{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 14702,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006802142674942607,
      "grad_norm": 0.24588888883590698,
      "learning_rate": 2.445652173913044e-07,
      "loss": 0.7888,
      "step": 10
    },
    {
      "epoch": 0.0013604285349885215,
      "grad_norm": 0.2492022067308426,
      "learning_rate": 5.16304347826087e-07,
      "loss": 0.7899,
      "step": 20
    },
    {
      "epoch": 0.002040642802482782,
      "grad_norm": 0.24171292781829834,
      "learning_rate": 7.880434782608697e-07,
      "loss": 0.7777,
      "step": 30
    },
    {
      "epoch": 0.002720857069977043,
      "grad_norm": 0.24095626175403595,
      "learning_rate": 1.0597826086956523e-06,
      "loss": 0.7771,
      "step": 40
    },
    {
      "epoch": 0.0034010713374713034,
      "grad_norm": 0.2470710128545761,
      "learning_rate": 1.3315217391304349e-06,
      "loss": 0.7868,
      "step": 50
    },
    {
      "epoch": 0.004081285604965564,
      "grad_norm": 0.23980823159217834,
      "learning_rate": 1.6032608695652175e-06,
      "loss": 0.7888,
      "step": 60
    },
    {
      "epoch": 0.004761499872459825,
      "grad_norm": 0.24573369324207306,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 0.7713,
      "step": 70
    },
    {
      "epoch": 0.005441714139954086,
      "grad_norm": 0.2483571171760559,
      "learning_rate": 2.146739130434783e-06,
      "loss": 0.7766,
      "step": 80
    },
    {
      "epoch": 0.006121928407448346,
      "grad_norm": 0.24448741972446442,
      "learning_rate": 2.4184782608695654e-06,
      "loss": 0.7937,
      "step": 90
    },
    {
      "epoch": 0.006802142674942607,
      "grad_norm": 0.2447367161512375,
      "learning_rate": 2.6902173913043476e-06,
      "loss": 0.7786,
      "step": 100
    },
    {
      "epoch": 0.007482356942436868,
      "grad_norm": 0.24462351202964783,
      "learning_rate": 2.9619565217391306e-06,
      "loss": 0.7701,
      "step": 110
    },
    {
      "epoch": 0.008162571209931128,
      "grad_norm": 0.23651371896266937,
      "learning_rate": 3.233695652173913e-06,
      "loss": 0.7855,
      "step": 120
    },
    {
      "epoch": 0.00884278547742539,
      "grad_norm": 0.25307324528694153,
      "learning_rate": 3.5054347826086958e-06,
      "loss": 0.776,
      "step": 130
    },
    {
      "epoch": 0.00952299974491965,
      "grad_norm": 0.24043457210063934,
      "learning_rate": 3.7771739130434788e-06,
      "loss": 0.7789,
      "step": 140
    },
    {
      "epoch": 0.01020321401241391,
      "grad_norm": 0.2457265853881836,
      "learning_rate": 4.048913043478261e-06,
      "loss": 0.7744,
      "step": 150
    },
    {
      "epoch": 0.010883428279908172,
      "grad_norm": 0.24025650322437286,
      "learning_rate": 4.320652173913044e-06,
      "loss": 0.7749,
      "step": 160
    },
    {
      "epoch": 0.011563642547402431,
      "grad_norm": 0.24393519759178162,
      "learning_rate": 4.5923913043478265e-06,
      "loss": 0.779,
      "step": 170
    },
    {
      "epoch": 0.012243856814896693,
      "grad_norm": 0.24013830721378326,
      "learning_rate": 4.864130434782609e-06,
      "loss": 0.7802,
      "step": 180
    },
    {
      "epoch": 0.012924071082390954,
      "grad_norm": 0.24367555975914001,
      "learning_rate": 5.135869565217392e-06,
      "loss": 0.7831,
      "step": 190
    },
    {
      "epoch": 0.013604285349885213,
      "grad_norm": 0.24819183349609375,
      "learning_rate": 5.407608695652174e-06,
      "loss": 0.7996,
      "step": 200
    },
    {
      "epoch": 0.014284499617379475,
      "grad_norm": 0.24860823154449463,
      "learning_rate": 5.679347826086957e-06,
      "loss": 0.799,
      "step": 210
    },
    {
      "epoch": 0.014964713884873736,
      "grad_norm": 0.2384982705116272,
      "learning_rate": 5.95108695652174e-06,
      "loss": 0.7795,
      "step": 220
    },
    {
      "epoch": 0.015644928152367996,
      "grad_norm": 0.25637149810791016,
      "learning_rate": 6.222826086956523e-06,
      "loss": 0.7843,
      "step": 230
    },
    {
      "epoch": 0.016325142419862257,
      "grad_norm": 0.24896763265132904,
      "learning_rate": 6.4945652173913055e-06,
      "loss": 0.7863,
      "step": 240
    },
    {
      "epoch": 0.017005356687356518,
      "grad_norm": 0.24670195579528809,
      "learning_rate": 6.766304347826087e-06,
      "loss": 0.7873,
      "step": 250
    },
    {
      "epoch": 0.01768557095485078,
      "grad_norm": 0.2428620159626007,
      "learning_rate": 7.03804347826087e-06,
      "loss": 0.7837,
      "step": 260
    },
    {
      "epoch": 0.018365785222345037,
      "grad_norm": 0.25439387559890747,
      "learning_rate": 7.309782608695652e-06,
      "loss": 0.7803,
      "step": 270
    },
    {
      "epoch": 0.0190459994898393,
      "grad_norm": 0.2559449076652527,
      "learning_rate": 7.581521739130435e-06,
      "loss": 0.7955,
      "step": 280
    },
    {
      "epoch": 0.01972621375733356,
      "grad_norm": 0.26467257738113403,
      "learning_rate": 7.853260869565218e-06,
      "loss": 0.7957,
      "step": 290
    },
    {
      "epoch": 0.02040642802482782,
      "grad_norm": 0.246280238032341,
      "learning_rate": 8.125000000000001e-06,
      "loss": 0.8032,
      "step": 300
    },
    {
      "epoch": 0.021086642292322082,
      "grad_norm": 0.2527225911617279,
      "learning_rate": 8.396739130434784e-06,
      "loss": 0.7802,
      "step": 310
    },
    {
      "epoch": 0.021766856559816344,
      "grad_norm": 0.2507527768611908,
      "learning_rate": 8.668478260869566e-06,
      "loss": 0.7968,
      "step": 320
    },
    {
      "epoch": 0.0224470708273106,
      "grad_norm": 0.25376781821250916,
      "learning_rate": 8.940217391304349e-06,
      "loss": 0.7903,
      "step": 330
    },
    {
      "epoch": 0.023127285094804863,
      "grad_norm": 0.24194824695587158,
      "learning_rate": 9.211956521739131e-06,
      "loss": 0.7857,
      "step": 340
    },
    {
      "epoch": 0.023807499362299124,
      "grad_norm": 0.25317472219467163,
      "learning_rate": 9.483695652173914e-06,
      "loss": 0.7788,
      "step": 350
    },
    {
      "epoch": 0.024487713629793385,
      "grad_norm": 0.26361075043678284,
      "learning_rate": 9.755434782608696e-06,
      "loss": 0.7841,
      "step": 360
    },
    {
      "epoch": 0.025167927897287647,
      "grad_norm": 0.2593362629413605,
      "learning_rate": 1.0027173913043479e-05,
      "loss": 0.7865,
      "step": 370
    },
    {
      "epoch": 0.025848142164781908,
      "grad_norm": 0.25769373774528503,
      "learning_rate": 1.0298913043478262e-05,
      "loss": 0.7821,
      "step": 380
    },
    {
      "epoch": 0.026528356432276166,
      "grad_norm": 0.25864651799201965,
      "learning_rate": 1.0570652173913046e-05,
      "loss": 0.7804,
      "step": 390
    },
    {
      "epoch": 0.027208570699770427,
      "grad_norm": 0.25476300716400146,
      "learning_rate": 1.0842391304347829e-05,
      "loss": 0.7827,
      "step": 400
    },
    {
      "epoch": 0.02788878496726469,
      "grad_norm": 0.2638179659843445,
      "learning_rate": 1.1114130434782611e-05,
      "loss": 0.7969,
      "step": 410
    },
    {
      "epoch": 0.02856899923475895,
      "grad_norm": 0.25068527460098267,
      "learning_rate": 1.1385869565217392e-05,
      "loss": 0.7707,
      "step": 420
    },
    {
      "epoch": 0.02924921350225321,
      "grad_norm": 0.2590930759906769,
      "learning_rate": 1.1657608695652175e-05,
      "loss": 0.7878,
      "step": 430
    },
    {
      "epoch": 0.029929427769747472,
      "grad_norm": 0.2491404116153717,
      "learning_rate": 1.1929347826086957e-05,
      "loss": 0.7877,
      "step": 440
    },
    {
      "epoch": 0.03060964203724173,
      "grad_norm": 0.2550014555454254,
      "learning_rate": 1.220108695652174e-05,
      "loss": 0.7755,
      "step": 450
    },
    {
      "epoch": 0.03128985630473599,
      "grad_norm": 0.25757303833961487,
      "learning_rate": 1.2472826086956522e-05,
      "loss": 0.7965,
      "step": 460
    },
    {
      "epoch": 0.03197007057223025,
      "grad_norm": 0.24262848496437073,
      "learning_rate": 1.2744565217391305e-05,
      "loss": 0.7695,
      "step": 470
    },
    {
      "epoch": 0.032650284839724514,
      "grad_norm": 0.2601710557937622,
      "learning_rate": 1.3016304347826088e-05,
      "loss": 0.795,
      "step": 480
    },
    {
      "epoch": 0.033330499107218775,
      "grad_norm": 0.2602006196975708,
      "learning_rate": 1.328804347826087e-05,
      "loss": 0.7838,
      "step": 490
    },
    {
      "epoch": 0.034010713374713036,
      "grad_norm": 0.2601083517074585,
      "learning_rate": 1.3559782608695653e-05,
      "loss": 0.7887,
      "step": 500
    },
    {
      "epoch": 0.0346909276422073,
      "grad_norm": 0.403003454208374,
      "learning_rate": 1.3831521739130435e-05,
      "loss": 0.7821,
      "step": 510
    },
    {
      "epoch": 0.03537114190970156,
      "grad_norm": 0.2679433822631836,
      "learning_rate": 1.4103260869565218e-05,
      "loss": 0.7813,
      "step": 520
    },
    {
      "epoch": 0.03605135617719581,
      "grad_norm": 0.25827887654304504,
      "learning_rate": 1.4375e-05,
      "loss": 0.7897,
      "step": 530
    },
    {
      "epoch": 0.036731570444690075,
      "grad_norm": 0.26387709379196167,
      "learning_rate": 1.4646739130434785e-05,
      "loss": 0.7902,
      "step": 540
    },
    {
      "epoch": 0.037411784712184336,
      "grad_norm": 0.25550195574760437,
      "learning_rate": 1.4918478260869567e-05,
      "loss": 0.7853,
      "step": 550
    },
    {
      "epoch": 0.0380919989796786,
      "grad_norm": 0.2730271518230438,
      "learning_rate": 1.519021739130435e-05,
      "loss": 0.8054,
      "step": 560
    },
    {
      "epoch": 0.03877221324717286,
      "grad_norm": 0.27456948161125183,
      "learning_rate": 1.546195652173913e-05,
      "loss": 0.7876,
      "step": 570
    },
    {
      "epoch": 0.03945242751466712,
      "grad_norm": 0.27266284823417664,
      "learning_rate": 1.5733695652173913e-05,
      "loss": 0.792,
      "step": 580
    },
    {
      "epoch": 0.04013264178216138,
      "grad_norm": 0.2583005726337433,
      "learning_rate": 1.6005434782608696e-05,
      "loss": 0.7855,
      "step": 590
    },
    {
      "epoch": 0.04081285604965564,
      "grad_norm": 0.2712930738925934,
      "learning_rate": 1.6277173913043482e-05,
      "loss": 0.7986,
      "step": 600
    },
    {
      "epoch": 0.041493070317149904,
      "grad_norm": 0.2771306335926056,
      "learning_rate": 1.6548913043478265e-05,
      "loss": 0.7951,
      "step": 610
    },
    {
      "epoch": 0.042173284584644165,
      "grad_norm": 0.2726171016693115,
      "learning_rate": 1.6820652173913047e-05,
      "loss": 0.7881,
      "step": 620
    },
    {
      "epoch": 0.042853498852138426,
      "grad_norm": 0.26444506645202637,
      "learning_rate": 1.709239130434783e-05,
      "loss": 0.7987,
      "step": 630
    },
    {
      "epoch": 0.04353371311963269,
      "grad_norm": 0.28685182332992554,
      "learning_rate": 1.7364130434782612e-05,
      "loss": 0.8071,
      "step": 640
    },
    {
      "epoch": 0.04421392738712694,
      "grad_norm": 0.2572171092033386,
      "learning_rate": 1.763586956521739e-05,
      "loss": 0.785,
      "step": 650
    },
    {
      "epoch": 0.0448941416546212,
      "grad_norm": 0.2599855065345764,
      "learning_rate": 1.7907608695652174e-05,
      "loss": 0.7958,
      "step": 660
    },
    {
      "epoch": 0.045574355922115464,
      "grad_norm": 0.2666078507900238,
      "learning_rate": 1.8179347826086957e-05,
      "loss": 0.7855,
      "step": 670
    },
    {
      "epoch": 0.046254570189609726,
      "grad_norm": 0.2504750192165375,
      "learning_rate": 1.845108695652174e-05,
      "loss": 0.7808,
      "step": 680
    },
    {
      "epoch": 0.04693478445710399,
      "grad_norm": 0.265756219625473,
      "learning_rate": 1.8722826086956522e-05,
      "loss": 0.8039,
      "step": 690
    },
    {
      "epoch": 0.04761499872459825,
      "grad_norm": 0.24935463070869446,
      "learning_rate": 1.8994565217391304e-05,
      "loss": 0.8042,
      "step": 700
    },
    {
      "epoch": 0.04829521299209251,
      "grad_norm": 0.2691420018672943,
      "learning_rate": 1.9266304347826087e-05,
      "loss": 0.79,
      "step": 710
    },
    {
      "epoch": 0.04897542725958677,
      "grad_norm": 0.2798922657966614,
      "learning_rate": 1.953804347826087e-05,
      "loss": 0.7957,
      "step": 720
    },
    {
      "epoch": 0.04965564152708103,
      "grad_norm": 0.2665148973464966,
      "learning_rate": 1.9809782608695652e-05,
      "loss": 0.7775,
      "step": 730
    },
    {
      "epoch": 0.05033585579457529,
      "grad_norm": 0.2752082347869873,
      "learning_rate": 1.999999772297307e-05,
      "loss": 0.7848,
      "step": 740
    },
    {
      "epoch": 0.051016070062069555,
      "grad_norm": 0.25373271107673645,
      "learning_rate": 1.9999957242523135e-05,
      "loss": 0.7894,
      "step": 750
    },
    {
      "epoch": 0.051696284329563816,
      "grad_norm": 0.27112025022506714,
      "learning_rate": 1.9999866161710487e-05,
      "loss": 0.7857,
      "step": 760
    },
    {
      "epoch": 0.05237649859705807,
      "grad_norm": 0.255633145570755,
      "learning_rate": 1.999972448099601e-05,
      "loss": 0.7895,
      "step": 770
    },
    {
      "epoch": 0.05305671286455233,
      "grad_norm": 0.25357896089553833,
      "learning_rate": 1.9999532201096614e-05,
      "loss": 0.7817,
      "step": 780
    },
    {
      "epoch": 0.05373692713204659,
      "grad_norm": 0.2669251263141632,
      "learning_rate": 1.999928932298524e-05,
      "loss": 0.8089,
      "step": 790
    },
    {
      "epoch": 0.054417141399540854,
      "grad_norm": 0.2620493173599243,
      "learning_rate": 1.9998995847890876e-05,
      "loss": 0.7836,
      "step": 800
    },
    {
      "epoch": 0.055097355667035115,
      "grad_norm": 0.25441357493400574,
      "learning_rate": 1.9998651777298512e-05,
      "loss": 0.7874,
      "step": 810
    },
    {
      "epoch": 0.05577756993452938,
      "grad_norm": 0.27190372347831726,
      "learning_rate": 1.9998257112949167e-05,
      "loss": 0.7841,
      "step": 820
    },
    {
      "epoch": 0.05645778420202364,
      "grad_norm": 0.26984357833862305,
      "learning_rate": 1.9997811856839875e-05,
      "loss": 0.7866,
      "step": 830
    },
    {
      "epoch": 0.0571379984695179,
      "grad_norm": 0.26123085618019104,
      "learning_rate": 1.9997316011223647e-05,
      "loss": 0.7975,
      "step": 840
    },
    {
      "epoch": 0.05781821273701216,
      "grad_norm": 0.2544929087162018,
      "learning_rate": 1.9996769578609494e-05,
      "loss": 0.7834,
      "step": 850
    },
    {
      "epoch": 0.05849842700450642,
      "grad_norm": 0.2855475842952728,
      "learning_rate": 1.99961725617624e-05,
      "loss": 0.795,
      "step": 860
    },
    {
      "epoch": 0.05917864127200068,
      "grad_norm": 0.26833781599998474,
      "learning_rate": 1.99955249637033e-05,
      "loss": 0.7934,
      "step": 870
    },
    {
      "epoch": 0.059858855539494944,
      "grad_norm": 0.2639079689979553,
      "learning_rate": 1.9994826787709082e-05,
      "loss": 0.7917,
      "step": 880
    },
    {
      "epoch": 0.0605390698069892,
      "grad_norm": 0.2570595443248749,
      "learning_rate": 1.9994078037312552e-05,
      "loss": 0.8031,
      "step": 890
    },
    {
      "epoch": 0.06121928407448346,
      "grad_norm": 0.26510873436927795,
      "learning_rate": 1.9993278716302435e-05,
      "loss": 0.7879,
      "step": 900
    },
    {
      "epoch": 0.06189949834197772,
      "grad_norm": 0.256365567445755,
      "learning_rate": 1.999242882872334e-05,
      "loss": 0.7806,
      "step": 910
    },
    {
      "epoch": 0.06257971260947198,
      "grad_norm": 0.25432026386260986,
      "learning_rate": 1.9991528378875747e-05,
      "loss": 0.79,
      "step": 920
    },
    {
      "epoch": 0.06325992687696624,
      "grad_norm": 0.25278228521347046,
      "learning_rate": 1.9990577371315983e-05,
      "loss": 0.7967,
      "step": 930
    },
    {
      "epoch": 0.0639401411444605,
      "grad_norm": 0.2596864104270935,
      "learning_rate": 1.99895758108562e-05,
      "loss": 0.786,
      "step": 940
    },
    {
      "epoch": 0.06462035541195477,
      "grad_norm": 0.25254207849502563,
      "learning_rate": 1.998852370256436e-05,
      "loss": 0.7775,
      "step": 950
    },
    {
      "epoch": 0.06530056967944903,
      "grad_norm": 0.2752808630466461,
      "learning_rate": 1.9987421051764184e-05,
      "loss": 0.7917,
      "step": 960
    },
    {
      "epoch": 0.06598078394694329,
      "grad_norm": 0.2659485638141632,
      "learning_rate": 1.998626786403515e-05,
      "loss": 0.7712,
      "step": 970
    },
    {
      "epoch": 0.06666099821443755,
      "grad_norm": 0.26159390807151794,
      "learning_rate": 1.998506414521246e-05,
      "loss": 0.8003,
      "step": 980
    },
    {
      "epoch": 0.06734121248193181,
      "grad_norm": 0.25408387184143066,
      "learning_rate": 1.9983809901386995e-05,
      "loss": 0.7882,
      "step": 990
    },
    {
      "epoch": 0.06802142674942607,
      "grad_norm": 0.2492450475692749,
      "learning_rate": 1.9982505138905303e-05,
      "loss": 0.8006,
      "step": 1000
    },
    {
      "epoch": 0.06870164101692033,
      "grad_norm": 0.26037201285362244,
      "learning_rate": 1.998114986436956e-05,
      "loss": 0.7897,
      "step": 1010
    },
    {
      "epoch": 0.0693818552844146,
      "grad_norm": 0.25039172172546387,
      "learning_rate": 1.997974408463753e-05,
      "loss": 0.7966,
      "step": 1020
    },
    {
      "epoch": 0.07006206955190886,
      "grad_norm": 0.25802841782569885,
      "learning_rate": 1.9978287806822544e-05,
      "loss": 0.7908,
      "step": 1030
    },
    {
      "epoch": 0.07074228381940312,
      "grad_norm": 0.25341302156448364,
      "learning_rate": 1.9976781038293444e-05,
      "loss": 0.7826,
      "step": 1040
    },
    {
      "epoch": 0.07142249808689738,
      "grad_norm": 0.2570743262767792,
      "learning_rate": 1.9975223786674576e-05,
      "loss": 0.7891,
      "step": 1050
    },
    {
      "epoch": 0.07210271235439163,
      "grad_norm": 0.25724539160728455,
      "learning_rate": 1.997361605984572e-05,
      "loss": 0.7906,
      "step": 1060
    },
    {
      "epoch": 0.07278292662188589,
      "grad_norm": 0.2633865773677826,
      "learning_rate": 1.9971957865942064e-05,
      "loss": 0.7928,
      "step": 1070
    },
    {
      "epoch": 0.07346314088938015,
      "grad_norm": 0.2706665098667145,
      "learning_rate": 1.9970249213354167e-05,
      "loss": 0.7876,
      "step": 1080
    },
    {
      "epoch": 0.07414335515687441,
      "grad_norm": 0.2597387135028839,
      "learning_rate": 1.996849011072791e-05,
      "loss": 0.7922,
      "step": 1090
    },
    {
      "epoch": 0.07482356942436867,
      "grad_norm": 0.2626490592956543,
      "learning_rate": 1.9966680566964456e-05,
      "loss": 0.7931,
      "step": 1100
    },
    {
      "epoch": 0.07550378369186293,
      "grad_norm": 0.26137858629226685,
      "learning_rate": 1.99648205912202e-05,
      "loss": 0.7881,
      "step": 1110
    },
    {
      "epoch": 0.0761839979593572,
      "grad_norm": 0.2648197412490845,
      "learning_rate": 1.9962910192906725e-05,
      "loss": 0.787,
      "step": 1120
    },
    {
      "epoch": 0.07686421222685146,
      "grad_norm": 0.25306472182273865,
      "learning_rate": 1.9960949381690757e-05,
      "loss": 0.7925,
      "step": 1130
    },
    {
      "epoch": 0.07754442649434572,
      "grad_norm": 0.2650945782661438,
      "learning_rate": 1.9958938167494118e-05,
      "loss": 0.7926,
      "step": 1140
    },
    {
      "epoch": 0.07822464076183998,
      "grad_norm": 0.24441677331924438,
      "learning_rate": 1.9956876560493668e-05,
      "loss": 0.7934,
      "step": 1150
    },
    {
      "epoch": 0.07890485502933424,
      "grad_norm": 0.25528618693351746,
      "learning_rate": 1.9954764571121254e-05,
      "loss": 0.7961,
      "step": 1160
    },
    {
      "epoch": 0.0795850692968285,
      "grad_norm": 0.2585403025150299,
      "learning_rate": 1.9952602210063664e-05,
      "loss": 0.7906,
      "step": 1170
    },
    {
      "epoch": 0.08026528356432276,
      "grad_norm": 0.26270559430122375,
      "learning_rate": 1.9950389488262576e-05,
      "loss": 0.7874,
      "step": 1180
    },
    {
      "epoch": 0.08094549783181702,
      "grad_norm": 0.259402871131897,
      "learning_rate": 1.9948126416914486e-05,
      "loss": 0.7946,
      "step": 1190
    },
    {
      "epoch": 0.08162571209931128,
      "grad_norm": 0.2569419741630554,
      "learning_rate": 1.9945813007470668e-05,
      "loss": 0.789,
      "step": 1200
    },
    {
      "epoch": 0.08230592636680555,
      "grad_norm": 0.2626119554042816,
      "learning_rate": 1.9943449271637103e-05,
      "loss": 0.7779,
      "step": 1210
    },
    {
      "epoch": 0.08298614063429981,
      "grad_norm": 0.2561872899532318,
      "learning_rate": 1.994103522137444e-05,
      "loss": 0.7798,
      "step": 1220
    },
    {
      "epoch": 0.08366635490179407,
      "grad_norm": 0.24945363402366638,
      "learning_rate": 1.9938570868897907e-05,
      "loss": 0.7823,
      "step": 1230
    },
    {
      "epoch": 0.08434656916928833,
      "grad_norm": 0.2486308068037033,
      "learning_rate": 1.993605622667727e-05,
      "loss": 0.7851,
      "step": 1240
    },
    {
      "epoch": 0.08502678343678259,
      "grad_norm": 0.2569667100906372,
      "learning_rate": 1.9933491307436768e-05,
      "loss": 0.7933,
      "step": 1250
    },
    {
      "epoch": 0.08570699770427685,
      "grad_norm": 0.24834266304969788,
      "learning_rate": 1.9930876124155037e-05,
      "loss": 0.796,
      "step": 1260
    },
    {
      "epoch": 0.08638721197177111,
      "grad_norm": 0.27963775396347046,
      "learning_rate": 1.9928210690065058e-05,
      "loss": 0.7944,
      "step": 1270
    },
    {
      "epoch": 0.08706742623926537,
      "grad_norm": 0.2627074718475342,
      "learning_rate": 1.992549501865408e-05,
      "loss": 0.7995,
      "step": 1280
    },
    {
      "epoch": 0.08774764050675964,
      "grad_norm": 0.25970980525016785,
      "learning_rate": 1.992272912366356e-05,
      "loss": 0.7823,
      "step": 1290
    },
    {
      "epoch": 0.08842785477425388,
      "grad_norm": 0.24787351489067078,
      "learning_rate": 1.9919913019089085e-05,
      "loss": 0.7839,
      "step": 1300
    },
    {
      "epoch": 0.08910806904174814,
      "grad_norm": 0.26264896988868713,
      "learning_rate": 1.991704671918031e-05,
      "loss": 0.7877,
      "step": 1310
    },
    {
      "epoch": 0.0897882833092424,
      "grad_norm": 0.26188844442367554,
      "learning_rate": 1.9914130238440874e-05,
      "loss": 0.7917,
      "step": 1320
    },
    {
      "epoch": 0.09046849757673667,
      "grad_norm": 0.2632116675376892,
      "learning_rate": 1.991116359162834e-05,
      "loss": 0.7889,
      "step": 1330
    },
    {
      "epoch": 0.09114871184423093,
      "grad_norm": 0.2512179911136627,
      "learning_rate": 1.9908146793754117e-05,
      "loss": 0.7885,
      "step": 1340
    },
    {
      "epoch": 0.09182892611172519,
      "grad_norm": 0.26141035556793213,
      "learning_rate": 1.9905079860083373e-05,
      "loss": 0.7858,
      "step": 1350
    },
    {
      "epoch": 0.09250914037921945,
      "grad_norm": 0.26588204503059387,
      "learning_rate": 1.9901962806134967e-05,
      "loss": 0.7848,
      "step": 1360
    },
    {
      "epoch": 0.09318935464671371,
      "grad_norm": 0.2591280937194824,
      "learning_rate": 1.9898795647681378e-05,
      "loss": 0.7953,
      "step": 1370
    },
    {
      "epoch": 0.09386956891420797,
      "grad_norm": 0.2677696645259857,
      "learning_rate": 1.989557840074861e-05,
      "loss": 0.7865,
      "step": 1380
    },
    {
      "epoch": 0.09454978318170223,
      "grad_norm": 0.2503071129322052,
      "learning_rate": 1.9892311081616116e-05,
      "loss": 0.7841,
      "step": 1390
    },
    {
      "epoch": 0.0952299974491965,
      "grad_norm": 0.26428544521331787,
      "learning_rate": 1.9888993706816722e-05,
      "loss": 0.7806,
      "step": 1400
    },
    {
      "epoch": 0.09591021171669076,
      "grad_norm": 0.24584265053272247,
      "learning_rate": 1.9885626293136537e-05,
      "loss": 0.7981,
      "step": 1410
    },
    {
      "epoch": 0.09659042598418502,
      "grad_norm": 0.27497342228889465,
      "learning_rate": 1.988220885761487e-05,
      "loss": 0.782,
      "step": 1420
    },
    {
      "epoch": 0.09727064025167928,
      "grad_norm": 0.2711544632911682,
      "learning_rate": 1.9878741417544138e-05,
      "loss": 0.7778,
      "step": 1430
    },
    {
      "epoch": 0.09795085451917354,
      "grad_norm": 0.2602938711643219,
      "learning_rate": 1.9875223990469797e-05,
      "loss": 0.7902,
      "step": 1440
    },
    {
      "epoch": 0.0986310687866678,
      "grad_norm": 0.2596087157726288,
      "learning_rate": 1.9871656594190222e-05,
      "loss": 0.7893,
      "step": 1450
    },
    {
      "epoch": 0.09931128305416206,
      "grad_norm": 0.2592509984970093,
      "learning_rate": 1.9868039246756652e-05,
      "loss": 0.7861,
      "step": 1460
    },
    {
      "epoch": 0.09999149732165633,
      "grad_norm": 0.2511366307735443,
      "learning_rate": 1.986437196647307e-05,
      "loss": 0.7953,
      "step": 1470
    },
    {
      "epoch": 0.10067171158915059,
      "grad_norm": 0.27958574891090393,
      "learning_rate": 1.986065477189613e-05,
      "loss": 0.7864,
      "step": 1480
    },
    {
      "epoch": 0.10135192585664485,
      "grad_norm": 0.252519428730011,
      "learning_rate": 1.9856887681835045e-05,
      "loss": 0.7883,
      "step": 1490
    },
    {
      "epoch": 0.10203214012413911,
      "grad_norm": 0.26588451862335205,
      "learning_rate": 1.9853070715351517e-05,
      "loss": 0.8007,
      "step": 1500
    },
    {
      "epoch": 0.10271235439163337,
      "grad_norm": 0.2513459026813507,
      "learning_rate": 1.9849203891759615e-05,
      "loss": 0.8079,
      "step": 1510
    },
    {
      "epoch": 0.10339256865912763,
      "grad_norm": 0.25905874371528625,
      "learning_rate": 1.984528723062569e-05,
      "loss": 0.7838,
      "step": 1520
    },
    {
      "epoch": 0.10407278292662188,
      "grad_norm": 0.2662612497806549,
      "learning_rate": 1.9841320751768277e-05,
      "loss": 0.7886,
      "step": 1530
    },
    {
      "epoch": 0.10475299719411614,
      "grad_norm": 0.2671913504600525,
      "learning_rate": 1.983730447525798e-05,
      "loss": 0.7926,
      "step": 1540
    },
    {
      "epoch": 0.1054332114616104,
      "grad_norm": 0.2774171233177185,
      "learning_rate": 1.98332384214174e-05,
      "loss": 0.7952,
      "step": 1550
    },
    {
      "epoch": 0.10611342572910466,
      "grad_norm": 0.25690150260925293,
      "learning_rate": 1.9829122610820998e-05,
      "loss": 0.7731,
      "step": 1560
    },
    {
      "epoch": 0.10679363999659892,
      "grad_norm": 0.26239097118377686,
      "learning_rate": 1.982495706429502e-05,
      "loss": 0.7777,
      "step": 1570
    },
    {
      "epoch": 0.10747385426409319,
      "grad_norm": 0.2668183147907257,
      "learning_rate": 1.9820741802917368e-05,
      "loss": 0.7892,
      "step": 1580
    },
    {
      "epoch": 0.10815406853158745,
      "grad_norm": 0.2633265256881714,
      "learning_rate": 1.981647684801751e-05,
      "loss": 0.8007,
      "step": 1590
    },
    {
      "epoch": 0.10883428279908171,
      "grad_norm": 0.26835504174232483,
      "learning_rate": 1.9812162221176368e-05,
      "loss": 0.7751,
      "step": 1600
    },
    {
      "epoch": 0.10951449706657597,
      "grad_norm": 0.26324793696403503,
      "learning_rate": 1.98077979442262e-05,
      "loss": 0.7892,
      "step": 1610
    },
    {
      "epoch": 0.11019471133407023,
      "grad_norm": 0.2817467451095581,
      "learning_rate": 1.9803384039250498e-05,
      "loss": 0.8094,
      "step": 1620
    },
    {
      "epoch": 0.11087492560156449,
      "grad_norm": 0.26144781708717346,
      "learning_rate": 1.979892052858388e-05,
      "loss": 0.7952,
      "step": 1630
    },
    {
      "epoch": 0.11155513986905875,
      "grad_norm": 0.25879013538360596,
      "learning_rate": 1.979440743481197e-05,
      "loss": 0.7944,
      "step": 1640
    },
    {
      "epoch": 0.11223535413655301,
      "grad_norm": 0.2659333348274231,
      "learning_rate": 1.9789844780771283e-05,
      "loss": 0.7818,
      "step": 1650
    },
    {
      "epoch": 0.11291556840404728,
      "grad_norm": 0.2678092420101166,
      "learning_rate": 1.978523258954911e-05,
      "loss": 0.7914,
      "step": 1660
    },
    {
      "epoch": 0.11359578267154154,
      "grad_norm": 0.26243963837623596,
      "learning_rate": 1.9780570884483404e-05,
      "loss": 0.7984,
      "step": 1670
    },
    {
      "epoch": 0.1142759969390358,
      "grad_norm": 0.2623418867588043,
      "learning_rate": 1.977585968916267e-05,
      "loss": 0.7931,
      "step": 1680
    },
    {
      "epoch": 0.11495621120653006,
      "grad_norm": 0.2606011629104614,
      "learning_rate": 1.9771099027425827e-05,
      "loss": 0.7915,
      "step": 1690
    },
    {
      "epoch": 0.11563642547402432,
      "grad_norm": 0.2516583800315857,
      "learning_rate": 1.9766288923362093e-05,
      "loss": 0.7943,
      "step": 1700
    },
    {
      "epoch": 0.11631663974151858,
      "grad_norm": 0.24944451451301575,
      "learning_rate": 1.9761429401310878e-05,
      "loss": 0.7905,
      "step": 1710
    },
    {
      "epoch": 0.11699685400901284,
      "grad_norm": 0.25479376316070557,
      "learning_rate": 1.9756520485861644e-05,
      "loss": 0.7965,
      "step": 1720
    },
    {
      "epoch": 0.1176770682765071,
      "grad_norm": 0.2506806552410126,
      "learning_rate": 1.9751562201853788e-05,
      "loss": 0.7835,
      "step": 1730
    },
    {
      "epoch": 0.11835728254400137,
      "grad_norm": 0.25013062357902527,
      "learning_rate": 1.9746554574376513e-05,
      "loss": 0.7787,
      "step": 1740
    },
    {
      "epoch": 0.11903749681149563,
      "grad_norm": 0.2511868178844452,
      "learning_rate": 1.9741497628768704e-05,
      "loss": 0.7874,
      "step": 1750
    },
    {
      "epoch": 0.11971771107898989,
      "grad_norm": 0.24284197390079498,
      "learning_rate": 1.97363913906188e-05,
      "loss": 0.7893,
      "step": 1760
    },
    {
      "epoch": 0.12039792534648414,
      "grad_norm": 0.25849926471710205,
      "learning_rate": 1.973123588576466e-05,
      "loss": 0.7953,
      "step": 1770
    },
    {
      "epoch": 0.1210781396139784,
      "grad_norm": 0.24535062909126282,
      "learning_rate": 1.9726031140293442e-05,
      "loss": 0.7783,
      "step": 1780
    },
    {
      "epoch": 0.12175835388147266,
      "grad_norm": 0.2653627097606659,
      "learning_rate": 1.9720777180541454e-05,
      "loss": 0.8002,
      "step": 1790
    },
    {
      "epoch": 0.12243856814896692,
      "grad_norm": 0.2720264196395874,
      "learning_rate": 1.9715474033094035e-05,
      "loss": 0.793,
      "step": 1800
    },
    {
      "epoch": 0.12311878241646118,
      "grad_norm": 0.2735433280467987,
      "learning_rate": 1.9710121724785422e-05,
      "loss": 0.7875,
      "step": 1810
    },
    {
      "epoch": 0.12379899668395544,
      "grad_norm": 0.2591589391231537,
      "learning_rate": 1.9704720282698607e-05,
      "loss": 0.7735,
      "step": 1820
    },
    {
      "epoch": 0.1244792109514497,
      "grad_norm": 0.2804940938949585,
      "learning_rate": 1.9699269734165195e-05,
      "loss": 0.7799,
      "step": 1830
    },
    {
      "epoch": 0.12515942521894396,
      "grad_norm": 0.2527720630168915,
      "learning_rate": 1.969377010676528e-05,
      "loss": 0.7986,
      "step": 1840
    },
    {
      "epoch": 0.12583963948643823,
      "grad_norm": 0.26848408579826355,
      "learning_rate": 1.9688221428327295e-05,
      "loss": 0.7955,
      "step": 1850
    },
    {
      "epoch": 0.1265198537539325,
      "grad_norm": 0.25075897574424744,
      "learning_rate": 1.9682623726927873e-05,
      "loss": 0.7967,
      "step": 1860
    },
    {
      "epoch": 0.12720006802142675,
      "grad_norm": 0.24642987549304962,
      "learning_rate": 1.9676977030891707e-05,
      "loss": 0.7908,
      "step": 1870
    },
    {
      "epoch": 0.127880282288921,
      "grad_norm": 0.2501721680164337,
      "learning_rate": 1.9671281368791407e-05,
      "loss": 0.7788,
      "step": 1880
    },
    {
      "epoch": 0.12856049655641527,
      "grad_norm": 0.2666166424751282,
      "learning_rate": 1.9665536769447357e-05,
      "loss": 0.7993,
      "step": 1890
    },
    {
      "epoch": 0.12924071082390953,
      "grad_norm": 0.26103368401527405,
      "learning_rate": 1.9659743261927553e-05,
      "loss": 0.8011,
      "step": 1900
    },
    {
      "epoch": 0.1299209250914038,
      "grad_norm": 0.2501685917377472,
      "learning_rate": 1.9653900875547486e-05,
      "loss": 0.7938,
      "step": 1910
    },
    {
      "epoch": 0.13060113935889806,
      "grad_norm": 0.2564981281757355,
      "learning_rate": 1.9648009639869966e-05,
      "loss": 0.7986,
      "step": 1920
    },
    {
      "epoch": 0.13128135362639232,
      "grad_norm": 0.24962271749973297,
      "learning_rate": 1.9642069584704985e-05,
      "loss": 0.7944,
      "step": 1930
    },
    {
      "epoch": 0.13196156789388658,
      "grad_norm": 0.26453691720962524,
      "learning_rate": 1.9636080740109575e-05,
      "loss": 0.8025,
      "step": 1940
    },
    {
      "epoch": 0.13264178216138084,
      "grad_norm": 0.2584055960178375,
      "learning_rate": 1.9630043136387623e-05,
      "loss": 0.7899,
      "step": 1950
    },
    {
      "epoch": 0.1333219964288751,
      "grad_norm": 0.24752554297447205,
      "learning_rate": 1.9623956804089766e-05,
      "loss": 0.7723,
      "step": 1960
    },
    {
      "epoch": 0.13400221069636936,
      "grad_norm": 0.2683132290840149,
      "learning_rate": 1.961782177401319e-05,
      "loss": 0.7847,
      "step": 1970
    },
    {
      "epoch": 0.13468242496386362,
      "grad_norm": 0.26384273171424866,
      "learning_rate": 1.9611638077201508e-05,
      "loss": 0.807,
      "step": 1980
    },
    {
      "epoch": 0.13536263923135788,
      "grad_norm": 0.2578035891056061,
      "learning_rate": 1.960540574494458e-05,
      "loss": 0.7938,
      "step": 1990
    },
    {
      "epoch": 0.13604285349885215,
      "grad_norm": 0.26255810260772705,
      "learning_rate": 1.9599124808778363e-05,
      "loss": 0.7994,
      "step": 2000
    },
    {
      "epoch": 0.1367230677663464,
      "grad_norm": 0.2509863078594208,
      "learning_rate": 1.959279530048477e-05,
      "loss": 0.7875,
      "step": 2010
    },
    {
      "epoch": 0.13740328203384067,
      "grad_norm": 52.004241943359375,
      "learning_rate": 1.958641725209147e-05,
      "loss": 0.7879,
      "step": 2020
    },
    {
      "epoch": 0.13808349630133493,
      "grad_norm": 0.2877405285835266,
      "learning_rate": 1.9579990695871768e-05,
      "loss": 0.7983,
      "step": 2030
    },
    {
      "epoch": 0.1387637105688292,
      "grad_norm": 0.25437185168266296,
      "learning_rate": 1.9573515664344396e-05,
      "loss": 0.7964,
      "step": 2040
    },
    {
      "epoch": 0.13944392483632345,
      "grad_norm": 0.24121397733688354,
      "learning_rate": 1.9566992190273394e-05,
      "loss": 0.7774,
      "step": 2050
    },
    {
      "epoch": 0.1401241391038177,
      "grad_norm": 0.2614583969116211,
      "learning_rate": 1.9560420306667924e-05,
      "loss": 0.7907,
      "step": 2060
    },
    {
      "epoch": 0.14080435337131197,
      "grad_norm": 0.25397762656211853,
      "learning_rate": 1.9553800046782093e-05,
      "loss": 0.7704,
      "step": 2070
    },
    {
      "epoch": 0.14148456763880624,
      "grad_norm": 0.2523746192455292,
      "learning_rate": 1.95471314441148e-05,
      "loss": 0.78,
      "step": 2080
    },
    {
      "epoch": 0.1421647819063005,
      "grad_norm": 0.2488732635974884,
      "learning_rate": 1.954041453240956e-05,
      "loss": 0.7877,
      "step": 2090
    },
    {
      "epoch": 0.14284499617379476,
      "grad_norm": 0.24965067207813263,
      "learning_rate": 1.9533649345654338e-05,
      "loss": 0.7849,
      "step": 2100
    },
    {
      "epoch": 0.14352521044128902,
      "grad_norm": 0.25302186608314514,
      "learning_rate": 1.9526835918081363e-05,
      "loss": 0.7885,
      "step": 2110
    },
    {
      "epoch": 0.14420542470878325,
      "grad_norm": 0.260049968957901,
      "learning_rate": 1.9519974284166982e-05,
      "loss": 0.7899,
      "step": 2120
    },
    {
      "epoch": 0.14488563897627751,
      "grad_norm": 0.26019856333732605,
      "learning_rate": 1.951306447863146e-05,
      "loss": 0.7943,
      "step": 2130
    },
    {
      "epoch": 0.14556585324377178,
      "grad_norm": 0.25930050015449524,
      "learning_rate": 1.9506106536438806e-05,
      "loss": 0.7877,
      "step": 2140
    },
    {
      "epoch": 0.14624606751126604,
      "grad_norm": 0.2671353816986084,
      "learning_rate": 1.9499100492796614e-05,
      "loss": 0.8049,
      "step": 2150
    },
    {
      "epoch": 0.1469262817787603,
      "grad_norm": 0.25898241996765137,
      "learning_rate": 1.949204638315587e-05,
      "loss": 0.7876,
      "step": 2160
    },
    {
      "epoch": 0.14760649604625456,
      "grad_norm": 0.24996714293956757,
      "learning_rate": 1.948494424321078e-05,
      "loss": 0.7861,
      "step": 2170
    },
    {
      "epoch": 0.14828671031374882,
      "grad_norm": 0.25589415431022644,
      "learning_rate": 1.947779410889858e-05,
      "loss": 0.79,
      "step": 2180
    },
    {
      "epoch": 0.14896692458124308,
      "grad_norm": 0.26438066363334656,
      "learning_rate": 1.9470596016399366e-05,
      "loss": 0.7955,
      "step": 2190
    },
    {
      "epoch": 0.14964713884873734,
      "grad_norm": 0.24316558241844177,
      "learning_rate": 1.9463350002135903e-05,
      "loss": 0.7886,
      "step": 2200
    },
    {
      "epoch": 0.1503273531162316,
      "grad_norm": 0.25717779994010925,
      "learning_rate": 1.945605610277344e-05,
      "loss": 0.7765,
      "step": 2210
    },
    {
      "epoch": 0.15100756738372587,
      "grad_norm": 0.2595769464969635,
      "learning_rate": 1.944871435521953e-05,
      "loss": 0.7874,
      "step": 2220
    },
    {
      "epoch": 0.15168778165122013,
      "grad_norm": 0.2691980302333832,
      "learning_rate": 1.9441324796623843e-05,
      "loss": 0.8013,
      "step": 2230
    },
    {
      "epoch": 0.1523679959187144,
      "grad_norm": 0.2670581638813019,
      "learning_rate": 1.9433887464377975e-05,
      "loss": 0.7922,
      "step": 2240
    },
    {
      "epoch": 0.15304821018620865,
      "grad_norm": 0.25826627016067505,
      "learning_rate": 1.9426402396115256e-05,
      "loss": 0.7823,
      "step": 2250
    },
    {
      "epoch": 0.1537284244537029,
      "grad_norm": 0.24645116925239563,
      "learning_rate": 1.941886962971056e-05,
      "loss": 0.7953,
      "step": 2260
    },
    {
      "epoch": 0.15440863872119717,
      "grad_norm": 0.26420411467552185,
      "learning_rate": 1.941128920328013e-05,
      "loss": 0.8027,
      "step": 2270
    },
    {
      "epoch": 0.15508885298869143,
      "grad_norm": 0.2580704987049103,
      "learning_rate": 1.940366115518136e-05,
      "loss": 0.7892,
      "step": 2280
    },
    {
      "epoch": 0.1557690672561857,
      "grad_norm": 0.2476886659860611,
      "learning_rate": 1.939598552401261e-05,
      "loss": 0.7859,
      "step": 2290
    },
    {
      "epoch": 0.15644928152367996,
      "grad_norm": 0.2571199834346771,
      "learning_rate": 1.9388262348613023e-05,
      "loss": 0.8004,
      "step": 2300
    },
    {
      "epoch": 0.15712949579117422,
      "grad_norm": 0.2546902000904083,
      "learning_rate": 1.9380491668062313e-05,
      "loss": 0.7731,
      "step": 2310
    },
    {
      "epoch": 0.15780971005866848,
      "grad_norm": 0.26125824451446533,
      "learning_rate": 1.937267352168057e-05,
      "loss": 0.7815,
      "step": 2320
    },
    {
      "epoch": 0.15848992432616274,
      "grad_norm": 0.2520924508571625,
      "learning_rate": 1.936480794902807e-05,
      "loss": 0.7699,
      "step": 2330
    },
    {
      "epoch": 0.159170138593657,
      "grad_norm": 0.2505636513233185,
      "learning_rate": 1.9356894989905072e-05,
      "loss": 0.8013,
      "step": 2340
    },
    {
      "epoch": 0.15985035286115126,
      "grad_norm": 0.2629729211330414,
      "learning_rate": 1.93489346843516e-05,
      "loss": 0.7784,
      "step": 2350
    },
    {
      "epoch": 0.16053056712864552,
      "grad_norm": 0.24999473989009857,
      "learning_rate": 1.934092707264726e-05,
      "loss": 0.7933,
      "step": 2360
    },
    {
      "epoch": 0.16121078139613979,
      "grad_norm": 1.5976718664169312,
      "learning_rate": 1.9332872195311038e-05,
      "loss": 0.8006,
      "step": 2370
    },
    {
      "epoch": 0.16189099566363405,
      "grad_norm": 0.2572176456451416,
      "learning_rate": 1.932477009310107e-05,
      "loss": 0.7872,
      "step": 2380
    },
    {
      "epoch": 0.1625712099311283,
      "grad_norm": 0.25545644760131836,
      "learning_rate": 1.931662080701447e-05,
      "loss": 0.7924,
      "step": 2390
    },
    {
      "epoch": 0.16325142419862257,
      "grad_norm": 0.2590372562408447,
      "learning_rate": 1.9308424378287098e-05,
      "loss": 0.784,
      "step": 2400
    },
    {
      "epoch": 0.16393163846611683,
      "grad_norm": 0.2718978822231293,
      "learning_rate": 1.9300180848393354e-05,
      "loss": 0.791,
      "step": 2410
    },
    {
      "epoch": 0.1646118527336111,
      "grad_norm": 0.26363012194633484,
      "learning_rate": 1.929189025904598e-05,
      "loss": 0.7921,
      "step": 2420
    },
    {
      "epoch": 0.16529206700110535,
      "grad_norm": 0.25173771381378174,
      "learning_rate": 1.928355265219583e-05,
      "loss": 0.7952,
      "step": 2430
    },
    {
      "epoch": 0.16597228126859961,
      "grad_norm": 0.26242753863334656,
      "learning_rate": 1.9275168070031692e-05,
      "loss": 0.7835,
      "step": 2440
    },
    {
      "epoch": 0.16665249553609388,
      "grad_norm": 0.26284611225128174,
      "learning_rate": 1.9266736554980024e-05,
      "loss": 0.778,
      "step": 2450
    },
    {
      "epoch": 0.16733270980358814,
      "grad_norm": 0.25407007336616516,
      "learning_rate": 1.9258258149704786e-05,
      "loss": 0.7824,
      "step": 2460
    },
    {
      "epoch": 0.1680129240710824,
      "grad_norm": 0.25445783138275146,
      "learning_rate": 1.9249732897107198e-05,
      "loss": 0.7996,
      "step": 2470
    },
    {
      "epoch": 0.16869313833857666,
      "grad_norm": 0.2684445083141327,
      "learning_rate": 1.9241160840325536e-05,
      "loss": 0.7902,
      "step": 2480
    },
    {
      "epoch": 0.16937335260607092,
      "grad_norm": 0.251264750957489,
      "learning_rate": 1.9232542022734897e-05,
      "loss": 0.7858,
      "step": 2490
    },
    {
      "epoch": 0.17005356687356518,
      "grad_norm": 0.25886887311935425,
      "learning_rate": 1.9223876487947002e-05,
      "loss": 0.7834,
      "step": 2500
    },
    {
      "epoch": 0.17073378114105944,
      "grad_norm": 0.24720726907253265,
      "learning_rate": 1.9215164279809958e-05,
      "loss": 0.7942,
      "step": 2510
    },
    {
      "epoch": 0.1714139954085537,
      "grad_norm": 0.2537200152873993,
      "learning_rate": 1.920640544240804e-05,
      "loss": 0.7936,
      "step": 2520
    },
    {
      "epoch": 0.17209420967604797,
      "grad_norm": 0.24554233253002167,
      "learning_rate": 1.9197600020061475e-05,
      "loss": 0.7779,
      "step": 2530
    },
    {
      "epoch": 0.17277442394354223,
      "grad_norm": 0.27647775411605835,
      "learning_rate": 1.9188748057326213e-05,
      "loss": 0.7887,
      "step": 2540
    },
    {
      "epoch": 0.1734546382110365,
      "grad_norm": 0.25730079412460327,
      "learning_rate": 1.917984959899369e-05,
      "loss": 0.789,
      "step": 2550
    },
    {
      "epoch": 0.17413485247853075,
      "grad_norm": 0.2522630989551544,
      "learning_rate": 1.9170904690090628e-05,
      "loss": 0.7823,
      "step": 2560
    },
    {
      "epoch": 0.174815066746025,
      "grad_norm": 0.2553759217262268,
      "learning_rate": 1.9161913375878777e-05,
      "loss": 0.7832,
      "step": 2570
    },
    {
      "epoch": 0.17549528101351927,
      "grad_norm": 0.25228244066238403,
      "learning_rate": 1.9152875701854715e-05,
      "loss": 0.7892,
      "step": 2580
    },
    {
      "epoch": 0.1761754952810135,
      "grad_norm": 0.25936052203178406,
      "learning_rate": 1.9143791713749592e-05,
      "loss": 0.7795,
      "step": 2590
    },
    {
      "epoch": 0.17685570954850777,
      "grad_norm": 0.2502341866493225,
      "learning_rate": 1.9134661457528914e-05,
      "loss": 0.7944,
      "step": 2600
    },
    {
      "epoch": 0.17753592381600203,
      "grad_norm": 0.24452045559883118,
      "learning_rate": 1.9125484979392306e-05,
      "loss": 0.7805,
      "step": 2610
    },
    {
      "epoch": 0.1782161380834963,
      "grad_norm": 0.25821682810783386,
      "learning_rate": 1.911626232577328e-05,
      "loss": 0.7858,
      "step": 2620
    },
    {
      "epoch": 0.17889635235099055,
      "grad_norm": 0.2633768320083618,
      "learning_rate": 1.9106993543338997e-05,
      "loss": 0.792,
      "step": 2630
    },
    {
      "epoch": 0.1795765666184848,
      "grad_norm": 0.2523728609085083,
      "learning_rate": 1.909767867899003e-05,
      "loss": 0.789,
      "step": 2640
    },
    {
      "epoch": 0.18025678088597907,
      "grad_norm": 0.2643498182296753,
      "learning_rate": 1.9088317779860136e-05,
      "loss": 0.7941,
      "step": 2650
    },
    {
      "epoch": 0.18093699515347333,
      "grad_norm": 0.2574792504310608,
      "learning_rate": 1.9078910893316005e-05,
      "loss": 0.7933,
      "step": 2660
    },
    {
      "epoch": 0.1816172094209676,
      "grad_norm": 0.24570371210575104,
      "learning_rate": 1.906945806695703e-05,
      "loss": 0.7811,
      "step": 2670
    },
    {
      "epoch": 0.18229742368846186,
      "grad_norm": 0.26268598437309265,
      "learning_rate": 1.9059959348615053e-05,
      "loss": 0.7853,
      "step": 2680
    },
    {
      "epoch": 0.18297763795595612,
      "grad_norm": 0.2532947063446045,
      "learning_rate": 1.9050414786354144e-05,
      "loss": 0.7887,
      "step": 2690
    },
    {
      "epoch": 0.18365785222345038,
      "grad_norm": 0.26706111431121826,
      "learning_rate": 1.904082442847034e-05,
      "loss": 0.8093,
      "step": 2700
    },
    {
      "epoch": 0.18433806649094464,
      "grad_norm": 0.2524626851081848,
      "learning_rate": 1.903118832349141e-05,
      "loss": 0.784,
      "step": 2710
    },
    {
      "epoch": 0.1850182807584389,
      "grad_norm": 0.2823905348777771,
      "learning_rate": 1.9021506520176598e-05,
      "loss": 0.7824,
      "step": 2720
    },
    {
      "epoch": 0.18569849502593316,
      "grad_norm": 0.25040870904922485,
      "learning_rate": 1.9011779067516392e-05,
      "loss": 0.7828,
      "step": 2730
    },
    {
      "epoch": 0.18637870929342742,
      "grad_norm": 0.255565345287323,
      "learning_rate": 1.9002006014732266e-05,
      "loss": 0.7721,
      "step": 2740
    },
    {
      "epoch": 0.18705892356092169,
      "grad_norm": 0.24715404212474823,
      "learning_rate": 1.8992187411276433e-05,
      "loss": 0.7908,
      "step": 2750
    },
    {
      "epoch": 0.18773913782841595,
      "grad_norm": 0.24670803546905518,
      "learning_rate": 1.8982323306831592e-05,
      "loss": 0.7905,
      "step": 2760
    },
    {
      "epoch": 0.1884193520959102,
      "grad_norm": 0.25133422017097473,
      "learning_rate": 1.897241375131069e-05,
      "loss": 0.7907,
      "step": 2770
    },
    {
      "epoch": 0.18909956636340447,
      "grad_norm": 0.24613183736801147,
      "learning_rate": 1.8962458794856645e-05,
      "loss": 0.7739,
      "step": 2780
    },
    {
      "epoch": 0.18977978063089873,
      "grad_norm": 0.24352288246154785,
      "learning_rate": 1.8952458487842114e-05,
      "loss": 0.7834,
      "step": 2790
    },
    {
      "epoch": 0.190459994898393,
      "grad_norm": 0.2572003901004791,
      "learning_rate": 1.894241288086923e-05,
      "loss": 0.7865,
      "step": 2800
    },
    {
      "epoch": 0.19114020916588725,
      "grad_norm": 0.24495874345302582,
      "learning_rate": 1.893232202476935e-05,
      "loss": 0.7822,
      "step": 2810
    },
    {
      "epoch": 0.19182042343338152,
      "grad_norm": 0.24594272673130035,
      "learning_rate": 1.8922185970602785e-05,
      "loss": 0.785,
      "step": 2820
    },
    {
      "epoch": 0.19250063770087578,
      "grad_norm": 0.24420033395290375,
      "learning_rate": 1.891200476965856e-05,
      "loss": 0.7849,
      "step": 2830
    },
    {
      "epoch": 0.19318085196837004,
      "grad_norm": 0.25174930691719055,
      "learning_rate": 1.8901778473454134e-05,
      "loss": 0.7726,
      "step": 2840
    },
    {
      "epoch": 0.1938610662358643,
      "grad_norm": 0.25425928831100464,
      "learning_rate": 1.889150713373517e-05,
      "loss": 0.7866,
      "step": 2850
    },
    {
      "epoch": 0.19454128050335856,
      "grad_norm": 0.2496025264263153,
      "learning_rate": 1.8881190802475227e-05,
      "loss": 0.7917,
      "step": 2860
    },
    {
      "epoch": 0.19522149477085282,
      "grad_norm": 0.25157952308654785,
      "learning_rate": 1.887082953187555e-05,
      "loss": 0.7875,
      "step": 2870
    },
    {
      "epoch": 0.19590170903834708,
      "grad_norm": 0.25684642791748047,
      "learning_rate": 1.8860423374364767e-05,
      "loss": 0.7775,
      "step": 2880
    },
    {
      "epoch": 0.19658192330584134,
      "grad_norm": 0.37965211272239685,
      "learning_rate": 1.8849972382598634e-05,
      "loss": 0.7809,
      "step": 2890
    },
    {
      "epoch": 0.1972621375733356,
      "grad_norm": 0.28048843145370483,
      "learning_rate": 1.8839476609459778e-05,
      "loss": 0.7934,
      "step": 2900
    },
    {
      "epoch": 0.19794235184082987,
      "grad_norm": 0.2760063111782074,
      "learning_rate": 1.8828936108057413e-05,
      "loss": 0.7938,
      "step": 2910
    },
    {
      "epoch": 0.19862256610832413,
      "grad_norm": 0.2563750147819519,
      "learning_rate": 1.8818350931727093e-05,
      "loss": 0.7887,
      "step": 2920
    },
    {
      "epoch": 0.1993027803758184,
      "grad_norm": 0.24341252446174622,
      "learning_rate": 1.8807721134030418e-05,
      "loss": 0.7989,
      "step": 2930
    },
    {
      "epoch": 0.19998299464331265,
      "grad_norm": 0.25352743268013,
      "learning_rate": 1.8797046768754783e-05,
      "loss": 0.762,
      "step": 2940
    },
    {
      "epoch": 0.2006632089108069,
      "grad_norm": 0.26104164123535156,
      "learning_rate": 1.8786327889913093e-05,
      "loss": 0.7762,
      "step": 2950
    },
    {
      "epoch": 0.20134342317830117,
      "grad_norm": 0.24420127272605896,
      "learning_rate": 1.8775564551743498e-05,
      "loss": 0.7812,
      "step": 2960
    },
    {
      "epoch": 0.20202363744579543,
      "grad_norm": 0.24496297538280487,
      "learning_rate": 1.876475680870911e-05,
      "loss": 0.7849,
      "step": 2970
    },
    {
      "epoch": 0.2027038517132897,
      "grad_norm": 0.24318674206733704,
      "learning_rate": 1.8753904715497734e-05,
      "loss": 0.8012,
      "step": 2980
    },
    {
      "epoch": 0.20338406598078396,
      "grad_norm": 0.25721314549446106,
      "learning_rate": 1.8743008327021594e-05,
      "loss": 0.788,
      "step": 2990
    },
    {
      "epoch": 0.20406428024827822,
      "grad_norm": 0.2569056749343872,
      "learning_rate": 1.8732067698417042e-05,
      "loss": 0.7852,
      "step": 3000
    },
    {
      "epoch": 0.20474449451577248,
      "grad_norm": 0.2654126286506653,
      "learning_rate": 1.8721082885044296e-05,
      "loss": 0.7908,
      "step": 3010
    },
    {
      "epoch": 0.20542470878326674,
      "grad_norm": 0.2526949942111969,
      "learning_rate": 1.8710053942487144e-05,
      "loss": 0.7924,
      "step": 3020
    },
    {
      "epoch": 0.206104923050761,
      "grad_norm": 0.2519407272338867,
      "learning_rate": 1.8698980926552675e-05,
      "loss": 0.781,
      "step": 3030
    },
    {
      "epoch": 0.20678513731825526,
      "grad_norm": 0.23384132981300354,
      "learning_rate": 1.8687863893270995e-05,
      "loss": 0.7733,
      "step": 3040
    },
    {
      "epoch": 0.20746535158574952,
      "grad_norm": 0.25337284803390503,
      "learning_rate": 1.867670289889493e-05,
      "loss": 0.7963,
      "step": 3050
    },
    {
      "epoch": 0.20814556585324376,
      "grad_norm": 0.2520735263824463,
      "learning_rate": 1.8665497999899763e-05,
      "loss": 0.7783,
      "step": 3060
    },
    {
      "epoch": 0.20882578012073802,
      "grad_norm": 0.25193944573402405,
      "learning_rate": 1.865424925298293e-05,
      "loss": 0.7874,
      "step": 3070
    },
    {
      "epoch": 0.20950599438823228,
      "grad_norm": 0.26063066720962524,
      "learning_rate": 1.864295671506374e-05,
      "loss": 0.7863,
      "step": 3080
    },
    {
      "epoch": 0.21018620865572654,
      "grad_norm": 0.2780202627182007,
      "learning_rate": 1.8631620443283093e-05,
      "loss": 0.7907,
      "step": 3090
    },
    {
      "epoch": 0.2108664229232208,
      "grad_norm": 0.25003233551979065,
      "learning_rate": 1.8620240495003172e-05,
      "loss": 0.7834,
      "step": 3100
    },
    {
      "epoch": 0.21154663719071506,
      "grad_norm": 0.2419746369123459,
      "learning_rate": 1.8608816927807182e-05,
      "loss": 0.7776,
      "step": 3110
    },
    {
      "epoch": 0.21222685145820933,
      "grad_norm": 0.2449108064174652,
      "learning_rate": 1.8597349799499037e-05,
      "loss": 0.779,
      "step": 3120
    },
    {
      "epoch": 0.2129070657257036,
      "grad_norm": 0.2526479959487915,
      "learning_rate": 1.858583916810306e-05,
      "loss": 0.7939,
      "step": 3130
    },
    {
      "epoch": 0.21358727999319785,
      "grad_norm": 0.2487485408782959,
      "learning_rate": 1.857428509186372e-05,
      "loss": 0.793,
      "step": 3140
    },
    {
      "epoch": 0.2142674942606921,
      "grad_norm": 0.25782719254493713,
      "learning_rate": 1.8562687629245312e-05,
      "loss": 0.7807,
      "step": 3150
    },
    {
      "epoch": 0.21494770852818637,
      "grad_norm": 0.24029205739498138,
      "learning_rate": 1.8551046838931665e-05,
      "loss": 0.7909,
      "step": 3160
    },
    {
      "epoch": 0.21562792279568063,
      "grad_norm": 0.24059611558914185,
      "learning_rate": 1.853936277982585e-05,
      "loss": 0.7798,
      "step": 3170
    },
    {
      "epoch": 0.2163081370631749,
      "grad_norm": 0.2464148998260498,
      "learning_rate": 1.8527635511049893e-05,
      "loss": 0.7797,
      "step": 3180
    },
    {
      "epoch": 0.21698835133066915,
      "grad_norm": 0.2505282759666443,
      "learning_rate": 1.851586509194444e-05,
      "loss": 0.7908,
      "step": 3190
    },
    {
      "epoch": 0.21766856559816342,
      "grad_norm": 0.47646915912628174,
      "learning_rate": 1.8504051582068502e-05,
      "loss": 0.7863,
      "step": 3200
    },
    {
      "epoch": 0.21834877986565768,
      "grad_norm": 0.2539674937725067,
      "learning_rate": 1.8492195041199122e-05,
      "loss": 0.7858,
      "step": 3210
    },
    {
      "epoch": 0.21902899413315194,
      "grad_norm": 0.2528616487979889,
      "learning_rate": 1.848029552933108e-05,
      "loss": 0.7894,
      "step": 3220
    },
    {
      "epoch": 0.2197092084006462,
      "grad_norm": 0.2563914358615875,
      "learning_rate": 1.84683531066766e-05,
      "loss": 0.7789,
      "step": 3230
    },
    {
      "epoch": 0.22038942266814046,
      "grad_norm": 0.25774049758911133,
      "learning_rate": 1.8456367833665024e-05,
      "loss": 0.7702,
      "step": 3240
    },
    {
      "epoch": 0.22106963693563472,
      "grad_norm": 0.25088292360305786,
      "learning_rate": 1.8444339770942537e-05,
      "loss": 0.7835,
      "step": 3250
    },
    {
      "epoch": 0.22174985120312898,
      "grad_norm": 0.24714317917823792,
      "learning_rate": 1.8432268979371825e-05,
      "loss": 0.7669,
      "step": 3260
    },
    {
      "epoch": 0.22243006547062324,
      "grad_norm": 0.25273606181144714,
      "learning_rate": 1.8420155520031803e-05,
      "loss": 0.7942,
      "step": 3270
    },
    {
      "epoch": 0.2231102797381175,
      "grad_norm": 0.25561535358428955,
      "learning_rate": 1.840799945421726e-05,
      "loss": 0.7879,
      "step": 3280
    },
    {
      "epoch": 0.22379049400561177,
      "grad_norm": 0.24375256896018982,
      "learning_rate": 1.8395800843438608e-05,
      "loss": 0.797,
      "step": 3290
    },
    {
      "epoch": 0.22447070827310603,
      "grad_norm": 0.25886252522468567,
      "learning_rate": 1.8383559749421517e-05,
      "loss": 0.7828,
      "step": 3300
    },
    {
      "epoch": 0.2251509225406003,
      "grad_norm": 0.2579536437988281,
      "learning_rate": 1.8371276234106634e-05,
      "loss": 0.7938,
      "step": 3310
    },
    {
      "epoch": 0.22583113680809455,
      "grad_norm": 0.2595989406108856,
      "learning_rate": 1.8358950359649256e-05,
      "loss": 0.779,
      "step": 3320
    },
    {
      "epoch": 0.2265113510755888,
      "grad_norm": 0.2517205476760864,
      "learning_rate": 1.8346582188419022e-05,
      "loss": 0.7743,
      "step": 3330
    },
    {
      "epoch": 0.22719156534308307,
      "grad_norm": 0.2425038069486618,
      "learning_rate": 1.8334171782999596e-05,
      "loss": 0.7698,
      "step": 3340
    },
    {
      "epoch": 0.22787177961057734,
      "grad_norm": 0.2507912516593933,
      "learning_rate": 1.8321719206188344e-05,
      "loss": 0.7834,
      "step": 3350
    },
    {
      "epoch": 0.2285519938780716,
      "grad_norm": 0.25281819701194763,
      "learning_rate": 1.8309224520996034e-05,
      "loss": 0.779,
      "step": 3360
    },
    {
      "epoch": 0.22923220814556586,
      "grad_norm": 0.24461200833320618,
      "learning_rate": 1.8296687790646487e-05,
      "loss": 0.7771,
      "step": 3370
    },
    {
      "epoch": 0.22991242241306012,
      "grad_norm": 0.2611219584941864,
      "learning_rate": 1.828410907857629e-05,
      "loss": 0.7872,
      "step": 3380
    },
    {
      "epoch": 0.23059263668055438,
      "grad_norm": 0.26623839139938354,
      "learning_rate": 1.8271488448434455e-05,
      "loss": 0.782,
      "step": 3390
    },
    {
      "epoch": 0.23127285094804864,
      "grad_norm": 0.2582928240299225,
      "learning_rate": 1.8258825964082096e-05,
      "loss": 0.784,
      "step": 3400
    },
    {
      "epoch": 0.2319530652155429,
      "grad_norm": 0.3065914809703827,
      "learning_rate": 1.8246121689592123e-05,
      "loss": 0.792,
      "step": 3410
    },
    {
      "epoch": 0.23263327948303716,
      "grad_norm": 0.25425055623054504,
      "learning_rate": 1.8233375689248896e-05,
      "loss": 0.7827,
      "step": 3420
    },
    {
      "epoch": 0.23331349375053143,
      "grad_norm": 0.2543756067752838,
      "learning_rate": 1.8220588027547917e-05,
      "loss": 0.7936,
      "step": 3430
    },
    {
      "epoch": 0.2339937080180257,
      "grad_norm": 0.24979901313781738,
      "learning_rate": 1.82077587691955e-05,
      "loss": 0.7934,
      "step": 3440
    },
    {
      "epoch": 0.23467392228551995,
      "grad_norm": 0.24882297217845917,
      "learning_rate": 1.8194887979108425e-05,
      "loss": 0.7794,
      "step": 3450
    },
    {
      "epoch": 0.2353541365530142,
      "grad_norm": 0.2424454540014267,
      "learning_rate": 1.818197572241364e-05,
      "loss": 0.7868,
      "step": 3460
    },
    {
      "epoch": 0.23603435082050847,
      "grad_norm": 0.2578614056110382,
      "learning_rate": 1.8169022064447913e-05,
      "loss": 0.7885,
      "step": 3470
    },
    {
      "epoch": 0.23671456508800273,
      "grad_norm": 0.26300248503685,
      "learning_rate": 1.81560270707575e-05,
      "loss": 0.7791,
      "step": 3480
    },
    {
      "epoch": 0.237394779355497,
      "grad_norm": 0.24765539169311523,
      "learning_rate": 1.814299080709783e-05,
      "loss": 0.7945,
      "step": 3490
    },
    {
      "epoch": 0.23807499362299125,
      "grad_norm": 0.24388448894023895,
      "learning_rate": 1.812991333943314e-05,
      "loss": 0.7902,
      "step": 3500
    },
    {
      "epoch": 0.23875520789048552,
      "grad_norm": 0.2515674829483032,
      "learning_rate": 1.811679473393618e-05,
      "loss": 0.7856,
      "step": 3510
    },
    {
      "epoch": 0.23943542215797978,
      "grad_norm": 0.2512035667896271,
      "learning_rate": 1.8103635056987855e-05,
      "loss": 0.7857,
      "step": 3520
    },
    {
      "epoch": 0.240115636425474,
      "grad_norm": 0.2631925940513611,
      "learning_rate": 1.8090434375176888e-05,
      "loss": 0.788,
      "step": 3530
    },
    {
      "epoch": 0.24079585069296827,
      "grad_norm": 0.27402493357658386,
      "learning_rate": 1.8077192755299494e-05,
      "loss": 0.7789,
      "step": 3540
    },
    {
      "epoch": 0.24147606496046253,
      "grad_norm": 0.2507264018058777,
      "learning_rate": 1.806391026435904e-05,
      "loss": 0.7833,
      "step": 3550
    },
    {
      "epoch": 0.2421562792279568,
      "grad_norm": 0.25710999965667725,
      "learning_rate": 1.8050586969565697e-05,
      "loss": 0.7866,
      "step": 3560
    },
    {
      "epoch": 0.24283649349545106,
      "grad_norm": 0.2526022791862488,
      "learning_rate": 1.8037222938336107e-05,
      "loss": 0.7902,
      "step": 3570
    },
    {
      "epoch": 0.24351670776294532,
      "grad_norm": 0.2536466121673584,
      "learning_rate": 1.8023818238293046e-05,
      "loss": 0.7903,
      "step": 3580
    },
    {
      "epoch": 0.24419692203043958,
      "grad_norm": 0.24735701084136963,
      "learning_rate": 1.8010372937265063e-05,
      "loss": 0.7826,
      "step": 3590
    },
    {
      "epoch": 0.24487713629793384,
      "grad_norm": 0.25265151262283325,
      "learning_rate": 1.7996887103286173e-05,
      "loss": 0.7926,
      "step": 3600
    },
    {
      "epoch": 0.2455573505654281,
      "grad_norm": 0.2552418112754822,
      "learning_rate": 1.7983360804595467e-05,
      "loss": 0.7847,
      "step": 3610
    },
    {
      "epoch": 0.24623756483292236,
      "grad_norm": 0.2783318758010864,
      "learning_rate": 1.7969794109636806e-05,
      "loss": 0.7984,
      "step": 3620
    },
    {
      "epoch": 0.24691777910041662,
      "grad_norm": 0.2513044476509094,
      "learning_rate": 1.7956187087058447e-05,
      "loss": 0.7863,
      "step": 3630
    },
    {
      "epoch": 0.24759799336791088,
      "grad_norm": 0.2544722855091095,
      "learning_rate": 1.7942539805712715e-05,
      "loss": 0.7847,
      "step": 3640
    },
    {
      "epoch": 0.24827820763540515,
      "grad_norm": 0.2505376935005188,
      "learning_rate": 1.792885233465564e-05,
      "loss": 0.7801,
      "step": 3650
    },
    {
      "epoch": 0.2489584219028994,
      "grad_norm": 0.2470908910036087,
      "learning_rate": 1.7915124743146623e-05,
      "loss": 0.778,
      "step": 3660
    },
    {
      "epoch": 0.24963863617039367,
      "grad_norm": 0.28139612078666687,
      "learning_rate": 1.790135710064807e-05,
      "loss": 0.7649,
      "step": 3670
    },
    {
      "epoch": 0.25031885043788793,
      "grad_norm": 0.2537873089313507,
      "learning_rate": 1.788754947682505e-05,
      "loss": 0.7771,
      "step": 3680
    },
    {
      "epoch": 0.2509990647053822,
      "grad_norm": 0.24972447752952576,
      "learning_rate": 1.7873701941544935e-05,
      "loss": 0.7949,
      "step": 3690
    },
    {
      "epoch": 0.25167927897287645,
      "grad_norm": 0.25351646542549133,
      "learning_rate": 1.7859814564877063e-05,
      "loss": 0.7751,
      "step": 3700
    },
    {
      "epoch": 0.2523594932403707,
      "grad_norm": 0.2693154215812683,
      "learning_rate": 1.7845887417092358e-05,
      "loss": 0.7709,
      "step": 3710
    },
    {
      "epoch": 0.253039707507865,
      "grad_norm": 0.3126961588859558,
      "learning_rate": 1.7831920568663e-05,
      "loss": 0.7929,
      "step": 3720
    },
    {
      "epoch": 0.25371992177535924,
      "grad_norm": 0.2440711259841919,
      "learning_rate": 1.781791409026204e-05,
      "loss": 0.7842,
      "step": 3730
    },
    {
      "epoch": 0.2544001360428535,
      "grad_norm": 0.2512514591217041,
      "learning_rate": 1.780386805276308e-05,
      "loss": 0.7765,
      "step": 3740
    },
    {
      "epoch": 0.25508035031034776,
      "grad_norm": 0.24332073330879211,
      "learning_rate": 1.778978252723988e-05,
      "loss": 0.7965,
      "step": 3750
    },
    {
      "epoch": 0.255760564577842,
      "grad_norm": 0.2476550042629242,
      "learning_rate": 1.7775657584966018e-05,
      "loss": 0.7736,
      "step": 3760
    },
    {
      "epoch": 0.2564407788453363,
      "grad_norm": 0.2414562702178955,
      "learning_rate": 1.776149329741451e-05,
      "loss": 0.771,
      "step": 3770
    },
    {
      "epoch": 0.25712099311283054,
      "grad_norm": 0.2415563464164734,
      "learning_rate": 1.7747289736257482e-05,
      "loss": 0.7706,
      "step": 3780
    },
    {
      "epoch": 0.2578012073803248,
      "grad_norm": 0.24708572030067444,
      "learning_rate": 1.773304697336577e-05,
      "loss": 0.7917,
      "step": 3790
    },
    {
      "epoch": 0.25848142164781907,
      "grad_norm": 0.25027525424957275,
      "learning_rate": 1.771876508080858e-05,
      "loss": 0.7804,
      "step": 3800
    },
    {
      "epoch": 0.2591616359153133,
      "grad_norm": 0.2515070140361786,
      "learning_rate": 1.770444413085312e-05,
      "loss": 0.7864,
      "step": 3810
    },
    {
      "epoch": 0.2598418501828076,
      "grad_norm": 0.24446500837802887,
      "learning_rate": 1.769008419596422e-05,
      "loss": 0.7862,
      "step": 3820
    },
    {
      "epoch": 0.26052206445030185,
      "grad_norm": 0.25847122073173523,
      "learning_rate": 1.767568534880398e-05,
      "loss": 0.7912,
      "step": 3830
    },
    {
      "epoch": 0.2612022787177961,
      "grad_norm": 0.25062787532806396,
      "learning_rate": 1.766124766223141e-05,
      "loss": 0.7788,
      "step": 3840
    },
    {
      "epoch": 0.26188249298529037,
      "grad_norm": 0.2614087164402008,
      "learning_rate": 1.764677120930203e-05,
      "loss": 0.7853,
      "step": 3850
    },
    {
      "epoch": 0.26256270725278463,
      "grad_norm": 0.2507074773311615,
      "learning_rate": 1.763225606326754e-05,
      "loss": 0.7881,
      "step": 3860
    },
    {
      "epoch": 0.2632429215202789,
      "grad_norm": 0.3656567931175232,
      "learning_rate": 1.7617702297575408e-05,
      "loss": 0.7816,
      "step": 3870
    },
    {
      "epoch": 0.26392313578777316,
      "grad_norm": 0.24729153513908386,
      "learning_rate": 1.7603109985868544e-05,
      "loss": 0.7723,
      "step": 3880
    },
    {
      "epoch": 0.2646033500552674,
      "grad_norm": 0.2480657994747162,
      "learning_rate": 1.758847920198488e-05,
      "loss": 0.7978,
      "step": 3890
    },
    {
      "epoch": 0.2652835643227617,
      "grad_norm": 0.24121278524398804,
      "learning_rate": 1.7573810019957033e-05,
      "loss": 0.7864,
      "step": 3900
    },
    {
      "epoch": 0.26596377859025594,
      "grad_norm": 0.2533305585384369,
      "learning_rate": 1.7559102514011913e-05,
      "loss": 0.7786,
      "step": 3910
    },
    {
      "epoch": 0.2666439928577502,
      "grad_norm": 0.26402154564857483,
      "learning_rate": 1.7544356758570355e-05,
      "loss": 0.7931,
      "step": 3920
    },
    {
      "epoch": 0.26732420712524446,
      "grad_norm": 0.258201539516449,
      "learning_rate": 1.752957282824673e-05,
      "loss": 0.801,
      "step": 3930
    },
    {
      "epoch": 0.2680044213927387,
      "grad_norm": 0.25085070729255676,
      "learning_rate": 1.7514750797848587e-05,
      "loss": 0.7904,
      "step": 3940
    },
    {
      "epoch": 0.268684635660233,
      "grad_norm": 0.2558985948562622,
      "learning_rate": 1.7499890742376254e-05,
      "loss": 0.7947,
      "step": 3950
    },
    {
      "epoch": 0.26936484992772725,
      "grad_norm": 0.25026050209999084,
      "learning_rate": 1.7484992737022474e-05,
      "loss": 0.7767,
      "step": 3960
    },
    {
      "epoch": 0.2700450641952215,
      "grad_norm": 0.250328004360199,
      "learning_rate": 1.7470056857172017e-05,
      "loss": 0.783,
      "step": 3970
    },
    {
      "epoch": 0.27072527846271577,
      "grad_norm": 0.2586681842803955,
      "learning_rate": 1.7455083178401295e-05,
      "loss": 0.7867,
      "step": 3980
    },
    {
      "epoch": 0.27140549273021003,
      "grad_norm": 0.2514212131500244,
      "learning_rate": 1.7440071776477996e-05,
      "loss": 0.7876,
      "step": 3990
    },
    {
      "epoch": 0.2720857069977043,
      "grad_norm": 0.2571597695350647,
      "learning_rate": 1.7425022727360684e-05,
      "loss": 0.7852,
      "step": 4000
    },
    {
      "epoch": 0.27276592126519855,
      "grad_norm": 0.25561052560806274,
      "learning_rate": 1.7409936107198416e-05,
      "loss": 0.7902,
      "step": 4010
    },
    {
      "epoch": 0.2734461355326928,
      "grad_norm": 0.2517993450164795,
      "learning_rate": 1.739481199233036e-05,
      "loss": 0.7811,
      "step": 4020
    },
    {
      "epoch": 0.2741263498001871,
      "grad_norm": 0.24940787255764008,
      "learning_rate": 1.7379650459285417e-05,
      "loss": 0.7839,
      "step": 4030
    },
    {
      "epoch": 0.27480656406768134,
      "grad_norm": 0.24489739537239075,
      "learning_rate": 1.736445158478182e-05,
      "loss": 0.7957,
      "step": 4040
    },
    {
      "epoch": 0.2754867783351756,
      "grad_norm": 0.2565753757953644,
      "learning_rate": 1.7349215445726763e-05,
      "loss": 0.7781,
      "step": 4050
    },
    {
      "epoch": 0.27616699260266986,
      "grad_norm": 0.2571543753147125,
      "learning_rate": 1.733394211921598e-05,
      "loss": 0.7808,
      "step": 4060
    },
    {
      "epoch": 0.2768472068701641,
      "grad_norm": 0.2582187354564667,
      "learning_rate": 1.731863168253339e-05,
      "loss": 0.7962,
      "step": 4070
    },
    {
      "epoch": 0.2775274211376584,
      "grad_norm": 0.24603897333145142,
      "learning_rate": 1.7303284213150695e-05,
      "loss": 0.7772,
      "step": 4080
    },
    {
      "epoch": 0.27820763540515264,
      "grad_norm": 0.24644562602043152,
      "learning_rate": 1.7287899788726975e-05,
      "loss": 0.788,
      "step": 4090
    },
    {
      "epoch": 0.2788878496726469,
      "grad_norm": 0.2610757052898407,
      "learning_rate": 1.727247848710831e-05,
      "loss": 0.79,
      "step": 4100
    },
    {
      "epoch": 0.27956806394014116,
      "grad_norm": 0.24853172898292542,
      "learning_rate": 1.7257020386327376e-05,
      "loss": 0.7931,
      "step": 4110
    },
    {
      "epoch": 0.2802482782076354,
      "grad_norm": 0.2530014216899872,
      "learning_rate": 1.7241525564603058e-05,
      "loss": 0.7923,
      "step": 4120
    },
    {
      "epoch": 0.2809284924751297,
      "grad_norm": 0.2790886163711548,
      "learning_rate": 1.722599410034006e-05,
      "loss": 0.7676,
      "step": 4130
    },
    {
      "epoch": 0.28160870674262395,
      "grad_norm": 0.2523122727870941,
      "learning_rate": 1.7210426072128476e-05,
      "loss": 0.7803,
      "step": 4140
    },
    {
      "epoch": 0.2822889210101182,
      "grad_norm": 0.2545037865638733,
      "learning_rate": 1.719482155874344e-05,
      "loss": 0.7851,
      "step": 4150
    },
    {
      "epoch": 0.28296913527761247,
      "grad_norm": 0.28469881415367126,
      "learning_rate": 1.717918063914469e-05,
      "loss": 0.7883,
      "step": 4160
    },
    {
      "epoch": 0.28364934954510673,
      "grad_norm": 0.2560071051120758,
      "learning_rate": 1.716350339247619e-05,
      "loss": 0.7746,
      "step": 4170
    },
    {
      "epoch": 0.284329563812601,
      "grad_norm": 0.2442929893732071,
      "learning_rate": 1.714778989806571e-05,
      "loss": 0.7922,
      "step": 4180
    },
    {
      "epoch": 0.28500977808009526,
      "grad_norm": 0.2452080398797989,
      "learning_rate": 1.7132040235424443e-05,
      "loss": 0.7871,
      "step": 4190
    },
    {
      "epoch": 0.2856899923475895,
      "grad_norm": 0.23994505405426025,
      "learning_rate": 1.7116254484246596e-05,
      "loss": 0.7627,
      "step": 4200
    },
    {
      "epoch": 0.2863702066150838,
      "grad_norm": 0.255530446767807,
      "learning_rate": 1.710043272440898e-05,
      "loss": 0.7911,
      "step": 4210
    },
    {
      "epoch": 0.28705042088257804,
      "grad_norm": 0.24917486310005188,
      "learning_rate": 1.7084575035970624e-05,
      "loss": 0.793,
      "step": 4220
    },
    {
      "epoch": 0.2877306351500723,
      "grad_norm": 0.2526813745498657,
      "learning_rate": 1.7068681499172338e-05,
      "loss": 0.7778,
      "step": 4230
    },
    {
      "epoch": 0.2884108494175665,
      "grad_norm": 0.24449491500854492,
      "learning_rate": 1.705275219443635e-05,
      "loss": 0.7811,
      "step": 4240
    },
    {
      "epoch": 0.28909106368506077,
      "grad_norm": 0.26806896924972534,
      "learning_rate": 1.7036787202365858e-05,
      "loss": 0.8003,
      "step": 4250
    },
    {
      "epoch": 0.28977127795255503,
      "grad_norm": 0.2438434511423111,
      "learning_rate": 1.7020786603744647e-05,
      "loss": 0.783,
      "step": 4260
    },
    {
      "epoch": 0.2904514922200493,
      "grad_norm": 0.2513818144798279,
      "learning_rate": 1.7004750479536675e-05,
      "loss": 0.7772,
      "step": 4270
    },
    {
      "epoch": 0.29113170648754355,
      "grad_norm": 0.2550441324710846,
      "learning_rate": 1.6988678910885654e-05,
      "loss": 0.7786,
      "step": 4280
    },
    {
      "epoch": 0.2918119207550378,
      "grad_norm": 0.2485378384590149,
      "learning_rate": 1.697257197911466e-05,
      "loss": 0.7736,
      "step": 4290
    },
    {
      "epoch": 0.2924921350225321,
      "grad_norm": 0.249145969748497,
      "learning_rate": 1.695642976572569e-05,
      "loss": 0.7912,
      "step": 4300
    },
    {
      "epoch": 0.29317234929002634,
      "grad_norm": 0.2479313313961029,
      "learning_rate": 1.694025235239929e-05,
      "loss": 0.7746,
      "step": 4310
    },
    {
      "epoch": 0.2938525635575206,
      "grad_norm": 0.24584801495075226,
      "learning_rate": 1.6924039820994097e-05,
      "loss": 0.7835,
      "step": 4320
    },
    {
      "epoch": 0.29453277782501486,
      "grad_norm": 0.24700278043746948,
      "learning_rate": 1.6907792253546467e-05,
      "loss": 0.7745,
      "step": 4330
    },
    {
      "epoch": 0.2952129920925091,
      "grad_norm": 0.2509085237979889,
      "learning_rate": 1.689150973227003e-05,
      "loss": 0.7878,
      "step": 4340
    },
    {
      "epoch": 0.2958932063600034,
      "grad_norm": 0.24491603672504425,
      "learning_rate": 1.6875192339555284e-05,
      "loss": 0.7862,
      "step": 4350
    },
    {
      "epoch": 0.29657342062749764,
      "grad_norm": 0.2515392303466797,
      "learning_rate": 1.6858840157969186e-05,
      "loss": 0.7915,
      "step": 4360
    },
    {
      "epoch": 0.2972536348949919,
      "grad_norm": 0.25104475021362305,
      "learning_rate": 1.6842453270254717e-05,
      "loss": 0.7854,
      "step": 4370
    },
    {
      "epoch": 0.29793384916248616,
      "grad_norm": 0.27672502398490906,
      "learning_rate": 1.6826031759330478e-05,
      "loss": 0.7857,
      "step": 4380
    },
    {
      "epoch": 0.2986140634299804,
      "grad_norm": 0.24459819495677948,
      "learning_rate": 1.680957570829027e-05,
      "loss": 0.7709,
      "step": 4390
    },
    {
      "epoch": 0.2992942776974747,
      "grad_norm": 0.25333377718925476,
      "learning_rate": 1.6793085200402658e-05,
      "loss": 0.7737,
      "step": 4400
    },
    {
      "epoch": 0.29997449196496895,
      "grad_norm": 0.26826807856559753,
      "learning_rate": 1.6776560319110568e-05,
      "loss": 0.7892,
      "step": 4410
    },
    {
      "epoch": 0.3006547062324632,
      "grad_norm": 0.2829219400882721,
      "learning_rate": 1.676000114803086e-05,
      "loss": 0.7924,
      "step": 4420
    },
    {
      "epoch": 0.30133492049995747,
      "grad_norm": 0.24529558420181274,
      "learning_rate": 1.6743407770953893e-05,
      "loss": 0.7827,
      "step": 4430
    },
    {
      "epoch": 0.30201513476745173,
      "grad_norm": 0.24346935749053955,
      "learning_rate": 1.6726780271843125e-05,
      "loss": 0.7948,
      "step": 4440
    },
    {
      "epoch": 0.302695349034946,
      "grad_norm": 0.257079541683197,
      "learning_rate": 1.6710118734834655e-05,
      "loss": 0.7796,
      "step": 4450
    },
    {
      "epoch": 0.30337556330244025,
      "grad_norm": 0.2608999013900757,
      "learning_rate": 1.669342324423683e-05,
      "loss": 0.7788,
      "step": 4460
    },
    {
      "epoch": 0.3040557775699345,
      "grad_norm": 0.25176891684532166,
      "learning_rate": 1.6676693884529795e-05,
      "loss": 0.779,
      "step": 4470
    },
    {
      "epoch": 0.3047359918374288,
      "grad_norm": 0.2465016096830368,
      "learning_rate": 1.6659930740365084e-05,
      "loss": 0.7929,
      "step": 4480
    },
    {
      "epoch": 0.30541620610492304,
      "grad_norm": 0.254658579826355,
      "learning_rate": 1.6643133896565174e-05,
      "loss": 0.7916,
      "step": 4490
    },
    {
      "epoch": 0.3060964203724173,
      "grad_norm": 0.2789342999458313,
      "learning_rate": 1.6626303438123067e-05,
      "loss": 0.774,
      "step": 4500
    },
    {
      "epoch": 0.30677663463991156,
      "grad_norm": 0.24634107947349548,
      "learning_rate": 1.6609439450201858e-05,
      "loss": 0.7752,
      "step": 4510
    },
    {
      "epoch": 0.3074568489074058,
      "grad_norm": 0.25199273228645325,
      "learning_rate": 1.6592542018134307e-05,
      "loss": 0.776,
      "step": 4520
    },
    {
      "epoch": 0.3081370631749001,
      "grad_norm": 0.25096115469932556,
      "learning_rate": 1.657561122742239e-05,
      "loss": 0.7851,
      "step": 4530
    },
    {
      "epoch": 0.30881727744239434,
      "grad_norm": 0.2556649148464203,
      "learning_rate": 1.6558647163736896e-05,
      "loss": 0.7796,
      "step": 4540
    },
    {
      "epoch": 0.3094974917098886,
      "grad_norm": 0.24998249113559723,
      "learning_rate": 1.6541649912916967e-05,
      "loss": 0.7771,
      "step": 4550
    },
    {
      "epoch": 0.31017770597738287,
      "grad_norm": 0.2417616993188858,
      "learning_rate": 1.6524619560969673e-05,
      "loss": 0.7801,
      "step": 4560
    },
    {
      "epoch": 0.31085792024487713,
      "grad_norm": 0.24307110905647278,
      "learning_rate": 1.6507556194069584e-05,
      "loss": 0.7843,
      "step": 4570
    },
    {
      "epoch": 0.3115381345123714,
      "grad_norm": 0.24328400194644928,
      "learning_rate": 1.6490459898558326e-05,
      "loss": 0.7737,
      "step": 4580
    },
    {
      "epoch": 0.31221834877986565,
      "grad_norm": 0.25381749868392944,
      "learning_rate": 1.6473330760944143e-05,
      "loss": 0.7746,
      "step": 4590
    },
    {
      "epoch": 0.3128985630473599,
      "grad_norm": 0.25124746561050415,
      "learning_rate": 1.645616886790146e-05,
      "loss": 0.7819,
      "step": 4600
    },
    {
      "epoch": 0.3135787773148542,
      "grad_norm": 0.2559330463409424,
      "learning_rate": 1.6438974306270453e-05,
      "loss": 0.7774,
      "step": 4610
    },
    {
      "epoch": 0.31425899158234843,
      "grad_norm": 0.24517956376075745,
      "learning_rate": 1.64217471630566e-05,
      "loss": 0.78,
      "step": 4620
    },
    {
      "epoch": 0.3149392058498427,
      "grad_norm": 0.2602207660675049,
      "learning_rate": 1.6404487525430248e-05,
      "loss": 0.7799,
      "step": 4630
    },
    {
      "epoch": 0.31561942011733696,
      "grad_norm": 0.2795751690864563,
      "learning_rate": 1.638719548072615e-05,
      "loss": 0.7832,
      "step": 4640
    },
    {
      "epoch": 0.3162996343848312,
      "grad_norm": 0.2633031904697418,
      "learning_rate": 1.6369871116443064e-05,
      "loss": 0.7741,
      "step": 4650
    },
    {
      "epoch": 0.3169798486523255,
      "grad_norm": 0.25834837555885315,
      "learning_rate": 1.635251452024327e-05,
      "loss": 0.7919,
      "step": 4660
    },
    {
      "epoch": 0.31766006291981974,
      "grad_norm": 0.2532989978790283,
      "learning_rate": 1.6335125779952153e-05,
      "loss": 0.7656,
      "step": 4670
    },
    {
      "epoch": 0.318340277187314,
      "grad_norm": 0.2523132264614105,
      "learning_rate": 1.631770498355775e-05,
      "loss": 0.7708,
      "step": 4680
    },
    {
      "epoch": 0.31902049145480826,
      "grad_norm": 0.25467121601104736,
      "learning_rate": 1.630025221921029e-05,
      "loss": 0.7786,
      "step": 4690
    },
    {
      "epoch": 0.3197007057223025,
      "grad_norm": 0.25180891156196594,
      "learning_rate": 1.628276757522178e-05,
      "loss": 0.7823,
      "step": 4700
    },
    {
      "epoch": 0.3203809199897968,
      "grad_norm": 0.2540765404701233,
      "learning_rate": 1.626525114006553e-05,
      "loss": 0.7902,
      "step": 4710
    },
    {
      "epoch": 0.32106113425729105,
      "grad_norm": 0.24241061508655548,
      "learning_rate": 1.624770300237571e-05,
      "loss": 0.7944,
      "step": 4720
    },
    {
      "epoch": 0.3217413485247853,
      "grad_norm": 0.2548539340496063,
      "learning_rate": 1.6230123250946913e-05,
      "loss": 0.7758,
      "step": 4730
    },
    {
      "epoch": 0.32242156279227957,
      "grad_norm": 0.24303984642028809,
      "learning_rate": 1.621251197473371e-05,
      "loss": 0.7802,
      "step": 4740
    },
    {
      "epoch": 0.32310177705977383,
      "grad_norm": 0.25173860788345337,
      "learning_rate": 1.6194869262850165e-05,
      "loss": 0.7695,
      "step": 4750
    },
    {
      "epoch": 0.3237819913272681,
      "grad_norm": 0.2511792480945587,
      "learning_rate": 1.6177195204569432e-05,
      "loss": 0.7739,
      "step": 4760
    },
    {
      "epoch": 0.32446220559476235,
      "grad_norm": 0.26270678639411926,
      "learning_rate": 1.6159489889323266e-05,
      "loss": 0.7921,
      "step": 4770
    },
    {
      "epoch": 0.3251424198622566,
      "grad_norm": 0.2512210011482239,
      "learning_rate": 1.6141753406701582e-05,
      "loss": 0.7776,
      "step": 4780
    },
    {
      "epoch": 0.3258226341297509,
      "grad_norm": 0.2432084083557129,
      "learning_rate": 1.612398584645202e-05,
      "loss": 0.7793,
      "step": 4790
    },
    {
      "epoch": 0.32650284839724514,
      "grad_norm": 0.24282826483249664,
      "learning_rate": 1.6106187298479455e-05,
      "loss": 0.7697,
      "step": 4800
    },
    {
      "epoch": 0.3271830626647394,
      "grad_norm": 0.27392762899398804,
      "learning_rate": 1.608835785284557e-05,
      "loss": 0.7847,
      "step": 4810
    },
    {
      "epoch": 0.32786327693223366,
      "grad_norm": 0.24720722436904907,
      "learning_rate": 1.60704975997684e-05,
      "loss": 0.7678,
      "step": 4820
    },
    {
      "epoch": 0.3285434911997279,
      "grad_norm": 0.2537555694580078,
      "learning_rate": 1.6052606629621856e-05,
      "loss": 0.7774,
      "step": 4830
    },
    {
      "epoch": 0.3292237054672222,
      "grad_norm": 0.2609635591506958,
      "learning_rate": 1.6034685032935277e-05,
      "loss": 0.7801,
      "step": 4840
    },
    {
      "epoch": 0.32990391973471644,
      "grad_norm": 0.2454914003610611,
      "learning_rate": 1.6016732900392987e-05,
      "loss": 0.7844,
      "step": 4850
    },
    {
      "epoch": 0.3305841340022107,
      "grad_norm": 0.2437964379787445,
      "learning_rate": 1.5998750322833814e-05,
      "loss": 0.7694,
      "step": 4860
    },
    {
      "epoch": 0.33126434826970497,
      "grad_norm": 0.2476215809583664,
      "learning_rate": 1.5980737391250643e-05,
      "loss": 0.7864,
      "step": 4870
    },
    {
      "epoch": 0.33194456253719923,
      "grad_norm": 0.24153448641300201,
      "learning_rate": 1.5962694196789947e-05,
      "loss": 0.7679,
      "step": 4880
    },
    {
      "epoch": 0.3326247768046935,
      "grad_norm": 0.26852676272392273,
      "learning_rate": 1.5944620830751342e-05,
      "loss": 0.7768,
      "step": 4890
    },
    {
      "epoch": 0.33330499107218775,
      "grad_norm": 0.2522974908351898,
      "learning_rate": 1.5926517384587097e-05,
      "loss": 0.7886,
      "step": 4900
    },
    {
      "epoch": 0.333985205339682,
      "grad_norm": 0.249611958861351,
      "learning_rate": 1.590838394990171e-05,
      "loss": 0.7885,
      "step": 4910
    },
    {
      "epoch": 0.3346654196071763,
      "grad_norm": 0.24254435300827026,
      "learning_rate": 1.5890220618451397e-05,
      "loss": 0.769,
      "step": 4920
    },
    {
      "epoch": 0.33534563387467053,
      "grad_norm": 0.2485640048980713,
      "learning_rate": 1.587202748214368e-05,
      "loss": 0.7748,
      "step": 4930
    },
    {
      "epoch": 0.3360258481421648,
      "grad_norm": 0.24997614324092865,
      "learning_rate": 1.5853804633036875e-05,
      "loss": 0.7668,
      "step": 4940
    },
    {
      "epoch": 0.33670606240965906,
      "grad_norm": 0.24659042060375214,
      "learning_rate": 1.5835552163339648e-05,
      "loss": 0.7923,
      "step": 4950
    },
    {
      "epoch": 0.3373862766771533,
      "grad_norm": 0.2528908848762512,
      "learning_rate": 1.5817270165410566e-05,
      "loss": 0.7734,
      "step": 4960
    },
    {
      "epoch": 0.3380664909446476,
      "grad_norm": 0.24560825526714325,
      "learning_rate": 1.5798958731757588e-05,
      "loss": 0.7791,
      "step": 4970
    },
    {
      "epoch": 0.33874670521214184,
      "grad_norm": 0.2493152767419815,
      "learning_rate": 1.578061795503763e-05,
      "loss": 0.7606,
      "step": 4980
    },
    {
      "epoch": 0.3394269194796361,
      "grad_norm": 0.24539536237716675,
      "learning_rate": 1.5762247928056085e-05,
      "loss": 0.7937,
      "step": 4990
    },
    {
      "epoch": 0.34010713374713036,
      "grad_norm": 0.24469323456287384,
      "learning_rate": 1.5743848743766358e-05,
      "loss": 0.7973,
      "step": 5000
    },
    {
      "epoch": 0.3407873480146246,
      "grad_norm": 0.2577314078807831,
      "learning_rate": 1.5725420495269378e-05,
      "loss": 0.7773,
      "step": 5010
    },
    {
      "epoch": 0.3414675622821189,
      "grad_norm": 0.2408699095249176,
      "learning_rate": 1.5706963275813155e-05,
      "loss": 0.7908,
      "step": 5020
    },
    {
      "epoch": 0.34214777654961315,
      "grad_norm": 0.24674849212169647,
      "learning_rate": 1.5688477178792288e-05,
      "loss": 0.7625,
      "step": 5030
    },
    {
      "epoch": 0.3428279908171074,
      "grad_norm": 0.24679632484912872,
      "learning_rate": 1.5669962297747497e-05,
      "loss": 0.7744,
      "step": 5040
    },
    {
      "epoch": 0.34350820508460167,
      "grad_norm": 0.25358954071998596,
      "learning_rate": 1.5651418726365155e-05,
      "loss": 0.7795,
      "step": 5050
    },
    {
      "epoch": 0.34418841935209593,
      "grad_norm": 0.25233152508735657,
      "learning_rate": 1.56328465584768e-05,
      "loss": 0.7797,
      "step": 5060
    },
    {
      "epoch": 0.3448686336195902,
      "grad_norm": 0.2582693099975586,
      "learning_rate": 1.5614245888058687e-05,
      "loss": 0.7811,
      "step": 5070
    },
    {
      "epoch": 0.34554884788708445,
      "grad_norm": 0.25043052434921265,
      "learning_rate": 1.5595616809231277e-05,
      "loss": 0.7994,
      "step": 5080
    },
    {
      "epoch": 0.3462290621545787,
      "grad_norm": 0.2512742877006531,
      "learning_rate": 1.5576959416258795e-05,
      "loss": 0.7787,
      "step": 5090
    },
    {
      "epoch": 0.346909276422073,
      "grad_norm": 0.24891647696495056,
      "learning_rate": 1.555827380354872e-05,
      "loss": 0.7867,
      "step": 5100
    },
    {
      "epoch": 0.34758949068956724,
      "grad_norm": 0.2560541331768036,
      "learning_rate": 1.5539560065651342e-05,
      "loss": 0.7905,
      "step": 5110
    },
    {
      "epoch": 0.3482697049570615,
      "grad_norm": 0.24090375006198883,
      "learning_rate": 1.552081829725926e-05,
      "loss": 0.7897,
      "step": 5120
    },
    {
      "epoch": 0.34894991922455576,
      "grad_norm": 0.2456362247467041,
      "learning_rate": 1.5502048593206897e-05,
      "loss": 0.7639,
      "step": 5130
    },
    {
      "epoch": 0.34963013349205,
      "grad_norm": 0.25189900398254395,
      "learning_rate": 1.5483251048470055e-05,
      "loss": 0.7778,
      "step": 5140
    },
    {
      "epoch": 0.3503103477595443,
      "grad_norm": 0.252087265253067,
      "learning_rate": 1.546442575816539e-05,
      "loss": 0.787,
      "step": 5150
    },
    {
      "epoch": 0.35099056202703854,
      "grad_norm": 0.25998103618621826,
      "learning_rate": 1.5445572817549963e-05,
      "loss": 0.7628,
      "step": 5160
    },
    {
      "epoch": 0.3516707762945328,
      "grad_norm": 0.25459837913513184,
      "learning_rate": 1.5426692322020748e-05,
      "loss": 0.7859,
      "step": 5170
    },
    {
      "epoch": 0.352350990562027,
      "grad_norm": 0.2553059756755829,
      "learning_rate": 1.5407784367114143e-05,
      "loss": 0.7805,
      "step": 5180
    },
    {
      "epoch": 0.3530312048295213,
      "grad_norm": 0.2508421242237091,
      "learning_rate": 1.5388849048505496e-05,
      "loss": 0.7674,
      "step": 5190
    },
    {
      "epoch": 0.35371141909701553,
      "grad_norm": 0.2463318556547165,
      "learning_rate": 1.5369886462008617e-05,
      "loss": 0.782,
      "step": 5200
    },
    {
      "epoch": 0.3543916333645098,
      "grad_norm": 0.2445809543132782,
      "learning_rate": 1.5350896703575288e-05,
      "loss": 0.7723,
      "step": 5210
    },
    {
      "epoch": 0.35507184763200406,
      "grad_norm": 0.24156180024147034,
      "learning_rate": 1.533187986929479e-05,
      "loss": 0.7796,
      "step": 5220
    },
    {
      "epoch": 0.3557520618994983,
      "grad_norm": 0.2451319843530655,
      "learning_rate": 1.5312836055393402e-05,
      "loss": 0.7788,
      "step": 5230
    },
    {
      "epoch": 0.3564322761669926,
      "grad_norm": 0.25641563534736633,
      "learning_rate": 1.5293765358233927e-05,
      "loss": 0.7802,
      "step": 5240
    },
    {
      "epoch": 0.35711249043448684,
      "grad_norm": 0.2478734701871872,
      "learning_rate": 1.5274667874315195e-05,
      "loss": 0.7858,
      "step": 5250
    },
    {
      "epoch": 0.3577927047019811,
      "grad_norm": 0.24530655145645142,
      "learning_rate": 1.525554370027159e-05,
      "loss": 0.7742,
      "step": 5260
    },
    {
      "epoch": 0.35847291896947536,
      "grad_norm": 0.24920684099197388,
      "learning_rate": 1.5236392932872525e-05,
      "loss": 0.7684,
      "step": 5270
    },
    {
      "epoch": 0.3591531332369696,
      "grad_norm": 0.26737821102142334,
      "learning_rate": 1.5217215669022002e-05,
      "loss": 0.7722,
      "step": 5280
    },
    {
      "epoch": 0.3598333475044639,
      "grad_norm": 0.2467416673898697,
      "learning_rate": 1.5198012005758083e-05,
      "loss": 0.7755,
      "step": 5290
    },
    {
      "epoch": 0.36051356177195815,
      "grad_norm": 0.26333990693092346,
      "learning_rate": 1.5178782040252417e-05,
      "loss": 0.7973,
      "step": 5300
    },
    {
      "epoch": 0.3611937760394524,
      "grad_norm": 0.2541069984436035,
      "learning_rate": 1.5159525869809746e-05,
      "loss": 0.7802,
      "step": 5310
    },
    {
      "epoch": 0.36187399030694667,
      "grad_norm": 0.24673090875148773,
      "learning_rate": 1.5140243591867405e-05,
      "loss": 0.7775,
      "step": 5320
    },
    {
      "epoch": 0.36255420457444093,
      "grad_norm": 0.2515685260295868,
      "learning_rate": 1.5120935303994833e-05,
      "loss": 0.7836,
      "step": 5330
    },
    {
      "epoch": 0.3632344188419352,
      "grad_norm": 0.27212855219841003,
      "learning_rate": 1.510160110389309e-05,
      "loss": 0.7738,
      "step": 5340
    },
    {
      "epoch": 0.36391463310942945,
      "grad_norm": 0.25533565878868103,
      "learning_rate": 1.5082241089394348e-05,
      "loss": 0.7702,
      "step": 5350
    },
    {
      "epoch": 0.3645948473769237,
      "grad_norm": 0.24824172258377075,
      "learning_rate": 1.5062855358461397e-05,
      "loss": 0.7695,
      "step": 5360
    },
    {
      "epoch": 0.365275061644418,
      "grad_norm": 0.2359800785779953,
      "learning_rate": 1.5043444009187163e-05,
      "loss": 0.7744,
      "step": 5370
    },
    {
      "epoch": 0.36595527591191224,
      "grad_norm": 0.2501261234283447,
      "learning_rate": 1.5024007139794195e-05,
      "loss": 0.7688,
      "step": 5380
    },
    {
      "epoch": 0.3666354901794065,
      "grad_norm": 0.26191291213035583,
      "learning_rate": 1.5004544848634177e-05,
      "loss": 0.7882,
      "step": 5390
    },
    {
      "epoch": 0.36731570444690076,
      "grad_norm": 0.2488580048084259,
      "learning_rate": 1.498505723418743e-05,
      "loss": 0.7738,
      "step": 5400
    },
    {
      "epoch": 0.367995918714395,
      "grad_norm": 0.26761361956596375,
      "learning_rate": 1.4965544395062412e-05,
      "loss": 0.7656,
      "step": 5410
    },
    {
      "epoch": 0.3686761329818893,
      "grad_norm": 0.24776126444339752,
      "learning_rate": 1.4946006429995219e-05,
      "loss": 0.7794,
      "step": 5420
    },
    {
      "epoch": 0.36935634724938354,
      "grad_norm": 0.2511827051639557,
      "learning_rate": 1.4926443437849087e-05,
      "loss": 0.7682,
      "step": 5430
    },
    {
      "epoch": 0.3700365615168778,
      "grad_norm": 0.25186964869499207,
      "learning_rate": 1.4906855517613885e-05,
      "loss": 0.7822,
      "step": 5440
    },
    {
      "epoch": 0.37071677578437207,
      "grad_norm": 0.25966915488243103,
      "learning_rate": 1.4887242768405627e-05,
      "loss": 0.7831,
      "step": 5450
    },
    {
      "epoch": 0.3713969900518663,
      "grad_norm": 0.2607783377170563,
      "learning_rate": 1.4867605289465958e-05,
      "loss": 0.7673,
      "step": 5460
    },
    {
      "epoch": 0.3720772043193606,
      "grad_norm": 0.24937234818935394,
      "learning_rate": 1.4847943180161661e-05,
      "loss": 0.7774,
      "step": 5470
    },
    {
      "epoch": 0.37275741858685485,
      "grad_norm": 0.2692229151725769,
      "learning_rate": 1.4828256539984146e-05,
      "loss": 0.759,
      "step": 5480
    },
    {
      "epoch": 0.3734376328543491,
      "grad_norm": 0.2526799142360687,
      "learning_rate": 1.480854546854895e-05,
      "loss": 0.7817,
      "step": 5490
    },
    {
      "epoch": 0.37411784712184337,
      "grad_norm": 0.25847598910331726,
      "learning_rate": 1.4788810065595238e-05,
      "loss": 0.7826,
      "step": 5500
    },
    {
      "epoch": 0.37479806138933763,
      "grad_norm": 0.27168703079223633,
      "learning_rate": 1.4769050430985292e-05,
      "loss": 0.7792,
      "step": 5510
    },
    {
      "epoch": 0.3754782756568319,
      "grad_norm": 0.25303542613983154,
      "learning_rate": 1.474926666470401e-05,
      "loss": 0.7704,
      "step": 5520
    },
    {
      "epoch": 0.37615848992432616,
      "grad_norm": 0.2534182071685791,
      "learning_rate": 1.4729458866858389e-05,
      "loss": 0.7783,
      "step": 5530
    },
    {
      "epoch": 0.3768387041918204,
      "grad_norm": 0.2554495334625244,
      "learning_rate": 1.4709627137677038e-05,
      "loss": 0.7737,
      "step": 5540
    },
    {
      "epoch": 0.3775189184593147,
      "grad_norm": 0.24515779316425323,
      "learning_rate": 1.468977157750965e-05,
      "loss": 0.7743,
      "step": 5550
    },
    {
      "epoch": 0.37819913272680894,
      "grad_norm": 0.26140493154525757,
      "learning_rate": 1.4669892286826516e-05,
      "loss": 0.7818,
      "step": 5560
    },
    {
      "epoch": 0.3788793469943032,
      "grad_norm": 0.30516576766967773,
      "learning_rate": 1.4649989366217993e-05,
      "loss": 0.7846,
      "step": 5570
    },
    {
      "epoch": 0.37955956126179746,
      "grad_norm": 0.25299209356307983,
      "learning_rate": 1.4630062916394014e-05,
      "loss": 0.7743,
      "step": 5580
    },
    {
      "epoch": 0.3802397755292917,
      "grad_norm": 0.26081040501594543,
      "learning_rate": 1.4610113038183569e-05,
      "loss": 0.784,
      "step": 5590
    },
    {
      "epoch": 0.380919989796786,
      "grad_norm": 0.2585929036140442,
      "learning_rate": 1.4590139832534194e-05,
      "loss": 0.7749,
      "step": 5600
    },
    {
      "epoch": 0.38160020406428025,
      "grad_norm": 0.2520377039909363,
      "learning_rate": 1.457014340051147e-05,
      "loss": 0.7669,
      "step": 5610
    },
    {
      "epoch": 0.3822804183317745,
      "grad_norm": 0.25388965010643005,
      "learning_rate": 1.4550123843298499e-05,
      "loss": 0.7754,
      "step": 5620
    },
    {
      "epoch": 0.38296063259926877,
      "grad_norm": 0.24941669404506683,
      "learning_rate": 1.45300812621954e-05,
      "loss": 0.772,
      "step": 5630
    },
    {
      "epoch": 0.38364084686676303,
      "grad_norm": 0.24526414275169373,
      "learning_rate": 1.4510015758618794e-05,
      "loss": 0.7767,
      "step": 5640
    },
    {
      "epoch": 0.3843210611342573,
      "grad_norm": 0.2574145495891571,
      "learning_rate": 1.448992743410129e-05,
      "loss": 0.7638,
      "step": 5650
    },
    {
      "epoch": 0.38500127540175155,
      "grad_norm": 0.25559088587760925,
      "learning_rate": 1.4469816390290973e-05,
      "loss": 0.7887,
      "step": 5660
    },
    {
      "epoch": 0.3856814896692458,
      "grad_norm": 0.244601309299469,
      "learning_rate": 1.4449682728950895e-05,
      "loss": 0.7873,
      "step": 5670
    },
    {
      "epoch": 0.3863617039367401,
      "grad_norm": 0.24688373506069183,
      "learning_rate": 1.4429526551958541e-05,
      "loss": 0.7796,
      "step": 5680
    },
    {
      "epoch": 0.38704191820423434,
      "grad_norm": 0.25341495871543884,
      "learning_rate": 1.4409347961305339e-05,
      "loss": 0.7797,
      "step": 5690
    },
    {
      "epoch": 0.3877221324717286,
      "grad_norm": 0.26516810059547424,
      "learning_rate": 1.4389147059096121e-05,
      "loss": 0.7776,
      "step": 5700
    },
    {
      "epoch": 0.38840234673922286,
      "grad_norm": 0.2658517062664032,
      "learning_rate": 1.4368923947548626e-05,
      "loss": 0.7665,
      "step": 5710
    },
    {
      "epoch": 0.3890825610067171,
      "grad_norm": 0.25798964500427246,
      "learning_rate": 1.4348678728992967e-05,
      "loss": 0.7956,
      "step": 5720
    },
    {
      "epoch": 0.3897627752742114,
      "grad_norm": 0.2526843845844269,
      "learning_rate": 1.4328411505871126e-05,
      "loss": 0.7783,
      "step": 5730
    },
    {
      "epoch": 0.39044298954170564,
      "grad_norm": 0.25203028321266174,
      "learning_rate": 1.4308122380736419e-05,
      "loss": 0.7608,
      "step": 5740
    },
    {
      "epoch": 0.3911232038091999,
      "grad_norm": 0.24806834757328033,
      "learning_rate": 1.4287811456253e-05,
      "loss": 0.7802,
      "step": 5750
    },
    {
      "epoch": 0.39180341807669417,
      "grad_norm": 0.24196073412895203,
      "learning_rate": 1.4267478835195321e-05,
      "loss": 0.772,
      "step": 5760
    },
    {
      "epoch": 0.3924836323441884,
      "grad_norm": 0.25286272168159485,
      "learning_rate": 1.424712462044762e-05,
      "loss": 0.7623,
      "step": 5770
    },
    {
      "epoch": 0.3931638466116827,
      "grad_norm": 0.25081154704093933,
      "learning_rate": 1.422674891500341e-05,
      "loss": 0.7876,
      "step": 5780
    },
    {
      "epoch": 0.39384406087917695,
      "grad_norm": 0.2481163740158081,
      "learning_rate": 1.420635182196493e-05,
      "loss": 0.7722,
      "step": 5790
    },
    {
      "epoch": 0.3945242751466712,
      "grad_norm": 0.27228838205337524,
      "learning_rate": 1.4185933444542657e-05,
      "loss": 0.7875,
      "step": 5800
    },
    {
      "epoch": 0.39520448941416547,
      "grad_norm": 0.24268393218517303,
      "learning_rate": 1.4165493886054763e-05,
      "loss": 0.777,
      "step": 5810
    },
    {
      "epoch": 0.39588470368165973,
      "grad_norm": 0.25057727098464966,
      "learning_rate": 1.4145033249926598e-05,
      "loss": 0.7737,
      "step": 5820
    },
    {
      "epoch": 0.396564917949154,
      "grad_norm": 0.2459910362958908,
      "learning_rate": 1.4124551639690166e-05,
      "loss": 0.7748,
      "step": 5830
    },
    {
      "epoch": 0.39724513221664826,
      "grad_norm": 0.25080984830856323,
      "learning_rate": 1.4104049158983596e-05,
      "loss": 0.7749,
      "step": 5840
    },
    {
      "epoch": 0.3979253464841425,
      "grad_norm": 0.254576712846756,
      "learning_rate": 1.4083525911550635e-05,
      "loss": 0.7779,
      "step": 5850
    },
    {
      "epoch": 0.3986055607516368,
      "grad_norm": 0.24936212599277496,
      "learning_rate": 1.4062982001240097e-05,
      "loss": 0.7789,
      "step": 5860
    },
    {
      "epoch": 0.39928577501913104,
      "grad_norm": 0.244483083486557,
      "learning_rate": 1.4042417532005361e-05,
      "loss": 0.7636,
      "step": 5870
    },
    {
      "epoch": 0.3999659892866253,
      "grad_norm": 0.25264477729797363,
      "learning_rate": 1.4021832607903831e-05,
      "loss": 0.7868,
      "step": 5880
    },
    {
      "epoch": 0.40064620355411956,
      "grad_norm": 0.26535698771476746,
      "learning_rate": 1.4001227333096416e-05,
      "loss": 0.7796,
      "step": 5890
    },
    {
      "epoch": 0.4013264178216138,
      "grad_norm": 0.2568298876285553,
      "learning_rate": 1.3980601811847001e-05,
      "loss": 0.7648,
      "step": 5900
    },
    {
      "epoch": 0.4020066320891081,
      "grad_norm": 0.2487412691116333,
      "learning_rate": 1.3959956148521916e-05,
      "loss": 0.7641,
      "step": 5910
    },
    {
      "epoch": 0.40268684635660235,
      "grad_norm": 0.24890504777431488,
      "learning_rate": 1.393929044758941e-05,
      "loss": 0.7741,
      "step": 5920
    },
    {
      "epoch": 0.4033670606240966,
      "grad_norm": 0.251572847366333,
      "learning_rate": 1.3918604813619132e-05,
      "loss": 0.796,
      "step": 5930
    },
    {
      "epoch": 0.40404727489159087,
      "grad_norm": 0.23951993882656097,
      "learning_rate": 1.3897899351281585e-05,
      "loss": 0.7578,
      "step": 5940
    },
    {
      "epoch": 0.40472748915908513,
      "grad_norm": 0.2541947066783905,
      "learning_rate": 1.3877174165347606e-05,
      "loss": 0.7668,
      "step": 5950
    },
    {
      "epoch": 0.4054077034265794,
      "grad_norm": 0.26087838411331177,
      "learning_rate": 1.385642936068784e-05,
      "loss": 0.7788,
      "step": 5960
    },
    {
      "epoch": 0.40608791769407365,
      "grad_norm": 0.24299927055835724,
      "learning_rate": 1.3835665042272195e-05,
      "loss": 0.7744,
      "step": 5970
    },
    {
      "epoch": 0.4067681319615679,
      "grad_norm": 0.2495705932378769,
      "learning_rate": 1.3814881315169328e-05,
      "loss": 0.767,
      "step": 5980
    },
    {
      "epoch": 0.4074483462290622,
      "grad_norm": 0.24570466578006744,
      "learning_rate": 1.37940782845461e-05,
      "loss": 0.7739,
      "step": 5990
    },
    {
      "epoch": 0.40812856049655644,
      "grad_norm": 0.24526531994342804,
      "learning_rate": 1.3773256055667053e-05,
      "loss": 0.7703,
      "step": 6000
    },
    {
      "epoch": 0.4088087747640507,
      "grad_norm": 0.2453545182943344,
      "learning_rate": 1.3752414733893866e-05,
      "loss": 0.7771,
      "step": 6010
    },
    {
      "epoch": 0.40948898903154496,
      "grad_norm": 0.26506248116493225,
      "learning_rate": 1.3731554424684839e-05,
      "loss": 0.7818,
      "step": 6020
    },
    {
      "epoch": 0.4101692032990392,
      "grad_norm": 0.2656206786632538,
      "learning_rate": 1.3710675233594342e-05,
      "loss": 0.7724,
      "step": 6030
    },
    {
      "epoch": 0.4108494175665335,
      "grad_norm": 0.24015671014785767,
      "learning_rate": 1.3689777266272292e-05,
      "loss": 0.7647,
      "step": 6040
    },
    {
      "epoch": 0.41152963183402774,
      "grad_norm": 0.24293836951255798,
      "learning_rate": 1.3668860628463611e-05,
      "loss": 0.7673,
      "step": 6050
    },
    {
      "epoch": 0.412209846101522,
      "grad_norm": 0.2461855262517929,
      "learning_rate": 1.3647925426007703e-05,
      "loss": 0.7577,
      "step": 6060
    },
    {
      "epoch": 0.41289006036901627,
      "grad_norm": 0.24776321649551392,
      "learning_rate": 1.3626971764837903e-05,
      "loss": 0.7826,
      "step": 6070
    },
    {
      "epoch": 0.4135702746365105,
      "grad_norm": 0.23962141573429108,
      "learning_rate": 1.3605999750980943e-05,
      "loss": 0.7703,
      "step": 6080
    },
    {
      "epoch": 0.4142504889040048,
      "grad_norm": 0.2501792013645172,
      "learning_rate": 1.3585009490556435e-05,
      "loss": 0.7685,
      "step": 6090
    },
    {
      "epoch": 0.41493070317149905,
      "grad_norm": 0.25442349910736084,
      "learning_rate": 1.3564001089776313e-05,
      "loss": 0.7585,
      "step": 6100
    },
    {
      "epoch": 0.4156109174389933,
      "grad_norm": 0.2500801682472229,
      "learning_rate": 1.3542974654944297e-05,
      "loss": 0.7649,
      "step": 6110
    },
    {
      "epoch": 0.4162911317064875,
      "grad_norm": 0.25212499499320984,
      "learning_rate": 1.3521930292455363e-05,
      "loss": 0.7607,
      "step": 6120
    },
    {
      "epoch": 0.4169713459739818,
      "grad_norm": 0.24636310338974,
      "learning_rate": 1.3500868108795213e-05,
      "loss": 0.7744,
      "step": 6130
    },
    {
      "epoch": 0.41765156024147604,
      "grad_norm": 0.24590402841567993,
      "learning_rate": 1.3479788210539705e-05,
      "loss": 0.7706,
      "step": 6140
    },
    {
      "epoch": 0.4183317745089703,
      "grad_norm": 0.2557471990585327,
      "learning_rate": 1.3458690704354353e-05,
      "loss": 0.7683,
      "step": 6150
    },
    {
      "epoch": 0.41901198877646456,
      "grad_norm": 0.250740110874176,
      "learning_rate": 1.3437575696993756e-05,
      "loss": 0.7672,
      "step": 6160
    },
    {
      "epoch": 0.4196922030439588,
      "grad_norm": 0.23877675831317902,
      "learning_rate": 1.3416443295301076e-05,
      "loss": 0.7613,
      "step": 6170
    },
    {
      "epoch": 0.4203724173114531,
      "grad_norm": 0.2432548552751541,
      "learning_rate": 1.3395293606207487e-05,
      "loss": 0.7752,
      "step": 6180
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 0.24699924886226654,
      "learning_rate": 1.3374126736731643e-05,
      "loss": 0.7823,
      "step": 6190
    },
    {
      "epoch": 0.4217328458464416,
      "grad_norm": 0.24633049964904785,
      "learning_rate": 1.3352942793979127e-05,
      "loss": 0.774,
      "step": 6200
    },
    {
      "epoch": 0.42241306011393587,
      "grad_norm": 0.25033462047576904,
      "learning_rate": 1.3331741885141915e-05,
      "loss": 0.7626,
      "step": 6210
    },
    {
      "epoch": 0.42309327438143013,
      "grad_norm": 0.2341737002134323,
      "learning_rate": 1.3310524117497839e-05,
      "loss": 0.7777,
      "step": 6220
    },
    {
      "epoch": 0.4237734886489244,
      "grad_norm": 0.2521795928478241,
      "learning_rate": 1.3289289598410024e-05,
      "loss": 0.7727,
      "step": 6230
    },
    {
      "epoch": 0.42445370291641865,
      "grad_norm": 0.24901489913463593,
      "learning_rate": 1.3268038435326374e-05,
      "loss": 0.7881,
      "step": 6240
    },
    {
      "epoch": 0.4251339171839129,
      "grad_norm": 0.2500702142715454,
      "learning_rate": 1.3246770735778998e-05,
      "loss": 0.7692,
      "step": 6250
    },
    {
      "epoch": 0.4258141314514072,
      "grad_norm": 0.24418380856513977,
      "learning_rate": 1.3225486607383693e-05,
      "loss": 0.7714,
      "step": 6260
    },
    {
      "epoch": 0.42649434571890144,
      "grad_norm": 0.24833431839942932,
      "learning_rate": 1.3204186157839379e-05,
      "loss": 0.7714,
      "step": 6270
    },
    {
      "epoch": 0.4271745599863957,
      "grad_norm": 0.2640336751937866,
      "learning_rate": 1.3182869494927562e-05,
      "loss": 0.7642,
      "step": 6280
    },
    {
      "epoch": 0.42785477425388996,
      "grad_norm": 0.2517968416213989,
      "learning_rate": 1.3161536726511791e-05,
      "loss": 0.782,
      "step": 6290
    },
    {
      "epoch": 0.4285349885213842,
      "grad_norm": 0.254301518201828,
      "learning_rate": 1.3140187960537113e-05,
      "loss": 0.7703,
      "step": 6300
    },
    {
      "epoch": 0.4292152027888785,
      "grad_norm": 0.24380703270435333,
      "learning_rate": 1.311882330502952e-05,
      "loss": 0.7648,
      "step": 6310
    },
    {
      "epoch": 0.42989541705637274,
      "grad_norm": 0.2415427565574646,
      "learning_rate": 1.3097442868095403e-05,
      "loss": 0.7791,
      "step": 6320
    },
    {
      "epoch": 0.430575631323867,
      "grad_norm": 0.24330158531665802,
      "learning_rate": 1.3076046757921011e-05,
      "loss": 0.7698,
      "step": 6330
    },
    {
      "epoch": 0.43125584559136126,
      "grad_norm": 0.25078320503234863,
      "learning_rate": 1.30546350827719e-05,
      "loss": 0.7741,
      "step": 6340
    },
    {
      "epoch": 0.4319360598588555,
      "grad_norm": 0.2747834026813507,
      "learning_rate": 1.303320795099239e-05,
      "loss": 0.7742,
      "step": 6350
    },
    {
      "epoch": 0.4326162741263498,
      "grad_norm": 0.24714337289333344,
      "learning_rate": 1.3011765471005004e-05,
      "loss": 0.7669,
      "step": 6360
    },
    {
      "epoch": 0.43329648839384405,
      "grad_norm": 0.25236451625823975,
      "learning_rate": 1.2990307751309936e-05,
      "loss": 0.7818,
      "step": 6370
    },
    {
      "epoch": 0.4339767026613383,
      "grad_norm": 0.26118069887161255,
      "learning_rate": 1.2968834900484489e-05,
      "loss": 0.7619,
      "step": 6380
    },
    {
      "epoch": 0.43465691692883257,
      "grad_norm": 0.25219205021858215,
      "learning_rate": 1.2947347027182531e-05,
      "loss": 0.7694,
      "step": 6390
    },
    {
      "epoch": 0.43533713119632683,
      "grad_norm": 0.24675573408603668,
      "learning_rate": 1.2925844240133945e-05,
      "loss": 0.7801,
      "step": 6400
    },
    {
      "epoch": 0.4360173454638211,
      "grad_norm": 0.24943795800209045,
      "learning_rate": 1.290432664814408e-05,
      "loss": 0.7956,
      "step": 6410
    },
    {
      "epoch": 0.43669755973131535,
      "grad_norm": 0.25274989008903503,
      "learning_rate": 1.2882794360093196e-05,
      "loss": 0.7821,
      "step": 6420
    },
    {
      "epoch": 0.4373777739988096,
      "grad_norm": 0.24581989645957947,
      "learning_rate": 1.2861247484935917e-05,
      "loss": 0.7698,
      "step": 6430
    },
    {
      "epoch": 0.4380579882663039,
      "grad_norm": 0.24845002591609955,
      "learning_rate": 1.283968613170068e-05,
      "loss": 0.7598,
      "step": 6440
    },
    {
      "epoch": 0.43873820253379814,
      "grad_norm": 0.24644219875335693,
      "learning_rate": 1.2818110409489184e-05,
      "loss": 0.766,
      "step": 6450
    },
    {
      "epoch": 0.4394184168012924,
      "grad_norm": 0.26946285367012024,
      "learning_rate": 1.2796520427475827e-05,
      "loss": 0.7848,
      "step": 6460
    },
    {
      "epoch": 0.44009863106878666,
      "grad_norm": 0.25460585951805115,
      "learning_rate": 1.2774916294907171e-05,
      "loss": 0.7803,
      "step": 6470
    },
    {
      "epoch": 0.4407788453362809,
      "grad_norm": 0.24377070367336273,
      "learning_rate": 1.2753298121101378e-05,
      "loss": 0.7768,
      "step": 6480
    },
    {
      "epoch": 0.4414590596037752,
      "grad_norm": 0.23929229378700256,
      "learning_rate": 1.2731666015447659e-05,
      "loss": 0.7773,
      "step": 6490
    },
    {
      "epoch": 0.44213927387126944,
      "grad_norm": 0.25252291560173035,
      "learning_rate": 1.2710020087405721e-05,
      "loss": 0.7825,
      "step": 6500
    },
    {
      "epoch": 0.4428194881387637,
      "grad_norm": 0.2470589131116867,
      "learning_rate": 1.2688360446505213e-05,
      "loss": 0.7599,
      "step": 6510
    },
    {
      "epoch": 0.44349970240625797,
      "grad_norm": 0.24808578193187714,
      "learning_rate": 1.2666687202345171e-05,
      "loss": 0.7761,
      "step": 6520
    },
    {
      "epoch": 0.44417991667375223,
      "grad_norm": 0.26062536239624023,
      "learning_rate": 1.264500046459347e-05,
      "loss": 0.7659,
      "step": 6530
    },
    {
      "epoch": 0.4448601309412465,
      "grad_norm": 0.2558269500732422,
      "learning_rate": 1.2623300342986256e-05,
      "loss": 0.77,
      "step": 6540
    },
    {
      "epoch": 0.44554034520874075,
      "grad_norm": 0.24331645667552948,
      "learning_rate": 1.26015869473274e-05,
      "loss": 0.7849,
      "step": 6550
    },
    {
      "epoch": 0.446220559476235,
      "grad_norm": 0.2490650862455368,
      "learning_rate": 1.2579860387487941e-05,
      "loss": 0.7692,
      "step": 6560
    },
    {
      "epoch": 0.4469007737437293,
      "grad_norm": 0.252430260181427,
      "learning_rate": 1.2558120773405536e-05,
      "loss": 0.7677,
      "step": 6570
    },
    {
      "epoch": 0.44758098801122354,
      "grad_norm": 0.24718795716762543,
      "learning_rate": 1.2536368215083883e-05,
      "loss": 0.7708,
      "step": 6580
    },
    {
      "epoch": 0.4482612022787178,
      "grad_norm": 0.25127896666526794,
      "learning_rate": 1.2514602822592189e-05,
      "loss": 0.7753,
      "step": 6590
    },
    {
      "epoch": 0.44894141654621206,
      "grad_norm": 0.2372049242258072,
      "learning_rate": 1.2492824706064602e-05,
      "loss": 0.7749,
      "step": 6600
    },
    {
      "epoch": 0.4496216308137063,
      "grad_norm": 0.23758196830749512,
      "learning_rate": 1.247103397569965e-05,
      "loss": 0.782,
      "step": 6610
    },
    {
      "epoch": 0.4503018450812006,
      "grad_norm": 0.2627807855606079,
      "learning_rate": 1.2449230741759694e-05,
      "loss": 0.7667,
      "step": 6620
    },
    {
      "epoch": 0.45098205934869484,
      "grad_norm": 0.2461169809103012,
      "learning_rate": 1.2427415114570361e-05,
      "loss": 0.761,
      "step": 6630
    },
    {
      "epoch": 0.4516622736161891,
      "grad_norm": 0.2529158890247345,
      "learning_rate": 1.2405587204519984e-05,
      "loss": 0.7693,
      "step": 6640
    },
    {
      "epoch": 0.45234248788368336,
      "grad_norm": 0.2508241832256317,
      "learning_rate": 1.2383747122059054e-05,
      "loss": 0.7662,
      "step": 6650
    },
    {
      "epoch": 0.4530227021511776,
      "grad_norm": 0.2500368058681488,
      "learning_rate": 1.2361894977699651e-05,
      "loss": 0.7759,
      "step": 6660
    },
    {
      "epoch": 0.4537029164186719,
      "grad_norm": 0.24708276987075806,
      "learning_rate": 1.2340030882014892e-05,
      "loss": 0.7745,
      "step": 6670
    },
    {
      "epoch": 0.45438313068616615,
      "grad_norm": 0.25018182396888733,
      "learning_rate": 1.2318154945638368e-05,
      "loss": 0.7888,
      "step": 6680
    },
    {
      "epoch": 0.4550633449536604,
      "grad_norm": 0.24344618618488312,
      "learning_rate": 1.2296267279263586e-05,
      "loss": 0.7786,
      "step": 6690
    },
    {
      "epoch": 0.45574355922115467,
      "grad_norm": 0.24326354265213013,
      "learning_rate": 1.2274367993643399e-05,
      "loss": 0.7673,
      "step": 6700
    },
    {
      "epoch": 0.45642377348864893,
      "grad_norm": 0.24483905732631683,
      "learning_rate": 1.2252457199589464e-05,
      "loss": 0.7721,
      "step": 6710
    },
    {
      "epoch": 0.4571039877561432,
      "grad_norm": 0.2535827159881592,
      "learning_rate": 1.223053500797167e-05,
      "loss": 0.7679,
      "step": 6720
    },
    {
      "epoch": 0.45778420202363745,
      "grad_norm": 0.2524265646934509,
      "learning_rate": 1.2208601529717573e-05,
      "loss": 0.7804,
      "step": 6730
    },
    {
      "epoch": 0.4584644162911317,
      "grad_norm": 0.2596147060394287,
      "learning_rate": 1.2186656875811841e-05,
      "loss": 0.7721,
      "step": 6740
    },
    {
      "epoch": 0.459144630558626,
      "grad_norm": 0.2450697124004364,
      "learning_rate": 1.21647011572957e-05,
      "loss": 0.7761,
      "step": 6750
    },
    {
      "epoch": 0.45982484482612024,
      "grad_norm": 0.25916561484336853,
      "learning_rate": 1.214273448526635e-05,
      "loss": 0.7834,
      "step": 6760
    },
    {
      "epoch": 0.4605050590936145,
      "grad_norm": 0.2503756284713745,
      "learning_rate": 1.2120756970876429e-05,
      "loss": 0.7829,
      "step": 6770
    },
    {
      "epoch": 0.46118527336110876,
      "grad_norm": 0.24557699263095856,
      "learning_rate": 1.209876872533343e-05,
      "loss": 0.7839,
      "step": 6780
    },
    {
      "epoch": 0.461865487628603,
      "grad_norm": 0.2489149421453476,
      "learning_rate": 1.2076769859899149e-05,
      "loss": 0.7775,
      "step": 6790
    },
    {
      "epoch": 0.4625457018960973,
      "grad_norm": 0.25012078881263733,
      "learning_rate": 1.2054760485889116e-05,
      "loss": 0.7724,
      "step": 6800
    },
    {
      "epoch": 0.46322591616359154,
      "grad_norm": 0.2612060308456421,
      "learning_rate": 1.2032740714672044e-05,
      "loss": 0.7624,
      "step": 6810
    },
    {
      "epoch": 0.4639061304310858,
      "grad_norm": 0.24063514173030853,
      "learning_rate": 1.2010710657669246e-05,
      "loss": 0.7899,
      "step": 6820
    },
    {
      "epoch": 0.46458634469858007,
      "grad_norm": 0.25464415550231934,
      "learning_rate": 1.1988670426354087e-05,
      "loss": 0.7932,
      "step": 6830
    },
    {
      "epoch": 0.46526655896607433,
      "grad_norm": 0.2486640214920044,
      "learning_rate": 1.1966620132251414e-05,
      "loss": 0.7872,
      "step": 6840
    },
    {
      "epoch": 0.4659467732335686,
      "grad_norm": 0.2476149946451187,
      "learning_rate": 1.194455988693699e-05,
      "loss": 0.7583,
      "step": 6850
    },
    {
      "epoch": 0.46662698750106285,
      "grad_norm": 0.2592043876647949,
      "learning_rate": 1.192248980203694e-05,
      "loss": 0.7682,
      "step": 6860
    },
    {
      "epoch": 0.4673072017685571,
      "grad_norm": 0.2530962824821472,
      "learning_rate": 1.1900409989227167e-05,
      "loss": 0.7763,
      "step": 6870
    },
    {
      "epoch": 0.4679874160360514,
      "grad_norm": 0.24381905794143677,
      "learning_rate": 1.1878320560232801e-05,
      "loss": 0.7773,
      "step": 6880
    },
    {
      "epoch": 0.46866763030354563,
      "grad_norm": 0.24828781187534332,
      "learning_rate": 1.1856221626827635e-05,
      "loss": 0.7681,
      "step": 6890
    },
    {
      "epoch": 0.4693478445710399,
      "grad_norm": 0.25107139348983765,
      "learning_rate": 1.183411330083355e-05,
      "loss": 0.7798,
      "step": 6900
    },
    {
      "epoch": 0.47002805883853416,
      "grad_norm": 0.2627694308757782,
      "learning_rate": 1.1811995694119956e-05,
      "loss": 0.7785,
      "step": 6910
    },
    {
      "epoch": 0.4707082731060284,
      "grad_norm": 0.34391316771507263,
      "learning_rate": 1.1789868918603223e-05,
      "loss": 0.7668,
      "step": 6920
    },
    {
      "epoch": 0.4713884873735227,
      "grad_norm": 0.24955721199512482,
      "learning_rate": 1.1767733086246116e-05,
      "loss": 0.7582,
      "step": 6930
    },
    {
      "epoch": 0.47206870164101694,
      "grad_norm": 0.26111745834350586,
      "learning_rate": 1.1745588309057232e-05,
      "loss": 0.7702,
      "step": 6940
    },
    {
      "epoch": 0.4727489159085112,
      "grad_norm": 0.2527529299259186,
      "learning_rate": 1.1723434699090423e-05,
      "loss": 0.7683,
      "step": 6950
    },
    {
      "epoch": 0.47342913017600546,
      "grad_norm": 0.2523750960826874,
      "learning_rate": 1.1701272368444239e-05,
      "loss": 0.7865,
      "step": 6960
    },
    {
      "epoch": 0.4741093444434997,
      "grad_norm": 0.24606585502624512,
      "learning_rate": 1.1679101429261357e-05,
      "loss": 0.7718,
      "step": 6970
    },
    {
      "epoch": 0.474789558710994,
      "grad_norm": 0.24632257223129272,
      "learning_rate": 1.1656921993728013e-05,
      "loss": 0.759,
      "step": 6980
    },
    {
      "epoch": 0.47546977297848825,
      "grad_norm": 0.2509964406490326,
      "learning_rate": 1.1634734174073431e-05,
      "loss": 0.7706,
      "step": 6990
    },
    {
      "epoch": 0.4761499872459825,
      "grad_norm": 0.24110184609889984,
      "learning_rate": 1.161253808256927e-05,
      "loss": 0.7766,
      "step": 7000
    },
    {
      "epoch": 0.47683020151347677,
      "grad_norm": 0.264956533908844,
      "learning_rate": 1.1590333831529034e-05,
      "loss": 0.7798,
      "step": 7010
    },
    {
      "epoch": 0.47751041578097103,
      "grad_norm": 0.2527983486652374,
      "learning_rate": 1.1568121533307522e-05,
      "loss": 0.7744,
      "step": 7020
    },
    {
      "epoch": 0.4781906300484653,
      "grad_norm": 0.24862922728061676,
      "learning_rate": 1.1545901300300244e-05,
      "loss": 0.766,
      "step": 7030
    },
    {
      "epoch": 0.47887084431595955,
      "grad_norm": 0.25299084186553955,
      "learning_rate": 1.152367324494287e-05,
      "loss": 0.7594,
      "step": 7040
    },
    {
      "epoch": 0.4795510585834538,
      "grad_norm": 0.24184784293174744,
      "learning_rate": 1.1501437479710646e-05,
      "loss": 0.7669,
      "step": 7050
    },
    {
      "epoch": 0.480231272850948,
      "grad_norm": 0.25336650013923645,
      "learning_rate": 1.1479194117117833e-05,
      "loss": 0.7811,
      "step": 7060
    },
    {
      "epoch": 0.4809114871184423,
      "grad_norm": 0.25547850131988525,
      "learning_rate": 1.1456943269717133e-05,
      "loss": 0.7631,
      "step": 7070
    },
    {
      "epoch": 0.48159170138593654,
      "grad_norm": 0.24883776903152466,
      "learning_rate": 1.1434685050099122e-05,
      "loss": 0.7724,
      "step": 7080
    },
    {
      "epoch": 0.4822719156534308,
      "grad_norm": 0.2447323352098465,
      "learning_rate": 1.141241957089168e-05,
      "loss": 0.7641,
      "step": 7090
    },
    {
      "epoch": 0.48295212992092507,
      "grad_norm": 0.25550904870033264,
      "learning_rate": 1.1390146944759421e-05,
      "loss": 0.7855,
      "step": 7100
    },
    {
      "epoch": 0.4836323441884193,
      "grad_norm": 0.26494309306144714,
      "learning_rate": 1.1367867284403123e-05,
      "loss": 0.7702,
      "step": 7110
    },
    {
      "epoch": 0.4843125584559136,
      "grad_norm": 0.2557946443557739,
      "learning_rate": 1.1345580702559158e-05,
      "loss": 0.7699,
      "step": 7120
    },
    {
      "epoch": 0.48499277272340785,
      "grad_norm": 0.25180885195732117,
      "learning_rate": 1.1323287311998921e-05,
      "loss": 0.7787,
      "step": 7130
    },
    {
      "epoch": 0.4856729869909021,
      "grad_norm": 0.24362744390964508,
      "learning_rate": 1.1300987225528257e-05,
      "loss": 0.7743,
      "step": 7140
    },
    {
      "epoch": 0.4863532012583964,
      "grad_norm": 0.2375001758337021,
      "learning_rate": 1.1278680555986898e-05,
      "loss": 0.7703,
      "step": 7150
    },
    {
      "epoch": 0.48703341552589063,
      "grad_norm": 0.2449880689382553,
      "learning_rate": 1.1256367416247882e-05,
      "loss": 0.766,
      "step": 7160
    },
    {
      "epoch": 0.4877136297933849,
      "grad_norm": 0.25702130794525146,
      "learning_rate": 1.1234047919216986e-05,
      "loss": 0.7898,
      "step": 7170
    },
    {
      "epoch": 0.48839384406087916,
      "grad_norm": 0.254113107919693,
      "learning_rate": 1.1211722177832163e-05,
      "loss": 0.7806,
      "step": 7180
    },
    {
      "epoch": 0.4890740583283734,
      "grad_norm": 0.253002405166626,
      "learning_rate": 1.118939030506295e-05,
      "loss": 0.7797,
      "step": 7190
    },
    {
      "epoch": 0.4897542725958677,
      "grad_norm": 0.24663403630256653,
      "learning_rate": 1.1167052413909925e-05,
      "loss": 0.764,
      "step": 7200
    },
    {
      "epoch": 0.49043448686336194,
      "grad_norm": 0.2504352331161499,
      "learning_rate": 1.1144708617404101e-05,
      "loss": 0.7721,
      "step": 7210
    },
    {
      "epoch": 0.4911147011308562,
      "grad_norm": 0.25107845664024353,
      "learning_rate": 1.1122359028606391e-05,
      "loss": 0.7737,
      "step": 7220
    },
    {
      "epoch": 0.49179491539835046,
      "grad_norm": 0.2497549057006836,
      "learning_rate": 1.1100003760607001e-05,
      "loss": 0.7644,
      "step": 7230
    },
    {
      "epoch": 0.4924751296658447,
      "grad_norm": 0.2447914481163025,
      "learning_rate": 1.1077642926524887e-05,
      "loss": 0.7779,
      "step": 7240
    },
    {
      "epoch": 0.493155343933339,
      "grad_norm": 0.25073736906051636,
      "learning_rate": 1.1055276639507165e-05,
      "loss": 0.7662,
      "step": 7250
    },
    {
      "epoch": 0.49383555820083325,
      "grad_norm": 0.24794918298721313,
      "learning_rate": 1.1032905012728536e-05,
      "loss": 0.7868,
      "step": 7260
    },
    {
      "epoch": 0.4945157724683275,
      "grad_norm": 0.2526129484176636,
      "learning_rate": 1.1010528159390732e-05,
      "loss": 0.7812,
      "step": 7270
    },
    {
      "epoch": 0.49519598673582177,
      "grad_norm": 0.24096263945102692,
      "learning_rate": 1.0988146192721927e-05,
      "loss": 0.7626,
      "step": 7280
    },
    {
      "epoch": 0.49587620100331603,
      "grad_norm": 0.25271764397621155,
      "learning_rate": 1.0965759225976165e-05,
      "loss": 0.7688,
      "step": 7290
    },
    {
      "epoch": 0.4965564152708103,
      "grad_norm": 0.24893783032894135,
      "learning_rate": 1.0943367372432796e-05,
      "loss": 0.7684,
      "step": 7300
    },
    {
      "epoch": 0.49723662953830455,
      "grad_norm": 0.24898919463157654,
      "learning_rate": 1.0920970745395897e-05,
      "loss": 0.7688,
      "step": 7310
    },
    {
      "epoch": 0.4979168438057988,
      "grad_norm": 0.24341994524002075,
      "learning_rate": 1.0898569458193695e-05,
      "loss": 0.7841,
      "step": 7320
    },
    {
      "epoch": 0.4985970580732931,
      "grad_norm": 0.25512826442718506,
      "learning_rate": 1.0876163624178004e-05,
      "loss": 0.7722,
      "step": 7330
    },
    {
      "epoch": 0.49927727234078734,
      "grad_norm": 0.2560203969478607,
      "learning_rate": 1.0853753356723637e-05,
      "loss": 0.7592,
      "step": 7340
    },
    {
      "epoch": 0.4999574866082816,
      "grad_norm": 0.25013265013694763,
      "learning_rate": 1.0831338769227848e-05,
      "loss": 0.7797,
      "step": 7350
    },
    {
      "epoch": 0.5006377008757759,
      "grad_norm": 0.24355962872505188,
      "learning_rate": 1.0808919975109746e-05,
      "loss": 0.7728,
      "step": 7360
    },
    {
      "epoch": 0.5013179151432702,
      "grad_norm": 0.2452683001756668,
      "learning_rate": 1.0786497087809731e-05,
      "loss": 0.7852,
      "step": 7370
    },
    {
      "epoch": 0.5019981294107644,
      "grad_norm": 0.2576746642589569,
      "learning_rate": 1.0764070220788906e-05,
      "loss": 0.7715,
      "step": 7380
    },
    {
      "epoch": 0.5026783436782587,
      "grad_norm": 0.25701427459716797,
      "learning_rate": 1.074163948752852e-05,
      "loss": 0.7653,
      "step": 7390
    },
    {
      "epoch": 0.5033585579457529,
      "grad_norm": 0.2555880844593048,
      "learning_rate": 1.0719205001529382e-05,
      "loss": 0.7741,
      "step": 7400
    },
    {
      "epoch": 0.5040387722132472,
      "grad_norm": 0.24719035625457764,
      "learning_rate": 1.069676687631129e-05,
      "loss": 0.7655,
      "step": 7410
    },
    {
      "epoch": 0.5047189864807414,
      "grad_norm": 0.24667803943157196,
      "learning_rate": 1.0674325225412456e-05,
      "loss": 0.7597,
      "step": 7420
    },
    {
      "epoch": 0.5053992007482357,
      "grad_norm": 0.2403658777475357,
      "learning_rate": 1.065188016238893e-05,
      "loss": 0.7848,
      "step": 7430
    },
    {
      "epoch": 0.50607941501573,
      "grad_norm": 0.24025067687034607,
      "learning_rate": 1.062943180081404e-05,
      "loss": 0.7586,
      "step": 7440
    },
    {
      "epoch": 0.5067596292832243,
      "grad_norm": 0.25513070821762085,
      "learning_rate": 1.060698025427779e-05,
      "loss": 0.7797,
      "step": 7450
    },
    {
      "epoch": 0.5074398435507185,
      "grad_norm": 0.2516796290874481,
      "learning_rate": 1.0584525636386302e-05,
      "loss": 0.7764,
      "step": 7460
    },
    {
      "epoch": 0.5081200578182128,
      "grad_norm": 0.2510603666305542,
      "learning_rate": 1.0562068060761249e-05,
      "loss": 0.7775,
      "step": 7470
    },
    {
      "epoch": 0.508800272085707,
      "grad_norm": 0.24026992917060852,
      "learning_rate": 1.0539607641039258e-05,
      "loss": 0.7642,
      "step": 7480
    },
    {
      "epoch": 0.5094804863532013,
      "grad_norm": 0.24698439240455627,
      "learning_rate": 1.0517144490871358e-05,
      "loss": 0.7826,
      "step": 7490
    },
    {
      "epoch": 0.5101607006206955,
      "grad_norm": 0.2510314881801605,
      "learning_rate": 1.049467872392239e-05,
      "loss": 0.7712,
      "step": 7500
    },
    {
      "epoch": 0.5108409148881898,
      "grad_norm": 0.2500401437282562,
      "learning_rate": 1.0472210453870433e-05,
      "loss": 0.7804,
      "step": 7510
    },
    {
      "epoch": 0.511521129155684,
      "grad_norm": 0.25299274921417236,
      "learning_rate": 1.0449739794406236e-05,
      "loss": 0.7746,
      "step": 7520
    },
    {
      "epoch": 0.5122013434231782,
      "grad_norm": 0.2486591637134552,
      "learning_rate": 1.0427266859232634e-05,
      "loss": 0.7668,
      "step": 7530
    },
    {
      "epoch": 0.5128815576906726,
      "grad_norm": 0.2587043344974518,
      "learning_rate": 1.0404791762063983e-05,
      "loss": 0.7554,
      "step": 7540
    },
    {
      "epoch": 0.5135617719581668,
      "grad_norm": 0.2467149943113327,
      "learning_rate": 1.038231461662558e-05,
      "loss": 0.7643,
      "step": 7550
    },
    {
      "epoch": 0.5142419862256611,
      "grad_norm": 0.24870260059833527,
      "learning_rate": 1.0359835536653073e-05,
      "loss": 0.7725,
      "step": 7560
    },
    {
      "epoch": 0.5149222004931553,
      "grad_norm": 0.24920696020126343,
      "learning_rate": 1.0337354635891919e-05,
      "loss": 0.7609,
      "step": 7570
    },
    {
      "epoch": 0.5156024147606496,
      "grad_norm": 0.24223774671554565,
      "learning_rate": 1.031487202809677e-05,
      "loss": 0.762,
      "step": 7580
    },
    {
      "epoch": 0.5162826290281438,
      "grad_norm": 0.24686753749847412,
      "learning_rate": 1.0292387827030926e-05,
      "loss": 0.7804,
      "step": 7590
    },
    {
      "epoch": 0.5169628432956381,
      "grad_norm": 0.2514561712741852,
      "learning_rate": 1.0269902146465744e-05,
      "loss": 0.7664,
      "step": 7600
    },
    {
      "epoch": 0.5176430575631323,
      "grad_norm": 0.24237996339797974,
      "learning_rate": 1.0247415100180073e-05,
      "loss": 0.7754,
      "step": 7610
    },
    {
      "epoch": 0.5183232718306267,
      "grad_norm": 0.2523074448108673,
      "learning_rate": 1.0224926801959672e-05,
      "loss": 0.7787,
      "step": 7620
    },
    {
      "epoch": 0.5190034860981209,
      "grad_norm": 0.25330936908721924,
      "learning_rate": 1.0202437365596625e-05,
      "loss": 0.78,
      "step": 7630
    },
    {
      "epoch": 0.5196837003656152,
      "grad_norm": 0.25370344519615173,
      "learning_rate": 1.0179946904888784e-05,
      "loss": 0.7701,
      "step": 7640
    },
    {
      "epoch": 0.5203639146331094,
      "grad_norm": 0.2527335584163666,
      "learning_rate": 1.0157455533639187e-05,
      "loss": 0.776,
      "step": 7650
    },
    {
      "epoch": 0.5210441289006037,
      "grad_norm": 0.2460501492023468,
      "learning_rate": 1.0134963365655473e-05,
      "loss": 0.7722,
      "step": 7660
    },
    {
      "epoch": 0.5217243431680979,
      "grad_norm": 0.26167675852775574,
      "learning_rate": 1.0112470514749311e-05,
      "loss": 0.7699,
      "step": 7670
    },
    {
      "epoch": 0.5224045574355922,
      "grad_norm": 0.24475277960300446,
      "learning_rate": 1.0089977094735833e-05,
      "loss": 0.7729,
      "step": 7680
    },
    {
      "epoch": 0.5230847717030864,
      "grad_norm": 0.24689598381519318,
      "learning_rate": 1.0067483219433045e-05,
      "loss": 0.7825,
      "step": 7690
    },
    {
      "epoch": 0.5237649859705807,
      "grad_norm": 0.2483471781015396,
      "learning_rate": 1.0044989002661258e-05,
      "loss": 0.7669,
      "step": 7700
    },
    {
      "epoch": 0.524445200238075,
      "grad_norm": 0.24557854235172272,
      "learning_rate": 1.0022494558242513e-05,
      "loss": 0.7779,
      "step": 7710
    },
    {
      "epoch": 0.5251254145055693,
      "grad_norm": 0.2462674230337143,
      "learning_rate": 1e-05,
      "loss": 0.7778,
      "step": 7720
    },
    {
      "epoch": 0.5258056287730635,
      "grad_norm": 0.2452927678823471,
      "learning_rate": 9.977505441757488e-06,
      "loss": 0.7751,
      "step": 7730
    },
    {
      "epoch": 0.5264858430405578,
      "grad_norm": 0.24400685727596283,
      "learning_rate": 9.955010997338745e-06,
      "loss": 0.7662,
      "step": 7740
    },
    {
      "epoch": 0.527166057308052,
      "grad_norm": 0.2558972239494324,
      "learning_rate": 9.932516780566959e-06,
      "loss": 0.7668,
      "step": 7750
    },
    {
      "epoch": 0.5278462715755463,
      "grad_norm": 0.24762921035289764,
      "learning_rate": 9.91002290526417e-06,
      "loss": 0.779,
      "step": 7760
    },
    {
      "epoch": 0.5285264858430405,
      "grad_norm": 0.26425403356552124,
      "learning_rate": 9.887529485250692e-06,
      "loss": 0.7688,
      "step": 7770
    },
    {
      "epoch": 0.5292067001105348,
      "grad_norm": 0.25996628403663635,
      "learning_rate": 9.86503663434453e-06,
      "loss": 0.765,
      "step": 7780
    },
    {
      "epoch": 0.529886914378029,
      "grad_norm": 0.24393202364444733,
      "learning_rate": 9.842544466360814e-06,
      "loss": 0.7821,
      "step": 7790
    },
    {
      "epoch": 0.5305671286455234,
      "grad_norm": 0.24665607511997223,
      "learning_rate": 9.820053095111218e-06,
      "loss": 0.7602,
      "step": 7800
    },
    {
      "epoch": 0.5312473429130176,
      "grad_norm": 0.2442709356546402,
      "learning_rate": 9.79756263440338e-06,
      "loss": 0.7492,
      "step": 7810
    },
    {
      "epoch": 0.5319275571805119,
      "grad_norm": 0.25507017970085144,
      "learning_rate": 9.775073198040332e-06,
      "loss": 0.7703,
      "step": 7820
    },
    {
      "epoch": 0.5326077714480061,
      "grad_norm": 0.24093185365200043,
      "learning_rate": 9.752584899819929e-06,
      "loss": 0.7641,
      "step": 7830
    },
    {
      "epoch": 0.5332879857155004,
      "grad_norm": 0.24089796841144562,
      "learning_rate": 9.730097853534258e-06,
      "loss": 0.75,
      "step": 7840
    },
    {
      "epoch": 0.5339681999829946,
      "grad_norm": 0.2501702606678009,
      "learning_rate": 9.707612172969077e-06,
      "loss": 0.7721,
      "step": 7850
    },
    {
      "epoch": 0.5346484142504889,
      "grad_norm": 0.24304954707622528,
      "learning_rate": 9.685127971903232e-06,
      "loss": 0.7709,
      "step": 7860
    },
    {
      "epoch": 0.5353286285179831,
      "grad_norm": 0.2563300132751465,
      "learning_rate": 9.662645364108086e-06,
      "loss": 0.778,
      "step": 7870
    },
    {
      "epoch": 0.5360088427854774,
      "grad_norm": 0.26332852244377136,
      "learning_rate": 9.640164463346929e-06,
      "loss": 0.7792,
      "step": 7880
    },
    {
      "epoch": 0.5366890570529717,
      "grad_norm": 0.2405518740415573,
      "learning_rate": 9.617685383374424e-06,
      "loss": 0.7722,
      "step": 7890
    },
    {
      "epoch": 0.537369271320466,
      "grad_norm": 0.24706117808818817,
      "learning_rate": 9.595208237936018e-06,
      "loss": 0.7503,
      "step": 7900
    },
    {
      "epoch": 0.5380494855879602,
      "grad_norm": 0.24360667169094086,
      "learning_rate": 9.57273314076737e-06,
      "loss": 0.7728,
      "step": 7910
    },
    {
      "epoch": 0.5387296998554545,
      "grad_norm": 0.2456994205713272,
      "learning_rate": 9.55026020559377e-06,
      "loss": 0.7776,
      "step": 7920
    },
    {
      "epoch": 0.5394099141229487,
      "grad_norm": 0.24596786499023438,
      "learning_rate": 9.52778954612957e-06,
      "loss": 0.7758,
      "step": 7930
    },
    {
      "epoch": 0.540090128390443,
      "grad_norm": 0.2587563991546631,
      "learning_rate": 9.505321276077614e-06,
      "loss": 0.784,
      "step": 7940
    },
    {
      "epoch": 0.5407703426579372,
      "grad_norm": 0.2651641368865967,
      "learning_rate": 9.482855509128644e-06,
      "loss": 0.7533,
      "step": 7950
    },
    {
      "epoch": 0.5414505569254315,
      "grad_norm": 0.30029022693634033,
      "learning_rate": 9.460392358960743e-06,
      "loss": 0.7892,
      "step": 7960
    },
    {
      "epoch": 0.5421307711929257,
      "grad_norm": 0.25703874230384827,
      "learning_rate": 9.437931939238755e-06,
      "loss": 0.775,
      "step": 7970
    },
    {
      "epoch": 0.5428109854604201,
      "grad_norm": 0.25282925367355347,
      "learning_rate": 9.415474363613701e-06,
      "loss": 0.7727,
      "step": 7980
    },
    {
      "epoch": 0.5434911997279143,
      "grad_norm": 0.2516460716724396,
      "learning_rate": 9.393019745722215e-06,
      "loss": 0.7732,
      "step": 7990
    },
    {
      "epoch": 0.5441714139954086,
      "grad_norm": 0.2425277680158615,
      "learning_rate": 9.370568199185963e-06,
      "loss": 0.7753,
      "step": 8000
    },
    {
      "epoch": 0.5448516282629028,
      "grad_norm": 0.24489714205265045,
      "learning_rate": 9.348119837611071e-06,
      "loss": 0.7874,
      "step": 8010
    },
    {
      "epoch": 0.5455318425303971,
      "grad_norm": 0.24812796711921692,
      "learning_rate": 9.32567477458755e-06,
      "loss": 0.7665,
      "step": 8020
    },
    {
      "epoch": 0.5462120567978913,
      "grad_norm": 2.061338424682617,
      "learning_rate": 9.303233123688716e-06,
      "loss": 0.7758,
      "step": 8030
    },
    {
      "epoch": 0.5468922710653856,
      "grad_norm": 0.24883481860160828,
      "learning_rate": 9.280794998470623e-06,
      "loss": 0.7652,
      "step": 8040
    },
    {
      "epoch": 0.5475724853328798,
      "grad_norm": 0.24857421219348907,
      "learning_rate": 9.258360512471484e-06,
      "loss": 0.7588,
      "step": 8050
    },
    {
      "epoch": 0.5482526996003741,
      "grad_norm": 0.2536051273345947,
      "learning_rate": 9.235929779211099e-06,
      "loss": 0.7665,
      "step": 8060
    },
    {
      "epoch": 0.5489329138678684,
      "grad_norm": 0.24272532761096954,
      "learning_rate": 9.213502912190275e-06,
      "loss": 0.7781,
      "step": 8070
    },
    {
      "epoch": 0.5496131281353627,
      "grad_norm": 0.24849000573158264,
      "learning_rate": 9.191080024890257e-06,
      "loss": 0.7675,
      "step": 8080
    },
    {
      "epoch": 0.5502933424028569,
      "grad_norm": 0.2545180022716522,
      "learning_rate": 9.168661230772157e-06,
      "loss": 0.7654,
      "step": 8090
    },
    {
      "epoch": 0.5509735566703512,
      "grad_norm": 0.2520526945590973,
      "learning_rate": 9.146246643276368e-06,
      "loss": 0.7686,
      "step": 8100
    },
    {
      "epoch": 0.5516537709378454,
      "grad_norm": 0.2500572204589844,
      "learning_rate": 9.123836375822002e-06,
      "loss": 0.769,
      "step": 8110
    },
    {
      "epoch": 0.5523339852053397,
      "grad_norm": 0.2473742514848709,
      "learning_rate": 9.101430541806308e-06,
      "loss": 0.7546,
      "step": 8120
    },
    {
      "epoch": 0.5530141994728339,
      "grad_norm": 0.27262288331985474,
      "learning_rate": 9.079029254604108e-06,
      "loss": 0.7804,
      "step": 8130
    },
    {
      "epoch": 0.5536944137403282,
      "grad_norm": 0.23986180126667023,
      "learning_rate": 9.056632627567209e-06,
      "loss": 0.7862,
      "step": 8140
    },
    {
      "epoch": 0.5543746280078224,
      "grad_norm": 0.2418127804994583,
      "learning_rate": 9.03424077402384e-06,
      "loss": 0.7768,
      "step": 8150
    },
    {
      "epoch": 0.5550548422753168,
      "grad_norm": 0.24957619607448578,
      "learning_rate": 9.01185380727808e-06,
      "loss": 0.7635,
      "step": 8160
    },
    {
      "epoch": 0.555735056542811,
      "grad_norm": 0.24887363612651825,
      "learning_rate": 8.989471840609273e-06,
      "loss": 0.7594,
      "step": 8170
    },
    {
      "epoch": 0.5564152708103053,
      "grad_norm": 0.24794262647628784,
      "learning_rate": 8.967094987271469e-06,
      "loss": 0.768,
      "step": 8180
    },
    {
      "epoch": 0.5570954850777995,
      "grad_norm": 0.24824637174606323,
      "learning_rate": 8.944723360492842e-06,
      "loss": 0.7641,
      "step": 8190
    },
    {
      "epoch": 0.5577756993452938,
      "grad_norm": 0.2551088333129883,
      "learning_rate": 8.922357073475116e-06,
      "loss": 0.7946,
      "step": 8200
    },
    {
      "epoch": 0.558455913612788,
      "grad_norm": 0.2511906325817108,
      "learning_rate": 8.899996239393002e-06,
      "loss": 0.7699,
      "step": 8210
    },
    {
      "epoch": 0.5591361278802823,
      "grad_norm": 0.24231868982315063,
      "learning_rate": 8.877640971393615e-06,
      "loss": 0.7673,
      "step": 8220
    },
    {
      "epoch": 0.5598163421477765,
      "grad_norm": 0.25138306617736816,
      "learning_rate": 8.855291382595904e-06,
      "loss": 0.7619,
      "step": 8230
    },
    {
      "epoch": 0.5604965564152709,
      "grad_norm": 0.2524348199367523,
      "learning_rate": 8.832947586090082e-06,
      "loss": 0.7649,
      "step": 8240
    },
    {
      "epoch": 0.5611767706827651,
      "grad_norm": 0.25453469157218933,
      "learning_rate": 8.810609694937053e-06,
      "loss": 0.7788,
      "step": 8250
    },
    {
      "epoch": 0.5618569849502594,
      "grad_norm": 0.2537672221660614,
      "learning_rate": 8.788277822167843e-06,
      "loss": 0.7643,
      "step": 8260
    },
    {
      "epoch": 0.5625371992177536,
      "grad_norm": 0.2489592581987381,
      "learning_rate": 8.765952080783019e-06,
      "loss": 0.7741,
      "step": 8270
    },
    {
      "epoch": 0.5632174134852479,
      "grad_norm": 0.24671168625354767,
      "learning_rate": 8.743632583752123e-06,
      "loss": 0.7627,
      "step": 8280
    },
    {
      "epoch": 0.5638976277527421,
      "grad_norm": 0.24434447288513184,
      "learning_rate": 8.721319444013108e-06,
      "loss": 0.7745,
      "step": 8290
    },
    {
      "epoch": 0.5645778420202364,
      "grad_norm": 0.24269433319568634,
      "learning_rate": 8.699012774471748e-06,
      "loss": 0.7675,
      "step": 8300
    },
    {
      "epoch": 0.5652580562877306,
      "grad_norm": 0.24096103012561798,
      "learning_rate": 8.676712688001086e-06,
      "loss": 0.7546,
      "step": 8310
    },
    {
      "epoch": 0.5659382705552249,
      "grad_norm": 0.2548983693122864,
      "learning_rate": 8.654419297440844e-06,
      "loss": 0.7611,
      "step": 8320
    },
    {
      "epoch": 0.5666184848227191,
      "grad_norm": 0.24676938354969025,
      "learning_rate": 8.63213271559688e-06,
      "loss": 0.7732,
      "step": 8330
    },
    {
      "epoch": 0.5672986990902135,
      "grad_norm": 0.24426358938217163,
      "learning_rate": 8.609853055240582e-06,
      "loss": 0.7803,
      "step": 8340
    },
    {
      "epoch": 0.5679789133577077,
      "grad_norm": 0.24262279272079468,
      "learning_rate": 8.587580429108323e-06,
      "loss": 0.7685,
      "step": 8350
    },
    {
      "epoch": 0.568659127625202,
      "grad_norm": 0.2493383139371872,
      "learning_rate": 8.56531494990088e-06,
      "loss": 0.7759,
      "step": 8360
    },
    {
      "epoch": 0.5693393418926962,
      "grad_norm": 0.2447572648525238,
      "learning_rate": 8.54305673028287e-06,
      "loss": 0.7655,
      "step": 8370
    },
    {
      "epoch": 0.5700195561601905,
      "grad_norm": 0.24559731781482697,
      "learning_rate": 8.520805882882168e-06,
      "loss": 0.7717,
      "step": 8380
    },
    {
      "epoch": 0.5706997704276847,
      "grad_norm": 0.23603376746177673,
      "learning_rate": 8.498562520289356e-06,
      "loss": 0.7702,
      "step": 8390
    },
    {
      "epoch": 0.571379984695179,
      "grad_norm": 0.24445731937885284,
      "learning_rate": 8.476326755057131e-06,
      "loss": 0.7674,
      "step": 8400
    },
    {
      "epoch": 0.5720601989626732,
      "grad_norm": 0.24234746396541595,
      "learning_rate": 8.454098699699759e-06,
      "loss": 0.7717,
      "step": 8410
    },
    {
      "epoch": 0.5727404132301676,
      "grad_norm": 0.2467491179704666,
      "learning_rate": 8.431878466692482e-06,
      "loss": 0.7736,
      "step": 8420
    },
    {
      "epoch": 0.5734206274976618,
      "grad_norm": 0.812633752822876,
      "learning_rate": 8.409666168470967e-06,
      "loss": 0.7578,
      "step": 8430
    },
    {
      "epoch": 0.5741008417651561,
      "grad_norm": 0.25730645656585693,
      "learning_rate": 8.387461917430732e-06,
      "loss": 0.7726,
      "step": 8440
    },
    {
      "epoch": 0.5747810560326503,
      "grad_norm": 0.24951675534248352,
      "learning_rate": 8.36526582592657e-06,
      "loss": 0.7812,
      "step": 8450
    },
    {
      "epoch": 0.5754612703001446,
      "grad_norm": 0.24625453352928162,
      "learning_rate": 8.343078006271989e-06,
      "loss": 0.764,
      "step": 8460
    },
    {
      "epoch": 0.5761414845676388,
      "grad_norm": 0.25189143419265747,
      "learning_rate": 8.320898570738645e-06,
      "loss": 0.7741,
      "step": 8470
    },
    {
      "epoch": 0.576821698835133,
      "grad_norm": 0.2517893314361572,
      "learning_rate": 8.298727631555761e-06,
      "loss": 0.772,
      "step": 8480
    },
    {
      "epoch": 0.5775019131026273,
      "grad_norm": 0.24455204606056213,
      "learning_rate": 8.276565300909577e-06,
      "loss": 0.7791,
      "step": 8490
    },
    {
      "epoch": 0.5781821273701215,
      "grad_norm": 0.24319756031036377,
      "learning_rate": 8.254411690942768e-06,
      "loss": 0.7697,
      "step": 8500
    },
    {
      "epoch": 0.5788623416376159,
      "grad_norm": 0.2484409362077713,
      "learning_rate": 8.232266913753883e-06,
      "loss": 0.7738,
      "step": 8510
    },
    {
      "epoch": 0.5795425559051101,
      "grad_norm": 0.25438666343688965,
      "learning_rate": 8.210131081396779e-06,
      "loss": 0.7666,
      "step": 8520
    },
    {
      "epoch": 0.5802227701726044,
      "grad_norm": 0.2526380717754364,
      "learning_rate": 8.188004305880046e-06,
      "loss": 0.7708,
      "step": 8530
    },
    {
      "epoch": 0.5809029844400986,
      "grad_norm": 0.2531062662601471,
      "learning_rate": 8.165886699166452e-06,
      "loss": 0.7688,
      "step": 8540
    },
    {
      "epoch": 0.5815831987075929,
      "grad_norm": 0.261444091796875,
      "learning_rate": 8.143778373172367e-06,
      "loss": 0.7732,
      "step": 8550
    },
    {
      "epoch": 0.5822634129750871,
      "grad_norm": 0.24088214337825775,
      "learning_rate": 8.1216794397672e-06,
      "loss": 0.7637,
      "step": 8560
    },
    {
      "epoch": 0.5829436272425814,
      "grad_norm": 0.2516765296459198,
      "learning_rate": 8.099590010772835e-06,
      "loss": 0.7683,
      "step": 8570
    },
    {
      "epoch": 0.5836238415100756,
      "grad_norm": 0.24979981780052185,
      "learning_rate": 8.077510197963061e-06,
      "loss": 0.7773,
      "step": 8580
    },
    {
      "epoch": 0.5843040557775699,
      "grad_norm": 0.25018417835235596,
      "learning_rate": 8.055440113063008e-06,
      "loss": 0.7824,
      "step": 8590
    },
    {
      "epoch": 0.5849842700450641,
      "grad_norm": 0.2482062429189682,
      "learning_rate": 8.033379867748587e-06,
      "loss": 0.7703,
      "step": 8600
    },
    {
      "epoch": 0.5856644843125585,
      "grad_norm": 0.24455052614212036,
      "learning_rate": 8.011329573645915e-06,
      "loss": 0.7753,
      "step": 8610
    },
    {
      "epoch": 0.5863446985800527,
      "grad_norm": 0.24979741871356964,
      "learning_rate": 7.989289342330757e-06,
      "loss": 0.7749,
      "step": 8620
    },
    {
      "epoch": 0.587024912847547,
      "grad_norm": 0.2415253072977066,
      "learning_rate": 7.96725928532796e-06,
      "loss": 0.7614,
      "step": 8630
    },
    {
      "epoch": 0.5877051271150412,
      "grad_norm": 0.24844114482402802,
      "learning_rate": 7.945239514110886e-06,
      "loss": 0.7707,
      "step": 8640
    },
    {
      "epoch": 0.5883853413825355,
      "grad_norm": 0.2513725459575653,
      "learning_rate": 7.923230140100855e-06,
      "loss": 0.7574,
      "step": 8650
    },
    {
      "epoch": 0.5890655556500297,
      "grad_norm": 0.2472066730260849,
      "learning_rate": 7.901231274666574e-06,
      "loss": 0.7729,
      "step": 8660
    },
    {
      "epoch": 0.589745769917524,
      "grad_norm": 0.2509872019290924,
      "learning_rate": 7.879243029123573e-06,
      "loss": 0.7644,
      "step": 8670
    },
    {
      "epoch": 0.5904259841850182,
      "grad_norm": 0.24720162153244019,
      "learning_rate": 7.857265514733651e-06,
      "loss": 0.7644,
      "step": 8680
    },
    {
      "epoch": 0.5911061984525126,
      "grad_norm": 0.2582946717739105,
      "learning_rate": 7.835298842704303e-06,
      "loss": 0.7824,
      "step": 8690
    },
    {
      "epoch": 0.5917864127200068,
      "grad_norm": 0.25611555576324463,
      "learning_rate": 7.81334312418816e-06,
      "loss": 0.7704,
      "step": 8700
    },
    {
      "epoch": 0.5924666269875011,
      "grad_norm": 0.24672120809555054,
      "learning_rate": 7.79139847028243e-06,
      "loss": 0.7535,
      "step": 8710
    },
    {
      "epoch": 0.5931468412549953,
      "grad_norm": 0.25224679708480835,
      "learning_rate": 7.769464992028332e-06,
      "loss": 0.7689,
      "step": 8720
    },
    {
      "epoch": 0.5938270555224896,
      "grad_norm": 0.2413455694913864,
      "learning_rate": 7.747542800410538e-06,
      "loss": 0.7705,
      "step": 8730
    },
    {
      "epoch": 0.5945072697899838,
      "grad_norm": 0.24807719886302948,
      "learning_rate": 7.725632006356603e-06,
      "loss": 0.7599,
      "step": 8740
    },
    {
      "epoch": 0.5951874840574781,
      "grad_norm": 0.2648003399372101,
      "learning_rate": 7.70373272073642e-06,
      "loss": 0.7611,
      "step": 8750
    },
    {
      "epoch": 0.5958676983249723,
      "grad_norm": 0.25700241327285767,
      "learning_rate": 7.681845054361634e-06,
      "loss": 0.7728,
      "step": 8760
    },
    {
      "epoch": 0.5965479125924666,
      "grad_norm": 0.25815171003341675,
      "learning_rate": 7.659969117985111e-06,
      "loss": 0.7685,
      "step": 8770
    },
    {
      "epoch": 0.5972281268599609,
      "grad_norm": 0.25158023834228516,
      "learning_rate": 7.638105022300352e-06,
      "loss": 0.7718,
      "step": 8780
    },
    {
      "epoch": 0.5979083411274552,
      "grad_norm": 0.24402719736099243,
      "learning_rate": 7.6162528779409506e-06,
      "loss": 0.7623,
      "step": 8790
    },
    {
      "epoch": 0.5985885553949494,
      "grad_norm": 0.24518702924251556,
      "learning_rate": 7.59441279548002e-06,
      "loss": 0.7688,
      "step": 8800
    },
    {
      "epoch": 0.5992687696624437,
      "grad_norm": 0.2488635927438736,
      "learning_rate": 7.5725848854296425e-06,
      "loss": 0.7697,
      "step": 8810
    },
    {
      "epoch": 0.5999489839299379,
      "grad_norm": 0.24867048859596252,
      "learning_rate": 7.550769258240307e-06,
      "loss": 0.7808,
      "step": 8820
    },
    {
      "epoch": 0.6006291981974322,
      "grad_norm": 0.2935914099216461,
      "learning_rate": 7.528966024300352e-06,
      "loss": 0.7627,
      "step": 8830
    },
    {
      "epoch": 0.6013094124649264,
      "grad_norm": 0.2575988173484802,
      "learning_rate": 7.507175293935401e-06,
      "loss": 0.7762,
      "step": 8840
    },
    {
      "epoch": 0.6019896267324207,
      "grad_norm": 0.24520112574100494,
      "learning_rate": 7.485397177407814e-06,
      "loss": 0.763,
      "step": 8850
    },
    {
      "epoch": 0.6026698409999149,
      "grad_norm": 0.2443533092737198,
      "learning_rate": 7.463631784916121e-06,
      "loss": 0.759,
      "step": 8860
    },
    {
      "epoch": 0.6033500552674093,
      "grad_norm": 0.24732626974582672,
      "learning_rate": 7.4418792265944686e-06,
      "loss": 0.7703,
      "step": 8870
    },
    {
      "epoch": 0.6040302695349035,
      "grad_norm": 0.24233926832675934,
      "learning_rate": 7.420139612512061e-06,
      "loss": 0.7449,
      "step": 8880
    },
    {
      "epoch": 0.6047104838023978,
      "grad_norm": 0.24057942628860474,
      "learning_rate": 7.398413052672604e-06,
      "loss": 0.7706,
      "step": 8890
    },
    {
      "epoch": 0.605390698069892,
      "grad_norm": 0.24792107939720154,
      "learning_rate": 7.376699657013747e-06,
      "loss": 0.7723,
      "step": 8900
    },
    {
      "epoch": 0.6060709123373863,
      "grad_norm": 0.24438060820102692,
      "learning_rate": 7.354999535406533e-06,
      "loss": 0.7583,
      "step": 8910
    },
    {
      "epoch": 0.6067511266048805,
      "grad_norm": 0.24966730177402496,
      "learning_rate": 7.3333127976548305e-06,
      "loss": 0.7619,
      "step": 8920
    },
    {
      "epoch": 0.6074313408723748,
      "grad_norm": 0.25231412053108215,
      "learning_rate": 7.31163955349479e-06,
      "loss": 0.7621,
      "step": 8930
    },
    {
      "epoch": 0.608111555139869,
      "grad_norm": 0.24898163974285126,
      "learning_rate": 7.289979912594283e-06,
      "loss": 0.7573,
      "step": 8940
    },
    {
      "epoch": 0.6087917694073633,
      "grad_norm": 0.248761385679245,
      "learning_rate": 7.268333984552345e-06,
      "loss": 0.7686,
      "step": 8950
    },
    {
      "epoch": 0.6094719836748576,
      "grad_norm": 0.24428556859493256,
      "learning_rate": 7.2467018788986256e-06,
      "loss": 0.7784,
      "step": 8960
    },
    {
      "epoch": 0.6101521979423519,
      "grad_norm": 0.2532806694507599,
      "learning_rate": 7.2250837050928324e-06,
      "loss": 0.7757,
      "step": 8970
    },
    {
      "epoch": 0.6108324122098461,
      "grad_norm": 0.2456384152173996,
      "learning_rate": 7.203479572524177e-06,
      "loss": 0.7804,
      "step": 8980
    },
    {
      "epoch": 0.6115126264773404,
      "grad_norm": 0.252175897359848,
      "learning_rate": 7.18188959051082e-06,
      "loss": 0.7613,
      "step": 8990
    },
    {
      "epoch": 0.6121928407448346,
      "grad_norm": 0.24484005570411682,
      "learning_rate": 7.160313868299321e-06,
      "loss": 0.753,
      "step": 9000
    },
    {
      "epoch": 0.6128730550123289,
      "grad_norm": 0.2501670718193054,
      "learning_rate": 7.138752515064085e-06,
      "loss": 0.7759,
      "step": 9010
    },
    {
      "epoch": 0.6135532692798231,
      "grad_norm": 0.24075950682163239,
      "learning_rate": 7.117205639906806e-06,
      "loss": 0.7577,
      "step": 9020
    },
    {
      "epoch": 0.6142334835473174,
      "grad_norm": 0.26213377714157104,
      "learning_rate": 7.095673351855923e-06,
      "loss": 0.7673,
      "step": 9030
    },
    {
      "epoch": 0.6149136978148116,
      "grad_norm": 0.23907162249088287,
      "learning_rate": 7.074155759866057e-06,
      "loss": 0.7665,
      "step": 9040
    },
    {
      "epoch": 0.615593912082306,
      "grad_norm": 0.25060412287712097,
      "learning_rate": 7.052652972817471e-06,
      "loss": 0.7637,
      "step": 9050
    },
    {
      "epoch": 0.6162741263498002,
      "grad_norm": 0.24636393785476685,
      "learning_rate": 7.031165099515514e-06,
      "loss": 0.7672,
      "step": 9060
    },
    {
      "epoch": 0.6169543406172945,
      "grad_norm": 0.25181058049201965,
      "learning_rate": 7.009692248690066e-06,
      "loss": 0.7782,
      "step": 9070
    },
    {
      "epoch": 0.6176345548847887,
      "grad_norm": 0.2514166235923767,
      "learning_rate": 6.988234528994997e-06,
      "loss": 0.7606,
      "step": 9080
    },
    {
      "epoch": 0.618314769152283,
      "grad_norm": 0.25581368803977966,
      "learning_rate": 6.966792049007613e-06,
      "loss": 0.7834,
      "step": 9090
    },
    {
      "epoch": 0.6189949834197772,
      "grad_norm": 0.24559302628040314,
      "learning_rate": 6.945364917228101e-06,
      "loss": 0.7753,
      "step": 9100
    },
    {
      "epoch": 0.6196751976872715,
      "grad_norm": 0.24848997592926025,
      "learning_rate": 6.923953242078992e-06,
      "loss": 0.7602,
      "step": 9110
    },
    {
      "epoch": 0.6203554119547657,
      "grad_norm": 0.2535330355167389,
      "learning_rate": 6.902557131904602e-06,
      "loss": 0.7782,
      "step": 9120
    },
    {
      "epoch": 0.62103562622226,
      "grad_norm": 0.248373344540596,
      "learning_rate": 6.881176694970483e-06,
      "loss": 0.7726,
      "step": 9130
    },
    {
      "epoch": 0.6217158404897543,
      "grad_norm": 0.24924324452877045,
      "learning_rate": 6.8598120394628875e-06,
      "loss": 0.7712,
      "step": 9140
    },
    {
      "epoch": 0.6223960547572486,
      "grad_norm": 0.24666650593280792,
      "learning_rate": 6.83846327348821e-06,
      "loss": 0.7541,
      "step": 9150
    },
    {
      "epoch": 0.6230762690247428,
      "grad_norm": 0.23946702480316162,
      "learning_rate": 6.817130505072442e-06,
      "loss": 0.7625,
      "step": 9160
    },
    {
      "epoch": 0.6237564832922371,
      "grad_norm": 0.24774959683418274,
      "learning_rate": 6.795813842160626e-06,
      "loss": 0.7658,
      "step": 9170
    },
    {
      "epoch": 0.6244366975597313,
      "grad_norm": 0.25880733132362366,
      "learning_rate": 6.774513392616311e-06,
      "loss": 0.7574,
      "step": 9180
    },
    {
      "epoch": 0.6251169118272256,
      "grad_norm": 0.24910889565944672,
      "learning_rate": 6.753229264221005e-06,
      "loss": 0.7717,
      "step": 9190
    },
    {
      "epoch": 0.6257971260947198,
      "grad_norm": 0.251426637172699,
      "learning_rate": 6.73196156467363e-06,
      "loss": 0.7683,
      "step": 9200
    },
    {
      "epoch": 0.6264773403622141,
      "grad_norm": 0.24830366671085358,
      "learning_rate": 6.7107104015899795e-06,
      "loss": 0.7615,
      "step": 9210
    },
    {
      "epoch": 0.6271575546297083,
      "grad_norm": 0.24772919714450836,
      "learning_rate": 6.689475882502167e-06,
      "loss": 0.7758,
      "step": 9220
    },
    {
      "epoch": 0.6278377688972027,
      "grad_norm": 0.2580617368221283,
      "learning_rate": 6.668258114858088e-06,
      "loss": 0.7655,
      "step": 9230
    },
    {
      "epoch": 0.6285179831646969,
      "grad_norm": 0.24163033068180084,
      "learning_rate": 6.64705720602088e-06,
      "loss": 0.7665,
      "step": 9240
    },
    {
      "epoch": 0.6291981974321912,
      "grad_norm": 0.25090691447257996,
      "learning_rate": 6.625873263268363e-06,
      "loss": 0.7555,
      "step": 9250
    },
    {
      "epoch": 0.6298784116996854,
      "grad_norm": 0.2470577210187912,
      "learning_rate": 6.604706393792517e-06,
      "loss": 0.7617,
      "step": 9260
    },
    {
      "epoch": 0.6305586259671797,
      "grad_norm": 0.2557064890861511,
      "learning_rate": 6.583556704698928e-06,
      "loss": 0.7596,
      "step": 9270
    },
    {
      "epoch": 0.6312388402346739,
      "grad_norm": 0.24683815240859985,
      "learning_rate": 6.5624243030062475e-06,
      "loss": 0.7646,
      "step": 9280
    },
    {
      "epoch": 0.6319190545021682,
      "grad_norm": 0.26031729578971863,
      "learning_rate": 6.54130929564565e-06,
      "loss": 0.7651,
      "step": 9290
    },
    {
      "epoch": 0.6325992687696624,
      "grad_norm": 0.25592610239982605,
      "learning_rate": 6.520211789460298e-06,
      "loss": 0.7561,
      "step": 9300
    },
    {
      "epoch": 0.6332794830371568,
      "grad_norm": 0.24974516034126282,
      "learning_rate": 6.499131891204792e-06,
      "loss": 0.7637,
      "step": 9310
    },
    {
      "epoch": 0.633959697304651,
      "grad_norm": 0.25895562767982483,
      "learning_rate": 6.478069707544639e-06,
      "loss": 0.7581,
      "step": 9320
    },
    {
      "epoch": 0.6346399115721453,
      "grad_norm": 0.25263655185699463,
      "learning_rate": 6.457025345055708e-06,
      "loss": 0.769,
      "step": 9330
    },
    {
      "epoch": 0.6353201258396395,
      "grad_norm": 0.24764102697372437,
      "learning_rate": 6.435998910223691e-06,
      "loss": 0.7653,
      "step": 9340
    },
    {
      "epoch": 0.6360003401071338,
      "grad_norm": 0.25428757071495056,
      "learning_rate": 6.414990509443566e-06,
      "loss": 0.7776,
      "step": 9350
    },
    {
      "epoch": 0.636680554374628,
      "grad_norm": 0.24425633251667023,
      "learning_rate": 6.394000249019059e-06,
      "loss": 0.7642,
      "step": 9360
    },
    {
      "epoch": 0.6373607686421223,
      "grad_norm": 0.24660134315490723,
      "learning_rate": 6.373028235162102e-06,
      "loss": 0.7751,
      "step": 9370
    },
    {
      "epoch": 0.6380409829096165,
      "grad_norm": 0.253810316324234,
      "learning_rate": 6.3520745739923e-06,
      "loss": 0.767,
      "step": 9380
    },
    {
      "epoch": 0.6387211971771108,
      "grad_norm": 0.24733662605285645,
      "learning_rate": 6.33113937153639e-06,
      "loss": 0.7664,
      "step": 9390
    },
    {
      "epoch": 0.639401411444605,
      "grad_norm": 0.24834582209587097,
      "learning_rate": 6.310222733727711e-06,
      "loss": 0.7662,
      "step": 9400
    },
    {
      "epoch": 0.6400816257120993,
      "grad_norm": 0.2556808888912201,
      "learning_rate": 6.28932476640566e-06,
      "loss": 0.7766,
      "step": 9410
    },
    {
      "epoch": 0.6407618399795936,
      "grad_norm": 0.2530834376811981,
      "learning_rate": 6.268445575315164e-06,
      "loss": 0.7675,
      "step": 9420
    },
    {
      "epoch": 0.6414420542470878,
      "grad_norm": 0.25645995140075684,
      "learning_rate": 6.2475852661061364e-06,
      "loss": 0.7681,
      "step": 9430
    },
    {
      "epoch": 0.6421222685145821,
      "grad_norm": 0.24543169140815735,
      "learning_rate": 6.226743944332954e-06,
      "loss": 0.771,
      "step": 9440
    },
    {
      "epoch": 0.6428024827820763,
      "grad_norm": 0.25770193338394165,
      "learning_rate": 6.205921715453904e-06,
      "loss": 0.7671,
      "step": 9450
    },
    {
      "epoch": 0.6434826970495706,
      "grad_norm": 0.2512899935245514,
      "learning_rate": 6.1851186848306775e-06,
      "loss": 0.7561,
      "step": 9460
    },
    {
      "epoch": 0.6441629113170648,
      "grad_norm": 0.2442668378353119,
      "learning_rate": 6.16433495772781e-06,
      "loss": 0.7712,
      "step": 9470
    },
    {
      "epoch": 0.6448431255845591,
      "grad_norm": 0.24152372777462006,
      "learning_rate": 6.143570639312166e-06,
      "loss": 0.7517,
      "step": 9480
    },
    {
      "epoch": 0.6455233398520533,
      "grad_norm": 0.24572031199932098,
      "learning_rate": 6.122825834652399e-06,
      "loss": 0.7654,
      "step": 9490
    },
    {
      "epoch": 0.6462035541195477,
      "grad_norm": 0.24191130697727203,
      "learning_rate": 6.102100648718416e-06,
      "loss": 0.7525,
      "step": 9500
    },
    {
      "epoch": 0.6468837683870419,
      "grad_norm": 0.25088223814964294,
      "learning_rate": 6.081395186380868e-06,
      "loss": 0.7699,
      "step": 9510
    },
    {
      "epoch": 0.6475639826545362,
      "grad_norm": 0.24233394861221313,
      "learning_rate": 6.060709552410591e-06,
      "loss": 0.7503,
      "step": 9520
    },
    {
      "epoch": 0.6482441969220304,
      "grad_norm": 0.2550160586833954,
      "learning_rate": 6.0400438514780854e-06,
      "loss": 0.7496,
      "step": 9530
    },
    {
      "epoch": 0.6489244111895247,
      "grad_norm": 0.251807302236557,
      "learning_rate": 6.019398188153e-06,
      "loss": 0.7519,
      "step": 9540
    },
    {
      "epoch": 0.6496046254570189,
      "grad_norm": 0.24086961150169373,
      "learning_rate": 5.998772666903583e-06,
      "loss": 0.754,
      "step": 9550
    },
    {
      "epoch": 0.6502848397245132,
      "grad_norm": 0.24868667125701904,
      "learning_rate": 5.978167392096168e-06,
      "loss": 0.7541,
      "step": 9560
    },
    {
      "epoch": 0.6509650539920074,
      "grad_norm": 0.24697424471378326,
      "learning_rate": 5.957582467994641e-06,
      "loss": 0.7684,
      "step": 9570
    },
    {
      "epoch": 0.6516452682595018,
      "grad_norm": 0.2440696656703949,
      "learning_rate": 5.937017998759904e-06,
      "loss": 0.7555,
      "step": 9580
    },
    {
      "epoch": 0.652325482526996,
      "grad_norm": 0.24430793523788452,
      "learning_rate": 5.916474088449366e-06,
      "loss": 0.7649,
      "step": 9590
    },
    {
      "epoch": 0.6530056967944903,
      "grad_norm": 0.24338242411613464,
      "learning_rate": 5.895950841016404e-06,
      "loss": 0.769,
      "step": 9600
    },
    {
      "epoch": 0.6536859110619845,
      "grad_norm": 0.24120405316352844,
      "learning_rate": 5.875448360309837e-06,
      "loss": 0.7618,
      "step": 9610
    },
    {
      "epoch": 0.6543661253294788,
      "grad_norm": 0.24697110056877136,
      "learning_rate": 5.854966750073403e-06,
      "loss": 0.7619,
      "step": 9620
    },
    {
      "epoch": 0.655046339596973,
      "grad_norm": 0.24529220163822174,
      "learning_rate": 5.834506113945237e-06,
      "loss": 0.7493,
      "step": 9630
    },
    {
      "epoch": 0.6557265538644673,
      "grad_norm": 0.2561686336994171,
      "learning_rate": 5.814066555457343e-06,
      "loss": 0.7613,
      "step": 9640
    },
    {
      "epoch": 0.6564067681319615,
      "grad_norm": 0.2434144914150238,
      "learning_rate": 5.7936481780350705e-06,
      "loss": 0.7611,
      "step": 9650
    },
    {
      "epoch": 0.6570869823994558,
      "grad_norm": 0.25305524468421936,
      "learning_rate": 5.773251084996591e-06,
      "loss": 0.7662,
      "step": 9660
    },
    {
      "epoch": 0.65776719666695,
      "grad_norm": 0.24415530264377594,
      "learning_rate": 5.752875379552378e-06,
      "loss": 0.7572,
      "step": 9670
    },
    {
      "epoch": 0.6584474109344444,
      "grad_norm": 0.25350406765937805,
      "learning_rate": 5.7325211648046795e-06,
      "loss": 0.7541,
      "step": 9680
    },
    {
      "epoch": 0.6591276252019386,
      "grad_norm": 0.24665938317775726,
      "learning_rate": 5.7121885437470015e-06,
      "loss": 0.753,
      "step": 9690
    },
    {
      "epoch": 0.6598078394694329,
      "grad_norm": 0.24728219211101532,
      "learning_rate": 5.691877619263581e-06,
      "loss": 0.7705,
      "step": 9700
    },
    {
      "epoch": 0.6604880537369271,
      "grad_norm": 0.24584805965423584,
      "learning_rate": 5.671588494128877e-06,
      "loss": 0.7659,
      "step": 9710
    },
    {
      "epoch": 0.6611682680044214,
      "grad_norm": 0.24218979477882385,
      "learning_rate": 5.651321271007034e-06,
      "loss": 0.7612,
      "step": 9720
    },
    {
      "epoch": 0.6618484822719156,
      "grad_norm": 0.25144141912460327,
      "learning_rate": 5.631076052451376e-06,
      "loss": 0.7681,
      "step": 9730
    },
    {
      "epoch": 0.6625286965394099,
      "grad_norm": 0.24840059876441956,
      "learning_rate": 5.610852940903881e-06,
      "loss": 0.7675,
      "step": 9740
    },
    {
      "epoch": 0.6632089108069041,
      "grad_norm": 0.24385057389736176,
      "learning_rate": 5.590652038694664e-06,
      "loss": 0.7585,
      "step": 9750
    },
    {
      "epoch": 0.6638891250743985,
      "grad_norm": 0.24427461624145508,
      "learning_rate": 5.57047344804146e-06,
      "loss": 0.7637,
      "step": 9760
    },
    {
      "epoch": 0.6645693393418927,
      "grad_norm": 0.2447434812784195,
      "learning_rate": 5.5503172710491064e-06,
      "loss": 0.7679,
      "step": 9770
    },
    {
      "epoch": 0.665249553609387,
      "grad_norm": 0.24962159991264343,
      "learning_rate": 5.530183609709025e-06,
      "loss": 0.7632,
      "step": 9780
    },
    {
      "epoch": 0.6659297678768812,
      "grad_norm": 0.2556682229042053,
      "learning_rate": 5.510072565898711e-06,
      "loss": 0.78,
      "step": 9790
    },
    {
      "epoch": 0.6666099821443755,
      "grad_norm": 0.2510310113430023,
      "learning_rate": 5.489984241381208e-06,
      "loss": 0.7674,
      "step": 9800
    },
    {
      "epoch": 0.6672901964118697,
      "grad_norm": 0.24962040781974792,
      "learning_rate": 5.469918737804601e-06,
      "loss": 0.7652,
      "step": 9810
    },
    {
      "epoch": 0.667970410679364,
      "grad_norm": 0.2444206029176712,
      "learning_rate": 5.449876156701501e-06,
      "loss": 0.7684,
      "step": 9820
    },
    {
      "epoch": 0.6686506249468582,
      "grad_norm": 0.26629745960235596,
      "learning_rate": 5.42985659948853e-06,
      "loss": 0.7695,
      "step": 9830
    },
    {
      "epoch": 0.6693308392143525,
      "grad_norm": 0.2458677440881729,
      "learning_rate": 5.409860167465806e-06,
      "loss": 0.7602,
      "step": 9840
    },
    {
      "epoch": 0.6700110534818468,
      "grad_norm": 0.24040453135967255,
      "learning_rate": 5.3898869618164325e-06,
      "loss": 0.7624,
      "step": 9850
    },
    {
      "epoch": 0.6706912677493411,
      "grad_norm": 0.23869656026363373,
      "learning_rate": 5.369937083605986e-06,
      "loss": 0.767,
      "step": 9860
    },
    {
      "epoch": 0.6713714820168353,
      "grad_norm": 0.24998657405376434,
      "learning_rate": 5.350010633782007e-06,
      "loss": 0.7638,
      "step": 9870
    },
    {
      "epoch": 0.6720516962843296,
      "grad_norm": 0.2390597015619278,
      "learning_rate": 5.3301077131734846e-06,
      "loss": 0.7647,
      "step": 9880
    },
    {
      "epoch": 0.6727319105518238,
      "grad_norm": 0.26572591066360474,
      "learning_rate": 5.310228422490349e-06,
      "loss": 0.7541,
      "step": 9890
    },
    {
      "epoch": 0.6734121248193181,
      "grad_norm": 0.24828065931797028,
      "learning_rate": 5.290372862322964e-06,
      "loss": 0.766,
      "step": 9900
    },
    {
      "epoch": 0.6740923390868123,
      "grad_norm": 0.24568676948547363,
      "learning_rate": 5.270541133141611e-06,
      "loss": 0.7575,
      "step": 9910
    },
    {
      "epoch": 0.6747725533543066,
      "grad_norm": 0.2494325488805771,
      "learning_rate": 5.250733335295992e-06,
      "loss": 0.774,
      "step": 9920
    },
    {
      "epoch": 0.6754527676218008,
      "grad_norm": 0.24998196959495544,
      "learning_rate": 5.230949569014707e-06,
      "loss": 0.7637,
      "step": 9930
    },
    {
      "epoch": 0.6761329818892952,
      "grad_norm": 0.24165105819702148,
      "learning_rate": 5.211189934404763e-06,
      "loss": 0.7714,
      "step": 9940
    },
    {
      "epoch": 0.6768131961567894,
      "grad_norm": 0.24361291527748108,
      "learning_rate": 5.1914545314510515e-06,
      "loss": 0.7698,
      "step": 9950
    },
    {
      "epoch": 0.6774934104242837,
      "grad_norm": 0.25533944368362427,
      "learning_rate": 5.171743460015857e-06,
      "loss": 0.7594,
      "step": 9960
    },
    {
      "epoch": 0.6781736246917779,
      "grad_norm": 0.24661685526371002,
      "learning_rate": 5.15205681983834e-06,
      "loss": 0.7473,
      "step": 9970
    },
    {
      "epoch": 0.6788538389592722,
      "grad_norm": 0.25577622652053833,
      "learning_rate": 5.132394710534041e-06,
      "loss": 0.7585,
      "step": 9980
    },
    {
      "epoch": 0.6795340532267664,
      "grad_norm": 0.256303995847702,
      "learning_rate": 5.112757231594374e-06,
      "loss": 0.7674,
      "step": 9990
    },
    {
      "epoch": 0.6802142674942607,
      "grad_norm": 0.24000784754753113,
      "learning_rate": 5.0931444823861155e-06,
      "loss": 0.7725,
      "step": 10000
    },
    {
      "epoch": 0.6808944817617549,
      "grad_norm": 0.23835529386997223,
      "learning_rate": 5.073556562150914e-06,
      "loss": 0.766,
      "step": 10010
    },
    {
      "epoch": 0.6815746960292492,
      "grad_norm": 0.2461031973361969,
      "learning_rate": 5.05399357000478e-06,
      "loss": 0.7606,
      "step": 10020
    },
    {
      "epoch": 0.6822549102967435,
      "grad_norm": 0.2479454129934311,
      "learning_rate": 5.034455604937587e-06,
      "loss": 0.7648,
      "step": 10030
    },
    {
      "epoch": 0.6829351245642378,
      "grad_norm": 0.2555757761001587,
      "learning_rate": 5.014942765812574e-06,
      "loss": 0.7726,
      "step": 10040
    },
    {
      "epoch": 0.683615338831732,
      "grad_norm": 0.25147545337677,
      "learning_rate": 4.995455151365828e-06,
      "loss": 0.7566,
      "step": 10050
    },
    {
      "epoch": 0.6842955530992263,
      "grad_norm": 0.23948884010314941,
      "learning_rate": 4.975992860205811e-06,
      "loss": 0.7648,
      "step": 10060
    },
    {
      "epoch": 0.6849757673667205,
      "grad_norm": 0.2528388500213623,
      "learning_rate": 4.956555990812842e-06,
      "loss": 0.7524,
      "step": 10070
    },
    {
      "epoch": 0.6856559816342148,
      "grad_norm": 0.25195372104644775,
      "learning_rate": 4.937144641538607e-06,
      "loss": 0.7588,
      "step": 10080
    },
    {
      "epoch": 0.686336195901709,
      "grad_norm": 0.2495216578245163,
      "learning_rate": 4.917758910605658e-06,
      "loss": 0.7605,
      "step": 10090
    },
    {
      "epoch": 0.6870164101692033,
      "grad_norm": 0.24449487030506134,
      "learning_rate": 4.898398896106914e-06,
      "loss": 0.7571,
      "step": 10100
    },
    {
      "epoch": 0.6876966244366975,
      "grad_norm": 0.24723927676677704,
      "learning_rate": 4.879064696005171e-06,
      "loss": 0.7487,
      "step": 10110
    },
    {
      "epoch": 0.6883768387041919,
      "grad_norm": 0.24362149834632874,
      "learning_rate": 4.859756408132601e-06,
      "loss": 0.7623,
      "step": 10120
    },
    {
      "epoch": 0.6890570529716861,
      "grad_norm": 0.24735696613788605,
      "learning_rate": 4.840474130190258e-06,
      "loss": 0.7534,
      "step": 10130
    },
    {
      "epoch": 0.6897372672391804,
      "grad_norm": 0.24546845257282257,
      "learning_rate": 4.821217959747585e-06,
      "loss": 0.7657,
      "step": 10140
    },
    {
      "epoch": 0.6904174815066746,
      "grad_norm": 0.2601371109485626,
      "learning_rate": 4.801987994241921e-06,
      "loss": 0.7578,
      "step": 10150
    },
    {
      "epoch": 0.6910976957741689,
      "grad_norm": 0.2417573481798172,
      "learning_rate": 4.782784330978003e-06,
      "loss": 0.7666,
      "step": 10160
    },
    {
      "epoch": 0.6917779100416631,
      "grad_norm": 0.23398882150650024,
      "learning_rate": 4.763607067127479e-06,
      "loss": 0.7694,
      "step": 10170
    },
    {
      "epoch": 0.6924581243091574,
      "grad_norm": 0.24017900228500366,
      "learning_rate": 4.744456299728417e-06,
      "loss": 0.7645,
      "step": 10180
    },
    {
      "epoch": 0.6931383385766516,
      "grad_norm": 0.2445467710494995,
      "learning_rate": 4.725332125684807e-06,
      "loss": 0.7655,
      "step": 10190
    },
    {
      "epoch": 0.693818552844146,
      "grad_norm": 0.25575804710388184,
      "learning_rate": 4.7062346417660775e-06,
      "loss": 0.7612,
      "step": 10200
    },
    {
      "epoch": 0.6944987671116402,
      "grad_norm": 0.24726071953773499,
      "learning_rate": 4.687163944606603e-06,
      "loss": 0.7607,
      "step": 10210
    },
    {
      "epoch": 0.6951789813791345,
      "grad_norm": 0.24544966220855713,
      "learning_rate": 4.6681201307052144e-06,
      "loss": 0.7523,
      "step": 10220
    },
    {
      "epoch": 0.6958591956466287,
      "grad_norm": 0.24802793562412262,
      "learning_rate": 4.649103296424716e-06,
      "loss": 0.77,
      "step": 10230
    },
    {
      "epoch": 0.696539409914123,
      "grad_norm": 0.24389968812465668,
      "learning_rate": 4.630113537991388e-06,
      "loss": 0.7582,
      "step": 10240
    },
    {
      "epoch": 0.6972196241816172,
      "grad_norm": 0.24656598269939423,
      "learning_rate": 4.611150951494506e-06,
      "loss": 0.7747,
      "step": 10250
    },
    {
      "epoch": 0.6978998384491115,
      "grad_norm": 0.24470894038677216,
      "learning_rate": 4.59221563288586e-06,
      "loss": 0.7651,
      "step": 10260
    },
    {
      "epoch": 0.6985800527166057,
      "grad_norm": 0.2406274378299713,
      "learning_rate": 4.573307677979255e-06,
      "loss": 0.7576,
      "step": 10270
    },
    {
      "epoch": 0.6992602669841,
      "grad_norm": 0.24292300641536713,
      "learning_rate": 4.554427182450039e-06,
      "loss": 0.7615,
      "step": 10280
    },
    {
      "epoch": 0.6999404812515942,
      "grad_norm": 0.2497672587633133,
      "learning_rate": 4.535574241834615e-06,
      "loss": 0.7625,
      "step": 10290
    },
    {
      "epoch": 0.7006206955190886,
      "grad_norm": 0.24484290182590485,
      "learning_rate": 4.51674895152995e-06,
      "loss": 0.758,
      "step": 10300
    },
    {
      "epoch": 0.7013009097865828,
      "grad_norm": 0.25799936056137085,
      "learning_rate": 4.497951406793105e-06,
      "loss": 0.7546,
      "step": 10310
    },
    {
      "epoch": 0.7019811240540771,
      "grad_norm": 0.24973861873149872,
      "learning_rate": 4.479181702740746e-06,
      "loss": 0.7709,
      "step": 10320
    },
    {
      "epoch": 0.7026613383215713,
      "grad_norm": 0.24379494786262512,
      "learning_rate": 4.460439934348661e-06,
      "loss": 0.7653,
      "step": 10330
    },
    {
      "epoch": 0.7033415525890656,
      "grad_norm": 0.24392396211624146,
      "learning_rate": 4.441726196451284e-06,
      "loss": 0.7573,
      "step": 10340
    },
    {
      "epoch": 0.7040217668565598,
      "grad_norm": 0.26352494955062866,
      "learning_rate": 4.423040583741211e-06,
      "loss": 0.7665,
      "step": 10350
    },
    {
      "epoch": 0.704701981124054,
      "grad_norm": 0.2395162284374237,
      "learning_rate": 4.4043831907687255e-06,
      "loss": 0.7475,
      "step": 10360
    },
    {
      "epoch": 0.7053821953915483,
      "grad_norm": 0.2522451877593994,
      "learning_rate": 4.385754111941316e-06,
      "loss": 0.7573,
      "step": 10370
    },
    {
      "epoch": 0.7060624096590425,
      "grad_norm": 0.2469368726015091,
      "learning_rate": 4.367153441523202e-06,
      "loss": 0.7517,
      "step": 10380
    },
    {
      "epoch": 0.7067426239265369,
      "grad_norm": 0.24479222297668457,
      "learning_rate": 4.34858127363485e-06,
      "loss": 0.7667,
      "step": 10390
    },
    {
      "epoch": 0.7074228381940311,
      "grad_norm": 0.24857044219970703,
      "learning_rate": 4.330037702252505e-06,
      "loss": 0.7595,
      "step": 10400
    },
    {
      "epoch": 0.7081030524615254,
      "grad_norm": 0.2452039271593094,
      "learning_rate": 4.311522821207715e-06,
      "loss": 0.7381,
      "step": 10410
    },
    {
      "epoch": 0.7087832667290196,
      "grad_norm": 0.24671368300914764,
      "learning_rate": 4.293036724186848e-06,
      "loss": 0.76,
      "step": 10420
    },
    {
      "epoch": 0.7094634809965139,
      "grad_norm": 0.25648754835128784,
      "learning_rate": 4.274579504730626e-06,
      "loss": 0.765,
      "step": 10430
    },
    {
      "epoch": 0.7101436952640081,
      "grad_norm": 0.2619253993034363,
      "learning_rate": 4.2561512562336475e-06,
      "loss": 0.7707,
      "step": 10440
    },
    {
      "epoch": 0.7108239095315024,
      "grad_norm": 0.2492845207452774,
      "learning_rate": 4.237752071943917e-06,
      "loss": 0.7582,
      "step": 10450
    },
    {
      "epoch": 0.7115041237989966,
      "grad_norm": 0.2471589595079422,
      "learning_rate": 4.219382044962374e-06,
      "loss": 0.7616,
      "step": 10460
    },
    {
      "epoch": 0.712184338066491,
      "grad_norm": 0.25441455841064453,
      "learning_rate": 4.201041268242418e-06,
      "loss": 0.7698,
      "step": 10470
    },
    {
      "epoch": 0.7128645523339852,
      "grad_norm": 0.25763577222824097,
      "learning_rate": 4.1827298345894394e-06,
      "loss": 0.7462,
      "step": 10480
    },
    {
      "epoch": 0.7135447666014795,
      "grad_norm": 0.24413402378559113,
      "learning_rate": 4.164447836660354e-06,
      "loss": 0.7664,
      "step": 10490
    },
    {
      "epoch": 0.7142249808689737,
      "grad_norm": 0.24375353753566742,
      "learning_rate": 4.146195366963133e-06,
      "loss": 0.7617,
      "step": 10500
    },
    {
      "epoch": 0.714905195136468,
      "grad_norm": 0.25362712144851685,
      "learning_rate": 4.127972517856326e-06,
      "loss": 0.7672,
      "step": 10510
    },
    {
      "epoch": 0.7155854094039622,
      "grad_norm": 0.25445815920829773,
      "learning_rate": 4.109779381548605e-06,
      "loss": 0.7639,
      "step": 10520
    },
    {
      "epoch": 0.7162656236714565,
      "grad_norm": 0.24999208748340607,
      "learning_rate": 4.091616050098296e-06,
      "loss": 0.7601,
      "step": 10530
    },
    {
      "epoch": 0.7169458379389507,
      "grad_norm": 0.2546180784702301,
      "learning_rate": 4.073482615412905e-06,
      "loss": 0.7778,
      "step": 10540
    },
    {
      "epoch": 0.717626052206445,
      "grad_norm": 0.24735789000988007,
      "learning_rate": 4.055379169248663e-06,
      "loss": 0.7719,
      "step": 10550
    },
    {
      "epoch": 0.7183062664739392,
      "grad_norm": 0.2446472942829132,
      "learning_rate": 4.037305803210057e-06,
      "loss": 0.7729,
      "step": 10560
    },
    {
      "epoch": 0.7189864807414336,
      "grad_norm": 0.2434018850326538,
      "learning_rate": 4.019262608749362e-06,
      "loss": 0.765,
      "step": 10570
    },
    {
      "epoch": 0.7196666950089278,
      "grad_norm": 0.24566373229026794,
      "learning_rate": 4.00124967716619e-06,
      "loss": 0.7629,
      "step": 10580
    },
    {
      "epoch": 0.7203469092764221,
      "grad_norm": 0.2555360496044159,
      "learning_rate": 3.9832670996070175e-06,
      "loss": 0.7533,
      "step": 10590
    },
    {
      "epoch": 0.7210271235439163,
      "grad_norm": 0.25441837310791016,
      "learning_rate": 3.9653149670647274e-06,
      "loss": 0.7723,
      "step": 10600
    },
    {
      "epoch": 0.7217073378114106,
      "grad_norm": 0.24170216917991638,
      "learning_rate": 3.947393370378151e-06,
      "loss": 0.7526,
      "step": 10610
    },
    {
      "epoch": 0.7223875520789048,
      "grad_norm": 0.24596717953681946,
      "learning_rate": 3.929502400231603e-06,
      "loss": 0.7587,
      "step": 10620
    },
    {
      "epoch": 0.7230677663463991,
      "grad_norm": 0.24249835312366486,
      "learning_rate": 3.911642147154431e-06,
      "loss": 0.7504,
      "step": 10630
    },
    {
      "epoch": 0.7237479806138933,
      "grad_norm": 0.23611494898796082,
      "learning_rate": 3.8938127015205484e-06,
      "loss": 0.7547,
      "step": 10640
    },
    {
      "epoch": 0.7244281948813877,
      "grad_norm": 0.24255642294883728,
      "learning_rate": 3.8760141535479835e-06,
      "loss": 0.7564,
      "step": 10650
    },
    {
      "epoch": 0.7251084091488819,
      "grad_norm": 0.24236130714416504,
      "learning_rate": 3.858246593298418e-06,
      "loss": 0.7567,
      "step": 10660
    },
    {
      "epoch": 0.7257886234163762,
      "grad_norm": 0.25300332903862,
      "learning_rate": 3.840510110676737e-06,
      "loss": 0.769,
      "step": 10670
    },
    {
      "epoch": 0.7264688376838704,
      "grad_norm": 0.24559025466442108,
      "learning_rate": 3.822804795430567e-06,
      "loss": 0.7617,
      "step": 10680
    },
    {
      "epoch": 0.7271490519513647,
      "grad_norm": 0.2505420446395874,
      "learning_rate": 3.8051307371498323e-06,
      "loss": 0.7678,
      "step": 10690
    },
    {
      "epoch": 0.7278292662188589,
      "grad_norm": 0.25066983699798584,
      "learning_rate": 3.7874880252662905e-06,
      "loss": 0.7725,
      "step": 10700
    },
    {
      "epoch": 0.7285094804863532,
      "grad_norm": 0.25248897075653076,
      "learning_rate": 3.769876749053084e-06,
      "loss": 0.7728,
      "step": 10710
    },
    {
      "epoch": 0.7291896947538474,
      "grad_norm": 0.23779115080833435,
      "learning_rate": 3.7522969976242917e-06,
      "loss": 0.7559,
      "step": 10720
    },
    {
      "epoch": 0.7298699090213417,
      "grad_norm": 0.248768612742424,
      "learning_rate": 3.7347488599344716e-06,
      "loss": 0.7719,
      "step": 10730
    },
    {
      "epoch": 0.730550123288836,
      "grad_norm": 0.245006263256073,
      "learning_rate": 3.7172324247782188e-06,
      "loss": 0.7632,
      "step": 10740
    },
    {
      "epoch": 0.7312303375563303,
      "grad_norm": 0.23913902044296265,
      "learning_rate": 3.6997477807897075e-06,
      "loss": 0.7488,
      "step": 10750
    },
    {
      "epoch": 0.7319105518238245,
      "grad_norm": 0.24731720983982086,
      "learning_rate": 3.682295016442251e-06,
      "loss": 0.7764,
      "step": 10760
    },
    {
      "epoch": 0.7325907660913188,
      "grad_norm": 0.23971058428287506,
      "learning_rate": 3.664874220047845e-06,
      "loss": 0.7626,
      "step": 10770
    },
    {
      "epoch": 0.733270980358813,
      "grad_norm": 0.24796296656131744,
      "learning_rate": 3.6474854797567305e-06,
      "loss": 0.7585,
      "step": 10780
    },
    {
      "epoch": 0.7339511946263073,
      "grad_norm": 0.24577690660953522,
      "learning_rate": 3.630128883556938e-06,
      "loss": 0.7627,
      "step": 10790
    },
    {
      "epoch": 0.7346314088938015,
      "grad_norm": 0.25564873218536377,
      "learning_rate": 3.6128045192738503e-06,
      "loss": 0.7571,
      "step": 10800
    },
    {
      "epoch": 0.7353116231612958,
      "grad_norm": 0.2462102174758911,
      "learning_rate": 3.5955124745697545e-06,
      "loss": 0.7675,
      "step": 10810
    },
    {
      "epoch": 0.73599183742879,
      "grad_norm": 0.29949715733528137,
      "learning_rate": 3.578252836943398e-06,
      "loss": 0.7658,
      "step": 10820
    },
    {
      "epoch": 0.7366720516962844,
      "grad_norm": 0.2516492009162903,
      "learning_rate": 3.561025693729545e-06,
      "loss": 0.7717,
      "step": 10830
    },
    {
      "epoch": 0.7373522659637786,
      "grad_norm": 0.2486589401960373,
      "learning_rate": 3.5438311320985397e-06,
      "loss": 0.7645,
      "step": 10840
    },
    {
      "epoch": 0.7380324802312729,
      "grad_norm": 0.24454742670059204,
      "learning_rate": 3.526669239055859e-06,
      "loss": 0.7699,
      "step": 10850
    },
    {
      "epoch": 0.7387126944987671,
      "grad_norm": 0.24663999676704407,
      "learning_rate": 3.5095401014416753e-06,
      "loss": 0.7823,
      "step": 10860
    },
    {
      "epoch": 0.7393929087662614,
      "grad_norm": 0.24899105727672577,
      "learning_rate": 3.4924438059304168e-06,
      "loss": 0.7652,
      "step": 10870
    },
    {
      "epoch": 0.7400731230337556,
      "grad_norm": 0.24318116903305054,
      "learning_rate": 3.475380439030328e-06,
      "loss": 0.7585,
      "step": 10880
    },
    {
      "epoch": 0.7407533373012499,
      "grad_norm": 0.24379071593284607,
      "learning_rate": 3.4583500870830356e-06,
      "loss": 0.7751,
      "step": 10890
    },
    {
      "epoch": 0.7414335515687441,
      "grad_norm": 0.24555222690105438,
      "learning_rate": 3.4413528362631043e-06,
      "loss": 0.7516,
      "step": 10900
    },
    {
      "epoch": 0.7421137658362384,
      "grad_norm": 0.2604420781135559,
      "learning_rate": 3.4243887725776093e-06,
      "loss": 0.7675,
      "step": 10910
    },
    {
      "epoch": 0.7427939801037327,
      "grad_norm": 0.25211283564567566,
      "learning_rate": 3.4074579818656972e-06,
      "loss": 0.7617,
      "step": 10920
    },
    {
      "epoch": 0.743474194371227,
      "grad_norm": 0.24558304250240326,
      "learning_rate": 3.390560549798143e-06,
      "loss": 0.7775,
      "step": 10930
    },
    {
      "epoch": 0.7441544086387212,
      "grad_norm": 0.25397711992263794,
      "learning_rate": 3.373696561876936e-06,
      "loss": 0.7494,
      "step": 10940
    },
    {
      "epoch": 0.7448346229062155,
      "grad_norm": 0.24445897340774536,
      "learning_rate": 3.3568661034348303e-06,
      "loss": 0.766,
      "step": 10950
    },
    {
      "epoch": 0.7455148371737097,
      "grad_norm": 0.24605581164360046,
      "learning_rate": 3.3400692596349206e-06,
      "loss": 0.7508,
      "step": 10960
    },
    {
      "epoch": 0.746195051441204,
      "grad_norm": 0.24277006089687347,
      "learning_rate": 3.3233061154702086e-06,
      "loss": 0.7616,
      "step": 10970
    },
    {
      "epoch": 0.7468752657086982,
      "grad_norm": 0.2509525418281555,
      "learning_rate": 3.3065767557631757e-06,
      "loss": 0.7653,
      "step": 10980
    },
    {
      "epoch": 0.7475554799761925,
      "grad_norm": 0.2534981071949005,
      "learning_rate": 3.289881265165349e-06,
      "loss": 0.7752,
      "step": 10990
    },
    {
      "epoch": 0.7482356942436867,
      "grad_norm": 0.24551300704479218,
      "learning_rate": 3.273219728156879e-06,
      "loss": 0.7719,
      "step": 11000
    },
    {
      "epoch": 0.7489159085111811,
      "grad_norm": 0.2448212206363678,
      "learning_rate": 3.256592229046108e-06,
      "loss": 0.7564,
      "step": 11010
    },
    {
      "epoch": 0.7495961227786753,
      "grad_norm": 0.254169762134552,
      "learning_rate": 3.239998851969144e-06,
      "loss": 0.7488,
      "step": 11020
    },
    {
      "epoch": 0.7502763370461696,
      "grad_norm": 0.24657952785491943,
      "learning_rate": 3.2234396808894353e-06,
      "loss": 0.7653,
      "step": 11030
    },
    {
      "epoch": 0.7509565513136638,
      "grad_norm": 0.2531900107860565,
      "learning_rate": 3.2069147995973473e-06,
      "loss": 0.7481,
      "step": 11040
    },
    {
      "epoch": 0.7516367655811581,
      "grad_norm": 0.25895965099334717,
      "learning_rate": 3.190424291709735e-06,
      "loss": 0.7577,
      "step": 11050
    },
    {
      "epoch": 0.7523169798486523,
      "grad_norm": 0.25525858998298645,
      "learning_rate": 3.173968240669525e-06,
      "loss": 0.7679,
      "step": 11060
    },
    {
      "epoch": 0.7529971941161466,
      "grad_norm": 0.24357037246227264,
      "learning_rate": 3.1575467297452868e-06,
      "loss": 0.7567,
      "step": 11070
    },
    {
      "epoch": 0.7536774083836408,
      "grad_norm": 0.24586918950080872,
      "learning_rate": 3.1411598420308175e-06,
      "loss": 0.7601,
      "step": 11080
    },
    {
      "epoch": 0.7543576226511352,
      "grad_norm": 0.2497064471244812,
      "learning_rate": 3.1248076604447177e-06,
      "loss": 0.7553,
      "step": 11090
    },
    {
      "epoch": 0.7550378369186294,
      "grad_norm": 0.2547801434993744,
      "learning_rate": 3.1084902677299733e-06,
      "loss": 0.769,
      "step": 11100
    },
    {
      "epoch": 0.7557180511861237,
      "grad_norm": 0.25114110112190247,
      "learning_rate": 3.092207746453535e-06,
      "loss": 0.7658,
      "step": 11110
    },
    {
      "epoch": 0.7563982654536179,
      "grad_norm": 0.2509963810443878,
      "learning_rate": 3.075960179005906e-06,
      "loss": 0.7631,
      "step": 11120
    },
    {
      "epoch": 0.7570784797211122,
      "grad_norm": 0.24080003798007965,
      "learning_rate": 3.0597476476007148e-06,
      "loss": 0.7514,
      "step": 11130
    },
    {
      "epoch": 0.7577586939886064,
      "grad_norm": 0.23864655196666718,
      "learning_rate": 3.0435702342743122e-06,
      "loss": 0.7471,
      "step": 11140
    },
    {
      "epoch": 0.7584389082561007,
      "grad_norm": 0.24954387545585632,
      "learning_rate": 3.0274280208853455e-06,
      "loss": 0.7598,
      "step": 11150
    },
    {
      "epoch": 0.7591191225235949,
      "grad_norm": 0.2520304322242737,
      "learning_rate": 3.0113210891143486e-06,
      "loss": 0.7677,
      "step": 11160
    },
    {
      "epoch": 0.7597993367910892,
      "grad_norm": 0.23881210386753082,
      "learning_rate": 2.9952495204633293e-06,
      "loss": 0.7517,
      "step": 11170
    },
    {
      "epoch": 0.7604795510585834,
      "grad_norm": 0.24030640721321106,
      "learning_rate": 2.979213396255356e-06,
      "loss": 0.7669,
      "step": 11180
    },
    {
      "epoch": 0.7611597653260778,
      "grad_norm": 0.24254277348518372,
      "learning_rate": 2.963212797634145e-06,
      "loss": 0.7777,
      "step": 11190
    },
    {
      "epoch": 0.761839979593572,
      "grad_norm": 0.2521003186702728,
      "learning_rate": 2.947247805563652e-06,
      "loss": 0.7735,
      "step": 11200
    },
    {
      "epoch": 0.7625201938610663,
      "grad_norm": 0.24352481961250305,
      "learning_rate": 2.9313185008276634e-06,
      "loss": 0.76,
      "step": 11210
    },
    {
      "epoch": 0.7632004081285605,
      "grad_norm": 0.24368160963058472,
      "learning_rate": 2.9154249640293806e-06,
      "loss": 0.7769,
      "step": 11220
    },
    {
      "epoch": 0.7638806223960548,
      "grad_norm": 0.2543738782405853,
      "learning_rate": 2.8995672755910208e-06,
      "loss": 0.7659,
      "step": 11230
    },
    {
      "epoch": 0.764560836663549,
      "grad_norm": 0.24566735327243805,
      "learning_rate": 2.883745515753408e-06,
      "loss": 0.7739,
      "step": 11240
    },
    {
      "epoch": 0.7652410509310433,
      "grad_norm": 0.24797692894935608,
      "learning_rate": 2.86795976457556e-06,
      "loss": 0.7712,
      "step": 11250
    },
    {
      "epoch": 0.7659212651985375,
      "grad_norm": 0.252880334854126,
      "learning_rate": 2.852210101934294e-06,
      "loss": 0.7662,
      "step": 11260
    },
    {
      "epoch": 0.7666014794660319,
      "grad_norm": 0.26727163791656494,
      "learning_rate": 2.836496607523813e-06,
      "loss": 0.7647,
      "step": 11270
    },
    {
      "epoch": 0.7672816937335261,
      "grad_norm": 0.2518885135650635,
      "learning_rate": 2.8208193608553115e-06,
      "loss": 0.7589,
      "step": 11280
    },
    {
      "epoch": 0.7679619080010204,
      "grad_norm": 0.25065723061561584,
      "learning_rate": 2.805178441256564e-06,
      "loss": 0.7597,
      "step": 11290
    },
    {
      "epoch": 0.7686421222685146,
      "grad_norm": 0.2463270127773285,
      "learning_rate": 2.789573927871528e-06,
      "loss": 0.7574,
      "step": 11300
    },
    {
      "epoch": 0.7693223365360088,
      "grad_norm": 0.254273921251297,
      "learning_rate": 2.7740058996599475e-06,
      "loss": 0.7617,
      "step": 11310
    },
    {
      "epoch": 0.7700025508035031,
      "grad_norm": 0.2485022246837616,
      "learning_rate": 2.758474435396944e-06,
      "loss": 0.7761,
      "step": 11320
    },
    {
      "epoch": 0.7706827650709973,
      "grad_norm": 0.23881711065769196,
      "learning_rate": 2.7429796136726283e-06,
      "loss": 0.7586,
      "step": 11330
    },
    {
      "epoch": 0.7713629793384916,
      "grad_norm": 0.24466168880462646,
      "learning_rate": 2.7275215128916943e-06,
      "loss": 0.7641,
      "step": 11340
    },
    {
      "epoch": 0.7720431936059858,
      "grad_norm": 0.24188317358493805,
      "learning_rate": 2.712100211273028e-06,
      "loss": 0.7547,
      "step": 11350
    },
    {
      "epoch": 0.7727234078734802,
      "grad_norm": 0.3558238744735718,
      "learning_rate": 2.6967157868493066e-06,
      "loss": 0.758,
      "step": 11360
    },
    {
      "epoch": 0.7734036221409744,
      "grad_norm": 0.24949957430362701,
      "learning_rate": 2.6813683174666106e-06,
      "loss": 0.7696,
      "step": 11370
    },
    {
      "epoch": 0.7740838364084687,
      "grad_norm": 0.25859785079956055,
      "learning_rate": 2.666057880784024e-06,
      "loss": 0.7612,
      "step": 11380
    },
    {
      "epoch": 0.7747640506759629,
      "grad_norm": 0.24926084280014038,
      "learning_rate": 2.650784554273241e-06,
      "loss": 0.751,
      "step": 11390
    },
    {
      "epoch": 0.7754442649434572,
      "grad_norm": 0.24230580031871796,
      "learning_rate": 2.63554841521818e-06,
      "loss": 0.7622,
      "step": 11400
    },
    {
      "epoch": 0.7761244792109514,
      "grad_norm": 0.24734055995941162,
      "learning_rate": 2.620349540714586e-06,
      "loss": 0.7696,
      "step": 11410
    },
    {
      "epoch": 0.7768046934784457,
      "grad_norm": 0.24714922904968262,
      "learning_rate": 2.6051880076696446e-06,
      "loss": 0.7726,
      "step": 11420
    },
    {
      "epoch": 0.7774849077459399,
      "grad_norm": 0.2531922161579132,
      "learning_rate": 2.59006389280159e-06,
      "loss": 0.7703,
      "step": 11430
    },
    {
      "epoch": 0.7781651220134342,
      "grad_norm": 0.24530744552612305,
      "learning_rate": 2.5749772726393198e-06,
      "loss": 0.7623,
      "step": 11440
    },
    {
      "epoch": 0.7788453362809284,
      "grad_norm": 0.23901546001434326,
      "learning_rate": 2.559928223522006e-06,
      "loss": 0.7695,
      "step": 11450
    },
    {
      "epoch": 0.7795255505484228,
      "grad_norm": 0.24316571652889252,
      "learning_rate": 2.5449168215987074e-06,
      "loss": 0.7451,
      "step": 11460
    },
    {
      "epoch": 0.780205764815917,
      "grad_norm": 0.2481565922498703,
      "learning_rate": 2.5299431428279884e-06,
      "loss": 0.7626,
      "step": 11470
    },
    {
      "epoch": 0.7808859790834113,
      "grad_norm": 0.24083758890628815,
      "learning_rate": 2.5150072629775314e-06,
      "loss": 0.7717,
      "step": 11480
    },
    {
      "epoch": 0.7815661933509055,
      "grad_norm": 0.24756860733032227,
      "learning_rate": 2.500109257623751e-06,
      "loss": 0.7647,
      "step": 11490
    },
    {
      "epoch": 0.7822464076183998,
      "grad_norm": 0.24836762249469757,
      "learning_rate": 2.485249202151417e-06,
      "loss": 0.7684,
      "step": 11500
    },
    {
      "epoch": 0.782926621885894,
      "grad_norm": 0.24777284264564514,
      "learning_rate": 2.4704271717532724e-06,
      "loss": 0.7641,
      "step": 11510
    },
    {
      "epoch": 0.7836068361533883,
      "grad_norm": 0.2474483698606491,
      "learning_rate": 2.4556432414296485e-06,
      "loss": 0.7708,
      "step": 11520
    },
    {
      "epoch": 0.7842870504208825,
      "grad_norm": 0.24180737137794495,
      "learning_rate": 2.4408974859880884e-06,
      "loss": 0.7556,
      "step": 11530
    },
    {
      "epoch": 0.7849672646883769,
      "grad_norm": 0.24215582013130188,
      "learning_rate": 2.4261899800429702e-06,
      "loss": 0.7626,
      "step": 11540
    },
    {
      "epoch": 0.7856474789558711,
      "grad_norm": 0.2542628347873688,
      "learning_rate": 2.4115207980151225e-06,
      "loss": 0.7549,
      "step": 11550
    },
    {
      "epoch": 0.7863276932233654,
      "grad_norm": 0.244821697473526,
      "learning_rate": 2.396890014131459e-06,
      "loss": 0.7607,
      "step": 11560
    },
    {
      "epoch": 0.7870079074908596,
      "grad_norm": 0.2469116747379303,
      "learning_rate": 2.382297702424591e-06,
      "loss": 0.7712,
      "step": 11570
    },
    {
      "epoch": 0.7876881217583539,
      "grad_norm": 0.24981118738651276,
      "learning_rate": 2.3677439367324628e-06,
      "loss": 0.7527,
      "step": 11580
    },
    {
      "epoch": 0.7883683360258481,
      "grad_norm": 0.24756966531276703,
      "learning_rate": 2.353228790697969e-06,
      "loss": 0.7731,
      "step": 11590
    },
    {
      "epoch": 0.7890485502933424,
      "grad_norm": 0.2556615173816681,
      "learning_rate": 2.3387523377685915e-06,
      "loss": 0.7545,
      "step": 11600
    },
    {
      "epoch": 0.7897287645608366,
      "grad_norm": 0.25671324133872986,
      "learning_rate": 2.324314651196019e-06,
      "loss": 0.7656,
      "step": 11610
    },
    {
      "epoch": 0.7904089788283309,
      "grad_norm": 0.2505514919757843,
      "learning_rate": 2.3099158040357827e-06,
      "loss": 0.7579,
      "step": 11620
    },
    {
      "epoch": 0.7910891930958251,
      "grad_norm": 0.2539149820804596,
      "learning_rate": 2.2955558691468827e-06,
      "loss": 0.7674,
      "step": 11630
    },
    {
      "epoch": 0.7917694073633195,
      "grad_norm": 0.24224653840065002,
      "learning_rate": 2.2812349191914197e-06,
      "loss": 0.7535,
      "step": 11640
    },
    {
      "epoch": 0.7924496216308137,
      "grad_norm": 0.24667827785015106,
      "learning_rate": 2.266953026634231e-06,
      "loss": 0.7617,
      "step": 11650
    },
    {
      "epoch": 0.793129835898308,
      "grad_norm": 0.5902561545372009,
      "learning_rate": 2.252710263742519e-06,
      "loss": 0.7465,
      "step": 11660
    },
    {
      "epoch": 0.7938100501658022,
      "grad_norm": 0.2432839274406433,
      "learning_rate": 2.23850670258549e-06,
      "loss": 0.756,
      "step": 11670
    },
    {
      "epoch": 0.7944902644332965,
      "grad_norm": 0.2495686411857605,
      "learning_rate": 2.2243424150339853e-06,
      "loss": 0.7635,
      "step": 11680
    },
    {
      "epoch": 0.7951704787007907,
      "grad_norm": 0.24666094779968262,
      "learning_rate": 2.210217472760121e-06,
      "loss": 0.7577,
      "step": 11690
    },
    {
      "epoch": 0.795850692968285,
      "grad_norm": 0.25573796033859253,
      "learning_rate": 2.1961319472369214e-06,
      "loss": 0.7694,
      "step": 11700
    },
    {
      "epoch": 0.7965309072357792,
      "grad_norm": 0.24457398056983948,
      "learning_rate": 2.182085909737961e-06,
      "loss": 0.7485,
      "step": 11710
    },
    {
      "epoch": 0.7972111215032736,
      "grad_norm": 0.24250029027462006,
      "learning_rate": 2.1680794313370044e-06,
      "loss": 0.7612,
      "step": 11720
    },
    {
      "epoch": 0.7978913357707678,
      "grad_norm": 0.2539270222187042,
      "learning_rate": 2.1541125829076436e-06,
      "loss": 0.7544,
      "step": 11730
    },
    {
      "epoch": 0.7985715500382621,
      "grad_norm": 0.2482372373342514,
      "learning_rate": 2.140185435122939e-06,
      "loss": 0.7616,
      "step": 11740
    },
    {
      "epoch": 0.7992517643057563,
      "grad_norm": 0.2496422976255417,
      "learning_rate": 2.126298058455066e-06,
      "loss": 0.7509,
      "step": 11750
    },
    {
      "epoch": 0.7999319785732506,
      "grad_norm": 0.24733223021030426,
      "learning_rate": 2.1124505231749537e-06,
      "loss": 0.7622,
      "step": 11760
    },
    {
      "epoch": 0.8006121928407448,
      "grad_norm": 0.2519744038581848,
      "learning_rate": 2.0986428993519324e-06,
      "loss": 0.7608,
      "step": 11770
    },
    {
      "epoch": 0.8012924071082391,
      "grad_norm": 0.250882089138031,
      "learning_rate": 2.0848752568533804e-06,
      "loss": 0.7626,
      "step": 11780
    },
    {
      "epoch": 0.8019726213757333,
      "grad_norm": 0.2523823380470276,
      "learning_rate": 2.071147665344363e-06,
      "loss": 0.7615,
      "step": 11790
    },
    {
      "epoch": 0.8026528356432276,
      "grad_norm": 0.252625972032547,
      "learning_rate": 2.0574601942872895e-06,
      "loss": 0.7655,
      "step": 11800
    },
    {
      "epoch": 0.8033330499107219,
      "grad_norm": 0.2436279058456421,
      "learning_rate": 2.0438129129415564e-06,
      "loss": 0.7458,
      "step": 11810
    },
    {
      "epoch": 0.8040132641782162,
      "grad_norm": 0.24941012263298035,
      "learning_rate": 2.030205890363197e-06,
      "loss": 0.7644,
      "step": 11820
    },
    {
      "epoch": 0.8046934784457104,
      "grad_norm": 0.24483945965766907,
      "learning_rate": 2.016639195404534e-06,
      "loss": 0.7511,
      "step": 11830
    },
    {
      "epoch": 0.8053736927132047,
      "grad_norm": 0.2422882467508316,
      "learning_rate": 2.003112896713829e-06,
      "loss": 0.7584,
      "step": 11840
    },
    {
      "epoch": 0.8060539069806989,
      "grad_norm": 0.246101513504982,
      "learning_rate": 1.9896270627349367e-06,
      "loss": 0.7698,
      "step": 11850
    },
    {
      "epoch": 0.8067341212481932,
      "grad_norm": 0.24442127346992493,
      "learning_rate": 1.976181761706959e-06,
      "loss": 0.7584,
      "step": 11860
    },
    {
      "epoch": 0.8074143355156874,
      "grad_norm": 0.24296994507312775,
      "learning_rate": 1.9627770616638953e-06,
      "loss": 0.7692,
      "step": 11870
    },
    {
      "epoch": 0.8080945497831817,
      "grad_norm": 0.24183446168899536,
      "learning_rate": 1.9494130304343064e-06,
      "loss": 0.748,
      "step": 11880
    },
    {
      "epoch": 0.8087747640506759,
      "grad_norm": 0.24936461448669434,
      "learning_rate": 1.9360897356409627e-06,
      "loss": 0.7562,
      "step": 11890
    },
    {
      "epoch": 0.8094549783181703,
      "grad_norm": 0.25846680998802185,
      "learning_rate": 1.9228072447005086e-06,
      "loss": 0.7644,
      "step": 11900
    },
    {
      "epoch": 0.8101351925856645,
      "grad_norm": 0.2411743700504303,
      "learning_rate": 1.9095656248231167e-06,
      "loss": 0.749,
      "step": 11910
    },
    {
      "epoch": 0.8108154068531588,
      "grad_norm": 0.24309656023979187,
      "learning_rate": 1.896364943012149e-06,
      "loss": 0.7574,
      "step": 11920
    },
    {
      "epoch": 0.811495621120653,
      "grad_norm": 0.24680094420909882,
      "learning_rate": 1.8832052660638223e-06,
      "loss": 0.7683,
      "step": 11930
    },
    {
      "epoch": 0.8121758353881473,
      "grad_norm": 0.2567564845085144,
      "learning_rate": 1.8700866605668622e-06,
      "loss": 0.7664,
      "step": 11940
    },
    {
      "epoch": 0.8128560496556415,
      "grad_norm": 0.24540911614894867,
      "learning_rate": 1.8570091929021738e-06,
      "loss": 0.7622,
      "step": 11950
    },
    {
      "epoch": 0.8135362639231358,
      "grad_norm": 0.24130462110042572,
      "learning_rate": 1.843972929242499e-06,
      "loss": 0.7405,
      "step": 11960
    },
    {
      "epoch": 0.81421647819063,
      "grad_norm": 0.23911410570144653,
      "learning_rate": 1.8309779355520885e-06,
      "loss": 0.7724,
      "step": 11970
    },
    {
      "epoch": 0.8148966924581243,
      "grad_norm": 0.24407736957073212,
      "learning_rate": 1.818024277586361e-06,
      "loss": 0.7581,
      "step": 11980
    },
    {
      "epoch": 0.8155769067256186,
      "grad_norm": 0.2502209544181824,
      "learning_rate": 1.8051120208915774e-06,
      "loss": 0.7427,
      "step": 11990
    },
    {
      "epoch": 0.8162571209931129,
      "grad_norm": 0.252231627702713,
      "learning_rate": 1.7922412308045035e-06,
      "loss": 0.7646,
      "step": 12000
    },
    {
      "epoch": 0.8169373352606071,
      "grad_norm": 0.2540828585624695,
      "learning_rate": 1.7794119724520832e-06,
      "loss": 0.7605,
      "step": 12010
    },
    {
      "epoch": 0.8176175495281014,
      "grad_norm": 0.24380894005298615,
      "learning_rate": 1.766624310751106e-06,
      "loss": 0.7566,
      "step": 12020
    },
    {
      "epoch": 0.8182977637955956,
      "grad_norm": 0.24279260635375977,
      "learning_rate": 1.7538783104078804e-06,
      "loss": 0.7372,
      "step": 12030
    },
    {
      "epoch": 0.8189779780630899,
      "grad_norm": 0.24932003021240234,
      "learning_rate": 1.7411740359179075e-06,
      "loss": 0.7704,
      "step": 12040
    },
    {
      "epoch": 0.8196581923305841,
      "grad_norm": 0.24557797610759735,
      "learning_rate": 1.7285115515655492e-06,
      "loss": 0.7509,
      "step": 12050
    },
    {
      "epoch": 0.8203384065980784,
      "grad_norm": 0.2418263554573059,
      "learning_rate": 1.715890921423713e-06,
      "loss": 0.7537,
      "step": 12060
    },
    {
      "epoch": 0.8210186208655726,
      "grad_norm": 0.25545066595077515,
      "learning_rate": 1.703312209353516e-06,
      "loss": 0.7598,
      "step": 12070
    },
    {
      "epoch": 0.821698835133067,
      "grad_norm": 0.24509641528129578,
      "learning_rate": 1.69077547900397e-06,
      "loss": 0.7559,
      "step": 12080
    },
    {
      "epoch": 0.8223790494005612,
      "grad_norm": 0.24777643382549286,
      "learning_rate": 1.6782807938116564e-06,
      "loss": 0.7566,
      "step": 12090
    },
    {
      "epoch": 0.8230592636680555,
      "grad_norm": 0.24264010787010193,
      "learning_rate": 1.665828217000407e-06,
      "loss": 0.771,
      "step": 12100
    },
    {
      "epoch": 0.8237394779355497,
      "grad_norm": 0.24878935515880585,
      "learning_rate": 1.6534178115809795e-06,
      "loss": 0.7517,
      "step": 12110
    },
    {
      "epoch": 0.824419692203044,
      "grad_norm": 0.25836506485939026,
      "learning_rate": 1.6410496403507459e-06,
      "loss": 0.7672,
      "step": 12120
    },
    {
      "epoch": 0.8250999064705382,
      "grad_norm": 0.24952253699302673,
      "learning_rate": 1.6287237658933685e-06,
      "loss": 0.7655,
      "step": 12130
    },
    {
      "epoch": 0.8257801207380325,
      "grad_norm": 0.2452516406774521,
      "learning_rate": 1.616440250578485e-06,
      "loss": 0.77,
      "step": 12140
    },
    {
      "epoch": 0.8264603350055267,
      "grad_norm": 0.25019338726997375,
      "learning_rate": 1.6041991565613935e-06,
      "loss": 0.7571,
      "step": 12150
    },
    {
      "epoch": 0.827140549273021,
      "grad_norm": 0.24410732090473175,
      "learning_rate": 1.5920005457827403e-06,
      "loss": 0.7615,
      "step": 12160
    },
    {
      "epoch": 0.8278207635405153,
      "grad_norm": 0.25065693259239197,
      "learning_rate": 1.5798444799682023e-06,
      "loss": 0.7517,
      "step": 12170
    },
    {
      "epoch": 0.8285009778080096,
      "grad_norm": 0.2452399581670761,
      "learning_rate": 1.5677310206281748e-06,
      "loss": 0.7664,
      "step": 12180
    },
    {
      "epoch": 0.8291811920755038,
      "grad_norm": 0.24665020406246185,
      "learning_rate": 1.555660229057465e-06,
      "loss": 0.7487,
      "step": 12190
    },
    {
      "epoch": 0.8298614063429981,
      "grad_norm": 0.24330231547355652,
      "learning_rate": 1.5436321663349763e-06,
      "loss": 0.7725,
      "step": 12200
    },
    {
      "epoch": 0.8305416206104923,
      "grad_norm": 0.255862295627594,
      "learning_rate": 1.5316468933234042e-06,
      "loss": 0.7628,
      "step": 12210
    },
    {
      "epoch": 0.8312218348779866,
      "grad_norm": 0.24069169163703918,
      "learning_rate": 1.5197044706689211e-06,
      "loss": 0.7448,
      "step": 12220
    },
    {
      "epoch": 0.8319020491454808,
      "grad_norm": 0.24543890357017517,
      "learning_rate": 1.5078049588008802e-06,
      "loss": 0.7637,
      "step": 12230
    },
    {
      "epoch": 0.832582263412975,
      "grad_norm": 0.26199522614479065,
      "learning_rate": 1.4959484179314988e-06,
      "loss": 0.7701,
      "step": 12240
    },
    {
      "epoch": 0.8332624776804693,
      "grad_norm": 0.2520271837711334,
      "learning_rate": 1.4841349080555612e-06,
      "loss": 0.7724,
      "step": 12250
    },
    {
      "epoch": 0.8339426919479636,
      "grad_norm": 0.24712850153446198,
      "learning_rate": 1.472364488950111e-06,
      "loss": 0.7597,
      "step": 12260
    },
    {
      "epoch": 0.8346229062154579,
      "grad_norm": 0.25130292773246765,
      "learning_rate": 1.4606372201741504e-06,
      "loss": 0.7623,
      "step": 12270
    },
    {
      "epoch": 0.8353031204829521,
      "grad_norm": 0.2628088891506195,
      "learning_rate": 1.4489531610683382e-06,
      "loss": 0.7531,
      "step": 12280
    },
    {
      "epoch": 0.8359833347504464,
      "grad_norm": 0.24563273787498474,
      "learning_rate": 1.4373123707546921e-06,
      "loss": 0.7542,
      "step": 12290
    },
    {
      "epoch": 0.8366635490179406,
      "grad_norm": 0.24699704349040985,
      "learning_rate": 1.425714908136282e-06,
      "loss": 0.7755,
      "step": 12300
    },
    {
      "epoch": 0.8373437632854349,
      "grad_norm": 0.24536138772964478,
      "learning_rate": 1.4141608318969425e-06,
      "loss": 0.764,
      "step": 12310
    },
    {
      "epoch": 0.8380239775529291,
      "grad_norm": 0.24678239226341248,
      "learning_rate": 1.4026502005009678e-06,
      "loss": 0.7652,
      "step": 12320
    },
    {
      "epoch": 0.8387041918204234,
      "grad_norm": 0.243879035115242,
      "learning_rate": 1.3911830721928187e-06,
      "loss": 0.7596,
      "step": 12330
    },
    {
      "epoch": 0.8393844060879176,
      "grad_norm": 0.24049773812294006,
      "learning_rate": 1.3797595049968305e-06,
      "loss": 0.7662,
      "step": 12340
    },
    {
      "epoch": 0.840064620355412,
      "grad_norm": 0.2362724244594574,
      "learning_rate": 1.3683795567169122e-06,
      "loss": 0.7542,
      "step": 12350
    },
    {
      "epoch": 0.8407448346229062,
      "grad_norm": 0.252137690782547,
      "learning_rate": 1.3570432849362626e-06,
      "loss": 0.7526,
      "step": 12360
    },
    {
      "epoch": 0.8414250488904005,
      "grad_norm": 0.2419818639755249,
      "learning_rate": 1.3457507470170728e-06,
      "loss": 0.7488,
      "step": 12370
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 0.2455890029668808,
      "learning_rate": 1.3345020001002385e-06,
      "loss": 0.7524,
      "step": 12380
    },
    {
      "epoch": 0.842785477425389,
      "grad_norm": 0.2423350065946579,
      "learning_rate": 1.3232971011050711e-06,
      "loss": 0.7573,
      "step": 12390
    },
    {
      "epoch": 0.8434656916928832,
      "grad_norm": 0.24948550760746002,
      "learning_rate": 1.3121361067290072e-06,
      "loss": 0.7668,
      "step": 12400
    },
    {
      "epoch": 0.8441459059603775,
      "grad_norm": 0.24597923457622528,
      "learning_rate": 1.3010190734473248e-06,
      "loss": 0.7516,
      "step": 12410
    },
    {
      "epoch": 0.8448261202278717,
      "grad_norm": 0.25187960267066956,
      "learning_rate": 1.2899460575128586e-06,
      "loss": 0.761,
      "step": 12420
    },
    {
      "epoch": 0.845506334495366,
      "grad_norm": 0.24187469482421875,
      "learning_rate": 1.2789171149557066e-06,
      "loss": 0.7577,
      "step": 12430
    },
    {
      "epoch": 0.8461865487628603,
      "grad_norm": 0.24686579406261444,
      "learning_rate": 1.2679323015829591e-06,
      "loss": 0.7494,
      "step": 12440
    },
    {
      "epoch": 0.8468667630303546,
      "grad_norm": 0.24481099843978882,
      "learning_rate": 1.2569916729784082e-06,
      "loss": 0.7539,
      "step": 12450
    },
    {
      "epoch": 0.8475469772978488,
      "grad_norm": 0.24312631785869598,
      "learning_rate": 1.2460952845022667e-06,
      "loss": 0.7723,
      "step": 12460
    },
    {
      "epoch": 0.8482271915653431,
      "grad_norm": 0.24266554415225983,
      "learning_rate": 1.2352431912908914e-06,
      "loss": 0.7552,
      "step": 12470
    },
    {
      "epoch": 0.8489074058328373,
      "grad_norm": 0.24375444650650024,
      "learning_rate": 1.2244354482565035e-06,
      "loss": 0.7617,
      "step": 12480
    },
    {
      "epoch": 0.8495876201003316,
      "grad_norm": 0.24153859913349152,
      "learning_rate": 1.2136721100869075e-06,
      "loss": 0.7582,
      "step": 12490
    },
    {
      "epoch": 0.8502678343678258,
      "grad_norm": 0.24547694623470306,
      "learning_rate": 1.2029532312452186e-06,
      "loss": 0.7741,
      "step": 12500
    },
    {
      "epoch": 0.8509480486353201,
      "grad_norm": 0.25353655219078064,
      "learning_rate": 1.1922788659695839e-06,
      "loss": 0.758,
      "step": 12510
    },
    {
      "epoch": 0.8516282629028143,
      "grad_norm": 0.24620665609836578,
      "learning_rate": 1.1816490682729097e-06,
      "loss": 0.7627,
      "step": 12520
    },
    {
      "epoch": 0.8523084771703087,
      "grad_norm": 0.2532989978790283,
      "learning_rate": 1.1710638919425887e-06,
      "loss": 0.7636,
      "step": 12530
    },
    {
      "epoch": 0.8529886914378029,
      "grad_norm": 0.24980738759040833,
      "learning_rate": 1.1605233905402258e-06,
      "loss": 0.7671,
      "step": 12540
    },
    {
      "epoch": 0.8536689057052972,
      "grad_norm": 0.23702862858772278,
      "learning_rate": 1.150027617401368e-06,
      "loss": 0.7698,
      "step": 12550
    },
    {
      "epoch": 0.8543491199727914,
      "grad_norm": 0.24570266902446747,
      "learning_rate": 1.1395766256352348e-06,
      "loss": 0.7594,
      "step": 12560
    },
    {
      "epoch": 0.8550293342402857,
      "grad_norm": 0.24833616614341736,
      "learning_rate": 1.1291704681244498e-06,
      "loss": 0.7583,
      "step": 12570
    },
    {
      "epoch": 0.8557095485077799,
      "grad_norm": 0.25779956579208374,
      "learning_rate": 1.118809197524774e-06,
      "loss": 0.7553,
      "step": 12580
    },
    {
      "epoch": 0.8563897627752742,
      "grad_norm": 0.2411714643239975,
      "learning_rate": 1.1084928662648352e-06,
      "loss": 0.7624,
      "step": 12590
    },
    {
      "epoch": 0.8570699770427684,
      "grad_norm": 0.24922004342079163,
      "learning_rate": 1.0982215265458683e-06,
      "loss": 0.7614,
      "step": 12600
    },
    {
      "epoch": 0.8577501913102628,
      "grad_norm": 0.24543525278568268,
      "learning_rate": 1.0879952303414454e-06,
      "loss": 0.7632,
      "step": 12610
    },
    {
      "epoch": 0.858430405577757,
      "grad_norm": 0.24459443986415863,
      "learning_rate": 1.0778140293972183e-06,
      "loss": 0.7583,
      "step": 12620
    },
    {
      "epoch": 0.8591106198452513,
      "grad_norm": 0.2454974502325058,
      "learning_rate": 1.0676779752306532e-06,
      "loss": 0.7566,
      "step": 12630
    },
    {
      "epoch": 0.8597908341127455,
      "grad_norm": 0.24637340009212494,
      "learning_rate": 1.0575871191307707e-06,
      "loss": 0.7587,
      "step": 12640
    },
    {
      "epoch": 0.8604710483802398,
      "grad_norm": 0.2507196068763733,
      "learning_rate": 1.047541512157888e-06,
      "loss": 0.7496,
      "step": 12650
    },
    {
      "epoch": 0.861151262647734,
      "grad_norm": 0.24969299137592316,
      "learning_rate": 1.0375412051433575e-06,
      "loss": 0.7478,
      "step": 12660
    },
    {
      "epoch": 0.8618314769152283,
      "grad_norm": 0.24901236593723297,
      "learning_rate": 1.0275862486893129e-06,
      "loss": 0.7587,
      "step": 12670
    },
    {
      "epoch": 0.8625116911827225,
      "grad_norm": 0.2548344135284424,
      "learning_rate": 1.0176766931684078e-06,
      "loss": 0.7506,
      "step": 12680
    },
    {
      "epoch": 0.8631919054502168,
      "grad_norm": 0.24074122309684753,
      "learning_rate": 1.007812588723569e-06,
      "loss": 0.7605,
      "step": 12690
    },
    {
      "epoch": 0.863872119717711,
      "grad_norm": 0.2506800889968872,
      "learning_rate": 9.979939852677356e-07,
      "loss": 0.7688,
      "step": 12700
    },
    {
      "epoch": 0.8645523339852054,
      "grad_norm": 0.2466941773891449,
      "learning_rate": 9.88220932483609e-07,
      "loss": 0.7667,
      "step": 12710
    },
    {
      "epoch": 0.8652325482526996,
      "grad_norm": 0.2489737868309021,
      "learning_rate": 9.784934798234037e-07,
      "loss": 0.7507,
      "step": 12720
    },
    {
      "epoch": 0.8659127625201939,
      "grad_norm": 0.2487376183271408,
      "learning_rate": 9.68811676508592e-07,
      "loss": 0.7506,
      "step": 12730
    },
    {
      "epoch": 0.8665929767876881,
      "grad_norm": 0.24357005953788757,
      "learning_rate": 9.591755715296601e-07,
      "loss": 0.7502,
      "step": 12740
    },
    {
      "epoch": 0.8672731910551824,
      "grad_norm": 0.23553359508514404,
      "learning_rate": 9.495852136458572e-07,
      "loss": 0.7688,
      "step": 12750
    },
    {
      "epoch": 0.8679534053226766,
      "grad_norm": 0.25732094049453735,
      "learning_rate": 9.400406513849491e-07,
      "loss": 0.7543,
      "step": 12760
    },
    {
      "epoch": 0.8686336195901709,
      "grad_norm": 0.24797244369983673,
      "learning_rate": 9.305419330429743e-07,
      "loss": 0.772,
      "step": 12770
    },
    {
      "epoch": 0.8693138338576651,
      "grad_norm": 0.2497953623533249,
      "learning_rate": 9.210891066839966e-07,
      "loss": 0.7647,
      "step": 12780
    },
    {
      "epoch": 0.8699940481251595,
      "grad_norm": 0.2506360113620758,
      "learning_rate": 9.116822201398657e-07,
      "loss": 0.7741,
      "step": 12790
    },
    {
      "epoch": 0.8706742623926537,
      "grad_norm": 0.24232223629951477,
      "learning_rate": 9.023213210099712e-07,
      "loss": 0.7718,
      "step": 12800
    },
    {
      "epoch": 0.871354476660148,
      "grad_norm": 0.25328123569488525,
      "learning_rate": 8.930064566610052e-07,
      "loss": 0.7645,
      "step": 12810
    },
    {
      "epoch": 0.8720346909276422,
      "grad_norm": 0.2413000762462616,
      "learning_rate": 8.837376742267212e-07,
      "loss": 0.7623,
      "step": 12820
    },
    {
      "epoch": 0.8727149051951365,
      "grad_norm": 0.24539200961589813,
      "learning_rate": 8.745150206076947e-07,
      "loss": 0.7682,
      "step": 12830
    },
    {
      "epoch": 0.8733951194626307,
      "grad_norm": 0.23920084536075592,
      "learning_rate": 8.65338542471087e-07,
      "loss": 0.7726,
      "step": 12840
    },
    {
      "epoch": 0.874075333730125,
      "grad_norm": 0.262353777885437,
      "learning_rate": 8.562082862504096e-07,
      "loss": 0.7617,
      "step": 12850
    },
    {
      "epoch": 0.8747555479976192,
      "grad_norm": 0.24785573780536652,
      "learning_rate": 8.471242981452854e-07,
      "loss": 0.7559,
      "step": 12860
    },
    {
      "epoch": 0.8754357622651135,
      "grad_norm": 0.33901122212409973,
      "learning_rate": 8.380866241212226e-07,
      "loss": 0.7594,
      "step": 12870
    },
    {
      "epoch": 0.8761159765326078,
      "grad_norm": 0.24244165420532227,
      "learning_rate": 8.290953099093746e-07,
      "loss": 0.7632,
      "step": 12880
    },
    {
      "epoch": 0.8767961908001021,
      "grad_norm": 0.24455446004867554,
      "learning_rate": 8.20150401006311e-07,
      "loss": 0.7587,
      "step": 12890
    },
    {
      "epoch": 0.8774764050675963,
      "grad_norm": 0.24641729891300201,
      "learning_rate": 8.112519426737897e-07,
      "loss": 0.7516,
      "step": 12900
    },
    {
      "epoch": 0.8781566193350906,
      "grad_norm": 0.24412564933300018,
      "learning_rate": 8.023999799385252e-07,
      "loss": 0.7477,
      "step": 12910
    },
    {
      "epoch": 0.8788368336025848,
      "grad_norm": 0.24342086911201477,
      "learning_rate": 7.935945575919634e-07,
      "loss": 0.7478,
      "step": 12920
    },
    {
      "epoch": 0.8795170478700791,
      "grad_norm": 0.25551480054855347,
      "learning_rate": 7.848357201900469e-07,
      "loss": 0.7607,
      "step": 12930
    },
    {
      "epoch": 0.8801972621375733,
      "grad_norm": 0.24093754589557648,
      "learning_rate": 7.76123512053002e-07,
      "loss": 0.7677,
      "step": 12940
    },
    {
      "epoch": 0.8808774764050676,
      "grad_norm": 0.2472139298915863,
      "learning_rate": 7.67457977265107e-07,
      "loss": 0.7578,
      "step": 12950
    },
    {
      "epoch": 0.8815576906725618,
      "grad_norm": 0.23935164511203766,
      "learning_rate": 7.588391596744693e-07,
      "loss": 0.7664,
      "step": 12960
    },
    {
      "epoch": 0.8822379049400562,
      "grad_norm": 0.2518456280231476,
      "learning_rate": 7.502671028928043e-07,
      "loss": 0.7545,
      "step": 12970
    },
    {
      "epoch": 0.8829181192075504,
      "grad_norm": 0.24603936076164246,
      "learning_rate": 7.417418502952179e-07,
      "loss": 0.7616,
      "step": 12980
    },
    {
      "epoch": 0.8835983334750447,
      "grad_norm": 0.24553045630455017,
      "learning_rate": 7.332634450199804e-07,
      "loss": 0.7628,
      "step": 12990
    },
    {
      "epoch": 0.8842785477425389,
      "grad_norm": 0.2492409199476242,
      "learning_rate": 7.248319299683126e-07,
      "loss": 0.7605,
      "step": 13000
    },
    {
      "epoch": 0.8849587620100332,
      "grad_norm": 0.25094982981681824,
      "learning_rate": 7.164473478041701e-07,
      "loss": 0.7599,
      "step": 13010
    },
    {
      "epoch": 0.8856389762775274,
      "grad_norm": 0.2430187612771988,
      "learning_rate": 7.08109740954025e-07,
      "loss": 0.7631,
      "step": 13020
    },
    {
      "epoch": 0.8863191905450217,
      "grad_norm": 0.2483309805393219,
      "learning_rate": 6.998191516066488e-07,
      "loss": 0.7599,
      "step": 13030
    },
    {
      "epoch": 0.8869994048125159,
      "grad_norm": 0.24521982669830322,
      "learning_rate": 6.915756217129044e-07,
      "loss": 0.7664,
      "step": 13040
    },
    {
      "epoch": 0.8876796190800103,
      "grad_norm": 0.2494121789932251,
      "learning_rate": 6.833791929855294e-07,
      "loss": 0.7571,
      "step": 13050
    },
    {
      "epoch": 0.8883598333475045,
      "grad_norm": 0.25339776277542114,
      "learning_rate": 6.752299068989309e-07,
      "loss": 0.7596,
      "step": 13060
    },
    {
      "epoch": 0.8890400476149988,
      "grad_norm": 0.24048739671707153,
      "learning_rate": 6.67127804688965e-07,
      "loss": 0.7606,
      "step": 13070
    },
    {
      "epoch": 0.889720261882493,
      "grad_norm": 0.24579575657844543,
      "learning_rate": 6.590729273527407e-07,
      "loss": 0.7668,
      "step": 13080
    },
    {
      "epoch": 0.8904004761499873,
      "grad_norm": 0.25283282995224,
      "learning_rate": 6.510653156484026e-07,
      "loss": 0.7769,
      "step": 13090
    },
    {
      "epoch": 0.8910806904174815,
      "grad_norm": 0.2448589950799942,
      "learning_rate": 6.431050100949298e-07,
      "loss": 0.7564,
      "step": 13100
    },
    {
      "epoch": 0.8917609046849758,
      "grad_norm": 0.24270819127559662,
      "learning_rate": 6.351920509719289e-07,
      "loss": 0.7629,
      "step": 13110
    },
    {
      "epoch": 0.89244111895247,
      "grad_norm": 0.24784918129444122,
      "learning_rate": 6.273264783194311e-07,
      "loss": 0.7613,
      "step": 13120
    },
    {
      "epoch": 0.8931213332199643,
      "grad_norm": 0.2509154975414276,
      "learning_rate": 6.195083319376905e-07,
      "loss": 0.7609,
      "step": 13130
    },
    {
      "epoch": 0.8938015474874585,
      "grad_norm": 0.23716795444488525,
      "learning_rate": 6.117376513869789e-07,
      "loss": 0.7457,
      "step": 13140
    },
    {
      "epoch": 0.8944817617549529,
      "grad_norm": 0.2542048394680023,
      "learning_rate": 6.040144759873911e-07,
      "loss": 0.7659,
      "step": 13150
    },
    {
      "epoch": 0.8951619760224471,
      "grad_norm": 0.2590234577655792,
      "learning_rate": 5.963388448186424e-07,
      "loss": 0.7558,
      "step": 13160
    },
    {
      "epoch": 0.8958421902899414,
      "grad_norm": 0.23940609395503998,
      "learning_rate": 5.887107967198702e-07,
      "loss": 0.7651,
      "step": 13170
    },
    {
      "epoch": 0.8965224045574356,
      "grad_norm": 0.24657294154167175,
      "learning_rate": 5.811303702894389e-07,
      "loss": 0.7667,
      "step": 13180
    },
    {
      "epoch": 0.8972026188249298,
      "grad_norm": 0.36665135622024536,
      "learning_rate": 5.735976038847457e-07,
      "loss": 0.7738,
      "step": 13190
    },
    {
      "epoch": 0.8978828330924241,
      "grad_norm": 0.24272586405277252,
      "learning_rate": 5.661125356220254e-07,
      "loss": 0.7576,
      "step": 13200
    },
    {
      "epoch": 0.8985630473599183,
      "grad_norm": 0.23903925716876984,
      "learning_rate": 5.586752033761566e-07,
      "loss": 0.7587,
      "step": 13210
    },
    {
      "epoch": 0.8992432616274126,
      "grad_norm": 0.2517642080783844,
      "learning_rate": 5.512856447804704e-07,
      "loss": 0.768,
      "step": 13220
    },
    {
      "epoch": 0.8999234758949068,
      "grad_norm": 0.24130401015281677,
      "learning_rate": 5.43943897226562e-07,
      "loss": 0.7568,
      "step": 13230
    },
    {
      "epoch": 0.9006036901624012,
      "grad_norm": 0.24629628658294678,
      "learning_rate": 5.366499978640994e-07,
      "loss": 0.7722,
      "step": 13240
    },
    {
      "epoch": 0.9012839044298954,
      "grad_norm": 0.24640581011772156,
      "learning_rate": 5.294039836006348e-07,
      "loss": 0.7556,
      "step": 13250
    },
    {
      "epoch": 0.9019641186973897,
      "grad_norm": 0.2520070970058441,
      "learning_rate": 5.222058911014194e-07,
      "loss": 0.7649,
      "step": 13260
    },
    {
      "epoch": 0.9026443329648839,
      "grad_norm": 0.2524298131465912,
      "learning_rate": 5.150557567892201e-07,
      "loss": 0.7546,
      "step": 13270
    },
    {
      "epoch": 0.9033245472323782,
      "grad_norm": 0.24579280614852905,
      "learning_rate": 5.079536168441301e-07,
      "loss": 0.7676,
      "step": 13280
    },
    {
      "epoch": 0.9040047614998724,
      "grad_norm": 0.24815985560417175,
      "learning_rate": 5.008995072033873e-07,
      "loss": 0.7557,
      "step": 13290
    },
    {
      "epoch": 0.9046849757673667,
      "grad_norm": 0.24681419134140015,
      "learning_rate": 4.938934635611958e-07,
      "loss": 0.7564,
      "step": 13300
    },
    {
      "epoch": 0.9053651900348609,
      "grad_norm": 0.2422610968351364,
      "learning_rate": 4.869355213685423e-07,
      "loss": 0.7613,
      "step": 13310
    },
    {
      "epoch": 0.9060454043023553,
      "grad_norm": 0.2551460862159729,
      "learning_rate": 4.800257158330179e-07,
      "loss": 0.7667,
      "step": 13320
    },
    {
      "epoch": 0.9067256185698495,
      "grad_norm": 0.2425679713487625,
      "learning_rate": 4.7316408191863674e-07,
      "loss": 0.7599,
      "step": 13330
    },
    {
      "epoch": 0.9074058328373438,
      "grad_norm": 0.24537131190299988,
      "learning_rate": 4.663506543456653e-07,
      "loss": 0.7559,
      "step": 13340
    },
    {
      "epoch": 0.908086047104838,
      "grad_norm": 0.2449687421321869,
      "learning_rate": 4.5958546759044207e-07,
      "loss": 0.7633,
      "step": 13350
    },
    {
      "epoch": 0.9087662613723323,
      "grad_norm": 0.24791422486305237,
      "learning_rate": 4.5286855588520195e-07,
      "loss": 0.7508,
      "step": 13360
    },
    {
      "epoch": 0.9094464756398265,
      "grad_norm": 0.25494876503944397,
      "learning_rate": 4.4619995321790844e-07,
      "loss": 0.7538,
      "step": 13370
    },
    {
      "epoch": 0.9101266899073208,
      "grad_norm": 0.24482247233390808,
      "learning_rate": 4.395796933320773e-07,
      "loss": 0.7689,
      "step": 13380
    },
    {
      "epoch": 0.910806904174815,
      "grad_norm": 0.2454451620578766,
      "learning_rate": 4.330078097266055e-07,
      "loss": 0.7637,
      "step": 13390
    },
    {
      "epoch": 0.9114871184423093,
      "grad_norm": 0.2453595995903015,
      "learning_rate": 4.26484335655607e-07,
      "loss": 0.7475,
      "step": 13400
    },
    {
      "epoch": 0.9121673327098035,
      "grad_norm": 0.24535664916038513,
      "learning_rate": 4.2000930412823605e-07,
      "loss": 0.7622,
      "step": 13410
    },
    {
      "epoch": 0.9128475469772979,
      "grad_norm": 0.2451518476009369,
      "learning_rate": 4.135827479085286e-07,
      "loss": 0.7575,
      "step": 13420
    },
    {
      "epoch": 0.9135277612447921,
      "grad_norm": 0.24552904069423676,
      "learning_rate": 4.072046995152301e-07,
      "loss": 0.7614,
      "step": 13430
    },
    {
      "epoch": 0.9142079755122864,
      "grad_norm": 0.2504306137561798,
      "learning_rate": 4.0087519122163575e-07,
      "loss": 0.7638,
      "step": 13440
    },
    {
      "epoch": 0.9148881897797806,
      "grad_norm": 0.24239565432071686,
      "learning_rate": 3.945942550554238e-07,
      "loss": 0.7635,
      "step": 13450
    },
    {
      "epoch": 0.9155684040472749,
      "grad_norm": 0.243515282869339,
      "learning_rate": 3.8836192279849473e-07,
      "loss": 0.7525,
      "step": 13460
    },
    {
      "epoch": 0.9162486183147691,
      "grad_norm": 0.24457940459251404,
      "learning_rate": 3.821782259868101e-07,
      "loss": 0.7524,
      "step": 13470
    },
    {
      "epoch": 0.9169288325822634,
      "grad_norm": 0.24360902607440948,
      "learning_rate": 3.7604319591023507e-07,
      "loss": 0.7686,
      "step": 13480
    },
    {
      "epoch": 0.9176090468497576,
      "grad_norm": 0.24603000283241272,
      "learning_rate": 3.6995686361237625e-07,
      "loss": 0.7613,
      "step": 13490
    },
    {
      "epoch": 0.918289261117252,
      "grad_norm": 0.2504574954509735,
      "learning_rate": 3.639192598904284e-07,
      "loss": 0.7537,
      "step": 13500
    },
    {
      "epoch": 0.9189694753847462,
      "grad_norm": 0.24181614816188812,
      "learning_rate": 3.579304152950147e-07,
      "loss": 0.7624,
      "step": 13510
    },
    {
      "epoch": 0.9196496896522405,
      "grad_norm": 0.24688316881656647,
      "learning_rate": 3.519903601300367e-07,
      "loss": 0.7634,
      "step": 13520
    },
    {
      "epoch": 0.9203299039197347,
      "grad_norm": 0.2513571083545685,
      "learning_rate": 3.4609912445251693e-07,
      "loss": 0.744,
      "step": 13530
    },
    {
      "epoch": 0.921010118187229,
      "grad_norm": 0.2379947006702423,
      "learning_rate": 3.402567380724486e-07,
      "loss": 0.7657,
      "step": 13540
    },
    {
      "epoch": 0.9216903324547232,
      "grad_norm": 0.24739056825637817,
      "learning_rate": 3.344632305526463e-07,
      "loss": 0.7682,
      "step": 13550
    },
    {
      "epoch": 0.9223705467222175,
      "grad_norm": 0.24008318781852722,
      "learning_rate": 3.2871863120859214e-07,
      "loss": 0.7713,
      "step": 13560
    },
    {
      "epoch": 0.9230507609897117,
      "grad_norm": 0.24716557562351227,
      "learning_rate": 3.230229691082931e-07,
      "loss": 0.7652,
      "step": 13570
    },
    {
      "epoch": 0.923730975257206,
      "grad_norm": 0.24023090302944183,
      "learning_rate": 3.173762730721275e-07,
      "loss": 0.7695,
      "step": 13580
    },
    {
      "epoch": 0.9244111895247002,
      "grad_norm": 0.24779115617275238,
      "learning_rate": 3.117785716727062e-07,
      "loss": 0.7586,
      "step": 13590
    },
    {
      "epoch": 0.9250914037921946,
      "grad_norm": 0.24192939698696136,
      "learning_rate": 3.0622989323472076e-07,
      "loss": 0.765,
      "step": 13600
    },
    {
      "epoch": 0.9257716180596888,
      "grad_norm": 0.25075802206993103,
      "learning_rate": 3.0073026583480547e-07,
      "loss": 0.764,
      "step": 13610
    },
    {
      "epoch": 0.9264518323271831,
      "grad_norm": 0.24366381764411926,
      "learning_rate": 2.952797173013944e-07,
      "loss": 0.765,
      "step": 13620
    },
    {
      "epoch": 0.9271320465946773,
      "grad_norm": 0.2407902032136917,
      "learning_rate": 2.898782752145779e-07,
      "loss": 0.7531,
      "step": 13630
    },
    {
      "epoch": 0.9278122608621716,
      "grad_norm": 0.23994383215904236,
      "learning_rate": 2.845259669059663e-07,
      "loss": 0.754,
      "step": 13640
    },
    {
      "epoch": 0.9284924751296658,
      "grad_norm": 0.24762782454490662,
      "learning_rate": 2.7922281945855e-07,
      "loss": 0.7682,
      "step": 13650
    },
    {
      "epoch": 0.9291726893971601,
      "grad_norm": 0.24440613389015198,
      "learning_rate": 2.739688597065604e-07,
      "loss": 0.7613,
      "step": 13660
    },
    {
      "epoch": 0.9298529036646543,
      "grad_norm": 0.24806998670101166,
      "learning_rate": 2.6876411423534035e-07,
      "loss": 0.7621,
      "step": 13670
    },
    {
      "epoch": 0.9305331179321487,
      "grad_norm": 0.24591968953609467,
      "learning_rate": 2.6360860938120183e-07,
      "loss": 0.7594,
      "step": 13680
    },
    {
      "epoch": 0.9312133321996429,
      "grad_norm": 0.2511826157569885,
      "learning_rate": 2.585023712312973e-07,
      "loss": 0.7696,
      "step": 13690
    },
    {
      "epoch": 0.9318935464671372,
      "grad_norm": 0.2496713250875473,
      "learning_rate": 2.5344542562348974e-07,
      "loss": 0.7709,
      "step": 13700
    },
    {
      "epoch": 0.9325737607346314,
      "grad_norm": 0.2496761530637741,
      "learning_rate": 2.4843779814621496e-07,
      "loss": 0.7508,
      "step": 13710
    },
    {
      "epoch": 0.9332539750021257,
      "grad_norm": 0.24858562648296356,
      "learning_rate": 2.434795141383584e-07,
      "loss": 0.7638,
      "step": 13720
    },
    {
      "epoch": 0.9339341892696199,
      "grad_norm": 0.2461564838886261,
      "learning_rate": 2.3857059868912403e-07,
      "loss": 0.7679,
      "step": 13730
    },
    {
      "epoch": 0.9346144035371142,
      "grad_norm": 0.25012731552124023,
      "learning_rate": 2.3371107663790915e-07,
      "loss": 0.7547,
      "step": 13740
    },
    {
      "epoch": 0.9352946178046084,
      "grad_norm": 0.24220655858516693,
      "learning_rate": 2.2890097257417642e-07,
      "loss": 0.7612,
      "step": 13750
    },
    {
      "epoch": 0.9359748320721027,
      "grad_norm": 0.24822300672531128,
      "learning_rate": 2.241403108373297e-07,
      "loss": 0.7692,
      "step": 13760
    },
    {
      "epoch": 0.936655046339597,
      "grad_norm": 0.24931180477142334,
      "learning_rate": 2.1942911551659506e-07,
      "loss": 0.7642,
      "step": 13770
    },
    {
      "epoch": 0.9373352606070913,
      "grad_norm": 0.23831304907798767,
      "learning_rate": 2.1476741045089233e-07,
      "loss": 0.7628,
      "step": 13780
    },
    {
      "epoch": 0.9380154748745855,
      "grad_norm": 0.24508658051490784,
      "learning_rate": 2.101552192287193e-07,
      "loss": 0.7631,
      "step": 13790
    },
    {
      "epoch": 0.9386956891420798,
      "grad_norm": 0.25515463948249817,
      "learning_rate": 2.0559256518803082e-07,
      "loss": 0.7509,
      "step": 13800
    },
    {
      "epoch": 0.939375903409574,
      "grad_norm": 0.23935982584953308,
      "learning_rate": 2.0107947141612017e-07,
      "loss": 0.7531,
      "step": 13810
    },
    {
      "epoch": 0.9400561176770683,
      "grad_norm": 0.24096937477588654,
      "learning_rate": 1.9661596074950328e-07,
      "loss": 0.7443,
      "step": 13820
    },
    {
      "epoch": 0.9407363319445625,
      "grad_norm": 0.2419290691614151,
      "learning_rate": 1.9220205577380356e-07,
      "loss": 0.7635,
      "step": 13830
    },
    {
      "epoch": 0.9414165462120568,
      "grad_norm": 0.2487664669752121,
      "learning_rate": 1.8783777882363407e-07,
      "loss": 0.7668,
      "step": 13840
    },
    {
      "epoch": 0.942096760479551,
      "grad_norm": 0.24574416875839233,
      "learning_rate": 1.8352315198248983e-07,
      "loss": 0.7743,
      "step": 13850
    },
    {
      "epoch": 0.9427769747470454,
      "grad_norm": 0.2422151267528534,
      "learning_rate": 1.7925819708263347e-07,
      "loss": 0.7618,
      "step": 13860
    },
    {
      "epoch": 0.9434571890145396,
      "grad_norm": 0.2462805211544037,
      "learning_rate": 1.750429357049821e-07,
      "loss": 0.7595,
      "step": 13870
    },
    {
      "epoch": 0.9441374032820339,
      "grad_norm": 0.24129988253116608,
      "learning_rate": 1.7087738917900277e-07,
      "loss": 0.7594,
      "step": 13880
    },
    {
      "epoch": 0.9448176175495281,
      "grad_norm": 0.24333396553993225,
      "learning_rate": 1.6676157858260157e-07,
      "loss": 0.7543,
      "step": 13890
    },
    {
      "epoch": 0.9454978318170224,
      "grad_norm": 0.2515409290790558,
      "learning_rate": 1.6269552474202033e-07,
      "loss": 0.7598,
      "step": 13900
    },
    {
      "epoch": 0.9461780460845166,
      "grad_norm": 0.24224993586540222,
      "learning_rate": 1.5867924823172565e-07,
      "loss": 0.7595,
      "step": 13910
    },
    {
      "epoch": 0.9468582603520109,
      "grad_norm": 0.24030137062072754,
      "learning_rate": 1.5471276937431e-07,
      "loss": 0.7549,
      "step": 13920
    },
    {
      "epoch": 0.9475384746195051,
      "grad_norm": 0.2618497312068939,
      "learning_rate": 1.5079610824038416e-07,
      "loss": 0.7499,
      "step": 13930
    },
    {
      "epoch": 0.9482186888869995,
      "grad_norm": 0.25244298577308655,
      "learning_rate": 1.4692928464848266e-07,
      "loss": 0.7574,
      "step": 13940
    },
    {
      "epoch": 0.9488989031544937,
      "grad_norm": 0.2502358555793762,
      "learning_rate": 1.431123181649552e-07,
      "loss": 0.7658,
      "step": 13950
    },
    {
      "epoch": 0.949579117421988,
      "grad_norm": 0.24570995569229126,
      "learning_rate": 1.3934522810387429e-07,
      "loss": 0.7608,
      "step": 13960
    },
    {
      "epoch": 0.9502593316894822,
      "grad_norm": 0.24009110033512115,
      "learning_rate": 1.356280335269333e-07,
      "loss": 0.754,
      "step": 13970
    },
    {
      "epoch": 0.9509395459569765,
      "grad_norm": 0.2586895823478699,
      "learning_rate": 1.319607532433509e-07,
      "loss": 0.7631,
      "step": 13980
    },
    {
      "epoch": 0.9516197602244707,
      "grad_norm": 0.26034995913505554,
      "learning_rate": 1.283434058097799e-07,
      "loss": 0.7558,
      "step": 13990
    },
    {
      "epoch": 0.952299974491965,
      "grad_norm": 0.24782530963420868,
      "learning_rate": 1.247760095302064e-07,
      "loss": 0.7628,
      "step": 14000
    },
    {
      "epoch": 0.9529801887594592,
      "grad_norm": 0.24033108353614807,
      "learning_rate": 1.2125858245586207e-07,
      "loss": 0.7767,
      "step": 14010
    },
    {
      "epoch": 0.9536604030269535,
      "grad_norm": 0.24754656851291656,
      "learning_rate": 1.1779114238513189e-07,
      "loss": 0.7616,
      "step": 14020
    },
    {
      "epoch": 0.9543406172944477,
      "grad_norm": 0.2364259511232376,
      "learning_rate": 1.1437370686346427e-07,
      "loss": 0.7622,
      "step": 14030
    },
    {
      "epoch": 0.9550208315619421,
      "grad_norm": 0.24411220848560333,
      "learning_rate": 1.1100629318327894e-07,
      "loss": 0.7598,
      "step": 14040
    },
    {
      "epoch": 0.9557010458294363,
      "grad_norm": 0.24533218145370483,
      "learning_rate": 1.0768891838388584e-07,
      "loss": 0.7563,
      "step": 14050
    },
    {
      "epoch": 0.9563812600969306,
      "grad_norm": 0.24469947814941406,
      "learning_rate": 1.0442159925139306e-07,
      "loss": 0.7602,
      "step": 14060
    },
    {
      "epoch": 0.9570614743644248,
      "grad_norm": 0.23895448446273804,
      "learning_rate": 1.0120435231862346e-07,
      "loss": 0.7565,
      "step": 14070
    },
    {
      "epoch": 0.9577416886319191,
      "grad_norm": 0.24606747925281525,
      "learning_rate": 9.803719386503485e-08,
      "loss": 0.7534,
      "step": 14080
    },
    {
      "epoch": 0.9584219028994133,
      "grad_norm": 0.2557947635650635,
      "learning_rate": 9.492013991663107e-08,
      "loss": 0.7647,
      "step": 14090
    },
    {
      "epoch": 0.9591021171669076,
      "grad_norm": 0.24558596312999725,
      "learning_rate": 9.185320624588545e-08,
      "loss": 0.7526,
      "step": 14100
    },
    {
      "epoch": 0.9597823314344018,
      "grad_norm": 0.24695099890232086,
      "learning_rate": 8.883640837166085e-08,
      "loss": 0.7526,
      "step": 14110
    },
    {
      "epoch": 0.960462545701896,
      "grad_norm": 0.25338950753211975,
      "learning_rate": 8.586976155912863e-08,
      "loss": 0.748,
      "step": 14120
    },
    {
      "epoch": 0.9611427599693904,
      "grad_norm": 0.2424427717924118,
      "learning_rate": 8.295328081969312e-08,
      "loss": 0.7648,
      "step": 14130
    },
    {
      "epoch": 0.9618229742368846,
      "grad_norm": 0.24449896812438965,
      "learning_rate": 8.008698091091727e-08,
      "loss": 0.7526,
      "step": 14140
    },
    {
      "epoch": 0.9625031885043789,
      "grad_norm": 0.24283750355243683,
      "learning_rate": 7.72708763364427e-08,
      "loss": 0.7646,
      "step": 14150
    },
    {
      "epoch": 0.9631834027718731,
      "grad_norm": 0.2458605021238327,
      "learning_rate": 7.450498134592198e-08,
      "loss": 0.7537,
      "step": 14160
    },
    {
      "epoch": 0.9638636170393674,
      "grad_norm": 0.2411893755197525,
      "learning_rate": 7.178930993494538e-08,
      "loss": 0.7512,
      "step": 14170
    },
    {
      "epoch": 0.9645438313068616,
      "grad_norm": 0.24779365956783295,
      "learning_rate": 6.912387584496527e-08,
      "loss": 0.7537,
      "step": 14180
    },
    {
      "epoch": 0.9652240455743559,
      "grad_norm": 0.25667867064476013,
      "learning_rate": 6.650869256323522e-08,
      "loss": 0.7616,
      "step": 14190
    },
    {
      "epoch": 0.9659042598418501,
      "grad_norm": 0.2500806748867035,
      "learning_rate": 6.394377332273216e-08,
      "loss": 0.771,
      "step": 14200
    },
    {
      "epoch": 0.9665844741093444,
      "grad_norm": 0.2422896921634674,
      "learning_rate": 6.142913110209537e-08,
      "loss": 0.7524,
      "step": 14210
    },
    {
      "epoch": 0.9672646883768387,
      "grad_norm": 0.2571762800216675,
      "learning_rate": 5.896477862556205e-08,
      "loss": 0.7581,
      "step": 14220
    },
    {
      "epoch": 0.967944902644333,
      "grad_norm": 0.2567151188850403,
      "learning_rate": 5.6550728362896325e-08,
      "loss": 0.767,
      "step": 14230
    },
    {
      "epoch": 0.9686251169118272,
      "grad_norm": 0.24308769404888153,
      "learning_rate": 5.418699252933479e-08,
      "loss": 0.757,
      "step": 14240
    },
    {
      "epoch": 0.9693053311793215,
      "grad_norm": 0.2470419555902481,
      "learning_rate": 5.18735830855166e-08,
      "loss": 0.763,
      "step": 14250
    },
    {
      "epoch": 0.9699855454468157,
      "grad_norm": 0.2530462443828583,
      "learning_rate": 4.9610511737425706e-08,
      "loss": 0.7682,
      "step": 14260
    },
    {
      "epoch": 0.97066575971431,
      "grad_norm": 0.2385081797838211,
      "learning_rate": 4.739778993633648e-08,
      "loss": 0.7407,
      "step": 14270
    },
    {
      "epoch": 0.9713459739818042,
      "grad_norm": 0.23911146819591522,
      "learning_rate": 4.523542887874932e-08,
      "loss": 0.7549,
      "step": 14280
    },
    {
      "epoch": 0.9720261882492985,
      "grad_norm": 0.24873803555965424,
      "learning_rate": 4.312343950633513e-08,
      "loss": 0.7566,
      "step": 14290
    },
    {
      "epoch": 0.9727064025167927,
      "grad_norm": 0.2478393018245697,
      "learning_rate": 4.1061832505883135e-08,
      "loss": 0.7756,
      "step": 14300
    },
    {
      "epoch": 0.9733866167842871,
      "grad_norm": 0.23787696659564972,
      "learning_rate": 3.9050618309243173e-08,
      "loss": 0.7621,
      "step": 14310
    },
    {
      "epoch": 0.9740668310517813,
      "grad_norm": 0.24838638305664062,
      "learning_rate": 3.708980709327681e-08,
      "loss": 0.7502,
      "step": 14320
    },
    {
      "epoch": 0.9747470453192756,
      "grad_norm": 0.24176611006259918,
      "learning_rate": 3.517940877980186e-08,
      "loss": 0.7471,
      "step": 14330
    },
    {
      "epoch": 0.9754272595867698,
      "grad_norm": 0.24439072608947754,
      "learning_rate": 3.331943303554464e-08,
      "loss": 0.7585,
      "step": 14340
    },
    {
      "epoch": 0.9761074738542641,
      "grad_norm": 0.24630014598369598,
      "learning_rate": 3.1509889272089975e-08,
      "loss": 0.7579,
      "step": 14350
    },
    {
      "epoch": 0.9767876881217583,
      "grad_norm": 0.24432967603206635,
      "learning_rate": 2.975078664583353e-08,
      "loss": 0.7632,
      "step": 14360
    },
    {
      "epoch": 0.9774679023892526,
      "grad_norm": 0.2476637065410614,
      "learning_rate": 2.8042134057937322e-08,
      "loss": 0.7748,
      "step": 14370
    },
    {
      "epoch": 0.9781481166567468,
      "grad_norm": 0.2439027577638626,
      "learning_rate": 2.638394015428203e-08,
      "loss": 0.7567,
      "step": 14380
    },
    {
      "epoch": 0.9788283309242412,
      "grad_norm": 0.24583202600479126,
      "learning_rate": 2.477621332542368e-08,
      "loss": 0.7638,
      "step": 14390
    },
    {
      "epoch": 0.9795085451917354,
      "grad_norm": 0.24024732410907745,
      "learning_rate": 2.321896170655591e-08,
      "loss": 0.7626,
      "step": 14400
    },
    {
      "epoch": 0.9801887594592297,
      "grad_norm": 0.4252866804599762,
      "learning_rate": 2.1712193177459985e-08,
      "loss": 0.7514,
      "step": 14410
    },
    {
      "epoch": 0.9808689737267239,
      "grad_norm": 0.24549369513988495,
      "learning_rate": 2.0255915362471513e-08,
      "loss": 0.7541,
      "step": 14420
    },
    {
      "epoch": 0.9815491879942182,
      "grad_norm": 0.24518539011478424,
      "learning_rate": 1.885013563044158e-08,
      "loss": 0.7499,
      "step": 14430
    },
    {
      "epoch": 0.9822294022617124,
      "grad_norm": 0.2515625059604645,
      "learning_rate": 1.749486109469789e-08,
      "loss": 0.7779,
      "step": 14440
    },
    {
      "epoch": 0.9829096165292067,
      "grad_norm": 0.23486600816249847,
      "learning_rate": 1.6190098613005913e-08,
      "loss": 0.7476,
      "step": 14450
    },
    {
      "epoch": 0.9835898307967009,
      "grad_norm": 0.24871693551540375,
      "learning_rate": 1.4935854787541114e-08,
      "loss": 0.7589,
      "step": 14460
    },
    {
      "epoch": 0.9842700450641952,
      "grad_norm": 0.24605490267276764,
      "learning_rate": 1.3732135964849013e-08,
      "loss": 0.7721,
      "step": 14470
    },
    {
      "epoch": 0.9849502593316894,
      "grad_norm": 0.24189579486846924,
      "learning_rate": 1.2578948235817402e-08,
      "loss": 0.7671,
      "step": 14480
    },
    {
      "epoch": 0.9856304735991838,
      "grad_norm": 0.24505004286766052,
      "learning_rate": 1.1476297435641937e-08,
      "loss": 0.7634,
      "step": 14490
    },
    {
      "epoch": 0.986310687866678,
      "grad_norm": 0.24500903487205505,
      "learning_rate": 1.0424189143799502e-08,
      "loss": 0.7758,
      "step": 14500
    },
    {
      "epoch": 0.9869909021341723,
      "grad_norm": 0.24318064749240875,
      "learning_rate": 9.42262868401933e-09,
      "loss": 0.743,
      "step": 14510
    },
    {
      "epoch": 0.9876711164016665,
      "grad_norm": 0.24455608427524567,
      "learning_rate": 8.471621124256368e-09,
      "loss": 0.7589,
      "step": 14520
    },
    {
      "epoch": 0.9883513306691608,
      "grad_norm": 0.24279429018497467,
      "learning_rate": 7.571171276662403e-09,
      "loss": 0.7716,
      "step": 14530
    },
    {
      "epoch": 0.989031544936655,
      "grad_norm": 0.24775396287441254,
      "learning_rate": 6.721283697566084e-09,
      "loss": 0.7686,
      "step": 14540
    },
    {
      "epoch": 0.9897117592041493,
      "grad_norm": 0.25869980454444885,
      "learning_rate": 5.92196268744849e-09,
      "loss": 0.7597,
      "step": 14550
    },
    {
      "epoch": 0.9903919734716435,
      "grad_norm": 0.2435072809457779,
      "learning_rate": 5.173212290919827e-09,
      "loss": 0.749,
      "step": 14560
    },
    {
      "epoch": 0.9910721877391379,
      "grad_norm": 0.24581976234912872,
      "learning_rate": 4.4750362967005415e-09,
      "loss": 0.7495,
      "step": 14570
    },
    {
      "epoch": 0.9917524020066321,
      "grad_norm": 0.241951122879982,
      "learning_rate": 3.827438237601344e-09,
      "loss": 0.757,
      "step": 14580
    },
    {
      "epoch": 0.9924326162741264,
      "grad_norm": 0.24407359957695007,
      "learning_rate": 3.230421390506555e-09,
      "loss": 0.7578,
      "step": 14590
    },
    {
      "epoch": 0.9931128305416206,
      "grad_norm": 0.24469362199306488,
      "learning_rate": 2.6839887763552287e-09,
      "loss": 0.7655,
      "step": 14600
    },
    {
      "epoch": 0.9937930448091149,
      "grad_norm": 0.23985865712165833,
      "learning_rate": 2.1881431601278315e-09,
      "loss": 0.7755,
      "step": 14610
    },
    {
      "epoch": 0.9944732590766091,
      "grad_norm": 0.2478683441877365,
      "learning_rate": 1.7428870508329199e-09,
      "loss": 0.7662,
      "step": 14620
    },
    {
      "epoch": 0.9951534733441034,
      "grad_norm": 0.24447156488895416,
      "learning_rate": 1.3482227014915973e-09,
      "loss": 0.7629,
      "step": 14630
    },
    {
      "epoch": 0.9958336876115976,
      "grad_norm": 0.24857555329799652,
      "learning_rate": 1.0041521091275209e-09,
      "loss": 0.7588,
      "step": 14640
    },
    {
      "epoch": 0.996513901879092,
      "grad_norm": 0.24451036751270294,
      "learning_rate": 7.106770147602415e-10,
      "loss": 0.7621,
      "step": 14650
    },
    {
      "epoch": 0.9971941161465862,
      "grad_norm": 0.2504979372024536,
      "learning_rate": 4.677989033874397e-10,
      "loss": 0.7528,
      "step": 14660
    },
    {
      "epoch": 0.9978743304140805,
      "grad_norm": 0.24821937084197998,
      "learning_rate": 2.7551900398936625e-10,
      "loss": 0.7599,
      "step": 14670
    },
    {
      "epoch": 0.9985545446815747,
      "grad_norm": 0.2521255612373352,
      "learning_rate": 1.3383828951218924e-10,
      "loss": 0.7507,
      "step": 14680
    },
    {
      "epoch": 0.999234758949069,
      "grad_norm": 0.2424987554550171,
      "learning_rate": 4.275747686799392e-11,
      "loss": 0.7597,
      "step": 14690
    },
    {
      "epoch": 0.9999149732165632,
      "grad_norm": 0.2516481280326843,
      "learning_rate": 2.277026931452042e-12,
      "loss": 0.7617,
      "step": 14700
    }
  ],
  "logging_steps": 10,
  "max_steps": 14702,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.302340990111685e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}