211862564,
      "learning_rate": 7.515900093914574e-05,
      "loss": 1.6267,
      "step": 127424
    },
    {
      "epoch": 0.2515103642420207,
      "grad_norm": 0.1065123975276947,
      "learning_rate": 7.515266071478104e-05,
      "loss": 1.6208,
      "step": 127456
    },
    {
      "epoch": 0.25157351020341717,
      "grad_norm": 0.11302441358566284,
      "learning_rate": 7.514632049041635e-05,
      "loss": 1.6248,
      "step": 127488
    },
    {
      "epoch": 0.25163665616481357,
      "grad_norm": 0.1007496640086174,
      "learning_rate": 7.513998026605167e-05,
      "loss": 1.6311,
      "step": 127520
    },
    {
      "epoch": 0.25169980212621,
      "grad_norm": 0.10568457096815109,
      "learning_rate": 7.513364004168698e-05,
      "loss": 1.6139,
      "step": 127552
    },
    {
      "epoch": 0.25176294808760646,
      "grad_norm": 0.1054898202419281,
      "learning_rate": 7.51272998173223e-05,
      "loss": 1.6233,
      "step": 127584
    },
    {
      "epoch": 0.2518260940490029,
      "grad_norm": 0.10831870883703232,
      "learning_rate": 7.51209595929576e-05,
      "loss": 1.6114,
      "step": 127616
    },
    {
      "epoch": 0.25188924001039936,
      "grad_norm": 0.1130576878786087,
      "learning_rate": 7.511461936859291e-05,
      "loss": 1.6112,
      "step": 127648
    },
    {
      "epoch": 0.2519523859717958,
      "grad_norm": 0.1042439192533493,
      "learning_rate": 7.510827914422823e-05,
      "loss": 1.6252,
      "step": 127680
    },
    {
      "epoch": 0.2520155319331922,
      "grad_norm": 0.10637133568525314,
      "learning_rate": 7.510193891986354e-05,
      "loss": 1.6173,
      "step": 127712
    },
    {
      "epoch": 0.25207867789458865,
      "grad_norm": 0.10836733877658844,
      "learning_rate": 7.509559869549884e-05,
      "loss": 1.6305,
      "step": 127744
    },
    {
      "epoch": 0.2521418238559851,
      "grad_norm": 0.10669766366481781,
      "learning_rate": 7.508925847113414e-05,
      "loss": 1.616,
      "step": 127776
    },
    {
      "epoch": 0.25220496981738155,
      "grad_norm": 0.111818328499794,
      "learning_rate": 7.508291824676946e-05,
      "loss": 1.6267,
      "step": 127808
    },
    {
      "epoch": 0.252268115778778,
      "grad_norm": 0.11270266771316528,
      "learning_rate": 7.507657802240477e-05,
      "loss": 1.603,
      "step": 127840
    },
    {
      "epoch": 0.25233126174017445,
      "grad_norm": 0.10993698984384537,
      "learning_rate": 7.507023779804007e-05,
      "loss": 1.6291,
      "step": 127872
    },
    {
      "epoch": 0.25239440770157084,
      "grad_norm": 0.09985803812742233,
      "learning_rate": 7.506389757367539e-05,
      "loss": 1.6218,
      "step": 127904
    },
    {
      "epoch": 0.2524575536629673,
      "grad_norm": 0.10485244542360306,
      "learning_rate": 7.50575573493107e-05,
      "loss": 1.634,
      "step": 127936
    },
    {
      "epoch": 0.25252069962436374,
      "grad_norm": 0.10270264744758606,
      "learning_rate": 7.505121712494602e-05,
      "loss": 1.6053,
      "step": 127968
    },
    {
      "epoch": 0.2525838455857602,
      "grad_norm": 0.10258591920137405,
      "learning_rate": 7.504487690058132e-05,
      "loss": 1.6209,
      "step": 128000
    },
    {
      "epoch": 0.25264699154715664,
      "grad_norm": 0.10999858379364014,
      "learning_rate": 7.503853667621663e-05,
      "loss": 1.6182,
      "step": 128032
    },
    {
      "epoch": 0.2527101375085531,
      "grad_norm": 0.10758821666240692,
      "learning_rate": 7.503219645185195e-05,
      "loss": 1.6074,
      "step": 128064
    },
    {
      "epoch": 0.2527732834699495,
      "grad_norm": 0.10051969438791275,
      "learning_rate": 7.502585622748726e-05,
      "loss": 1.6216,
      "step": 128096
    },
    {
      "epoch": 0.25283642943134593,
      "grad_norm": 0.10678692907094955,
      "learning_rate": 7.501951600312256e-05,
      "loss": 1.5979,
      "step": 128128
    },
    {
      "epoch": 0.2528995753927424,
      "grad_norm": 0.09789399802684784,
      "learning_rate": 7.501317577875788e-05,
      "loss": 1.5982,
      "step": 128160
    },
    {
      "epoch": 0.25296272135413883,
      "grad_norm": 0.10215232521295547,
      "learning_rate": 7.500683555439318e-05,
      "loss": 1.614,
      "step": 128192
    },
    {
      "epoch": 0.2530258673155353,
      "grad_norm": 0.10916431248188019,
      "learning_rate": 7.500049533002849e-05,
      "loss": 1.6092,
      "step": 128224
    },
    {
      "epoch": 0.25308901327693173,
      "grad_norm": 0.1081867516040802,
      "learning_rate": 7.499415510566381e-05,
      "loss": 1.6237,
      "step": 128256
    },
    {
      "epoch": 0.2531521592383282,
      "grad_norm": 0.11384835094213486,
      "learning_rate": 7.498781488129911e-05,
      "loss": 1.6185,
      "step": 128288
    },
    {
      "epoch": 0.25321530519972457,
      "grad_norm": 0.10165699571371078,
      "learning_rate": 7.498147465693442e-05,
      "loss": 1.6294,
      "step": 128320
    },
    {
      "epoch": 0.253278451161121,
      "grad_norm": 0.10673161596059799,
      "learning_rate": 7.497513443256974e-05,
      "loss": 1.6227,
      "step": 128352
    },
    {
      "epoch": 0.25334159712251747,
      "grad_norm": 0.1099945604801178,
      "learning_rate": 7.496879420820505e-05,
      "loss": 1.6248,
      "step": 128384
    },
    {
      "epoch": 0.2534047430839139,
      "grad_norm": 0.10666900873184204,
      "learning_rate": 7.496245398384035e-05,
      "loss": 1.6355,
      "step": 128416
    },
    {
      "epoch": 0.25346788904531037,
      "grad_norm": 0.10574255883693695,
      "learning_rate": 7.495611375947567e-05,
      "loss": 1.6118,
      "step": 128448
    },
    {
      "epoch": 0.2535310350067068,
      "grad_norm": 0.10493651032447815,
      "learning_rate": 7.494977353511098e-05,
      "loss": 1.5986,
      "step": 128480
    },
    {
      "epoch": 0.2535941809681032,
      "grad_norm": 0.1075049415230751,
      "learning_rate": 7.49434333107463e-05,
      "loss": 1.6079,
      "step": 128512
    },
    {
      "epoch": 0.25365732692949966,
      "grad_norm": 0.131600484251976,
      "learning_rate": 7.49370930863816e-05,
      "loss": 1.6193,
      "step": 128544
    },
    {
      "epoch": 0.2537204728908961,
      "grad_norm": 0.1113981083035469,
      "learning_rate": 7.493075286201691e-05,
      "loss": 1.6224,
      "step": 128576
    },
    {
      "epoch": 0.25378361885229256,
      "grad_norm": 0.10733971744775772,
      "learning_rate": 7.492441263765221e-05,
      "loss": 1.6104,
      "step": 128608
    },
    {
      "epoch": 0.253846764813689,
      "grad_norm": 0.10786531865596771,
      "learning_rate": 7.491807241328753e-05,
      "loss": 1.5924,
      "step": 128640
    },
    {
      "epoch": 0.25390991077508546,
      "grad_norm": 0.10711334645748138,
      "learning_rate": 7.491173218892283e-05,
      "loss": 1.6181,
      "step": 128672
    },
    {
      "epoch": 0.25397305673648185,
      "grad_norm": 0.10400371253490448,
      "learning_rate": 7.490539196455814e-05,
      "loss": 1.6096,
      "step": 128704
    },
    {
      "epoch": 0.2540362026978783,
      "grad_norm": 0.10419753938913345,
      "learning_rate": 7.489905174019346e-05,
      "loss": 1.6029,
      "step": 128736
    },
    {
      "epoch": 0.25409934865927475,
      "grad_norm": 0.11558572947978973,
      "learning_rate": 7.489271151582877e-05,
      "loss": 1.6147,
      "step": 128768
    },
    {
      "epoch": 0.2541624946206712,
      "grad_norm": 0.10490277409553528,
      "learning_rate": 7.488637129146409e-05,
      "loss": 1.6276,
      "step": 128800
    },
    {
      "epoch": 0.25422564058206765,
      "grad_norm": 0.11033342778682709,
      "learning_rate": 7.488003106709939e-05,
      "loss": 1.605,
      "step": 128832
    },
    {
      "epoch": 0.2542887865434641,
      "grad_norm": 0.11046944558620453,
      "learning_rate": 7.48736908427347e-05,
      "loss": 1.6261,
      "step": 128864
    },
    {
      "epoch": 0.2543519325048605,
      "grad_norm": 0.108438640832901,
      "learning_rate": 7.486735061837002e-05,
      "loss": 1.6084,
      "step": 128896
    },
    {
      "epoch": 0.25441507846625694,
      "grad_norm": 0.10542812943458557,
      "learning_rate": 7.486101039400533e-05,
      "loss": 1.5991,
      "step": 128928
    },
    {
      "epoch": 0.2544782244276534,
      "grad_norm": 0.10477355122566223,
      "learning_rate": 7.485467016964063e-05,
      "loss": 1.6154,
      "step": 128960
    },
    {
      "epoch": 0.25454137038904984,
      "grad_norm": 0.10619034618139267,
      "learning_rate": 7.484832994527595e-05,
      "loss": 1.6184,
      "step": 128992
    },
    {
      "epoch": 0.2546045163504463,
      "grad_norm": 0.09891405701637268,
      "learning_rate": 7.484198972091125e-05,
      "loss": 1.6136,
      "step": 129024
    },
    {
      "epoch": 0.25466766231184274,
      "grad_norm": 0.10589949041604996,
      "learning_rate": 7.483564949654656e-05,
      "loss": 1.6279,
      "step": 129056
    },
    {
      "epoch": 0.25473080827323913,
      "grad_norm": 0.11763922870159149,
      "learning_rate": 7.482930927218186e-05,
      "loss": 1.6233,
      "step": 129088
    },
    {
      "epoch": 0.2547939542346356,
      "grad_norm": 0.104377880692482,
      "learning_rate": 7.482296904781718e-05,
      "loss": 1.6032,
      "step": 129120
    },
    {
      "epoch": 0.254857100196032,
      "grad_norm": 0.10092160105705261,
      "learning_rate": 7.481662882345249e-05,
      "loss": 1.6197,
      "step": 129152
    },
    {
      "epoch": 0.2549202461574285,
      "grad_norm": 0.10802380740642548,
      "learning_rate": 7.481028859908781e-05,
      "loss": 1.6222,
      "step": 129184
    },
    {
      "epoch": 0.2549833921188249,
      "grad_norm": 0.11016335338354111,
      "learning_rate": 7.480394837472311e-05,
      "loss": 1.6021,
      "step": 129216
    },
    {
      "epoch": 0.2550465380802214,
      "grad_norm": 0.10332827270030975,
      "learning_rate": 7.479760815035842e-05,
      "loss": 1.6129,
      "step": 129248
    },
    {
      "epoch": 0.25510968404161777,
      "grad_norm": 0.10060596466064453,
      "learning_rate": 7.479126792599374e-05,
      "loss": 1.606,
      "step": 129280
    },
    {
      "epoch": 0.2551728300030142,
      "grad_norm": 0.10596133768558502,
      "learning_rate": 7.478492770162905e-05,
      "loss": 1.6114,
      "step": 129312
    },
    {
      "epoch": 0.25523597596441067,
      "grad_norm": 0.1052638590335846,
      "learning_rate": 7.477858747726435e-05,
      "loss": 1.6143,
      "step": 129344
    },
    {
      "epoch": 0.2552991219258071,
      "grad_norm": 0.09929129481315613,
      "learning_rate": 7.477224725289967e-05,
      "loss": 1.6181,
      "step": 129376
    },
    {
      "epoch": 0.25536226788720356,
      "grad_norm": 0.10516052693128586,
      "learning_rate": 7.476590702853498e-05,
      "loss": 1.6074,
      "step": 129408
    },
    {
      "epoch": 0.2554254138486,
      "grad_norm": 0.10188374668359756,
      "learning_rate": 7.475956680417028e-05,
      "loss": 1.6095,
      "step": 129440
    },
    {
      "epoch": 0.2554885598099964,
      "grad_norm": 0.10364662110805511,
      "learning_rate": 7.47532265798056e-05,
      "loss": 1.6175,
      "step": 129472
    },
    {
      "epoch": 0.25555170577139286,
      "grad_norm": 0.11409908533096313,
      "learning_rate": 7.47468863554409e-05,
      "loss": 1.6104,
      "step": 129504
    },
    {
      "epoch": 0.2556148517327893,
      "grad_norm": 0.11226414144039154,
      "learning_rate": 7.474054613107621e-05,
      "loss": 1.6139,
      "step": 129536
    },
    {
      "epoch": 0.25567799769418575,
      "grad_norm": 0.10771206766366959,
      "learning_rate": 7.473420590671153e-05,
      "loss": 1.6138,
      "step": 129568
    },
    {
      "epoch": 0.2557411436555822,
      "grad_norm": 0.10493512451648712,
      "learning_rate": 7.472786568234684e-05,
      "loss": 1.6201,
      "step": 129600
    },
    {
      "epoch": 0.25580428961697865,
      "grad_norm": 0.12275021523237228,
      "learning_rate": 7.472152545798214e-05,
      "loss": 1.6149,
      "step": 129632
    },
    {
      "epoch": 0.25586743557837505,
      "grad_norm": 0.1040772870182991,
      "learning_rate": 7.471518523361746e-05,
      "loss": 1.6217,
      "step": 129664
    },
    {
      "epoch": 0.2559305815397715,
      "grad_norm": 0.11437197029590607,
      "learning_rate": 7.470884500925277e-05,
      "loss": 1.6207,
      "step": 129696
    },
    {
      "epoch": 0.25599372750116794,
      "grad_norm": 0.12347360700368881,
      "learning_rate": 7.470250478488809e-05,
      "loss": 1.6151,
      "step": 129728
    },
    {
      "epoch": 0.2560568734625644,
      "grad_norm": 0.10722575336694717,
      "learning_rate": 7.469616456052339e-05,
      "loss": 1.6343,
      "step": 129760
    },
    {
      "epoch": 0.25612001942396084,
      "grad_norm": 0.10563173145055771,
      "learning_rate": 7.46898243361587e-05,
      "loss": 1.6104,
      "step": 129792
    },
    {
      "epoch": 0.2561831653853573,
      "grad_norm": 0.09876292943954468,
      "learning_rate": 7.468348411179402e-05,
      "loss": 1.6121,
      "step": 129824
    },
    {
      "epoch": 0.25624631134675374,
      "grad_norm": 0.11074431985616684,
      "learning_rate": 7.467714388742932e-05,
      "loss": 1.6155,
      "step": 129856
    },
    {
      "epoch": 0.25630945730815013,
      "grad_norm": 0.10697343945503235,
      "learning_rate": 7.467080366306462e-05,
      "loss": 1.6122,
      "step": 129888
    },
    {
      "epoch": 0.2563726032695466,
      "grad_norm": 0.10713440924882889,
      "learning_rate": 7.466446343869993e-05,
      "loss": 1.6095,
      "step": 129920
    },
    {
      "epoch": 0.25643574923094303,
      "grad_norm": 0.10320103168487549,
      "learning_rate": 7.465812321433525e-05,
      "loss": 1.591,
      "step": 129952
    },
    {
      "epoch": 0.2564988951923395,
      "grad_norm": 0.10894148051738739,
      "learning_rate": 7.465178298997056e-05,
      "loss": 1.6067,
      "step": 129984
    },
    {
      "epoch": 0.25656204115373593,
      "grad_norm": 0.10409074276685715,
      "learning_rate": 7.464544276560586e-05,
      "loss": 1.5955,
      "step": 130016
    },
    {
      "epoch": 0.2566251871151324,
      "grad_norm": 0.11294655501842499,
      "learning_rate": 7.463910254124118e-05,
      "loss": 1.6076,
      "step": 130048
    },
    {
      "epoch": 0.2566883330765288,
      "grad_norm": 0.10786151140928268,
      "learning_rate": 7.46327623168765e-05,
      "loss": 1.6136,
      "step": 130080
    },
    {
      "epoch": 0.2567514790379252,
      "grad_norm": 0.10605989396572113,
      "learning_rate": 7.462642209251181e-05,
      "loss": 1.6146,
      "step": 130112
    },
    {
      "epoch": 0.2568146249993217,
      "grad_norm": 0.11035019159317017,
      "learning_rate": 7.462008186814712e-05,
      "loss": 1.611,
      "step": 130144
    },
    {
      "epoch": 0.2568777709607181,
      "grad_norm": 0.11705031245946884,
      "learning_rate": 7.461374164378242e-05,
      "loss": 1.6104,
      "step": 130176
    },
    {
      "epoch": 0.25694091692211457,
      "grad_norm": 0.10468898713588715,
      "learning_rate": 7.460740141941774e-05,
      "loss": 1.6111,
      "step": 130208
    },
    {
      "epoch": 0.257004062883511,
      "grad_norm": 0.10858344286680222,
      "learning_rate": 7.460106119505304e-05,
      "loss": 1.6115,
      "step": 130240
    },
    {
      "epoch": 0.2570672088449074,
      "grad_norm": 0.10601144284009933,
      "learning_rate": 7.459472097068835e-05,
      "loss": 1.6158,
      "step": 130272
    },
    {
      "epoch": 0.25713035480630386,
      "grad_norm": 0.10758187621831894,
      "learning_rate": 7.458838074632365e-05,
      "loss": 1.6208,
      "step": 130304
    },
    {
      "epoch": 0.2571935007677003,
      "grad_norm": 0.10547227412462234,
      "learning_rate": 7.458204052195897e-05,
      "loss": 1.6207,
      "step": 130336
    },
    {
      "epoch": 0.25725664672909676,
      "grad_norm": 0.10425049066543579,
      "learning_rate": 7.457570029759428e-05,
      "loss": 1.6184,
      "step": 130368
    },
    {
      "epoch": 0.2573197926904932,
      "grad_norm": 0.10619667172431946,
      "learning_rate": 7.45693600732296e-05,
      "loss": 1.6157,
      "step": 130400
    },
    {
      "epoch": 0.25738293865188966,
      "grad_norm": 0.11281777918338776,
      "learning_rate": 7.45630198488649e-05,
      "loss": 1.6193,
      "step": 130432
    },
    {
      "epoch": 0.25744608461328605,
      "grad_norm": 0.10736407339572906,
      "learning_rate": 7.455667962450021e-05,
      "loss": 1.6158,
      "step": 130464
    },
    {
      "epoch": 0.2575092305746825,
      "grad_norm": 0.10818787664175034,
      "learning_rate": 7.455033940013553e-05,
      "loss": 1.6045,
      "step": 130496
    },
    {
      "epoch": 0.25757237653607895,
      "grad_norm": 0.11346203088760376,
      "learning_rate": 7.454399917577084e-05,
      "loss": 1.6309,
      "step": 130528
    },
    {
      "epoch": 0.2576355224974754,
      "grad_norm": 0.105833500623703,
      "learning_rate": 7.453765895140614e-05,
      "loss": 1.6085,
      "step": 130560
    },
    {
      "epoch": 0.25769866845887185,
      "grad_norm": 0.09954508394002914,
      "learning_rate": 7.453131872704146e-05,
      "loss": 1.6079,
      "step": 130592
    },
    {
      "epoch": 0.2577618144202683,
      "grad_norm": 0.11163540929555893,
      "learning_rate": 7.452497850267677e-05,
      "loss": 1.6159,
      "step": 130624
    },
    {
      "epoch": 0.2578249603816647,
      "grad_norm": 0.12326701730489731,
      "learning_rate": 7.451863827831207e-05,
      "loss": 1.6338,
      "step": 130656
    },
    {
      "epoch": 0.25788810634306114,
      "grad_norm": 0.10490484535694122,
      "learning_rate": 7.451229805394739e-05,
      "loss": 1.6089,
      "step": 130688
    },
    {
      "epoch": 0.2579512523044576,
      "grad_norm": 0.11521980166435242,
      "learning_rate": 7.450595782958269e-05,
      "loss": 1.6072,
      "step": 130720
    },
    {
      "epoch": 0.25801439826585404,
      "grad_norm": 0.11303827166557312,
      "learning_rate": 7.4499617605218e-05,
      "loss": 1.6084,
      "step": 130752
    },
    {
      "epoch": 0.2580775442272505,
      "grad_norm": 0.1141236275434494,
      "learning_rate": 7.449327738085332e-05,
      "loss": 1.6111,
      "step": 130784
    },
    {
      "epoch": 0.25814069018864694,
      "grad_norm": 0.11162339895963669,
      "learning_rate": 7.448693715648863e-05,
      "loss": 1.6252,
      "step": 130816
    },
    {
      "epoch": 0.25820383615004333,
      "grad_norm": 0.10696367919445038,
      "learning_rate": 7.448059693212393e-05,
      "loss": 1.6133,
      "step": 130848
    },
    {
      "epoch": 0.2582669821114398,
      "grad_norm": 0.1008777990937233,
      "learning_rate": 7.447425670775925e-05,
      "loss": 1.6059,
      "step": 130880
    },
    {
      "epoch": 0.25833012807283623,
      "grad_norm": 0.11081548780202866,
      "learning_rate": 7.446791648339456e-05,
      "loss": 1.6105,
      "step": 130912
    },
    {
      "epoch": 0.2583932740342327,
      "grad_norm": 0.11868808418512344,
      "learning_rate": 7.446157625902988e-05,
      "loss": 1.5982,
      "step": 130944
    },
    {
      "epoch": 0.2584564199956291,
      "grad_norm": 0.10667791962623596,
      "learning_rate": 7.445523603466518e-05,
      "loss": 1.6054,
      "step": 130976
    },
    {
      "epoch": 0.2585195659570256,
      "grad_norm": 0.1071515679359436,
      "learning_rate": 7.44488958103005e-05,
      "loss": 1.627,
      "step": 131008
    },
    {
      "epoch": 0.25858271191842197,
      "grad_norm": 0.10900339484214783,
      "learning_rate": 7.444255558593581e-05,
      "loss": 1.6205,
      "step": 131040
    },
    {
      "epoch": 0.2586458578798184,
      "grad_norm": 0.10290917754173279,
      "learning_rate": 7.443621536157111e-05,
      "loss": 1.6222,
      "step": 131072
    },
    {
      "epoch": 0.25870900384121487,
      "grad_norm": 0.1087384894490242,
      "learning_rate": 7.442987513720642e-05,
      "loss": 1.61,
      "step": 131104
    },
    {
      "epoch": 0.2587721498026113,
      "grad_norm": 0.11124587059020996,
      "learning_rate": 7.442353491284172e-05,
      "loss": 1.6075,
      "step": 131136
    },
    {
      "epoch": 0.25883529576400777,
      "grad_norm": 0.11455298215150833,
      "learning_rate": 7.441719468847704e-05,
      "loss": 1.6155,
      "step": 131168
    },
    {
      "epoch": 0.2588984417254042,
      "grad_norm": 0.10650830715894699,
      "learning_rate": 7.441085446411235e-05,
      "loss": 1.5983,
      "step": 131200
    },
    {
      "epoch": 0.2589615876868006,
      "grad_norm": 0.11324498057365417,
      "learning_rate": 7.440451423974766e-05,
      "loss": 1.6035,
      "step": 131232
    },
    {
      "epoch": 0.25902473364819706,
      "grad_norm": 0.10617034137248993,
      "learning_rate": 7.439817401538297e-05,
      "loss": 1.6124,
      "step": 131264
    },
    {
      "epoch": 0.2590878796095935,
      "grad_norm": 0.11221248656511307,
      "learning_rate": 7.439183379101828e-05,
      "loss": 1.6133,
      "step": 131296
    },
    {
      "epoch": 0.25915102557098996,
      "grad_norm": 0.11007510870695114,
      "learning_rate": 7.43854935666536e-05,
      "loss": 1.6133,
      "step": 131328
    },
    {
      "epoch": 0.2592141715323864,
      "grad_norm": 0.10629601776599884,
      "learning_rate": 7.43791533422889e-05,
      "loss": 1.6237,
      "step": 131360
    },
    {
      "epoch": 0.25927731749378286,
      "grad_norm": 0.11356817185878754,
      "learning_rate": 7.437281311792421e-05,
      "loss": 1.593,
      "step": 131392
    },
    {
      "epoch": 0.25934046345517925,
      "grad_norm": 0.10988583415746689,
      "learning_rate": 7.436647289355953e-05,
      "loss": 1.6087,
      "step": 131424
    },
    {
      "epoch": 0.2594036094165757,
      "grad_norm": 0.10281365364789963,
      "learning_rate": 7.436013266919484e-05,
      "loss": 1.6085,
      "step": 131456
    },
    {
      "epoch": 0.25946675537797215,
      "grad_norm": 0.1106758564710617,
      "learning_rate": 7.435379244483014e-05,
      "loss": 1.6023,
      "step": 131488
    },
    {
      "epoch": 0.2595299013393686,
      "grad_norm": 0.1066455990076065,
      "learning_rate": 7.434745222046545e-05,
      "loss": 1.6162,
      "step": 131520
    },
    {
      "epoch": 0.25959304730076505,
      "grad_norm": 0.11346118897199631,
      "learning_rate": 7.434111199610076e-05,
      "loss": 1.6166,
      "step": 131552
    },
    {
      "epoch": 0.2596561932621615,
      "grad_norm": 0.10770457983016968,
      "learning_rate": 7.433477177173607e-05,
      "loss": 1.6093,
      "step": 131584
    },
    {
      "epoch": 0.25971933922355794,
      "grad_norm": 0.1185644268989563,
      "learning_rate": 7.432843154737139e-05,
      "loss": 1.609,
      "step": 131616
    },
    {
      "epoch": 0.25978248518495434,
      "grad_norm": 0.11142583191394806,
      "learning_rate": 7.432209132300669e-05,
      "loss": 1.6096,
      "step": 131648
    },
    {
      "epoch": 0.2598456311463508,
      "grad_norm": 0.11487526446580887,
      "learning_rate": 7.4315751098642e-05,
      "loss": 1.6182,
      "step": 131680
    },
    {
      "epoch": 0.25990877710774724,
      "grad_norm": 0.11717316508293152,
      "learning_rate": 7.430941087427732e-05,
      "loss": 1.6111,
      "step": 131712
    },
    {
      "epoch": 0.2599719230691437,
      "grad_norm": 0.11668983846902847,
      "learning_rate": 7.430307064991263e-05,
      "loss": 1.6116,
      "step": 131744
    },
    {
      "epoch": 0.26003506903054013,
      "grad_norm": 0.12166266143321991,
      "learning_rate": 7.429673042554793e-05,
      "loss": 1.6101,
      "step": 131776
    },
    {
      "epoch": 0.2600982149919366,
      "grad_norm": 0.10517725348472595,
      "learning_rate": 7.429039020118325e-05,
      "loss": 1.6167,
      "step": 131808
    },
    {
      "epoch": 0.260161360953333,
      "grad_norm": 0.11283927410840988,
      "learning_rate": 7.428404997681856e-05,
      "loss": 1.6091,
      "step": 131840
    },
    {
      "epoch": 0.2602245069147294,
      "grad_norm": 0.10576330870389938,
      "learning_rate": 7.427770975245388e-05,
      "loss": 1.6239,
      "step": 131872
    },
    {
      "epoch": 0.2602876528761259,
      "grad_norm": 0.10896219313144684,
      "learning_rate": 7.427136952808918e-05,
      "loss": 1.6061,
      "step": 131904
    },
    {
      "epoch": 0.2603507988375223,
      "grad_norm": 0.10660555958747864,
      "learning_rate": 7.426502930372448e-05,
      "loss": 1.6274,
      "step": 131936
    },
    {
      "epoch": 0.2604139447989188,
      "grad_norm": 0.11188357323408127,
      "learning_rate": 7.42586890793598e-05,
      "loss": 1.6155,
      "step": 131968
    },
    {
      "epoch": 0.2604770907603152,
      "grad_norm": 0.10984139889478683,
      "learning_rate": 7.425234885499511e-05,
      "loss": 1.6091,
      "step": 132000
    },
    {
      "epoch": 0.2605402367217116,
      "grad_norm": 0.10720211267471313,
      "learning_rate": 7.424600863063041e-05,
      "loss": 1.6065,
      "step": 132032
    },
    {
      "epoch": 0.26060338268310806,
      "grad_norm": 0.11262527108192444,
      "learning_rate": 7.423966840626573e-05,
      "loss": 1.6231,
      "step": 132064
    },
    {
      "epoch": 0.2606665286445045,
      "grad_norm": 0.1069030910730362,
      "learning_rate": 7.423332818190104e-05,
      "loss": 1.6108,
      "step": 132096
    },
    {
      "epoch": 0.26072967460590096,
      "grad_norm": 0.10583042353391647,
      "learning_rate": 7.422698795753635e-05,
      "loss": 1.6128,
      "step": 132128
    },
    {
      "epoch": 0.2607928205672974,
      "grad_norm": 0.10477813333272934,
      "learning_rate": 7.422064773317167e-05,
      "loss": 1.6224,
      "step": 132160
    },
    {
      "epoch": 0.26085596652869386,
      "grad_norm": 0.10200940817594528,
      "learning_rate": 7.421430750880697e-05,
      "loss": 1.6075,
      "step": 132192
    },
    {
      "epoch": 0.26091911249009025,
      "grad_norm": 0.1078220009803772,
      "learning_rate": 7.420796728444228e-05,
      "loss": 1.618,
      "step": 132224
    },
    {
      "epoch": 0.2609822584514867,
      "grad_norm": 0.10805245488882065,
      "learning_rate": 7.42016270600776e-05,
      "loss": 1.6199,
      "step": 132256
    },
    {
      "epoch": 0.26104540441288315,
      "grad_norm": 0.10207384824752808,
      "learning_rate": 7.419528683571291e-05,
      "loss": 1.6153,
      "step": 132288
    },
    {
      "epoch": 0.2611085503742796,
      "grad_norm": 0.10227125883102417,
      "learning_rate": 7.418894661134821e-05,
      "loss": 1.6254,
      "step": 132320
    },
    {
      "epoch": 0.26117169633567605,
      "grad_norm": 0.10752616822719574,
      "learning_rate": 7.418260638698352e-05,
      "loss": 1.6081,
      "step": 132352
    },
    {
      "epoch": 0.2612348422970725,
      "grad_norm": 0.11380825936794281,
      "learning_rate": 7.417626616261883e-05,
      "loss": 1.5994,
      "step": 132384
    },
    {
      "epoch": 0.2612979882584689,
      "grad_norm": 0.11722514778375626,
      "learning_rate": 7.416992593825414e-05,
      "loss": 1.6214,
      "step": 132416
    },
    {
      "epoch": 0.26136113421986534,
      "grad_norm": 0.09964317083358765,
      "learning_rate": 7.416358571388945e-05,
      "loss": 1.6121,
      "step": 132448
    },
    {
      "epoch": 0.2614242801812618,
      "grad_norm": 0.10399201512336731,
      "learning_rate": 7.415724548952476e-05,
      "loss": 1.5978,
      "step": 132480
    },
    {
      "epoch": 0.26148742614265824,
      "grad_norm": 0.1094483882188797,
      "learning_rate": 7.415090526516007e-05,
      "loss": 1.6134,
      "step": 132512
    },
    {
      "epoch": 0.2615505721040547,
      "grad_norm": 0.11518104374408722,
      "learning_rate": 7.414456504079539e-05,
      "loss": 1.6098,
      "step": 132544
    },
    {
      "epoch": 0.26161371806545114,
      "grad_norm": 0.111690953373909,
      "learning_rate": 7.413822481643069e-05,
      "loss": 1.6153,
      "step": 132576
    },
    {
      "epoch": 0.26167686402684753,
      "grad_norm": 0.11402498185634613,
      "learning_rate": 7.4131884592066e-05,
      "loss": 1.6175,
      "step": 132608
    },
    {
      "epoch": 0.261740009988244,
      "grad_norm": 0.10443051159381866,
      "learning_rate": 7.412554436770132e-05,
      "loss": 1.6043,
      "step": 132640
    },
    {
      "epoch": 0.26180315594964043,
      "grad_norm": 0.10866302996873856,
      "learning_rate": 7.411920414333663e-05,
      "loss": 1.613,
      "step": 132672
    },
    {
      "epoch": 0.2618663019110369,
      "grad_norm": 0.10197242349386215,
      "learning_rate": 7.411286391897194e-05,
      "loss": 1.6144,
      "step": 132704
    },
    {
      "epoch": 0.26192944787243333,
      "grad_norm": 0.11655410379171371,
      "learning_rate": 7.410652369460725e-05,
      "loss": 1.627,
      "step": 132736
    },
    {
      "epoch": 0.2619925938338298,
      "grad_norm": 0.10013627260923386,
      "learning_rate": 7.410018347024255e-05,
      "loss": 1.6099,
      "step": 132768
    },
    {
      "epoch": 0.2620557397952262,
      "grad_norm": 0.10514095425605774,
      "learning_rate": 7.409384324587787e-05,
      "loss": 1.6214,
      "step": 132800
    },
    {
      "epoch": 0.2621188857566226,
      "grad_norm": 0.10498439520597458,
      "learning_rate": 7.408750302151318e-05,
      "loss": 1.6091,
      "step": 132832
    },
    {
      "epoch": 0.26218203171801907,
      "grad_norm": 0.10498408228158951,
      "learning_rate": 7.408116279714848e-05,
      "loss": 1.608,
      "step": 132864
    },
    {
      "epoch": 0.2622451776794155,
      "grad_norm": 0.10708972811698914,
      "learning_rate": 7.40748225727838e-05,
      "loss": 1.6168,
      "step": 132896
    },
    {
      "epoch": 0.26230832364081197,
      "grad_norm": 0.10591749101877213,
      "learning_rate": 7.406848234841911e-05,
      "loss": 1.6106,
      "step": 132928
    },
    {
      "epoch": 0.2623714696022084,
      "grad_norm": 0.11292807012796402,
      "learning_rate": 7.406214212405442e-05,
      "loss": 1.6064,
      "step": 132960
    },
    {
      "epoch": 0.2624346155636048,
      "grad_norm": 0.1074332743883133,
      "learning_rate": 7.405580189968973e-05,
      "loss": 1.6287,
      "step": 132992
    },
    {
      "epoch": 0.26249776152500126,
      "grad_norm": 0.10224854201078415,
      "learning_rate": 7.404946167532504e-05,
      "loss": 1.6019,
      "step": 133024
    },
    {
      "epoch": 0.2625609074863977,
      "grad_norm": 0.11157242208719254,
      "learning_rate": 7.404312145096035e-05,
      "loss": 1.6122,
      "step": 133056
    },
    {
      "epoch": 0.26262405344779416,
      "grad_norm": 0.10429177433252335,
      "learning_rate": 7.403678122659567e-05,
      "loss": 1.6218,
      "step": 133088
    },
    {
      "epoch": 0.2626871994091906,
      "grad_norm": 0.10565786808729172,
      "learning_rate": 7.403044100223097e-05,
      "loss": 1.6105,
      "step": 133120
    },
    {
      "epoch": 0.26275034537058706,
      "grad_norm": 0.1053301990032196,
      "learning_rate": 7.402410077786628e-05,
      "loss": 1.6069,
      "step": 133152
    },
    {
      "epoch": 0.2628134913319835,
      "grad_norm": 0.10900627821683884,
      "learning_rate": 7.401776055350159e-05,
      "loss": 1.6003,
      "step": 133184
    },
    {
      "epoch": 0.2628766372933799,
      "grad_norm": 0.10631921142339706,
      "learning_rate": 7.40114203291369e-05,
      "loss": 1.6104,
      "step": 133216
    },
    {
      "epoch": 0.26293978325477635,
      "grad_norm": 0.10570721328258514,
      "learning_rate": 7.40050801047722e-05,
      "loss": 1.6092,
      "step": 133248
    },
    {
      "epoch": 0.2630029292161728,
      "grad_norm": 0.11719290167093277,
      "learning_rate": 7.399873988040752e-05,
      "loss": 1.6074,
      "step": 133280
    },
    {
      "epoch": 0.26306607517756925,
      "grad_norm": 0.11129261553287506,
      "learning_rate": 7.399239965604283e-05,
      "loss": 1.6005,
      "step": 133312
    },
    {
      "epoch": 0.2631292211389657,
      "grad_norm": 0.10646923631429672,
      "learning_rate": 7.398605943167815e-05,
      "loss": 1.616,
      "step": 133344
    },
    {
      "epoch": 0.26319236710036215,
      "grad_norm": 0.10687491297721863,
      "learning_rate": 7.397971920731345e-05,
      "loss": 1.6168,
      "step": 133376
    },
    {
      "epoch": 0.26325551306175854,
      "grad_norm": 0.1165340393781662,
      "learning_rate": 7.397337898294876e-05,
      "loss": 1.6099,
      "step": 133408
    },
    {
      "epoch": 0.263318659023155,
      "grad_norm": 0.10726451873779297,
      "learning_rate": 7.396703875858408e-05,
      "loss": 1.6035,
      "step": 133440
    },
    {
      "epoch": 0.26338180498455144,
      "grad_norm": 0.10097071528434753,
      "learning_rate": 7.396069853421939e-05,
      "loss": 1.6208,
      "step": 133472
    },
    {
      "epoch": 0.2634449509459479,
      "grad_norm": 0.10605879127979279,
      "learning_rate": 7.39543583098547e-05,
      "loss": 1.6055,
      "step": 133504
    },
    {
      "epoch": 0.26350809690734434,
      "grad_norm": 0.10595313459634781,
      "learning_rate": 7.394801808549e-05,
      "loss": 1.6149,
      "step": 133536
    },
    {
      "epoch": 0.2635712428687408,
      "grad_norm": 0.1006612628698349,
      "learning_rate": 7.394167786112532e-05,
      "loss": 1.6089,
      "step": 133568
    },
    {
      "epoch": 0.2636343888301372,
      "grad_norm": 0.10585619509220123,
      "learning_rate": 7.393533763676062e-05,
      "loss": 1.6101,
      "step": 133600
    },
    {
      "epoch": 0.2636975347915336,
      "grad_norm": 0.10954471677541733,
      "learning_rate": 7.392899741239594e-05,
      "loss": 1.6046,
      "step": 133632
    },
    {
      "epoch": 0.2637606807529301,
      "grad_norm": 0.11476067453622818,
      "learning_rate": 7.392265718803124e-05,
      "loss": 1.6099,
      "step": 133664
    },
    {
      "epoch": 0.2638238267143265,
      "grad_norm": 0.11173353344202042,
      "learning_rate": 7.391631696366655e-05,
      "loss": 1.6041,
      "step": 133696
    },
    {
      "epoch": 0.263886972675723,
      "grad_norm": 0.11101987957954407,
      "learning_rate": 7.390997673930187e-05,
      "loss": 1.6081,
      "step": 133728
    },
    {
      "epoch": 0.2639501186371194,
      "grad_norm": 0.10786333680152893,
      "learning_rate": 7.390363651493718e-05,
      "loss": 1.6137,
      "step": 133760
    },
    {
      "epoch": 0.2640132645985158,
      "grad_norm": 0.10600059479475021,
      "learning_rate": 7.389729629057248e-05,
      "loss": 1.6069,
      "step": 133792
    },
    {
      "epoch": 0.26407641055991227,
      "grad_norm": 0.10257264226675034,
      "learning_rate": 7.38909560662078e-05,
      "loss": 1.6047,
      "step": 133824
    },
    {
      "epoch": 0.2641395565213087,
      "grad_norm": 0.10793018341064453,
      "learning_rate": 7.388461584184311e-05,
      "loss": 1.6041,
      "step": 133856
    },
    {
      "epoch": 0.26420270248270517,
      "grad_norm": 0.1037888303399086,
      "learning_rate": 7.387827561747842e-05,
      "loss": 1.6145,
      "step": 133888
    },
    {
      "epoch": 0.2642658484441016,
      "grad_norm": 0.10750699788331985,
      "learning_rate": 7.387193539311373e-05,
      "loss": 1.603,
      "step": 133920
    },
    {
      "epoch": 0.26432899440549806,
      "grad_norm": 0.10909721255302429,
      "learning_rate": 7.386559516874904e-05,
      "loss": 1.609,
      "step": 133952
    },
    {
      "epoch": 0.26439214036689446,
      "grad_norm": 0.10319609194993973,
      "learning_rate": 7.385925494438436e-05,
      "loss": 1.6222,
      "step": 133984
    },
    {
      "epoch": 0.2644552863282909,
      "grad_norm": 0.10609913617372513,
      "learning_rate": 7.385291472001966e-05,
      "loss": 1.608,
      "step": 134016
    },
    {
      "epoch": 0.26451843228968736,
      "grad_norm": 0.1000935509800911,
      "learning_rate": 7.384657449565496e-05,
      "loss": 1.6163,
      "step": 134048
    },
    {
      "epoch": 0.2645815782510838,
      "grad_norm": 0.10459282994270325,
      "learning_rate": 7.384023427129027e-05,
      "loss": 1.5992,
      "step": 134080
    },
    {
      "epoch": 0.26464472421248025,
      "grad_norm": 0.10995563119649887,
      "learning_rate": 7.383389404692559e-05,
      "loss": 1.6024,
      "step": 134112
    },
    {
      "epoch": 0.2647078701738767,
      "grad_norm": 0.1150684803724289,
      "learning_rate": 7.38275538225609e-05,
      "loss": 1.6127,
      "step": 134144
    },
    {
      "epoch": 0.2647710161352731,
      "grad_norm": 0.1176714152097702,
      "learning_rate": 7.382121359819622e-05,
      "loss": 1.5986,
      "step": 134176
    },
    {
      "epoch": 0.26483416209666955,
      "grad_norm": 0.11197208613157272,
      "learning_rate": 7.381487337383152e-05,
      "loss": 1.6088,
      "step": 134208
    },
    {
      "epoch": 0.264897308058066,
      "grad_norm": 0.11053857952356339,
      "learning_rate": 7.380853314946683e-05,
      "loss": 1.6074,
      "step": 134240
    },
    {
      "epoch": 0.26496045401946244,
      "grad_norm": 0.10849561542272568,
      "learning_rate": 7.380219292510215e-05,
      "loss": 1.6053,
      "step": 134272
    },
    {
      "epoch": 0.2650235999808589,
      "grad_norm": 0.10623572766780853,
      "learning_rate": 7.379585270073746e-05,
      "loss": 1.6164,
      "step": 134304
    },
    {
      "epoch": 0.26508674594225534,
      "grad_norm": 0.11052551865577698,
      "learning_rate": 7.378951247637276e-05,
      "loss": 1.6037,
      "step": 134336
    },
    {
      "epoch": 0.26514989190365174,
      "grad_norm": 0.11220145970582962,
      "learning_rate": 7.378317225200808e-05,
      "loss": 1.6159,
      "step": 134368
    },
    {
      "epoch": 0.2652130378650482,
      "grad_norm": 0.10774993151426315,
      "learning_rate": 7.377683202764338e-05,
      "loss": 1.6097,
      "step": 134400
    },
    {
      "epoch": 0.26527618382644463,
      "grad_norm": 0.1051446795463562,
      "learning_rate": 7.377049180327869e-05,
      "loss": 1.6172,
      "step": 134432
    },
    {
      "epoch": 0.2653393297878411,
      "grad_norm": 0.11249789595603943,
      "learning_rate": 7.376415157891399e-05,
      "loss": 1.6108,
      "step": 134464
    },
    {
      "epoch": 0.26540247574923753,
      "grad_norm": 0.10095071792602539,
      "learning_rate": 7.37578113545493e-05,
      "loss": 1.593,
      "step": 134496
    },
    {
      "epoch": 0.265465621710634,
      "grad_norm": 0.10578959435224533,
      "learning_rate": 7.375147113018462e-05,
      "loss": 1.6181,
      "step": 134528
    },
    {
      "epoch": 0.2655287676720304,
      "grad_norm": 0.10583876073360443,
      "learning_rate": 7.374513090581994e-05,
      "loss": 1.6035,
      "step": 134560
    },
    {
      "epoch": 0.2655919136334268,
      "grad_norm": 0.1076938733458519,
      "learning_rate": 7.373879068145524e-05,
      "loss": 1.6161,
      "step": 134592
    },
    {
      "epoch": 0.2656550595948233,
      "grad_norm": 0.1081429272890091,
      "learning_rate": 7.373245045709055e-05,
      "loss": 1.6117,
      "step": 134624
    },
    {
      "epoch": 0.2657182055562197,
      "grad_norm": 0.10474961996078491,
      "learning_rate": 7.372611023272587e-05,
      "loss": 1.6064,
      "step": 134656
    },
    {
      "epoch": 0.26578135151761617,
      "grad_norm": 0.10079970210790634,
      "learning_rate": 7.371977000836118e-05,
      "loss": 1.61,
      "step": 134688
    },
    {
      "epoch": 0.2658444974790126,
      "grad_norm": 0.10435830056667328,
      "learning_rate": 7.371342978399648e-05,
      "loss": 1.6155,
      "step": 134720
    },
    {
      "epoch": 0.265907643440409,
      "grad_norm": 0.10284986346960068,
      "learning_rate": 7.37070895596318e-05,
      "loss": 1.6114,
      "step": 134752
    },
    {
      "epoch": 0.26597078940180546,
      "grad_norm": 0.10612160712480545,
      "learning_rate": 7.370074933526711e-05,
      "loss": 1.6122,
      "step": 134784
    },
    {
      "epoch": 0.2660339353632019,
      "grad_norm": 0.10711552947759628,
      "learning_rate": 7.369440911090241e-05,
      "loss": 1.6051,
      "step": 134816
    },
    {
      "epoch": 0.26609708132459836,
      "grad_norm": 0.11176897585391998,
      "learning_rate": 7.368806888653773e-05,
      "loss": 1.6114,
      "step": 134848
    },
    {
      "epoch": 0.2661602272859948,
      "grad_norm": 0.10485158860683441,
      "learning_rate": 7.368172866217303e-05,
      "loss": 1.6178,
      "step": 134880
    },
    {
      "epoch": 0.26622337324739126,
      "grad_norm": 0.11469212919473648,
      "learning_rate": 7.367538843780834e-05,
      "loss": 1.6081,
      "step": 134912
    },
    {
      "epoch": 0.2662865192087877,
      "grad_norm": 0.10416357219219208,
      "learning_rate": 7.366904821344366e-05,
      "loss": 1.6269,
      "step": 134944
    },
    {
      "epoch": 0.2663496651701841,
      "grad_norm": 0.11275526881217957,
      "learning_rate": 7.366270798907897e-05,
      "loss": 1.6018,
      "step": 134976
    },
    {
      "epoch": 0.26641281113158055,
      "grad_norm": 0.1164131686091423,
      "learning_rate": 7.365636776471427e-05,
      "loss": 1.5981,
      "step": 135008
    },
    {
      "epoch": 0.266475957092977,
      "grad_norm": 0.10406443476676941,
      "learning_rate": 7.365002754034959e-05,
      "loss": 1.6087,
      "step": 135040
    },
    {
      "epoch": 0.26653910305437345,
      "grad_norm": 0.10632601380348206,
      "learning_rate": 7.36436873159849e-05,
      "loss": 1.6069,
      "step": 135072
    },
    {
      "epoch": 0.2666022490157699,
      "grad_norm": 0.10499055683612823,
      "learning_rate": 7.363734709162022e-05,
      "loss": 1.5964,
      "step": 135104
    },
    {
      "epoch": 0.26666539497716635,
      "grad_norm": 0.11918951570987701,
      "learning_rate": 7.363100686725552e-05,
      "loss": 1.6104,
      "step": 135136
    },
    {
      "epoch": 0.26672854093856274,
      "grad_norm": 0.11259063333272934,
      "learning_rate": 7.362466664289083e-05,
      "loss": 1.6144,
      "step": 135168
    },
    {
      "epoch": 0.2667916868999592,
      "grad_norm": 0.10633876919746399,
      "learning_rate": 7.361832641852615e-05,
      "loss": 1.6119,
      "step": 135200
    },
    {
      "epoch": 0.26685483286135564,
      "grad_norm": 0.1007336750626564,
      "learning_rate": 7.361198619416145e-05,
      "loss": 1.597,
      "step": 135232
    },
    {
      "epoch": 0.2669179788227521,
      "grad_norm": 0.09957956522703171,
      "learning_rate": 7.360564596979676e-05,
      "loss": 1.6002,
      "step": 135264
    },
    {
      "epoch": 0.26698112478414854,
      "grad_norm": 0.1052396297454834,
      "learning_rate": 7.359930574543206e-05,
      "loss": 1.5996,
      "step": 135296
    },
    {
      "epoch": 0.267044270745545,
      "grad_norm": 0.10251233726739883,
      "learning_rate": 7.359296552106738e-05,
      "loss": 1.602,
      "step": 135328
    },
    {
      "epoch": 0.2671074167069414,
      "grad_norm": 0.10895755887031555,
      "learning_rate": 7.358662529670269e-05,
      "loss": 1.6117,
      "step": 135360
    },
    {
      "epoch": 0.26717056266833783,
      "grad_norm": 0.09954611957073212,
      "learning_rate": 7.358028507233799e-05,
      "loss": 1.6074,
      "step": 135392
    },
    {
      "epoch": 0.2672337086297343,
      "grad_norm": 0.11349955946207047,
      "learning_rate": 7.357394484797331e-05,
      "loss": 1.6193,
      "step": 135424
    },
    {
      "epoch": 0.26729685459113073,
      "grad_norm": 0.10204102098941803,
      "learning_rate": 7.356760462360862e-05,
      "loss": 1.6173,
      "step": 135456
    },
    {
      "epoch": 0.2673600005525272,
      "grad_norm": 0.1054556742310524,
      "learning_rate": 7.356126439924394e-05,
      "loss": 1.6119,
      "step": 135488
    },
    {
      "epoch": 0.2674231465139236,
      "grad_norm": 0.10490527749061584,
      "learning_rate": 7.355492417487925e-05,
      "loss": 1.6047,
      "step": 135520
    },
    {
      "epoch": 0.26748629247532,
      "grad_norm": 0.10282700508832932,
      "learning_rate": 7.354858395051455e-05,
      "loss": 1.6068,
      "step": 135552
    },
    {
      "epoch": 0.26754943843671647,
      "grad_norm": 0.10347724705934525,
      "learning_rate": 7.354224372614987e-05,
      "loss": 1.5943,
      "step": 135584
    },
    {
      "epoch": 0.2676125843981129,
      "grad_norm": 0.11144138127565384,
      "learning_rate": 7.353590350178518e-05,
      "loss": 1.6076,
      "step": 135616
    },
    {
      "epoch": 0.26767573035950937,
      "grad_norm": 0.11879058182239532,
      "learning_rate": 7.352956327742048e-05,
      "loss": 1.6028,
      "step": 135648
    },
    {
      "epoch": 0.2677388763209058,
      "grad_norm": 0.10743587464094162,
      "learning_rate": 7.35232230530558e-05,
      "loss": 1.609,
      "step": 135680
    },
    {
      "epoch": 0.26780202228230227,
      "grad_norm": 0.1044100821018219,
      "learning_rate": 7.35168828286911e-05,
      "loss": 1.609,
      "step": 135712
    },
    {
      "epoch": 0.26786516824369866,
      "grad_norm": 0.10664186626672745,
      "learning_rate": 7.351054260432641e-05,
      "loss": 1.6088,
      "step": 135744
    },
    {
      "epoch": 0.2679283142050951,
      "grad_norm": 0.10664991289377213,
      "learning_rate": 7.350420237996173e-05,
      "loss": 1.6039,
      "step": 135776
    },
    {
      "epoch": 0.26799146016649156,
      "grad_norm": 0.10977598279714584,
      "learning_rate": 7.349786215559703e-05,
      "loss": 1.6142,
      "step": 135808
    },
    {
      "epoch": 0.268054606127888,
      "grad_norm": 0.10533903539180756,
      "learning_rate": 7.349152193123234e-05,
      "loss": 1.6086,
      "step": 135840
    },
    {
      "epoch": 0.26811775208928446,
      "grad_norm": 0.10877023637294769,
      "learning_rate": 7.348518170686766e-05,
      "loss": 1.6084,
      "step": 135872
    },
    {
      "epoch": 0.2681808980506809,
      "grad_norm": 0.11798279732465744,
      "learning_rate": 7.347884148250297e-05,
      "loss": 1.6036,
      "step": 135904
    },
    {
      "epoch": 0.2682440440120773,
      "grad_norm": 0.1045432984828949,
      "learning_rate": 7.347250125813827e-05,
      "loss": 1.6115,
      "step": 135936
    },
    {
      "epoch": 0.26830718997347375,
      "grad_norm": 0.115162692964077,
      "learning_rate": 7.346616103377359e-05,
      "loss": 1.6136,
      "step": 135968
    },
    {
      "epoch": 0.2683703359348702,
      "grad_norm": 0.10045327991247177,
      "learning_rate": 7.34598208094089e-05,
      "loss": 1.6187,
      "step": 136000
    },
    {
      "epoch": 0.26843348189626665,
      "grad_norm": 0.10821972787380219,
      "learning_rate": 7.345348058504422e-05,
      "loss": 1.6134,
      "step": 136032
    },
    {
      "epoch": 0.2684966278576631,
      "grad_norm": 0.11383935809135437,
      "learning_rate": 7.344714036067952e-05,
      "loss": 1.6132,
      "step": 136064
    },
    {
      "epoch": 0.26855977381905954,
      "grad_norm": 0.09907317161560059,
      "learning_rate": 7.344080013631482e-05,
      "loss": 1.6038,
      "step": 136096
    },
    {
      "epoch": 0.26862291978045594,
      "grad_norm": 0.10620388388633728,
      "learning_rate": 7.343445991195013e-05,
      "loss": 1.6052,
      "step": 136128
    },
    {
      "epoch": 0.2686860657418524,
      "grad_norm": 0.10281985998153687,
      "learning_rate": 7.342811968758545e-05,
      "loss": 1.6005,
      "step": 136160
    },
    {
      "epoch": 0.26874921170324884,
      "grad_norm": 0.1163906455039978,
      "learning_rate": 7.342177946322076e-05,
      "loss": 1.5965,
      "step": 136192
    },
    {
      "epoch": 0.2688123576646453,
      "grad_norm": 0.10981487482786179,
      "learning_rate": 7.341543923885606e-05,
      "loss": 1.6019,
      "step": 136224
    },
    {
      "epoch": 0.26887550362604173,
      "grad_norm": 0.11343888938426971,
      "learning_rate": 7.340909901449138e-05,
      "loss": 1.6161,
      "step": 136256
    },
    {
      "epoch": 0.2689386495874382,
      "grad_norm": 0.11012143641710281,
      "learning_rate": 7.340275879012669e-05,
      "loss": 1.6021,
      "step": 136288
    },
    {
      "epoch": 0.2690017955488346,
      "grad_norm": 0.12245097756385803,
      "learning_rate": 7.3396418565762e-05,
      "loss": 1.5999,
      "step": 136320
    },
    {
      "epoch": 0.269064941510231,
      "grad_norm": 0.10646633803844452,
      "learning_rate": 7.339007834139731e-05,
      "loss": 1.6145,
      "step": 136352
    },
    {
      "epoch": 0.2691280874716275,
      "grad_norm": 0.10820487141609192,
      "learning_rate": 7.338373811703262e-05,
      "loss": 1.6093,
      "step": 136384
    },
    {
      "epoch": 0.2691912334330239,
      "grad_norm": 0.10841498523950577,
      "learning_rate": 7.337739789266794e-05,
      "loss": 1.6067,
      "step": 136416
    },
    {
      "epoch": 0.2692543793944204,
      "grad_norm": 0.12096355110406876,
      "learning_rate": 7.337105766830325e-05,
      "loss": 1.6078,
      "step": 136448
    },
    {
      "epoch": 0.2693175253558168,
      "grad_norm": 0.11072580516338348,
      "learning_rate": 7.336471744393855e-05,
      "loss": 1.6083,
      "step": 136480
    },
    {
      "epoch": 0.2693806713172132,
      "grad_norm": 0.11263925582170486,
      "learning_rate": 7.335837721957385e-05,
      "loss": 1.6112,
      "step": 136512
    },
    {
      "epoch": 0.26944381727860967,
      "grad_norm": 0.11243830621242523,
      "learning_rate": 7.335203699520917e-05,
      "loss": 1.6184,
      "step": 136544
    },
    {
      "epoch": 0.2695069632400061,
      "grad_norm": 0.10702415555715561,
      "learning_rate": 7.334569677084448e-05,
      "loss": 1.603,
      "step": 136576
    },
    {
      "epoch": 0.26957010920140256,
      "grad_norm": 0.09745298326015472,
      "learning_rate": 7.333935654647978e-05,
      "loss": 1.606,
      "step": 136608
    },
    {
      "epoch": 0.269633255162799,
      "grad_norm": 0.11158228665590286,
      "learning_rate": 7.33330163221151e-05,
      "loss": 1.6055,
      "step": 136640
    },
    {
      "epoch": 0.26969640112419546,
      "grad_norm": 0.11331474781036377,
      "learning_rate": 7.332667609775041e-05,
      "loss": 1.6143,
      "step": 136672
    },
    {
      "epoch": 0.2697595470855919,
      "grad_norm": 0.1052720844745636,
      "learning_rate": 7.332033587338573e-05,
      "loss": 1.6072,
      "step": 136704
    },
    {
      "epoch": 0.2698226930469883,
      "grad_norm": 0.10362829267978668,
      "learning_rate": 7.331399564902103e-05,
      "loss": 1.6147,
      "step": 136736
    },
    {
      "epoch": 0.26988583900838475,
      "grad_norm": 0.1109706461429596,
      "learning_rate": 7.330765542465634e-05,
      "loss": 1.5882,
      "step": 136768
    },
    {
      "epoch": 0.2699489849697812,
      "grad_norm": 0.10616286098957062,
      "learning_rate": 7.330131520029166e-05,
      "loss": 1.6123,
      "step": 136800
    },
    {
      "epoch": 0.27001213093117765,
      "grad_norm": 0.11944980174303055,
      "learning_rate": 7.329497497592697e-05,
      "loss": 1.5988,
      "step": 136832
    },
    {
      "epoch": 0.2700752768925741,
      "grad_norm": 0.11348874121904373,
      "learning_rate": 7.328863475156229e-05,
      "loss": 1.6108,
      "step": 136864
    },
    {
      "epoch": 0.27013842285397055,
      "grad_norm": 0.10996314883232117,
      "learning_rate": 7.328229452719759e-05,
      "loss": 1.6109,
      "step": 136896
    },
    {
      "epoch": 0.27020156881536694,
      "grad_norm": 0.11196240037679672,
      "learning_rate": 7.327595430283289e-05,
      "loss": 1.6144,
      "step": 136928
    },
    {
      "epoch": 0.2702647147767634,
      "grad_norm": 0.10875514894723892,
      "learning_rate": 7.32696140784682e-05,
      "loss": 1.6022,
      "step": 136960
    },
    {
      "epoch": 0.27032786073815984,
      "grad_norm": 0.10737422108650208,
      "learning_rate": 7.326327385410352e-05,
      "loss": 1.6134,
      "step": 136992
    },
    {
      "epoch": 0.2703910066995563,
      "grad_norm": 0.1196058839559555,
      "learning_rate": 7.325693362973882e-05,
      "loss": 1.6186,
      "step": 137024
    },
    {
      "epoch": 0.27045415266095274,
      "grad_norm": 0.11066177487373352,
      "learning_rate": 7.325059340537413e-05,
      "loss": 1.6142,
      "step": 137056
    },
    {
      "epoch": 0.2705172986223492,
      "grad_norm": 0.10822766274213791,
      "learning_rate": 7.324425318100945e-05,
      "loss": 1.6053,
      "step": 137088
    },
    {
      "epoch": 0.2705804445837456,
      "grad_norm": 0.10514150559902191,
      "learning_rate": 7.323791295664476e-05,
      "loss": 1.6163,
      "step": 137120
    },
    {
      "epoch": 0.27064359054514203,
      "grad_norm": 0.10512989014387131,
      "learning_rate": 7.323157273228006e-05,
      "loss": 1.6124,
      "step": 137152
    },
    {
      "epoch": 0.2707067365065385,
      "grad_norm": 0.10496227443218231,
      "learning_rate": 7.322523250791538e-05,
      "loss": 1.5942,
      "step": 137184
    },
    {
      "epoch": 0.27076988246793493,
      "grad_norm": 0.11064368486404419,
      "learning_rate": 7.321889228355069e-05,
      "loss": 1.6046,
      "step": 137216
    },
    {
      "epoch": 0.2708330284293314,
      "grad_norm": 0.11531023681163788,
      "learning_rate": 7.3212552059186e-05,
      "loss": 1.6056,
      "step": 137248
    },
    {
      "epoch": 0.27089617439072783,
      "grad_norm": 0.11127570271492004,
      "learning_rate": 7.320621183482131e-05,
      "loss": 1.6244,
      "step": 137280
    },
    {
      "epoch": 0.2709593203521242,
      "grad_norm": 0.10632797330617905,
      "learning_rate": 7.319987161045662e-05,
      "loss": 1.5886,
      "step": 137312
    },
    {
      "epoch": 0.27102246631352067,
      "grad_norm": 0.10544941574335098,
      "learning_rate": 7.319353138609192e-05,
      "loss": 1.6034,
      "step": 137344
    },
    {
      "epoch": 0.2710856122749171,
      "grad_norm": 0.11425535380840302,
      "learning_rate": 7.318719116172724e-05,
      "loss": 1.6073,
      "step": 137376
    },
    {
      "epoch": 0.27114875823631357,
      "grad_norm": 0.10999225080013275,
      "learning_rate": 7.318085093736254e-05,
      "loss": 1.5998,
      "step": 137408
    },
    {
      "epoch": 0.27121190419771,
      "grad_norm": 0.10617484897375107,
      "learning_rate": 7.317451071299785e-05,
      "loss": 1.6185,
      "step": 137440
    },
    {
      "epoch": 0.27127505015910647,
      "grad_norm": 0.10912901908159256,
      "learning_rate": 7.316817048863317e-05,
      "loss": 1.6149,
      "step": 137472
    },
    {
      "epoch": 0.27133819612050286,
      "grad_norm": 0.11024027317762375,
      "learning_rate": 7.316183026426848e-05,
      "loss": 1.602,
      "step": 137504
    },
    {
      "epoch": 0.2714013420818993,
      "grad_norm": 0.11293243616819382,
      "learning_rate": 7.31554900399038e-05,
      "loss": 1.6069,
      "step": 137536
    },
    {
      "epoch": 0.27146448804329576,
      "grad_norm": 0.11466856300830841,
      "learning_rate": 7.31491498155391e-05,
      "loss": 1.6005,
      "step": 137568
    },
    {
      "epoch": 0.2715276340046922,
      "grad_norm": 0.10746665298938751,
      "learning_rate": 7.314280959117441e-05,
      "loss": 1.6253,
      "step": 137600
    },
    {
      "epoch": 0.27159077996608866,
      "grad_norm": 0.1084543839097023,
      "learning_rate": 7.313646936680973e-05,
      "loss": 1.6148,
      "step": 137632
    },
    {
      "epoch": 0.2716539259274851,
      "grad_norm": 0.10356723517179489,
      "learning_rate": 7.313012914244504e-05,
      "loss": 1.6069,
      "step": 137664
    },
    {
      "epoch": 0.2717170718888815,
      "grad_norm": 0.1094108298420906,
      "learning_rate": 7.312378891808034e-05,
      "loss": 1.6046,
      "step": 137696
    },
    {
      "epoch": 0.27178021785027795,
      "grad_norm": 0.10308845341205597,
      "learning_rate": 7.311744869371566e-05,
      "loss": 1.5995,
      "step": 137728
    },
    {
      "epoch": 0.2718433638116744,
      "grad_norm": 0.10088953375816345,
      "learning_rate": 7.311110846935096e-05,
      "loss": 1.6166,
      "step": 137760
    },
    {
      "epoch": 0.27190650977307085,
      "grad_norm": 0.10615668445825577,
      "learning_rate": 7.310476824498627e-05,
      "loss": 1.6138,
      "step": 137792
    },
    {
      "epoch": 0.2719696557344673,
      "grad_norm": 0.11264284700155258,
      "learning_rate": 7.309842802062157e-05,
      "loss": 1.6274,
      "step": 137824
    },
    {
      "epoch": 0.27203280169586375,
      "grad_norm": 0.1094990149140358,
      "learning_rate": 7.309208779625689e-05,
      "loss": 1.6043,
      "step": 137856
    },
    {
      "epoch": 0.27209594765726014,
      "grad_norm": 0.10782024264335632,
      "learning_rate": 7.30857475718922e-05,
      "loss": 1.6034,
      "step": 137888
    },
    {
      "epoch": 0.2721590936186566,
      "grad_norm": 0.10610847920179367,
      "learning_rate": 7.307940734752752e-05,
      "loss": 1.5993,
      "step": 137920
    },
    {
      "epoch": 0.27222223958005304,
      "grad_norm": 0.1120687946677208,
      "learning_rate": 7.307306712316282e-05,
      "loss": 1.612,
      "step": 137952
    },
    {
      "epoch": 0.2722853855414495,
      "grad_norm": 0.10741465538740158,
      "learning_rate": 7.306672689879813e-05,
      "loss": 1.6156,
      "step": 137984
    },
    {
      "epoch": 0.27234853150284594,
      "grad_norm": 0.101494200527668,
      "learning_rate": 7.306038667443345e-05,
      "loss": 1.6064,
      "step": 138016
    },
    {
      "epoch": 0.2724116774642424,
      "grad_norm": 0.10113751888275146,
      "learning_rate": 7.305404645006876e-05,
      "loss": 1.6121,
      "step": 138048
    },
    {
      "epoch": 0.2724748234256388,
      "grad_norm": 0.12258037179708481,
      "learning_rate": 7.304770622570406e-05,
      "loss": 1.6035,
      "step": 138080
    },
    {
      "epoch": 0.27253796938703523,
      "grad_norm": 0.10584273189306259,
      "learning_rate": 7.304136600133938e-05,
      "loss": 1.6019,
      "step": 138112
    },
    {
      "epoch": 0.2726011153484317,
      "grad_norm": 0.10857692360877991,
      "learning_rate": 7.303502577697469e-05,
      "loss": 1.5972,
      "step": 138144
    },
    {
      "epoch": 0.2726642613098281,
      "grad_norm": 0.10216984152793884,
      "learning_rate": 7.302868555261e-05,
      "loss": 1.6103,
      "step": 138176
    },
    {
      "epoch": 0.2727274072712246,
      "grad_norm": 0.1066039502620697,
      "learning_rate": 7.302234532824531e-05,
      "loss": 1.5957,
      "step": 138208
    },
    {
      "epoch": 0.272790553232621,
      "grad_norm": 0.10498004406690598,
      "learning_rate": 7.301600510388061e-05,
      "loss": 1.5859,
      "step": 138240
    },
    {
      "epoch": 0.2728536991940175,
      "grad_norm": 0.10656807571649551,
      "learning_rate": 7.300966487951592e-05,
      "loss": 1.6159,
      "step": 138272
    },
    {
      "epoch": 0.27291684515541387,
      "grad_norm": 0.11368874460458755,
      "learning_rate": 7.300332465515124e-05,
      "loss": 1.6122,
      "step": 138304
    },
    {
      "epoch": 0.2729799911168103,
      "grad_norm": 0.10525816679000854,
      "learning_rate": 7.299698443078655e-05,
      "loss": 1.6103,
      "step": 138336
    },
    {
      "epoch": 0.27304313707820677,
      "grad_norm": 0.11544210463762283,
      "learning_rate": 7.299064420642185e-05,
      "loss": 1.6156,
      "step": 138368
    },
    {
      "epoch": 0.2731062830396032,
      "grad_norm": 0.10521316528320312,
      "learning_rate": 7.298430398205717e-05,
      "loss": 1.5968,
      "step": 138400
    },
    {
      "epoch": 0.27316942900099966,
      "grad_norm": 0.10763830691576004,
      "learning_rate": 7.297796375769248e-05,
      "loss": 1.6112,
      "step": 138432
    },
    {
      "epoch": 0.2732325749623961,
      "grad_norm": 0.1082969531416893,
      "learning_rate": 7.29716235333278e-05,
      "loss": 1.6161,
      "step": 138464
    },
    {
      "epoch": 0.2732957209237925,
      "grad_norm": 0.11077774316072464,
      "learning_rate": 7.29652833089631e-05,
      "loss": 1.632,
      "step": 138496
    },
    {
      "epoch": 0.27335886688518896,
      "grad_norm": 0.11002691835165024,
      "learning_rate": 7.295894308459841e-05,
      "loss": 1.6073,
      "step": 138528
    },
    {
      "epoch": 0.2734220128465854,
      "grad_norm": 0.1076987013220787,
      "learning_rate": 7.295260286023371e-05,
      "loss": 1.6163,
      "step": 138560
    },
    {
      "epoch": 0.27348515880798185,
      "grad_norm": 0.11410069465637207,
      "learning_rate": 7.294626263586903e-05,
      "loss": 1.6108,
      "step": 138592
    },
    {
      "epoch": 0.2735483047693783,
      "grad_norm": 0.1137956976890564,
      "learning_rate": 7.293992241150433e-05,
      "loss": 1.6132,
      "step": 138624
    },
    {
      "epoch": 0.27361145073077475,
      "grad_norm": 0.10526256263256073,
      "learning_rate": 7.293358218713964e-05,
      "loss": 1.6099,
      "step": 138656
    },
    {
      "epoch": 0.27367459669217115,
      "grad_norm": 0.10719609260559082,
      "learning_rate": 7.292724196277496e-05,
      "loss": 1.6187,
      "step": 138688
    },
    {
      "epoch": 0.2737377426535676,
      "grad_norm": 0.10259390622377396,
      "learning_rate": 7.292090173841027e-05,
      "loss": 1.6231,
      "step": 138720
    },
    {
      "epoch": 0.27380088861496404,
      "grad_norm": 0.10427749902009964,
      "learning_rate": 7.291456151404559e-05,
      "loss": 1.5988,
      "step": 138752
    },
    {
      "epoch": 0.2738640345763605,
      "grad_norm": 0.10385967791080475,
      "learning_rate": 7.290822128968089e-05,
      "loss": 1.6039,
      "step": 138784
    },
    {
      "epoch": 0.27392718053775694,
      "grad_norm": 0.10709089785814285,
      "learning_rate": 7.29018810653162e-05,
      "loss": 1.6031,
      "step": 138816
    },
    {
      "epoch": 0.2739903264991534,
      "grad_norm": 0.10972519218921661,
      "learning_rate": 7.289554084095152e-05,
      "loss": 1.6081,
      "step": 138848
    },
    {
      "epoch": 0.2740534724605498,
      "grad_norm": 0.1053602397441864,
      "learning_rate": 7.288920061658683e-05,
      "loss": 1.6014,
      "step": 138880
    },
    {
      "epoch": 0.27411661842194623,
      "grad_norm": 0.1116967722773552,
      "learning_rate": 7.288286039222213e-05,
      "loss": 1.602,
      "step": 138912
    },
    {
      "epoch": 0.2741797643833427,
      "grad_norm": 0.11738558113574982,
      "learning_rate": 7.287652016785745e-05,
      "loss": 1.5972,
      "step": 138944
    },
    {
      "epoch": 0.27424291034473913,
      "grad_norm": 0.10538122057914734,
      "learning_rate": 7.287017994349275e-05,
      "loss": 1.6196,
      "step": 138976
    },
    {
      "epoch": 0.2743060563061356,
      "grad_norm": 0.10250070691108704,
      "learning_rate": 7.286383971912806e-05,
      "loss": 1.6146,
      "step": 139008
    },
    {
      "epoch": 0.27436920226753203,
      "grad_norm": 0.10746382921934128,
      "learning_rate": 7.285749949476336e-05,
      "loss": 1.61,
      "step": 139040
    },
    {
      "epoch": 0.2744323482289284,
      "grad_norm": 0.1076064258813858,
      "learning_rate": 7.285115927039868e-05,
      "loss": 1.6043,
      "step": 139072
    },
    {
      "epoch": 0.2744954941903249,
      "grad_norm": 0.10753529518842697,
      "learning_rate": 7.2844819046034e-05,
      "loss": 1.6063,
      "step": 139104
    },
    {
      "epoch": 0.2745586401517213,
      "grad_norm": 0.11315683275461197,
      "learning_rate": 7.283847882166931e-05,
      "loss": 1.6136,
      "step": 139136
    },
    {
      "epoch": 0.27462178611311777,
      "grad_norm": 0.11320410668849945,
      "learning_rate": 7.283213859730461e-05,
      "loss": 1.6001,
      "step": 139168
    },
    {
      "epoch": 0.2746849320745142,
      "grad_norm": 0.10548532009124756,
      "learning_rate": 7.282579837293992e-05,
      "loss": 1.608,
      "step": 139200
    },
    {
      "epoch": 0.27474807803591067,
      "grad_norm": 0.09956170618534088,
      "learning_rate": 7.281945814857524e-05,
      "loss": 1.6043,
      "step": 139232
    },
    {
      "epoch": 0.27481122399730706,
      "grad_norm": 0.10983692854642868,
      "learning_rate": 7.281311792421055e-05,
      "loss": 1.6079,
      "step": 139264
    },
    {
      "epoch": 0.2748743699587035,
      "grad_norm": 0.10724955797195435,
      "learning_rate": 7.280677769984585e-05,
      "loss": 1.6105,
      "step": 139296
    },
    {
      "epoch": 0.27493751592009996,
      "grad_norm": 0.10298347473144531,
      "learning_rate": 7.280043747548117e-05,
      "loss": 1.592,
      "step": 139328
    },
    {
      "epoch": 0.2750006618814964,
      "grad_norm": 0.11434242874383926,
      "learning_rate": 7.279409725111648e-05,
      "loss": 1.5987,
      "step": 139360
    },
    {
      "epoch": 0.27506380784289286,
      "grad_norm": 0.10061415284872055,
      "learning_rate": 7.278775702675178e-05,
      "loss": 1.6016,
      "step": 139392
    },
    {
      "epoch": 0.2751269538042893,
      "grad_norm": 0.11188656836748123,
      "learning_rate": 7.27814168023871e-05,
      "loss": 1.6159,
      "step": 139424
    },
    {
      "epoch": 0.2751900997656857,
      "grad_norm": 0.11809300631284714,
      "learning_rate": 7.27750765780224e-05,
      "loss": 1.6029,
      "step": 139456
    },
    {
      "epoch": 0.27525324572708215,
      "grad_norm": 0.11942886561155319,
      "learning_rate": 7.276873635365771e-05,
      "loss": 1.6032,
      "step": 139488
    },
    {
      "epoch": 0.2753163916884786,
      "grad_norm": 0.11183831840753555,
      "learning_rate": 7.276239612929303e-05,
      "loss": 1.6125,
      "step": 139520
    },
    {
      "epoch": 0.27537953764987505,
      "grad_norm": 0.11295828968286514,
      "learning_rate": 7.275605590492834e-05,
      "loss": 1.6269,
      "step": 139552
    },
    {
      "epoch": 0.2754426836112715,
      "grad_norm": 0.10338342934846878,
      "learning_rate": 7.274971568056364e-05,
      "loss": 1.6063,
      "step": 139584
    },
    {
      "epoch": 0.27550582957266795,
      "grad_norm": 0.10053700953722,
      "learning_rate": 7.274337545619896e-05,
      "loss": 1.6102,
      "step": 139616
    },
    {
      "epoch": 0.27556897553406434,
      "grad_norm": 0.10031751543283463,
      "learning_rate": 7.273703523183427e-05,
      "loss": 1.6153,
      "step": 139648
    },
    {
      "epoch": 0.2756321214954608,
      "grad_norm": 0.10287375003099442,
      "learning_rate": 7.273069500746959e-05,
      "loss": 1.6033,
      "step": 139680
    },
    {
      "epoch": 0.27569526745685724,
      "grad_norm": 0.11539004743099213,
      "learning_rate": 7.272435478310489e-05,
      "loss": 1.5963,
      "step": 139712
    },
    {
      "epoch": 0.2757584134182537,
      "grad_norm": 0.10440555214881897,
      "learning_rate": 7.27180145587402e-05,
      "loss": 1.6039,
      "step": 139744
    },
    {
      "epoch": 0.27582155937965014,
      "grad_norm": 0.11029437929391861,
      "learning_rate": 7.271167433437552e-05,
      "loss": 1.6046,
      "step": 139776
    },
    {
      "epoch": 0.2758847053410466,
      "grad_norm": 0.10828778147697449,
      "learning_rate": 7.270533411001082e-05,
      "loss": 1.5994,
      "step": 139808
    },
    {
      "epoch": 0.275947851302443,
      "grad_norm": 0.11068389564752579,
      "learning_rate": 7.269899388564613e-05,
      "loss": 1.6034,
      "step": 139840
    },
    {
      "epoch": 0.27601099726383943,
      "grad_norm": 0.10843534767627716,
      "learning_rate": 7.269265366128143e-05,
      "loss": 1.6079,
      "step": 139872
    },
    {
      "epoch": 0.2760741432252359,
      "grad_norm": 0.10967250168323517,
      "learning_rate": 7.268631343691675e-05,
      "loss": 1.6179,
      "step": 139904
    },
    {
      "epoch": 0.27613728918663233,
      "grad_norm": 0.10577895492315292,
      "learning_rate": 7.267997321255206e-05,
      "loss": 1.6053,
      "step": 139936
    },
    {
      "epoch": 0.2762004351480288,
      "grad_norm": 0.10573165118694305,
      "learning_rate": 7.267363298818737e-05,
      "loss": 1.5949,
      "step": 139968
    },
    {
      "epoch": 0.2762635811094252,
      "grad_norm": 0.10600164532661438,
      "learning_rate": 7.266729276382268e-05,
      "loss": 1.6017,
      "step": 140000
    },
    {
      "epoch": 0.2763267270708217,
      "grad_norm": 0.11545202881097794,
      "learning_rate": 7.2660952539458e-05,
      "loss": 1.6059,
      "step": 140032
    },
    {
      "epoch": 0.27638987303221807,
      "grad_norm": 0.10129143297672272,
      "learning_rate": 7.265461231509331e-05,
      "loss": 1.6038,
      "step": 140064
    },
    {
      "epoch": 0.2764530189936145,
      "grad_norm": 0.11578719317913055,
      "learning_rate": 7.264827209072862e-05,
      "loss": 1.6169,
      "step": 140096
    },
    {
      "epoch": 0.27651616495501097,
      "grad_norm": 0.10591530054807663,
      "learning_rate": 7.264193186636392e-05,
      "loss": 1.6087,
      "step": 140128
    },
    {
      "epoch": 0.2765793109164074,
      "grad_norm": 0.10603789985179901,
      "learning_rate": 7.263559164199924e-05,
      "loss": 1.6054,
      "step": 140160
    },
    {
      "epoch": 0.27664245687780387,
      "grad_norm": 0.10827932506799698,
      "learning_rate": 7.262925141763455e-05,
      "loss": 1.6216,
      "step": 140192
    },
    {
      "epoch": 0.2767056028392003,
      "grad_norm": 0.11001746356487274,
      "learning_rate": 7.262291119326985e-05,
      "loss": 1.6215,
      "step": 140224
    },
    {
      "epoch": 0.2767687488005967,
      "grad_norm": 0.10563153028488159,
      "learning_rate": 7.261657096890516e-05,
      "loss": 1.6043,
      "step": 140256
    },
    {
      "epoch": 0.27683189476199316,
      "grad_norm": 0.10176856070756912,
      "learning_rate": 7.261023074454047e-05,
      "loss": 1.6057,
      "step": 140288
    },
    {
      "epoch": 0.2768950407233896,
      "grad_norm": 0.10194772481918335,
      "learning_rate": 7.260389052017578e-05,
      "loss": 1.6102,
      "step": 140320
    },
    {
      "epoch": 0.27695818668478606,
      "grad_norm": 0.10760031640529633,
      "learning_rate": 7.25975502958111e-05,
      "loss": 1.5999,
      "step": 140352
    },
    {
      "epoch": 0.2770213326461825,
      "grad_norm": 0.11413019150495529,
      "learning_rate": 7.25912100714464e-05,
      "loss": 1.6143,
      "step": 140384
    },
    {
      "epoch": 0.27708447860757895,
      "grad_norm": 0.11079232394695282,
      "learning_rate": 7.258486984708171e-05,
      "loss": 1.5972,
      "step": 140416
    },
    {
      "epoch": 0.27714762456897535,
      "grad_norm": 0.11735320836305618,
      "learning_rate": 7.257852962271703e-05,
      "loss": 1.6214,
      "step": 140448
    },
    {
      "epoch": 0.2772107705303718,
      "grad_norm": 0.11206310242414474,
      "learning_rate": 7.257218939835234e-05,
      "loss": 1.5947,
      "step": 140480
    },
    {
      "epoch": 0.27727391649176825,
      "grad_norm": 0.10074267536401749,
      "learning_rate": 7.256584917398764e-05,
      "loss": 1.6058,
      "step": 140512
    },
    {
      "epoch": 0.2773370624531647,
      "grad_norm": 0.10533620417118073,
      "learning_rate": 7.255950894962296e-05,
      "loss": 1.614,
      "step": 140544
    },
    {
      "epoch": 0.27740020841456114,
      "grad_norm": 0.10960860550403595,
      "learning_rate": 7.255316872525827e-05,
      "loss": 1.6131,
      "step": 140576
    },
    {
      "epoch": 0.2774633543759576,
      "grad_norm": 0.10648467391729355,
      "learning_rate": 7.254682850089359e-05,
      "loss": 1.614,
      "step": 140608
    },
    {
      "epoch": 0.277526500337354,
      "grad_norm": 0.1147274300456047,
      "learning_rate": 7.254048827652889e-05,
      "loss": 1.6094,
      "step": 140640
    },
    {
      "epoch": 0.27758964629875044,
      "grad_norm": 0.11199229210615158,
      "learning_rate": 7.253414805216419e-05,
      "loss": 1.6028,
      "step": 140672
    },
    {
      "epoch": 0.2776527922601469,
      "grad_norm": 0.09860625863075256,
      "learning_rate": 7.25278078277995e-05,
      "loss": 1.6023,
      "step": 140704
    },
    {
      "epoch": 0.27771593822154333,
      "grad_norm": 0.10627433657646179,
      "learning_rate": 7.252146760343482e-05,
      "loss": 1.5919,
      "step": 140736
    },
    {
      "epoch": 0.2777790841829398,
      "grad_norm": 0.11265649646520615,
      "learning_rate": 7.251512737907013e-05,
      "loss": 1.6077,
      "step": 140768
    },
    {
      "epoch": 0.27784223014433623,
      "grad_norm": 0.11177830398082733,
      "learning_rate": 7.250878715470544e-05,
      "loss": 1.5975,
      "step": 140800
    },
    {
      "epoch": 0.2779053761057326,
      "grad_norm": 0.10543101280927658,
      "learning_rate": 7.250244693034075e-05,
      "loss": 1.6101,
      "step": 140832
    },
    {
      "epoch": 0.2779685220671291,
      "grad_norm": 0.1085662916302681,
      "learning_rate": 7.249610670597606e-05,
      "loss": 1.602,
      "step": 140864
    },
    {
      "epoch": 0.2780316680285255,
      "grad_norm": 0.10149335116147995,
      "learning_rate": 7.248976648161138e-05,
      "loss": 1.5982,
      "step": 140896
    },
    {
      "epoch": 0.278094813989922,
      "grad_norm": 0.10375605523586273,
      "learning_rate": 7.248342625724668e-05,
      "loss": 1.6002,
      "step": 140928
    },
    {
      "epoch": 0.2781579599513184,
      "grad_norm": 0.10206574946641922,
      "learning_rate": 7.2477086032882e-05,
      "loss": 1.587,
      "step": 140960
    },
    {
      "epoch": 0.27822110591271487,
      "grad_norm": 0.10288719832897186,
      "learning_rate": 7.247074580851731e-05,
      "loss": 1.5879,
      "step": 140992
    },
    {
      "epoch": 0.27828425187411127,
      "grad_norm": 0.10083068162202835,
      "learning_rate": 7.246440558415262e-05,
      "loss": 1.616,
      "step": 141024
    },
    {
      "epoch": 0.2783473978355077,
      "grad_norm": 0.10225941240787506,
      "learning_rate": 7.245806535978792e-05,
      "loss": 1.6111,
      "step": 141056
    },
    {
      "epoch": 0.27841054379690416,
      "grad_norm": 0.1042669340968132,
      "learning_rate": 7.245172513542323e-05,
      "loss": 1.6184,
      "step": 141088
    },
    {
      "epoch": 0.2784736897583006,
      "grad_norm": 0.10268131643533707,
      "learning_rate": 7.244538491105854e-05,
      "loss": 1.6174,
      "step": 141120
    },
    {
      "epoch": 0.27853683571969706,
      "grad_norm": 0.10237900912761688,
      "learning_rate": 7.243904468669385e-05,
      "loss": 1.6018,
      "step": 141152
    },
    {
      "epoch": 0.2785999816810935,
      "grad_norm": 0.1135939434170723,
      "learning_rate": 7.243270446232916e-05,
      "loss": 1.6032,
      "step": 141184
    },
    {
      "epoch": 0.2786631276424899,
      "grad_norm": 0.10880962759256363,
      "learning_rate": 7.242636423796447e-05,
      "loss": 1.594,
      "step": 141216
    },
    {
      "epoch": 0.27872627360388635,
      "grad_norm": 0.10294250398874283,
      "learning_rate": 7.242002401359978e-05,
      "loss": 1.6094,
      "step": 141248
    },
    {
      "epoch": 0.2787894195652828,
      "grad_norm": 0.10812443494796753,
      "learning_rate": 7.24136837892351e-05,
      "loss": 1.5994,
      "step": 141280
    },
    {
      "epoch": 0.27885256552667925,
      "grad_norm": 0.1082243099808693,
      "learning_rate": 7.24073435648704e-05,
      "loss": 1.6011,
      "step": 141312
    },
    {
      "epoch": 0.2789157114880757,
      "grad_norm": 0.10394982248544693,
      "learning_rate": 7.240100334050572e-05,
      "loss": 1.607,
      "step": 141344
    },
    {
      "epoch": 0.27897885744947215,
      "grad_norm": 0.10719755291938782,
      "learning_rate": 7.239466311614103e-05,
      "loss": 1.6069,
      "step": 141376
    },
    {
      "epoch": 0.27904200341086854,
      "grad_norm": 0.11045152693986893,
      "learning_rate": 7.238832289177634e-05,
      "loss": 1.6176,
      "step": 141408
    },
    {
      "epoch": 0.279105149372265,
      "grad_norm": 0.11409204453229904,
      "learning_rate": 7.238198266741165e-05,
      "loss": 1.6184,
      "step": 141440
    },
    {
      "epoch": 0.27916829533366144,
      "grad_norm": 0.11152160912752151,
      "learning_rate": 7.237564244304696e-05,
      "loss": 1.6048,
      "step": 141472
    },
    {
      "epoch": 0.2792314412950579,
      "grad_norm": 0.11334708333015442,
      "learning_rate": 7.236930221868226e-05,
      "loss": 1.6048,
      "step": 141504
    },
    {
      "epoch": 0.27929458725645434,
      "grad_norm": 0.11197617650032043,
      "learning_rate": 7.236296199431758e-05,
      "loss": 1.6035,
      "step": 141536
    },
    {
      "epoch": 0.2793577332178508,
      "grad_norm": 0.11193860322237015,
      "learning_rate": 7.235662176995289e-05,
      "loss": 1.6022,
      "step": 141568
    },
    {
      "epoch": 0.27942087917924724,
      "grad_norm": 0.12336414307355881,
      "learning_rate": 7.235028154558819e-05,
      "loss": 1.6061,
      "step": 141600
    },
    {
      "epoch": 0.27948402514064363,
      "grad_norm": 0.10348925739526749,
      "learning_rate": 7.23439413212235e-05,
      "loss": 1.6045,
      "step": 141632
    },
    {
      "epoch": 0.2795471711020401,
      "grad_norm": 0.10907229036092758,
      "learning_rate": 7.233760109685882e-05,
      "loss": 1.6184,
      "step": 141664
    },
    {
      "epoch": 0.27961031706343653,
      "grad_norm": 0.11674027144908905,
      "learning_rate": 7.233126087249413e-05,
      "loss": 1.6027,
      "step": 141696
    },
    {
      "epoch": 0.279673463024833,
      "grad_norm": 0.10845894366502762,
      "learning_rate": 7.232492064812944e-05,
      "loss": 1.6114,
      "step": 141728
    },
    {
      "epoch": 0.27973660898622943,
      "grad_norm": 0.1105903759598732,
      "learning_rate": 7.231858042376475e-05,
      "loss": 1.6059,
      "step": 141760
    },
    {
      "epoch": 0.2797997549476259,
      "grad_norm": 0.11820506304502487,
      "learning_rate": 7.231224019940006e-05,
      "loss": 1.6071,
      "step": 141792
    },
    {
      "epoch": 0.27986290090902227,
      "grad_norm": 0.11597861349582672,
      "learning_rate": 7.230589997503538e-05,
      "loss": 1.6018,
      "step": 141824
    },
    {
      "epoch": 0.2799260468704187,
      "grad_norm": 0.10703887045383453,
      "learning_rate": 7.229955975067068e-05,
      "loss": 1.6101,
      "step": 141856
    },
    {
      "epoch": 0.27998919283181517,
      "grad_norm": 0.10526274144649506,
      "learning_rate": 7.2293219526306e-05,
      "loss": 1.5902,
      "step": 141888
    },
    {
      "epoch": 0.2800523387932116,
      "grad_norm": 0.10980308800935745,
      "learning_rate": 7.22868793019413e-05,
      "loss": 1.6085,
      "step": 141920
    },
    {
      "epoch": 0.28011548475460807,
      "grad_norm": 0.11822338402271271,
      "learning_rate": 7.228053907757661e-05,
      "loss": 1.6007,
      "step": 141952
    },
    {
      "epoch": 0.2801786307160045,
      "grad_norm": 0.10629415512084961,
      "learning_rate": 7.227419885321191e-05,
      "loss": 1.6061,
      "step": 141984
    },
    {
      "epoch": 0.2802417766774009,
      "grad_norm": 0.10353866219520569,
      "learning_rate": 7.226785862884723e-05,
      "loss": 1.5906,
      "step": 142016
    },
    {
      "epoch": 0.28030492263879736,
      "grad_norm": 0.1042291447520256,
      "learning_rate": 7.226151840448254e-05,
      "loss": 1.6029,
      "step": 142048
    },
    {
      "epoch": 0.2803680686001938,
      "grad_norm": 0.11149253696203232,
      "learning_rate": 7.225517818011786e-05,
      "loss": 1.6022,
      "step": 142080
    },
    {
      "epoch": 0.28043121456159026,
      "grad_norm": 0.10171899199485779,
      "learning_rate": 7.224883795575317e-05,
      "loss": 1.6227,
      "step": 142112
    },
    {
      "epoch": 0.2804943605229867,
      "grad_norm": 0.1131385862827301,
      "learning_rate": 7.224249773138847e-05,
      "loss": 1.5984,
      "step": 142144
    },
    {
      "epoch": 0.28055750648438316,
      "grad_norm": 0.11076127737760544,
      "learning_rate": 7.223615750702379e-05,
      "loss": 1.603,
      "step": 142176
    },
    {
      "epoch": 0.28062065244577955,
      "grad_norm": 0.10894398391246796,
      "learning_rate": 7.22298172826591e-05,
      "loss": 1.6074,
      "step": 142208
    },
    {
      "epoch": 0.280683798407176,
      "grad_norm": 0.107597716152668,
      "learning_rate": 7.222347705829441e-05,
      "loss": 1.6097,
      "step": 142240
    },
    {
      "epoch": 0.28074694436857245,
      "grad_norm": 0.10396246612071991,
      "learning_rate": 7.221713683392972e-05,
      "loss": 1.5811,
      "step": 142272
    },
    {
      "epoch": 0.2808100903299689,
      "grad_norm": 0.11265064775943756,
      "learning_rate": 7.221079660956503e-05,
      "loss": 1.6068,
      "step": 142304
    },
    {
      "epoch": 0.28087323629136535,
      "grad_norm": 0.11541104316711426,
      "learning_rate": 7.220445638520033e-05,
      "loss": 1.6007,
      "step": 142336
    },
    {
      "epoch": 0.2809363822527618,
      "grad_norm": 0.10701020807027817,
      "learning_rate": 7.219811616083565e-05,
      "loss": 1.6005,
      "step": 142368
    },
    {
      "epoch": 0.2809995282141582,
      "grad_norm": 0.10639555752277374,
      "learning_rate": 7.219177593647095e-05,
      "loss": 1.59,
      "step": 142400
    },
    {
      "epoch": 0.28106267417555464,
      "grad_norm": 0.11335942149162292,
      "learning_rate": 7.218543571210626e-05,
      "loss": 1.6057,
      "step": 142432
    },
    {
      "epoch": 0.2811258201369511,
      "grad_norm": 0.10396453738212585,
      "learning_rate": 7.217909548774158e-05,
      "loss": 1.6,
      "step": 142464
    },
    {
      "epoch": 0.28118896609834754,
      "grad_norm": 0.11395416408777237,
      "learning_rate": 7.217275526337689e-05,
      "loss": 1.6118,
      "step": 142496
    },
    {
      "epoch": 0.281252112059744,
      "grad_norm": 0.10711883008480072,
      "learning_rate": 7.216641503901219e-05,
      "loss": 1.6113,
      "step": 142528
    },
    {
      "epoch": 0.28131525802114044,
      "grad_norm": 0.11413022875785828,
      "learning_rate": 7.21600748146475e-05,
      "loss": 1.6212,
      "step": 142560
    },
    {
      "epoch": 0.28137840398253683,
      "grad_norm": 0.11210522055625916,
      "learning_rate": 7.215373459028282e-05,
      "loss": 1.6146,
      "step": 142592
    },
    {
      "epoch": 0.2814415499439333,
      "grad_norm": 0.10961081087589264,
      "learning_rate": 7.214739436591813e-05,
      "loss": 1.6007,
      "step": 142624
    },
    {
      "epoch": 0.2815046959053297,
      "grad_norm": 0.10824732482433319,
      "learning_rate": 7.214105414155344e-05,
      "loss": 1.5992,
      "step": 142656
    },
    {
      "epoch": 0.2815678418667262,
      "grad_norm": 0.10795754194259644,
      "learning_rate": 7.213471391718875e-05,
      "loss": 1.6049,
      "step": 142688
    },
    {
      "epoch": 0.2816309878281226,
      "grad_norm": 0.10979624837636948,
      "learning_rate": 7.212837369282405e-05,
      "loss": 1.6099,
      "step": 142720
    },
    {
      "epoch": 0.2816941337895191,
      "grad_norm": 0.10198965668678284,
      "learning_rate": 7.212203346845937e-05,
      "loss": 1.6163,
      "step": 142752
    },
    {
      "epoch": 0.28175727975091547,
      "grad_norm": 0.1076868399977684,
      "learning_rate": 7.211569324409468e-05,
      "loss": 1.6022,
      "step": 142784
    },
    {
      "epoch": 0.2818204257123119,
      "grad_norm": 0.10252309590578079,
      "learning_rate": 7.210935301972998e-05,
      "loss": 1.6058,
      "step": 142816
    },
    {
      "epoch": 0.28188357167370837,
      "grad_norm": 0.10857546329498291,
      "learning_rate": 7.21030127953653e-05,
      "loss": 1.6054,
      "step": 142848
    },
    {
      "epoch": 0.2819467176351048,
      "grad_norm": 0.10896912962198257,
      "learning_rate": 7.209667257100061e-05,
      "loss": 1.6055,
      "step": 142880
    },
    {
      "epoch": 0.28200986359650126,
      "grad_norm": 0.1093268096446991,
      "learning_rate": 7.209033234663593e-05,
      "loss": 1.6093,
      "step": 142912
    },
    {
      "epoch": 0.2820730095578977,
      "grad_norm": 0.10696156322956085,
      "learning_rate": 7.208399212227123e-05,
      "loss": 1.5884,
      "step": 142944
    },
    {
      "epoch": 0.2821361555192941,
      "grad_norm": 0.1142810732126236,
      "learning_rate": 7.207765189790654e-05,
      "loss": 1.6073,
      "step": 142976
    },
    {
      "epoch": 0.28219930148069056,
      "grad_norm": 0.1061597615480423,
      "learning_rate": 7.207131167354186e-05,
      "loss": 1.6091,
      "step": 143008
    },
    {
      "epoch": 0.282262447442087,
      "grad_norm": 0.11574750393629074,
      "learning_rate": 7.206497144917717e-05,
      "loss": 1.596,
      "step": 143040
    },
    {
      "epoch": 0.28232559340348345,
      "grad_norm": 0.10675003379583359,
      "learning_rate": 7.205863122481247e-05,
      "loss": 1.6057,
      "step": 143072
    },
    {
      "epoch": 0.2823887393648799,
      "grad_norm": 0.11051331460475922,
      "learning_rate": 7.205229100044779e-05,
      "loss": 1.6074,
      "step": 143104
    },
    {
      "epoch": 0.28245188532627635,
      "grad_norm": 0.10394987463951111,
      "learning_rate": 7.204595077608309e-05,
      "loss": 1.6022,
      "step": 143136
    },
    {
      "epoch": 0.28251503128767275,
      "grad_norm": 0.10684893280267715,
      "learning_rate": 7.20396105517184e-05,
      "loss": 1.5979,
      "step": 143168
    },
    {
      "epoch": 0.2825781772490692,
      "grad_norm": 0.11293821036815643,
      "learning_rate": 7.20332703273537e-05,
      "loss": 1.609,
      "step": 143200
    },
    {
      "epoch": 0.28264132321046564,
      "grad_norm": 0.10789318382740021,
      "learning_rate": 7.202693010298902e-05,
      "loss": 1.6146,
      "step": 143232
    },
    {
      "epoch": 0.2827044691718621,
      "grad_norm": 0.0996130034327507,
      "learning_rate": 7.202058987862433e-05,
      "loss": 1.5966,
      "step": 143264
    },
    {
      "epoch": 0.28276761513325854,
      "grad_norm": 0.1046484187245369,
      "learning_rate": 7.201424965425965e-05,
      "loss": 1.6021,
      "step": 143296
    },
    {
      "epoch": 0.282830761094655,
      "grad_norm": 0.1035534217953682,
      "learning_rate": 7.200790942989495e-05,
      "loss": 1.5998,
      "step": 143328
    },
    {
      "epoch": 0.28289390705605144,
      "grad_norm": 0.1182725727558136,
      "learning_rate": 7.200156920553026e-05,
      "loss": 1.5969,
      "step": 143360
    },
    {
      "epoch": 0.28295705301744783,
      "grad_norm": 0.11813196539878845,
      "learning_rate": 7.199522898116558e-05,
      "loss": 1.5962,
      "step": 143392
    },
    {
      "epoch": 0.2830201989788443,
      "grad_norm": 0.11393732577562332,
      "learning_rate": 7.198888875680089e-05,
      "loss": 1.607,
      "step": 143424
    },
    {
      "epoch": 0.28308334494024073,
      "grad_norm": 0.11106444895267487,
      "learning_rate": 7.19825485324362e-05,
      "loss": 1.5973,
      "step": 143456
    },
    {
      "epoch": 0.2831464909016372,
      "grad_norm": 0.10667029768228531,
      "learning_rate": 7.19762083080715e-05,
      "loss": 1.6212,
      "step": 143488
    },
    {
      "epoch": 0.28320963686303363,
      "grad_norm": 0.10621143877506256,
      "learning_rate": 7.196986808370682e-05,
      "loss": 1.5952,
      "step": 143520
    },
    {
      "epoch": 0.2832727828244301,
      "grad_norm": 0.1070103719830513,
      "learning_rate": 7.196352785934212e-05,
      "loss": 1.6074,
      "step": 143552
    },
    {
      "epoch": 0.2833359287858265,
      "grad_norm": 0.10600332170724869,
      "learning_rate": 7.195718763497744e-05,
      "loss": 1.5992,
      "step": 143584
    },
    {
      "epoch": 0.2833990747472229,
      "grad_norm": 0.12146126478910446,
      "learning_rate": 7.195084741061274e-05,
      "loss": 1.5951,
      "step": 143616
    },
    {
      "epoch": 0.2834622207086194,
      "grad_norm": 0.104829341173172,
      "learning_rate": 7.194450718624805e-05,
      "loss": 1.604,
      "step": 143648
    },
    {
      "epoch": 0.2835253666700158,
      "grad_norm": 0.10839908570051193,
      "learning_rate": 7.193816696188337e-05,
      "loss": 1.6097,
      "step": 143680
    },
    {
      "epoch": 0.28358851263141227,
      "grad_norm": 0.11826279014348984,
      "learning_rate": 7.193182673751868e-05,
      "loss": 1.6124,
      "step": 143712
    },
    {
      "epoch": 0.2836516585928087,
      "grad_norm": 0.10221344232559204,
      "learning_rate": 7.192548651315398e-05,
      "loss": 1.6053,
      "step": 143744
    },
    {
      "epoch": 0.2837148045542051,
      "grad_norm": 0.1202864870429039,
      "learning_rate": 7.19191462887893e-05,
      "loss": 1.6076,
      "step": 143776
    },
    {
      "epoch": 0.28377795051560156,
      "grad_norm": 0.10724340379238129,
      "learning_rate": 7.191280606442461e-05,
      "loss": 1.5984,
      "step": 143808
    },
    {
      "epoch": 0.283841096476998,
      "grad_norm": 0.1065857782959938,
      "learning_rate": 7.190646584005993e-05,
      "loss": 1.6075,
      "step": 143840
    },
    {
      "epoch": 0.28390424243839446,
      "grad_norm": 0.10594381392002106,
      "learning_rate": 7.190012561569523e-05,
      "loss": 1.6038,
      "step": 143872
    },
    {
      "epoch": 0.2839673883997909,
      "grad_norm": 0.10046811401844025,
      "learning_rate": 7.189378539133054e-05,
      "loss": 1.6021,
      "step": 143904
    },
    {
      "epoch": 0.28403053436118736,
      "grad_norm": 0.10927850008010864,
      "learning_rate": 7.188744516696586e-05,
      "loss": 1.606,
      "step": 143936
    },
    {
      "epoch": 0.28409368032258375,
      "grad_norm": 0.11072222888469696,
      "learning_rate": 7.188110494260116e-05,
      "loss": 1.6049,
      "step": 143968
    },
    {
      "epoch": 0.2841568262839802,
      "grad_norm": 0.10642096400260925,
      "learning_rate": 7.187476471823647e-05,
      "loss": 1.6096,
      "step": 144000
    },
    {
      "epoch": 0.28421997224537665,
      "grad_norm": 0.11058561503887177,
      "learning_rate": 7.186842449387177e-05,
      "loss": 1.6056,
      "step": 144032
    },
    {
      "epoch": 0.2842831182067731,
      "grad_norm": 0.10767220705747604,
      "learning_rate": 7.186208426950709e-05,
      "loss": 1.6095,
      "step": 144064
    },
    {
      "epoch": 0.28434626416816955,
      "grad_norm": 0.10795776546001434,
      "learning_rate": 7.18557440451424e-05,
      "loss": 1.6033,
      "step": 144096
    },
    {
      "epoch": 0.284409410129566,
      "grad_norm": 0.10739440470933914,
      "learning_rate": 7.184940382077772e-05,
      "loss": 1.6001,
      "step": 144128
    },
    {
      "epoch": 0.2844725560909624,
      "grad_norm": 0.1105915904045105,
      "learning_rate": 7.184306359641302e-05,
      "loss": 1.6063,
      "step": 144160
    },
    {
      "epoch": 0.28453570205235884,
      "grad_norm": 0.10494209825992584,
      "learning_rate": 7.183672337204833e-05,
      "loss": 1.6111,
      "step": 144192
    },
    {
      "epoch": 0.2845988480137553,
      "grad_norm": 0.10192414373159409,
      "learning_rate": 7.183038314768365e-05,
      "loss": 1.6148,
      "step": 144224
    },
    {
      "epoch": 0.28466199397515174,
      "grad_norm": 0.11192287504673004,
      "learning_rate": 7.182404292331896e-05,
      "loss": 1.6066,
      "step": 144256
    },
    {
      "epoch": 0.2847251399365482,
      "grad_norm": 0.10883938521146774,
      "learning_rate": 7.181770269895426e-05,
      "loss": 1.5984,
      "step": 144288
    },
    {
      "epoch": 0.28478828589794464,
      "grad_norm": 0.10139583051204681,
      "learning_rate": 7.181136247458958e-05,
      "loss": 1.6042,
      "step": 144320
    },
    {
      "epoch": 0.28485143185934103,
      "grad_norm": 0.1154075488448143,
      "learning_rate": 7.180502225022489e-05,
      "loss": 1.6124,
      "step": 144352
    },
    {
      "epoch": 0.2849145778207375,
      "grad_norm": 0.10956396162509918,
      "learning_rate": 7.179868202586019e-05,
      "loss": 1.605,
      "step": 144384
    },
    {
      "epoch": 0.28497772378213393,
      "grad_norm": 0.11622713506221771,
      "learning_rate": 7.179234180149549e-05,
      "loss": 1.597,
      "step": 144416
    },
    {
      "epoch": 0.2850408697435304,
      "grad_norm": 0.11885629594326019,
      "learning_rate": 7.178600157713081e-05,
      "loss": 1.606,
      "step": 144448
    },
    {
      "epoch": 0.2851040157049268,
      "grad_norm": 0.11188899725675583,
      "learning_rate": 7.177966135276612e-05,
      "loss": 1.6088,
      "step": 144480
    },
    {
      "epoch": 0.2851671616663233,
      "grad_norm": 0.11456351727247238,
      "learning_rate": 7.177332112840144e-05,
      "loss": 1.5963,
      "step": 144512
    },
    {
      "epoch": 0.28523030762771967,
      "grad_norm": 0.11649195104837418,
      "learning_rate": 7.176698090403674e-05,
      "loss": 1.6048,
      "step": 144544
    },
    {
      "epoch": 0.2852934535891161,
      "grad_norm": 0.10673061013221741,
      "learning_rate": 7.176064067967205e-05,
      "loss": 1.5998,
      "step": 144576
    },
    {
      "epoch": 0.28535659955051257,
      "grad_norm": 0.11773225665092468,
      "learning_rate": 7.175430045530737e-05,
      "loss": 1.6071,
      "step": 144608
    },
    {
      "epoch": 0.285419745511909,
      "grad_norm": 0.11230260878801346,
      "learning_rate": 7.174796023094268e-05,
      "loss": 1.6029,
      "step": 144640
    },
    {
      "epoch": 0.28548289147330547,
      "grad_norm": 0.11644361168146133,
      "learning_rate": 7.174162000657798e-05,
      "loss": 1.5975,
      "step": 144672
    },
    {
      "epoch": 0.2855460374347019,
      "grad_norm": 0.11056544631719589,
      "learning_rate": 7.17352797822133e-05,
      "loss": 1.5879,
      "step": 144704
    },
    {
      "epoch": 0.2856091833960983,
      "grad_norm": 0.11167342960834503,
      "learning_rate": 7.172893955784861e-05,
      "loss": 1.5982,
      "step": 144736
    },
    {
      "epoch": 0.28567232935749476,
      "grad_norm": 0.10501691699028015,
      "learning_rate": 7.172259933348393e-05,
      "loss": 1.5986,
      "step": 144768
    },
    {
      "epoch": 0.2857354753188912,
      "grad_norm": 0.11405222117900848,
      "learning_rate": 7.171625910911923e-05,
      "loss": 1.6024,
      "step": 144800
    },
    {
      "epoch": 0.28579862128028766,
      "grad_norm": 0.10829059779644012,
      "learning_rate": 7.170991888475453e-05,
      "loss": 1.6031,
      "step": 144832
    },
    {
      "epoch": 0.2858617672416841,
      "grad_norm": 0.12299740314483643,
      "learning_rate": 7.170357866038984e-05,
      "loss": 1.5998,
      "step": 144864
    },
    {
      "epoch": 0.28592491320308056,
      "grad_norm": 0.10545315593481064,
      "learning_rate": 7.169723843602516e-05,
      "loss": 1.6091,
      "step": 144896
    },
    {
      "epoch": 0.28598805916447695,
      "grad_norm": 0.10883430391550064,
      "learning_rate": 7.169089821166047e-05,
      "loss": 1.5987,
      "step": 144928
    },
    {
      "epoch": 0.2860512051258734,
      "grad_norm": 0.10413713753223419,
      "learning_rate": 7.168455798729577e-05,
      "loss": 1.609,
      "step": 144960
    },
    {
      "epoch": 0.28611435108726985,
      "grad_norm": 0.1098276749253273,
      "learning_rate": 7.167821776293109e-05,
      "loss": 1.6083,
      "step": 144992
    },
    {
      "epoch": 0.2861774970486663,
      "grad_norm": 0.10545022785663605,
      "learning_rate": 7.16718775385664e-05,
      "loss": 1.6115,
      "step": 145024
    },
    {
      "epoch": 0.28624064301006275,
      "grad_norm": 0.11426562070846558,
      "learning_rate": 7.166553731420172e-05,
      "loss": 1.6015,
      "step": 145056
    },
    {
      "epoch": 0.2863037889714592,
      "grad_norm": 0.11089000105857849,
      "learning_rate": 7.165919708983702e-05,
      "loss": 1.6013,
      "step": 145088
    },
    {
      "epoch": 0.28636693493285564,
      "grad_norm": 0.1029604896903038,
      "learning_rate": 7.165285686547233e-05,
      "loss": 1.6078,
      "step": 145120
    },
    {
      "epoch": 0.28643008089425204,
      "grad_norm": 0.11236855387687683,
      "learning_rate": 7.164651664110765e-05,
      "loss": 1.5883,
      "step": 145152
    },
    {
      "epoch": 0.2864932268556485,
      "grad_norm": 0.10302947461605072,
      "learning_rate": 7.164017641674296e-05,
      "loss": 1.6048,
      "step": 145184
    },
    {
      "epoch": 0.28655637281704494,
      "grad_norm": 0.10683345049619675,
      "learning_rate": 7.163383619237826e-05,
      "loss": 1.6015,
      "step": 145216
    },
    {
      "epoch": 0.2866195187784414,
      "grad_norm": 0.1122296154499054,
      "learning_rate": 7.162749596801356e-05,
      "loss": 1.6076,
      "step": 145248
    },
    {
      "epoch": 0.28668266473983783,
      "grad_norm": 0.1146034374833107,
      "learning_rate": 7.162115574364888e-05,
      "loss": 1.6012,
      "step": 145280
    },
    {
      "epoch": 0.2867458107012343,
      "grad_norm": 0.10625258833169937,
      "learning_rate": 7.161481551928419e-05,
      "loss": 1.6087,
      "step": 145312
    },
    {
      "epoch": 0.2868089566626307,
      "grad_norm": 0.1057293638586998,
      "learning_rate": 7.16084752949195e-05,
      "loss": 1.5994,
      "step": 145344
    },
    {
      "epoch": 0.2868721026240271,
      "grad_norm": 0.1120283454656601,
      "learning_rate": 7.160213507055481e-05,
      "loss": 1.596,
      "step": 145376
    },
    {
      "epoch": 0.2869352485854236,
      "grad_norm": 0.10851830989122391,
      "learning_rate": 7.159579484619012e-05,
      "loss": 1.6061,
      "step": 145408
    },
    {
      "epoch": 0.28699839454682,
      "grad_norm": 0.11628100275993347,
      "learning_rate": 7.158945462182544e-05,
      "loss": 1.6189,
      "step": 145440
    },
    {
      "epoch": 0.2870615405082165,
      "grad_norm": 0.11050492525100708,
      "learning_rate": 7.158311439746075e-05,
      "loss": 1.6009,
      "step": 145472
    },
    {
      "epoch": 0.2871246864696129,
      "grad_norm": 0.10678663849830627,
      "learning_rate": 7.157677417309605e-05,
      "loss": 1.6039,
      "step": 145504
    },
    {
      "epoch": 0.2871878324310093,
      "grad_norm": 0.10617556422948837,
      "learning_rate": 7.157043394873137e-05,
      "loss": 1.6002,
      "step": 145536
    },
    {
      "epoch": 0.28725097839240576,
      "grad_norm": 0.1099933609366417,
      "learning_rate": 7.156409372436668e-05,
      "loss": 1.5951,
      "step": 145568
    },
    {
      "epoch": 0.2873141243538022,
      "grad_norm": 0.10714271664619446,
      "learning_rate": 7.155775350000198e-05,
      "loss": 1.5954,
      "step": 145600
    },
    {
      "epoch": 0.28737727031519866,
      "grad_norm": 0.1074344664812088,
      "learning_rate": 7.15514132756373e-05,
      "loss": 1.5993,
      "step": 145632
    },
    {
      "epoch": 0.2874404162765951,
      "grad_norm": 0.10610081255435944,
      "learning_rate": 7.15450730512726e-05,
      "loss": 1.6006,
      "step": 145664
    },
    {
      "epoch": 0.28750356223799156,
      "grad_norm": 0.11171463876962662,
      "learning_rate": 7.153873282690791e-05,
      "loss": 1.6037,
      "step": 145696
    },
    {
      "epoch": 0.28756670819938795,
      "grad_norm": 0.10531670600175858,
      "learning_rate": 7.153239260254323e-05,
      "loss": 1.5984,
      "step": 145728
    },
    {
      "epoch": 0.2876298541607844,
      "grad_norm": 0.11367079615592957,
      "learning_rate": 7.152605237817853e-05,
      "loss": 1.6137,
      "step": 145760
    },
    {
      "epoch": 0.28769300012218085,
      "grad_norm": 0.11154003441333771,
      "learning_rate": 7.151971215381384e-05,
      "loss": 1.5993,
      "step": 145792
    },
    {
      "epoch": 0.2877561460835773,
      "grad_norm": 0.12410751730203629,
      "learning_rate": 7.151337192944916e-05,
      "loss": 1.6105,
      "step": 145824
    },
    {
      "epoch": 0.28781929204497375,
      "grad_norm": 0.1111689880490303,
      "learning_rate": 7.150703170508447e-05,
      "loss": 1.5901,
      "step": 145856
    },
    {
      "epoch": 0.2878824380063702,
      "grad_norm": 0.1172381192445755,
      "learning_rate": 7.150069148071977e-05,
      "loss": 1.6027,
      "step": 145888
    },
    {
      "epoch": 0.2879455839677666,
      "grad_norm": 0.11613024026155472,
      "learning_rate": 7.149435125635509e-05,
      "loss": 1.6001,
      "step": 145920
    },
    {
      "epoch": 0.28800872992916304,
      "grad_norm": 0.11064592748880386,
      "learning_rate": 7.14880110319904e-05,
      "loss": 1.606,
      "step": 145952
    },
    {
      "epoch": 0.2880718758905595,
      "grad_norm": 0.11777277290821075,
      "learning_rate": 7.148167080762572e-05,
      "loss": 1.6022,
      "step": 145984
    },
    {
      "epoch": 0.28813502185195594,
      "grad_norm": 0.10947234183549881,
      "learning_rate": 7.147533058326102e-05,
      "loss": 1.6143,
      "step": 146016
    },
    {
      "epoch": 0.2881981678133524,
      "grad_norm": 0.10328909754753113,
      "learning_rate": 7.146899035889633e-05,
      "loss": 1.6007,
      "step": 146048
    },
    {
      "epoch": 0.28826131377474884,
      "grad_norm": 0.11179718375205994,
      "learning_rate": 7.146265013453163e-05,
      "loss": 1.6088,
      "step": 146080
    },
    {
      "epoch": 0.28832445973614523,
      "grad_norm": 0.11844056099653244,
      "learning_rate": 7.145630991016695e-05,
      "loss": 1.6014,
      "step": 146112
    },
    {
      "epoch": 0.2883876056975417,
      "grad_norm": 0.12668026983737946,
      "learning_rate": 7.144996968580226e-05,
      "loss": 1.62,
      "step": 146144
    },
    {
      "epoch": 0.28845075165893813,
      "grad_norm": 0.10801265388727188,
      "learning_rate": 7.144362946143756e-05,
      "loss": 1.6127,
      "step": 146176
    },
    {
      "epoch": 0.2885138976203346,
      "grad_norm": 0.1138705238699913,
      "learning_rate": 7.143728923707288e-05,
      "loss": 1.6062,
      "step": 146208
    },
    {
      "epoch": 0.28857704358173103,
      "grad_norm": 0.11289960891008377,
      "learning_rate": 7.143094901270819e-05,
      "loss": 1.6037,
      "step": 146240
    },
    {
      "epoch": 0.2886401895431275,
      "grad_norm": 0.11056394875049591,
      "learning_rate": 7.142460878834351e-05,
      "loss": 1.6001,
      "step": 146272
    },
    {
      "epoch": 0.2887033355045239,
      "grad_norm": 0.10921726375818253,
      "learning_rate": 7.141826856397881e-05,
      "loss": 1.619,
      "step": 146304
    },
    {
      "epoch": 0.2887664814659203,
      "grad_norm": 0.10466153174638748,
      "learning_rate": 7.141192833961412e-05,
      "loss": 1.5916,
      "step": 146336
    },
    {
      "epoch": 0.28882962742731677,
      "grad_norm": 0.1139112189412117,
      "learning_rate": 7.140558811524944e-05,
      "loss": 1.5999,
      "step": 146368
    },
    {
      "epoch": 0.2888927733887132,
      "grad_norm": 0.10813932865858078,
      "learning_rate": 7.139924789088475e-05,
      "loss": 1.6015,
      "step": 146400
    },
    {
      "epoch": 0.28895591935010967,
      "grad_norm": 0.1114303320646286,
      "learning_rate": 7.139290766652005e-05,
      "loss": 1.606,
      "step": 146432
    },
    {
      "epoch": 0.2890190653115061,
      "grad_norm": 0.10804945975542068,
      "learning_rate": 7.138656744215537e-05,
      "loss": 1.5917,
      "step": 146464
    },
    {
      "epoch": 0.2890822112729025,
      "grad_norm": 0.10753102600574493,
      "learning_rate": 7.138022721779067e-05,
      "loss": 1.5981,
      "step": 146496
    },
    {
      "epoch": 0.28914535723429896,
      "grad_norm": 0.10849519073963165,
      "learning_rate": 7.137388699342598e-05,
      "loss": 1.6072,
      "step": 146528
    },
    {
      "epoch": 0.2892085031956954,
      "grad_norm": 0.10747122019529343,
      "learning_rate": 7.136754676906128e-05,
      "loss": 1.5937,
      "step": 146560
    },
    {
      "epoch": 0.28927164915709186,
      "grad_norm": 0.10553805530071259,
      "learning_rate": 7.13612065446966e-05,
      "loss": 1.5914,
      "step": 146592
    },
    {
      "epoch": 0.2893347951184883,
      "grad_norm": 0.11188548803329468,
      "learning_rate": 7.135486632033191e-05,
      "loss": 1.6041,
      "step": 146624
    },
    {
      "epoch": 0.28939794107988476,
      "grad_norm": 0.10540486872196198,
      "learning_rate": 7.134852609596723e-05,
      "loss": 1.5981,
      "step": 146656
    },
    {
      "epoch": 0.2894610870412812,
      "grad_norm": 0.1163511872291565,
      "learning_rate": 7.134218587160253e-05,
      "loss": 1.6033,
      "step": 146688
    },
    {
      "epoch": 0.2895242330026776,
      "grad_norm": 0.10686226189136505,
      "learning_rate": 7.133584564723784e-05,
      "loss": 1.6045,
      "step": 146720
    },
    {
      "epoch": 0.28958737896407405,
      "grad_norm": 0.10692430287599564,
      "learning_rate": 7.132950542287316e-05,
      "loss": 1.586,
      "step": 146752
    },
    {
      "epoch": 0.2896505249254705,
      "grad_norm": 0.10165756940841675,
      "learning_rate": 7.132316519850847e-05,
      "loss": 1.6093,
      "step": 146784
    },
    {
      "epoch": 0.28971367088686695,
      "grad_norm": 0.11457328498363495,
      "learning_rate": 7.131682497414379e-05,
      "loss": 1.6065,
      "step": 146816
    },
    {
      "epoch": 0.2897768168482634,
      "grad_norm": 0.1179886907339096,
      "learning_rate": 7.131048474977909e-05,
      "loss": 1.5993,
      "step": 146848
    },
    {
      "epoch": 0.28983996280965985,
      "grad_norm": 0.12959076464176178,
      "learning_rate": 7.13041445254144e-05,
      "loss": 1.5865,
      "step": 146880
    },
    {
      "epoch": 0.28990310877105624,
      "grad_norm": 0.10887438803911209,
      "learning_rate": 7.12978043010497e-05,
      "loss": 1.5928,
      "step": 146912
    },
    {
      "epoch": 0.2899662547324527,
      "grad_norm": 0.11446130275726318,
      "learning_rate": 7.129146407668502e-05,
      "loss": 1.6029,
      "step": 146944
    },
    {
      "epoch": 0.29002940069384914,
      "grad_norm": 0.10503789782524109,
      "learning_rate": 7.128512385232032e-05,
      "loss": 1.5972,
      "step": 146976
    },
    {
      "epoch": 0.2900925466552456,
      "grad_norm": 0.11018822342157364,
      "learning_rate": 7.127878362795563e-05,
      "loss": 1.5938,
      "step": 147008
    },
    {
      "epoch": 0.29015569261664204,
      "grad_norm": 0.10738013684749603,
      "learning_rate": 7.127244340359095e-05,
      "loss": 1.5985,
      "step": 147040
    },
    {
      "epoch": 0.2902188385780385,
      "grad_norm": 0.11055608093738556,
      "learning_rate": 7.126610317922626e-05,
      "loss": 1.6051,
      "step": 147072
    },
    {
      "epoch": 0.2902819845394349,
      "grad_norm": 0.10610676556825638,
      "learning_rate": 7.125976295486156e-05,
      "loss": 1.5933,
      "step": 147104
    },
    {
      "epoch": 0.2903451305008313,
      "grad_norm": 0.10864364355802536,
      "learning_rate": 7.125342273049688e-05,
      "loss": 1.593,
      "step": 147136
    },
    {
      "epoch": 0.2904082764622278,
      "grad_norm": 0.10153425484895706,
      "learning_rate": 7.124708250613219e-05,
      "loss": 1.6011,
      "step": 147168
    },
    {
      "epoch": 0.2904714224236242,
      "grad_norm": 0.1137988343834877,
      "learning_rate": 7.124074228176751e-05,
      "loss": 1.6008,
      "step": 147200
    },
    {
      "epoch": 0.2905345683850207,
      "grad_norm": 0.10674472898244858,
      "learning_rate": 7.123440205740281e-05,
      "loss": 1.6047,
      "step": 147232
    },
    {
      "epoch": 0.2905977143464171,
      "grad_norm": 0.10434059798717499,
      "learning_rate": 7.122806183303812e-05,
      "loss": 1.6037,
      "step": 147264
    },
    {
      "epoch": 0.2906608603078135,
      "grad_norm": 0.11284546554088593,
      "learning_rate": 7.122172160867342e-05,
      "loss": 1.6025,
      "step": 147296
    },
    {
      "epoch": 0.29072400626920997,
      "grad_norm": 0.1067948043346405,
      "learning_rate": 7.121538138430874e-05,
      "loss": 1.612,
      "step": 147328
    },
    {
      "epoch": 0.2907871522306064,
      "grad_norm": 0.11101321876049042,
      "learning_rate": 7.120904115994404e-05,
      "loss": 1.6022,
      "step": 147360
    },
    {
      "epoch": 0.29085029819200287,
      "grad_norm": 0.1162412166595459,
      "learning_rate": 7.120270093557935e-05,
      "loss": 1.6005,
      "step": 147392
    },
    {
      "epoch": 0.2909134441533993,
      "grad_norm": 0.10932154208421707,
      "learning_rate": 7.119636071121467e-05,
      "loss": 1.6055,
      "step": 147424
    },
    {
      "epoch": 0.29097659011479576,
      "grad_norm": 0.1102326512336731,
      "learning_rate": 7.119002048684998e-05,
      "loss": 1.6023,
      "step": 147456
    },
    {
      "epoch": 0.29103973607619216,
      "grad_norm": 0.1088930070400238,
      "learning_rate": 7.11836802624853e-05,
      "loss": 1.605,
      "step": 147488
    },
    {
      "epoch": 0.2911028820375886,
      "grad_norm": 0.10465831309556961,
      "learning_rate": 7.11773400381206e-05,
      "loss": 1.6082,
      "step": 147520
    },
    {
      "epoch": 0.29116602799898506,
      "grad_norm": 0.1064116582274437,
      "learning_rate": 7.117099981375591e-05,
      "loss": 1.6036,
      "step": 147552
    },
    {
      "epoch": 0.2912291739603815,
      "grad_norm": 0.11028715968132019,
      "learning_rate": 7.116465958939123e-05,
      "loss": 1.595,
      "step": 147584
    },
    {
      "epoch": 0.29129231992177795,
      "grad_norm": 0.10586348921060562,
      "learning_rate": 7.115831936502654e-05,
      "loss": 1.5992,
      "step": 147616
    },
    {
      "epoch": 0.2913554658831744,
      "grad_norm": 0.10846854746341705,
      "learning_rate": 7.115197914066184e-05,
      "loss": 1.5942,
      "step": 147648
    },
    {
      "epoch": 0.2914186118445708,
      "grad_norm": 0.11915256083011627,
      "learning_rate": 7.114563891629716e-05,
      "loss": 1.5939,
      "step": 147680
    },
    {
      "epoch": 0.29148175780596725,
      "grad_norm": 0.10486552119255066,
      "learning_rate": 7.113929869193246e-05,
      "loss": 1.6025,
      "step": 147712
    },
    {
      "epoch": 0.2915449037673637,
      "grad_norm": 0.10865342617034912,
      "learning_rate": 7.113295846756777e-05,
      "loss": 1.6031,
      "step": 147744
    },
    {
      "epoch": 0.29160804972876014,
      "grad_norm": 0.11034132540225983,
      "learning_rate": 7.112661824320307e-05,
      "loss": 1.5948,
      "step": 147776
    },
    {
      "epoch": 0.2916711956901566,
      "grad_norm": 0.10538811981678009,
      "learning_rate": 7.112027801883839e-05,
      "loss": 1.6005,
      "step": 147808
    },
    {
      "epoch": 0.29173434165155304,
      "grad_norm": 0.10983075946569443,
      "learning_rate": 7.11139377944737e-05,
      "loss": 1.6124,
      "step": 147840
    },
    {
      "epoch": 0.29179748761294944,
      "grad_norm": 0.11809743940830231,
      "learning_rate": 7.110759757010902e-05,
      "loss": 1.6116,
      "step": 147872
    },
    {
      "epoch": 0.2918606335743459,
      "grad_norm": 0.11948984861373901,
      "learning_rate": 7.110125734574432e-05,
      "loss": 1.615,
      "step": 147904
    },
    {
      "epoch": 0.29192377953574233,
      "grad_norm": 0.11344622820615768,
      "learning_rate": 7.109491712137963e-05,
      "loss": 1.596,
      "step": 147936
    },
    {
      "epoch": 0.2919869254971388,
      "grad_norm": 0.10580223053693771,
      "learning_rate": 7.108857689701495e-05,
      "loss": 1.6069,
      "step": 147968
    },
    {
      "epoch": 0.29205007145853523,
      "grad_norm": 0.10495791584253311,
      "learning_rate": 7.108223667265026e-05,
      "loss": 1.6072,
      "step": 148000
    },
    {
      "epoch": 0.2921132174199317,
      "grad_norm": 0.10190922766923904,
      "learning_rate": 7.107589644828556e-05,
      "loss": 1.6053,
      "step": 148032
    },
    {
      "epoch": 0.2921763633813281,
      "grad_norm": 0.12029528617858887,
      "learning_rate": 7.106955622392088e-05,
      "loss": 1.6044,
      "step": 148064
    },
    {
      "epoch": 0.2922395093427245,
      "grad_norm": 0.11414968967437744,
      "learning_rate": 7.10632159995562e-05,
      "loss": 1.6161,
      "step": 148096
    },
    {
      "epoch": 0.292302655304121,
      "grad_norm": 0.10950406640768051,
      "learning_rate": 7.10568757751915e-05,
      "loss": 1.6007,
      "step": 148128
    },
    {
      "epoch": 0.2923658012655174,
      "grad_norm": 0.11942359805107117,
      "learning_rate": 7.105053555082681e-05,
      "loss": 1.5998,
      "step": 148160
    },
    {
      "epoch": 0.29242894722691387,
      "grad_norm": 0.10723134875297546,
      "learning_rate": 7.104419532646211e-05,
      "loss": 1.5952,
      "step": 148192
    },
    {
      "epoch": 0.2924920931883103,
      "grad_norm": 0.10833689570426941,
      "learning_rate": 7.103785510209742e-05,
      "loss": 1.6096,
      "step": 148224
    },
    {
      "epoch": 0.2925552391497067,
      "grad_norm": 0.11473509669303894,
      "learning_rate": 7.103151487773274e-05,
      "loss": 1.6043,
      "step": 148256
    },
    {
      "epoch": 0.29261838511110316,
      "grad_norm": 0.11084617674350739,
      "learning_rate": 7.102517465336805e-05,
      "loss": 1.5959,
      "step": 148288
    },
    {
      "epoch": 0.2926815310724996,
      "grad_norm": 0.1038217842578888,
      "learning_rate": 7.101883442900335e-05,
      "loss": 1.6055,
      "step": 148320
    },
    {
      "epoch": 0.29274467703389606,
      "grad_norm": 0.1097683310508728,
      "learning_rate": 7.101249420463867e-05,
      "loss": 1.5946,
      "step": 148352
    },
    {
      "epoch": 0.2928078229952925,
      "grad_norm": 0.1091427281498909,
      "learning_rate": 7.100615398027398e-05,
      "loss": 1.6058,
      "step": 148384
    },
    {
      "epoch": 0.29287096895668896,
      "grad_norm": 0.11220090836286545,
      "learning_rate": 7.09998137559093e-05,
      "loss": 1.5975,
      "step": 148416
    },
    {
      "epoch": 0.2929341149180854,
      "grad_norm": 0.10755281895399094,
      "learning_rate": 7.09934735315446e-05,
      "loss": 1.5954,
      "step": 148448
    },
    {
      "epoch": 0.2929972608794818,
      "grad_norm": 0.1037331148982048,
      "learning_rate": 7.098713330717991e-05,
      "loss": 1.5875,
      "step": 148480
    },
    {
      "epoch": 0.29306040684087825,
      "grad_norm": 0.10771684348583221,
      "learning_rate": 7.098079308281523e-05,
      "loss": 1.6043,
      "step": 148512
    },
    {
      "epoch": 0.2931235528022747,
      "grad_norm": 0.10596349835395813,
      "learning_rate": 7.097445285845053e-05,
      "loss": 1.6006,
      "step": 148544
    },
    {
      "epoch": 0.29318669876367115,
      "grad_norm": 0.10761568695306778,
      "learning_rate": 7.096811263408583e-05,
      "loss": 1.6281,
      "step": 148576
    },
    {
      "epoch": 0.2932498447250676,
      "grad_norm": 0.09952478110790253,
      "learning_rate": 7.096177240972115e-05,
      "loss": 1.5996,
      "step": 148608
    },
    {
      "epoch": 0.29331299068646405,
      "grad_norm": 0.10850663483142853,
      "learning_rate": 7.095543218535646e-05,
      "loss": 1.5776,
      "step": 148640
    },
    {
      "epoch": 0.29337613664786044,
      "grad_norm": 0.10966360569000244,
      "learning_rate": 7.094909196099177e-05,
      "loss": 1.5965,
      "step": 148672
    },
    {
      "epoch": 0.2934392826092569,
      "grad_norm": 0.10936962068080902,
      "learning_rate": 7.094275173662708e-05,
      "loss": 1.613,
      "step": 148704
    },
    {
      "epoch": 0.29350242857065334,
      "grad_norm": 0.11115074157714844,
      "learning_rate": 7.093641151226239e-05,
      "loss": 1.6183,
      "step": 148736
    },
    {
      "epoch": 0.2935655745320498,
      "grad_norm": 0.1090892106294632,
      "learning_rate": 7.09300712878977e-05,
      "loss": 1.582,
      "step": 148768
    },
    {
      "epoch": 0.29362872049344624,
      "grad_norm": 0.1073753759264946,
      "learning_rate": 7.092373106353302e-05,
      "loss": 1.5966,
      "step": 148800
    },
    {
      "epoch": 0.2936918664548427,
      "grad_norm": 0.10955362021923065,
      "learning_rate": 7.091739083916833e-05,
      "loss": 1.6111,
      "step": 148832
    },
    {
      "epoch": 0.2937550124162391,
      "grad_norm": 0.10309655219316483,
      "learning_rate": 7.091105061480363e-05,
      "loss": 1.6129,
      "step": 148864
    },
    {
      "epoch": 0.29381815837763553,
      "grad_norm": 0.11600329726934433,
      "learning_rate": 7.090471039043895e-05,
      "loss": 1.6014,
      "step": 148896
    },
    {
      "epoch": 0.293881304339032,
      "grad_norm": 0.10746400058269501,
      "learning_rate": 7.089837016607426e-05,
      "loss": 1.5834,
      "step": 148928
    },
    {
      "epoch": 0.29394445030042843,
      "grad_norm": 0.11148227006196976,
      "learning_rate": 7.089202994170956e-05,
      "loss": 1.5993,
      "step": 148960
    },
    {
      "epoch": 0.2940075962618249,
      "grad_norm": 0.10850510746240616,
      "learning_rate": 7.088568971734487e-05,
      "loss": 1.5943,
      "step": 148992
    },
    {
      "epoch": 0.2940707422232213,
      "grad_norm": 0.10176557302474976,
      "learning_rate": 7.087934949298018e-05,
      "loss": 1.5972,
      "step": 149024
    },
    {
      "epoch": 0.2941338881846177,
      "grad_norm": 0.11594806611537933,
      "learning_rate": 7.08730092686155e-05,
      "loss": 1.5986,
      "step": 149056
    },
    {
      "epoch": 0.29419703414601417,
      "grad_norm": 0.110724076628685,
      "learning_rate": 7.086666904425081e-05,
      "loss": 1.6094,
      "step": 149088
    },
    {
      "epoch": 0.2942601801074106,
      "grad_norm": 0.11467071622610092,
      "learning_rate": 7.086032881988611e-05,
      "loss": 1.597,
      "step": 149120
    },
    {
      "epoch": 0.29432332606880707,
      "grad_norm": 0.11094608902931213,
      "learning_rate": 7.085398859552142e-05,
      "loss": 1.5972,
      "step": 149152
    },
    {
      "epoch": 0.2943864720302035,
      "grad_norm": 0.11065344512462616,
      "learning_rate": 7.084764837115674e-05,
      "loss": 1.5825,
      "step": 149184
    },
    {
      "epoch": 0.29444961799159997,
      "grad_norm": 0.11137869209051132,
      "learning_rate": 7.084130814679205e-05,
      "loss": 1.6025,
      "step": 149216
    },
    {
      "epoch": 0.29451276395299636,
      "grad_norm": 0.10236819833517075,
      "learning_rate": 7.083496792242735e-05,
      "loss": 1.5962,
      "step": 149248
    },
    {
      "epoch": 0.2945759099143928,
      "grad_norm": 0.1143043264746666,
      "learning_rate": 7.082862769806267e-05,
      "loss": 1.6085,
      "step": 149280
    },
    {
      "epoch": 0.29463905587578926,
      "grad_norm": 0.10390573740005493,
      "learning_rate": 7.082228747369798e-05,
      "loss": 1.5948,
      "step": 149312
    },
    {
      "epoch": 0.2947022018371857,
      "grad_norm": 0.11138554662466049,
      "learning_rate": 7.08159472493333e-05,
      "loss": 1.5891,
      "step": 149344
    },
    {
      "epoch": 0.29476534779858216,
      "grad_norm": 0.1055331900715828,
      "learning_rate": 7.08096070249686e-05,
      "loss": 1.6031,
      "step": 149376
    },
    {
      "epoch": 0.2948284937599786,
      "grad_norm": 0.10739952325820923,
      "learning_rate": 7.08032668006039e-05,
      "loss": 1.5874,
      "step": 149408
    },
    {
      "epoch": 0.294891639721375,
      "grad_norm": 0.1127571314573288,
      "learning_rate": 7.079692657623922e-05,
      "loss": 1.598,
      "step": 149440
    },
    {
      "epoch": 0.29495478568277145,
      "grad_norm": 0.10634210705757141,
      "learning_rate": 7.079058635187453e-05,
      "loss": 1.6079,
      "step": 149472
    },
    {
      "epoch": 0.2950179316441679,
      "grad_norm": 0.11272866278886795,
      "learning_rate": 7.078424612750984e-05,
      "loss": 1.6142,
      "step": 149504
    },
    {
      "epoch": 0.29508107760556435,
      "grad_norm": 0.1091655045747757,
      "learning_rate": 7.077790590314515e-05,
      "loss": 1.6018,
      "step": 149536
    },
    {
      "epoch": 0.2951442235669608,
      "grad_norm": 0.11715688556432724,
      "learning_rate": 7.077156567878046e-05,
      "loss": 1.6003,
      "step": 149568
    },
    {
      "epoch": 0.29520736952835724,
      "grad_norm": 0.10742654651403427,
      "learning_rate": 7.076522545441577e-05,
      "loss": 1.6068,
      "step": 149600
    },
    {
      "epoch": 0.29527051548975364,
      "grad_norm": 0.10077372193336487,
      "learning_rate": 7.075888523005109e-05,
      "loss": 1.6075,
      "step": 149632
    },
    {
      "epoch": 0.2953336614511501,
      "grad_norm": 0.11272387951612473,
      "learning_rate": 7.075254500568639e-05,
      "loss": 1.5946,
      "step": 149664
    },
    {
      "epoch": 0.29539680741254654,
      "grad_norm": 0.11600866168737411,
      "learning_rate": 7.07462047813217e-05,
      "loss": 1.6153,
      "step": 149696
    },
    {
      "epoch": 0.295459953373943,
      "grad_norm": 0.11558971554040909,
      "learning_rate": 7.073986455695702e-05,
      "loss": 1.5928,
      "step": 149728
    },
    {
      "epoch": 0.29552309933533943,
      "grad_norm": 0.10414761304855347,
      "learning_rate": 7.073352433259232e-05,
      "loss": 1.5996,
      "step": 149760
    },
    {
      "epoch": 0.2955862452967359,
      "grad_norm": 0.11273051053285599,
      "learning_rate": 7.072718410822763e-05,
      "loss": 1.5946,
      "step": 149792
    },
    {
      "epoch": 0.2956493912581323,
      "grad_norm": 0.11486554890871048,
      "learning_rate": 7.072084388386294e-05,
      "loss": 1.5996,
      "step": 149824
    },
    {
      "epoch": 0.2957125372195287,
      "grad_norm": 0.10669771581888199,
      "learning_rate": 7.071450365949825e-05,
      "loss": 1.5841,
      "step": 149856
    },
    {
      "epoch": 0.2957756831809252,
      "grad_norm": 0.11150334775447845,
      "learning_rate": 7.070816343513356e-05,
      "loss": 1.6167,
      "step": 149888
    },
    {
      "epoch": 0.2958388291423216,
      "grad_norm": 0.11866706609725952,
      "learning_rate": 7.070182321076887e-05,
      "loss": 1.6098,
      "step": 149920
    },
    {
      "epoch": 0.2959019751037181,
      "grad_norm": 0.10640261322259903,
      "learning_rate": 7.069548298640418e-05,
      "loss": 1.5874,
      "step": 149952
    },
    {
      "epoch": 0.2959651210651145,
      "grad_norm": 0.10304625332355499,
      "learning_rate": 7.06891427620395e-05,
      "loss": 1.595,
      "step": 149984
    },
    {
      "epoch": 0.2960282670265109,
      "grad_norm": 0.11005029082298279,
      "learning_rate": 7.068280253767481e-05,
      "loss": 1.5901,
      "step": 150016
    },
    {
      "epoch": 0.29609141298790737,
      "grad_norm": 0.10300908982753754,
      "learning_rate": 7.067646231331012e-05,
      "loss": 1.6064,
      "step": 150048
    },
    {
      "epoch": 0.2961545589493038,
      "grad_norm": 0.10868167132139206,
      "learning_rate": 7.067012208894543e-05,
      "loss": 1.591,
      "step": 150080
    },
    {
      "epoch": 0.29621770491070026,
      "grad_norm": 0.10578823834657669,
      "learning_rate": 7.066378186458074e-05,
      "loss": 1.5964,
      "step": 150112
    },
    {
      "epoch": 0.2962808508720967,
      "grad_norm": 0.11526907980442047,
      "learning_rate": 7.065744164021605e-05,
      "loss": 1.603,
      "step": 150144
    },
    {
      "epoch": 0.29634399683349316,
      "grad_norm": 0.1077684685587883,
      "learning_rate": 7.065110141585136e-05,
      "loss": 1.6009,
      "step": 150176
    },
    {
      "epoch": 0.2964071427948896,
      "grad_norm": 0.10519464313983917,
      "learning_rate": 7.064476119148667e-05,
      "loss": 1.6029,
      "step": 150208
    },
    {
      "epoch": 0.296470288756286,
      "grad_norm": 0.1093609631061554,
      "learning_rate": 7.063842096712197e-05,
      "loss": 1.5931,
      "step": 150240
    },
    {
      "epoch": 0.29653343471768245,
      "grad_norm": 0.11715901643037796,
      "learning_rate": 7.063208074275729e-05,
      "loss": 1.6072,
      "step": 150272
    },
    {
      "epoch": 0.2965965806790789,
      "grad_norm": 0.1089400202035904,
      "learning_rate": 7.06257405183926e-05,
      "loss": 1.5947,
      "step": 150304
    },
    {
      "epoch": 0.29665972664047535,
      "grad_norm": 0.11363311111927032,
      "learning_rate": 7.06194002940279e-05,
      "loss": 1.6053,
      "step": 150336
    },
    {
      "epoch": 0.2967228726018718,
      "grad_norm": 0.10174848884344101,
      "learning_rate": 7.061306006966322e-05,
      "loss": 1.6003,
      "step": 150368
    },
    {
      "epoch": 0.29678601856326825,
      "grad_norm": 0.10809173434972763,
      "learning_rate": 7.060671984529853e-05,
      "loss": 1.6008,
      "step": 150400
    },
    {
      "epoch": 0.29684916452466464,
      "grad_norm": 0.11010542511940002,
      "learning_rate": 7.060037962093384e-05,
      "loss": 1.6032,
      "step": 150432
    },
    {
      "epoch": 0.2969123104860611,
      "grad_norm": 0.11021538823843002,
      "learning_rate": 7.059403939656915e-05,
      "loss": 1.5926,
      "step": 150464
    },
    {
      "epoch": 0.29697545644745754,
      "grad_norm": 0.10456708073616028,
      "learning_rate": 7.058769917220446e-05,
      "loss": 1.5978,
      "step": 150496
    },
    {
      "epoch": 0.297038602408854,
      "grad_norm": 0.11134756356477737,
      "learning_rate": 7.058135894783977e-05,
      "loss": 1.597,
      "step": 150528
    },
    {
      "epoch": 0.29710174837025044,
      "grad_norm": 0.10560545325279236,
      "learning_rate": 7.057501872347509e-05,
      "loss": 1.6003,
      "step": 150560
    },
    {
      "epoch": 0.2971648943316469,
      "grad_norm": 0.11020217835903168,
      "learning_rate": 7.056867849911039e-05,
      "loss": 1.6011,
      "step": 150592
    },
    {
      "epoch": 0.2972280402930433,
      "grad_norm": 0.10672367364168167,
      "learning_rate": 7.05623382747457e-05,
      "loss": 1.5971,
      "step": 150624
    },
    {
      "epoch": 0.29729118625443973,
      "grad_norm": 0.10412941128015518,
      "learning_rate": 7.0555998050381e-05,
      "loss": 1.5943,
      "step": 150656
    },
    {
      "epoch": 0.2973543322158362,
      "grad_norm": 0.1048407033085823,
      "learning_rate": 7.054965782601632e-05,
      "loss": 1.5907,
      "step": 150688
    },
    {
      "epoch": 0.29741747817723263,
      "grad_norm": 0.10758503526449203,
      "learning_rate": 7.054331760165164e-05,
      "loss": 1.6024,
      "step": 150720
    },
    {
      "epoch": 0.2974806241386291,
      "grad_norm": 0.10499800741672516,
      "learning_rate": 7.053697737728694e-05,
      "loss": 1.5926,
      "step": 150752
    },
    {
      "epoch": 0.29754377010002553,
      "grad_norm": 0.10877201706171036,
      "learning_rate": 7.053063715292225e-05,
      "loss": 1.6196,
      "step": 150784
    },
    {
      "epoch": 0.2976069160614219,
      "grad_norm": 0.11338555812835693,
      "learning_rate": 7.052429692855757e-05,
      "loss": 1.5973,
      "step": 150816
    },
    {
      "epoch": 0.29767006202281837,
      "grad_norm": 0.11133240908384323,
      "learning_rate": 7.051795670419288e-05,
      "loss": 1.6117,
      "step": 150848
    },
    {
      "epoch": 0.2977332079842148,
      "grad_norm": 0.1142796203494072,
      "learning_rate": 7.051161647982818e-05,
      "loss": 1.6013,
      "step": 150880
    },
    {
      "epoch": 0.29779635394561127,
      "grad_norm": 0.12019713968038559,
      "learning_rate": 7.05052762554635e-05,
      "loss": 1.5995,
      "step": 150912
    },
    {
      "epoch": 0.2978594999070077,
      "grad_norm": 0.11198505759239197,
      "learning_rate": 7.049893603109881e-05,
      "loss": 1.6067,
      "step": 150944
    },
    {
      "epoch": 0.29792264586840417,
      "grad_norm": 0.11616820842027664,
      "learning_rate": 7.049259580673412e-05,
      "loss": 1.5935,
      "step": 150976
    },
    {
      "epoch": 0.29798579182980056,
      "grad_norm": 0.10942426323890686,
      "learning_rate": 7.048625558236943e-05,
      "loss": 1.6111,
      "step": 151008
    },
    {
      "epoch": 0.298048937791197,
      "grad_norm": 0.10377158969640732,
      "learning_rate": 7.047991535800474e-05,
      "loss": 1.5991,
      "step": 151040
    },
    {
      "epoch": 0.29811208375259346,
      "grad_norm": 0.10230506956577301,
      "learning_rate": 7.047357513364004e-05,
      "loss": 1.6147,
      "step": 151072
    },
    {
      "epoch": 0.2981752297139899,
      "grad_norm": 0.1174510195851326,
      "learning_rate": 7.046723490927536e-05,
      "loss": 1.6124,
      "step": 151104
    },
    {
      "epoch": 0.29823837567538636,
      "grad_norm": 0.10018105804920197,
      "learning_rate": 7.046089468491066e-05,
      "loss": 1.5959,
      "step": 151136
    },
    {
      "epoch": 0.2983015216367828,
      "grad_norm": 0.10968267917633057,
      "learning_rate": 7.045455446054597e-05,
      "loss": 1.6065,
      "step": 151168
    },
    {
      "epoch": 0.2983646675981792,
      "grad_norm": 0.106751449406147,
      "learning_rate": 7.044821423618129e-05,
      "loss": 1.6052,
      "step": 151200
    },
    {
      "epoch": 0.29842781355957565,
      "grad_norm": 0.10615841299295425,
      "learning_rate": 7.04418740118166e-05,
      "loss": 1.6045,
      "step": 151232
    },
    {
      "epoch": 0.2984909595209721,
      "grad_norm": 0.10571028292179108,
      "learning_rate": 7.04355337874519e-05,
      "loss": 1.5928,
      "step": 151264
    },
    {
      "epoch": 0.29855410548236855,
      "grad_norm": 0.11286832392215729,
      "learning_rate": 7.042919356308722e-05,
      "loss": 1.5954,
      "step": 151296
    },
    {
      "epoch": 0.298617251443765,
      "grad_norm": 0.11962977796792984,
      "learning_rate": 7.042285333872253e-05,
      "loss": 1.5935,
      "step": 151328
    },
    {
      "epoch": 0.29868039740516145,
      "grad_norm": 0.10393813252449036,
      "learning_rate": 7.041651311435785e-05,
      "loss": 1.591,
      "step": 151360
    },
    {
      "epoch": 0.29874354336655784,
      "grad_norm": 0.11043057590723038,
      "learning_rate": 7.041017288999316e-05,
      "loss": 1.5955,
      "step": 151392
    },
    {
      "epoch": 0.2988066893279543,
      "grad_norm": 0.11294026672840118,
      "learning_rate": 7.040383266562846e-05,
      "loss": 1.5969,
      "step": 151424
    },
    {
      "epoch": 0.29886983528935074,
      "grad_norm": 0.10951151698827744,
      "learning_rate": 7.039749244126376e-05,
      "loss": 1.6099,
      "step": 151456
    },
    {
      "epoch": 0.2989329812507472,
      "grad_norm": 0.11363144218921661,
      "learning_rate": 7.039115221689908e-05,
      "loss": 1.6062,
      "step": 151488
    },
    {
      "epoch": 0.29899612721214364,
      "grad_norm": 0.10769317299127579,
      "learning_rate": 7.038481199253439e-05,
      "loss": 1.6066,
      "step": 151520
    },
    {
      "epoch": 0.2990592731735401,
      "grad_norm": 0.10310589522123337,
      "learning_rate": 7.037847176816969e-05,
      "loss": 1.5996,
      "step": 151552
    },
    {
      "epoch": 0.2991224191349365,
      "grad_norm": 0.10610410571098328,
      "learning_rate": 7.0372131543805e-05,
      "loss": 1.5932,
      "step": 151584
    },
    {
      "epoch": 0.29918556509633293,
      "grad_norm": 0.11068467795848846,
      "learning_rate": 7.036579131944032e-05,
      "loss": 1.6125,
      "step": 151616
    },
    {
      "epoch": 0.2992487110577294,
      "grad_norm": 0.10848487168550491,
      "learning_rate": 7.035945109507564e-05,
      "loss": 1.5865,
      "step": 151648
    },
    {
      "epoch": 0.2993118570191258,
      "grad_norm": 0.10966099798679352,
      "learning_rate": 7.035311087071094e-05,
      "loss": 1.5881,
      "step": 151680
    },
    {
      "epoch": 0.2993750029805223,
      "grad_norm": 0.10770150274038315,
      "learning_rate": 7.034677064634625e-05,
      "loss": 1.6034,
      "step": 151712
    },
    {
      "epoch": 0.2994381489419187,
      "grad_norm": 0.10797891765832901,
      "learning_rate": 7.034043042198157e-05,
      "loss": 1.5958,
      "step": 151744
    },
    {
      "epoch": 0.2995012949033152,
      "grad_norm": 0.11336114257574081,
      "learning_rate": 7.033409019761688e-05,
      "loss": 1.6104,
      "step": 151776
    },
    {
      "epoch": 0.29956444086471157,
      "grad_norm": 0.10820356011390686,
      "learning_rate": 7.032774997325218e-05,
      "loss": 1.6075,
      "step": 151808
    },
    {
      "epoch": 0.299627586826108,
      "grad_norm": 0.10891455411911011,
      "learning_rate": 7.03214097488875e-05,
      "loss": 1.6026,
      "step": 151840
    },
    {
      "epoch": 0.29969073278750447,
      "grad_norm": 0.10543946176767349,
      "learning_rate": 7.03150695245228e-05,
      "loss": 1.5985,
      "step": 151872
    },
    {
      "epoch": 0.2997538787489009,
      "grad_norm": 0.1041947528719902,
      "learning_rate": 7.030872930015811e-05,
      "loss": 1.5802,
      "step": 151904
    },
    {
      "epoch": 0.29981702471029736,
      "grad_norm": 0.10467489063739777,
      "learning_rate": 7.030238907579341e-05,
      "loss": 1.5884,
      "step": 151936
    },
    {
      "epoch": 0.2998801706716938,
      "grad_norm": 0.11704521626234055,
      "learning_rate": 7.029604885142873e-05,
      "loss": 1.5988,
      "step": 151968
    },
    {
      "epoch": 0.2999433166330902,
      "grad_norm": 0.10920777171850204,
      "learning_rate": 7.028970862706404e-05,
      "loss": 1.5922,
      "step": 152000
    },
    {
      "epoch": 0.30000646259448666,
      "grad_norm": 0.10438364744186401,
      "learning_rate": 7.028336840269936e-05,
      "loss": 1.5932,
      "step": 152032
    },
    {
      "epoch": 0.3000696085558831,
      "grad_norm": 0.1106170192360878,
      "learning_rate": 7.027702817833467e-05,
      "loss": 1.601,
      "step": 152064
    },
    {
      "epoch": 0.30013275451727955,
      "grad_norm": 0.10635095089673996,
      "learning_rate": 7.027068795396997e-05,
      "loss": 1.6003,
      "step": 152096
    },
    {
      "epoch": 0.300195900478676,
      "grad_norm": 0.11621138453483582,
      "learning_rate": 7.026434772960529e-05,
      "loss": 1.6044,
      "step": 152128
    },
    {
      "epoch": 0.30025904644007245,
      "grad_norm": 0.11256194114685059,
      "learning_rate": 7.02580075052406e-05,
      "loss": 1.6098,
      "step": 152160
    },
    {
      "epoch": 0.30032219240146885,
      "grad_norm": 0.11211389303207397,
      "learning_rate": 7.025166728087592e-05,
      "loss": 1.5969,
      "step": 152192
    },
    {
      "epoch": 0.3003853383628653,
      "grad_norm": 0.10912995040416718,
      "learning_rate": 7.024532705651122e-05,
      "loss": 1.606,
      "step": 152224
    },
    {
      "epoch": 0.30044848432426174,
      "grad_norm": 0.12073302268981934,
      "learning_rate": 7.023898683214653e-05,
      "loss": 1.5912,
      "step": 152256
    },
    {
      "epoch": 0.3005116302856582,
      "grad_norm": 0.11587313562631607,
      "learning_rate": 7.023264660778183e-05,
      "loss": 1.6032,
      "step": 152288
    },
    {
      "epoch": 0.30057477624705464,
      "grad_norm": 0.11150822043418884,
      "learning_rate": 7.022630638341715e-05,
      "loss": 1.5906,
      "step": 152320
    },
    {
      "epoch": 0.3006379222084511,
      "grad_norm": 0.11763504892587662,
      "learning_rate": 7.021996615905245e-05,
      "loss": 1.5941,
      "step": 152352
    },
    {
      "epoch": 0.3007010681698475,
      "grad_norm": 0.10895249992609024,
      "learning_rate": 7.021362593468776e-05,
      "loss": 1.6021,
      "step": 152384
    },
    {
      "epoch": 0.30076421413124393,
      "grad_norm": 0.10470119118690491,
      "learning_rate": 7.020728571032308e-05,
      "loss": 1.6011,
      "step": 152416
    },
    {
      "epoch": 0.3008273600926404,
      "grad_norm": 0.11427845060825348,
      "learning_rate": 7.020094548595839e-05,
      "loss": 1.6003,
      "step": 152448
    },
    {
      "epoch": 0.30089050605403683,
      "grad_norm": 0.10430949926376343,
      "learning_rate": 7.019460526159369e-05,
      "loss": 1.6075,
      "step": 152480
    },
    {
      "epoch": 0.3009536520154333,
      "grad_norm": 0.11266857385635376,
      "learning_rate": 7.0188265037229e-05,
      "loss": 1.6047,
      "step": 152512
    },
    {
      "epoch": 0.30101679797682973,
      "grad_norm": 0.1131991520524025,
      "learning_rate": 7.018192481286432e-05,
      "loss": 1.6033,
      "step": 152544
    },
    {
      "epoch": 0.3010799439382261,
      "grad_norm": 0.11302390694618225,
      "learning_rate": 7.017558458849964e-05,
      "loss": 1.6041,
      "step": 152576
    },
    {
      "epoch": 0.3011430898996226,
      "grad_norm": 0.11738061904907227,
      "learning_rate": 7.016924436413494e-05,
      "loss": 1.6032,
      "step": 152608
    },
    {
      "epoch": 0.301206235861019,
      "grad_norm": 0.10957682877779007,
      "learning_rate": 7.016290413977025e-05,
      "loss": 1.5962,
      "step": 152640
    },
    {
      "epoch": 0.30126938182241547,
      "grad_norm": 0.11225590854883194,
      "learning_rate": 7.015656391540557e-05,
      "loss": 1.5921,
      "step": 152672
    },
    {
      "epoch": 0.3013325277838119,
      "grad_norm": 0.11029886454343796,
      "learning_rate": 7.015022369104087e-05,
      "loss": 1.6007,
      "step": 152704
    },
    {
      "epoch": 0.30139567374520837,
      "grad_norm": 0.1058514267206192,
      "learning_rate": 7.014388346667618e-05,
      "loss": 1.6044,
      "step": 152736
    },
    {
      "epoch": 0.30145881970660476,
      "grad_norm": 0.10820958018302917,
      "learning_rate": 7.013754324231148e-05,
      "loss": 1.6066,
      "step": 152768
    },
    {
      "epoch": 0.3015219656680012,
      "grad_norm": 0.10125933587551117,
      "learning_rate": 7.01312030179468e-05,
      "loss": 1.5946,
      "step": 152800
    },
    {
      "epoch": 0.30158511162939766,
      "grad_norm": 0.109490305185318,
      "learning_rate": 7.012486279358211e-05,
      "loss": 1.6039,
      "step": 152832
    },
    {
      "epoch": 0.3016482575907941,
      "grad_norm": 0.1123470664024353,
      "learning_rate": 7.011852256921743e-05,
      "loss": 1.6029,
      "step": 152864
    },
    {
      "epoch": 0.30171140355219056,
      "grad_norm": 0.10927826911211014,
      "learning_rate": 7.011218234485273e-05,
      "loss": 1.6048,
      "step": 152896
    },
    {
      "epoch": 0.301774549513587,
      "grad_norm": 0.10237983614206314,
      "learning_rate": 7.010584212048804e-05,
      "loss": 1.594,
      "step": 152928
    },
    {
      "epoch": 0.3018376954749834,
      "grad_norm": 0.11437680572271347,
      "learning_rate": 7.009950189612336e-05,
      "loss": 1.5965,
      "step": 152960
    },
    {
      "epoch": 0.30190084143637985,
      "grad_norm": 0.10745229572057724,
      "learning_rate": 7.009316167175867e-05,
      "loss": 1.6127,
      "step": 152992
    },
    {
      "epoch": 0.3019639873977763,
      "grad_norm": 0.10919062793254852,
      "learning_rate": 7.008682144739397e-05,
      "loss": 1.5869,
      "step": 153024
    },
    {
      "epoch": 0.30202713335917275,
      "grad_norm": 0.1046941876411438,
      "learning_rate": 7.008048122302929e-05,
      "loss": 1.5856,
      "step": 153056
    },
    {
      "epoch": 0.3020902793205692,
      "grad_norm": 0.10780448466539383,
      "learning_rate": 7.00741409986646e-05,
      "loss": 1.5898,
      "step": 153088
    },
    {
      "epoch": 0.30215342528196565,
      "grad_norm": 0.10609892755746841,
      "learning_rate": 7.00678007742999e-05,
      "loss": 1.5981,
      "step": 153120
    },
    {
      "epoch": 0.30221657124336204,
      "grad_norm": 0.10885103791952133,
      "learning_rate": 7.00614605499352e-05,
      "loss": 1.5931,
      "step": 153152
    },
    {
      "epoch": 0.3022797172047585,
      "grad_norm": 0.10555577278137207,
      "learning_rate": 7.005512032557052e-05,
      "loss": 1.5907,
      "step": 153184
    },
    {
      "epoch": 0.30234286316615494,
      "grad_norm": 0.10858432948589325,
      "learning_rate": 7.004878010120583e-05,
      "loss": 1.5931,
      "step": 153216
    },
    {
      "epoch": 0.3024060091275514,
      "grad_norm": 0.12543374300003052,
      "learning_rate": 7.004243987684115e-05,
      "loss": 1.5937,
      "step": 153248
    },
    {
      "epoch": 0.30246915508894784,
      "grad_norm": 0.10381802171468735,
      "learning_rate": 7.003609965247645e-05,
      "loss": 1.5938,
      "step": 153280
    },
    {
      "epoch": 0.3025323010503443,
      "grad_norm": 0.12593108415603638,
      "learning_rate": 7.002975942811176e-05,
      "loss": 1.5886,
      "step": 153312
    },
    {
      "epoch": 0.3025954470117407,
      "grad_norm": 0.12112244218587875,
      "learning_rate": 7.002341920374708e-05,
      "loss": 1.6135,
      "step": 153344
    },
    {
      "epoch": 0.30265859297313713,
      "grad_norm": 0.10223555564880371,
      "learning_rate": 7.001707897938239e-05,
      "loss": 1.5847,
      "step": 153376
    },
    {
      "epoch": 0.3027217389345336,
      "grad_norm": 0.11031569540500641,
      "learning_rate": 7.00107387550177e-05,
      "loss": 1.5877,
      "step": 153408
    },
    {
      "epoch": 0.30278488489593003,
      "grad_norm": 0.10166694968938828,
      "learning_rate": 7.000439853065301e-05,
      "loss": 1.6019,
      "step": 153440
    },
    {
      "epoch": 0.3028480308573265,
      "grad_norm": 0.10694776475429535,
      "learning_rate": 6.999805830628832e-05,
      "loss": 1.6039,
      "step": 153472
    },
    {
      "epoch": 0.3029111768187229,
      "grad_norm": 0.11236537992954254,
      "learning_rate": 6.999171808192364e-05,
      "loss": 1.602,
      "step": 153504
    },
    {
      "epoch": 0.3029743227801194,
      "grad_norm": 0.11643679440021515,
      "learning_rate": 6.998537785755894e-05,
      "loss": 1.6087,
      "step": 153536
    },
    {
      "epoch": 0.30303746874151577,
      "grad_norm": 0.1105060875415802,
      "learning_rate": 6.997903763319424e-05,
      "loss": 1.6024,
      "step": 153568
    },
    {
      "epoch": 0.3031006147029122,
      "grad_norm": 0.11364857852458954,
      "learning_rate": 6.997269740882955e-05,
      "loss": 1.6041,
      "step": 153600
    },
    {
      "epoch": 0.30316376066430867,
      "grad_norm": 0.11282741278409958,
      "learning_rate": 6.996635718446487e-05,
      "loss": 1.6065,
      "step": 153632
    },
    {
      "epoch": 0.3032269066257051,
      "grad_norm": 0.11782099306583405,
      "learning_rate": 6.996001696010018e-05,
      "loss": 1.6092,
      "step": 153664
    },
    {
      "epoch": 0.30329005258710157,
      "grad_norm": 0.10855941474437714,
      "learning_rate": 6.995367673573548e-05,
      "loss": 1.6017,
      "step": 153696
    },
    {
      "epoch": 0.303353198548498,
      "grad_norm": 0.11252662539482117,
      "learning_rate": 6.99473365113708e-05,
      "loss": 1.6067,
      "step": 153728
    },
    {
      "epoch": 0.3034163445098944,
      "grad_norm": 0.11697684228420258,
      "learning_rate": 6.994099628700611e-05,
      "loss": 1.5947,
      "step": 153760
    },
    {
      "epoch": 0.30347949047129086,
      "grad_norm": 0.10977371037006378,
      "learning_rate": 6.993465606264143e-05,
      "loss": 1.6086,
      "step": 153792
    },
    {
      "epoch": 0.3035426364326873,
      "grad_norm": 0.10438380390405655,
      "learning_rate": 6.992831583827673e-05,
      "loss": 1.5985,
      "step": 153824
    },
    {
      "epoch": 0.30360578239408376,
      "grad_norm": 0.11717565357685089,
      "learning_rate": 6.992197561391204e-05,
      "loss": 1.607,
      "step": 153856
    },
    {
      "epoch": 0.3036689283554802,
      "grad_norm": 0.11341336369514465,
      "learning_rate": 6.991563538954736e-05,
      "loss": 1.5939,
      "step": 153888
    },
    {
      "epoch": 0.30373207431687665,
      "grad_norm": 0.10679785162210464,
      "learning_rate": 6.990929516518266e-05,
      "loss": 1.5883,
      "step": 153920
    },
    {
      "epoch": 0.30379522027827305,
      "grad_norm": 0.100877545773983,
      "learning_rate": 6.990295494081797e-05,
      "loss": 1.5899,
      "step": 153952
    },
    {
      "epoch": 0.3038583662396695,
      "grad_norm": 0.10690692812204361,
      "learning_rate": 6.989661471645327e-05,
      "loss": 1.614,
      "step": 153984
    },
    {
      "epoch": 0.30392151220106595,
      "grad_norm": 0.10297846049070358,
      "learning_rate": 6.989027449208859e-05,
      "loss": 1.6168,
      "step": 154016
    },
    {
      "epoch": 0.3039846581624624,
      "grad_norm": 0.10769422352313995,
      "learning_rate": 6.98839342677239e-05,
      "loss": 1.5998,
      "step": 154048
    },
    {
      "epoch": 0.30404780412385884,
      "grad_norm": 0.11395012587308884,
      "learning_rate": 6.987759404335922e-05,
      "loss": 1.5858,
      "step": 154080
    },
    {
      "epoch": 0.3041109500852553,
      "grad_norm": 0.110221728682518,
      "learning_rate": 6.987125381899452e-05,
      "loss": 1.6014,
      "step": 154112
    },
    {
      "epoch": 0.3041740960466517,
      "grad_norm": 0.10333923995494843,
      "learning_rate": 6.986491359462983e-05,
      "loss": 1.5963,
      "step": 154144
    },
    {
      "epoch": 0.30423724200804814,
      "grad_norm": 0.11066845059394836,
      "learning_rate": 6.985857337026515e-05,
      "loss": 1.5952,
      "step": 154176
    },
    {
      "epoch": 0.3043003879694446,
      "grad_norm": 0.1154382973909378,
      "learning_rate": 6.985223314590046e-05,
      "loss": 1.5841,
      "step": 154208
    },
    {
      "epoch": 0.30436353393084103,
      "grad_norm": 0.10902144014835358,
      "learning_rate": 6.984589292153576e-05,
      "loss": 1.5901,
      "step": 154240
    },
    {
      "epoch": 0.3044266798922375,
      "grad_norm": 0.1055220514535904,
      "learning_rate": 6.983955269717108e-05,
      "loss": 1.6,
      "step": 154272
    },
    {
      "epoch": 0.30448982585363393,
      "grad_norm": 0.10117088258266449,
      "learning_rate": 6.983321247280639e-05,
      "loss": 1.5832,
      "step": 154304
    },
    {
      "epoch": 0.3045529718150303,
      "grad_norm": 0.12125889211893082,
      "learning_rate": 6.982687224844169e-05,
      "loss": 1.5974,
      "step": 154336
    },
    {
      "epoch": 0.3046161177764268,
      "grad_norm": 0.11177989840507507,
      "learning_rate": 6.982053202407701e-05,
      "loss": 1.5952,
      "step": 154368
    },
    {
      "epoch": 0.3046792637378232,
      "grad_norm": 0.10721558332443237,
      "learning_rate": 6.981419179971231e-05,
      "loss": 1.6077,
      "step": 154400
    },
    {
      "epoch": 0.3047424096992197,
      "grad_norm": 0.11649288237094879,
      "learning_rate": 6.980785157534762e-05,
      "loss": 1.6024,
      "step": 154432
    },
    {
      "epoch": 0.3048055556606161,
      "grad_norm": 0.11103370040655136,
      "learning_rate": 6.980151135098294e-05,
      "loss": 1.5923,
      "step": 154464
    },
    {
      "epoch": 0.3048687016220126,
      "grad_norm": 0.10063661634922028,
      "learning_rate": 6.979517112661824e-05,
      "loss": 1.6024,
      "step": 154496
    },
    {
      "epoch": 0.30493184758340897,
      "grad_norm": 0.10404430329799652,
      "learning_rate": 6.978883090225355e-05,
      "loss": 1.6124,
      "step": 154528
    },
    {
      "epoch": 0.3049949935448054,
      "grad_norm": 0.11656454205513,
      "learning_rate": 6.978249067788887e-05,
      "loss": 1.6003,
      "step": 154560
    },
    {
      "epoch": 0.30505813950620186,
      "grad_norm": 0.10750927776098251,
      "learning_rate": 6.977615045352418e-05,
      "loss": 1.5853,
      "step": 154592
    },
    {
      "epoch": 0.3051212854675983,
      "grad_norm": 0.11315322667360306,
      "learning_rate": 6.976981022915948e-05,
      "loss": 1.5985,
      "step": 154624
    },
    {
      "epoch": 0.30518443142899476,
      "grad_norm": 0.10713580995798111,
      "learning_rate": 6.97634700047948e-05,
      "loss": 1.5903,
      "step": 154656
    },
    {
      "epoch": 0.3052475773903912,
      "grad_norm": 0.10717027634382248,
      "learning_rate": 6.975712978043011e-05,
      "loss": 1.5922,
      "step": 154688
    },
    {
      "epoch": 0.3053107233517876,
      "grad_norm": 0.11209490150213242,
      "learning_rate": 6.975078955606543e-05,
      "loss": 1.5918,
      "step": 154720
    },
    {
      "epoch": 0.30537386931318405,
      "grad_norm": 0.10829471796751022,
      "learning_rate": 6.974444933170073e-05,
      "loss": 1.5994,
      "step": 154752
    },
    {
      "epoch": 0.3054370152745805,
      "grad_norm": 0.11069749295711517,
      "learning_rate": 6.973810910733604e-05,
      "loss": 1.6018,
      "step": 154784
    },
    {
      "epoch": 0.30550016123597695,
      "grad_norm": 0.11267156153917313,
      "learning_rate": 6.973176888297134e-05,
      "loss": 1.5934,
      "step": 154816
    },
    {
      "epoch": 0.3055633071973734,
      "grad_norm": 0.11649084091186523,
      "learning_rate": 6.972542865860666e-05,
      "loss": 1.5759,
      "step": 154848
    },
    {
      "epoch": 0.30562645315876985,
      "grad_norm": 0.1033872663974762,
      "learning_rate": 6.971908843424197e-05,
      "loss": 1.5999,
      "step": 154880
    },
    {
      "epoch": 0.30568959912016624,
      "grad_norm": 0.10910879075527191,
      "learning_rate": 6.971274820987727e-05,
      "loss": 1.591,
      "step": 154912
    },
    {
      "epoch": 0.3057527450815627,
      "grad_norm": 0.1053755134344101,
      "learning_rate": 6.970640798551259e-05,
      "loss": 1.5868,
      "step": 154944
    },
    {
      "epoch": 0.30581589104295914,
      "grad_norm": 0.11260037869215012,
      "learning_rate": 6.97000677611479e-05,
      "loss": 1.6118,
      "step": 154976
    },
    {
      "epoch": 0.3058790370043556,
      "grad_norm": 0.10938689112663269,
      "learning_rate": 6.969372753678322e-05,
      "loss": 1.6104,
      "step": 155008
    },
    {
      "epoch": 0.30594218296575204,
      "grad_norm": 0.10422463715076447,
      "learning_rate": 6.968738731241852e-05,
      "loss": 1.598,
      "step": 155040
    },
    {
      "epoch": 0.3060053289271485,
      "grad_norm": 0.10344479233026505,
      "learning_rate": 6.968104708805383e-05,
      "loss": 1.6079,
      "step": 155072
    },
    {
      "epoch": 0.30606847488854494,
      "grad_norm": 0.11008403450250626,
      "learning_rate": 6.967470686368915e-05,
      "loss": 1.6025,
      "step": 155104
    },
    {
      "epoch": 0.30613162084994133,
      "grad_norm": 0.11175358295440674,
      "learning_rate": 6.966836663932446e-05,
      "loss": 1.5771,
      "step": 155136
    },
    {
      "epoch": 0.3061947668113378,
      "grad_norm": 0.10845853388309479,
      "learning_rate": 6.966202641495976e-05,
      "loss": 1.5986,
      "step": 155168
    },
    {
      "epoch": 0.30625791277273423,
      "grad_norm": 0.11384795606136322,
      "learning_rate": 6.965568619059508e-05,
      "loss": 1.5969,
      "step": 155200
    },
    {
      "epoch": 0.3063210587341307,
      "grad_norm": 0.11259113997220993,
      "learning_rate": 6.964934596623038e-05,
      "loss": 1.5955,
      "step": 155232
    },
    {
      "epoch": 0.30638420469552713,
      "grad_norm": 0.10678324103355408,
      "learning_rate": 6.964300574186569e-05,
      "loss": 1.598,
      "step": 155264
    },
    {
      "epoch": 0.3064473506569236,
      "grad_norm": 0.11580517888069153,
      "learning_rate": 6.9636665517501e-05,
      "loss": 1.5881,
      "step": 155296
    },
    {
      "epoch": 0.30651049661831997,
      "grad_norm": 0.1062786802649498,
      "learning_rate": 6.963032529313631e-05,
      "loss": 1.5901,
      "step": 155328
    },
    {
      "epoch": 0.3065736425797164,
      "grad_norm": 0.1118519976735115,
      "learning_rate": 6.962398506877162e-05,
      "loss": 1.5993,
      "step": 155360
    },
    {
      "epoch": 0.30663678854111287,
      "grad_norm": 0.10496975481510162,
      "learning_rate": 6.961764484440694e-05,
      "loss": 1.5917,
      "step": 155392
    },
    {
      "epoch": 0.3066999345025093,
      "grad_norm": 0.11165449023246765,
      "learning_rate": 6.961130462004225e-05,
      "loss": 1.5978,
      "step": 155424
    },
    {
      "epoch": 0.30676308046390577,
      "grad_norm": 0.11271575838327408,
      "learning_rate": 6.960496439567755e-05,
      "loss": 1.5939,
      "step": 155456
    },
    {
      "epoch": 0.3068262264253022,
      "grad_norm": 0.11157065629959106,
      "learning_rate": 6.959862417131287e-05,
      "loss": 1.5995,
      "step": 155488
    },
    {
      "epoch": 0.3068893723866986,
      "grad_norm": 0.10466542840003967,
      "learning_rate": 6.959228394694818e-05,
      "loss": 1.6123,
      "step": 155520
    },
    {
      "epoch": 0.30695251834809506,
      "grad_norm": 0.10684055089950562,
      "learning_rate": 6.95859437225835e-05,
      "loss": 1.5835,
      "step": 155552
    },
    {
      "epoch": 0.3070156643094915,
      "grad_norm": 0.11516595631837845,
      "learning_rate": 6.95796034982188e-05,
      "loss": 1.6044,
      "step": 155584
    },
    {
      "epoch": 0.30707881027088796,
      "grad_norm": 0.10631965845823288,
      "learning_rate": 6.95732632738541e-05,
      "loss": 1.6077,
      "step": 155616
    },
    {
      "epoch": 0.3071419562322844,
      "grad_norm": 0.10585325956344604,
      "learning_rate": 6.956692304948941e-05,
      "loss": 1.5988,
      "step": 155648
    },
    {
      "epoch": 0.30720510219368086,
      "grad_norm": 0.10833197087049484,
      "learning_rate": 6.956058282512473e-05,
      "loss": 1.5951,
      "step": 155680
    },
    {
      "epoch": 0.30726824815507725,
      "grad_norm": 0.11436755955219269,
      "learning_rate": 6.955424260076003e-05,
      "loss": 1.6001,
      "step": 155712
    },
    {
      "epoch": 0.3073313941164737,
      "grad_norm": 0.10563050210475922,
      "learning_rate": 6.954790237639534e-05,
      "loss": 1.5977,
      "step": 155744
    },
    {
      "epoch": 0.30739454007787015,
      "grad_norm": 0.10954467207193375,
      "learning_rate": 6.954156215203066e-05,
      "loss": 1.5903,
      "step": 155776
    },
    {
      "epoch": 0.3074576860392666,
      "grad_norm": 0.1087346225976944,
      "learning_rate": 6.953522192766597e-05,
      "loss": 1.6036,
      "step": 155808
    },
    {
      "epoch": 0.30752083200066305,
      "grad_norm": 0.1167093887925148,
      "learning_rate": 6.952888170330127e-05,
      "loss": 1.5952,
      "step": 155840
    },
    {
      "epoch": 0.3075839779620595,
      "grad_norm": 0.11237960308790207,
      "learning_rate": 6.952254147893659e-05,
      "loss": 1.597,
      "step": 155872
    },
    {
      "epoch": 0.3076471239234559,
      "grad_norm": 0.10673242807388306,
      "learning_rate": 6.95162012545719e-05,
      "loss": 1.5953,
      "step": 155904
    },
    {
      "epoch": 0.30771026988485234,
      "grad_norm": 0.11010044068098068,
      "learning_rate": 6.950986103020722e-05,
      "loss": 1.5848,
      "step": 155936
    },
    {
      "epoch": 0.3077734158462488,
      "grad_norm": 0.1044807955622673,
      "learning_rate": 6.950352080584252e-05,
      "loss": 1.5741,
      "step": 155968
    },
    {
      "epoch": 0.30783656180764524,
      "grad_norm": 0.10707549750804901,
      "learning_rate": 6.949718058147783e-05,
      "loss": 1.5924,
      "step": 156000
    },
    {
      "epoch": 0.3078997077690417,
      "grad_norm": 0.10744640976190567,
      "learning_rate": 6.949084035711313e-05,
      "loss": 1.6009,
      "step": 156032
    },
    {
      "epoch": 0.30796285373043814,
      "grad_norm": 0.10549584776163101,
      "learning_rate": 6.948450013274845e-05,
      "loss": 1.5933,
      "step": 156064
    },
    {
      "epoch": 0.30802599969183453,
      "grad_norm": 0.1125851422548294,
      "learning_rate": 6.947815990838376e-05,
      "loss": 1.5983,
      "step": 156096
    },
    {
      "epoch": 0.308089145653231,
      "grad_norm": 0.11948680132627487,
      "learning_rate": 6.947181968401906e-05,
      "loss": 1.597,
      "step": 156128
    },
    {
      "epoch": 0.3081522916146274,
      "grad_norm": 0.10488571226596832,
      "learning_rate": 6.946547945965438e-05,
      "loss": 1.5984,
      "step": 156160
    },
    {
      "epoch": 0.3082154375760239,
      "grad_norm": 0.10445652157068253,
      "learning_rate": 6.94591392352897e-05,
      "loss": 1.5849,
      "step": 156192
    },
    {
      "epoch": 0.3082785835374203,
      "grad_norm": 0.11123156547546387,
      "learning_rate": 6.945279901092501e-05,
      "loss": 1.5971,
      "step": 156224
    },
    {
      "epoch": 0.3083417294988168,
      "grad_norm": 0.1111505851149559,
      "learning_rate": 6.944645878656031e-05,
      "loss": 1.5985,
      "step": 156256
    },
    {
      "epoch": 0.30840487546021317,
      "grad_norm": 0.10809691250324249,
      "learning_rate": 6.944011856219562e-05,
      "loss": 1.5918,
      "step": 156288
    },
    {
      "epoch": 0.3084680214216096,
      "grad_norm": 0.10672581195831299,
      "learning_rate": 6.943377833783094e-05,
      "loss": 1.5942,
      "step": 156320
    },
    {
      "epoch": 0.30853116738300607,
      "grad_norm": 0.10128645598888397,
      "learning_rate": 6.942743811346625e-05,
      "loss": 1.5874,
      "step": 156352
    },
    {
      "epoch": 0.3085943133444025,
      "grad_norm": 0.1070328876376152,
      "learning_rate": 6.942109788910155e-05,
      "loss": 1.5951,
      "step": 156384
    },
    {
      "epoch": 0.30865745930579896,
      "grad_norm": 0.1111120730638504,
      "learning_rate": 6.941475766473687e-05,
      "loss": 1.6038,
      "step": 156416
    },
    {
      "epoch": 0.3087206052671954,
      "grad_norm": 0.10825317353010178,
      "learning_rate": 6.940841744037217e-05,
      "loss": 1.6056,
      "step": 156448
    },
    {
      "epoch": 0.3087837512285918,
      "grad_norm": 0.11818309873342514,
      "learning_rate": 6.940207721600748e-05,
      "loss": 1.5906,
      "step": 156480
    },
    {
      "epoch": 0.30884689718998826,
      "grad_norm": 0.11527826637029648,
      "learning_rate": 6.939573699164278e-05,
      "loss": 1.5933,
      "step": 156512
    },
    {
      "epoch": 0.3089100431513847,
      "grad_norm": 0.11291716992855072,
      "learning_rate": 6.93893967672781e-05,
      "loss": 1.5826,
      "step": 156544
    },
    {
      "epoch": 0.30897318911278115,
      "grad_norm": 0.10666730254888535,
      "learning_rate": 6.938305654291341e-05,
      "loss": 1.586,
      "step": 156576
    },
    {
      "epoch": 0.3090363350741776,
      "grad_norm": 0.10332287847995758,
      "learning_rate": 6.937671631854873e-05,
      "loss": 1.595,
      "step": 156608
    },
    {
      "epoch": 0.30909948103557405,
      "grad_norm": 0.1033509150147438,
      "learning_rate": 6.937037609418403e-05,
      "loss": 1.6067,
      "step": 156640
    },
    {
      "epoch": 0.30916262699697045,
      "grad_norm": 0.10919076949357986,
      "learning_rate": 6.936403586981934e-05,
      "loss": 1.604,
      "step": 156672
    },
    {
      "epoch": 0.3092257729583669,
      "grad_norm": 0.10804831236600876,
      "learning_rate": 6.935769564545466e-05,
      "loss": 1.5966,
      "step": 156704
    },
    {
      "epoch": 0.30928891891976334,
      "grad_norm": 0.11538102477788925,
      "learning_rate": 6.935135542108997e-05,
      "loss": 1.5889,
      "step": 156736
    },
    {
      "epoch": 0.3093520648811598,
      "grad_norm": 0.10929667949676514,
      "learning_rate": 6.934501519672529e-05,
      "loss": 1.6088,
      "step": 156768
    },
    {
      "epoch": 0.30941521084255624,
      "grad_norm": 0.10778364539146423,
      "learning_rate": 6.933867497236059e-05,
      "loss": 1.6032,
      "step": 156800
    },
    {
      "epoch": 0.3094783568039527,
      "grad_norm": 0.12172455340623856,
      "learning_rate": 6.93323347479959e-05,
      "loss": 1.5986,
      "step": 156832
    },
    {
      "epoch": 0.30954150276534914,
      "grad_norm": 0.10607221722602844,
      "learning_rate": 6.93259945236312e-05,
      "loss": 1.6037,
      "step": 156864
    },
    {
      "epoch": 0.30960464872674553,
      "grad_norm": 0.10583367943763733,
      "learning_rate": 6.931965429926652e-05,
      "loss": 1.5868,
      "step": 156896
    },
    {
      "epoch": 0.309667794688142,
      "grad_norm": 0.10794432461261749,
      "learning_rate": 6.931331407490182e-05,
      "loss": 1.5992,
      "step": 156928
    },
    {
      "epoch": 0.30973094064953843,
      "grad_norm": 0.10778047889471054,
      "learning_rate": 6.930697385053713e-05,
      "loss": 1.5858,
      "step": 156960
    },
    {
      "epoch": 0.3097940866109349,
      "grad_norm": 0.11203970015048981,
      "learning_rate": 6.930063362617245e-05,
      "loss": 1.596,
      "step": 156992
    },
    {
      "epoch": 0.30985723257233133,
      "grad_norm": 0.11195708066225052,
      "learning_rate": 6.929429340180776e-05,
      "loss": 1.6011,
      "step": 157024
    },
    {
      "epoch": 0.3099203785337278,
      "grad_norm": 0.11102274805307388,
      "learning_rate": 6.928795317744306e-05,
      "loss": 1.596,
      "step": 157056
    },
    {
      "epoch": 0.3099835244951242,
      "grad_norm": 0.1033940240740776,
      "learning_rate": 6.928161295307838e-05,
      "loss": 1.5889,
      "step": 157088
    },
    {
      "epoch": 0.3100466704565206,
      "grad_norm": 0.11119741946458817,
      "learning_rate": 6.92752727287137e-05,
      "loss": 1.5916,
      "step": 157120
    },
    {
      "epoch": 0.3101098164179171,
      "grad_norm": 0.11409356445074081,
      "learning_rate": 6.926893250434901e-05,
      "loss": 1.6053,
      "step": 157152
    },
    {
      "epoch": 0.3101729623793135,
      "grad_norm": 0.1127903163433075,
      "learning_rate": 6.926259227998431e-05,
      "loss": 1.5831,
      "step": 157184
    },
    {
      "epoch": 0.31023610834070997,
      "grad_norm": 0.11782166361808777,
      "learning_rate": 6.925625205561962e-05,
      "loss": 1.595,
      "step": 157216
    },
    {
      "epoch": 0.3102992543021064,
      "grad_norm": 0.11755773425102234,
      "learning_rate": 6.924991183125494e-05,
      "loss": 1.6002,
      "step": 157248
    },
    {
      "epoch": 0.3103624002635028,
      "grad_norm": 0.10300610214471817,
      "learning_rate": 6.924357160689024e-05,
      "loss": 1.5965,
      "step": 157280
    },
    {
      "epoch": 0.31042554622489926,
      "grad_norm": 0.10954927653074265,
      "learning_rate": 6.923723138252554e-05,
      "loss": 1.5879,
      "step": 157312
    },
    {
      "epoch": 0.3104886921862957,
      "grad_norm": 0.10760588198900223,
      "learning_rate": 6.923089115816086e-05,
      "loss": 1.5973,
      "step": 157344
    },
    {
      "epoch": 0.31055183814769216,
      "grad_norm": 0.1067931205034256,
      "learning_rate": 6.922455093379617e-05,
      "loss": 1.5779,
      "step": 157376
    },
    {
      "epoch": 0.3106149841090886,
      "grad_norm": 0.1142549142241478,
      "learning_rate": 6.921821070943148e-05,
      "loss": 1.5802,
      "step": 157408
    },
    {
      "epoch": 0.31067813007048506,
      "grad_norm": 0.11532670259475708,
      "learning_rate": 6.92118704850668e-05,
      "loss": 1.6029,
      "step": 157440
    },
    {
      "epoch": 0.31074127603188145,
      "grad_norm": 0.10625465214252472,
      "learning_rate": 6.92055302607021e-05,
      "loss": 1.5931,
      "step": 157472
    },
    {
      "epoch": 0.3108044219932779,
      "grad_norm": 0.10834839940071106,
      "learning_rate": 6.919919003633741e-05,
      "loss": 1.6007,
      "step": 157504
    },
    {
      "epoch": 0.31086756795467435,
      "grad_norm": 0.11619570106267929,
      "learning_rate": 6.919284981197273e-05,
      "loss": 1.5911,
      "step": 157536
    },
    {
      "epoch": 0.3109307139160708,
      "grad_norm": 0.1049533560872078,
      "learning_rate": 6.918650958760804e-05,
      "loss": 1.5836,
      "step": 157568
    },
    {
      "epoch": 0.31099385987746725,
      "grad_norm": 0.10432977974414825,
      "learning_rate": 6.918016936324334e-05,
      "loss": 1.5901,
      "step": 157600
    },
    {
      "epoch": 0.3110570058388637,
      "grad_norm": 0.11452938616275787,
      "learning_rate": 6.917382913887866e-05,
      "loss": 1.6011,
      "step": 157632
    },
    {
      "epoch": 0.3111201518002601,
      "grad_norm": 0.10731777548789978,
      "learning_rate": 6.916748891451397e-05,
      "loss": 1.5998,
      "step": 157664
    },
    {
      "epoch": 0.31118329776165654,
      "grad_norm": 0.1118592843413353,
      "learning_rate": 6.916114869014927e-05,
      "loss": 1.6049,
      "step": 157696
    },
    {
      "epoch": 0.311246443723053,
      "grad_norm": 0.10823763161897659,
      "learning_rate": 6.915480846578458e-05,
      "loss": 1.596,
      "step": 157728
    },
    {
      "epoch": 0.31130958968444944,
      "grad_norm": 0.11033618450164795,
      "learning_rate": 6.914846824141989e-05,
      "loss": 1.5786,
      "step": 157760
    },
    {
      "epoch": 0.3113727356458459,
      "grad_norm": 0.11134444177150726,
      "learning_rate": 6.91421280170552e-05,
      "loss": 1.5877,
      "step": 157792
    },
    {
      "epoch": 0.31143588160724234,
      "grad_norm": 0.10570427775382996,
      "learning_rate": 6.913578779269052e-05,
      "loss": 1.5932,
      "step": 157824
    },
    {
      "epoch": 0.31149902756863873,
      "grad_norm": 0.10497938096523285,
      "learning_rate": 6.912944756832582e-05,
      "loss": 1.5891,
      "step": 157856
    },
    {
      "epoch": 0.3115621735300352,
      "grad_norm": 0.10699564963579178,
      "learning_rate": 6.912310734396113e-05,
      "loss": 1.5875,
      "step": 157888
    },
    {
      "epoch": 0.31162531949143163,
      "grad_norm": 0.11236100643873215,
      "learning_rate": 6.911676711959645e-05,
      "loss": 1.5887,
      "step": 157920
    },
    {
      "epoch": 0.3116884654528281,
      "grad_norm": 0.12021209299564362,
      "learning_rate": 6.911042689523176e-05,
      "loss": 1.602,
      "step": 157952
    },
    {
      "epoch": 0.3117516114142245,
      "grad_norm": 0.11001028120517731,
      "learning_rate": 6.910408667086707e-05,
      "loss": 1.6039,
      "step": 157984
    },
    {
      "epoch": 0.311814757375621,
      "grad_norm": 0.114458367228508,
      "learning_rate": 6.909774644650238e-05,
      "loss": 1.6109,
      "step": 158016
    },
    {
      "epoch": 0.31187790333701737,
      "grad_norm": 0.10677354037761688,
      "learning_rate": 6.90914062221377e-05,
      "loss": 1.5957,
      "step": 158048
    },
    {
      "epoch": 0.3119410492984138,
      "grad_norm": 0.11993306130170822,
      "learning_rate": 6.9085065997773e-05,
      "loss": 1.6017,
      "step": 158080
    },
    {
      "epoch": 0.31200419525981027,
      "grad_norm": 0.11413183063268661,
      "learning_rate": 6.907872577340831e-05,
      "loss": 1.6018,
      "step": 158112
    },
    {
      "epoch": 0.3120673412212067,
      "grad_norm": 0.11328325420618057,
      "learning_rate": 6.907238554904361e-05,
      "loss": 1.597,
      "step": 158144
    },
    {
      "epoch": 0.31213048718260317,
      "grad_norm": 0.10782735794782639,
      "learning_rate": 6.906604532467893e-05,
      "loss": 1.5913,
      "step": 158176
    },
    {
      "epoch": 0.3121936331439996,
      "grad_norm": 0.11115462332963943,
      "learning_rate": 6.905970510031424e-05,
      "loss": 1.5905,
      "step": 158208
    },
    {
      "epoch": 0.312256779105396,
      "grad_norm": 0.10908707231283188,
      "learning_rate": 6.905336487594955e-05,
      "loss": 1.5783,
      "step": 158240
    },
    {
      "epoch": 0.31231992506679246,
      "grad_norm": 0.10839131474494934,
      "learning_rate": 6.904702465158486e-05,
      "loss": 1.5801,
      "step": 158272
    },
    {
      "epoch": 0.3123830710281889,
      "grad_norm": 0.10044689476490021,
      "learning_rate": 6.904068442722017e-05,
      "loss": 1.5916,
      "step": 158304
    },
    {
      "epoch": 0.31244621698958536,
      "grad_norm": 0.11105819046497345,
      "learning_rate": 6.903434420285548e-05,
      "loss": 1.5897,
      "step": 158336
    },
    {
      "epoch": 0.3125093629509818,
      "grad_norm": 0.1068510115146637,
      "learning_rate": 6.90280039784908e-05,
      "loss": 1.5842,
      "step": 158368
    },
    {
      "epoch": 0.31257250891237826,
      "grad_norm": 0.10635650157928467,
      "learning_rate": 6.90216637541261e-05,
      "loss": 1.5992,
      "step": 158400
    },
    {
      "epoch": 0.31263565487377465,
      "grad_norm": 0.10454981029033661,
      "learning_rate": 6.901532352976141e-05,
      "loss": 1.6038,
      "step": 158432
    },
    {
      "epoch": 0.3126988008351711,
      "grad_norm": 0.11232787370681763,
      "learning_rate": 6.900898330539673e-05,
      "loss": 1.6034,
      "step": 158464
    },
    {
      "epoch": 0.31276194679656755,
      "grad_norm": 0.11920122802257538,
      "learning_rate": 6.900264308103203e-05,
      "loss": 1.5945,
      "step": 158496
    },
    {
      "epoch": 0.312825092757964,
      "grad_norm": 0.11414216458797455,
      "learning_rate": 6.899630285666734e-05,
      "loss": 1.5997,
      "step": 158528
    },
    {
      "epoch": 0.31288823871936045,
      "grad_norm": 0.1111496314406395,
      "learning_rate": 6.898996263230265e-05,
      "loss": 1.594,
      "step": 158560
    },
    {
      "epoch": 0.3129513846807569,
      "grad_norm": 0.11295372247695923,
      "learning_rate": 6.898362240793796e-05,
      "loss": 1.6024,
      "step": 158592
    },
    {
      "epoch": 0.31301453064215334,
      "grad_norm": 0.10866402834653854,
      "learning_rate": 6.897728218357327e-05,
      "loss": 1.5901,
      "step": 158624
    },
    {
      "epoch": 0.31307767660354974,
      "grad_norm": 0.11278446018695831,
      "learning_rate": 6.897094195920858e-05,
      "loss": 1.5909,
      "step": 158656
    },
    {
      "epoch": 0.3131408225649462,
      "grad_norm": 0.1106196865439415,
      "learning_rate": 6.896460173484389e-05,
      "loss": 1.5888,
      "step": 158688
    },
    {
      "epoch": 0.31320396852634264,
      "grad_norm": 0.1180410087108612,
      "learning_rate": 6.89582615104792e-05,
      "loss": 1.5948,
      "step": 158720
    },
    {
      "epoch": 0.3132671144877391,
      "grad_norm": 0.10209639370441437,
      "learning_rate": 6.895192128611452e-05,
      "loss": 1.5947,
      "step": 158752
    },
    {
      "epoch": 0.31333026044913553,
      "grad_norm": 0.10905079543590546,
      "learning_rate": 6.894558106174983e-05,
      "loss": 1.59,
      "step": 158784
    },
    {
      "epoch": 0.313393406410532,
      "grad_norm": 0.10796741396188736,
      "learning_rate": 6.893924083738514e-05,
      "loss": 1.5868,
      "step": 158816
    },
    {
      "epoch": 0.3134565523719284,
      "grad_norm": 0.11838201433420181,
      "learning_rate": 6.893290061302045e-05,
      "loss": 1.584,
      "step": 158848
    },
    {
      "epoch": 0.3135196983333248,
      "grad_norm": 0.11917586624622345,
      "learning_rate": 6.892656038865576e-05,
      "loss": 1.5853,
      "step": 158880
    },
    {
      "epoch": 0.3135828442947213,
      "grad_norm": 0.10489118099212646,
      "learning_rate": 6.892022016429107e-05,
      "loss": 1.597,
      "step": 158912
    },
    {
      "epoch": 0.3136459902561177,
      "grad_norm": 0.10635361820459366,
      "learning_rate": 6.891387993992638e-05,
      "loss": 1.6046,
      "step": 158944
    },
    {
      "epoch": 0.3137091362175142,
      "grad_norm": 0.1102752611041069,
      "learning_rate": 6.890753971556168e-05,
      "loss": 1.594,
      "step": 158976
    },
    {
      "epoch": 0.3137722821789106,
      "grad_norm": 0.10674628615379333,
      "learning_rate": 6.8901199491197e-05,
      "loss": 1.5869,
      "step": 159008
    },
    {
      "epoch": 0.313835428140307,
      "grad_norm": 0.11249663680791855,
      "learning_rate": 6.889485926683231e-05,
      "loss": 1.5928,
      "step": 159040
    },
    {
      "epoch": 0.31389857410170346,
      "grad_norm": 0.11040057986974716,
      "learning_rate": 6.888851904246761e-05,
      "loss": 1.5908,
      "step": 159072
    },
    {
      "epoch": 0.3139617200630999,
      "grad_norm": 0.11433102190494537,
      "learning_rate": 6.888217881810293e-05,
      "loss": 1.5853,
      "step": 159104
    },
    {
      "epoch": 0.31402486602449636,
      "grad_norm": 0.10093247145414352,
      "learning_rate": 6.887583859373824e-05,
      "loss": 1.5982,
      "step": 159136
    },
    {
      "epoch": 0.3140880119858928,
      "grad_norm": 0.10722556710243225,
      "learning_rate": 6.886949836937355e-05,
      "loss": 1.5972,
      "step": 159168
    },
    {
      "epoch": 0.31415115794728926,
      "grad_norm": 0.11650107800960541,
      "learning_rate": 6.886315814500886e-05,
      "loss": 1.5906,
      "step": 159200
    },
    {
      "epoch": 0.31421430390868565,
      "grad_norm": 0.11217987537384033,
      "learning_rate": 6.885681792064417e-05,
      "loss": 1.6025,
      "step": 159232
    },
    {
      "epoch": 0.3142774498700821,
      "grad_norm": 0.1122429296374321,
      "learning_rate": 6.885047769627948e-05,
      "loss": 1.5926,
      "step": 159264
    },
    {
      "epoch": 0.31434059583147855,
      "grad_norm": 0.11013469845056534,
      "learning_rate": 6.88441374719148e-05,
      "loss": 1.5922,
      "step": 159296
    },
    {
      "epoch": 0.314403741792875,
      "grad_norm": 0.10739205032587051,
      "learning_rate": 6.88377972475501e-05,
      "loss": 1.609,
      "step": 159328
    },
    {
      "epoch": 0.31446688775427145,
      "grad_norm": 0.11144138872623444,
      "learning_rate": 6.883145702318542e-05,
      "loss": 1.5856,
      "step": 159360
    },
    {
      "epoch": 0.3145300337156679,
      "grad_norm": 0.11184258759021759,
      "learning_rate": 6.882511679882072e-05,
      "loss": 1.5908,
      "step": 159392
    },
    {
      "epoch": 0.3145931796770643,
      "grad_norm": 0.11355500668287277,
      "learning_rate": 6.881877657445603e-05,
      "loss": 1.5909,
      "step": 159424
    },
    {
      "epoch": 0.31465632563846074,
      "grad_norm": 0.1178479790687561,
      "learning_rate": 6.881243635009135e-05,
      "loss": 1.5941,
      "step": 159456
    },
    {
      "epoch": 0.3147194715998572,
      "grad_norm": 0.11037848144769669,
      "learning_rate": 6.880609612572665e-05,
      "loss": 1.5923,
      "step": 159488
    },
    {
      "epoch": 0.31478261756125364,
      "grad_norm": 0.10712146013975143,
      "learning_rate": 6.879975590136196e-05,
      "loss": 1.5997,
      "step": 159520
    },
    {
      "epoch": 0.3148457635226501,
      "grad_norm": 0.111323781311512,
      "learning_rate": 6.879341567699728e-05,
      "loss": 1.6061,
      "step": 159552
    },
    {
      "epoch": 0.31490890948404654,
      "grad_norm": 0.11086145788431168,
      "learning_rate": 6.878707545263259e-05,
      "loss": 1.5827,
      "step": 159584
    },
    {
      "epoch": 0.31497205544544293,
      "grad_norm": 0.10983403027057648,
      "learning_rate": 6.878073522826789e-05,
      "loss": 1.5836,
      "step": 159616
    },
    {
      "epoch": 0.3150352014068394,
      "grad_norm": 0.10668952018022537,
      "learning_rate": 6.87743950039032e-05,
      "loss": 1.5942,
      "step": 159648
    },
    {
      "epoch": 0.31509834736823583,
      "grad_norm": 0.11072690039873123,
      "learning_rate": 6.876805477953852e-05,
      "loss": 1.6003,
      "step": 159680
    },
    {
      "epoch": 0.3151614933296323,
      "grad_norm": 0.11020134389400482,
      "learning_rate": 6.876171455517383e-05,
      "loss": 1.6023,
      "step": 159712
    },
    {
      "epoch": 0.31522463929102873,
      "grad_norm": 0.11055891215801239,
      "learning_rate": 6.875537433080914e-05,
      "loss": 1.6019,
      "step": 159744
    },
    {
      "epoch": 0.3152877852524252,
      "grad_norm": 0.10114159435033798,
      "learning_rate": 6.874903410644444e-05,
      "loss": 1.5761,
      "step": 159776
    },
    {
      "epoch": 0.3153509312138216,
      "grad_norm": 0.10959020256996155,
      "learning_rate": 6.874269388207975e-05,
      "loss": 1.5921,
      "step": 159808
    },
    {
      "epoch": 0.315414077175218,
      "grad_norm": 0.112167127430439,
      "learning_rate": 6.873635365771507e-05,
      "loss": 1.594,
      "step": 159840
    },
    {
      "epoch": 0.31547722313661447,
      "grad_norm": 0.11071991175413132,
      "learning_rate": 6.873001343335037e-05,
      "loss": 1.5917,
      "step": 159872
    },
    {
      "epoch": 0.3155403690980109,
      "grad_norm": 0.11651685833930969,
      "learning_rate": 6.872367320898568e-05,
      "loss": 1.5957,
      "step": 159904
    },
    {
      "epoch": 0.31560351505940737,
      "grad_norm": 0.1150650829076767,
      "learning_rate": 6.8717332984621e-05,
      "loss": 1.5979,
      "step": 159936
    },
    {
      "epoch": 0.3156666610208038,
      "grad_norm": 0.1077534481883049,
      "learning_rate": 6.871099276025631e-05,
      "loss": 1.5877,
      "step": 159968
    },
    {
      "epoch": 0.3157298069822002,
      "grad_norm": 0.11488234251737595,
      "learning_rate": 6.870465253589162e-05,
      "loss": 1.592,
      "step": 160000
    },
    {
      "epoch": 0.31579295294359666,
      "grad_norm": 0.11051345616579056,
      "learning_rate": 6.869831231152693e-05,
      "loss": 1.592,
      "step": 160032
    },
    {
      "epoch": 0.3158560989049931,
      "grad_norm": 0.10786715149879456,
      "learning_rate": 6.869197208716224e-05,
      "loss": 1.5975,
      "step": 160064
    },
    {
      "epoch": 0.31591924486638956,
      "grad_norm": 0.11121596395969391,
      "learning_rate": 6.868563186279756e-05,
      "loss": 1.5963,
      "step": 160096
    },
    {
      "epoch": 0.315982390827786,
      "grad_norm": 0.11501624435186386,
      "learning_rate": 6.867929163843287e-05,
      "loss": 1.5988,
      "step": 160128
    },
    {
      "epoch": 0.31604553678918246,
      "grad_norm": 0.11093879491090775,
      "learning_rate": 6.867295141406817e-05,
      "loss": 1.5823,
      "step": 160160
    },
    {
      "epoch": 0.3161086827505789,
      "grad_norm": 0.1130003109574318,
      "learning_rate": 6.866661118970347e-05,
      "loss": 1.6006,
      "step": 160192
    },
    {
      "epoch": 0.3161718287119753,
      "grad_norm": 0.11832395941019058,
      "learning_rate": 6.866027096533879e-05,
      "loss": 1.5877,
      "step": 160224
    },
    {
      "epoch": 0.31623497467337175,
      "grad_norm": 0.10775455832481384,
      "learning_rate": 6.86539307409741e-05,
      "loss": 1.5882,
      "step": 160256
    },
    {
      "epoch": 0.3162981206347682,
      "grad_norm": 0.11109977215528488,
      "learning_rate": 6.86475905166094e-05,
      "loss": 1.6057,
      "step": 160288
    },
    {
      "epoch": 0.31636126659616465,
      "grad_norm": 0.11257420480251312,
      "learning_rate": 6.864125029224472e-05,
      "loss": 1.5989,
      "step": 160320
    },
    {
      "epoch": 0.3164244125575611,
      "grad_norm": 0.10840550810098648,
      "learning_rate": 6.863491006788003e-05,
      "loss": 1.5926,
      "step": 160352
    },
    {
      "epoch": 0.31648755851895755,
      "grad_norm": 0.10533870756626129,
      "learning_rate": 6.862856984351535e-05,
      "loss": 1.613,
      "step": 160384
    },
    {
      "epoch": 0.31655070448035394,
      "grad_norm": 0.1078510731458664,
      "learning_rate": 6.862222961915065e-05,
      "loss": 1.6042,
      "step": 160416
    },
    {
      "epoch": 0.3166138504417504,
      "grad_norm": 0.10418468713760376,
      "learning_rate": 6.861588939478596e-05,
      "loss": 1.5873,
      "step": 160448
    },
    {
      "epoch": 0.31667699640314684,
      "grad_norm": 0.11282748728990555,
      "learning_rate": 6.860954917042128e-05,
      "loss": 1.5813,
      "step": 160480
    },
    {
      "epoch": 0.3167401423645433,
      "grad_norm": 0.10563211143016815,
      "learning_rate": 6.860320894605659e-05,
      "loss": 1.5971,
      "step": 160512
    },
    {
      "epoch": 0.31680328832593974,
      "grad_norm": 0.10779976844787598,
      "learning_rate": 6.859686872169189e-05,
      "loss": 1.6065,
      "step": 160544
    },
    {
      "epoch": 0.3168664342873362,
      "grad_norm": 0.10460280627012253,
      "learning_rate": 6.85905284973272e-05,
      "loss": 1.5941,
      "step": 160576
    },
    {
      "epoch": 0.3169295802487326,
      "grad_norm": 0.1153113842010498,
      "learning_rate": 6.858418827296251e-05,
      "loss": 1.5903,
      "step": 160608
    },
    {
      "epoch": 0.316992726210129,
      "grad_norm": 0.10854650288820267,
      "learning_rate": 6.857784804859782e-05,
      "loss": 1.6013,
      "step": 160640
    },
    {
      "epoch": 0.3170558721715255,
      "grad_norm": 0.10419408231973648,
      "learning_rate": 6.857150782423314e-05,
      "loss": 1.5905,
      "step": 160672
    },
    {
      "epoch": 0.3171190181329219,
      "grad_norm": 0.11376214772462845,
      "learning_rate": 6.856516759986844e-05,
      "loss": 1.598,
      "step": 160704
    },
    {
      "epoch": 0.3171821640943184,
      "grad_norm": 0.11283624172210693,
      "learning_rate": 6.855882737550375e-05,
      "loss": 1.6017,
      "step": 160736
    },
    {
      "epoch": 0.3172453100557148,
      "grad_norm": 0.11211321502923965,
      "learning_rate": 6.855248715113907e-05,
      "loss": 1.593,
      "step": 160768
    },
    {
      "epoch": 0.3173084560171112,
      "grad_norm": 0.108080193400383,
      "learning_rate": 6.854614692677438e-05,
      "loss": 1.6065,
      "step": 160800
    },
    {
      "epoch": 0.31737160197850767,
      "grad_norm": 0.1091194823384285,
      "learning_rate": 6.853980670240968e-05,
      "loss": 1.5901,
      "step": 160832
    },
    {
      "epoch": 0.3174347479399041,
      "grad_norm": 0.10642543435096741,
      "learning_rate": 6.8533466478045e-05,
      "loss": 1.6039,
      "step": 160864
    },
    {
      "epoch": 0.31749789390130057,
      "grad_norm": 0.11542787402868271,
      "learning_rate": 6.852712625368031e-05,
      "loss": 1.5977,
      "step": 160896
    },
    {
      "epoch": 0.317561039862697,
      "grad_norm": 0.11110210418701172,
      "learning_rate": 6.852078602931563e-05,
      "loss": 1.5842,
      "step": 160928
    },
    {
      "epoch": 0.31762418582409346,
      "grad_norm": 0.10630090534687042,
      "learning_rate": 6.851444580495093e-05,
      "loss": 1.6028,
      "step": 160960
    },
    {
      "epoch": 0.31768733178548986,
      "grad_norm": 0.1142411008477211,
      "learning_rate": 6.850810558058624e-05,
      "loss": 1.5819,
      "step": 160992
    },
    {
      "epoch": 0.3177504777468863,
      "grad_norm": 0.10625431686639786,
      "learning_rate": 6.850176535622154e-05,
      "loss": 1.5893,
      "step": 161024
    },
    {
      "epoch": 0.31781362370828276,
      "grad_norm": 0.10537683218717575,
      "learning_rate": 6.849542513185686e-05,
      "loss": 1.5942,
      "step": 161056
    },
    {
      "epoch": 0.3178767696696792,
      "grad_norm": 0.10894153267145157,
      "learning_rate": 6.848908490749216e-05,
      "loss": 1.5887,
      "step": 161088
    },
    {
      "epoch": 0.31793991563107565,
      "grad_norm": 0.10605444759130478,
      "learning_rate": 6.848274468312747e-05,
      "loss": 1.5982,
      "step": 161120
    },
    {
      "epoch": 0.3180030615924721,
      "grad_norm": 0.11229405552148819,
      "learning_rate": 6.847640445876279e-05,
      "loss": 1.5946,
      "step": 161152
    },
    {
      "epoch": 0.3180662075538685,
      "grad_norm": 0.10472061485052109,
      "learning_rate": 6.84700642343981e-05,
      "loss": 1.6101,
      "step": 161184
    },
    {
      "epoch": 0.31812935351526495,
      "grad_norm": 0.11303604394197464,
      "learning_rate": 6.84637240100334e-05,
      "loss": 1.6002,
      "step": 161216
    },
    {
      "epoch": 0.3181924994766614,
      "grad_norm": 0.11491641402244568,
      "learning_rate": 6.845738378566872e-05,
      "loss": 1.5865,
      "step": 161248
    },
    {
      "epoch": 0.31825564543805784,
      "grad_norm": 0.11333022266626358,
      "learning_rate": 6.845104356130403e-05,
      "loss": 1.5895,
      "step": 161280
    },
    {
      "epoch": 0.3183187913994543,
      "grad_norm": 0.11232884228229523,
      "learning_rate": 6.844470333693935e-05,
      "loss": 1.5993,
      "step": 161312
    },
    {
      "epoch": 0.31838193736085074,
      "grad_norm": 0.11450444161891937,
      "learning_rate": 6.843836311257466e-05,
      "loss": 1.5887,
      "step": 161344
    },
    {
      "epoch": 0.31844508332224714,
      "grad_norm": 0.11097846180200577,
      "learning_rate": 6.843202288820996e-05,
      "loss": 1.587,
      "step": 161376
    },
    {
      "epoch": 0.3185082292836436,
      "grad_norm": 0.10028678178787231,
      "learning_rate": 6.842568266384528e-05,
      "loss": 1.5876,
      "step": 161408
    },
    {
      "epoch": 0.31857137524504003,
      "grad_norm": 0.10802214592695236,
      "learning_rate": 6.841934243948058e-05,
      "loss": 1.603,
      "step": 161440
    },
    {
      "epoch": 0.3186345212064365,
      "grad_norm": 0.10867012292146683,
      "learning_rate": 6.841300221511589e-05,
      "loss": 1.586,
      "step": 161472
    },
    {
      "epoch": 0.31869766716783293,
      "grad_norm": 0.10596240311861038,
      "learning_rate": 6.840666199075119e-05,
      "loss": 1.5819,
      "step": 161504
    },
    {
      "epoch": 0.3187608131292294,
      "grad_norm": 0.10755529999732971,
      "learning_rate": 6.840032176638651e-05,
      "loss": 1.5902,
      "step": 161536
    },
    {
      "epoch": 0.3188239590906258,
      "grad_norm": 0.11589985340833664,
      "learning_rate": 6.839398154202182e-05,
      "loss": 1.5871,
      "step": 161568
    },
    {
      "epoch": 0.3188871050520222,
      "grad_norm": 0.11852046847343445,
      "learning_rate": 6.838764131765714e-05,
      "loss": 1.5944,
      "step": 161600
    },
    {
      "epoch": 0.3189502510134187,
      "grad_norm": 0.10627137124538422,
      "learning_rate": 6.838130109329244e-05,
      "loss": 1.602,
      "step": 161632
    },
    {
      "epoch": 0.3190133969748151,
      "grad_norm": 0.10510344803333282,
      "learning_rate": 6.837496086892775e-05,
      "loss": 1.6018,
      "step": 161664
    },
    {
      "epoch": 0.31907654293621157,
      "grad_norm": 0.11405289173126221,
      "learning_rate": 6.836862064456307e-05,
      "loss": 1.586,
      "step": 161696
    },
    {
      "epoch": 0.319139688897608,
      "grad_norm": 0.12056142836809158,
      "learning_rate": 6.836228042019838e-05,
      "loss": 1.5878,
      "step": 161728
    },
    {
      "epoch": 0.3192028348590044,
      "grad_norm": 0.10971621423959732,
      "learning_rate": 6.835594019583368e-05,
      "loss": 1.5892,
      "step": 161760
    },
    {
      "epoch": 0.31926598082040086,
      "grad_norm": 0.11604329198598862,
      "learning_rate": 6.8349599971469e-05,
      "loss": 1.5927,
      "step": 161792
    },
    {
      "epoch": 0.3193291267817973,
      "grad_norm": 0.1260664165019989,
      "learning_rate": 6.834325974710431e-05,
      "loss": 1.5922,
      "step": 161824
    },
    {
      "epoch": 0.31939227274319376,
      "grad_norm": 0.10880137979984283,
      "learning_rate": 6.833691952273961e-05,
      "loss": 1.5963,
      "step": 161856
    },
    {
      "epoch": 0.3194554187045902,
      "grad_norm": 0.10119996219873428,
      "learning_rate": 6.833057929837491e-05,
      "loss": 1.5977,
      "step": 161888
    },
    {
      "epoch": 0.31951856466598666,
      "grad_norm": 0.11640535295009613,
      "learning_rate": 6.832423907401023e-05,
      "loss": 1.5963,
      "step": 161920
    },
    {
      "epoch": 0.3195817106273831,
      "grad_norm": 0.1023346483707428,
      "learning_rate": 6.831789884964554e-05,
      "loss": 1.5878,
      "step": 161952
    },
    {
      "epoch": 0.3196448565887795,
      "grad_norm": 0.1058148518204689,
      "learning_rate": 6.831155862528086e-05,
      "loss": 1.5916,
      "step": 161984
    },
    {
      "epoch": 0.31970800255017595,
      "grad_norm": 0.10869254916906357,
      "learning_rate": 6.830521840091617e-05,
      "loss": 1.5805,
      "step": 162016
    },
    {
      "epoch": 0.3197711485115724,
      "grad_norm": 0.1046665757894516,
      "learning_rate": 6.829887817655147e-05,
      "loss": 1.5931,
      "step": 162048
    },
    {
      "epoch": 0.31983429447296885,
      "grad_norm": 0.11770608276128769,
      "learning_rate": 6.829253795218679e-05,
      "loss": 1.5933,
      "step": 162080
    },
    {
      "epoch": 0.3198974404343653,
      "grad_norm": 0.11175259947776794,
      "learning_rate": 6.82861977278221e-05,
      "loss": 1.5951,
      "step": 162112
    },
    {
      "epoch": 0.31996058639576175,
      "grad_norm": 0.11490031331777573,
      "learning_rate": 6.827985750345742e-05,
      "loss": 1.5936,
      "step": 162144
    },
    {
      "epoch": 0.32002373235715814,
      "grad_norm": 0.11524728685617447,
      "learning_rate": 6.827351727909272e-05,
      "loss": 1.5938,
      "step": 162176
    },
    {
      "epoch": 0.3200868783185546,
      "grad_norm": 0.11893665790557861,
      "learning_rate": 6.826717705472803e-05,
      "loss": 1.5884,
      "step": 162208
    },
    {
      "epoch": 0.32015002427995104,
      "grad_norm": 0.11445010453462601,
      "learning_rate": 6.826083683036335e-05,
      "loss": 1.59,
      "step": 162240
    },
    {
      "epoch": 0.3202131702413475,
      "grad_norm": 0.11429999768733978,
      "learning_rate": 6.825449660599865e-05,
      "loss": 1.586,
      "step": 162272
    },
    {
      "epoch": 0.32027631620274394,
      "grad_norm": 0.10755852609872818,
      "learning_rate": 6.824815638163395e-05,
      "loss": 1.5944,
      "step": 162304
    },
    {
      "epoch": 0.3203394621641404,
      "grad_norm": 0.10976274311542511,
      "learning_rate": 6.824181615726926e-05,
      "loss": 1.5985,
      "step": 162336
    },
    {
      "epoch": 0.3204026081255368,
      "grad_norm": 0.1098981723189354,
      "learning_rate": 6.823547593290458e-05,
      "loss": 1.5919,
      "step": 162368
    },
    {
      "epoch": 0.32046575408693323,
      "grad_norm": 0.1229216530919075,
      "learning_rate": 6.822913570853989e-05,
      "loss": 1.6008,
      "step": 162400
    },
    {
      "epoch": 0.3205289000483297,
      "grad_norm": 0.11485423892736435,
      "learning_rate": 6.822279548417519e-05,
      "loss": 1.6001,
      "step": 162432
    },
    {
      "epoch": 0.32059204600972613,
      "grad_norm": 0.11030910909175873,
      "learning_rate": 6.821645525981051e-05,
      "loss": 1.5969,
      "step": 162464
    },
    {
      "epoch": 0.3206551919711226,
      "grad_norm": 0.11059672385454178,
      "learning_rate": 6.821011503544582e-05,
      "loss": 1.5781,
      "step": 162496
    },
    {
      "epoch": 0.320718337932519,
      "grad_norm": 0.11908461898565292,
      "learning_rate": 6.820377481108114e-05,
      "loss": 1.5884,
      "step": 162528
    },
    {
      "epoch": 0.3207814838939154,
      "grad_norm": 0.11256904900074005,
      "learning_rate": 6.819743458671644e-05,
      "loss": 1.5965,
      "step": 162560
    },
    {
      "epoch": 0.32084462985531187,
      "grad_norm": 0.11307980120182037,
      "learning_rate": 6.819109436235175e-05,
      "loss": 1.6062,
      "step": 162592
    },
    {
      "epoch": 0.3209077758167083,
      "grad_norm": 0.10993563383817673,
      "learning_rate": 6.818475413798707e-05,
      "loss": 1.5855,
      "step": 162624
    },
    {
      "epoch": 0.32097092177810477,
      "grad_norm": 0.10647181421518326,
      "learning_rate": 6.817841391362237e-05,
      "loss": 1.595,
      "step": 162656
    },
    {
      "epoch": 0.3210340677395012,
      "grad_norm": 0.10353179275989532,
      "learning_rate": 6.817207368925768e-05,
      "loss": 1.5861,
      "step": 162688
    },
    {
      "epoch": 0.32109721370089767,
      "grad_norm": 0.11519227176904678,
      "learning_rate": 6.816573346489298e-05,
      "loss": 1.6057,
      "step": 162720
    },
    {
      "epoch": 0.32116035966229406,
      "grad_norm": 0.10504128783941269,
      "learning_rate": 6.81593932405283e-05,
      "loss": 1.5845,
      "step": 162752
    },
    {
      "epoch": 0.3212235056236905,
      "grad_norm": 0.11948299407958984,
      "learning_rate": 6.815305301616361e-05,
      "loss": 1.5912,
      "step": 162784
    },
    {
      "epoch": 0.32128665158508696,
      "grad_norm": 0.10750807821750641,
      "learning_rate": 6.814671279179893e-05,
      "loss": 1.5895,
      "step": 162816
    },
    {
      "epoch": 0.3213497975464834,
      "grad_norm": 0.09923849254846573,
      "learning_rate": 6.814037256743423e-05,
      "loss": 1.5845,
      "step": 162848
    },
    {
      "epoch": 0.32141294350787986,
      "grad_norm": 0.10898289084434509,
      "learning_rate": 6.813403234306954e-05,
      "loss": 1.6015,
      "step": 162880
    },
    {
      "epoch": 0.3214760894692763,
      "grad_norm": 0.11153297871351242,
      "learning_rate": 6.812769211870486e-05,
      "loss": 1.5945,
      "step": 162912
    },
    {
      "epoch": 0.3215392354306727,
      "grad_norm": 0.10640190541744232,
      "learning_rate": 6.812135189434017e-05,
      "loss": 1.5921,
      "step": 162944
    },
    {
      "epoch": 0.32160238139206915,
      "grad_norm": 0.11280359327793121,
      "learning_rate": 6.811501166997547e-05,
      "loss": 1.5863,
      "step": 162976
    },
    {
      "epoch": 0.3216655273534656,
      "grad_norm": 0.1157355085015297,
      "learning_rate": 6.810867144561079e-05,
      "loss": 1.6002,
      "step": 163008
    },
    {
      "epoch": 0.32172867331486205,
      "grad_norm": 0.10953034460544586,
      "learning_rate": 6.81023312212461e-05,
      "loss": 1.6148,
      "step": 163040
    },
    {
      "epoch": 0.3217918192762585,
      "grad_norm": 0.11689390242099762,
      "learning_rate": 6.80959909968814e-05,
      "loss": 1.5786,
      "step": 163072
    },
    {
      "epoch": 0.32185496523765494,
      "grad_norm": 0.10442730039358139,
      "learning_rate": 6.808965077251672e-05,
      "loss": 1.59,
      "step": 163104
    },
    {
      "epoch": 0.32191811119905134,
      "grad_norm": 0.10704589635133743,
      "learning_rate": 6.808331054815202e-05,
      "loss": 1.5943,
      "step": 163136
    },
    {
      "epoch": 0.3219812571604478,
      "grad_norm": 0.12023435533046722,
      "learning_rate": 6.807697032378733e-05,
      "loss": 1.5835,
      "step": 163168
    },
    {
      "epoch": 0.32204440312184424,
      "grad_norm": 0.1139121726155281,
      "learning_rate": 6.807063009942265e-05,
      "loss": 1.5974,
      "step": 163200
    },
    {
      "epoch": 0.3221075490832407,
      "grad_norm": 0.11478232592344284,
      "learning_rate": 6.806428987505795e-05,
      "loss": 1.5839,
      "step": 163232
    },
    {
      "epoch": 0.32217069504463713,
      "grad_norm": 0.1022125855088234,
      "learning_rate": 6.805794965069326e-05,
      "loss": 1.5927,
      "step": 163264
    },
    {
      "epoch": 0.3222338410060336,
      "grad_norm": 0.10918080061674118,
      "learning_rate": 6.805160942632858e-05,
      "loss": 1.5942,
      "step": 163296
    },
    {
      "epoch": 0.32229698696743,
      "grad_norm": 0.11021580547094345,
      "learning_rate": 6.804526920196389e-05,
      "loss": 1.5904,
      "step": 163328
    },
    {
      "epoch": 0.3223601329288264,
      "grad_norm": 0.11104167997837067,
      "learning_rate": 6.803892897759921e-05,
      "loss": 1.6126,
      "step": 163360
    },
    {
      "epoch": 0.3224232788902229,
      "grad_norm": 0.10933711379766464,
      "learning_rate": 6.803258875323451e-05,
      "loss": 1.5954,
      "step": 163392
    },
    {
      "epoch": 0.3224864248516193,
      "grad_norm": 0.10540857166051865,
      "learning_rate": 6.802624852886982e-05,
      "loss": 1.6208,
      "step": 163424
    },
    {
      "epoch": 0.3225495708130158,
      "grad_norm": 0.11258790642023087,
      "learning_rate": 6.801990830450514e-05,
      "loss": 1.5844,
      "step": 163456
    },
    {
      "epoch": 0.3226127167744122,
      "grad_norm": 0.12042684853076935,
      "learning_rate": 6.801356808014044e-05,
      "loss": 1.5929,
      "step": 163488
    },
    {
      "epoch": 0.32267586273580867,
      "grad_norm": 0.11128998547792435,
      "learning_rate": 6.800722785577575e-05,
      "loss": 1.5868,
      "step": 163520
    },
    {
      "epoch": 0.32273900869720507,
      "grad_norm": 0.10672010481357574,
      "learning_rate": 6.800088763141105e-05,
      "loss": 1.5838,
      "step": 163552
    },
    {
      "epoch": 0.3228021546586015,
      "grad_norm": 0.1076037660241127,
      "learning_rate": 6.799454740704637e-05,
      "loss": 1.5865,
      "step": 163584
    },
    {
      "epoch": 0.32286530061999796,
      "grad_norm": 0.11323705315589905,
      "learning_rate": 6.798820718268168e-05,
      "loss": 1.5918,
      "step": 163616
    },
    {
      "epoch": 0.3229284465813944,
      "grad_norm": 0.10990415513515472,
      "learning_rate": 6.798186695831698e-05,
      "loss": 1.5998,
      "step": 163648
    },
    {
      "epoch": 0.32299159254279086,
      "grad_norm": 0.1031436026096344,
      "learning_rate": 6.79755267339523e-05,
      "loss": 1.5932,
      "step": 163680
    },
    {
      "epoch": 0.3230547385041873,
      "grad_norm": 0.11307321488857269,
      "learning_rate": 6.796918650958761e-05,
      "loss": 1.5919,
      "step": 163712
    },
    {
      "epoch": 0.3231178844655837,
      "grad_norm": 0.11065627634525299,
      "learning_rate": 6.796284628522293e-05,
      "loss": 1.592,
      "step": 163744
    },
    {
      "epoch": 0.32318103042698015,
      "grad_norm": 0.10902044922113419,
      "learning_rate": 6.795650606085823e-05,
      "loss": 1.5917,
      "step": 163776
    },
    {
      "epoch": 0.3232441763883766,
      "grad_norm": 0.10561446845531464,
      "learning_rate": 6.795016583649354e-05,
      "loss": 1.584,
      "step": 163808
    },
    {
      "epoch": 0.32330732234977305,
      "grad_norm": 0.1130950003862381,
      "learning_rate": 6.794382561212886e-05,
      "loss": 1.5859,
      "step": 163840
    },
    {
      "epoch": 0.3233704683111695,
      "grad_norm": 0.11332274973392487,
      "learning_rate": 6.793748538776417e-05,
      "loss": 1.6003,
      "step": 163872
    },
    {
      "epoch": 0.32343361427256595,
      "grad_norm": 0.10275166481733322,
      "learning_rate": 6.793114516339947e-05,
      "loss": 1.5933,
      "step": 163904
    },
    {
      "epoch": 0.32349676023396234,
      "grad_norm": 0.10680346935987473,
      "learning_rate": 6.792480493903477e-05,
      "loss": 1.5929,
      "step": 163936
    },
    {
      "epoch": 0.3235599061953588,
      "grad_norm": 0.10647042095661163,
      "learning_rate": 6.791846471467009e-05,
      "loss": 1.5995,
      "step": 163968
    },
    {
      "epoch": 0.32362305215675524,
      "grad_norm": 0.10803354531526566,
      "learning_rate": 6.79121244903054e-05,
      "loss": 1.6,
      "step": 164000
    },
    {
      "epoch": 0.3236861981181517,
      "grad_norm": 0.11125516146421432,
      "learning_rate": 6.790578426594072e-05,
      "loss": 1.5958,
      "step": 164032
    },
    {
      "epoch": 0.32374934407954814,
      "grad_norm": 0.11516519635915756,
      "learning_rate": 6.789944404157602e-05,
      "loss": 1.5857,
      "step": 164064
    },
    {
      "epoch": 0.3238124900409446,
      "grad_norm": 0.10491325706243515,
      "learning_rate": 6.789310381721133e-05,
      "loss": 1.5887,
      "step": 164096
    },
    {
      "epoch": 0.323875636002341,
      "grad_norm": 0.11010641604661942,
      "learning_rate": 6.788676359284665e-05,
      "loss": 1.5963,
      "step": 164128
    },
    {
      "epoch": 0.32393878196373743,
      "grad_norm": 0.11462292075157166,
      "learning_rate": 6.788042336848196e-05,
      "loss": 1.5922,
      "step": 164160
    },
    {
      "epoch": 0.3240019279251339,
      "grad_norm": 0.10731970518827438,
      "learning_rate": 6.787408314411726e-05,
      "loss": 1.6029,
      "step": 164192
    },
    {
      "epoch": 0.32406507388653033,
      "grad_norm": 0.10830609500408173,
      "learning_rate": 6.786774291975258e-05,
      "loss": 1.5923,
      "step": 164224
    },
    {
      "epoch": 0.3241282198479268,
      "grad_norm": 0.10924004763364792,
      "learning_rate": 6.786140269538789e-05,
      "loss": 1.5822,
      "step": 164256
    },
    {
      "epoch": 0.32419136580932323,
      "grad_norm": 0.1176849901676178,
      "learning_rate": 6.785506247102321e-05,
      "loss": 1.5892,
      "step": 164288
    },
    {
      "epoch": 0.3242545117707196,
      "grad_norm": 0.10422728210687637,
      "learning_rate": 6.784872224665851e-05,
      "loss": 1.5898,
      "step": 164320
    },
    {
      "epoch": 0.32431765773211607,
      "grad_norm": 0.11070388555526733,
      "learning_rate": 6.784238202229381e-05,
      "loss": 1.5993,
      "step": 164352
    },
    {
      "epoch": 0.3243808036935125,
      "grad_norm": 0.11063247174024582,
      "learning_rate": 6.783604179792912e-05,
      "loss": 1.5929,
      "step": 164384
    },
    {
      "epoch": 0.32444394965490897,
      "grad_norm": 0.11325111985206604,
      "learning_rate": 6.782970157356444e-05,
      "loss": 1.6053,
      "step": 164416
    },
    {
      "epoch": 0.3245070956163054,
      "grad_norm": 0.11281933635473251,
      "learning_rate": 6.782336134919974e-05,
      "loss": 1.5871,
      "step": 164448
    },
    {
      "epoch": 0.32457024157770187,
      "grad_norm": 0.10405316203832626,
      "learning_rate": 6.781702112483505e-05,
      "loss": 1.5913,
      "step": 164480
    },
    {
      "epoch": 0.32463338753909826,
      "grad_norm": 0.10825597494840622,
      "learning_rate": 6.781068090047037e-05,
      "loss": 1.5974,
      "step": 164512
    },
    {
      "epoch": 0.3246965335004947,
      "grad_norm": 0.10484517365694046,
      "learning_rate": 6.780434067610568e-05,
      "loss": 1.6022,
      "step": 164544
    },
    {
      "epoch": 0.32475967946189116,
      "grad_norm": 0.12093804031610489,
      "learning_rate": 6.779800045174098e-05,
      "loss": 1.6005,
      "step": 164576
    },
    {
      "epoch": 0.3248228254232876,
      "grad_norm": 0.11590174585580826,
      "learning_rate": 6.77916602273763e-05,
      "loss": 1.5919,
      "step": 164608
    },
    {
      "epoch": 0.32488597138468406,
      "grad_norm": 0.10578962415456772,
      "learning_rate": 6.778532000301161e-05,
      "loss": 1.5993,
      "step": 164640
    },
    {
      "epoch": 0.3249491173460805,
      "grad_norm": 0.12327004224061966,
      "learning_rate": 6.777897977864693e-05,
      "loss": 1.5802,
      "step": 164672
    },
    {
      "epoch": 0.3250122633074769,
      "grad_norm": 0.11776653677225113,
      "learning_rate": 6.777263955428224e-05,
      "loss": 1.5939,
      "step": 164704
    },
    {
      "epoch": 0.32507540926887335,
      "grad_norm": 0.1176818385720253,
      "learning_rate": 6.776629932991754e-05,
      "loss": 1.5872,
      "step": 164736
    },
    {
      "epoch": 0.3251385552302698,
      "grad_norm": 0.11193326115608215,
      "learning_rate": 6.775995910555284e-05,
      "loss": 1.5989,
      "step": 164768
    },
    {
      "epoch": 0.32520170119166625,
      "grad_norm": 0.11355994641780853,
      "learning_rate": 6.775361888118816e-05,
      "loss": 1.6054,
      "step": 164800
    },
    {
      "epoch": 0.3252648471530627,
      "grad_norm": 0.11177047342061996,
      "learning_rate": 6.774727865682347e-05,
      "loss": 1.5805,
      "step": 164832
    },
    {
      "epoch": 0.32532799311445915,
      "grad_norm": 0.11082953214645386,
      "learning_rate": 6.774093843245877e-05,
      "loss": 1.589,
      "step": 164864
    },
    {
      "epoch": 0.32539113907585554,
      "grad_norm": 0.11140895634889603,
      "learning_rate": 6.773459820809409e-05,
      "loss": 1.5966,
      "step": 164896
    },
    {
      "epoch": 0.325454285037252,
      "grad_norm": 0.10152396559715271,
      "learning_rate": 6.77282579837294e-05,
      "loss": 1.5884,
      "step": 164928
    },
    {
      "epoch": 0.32551743099864844,
      "grad_norm": 0.11770064383745193,
      "learning_rate": 6.772191775936472e-05,
      "loss": 1.6033,
      "step": 164960
    },
    {
      "epoch": 0.3255805769600449,
      "grad_norm": 0.10735750198364258,
      "learning_rate": 6.771557753500002e-05,
      "loss": 1.5941,
      "step": 164992
    },
    {
      "epoch": 0.32564372292144134,
      "grad_norm": 0.10940711200237274,
      "learning_rate": 6.770923731063533e-05,
      "loss": 1.5832,
      "step": 165024
    },
    {
      "epoch": 0.3257068688828378,
      "grad_norm": 0.10065314173698425,
      "learning_rate": 6.770289708627065e-05,
      "loss": 1.5954,
      "step": 165056
    },
    {
      "epoch": 0.3257700148442342,
      "grad_norm": 0.11828918009996414,
      "learning_rate": 6.769655686190596e-05,
      "loss": 1.581,
      "step": 165088
    },
    {
      "epoch": 0.32583316080563063,
      "grad_norm": 0.11608125269412994,
      "learning_rate": 6.769021663754126e-05,
      "loss": 1.5859,
      "step": 165120
    },
    {
      "epoch": 0.3258963067670271,
      "grad_norm": 0.10929804295301437,
      "learning_rate": 6.768387641317658e-05,
      "loss": 1.5849,
      "step": 165152
    },
    {
      "epoch": 0.3259594527284235,
      "grad_norm": 0.11840365827083588,
      "learning_rate": 6.767753618881188e-05,
      "loss": 1.5808,
      "step": 165184
    },
    {
      "epoch": 0.32602259868982,
      "grad_norm": 0.11759553104639053,
      "learning_rate": 6.76711959644472e-05,
      "loss": 1.5918,
      "step": 165216
    },
    {
      "epoch": 0.3260857446512164,
      "grad_norm": 0.11764785647392273,
      "learning_rate": 6.76648557400825e-05,
      "loss": 1.5761,
      "step": 165248
    },
    {
      "epoch": 0.3261488906126129,
      "grad_norm": 0.11398757249116898,
      "learning_rate": 6.765851551571781e-05,
      "loss": 1.5931,
      "step": 165280
    },
    {
      "epoch": 0.32621203657400927,
      "grad_norm": 0.11280125379562378,
      "learning_rate": 6.765217529135312e-05,
      "loss": 1.587,
      "step": 165312
    },
    {
      "epoch": 0.3262751825354057,
      "grad_norm": 0.113761767745018,
      "learning_rate": 6.764583506698844e-05,
      "loss": 1.588,
      "step": 165344
    },
    {
      "epoch": 0.32633832849680217,
      "grad_norm": 0.10702557861804962,
      "learning_rate": 6.763949484262375e-05,
      "loss": 1.5831,
      "step": 165376
    },
    {
      "epoch": 0.3264014744581986,
      "grad_norm": 0.1112300306558609,
      "learning_rate": 6.763315461825905e-05,
      "loss": 1.5868,
      "step": 165408
    },
    {
      "epoch": 0.32646462041959506,
      "grad_norm": 0.10162105411291122,
      "learning_rate": 6.762681439389437e-05,
      "loss": 1.5909,
      "step": 165440
    },
    {
      "epoch": 0.3265277663809915,
      "grad_norm": 0.11442627012729645,
      "learning_rate": 6.762047416952968e-05,
      "loss": 1.5919,
      "step": 165472
    },
    {
      "epoch": 0.3265909123423879,
      "grad_norm": 0.1244163066148758,
      "learning_rate": 6.7614133945165e-05,
      "loss": 1.5834,
      "step": 165504
    },
    {
      "epoch": 0.32665405830378436,
      "grad_norm": 0.11429568380117416,
      "learning_rate": 6.76077937208003e-05,
      "loss": 1.5795,
      "step": 165536
    },
    {
      "epoch": 0.3267172042651808,
      "grad_norm": 0.10941269993782043,
      "learning_rate": 6.760145349643561e-05,
      "loss": 1.5955,
      "step": 165568
    },
    {
      "epoch": 0.32678035022657725,
      "grad_norm": 0.11360456049442291,
      "learning_rate": 6.759511327207091e-05,
      "loss": 1.5839,
      "step": 165600
    },
    {
      "epoch": 0.3268434961879737,
      "grad_norm": 0.10601862519979477,
      "learning_rate": 6.758877304770623e-05,
      "loss": 1.5993,
      "step": 165632
    },
    {
      "epoch": 0.32690664214937015,
      "grad_norm": 0.10836266726255417,
      "learning_rate": 6.758243282334153e-05,
      "loss": 1.5845,
      "step": 165664
    },
    {
      "epoch": 0.32696978811076655,
      "grad_norm": 0.1072690412402153,
      "learning_rate": 6.757609259897684e-05,
      "loss": 1.5876,
      "step": 165696
    },
    {
      "epoch": 0.327032934072163,
      "grad_norm": 0.10807265341281891,
      "learning_rate": 6.756975237461216e-05,
      "loss": 1.5803,
      "step": 165728
    },
    {
      "epoch": 0.32709608003355944,
      "grad_norm": 0.10372138023376465,
      "learning_rate": 6.756341215024747e-05,
      "loss": 1.5752,
      "step": 165760
    },
    {
      "epoch": 0.3271592259949559,
      "grad_norm": 0.10341465473175049,
      "learning_rate": 6.755707192588277e-05,
      "loss": 1.5946,
      "step": 165792
    },
    {
      "epoch": 0.32722237195635234,
      "grad_norm": 0.10707486420869827,
      "learning_rate": 6.755073170151809e-05,
      "loss": 1.6083,
      "step": 165824
    },
    {
      "epoch": 0.3272855179177488,
      "grad_norm": 0.11649633944034576,
      "learning_rate": 6.75443914771534e-05,
      "loss": 1.5898,
      "step": 165856
    },
    {
      "epoch": 0.3273486638791452,
      "grad_norm": 0.1109938770532608,
      "learning_rate": 6.753805125278872e-05,
      "loss": 1.5922,
      "step": 165888
    },
    {
      "epoch": 0.32741180984054163,
      "grad_norm": 0.11330954730510712,
      "learning_rate": 6.753171102842402e-05,
      "loss": 1.5796,
      "step": 165920
    },
    {
      "epoch": 0.3274749558019381,
      "grad_norm": 0.10658450424671173,
      "learning_rate": 6.752537080405933e-05,
      "loss": 1.5737,
      "step": 165952
    },
    {
      "epoch": 0.32753810176333453,
      "grad_norm": 0.11703871190547943,
      "learning_rate": 6.751903057969465e-05,
      "loss": 1.5849,
      "step": 165984
    },
    {
      "epoch": 0.327601247724731,
      "grad_norm": 0.11008935421705246,
      "learning_rate": 6.751269035532995e-05,
      "loss": 1.5892,
      "step": 166016
    },
    {
      "epoch": 0.32766439368612743,
      "grad_norm": 0.12122257798910141,
      "learning_rate": 6.750635013096526e-05,
      "loss": 1.5914,
      "step": 166048
    },
    {
      "epoch": 0.3277275396475238,
      "grad_norm": 0.1071445420384407,
      "learning_rate": 6.750000990660057e-05,
      "loss": 1.5923,
      "step": 166080
    },
    {
      "epoch": 0.3277906856089203,
      "grad_norm": 0.09923968464136124,
      "learning_rate": 6.749366968223588e-05,
      "loss": 1.5963,
      "step": 166112
    },
    {
      "epoch": 0.3278538315703167,
      "grad_norm": 0.10645250976085663,
      "learning_rate": 6.74873294578712e-05,
      "loss": 1.5945,
      "step": 166144
    },
    {
      "epoch": 0.32791697753171317,
      "grad_norm": 0.11301776766777039,
      "learning_rate": 6.748098923350651e-05,
      "loss": 1.5893,
      "step": 166176
    },
    {
      "epoch": 0.3279801234931096,
      "grad_norm": 0.11489652097225189,
      "learning_rate": 6.747464900914181e-05,
      "loss": 1.5852,
      "step": 166208
    },
    {
      "epoch": 0.32804326945450607,
      "grad_norm": 0.10619910806417465,
      "learning_rate": 6.746830878477712e-05,
      "loss": 1.5943,
      "step": 166240
    },
    {
      "epoch": 0.32810641541590246,
      "grad_norm": 0.10980654507875443,
      "learning_rate": 6.746196856041244e-05,
      "loss": 1.5928,
      "step": 166272
    },
    {
      "epoch": 0.3281695613772989,
      "grad_norm": 0.10765120387077332,
      "learning_rate": 6.745562833604775e-05,
      "loss": 1.5987,
      "step": 166304
    },
    {
      "epoch": 0.32823270733869536,
      "grad_norm": 0.10863944888114929,
      "learning_rate": 6.744928811168305e-05,
      "loss": 1.5807,
      "step": 166336
    },
    {
      "epoch": 0.3282958533000918,
      "grad_norm": 0.11275624483823776,
      "learning_rate": 6.744294788731837e-05,
      "loss": 1.5835,
      "step": 166368
    },
    {
      "epoch": 0.32835899926148826,
      "grad_norm": 0.10429783910512924,
      "learning_rate": 6.743660766295368e-05,
      "loss": 1.5894,
      "step": 166400
    },
    {
      "epoch": 0.3284221452228847,
      "grad_norm": 0.10573472082614899,
      "learning_rate": 6.743026743858898e-05,
      "loss": 1.591,
      "step": 166432
    },
    {
      "epoch": 0.3284852911842811,
      "grad_norm": 0.11760779470205307,
      "learning_rate": 6.742392721422429e-05,
      "loss": 1.6008,
      "step": 166464
    },
    {
      "epoch": 0.32854843714567755,
      "grad_norm": 0.11851943284273148,
      "learning_rate": 6.74175869898596e-05,
      "loss": 1.5922,
      "step": 166496
    },
    {
      "epoch": 0.328611583107074,
      "grad_norm": 0.10839410871267319,
      "learning_rate": 6.741124676549491e-05,
      "loss": 1.5847,
      "step": 166528
    },
    {
      "epoch": 0.32867472906847045,
      "grad_norm": 0.10711562633514404,
      "learning_rate": 6.740490654113023e-05,
      "loss": 1.5851,
      "step": 166560
    },
    {
      "epoch": 0.3287378750298669,
      "grad_norm": 0.10874127596616745,
      "learning_rate": 6.739856631676553e-05,
      "loss": 1.5985,
      "step": 166592
    },
    {
      "epoch": 0.32880102099126335,
      "grad_norm": 0.11116418987512589,
      "learning_rate": 6.739222609240084e-05,
      "loss": 1.5868,
      "step": 166624
    },
    {
      "epoch": 0.32886416695265974,
      "grad_norm": 0.10855557769536972,
      "learning_rate": 6.738588586803616e-05,
      "loss": 1.5928,
      "step": 166656
    },
    {
      "epoch": 0.3289273129140562,
      "grad_norm": 0.10828322917222977,
      "learning_rate": 6.737954564367147e-05,
      "loss": 1.5989,
      "step": 166688
    },
    {
      "epoch": 0.32899045887545264,
      "grad_norm": 0.1115008071064949,
      "learning_rate": 6.737320541930679e-05,
      "loss": 1.5877,
      "step": 166720
    },
    {
      "epoch": 0.3290536048368491,
      "grad_norm": 0.10906339436769485,
      "learning_rate": 6.736686519494209e-05,
      "loss": 1.596,
      "step": 166752
    },
    {
      "epoch": 0.32911675079824554,
      "grad_norm": 0.11613047122955322,
      "learning_rate": 6.73605249705774e-05,
      "loss": 1.5843,
      "step": 166784
    },
    {
      "epoch": 0.329179896759642,
      "grad_norm": 0.10479158908128738,
      "learning_rate": 6.73541847462127e-05,
      "loss": 1.6021,
      "step": 166816
    },
    {
      "epoch": 0.3292430427210384,
      "grad_norm": 0.1132587343454361,
      "learning_rate": 6.734784452184802e-05,
      "loss": 1.5797,
      "step": 166848
    },
    {
      "epoch": 0.32930618868243483,
      "grad_norm": 0.1093219593167305,
      "learning_rate": 6.734150429748332e-05,
      "loss": 1.5849,
      "step": 166880
    },
    {
      "epoch": 0.3293693346438313,
      "grad_norm": 0.11402468383312225,
      "learning_rate": 6.733516407311864e-05,
      "loss": 1.592,
      "step": 166912
    },
    {
      "epoch": 0.32943248060522773,
      "grad_norm": 0.12217032164335251,
      "learning_rate": 6.732882384875395e-05,
      "loss": 1.6057,
      "step": 166944
    },
    {
      "epoch": 0.3294956265666242,
      "grad_norm": 0.1209881380200386,
      "learning_rate": 6.732248362438926e-05,
      "loss": 1.5939,
      "step": 166976
    },
    {
      "epoch": 0.3295587725280206,
      "grad_norm": 0.10653063654899597,
      "learning_rate": 6.731614340002457e-05,
      "loss": 1.5924,
      "step": 167008
    },
    {
      "epoch": 0.3296219184894171,
      "grad_norm": 0.11813493818044662,
      "learning_rate": 6.730980317565988e-05,
      "loss": 1.5961,
      "step": 167040
    },
    {
      "epoch": 0.32968506445081347,
      "grad_norm": 0.10556158423423767,
      "learning_rate": 6.73034629512952e-05,
      "loss": 1.5857,
      "step": 167072
    },
    {
      "epoch": 0.3297482104122099,
      "grad_norm": 0.10984716564416885,
      "learning_rate": 6.729712272693051e-05,
      "loss": 1.5894,
      "step": 167104
    },
    {
      "epoch": 0.32981135637360637,
      "grad_norm": 0.10736513882875443,
      "learning_rate": 6.729078250256581e-05,
      "loss": 1.5894,
      "step": 167136
    },
    {
      "epoch": 0.3298745023350028,
      "grad_norm": 0.10866238921880722,
      "learning_rate": 6.728444227820112e-05,
      "loss": 1.599,
      "step": 167168
    },
    {
      "epoch": 0.32993764829639927,
      "grad_norm": 0.11599130928516388,
      "learning_rate": 6.727810205383644e-05,
      "loss": 1.6001,
      "step": 167200
    },
    {
      "epoch": 0.3300007942577957,
      "grad_norm": 0.12034738808870316,
      "learning_rate": 6.727176182947174e-05,
      "loss": 1.5993,
      "step": 167232
    },
    {
      "epoch": 0.3300639402191921,
      "grad_norm": 0.11284483969211578,
      "learning_rate": 6.726542160510705e-05,
      "loss": 1.5889,
      "step": 167264
    },
    {
      "epoch": 0.33012708618058856,
      "grad_norm": 0.10635127872228622,
      "learning_rate": 6.725908138074236e-05,
      "loss": 1.5977,
      "step": 167296
    },
    {
      "epoch": 0.330190232141985,
      "grad_norm": 0.10802595317363739,
      "learning_rate": 6.725274115637767e-05,
      "loss": 1.5873,
      "step": 167328
    },
    {
      "epoch": 0.33025337810338146,
      "grad_norm": 0.11855754256248474,
      "learning_rate": 6.724640093201299e-05,
      "loss": 1.5866,
      "step": 167360
    },
    {
      "epoch": 0.3303165240647779,
      "grad_norm": 0.10876497626304626,
      "learning_rate": 6.72400607076483e-05,
      "loss": 1.5825,
      "step": 167392
    },
    {
      "epoch": 0.33037967002617435,
      "grad_norm": 0.11352695524692535,
      "learning_rate": 6.72337204832836e-05,
      "loss": 1.575,
      "step": 167424
    },
    {
      "epoch": 0.33044281598757075,
      "grad_norm": 0.10477931052446365,
      "learning_rate": 6.722738025891892e-05,
      "loss": 1.582,
      "step": 167456
    },
    {
      "epoch": 0.3305059619489672,
      "grad_norm": 0.11220613867044449,
      "learning_rate": 6.722104003455423e-05,
      "loss": 1.5847,
      "step": 167488
    },
    {
      "epoch": 0.33056910791036365,
      "grad_norm": 0.10531621426343918,
      "learning_rate": 6.721469981018954e-05,
      "loss": 1.5865,
      "step": 167520
    },
    {
      "epoch": 0.3306322538717601,
      "grad_norm": 0.11641617864370346,
      "learning_rate": 6.720835958582485e-05,
      "loss": 1.5926,
      "step": 167552
    },
    {
      "epoch": 0.33069539983315654,
      "grad_norm": 0.10772272944450378,
      "learning_rate": 6.720201936146016e-05,
      "loss": 1.5994,
      "step": 167584
    },
    {
      "epoch": 0.330758545794553,
      "grad_norm": 0.10645993798971176,
      "learning_rate": 6.719567913709547e-05,
      "loss": 1.59,
      "step": 167616
    },
    {
      "epoch": 0.3308216917559494,
      "grad_norm": 0.11529100686311722,
      "learning_rate": 6.718933891273078e-05,
      "loss": 1.5909,
      "step": 167648
    },
    {
      "epoch": 0.33088483771734584,
      "grad_norm": 0.11064822971820831,
      "learning_rate": 6.718299868836609e-05,
      "loss": 1.5836,
      "step": 167680
    },
    {
      "epoch": 0.3309479836787423,
      "grad_norm": 0.11568716913461685,
      "learning_rate": 6.717665846400139e-05,
      "loss": 1.5875,
      "step": 167712
    },
    {
      "epoch": 0.33101112964013873,
      "grad_norm": 0.10499885678291321,
      "learning_rate": 6.71703182396367e-05,
      "loss": 1.5965,
      "step": 167744
    },
    {
      "epoch": 0.3310742756015352,
      "grad_norm": 0.10890348255634308,
      "learning_rate": 6.716397801527202e-05,
      "loss": 1.5876,
      "step": 167776
    },
    {
      "epoch": 0.33113742156293163,
      "grad_norm": 0.111075758934021,
      "learning_rate": 6.715763779090732e-05,
      "loss": 1.5943,
      "step": 167808
    },
    {
      "epoch": 0.331200567524328,
      "grad_norm": 0.10671944916248322,
      "learning_rate": 6.715129756654264e-05,
      "loss": 1.5835,
      "step": 167840
    },
    {
      "epoch": 0.3312637134857245,
      "grad_norm": 0.11546798795461655,
      "learning_rate": 6.714495734217795e-05,
      "loss": 1.5701,
      "step": 167872
    },
    {
      "epoch": 0.3313268594471209,
      "grad_norm": 0.1191597431898117,
      "learning_rate": 6.713861711781326e-05,
      "loss": 1.5842,
      "step": 167904
    },
    {
      "epoch": 0.3313900054085174,
      "grad_norm": 0.11169959604740143,
      "learning_rate": 6.713227689344857e-05,
      "loss": 1.5893,
      "step": 167936
    },
    {
      "epoch": 0.3314531513699138,
      "grad_norm": 0.10798674821853638,
      "learning_rate": 6.712593666908388e-05,
      "loss": 1.5871,
      "step": 167968
    },
    {
      "epoch": 0.3315162973313103,
      "grad_norm": 0.11244769394397736,
      "learning_rate": 6.71195964447192e-05,
      "loss": 1.5949,
      "step": 168000
    },
    {
      "epoch": 0.33157944329270667,
      "grad_norm": 0.11217678338289261,
      "learning_rate": 6.711325622035451e-05,
      "loss": 1.5916,
      "step": 168032
    },
    {
      "epoch": 0.3316425892541031,
      "grad_norm": 0.1120830848813057,
      "learning_rate": 6.710691599598981e-05,
      "loss": 1.5848,
      "step": 168064
    },
    {
      "epoch": 0.33170573521549956,
      "grad_norm": 0.10977490246295929,
      "learning_rate": 6.710057577162511e-05,
      "loss": 1.5986,
      "step": 168096
    },
    {
      "epoch": 0.331768881176896,
      "grad_norm": 0.10606007277965546,
      "learning_rate": 6.709423554726043e-05,
      "loss": 1.5885,
      "step": 168128
    },
    {
      "epoch": 0.33183202713829246,
      "grad_norm": 0.11085110157728195,
      "learning_rate": 6.708789532289574e-05,
      "loss": 1.6018,
      "step": 168160
    },
    {
      "epoch": 0.3318951730996889,
      "grad_norm": 0.11350330710411072,
      "learning_rate": 6.708155509853106e-05,
      "loss": 1.5888,
      "step": 168192
    },
    {
      "epoch": 0.3319583190610853,
      "grad_norm": 0.10424413532018661,
      "learning_rate": 6.707521487416636e-05,
      "loss": 1.5886,
      "step": 168224
    },
    {
      "epoch": 0.33202146502248175,
      "grad_norm": 0.11231934279203415,
      "learning_rate": 6.706887464980167e-05,
      "loss": 1.5996,
      "step": 168256
    },
    {
      "epoch": 0.3320846109838782,
      "grad_norm": 0.1062673032283783,
      "learning_rate": 6.706253442543699e-05,
      "loss": 1.5897,
      "step": 168288
    },
    {
      "epoch": 0.33214775694527465,
      "grad_norm": 0.11131986230611801,
      "learning_rate": 6.70561942010723e-05,
      "loss": 1.5764,
      "step": 168320
    },
    {
      "epoch": 0.3322109029066711,
      "grad_norm": 0.11383592337369919,
      "learning_rate": 6.70498539767076e-05,
      "loss": 1.5983,
      "step": 168352
    },
    {
      "epoch": 0.33227404886806755,
      "grad_norm": 0.1052570566534996,
      "learning_rate": 6.704351375234292e-05,
      "loss": 1.593,
      "step": 168384
    },
    {
      "epoch": 0.33233719482946394,
      "grad_norm": 0.11455070972442627,
      "learning_rate": 6.703717352797823e-05,
      "loss": 1.6021,
      "step": 168416
    },
    {
      "epoch": 0.3324003407908604,
      "grad_norm": 0.11270928382873535,
      "learning_rate": 6.703083330361354e-05,
      "loss": 1.5952,
      "step": 168448
    },
    {
      "epoch": 0.33246348675225684,
      "grad_norm": 0.10765434056520462,
      "learning_rate": 6.702449307924885e-05,
      "loss": 1.6175,
      "step": 168480
    },
    {
      "epoch": 0.3325266327136533,
      "grad_norm": 0.10825809836387634,
      "learning_rate": 6.701815285488415e-05,
      "loss": 1.5961,
      "step": 168512
    },
    {
      "epoch": 0.33258977867504974,
      "grad_norm": 0.11404110491275787,
      "learning_rate": 6.701181263051946e-05,
      "loss": 1.5804,
      "step": 168544
    },
    {
      "epoch": 0.3326529246364462,
      "grad_norm": 0.10480169951915741,
      "learning_rate": 6.700547240615478e-05,
      "loss": 1.5975,
      "step": 168576
    },
    {
      "epoch": 0.33271607059784264,
      "grad_norm": 0.10455060750246048,
      "learning_rate": 6.699913218179008e-05,
      "loss": 1.5862,
      "step": 168608
    },
    {
      "epoch": 0.33277921655923903,
      "grad_norm": 0.11547906696796417,
      "learning_rate": 6.699279195742539e-05,
      "loss": 1.5853,
      "step": 168640
    },
    {
      "epoch": 0.3328423625206355,
      "grad_norm": 0.10812269151210785,
      "learning_rate": 6.69864517330607e-05,
      "loss": 1.5986,
      "step": 168672
    },
    {
      "epoch": 0.33290550848203193,
      "grad_norm": 0.10520678013563156,
      "learning_rate": 6.698011150869602e-05,
      "loss": 1.5977,
      "step": 168704
    },
    {
      "epoch": 0.3329686544434284,
      "grad_norm": 0.10478810966014862,
      "learning_rate": 6.697377128433134e-05,
      "loss": 1.5787,
      "step": 168736
    },
    {
      "epoch": 0.33303180040482483,
      "grad_norm": 0.1120506078004837,
      "learning_rate": 6.696743105996664e-05,
      "loss": 1.5932,
      "step": 168768
    },
    {
      "epoch": 0.3330949463662213,
      "grad_norm": 0.11565958708524704,
      "learning_rate": 6.696109083560195e-05,
      "loss": 1.584,
      "step": 168800
    },
    {
      "epoch": 0.33315809232761767,
      "grad_norm": 0.11421355605125427,
      "learning_rate": 6.695475061123727e-05,
      "loss": 1.595,
      "step": 168832
    },
    {
      "epoch": 0.3332212382890141,
      "grad_norm": 0.1072520837187767,
      "learning_rate": 6.694841038687258e-05,
      "loss": 1.5862,
      "step": 168864
    },
    {
      "epoch": 0.33328438425041057,
      "grad_norm": 0.10841010510921478,
      "learning_rate": 6.694207016250788e-05,
      "loss": 1.5998,
      "step": 168896
    },
    {
      "epoch": 0.333347530211807,
      "grad_norm": 0.10595608502626419,
      "learning_rate": 6.693572993814318e-05,
      "loss": 1.5926,
      "step": 168928
    },
    {
      "epoch": 0.33341067617320347,
      "grad_norm": 0.10938339680433273,
      "learning_rate": 6.69293897137785e-05,
      "loss": 1.5928,
      "step": 168960
    },
    {
      "epoch": 0.3334738221345999,
      "grad_norm": 0.12203320115804672,
      "learning_rate": 6.692304948941381e-05,
      "loss": 1.5884,
      "step": 168992
    },
    {
      "epoch": 0.3335369680959963,
      "grad_norm": 0.11459293961524963,
      "learning_rate": 6.691670926504911e-05,
      "loss": 1.5835,
      "step": 169024
    },
    {
      "epoch": 0.33360011405739276,
      "grad_norm": 0.10714321583509445,
      "learning_rate": 6.691036904068443e-05,
      "loss": 1.5894,
      "step": 169056
    },
    {
      "epoch": 0.3336632600187892,
      "grad_norm": 0.11945177614688873,
      "learning_rate": 6.690402881631974e-05,
      "loss": 1.5873,
      "step": 169088
    },
    {
      "epoch": 0.33372640598018566,
      "grad_norm": 0.10785206407308578,
      "learning_rate": 6.689768859195506e-05,
      "loss": 1.5944,
      "step": 169120
    },
    {
      "epoch": 0.3337895519415821,
      "grad_norm": 0.11360948532819748,
      "learning_rate": 6.689134836759036e-05,
      "loss": 1.579,
      "step": 169152
    },
    {
      "epoch": 0.33385269790297856,
      "grad_norm": 0.10632267594337463,
      "learning_rate": 6.688500814322567e-05,
      "loss": 1.5923,
      "step": 169184
    },
    {
      "epoch": 0.33391584386437495,
      "grad_norm": 0.11562076956033707,
      "learning_rate": 6.687866791886099e-05,
      "loss": 1.5915,
      "step": 169216
    },
    {
      "epoch": 0.3339789898257714,
      "grad_norm": 0.11135333776473999,
      "learning_rate": 6.68723276944963e-05,
      "loss": 1.589,
      "step": 169248
    },
    {
      "epoch": 0.33404213578716785,
      "grad_norm": 0.11263851821422577,
      "learning_rate": 6.68659874701316e-05,
      "loss": 1.5813,
      "step": 169280
    },
    {
      "epoch": 0.3341052817485643,
      "grad_norm": 0.10896914452314377,
      "learning_rate": 6.685964724576692e-05,
      "loss": 1.5907,
      "step": 169312
    },
    {
      "epoch": 0.33416842770996075,
      "grad_norm": 0.11386636644601822,
      "learning_rate": 6.685330702140222e-05,
      "loss": 1.5944,
      "step": 169344
    },
    {
      "epoch": 0.3342315736713572,
      "grad_norm": 0.11390720307826996,
      "learning_rate": 6.684696679703753e-05,
      "loss": 1.6007,
      "step": 169376
    },
    {
      "epoch": 0.3342947196327536,
      "grad_norm": 0.1162019595503807,
      "learning_rate": 6.684062657267285e-05,
      "loss": 1.5965,
      "step": 169408
    },
    {
      "epoch": 0.33435786559415004,
      "grad_norm": 0.10611633211374283,
      "learning_rate": 6.683428634830815e-05,
      "loss": 1.582,
      "step": 169440
    },
    {
      "epoch": 0.3344210115555465,
      "grad_norm": 0.10824020206928253,
      "learning_rate": 6.682794612394346e-05,
      "loss": 1.5698,
      "step": 169472
    },
    {
      "epoch": 0.33448415751694294,
      "grad_norm": 0.1087648794054985,
      "learning_rate": 6.682160589957878e-05,
      "loss": 1.5906,
      "step": 169504
    },
    {
      "epoch": 0.3345473034783394,
      "grad_norm": 0.10838939994573593,
      "learning_rate": 6.681526567521409e-05,
      "loss": 1.5966,
      "step": 169536
    },
    {
      "epoch": 0.33461044943973584,
      "grad_norm": 0.10725168138742447,
      "learning_rate": 6.680892545084939e-05,
      "loss": 1.5775,
      "step": 169568
    },
    {
      "epoch": 0.33467359540113223,
      "grad_norm": 0.10876808315515518,
      "learning_rate": 6.68025852264847e-05,
      "loss": 1.5843,
      "step": 169600
    },
    {
      "epoch": 0.3347367413625287,
      "grad_norm": 0.1187715157866478,
      "learning_rate": 6.679624500212002e-05,
      "loss": 1.5876,
      "step": 169632
    },
    {
      "epoch": 0.3347998873239251,
      "grad_norm": 0.11132321506738663,
      "learning_rate": 6.678990477775534e-05,
      "loss": 1.5924,
      "step": 169664
    },
    {
      "epoch": 0.3348630332853216,
      "grad_norm": 0.11269792169332504,
      "learning_rate": 6.678356455339064e-05,
      "loss": 1.5955,
      "step": 169696
    },
    {
      "epoch": 0.334926179246718,
      "grad_norm": 0.11117807775735855,
      "learning_rate": 6.677722432902595e-05,
      "loss": 1.586,
      "step": 169728
    },
    {
      "epoch": 0.3349893252081145,
      "grad_norm": 0.10820810496807098,
      "learning_rate": 6.677088410466125e-05,
      "loss": 1.5881,
      "step": 169760
    },
    {
      "epoch": 0.33505247116951087,
      "grad_norm": 0.11064624041318893,
      "learning_rate": 6.676454388029657e-05,
      "loss": 1.593,
      "step": 169792
    },
    {
      "epoch": 0.3351156171309073,
      "grad_norm": 0.1094837561249733,
      "learning_rate": 6.675820365593187e-05,
      "loss": 1.6042,
      "step": 169824
    },
    {
      "epoch": 0.33517876309230377,
      "grad_norm": 0.11262187361717224,
      "learning_rate": 6.675186343156718e-05,
      "loss": 1.6065,
      "step": 169856
    },
    {
      "epoch": 0.3352419090537002,
      "grad_norm": 0.11713772267103195,
      "learning_rate": 6.67455232072025e-05,
      "loss": 1.5779,
      "step": 169888
    },
    {
      "epoch": 0.33530505501509666,
      "grad_norm": 0.11381383240222931,
      "learning_rate": 6.673918298283781e-05,
      "loss": 1.5952,
      "step": 169920
    },
    {
      "epoch": 0.3353682009764931,
      "grad_norm": 0.09947804361581802,
      "learning_rate": 6.673284275847313e-05,
      "loss": 1.5748,
      "step": 169952
    },
    {
      "epoch": 0.3354313469378895,
      "grad_norm": 0.11649797856807709,
      "learning_rate": 6.672650253410843e-05,
      "loss": 1.5861,
      "step": 169984
    },
    {
      "epoch": 0.33549449289928596,
      "grad_norm": 0.10755553841590881,
      "learning_rate": 6.672016230974374e-05,
      "loss": 1.5873,
      "step": 170016
    },
    {
      "epoch": 0.3355576388606824,
      "grad_norm": 0.1077168732881546,
      "learning_rate": 6.671382208537906e-05,
      "loss": 1.5884,
      "step": 170048
    },
    {
      "epoch": 0.33562078482207885,
      "grad_norm": 0.11261429637670517,
      "learning_rate": 6.670748186101437e-05,
      "loss": 1.5751,
      "step": 170080
    },
    {
      "epoch": 0.3356839307834753,
      "grad_norm": 0.10956179350614548,
      "learning_rate": 6.670114163664967e-05,
      "loss": 1.5822,
      "step": 170112
    },
    {
      "epoch": 0.33574707674487175,
      "grad_norm": 0.10310094058513641,
      "learning_rate": 6.669480141228499e-05,
      "loss": 1.584,
      "step": 170144
    },
    {
      "epoch": 0.33581022270626815,
      "grad_norm": 0.10348838567733765,
      "learning_rate": 6.668846118792029e-05,
      "loss": 1.5879,
      "step": 170176
    },
    {
      "epoch": 0.3358733686676646,
      "grad_norm": 0.10774225741624832,
      "learning_rate": 6.66821209635556e-05,
      "loss": 1.5743,
      "step": 170208
    },
    {
      "epoch": 0.33593651462906104,
      "grad_norm": 0.10263609886169434,
      "learning_rate": 6.66757807391909e-05,
      "loss": 1.5811,
      "step": 170240
    },
    {
      "epoch": 0.3359996605904575,
      "grad_norm": 0.11636590957641602,
      "learning_rate": 6.666944051482622e-05,
      "loss": 1.5751,
      "step": 170272
    },
    {
      "epoch": 0.33606280655185394,
      "grad_norm": 0.11100584268569946,
      "learning_rate": 6.666310029046153e-05,
      "loss": 1.5957,
      "step": 170304
    },
    {
      "epoch": 0.3361259525132504,
      "grad_norm": 0.1203126311302185,
      "learning_rate": 6.665676006609685e-05,
      "loss": 1.5954,
      "step": 170336
    },
    {
      "epoch": 0.33618909847464684,
      "grad_norm": 0.10767781734466553,
      "learning_rate": 6.665041984173215e-05,
      "loss": 1.5838,
      "step": 170368
    },
    {
      "epoch": 0.33625224443604323,
      "grad_norm": 0.10694015026092529,
      "learning_rate": 6.664407961736746e-05,
      "loss": 1.5816,
      "step": 170400
    },
    {
      "epoch": 0.3363153903974397,
      "grad_norm": 0.10093408077955246,
      "learning_rate": 6.663773939300278e-05,
      "loss": 1.5951,
      "step": 170432
    },
    {
      "epoch": 0.33637853635883613,
      "grad_norm": 0.1062542200088501,
      "learning_rate": 6.663139916863809e-05,
      "loss": 1.5942,
      "step": 170464
    },
    {
      "epoch": 0.3364416823202326,
      "grad_norm": 0.10482421517372131,
      "learning_rate": 6.662505894427339e-05,
      "loss": 1.5658,
      "step": 170496
    },
    {
      "epoch": 0.33650482828162903,
      "grad_norm": 0.10923250019550323,
      "learning_rate": 6.66187187199087e-05,
      "loss": 1.5838,
      "step": 170528
    },
    {
      "epoch": 0.3365679742430255,
      "grad_norm": 0.11676891148090363,
      "learning_rate": 6.661237849554402e-05,
      "loss": 1.5876,
      "step": 170560
    },
    {
      "epoch": 0.3366311202044219,
      "grad_norm": 0.1030052974820137,
      "learning_rate": 6.660603827117932e-05,
      "loss": 1.5828,
      "step": 170592
    },
    {
      "epoch": 0.3366942661658183,
      "grad_norm": 0.10722127556800842,
      "learning_rate": 6.659969804681464e-05,
      "loss": 1.5908,
      "step": 170624
    },
    {
      "epoch": 0.3367574121272148,
      "grad_norm": 0.10357983410358429,
      "learning_rate": 6.659335782244994e-05,
      "loss": 1.589,
      "step": 170656
    },
    {
      "epoch": 0.3368205580886112,
      "grad_norm": 0.1128392219543457,
      "learning_rate": 6.658701759808525e-05,
      "loss": 1.5866,
      "step": 170688
    },
    {
      "epoch": 0.33688370405000767,
      "grad_norm": 0.11493240296840668,
      "learning_rate": 6.658067737372057e-05,
      "loss": 1.5916,
      "step": 170720
    },
    {
      "epoch": 0.3369468500114041,
      "grad_norm": 0.11089370399713516,
      "learning_rate": 6.657433714935588e-05,
      "loss": 1.5811,
      "step": 170752
    },
    {
      "epoch": 0.3370099959728005,
      "grad_norm": 0.11617998778820038,
      "learning_rate": 6.656799692499118e-05,
      "loss": 1.6027,
      "step": 170784
    },
    {
      "epoch": 0.33707314193419696,
      "grad_norm": 0.11355898529291153,
      "learning_rate": 6.65616567006265e-05,
      "loss": 1.5896,
      "step": 170816
    },
    {
      "epoch": 0.3371362878955934,
      "grad_norm": 0.11810681223869324,
      "learning_rate": 6.655531647626181e-05,
      "loss": 1.5916,
      "step": 170848
    },
    {
      "epoch": 0.33719943385698986,
      "grad_norm": 0.11200131475925446,
      "learning_rate": 6.654897625189713e-05,
      "loss": 1.5928,
      "step": 170880
    },
    {
      "epoch": 0.3372625798183863,
      "grad_norm": 0.10641849040985107,
      "learning_rate": 6.654263602753243e-05,
      "loss": 1.5873,
      "step": 170912
    },
    {
      "epoch": 0.33732572577978276,
      "grad_norm": 0.11036920547485352,
      "learning_rate": 6.653629580316774e-05,
      "loss": 1.5941,
      "step": 170944
    },
    {
      "epoch": 0.33738887174117915,
      "grad_norm": 0.10746380686759949,
      "learning_rate": 6.652995557880304e-05,
      "loss": 1.5897,
      "step": 170976
    },
    {
      "epoch": 0.3374520177025756,
      "grad_norm": 0.11281225830316544,
      "learning_rate": 6.652361535443836e-05,
      "loss": 1.5905,
      "step": 171008
    },
    {
      "epoch": 0.33751516366397205,
      "grad_norm": 0.10482367128133774,
      "learning_rate": 6.651727513007366e-05,
      "loss": 1.5867,
      "step": 171040
    },
    {
      "epoch": 0.3375783096253685,
      "grad_norm": 0.1079397052526474,
      "learning_rate": 6.651093490570897e-05,
      "loss": 1.5895,
      "step": 171072
    },
    {
      "epoch": 0.33764145558676495,
      "grad_norm": 0.11590129882097244,
      "learning_rate": 6.650459468134429e-05,
      "loss": 1.6014,
      "step": 171104
    },
    {
      "epoch": 0.3377046015481614,
      "grad_norm": 0.1105310395359993,
      "learning_rate": 6.64982544569796e-05,
      "loss": 1.5813,
      "step": 171136
    },
    {
      "epoch": 0.3377677475095578,
      "grad_norm": 0.11263422667980194,
      "learning_rate": 6.64919142326149e-05,
      "loss": 1.5888,
      "step": 171168
    },
    {
      "epoch": 0.33783089347095424,
      "grad_norm": 0.10923925787210464,
      "learning_rate": 6.648557400825022e-05,
      "loss": 1.5984,
      "step": 171200
    },
    {
      "epoch": 0.3378940394323507,
      "grad_norm": 0.11082915961742401,
      "learning_rate": 6.647923378388553e-05,
      "loss": 1.5982,
      "step": 171232
    },
    {
      "epoch": 0.33795718539374714,
      "grad_norm": 0.11139413714408875,
      "learning_rate": 6.647289355952085e-05,
      "loss": 1.5896,
      "step": 171264
    },
    {
      "epoch": 0.3380203313551436,
      "grad_norm": 0.1109723448753357,
      "learning_rate": 6.646655333515616e-05,
      "loss": 1.5747,
      "step": 171296
    },
    {
      "epoch": 0.33808347731654004,
      "grad_norm": 0.1085285171866417,
      "learning_rate": 6.646021311079146e-05,
      "loss": 1.5878,
      "step": 171328
    },
    {
      "epoch": 0.33814662327793643,
      "grad_norm": 0.11112835258245468,
      "learning_rate": 6.645387288642678e-05,
      "loss": 1.5816,
      "step": 171360
    },
    {
      "epoch": 0.3382097692393329,
      "grad_norm": 0.10931267589330673,
      "learning_rate": 6.644753266206208e-05,
      "loss": 1.5832,
      "step": 171392
    },
    {
      "epoch": 0.33827291520072933,
      "grad_norm": 0.1158394068479538,
      "learning_rate": 6.644119243769739e-05,
      "loss": 1.5812,
      "step": 171424
    },
    {
      "epoch": 0.3383360611621258,
      "grad_norm": 0.10848306119441986,
      "learning_rate": 6.64348522133327e-05,
      "loss": 1.5878,
      "step": 171456
    },
    {
      "epoch": 0.3383992071235222,
      "grad_norm": 0.1074216291308403,
      "learning_rate": 6.642851198896801e-05,
      "loss": 1.5807,
      "step": 171488
    },
    {
      "epoch": 0.3384623530849187,
      "grad_norm": 0.11576060950756073,
      "learning_rate": 6.642217176460332e-05,
      "loss": 1.5916,
      "step": 171520
    },
    {
      "epoch": 0.33852549904631507,
      "grad_norm": 0.11410527676343918,
      "learning_rate": 6.641583154023864e-05,
      "loss": 1.5916,
      "step": 171552
    },
    {
      "epoch": 0.3385886450077115,
      "grad_norm": 0.10908595472574234,
      "learning_rate": 6.640949131587394e-05,
      "loss": 1.588,
      "step": 171584
    },
    {
      "epoch": 0.33865179096910797,
      "grad_norm": 0.11122795194387436,
      "learning_rate": 6.640315109150925e-05,
      "loss": 1.6002,
      "step": 171616
    },
    {
      "epoch": 0.3387149369305044,
      "grad_norm": 0.10948388278484344,
      "learning_rate": 6.639681086714457e-05,
      "loss": 1.5835,
      "step": 171648
    },
    {
      "epoch": 0.33877808289190087,
      "grad_norm": 0.10557802766561508,
      "learning_rate": 6.639047064277988e-05,
      "loss": 1.5692,
      "step": 171680
    },
    {
      "epoch": 0.3388412288532973,
      "grad_norm": 0.10576324909925461,
      "learning_rate": 6.638413041841518e-05,
      "loss": 1.5869,
      "step": 171712
    },
    {
      "epoch": 0.3389043748146937,
      "grad_norm": 0.11505261808633804,
      "learning_rate": 6.63777901940505e-05,
      "loss": 1.5859,
      "step": 171744
    },
    {
      "epoch": 0.33896752077609016,
      "grad_norm": 0.11075065284967422,
      "learning_rate": 6.637144996968581e-05,
      "loss": 1.5678,
      "step": 171776
    },
    {
      "epoch": 0.3390306667374866,
      "grad_norm": 0.11918734759092331,
      "learning_rate": 6.636510974532111e-05,
      "loss": 1.5872,
      "step": 171808
    },
    {
      "epoch": 0.33909381269888306,
      "grad_norm": 0.10720392316579819,
      "learning_rate": 6.635876952095643e-05,
      "loss": 1.5761,
      "step": 171840
    },
    {
      "epoch": 0.3391569586602795,
      "grad_norm": 0.11589060723781586,
      "learning_rate": 6.635242929659173e-05,
      "loss": 1.573,
      "step": 171872
    },
    {
      "epoch": 0.33922010462167596,
      "grad_norm": 0.1120370551943779,
      "learning_rate": 6.634608907222704e-05,
      "loss": 1.6045,
      "step": 171904
    },
    {
      "epoch": 0.3392832505830724,
      "grad_norm": 0.11259763687849045,
      "learning_rate": 6.633974884786236e-05,
      "loss": 1.5857,
      "step": 171936
    },
    {
      "epoch": 0.3393463965444688,
      "grad_norm": 0.11293957382440567,
      "learning_rate": 6.633340862349767e-05,
      "loss": 1.581,
      "step": 171968
    },
    {
      "epoch": 0.33940954250586525,
      "grad_norm": 0.10410208255052567,
      "learning_rate": 6.632706839913297e-05,
      "loss": 1.6065,
      "step": 172000
    },
    {
      "epoch": 0.3394726884672617,
      "grad_norm": 0.11103110015392303,
      "learning_rate": 6.632072817476829e-05,
      "loss": 1.5928,
      "step": 172032
    },
    {
      "epoch": 0.33953583442865815,
      "grad_norm": 0.11333668231964111,
      "learning_rate": 6.63143879504036e-05,
      "loss": 1.5822,
      "step": 172064
    },
    {
      "epoch": 0.3395989803900546,
      "grad_norm": 0.11592777818441391,
      "learning_rate": 6.630804772603892e-05,
      "loss": 1.5983,
      "step": 172096
    },
    {
      "epoch": 0.33966212635145104,
      "grad_norm": 0.10337577760219574,
      "learning_rate": 6.630170750167422e-05,
      "loss": 1.5903,
      "step": 172128
    },
    {
      "epoch": 0.33972527231284744,
      "grad_norm": 0.10988391935825348,
      "learning_rate": 6.629536727730953e-05,
      "loss": 1.5935,
      "step": 172160
    },
    {
      "epoch": 0.3397884182742439,
      "grad_norm": 0.10925642400979996,
      "learning_rate": 6.628902705294485e-05,
      "loss": 1.5778,
      "step": 172192
    },
    {
      "epoch": 0.33985156423564034,
      "grad_norm": 0.11613163352012634,
      "learning_rate": 6.628268682858015e-05,
      "loss": 1.5786,
      "step": 172224
    },
    {
      "epoch": 0.3399147101970368,
      "grad_norm": 0.12016303837299347,
      "learning_rate": 6.627634660421545e-05,
      "loss": 1.5803,
      "step": 172256
    },
    {
      "epoch": 0.33997785615843323,
      "grad_norm": 0.11270086467266083,
      "learning_rate": 6.627000637985076e-05,
      "loss": 1.5938,
      "step": 172288
    },
    {
      "epoch": 0.3400410021198297,
      "grad_norm": 0.10815758258104324,
      "learning_rate": 6.626366615548608e-05,
      "loss": 1.5804,
      "step": 172320
    },
    {
      "epoch": 0.3401041480812261,
      "grad_norm": 0.11350827664136887,
      "learning_rate": 6.625732593112139e-05,
      "loss": 1.5949,
      "step": 172352
    },
    {
      "epoch": 0.3401672940426225,
      "grad_norm": 0.11088835448026657,
      "learning_rate": 6.62509857067567e-05,
      "loss": 1.5931,
      "step": 172384
    },
    {
      "epoch": 0.340230440004019,
      "grad_norm": 0.12247086316347122,
      "learning_rate": 6.624464548239201e-05,
      "loss": 1.5896,
      "step": 172416
    },
    {
      "epoch": 0.3402935859654154,
      "grad_norm": 0.1066143736243248,
      "learning_rate": 6.623830525802732e-05,
      "loss": 1.5795,
      "step": 172448
    },
    {
      "epoch": 0.3403567319268119,
      "grad_norm": 0.11791466176509857,
      "learning_rate": 6.623196503366264e-05,
      "loss": 1.5887,
      "step": 172480
    },
    {
      "epoch": 0.3404198778882083,
      "grad_norm": 0.11066970229148865,
      "learning_rate": 6.622562480929794e-05,
      "loss": 1.5897,
      "step": 172512
    },
    {
      "epoch": 0.3404830238496047,
      "grad_norm": 0.1076154038310051,
      "learning_rate": 6.621928458493325e-05,
      "loss": 1.5906,
      "step": 172544
    },
    {
      "epoch": 0.34054616981100116,
      "grad_norm": 0.11044386774301529,
      "learning_rate": 6.621294436056857e-05,
      "loss": 1.5741,
      "step": 172576
    },
    {
      "epoch": 0.3406093157723976,
      "grad_norm": 0.11065259575843811,
      "learning_rate": 6.620660413620388e-05,
      "loss": 1.5854,
      "step": 172608
    },
    {
      "epoch": 0.34067246173379406,
      "grad_norm": 0.12373436242341995,
      "learning_rate": 6.620026391183918e-05,
      "loss": 1.5941,
      "step": 172640
    },
    {
      "epoch": 0.3407356076951905,
      "grad_norm": 0.115959532558918,
      "learning_rate": 6.619392368747448e-05,
      "loss": 1.5816,
      "step": 172672
    },
    {
      "epoch": 0.34079875365658696,
      "grad_norm": 0.10721660405397415,
      "learning_rate": 6.61875834631098e-05,
      "loss": 1.586,
      "step": 172704
    },
    {
      "epoch": 0.34086189961798335,
      "grad_norm": 0.10889130085706711,
      "learning_rate": 6.618124323874511e-05,
      "loss": 1.5881,
      "step": 172736
    },
    {
      "epoch": 0.3409250455793798,
      "grad_norm": 0.10835132747888565,
      "learning_rate": 6.617490301438043e-05,
      "loss": 1.5991,
      "step": 172768
    },
    {
      "epoch": 0.34098819154077625,
      "grad_norm": 0.11941508948802948,
      "learning_rate": 6.616856279001573e-05,
      "loss": 1.575,
      "step": 172800
    },
    {
      "epoch": 0.3410513375021727,
      "grad_norm": 0.10933182388544083,
      "learning_rate": 6.616222256565104e-05,
      "loss": 1.5795,
      "step": 172832
    },
    {
      "epoch": 0.34111448346356915,
      "grad_norm": 0.1083269789814949,
      "learning_rate": 6.615588234128636e-05,
      "loss": 1.5838,
      "step": 172864
    },
    {
      "epoch": 0.3411776294249656,
      "grad_norm": 0.10989560186862946,
      "learning_rate": 6.614954211692167e-05,
      "loss": 1.5876,
      "step": 172896
    },
    {
      "epoch": 0.341240775386362,
      "grad_norm": 0.11485954374074936,
      "learning_rate": 6.614320189255697e-05,
      "loss": 1.6061,
      "step": 172928
    },
    {
      "epoch": 0.34130392134775844,
      "grad_norm": 0.10763099044561386,
      "learning_rate": 6.613686166819229e-05,
      "loss": 1.5744,
      "step": 172960
    },
    {
      "epoch": 0.3413670673091549,
      "grad_norm": 0.10989924520254135,
      "learning_rate": 6.61305214438276e-05,
      "loss": 1.584,
      "step": 172992
    },
    {
      "epoch": 0.34143021327055134,
      "grad_norm": 0.1087680235505104,
      "learning_rate": 6.612418121946292e-05,
      "loss": 1.5735,
      "step": 173024
    },
    {
      "epoch": 0.3414933592319478,
      "grad_norm": 0.10286132991313934,
      "learning_rate": 6.611784099509822e-05,
      "loss": 1.5747,
      "step": 173056
    },
    {
      "epoch": 0.34155650519334424,
      "grad_norm": 0.1099427342414856,
      "learning_rate": 6.611150077073352e-05,
      "loss": 1.5892,
      "step": 173088
    },
    {
      "epoch": 0.34161965115474063,
      "grad_norm": 0.1077076867222786,
      "learning_rate": 6.610516054636883e-05,
      "loss": 1.5825,
      "step": 173120
    },
    {
      "epoch": 0.3416827971161371,
      "grad_norm": 0.11097551882266998,
      "learning_rate": 6.609882032200415e-05,
      "loss": 1.5871,
      "step": 173152
    },
    {
      "epoch": 0.34174594307753353,
      "grad_norm": 0.10670367628335953,
      "learning_rate": 6.609248009763945e-05,
      "loss": 1.5827,
      "step": 173184
    },
    {
      "epoch": 0.34180908903893,
      "grad_norm": 0.10981249064207077,
      "learning_rate": 6.608613987327476e-05,
      "loss": 1.5912,
      "step": 173216
    },
    {
      "epoch": 0.34187223500032643,
      "grad_norm": 0.10873278230428696,
      "learning_rate": 6.607979964891008e-05,
      "loss": 1.5774,
      "step": 173248
    },
    {
      "epoch": 0.3419353809617229,
      "grad_norm": 0.11112535744905472,
      "learning_rate": 6.607345942454539e-05,
      "loss": 1.5902,
      "step": 173280
    },
    {
      "epoch": 0.3419985269231193,
      "grad_norm": 0.12162293493747711,
      "learning_rate": 6.606711920018071e-05,
      "loss": 1.5909,
      "step": 173312
    },
    {
      "epoch": 0.3420616728845157,
      "grad_norm": 0.11760684847831726,
      "learning_rate": 6.606077897581601e-05,
      "loss": 1.587,
      "step": 173344
    },
    {
      "epoch": 0.34212481884591217,
      "grad_norm": 0.10883896797895432,
      "learning_rate": 6.605443875145132e-05,
      "loss": 1.6,
      "step": 173376
    },
    {
      "epoch": 0.3421879648073086,
      "grad_norm": 0.1142784133553505,
      "learning_rate": 6.604809852708664e-05,
      "loss": 1.5971,
      "step": 173408
    },
    {
      "epoch": 0.34225111076870507,
      "grad_norm": 0.10538744181394577,
      "learning_rate": 6.604175830272194e-05,
      "loss": 1.5685,
      "step": 173440
    },
    {
      "epoch": 0.3423142567301015,
      "grad_norm": 0.11750444024801254,
      "learning_rate": 6.603541807835725e-05,
      "loss": 1.5901,
      "step": 173472
    },
    {
      "epoch": 0.3423774026914979,
      "grad_norm": 0.11520704627037048,
      "learning_rate": 6.602907785399255e-05,
      "loss": 1.5924,
      "step": 173504
    },
    {
      "epoch": 0.34244054865289436,
      "grad_norm": 0.11306174844503403,
      "learning_rate": 6.602273762962787e-05,
      "loss": 1.5865,
      "step": 173536
    },
    {
      "epoch": 0.3425036946142908,
      "grad_norm": 0.11422741413116455,
      "learning_rate": 6.601639740526318e-05,
      "loss": 1.5979,
      "step": 173568
    },
    {
      "epoch": 0.34256684057568726,
      "grad_norm": 0.10371016710996628,
      "learning_rate": 6.601005718089848e-05,
      "loss": 1.5848,
      "step": 173600
    },
    {
      "epoch": 0.3426299865370837,
      "grad_norm": 0.10575991123914719,
      "learning_rate": 6.60037169565338e-05,
      "loss": 1.5968,
      "step": 173632
    },
    {
      "epoch": 0.34269313249848016,
      "grad_norm": 0.10149675607681274,
      "learning_rate": 6.599737673216911e-05,
      "loss": 1.5909,
      "step": 173664
    },
    {
      "epoch": 0.3427562784598766,
      "grad_norm": 0.10535045713186264,
      "learning_rate": 6.599103650780443e-05,
      "loss": 1.5913,
      "step": 173696
    },
    {
      "epoch": 0.342819424421273,
      "grad_norm": 0.10517001152038574,
      "learning_rate": 6.598469628343973e-05,
      "loss": 1.589,
      "step": 173728
    },
    {
      "epoch": 0.34288257038266945,
      "grad_norm": 0.13139063119888306,
      "learning_rate": 6.597835605907504e-05,
      "loss": 1.5818,
      "step": 173760
    },
    {
      "epoch": 0.3429457163440659,
      "grad_norm": 0.1072242334485054,
      "learning_rate": 6.597201583471036e-05,
      "loss": 1.5891,
      "step": 173792
    },
    {
      "epoch": 0.34300886230546235,
      "grad_norm": 0.10862624645233154,
      "learning_rate": 6.596567561034567e-05,
      "loss": 1.5793,
      "step": 173824
    },
    {
      "epoch": 0.3430720082668588,
      "grad_norm": 0.11293728649616241,
      "learning_rate": 6.595933538598097e-05,
      "loss": 1.5897,
      "step": 173856
    },
    {
      "epoch": 0.34313515422825525,
      "grad_norm": 0.10044048726558685,
      "learning_rate": 6.595299516161629e-05,
      "loss": 1.6005,
      "step": 173888
    },
    {
      "epoch": 0.34319830018965164,
      "grad_norm": 0.11105641722679138,
      "learning_rate": 6.594665493725159e-05,
      "loss": 1.5954,
      "step": 173920
    },
    {
      "epoch": 0.3432614461510481,
      "grad_norm": 0.10460119694471359,
      "learning_rate": 6.59403147128869e-05,
      "loss": 1.5827,
      "step": 173952
    },
    {
      "epoch": 0.34332459211244454,
      "grad_norm": 0.11204526573419571,
      "learning_rate": 6.593397448852222e-05,
      "loss": 1.5883,
      "step": 173984
    },
    {
      "epoch": 0.343387738073841,
      "grad_norm": 0.11562290787696838,
      "learning_rate": 6.592763426415752e-05,
      "loss": 1.5798,
      "step": 174016
    },
    {
      "epoch": 0.34345088403523744,
      "grad_norm": 0.1139347031712532,
      "learning_rate": 6.592129403979283e-05,
      "loss": 1.5852,
      "step": 174048
    },
    {
      "epoch": 0.3435140299966339,
      "grad_norm": 0.11531013995409012,
      "learning_rate": 6.591495381542815e-05,
      "loss": 1.5861,
      "step": 174080
    },
    {
      "epoch": 0.3435771759580303,
      "grad_norm": 0.09973633289337158,
      "learning_rate": 6.590861359106346e-05,
      "loss": 1.5897,
      "step": 174112
    },
    {
      "epoch": 0.34364032191942673,
      "grad_norm": 0.11261215060949326,
      "learning_rate": 6.590227336669876e-05,
      "loss": 1.5909,
      "step": 174144
    },
    {
      "epoch": 0.3437034678808232,
      "grad_norm": 0.11859069764614105,
      "learning_rate": 6.589593314233408e-05,
      "loss": 1.5881,
      "step": 174176
    },
    {
      "epoch": 0.3437666138422196,
      "grad_norm": 0.10515476018190384,
      "learning_rate": 6.58895929179694e-05,
      "loss": 1.5974,
      "step": 174208
    },
    {
      "epoch": 0.3438297598036161,
      "grad_norm": 0.1277274787425995,
      "learning_rate": 6.588325269360471e-05,
      "loss": 1.5826,
      "step": 174240
    },
    {
      "epoch": 0.3438929057650125,
      "grad_norm": 0.11068448424339294,
      "learning_rate": 6.587691246924001e-05,
      "loss": 1.5834,
      "step": 174272
    },
    {
      "epoch": 0.3439560517264089,
      "grad_norm": 0.11146080493927002,
      "learning_rate": 6.587057224487532e-05,
      "loss": 1.5855,
      "step": 174304
    },
    {
      "epoch": 0.34401919768780537,
      "grad_norm": 0.10851216316223145,
      "learning_rate": 6.586423202051062e-05,
      "loss": 1.5915,
      "step": 174336
    },
    {
      "epoch": 0.3440823436492018,
      "grad_norm": 0.10507667809724808,
      "learning_rate": 6.585789179614594e-05,
      "loss": 1.5883,
      "step": 174368
    },
    {
      "epoch": 0.34414548961059827,
      "grad_norm": 0.11206206679344177,
      "learning_rate": 6.585155157178124e-05,
      "loss": 1.5825,
      "step": 174400
    },
    {
      "epoch": 0.3442086355719947,
      "grad_norm": 0.11833879351615906,
      "learning_rate": 6.584521134741655e-05,
      "loss": 1.5917,
      "step": 174432
    },
    {
      "epoch": 0.34427178153339116,
      "grad_norm": 0.10395336151123047,
      "learning_rate": 6.583887112305187e-05,
      "loss": 1.5918,
      "step": 174464
    },
    {
      "epoch": 0.34433492749478756,
      "grad_norm": 0.11111441254615784,
      "learning_rate": 6.583253089868718e-05,
      "loss": 1.5915,
      "step": 174496
    },
    {
      "epoch": 0.344398073456184,
      "grad_norm": 0.10616868734359741,
      "learning_rate": 6.582619067432248e-05,
      "loss": 1.5944,
      "step": 174528
    },
    {
      "epoch": 0.34446121941758046,
      "grad_norm": 0.12092773616313934,
      "learning_rate": 6.58198504499578e-05,
      "loss": 1.5807,
      "step": 174560
    },
    {
      "epoch": 0.3445243653789769,
      "grad_norm": 0.11427041888237,
      "learning_rate": 6.581351022559311e-05,
      "loss": 1.5916,
      "step": 174592
    },
    {
      "epoch": 0.34458751134037335,
      "grad_norm": 0.11512736976146698,
      "learning_rate": 6.580717000122843e-05,
      "loss": 1.5743,
      "step": 174624
    },
    {
      "epoch": 0.3446506573017698,
      "grad_norm": 0.10949216037988663,
      "learning_rate": 6.580082977686374e-05,
      "loss": 1.5849,
      "step": 174656
    },
    {
      "epoch": 0.3447138032631662,
      "grad_norm": 0.11290450394153595,
      "learning_rate": 6.579448955249904e-05,
      "loss": 1.5951,
      "step": 174688
    },
    {
      "epoch": 0.34477694922456265,
      "grad_norm": 0.1092032715678215,
      "learning_rate": 6.578814932813436e-05,
      "loss": 1.5981,
      "step": 174720
    },
    {
      "epoch": 0.3448400951859591,
      "grad_norm": 0.10919303447008133,
      "learning_rate": 6.578180910376966e-05,
      "loss": 1.5905,
      "step": 174752
    },
    {
      "epoch": 0.34490324114735554,
      "grad_norm": 0.10852492600679398,
      "learning_rate": 6.577546887940497e-05,
      "loss": 1.5897,
      "step": 174784
    },
    {
      "epoch": 0.344966387108752,
      "grad_norm": 0.10703276097774506,
      "learning_rate": 6.576912865504028e-05,
      "loss": 1.5889,
      "step": 174816
    },
    {
      "epoch": 0.34502953307014844,
      "grad_norm": 0.10765741020441055,
      "learning_rate": 6.576278843067559e-05,
      "loss": 1.5733,
      "step": 174848
    },
    {
      "epoch": 0.34509267903154484,
      "grad_norm": 0.1112956702709198,
      "learning_rate": 6.57564482063109e-05,
      "loss": 1.5889,
      "step": 174880
    },
    {
      "epoch": 0.3451558249929413,
      "grad_norm": 0.11600566655397415,
      "learning_rate": 6.575010798194622e-05,
      "loss": 1.5851,
      "step": 174912
    },
    {
      "epoch": 0.34521897095433773,
      "grad_norm": 0.10642077028751373,
      "learning_rate": 6.574376775758152e-05,
      "loss": 1.5886,
      "step": 174944
    },
    {
      "epoch": 0.3452821169157342,
      "grad_norm": 0.1070687472820282,
      "learning_rate": 6.573742753321683e-05,
      "loss": 1.5836,
      "step": 174976
    },
    {
      "epoch": 0.34534526287713063,
      "grad_norm": 0.10959191620349884,
      "learning_rate": 6.573108730885215e-05,
      "loss": 1.5849,
      "step": 175008
    },
    {
      "epoch": 0.3454084088385271,
      "grad_norm": 0.10763245075941086,
      "learning_rate": 6.572474708448746e-05,
      "loss": 1.5852,
      "step": 175040
    },
    {
      "epoch": 0.3454715547999235,
      "grad_norm": 0.10724011063575745,
      "learning_rate": 6.571840686012276e-05,
      "loss": 1.5882,
      "step": 175072
    },
    {
      "epoch": 0.3455347007613199,
      "grad_norm": 0.1093960627913475,
      "learning_rate": 6.571206663575808e-05,
      "loss": 1.5837,
      "step": 175104
    },
    {
      "epoch": 0.3455978467227164,
      "grad_norm": 0.11032553762197495,
      "learning_rate": 6.570572641139338e-05,
      "loss": 1.5919,
      "step": 175136
    },
    {
      "epoch": 0.3456609926841128,
      "grad_norm": 0.11518137156963348,
      "learning_rate": 6.56993861870287e-05,
      "loss": 1.6009,
      "step": 175168
    },
    {
      "epoch": 0.34572413864550927,
      "grad_norm": 0.10641338676214218,
      "learning_rate": 6.5693045962664e-05,
      "loss": 1.5828,
      "step": 175200
    },
    {
      "epoch": 0.3457872846069057,
      "grad_norm": 0.10659778118133545,
      "learning_rate": 6.568670573829931e-05,
      "loss": 1.5957,
      "step": 175232
    },
    {
      "epoch": 0.3458504305683021,
      "grad_norm": 0.11084144562482834,
      "learning_rate": 6.568036551393462e-05,
      "loss": 1.5882,
      "step": 175264
    },
    {
      "epoch": 0.34591357652969856,
      "grad_norm": 0.10810660570859909,
      "learning_rate": 6.567402528956994e-05,
      "loss": 1.5896,
      "step": 175296
    },
    {
      "epoch": 0.345976722491095,
      "grad_norm": 0.10464351624250412,
      "learning_rate": 6.566768506520525e-05,
      "loss": 1.5886,
      "step": 175328
    },
    {
      "epoch": 0.34603986845249146,
      "grad_norm": 0.12545694410800934,
      "learning_rate": 6.566134484084056e-05,
      "loss": 1.5837,
      "step": 175360
    },
    {
      "epoch": 0.3461030144138879,
      "grad_norm": 0.10950300842523575,
      "learning_rate": 6.565500461647587e-05,
      "loss": 1.581,
      "step": 175392
    },
    {
      "epoch": 0.34616616037528436,
      "grad_norm": 0.11077903211116791,
      "learning_rate": 6.564866439211118e-05,
      "loss": 1.5854,
      "step": 175424
    },
    {
      "epoch": 0.3462293063366808,
      "grad_norm": 0.128065824508667,
      "learning_rate": 6.56423241677465e-05,
      "loss": 1.5919,
      "step": 175456
    },
    {
      "epoch": 0.3462924522980772,
      "grad_norm": 0.11247207969427109,
      "learning_rate": 6.56359839433818e-05,
      "loss": 1.5747,
      "step": 175488
    },
    {
      "epoch": 0.34635559825947365,
      "grad_norm": 0.10944617539644241,
      "learning_rate": 6.562964371901711e-05,
      "loss": 1.5823,
      "step": 175520
    },
    {
      "epoch": 0.3464187442208701,
      "grad_norm": 0.10773425549268723,
      "learning_rate": 6.562330349465242e-05,
      "loss": 1.587,
      "step": 175552
    },
    {
      "epoch": 0.34648189018226655,
      "grad_norm": 0.11003351956605911,
      "learning_rate": 6.561696327028773e-05,
      "loss": 1.5934,
      "step": 175584
    },
    {
      "epoch": 0.346545036143663,
      "grad_norm": 0.12288950383663177,
      "learning_rate": 6.561062304592303e-05,
      "loss": 1.5897,
      "step": 175616
    },
    {
      "epoch": 0.34660818210505945,
      "grad_norm": 0.11313644051551819,
      "learning_rate": 6.560428282155835e-05,
      "loss": 1.5944,
      "step": 175648
    },
    {
      "epoch": 0.34667132806645584,
      "grad_norm": 0.10772547125816345,
      "learning_rate": 6.559794259719366e-05,
      "loss": 1.5955,
      "step": 175680
    },
    {
      "epoch": 0.3467344740278523,
      "grad_norm": 0.11855486780405045,
      "learning_rate": 6.559160237282897e-05,
      "loss": 1.5885,
      "step": 175712
    },
    {
      "epoch": 0.34679761998924874,
      "grad_norm": 0.11729282885789871,
      "learning_rate": 6.558526214846428e-05,
      "loss": 1.5808,
      "step": 175744
    },
    {
      "epoch": 0.3468607659506452,
      "grad_norm": 0.1073223426938057,
      "learning_rate": 6.557892192409959e-05,
      "loss": 1.5937,
      "step": 175776
    },
    {
      "epoch": 0.34692391191204164,
      "grad_norm": 0.10576095432043076,
      "learning_rate": 6.55725816997349e-05,
      "loss": 1.6005,
      "step": 175808
    },
    {
      "epoch": 0.3469870578734381,
      "grad_norm": 0.1085980162024498,
      "learning_rate": 6.556624147537022e-05,
      "loss": 1.59,
      "step": 175840
    },
    {
      "epoch": 0.3470502038348345,
      "grad_norm": 0.12293184548616409,
      "learning_rate": 6.555990125100552e-05,
      "loss": 1.588,
      "step": 175872
    },
    {
      "epoch": 0.34711334979623093,
      "grad_norm": 0.11466878652572632,
      "learning_rate": 6.555356102664083e-05,
      "loss": 1.5894,
      "step": 175904
    },
    {
      "epoch": 0.3471764957576274,
      "grad_norm": 0.10859908908605576,
      "learning_rate": 6.554722080227615e-05,
      "loss": 1.5913,
      "step": 175936
    },
    {
      "epoch": 0.34723964171902383,
      "grad_norm": 0.11201232671737671,
      "learning_rate": 6.554088057791145e-05,
      "loss": 1.581,
      "step": 175968
    },
    {
      "epoch": 0.3473027876804203,
      "grad_norm": 0.10304196923971176,
      "learning_rate": 6.553454035354676e-05,
      "loss": 1.5722,
      "step": 176000
    },
    {
      "epoch": 0.3473659336418167,
      "grad_norm": 0.10468000173568726,
      "learning_rate": 6.552820012918207e-05,
      "loss": 1.598,
      "step": 176032
    },
    {
      "epoch": 0.3474290796032131,
      "grad_norm": 0.10960904508829117,
      "learning_rate": 6.552185990481738e-05,
      "loss": 1.6054,
      "step": 176064
    },
    {
      "epoch": 0.34749222556460957,
      "grad_norm": 0.11137548834085464,
      "learning_rate": 6.55155196804527e-05,
      "loss": 1.5967,
      "step": 176096
    },
    {
      "epoch": 0.347555371526006,
      "grad_norm": 0.10918192565441132,
      "learning_rate": 6.550917945608801e-05,
      "loss": 1.5815,
      "step": 176128
    },
    {
      "epoch": 0.34761851748740247,
      "grad_norm": 0.11181139945983887,
      "learning_rate": 6.550283923172331e-05,
      "loss": 1.5851,
      "step": 176160
    },
    {
      "epoch": 0.3476816634487989,
      "grad_norm": 0.10973136126995087,
      "learning_rate": 6.549649900735863e-05,
      "loss": 1.5897,
      "step": 176192
    },
    {
      "epoch": 0.34774480941019537,
      "grad_norm": 0.11366064846515656,
      "learning_rate": 6.549015878299394e-05,
      "loss": 1.584,
      "step": 176224
    },
    {
      "epoch": 0.34780795537159176,
      "grad_norm": 0.1052897498011589,
      "learning_rate": 6.548381855862925e-05,
      "loss": 1.5741,
      "step": 176256
    },
    {
      "epoch": 0.3478711013329882,
      "grad_norm": 0.1174076721072197,
      "learning_rate": 6.547747833426456e-05,
      "loss": 1.5786,
      "step": 176288
    },
    {
      "epoch": 0.34793424729438466,
      "grad_norm": 0.11128125339746475,
      "learning_rate": 6.547113810989987e-05,
      "loss": 1.5902,
      "step": 176320
    },
    {
      "epoch": 0.3479973932557811,
      "grad_norm": 0.11117946356534958,
      "learning_rate": 6.546479788553518e-05,
      "loss": 1.5909,
      "step": 176352
    },
    {
      "epoch": 0.34806053921717756,
      "grad_norm": 0.1188756674528122,
      "learning_rate": 6.545845766117049e-05,
      "loss": 1.5779,
      "step": 176384
    },
    {
      "epoch": 0.348123685178574,
      "grad_norm": 0.11214076727628708,
      "learning_rate": 6.545211743680579e-05,
      "loss": 1.589,
      "step": 176416
    },
    {
      "epoch": 0.3481868311399704,
      "grad_norm": 0.11385166645050049,
      "learning_rate": 6.54457772124411e-05,
      "loss": 1.5874,
      "step": 176448
    },
    {
      "epoch": 0.34824997710136685,
      "grad_norm": 0.1099756583571434,
      "learning_rate": 6.543943698807642e-05,
      "loss": 1.5779,
      "step": 176480
    },
    {
      "epoch": 0.3483131230627633,
      "grad_norm": 0.11098700761795044,
      "learning_rate": 6.543309676371173e-05,
      "loss": 1.5763,
      "step": 176512
    },
    {
      "epoch": 0.34837626902415975,
      "grad_norm": 0.10740330815315247,
      "learning_rate": 6.542675653934703e-05,
      "loss": 1.587,
      "step": 176544
    },
    {
      "epoch": 0.3484394149855562,
      "grad_norm": 0.11414963752031326,
      "learning_rate": 6.542041631498235e-05,
      "loss": 1.5904,
      "step": 176576
    },
    {
      "epoch": 0.34850256094695264,
      "grad_norm": 0.11507175862789154,
      "learning_rate": 6.541407609061766e-05,
      "loss": 1.593,
      "step": 176608
    },
    {
      "epoch": 0.34856570690834904,
      "grad_norm": 0.10829048603773117,
      "learning_rate": 6.540773586625297e-05,
      "loss": 1.5822,
      "step": 176640
    },
    {
      "epoch": 0.3486288528697455,
      "grad_norm": 0.11237678676843643,
      "learning_rate": 6.540139564188829e-05,
      "loss": 1.5847,
      "step": 176672
    },
    {
      "epoch": 0.34869199883114194,
      "grad_norm": 0.10770285129547119,
      "learning_rate": 6.539505541752359e-05,
      "loss": 1.5895,
      "step": 176704
    },
    {
      "epoch": 0.3487551447925384,
      "grad_norm": 0.11052990704774857,
      "learning_rate": 6.53887151931589e-05,
      "loss": 1.5883,
      "step": 176736
    },
    {
      "epoch": 0.34881829075393483,
      "grad_norm": 0.10141550749540329,
      "learning_rate": 6.538237496879422e-05,
      "loss": 1.5772,
      "step": 176768
    },
    {
      "epoch": 0.3488814367153313,
      "grad_norm": 0.10469438135623932,
      "learning_rate": 6.537603474442952e-05,
      "loss": 1.5846,
      "step": 176800
    },
    {
      "epoch": 0.3489445826767277,
      "grad_norm": 0.11447334289550781,
      "learning_rate": 6.536969452006482e-05,
      "loss": 1.5869,
      "step": 176832
    },
    {
      "epoch": 0.3490077286381241,
      "grad_norm": 0.10925020277500153,
      "learning_rate": 6.536335429570014e-05,
      "loss": 1.6061,
      "step": 176864
    },
    {
      "epoch": 0.3490708745995206,
      "grad_norm": 0.10936908423900604,
      "learning_rate": 6.535701407133545e-05,
      "loss": 1.5985,
      "step": 176896
    },
    {
      "epoch": 0.349134020560917,
      "grad_norm": 0.12269753962755203,
      "learning_rate": 6.535067384697077e-05,
      "loss": 1.5808,
      "step": 176928
    },
    {
      "epoch": 0.3491971665223135,
      "grad_norm": 0.1172834113240242,
      "learning_rate": 6.534433362260607e-05,
      "loss": 1.5936,
      "step": 176960
    },
    {
      "epoch": 0.3492603124837099,
      "grad_norm": 0.1211320161819458,
      "learning_rate": 6.533799339824138e-05,
      "loss": 1.5749,
      "step": 176992
    },
    {
      "epoch": 0.34932345844510637,
      "grad_norm": 0.11673060059547424,
      "learning_rate": 6.53316531738767e-05,
      "loss": 1.5905,
      "step": 177024
    },
    {
      "epoch": 0.34938660440650277,
      "grad_norm": 0.10510196536779404,
      "learning_rate": 6.532531294951201e-05,
      "loss": 1.5782,
      "step": 177056
    },
    {
      "epoch": 0.3494497503678992,
      "grad_norm": 0.11799222975969315,
      "learning_rate": 6.531897272514731e-05,
      "loss": 1.5822,
      "step": 177088
    },
    {
      "epoch": 0.34951289632929566,
      "grad_norm": 0.102109894156456,
      "learning_rate": 6.531263250078263e-05,
      "loss": 1.5913,
      "step": 177120
    },
    {
      "epoch": 0.3495760422906921,
      "grad_norm": 0.11021828651428223,
      "learning_rate": 6.530629227641794e-05,
      "loss": 1.5971,
      "step": 177152
    },
    {
      "epoch": 0.34963918825208856,
      "grad_norm": 0.10070045292377472,
      "learning_rate": 6.529995205205325e-05,
      "loss": 1.5885,
      "step": 177184
    },
    {
      "epoch": 0.349702334213485,
      "grad_norm": 0.10902727395296097,
      "learning_rate": 6.529361182768856e-05,
      "loss": 1.5893,
      "step": 177216
    },
    {
      "epoch": 0.3497654801748814,
      "grad_norm": 0.10452146828174591,
      "learning_rate": 6.528727160332386e-05,
      "loss": 1.5858,
      "step": 177248
    },
    {
      "epoch": 0.34982862613627785,
      "grad_norm": 0.11428716033697128,
      "learning_rate": 6.528093137895917e-05,
      "loss": 1.5676,
      "step": 177280
    },
    {
      "epoch": 0.3498917720976743,
      "grad_norm": 0.10437949746847153,
      "learning_rate": 6.527459115459449e-05,
      "loss": 1.5789,
      "step": 177312
    },
    {
      "epoch": 0.34995491805907075,
      "grad_norm": 0.1125776618719101,
      "learning_rate": 6.52682509302298e-05,
      "loss": 1.5818,
      "step": 177344
    },
    {
      "epoch": 0.3500180640204672,
      "grad_norm": 0.10946209728717804,
      "learning_rate": 6.52619107058651e-05,
      "loss": 1.5825,
      "step": 177376
    },
    {
      "epoch": 0.35008120998186365,
      "grad_norm": 0.10685037821531296,
      "learning_rate": 6.525557048150042e-05,
      "loss": 1.5807,
      "step": 177408
    },
    {
      "epoch": 0.35014435594326004,
      "grad_norm": 0.11477645486593246,
      "learning_rate": 6.524923025713573e-05,
      "loss": 1.5913,
      "step": 177440
    },
    {
      "epoch": 0.3502075019046565,
      "grad_norm": 0.1101703867316246,
      "learning_rate": 6.524289003277105e-05,
      "loss": 1.5832,
      "step": 177472
    },
    {
      "epoch": 0.35027064786605294,
      "grad_norm": 0.12006646394729614,
      "learning_rate": 6.523654980840635e-05,
      "loss": 1.5885,
      "step": 177504
    },
    {
      "epoch": 0.3503337938274494,
      "grad_norm": 0.12149064242839813,
      "learning_rate": 6.523020958404166e-05,
      "loss": 1.5878,
      "step": 177536
    },
    {
      "epoch": 0.35039693978884584,
      "grad_norm": 0.10605932772159576,
      "learning_rate": 6.522386935967698e-05,
      "loss": 1.5874,
      "step": 177568
    },
    {
      "epoch": 0.3504600857502423,
      "grad_norm": 0.1071997582912445,
      "learning_rate": 6.521752913531229e-05,
      "loss": 1.5831,
      "step": 177600
    },
    {
      "epoch": 0.3505232317116387,
      "grad_norm": 0.10740258544683456,
      "learning_rate": 6.521118891094759e-05,
      "loss": 1.5896,
      "step": 177632
    },
    {
      "epoch": 0.35058637767303513,
      "grad_norm": 0.11575182527303696,
      "learning_rate": 6.520484868658289e-05,
      "loss": 1.5911,
      "step": 177664
    },
    {
      "epoch": 0.3506495236344316,
      "grad_norm": 0.11003038287162781,
      "learning_rate": 6.51985084622182e-05,
      "loss": 1.5907,
      "step": 177696
    },
    {
      "epoch": 0.35071266959582803,
      "grad_norm": 0.12044735252857208,
      "learning_rate": 6.519216823785352e-05,
      "loss": 1.5838,
      "step": 177728
    },
    {
      "epoch": 0.3507758155572245,
      "grad_norm": 0.109721340239048,
      "learning_rate": 6.518582801348882e-05,
      "loss": 1.5891,
      "step": 177760
    },
    {
      "epoch": 0.35083896151862093,
      "grad_norm": 0.11115572601556778,
      "learning_rate": 6.517948778912414e-05,
      "loss": 1.583,
      "step": 177792
    },
    {
      "epoch": 0.3509021074800173,
      "grad_norm": 0.11124666780233383,
      "learning_rate": 6.517314756475945e-05,
      "loss": 1.5853,
      "step": 177824
    },
    {
      "epoch": 0.35096525344141377,
      "grad_norm": 0.11155048757791519,
      "learning_rate": 6.516680734039477e-05,
      "loss": 1.5947,
      "step": 177856
    },
    {
      "epoch": 0.3510283994028102,
      "grad_norm": 0.11156784743070602,
      "learning_rate": 6.516046711603007e-05,
      "loss": 1.5877,
      "step": 177888
    },
    {
      "epoch": 0.35109154536420667,
      "grad_norm": 0.10897670686244965,
      "learning_rate": 6.515412689166538e-05,
      "loss": 1.5977,
      "step": 177920
    },
    {
      "epoch": 0.3511546913256031,
      "grad_norm": 0.1054021343588829,
      "learning_rate": 6.51477866673007e-05,
      "loss": 1.5833,
      "step": 177952
    },
    {
      "epoch": 0.35121783728699957,
      "grad_norm": 0.11119698733091354,
      "learning_rate": 6.514144644293601e-05,
      "loss": 1.5949,
      "step": 177984
    },
    {
      "epoch": 0.35128098324839596,
      "grad_norm": 0.10701705515384674,
      "learning_rate": 6.513510621857131e-05,
      "loss": 1.5924,
      "step": 178016
    },
    {
      "epoch": 0.3513441292097924,
      "grad_norm": 0.10838789492845535,
      "learning_rate": 6.512876599420663e-05,
      "loss": 1.5879,
      "step": 178048
    },
    {
      "epoch": 0.35140727517118886,
      "grad_norm": 0.10813632607460022,
      "learning_rate": 6.512242576984193e-05,
      "loss": 1.5943,
      "step": 178080
    },
    {
      "epoch": 0.3514704211325853,
      "grad_norm": 0.10789559781551361,
      "learning_rate": 6.511608554547724e-05,
      "loss": 1.5738,
      "step": 178112
    },
    {
      "epoch": 0.35153356709398176,
      "grad_norm": 0.11196893453598022,
      "learning_rate": 6.510974532111256e-05,
      "loss": 1.592,
      "step": 178144
    },
    {
      "epoch": 0.3515967130553782,
      "grad_norm": 0.10628174245357513,
      "learning_rate": 6.510340509674786e-05,
      "loss": 1.5911,
      "step": 178176
    },
    {
      "epoch": 0.3516598590167746,
      "grad_norm": 0.11501415073871613,
      "learning_rate": 6.509706487238317e-05,
      "loss": 1.5891,
      "step": 178208
    },
    {
      "epoch": 0.35172300497817105,
      "grad_norm": 0.10158290714025497,
      "learning_rate": 6.509072464801849e-05,
      "loss": 1.5757,
      "step": 178240
    },
    {
      "epoch": 0.3517861509395675,
      "grad_norm": 0.12369918823242188,
      "learning_rate": 6.50843844236538e-05,
      "loss": 1.5879,
      "step": 178272
    },
    {
      "epoch": 0.35184929690096395,
      "grad_norm": 0.11359665542840958,
      "learning_rate": 6.50780441992891e-05,
      "loss": 1.5913,
      "step": 178304
    },
    {
      "epoch": 0.3519124428623604,
      "grad_norm": 0.10677289217710495,
      "learning_rate": 6.507170397492442e-05,
      "loss": 1.5814,
      "step": 178336
    },
    {
      "epoch": 0.35197558882375685,
      "grad_norm": 0.11651216447353363,
      "learning_rate": 6.506536375055973e-05,
      "loss": 1.5808,
      "step": 178368
    },
    {
      "epoch": 0.35203873478515324,
      "grad_norm": 0.10844559222459793,
      "learning_rate": 6.505902352619505e-05,
      "loss": 1.5811,
      "step": 178400
    },
    {
      "epoch": 0.3521018807465497,
      "grad_norm": 0.10941294580698013,
      "learning_rate": 6.505268330183035e-05,
      "loss": 1.593,
      "step": 178432
    },
    {
      "epoch": 0.35216502670794614,
      "grad_norm": 0.111540287733078,
      "learning_rate": 6.504634307746566e-05,
      "loss": 1.5875,
      "step": 178464
    },
    {
      "epoch": 0.3522281726693426,
      "grad_norm": 0.10734125226736069,
      "learning_rate": 6.504000285310096e-05,
      "loss": 1.5803,
      "step": 178496
    },
    {
      "epoch": 0.35229131863073904,
      "grad_norm": 0.10194408893585205,
      "learning_rate": 6.503366262873628e-05,
      "loss": 1.5845,
      "step": 178528
    },
    {
      "epoch": 0.3523544645921355,
      "grad_norm": 0.12050027400255203,
      "learning_rate": 6.502732240437158e-05,
      "loss": 1.5886,
      "step": 178560
    },
    {
      "epoch": 0.3524176105535319,
      "grad_norm": 0.10722140967845917,
      "learning_rate": 6.502098218000689e-05,
      "loss": 1.5785,
      "step": 178592
    },
    {
      "epoch": 0.35248075651492833,
      "grad_norm": 0.10467955470085144,
      "learning_rate": 6.50146419556422e-05,
      "loss": 1.574,
      "step": 178624
    },
    {
      "epoch": 0.3525439024763248,
      "grad_norm": 0.10946305841207504,
      "learning_rate": 6.500830173127752e-05,
      "loss": 1.5818,
      "step": 178656
    },
    {
      "epoch": 0.3526070484377212,
      "grad_norm": 0.10749588906764984,
      "learning_rate": 6.500196150691284e-05,
      "loss": 1.574,
      "step": 178688
    },
    {
      "epoch": 0.3526701943991177,
      "grad_norm": 0.10729070752859116,
      "learning_rate": 6.499562128254814e-05,
      "loss": 1.5817,
      "step": 178720
    },
    {
      "epoch": 0.3527333403605141,
      "grad_norm": 0.11417505890130997,
      "learning_rate": 6.498928105818345e-05,
      "loss": 1.5789,
      "step": 178752
    },
    {
      "epoch": 0.3527964863219106,
      "grad_norm": 0.11822691559791565,
      "learning_rate": 6.498294083381877e-05,
      "loss": 1.5818,
      "step": 178784
    },
    {
      "epoch": 0.35285963228330697,
      "grad_norm": 0.11468329280614853,
      "learning_rate": 6.497660060945408e-05,
      "loss": 1.5808,
      "step": 178816
    },
    {
      "epoch": 0.3529227782447034,
      "grad_norm": 0.10972462594509125,
      "learning_rate": 6.497026038508938e-05,
      "loss": 1.5785,
      "step": 178848
    },
    {
      "epoch": 0.35298592420609987,
      "grad_norm": 0.11151985824108124,
      "learning_rate": 6.49639201607247e-05,
      "loss": 1.5893,
      "step": 178880
    },
    {
      "epoch": 0.3530490701674963,
      "grad_norm": 0.11051822453737259,
      "learning_rate": 6.495757993636e-05,
      "loss": 1.5987,
      "step": 178912
    },
    {
      "epoch": 0.35311221612889276,
      "grad_norm": 0.1133396327495575,
      "learning_rate": 6.495123971199531e-05,
      "loss": 1.5903,
      "step": 178944
    },
    {
      "epoch": 0.3531753620902892,
      "grad_norm": 0.11220772564411163,
      "learning_rate": 6.494489948763061e-05,
      "loss": 1.584,
      "step": 178976
    },
    {
      "epoch": 0.3532385080516856,
      "grad_norm": 0.11535008251667023,
      "learning_rate": 6.493855926326593e-05,
      "loss": 1.5819,
      "step": 179008
    },
    {
      "epoch": 0.35330165401308206,
      "grad_norm": 0.10733187198638916,
      "learning_rate": 6.493221903890124e-05,
      "loss": 1.5979,
      "step": 179040
    },
    {
      "epoch": 0.3533647999744785,
      "grad_norm": 0.10744732618331909,
      "learning_rate": 6.492587881453656e-05,
      "loss": 1.587,
      "step": 179072
    },
    {
      "epoch": 0.35342794593587495,
      "grad_norm": 0.10880184173583984,
      "learning_rate": 6.491953859017186e-05,
      "loss": 1.5961,
      "step": 179104
    },
    {
      "epoch": 0.3534910918972714,
      "grad_norm": 0.1068880558013916,
      "learning_rate": 6.491319836580717e-05,
      "loss": 1.5836,
      "step": 179136
    },
    {
      "epoch": 0.35355423785866785,
      "grad_norm": 0.11626382917165756,
      "learning_rate": 6.490685814144249e-05,
      "loss": 1.5749,
      "step": 179168
    },
    {
      "epoch": 0.35361738382006425,
      "grad_norm": 0.10463514924049377,
      "learning_rate": 6.49005179170778e-05,
      "loss": 1.5714,
      "step": 179200
    },
    {
      "epoch": 0.3536805297814607,
      "grad_norm": 0.11262864619493484,
      "learning_rate": 6.489417769271312e-05,
      "loss": 1.5738,
      "step": 179232
    },
    {
      "epoch": 0.35374367574285714,
      "grad_norm": 0.10983620584011078,
      "learning_rate": 6.488783746834842e-05,
      "loss": 1.5801,
      "step": 179264
    },
    {
      "epoch": 0.3538068217042536,
      "grad_norm": 0.11021733283996582,
      "learning_rate": 6.488149724398372e-05,
      "loss": 1.597,
      "step": 179296
    },
    {
      "epoch": 0.35386996766565004,
      "grad_norm": 0.10846909880638123,
      "learning_rate": 6.487515701961903e-05,
      "loss": 1.5888,
      "step": 179328
    },
    {
      "epoch": 0.3539331136270465,
      "grad_norm": 0.1092953160405159,
      "learning_rate": 6.486881679525435e-05,
      "loss": 1.5821,
      "step": 179360
    },
    {
      "epoch": 0.3539962595884429,
      "grad_norm": 0.11354577541351318,
      "learning_rate": 6.486247657088965e-05,
      "loss": 1.5841,
      "step": 179392
    },
    {
      "epoch": 0.35405940554983933,
      "grad_norm": 0.1108522117137909,
      "learning_rate": 6.485613634652496e-05,
      "loss": 1.5808,
      "step": 179424
    },
    {
      "epoch": 0.3541225515112358,
      "grad_norm": 0.11519111692905426,
      "learning_rate": 6.484979612216028e-05,
      "loss": 1.5882,
      "step": 179456
    },
    {
      "epoch": 0.35418569747263223,
      "grad_norm": 0.1082671582698822,
      "learning_rate": 6.484345589779559e-05,
      "loss": 1.5898,
      "step": 179488
    },
    {
      "epoch": 0.3542488434340287,
      "grad_norm": 0.11377017945051193,
      "learning_rate": 6.483711567343089e-05,
      "loss": 1.5849,
      "step": 179520
    },
    {
      "epoch": 0.35431198939542513,
      "grad_norm": 0.11119108647108078,
      "learning_rate": 6.483077544906621e-05,
      "loss": 1.5893,
      "step": 179552
    },
    {
      "epoch": 0.3543751353568215,
      "grad_norm": 0.11232598125934601,
      "learning_rate": 6.482443522470152e-05,
      "loss": 1.589,
      "step": 179584
    },
    {
      "epoch": 0.354438281318218,
      "grad_norm": 0.10830990970134735,
      "learning_rate": 6.481809500033684e-05,
      "loss": 1.5762,
      "step": 179616
    },
    {
      "epoch": 0.3545014272796144,
      "grad_norm": 0.11087454855442047,
      "learning_rate": 6.481175477597214e-05,
      "loss": 1.5785,
      "step": 179648
    },
    {
      "epoch": 0.35456457324101087,
      "grad_norm": 0.11044250428676605,
      "learning_rate": 6.480541455160745e-05,
      "loss": 1.5842,
      "step": 179680
    },
    {
      "epoch": 0.3546277192024073,
      "grad_norm": 0.11231352388858795,
      "learning_rate": 6.479907432724275e-05,
      "loss": 1.5774,
      "step": 179712
    },
    {
      "epoch": 0.35469086516380377,
      "grad_norm": 0.11182203143835068,
      "learning_rate": 6.479273410287807e-05,
      "loss": 1.578,
      "step": 179744
    },
    {
      "epoch": 0.35475401112520016,
      "grad_norm": 0.1104021817445755,
      "learning_rate": 6.478639387851337e-05,
      "loss": 1.5822,
      "step": 179776
    },
    {
      "epoch": 0.3548171570865966,
      "grad_norm": 0.11247829347848892,
      "learning_rate": 6.478005365414868e-05,
      "loss": 1.5873,
      "step": 179808
    },
    {
      "epoch": 0.35488030304799306,
      "grad_norm": 0.11255551874637604,
      "learning_rate": 6.4773713429784e-05,
      "loss": 1.5823,
      "step": 179840
    },
    {
      "epoch": 0.3549434490093895,
      "grad_norm": 0.10529974848031998,
      "learning_rate": 6.476737320541931e-05,
      "loss": 1.5871,
      "step": 179872
    },
    {
      "epoch": 0.35500659497078596,
      "grad_norm": 0.10930514335632324,
      "learning_rate": 6.476103298105463e-05,
      "loss": 1.585,
      "step": 179904
    },
    {
      "epoch": 0.3550697409321824,
      "grad_norm": 0.10999929904937744,
      "learning_rate": 6.475469275668993e-05,
      "loss": 1.5813,
      "step": 179936
    },
    {
      "epoch": 0.3551328868935788,
      "grad_norm": 0.11748002469539642,
      "learning_rate": 6.474835253232524e-05,
      "loss": 1.5888,
      "step": 179968
    },
    {
      "epoch": 0.35519603285497525,
      "grad_norm": 0.11176477372646332,
      "learning_rate": 6.474201230796056e-05,
      "loss": 1.5967,
      "step": 180000
    },
    {
      "epoch": 0.3552591788163717,
      "grad_norm": 0.11481215059757233,
      "learning_rate": 6.473567208359587e-05,
      "loss": 1.5936,
      "step": 180032
    },
    {
      "epoch": 0.35532232477776815,
      "grad_norm": 0.10812889039516449,
      "learning_rate": 6.472933185923117e-05,
      "loss": 1.5836,
      "step": 180064
    },
    {
      "epoch": 0.3553854707391646,
      "grad_norm": 0.10911864042282104,
      "learning_rate": 6.472299163486649e-05,
      "loss": 1.5993,
      "step": 180096
    },
    {
      "epoch": 0.35544861670056105,
      "grad_norm": 0.10474446415901184,
      "learning_rate": 6.471665141050179e-05,
      "loss": 1.5831,
      "step": 180128
    },
    {
      "epoch": 0.35551176266195744,
      "grad_norm": 0.11591237783432007,
      "learning_rate": 6.47103111861371e-05,
      "loss": 1.5909,
      "step": 180160
    },
    {
      "epoch": 0.3555749086233539,
      "grad_norm": 0.10848182439804077,
      "learning_rate": 6.47039709617724e-05,
      "loss": 1.5913,
      "step": 180192
    },
    {
      "epoch": 0.35563805458475034,
      "grad_norm": 0.10682623088359833,
      "learning_rate": 6.469763073740772e-05,
      "loss": 1.5791,
      "step": 180224
    },
    {
      "epoch": 0.3557012005461468,
      "grad_norm": 0.1151672974228859,
      "learning_rate": 6.469129051304303e-05,
      "loss": 1.5782,
      "step": 180256
    },
    {
      "epoch": 0.35576434650754324,
      "grad_norm": 0.11238664388656616,
      "learning_rate": 6.468495028867835e-05,
      "loss": 1.5824,
      "step": 180288
    },
    {
      "epoch": 0.3558274924689397,
      "grad_norm": 0.110641248524189,
      "learning_rate": 6.467861006431365e-05,
      "loss": 1.5814,
      "step": 180320
    },
    {
      "epoch": 0.3558906384303361,
      "grad_norm": 0.10974923521280289,
      "learning_rate": 6.467226983994896e-05,
      "loss": 1.5838,
      "step": 180352
    },
    {
      "epoch": 0.35595378439173253,
      "grad_norm": 0.1073017567396164,
      "learning_rate": 6.466592961558428e-05,
      "loss": 1.5786,
      "step": 180384
    },
    {
      "epoch": 0.356016930353129,
      "grad_norm": 0.10749690234661102,
      "learning_rate": 6.465958939121959e-05,
      "loss": 1.574,
      "step": 180416
    },
    {
      "epoch": 0.35608007631452543,
      "grad_norm": 0.10572580993175507,
      "learning_rate": 6.465324916685489e-05,
      "loss": 1.5825,
      "step": 180448
    },
    {
      "epoch": 0.3561432222759219,
      "grad_norm": 0.10467702150344849,
      "learning_rate": 6.464690894249021e-05,
      "loss": 1.5886,
      "step": 180480
    },
    {
      "epoch": 0.3562063682373183,
      "grad_norm": 0.11328021436929703,
      "learning_rate": 6.464056871812552e-05,
      "loss": 1.5801,
      "step": 180512
    },
    {
      "epoch": 0.3562695141987148,
      "grad_norm": 0.1265615075826645,
      "learning_rate": 6.463422849376082e-05,
      "loss": 1.584,
      "step": 180544
    },
    {
      "epoch": 0.35633266016011117,
      "grad_norm": 0.1107073575258255,
      "learning_rate": 6.462788826939614e-05,
      "loss": 1.5918,
      "step": 180576
    },
    {
      "epoch": 0.3563958061215076,
      "grad_norm": 0.1132175475358963,
      "learning_rate": 6.462154804503144e-05,
      "loss": 1.5858,
      "step": 180608
    },
    {
      "epoch": 0.35645895208290407,
      "grad_norm": 0.1220729649066925,
      "learning_rate": 6.461520782066675e-05,
      "loss": 1.5916,
      "step": 180640
    },
    {
      "epoch": 0.3565220980443005,
      "grad_norm": 0.10473572462797165,
      "learning_rate": 6.460886759630207e-05,
      "loss": 1.5934,
      "step": 180672
    },
    {
      "epoch": 0.35658524400569697,
      "grad_norm": 0.10555895417928696,
      "learning_rate": 6.460252737193738e-05,
      "loss": 1.5913,
      "step": 180704
    },
    {
      "epoch": 0.3566483899670934,
      "grad_norm": 0.11519496142864227,
      "learning_rate": 6.459618714757268e-05,
      "loss": 1.584,
      "step": 180736
    },
    {
      "epoch": 0.3567115359284898,
      "grad_norm": 0.10348035395145416,
      "learning_rate": 6.4589846923208e-05,
      "loss": 1.5882,
      "step": 180768
    },
    {
      "epoch": 0.35677468188988626,
      "grad_norm": 0.10846167057752609,
      "learning_rate": 6.458350669884331e-05,
      "loss": 1.5798,
      "step": 180800
    },
    {
      "epoch": 0.3568378278512827,
      "grad_norm": 0.10748612135648727,
      "learning_rate": 6.457716647447863e-05,
      "loss": 1.5933,
      "step": 180832
    },
    {
      "epoch": 0.35690097381267916,
      "grad_norm": 0.10867191851139069,
      "learning_rate": 6.457082625011393e-05,
      "loss": 1.5794,
      "step": 180864
    },
    {
      "epoch": 0.3569641197740756,
      "grad_norm": 0.10982093214988708,
      "learning_rate": 6.456448602574924e-05,
      "loss": 1.5834,
      "step": 180896
    },
    {
      "epoch": 0.35702726573547205,
      "grad_norm": 0.11110133677721024,
      "learning_rate": 6.455814580138456e-05,
      "loss": 1.5785,
      "step": 180928
    },
    {
      "epoch": 0.35709041169686845,
      "grad_norm": 0.11195295304059982,
      "learning_rate": 6.455180557701986e-05,
      "loss": 1.5856,
      "step": 180960
    },
    {
      "epoch": 0.3571535576582649,
      "grad_norm": 0.12834830582141876,
      "learning_rate": 6.454546535265516e-05,
      "loss": 1.5893,
      "step": 180992
    },
    {
      "epoch": 0.35721670361966135,
      "grad_norm": 0.10547395795583725,
      "learning_rate": 6.453912512829047e-05,
      "loss": 1.5856,
      "step": 181024
    },
    {
      "epoch": 0.3572798495810578,
      "grad_norm": 0.11166314035654068,
      "learning_rate": 6.453278490392579e-05,
      "loss": 1.5797,
      "step": 181056
    },
    {
      "epoch": 0.35734299554245424,
      "grad_norm": 0.11106620728969574,
      "learning_rate": 6.45264446795611e-05,
      "loss": 1.5852,
      "step": 181088
    },
    {
      "epoch": 0.3574061415038507,
      "grad_norm": 0.10392887145280838,
      "learning_rate": 6.45201044551964e-05,
      "loss": 1.5699,
      "step": 181120
    },
    {
      "epoch": 0.3574692874652471,
      "grad_norm": 0.11312718689441681,
      "learning_rate": 6.451376423083172e-05,
      "loss": 1.5799,
      "step": 181152
    },
    {
      "epoch": 0.35753243342664354,
      "grad_norm": 0.1205592006444931,
      "learning_rate": 6.450742400646703e-05,
      "loss": 1.5844,
      "step": 181184
    },
    {
      "epoch": 0.35759557938804,
      "grad_norm": 0.11224983632564545,
      "learning_rate": 6.450108378210235e-05,
      "loss": 1.5688,
      "step": 181216
    },
    {
      "epoch": 0.35765872534943643,
      "grad_norm": 0.11771758645772934,
      "learning_rate": 6.449474355773766e-05,
      "loss": 1.5784,
      "step": 181248
    },
    {
      "epoch": 0.3577218713108329,
      "grad_norm": 0.11228399723768234,
      "learning_rate": 6.448840333337296e-05,
      "loss": 1.5861,
      "step": 181280
    },
    {
      "epoch": 0.35778501727222933,
      "grad_norm": 0.11208807677030563,
      "learning_rate": 6.448206310900828e-05,
      "loss": 1.5612,
      "step": 181312
    },
    {
      "epoch": 0.3578481632336257,
      "grad_norm": 0.10504967719316483,
      "learning_rate": 6.447572288464359e-05,
      "loss": 1.5812,
      "step": 181344
    },
    {
      "epoch": 0.3579113091950222,
      "grad_norm": 0.10955792665481567,
      "learning_rate": 6.446938266027889e-05,
      "loss": 1.5809,
      "step": 181376
    },
    {
      "epoch": 0.3579744551564186,
      "grad_norm": 0.11303912103176117,
      "learning_rate": 6.44630424359142e-05,
      "loss": 1.5843,
      "step": 181408
    },
    {
      "epoch": 0.3580376011178151,
      "grad_norm": 0.1057058796286583,
      "learning_rate": 6.445670221154951e-05,
      "loss": 1.5789,
      "step": 181440
    },
    {
      "epoch": 0.3581007470792115,
      "grad_norm": 0.11012724041938782,
      "learning_rate": 6.445036198718482e-05,
      "loss": 1.5933,
      "step": 181472
    },
    {
      "epoch": 0.358163893040608,
      "grad_norm": 0.11209618300199509,
      "learning_rate": 6.444402176282014e-05,
      "loss": 1.589,
      "step": 181504
    },
    {
      "epoch": 0.35822703900200437,
      "grad_norm": 0.11269824951887131,
      "learning_rate": 6.443768153845544e-05,
      "loss": 1.591,
      "step": 181536
    },
    {
      "epoch": 0.3582901849634008,
      "grad_norm": 0.11304093897342682,
      "learning_rate": 6.443134131409075e-05,
      "loss": 1.609,
      "step": 181568
    },
    {
      "epoch": 0.35835333092479726,
      "grad_norm": 0.11303514242172241,
      "learning_rate": 6.442500108972607e-05,
      "loss": 1.5863,
      "step": 181600
    },
    {
      "epoch": 0.3584164768861937,
      "grad_norm": 0.10825137048959732,
      "learning_rate": 6.441866086536138e-05,
      "loss": 1.5748,
      "step": 181632
    },
    {
      "epoch": 0.35847962284759016,
      "grad_norm": 0.12855233252048492,
      "learning_rate": 6.441232064099668e-05,
      "loss": 1.588,
      "step": 181664
    },
    {
      "epoch": 0.3585427688089866,
      "grad_norm": 0.10915649682283401,
      "learning_rate": 6.4405980416632e-05,
      "loss": 1.5745,
      "step": 181696
    },
    {
      "epoch": 0.358605914770383,
      "grad_norm": 0.1154591366648674,
      "learning_rate": 6.439964019226731e-05,
      "loss": 1.5836,
      "step": 181728
    },
    {
      "epoch": 0.35866906073177945,
      "grad_norm": 0.11371373385190964,
      "learning_rate": 6.439329996790263e-05,
      "loss": 1.5863,
      "step": 181760
    },
    {
      "epoch": 0.3587322066931759,
      "grad_norm": 0.1083037257194519,
      "learning_rate": 6.438695974353793e-05,
      "loss": 1.587,
      "step": 181792
    },
    {
      "epoch": 0.35879535265457235,
      "grad_norm": 0.10600948333740234,
      "learning_rate": 6.438061951917323e-05,
      "loss": 1.585,
      "step": 181824
    },
    {
      "epoch": 0.3588584986159688,
      "grad_norm": 0.110209159553051,
      "learning_rate": 6.437427929480854e-05,
      "loss": 1.5848,
      "step": 181856
    },
    {
      "epoch": 0.35892164457736525,
      "grad_norm": 0.10848981142044067,
      "learning_rate": 6.436793907044386e-05,
      "loss": 1.5907,
      "step": 181888
    },
    {
      "epoch": 0.35898479053876164,
      "grad_norm": 0.10719979554414749,
      "learning_rate": 6.436159884607917e-05,
      "loss": 1.5727,
      "step": 181920
    },
    {
      "epoch": 0.3590479365001581,
      "grad_norm": 0.10645796358585358,
      "learning_rate": 6.435525862171447e-05,
      "loss": 1.5879,
      "step": 181952
    },
    {
      "epoch": 0.35911108246155454,
      "grad_norm": 0.1148567944765091,
      "learning_rate": 6.434891839734979e-05,
      "loss": 1.5841,
      "step": 181984
    },
    {
      "epoch": 0.359174228422951,
      "grad_norm": 0.11215532571077347,
      "learning_rate": 6.43425781729851e-05,
      "loss": 1.5902,
      "step": 182016
    },
    {
      "epoch": 0.35923737438434744,
      "grad_norm": 0.11848314106464386,
      "learning_rate": 6.433623794862042e-05,
      "loss": 1.5895,
      "step": 182048
    },
    {
      "epoch": 0.3593005203457439,
      "grad_norm": 0.12311811000108719,
      "learning_rate": 6.432989772425572e-05,
      "loss": 1.5816,
      "step": 182080
    },
    {
      "epoch": 0.35936366630714034,
      "grad_norm": 0.10764306783676147,
      "learning_rate": 6.432355749989103e-05,
      "loss": 1.5952,
      "step": 182112
    },
    {
      "epoch": 0.35942681226853673,
      "grad_norm": 0.10449346899986267,
      "learning_rate": 6.431721727552635e-05,
      "loss": 1.5855,
      "step": 182144
    },
    {
      "epoch": 0.3594899582299332,
      "grad_norm": 0.11061811447143555,
      "learning_rate": 6.431087705116165e-05,
      "loss": 1.5842,
      "step": 182176
    },
    {
      "epoch": 0.35955310419132963,
      "grad_norm": 0.1019996777176857,
      "learning_rate": 6.430453682679696e-05,
      "loss": 1.6047,
      "step": 182208
    },
    {
      "epoch": 0.3596162501527261,
      "grad_norm": 0.10651903599500656,
      "learning_rate": 6.429819660243226e-05,
      "loss": 1.5804,
      "step": 182240
    },
    {
      "epoch": 0.35967939611412253,
      "grad_norm": 0.11617961525917053,
      "learning_rate": 6.429185637806758e-05,
      "loss": 1.5903,
      "step": 182272
    },
    {
      "epoch": 0.359742542075519,
      "grad_norm": 0.1261436939239502,
      "learning_rate": 6.42855161537029e-05,
      "loss": 1.5817,
      "step": 182304
    },
    {
      "epoch": 0.35980568803691537,
      "grad_norm": 0.1021326333284378,
      "learning_rate": 6.42791759293382e-05,
      "loss": 1.5903,
      "step": 182336
    },
    {
      "epoch": 0.3598688339983118,
      "grad_norm": 0.1128995269536972,
      "learning_rate": 6.427283570497351e-05,
      "loss": 1.5831,
      "step": 182368
    },
    {
      "epoch": 0.35993197995970827,
      "grad_norm": 0.10895900428295135,
      "learning_rate": 6.426649548060882e-05,
      "loss": 1.5926,
      "step": 182400
    },
    {
      "epoch": 0.3599951259211047,
      "grad_norm": 0.11161261051893234,
      "learning_rate": 6.426015525624414e-05,
      "loss": 1.5923,
      "step": 182432
    },
    {
      "epoch": 0.36005827188250117,
      "grad_norm": 0.12293332815170288,
      "learning_rate": 6.425381503187944e-05,
      "loss": 1.5899,
      "step": 182464
    },
    {
      "epoch": 0.3601214178438976,
      "grad_norm": 0.11045447736978531,
      "learning_rate": 6.424747480751475e-05,
      "loss": 1.585,
      "step": 182496
    },
    {
      "epoch": 0.360184563805294,
      "grad_norm": 0.11175920814275742,
      "learning_rate": 6.424113458315007e-05,
      "loss": 1.5801,
      "step": 182528
    },
    {
      "epoch": 0.36024770976669046,
      "grad_norm": 0.11152239143848419,
      "learning_rate": 6.423479435878538e-05,
      "loss": 1.5836,
      "step": 182560
    },
    {
      "epoch": 0.3603108557280869,
      "grad_norm": 0.10802054405212402,
      "learning_rate": 6.422845413442068e-05,
      "loss": 1.576,
      "step": 182592
    },
    {
      "epoch": 0.36037400168948336,
      "grad_norm": 0.10338658094406128,
      "learning_rate": 6.4222113910056e-05,
      "loss": 1.5804,
      "step": 182624
    },
    {
      "epoch": 0.3604371476508798,
      "grad_norm": 0.11012491583824158,
      "learning_rate": 6.42157736856913e-05,
      "loss": 1.5894,
      "step": 182656
    },
    {
      "epoch": 0.36050029361227626,
      "grad_norm": 0.10463359206914902,
      "learning_rate": 6.420943346132661e-05,
      "loss": 1.5779,
      "step": 182688
    },
    {
      "epoch": 0.36056343957367265,
      "grad_norm": 0.10699468106031418,
      "learning_rate": 6.420309323696193e-05,
      "loss": 1.5997,
      "step": 182720
    },
    {
      "epoch": 0.3606265855350691,
      "grad_norm": 0.11880798637866974,
      "learning_rate": 6.419675301259723e-05,
      "loss": 1.5779,
      "step": 182752
    },
    {
      "epoch": 0.36068973149646555,
      "grad_norm": 0.1022857204079628,
      "learning_rate": 6.419041278823254e-05,
      "loss": 1.5948,
      "step": 182784
    },
    {
      "epoch": 0.360752877457862,
      "grad_norm": 0.11548960953950882,
      "learning_rate": 6.418407256386786e-05,
      "loss": 1.5934,
      "step": 182816
    },
    {
      "epoch": 0.36081602341925845,
      "grad_norm": 0.11352226883172989,
      "learning_rate": 6.417773233950317e-05,
      "loss": 1.5802,
      "step": 182848
    },
    {
      "epoch": 0.3608791693806549,
      "grad_norm": 0.12137927860021591,
      "learning_rate": 6.417139211513847e-05,
      "loss": 1.5766,
      "step": 182880
    },
    {
      "epoch": 0.3609423153420513,
      "grad_norm": 0.11009381711483002,
      "learning_rate": 6.416505189077379e-05,
      "loss": 1.5922,
      "step": 182912
    },
    {
      "epoch": 0.36100546130344774,
      "grad_norm": 0.10775876045227051,
      "learning_rate": 6.41587116664091e-05,
      "loss": 1.58,
      "step": 182944
    },
    {
      "epoch": 0.3610686072648442,
      "grad_norm": 0.10785428434610367,
      "learning_rate": 6.415237144204442e-05,
      "loss": 1.5828,
      "step": 182976
    },
    {
      "epoch": 0.36113175322624064,
      "grad_norm": 0.1145273894071579,
      "learning_rate": 6.414603121767972e-05,
      "loss": 1.5927,
      "step": 183008
    },
    {
      "epoch": 0.3611948991876371,
      "grad_norm": 0.10673095285892487,
      "learning_rate": 6.413969099331503e-05,
      "loss": 1.5785,
      "step": 183040
    },
    {
      "epoch": 0.36125804514903354,
      "grad_norm": 0.11440155655145645,
      "learning_rate": 6.413335076895033e-05,
      "loss": 1.5913,
      "step": 183072
    },
    {
      "epoch": 0.36132119111042993,
      "grad_norm": 0.11792333424091339,
      "learning_rate": 6.412701054458565e-05,
      "loss": 1.5923,
      "step": 183104
    },
    {
      "epoch": 0.3613843370718264,
      "grad_norm": 0.10590378940105438,
      "learning_rate": 6.412067032022095e-05,
      "loss": 1.5758,
      "step": 183136
    },
    {
      "epoch": 0.3614474830332228,
      "grad_norm": 0.10517742484807968,
      "learning_rate": 6.411433009585626e-05,
      "loss": 1.5954,
      "step": 183168
    },
    {
      "epoch": 0.3615106289946193,
      "grad_norm": 0.11259836703538895,
      "learning_rate": 6.410798987149158e-05,
      "loss": 1.5927,
      "step": 183200
    },
    {
      "epoch": 0.3615737749560157,
      "grad_norm": 0.10897098481655121,
      "learning_rate": 6.41016496471269e-05,
      "loss": 1.5876,
      "step": 183232
    },
    {
      "epoch": 0.3616369209174122,
      "grad_norm": 0.10789348930120468,
      "learning_rate": 6.409530942276221e-05,
      "loss": 1.5839,
      "step": 183264
    },
    {
      "epoch": 0.36170006687880857,
      "grad_norm": 0.10714706778526306,
      "learning_rate": 6.408896919839751e-05,
      "loss": 1.5846,
      "step": 183296
    },
    {
      "epoch": 0.361763212840205,
      "grad_norm": 0.11360534280538559,
      "learning_rate": 6.408262897403282e-05,
      "loss": 1.5952,
      "step": 183328
    },
    {
      "epoch": 0.36182635880160147,
      "grad_norm": 0.11198006570339203,
      "learning_rate": 6.407628874966814e-05,
      "loss": 1.5902,
      "step": 183360
    },
    {
      "epoch": 0.3618895047629979,
      "grad_norm": 0.11150224506855011,
      "learning_rate": 6.406994852530345e-05,
      "loss": 1.5863,
      "step": 183392
    },
    {
      "epoch": 0.36195265072439436,
      "grad_norm": 0.11451698839664459,
      "learning_rate": 6.406360830093875e-05,
      "loss": 1.5852,
      "step": 183424
    },
    {
      "epoch": 0.3620157966857908,
      "grad_norm": 0.11168801784515381,
      "learning_rate": 6.405726807657406e-05,
      "loss": 1.5928,
      "step": 183456
    },
    {
      "epoch": 0.3620789426471872,
      "grad_norm": 0.11182107776403427,
      "learning_rate": 6.405092785220937e-05,
      "loss": 1.5982,
      "step": 183488
    },
    {
      "epoch": 0.36214208860858366,
      "grad_norm": 0.11775576323270798,
      "learning_rate": 6.404458762784468e-05,
      "loss": 1.5818,
      "step": 183520
    },
    {
      "epoch": 0.3622052345699801,
      "grad_norm": 0.10620637983083725,
      "learning_rate": 6.403824740347999e-05,
      "loss": 1.5794,
      "step": 183552
    },
    {
      "epoch": 0.36226838053137655,
      "grad_norm": 0.11593019962310791,
      "learning_rate": 6.40319071791153e-05,
      "loss": 1.5875,
      "step": 183584
    },
    {
      "epoch": 0.362331526492773,
      "grad_norm": 0.11101772636175156,
      "learning_rate": 6.402556695475061e-05,
      "loss": 1.5867,
      "step": 183616
    },
    {
      "epoch": 0.36239467245416945,
      "grad_norm": 0.11641628295183182,
      "learning_rate": 6.401922673038593e-05,
      "loss": 1.5832,
      "step": 183648
    },
    {
      "epoch": 0.36245781841556585,
      "grad_norm": 0.10823922604322433,
      "learning_rate": 6.401288650602123e-05,
      "loss": 1.5849,
      "step": 183680
    },
    {
      "epoch": 0.3625209643769623,
      "grad_norm": 0.10185832530260086,
      "learning_rate": 6.400654628165654e-05,
      "loss": 1.5705,
      "step": 183712
    },
    {
      "epoch": 0.36258411033835874,
      "grad_norm": 0.1149238720536232,
      "learning_rate": 6.400020605729186e-05,
      "loss": 1.5872,
      "step": 183744
    },
    {
      "epoch": 0.3626472562997552,
      "grad_norm": 0.10884139686822891,
      "learning_rate": 6.399386583292717e-05,
      "loss": 1.5804,
      "step": 183776
    },
    {
      "epoch": 0.36271040226115164,
      "grad_norm": 0.11320940405130386,
      "learning_rate": 6.398752560856247e-05,
      "loss": 1.5896,
      "step": 183808
    },
    {
      "epoch": 0.3627735482225481,
      "grad_norm": 0.12907955050468445,
      "learning_rate": 6.398118538419779e-05,
      "loss": 1.5802,
      "step": 183840
    },
    {
      "epoch": 0.36283669418394454,
      "grad_norm": 0.11195775121450424,
      "learning_rate": 6.397484515983309e-05,
      "loss": 1.5783,
      "step": 183872
    },
    {
      "epoch": 0.36289984014534094,
      "grad_norm": 0.11155863851308823,
      "learning_rate": 6.39685049354684e-05,
      "loss": 1.5717,
      "step": 183904
    },
    {
      "epoch": 0.3629629861067374,
      "grad_norm": 0.11714331060647964,
      "learning_rate": 6.396216471110372e-05,
      "loss": 1.5828,
      "step": 183936
    },
    {
      "epoch": 0.36302613206813383,
      "grad_norm": 0.1108788549900055,
      "learning_rate": 6.395582448673902e-05,
      "loss": 1.5884,
      "step": 183968
    },
    {
      "epoch": 0.3630892780295303,
      "grad_norm": 0.11498186737298965,
      "learning_rate": 6.394948426237433e-05,
      "loss": 1.5828,
      "step": 184000
    },
    {
      "epoch": 0.36315242399092673,
      "grad_norm": 0.10536820441484451,
      "learning_rate": 6.394314403800965e-05,
      "loss": 1.5842,
      "step": 184032
    },
    {
      "epoch": 0.3632155699523232,
      "grad_norm": 0.11589545011520386,
      "learning_rate": 6.393680381364496e-05,
      "loss": 1.5773,
      "step": 184064
    },
    {
      "epoch": 0.3632787159137196,
      "grad_norm": 0.11279499530792236,
      "learning_rate": 6.393046358928027e-05,
      "loss": 1.5928,
      "step": 184096
    },
    {
      "epoch": 0.363341861875116,
      "grad_norm": 0.10008393973112106,
      "learning_rate": 6.392412336491558e-05,
      "loss": 1.5799,
      "step": 184128
    },
    {
      "epoch": 0.3634050078365125,
      "grad_norm": 0.10640986263751984,
      "learning_rate": 6.39177831405509e-05,
      "loss": 1.5927,
      "step": 184160
    },
    {
      "epoch": 0.3634681537979089,
      "grad_norm": 0.11636866629123688,
      "learning_rate": 6.391144291618621e-05,
      "loss": 1.5905,
      "step": 184192
    },
    {
      "epoch": 0.36353129975930537,
      "grad_norm": 0.11361801624298096,
      "learning_rate": 6.390510269182151e-05,
      "loss": 1.5733,
      "step": 184224
    },
    {
      "epoch": 0.3635944457207018,
      "grad_norm": 0.113465316593647,
      "learning_rate": 6.389876246745682e-05,
      "loss": 1.5875,
      "step": 184256
    },
    {
      "epoch": 0.3636575916820982,
      "grad_norm": 0.10863377153873444,
      "learning_rate": 6.389242224309213e-05,
      "loss": 1.5766,
      "step": 184288
    },
    {
      "epoch": 0.36372073764349466,
      "grad_norm": 0.11061983555555344,
      "learning_rate": 6.388608201872744e-05,
      "loss": 1.5831,
      "step": 184320
    },
    {
      "epoch": 0.3637838836048911,
      "grad_norm": 0.1096784844994545,
      "learning_rate": 6.387974179436274e-05,
      "loss": 1.5817,
      "step": 184352
    },
    {
      "epoch": 0.36384702956628756,
      "grad_norm": 0.10557596385478973,
      "learning_rate": 6.387340156999806e-05,
      "loss": 1.5825,
      "step": 184384
    },
    {
      "epoch": 0.363910175527684,
      "grad_norm": 0.11243075132369995,
      "learning_rate": 6.386706134563337e-05,
      "loss": 1.597,
      "step": 184416
    },
    {
      "epoch": 0.36397332148908046,
      "grad_norm": 0.11064084619283676,
      "learning_rate": 6.386072112126868e-05,
      "loss": 1.5931,
      "step": 184448
    },
    {
      "epoch": 0.36403646745047685,
      "grad_norm": 0.10731770843267441,
      "learning_rate": 6.385438089690399e-05,
      "loss": 1.5895,
      "step": 184480
    },
    {
      "epoch": 0.3640996134118733,
      "grad_norm": 0.11211124062538147,
      "learning_rate": 6.38480406725393e-05,
      "loss": 1.5773,
      "step": 184512
    },
    {
      "epoch": 0.36416275937326975,
      "grad_norm": 0.11388519406318665,
      "learning_rate": 6.384170044817461e-05,
      "loss": 1.5924,
      "step": 184544
    },
    {
      "epoch": 0.3642259053346662,
      "grad_norm": 0.12159216403961182,
      "learning_rate": 6.383536022380993e-05,
      "loss": 1.5875,
      "step": 184576
    },
    {
      "epoch": 0.36428905129606265,
      "grad_norm": 0.11033236235380173,
      "learning_rate": 6.382901999944524e-05,
      "loss": 1.5909,
      "step": 184608
    },
    {
      "epoch": 0.3643521972574591,
      "grad_norm": 0.10566152632236481,
      "learning_rate": 6.382267977508054e-05,
      "loss": 1.5676,
      "step": 184640
    },
    {
      "epoch": 0.3644153432188555,
      "grad_norm": 0.1056530699133873,
      "learning_rate": 6.381633955071586e-05,
      "loss": 1.6003,
      "step": 184672
    },
    {
      "epoch": 0.36447848918025194,
      "grad_norm": 0.10893242806196213,
      "learning_rate": 6.380999932635116e-05,
      "loss": 1.5796,
      "step": 184704
    },
    {
      "epoch": 0.3645416351416484,
      "grad_norm": 0.10994032770395279,
      "learning_rate": 6.380365910198648e-05,
      "loss": 1.5853,
      "step": 184736
    },
    {
      "epoch": 0.36460478110304484,
      "grad_norm": 0.11297830194234848,
      "learning_rate": 6.379731887762178e-05,
      "loss": 1.5887,
      "step": 184768
    },
    {
      "epoch": 0.3646679270644413,
      "grad_norm": 0.11538960784673691,
      "learning_rate": 6.379097865325709e-05,
      "loss": 1.5807,
      "step": 184800
    },
    {
      "epoch": 0.36473107302583774,
      "grad_norm": 0.11083732545375824,
      "learning_rate": 6.37846384288924e-05,
      "loss": 1.5869,
      "step": 184832
    },
    {
      "epoch": 0.36479421898723413,
      "grad_norm": 0.11004041135311127,
      "learning_rate": 6.377829820452772e-05,
      "loss": 1.5857,
      "step": 184864
    },
    {
      "epoch": 0.3648573649486306,
      "grad_norm": 0.10743048787117004,
      "learning_rate": 6.377195798016302e-05,
      "loss": 1.5787,
      "step": 184896
    },
    {
      "epoch": 0.36492051091002703,
      "grad_norm": 0.1123712956905365,
      "learning_rate": 6.376561775579834e-05,
      "loss": 1.583,
      "step": 184928
    },
    {
      "epoch": 0.3649836568714235,
      "grad_norm": 0.11037390679121017,
      "learning_rate": 6.375927753143365e-05,
      "loss": 1.5869,
      "step": 184960
    },
    {
      "epoch": 0.3650468028328199,
      "grad_norm": 0.10735499858856201,
      "learning_rate": 6.375293730706896e-05,
      "loss": 1.5662,
      "step": 184992
    },
    {
      "epoch": 0.3651099487942164,
      "grad_norm": 0.10350576043128967,
      "learning_rate": 6.374659708270427e-05,
      "loss": 1.5813,
      "step": 185024
    },
    {
      "epoch": 0.36517309475561277,
      "grad_norm": 0.11766477674245834,
      "learning_rate": 6.374025685833958e-05,
      "loss": 1.5849,
      "step": 185056
    },
    {
      "epoch": 0.3652362407170092,
      "grad_norm": 0.11200088262557983,
      "learning_rate": 6.37339166339749e-05,
      "loss": 1.5904,
      "step": 185088
    },
    {
      "epoch": 0.36529938667840567,
      "grad_norm": 0.1167105957865715,
      "learning_rate": 6.37275764096102e-05,
      "loss": 1.5947,
      "step": 185120
    },
    {
      "epoch": 0.3653625326398021,
      "grad_norm": 0.11037057638168335,
      "learning_rate": 6.37212361852455e-05,
      "loss": 1.5628,
      "step": 185152
    },
    {
      "epoch": 0.36542567860119857,
      "grad_norm": 0.1177527904510498,
      "learning_rate": 6.371489596088081e-05,
      "loss": 1.5811,
      "step": 185184
    },
    {
      "epoch": 0.365488824562595,
      "grad_norm": 0.10527502745389938,
      "learning_rate": 6.370855573651613e-05,
      "loss": 1.5818,
      "step": 185216
    },
    {
      "epoch": 0.3655519705239914,
      "grad_norm": 0.10623817145824432,
      "learning_rate": 6.370221551215144e-05,
      "loss": 1.5764,
      "step": 185248
    },
    {
      "epoch": 0.36561511648538786,
      "grad_norm": 0.10605397820472717,
      "learning_rate": 6.369587528778675e-05,
      "loss": 1.5891,
      "step": 185280
    },
    {
      "epoch": 0.3656782624467843,
      "grad_norm": 0.11802637577056885,
      "learning_rate": 6.368953506342206e-05,
      "loss": 1.5709,
      "step": 185312
    },
    {
      "epoch": 0.36574140840818076,
      "grad_norm": 0.11343847215175629,
      "learning_rate": 6.368319483905737e-05,
      "loss": 1.5856,
      "step": 185344
    },
    {
      "epoch": 0.3658045543695772,
      "grad_norm": 0.10883905738592148,
      "learning_rate": 6.367685461469268e-05,
      "loss": 1.589,
      "step": 185376
    },
    {
      "epoch": 0.36586770033097366,
      "grad_norm": 0.10828329622745514,
      "learning_rate": 6.3670514390328e-05,
      "loss": 1.5672,
      "step": 185408
    },
    {
      "epoch": 0.3659308462923701,
      "grad_norm": 0.11625433713197708,
      "learning_rate": 6.36641741659633e-05,
      "loss": 1.5741,
      "step": 185440
    },
    {
      "epoch": 0.3659939922537665,
      "grad_norm": 0.10105549544095993,
      "learning_rate": 6.365783394159862e-05,
      "loss": 1.5741,
      "step": 185472
    },
    {
      "epoch": 0.36605713821516295,
      "grad_norm": 0.10638256371021271,
      "learning_rate": 6.365149371723393e-05,
      "loss": 1.5937,
      "step": 185504
    },
    {
      "epoch": 0.3661202841765594,
      "grad_norm": 0.10467677563428879,
      "learning_rate": 6.364515349286923e-05,
      "loss": 1.584,
      "step": 185536
    },
    {
      "epoch": 0.36618343013795585,
      "grad_norm": 0.11417409032583237,
      "learning_rate": 6.363881326850453e-05,
      "loss": 1.595,
      "step": 185568
    },
    {
      "epoch": 0.3662465760993523,
      "grad_norm": 0.10955195873975754,
      "learning_rate": 6.363247304413985e-05,
      "loss": 1.5834,
      "step": 185600
    },
    {
      "epoch": 0.36630972206074874,
      "grad_norm": 0.11689722537994385,
      "learning_rate": 6.362613281977516e-05,
      "loss": 1.5864,
      "step": 185632
    },
    {
      "epoch": 0.36637286802214514,
      "grad_norm": 0.11240193992853165,
      "learning_rate": 6.361979259541048e-05,
      "loss": 1.5856,
      "step": 185664
    },
    {
      "epoch": 0.3664360139835416,
      "grad_norm": 0.10780593752861023,
      "learning_rate": 6.361345237104578e-05,
      "loss": 1.5942,
      "step": 185696
    },
    {
      "epoch": 0.36649915994493804,
      "grad_norm": 0.11237487941980362,
      "learning_rate": 6.360711214668109e-05,
      "loss": 1.5855,
      "step": 185728
    },
    {
      "epoch": 0.3665623059063345,
      "grad_norm": 0.11068807542324066,
      "learning_rate": 6.36007719223164e-05,
      "loss": 1.5895,
      "step": 185760
    },
    {
      "epoch": 0.36662545186773093,
      "grad_norm": 0.11342378705739975,
      "learning_rate": 6.359443169795172e-05,
      "loss": 1.5798,
      "step": 185792
    },
    {
      "epoch": 0.3666885978291274,
      "grad_norm": 0.10851754993200302,
      "learning_rate": 6.358809147358702e-05,
      "loss": 1.5823,
      "step": 185824
    },
    {
      "epoch": 0.3667517437905238,
      "grad_norm": 0.10945183783769608,
      "learning_rate": 6.358175124922234e-05,
      "loss": 1.5779,
      "step": 185856
    },
    {
      "epoch": 0.3668148897519202,
      "grad_norm": 0.11009562760591507,
      "learning_rate": 6.357541102485765e-05,
      "loss": 1.5762,
      "step": 185888
    },
    {
      "epoch": 0.3668780357133167,
      "grad_norm": 0.10616668313741684,
      "learning_rate": 6.356907080049296e-05,
      "loss": 1.5855,
      "step": 185920
    },
    {
      "epoch": 0.3669411816747131,
      "grad_norm": 0.11221147328615189,
      "learning_rate": 6.356273057612827e-05,
      "loss": 1.5763,
      "step": 185952
    },
    {
      "epoch": 0.3670043276361096,
      "grad_norm": 0.12302812188863754,
      "learning_rate": 6.355639035176357e-05,
      "loss": 1.5915,
      "step": 185984
    },
    {
      "epoch": 0.367067473597506,
      "grad_norm": 0.10927576571702957,
      "learning_rate": 6.355005012739888e-05,
      "loss": 1.5831,
      "step": 186016
    },
    {
      "epoch": 0.3671306195589024,
      "grad_norm": 0.11248224228620529,
      "learning_rate": 6.35437099030342e-05,
      "loss": 1.5675,
      "step": 186048
    },
    {
      "epoch": 0.36719376552029886,
      "grad_norm": 0.1288052648305893,
      "learning_rate": 6.353736967866951e-05,
      "loss": 1.5896,
      "step": 186080
    },
    {
      "epoch": 0.3672569114816953,
      "grad_norm": 0.1102360412478447,
      "learning_rate": 6.353102945430481e-05,
      "loss": 1.5953,
      "step": 186112
    },
    {
      "epoch": 0.36732005744309176,
      "grad_norm": 0.10656482726335526,
      "learning_rate": 6.352468922994013e-05,
      "loss": 1.5756,
      "step": 186144
    },
    {
      "epoch": 0.3673832034044882,
      "grad_norm": 0.108738012611866,
      "learning_rate": 6.351834900557544e-05,
      "loss": 1.5688,
      "step": 186176
    },
    {
      "epoch": 0.36744634936588466,
      "grad_norm": 0.1082652285695076,
      "learning_rate": 6.351200878121076e-05,
      "loss": 1.577,
      "step": 186208
    },
    {
      "epoch": 0.36750949532728105,
      "grad_norm": 0.11207770556211472,
      "learning_rate": 6.350566855684606e-05,
      "loss": 1.5827,
      "step": 186240
    },
    {
      "epoch": 0.3675726412886775,
      "grad_norm": 0.11149799823760986,
      "learning_rate": 6.349932833248137e-05,
      "loss": 1.5718,
      "step": 186272
    },
    {
      "epoch": 0.36763578725007395,
      "grad_norm": 0.12550869584083557,
      "learning_rate": 6.349298810811669e-05,
      "loss": 1.5937,
      "step": 186304
    },
    {
      "epoch": 0.3676989332114704,
      "grad_norm": 0.10630583763122559,
      "learning_rate": 6.348664788375199e-05,
      "loss": 1.5784,
      "step": 186336
    },
    {
      "epoch": 0.36776207917286685,
      "grad_norm": 0.11240078508853912,
      "learning_rate": 6.34803076593873e-05,
      "loss": 1.576,
      "step": 186368
    },
    {
      "epoch": 0.3678252251342633,
      "grad_norm": 0.11415791511535645,
      "learning_rate": 6.34739674350226e-05,
      "loss": 1.5774,
      "step": 186400
    },
    {
      "epoch": 0.3678883710956597,
      "grad_norm": 0.11179355531930923,
      "learning_rate": 6.346762721065792e-05,
      "loss": 1.5878,
      "step": 186432
    },
    {
      "epoch": 0.36795151705705614,
      "grad_norm": 0.10920584201812744,
      "learning_rate": 6.346128698629323e-05,
      "loss": 1.5845,
      "step": 186464
    },
    {
      "epoch": 0.3680146630184526,
      "grad_norm": 0.10972471535205841,
      "learning_rate": 6.345494676192853e-05,
      "loss": 1.583,
      "step": 186496
    },
    {
      "epoch": 0.36807780897984904,
      "grad_norm": 0.10917947441339493,
      "learning_rate": 6.344860653756385e-05,
      "loss": 1.5843,
      "step": 186528
    },
    {
      "epoch": 0.3681409549412455,
      "grad_norm": 0.10343483090400696,
      "learning_rate": 6.344226631319916e-05,
      "loss": 1.5913,
      "step": 186560
    },
    {
      "epoch": 0.36820410090264194,
      "grad_norm": 0.11454430967569351,
      "learning_rate": 6.343592608883448e-05,
      "loss": 1.5826,
      "step": 186592
    },
    {
      "epoch": 0.36826724686403833,
      "grad_norm": 0.1080147922039032,
      "learning_rate": 6.342958586446979e-05,
      "loss": 1.5873,
      "step": 186624
    },
    {
      "epoch": 0.3683303928254348,
      "grad_norm": 0.11471916735172272,
      "learning_rate": 6.342324564010509e-05,
      "loss": 1.5831,
      "step": 186656
    },
    {
      "epoch": 0.36839353878683123,
      "grad_norm": 0.10837042331695557,
      "learning_rate": 6.34169054157404e-05,
      "loss": 1.5866,
      "step": 186688
    },
    {
      "epoch": 0.3684566847482277,
      "grad_norm": 0.10801560431718826,
      "learning_rate": 6.341056519137572e-05,
      "loss": 1.6044,
      "step": 186720
    },
    {
      "epoch": 0.36851983070962413,
      "grad_norm": 0.11287467181682587,
      "learning_rate": 6.340422496701102e-05,
      "loss": 1.5897,
      "step": 186752
    },
    {
      "epoch": 0.3685829766710206,
      "grad_norm": 0.10864336788654327,
      "learning_rate": 6.339788474264634e-05,
      "loss": 1.5779,
      "step": 186784
    },
    {
      "epoch": 0.368646122632417,
      "grad_norm": 0.11014921218156815,
      "learning_rate": 6.339154451828164e-05,
      "loss": 1.5901,
      "step": 186816
    },
    {
      "epoch": 0.3687092685938134,
      "grad_norm": 0.11572349071502686,
      "learning_rate": 6.338520429391695e-05,
      "loss": 1.5711,
      "step": 186848
    },
    {
      "epoch": 0.36877241455520987,
      "grad_norm": 0.1091943010687828,
      "learning_rate": 6.337886406955227e-05,
      "loss": 1.5754,
      "step": 186880
    },
    {
      "epoch": 0.3688355605166063,
      "grad_norm": 0.11174189299345016,
      "learning_rate": 6.337252384518757e-05,
      "loss": 1.5924,
      "step": 186912
    },
    {
      "epoch": 0.36889870647800277,
      "grad_norm": 0.11803366988897324,
      "learning_rate": 6.336618362082288e-05,
      "loss": 1.5802,
      "step": 186944
    },
    {
      "epoch": 0.3689618524393992,
      "grad_norm": 0.112164206802845,
      "learning_rate": 6.33598433964582e-05,
      "loss": 1.5938,
      "step": 186976
    },
    {
      "epoch": 0.3690249984007956,
      "grad_norm": 0.1092149019241333,
      "learning_rate": 6.335350317209351e-05,
      "loss": 1.5906,
      "step": 187008
    },
    {
      "epoch": 0.36908814436219206,
      "grad_norm": 0.10716573894023895,
      "learning_rate": 6.334716294772881e-05,
      "loss": 1.5869,
      "step": 187040
    },
    {
      "epoch": 0.3691512903235885,
      "grad_norm": 0.12248466908931732,
      "learning_rate": 6.334082272336413e-05,
      "loss": 1.5761,
      "step": 187072
    },
    {
      "epoch": 0.36921443628498496,
      "grad_norm": 0.10569100081920624,
      "learning_rate": 6.333448249899944e-05,
      "loss": 1.5766,
      "step": 187104
    },
    {
      "epoch": 0.3692775822463814,
      "grad_norm": 0.11047903448343277,
      "learning_rate": 6.332814227463476e-05,
      "loss": 1.577,
      "step": 187136
    },
    {
      "epoch": 0.36934072820777786,
      "grad_norm": 0.11852037161588669,
      "learning_rate": 6.332180205027006e-05,
      "loss": 1.5805,
      "step": 187168
    },
    {
      "epoch": 0.3694038741691743,
      "grad_norm": 0.11280177533626556,
      "learning_rate": 6.331546182590537e-05,
      "loss": 1.5888,
      "step": 187200
    },
    {
      "epoch": 0.3694670201305707,
      "grad_norm": 0.11293172836303711,
      "learning_rate": 6.330912160154067e-05,
      "loss": 1.5788,
      "step": 187232
    },
    {
      "epoch": 0.36953016609196715,
      "grad_norm": 0.10566996783018112,
      "learning_rate": 6.330278137717599e-05,
      "loss": 1.5906,
      "step": 187264
    },
    {
      "epoch": 0.3695933120533636,
      "grad_norm": 0.1050875261425972,
      "learning_rate": 6.32964411528113e-05,
      "loss": 1.5784,
      "step": 187296
    },
    {
      "epoch": 0.36965645801476005,
      "grad_norm": 0.11344988644123077,
      "learning_rate": 6.32901009284466e-05,
      "loss": 1.5813,
      "step": 187328
    },
    {
      "epoch": 0.3697196039761565,
      "grad_norm": 0.10801728069782257,
      "learning_rate": 6.328376070408192e-05,
      "loss": 1.5862,
      "step": 187360
    },
    {
      "epoch": 0.36978274993755295,
      "grad_norm": 0.11555488407611847,
      "learning_rate": 6.327742047971723e-05,
      "loss": 1.5682,
      "step": 187392
    },
    {
      "epoch": 0.36984589589894934,
      "grad_norm": 0.1096230298280716,
      "learning_rate": 6.327108025535255e-05,
      "loss": 1.5892,
      "step": 187424
    },
    {
      "epoch": 0.3699090418603458,
      "grad_norm": 0.10769311338663101,
      "learning_rate": 6.326474003098785e-05,
      "loss": 1.5902,
      "step": 187456
    },
    {
      "epoch": 0.36997218782174224,
      "grad_norm": 0.10758940875530243,
      "learning_rate": 6.325839980662316e-05,
      "loss": 1.5808,
      "step": 187488
    },
    {
      "epoch": 0.3700353337831387,
      "grad_norm": 0.11722041666507721,
      "learning_rate": 6.325205958225848e-05,
      "loss": 1.5754,
      "step": 187520
    },
    {
      "epoch": 0.37009847974453514,
      "grad_norm": 0.10478208214044571,
      "learning_rate": 6.324571935789379e-05,
      "loss": 1.5898,
      "step": 187552
    },
    {
      "epoch": 0.3701616257059316,
      "grad_norm": 0.11724715679883957,
      "learning_rate": 6.323937913352909e-05,
      "loss": 1.5778,
      "step": 187584
    },
    {
      "epoch": 0.370224771667328,
      "grad_norm": 0.11673936247825623,
      "learning_rate": 6.323303890916439e-05,
      "loss": 1.5776,
      "step": 187616
    },
    {
      "epoch": 0.37028791762872443,
      "grad_norm": 0.11095519363880157,
      "learning_rate": 6.322669868479971e-05,
      "loss": 1.5778,
      "step": 187648
    },
    {
      "epoch": 0.3703510635901209,
      "grad_norm": 0.1110343262553215,
      "learning_rate": 6.322035846043502e-05,
      "loss": 1.5813,
      "step": 187680
    },
    {
      "epoch": 0.3704142095515173,
      "grad_norm": 0.11634550988674164,
      "learning_rate": 6.321401823607032e-05,
      "loss": 1.5733,
      "step": 187712
    },
    {
      "epoch": 0.3704773555129138,
      "grad_norm": 0.10587144643068314,
      "learning_rate": 6.320767801170564e-05,
      "loss": 1.5924,
      "step": 187744
    },
    {
      "epoch": 0.3705405014743102,
      "grad_norm": 0.11117944121360779,
      "learning_rate": 6.320133778734095e-05,
      "loss": 1.579,
      "step": 187776
    },
    {
      "epoch": 0.3706036474357066,
      "grad_norm": 0.11027168482542038,
      "learning_rate": 6.319499756297627e-05,
      "loss": 1.5792,
      "step": 187808
    },
    {
      "epoch": 0.37066679339710307,
      "grad_norm": 0.11100541800260544,
      "learning_rate": 6.318865733861157e-05,
      "loss": 1.5791,
      "step": 187840
    },
    {
      "epoch": 0.3707299393584995,
      "grad_norm": 0.11651395261287689,
      "learning_rate": 6.318231711424688e-05,
      "loss": 1.5806,
      "step": 187872
    },
    {
      "epoch": 0.37079308531989597,
      "grad_norm": 0.1091759130358696,
      "learning_rate": 6.31759768898822e-05,
      "loss": 1.5852,
      "step": 187904
    },
    {
      "epoch": 0.3708562312812924,
      "grad_norm": 0.11268169432878494,
      "learning_rate": 6.316963666551751e-05,
      "loss": 1.5655,
      "step": 187936
    },
    {
      "epoch": 0.37091937724268886,
      "grad_norm": 0.10827192664146423,
      "learning_rate": 6.316329644115283e-05,
      "loss": 1.5883,
      "step": 187968
    },
    {
      "epoch": 0.37098252320408526,
      "grad_norm": 0.1223548874258995,
      "learning_rate": 6.315695621678813e-05,
      "loss": 1.5881,
      "step": 188000
    },
    {
      "epoch": 0.3710456691654817,
      "grad_norm": 0.11432972550392151,
      "learning_rate": 6.315061599242343e-05,
      "loss": 1.5763,
      "step": 188032
    },
    {
      "epoch": 0.37110881512687816,
      "grad_norm": 0.11161147058010101,
      "learning_rate": 6.314427576805874e-05,
      "loss": 1.5721,
      "step": 188064
    },
    {
      "epoch": 0.3711719610882746,
      "grad_norm": 0.10936208069324493,
      "learning_rate": 6.313793554369406e-05,
      "loss": 1.5879,
      "step": 188096
    },
    {
      "epoch": 0.37123510704967105,
      "grad_norm": 0.10863802582025528,
      "learning_rate": 6.313159531932936e-05,
      "loss": 1.5788,
      "step": 188128
    },
    {
      "epoch": 0.3712982530110675,
      "grad_norm": 0.1166510283946991,
      "learning_rate": 6.312525509496467e-05,
      "loss": 1.5797,
      "step": 188160
    },
    {
      "epoch": 0.3713613989724639,
      "grad_norm": 0.11168581247329712,
      "learning_rate": 6.311891487059999e-05,
      "loss": 1.5825,
      "step": 188192
    },
    {
      "epoch": 0.37142454493386035,
      "grad_norm": 0.10761485248804092,
      "learning_rate": 6.31125746462353e-05,
      "loss": 1.5883,
      "step": 188224
    },
    {
      "epoch": 0.3714876908952568,
      "grad_norm": 0.12380202114582062,
      "learning_rate": 6.31062344218706e-05,
      "loss": 1.5718,
      "step": 188256
    },
    {
      "epoch": 0.37155083685665324,
      "grad_norm": 0.10772494971752167,
      "learning_rate": 6.309989419750592e-05,
      "loss": 1.5724,
      "step": 188288
    },
    {
      "epoch": 0.3716139828180497,
      "grad_norm": 0.11718533933162689,
      "learning_rate": 6.309355397314123e-05,
      "loss": 1.5922,
      "step": 188320
    },
    {
      "epoch": 0.37167712877944614,
      "grad_norm": 0.10808006674051285,
      "learning_rate": 6.308721374877655e-05,
      "loss": 1.5777,
      "step": 188352
    },
    {
      "epoch": 0.37174027474084254,
      "grad_norm": 0.11395537108182907,
      "learning_rate": 6.308087352441185e-05,
      "loss": 1.5815,
      "step": 188384
    },
    {
      "epoch": 0.371803420702239,
      "grad_norm": 0.11253679543733597,
      "learning_rate": 6.307453330004716e-05,
      "loss": 1.5827,
      "step": 188416
    },
    {
      "epoch": 0.37186656666363543,
      "grad_norm": 0.1086331456899643,
      "learning_rate": 6.306819307568246e-05,
      "loss": 1.5627,
      "step": 188448
    },
    {
      "epoch": 0.3719297126250319,
      "grad_norm": 0.1119273379445076,
      "learning_rate": 6.306185285131778e-05,
      "loss": 1.5649,
      "step": 188480
    },
    {
      "epoch": 0.37199285858642833,
      "grad_norm": 0.11097158491611481,
      "learning_rate": 6.305551262695308e-05,
      "loss": 1.578,
      "step": 188512
    },
    {
      "epoch": 0.3720560045478248,
      "grad_norm": 0.11210909485816956,
      "learning_rate": 6.304917240258839e-05,
      "loss": 1.5931,
      "step": 188544
    },
    {
      "epoch": 0.3721191505092212,
      "grad_norm": 0.10717438906431198,
      "learning_rate": 6.304283217822371e-05,
      "loss": 1.5813,
      "step": 188576
    },
    {
      "epoch": 0.3721822964706176,
      "grad_norm": 0.10911908000707626,
      "learning_rate": 6.303649195385902e-05,
      "loss": 1.5902,
      "step": 188608
    },
    {
      "epoch": 0.3722454424320141,
      "grad_norm": 0.11350075900554657,
      "learning_rate": 6.303015172949434e-05,
      "loss": 1.5821,
      "step": 188640
    },
    {
      "epoch": 0.3723085883934105,
      "grad_norm": 0.10877273231744766,
      "learning_rate": 6.302381150512964e-05,
      "loss": 1.5749,
      "step": 188672
    },
    {
      "epoch": 0.37237173435480697,
      "grad_norm": 0.10497533529996872,
      "learning_rate": 6.301747128076495e-05,
      "loss": 1.5787,
      "step": 188704
    },
    {
      "epoch": 0.3724348803162034,
      "grad_norm": 0.10719005018472672,
      "learning_rate": 6.301113105640027e-05,
      "loss": 1.5939,
      "step": 188736
    },
    {
      "epoch": 0.3724980262775998,
      "grad_norm": 0.11375787109136581,
      "learning_rate": 6.300479083203558e-05,
      "loss": 1.5854,
      "step": 188768
    },
    {
      "epoch": 0.37256117223899626,
      "grad_norm": 0.11353506147861481,
      "learning_rate": 6.299845060767088e-05,
      "loss": 1.5841,
      "step": 188800
    },
    {
      "epoch": 0.3726243182003927,
      "grad_norm": 0.11330796778202057,
      "learning_rate": 6.29921103833062e-05,
      "loss": 1.5835,
      "step": 188832
    },
    {
      "epoch": 0.37268746416178916,
      "grad_norm": 0.10765306651592255,
      "learning_rate": 6.29857701589415e-05,
      "loss": 1.5716,
      "step": 188864
    },
    {
      "epoch": 0.3727506101231856,
      "grad_norm": 0.115536168217659,
      "learning_rate": 6.297942993457681e-05,
      "loss": 1.5767,
      "step": 188896
    },
    {
      "epoch": 0.37281375608458206,
      "grad_norm": 0.11861121654510498,
      "learning_rate": 6.297308971021211e-05,
      "loss": 1.5825,
      "step": 188928
    },
    {
      "epoch": 0.3728769020459785,
      "grad_norm": 0.10668471455574036,
      "learning_rate": 6.296674948584743e-05,
      "loss": 1.5745,
      "step": 188960
    },
    {
      "epoch": 0.3729400480073749,
      "grad_norm": 0.1124548688530922,
      "learning_rate": 6.296040926148274e-05,
      "loss": 1.5855,
      "step": 188992
    },
    {
      "epoch": 0.37300319396877135,
      "grad_norm": 0.11227509379386902,
      "learning_rate": 6.295406903711806e-05,
      "loss": 1.5819,
      "step": 189024
    },
    {
      "epoch": 0.3730663399301678,
      "grad_norm": 0.12284498661756516,
      "learning_rate": 6.294772881275336e-05,
      "loss": 1.5829,
      "step": 189056
    },
    {
      "epoch": 0.37312948589156425,
      "grad_norm": 0.11107189208269119,
      "learning_rate": 6.294138858838867e-05,
      "loss": 1.5859,
      "step": 189088
    },
    {
      "epoch": 0.3731926318529607,
      "grad_norm": 0.10985364764928818,
      "learning_rate": 6.293504836402399e-05,
      "loss": 1.5772,
      "step": 189120
    },
    {
      "epoch": 0.37325577781435715,
      "grad_norm": 0.10888254642486572,
      "learning_rate": 6.29287081396593e-05,
      "loss": 1.5831,
      "step": 189152
    },
    {
      "epoch": 0.37331892377575354,
      "grad_norm": 0.1114615648984909,
      "learning_rate": 6.29223679152946e-05,
      "loss": 1.5767,
      "step": 189184
    },
    {
      "epoch": 0.37338206973715,
      "grad_norm": 0.11195087432861328,
      "learning_rate": 6.291602769092992e-05,
      "loss": 1.5908,
      "step": 189216
    },
    {
      "epoch": 0.37344521569854644,
      "grad_norm": 0.11618414521217346,
      "learning_rate": 6.290968746656523e-05,
      "loss": 1.5838,
      "step": 189248
    },
    {
      "epoch": 0.3735083616599429,
      "grad_norm": 0.11566731333732605,
      "learning_rate": 6.290334724220053e-05,
      "loss": 1.5942,
      "step": 189280
    },
    {
      "epoch": 0.37357150762133934,
      "grad_norm": 0.11618366837501526,
      "learning_rate": 6.289700701783585e-05,
      "loss": 1.584,
      "step": 189312
    },
    {
      "epoch": 0.3736346535827358,
      "grad_norm": 0.11000082641839981,
      "learning_rate": 6.289066679347115e-05,
      "loss": 1.5809,
      "step": 189344
    },
    {
      "epoch": 0.3736977995441322,
      "grad_norm": 0.11952618509531021,
      "learning_rate": 6.288432656910646e-05,
      "loss": 1.578,
      "step": 189376
    },
    {
      "epoch": 0.37376094550552863,
      "grad_norm": 0.10670780390501022,
      "learning_rate": 6.287798634474178e-05,
      "loss": 1.5737,
      "step": 189408
    },
    {
      "epoch": 0.3738240914669251,
      "grad_norm": 0.11222748458385468,
      "learning_rate": 6.287164612037709e-05,
      "loss": 1.591,
      "step": 189440
    },
    {
      "epoch": 0.37388723742832153,
      "grad_norm": 0.11076555401086807,
      "learning_rate": 6.28653058960124e-05,
      "loss": 1.5819,
      "step": 189472
    },
    {
      "epoch": 0.373950383389718,
      "grad_norm": 0.11821553856134415,
      "learning_rate": 6.285896567164771e-05,
      "loss": 1.5873,
      "step": 189504
    },
    {
      "epoch": 0.3740135293511144,
      "grad_norm": 0.12025181949138641,
      "learning_rate": 6.285262544728302e-05,
      "loss": 1.5672,
      "step": 189536
    },
    {
      "epoch": 0.3740766753125108,
      "grad_norm": 0.1173589751124382,
      "learning_rate": 6.284628522291834e-05,
      "loss": 1.5903,
      "step": 189568
    },
    {
      "epoch": 0.37413982127390727,
      "grad_norm": 0.11068599671125412,
      "learning_rate": 6.283994499855364e-05,
      "loss": 1.5772,
      "step": 189600
    },
    {
      "epoch": 0.3742029672353037,
      "grad_norm": 0.10940337926149368,
      "learning_rate": 6.283360477418895e-05,
      "loss": 1.5761,
      "step": 189632
    },
    {
      "epoch": 0.37426611319670017,
      "grad_norm": 0.1058943122625351,
      "learning_rate": 6.282726454982427e-05,
      "loss": 1.5879,
      "step": 189664
    },
    {
      "epoch": 0.3743292591580966,
      "grad_norm": 0.11430080980062485,
      "learning_rate": 6.282092432545957e-05,
      "loss": 1.5982,
      "step": 189696
    },
    {
      "epoch": 0.37439240511949307,
      "grad_norm": 0.11159449815750122,
      "learning_rate": 6.281458410109487e-05,
      "loss": 1.5769,
      "step": 189728
    },
    {
      "epoch": 0.37445555108088946,
      "grad_norm": 0.11083097755908966,
      "learning_rate": 6.280824387673018e-05,
      "loss": 1.5755,
      "step": 189760
    },
    {
      "epoch": 0.3745186970422859,
      "grad_norm": 0.11005783826112747,
      "learning_rate": 6.28019036523655e-05,
      "loss": 1.5858,
      "step": 189792
    },
    {
      "epoch": 0.37458184300368236,
      "grad_norm": 0.106862373650074,
      "learning_rate": 6.279556342800081e-05,
      "loss": 1.5724,
      "step": 189824
    },
    {
      "epoch": 0.3746449889650788,
      "grad_norm": 0.11401025205850601,
      "learning_rate": 6.278922320363613e-05,
      "loss": 1.5833,
      "step": 189856
    },
    {
      "epoch": 0.37470813492647526,
      "grad_norm": 0.11320473253726959,
      "learning_rate": 6.278288297927143e-05,
      "loss": 1.5728,
      "step": 189888
    },
    {
      "epoch": 0.3747712808878717,
      "grad_norm": 0.11194812506437302,
      "learning_rate": 6.277654275490674e-05,
      "loss": 1.6002,
      "step": 189920
    },
    {
      "epoch": 0.3748344268492681,
      "grad_norm": 0.10899345576763153,
      "learning_rate": 6.277020253054206e-05,
      "loss": 1.5887,
      "step": 189952
    },
    {
      "epoch": 0.37489757281066455,
      "grad_norm": 0.10808474570512772,
      "learning_rate": 6.276386230617737e-05,
      "loss": 1.5686,
      "step": 189984
    },
    {
      "epoch": 0.374960718772061,
      "grad_norm": 0.11248473078012466,
      "learning_rate": 6.275752208181267e-05,
      "loss": 1.5782,
      "step": 190016
    },
    {
      "epoch": 0.37502386473345745,
      "grad_norm": 0.1137518584728241,
      "learning_rate": 6.275118185744799e-05,
      "loss": 1.586,
      "step": 190048
    },
    {
      "epoch": 0.3750870106948539,
      "grad_norm": 0.1211237832903862,
      "learning_rate": 6.27448416330833e-05,
      "loss": 1.5788,
      "step": 190080
    },
    {
      "epoch": 0.37515015665625034,
      "grad_norm": 0.11008920520544052,
      "learning_rate": 6.27385014087186e-05,
      "loss": 1.5796,
      "step": 190112
    },
    {
      "epoch": 0.37521330261764674,
      "grad_norm": 0.10260678082704544,
      "learning_rate": 6.27321611843539e-05,
      "loss": 1.5874,
      "step": 190144
    },
    {
      "epoch": 0.3752764485790432,
      "grad_norm": 0.11161653697490692,
      "learning_rate": 6.272582095998922e-05,
      "loss": 1.577,
      "step": 190176
    },
    {
      "epoch": 0.37533959454043964,
      "grad_norm": 0.12450781464576721,
      "learning_rate": 6.271948073562453e-05,
      "loss": 1.5882,
      "step": 190208
    },
    {
      "epoch": 0.3754027405018361,
      "grad_norm": 0.11262410134077072,
      "learning_rate": 6.271314051125985e-05,
      "loss": 1.5839,
      "step": 190240
    },
    {
      "epoch": 0.37546588646323253,
      "grad_norm": 0.11589303612709045,
      "learning_rate": 6.270680028689515e-05,
      "loss": 1.5746,
      "step": 190272
    },
    {
      "epoch": 0.375529032424629,
      "grad_norm": 0.11190151423215866,
      "learning_rate": 6.270046006253046e-05,
      "loss": 1.5815,
      "step": 190304
    },
    {
      "epoch": 0.3755921783860254,
      "grad_norm": 0.1065768450498581,
      "learning_rate": 6.269411983816578e-05,
      "loss": 1.5759,
      "step": 190336
    },
    {
      "epoch": 0.3756553243474218,
      "grad_norm": 0.11990329623222351,
      "learning_rate": 6.268777961380109e-05,
      "loss": 1.5816,
      "step": 190368
    },
    {
      "epoch": 0.3757184703088183,
      "grad_norm": 0.10311850905418396,
      "learning_rate": 6.26814393894364e-05,
      "loss": 1.5719,
      "step": 190400
    },
    {
      "epoch": 0.3757816162702147,
      "grad_norm": 0.10978825390338898,
      "learning_rate": 6.267509916507171e-05,
      "loss": 1.5798,
      "step": 190432
    },
    {
      "epoch": 0.3758447622316112,
      "grad_norm": 0.11228978633880615,
      "learning_rate": 6.266875894070702e-05,
      "loss": 1.5829,
      "step": 190464
    },
    {
      "epoch": 0.3759079081930076,
      "grad_norm": 0.10573288053274155,
      "learning_rate": 6.266241871634232e-05,
      "loss": 1.592,
      "step": 190496
    },
    {
      "epoch": 0.37597105415440407,
      "grad_norm": 0.11256791651248932,
      "learning_rate": 6.265607849197764e-05,
      "loss": 1.5982,
      "step": 190528
    },
    {
      "epoch": 0.37603420011580047,
      "grad_norm": 0.10375548154115677,
      "learning_rate": 6.264973826761294e-05,
      "loss": 1.5888,
      "step": 190560
    },
    {
      "epoch": 0.3760973460771969,
      "grad_norm": 0.10288132727146149,
      "learning_rate": 6.264339804324825e-05,
      "loss": 1.5796,
      "step": 190592
    },
    {
      "epoch": 0.37616049203859336,
      "grad_norm": 0.1160019040107727,
      "learning_rate": 6.263705781888357e-05,
      "loss": 1.5684,
      "step": 190624
    },
    {
      "epoch": 0.3762236379999898,
      "grad_norm": 0.11431177705526352,
      "learning_rate": 6.263071759451888e-05,
      "loss": 1.5761,
      "step": 190656
    },
    {
      "epoch": 0.37628678396138626,
      "grad_norm": 0.11775610595941544,
      "learning_rate": 6.262437737015418e-05,
      "loss": 1.5926,
      "step": 190688
    },
    {
      "epoch": 0.3763499299227827,
      "grad_norm": 0.11620460450649261,
      "learning_rate": 6.26180371457895e-05,
      "loss": 1.5806,
      "step": 190720
    },
    {
      "epoch": 0.3764130758841791,
      "grad_norm": 0.11400526762008667,
      "learning_rate": 6.261169692142481e-05,
      "loss": 1.5613,
      "step": 190752
    },
    {
      "epoch": 0.37647622184557555,
      "grad_norm": 0.11397536098957062,
      "learning_rate": 6.260535669706013e-05,
      "loss": 1.5769,
      "step": 190784
    },
    {
      "epoch": 0.376539367806972,
      "grad_norm": 0.11190059036016464,
      "learning_rate": 6.259901647269543e-05,
      "loss": 1.5664,
      "step": 190816
    },
    {
      "epoch": 0.37660251376836845,
      "grad_norm": 0.11497852951288223,
      "learning_rate": 6.259267624833074e-05,
      "loss": 1.5884,
      "step": 190848
    },
    {
      "epoch": 0.3766656597297649,
      "grad_norm": 0.11309961974620819,
      "learning_rate": 6.258633602396606e-05,
      "loss": 1.5801,
      "step": 190880
    },
    {
      "epoch": 0.37672880569116135,
      "grad_norm": 0.1095379963517189,
      "learning_rate": 6.257999579960136e-05,
      "loss": 1.5813,
      "step": 190912
    },
    {
      "epoch": 0.37679195165255774,
      "grad_norm": 0.11139626801013947,
      "learning_rate": 6.257365557523667e-05,
      "loss": 1.5693,
      "step": 190944
    },
    {
      "epoch": 0.3768550976139542,
      "grad_norm": 0.11057497560977936,
      "learning_rate": 6.256731535087197e-05,
      "loss": 1.5683,
      "step": 190976
    },
    {
      "epoch": 0.37691824357535064,
      "grad_norm": 0.110712930560112,
      "learning_rate": 6.256097512650729e-05,
      "loss": 1.5796,
      "step": 191008
    },
    {
      "epoch": 0.3769813895367471,
      "grad_norm": 0.11229637265205383,
      "learning_rate": 6.25546349021426e-05,
      "loss": 1.5879,
      "step": 191040
    },
    {
      "epoch": 0.37704453549814354,
      "grad_norm": 0.11921262741088867,
      "learning_rate": 6.25482946777779e-05,
      "loss": 1.5832,
      "step": 191072
    },
    {
      "epoch": 0.37710768145954,
      "grad_norm": 0.11263344436883926,
      "learning_rate": 6.254195445341322e-05,
      "loss": 1.5868,
      "step": 191104
    },
    {
      "epoch": 0.3771708274209364,
      "grad_norm": 0.11202883720397949,
      "learning_rate": 6.253561422904853e-05,
      "loss": 1.5725,
      "step": 191136
    },
    {
      "epoch": 0.37723397338233283,
      "grad_norm": 0.11726159602403641,
      "learning_rate": 6.252927400468385e-05,
      "loss": 1.5763,
      "step": 191168
    },
    {
      "epoch": 0.3772971193437293,
      "grad_norm": 0.10986708104610443,
      "learning_rate": 6.252293378031916e-05,
      "loss": 1.5753,
      "step": 191200
    },
    {
      "epoch": 0.37736026530512573,
      "grad_norm": 0.11207885295152664,
      "learning_rate": 6.251659355595446e-05,
      "loss": 1.582,
      "step": 191232
    },
    {
      "epoch": 0.3774234112665222,
      "grad_norm": 0.1060028225183487,
      "learning_rate": 6.251025333158978e-05,
      "loss": 1.5728,
      "step": 191264
    },
    {
      "epoch": 0.37748655722791863,
      "grad_norm": 0.10829560458660126,
      "learning_rate": 6.250391310722509e-05,
      "loss": 1.5865,
      "step": 191296
    },
    {
      "epoch": 0.377549703189315,
      "grad_norm": 0.11095945537090302,
      "learning_rate": 6.24975728828604e-05,
      "loss": 1.5856,
      "step": 191328
    },
    {
      "epoch": 0.37761284915071147,
      "grad_norm": 0.10997764021158218,
      "learning_rate": 6.249123265849571e-05,
      "loss": 1.5802,
      "step": 191360
    },
    {
      "epoch": 0.3776759951121079,
      "grad_norm": 0.1037118136882782,
      "learning_rate": 6.248489243413101e-05,
      "loss": 1.5782,
      "step": 191392
    },
    {
      "epoch": 0.37773914107350437,
      "grad_norm": 0.10766377300024033,
      "learning_rate": 6.247855220976632e-05,
      "loss": 1.5982,
      "step": 191424
    },
    {
      "epoch": 0.3778022870349008,
      "grad_norm": 0.10773643106222153,
      "learning_rate": 6.247221198540164e-05,
      "loss": 1.5771,
      "step": 191456
    },
    {
      "epoch": 0.37786543299629727,
      "grad_norm": 0.10769221931695938,
      "learning_rate": 6.246587176103694e-05,
      "loss": 1.5762,
      "step": 191488
    },
    {
      "epoch": 0.37792857895769366,
      "grad_norm": 0.11420951038599014,
      "learning_rate": 6.245953153667225e-05,
      "loss": 1.5734,
      "step": 191520
    },
    {
      "epoch": 0.3779917249190901,
      "grad_norm": 0.11754541844129562,
      "learning_rate": 6.245319131230757e-05,
      "loss": 1.5967,
      "step": 191552
    },
    {
      "epoch": 0.37805487088048656,
      "grad_norm": 0.11201565712690353,
      "learning_rate": 6.244685108794288e-05,
      "loss": 1.5828,
      "step": 191584
    },
    {
      "epoch": 0.378118016841883,
      "grad_norm": 0.10708951205015182,
      "learning_rate": 6.244051086357818e-05,
      "loss": 1.5724,
      "step": 191616
    },
    {
      "epoch": 0.37818116280327946,
      "grad_norm": 0.11266741156578064,
      "learning_rate": 6.24341706392135e-05,
      "loss": 1.5759,
      "step": 191648
    },
    {
      "epoch": 0.3782443087646759,
      "grad_norm": 0.11262353509664536,
      "learning_rate": 6.242783041484881e-05,
      "loss": 1.5853,
      "step": 191680
    },
    {
      "epoch": 0.3783074547260723,
      "grad_norm": 0.11155575513839722,
      "learning_rate": 6.242149019048413e-05,
      "loss": 1.5882,
      "step": 191712
    },
    {
      "epoch": 0.37837060068746875,
      "grad_norm": 0.11081994324922562,
      "learning_rate": 6.241514996611943e-05,
      "loss": 1.5734,
      "step": 191744
    },
    {
      "epoch": 0.3784337466488652,
      "grad_norm": 0.11958064883947372,
      "learning_rate": 6.240880974175473e-05,
      "loss": 1.5889,
      "step": 191776
    },
    {
      "epoch": 0.37849689261026165,
      "grad_norm": 0.1084212213754654,
      "learning_rate": 6.240246951739004e-05,
      "loss": 1.5814,
      "step": 191808
    },
    {
      "epoch": 0.3785600385716581,
      "grad_norm": 0.10623104870319366,
      "learning_rate": 6.239612929302536e-05,
      "loss": 1.5779,
      "step": 191840
    },
    {
      "epoch": 0.37862318453305455,
      "grad_norm": 0.10781396925449371,
      "learning_rate": 6.238978906866067e-05,
      "loss": 1.5822,
      "step": 191872
    },
    {
      "epoch": 0.37868633049445094,
      "grad_norm": 0.1161232590675354,
      "learning_rate": 6.238344884429597e-05,
      "loss": 1.591,
      "step": 191904
    },
    {
      "epoch": 0.3787494764558474,
      "grad_norm": 0.11555363237857819,
      "learning_rate": 6.237710861993129e-05,
      "loss": 1.5881,
      "step": 191936
    },
    {
      "epoch": 0.37881262241724384,
      "grad_norm": 0.10289894044399261,
      "learning_rate": 6.23707683955666e-05,
      "loss": 1.5781,
      "step": 191968
    },
    {
      "epoch": 0.3788757683786403,
      "grad_norm": 0.10856793075799942,
      "learning_rate": 6.236442817120192e-05,
      "loss": 1.5751,
      "step": 192000
    },
    {
      "epoch": 0.37893891434003674,
      "grad_norm": 0.11440318822860718,
      "learning_rate": 6.235808794683722e-05,
      "loss": 1.5916,
      "step": 192032
    },
    {
      "epoch": 0.3790020603014332,
      "grad_norm": 0.11252167820930481,
      "learning_rate": 6.235174772247253e-05,
      "loss": 1.5816,
      "step": 192064
    },
    {
      "epoch": 0.3790652062628296,
      "grad_norm": 0.11253858357667923,
      "learning_rate": 6.234540749810785e-05,
      "loss": 1.5755,
      "step": 192096
    },
    {
      "epoch": 0.37912835222422603,
      "grad_norm": 0.12229254841804504,
      "learning_rate": 6.233906727374316e-05,
      "loss": 1.5863,
      "step": 192128
    },
    {
      "epoch": 0.3791914981856225,
      "grad_norm": 0.10934390872716904,
      "learning_rate": 6.233272704937846e-05,
      "loss": 1.5826,
      "step": 192160
    },
    {
      "epoch": 0.3792546441470189,
      "grad_norm": 0.10436520725488663,
      "learning_rate": 6.232638682501377e-05,
      "loss": 1.5719,
      "step": 192192
    },
    {
      "epoch": 0.3793177901084154,
      "grad_norm": 0.1138714849948883,
      "learning_rate": 6.232004660064908e-05,
      "loss": 1.5733,
      "step": 192224
    },
    {
      "epoch": 0.3793809360698118,
      "grad_norm": 0.10553093254566193,
      "learning_rate": 6.23137063762844e-05,
      "loss": 1.5742,
      "step": 192256
    },
    {
      "epoch": 0.3794440820312083,
      "grad_norm": 0.10531999915838242,
      "learning_rate": 6.23073661519197e-05,
      "loss": 1.5729,
      "step": 192288
    },
    {
      "epoch": 0.37950722799260467,
      "grad_norm": 0.11378216743469238,
      "learning_rate": 6.230102592755501e-05,
      "loss": 1.5691,
      "step": 192320
    },
    {
      "epoch": 0.3795703739540011,
      "grad_norm": 0.11386200040578842,
      "learning_rate": 6.229468570319032e-05,
      "loss": 1.5916,
      "step": 192352
    },
    {
      "epoch": 0.37963351991539757,
      "grad_norm": 0.11795624345541,
      "learning_rate": 6.228834547882564e-05,
      "loss": 1.5784,
      "step": 192384
    },
    {
      "epoch": 0.379696665876794,
      "grad_norm": 0.12270095199346542,
      "learning_rate": 6.228200525446094e-05,
      "loss": 1.5816,
      "step": 192416
    },
    {
      "epoch": 0.37975981183819046,
      "grad_norm": 0.11291665583848953,
      "learning_rate": 6.227566503009625e-05,
      "loss": 1.5791,
      "step": 192448
    },
    {
      "epoch": 0.3798229577995869,
      "grad_norm": 0.11110670864582062,
      "learning_rate": 6.226932480573157e-05,
      "loss": 1.5834,
      "step": 192480
    },
    {
      "epoch": 0.3798861037609833,
      "grad_norm": 0.11037114262580872,
      "learning_rate": 6.226298458136688e-05,
      "loss": 1.583,
      "step": 192512
    },
    {
      "epoch": 0.37994924972237976,
      "grad_norm": 0.10810991376638412,
      "learning_rate": 6.22566443570022e-05,
      "loss": 1.578,
      "step": 192544
    },
    {
      "epoch": 0.3800123956837762,
      "grad_norm": 0.10720200836658478,
      "learning_rate": 6.22503041326375e-05,
      "loss": 1.5839,
      "step": 192576
    },
    {
      "epoch": 0.38007554164517265,
      "grad_norm": 0.11847103387117386,
      "learning_rate": 6.22439639082728e-05,
      "loss": 1.5784,
      "step": 192608
    },
    {
      "epoch": 0.3801386876065691,
      "grad_norm": 0.11450802534818649,
      "learning_rate": 6.223762368390811e-05,
      "loss": 1.5781,
      "step": 192640
    },
    {
      "epoch": 0.38020183356796555,
      "grad_norm": 0.11172592639923096,
      "learning_rate": 6.223128345954343e-05,
      "loss": 1.5805,
      "step": 192672
    },
    {
      "epoch": 0.38026497952936195,
      "grad_norm": 0.11719383299350739,
      "learning_rate": 6.222494323517873e-05,
      "loss": 1.5755,
      "step": 192704
    },
    {
      "epoch": 0.3803281254907584,
      "grad_norm": 0.11797864735126495,
      "learning_rate": 6.221860301081405e-05,
      "loss": 1.5758,
      "step": 192736
    },
    {
      "epoch": 0.38039127145215484,
      "grad_norm": 0.11396016925573349,
      "learning_rate": 6.221226278644936e-05,
      "loss": 1.5732,
      "step": 192768
    },
    {
      "epoch": 0.3804544174135513,
      "grad_norm": 0.11699241399765015,
      "learning_rate": 6.220592256208467e-05,
      "loss": 1.5837,
      "step": 192800
    },
    {
      "epoch": 0.38051756337494774,
      "grad_norm": 0.11038757860660553,
      "learning_rate": 6.219958233771998e-05,
      "loss": 1.5879,
      "step": 192832
    },
    {
      "epoch": 0.3805807093363442,
      "grad_norm": 0.11245683580636978,
      "learning_rate": 6.219324211335529e-05,
      "loss": 1.5719,
      "step": 192864
    },
    {
      "epoch": 0.3806438552977406,
      "grad_norm": 0.11117320507764816,
      "learning_rate": 6.21869018889906e-05,
      "loss": 1.5674,
      "step": 192896
    },
    {
      "epoch": 0.38070700125913703,
      "grad_norm": 0.10423646867275238,
      "learning_rate": 6.218056166462592e-05,
      "loss": 1.5692,
      "step": 192928
    },
    {
      "epoch": 0.3807701472205335,
      "grad_norm": 0.11134912818670273,
      "learning_rate": 6.217422144026122e-05,
      "loss": 1.5853,
      "step": 192960
    },
    {
      "epoch": 0.38083329318192993,
      "grad_norm": 0.11681429296731949,
      "learning_rate": 6.216788121589653e-05,
      "loss": 1.5763,
      "step": 192992
    },
    {
      "epoch": 0.3808964391433264,
      "grad_norm": 0.11456853896379471,
      "learning_rate": 6.216154099153184e-05,
      "loss": 1.5776,
      "step": 193024
    },
    {
      "epoch": 0.38095958510472283,
      "grad_norm": 0.11466328799724579,
      "learning_rate": 6.215520076716715e-05,
      "loss": 1.5796,
      "step": 193056
    },
    {
      "epoch": 0.3810227310661192,
      "grad_norm": 0.10899902880191803,
      "learning_rate": 6.214886054280245e-05,
      "loss": 1.5761,
      "step": 193088
    },
    {
      "epoch": 0.3810858770275157,
      "grad_norm": 0.11018490046262741,
      "learning_rate": 6.214252031843777e-05,
      "loss": 1.5764,
      "step": 193120
    },
    {
      "epoch": 0.3811490229889121,
      "grad_norm": 0.12358193099498749,
      "learning_rate": 6.213618009407308e-05,
      "loss": 1.5704,
      "step": 193152
    },
    {
      "epoch": 0.38121216895030857,
      "grad_norm": 0.12298901379108429,
      "learning_rate": 6.21298398697084e-05,
      "loss": 1.5744,
      "step": 193184
    },
    {
      "epoch": 0.381275314911705,
      "grad_norm": 0.10603820532560349,
      "learning_rate": 6.212349964534371e-05,
      "loss": 1.572,
      "step": 193216
    },
    {
      "epoch": 0.38133846087310147,
      "grad_norm": 0.12213683873414993,
      "learning_rate": 6.211715942097901e-05,
      "loss": 1.5836,
      "step": 193248
    },
    {
      "epoch": 0.38140160683449786,
      "grad_norm": 0.1066909059882164,
      "learning_rate": 6.211081919661432e-05,
      "loss": 1.5694,
      "step": 193280
    },
    {
      "epoch": 0.3814647527958943,
      "grad_norm": 0.13556654751300812,
      "learning_rate": 6.210447897224964e-05,
      "loss": 1.5812,
      "step": 193312
    },
    {
      "epoch": 0.38152789875729076,
      "grad_norm": 0.11676943302154541,
      "learning_rate": 6.209813874788495e-05,
      "loss": 1.5741,
      "step": 193344
    },
    {
      "epoch": 0.3815910447186872,
      "grad_norm": 0.10768906027078629,
      "learning_rate": 6.209179852352025e-05,
      "loss": 1.573,
      "step": 193376
    },
    {
      "epoch": 0.38165419068008366,
      "grad_norm": 0.12190432846546173,
      "learning_rate": 6.208545829915557e-05,
      "loss": 1.5749,
      "step": 193408
    },
    {
      "epoch": 0.3817173366414801,
      "grad_norm": 0.11266548931598663,
      "learning_rate": 6.207911807479087e-05,
      "loss": 1.5922,
      "step": 193440
    },
    {
      "epoch": 0.3817804826028765,
      "grad_norm": 0.10480077564716339,
      "learning_rate": 6.207277785042619e-05,
      "loss": 1.5857,
      "step": 193472
    },
    {
      "epoch": 0.38184362856427295,
      "grad_norm": 0.10797309130430222,
      "learning_rate": 6.206643762606149e-05,
      "loss": 1.5905,
      "step": 193504
    },
    {
      "epoch": 0.3819067745256694,
      "grad_norm": 0.10765618830919266,
      "learning_rate": 6.20600974016968e-05,
      "loss": 1.5811,
      "step": 193536
    },
    {
      "epoch": 0.38196992048706585,
      "grad_norm": 0.11219711601734161,
      "learning_rate": 6.205375717733212e-05,
      "loss": 1.5647,
      "step": 193568
    },
    {
      "epoch": 0.3820330664484623,
      "grad_norm": 0.10743927210569382,
      "learning_rate": 6.204741695296743e-05,
      "loss": 1.5866,
      "step": 193600
    },
    {
      "epoch": 0.38209621240985875,
      "grad_norm": 0.11673557013273239,
      "learning_rate": 6.204107672860273e-05,
      "loss": 1.5766,
      "step": 193632
    },
    {
      "epoch": 0.38215935837125514,
      "grad_norm": 0.11123722046613693,
      "learning_rate": 6.203473650423805e-05,
      "loss": 1.579,
      "step": 193664
    },
    {
      "epoch": 0.3822225043326516,
      "grad_norm": 0.11459823697805405,
      "learning_rate": 6.202839627987336e-05,
      "loss": 1.5669,
      "step": 193696
    },
    {
      "epoch": 0.38228565029404804,
      "grad_norm": 0.11273475736379623,
      "learning_rate": 6.202205605550867e-05,
      "loss": 1.5746,
      "step": 193728
    },
    {
      "epoch": 0.3823487962554445,
      "grad_norm": 0.11307339370250702,
      "learning_rate": 6.201571583114398e-05,
      "loss": 1.597,
      "step": 193760
    },
    {
      "epoch": 0.38241194221684094,
      "grad_norm": 0.10869086533784866,
      "learning_rate": 6.200937560677929e-05,
      "loss": 1.5897,
      "step": 193792
    },
    {
      "epoch": 0.3824750881782374,
      "grad_norm": 0.10897192358970642,
      "learning_rate": 6.20030353824146e-05,
      "loss": 1.583,
      "step": 193824
    },
    {
      "epoch": 0.38253823413963384,
      "grad_norm": 0.11251529306173325,
      "learning_rate": 6.19966951580499e-05,
      "loss": 1.5786,
      "step": 193856
    },
    {
      "epoch": 0.38260138010103023,
      "grad_norm": 0.1086493730545044,
      "learning_rate": 6.199035493368522e-05,
      "loss": 1.5915,
      "step": 193888
    },
    {
      "epoch": 0.3826645260624267,
      "grad_norm": 0.10941314697265625,
      "learning_rate": 6.198401470932052e-05,
      "loss": 1.5868,
      "step": 193920
    },
    {
      "epoch": 0.38272767202382313,
      "grad_norm": 0.11620660126209259,
      "learning_rate": 6.197767448495584e-05,
      "loss": 1.5681,
      "step": 193952
    },
    {
      "epoch": 0.3827908179852196,
      "grad_norm": 0.10568537563085556,
      "learning_rate": 6.197133426059115e-05,
      "loss": 1.5756,
      "step": 193984
    },
    {
      "epoch": 0.382853963946616,
      "grad_norm": 0.114474818110466,
      "learning_rate": 6.196499403622646e-05,
      "loss": 1.5711,
      "step": 194016
    },
    {
      "epoch": 0.3829171099080125,
      "grad_norm": 0.11805200576782227,
      "learning_rate": 6.195865381186177e-05,
      "loss": 1.5777,
      "step": 194048
    },
    {
      "epoch": 0.38298025586940887,
      "grad_norm": 0.1138908788561821,
      "learning_rate": 6.195231358749708e-05,
      "loss": 1.5711,
      "step": 194080
    },
    {
      "epoch": 0.3830434018308053,
      "grad_norm": 0.11321444809436798,
      "learning_rate": 6.19459733631324e-05,
      "loss": 1.5834,
      "step": 194112
    },
    {
      "epoch": 0.38310654779220177,
      "grad_norm": 0.10935060679912567,
      "learning_rate": 6.193963313876771e-05,
      "loss": 1.567,
      "step": 194144
    },
    {
      "epoch": 0.3831696937535982,
      "grad_norm": 0.1189296543598175,
      "learning_rate": 6.193329291440301e-05,
      "loss": 1.5924,
      "step": 194176
    },
    {
      "epoch": 0.38323283971499467,
      "grad_norm": 0.11690651625394821,
      "learning_rate": 6.192695269003833e-05,
      "loss": 1.5853,
      "step": 194208
    },
    {
      "epoch": 0.3832959856763911,
      "grad_norm": 0.10909455269575119,
      "learning_rate": 6.192061246567364e-05,
      "loss": 1.5802,
      "step": 194240
    },
    {
      "epoch": 0.3833591316377875,
      "grad_norm": 0.1151445284485817,
      "learning_rate": 6.191427224130894e-05,
      "loss": 1.5769,
      "step": 194272
    },
    {
      "epoch": 0.38342227759918396,
      "grad_norm": 0.1197195053100586,
      "learning_rate": 6.190793201694424e-05,
      "loss": 1.5814,
      "step": 194304
    },
    {
      "epoch": 0.3834854235605804,
      "grad_norm": 0.11474685370922089,
      "learning_rate": 6.190159179257956e-05,
      "loss": 1.5771,
      "step": 194336
    },
    {
      "epoch": 0.38354856952197686,
      "grad_norm": 0.11637883633375168,
      "learning_rate": 6.189525156821487e-05,
      "loss": 1.5773,
      "step": 194368
    },
    {
      "epoch": 0.3836117154833733,
      "grad_norm": 0.11324753612279892,
      "learning_rate": 6.188891134385019e-05,
      "loss": 1.5862,
      "step": 194400
    },
    {
      "epoch": 0.38367486144476975,
      "grad_norm": 0.10944308340549469,
      "learning_rate": 6.188257111948549e-05,
      "loss": 1.5815,
      "step": 194432
    },
    {
      "epoch": 0.38373800740616615,
      "grad_norm": 0.1073647141456604,
      "learning_rate": 6.18762308951208e-05,
      "loss": 1.5849,
      "step": 194464
    },
    {
      "epoch": 0.3838011533675626,
      "grad_norm": 0.11327075958251953,
      "learning_rate": 6.186989067075612e-05,
      "loss": 1.5674,
      "step": 194496
    },
    {
      "epoch": 0.38386429932895905,
      "grad_norm": 0.12217191606760025,
      "learning_rate": 6.186355044639143e-05,
      "loss": 1.578,
      "step": 194528
    },
    {
      "epoch": 0.3839274452903555,
      "grad_norm": 0.11427702009677887,
      "learning_rate": 6.185721022202674e-05,
      "loss": 1.5842,
      "step": 194560
    },
    {
      "epoch": 0.38399059125175194,
      "grad_norm": 0.11438681185245514,
      "learning_rate": 6.185086999766205e-05,
      "loss": 1.5772,
      "step": 194592
    },
    {
      "epoch": 0.3840537372131484,
      "grad_norm": 0.11550986021757126,
      "learning_rate": 6.184452977329736e-05,
      "loss": 1.5699,
      "step": 194624
    },
    {
      "epoch": 0.3841168831745448,
      "grad_norm": 0.11881895363330841,
      "learning_rate": 6.183818954893266e-05,
      "loss": 1.5809,
      "step": 194656
    },
    {
      "epoch": 0.38418002913594124,
      "grad_norm": 0.11112011224031448,
      "learning_rate": 6.183184932456798e-05,
      "loss": 1.5933,
      "step": 194688
    },
    {
      "epoch": 0.3842431750973377,
      "grad_norm": 0.11052552610635757,
      "learning_rate": 6.182550910020328e-05,
      "loss": 1.5922,
      "step": 194720
    },
    {
      "epoch": 0.38430632105873413,
      "grad_norm": 0.11437926441431046,
      "learning_rate": 6.181916887583859e-05,
      "loss": 1.5738,
      "step": 194752
    },
    {
      "epoch": 0.3843694670201306,
      "grad_norm": 0.11414770036935806,
      "learning_rate": 6.18128286514739e-05,
      "loss": 1.5811,
      "step": 194784
    },
    {
      "epoch": 0.38443261298152703,
      "grad_norm": 0.10907621681690216,
      "learning_rate": 6.180648842710922e-05,
      "loss": 1.5852,
      "step": 194816
    },
    {
      "epoch": 0.3844957589429234,
      "grad_norm": 0.10684679448604584,
      "learning_rate": 6.180014820274452e-05,
      "loss": 1.585,
      "step": 194848
    },
    {
      "epoch": 0.3845589049043199,
      "grad_norm": 0.1143089011311531,
      "learning_rate": 6.179380797837984e-05,
      "loss": 1.5803,
      "step": 194880
    },
    {
      "epoch": 0.3846220508657163,
      "grad_norm": 0.11577370762825012,
      "learning_rate": 6.178746775401515e-05,
      "loss": 1.5828,
      "step": 194912
    },
    {
      "epoch": 0.3846851968271128,
      "grad_norm": 0.11089134961366653,
      "learning_rate": 6.178112752965047e-05,
      "loss": 1.5771,
      "step": 194944
    },
    {
      "epoch": 0.3847483427885092,
      "grad_norm": 0.10856077820062637,
      "learning_rate": 6.177478730528577e-05,
      "loss": 1.5656,
      "step": 194976
    },
    {
      "epoch": 0.3848114887499057,
      "grad_norm": 0.11164447665214539,
      "learning_rate": 6.176844708092108e-05,
      "loss": 1.5736,
      "step": 195008
    },
    {
      "epoch": 0.38487463471130207,
      "grad_norm": 0.10956843197345734,
      "learning_rate": 6.17621068565564e-05,
      "loss": 1.5742,
      "step": 195040
    },
    {
      "epoch": 0.3849377806726985,
      "grad_norm": 0.10592419654130936,
      "learning_rate": 6.17557666321917e-05,
      "loss": 1.5705,
      "step": 195072
    },
    {
      "epoch": 0.38500092663409496,
      "grad_norm": 0.10649097710847855,
      "learning_rate": 6.174942640782701e-05,
      "loss": 1.5725,
      "step": 195104
    },
    {
      "epoch": 0.3850640725954914,
      "grad_norm": 0.11064901202917099,
      "learning_rate": 6.174308618346231e-05,
      "loss": 1.5805,
      "step": 195136
    },
    {
      "epoch": 0.38512721855688786,
      "grad_norm": 0.11852394044399261,
      "learning_rate": 6.173674595909763e-05,
      "loss": 1.5839,
      "step": 195168
    },
    {
      "epoch": 0.3851903645182843,
      "grad_norm": 0.11115214228630066,
      "learning_rate": 6.173040573473294e-05,
      "loss": 1.582,
      "step": 195200
    },
    {
      "epoch": 0.3852535104796807,
      "grad_norm": 0.11130701750516891,
      "learning_rate": 6.172406551036826e-05,
      "loss": 1.5936,
      "step": 195232
    },
    {
      "epoch": 0.38531665644107715,
      "grad_norm": 0.1142987310886383,
      "learning_rate": 6.171772528600356e-05,
      "loss": 1.5739,
      "step": 195264
    },
    {
      "epoch": 0.3853798024024736,
      "grad_norm": 0.11486975848674774,
      "learning_rate": 6.171138506163887e-05,
      "loss": 1.5883,
      "step": 195296
    },
    {
      "epoch": 0.38544294836387005,
      "grad_norm": 0.10437566041946411,
      "learning_rate": 6.170504483727419e-05,
      "loss": 1.5796,
      "step": 195328
    },
    {
      "epoch": 0.3855060943252665,
      "grad_norm": 0.10351017117500305,
      "learning_rate": 6.16987046129095e-05,
      "loss": 1.5845,
      "step": 195360
    },
    {
      "epoch": 0.38556924028666295,
      "grad_norm": 0.11035379767417908,
      "learning_rate": 6.16923643885448e-05,
      "loss": 1.5936,
      "step": 195392
    },
    {
      "epoch": 0.38563238624805934,
      "grad_norm": 0.11144288629293442,
      "learning_rate": 6.168602416418012e-05,
      "loss": 1.5755,
      "step": 195424
    },
    {
      "epoch": 0.3856955322094558,
      "grad_norm": 0.10670211166143417,
      "learning_rate": 6.167968393981543e-05,
      "loss": 1.5687,
      "step": 195456
    },
    {
      "epoch": 0.38575867817085224,
      "grad_norm": 0.11042036861181259,
      "learning_rate": 6.167334371545073e-05,
      "loss": 1.5799,
      "step": 195488
    },
    {
      "epoch": 0.3858218241322487,
      "grad_norm": 0.11746933311223984,
      "learning_rate": 6.166700349108605e-05,
      "loss": 1.5903,
      "step": 195520
    },
    {
      "epoch": 0.38588497009364514,
      "grad_norm": 0.11040772497653961,
      "learning_rate": 6.166066326672135e-05,
      "loss": 1.5742,
      "step": 195552
    },
    {
      "epoch": 0.3859481160550416,
      "grad_norm": 0.11326251924037933,
      "learning_rate": 6.165432304235666e-05,
      "loss": 1.5886,
      "step": 195584
    },
    {
      "epoch": 0.38601126201643804,
      "grad_norm": 0.11498472094535828,
      "learning_rate": 6.164798281799198e-05,
      "loss": 1.5862,
      "step": 195616
    },
    {
      "epoch": 0.38607440797783443,
      "grad_norm": 0.10761165618896484,
      "learning_rate": 6.164164259362728e-05,
      "loss": 1.5752,
      "step": 195648
    },
    {
      "epoch": 0.3861375539392309,
      "grad_norm": 0.1089029312133789,
      "learning_rate": 6.163530236926259e-05,
      "loss": 1.588,
      "step": 195680
    },
    {
      "epoch": 0.38620069990062733,
      "grad_norm": 0.10804031044244766,
      "learning_rate": 6.16289621448979e-05,
      "loss": 1.5642,
      "step": 195712
    },
    {
      "epoch": 0.3862638458620238,
      "grad_norm": 0.10668666660785675,
      "learning_rate": 6.162262192053322e-05,
      "loss": 1.5716,
      "step": 195744
    },
    {
      "epoch": 0.38632699182342023,
      "grad_norm": 0.11377915740013123,
      "learning_rate": 6.161628169616852e-05,
      "loss": 1.5715,
      "step": 195776
    },
    {
      "epoch": 0.3863901377848167,
      "grad_norm": 0.11925654113292694,
      "learning_rate": 6.160994147180384e-05,
      "loss": 1.5834,
      "step": 195808
    },
    {
      "epoch": 0.38645328374621307,
      "grad_norm": 0.11272826790809631,
      "learning_rate": 6.160360124743915e-05,
      "loss": 1.5727,
      "step": 195840
    },
    {
      "epoch": 0.3865164297076095,
      "grad_norm": 0.11087898910045624,
      "learning_rate": 6.159726102307447e-05,
      "loss": 1.5846,
      "step": 195872
    },
    {
      "epoch": 0.38657957566900597,
      "grad_norm": 0.1085660457611084,
      "learning_rate": 6.159092079870977e-05,
      "loss": 1.5751,
      "step": 195904
    },
    {
      "epoch": 0.3866427216304024,
      "grad_norm": 0.11042071133852005,
      "learning_rate": 6.158458057434507e-05,
      "loss": 1.5622,
      "step": 195936
    },
    {
      "epoch": 0.38670586759179887,
      "grad_norm": 0.12418434768915176,
      "learning_rate": 6.157824034998038e-05,
      "loss": 1.5807,
      "step": 195968
    },
    {
      "epoch": 0.3867690135531953,
      "grad_norm": 0.11331064999103546,
      "learning_rate": 6.15719001256157e-05,
      "loss": 1.5893,
      "step": 196000
    },
    {
      "epoch": 0.3868321595145917,
      "grad_norm": 0.11731262505054474,
      "learning_rate": 6.156555990125101e-05,
      "loss": 1.5804,
      "step": 196032
    },
    {
      "epoch": 0.38689530547598816,
      "grad_norm": 0.10680776834487915,
      "learning_rate": 6.155921967688631e-05,
      "loss": 1.5781,
      "step": 196064
    },
    {
      "epoch": 0.3869584514373846,
      "grad_norm": 0.11508776247501373,
      "learning_rate": 6.155287945252163e-05,
      "loss": 1.5825,
      "step": 196096
    },
    {
      "epoch": 0.38702159739878106,
      "grad_norm": 0.11240911483764648,
      "learning_rate": 6.154653922815694e-05,
      "loss": 1.5853,
      "step": 196128
    },
    {
      "epoch": 0.3870847433601775,
      "grad_norm": 0.10966435819864273,
      "learning_rate": 6.154019900379226e-05,
      "loss": 1.5899,
      "step": 196160
    },
    {
      "epoch": 0.38714788932157396,
      "grad_norm": 0.10919499397277832,
      "learning_rate": 6.153385877942756e-05,
      "loss": 1.5804,
      "step": 196192
    },
    {
      "epoch": 0.38721103528297035,
      "grad_norm": 0.11427536606788635,
      "learning_rate": 6.152751855506287e-05,
      "loss": 1.5607,
      "step": 196224
    },
    {
      "epoch": 0.3872741812443668,
      "grad_norm": 0.11667704582214355,
      "learning_rate": 6.152117833069819e-05,
      "loss": 1.5761,
      "step": 196256
    },
    {
      "epoch": 0.38733732720576325,
      "grad_norm": 0.10493740439414978,
      "learning_rate": 6.15148381063335e-05,
      "loss": 1.5779,
      "step": 196288
    },
    {
      "epoch": 0.3874004731671597,
      "grad_norm": 0.11263034492731094,
      "learning_rate": 6.15084978819688e-05,
      "loss": 1.5686,
      "step": 196320
    },
    {
      "epoch": 0.38746361912855615,
      "grad_norm": 0.11960586905479431,
      "learning_rate": 6.15021576576041e-05,
      "loss": 1.5873,
      "step": 196352
    },
    {
      "epoch": 0.3875267650899526,
      "grad_norm": 0.12207693606615067,
      "learning_rate": 6.149581743323942e-05,
      "loss": 1.5862,
      "step": 196384
    },
    {
      "epoch": 0.387589911051349,
      "grad_norm": 0.11116329580545425,
      "learning_rate": 6.148947720887473e-05,
      "loss": 1.5743,
      "step": 196416
    },
    {
      "epoch": 0.38765305701274544,
      "grad_norm": 0.10941682010889053,
      "learning_rate": 6.148313698451003e-05,
      "loss": 1.5865,
      "step": 196448
    },
    {
      "epoch": 0.3877162029741419,
      "grad_norm": 0.11591745167970657,
      "learning_rate": 6.147679676014535e-05,
      "loss": 1.5834,
      "step": 196480
    },
    {
      "epoch": 0.38777934893553834,
      "grad_norm": 0.11391013115644455,
      "learning_rate": 6.147045653578066e-05,
      "loss": 1.577,
      "step": 196512
    },
    {
      "epoch": 0.3878424948969348,
      "grad_norm": 0.10873373597860336,
      "learning_rate": 6.146411631141598e-05,
      "loss": 1.5747,
      "step": 196544
    },
    {
      "epoch": 0.38790564085833124,
      "grad_norm": 0.11320281773805618,
      "learning_rate": 6.145777608705129e-05,
      "loss": 1.5858,
      "step": 196576
    },
    {
      "epoch": 0.38796878681972763,
      "grad_norm": 0.10731580853462219,
      "learning_rate": 6.145143586268659e-05,
      "loss": 1.5817,
      "step": 196608
    },
    {
      "epoch": 0.3880319327811241,
      "grad_norm": 0.10921253263950348,
      "learning_rate": 6.14450956383219e-05,
      "loss": 1.5692,
      "step": 196640
    },
    {
      "epoch": 0.3880950787425205,
      "grad_norm": 0.10596117377281189,
      "learning_rate": 6.143875541395722e-05,
      "loss": 1.5832,
      "step": 196672
    },
    {
      "epoch": 0.388158224703917,
      "grad_norm": 0.11492400616407394,
      "learning_rate": 6.143241518959254e-05,
      "loss": 1.5813,
      "step": 196704
    },
    {
      "epoch": 0.3882213706653134,
      "grad_norm": 0.11025400459766388,
      "learning_rate": 6.142607496522784e-05,
      "loss": 1.5814,
      "step": 196736
    },
    {
      "epoch": 0.3882845166267099,
      "grad_norm": 0.10787409543991089,
      "learning_rate": 6.141973474086314e-05,
      "loss": 1.5634,
      "step": 196768
    },
    {
      "epoch": 0.38834766258810627,
      "grad_norm": 0.1080102026462555,
      "learning_rate": 6.141339451649845e-05,
      "loss": 1.582,
      "step": 196800
    },
    {
      "epoch": 0.3884108085495027,
      "grad_norm": 0.10854236781597137,
      "learning_rate": 6.140705429213377e-05,
      "loss": 1.5647,
      "step": 196832
    },
    {
      "epoch": 0.38847395451089917,
      "grad_norm": 0.1165924221277237,
      "learning_rate": 6.140071406776907e-05,
      "loss": 1.5821,
      "step": 196864
    },
    {
      "epoch": 0.3885371004722956,
      "grad_norm": 0.11356721818447113,
      "learning_rate": 6.139437384340438e-05,
      "loss": 1.5894,
      "step": 196896
    },
    {
      "epoch": 0.38860024643369206,
      "grad_norm": 0.11147555708885193,
      "learning_rate": 6.13880336190397e-05,
      "loss": 1.5829,
      "step": 196928
    },
    {
      "epoch": 0.3886633923950885,
      "grad_norm": 0.11362142115831375,
      "learning_rate": 6.138169339467501e-05,
      "loss": 1.5818,
      "step": 196960
    },
    {
      "epoch": 0.3887265383564849,
      "grad_norm": 0.10542314499616623,
      "learning_rate": 6.137535317031031e-05,
      "loss": 1.5767,
      "step": 196992
    },
    {
      "epoch": 0.38878968431788136,
      "grad_norm": 0.11809996515512466,
      "learning_rate": 6.136901294594563e-05,
      "loss": 1.5883,
      "step": 197024
    },
    {
      "epoch": 0.3888528302792778,
      "grad_norm": 0.1132158488035202,
      "learning_rate": 6.136267272158094e-05,
      "loss": 1.579,
      "step": 197056
    },
    {
      "epoch": 0.38891597624067425,
      "grad_norm": 0.1142975464463234,
      "learning_rate": 6.135633249721626e-05,
      "loss": 1.574,
      "step": 197088
    },
    {
      "epoch": 0.3889791222020707,
      "grad_norm": 0.1133769080042839,
      "learning_rate": 6.134999227285156e-05,
      "loss": 1.5874,
      "step": 197120
    },
    {
      "epoch": 0.38904226816346715,
      "grad_norm": 0.11044921725988388,
      "learning_rate": 6.134365204848687e-05,
      "loss": 1.598,
      "step": 197152
    },
    {
      "epoch": 0.38910541412486355,
      "grad_norm": 0.10900735855102539,
      "learning_rate": 6.133731182412217e-05,
      "loss": 1.5843,
      "step": 197184
    },
    {
      "epoch": 0.38916856008626,
      "grad_norm": 0.10551530122756958,
      "learning_rate": 6.133097159975749e-05,
      "loss": 1.5755,
      "step": 197216
    },
    {
      "epoch": 0.38923170604765644,
      "grad_norm": 0.11120200157165527,
      "learning_rate": 6.13246313753928e-05,
      "loss": 1.5646,
      "step": 197248
    },
    {
      "epoch": 0.3892948520090529,
      "grad_norm": 0.11255236715078354,
      "learning_rate": 6.13182911510281e-05,
      "loss": 1.5818,
      "step": 197280
    },
    {
      "epoch": 0.38935799797044934,
      "grad_norm": 0.10892856866121292,
      "learning_rate": 6.131195092666342e-05,
      "loss": 1.5796,
      "step": 197312
    },
    {
      "epoch": 0.3894211439318458,
      "grad_norm": 0.11646037548780441,
      "learning_rate": 6.130561070229873e-05,
      "loss": 1.591,
      "step": 197344
    },
    {
      "epoch": 0.38948428989324224,
      "grad_norm": 0.12235072255134583,
      "learning_rate": 6.129927047793405e-05,
      "loss": 1.5769,
      "step": 197376
    },
    {
      "epoch": 0.38954743585463864,
      "grad_norm": 0.10207653045654297,
      "learning_rate": 6.129293025356935e-05,
      "loss": 1.5696,
      "step": 197408
    },
    {
      "epoch": 0.3896105818160351,
      "grad_norm": 0.11503254622220993,
      "learning_rate": 6.128659002920466e-05,
      "loss": 1.5717,
      "step": 197440
    },
    {
      "epoch": 0.38967372777743153,
      "grad_norm": 0.10136708617210388,
      "learning_rate": 6.128024980483998e-05,
      "loss": 1.564,
      "step": 197472
    },
    {
      "epoch": 0.389736873738828,
      "grad_norm": 0.11082805693149567,
      "learning_rate": 6.127390958047529e-05,
      "loss": 1.587,
      "step": 197504
    },
    {
      "epoch": 0.38980001970022443,
      "grad_norm": 0.12247280776500702,
      "learning_rate": 6.126756935611059e-05,
      "loss": 1.5784,
      "step": 197536
    },
    {
      "epoch": 0.3898631656616209,
      "grad_norm": 0.10812677443027496,
      "learning_rate": 6.126122913174591e-05,
      "loss": 1.5669,
      "step": 197568
    },
    {
      "epoch": 0.3899263116230173,
      "grad_norm": 0.10772112756967545,
      "learning_rate": 6.125488890738121e-05,
      "loss": 1.5817,
      "step": 197600
    },
    {
      "epoch": 0.3899894575844137,
      "grad_norm": 0.10378552973270416,
      "learning_rate": 6.124854868301652e-05,
      "loss": 1.5782,
      "step": 197632
    },
    {
      "epoch": 0.3900526035458102,
      "grad_norm": 0.10763803124427795,
      "learning_rate": 6.124220845865182e-05,
      "loss": 1.584,
      "step": 197664
    },
    {
      "epoch": 0.3901157495072066,
      "grad_norm": 0.11229921877384186,
      "learning_rate": 6.123586823428714e-05,
      "loss": 1.5807,
      "step": 197696
    },
    {
      "epoch": 0.39017889546860307,
      "grad_norm": 0.12197394669055939,
      "learning_rate": 6.122952800992245e-05,
      "loss": 1.5701,
      "step": 197728
    },
    {
      "epoch": 0.3902420414299995,
      "grad_norm": 0.11499131470918655,
      "learning_rate": 6.122318778555777e-05,
      "loss": 1.5824,
      "step": 197760
    },
    {
      "epoch": 0.3903051873913959,
      "grad_norm": 0.11056692153215408,
      "learning_rate": 6.121684756119307e-05,
      "loss": 1.579,
      "step": 197792
    },
    {
      "epoch": 0.39036833335279236,
      "grad_norm": 0.1076459214091301,
      "learning_rate": 6.121050733682838e-05,
      "loss": 1.5627,
      "step": 197824
    },
    {
      "epoch": 0.3904314793141888,
      "grad_norm": 0.11634511500597,
      "learning_rate": 6.12041671124637e-05,
      "loss": 1.5706,
      "step": 197856
    },
    {
      "epoch": 0.39049462527558526,
      "grad_norm": 0.11486385017633438,
      "learning_rate": 6.119782688809901e-05,
      "loss": 1.5729,
      "step": 197888
    },
    {
      "epoch": 0.3905577712369817,
      "grad_norm": 0.11045890301465988,
      "learning_rate": 6.119148666373433e-05,
      "loss": 1.5867,
      "step": 197920
    },
    {
      "epoch": 0.39062091719837816,
      "grad_norm": 0.10614673793315887,
      "learning_rate": 6.118514643936963e-05,
      "loss": 1.573,
      "step": 197952
    },
    {
      "epoch": 0.39068406315977455,
      "grad_norm": 0.12246210873126984,
      "learning_rate": 6.117880621500494e-05,
      "loss": 1.5937,
      "step": 197984
    },
    {
      "epoch": 0.390747209121171,
      "grad_norm": 0.11638857424259186,
      "learning_rate": 6.117246599064024e-05,
      "loss": 1.5845,
      "step": 198016
    },
    {
      "epoch": 0.39081035508256745,
      "grad_norm": 0.1108914241194725,
      "learning_rate": 6.116612576627556e-05,
      "loss": 1.5858,
      "step": 198048
    },
    {
      "epoch": 0.3908735010439639,
      "grad_norm": 0.11220531910657883,
      "learning_rate": 6.115978554191086e-05,
      "loss": 1.5848,
      "step": 198080
    },
    {
      "epoch": 0.39093664700536035,
      "grad_norm": 0.1126028373837471,
      "learning_rate": 6.115344531754617e-05,
      "loss": 1.5653,
      "step": 198112
    },
    {
      "epoch": 0.3909997929667568,
      "grad_norm": 0.10485746711492538,
      "learning_rate": 6.114710509318149e-05,
      "loss": 1.5839,
      "step": 198144
    },
    {
      "epoch": 0.3910629389281532,
      "grad_norm": 0.11530809104442596,
      "learning_rate": 6.11407648688168e-05,
      "loss": 1.5815,
      "step": 198176
    },
    {
      "epoch": 0.39112608488954964,
      "grad_norm": 0.11372833698987961,
      "learning_rate": 6.11344246444521e-05,
      "loss": 1.5826,
      "step": 198208
    },
    {
      "epoch": 0.3911892308509461,
      "grad_norm": 0.11040119081735611,
      "learning_rate": 6.112808442008742e-05,
      "loss": 1.5814,
      "step": 198240
    },
    {
      "epoch": 0.39125237681234254,
      "grad_norm": 0.11814737319946289,
      "learning_rate": 6.112174419572273e-05,
      "loss": 1.5835,
      "step": 198272
    },
    {
      "epoch": 0.391315522773739,
      "grad_norm": 0.11235001683235168,
      "learning_rate": 6.111540397135805e-05,
      "loss": 1.5842,
      "step": 198304
    },
    {
      "epoch": 0.39137866873513544,
      "grad_norm": 0.11039021611213684,
      "learning_rate": 6.110906374699335e-05,
      "loss": 1.5902,
      "step": 198336
    },
    {
      "epoch": 0.39144181469653183,
      "grad_norm": 0.12517079710960388,
      "learning_rate": 6.110272352262866e-05,
      "loss": 1.5619,
      "step": 198368
    },
    {
      "epoch": 0.3915049606579283,
      "grad_norm": 0.1051490530371666,
      "learning_rate": 6.109638329826398e-05,
      "loss": 1.5728,
      "step": 198400
    },
    {
      "epoch": 0.39156810661932473,
      "grad_norm": 0.11648889631032944,
      "learning_rate": 6.109004307389928e-05,
      "loss": 1.5799,
      "step": 198432
    },
    {
      "epoch": 0.3916312525807212,
      "grad_norm": 0.11079149693250656,
      "learning_rate": 6.108370284953458e-05,
      "loss": 1.5703,
      "step": 198464
    },
    {
      "epoch": 0.3916943985421176,
      "grad_norm": 0.11616569012403488,
      "learning_rate": 6.10773626251699e-05,
      "loss": 1.5915,
      "step": 198496
    },
    {
      "epoch": 0.3917575445035141,
      "grad_norm": 0.11501801013946533,
      "learning_rate": 6.107102240080521e-05,
      "loss": 1.5724,
      "step": 198528
    },
    {
      "epoch": 0.39182069046491047,
      "grad_norm": 0.11039219051599503,
      "learning_rate": 6.106468217644052e-05,
      "loss": 1.5906,
      "step": 198560
    },
    {
      "epoch": 0.3918838364263069,
      "grad_norm": 0.10678219795227051,
      "learning_rate": 6.105834195207584e-05,
      "loss": 1.5669,
      "step": 198592
    },
    {
      "epoch": 0.39194698238770337,
      "grad_norm": 0.1385871320962906,
      "learning_rate": 6.105200172771114e-05,
      "loss": 1.5865,
      "step": 198624
    },
    {
      "epoch": 0.3920101283490998,
      "grad_norm": 0.11026018857955933,
      "learning_rate": 6.104566150334645e-05,
      "loss": 1.5876,
      "step": 198656
    },
    {
      "epoch": 0.39207327431049627,
      "grad_norm": 0.10629809647798538,
      "learning_rate": 6.103932127898177e-05,
      "loss": 1.5842,
      "step": 198688
    },
    {
      "epoch": 0.3921364202718927,
      "grad_norm": 0.12065635621547699,
      "learning_rate": 6.1032981054617075e-05,
      "loss": 1.5825,
      "step": 198720
    },
    {
      "epoch": 0.3921995662332891,
      "grad_norm": 0.11541282385587692,
      "learning_rate": 6.1026640830252376e-05,
      "loss": 1.57,
      "step": 198752
    },
    {
      "epoch": 0.39226271219468556,
      "grad_norm": 0.11166813224554062,
      "learning_rate": 6.102030060588769e-05,
      "loss": 1.5764,
      "step": 198784
    },
    {
      "epoch": 0.392325858156082,
      "grad_norm": 0.11553598195314407,
      "learning_rate": 6.1013960381523006e-05,
      "loss": 1.5756,
      "step": 198816
    },
    {
      "epoch": 0.39238900411747846,
      "grad_norm": 0.11878856271505356,
      "learning_rate": 6.100762015715832e-05,
      "loss": 1.5705,
      "step": 198848
    },
    {
      "epoch": 0.3924521500788749,
      "grad_norm": 0.11386990547180176,
      "learning_rate": 6.100127993279362e-05,
      "loss": 1.5845,
      "step": 198880
    },
    {
      "epoch": 0.39251529604027136,
      "grad_norm": 0.1106158047914505,
      "learning_rate": 6.0994939708428936e-05,
      "loss": 1.5732,
      "step": 198912
    },
    {
      "epoch": 0.3925784420016678,
      "grad_norm": 0.11698216944932938,
      "learning_rate": 6.0988599484064243e-05,
      "loss": 1.57,
      "step": 198944
    },
    {
      "epoch": 0.3926415879630642,
      "grad_norm": 0.11797782778739929,
      "learning_rate": 6.098225925969956e-05,
      "loss": 1.5769,
      "step": 198976
    },
    {
      "epoch": 0.39270473392446065,
      "grad_norm": 0.10832861810922623,
      "learning_rate": 6.097591903533486e-05,
      "loss": 1.5623,
      "step": 199008
    },
    {
      "epoch": 0.3927678798858571,
      "grad_norm": 0.11208319664001465,
      "learning_rate": 6.0969578810970174e-05,
      "loss": 1.5737,
      "step": 199040
    },
    {
      "epoch": 0.39283102584725355,
      "grad_norm": 0.11608047038316727,
      "learning_rate": 6.096323858660549e-05,
      "loss": 1.5689,
      "step": 199072
    },
    {
      "epoch": 0.39289417180865,
      "grad_norm": 0.12055353820323944,
      "learning_rate": 6.09568983622408e-05,
      "loss": 1.5836,
      "step": 199104
    },
    {
      "epoch": 0.39295731777004644,
      "grad_norm": 0.11328496038913727,
      "learning_rate": 6.0950558137876104e-05,
      "loss": 1.5757,
      "step": 199136
    },
    {
      "epoch": 0.39302046373144284,
      "grad_norm": 0.10745330899953842,
      "learning_rate": 6.094421791351141e-05,
      "loss": 1.5712,
      "step": 199168
    },
    {
      "epoch": 0.3930836096928393,
      "grad_norm": 0.10427608340978622,
      "learning_rate": 6.0937877689146726e-05,
      "loss": 1.5799,
      "step": 199200
    },
    {
      "epoch": 0.39314675565423574,
      "grad_norm": 0.12024036794900894,
      "learning_rate": 6.093153746478204e-05,
      "loss": 1.5823,
      "step": 199232
    },
    {
      "epoch": 0.3932099016156322,
      "grad_norm": 0.11198412626981735,
      "learning_rate": 6.0925197240417355e-05,
      "loss": 1.5721,
      "step": 199264
    },
    {
      "epoch": 0.39327304757702863,
      "grad_norm": 0.1111232116818428,
      "learning_rate": 6.0918857016052656e-05,
      "loss": 1.5724,
      "step": 199296
    },
    {
      "epoch": 0.3933361935384251,
      "grad_norm": 0.11201640963554382,
      "learning_rate": 6.091251679168797e-05,
      "loss": 1.572,
      "step": 199328
    },
    {
      "epoch": 0.3933993394998215,
      "grad_norm": 0.10741990804672241,
      "learning_rate": 6.090617656732328e-05,
      "loss": 1.5742,
      "step": 199360
    },
    {
      "epoch": 0.3934624854612179,
      "grad_norm": 0.108777716755867,
      "learning_rate": 6.089983634295859e-05,
      "loss": 1.5682,
      "step": 199392
    },
    {
      "epoch": 0.3935256314226144,
      "grad_norm": 0.11616198718547821,
      "learning_rate": 6.0893496118593894e-05,
      "loss": 1.566,
      "step": 199424
    },
    {
      "epoch": 0.3935887773840108,
      "grad_norm": 0.11209969967603683,
      "learning_rate": 6.088715589422921e-05,
      "loss": 1.5812,
      "step": 199456
    },
    {
      "epoch": 0.3936519233454073,
      "grad_norm": 0.11173689365386963,
      "learning_rate": 6.088081566986452e-05,
      "loss": 1.5821,
      "step": 199488
    },
    {
      "epoch": 0.3937150693068037,
      "grad_norm": 0.10855965316295624,
      "learning_rate": 6.087447544549984e-05,
      "loss": 1.5813,
      "step": 199520
    },
    {
      "epoch": 0.3937782152682001,
      "grad_norm": 0.11043226718902588,
      "learning_rate": 6.086813522113514e-05,
      "loss": 1.5777,
      "step": 199552
    },
    {
      "epoch": 0.39384136122959656,
      "grad_norm": 0.11845913529396057,
      "learning_rate": 6.0861794996770447e-05,
      "loss": 1.5789,
      "step": 199584
    },
    {
      "epoch": 0.393904507190993,
      "grad_norm": 0.11043913662433624,
      "learning_rate": 6.085545477240576e-05,
      "loss": 1.591,
      "step": 199616
    },
    {
      "epoch": 0.39396765315238946,
      "grad_norm": 0.10733698308467865,
      "learning_rate": 6.0849114548041076e-05,
      "loss": 1.577,
      "step": 199648
    },
    {
      "epoch": 0.3940307991137859,
      "grad_norm": 0.11167687922716141,
      "learning_rate": 6.084277432367638e-05,
      "loss": 1.5764,
      "step": 199680
    },
    {
      "epoch": 0.39409394507518236,
      "grad_norm": 0.10581743717193604,
      "learning_rate": 6.083643409931169e-05,
      "loss": 1.5779,
      "step": 199712
    },
    {
      "epoch": 0.39415709103657876,
      "grad_norm": 0.1120852380990982,
      "learning_rate": 6.0830093874947006e-05,
      "loss": 1.5852,
      "step": 199744
    },
    {
      "epoch": 0.3942202369979752,
      "grad_norm": 0.11101929843425751,
      "learning_rate": 6.0823753650582314e-05,
      "loss": 1.5714,
      "step": 199776
    },
    {
      "epoch": 0.39428338295937165,
      "grad_norm": 0.10848644375801086,
      "learning_rate": 6.081741342621763e-05,
      "loss": 1.5767,
      "step": 199808
    },
    {
      "epoch": 0.3943465289207681,
      "grad_norm": 0.1133238673210144,
      "learning_rate": 6.081107320185293e-05,
      "loss": 1.5696,
      "step": 199840
    },
    {
      "epoch": 0.39440967488216455,
      "grad_norm": 0.11962737143039703,
      "learning_rate": 6.0804732977488244e-05,
      "loss": 1.5851,
      "step": 199872
    },
    {
      "epoch": 0.394472820843561,
      "grad_norm": 0.10968546569347382,
      "learning_rate": 6.079839275312356e-05,
      "loss": 1.5791,
      "step": 199904
    },
    {
      "epoch": 0.3945359668049574,
      "grad_norm": 0.10896339267492294,
      "learning_rate": 6.0792052528758866e-05,
      "loss": 1.5708,
      "step": 199936
    },
    {
      "epoch": 0.39459911276635384,
      "grad_norm": 0.10600137710571289,
      "learning_rate": 6.0785712304394174e-05,
      "loss": 1.5842,
      "step": 199968
    },
    {
      "epoch": 0.3946622587277503,
      "grad_norm": 0.10372083634138107,
      "learning_rate": 6.077937208002948e-05,
      "loss": 1.5675,
      "step": 200000
    },
    {
      "epoch": 0.39472540468914674,
      "grad_norm": 0.10900106281042099,
      "learning_rate": 6.0773031855664796e-05,
      "loss": 1.5709,
      "step": 200032
    },
    {
      "epoch": 0.3947885506505432,
      "grad_norm": 0.11632402241230011,
      "learning_rate": 6.076669163130011e-05,
      "loss": 1.5768,
      "step": 200064
    },
    {
      "epoch": 0.39485169661193964,
      "grad_norm": 0.10969655215740204,
      "learning_rate": 6.076035140693541e-05,
      "loss": 1.579,
      "step": 200096
    },
    {
      "epoch": 0.39491484257333603,
      "grad_norm": 0.10480445623397827,
      "learning_rate": 6.0754011182570726e-05,
      "loss": 1.5708,
      "step": 200128
    },
    {
      "epoch": 0.3949779885347325,
      "grad_norm": 0.1100243628025055,
      "learning_rate": 6.074767095820604e-05,
      "loss": 1.5842,
      "step": 200160
    },
    {
      "epoch": 0.39504113449612893,
      "grad_norm": 0.11034812033176422,
      "learning_rate": 6.074133073384135e-05,
      "loss": 1.5744,
      "step": 200192
    },
    {
      "epoch": 0.3951042804575254,
      "grad_norm": 0.1140894666314125,
      "learning_rate": 6.073499050947665e-05,
      "loss": 1.5799,
      "step": 200224
    },
    {
      "epoch": 0.39516742641892183,
      "grad_norm": 0.10401913523674011,
      "learning_rate": 6.0728650285111964e-05,
      "loss": 1.576,
      "step": 200256
    },
    {
      "epoch": 0.3952305723803183,
      "grad_norm": 0.11680079251527786,
      "learning_rate": 6.072231006074728e-05,
      "loss": 1.5888,
      "step": 200288
    },
    {
      "epoch": 0.3952937183417147,
      "grad_norm": 0.10789495706558228,
      "learning_rate": 6.0715969836382593e-05,
      "loss": 1.5715,
      "step": 200320
    },
    {
      "epoch": 0.3953568643031111,
      "grad_norm": 0.11175715923309326,
      "learning_rate": 6.0709629612017894e-05,
      "loss": 1.5688,
      "step": 200352
    },
    {
      "epoch": 0.39542001026450757,
      "grad_norm": 0.10848549008369446,
      "learning_rate": 6.070328938765321e-05,
      "loss": 1.5862,
      "step": 200384
    },
    {
      "epoch": 0.395483156225904,
      "grad_norm": 0.11460563540458679,
      "learning_rate": 6.069694916328852e-05,
      "loss": 1.5731,
      "step": 200416
    },
    {
      "epoch": 0.39554630218730047,
      "grad_norm": 0.11236392706632614,
      "learning_rate": 6.069060893892383e-05,
      "loss": 1.575,
      "step": 200448
    },
    {
      "epoch": 0.3956094481486969,
      "grad_norm": 0.11141322553157806,
      "learning_rate": 6.0684268714559146e-05,
      "loss": 1.5734,
      "step": 200480
    },
    {
      "epoch": 0.3956725941100933,
      "grad_norm": 0.12026748806238174,
      "learning_rate": 6.067792849019445e-05,
      "loss": 1.5738,
      "step": 200512
    },
    {
      "epoch": 0.39573574007148976,
      "grad_norm": 0.11951130628585815,
      "learning_rate": 6.067158826582976e-05,
      "loss": 1.5958,
      "step": 200544
    },
    {
      "epoch": 0.3957988860328862,
      "grad_norm": 0.1094803735613823,
      "learning_rate": 6.066524804146507e-05,
      "loss": 1.5785,
      "step": 200576
    },
    {
      "epoch": 0.39586203199428266,
      "grad_norm": 0.1262432336807251,
      "learning_rate": 6.0658907817100384e-05,
      "loss": 1.5771,
      "step": 200608
    },
    {
      "epoch": 0.3959251779556791,
      "grad_norm": 0.10395687818527222,
      "learning_rate": 6.0652567592735685e-05,
      "loss": 1.5709,
      "step": 200640
    },
    {
      "epoch": 0.39598832391707556,
      "grad_norm": 0.10931379348039627,
      "learning_rate": 6.0646227368371e-05,
      "loss": 1.5707,
      "step": 200672
    },
    {
      "epoch": 0.396051469878472,
      "grad_norm": 0.12008059769868851,
      "learning_rate": 6.0639887144006314e-05,
      "loss": 1.5895,
      "step": 200704
    },
    {
      "epoch": 0.3961146158398684,
      "grad_norm": 0.12564410269260406,
      "learning_rate": 6.063354691964163e-05,
      "loss": 1.5916,
      "step": 200736
    },
    {
      "epoch": 0.39617776180126485,
      "grad_norm": 0.11612212657928467,
      "learning_rate": 6.062720669527693e-05,
      "loss": 1.5876,
      "step": 200768
    },
    {
      "epoch": 0.3962409077626613,
      "grad_norm": 0.11274783313274384,
      "learning_rate": 6.0620866470912244e-05,
      "loss": 1.5669,
      "step": 200800
    },
    {
      "epoch": 0.39630405372405775,
      "grad_norm": 0.11187774688005447,
      "learning_rate": 6.061452624654755e-05,
      "loss": 1.5878,
      "step": 200832
    },
    {
      "epoch": 0.3963671996854542,
      "grad_norm": 0.10816112160682678,
      "learning_rate": 6.0608186022182866e-05,
      "loss": 1.5697,
      "step": 200864
    },
    {
      "epoch": 0.39643034564685065,
      "grad_norm": 0.10745732486248016,
      "learning_rate": 6.060184579781817e-05,
      "loss": 1.578,
      "step": 200896
    },
    {
      "epoch": 0.39649349160824704,
      "grad_norm": 0.10564573854207993,
      "learning_rate": 6.059550557345348e-05,
      "loss": 1.5656,
      "step": 200928
    },
    {
      "epoch": 0.3965566375696435,
      "grad_norm": 0.11073864251375198,
      "learning_rate": 6.0589165349088797e-05,
      "loss": 1.5802,
      "step": 200960
    },
    {
      "epoch": 0.39661978353103994,
      "grad_norm": 0.11043330281972885,
      "learning_rate": 6.0582825124724104e-05,
      "loss": 1.5776,
      "step": 200992
    },
    {
      "epoch": 0.3966829294924364,
      "grad_norm": 0.11076078563928604,
      "learning_rate": 6.057648490035941e-05,
      "loss": 1.5665,
      "step": 201024
    },
    {
      "epoch": 0.39674607545383284,
      "grad_norm": 0.11658389121294022,
      "learning_rate": 6.057014467599472e-05,
      "loss": 1.5756,
      "step": 201056
    },
    {
      "epoch": 0.3968092214152293,
      "grad_norm": 0.10756539553403854,
      "learning_rate": 6.0563804451630035e-05,
      "loss": 1.5681,
      "step": 201088
    },
    {
      "epoch": 0.3968723673766257,
      "grad_norm": 0.10969916731119156,
      "learning_rate": 6.055746422726535e-05,
      "loss": 1.5819,
      "step": 201120
    },
    {
      "epoch": 0.39693551333802213,
      "grad_norm": 0.12701939046382904,
      "learning_rate": 6.0551124002900664e-05,
      "loss": 1.5707,
      "step": 201152
    },
    {
      "epoch": 0.3969986592994186,
      "grad_norm": 0.11056146025657654,
      "learning_rate": 6.0544783778535965e-05,
      "loss": 1.5809,
      "step": 201184
    },
    {
      "epoch": 0.397061805260815,
      "grad_norm": 0.10754179954528809,
      "learning_rate": 6.053844355417127e-05,
      "loss": 1.5782,
      "step": 201216
    },
    {
      "epoch": 0.3971249512222115,
      "grad_norm": 0.1104857474565506,
      "learning_rate": 6.053210332980659e-05,
      "loss": 1.5679,
      "step": 201248
    },
    {
      "epoch": 0.3971880971836079,
      "grad_norm": 0.11548054218292236,
      "learning_rate": 6.05257631054419e-05,
      "loss": 1.5762,
      "step": 201280
    },
    {
      "epoch": 0.3972512431450043,
      "grad_norm": 0.11069951951503754,
      "learning_rate": 6.05194228810772e-05,
      "loss": 1.5914,
      "step": 201312
    },
    {
      "epoch": 0.39731438910640077,
      "grad_norm": 0.1240776926279068,
      "learning_rate": 6.051308265671252e-05,
      "loss": 1.5787,
      "step": 201344
    },
    {
      "epoch": 0.3973775350677972,
      "grad_norm": 0.11726592481136322,
      "learning_rate": 6.050674243234783e-05,
      "loss": 1.5733,
      "step": 201376
    },
    {
      "epoch": 0.39744068102919367,
      "grad_norm": 0.11535972356796265,
      "learning_rate": 6.050040220798314e-05,
      "loss": 1.5776,
      "step": 201408
    },
    {
      "epoch": 0.3975038269905901,
      "grad_norm": 0.11283685266971588,
      "learning_rate": 6.049406198361845e-05,
      "loss": 1.576,
      "step": 201440
    },
    {
      "epoch": 0.39756697295198656,
      "grad_norm": 0.11160247772932053,
      "learning_rate": 6.0487721759253755e-05,
      "loss": 1.5798,
      "step": 201472
    },
    {
      "epoch": 0.39763011891338296,
      "grad_norm": 0.11608602851629257,
      "learning_rate": 6.048138153488907e-05,
      "loss": 1.5868,
      "step": 201504
    },
    {
      "epoch": 0.3976932648747794,
      "grad_norm": 0.11454786360263824,
      "learning_rate": 6.0475041310524384e-05,
      "loss": 1.57,
      "step": 201536
    },
    {
      "epoch": 0.39775641083617586,
      "grad_norm": 0.12565377354621887,
      "learning_rate": 6.0468701086159685e-05,
      "loss": 1.5768,
      "step": 201568
    },
    {
      "epoch": 0.3978195567975723,
      "grad_norm": 0.10945910215377808,
      "learning_rate": 6.0462360861795e-05,
      "loss": 1.5659,
      "step": 201600
    },
    {
      "epoch": 0.39788270275896875,
      "grad_norm": 0.129023939371109,
      "learning_rate": 6.045602063743031e-05,
      "loss": 1.5792,
      "step": 201632
    },
    {
      "epoch": 0.3979458487203652,
      "grad_norm": 0.1132936179637909,
      "learning_rate": 6.044968041306562e-05,
      "loss": 1.5647,
      "step": 201664
    },
    {
      "epoch": 0.3980089946817616,
      "grad_norm": 0.11813409626483917,
      "learning_rate": 6.044334018870092e-05,
      "loss": 1.5677,
      "step": 201696
    },
    {
      "epoch": 0.39807214064315805,
      "grad_norm": 0.12388277053833008,
      "learning_rate": 6.043699996433624e-05,
      "loss": 1.5806,
      "step": 201728
    },
    {
      "epoch": 0.3981352866045545,
      "grad_norm": 0.10976135730743408,
      "learning_rate": 6.043065973997155e-05,
      "loss": 1.5794,
      "step": 201760
    },
    {
      "epoch": 0.39819843256595094,
      "grad_norm": 0.10981694608926773,
      "learning_rate": 6.042431951560687e-05,
      "loss": 1.5687,
      "step": 201792
    },
    {
      "epoch": 0.3982615785273474,
      "grad_norm": 0.10986041277647018,
      "learning_rate": 6.0417979291242175e-05,
      "loss": 1.5913,
      "step": 201824
    },
    {
      "epoch": 0.39832472448874384,
      "grad_norm": 0.12826699018478394,
      "learning_rate": 6.0411639066877476e-05,
      "loss": 1.578,
      "step": 201856
    },
    {
      "epoch": 0.39838787045014024,
      "grad_norm": 0.10725842416286469,
      "learning_rate": 6.040529884251279e-05,
      "loss": 1.5701,
      "step": 201888
    },
    {
      "epoch": 0.3984510164115367,
      "grad_norm": 0.1162799596786499,
      "learning_rate": 6.0398958618148105e-05,
      "loss": 1.5872,
      "step": 201920
    },
    {
      "epoch": 0.39851416237293313,
      "grad_norm": 0.11675944924354553,
      "learning_rate": 6.039261839378342e-05,
      "loss": 1.5743,
      "step": 201952
    },
    {
      "epoch": 0.3985773083343296,
      "grad_norm": 0.10371293872594833,
      "learning_rate": 6.038627816941872e-05,
      "loss": 1.5736,
      "step": 201984
    },
    {
      "epoch": 0.39864045429572603,
      "grad_norm": 0.1117289662361145,
      "learning_rate": 6.0379937945054035e-05,
      "loss": 1.5746,
      "step": 202016
    },
    {
      "epoch": 0.3987036002571225,
      "grad_norm": 0.10964484512805939,
      "learning_rate": 6.037359772068934e-05,
      "loss": 1.5646,
      "step": 202048
    },
    {
      "epoch": 0.3987667462185189,
      "grad_norm": 0.10782244801521301,
      "learning_rate": 6.036725749632466e-05,
      "loss": 1.5647,
      "step": 202080
    },
    {
      "epoch": 0.3988298921799153,
      "grad_norm": 0.1156175285577774,
      "learning_rate": 6.036091727195996e-05,
      "loss": 1.5759,
      "step": 202112
    },
    {
      "epoch": 0.3988930381413118,
      "grad_norm": 0.10407830774784088,
      "learning_rate": 6.035457704759527e-05,
      "loss": 1.5801,
      "step": 202144
    },
    {
      "epoch": 0.3989561841027082,
      "grad_norm": 0.10547314584255219,
      "learning_rate": 6.034823682323059e-05,
      "loss": 1.577,
      "step": 202176
    },
    {
      "epoch": 0.39901933006410467,
      "grad_norm": 0.1167195662856102,
      "learning_rate": 6.03418965988659e-05,
      "loss": 1.5745,
      "step": 202208
    },
    {
      "epoch": 0.3990824760255011,
      "grad_norm": 0.11346475780010223,
      "learning_rate": 6.03355563745012e-05,
      "loss": 1.5798,
      "step": 202240
    },
    {
      "epoch": 0.3991456219868975,
      "grad_norm": 0.11195581406354904,
      "learning_rate": 6.032921615013651e-05,
      "loss": 1.5648,
      "step": 202272
    },
    {
      "epoch": 0.39920876794829396,
      "grad_norm": 0.11395737528800964,
      "learning_rate": 6.0322875925771825e-05,
      "loss": 1.5743,
      "step": 202304
    },
    {
      "epoch": 0.3992719139096904,
      "grad_norm": 0.11736161261796951,
      "learning_rate": 6.031653570140714e-05,
      "loss": 1.5795,
      "step": 202336
    },
    {
      "epoch": 0.39933505987108686,
      "grad_norm": 0.13533258438110352,
      "learning_rate": 6.031019547704244e-05,
      "loss": 1.5762,
      "step": 202368
    },
    {
      "epoch": 0.3993982058324833,
      "grad_norm": 0.10975204408168793,
      "learning_rate": 6.0303855252677755e-05,
      "loss": 1.5763,
      "step": 202400
    },
    {
      "epoch": 0.39946135179387976,
      "grad_norm": 0.11502835154533386,
      "learning_rate": 6.029751502831307e-05,
      "loss": 1.5715,
      "step": 202432
    },
    {
      "epoch": 0.3995244977552762,
      "grad_norm": 0.11812707036733627,
      "learning_rate": 6.029117480394838e-05,
      "loss": 1.5798,
      "step": 202464
    },
    {
      "epoch": 0.3995876437166726,
      "grad_norm": 0.12040771543979645,
      "learning_rate": 6.028483457958369e-05,
      "loss": 1.5762,
      "step": 202496
    },
    {
      "epoch": 0.39965078967806905,
      "grad_norm": 0.11257939785718918,
      "learning_rate": 6.027849435521899e-05,
      "loss": 1.5734,
      "step": 202528
    },
    {
      "epoch": 0.3997139356394655,
      "grad_norm": 0.10618849843740463,
      "learning_rate": 6.027215413085431e-05,
      "loss": 1.579,
      "step": 202560
    },
    {
      "epoch": 0.39977708160086195,
      "grad_norm": 0.11009246110916138,
      "learning_rate": 6.026581390648962e-05,
      "loss": 1.5727,
      "step": 202592
    },
    {
      "epoch": 0.3998402275622584,
      "grad_norm": 0.10986966639757156,
      "learning_rate": 6.025947368212494e-05,
      "loss": 1.5919,
      "step": 202624
    },
    {
      "epoch": 0.39990337352365485,
      "grad_norm": 0.10643728077411652,
      "learning_rate": 6.025313345776024e-05,
      "loss": 1.585,
      "step": 202656
    },
    {
      "epoch": 0.39996651948505124,
      "grad_norm": 0.11467166244983673,
      "learning_rate": 6.0246793233395546e-05,
      "loss": 1.5767,
      "step": 202688
    },
    {
      "epoch": 0.4000296654464477,
      "grad_norm": 0.10692814737558365,
      "learning_rate": 6.024045300903086e-05,
      "loss": 1.5665,
      "step": 202720
    },
    {
      "epoch": 0.40009281140784414,
      "grad_norm": 0.11108317971229553,
      "learning_rate": 6.0234112784666175e-05,
      "loss": 1.5812,
      "step": 202752
    },
    {
      "epoch": 0.4001559573692406,
      "grad_norm": 0.11161785572767258,
      "learning_rate": 6.0227772560301476e-05,
      "loss": 1.5857,
      "step": 202784
    },
    {
      "epoch": 0.40021910333063704,
      "grad_norm": 0.11147263646125793,
      "learning_rate": 6.022143233593679e-05,
      "loss": 1.5731,
      "step": 202816
    },
    {
      "epoch": 0.4002822492920335,
      "grad_norm": 0.10793454945087433,
      "learning_rate": 6.0215092111572105e-05,
      "loss": 1.5718,
      "step": 202848
    },
    {
      "epoch": 0.4003453952534299,
      "grad_norm": 0.12187591195106506,
      "learning_rate": 6.020875188720741e-05,
      "loss": 1.566,
      "step": 202880
    },
    {
      "epoch": 0.40040854121482633,
      "grad_norm": 0.11108333617448807,
      "learning_rate": 6.0202411662842714e-05,
      "loss": 1.5865,
      "step": 202912
    },
    {
      "epoch": 0.4004716871762228,
      "grad_norm": 0.1154702678322792,
      "learning_rate": 6.019607143847803e-05,
      "loss": 1.5675,
      "step": 202944
    },
    {
      "epoch": 0.40053483313761923,
      "grad_norm": 0.10599309951066971,
      "learning_rate": 6.018973121411334e-05,
      "loss": 1.5795,
      "step": 202976
    },
    {
      "epoch": 0.4005979790990157,
      "grad_norm": 0.11422599107027054,
      "learning_rate": 6.018339098974866e-05,
      "loss": 1.5697,
      "step": 203008
    },
    {
      "epoch": 0.4006611250604121,
      "grad_norm": 0.12432745844125748,
      "learning_rate": 6.017705076538396e-05,
      "loss": 1.5739,
      "step": 203040
    },
    {
      "epoch": 0.4007242710218085,
      "grad_norm": 0.11741861701011658,
      "learning_rate": 6.017071054101927e-05,
      "loss": 1.5793,
      "step": 203072
    },
    {
      "epoch": 0.40078741698320497,
      "grad_norm": 0.11207618564367294,
      "learning_rate": 6.016437031665458e-05,
      "loss": 1.5718,
      "step": 203104
    },
    {
      "epoch": 0.4008505629446014,
      "grad_norm": 0.10925807803869247,
      "learning_rate": 6.0158030092289895e-05,
      "loss": 1.5888,
      "step": 203136
    },
    {
      "epoch": 0.40091370890599787,
      "grad_norm": 0.11147225648164749,
      "learning_rate": 6.015168986792521e-05,
      "loss": 1.5831,
      "step": 203168
    },
    {
      "epoch": 0.4009768548673943,
      "grad_norm": 0.11625019460916519,
      "learning_rate": 6.014534964356051e-05,
      "loss": 1.5781,
      "step": 203200
    },
    {
      "epoch": 0.40104000082879077,
      "grad_norm": 0.11803005635738373,
      "learning_rate": 6.0139009419195826e-05,
      "loss": 1.5738,
      "step": 203232
    },
    {
      "epoch": 0.40110314679018716,
      "grad_norm": 0.11831724643707275,
      "learning_rate": 6.013266919483114e-05,
      "loss": 1.5708,
      "step": 203264
    },
    {
      "epoch": 0.4011662927515836,
      "grad_norm": 0.11004021763801575,
      "learning_rate": 6.012632897046645e-05,
      "loss": 1.5755,
      "step": 203296
    },
    {
      "epoch": 0.40122943871298006,
      "grad_norm": 0.1137085109949112,
      "learning_rate": 6.011998874610175e-05,
      "loss": 1.5873,
      "step": 203328
    },
    {
      "epoch": 0.4012925846743765,
      "grad_norm": 0.11433461308479309,
      "learning_rate": 6.0113648521737064e-05,
      "loss": 1.5818,
      "step": 203360
    },
    {
      "epoch": 0.40135573063577296,
      "grad_norm": 0.1072268933057785,
      "learning_rate": 6.010730829737238e-05,
      "loss": 1.5763,
      "step": 203392
    },
    {
      "epoch": 0.4014188765971694,
      "grad_norm": 0.10801132768392563,
      "learning_rate": 6.010096807300769e-05,
      "loss": 1.5786,
      "step": 203424
    },
    {
      "epoch": 0.4014820225585658,
      "grad_norm": 0.11331193894147873,
      "learning_rate": 6.0094627848642994e-05,
      "loss": 1.5681,
      "step": 203456
    },
    {
      "epoch": 0.40154516851996225,
      "grad_norm": 0.11307258158922195,
      "learning_rate": 6.008828762427831e-05,
      "loss": 1.5718,
      "step": 203488
    },
    {
      "epoch": 0.4016083144813587,
      "grad_norm": 0.10479161143302917,
      "learning_rate": 6.0081947399913616e-05,
      "loss": 1.5695,
      "step": 203520
    },
    {
      "epoch": 0.40167146044275515,
      "grad_norm": 0.11359573900699615,
      "learning_rate": 6.007560717554893e-05,
      "loss": 1.5744,
      "step": 203552
    },
    {
      "epoch": 0.4017346064041516,
      "grad_norm": 0.10413649678230286,
      "learning_rate": 6.006926695118423e-05,
      "loss": 1.5608,
      "step": 203584
    },
    {
      "epoch": 0.40179775236554804,
      "grad_norm": 0.10917824506759644,
      "learning_rate": 6.0062926726819546e-05,
      "loss": 1.5695,
      "step": 203616
    },
    {
      "epoch": 0.40186089832694444,
      "grad_norm": 0.11242704838514328,
      "learning_rate": 6.005658650245486e-05,
      "loss": 1.5775,
      "step": 203648
    },
    {
      "epoch": 0.4019240442883409,
      "grad_norm": 0.129756361246109,
      "learning_rate": 6.0050246278090175e-05,
      "loss": 1.5791,
      "step": 203680
    },
    {
      "epoch": 0.40198719024973734,
      "grad_norm": 0.10912884771823883,
      "learning_rate": 6.0043906053725476e-05,
      "loss": 1.5649,
      "step": 203712
    },
    {
      "epoch": 0.4020503362111338,
      "grad_norm": 0.11269836872816086,
      "learning_rate": 6.0037565829360784e-05,
      "loss": 1.5732,
      "step": 203744
    },
    {
      "epoch": 0.40211348217253023,
      "grad_norm": 0.1197069063782692,
      "learning_rate": 6.00312256049961e-05,
      "loss": 1.5762,
      "step": 203776
    },
    {
      "epoch": 0.4021766281339267,
      "grad_norm": 0.11089413613080978,
      "learning_rate": 6.002488538063141e-05,
      "loss": 1.5678,
      "step": 203808
    },
    {
      "epoch": 0.4022397740953231,
      "grad_norm": 0.11494425684213638,
      "learning_rate": 6.001854515626673e-05,
      "loss": 1.5886,
      "step": 203840
    },
    {
      "epoch": 0.4023029200567195,
      "grad_norm": 0.11617664992809296,
      "learning_rate": 6.001220493190203e-05,
      "loss": 1.5773,
      "step": 203872
    },
    {
      "epoch": 0.402366066018116,
      "grad_norm": 0.1175367683172226,
      "learning_rate": 6.000586470753734e-05,
      "loss": 1.5683,
      "step": 203904
    },
    {
      "epoch": 0.4024292119795124,
      "grad_norm": 0.11487821489572525,
      "learning_rate": 5.999952448317265e-05,
      "loss": 1.572,
      "step": 203936
    },
    {
      "epoch": 0.4024923579409089,
      "grad_norm": 0.11277016252279282,
      "learning_rate": 5.9993184258807966e-05,
      "loss": 1.5746,
      "step": 203968
    },
    {
      "epoch": 0.4025555039023053,
      "grad_norm": 0.10945415496826172,
      "learning_rate": 5.998684403444327e-05,
      "loss": 1.5769,
      "step": 204000
    },
    {
      "epoch": 0.40261864986370177,
      "grad_norm": 0.1095406711101532,
      "learning_rate": 5.998050381007858e-05,
      "loss": 1.5735,
      "step": 204032
    },
    {
      "epoch": 0.40268179582509817,
      "grad_norm": 0.11577536910772324,
      "learning_rate": 5.9974163585713896e-05,
      "loss": 1.5988,
      "step": 204064
    },
    {
      "epoch": 0.4027449417864946,
      "grad_norm": 0.11286237090826035,
      "learning_rate": 5.9967823361349204e-05,
      "loss": 1.566,
      "step": 204096
    },
    {
      "epoch": 0.40280808774789106,
      "grad_norm": 0.1081601157784462,
      "learning_rate": 5.996148313698451e-05,
      "loss": 1.5745,
      "step": 204128
    },
    {
      "epoch": 0.4028712337092875,
      "grad_norm": 0.11012493073940277,
      "learning_rate": 5.995514291261982e-05,
      "loss": 1.5732,
      "step": 204160
    },
    {
      "epoch": 0.40293437967068396,
      "grad_norm": 0.11891575902700424,
      "learning_rate": 5.9948802688255134e-05,
      "loss": 1.569,
      "step": 204192
    },
    {
      "epoch": 0.4029975256320804,
      "grad_norm": 0.11340872943401337,
      "learning_rate": 5.994246246389045e-05,
      "loss": 1.572,
      "step": 204224
    },
    {
      "epoch": 0.4030606715934768,
      "grad_norm": 0.11022622138261795,
      "learning_rate": 5.993612223952575e-05,
      "loss": 1.5883,
      "step": 204256
    },
    {
      "epoch": 0.40312381755487325,
      "grad_norm": 0.1104617491364479,
      "learning_rate": 5.9929782015161064e-05,
      "loss": 1.5858,
      "step": 204288
    },
    {
      "epoch": 0.4031869635162697,
      "grad_norm": 0.11066514253616333,
      "learning_rate": 5.992344179079638e-05,
      "loss": 1.5836,
      "step": 204320
    },
    {
      "epoch": 0.40325010947766615,
      "grad_norm": 0.11127970367670059,
      "learning_rate": 5.9917101566431686e-05,
      "loss": 1.5755,
      "step": 204352
    },
    {
      "epoch": 0.4033132554390626,
      "grad_norm": 0.11764411628246307,
      "learning_rate": 5.991076134206699e-05,
      "loss": 1.5856,
      "step": 204384
    },
    {
      "epoch": 0.40337640140045905,
      "grad_norm": 0.11639747023582458,
      "learning_rate": 5.99044211177023e-05,
      "loss": 1.5748,
      "step": 204416
    },
    {
      "epoch": 0.40343954736185544,
      "grad_norm": 0.10433321446180344,
      "learning_rate": 5.9898080893337616e-05,
      "loss": 1.5755,
      "step": 204448
    },
    {
      "epoch": 0.4035026933232519,
      "grad_norm": 0.11697738617658615,
      "learning_rate": 5.989174066897293e-05,
      "loss": 1.5729,
      "step": 204480
    },
    {
      "epoch": 0.40356583928464834,
      "grad_norm": 0.1136200875043869,
      "learning_rate": 5.988540044460824e-05,
      "loss": 1.5765,
      "step": 204512
    },
    {
      "epoch": 0.4036289852460448,
      "grad_norm": 0.12079986184835434,
      "learning_rate": 5.9879060220243547e-05,
      "loss": 1.5886,
      "step": 204544
    },
    {
      "epoch": 0.40369213120744124,
      "grad_norm": 0.10879378020763397,
      "learning_rate": 5.9872719995878854e-05,
      "loss": 1.5786,
      "step": 204576
    },
    {
      "epoch": 0.4037552771688377,
      "grad_norm": 0.11408454924821854,
      "learning_rate": 5.986637977151417e-05,
      "loss": 1.575,
      "step": 204608
    },
    {
      "epoch": 0.4038184231302341,
      "grad_norm": 0.11341025680303574,
      "learning_rate": 5.9860039547149483e-05,
      "loss": 1.5763,
      "step": 204640
    },
    {
      "epoch": 0.40388156909163053,
      "grad_norm": 0.10993221402168274,
      "learning_rate": 5.9853699322784784e-05,
      "loss": 1.5835,
      "step": 204672
    },
    {
      "epoch": 0.403944715053027,
      "grad_norm": 0.11652026325464249,
      "learning_rate": 5.98473590984201e-05,
      "loss": 1.5699,
      "step": 204704
    },
    {
      "epoch": 0.40400786101442343,
      "grad_norm": 0.10848741233348846,
      "learning_rate": 5.984101887405541e-05,
      "loss": 1.5737,
      "step": 204736
    },
    {
      "epoch": 0.4040710069758199,
      "grad_norm": 0.11425282806158066,
      "learning_rate": 5.983467864969072e-05,
      "loss": 1.5699,
      "step": 204768
    },
    {
      "epoch": 0.40413415293721633,
      "grad_norm": 0.10717558860778809,
      "learning_rate": 5.982833842532602e-05,
      "loss": 1.5696,
      "step": 204800
    },
    {
      "epoch": 0.4041972988986127,
      "grad_norm": 0.12416622787714005,
      "learning_rate": 5.982199820096134e-05,
      "loss": 1.571,
      "step": 204832
    },
    {
      "epoch": 0.40426044486000917,
      "grad_norm": 0.10836885124444962,
      "learning_rate": 5.981565797659665e-05,
      "loss": 1.5564,
      "step": 204864
    },
    {
      "epoch": 0.4043235908214056,
      "grad_norm": 0.11000969260931015,
      "learning_rate": 5.9809317752231966e-05,
      "loss": 1.5837,
      "step": 204896
    },
    {
      "epoch": 0.40438673678280207,
      "grad_norm": 0.11707895249128342,
      "learning_rate": 5.980297752786727e-05,
      "loss": 1.5673,
      "step": 204928
    },
    {
      "epoch": 0.4044498827441985,
      "grad_norm": 0.1101837307214737,
      "learning_rate": 5.979663730350258e-05,
      "loss": 1.573,
      "step": 204960
    },
    {
      "epoch": 0.40451302870559497,
      "grad_norm": 0.11137574911117554,
      "learning_rate": 5.979029707913789e-05,
      "loss": 1.5783,
      "step": 204992
    },
    {
      "epoch": 0.40457617466699136,
      "grad_norm": 0.10906076431274414,
      "learning_rate": 5.9783956854773204e-05,
      "loss": 1.5864,
      "step": 205024
    },
    {
      "epoch": 0.4046393206283878,
      "grad_norm": 0.11517034471035004,
      "learning_rate": 5.9777616630408505e-05,
      "loss": 1.581,
      "step": 205056
    },
    {
      "epoch": 0.40470246658978426,
      "grad_norm": 0.12411621958017349,
      "learning_rate": 5.977127640604382e-05,
      "loss": 1.5717,
      "step": 205088
    },
    {
      "epoch": 0.4047656125511807,
      "grad_norm": 0.11467491835355759,
      "learning_rate": 5.9764936181679134e-05,
      "loss": 1.5776,
      "step": 205120
    },
    {
      "epoch": 0.40482875851257716,
      "grad_norm": 0.11423167586326599,
      "learning_rate": 5.975859595731444e-05,
      "loss": 1.5794,
      "step": 205152
    },
    {
      "epoch": 0.4048919044739736,
      "grad_norm": 0.11539997905492783,
      "learning_rate": 5.9752255732949756e-05,
      "loss": 1.5782,
      "step": 205184
    },
    {
      "epoch": 0.40495505043537,
      "grad_norm": 0.1092941015958786,
      "learning_rate": 5.974591550858506e-05,
      "loss": 1.5667,
      "step": 205216
    },
    {
      "epoch": 0.40501819639676645,
      "grad_norm": 0.11151275783777237,
      "learning_rate": 5.973957528422037e-05,
      "loss": 1.5782,
      "step": 205248
    },
    {
      "epoch": 0.4050813423581629,
      "grad_norm": 0.10875088721513748,
      "learning_rate": 5.9733235059855687e-05,
      "loss": 1.5535,
      "step": 205280
    },
    {
      "epoch": 0.40514448831955935,
      "grad_norm": 0.11437881737947464,
      "learning_rate": 5.9726894835491e-05,
      "loss": 1.5796,
      "step": 205312
    },
    {
      "epoch": 0.4052076342809558,
      "grad_norm": 0.11793049424886703,
      "learning_rate": 5.97205546111263e-05,
      "loss": 1.5737,
      "step": 205344
    },
    {
      "epoch": 0.40527078024235225,
      "grad_norm": 0.10634225606918335,
      "learning_rate": 5.971421438676161e-05,
      "loss": 1.5723,
      "step": 205376
    },
    {
      "epoch": 0.40533392620374864,
      "grad_norm": 0.11165452748537064,
      "learning_rate": 5.9707874162396924e-05,
      "loss": 1.5781,
      "step": 205408
    },
    {
      "epoch": 0.4053970721651451,
      "grad_norm": 0.11905699223279953,
      "learning_rate": 5.970153393803224e-05,
      "loss": 1.5718,
      "step": 205440
    },
    {
      "epoch": 0.40546021812654154,
      "grad_norm": 0.11469445377588272,
      "learning_rate": 5.969519371366754e-05,
      "loss": 1.5696,
      "step": 205472
    },
    {
      "epoch": 0.405523364087938,
      "grad_norm": 0.11540370434522629,
      "learning_rate": 5.9688853489302855e-05,
      "loss": 1.5856,
      "step": 205504
    },
    {
      "epoch": 0.40558651004933444,
      "grad_norm": 0.10299599170684814,
      "learning_rate": 5.968251326493817e-05,
      "loss": 1.5727,
      "step": 205536
    },
    {
      "epoch": 0.4056496560107309,
      "grad_norm": 0.11487219482660294,
      "learning_rate": 5.967617304057348e-05,
      "loss": 1.5939,
      "step": 205568
    },
    {
      "epoch": 0.4057128019721273,
      "grad_norm": 0.11906348168849945,
      "learning_rate": 5.9669832816208785e-05,
      "loss": 1.5934,
      "step": 205600
    },
    {
      "epoch": 0.40577594793352373,
      "grad_norm": 0.10434137284755707,
      "learning_rate": 5.966349259184409e-05,
      "loss": 1.5803,
      "step": 205632
    },
    {
      "epoch": 0.4058390938949202,
      "grad_norm": 0.10675708949565887,
      "learning_rate": 5.965715236747941e-05,
      "loss": 1.5797,
      "step": 205664
    },
    {
      "epoch": 0.4059022398563166,
      "grad_norm": 0.1201130822300911,
      "learning_rate": 5.965081214311472e-05,
      "loss": 1.5773,
      "step": 205696
    },
    {
      "epoch": 0.4059653858177131,
      "grad_norm": 0.11443489789962769,
      "learning_rate": 5.964447191875002e-05,
      "loss": 1.5843,
      "step": 205728
    },
    {
      "epoch": 0.4060285317791095,
      "grad_norm": 0.1270044595003128,
      "learning_rate": 5.963813169438534e-05,
      "loss": 1.5832,
      "step": 205760
    },
    {
      "epoch": 0.406091677740506,
      "grad_norm": 0.11829876154661179,
      "learning_rate": 5.9631791470020645e-05,
      "loss": 1.5638,
      "step": 205792
    },
    {
      "epoch": 0.40615482370190237,
      "grad_norm": 0.1094130277633667,
      "learning_rate": 5.962545124565596e-05,
      "loss": 1.5852,
      "step": 205824
    },
    {
      "epoch": 0.4062179696632988,
      "grad_norm": 0.10748355090618134,
      "learning_rate": 5.9619111021291274e-05,
      "loss": 1.5676,
      "step": 205856
    },
    {
      "epoch": 0.40628111562469527,
      "grad_norm": 0.11317727714776993,
      "learning_rate": 5.9612770796926575e-05,
      "loss": 1.5662,
      "step": 205888
    },
    {
      "epoch": 0.4063442615860917,
      "grad_norm": 0.1065066009759903,
      "learning_rate": 5.960643057256189e-05,
      "loss": 1.571,
      "step": 205920
    },
    {
      "epoch": 0.40640740754748816,
      "grad_norm": 0.10893092304468155,
      "learning_rate": 5.9600090348197204e-05,
      "loss": 1.5706,
      "step": 205952
    },
    {
      "epoch": 0.4064705535088846,
      "grad_norm": 0.11343265324831009,
      "learning_rate": 5.959375012383251e-05,
      "loss": 1.5589,
      "step": 205984
    },
    {
      "epoch": 0.406533699470281,
      "grad_norm": 0.11336424946784973,
      "learning_rate": 5.958740989946782e-05,
      "loss": 1.5799,
      "step": 206016
    },
    {
      "epoch": 0.40659684543167746,
      "grad_norm": 0.10973262786865234,
      "learning_rate": 5.958106967510313e-05,
      "loss": 1.5755,
      "step": 206048
    },
    {
      "epoch": 0.4066599913930739,
      "grad_norm": 0.10886547714471817,
      "learning_rate": 5.957472945073844e-05,
      "loss": 1.5775,
      "step": 206080
    },
    {
      "epoch": 0.40672313735447035,
      "grad_norm": 0.11491454392671585,
      "learning_rate": 5.956838922637376e-05,
      "loss": 1.5713,
      "step": 206112
    },
    {
      "epoch": 0.4067862833158668,
      "grad_norm": 0.1137968972325325,
      "learning_rate": 5.956204900200906e-05,
      "loss": 1.5726,
      "step": 206144
    },
    {
      "epoch": 0.40684942927726325,
      "grad_norm": 0.1123242974281311,
      "learning_rate": 5.955570877764437e-05,
      "loss": 1.571,
      "step": 206176
    },
    {
      "epoch": 0.40691257523865965,
      "grad_norm": 0.11520496755838394,
      "learning_rate": 5.954936855327968e-05,
      "loss": 1.5909,
      "step": 206208
    },
    {
      "epoch": 0.4069757212000561,
      "grad_norm": 0.11327225714921951,
      "learning_rate": 5.9543028328914995e-05,
      "loss": 1.5996,
      "step": 206240
    },
    {
      "epoch": 0.40703886716145254,
      "grad_norm": 0.11600563675165176,
      "learning_rate": 5.9536688104550296e-05,
      "loss": 1.5643,
      "step": 206272
    },
    {
      "epoch": 0.407102013122849,
      "grad_norm": 0.11213836818933487,
      "learning_rate": 5.953034788018561e-05,
      "loss": 1.5694,
      "step": 206304
    },
    {
      "epoch": 0.40716515908424544,
      "grad_norm": 0.12320289760828018,
      "learning_rate": 5.9524007655820925e-05,
      "loss": 1.5738,
      "step": 206336
    },
    {
      "epoch": 0.4072283050456419,
      "grad_norm": 0.11507931351661682,
      "learning_rate": 5.951766743145624e-05,
      "loss": 1.5807,
      "step": 206368
    },
    {
      "epoch": 0.4072914510070383,
      "grad_norm": 0.11074212193489075,
      "learning_rate": 5.951132720709154e-05,
      "loss": 1.5629,
      "step": 206400
    },
    {
      "epoch": 0.40735459696843473,
      "grad_norm": 0.10792914777994156,
      "learning_rate": 5.950498698272685e-05,
      "loss": 1.5629,
      "step": 206432
    },
    {
      "epoch": 0.4074177429298312,
      "grad_norm": 0.1168747991323471,
      "learning_rate": 5.949864675836216e-05,
      "loss": 1.5687,
      "step": 206464
    },
    {
      "epoch": 0.40748088889122763,
      "grad_norm": 0.11281394958496094,
      "learning_rate": 5.949230653399748e-05,
      "loss": 1.5668,
      "step": 206496
    },
    {
      "epoch": 0.4075440348526241,
      "grad_norm": 0.10642575472593307,
      "learning_rate": 5.948596630963279e-05,
      "loss": 1.5818,
      "step": 206528
    },
    {
      "epoch": 0.40760718081402053,
      "grad_norm": 0.11196409910917282,
      "learning_rate": 5.947962608526809e-05,
      "loss": 1.5814,
      "step": 206560
    },
    {
      "epoch": 0.4076703267754169,
      "grad_norm": 0.11565840989351273,
      "learning_rate": 5.947328586090341e-05,
      "loss": 1.5831,
      "step": 206592
    },
    {
      "epoch": 0.4077334727368134,
      "grad_norm": 0.10333291441202164,
      "learning_rate": 5.9466945636538715e-05,
      "loss": 1.5775,
      "step": 206624
    },
    {
      "epoch": 0.4077966186982098,
      "grad_norm": 0.1070685088634491,
      "learning_rate": 5.946060541217403e-05,
      "loss": 1.5732,
      "step": 206656
    },
    {
      "epoch": 0.40785976465960627,
      "grad_norm": 0.11558514833450317,
      "learning_rate": 5.945426518780933e-05,
      "loss": 1.5785,
      "step": 206688
    },
    {
      "epoch": 0.4079229106210027,
      "grad_norm": 0.10835957527160645,
      "learning_rate": 5.9447924963444645e-05,
      "loss": 1.5648,
      "step": 206720
    },
    {
      "epoch": 0.40798605658239917,
      "grad_norm": 0.11364418268203735,
      "learning_rate": 5.944158473907996e-05,
      "loss": 1.5757,
      "step": 206752
    },
    {
      "epoch": 0.40804920254379556,
      "grad_norm": 0.12189406901597977,
      "learning_rate": 5.9435244514715275e-05,
      "loss": 1.5745,
      "step": 206784
    },
    {
      "epoch": 0.408112348505192,
      "grad_norm": 0.1169697642326355,
      "learning_rate": 5.9428904290350576e-05,
      "loss": 1.5685,
      "step": 206816
    },
    {
      "epoch": 0.40817549446658846,
      "grad_norm": 0.1133742704987526,
      "learning_rate": 5.942256406598588e-05,
      "loss": 1.5802,
      "step": 206848
    },
    {
      "epoch": 0.4082386404279849,
      "grad_norm": 0.10903415083885193,
      "learning_rate": 5.94162238416212e-05,
      "loss": 1.5596,
      "step": 206880
    },
    {
      "epoch": 0.40830178638938136,
      "grad_norm": 0.11916998028755188,
      "learning_rate": 5.940988361725651e-05,
      "loss": 1.5695,
      "step": 206912
    },
    {
      "epoch": 0.4083649323507778,
      "grad_norm": 0.11026991903781891,
      "learning_rate": 5.9403543392891813e-05,
      "loss": 1.5548,
      "step": 206944
    },
    {
      "epoch": 0.4084280783121742,
      "grad_norm": 0.10510311275720596,
      "learning_rate": 5.939720316852713e-05,
      "loss": 1.573,
      "step": 206976
    },
    {
      "epoch": 0.40849122427357065,
      "grad_norm": 0.11702536791563034,
      "learning_rate": 5.939086294416244e-05,
      "loss": 1.5641,
      "step": 207008
    },
    {
      "epoch": 0.4085543702349671,
      "grad_norm": 0.1176532730460167,
      "learning_rate": 5.938452271979775e-05,
      "loss": 1.5828,
      "step": 207040
    },
    {
      "epoch": 0.40861751619636355,
      "grad_norm": 0.11026246845722198,
      "learning_rate": 5.937818249543305e-05,
      "loss": 1.577,
      "step": 207072
    },
    {
      "epoch": 0.40868066215776,
      "grad_norm": 0.10813239961862564,
      "learning_rate": 5.9371842271068366e-05,
      "loss": 1.5833,
      "step": 207104
    },
    {
      "epoch": 0.40874380811915645,
      "grad_norm": 0.11306612193584442,
      "learning_rate": 5.936550204670368e-05,
      "loss": 1.5761,
      "step": 207136
    },
    {
      "epoch": 0.40880695408055284,
      "grad_norm": 0.10613807290792465,
      "learning_rate": 5.9359161822338995e-05,
      "loss": 1.5735,
      "step": 207168
    },
    {
      "epoch": 0.4088701000419493,
      "grad_norm": 0.1124691516160965,
      "learning_rate": 5.935282159797431e-05,
      "loss": 1.579,
      "step": 207200
    },
    {
      "epoch": 0.40893324600334574,
      "grad_norm": 0.1268845945596695,
      "learning_rate": 5.934648137360961e-05,
      "loss": 1.5866,
      "step": 207232
    },
    {
      "epoch": 0.4089963919647422,
      "grad_norm": 0.10868234932422638,
      "learning_rate": 5.934014114924492e-05,
      "loss": 1.5733,
      "step": 207264
    },
    {
      "epoch": 0.40905953792613864,
      "grad_norm": 0.10653870552778244,
      "learning_rate": 5.933380092488023e-05,
      "loss": 1.5721,
      "step": 207296
    },
    {
      "epoch": 0.4091226838875351,
      "grad_norm": 0.1217452734708786,
      "learning_rate": 5.932746070051555e-05,
      "loss": 1.5902,
      "step": 207328
    },
    {
      "epoch": 0.40918582984893154,
      "grad_norm": 0.10850151628255844,
      "learning_rate": 5.932112047615085e-05,
      "loss": 1.5777,
      "step": 207360
    },
    {
      "epoch": 0.40924897581032793,
      "grad_norm": 0.11223257333040237,
      "learning_rate": 5.931478025178616e-05,
      "loss": 1.5711,
      "step": 207392
    },
    {
      "epoch": 0.4093121217717244,
      "grad_norm": 0.11708072572946548,
      "learning_rate": 5.930844002742148e-05,
      "loss": 1.5755,
      "step": 207424
    },
    {
      "epoch": 0.40937526773312083,
      "grad_norm": 0.12039277702569962,
      "learning_rate": 5.9302099803056785e-05,
      "loss": 1.5776,
      "step": 207456
    },
    {
      "epoch": 0.4094384136945173,
      "grad_norm": 0.11246675252914429,
      "learning_rate": 5.9295759578692086e-05,
      "loss": 1.5727,
      "step": 207488
    },
    {
      "epoch": 0.4095015596559137,
      "grad_norm": 0.11134649813175201,
      "learning_rate": 5.92894193543274e-05,
      "loss": 1.5809,
      "step": 207520
    },
    {
      "epoch": 0.4095647056173102,
      "grad_norm": 0.11862438917160034,
      "learning_rate": 5.9283079129962716e-05,
      "loss": 1.5921,
      "step": 207552
    },
    {
      "epoch": 0.40962785157870657,
      "grad_norm": 0.11024701595306396,
      "learning_rate": 5.927673890559803e-05,
      "loss": 1.5748,
      "step": 207584
    },
    {
      "epoch": 0.409690997540103,
      "grad_norm": 0.1241542175412178,
      "learning_rate": 5.927039868123333e-05,
      "loss": 1.5823,
      "step": 207616
    },
    {
      "epoch": 0.40975414350149947,
      "grad_norm": 0.11487919837236404,
      "learning_rate": 5.9264058456868646e-05,
      "loss": 1.5594,
      "step": 207648
    },
    {
      "epoch": 0.4098172894628959,
      "grad_norm": 0.11190209537744522,
      "learning_rate": 5.9257718232503954e-05,
      "loss": 1.5626,
      "step": 207680
    },
    {
      "epoch": 0.40988043542429237,
      "grad_norm": 0.1269836276769638,
      "learning_rate": 5.925137800813927e-05,
      "loss": 1.5773,
      "step": 207712
    },
    {
      "epoch": 0.4099435813856888,
      "grad_norm": 0.11494969576597214,
      "learning_rate": 5.924503778377457e-05,
      "loss": 1.581,
      "step": 207744
    },
    {
      "epoch": 0.4100067273470852,
      "grad_norm": 0.10899503529071808,
      "learning_rate": 5.9238697559409884e-05,
      "loss": 1.5652,
      "step": 207776
    },
    {
      "epoch": 0.41006987330848166,
      "grad_norm": 0.1093950867652893,
      "learning_rate": 5.92323573350452e-05,
      "loss": 1.5775,
      "step": 207808
    },
    {
      "epoch": 0.4101330192698781,
      "grad_norm": 0.1146049052476883,
      "learning_rate": 5.922601711068051e-05,
      "loss": 1.5834,
      "step": 207840
    },
    {
      "epoch": 0.41019616523127456,
      "grad_norm": 0.10726942867040634,
      "learning_rate": 5.921967688631582e-05,
      "loss": 1.5782,
      "step": 207872
    },
    {
      "epoch": 0.410259311192671,
      "grad_norm": 0.11150894314050674,
      "learning_rate": 5.921333666195112e-05,
      "loss": 1.5709,
      "step": 207904
    },
    {
      "epoch": 0.41032245715406745,
      "grad_norm": 0.12000482529401779,
      "learning_rate": 5.9206996437586436e-05,
      "loss": 1.5766,
      "step": 207936
    },
    {
      "epoch": 0.41038560311546385,
      "grad_norm": 0.11264950037002563,
      "learning_rate": 5.920065621322175e-05,
      "loss": 1.5626,
      "step": 207968
    },
    {
      "epoch": 0.4104487490768603,
      "grad_norm": 0.1066565290093422,
      "learning_rate": 5.9194315988857065e-05,
      "loss": 1.5745,
      "step": 208000
    },
    {
      "epoch": 0.41051189503825675,
      "grad_norm": 0.10454881191253662,
      "learning_rate": 5.9187975764492366e-05,
      "loss": 1.577,
      "step": 208032
    },
    {
      "epoch": 0.4105750409996532,
      "grad_norm": 0.1147308200597763,
      "learning_rate": 5.918163554012768e-05,
      "loss": 1.5574,
      "step": 208064
    },
    {
      "epoch": 0.41063818696104964,
      "grad_norm": 0.11452020704746246,
      "learning_rate": 5.917529531576299e-05,
      "loss": 1.5672,
      "step": 208096
    },
    {
      "epoch": 0.4107013329224461,
      "grad_norm": 0.12036538124084473,
      "learning_rate": 5.91689550913983e-05,
      "loss": 1.5739,
      "step": 208128
    },
    {
      "epoch": 0.4107644788838425,
      "grad_norm": 0.11893625557422638,
      "learning_rate": 5.9162614867033604e-05,
      "loss": 1.5759,
      "step": 208160
    },
    {
      "epoch": 0.41082762484523894,
      "grad_norm": 0.10815265029668808,
      "learning_rate": 5.915627464266892e-05,
      "loss": 1.5693,
      "step": 208192
    },
    {
      "epoch": 0.4108907708066354,
      "grad_norm": 0.11385013163089752,
      "learning_rate": 5.914993441830423e-05,
      "loss": 1.5845,
      "step": 208224
    },
    {
      "epoch": 0.41095391676803184,
      "grad_norm": 0.11224812269210815,
      "learning_rate": 5.914359419393954e-05,
      "loss": 1.5697,
      "step": 208256
    },
    {
      "epoch": 0.4110170627294283,
      "grad_norm": 0.11066325008869171,
      "learning_rate": 5.913725396957485e-05,
      "loss": 1.5756,
      "step": 208288
    },
    {
      "epoch": 0.41108020869082473,
      "grad_norm": 0.11634143441915512,
      "learning_rate": 5.913091374521016e-05,
      "loss": 1.5866,
      "step": 208320
    },
    {
      "epoch": 0.4111433546522211,
      "grad_norm": 0.10700631141662598,
      "learning_rate": 5.912457352084547e-05,
      "loss": 1.5725,
      "step": 208352
    },
    {
      "epoch": 0.4112065006136176,
      "grad_norm": 0.11063332855701447,
      "learning_rate": 5.9118233296480786e-05,
      "loss": 1.582,
      "step": 208384
    },
    {
      "epoch": 0.411269646575014,
      "grad_norm": 0.11440924555063248,
      "learning_rate": 5.911189307211609e-05,
      "loss": 1.5804,
      "step": 208416
    },
    {
      "epoch": 0.4113327925364105,
      "grad_norm": 0.11152104288339615,
      "learning_rate": 5.91055528477514e-05,
      "loss": 1.5719,
      "step": 208448
    },
    {
      "epoch": 0.4113959384978069,
      "grad_norm": 0.10765744745731354,
      "learning_rate": 5.9099212623386716e-05,
      "loss": 1.5747,
      "step": 208480
    },
    {
      "epoch": 0.4114590844592034,
      "grad_norm": 0.11994259059429169,
      "learning_rate": 5.9092872399022024e-05,
      "loss": 1.5839,
      "step": 208512
    },
    {
      "epoch": 0.41152223042059977,
      "grad_norm": 0.11146427690982819,
      "learning_rate": 5.908653217465734e-05,
      "loss": 1.5714,
      "step": 208544
    },
    {
      "epoch": 0.4115853763819962,
      "grad_norm": 0.11730946600437164,
      "learning_rate": 5.908019195029264e-05,
      "loss": 1.5753,
      "step": 208576
    },
    {
      "epoch": 0.41164852234339266,
      "grad_norm": 0.11481603980064392,
      "learning_rate": 5.9073851725927954e-05,
      "loss": 1.5837,
      "step": 208608
    },
    {
      "epoch": 0.4117116683047891,
      "grad_norm": 0.10346874594688416,
      "learning_rate": 5.906751150156327e-05,
      "loss": 1.5672,
      "step": 208640
    },
    {
      "epoch": 0.41177481426618556,
      "grad_norm": 0.11606725305318832,
      "learning_rate": 5.9061171277198576e-05,
      "loss": 1.5687,
      "step": 208672
    },
    {
      "epoch": 0.411837960227582,
      "grad_norm": 0.11690467596054077,
      "learning_rate": 5.9054831052833884e-05,
      "loss": 1.5747,
      "step": 208704
    },
    {
      "epoch": 0.4119011061889784,
      "grad_norm": 0.11952147632837296,
      "learning_rate": 5.904849082846919e-05,
      "loss": 1.559,
      "step": 208736
    },
    {
      "epoch": 0.41196425215037485,
      "grad_norm": 0.10767176002264023,
      "learning_rate": 5.9042150604104506e-05,
      "loss": 1.5723,
      "step": 208768
    },
    {
      "epoch": 0.4120273981117713,
      "grad_norm": 0.11477220058441162,
      "learning_rate": 5.903581037973982e-05,
      "loss": 1.5767,
      "step": 208800
    },
    {
      "epoch": 0.41209054407316775,
      "grad_norm": 0.1156458705663681,
      "learning_rate": 5.902947015537512e-05,
      "loss": 1.5881,
      "step": 208832
    },
    {
      "epoch": 0.4121536900345642,
      "grad_norm": 0.11916563659906387,
      "learning_rate": 5.9023129931010436e-05,
      "loss": 1.5851,
      "step": 208864
    },
    {
      "epoch": 0.41221683599596065,
      "grad_norm": 0.11724422872066498,
      "learning_rate": 5.9016789706645744e-05,
      "loss": 1.5702,
      "step": 208896
    },
    {
      "epoch": 0.41227998195735704,
      "grad_norm": 0.11063949018716812,
      "learning_rate": 5.901044948228106e-05,
      "loss": 1.5661,
      "step": 208928
    },
    {
      "epoch": 0.4123431279187535,
      "grad_norm": 0.11034804582595825,
      "learning_rate": 5.900410925791636e-05,
      "loss": 1.5789,
      "step": 208960
    },
    {
      "epoch": 0.41240627388014994,
      "grad_norm": 0.11543915420770645,
      "learning_rate": 5.8997769033551674e-05,
      "loss": 1.5768,
      "step": 208992
    },
    {
      "epoch": 0.4124694198415464,
      "grad_norm": 0.11726246029138565,
      "learning_rate": 5.899142880918699e-05,
      "loss": 1.5754,
      "step": 209024
    },
    {
      "epoch": 0.41253256580294284,
      "grad_norm": 0.12239725887775421,
      "learning_rate": 5.8985088584822304e-05,
      "loss": 1.5755,
      "step": 209056
    },
    {
      "epoch": 0.4125957117643393,
      "grad_norm": 0.11276612430810928,
      "learning_rate": 5.8978748360457605e-05,
      "loss": 1.5709,
      "step": 209088
    },
    {
      "epoch": 0.41265885772573574,
      "grad_norm": 0.11039990931749344,
      "learning_rate": 5.897240813609292e-05,
      "loss": 1.5782,
      "step": 209120
    },
    {
      "epoch": 0.41272200368713213,
      "grad_norm": 0.12096954137086868,
      "learning_rate": 5.896606791172823e-05,
      "loss": 1.5711,
      "step": 209152
    },
    {
      "epoch": 0.4127851496485286,
      "grad_norm": 0.10877064615488052,
      "learning_rate": 5.895972768736354e-05,
      "loss": 1.5918,
      "step": 209184
    },
    {
      "epoch": 0.41284829560992503,
      "grad_norm": 0.11142287403345108,
      "learning_rate": 5.8953387462998856e-05,
      "loss": 1.5739,
      "step": 209216
    },
    {
      "epoch": 0.4129114415713215,
      "grad_norm": 0.10661587119102478,
      "learning_rate": 5.894704723863416e-05,
      "loss": 1.5662,
      "step": 209248
    },
    {
      "epoch": 0.41297458753271793,
      "grad_norm": 0.11213047802448273,
      "learning_rate": 5.894070701426947e-05,
      "loss": 1.5721,
      "step": 209280
    },
    {
      "epoch": 0.4130377334941144,
      "grad_norm": 0.1153879463672638,
      "learning_rate": 5.893436678990478e-05,
      "loss": 1.5727,
      "step": 209312
    },
    {
      "epoch": 0.41310087945551077,
      "grad_norm": 0.11843599379062653,
      "learning_rate": 5.8928026565540094e-05,
      "loss": 1.5742,
      "step": 209344
    },
    {
      "epoch": 0.4131640254169072,
      "grad_norm": 0.10982224345207214,
      "learning_rate": 5.8921686341175395e-05,
      "loss": 1.5739,
      "step": 209376
    },
    {
      "epoch": 0.41322717137830367,
      "grad_norm": 0.10911291837692261,
      "learning_rate": 5.891534611681071e-05,
      "loss": 1.5661,
      "step": 209408
    },
    {
      "epoch": 0.4132903173397001,
      "grad_norm": 0.10778553783893585,
      "learning_rate": 5.8909005892446024e-05,
      "loss": 1.5761,
      "step": 209440
    },
    {
      "epoch": 0.41335346330109657,
      "grad_norm": 0.1131877675652504,
      "learning_rate": 5.890266566808134e-05,
      "loss": 1.579,
      "step": 209472
    },
    {
      "epoch": 0.413416609262493,
      "grad_norm": 0.11634394526481628,
      "learning_rate": 5.889632544371664e-05,
      "loss": 1.573,
      "step": 209504
    },
    {
      "epoch": 0.4134797552238894,
      "grad_norm": 0.10641346126794815,
      "learning_rate": 5.888998521935195e-05,
      "loss": 1.5679,
      "step": 209536
    },
    {
      "epoch": 0.41354290118528586,
      "grad_norm": 0.11223439872264862,
      "learning_rate": 5.888364499498726e-05,
      "loss": 1.5733,
      "step": 209568
    },
    {
      "epoch": 0.4136060471466823,
      "grad_norm": 0.1170460656285286,
      "learning_rate": 5.8877304770622577e-05,
      "loss": 1.5669,
      "step": 209600
    },
    {
      "epoch": 0.41366919310807876,
      "grad_norm": 0.10966886579990387,
      "learning_rate": 5.887096454625788e-05,
      "loss": 1.5727,
      "step": 209632
    },
    {
      "epoch": 0.4137323390694752,
      "grad_norm": 0.11373679339885712,
      "learning_rate": 5.886462432189319e-05,
      "loss": 1.5856,
      "step": 209664
    },
    {
      "epoch": 0.41379548503087166,
      "grad_norm": 0.11283385008573532,
      "learning_rate": 5.885828409752851e-05,
      "loss": 1.5671,
      "step": 209696
    },
    {
      "epoch": 0.41385863099226805,
      "grad_norm": 0.1171746626496315,
      "learning_rate": 5.8851943873163814e-05,
      "loss": 1.5687,
      "step": 209728
    },
    {
      "epoch": 0.4139217769536645,
      "grad_norm": 0.10731907933950424,
      "learning_rate": 5.884560364879913e-05,
      "loss": 1.5654,
      "step": 209760
    },
    {
      "epoch": 0.41398492291506095,
      "grad_norm": 0.11476921290159225,
      "learning_rate": 5.883926342443443e-05,
      "loss": 1.5884,
      "step": 209792
    },
    {
      "epoch": 0.4140480688764574,
      "grad_norm": 0.10662003606557846,
      "learning_rate": 5.8832923200069745e-05,
      "loss": 1.5797,
      "step": 209824
    },
    {
      "epoch": 0.41411121483785385,
      "grad_norm": 0.1128525584936142,
      "learning_rate": 5.882658297570506e-05,
      "loss": 1.5782,
      "step": 209856
    },
    {
      "epoch": 0.4141743607992503,
      "grad_norm": 0.11338004469871521,
      "learning_rate": 5.8820242751340374e-05,
      "loss": 1.5969,
      "step": 209888
    },
    {
      "epoch": 0.4142375067606467,
      "grad_norm": 0.11333412677049637,
      "learning_rate": 5.8813902526975675e-05,
      "loss": 1.5711,
      "step": 209920
    },
    {
      "epoch": 0.41430065272204314,
      "grad_norm": 0.11807472258806229,
      "learning_rate": 5.880756230261098e-05,
      "loss": 1.5805,
      "step": 209952
    },
    {
      "epoch": 0.4143637986834396,
      "grad_norm": 0.1206955760717392,
      "learning_rate": 5.88012220782463e-05,
      "loss": 1.5732,
      "step": 209984
    },
    {
      "epoch": 0.41442694464483604,
      "grad_norm": 0.10973561555147171,
      "learning_rate": 5.879488185388161e-05,
      "loss": 1.5799,
      "step": 210016
    },
    {
      "epoch": 0.4144900906062325,
      "grad_norm": 0.11385183781385422,
      "learning_rate": 5.878854162951691e-05,
      "loss": 1.5707,
      "step": 210048
    },
    {
      "epoch": 0.41455323656762894,
      "grad_norm": 0.1270441859960556,
      "learning_rate": 5.878220140515223e-05,
      "loss": 1.574,
      "step": 210080
    },
    {
      "epoch": 0.41461638252902533,
      "grad_norm": 0.11708077043294907,
      "learning_rate": 5.877586118078754e-05,
      "loss": 1.576,
      "step": 210112
    },
    {
      "epoch": 0.4146795284904218,
      "grad_norm": 0.10859125107526779,
      "learning_rate": 5.876952095642285e-05,
      "loss": 1.5623,
      "step": 210144
    },
    {
      "epoch": 0.4147426744518182,
      "grad_norm": 0.11469090729951859,
      "learning_rate": 5.876318073205816e-05,
      "loss": 1.5896,
      "step": 210176
    },
    {
      "epoch": 0.4148058204132147,
      "grad_norm": 0.12022719532251358,
      "learning_rate": 5.8756840507693465e-05,
      "loss": 1.5685,
      "step": 210208
    },
    {
      "epoch": 0.4148689663746111,
      "grad_norm": 0.11533736437559128,
      "learning_rate": 5.875050028332878e-05,
      "loss": 1.571,
      "step": 210240
    },
    {
      "epoch": 0.4149321123360076,
      "grad_norm": 0.11421668529510498,
      "learning_rate": 5.8744160058964094e-05,
      "loss": 1.5799,
      "step": 210272
    },
    {
      "epoch": 0.41499525829740397,
      "grad_norm": 0.10821504890918732,
      "learning_rate": 5.8737819834599395e-05,
      "loss": 1.5884,
      "step": 210304
    },
    {
      "epoch": 0.4150584042588004,
      "grad_norm": 0.11552296578884125,
      "learning_rate": 5.873147961023471e-05,
      "loss": 1.5645,
      "step": 210336
    },
    {
      "epoch": 0.41512155022019687,
      "grad_norm": 0.11359403282403946,
      "learning_rate": 5.872513938587002e-05,
      "loss": 1.5773,
      "step": 210368
    },
    {
      "epoch": 0.4151846961815933,
      "grad_norm": 0.11828986555337906,
      "learning_rate": 5.871879916150533e-05,
      "loss": 1.5743,
      "step": 210400
    },
    {
      "epoch": 0.41524784214298976,
      "grad_norm": 0.11255080252885818,
      "learning_rate": 5.871245893714065e-05,
      "loss": 1.582,
      "step": 210432
    },
    {
      "epoch": 0.4153109881043862,
      "grad_norm": 0.111783467233181,
      "learning_rate": 5.870611871277595e-05,
      "loss": 1.5702,
      "step": 210464
    },
    {
      "epoch": 0.4153741340657826,
      "grad_norm": 0.12021353840827942,
      "learning_rate": 5.869977848841126e-05,
      "loss": 1.5721,
      "step": 210496
    },
    {
      "epoch": 0.41543728002717906,
      "grad_norm": 0.11928527802228928,
      "learning_rate": 5.869343826404658e-05,
      "loss": 1.5638,
      "step": 210528
    },
    {
      "epoch": 0.4155004259885755,
      "grad_norm": 0.11085432767868042,
      "learning_rate": 5.8687098039681885e-05,
      "loss": 1.5623,
      "step": 210560
    },
    {
      "epoch": 0.41556357194997195,
      "grad_norm": 0.10697130113840103,
      "learning_rate": 5.8680757815317186e-05,
      "loss": 1.5673,
      "step": 210592
    },
    {
      "epoch": 0.4156267179113684,
      "grad_norm": 0.10721738636493683,
      "learning_rate": 5.86744175909525e-05,
      "loss": 1.5802,
      "step": 210624
    },
    {
      "epoch": 0.41568986387276485,
      "grad_norm": 0.11581702530384064,
      "learning_rate": 5.8668077366587815e-05,
      "loss": 1.5853,
      "step": 210656
    },
    {
      "epoch": 0.41575300983416125,
      "grad_norm": 0.11685071140527725,
      "learning_rate": 5.866173714222313e-05,
      "loss": 1.573,
      "step": 210688
    },
    {
      "epoch": 0.4158161557955577,
      "grad_norm": 0.12459532171487808,
      "learning_rate": 5.865539691785843e-05,
      "loss": 1.5747,
      "step": 210720
    },
    {
      "epoch": 0.41587930175695415,
      "grad_norm": 0.11517032235860825,
      "learning_rate": 5.8649056693493745e-05,
      "loss": 1.5769,
      "step": 210752
    },
    {
      "epoch": 0.4159424477183506,
      "grad_norm": 0.11076007783412933,
      "learning_rate": 5.864271646912905e-05,
      "loss": 1.5768,
      "step": 210784
    },
    {
      "epoch": 0.41600559367974704,
      "grad_norm": 0.11767956614494324,
      "learning_rate": 5.863637624476437e-05,
      "loss": 1.5665,
      "step": 210816
    },
    {
      "epoch": 0.4160687396411435,
      "grad_norm": 0.11118356883525848,
      "learning_rate": 5.863003602039967e-05,
      "loss": 1.5862,
      "step": 210848
    },
    {
      "epoch": 0.41613188560253994,
      "grad_norm": 0.11311709135770798,
      "learning_rate": 5.862369579603498e-05,
      "loss": 1.5821,
      "step": 210880
    },
    {
      "epoch": 0.41619503156393634,
      "grad_norm": 0.11672328412532806,
      "learning_rate": 5.86173555716703e-05,
      "loss": 1.5687,
      "step": 210912
    },
    {
      "epoch": 0.4162581775253328,
      "grad_norm": 0.11094388365745544,
      "learning_rate": 5.861101534730561e-05,
      "loss": 1.5781,
      "step": 210944
    },
    {
      "epoch": 0.41632132348672923,
      "grad_norm": 0.11598347127437592,
      "learning_rate": 5.860467512294091e-05,
      "loss": 1.5817,
      "step": 210976
    },
    {
      "epoch": 0.4163844694481257,
      "grad_norm": 0.1184142604470253,
      "learning_rate": 5.859833489857622e-05,
      "loss": 1.5699,
      "step": 211008
    },
    {
      "epoch": 0.41644761540952213,
      "grad_norm": 0.11470816284418106,
      "learning_rate": 5.8591994674211535e-05,
      "loss": 1.5742,
      "step": 211040
    },
    {
      "epoch": 0.4165107613709186,
      "grad_norm": 0.11247041076421738,
      "learning_rate": 5.858565444984685e-05,
      "loss": 1.5899,
      "step": 211072
    },
    {
      "epoch": 0.416573907332315,
      "grad_norm": 0.10590722411870956,
      "learning_rate": 5.8579314225482164e-05,
      "loss": 1.5901,
      "step": 211104
    },
    {
      "epoch": 0.4166370532937114,
      "grad_norm": 0.10712088644504547,
      "learning_rate": 5.8572974001117465e-05,
      "loss": 1.5613,
      "step": 211136
    },
    {
      "epoch": 0.4167001992551079,
      "grad_norm": 0.11482211202383041,
      "learning_rate": 5.856663377675278e-05,
      "loss": 1.5741,
      "step": 211168
    },
    {
      "epoch": 0.4167633452165043,
      "grad_norm": 0.11243411898612976,
      "learning_rate": 5.856029355238809e-05,
      "loss": 1.5825,
      "step": 211200
    },
    {
      "epoch": 0.41682649117790077,
      "grad_norm": 0.11756385862827301,
      "learning_rate": 5.85539533280234e-05,
      "loss": 1.5844,
      "step": 211232
    },
    {
      "epoch": 0.4168896371392972,
      "grad_norm": 0.12397975474596024,
      "learning_rate": 5.8547613103658703e-05,
      "loss": 1.5693,
      "step": 211264
    },
    {
      "epoch": 0.4169527831006936,
      "grad_norm": 0.12299605458974838,
      "learning_rate": 5.854127287929402e-05,
      "loss": 1.5804,
      "step": 211296
    },
    {
      "epoch": 0.41701592906209006,
      "grad_norm": 0.11917003244161606,
      "learning_rate": 5.853493265492933e-05,
      "loss": 1.5757,
      "step": 211328
    },
    {
      "epoch": 0.4170790750234865,
      "grad_norm": 0.10854147374629974,
      "learning_rate": 5.852859243056465e-05,
      "loss": 1.5834,
      "step": 211360
    },
    {
      "epoch": 0.41714222098488296,
      "grad_norm": 0.11584977805614471,
      "learning_rate": 5.852225220619995e-05,
      "loss": 1.5885,
      "step": 211392
    },
    {
      "epoch": 0.4172053669462794,
      "grad_norm": 0.11360142379999161,
      "learning_rate": 5.8515911981835256e-05,
      "loss": 1.5766,
      "step": 211424
    },
    {
      "epoch": 0.41726851290767586,
      "grad_norm": 0.11507585644721985,
      "learning_rate": 5.850957175747057e-05,
      "loss": 1.5673,
      "step": 211456
    },
    {
      "epoch": 0.41733165886907225,
      "grad_norm": 0.1230563074350357,
      "learning_rate": 5.8503231533105885e-05,
      "loss": 1.5702,
      "step": 211488
    },
    {
      "epoch": 0.4173948048304687,
      "grad_norm": 0.10865850746631622,
      "learning_rate": 5.8496891308741186e-05,
      "loss": 1.5718,
      "step": 211520
    },
    {
      "epoch": 0.41745795079186515,
      "grad_norm": 0.1129273921251297,
      "learning_rate": 5.84905510843765e-05,
      "loss": 1.5635,
      "step": 211552
    },
    {
      "epoch": 0.4175210967532616,
      "grad_norm": 0.10906870663166046,
      "learning_rate": 5.8484210860011815e-05,
      "loss": 1.5698,
      "step": 211584
    },
    {
      "epoch": 0.41758424271465805,
      "grad_norm": 0.11242571473121643,
      "learning_rate": 5.847787063564712e-05,
      "loss": 1.564,
      "step": 211616
    },
    {
      "epoch": 0.4176473886760545,
      "grad_norm": 0.11419042944908142,
      "learning_rate": 5.8471530411282424e-05,
      "loss": 1.5768,
      "step": 211648
    },
    {
      "epoch": 0.4177105346374509,
      "grad_norm": 0.11575905233621597,
      "learning_rate": 5.846519018691774e-05,
      "loss": 1.5718,
      "step": 211680
    },
    {
      "epoch": 0.41777368059884734,
      "grad_norm": 0.10888403654098511,
      "learning_rate": 5.845884996255305e-05,
      "loss": 1.5788,
      "step": 211712
    },
    {
      "epoch": 0.4178368265602438,
      "grad_norm": 0.11863566190004349,
      "learning_rate": 5.845250973818837e-05,
      "loss": 1.5726,
      "step": 211744
    },
    {
      "epoch": 0.41789997252164024,
      "grad_norm": 0.12143582105636597,
      "learning_rate": 5.8446169513823675e-05,
      "loss": 1.5803,
      "step": 211776
    },
    {
      "epoch": 0.4179631184830367,
      "grad_norm": 0.11661967635154724,
      "learning_rate": 5.843982928945898e-05,
      "loss": 1.57,
      "step": 211808
    },
    {
      "epoch": 0.41802626444443314,
      "grad_norm": 0.12119210511445999,
      "learning_rate": 5.843348906509429e-05,
      "loss": 1.5678,
      "step": 211840
    },
    {
      "epoch": 0.41808941040582953,
      "grad_norm": 0.10788056999444962,
      "learning_rate": 5.8427148840729606e-05,
      "loss": 1.5778,
      "step": 211872
    },
    {
      "epoch": 0.418152556367226,
      "grad_norm": 0.12212945520877838,
      "learning_rate": 5.842080861636492e-05,
      "loss": 1.5855,
      "step": 211904
    },
    {
      "epoch": 0.41821570232862243,
      "grad_norm": 0.11086183786392212,
      "learning_rate": 5.841446839200022e-05,
      "loss": 1.5835,
      "step": 211936
    },
    {
      "epoch": 0.4182788482900189,
      "grad_norm": 0.11380373686552048,
      "learning_rate": 5.8408128167635536e-05,
      "loss": 1.5733,
      "step": 211968
    },
    {
      "epoch": 0.41834199425141533,
      "grad_norm": 0.12136074900627136,
      "learning_rate": 5.840178794327085e-05,
      "loss": 1.5837,
      "step": 212000
    },
    {
      "epoch": 0.4184051402128118,
      "grad_norm": 0.11972972005605698,
      "learning_rate": 5.839544771890616e-05,
      "loss": 1.5803,
      "step": 212032
    },
    {
      "epoch": 0.41846828617420817,
      "grad_norm": 0.12095468491315842,
      "learning_rate": 5.838910749454146e-05,
      "loss": 1.5969,
      "step": 212064
    },
    {
      "epoch": 0.4185314321356046,
      "grad_norm": 0.11016728729009628,
      "learning_rate": 5.8382767270176774e-05,
      "loss": 1.5783,
      "step": 212096
    },
    {
      "epoch": 0.41859457809700107,
      "grad_norm": 0.11938399076461792,
      "learning_rate": 5.837642704581209e-05,
      "loss": 1.585,
      "step": 212128
    },
    {
      "epoch": 0.4186577240583975,
      "grad_norm": 0.11566765606403351,
      "learning_rate": 5.83700868214474e-05,
      "loss": 1.5644,
      "step": 212160
    },
    {
      "epoch": 0.41872087001979397,
      "grad_norm": 0.11348102241754532,
      "learning_rate": 5.8363746597082704e-05,
      "loss": 1.5819,
      "step": 212192
    },
    {
      "epoch": 0.4187840159811904,
      "grad_norm": 0.11203183978796005,
      "learning_rate": 5.835740637271802e-05,
      "loss": 1.5709,
      "step": 212224
    },
    {
      "epoch": 0.4188471619425868,
      "grad_norm": 0.11854086071252823,
      "learning_rate": 5.8351066148353326e-05,
      "loss": 1.5731,
      "step": 212256
    },
    {
      "epoch": 0.41891030790398326,
      "grad_norm": 0.11306841671466827,
      "learning_rate": 5.834472592398864e-05,
      "loss": 1.5692,
      "step": 212288
    },
    {
      "epoch": 0.4189734538653797,
      "grad_norm": 0.109061598777771,
      "learning_rate": 5.833838569962394e-05,
      "loss": 1.5798,
      "step": 212320
    },
    {
      "epoch": 0.41903659982677616,
      "grad_norm": 0.11595797538757324,
      "learning_rate": 5.8332045475259256e-05,
      "loss": 1.5812,
      "step": 212352
    },
    {
      "epoch": 0.4190997457881726,
      "grad_norm": 0.11549539864063263,
      "learning_rate": 5.832570525089457e-05,
      "loss": 1.5844,
      "step": 212384
    },
    {
      "epoch": 0.41916289174956906,
      "grad_norm": 0.11134572327136993,
      "learning_rate": 5.831936502652988e-05,
      "loss": 1.5708,
      "step": 212416
    },
    {
      "epoch": 0.4192260377109655,
      "grad_norm": 0.11597440391778946,
      "learning_rate": 5.831302480216519e-05,
      "loss": 1.5806,
      "step": 212448
    },
    {
      "epoch": 0.4192891836723619,
      "grad_norm": 0.11357767879962921,
      "learning_rate": 5.8306684577800494e-05,
      "loss": 1.5736,
      "step": 212480
    },
    {
      "epoch": 0.41935232963375835,
      "grad_norm": 0.11139217019081116,
      "learning_rate": 5.830034435343581e-05,
      "loss": 1.577,
      "step": 212512
    },
    {
      "epoch": 0.4194154755951548,
      "grad_norm": 0.11596512049436569,
      "learning_rate": 5.829400412907112e-05,
      "loss": 1.5669,
      "step": 212544
    },
    {
      "epoch": 0.41947862155655125,
      "grad_norm": 0.11166244000196457,
      "learning_rate": 5.828766390470644e-05,
      "loss": 1.5691,
      "step": 212576
    },
    {
      "epoch": 0.4195417675179477,
      "grad_norm": 0.10859309136867523,
      "learning_rate": 5.828132368034174e-05,
      "loss": 1.5687,
      "step": 212608
    },
    {
      "epoch": 0.41960491347934414,
      "grad_norm": 0.11687473952770233,
      "learning_rate": 5.8274983455977053e-05,
      "loss": 1.5849,
      "step": 212640
    },
    {
      "epoch": 0.41966805944074054,
      "grad_norm": 0.11086977273225784,
      "learning_rate": 5.826864323161236e-05,
      "loss": 1.5768,
      "step": 212672
    },
    {
      "epoch": 0.419731205402137,
      "grad_norm": 0.1037912368774414,
      "learning_rate": 5.8262303007247676e-05,
      "loss": 1.5645,
      "step": 212704
    },
    {
      "epoch": 0.41979435136353344,
      "grad_norm": 0.13082370162010193,
      "learning_rate": 5.825596278288298e-05,
      "loss": 1.5722,
      "step": 212736
    },
    {
      "epoch": 0.4198574973249299,
      "grad_norm": 0.11623410135507584,
      "learning_rate": 5.824962255851829e-05,
      "loss": 1.5809,
      "step": 212768
    },
    {
      "epoch": 0.41992064328632633,
      "grad_norm": 0.11510179936885834,
      "learning_rate": 5.8243282334153606e-05,
      "loss": 1.5852,
      "step": 212800
    },
    {
      "epoch": 0.4199837892477228,
      "grad_norm": 0.11736845225095749,
      "learning_rate": 5.8236942109788914e-05,
      "loss": 1.5779,
      "step": 212832
    },
    {
      "epoch": 0.4200469352091192,
      "grad_norm": 0.10826859623193741,
      "learning_rate": 5.823060188542422e-05,
      "loss": 1.5655,
      "step": 212864
    },
    {
      "epoch": 0.4201100811705156,
      "grad_norm": 0.10915494710206985,
      "learning_rate": 5.822426166105953e-05,
      "loss": 1.575,
      "step": 212896
    },
    {
      "epoch": 0.4201732271319121,
      "grad_norm": 0.11843578517436981,
      "learning_rate": 5.8217921436694844e-05,
      "loss": 1.5774,
      "step": 212928
    },
    {
      "epoch": 0.4202363730933085,
      "grad_norm": 0.10358286648988724,
      "learning_rate": 5.821158121233016e-05,
      "loss": 1.566,
      "step": 212960
    },
    {
      "epoch": 0.420299519054705,
      "grad_norm": 0.11201329529285431,
      "learning_rate": 5.820524098796546e-05,
      "loss": 1.5822,
      "step": 212992
    },
    {
      "epoch": 0.4203626650161014,
      "grad_norm": 0.11048104614019394,
      "learning_rate": 5.8198900763600774e-05,
      "loss": 1.568,
      "step": 213024
    },
    {
      "epoch": 0.4204258109774978,
      "grad_norm": 0.11986801028251648,
      "learning_rate": 5.819256053923608e-05,
      "loss": 1.5719,
      "step": 213056
    },
    {
      "epoch": 0.42048895693889426,
      "grad_norm": 0.10660749673843384,
      "learning_rate": 5.8186220314871396e-05,
      "loss": 1.5737,
      "step": 213088
    },
    {
      "epoch": 0.4205521029002907,
      "grad_norm": 0.11020977795124054,
      "learning_rate": 5.817988009050671e-05,
      "loss": 1.5593,
      "step": 213120
    },
    {
      "epoch": 0.42061524886168716,
      "grad_norm": 0.13178670406341553,
      "learning_rate": 5.817353986614201e-05,
      "loss": 1.5729,
      "step": 213152
    },
    {
      "epoch": 0.4206783948230836,
      "grad_norm": 0.11084149777889252,
      "learning_rate": 5.8167199641777326e-05,
      "loss": 1.5542,
      "step": 213184
    },
    {
      "epoch": 0.42074154078448006,
      "grad_norm": 0.11329176276922226,
      "learning_rate": 5.816085941741264e-05,
      "loss": 1.5791,
      "step": 213216
    },
    {
      "epoch": 0.42080468674587646,
      "grad_norm": 0.11462894082069397,
      "learning_rate": 5.815451919304795e-05,
      "loss": 1.5838,
      "step": 213248
    },
    {
      "epoch": 0.4208678327072729,
      "grad_norm": 0.11463163793087006,
      "learning_rate": 5.8148178968683257e-05,
      "loss": 1.5662,
      "step": 213280
    },
    {
      "epoch": 0.42093097866866935,
      "grad_norm": 0.11173997074365616,
      "learning_rate": 5.8141838744318564e-05,
      "loss": 1.5656,
      "step": 213312
    },
    {
      "epoch": 0.4209941246300658,
      "grad_norm": 0.1229645386338234,
      "learning_rate": 5.813549851995388e-05,
      "loss": 1.5708,
      "step": 213344
    },
    {
      "epoch": 0.42105727059146225,
      "grad_norm": 0.11651979386806488,
      "learning_rate": 5.8129158295589193e-05,
      "loss": 1.5783,
      "step": 213376
    },
    {
      "epoch": 0.4211204165528587,
      "grad_norm": 0.11476512998342514,
      "learning_rate": 5.8122818071224494e-05,
      "loss": 1.5689,
      "step": 213408
    },
    {
      "epoch": 0.4211835625142551,
      "grad_norm": 0.11378980427980423,
      "learning_rate": 5.811647784685981e-05,
      "loss": 1.5692,
      "step": 213440
    },
    {
      "epoch": 0.42124670847565154,
      "grad_norm": 0.1159764975309372,
      "learning_rate": 5.811013762249512e-05,
      "loss": 1.5731,
      "step": 213472
    },
    {
      "epoch": 0.421309854437048,
      "grad_norm": 0.11891423910856247,
      "learning_rate": 5.810379739813043e-05,
      "loss": 1.5789,
      "step": 213504
    },
    {
      "epoch": 0.42137300039844444,
      "grad_norm": 0.11131199449300766,
      "learning_rate": 5.809745717376573e-05,
      "loss": 1.5839,
      "step": 213536
    },
    {
      "epoch": 0.4214361463598409,
      "grad_norm": 0.11010655015707016,
      "learning_rate": 5.809111694940105e-05,
      "loss": 1.5775,
      "step": 213568
    },
    {
      "epoch": 0.42149929232123734,
      "grad_norm": 0.11034880578517914,
      "learning_rate": 5.808477672503636e-05,
      "loss": 1.5776,
      "step": 213600
    },
    {
      "epoch": 0.42156243828263373,
      "grad_norm": 0.11275182664394379,
      "learning_rate": 5.8078436500671676e-05,
      "loss": 1.5695,
      "step": 213632
    },
    {
      "epoch": 0.4216255842440302,
      "grad_norm": 0.11139434576034546,
      "learning_rate": 5.807209627630698e-05,
      "loss": 1.5716,
      "step": 213664
    },
    {
      "epoch": 0.42168873020542663,
      "grad_norm": 0.11640458554029465,
      "learning_rate": 5.806575605194229e-05,
      "loss": 1.5729,
      "step": 213696
    },
    {
      "epoch": 0.4217518761668231,
      "grad_norm": 0.11134900152683258,
      "learning_rate": 5.80594158275776e-05,
      "loss": 1.57,
      "step": 213728
    },
    {
      "epoch": 0.42181502212821953,
      "grad_norm": 0.11026263236999512,
      "learning_rate": 5.8053075603212914e-05,
      "loss": 1.5736,
      "step": 213760
    },
    {
      "epoch": 0.421878168089616,
      "grad_norm": 0.11456513404846191,
      "learning_rate": 5.804673537884823e-05,
      "loss": 1.5746,
      "step": 213792
    },
    {
      "epoch": 0.4219413140510124,
      "grad_norm": 0.11510647088289261,
      "learning_rate": 5.804039515448353e-05,
      "loss": 1.565,
      "step": 213824
    },
    {
      "epoch": 0.4220044600124088,
      "grad_norm": 0.11348488181829453,
      "learning_rate": 5.8034054930118844e-05,
      "loss": 1.5789,
      "step": 213856
    },
    {
      "epoch": 0.42206760597380527,
      "grad_norm": 0.11129774898290634,
      "learning_rate": 5.802771470575415e-05,
      "loss": 1.5756,
      "step": 213888
    },
    {
      "epoch": 0.4221307519352017,
      "grad_norm": 0.10939408093690872,
      "learning_rate": 5.8021374481389467e-05,
      "loss": 1.5751,
      "step": 213920
    },
    {
      "epoch": 0.42219389789659817,
      "grad_norm": 0.11541156470775604,
      "learning_rate": 5.801503425702477e-05,
      "loss": 1.579,
      "step": 213952
    },
    {
      "epoch": 0.4222570438579946,
      "grad_norm": 0.11404352635145187,
      "learning_rate": 5.800869403266008e-05,
      "loss": 1.5834,
      "step": 213984
    },
    {
      "epoch": 0.422320189819391,
      "grad_norm": 0.10567382723093033,
      "learning_rate": 5.80023538082954e-05,
      "loss": 1.5848,
      "step": 214016
    },
    {
      "epoch": 0.42238333578078746,
      "grad_norm": 0.1112348660826683,
      "learning_rate": 5.799601358393071e-05,
      "loss": 1.5696,
      "step": 214048
    },
    {
      "epoch": 0.4224464817421839,
      "grad_norm": 0.11303667724132538,
      "learning_rate": 5.798967335956601e-05,
      "loss": 1.572,
      "step": 214080
    },
    {
      "epoch": 0.42250962770358036,
      "grad_norm": 0.11266178637742996,
      "learning_rate": 5.798333313520132e-05,
      "loss": 1.5786,
      "step": 214112
    },
    {
      "epoch": 0.4225727736649768,
      "grad_norm": 0.111946240067482,
      "learning_rate": 5.7976992910836635e-05,
      "loss": 1.5714,
      "step": 214144
    },
    {
      "epoch": 0.42263591962637326,
      "grad_norm": 0.11566639691591263,
      "learning_rate": 5.797065268647195e-05,
      "loss": 1.5765,
      "step": 214176
    },
    {
      "epoch": 0.4226990655877697,
      "grad_norm": 0.10516596585512161,
      "learning_rate": 5.796431246210725e-05,
      "loss": 1.5638,
      "step": 214208
    },
    {
      "epoch": 0.4227622115491661,
      "grad_norm": 0.11059646308422089,
      "learning_rate": 5.7957972237742565e-05,
      "loss": 1.5635,
      "step": 214240
    },
    {
      "epoch": 0.42282535751056255,
      "grad_norm": 0.12049796432256699,
      "learning_rate": 5.795163201337788e-05,
      "loss": 1.5678,
      "step": 214272
    },
    {
      "epoch": 0.422888503471959,
      "grad_norm": 0.11232086271047592,
      "learning_rate": 5.794529178901319e-05,
      "loss": 1.5747,
      "step": 214304
    },
    {
      "epoch": 0.42295164943335545,
      "grad_norm": 0.10908524692058563,
      "learning_rate": 5.7938951564648495e-05,
      "loss": 1.5734,
      "step": 214336
    },
    {
      "epoch": 0.4230147953947519,
      "grad_norm": 0.10831624269485474,
      "learning_rate": 5.79326113402838e-05,
      "loss": 1.5613,
      "step": 214368
    },
    {
      "epoch": 0.42307794135614835,
      "grad_norm": 0.12005595862865448,
      "learning_rate": 5.792627111591912e-05,
      "loss": 1.5769,
      "step": 214400
    },
    {
      "epoch": 0.42314108731754474,
      "grad_norm": 0.10607882589101791,
      "learning_rate": 5.791993089155443e-05,
      "loss": 1.566,
      "step": 214432
    },
    {
      "epoch": 0.4232042332789412,
      "grad_norm": 0.11583924293518066,
      "learning_rate": 5.7913590667189746e-05,
      "loss": 1.5665,
      "step": 214464
    },
    {
      "epoch": 0.42326737924033764,
      "grad_norm": 0.11221151053905487,
      "learning_rate": 5.790725044282505e-05,
      "loss": 1.5691,
      "step": 214496
    },
    {
      "epoch": 0.4233305252017341,
      "grad_norm": 0.10671308636665344,
      "learning_rate": 5.7900910218460355e-05,
      "loss": 1.5757,
      "step": 214528
    },
    {
      "epoch": 0.42339367116313054,
      "grad_norm": 0.11174337565898895,
      "learning_rate": 5.789456999409567e-05,
      "loss": 1.576,
      "step": 214560
    },
    {
      "epoch": 0.423456817124527,
      "grad_norm": 0.10919161885976791,
      "learning_rate": 5.7888229769730984e-05,
      "loss": 1.5788,
      "step": 214592
    },
    {
      "epoch": 0.4235199630859234,
      "grad_norm": 0.11323666572570801,
      "learning_rate": 5.7881889545366285e-05,
      "loss": 1.5663,
      "step": 214624
    },
    {
      "epoch": 0.42358310904731983,
      "grad_norm": 0.10981086641550064,
      "learning_rate": 5.78755493210016e-05,
      "loss": 1.5776,
      "step": 214656
    },
    {
      "epoch": 0.4236462550087163,
      "grad_norm": 0.12076502293348312,
      "learning_rate": 5.7869209096636914e-05,
      "loss": 1.5684,
      "step": 214688
    },
    {
      "epoch": 0.4237094009701127,
      "grad_norm": 0.11037950217723846,
      "learning_rate": 5.786286887227222e-05,
      "loss": 1.5704,
      "step": 214720
    },
    {
      "epoch": 0.4237725469315092,
      "grad_norm": 0.10478291660547256,
      "learning_rate": 5.785652864790752e-05,
      "loss": 1.5835,
      "step": 214752
    },
    {
      "epoch": 0.4238356928929056,
      "grad_norm": 0.11317232996225357,
      "learning_rate": 5.785018842354284e-05,
      "loss": 1.5653,
      "step": 214784
    },
    {
      "epoch": 0.423898838854302,
      "grad_norm": 0.1120942011475563,
      "learning_rate": 5.784384819917815e-05,
      "loss": 1.5616,
      "step": 214816
    },
    {
      "epoch": 0.42396198481569847,
      "grad_norm": 0.11217595636844635,
      "learning_rate": 5.783750797481347e-05,
      "loss": 1.5706,
      "step": 214848
    },
    {
      "epoch": 0.4240251307770949,
      "grad_norm": 0.10993542522192001,
      "learning_rate": 5.783116775044877e-05,
      "loss": 1.5764,
      "step": 214880
    },
    {
      "epoch": 0.42408827673849137,
      "grad_norm": 0.1143922507762909,
      "learning_rate": 5.782482752608408e-05,
      "loss": 1.5797,
      "step": 214912
    },
    {
      "epoch": 0.4241514226998878,
      "grad_norm": 0.11156270653009415,
      "learning_rate": 5.781848730171939e-05,
      "loss": 1.5682,
      "step": 214944
    },
    {
      "epoch": 0.42421456866128426,
      "grad_norm": 0.1151362732052803,
      "learning_rate": 5.7812147077354705e-05,
      "loss": 1.5718,
      "step": 214976
    },
    {
      "epoch": 0.42427771462268066,
      "grad_norm": 0.13502606749534607,
      "learning_rate": 5.7805806852990006e-05,
      "loss": 1.576,
      "step": 215008
    },
    {
      "epoch": 0.4243408605840771,
      "grad_norm": 0.11473239958286285,
      "learning_rate": 5.779946662862532e-05,
      "loss": 1.5754,
      "step": 215040
    },
    {
      "epoch": 0.42440400654547356,
      "grad_norm": 0.11435826867818832,
      "learning_rate": 5.7793126404260635e-05,
      "loss": 1.5791,
      "step": 215072
    },
    {
      "epoch": 0.42446715250687,
      "grad_norm": 0.11932343244552612,
      "learning_rate": 5.778678617989595e-05,
      "loss": 1.5833,
      "step": 215104
    },
    {
      "epoch": 0.42453029846826645,
      "grad_norm": 0.10759299248456955,
      "learning_rate": 5.778044595553126e-05,
      "loss": 1.5845,
      "step": 215136
    },
    {
      "epoch": 0.4245934444296629,
      "grad_norm": 0.11628827452659607,
      "learning_rate": 5.777410573116656e-05,
      "loss": 1.5732,
      "step": 215168
    },
    {
      "epoch": 0.4246565903910593,
      "grad_norm": 0.10682403296232224,
      "learning_rate": 5.776776550680187e-05,
      "loss": 1.5689,
      "step": 215200
    },
    {
      "epoch": 0.42471973635245575,
      "grad_norm": 0.1116858646273613,
      "learning_rate": 5.776142528243719e-05,
      "loss": 1.5724,
      "step": 215232
    },
    {
      "epoch": 0.4247828823138522,
      "grad_norm": 0.11500170081853867,
      "learning_rate": 5.77550850580725e-05,
      "loss": 1.5661,
      "step": 215264
    },
    {
      "epoch": 0.42484602827524864,
      "grad_norm": 0.1105748862028122,
      "learning_rate": 5.77487448337078e-05,
      "loss": 1.5663,
      "step": 215296
    },
    {
      "epoch": 0.4249091742366451,
      "grad_norm": 0.12328539043664932,
      "learning_rate": 5.774240460934312e-05,
      "loss": 1.5587,
      "step": 215328
    },
    {
      "epoch": 0.42497232019804154,
      "grad_norm": 0.11484191566705704,
      "learning_rate": 5.7736064384978425e-05,
      "loss": 1.5744,
      "step": 215360
    },
    {
      "epoch": 0.42503546615943794,
      "grad_norm": 0.1185460016131401,
      "learning_rate": 5.772972416061374e-05,
      "loss": 1.5848,
      "step": 215392
    },
    {
      "epoch": 0.4250986121208344,
      "grad_norm": 0.10631705820560455,
      "learning_rate": 5.772338393624904e-05,
      "loss": 1.5652,
      "step": 215424
    },
    {
      "epoch": 0.42516175808223083,
      "grad_norm": 0.1144087016582489,
      "learning_rate": 5.7717043711884355e-05,
      "loss": 1.5786,
      "step": 215456
    },
    {
      "epoch": 0.4252249040436273,
      "grad_norm": 0.1128242239356041,
      "learning_rate": 5.771070348751967e-05,
      "loss": 1.5704,
      "step": 215488
    },
    {
      "epoch": 0.42528805000502373,
      "grad_norm": 0.10929341614246368,
      "learning_rate": 5.7704363263154985e-05,
      "loss": 1.56,
      "step": 215520
    },
    {
      "epoch": 0.4253511959664202,
      "grad_norm": 0.11630631238222122,
      "learning_rate": 5.7698023038790286e-05,
      "loss": 1.5629,
      "step": 215552
    },
    {
      "epoch": 0.4254143419278166,
      "grad_norm": 0.11064088344573975,
      "learning_rate": 5.769168281442559e-05,
      "loss": 1.5796,
      "step": 215584
    },
    {
      "epoch": 0.425477487889213,
      "grad_norm": 0.12014969438314438,
      "learning_rate": 5.768534259006091e-05,
      "loss": 1.5718,
      "step": 215616
    },
    {
      "epoch": 0.4255406338506095,
      "grad_norm": 0.11737433075904846,
      "learning_rate": 5.767900236569622e-05,
      "loss": 1.5749,
      "step": 215648
    },
    {
      "epoch": 0.4256037798120059,
      "grad_norm": 0.11125744879245758,
      "learning_rate": 5.7672662141331523e-05,
      "loss": 1.5777,
      "step": 215680
    },
    {
      "epoch": 0.42566692577340237,
      "grad_norm": 0.11548735201358795,
      "learning_rate": 5.766632191696684e-05,
      "loss": 1.5867,
      "step": 215712
    },
    {
      "epoch": 0.4257300717347988,
      "grad_norm": 0.11033716052770615,
      "learning_rate": 5.765998169260215e-05,
      "loss": 1.5676,
      "step": 215744
    },
    {
      "epoch": 0.42579321769619527,
      "grad_norm": 0.11361586302518845,
      "learning_rate": 5.765364146823746e-05,
      "loss": 1.5622,
      "step": 215776
    },
    {
      "epoch": 0.42585636365759166,
      "grad_norm": 0.11350362002849579,
      "learning_rate": 5.7647301243872775e-05,
      "loss": 1.5681,
      "step": 215808
    },
    {
      "epoch": 0.4259195096189881,
      "grad_norm": 0.11969402432441711,
      "learning_rate": 5.7640961019508076e-05,
      "loss": 1.579,
      "step": 215840
    },
    {
      "epoch": 0.42598265558038456,
      "grad_norm": 0.11505340039730072,
      "learning_rate": 5.763462079514339e-05,
      "loss": 1.5593,
      "step": 215872
    },
    {
      "epoch": 0.426045801541781,
      "grad_norm": 0.11213871836662292,
      "learning_rate": 5.7628280570778705e-05,
      "loss": 1.5783,
      "step": 215904
    },
    {
      "epoch": 0.42610894750317746,
      "grad_norm": 0.11875083297491074,
      "learning_rate": 5.762194034641401e-05,
      "loss": 1.5729,
      "step": 215936
    },
    {
      "epoch": 0.4261720934645739,
      "grad_norm": 0.10955949872732162,
      "learning_rate": 5.761560012204932e-05,
      "loss": 1.5589,
      "step": 215968
    },
    {
      "epoch": 0.4262352394259703,
      "grad_norm": 0.11102863401174545,
      "learning_rate": 5.760925989768463e-05,
      "loss": 1.5806,
      "step": 216000
    },
    {
      "epoch": 0.42629838538736675,
      "grad_norm": 0.10898289829492569,
      "learning_rate": 5.760291967331994e-05,
      "loss": 1.5577,
      "step": 216032
    },
    {
      "epoch": 0.4263615313487632,
      "grad_norm": 0.11119260638952255,
      "learning_rate": 5.759657944895526e-05,
      "loss": 1.5503,
      "step": 216064
    },
    {
      "epoch": 0.42642467731015965,
      "grad_norm": 0.1125345304608345,
      "learning_rate": 5.759023922459056e-05,
      "loss": 1.5698,
      "step": 216096
    },
    {
      "epoch": 0.4264878232715561,
      "grad_norm": 0.12304135411977768,
      "learning_rate": 5.758389900022587e-05,
      "loss": 1.5855,
      "step": 216128
    },
    {
      "epoch": 0.42655096923295255,
      "grad_norm": 0.10996831953525543,
      "learning_rate": 5.757755877586119e-05,
      "loss": 1.5827,
      "step": 216160
    },
    {
      "epoch": 0.42661411519434894,
      "grad_norm": 0.11370484530925751,
      "learning_rate": 5.7571218551496496e-05,
      "loss": 1.5817,
      "step": 216192
    },
    {
      "epoch": 0.4266772611557454,
      "grad_norm": 0.10931141674518585,
      "learning_rate": 5.7564878327131797e-05,
      "loss": 1.5655,
      "step": 216224
    },
    {
      "epoch": 0.42674040711714184,
      "grad_norm": 0.11213477700948715,
      "learning_rate": 5.755853810276711e-05,
      "loss": 1.5803,
      "step": 216256
    },
    {
      "epoch": 0.4268035530785383,
      "grad_norm": 0.11812477558851242,
      "learning_rate": 5.7552197878402426e-05,
      "loss": 1.5706,
      "step": 216288
    },
    {
      "epoch": 0.42686669903993474,
      "grad_norm": 0.12508949637413025,
      "learning_rate": 5.754585765403774e-05,
      "loss": 1.5781,
      "step": 216320
    },
    {
      "epoch": 0.4269298450013312,
      "grad_norm": 0.12047788500785828,
      "learning_rate": 5.753951742967304e-05,
      "loss": 1.5608,
      "step": 216352
    },
    {
      "epoch": 0.4269929909627276,
      "grad_norm": 0.11848469823598862,
      "learning_rate": 5.7533177205308356e-05,
      "loss": 1.5662,
      "step": 216384
    },
    {
      "epoch": 0.42705613692412403,
      "grad_norm": 0.1096002534031868,
      "learning_rate": 5.7526836980943664e-05,
      "loss": 1.5619,
      "step": 216416
    },
    {
      "epoch": 0.4271192828855205,
      "grad_norm": 0.11447635293006897,
      "learning_rate": 5.752049675657898e-05,
      "loss": 1.5887,
      "step": 216448
    },
    {
      "epoch": 0.42718242884691693,
      "grad_norm": 0.10712642967700958,
      "learning_rate": 5.751415653221429e-05,
      "loss": 1.5771,
      "step": 216480
    },
    {
      "epoch": 0.4272455748083134,
      "grad_norm": 0.112143874168396,
      "learning_rate": 5.7507816307849594e-05,
      "loss": 1.5499,
      "step": 216512
    },
    {
      "epoch": 0.4273087207697098,
      "grad_norm": 0.11540811508893967,
      "learning_rate": 5.750147608348491e-05,
      "loss": 1.5802,
      "step": 216544
    },
    {
      "epoch": 0.4273718667311062,
      "grad_norm": 0.1156095638871193,
      "learning_rate": 5.7495135859120216e-05,
      "loss": 1.5744,
      "step": 216576
    },
    {
      "epoch": 0.42743501269250267,
      "grad_norm": 0.10763511806726456,
      "learning_rate": 5.748879563475553e-05,
      "loss": 1.5668,
      "step": 216608
    },
    {
      "epoch": 0.4274981586538991,
      "grad_norm": 0.11237064003944397,
      "learning_rate": 5.748245541039083e-05,
      "loss": 1.5695,
      "step": 216640
    },
    {
      "epoch": 0.42756130461529557,
      "grad_norm": 0.1260167509317398,
      "learning_rate": 5.7476115186026146e-05,
      "loss": 1.577,
      "step": 216672
    },
    {
      "epoch": 0.427624450576692,
      "grad_norm": 0.120932936668396,
      "learning_rate": 5.746977496166146e-05,
      "loss": 1.5782,
      "step": 216704
    },
    {
      "epoch": 0.42768759653808847,
      "grad_norm": 0.11022268980741501,
      "learning_rate": 5.7463434737296775e-05,
      "loss": 1.5807,
      "step": 216736
    },
    {
      "epoch": 0.42775074249948486,
      "grad_norm": 0.11237247288227081,
      "learning_rate": 5.7457094512932076e-05,
      "loss": 1.5672,
      "step": 216768
    },
    {
      "epoch": 0.4278138884608813,
      "grad_norm": 0.10768884420394897,
      "learning_rate": 5.745075428856739e-05,
      "loss": 1.5711,
      "step": 216800
    },
    {
      "epoch": 0.42787703442227776,
      "grad_norm": 0.11572138965129852,
      "learning_rate": 5.74444140642027e-05,
      "loss": 1.5737,
      "step": 216832
    },
    {
      "epoch": 0.4279401803836742,
      "grad_norm": 0.11492817848920822,
      "learning_rate": 5.743807383983801e-05,
      "loss": 1.5657,
      "step": 216864
    },
    {
      "epoch": 0.42800332634507066,
      "grad_norm": 0.1095789223909378,
      "learning_rate": 5.7431733615473314e-05,
      "loss": 1.5735,
      "step": 216896
    },
    {
      "epoch": 0.4280664723064671,
      "grad_norm": 0.10906712710857391,
      "learning_rate": 5.742539339110863e-05,
      "loss": 1.5787,
      "step": 216928
    },
    {
      "epoch": 0.4281296182678635,
      "grad_norm": 0.10804124921560287,
      "learning_rate": 5.741905316674394e-05,
      "loss": 1.5786,
      "step": 216960
    },
    {
      "epoch": 0.42819276422925995,
      "grad_norm": 0.12142904847860336,
      "learning_rate": 5.741271294237925e-05,
      "loss": 1.5761,
      "step": 216992
    },
    {
      "epoch": 0.4282559101906564,
      "grad_norm": 0.11196298897266388,
      "learning_rate": 5.740637271801456e-05,
      "loss": 1.5704,
      "step": 217024
    },
    {
      "epoch": 0.42831905615205285,
      "grad_norm": 0.1128426119685173,
      "learning_rate": 5.740003249364987e-05,
      "loss": 1.5792,
      "step": 217056
    },
    {
      "epoch": 0.4283822021134493,
      "grad_norm": 0.12256114184856415,
      "learning_rate": 5.739369226928518e-05,
      "loss": 1.5865,
      "step": 217088
    },
    {
      "epoch": 0.42844534807484574,
      "grad_norm": 0.11635429412126541,
      "learning_rate": 5.7387352044920496e-05,
      "loss": 1.5817,
      "step": 217120
    },
    {
      "epoch": 0.42850849403624214,
      "grad_norm": 0.1104697734117508,
      "learning_rate": 5.738101182055581e-05,
      "loss": 1.5711,
      "step": 217152
    },
    {
      "epoch": 0.4285716399976386,
      "grad_norm": 0.111278235912323,
      "learning_rate": 5.737467159619111e-05,
      "loss": 1.5672,
      "step": 217184
    },
    {
      "epoch": 0.42863478595903504,
      "grad_norm": 0.10895587503910065,
      "learning_rate": 5.736833137182642e-05,
      "loss": 1.5732,
      "step": 217216
    },
    {
      "epoch": 0.4286979319204315,
      "grad_norm": 0.117510586977005,
      "learning_rate": 5.7361991147461734e-05,
      "loss": 1.5647,
      "step": 217248
    },
    {
      "epoch": 0.42876107788182793,
      "grad_norm": 0.10830023884773254,
      "learning_rate": 5.735565092309705e-05,
      "loss": 1.5885,
      "step": 217280
    },
    {
      "epoch": 0.4288242238432244,
      "grad_norm": 0.1161816194653511,
      "learning_rate": 5.734931069873235e-05,
      "loss": 1.563,
      "step": 217312
    },
    {
      "epoch": 0.4288873698046208,
      "grad_norm": 0.11264339834451675,
      "learning_rate": 5.7342970474367664e-05,
      "loss": 1.5744,
      "step": 217344
    },
    {
      "epoch": 0.4289505157660172,
      "grad_norm": 0.11148174107074738,
      "learning_rate": 5.733663025000298e-05,
      "loss": 1.5597,
      "step": 217376
    },
    {
      "epoch": 0.4290136617274137,
      "grad_norm": 0.11672309786081314,
      "learning_rate": 5.7330290025638286e-05,
      "loss": 1.5657,
      "step": 217408
    },
    {
      "epoch": 0.4290768076888101,
      "grad_norm": 0.1104683056473732,
      "learning_rate": 5.7323949801273594e-05,
      "loss": 1.5698,
      "step": 217440
    },
    {
      "epoch": 0.4291399536502066,
      "grad_norm": 0.109275221824646,
      "learning_rate": 5.73176095769089e-05,
      "loss": 1.5652,
      "step": 217472
    },
    {
      "epoch": 0.429203099611603,
      "grad_norm": 0.11759917438030243,
      "learning_rate": 5.7311269352544216e-05,
      "loss": 1.5744,
      "step": 217504
    },
    {
      "epoch": 0.42926624557299947,
      "grad_norm": 0.11129330098628998,
      "learning_rate": 5.730492912817953e-05,
      "loss": 1.5656,
      "step": 217536
    },
    {
      "epoch": 0.42932939153439587,
      "grad_norm": 0.11111228168010712,
      "learning_rate": 5.729858890381483e-05,
      "loss": 1.5607,
      "step": 217568
    },
    {
      "epoch": 0.4293925374957923,
      "grad_norm": 0.11939235031604767,
      "learning_rate": 5.7292248679450147e-05,
      "loss": 1.5801,
      "step": 217600
    },
    {
      "epoch": 0.42945568345718876,
      "grad_norm": 0.11422089487314224,
      "learning_rate": 5.7285908455085454e-05,
      "loss": 1.5748,
      "step": 217632
    },
    {
      "epoch": 0.4295188294185852,
      "grad_norm": 0.12160011380910873,
      "learning_rate": 5.727956823072077e-05,
      "loss": 1.572,
      "step": 217664
    },
    {
      "epoch": 0.42958197537998166,
      "grad_norm": 0.1145854964852333,
      "learning_rate": 5.727322800635607e-05,
      "loss": 1.5721,
      "step": 217696
    },
    {
      "epoch": 0.4296451213413781,
      "grad_norm": 0.1209818497300148,
      "learning_rate": 5.7266887781991384e-05,
      "loss": 1.5702,
      "step": 217728
    },
    {
      "epoch": 0.4297082673027745,
      "grad_norm": 0.11725013703107834,
      "learning_rate": 5.72605475576267e-05,
      "loss": 1.5752,
      "step": 217760
    },
    {
      "epoch": 0.42977141326417095,
      "grad_norm": 0.11543697118759155,
      "learning_rate": 5.7254207333262014e-05,
      "loss": 1.5685,
      "step": 217792
    },
    {
      "epoch": 0.4298345592255674,
      "grad_norm": 0.11653497815132141,
      "learning_rate": 5.724786710889732e-05,
      "loss": 1.5775,
      "step": 217824
    },
    {
      "epoch": 0.42989770518696385,
      "grad_norm": 0.11175195127725601,
      "learning_rate": 5.724152688453263e-05,
      "loss": 1.5761,
      "step": 217856
    },
    {
      "epoch": 0.4299608511483603,
      "grad_norm": 0.11420013755559921,
      "learning_rate": 5.723518666016794e-05,
      "loss": 1.5821,
      "step": 217888
    },
    {
      "epoch": 0.43002399710975675,
      "grad_norm": 0.12244796007871628,
      "learning_rate": 5.722884643580325e-05,
      "loss": 1.5757,
      "step": 217920
    },
    {
      "epoch": 0.43008714307115314,
      "grad_norm": 0.11154807358980179,
      "learning_rate": 5.7222506211438566e-05,
      "loss": 1.557,
      "step": 217952
    },
    {
      "epoch": 0.4301502890325496,
      "grad_norm": 0.11421848833560944,
      "learning_rate": 5.721616598707387e-05,
      "loss": 1.5815,
      "step": 217984
    },
    {
      "epoch": 0.43021343499394604,
      "grad_norm": 0.1296013742685318,
      "learning_rate": 5.720982576270918e-05,
      "loss": 1.5722,
      "step": 218016
    },
    {
      "epoch": 0.4302765809553425,
      "grad_norm": 0.12777550518512726,
      "learning_rate": 5.720348553834449e-05,
      "loss": 1.5817,
      "step": 218048
    },
    {
      "epoch": 0.43033972691673894,
      "grad_norm": 0.11427032202482224,
      "learning_rate": 5.7197145313979804e-05,
      "loss": 1.5782,
      "step": 218080
    },
    {
      "epoch": 0.4304028728781354,
      "grad_norm": 0.11090344190597534,
      "learning_rate": 5.7190805089615105e-05,
      "loss": 1.5622,
      "step": 218112
    },
    {
      "epoch": 0.4304660188395318,
      "grad_norm": 0.11822871118783951,
      "learning_rate": 5.718446486525042e-05,
      "loss": 1.5732,
      "step": 218144
    },
    {
      "epoch": 0.43052916480092823,
      "grad_norm": 0.11690302193164825,
      "learning_rate": 5.7178124640885734e-05,
      "loss": 1.5688,
      "step": 218176
    },
    {
      "epoch": 0.4305923107623247,
      "grad_norm": 0.10858042538166046,
      "learning_rate": 5.717178441652105e-05,
      "loss": 1.574,
      "step": 218208
    },
    {
      "epoch": 0.43065545672372113,
      "grad_norm": 0.1063372790813446,
      "learning_rate": 5.716544419215635e-05,
      "loss": 1.5689,
      "step": 218240
    },
    {
      "epoch": 0.4307186026851176,
      "grad_norm": 0.11257363110780716,
      "learning_rate": 5.715910396779166e-05,
      "loss": 1.5583,
      "step": 218272
    },
    {
      "epoch": 0.43078174864651403,
      "grad_norm": 0.10806793719530106,
      "learning_rate": 5.715276374342697e-05,
      "loss": 1.5576,
      "step": 218304
    },
    {
      "epoch": 0.4308448946079104,
      "grad_norm": 0.1094035729765892,
      "learning_rate": 5.7146423519062287e-05,
      "loss": 1.5715,
      "step": 218336
    },
    {
      "epoch": 0.43090804056930687,
      "grad_norm": 0.1087304875254631,
      "learning_rate": 5.714008329469759e-05,
      "loss": 1.5677,
      "step": 218368
    },
    {
      "epoch": 0.4309711865307033,
      "grad_norm": 0.11429969221353531,
      "learning_rate": 5.71337430703329e-05,
      "loss": 1.5745,
      "step": 218400
    },
    {
      "epoch": 0.43103433249209977,
      "grad_norm": 0.11125301569700241,
      "learning_rate": 5.712740284596822e-05,
      "loss": 1.5671,
      "step": 218432
    },
    {
      "epoch": 0.4310974784534962,
      "grad_norm": 0.11256994307041168,
      "learning_rate": 5.7121062621603525e-05,
      "loss": 1.5491,
      "step": 218464
    },
    {
      "epoch": 0.43116062441489267,
      "grad_norm": 0.11583176255226135,
      "learning_rate": 5.711472239723884e-05,
      "loss": 1.5792,
      "step": 218496
    },
    {
      "epoch": 0.43122377037628906,
      "grad_norm": 0.11268655955791473,
      "learning_rate": 5.710838217287414e-05,
      "loss": 1.5839,
      "step": 218528
    },
    {
      "epoch": 0.4312869163376855,
      "grad_norm": 0.1278001070022583,
      "learning_rate": 5.7102041948509455e-05,
      "loss": 1.5753,
      "step": 218560
    },
    {
      "epoch": 0.43135006229908196,
      "grad_norm": 0.11566419154405594,
      "learning_rate": 5.709570172414477e-05,
      "loss": 1.5705,
      "step": 218592
    },
    {
      "epoch": 0.4314132082604784,
      "grad_norm": 0.11827423423528671,
      "learning_rate": 5.7089361499780084e-05,
      "loss": 1.5763,
      "step": 218624
    },
    {
      "epoch": 0.43147635422187486,
      "grad_norm": 0.10856103152036667,
      "learning_rate": 5.7083021275415385e-05,
      "loss": 1.5671,
      "step": 218656
    },
    {
      "epoch": 0.4315395001832713,
      "grad_norm": 0.11640467494726181,
      "learning_rate": 5.707668105105069e-05,
      "loss": 1.5718,
      "step": 218688
    },
    {
      "epoch": 0.4316026461446677,
      "grad_norm": 0.12964989244937897,
      "learning_rate": 5.707034082668601e-05,
      "loss": 1.5773,
      "step": 218720
    },
    {
      "epoch": 0.43166579210606415,
      "grad_norm": 0.11477820575237274,
      "learning_rate": 5.706400060232132e-05,
      "loss": 1.5856,
      "step": 218752
    },
    {
      "epoch": 0.4317289380674606,
      "grad_norm": 0.11888802796602249,
      "learning_rate": 5.705766037795662e-05,
      "loss": 1.5871,
      "step": 218784
    },
    {
      "epoch": 0.43179208402885705,
      "grad_norm": 0.11207152903079987,
      "learning_rate": 5.705132015359194e-05,
      "loss": 1.5712,
      "step": 218816
    },
    {
      "epoch": 0.4318552299902535,
      "grad_norm": 0.11219317466020584,
      "learning_rate": 5.704497992922725e-05,
      "loss": 1.583,
      "step": 218848
    },
    {
      "epoch": 0.43191837595164995,
      "grad_norm": 0.10959234088659286,
      "learning_rate": 5.703863970486256e-05,
      "loss": 1.5733,
      "step": 218880
    },
    {
      "epoch": 0.43198152191304634,
      "grad_norm": 0.10953658074140549,
      "learning_rate": 5.703229948049786e-05,
      "loss": 1.5606,
      "step": 218912
    },
    {
      "epoch": 0.4320446678744428,
      "grad_norm": 0.11791291832923889,
      "learning_rate": 5.7025959256133175e-05,
      "loss": 1.5746,
      "step": 218944
    },
    {
      "epoch": 0.43210781383583924,
      "grad_norm": 0.12534630298614502,
      "learning_rate": 5.701961903176849e-05,
      "loss": 1.579,
      "step": 218976
    },
    {
      "epoch": 0.4321709597972357,
      "grad_norm": 0.119520403444767,
      "learning_rate": 5.7013278807403804e-05,
      "loss": 1.5841,
      "step": 219008
    },
    {
      "epoch": 0.43223410575863214,
      "grad_norm": 0.1143878623843193,
      "learning_rate": 5.7006938583039105e-05,
      "loss": 1.5856,
      "step": 219040
    },
    {
      "epoch": 0.4322972517200286,
      "grad_norm": 0.1126607283949852,
      "learning_rate": 5.700059835867442e-05,
      "loss": 1.5805,
      "step": 219072
    },
    {
      "epoch": 0.432360397681425,
      "grad_norm": 0.1200566217303276,
      "learning_rate": 5.699425813430973e-05,
      "loss": 1.5676,
      "step": 219104
    },
    {
      "epoch": 0.43242354364282143,
      "grad_norm": 0.10853078216314316,
      "learning_rate": 5.698791790994504e-05,
      "loss": 1.5833,
      "step": 219136
    },
    {
      "epoch": 0.4324866896042179,
      "grad_norm": 0.11355725675821304,
      "learning_rate": 5.698157768558036e-05,
      "loss": 1.5714,
      "step": 219168
    },
    {
      "epoch": 0.4325498355656143,
      "grad_norm": 0.10941006988286972,
      "learning_rate": 5.697523746121566e-05,
      "loss": 1.5691,
      "step": 219200
    },
    {
      "epoch": 0.4326129815270108,
      "grad_norm": 0.1205168291926384,
      "learning_rate": 5.696889723685097e-05,
      "loss": 1.5728,
      "step": 219232
    },
    {
      "epoch": 0.4326761274884072,
      "grad_norm": 0.1134066954255104,
      "learning_rate": 5.696255701248629e-05,
      "loss": 1.5725,
      "step": 219264
    },
    {
      "epoch": 0.4327392734498037,
      "grad_norm": 0.11044749617576599,
      "learning_rate": 5.6956216788121595e-05,
      "loss": 1.5726,
      "step": 219296
    },
    {
      "epoch": 0.43280241941120007,
      "grad_norm": 0.11439785361289978,
      "learning_rate": 5.6949876563756896e-05,
      "loss": 1.5813,
      "step": 219328
    },
    {
      "epoch": 0.4328655653725965,
      "grad_norm": 0.11201561987400055,
      "learning_rate": 5.694353633939221e-05,
      "loss": 1.5738,
      "step": 219360
    },
    {
      "epoch": 0.43292871133399297,
      "grad_norm": 0.1089145615696907,
      "learning_rate": 5.6937196115027525e-05,
      "loss": 1.5848,
      "step": 219392
    },
    {
      "epoch": 0.4329918572953894,
      "grad_norm": 0.10721936076879501,
      "learning_rate": 5.693085589066284e-05,
      "loss": 1.5625,
      "step": 219424
    },
    {
      "epoch": 0.43305500325678586,
      "grad_norm": 0.12367662787437439,
      "learning_rate": 5.692451566629814e-05,
      "loss": 1.57,
      "step": 219456
    },
    {
      "epoch": 0.4331181492181823,
      "grad_norm": 0.11392983049154282,
      "learning_rate": 5.6918175441933455e-05,
      "loss": 1.5743,
      "step": 219488
    },
    {
      "epoch": 0.4331812951795787,
      "grad_norm": 0.11306256800889969,
      "learning_rate": 5.691183521756876e-05,
      "loss": 1.5713,
      "step": 219520
    },
    {
      "epoch": 0.43324444114097516,
      "grad_norm": 0.1093650683760643,
      "learning_rate": 5.690549499320408e-05,
      "loss": 1.5758,
      "step": 219552
    },
    {
      "epoch": 0.4333075871023716,
      "grad_norm": 0.1123494878411293,
      "learning_rate": 5.689915476883938e-05,
      "loss": 1.5713,
      "step": 219584
    },
    {
      "epoch": 0.43337073306376805,
      "grad_norm": 0.1146850511431694,
      "learning_rate": 5.689281454447469e-05,
      "loss": 1.5835,
      "step": 219616
    },
    {
      "epoch": 0.4334338790251645,
      "grad_norm": 0.10450323671102524,
      "learning_rate": 5.688647432011001e-05,
      "loss": 1.5675,
      "step": 219648
    },
    {
      "epoch": 0.43349702498656095,
      "grad_norm": 0.10758733749389648,
      "learning_rate": 5.688013409574532e-05,
      "loss": 1.5722,
      "step": 219680
    },
    {
      "epoch": 0.43356017094795735,
      "grad_norm": 0.11030951887369156,
      "learning_rate": 5.687379387138063e-05,
      "loss": 1.5847,
      "step": 219712
    },
    {
      "epoch": 0.4336233169093538,
      "grad_norm": 0.11307337135076523,
      "learning_rate": 5.686745364701593e-05,
      "loss": 1.5769,
      "step": 219744
    },
    {
      "epoch": 0.43368646287075024,
      "grad_norm": 0.10753865540027618,
      "learning_rate": 5.6861113422651245e-05,
      "loss": 1.5717,
      "step": 219776
    },
    {
      "epoch": 0.4337496088321467,
      "grad_norm": 0.11488639563322067,
      "learning_rate": 5.685477319828656e-05,
      "loss": 1.5746,
      "step": 219808
    },
    {
      "epoch": 0.43381275479354314,
      "grad_norm": 0.11072096228599548,
      "learning_rate": 5.6848432973921875e-05,
      "loss": 1.5644,
      "step": 219840
    },
    {
      "epoch": 0.4338759007549396,
      "grad_norm": 0.11318612098693848,
      "learning_rate": 5.6842092749557176e-05,
      "loss": 1.5757,
      "step": 219872
    },
    {
      "epoch": 0.433939046716336,
      "grad_norm": 0.11152063310146332,
      "learning_rate": 5.683575252519249e-05,
      "loss": 1.582,
      "step": 219904
    },
    {
      "epoch": 0.43400219267773243,
      "grad_norm": 0.10941016674041748,
      "learning_rate": 5.68294123008278e-05,
      "loss": 1.5613,
      "step": 219936
    },
    {
      "epoch": 0.4340653386391289,
      "grad_norm": 0.1090482622385025,
      "learning_rate": 5.682307207646311e-05,
      "loss": 1.5801,
      "step": 219968
    },
    {
      "epoch": 0.43412848460052533,
      "grad_norm": 0.11290223151445389,
      "learning_rate": 5.6816731852098413e-05,
      "loss": 1.5793,
      "step": 220000
    },
    {
      "epoch": 0.4341916305619218,
      "grad_norm": 0.11280801147222519,
      "learning_rate": 5.681039162773373e-05,
      "loss": 1.5752,
      "step": 220032
    },
    {
      "epoch": 0.43425477652331823,
      "grad_norm": 0.1137833371758461,
      "learning_rate": 5.680405140336904e-05,
      "loss": 1.5715,
      "step": 220064
    },
    {
      "epoch": 0.4343179224847146,
      "grad_norm": 0.11288630217313766,
      "learning_rate": 5.679771117900435e-05,
      "loss": 1.5667,
      "step": 220096
    },
    {
      "epoch": 0.4343810684461111,
      "grad_norm": 0.1094576045870781,
      "learning_rate": 5.679137095463966e-05,
      "loss": 1.5807,
      "step": 220128
    },
    {
      "epoch": 0.4344442144075075,
      "grad_norm": 0.11282413452863693,
      "learning_rate": 5.6785030730274966e-05,
      "loss": 1.5823,
      "step": 220160
    },
    {
      "epoch": 0.43450736036890397,
      "grad_norm": 0.11341885477304459,
      "learning_rate": 5.677869050591028e-05,
      "loss": 1.5733,
      "step": 220192
    },
    {
      "epoch": 0.4345705063303004,
      "grad_norm": 0.11951928585767746,
      "learning_rate": 5.6772350281545595e-05,
      "loss": 1.573,
      "step": 220224
    },
    {
      "epoch": 0.43463365229169687,
      "grad_norm": 0.11937609314918518,
      "learning_rate": 5.6766010057180896e-05,
      "loss": 1.5716,
      "step": 220256
    },
    {
      "epoch": 0.43469679825309326,
      "grad_norm": 0.11358558386564255,
      "learning_rate": 5.675966983281621e-05,
      "loss": 1.5756,
      "step": 220288
    },
    {
      "epoch": 0.4347599442144897,
      "grad_norm": 0.11427044868469238,
      "learning_rate": 5.6753329608451525e-05,
      "loss": 1.5769,
      "step": 220320
    },
    {
      "epoch": 0.43482309017588616,
      "grad_norm": 0.1126299500465393,
      "learning_rate": 5.674698938408683e-05,
      "loss": 1.573,
      "step": 220352
    },
    {
      "epoch": 0.4348862361372826,
      "grad_norm": 0.12942196428775787,
      "learning_rate": 5.674064915972215e-05,
      "loss": 1.5747,
      "step": 220384
    },
    {
      "epoch": 0.43494938209867906,
      "grad_norm": 0.11733990907669067,
      "learning_rate": 5.673430893535745e-05,
      "loss": 1.5634,
      "step": 220416
    },
    {
      "epoch": 0.4350125280600755,
      "grad_norm": 0.1132250502705574,
      "learning_rate": 5.672796871099276e-05,
      "loss": 1.5759,
      "step": 220448
    },
    {
      "epoch": 0.4350756740214719,
      "grad_norm": 0.1102571189403534,
      "learning_rate": 5.672162848662808e-05,
      "loss": 1.5608,
      "step": 220480
    },
    {
      "epoch": 0.43513881998286835,
      "grad_norm": 0.11909134685993195,
      "learning_rate": 5.6715288262263385e-05,
      "loss": 1.5681,
      "step": 220512
    },
    {
      "epoch": 0.4352019659442648,
      "grad_norm": 0.10762350261211395,
      "learning_rate": 5.670894803789869e-05,
      "loss": 1.5777,
      "step": 220544
    },
    {
      "epoch": 0.43526511190566125,
      "grad_norm": 0.10702694952487946,
      "learning_rate": 5.6702607813534e-05,
      "loss": 1.5708,
      "step": 220576
    },
    {
      "epoch": 0.4353282578670577,
      "grad_norm": 0.11215417087078094,
      "learning_rate": 5.6696267589169316e-05,
      "loss": 1.5607,
      "step": 220608
    },
    {
      "epoch": 0.43539140382845415,
      "grad_norm": 0.11287208646535873,
      "learning_rate": 5.668992736480463e-05,
      "loss": 1.565,
      "step": 220640
    },
    {
      "epoch": 0.43545454978985054,
      "grad_norm": 0.11429022997617722,
      "learning_rate": 5.668358714043993e-05,
      "loss": 1.5766,
      "step": 220672
    },
    {
      "epoch": 0.435517695751247,
      "grad_norm": 0.1195773184299469,
      "learning_rate": 5.6677246916075246e-05,
      "loss": 1.5743,
      "step": 220704
    },
    {
      "epoch": 0.43558084171264344,
      "grad_norm": 0.11271505802869797,
      "learning_rate": 5.6670906691710554e-05,
      "loss": 1.5755,
      "step": 220736
    },
    {
      "epoch": 0.4356439876740399,
      "grad_norm": 0.11656443774700165,
      "learning_rate": 5.666456646734587e-05,
      "loss": 1.5646,
      "step": 220768
    },
    {
      "epoch": 0.43570713363543634,
      "grad_norm": 0.11756814271211624,
      "learning_rate": 5.665822624298117e-05,
      "loss": 1.5645,
      "step": 220800
    },
    {
      "epoch": 0.4357702795968328,
      "grad_norm": 0.11012377589941025,
      "learning_rate": 5.6651886018616484e-05,
      "loss": 1.5781,
      "step": 220832
    },
    {
      "epoch": 0.43583342555822924,
      "grad_norm": 0.11887282133102417,
      "learning_rate": 5.66455457942518e-05,
      "loss": 1.5688,
      "step": 220864
    },
    {
      "epoch": 0.43589657151962563,
      "grad_norm": 0.11284030973911285,
      "learning_rate": 5.663920556988711e-05,
      "loss": 1.5656,
      "step": 220896
    },
    {
      "epoch": 0.4359597174810221,
      "grad_norm": 0.11179853230714798,
      "learning_rate": 5.6632865345522414e-05,
      "loss": 1.5605,
      "step": 220928
    },
    {
      "epoch": 0.43602286344241853,
      "grad_norm": 0.11030498892068863,
      "learning_rate": 5.662652512115773e-05,
      "loss": 1.5778,
      "step": 220960
    },
    {
      "epoch": 0.436086009403815,
      "grad_norm": 0.11525856703519821,
      "learning_rate": 5.6620184896793036e-05,
      "loss": 1.567,
      "step": 220992
    },
    {
      "epoch": 0.4361491553652114,
      "grad_norm": 0.11659888923168182,
      "learning_rate": 5.661384467242835e-05,
      "loss": 1.5755,
      "step": 221024
    },
    {
      "epoch": 0.4362123013266079,
      "grad_norm": 0.11479858309030533,
      "learning_rate": 5.6607504448063665e-05,
      "loss": 1.5653,
      "step": 221056
    },
    {
      "epoch": 0.43627544728800427,
      "grad_norm": 0.10940715670585632,
      "learning_rate": 5.6601164223698966e-05,
      "loss": 1.5795,
      "step": 221088
    },
    {
      "epoch": 0.4363385932494007,
      "grad_norm": 0.12156342715024948,
      "learning_rate": 5.659482399933428e-05,
      "loss": 1.5707,
      "step": 221120
    },
    {
      "epoch": 0.43640173921079717,
      "grad_norm": 0.11272057890892029,
      "learning_rate": 5.658848377496959e-05,
      "loss": 1.5699,
      "step": 221152
    },
    {
      "epoch": 0.4364648851721936,
      "grad_norm": 0.10929644852876663,
      "learning_rate": 5.65821435506049e-05,
      "loss": 1.575,
      "step": 221184
    },
    {
      "epoch": 0.43652803113359007,
      "grad_norm": 0.12085824459791183,
      "learning_rate": 5.6575803326240204e-05,
      "loss": 1.5657,
      "step": 221216
    },
    {
      "epoch": 0.4365911770949865,
      "grad_norm": 0.11290782690048218,
      "learning_rate": 5.656946310187552e-05,
      "loss": 1.5715,
      "step": 221248
    },
    {
      "epoch": 0.4366543230563829,
      "grad_norm": 0.11898967623710632,
      "learning_rate": 5.656312287751083e-05,
      "loss": 1.5683,
      "step": 221280
    },
    {
      "epoch": 0.43671746901777936,
      "grad_norm": 0.1148371770977974,
      "learning_rate": 5.655678265314615e-05,
      "loss": 1.5605,
      "step": 221312
    },
    {
      "epoch": 0.4367806149791758,
      "grad_norm": 0.11350247263908386,
      "learning_rate": 5.655044242878145e-05,
      "loss": 1.5729,
      "step": 221344
    },
    {
      "epoch": 0.43684376094057226,
      "grad_norm": 0.11910045146942139,
      "learning_rate": 5.6544102204416763e-05,
      "loss": 1.5643,
      "step": 221376
    },
    {
      "epoch": 0.4369069069019687,
      "grad_norm": 0.11791671067476273,
      "learning_rate": 5.653776198005207e-05,
      "loss": 1.5709,
      "step": 221408
    },
    {
      "epoch": 0.43697005286336515,
      "grad_norm": 0.10673581808805466,
      "learning_rate": 5.6531421755687386e-05,
      "loss": 1.5751,
      "step": 221440
    },
    {
      "epoch": 0.43703319882476155,
      "grad_norm": 0.11276498436927795,
      "learning_rate": 5.652508153132269e-05,
      "loss": 1.5636,
      "step": 221472
    },
    {
      "epoch": 0.437096344786158,
      "grad_norm": 0.11368037760257721,
      "learning_rate": 5.6518741306958e-05,
      "loss": 1.5747,
      "step": 221504
    },
    {
      "epoch": 0.43715949074755445,
      "grad_norm": 0.12395352125167847,
      "learning_rate": 5.6512401082593316e-05,
      "loss": 1.5554,
      "step": 221536
    },
    {
      "epoch": 0.4372226367089509,
      "grad_norm": 0.10890449583530426,
      "learning_rate": 5.6506060858228624e-05,
      "loss": 1.5659,
      "step": 221568
    },
    {
      "epoch": 0.43728578267034734,
      "grad_norm": 0.11172203719615936,
      "learning_rate": 5.649972063386393e-05,
      "loss": 1.5746,
      "step": 221600
    },
    {
      "epoch": 0.4373489286317438,
      "grad_norm": 0.10597629845142365,
      "learning_rate": 5.649338040949924e-05,
      "loss": 1.5841,
      "step": 221632
    },
    {
      "epoch": 0.4374120745931402,
      "grad_norm": 0.11421770602464676,
      "learning_rate": 5.6487040185134554e-05,
      "loss": 1.5672,
      "step": 221664
    },
    {
      "epoch": 0.43747522055453664,
      "grad_norm": 0.12479004263877869,
      "learning_rate": 5.648069996076987e-05,
      "loss": 1.561,
      "step": 221696
    },
    {
      "epoch": 0.4375383665159331,
      "grad_norm": 0.11543096601963043,
      "learning_rate": 5.647435973640518e-05,
      "loss": 1.5819,
      "step": 221728
    },
    {
      "epoch": 0.43760151247732954,
      "grad_norm": 0.11432941257953644,
      "learning_rate": 5.6468019512040484e-05,
      "loss": 1.5856,
      "step": 221760
    },
    {
      "epoch": 0.437664658438726,
      "grad_norm": 0.11683588474988937,
      "learning_rate": 5.646167928767579e-05,
      "loss": 1.5702,
      "step": 221792
    },
    {
      "epoch": 0.43772780440012243,
      "grad_norm": 0.105549655854702,
      "learning_rate": 5.6455339063311106e-05,
      "loss": 1.5828,
      "step": 221824
    },
    {
      "epoch": 0.4377909503615188,
      "grad_norm": 0.1098296195268631,
      "learning_rate": 5.644899883894642e-05,
      "loss": 1.5715,
      "step": 221856
    },
    {
      "epoch": 0.4378540963229153,
      "grad_norm": 0.11159711331129074,
      "learning_rate": 5.644265861458172e-05,
      "loss": 1.5694,
      "step": 221888
    },
    {
      "epoch": 0.4379172422843117,
      "grad_norm": 0.10768149048089981,
      "learning_rate": 5.6436318390217036e-05,
      "loss": 1.5761,
      "step": 221920
    },
    {
      "epoch": 0.4379803882457082,
      "grad_norm": 0.1107068657875061,
      "learning_rate": 5.642997816585235e-05,
      "loss": 1.5558,
      "step": 221952
    },
    {
      "epoch": 0.4380435342071046,
      "grad_norm": 0.11191987246274948,
      "learning_rate": 5.642363794148766e-05,
      "loss": 1.5752,
      "step": 221984
    },
    {
      "epoch": 0.4381066801685011,
      "grad_norm": 0.10686592757701874,
      "learning_rate": 5.641729771712297e-05,
      "loss": 1.566,
      "step": 222016
    },
    {
      "epoch": 0.43816982612989747,
      "grad_norm": 0.1115211769938469,
      "learning_rate": 5.6410957492758274e-05,
      "loss": 1.5667,
      "step": 222048
    },
    {
      "epoch": 0.4382329720912939,
      "grad_norm": 0.11843854933977127,
      "learning_rate": 5.640461726839359e-05,
      "loss": 1.5657,
      "step": 222080
    },
    {
      "epoch": 0.43829611805269036,
      "grad_norm": 0.1118033155798912,
      "learning_rate": 5.6398277044028904e-05,
      "loss": 1.5741,
      "step": 222112
    },
    {
      "epoch": 0.4383592640140868,
      "grad_norm": 0.11383120715618134,
      "learning_rate": 5.6391936819664205e-05,
      "loss": 1.5662,
      "step": 222144
    },
    {
      "epoch": 0.43842240997548326,
      "grad_norm": 0.1308271735906601,
      "learning_rate": 5.638559659529952e-05,
      "loss": 1.5664,
      "step": 222176
    },
    {
      "epoch": 0.4384855559368797,
      "grad_norm": 0.11619142442941666,
      "learning_rate": 5.637925637093483e-05,
      "loss": 1.5571,
      "step": 222208
    },
    {
      "epoch": 0.4385487018982761,
      "grad_norm": 0.10571594536304474,
      "learning_rate": 5.637291614657014e-05,
      "loss": 1.5656,
      "step": 222240
    },
    {
      "epoch": 0.43861184785967255,
      "grad_norm": 0.11335493624210358,
      "learning_rate": 5.636657592220544e-05,
      "loss": 1.5706,
      "step": 222272
    },
    {
      "epoch": 0.438674993821069,
      "grad_norm": 0.1135411486029625,
      "learning_rate": 5.636023569784076e-05,
      "loss": 1.5781,
      "step": 222304
    },
    {
      "epoch": 0.43873813978246545,
      "grad_norm": 0.11494645476341248,
      "learning_rate": 5.635389547347607e-05,
      "loss": 1.5787,
      "step": 222336
    },
    {
      "epoch": 0.4388012857438619,
      "grad_norm": 0.11322207748889923,
      "learning_rate": 5.6347555249111386e-05,
      "loss": 1.5772,
      "step": 222368
    },
    {
      "epoch": 0.43886443170525835,
      "grad_norm": 0.11661253124475479,
      "learning_rate": 5.6341215024746694e-05,
      "loss": 1.5762,
      "step": 222400
    },
    {
      "epoch": 0.43892757766665474,
      "grad_norm": 0.11384445428848267,
      "learning_rate": 5.6334874800381995e-05,
      "loss": 1.5707,
      "step": 222432
    },
    {
      "epoch": 0.4389907236280512,
      "grad_norm": 0.10868734866380692,
      "learning_rate": 5.632853457601731e-05,
      "loss": 1.5545,
      "step": 222464
    },
    {
      "epoch": 0.43905386958944764,
      "grad_norm": 0.10969744622707367,
      "learning_rate": 5.6322194351652624e-05,
      "loss": 1.574,
      "step": 222496
    },
    {
      "epoch": 0.4391170155508441,
      "grad_norm": 0.11876323819160461,
      "learning_rate": 5.631585412728794e-05,
      "loss": 1.5576,
      "step": 222528
    },
    {
      "epoch": 0.43918016151224054,
      "grad_norm": 0.11136127263307571,
      "learning_rate": 5.630951390292324e-05,
      "loss": 1.572,
      "step": 222560
    },
    {
      "epoch": 0.439243307473637,
      "grad_norm": 0.12057178467512131,
      "learning_rate": 5.6303173678558554e-05,
      "loss": 1.5735,
      "step": 222592
    },
    {
      "epoch": 0.43930645343503344,
      "grad_norm": 0.1075592190027237,
      "learning_rate": 5.629683345419386e-05,
      "loss": 1.5741,
      "step": 222624
    },
    {
      "epoch": 0.43936959939642983,
      "grad_norm": 0.11222700774669647,
      "learning_rate": 5.6290493229829177e-05,
      "loss": 1.5753,
      "step": 222656
    },
    {
      "epoch": 0.4394327453578263,
      "grad_norm": 0.1292693316936493,
      "learning_rate": 5.628415300546448e-05,
      "loss": 1.5765,
      "step": 222688
    },
    {
      "epoch": 0.43949589131922273,
      "grad_norm": 0.12229131162166595,
      "learning_rate": 5.627781278109979e-05,
      "loss": 1.5642,
      "step": 222720
    },
    {
      "epoch": 0.4395590372806192,
      "grad_norm": 0.10908335447311401,
      "learning_rate": 5.627147255673511e-05,
      "loss": 1.5639,
      "step": 222752
    },
    {
      "epoch": 0.43962218324201563,
      "grad_norm": 0.10967562347650528,
      "learning_rate": 5.626513233237042e-05,
      "loss": 1.5617,
      "step": 222784
    },
    {
      "epoch": 0.4396853292034121,
      "grad_norm": 0.10927215963602066,
      "learning_rate": 5.625879210800572e-05,
      "loss": 1.5724,
      "step": 222816
    },
    {
      "epoch": 0.43974847516480847,
      "grad_norm": 0.12057118117809296,
      "learning_rate": 5.625245188364103e-05,
      "loss": 1.5673,
      "step": 222848
    },
    {
      "epoch": 0.4398116211262049,
      "grad_norm": 0.10720538347959518,
      "learning_rate": 5.6246111659276345e-05,
      "loss": 1.5737,
      "step": 222880
    },
    {
      "epoch": 0.43987476708760137,
      "grad_norm": 0.10852023959159851,
      "learning_rate": 5.623977143491166e-05,
      "loss": 1.5879,
      "step": 222912
    },
    {
      "epoch": 0.4399379130489978,
      "grad_norm": 0.11148054152727127,
      "learning_rate": 5.623343121054696e-05,
      "loss": 1.5604,
      "step": 222944
    },
    {
      "epoch": 0.44000105901039427,
      "grad_norm": 0.1090625748038292,
      "learning_rate": 5.6227090986182275e-05,
      "loss": 1.5623,
      "step": 222976
    },
    {
      "epoch": 0.4400642049717907,
      "grad_norm": 0.11345841735601425,
      "learning_rate": 5.622075076181759e-05,
      "loss": 1.5586,
      "step": 223008
    },
    {
      "epoch": 0.4401273509331871,
      "grad_norm": 0.1110571026802063,
      "learning_rate": 5.62144105374529e-05,
      "loss": 1.5646,
      "step": 223040
    },
    {
      "epoch": 0.44019049689458356,
      "grad_norm": 0.1142256110906601,
      "learning_rate": 5.620807031308821e-05,
      "loss": 1.5691,
      "step": 223072
    },
    {
      "epoch": 0.44025364285598,
      "grad_norm": 0.11341311782598495,
      "learning_rate": 5.620173008872351e-05,
      "loss": 1.5704,
      "step": 223104
    },
    {
      "epoch": 0.44031678881737646,
      "grad_norm": 0.1143636628985405,
      "learning_rate": 5.619538986435883e-05,
      "loss": 1.5726,
      "step": 223136
    },
    {
      "epoch": 0.4403799347787729,
      "grad_norm": 0.11442039161920547,
      "learning_rate": 5.618904963999414e-05,
      "loss": 1.5664,
      "step": 223168
    },
    {
      "epoch": 0.44044308074016936,
      "grad_norm": 0.1149403378367424,
      "learning_rate": 5.6182709415629456e-05,
      "loss": 1.5679,
      "step": 223200
    },
    {
      "epoch": 0.44050622670156575,
      "grad_norm": 0.12742525339126587,
      "learning_rate": 5.617636919126476e-05,
      "loss": 1.5689,
      "step": 223232
    },
    {
      "epoch": 0.4405693726629622,
      "grad_norm": 0.10830821096897125,
      "learning_rate": 5.6170028966900065e-05,
      "loss": 1.5683,
      "step": 223264
    },
    {
      "epoch": 0.44063251862435865,
      "grad_norm": 0.11148485541343689,
      "learning_rate": 5.616368874253538e-05,
      "loss": 1.5631,
      "step": 223296
    },
    {
      "epoch": 0.4406956645857551,
      "grad_norm": 0.11369830369949341,
      "learning_rate": 5.6157348518170694e-05,
      "loss": 1.5668,
      "step": 223328
    },
    {
      "epoch": 0.44075881054715155,
      "grad_norm": 0.11327175796031952,
      "learning_rate": 5.6151008293805995e-05,
      "loss": 1.5615,
      "step": 223360
    },
    {
      "epoch": 0.440821956508548,
      "grad_norm": 0.11536484211683273,
      "learning_rate": 5.614466806944131e-05,
      "loss": 1.5665,
      "step": 223392
    },
    {
      "epoch": 0.4408851024699444,
      "grad_norm": 0.11269336193799973,
      "learning_rate": 5.6138327845076624e-05,
      "loss": 1.5691,
      "step": 223424
    },
    {
      "epoch": 0.44094824843134084,
      "grad_norm": 0.10846231132745743,
      "learning_rate": 5.613198762071193e-05,
      "loss": 1.5747,
      "step": 223456
    },
    {
      "epoch": 0.4410113943927373,
      "grad_norm": 0.11808638274669647,
      "learning_rate": 5.612564739634723e-05,
      "loss": 1.5587,
      "step": 223488
    },
    {
      "epoch": 0.44107454035413374,
      "grad_norm": 0.10712187737226486,
      "learning_rate": 5.611930717198255e-05,
      "loss": 1.5672,
      "step": 223520
    },
    {
      "epoch": 0.4411376863155302,
      "grad_norm": 0.1203143373131752,
      "learning_rate": 5.611296694761786e-05,
      "loss": 1.5547,
      "step": 223552
    },
    {
      "epoch": 0.44120083227692664,
      "grad_norm": 0.1094353124499321,
      "learning_rate": 5.610662672325318e-05,
      "loss": 1.5624,
      "step": 223584
    },
    {
      "epoch": 0.44126397823832303,
      "grad_norm": 0.11553500592708588,
      "learning_rate": 5.610028649888848e-05,
      "loss": 1.5741,
      "step": 223616
    },
    {
      "epoch": 0.4413271241997195,
      "grad_norm": 0.11306028068065643,
      "learning_rate": 5.609394627452379e-05,
      "loss": 1.5613,
      "step": 223648
    },
    {
      "epoch": 0.4413902701611159,
      "grad_norm": 0.11836506426334381,
      "learning_rate": 5.60876060501591e-05,
      "loss": 1.5698,
      "step": 223680
    },
    {
      "epoch": 0.4414534161225124,
      "grad_norm": 0.11492486298084259,
      "learning_rate": 5.6081265825794415e-05,
      "loss": 1.5777,
      "step": 223712
    },
    {
      "epoch": 0.4415165620839088,
      "grad_norm": 0.11137628555297852,
      "learning_rate": 5.607492560142973e-05,
      "loss": 1.5652,
      "step": 223744
    },
    {
      "epoch": 0.4415797080453053,
      "grad_norm": 0.10922759771347046,
      "learning_rate": 5.606858537706503e-05,
      "loss": 1.5645,
      "step": 223776
    },
    {
      "epoch": 0.44164285400670167,
      "grad_norm": 0.12418708950281143,
      "learning_rate": 5.6062245152700345e-05,
      "loss": 1.5639,
      "step": 223808
    },
    {
      "epoch": 0.4417059999680981,
      "grad_norm": 0.11384117603302002,
      "learning_rate": 5.605590492833566e-05,
      "loss": 1.5717,
      "step": 223840
    },
    {
      "epoch": 0.44176914592949457,
      "grad_norm": 0.12185539305210114,
      "learning_rate": 5.604956470397097e-05,
      "loss": 1.5631,
      "step": 223872
    },
    {
      "epoch": 0.441832291890891,
      "grad_norm": 0.11555115133523941,
      "learning_rate": 5.604322447960627e-05,
      "loss": 1.575,
      "step": 223904
    },
    {
      "epoch": 0.44189543785228746,
      "grad_norm": 0.11330656707286835,
      "learning_rate": 5.603688425524158e-05,
      "loss": 1.5627,
      "step": 223936
    },
    {
      "epoch": 0.4419585838136839,
      "grad_norm": 0.11715026199817657,
      "learning_rate": 5.60305440308769e-05,
      "loss": 1.5715,
      "step": 223968
    },
    {
      "epoch": 0.4420217297750803,
      "grad_norm": 0.11196105182170868,
      "learning_rate": 5.602420380651221e-05,
      "loss": 1.5772,
      "step": 224000
    },
    {
      "epoch": 0.44208487573647676,
      "grad_norm": 0.11672034114599228,
      "learning_rate": 5.601786358214751e-05,
      "loss": 1.5755,
      "step": 224032
    },
    {
      "epoch": 0.4421480216978732,
      "grad_norm": 0.11318805813789368,
      "learning_rate": 5.601152335778283e-05,
      "loss": 1.5604,
      "step": 224064
    },
    {
      "epoch": 0.44221116765926965,
      "grad_norm": 0.11469605565071106,
      "learning_rate": 5.6005183133418135e-05,
      "loss": 1.5569,
      "step": 224096
    },
    {
      "epoch": 0.4422743136206661,
      "grad_norm": 0.10346221923828125,
      "learning_rate": 5.599884290905345e-05,
      "loss": 1.5855,
      "step": 224128
    },
    {
      "epoch": 0.44233745958206255,
      "grad_norm": 0.10945054143667221,
      "learning_rate": 5.599250268468875e-05,
      "loss": 1.5611,
      "step": 224160
    },
    {
      "epoch": 0.442400605543459,
      "grad_norm": 0.11730410158634186,
      "learning_rate": 5.5986162460324066e-05,
      "loss": 1.5697,
      "step": 224192
    },
    {
      "epoch": 0.4424637515048554,
      "grad_norm": 0.1142803505063057,
      "learning_rate": 5.597982223595938e-05,
      "loss": 1.5713,
      "step": 224224
    },
    {
      "epoch": 0.44252689746625185,
      "grad_norm": 0.1277397722005844,
      "learning_rate": 5.597348201159469e-05,
      "loss": 1.5721,
      "step": 224256
    },
    {
      "epoch": 0.4425900434276483,
      "grad_norm": 0.11798173189163208,
      "learning_rate": 5.5967141787229996e-05,
      "loss": 1.5756,
      "step": 224288
    },
    {
      "epoch": 0.44265318938904474,
      "grad_norm": 0.11593777686357498,
      "learning_rate": 5.5960801562865303e-05,
      "loss": 1.5678,
      "step": 224320
    },
    {
      "epoch": 0.4427163353504412,
      "grad_norm": 0.1148558184504509,
      "learning_rate": 5.595446133850062e-05,
      "loss": 1.5768,
      "step": 224352
    },
    {
      "epoch": 0.44277948131183764,
      "grad_norm": 0.11958102136850357,
      "learning_rate": 5.594812111413593e-05,
      "loss": 1.5714,
      "step": 224384
    },
    {
      "epoch": 0.44284262727323404,
      "grad_norm": 0.11051355302333832,
      "learning_rate": 5.594178088977125e-05,
      "loss": 1.5574,
      "step": 224416
    },
    {
      "epoch": 0.4429057732346305,
      "grad_norm": 0.11384843289852142,
      "learning_rate": 5.593544066540655e-05,
      "loss": 1.5627,
      "step": 224448
    },
    {
      "epoch": 0.44296891919602693,
      "grad_norm": 0.11094524711370468,
      "learning_rate": 5.592910044104186e-05,
      "loss": 1.571,
      "step": 224480
    },
    {
      "epoch": 0.4430320651574234,
      "grad_norm": 0.11181817203760147,
      "learning_rate": 5.592276021667717e-05,
      "loss": 1.5768,
      "step": 224512
    },
    {
      "epoch": 0.44309521111881983,
      "grad_norm": 0.11415065079927444,
      "learning_rate": 5.5916419992312485e-05,
      "loss": 1.5852,
      "step": 224544
    },
    {
      "epoch": 0.4431583570802163,
      "grad_norm": 0.11386341601610184,
      "learning_rate": 5.5910079767947786e-05,
      "loss": 1.5603,
      "step": 224576
    },
    {
      "epoch": 0.4432215030416127,
      "grad_norm": 0.10986695438623428,
      "learning_rate": 5.59037395435831e-05,
      "loss": 1.5747,
      "step": 224608
    },
    {
      "epoch": 0.4432846490030091,
      "grad_norm": 0.10902444273233414,
      "learning_rate": 5.5897399319218415e-05,
      "loss": 1.574,
      "step": 224640
    },
    {
      "epoch": 0.4433477949644056,
      "grad_norm": 0.11428419500589371,
      "learning_rate": 5.589105909485372e-05,
      "loss": 1.5682,
      "step": 224672
    },
    {
      "epoch": 0.443410940925802,
      "grad_norm": 0.11163070797920227,
      "learning_rate": 5.588471887048903e-05,
      "loss": 1.5745,
      "step": 224704
    },
    {
      "epoch": 0.44347408688719847,
      "grad_norm": 0.11591023206710815,
      "learning_rate": 5.587837864612434e-05,
      "loss": 1.5763,
      "step": 224736
    },
    {
      "epoch": 0.4435372328485949,
      "grad_norm": 0.11970361322164536,
      "learning_rate": 5.587203842175965e-05,
      "loss": 1.5783,
      "step": 224768
    },
    {
      "epoch": 0.4436003788099913,
      "grad_norm": 0.12001791596412659,
      "learning_rate": 5.586569819739497e-05,
      "loss": 1.5835,
      "step": 224800
    },
    {
      "epoch": 0.44366352477138776,
      "grad_norm": 0.11000420898199081,
      "learning_rate": 5.585935797303027e-05,
      "loss": 1.5723,
      "step": 224832
    },
    {
      "epoch": 0.4437266707327842,
      "grad_norm": 0.1081654354929924,
      "learning_rate": 5.585301774866558e-05,
      "loss": 1.5644,
      "step": 224864
    },
    {
      "epoch": 0.44378981669418066,
      "grad_norm": 0.1101280227303505,
      "learning_rate": 5.584667752430089e-05,
      "loss": 1.5772,
      "step": 224896
    },
    {
      "epoch": 0.4438529626555771,
      "grad_norm": 0.12191268056631088,
      "learning_rate": 5.5840337299936206e-05,
      "loss": 1.5798,
      "step": 224928
    },
    {
      "epoch": 0.44391610861697356,
      "grad_norm": 0.11502295732498169,
      "learning_rate": 5.5833997075571507e-05,
      "loss": 1.5643,
      "step": 224960
    },
    {
      "epoch": 0.44397925457836995,
      "grad_norm": 0.11195903271436691,
      "learning_rate": 5.582765685120682e-05,
      "loss": 1.5854,
      "step": 224992
    },
    {
      "epoch": 0.4440424005397664,
      "grad_norm": 0.1128184124827385,
      "learning_rate": 5.5821316626842136e-05,
      "loss": 1.5732,
      "step": 225024
    },
    {
      "epoch": 0.44410554650116285,
      "grad_norm": 0.12109333276748657,
      "learning_rate": 5.581497640247745e-05,
      "loss": 1.5761,
      "step": 225056
    },
    {
      "epoch": 0.4441686924625593,
      "grad_norm": 0.10844786465167999,
      "learning_rate": 5.580863617811276e-05,
      "loss": 1.5692,
      "step": 225088
    },
    {
      "epoch": 0.44423183842395575,
      "grad_norm": 0.11165008693933487,
      "learning_rate": 5.5802295953748066e-05,
      "loss": 1.5773,
      "step": 225120
    },
    {
      "epoch": 0.4442949843853522,
      "grad_norm": 0.11832918226718903,
      "learning_rate": 5.5795955729383374e-05,
      "loss": 1.5771,
      "step": 225152
    },
    {
      "epoch": 0.4443581303467486,
      "grad_norm": 0.12097255140542984,
      "learning_rate": 5.578961550501869e-05,
      "loss": 1.5842,
      "step": 225184
    },
    {
      "epoch": 0.44442127630814504,
      "grad_norm": 0.11156468838453293,
      "learning_rate": 5.5783275280654e-05,
      "loss": 1.5732,
      "step": 225216
    },
    {
      "epoch": 0.4444844222695415,
      "grad_norm": 0.11373218894004822,
      "learning_rate": 5.5776935056289304e-05,
      "loss": 1.5621,
      "step": 225248
    },
    {
      "epoch": 0.44454756823093794,
      "grad_norm": 0.1159541979432106,
      "learning_rate": 5.577059483192462e-05,
      "loss": 1.56,
      "step": 225280
    },
    {
      "epoch": 0.4446107141923344,
      "grad_norm": 0.10900960117578506,
      "learning_rate": 5.5764254607559926e-05,
      "loss": 1.5613,
      "step": 225312
    },
    {
      "epoch": 0.44467386015373084,
      "grad_norm": 0.10891033709049225,
      "learning_rate": 5.575791438319524e-05,
      "loss": 1.5664,
      "step": 225344
    },
    {
      "epoch": 0.44473700611512723,
      "grad_norm": 0.11245907843112946,
      "learning_rate": 5.575157415883054e-05,
      "loss": 1.5779,
      "step": 225376
    },
    {
      "epoch": 0.4448001520765237,
      "grad_norm": 0.10898572951555252,
      "learning_rate": 5.5745233934465856e-05,
      "loss": 1.5696,
      "step": 225408
    },
    {
      "epoch": 0.44486329803792013,
      "grad_norm": 0.11658014357089996,
      "learning_rate": 5.573889371010117e-05,
      "loss": 1.5656,
      "step": 225440
    },
    {
      "epoch": 0.4449264439993166,
      "grad_norm": 0.10766708850860596,
      "learning_rate": 5.5732553485736485e-05,
      "loss": 1.5705,
      "step": 225472
    },
    {
      "epoch": 0.44498958996071303,
      "grad_norm": 0.12098678946495056,
      "learning_rate": 5.5726213261371786e-05,
      "loss": 1.5676,
      "step": 225504
    },
    {
      "epoch": 0.4450527359221095,
      "grad_norm": 0.1107996255159378,
      "learning_rate": 5.57198730370071e-05,
      "loss": 1.5779,
      "step": 225536
    },
    {
      "epoch": 0.44511588188350587,
      "grad_norm": 0.11110315471887589,
      "learning_rate": 5.571353281264241e-05,
      "loss": 1.5708,
      "step": 225568
    },
    {
      "epoch": 0.4451790278449023,
      "grad_norm": 0.11498510837554932,
      "learning_rate": 5.570719258827772e-05,
      "loss": 1.5671,
      "step": 225600
    },
    {
      "epoch": 0.44524217380629877,
      "grad_norm": 0.11473532766103745,
      "learning_rate": 5.5700852363913024e-05,
      "loss": 1.5821,
      "step": 225632
    },
    {
      "epoch": 0.4453053197676952,
      "grad_norm": 0.11622277647256851,
      "learning_rate": 5.569451213954834e-05,
      "loss": 1.5739,
      "step": 225664
    },
    {
      "epoch": 0.44536846572909167,
      "grad_norm": 0.10744191706180573,
      "learning_rate": 5.5688171915183653e-05,
      "loss": 1.5738,
      "step": 225696
    },
    {
      "epoch": 0.4454316116904881,
      "grad_norm": 0.11692357063293457,
      "learning_rate": 5.568183169081896e-05,
      "loss": 1.5703,
      "step": 225728
    },
    {
      "epoch": 0.4454947576518845,
      "grad_norm": 0.10964972525835037,
      "learning_rate": 5.5675491466454276e-05,
      "loss": 1.5685,
      "step": 225760
    },
    {
      "epoch": 0.44555790361328096,
      "grad_norm": 0.11665034294128418,
      "learning_rate": 5.566915124208958e-05,
      "loss": 1.5583,
      "step": 225792
    },
    {
      "epoch": 0.4456210495746774,
      "grad_norm": 0.11089098453521729,
      "learning_rate": 5.566281101772489e-05,
      "loss": 1.5808,
      "step": 225824
    },
    {
      "epoch": 0.44568419553607386,
      "grad_norm": 0.12039243429899216,
      "learning_rate": 5.5656470793360206e-05,
      "loss": 1.5795,
      "step": 225856
    },
    {
      "epoch": 0.4457473414974703,
      "grad_norm": 0.12202098965644836,
      "learning_rate": 5.565013056899552e-05,
      "loss": 1.5655,
      "step": 225888
    },
    {
      "epoch": 0.44581048745886676,
      "grad_norm": 0.12360382825136185,
      "learning_rate": 5.564379034463082e-05,
      "loss": 1.5789,
      "step": 225920
    },
    {
      "epoch": 0.4458736334202632,
      "grad_norm": 0.10443336516618729,
      "learning_rate": 5.563745012026613e-05,
      "loss": 1.5763,
      "step": 225952
    },
    {
      "epoch": 0.4459367793816596,
      "grad_norm": 0.10910505801439285,
      "learning_rate": 5.5631109895901444e-05,
      "loss": 1.5589,
      "step": 225984
    },
    {
      "epoch": 0.44599992534305605,
      "grad_norm": 0.11002499610185623,
      "learning_rate": 5.562476967153676e-05,
      "loss": 1.5648,
      "step": 226016
    },
    {
      "epoch": 0.4460630713044525,
      "grad_norm": 0.12160021811723709,
      "learning_rate": 5.561842944717206e-05,
      "loss": 1.5655,
      "step": 226048
    },
    {
      "epoch": 0.44612621726584895,
      "grad_norm": 0.11697404831647873,
      "learning_rate": 5.5612089222807374e-05,
      "loss": 1.5684,
      "step": 226080
    },
    {
      "epoch": 0.4461893632272454,
      "grad_norm": 0.1245870590209961,
      "learning_rate": 5.560574899844269e-05,
      "loss": 1.573,
      "step": 226112
    },
    {
      "epoch": 0.44625250918864184,
      "grad_norm": 0.11575856059789658,
      "learning_rate": 5.5599408774077996e-05,
      "loss": 1.5684,
      "step": 226144
    },
    {
      "epoch": 0.44631565515003824,
      "grad_norm": 0.11695855110883713,
      "learning_rate": 5.5593068549713304e-05,
      "loss": 1.5709,
      "step": 226176
    },
    {
      "epoch": 0.4463788011114347,
      "grad_norm": 0.1154797151684761,
      "learning_rate": 5.558672832534861e-05,
      "loss": 1.5726,
      "step": 226208
    },
    {
      "epoch": 0.44644194707283114,
      "grad_norm": 0.12129583954811096,
      "learning_rate": 5.5580388100983926e-05,
      "loss": 1.5703,
      "step": 226240
    },
    {
      "epoch": 0.4465050930342276,
      "grad_norm": 0.1199704259634018,
      "learning_rate": 5.557404787661924e-05,
      "loss": 1.57,
      "step": 226272
    },
    {
      "epoch": 0.44656823899562403,
      "grad_norm": 0.11570403724908829,
      "learning_rate": 5.556770765225454e-05,
      "loss": 1.5578,
      "step": 226304
    },
    {
      "epoch": 0.4466313849570205,
      "grad_norm": 0.10727885365486145,
      "learning_rate": 5.5561367427889857e-05,
      "loss": 1.5598,
      "step": 226336
    },
    {
      "epoch": 0.4466945309184169,
      "grad_norm": 0.12253036350011826,
      "learning_rate": 5.5555027203525164e-05,
      "loss": 1.5698,
      "step": 226368
    },
    {
      "epoch": 0.4467576768798133,
      "grad_norm": 0.10797633230686188,
      "learning_rate": 5.554868697916048e-05,
      "loss": 1.5617,
      "step": 226400
    },
    {
      "epoch": 0.4468208228412098,
      "grad_norm": 0.11121440678834915,
      "learning_rate": 5.5542346754795794e-05,
      "loss": 1.563,
      "step": 226432
    },
    {
      "epoch": 0.4468839688026062,
      "grad_norm": 0.1147451102733612,
      "learning_rate": 5.5536006530431095e-05,
      "loss": 1.5654,
      "step": 226464
    },
    {
      "epoch": 0.4469471147640027,
      "grad_norm": 0.10988610237836838,
      "learning_rate": 5.552966630606641e-05,
      "loss": 1.565,
      "step": 226496
    },
    {
      "epoch": 0.4470102607253991,
      "grad_norm": 0.11073459684848785,
      "learning_rate": 5.5523326081701724e-05,
      "loss": 1.579,
      "step": 226528
    },
    {
      "epoch": 0.4470734066867955,
      "grad_norm": 0.12059857696294785,
      "learning_rate": 5.551698585733703e-05,
      "loss": 1.5728,
      "step": 226560
    },
    {
      "epoch": 0.44713655264819196,
      "grad_norm": 0.11194503307342529,
      "learning_rate": 5.551064563297233e-05,
      "loss": 1.5695,
      "step": 226592
    },
    {
      "epoch": 0.4471996986095884,
      "grad_norm": 0.11144059896469116,
      "learning_rate": 5.550430540860765e-05,
      "loss": 1.5699,
      "step": 226624
    },
    {
      "epoch": 0.44726284457098486,
      "grad_norm": 0.11113991588354111,
      "learning_rate": 5.549796518424296e-05,
      "loss": 1.5815,
      "step": 226656
    },
    {
      "epoch": 0.4473259905323813,
      "grad_norm": 0.1123872920870781,
      "learning_rate": 5.5491624959878276e-05,
      "loss": 1.5779,
      "step": 226688
    },
    {
      "epoch": 0.44738913649377776,
      "grad_norm": 0.11324728280305862,
      "learning_rate": 5.548528473551358e-05,
      "loss": 1.5608,
      "step": 226720
    },
    {
      "epoch": 0.44745228245517416,
      "grad_norm": 0.11159668117761612,
      "learning_rate": 5.547894451114889e-05,
      "loss": 1.5596,
      "step": 226752
    },
    {
      "epoch": 0.4475154284165706,
      "grad_norm": 0.10906250774860382,
      "learning_rate": 5.54726042867842e-05,
      "loss": 1.5752,
      "step": 226784
    },
    {
      "epoch": 0.44757857437796705,
      "grad_norm": 0.10427848249673843,
      "learning_rate": 5.5466264062419514e-05,
      "loss": 1.5719,
      "step": 226816
    },
    {
      "epoch": 0.4476417203393635,
      "grad_norm": 0.11987894773483276,
      "learning_rate": 5.5459923838054815e-05,
      "loss": 1.585,
      "step": 226848
    },
    {
      "epoch": 0.44770486630075995,
      "grad_norm": 0.112233005464077,
      "learning_rate": 5.545358361369013e-05,
      "loss": 1.5783,
      "step": 226880
    },
    {
      "epoch": 0.4477680122621564,
      "grad_norm": 0.11146263778209686,
      "learning_rate": 5.5447243389325444e-05,
      "loss": 1.5782,
      "step": 226912
    },
    {
      "epoch": 0.4478311582235528,
      "grad_norm": 0.10751260817050934,
      "learning_rate": 5.544090316496076e-05,
      "loss": 1.5837,
      "step": 226944
    },
    {
      "epoch": 0.44789430418494924,
      "grad_norm": 0.1107482835650444,
      "learning_rate": 5.543456294059606e-05,
      "loss": 1.5846,
      "step": 226976
    },
    {
      "epoch": 0.4479574501463457,
      "grad_norm": 0.10877012461423874,
      "learning_rate": 5.542822271623137e-05,
      "loss": 1.5757,
      "step": 227008
    },
    {
      "epoch": 0.44802059610774214,
      "grad_norm": 0.10649998486042023,
      "learning_rate": 5.542188249186668e-05,
      "loss": 1.5527,
      "step": 227040
    },
    {
      "epoch": 0.4480837420691386,
      "grad_norm": 0.11101628094911575,
      "learning_rate": 5.5415542267502e-05,
      "loss": 1.5585,
      "step": 227072
    },
    {
      "epoch": 0.44814688803053504,
      "grad_norm": 0.10834810137748718,
      "learning_rate": 5.540920204313731e-05,
      "loss": 1.5686,
      "step": 227104
    },
    {
      "epoch": 0.44821003399193143,
      "grad_norm": 0.11642777174711227,
      "learning_rate": 5.540286181877261e-05,
      "loss": 1.5727,
      "step": 227136
    },
    {
      "epoch": 0.4482731799533279,
      "grad_norm": 0.11910204589366913,
      "learning_rate": 5.539652159440793e-05,
      "loss": 1.5733,
      "step": 227168
    },
    {
      "epoch": 0.44833632591472433,
      "grad_norm": 0.12381482869386673,
      "learning_rate": 5.5390181370043235e-05,
      "loss": 1.5716,
      "step": 227200
    },
    {
      "epoch": 0.4483994718761208,
      "grad_norm": 0.112065888941288,
      "learning_rate": 5.538384114567855e-05,
      "loss": 1.5726,
      "step": 227232
    },
    {
      "epoch": 0.44846261783751723,
      "grad_norm": 0.12411680072546005,
      "learning_rate": 5.537750092131385e-05,
      "loss": 1.5627,
      "step": 227264
    },
    {
      "epoch": 0.4485257637989137,
      "grad_norm": 0.11828617006540298,
      "learning_rate": 5.5371160696949165e-05,
      "loss": 1.5684,
      "step": 227296
    },
    {
      "epoch": 0.4485889097603101,
      "grad_norm": 0.11670609563589096,
      "learning_rate": 5.536482047258448e-05,
      "loss": 1.5606,
      "step": 227328
    },
    {
      "epoch": 0.4486520557217065,
      "grad_norm": 0.10628534853458405,
      "learning_rate": 5.5358480248219794e-05,
      "loss": 1.5666,
      "step": 227360
    },
    {
      "epoch": 0.44871520168310297,
      "grad_norm": 0.11339747905731201,
      "learning_rate": 5.5352140023855095e-05,
      "loss": 1.5715,
      "step": 227392
    },
    {
      "epoch": 0.4487783476444994,
      "grad_norm": 0.1146375760436058,
      "learning_rate": 5.53457997994904e-05,
      "loss": 1.5642,
      "step": 227424
    },
    {
      "epoch": 0.44884149360589587,
      "grad_norm": 0.1137080267071724,
      "learning_rate": 5.533945957512572e-05,
      "loss": 1.5756,
      "step": 227456
    },
    {
      "epoch": 0.4489046395672923,
      "grad_norm": 0.11582580208778381,
      "learning_rate": 5.533311935076103e-05,
      "loss": 1.5603,
      "step": 227488
    },
    {
      "epoch": 0.4489677855286887,
      "grad_norm": 0.11359766870737076,
      "learning_rate": 5.532677912639633e-05,
      "loss": 1.5597,
      "step": 227520
    },
    {
      "epoch": 0.44903093149008516,
      "grad_norm": 0.11395041644573212,
      "learning_rate": 5.532043890203165e-05,
      "loss": 1.5744,
      "step": 227552
    },
    {
      "epoch": 0.4490940774514816,
      "grad_norm": 0.11050038039684296,
      "learning_rate": 5.531409867766696e-05,
      "loss": 1.5772,
      "step": 227584
    },
    {
      "epoch": 0.44915722341287806,
      "grad_norm": 0.117493636906147,
      "learning_rate": 5.530775845330227e-05,
      "loss": 1.5722,
      "step": 227616
    },
    {
      "epoch": 0.4492203693742745,
      "grad_norm": 0.11625221371650696,
      "learning_rate": 5.530141822893757e-05,
      "loss": 1.5841,
      "step": 227648
    },
    {
      "epoch": 0.44928351533567096,
      "grad_norm": 0.11636626720428467,
      "learning_rate": 5.5295078004572885e-05,
      "loss": 1.5579,
      "step": 227680
    },
    {
      "epoch": 0.4493466612970674,
      "grad_norm": 0.12009144574403763,
      "learning_rate": 5.52887377802082e-05,
      "loss": 1.5821,
      "step": 227712
    },
    {
      "epoch": 0.4494098072584638,
      "grad_norm": 0.11495005339384079,
      "learning_rate": 5.5282397555843514e-05,
      "loss": 1.5779,
      "step": 227744
    },
    {
      "epoch": 0.44947295321986025,
      "grad_norm": 0.11084603518247604,
      "learning_rate": 5.527605733147882e-05,
      "loss": 1.5736,
      "step": 227776
    },
    {
      "epoch": 0.4495360991812567,
      "grad_norm": 0.10887359082698822,
      "learning_rate": 5.526971710711413e-05,
      "loss": 1.5741,
      "step": 227808
    },
    {
      "epoch": 0.44959924514265315,
      "grad_norm": 0.114344023168087,
      "learning_rate": 5.526337688274944e-05,
      "loss": 1.5728,
      "step": 227840
    },
    {
      "epoch": 0.4496623911040496,
      "grad_norm": 0.11170511692762375,
      "learning_rate": 5.525703665838475e-05,
      "loss": 1.559,
      "step": 227872
    },
    {
      "epoch": 0.44972553706544605,
      "grad_norm": 0.11704278737306595,
      "learning_rate": 5.525069643402007e-05,
      "loss": 1.568,
      "step": 227904
    },
    {
      "epoch": 0.44978868302684244,
      "grad_norm": 0.10768473893404007,
      "learning_rate": 5.524435620965537e-05,
      "loss": 1.562,
      "step": 227936
    },
    {
      "epoch": 0.4498518289882389,
      "grad_norm": 0.12023501098155975,
      "learning_rate": 5.523801598529068e-05,
      "loss": 1.5663,
      "step": 227968
    },
    {
      "epoch": 0.44991497494963534,
      "grad_norm": 0.11995047330856323,
      "learning_rate": 5.5231675760926e-05,
      "loss": 1.5811,
      "step": 228000
    },
    {
      "epoch": 0.4499781209110318,
      "grad_norm": 0.1100674718618393,
      "learning_rate": 5.5225335536561305e-05,
      "loss": 1.5721,
      "step": 228032
    },
    {
      "epoch": 0.45004126687242824,
      "grad_norm": 0.11524344235658646,
      "learning_rate": 5.5218995312196606e-05,
      "loss": 1.5699,
      "step": 228064
    },
    {
      "epoch": 0.4501044128338247,
      "grad_norm": 0.11840048432350159,
      "learning_rate": 5.521265508783192e-05,
      "loss": 1.5662,
      "step": 228096
    },
    {
      "epoch": 0.4501675587952211,
      "grad_norm": 0.11256358027458191,
      "learning_rate": 5.5206314863467235e-05,
      "loss": 1.5577,
      "step": 228128
    },
    {
      "epoch": 0.45023070475661753,
      "grad_norm": 0.12191502004861832,
      "learning_rate": 5.519997463910255e-05,
      "loss": 1.5737,
      "step": 228160
    },
    {
      "epoch": 0.450293850718014,
      "grad_norm": 0.11852116882801056,
      "learning_rate": 5.519363441473785e-05,
      "loss": 1.5714,
      "step": 228192
    },
    {
      "epoch": 0.4503569966794104,
      "grad_norm": 0.12248721718788147,
      "learning_rate": 5.5187294190373165e-05,
      "loss": 1.5668,
      "step": 228224
    },
    {
      "epoch": 0.4504201426408069,
      "grad_norm": 0.11424346268177032,
      "learning_rate": 5.518095396600847e-05,
      "loss": 1.5646,
      "step": 228256
    },
    {
      "epoch": 0.4504832886022033,
      "grad_norm": 0.11144455522298813,
      "learning_rate": 5.517461374164379e-05,
      "loss": 1.5616,
      "step": 228288
    },
    {
      "epoch": 0.4505464345635997,
      "grad_norm": 0.11450143158435822,
      "learning_rate": 5.516827351727909e-05,
      "loss": 1.5719,
      "step": 228320
    },
    {
      "epoch": 0.45060958052499617,
      "grad_norm": 0.11503936350345612,
      "learning_rate": 5.51619332929144e-05,
      "loss": 1.5632,
      "step": 228352
    },
    {
      "epoch": 0.4506727264863926,
      "grad_norm": 0.11710304766893387,
      "learning_rate": 5.515559306854972e-05,
      "loss": 1.577,
      "step": 228384
    },
    {
      "epoch": 0.45073587244778907,
      "grad_norm": 0.1098899394273758,
      "learning_rate": 5.5149252844185025e-05,
      "loss": 1.5784,
      "step": 228416
    },
    {
      "epoch": 0.4507990184091855,
      "grad_norm": 0.10603275895118713,
      "learning_rate": 5.514291261982034e-05,
      "loss": 1.5683,
      "step": 228448
    },
    {
      "epoch": 0.45086216437058196,
      "grad_norm": 0.1132390946149826,
      "learning_rate": 5.513657239545564e-05,
      "loss": 1.5684,
      "step": 228480
    },
    {
      "epoch": 0.45092531033197836,
      "grad_norm": 0.11591226607561111,
      "learning_rate": 5.5130232171090955e-05,
      "loss": 1.5903,
      "step": 228512
    },
    {
      "epoch": 0.4509884562933748,
      "grad_norm": 0.10810161381959915,
      "learning_rate": 5.512389194672627e-05,
      "loss": 1.5749,
      "step": 228544
    },
    {
      "epoch": 0.45105160225477126,
      "grad_norm": 0.12791648507118225,
      "learning_rate": 5.5117551722361585e-05,
      "loss": 1.5773,
      "step": 228576
    },
    {
      "epoch": 0.4511147482161677,
      "grad_norm": 0.112993985414505,
      "learning_rate": 5.5111211497996886e-05,
      "loss": 1.5816,
      "step": 228608
    },
    {
      "epoch": 0.45117789417756415,
      "grad_norm": 0.11797860264778137,
      "learning_rate": 5.51048712736322e-05,
      "loss": 1.5708,
      "step": 228640
    },
    {
      "epoch": 0.4512410401389606,
      "grad_norm": 0.11705265194177628,
      "learning_rate": 5.509853104926751e-05,
      "loss": 1.5576,
      "step": 228672
    },
    {
      "epoch": 0.451304186100357,
      "grad_norm": 0.11175525188446045,
      "learning_rate": 5.509219082490282e-05,
      "loss": 1.5717,
      "step": 228704
    },
    {
      "epoch": 0.45136733206175345,
      "grad_norm": 0.11266481131315231,
      "learning_rate": 5.5085850600538124e-05,
      "loss": 1.5624,
      "step": 228736
    },
    {
      "epoch": 0.4514304780231499,
      "grad_norm": 0.11547210067510605,
      "learning_rate": 5.507951037617344e-05,
      "loss": 1.5707,
      "step": 228768
    },
    {
      "epoch": 0.45149362398454634,
      "grad_norm": 0.12393493950366974,
      "learning_rate": 5.507317015180875e-05,
      "loss": 1.5673,
      "step": 228800
    },
    {
      "epoch": 0.4515567699459428,
      "grad_norm": 0.11057010293006897,
      "learning_rate": 5.506682992744406e-05,
      "loss": 1.5623,
      "step": 228832
    },
    {
      "epoch": 0.45161991590733924,
      "grad_norm": 0.10849099606275558,
      "learning_rate": 5.506048970307937e-05,
      "loss": 1.5706,
      "step": 228864
    },
    {
      "epoch": 0.45168306186873564,
      "grad_norm": 0.11320403218269348,
      "learning_rate": 5.5054149478714676e-05,
      "loss": 1.5733,
      "step": 228896
    },
    {
      "epoch": 0.4517462078301321,
      "grad_norm": 0.11227769404649734,
      "learning_rate": 5.504780925434999e-05,
      "loss": 1.57,
      "step": 228928
    },
    {
      "epoch": 0.45180935379152853,
      "grad_norm": 0.10526099056005478,
      "learning_rate": 5.5041469029985305e-05,
      "loss": 1.5743,
      "step": 228960
    },
    {
      "epoch": 0.451872499752925,
      "grad_norm": 0.1137545257806778,
      "learning_rate": 5.5035128805620606e-05,
      "loss": 1.5719,
      "step": 228992
    },
    {
      "epoch": 0.45193564571432143,
      "grad_norm": 0.11737561225891113,
      "learning_rate": 5.502878858125592e-05,
      "loss": 1.5638,
      "step": 229024
    },
    {
      "epoch": 0.4519987916757179,
      "grad_norm": 0.11427541822195053,
      "learning_rate": 5.5022448356891235e-05,
      "loss": 1.5721,
      "step": 229056
    },
    {
      "epoch": 0.4520619376371143,
      "grad_norm": 0.10818526893854141,
      "learning_rate": 5.501610813252654e-05,
      "loss": 1.5639,
      "step": 229088
    },
    {
      "epoch": 0.4521250835985107,
      "grad_norm": 0.1136135458946228,
      "learning_rate": 5.500976790816186e-05,
      "loss": 1.5714,
      "step": 229120
    },
    {
      "epoch": 0.4521882295599072,
      "grad_norm": 0.10835583508014679,
      "learning_rate": 5.500342768379716e-05,
      "loss": 1.5579,
      "step": 229152
    },
    {
      "epoch": 0.4522513755213036,
      "grad_norm": 0.10844231396913528,
      "learning_rate": 5.499708745943247e-05,
      "loss": 1.5799,
      "step": 229184
    },
    {
      "epoch": 0.45231452148270007,
      "grad_norm": 0.11937616765499115,
      "learning_rate": 5.499074723506779e-05,
      "loss": 1.562,
      "step": 229216
    },
    {
      "epoch": 0.4523776674440965,
      "grad_norm": 0.11806692183017731,
      "learning_rate": 5.4984407010703096e-05,
      "loss": 1.5561,
      "step": 229248
    },
    {
      "epoch": 0.45244081340549297,
      "grad_norm": 0.1146462932229042,
      "learning_rate": 5.49780667863384e-05,
      "loss": 1.5676,
      "step": 229280
    },
    {
      "epoch": 0.45250395936688936,
      "grad_norm": 0.11114591360092163,
      "learning_rate": 5.497172656197371e-05,
      "loss": 1.5774,
      "step": 229312
    },
    {
      "epoch": 0.4525671053282858,
      "grad_norm": 0.11399415135383606,
      "learning_rate": 5.4965386337609026e-05,
      "loss": 1.5711,
      "step": 229344
    },
    {
      "epoch": 0.45263025128968226,
      "grad_norm": 0.11704014241695404,
      "learning_rate": 5.495904611324434e-05,
      "loss": 1.5633,
      "step": 229376
    }
  ],
  "logging_steps": 32,
  "max_steps": 506762,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 4096,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5.505181661216427e+19,
  "train_batch_size": 15,
  "trial_name": null,
  "trial_params": null
}