{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.25250227479526843,
  "eval_steps": 500,
  "global_step": 555,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00045495905368516835,
      "grad_norm": 9.461428161462043,
      "learning_rate": 1e-05,
      "loss": 0.1263,
      "step": 1
    },
    {
      "epoch": 0.0009099181073703367,
      "grad_norm": 5.190780450250769,
      "learning_rate": 9.99999979571129e-06,
      "loss": 0.1723,
      "step": 2
    },
    {
      "epoch": 0.001364877161055505,
      "grad_norm": 7.521926017130347,
      "learning_rate": 9.999999182845177e-06,
      "loss": 0.1327,
      "step": 3
    },
    {
      "epoch": 0.0018198362147406734,
      "grad_norm": 2.5665810200307217,
      "learning_rate": 9.99999816140171e-06,
      "loss": 0.1095,
      "step": 4
    },
    {
      "epoch": 0.0022747952684258415,
      "grad_norm": 2.738508706395883,
      "learning_rate": 9.999996731380973e-06,
      "loss": 0.1151,
      "step": 5
    },
    {
      "epoch": 0.00272975432211101,
      "grad_norm": 2.67941899677245,
      "learning_rate": 9.999994892783083e-06,
      "loss": 0.0821,
      "step": 6
    },
    {
      "epoch": 0.0031847133757961785,
      "grad_norm": 2.137586234420784,
      "learning_rate": 9.99999264560819e-06,
      "loss": 0.0729,
      "step": 7
    },
    {
      "epoch": 0.003639672429481347,
      "grad_norm": 2.8221590420989164,
      "learning_rate": 9.999989989856477e-06,
      "loss": 0.0929,
      "step": 8
    },
    {
      "epoch": 0.004094631483166515,
      "grad_norm": 1.6167314639784554,
      "learning_rate": 9.999986925528164e-06,
      "loss": 0.0466,
      "step": 9
    },
    {
      "epoch": 0.004549590536851683,
      "grad_norm": 2.1773262431631313,
      "learning_rate": 9.999983452623498e-06,
      "loss": 0.0709,
      "step": 10
    },
    {
      "epoch": 0.005004549590536852,
      "grad_norm": 7.6444390817806465,
      "learning_rate": 9.999979571142765e-06,
      "loss": 0.0809,
      "step": 11
    },
    {
      "epoch": 0.00545950864422202,
      "grad_norm": 2.034523884241798,
      "learning_rate": 9.999975281086278e-06,
      "loss": 0.0839,
      "step": 12
    },
    {
      "epoch": 0.005914467697907188,
      "grad_norm": 3.576108282005355,
      "learning_rate": 9.999970582454392e-06,
      "loss": 0.0728,
      "step": 13
    },
    {
      "epoch": 0.006369426751592357,
      "grad_norm": 2.623641566468802,
      "learning_rate": 9.999965475247491e-06,
      "loss": 0.1052,
      "step": 14
    },
    {
      "epoch": 0.006824385805277525,
      "grad_norm": 2.1413574998269085,
      "learning_rate": 9.99995995946599e-06,
      "loss": 0.0885,
      "step": 15
    },
    {
      "epoch": 0.007279344858962694,
      "grad_norm": 1.4859066724415246,
      "learning_rate": 9.999954035110342e-06,
      "loss": 0.0644,
      "step": 16
    },
    {
      "epoch": 0.0077343039126478615,
      "grad_norm": 2.851793157608408,
      "learning_rate": 9.999947702181027e-06,
      "loss": 0.1057,
      "step": 17
    },
    {
      "epoch": 0.00818926296633303,
      "grad_norm": 4.693829546662477,
      "learning_rate": 9.999940960678568e-06,
      "loss": 0.0867,
      "step": 18
    },
    {
      "epoch": 0.008644222020018199,
      "grad_norm": 2.2728033563417362,
      "learning_rate": 9.999933810603513e-06,
      "loss": 0.0789,
      "step": 19
    },
    {
      "epoch": 0.009099181073703366,
      "grad_norm": 1.6705986173507794,
      "learning_rate": 9.999926251956447e-06,
      "loss": 0.0683,
      "step": 20
    },
    {
      "epoch": 0.009554140127388535,
      "grad_norm": 2.187579869114393,
      "learning_rate": 9.999918284737986e-06,
      "loss": 0.0984,
      "step": 21
    },
    {
      "epoch": 0.010009099181073703,
      "grad_norm": 2.328040268012338,
      "learning_rate": 9.999909908948782e-06,
      "loss": 0.0699,
      "step": 22
    },
    {
      "epoch": 0.010464058234758872,
      "grad_norm": 5.572389775693198,
      "learning_rate": 9.999901124589519e-06,
      "loss": 0.0912,
      "step": 23
    },
    {
      "epoch": 0.01091901728844404,
      "grad_norm": 1.84796719674859,
      "learning_rate": 9.999891931660916e-06,
      "loss": 0.1015,
      "step": 24
    },
    {
      "epoch": 0.011373976342129208,
      "grad_norm": 1.7501762990792236,
      "learning_rate": 9.999882330163725e-06,
      "loss": 0.0909,
      "step": 25
    },
    {
      "epoch": 0.011828935395814377,
      "grad_norm": 0.9922115950592263,
      "learning_rate": 9.999872320098729e-06,
      "loss": 0.0656,
      "step": 26
    },
    {
      "epoch": 0.012283894449499545,
      "grad_norm": 1.5612370560987539,
      "learning_rate": 9.999861901466746e-06,
      "loss": 0.0974,
      "step": 27
    },
    {
      "epoch": 0.012738853503184714,
      "grad_norm": 1.4617271794930395,
      "learning_rate": 9.999851074268625e-06,
      "loss": 0.0853,
      "step": 28
    },
    {
      "epoch": 0.013193812556869881,
      "grad_norm": 1.8127085104491556,
      "learning_rate": 9.999839838505257e-06,
      "loss": 0.1081,
      "step": 29
    },
    {
      "epoch": 0.01364877161055505,
      "grad_norm": 1.4710105512612208,
      "learning_rate": 9.999828194177555e-06,
      "loss": 0.0868,
      "step": 30
    },
    {
      "epoch": 0.014103730664240218,
      "grad_norm": 1.3474487189311888,
      "learning_rate": 9.999816141286472e-06,
      "loss": 0.0817,
      "step": 31
    },
    {
      "epoch": 0.014558689717925387,
      "grad_norm": 1.0967596652549403,
      "learning_rate": 9.99980367983299e-06,
      "loss": 0.0637,
      "step": 32
    },
    {
      "epoch": 0.015013648771610554,
      "grad_norm": 3.179425671823194,
      "learning_rate": 9.999790809818134e-06,
      "loss": 0.069,
      "step": 33
    },
    {
      "epoch": 0.015468607825295723,
      "grad_norm": 4.482257681577152,
      "learning_rate": 9.999777531242951e-06,
      "loss": 0.0915,
      "step": 34
    },
    {
      "epoch": 0.01592356687898089,
      "grad_norm": 3.953299040475791,
      "learning_rate": 9.999763844108528e-06,
      "loss": 0.0562,
      "step": 35
    },
    {
      "epoch": 0.01637852593266606,
      "grad_norm": 1.1127201050382067,
      "learning_rate": 9.999749748415982e-06,
      "loss": 0.0556,
      "step": 36
    },
    {
      "epoch": 0.01683348498635123,
      "grad_norm": 79.45756094624792,
      "learning_rate": 9.999735244166464e-06,
      "loss": 0.1223,
      "step": 37
    },
    {
      "epoch": 0.017288444040036398,
      "grad_norm": 2777.9092912017113,
      "learning_rate": 9.99972033136116e-06,
      "loss": 0.3211,
      "step": 38
    },
    {
      "epoch": 0.017743403093721567,
      "grad_norm": 2.5204693177238466,
      "learning_rate": 9.999705010001291e-06,
      "loss": 0.0723,
      "step": 39
    },
    {
      "epoch": 0.018198362147406732,
      "grad_norm": 2.2975907071135655,
      "learning_rate": 9.999689280088105e-06,
      "loss": 0.0696,
      "step": 40
    },
    {
      "epoch": 0.0186533212010919,
      "grad_norm": 2.998434349074003,
      "learning_rate": 9.99967314162289e-06,
      "loss": 0.083,
      "step": 41
    },
    {
      "epoch": 0.01910828025477707,
      "grad_norm": 3.882239448575704,
      "learning_rate": 9.999656594606966e-06,
      "loss": 0.1015,
      "step": 42
    },
    {
      "epoch": 0.019563239308462238,
      "grad_norm": 3.5286596480512493,
      "learning_rate": 9.999639639041681e-06,
      "loss": 0.0817,
      "step": 43
    },
    {
      "epoch": 0.020018198362147407,
      "grad_norm": 1.6933989447443707,
      "learning_rate": 9.999622274928424e-06,
      "loss": 0.1003,
      "step": 44
    },
    {
      "epoch": 0.020473157415832575,
      "grad_norm": 1.2483160046323276,
      "learning_rate": 9.999604502268614e-06,
      "loss": 0.0952,
      "step": 45
    },
    {
      "epoch": 0.020928116469517744,
      "grad_norm": 0.9417906124383243,
      "learning_rate": 9.9995863210637e-06,
      "loss": 0.0731,
      "step": 46
    },
    {
      "epoch": 0.021383075523202913,
      "grad_norm": 2.8195414757816897,
      "learning_rate": 9.99956773131517e-06,
      "loss": 0.1845,
      "step": 47
    },
    {
      "epoch": 0.02183803457688808,
      "grad_norm": 2.74390379471345,
      "learning_rate": 9.999548733024545e-06,
      "loss": 0.1826,
      "step": 48
    },
    {
      "epoch": 0.022292993630573247,
      "grad_norm": 1.5138494619527987,
      "learning_rate": 9.999529326193373e-06,
      "loss": 0.0857,
      "step": 49
    },
    {
      "epoch": 0.022747952684258416,
      "grad_norm": 1.215379974181271,
      "learning_rate": 9.999509510823242e-06,
      "loss": 0.0686,
      "step": 50
    },
    {
      "epoch": 0.023202911737943584,
      "grad_norm": 1.292187967807859,
      "learning_rate": 9.999489286915773e-06,
      "loss": 0.0707,
      "step": 51
    },
    {
      "epoch": 0.023657870791628753,
      "grad_norm": 1.7888013203563982,
      "learning_rate": 9.999468654472614e-06,
      "loss": 0.0682,
      "step": 52
    },
    {
      "epoch": 0.024112829845313922,
      "grad_norm": 0.8979425621703144,
      "learning_rate": 9.999447613495457e-06,
      "loss": 0.0508,
      "step": 53
    },
    {
      "epoch": 0.02456778889899909,
      "grad_norm": 1.9123835444775663,
      "learning_rate": 9.99942616398602e-06,
      "loss": 0.0689,
      "step": 54
    },
    {
      "epoch": 0.02502274795268426,
      "grad_norm": 0.9393581994096443,
      "learning_rate": 9.99940430594605e-06,
      "loss": 0.0496,
      "step": 55
    },
    {
      "epoch": 0.025477707006369428,
      "grad_norm": 1.0234476513644222,
      "learning_rate": 9.999382039377339e-06,
      "loss": 0.0601,
      "step": 56
    },
    {
      "epoch": 0.025932666060054597,
      "grad_norm": 0.9291387208138827,
      "learning_rate": 9.999359364281704e-06,
      "loss": 0.0377,
      "step": 57
    },
    {
      "epoch": 0.026387625113739762,
      "grad_norm": 1.8209170803663992,
      "learning_rate": 9.999336280660999e-06,
      "loss": 0.1144,
      "step": 58
    },
    {
      "epoch": 0.02684258416742493,
      "grad_norm": 1.1214625046464874,
      "learning_rate": 9.99931278851711e-06,
      "loss": 0.0622,
      "step": 59
    },
    {
      "epoch": 0.0272975432211101,
      "grad_norm": 1.0331723997917317,
      "learning_rate": 9.999288887851956e-06,
      "loss": 0.0667,
      "step": 60
    },
    {
      "epoch": 0.027752502274795268,
      "grad_norm": 1.0412381501406744,
      "learning_rate": 9.999264578667493e-06,
      "loss": 0.0566,
      "step": 61
    },
    {
      "epoch": 0.028207461328480437,
      "grad_norm": 1.4510603110658047,
      "learning_rate": 9.999239860965703e-06,
      "loss": 0.0845,
      "step": 62
    },
    {
      "epoch": 0.028662420382165606,
      "grad_norm": 1.301162540669183,
      "learning_rate": 9.999214734748609e-06,
      "loss": 0.0759,
      "step": 63
    },
    {
      "epoch": 0.029117379435850774,
      "grad_norm": 0.9977688847603402,
      "learning_rate": 9.999189200018263e-06,
      "loss": 0.0528,
      "step": 64
    },
    {
      "epoch": 0.029572338489535943,
      "grad_norm": 1.2894688842348854,
      "learning_rate": 9.99916325677675e-06,
      "loss": 0.0899,
      "step": 65
    },
    {
      "epoch": 0.03002729754322111,
      "grad_norm": 1.4627871680702638,
      "learning_rate": 9.999136905026194e-06,
      "loss": 0.1456,
      "step": 66
    },
    {
      "epoch": 0.030482256596906277,
      "grad_norm": 1.2304385710214434,
      "learning_rate": 9.999110144768745e-06,
      "loss": 0.079,
      "step": 67
    },
    {
      "epoch": 0.030937215650591446,
      "grad_norm": 1.085016380732753,
      "learning_rate": 9.99908297600659e-06,
      "loss": 0.0696,
      "step": 68
    },
    {
      "epoch": 0.03139217470427662,
      "grad_norm": 0.989450558642297,
      "learning_rate": 9.99905539874195e-06,
      "loss": 0.069,
      "step": 69
    },
    {
      "epoch": 0.03184713375796178,
      "grad_norm": 1.0510491151133208,
      "learning_rate": 9.99902741297708e-06,
      "loss": 0.0555,
      "step": 70
    },
    {
      "epoch": 0.03230209281164695,
      "grad_norm": 0.8938033562648371,
      "learning_rate": 9.998999018714264e-06,
      "loss": 0.0783,
      "step": 71
    },
    {
      "epoch": 0.03275705186533212,
      "grad_norm": 2.902512108322722,
      "learning_rate": 9.998970215955824e-06,
      "loss": 0.0702,
      "step": 72
    },
    {
      "epoch": 0.033212010919017286,
      "grad_norm": 0.7661831894133686,
      "learning_rate": 9.998941004704113e-06,
      "loss": 0.0519,
      "step": 73
    },
    {
      "epoch": 0.03366696997270246,
      "grad_norm": 1.1047249497744047,
      "learning_rate": 9.998911384961518e-06,
      "loss": 0.0773,
      "step": 74
    },
    {
      "epoch": 0.034121929026387623,
      "grad_norm": 0.7750047299312716,
      "learning_rate": 9.998881356730458e-06,
      "loss": 0.0598,
      "step": 75
    },
    {
      "epoch": 0.034576888080072796,
      "grad_norm": 0.9815801555720315,
      "learning_rate": 9.99885092001339e-06,
      "loss": 0.0661,
      "step": 76
    },
    {
      "epoch": 0.03503184713375796,
      "grad_norm": 1.3090963451351905,
      "learning_rate": 9.998820074812799e-06,
      "loss": 0.0713,
      "step": 77
    },
    {
      "epoch": 0.03548680618744313,
      "grad_norm": 1.1489338732270693,
      "learning_rate": 9.998788821131207e-06,
      "loss": 0.0946,
      "step": 78
    },
    {
      "epoch": 0.0359417652411283,
      "grad_norm": 0.9040381990998293,
      "learning_rate": 9.998757158971164e-06,
      "loss": 0.067,
      "step": 79
    },
    {
      "epoch": 0.036396724294813464,
      "grad_norm": 1.1019926198229115,
      "learning_rate": 9.998725088335263e-06,
      "loss": 0.0874,
      "step": 80
    },
    {
      "epoch": 0.036851683348498636,
      "grad_norm": 0.5779852750462403,
      "learning_rate": 9.99869260922612e-06,
      "loss": 0.0492,
      "step": 81
    },
    {
      "epoch": 0.0373066424021838,
      "grad_norm": 1.2769852710418472,
      "learning_rate": 9.998659721646393e-06,
      "loss": 0.0781,
      "step": 82
    },
    {
      "epoch": 0.03776160145586897,
      "grad_norm": 0.9020624084974485,
      "learning_rate": 9.998626425598766e-06,
      "loss": 0.0734,
      "step": 83
    },
    {
      "epoch": 0.03821656050955414,
      "grad_norm": 0.9626764462141776,
      "learning_rate": 9.99859272108596e-06,
      "loss": 0.0719,
      "step": 84
    },
    {
      "epoch": 0.03867151956323931,
      "grad_norm": 0.9435885887029873,
      "learning_rate": 9.998558608110733e-06,
      "loss": 0.0835,
      "step": 85
    },
    {
      "epoch": 0.039126478616924476,
      "grad_norm": 1.0578725525123687,
      "learning_rate": 9.998524086675867e-06,
      "loss": 0.0746,
      "step": 86
    },
    {
      "epoch": 0.03958143767060965,
      "grad_norm": 1.0366588534208079,
      "learning_rate": 9.998489156784188e-06,
      "loss": 0.0933,
      "step": 87
    },
    {
      "epoch": 0.040036396724294813,
      "grad_norm": 1.0595948680723846,
      "learning_rate": 9.998453818438547e-06,
      "loss": 0.0846,
      "step": 88
    },
    {
      "epoch": 0.04049135577797998,
      "grad_norm": 0.8807515753016749,
      "learning_rate": 9.998418071641833e-06,
      "loss": 0.0649,
      "step": 89
    },
    {
      "epoch": 0.04094631483166515,
      "grad_norm": 0.9034225145874141,
      "learning_rate": 9.998381916396967e-06,
      "loss": 0.0621,
      "step": 90
    },
    {
      "epoch": 0.041401273885350316,
      "grad_norm": 0.6732889821553815,
      "learning_rate": 9.998345352706901e-06,
      "loss": 0.0367,
      "step": 91
    },
    {
      "epoch": 0.04185623293903549,
      "grad_norm": 0.7136967603743426,
      "learning_rate": 9.998308380574628e-06,
      "loss": 0.0569,
      "step": 92
    },
    {
      "epoch": 0.042311191992720654,
      "grad_norm": 1.1459385364035048,
      "learning_rate": 9.998271000003166e-06,
      "loss": 0.1184,
      "step": 93
    },
    {
      "epoch": 0.042766151046405826,
      "grad_norm": 0.8224906129097734,
      "learning_rate": 9.998233210995569e-06,
      "loss": 0.0682,
      "step": 94
    },
    {
      "epoch": 0.04322111010009099,
      "grad_norm": 1.5182946932236698,
      "learning_rate": 9.998195013554926e-06,
      "loss": 0.0875,
      "step": 95
    },
    {
      "epoch": 0.04367606915377616,
      "grad_norm": 0.9355855711018981,
      "learning_rate": 9.998156407684359e-06,
      "loss": 0.0939,
      "step": 96
    },
    {
      "epoch": 0.04413102820746133,
      "grad_norm": 0.7329840867165283,
      "learning_rate": 9.998117393387022e-06,
      "loss": 0.0466,
      "step": 97
    },
    {
      "epoch": 0.044585987261146494,
      "grad_norm": 0.8701001036058451,
      "learning_rate": 9.9980779706661e-06,
      "loss": 0.0729,
      "step": 98
    },
    {
      "epoch": 0.045040946314831666,
      "grad_norm": 1.0218896298663185,
      "learning_rate": 9.99803813952482e-06,
      "loss": 0.0828,
      "step": 99
    },
    {
      "epoch": 0.04549590536851683,
      "grad_norm": 0.9044995357273884,
      "learning_rate": 9.997997899966433e-06,
      "loss": 0.0709,
      "step": 100
    },
    {
      "epoch": 0.045950864422202004,
      "grad_norm": 0.9877796099816964,
      "learning_rate": 9.99795725199423e-06,
      "loss": 0.0903,
      "step": 101
    },
    {
      "epoch": 0.04640582347588717,
      "grad_norm": 1.0061501994463906,
      "learning_rate": 9.99791619561153e-06,
      "loss": 0.0831,
      "step": 102
    },
    {
      "epoch": 0.04686078252957234,
      "grad_norm": 0.8789173954818107,
      "learning_rate": 9.997874730821689e-06,
      "loss": 0.0714,
      "step": 103
    },
    {
      "epoch": 0.047315741583257506,
      "grad_norm": 15.480920098194954,
      "learning_rate": 9.997832857628093e-06,
      "loss": 0.2603,
      "step": 104
    },
    {
      "epoch": 0.04777070063694268,
      "grad_norm": 1.3806761301603454,
      "learning_rate": 9.99779057603417e-06,
      "loss": 0.1227,
      "step": 105
    },
    {
      "epoch": 0.048225659690627844,
      "grad_norm": 0.8462176607269959,
      "learning_rate": 9.997747886043368e-06,
      "loss": 0.0605,
      "step": 106
    },
    {
      "epoch": 0.04868061874431301,
      "grad_norm": 0.7467169847716549,
      "learning_rate": 9.997704787659179e-06,
      "loss": 0.0618,
      "step": 107
    },
    {
      "epoch": 0.04913557779799818,
      "grad_norm": 1.5653334818977065,
      "learning_rate": 9.997661280885125e-06,
      "loss": 0.1253,
      "step": 108
    },
    {
      "epoch": 0.049590536851683346,
      "grad_norm": 0.871706038604149,
      "learning_rate": 9.99761736572476e-06,
      "loss": 0.0716,
      "step": 109
    },
    {
      "epoch": 0.05004549590536852,
      "grad_norm": 1.1398296008355844,
      "learning_rate": 9.997573042181672e-06,
      "loss": 0.0698,
      "step": 110
    },
    {
      "epoch": 0.050500454959053684,
      "grad_norm": 1.0487992691419916,
      "learning_rate": 9.997528310259485e-06,
      "loss": 0.1102,
      "step": 111
    },
    {
      "epoch": 0.050955414012738856,
      "grad_norm": 0.9112684449646818,
      "learning_rate": 9.997483169961852e-06,
      "loss": 0.1032,
      "step": 112
    },
    {
      "epoch": 0.05141037306642402,
      "grad_norm": 0.9418790141923585,
      "learning_rate": 9.997437621292463e-06,
      "loss": 0.0771,
      "step": 113
    },
    {
      "epoch": 0.051865332120109194,
      "grad_norm": 0.7796140692842074,
      "learning_rate": 9.99739166425504e-06,
      "loss": 0.0627,
      "step": 114
    },
    {
      "epoch": 0.05232029117379436,
      "grad_norm": 1.5434421216734795,
      "learning_rate": 9.997345298853339e-06,
      "loss": 0.1495,
      "step": 115
    },
    {
      "epoch": 0.052775250227479524,
      "grad_norm": 0.8898179660551836,
      "learning_rate": 9.997298525091148e-06,
      "loss": 0.0735,
      "step": 116
    },
    {
      "epoch": 0.053230209281164696,
      "grad_norm": 0.8585916871524272,
      "learning_rate": 9.997251342972288e-06,
      "loss": 0.068,
      "step": 117
    },
    {
      "epoch": 0.05368516833484986,
      "grad_norm": 0.812806800238708,
      "learning_rate": 9.997203752500616e-06,
      "loss": 0.0689,
      "step": 118
    },
    {
      "epoch": 0.054140127388535034,
      "grad_norm": 0.9677722064277628,
      "learning_rate": 9.997155753680021e-06,
      "loss": 0.0795,
      "step": 119
    },
    {
      "epoch": 0.0545950864422202,
      "grad_norm": 1.621934591654054,
      "learning_rate": 9.997107346514425e-06,
      "loss": 0.0707,
      "step": 120
    },
    {
      "epoch": 0.05505004549590537,
      "grad_norm": 0.6750452750311531,
      "learning_rate": 9.997058531007782e-06,
      "loss": 0.0588,
      "step": 121
    },
    {
      "epoch": 0.055505004549590536,
      "grad_norm": 0.9583870506818666,
      "learning_rate": 9.997009307164083e-06,
      "loss": 0.0859,
      "step": 122
    },
    {
      "epoch": 0.05595996360327571,
      "grad_norm": 1.247483970027119,
      "learning_rate": 9.99695967498735e-06,
      "loss": 0.0952,
      "step": 123
    },
    {
      "epoch": 0.056414922656960874,
      "grad_norm": 0.7937903902273558,
      "learning_rate": 9.996909634481639e-06,
      "loss": 0.0614,
      "step": 124
    },
    {
      "epoch": 0.05686988171064604,
      "grad_norm": 4.855426128828546,
      "learning_rate": 9.996859185651038e-06,
      "loss": 0.1629,
      "step": 125
    },
    {
      "epoch": 0.05732484076433121,
      "grad_norm": 1.0499970639607177,
      "learning_rate": 9.99680832849967e-06,
      "loss": 0.1031,
      "step": 126
    },
    {
      "epoch": 0.05777979981801638,
      "grad_norm": 0.8730447821488512,
      "learning_rate": 9.99675706303169e-06,
      "loss": 0.0606,
      "step": 127
    },
    {
      "epoch": 0.05823475887170155,
      "grad_norm": 1.2779985416162813,
      "learning_rate": 9.99670538925129e-06,
      "loss": 0.074,
      "step": 128
    },
    {
      "epoch": 0.058689717925386714,
      "grad_norm": 0.8606157718419157,
      "learning_rate": 9.996653307162687e-06,
      "loss": 0.0703,
      "step": 129
    },
    {
      "epoch": 0.059144676979071886,
      "grad_norm": 0.8920761218762643,
      "learning_rate": 9.996600816770144e-06,
      "loss": 0.0818,
      "step": 130
    },
    {
      "epoch": 0.05959963603275705,
      "grad_norm": 1.1603462045917847,
      "learning_rate": 9.996547918077944e-06,
      "loss": 0.1148,
      "step": 131
    },
    {
      "epoch": 0.06005459508644222,
      "grad_norm": 0.9108713801214797,
      "learning_rate": 9.996494611090414e-06,
      "loss": 0.0884,
      "step": 132
    },
    {
      "epoch": 0.06050955414012739,
      "grad_norm": 0.6523725468628359,
      "learning_rate": 9.996440895811907e-06,
      "loss": 0.0535,
      "step": 133
    },
    {
      "epoch": 0.060964513193812554,
      "grad_norm": 0.8812777694752004,
      "learning_rate": 9.996386772246816e-06,
      "loss": 0.087,
      "step": 134
    },
    {
      "epoch": 0.061419472247497726,
      "grad_norm": 1.0622191207422995,
      "learning_rate": 9.99633224039956e-06,
      "loss": 0.0982,
      "step": 135
    },
    {
      "epoch": 0.06187443130118289,
      "grad_norm": 3.7961077321923025,
      "learning_rate": 9.996277300274596e-06,
      "loss": 0.1526,
      "step": 136
    },
    {
      "epoch": 0.062329390354868064,
      "grad_norm": 0.9444433559435487,
      "learning_rate": 9.996221951876415e-06,
      "loss": 0.0996,
      "step": 137
    },
    {
      "epoch": 0.06278434940855324,
      "grad_norm": 1.444871481552235,
      "learning_rate": 9.996166195209539e-06,
      "loss": 0.1075,
      "step": 138
    },
    {
      "epoch": 0.0632393084622384,
      "grad_norm": 0.7446446480732116,
      "learning_rate": 9.996110030278522e-06,
      "loss": 0.0561,
      "step": 139
    },
    {
      "epoch": 0.06369426751592357,
      "grad_norm": 0.8913010543094952,
      "learning_rate": 9.996053457087958e-06,
      "loss": 0.0715,
      "step": 140
    },
    {
      "epoch": 0.06414922656960874,
      "grad_norm": 0.7815821404043856,
      "learning_rate": 9.995996475642466e-06,
      "loss": 0.0796,
      "step": 141
    },
    {
      "epoch": 0.0646041856232939,
      "grad_norm": 0.74337588448595,
      "learning_rate": 9.995939085946704e-06,
      "loss": 0.0661,
      "step": 142
    },
    {
      "epoch": 0.06505914467697907,
      "grad_norm": 0.9974255688753435,
      "learning_rate": 9.995881288005363e-06,
      "loss": 0.0869,
      "step": 143
    },
    {
      "epoch": 0.06551410373066424,
      "grad_norm": 1.2260290141946268,
      "learning_rate": 9.995823081823162e-06,
      "loss": 0.0766,
      "step": 144
    },
    {
      "epoch": 0.06596906278434941,
      "grad_norm": 0.9751795993584637,
      "learning_rate": 9.99576446740486e-06,
      "loss": 0.091,
      "step": 145
    },
    {
      "epoch": 0.06642402183803457,
      "grad_norm": 1.6175476325168967,
      "learning_rate": 9.995705444755249e-06,
      "loss": 0.1208,
      "step": 146
    },
    {
      "epoch": 0.06687898089171974,
      "grad_norm": 0.7580083688127299,
      "learning_rate": 9.995646013879147e-06,
      "loss": 0.0622,
      "step": 147
    },
    {
      "epoch": 0.06733393994540492,
      "grad_norm": 1.0194887039793072,
      "learning_rate": 9.995586174781413e-06,
      "loss": 0.0753,
      "step": 148
    },
    {
      "epoch": 0.06778889899909009,
      "grad_norm": 0.9065646408503975,
      "learning_rate": 9.995525927466936e-06,
      "loss": 0.0848,
      "step": 149
    },
    {
      "epoch": 0.06824385805277525,
      "grad_norm": 0.8871078738477127,
      "learning_rate": 9.995465271940641e-06,
      "loss": 0.0607,
      "step": 150
    },
    {
      "epoch": 0.06869881710646042,
      "grad_norm": 1.1486707652049646,
      "learning_rate": 9.995404208207485e-06,
      "loss": 0.0809,
      "step": 151
    },
    {
      "epoch": 0.06915377616014559,
      "grad_norm": 1.1473150526096232,
      "learning_rate": 9.995342736272453e-06,
      "loss": 0.1035,
      "step": 152
    },
    {
      "epoch": 0.06960873521383075,
      "grad_norm": 1.3025683052462544,
      "learning_rate": 9.995280856140572e-06,
      "loss": 0.1197,
      "step": 153
    },
    {
      "epoch": 0.07006369426751592,
      "grad_norm": 0.8069596755970996,
      "learning_rate": 9.9952185678169e-06,
      "loss": 0.0526,
      "step": 154
    },
    {
      "epoch": 0.0705186533212011,
      "grad_norm": 0.8153700064848134,
      "learning_rate": 9.995155871306524e-06,
      "loss": 0.0613,
      "step": 155
    },
    {
      "epoch": 0.07097361237488627,
      "grad_norm": 0.7319023745966868,
      "learning_rate": 9.995092766614567e-06,
      "loss": 0.0512,
      "step": 156
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 1.0146656175738817,
      "learning_rate": 9.995029253746186e-06,
      "loss": 0.0846,
      "step": 157
    },
    {
      "epoch": 0.0718835304822566,
      "grad_norm": 0.8015254985373994,
      "learning_rate": 9.994965332706574e-06,
      "loss": 0.0619,
      "step": 158
    },
    {
      "epoch": 0.07233848953594177,
      "grad_norm": 1.0630207312416284,
      "learning_rate": 9.994901003500952e-06,
      "loss": 0.0796,
      "step": 159
    },
    {
      "epoch": 0.07279344858962693,
      "grad_norm": 0.9431304991088505,
      "learning_rate": 9.994836266134575e-06,
      "loss": 0.0743,
      "step": 160
    },
    {
      "epoch": 0.0732484076433121,
      "grad_norm": 1.023738915097686,
      "learning_rate": 9.994771120612737e-06,
      "loss": 0.0888,
      "step": 161
    },
    {
      "epoch": 0.07370336669699727,
      "grad_norm": 0.9272637744585672,
      "learning_rate": 9.994705566940757e-06,
      "loss": 0.084,
      "step": 162
    },
    {
      "epoch": 0.07415832575068244,
      "grad_norm": 1.122378326253592,
      "learning_rate": 9.994639605123994e-06,
      "loss": 0.0961,
      "step": 163
    },
    {
      "epoch": 0.0746132848043676,
      "grad_norm": 0.753531768411978,
      "learning_rate": 9.994573235167839e-06,
      "loss": 0.0736,
      "step": 164
    },
    {
      "epoch": 0.07506824385805277,
      "grad_norm": 0.9314766958597749,
      "learning_rate": 9.994506457077715e-06,
      "loss": 0.0838,
      "step": 165
    },
    {
      "epoch": 0.07552320291173795,
      "grad_norm": 0.996008388557059,
      "learning_rate": 9.994439270859077e-06,
      "loss": 0.1076,
      "step": 166
    },
    {
      "epoch": 0.07597816196542312,
      "grad_norm": 0.9199332464612126,
      "learning_rate": 9.994371676517418e-06,
      "loss": 0.0724,
      "step": 167
    },
    {
      "epoch": 0.07643312101910828,
      "grad_norm": 0.8652292283168678,
      "learning_rate": 9.994303674058259e-06,
      "loss": 0.0628,
      "step": 168
    },
    {
      "epoch": 0.07688808007279345,
      "grad_norm": 0.8176262426438138,
      "learning_rate": 9.994235263487158e-06,
      "loss": 0.0743,
      "step": 169
    },
    {
      "epoch": 0.07734303912647862,
      "grad_norm": 0.8147855247941459,
      "learning_rate": 9.994166444809705e-06,
      "loss": 0.0559,
      "step": 170
    },
    {
      "epoch": 0.07779799818016378,
      "grad_norm": 0.7853019575635352,
      "learning_rate": 9.994097218031524e-06,
      "loss": 0.0681,
      "step": 171
    },
    {
      "epoch": 0.07825295723384895,
      "grad_norm": 0.8445610480134321,
      "learning_rate": 9.994027583158272e-06,
      "loss": 0.0785,
      "step": 172
    },
    {
      "epoch": 0.07870791628753412,
      "grad_norm": 0.8555498692388026,
      "learning_rate": 9.993957540195638e-06,
      "loss": 0.077,
      "step": 173
    },
    {
      "epoch": 0.0791628753412193,
      "grad_norm": 0.8281270493499452,
      "learning_rate": 9.993887089149346e-06,
      "loss": 0.0848,
      "step": 174
    },
    {
      "epoch": 0.07961783439490445,
      "grad_norm": 0.7180425978661062,
      "learning_rate": 9.993816230025152e-06,
      "loss": 0.0588,
      "step": 175
    },
    {
      "epoch": 0.08007279344858963,
      "grad_norm": 0.9287545326980071,
      "learning_rate": 9.99374496282885e-06,
      "loss": 0.0874,
      "step": 176
    },
    {
      "epoch": 0.0805277525022748,
      "grad_norm": 1.5950603980195528,
      "learning_rate": 9.993673287566261e-06,
      "loss": 0.1301,
      "step": 177
    },
    {
      "epoch": 0.08098271155595996,
      "grad_norm": 0.505966633973175,
      "learning_rate": 9.99360120424324e-06,
      "loss": 0.0459,
      "step": 178
    },
    {
      "epoch": 0.08143767060964513,
      "grad_norm": 0.6170796905443107,
      "learning_rate": 9.993528712865681e-06,
      "loss": 0.0666,
      "step": 179
    },
    {
      "epoch": 0.0818926296633303,
      "grad_norm": 0.8965600572228928,
      "learning_rate": 9.993455813439507e-06,
      "loss": 0.0648,
      "step": 180
    },
    {
      "epoch": 0.08234758871701547,
      "grad_norm": 0.7555745664692847,
      "learning_rate": 9.993382505970673e-06,
      "loss": 0.0479,
      "step": 181
    },
    {
      "epoch": 0.08280254777070063,
      "grad_norm": 0.7885826993774436,
      "learning_rate": 9.99330879046517e-06,
      "loss": 0.0605,
      "step": 182
    },
    {
      "epoch": 0.0832575068243858,
      "grad_norm": 0.6970911126559147,
      "learning_rate": 9.993234666929024e-06,
      "loss": 0.0545,
      "step": 183
    },
    {
      "epoch": 0.08371246587807098,
      "grad_norm": 0.8281240642020996,
      "learning_rate": 9.99316013536829e-06,
      "loss": 0.0651,
      "step": 184
    },
    {
      "epoch": 0.08416742493175614,
      "grad_norm": 0.8497823551734951,
      "learning_rate": 9.993085195789057e-06,
      "loss": 0.098,
      "step": 185
    },
    {
      "epoch": 0.08462238398544131,
      "grad_norm": 0.8425278224044996,
      "learning_rate": 9.993009848197452e-06,
      "loss": 0.0861,
      "step": 186
    },
    {
      "epoch": 0.08507734303912648,
      "grad_norm": 0.729342450692031,
      "learning_rate": 9.992934092599629e-06,
      "loss": 0.0651,
      "step": 187
    },
    {
      "epoch": 0.08553230209281165,
      "grad_norm": 0.8810253378927329,
      "learning_rate": 9.99285792900178e-06,
      "loss": 0.0995,
      "step": 188
    },
    {
      "epoch": 0.08598726114649681,
      "grad_norm": 1.0402457083445067,
      "learning_rate": 9.992781357410131e-06,
      "loss": 0.1061,
      "step": 189
    },
    {
      "epoch": 0.08644222020018198,
      "grad_norm": 0.7397036090930822,
      "learning_rate": 9.992704377830934e-06,
      "loss": 0.0571,
      "step": 190
    },
    {
      "epoch": 0.08689717925386715,
      "grad_norm": 1.4783630598693296,
      "learning_rate": 9.992626990270484e-06,
      "loss": 0.1154,
      "step": 191
    },
    {
      "epoch": 0.08735213830755233,
      "grad_norm": 1.1100322283473036,
      "learning_rate": 9.992549194735101e-06,
      "loss": 0.1179,
      "step": 192
    },
    {
      "epoch": 0.08780709736123748,
      "grad_norm": 0.5797984556503705,
      "learning_rate": 9.992470991231144e-06,
      "loss": 0.0466,
      "step": 193
    },
    {
      "epoch": 0.08826205641492266,
      "grad_norm": 1.059908713900853,
      "learning_rate": 9.992392379765005e-06,
      "loss": 0.0994,
      "step": 194
    },
    {
      "epoch": 0.08871701546860783,
      "grad_norm": 1.1187885391430794,
      "learning_rate": 9.992313360343104e-06,
      "loss": 0.0986,
      "step": 195
    },
    {
      "epoch": 0.08917197452229299,
      "grad_norm": 0.7509441330173129,
      "learning_rate": 9.992233932971901e-06,
      "loss": 0.0634,
      "step": 196
    },
    {
      "epoch": 0.08962693357597816,
      "grad_norm": 0.9426276516690344,
      "learning_rate": 9.992154097657888e-06,
      "loss": 0.0857,
      "step": 197
    },
    {
      "epoch": 0.09008189262966333,
      "grad_norm": 0.8754039034503873,
      "learning_rate": 9.992073854407585e-06,
      "loss": 0.0881,
      "step": 198
    },
    {
      "epoch": 0.0905368516833485,
      "grad_norm": 2.8697219156120712,
      "learning_rate": 9.99199320322755e-06,
      "loss": 0.0851,
      "step": 199
    },
    {
      "epoch": 0.09099181073703366,
      "grad_norm": 0.7429242681646778,
      "learning_rate": 9.991912144124375e-06,
      "loss": 0.0729,
      "step": 200
    },
    {
      "epoch": 0.09144676979071883,
      "grad_norm": 1.0552979449251756,
      "learning_rate": 9.991830677104682e-06,
      "loss": 0.1066,
      "step": 201
    },
    {
      "epoch": 0.09190172884440401,
      "grad_norm": 0.8812651371324355,
      "learning_rate": 9.99174880217513e-06,
      "loss": 0.0732,
      "step": 202
    },
    {
      "epoch": 0.09235668789808917,
      "grad_norm": 1.0755107845413352,
      "learning_rate": 9.991666519342407e-06,
      "loss": 0.0977,
      "step": 203
    },
    {
      "epoch": 0.09281164695177434,
      "grad_norm": 0.8925063431256136,
      "learning_rate": 9.99158382861324e-06,
      "loss": 0.0904,
      "step": 204
    },
    {
      "epoch": 0.09326660600545951,
      "grad_norm": 0.8190206986922173,
      "learning_rate": 9.991500729994384e-06,
      "loss": 0.0729,
      "step": 205
    },
    {
      "epoch": 0.09372156505914468,
      "grad_norm": 0.6635798147425112,
      "learning_rate": 9.991417223492629e-06,
      "loss": 0.0631,
      "step": 206
    },
    {
      "epoch": 0.09417652411282984,
      "grad_norm": 1.0314655306023923,
      "learning_rate": 9.991333309114798e-06,
      "loss": 0.0852,
      "step": 207
    },
    {
      "epoch": 0.09463148316651501,
      "grad_norm": 0.8533496857694978,
      "learning_rate": 9.991248986867753e-06,
      "loss": 0.0868,
      "step": 208
    },
    {
      "epoch": 0.09508644222020018,
      "grad_norm": 1.039085255997433,
      "learning_rate": 9.991164256758378e-06,
      "loss": 0.095,
      "step": 209
    },
    {
      "epoch": 0.09554140127388536,
      "grad_norm": 1.1484522866350177,
      "learning_rate": 9.9910791187936e-06,
      "loss": 0.1333,
      "step": 210
    },
    {
      "epoch": 0.09599636032757052,
      "grad_norm": 0.8277820800102422,
      "learning_rate": 9.99099357298038e-06,
      "loss": 0.0664,
      "step": 211
    },
    {
      "epoch": 0.09645131938125569,
      "grad_norm": 0.821796111319934,
      "learning_rate": 9.9909076193257e-06,
      "loss": 0.083,
      "step": 212
    },
    {
      "epoch": 0.09690627843494086,
      "grad_norm": 0.9448800546720313,
      "learning_rate": 9.990821257836589e-06,
      "loss": 0.0873,
      "step": 213
    },
    {
      "epoch": 0.09736123748862602,
      "grad_norm": 0.9002810379340489,
      "learning_rate": 9.990734488520103e-06,
      "loss": 0.099,
      "step": 214
    },
    {
      "epoch": 0.09781619654231119,
      "grad_norm": 0.6145149717344348,
      "learning_rate": 9.990647311383334e-06,
      "loss": 0.0425,
      "step": 215
    },
    {
      "epoch": 0.09827115559599636,
      "grad_norm": 1.1377497370761045,
      "learning_rate": 9.990559726433404e-06,
      "loss": 0.0903,
      "step": 216
    },
    {
      "epoch": 0.09872611464968153,
      "grad_norm": 0.8401357673155365,
      "learning_rate": 9.99047173367747e-06,
      "loss": 0.0812,
      "step": 217
    },
    {
      "epoch": 0.09918107370336669,
      "grad_norm": 0.6977882365614015,
      "learning_rate": 9.990383333122722e-06,
      "loss": 0.0613,
      "step": 218
    },
    {
      "epoch": 0.09963603275705187,
      "grad_norm": 0.6751056796776193,
      "learning_rate": 9.990294524776384e-06,
      "loss": 0.0636,
      "step": 219
    },
    {
      "epoch": 0.10009099181073704,
      "grad_norm": 0.7973250315161167,
      "learning_rate": 9.990205308645716e-06,
      "loss": 0.0655,
      "step": 220
    },
    {
      "epoch": 0.1005459508644222,
      "grad_norm": 0.6494979859380491,
      "learning_rate": 9.990115684738005e-06,
      "loss": 0.0461,
      "step": 221
    },
    {
      "epoch": 0.10100090991810737,
      "grad_norm": 0.7863907355652456,
      "learning_rate": 9.990025653060574e-06,
      "loss": 0.0881,
      "step": 222
    },
    {
      "epoch": 0.10145586897179254,
      "grad_norm": 1.2756737972223395,
      "learning_rate": 9.98993521362078e-06,
      "loss": 0.1102,
      "step": 223
    },
    {
      "epoch": 0.10191082802547771,
      "grad_norm": 1.1992554133605928,
      "learning_rate": 9.989844366426018e-06,
      "loss": 0.1147,
      "step": 224
    },
    {
      "epoch": 0.10236578707916287,
      "grad_norm": 0.5034605400337953,
      "learning_rate": 9.989753111483707e-06,
      "loss": 0.0462,
      "step": 225
    },
    {
      "epoch": 0.10282074613284804,
      "grad_norm": 0.9881921480518578,
      "learning_rate": 9.989661448801305e-06,
      "loss": 0.0848,
      "step": 226
    },
    {
      "epoch": 0.10327570518653321,
      "grad_norm": 0.7581777568438945,
      "learning_rate": 9.989569378386303e-06,
      "loss": 0.079,
      "step": 227
    },
    {
      "epoch": 0.10373066424021839,
      "grad_norm": 0.6464731162067388,
      "learning_rate": 9.989476900246223e-06,
      "loss": 0.0617,
      "step": 228
    },
    {
      "epoch": 0.10418562329390355,
      "grad_norm": 0.8780639185859085,
      "learning_rate": 9.989384014388624e-06,
      "loss": 0.086,
      "step": 229
    },
    {
      "epoch": 0.10464058234758872,
      "grad_norm": 0.6623808171307163,
      "learning_rate": 9.989290720821095e-06,
      "loss": 0.0694,
      "step": 230
    },
    {
      "epoch": 0.10509554140127389,
      "grad_norm": 0.721054554263859,
      "learning_rate": 9.98919701955126e-06,
      "loss": 0.0735,
      "step": 231
    },
    {
      "epoch": 0.10555050045495905,
      "grad_norm": 0.7868134014829404,
      "learning_rate": 9.989102910586776e-06,
      "loss": 0.0546,
      "step": 232
    },
    {
      "epoch": 0.10600545950864422,
      "grad_norm": 0.9137158371163484,
      "learning_rate": 9.989008393935331e-06,
      "loss": 0.0771,
      "step": 233
    },
    {
      "epoch": 0.10646041856232939,
      "grad_norm": 0.8326009579593463,
      "learning_rate": 9.98891346960465e-06,
      "loss": 0.0667,
      "step": 234
    },
    {
      "epoch": 0.10691537761601456,
      "grad_norm": 0.6462724580348628,
      "learning_rate": 9.988818137602494e-06,
      "loss": 0.0717,
      "step": 235
    },
    {
      "epoch": 0.10737033666969972,
      "grad_norm": 0.7513725247558808,
      "learning_rate": 9.988722397936646e-06,
      "loss": 0.0733,
      "step": 236
    },
    {
      "epoch": 0.1078252957233849,
      "grad_norm": 1.094509848236789,
      "learning_rate": 9.988626250614932e-06,
      "loss": 0.1009,
      "step": 237
    },
    {
      "epoch": 0.10828025477707007,
      "grad_norm": 0.8200579138639758,
      "learning_rate": 9.98852969564521e-06,
      "loss": 0.0844,
      "step": 238
    },
    {
      "epoch": 0.10873521383075523,
      "grad_norm": 0.7417763562196316,
      "learning_rate": 9.988432733035369e-06,
      "loss": 0.0611,
      "step": 239
    },
    {
      "epoch": 0.1091901728844404,
      "grad_norm": 0.8476475869820355,
      "learning_rate": 9.988335362793333e-06,
      "loss": 0.0863,
      "step": 240
    },
    {
      "epoch": 0.10964513193812557,
      "grad_norm": 0.9998642783878469,
      "learning_rate": 9.988237584927058e-06,
      "loss": 0.0909,
      "step": 241
    },
    {
      "epoch": 0.11010009099181074,
      "grad_norm": 1.1689324698997519,
      "learning_rate": 9.988139399444534e-06,
      "loss": 0.124,
      "step": 242
    },
    {
      "epoch": 0.1105550500454959,
      "grad_norm": 0.790901332269412,
      "learning_rate": 9.988040806353786e-06,
      "loss": 0.0855,
      "step": 243
    },
    {
      "epoch": 0.11101000909918107,
      "grad_norm": 0.8931785977847209,
      "learning_rate": 9.987941805662869e-06,
      "loss": 0.1023,
      "step": 244
    },
    {
      "epoch": 0.11146496815286625,
      "grad_norm": 0.7352781929773609,
      "learning_rate": 9.98784239737987e-06,
      "loss": 0.0563,
      "step": 245
    },
    {
      "epoch": 0.11191992720655142,
      "grad_norm": 0.7169092611535308,
      "learning_rate": 9.987742581512919e-06,
      "loss": 0.0683,
      "step": 246
    },
    {
      "epoch": 0.11237488626023658,
      "grad_norm": 0.6767560569792272,
      "learning_rate": 9.987642358070167e-06,
      "loss": 0.0669,
      "step": 247
    },
    {
      "epoch": 0.11282984531392175,
      "grad_norm": 0.8442319805699996,
      "learning_rate": 9.987541727059805e-06,
      "loss": 0.0768,
      "step": 248
    },
    {
      "epoch": 0.11328480436760692,
      "grad_norm": 0.7700876798522618,
      "learning_rate": 9.987440688490058e-06,
      "loss": 0.0643,
      "step": 249
    },
    {
      "epoch": 0.11373976342129208,
      "grad_norm": 0.7286087978317647,
      "learning_rate": 9.98733924236918e-06,
      "loss": 0.0698,
      "step": 250
    },
    {
      "epoch": 0.11419472247497725,
      "grad_norm": 0.7917355018437868,
      "learning_rate": 9.98723738870546e-06,
      "loss": 0.0791,
      "step": 251
    },
    {
      "epoch": 0.11464968152866242,
      "grad_norm": 1.0469499693242315,
      "learning_rate": 9.987135127507226e-06,
      "loss": 0.0761,
      "step": 252
    },
    {
      "epoch": 0.1151046405823476,
      "grad_norm": 0.8361714930383379,
      "learning_rate": 9.987032458782828e-06,
      "loss": 0.0789,
      "step": 253
    },
    {
      "epoch": 0.11555959963603275,
      "grad_norm": 0.5902853873046482,
      "learning_rate": 9.986929382540662e-06,
      "loss": 0.0479,
      "step": 254
    },
    {
      "epoch": 0.11601455868971793,
      "grad_norm": 0.7349436304465384,
      "learning_rate": 9.986825898789145e-06,
      "loss": 0.0668,
      "step": 255
    },
    {
      "epoch": 0.1164695177434031,
      "grad_norm": 0.7657107039148755,
      "learning_rate": 9.986722007536737e-06,
      "loss": 0.0617,
      "step": 256
    },
    {
      "epoch": 0.11692447679708826,
      "grad_norm": 0.6450631027744769,
      "learning_rate": 9.986617708791926e-06,
      "loss": 0.0679,
      "step": 257
    },
    {
      "epoch": 0.11737943585077343,
      "grad_norm": 0.6292930010016882,
      "learning_rate": 9.986513002563236e-06,
      "loss": 0.0482,
      "step": 258
    },
    {
      "epoch": 0.1178343949044586,
      "grad_norm": 0.8758541343517451,
      "learning_rate": 9.986407888859221e-06,
      "loss": 0.0994,
      "step": 259
    },
    {
      "epoch": 0.11828935395814377,
      "grad_norm": 0.6537445862223847,
      "learning_rate": 9.986302367688473e-06,
      "loss": 0.07,
      "step": 260
    },
    {
      "epoch": 0.11874431301182893,
      "grad_norm": 0.8029660816844667,
      "learning_rate": 9.986196439059613e-06,
      "loss": 0.0623,
      "step": 261
    },
    {
      "epoch": 0.1191992720655141,
      "grad_norm": 0.7339528606524214,
      "learning_rate": 9.986090102981297e-06,
      "loss": 0.0791,
      "step": 262
    },
    {
      "epoch": 0.11965423111919928,
      "grad_norm": 0.7934112522002073,
      "learning_rate": 9.985983359462215e-06,
      "loss": 0.0672,
      "step": 263
    },
    {
      "epoch": 0.12010919017288443,
      "grad_norm": 1.0186962263060808,
      "learning_rate": 9.98587620851109e-06,
      "loss": 0.1213,
      "step": 264
    },
    {
      "epoch": 0.1205641492265696,
      "grad_norm": 0.6769843647605545,
      "learning_rate": 9.985768650136679e-06,
      "loss": 0.0685,
      "step": 265
    },
    {
      "epoch": 0.12101910828025478,
      "grad_norm": 0.7543020935976431,
      "learning_rate": 9.985660684347765e-06,
      "loss": 0.0861,
      "step": 266
    },
    {
      "epoch": 0.12147406733393995,
      "grad_norm": 0.9552124731299731,
      "learning_rate": 9.985552311153178e-06,
      "loss": 0.0922,
      "step": 267
    },
    {
      "epoch": 0.12192902638762511,
      "grad_norm": 0.7436699167226903,
      "learning_rate": 9.985443530561769e-06,
      "loss": 0.0885,
      "step": 268
    },
    {
      "epoch": 0.12238398544131028,
      "grad_norm": 1.329058937551934,
      "learning_rate": 9.98533434258243e-06,
      "loss": 0.1115,
      "step": 269
    },
    {
      "epoch": 0.12283894449499545,
      "grad_norm": 0.6835909813818813,
      "learning_rate": 9.985224747224083e-06,
      "loss": 0.0586,
      "step": 270
    },
    {
      "epoch": 0.12329390354868063,
      "grad_norm": 1.0733107060854794,
      "learning_rate": 9.98511474449568e-06,
      "loss": 0.0811,
      "step": 271
    },
    {
      "epoch": 0.12374886260236578,
      "grad_norm": 0.5916007278667166,
      "learning_rate": 9.985004334406215e-06,
      "loss": 0.0696,
      "step": 272
    },
    {
      "epoch": 0.12420382165605096,
      "grad_norm": 0.9149357508392912,
      "learning_rate": 9.984893516964707e-06,
      "loss": 0.0704,
      "step": 273
    },
    {
      "epoch": 0.12465878070973613,
      "grad_norm": 1.1634742377762608,
      "learning_rate": 9.984782292180212e-06,
      "loss": 0.1178,
      "step": 274
    },
    {
      "epoch": 0.1251137397634213,
      "grad_norm": 0.603957454908005,
      "learning_rate": 9.98467066006182e-06,
      "loss": 0.0585,
      "step": 275
    },
    {
      "epoch": 0.12556869881710647,
      "grad_norm": 0.7735087790025026,
      "learning_rate": 9.984558620618651e-06,
      "loss": 0.0953,
      "step": 276
    },
    {
      "epoch": 0.12602365787079162,
      "grad_norm": 1.2570182633873541,
      "learning_rate": 9.984446173859863e-06,
      "loss": 0.1353,
      "step": 277
    },
    {
      "epoch": 0.1264786169244768,
      "grad_norm": 0.7275895818672663,
      "learning_rate": 9.984333319794642e-06,
      "loss": 0.0774,
      "step": 278
    },
    {
      "epoch": 0.12693357597816196,
      "grad_norm": 0.6395006056363333,
      "learning_rate": 9.984220058432212e-06,
      "loss": 0.0591,
      "step": 279
    },
    {
      "epoch": 0.12738853503184713,
      "grad_norm": 0.6563921850032347,
      "learning_rate": 9.984106389781828e-06,
      "loss": 0.0573,
      "step": 280
    },
    {
      "epoch": 0.1278434940855323,
      "grad_norm": 0.9399157526953884,
      "learning_rate": 9.983992313852776e-06,
      "loss": 0.0793,
      "step": 281
    },
    {
      "epoch": 0.12829845313921748,
      "grad_norm": 0.93528061821534,
      "learning_rate": 9.983877830654381e-06,
      "loss": 0.0807,
      "step": 282
    },
    {
      "epoch": 0.12875341219290265,
      "grad_norm": 0.7192448233352142,
      "learning_rate": 9.983762940195996e-06,
      "loss": 0.0773,
      "step": 283
    },
    {
      "epoch": 0.1292083712465878,
      "grad_norm": 0.7097381072031733,
      "learning_rate": 9.98364764248701e-06,
      "loss": 0.0698,
      "step": 284
    },
    {
      "epoch": 0.12966333030027297,
      "grad_norm": 1.1635566012920768,
      "learning_rate": 9.983531937536844e-06,
      "loss": 0.0893,
      "step": 285
    },
    {
      "epoch": 0.13011828935395814,
      "grad_norm": 0.8456555685011555,
      "learning_rate": 9.983415825354954e-06,
      "loss": 0.0628,
      "step": 286
    },
    {
      "epoch": 0.1305732484076433,
      "grad_norm": 0.7151838393189083,
      "learning_rate": 9.983299305950828e-06,
      "loss": 0.0557,
      "step": 287
    },
    {
      "epoch": 0.13102820746132848,
      "grad_norm": 0.7095193783870621,
      "learning_rate": 9.983182379333989e-06,
      "loss": 0.0604,
      "step": 288
    },
    {
      "epoch": 0.13148316651501366,
      "grad_norm": 0.8581434444337498,
      "learning_rate": 9.983065045513986e-06,
      "loss": 0.0781,
      "step": 289
    },
    {
      "epoch": 0.13193812556869883,
      "grad_norm": 0.5600994934804626,
      "learning_rate": 9.982947304500414e-06,
      "loss": 0.0498,
      "step": 290
    },
    {
      "epoch": 0.13239308462238397,
      "grad_norm": 0.7355720212694087,
      "learning_rate": 9.98282915630289e-06,
      "loss": 0.0692,
      "step": 291
    },
    {
      "epoch": 0.13284804367606914,
      "grad_norm": 1.6846985851500909,
      "learning_rate": 9.98271060093107e-06,
      "loss": 0.1687,
      "step": 292
    },
    {
      "epoch": 0.13330300272975432,
      "grad_norm": 0.7959406174268434,
      "learning_rate": 9.98259163839464e-06,
      "loss": 0.0718,
      "step": 293
    },
    {
      "epoch": 0.1337579617834395,
      "grad_norm": 0.6005858848115938,
      "learning_rate": 9.982472268703323e-06,
      "loss": 0.0465,
      "step": 294
    },
    {
      "epoch": 0.13421292083712466,
      "grad_norm": 0.7865103977061746,
      "learning_rate": 9.982352491866874e-06,
      "loss": 0.071,
      "step": 295
    },
    {
      "epoch": 0.13466787989080983,
      "grad_norm": 0.7167219429964851,
      "learning_rate": 9.982232307895077e-06,
      "loss": 0.0658,
      "step": 296
    },
    {
      "epoch": 0.135122838944495,
      "grad_norm": 1.206398567596641,
      "learning_rate": 9.982111716797758e-06,
      "loss": 0.101,
      "step": 297
    },
    {
      "epoch": 0.13557779799818018,
      "grad_norm": 1.0085912508470862,
      "learning_rate": 9.981990718584768e-06,
      "loss": 0.0959,
      "step": 298
    },
    {
      "epoch": 0.13603275705186532,
      "grad_norm": 0.8594135430057543,
      "learning_rate": 9.981869313265995e-06,
      "loss": 0.0912,
      "step": 299
    },
    {
      "epoch": 0.1364877161055505,
      "grad_norm": 0.9903339586980618,
      "learning_rate": 9.981747500851357e-06,
      "loss": 0.0692,
      "step": 300
    },
    {
      "epoch": 0.13694267515923567,
      "grad_norm": 0.7623380548666351,
      "learning_rate": 9.981625281350812e-06,
      "loss": 0.0699,
      "step": 301
    },
    {
      "epoch": 0.13739763421292084,
      "grad_norm": 0.6267143484055344,
      "learning_rate": 9.981502654774349e-06,
      "loss": 0.0499,
      "step": 302
    },
    {
      "epoch": 0.137852593266606,
      "grad_norm": 0.8234150836820757,
      "learning_rate": 9.98137962113198e-06,
      "loss": 0.0788,
      "step": 303
    },
    {
      "epoch": 0.13830755232029118,
      "grad_norm": 0.8158733102806115,
      "learning_rate": 9.98125618043377e-06,
      "loss": 0.089,
      "step": 304
    },
    {
      "epoch": 0.13876251137397635,
      "grad_norm": 0.6372656549463032,
      "learning_rate": 9.981132332689796e-06,
      "loss": 0.0517,
      "step": 305
    },
    {
      "epoch": 0.1392174704276615,
      "grad_norm": 0.7713863813548327,
      "learning_rate": 9.981008077910184e-06,
      "loss": 0.0769,
      "step": 306
    },
    {
      "epoch": 0.13967242948134667,
      "grad_norm": 0.8883775702857831,
      "learning_rate": 9.980883416105084e-06,
      "loss": 0.0828,
      "step": 307
    },
    {
      "epoch": 0.14012738853503184,
      "grad_norm": 0.6490936355626988,
      "learning_rate": 9.980758347284687e-06,
      "loss": 0.0618,
      "step": 308
    },
    {
      "epoch": 0.14058234758871702,
      "grad_norm": 0.8359554084586713,
      "learning_rate": 9.980632871459209e-06,
      "loss": 0.0714,
      "step": 309
    },
    {
      "epoch": 0.1410373066424022,
      "grad_norm": 0.7373523328454649,
      "learning_rate": 9.980506988638906e-06,
      "loss": 0.0836,
      "step": 310
    },
    {
      "epoch": 0.14149226569608736,
      "grad_norm": 0.6644370731485183,
      "learning_rate": 9.980380698834064e-06,
      "loss": 0.0777,
      "step": 311
    },
    {
      "epoch": 0.14194722474977253,
      "grad_norm": 0.870883965477211,
      "learning_rate": 9.980254002055003e-06,
      "loss": 0.0847,
      "step": 312
    },
    {
      "epoch": 0.14240218380345768,
      "grad_norm": 0.6021065409531002,
      "learning_rate": 9.980126898312074e-06,
      "loss": 0.0583,
      "step": 313
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.8705461588189498,
      "learning_rate": 9.979999387615665e-06,
      "loss": 0.0895,
      "step": 314
    },
    {
      "epoch": 0.14331210191082802,
      "grad_norm": 0.9639410731114018,
      "learning_rate": 9.979871469976197e-06,
      "loss": 0.0901,
      "step": 315
    },
    {
      "epoch": 0.1437670609645132,
      "grad_norm": 0.7554126383153169,
      "learning_rate": 9.97974314540412e-06,
      "loss": 0.0699,
      "step": 316
    },
    {
      "epoch": 0.14422202001819837,
      "grad_norm": 1.1039648440512544,
      "learning_rate": 9.979614413909922e-06,
      "loss": 0.1013,
      "step": 317
    },
    {
      "epoch": 0.14467697907188354,
      "grad_norm": 0.5258831871743486,
      "learning_rate": 9.979485275504121e-06,
      "loss": 0.0544,
      "step": 318
    },
    {
      "epoch": 0.1451319381255687,
      "grad_norm": 1.3025897394440575,
      "learning_rate": 9.979355730197271e-06,
      "loss": 0.1067,
      "step": 319
    },
    {
      "epoch": 0.14558689717925385,
      "grad_norm": 0.5206132423310033,
      "learning_rate": 9.979225777999956e-06,
      "loss": 0.0497,
      "step": 320
    },
    {
      "epoch": 0.14604185623293903,
      "grad_norm": 0.7202189397663867,
      "learning_rate": 9.9790954189228e-06,
      "loss": 0.0807,
      "step": 321
    },
    {
      "epoch": 0.1464968152866242,
      "grad_norm": 0.5738667169449175,
      "learning_rate": 9.97896465297645e-06,
      "loss": 0.0614,
      "step": 322
    },
    {
      "epoch": 0.14695177434030937,
      "grad_norm": 0.7972440737628133,
      "learning_rate": 9.978833480171592e-06,
      "loss": 0.0906,
      "step": 323
    },
    {
      "epoch": 0.14740673339399454,
      "grad_norm": 0.7697423454053598,
      "learning_rate": 9.978701900518947e-06,
      "loss": 0.0632,
      "step": 324
    },
    {
      "epoch": 0.14786169244767972,
      "grad_norm": 0.8259885564233931,
      "learning_rate": 9.978569914029267e-06,
      "loss": 0.0944,
      "step": 325
    },
    {
      "epoch": 0.1483166515013649,
      "grad_norm": 0.8450006655868962,
      "learning_rate": 9.978437520713335e-06,
      "loss": 0.0862,
      "step": 326
    },
    {
      "epoch": 0.14877161055505003,
      "grad_norm": 0.7746078278616594,
      "learning_rate": 9.978304720581973e-06,
      "loss": 0.088,
      "step": 327
    },
    {
      "epoch": 0.1492265696087352,
      "grad_norm": 0.9977734940815816,
      "learning_rate": 9.97817151364603e-06,
      "loss": 0.1036,
      "step": 328
    },
    {
      "epoch": 0.14968152866242038,
      "grad_norm": 0.7800752301510507,
      "learning_rate": 9.978037899916393e-06,
      "loss": 0.0778,
      "step": 329
    },
    {
      "epoch": 0.15013648771610555,
      "grad_norm": 0.7521153273438224,
      "learning_rate": 9.97790387940398e-06,
      "loss": 0.0532,
      "step": 330
    },
    {
      "epoch": 0.15059144676979072,
      "grad_norm": 0.8046420256419254,
      "learning_rate": 9.977769452119741e-06,
      "loss": 0.0708,
      "step": 331
    },
    {
      "epoch": 0.1510464058234759,
      "grad_norm": 0.9071770528791517,
      "learning_rate": 9.97763461807466e-06,
      "loss": 0.1006,
      "step": 332
    },
    {
      "epoch": 0.15150136487716107,
      "grad_norm": 0.8824570234268595,
      "learning_rate": 9.97749937727976e-06,
      "loss": 0.0855,
      "step": 333
    },
    {
      "epoch": 0.15195632393084624,
      "grad_norm": 0.8286075823730068,
      "learning_rate": 9.977363729746088e-06,
      "loss": 0.077,
      "step": 334
    },
    {
      "epoch": 0.15241128298453138,
      "grad_norm": 0.6791233851472963,
      "learning_rate": 9.977227675484729e-06,
      "loss": 0.0698,
      "step": 335
    },
    {
      "epoch": 0.15286624203821655,
      "grad_norm": 0.9813875260679181,
      "learning_rate": 9.977091214506803e-06,
      "loss": 0.0838,
      "step": 336
    },
    {
      "epoch": 0.15332120109190173,
      "grad_norm": 0.9986284190120469,
      "learning_rate": 9.976954346823456e-06,
      "loss": 0.0789,
      "step": 337
    },
    {
      "epoch": 0.1537761601455869,
      "grad_norm": 0.6456071732838817,
      "learning_rate": 9.976817072445878e-06,
      "loss": 0.0566,
      "step": 338
    },
    {
      "epoch": 0.15423111919927207,
      "grad_norm": 0.7707362352402762,
      "learning_rate": 9.976679391385283e-06,
      "loss": 0.0677,
      "step": 339
    },
    {
      "epoch": 0.15468607825295724,
      "grad_norm": 0.5804713825378958,
      "learning_rate": 9.976541303652923e-06,
      "loss": 0.0547,
      "step": 340
    },
    {
      "epoch": 0.15514103730664242,
      "grad_norm": 0.7705377953828665,
      "learning_rate": 9.976402809260083e-06,
      "loss": 0.0673,
      "step": 341
    },
    {
      "epoch": 0.15559599636032756,
      "grad_norm": 0.651002355082985,
      "learning_rate": 9.976263908218076e-06,
      "loss": 0.066,
      "step": 342
    },
    {
      "epoch": 0.15605095541401273,
      "grad_norm": 1.0075230687249708,
      "learning_rate": 9.976124600538257e-06,
      "loss": 0.1151,
      "step": 343
    },
    {
      "epoch": 0.1565059144676979,
      "grad_norm": 0.7110146200064966,
      "learning_rate": 9.975984886232006e-06,
      "loss": 0.0693,
      "step": 344
    },
    {
      "epoch": 0.15696087352138308,
      "grad_norm": 0.782615076662302,
      "learning_rate": 9.975844765310743e-06,
      "loss": 0.071,
      "step": 345
    },
    {
      "epoch": 0.15741583257506825,
      "grad_norm": 1.091513822496144,
      "learning_rate": 9.975704237785915e-06,
      "loss": 0.1277,
      "step": 346
    },
    {
      "epoch": 0.15787079162875342,
      "grad_norm": 0.8244942271322709,
      "learning_rate": 9.975563303669006e-06,
      "loss": 0.092,
      "step": 347
    },
    {
      "epoch": 0.1583257506824386,
      "grad_norm": 1.0997264747524325,
      "learning_rate": 9.975421962971536e-06,
      "loss": 0.102,
      "step": 348
    },
    {
      "epoch": 0.15878070973612374,
      "grad_norm": 1.0471722358260585,
      "learning_rate": 9.97528021570505e-06,
      "loss": 0.1112,
      "step": 349
    },
    {
      "epoch": 0.1592356687898089,
      "grad_norm": 0.6366013160292697,
      "learning_rate": 9.975138061881135e-06,
      "loss": 0.0629,
      "step": 350
    },
    {
      "epoch": 0.15969062784349408,
      "grad_norm": 0.7145502784859615,
      "learning_rate": 9.974995501511404e-06,
      "loss": 0.0567,
      "step": 351
    },
    {
      "epoch": 0.16014558689717925,
      "grad_norm": 1.0825694007542435,
      "learning_rate": 9.974852534607506e-06,
      "loss": 0.0897,
      "step": 352
    },
    {
      "epoch": 0.16060054595086443,
      "grad_norm": 0.8874195306329471,
      "learning_rate": 9.974709161181126e-06,
      "loss": 0.0879,
      "step": 353
    },
    {
      "epoch": 0.1610555050045496,
      "grad_norm": 0.8193025449594961,
      "learning_rate": 9.974565381243982e-06,
      "loss": 0.0969,
      "step": 354
    },
    {
      "epoch": 0.16151046405823477,
      "grad_norm": 0.76528422131405,
      "learning_rate": 9.974421194807815e-06,
      "loss": 0.0786,
      "step": 355
    },
    {
      "epoch": 0.16196542311191992,
      "grad_norm": 0.8836543328533641,
      "learning_rate": 9.974276601884416e-06,
      "loss": 0.0744,
      "step": 356
    },
    {
      "epoch": 0.1624203821656051,
      "grad_norm": 0.7482952108426273,
      "learning_rate": 9.974131602485596e-06,
      "loss": 0.0772,
      "step": 357
    },
    {
      "epoch": 0.16287534121929026,
      "grad_norm": 0.9122723647083647,
      "learning_rate": 9.973986196623203e-06,
      "loss": 0.0851,
      "step": 358
    },
    {
      "epoch": 0.16333030027297543,
      "grad_norm": 0.8373653902978805,
      "learning_rate": 9.973840384309121e-06,
      "loss": 0.0865,
      "step": 359
    },
    {
      "epoch": 0.1637852593266606,
      "grad_norm": 0.6360069343077157,
      "learning_rate": 9.973694165555264e-06,
      "loss": 0.0618,
      "step": 360
    },
    {
      "epoch": 0.16424021838034578,
      "grad_norm": 0.7967304456611868,
      "learning_rate": 9.973547540373582e-06,
      "loss": 0.0865,
      "step": 361
    },
    {
      "epoch": 0.16469517743403095,
      "grad_norm": 1.1699452577832765,
      "learning_rate": 9.973400508776054e-06,
      "loss": 0.1144,
      "step": 362
    },
    {
      "epoch": 0.1651501364877161,
      "grad_norm": 0.6282867599706373,
      "learning_rate": 9.973253070774698e-06,
      "loss": 0.0633,
      "step": 363
    },
    {
      "epoch": 0.16560509554140126,
      "grad_norm": 0.79942272506218,
      "learning_rate": 9.973105226381559e-06,
      "loss": 0.069,
      "step": 364
    },
    {
      "epoch": 0.16606005459508644,
      "grad_norm": 0.9348674828410355,
      "learning_rate": 9.972956975608719e-06,
      "loss": 0.1019,
      "step": 365
    },
    {
      "epoch": 0.1665150136487716,
      "grad_norm": 1.0942665884463076,
      "learning_rate": 9.972808318468292e-06,
      "loss": 0.0859,
      "step": 366
    },
    {
      "epoch": 0.16696997270245678,
      "grad_norm": 0.6283579225277517,
      "learning_rate": 9.972659254972426e-06,
      "loss": 0.0589,
      "step": 367
    },
    {
      "epoch": 0.16742493175614195,
      "grad_norm": 1.0989677054167046,
      "learning_rate": 9.972509785133304e-06,
      "loss": 0.1081,
      "step": 368
    },
    {
      "epoch": 0.16787989080982713,
      "grad_norm": 0.7310198219540203,
      "learning_rate": 9.972359908963137e-06,
      "loss": 0.0675,
      "step": 369
    },
    {
      "epoch": 0.16833484986351227,
      "grad_norm": 0.757671629194488,
      "learning_rate": 9.972209626474172e-06,
      "loss": 0.0734,
      "step": 370
    },
    {
      "epoch": 0.16878980891719744,
      "grad_norm": 0.7966175159886519,
      "learning_rate": 9.972058937678692e-06,
      "loss": 0.075,
      "step": 371
    },
    {
      "epoch": 0.16924476797088261,
      "grad_norm": 0.9805514159267839,
      "learning_rate": 9.97190784258901e-06,
      "loss": 0.1071,
      "step": 372
    },
    {
      "epoch": 0.1696997270245678,
      "grad_norm": 0.7000612574442994,
      "learning_rate": 9.971756341217471e-06,
      "loss": 0.0526,
      "step": 373
    },
    {
      "epoch": 0.17015468607825296,
      "grad_norm": 0.7917466702374949,
      "learning_rate": 9.971604433576456e-06,
      "loss": 0.0698,
      "step": 374
    },
    {
      "epoch": 0.17060964513193813,
      "grad_norm": 0.8412692631182211,
      "learning_rate": 9.97145211967838e-06,
      "loss": 0.0783,
      "step": 375
    },
    {
      "epoch": 0.1710646041856233,
      "grad_norm": 0.5615038895232536,
      "learning_rate": 9.971299399535685e-06,
      "loss": 0.053,
      "step": 376
    },
    {
      "epoch": 0.17151956323930848,
      "grad_norm": 0.6849745369298482,
      "learning_rate": 9.971146273160854e-06,
      "loss": 0.0774,
      "step": 377
    },
    {
      "epoch": 0.17197452229299362,
      "grad_norm": 0.6466596777060115,
      "learning_rate": 9.9709927405664e-06,
      "loss": 0.0606,
      "step": 378
    },
    {
      "epoch": 0.1724294813466788,
      "grad_norm": 0.7169884074840761,
      "learning_rate": 9.970838801764866e-06,
      "loss": 0.0839,
      "step": 379
    },
    {
      "epoch": 0.17288444040036396,
      "grad_norm": 0.9393396355410675,
      "learning_rate": 9.970684456768836e-06,
      "loss": 0.1132,
      "step": 380
    },
    {
      "epoch": 0.17333939945404914,
      "grad_norm": 12.197098173453568,
      "learning_rate": 9.970529705590918e-06,
      "loss": 0.4858,
      "step": 381
    },
    {
      "epoch": 0.1737943585077343,
      "grad_norm": 0.7355841274771772,
      "learning_rate": 9.97037454824376e-06,
      "loss": 0.0714,
      "step": 382
    },
    {
      "epoch": 0.17424931756141948,
      "grad_norm": 1.050385265783733,
      "learning_rate": 9.97021898474004e-06,
      "loss": 0.1024,
      "step": 383
    },
    {
      "epoch": 0.17470427661510465,
      "grad_norm": 0.8612087678995594,
      "learning_rate": 9.970063015092469e-06,
      "loss": 0.085,
      "step": 384
    },
    {
      "epoch": 0.1751592356687898,
      "grad_norm": 1.3886472100476919,
      "learning_rate": 9.969906639313793e-06,
      "loss": 0.1212,
      "step": 385
    },
    {
      "epoch": 0.17561419472247497,
      "grad_norm": 0.8238176964814595,
      "learning_rate": 9.96974985741679e-06,
      "loss": 0.0721,
      "step": 386
    },
    {
      "epoch": 0.17606915377616014,
      "grad_norm": 0.8718897735731601,
      "learning_rate": 9.969592669414272e-06,
      "loss": 0.0959,
      "step": 387
    },
    {
      "epoch": 0.17652411282984531,
      "grad_norm": 6.796752422837202,
      "learning_rate": 9.969435075319083e-06,
      "loss": 0.115,
      "step": 388
    },
    {
      "epoch": 0.1769790718835305,
      "grad_norm": 0.58176536820322,
      "learning_rate": 9.969277075144104e-06,
      "loss": 0.0459,
      "step": 389
    },
    {
      "epoch": 0.17743403093721566,
      "grad_norm": 0.7267253435076165,
      "learning_rate": 9.969118668902242e-06,
      "loss": 0.07,
      "step": 390
    },
    {
      "epoch": 0.17788898999090083,
      "grad_norm": 0.7682389367523258,
      "learning_rate": 9.968959856606442e-06,
      "loss": 0.0542,
      "step": 391
    },
    {
      "epoch": 0.17834394904458598,
      "grad_norm": 0.7873348185837048,
      "learning_rate": 9.968800638269682e-06,
      "loss": 0.0598,
      "step": 392
    },
    {
      "epoch": 0.17879890809827115,
      "grad_norm": 1.287713292390112,
      "learning_rate": 9.968641013904974e-06,
      "loss": 0.1442,
      "step": 393
    },
    {
      "epoch": 0.17925386715195632,
      "grad_norm": 1.085650814952146,
      "learning_rate": 9.968480983525359e-06,
      "loss": 0.0926,
      "step": 394
    },
    {
      "epoch": 0.1797088262056415,
      "grad_norm": 0.6716676596759695,
      "learning_rate": 9.968320547143918e-06,
      "loss": 0.0767,
      "step": 395
    },
    {
      "epoch": 0.18016378525932666,
      "grad_norm": 0.8467396807693714,
      "learning_rate": 9.968159704773757e-06,
      "loss": 0.0977,
      "step": 396
    },
    {
      "epoch": 0.18061874431301184,
      "grad_norm": 0.6438855833782786,
      "learning_rate": 9.967998456428021e-06,
      "loss": 0.0586,
      "step": 397
    },
    {
      "epoch": 0.181073703366697,
      "grad_norm": 0.7254140122399564,
      "learning_rate": 9.967836802119886e-06,
      "loss": 0.06,
      "step": 398
    },
    {
      "epoch": 0.18152866242038215,
      "grad_norm": 0.87517545358881,
      "learning_rate": 9.967674741862563e-06,
      "loss": 0.1016,
      "step": 399
    },
    {
      "epoch": 0.18198362147406733,
      "grad_norm": 1.0624206936058178,
      "learning_rate": 9.967512275669294e-06,
      "loss": 0.1296,
      "step": 400
    },
    {
      "epoch": 0.1824385805277525,
      "grad_norm": 1.0284720738314184,
      "learning_rate": 9.967349403553353e-06,
      "loss": 0.0862,
      "step": 401
    },
    {
      "epoch": 0.18289353958143767,
      "grad_norm": 0.8342932737384292,
      "learning_rate": 9.967186125528053e-06,
      "loss": 0.0873,
      "step": 402
    },
    {
      "epoch": 0.18334849863512284,
      "grad_norm": 1.543095569701571,
      "learning_rate": 9.967022441606734e-06,
      "loss": 0.1209,
      "step": 403
    },
    {
      "epoch": 0.18380345768880801,
      "grad_norm": 0.70731586616612,
      "learning_rate": 9.966858351802773e-06,
      "loss": 0.0726,
      "step": 404
    },
    {
      "epoch": 0.1842584167424932,
      "grad_norm": 0.6660531988680356,
      "learning_rate": 9.966693856129576e-06,
      "loss": 0.0562,
      "step": 405
    },
    {
      "epoch": 0.18471337579617833,
      "grad_norm": 0.8503640969928286,
      "learning_rate": 9.966528954600587e-06,
      "loss": 0.0838,
      "step": 406
    },
    {
      "epoch": 0.1851683348498635,
      "grad_norm": 0.6021534124846688,
      "learning_rate": 9.96636364722928e-06,
      "loss": 0.0673,
      "step": 407
    },
    {
      "epoch": 0.18562329390354868,
      "grad_norm": 0.8782816795828058,
      "learning_rate": 9.966197934029165e-06,
      "loss": 0.0845,
      "step": 408
    },
    {
      "epoch": 0.18607825295723385,
      "grad_norm": 0.9030990654346936,
      "learning_rate": 9.966031815013781e-06,
      "loss": 0.0839,
      "step": 409
    },
    {
      "epoch": 0.18653321201091902,
      "grad_norm": 0.8567507299712805,
      "learning_rate": 9.965865290196703e-06,
      "loss": 0.0935,
      "step": 410
    },
    {
      "epoch": 0.1869881710646042,
      "grad_norm": 0.8099856489670021,
      "learning_rate": 9.96569835959154e-06,
      "loss": 0.0747,
      "step": 411
    },
    {
      "epoch": 0.18744313011828936,
      "grad_norm": 0.8938878675243255,
      "learning_rate": 9.965531023211931e-06,
      "loss": 0.0854,
      "step": 412
    },
    {
      "epoch": 0.18789808917197454,
      "grad_norm": 0.735313860104022,
      "learning_rate": 9.965363281071551e-06,
      "loss": 0.0865,
      "step": 413
    },
    {
      "epoch": 0.18835304822565968,
      "grad_norm": 0.5495229598132649,
      "learning_rate": 9.965195133184108e-06,
      "loss": 0.0403,
      "step": 414
    },
    {
      "epoch": 0.18880800727934485,
      "grad_norm": 1.0700416713113117,
      "learning_rate": 9.965026579563342e-06,
      "loss": 0.1086,
      "step": 415
    },
    {
      "epoch": 0.18926296633303002,
      "grad_norm": 0.7118653717355078,
      "learning_rate": 9.964857620223024e-06,
      "loss": 0.0691,
      "step": 416
    },
    {
      "epoch": 0.1897179253867152,
      "grad_norm": 0.6871481686027417,
      "learning_rate": 9.964688255176963e-06,
      "loss": 0.0667,
      "step": 417
    },
    {
      "epoch": 0.19017288444040037,
      "grad_norm": 0.9848841869658392,
      "learning_rate": 9.964518484438998e-06,
      "loss": 0.0813,
      "step": 418
    },
    {
      "epoch": 0.19062784349408554,
      "grad_norm": 0.6311750922074311,
      "learning_rate": 9.964348308023001e-06,
      "loss": 0.0592,
      "step": 419
    },
    {
      "epoch": 0.1910828025477707,
      "grad_norm": 0.7813168734245782,
      "learning_rate": 9.964177725942881e-06,
      "loss": 0.0826,
      "step": 420
    },
    {
      "epoch": 0.19153776160145586,
      "grad_norm": 0.8572110622332836,
      "learning_rate": 9.964006738212574e-06,
      "loss": 0.0853,
      "step": 421
    },
    {
      "epoch": 0.19199272065514103,
      "grad_norm": 0.5304433423014596,
      "learning_rate": 9.963835344846056e-06,
      "loss": 0.048,
      "step": 422
    },
    {
      "epoch": 0.1924476797088262,
      "grad_norm": 0.7598521228122416,
      "learning_rate": 9.963663545857328e-06,
      "loss": 0.0757,
      "step": 423
    },
    {
      "epoch": 0.19290263876251137,
      "grad_norm": 1.1542546683489703,
      "learning_rate": 9.963491341260432e-06,
      "loss": 0.104,
      "step": 424
    },
    {
      "epoch": 0.19335759781619655,
      "grad_norm": 0.7766563582253432,
      "learning_rate": 9.963318731069437e-06,
      "loss": 0.0952,
      "step": 425
    },
    {
      "epoch": 0.19381255686988172,
      "grad_norm": 1.1319194983916299,
      "learning_rate": 9.96314571529845e-06,
      "loss": 0.1005,
      "step": 426
    },
    {
      "epoch": 0.1942675159235669,
      "grad_norm": 0.7230559135257585,
      "learning_rate": 9.962972293961608e-06,
      "loss": 0.0647,
      "step": 427
    },
    {
      "epoch": 0.19472247497725204,
      "grad_norm": 0.9863934566369588,
      "learning_rate": 9.962798467073083e-06,
      "loss": 0.0763,
      "step": 428
    },
    {
      "epoch": 0.1951774340309372,
      "grad_norm": 0.8259784410005646,
      "learning_rate": 9.96262423464708e-06,
      "loss": 0.087,
      "step": 429
    },
    {
      "epoch": 0.19563239308462238,
      "grad_norm": 0.7987139095182185,
      "learning_rate": 9.962449596697834e-06,
      "loss": 0.0671,
      "step": 430
    },
    {
      "epoch": 0.19608735213830755,
      "grad_norm": 1.130208173229934,
      "learning_rate": 9.962274553239619e-06,
      "loss": 0.119,
      "step": 431
    },
    {
      "epoch": 0.19654231119199272,
      "grad_norm": 0.7399696243677417,
      "learning_rate": 9.962099104286735e-06,
      "loss": 0.064,
      "step": 432
    },
    {
      "epoch": 0.1969972702456779,
      "grad_norm": 1.156015767405528,
      "learning_rate": 9.961923249853523e-06,
      "loss": 0.1102,
      "step": 433
    },
    {
      "epoch": 0.19745222929936307,
      "grad_norm": 0.972422739757894,
      "learning_rate": 9.961746989954349e-06,
      "loss": 0.1093,
      "step": 434
    },
    {
      "epoch": 0.1979071883530482,
      "grad_norm": 0.7766700420403171,
      "learning_rate": 9.96157032460362e-06,
      "loss": 0.0655,
      "step": 435
    },
    {
      "epoch": 0.19836214740673339,
      "grad_norm": 0.7460679115751414,
      "learning_rate": 9.961393253815767e-06,
      "loss": 0.0751,
      "step": 436
    },
    {
      "epoch": 0.19881710646041856,
      "grad_norm": 1.0684214450487566,
      "learning_rate": 9.961215777605266e-06,
      "loss": 0.0789,
      "step": 437
    },
    {
      "epoch": 0.19927206551410373,
      "grad_norm": 0.7683994291392229,
      "learning_rate": 9.961037895986615e-06,
      "loss": 0.0849,
      "step": 438
    },
    {
      "epoch": 0.1997270245677889,
      "grad_norm": 0.7270368453251704,
      "learning_rate": 9.960859608974352e-06,
      "loss": 0.0779,
      "step": 439
    },
    {
      "epoch": 0.20018198362147407,
      "grad_norm": 0.701460207303568,
      "learning_rate": 9.960680916583042e-06,
      "loss": 0.0639,
      "step": 440
    },
    {
      "epoch": 0.20063694267515925,
      "grad_norm": 0.6784619280926262,
      "learning_rate": 9.960501818827292e-06,
      "loss": 0.077,
      "step": 441
    },
    {
      "epoch": 0.2010919017288444,
      "grad_norm": 0.8064075868568972,
      "learning_rate": 9.960322315721735e-06,
      "loss": 0.0827,
      "step": 442
    },
    {
      "epoch": 0.20154686078252956,
      "grad_norm": 0.9155026735417204,
      "learning_rate": 9.960142407281039e-06,
      "loss": 0.0841,
      "step": 443
    },
    {
      "epoch": 0.20200181983621474,
      "grad_norm": 0.6167749294869733,
      "learning_rate": 9.959962093519904e-06,
      "loss": 0.054,
      "step": 444
    },
    {
      "epoch": 0.2024567788898999,
      "grad_norm": 0.8127781985331358,
      "learning_rate": 9.959781374453066e-06,
      "loss": 0.0751,
      "step": 445
    },
    {
      "epoch": 0.20291173794358508,
      "grad_norm": 0.98306444688532,
      "learning_rate": 9.959600250095294e-06,
      "loss": 0.075,
      "step": 446
    },
    {
      "epoch": 0.20336669699727025,
      "grad_norm": 0.7982130269360888,
      "learning_rate": 9.959418720461384e-06,
      "loss": 0.0834,
      "step": 447
    },
    {
      "epoch": 0.20382165605095542,
      "grad_norm": 0.7862225023823932,
      "learning_rate": 9.959236785566175e-06,
      "loss": 0.0704,
      "step": 448
    },
    {
      "epoch": 0.20427661510464057,
      "grad_norm": 0.562107514296544,
      "learning_rate": 9.959054445424532e-06,
      "loss": 0.0644,
      "step": 449
    },
    {
      "epoch": 0.20473157415832574,
      "grad_norm": 0.6089607791855781,
      "learning_rate": 9.958871700051353e-06,
      "loss": 0.0512,
      "step": 450
    },
    {
      "epoch": 0.2051865332120109,
      "grad_norm": 0.6962095067981563,
      "learning_rate": 9.958688549461573e-06,
      "loss": 0.0712,
      "step": 451
    },
    {
      "epoch": 0.20564149226569609,
      "grad_norm": 1.155217046291275,
      "learning_rate": 9.958504993670158e-06,
      "loss": 0.1049,
      "step": 452
    },
    {
      "epoch": 0.20609645131938126,
      "grad_norm": 1.0913314226134752,
      "learning_rate": 9.958321032692107e-06,
      "loss": 0.1226,
      "step": 453
    },
    {
      "epoch": 0.20655141037306643,
      "grad_norm": 22.735025633907238,
      "learning_rate": 9.958136666542455e-06,
      "loss": 0.8419,
      "step": 454
    },
    {
      "epoch": 0.2070063694267516,
      "grad_norm": 1.184019553325164,
      "learning_rate": 9.957951895236262e-06,
      "loss": 0.1113,
      "step": 455
    },
    {
      "epoch": 0.20746132848043677,
      "grad_norm": 0.7664792046331882,
      "learning_rate": 9.957766718788632e-06,
      "loss": 0.104,
      "step": 456
    },
    {
      "epoch": 0.20791628753412192,
      "grad_norm": 0.8672883026786035,
      "learning_rate": 9.957581137214695e-06,
      "loss": 0.074,
      "step": 457
    },
    {
      "epoch": 0.2083712465878071,
      "grad_norm": 0.8772220264781722,
      "learning_rate": 9.957395150529615e-06,
      "loss": 0.0986,
      "step": 458
    },
    {
      "epoch": 0.20882620564149226,
      "grad_norm": 0.7016331971826193,
      "learning_rate": 9.95720875874859e-06,
      "loss": 0.0752,
      "step": 459
    },
    {
      "epoch": 0.20928116469517744,
      "grad_norm": 0.6308822051977305,
      "learning_rate": 9.957021961886855e-06,
      "loss": 0.0608,
      "step": 460
    },
    {
      "epoch": 0.2097361237488626,
      "grad_norm": 0.9803601042372939,
      "learning_rate": 9.956834759959669e-06,
      "loss": 0.0908,
      "step": 461
    },
    {
      "epoch": 0.21019108280254778,
      "grad_norm": 0.7674462109758159,
      "learning_rate": 9.95664715298233e-06,
      "loss": 0.074,
      "step": 462
    },
    {
      "epoch": 0.21064604185623295,
      "grad_norm": 0.7450186566335193,
      "learning_rate": 9.95645914097017e-06,
      "loss": 0.0817,
      "step": 463
    },
    {
      "epoch": 0.2111010009099181,
      "grad_norm": 0.7225723661612439,
      "learning_rate": 9.956270723938553e-06,
      "loss": 0.0849,
      "step": 464
    },
    {
      "epoch": 0.21155595996360327,
      "grad_norm": 0.7190355211871646,
      "learning_rate": 9.956081901902875e-06,
      "loss": 0.0748,
      "step": 465
    },
    {
      "epoch": 0.21201091901728844,
      "grad_norm": 1.210684562087392,
      "learning_rate": 9.955892674878565e-06,
      "loss": 0.1272,
      "step": 466
    },
    {
      "epoch": 0.2124658780709736,
      "grad_norm": 0.834170476650907,
      "learning_rate": 9.955703042881087e-06,
      "loss": 0.0992,
      "step": 467
    },
    {
      "epoch": 0.21292083712465878,
      "grad_norm": 0.874478173291907,
      "learning_rate": 9.955513005925934e-06,
      "loss": 0.0858,
      "step": 468
    },
    {
      "epoch": 0.21337579617834396,
      "grad_norm": 0.5510320150423565,
      "learning_rate": 9.95532256402864e-06,
      "loss": 0.0574,
      "step": 469
    },
    {
      "epoch": 0.21383075523202913,
      "grad_norm": 0.5657171871822584,
      "learning_rate": 9.955131717204762e-06,
      "loss": 0.0671,
      "step": 470
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 0.7564664653864259,
      "learning_rate": 9.954940465469898e-06,
      "loss": 0.085,
      "step": 471
    },
    {
      "epoch": 0.21474067333939945,
      "grad_norm": 0.7594501005901694,
      "learning_rate": 9.954748808839675e-06,
      "loss": 0.0733,
      "step": 472
    },
    {
      "epoch": 0.21519563239308462,
      "grad_norm": 0.6748092428366178,
      "learning_rate": 9.954556747329754e-06,
      "loss": 0.0707,
      "step": 473
    },
    {
      "epoch": 0.2156505914467698,
      "grad_norm": 1.715089789819449,
      "learning_rate": 9.954364280955832e-06,
      "loss": 0.1045,
      "step": 474
    },
    {
      "epoch": 0.21610555050045496,
      "grad_norm": 0.6668751648778155,
      "learning_rate": 9.954171409733634e-06,
      "loss": 0.0573,
      "step": 475
    },
    {
      "epoch": 0.21656050955414013,
      "grad_norm": 0.5963716475430643,
      "learning_rate": 9.95397813367892e-06,
      "loss": 0.0752,
      "step": 476
    },
    {
      "epoch": 0.2170154686078253,
      "grad_norm": 0.9917190233932158,
      "learning_rate": 9.953784452807487e-06,
      "loss": 0.1049,
      "step": 477
    },
    {
      "epoch": 0.21747042766151045,
      "grad_norm": 0.5638529401686616,
      "learning_rate": 9.953590367135159e-06,
      "loss": 0.0547,
      "step": 478
    },
    {
      "epoch": 0.21792538671519562,
      "grad_norm": 0.6477110515460727,
      "learning_rate": 9.953395876677796e-06,
      "loss": 0.0564,
      "step": 479
    },
    {
      "epoch": 0.2183803457688808,
      "grad_norm": 0.5492055118574499,
      "learning_rate": 9.95320098145129e-06,
      "loss": 0.0505,
      "step": 480
    },
    {
      "epoch": 0.21883530482256597,
      "grad_norm": 0.8954528378372288,
      "learning_rate": 9.95300568147157e-06,
      "loss": 0.126,
      "step": 481
    },
    {
      "epoch": 0.21929026387625114,
      "grad_norm": 0.6155736143826033,
      "learning_rate": 9.952809976754593e-06,
      "loss": 0.0518,
      "step": 482
    },
    {
      "epoch": 0.2197452229299363,
      "grad_norm": 1.1486004986445648,
      "learning_rate": 9.952613867316351e-06,
      "loss": 0.1142,
      "step": 483
    },
    {
      "epoch": 0.22020018198362148,
      "grad_norm": 0.8236924325360948,
      "learning_rate": 9.95241735317287e-06,
      "loss": 0.1047,
      "step": 484
    },
    {
      "epoch": 0.22065514103730663,
      "grad_norm": 0.832372102653505,
      "learning_rate": 9.952220434340209e-06,
      "loss": 0.0729,
      "step": 485
    },
    {
      "epoch": 0.2211101000909918,
      "grad_norm": 0.7288716722109786,
      "learning_rate": 9.952023110834456e-06,
      "loss": 0.068,
      "step": 486
    },
    {
      "epoch": 0.22156505914467697,
      "grad_norm": 0.5327254294033283,
      "learning_rate": 9.951825382671739e-06,
      "loss": 0.0614,
      "step": 487
    },
    {
      "epoch": 0.22202001819836215,
      "grad_norm": 0.7204991379763186,
      "learning_rate": 9.951627249868213e-06,
      "loss": 0.0666,
      "step": 488
    },
    {
      "epoch": 0.22247497725204732,
      "grad_norm": 0.7485835393026234,
      "learning_rate": 9.95142871244007e-06,
      "loss": 0.068,
      "step": 489
    },
    {
      "epoch": 0.2229299363057325,
      "grad_norm": 0.45602532896445397,
      "learning_rate": 9.951229770403531e-06,
      "loss": 0.0414,
      "step": 490
    },
    {
      "epoch": 0.22338489535941766,
      "grad_norm": 0.7240661348572547,
      "learning_rate": 9.951030423774858e-06,
      "loss": 0.0798,
      "step": 491
    },
    {
      "epoch": 0.22383985441310283,
      "grad_norm": 0.7716352477687572,
      "learning_rate": 9.950830672570337e-06,
      "loss": 0.071,
      "step": 492
    },
    {
      "epoch": 0.22429481346678798,
      "grad_norm": 1.22677184750836,
      "learning_rate": 9.95063051680629e-06,
      "loss": 0.1373,
      "step": 493
    },
    {
      "epoch": 0.22474977252047315,
      "grad_norm": 0.7365431233953595,
      "learning_rate": 9.950429956499074e-06,
      "loss": 0.0699,
      "step": 494
    },
    {
      "epoch": 0.22520473157415832,
      "grad_norm": 0.705654951368504,
      "learning_rate": 9.950228991665078e-06,
      "loss": 0.0741,
      "step": 495
    },
    {
      "epoch": 0.2256596906278435,
      "grad_norm": 0.8261497906057415,
      "learning_rate": 9.950027622320724e-06,
      "loss": 0.0764,
      "step": 496
    },
    {
      "epoch": 0.22611464968152867,
      "grad_norm": 0.9965395262255518,
      "learning_rate": 9.949825848482465e-06,
      "loss": 0.0852,
      "step": 497
    },
    {
      "epoch": 0.22656960873521384,
      "grad_norm": 0.6807161957389707,
      "learning_rate": 9.949623670166794e-06,
      "loss": 0.074,
      "step": 498
    },
    {
      "epoch": 0.227024567788899,
      "grad_norm": 1.1216390709095547,
      "learning_rate": 9.949421087390228e-06,
      "loss": 0.0931,
      "step": 499
    },
    {
      "epoch": 0.22747952684258416,
      "grad_norm": 1.1278655216416786,
      "learning_rate": 9.949218100169322e-06,
      "loss": 0.1177,
      "step": 500
    },
    {
      "epoch": 0.22793448589626933,
      "grad_norm": 0.9160591457448575,
      "learning_rate": 9.949014708520664e-06,
      "loss": 0.1015,
      "step": 501
    },
    {
      "epoch": 0.2283894449499545,
      "grad_norm": 0.9377363057118697,
      "learning_rate": 9.948810912460872e-06,
      "loss": 0.1059,
      "step": 502
    },
    {
      "epoch": 0.22884440400363967,
      "grad_norm": 0.8760932101779023,
      "learning_rate": 9.948606712006601e-06,
      "loss": 0.0812,
      "step": 503
    },
    {
      "epoch": 0.22929936305732485,
      "grad_norm": 0.6962605051289937,
      "learning_rate": 9.948402107174537e-06,
      "loss": 0.0735,
      "step": 504
    },
    {
      "epoch": 0.22975432211101002,
      "grad_norm": 0.6501265713488487,
      "learning_rate": 9.948197097981401e-06,
      "loss": 0.0551,
      "step": 505
    },
    {
      "epoch": 0.2302092811646952,
      "grad_norm": 1.2156011775652311,
      "learning_rate": 9.947991684443942e-06,
      "loss": 0.1066,
      "step": 506
    },
    {
      "epoch": 0.23066424021838033,
      "grad_norm": 0.9679794435610901,
      "learning_rate": 9.947785866578951e-06,
      "loss": 0.0981,
      "step": 507
    },
    {
      "epoch": 0.2311191992720655,
      "grad_norm": 0.7195724631231237,
      "learning_rate": 9.94757964440324e-06,
      "loss": 0.0777,
      "step": 508
    },
    {
      "epoch": 0.23157415832575068,
      "grad_norm": 0.549427502610929,
      "learning_rate": 9.947373017933665e-06,
      "loss": 0.0516,
      "step": 509
    },
    {
      "epoch": 0.23202911737943585,
      "grad_norm": 0.5667212336170355,
      "learning_rate": 9.947165987187108e-06,
      "loss": 0.0583,
      "step": 510
    },
    {
      "epoch": 0.23248407643312102,
      "grad_norm": 0.6638127935874616,
      "learning_rate": 9.946958552180489e-06,
      "loss": 0.0723,
      "step": 511
    },
    {
      "epoch": 0.2329390354868062,
      "grad_norm": 0.5226768129517959,
      "learning_rate": 9.946750712930756e-06,
      "loss": 0.0482,
      "step": 512
    },
    {
      "epoch": 0.23339399454049137,
      "grad_norm": 0.8358986518129136,
      "learning_rate": 9.946542469454894e-06,
      "loss": 0.1037,
      "step": 513
    },
    {
      "epoch": 0.2338489535941765,
      "grad_norm": 0.6695809647699968,
      "learning_rate": 9.94633382176992e-06,
      "loss": 0.0728,
      "step": 514
    },
    {
      "epoch": 0.23430391264786168,
      "grad_norm": 1.0608546974350634,
      "learning_rate": 9.946124769892884e-06,
      "loss": 0.1192,
      "step": 515
    },
    {
      "epoch": 0.23475887170154686,
      "grad_norm": 0.5090717025630993,
      "learning_rate": 9.945915313840869e-06,
      "loss": 0.0612,
      "step": 516
    },
    {
      "epoch": 0.23521383075523203,
      "grad_norm": 0.8105130307542814,
      "learning_rate": 9.94570545363099e-06,
      "loss": 0.0838,
      "step": 517
    },
    {
      "epoch": 0.2356687898089172,
      "grad_norm": 0.7752986876049957,
      "learning_rate": 9.945495189280394e-06,
      "loss": 0.092,
      "step": 518
    },
    {
      "epoch": 0.23612374886260237,
      "grad_norm": 0.869801315379322,
      "learning_rate": 9.945284520806267e-06,
      "loss": 0.077,
      "step": 519
    },
    {
      "epoch": 0.23657870791628755,
      "grad_norm": 0.5427153243822386,
      "learning_rate": 9.94507344822582e-06,
      "loss": 0.0592,
      "step": 520
    },
    {
      "epoch": 0.2370336669699727,
      "grad_norm": 0.7368670007832758,
      "learning_rate": 9.944861971556305e-06,
      "loss": 0.0608,
      "step": 521
    },
    {
      "epoch": 0.23748862602365786,
      "grad_norm": 0.8141430793460733,
      "learning_rate": 9.944650090814998e-06,
      "loss": 0.0616,
      "step": 522
    },
    {
      "epoch": 0.23794358507734303,
      "grad_norm": 2.1096588720516425,
      "learning_rate": 9.944437806019216e-06,
      "loss": 0.0938,
      "step": 523
    },
    {
      "epoch": 0.2383985441310282,
      "grad_norm": 0.7014907085161215,
      "learning_rate": 9.944225117186306e-06,
      "loss": 0.0812,
      "step": 524
    },
    {
      "epoch": 0.23885350318471338,
      "grad_norm": 0.5078467158211916,
      "learning_rate": 9.944012024333647e-06,
      "loss": 0.0561,
      "step": 525
    },
    {
      "epoch": 0.23930846223839855,
      "grad_norm": 0.6379031604907951,
      "learning_rate": 9.943798527478652e-06,
      "loss": 0.0678,
      "step": 526
    },
    {
      "epoch": 0.23976342129208372,
      "grad_norm": 0.799876019099874,
      "learning_rate": 9.943584626638768e-06,
      "loss": 0.0914,
      "step": 527
    },
    {
      "epoch": 0.24021838034576887,
      "grad_norm": 0.6550229607349646,
      "learning_rate": 9.943370321831474e-06,
      "loss": 0.0668,
      "step": 528
    },
    {
      "epoch": 0.24067333939945404,
      "grad_norm": 0.767534839542607,
      "learning_rate": 9.943155613074279e-06,
      "loss": 0.0711,
      "step": 529
    },
    {
      "epoch": 0.2411282984531392,
      "grad_norm": 0.7571838990000624,
      "learning_rate": 9.942940500384733e-06,
      "loss": 0.0893,
      "step": 530
    },
    {
      "epoch": 0.24158325750682438,
      "grad_norm": 17.807000846945513,
      "learning_rate": 9.942724983780409e-06,
      "loss": 0.3419,
      "step": 531
    },
    {
      "epoch": 0.24203821656050956,
      "grad_norm": 1.2088422410181228,
      "learning_rate": 9.942509063278922e-06,
      "loss": 0.1173,
      "step": 532
    },
    {
      "epoch": 0.24249317561419473,
      "grad_norm": 0.8811842157145667,
      "learning_rate": 9.942292738897914e-06,
      "loss": 0.1006,
      "step": 533
    },
    {
      "epoch": 0.2429481346678799,
      "grad_norm": 0.7726281786442553,
      "learning_rate": 9.942076010655063e-06,
      "loss": 0.0909,
      "step": 534
    },
    {
      "epoch": 0.24340309372156507,
      "grad_norm": 0.9942256398778268,
      "learning_rate": 9.941858878568078e-06,
      "loss": 0.134,
      "step": 535
    },
    {
      "epoch": 0.24385805277525022,
      "grad_norm": 1.001596627292525,
      "learning_rate": 9.941641342654702e-06,
      "loss": 0.0977,
      "step": 536
    },
    {
      "epoch": 0.2443130118289354,
      "grad_norm": 0.5064863363900076,
      "learning_rate": 9.941423402932713e-06,
      "loss": 0.0559,
      "step": 537
    },
    {
      "epoch": 0.24476797088262056,
      "grad_norm": 0.8589680374278897,
      "learning_rate": 9.94120505941992e-06,
      "loss": 0.0992,
      "step": 538
    },
    {
      "epoch": 0.24522292993630573,
      "grad_norm": 0.7830880681851201,
      "learning_rate": 9.940986312134162e-06,
      "loss": 0.0825,
      "step": 539
    },
    {
      "epoch": 0.2456778889899909,
      "grad_norm": 0.5778344550660577,
      "learning_rate": 9.940767161093316e-06,
      "loss": 0.0637,
      "step": 540
    },
    {
      "epoch": 0.24613284804367608,
      "grad_norm": 0.8661775200374767,
      "learning_rate": 9.94054760631529e-06,
      "loss": 0.0958,
      "step": 541
    },
    {
      "epoch": 0.24658780709736125,
      "grad_norm": 0.6976226834296251,
      "learning_rate": 9.940327647818026e-06,
      "loss": 0.0752,
      "step": 542
    },
    {
      "epoch": 0.2470427661510464,
      "grad_norm": 0.7530160135685138,
      "learning_rate": 9.940107285619495e-06,
      "loss": 0.077,
      "step": 543
    },
    {
      "epoch": 0.24749772520473157,
      "grad_norm": 0.7997106896354084,
      "learning_rate": 9.939886519737707e-06,
      "loss": 0.0958,
      "step": 544
    },
    {
      "epoch": 0.24795268425841674,
      "grad_norm": 0.8918061918047896,
      "learning_rate": 9.939665350190702e-06,
      "loss": 0.0822,
      "step": 545
    },
    {
      "epoch": 0.2484076433121019,
      "grad_norm": 0.804115756264787,
      "learning_rate": 9.93944377699655e-06,
      "loss": 0.0915,
      "step": 546
    },
    {
      "epoch": 0.24886260236578708,
      "grad_norm": 0.6234057941022288,
      "learning_rate": 9.93922180017336e-06,
      "loss": 0.0672,
      "step": 547
    },
    {
      "epoch": 0.24931756141947226,
      "grad_norm": 0.8269450754551354,
      "learning_rate": 9.93899941973927e-06,
      "loss": 0.1102,
      "step": 548
    },
    {
      "epoch": 0.24977252047315743,
      "grad_norm": 0.9233841316663005,
      "learning_rate": 9.93877663571245e-06,
      "loss": 0.0963,
      "step": 549
    },
    {
      "epoch": 0.2502274795268426,
      "grad_norm": 0.9944861568923805,
      "learning_rate": 9.938553448111108e-06,
      "loss": 0.1127,
      "step": 550
    },
    {
      "epoch": 0.25068243858052774,
      "grad_norm": 0.8423641298780182,
      "learning_rate": 9.938329856953482e-06,
      "loss": 0.0788,
      "step": 551
    },
    {
      "epoch": 0.25113739763421294,
      "grad_norm": 0.8124861649110975,
      "learning_rate": 9.938105862257839e-06,
      "loss": 0.0831,
      "step": 552
    },
    {
      "epoch": 0.2515923566878981,
      "grad_norm": 0.6612222253979325,
      "learning_rate": 9.937881464042485e-06,
      "loss": 0.0703,
      "step": 553
    },
    {
      "epoch": 0.25204731574158323,
      "grad_norm": 0.854447666921162,
      "learning_rate": 9.937656662325759e-06,
      "loss": 0.1074,
      "step": 554
    },
    {
      "epoch": 0.25250227479526843,
      "grad_norm": 0.74521770368624,
      "learning_rate": 9.937431457126028e-06,
      "loss": 0.0777,
      "step": 555
    }
  ],
  "logging_steps": 1,
  "max_steps": 10990,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 555,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3666645319680.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}